2022 Volume 70 Issue 11 Pages 796-804
We have developed a simple and accurate method for quantifying sugars in herbal medicines, which have hitherto been difficult to quantify. Using ultra performance liquid chromatography-quadrupole-time-of-flight (UPLC-Q-TOF)-MS and two types of columns with different chemical properties, we determined the optimum conditions for separating nine sugars (fructose, galactose, glucose, mannitol, sucrose, melibiose, raffinose, manninotriose, and stachyose) commonly found in herbal medicines. Separation was completed within 10 min when an apHera NH2 HPLC column was used, although galactose and glucose could not be separated. On the other hand, the nine sugars were completely separated within 16 min when a hydrophilic interaction chromatography (HILIC)pak VG-50 2D column was used. The calibration curves obtained using those two columns gave good linearity for the sugar standards, and the coefficient of determination was 0.995 or higher. Both columns showed excellent performance with short analysis time and high sensitivity. Using our developed method, we were able to quantify sugars in galactose-free herbal medicines within 10 min and in herbal medicines containing galactose within 16 min. We revealed that our method could be used for the analysis of sugars in Angelica acutiloba and Rehmannia glutinosa roots.
Plants require primary and secondary metabolites for growth and development. Examples of primary metabolites include sugars and amino acids. Glucose derived from photosynthesis is an important energy source. Amino acids are composed of amino groups and carboxyl groups, and are important building blocks of proteins in living organisms. Secondary metabolites, on the other hand, are not directly involved in growth and development, but rather are generated as defense against infection and damage by insects, and in response to stress such as desiccation and ultraviolet irradiation. Alkaloids, terpenoids, and flavonoids have a variety of chemical structures and biological activities. In general, humans utilize primary metabolites for growth and development and secondary metabolites for regulating body function. Primary and secondary metabolites in herbal medicine are expected for the pharmaceutical effects. As herbal medicine contains natural products, there is much variability among individuals depending on the production area, the processing method, etc., and appropriate evaluation methods are required to ensure stable quality. Secondary metabolites are often used as indicator compounds for the quality control of herbal medicine; one example is paeoniflorin in Paeoniae radix. However, the medicinal effects of herbal medicine cannot be always explained only by secondary metabolites.
In the past when analytical techniques were lacking, the presence of experts who were able to evaluate the quality of herbal medicine by their five senses was critical. Among the five senses of sight, touch, taste, smell, and hearing, taste was useful and it could be further subdivided into five basic types such as sweet, bitter, sour, spicy, and salty. In this regard, it is necessary to measure sugar content as an indicator of sweetness. For example, sweetness is considered one of the indicators of the high quality of Japanese Angelica Root (当帰, Toki in Japanese)1) (identified as the root of Angelica acutiloba Kitag.) because the sugars components of this herbal medicine vary depending on the processing method.2) As sucrose (Suc) in A. acutiloba root is hydrolyzed into glucose (Glc) and fructose (Fru) during root processing,3) these sugars possessing a sweetness are important indicators of quality. On the other hand, it has been reported that the taste of Rehmannia Root (地黄, Jio in Japanese) (derived from the root of Rehmannia glutinosa Libosch. ex Fisch. et C.A.Mey.) was changed by processing as well, and that stachyose (Sta) was hydrolyzed into raffinose (Raf), manninotriose (Mnt), Suc, melibiose (Mel), galactose (Gal), Glc, and Fru4) (Fig. 1). Despite the necessity of analyzing sugars as indicators of quality, there are few reports on the quantification of sugars in the quality evaluation of herbal medicine due to the lack of an established method that enables simple and accurate quantification.
One of the available methods for the determination of sugars is the enzymatic method, which is mainly applied to foods5) and biological tests such as blood glucose determination.6) However, the enzymatic method requires a specific enzyme for the substrate, so the target sugar is limited to Fru, Glc, etc. Moreover, the analysis is time-consuming. GC is a general method for sugar analysis,7,8) but because trimethylsilyl (TMS) derivatization is required for pretreatment, operability and stability issues exist. Furthermore, it takes a long time to analyze oligosaccharides such as Sta (a tetrasaccharide). In the case of LC of sugars, two problems exist, namely, weak UV absorption (lack of UV absorption on chemical structures of sugars) and the lack of an appropriate separation column (to separate isomers). As regards the detector, a refractive index detector is generally used in conventional methods,9,10) and some studies have used an evaporative light scattering detector11,12) and a charged aerosol detector.13) Those three methods have the advantage of being able to measure compounds that do not have UV absorption. However, they have disadvantages as well, such as their unsuitability for gradient elution, the inconsistency of the retention times, and difficulty of separation if the sample contains interfering compounds. Therefore, they cannot be applied to a wide variety of samples. Gal and Glc are stereoisomers that differ only in the configuration of the hydroxyl group at 4-position, and this property has made complete separation difficult. Generally, ion-exchange columns are used to separate Gal and Glc, but because only water can be used as the mobile phase, the analysis of sugars with high molecular weights is time-consuming.14) Although the importance of the quantitative analysis of sugars in herbal medicine is apparent, there have been few reports so far.
In this study, we attempted to qualitatively analyze sugars in herbal medicines by LC using an optimum detection technique and separation column. In recent years, LC/MS has been applied to plant materials and food products.15–17) Because LC/MS detects the molecular weight of a compound, compounds that do not have UV absorption can be detected with high sensitivity. With regard to the separation column, we examined two columns: One was a normal-phase column based on covalent polyamine (Method 1), and the other was a polymer-based column in the hydrophilic interaction chromatography (HILIC) mode (Method 2). The mobile phases and the gradient systems commonly used in LC-MS could be applied to these columns. We initially determined the optimum conditions by using sugar standards and then applied those conditions to herbal medicine samples.
The following sugars were adopted as standards: monosaccharides (Fru, Gal, Glc), disaccharides (Suc, Mel), trisaccharides (Raf, Mnt), tetrasaccharide (Sta), and a sugar alcohol (mannitol (Mal)). Two kinds of herbal medicine, Toki and Jio, were selected as the experimental materials. The plants of origin of Jio were cultivated and processed in the Medicinal Plant Garden of the School of Pharmacy, Kanazawa University under the same conditions. Fru, Glc, and Suc are commonly found in plants as well as herbal medicines such as Toki. In addition to these sugars, Jio contained Gal, Suc, Mel, Raf, Mnt, and Sta, so these two herbal medicines were good experimental materials for checking the separation ability under the optimum LC-MS conditions.
Through the quantitative determination of sugars in Toki, it could suggest the validity of the sugar content of Angelicae acutilobae Radix specified in the 18th edition of the Japanese Pharmacopoeia (JP18), namely, no less than 35.0% in dilute ethanol-soluble extract.18) It is considered that the dilute ethanol-soluble extract in the Test for Crude Drugs in JP18 was specified because it was difficult to measure the sugar content as the sweetness of Toki.
Fructose (purity ≥99.0%), galactose (purity ≥98.0%), glucose (purity ≥98.0%), sucrose (purity ≤100%), and raffinose pentahydrate (purity ≤100%) were purchased from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan). Mannitol (purity ≥98.0%) and melibiose monohydrate (purity ≥99%) were purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Manninotriose (purity ≥98%) and stachyose tetrahydrate (purity unknown) were purchased from Ark Pharm, Inc. (IL, U.S.A.) and MP Biomedicals (CA, U.S.A.), respectively. 2-Aminoethanesulfonic acid (purity ≥98.0%) and cyanuric acid (purity ≥99.0%) as the internal standard were purchased from Tokyo Chemical Industry Co., Ltd. Ammonium hydroxide was purchased from Hayashi Pure Chemical Ind., Ltd. (Osaka, Japan). Acetonitrile and water of LC-MS grade were purchased from FUJIFILM Wako Pure Chemical Corporation. All other reagents were of analytical grade or higher.
Herbal medicines derived from the roots of A. acutiloba (Toki, processed by hot water), A. acutiloba var. sugiyamae (Toki, Hokkai grade without processing), and R. glutinosa (Jio) were purchased from Uchida Wakan-yaku Co., Ltd. (Tokyo, Japan; abbreviated UCW hereinafter) and Tochimoto Tenkaido Co., Ltd. (Osaka, Japan; abbreviated TMT hereinafter). The sample numbers were as follows: T1 (Lot no. 008012008), Toki (Hokkai grade, heat-dried) from TMT; T2 (Lot no. 008013008), Toki (processed) from TMT; T3 (Lot no. 008014005), Toki (processed) from TMT; T4 (Lot no. E260330), Toki (processed) from UCW; T5 (Lot no. 008016015), Toki (processed) from TMT; J1 (Lot no. 005019006), Jio (heat-dried) from TMT; J2 (Lot no. P011103291), dried Jio (heat-dried) from TMT; J3 (Lot no. 1403C005001), Juku-Jio (processed) from TMT; and J4 (Lot no. 1312K005001), Juku-Jio (processed) from TMT.
Herbal medicines prepared in the Medicinal Plant Garden of the School of Pharmacy, Kanazawa University were as follows19,20): J5, freeze-dried Jio; J6, oven-dried Jio (50 °C, 24 h).
All samples were identified by Professor Yohei Sasaki, and the materials used in this study were stored in the Specimen Room, Kanazawa University, Japan.
Ultra Performance Liquid Chromatography-Quadrupole-Time-of-Flight (UPLC-QTOF)-MS AnalysisWe performed analyses on a UPLC-Q-TOF-MS, which consisted of an ACQUITY UPLC H-Class system and a Xevo G2-XS Q-Tof system (Waters, MA, U.S.A.). Negative electrospray ionization-MS analysis using Q-TOF-MS was performed in the full scan mass mode in the m/z range of 100 to 1000. Extracted ion chromatograms were used to obtain the molecular ion of interest in the full scan mass mode at the solation width of m/z 0.05.
Two columns and different conditions were used for analysis. In [Method 1], we carried out apHera NH2 HPLC (ϕ2 × 150 mm, 5 µm, Sigma-Aldrich Co., MO, U.S.A.) at 40 °C. The mobile phase consisted of solvent A (water) and solvent B (acetonitrile), and the following gradient elution program was used: 75% B from 0 to 5 min; 75–65% B from 5 to 10 min; 65–75% B from 10 to 10.1 min; 75% B from 10.1 to 15 min. The flow rate was 0.5 mL/min. In (Method 2), we performed the separation using HILICpak VG-50 2D (ϕ2 × 150 mm, 5 µm, Showa Denko K. K., Tokyo, Japan) at 60 °C. The mobile phase consisted of solvent A (67% methanol mixed with 1.25% ammonium hydroxide) and solvent B (acetonitrile with 1.25% ammonium hydroxide), and the following gradient elution program was used: 97–96% B from 0 to 8.5 min; 96–60% B from 8 to 12.5 min; 60–30% B from 12.5 to 13 min; 30% B from 13 to 15 min; 30–20% B from 15 to 15.5 min; 20% B from 15.5 to 20.5 min; and returned to 97% B at 26 min. The flow rate was 0.4 mL/min. The column injection volume was 2 µL.
Preparation of Standard Solutions and SamplesEach sugar standard was dissolved in water (10 mM) and diluted with water–acetonitrile (2 : 8, v/v). Fru, Gal, and Glc, which were difficult to separate and had the same molecular weight, were premixed. Solutions of eight concentrations were analyzed in duplicate, and calibration curves for the linearity test were prepared by plotting the peak area ratio of the standard of known concentration to its internal standard on the vertical axis and the standard concentration (Method 1 : 0.5–64.0 µM, Method 2: 1.0–64.0 µM) on the horizontal axis.
Toki powder was sonicated with purified water or dilute ethanol (47.5% ethanol) for 30 min, and the suspension was centrifuged at 15900 × g for 5 min. The supernatant was diluted with acetonitrile and mixed with the internal standard in 1 : 1 ratio (final sample concentration 50 µg/mL). Jio was prepared in the same way (final sample concentration 100 µg/mL). Here, since the ethanol concentration for the dilute ethanol-soluble extract in JP18 was 47.5%, the same concentration was adopted in this dilute ethanol extract.
Simple Method for Determination of Dilute Ethanol-Soluble Extract ContentThe dilute ethanol-soluble extract content was determined by adding 1.4 mL of dilute ethanol to 46 mg of powdered sample and performing ultrasonication for 30 min at room temperature. The supernatant obtained after centrifugation at 15900 × g for 5 min was transferred into a volumetric flask. Dilute ethanol was added again into the centrifuge tube containing the pellet and the supernatant obtained after a second centrifugation at 15900 × g for 5 min was transferred into the same volumetric flask. Dilute ethanol was added to make 2 mL. Half the volume of the diluted supernatant was heat-dried with a heat block at 105 °C until ethanol was almost completely removed. Finally, the residue was heat-dried at 105 °C for 4 h in an oven to completely remove the dilute ethanol, and its weight was measured. The dilute ethanol-soluble extract content was calculated from the weights of the powder and the residue. We were able to confirm the equivalence of our simple method and the JP18 method (simple method; 46.8% ± 0.55, JP18 method; 47.3% ± 0.53, n = 5, mean ± standard deviation (S.D.)).
Verification of Precision and RepeatabilityThe inter-day precision of measuring sugar content was evaluated by measuring each standard solution twice a day for three consecutive days. Repeatability was determined by measuring sugar content in six independently prepared extracts of identical commercial products.
Optimization of the analytical conditions for Method 1 and Method 2 was carried out using nine standards of sugars contained in herbal medicines. The chromatograms of the nine sugar standards are shown in Figs. 2 (Method 1) and 3 (Method 2). The chromatograms were selected on the basis of the qualifier ion mass (m/z). The chromatograms obtained using Method 1 showed peaks having a highly symmetrical shape, and the nine standards could be analyzed within 10 min (Fig. 2). Only Gal and Glc were detected as one overlapping peak (peak 2 + 3 in topmost chromatogram of Fig. 2). Furthermore, this high-sensitivity analysis revealed that Sta standard contained 0.6% Fru, 0.5% Gal + Glc, 4.8% Suc, 6.6% Raf, and 1.9% Mnt. Sta content was calculated to be 85.5% although the purity of Sta standard was unknown. The expanded chromatograms selected on the basis of the qualifier ion mass of each impurity are shown in Fig. 2, right. An accurate calibration curve for Sta standard could be created on the basis of the calculated purity of Sta standard.
Left: Chromatograms selected on the basis of qualifier ion mass (m/z) specific to each standard, Fru (peak 1), Gal (peak 2), and Glc (peak 3) were premixed. Right: Chromatograms of impurities contained in Sta standard, selected on the basis of ion mass (m/z) specific to each sugars. Peak numbers 1: Fru, 2: Gal, 3: Glc, 4: Mal, 5: Suc, 6: Mel, 7: Raf, 8: Mnt, 9: Sta
In the chromatograms obtained using Method 2, the peaks of Fru (peak 1), Gal (peak 2), and Glc (peak 3) showed tailing (Fig. 3), although the peaks of Gal and Glc could be clearly separated (peaks 2 and 3 in the topmost chromatogram, Fig. 3), and the nine standards could be analyzed within 16 min (Fig. 3). Peak 4 (Mal), peak 5 (Suc) and peak 6 (Mel) showed splitting. However, it was unlikely to contain large amounts of impurities for Mal, Suc and Mel, because those standards showed a single and clear symmetric peak in the chromatograms obtained by using Method 1 (Fig. 2). Therefore, we postulated that the peak splitting was caused by the chemical properties of the HILIC column. On the other hand, impurities were detected from Sta standard using Methods 1 and 2 (Figs. 2, 3 right). The purity of Sta standard was calculated by the content of impurities and found to be about 83.2% using Method 2.
Left: Chromatograms selected on the basis of qualifier ion mass (m/z) specific to each standard, Fru (peak 1), Gal (peak 2), and Glc (peak 3) were premixed. Right: Chromatograms of impurities contained in Sta standard, selected on the basis of ion mass (m/z) specific to each sugars. Peak numbers 1: Fru, 2: Gal, 3: Glc, 4: Mal, 5: Suc, 6: Mel, 7: Raf, 8: Mnt, 9: Sta. *Peaks 5 and 6 were single peaks, and splitting seemed to be caused by the chemical properties of the column.
The calibration curves for the sugar standards were prepared by using Method 1 and Method 2, respectively. When the sugar standards were prepared at concentrations of 0.5 to 64 µM (Method 1), 1.0 to 64 µM (Method 2) and analyzed, the coefficient of determination (R2) was higher than 0.995 for both columns, and the calibration curve showed good linearity (Table 1). In addition, the use of [M + Cl]− for Fru, Gal, Glc, and Mel (Method 1&2), [M + Cl]− for Mnt and Sta (Method 1), and [M − H]− for the other sugars resulted in high sensitivity and quantification. The limit of detection (LOD) and the limit of quantitation (LOQ) for each sugar standard was set to three and ten times the signal-to-noise ratio, respectively. The LOD of each standard for Method 1 ranged from 0.01 to 0.07 µM, and that for Method 2 ranged from 0.03 to 0.26 µM. The LOQ for Method 1 ranged from 0.03 to 0.20 µM, and that for Method 2 ranged from 0.09 to 0.81 µM.
| Peak no. | Compounds | Retention time (min) | Molecular formula | MW | Detection ion mass (m/z) | Ion selected for calibration curve (m/z) | R2 | LOD (µM) | LOQ (µM) | |
|---|---|---|---|---|---|---|---|---|---|---|
| Method 1 | 1 | Fructose (Fru) | 1.76 | C6H12O6 | 180.0634 | [M − H]− 179.0693 | [M + Cl]− 215.0490 | 0.9976 | 0.0426 | 0.1235 |
| [M + Cl]− 215.0490 | ||||||||||
| 2 | Galactose (Gal) | 2.13 | C6H12O6 | 180.0634 | [M − H]− 179.0682 | [M + Cl]− 215.0483 | 0.9987 | 0.0343 | 0.1034 | |
| [M + Cl]− 215.0483 | ||||||||||
| 3 | Glucose (Glc) | 2.13 | C6H12O6 | 180.0634 | [M − H]− 179.0682 | [M + Cl]− 215.0483 | 0.9991 | 0.0319 | 0.1078 | |
| [M + Cl]− 215.0483 | ||||||||||
| 4 | Mannitol (Mal) | 1.97 | C6H14O6 | 182.0790 | [M − H]− 181.0850 | [M − H]− 181.0850 | 0.9953 | 0.0295 | 0.0890 | |
| [M + Cl]− 217.0645 | ||||||||||
| 5 | Sucrose (Suc) | 2.87 | C12H22O11 | 342.1162 | [M − H]− 341.1348 | [M − H]− 341.1348 | 0.9992 | 0.0136 | 0.0358 | |
| [M + Cl]− 377.1143 | ||||||||||
| 6 | Melibiose (Mel) | 3.89 | C12H22O11 | 342.1162 | [M − H]− 341.0327 | [M + Cl]− 377.0067 | 0.9969 | 0.0299 | 0.0892 | |
| [M + Cl]− 377.0067 | ||||||||||
| 7 | Raffinose (Raf) | 5.02 | C18H32O16 | 504.1690 | [M − H]− 503.2002 | [M − H]− 503.2002 | 0.9994 | 0.0174 | 0.0558 | |
| [M + Cl]−539.2106 | ||||||||||
| 8 | Manninotriose (Mnt) | 7.12 | C18H32O16 | 504.1690 | [M − H]− 503.1823 | [M + Cl]− 539.1507 | 0.9973 | 0.0650 | 0.1958 | |
| [M + Cl]−539.1507 | ||||||||||
| 9 | Stachyose (Sta) | 8.79 | C24H42O21 | 666.2219 | [M − H]− 665.2306 | [M + Cl]−701.2072 | 0.9991 | 0.0102 | 0.0289 | |
| [M + Cl]− 701.2072 | ||||||||||
| Method 2 | 1 | Fructose (Fru) | 5.75 | C6H12O6 | 180.0634 | [M − H]− 179.0658 | [M + Cl]− 215.0454 | 0.9980 | 0.2159 | 0.6736 |
| [M + Cl]− 215.0454 | ||||||||||
| 2 | Galactose (Gal) | 8.12 | C6H12O6 | 180.0634 | [M − H]− 179.0620 | [M + Cl]− 215.0445 | 0.9999 | 0.2578 | 0.8059 | |
| [M + Cl]− 215.0445 | ||||||||||
| 3 | Glucose (Glc) | 9.7 | C6H12O6 | 180.0634 | [M − H]− 179.0641 | [M + Cl]− 215.0437 | 0.9998 | 0.2254 | 0.7556 | |
| [M + Cl]− 215.0437 | ||||||||||
| 4 | Mannitol (Mal) | 9.13 | C6H14O6 | 182.079 | [M − H]− 181.0824 | [M − H]− 181.0824 | 0.9994 | 0.0514 | 0.1577 | |
| [M + Cl]− 217.0604 | ||||||||||
| 5 | Sucrose (Suc) | 12.69 | C12H22O11 | 342.1162 | [M − H]− 341.1309 | [M − H]− 341.1309 | 0.9992 | 0.0477 | 0.1587 | |
| [M + Cl]− 377.1087 | ||||||||||
| 6 | Melibiose (Mel) | 13.22 | C12H22O11 | 342.1162 | [M − H]− 341.1214 | [M + Cl]− 377.993 | 0.9999 | 0.0848 | 0.2564 | |
| [M + Cl]− 377.993 | ||||||||||
| 7 | Raffinose (Raf) | 13.62 | C18H32O16 | 504.169 | [M − H]− 503.1942 | [M − H]− 503.1942 | 0.9994 | 0.0294 | 0.0961 | |
| [M + Cl]−539.1682 | ||||||||||
| 8 | Manninotriose (Mnt) | 13.94 | C18H32O16 | 504.169 | [M − H]− 503.1749 | [M − H]− 503.1749 | 0.9986 | 0.1787 | 0.5782 | |
| [M + Cl]−539.1539 | ||||||||||
| 9 | Stachyose (Sta) | 14.13 | C24H42O21 | 666.2219 | [M − H]− 665.2076 | [M − H]− 665.2076 | 0.9999 | 0.0274 | 0.0857 | |
| [M + Cl]− 701.1812 |
The calculated inter-day precision for each standard as relative standard deviation (RSD) was within 2.0% for both methods. In repeatability analyzed commercial products of Toki and Jio, all the RSD values were within 2.0% when the extracts of T1 were used as samples in Method 1, and all the RSD values were within 12.0% when the extracts of J3 were used in Method 2 (Table 2).
| Inter-day precision (%) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Fru | Gal | Glc | Mal | Suc | Mel | Raf | Mnt | Sta | ||
| Method 1 | Standard | 1.2 | 0.8 | 1.1 | 0.5 | 1.9 | 0.7 | 1.1 | 0.9 | 1.2 |
| Method 2 | 1.4 | 0.9 | 1.5 | 0.6 | 1.2 | 0.5 | 1.8 | 0.8 | 1.7 | |
| Repeatability (%) | ||||||||||
| Fru | Gal | Glc | Ma | Suc | Mel | Raf | Mnt | Sta | ||
| Method 1 | T1 | 0.7 | — | 0.8 | — | 1.8 | — | — | — | — |
| Method 2 | J3 | 1.6 | 1.70 | 2.2 | 5.3 | 4.3 | 10.8 | 10.7 | 12.0 | 1.5 |
Because Method 1 was suitable for Gal-free samples and Method 2 was suitable for Gal-containing samples, Method 1 was applied to the analysis of Toki of various qualities and Method 2, to the analysis of Jio of various qualities. It has been reported that Toki contained Fru, Glc, and Suc, and the content varies by processing.3) And we confirmed that the Gal content in Toki is low enough by Method 2 in advance. Therefore, we analyzed the three sugars using Method 1 (Fig. 4). Samples T1 to T5 (commercial products, T1 is unprocessed and T2–T5 is processed) had the following sugar content: Fru: 1.8–4.3%, Glc: 0.3–1.4%, and Suc: 28.0–36.4%. On the other hand, it has been reported that Jio contained Fru, Gal, Glc, Mal, Suc, Mel, Raf, Mnt, Sta, and verbascose and that the oligosaccharides were cleaved during processing.4) Therefore, we analyzed nine of these sugars by using Method 2 (Fig. 5).
T1–T5: commercial products.
J5: freeze-dried in Kanazawa Univ., J6: oven-dried in Kanazawa Univ., J1–J4: commercial products, J1 and J2: heat-dried, J3 and J4: processed.
Sample J5 (freeze dried) contained 0.2% Fru, 0.2% Gal, 0.1% Glc, 0.5% Mal, 4.1% Suc, 0.0% Mel, 0.9% Raf, 0.8% Mnt, and 42.5% Sta. Sample J6 (oven-dried) contained 3.2% Fru, 2.2% Gal, 3.1% Glc, 0.5% Mal, 6.9% Suc, 0.0% Mel, 2.6% Raf, 3.2% Mnt, and 21.4% Sta. On the other hand, samples J1 and J2 (commercial products, heat-dried) had the following sugar content: Fru: 1.7, 3.2%, Gal: 1.5, 2.2%, Glc: 1.4, 3.1%, Mal: 1.1, 0.5%, Suc: 7.8, 6.9%, Mel: 0.0, 0.0%, Raf: 2.4, 2.6%, Mnt: 3.2, 2.4%, and Sta: 28.7, 21.4%. Samples J3 and J4 (commercial products, processed) had the following sugar content: Fru: 11.4, 10.8%, Gal: 2.0, 1.6%, Glc: 4.2, 2.4%, Mal: 0.5, 0.5%, Suc: 0.0, 0.3%, Mel: 2.0, 1.8%, Raf: 0.4, 0.7%, Mnt: 26.0, 17.4%, and Sta: 0.0, 2.0%.
In the quality evaluation of Toki, JP18 specifies that the dilute ethanol-soluble extract content, which should be no less than 35.0%.21) It is considered that the dilute ethanol-soluble extract in the Test for Crude Drugs in JP18 was specified because it was difficult to measure the sugar content as the sweetness of Toki. Because the LC-MS results described above were for purified water extract (PWE), dilute ethanol extract (DEE) was also prepared for comparison with the dilute ethanol-soluble extract defined in JP18 Test for Crude Drugs. We measured and compared the sugar content of PWE and DEE of Toki by using Method 1 (Fig. 6A). The total sugar content of the PWE was 34.3%, and that of the DEE was 34.2%. The total sugar content was calculated from the content of Fru, Glc, and Suc. The chromatograms of both extracts are also shown, and we have confirmed that the PWE and the DEE exhibited no difference in peak patterns (Fig. 6B). In case of the simple method of the dilute ethanol-soluble extract, The DEE content of T1 to T5 ranged from 41.2 to 46.6%, and the DEE were composed of 4.0 to 11.4% Fru; 0.6 to 3.6% Glc; 62.7 to 82.4% Suc; and 5.7 to 29.1% other components (Fig. 7).
A: Sugar content of T3, B: Chromatograms of T3 measured by LC/MS. B1: PWE, B2: DEE, Peak numbers 1: Fru, 3: Glc, 5: Suc.
Sugar composition was calculated from DEE content and each sugar content. T1–T5: commercial products.
We also confirmed the equivalence of the sugar content of PWE and DEE of Jio (Fig. 8A). The total sugar content of the PWE was 39.4% and that of the DEE was 37.6%. The LC-MS chromatograms of the PWE and the DEE of Jio (J4 as representative) are shown in Fig. 8B. The symmetrical peaks in the chromatograms of the DEE can be attributed to the fact that the DEE contained fewer impurities than the PWE. Overall, we found similar peak patterns in the two chromatograms. On the other hand, in the quality evaluation of Jio, JP18 indicates that Sta is present in dried Jio, and Fru and Mnt are present in processed Jio by TLC. Therefore, we also measured the sugar content of the DEE of Jio and the sugar composition of Jio samples by using Method 2 (Fig. 9). The DEE content of unprocessed J5 was 79.0%, and the extract was composed of 0.7% Fru, 1.9% Gal, 0.4% Glc, 0.9% Mal, 8.7% Suc, 0.0% Mel, 1.2% Raf, 0.0% Mnt, 53.9% Sta, and 32.3% other components. The DEE content of sample J6 was 75.9%, and the extract was composed of 4.2% Fru, 2.8% Gal, 4.1% Glc, 0.5% Mal, 9.1% Suc, 0.0% Mel, 3.5% Raf, 3.2% Mnt, 28.2% Sta, and 44.4% other components. The DEE content of J1 and J2 (commercial products, heat-dried) are 75.9, 75.0%, and the extracts were composed of: Fru: 2.1, 4.1%, Gal: 1.9, 2.7%, Glc: 1.8, 3.9%, Mal: 1.0, 0.5%, Suc: 9.8, 8.7%, Mel: 0.0, 0.0%, Raf: 3.0, 3.3%, Mnt: 4.1, 3.0%, Sta: 36.4, 27.1%, and other components: 39.8, 46.6%, respectively. The DEE content of J3 and J4 (commercial products, processed) are 78.9, 78.9%, and the extracts were composed of: Fru: 15.0, 14.4%, Gal: 2.6, 2.1%, Glc: 5.5, 5.9%, Mal: 0.5, 0.5%, Suc: 0.0, 0.4%, Mel: 2.6, 2.4%, Raf: 0.5, 0.9%, Mnt: 34.2, 23.2%, Sta: 0.0, 2.5%, and other components: 39.1, 47.6%, respectively.
A: Sugar content of J4, B: Chromatograms of J4 measured by LC/MS. B1: PWE, B2: DEE, Peak numbers 1: Fru, 2: Gal, 3: Glc, 4: Mal, 5: Suc, 6: Mel, 7: Raf, 8: Mnt, 9: Sta.
Sugar composition was calculated from DEE content and each sugar content. J5: freeze-dried in Kanazawa Univ., J6: oven-dried in Kanazawa Univ., J1–J4: commercial products, J1 and J2: heat-dried, J3 and J4: processed.
In general, the qualitative and quantitative analyses of sugars in herbal medicines by LC/Evaporative light scattering and LC/Refractive index have low sensitivity. In this study, we used MS as the detector and an amino column and a HILIC column as the separation columns for LC of sugars. Our LC-MS method enabled the quantitative analysis of representative sugars in a small amount of material without pretreatment. Specifically, Method 1 enabled the analysis of Fru, Glc, and Suc within 10 min in Gal-free materials. Because our method adopted MS for detection, accurate quantitative analysis was possible even for compounds with similar retention times, such as Fru, Gal, Glc, and Mal. In Method 1, the chemical modification of the amino group of the amino column did not lead to anomer formation of sugars and resulted in relatively sharp peaks. However, because this amino group reacted with an aldehyde group to form a Schiff salt, repeated measurements led to column deterioration. It is known that Toki contains Fru, Glc, and Suc, and the fact that Suc content is much higher than Fru and Glc content has made it difficult to accurately measure Fru and Glc. Using Method 1, we were able to measure sugar content of Gal-free samples like Toki within 10 min, although we were unable to separate Glc from Gal in the Gal-containing samples.
We have developed Method 2 for samples containing both Gal and Glc. Using Method 2, Gal and Glc could be clearly separated, and oligosaccharides could be separated from mono-, di-, tri- and tetrasaccharides within 16 min. In the HILIC column, anomer formation of sugars occurred when mobile phases such as formic acid/acetonitrile systems commonly used in LC-MS were used. Therefore, it was necessary to make the mobile phase basic and to perform the analysis at high temperatures. Under such conditions, the recovery rate of reducing sugars was high, and the column was chemically stable with little degradation over time. Although peak tailing was found for Fru (peak 1), Gal (peak 2), Glc (peak 3), and Mal (peak 4) compared with Method 1, we postulated that the peak tailing was due to the chemical properties of the column. Nevertheless, we thought this was not a major problem because baseline separation was achieved and the quantification results were almost identical with those of Method 1.
We were able to quantify various impurities present in commercially purchased standards of Mel, Mnt, and Sta using our method. In other words, we were able to calculate the exact purity of the commercial products and the exact sugar content of the experimental materials. Detection sensitivity was high, and Sta was detected down to 0.01 µM. In research conducted so far, the accurate quantification of Sta has not been achieved; even the exact purity of Sta standard is not known. Using our method, we found that the purity of Sta standard is 85.5% (Method 1) and 83.2% (Method 2). When calibration curves were prepared using both columns, the coefficient of determination (R2) was higher than 0.995 for all sugars in the concentration range of 0.5 to 64 µM, and the calibration curves showed good linearity. There are several reports on the analysis of sugars by LC-MS, but many of them do not provide calibration curves or coefficients of determination.16,17) To check the accuracy of our method, the RSD values of the inter-day precision and the repeatability were calculated. In Method 1 and Method 2, the RSD values of the inter-day precision for all the sugar standards were lower than 2%, indicating that the target sugars could be quantified with high accuracy. The repeatability for Method 1 was lower than 2.0% RSD, and that for Method 2 was 12.0% RSD. Even though the herbal medicine samples contained natural medicine of nonuniform quality, we were able to obtain accurate results.
We chose Toki as the Gal-free sample and Jio as the sample containing both Gal and Glc. In fact, the analysis of commercial Toki (T1–T5) revealed the following sugar content: 1.8–4.3% Fru, 0.3–1.4% Glc, and 28.0–36.4% Suc, all of which were within the range of quantification of the calibration curve. On the other hand, it is known that the Suc content of Toki varies widely depending on the processing conditions, and therefore our method needs to be able to measure such materials accurately. Toki had 12 times higher Suc content than Glc content, which was the lowest of all sugars contained in Toki; nevertheless, Glc could be quantified in a single analysis without having to adjust its concentration. On the other hand, there is a report that Gal was detected in the roots of Angelica acutiloba purchased from a Japanese market,22) in this manuscript, compared with the Glc content of 4.44 mg/g, the Gal content was as low as 0.088 mg/g, which is below the limit of detection (LOD) of our method (Table 1, LOD of Gal was 0.2578 µM). Therefore, Toki can be regarded as Gal-free.
Jio contains various sugars including Fru, Gal, Glc, Mal, Suc, Mel, Raf, Mnt, Sta, verbascose, and others.23) In addition, it has been reported that the major sugar in Jio was changed by processing. In our analysis of the commercial products of Jio, we found that Sta content was highest in freeze-dried Jio (J5), and Mnt content was highest in processed Jio (J3, J4). Therefore, when analyzing Jio, it is necessary to separate the sugars and have a wide quantification range. The content of the most abundant sugar was approximately 20 times higher than that of the least abundant sugar; nevertheless, in our method, any sugar could be quantified in a single analysis without having to adjust its concentration. As described above, the wide quantification range of the calibration curve has made it possible to analyze herbal medicines having various sugar content.
We also adopted the simple method of 30-min ultrasonication in 47.5% EtOH as an equivalent test to the JP18-specified quality evaluation of Toki using dilute ethanol-soluble extracts. We examined whether the quality of Toki could be evaluated on the basis of the DEE content (47.5%) or the dilute ethanol-soluble extract defined in JP18, and found that when the total sugar composition shows little difference among samples, the DEE content will have little difference as well. This is because Fru, Glc, and Suc content account for no less than 70% of the DEE content. Furthermore, on the basis of the report that Suc content, the major sugar in Toki, has a strong positive correlation with the DEE content,24) we judged that the dilute ethanol-soluble extract defined in JP18 was the appropriate index for the quality evaluation of Toki. The dilute ethanol-soluble extract defined in JP18 is used as one of the quality indicators of Toki in a number of studies. In previous studies, the dilute ethanol-soluble extract is used only as a comprehensive evaluation method.3,24,25) In this study, we proved that the dilute ethanol-soluble extract can be used reliably for the determination of sugar content in Toki.
The sugar composition of processed herbal medicines may differ greatly depending on the origin or the processing location/conditions, even if the dilute ethanol-soluble extracts content is the same.23,26–28) An example is Jio, which is processed for used as herbal medicine. Toki may also be processed, but sugar composition is likely easier to change in Jio than in Toki. Our method enabled the separation of Gal and Glc and allowed us to accurately grasp changes in monosaccharide content due to the hydrolysis of sugars in Jio. Sugar composition in Jio varied greatly depending on the processing conditions. However, the DEE content did not change significantly even though the sugar composition was changed, indicating that the dilute ethanol-soluble extract could not be used for the quality evaluation of Jio. Since DEE has mostly consisted of sugars, it is an appropriate indicator for evaluating the total sugar content. On the other hand, DEE cannot represent the change in the sugar composition. Therefore, it seems that DEE is specified for herbal medicine such as Toki, whose sugar composition does not significantly vary by processing, but not for herbal medicine as Jio, whose sugar composition varies by processing. By contrast, our method, which enables accurate quantitative analysis of sugars, may be used for the quality evaluation of various herbal medicines and comparison of sugar composition among processing conditions.
We have established a method for the simple and accurate quantification of nine sugars (fructose, galactose, glucose, mannitol, sucrose, melibiose, raffinose, manninotriose, and stachyose) mainly contained in herbal medicines by UPLC-Q-TOF-MS using two types of separation columns with different chemical properties: apHera NH2 HPLC and HILICpak VG-50 2D. The coefficient of determination (R2) exceeded 0.995 for the calibration curves created for the sugar standards using the two columns in the sugar concentration range of 0.5 to 64.0 µM (Method 1), 1.0 to 64.0 µM (Method 2), and the calibration curves showed good linearity. As it was difficult to separate galactose and glucose, two types of separation columns were used. Our method enabled analysis within 10 min for galactose-free materials such as Toki, and within 16 min for materials containing both glucose and galactose, such as Jio. Single measurement was achieved even in samples with large differences in Glc and Suc content (Suc content was 12 times higher than Glc content) such as Toki, and samples with large differences in Glc and Sta content (Sta content was 20 times higher than Glc content) such as Jio. Our method can be widely used for sugar quantification in the quality evaluation of herbal medicine.
This work was supported by a Grant-in-Aid for Scientific Research (C), No. 18K06730, for 2018–2020 from the Japan Society for the Promotion of Science.
The authors declare no conflict of interest.
