Design of Cinnamic Acid Derivatives and Molecular Docking Toward Human-Neutral α-Glucosidase by Using Homology Modeling

Introduction

α-Glucosidase enzymes (EC 3.2.1.20) are classified into GH13 and GH31 classes.¹   These enzymes are the kind of exo-glycosidase that hydrolyzes α-D-glucopyranoside. During the hydrolysis of α-glucosidase, GH13 and GH31 α-glucosidase enzymes generate α-D-glucopyranose with retention of the configuration of the anomeric position.²  Inhibition of the α-glucosidase enzyme make glucose levels in the blood return within normal range.³ Glucosidase is responsible for the catalytic cracking glycosidic bonds specifically depending on the number of monosaccharides, the position of the cleavage site, and the configuration of the hydroxyl group in the substrate.⁴ The main mechanism of enzymatic hydrolysis of glycosidic bonds proposed by Koshland is explained through the mechanism of detention and inversion in anomeric configuration.⁵ Compounds that can inhibit α-glucosidase enzyme called α-glucosidase inhibitors (AGIs). AGIs compounds are widely used for the treatment of type 2 diabetes patients.^6-8 These inhibitors play a vital role in glycemic control. α-Glucosidase inhibitor compound inhibits α-glucosidase enzyme contained in the intestinal wall. Diabetes drugs functioning as α-glucosidase inhibitors are acarbose, miglitol, and voglibose.^3,9-11 The α-glucosidase inhibitor is recommended as first-line therapy by the International Diabetes Federation (IDF) and the American Association of Clinical Endocrinologists (AACE). AGI’s efficacy, safety, tolerability cardiovascular benefits, and lack of hypoglycemia make them suitable for diabetics ¹².

Research in the field of organic chemistry of the cinnamic acid derivatives has also been growing. In the last decades, a lot of research and development of the cinnamic acid derivatives as α-glucosidase inhibitor has been done. Some cinnamic acid derivatives which have inhibitory activity against α-glucosidase; such as aegeline, aegelinoside A and B, phenylethylcinnamide, aloeresin A, flavonol glycosides cinnamate.¹³ Several studies of structure activity relationship of α-glucosidase inhibitor class of cinnamic acid derivatives indicated that the substituent hydroxy or methoxy at position 4 in the trans-cinnamic acid could increase the activity of inhibitory of α-glucosidase enzyme. Likewise, the addition of ethylester group on the carboxylic group of cinnamic acid plays an important role to the inhibition of α-glucosidase ¹⁴.

Based on the results of the literature study, it was not yet known whether the cinnamic acid compounds whose carboxylic acid moeity was replaced by alkyl amine had good activity on α-glucosidase inhibitor or not. In this study, we analyzed 50 designs of cinnamic acid derivatives using molecular docking simulation that involves homology modeling and ligand-based virtual screening. Molecular docking remained an important tool for structure ligand based screening to find new active ligand. Molecular docking was widely used to predict protein-ligand complexes and to screen several compounds that would arrange the activity of a bi-logical receptor.¹⁵  Ligand-based virtual screening approach was used in here to calculate the lowest binding energy between α-glucosidase enzyme with ligand and also to study their binding affinity ¹⁶. We chose the enzyme α-glucosidase C-neutral protein in humans as the target for virtual screening.^17-24

Materials and Method

Homology Modeling

The target protein to be used was α-glucosidase derived from human neutral α-glucosidase. This protein was not yet available at the provider’s site Protein Data Bank (PDB) so we were looking for a target protein in Gene Bank sequence database which could be accessed at http://www.ncbi.nlm.nih.gov/genbank/. This database was created by the National Center for Biotechnology Information (NCBI). We chose the corresponding protein to be made using Swiss-Model homology with existing mold in the Protein Data Bank (PDB) which could be accessed at www.pdb.org. For the target protein we chose UniprotKB/Swiss-Prot: Q8TET4 then it was made as homology model with existing mold on PDB ID 2G3M. Model templates made by Swiss-Model had a similarity sequence identity of 31.35%.

Validation

Validation was done by comparing the value of the bond between the active ligand- receptor ligand binding site that comparison receptor ligand binding site. Analysis of the data was revealed by RMSD (Rate Mean Square Deviation) value ratio. Docking method is said to be good if its RMSD value is less than or equal to 2.0 (≤ 2.0).  Macromolecules that have been superposed were prepared for molecular docking. Macromolecules were optimized by using software Autodock 4.2. 3D structure of macromolecules added hydrogen atoms was then repaired of the charge by adding the partial charge Gasteiger and given force field using Autodock. Macromolecular structure and ligand to be docked firstly were optimized then after that each macromolecules and ligand were made their PDBQT file, also grid file parameter file (GPF), and docking file parameter file (DPF). GPF would inform Autogrid which potential receptors needed to be calculated and the type of map types that must be counted and their location. While the DPF would inform Autodock which folder to be used, which ligand to be moved, including the center and the torsion of the ligand, what docking algorithm to be used, and the number of docking should be done.

Visualization

The results of molecular docking were visualized by using Autodock and Chimera. Docking outcome parameters were analyzed. Receptor-ligand binding energy obtained from the docking was sorted by their lowest energy. And through Chimera, it was observed that amino acid residue is the closest to the ligand.

Materials

The molecular docking studies were carried out to understand the binding interaction modes of α-glucosidase enzyme with α-glucosidase inhibitors. The structures of known α-glucosidase inhibitors such as acarbose, 1-deoxynojirimycin, miglitol and voglibose were used as references, shown in Table 1. The structures of cinnamic acid derivatives compounds already have the IC₅₀ data, shown in Table 2. The structures of cinnamic acid derivatives that were to be tested are shown in Table 3.

Result and Discussion

Most of the α-glucosidase inhibitors testing were performed on yeast and rat intestinal. However, 3D protein structure of α-glucosidase unfortunately so far was not available hence the structure homology modeling for α-glucosidase were still being developed. In this paper, we observed molecular interactions of human α-glucosidase when it was inhibited by cinnamide derivatives using lead compound of cinnamic acid. The model of the enzyme was constructed based on the crystal structure of the S.solphataricus α-glucosidase, Ma1A, and human N-terminal subunit of Maltase Glucoamylase (NtMGAM). Moreover, the model were also including homology modeling of human α-glucosidase (GANC) with 914 amino acid residues from the Swiss-Prot Q8TET4 identity and using the template mold 2G3M (PDB ID).  2G3M structure is a crystal structure of α-glucosidase from Sulfolobus solfataricus Maia which was taken as a template for the Swiss-Model showed the similarity of 31.35%. Ramachandaran plot of the results of the model obtained showed the following results. The residues on the most favored regions have 391 residues with a percentage of 88.5%. Meanwhile, the residues in additional allowed regions have as much as 46 residues with percentage of 10.4%. In addition, the residues in generously allowed regions have as much as 2 residues with percentage of 0.5%. Finally, the residues in disallowed regions have as much as 3 residues with percentage of 0.7%.

Figure 1: Ribbon diagrams of the template structure of the α-glucosidase the homology modeled structure  GANC-human Click here to View figure

 

In this paper, the co-crystal ligand was detached from the active site and re-docked using AutoDock 4.2 into the binding pocket to calculate the rmsd value. The rmsd value was found to be 1.458 Å that showed the extent of overlapping of docked pose and actual pose of the co-crystal ligand. Ribbon diagrams of the template of the homology modeled structure of GANC-human α-glucosidase are shown in Figure 1.

The molecular docking studies were used to predict the interactions between cinnamic acid derivatives with the human C-neutral (GANC) α-glucosidase enzyme. Herein, we used reference compound as the standard. Results of the analysis indicated by the value of molecular docking Gibs free energy (ΔG) was shown in the table 1 below:

Table 1: Docking Scores of The known α-glucosidase inhibitor

No.	Reference sample	Result of Docking		IC50 (mM)
		-∆G (kcal/mol)	Ki (µM)
1	Acarbose (1)	-2.88	7.77	199.81
2	1-deoxynojirimycin (2)	-6.07	35.53	5.60
3	Miglitol (3)	-5.04	201.15	14.47
4	Voglibose (4)	-3.55	2.50	276.86

 

Parameters measured were free energy Gibs (-ΔG in kcal/mol) which were resulted from receptor-ligand interactions. Table 1 showed the result of docking simulation in column 2 and 3 and results from in vitro testing of references samples in column 4. It can be seen that there is significant corelation between docking simulation results and in vitro testing results.  Table 1  also showed that 1-deoxynojirimycin capable of inhibiting the human-neutral α-glucosidase enzyme and also was ranked as the highest among all the known α-glucosidase inhibitors. For reference compounds such as miglitol (3), voglibose (4) and acarbose (1), they also have the ability to inhibit the human-neutral α-glucosidase enzyme but it is still below the inhibition levels of 1-deoxynojirimycin because they have larger value Gibs free energy (ΔG) than that of 1-deoxynojirimycin. Their Gibs free energies (ΔG) were -5.04 kcal/mol, -3.55 kcal/mol, and -2.88 kcal/mol respectively.

The best ranked for the known α-glucosidase inhibitor was shown in Figure 2. This figure showed that 1-deoxynojirimicyn (1) compound was reported to form hydrogen bonds (shown as a green breaking point) with the amino acid residues HIS645, ASP398, ASP 511, ASP587, and ARG571.

Figure 2: Binding mode of  1-deoxynojirimicyn in binding pocket human α-glucosidase enzyme Click here to View figure

The simulation analysis results of molecular docking of the tested compounds are shown in the Table 2. 20 compounds which were already known of their in-vitro activities against α-glucosidase enzyme are simulated in molecular docking software in order to compare the results from in-vitro testing.  We could estimate that the extent to which the activeness of the sample compounds of docking results and the predictability of its activeness in inhibiting human α-glucosidase enzyme. The results of the docking simulation of 20 tested compounds that have Gibs free energy values (ΔG) better than that of 1-deoxynojirimicyn are compounds of 21, 22 and 24 with a value of ΔG = -6.11 kcal/mol, -6.67 kcal/mol, and -6.76 kcal/mol respectively. It can be seen that there is significant correlation between docking simulation results (which have low Gibs free energy value and low kinetic inhibition value) and in vitro testing results.

Table 2:  Docking Scores of Tested Compound against human α-glucosidase enzyme

No	Tested Samples	R1	R2	R3	R4	R5	Log P	IC50	∆G (kcal/mol)	Ki(µM)
1	Compound  5	H	H	H	H	OH	1.93	> 5   mM	-3.42	3.12
2	Compound  6	H	H	H	H	OC2H5	2.95	> 5   mM	-3.79	1.67
3	Compound  7	H	H	H	OH	OH	1.54	> 5   mM	-3.80	1.65
4	Compound  8	H	H	OH	H	OH	1.54	1.27 mM	-3.83	1.57
5	Compound  9	H	OH	H	H	OH	1.54	0.20 mM	-3.47	2.86
6	Compound  10	H	H	H	OCH3	OH	1.81	4.34 mM	-3.59	2.35
7	Compound  11	H	H	OCH3	H	OH	1.81	0.58 mM	-3.70	1.93
8	Compound  12	H	OCH3	H	H	OH	1.81	0.04 mM	-3.26	4.06
9	Compound  13	H	OCH3	H	H	OC2H5	2.41	0.05 mM	-3.74	1.82
10	Compound  14	H	NO2	H	H	OH	1.92	> 5   mM	-4.17	876.64
11	Compound  15	H	OC6H13	H	H	H	3.88	> 5   mM	-4.56	453.13
12	Compound  16	H	OC4H9	H	H	OH	3.05	> 5   mM	-4.11	973.62
13	Compound  17	H	F	H	H	OH	2.09	0.27 mM	-3.43	3.07
14	Compound  18	H	Cl	H	H	OH	2.49	0.39 mM	-4.30	702.4
15	Compound  19	H	Br	H	H	OH	2.76	> 5   mM	-3.66	2.09
16	Compound  20	H	OPhe	H	H	OH	3.47	0.44 mM	-5.19	157.79
17	Compound  21	H	OBn	H	H	OH	3.54	> 5   mM	-6.11	3.72
18	Compound  22	H	H	H	H	NHCHCHPhe	3.66	0.0358  mM	-6.67	12.96
19	Compound  23	H	OH	H	H	Phe	3.2	0.0410  mM	-5.21	151.11
20	Compound  24	OH	OH	H	H	Phe	2.81	0.0621  mM	-6.76	11.06

 

The results for the simulation of the 20 tested compounds are shown in Fig.3. Compound 22 was reported to form hydrogen bonds, shown as a green breaking point, with the amino acid residues ASP537 while the compound 24 was reported to form hydrogen bonds, shown as a green breaking point, with the amino acid residues ASP (398, 511) and TRP472.

Figure 3: Binding mode of compound 22 (a) and  24 (b) in binding human α-glucosidase enzyme Click here to View figure

Table 3 showed the results of docking simulation of 50 cinnamic acid derivatives which were molecular docked against C-neutral-human α-glucosidase enzyme. There were as many as 13 compounds (compounds of 27, 28, 37, 51, 52, 55, 58, 59, 64, 65, 66, 67 and 68) that have free bond energy value (ΔG) better than that of 1-deoxynojirimycin compound. It can be seen that there is significant correlation between docking simulation results (which have low Gibs free energy value and low kinetic inhibition value) and in vitro testing results. This means that these 13 cinnamic acid derivative compounds would tend to bind to the receptor C-humans α-glucosidase than the other cinnamic acid derivatives. Gibs free energy value (ΔG) of the docking results showed that 13 cinnamic acid derivatives have affinities for the receptor of C-Neutral human α-glucosidase that could act as α-glucosidase inhibitor.

Table 3: Docking Scores of sample compounds ligand against human α-glucosidase enzyme

No	Sample compounds	R1	R2	R3	log P	∆G (kcal/mol)	Ki (µM)
1	Compound 25	H	NH2	OH	1.13	-2.15	26.7
2	Compound  26	NH2	H	OH	1.13	-2.52	14.12
3	Compound  27	H	NH2	NH2	0.48	-6.85	9.55
4	Compound  28	H	OCH3	NH2	1.16	-6.60	14.58
5	Compound  29	H	OCH3	NHCH3	1.39	-4.07	1.04
6	Compound  30	H	OCH3	NHC2H5	1.73	-4.30	703.98
7	Compound  31	H	OCH3	NHC3H7	2.22	-4.56	452.27
8	Compound  32	H	OCH3	NHCH(CH3)2	2.05	-4.62	408.15
9	Compound  33	H	OCH3	NHC8H17	2.05	-5.71	65.77
10	Compound  34	H	OCH3	NH-Phe	3.05	-6.03	37.81
11	Compound  35	H	OH	NH2	0.89	-3.39	3.29
12	Compound  36	H	OH	NHCH3	1.13	-4.15	914.98
13	Compound  37	OH	H	NHCH3	1.13	-6.13	32.32
14	Compound  38	H	OH	NHC2H5	1.47	-4.31	697.19
15	Compound  39	H	OH	NHC3H7	1.95	-4.52	488.08
16	Compound  40	H	OH	NHC8H17	4.04	-5.86	50.36
17	Compound  41	H	NH2	NHCH3	0.71	-5.43	105.05
18	Compound 42	H	NH2	NHC2H5	1.05	-5.76	59.82
19	Compound 43	H	NH2	NHC8H17	3.63	-5.91	46.31
20	Compound 44	H	OH	NH-Phe	2.79	-5.00	216.62
21	Compound 45	H	NH2	NH-Phe	2.38	-3.58	2.39
22	Compound 46	H	OCH3	NH-Phe-OCH3	2.93	-5.02	207.54
23	Compound 47	H	NH2	NH-Phe-OCH3	2.25	-5.98	41.14
24	Compound 48	H	OCH3	NH-Glu	-0.25	-5.71	65.37
25	Compound 49	H	NH2	NH-Glu	-0.93	-5.24	143.35
26	Compound 50	H	OCH3	O-Glu	0.43	-5.66	71.17
27	Compound 51	H	NH2	O-Glu	-0.25	-6.22	27.65
28	Compound 52	NO2	H	OH	1.92	-7.06	6.71
29	Compound 53	H	NO2	OH	1.92	-4.15	866.02
30	Compound 54	NO2	NO2	NH2	1.28	-4.93	241.40
31	Compound 55	NO2	H	NH2	1.28	-7.78	1.97
32	Compound 56	H	NO2	NHCH3	1.71	-5.93	45.32
33	Compound 57	H	NO2	NHC2H5	1.92	-6.05	36.74
34	Compound 58	H	NO2	NHC3H7	2.37	-6.10	33.99
35	Compound 59	H	NO2	NHCH(CH3)2	2.24	-6.17	30.24
36	Compound 60	H	NO2	NHC8H17	2.24	-5.70	65.87
37	Compound 61	H	NO2	NH-Phe	3.27	-5.47	97.29
38	Compound 62	H	H	NHCH3	1.52	-4.07	1.04
39	Compound 63	NO2	H	NHCH3	1.71	-5.88	48.83
40	Compound 64	NO2	H	NHC2H5	1.92	-6.40	20.48
41	Compound 65	NO2	H	NHC3H7	2.37	-6.12	32.93
42	Compound 66	NO2	H	NHC8H17	4.65	-6.67	12.87
43	Compound 67	NO2	H	NH-Phe	3.27	-7.25	4.85
44	Compound 68	NO2	H	NH-Phe-OCH3	3.40	-6.63	13.78
45	Compound 69	H	NO2	NH-Phe-OCH3	3.40	-5.58	81.73
46	Compound 70	H	H	NHC2H5	1.86	-4.27	738.43
47	Compound 71	H	H	NHC3H7	2.34	-4.49	513.16
48	Compound 72	H	H	NHCH(CH3)2	2.17	-4.62	411.26
49	Compound 73	H	H	NHC8H17	4.43	-5.73	62.67
50	Compound 74	H	H	NH-Phe	3.18	-5.27	136.67

 

The best compounds from the docking simulation are shown in Fig.4. Compound 27 was reported to form hydrogen bonds, shown as a green breaking point, with the amino acid residues ASP587 and ARG643. Compound 52 was reported to form hydrogen bonds, shown as a green breaking point, with the amino acid residues TRP584 and ARG571. Compound 55 was reported to form hydrogen bonds, shown as a green breaking point, with the amino acid residues ASP (587, 616).  Compound 67 were reported to form hydrogen bonds, shown as a green breaking point, with the amino acid residues ASP (398, 587).

Figure 4: Binding mode of compounds 27(a), 52(b), 55(c) and 67(d) in binding pocket human α-glucosidase enzyme Click here to View figure

 

Conclusion

In this paper, the molecular studies were conducted on 50 cinnamic acid derivatives against C-Neutral human α-glucosidase enzyme. The docking analysis resulted in the ligand interactions with the binding site. The binding interaction of the ligands to the enzyme indicated that conserved amino acidw residues in GANC-human protein played an important role in maintaining a functional conformation and were involved in binding to the inhibitors. The interactions mechanism of the protein and inhibitors in this study were useful to understand the potential mechanism of the protein of GANC-human and α-glucosidase inhibitors interactions. Two parameters from docking simulation that can predict the inhibition activity of α-glucosidase enzyme are: low free energy Gibs and low kinetic inhibition value. The Gibs free energy value (ΔG) obtained from the docking simulation showed that 13 cinnamic acid derivatives compounds had an affinity for the receptor α-glucosidase so it could act as α-glucosidase inhibitor. There were depicted significant interactions with the important residues of binding site.

Acknowledgments

We are very grateful for the scholarship that was given by The Ministry of Research and Technology for fiscal year of 2014-2018. In addition, I also am very thankful for all the support and help from Research Center for Chemistry – Indonesian Sciences and Institute.

Conflict of Interest

We declare no conflict of interest

References

1. Frandsen, T. P. & Svensson, B. Plant Mol. Biol. 1988, 37, 1–13.
   CrossRef
2. Nashiru, O., Koh, S., Lee, S. & Lee, D. J. Biochem. Mol. Biol. 2001, 34, 347–354.
3. Bösenberg, L. H. & Zyl, D. G. van. J. Endocrinol. Metab. Diabetes South Africa.2008,  13, 3, 80-88.
   CrossRef
4. Park, H. et al. Bioorg. Med. Chem. 2008, 16, 284–292.
   CrossRef
5. Daviess, G. & Henrissat, B. Structure. 1995, 3, 853–859.
   CrossRef
6. Laar, F. A. van De et al. Diabetes Care.2005, 28, 166–175.
7. Luna, B. & Feinglos, M. N. Am. Fam. Phisician. 2011, 63, 1747–1756.
8. Andrade-cetto, A. & Becerra-jim, J. J. Ethnopharmacol.2008, 116, 27–32.
   CrossRef
9. Chiasson, J. et al. Lance. 2002, 359, 2072–2077.
   CrossRef
10. Dabhi, A. S., Bhat, N. R. & Shah, M. J. J. Clin. Diagnostic Res.2013, 7, 3023–3027.
11. Liu, Y. et al. Bioorg. Med. Chem. 2006, 14, 5683–5690.
    CrossRef
12. Kalra, S. in J Pak Med Assoc. 2014, 64, 474–476.
13. Phuwapraisirisan, P., Puksasook, T., Jong-aramruang, J. & Kokpol, U.. Bioorg. Med. Chem. Lett. 2008, 18, 4956–4958.
    CrossRef
14. Adisakwattana, S. et al. Bioorg. Med. Chem. Lett. 2004, 14, 2893–2896.
    CrossRef
15. Coleman, R. G., Carchia, M., Sterling, T., Irwin, J. J. & Shoichet, B. K. PLOS ONE. 2013, 8, 1–19.
    CrossRef
16. Mohammed, K. et al. Eur. J. Med. Chem. 2014, 81, 245–252.
    CrossRef
17. Yoshimizu, M., Tajima, Y., Matsuzawa, F. & Aikawa, S. Clin. Chim. Acta. 2008, 391, 68–73.
    CrossRef
18. Sim, L. Disertation. 2010,  1–140, University of Toronto.
19. Saqib, U. & Siddiqi, M. I. Int. J. Integr. Biol. 2009,  5, 13–19.
20. Saqib, U. & Siddiqi, M. I. Int. J. Integr. Biol.2008, 2, 116–121.
21. Azam, S. S., Uddin, R., Syed, A. A. S., Zaheer-ul-Haq. Med Chem Res.2012, 21, 1677–1683.
    CrossRef
22. Walder, K. & McMillan, J. Inter. J. Obes Relat metab Disord. 2012, 26, 442–449.
23. Mooradian, A. & Thurman, J. Drugs. 1999, 57, 19–29.
    CrossRef
24. Hirschhorn, R., Huie, M. & Kasper, J. Proc.Natl.Acad.Sci . 2002, 99, 13642–13646.
    CrossRef

---

This work is licensed under a Creative Commons Attribution 4.0 International License.