Glucosinolates with their Hydrolysis Products from Two Cruciferous Plants with Study of Antidiabetic Activity Based on Molecular Docking

1Department of Chemistry, Faculty of Sciences and Arts, Mukhwah, Albaha university, Albaha, Saudi Arabia. 2Department Chemistry of Medicinal Plants, Pharmaceutical Industries Div., National Research Centre, 33 El Buhouth St.(Former El Tahrir St.), 12622-Dokki, Giza, Egypt. 3Chemistry Department, Faculty of Science, Al-Azhar University (Boys Branch), Nasr City, Cairo, Egypt. *Corresponding Author E-mail: khabdelhady@gmail.com

Many plants of the Cruciferae (formerly Brassicaceae) family contain glucosinolates which are precursors of strongly odorous pungent isothiocyanates formed by the action of myrosinase on glucosinolate when the plant tissues are ruptured 1 . The Cruciferae family is considered one of the largest angiosperm families and includes many economically important plants, such as salad vegetables and crop species. These plants have many biological activities, with anticancer, antibacterial, antifungal, antirheumatic, and antidiabetic properties 2 . Carrichtera and Farsetia are two genera belonging to the Crucifereae family. In Egypt, there is only one Carrichtera species, C. annua L.(DC) grown in the Sinai peninsula, especially in the Elarish region, while the Farsetia genus is represented by three species, of which, F. aegyptia Turra. is more popular and grown in many localities throughout Egypt 3 . Both plants have been used by the native Bedouins for their antidiabetic and antispasmodic effects. Moreover, F. aegyptia is used to relieve rheumatic pains and taken as a cooling medicine after pounding. F. aegyptia is also prepared as a decoction for toothache, gingivitis, and sore eyes, and mixed with alum for use as a mouth rinse and disinfectant [2][3][4] .
The glucosinolates and their hydrolysis products in suspension culture following elicitation of F. aegyptia were found to contain glucotropaeolin (bezyl Gls) and gucocheirolin (3-methylsulfonylpropyl-Gls), isobutyl Gls and gluconasturtiin in addition to twenty-two hydrolysis products [9][10] . Total glucosinolates of C. annua and F. aegyptia exhibit strong antifeedant effects against the 4th instar larvae of the Egyptian cotton leaf worm, S. littoralis 4 . D i a b e t e s m e l l i t u s i s a c h r o n i c endocrine disorder that affects the metabolism of carbohydrates, proteins, fat, electrolytes, and water. It is characterized by hyperglycemia in which blood sugar levels are elevated because the pancreas does not produce enough insulin or cells do not respond to the produced insulin 11 . Therefore, a therapeutic approach to treat diabetes is to decrease postprandial hyperglycemia12, which can be achieved by the inhibition of carbohydrate hydrolyzing enzymes like á-amylase and á-glucosidase, important enzymes involved in the digestion of carbohydrates.
Docking and DFT simulations are powerful computational tools that have been proven to be inexpensive and efficient approaches for understanding fundamental characters of biomolecules 13 . DFT calculations provide information regarding important topographic and molecular characteristics such as optimization geometry, charge-transformation, global and local reactivities [14][15][16] . Molecular docking studies explore the possible binding mode of the compound via its target protein, predicting the ligand-receptor orientation that is useful to guide improved compound features for drug design 17 . The present study isolated and identified glucosinolates and their hydrolysis products of C. annua and F. aegyptia, combining DFT, ADMET, and molecular docking analysis to investigate the potential antidiabetic properties.

Plant Materials
The aerial parts of both plants (Carrichtera annua (L.) DC. and Farsetia aegyptia Turra) were collected from Alariesh airport region (North Saini governate) and wadi Hagol (at Cairo-Sues Road, 30km), respectively, through the flowering stage at the end of the winter season (25 th March). The plants were kindly, identified by Prof. Ibrahim Elgarf, professor of taxonomy at the Botany Dept., Faculty of Science, Cairo University, and the voucher specimens were deposited at NRC herbarium. Each plant was air-dried and ground to a fine powder. instruments and reagents 1 H and 13 C NMR spectra were measured using Joel JNM EX270FTNMR system at 270 MHz for 1 H and 70 MHz for 13 C at the central lab-NRC, Cairo, Egypt. The sample was dissolved in DMSO and the chemical shifts are given in ä (ppm) relative to tetramethylsilane (Me 4 Si). The UV spectra were recorded on a Shimadzu model UV-240 and a 2401 PC spectrophotometer (Shimadzu Inc., Tokyo, Japan) in methanol in the range of 200-450 nm.

Conditions for GC/Ms analysis of the isothiocyanates apparatus
G C / M S F i n n i n g a n S S Q 7 0 0 0 , Digital DEC3000 with column: DB5 capillary (I.d.0.25mm), ionization mode: EI at 70ev, temperature program starting at: 50ÚC up to 250 ÚC with a rate of 4 ÚC/min., Detector: MS, sample volume 2 ?l and the mass from 40-400.
• modified silver nitrate reagent 12 in which the Gls. appears as brownish spots on a yellowish background.
• Solvent systems: S1, Butanol: acetic acid : water 4:1:5 upper layer (which contain the organic solvent saturated with water) and S2, Butanol: ethanol: water 4:1:3. Acidic aluminium oxide (anionotropic 1.5 kg, activity grade I, type WA-I acidic, Sigma chemical Co.), Cellulose for column (Merck, 64271 Darmstadt, Germany) and DEAE Sephadex A-25 (Germany). myrosinase enzyme (Thioglucosidase, Sigma, T -4528). extraction and isolation of Glucosinolates About 2.0 kg of air-dried powdered herb of both C. annua and F. aegyptia were defatted, separately, with petroleum ether (br. 40-60°C) in a Soxhlet apparatus for three days (operated 12 hrs. daily). The defatted powders were air -dried and extracted (separately) with methyl alcohol (80%, 3x3L, 3 days each), the combined methanol extract of each plant was evaporated in vacuo at 40 ºC and the residue was dissolved in hot distilled water (750 ml), left in the refrigerator for 24 hrs, and filtered the gummy ppt. The aqueous filtrates were allowed to pass slowly through an acidic aluminium oxide column (Aluminium oxide, for chromatography, acidic, Brockmann I, 50-200 µm, 60A, Acros Organics, Fluca, 7.5x60) to remove the coloring matter, then the total Gls were eluted with potassium sulfate solution (2%, 2.5L) until the brown zone (containing the total glucosinolates) reached the bottom of the column, and the eluate containing the brown zone was evaporated under reduced pressure till dryness. The residue for each plant was dissolved with hot methanol, filtered through a central glass funnel (G3) and evaporated in vacuo at 40°C till dryness to leave a faint brownish residue (2.1 g for CA and 500 mg for FA)

Purification of Gls
The obtained brownish residue was further purified using a cellulose column eluted with the upper layer of the solvent S2, after the complete elution, the eluate was concentrated and subjected to preparative paper chromatography (PPC) using chromatographic paper Whatmann 3 mm developed in the upper layer of solvent S1 (two runs) by applying the descending technique. Two main zones were determined with a pencil cut into small pieces and eluted with methanol (90%). Each zone was further purified over a DEAE Sephadex A-25 column eluted with 0.5% ammonium acetate soln., to afford two compounds (CA1 and CA2, R b 0.52 and 0.65, relative to benzyl Gls, 17mg and 3mg, respectively).

enzymatic hydrolysis of Gls
Due to the small amounts of the compound A2 and the residue of total Gls from FA, so compound A2 and the total Gls from FA were subjected, separately, to enzymatic hydrolysis using a myrosinase enzyme. The aqueous acidic solution after separation of the aglucones was neutralized with barium carbonate and filtered. The clear filtrate was evaporated till dryness. The residue was dissolved in 10% isopropanol and subjected to paper chromatography using ethyl acetate: pyridine: water (12: 5: 4) as the developing solvent and different authentic sugars. The chromatogram was visualized by spraying with aniline phthalate, and heated at 105 °C for a few minutes. Only glucose was detected as a sugar and the presence of sulfate ions in the aqueous solution was detected by the addition of a few drops of barium chloride soln., where a white ppt. was noticed, proving the presence of sulphate ions.
Also, the total Gls. were extracted from F. aegyptia using the same method, which were purified using PPC with different solvents and repeated Sephadex LH-20 columns to afford five compounds in small amounts, so they were subjected to enzymatic hydrolysis as before, and the hydrolysis products were extracted with ether and identified by GC/MS. The aqueous layer was found to contain glucose and sulphate ions.

Computational study Preparation of the small Molecule
The target compound was built and minimized using the PM3 semi-empirical Hamiltonian molecular orbital calculation MOPAC16 package [32], then employing the density function theory in the Gaussian 09 program package [28] with the Becke3-Lee-Yang-parr (B3LYP) level using a 6-311G* basis, as implemented in the MOE 2015 package 17 . The optimization Geometry for molecular structures was applied to improve the knowledge of chemical structures. Our compounds were introduced into the binding sites according to the published crystal structures. The global chemical reactivity descriptors for molecules were computed, like: S; softness (measures the stability of molecules and chemical reactivity with direct proportional), ç; hardness (reciprocal of softness), µ; chemical potential, ÷; electronegativity strength catching electrons, the ì"; electron donating power, ì+; potency for catching the electron, ù"; electro donating capacity, ù+; electro accepting capacity, ù ± ; net electrophilicity (measuring the relative power between electron acceptance and electron donatation) 15 , wi; an electrophilicity index in a ground state (determining the decreasing energy obtained from the maximal electrons' current movement between the donor and acceptor media), wi VS ; and the electrophilicity index in the valance state. These parameters are represented in terms I; ionization potential, and A; electron affinity 15 , as the previous terms represented.

selection of protein structures
A docking experiment was carried out for the target active site into á-glucosidase (PDB: 4yvx) and â-amylase (PDB: 4gqr) using MOE 2015 17 . The errors of the active sites were corrected by the structure preparation process in MOE. After the correction, hydrogens were added and the partial charges (Amber12:EHT) were calculated. Energy minimization (AMBER12:EHT, the root mean square gradient: 0.100) was performed.

Binding site analysis
The binding site of each receptor was identified through the MOE Site Finder program, which uses a geometric approach to calculate the putative binding sites in a protein, starting from its tridimensional structure. This method is not based on energy models, but only on alpha spheres, which are a generalization of convex hulls. The prediction of the binding sites, performed by the MOE Site Finder module, confirmed the binding sites defined by the co-crystallized ligands in the holo-forms of the investigated proteins.

Moe stepwise docking Method
The crystal structures of the enzymes were obtained. Water and the inhibitors' molecule were removed, and hydrogen atoms were added. The parameters and charges were assigned with the MMFF94x force field. After that, the alphasite spheres were generated using the site finder module of the MOE. The optimized 3D structures of the molecules were subjected to generate different poses of ligands using the triangular matcher placement method, which generates poses by aligning the ligand triplets of the atoms with triplets of the alpha spheres represented at the receptor site points. A random triplet of the alpha sphere center was used to determine the pose during each iteration. The pose generated was rescored using the London dG. scoring function. The poses generated were refined with the MMFF94x forcefield; also, the solvation effects were treated. The Born solvation model (GB/VI) was used to calculate the final energy, and the finally assigned poses were assigned a score based on the free energy in Kcal/mol.

adMet predictions
The ADMET in silico profile was applied using "MOE" and "admetSAR" tools to predict the pharmacokinetic and ADMET characters (absorption, distribution, metabolism, excretion, and toxicity).

Isolation and identification of Gls
The total Gls of both plants were isolated and further purified yielding two compounds (CA1 and CA2) from CA and five compounds (FA1-FA5) from FA as follows: CA1 -4-methylthio-3-butenyl Gls: This off-white amorphous powder appeared as a brown spot (R b = 0.52) after spraying with a modified AgNO 3 reagent. The UV absorption spectrum of the compound in methanol had a ë max . at 208 nm which shifted to 217 nm on the addition of NaOH. When silver nitrate solution (1%) was added a new maximum appeared at 260 nm in addition to the first peak at 215 nm with an increase in intensity indicating the glucosinolate nature of the compound 18 .
The 1 H-NMR spectrum (DMSO) showed signals at ?in ppm 2.13 (3H, s, CH 3 , the anomeric proton of glucose was displayed as a doublet at 5.1ppm. The complex group of signals between 3.1 and 3.85 is characteristic for the rest proton of the glucose moiety. The 13 C-NMR spectrum gave the anomeric carbon of the glucose unit at 82.16, the two carbon atoms (C-3 and C-4) of the double bond appeared at 137 and 115.6 ppm respectively, and the central carbon of the Gls was assigned at 156.21 ppm. The other data were summarized in Table-1. These findings are in line with a report by Manuela et al. 11 . The GC/MS of the isolated aglucone of this compound after the enzymatic hydrolysis identified the corresponding ITC with the following fragmentation pattern: the mass spectrum displayed a mass to charge ratio (m/z) of 159 due to the molecular ion peak M + which corresponded to the molecular formula C 5 H 7 S 2 N and fragment ion peaks at m/z 144 (M + -CH 3 ), 101 (M + -NCS) and 73 (M + -CH 2 NCS). By comparing these data with the reported data, the compound CA1 was identified as 4-methylthio-3-butenyl glucosinolate.
CA2 -6-methylsulfonylhexyl Gls: This compound was isolated in a small amount (~ 3 mg) as a white-yellowish powder which was not sufficient for NMR measurements, so was subjected to enzymatic hydrolysis. The resultant ITC was identified as follows: the mass spectrum exhibited M + at m/z of 221 (20.5%) which fits the molecular formula C 8 H 15 O 2 NS 2 . The other fragments were at m/z = 220 (M + -1, 23%), 205 (M + -CH 3 , 100%), 163 (M + -NCS, 40%) and 142 (M + -CH 3 SO 2 , 35%). This fragmentation pattern confirmed the identification of 6-methylsulfonylhexyl ITC [CH 3 SO 2 -CH 2 -(CH 2 ) 4 -CH 2 -NCS] and accordingly the compound was identified as 6-methylsulfonylhexyl Gls. This is the first report of the isolation of these compounds from C. annua grown in Egypt.
The GC/MS data of the Gls confirmed that the FA hydrolysis products matched previously published data [18][19] and included only isothiocyanates which were identified in an increasing retention manner as follows: FA1 -6-methylsulfonyl6-hydroxy-hexyl ITC: R t . 15.3 min, the mass spectrum displayed
FA3 -3-Methylthio propyl ITC: R t . 16.9 min, the mass spectrum gave M+ at m/z of 147(12%) which fits with the molecular formula
The variation in the glucosinolate hydrolysis products between previous studies and the present work may be due to geographical and seasonal differences, environment, soil type, stress, and plant part examined, or due to the differences in the experimental conditions. Computational studies Frontier orbital analysis "FMOs" frontier molecular orbitals through two types HOMO (donating electron) and LUMO (accepting electron) are crucial orbitals for molecules, displaying the binding properties of biomolecules with a receptor. The FMOs were characterized by the chemical reactivity and kinetic stability of the molecule [19][20] .
The HOMO was localized over the methyl(prop-1-en-1-yl)sulfane region (Figure 1), while the sulfanecarbaldehyde O-((oxidaneyl) Fig. 2. Interaction between A.) reference dug and B.) Gls. with binding site of αglucosidase dioxo-l6-sulfanyl) oxime fragment was covered by the LUMO zone, thus, electron transition (HOMO'!LUMO) from sulfonyl to sulfonyl groups. The low DG enhances the interaction between the HOMO ligand and the LUMO receptor (Table 2) with an inversed mode [20]. The charge transfer was represented by the reactivity index term "DN max ", which was restricted by the stabilization energy during the system gaining charge from the biological media ( Table 2). The low softness value may be combined with a high potential against a biological environment ( Table  2). The ligand had a high electrophilicity value "ù±= 6.741", increasing the electron-donating Fig. 3. Interaction between A.) reference dug and B.) Gls. with binding site of α-amylase power (µ-=7.02ev) with increasing donating capacity (ù -= 6.74ev), which may increase the attacking power against a polar residue in the receptor, thereby increasing the potency.

Molecular electrostatic potential (MeP)
The repulsive and attractive forces were determined by the molecular electrostatic potential (MEP) for 4-methylthio-3-butenylglucosinolate ( Figure 1). Orange, yellow, red colors depict the high electron density area, while the blue color represents the positive potential, and the green color represents the intermediate potential. The negative charge capped most of the molecule area, while the formaldhydoxime area was covered by a positive charge. The variation in color of the MEP surface is associated with a difference in the electrostatic potential. This force is accountable for the interaction of the substrate with the receptor binding site, hence promoting the electrostatic bond between the substrate and receptor 21 . In vitro Ames test result for each TA100 strain (Metabolic activation by rat liver homogenate), TA100 strain (No metabolic activation), TA1535 strain (Metabolic activation by rat liver homogenate) and TA1535 strain (No metabolic activation) [9]. d Toxicity Class ranging from 1 to 6 according to the Global Harmony System (GHS) [8].

docking studies
The The docking study targeted á-glucosidase and â-amylase to examine the mode of action of 4-methylthio-3-butenylglucosinolate.
The ligand-protein interaction was estimated based on the gold score function as implemented in MOE 2015.10 22 . All calculations are presented in Table 2. The crystal structures for á-glucosidase (PDB: 4yvx 23 ) and á-amylase (PDB: 4gqr 24 ) were obtained from the protein databank and they were complexed with glimipride and myricetin as reference drugs, respectively. The tested ligand docked into the active site of the two enzymes. The active site of these enzymes was defined to include residues within a 3.5 Å radius around the reference drug atoms. The molecular dynamic (MD) using an (MMFF94) force field to the 0.05 Kcal/mol gradient convergence range was applied to minimize the energy for the obtained ligand-enzyme complexes. The binding affinity of the ligands tested was determined by the highest MOE scoring function ( Table 3).
The premier ligands had a MOE score of -7.511 and -6.157 Kcal/mol, respectively, for á-glucosidase and â-amylase. In the á-glucosidase binding site, glimipride forms an important strong hydrogen bond with His117 and Tyr55, at the time this ligand forms p-p bonds with Tyr24 and Trp22. The analysis of the binding site of â-amylase-myricetin showed that the ligand interacted with an important amino acid backbone Asp197, which contacted 2-H-bonds. The tested compound exhibited a higher binding affinity with a MOE score of -6.611 Kcal/mol than the reference drug (-6.115 Kcal/mol.). This ligand formed an important hydrogen bond with Asp ( Figure 3) and was arranged in a binding pocket in a parallel mode with Asp ( Figure 3).
The hydrophilicity of the binding pocket indicates that the hydrophobicity and membrane permeability are important pharmacokinetic characteristics for absorption molecules in biological systems. These results demonstrate that the amino acid residues close to the reference molecules are mostly the same as those observed in the test compounds. in silico toxicological study

In silico pharmacokinetic Profile
T h e c a l c u l a t e d d e s c r i p t o r s f o r 4-methylthio-3-butenylglucosinolate and glimipride were calculated by MOE, SwissADME 25 , and the admet-SAR model 26 , as shown in Table 4. The physicochemical and ADME parameters for 4-methythio-3-butenyl glucosinolate glimipride (reference drug) indicate that this ligand is suitable for Lipinski's rule with one violation related to molecular weight. Furthermore, when the Ghose Veber, Egan and Muegge rules were applied only glimipride qualified for the Muegge rule. Consequently, the test compound exhibited bioavailability scores of 0.11 and 0.55 with biodegradation values 0.6362 and 0.4523, confirming good oral bioavailability. The leadlikeness profile plotted the Bioavailability Radar planner for the test compounds (Figure 4) showing the relationship between polarity, size, lipophilicity, solubility, saturation, and ûexibility 20 , with the optimal range for each parameter represented in pink. From Figure 4, 4-methythio-3-butenyl glucosinolate and glimipride exhibited deviation for polarity and flexibility, respectively.
Both compounds exhibited a high degree of saturation (Fraction Csp3 = 0.75,0.45, respectively), which pass the filter of Fraction Csp3< 0.25 27 . The solubility parameters are important for absorption, with 4-methythio-3butenyl glucosinolate being strongly soluble in H 2 O when the ESOL topological model was applied with absorption% (%ABS=44.48) 28 , whereas glimipride showed poor solubility in this model (Table 4).
Also, the lipophilicity parameters in medicinal chemistry were incorporated as a leadlikeness filter to identify problematic fragments in bioactive molecules based on two structural alerts, PAINS and Brenk's filters 20 . The tested compounds exhibited no structural alerts against PAINS and Brenk's filters.
The mutagenic, tumorigenic, reproductive, and irritant properties and human intestinal absorption were investigated in silico using ADMET-SAR pharmacokinetic parameters 29 . Furthermore, the SVM algorithm was applied to identify whether the substrate or non-substrate could permeate the skin (Log Kp), Caco-2, bloodbrain barrier (BBB), and p-glycoprotein (P-gp), as well as determine the inhibitory effect against the main cytochromes P450 isoenzymes (CYP1A2, CYP2C19, CYP2C9, CYP2D6, CYP3A4). The results presented in Table 3 show that glimipride inhibited both CYP2C9 and CYP3A4 but the test ligand did not inhibit any P450 isoenzymes. The BOILED-Egg model represents the relationship for WLOGP vs TPSA in Figure 4, which demonstrated high BBB permeability, high GI absorption, high brain penetration, and ambiguous inhibition action against human Ether-à-go-go-Related Gene (hERG). Glimipride activated P-gp (multidrug resistance protein 1) but the test compound acted as a P-gp inhibitor. Furthermore, the skin permeability (Kp) results in Table 4 showed that glimipride has higher skin permeability than the test ligand. The tested compounds exhibited no carcinogenicity, mutagenicity, and tumorigenicity effects 30 . oral toxicity prediction The possible toxicity was predicted [31] by estimating the rodent oral toxicity based on data extracted from the Chemical European Biology Laboratory (ChEMBL) database [26]. Also, the median lethal doses (LD 50 ) were estimated in rodents, with the test ligand (LD50 = 16 mg/Kg) showing a lower value than the reference drug (LD 50 = 3250 mg/Kg). The calculated toxicity using this database depends on the highest endpoints including 33 models ( Table 4). The calculated schematic is divided into several stages of toxicity, "toxicity, toxicological endpoints (mutagenicity, carcinotoxicity, organ toxicity (hepatotoxicity), cytotoxicity and immunotoxicity), toxicological pathways (AOPs) and toxicity targets thereby" 25 providing information regarding the possible molecular mechanism as well as toxic response (Table 5). Glimipride was located in the lower toxicity class 5 than 4-methythio-3butenyl glucosinolate in class (No. 4), with both compounds not exhibiting any toxic fragments without non-binding to any toxicity targets (Table 5).
In general, the tested compounds have good oral bioavailability, high ability BBB transport, and no marked health effects observed for rodent toxicity profiles.

ConClusion
The glucosinolates of Carrichtera annua and Farsetia aegyptia (Crucifereae family) were identified as 4-methylthio-3-butenyl Gls and 6-methyl sulfonylhexyl Gls from CA, 6-methylsulfonyl-6-hydroxy hexyl ITC, 4-pentenyl Gls, 3-methylthio propyl ITC, 5-hydroxy pentyl ITC and 4-methylsulphinyl butyl ITC from FA. Future experiments will be conducted to isolate a sufficient quantity of the glucosinolates of Farsetia aegyptia for identification and to investigate their antitumor activity. The obtained docking data showed that the ability of Gls to interact with á-glucosidase and amylase, thus this Gls may be a suitable inhibitor and used as an antidiabetic agent. The ADMET in silico showed that these compounds have good oral bioavailability and a high ability for BBB transport. Furthermore, these compounds had no observed carcinogenic and health effects.

Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.