Anti-α-Glucosidase, SAR Analysis, and Mechanism Investigation of Indolo[1,2-b]isoquinoline Derivatives

To find potential α-glucosidase inhibitors, indolo[1,2-b]isoquinoline derivatives (1–20) were screened for their α-glucosidase inhibitory effects. All derivatives presented potential α-glucosidase inhibitory effects with IC50 values of 3.44 ± 0.36~41.24 ± 0.26 μM compared to the positive control acarbose (IC50 value: 640.57 ± 5.13 μM). In particular, compound 11 displayed the strongest anti-α-glucosidase activity, being ~186 times stronger than acarbose. Kinetic studies found that compounds 9, 11, 13, 18, and 19 were all reversible mix-type inhibitors. The 3D fluorescence spectra and CD spectra results revealed that the interaction between compounds 9, 11, 13, 18, and 19 and α-glucosidase changed the conformational changes of α-glucosidase. Molecular docking and molecular dynamics simulation results indicated the interaction between compounds and α-glucosidase. In addition, cell cytotoxicity and drug-like properties of compound 11 were also investigated.


Introduction
Diabetes mellitus, one of the most common endocrine and metabolic diseases, is induced by insulin resistance or abnormality [1]. Diabetes mellitus patients have abnormally elevated blood glucose, and long-term high blood glucose levels can harm vascular endothelium, which can then lead to various complications. Using hypoglycemic drugs to regulate postprandial hyperglycemia is an important strategy for type 2 diabetes patients [2][3][4].
α-Glucosidase, a crucial membrane-bound enzyme in the small intestine, is one of the most important therapeutic targets for diabetes [5][6][7]. It is responsible for the hydrolysis of glycosidic linkage bonds of carbohydrates, especially disaccharides and polysaccharides, thereby releasing absorbable monosaccharides, which are absorbed to cause postprandial hyperglycemia [8,9]. Therefore, the inhibition of α-glucosidase activity can delay carbohydrate ingestion and reduce postprandial hyperglycemia [10]. Up to now, increasingly more α-glucosidase inhibitors have been developed, but only a few inhibitors are clinically used for the treatment of type 2 diabetes, such as acarbose, voglibose, and miglitol [11][12][13]. Moreover, long-term use of these drugs also leads to some gastrointestinal side effects, including diarrhea and flatulence [14,15]. This situation has encouraged the authors to develop more efficient and safer α-glucosidase inhibitors [16][17][18].

α-Glucosidase Activity Evaluation
Because commercial human α-glucosidase is still lacking, the anti-α-glucosidase activity of indolo[1,2-b]isoquinoline derivatives (1-20) was first evaluated using α-glucosidase from Saccharomyces cerevisiae, due to its similar active region structure to human αglucosidase. Acarbose was selected as the positive control. The 50% inhibition concentration (IC50) of all derivatives (1-20) was obtained and is illustrated in

Inhibition Kinetics Study
Compounds 9, 11, 13, 18, and 19 with different α-glucosidase inhibitory activity were selected as representative compounds to reveal the inhibition mechanism of indolo[1,2b]isoquinoline derivatives (1-21) on α-glucosidase by the implementation of an inhibition kinetics study. For enzyme kinetics (Figure 5a,c,e,g,i), the plots of remaining enzyme activity versus enzyme concentration treated with compounds 9, 11, 13, 18, and 19 all passed through the origin point, which indicated that the inhibition of compounds 9, 11, 13, 18, and 19 was reversible, respectively. For substrate kinetics (Figure 5b,d,f,h,j), the Lineweaver-Burk plots of remaining enzyme activity versus enzyme concentration treated with compounds 9, 11, 13, 18, and 19 intersected at one point in the second quadrant, respectively. The inhibition kinetics study was an effective method by which to understand the inhibition mechanism of the inhibitor against the target protein. In a substrate kinetics study, different inhibition mechanisms of inhibitors manifested as different intersection positions of Lineweaver-Burk plots. Further research confirmed that the intersection position of the Lineweaver-Burk plots was located in four quadrants, meaning mixed-type inhibition [31]. Therefore, our results declared a mixed-type inhibition of compounds 9, 11, 13, 18, and 19, respectively. Moreover, compounds 9, 11, 13, 18, and 19 bind with both free enzyme and enzyme-substrate complexes to inhibit α-glucosidase. The inhibition kinetics constants of compounds 9, 11, 13, 18, and 19 were obtained and are listed in Table 2, including K i , Kis, Km, and Vmax values, which might be helpful to understand the inhibition mechanism.

Inhibition Kinetics Study
Compounds 9, 11, 13, 18, and 19 with different α-glucosidase inhibitory activity were selected as representative compounds to reveal the inhibition mechanism of indolo[1,2b]isoquinoline derivatives (1-21) on α-glucosidase by the implementation of an inhibition kinetics study. For enzyme kinetics (Figure 5a,c,e,g,i), the plots of remaining enzyme activity versus enzyme concentration treated with compounds 9, 11, 13, 18, and 19 all passed through the origin point, which indicated that the inhibition of compounds 9, 11, 13, 18, and 19 was reversible, respectively. For substrate kinetics (Figure 5b,d,f,h,j), the Lineweaver-Burk plots of remaining enzyme activity versus enzyme concentration treated with compounds 9, 11, 13, 18, and 19 intersected at one point in the second quadrant, respectively. The inhibition kinetics study was an effective method by which to understand the inhibition mechanism of the inhibitor against the target protein. In a substrate kinetics study, different inhibition mechanisms of inhibitors manifested as different intersection positions of Lineweaver-Burk plots. Further research confirmed that the intersection position of the Lineweaver-Burk plots was located in four quadrants, meaning mixedtype inhibition [31]. Therefore, our results declared a mixed-type inhibition of compounds 9, 11, 13, 18, and 19, respectively. Moreover, compounds 9, 11, 13, 18, and 19 bind with both free enzyme and enzyme-substrate complexes to inhibit α-glucosidase. The inhibition kinetics constants of compounds 9, 11, 13, 18, and 19 were obtained and are listed in Table 2, including Ki, Kis, Km, and Vmax values, which might be helpful to understand the inhibition mechanism.

3D Fluorescence Spectra Assay
The 3D fluorescence spectra of α-glucosidase with compounds 9, 11, 13, 18, and 19 was investigated to analyze the effect of compounds on the structure of α-glucosidase, respectively. Two important characteristic peaks appeared in the 3D fluorescence spectra of α-glucosidase (Figure 6a,c,e,g,i), including Peak 1 (λ ex = 335 nm, λ em = 230 nm), corresponding to the main chain structure of the polypeptide, and Peak 2 (λ ex = 335 nm, λ em = 277.5 nm), corresponding to tyrosine and tryptophan residues. While treatment with compounds 9, 11, 13, 18, and 19 could decrease the fluorescence intensity of Peak 1 and Peak 2, respectively. (Figure 6b,d,f,h,j). Previously, 3D fluorescence spectra have been used to determine protein conformation changes. Our results indicated that the interaction of inhibitors with α-glucosidase changed the microenvironment and structure of α-glucosidase, consistent with previous research [32], which also indicated that the interaction of the inhibitor with α-glucosidase reduced the intensity of the 3D fluorescence spectra characteristic peaks of α-glucosidase.

3D Fluorescence Spectra Assay
The 3D fluorescence spectra of α-glucosidase with compounds 9, 11, 13, 18, and 19 was investigated to analyze the effect of compounds on the structure of α-glucosidase, respectively. Two important characteristic peaks appeared in the 3D fluorescence spectra of α-glucosidase (Figure 6a, c, e, g, i), including Peak 1 ( ex = 335 nm, em = 230 nm), corresponding to the main chain structure of the polypeptide, and Peak 2 ( ex = 335 nm, em = 277.5 nm), corresponding to tyrosine and tryptophan residues. While treatment with compounds 9, 11, 13, 18, and 19 could decrease the fluorescence intensity of Peak 1 and Peak 2, respectively. (Figure 6b, d, f, h, j). Previously, 3D fluorescence spectra have been used to determine protein conformation changes. Our results indicated that the interaction of inhibitors with α-glucosidase changed the microenvironment and structure of α-glucosidase, consistent with previous research [32], which also indicated that the interaction of the inhibitor with α-glucosidase reduced the intensity of the 3D fluorescence spectra characteristic peaks of α-glucosidase.

CD Spectra Assay
CD spectra were also monitored to study the effect of compounds 9, 11, 13, 18, and 19 on the conformational changes of α-glucosidase. As shown in Figure 7a-e, α-glucosidase presented two negative CD bands in the region of 190~280 nm, which was owed to the electronic transitions of n→π* of α-helical bonds. Treatment of compound 11 resulted in a concentration-dependent increase in the CD band intensity of α-glucosidase, while treatment of compounds 9, 13, 18, and 19 led to a decrease in CD band intensity. For specific secondary structure change, treatment with compound 11 (molar ratio: 3:1) reduced α-helix (from 8.50 to 7.80%), β-sheet (from 34.10 to 35.20%), and β-turn (from 19.80 to 20.10%), and increased random coils (from 35.50 to 35.80%), respectively (Table 3). While treatment of compounds 9, 13, 18, and 19 led to an increase of α-helix and a reduction of β-sheet, β-turn, and random coils, respectively (Table 3). CD spectra have been an important method by which to study protein structural changes and secondary structural content. Previous research has shown that the interaction of an inhibitor with α-glucosidase caused the partial folding and loosening of the α-glucosidase structure [33]. Our results also revealed that CD bands of α-glucosidase could be changed by the addition of compounds 9, 11, 13, 18, and 19, suggesting an effective interaction between compounds 9, 11, 13, 18, and 19 with α-glucosidase, respectively.

CD Spectra Assay
CD spectra were also monitored to study the effect of compounds 9, 11, 13, 18, and 19 on the conformational changes of α-glucosidase. As shown in Figure 7a-e, α-glucosidase presented two negative CD bands in the region of 190~280 nm, which was owed to the electronic transitions of n→π* of α-helical bonds. Treatment of compound 11 resulted in a concentration-dependent increase in the CD band intensity of α-glucosidase, while treatment of compounds 9, 13, 18, and 19 led to a decrease in CD band intensity. For specific secondary structure change, treatment with compound 11 (molar ratio: 3:1) reduced α-helix (from 8.50 to 7.80%), β-sheet (from 34.10 to 35.20%), and β-turn (from 19.80 to 20.10%), and increased random coils (from 35.50 to 35.80%), respectively (Table 3). While treatment of compounds 9, 13, 18, and 19 led to an increase of α-helix and a reduction of β-sheet, β-turn, and random coils, respectively (Table 3). CD spectra have been an important method by which to study protein structural changes and secondary structural content. Previous research has shown that the interaction of an inhibitor with α-glucosidase caused the partial folding and loosening of the α-glucosidase structure [33]. Our results also revealed that CD bands of α-glucosidase could be changed by the addition of compounds 9, 11, 13, 18, and 19, suggesting an effective interaction between compounds 9, 11, 13, 18, and 19 with α-glucosidase, respectively.
Compound 18 formed a hydrogen bond with Arg312 (2.3 Å) and one π-π bond with Phe300 (3.8 Å), respectively (Figure 8h). Compound 19 also formed a hydrogen bond with Arg312 (2.3 Å) and one π-π bond with Phe300 (4.1 Å), respectively (Figure 8i). The different binding interactions between compounds 9, 11, 13, 18, and 19, with α-glucosidase would be helpful to explain the different inhibitory activites. To better validate the binding between compounds and α-glucosidase, human αglucosidase was used as a target protein to simulate the docking; the results are shown in Figure 9. Because the 3D structure of human α-glucosidase has still not been characterized, a homologous model of human α-glucosidase was also built based on the existing enzyme. Although compounds 9, 11, 13, 18, 19, and acarbose were embedded in the active pocket of α-glucosidase, it was observed that compounds 9, 19, and acarbose were located in the interior of the active pocket, and compounds 11, 13, and 18 were located in the exterior of the active pocket (Figure 9a (Figure 9i). It was observed that there were many differences between the two docking results, especially the amino acid residues in the active pocket, which formed hydrogen or π-π bonds. To better validate the binding between compounds and α-glucosidase, human α-glucosidase was used as a target protein to simulate the docking; the results are shown in Figure 9. Because the 3D structure of human α-glucosidase has still not been characterized, a homologous model of human α-glucosidase was also built based on the existing enzyme. Although compounds 9, 11, 13, 18, 19, and acarbose were embedded in the active pocket of α-glucosidase, it was observed that compounds 9, 19, and acarbose were located in the interior of the active pocket, and compounds 11, 13, and 18 were located in the  (Figure 9e). Inhibitor 11 formed a hydrogen bond with Asn236 (2.0 Å) and a π-π bond with Phe238 (3.6 Å), respectively (Figure 9f). Compound 13 only formed hydrophobic bonds (Figure 9g). Compound 18 formed two hydrogen bonds with Asn236 (2.1 Å) and Asp260 (2.3 Å), and a π-π bond with Phe238 (3.6 Å), respectively (Figure 9h). Compound 19 formed two hydrogen bonds with Arg44 (1.8 and 2.7 Å) (Figure 9i). It was observed that there were many differences between the two docking results, especially the amino acid residues in the active pocket, which formed hydrogen or π-π bonds.

Molecular Dynamics Simulation
To analyze the contribution of amino acid residues in the active pocket to substrate binding, the docking results of compounds 9, 11, 13, 18, and 19 in the complexes with α-glucosidase were further simulated using molecular dynamics simulation for 100 ns. The root-mean-square deviation (RMSD) results are presented in Figure 10a-e, illustrating the equilibration of the systems. The calculated RMSD values confirm that these systems reached a state of structural equilibrium. Specifically, it was observed that the compound 11α-glucosidase systems achieved stability after approximately 45 ns. The RMSD values for the free α-glucosidase and compound 11-α-glucosidase were measured at 2.7 Å and 1.7 Å, respectively. Moreover, the overall RMSD fluctuations of the compound 11-α-glucosidase remained within the range of 1-1.7 Å, indicating the backbone stability of α-glucosidase during the docking process. The RMSF value was also observed to characterize local changes in the protein chain (Figure 10f-j). The overall RMSF value indicated few fluctuations in the N-and C-terminal loop regions. Subsequently, the binding free energies of the complexes were calculated using the molecular mechanics−generalized Born surface area (MM-GBSA) method (Table 4). The total binding free energies were determined as follows: −47.23 kcal/mol for compound 9, −66.94 kcal/mol for compound 11, −60.27 kcal/mol for compound 13, −52.87 kcal/mol for compound 18, and −51.49 kcal/mol for compound 19 in their respective complexes with α-glucosidase. The results obtained from the molecular dynamics simulations are in agreement with the experimental observations. Notably, the contributions that favored ligand binding included van der Waals energy, electrostatic interaction energy, and nonpolar solvation interaction. Conversely, the polar solvation interaction had a detrimental effect on the binding with the targets. Given that these compounds consist of the indolo[1,2-b]isoquinoline scaffold, they were able to establish hydrophobic interactions with α-glucosidase. Therefore, van der Waals and nonpolar solvation energies emerged as the two crucial components of the overall binding free energy.   ∆G bind (free energy of binding), ∆E VDW (van der Waals energy), ∆E ele (electrostatic energy), ∆E GB (polar solvation energy), ∆E GA (nonpolar solvation energy).

In Vitro Cytotoxicity
The in vitro cytotoxicity of compounds 9, 11, 13, 18, and 19 against hepatocytes LO2 cells was detected using the MTT method. From Figure 11, compounds 9, 11, 13, 18, and 19 had no significant effect on cell viability up to a concentration of 32 µM, suggesting the safety of these compounds at low concentrations.

D Fluorescence Spectra
Compounds 9, 11, 13, 18, and 19 were added to α-glucosidase (5 µM) in 100 µL of PBS and incubated for 5 min, then the 3D fluorescence spectra of the mixture were recorded [32] . The excitation and emission wavelengths were 200-500 nm and the slit width was 2.5 nm. All data were imported into Matlab for processing.

CD Spectra
Compounds 9, 11, 13, 18, and 19 with different concentrations were added into αglucosidase solution (31 µM) and incubated for 5 min, respectively, then the CD spectra of the mixture were recorded [33]. CDNN was used to analyze the ratio of protein secondary conformations.

Molecular Docking
Molecular docking between compounds 9, 11, 13, 18, and 19 and α-glucosidase was conducted using SYBYL software [41][42][43]. The homology model of α-glucosidase had been built in our previous works. Compounds 9, 11, 13, 18, and 19 were built and treated with energy minimization. α-Glucosidase was also optimized using the internal program, followed by the generation of the active pocket. Molecular docking was conducted in the default format.
For the homology model of human α-glucosidase, the sequence in FASTA format was obtained from UniProt (ID O33830), and alpha-glucosidase A (ID: 1OBB) was selected as the template. The human α-glucosidase homology models were constructed using Modeler 10.1 software (http://salilab.org/modeller/, accessed on 20 June 2023). Then, their qualities were verified by using a Ramachandran plot (http://services.mbi.ucla.edu/PROCHECK/, accessed on 20 June 2023). The optimal homology mode with a Phi (degrees) of 93.3% ( Figure 12) was selected for subsequent docking.
width was 2.5 nm. All data were imported into Matlab for processing.

CD Spectra
Compounds 9, 11, 13, 18, and 19 with different concentrations were added into αglucosidase solution (31 µM) and incubated for 5 min, respectively, then the CD spectra of the mixture were recorded [33]. CDNN was used to analyze the ratio of protein secondary conformations.

Molecular Docking
Molecular docking between compounds 9, 11, 13, 18, and 19 and α-glucosidase was conducted using SYBYL software [41][42][43]. The homology model of α-glucosidase had been built in our previous works. Compounds 9, 11, 13, 18, and 19 were built and treated with energy minimization. α-Glucosidase was also optimized using the internal program, followed by the generation of the active pocket. Molecular docking was conducted in the default format.
For the homology model of human α-glucosidase, the sequence in FASTA format was obtained from UniProt (ID O33830), and alpha-glucosidase A (ID: 1OBB) was selected as the template. The human α-glucosidase homology models were constructed using Modeler 10.1 software (http://salilab.org/modeller/, accessed on 20 June 2023). Then, their qualities were verified by using a Ramachandran plot (http://services.mbi.ucla.edu/PROCHECK/, accessed on 20 June 2023). The optimal homology mode with a Phi (degrees) of 93.3% (Figure 12) was selected for subsequent docking.

Molecular Dynamics Simulation
A molecular dynamics simulation was carried out to analyze the protein backbone stability in docking compounds 9, 11, 13, 18, and 19 using the Desmond simulation. The complex was treated with internal procedures, including filling water, cleaning total charge, and minimizing energy. Then, the dynamics simulation was run with a simulation length of 100 ns and a relaxation time of 1 ps.

MTT Assay
The in vitro cytotoxicity of compound 11 against hepatocytes LO2 cells was assayed using the MTT method [44,45]. LO2 cells were cultured in DMEM containing 10% FBS, 100 IU/mL penicillin, and 100 IU/mL streptomycin at 37 • C under 5% CO2. An amount of 100 µL of LO2 cells was seeded into a 96-well plate (5000 per well) for 24 h, and was then treated with compounds 9, 11, 13, 18, and 19 for 24 h, respectively. Then, MTT solution was added into each well and incubated for 4 h. An amount of 100 mL of DMSO was used to dissolve the obtained crystallization. Then, this absorbance was determined at 490 nm. Each sample was performed in triplicate.