Activation of a Sweet Taste Receptor by Oleanane-Type Glycosides from Wisteria sinensis

The phytochemical study of Wisteria sinensis (Sims) DC. (Fabaceae), commonly known as the Chinese Wisteria, led to the isolation of seven oleanane-type glycosides from an aqueous-ethanolic extract of the roots. Among the seven isolated saponins, two have never been reported before: 3-O-α-L-rhamnopyranosyl-(1→2)-β-D-glucopyranosyl-(1→2)-β-D-glucuronopyranosyl-22-O-acetylolean-12-ene-3β,16β,22β,30-tetrol, and 3-O-β-D-xylopyranosyl-(1→2)-β-D-glucuronopyranosylwistariasapogenol A. Based on the close structures between the saponins from W. sinensis, and the glycyrrhizin from licorice, the stimulation of the sweet taste receptor TAS1R2/TAS1R3 by these glycosides was evaluated.


Introduction
Since the end of the 20th century, obesity has been one of the biggest health problems worldwide and has been linked in the long term to many different diseases such as type 2 diabetes, cardiovascular disease, hypertension, metabolic syndrome, and dyslipidemia [1,2]. However, reducing sugar intake can be very difficult, knowing that people's taste preference for sweetness is innate [3]. Thus, to limit the prevalence of diseases linked to this excessive consumption of sugar, the search for natural non-caloric compounds with a sweet taste and the development of artificial sweeteners has increased [4]. Some natural triterpene glycosides activate sweet taste receptors, such as glycyrrhizin from licorice (Glycyrrhiza glabra L., Fabaceae), which stimulates the heterodimer TAS1R2/TAS1R3 [5,6]. The Fabaceae family is one of the most important families of Angiosperms, being the third largest of this group with 727 genera and about 19,325 species [7]. From a chemotaxonomic point of view, the two genera Wisteria Nutt. and Glycyrrhiza L. belong to the Faboideae subfamily, so they are phylogenetically very close [8]. Moreover, oleanane-type glycosides isolated from Wisteria species like W. frutescens [9], W. floribunda cultivars [10], and W. brachybotrys [11], showed some structural similarities with glycyrrhizin [12]: a β-carboxyl group at position C-30, a ketone function, and a 3-O-β-D-glucuronopyranosyl moiety.

Results and Discussion
The saponins 1-7 were isolated from an aqueous-ethanolic extract of the roots of W. sinensis by various solid/liquid chromatographic methods, vacuum liquid chromatography (VLC), medium pressure liquid chromatography (MPLC), on normal and reverse phase RP-18 silica gel, and size exclusion chromatography on Sephadex LH-20.
The mass spectrum of compound 1 in HRESIMS (positive mode), reveals a quasimolecular ion at m/z 1023.5051 [M + Na] + , in agreement with the molecular formula C50H80O20Na. This suggests a molecular mass of 1000 g/mol.
The structure of the aglycone was determined using 1D and 2D NMR spectra, mainly COSY, TOCSY, ROESY, HSQC and HMBC ( Table 1)

Results and Discussion
The saponins 1-7 were isolated from an aqueous-ethanolic extract of the roots of W. sinensis by various solid/liquid chromatographic methods, vacuum liquid chromatography (VLC), medium pressure liquid chromatography (MPLC), on normal and reverse phase RP-18 silica gel, and size exclusion chromatography on Sephadex LH-20.
The mass spectrum of compound 1 in HRESIMS (positive mode), reveals a quasimolecular ion at m/z 1023.5051 [M + Na] + , in agreement with the molecular formula C 50 H 80 O 20 Na. This suggests a molecular mass of 1000 g/mol.
In the osidic part of compound 1, the HSQC spectrum showed three anomeric signals at δH 4.42 (1H, d, J = 7.6 Hz)/δC 104.3, δH 4.88 (1H, d, J = 7.6 Hz)/δC 100.7 and δH 5.19 (1H, br s)/δC 100.6. The ring protons of the monosaccharide residues were assigned mainly by COSY, TOCSY, HSQC, and HMBC experiments, which allowed the identification of one glucuronopyranosyl (GlcA), one glucopyranosyl (Glc), and one rhamnopyranosyl (Rha) units ( Table 1). The large 3 JH-1,H-2 values in the 1 H NMR spectrum of glucuronic acid and glucose in their pyranose form (7.6 Hz) indicated their β anomeric orientation. The large 1 JH-1,C-1 value of the Rha (167 Hz) confirmed that the anomeric proton was equatorial (αpyranoid anomeric form). The absolute configurations of the sugars were determined to be D for GlcA and Glc, and L for Rha (Experimental section). The same protocol was used for the identification of the monosaccharides of compound 2.
The HMBC correlations at   In the osidic part of compound 1, the HSQC spectrum showed three anomeric signals at δ H 4.42 (1H, d, J = 7.6 Hz)/δ C 104.3, δ H 4.88 (1H, d, J = 7.6 Hz)/δ C 100.7 and δ H 5.19 (1H, br s)/δ C 100.6. The ring protons of the monosaccharide residues were assigned mainly by COSY, TOCSY, HSQC, and HMBC experiments, which allowed the identification of one glucuronopyranosyl (GlcA), one glucopyranosyl (Glc), and one rhamnopyranosyl (Rha) units ( Table 1). The large 3 J H-1,H-2 values in the 1 H NMR spectrum of glucuronic acid and glucose in their pyranose form (7.6 Hz) indicated their β anomeric orientation. The large 1 J H-1,C-1 value of the Rha (167 Hz) confirmed that the anomeric proton was equatorial (α-pyranoid anomeric form). The absolute configurations of the sugars were determined to be D for GlcA and Glc, and L for Rha (Experimental section). The same protocol was used for the identification of the monosaccharides of compound 2.
The sweet taste properties of the saponins with the higher amounts after purification, compounds 1-3, were evaluated using the stimulation of the human taste heterodimer receptor TAS1R2/TAS1R3, with sucralose as reference (EC 50 = 16 ± 2 µg/mL). They were applied on a cell-based heterologous expression system and compared to glycyrrhizin. Glycosides 1-3 shared a common oleanane-type aglycone with a primary alcoholic function at the 30-position, a 3-O-β-D-glucuronopyranosyl linkage. However, only saponins 2 and 3 activated the sweet taste receptor with EC 50 values at 28 ± 2 µg/mL and 29 ± 7 µg/mL, respectively, both in the same range as glycyrrhizin (EC 50 = 34 ± 3 µg/mL) (Figure 3). Comparing compound 1 with compounds 2 and 3, compound 1 possesses a 16β-OH, a  22β-O-acetyl, and a free 24β-CH 3 group, instead of a 22-ketone, a free 16-CH 2 , and a 24β-CH 2 OH function in compounds 2 and 3. Glycosides 1-3 shared a common oleanane-type aglycone with a primary alcoholic function at the 30-position, a 3-O-β-D-glucuronopyranosyl linkage. However, only saponins 2 and 3 activated the sweet taste receptor with EC50 values at 28 ± 2 μg/mL and 29 ± 7 μg/mL, respectively, both in the same range as glycyrrhizin (EC50 = 34 ± 3 μg/mL) (Figure 3). Com- paring compound 1 with compounds 2 and 3, compound 1 possesses a 16β-OH, a 22β-Oacetyl, and a free 24β-CH3 group, instead of a 22-ketone, a free 16-CH2, and a 24β-CH2OH function in compounds 2 and 3. Structurally, compounds 2 and 3 shared with glycyrrhizin a 3-O-β-D-glucuronopyranosyl linkage, a ketone group, and an oxidated 30β-CH3 (Figure 1). Previous studies highlighted the key role of the 3-O-β-D-glucuronopyranosyl group, the ketone function, and the oxidation of the 24β-CH3 and the 30β-CH3 groups [12]. These results are very promising since saponins 2 and 3 can activate the sweet-taste receptor TAS1R2/TAS1R3 with EC50 values in the micromolar range. These values are close to those measured for sucralose, a sweetener widely used by the food industry, and the glycyrrhizin highly appreciated for its sweetness with a typical licorice taste. However, saponins are known for their toxicity on cell membranes, so, before any further investigations, the toxicity of these molecules has to be evaluated.

General Experimental Procedures
NMR spectra were recorded on a Varian INOVA 600 MHz spectrometer (Agilent Technologies) equipped with 3 mm triple resonance inverse and 3 mm dual broadband probe heads. Spectra were recorded in methanol-d4, and all spectra were recorded at T = Structurally, compounds 2 and 3 shared with glycyrrhizin a 3-O-β-D-glucuronopyranosyl linkage, a ketone group, and an oxidated 30β-CH 3 (Figure 1). Previous studies highlighted the key role of the 3-O-β-D-glucuronopyranosyl group, the ketone function, and the oxidation of the 24β-CH 3 and the 30β-CH 3 groups [12]. These results are very promising since saponins 2 and 3 can activate the sweet-taste receptor TAS1R2/TAS1R3 with EC 50 values in the micromolar range. These values are close to those measured for sucralose, a sweetener widely used by the food industry, and the glycyrrhizin highly appreciated for its sweetness with a typical licorice taste. However, saponins are known for their toxicity on cell membranes, so, before any further investigations, the toxicity of these molecules has to be evaluated.

General Experimental Procedures
NMR spectra were recorded on a Varian INOVA 600 MHz spectrometer (Agilent Technologies) equipped with 3 mm triple resonance inverse and 3 mm dual broadband probe heads. Spectra were recorded in methanol-d 4 , and all spectra were recorded at T = 308.15 K. Pulse sequences were taken from the Varian pulse sequence library (gCOSY; gHSQCAD and gHMBCAD with adiabatic pulses CRISIS-HSQC and CRISIS-HMBC). TOCSY spectra were acquired using DIPSI spin-lock and 150 ms mixing time. Mixing time in ROESY experiments was 300 ms. Chemical shifts were reported in δ units and coupling constants (J) in Hz. HR-ESIMS (positive-ion mode) and ESIMS (positive-and negative-ion mode) were carried out on a Bruker micrOTOF mass spectrometer. A MARS 6 microwave apparatus (CEM) was used for the extractions. Isolations of the compounds were carried out using column chromatography (CC) with Sephadex LH-20 (550 mm × 20 mm, GE Healthcare Bio-Sciences AB), and vacuum liquid chromatography (VLC) with reversed-phase RP-18 silica gel 50:1). The HPLC was performed on an Agilent 1260 instrument, equipped with a degasser, a quaternary pump, a sample changer, and a UV detector (210 nm). The chromatographic separation for the analytical part was carried out on a C18 column (250 mm × 4.6 mm internal diameter, 5 µm; Phenomenex LUNA) at room temperature and protected by a guard column. The mobile phase consists of (A) 0.01% (v/v) aqueous trifluoroacetic acid and (B) acetonitrile delivered at 1 mL/min going from 30% to 80% B in 30 min. The injection volume was 10 µL.

Plant Material
Wisteria sinensis was purchased from Botanic ® (Quetigny, France) in September 2019, and a sample was deposited in the herbarium at the Laboratory of Pharmacognosy, Université de Bourgogne Franche-Comté, Dijon, France, under the number N • 2019/09/06.

Extraction and Isolation
Microwave-assisted extraction of 47.07 g of dried pulverized roots was carried out three times, with a mixture of EtOH/H 2 O (75:35; 500 mL). The microwave apparatus was programmed to reach 60 • C in 10 min, and then maintain this temperature for another 30 min with moderate agitation. After evaporation of the solvent under vacuum, the resulting extract (6.65 g) was submitted to VLC (RP-18

Acid Hydrolysis and Absolute Configuration Determination
An aliquot (150 mg) of a rich saponin fraction was hydrolyzed with 2N aqueous CF 3 COOH (25 mL) for 3 h at 95 • C. After extraction with CH 2 Cl 2 (3 × 15 mL), the aqueous layer was evaporated to dryness with H 2 O until neutral to give the sugar residue (55 mg). Glucuronic acid, glucose, xylose and rhamnose were identified by comparison with authentic samples by TLC using CH 3 COOEt/CH 3 COOH/CH 3

Bioactivity Assay
We investigated the ability of purified compounds to activate the sweet taste receptor using heterologous expression of TAS1R2/TAS1R3 and functional calcium imaging. As previously described, the cDNAs for TAS1Rs and the plasmid pGP-CMV-GCaMP6S (Addgene #40753) coding for a calcium biosensor, were transiently transfected into HEK293T cells stably expressing the chimeric G-protein subunit Gα16gust44, using Fugene HD (Promega) [6]. Cells transfected only with calcium indicator vector served as negative control. Prior to the stimulation, the transfected cells were washed with C1 buffer (130 mM NaCl, 5 mM KCl, 10 mM Hepes, 2 mM CaCl 2 , 5 mM sodium pyruvate, pH 7.4). Then, we monitored calcium mobilization following sweet taste receptor activation after automatic injection of test substances with a Molecular Devices FlexStation 3 system. Compounds 2 and 3 were dissolved first at 10 mg/mL in DMSO with good solubility. Further dilutions were prepared in C1 solution. The compounds 2 and 3 were evaluated up to a maximum range of 50 µg/mL, because they elicited non-specific calcium responses in mock-transfected cells at concentration above 100µg/mL. Data were collected from at least three independent experiments carried out in duplicate. The concentration-response curves were obtained after correction of calcium signals for the response of mock transfected cells and normalization to the fluorescence of cells prior to the stimulation. EC 50 values were calculated using a four-parameter logistic nonlinear regression with equation [f(x) = min + (max-min)/(1 + (x/EC 50 ) nH )] with curves fitting of Sigma Plot software.