Solvent Engineering for Nonpolar Substrate Glycosylation Catalyzed by the UDP-Glucose-Dependent Glycosyltransferase UGT71E5: Intensification of the Synthesis of 15-Hydroxy Cinmethylin β-d-Glucoside

Sugar nucleotide-dependent glycosyltransferases are powerful catalysts of the glycosylation of natural products and xenobiotics. The low solubility of the aglycone substrate often limits the synthetic efficiency of the transformation catalyzed. Here, we explored different approaches of solvent engineering for reaction intensification of β-glycosylation of 15HCM (a C15-hydroxylated, plant detoxification metabolite of the herbicide cinmethylin) catalyzed by safflower UGT71E5 using UDP-glucose as the donor substrate. Use of a cosolvent (DMSO, ethanol, and acetonitrile; ≤50 vol %) or a water-immiscible solvent (n-dodecane, n-heptane, n-hexane, and 1-hexene) was ineffective due to enzyme activity and stability, both impaired ≥10-fold compared to a pure aqueous solvent. Complexation in 2-hydroxypropyl-β-cyclodextrin enabled dissolution of 50 mM 15HCM while retaining the UGT71E5 activity (∼0.32 U/mg) and stability. Using UDP-glucose recycling, 15HCM was converted completely, and 15HCM β-d-glucoside was isolated in 90% yield (∼150 mg). Collectively, this study highlights the requirement for a mild, enzyme-compatible strategy for aglycone solubility enhancement in glycosyltransferase catalysis applied to glycoside synthesis.


■ INTRODUCTION
−6 With the exception of organic−aqueous interfacial reactions notwithstanding, 1,7 the enzymatic transformations generally happen in the bulk water phase. 2Whenever nonpolar substrates are involved, the solubility enhancement thus becomes a requirement so that the reaction efficiency meets the demands of a practical synthesis. 2,3Considering the fundamental importance and applied significance a generally applicable method would entail, numerous strategies of substrate solubilization have been proposed. 1,2,8,9These strategies can be broadly categorized according to whether they achieve solubilization in a single aqueous phase or introduce an additional, water-immiscible liquid or solid phase that solubilizes the substrate better than water does. 1,2,10There exist countless varieties of either type of the strategy.−14 The challenge, therefore, is to find a strategy that combines solubilization with the retention of the enzyme function.Solvent engineering therefore plays a key role in applied biocatalysis development. 1,2While general guidelines for enzyme usage in nonconventional media have emerged from the aggregate evidence of a large set of diverse studies, 1,2,8 the concrete strategy of substrate solubilization always necessitates a case-specific development under method adaptation to the requirements of the particular enzyme(s) used. 15,16Enzyme robustness may not be uniform to the various stressors that occur in the biocatalytic process.1][2][3]14,15 Sugar nucleotide-dependent (Leloir) glycosyltransferases (GTs) catalyze the glycosylation of acceptor substrates that represent a broad diversity of chemical structures. 17−19ertain classes of GT acceptor, including natural products and xenobiotics in particular, involve a highly nonpolar aglycone core.17,20−24 Due to the specificity and flexibility in catalysis offered in useful combination, GTs are promising enzymes for the synthesis of the corresponding acceptor glycosides.17,22,25− 27 The glycosides involve considerable interest for the diverse uses they have, ranging from chemical reference for studies of the biological metabolism 28,29 to functional ingredients in food 23,26,30,31 and cosmetic applications.23,30,32,33 However, GTs are generally perceived as "difficult" enzymes for use in applied biocatalysis. Be35 Low robustness of GTs may have been the reason that solvent engineering strategies have not been widely explored with these enzymes.36 Nevertheless, glycosylation reactions with nonpolar acceptors are expected to benefit strongly from enhanced solubilization of the substrate into the aqueous phase containing the enzyme.17,[22][23][24]27 In the current study, we focused on β-D-glucosylation of 15hydroxy cinmethylin (15HCM) from uridine 5′-diphosphate (UDP)-glucose by UGT71E5 of safflower plant (Carthamus tinctorius), as shown in Figure 1A.37,38 Cinmethylin (Figure 1B) is an agricultural herbicide, and its detoxification in planta involves C15 hydroxylation and glycosylation. 39,40 15HCM β-D-glucoside is therefore an important chemical and biological reference of xenobiotic metabolism. UG71E5 catalyzes the synthesis of 15HCM β-D-glucoside at a useful specific activity (0.43 U/mg) as well as in high selectivity.37 The reaction, however, confronts the major challenge of extremely low 15HCM solubility (≤0.1 g/L, in water at 25 °C).To improve the efficiency of 15HCM transformation into 15HCM β-Dglucoside, we here examine the two main strategies of substrate solubilization.We use organic cosolvents (DMSO, ethanol, and acetonitrile) in different concentrations. Additionally, we use water-immiscible organic solvents (n-hexane, 1-hexene, nheptane, and n-dodecane) and explore different water−organic phase ratios and incubation conditions. Lat, we use inclusion complexation 41,42 in cyclodextrin to enhance solubilization.Earlier works from this laboratory have shown the complexation of phenolic acceptors in 2-hydroxypropyl-β-cyclodextrin (HPβCD) in order to achieve intensification of GT-catalyzed C-glycosylation reactions.44 Moreover, HPβCD was readily removed during the product isolation.36,46 We find here that organic (co)solvents were not very compatible with UGT71E5 activity and that enzyme stability was also low under the alternative solvent conditions. Applied as n inclusion complex with 2-hydroxypropyl-βcyclodextrin (HPβCD), 15HCM was dissolved to ∼100 mM in a single aqueous phase, and the UGT71E5 activity was largely unimpaired by the conditions used. Full converion of 15HCM (50 mM) was shown, and 15HCM β-D-glucoside was recovered in 90% yield (∼150 mg scale).Compared to a reference reaction done with 10 mM 15HCM in 10 vol % DMSO, 37 the HPβCD-based strategy of solubility enhancement achieved ∼5-fold intensification of the synthetic transformation.In summary, therefore, our study highlights the requirement for a mild, enzyme-compatible strategy of solubilization of the acceptor substrate for the efficient application of GT catalysis to glycoside synthesis.
Enzyme Production.UGT71E5 from C. tinctorius (GenBank accession number: KX610759.1)was produced and purified with a slightly modified protocol from that described in the literature. 47The codon-optimized synthetic gene of UGT71E5 (Supporting Information) in a pET28a(+) expression vector was expressed in E. coli BL21(DE3) cells using Terrific Broth (TB) medium.The enzyme was purified by utilizing its N-terminal 6xHis-tag.Full details of the expression and purification conditions are given in the Supporting Information.Sucrose synthase from soybean (Glycine max; GmSuSy, GenBank accession number: AF030231.1)equipped with an Nterminal Strep-Tag II was produced and purified as described in an earlier work. 36,48The size and purity of UGT71E5 and GmSuSy were confirmed by SDS-PAGE (Figure S1).Protein concentration was determined based on the absorption at 280 nm on a spectrophotometer (NanoDrop One, Biozym, Vienna, Austria) using a molar extinction coefficient (GmSuSy, 104,210 M −1 cm −1 ; UGT71E5, 45,170 M −1 cm −1 ) and molecular mass (GmSuSy, 92,244 Da; UGT71E5, 55,167 Da) calculated from the amino acid sequence.
Enzyme Activity Assays.The specific activity of UGT71E5 toward free and HPβCD-encapsulated 15HCM was determined from a reaction containing 1.0 mM 15HCM and 2.0 mM UDP-glucose. 37he activity of GmSuSy on UDP (2.0 mM) was determined from a reaction with 500 mM sucrose.Both activity assays were carried out in Tris buffer (50 mM, pH 7.4) supplemented with 5 mM MgCl 2 .Full details of the activity assays are provided in the Supporting Information.The results are shown in Figures S2 and S3.
Reactions with Substrate Feeding.The experiments with 15HCM feeding included UDP-glucose regeneration by GmSuSy.The reaction  .e Organic phase was 0.1 mL.
Aqueous phase was 0.4 mL, containing 250 mM UDP-glucose and 1.0 mg/mL UGT71E5.f Organic and aqueous phases were each 0.4 mL.In the aqueous phase, UDP-glucose was 70 mM and UGT71E5 was 1.5 mg/mL.g Values were obtained from both aqueous and organic phases as 15HCM β-D-glucoside was observed in the organic phase.
Encapsulation of 15HCM by HPβCD.HPβCD inclusion complexes of 15HCM were prepared with a protocol adapted from the literature. 36The host−guest molar ratio was varied between 1:1 and 4:1.For the complexation at 3:1 ratio (used for preparing the substrate for cascade reactions), HPβCD (420 mg, 0.30 mmol) was dissolved in deionized water (0.5 mL) by microwave heating (Micro-Chef V98, Moulinex, Austria; 20 s at 750 W) and mixed by inversion every 15 min.For inclusion complex formation, 15HCM (29.04 mg, 0.10 mmol) was added and dissolved as described for HPβCD.The final volume was set to 1.0 mL with deionized water, and the mixture was equilibrated at 70 °C for 30 min.The mixture was further incubated at 22 °C for 3 h and centrifuged for 5 min (21,130 g, 22 °C).The concentration of complexed 15HCM was measured based on a calibration curve of free 15HCM on high-performance liquid chromatography (HPLC) (Figures S9 and S10).
Cascade Reactions of UGT71E5 and GmSuSy on HPβCD-Encapsulated 15HCM.The reactions were performed in Tris buffer (50 mM, pH 7.4) containing MgCl 2 (5 mM) in the final volume of 0.3 mL.Sucrose (500 mM), 15HCM-HPβCD complex (5−50 mM), and UDP (1.0 mM) were dissolved in the reaction buffer, followed by the addition of UGT71E5 (1.2−3.7 mg/mL).The reactions were started with GmSuSy (0.25−2.1 mg/mL) addition.In the 50 mM 15HCM-HPβCD reaction with batch addition of the enzymes, 3.7  mg/mL UGT71E5 and 2.1 mg/mL GmSuSy were supplied at the start and 1.1 mg/mL UGT71E5 after 8 h of reaction.The reactions were performed at 30 °C with agitation (500 rpm, Thermomixer) and sampled as described in the Supporting Information under "Enzyme Activity Assays".Control reactions in the absence of UGT71E5 and GmSuSy were treated identically with the reactions in the presence of the enzymes.The results are listed in Figure 5.
Production and Isolation of 15HCM β-D-Glucoside.The reactions (24 × 0.3 mL, 7.2 mL total volume) were carried out under the following conditions: 50 mM 15HCM (in HPβCD complex), 500 mM sucrose, and 1.0 mM UDP in 50 mM Tris buffer (pH 7.4, containing 5 mM MgCl 2 ) at 30 °C with agitation (500 rpm).UGT71E5 (2.5 mg/mL) and GmSuSy (0.6 mg/mL) were added at start and further supplied after 8 h of reaction (1.6 mg/mL UGT71E5 and 0.4 mg/mL GmSuSy).The reaction was sampled at desired time points by quenching 10 μL of reaction mixture with 90 μL of acetonitrile, and the samples were subjected to thin-layer chromatography (TLC) and HPLC analysis.After reaching full conversion to the glycoside product (after 23 h, Figure S11), the reaction mixtures were pooled, and the enzymes were removed by ultrafiltration (Vivaspin 10 kDa cutoff, Sartorius, Goẗtingen, Germany).The enzyme-free solution was lyophilized (Christ Alpha 1−4 lyophilizer, B. Braun Biotech International, Melsungen, Germany) and redissolved in ∼4 mL of deionized water.The product mixture was loaded into a column containing silica gel C18 (45 g, 0.035−0.070mm, Carl-Roth), and deionized water (270 mL) was used for washing out impurities (fructose and sucrose).The target product was eluted with 270 mL of acetonitrile/deionized water (1:1) mixture.The fractions (each 15 mL) were analyzed on TLC and the compounds visualized by UV (15HCM β-D-glucoside) and staining solution (sucrose, fructose, and HPβCD), as shown in Figure S12 A. Fractions containing 15HCM β-D-glucoside were concentrated under reduced pressure (40 °C; Heidolph Laborota 4000 rotary evaporator equipped with a Vacuubrand PC2001 pump and a CVC2000II controller, Wertheim, Schwabach, Germany), lyophilized, and redissolved in a solvent mixture of 1-butanol/acetic acid/deionized water (2:1:1).The product was loaded into a silica 60 column (30 g, 0.04−0.063mm, Machery-Nagel, Duren, Germany) for HPβCD removal.The column was washed with a 1-butanol/acetic acid/deionized water (2:1:1) mixture until the target product was eluted in the fractions (10 mL, analysis on TLC, Figure S12B).The product-containing fractions were pooled, and the solvent removed was by rotary evaporation and lyophilization.The obtained solid material was dissolved in 150 mL of methanol (for silica removal), and the supernatant was collected after centrifugation (20 min, 16,100 g).The pure product was recovered from methanol after rotary evaporation and lyophilization and analyzed by nuclear magnetic resonance (NMR) (Figures S13−S16).
Analytical Methods.Reversed-Phase HPLC.The acceptor substrate (15HCM) and its β-D-glucoside product were separated with an Agilent 1200 Series HPLC-UV system on a Kinetex EVO C18 column (5 μm, 100 Å, 150 × 4.6 mm; Phenomenex, Aschaffenburg, Germany) using a gradient method (1.0 mL/min) of water and acetonitrile (each containing 0.1% (v/v) formic acid) as the mobile phase. 37,38Gradient: 20−75% acetonitrile in 5.5 min, 75% acetonitrile for 2 min, 75−20% acetonitrile in 0.1 min, and 20% acetonitrile for 4.5 min.The acceptor substrate and the glucoside product were detected by UV at a 203 nm wavelength.UDP-glucose and UDP were separated on a Kinetex C18 column (5 μm, 100 Å, 50 × 4.6 mm; Phenomenex) using an isocratic method (2.0 mL/min) with 5% acetonitrile and 95% tetrabutylammonium bromide (TBAB) buffer (40 mM TBAB, 20 mM K 2 HPO 4 /KH 2 PO 4 , pH 5.9) as the mobile phase.UDP-glucose and UDP were detected by UV at a 262 nm wavelength.The amount of 15HCM/15HCM β-D-glucoside/UDP/ UDP-glucose in monophasic reaction systems was determined based on the relative integrated peak areas.The substrate and product concentrations in biphasic reactions were determined by referring to the peak areas of authentic standards and based on the general principle of mass balance.
Nuclear Magnetic Resonance.The acquisitions of 15HCM β-Dglucoside were carried out in DMSO-d 6 (99.8%D) on a Bruker Avance III NMR spectrometer (Bruker, Rheinstetten, Germany); 300 MHz for 1 H NMR, COSY, and HSQC measurements and 75 MHz for a 13 C NMR measurement were used.The spectra were analyzed using MestReNova 14.

■ RESULTS AND DISCUSSION
Three approaches were pursued with the aim of improving the efficiency of the 15HCM β-D-glucoside synthesis.First, the use of organic cosolvents was investigated.Second, the use of water-immiscible organic solvents was examined.Last, inclusion complexation in HPβCD was studied.The scope of each approach was investigated in detail.
Organic Cosolvents.The UGT71E5 reaction for 15HCM β-D-glucoside synthesis was previously performed using DMSO cosolvent in the range 4−16 vol %. 37 The 15HCM concentration was 1.0 or 10 mM with UDP-glucose used in a 1.5-fold molar excess.Conversion into 15HCM β-D-glucoside was ≥90% (9.0 mM; 4.1 g/L).Enzyme activity was decreased by ∼30% when DMSO was increased from 4 vol % (0.43 U/ mg) to 16 vol % (Table S1). 37Here, DMSO was used at 15, 20, and 50 vol %, and to further increase the acceptor concentration, 15HCM was dissolved at up to 30 mM.Time courses of the UGT71E5 reaction are shown in Figure S4.Activities and conversion parameters are summarized in Table S1.UGT71E5 was inactive at 50 vol % DMSO.The reaction at 20 vol % DMSO gave ∼20.5 mM product (∼9.3g/L; ∼70% conversion).The UGT71E5 activity (0.27 ± 0.02 U/mg; N = 3) was 63% that of the standard activity (Table S1).Reactions at 15 and 20 vol % DMSO slowed down strongly after ∼6 h (Figure S4), despite incomplete conversion of 15HCM (≤53%) and UDP-glucose remaining in excess at this point.The results thus suggest that reaction progress was limited by loss of enzyme activity under the conditions used.We therefore explored additional cosolvents (ethanol and acetonitrile) and used each at 15 vol % to dissolve 20 mM 15HCM.Both cosolvents are generally known to be denaturing to proteins, but the immediate effect on UGT71E5 was impossible to predict and thus required experimental assessment.As shown in Figure S4 and Table S1, UGT71E5 exhibited low activity under these conditions.Overall, therefore, the cosolvent approach encountered a major obstacle due to enzyme instability and thus seemed to be of limited use for reaction intensification.We did not consider more-specialized, water-miscible solvents 1−6 such as ionic liquids or deep eutectic solvents at this stage, in particular because there was no immediate suggestion from the literature that these solvents would be fundamentally less detrimental to enzyme activity than the classical solvents already examined.
Substrate Feeding.Given the low cosolvent tolerance of UGT71E5, there remained the possibility to supply the total amount of 15HCM substrate not at once but in smaller portions.15HCM feeding in the way considered would allow for a lower DMSO cosolvent concentration to be used that might be better compatible with enzyme stability.Here, a batch reaction of 10 mM 15HCM was performed in 10 vol % DMSO (Figure 2), and after 24 h, the same amount of 15HCM (dissolved in pure DMSO) was added to increase the total cosolvent concentration by 2 vol % (Figure 2A) and 10 vol % (Figure 2B).The reactions were tracked by analyzing 15HCM and 15HCM β-D-glucoside in solution, and the insoluble (precipitated) 15HCM was calculated from the mass balance.Figure 2A shows that after the first substrate addition, 15HCM precipitated in large amounts.This was unexpected because the 15HCM concentration was not increased substantially above the one used initially, and the DMSO concentration was increased from 10 to 12 vol %.As shown in Figure 2A, the release of 15HCM β-D-glucoside continued at a slow rate.The 15HCM addition was repeated after 48 h, leading to further precipitation and no increase in the 15HCM consumption rate.Performing the 15HCM addition with increase in total DMSO concentration to 20 vol % (Figure 2B) resulted in a smaller amount of substrate precipitation; yet, the 15HCM β-D-glucoside formation was slow.The product concentrations (≤13 mM) and conversions (≤43%) reached in the experiments were not convincing to justify a relatively complex mode of operating the reaction.Solubility control of 15HCM appeared to be difficult under conditions of accumulating 15HCM β-D-glucoside.Slowdown of the reaction was not prevented, suggesting impaired UGT71E5 activity.Earlier work on the GT-catalyzed natural product Cglycosylation 43 has shown that the enzyme activity and stability can be affected negatively by the insoluble substrate precipitated in the reaction.Considering the goal of this study to enhance the 15HCM β-D-glucoside synthesis by ∼10-

Journal of Agricultural and Food Chemistry
fold to a product concentration of ∼50 mM, we realized the requirement of a different approach for 15HCM solubility enhancement.
Water-Immiscible Organic Solvents.The aqueous− organic two-phase reaction does not strictly enhance the solubility in the aqueous phase; yet, it constitutes a convenient way of enhancing the amount of dissolved substrate introduced into the reaction mixture.Substrate supply into the aqueous phase occurs by liquid−liquid mass transfer that is typically more efficient and more easily controlled than substrate dissolution from a solid precipitate.Following the wellestablished general guidelines of a water-immiscible solvent use for enzymatic reactions, 11,15,49,50 we selected strongly hydrophobic solvents characterized by a log P OW (octanol− water partition coefficient) of ≥2 (see Table S2 for a parameter summary of the used solvent).There is furthermore evidence that linear nonbranched alkanes are less detrimental to enzymes than esters, cyclic ethers, simple ethers, aromatic chemicals, and cyclic alkanes. 15We therefore chose 1-hexene, n-hexane, n-heptane, and n-dodecane covering the log P OW range of ∼3 to ∼7 (Table S2).We note the relatively high evaporation rate of all solvents except n-dodecane (Table S2), rendering them nonpreferred from the health−environment− safety point of view. 49However, for a preliminary solvent screening to assess a two-phase reaction for 15HCM β-Dglucoside synthesis in principle, all solvents were suitable.As discussed later, a possible advantage of the two-phase mixture is selective partitioning of 15HCM and 15HCM β-D-glucoside to the organic and the aqueous phase, respectively.
The first set of experiments was performed at 4:1 ratio of aqueous and organic phases, with agitation at 300 rpm (Thermomixer) and supply of 15HCM solely through the organic solvent at 20 or 100 mM.The enzyme activities determined from the initial reaction rates were ∼28 mU/mg for n-dodecane, ∼33 mU/mg for n-heptane, and ∼18 mU/mg for 1-hexene.The activities increased up to 3.8-fold in response to the 5-fold increase in 15HCM concentration in the organic solvent (Table 1).Note that variation in the UDP-glucose concentration used (30 or 250 mM; Table 1) did not cause change in the reaction rate.Activities measured in the twophase reactions were lowered by ≥13-fold compared to the standard activity determined in a single aqueous phase (1.0 mM 15HCM, 4 vol % DMSO, Table S1).The results suggested that the 15HCM concentration in the aqueous phase might have been too low to saturate UGT71E5 for full activity.Time courses of the enzymatic reactions at 20 mM 15HCM in the organic phase are shown in Figure 3. 15HCM β-Dglucoside accumulating in the aqueous phase indicated high conversion (≥81%) of 15HCM supplied.Direct measurements of 15HCM from the organic phase confirmed the results based on the close mass balance.Partial evaporation of n-hexane (≥20%) and 1-hexene (≥50%) was, however, noted.The conversion was lower (60−69%) when 15HCM was supplied at 100 mM in the organic phase (Table 1 and Figure S5).The agitation rate was varied in the range of 300−800 rpm, considering that enhanced energy input by agitation could benefit the interfacial mass transfer through its combined effect on the overall mass-transfer coefficient and the specific liquid− liquid exchange surface area.An increase in specific surface area could, however, also result in enhanced enzyme inactivation.Results in Figure S5B show that the UGT71E5 activity was increased (∼1.8-fold) due to enhanced agitation, consistent with the idea that liquid−liquid mass transfer is partially rate-limiting for the overall reaction rate ("activity") under standard agitation conditions (300 rpm).The final 15HCM β-D-glucoside concentration reached was hardly dependent on the agitation rate (Figure S5).Time courses of reaction (Figure S5C−E) show differences in 15HCM β-Dglucoside release at varied agitation rates, primarily in the early phase of the conversion.We interpret these results to indicate that increased agitation did not accelerate the enzyme inactivation substantially.If this were the case, one would expect the 15HCM conversion to decrease at an increased agitation rate.
Having shown the two-phase UGT71E5 reaction in principle, we increased in the next step the volume portion of the organic solvent to an aqueous−organic phase ratio of 1:1.The 15HCM concentration in the organic solvent was 30 mM and 75 mM.The enzyme activities were slightly increased in n-dodecane (70 mU/mg, 75 mM 15HCM) and even decreased up to 1.9-fold in n-heptane (57 mU/mg, 75 mM 15HCM) and 1-hexene (37 mU/mg, 75 mM 15HCM) compared to the previous reactions at 4:1 phase ratio, as shown in Table 1, and they still did not reach the benchmark of the standard assay (431 mU/mg).The product yields decreased with increasing 15HCM concentration in the organic phase, from 37−55% at 30 mM to 27−33% at 75 mM.Time courses of the 30 mM reactions (Figure S6) show that the 15HCM conversion into β-D-glucoside slowed down after ∼3 h, and there was little progress of reaction between 6 and 24 h.The kinetic behavior was highly similar for all reactions and appeared to be independent of the organic solvent used.It was comparable to the kinetic behavior seen in the other time courses previously shown in Figures 3 and S5C−E.The point of decline in the reaction rate was always localized at a similar time; yet, it seemed to be unrelated to the degree of conversion reached or the concentration of 15HCM β-D-glucoside released.The plausible explanation of the observed behavior was general instability of UGT71E5 under the conditions used.We also analyzed the distribution of 15HCM and 15HCM β-D-glucoside between the aqueous and organic phase in the different solvent reactions.Results in Figure S7 reveal that except for minor concentrations found in 1-hexene (0.10 ± 0.02 mM; N ≥ 3) and n-hexane (0.11 ± 0.03 mM; N ≥ 3), 15HCM β-D-glucoside was present exclusively in the aqueous phase.15HCM by contrast was present primarily in the organic phase, with only small concentrations found in the water phase (1.4−1.7 mM; N ≥ 3).
Considering the possibility that the two-phase reactions were limited by the 15HCM substrate available in the aqueous phase, we performed further experiments with n-dodecane solvent at 30 mM 15HCM and used DMSO (10 vol % of aqueous phase volume) as cosolvent additionally.The idea was to tune the phase partitioning of 15HCM so that the substrate concentration in the water phase was enhanced.The aggregate evidence up to this point suggested that n-dodecane was preferred among the series of solvents here explored.The aqueous/organic phase ratio varied from 1:1 to 1:3.Time courses of 15HCM β-D-glucoside release are shown in Figure 4.The UGT71E5 activity did not benefit from the change in the reaction conditions (Table 1).The product concentration in the aqueous phase increased from 17 to 27 mM as the volume portion of n-dodecane increased 3-fold.However, the 15HCM β-D-glucoside yield based on the total 15HCM converted decreased from 57 to 30% (Table 1).Analysis of the phase partitioning of 15HCM and 15HCM β-D-glucoside Journal of Agricultural and Food Chemistry (Figure S8) showed an increase in 15HCM available in the water phase due to the change in the phase ratio by 1.9-fold, from 1.9 mM at 1:1 ratio (panels A and D) to 3.7 mM at 1:3 ratio (panels C and F).15HCM β-D-glucoside was present in the water phase.The effect of the DMSO cosolvent on the aqueous 15HCM concentration was moderate, and a 36% increase was calculated from the results in Figure S7 (panel E; ∼1.4 mM) and the Supporting Information, Figure S8 (panel D; ∼1.9 mM).
As an intermediate conclusion, the study of the UGT71E5 reaction in an aqueous−organic two-phase mixture provided an important insight into substrate and product phase partitioning and the kinetic behavior of the enzymatic transformations under the different conditions of organic solvent and reaction operation used.At this stage, however, we also realized that it would be difficult to use the two-phase reaction for a 15HCM β-D-glucoside synthesis that combined high product concentration (target: ≥50 mM) with a suitable conversion of the 15HCM substrate (target: ≥90%).Although the range of possible solvents to be examined is still large, with deep eutectic solvents and emerging biobased solvents such as Cyrene representing interesting classes, 1−6 we were not convinced based on the evidence already shown that the approach of water-immiscible solvent would hold promise in particular.We therefore considered 15HCM solubility enhancement by inclusion complexation in HPβCD as an alternative.
HPβCD Inclusion Complex.−46 Unlike here, solvent engineering approaches have not been investigated in detail for C-glycosylation.By applying the encapsulation of phloretin (a flavonoid) in HPβCD, enzymatic syntheses of the phloretin 3′-β-D-glucoside (nothofagin) 36,44,45 and the phloretin 3′,5′-diβ-D-glucoside 46 were shown at concentrations of up to 50 mM and higher, exceeding the water solubility of the noncomplexed phloretin by about 10 2 -fold.Conversion of the phloretin acceptor was excellent as well (≥90%). 36,44,46Solubilization by inclusion complex formation with HPβCD, clearly, increases the complexity of the overall synthetic procedure compared to the use of a standard organic (co)-solvent, 36,43,46 and it would also add to the production costs of a biocatalytic process.Given the evidence from previous research that (a) HPβCD was well tolerated by different GT enzymes (including the sucrose synthase used here for UDP-glucose regeneration 36,43,44 ) and (b) it was removed relatively easily during the product isolation, 36,46 we considered the strategy of complexation by HPβCD to be promising for reaction intensification in 15HCM β-D-glucoside synthesis.
15HCM complexation was examined at a concentration of 50 mM, and the degree of solubilization was measured in the presence of varied HPβCD concentrations, representing a host−guest molar ratio between 1:1 and 4:1.A 3-fold excess of HPβCD was required to dissolve 15HCM to ≥95% (Figures S9 and S10).Using the host−guest molar ratio of 3:1, 15HCM was successfully encapsulated up to a 100 mM concentration.When offered 15HCM as an inclusion complex with HPβCD, the UGT71E5 activity was retained (0.32 ± 0.02 U/mg; ∼79 ± 6%; N = 2) compared to the reaction in the standard assay (Figure S2).Reactions were performed at different 15HCM concentrations in the range of 5.0−50 mM.The UDP-glucose donor substrate was generated in situ from UDP (1.0 mM) by the sucrose synthase reaction from sucrose present in excess (500 mM; GmSuSy activity assay in Figure S3).The results are shown in Figure 5.Besides 15HCM and 15HCM β-Dglucoside, UDP and the intermediary UDP-glucose were also tracked analytically.The conversion of 15HCM into β-Dglucoside proceeded to completion in all reactions (Figure 5A−E).The initial rate of 15HCM β-D-glucoside release corresponded to expectation from the enzyme concentration used and the specific activity of UGT71E5.The results show that the UGT71E5 activity was utilized efficiently in the reactions, enabled by a suitable supply of UDP-glucose from sucrose and UDP.The results also imply that the UDP-glucose formation was never rate-limiting for the overall conversion, supporting the design of the cascade reactions in Figure 5 with GmSuSy activity used in ≥3-fold excess.Measurements of UDP and UDP-glucose affirm the conclusion of rate limitation.In all reactions (Figure 5A−E), the initial UDP concentration (∼1.0 mM) dropped to a much lower value that remained largely constant throughout the conversion.The distribution of total UDP into the steady-state levels of UDP-glucose and UDP implies a UDP-glucose formation rate (GmSuSy) that exceeded the corresponding consumption rate (UGT71E5).
All reactions in Figure 5 feature a decrease in the product release rate at some point of the time course, typically when ∼50% conversion of the available 15HCM was achieved.The decline in rate might arise due to several factors in combination that time course analysis alone cannot resolve.The availability of a free 15HCM substrate from the inclusion complex with HPβCD to UGT71E5 could be an issue.Further work beyond the scope of this research is needed for clarification.However, the experiment with the addition of fresh enzyme after 8 h (Figure 5E) shows that inactivation of UGT71E5 does not control the productivity of the overall reaction of the HPβCD complex of 15HCM.The reaction rate was only modestly enhanced upon the enzyme addition to increase of the total loading of UGT71E5 by 30% (Figure 5E).Difference to enzymatic reactions in the presence of organic (co)solvent is noted for these reactions involve loss of UGT71E5 activity as a main factor of conversion efficiency.The HPLC traces (Figure 5F) display the successful conversion of 50 mM 15HCM into the corresponding β-Dglucoside achieved in 24 h, while the control reaction in the absence of enzymes shows intact 15HCM (Figure 5F).Overall, the UGT71E5-GmSuSy cascade reaction was efficient for 15HCM β-D-glucoside synthesis at substantial intensification of the final product concentration (≥5-fold) when the 15HCM substrate was supplied as an inclusion complex with HPβCD.
Preparative Biotransformation with Product Isolation.Preparative synthesis was performed at ∼160 mg scale of 15HCM β-D-glucoside.15HCM was offered as the HPβCD inclusion complex at 50 mM concentration of the acceptor substrate.Conversion of 15HCM to the corresponding β-Dglucoside was monitored by HPLC and TLC, and the enzymes were removed at the point of complete conversion (≥99%, after 23 h; Figure S11).The product was isolated by two-step silica column chromatography.The first step was performed on a silica C18 resin, from which the target product was coeluted with HPβCD (Figure S12A).Secondary chromatography was conducted on a silica 60 column using the TLC eluent as a mobile phase, leading to successful separation of 15HCM β-Dglucoside from HPβCD (Figure S12B).However, operating the column with a mobile phase consisting of 1-butanol, acetic acid, and water partially dissolved silica into the product fractions.Methanol extraction was used for removing the dissolved silica, and the final extract showed over 95% purity of 15HCM β-D-glucoside.The target product was obtained in excellent yield (147 mg, 90%); yet, the explorative purification method could be further streamlined by excluding the C18 column.Identity of the obtained 15HCM β-D-glucoside was confirmed by 1 H-and 13 C NMR, HSQC, and COSY analysis (Figures S13−S16).
In summary, we have performed a systematic study of 15HCM solubility enhancement for UGT71E5 reaction intensification, with the aim of making 15HCM β-D-glucoside synthesis more efficient.−46 Here, approaches based on organic solvents miscible or immiscible with the aqueous buffer failed due to limited enzyme robustness.Inclusion complexation with HPβCD represented a mild, enzyme-compatible approach to 15HCM solubility enhancement.The study of 15HCM glycosylation by UGT71E5 can be relevant in general for the cascade catalysis of Leloir glycosyltransferases applied to glycoside synthesis.

■ ASSOCIATED CONTENT
* sı Supporting Information

Figure 1 .
Figure 1.(A) Reaction scheme for the synthesis of 15HCM β-D-glucoside by glycosyltransferase UGT71E5 coupled with UDP-glucose production and regeneration by a sucrose synthase GmSuSy.(B) Chemical structures of cinmethylin/15HCM enantiomers.

Figure 3 .
Figure 3. Glycosylation of 15HCM in a 1:4 ratio of organic solvent to aqueous solution in the biphasic system.15HCM (20 mM: closed circles) was dissolved in organic solvents [0.1 mL: (A) n-dodecane, (B) n-heptane, and (C) 1-hexene].UGT71E5 (1 mg/mL) and UDP-glucose (30 mM) were dissolved in an aqueous solution (0.4 mL).15HCM β-D-glucoside (open circles) was formed and collected in the aqueous phase.The data are summarized in Table 1.The standard deviations shown are from the analytical determination by HPLC (n ≥ 2).The number N of experiments was 1.

Table 1 .
Glycosylation of 15HCM in the Organic−Aqueous Biphasic Reaction with Water-Immiscible Organic Solvents a Volumetric percentage of solvent in total volume.b Standard deviations are from N ≥ 2 determinations in the initial phase (≤6 h) of the reaction.c Standard deviations are from n ≥ 2 analytical determinations.d = × [ ] [ ] Yield in the aqueous phase (%) 100 observed 15HCM D glucoside in the aqueous phase theoretical 15HCM D glucoside in the aqueous phase