Abstract
Bee pollen is a health food with a wide range of nutritional and therapeutic properties. However, the bioactive compounds of bee pollen have not been extensively revealed due to low efficacy in separation. High-speed counter-current chromatography (HSCCC) and solvent extraction were applied to separate tyrosinase inhibitors from camellia pollen in this study. The camellia pollen extracts prepared with petroleum ether, ethyl acetate, and n-BuOH have tyrosinase inhibitory activity. Acidic hydrolysis could promote the tyrosinase inhibitory activity of crude sample. Three fractions with tyrosinase inhibitory activity were separated from the hydrolysate by a one-step HSCCC procedure. Among the fractions, two chemicals were sufficiently purified and identified to be levulinic acid (LA) and 5-hydroxymethylfurfural (5-HMF). The recovery was 0.80 g kg−1 pollen for LA and 1.75 g kg−1 pollen for 5-HMF; and their purity was all over 98%. The study demonstrates that HSCCC method is powerful for preparative separation of tyrosinase inhibitors from camellia pollen.
Introduction
Bee pollen is a well-known health food in China, which contains all nutritionally essential substances that are necessary for plant growth and development [1, 2]. Due to its nutritional and therapeutic properties, bee pollen has been used for many years in traditional medicine and supplementary nutrient. Previous studies have shown that bee pollen has a beneficial effect on prostate problems [3]. Others showed that pollen could improve body immunity and the cardiovascular and digestive systems [4, 5]. In addition, bee pollen is closely related to various human health-promoting effects, such as antimicrobial, antioxidant, anti-inflammatory, tyrosinase inhibitory activity, and chemoprotective activities [6–8]. Despite the popularity of usage, relevant investigation of bioactive components of bee pollen is still very limited due to the fact that most of bioactive chemicals are present at low concentration and difficult to be separated from the major constituents.
To date, normal-phase liquid chromatography [9] and solvent liquid–liquid extraction coupled with semi-preparative high-performance liquid chromatography [10] have been developed for separation and purification of bioactive components. However, the processes of these separation methods are time-consuming and result in large amounts of organic solvent wastage, as well as often require multiple steps. Therefore, alternative method has gained growing importance. High-speed counter-current chromatography (HSCCC) is a continuous liquid–liquid partition chromatography with no solid support. Unlike traditional separation methods, it has been shown that HSCCC can eliminate the irreversible adsorption and denaturation of the compounds on the solid stationary phase [11, 12]. What is more, HSCCC offers many other advantages such as low solvent consumption, rapid separation, high recovery, and ease of scaling up [13, 14]. Due to the abovementioned advantages, more and more attempts have been made to use HSCCC to separate active components from natural products [15–17].
Tyrosinase (EC: 1.14.18.1), known as polyphenol oxidase, is widely distributed in plants, animals, and microorganisms. Tyrosinase catalyzes both the hydroxylation of tyrosine to 3,4-dihydroxyphenylalanine (l-DOPA) and the oxidation of l-DOPA to dopaquinone [18, 19]. In human body, the overexpression of tyrosinase can lead to excessive accumulation of melanin, causing skin disorders such as age spots, freckles, melasma, and malignant melanoma [20, 21]. Besides that, tyrosinase involves in the browning of fruits and vegetables; this unfavorable darkening from enzyme oxidation generally results in a less attractive appearance and a loss in nutritional quality [22, 23]. Therefore, tyrosinase inhibitors have been established as important constituents in cosmetic for whitening and depigmentation and have potential uses as food preservatives [24, 25]. Hence, the isolation and purification of the tyrosinase inhibitors from nature products are of great interests.
Although pollen has been reported to have the tyrosinase inhibitory activity [8], the compound responsible for this activity has not been illustrated. The aim of the present study was to separate tyrosinase inhibitor from pollen extract using HSCCC. The specific content included: (1) solvent extraction of tyrosinase inhibitor; (2) HSCCC separation of tyrosinase inhibitor; (3) analysis of IC50 of the tyrosinase inhibition; (4) identification of the structure of tyrosinase inhibitor; and (5) analysis of the purity and recovery of tyrosinase inhibitor.
Experimental
Reagents and Materials
Organic solvents, including ethanol, petroleum ether (60–90 °C), ethyl acetate, n-butanol, and hydrochloric acid were all analytical grade (Sinopharm Chemical Reagent, Shanghai, China). Chromatographic grade acetonitrile (Tedia Company Inc., Fairfield, USA) was used for high-performance liquid chromatography (HPLC) analysis. Reverse osmosis Milli-Q water (Millipore Bedford, MA, USA) was used for all solutions and dilutions. Arbutin, standards LA (98% purity), 5-hydroxymethylfurfural (5-HMF, 99% purity), l-DOPA, and tyrosinase (E.C.1.14.18.1) from mushroom were obtained from Sigma Chemical Co. (St. Louis, MO). The traditional bee pollen (camellia pollen) was purchased from Beijing Tong Ren Tang Co., Ltd. (Beijing, China).
Preparing Crude Extract
One kilogram of dried pollen was milled to power (about 40 mesh) and extracted with 10 L 95% ethanol for three times (8 h each extraction) at the temperature of 50 °C. The extracts were concentrated in a rotary evaporator at 50 °C under reduced pressure. The concentrated extract was resuspended in water and successively partitioned with petroleum ether, ethyl acetate, and n-butanol, followed by concentration in the rotary evaporator, respectively. The dried fractions were stored in a refrigerator (4 °C) until further experiments.
Hydrolyzing Tyrosinase Inhibition Extract
The concentrated ethyl acetate extract was subjected for acid hydrolysis using the method of a previous study with minor modifications [26]. Briefly, 10 g ethyl acetate extract was dissolved in 500 mL of ethanol and then mixed with 500 mL of 4 M hydrochloric acid, sealed, and incubated in a water bath at 90 °C for 90 min. After cooling at room temperature, the hydrolysate was evaporated in a rotary evaporator at 50 °C under vacuum. The dried extract was stored in a refrigerator (4 °C) for HSCCC separation.
Measuring Partition Coefficients (K) of Different Solvent Systems
The two-phase solvent system was investigated in the partition coefficients (K) of the target compound. About 5 mg hydrolysate was dropped into a 10 mL test tube containing 5 mL of each phases. The test tube was shaken vigorously and then stood at room temperature until the separation of two phases. After the separation, the two phases were collected and concentrated at 40 °C in vacuum, followed by redissolution in methanol. After the content analyses of each compounds by HPLC, K value of the target compound was calculated, which was defined as the concentration ratio of the lower phase (CL) against the upper phase (CU).
HSCCC Separation
A TBE-300C high-speed counter-current chromatograph (Shanghai Tauto Biotechnique Co. Ltd., Shanghai, China) was used, which was equipped with 3 polytetrafluoroethylene multi-layer coil separation columns connected in series (internal diameter of the tubing, 1.8 mm; total capacity, 300 mL) and a 20 mL manual sample loop. The revolution radius or the distance between the holder axis and central axis of the centrifuge (R) is 5 cm, and the b values of the multilayer coil varied from 0.5 at the internal terminal to 0.8 at the external terminal (b = r/R, where r is the distance from the coil to the holder shaft). The HSCCC system was equipped with a TBP-5002 constant flow pump, a TBD-2000 ultraviolet (UV) detector, a DC-0506 constant temperature regulator (Shanghai Sanotac Scientific Instruments Co. Ltd., Shanghai, China), and a V2.2.0B workstation (Shanghai Sanotac Scientific Instruments Co. Ltd., Shanghai, China). In the separation process, the temperature of separation columns was maintained at 25 °C, and the effluent was collected and determined at 280 nm. An optimum speed of 800 rpm was used in the experiment.
The HSCCC separation was conducted in the normal phase model. The selected two-phase solvent systems of HSCCC were prepared by adding all the solvents into a separation funnel at the selected volume ratios. Then, the solvent mixture was thoroughly equilibrated at room temperature; the upper phase and the lower phase were separated and degassed by sonication for 15 min before used. The sample solutions for HSCCC separation were prepared by dissolving 0.15 g of the hydrolysate into 10 mL each phase. HSCCC was performed as follows: the multilayer coiled column was first entirely filled with the lower phase, the apparatus was then rotated at 800 rpm, while the upper phase (mobile phase) was pumped into the column at a flow rate of 8 mL min−1. After hydrodynamic reached equilibrium, the sample solution (0.15 g of the hydrolysate in 10 mL each phase) was injected through the sample port. The effluent from the outlet of the column was continuously monitored with a UV detector at 280 nm. Each peak fraction was manually collected according to the UV absorbance profile, followed by analyzing the purity and tyrosinase inhibition activity.
TLC Analysis
Thin-layer chromatography (TLC) analysis was performed on a GF254 silica gel plate (Qingdao Haiyang Chemical Co., Qingdao, China). The standards and samples were spotted to a 5 mm diameter dot using a 2 μL capillary tube. The developing was conducted using the mixed solvent consisting of chloroform and methanol in the ratio of 10:1.2. After drying at room temperature, the plate was colored by iodine vapor.
HPLC Analysis
An Agilent 1260 HPLC instrument (Agilent Technologies Co., Milford, USA) coupled with a UV-G1314F VWD detector and a reversed-phase Symmetry C18 column (150 mm × 4.6 mm i.d., 3.5 μm, Waters, Milford, MA, USA) were used. The solvent system was composed of ultra-pure water (mobile phase A) and acetonitrile (mobile phase B). The elution was carried out using a gradient program: 5% B, 0–5 min; 5–25% B, 5–10 min; 25–45% B, 10–35 min; 45–5% B, 35–40 min. The column temperature was 35 °C, and injection volume was 20 μL. The elution flow rate was 0.4 mL min−1. The eluent was monitored using the UV detector at 280 nm.
Quantification was based on the external standard method. A stock solution was prepared with 10.0 mg of standard, which was dissolved and diluted to 5 mL with ultra-pure water. Aliquots of the standard stock solution were pipetted to prepare working solutions at series concentration with ultra-pure water. After analysis using the HPLC, the peak area responses (y) were applied to build the calibration curve against the concentration (x), which was further used to calculate the concentration in the samples.
Identification of Tyrosinase Inhibitors by NMR
The nuclear magnetic resonance (NMR) spectra were performed with tetramethylsilane as internal standard and CDCl3 as solvent using a Bruker AV400NMR spectrometer at 400 MHz (Bruker BioSpin Corporation, Billerica, MA).
Mass data were acquired in the positive ion mode from 6460 Triple Quad LC/MS instrument (Agilent Technologies, USA). The flow rate was 0.5 mL min−1; the injection volume was 2.0 μL; Agilent Jet Stream electrospray ionization (ESI) was conducted with a gas temperature, 300 °C; gas flow, 10 L min−1; nebulizer, 45 psi; sheath gas temperature, 350 °C; sheath gas flow, 12 L min−1; capillary voltage, 3000 V; and nozzle voltage, 450 V.
Determination of Tyrosinase Inhibition Activity
Statistical Analysis
All the assays were carried out in triplicate. The results were expressed as mean value with standard deviation (mean ± SD). Differences were analyzed using Duncan's honestly significant difference post hoc test with p = 0.05. All the analyses were done by using SPSS v17.0 software.
Results and Discussion
Solvent Extraction and Acidic Hydrolysis of Tyrosinase Inhibitory Compounds
Enzymatic active assay showed that the fractions of petroleum ether, ethyl acetate, and n-BuOH had notable tyrosinase inhibitory activity, which is consistent that pollen has strong tyrosinase inhibitory activity [28]. Furthermore, the ethyl acetate fraction was detected to have the IC50 value (4.04 mg mL−1) significantly lower than those (≥5 mg mL−1) of petroleum ether and n-BuOH fractions (Table 1), indicating the ethyl acetate fraction has the strongest activity of inhibiting tyrosinase. HPLC analysis showed that the ethyl acetate fraction contained complicated components with close retention times (Figure 1A). After the hydrolysis by acid, both the components and IC50 value (3.08 mg mL−1) markedly reduced (Figure 1B), which means that hydrolysis can promote tyrosinase inhibitory activity of pollen. This result is consistent with previous researches about some flavonoids responsible for many bioactivities. For instance, Lv et al. has indicated that acid hydrolysis is an effective method to transform flavonoid glycosides to flavonoids aglycones [26], and Kathirvel and Richards has shown the lipid oxidation inhibitory activity of flavonol aglycones is more effective than that of flavonol glycosides [29].
Tyrosinase inhibitory activity of different fractions of camellia pollen
| Samples | IC50 of tyrosinase inhibition (mg mL−1) |
|---|---|
| Petroleum ether fraction | 9.43 ± 0.81 |
| n-BuOH fraction | 8.96 ± 0.63 |
| Crude ethanol extract of camellia pollen | 5.90 ± 0.33 |
| Ethyl acetate fraction | 4.04 ± 0.32 |
| Hydrolysate of ethyl acetate fraction | 3.08 ± 0.27 |
| Fr. I | 0.17 ± 0.02 |
| Fr. II (LA) | 4.02 ± 0.03 |
| Fr. III (5-HMF) | 2.03 ± 0.12 |
| Arbutin | 0.21 ± 0.03 |
HPLC chromatograms of (A) ethyl acetate fraction of camellia pollen, (B) hydrolysate of ethyl acetate fraction, (C) peak fraction I by HSCCC separation, (D) peak fraction II by HSCCC separation, and (E) peak fraction III by HSCCC separation. Insert: the UV spectrum of corresponding peak
Citation: Acta Chromatographica Acta Chromatographica 29, 4; 10.1556/1326.2017.29407
Selection of Two-Phase Solvent System and HSCCC Separation
For successful resolution of target compounds in HSCCC separation, two sets of solvent system, i.e., ethyl acetate–n-butanol–water and n-hexane–ethyl acetate–methanol–water systems, were tested for their capability to separate the main compounds (i.e., compounds a, b, and c) in Figure 1 B. As shown in Table 2, the solvent systems of n-hexane–ethyl acetate–methanol–water provided large partition coefficient values (K values) for all the three compounds, which may lead to extended elution time as the solutes will be eluted in excessively broad peaks [30]. By contrast, solvent systems of ethyl acetate–n-butanol–water showed smaller K values (Table 2). Researchers have illustrated that a solvent system having the partition coefficient (K) in the range of 0.5–2.0 could get an efficient HSCCC separation [31, 32]. From this point of view, the solvent system of ethyl acetate–n-butanol–water at the volume ratio of 2:1:1 and 2:1:2 was not desirable to operate the separation, because the K values for compound b and compound c were much smaller than 0.5 (Table 2). When ethyl acetate–n-butanol–water was at the volume ratio of 2:1:3, the solvent provided suitable K values for the HSCCC separation of compound a (K = 0.86) and compound b (K = 0.57) (Table 2). After the HSCCC separation using the solvent of ethyl acetate–n-butanol–water (2:1:3, v/v/v), three fractions were eluted within 50 min (Figure 2). Furthermore, all of the three fractions (Fr. I–Fr. III) exhibited tyrosinase inhibitory activity (Figure 2B). The IC50 values were determined to be 0.17, 4.02, and 2.03 mg mL−1 for Fr. I, II, and III (Table 1), respectively. In comparison, the IC50 value of Fr. I is more than most of other tyrosinase inhibitors such as the C-glycosylated flavonoid from immature calamondin peel (IC50 = 0.87 mg mL−1) [33], the fucoidan from kelp polysaccharide (IC50 = 0.82 mg mL−1) [34], and the commercial cosmetic whitening agent arbutin (IC50 = 0.21 mg mL−1).
The partition coefficients (K) of target compounds in different solvent systems
| Solvent systems (v/v) |
K values |
||
|---|---|---|---|
| Compound a | Compound b | Compound c | |
| Ethyl acetate–n-butanol–water (2:1:1) | 0.92 | 0.37 | 0.31 |
| Ethyl acetate–n-butanol–water (2:1:2) | 0.90 | 0.41 | 0.36 |
| Ethyl acetate–n-butanol–water (2:1:3) | 0.86 | 0.57 | 0.45 |
| n-Hexane–ethyl acetate–methanol–water (2:3:2:3) | 1.20 | 4.16 | 2.22 |
| n-Hexane–ethyl acetate–methanol–water (2:3:3:2) | 2.56 | 7.69 | 2.56 |
| n-Hexane–ethyl acetate–methanol–water (2:2:3:3) | 3.03 | 10 | 3.57 |
Absorbance (A) and tyrosinase inhibitory activity (B) of HSCCC separations
Citation: Acta Chromatographica Acta Chromatographica 29, 4; 10.1556/1326.2017.29407
The analysis of TLC (Figure 3) and HPLC (Figure 1) showed that Fr. I was a mixture containing several components, whereas Fr. II and Fr. III consisted of single component. Retention time analysis indicated that Fr. II and Fr. III from HSCCC were compounds b and a (Figure 1B), respectively. However, Fr. I was a mixture containing compound c and some other compounds.
TLC analysis of (A) ethyl acetate fraction, (B) hydrolysate of ethyl acetate fraction, (C) peak fraction I by HSCCC separation, (D) peak fraction II by HSCCC separation, and (E) peak fraction III by HSCCC separation
Citation: Acta Chromatographica Acta Chromatographica 29, 4; 10.1556/1326.2017.29407
Identification of the HSCCC Fractions
Fr. II (Figure 2) was obtained as a white oil, UV (λmaxWater): 209 nm, 267 nm (Figure 1E). Positive ESI–MS (m/z): [M + H]+116.12, [M + Na]+ 138.12. 1H-NMR (CDCl3, 400 MHz): δ 10.36 (1H, s, COOH), 2.77 (2H, t, J = 4.0 Hz, H-3), 2.64 (2H, t, J = 4.0 Hz, H-2), 2.21 (3H, s, H-5) (Figure 4A). 13C-NMR (CDCl3, 125 MHz): δ 206.67 (CO), 178.38 (COOH), 37.69 (C-3), 29.82 (C-2), 27.75 (C-5) (Figure 4B). Based on this information, Fr. II was identified as levulinic acid (LA) and the structure was shown in Figure 4.
1H-NMR (A) and 13C-NMR (B) spectrums of 5-hydroxymethylfurfural. Insert: the chemical structures of levulinic acid
Citation: Acta Chromatographica Acta Chromatographica 29, 4; 10.1556/1326.2017.29407
Fr. III (Figure 2) was obtained as a yellow oil, UV (λmaxWater): 232 nm, 284 nm (Figure 1D). Positive ESI–MS (m/z): [M + H]+ 127.2, [M + Na]+ 149.2. 1H-NMR (CDCl3, 400 MHz): δ 9.59 (1H, s, CHO), 7.24 (1H, d, J = 4.0 Hz, H-3), 6.53 (1H, d, J = 4.0 Hz, H-4), 4.73 (2H, s, H-6) (Figure 5A). 13C-NMR (CDCl3, 125 MHz): δ 177.77 (CHO), 160.52 (C-5), 152.61 (C-2), 122.69 (C-3), 110.14 (C-4), 57.86 (C-6) (Figure 5B). These data indicated that Fr. III is 5-HMF and the structure was shown in Figure 5.
1H-NMR (A) and 13C-NMR (B) of levulinic acid. Insert: the chemical structures of 5-hydroxymethylfurfural
Citation: Acta Chromatographica Acta Chromatographica 29, 4; 10.1556/1326.2017.29407
Researchers have already separated some bioactive components from bee pollen. Han et al. have isolated five flavonoids from pollen that could inhibit prostate-specific antigen secretion [3]; Li et al. found that the polysaccharide separated from bee pollen possessed strong immunostimulatory activity [4]; Kim et al. suggested that phenolic compounds were the active constituents of bee pollen for tyrosinase inhibition [35]. In this study, LA and 5-HMF were identified to inhibit tyrosinase, which is different in structure from those of previous studies, indicating that bee pollen still contained lots of bioactive compounds unknown to public and HSCCC is an effective approach to separate new bioactive compound from nature.
Analysis of Purity and Recovery
After the separation of HSCCC, 6.04 mg LA and 13.21 mg 5-HMF were obtained from 150 mg hydrolysate, and the purity analysis showed that LA and 5-HMF were separated with the purity of 98.72% and 99.18%, respectively. Furthermore, the HPLC analysis showed that LA and 5-HMF were hardly detected in the crude extract of pollen (Figure 1A), which are much lower than those of the polyphenolic (4.4~16.4 g kg−1) and flavonoid (2.8~12.7 g kg−1) in bee pollen [36]. However, after solvent extraction and HSCCC separation, LA and 5-HMF were analyzed to have their recoveries of 0.80 g kg−1 and 1.75 g kg−1 from camellia pollen (Table 3). This result demonstrated that HSCCC is effective to purify bioactive compounds existing at a low concentration.
Yield and purity of different fraction of camellia pollen
| Sample | Recovery (g kg−1 pollen) | Purity (%) |
|---|---|---|
| Fr. II (LA) | 0.80 | 98.72 |
| Fr. III (5-HMF) | 1.75 | 99.18 |
Researchers have suggested that a combination of two or more chromatographic separation was always required for the purification of high-purity compounds. For example, silica gel has been repeatedly used for the separation of tyrosinase inhibitors from Neolitsea aciculata [9]; Husni et al. showed that a procedure consisting of repeated Sephadex LH-20 chromatography and preparative HPLC was required for the purification of bioactive components from Stichopus japonicus [37]. Besides that, Han et al. demonstrated that the combination of chromatographic columns, such as silica gel, Sephadex LH-20, and ODS MPLC columns, were required for the separation of bioactive compounds from the pollen of Brassica napus L. [3]. Despite that Fr. I, which showed the greatest tyrosinase inhibition activity, was not purified to homogeneity, two tyrosinase inhibitors were sufficiently purified in one-step separation within 50 min by the use of HSCCC, which suggested that HSCCC is a simple and effective approach to separating bioactive components from camellia pollen.
Conclusions
The camellia pollen extracts prepared with petroleum ether, ethyl acetate, and n-BuOH had tyrosinase inhibitory activity. Acidic hydrolysis could promote the tyrosinase inhibitory activity of ethyl acetate fraction. After the one-step separation of HSCCC (solvent system: ethyl acetate–n-butanol–water [2:1:3, v/v/v]), three fractions were separated to have tyrosinase inhibitory activity from the hydrolysate, of which two chemicals (>98% purity) were sufficiently purified and identified to be LA and 5-HMF. The recovery was 0.80 g kg−1 pollen for LA and 1.75 g kg−1 pollen for 5-HMF. These results demonstrated that HSCCC is an efficient method for isolation and purification of bioactive components from nature products.
Acknowledgments
The authors acknowledge the financial support from the Project of Ministry of Agriculture (Apicuture) of China (No. 30771636) and the Foundation for Innovative Research Team of Jimei University (No. 2010A006).
References
- 1.↑
Pascoal A. ; Rodrigues S.; Teixeira A.; Feás X.; Estevinhoa L. M. Food. Chem. Toxicol. 2014, 233, 63.
- 2.↑
Almeida-Muradian L. B. ; Pamplona L. C.; Coimbra S.; Barth O. M. J. Food. Compos. Anal. 2005, 105, 18.
- 5.↑
Kostić A. Ž. ; Barać M. B.; Stanojević S. P.; Milojković-Opsenica D. M.; Tešić Z. M.; Šikoparija Z. L.; Radišić P.; Prentović M.; Pešić M. P. LWT-Food. Sci. Technol., 2015, 301, 62.
- 7.
Morais M. ; Moreira L.; Feás X.; Estevinho L. M. Food. Chem. Toxicol. 2011, 1096, 5.
- 12.↑
Zhang J. ; Zhu X.; Luo F.; Sun C. D.; Huang J. Z.; Li X.; , Chen K. S. J. Sep. Sci. 2012, 128, 35.
- 16.
Shaheen N. ; Yin L.; Gu Y.; Rwigimba E.; Xie Q.; Wei Y. J. Sep. Sci. 2015 1933, 38.
- 17.↑
He K. ; Ye X.; Li X.; Chen H. Y.; Yuan L. J.; Deng Y. F.; Chen X.; Lia X. D. J. Chromatogr. B 2012, 49, 881.
- 20.↑
Hridya H. ; Amrita A.; Sankari M.; Doss C. G.; Gopalakrishnanb M.; Gopalakrishnana C.; Siva R. Int. J. Biol. Macromol. 2015, 228, 81.
- 23.↑
Hu Y. H. ; Liu X.; Jia Y. L.; Guo Y. J.; Wang Q.; Chen Q. X. J. Biosci. Bioeng. 2014, 142, 117.
- 24.↑
Huang X. X. ; Liu Q. B.; Wu J.; Yu L. H.; Cong Q.; Zhang Y.; Lou L. L.; Li L. Z.; Song S. J. Planta. Med. 2014, 1732, 80.
- 27.↑
Wang B. S. ; Chang L. W.; Wu H. C.; Huang S. L.; Chu H. L.; Huang M. H. Food Chem. 2011, 141, 127.
- 35.↑
Kim S. B. ; Jo Y. H.; Liu Q.; Ahn J. H.; Hong I. P.; Han S. M.; Hwang B. Y.; Lee M. K. Molecules 2015, 19764, 20.
- 36.↑
MĂrghitaş L. A. ; Stanciu O. G.; Dezmirean D. S.; Bobis O.; Popescu O.; Bogdanov S.; Campos M. G. Food Chem. 2009, 878, 115.
