Evaluation of the Antidiabetic Activity and Chemical Composition of Geranium collinum Root Extracts—Computational and Experimental Investigations

The root of Geranium collinum Steph is known in Tajik traditional medicine for its hepatoprotective, antioxidant, and anti-inflammatory therapeutic effects. The present study was conducted to evaluate of potential antidiabetic, antioxidant activities, total polyphenolic and flavonoid content from the different extracts (aqueous, aqueous-ethanolic) and individual compounds isolated of the root parts of G. collinum. The 50% aqueous-ethanolic extract possesses potent antidiabetic activity, with IC50 values of 0.10 μg/mL and 0.09 μg/mL for the enzymes protein-tyrosine phosphatase (1B PTP-1B) and α-glucosidase, respectively. Phytochemical investigations of the 50% aqueous-ethanolic extract of G. collinum, led to the isolation of ten pure compounds identified as 3,3′,4,4′-tetra-O-methylellagic acid (1), 3,3′-di-O-methylellagic acid (2), quercetin (3), caffeic acid (4), (+)-catechin (5), (–)-epicatechin (6), (–)-epigallocatechin (7), gallic acid (8), β-sitosterol-3-O-β-d-glucopyranoside (9), and corilagin (10). Their structures were determined based on 1D and 2D NMR and mass spectrometric analyses. Three isolated compounds exhibited strong inhibitory activity against PTP-1B, with IC50 values below 0.9 μg/mL, more effective than the positive control (1.46 μg/mL). Molecular docking analysis suggests polyphenolic compounds such as corilagin, catechin and caffeic acid inhibit PTP-1B and β-sitosterol-3-O-β-d-gluco-pyranoside inhibits α-glucosidase. The experimental results suggest that the biological activity of G. collinum is related to its polyphenol contents. The results are also in agreement with computational investigations. Furthermore, the potent antidiabetic activity of the 50% aqueous-ethanolic extract from G. collinum shows promise for its future application in medicine. To the best of our knowledge, we hereby report, for the first time, the antidiabetic activity of G. collinum.

Extraction was conducted for various extraction time durations using the optimal extraction conditions. The extract recovery yield appeared to increase as the extraction time increased up to 3 h, but further extension of the extraction time did not result in any further increase of extract recovery yield. No big difference in the extractive yield was observed between 2 h and 3 h and in the case of 70% ethanol 2 h extraction delivered more yield than 3 h. Therefore, in order to save energy and time, the optimal extraction time was set to be 2 h. All of the extracts obtained with 2 h extraction times were used for quantification of total polyphenols, flavonoids and their pharmacological activities. Figure 1 shows a graphical comparison of the extraction yields of various combinations of solvents and extraction time. Using 2 h extraction 70% ethanol delivered the highest extraction yield (31 g/100 g) among all the tested solvents, as the extraction values for H 2 O, 30%, 50%, and 100% ethanol were 19 g/100 g, 21 g/100 g, 28 g/100 g and 22 g/100 g, respectively.
Molecules 2017, 22, 983 3 of 12 conditions. The extract recovery yield appeared to increase as the extraction time increased up to 3 h, but further extension of the extraction time did not result in any further increase of extract recovery yield. No big difference in the extractive yield was observed between 2 h and 3 h and in the case of 70% ethanol 2 h extraction delivered more yield than 3 h. Therefore, in order to save energy and time, the optimal extraction time was set to be 2 h. All of the extracts obtained with 2 h extraction times were used for quantification of total polyphenols, flavonoids and their pharmacological activities. Figure 1 shows a graphical comparison of the extraction yields of various combinations of solvents and extraction time. Using 2 h extraction 70% ethanol delivered the highest extraction yield (31 g/100 g) among all the tested solvents, as the extraction values for H2O, 30%, 50%, and 100% ethanol were 19 g/100 g, 21 g/100 g, 28 g/100 g and 22 g/100 g, respectively.

Spectral Identification of the Isolated Compounds
The structures of all isolated compounds were elucidated by analyzing their spectral data (MS, 1 Hand 13 C-NMR, including HMQC, HMBC and DEPT) and by comparison with published literature data. They were thus identified as 3,3 ,4,4 -tetra-O-methylellagic acid [10], 3,3 -di-O-methylellagic acid [11], caffeic acid [12], quercetin [13], (+)-catechin [14], (-)-epicatechin [13], (-)-epigallocatechin [15], gallic acid [16], β-sitosterol-3-O-β-D-glucopyranoside [17] and corilagin [18] (Figure 2). conditions. The extract recovery yield appeared to increase as the extraction time increased up to 3 h, but further extension of the extraction time did not result in any further increase of extract recovery yield. No big difference in the extractive yield was observed between 2 h and 3 h and in the case of 70% ethanol 2 h extraction delivered more yield than 3 h. Therefore, in order to save energy and time, the optimal extraction time was set to be 2 h. All of the extracts obtained with 2 h extraction times were used for quantification of total polyphenols, flavonoids and their pharmacological activities. Figure 1 shows a graphical comparison of the extraction yields of various combinations of solvents and extraction time. Using 2 h extraction 70% ethanol delivered the highest extraction yield (31 g/100 g) among all the tested solvents, as the extraction values for H2O, 30%, 50%, and 100% ethanol were 19 g/100 g, 21 g/100 g, 28 g/100 g and 22 g/100 g, respectively.

Total Polyphenolic Compounds
Total polyphenolic compounds were estimated as gallic acid equivalents, using the Folin-Ciocalteu method, whereby polyphenolic compounds produce a blue-colored complex with a wavelength maximum at 740 nm. Seven point calibration curves were constructed using Microsoft Excel spreadsheets and forcing the curve through zero. The curve delivered a R 2 value of 0.9956 and a straight line equation (y = 10.909x). The highest amounts of total polyphenolic compounds were calculated as 349.84 ± 0.21 and 180.14 ± 0.11 mg GAE/g of the dried extract in 50% ethanol and 70% ethanol extracts, respectively (Table 1). Total polyphenolic compounds quantified in water, 30% and absolute ethanol extracts were: 12.21 ± 0.10, 83.74 ± 0.18 and 100.42 ± 0.14 mg GAE/g of the dried extract, respectively. Table 1.
Results of the total polyphenolic compounds, total flavonoids, antioxidant and antidiabetic activities.

Total Flavonoids Contents
The quantification of contents of total flavonoids was based on the complex formation between AlCl 3 and flavonoids present in the extracts. Seven point calibration curve was obtained with an R 2 value as 0.9971 producing a straight line equation (y = 9.157x). Using this equation, quantity of total flavonoids (x) with high contents in 50% and 70% ethanol extracts were determined as 96.07 ± 0.08 and 75.31 ± 0.07 mg quercetin equivalents (QE)/g of the dried extract (Table 1). In water, 30% and absolute ethanol extracts, the estimated total flavonoids contents were: 3.31 ± 0.04, 42.77 ± 0.12 and 55.68 ± 0.02 mg quercetin equivalents (QE)/g of the dried extract, respectively.

Antioxidant Activity
The antioxidant activity of several phenolic compounds such as catechin, gallic acid, caffeic acid and chlorogenic acid have been previously evaluated [19]. The majority of the antioxidant activities (such as free radical scavenging) of herbal medicines, vegetables and fruits may be due to their phenolic compounds. A comparative evaluation of the antioxidant activities of the extracts after 2 h of extraction is tabulated in Table 1. A lower IC 50 value than the standard vitamin C implied a significant antioxidant activity of the tested samples. Among the understudied extracts the 30% ethanol extract was comparably active, with an IC 50 value of 10.89 ± 0.63 µg/mL. The IC 50 values (µg/mL) of the 50% ethanol, 70% ethanol, absolute ethanol and water extracts were: 11.21 ± 0.49, 12.69 ± 0.6, 11.23 ± 0.7 and 15.17 ± 0.84, respectively.

Antidiabetic Activities of Crude Extracts and Isolated Individual Compounds
Flavonoids and phenolic glycosides of plant extracts have shown promising effect as inhibitors of PTP-1B and α-glucosidase. They can be used in pharmaceutical formulations against type 2 diabetes mellitus [14,20,21]. The antidiabetic effect of gallic acid was evaluated in vitro on inhibitory enzymes (α-glucosidase and α-amylase) [22]. Antidiabetic activity was determined using PTP-1B inhibition and α-glucosidase procedures. PTP-1B is a negative regulator of the insulin signaling pathway [23]. All extracts of G. collinum were obtained after a period of extraction of 2 h. They were assayed for their inhibitory activities against PTP-1B and α-glycosidase, delivering strong activity, with IC 50 values ranging from 0.1 µg/mL to 1.98 µg/mL (Table 1). Among the five tested samples, water and 50% ethanol extracts showed strongest antidiabetic (PTP-1B inhibition) activity with an IC 50 0.13 ± 0.01 µg/mL and 0.10 ± 0.01 µg/mL, respectively. That is more effective than the positive control for the PTP-1B inhibitor (1.46 ± 0.40 µg/mL). Results of α-glucosidase inhibitory activities of root extracts are presented on Table 1. In addition, in vitro inhibitory effect on the enzymes PTP-1B and α-glucosidase of the pure isolated compounds 1-10 from G. collinum were evaluated. Catechin, epicatechin and corilagin delivered the strongest activities, with IC 50 values of 0.62, 0.23 and 0.87, respectively. That was more than the known PTP-1B inhibitor (1.46 µg/mL). Catechin, epicatechin, corilagin and quercetin have shown potent antidiabetic activities in the α-glucosidase inhibition assay. IC 50 values of the all the samples are presented in Table 2. The results of the in vitro assays, antidiabetic analyses of the isolated pure compounds, confirmed the results of the in silico computational investigation. It is well known that polyphenols can provide health benefits to humans attributed to their strong radical-scavenging and antioxidant effects [24]. These effects may also contribute to the prevention of diseases, such as diabetes. Additionally, intake of catechin-rich non toxic plants slightly inhibited postprandial elevation of blood glucose levels and oxidative products. Therefore these effect suggest that consumption of decoctions containing catechins could reduce the risk of type 2 diabetes.  The experimental results suggest promise for the future application of plant extracts for the development of phytopharmaceuticals and food formulations against diabetes. Nevertheless, toxicological studies should be carried out to ascertain the boundary between health beneficial effects and the risk of possible toxicity.

Molecular Docking
In order to provide some insight as to which compound(s) may be responsible for the antidiabetic activity, a molecular docking study was carried out on two different protein structures of human PTP-1B and two different protein structures of human α-glucosidase. Molegro re-rank energies (E dock ) and normalized docking scores (DS norm ) of the G. collinum ligands with the protein targets are summarized in Table 3.  Polyphenolic ligands such as epigallocatechin or caffeic acid showed preferential docking with HsPTP1B, although the docking energies were not as exothermic as the co-crystallized ligands for these proteins. Daucosterol (β-sitosterol-3-O-β-D-glucopyranoside), on the other hand, showed docking preference for α-glucosidase.
The docking energy for daucosterol was comparable to the co-crystallized ligand, α-acarbose. The β-D-glucose moiety of daucosterol overlays the 6-amino-4-(hydroxymethyl)-4-cyclo-hexene-1,2,3-triol group of the co-crystallized α-acarbose, occupying a cavity on the protein surrounded by hydrogen-bonding amino acids His1584, Asp1526, and Asp1279 ( Figure 3). Although corilagin showed strong docking to both PTP-1B and α-glucosidase, this compound is a hydrolyzable tannin, which is well known to bind indiscriminately to proteins and shows non-selective docking. 1,2,3-triol group of the co-crystallized α-acarbose, occupying a cavity on the protein surrounded by hydrogen-bonding amino acids His1584, Asp1526, and Asp1279 ( Figure 3). Although corilagin showed strong docking to both PTP-1B and α-glucosidase, this compound is a hydrolyzable tannin, which is well known to bind indiscriminately to proteins and shows non-selective docking.

General Procedures
Absolute ethanol and acetone were purchased from Tianjinshi Baishi Chemicals Company (Urumqi, China). DMSO, DPPH, pNPP, NaOH, NaCl, vitamin C, quercetin (98%), gallic acid (≥97%), aluminum chloride, sodium acetate and Folin-Ciocalteau reagent (2 N) were purchased from Sigma-Aldrich GmbH (Steinheim, Germany). All the chemicals and reagents were of analytical grade and double-distilled water was used throughout the experiment. NMR spectra recorded on Varian MR-400 (400 MHz for 1 H and 100 MHz for 13 C) spectrometer. Melting points were determined using a BUCHI Melting Point B-540 apparatus (Sigma-Aldrich, Darmstadt, Germany). Column chromatographic separation was performed on Sephadex LH-20 gel (Amersham Pharmacia Biotech, Stockholm, Sweden) and silica gel (100-200 mesh, Qingdao Haiyang Chemical Factory, Qingdao, China). The spots on the TLC plate were identified by spraying with 5% H 2 SO 4 solution in EtOH and heating the plate at about 105 • C. UPLC analysis was performed on a Waters Acquity UPLC™ system (Waters, Milford, MA, USA) equipped with binary solvent delivery pump, an auto sampler, and a photodiode array detector (PAD). The instrument was controlled by the Waters Empower 2 software. The chromatographic separation was performed using a Waters Acquity BEH Shield C18 column (100 mm 2.1 mm i.d., 1.8 µm, Waters), operated at 35 • C.

Plant Material
Underground parts of G. collinum were collected from Takob Valley in the Republic of Tajikistan in October 2015. The roots were authenticated by Professor Yusuf Nuraliev and voucher sample has been deposited in the herbarium of the Xinjiang Technical Institute of Physics and Chemistry Urumqi, Chinese Academy of Science.

Determination of Total Polyphenolic Compounds and Total Contents of Flavonoids
Each dried root extract (1 g each) of G. collinum was reconstituted in 20 mL of 70% methanol and this extract was used for the determination of total polyphenolic compounds and total flavonoids contents. Total polyphenolic compounds were determined by Folin-Ciocalteau [25,26] method using gallic acid as the reference standard in the concentration range of 0.02 mg/mL to 0.2 mg/mL in water. Amount of total flavonoids were estimated using the procedure adopted by Numonov et al. [27].

Molecular Docking
Protein-ligand docking studies were carried out based on the structures of two human protein tyrosine phosphatase 1B crystal structures (HsPTP1B, PDB 3CWE [31] and PDB 4Y14) [32] and two α-glucosidase crystal structures (human sucrose-isomaltase, HsSI, PDB 3LPP [33], and human maltase-glucoamylase, HsMGAM, PDB 3TOP [34]). Prior to docking, all solvent molecules and the co-crystallized ligands were removed from the structures. Molecular docking calculations for all compounds with each of the proteins were undertaken using Molegro Virtual Docker (version 6.0.1, Molegro ApS, Aarhus, Denmark) [35], with a sphere (15 Å radius) large enough to accommodate the cavity centered on the binding sites of each protein structure in order to allow each ligand to search. Standard protonation states of the proteins based on neutral pH were used in the docking studies. Each protein was used as a rigid model structure; no relaxation of the protein was performed. Assignments of the charges on each protein were based on standard templates as part of the Molegro Virtual Docker program; no other charges were necessary to be set. Each ligand structure was built using Spartan16 for Windows (version 2.0.1, Wavefunction Inc., Irvine, CA, USA). For each ligand, a conformational search and geometry optimization was carried out using the MMFF force field [36]. Flexible ligand models were used in the docking and subsequent optimization scheme. Variable orientations of each of the ligands were searched and ranked based on their re-rank score. For each docking simulation the maximum number of iterations for the docking algorithm was set to 1500, with a maximum population size of 50, and 30 runs per ligand. The RMSD threshold for multiple poses was set to 1.00 Å. The generated poses from each ligand were sorted by the calculated re-rank score. In order to correct for the known biasing of docking energies (Edock) with increasing molecular weight (MW) [37], we have also determined a normalized docking score (DS norm ) based on the molecular weight: DS norm = 7.2 × Edock/MW 1 / 3 [38].

α-Glucosidase Inhibition Assay
α-Glucosidase inhibitory activities were evaluated according to the chromogenic method described by [39].

Antioxidant Activity
Antioxidant activity was measured using the DPPH (1,1-diphenyl-2-picrylhydrazyl) scavenging assay procedure as published in the literature [40,41]. The extracts were dissolved in DMSO in a concentration of 100 µg/mL. Different concentrations of each extract were prepared in DMSO. Sample solution in different concentrations was added at 2.5 mL in each well of the 96 micro well plate. Into each well was added 1 mL of 0.3 mM DPPH solution in ethanol to produce the test solutions. DMSO (1 mL) was added to produce the blank solutions. Negative control sample was prepared by mixing 1 mL of DPPH solution with 2.5 mL of DMSO. The solutions were kept in the dark at room temperature for 30 min. Absorbance was measured at the wavelength of 517 nm. Ascorbic acid (vitamin C) was used as the standard sample.

Conclusions
The underground part of G. collinum growing in Tajikistan can provide a natural source of antidibetical drugs and antioxidants. Our investigations suggest the development of herbal formulations based on the roots parts of G. collinum growing in Tajikistan, however, further toxicological and dose determination studies are required to realize an effective phytopharmaceutical. The 50% ethanol extract delivered 10 pure biologically active compounds. Based on molecular docking, we conclude that polyphenolic constituents are likely responsible for the inhibition of PTP-1B while the sterol glucoside (daucosterol) is a likely inhibitor of human α-glucosidase. Our investigation clearly showed that other fractions also have potent activities and deserve further investigation in the future. In the present investigation, phenolic compounds and steroid glycoside were identified for the first time from roots of G. collinum. The alcoholic extracts of G. collinum exhibited strong antidiabetic activity.