A Newly Discovered Phenylethanoid Glycoside from Stevia rebaudiana Bertoni Affects Insulin Secretion in Rat INS-1 Islet β Cells

The tea-like beverage Stevia rebaudiana Bertoni (Stevia) is popular in China because it reduces blood glucose and has a sweet taste. In this work, a comprehensive quality assessment of Stevia led to the discovery of five phenylethanoid glycosides, namely steviophethanoside (1), cuchiloside (2), salidroside (3), icariside D (4), and tyrosol (5). Of them, compound 1 is a novel compound. Mass spectrometry and NMR spectroscopy were employed to confirm the absolute configuration. A hydrolytic step with 4 N TFA at 95 °C for 4 h was used to confirm the monosaccharides. In addition, Discovery Studio 4.0 was used to predict the ADME and toxicity activity of compound 1. The results suggested that compound 1 was biocompatible and had poor toxicity, which was verified by rat INS-1 islet β cells through an MTT assay. Meanwhile, a significant stimulatory effect on INS-1 cells was observed, which indicated a hypoglycemic effect of compound 1. This is the first report that describes a natural, novel, and hypoglycemic phenylethanoid glycoside in Stevia.


Introduction
Stevia rebaudiana Bertoni (Stevia), has been used as a non-cariogenic sweetener for many years [1]. Early in the development of sweeteners, several non-cariogenic sweeteners, such as steviol glycosides, stevioside, and rebaudioside (A to F), were isolated from Stevia. Steviol glycosides have been used to replace sucrose in the treatment of diabetes mellitus and obesity [2]. Recently, several biological activities such as the beneficial effects of blood glucose regulation, antioxidant, and renal protective properties have attracted attention [3][4][5]. It is reported that the crude extract of Stevia had the ability to decrease blood glucose [6]. In addition, stevioside may indirectly contribute to anti-hyperglycemic action [7]. The aqueous extracts of the plant have been extensively used in Paraguay, and the clinical efficacy of regulating blood glucose was confirmed in related studies [8,9].
Normally, blood glucose was decreased by promoting insulin secretion, inhibiting glucose absorption, and promoting glucose decomposition. Among them, insulin is an important and effective

Pharmacokinetic Properties
The predicted pharmacokinetic properties include HIA (human intestinal absorption), BBB (blood brain barrier), PPB (plasma protein binding), CYP2D6, aqueous solubility, and toxicity. Our results showed that compound 1 was found to be non-hepatotoxic (Hepatotoxicity level of 0) and non-inhibitors of CYP2D6 (CYP2D6 level of 0). In addition, the binding was below 90% to carrier proteins. It is well known that the liver plays a critical role in transforming and clearing chemicals, and it is susceptible to the toxicity from these agents. Chemicals that cause liver injury may also induce the injury of other organs (like kidney). In early studies, traditional Chinese medicine presenting with high security and low toxicity compounds even without toxicity, such as phenylethanoid glycosides, which existed in purified fraction of goldengermander, were confirmed to show a lack of hepatotoxicity [28]. Combined with the current results of pharmacokinetic properties, compound 1 could be considered as the one that had low hepatotoxicity, good bioavailability [29], and involved less drug-drug interaction. In addition, the results revealed that compound 1 had optimal aqueous solubility of 4 with undefined BBB level at 4 and very poor absorption (HIA levels of 3), which might be considered as more hydrogen bond donors and acceptors. The property of high solubility in water was probably due to the lower log p value of the compounds, which depicted lower lipophilicity [30].
TOPKAT provides the capability to rapidly and confidently evaluate the toxic effects of chemicals directly from the molecular structure. In this study, compound 1 was found to be non-mutagenic (NM), non-carcinogenic (NC), and non-skin irritant in the toxicity parameter. Besides, the calculated LD 50 (oral administration to rats) value of compound 1 was 5.61 g/kg. According to the different categories defined by the GHS, the category of compound 1 is higher than 5, which means that compound 1 is non-toxic after deglutition [31]. Although the prediction suggests that compound 1 may involve skin sensitization, skin, and ocular irritancy, i.e., some kind of developmental toxicity potential, the toxic dose is currently unclear ( Table 1).

Cytotoxicity Assay
To investigate the cytotoxic effects of compound 1, an MTT assay was used. Regarding the similarity structure, phenylethanoid glycosides (PhGs) (including compound 1 and 2) were evaluated together. After 48 h of exposure, no significant decreases in cell viability was observed dealing with various concentrations of compound 1 and PhGs (Figure 3). This means that both compound 1 and PhGs were not toxic to the INS-1 cells. The result indicated that compound 1 and compound 2 could not exert harmful effects on rat INS-1 cells even at the higher concentration of 400 µg/mL. Although there is no significant difference between compound 1 and PhGs, PhGs presented a trend of more secure effects than single compound 1. Those results provided another proof of low toxicity.

Insulin Secretion in Vitro
Shivanna found that when rats were fed with whole leaves powder and extracted polyphenols of Stevia, a reduction of blood glucose, alanine transaminase (ALT), aspartate transaminase (AST), and increment of insulin level were observed [4]. In this study, the results for the glucose-mediated insulin release from INS-1 cells are presented in Figure 4. Substimulatory glucose (2.8 mM, BIS) showed less insulin release (0.99 µIU/mg of protein/h) than that (1.99 µIU/mg of protein/h) at high glucose levels (16.7 mM, GSIS).  Figure 4A). The rate of basic insulin secretion (BIS) was only slightly changed relative to the rate found in the glucose stimulated insulin secretion (GSIS). In a high-glucose environment, there was a significant increase of insulin secretion in 200 µg/mL compared to the other two concentrations. Although both compound 1 and PhGs group stimulated insulin secretion of INS-1 cells, a significant difference in the rates of the two groups were observed at the same dose (200 µg/mL) of 16.7 mM glucose. These results are in agreement with those expected from hypoglycemic agents. The effects of phenylethanoid glycoside on insulin secretion have previously been reported with type 2 diabetic db/db mice [32]. The results demonstrated that the investigation could be a starting point for an interesting investigation about the potential antidiabetic activity of the new compound. In early studies, the hypoglycaemic chemicals of Stevia rebaudiana Bertoni have been reported to be steviol and polyphenols, known as secretagogues and facilitate insulin release, protecting pancrease, etc. [33][34][35]. This paper demostrated that phenylethanoid glycosides could directly stimulate insulin secretion dose dependently. Those discoveries are meaningful for the investigation of hypoglycemic active component group of Stevia.

Material and Chemicals
Stevia were collected from Ganzhou Julong High-tech Industrial Co., Ltd. in Jiangxi province, China, and were naturally dried at room temperature. Samples were identified by Assistant Prof.

Isolation and Purification of Compounds 1-5
The phenolic extracts of Stevia were enriched from the plant as described previously [36]. Compounds 1-5 were isolated from the phenolic extracts. In detail, the phenolic extracts were dispersed with H 2 O and then subjected to macroporous resin Diaion HP-20 eluted with a step gradient of H 2 O/MeOH (1:0→0:1) to obtain eight fractions (Fr. A-Fr. H). Fr. B (55.7 g) was separated by a Sephadex LH-20 column (5 × 80 cm), eluted with MeOH, and then purified with a preparative high-performance liquid chromatography, which was performed on a CXTH 3000 chromatograph (Beijing Chuangxintongheng Co., Ltd. Beijing, China) equipped with UVD detector. A YMC C 18 column (250 × 20 mm, 5 µm) was used at a flow rate of 8 mL/min to give compound 2 (30 mg), 3

Structure Validation and Property Prediction
3.3.1. Identification of the Isolated Compounds 1 H and 13 C NMR spectra, as well as a DEPT spectra, were recorded in D 2 O with Bruker DRX-500 spectrometers operating at 500 MHz for 1 H, and 125 MHz for 13 C, respectively. Coupling constants were expressed in Hertz (Hz) and chemical shifts were given on a (ppm) scale with tetramethyl silane as internal standard. Positive ion ESI-MS and HR ESI-MS were recorded on an AutoSpec 3000 spectrometer (VG, Manchester, UK) with glycerol as the matrix. UV spectra were obtained on a UV 210A Shimadzu spectrometer. ESI-MS measurements were done by positive ionization with the capillary temperature set at 200 • C, Electrospray voltage 4.5 kV/cm; nitrogen was used as nebulizer gas at 20 arb. Collision-induced dissociation was optimized for each compound around 26-30%. The MS was recorded in the range of 80 to 1000 m/z.

Hydrolysis of Phenylethanoid Glycoside
The phenylethanoid glycoside (5 mg) was heated in 4 N TFA (trifluoroacetic acid, aqueous solution, 3 mL) at 95 • C for 4 h. The reaction mixture was extracted with CHCl 3 (3 × 3 mL). Each remaining aqueous layer was concentrated to dryness to yield a residue that was subsequently dissolved in pyridine (2 mL), and L-cysteine methyl ester hydrochloride (2 mg) was added to the solution. The mixture was heated at 60 • C for 2 h, chlorotrimethylsilane (0.3 mL) was added, and the mixture was further heated at 60 • C for 2 h. After the reactions had been performed, the supernatant was analyzed by GC, which was performed on an Agilent Technology 7890 (Agilent Technologies, Palo Alto, CA, USA). The conditions were as follows: The column temperature was maintained at 80 • C for 5 min, then increased from 80 to 280 • C at a rate of 25 • C/min, and finally maintained at 280 • C for 5 min; the carrier gas was N 2 (1.4 mL/min); the split ratio was 1/20; the injection temperature was 250 • C; and the injection volume was 1 µL. The absolute configurations of the monosaccharides were confirmed to be all D-glu and L-ara by comparison of the retention times of the monosaccharide derivatives with those of the standard samples D-glu (13.489 min) and L-ara (17.376 min).

ADMET Prediction
In this work, both ADMET (absorption, distribution, metabolism, excretion, and toxicity) and TOPKAT tests were performed using Discovery Studio ® 4.0 (Accelrys, San Diego, CA, USA) [37,38]. The Parameter values from six parts including aqueous solubility, human intestinal absorption (HIA), hepatotoxicity, blood brain barrier (BBB) penetration, cytochrome P450 2D6 inhibition, and plasma protein binding (PPB) were evaluated through ADMET descriptors algorithm [39]. In toxicity prediction, DS TOPKAT was used to evaluate toxicities under small molecules. Compound 1 was selected in models section containing screening for developmental toxicity potentials (DTP), rodent carcinogenicity, rat chronic oral lowest observed adverse effect level, rat oral LD 50, Ames mutagenicity, skin sensitization (GPMT), skin irritancy, and ocular irritancy. Ultimately, all the parameters were analyzed in order to predict the new compound's properties.
The potential cellular toxicity of compound 1 on INS-1 cells was assessed with the MTT method. It is based on the conversion of MTT into formazan crystals by living cells. In this paper, INS-1 (1 × 10 5 cells per well) were cultured in a 96-well culture plate for 24 h at 37 • C in atmosphere of 5% CO 2 . Then, different concentrations of compound 1 were added for culture another 48 h. In parallel, different concentrations of phenylethanoid glycosides (PhGs), which contain compound 1 and compound 2, were added. At the end of the treatment, supernatants were removed, a 5 mg/mL MTT was added, and incubated for an additional 4 h. The purple formazan crystals that developed from tetrazolium within the cells by the action of mitochondrial succinate dehydrogenase were extracted into DMSO. The absorbance was measured at 490 nm using a Multiskan Go microplate spectrophotometer (Thermo Fisher scientific, Waltham, MA, USA). All measurements were performed three times.

Insulin Secretion Assay
Insulin secretion by the INS-1 cells was studied with a slight modification of the method described before [40]. In brief, 2 × 10 5 of INS-1 cells per well were grown in RPMI 1640 medium, which was supplemented with 10% (v/v) fetal calf serum, penicillin (100 U/mL), and streptomycin (0.1 mg/mL), at 37 • C with 5% CO 2 in a humidified atmosphere. Prior to the experiment, the cells were washed three times with phosphate-buffered saline and then incubated in Krebs-Ringer bicarbonate HEPES buffer (KRBH), which contained 115 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 24 mM NaHCO 3 , 2.5mM CaCl 2 , and 10 mM HEPES. In order to measure insulin secretion, half-confluent cells in a 12-well cell plate were incubated for 2 h at 37 • C in the KRBH buffer. After that, different compounds with 2.8 mM (substimulatory concentration) or 16.7 mM (high concentration) glucose were added in the treatment groups for another 1 h; PhGs was added at various concentrations ranging from 100 to 400 µg/mL. Control experiments (with no compounds) were performed using KRBH buffer containing 2.8 mM and 16.7 mM glucose. Both medium and cell debris were separated, collected, and centrifuged at 4 • C, 700 g, 5 min. Total proteins per well were measured through cell debris to corrected insulin. Insulin secretion was assessed by radio-immunoassay (RIA) as the concentration of insulin in both the culture medium and the cell debris.

Statistical Analysis
Values are given as the mean ± SEM, and n values refer to the number of independent experiments. In the case of representative experiments, values are the mean ± SD. The significance of differences was assessed by t test for unpaired data.

Conclusions
A new phenylethanoid glycoside, 4-hydroxyphenyl ethyl-8-O-[αl-arabinopyranosyl-(1→6)]β-d-glucopyranoside, and four phenylethanolyl glycosides were isolated from the leaves of Stevia. The new compound 1 showed moderate properties for its low toxicity on both DS prediction and cytotoxicity assay. In addition, a significant stimulatory effect was observed on the INS-1 cell line. In summary, this is the first report that describes a natural, novel phenylethanoid glycoside in Stevia. The research is a starting point for further investigation into the potential anti-diabetic activity of steviophethanoside, as well as the investigation of hypoglycemic active component group of Stevia.

Patents
Ren-bing Shi has a pending patent application "Preparation, activity, application and quality control of a new compound", CN 104447900 A.