One-Pot Bi-Enzymatic Cascade Synthesis of Novel Ganoderma Triterpenoid Saponins

Ganoderma lucidum is a medicinal fungus whose numerous triterpenoids are its main bioactive constituents. Although hundreds of Ganoderma triterpenoids have been identified, Ganoderma triterpenoid glycosides, also named triterpenoid saponins, have been rarely found. Ganoderic acid A (GAA), a major Ganoderma triterpenoid, was synthetically cascaded to form GAA-15-O-β-glucopyranoside (GAA-15-G) by glycosyltransferase (BtGT_16345) from Bacillus thuringiensis GA A07 and subsequently biotransformed into a series of GAA glucosides by cyclodextrin glucanotransferase (Toruzyme® 3.0 L) from Thermoanaerobacter sp. The optimal reaction conditions for the second-step biotransformation of GAA-15-G were found to be 20% of maltose; pH 5; 60 °C. A series of GAA glucosides (GAA-G2, GAA-G3, and GAA-G4) could be purified with preparative high-performance liquid chromatography (HPLC) and identified by mass and nucleic magnetic resonance (NMR) spectral analysis. The major product, GAA-15-O-[α-glucopyranosyl-(1→4)-β-glucopyranoside] (GAA-G2), showed over 4554-fold higher aqueous solubility than GAA. The present study demonstrated that multiple Ganoderma triterpenoid saponins could be produced by sequential actions of BtGT_16345 and Toruzyme®, and the synthetic strategy that we proposed might be applied to many other Ganoderma triterpenoids to produce numerous novel Ganoderma triterpenoid saponins in the future.


Introduction
Ganoderma lucidum is a medicinal fungus which has been used as a dietary supplement to promote healthiness in Asia for thousands of years [1]. In the modern age, many triterpenoids are purified from the fungus and identified to possess benevolent bioactivities, such as anti-cancer, antiinflammation, anti-bacteria, antioxidant, and immune-regulation activities [2]. Ganoderic acid A (GAA) is a major Ganoderma triterpenoid and recently has been shown to be bioactive on medical applications, including inhibiting bleomycininduced lung fibrosis in mice [3], attenuating lipopolysaccharide-induced lung injury in mice [4], retarding renal cyst development in polycystic kidney disease [5], and protecting neural cells against nitric oxide (NO) stress injury [6].
On the other hand, even though more than 300 Ganoderma triterpenoids have already been identified, very few Ganoderma triterpenoid saponins have been found [20,21]. Even though it is still not clear why G. lucidum does not biosynthesize Ganoderma triterpenoid saponins in vivo, developing ex vivo biotransformation strategies to produce novel Ganoderma triterpenoid saponins with strengthened bioactivities has become an issue of interest.
In nature, glycosylation of small molecules is usually catalyzed by Leloir glycosyltransferase (GT), which uses activated sugar, such as uridine diphosphate-glucose (UDP-G), as a sugar donor to glycosylate small molecules (sugar acceptors), such as flavonoids or triterpenoids [16][17][18][19]. Until now, dozens of plant GTs [22] and seven bacterial GTs [23] have been demonstrated to catalyze the glycosylation of triterpenoids. Among them, all identified plant or bacterial GTs belong to the GT1 family, except BtGT_16345, a novel GT28 from Bacillus thuringiensis GA A07, that regioselectively catalyzes C-15 glycosylation of GAA [24]. On the other hand, due to the high cost of UDP-G, some scientists focus on finding novel enzymes that use inexpensive sugar, such as starch, dextrin, maltose, or sucrose, as sugar donors to catalyze the glycosylation of small molecules [18,19,25]. Most of these novel enzymes belong to either glycoside-hydrolases (GH) family 13 (GH13), such as α-amylase (E.C. 3 [18,19,25]. As described above, producing Ganoderma triterpenoid saponins could be approached via synthetic biology. Furthermore, to the best of our knowledge, no study has attempted to produce Ganoderma triterpenoid saponins using GH enzymes yet. In the present study, three commercial and one recombinant GH13 enzymes were assayed for the ability to glycosylate Ganoderma triterpenoids using inexpensive sugar as sugar donors. Among them, cyclodextrin glucanotransferase from Thermoanaerobacter sp. (Toruzyme ® 3.0 L from Novozyme company) was confirmed to catalyze the glycosylation of GAA-15-Oβ-glucopyranoside (GAA-15-G), which was the glycosylated product of GAA biotransformed by BtGT_16345 [24]. The catalytic conditions for Toruzyme ® were optimized and the products were purified with preparative high-performance liquid chromatography (HPLC). Both the chemical structures and the aqueous solubility of the products were determined further.
On the other hand, the four GH13 enzymes were also reported to catalyze the glycosylation of flavonoid glucosides [30,33,34], steroidal saponins [26,31], and triterpenoid saponin ginsenoside [32]; therefore, the four enzymes were re-evaluated for the ability to catalyze the glycosylation of a GAA saponin, GAA-15-G, which can be biotransformed from GAA by BtGT_16345 [24]. During our test using a one-pot reaction, GAA was first biotransformed by BtGT_16345; then, the four GH13 enzymes were added to start the second biotransformation in a one-pot reaction. We found that only Toruzyme ® could biotransform GAA-15-G. Figure 1 shows the HPLC analysis of GAA (1), the first reaction product GAA-15-G biotransformed from GAA by BtGT_16345 (2), and the second reaction product GAA-G2 biotransformed sequentially from GAA by BtGT_16345 and then by Toruzyme ® with 2% maltose (3).
Catalysts 2021, 11, x FOR PEER REVIEW 3 of 12 saponin ginsenoside [32]; therefore, the four enzymes were re-evaluated for the ability to catalyze the glycosylation of a GAA saponin, GAA-15-G, which can be biotransformed from GAA by BtGT_16345 [24]. During our test using a one-pot reaction, GAA was first biotransformed by BtGT_16345; then, the four GH13 enzymes were added to start the second biotransformation in a one-pot reaction. We found that only Toruzyme® could biotransform GAA-15-G. Figure 1 shows the HPLC analysis of GAA (1), the first reaction product GAA-15-G biotransformed from GAA by BtGT_16345 (2), and the second reaction product GAA-G2 biotransformed sequentially from GAA by BtGT_16345 and then by Toruzyme® with 2% maltose (3). To increase the conversion of GAA-15-G into GAA-G2 by Toruzyme®, different sugars, sugar concentrations, reaction pH, and reaction temperatures were further evaluated. When using different sugars with 2% concentration, the highest conversion was observed for the reaction with maltose ( Figure 2a). We thus assayed different concentrations of maltose and found that maltose concentration dramatically affected the biotransformation of GAA-15-G by Toruzyme® (Figure 2b). An increase in maltose concentration from 2% to 20% resulted in an increase in both conversion and the types of products. Figure 1 (4) shows the HPLC analysis of the reaction products biotransformed from GAA-15-G by Toruzyme® with 20% maltose, where multiple derivatives, GAA-G2 to GAA-G4, were produced. In addition, conversion was stable between 50 °C and 60 °C ( Figure 2c) and exhibited optimal transformation activity at pH 5 ( Figure 2d) with 0.3 KNU (kilo novo units)/ml Toruzyme® and 20% maltose. Finally, the effects of reaction time on the conversion and the yields of different products were evaluated. The results showed that both the conversion and yields of the products increased quickly during 0 to 4 hr, and then reached To increase the conversion of GAA-15-G into GAA-G2 by Toruzyme ® , different sugars, sugar concentrations, reaction pH, and reaction temperatures were further evaluated. When using different sugars with 2% concentration, the highest conversion was observed for the reaction with maltose ( Figure 2a). We thus assayed different concentrations of maltose and found that maltose concentration dramatically affected the biotransformation of GAA-15-G by Toruzyme ® (Figure 2b). An increase in maltose concentration from 2% to 20% resulted in an increase in both conversion and the types of products. Figure 1 (4) shows the HPLC analysis of the reaction products biotransformed from GAA-15-G by Toruzyme ® with 20% maltose, where multiple derivatives, GAA-G2 to GAA-G4, were produced. In addition, conversion was stable between 50 • C and 60 • C ( Figure 2c) and exhibited optimal transformation activity at pH 5 ( Figure 2d) with 0.3 KNU (kilo novo units)/mL Toruzyme ® and 20% maltose. Finally, the effects of reaction time on the conversion and the yields of different products were evaluated. The results showed that both the conversion and yields of the products increased quickly during 0 to 4 h, and then reached maximum conversion and yields from 4 to 8 h ( Figure 3). In summary, the optimal reaction condition for Toruzyme ® catalyzing GAA-15-G was 20% maltose, pH 5 at 60 • C for 6 h, where conversion was 58.9% and the yields of GAA-G2 to GAA-G4 were 27.8%, 17.5%, and 7.8%, respectively.

Purification and Identification of the Biotransformed Products
To determine the chemical structures of the products, the reaction was scaled-up to 30 ml, and the products were purified using preparative HPLC method. The molecular weights and the structures of the purified products were then determined by mass and mass-mass spectral analysis. The mass detector showed that the molecular weights  Figure S6). Thus, GAA-G2 to GAA-G4 were diglucosyl, triglucosyl, and tetraglucosyl GAA saponins, respectively. The structure of GAA-G2 was further determined by nucleic magnetic resonance (NMR) spectral analysis. The 1 H and 13 C NMR, including distortionless enhancement by polarization transfer (DEPT), heteronuclear single quantum coherence (HSQC), heteronuclear multiple bond connectivity (HMBC), correlation spectroscopy (COSY), and nuclear Overhauser effect spectroscopy (NOESY), spectra were obtained (shown in Figures S7-S13). In addition to the signals of the GAA, fourteen proton signals (from 3.77 to 5.88 ppm) and twelve carbon signals (from 61.9 to 105.0 ppm) corresponding to two glucose moiety structures were observed. The first glucosyl moiety connected to GAA was confirmed according to the NOE connectivity between its anomeric proton H-1' at δ

Purification and Identification of the Biotransformed Products
To determine the chemical structures of the products, the reaction was scaled-up to 30 mL, and the products were purified using preparative HPLC method. The molecular weights and the structures of the purified products were then determined by mass and mass-mass spectral analysis. The mass detector showed that the molecular weights [M-H] − of compounds GAA-G2 to GAA-G4 were 839.5 ( Figures S1 and S2), 1001.5 (Figures S3 and S4), and 1163.6 ( Figures S5 and S6) m/z, respectively. Therefore, the mass data supported the implication observed in the HPLC profile that Toruzyme ® biotransformed GAA saponins by increasing one glucose residue (m/z = 162, [Glucose-OH] − ) at a time. Moreover, the mass-mass data also showed the corresponding molecules with glucoside breakdown in each GAA saponin's analysis. In GAA-G2 mass-mass analysis, both GAAglucoside (m/z 677.  Figure S6). Thus, GAA-G2 to GAA-G4 were diglucosyl, triglucosyl, and tetraglucosyl GAA saponins, respectively. The structure of GAA-G2 was further determined by nucleic magnetic resonance (NMR) spectral analysis. The 1 H and 13 C NMR, including distortionless enhancement by polarization transfer (DEPT), heteronuclear single quantum coherence (HSQC), heteronuclear multiple bond connectivity (HMBC), correlation spectroscopy (COSY), and nuclear Overhauser effect spectroscopy (NOESY), spectra were obtained (shown in Figures S7-S13). In addition to the signals of the GAA, fourteen proton signals (from 3.77 to 5.88 ppm) and twelve carbon signals (from 61.9 to 105.0 ppm) corresponding to two glucose moiety structures were observed. The first glucosyl moiety connected to GAA was confirmed according to the NOE connectivity between its anomeric proton H-1 at δ  Table S1. The compound GAA-G2 was thus confirmed as GAA-15-O-[αglucopyranosyl-(1→4)-β-glucopyranoside]. The results showed that the glycosylation of GAA-15-G by Toruzyme ® was the same as the starch linkage. Figure 4 summarizes the sequential biotransformation process of GAA by BtGT_16345 and Toruzyme ® .  Table S1. The compound GAA-G2 was thus confirmed as GAA-15-O-[α-glucopyranosyl-(1→4)-β-glucopyranoside]. The results showed that the glycosylation of GAA-15-G by Toruzyme® was the same as the starch linkage. Figure 4 summarizes the sequential biotransformation process of GAA by BtGT_16345 and Toruzyme®.

Determination of Aqueous Solubility of the Products
It has been reported that glycosylation of small molecules would increase the aqueous solubility of the molecule. Therefore, the aqueous solubility of the major product, GAA-G2, was determined by using the HPLC method. Table 1 shows that the aqueous solubility of GAA-G2 was over the maximal detection (70,000 mg/L) in this study, which was higher than 4554-fold of that of GAA.  1 The fold of aqueous solubility of GAA glucoside derivatives is expressed as relative to that of GAA, normalized to 1.

Discussion
Glycosylation of small molecules not only increases the aqueous solubility [10-13] but may also improve the bioavailability of the molecule [14,15]. In addition, different types, numbers, and linkage sites of sugars may also affect the bioactivities of the molecule [7][8][9][10]. Even though G. lucidum possesses hundreds of triterpenoids, very few triterpenoid saponins have been discovered [20,21]. Thus, applying glycosylation on Ganoderma triterpenoids may be helpful in the synthesis of biomedicine derivatives.

Determination of Aqueous Solubility of the Products
It has been reported that glycosylation of small molecules would increase the aqueous solubility of the molecule. Therefore, the aqueous solubility of the major product, GAA-G2, was determined by using the HPLC method. Table 1 shows that the aqueous solubility of GAA-G2 was over the maximal detection (70,000 mg/L) in this study, which was higher than 4554-fold of that of GAA. GAA-G2 >70,000 >4554.3 This study 1 The fold of aqueous solubility of GAA glucoside derivatives is expressed as relative to that of GAA, normalized to 1.

Discussion
Glycosylation of small molecules not only increases the aqueous solubility [10][11][12][13] but may also improve the bioavailability of the molecule [14,15]. In addition, different types, numbers, and linkage sites of sugars may also affect the bioactivities of the molecule [7][8][9][10]. Even though G. lucidum possesses hundreds of triterpenoids, very few triterpenoid saponins have been discovered [20,21]. Thus, applying glycosylation on Ganoderma triterpenoids may be helpful in the synthesis of biomedicine derivatives. There are two gene families of enzymes catalyzing the glycosylation of small molecules: GT [16,17,22,23] and GH [18,19,25]. Using low-cost sugar as sugar donors on GH enzymes is considered more cost-effective than GTs, which requires expensive sugar donors such as UDP-G. Although some GH13 or GH70 enzymes have been reported to catalyze the glycosylation of simple phenolics or flavonoids [18,19,25], no GHs have been found to catalyze the glycosylation of triterpenoid aglycones yet. This study also showed that none the four selected GH13 enzymes could catalyze the glycosylation of GAA [18,19,25]. Hence, designing synthetic strategies via GH enzymes to catalyze the glycosylation of triterpenoid aglycones has become a valuable strategy.
Previous reports have described the glycosylation of flavonoid glycosides by amylosucrase [33,34] or Toruzyme ® [30]; the glycosylation of steroidal saponins by α-amylases [26] or Toruzyme ® [26,31]; and the glycosylation of Ginseng triterpenoid saponin by Toruzyme ® [32]. In the present study, only Toruzyme ® could biotransform Ganoderma triterpenoid saponin, GAA-15-G (Figure 1). The result revealed that Toruzyme ® catalyzed glycosylation toward a broader substrate range than that of amylosucrase or α-amylase. In addition, previous studies have shown that Toruzyme ® catalyzed the α-(1→4) glycosylation linkage on the sugar acceptors of flavonoid glucoside [30] or steroidal saponins [26,31]. The result of the present study showing that Toruzyme ® catalyzed α-(1→4) glycosylation linkage on the sugar acceptor of GAA-15-G (Figure 4) was consistent with the previous studies. Meanwhile, other scientists also conducted a similar one-pot enzymatic reaction through cyclodextrin glucanotransferase stepwise with other enzymes to produce novel fine chemicals. Koh et al. combined amylosucrase and cyclodextrin glucanotransferase to produce cyclodextrin from sucrose [36]. Gonzalez-Alfonso et al. also combined cyclodextrin glucanotransferase and amyloglucosidase to produce pterostilbene monoglucoside [28]. Shimoda et al. combined a plant glucosyltransferase and cyclodextrin glucanotransferaase to produce resveratrol glucosides [37]. Thus, cyclodextrin glucanotransferase in producing novel glucosylated products could be developed through combinations with other enzymes in the future.
There are several advantages of combining BtGT_16345 and Toruzyme ® in catalyzing the glycosylation of triterpenoids. Firstly, most of the microbial GTs proven to catalyze the glycosylation of triterpenoids have less regioselectivity toward the triterpenoid substrates compared with the BtGT_16345 used in the present study. For example, UGT109A1 from Bacillus subtilis catalyzes the glycosylation of ginsenosides at either C-3, C-12, or C-20 [38,39]; BsYjiC from B. subtilis catalyzed the glycosylation of ginsenosides at either C-3, C-6, C-12, or C-20 [40,41]. In contrast, BtGT_16345 from B. thuringiensis regioselectively catalyzed C-15 glycosylation of GAA among its multiple glycosylation sites [24]. Secondly, present and previous studies all show that Toruzyme ® catalyzes chain-growth α-(1→4) glycosylation linkages toward the sugar acceptor of the saponins [26,30,31]. Accordingly, combining BtGT_16345 and Toruzyme ® achieved the specific glycosylation of GAA with maltogenic, maltotriogenic, and so on, sugar moieties. Finally, an additional advantage coming from the combination of a retaining α-glycosidase (Toruzyme ® ) with a glycosyltransferase (BtGT_16345) adding the first sugar unit by a β bond is that it generates saponins resistant to α-glycosidase, hence stable in the following reaction process. Therefore, the strategy could be applied to other triterpenoids in producing regioselectively polyglycosylated triterpenoid saponins.
According to the literature [26][27][28][29][30][31][32], the reaction conditions for transglycosylation catalyzed by Toruzyme ® , including various sugar donors, pH 5-7, and 50-60 • C, have been assessed. Here, we designed our experiments seeking the optimal reaction conditions based on the literature. First, various sugar donors, including soluble starch, maltose, dextrin, and cyclodextrin, were tested in the reaction conducted by 0.3 KNU/mL of Toruzyme ® in 50 mM acetate buffer of pH 5, at 60 • C, for 4 h. Among these sugar donors, maltose gave the highest conversion. Therefore, different concentrations of maltose were evaluated next in the same conditions mentioned above. Based on the result, the reaction condition with the best conversion using 20% (w/v) maltose was conducted, except that the reaction temperatures were different. Finally, the reaction buffers with different pH values were tested in the optimized reaction conditions established above. From the results (Figure 2), the optimal reaction conditions, pH 5 and reaction temperature 50 to 60 • C, are similar to the literature. The only factor has not been assessed previously is that the maltose concentrations in the range of 2-40% (w/v) affect significantly the reaction efficiency of Toruzyme ® . Here, we found that increasing the maltose concentration from 2% to 20% would dramatically increase the conversion yields, as well as producing more derivatives during the biotransformation of GAA-15-G by Toruzyme ® (Figure 2). The optimal molar ratio between maltose and GAA-15-G is 600 (for 20% maltose: 584.3 mM of maltose/0.97 mM of GAA-15-G), which was dramatically higher than similar studies using Toruzyme ® as a biocatalyst [26][27][28][29][30][31][32]. Choung et al. used maltose (1% w/v and 6.6 molar ratio) as a sugar donor to conduct the glycosylation reaction, where the sugar acceptor was astragalin, a flavonoid glucoside [30]. Zhou et al. [26], Wang et al. [31], and Moon et al. [32] used 0.3-0.7% dextrin on the glycosylation of steroidal saponins or Ginseng saponins, but the molar ratio between dextrin and the sugar acceptors was below 10. Gonzalez-Alfonso et al. studied the glycosylation of epigallocatechin gallate (EGCG) by Toruzyme ® and found that changing the sugar donor (starch) concentration from 1% to 30% did not alter the glycosylation activity [27]. The possible reason that previous studies did not find positive effects on Toruzyme ® activity by the high concentrations of sugar donors is that starch or dextrin are low-soluble sugars (less than 2% in our experience). In contrast, maltose is highly soluble, in which its aqueous solubility can reach 40% or beyond. This explains why the effect of maltose concentration on the glycosylation activity by Toruzyme ® could be observed in our study. Our results suggest that high concentrations of maltose can potentially be applied in the glycosylation of small molecules by Toruzyme ® .
It is well known that glycosylation would increase the aqueous solubility of the modified molecules [10,13,33]. This study revealed that the aqueous solubility of the Ganoderma saponin GAA-G2 was over 4554-fold higher than that of GAA, whereas the aqueous solubility of another GAA saponin, GAA-15,26-diglucoside, was previously found to be 1024-fold higher than that of GAA (Table 1) [35]. It is intriguing that the solubility of GAA-15,26-diglucoside was lower than that of GAA-G2 (diglucoside isomer). Nevertheless, the comparison showed that not only the number of side chain sugars but also the position of sugars would affect the aqueous solubility of the modified derivatives.
Herein, a series of Ganoderma triterpenoid saponins were produced in a one-pot reaction by stepwise reactions of BtGT_16345 from Bacillus thuringiensis GA A07 and Toruzyme ® from Thermoanaerobacter sp. The produced GAA triterpenoid saponins possess higher aqueous solubility than GAA. The strategy could be applied to other Ganoderma triterpenoids in producing more novel Ganoderma triterpenoid saponins. Moreover, the produced novel Ganoderma saponins might have unique bioactivities that could benefit the medicinal industry.

Biotransformation
For biotransformation of GAA by α-amylase from A. oryzae (Sigma 10065), 50 mM phosphate buffer (PB) of pH 6, 20% maltose, 3400 U/mL enzyme, and 1 mg/mL of GAA were incubated at 30 • C for 4 h. For biotransformation of GAA by α-amylase from B. licheniformis (Sigma A3306), 50 mM PB of pH 6, 20% maltose,~4000 U/mL enzyme, and 1 mg/mL of GAA were incubated at 60 • C for 4 h. For biotransformation of GAA by DgAS, 50 mM PB of pH 7, 20% sucrose, 10 µg/mL enzyme, and 1 mg/mL of GAA were incubated at 30 • C for 4 h. For biotransformation of GAA by Toruzyme ® , 50 mM acetate buffer of pH 5, 20% maltose, 10% (v/v) enzyme, and 1 mg/mL of GAA were incubated at 60 • C for 4 h. After the reactions, the reaction mixtures were mixed with equal volume of methanol and analyzed with HPLC.
The biotransformation of GAA by the combined actions of BtGT_16345 and Toruzyme ® was performed in a one-pot solution via two sequential steps. First, the biotransformation was conducted in 0.1 mL reaction mixture containing 1 mg/mL GAA, 15 µg/mL BtGT_16345, 10 mM MgCl 2 , and 10 mM UDP-G at pH 7.4 (10 mM PB). The reaction was performed at 30 • C for 30 min. Afterwards, the reaction mixture was mixed with equal volume of second reaction buffer containing 2% different sugar and 0.3 KNU/mL Toruzyme ® in 50 mM of acetate buffer at pH 5. The second reaction was performed at 60 • C for 4 h and then analyzed with HPLC. For optimization experiments of pH, 50 mM of acetate buffer (pH 5), phosphate buffer (pH 6 and pH 7), and Tris buffer (pH 8) were used. For optimization experiments of sugar concentration, 2-40% of maltose was used.

High-Performance Liquid Chromatography (HPLC)
HPLC was performed with an Agilent ® 1100 series HPLC system (Santa Clara, CA, USA) equipped with a gradient pump (Waters 600, Waters, Milford, MA, USA). The stationary phase was a C18 column (Sharpsil H-C18, 5 µm, 4.6 i.d. × 250 mm, Sharpsil, Beijing, China), and the mobile phase was 1% acetic acid in water (A) and methanol (B). The elution condition was a linear gradient from 0 min with 40% B to 20 min with 70% B; isocratic elution from 20 min to 25 min with 70% B; linear gradient from 25 min with 70% B to 28 min with 40% B; and isocratic elution from 28 min to 35 min with 40% B. All elutes were at a flow rate of 1 mL/min. Sample volume was 10 µL. The detection condition was set at 254 nm.

Purification and Identification of the Glycosylated Products
First, the reaction mixture (30 mL containing 1 mg/mL GAA, 15 µg/mL BsGT110, 10 mM UDP-glucose, 10 mM MgCl 2 , 10 mM phosphate buffer at pH 7.4) was incubated at 30 • C for 30 min. Next, 30 mL of 50 mM acetate buffer (pH 5) containing 20% maltose and 0.3 KNU/mL Toruzyme ® was added into the reaction mixture to start the second biotransformation. After 4 h incubation at 60 • C, the reaction was stopped by adding 60 mL of methanol into the mixture. Then, the reaction mixture, 120 mL with 50% methanol, was resolved by the preparative YL9100 HPLC system (YoungLin, Gyeonggi-do, Korea). In this system, Inertsil ODS 3 column (10 mm, 20 i.d. × 250 mm, GL Sciences, Eindhoven, The Netherlands) was the stationary phase. The same mobile phase as that in the HPLC system, but with a flow rate of 15 mL/min was applied. 5 mL sample was injected each time and the products were detected by optical absorbance at 254 nm. After products collected, they were concentrated by a vacuum pump then lyophilized by a freeze dryer. Finally, 9.39 mg and 4.41 mg of the products, GAA-G2 and GAA-G3, were purified, while the amount of the purified GAA-G4 was less than 1 mg. The chemical structures of the products were determined with mass, mass-mass, and NMR spectral analyses. The Finnigan LCQ Duo mass spectrometer (ThermoQuest Corp., San Jose, CA, USA) with ESI was used for the mass spectral analysis. The Bruker AV-700 NMR spectrometer (Bruker Corp., Billerica, MA, USA) recorded 1 H-and 13 C-NMR, DEPT, HSQC, HMBC, COSY, and NOESY spectra at ambient temperature. The NMR experiments used standard pulse sequences and parameters (Topspin program 4.0), and reported all chemical shifts in parts per million (ppm, δ).

Determination of Aqueous Solubility of the Glycosylated Products
Aqueous solubility and stability were assessed with the methods that we used in a previous study [12]. For aqueous solubility assay, 7 mg of GAA-G2 was vortexed in 0.1 mL of distilled deionized (d.d.) H 2 O for 10 min at 25 • C and then centrifuged at 14,000× g. The supernatant was filtrated through a 0.2 µm nylon membrane, diluted 200-fold with 50% of methanol, and then analyzed with HPLC. The concentrations of the tested compounds were calculated based on their peak areas using calibration curves prepared with HPLC analyses of the same compounds dissolved in dimethyl sulfoxide (DMSO). After assay, all GAA-G2 in water was recovered by lyophilization with a freeze dryer and the assay was repeated once.