Synthesis, in vitro inhibitory activity, kinetic study and molecular docking of novel N-alkyl–deoxynojirimycin derivatives as potential α-glucosidase inhibitors

Abstract A series of novel N-alkyl-1-deoxynojirimycin derivatives 25 ∼ 44 were synthesised and evaluated for their in vitro α-glucosidase inhibitory activity to develop α-glucosidase inhibitors with high activity. All twenty compounds exhibited α-glucosidase inhibitory activity with IC50 values ranging from 30.0 ± 0.6 µM to 2000 µM as compared to standard acarbose (IC50 = 822.0 ± 1.5 µM). The most active compound 43 was ∼27-fold more active than acarbose. Kinetic study revealed that compounds 43, 40, and 34 were all competitive inhibitors on α-glucosidase with Kiof 10 µM, 52 µM, and 150 µM, respectively. Molecular docking demonstrated that the high active inhibitors interacted with α-glucosidase by four types of interactions, including hydrogen bonds, π–π stacking interactions, hydrophobic interactions, and electrostatic interaction. Among all the interactions, the π–π stacking interaction and hydrogen bond played a significant role in a various range of activities of the compounds.


Introduction
a-Glucosidase is a type of glucosidases that acts on 1,4-a-bonds, which locate on the brush edge of the small intestine and play a critical role in digestion and absorption of carbohydrates 1 . Inhibition of a-glucosidase is one approach to delay the absorption of glucose and decrease the postprandial blood glucose level 2 . Therefore, a-glucosidase inhibitors are widely used for the prevention and treatment of typeII diabetes millitus 3 . Besides, a-glucosidase participates in other physical and biological processes as well and may also be used as a therapeutic agent for other diseases, such as cancer 4 and HIV 5 . Today, several types of a-glucosidase inhibitors are being clinically used for the treatment of typeII diabetes millitus, such as acarbose, voglibose, and miglitol 6 . However, these medications also have adverse effects, including abdominal discomfort, diarrhoea, and flatulence 7 . So, developing novel a-glucosidase inhibitors is critical and attractive.
Iminosugars are sugars in which the endocyclic oxygen is replaced by a basic nitrogen atom 8 . As a-glucosidase inhibitor, the best known naturally occurring iminosugar was 1-deoxynojirimycin (1-DNJ), which was first isolated from the roots of mulberry trees 9 . 1-DNJ is currently under clinical evaluation, not only acting as an a-glucosidase inhibitor 10 but also a potent drug for cancer 11 and HIV 12 . The accepted mechanism is that 1-DNJ inhibits a-glucosidase by competitively blocking the active site of the enzyme 13 , and the nitrogen atom can mimic the charge of proposed transition states of oxocarbonium ion formed during hydrolysis 14 . Pharmacokinetic studies showed that DNJ & DMJ (structurally related to DNJ) were rapidly eliminated from the body in an intact form by renal excretion, resulting in a weak effect on reducing blood sugar in vivo (Nakagawa et al. 15 and Faber et al. 16 ). Modification of 1-DNJ by increasing the alkalinity and introducing hydrophobic groups led to significant changes in the potency and specificity of inhibition. Ardes et al. 17 and Rawlings et al. 18 synthesised a number of deoxynojirimycin derivatives by combining different groups with a series of the N-alkyl chain length, which showed various degrees of inhibition on a-glucosidase and other enzymes. Zhang et al. 19 synthesised hybrids of 1-DNJ and quinazoline and obtained fifteen compounds, and some compounds exhibited significant inhibitory activities against the epidermal growth factor receptor (EGFR) tyrosine kinase and a-glucosidase. Some new N-alkyl, alkenyl, and benzyl substituted DNJ derivatives incorporating a silicon atom in the substituent were synthesised, which showed activity as potent and selective inhibitors of intestinal glucosidase 20 .
Here, we report for the first time the synthesis of a novel series of 1-DNJ derivatives with benzylidene acetone backbone groups (i.e. cinnamic acid, methyl 4-hydroxycinnamate, Ethyl 4 0 -hydroxy-3 0 -methoxycinnamate, and 2 0 -hydroxychalcone) and different length of alkyl chains, and obtained compounds 25-44. Moreover, all the compounds were evaluated for their a-glucosidase inhibitory activities. Furthermore, kinetic study and molecular docking were also performed to study the mechanism and discussed the structure-activity relationship.

Chemistry
The N-alkyl-deoxynojirimycin derivatives 25-44 were synthesised as shown in Scheme 1.

Results and discussion
3.1. In vitro a-glucosidase inhibitory activity All the synthesised target products 25-44 were screened to check theirs in vitro a-glucosidase inhibitory activity. All the synthesised compounds show activity on a-glucosidase with IC 50 ranging from 30 ± 0.60 mM to 2000 mM as compared with acarbose (IC 50 ¼ 822.0 ± 1.5 mM). The results were shown in Table 1.
Among all the tested compounds, compound 43 exhibited high a-glucosidase inhibitory activity with IC 50 of 30.0 ± 0.60 mM which is $27-fold higher than acarbose. Similarly, compound 40 showed an excellent activity with IC 50 of 160.5 ± 0.60 mM, around 5-fold better than acarbose. Others also exhibited inhibitory activities.
In group "B," compound 34 with the length of six carbon (n ¼ 5) showed the most inhibitory activity (IC 50 ¼ 417.0 ± 0.14 mM), and was also the third most active compound among all the synthesised compounds. It is the same as the group "A," the inhibitory activity was poor for the compounds with the alkyl chain containing less than four carbon. While the alkyl chain was long with more than four carbon, the inhibitory activity of the compounds increases as the length of the alkyl chain increases.
In group "C," only compound 35 (purity: 91.2%) showed a low activity with an IC 50 value of 966.2 ± 0.40 mM. Changing the alkyl chain length of the compounds led to no significant improvement in inhibitory activity. Compared with group "B," it was suggested that the compounds without a methoxy group in the 3 0 -position of phenyl ring were more active than that with a methoxy group.
In the case of group "D," all compounds were found to have excellent inhibitory activity with IC 50 values between 30.0 ± 0.60 mM and 571.6 ± 0.60 mM when compared with acarbose (IC 50 ¼822.0 ± 1.5 mM). Notably, compound 43 with the length of five carbon (n ¼ 4) displayed the highest activity with IC 50 value of 30.0 ± 0.60 mM. This compound was also the most active compound among all the synthesised compounds. In addition, compound 40 with an alkyl chain of two carbon (n ¼ 1) was the second most active compound among all the synthesised compounds (IC 50 ¼ 160.5 ± 0.60 mM). The two compounds were both active than 1-DNJ (IC 50 ¼ 222.4 ± 0.50 mM) in inhibiting a-glucosidase. Compared to group "D" with other groups, compounds in group "D" have two phenyl rings, but compounds in other groups have only one, suggesting that the number of phenyl rings plays an important role in compounds' inhibitory activity.

Kinetic study
To study the inhibition mode of synthesised compounds on a-glucosidase, kinetic studies were performed with the three most active compounds 43, 40, and 34. The type of inhibition and value of K i were determined by Lineweaver-Burk plots. As shown in Figure 2, when increasing concentrations of compound 43, 40, and 34, the V max was not affected, while the K m increased, indicating that all these three compounds were competitive inhibitors for a-glucosidase. The K i values of 43, 40, and 34 were 10 mM, 52 mM, and 150 mM, respectively.

Docking study
In order to clarify the interactions between compounds and amino acids in the substrate-binding pocket of a-glucosidase at the molecular level, a molecular docking study was carried out using Autodock Vina 21 . Since the X-ray crystallographic structure of Saccharomyces cerevisiae a-glucosidase we used in the experiments has not been reported yet, the 3 D structure of a-glucosidase was conducted with SWISS-MODEL 22 .
Acarbose and the most potent compounds 43, 40, and 34 were docked in the active site of the a-glucosidase. In order to explore the structure-activity relationship, compound 41 was also docked. Table 2 showed the results of the molecular docking and detailed interactions, including hydrogen bonds, p-p stacking interactions, hydrophobic interactions, and electrostatic interactions. From the docking study, it was observed that acarbose ( Figure 4(A)) interacted with the active site of a-glucosidase via six hydrogen bonds with residues Gln350, Arg312, and Asn241. Additionally, the compound formed several electrostatic interactions with residues Phe157, Phe158, and Phe300.   The most active compound 43 was well accommodated inside the active site of a-glucosidase (Figure 3(A)) and established nine hydrogen bonds with residues Glu276, Asp349, Arg439, His279, Glu304, Pro309, and Arg312 (Figure 3(B)). Additionally, the phenyl rings of the compound formed two p-p stacking interactions with Phe300 and Phe157. Furthermore, hydrophobic interactions and electrostatic interaction were observed between compound 43 and residues Leu218, Ala 278, Phe177, Phe158, and Asp349. The compound 43 has lower binding energy (À9.2 kcal/mol) than acarbose (À7.8 kcal/mol), suggesting that compound 43 was binding with enzyme more easily and strongly than acarbose.
In order to explore how the length of the alkyl chain affects the activity on a-glucosidase, compound 41(n ¼ 2), which belongs to the same group as the compounds 43(n ¼ 4) and 40 (n ¼ 1), was also docked. As seen in Figure 4(C), except for the number of hydrogen bonds, other interactions were nearly the same as that for compounds 43 and 40. According to the IC 50 values, structures, and docking study results, we can infer that the length of  the alkyl chain of the compounds affected their a-glucosidase inhibitory activity by affecting the number of hydrogen bonds between the compounds and the enzyme.
In the case of compound 34, the third most active compound, which formed nine hydrogen bonds with the binding site residues His348, Asp349, Asp214, His279, Glu304, Asn241, His239 and also one p-p interaction with residues Phe157. In addition, several hydrophobic interactions were also observed between molecules and residues Phe158, Phe177, Tyr71, Phe300, and Leu437 ( Figure  4(B)). Compared with compound 43, they both have nine hydrogen bonds, while the most significant difference between them was the number of p-p stacking interactions, the number of p-p stacking interactions for compound 34 was less than that for compound 43, suggesting that p-p stacking interaction played a significant effect on the inhibitory activity of the compounds.
Studies of the biological activity and molecular docking of these compounds showed that the inhibitory activities were highly dependent on the length of the alkyl chain, but without a trend correlation. Besides, the more phenyl rings in the molecules, the more probabilities to establish p-p stacking interactions between molecules and enzymes, which were responsible for high activity on a-glucosidase. The compounds without a methoxy group in the 3 0 -position of phenyl ring were more active than that with a methoxy group. The docking results were similar with previous studies. Zeng et al. synthesised a series of N-benzyl-deoxynojirimycin derivatives, and the most active compound also established a p-p stacking interactions between molecules and enzymes, which gave the most active compound a strong inhibitory activity to a-glucosidase 23 . Besides, the hydrogen bonds between compounds a-glucosidase and were also played important roles in high activity on a-glucosidase. Shahzad et al. synthesised a series of symmetrical salicylaldehyde-bishydrazine azo molecules. The high activity of compounds is mainly caused by hydrogen bonds 24 , which were the same with us.

Conclusions
In conclusion, we synthesised a series of novel N-alkyl-1-DNJ derivatives 25-44, all the compounds were tested for their a-glucosidase inhibitory activity. Among them, the compound 43 (IC 50 ¼ 30.0 ± 0.60 mM) was the most active compound, which was $27-fold more active than acarbose (IC 50 ¼ 822.0 ± 1.5 mM) and $7-fold more active than 1-DNJ (IC 50 ¼ 222.4 ± 0.5 mM). The kinetic study revealed that compounds 43, 40, and 34 inhibit a-glucosidase via a competitive mechanism with K i of 10 mM, 52 mM, and 150 mM, respectively. The docking study showed that hydrogen bond and p-p stacking interaction played a significant role in the anti-a-glucosidase activity of the synthesised compounds. The numbers of hydrogen bonds and p-p stacking interactions were correlated with and responsible for the compounds' activities, and the compounds without methoxy group in the 3 0 -position of phenyl ring were more active than that with a methoxy group.

Experimental
All starting materials and reagents were purchased from commercial suppliers. a-glucosidase (EC 3.2.1.20) was purchased from Sigma-Aldrich. TLC was performed on Silica gel F-254. Melting points were measured on a microscopic melting point apparatus. The 1 H NMR and 13 C NMR were measured (DMSO solution) with Bruker spectrometer (500 MHz 1 H, 125 MHz 13 C). HRMS was performed on AB SCIEX Triple TOF 5600þ with electron spray ionisation (ESI) as the ion source.
5.1. General experimental procedure for the syntheses of intermediates 5-9 A solution of cinnamic acid 1 (1 mmol), Et 3 N(3 mmol), dibromo alkane (4 $ 5 mmol) in acetone was heated at 65 C, overnight. After the reaction completed, the mixture was cooled down to room temperature. Water and ethyl acetate were added and extracted three times. The combined organic extracts were dried over Na 2 SO 4 and then concentrated. Further purification by flash chromatography gave the title compounds.

General experimental procedure for the syntheses of intermediates 10-24
A solution of methyl 4-hydroxycinnamate 2 (1 mmol), K 2 CO 3 (2 mmol), dibromo alkane (4 $ 5 mmol) in acetone was heated at 65 C, overnight. After the reaction completed, the mixture was cooled down to room temperature. Water and ethyl acetate were added and extracted three times. The combined organic extracts were dried over Na 2 SO 4 and then concentrated. Further purification by flash chromatography gave the compounds 10-14.

In vitro assay of a-glucosidase inhibitory activity
The method we used was reported before in our team 23 , a-glucosidase inhibitory activity was measured by using 0.1 mM phosphate buffer (pH 6.8) at 37 C. The a-glucosidase enzyme (EC 3.2.1.20, 1 U/ ml, 10 mL) in phosphate buffer was incubated with various concentrations of tested compounds (dissolved in 1% DMSO) at 37 C for 20 min, then PNPG (10 mM, 20 mL) was added to the mixture as substrate. Finally, the absorbance was measured at 405 nm by using a spectrophotometer. The sample solution was replaced by DMSO as the control. Acarbose and 1-DNJ were used as standard drugs. The inhibition has been obtained using the formula: Inhibition ð%Þ ¼ ðDA control À DA sample Þ=DA control Â 100: The IC 50 fitted with SPSS:

Molecular docking
Since the X-ray crystallographic structure of Saccharomyces cerevisiae a-glucosidase we used in the experiments has not been reported yet, the 3 D structural modelling of a-glucosidase was conducted with SWISS-MODEL 22 . The sequence in the FASTA format of a-glucosidase was download from NCBI. Isomaltase from Saccharomyces cerevisiae(PDB code 3AJ7) shows high sequence similarity (72.51%) with a-glucosidase, which structure was selected as the template for homology modelling, and the quality of the obtained homology model was verified using PROCHECK 25 . The 3 D structures of acarbose and synthesised compounds were built by ChemBioDraw Ultra and ChemBio3D Ultra software. The AutoDock Tool 1.5.6 package was employed to generate the docking input files. Docking studies were performed using Autodock Vina 21 . The centre of the grid box 26 was placed at cen-tre_x ¼ 12.5825, centre_y ¼ À7.8955, centre_z ¼ 12.5190 with dimensions size_x ¼ 40, size_y ¼ 40, size_z ¼ 40. The best-scoring poses as judged by the Vina docking score were chosen and visually analysed using PyMOL 1.8.0 software (http://www.pymol.org/).

Disclosure statement
No potential conflict of interest was reported by the author(s).