Rubus suavissimus S. Lee belongs to Rubus, a large genus of flowering plants in the rose family, Rosaceae , subfamily Rosoideae. Raspberries, blackberries, and dewberries are common, widely distributed members of this genus. R. suavissimus is a perennial shrub grows widely grown in Guang-xi and Guang-dong, China [1]. The leaves of R. suavissimus are used to make beverage leaf tea by the local residents because of its intensely sweet flavor. It is generally known as tiancha in Chinese or Chinese sweet tea. Previous phytochemical studies of this plant mainly showed the presence of diterpene and triterpene glycosides as well as phenolic compounds [2-4]. The major constituent of this plant is the sweet diterpenoid glycoside rubusoside with an aglycone moiety belongs to the class of the diterpene, ent-13-hydroxykaur-16-en-19-oic acid, known as steviol [5]. As a part of our research to discover natural sweeteners, we have recently reported several ditepene glycosides from S. rebaudiana [6-8] and triterpene glycosides from Siraitia grosvenorii [9].

In our continuing research to isolate natural compounds from various sweet taste plants collected from all over the World, we have isolated three diterpene glycosides rubusoside (1), suavioside-A (2) and sugeroside (3) as well as the phenolic and lignin glycosides namely quercetrin (4) and arctiin (5) respectively from the polar fraction of the aqueous extract of the leaves of R. suavissimus obtained from Chengdu Biopurify Phytochemicals Limited, China. This paper describes the isolation and structure elucidation of the isolated glycosides 1-5 (Figure 1) on the basis of extensive spectroscopic and chemical studies as well as in comparison of their physical and spectral properties reported from the literature. Also, we are herewith reporting the sweetness recognition threshold (SRT) and sweetness enhancement effect (SEE) of the predominant constituent of the plant, rubusoside (1).

Figure 1: Structures of 1-5 and other compounds.

Compound 1 was isolated as a white powder and its molecular formula has been deduced as C₃₂H₅₀O₁₃ on the basis of its HRMS which showed [M+NH₄]⁺ and [M+Na]⁺ ions at m/z 660.3590 and 665.3136 respectively, and this was supported by the ¹³C NMR spectral data. The ¹H NMR spectrum of 1 showed the presence of two methyl singlets at δ 1.24 and 1.26, two olefinic protons as singlets at δ 5.03 and 5.56 of an exocyclic double bond, nine methylene and two methine protons between δ 0.79-2.72 characteristic for the ent-kaurane diterpenoids isolated earlier from the plants belongs to the genus Stevia and Rubus [6-8,10,11]. The basic skeleton of kaurane diterpenoids was supported by COSY (H-1/H-2; H-2/H-3; H-5/H-6; H-6/H-7; H-9/H-11; H-11/H-12) and HMBC (H-1/C-2, C-10; H-3/C-1, C-2, C-4, C-5, C-18, C-19; H-5/C-4, C-6, C-7, C-9, C-10, C-18, C-19, C-20; H-9/C-8, C-10, C-11, C-12, C-14, C-15; H-14/C-8, C-9, C-13, C-15, C-16 and H-17/C-13, C-15, C-16) correlations. The fragment ions observed at m/z 481 and 319 in the positive ESI mode MS/MS spectrum of 1 indicating the presence of two hexose sugars in its structure. This was further supported by the ¹H NMR spectrum of 1 which showed the presence of two anomeric protons at δ 5.13, and 6.15. Enzymatic hydrolysis of 1 furnished an aglycone which was identified as steviol (6) by comparison of ¹H NMR [10] spectral data reported in the literature and co-TLC with standard compound. Acid hydrolysis of 1 afforded D-glucose that was identified by preparing the corresponding thiocarbamoyl-thiazolidine carboxylate derivative with L-cysteine methyl ester and O-tolyl isothiocyanate, and in comparison of its retention time with the standard sugars as described in the literature comparison [12]. The large coupling constants observed for the two anomeric protons at δ 5.13 (J=7.8 Hz), and 6.15 (d, J=8.5 Hz) suggested their β-orientation as reported for steviol glycosides [6-8].

The ¹³C NMR values for all the carbons were assigned on the basis of HSQC and HMBC correlations (Table 1 and Table 2). From COSY, and HMBC correlations as shown in Figure 2, 1was found to have a steviol aglycone moiety having a β-D-glucopyranosyl unit attached C-13 hydroxyl and another β-D-glucopyranosyl moiety in the form of an ester at C-19. This was supported by the HMBC correlations: H-1’/C-19, C-2’, C-3’; and H-1’’/C-13, C-2’’, C-3’’. A close comparison of the NMR spectral data of 1 with the reported literature values for rubusoside confirmed its structure.

Figure 2: Key COSY and HMBC correlations of 1.

Table 1: ¹H NMR chemical shift values for 1–3 isolated from Rubus suavissimus recorded in C₅D₅N^a-c.

Table 2: ¹³C NMR chemical shift values for 1–3 isolated from Rubus suavissimus recorded in C₅D₅N ^a-b.

The molecular formula of compound 2 was established as C₂₆H₄₄O₈ from its HRMS spectral data which showed [M+NH₄]⁺ and [M+Na]⁺ ions at m/z 502.3327 and 507.2924 respectively. The ¹H NMR spectrum of 2 showed the presence of three methyl singlets at δ 0.87, 1.02, and 1.17, eight methylene and two methine protons between δ 0.89-2.43, similar to 1. The ¹H NMR of 1 also showed the presence of signal at δ 4.12 as a triplet like with W_1/2=2.6 Hz and a pair of doublets corresponding to an oxymethylene at δ 3.91 (J=10.6 Hz), 4.49 (J=10.2 Hz) and an anomeric proton as a doublet at δ 5.05. Acid hydrolysis of 2 afforded D-glucose that was identified by preparing its corresponding thiocarbamoyl-thiazolidine carboxylate derivative as in 1, and the coupling constant observed for the anomeric proton J=8.2 Hz suggested the β-orientation of the D-glucosyl unit. The ¹³C NMR spectrum of 2 showed the presence of nine oxygenated carbons between δ 63.2 and 107.2 of which six were assigned to the β-Dglucopyranosyl unit, leaving the assignment of the other three carbons. From HSQC and HMBC spectral data, it was found that the other three oxygenated carbons of 2 as one oxymethylene, one oxymethine and a tertiary hydroxyl resonating at δ 76.0, 75.6 and 81.3 respectively in its ¹³C NMR spectrum. From the above spectral data and chemical studies, the structure was identified as an ent-kaurane diterpenenoid skeleton having a β-D-glucopyranosyl unit and three oxygenated carbons as mentioned above. A search from the literature indicated the presence of two compounds with the above functional groups namely ent-kaurane-3α, 16β, 17-triol-17-O-β-D-glucoside (suavioside A) and ent-kaurane- 3β, 16β, 17-triol-17-O-β-D-glucoside (iwayoside A, 7) isolated from R. suavissimus [13] and Artemisia iwayomogi [14] respectively. The key COSY and HMBC correlations as displayed in Figure 3 supported the basic skeleton of ent-kaurane-3, 16, 17-triol-17-O-β-D-glucoside. Since the spectral data for suavioside A and iwayoside A were reported in CD₃OD and C₅D₅N respectively and in order to compare the NMR values of 2 with the isolated compounds, its ¹H and ¹³C NMR were also acquired in both the solvents. A close comparison of the NMR spectral data and optical rotation of 2 confirmed the structure as suavioside A.

Figure 3: Key COSY and HMBC correlation of 2.

Compound 3 was also obtained as a white amorphous powder and its molecular formula was inferred as C₂₆H₄₂O₈ from its HRMS spectral data that showed [M+NH₄]⁺ and [M+Na]⁺ ions at m/z 500.3220 and 505.2768 respectively, and this molecular composition was supported by the ¹³C NMR spectral data. The ¹H NMR spectrum of 3 showed the presence three methyl singlets, eight methylene and two methine protons, two protons corresponding to an oxymethylene group and an anomeric proton; identical to 2. Acid hydrolysis confirmed the sugar and its configuration as D-glucose. A close comparison of the ¹H and ¹³C NMR chemical shift values of 2 and 3 together with the ESI-MS data which has 2 amu difference, suggested the presence of a carbonyl group in 3 at C-3 position in place of an oxymethine proton in 2. This was further supported by the presence of the peak observed in its ¹³C NMR spectrum at δ 217.3, and the absence of a triplet like signal for the oxymethine proton at C-3 position and its corresponding carbon in the respective proton and carbon NMR spectral data of 3. The above spectral and chemical data suggested its structure as 3, which was reported as sugeroside earlier from Ilex sugerokii var. brevipedunculata [15] and R. suavissimus [13].

Compound 4 was obtained as yellow amorphous powder and its molecular formula was established as C₂₁H₂₀O₁₁ from its HRMS spectral data that showed [M+H]⁺ and [M+Na]⁺ ions at m/z 449.1075 and m/z 471.0892 respectively; this was supported by the ¹³C NMR spectral data. The 1H NMR spectrum of 4 showed the presence of three meta coupled aromatic protons at δ 6.65 (J=1.6 Hz), 6.72 (J=1.8 Hz) and 8.03 (J=1.7 Hz); one ortho coupled aromatic proton at δ 7.31 (J=8.1 Hz); another ortho and meta coupled aromatic proton as doublet of doublets at δ 7.73 (J=7.5, 1.8 Hz); characteristic for the 3-substituted flavone. The ¹H NMR of 4 also showed the presence of an anomeric proton as a doublet at δ 6.29 suggesting a sugar residue in its structure which was identified as L-rhamnosyl moiety on the basis of acid hydrolysis and by preparing the corresponding thiocarbamoyl-thiazolidine carboxylate derivative with L-cysteine methyl ester and O-tolyl isothiocyanate, and in comparison of its retention time with the standard sugars as described for compounds 1-3. The presence of L-rhamnosyl moiety was further supported by the presence of the secondary methyl group as a doublet at δ 1.49 (J=6.0 Hz) as well as the absence of oxymethylene group at C-5 position of the sugar unit. The anomeric proton had a coupling constant of 1.5 Hz, similar to dulcosides A and B isolated from S.rebaudiana [16] confirming the α-orientation L-rhamnosyl moiety. The placement of the L-rhamnosyl moiety was identified at C-3 position on the basis of key HMBC correlations: H-2’/C-2, C-1’, C-3’ and H-1’’/C-2, C-3, C-2’’ (Figure 4). Thus, based on the above spectral data the structure of 4 was assigned as quercetin-3-O-α-L-rhamnoside (quercetrin) consistent to the reported literature values [17,18].

Figure 4: Key HMBC correlation of 4.

Compound 5 was obtained as a hard gum and its molecular formula was inferred as C₂₇H₃₄O₁₁ from its HRMS spectral data that showed [M+NH₄]⁺ and [M+Na]⁺ ions at m/z 552.2437 and m/z 557.1986 respectively. The 1H NMR spectrum of 5 showed the presence of six aromatic protons between δ 6.72-7.58; three methoxyl groups at δ 3.75, 3.78 and 3.80; two protons of an oxymethylene group at δ 3.93 and 4.20; two methylenes and two methines between δ 2.52-3.07; very similar to the lignan arctigenin. Acid hydrolysis of 5 furnished D-glucose and was identified as having β-orientation from the coupling constant of its anomeric proton appeared at δ 5.69 (J=7.2 Hz). From the above spectral data and chemical studies it was evident that the structure of 5 should contain a β-D-glucopyranosyl unit that has been attached to the aglycone moiety of arctigenin. From the HMBC correlations shown in Figure 5, the presence of β-D-glucopyranosyl unit was suggested at C-6’ position unambiguously as in arctiin; confirmed its structure completely [19,20].

Figure 5:Key HMBC correlations of 5.

This is the first report of the isolation of quercetrin and arctiin not only from the plant R. suavissimus but also from the genus Rubus. Further, the detailed NMR characterizations have not been studied on some of the isolated glycosides and herewith we have assigned the entire proton and carbon values on the basis of COSY, HSQC and HMBC spectral data as well as confirmed the sugars and their configuration by hydrolysis studies. Also, we are reporting the ¹H and ¹³C NMR data for all the isolated five compounds (1-5) in C₅D₅N in this article.

General

Melting points were measured using a SRS Optimelt MPA 100 instrument and are uncorrected. Optical rotations were recorded using a Rudolph Autopol V at 25 °C and NMR spectra were acquired on a Varian Unity Plus 600 MHz instrument using standard pulse sequences at ambient temperature. Chemical shifts are given in δ (ppm), and coupling constants are reported in Hz. HRMS data was generated with a Thermo LTQ Orbitrap Discovery mass spectrometer in the positive positive ion mode electrospray. Instrument was mass calibrated with a mixture of Ultramark 1621, MRFA [a peptide], and caffeine immediately prior to accurate mass measurements of the samples. Samples were diluted with water:acetonitrile:methanol (1:2:2) and prepared a stock solution of 50 ul concentration for each sample. Each sample (25 ul) was introduced via infusion using the onboard syringe pump at a flow injection rate of 120 ul/min. Low pressure chromatography was performed on a Biotage Flash system using a C-18 cartridge (40+ M, 35-70 µm). TLC was performed on Baker Si- C18F plates and identification of the spots on the TLC plate was carried out by spraying 10% H₂SO₄ in EtOH and heating the plate at about 80°C. Analytical HPLC for sugar analysis was carried out with a Waters 600E multisolvent delivery system using a Phenomenex Luna C₁₈ nonchiral (150 x 4.6 mm, 5 µm) column.

Plant material

The commercial sample consisting of the aqueous extract of the leaves of R. suavissimus was purchased from Chengdu Biopurify Phytochemicals, China. The plant material was identified by Professor Weiping He, Natural Plant Scientific Institute, Guangdong Ocean University, Guangxi, China and a voucher specimen is deposited at The Coca Cola Company, No. VSPC-3166-68.

Isolation

The aqueous extract of the leaves of R. suavissimus (10 g) was suspended in 100 ml water and extracted successively with n-hexane (3 x 100 ml), CH₂Cl₂ (3 x 100 ml) and n-BuOH (2 x 100 ml). The n-BuOH layer was concentrated under vacuum furnished a residue (2.5 g) which was purified on a Biotage flash chromatography system using C-18 (100 g) column (solvent system: gradient from 20-80 MeOH-water to 100% MeOH at 60 ml/min. detection at UV 210 nm) for 40 min. Fractions 13-20 (0.2 g) were combined and further subjected to repeated flash chromatography purification with gradient from 40-80% MeOHwater at 30 ml/min for 30 min afforded suavioside A (2, tR, 18.6 min, 10.2 mg). Fractions 21-23 (1.4 g) were combined and crystallized with MeOH furnished rubusoside (1, 1.1 g). Purification of the combined fractions 24-27 (0.12 g) using the gradient from 40-80% MeOH-water at 30 ml/min for 40 min furnished sugeroside (3, tR 16.4 min, 11.5 mg). Fractions 46-50 and 52-56 were combined to get residues 0.14 g and 0.11 g respectively, which on repeated purification using the gradient 60-90% MeOH-water at 30 ml/min for 40 min resulted quercetrin (4, tR 22.4 min, 6.8 mg), and arctiin (5, tR 19.6 min, 8.4 mg), respectively.

Identification

Rubusoside: White powder, mp 177-179°C [reported mp 178-181°C]; [α]_D ²⁵ -37.20 (c 1.0, MeOH) [reported [α]_D ¹⁸ -40.30 (c 0.8, MeOH)]; ¹H NMR (600 MHz, C₅D₅N, δ ppm) and ¹³C NMR (150 MHz, C₅D₅N, δ ppm) spectroscopic data see Table 1 and Table 2; HRMS m/z [M+NH₄ ⁺] calcd for C₃₂H₅₄O₁₃N: 660.3595; found 660.3590 and [M+Na⁺] calcd for C₃₂H₅₀O₁₃Na: 665.3159; found 665.3136.

Enzymatic hydrolysis of 1. A solution of 1 (500 µg) was dissolved in 5 ml of 0.1 M sodium acetate buffer, pH 4.5 and crude pectinase from Aspergillus niger (100 uL, Sigma-Aldrich, P2736) was added. The mixture was stirred at 50°C for 48 h. The product precipitated out during the reaction was filtered and then crystallized from methanol (MeOH). The resulting steviol (6, 73 µg) was identical to an authentic sample by co-TLC and ¹H NMR [10].

Suavioside A: White powder, [α]_D²⁵ +45.62 (c 0.15, C₅H₅N)[reported [α]_D²⁵ +47.21 (c 0.14, C₅H₅N)]; ¹H NMR (600 MHz, C₅H₅N, δ ppm) and ¹³C NMR (150 MHz, C₅D₅N, δ ppm) spectroscopic data see Table 1 and Table 2; HRMS m/z HRMS m/z [M+NH₄ ⁺] calcd for C₂₆H₄₈O₈N: 502.3380; found 502.3327 and [M+Na⁺] calcd for C₂₆H₄₄O₈Na: 507.2934; found 507.2924.

Sugeroside: White powder, [α]_D²⁵ -53.20 (c 0.35, C₅H₅N) [reported [_D²³ -55.60 (c 0.36, C₅H₅N)]; ¹H NMR (600 MHz, CD₅N, δ ppm) and ¹³C NMR (150 MHz, CD₅N, δ ppm) spectroscopic data see Table 1 and Table 2; HRMS m/z [M+NH₄⁺] calcd for C₂₆H₄₆O₈N: 500.3223; found 500.3220 [M+Na⁺] calcd for C₂₆H₄₂O₈Na: 505.2777; found 505.2768.

Determination of the configuration of sugars in 1-3: Each compound 1-3 (500 µg) was hydrolyzed with 0.5 M HCl (0.5 mL) for 1.5 h. After cooling, the mixture was diluted with 5 ml water, passed through an Amberlite IRA400 column and the eluate was lyophilized. The residue was dissolved in pyridine (0.25 mL) and heated with L-cysteine methyl ester HCl (2.5 mg) at 60ºC for 1.5 h, and then O-tolyl isothiocyanate (12.5 uL) was added to the mixture and heated at 60ºC for an additional 1.5 h. The reaction mixture was analyzed by HPLC: column Phenomenex Luna C₁₈, 150 x 4.6 mm (5 u); 25% acetonitrile-0.2% TFA water, 1 mL/min; UV detection at 250 nm. The sugar was identified as D-glucose in each experiment (tR, 12.26 to 12.42 min) [authentic samples, D-glucose (tR, 12.35) and L-glucose (tR, 11.12 min) [12].

Quercetrin : Yellow powder, mp 179.4-181.2°C [reported mp 180-182 °C]; ¹H NMR (600 MHz, C₅D₅N, δ ppm) and ¹³C NMR (150 MHz, C₅D₅N, δ ppm) spectroscopic data see Table 3; HRMS m/z [M+H⁺] calcd for C₂₁H₂₁O₁₁: 449.1084; found 449.1075 and [M+Na⁺] calcd for C₂₁H₂₀O₁₁Na: m/z 471.0903; found 471.0892.

Table 3: ¹H and ¹³C NMR chemical shift values for 4–5 isolated from Rubus suavissimus recorded in C₅D₅N^a-c.

Arctiin : Hard gum, [α]_D²⁵ -52.60 (c 0.003, MeOH) [reported [α]_D²⁵-55.30 (c 0.0033, MeOH)]; ¹H NMR (600 MHz, C₅D₅N, δ ppm) and ¹³C NMR (150 MHz, C₅D₅N, δ ppm) spectroscopic data see Table 3; HRMS m/z [M+NH₄ ⁺] calcd for C₂₆H₄₈O₈N: 552.2445; found 552.2437 and [M+Na⁺] calcd for C₂₇H₃₄O₁₁Na: m/z 557.1999; found 557.1986.

Determination of the configuration of sugars in 4-5: Each compound 4 and 5 (1 mg) was hydrolyzed with 2 M HCl (2 mL) for 18 h. After cooling, the mixture was diluted with 5 ml water, passed through an Amberlite IRA400 column, lyophilized and the residues obtained were converted to the corresponding thiocarbamoyl-thiazolidine carboxylate derivative with L-cysteine methyl ester and O-tolyl isothiocyanate as described above. The sugars were identified as L-rhamnose (tR, 21.32 min) and D-glucose (tR, 12.21 min) respectively from the hydrolysis experiments with 4 and 5 [authentic samples: D- rhamnose (tR, 11.73 min) and L- rhamnose (tR, 21.64 min); D-glucose (tR, 12.35) and L-glucose (tR, 11.12 min)] [12].

Sweetness Recognition Threshold (SRT) measurement of 1: The sweetness recognition threshold of 1 was measured by three experienced panelists in duplicated runs. All solutions were made in carbon-treated water and used at room temperature. Each of three subjects were asked to isosweet the random, different order blind samples against standard sugar solutions at 0.5%, 1.0% and 1.5% (w/v). The subjects were asked to focus on second sip of each sample and to rinse their mouths with water in between samples. The blind results indicated both duplicated runs yielded consistent results among samples at three different concentrations of 35, 50 and 65 ppm of 1 and the overall % sweetness equivalence (SE) averages were 0.59, 0.92 and 1.06, respectively. As a result, the SRT at 0.75% SE of sugar in water is estimated to be 42 ppm for 1. Similarly, the SRT of 1 in carbonated lemon-lime (LL) soda prototypes was evaluated and determined to be 150 ppm.

Sweetness Enhancement Effect (SEE) measurement of 1: In order to find the SEE, test solutions of glucose, fructose and sucrose were prepared equivalent to 6% SE with carbon-treated water at room temperature at SRT of 1 in each carbohydrate solution. The same 6% SE solutions with the above three carbohydrates were prepared without 1. Experimental results indicated that 1 was found to have a slight, positive SEE (ca. 1% SE more) in glucose and fructose solutions whereas a relatively larger SEE in sucrose by ca. 2% SE compared to the solutions of glucose, fructose and sucrose without 1. Likewise, the SEE in carbonated lemon-lime (LL) soda prototypes was evaluated for sensory data in a trained, descriptive analysis of 10 panelists by preparing 8% SE solution using high fructose corn syrup (HFCS), with and without 1 as described above. Each assessor evaluated all the beverage products in triplicate runs with 8 min break time between testing samples. Unsalted cracker, 0.75% saline solution and mineral grade water was used as a mouth rinse and refresher before testing each sample. Sensory data revealed that the 8% HFCS containing 1 had a SE of 9.2% which is 1.15 times as sweet as the prototype without 1.

Five glycosides including three diterpene, a phenolic and a lignan were isolated from the commercial extract obtained from the leaves of R. suavissimus obtained from Chengdu Biopurify Phytochemicals Limited, China. The structures of all the isolated new compounds were identified as rubusoside (1), suavioside-A (2), sugeroside (3), quercetrin (4) and arctiin (5) on the basis of spectroscopic and chemical studies as well as by comparing their physical properties reported in the literature. This is the first report of the isolation of quercetrin and arctiin from R. suavissimus in nature. The complete ¹H and ¹³C NMR spectral assignments of all the isolated compounds are reported herewith in CD₅N₅ based on COSY, HSQC, HMBC, and MS/MS spectroscopic data as well as chemical studies. The sensory evaluation results demonstrated that the SRT of 1 in water and LL soda matrixes are 50 and 150 ppm respectively. Also, 1 showed 1% SEE in glucose and fructose, and 2% SEE in sucrose in aqueous solutions; whereas it showed 1.15 times SEE in LL soda prototypes at its SRT.

We wish to thank Chengdu Biopurify Phytochemicals Limited, China for providing the Rubus suavissimus aqueous extract.