Mutational Analysis of the Medicago Glycosyltransferase UGT71G1 Reveals Residues That Control Regioselectivity for (Iso)flavonoid Glycosylation*

The plant glycosyltransferase UGT71G1 from the model legume barrel medic (Medicago truncatula) glycosylates flavonoids, isoflavonoids, and triterpenes. It can transfer glucose to each of the five hydroxyl groups of the flavonol quercetin, with the 3′-O-glucoside as the major product, and to the A-ring 7-hydroxyl of the isoflavone genistein. The sugar donor and acceptor binding pockets are located in the N and C termini, respectively, of the recently determined crystal structure of UGT71G1. The residues forming the binding pockets of UGT71G1 were systematically altered by site-directed mutagenesis. Mutation of Phe148 to Val, or Tyr202 to Ala, drastically changed the regioselectivity for quercetin glycosylation from predominantly the 3′-O-position of the B-ring to the 3-O-position of the C ring. The Y202A mutant exhibited comparable catalytic efficiency with quercetin to the wild-type enzyme, whereas efficiency was reduced 3-4-fold in the F148V mutant. The Y202A mutant gained the ability to glycosylate the 5-hydroxyl of genistein. Additional mutations affected the relative specificities for the sugar donors UDP-galactose and UDP-glucuronic acid, although UDP-glucose was always preferred. The results are discussed in relation to the design of novel biocatalysts for production of therapeutic flavonoids.

Glycosylation is among the most important chemical reactions in plants. It involves transfer of a nucleotide diphosphateactivated sugar moiety to an acceptor molecule and is catalyzed by a diverse group of uridine diphosphate glycosyltransferases (UGTs). 2 All major classes of plant secondary metabolites, including flavonoids (1, 2), terpenoids (3)(4)(5), and alkaloids (6) can be modified by glycosylation, and more than 300 different flavonoid glycosides have been identified from plants (7).
The flavonol quercetin is a common constituent of plants. It has five hydroxyl groups, at positions of 3, 5, 7, 3Ј, and 4Ј (Fig.  1A). As one of the major antioxidants derived from plant sources, quercetin is beneficial to human health and has been ascribed anticancer (20 -22), anti-inflammatory (23,24), and anti-allergic activity (25,26). In nature quercetin is often glycosylated on one or more of the five hydroxyl groups to increase its solubility, stability, and bioavailability. The position of conjugation has a significant impact on its biological activity and potential health benefits for humans (20,21).
A single amino acid mutation in a UDP-galactose:anthocyanin galactosyltransferase from Aralia cordata was reported to alter sugar donor specificity (18). However, the relationship between the primary sequence of plant UGTs and acceptor substrate regioselectivity is not clear. For example, among the 11 Arabidopsis UGTs that catalyze glycosylation on the 3-hydroxyl of quercetin, identity at the amino acid level is from 20 to 72%, and the betalain (betanidin) 6-O-and 5-O-glucosyltransferases from Dorotheanthus bellidiformis share only 19% amino acid sequence identity (25).
UGT71G1 is a relatively promiscuous UGT from the model legume Medicago truncatula. Its highest in vitro activity appears to be with quercetin, but it also efficiently catalyzes glycosylation of the isoflavone phytoestrogen genistein (5). Incubation of UGT71G1 with quercetin and UDP-glucose results in the formation of all five possible monoglucosides, with the 3Ј-O-glucoside predominating, whereas genistein is only glucosylated on the A-ring 7-hydroxyl (5,30). The x-ray crystal structure of UGT71G1 was recently determined, and the donor and acceptor pockets defined (Fig. 1, B-D) (30). We here employ systematic site-directed alteration of amino acid residues around these pockets to dissect the structural basis for substrate regioselectivity in this versatile biocatalyst. Alter-ations to the length of the acceptor binding pocket, by replacing residues Phe 148 or Tyr 202 with smaller amino acids, appear to allow the substrate to slide further into the pocket and alter the predominant position of glycosylation of quercetin from 3Ј (B-ring) to 3 (C-ring).
Site-directed Mutagenesis-Sitedirected mutagenesis was performed using the QuikChange II XL site-directed mutagenesis kit (Stratagene). The UGT71G1 cDNA (30) cloned into pET28a vector with a hexa-histidine tag (Novagen, Milwaukee, WI) was used as template. Synthetic oligonucleotides used for mutagenesis are listed in Table 1. Mutations were confirmed by sequencing.
Expression of Recombinant Mutant Enzymes-Escherichia coli BL21(DE3) cells harboring the wildtype and mutated constructs were cultured in LB medium at 37°C until A 600 reached 0.6 -0.8. Isopropyl 1-thio-␤-galactopyranoside was added to a final concentration of 0.5 mM and the cultures incubated overnight at 16°C. UGT protein was purified using the MagneHis protein purification system according to the manufacturer's instructions (Promega). pET28a-transformed E. coli BL21(DE3) cells were treated in parallel as a control. Protein concentration was determined with the Bio-Rad protein dye-binding assay (Bio-Rad) using bovine serum albumin as standard.
Assay of Enzyme Activity-Assays for determination of substrate specificity and regioselectivity were carried out in 50 mM Tris-HCl buffer (pH 7.0) containing 4 g of purified recombinant protein, 500 M UDPG, UDP-galactose, or UDP-glucuronic acid, 250 M of either quercetin or genistein, and 14 mM mercaptoethanol in a final volume of 200 l for 1 h at 30°C. Reactions were stopped with 5 l of trichloroacetic acid (240 mg/ml) and extracted with 250 l of ethyl acetate. The extracted residues were resuspended in methanol and analyzed by reverse-phase HPLC (Hewlett Packard 1100 system) on a

Effects of Mutation of Acceptor Binding Pocket Residues on Enzyme Activity and Regioselectivity for Quercetin-In in vitro
assays with the flavonol quercetin as acceptor, UGT71G1 catalyzes formation of all five potential monoglucosides (30), with the 3Ј-O-glucoside accounting for ϳ70% of the total activity (Table 2, Fig. 2A).
To investigate the structural basis for the substrate specificity and product regioselectivity of UGT71G1, each of the amino acid side chains of the residues around the acceptor binding pocket were independently altered by site-directed mutagene-

TABLE 1 Primers used for introducing amino acid changes into UGT71G1
The modified bases are underlined.

Mutation
Site a Primer sequence D GACATGGCCTATTTATGCAGAACCACAGCTTAATGCTTTTAGGTTGG a Mutation in sugar donor site (D) or phenolic acceptor site (A).

TABLE 2 Relative activities of wild-type and mutant UGT71G1 enzymes for glycosylation of the hydroxyl groups of quercetin with UDP-glucose as sugar donor
Results are expressed as HPLC peak area (average of three determinations) for each product as a % of total product peak area (% T), or as a % of the peak area of the same product from an identical incubation with an equal amount of wild-type enzyme (% W).    (Fig. 1B, Table 1). Soluble enzyme expressed from wild-type and mutant enzyme constructs was obtained in yields from 4 to 24 g/10 ml of induced culture, and SDS-PAGE analysis confirmed that a single band of the correct molecular weight was seen in the purified preparations of each mutant, suggesting that the single amino acid changes did not affect protein stability.

Mutagenesis of UGT71G1
Reducing the size of large aromatic side chains at one end of the pocket dramatically altered regioselectivity with quercetin as acceptor, from predominantly B-ring 3Ј-O-glucosylation to predominantly C-ring 3-O-glucosylation (Table 2, Fig. 2A). Y202A retained 96% of wild-type activity with the 3-O-glucoside accounting for 95% of the conversion and production of the 3Ј-O-glucoside reduced to just 1% of the wild-type value ( Table  2, Fig. 2A). Although the F148V mutant lost 80% of the wildtype activity, it produced a single product identified as quercetin 3-O-glucoside ( Fig. 2A).
Kinetic analysis indicated that Y202A exhibited comparable catalytic efficiency (K cat /K m ) to the wild-type enzyme, but that F148V exhibited 3.6-fold reduced catalytic efficiency ( Table 3). The K m values for both mutant enzymes were 2-3-fold lower than for the wild-type enzyme, with the loss in efficiency of F148V resulting from a large decrease in K cat .
The overall activities of the P18A, I20V, F49V, G51R, M52L, P53A, and P88A mutants were from 40 to 90% of wild-type activity. The activity of F54V was reduced by about 80% ( Table  2). All these mutants preferentially glucosylated the 3Ј-hydroxyl of quercetin, with this reaction accounting for 48 -76% of total products formed ( Table 2). They also retained their ability to glucosylate the A-ring 5-hydroxyl of quercetin. Pro 18 and Phe 49 are located in the middle of the acceptor binding pocket, close to each other in the three-dimensional structure (Fig. 1). The P18A and F49V mutations caused a 2-3-fold increase in 7-O-glucosylation compared with the wild-type enzyme ( Table 2). E89L had no detectable activity.
Effects of Mutation of Acceptor Binding Pocket Residues on Enzyme Activity and Regioselectivity for Genistein-Wild-type UGT71G1 only glucosylates the A-ring 7-hydroxyl of the isoflavone genistein (5, 30) (Fig. 2B), consistent with the orientation of genistein docked into the acceptor binding pocket (Fig. 1B). However, Y202A converted genistein to two products, the 7-Oglucoside and a compound tentatively identified as the 5-Oglucoside (comparison with an authentic standard ruled out production of the other possible product, the 4Ј-O-glucoside), with a total activity of 23% of that of the wild-type enzyme ( Table 4, Fig. 2B). In contrast to its retention of activity with quercetin, F148V had no detectable activity with genistein ( Table 4).

Effects of Mutation of Sugar Donor Binding Pocket Residues on Enzyme Activity and Sugar Donor Specificity-UGT71G1
displayed 1-4% activity with quercetin when UDPG was replaced by UDP-galactose or UDP-glucuronic acid. The 3Ј-Oglycoside was the predominant product (Fig. 2, C and D).
To investigate the roles of residues around the sugar binding pocket in sugar donor specificity and enzyme activity with quercetin as acceptor, site-directed mutagenesis was used to alter the side chains of those residues with close contacts to the sugar moiety (30) (Fig. 1D, Table 1). M286L retained full activity with UDP-glucose as sugar donor (Table 2). However, it exhibited 2-3-fold higher activity than the wild-type enzyme with UDP-galactose or UDP-glucuronic acid (Table 5). Y379A retained 65% of wild-type activity with UDP-galactose, but no activity was detected with UDP-glucuronic acid ( Table 5). The S285A, W339G, Q342A, H357D, W360G, N361A, S362A, E365P, and Q382P mutants had no detectable activity with quercetin and UDP-glucose, UDP-galactose, or UDP-glucuronic acid (data not shown).

DISCUSSION
Based on the x-ray crystal structure of UGT71G1, a series of mutant enzymes with single alterations in amino acids surrounding the sugar donor and phenolic acceptor pockets were   generated. Through enzyme activity assay with quercetin, residues Phe 148 and Tyr 202 were identified as crucial in determining the preferred 3Ј-O-regioselectivity of the wild-type enzyme. The Y202A mutant exhibited 13-14-fold higher activity toward the 3-hydroxyl group than did the wild-type enzyme, with a reduction in the dominant glucosylation of the 3Ј-hydroxyl to only 1% of the wild-type activity while maintaining 96% total activity and efficient kinetic constants. Only the 3-O-glucoside was detected in incubations with F148V; the mutant enzyme had a decreased K m value but strongly reduced K cat , such that its total activity was only about 20% of the wild-type. However, the expression level of this enzyme in E. coli was identical to that of the Y2023A mutant (23 g/10 ml of culture), suggesting that its loss of activity does not reflect reduced stability. Detection of a single product with F148V is more likely due to lack of sensitivity of the HPLC assay than to a change of regioselectivity, because reducing the size of the amino acid at positions 148 or 202 would be predicted to have similar effects. Residues Phe 148 and Tyr 202 are close to each other in the threedimensional structure of UGT71G1 (Fig. 1B), located at one end of the acceptor binding pocket (30). Reducing the size at either of these positions would increase the volume of the binding pocket around where the C-ring of quercetin is located. This binding pocket volume change would allow the 3-hydroxyl to move closer to the C1 reaction center on the UDP-glucose and be more favorably placed for glycosylation. Active site cleft volume was previously reported to determine the substrate specificity of human UDP-galactose 4Ј-epimerase, which interconverts UDP-galactose and UDP-glucose, and UDP-N-acetylgalactosamine and UDP-N-acetylglucosamine (31). Cys 307 in the active site of the epimerase acts as a gatekeeper for substrate access to the site. The C307Y mutant completely lost activity for UDP-N-acetylgalactosamine, the larger substrate, but retained normal activity for UDP-galactose (31).
Met 286 is located between the N-and C-terminal domains of UGT71G1, close to the phosphate group and the C1 reaction center of UDP-glucose. Mutation at this position may therefore affect donor specificity, as seen by the small (2-3-fold) increase in activity with UDP-galactose or UDP-glucuronic acid, although the similar size of Met and Leu results in no difference in activity with the preferred donor UDPG.
The W360G mutation in the donor binding pocket had no detectable activity with quercetin and any of three UPD-sugars tested. Trp 360 forms a hydrogen bond with the sugar donor through its main chain nitrogen atom (37). The W360G mutant would be expected to maintain this same interaction, unless this mutation altered the overall conformation of the donor binding pocket due to the large difference in size between Gly and Trp.
The Y202A mutant retained activity with the isoflavone genistein, but genistein 5-O-glucoside was formed in addition to the 7-O-glucoside, the only product formed by the wild-type enzyme. Structural analysis and molecular docking indicate that the 7-OH of genistein may be docked into the active site in a position to accept the glucose moiety from UDPG, but the proximity of Glu 89 and Ala 380 to the A-ring of genistein provide steric constraints for correct alignment of the 5-hydroxyl for glycosylation (Fig. 1B). Mutation of Tyr 202 to alanine creates more available space in the acceptor pocket and may allow the 5-OH to fit into the active site without these potential spatial conflicts.
With quercetin as acceptor, F148V only produced the 3-Oglucoside, and lost activity for other positions (including the 7-OH). The 7-OH of genistein can still be docked into the active site of the F148V mutant enzyme, and the reason for the loss of activity of this mutant is therefore not clear.
The E89K mutant lost activity with both genistein and quercetin. Glu 89 is located inside the acceptor binding pocket, close to the 5-OH or 7-OH groups of genistein docked into the active site. Although modeling shows that the E89K mutant might retain a similar overall acceptor binding pocket conformation, the pocket may become crowded due to the long side chain of lysine, and the change in electrostatic properties from negative to positive charge might negatively affect acceptor binding.
Quercetin is one of the major antioxidants in fruits and vegetables, and is consumed in forms in which one or more of its hydroxyl groups are glycosylated. The nature of the position of glycosylation significantly affects its absorption and utilization in humans (32)(33)(34). The chemical synthesis of specific quercetin glucosides is both complex and costly (35,36). Recently, there has been increasing interest in biocatalytic synthesis of glucosides of quercetin (27,37) and other phenolic compounds (38) due to their potential therapeutic value in cardiovascular disease and cancer, and their antimicrobial properties. Specific quercetin mono-or diglucosides have been successfully produced in an E. coli fermentation system employing regiospecific glycosyltransferases (27,37) The majority of the soluble glucoside products are secreted to the medium that facilitates their purification. We have now demonstrated that new glycosyltransferases with altered regioselectivity for quercetin can be obtained through targeted mutagenesis. Such mutant enzymes may prove useful in the biocatalytic synthesis of specific quercetin glucosides, because the position of conjugation affects both the bioavailability and health beneficial activity of quercetin for humans (20,21). It remains to be seen whether the basic structure of UGT71G1 can be altered for preferential formation of the A-ring and B-ring (4Ј-O) glucosides.