Atom based 3D QSAR and Fingerprint based 2D QSAR of Novel Molecules as MmpL3 receptor inhibitors for Mycobacterium tuberculosis

 

K Poojita¹, Fajeelath Fathima¹, Rajdeep Ray¹, Lalit Kumar², Ruchi Verma^1*

¹Department of Pharmaceutical Chemistry, Manipal College of Pharmaceutical Sciences,

Manipal Academy of Higher Education, Madhav Nagar - 576104, Manipal, Udupi, Karnataka, India.

²Department of Pharmaceutics, Manipal College of Pharmaceutical Sciences, Manipal Academy of Higher Education, Madhav Nagar - 576104, Manipal, Udupi, Karnataka, India.

 

ABSTRACT:

Tuberculosis is one of the leading cause of increase in mortality rate in today’s health care scenario. Due to increase frequency of drug resistant TB it is prudent to find new targets and promising targets for anti-tubercular activity.  MmpL3 (Mycobacterial Membrane Protein Large 3) is one of the most effective and promiscuous targets for development of new drug for anti-tubercular therapy due to its cross resistance inhibition property. In this study we have presented atom based 3D QSAR and finger print based 2D QSAR models to study different structural and functional groups of Adamantyl urea derivatives and their action in MmpL3 inhibitory activity which will provide us the insight for designing better and far more effective anti TB drugs.

 

 

INTRODUCTION:

Tuberculosis (TB) is an infectious disease which primarily affects the lungs. This contagious disease is caused by the microorganism Mycobacterium tuberculosis. Tuberculosis is one of the leading cause of death in today’s health care scenario.¹

 

According to Global Tuberculosis Report by WHO in 2018 nearly 10 million people suffered from Tuberculosis with an overall mean of 130 cases per 100,000 population per year.²

 

Hence, with the increasing prevalence of MDRTB (Multi Drug Resistance TB) and TDRTB (Total Drug Resistant TB) it is of utmost importance to develop new drugs with novel mechanism of action targeting MTb disease.

 

One of most effective and important target for anti TB drugs is the mycobacterial cell wall. Due to its unique structure it plays a very important role in Tuberculosis pathogenesis.  The cell wall mainly contains mycolic acid which is responsible for cell viability.

 

This cell wall is extremely hydrophobic and is responsible for its survival and virulence. The outer cell wall includes a long chain of mycolic acid (MAs) which are transported from the inner membrane as Trehalose Monomycolates (TMMs) with the help of a transporter to outside the cell wall where it gets incorporated as Trehalose Dimycolates (TDMs), along with noncovalently binded lipids. This distinct structure of the outer cell wall of Mtb is very rigid and extremely impermeable to many anti TB Drugs. Hence, TMMs transport and biosynthesis is essential for the survival of Mtb.^{3, 4}

 

MmpL3 (Mycobacterial Membrane Protein Large 3) which belongs to RND family (Resistance Nodulation and cell Division) is a lipid transporter that binds with the Trehalose Monomycolates and transfer it to the outer cell wall from the cytoplasm where it is required for cell wall integrity and biosynthesis. MmpL3 is one of the most recent targets for development of new drug for anti-tubercular therapy. Several compound which act as inhibitors of MmpL3 have been reported so far inhibitors like BM 212, SQ109, adamantyl urea derivatives that have shown promising inhibitory activity in vitro.

 

Over the years, the study of QSAR (Quantitative Structure Activity Relationship) has emerged as an adept technique which used as an important and efficient tool in for drug discovery and development.⁵

 

QSAR is a mathematical model of statistical correlation between biological activity and deviation structural properties in a series of chemical compounds. The QSAR helps in predicting the activity of unknown compounds using a structurally similar QSAR model. The molecular descriptors (the molecular properties used in QSAR) can be either 2D (Hydrophobicity, Bond length, bond angle, dipole moments, steric effects, electronic properties, pKa in ionic compounds etc.) or 3D (involves stereochemistry, optical activity, active site interaction etc.)^6,7,8,9_.

 

Here, a set of 1-Adamantyl-3-Heteroaryl Urea derivatives with MmpL3 inhibitory activity were selected from the literature to study and design 3D and 2D QSAR models and structural features required for MmpL3 inhibitory activity were identified. The data reported by the various QSAR models provides guidance for the designing of structurally new adamantyl urea inhibitors with potential activity against MmpL3 of M. tuberculosis.

 

Experimental:

MATERIALS AND METHODS:

All the in silico experiments were conducted using the Schrodinger Software. The chemical structures were drawn and prepared using Marvin Sketch from ChemAxon. The alignment of the ligand, generation of atom based 3D QSAR models and fingerprint based 2D QSAR were performed using Maestro version 11.4 (Schrodinger Inc.) and Canvas software from the Schrodinger Package in using an Intel Core i3-4160 processor with 4GB RAM and Intel Haswell graphics card using a Linux Ubuntu 18.04.1 LTS operating system^10,11,12,13_..

 

Molecular Dataset:

A set of 48 structures Table 1. Of 1-Adamantyl-3-Heteroaryl Urea derivatives having MmpL3 inhibitory property in M. tuberculosis H7RV strain were selected from the literature to generate QSAR models. The structure variation and the relation to its wide range of biological activity served as an ideal data for generating suitable QSAR models for predicting activity. The MIC activity data [MIC H37RV (µM)] for the above derivatives was used after being converted to the logarithmic scale pMIC H37RV as the depending variables in 3D and 2D QSAR studies^{14, 15,16,17,18.}.

 

Ligand Alignment:

It is one of the most crucial step to ensure the generation of most precise and accurate 3D QSAR model. It helps in comparing and studying the relation between the deviations of the structures in the dataset. The structures in the data set are aligned in such a way that they are superimposed over each other, this was achieved using flexible ligand alignment in the maestro program of the Schrodinger software. [Fig.1.]

 

Fig.1. Flexible Ligand Alignment of the selected dataset from the literature

 

Building Atom based 3D QSAR model.

Hereby, using the above dataset atom based 3D QSAR was built. From this dataset 70% were chosen as training set and 30% as test set to generate the best possible QSAR model with 6 PLS (Partial least square) factors to obtain a more precise cross validated coefficient R² with minimum standard deviation (<0.3-0.2). The QSAR model with the value of R² >> 0.9 or equal to 1 is usually preferred. This in turn helps in generating a better and more accurate cross validated value of q² (>>0.7). This was carried out using Atom Based QSAR in Phase application of Maestro (Schrodinger).^{19, 20, 21} Refer Table1.

 

Building Fingerprint based 2D QSAR Models:

The Canvas provides seven 2D fingerprint based models. They are explained briefly as follows: Atom pair, Atom Triplet, Dendritic, linear, molprint, radial and topological fingerprint models.

 

Here the 2D canvas fingerprints are associated with Kernel based PLS in order to generate precise and suitable QSAR models that represents the atoms in the structure and give information about the overall beneficial and non-beneficial attributes of the given dataset. The same set of test and training used in atom based 3D QSAR was utilised in the fingerprint based 2D QSAR model generation. 2D QSAR analysis see Table 2.

Table 1. Structure, Predicted and experimental activity of 1-Adamantyl-3-Heteroaryl Urea derivatives with anti-tubercular activity utilised in the test and training set in QSAR model.

Compound	Structure	MIC H37RV  (µM)	pMIC H37RV	QSAR Set	Predicted Activity	Predicted Error
1		0.00003	10.511	Training	10.2366	-0.274413
46		0.00032	9.502	Test	8.63846	-0.863198
6		0.00116	8.935	Training	8.91062	-0.0244011
2		0.00123	8.909	Training	9.25025	0.341301
4		0.00135	8.868	Training	9.16503	0.295597
28		0.00237	8.626	Training	8.40991	-0.215621
35		0.00238	8.623	Test	0.096095	0.437895
51		0.00247	8.608	Training	7.96568	-0.642469
24		0.00491	8.309	Training	8.63846	0.329927
43		0.00493	8.307	Test	8.21433	-0.0928476
37		0.00512	8.29	Training	7.89887	-0.391441
33		0.00535	8.271	Training	8.31264	0.0412451
50		0.01035	7.985	Test	7.55453	-0.430476
52		0.01035	7.985	Training	7.65847	-0.326596
34		0.01074	7.969	Training	7.91737	-0.0516071
32		0.01128	7.948	Test	7.90751	-0.0400336
13		0.01469	7.833	Training	7.61688	-0.216196
47		0.0227	7.644	Test	6.80983	-0.834181
21		0.02392	7.621	Training	7.35433	-0.266974
39		0.03951	7.403	Training	7.96568	0.562346
38		0.04134	7.384	Training	7.55453	0.170889
42		0.04133	7.384	Training	7.65847	0.274769
27		0.0454	7.343	Training	7.02542	-0.317557
15		0.04606	7.337	Test	7.24852	-0.0881216
45		0.04784	7.32	Training	7.35433	0.0340912
7		0.05412	7.267	Training	7.38997	0.123288
8		0.05464	7.263	Training	7.18819	-0.074317
12		0.05624	7.25	Training	7.23276	-0.0172287
9		0.05807	7.236	Training	7.19772	-0.0383465
14		0.06633	7.178	Training	7.00804	-0.170238
41		0.08669	7.062	Training	7.25519	0.193142
36		0.09047	7.044	Test	6.94194	-0.101564
18		0.09179	7.037	Training	6.75326	-0.283934
23		0.09806	7.009	Test	6.80983	-0.198697
44		0.18224	6.739	Training	6.8286	0.0892343
22		0.19133	6.718	Training	6.77585	0.0576337
25		0.28228	6.549	Test	6.81061	0.261293
40		0.29033	6.537	Training	6.29932	-0.237789
29		0.32279	6.491	Training	6.58683	0.0957547
30		0.33293	6.478	Test	6.54856	0.0709141
31		0.34557	6.461	Training	6.47593	0.0144623
19		0.45731	6.34	Test	6.37966	0.03987
20		0.73169	6.136	Test	6.70511	0.569433
17		0.73434	6.134	Training	6.43449	0.300391
16		0.73701	6.133	Training	6.80809	0.675562

 

Table 2.  Fingerprint based 2D QSAR data set.

Ligand molecule	MIC H37Rv  (µM)	pMIC	fp linear 1	fp radial 2	fp dendritic 3	fp molprint    2 d4	fp atompairs	fp atomtripleets 5	fp  topological       7
1	0.0000	10.511	266	67	188	15	106	767	22
46	0.000	9.502	237	61	180	15	99	539	22
6	0.001	8.935	292	70	226	19	133	893	29
2	0.001	8.909	249	61	188	15	103	682	22
4	0.001	8.868	184	66	127	14	95	545	15
28	0.002	8.626	237	61	180	15	99	538	22
35	0.002	8.63	352	69	224	16	90	569	28
51	0.002	8.608	279	66	184	16	103	617	23
24	0.005	8.309	237	61	180	15	99	539	22
43	0.005	8.290	277	70	207	17	120	723	26
37	0.005	8.290	229	60	168	15	96	532	20
33	0.005	8.271	218	57	157	14	84	432	20
50	0.010	7.985	269	59	177	15	94	518	21
52	0.010	7.985	261	64	198	16	105	617	25
34	0.010	7.969	218	57	157	14	84	432	20
32	0.011	7.948	197	52	37	13	66	315	18
13	0.015	7.833	219	72	158	16	142	1084	17
47	0.023	7.833	219	55	157	14	84	432	20
21	0.024	7.621	197	49	137	13	66	315	18
39	0.040	7.403	279	66	184	16	103	620	23
38	0.041	7.384	269	59	177	15	94	521	21
42	0.041	7.384	261	64	198	16	105	617	25
27	0.045	7.343	218	55	157	14	84	432	20
15	0.046	7.337	196	59	136	14	72	981	18
45	0.048	7.320	197	49	137	13	66	315	18
7	0.054	7.267	310	84	230	22	203	1629	27
8	0.055	7.263	325	84	237	2	217	1835	28
12	0.056	7.250	331	81	270	22	207	1627	33
9	0.058	7.236	311	80	239	21	191	1443	29
14	0.066	7.178	164	55	116	12	65	326	14
41	0.087	7.062	237	63	170	15	96	518	22
36	0.090	7.044	210	54	149	14	81	428	18
18	0.092	7.037	207	59	143	14	73	381	19
23	0.098	7.009	218	55	157	14	84	432	20
44	0.182	6.739	208	57	151	14	79	399	19
22	0.191	6.718	197	51	137	13	74	353	18
25	0.282	6.549	240	57	189	15	95	512	24
40	0.290	6.537	344	70	227	18	123	771	29
29	0.323	6.491	240	57	189	15	95	512	24
30	0.333	6.478	260	62	207	16	109	624	25
31	0.346	6.461	237	55	186	15	92	508	24
19	0.457	6.340	163	55	115	12	61	321	14
20	0.732	6.136	196	58	139	14	76	383	18
17	0.734	6.134	164	55	116	17	66	324	14
16	0.737	6.133	206	58	142	14	72	386	19

 

RESULT AND DISCUSSION:

Atom Based 3D QSAR analysis:

The atom based 3D QSAR models were developed on the basis of the flexible ligand alignment^19,20,21. The predicted activity as well the predicted errors along with the test and training set selection of the dataset is displayed on Table 1. This model was chosen due to its high Q² (0.8070) and R² (0.9128) which proves it was credible and precise. The statistical parameters of the atom based QSAR constructed based on 6 PLS factors are presented at the Table 3. The F factor and the standard deviation were 47.1 and 0.3078 respectively. The scatter plots of the test and training set of the atom based 3D QSAR are shown in Fig2.            

 

Table 3. Atom based 3D QSAR Statistics

Factors	SD	R2	F	Q^2
1	0.7185	0.4369	24.8	0.1978
2	0.6280	0.5833	21.7	0.4807
3	0.5170	0.7267	26.6	0.6520
4	0.3920	0.8481	40.5	0.7925
5	0.3650	0.8728	38.4	0.7925
6	0.3078	0.9128	47.1	0.8070

 

 

Test data set plot

 

Training Dataset Plot

Figure 2. Scatter plot for Atom Based 3D QSAR showing relation between activity observed and the predicted activity 6

 

3D QSAR visualization:

The statistics Atom type fractions are seen in Table 4. The QSAR study was visualised in fields such as Hydrogen Bond Donor, negative ionic, hydrophobic, electron withdrawing and others factors which helped us to gain insight regarding the regions of the chemical framework of the adamantyl urea derivatives in the dataset and information regarding addition or removal of particular groups which causes either positive or negative effect on the anti-tubercular activity of the molecule. The visualisation of the QSAR model are shown in Figure3. The analysis was done using the ligand molecule 1 as the reference as it has the highest MmpL3 inhibitory activity among the chosen data set and 16 as molecule with the least inhibitory activity and 42 as the molecule showing the average MmpL3 inhibitory activity. The visualisation was focused on the Hydrogen Bond Donor, negative ionic, hydrophobic, electron withdrawing and others factors in which it is shown that the dark blue and red colour regions represent the positive and negative effects on the activity.

 

Table 4. Atom Type Fractions

Factors	H-bond donor	Hydrophobic/non-polar	Negative ionic	Electron-withdrawing	Other
1.	0.100787	0.455793	0.007573	0.396786	`0.039061
2.	0.054826	0.461581	0.016217	0.384467	0.082910
3.	0.022115	0.508655	0.022350	0.368735	0.078146
4.	0.013001	0.573151	0.019799	0.308392	0.085658
5.	0.012788	0.570752	0.004901	0.320661	0.090899
6.	0.009478	0.606266	0.010302	0.268104	0.105851

 

 

3.1 Hydrogen bond Donor

 

3.2 Hydrophobic

 

3.3 Electron Withdrawing

 

3.4 Negative ionic:

Figure 3. Atom based 3D QSAR visualisation in various fields as mentioned above.

 

Fingerprint based 2D QSAR study and visualization:

The fingerprint based models in atom pair, atom triplet, dendritic, molprint, linear, radial and topological models were developed using the dataset in Table2^{22, 23, 24,25}. The topological model was found out to be the most suitable model among all the fingerprints based 2D QSAR model which helps in predicting the activity more precisely with the validating factors of training set R² and standard deviation values as 0.9644 and 0.2005 respectively and test validating set Q² and RMSE values as 0.6758 and 0.5386 respectively with six KPLS factors. The statistical data of the Fingerprint based 2D QSAR model are presented at the Table 5. The 2D QSAR is visualised for all the fingerprint model with the ligand molecule 1 as the reference as it has the highest MmpL3 inhibitory activity among the chosen data set and 16 as the least inhibitory activity and 42 and the average. The 2D QSAR fingerprint visualisation of the topological model can be seen in figure 4. It is shown such that the red and blue colour regions represent the positive and negative effects on the activity.

 

Table 5. Fingerprint based 2d qsar model statistics

Fingerprint based 2D QSAR (fp topological_7 model) Statistics

                             Training Set                        Test Set

Kpls Factors	SD	R2
1	0.3896	0.8406
2	0.2869	0.9163
3	0.2304	0.9478
4	0.2159	0.9557
5	0.2075	0.9604
6	0.2005	0.9644

 

Kpls Factors	RMSE	Q2
1	0.634	0.5506
2	0.566	0.6419
3	0.5036	0.7166
4	0.5093	0.7102
5	0.5288	0.6874
6	0.5386	0.6758

 

Figure 4. Fingerprint based 2D QSAR topological based model visualisation

 

CONCLUSION:

The atom based 3D QSAR model and fingerprint based canvas 2D QSAR model were designed and the most precise and suitable model was generated to predict the MmpL3 inhibitory activity of 1-Adamantyl-3-Heteroaryl Urea derivatives. The given study indicates that the ligand molecule 1 has high anti-tubercular activity with various possibilities of structural alteration to develop potential molecule with significant MmpL3 inhibitory activity and also predict the activity of any unknown derivative and determine its pMIC values. The data reported by the above QSAR models provides guidance for the designing of structurally new adamantyl urea inhibitors with potential activity against MmpL3 of M. tuberculosis.

 

REFERENCES:

1.      Singh S, Kumar S. Tuberculosis in India: Road to elimination. Int J Prev Med 2019; 10:114.

2.      World Health Organisation. Global Tuberculosis Report 2019. Accessed on January 2020.

3.      Chih-Chia Su, Philip A. Klenotic, Jani Reddy Bolla, Georgiana E. Purdy, Carol V. Robinson, Edward W. MmpL3 is a lipid transporter that binds Trehalose Monomycolates and phosphatidylethanolamine. Proceedings of the National Academy of Sciences 2019;116 (23) 11241-11246

4.      Grzegorzewicz AE, Pham H, Gundi VA, et al. Inhibition of mycolic acid transport across the Mycobacterium tuberculosis plasma membrane. Nat Chem Biol. 2012; 8(4):334‐341.

5.      Rayasam GV. MmpL3 a potential new target for development of novel anti-tuberculosis drugs. Expert Opin Ther Targets. 2014; 18(3):247‐256.

6.      Kumar L, Verma R. Molecular docking based approach for the design of novel flavone analogues as inhibitor of beta-hydroxyacyl-ACP dehydratase HadAB complex. Research J. of Pharm. and tech. 10(8); 2439-2445.

7.      Lohit T, Kumar L, Verma R. Design, Molecular Docking, ADME Analysis and Molecular Dynamics Studies of Novel Acetylated Schiff bases as COX-2 inhibitors. Research J. Pharm. and Tech. 13(4); 1901-1906.

8.      Ruchi V, Indira B, Mradul T, Varadaraj BG· Gautham GS. In silico studies, synthesis and anticancer activity of novel diphenyl ether-based pyridine derivatives. Molecular diversity. 23; 2019: 541–554.

9.      Ruchi V, Helena IMB, Kriti A, Indira B., Mradul T, Varadral GB, Gautham GS. Synthesis, antitubercular evaluation, molecular docking and molecular dynamics studies of 4,6-disubstituted-2-oxo-dihydropyridine-3-carbonitriles. Journal of molecular structure.1197:2019:117-133.

10.   Ruchi V, Helena IMB, Kriti A, Indira B., Mradul T, Varadral GB, Gautham GS. Synthesis, evaluation, molecular docking, and molecular dynamics studies of novel N-(4-[pyridin-2-yloxy]benzyl) arylamine derivatives as potential antitubercular agents. Drug Dev Res. 2020;81:315–328.

11.   Prem SM, Ravichandiran V, Aanandhi MV. Design, Synthesis and in silico molecular docking study of N-carbamoyl-6-oxo-1-phenyl-1, 6-dihydropyridine-3-carboxamide derivatives as fibroblast growth factor 1 inhibitor. Research Journal of Pharmacy and technology. 10(8); 2017:2527-2534.

12.   Habeela JN, Raja MKMM. In silico molecular docking studies on the chemical constituents of clerodendrum    phlomidis for its cytotoxic potential against breast cancer markers. Research Journal of Pharmacy and technology. 11 (4); 1612-1618:2018.

13.   Surakanti R, Eppakayala L, Ramchander M, Venkat RP. Synthesis and molecular docking for anti-inflammatory and anti-mitotic activities of (S)-(2-Methyl-4-(1-Phenyl-1h-Thieno/Furan [3, 2-C] Pyrazol-3-yl) Piperazin-1-yl) (Pyridin-2-yl) Methanone. Asian Journal of Research in Chemistry. 10(4); 2017:582-586.

14.   Hemalatha K, Chakkaravarthi V, Ganesa Murthy K, Kayatri R, Girija K. Molecular Properties and Docking Studies of Benzimidazole Derivatives as Potential Peptide Deformylase Inhibitors. Asian Journal of Research in Chemistry. 7(7); 2014: 644-648.

15.   North, E.J., Scherman, M.S., Bruhn, D.F., Scarborough, J.S., Maddox, M.M., Jones, V., Grzegorzewicz, A., Yang, L., Hess, T., Morisseau, C., Jackson, M., McNeil, M.R., Lee, R.E. Design, synthesis and anti-tuberculosis activity of 1-adamantyl-3- heteroaryl ureas with improved in vitro pharmacokinetic properties. Bioorg Med Chem 2013; 21 (9), pp. 2587-2599

16.   Wadhwa P, Bagchi S, Sharma A. 3D-QSAR Selectivity Analysis of 1-Adamantyl-3-Heteroaryl Urea Analogs as Potent Inhibitors of Mycobacterium tuberculosis. Curr Comput Aided Drug Des. 2015; 11(2):164‐183.

17.   Ul-Haq Z, Effendi JS, Ashraf S, Bkhaitan MM. Atom and receptor based 3D QSAR models for generating new conformations from pyrazolopyrimidine as IL-2 inducible tyrosine kinase inhibitors. J Mol Graph Model. 2017; 74: 379‐395.

18.   Wang M, Li W, Wang Y, Song Y, Wang J, Cheng M. In silico insight into voltage-gated sodium channel 1.7 inhibition for anti-pain drug discovery. J Mol Graph Model. 2018; 84:18‐28.

19.   Kristam R, Parmar V, Viswanadhan VN. 3D-QSAR analysis of TRPV1 inhibitors reveals a pharmacophore applicable to diverse scaffolds and clinical candidates. J Mol Graph Model. 2013; 45: 157‐172.

20.   An Y, Sherman W, Dixon SL. Kernel-based partial least squares: application to fingerprint-based QSAR with model visualization. J Chem Inf Model. 2013; 53(9):2312‐2321.

21.   Kunal Roy, Rudra Narayan Das. A Review on Principles, Theory and Practices of 2D-QSAR. Curr. Drug Metab, 2014; 15, 346-379.  

22.   Ganatra S. H., Patle M. R, Bhagat G. K. Studies of Quantitative Structure-Activity Relationship (QSAR) of Hydantoin Based Active Anti-Cancer Drugs. Asian J. Research Chem. 2011; 4(10): 1643-1648.

23.   R. S. Kalkotwar, R. B. Saudagar. Synthesis and QSAR Studies of Some 2,5-Diaryl Substituted-1,3,4-Oxadiazole Derivatives. Asian J. Research Chem. 2013, (11): 985-991.

24.   Ashish Mullani, J. I. Disouza. Synthesis and QSAR study of N-Substituted [5-(1H-1,2,4-Triazol-5-yl)pyridine-2-YL]methanimine Derivatives as potential Antibacterial. Asian J. Research Chem. 2015, 8(9): 561-565.

25.   V. S. Kawade, S. S. Kumbhar, P. B. Choudhari, M. S. Bhatia. 3D QSAR and Pharmacophore Modelling of some Pyrimidine Analogs as CDK4 Inhibitors. Asian J. Research Chem 2015; 8(4): 231-235.

 

 

Accepted on 02.01.2021           © RJPT All right reserved

Research J. Pharm.and Tech 2021; 14(12):6321-6329.