Computer-Aided Design of Negative Allosteric Modulators of Metabotropic Glutamate Receptor 5 (mGluR5): Comparative Molecular Field Analysis of Aryl Ether Derivatives

Abstract

The metabotropic glutamate receptors (mGlu receptors) have emerged as attractive targets for number of neurological and psychiatric disorders. Recently, mGluR5 negative allosteric modulators (NAMs) have gained considerable attention in pharmacological research. Comparative Molecular Field Analysis (CoMFA) was performed on 73 analogues of aryl ether which were reported as mGluR5 NAMs. The study produced a statistically significant model with high correlation coefficient and good predictive abilities.

Graphical abstract

The metabotropic gluatamate receptors (mGlu receptors) have emerged as attractive therapeutic targets for several psychiatric and neurological problems.¹ mGlus are family of eight G-protein-coupled receptors (GPCRs) that are clustered into group1, group 2 and group 3.² Glutamate (L-glutamic acid) is the major excitatory neurotransmitter in the central nervous system. It binds to the orthosteric binding site of mGluR which are highly conserved. Thus, designing selective compounds targeting the orthosteric binding sites is highly challenging. To overcome the selectivity problems associated with the ligands binding to orthosteric binding sites, allosteric modulators are being developed.^3,4

Recently, mGluR5 negative allosteric modulators (NAMs) have gained considerable attention in pharmacological research. Todate, numerous acetylenic and non-acetylenic NAMs have been identified.^1,5-10 The recent revelation of the crystal structures of mGluR5 bound with NAMs explains the crucial interactions.^10,11 Previous study provided structure activity relationship on many mGluR5 NAMs. mGluR5 negative allosteric modulators (NAMs) have now entered human clinical trials. Mavoglurant (Novartis) has completed phase III trial and basimglurant (Roche) (Figure 1) has completed phase II trial.^12,13 Clinical trials with Dipraglurant (Addex), is ongoing. Despite the huge amount of work, there is no comprehensive report on quantitative structure activity relationship study of mGluR5 NAMs in the literature.

Figure 1.

Structures of some known NAMs of mGluR5

Our aim was to develop predictive QSAR models using comparative molecular field analysis (CoMFA).

We have been involved in the design of COX-2 inhibitors,^14-17 HMGR inhibitors¹⁸ using 3-D QSAR studies. In this study, CoMFA¹⁹ was applied to a series of mGluR5 NAMs belong to aryl ethers. The predictive ability of the models was validated by using external test set of molecules.

To develop 3-D QSAR models for mGluR5 NAMs, we selected a series of aryl ethers⁷ that encompasses compounds with structural diversity and wider range (pIC₅₀ range: 4.5-8.13) of biological activity. The chemical structures and pIC₅₀ values of the compounds used for the present study are given in Table 1. We used sybyl²⁰ software (X2.1.1) of Tripos Inc for molecular modeling and 3-D QSAR studies. Tripos force field, Powel method including Geometry optimization was done using Tripos force field, Powel method²¹ including Gasteiger-Hückel method. Out of 73 molecules, training set was constructed by taking 58 molecules and the remaining molecules were used as test compounds. We used “random” option during the CoMFA in order to select test set of molecules and many such trials were attempted till we get statistically significant models. We finally ensured that the selected molecules for test set have enough structural diversity and wider range of pIC₅₀ values. One of the most active molecules (78) was subjected to conformational analysis. The lowest energy conformation was used as a template. All other compounds were built by incorporating appropriate structural modifications on the template followed by minimization using Tripos force field.

Table 1.

The structures and actual and predicted inhibitory activities

No	Series	R1	R2	A	Actual pIC50	Pred pIC50	Residual	Set
3	I				5.71	5.73	-0.02	TR
9	I				5.66	5.23	0.43	TR
10	I				6.07	5.82	0.25	TR
11	I				<5.0	5.25	-0.25	TR
12	I				6.45	6.23	0.22	TS
13	I				<5.0	4.70	0.30	TR
14	I				5.73	5.67	0.06	TR
15	I				5.44	5.41	0.03	TR
16	I				5.17	5.35	-0.18	TR
17	I				<5.0	5.06	-0.06	TR
18	I				6.11	6.32	-0.21	TR
19	I				<4.5	4.19	0.31	TR
20	I				<4.5	4.63	-0.13	TR
21	I				<4.5	4.47	0.03	TR
22	I				<4.5	5.06	-0.56	TS
23	II	H			6.69	7.22	-0.53	TR
24	II	F			7.62	7.43	0.19	TR
25	II	Cl			7.94	7.99	-0.05	TR
26	II	CF3			6.55	6.50	0.05	TR
27	III	F			6.52	6.42	0.10	TR
28	III	Cl			7.49	6.97	0.52	TR
29	III	CH3			7.72	7.57	0.15	TS
30	III	CF3			5.26	5.48	-0.22	TR
31	III	CN			6.50	6.35	0.15	TR
32	IV	H	F		5.75	6.21	-0.46	TR
33	IV	4-(C3H5)	F		6.62	5.78	0.84	TS
34	IV	4-CH3, 5-F	F		7.27	7.17	0.10	TR
35	IV	4-CH2F	F		6.66	6.75	-0.09	TR
36	IV	4-CHF2	F		5.66	5.82	-0.16	TS
37	IV	4-CF3	F		5.49	5.77	-0.28	TS
38	IV	4-(C3H5)	Cl		7.09	6.35	0.74	TS
39	IV	4-CH3, 5-F	Cl		7.44	7.72	-0.28	TS
40	IV	4-CH2F	Cl		7.01	7.31	-0.30	TS
41	IV	4-CHF2	Cl		6.51	6.37	0.14	TR
42	IV	4-CF3	Cl		6.68	6.34	0.34	TR
43	V	H	F		7.02	6.94	0.08	TR
44	V	3-F	F		<5.0	6.90	-1.90	Outlier
45	V	4-Cl	F		7.38	6.77	0.61	TR
46	V	4-CH3	F		6.67	6.34	0.33	TR
47	V	4-CF3	F		5.98	5.49	0.49	TR
48	V	5-Cl	F		5.97	6.33	-0.36	TS
49	V	5-CN	F		<4.5	4.60	-0.10	TR
50	V	6-F	F		6.59	7.13	-0.54	TR
51	V	6-Cl	F		7.35	7.15	0.20	TR
52	V	6-OCH3	F		6.18	6.17	0.01	TR
53	V	6-CHF2	F		6.83	6.68	0.15	TR
54	V	6-CF3	F		5.64	5.99	-0.35	TR
55	V	H	Cl		7.77	7.46	0.31	TR
56	V	3-F	Cl		<5.0	7.41	-2.41	Outlier
57	V	4-F	Cl		6.47	6.55	-0.08	TR
58	V	4-Cl	Cl		6.77	7.33	-0.56	TR
59	V	4-CH3	Cl		6.85	6.84	0.01	TR
60	V	4-CH2CH3	Cl		5.35	7.05	-1.70	Outlier
61	V	4-CF3	Cl		5.36	5.92	-0.56	TR
62	V	5-Cl	Cl		6.99	6.85	0.14	TR
63	V	5-CF3	Cl		<4.5	4.59	-0.09	TR
64	V	6-F	Cl		8.00	7.66	0.34	TR
65	V	6-Cl	Cl		7.86	7.67	0.19	TR
66	V	6-CH3	Cl		7.75	7.71	0.04	TR
67	V	6-CH2F	Cl		7.74	7.62	0.12	TR
68	V	6-CHF2	Cl		7.44	7.16	0.28	TR
69	V	6-CF3	Cl		6.51	6.55	-0.04	TR
70	V	6-OCH3	Cl		7.27	7.28	-0.01	TS
71	V	4-CH3, 5-F	Cl		5.49	6.19	-0.70	TR
72	V	5-F, 6-CH3	Cl		7.30	7.28	0.02	TR
73	V	5-F, 6-CH2F	Cl		6.84	6.76	0.08	TR
74	V	5-F, 6-CHF2	Cl		6.30	6.78	-0.48	TR
75	VI	5-F	CH3	N	5.35	6.08	-0.73	TS
76	VI	6-CH3	CH3	N	7.02	6.87	0.15	TR
77	VI	5-F	H	CF	7.61	7.44	0.17	TR
78	VI	6-CH3	H	CF	8.13	8.45	-0.32	TR
79	VI	5-F	H	CCN	7.57	7.51	0.06	TR
80	VI	6-CH3	H	CCN	7.99	8.30	-0.31	TR

Molecular alignment was performed using database alignment method by taking the most active molecule 78. Molecular alignment was done by taking the common fragment shown in figure 2a and the alignment is shown in figure 2b. Standard procedures were followed to generate the CoMFA fields.¹⁹ The partial least square (PLS) algorithm with leave-one-out method was used to find out the cross validated r² (Q²) . The statistical metrics are given in table 2. The cross validated r² (Q²) was found to be 0.628 and the conventional R² value was found to be 0.925.

Figure 2.

(a) Common fragment used for alignment (b) Aligned molecules.

Table 2.

The summary of the results of CoMFA

	CoMFA
r2cv	0.628
NOC	8
r2_ Conv	0.925
SEE	0.311
r2 Pred	0.762

The abilities to predict the molecules that were not included in training the models were assessed by predicting the biological activities of test set molecules. Predictive r² was calculated using the following formula.¹⁴

$$r^2\mathrm{predictive}=\frac{\mathrm{SD}-\mathrm{PRESS}}{\mathrm{SD}}$$

The CoMFA method generated the contour plots of steric and electrostatic interactions which are depicted in Figure 3. Green- and yellow-colored contours represent the steric interactions. Bulky substituents in the regions shaded green are likely to enhance biological activity. A green contour near the meta position of the phenyl core characterizes the regions where bulky substituents would increase the activity. The predicted activities from the CoMFA model for the molecules with substitutions on phenyl ring at meta position, 24-31, are in reasonable agreement with the actual activity. Green-shaded regions located near the amino aryl/heteroaryl rings indicate that substitutents (6^th position, series V, VI and 4th position of series IV) on this ring would enhance the biological activity. The predicted activities from the model for compounds having substituents on pyridyl ring (next to amide) at 6^th position, 64-70, 76, 78, 80 and for compounds having substituents on thiazole ring at 4th position, 33, 35, 38 and 40 are in reasonable agreement with the actual activity. Model failed to predict the activities of three compounds, namely, 44, 56 , and 60. These three molecules can be treated as outliers. 44 and 56 are the only two compounds in the selected data set having substituent in 3^rd position of the pyridyl ring that may be the reason for the incorrect predictions.

Figure 3.

CoMFA steric and electrostatic contour plots (STDEV*COEFF). Molecule 78 is displayed in the background for reference

The electrostatic fields based on PLS analyses are represented by blue and red contours (Figure 3). Red-shaded region near the meta position of the ether linked pyridyl ring suggest that the electronegative substituents are likely to enhance biological activity. The predicted activities of compounds with flourine and nitrile substituents, 77-80, are good. Electropositive substituents in the regions shaded blue are likely to boost biological activity. Blue-shaded regions near the amide pyridyl ring indicate that the electropositive substituents are likely to enhance biological activity. The predicted activities from the model for the molecules 33, 46, 59, 66, 76, 78 and 80 are in reasonable agreement with the actual activity. Similarly, another blue contour near the meta position of the amide phenyl ring characterizes the region where electropositive substituents would increase the activity. The predicted activity of compound 29 is good. Figure 4 shows a plot between actual and predicted activities of all the molecules.

Figure 4.

Plot between actual and predicted inhibitory activities (pIC₅₀) for all the molecules

In summary, CoMFA of 73 analogues of aryl ethers produced good model with high predictive ability. The resulted contour maps highlight the structural features pertinent to biological activities. The information gathered from this study might be useful to expand the design and development of novel mGluR5 NAMs. The current study is purely of ligand based approach, understanding the active site interactions and prediction of binding mode are currently undergoing that will be published in near future.