Modeling of vascular endothelial growth factor receptor 2 (VEGFR2) kinase inhibitory activity of 2-anilino-5-aryloxazoles using chemometric tools

Figures & data

Figure 1.  Structure of 2-anilino-5-aryloxazole derivatives.

Table I.  The observed, calculated and predicted VEGFR2 inhibition activity values of 2-anilino-5-aryloxazole derivatives (Figure 1 for structures).

			pIC50 (M)
				Calcd
S. No.	R1	R2	Obsd^*	Eq(19)	Eq(20)	Prctd LOO
1	H	H	5.92	6.00	5.98	5.99
2	2-CN	H	6.06	6.20	6.18	6.20
3	2-OMe	H	6.42	6.34	6.33	6.32
4	3-CN	H	6.03	6.01	5.99	5.98
5	3-OPh	H	5.85	6.02	5.99	6.03
6	3-CH2OH	H	6.30	6.08	6.06	6.03
7	3-SO2Et	H	6.14	6.26	6.24	6.25
8	3-CH3	H	6.06	5.96	5.94	5.92
9	4-SO2NH2	H	6.30	6.51	6.50	6.52
10	4-OPh	H	–^†	6.05	6.02	–
11	3,4-(OMe)2	H	6.85	6.71	6.71	6.67
12	3,5-(OMe)2	H	6.77	6.71	6.71	6.70
13	3,4,5-(OMe)3	H	7.00	7.07	7.07	7.09
14	2-OMe, 5-SO2NEt2	H	7.64	7.57	7.54	7.53
15	2-OMe, 5-SO2Et	H	7.30	7.01	6.98	6.97
16	2-OMe, 5-SO2Et	2-Cl	6.44	7.07	7.05	7.07
17^‡	2-OMe, 5-SO2Et	3-Cl	7.66	6.96	6.93	7.07
18	2-OMe, 5-SO2Et	3-CN	6.85	7.16	7.13	7.15
19	2-OMe, 5-SO2Et	3-OMe	7.44	7.37	7.35	7.34
20	2-OMe, 5-SO2Et	3-F	7.03	7.01	6.99	6.99
21	2-OMe, 5-SO2Et	4-OMe	7.06	7.35	7.33	7.35
22	2-OMe, 5-SO2Et	4-Cl	7.09	7.12	7.10	7.10
23	2-OMe, 5-SO2Et	4-F	7.10	7.05	7.03	7.03
24	2-OMe, 5-SO2Et	4-CN	7.29	7.24	7.22	7.21
25	2-OMe, 5-SO2Et	4-CONH2	7.74	7.35	7.33	7.30
26	2-OMe, 5-SO2Et	3-Et	7.54	7.28	7.25	7.23
27	2-OMe, 5-SO2Et	3-COMe	7.04	7.32	7.30	7.31
28	2-OMe, 5-SO2Et	3-(2-thiophenyl)	7.80	7.74	7.74	7.71
29	2-OMe, 5-SO2Et	3-(3-thiophenyl)	7.82	7.70	7.70	7.64
30	2-OMe, 5-SO2Et	3-(2-pyridyl)	7.66	7.24	7.20	7.15
31	2-OMe, 5-SO2Et	3-(3-pyridyl)	7.12	7.27	7.24	7.25
32	2-OMe, 5-SO2Et	3-(4-pyridyl)	7.11	7.31	7.27	7.29
33	2-OMe, 5-SO2Et	3-{5-(1-methylimidazole)}	7.60	7.78	7.78	7.86
34	2-OMe, 5-SO2Et	3-phenyl	7.10	7.27	7.23	7.25
35	2-OMe, 5-SO2Et	3-(2-F-phenyl)	7.05	7.28	7.24	7.26
36	2-OMe, 5-SO2Et	3-(2-Cl-phenyl)	7.41	7.25	7.21	7.19

* The inhibition concentration on molar scale, taken from Ref. [11]

^† Uncertain activity

^‡ ‘Outllier’ compound of present study.

Table II.  Descriptor classes used for the analysis of VEGFR2 inhibitory activity of derivatives of 2-anilino-5-aryloxazole and identified categories in modeling the activity.

Descriptor class (acronyms)^*	Definition and scope	Descriptors' category^†
Constitutional (CONS)	Dimensionless or 0D descriptors; independent from molecular connectivity and conformations	I
Topological and (TOPO)	2D-descriptor from molecular graphs independent conformations	I
Molecular walk walks counts (MWC)	2D-descriptors representing self-returning counts of different lengths	I
Modified Burden eigenvalues (BCUT)	2D-descriptors representing positive and negative eigenvalues of the adjacency matrix, weights the diagonal elements and atoms	I
Galvez topological charge indices (GVZ)	2D-descriptors representing the first 10 eigenvalues of corrected adjacency matrix	I
2D-autocorrelations (2DAUTO)	Molecular descriptors calculated from the molecular graphs by summing the products of atom weights of the terminal atoms of all the paths of the considered path length (the lag)	I
Functional groups (FUN)	Molecular descriptors based on the counting of the chemical functional groups	I
Atom centered fragments (ACF)	Molecular descriptors based on the counting of 120 atom centered fragments, as defined by Ghose-Crippen	I
Empirical (EMP)	1D-descriptors represent the counts of non-single bonds, hydrophilic groups and ratio of the number of aromatic bonds and total bonds in an H-depleted molecule	I
Properties (PROP)	1D-descriptors representing molecular properties of a molecule	I

* Reference [12]

^† Descriptor categories identified at the end of second stage; in this the filter values are: filter-1 as 0.3, filter-2 as 2.0, filter-3 as 0.71, and filter-4 as 0.3 ≤ Q² ≤ 1.0, the number of compounds in each dataset was 35.

Table III.  The QSAR models emerged in primary descriptors* from ten different classes.

Des. class	Constant	Des-1	Des-2	r	Q^2	s	F	AIC	FIT	Eqn.
CONS	5.343	0.249 RBN		0.829	0.659	0.337	72.51	0.128	2.014	(1)
	4.734	0.237 RBN	0.633 nR05	0.881	0.741	0.289	55.87	0.099	2.865	(2)
TOPO	− 39.869	19.482 LP1		0.849	0.691	0.318	85.24	0.114	2.368	(3)
	4.382	0.012 TIC2	0.826 CIC5	0.875	0.735	0.297	52.11	0.105	2.673	(4)
MWC	4.877	3.359 MWC08		0.809	0.616	0.354	62.55	0.141	1.738	(5)
	4.877	9.252 MWC09	0.051 SRW05	0.837	0.656	0.334	37.60	0.133	1.928	(6)
BCUT	2.760	4.107 BELe6		0.862	0.716	0.306	95.04	0.105	2.640	(7)
	51.406	− 25.536 BELm2	1.333 BEHv3	0.794	0.536	0.372	27.33	0.164	1.402	(8)
GVZ	5.350	1.325 GGI4		0.848	0.691	0.319	84.54	0.114	2.348	(9)
2DAUTO	4.975	0.020 ATS5e		0.845	0.686	0.322	82.55	0.116	2.293	(10)
	7.960	0.032 ATS7m	− 2.290 GATS2p	0.870	0.725	0.300	50.25	0.108	2.577	(11)
FUN	6.108	0.480 nCp		0.786	0.582	0.372	53.61	0.155	1.489	(12)
	4.887	0.871 nRSR	0.342 nHAcc	0.848	0.694	0.324	41.16	0.125	2.111	(13)
ACF	6.032	0.398 C-005	0.728 C-006	0.832	0.645	0.340	35.89	0.137	1.841	(14)
EMP	5.345	1.172 Ui	− 5.464 ARR	0.830	0.640	0.342	35.33	0.139	1.812	(15)
PROP	5.560	0.019 PSA		0.774	0.557	0.382	49.36	0.163	1.371	(16)
	5.095	0.029 MR	− 0.353 MLOGP	0.809	0.592	0.360	30.38	0.154	1.558	(17)

* RBN, number of rotatable bonds; nR05, number of five membered rings; LP1, eigenvalue based Lovasz-Pelikan index; TIC2, total information content index of neighborhood symmetry of order-2; CIC5, complementary information content of neighborhood symmetry of order-5; MWC08 and MWC09, molecular walk count of order-08 and -09 respectively; SRW05, self returning walk count of order-5; BELe6, lowest eigenvalue no. 6 of Burden matrix/weighted by atomic Sanderson electronegativities; BELm2, lowest eigenvalue no. 2 of Burden matrix/weighted by atomic masses; BEHv3, highest eigenvalue no. 3 of Burden matrix/weighted by atomic van der Waals volumes; GGI4, topological charge index of order-4; ATS5e, Broto-Moreau autocorrelation of lag 5/weighted by atomic Sanderson electronegativities; ATS7m, Broto-Moreau autocorrelation of lag 7/weighted by atomic masses; GATS2p, Geary autocorrelation of lag 2/weighted by atomic polarizabilities; nCp, number of total primary sp³ carbon atoms; nRSR, number of sulphurs; nHAcc, number of hydrogen bond acceptor N, O, F atoms; C-005, -CH₃X; C-006, -CH₂X-; Ui, the unsaturation index; ARR, ratio of the number of aromatic bonds over the total number of non-H bonds; PSA, fragment based polar surface area; MR, Ghose-Crippon molar refractivity; MLOGP, Moriguchi octanol-water partition coefficient.

Table IV.  Descriptors identified for modeling the VEGFR2 inhibitory activity along with the average regression coefficient, standard deviation and the total incidence.

Descriptor^*	Avg reg coeff(sd) total incidence^†		Descriptor^*	Avg reg coeff(sd) total incidence^†
nR05	0.558(0.082)6	CONS	nBnz	− 0.407(0.066)2
X4v	0.534(0.000)1	TOPO	TIC1	0.018(0.000)1
SRW07	0.006(0.001)5	MWC	SRW09	0.001(0.000)2
BEHe7	3.875(0.158)7	BCUT	–	–
GGI6	2.004(0.169)3	GVZ	–	–
ATS7m	0.035(0.000)1	2DAUTO	ATS5e	0.019(0.000)1
ATS8e	0.020(0.000)2		MATS4v	− 3.279(0.365)4
MATS4p	− 3.247(0.824)2		GATS4v	2.974(0.501)4
GATS2e	0.893(0.022)2		GATS2p	− 1.822(0.000)1
nCaH	− 0.103(0.007)3	FUN	–	–
C-024	− 0.100(0.000)1	ACF	–	–

* The descriptors are identified from the three parameter models emerged from CP-MLR protocol with filter-1 as 0.3; filter-2 as 2.0; filter-3 as 0.89; filter-4 as 0.3 ≤ Q² ≤ 1.0; number of compounds in the study are 35; CONST: nR05 is number of 5-membered rings; TOPO: X4v is valence connectivity index chi-4; TIC1 was total information content index (neighborhood symmetry of 1-order); MWC: SRW07 and SRW09 are self-returning walk count of order 07 and 09 respectively; BCUT: BEHe7 highest eigenvalue n.7 of Burden matrix/weighted by atomic Sanderson electronegativities; GALV: GGI6 topological charge index of order 6; 2DAUTO: ATS7m is Broto-Moreau autocorrelation of a topological structure—lag 7/weighted by atomic masses; ATS5e and ATS8e are Broto-Moreau autocorrelation of a topological structure—lag 5 and 8 respectively/weighted by atomic Sanderson electronegativities, MATS4v and MATS4p are Moran autocorrelation—lag 4/weighted by atomic van der Waals volumes and polarizabilities respectively, GATS4v Geary autocorrelation—lag 4/weighted by atomic van der Waals volumes, GATS2e and GATS2p are Geary autocorrelation—lag 2/weighted by Sandersons electronegativities and polarizabilities, respectively; FUN: nCaH number of unsubstituted aromatic C(sp2); ACF: C-024 R-CH-R; also see ref. [23]

^† The average regression coefficient of the descriptor corresponding to all models and the total number of its incidences; the arithmetic sign represents the sign of the regression coefficient in the models.

Figure 2.  Plots of observed versus calculated and predicted pIC₅₀ values.

Table V.  Predicted residual activity of different test sets (9 compounds each) of the compounds listed in Table I.

	Residual^*
Compd. No.	Des Clus^†	Act Clus^‡	Random Clus^¶
1	− 0.01		− 0.11
3		0.13
5	− 0.15	− 0.19	− 0.16
8		0.10
9	− 0.23		− 0.25
10	6.03	6.07	6.05
13			− 0.10
14		0.05
15	0.34	0.30
17	6.90	6.96	6.93
18		− 0.32	− 0.25
19	0.14
22	0.01		− 0.01
26			0.31
28	0.09	0.09
30	0.52		0.50
31	− 0.06
34		− 0.19	− 0.10
35		− 0.25
Training set^§
r	0.918	0.917	0.929
s	0.231	0.244	0.226
F	37.828	37.112	44.283
Q^2	0.795	0.788	0.828
Test set^∥
r^2	0.883	0.899	0.856

* The difference of observed and predicted pIC₅₀ values; the training model have 25 compounds each

^† Test set from the cluster analysis of full descriptor data set of the compounds

^‡ Test set from the cluster analysis of the activity of the compounds

^¶ Test set from the random selection of the compounds

^§ The statistics for the training sets were derived with the descriptors of Equation (20)

^∥ In computing the r² values, the descriptors were considered as in Equation (20) for nine compounds in each of three test sets.

(a) AR Katritzky, V Lobnov, and M Karelson. CODESSA (Comprehensive descriptors for structural and statistical analysis). Gainesville, FL: University of Florida; (1994) (b) AR Katritzky, S Perumal, R Petrukhin, E Kleinpeter. CODESSA-based theoretical QSPR model for hydantoin HPLC-RT lipophilicities. J Chem Inf Comput Sci 2001; 41:569–574.

 Google Scholar

DRAGON software version 3.0-2003. By R Todeschini, V Consonni, A Mauri, M Pavan. Milano, Italy. http//disat.unimib.it/chm/Dragon.htm..

 Google Scholar

H Kubinyi. (1994). Variable selection in QSAR studies. II. A highly efficient combination of systematic search and evolution. Quant Struct–Act Relat 13:393–401.

 Google Scholar