Synthesis of alkoxy-isoflavones as potential α-glucosidase inhibitors

The aim of the present study was to synthesize isoflavone-enaminones 3a-c and 7-alkoxy-isoflavones 4a-c, evaluate their inhibition of α-glucosidase, and analyze the bioisosteric effect of the presence versus absence of aromatic moieties in these benzopyran derivatives. All the test compounds exhibited greater inhibition of α-glucosidase than the positive control acarbose. The series of isoflavones 3a-c and 4a-c showed higher inhibitory activity (IC50 = 6.3–87.6 µM) than the parental 7-hydroxyisoflavones 2a-c (IC50 = 109.4–173.2 µM), suggesting that the attachment of a 4’-chloroacetophenone moiety to the 7-hydroxyl group of 2a-c is an efficient way to increase the inhibition of α-glucosidase. Furthermore, the aromatic moieties of the series of compounds 3 and 4 enhance inhibitory activity by hydrophobic effects, according to docking calculations. Graphical abstract Graphical abstract

Introduction α-Glucosidase is an enzyme that catalyzes the hydrolysis of glycosidic bonds from the non-reducing portion of the oligosaccharide substrates. It plays a key role in the digestion of dietary carbohydrates and the processing of glycoproteins. By inhibiting this enzyme, the digestion of starch and other dietary sugars decreases, which helps to avoid hyperglycemia by maintaining blood glucose at normal levels [1]. The reduction of sugar digestion makes α-glucosidase an attractive therapeutic target for the treatment of type 2 diabetes [2], cardiovascular diseases [3], and cancer [4,5]. Regarding diabetes, about 422 million people worldwide are affected by this metabolic disease, characterized by high levels of blood glucose (hyperglycemia) derived from a deficiency in the secretion and/or action of insulin [6]. Therefore, α-glucosidase inhibitors may represent an effective therapeutic strategy for reducing postprandial hyperglycemia.
Recently, our group reported that enaminone-chromone derivatives can potentially act as α-glucosidase inhibitors [15]. Several interactions were found between the chromone derivatives and α-glucosidase, including π-stacking and hydrophobic effects. Such interactions are key for enhancing inhibitory activity. For instance, enaminone-chromone 1c bears a phenyl group at the C-3 position of the chromone ring system (isoflavone skeleton) and a 4-chlorophenyl group at the enaminone moiety. It displays considerably greater inhibition of α-glucosidase than compounds 1a-b (Fig. 1A).
In accordance with the aforementioned, the 7-hydroxy group and its 7-alkoxylated isoflavone derivatives as well as 6-methoxy-containing or 6-methoxy-free analogs may generate a significant inhibition of α-glucosidase. This largely depends on the nature of the C-7 alkoxy groups and the C-6 substituents in such compounds. For the choice of compounds to be tested presently, it was taken into account that functionalized benzopyrans with a 2-(4-chlorophenyl)-2-oxoethoxy group at the C-7 position (i.e., compound 1c) increased the inhibition of the enzyme, focusing particular attention on the role played by the enaminone moiety. Hence, the aim of the current contribution was to synthesize a series of enaminone-isoflavones 3a-c and 7-(2-(4-chlorophenyl)-2-oxoethoxy)isoflavones 4a-c, evaluate them as potential inhibitors of α-glucosidase (Fig. 1B), and test them for their antioxidant effect. The analysis of the results on the inhibition of α-glucosidase provided insights into the structure-activity relationship of the isoflavone-based derivatives. Additionally, docking studies were carried out to better understand the mechanism of binding of the 7alkoxy-isoflavone derivatives to the crystal structure of isomaltase, an α-glucosidase of Saccharomyces cerevisiae.

Results and discussion
Chemistry Compounds 7a-c were prepared in good yields by the acylation of resorcinols 5a-c with 2-(4-methoxyphenyl) acetic acid (6), in the presence of boron trifluoride On the other hand, the treatment of 2,4-dihydroxyacetophenones 7a-c with DMFDMA gave the series of isoflavones 2a-c in high yields, along with 7-methoxyisoflavones 10a-b as by-products in low yields [32]. Then O-alkylation of isoflavones 2a-c with 8 provided the target compounds 4a-c (Scheme 2).
The structure of all test compounds was elucidated by 1D and 2D NMR techniques and HRMS. The intramolecular cyclization and condensation of the α-arylacetophenones 9a-c led to the enaminone-isoflavones 3a-c as single stereoisomers. The Z geometry of the latter compounds was established by NOE experiments. The irradiation of the signal assigned to the methyl protons of the dimethylamino group produced an enhancement of the signal corresponding to the H-6 (for 3a) and H-8 protons of the benzene ring of the chromen-4-one scaffold. This stereoselectivity has also been observed in similar systems, which is probably due to the higher stability achieved by the planar π-conjugated acrylate system when the bulky dimethylamino group is located at the opposite side of the carbonyl group [15].
Compared to their precursors 7a-c, compounds 2a-c (with a benzopyran ring moiety) showed greater inhibition of α-glucosidase. In order to explore the possibility that the enaminone moiety in 3a-c favors inhibition, derivatives 4ac (without this fragment) were evaluated. The latter compounds exhibited better inhibitory activity than compounds 3a and 3b. The best inhibition was found with 4a, the IC 50 value (6.3 µM) of which indicated a 49-fold greater inhibition of α-glucosidase than 12.
Regarding the inhibitory effect of compounds 3a-b and 4a-b, the analysis of the structure-activity relationship revealed that the polar enaminone group is not as relevant as the (4-chlorophenyl)-2-oxoethoxy group. In contrast, the enaminone group and the chlorine atoms in 3c favored improved inhibition with respect to 4c. Furthermore, the Scheme 1 Synthesis of enaminone-isoflavones 3a-c π-stacking or hydrophobic effect of the additional aromatic rings at C-3 and C-7 of the isoflavone ring (series 3 and 4) may have played a more important role than the H-bond donor effect of 7-OH (series 2) in the interaction with the enzyme. At the C-6 position, the absence of a substituent in the isoflavone and benzene rings in 3a, 4a, and 9a led to better inhibitory activity than the presence of a methoxy group in 3b, 4b, and 9b or a chlorine atom in 4c and 9c.

Evaluation of the antioxidant activity
Antioxidants are responsible for protecting the body against oxidative stress and many disorders (e.g., diabetes). To determine the potential of the presently synthesized compounds as antioxidants, their free radical scavenging was examined by the DPPH radical assay, with butylated hydroxytoluene as the positive control. The reduction in color intensity represented the percentage of inhibition of DPPH (Table 1), finding that the 2-acylphenols 7b-c, αarylacetophenones 9a-c, and isoflavones 2a-c, 3a-c, 4a-c, and 10a-b did not have any significant antiradical activity up to the maximum concentration tested. Only compound 7a showed some inhibition (61% at 2.5 mM), suggesting that this effect is mainly related to the hydroxyl groups substituted in the benzene ring.

Docking results
To identify the binding mode of the compounds with the greatest inhibition of α-glucosidase, molecular docking studies were carried out. The crystal structure of isomaltase from S. cerevisiae served as the target protein because it is a type of α-glucosidase. The positive controls were maltose (11) and acarbose (12), two competitive inhibitors of isomaltase. The results of the interactions are illustrated in 2D and 3D (Fig. 2). Compounds 2a, 3a, 4a, and 9c as well as the controls 11 and 12 recognized many of the amino acid residues at the active site. Among such residues were Asp64, His112, Arg213, Asp215, Glu277, His351, Asp352, and Arg442, as reported in other studies on isomaltose inhibitors [33][34][35][36]. For all the compounds, the two residues Scheme 2 Synthesis of 7arylalkoxy-isoflavones 4a-c nt not tested a IC 50 = 0.84 ± 0.08 mM Fig. 2 Portrayal of the interactions within the active site of isomaltase by isoflavone derivatives 2a, 3a, and 4a, aryloxycarbonyl compound 9c, and controls 11 and 12. The 3D models illustrate the hydrogen bonds and the amino acid residues at the active site of the enzyme. The non-hydrogen bonds were omitted for clarity. In the 2D model, the following bonds are displayed with dotted lines: conventional hydrogen (green), carbon-hydrogen (brown), π-sigma (purple), π-π T-shaped (fuchsia), π-alkyl (pink), and halogens (cyan). The amino acids are depicted with circles: basic (pink), acid (orange), polar (dark blue), and non-polar (yellow) involved in catalysis (Glu277 and Asp352) are active in the binding process, suggesting that the derivatives of 7-alkoxyand 7-hydroxyisoflavones bind to the active site of the enzyme to inhibit α-glucosidase [33].
The binding energy values (ΔG) and the type of interactions are shown in Table 2. The best ΔG values were observed for 2a (−7.51 Kcal/mol), 3a (−9.82 Kcal/mol), 4a (−9.58 Kcal/mol), and 9c (−9.83 Kcal/mol), each being better than the corresponding value for maltose (11) (−5.55 Kcal/mol), a competitive inhibitor. The best binding to the active site of isomaltase was found with derivative 9c. There were hydrophilic and hydrophobic interactions in isoflavones 2a, 3a, 4a, and 9c, but not in the two controls. The predominant hydrophilic interactions in all the derivatives and maltose (11) are those between some substituent of the compounds and the Asp352 or Arg442 residues. There is a hydrophobic interaction with Phe178 for these four derivatives, as reported previously [34][35][36].
The results of the docking study coincide with those of the enzyme inhibition assays. That is, among the four aforementioned compounds with the binding energy, 3a, 4a, and 9c produced the greatest inhibition of the α-glucosidase enzyme. The corresponding IC 50 values were 18.8 ± 0.13, 6.3 ± 0.11, and 34.1 ± 0.013 µM, respectively, much lower than the IC 50 of 12 (309.2 ± 0.9 µM). The analysis of the docking results for compounds 3a, 4a, and 9c showed that the 4-methoxyphenyl group at C-3 of the benzopyran ring (C-1 for derivative 9c) exhibits a hydrophilic interaction with a hydrogen atom of group methoxy between carboxamide group of Asn415. The hydrophobic interactions π-π stacked and π-alkyl were observed with Arg315, Phe314, and Lys156. The presence Table 2 Docking results of isoflavones 2a, 3a, and 4a, aryloxycarbonyl compound 9c, maltose (11), and acarbose (12) Asp69, Tyr72, His112, Tyr158, Phe159, Phe178, Arg213, Asp215, Val216, Glu277, Gln279, His 280, Phe303, Asp307, Arg315, Tyr316, His351, Asp352, Gln353, Glu411, Arg442, Arg446 Halogen bonding Asp69 Halogen bonding Asp69 of the (4-chlorophenyl)-2-oxoethoxy fragment at C-7 of the isoflavone ring presents hydrophobic interactions with the active site of α-glucosidase such as π-sigma interactions with Val216, π-alkyl with Tyr72, His112 and Phe178, and halogen bonding with Asp69. Hydrophilic interactions of conventional hydrogen bond type and carbon-hydrogen bond were observed with Gln279 with Asp352, respectively. Considering these findings, the presence of polar methoxy or hydroxy groups on the aromatic ring at C-3 of the benzopyran system could increase the affinity with α-glucosidase. Furthermore, analogs with a chlorine substituent at C-6 of the benzopyran ring increase their affinity for the enzyme. On the other hand, bioisosteric replacement of the (4-chlorophenyl)-2-oxoethoxy group by a (4-chlorophenyl)-2-oxoacetamide group at C-7 of the isoflavone ring could be explored, since that oxoacetamide moiety may show hydrophilic interactions that stabilize binding energies at the active site of the enzyme.
These results suggest that isoflavones 2a and 4a (without a substituent at C-6), as well as compounds 7a and 9a (without a substituent on the catechol ring), exhibited a greater inhibitory effect than their chlorine-substituted analogs (2c, 4c, 7c, and 9c) and methoxy-substituted analogs (2b, 4b, 7b, and 9b). In addition, compounds 2c and 3c with the chlorine substituent at C-6 of the benzopyran ring (C-5 for compound 9c) exhibited better enzyme inhibition than compounds 2b and 3b with the methoxy substituent (C-5 for compound 9b), which indicates that the negative inductive effect of the chlorine atom increases its affinity with the enzyme compared to the mesomeric effect of the methoxy group. Therefore, these scaffolds provide promise for the development of new drugs to treat diabetes mellitus.

Physicochemical properties
The physicochemical properties used to evaluate the druglikeness of the synthesized compounds are shown in Table 3. The octanol-water partition coefficient Log P (expressed as Log Po/w) is defined as the partition coefficient of an un-ionized compound in two immiscible phases (octanol and water/buffer) at equilibrium.
Lipophilicity is an important parameter in medicinal chemistry due to its correlation with molecular hydrophobicity, which affects the bioavailability, permeability, and metabolism of drugs as well as their toxicity [37,38]. Permeability is considered high with Log P values close to 5 and low with negative values. All compounds showed acceptable Log P values within a range of 2.17-4.95. For the series 3a-c, 4a-c, and 9a-c, the presence of the (4chlorophenyl)-2-oxoethoxy group and chlorine atoms increased the Log P values, particularly in 3c and 4c (Log P 4.95 and 4.69, respectively).
Log S is a descriptor of the octanol-water distribution coefficient for the partitioning of ionizable species in biphasic media. It reflects the true behavior of ionizable compounds in solution at a given pH value or range [39]. Compounds 2a-c, 3a-c, 4a-c, 7a-c, 9a-c, and 10a-b showed a slight to moderate solubility in water with their Log S ranging from −2.86 to −6.79. The controls, 11 and 12, exhibited values of 0.25 and 0.58, respectively (their polar groups form hydrogen bonds with the water).
The polar surface area (PSA) defines the sum of the surfaces of the polar atoms (N and O) and slightly polar atoms (S and P), and the hydrogen atoms attached to them. This parameter predicts intestinal absorption and the capacity to cross the blood-brain barrier [40]. A PSA below 90 Å 2 is indicative of the capacity of a drug to cross the blood-brain barrier. All of the present derivatives meet this requirement. Interestingly, the compounds with a methoxy group at the C-6 position of the benzopyran ring showed higher PSA values than those with a chlorine or hydrogen atom. For acarbose (12), the PSA value obtained indicates poor penetration of the central nervous system.
Finally, the bioavailability of these compounds was evaluated using Lipinski's rule of 5, which states that a  (4) <10 hydrogen bond acceptors [38]. As can be appreciated in Table 3, the synthesized compounds are in agreement with these guidelines (except 3b and 3c with a molecular weight >500 g/mol).

Conclusions
The series of isoflavones 2-4 were easily synthesized in good yields. In vitro assays showed a greater inhibition of α-glucosidase by these derivatives than the positive control acarbose (12). Furthermore, alkoxy-isoflavone derivatives 4a-b displayed better inhibition than enaminoneisoflavones 3a-b. Docking studies revealed that the πstacking and hydrophobic effects of the aromatic moieties at C-3 and C-7 of the chromone ring play a major role in the interaction with the active site of the enzyme. Compound 7a showed a low degree of antioxidant activity. Hence, the series of α-arylacetophenones and isoflavones herein prepared can be considered promising scaffolds for the development of new compounds with enhanced pharmacological activity.

General experimental procedures
Raw data measurements of melting points were taken on an Electrothermal apparatus and are uncorrected. NMR spectra of 1 H (300, 500, 600 or 750 MHz) and 13  General method for the synthesis of 2-4dihydroxyacetophenones (7a-c) Under a nitrogen atmosphere and at 0°C, BF 3 • OEt 2 (2.0 mol equiv.) was added to a solution of phenols 5a-c (1.0 mol equiv.) and 2-(4-methoxyphenyl)acetic acid (6) (1.2 mol equiv.), and the resulting mixture was stirred at 80°C for 3 h. The reaction crude was poured into cold water (10 ml) and adjusted to neutral pH with an aqueous saturated solution of NaHCO 3 (10%), followed by extraction with EtOAc (3 × 30 ml). The organic layer was dried (Na 2 SO 4 ) and concentrated under vacuum. The residue was purified by flash chromatography over silica gel (hexane/EtOAc, 8:2 or 98:2) to provide the corresponding products 7a-c.

Biological evaluation
Inhibition of α-glucosidase α-Glucosidase inhibition was determined according to a method described by Salehi et al. with some modifications [47]. For this assay, α-glucosidase from S. cerevisiae was employed. A reaction was prepared by mixing 20 µl of αglucosidase solution (0.5 unit/ml), 120 µl of 0.1 M phosphate buffer (pH 6.9), and 10 µl of the sample at several concentrations. The solution was incubated in a 96-well microplate at 37°C for 15 min. Subsequently, the enzymatic reaction was initiated by adding 20 µl of 5 mM PNP solution in 0.1 M phosphate buffer (pH 6.9), followed by incubation at 37°C for 15 min. The reaction was stopped by adding 80 µl of 0.2 M sodium carbonate solution and absorbance was read at 405 nm in a microplate reader (Epoch, BioTek ® ). The reaction system without any test compounds was used as the control, and the system without α-glucosidase served as the blank for correcting the background absorbance. The rate of inhibition on α-glucosidase by the sample was calculated with Eq. (1): DPPH radical scavenging assay The scavenging of free radicals by the synthesized compounds was assessed with the previously reported 2,2diphenyl-1-picrylhydrazyl radical (DPPH) assay, with some modifications [48]. After a solution of each compound was elaborated at a concentration of 2.5 mM in DMSO, a solution of DPPH (133.33 µM in absolute ethanol) was added at a ratio of 1:3 (v/v). The mixture was incubated at 37°C for 30 min before reading the absorbance in a microplate reader (Epoch, BioTek®) at 517 nm. Butylhydroxytoluene served as the positive control. Scavenging capacity was expressed as the percent decrease in DPPH at 2.5 mM: where A control is the absorbance of the DPPH solution (control) and A test is the absorbance of the DPPH solution plus a compound.

Molecular docking studies
Molecular docking simulations were carried out on the AutoDock 4 program [49]. The α-glucosidase homology model was constructed by using the crystal structure of isomaltase from Saccharomyces cerevisiae in complex with its competitive inhibitor maltose (Protein Data Bank (PDB): 3A4A), which was retrieved from the PDB (http://www. rcsb.org/). Docking was validated with maltose to identify the main side chains present at the active site of the enzyme. Water molecules were removed, hydrogen atoms were added to the polar atoms (considering pH at 7.4), and Kollman charges were assigned with AutoDock Tools 1.5.6. The 3D structures of maltose and acarbose were downloaded from the ZINC 15 database [50]. The alkoxyisoflavone derivatives were sketched in two dimensions with ChemSketch (https://www.acdlabs.com/resources/ freeware/chemsketch/) and converted into 3D mol2 format in the Open Babel GUI program [51]. The maltose and acarbose ligands were optimized with PM6 on Gaussian 9875 software to obtain the minimum energy conformation for the docking studies. All the possible rotatable bonds, torsion angles, atomic partial charges and non-polar hydrogens were determined for each ligand. The grid dimensions in AutoDockTools were 60 × 72 × 66 Å 3 with points separated by 0.375 Å, centered at: X = 24.0, Y = −8.0 and Z = 23.028. The hybrid Lamarckian genetic algorithm was applied for minimization, utilizing default parameters. A total of 100 docking runs were conducted, adopting the conformation with the lowest binding energy (kcal/mol) for all further simulations. AutoDockTools was used to prepare the script and files as well as to visualize the docking results, which were edited in Discovery 4.0 Client.