New Molecules of Diterpene Origin with Inhibitory Properties toward α-Glucosidase

The incidence of diabetes mellitus (DM), one of the most common chronic metabolic disorders, has increased dramatically over the past decade and has resulted in higher rates of morbidity and mortality worldwide. The enzyme, α-Glucosidase (α-GLy), is considered a therapeutic target for the treatment of type 2 DM. Herein, we synthesized arylidene, heterocyclic, cyanoetoxy- and propargylated derivatives of quinopimaric acid (levopimaric acid diene adduct with p-benzoquinone) 1–50 and, first, evaluated their ability to inhibit α-GLy. Among the tested compounds, quinopimaric acid 1, 2,3-dihydroquinopimaric acid 8 and its amide and heterocyclic derivatives 9, 30, 33, 39, 44, with IC50 values of 35.57–65.98 μM, emerged as being good inhibitors of α-GLy. Arylidene 1β-hydroxy and 1β,13α-epoxy methyl dihydroquinopimarate derivatives 6, 7, 26–29, thiadiazole 32, 1a,4a-dehydroquinopimaric acid 40 and its indole, nitrile and propargyl hybrids 35–38, 42, 45, 48, and 50 showed excellent inhibitory activities. The most active compounds 38, 45, 48, and 50 displayed IC50 values of 0.15 to 0.68 μM, being 1206 to 266 more active than acarbose (IC50 of 181.02 μM). Kinetic analysis revealed the most active diterpene indole with an alkyne substituent 45 as a competitive inhibitor with Ki of 50.45 μM. Molecular modeling supported this finding and suggested that the indole core plays a key role in the binding. Compound 45 also has favorable pharmacokinetic and safety properties, according to the computational ADMET profiling. The results suggested that quinopimaric acid derivatives should be considered as potential candidates for novel alternative therapies in the treatment of type 2 diabetes.


Introduction
Enzymes responsible for breaking down proteins, carbohydrates and lipids into smaller and more readily absorbable molecules are a key component of the digestive system. The main cause of many metabolic diseases is abnormal changes in the activity of these enzymes. Inhibition of metabolic enzymes, such as α-glucosidase (α-GLy), is one of the accepted approaches in the treatment of diabetes mellitus (DM), which is one of the most common chronic endocrine diseases, along with arterial hypertension and obesity [1]. The number of people with disorders of carbohydrate metabolism and the incidence of DM are constantly growing, which is primarily due to an increase in the number of patients with obesity, as well as in average life expectancy [2]. Type II diabetes mellitus (DM2), accounting for about 90% of all cases of diabetes, is caused by a decrease in insulin sensitivity in target organs, such as the liver, muscle, and adipose tissue, as well as a deficiency in insulin secretion [3][4][5]. Medicinal agents with the ability to stimulate glucose uptake in these tissues can be used to improve insulin resistance and, therefore, to treat DM2 [6]. Today, a vast number of synthetic antidiabetic agents, such as acarbose, miglitol, sulfonylurea, metformin, and thiozolidinedione, are readily available on the market [7][8][9]. However, their effectiveness is limited, due to low bioavailability and unwanted side effects [10][11][12]. Therefore, there is a great need to develop alternative and more active antidiabetic drugs from natural sources.
Abietane diterpenoids are classes of compounds which are mainly found in the conifer family and have long been used to treat a variety of ailments [13]. Their derivatives are characterized by a wide range of biological activities, like anticancer, antiviral, antimicrobial, antileishmanial, antiplasmodial, antifungal, antitumour, cytotoxicity, antiulcer, cardiovascular, antioxidant, anti-inflammatory and antidiabetic activities [14][15][16][17][18][19][20][21]. Abietic and dehydroabietic acids have been reported to decrease the activity of glucose-6-phosphatase and to stimulate glycogen synthase [22]. Carnosic acid derivatives are very effective in treating diabetic complications by improving insulin secretion [23] and glucose homeostasis or by stimulating glucose uptake by increasing peripheral glucose clearance in tissues [24]. Abietic and carnosic acids also significantly activate nuclear receptor peroxisome proliferator-activated receptor PPAR-γ by exerting its beneficial effect on lipid and glucose homeostasis through PPAR-γ-mediated pathways [25,26]. Carnosol stimulates glucose uptake [27], improves diabetes and its complications by the regulation of oxidative stress and inflammatory responses [28] and suppresses forskolin-induced luciferase expression, when monitored by the cAMP/response element, and glucose-6-phosphatase gene promoters [29][30][31]. Tanshinones exhibited potent protein tyrosine phosphatase 1B inhibitory activity [32] as well as increased the activity of insulin on the tyrosine phosphorylation of the insulin receptor in addition to the activation of the kinases Akt, ERK1/2, and GSK3beta and may be very useful for developing new anti-diabetic agents as specific insulin receptor activators [33].
Studies evaluating the antidiabetic properties of abietane diterpenoids in an animal model using rats and mice showed that dehydroabietic acid reduces plasma glucose and insulin levels, as well as plasma and hepatic triglyceride levels, by suppressing the production of monocyte chemoattractant protein-1 and tumor necrosis factor-alpha and increasing that of adiponectin, through decrease in macrophage infiltration into adipose tissues [34]. Carnosol and carnosic acid reduced plasma glucose, total cholesterol, and triglycerides in a diabetic group of rats and suppressed inflammation and lipogenesis in mice administered a high-fat diet, through C-kinase substrate regulation [23,28,35]. Tanshinone analogs also demonstrated a significant decrease in blood glucose level, total cholesterol and triglyceride, free fatty acids, and insulin receptor substrate 1 expression, body weight loss and higher insulin resistance when administered to type 2 diabetic rats, with oral administration, resulting in the activation of AMP-activated protein kinase in aortas from ob/ob or db/db mice [35,36]. Data on systematic studies of the enzymatic activity of abietane diterpenoid derivatives, obtained as a result of various modifications of the native core, in particular, on levopimaric acid derivatives as potential inhibitors of α-GLy, are practically absent in the literature. Therefore, herein, we describe the synthesis of abietane type derivatives with arylidene, heterocyclic, nitrile and acetylene fragments. These derivatives were, then, first evaluated for in-vitro α-GLy inhibition. The mechanism of inhibition and enzyme binding were investigated with kinetic and molecular modeling approaches.

Chemistry
Since the quinopimaric acid structure (the Diels-Alder reaction product of levopimaric acid and p-benzoquinone) contains major reaction centers at the C-1, C-3, C-4, and C-20 atoms, we planned to functionalize these positions for better understanding of the structure-activity relationship and to reveal new promising molecules with antidiabetic activity (Figure 1).

Chemistry
Since the quinopimaric acid structure (the Diels-Alder reaction product of levopimaric acid and p-benzoquinone) contains major reaction centers at the C-1, C-3, C-4, and C-20 atoms, we planned to functionalize these positions for better understanding of the structure-activity relationship and to reveal new promising molecules with antidiabetic activity (Figure 1). Modifications of these sites involved the synthesis of arylidene and heterocyclic derivatives and quinopimaric acid cyanoetoxy-and propargylated analogs. Figure 2 shows the structures of quinopimaric acid 1 and its analogs 2-30, modified at position C-1, C-3, C-4, and C-20.  Modifications of these sites involved the synthesis of arylidene and heterocyclic derivatives and quinopimaric acid cyanoetoxy-and propargylated analogs. Figure 2 shows the structures of quinopimaric acid 1 and its analogs 2-30, modified at position C-1, C-3, C-4, and C-20.

Chemistry
Since the quinopimaric acid structure (the Diels-Alder reaction product of levopimaric acid and p-benzoquinone) contains major reaction centers at the C-1, C-3, C-4, and C-20 atoms, we planned to functionalize these positions for better understanding of the structure-activity relationship and to reveal new promising molecules with antidiabetic activity (Figure 1). Modifications of these sites involved the synthesis of arylidene and heterocyclic derivatives and quinopimaric acid cyanoetoxy-and propargylated analogs. Figure 2 shows the structures of quinopimaric acid 1 and its analogs 2-30, modified at position C-1, C-3, C-4, and C-20.     We planned to use the Nenitzescu reaction [37] for the synthesis of new diterpene indoles. In this reaction, 1a,4a-dehydroquinopimaric acid 40, easily formed in two steps from quinopimaric acid 1 [38], was used as the quinone component, as well as ethyl 3aminocrotonate or 3-aminocrotononitrile being were used as the new enamine components. Under the conditions of the Nenitzescu indole synthesis by the reaction of 1a,4a-dehydroquinopimaric acid 40 with the corresponding enamine in glacial AcOH, at room temperature, diterpene indoles 42, 44 were synthesized in 76 and 69% yields, respectively. Methyl ester of diterpene indole 43 was obtained in quantitative yield by treating compound 42 with methyl iodide during reflux in acetone for 2 h in the presence of potash (Scheme 1), or direct synthesis from 1a,4a-dehydroquinopimaric acid methyl ester 41, similar to the preparation of compounds 42 and 44.  We planned to use the Nenitzescu reaction [37] for the synthesis of new diterpene indoles. In this reaction, 1a,4a-dehydroquinopimaric acid 40, easily formed in two steps from quinopimaric acid 1 [38], was used as the quinone component, as well as ethyl 3-aminocrotonate or 3-aminocrotononitrile being were used as the new enamine components. Under the conditions of the Nenitzescu indole synthesis by the reaction of 1a,4a-dehydroquinopimaric acid 40 with the corresponding enamine in glacial AcOH, at room temperature, diterpene indoles 42, 44 were synthesized in 76 and 69% yields, respectively. Methyl ester of diterpene indole 43 was obtained in quantitative yield by treating compound 42 with methyl iodide during reflux in acetone for 2 h in the presence of potash (Scheme 1), or direct synthesis from 1a,4a-dehydroquinopimaric acid methyl ester 41, similar to the preparation of compounds 42 and 44. Figure 3 shows the structures of quinopimaric acid heterocyclic derivatives obtained as a result of interaction with hydrazine hydrate 31, thiourea 32, and using the Fischer reaction (indoles 33, 34), Nenitzescu reaction (indoles 35-38) and Beckmann rearrangement (lactam 39). We planned to use the Nenitzescu reaction [37] for the synthesis of new diterpene indoles. In this reaction, 1a,4a-dehydroquinopimaric acid 40, easily formed in two steps from quinopimaric acid 1 [38], was used as the quinone component, as well as ethyl 3aminocrotonate or 3-aminocrotononitrile being were used as the new enamine components. Under the conditions of the Nenitzescu indole synthesis by the reaction of 1a,4a-dehydroquinopimaric acid 40 with the corresponding enamine in glacial AcOH, at room temperature, diterpene indoles 42, 44 were synthesized in 76 and 69% yields, respectively. Methyl ester of diterpene indole 43 was obtained in quantitative yield by treating compound 42 with methyl iodide during reflux in acetone for 2 h in the presence of potash (Scheme 1), or direct synthesis from 1a,4a-dehydroquinopimaric acid methyl ester 41, similar to the preparation of compounds 42 and 44. Propargyl derivatives 45, 46, 48 were obtained in 79-83% yields by the reaction of diterpene indoles 36, 43 and quinone 40 with propargyl bromide during reflux in dimethylformamide in the presence of K2CO3. Cyanoethyloxy derivatives 47, 50 were prepared by adding acrylonitrile in 1,4-dioxane. at room temperature, to the diterpene indole 37 or aromatic derivative 49 in the occurrence of phase transfer catalyst triethylbenzylammonium chloride in combination with an alkali (30% KOH) (Scheme 2). The structures of the synthesized compounds were confirmed using mass spectrometry, and one-and two-dimensional (COSY, NOESY, 1 H- 13 C HSQC, 1 H- 13 C HMBC) NMR spectroscopy. Thus, the signal of the C-2 carbon atom of the aromatic ring in the 13 C NMR spectra of compound 42-44 appeared at δ 99.7-103.3 ppm, and correlated with the signal of the H-2 proton at δ 6.83-7.28 ppm in the 1 H- 13 C HSQC spectra. The 1 H NMR spectra showed characteristic signals of methyl group protons at δ 2.51-2.71 (3'-CH3), as well as broadened signals of the hydroxyl group and NH group at δ 9.12-9.35 and 12.13 ppm, respectively. The 1 H NMR spectra of compound 43 contained an additional signal of the protons of the methyl ester group at δ 3.76 ppm, which, in the 1 H- 13 C HSQC spectrum, correlated with the signal of the C-21 atom at δ 15.5 ppm. In the 13 C NMR spectra of compound 44, a carbon signal of the CN-group was observed at δ 117.8 ppm. The 1 H NMR spectra of propargyl derivatives 45, 46, 48 contained the methylene group proton signal in the region δ 4.67-4.82 ppm, while in the 13 C NMR the triple bond carbon signals appeared at δ 74.4-74.9 and 77.9-79.3 ppm, respectively. The signals of the cyanoethyl methylene groups in the 1 H NMR spectra of compounds 47, 50 were observed in the region δ 2.80-2.91 and 4.05-4.30 ppm, and the characteristic carbon signal of the nitrile group in the 13 C NMR spectra was observed at δ 117.4-117.6 ppm (Figures S1-S18, Supplementary Materials).

Inhibition of Yeast α-Glucosidase
All the synthesized compounds 1-50 were tested for their inhibitory potential against yeast α-GLy. Acarbose served as a control drug in this experiment. The IC50 values of compounds are provided in Table 1 The structures of the synthesized compounds were confirmed using mass spectrometry, and one-and two-dimensional (COSY, NOESY, 1 H- 13 C HSQC, 1 H- 13 C HMBC) NMR spectroscopy. Thus, the signal of the C-2 carbon atom of the aromatic ring in the 13 C NMR spectra of compound 42-44 appeared at δ 99.7-103.3 ppm, and correlated with the signal of the H-2 proton at δ 6.83-7.28 ppm in the 1 H- 13 C HSQC spectra. The 1 H NMR spectra showed characteristic signals of methyl group protons at δ 2.51-2.71 (3 -CH 3 ), as well as broadened signals of the hydroxyl group and NH group at δ 9.12-9.35 and 12.13 ppm, respectively. The 1 H NMR spectra of compound 43 contained an additional signal of the protons of the methyl ester group at δ 3.76 ppm, which, in the 1 H- 13 C HSQC spectrum, correlated with the signal of the C-21 atom at δ 15.5 ppm. In the 13

Inhibition of Yeast α-Glucosidase
All the synthesized compounds 1-50 were tested for their inhibitory potential against yeast α-GLy. Acarbose served as a control drug in this experiment. The IC 50 values of compounds are provided in Table 1.  As shown in Table 1, quinopimaric acid 1 had an activity against α-GLy three times higher than that of acarbose. Its simplest modifications, namely, the reduction of the C2- For compounds 38, 45, 48 and 50, which showed the highest activity against α-GLy, studies of their anti-antioxidant, antimicrobial and cytotoxic activity were carried out (Tables S1-S3, see Supplementary Materials).

Docking Studies for Compound 45
We performed a molecular modeling study to gain insight into the structural basis of interactions between the lead compound 45 and α-GLy enzyme. Since "structure cannot be predicted from kinetics" [45], we avoided preconceived competitive mechanism assumptions and subjected the whole protein surface to a docking procedure ( Figure 5). Nevertheless, docking proposed that diterpene derivative 45 shared a favorable binding site with acarbose. Moreover, the indole hydroxyl group appeared to form a conventional H-bond with carboxyl of the catalytic Asp1100 residue. The indole core itself contributed to the binding the most. It was anchored by strong π-π parallel stacking with a Trp1312 side chain and T-shaped π-stacking with a Phe1370 side chain. The amino group of Lys1403 formed an H-bond with the ester substituent and charged π-cation interaction with the indole aromatic system. The dodecahydrophenanthrene part of the molecule was also stabilized by Van der Waals forces with multiple lipophilic residues (Pro1102, Trp1298, Trp1312, Phe1503). Both ester moieties of compound 45 pointed towards the solvent-accessible area of the pocket, providing an opportunity for the introduction of polar fragments. This modification might improve water solubility Nevertheless, docking proposed that diterpene derivative 45 shared a favorable binding site with acarbose. Moreover, the indole hydroxyl group appeared to form a conventional Hbond with carboxyl of the catalytic Asp1100 residue. The indole core itself contributed to the binding the most. It was anchored by strong π-π parallel stacking with a Trp1312 side chain and T-shaped π-stacking with a Phe1370 side chain. The amino group of Lys1403 formed an H-bond with the ester substituent and charged π-cation interaction with the indole aromatic system. The dodecahydrophenanthrene part of the molecule was also stabilized by Van der Waals forces with multiple lipophilic residues (Pro1102, Trp1298, Trp1312, Phe1503). Both ester moieties of compound 45 pointed towards the solvent-accessible area of the pocket, providing an opportunity for the introduction of polar fragments. This modification might improve water solubility without hampering enzyme binding. To sum up, molecular modeling confirmed the competitive mechanism of action revealed in the kinetic experiment and provided guidance for future structural optimization.

ADMET Profiling of Compound 45
We assessed drug-like, pharmacokinetic and toxicological properties of the lead compound 45 using a consensus of predictive services that took into account different computational strategies ( Table 2). Compound 45 fulfilled Lipinski's rule of 5 (with the exception of molecular weight <500) and Pfizer's rules. At the same time, GSK and "Golden triangle" rules were violated. Mean water solubility was acceptable. Importantly, there was a good consensus on low intestinal absorption and oral bioavailability, which could avoid systemic exposure to the substance. In the case of entering systemic circulation, indole 45 was anticipated to be bound to plasma proteins, likely due to high lipophilicity. Blood-brain barrier penetration was unlikely. Liver metabolism was to be mediated by cytochrome P450 3A4. Acute oral toxicity was predicted to be sufficiently low to achieve a wide therapeutic window. There were no alerts for toxicity to the liver and heart, nor mutagenicity and carcinogenicity. Hence, compound 45's calculated ADMET profile was favorable for the proposed mechanism of action.

Synthesis of Compounds 42 and 44
A threefold excess of ethyl 3-aminocrotonate (0.387 g, 3 mmol) or 3-aminocrotononitrile (0.246 g, 3 mmol) was added with stirring to a solution of compound 40 (0.408 g, 1 mmol) in glacial AcOH (20 mL). The reaction mixture was stirred at room temperature for 20 h, and then poured into H 2 O. The precipitate was filtered off, washed until neutral, and the residue was air-dried. The reaction product was chromatographed on a silica gel column, eluent CHCl 3 -MeOH, 40: 1.
Procedure B. A threefold excess of 3-aminocrotononitrile (0.246 g, 3 mmol) was added with stirring to a solution of compound 41 (0.422 g, 1 mmol) in glacial AcOH (20 mL). The reaction mixture was stirred at room temperature for 20 h, and then poured into H 2 O. The precipitate was filtered off, washed until neutral, and the residue was air-dried. The reaction product was chromatographed on a silica gel column, eluent CHCl 3 -MeOH, 40 13

Conclusions
The screening of a series of 50 semisynthetic derivatives of levopimaric acid revealed that, in contrast to the majority of previously reported diterpene α-GLy inhibitors, a lead diterpene indole with an alkyne substituent 45 was identified as a competitive inhibitor. As a consequence, one might hope for better translatability to animal and clinical settings, since the active site of yeast α-GLy and intestinal mammalian maltase-glucoamylase are conserved, while allosteric sites are likely to be different. In addition, compound 45 is anticipated to have low intestinal absorption that benefits high concentration of the drug in the target area and helps to avoid systemic exposure. Additional experiments are warranted to confirm antihyperglycemic properties of compound 45 in vivo. In the event of the efficacy and safety being confirmed, novel glucosidase inhibitors open a promising venue to antidiabetic agents able not only to ameliorate postprandial hyperglycemia, but also reduce secretory load on pancreatic beta-cells.