Expanded investigations of the aglycon promiscuity and catalysis characteristic of flavonol 3-O-rhamnosyltransferase AtUGT78D1 from Arabidopsis thaliana

Abstract

Rhamnosides usually possess better bioavailabilities and improved solubilities compared with their aglycons and are a major source of bioactive natural products. However, biosynthesis of rhamnosides is hindered by the commercially expensive UDP-rhamnose (UDP-Rha) donor and a lack of universal rhamnosyltransferases. In the present study, an efficient UDP-Rha production system via a two-step enzymatic reactions using UDP-glucose (UDP-Glc) as a substrate was constructed. Extensive in vitro enzymatic assays and preparative reactions using the obtained UDP-Rha/UDP-Glc highlighted the robust glycosylation promiscuity of the reported rhamnosyltransferase AtUGT78D1. Based on HPLC-UV and HR-MS analyses, 30 of the tested aromatic compounds belonging to 7 structural types, including flavonoids, flavonoid glycosides, phenylethyl chromones, benzophenones, coumarins, lignanoids, and anthraquinones, were accepted by AtUGT78D1 to conduct the corresponding rhamnosylation and/or glucosylation with one or more glycosyl substitutions at different positions. Further preparative reactions expanded the catalytic characteristic of AtUGT78D1 since it can catalyse the rhamnosylation at the 3-OH position of the flavonols, glucosylation at the 7-OH position of the flavone baicalein, and multiple hydroxyl substitutions for diverse types of aromatics. Interestingly, a unique reversible catalysis activity of AtUGT78D1 was observed, and it has been effectively used in one-pot rhamnosylation of the desired rhamnoside. The enzymatic rhamnosylations of diverse “drug-like” scaffolds as well as bidirectional catalysis for one-pot rhamnosylations by plant rhamnosyltransferase were rarely reported before, which indicated that AtUGT78D1 was expected to be a universal and effective tool for chemo-enzymatic synthesis of diverse bioactive rhamnosylated derivatives for drug discovery.

---

1. Introduction

Glycosylated aromatics, which comprise multiple types of secondary metabolites, are widely distributed in nature and often possess various biological properties, such as anti-cancer, anti-psychotic, anti-inflammatory, and anti-diabetic activities.^1,2 The remarkable properties of these compounds are thought to reside in their increased aqueous solubilities, stabilities, and enhanced bioavailabilities, which are likely a result of the presence of a bulky hydrophilic sugar moiety.^3,4 The glycosylation of aromatics not only contributes significantly to the structural diversity of these valuable natural products, but it may also be a key mechanism in determining the structural complexity and remarkable biological activities compared with their non-glycosylated analogs.^5,6 Thus, exploring the rich pool of bioactive glycosylated aromatics for the design of new candidate drugs is a compelling target for both basic research and pharmacological applications. However, many potentially useful glycosylated compounds occur only in trace amounts in natural sources and are not easily amenable to chemical synthesis. In the biosynthesis of bioactive natural glycosides, glycosyltransferases (GTs), which transfer one or more sugar moieties onto aromatic acceptor molecules, are often involved in the last step modification of an aglycon, leading to the corresponding ultimate bioactive molecules. Thus, the manipulation of glycosylation reactions catalysed by GTs might represent an advantageous shortcut for the synthesis of glycosylated aromatic compounds.

In plants, glycosylation is catalysed by family I glycosyltransferases, which commonly utilize small molecular compounds as acceptor substrates and UDP-sugars as donors.⁷ These enzymes are characterized by the presence of a C-terminal consensus sequence comprised of 44 amino acid residues that is termed as the plant secondary product glycosyltransferase (PSPG) box.⁸ In recent years, a large number of genes encoding uridine diphosphate (UDP)–glycosyltransferases (UGTs) have been identified in various plant species to glycosylate multiple plant secondary metabolites like flavonoids, terpenoids, steroids, xanthonoids, etc.⁹ The sugar moieties recognized by these UGTs are diverse, including UDP-glucose (UDP-Glc, the most common sugar donor), UDP-glucuronic acid (UDP-GlcA), UDP-rhamnose (UDP-Rha), UDP-xylose (UDP-Xyl), UDP-galactose (UDP-Gal) and so on.

UDP-Rha is a ubiquitous sugar present in plants and many bioactive leading compounds contain a rhamnose moiety in their structures, e.g., the antimicrobial agent quercetin 3-O-rhamnoside,¹⁰ kaempferol 3-O-rhamnoside¹¹ and icariin.¹² Rhamnosylation substitution plays an important role in the structural stability, solubility improvement, intracellular transport, and bioavailability regulation of these natural products. However, compared with the mostly reported glucosyltransferase, rhamnosyltransferase, which is involved in the enzymatic synthesis of rhamnoside products, has not been explored extensively because UDP-Rha is commercially expensive and its chemical synthesis is multistep, time-consuming and low-yielding.^13,14 Combinatorial biosynthesis of UDP-Rha synthase and rhamnosyltransferase would offer an alternative source for bioactive rhamnosides. As one of the model plants, about 120 glycosyltransferase genes belonging to the GTI family were annotated in Arabidopsis thaliana. Among them, two rhamnosyltransferases, AtUGT78D1 and AtUGT89C1, were functionally characterized to catalyse the O-rhamnosylation of flavonol at the 3-OH or 7-OH position, respectively.¹⁵ Further investigation of AtUGT89C1 revealed its donor (TDP-rhamnose and UDP-rhamnose) and acceptor (various classes of flavonoids including flavonols, flavones and flavanones) substrate flexibilities via microbial biotransformation to generate a library of flavonoid 7-O-rhamnosides. This offered proof that plant UGTs may have potential catalysis promiscuities towards sugar donors and/or acceptors to be used in universal glycosylation of natural products based on many identified promiscuous UGTs.^16–18

As far as AtUGT78D1, it was identified as a flavonol-specific glycosyltransferase that is responsible for transferring rhamnose or glucose to the 3-OH position of flavonol in vitro. In recent reports, it has been successfully used in the production of flavonol 3-O-glucosides or 3-O-rhamnosides in engineered bio-catalysis systems.^19–22

In this article, aiming to thoroughly investigate the substrate flexibilities of AtUGT78D1 in enzymatic rhamnosylation of aromatic natural products, an efficient UDP-Rha production system using a two-step enzymatic reaction catalysed by A. thaliana rhamnose synthase 2 (AtRHM2) with UDP-Glc as a substrate was constructed (Scheme 1). The obtained UDP-Rha product was subsequently used in the in vitro enzymatic assay performed by AtUGT78D1 with a wide range of aromatic substrates as glycosyl acceptors. The expanded assays revealed the significant substrate promiscuities of AtUGT78D1 and 30 of the tested aromatic compounds belonging to 7 structural types, including flavonoids (flavones, flavonones, flavonols, isoflavones, chalcones, methylflavone), flavonoid glycosides (both mono- and di-glycosides), benzophenones, coumarins, lignanoids, anthraquinones, were accepted by AtUGT78D1 as glycosyl acceptors to conduct the corresponding rhamnosylation using UDP-Rha as a glycosyl donor. Among them, 10 flavonoids were also accepted by AtUGT78D1 using UDP-Glc as a glycosyl donor, and the rhamnosylation efficiency was much higher than the glucosylation efficiency. Besides, it was also worth mentioning that the enzymatic rhamnosylation by recombinant proteins of methylflavone, phenylethyl chromones, coumarins, lignanoids and anthraquinones were rarely reported before. Considering that AtUGT78D1 has been described as a flavonol specific glycosyltransferase at the 3-OH position, two flavonol compounds, quercetin (7) and kaempferol (10), along with a flavone baicalein, which has a relatively high conversion yield, were picked out to be used in the preparative scale enzymatic synthesis. The corresponding glycosylated products were separated and identified. Structure elucidation revealed that AtUGT78D1 could transfer not only the rhamnose moiety onto the 3-OH position of flavonols but also the glucose moiety onto the 7-OH position of the flavone baicalein (1). Besides, the reaction reversibility of AtUGT78D1 was observed to catalyse the bidirectional rhamnosylation in the one-pot reaction to produce the diverse, desired bioactive rhamnosides without adding the expensive UDP-Rha. The remarkable promiscuity of AtUGT78D1 towards diverse aromatic acceptors endows it with a versatile tool for chemo-enzymatic synthesis of diverse rhamnosylated aromatic derivatives. It also can act as a potentially general functional part for secondary metabolism pathway design and construction in synthetic biology of bioactive natural and unnatural products for lead compound discovery in drug R&D.

Scheme 1 UDP-rhamnose biosynthetic pathways from UDP-glucose in A. thaliana. RHM2-N has the first activity as an UDP-D-glucose 4,6-dehydratase, while RHM2-C has the two following UDP-4-keto-6-deoxy-D-glucose 3,5-epimerase, and UDP-4-keto-L-rhamnose 4-keto-reductase activities.^23–25

2. Materials and methods

2.1. Chemicals

UDP-glucose and NAD⁺ were purchased from Sigma-Aldrich (St. Louis, USA). NADPH was purchased from Biodee Biotechnology Co., Ltd. (Beijing, China), and various substrates tested for the enzymatic reaction were purchased from Chengdu Push Biotechnology Co., Ltd. (Chengdu, China) and Chengdu Biopurify Phytochemicals Ltd. (Chengdu, China).

2.2. Gene cloning and heterogenous expression of AtRHM2-N, AtRHM2-C and AtUGT78D1

The total RNA of A. thaliana was isolated by Plant RNA Kit (OMEQA Bio-Tek, Inc. GA, USA). All PCR were done using high fidelity KOD-Plus-Neo DNA polymerase (TOYOBO, Osaka, Japan). The AtRHM2-N, AtRHM2-C and AtUGT78D1 genes were amplified by RT-PCR based on the reported mRNA sequences described before and using cDNA of A. thaliana as a template with the primer pairs as follows: AtRHM2-N-BamHI-F (5′-GGA TCC ATG GAT GAT ACT ACG TAT AAG C-3′), AtRHM2-N-XhoI-R (5′-CTC GAG TCA TAC AAC CGT AAA TGT CTG G-3′); AtRHM2-C-BamHI-F (5′-GGA TCC ATG ACA CCT AAG AAT GGT G-3′), AtRHM2-C-XhoI-R (5′-CTC GAG GGT TCT CTT GTT TGG TTC-3′); AtUGT78D1-EcoRI-F (5′-GAA TTC ATG ACC AAA TTC TCC GAG CC-3′), AtUGT78D1-XhoI-R (5′-CTC GAG AAC TTT CAC AAT TTC GTC C-3′). Underlining indicates restriction sites (BamHI, EcoRI and XhoI, respectively). The amplified DNA fragments were digested with the corresponding restriction enzymes (Takara, Dalian, China) and inserted into the counterpart sites in the pET-28a vector (Novagen). The fragments of the vector and genes were recovered with a Nucleic Acid Purification Kit (Axygen, Union City, USA) and then ligated with T₄ DNA ligase (BioLabs, Beijing, China) at 16 °C for 6 hours. The recombinant plasmid was transformed into Escherichia coli BL21 (DE3) (TransGen, Beijing, China) before being confirmed.

2.3. Expression and purification of recombinant proteins of AtRHM2-N, AtRHM2-C and AtUGT78D1

A single isolated bacterial colony was inoculated in a liquid LB medium supplemented with kanamycin (50 mg L⁻¹) and grown for 5 h at 37 °C, 200 rpm. A portion of the cultured cells was transferred into fresh LB liquid medium supplemented with the same antibiotics at a ratio of 1 : 100 and then grown at 37 °C, 200 rpm until the optical density at 600 nm reached 0.4–0.6. Recombinant protein expression was induced by addition of 1.0 mM isopropyl β-D-thiogalactopyranoside (IPTG) and grown for 16 h at 25 °C, 200 rpm. Cells were harvested by centrifugation at 7000 × g for 20 min before being suspended in a lysis buffer (20 mM phosphate buffer, 20 mM imidazole, 500 mM NaCl, 0.5 mM phenylmethylsulfonyl fluoride, 10% glycerol, pH 7.4). Cells were lysed in an ice bath with an Ultrasonic Processor (VERNON HILLS, Illinois, USA) equipped with a microtip probe using 200 sonication. Supernatants containing soluble RNM2-N, RNM2-C and AtUGT78D1 proteins were filtered through 0.45 μm Millex Syringe-driven Filter Units (Merck KGaA, Darmstadt, Germany) after centrifuging at 10 000 × g for 40 min at 4 °C.

The supernatants of the proteins were purified by the Histrap column (GE Healthcare, Uppsala, Sweden) previously equilibrated in a binding buffer (20 mM phosphate buffer, 20 mM imidazole, 500 mM NaCl, 3% (v/v) glycerol, pH 7.4). After loading the protein solution, the column was rinsed with binding buffer to remove the unbound contaminant proteins. The bound His-tagged protein was then eluted with an elution buffer (20 mM phosphate buffer, linear gradient of 20–500 mM imidazole, 500 mM NaCl, 3% (v/v) glycerol, pH 7.4). The fractions containing the targeted proteins were concentrated in Centrifugal Filter Units (Merck KGaA, Darmstadt, Germany) at 4500 × g, 4 °C. The obtained supernatant was subsequently diluted with a desalting buffer (50 mM Tris–HCl, 100 mM NaCl, 1 mM DTT, 3% (v/v) glycerol, pH 7.4) using a PD-10 column (GE Healthcare, Uppsala, Sweden), and the obtained enzymes were stored at −80 °C. The purity of the targeted protein was detected by SDS-PAGE to be >90% (Fig. S2†), and the protein concentration for all studies was determined as described before.

2.4. Chemo-enzymatic synthesis, HPLC-UV and HR-MS analysis of UDP-Rha

UDP-Rha was enzymatically synthesized by a two-step reaction using UDP-Glc as a substrate. For this purpose, the UDP-glucose-4,6-dehydratase (AtRHM2-N), UDP-4-keto-6-deoxyglucose-3,5-epimerase and UDP-4-keto-rhamnose-4-keto-reductase domains (AtRHM2-C) of A. thaliana UDP-Rha synthase were heterogeneously expressed. For the first step in the reaction, UDP-4-keto-6-deoxy-glucose was synthesized using 1 mM UDP-Glc, 30 μg of AtRHM2-N, 0.5 mM NAD⁺, and 10 mM Tris–HCl (pH 8.5) in a total volume of 2.0 mL. After overnight incubation at 25 °C, the reaction was stopped by heat inactivation for 10 min at 95 °C. The volume of the reaction mixture was then adjusted to 10.0 mL, containing 1 mM NADPH, 60 μg of AtRHM2-C, and 10 mM Tris–HCl (pH 8.5). After overnight incubation at 25 °C, the reaction was stopped by heat inactivation for 10 min at 95 °C.

The obtained UDP-Rha product was subsequently analysed using a CAPCELL PAK ADME S5 column (4.6 mm I.D. × 250 mm, SHISEIDO Co., Ltd, Tokyo, Japan) eluted with ammonium formate (5 mM) over 10 min at a flow rate of 0.3 mL min⁻¹, using an Agilent 1260 Series HPLC system equipped with an autosampler, diode array detector and ChemStation software. During chromatographic runs, the injection volume was 5.0 μL, the plate cooler temperature was set to 4 °C, and column compartment was 25 °C. Nucleotide sugars were detected at 260 nm and coupled directly to the mass spectrometer for analysis. Ultra-high purity helium (He) was used as the collision gas, and high purity nitrogen (N₂) was used as the nebulizing gas. The optimized ESI† source parameters were as follows: sheath gas flow rate, 1.5 mL min⁻¹; auxiliary gas flow rate, 1.5 mL min⁻¹; spray voltage, 4.5 kV; capillary temperature, 200 °C. The spectra were recorded in the 100–1000 m/z range for a full scan MS analysis.

2.5. Expanded enzymatic assays for substrate flexibility of AtUGT78D1

The reaction mixture routinely contained 0.4 mM aromatic substrate, 0.8 mM UDP-Glc or UDP-Rha and 60 μg AtUGT78D1 proteins in 200 μL reaction buffer (50 mM Tris–HCl, 100 mM NaCl, 1 mM DTT, 1.5% (v/v) glycerol, pH 7.4). Assays with the heat-inactivated enzyme were used as negative controls. After incubation for 16 h at 30 °C, the reaction was terminated by the addition of 400 μL of ethyl acetate and then vortexed. The supernatant was separated and evaporated to dryness under reduced pressure. The resultant residues were re-dissolved in methanol and centrifuged at 12 000 × g for 30 min. The supernatant was analysed by HPLC-UV and HR-ESI-MS. HPLC analyses were performed on a CAPCELL PAK C₁₈ column (4.6 mm I.D. × 250 mm, SHISEIDO Co., Ltd, Tokyo, Japan) at a flow rate of 1 mL min⁻¹, and the column temperature was maintained at 30 °C. The mobile phase was methanol (A) and 0.1% formic acid in H₂O (B). The gradient elution A : B (v/v) was as follows: 0 min (30 : 70)–25 min (90 : 10)–27 min (100 : 0).

2.6. In vitro preparation of enzymatic products by AtUGT78D1

The preparative scale reactions contained 0.03 mmol aromatic substrate (baicalein 1, quercetin 7 and kaempferol 10), 0.06 mmol UDP-Glc or UDP-Rha and 35 mg AtUGT78D1 proteins in 50 mL reaction buffer (50 mM Tris–HCl, 100 mM NaCl, 1 mM DTT, 1.5% (v/v) glycerol, pH 7.4). After incubation for 16 h at 30 °C, the reaction supernatant were separated by column chromatography with a macroporous resin (MCI GEL CHP) after it was centrifuged at 12 000 × g for 30 min. The mobile phase was a gradient elution of 100% water to 100% methanol, and each aliquot was analysed by HPLC-UV. Aliquots including the targeted product were evaporated to dryness under reduced pressure. The resultant residues were re-dissolved in methanol and purified by reverse phase semi-preparative HPLC, which was performed on a YMC-Pack ODS-A HPLC column (10 mm I.D. × 250 mm, S-5 μm, 12 nm, Co., Ltd, Tokyo, Japan) at a flow rate of 3 mL min⁻¹. The column temperature was maintained at 25 °C. The mobile phase was methanol (A) and water (B). The gradient elution A : B (v/v) was as follows: 0 min (30 : 70)–10 min (90 : 10)–20 min (100 : 0). Then, the obtained products were confirmed by ¹H and ¹³C nuclear magnetic resonance or followed by HMBC (heteronuclear multiple bond correlation).

2.7. One-pot reactions catalysed by AtUGT78D1

The one-pot reaction was composed of 0.4 mM quercitrin (18), 0.4 mM kaempferol (10), 0.4 mM UDP and 60 μg purified AtUGT78D1 in a total volume of 200 μL. After incubation at 30 °C for 12 h, the reaction was terminated with 400 μL of ice-cold methanol and centrifuged at 15 000 × g for 1 h. The obtained supernatant was analysed by HPLC-UV and HR-MS as described above.

2.8. ¹H and ¹³C-NMR data of prepared glycosylated products

Baicalein-7-O-β-glucopyranoside (1a). Yellow powder, ¹H-NMR (DMSO-d₆, 500 MHz) δ 12.56 (1H, s, OH-5), 8.56 (1H, s, OH-6), 8.07 (2H, dd, J = 7.0, 1.0 Hz, H-2′, 6′), 7.61 (3H, m, H-3′, 4′, 5′), 7.05 (1H, s, H-8), 7.00 (1H, s, H-3), 5.40 (1H, brs, 2′′-OH), 5.11 (1H, d, J = 5.0 Hz, 3′′-OH), 5.08 (1H, d, J = 5.5 Hz, 4′′-OH), 5.02 (1H, d, J = 7.5 Hz, H-1′′), 4.64 (1H, d, J = 6.0 Hz, 6′′-OH), 3.76 (1H, m, H-6′′a), 3.50 (2H, m, H-6′′b, H-3′′), 3.37–3.19 (3H, m, H-2′′, 4′′, 5′′). ¹³C-NMR (DMSO-d₆, 125 MHz) δ 182.52 (C-4), 163.44 (C-2), 151.58 (C-7), 149.17 (C-9), 146.47 (C-5), 132.01 (C-4′), 130.81 (C-6), 130.58 (C-1′), 129.12 (C-3′, 5′), 126.34 (C-2′, 6′), 106.07 (C-10), 104.70 (C-3), 100.95 (C-1′′), 94.24 (C-8), 77.32 (C-5′′), 75.86 (C-3′′), 73.16 (C-2′′), 69.67 (C-4′′), 60.63 (C-6′′).

Quercetin-3-O-rhamnoside (7a). Yellow powder, ¹H-NMR (methanol-d₄, 500 MHz) δ 7.34 (1H, d, J = 2.0 Hz, H-2′), 7.31 (1H, dd, J = 8.5, 2.0 Hz, H-6′), 6.91 (1H, d, J = 8.5 Hz, H-5′), 6.37 (1H, d, J = 2.0 Hz, H-8), 6.20 (1H, d, J = 2.0 Hz, H-6), 5.35 (1H, brs, H-1′′), 4.21 (1H, brd, J = 1.5 Hz, H-3′′), 3.74 (1H, dd, J = 9.5, 3.0 Hz, H-2′′), 3.42 (1H, m, H-5′′), 3.34 (1H, m, H-4′′), 0.94 (3H, d, J = 6.0 Hz, H-6′′). ¹³C-NMR (methanol-d₄, 125 MHz) δ 179.80 (C-4), 166.04 (C-7), 163.45 (C-5), 159.46 (C-9), 158.68 (C-2), 149.96 (C-4′), 146.58 (C-3′), 136.38 (C-3), 123.10 (C-1′), 122.99 (C-6′), 117.05 (C-5′), 116.50 (C-2′), 106.03 (C-10), 103.69 (C-1′′), 99.9 (C-6), 94.84 (C-8), 73.39 (C-4′′), 72.25 (C-3′′), 72.19 (C-2′′), 72.05 (C-5′′), 17.81 (C-6′′).

Kaempferol-3-O-rhamnoside (10a). Yellow powder, ¹H-NMR (methanol-d₄, 500 MHz) δ 7.76 (2H, d, J = 8.5 Hz, H-2′, 6′), 6.93 (2H, d, J = 8.5 Hz, H-3′, 5′), 6.38 (1H, s, H-8), 6.20 (1H, s, H-6), 5.38 (1H, brs, H-1′′), 4.22 (1H, brs, H-2′′), 3.71 (1H, d, J = 5.5 Hz, H-3′′), 3.33 (2H, m, H-4′′, 5′′), 0.92 (3H, d, J = 4.5 Hz, H-6′′). ¹³C-NMR (methanol-d₄, 125 MHz) δ 179.61 (C-4), 165.86 (C-7), 163.22 (C-5), 161.58 (C-4′), 159.27 (C-9), 158.54 (C-2), 136.20 (C-3), 131.89 (C-2′, 6′), 122.61 (C-1′), 116.51 (C-3′, 5′), 105.92 (C-10), 103.49 (C-1′′), 99.81 (C-6), 94.73 (C-8), 73.17 (C-4′′), 72.10 (C-2′′), 72.03 (C-3′′), 71.92 (C-5′′), 17.65 (C-6′′).

Quercetin-3-O-glucopyranoside (18a). Yellow powder, ¹H-NMR (methanol-d₄, 500 MHz) δ 7.72 (1H, d, J = 2.0 Hz, H-2′), 7.59 (1H, dd, J = 8.5, 2.0 Hz, H-6′), 6.87 (1H, d, J = 8.5 Hz, H-5′), 6.39 (1H, d, J = 1.5 Hz, H-8), 6.20 (1H, d, J = 1.5 Hz, H-6), 5.26 (1H, d, J = 7.5 Hz, H-1′′), 3.72 (1H, brd, J = 12.0 Hz, H-6′′a), 3.58 (1H, dd, J = 12.0, 5.0 Hz, H-6′′b), 3.52–3.42 (2H, m, H-2′′, 3′′), 3.36 (2H, m, H-4′′, 5′′). ¹³C-NMR (methanol-d₄, 125 MHz) δ 179.62 (C-4), 166.11 (C-7), 163.17 (C-5), 159.15 (C-9), 158.58 (C-2), 149.97 (C-4′), 146.03 (C-3′), 135.76 (C-3), 123.33 (C-1′), 123.21 (C-6′), 117.70 (C-5′), 116.13 (C-2′), 105.83 (C-10), 104.46 (C-1′′), 100.01 (C-6), 94.84 (C-8), 78.52 (C-5′′), 78.26 (C-3′′), 75.87 (C-2′′), 71.36 (C-4′′), 62.70 (C-6′′).

3. Results and discussion

3.1. Gene cloning and heterogenous expression of AtRHM2

A. thaliana contains three bi-functional enzymes that catalyse the synthesis of UDP-L-rhamnose from UDP-D-glucose: AtRHM1 (At1g78570; GeneBank accession AY042833), AtRHM2 (At1g53500; GeneBank accession AY328518) and AtRHM3 (At3g14790; GeneBank accession AY078958).²³ As illustrated in Scheme 1,^23–25 the functional domain analysis revealed that the N-terminal region of RHM2 (RHM2-N; amino acids 1–370) has an activity like an UDP-D-glucose-4,6-dehydratase, while the C-terminal region of RHM2 (RHM2-C; amino acids 371–667) has the two following activities: UDP-4-keto-6-deoxy-D-glucose 3,5-epimerase, and UDP-4-keto-L-rhamnose 4-keto-reductase.²³ Here, we amplified the coding region of RHM2-N and RHM2-C separately based on the known sequence to obtain the 1113 bp and 897 bp open reading frame by using PCR from the cDNA of A. thaliana before being cloned into the expression vector pET-28a (Fig. S1†). Soluble proteins were obtained from transformants of E. coli after overnight induction with 1.0 mM IPTG at 25 °C. Purified polypeptide was obtained by Ni-NTA-agarose to generate a product with apparent homogeneity by SDS-PAGE. The observed molecular mass of AtRHM2-N and AtRHM2-C were 42.52 kDa (Fig. S2A†) and 35.32 kDa (Fig. S2B†) with a yield of 571 mg L⁻¹ culture and 908 mg L⁻¹ culture, respectively.

3.2. Chemo-enzymatic synthesis of UDP-Rha by the AtRHM2 protein

The UDP-Rha synthetic activity of AtRHM2 was assayed by a two-step reaction using the homogeneous protein of AtRHM2-N and AtRHM2-C as catalysts, UDP-Glc as a substrate, NAD⁺ and NADPH as co-factors. A control reaction was performed under the same conditions using the high temperature deactivated protein of AtRHM2.

In the previously reported HPLC analysis of nucleotide sugars, anion exchange columns and reversed phase columns were used. KH₂PO₄ or ion-pairing reagents such as tetra-, octa- and hexadecyltrimethyl-ammonium bromide in the presence of borate and tetrabutyl-ammonium salts were used in the mobile phases.^24,26–29 Thus, the post-treatment of the enzymatic products such as desalting or ion-exchange was needed since the existing ions were not suitable for direct mass spectrometry analysis. Here, the obtained UDP-Rha product was directly subjected to HPLC-UV and HR-ESI-MS analysis using ammonium formate (5 mM) as the mobile phase at a flow rate of 0.3 mL min⁻¹. Nucleotide sugars were detected at 260 nm. The retention time for the standard UDP-Glc was 9.774 min (Fig. 1A) and compared to the blank control (Fig. 1C). The reaction sample had a new peak at 10.570 min, which suggested an enzymatic reaction product (Fig. 1B). HR-ESI-MS analyses in the negative ion mode (de-protonated [M − H]⁻) and total MS ions of UDP-Glc standards gave the [M − H]⁻ ion at m/z 565.0455 (calcd 565.0477 for C₁₅H₂₄N₂O₁₇P₂). The CID-fragments formed from the parent ion gave a product ion at a m/z of 322.9817, corresponding to UMP [M − H]⁻ (Fig. 1E). Similarly, the total MS ions of the enzymatic product gave a [M − H]⁻ ion at m/z 549.0515 (the theoretical molecular weight of UDP-Rha was 549.0528 for C₁₅H₂₄N₂O₁₆P₂), and the CID-fragments formed from the parent ion gave a product ion at a m/z of 322.9814, corresponding to UMP [M − H]⁻ (Fig. 1D). The mass spectrometry behaviour of the new peak at 10.570 min was consistent with UDP-Rha, which confirmed that the enzymatic product was UDP-Rha with a yield of 39.08%.

Fig. 1 HPLC-UV and HR-ESI-MS analysis of chemo-enzymatic synthesis reaction of UDP-Rha by UDP-Glc. (A) HPLC chromatogram and UV spectra of authentic UDP-Glc; (B) HPLC chromatogram and UV spectra of chemo-enzymatic synthesis reaction of UDP-Rha by UDP-Glc; (C) HPLC chromatogram in negative control; (D), HR-ESI-MS (negative) spectrum and typical negative ion MS² spectrum of UDP-Rha product; (E) HR-ESI-MS (negative) spectrum and typical negative ion MS² spectrum of authentic UDP-Glc.

3.3. Gene cloning and heterogenous expression of AtUGT78D1

The coding region of AtUGT78D1 (At1g30530; GeneBank accession AY056312) was amplified using PCR from cDNA of A. thaliana. For gene expression, the 1362 bp coding sequence of AtUGT78D1 was cloned into the expression vector pET-28a. E. coli cells harboring the expression plasmid were induced by 1.0 mM IPTG at 25 °C. Protein induction was clearly observed and subsequently purified on nickel-NTA agarose to apparent homogeneity as judged by SDS-PAGE (Fig. S2C†). A yield of 668 mg purified His₆-tagged AtUGT78D1 per litre of culture was obtained. The observed molecular mass reasonably corresponded to the calculated mass of 51.74 kDa for His₆-AtUGT78D1.

3.4. Expanded enzymatic assays for substrate flexibility of AtUGT78D1

An acceptor library with structurally diverse “drug-like” aromatic scaffolds was established to investigate the substrate flexibility and synthetic utility of AtUGT78D1. Thirty different members (their structures are shown in Fig. 2B), including flavones (baicalein 1, diosmetin 2, chrysin 3), flavonones (pinocembrin 4, hesperetin 5, naringenin 6), flavonols (quercetin 7, morin hydrate 8, fisetin 9, kaempferol 10), isoflavones (formononetin 11, genistein 12), chalcones (isoliquiritigenin 13), flavonoid glycosides (vincetoxicoside B 14, isovitexin 15, hyperin 16, swertisin 17, quercitrin 18, scutellarin 19, sopho-ricoside 20), di-glycosides (rhoifolin 21), methylflavone (methylophiopogonone A 22), phenylethyl chromones (7-hydroxy-2-(2-phenylethyl)chromen-4-one 23, 8-chloranyl-6-hydroxy-2-(2-phenylethyl)chromen-4-one 24), benzophenones (4,4′-di-hydroxybenzophenone 25), 2,4-dihydroxybenzophenone 26), coumarins (7-hydroxycoumarin 27, 7-hydroxy-4-methylcoumarin 28), lignanoids (magnolol 29), and anthraquinones (emodin 30) were used for expanded enzymatic assays, and both UDP-Rha and UDP-Glc were tested as sugar donors.

Fig. 2 Exploring the aglycon promiscuity of AtUGT78D1 using UDP-Glc and UDP-Rha as sugar donor respectively. (A) Conversions of glycosylated products catalysed by AtUGT78D1. Members are listed with the number that corresponds to the structures listed in part (B). The dark color columns represent the percent conversion using UDP-Rha as sugar donor, while the light color represents the percent conversion using UDP-Glc as sugar donor. “*” represents the glycosylated products that were prepared and confirmed by HPLC-UV, HR-MS, and NMR spectroscopic data analyses (1a, 7a and 10a). “#” represents the glycosylated products that were structurally identified by HPLC-UV, HR-MS, and typical ion MS² data analyses (23a, 24a, 27a and 28a), or by comparison with the authentic sample (18a). “§” represents that the mono-rhamnoside and di-rhamnoside of 1 were produced with a yield of 50.74% and 37.01%, respectively. (B) The structures of tested aglycons and corresponding glycosylated products.

From HPLC-UV analyses, it was revealed that when using the optimum UDP-Rha as a sugar donor, AtUGT78D1 could catalyse the rhamnosylation of all 30 of the tested aromatic compounds belonging to the 7 structural types except for 19. All of the predicted rhamnosylated products were further identified by high resolution quadruple time-of-flight electrospray ionization mass spectrometry (HR-QTOF-ESI-MS). The molecular mass found for the novel peaks matched with their calculated mass exactly (Table 1). Besides, HR-MS analysis confirmed that both mono-rhamnoside and di-rhamnoside were produced when using baicalein (1) as a sugar acceptor with UDP-Rha as a sugar donor (mono-rhamnoside of baicalein: calcd for C₂₁H₂₀O₉ [M − H]⁻ 415.1035, found 415.1027, with a yield of 50.74%; di-rhamnoside of baicalein: calcd for C₂₇H₃₀O₁₃ [M + H]⁺ 563.1759, found 563.1754, with a yield of 37.01%, Table 1).

Table 1 Characterization of enzymatic rhamnosylated products by HPLC-ESI-MSⁿ analysis

Substrate (no)	Substrate RT (min)	Rha-product RT (min)	Rha-product yield%	Measured m/z	Predicted formula	Theoretical m/z	Error (ppm)
a N.P. represents as no product.
1	21.1	19.5	50.74	415.1027 [M − H]^−	C21H20O9	415.1035	−1.93
		20.6	37.01	563.1754 [M + H]^+	C27H30O13	563.1759	−0.89
2	20.9	19.8	24.28	469.1099 [M + Na]^+	C22H22O10	469.1105	−1.28
3	24.0	22.8	22.02	445.1143 [M + HCOO]^−	C21H20O8	455.1140	0.67
4	22.5	21.7	14.38	425.1205 [M + Na]^+	C21H22O8	425.1207	−0.47
5	18.4	17.8	11.94	447.1276 [M − H]^−	C22H24O10	477.1297	−4.70
6	17.8	17.3	6.17	441.1162 [M + Na]^+	C21H22O9	441.1156	1.36
7	17.8	15.0	96.38	447.0930 [M − H]^−	C21H20O11	447.0933	−0.67
8	17.1	15.1	95.63	471.0920 [M + Na]^+	C21H20O11	471.0898	4.67
9	15.6	12.8	79.25	455.0969 [M + Na]^+	C21H20O10	455.0949	4.39
10	20.2	17.0	40.01	431.0974 [M − H]^−	C21H20O10	431.0984	−2.32
11	21.3	19.6	45.32	437.1225 [M + Na]^+	C22H22O8	437.1207	4.12
12	18.9	16.9	25.65	461.1074 [M + HCOO]^−	C21H20O9	461.1089	−3.25
13	20.4	19.5	3.99	401.1224 [M − H]^−	C21H22O8	401.1242	−4.49
14	16.8	13.5	98.86	595.1662 [M + H]^+	C27H30O15	595.1657	0.84
15	12.3	10.9	4.45	577.1532 [M − H]^−	C27H30O14	577.1563	−5.37
16	13.4	11.0	8.00	609.1348 [M − H]^−	C27H30O16	609.1461	−3.78
17	12.4	10.6	6.27	591.1696 [M − H]^−	C28H32O14	591.1719	−3.89
18	15.1	12.8	5.83	593.1510 [M − H]^−	C27H30O15	593.1512	−0.34
19	13.4	N.P.^a
20	13.8	11.4	13.09	579.1689 [M + H]^+	C27H30O14	579.1708	−3.28
21	14.2	13.2	9.85	725.2313 [M + H]^+	C33H40O18	725.2287	3.59
22	27.4	26.2	9.61	509.1439 [M + Na]^+	C25H26O10	509.1418	4.12
23	22.9	21.6	12.08	435.1459 [M + Na]^+	C23H24O7	435.1414	10.34
24	26.6	24.9	4.82	491.1100 [M + HCOO]^−	C23H23O7Cl	491.1114	−2.85
25	21.4	20.8	20.59	383.1110 [M + Na]^+	C19H20O7	383.1101	2.35
26	14.2	13.4	14.68	361.1276 [M + H]^+	C19H20O7	361.1282	−1.66
27	10.8	10.0	22.58	309.0966 [M + H]^+	C15H16O7	309.0969	−0.97
28	13.4	12.4	22.47	323.1138 [M + H]^+	C16H18O7	323.1125	4.02
29	27.2	25.8	25.23	457.1845 [M + HCOO]^−	C24H28O6	457.1868	−5.03
30	29.1	27.3	13.59	439.1006 [M + Na]^+	C21H20O9	439.1000	1.37

---

Moreover, AtUGT78D1 exhibited a high substrate conversion rate (>40%) with 7 out of 30 substrates, including the reported optimum substrate flavonols (7–10), as well as flavones 1, isoflavones 11, and flavonoid glycosides 14.

Considering the flexibility of AtUGT78D1 in recognizing sugar acceptors, the donor specificity of AtUGT78D1 was also examined with UDP-Glc, UDP-Gal, UDP-GlcA and UDP-N-acetylglucosamine (UDP-GlcNAc). When using UDP-Glc as a glycosyl donor, the HPLC-UV chromatogram generated a single product peak from all the flavonol (7–10) reaction mixtures, as well as flavones 1, flavonones 4, isoflavones 12, and flavonoid glycosides 14, 18, 19. These reaction mixtures were then subjected to high-resolution mass analyses using HR-QTOF-ESI-MS. We found the exact mass spectrum matched with the calculated mass of each flavonoid glucoside (Table 2). Among the ten positive substrates, the glucosylated product yield of compounds 1, 8, 9, 14 were highest at 64.96%, 100%, 87.34%, and 96.84% respectively. Meanwhile, no obvious catalysis activity was observed with the three sugar donors of UDP-Gal, UDP-GlcA and UDP-GlcNAc.

Table 2 Characterization of enzymatic glucosylated products by HPLC-ESI-MSⁿ analysis

Substrate (no)	Substrate RT (min)	Glc-product RT (min)	Glc-product yield%	Measured m/z	Predicted formula	Theoretical m/z	Error (ppm)
a N.P. represents as no product.
1	21.1	19.6	64.96	431.0973 [M − H]^−	C21H20O10	431.0984	−2.55
2	20.9	N.P.^a
3	24.0	N.P.
4	22.5	16.9	1.92	463.1233 [M + HCOO]^−	C21H22O9	463.1246	−2.81
5	18.4	N.P.
6	17.8	N.P.
7	17.8	13.5	36.28	465.1038 [M + H]^+	C21H20O12	465.1028	2.15
8	17.1	13.4	100.00	463.0857 [M − H]^−	C21H20O12	463.0882	−5.40
9	15.6	11.7	87.34	471.0917 [M + Na]^+	C21H20O11	471.0898	4.03
10	20.2	15.2	22.24	447.0920 [M − H]^−	C21H20O11	447.0933	−2.91
11	21.3	N.P.
12	18.9	11.6	5.03	477.1016 [M + HCOO]^−	C21H20O10	477.1038	−4.61
13	20.4	N.P.
14	16.8	11.0	96.84	609.1434 [M − H]^−	C27H30O16	609.1461	−4.43
15	12.3	N.P.
16	13.4	N.P.
17	12.4	N.P.
18	15.1	13.5	8.09	465.1043 [M + H]^+	C21H20O12	465.1028	3.23
19	13.4	10.8	7.11	623.1218 [M − H]^−	C27H28O17	623.1254	−5.78
20–30	—	N.P.

---

Our extensive enzymatic reactions effectively expanded the substrate spectrum of AtUGT78D1 and clarified its catalysis behavior towards both sugar donors and acceptors. In addition to flavonols, diverse flavones, flavonones, isoflavones, chalcones, flavonoid glycosides, di-glycosides, methylflavone, phenylethyl chromones, benzophenones, coumarins, lignanoids and anthraquinones with active hydroxyl groups in their structures were also accepted by AtUGT78D1 to yield the corresponding rhamnosylated products. According to HPLC-UV and HR-QTOF-ESI-MSⁿ analyses, all of the generated aromatic rhamnosides were presumed to be O-rhamnosides. For instance, as can be seen in Fig. S9,† the parent-molecule ion [M + H]⁺ of the rhamnosylated product of 7-hydroxycoumarin (27) at m/z 309.0966 yielded MS² fragment ions at m/z 163.0401, corresponding to the loss of a rhamnosyl moiety (146 amu). Since only one active hydroxyl group existed in 27, the rhamnosylation substitution was inferred to occur at the 7-OH group in product 27a. Similar situations were also observed in substrates 23 (Fig. S7†), 24 (Fig. S8†) and 28 (Fig. S10†), and their rhamnosylated products were presumed to be 7-O-rhamnosyloxy-2-(2-phenylethyl)chromen-4-one (23a), 8-chloranyl-6-O-rhamnosyloxy-2-(2-phenylethyl)chromen-4-one (24a) and 7-O-rhamnosyloxy-4-methylcoumarin (28a), respectively.

Besides, our extensive enzymatic reactions also revealed that the UDP-Rha and UDP-Glc were also accepted by AtUGT78D1 to catalyse the glycosylation of 10 flavonoids while the rhamnosylation efficiency was much higher than the glucosylation efficiency, which indicated that UDP-Rha was the optimum sugar donor. Similarly, the rhamnosylation and/or glucosylation efficiency of flavonols was much higher than other substrate structure types, which indicated that flavonols were the optimum sugar acceptor of AtUGT78D1.

Of particular note is that the enzyme exhibited the capability to glycosylate natural “drug-like” scaffolds including methylflavone (22), phenylethyl chromones (23–24), coumarins (27–28), magnolol (29) and emodin (30). Besides, the rhamnosylation of diverse flavonoid glycosides, including both mono- and di-glycosides, resulted in the formation of polysaccharides with one or more glycosyl substitutions at different positions, which usually lead to the improved aqueous solubilities, stabilities, and enhanced bioavailabilities. To the best of our knowledge, the rhamnosylations of these “drug-like” scaffolds by plant GTs were rarely reported before, and the unusual substrate promiscuity renders AtUGT78D1 a promising enzyme for the creation of structurally diverse bioactive glycosides especially for various rhamnosides.

3.5. In vitro preparation of enzymatic products by AtUGT78D1

As mentioned above, our extensive enzymatic reactions expanded the recognition of AtUGT78D1 as a flavonol-3-O-rhamnosyltransferase. To further reveal the catalytic characteristic of AtUGT78D1 in vitro, a flavone baicalein (1) without a hydroxyl group at the C-3 position as well as two flavonols (quercetin 7 and kaempferol 10) with a hydroxyl group at the C-3 position were picked out as sugar acceptors and subjected to the preparative scale enzymatic reactions using UDP-Glc and UDP-Rha as sugar donors. When using baicalein (1) as a sugar acceptor and UDP-Glc as a sugar donor, one glucosylated product (1a) was obtained and was structurally analysed by ¹H and ¹³C-NMR followed by HMBC (heteronuclear multiple bond correlation). When comparing the ¹H-NMR of standard baicalein (1) and the reaction product (1a), signals at δ 12.56 (1H, s, 5-OH) and δ 8.56 (1H, s, 6-OH) were detected but the 7-OH signal was missing in the reaction product (Fig. S45†). This indicated that the sugar moiety could have replaced the proton of this hydroxyl group. The anomeric proton (H-1′′) was consistent with δ 5.02 (d, J = 7.5 Hz, 1H), and the anomeric proton coupling constant (J = 7.5 Hz) confirms the conjugation of the glucose moiety to be in the β-configuration. To determine the position of the sugar conjugation, the HMBC data of 1a was further analysed. Correlation of the anomeric proton H-1′′ (δ 5.02) with C-7 (δ 151.58) in HMBC confirmed the attachment of the glucose moiety at the 7-OH position of baicalein (1, Fig. 3). Therefore, the final product was determined to be baicalein-7-O-β-glucopyranoside.

Fig. 3 HMBC of the glucosylated product of baicalein (1), baicalein-7-O-β-glucopyranoside (1a), showing the correlation with the anomeric proton and C-7 of baicalein.

The rhamnosylation of two flavonols (quercetin 7 and kaempferol 10) resulted in two rhamnosylated products, 7a and 10a. Their structures were identified as quercetin 3-O-rhamnoside (7a) and kaempferol 3-O-rhamnoside (10a) using HR-MS, ¹H and ¹³C-NMR spectroscopic data analyses (Fig. S48–S51,† NMR data shown in Section 2.8), which confirmed that the rhamnosylation specifically occurred at the 3-OH position when using flavonols as substrates.

It is important to note that most glycosylated products exhibit improved biological activities or water solubilities compared to the parent molecules. For instance, in vivo quercetin 3-O-rhamnoside (7a) can be a more important antioxidant and neuro-protective agent than quercetin (7) because of its high bioavailability in the digestive track.³⁰ Kaempferol 3-O-rhamnoside (10a) showed higher inhibitory activities against aldose reductase (AR) than flavonoid aglycone kaempferol (10) and could be a useful natural source in the development of a novel AR inhibitory agent against diabetic complications.³¹ The solubility of baicalein 7-O-β-glucopyranoside (1a) in water (50 mg L⁻¹) was 10 times greater than baicalein (1, 5.4 mg L⁻¹), which implies that the attachment of a glucosyl residue to baicalein effectively enhanced the water solubility of the original compound.³²

Our results expanded the catalytic characteristics of AtUGT78D1 in vitro. It can catalyse not only the rhamnosylation at the 3-OH position of flavonols, but also glucosylation at the 7-OH position of baicalein, as well as multiple hydroxyl substitutions on diverse types of aromatics (discussed in Section 3.4) to afford various glycosylated products that always exhibit improved pharmacological activities compared to their precursors. These findings provide strong hints that AtUGT78D1 could be used as a powerful tool for the glycosylation of diverse aromatics for structural modifications and drug discovery.

3.6. One-pot reaction catalysed by AtUGT78D1

In our aforementioned extensive enzymatic assays for substrate flexibility of AtUGT78D1, when quercitrin (18, quercetin 3-O-rhamnoside, C₂₁H₂₀O₁₁) was tested as sugar acceptor with UDP-Glc as a sugar donor, a glycosylated product, 18a, was formed compared to the negative control at t_R13.5 min in a yield of 8.09%. However, the HR-MS analysis revealed that the molecular formula of 18a was predicted to be C₂₁H₂₀O₁₂ (calcd for C₂₁H₂₀O₁₂ [M + H]⁺ 465.1028, found 465.1043), which possessed an extra oxygen atom compared to substrate 18. Meanwhile, as can be seen in Fig. S6,† the parent-molecule ion [M + H]⁻ of 18a at m/z 465.1028 yielded MS² fragment ions at m/z 303.0458, corresponding to the loss of the glucosyl moiety (C₆H₁₀O₅, 162 amu). These results indicated that the glucose moiety was transferred into quercetin instead of quercitrin. Based on further chromatographic and spectral analyses, as well as comparison and co-HPLC with the authentic samples (Fig. S43†), product 18a was identified to be quercetin 3-O-glucopyranoside, (Fig. S52 and S53,† NMR data shown in Section 2.8), which confirmed that de-rhamnosylation of quercitrin (18) occurred in the reaction to give the aglycon quercetin (7) before subsequent glucosylation at the 3-OH position.

Generally, GTs are perceived as unidirectional catalysts that drive the formation of glycosidic bonds from NDP-sugar donors and aglycon acceptors.³³ However, more and more reported GTs have been observed to catalyse reversible, even bidirectional reactions.^34–37 The observed de-rhamnosylation of quercitrin 18 reminded us that AtUGT78D1 may be employed for de-glycosylation in certain cases to transfer a sugar moiety from rhamnoside scaffolds to aglycons. To prove this speculation, the reverse assays of AtUGT78D1 catalysed reactions were conducted in the presence of UDP and quercitrin (18). As expected, the de-rhamnosylated product quercetin (7) was detected by HPLC-UV analysis (Fig. S44A†), confirming the reaction reversibility of AtUGT78D1.

Removal of the rhamnose unit from the 3-O position of quercitrin (18) affords AtUGT78D1 the availability for a molecule “one-pot” reaction. The simple aromatic substrate kaempferol (10) was selected as an example and was subjected to coupled reactions with UDP and quercitrin (18) to further exploit the application of AtUGT78D1 in generating bioactive rhamnosides through one-pot reactions. Interestingly, in the coupled reaction (Fig. 4), UDP-Rha was generated by AtUGT78D1 catalysed de-rhamnosylation of 18 in the presence of UDP, and the rhamnose moiety was intermediately transferred to kaempferol (10) through subsequent rhamnosylation catalysed by AtUGT78D1 to generate the product kaempferol-3-O-rhamnoside (10a) with a high yield of 50.17% (Fig. S44†). This one-pot reaction, providing the desired products by reusing the byproduct without adding the expensive UDP-Rha, was economic and environmental-friendly. As far as we know, it is the first report of a rhamnosyltransferase with bidirectional catalysis activities to be used in one-pot rhamnosylation, and the results indicated that AtUGT78D1 could serve as a cost-effective and applicable tool to generate bioactive rhamnosides.

Fig. 4 Catalytic reverse reaction and coupled one-pot reactions catalysed by AtUGT78D1. The bioactive glycosides were generated from the simple sugar donor quercitrin (18) with the presence of UDP and aglycon kaempferol (10). The HPLC-UV chromatograms (Fig. S44†) are shown in the ESI.†

4. Conclusions

In summary, an efficient UDP-rhamnose two-step enzymatic reaction production system was well established. Extensive in vitro enzymatic assays and preparative reactions using the obtained UDP-Rha/UDP-Glc highlighted the robust glycosylation promiscuity of AtUGT78D1 toward diverse “drug-like” compounds, as well as expanded the catalysis characteristics of AtUGT78D1. Combined with the obvious reaction reversibility for one-pot production, AtUGT78D1 was expected to be a universal and effective tool for chemo-enzymatic synthesis of diverse desired bioactive rhamnosylated derivatives. Our studies also provided useful hints for further enzyme engineering to develop powerful biocatalysts in the combinatorial biosynthesis of leading compounds for drug R&D.

Acknowledgements

We are thankful to the financial support from the National Natural Science Foundation of China (Grant No. 81402809) and graduate independent project at Beijing University of Chinese Medicine (No. 2016-JYB-XS077).

References

1. A. Luzhetskyy, C. Méndez, J. A. Salas and A. Bechthold, Curr. Top. Med. Chem., 2008, 8, 680.
2. X. Q. Wang, FEBS Lett., 2009, 583, 3303.
3. P. C. H. Hollman, J. H. M. de Vries, S. D. van Leeuwen, M. J. B. Mengelers and M. B. Katan, Am. J. Clin. Nutr., 1995, 62, 1276.
4. P. C. H. Hollman, M. N. C. P. Bijsman, Y. van Gameren, E. P. J. Cnossen, J. H. M. de Vries and M. B. Katan, Free Radical Res., 1999, 31, 569.
5. K. Yonekura-Sakakibara, T. Tohge, R. Niida and K. Saito, J. Biol. Chem., 2007, 282, 14932.
6. E. K. Lim, D. A. Ashford, B. K. Hou, R. G. Jackson and D. J. Bowles, Biotechnol. Bioeng., 2004, 87, 623.
7. T. Ohashi, Y. Hasegawa, R. Misaki and K. Fujiyama, Appl. Microbiol. Biotechnol., 2016, 100, 687.
8. S. M. Paquette, K. Jensen and S. Bak, Phytochemistry, 2009, 70, 1940.
9. K. Yonekura-Sakakibara and K. Hanada, Plant J., 2011, 66, 182.
10. H. J. Choi, J. H. Song, K. S. Park and D. H. Kwon, Eur. J. Pharm. Sci., 2009, 37, 329.
11. S. J. N. Tatsimo, J. D. Tamokou, L. Havyarimana, D. Csupor, P. Forgo, J. J. R. Kuiate and P. Tane, BMC Res. Notes, 2012, 5, 158.
12. T. Coenye, G. Brackman, P. Rigole, E. De Witte, K. Honraet, B. Rossel and H. J. Nelis, Phytomedicine, 2012, 19, 409.
13. E. K. Lim, D. A. Ashford and D. J. Bowles, ChemBioChem, 2006, 7, 1181.
14. W. Offen, C. Martinez-Fleites, M. Yang, E. Kiat-Lim, B. G. Davis, C. A. Tarling, C. M. Ford, D. J. Bowles and G. J. Davies, EMBO J., 2006, 25, 1396.
15. P. Jones, B. Messner, J. Nakajima, A. R. Schäffner and K. Saito, J. Biol. Chem., 2003, 278, 43910.
16. P. Parajuli, R. P. Pandey, N. T. H. Trang, T. J. Oh and J. K. Sohng, Carbohydr. Res., 2015, 418, 13.
17. H. J. Kim, B. G. Kim and J. H. Ahn, Appl. Microbiol. Biotechnol., 2013, 97, 5275.
18. R. W. Gantt, R. D. Goff, G. J. Williams and J. S. Thorson, Angew. Chem., Int. Ed. Engl., 2008, 47, 8889.
19. G. X. Ren, J. L. Hou, Q. H. Fang, H. Sun, X. Y. Liu, L. W. Zhang and P. G. Wang, Glycoconjugate J., 2012, 29, 425.
20. B. G. Kim, H. J. Kim and J. H. Ahn, J. Agric. Food Chem., 2012, 60, 11143.
21. S. M. Yang, S. H. Han, B. G Kim and J. H. Ahn, J. Ind. Microbiol. Biotechnol., 2014, 41, 1311.
22. R. Yin, B. Messner, T. Faus-Kessler, T. Hoffmann, W. Schwab, M. R. Hajirezaei, V. S. Paul, W. Heller and A. R. Schäffner, J. Exp. Bot., 2012, 63, 2465.
23. T. Oka, T. Nemoto and Y. Jigami, J. Biol. Chem., 2007, 282, 5389.
24. G. Watt, C. Leoff, A. D. Harper and M. Bar-Peled, Plant Physiol., 2004, 134, 1337.
25. V. Martinez, M. Ingwers, J. Smith, J. Glushka, T. Yang and M. Bar-Peled, J. Biol. Chem., 2012, 287, 879.
26. J. Räbinä, M. Mäki, E. M. Savilahti, N. Järvinen, L. Penttilä and R. Renkonen, Glycoconjugate J., 2001, 18, 799.
27. C. Rautengarten, B. Ebert, T. Herter, C. J. Petzold, T. Ishii, A. Mukhopadhyay, B. Usadel and H. V. Scheller, Plant Cell, 2011, 23, 1373.
28. T. Oka and Y. Jigami, FEBS J., 2006, 273, 2645.
29. N. Tomiya, E. Ailor, S. M. Lawrence, M. J. Betenbaugh and Y. C. Lee, Anal. Biochem., 2001, 293, 129.
30. C. Wagnera, R. Fachinettoa, C. L. Dalla Cortea, V. B. Britoa, D. Severoa, G. de Oliveira Costa Diasb, A. F. Morelb, C. W. Nogueiraa and J. B. T. Rochaa, Brain Res., 2006, 1107, 192.
31. S. Y. Mok and S. Lee, Food Chem., 2013, 136, 969.
32. K. H. Kim, Y. D. Park, H. Park, K. O. Moon, K. T. Ha, N. I. Baek, C. S. Park, M. Joo and J. Cha, Eur. J. Pharmacol., 2014, 744, 147.
33. K. M. Koeller and C. H. Wong, Chem. Rev., 2000, 100, 4465.
34. A. Minami, K. Kakinuma and T. Eguchi, Tetrahedron Lett., 2005, 46, 6187.
35. C. Zhang, B. R. Griffith, Q. Fu, C. Albermann, X. Fu, I. K. Lee, L. Li and J. S. Thorson, Science, 2006, 313, 1291.
36. L. L. Sun, D. W. Chen, R. D. Chen, K. B. Xie, L. Yang and J. G. Dai, Tetrahedron Lett., 2016, 57, 1518.
37. K. B. Xie, R. D. Chen, J. H. Li, R. S. Wang, D. W. Chen, X. X. Dou and J. G. Dai, Org. Lett., 2014, 16, 4874.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra16251g

‡ Ting Mo and Xiao Liu contributed equally to the work.

---