Glucosylation of (±)-Menthol by Uridine-Diphosphate-Sugar Dependent Glucosyltransferases from Plants

Menthol is a cyclic monoterpene alcohol of the essential oils of plants of the genus Mentha, which is in demand by various industries due to its diverse sensorial and physiological properties. However, its poor water solubility and its toxic effect limit possible applications. Glycosylation offers a solution as the binding of a sugar residue to small molecules increases their water solubility and stability, renders aroma components odorless and modifies bioactivity. In order to identify plant enzymes that catalyze this reaction, a glycosyltransferase library containing 57 uridine diphosphate sugar-dependent enzymes (UGTs) was screened with (±)-menthol. The identity of the products was confirmed by mass spectrometry and nuclear magnetic resonance spectroscopy. Five enzymes were able to form (±)-menthyl-β-d-glucopyranoside in whole-cell biotransformations: UGT93Y1, UGT93Y2, UGT85K11, UGT72B27 and UGT73B24. In vitro enzyme activity assays revealed highest catalytic activity for UGT93Y1 (7.6 nkat/mg) from Camellia sinensis towards menthol and its isomeric forms. Although UGT93Y2 shares 70% sequence identity with UGT93Y1, it was less efficient. Of the five enzymes, UGT93Y1 stood out because of its high in vivo and in vitro biotransformation rate. The identification of novel menthol glycosyltransferases from the tea plant opens new perspectives for the biotechnological production of menthyl glucoside.


Cloning and Expression of UGT93Y1 and UGT93Y2
To add additional members to an already existing UGT library, putative UGT sequences in Camellia sinensis var. sinensis were searched by BLAST search in the Tea Plant Information Archive database (http://tpia.teaplant.org; accessed on 8 September 2021). The coding sequences of the full-length candidate genes TEA009739 and TEA009753 were randomly selected. Their cDNA sequences were obtained from the leaves of the tea plant C. sinensis var. sinensis, cloned into the pGEX-4T1-vector and the encoded UGTs were successfully produced as N-terminal glutathione S-transferase (GST) fusion proteins in Escherichia coli BL21(DE3)pLysS. After affinity purification and verification by SDS-PAGE, the analyses showed clear bands for the fusion proteins at around 80 kDa (Supplementary material Figure S1). In addition, a band of about 27 kDa was visible for the GST, and few bands showed co-purified proteins. The production of UGT-GST fusion proteins was confirmed by Western blot analysis by using a monoclonal anti-GST antibody (Supplementary material Figure S1). The encoded proteins were subsequently assigned UGT93Y1 and UGT93Y2 by the UGT Nomenclature Committee. UGT93Y1 and UGT93Y2 are 1404 and 1413 base pairs long, respectively, encode proteins with a length of 467 and 470 amino acids, respectively, and share a sequence identity of 74.3%. Both sequences show the characteristic features of functional UGTs of the CAZy family 1, including the catalytically active amino acids histidine at position 29 [38], the activating aspartic acid (position 134) and the 44 amino acid long PSPG box [19,31] (Supplementary material Figure S2). The GSS motif, a feature of mono-glucosyltransferases [39], is replaced with TSS at position 413-415 (UGT93Y1) and ASS at position 416-418 (UGT93Y2). Eventually, Escherichia coli Waksman cells were transformed with the respective plasmid carrying UGT93Y1 or UGT93Y2 for whole-cell biotransformation.

In Vivo Substrate Screening Using E. coli Waksman
Escherichia coli Waksman cells expressing recombinant plant UGTs were employed as whole-cell biocatalysts in order to identify (±)-menthol UGTs, similar to the method described in [37]. Menthyl glucoside produced by the biocatalysts was identified by LC-MS (Supplementary material Figure S3). By comparing the intensities of the signals for the pseudo-molecular ion m/z 363 [M + HCOO] − of (±)-menthyl glucoside, the relative biotransformation activities of the biocatalysts were calculated, with the highest value set at 100% (Supplementary material Figure S4). The sreening of a library of family 1 UGTs containing enzymes from Arabidopsis thaliana, C. sinensis, Catharanthus roseus, Fragaria x ananassa, Fragaria x vesca, Nicotiana benthamiana, Rubus idaeus, Starmerella bombicola and Vitis vinifera revealed that only 5 out of 57 UGTs were able to glucosylate (±)-menthol in vivo (Supplementary material Figure S4). The most efficient biocatalyst was the strain expressing the newly identified UGT93Y1, followed by UGT85K11 and UGT93Y2. All three enzymes originated from the tea plant C. sinensis: UGT72B27 from V. vinifera and UGT73B24 from F. x ananassa glucosylate menthol to a minor extent. The (±)-menthyl glucoside was not produced by the other 52 UGTs after 48 h of biotransformation. In order to test the stereospecificity of the five biocatalysts, UGTs were incubated individually with (±)-menthol and (−)-menthol ( Figure 1). As early as six hours after substrate addition, the corresponding menthyl glucosides were already detected in the cell supernatant, and their concentration increased for at least up to 48 h. Similar biotransformation rates were calculated after the addition of racemic and (−)-menthol, suggesting that the enzymes are not stereospecific for menthol enantiomers.
Molecules 2021, 26, x FOR PEER REVIEW stereospecificity of the five biocatalysts, UGTs were incubated individually with (±) thol and (-)-menthol ( Figure 1). As early as six hours after substrate addition, the sponding menthyl glucosides were already detected in the cell supernatant, and the centration increased for at least up to 48 h. Similar biotransformation rates were calc after the addition of racemic and (-)-menthol, suggesting that the enzymes are not s specific for menthol enantiomers.

Qualitative Substrate Screening of UGT93Y1 and UGT93Y2 by LC-MS
Since the two recombinant UGTs identified in C. sinensis var. sinensis (UGT93Y UGT93Y2) efficiently glucosylated menthol, they were analyzed in more detail. Sub screening was performed by incubating the purified protein extract of UGT93Y UGT93Y2 with the sugar donor UDP-glucose and various potential acceptor (mono nol) substrates. The reaction products of the glycosyltransferase reaction were an by LC-MS, and glucosides were identified by their MS and MS2 spectra, relative ret time and by comparison with the spectral data of authentic reference material. UG revealed superior enzymatic activity towards the monocyclic monoterpenoid (±)-me ( Figure 2A). Moreover, the enzyme was able to glucosylate the menthol isome isomenthol and (+)-neomenthol, as well as the bicyclic monoterpenoid fenchyl a ( Figure 2B-D). The three glucosides of the menthol isomers showed correspo pseudo-molecular ions for the glucoside product at m/z 363 [M+HCOO] -, and fench coside was detected by means of the pseudo-molecular ion m/z 361 [M+HCOO] -. D the high amino acid sequence similarity, UGT93Y2 glycosylated menthol only to a extend (Supplementary material Figure S5). Therefore, the following quantitativ strate screening using UDP Glo™ Glycosyltransferase assay was performed UGT93Y1 only.

Qualitative Substrate Screening of UGT93Y1 and UGT93Y2 by LC-MS
Since the two recombinant UGTs identified in C. sinensis var. sinensis (UGT93Y1 and UGT93Y2) efficiently glucosylated menthol, they were analyzed in more detail. Substrate screening was performed by incubating the purified protein extract of UGT93Y1 and UGT93Y2 with the sugar donor UDP-glucose and various potential acceptor (monoterpenol) substrates. The reaction products of the glycosyltransferase reaction were analysed by LC-MS, and glucosides were identified by their MS and MS2 spectra, relative retention time and by comparison with the spectral data of authentic reference material. UGT93Y1 revealed superior enzymatic activity towards the monocyclic monoterpenoid (±)-menthol, (Figure 2A). Moreover, the enzyme was able to glucosylate the menthol isomers (+)-isomenthol and (+)-neomenthol, as well as the bicyclic monoterpenoid fenchyl alcohol ( Figure 2B-D). The three glucosides of the menthol isomers showed corresponding pseudo-molecular ions for the glucoside product at m/z 363 [M + HCOO] − , and fenchyl glucoside was detected by means of the pseudo-molecular ion m/z 361 [M + HCOO] − . Despite the high amino acid sequence similarity, UGT93Y2 glycosylated menthol only to a lesser extend (Supplementary material Figure S5). Therefore, the following quantitative substrate screening using UDP Glo™ Glycosyltransferase assay was performed with UGT93Y1 only.

UDP Glo™ Glycosyltransferase Assay
In order to characterize the enzyme activity of UGT93Y1 in more detail with respect to its particular substrate specificity towards menthol, a quantitative substrate screening was performed using the UDP Glo™ Glycosyltransferase assay [40]. In this assay, the enzymatically released amount of UDP is determined to calculate relative activities. The preferred substrate for UGT93Y1 was (±)-menthol (100% corresponding to 0.11 nkat mg −1 ), whereby the enzyme activity towards the two isomeric forms was lower ( Figure 3A). In addition, the recombinant UGT93Y1 showed UDP-α-D-glucose hydrolase activity when no acceptor substrate was added to the enzyme reaction as a low amount of UDP was detectable. The glycosyltransferase reaction was optimized with respect to protein amount, incubation time, incubation temperature and pH, and the enzyme kinetics for UGT93Y1 were determined using 0 to 1000 µM (±)-menthol ( Figure 3B). The enzyme activity versus substrate concentration graph shows a substrate inhibition curve [41]. Calculations of the kinetic data assuming a two-site sequential enzyme model with identical dissociation constants produced 7.6 and 1.5 nkat/mg for the maximum reaction velocities v max1 and v max2 , respectively, while the dissociation constant K D was 453.1 µM.

UDP Glo™ Glycosyltransferase Assay
In order to characterize the enzyme activity of UGT93Y1 in more detail with resp to its particular substrate specificity towards menthol, a quantitative substrate screen was performed using the UDP Glo™ Glycosyltransferase assay [40]. In this assay, the zymatically released amount of UDP is determined to calculate relative activities. The p ferred substrate for UGT93Y1 was (±)-menthol (100% corresponding to 0.11 nkat mg whereby the enzyme activity towards the two isomeric forms was lower ( Figure 3A) addition, the recombinant UGT93Y1 showed UDP-α-D-glucose hydrolase activity w dissociation constants produced 7.6 and 1.5 nkat/mg for the maximum reaction veloc vmax1 and vmax2, respectively, while the dissociation constant KD was 453.1 μM.
The in vitro enzyme activity assays by qualitative LC-MS analysis ( Figure 2), qu tative substrate screening ( Figure 3A) and kinetic analysis performed by UDP Glo™ cosyltransferase assay ( Figure 3B) as well as in vivo experiments using E. coli Waks cells ( Figure 1) revealed high glycosyltransferase activity of UGT93Y1 towards men resulting in the formation of menthyl glucoside. Figure 3. Determination of the relative enzymatic activities with UDP Glo™ Glycosyltransferase assay (A) and UGT93Y1 enzyme kinetics for (±)-menthol (B). The relative enzyme activity was calculated after measuring the release of UDP during the glucosylation reaction. The highest enzymatic activity was set at 100%, and the relative enzyme activities calculated (A). The enzyme activity versus substrate concentration graph for UGT93Y1 from C. sinensis var. sinensis was determined using the UDP Glo™ Glycosyltransferase assay. The calculation was performed with KaleidaGraph (https://www.synergy.com/; v4.5.4; accessed on 9 September 2021).

Characterization of Menthyl Glucoside by NMR
The biotransformation product of (±)-menthol was produced on a larger scale isolated by solid phase extraction. The structures of the glucosides were analyzed by N spectroscopy (Figure 4). The 1 H-NMR and 13 C-NMR signals of the products were assig by studies of the 1 H-1 H COSY, HMQC and HMBC spectra, as well as reference data f the literature [14,32,34,[42][43][44]. The 1 H-NMR spectra (400 MHz, MeOD) of the glucosyl racemic menthol showed the characteristic signals due to β-D-glucopyranosyl residu  Figure 4). These signals reflect the production of two diastereomers from cemic menthol by the conjunction with β-D-glucopyranose. Splitted signals were also served for all carbons. The signals for the anomeric 1' carbons showed almost the s intensity. Based on the NMR and literature data, the biotransformation products for by the UGT93Y1 whole-cell biocatalyst were established to be (±)-menthyl β-D-glucop nosides. Since the racemic acceptor substrate was added in excess, the results imply UGT93AY1 shows low stereospecificity for (±)-menthol ( Figure 4). Determination of the relative enzymatic activities with UDP Glo™ Glycosyltransferase assay (A) and UGT93Y1 enzyme kinetics for (±)-menthol (B). The relative enzyme activity was calculated after measuring the release of UDP during the glucosylation reaction. The highest enzymatic activity was set at 100%, and the relative enzyme activities calculated (A). The enzyme activity versus substrate concentration graph for UGT93Y1 from C. sinensis var. sinensis was determined using the UDP Glo™ Glycosyltransferase assay. The calculation was performed with KaleidaGraph (https://www.synergy.com/; v4.5.4; accessed on 8 September 2021).

Characterization of Menthyl Glucoside by NMR
The biotransformation product of (±)-menthol was produced on a larger scale and isolated by solid phase extraction. The structures of the glucosides were analyzed by NMR spectroscopy (Figure 4). The 1 H-NMR and 13 C-NMR signals of the products were assigned by studies of the 1 H-1 H COSY, HMQC and HMBC spectra, as well as reference data from the literature [14,32,34,[42][43][44]. The 1 H-NMR spectra (400 MHz, MeOD) of the glucosylated racemic menthol showed the characteristic signals due to β-D-glucopyranosyl residues of δ 4.38 ppm (d, J = 7.8 Hz, 1H) for the (−)-menthyl conjugate and of δ 4.34 ppm (d, J = 7.7 Hz, 1H) for the (+)-menthyl derivative. The 13 C-NMR spectra exhibited signals for the anomeric carbons at 101.31 and 105.65 ppm (C1') for (−)-and (+)-menthyl β-D-glucoside, respectively ( Figure 4). These signals reflect the production of two diastereomers from racemic menthol by the conjunction with β-D-glucopyranose. Splitted signals were also observed for all carbons. The signals for the anomeric 1' carbons showed almost the same intensity. Based on the NMR and literature data, the biotransformation products formed by the UGT93Y1 whole-cell biocatalyst were established to be (±)-menthyl β-D-glucopyranosides. Since the racemic acceptor substrate was added in excess, the results imply that UGT93AY1 shows low stereospecificity for (±)-menthol ( Figure 4).

Homology Modeling and Ligand Docking
To better understand the structural requirements of enzymatic glucosylation of racemic menthol, homology modelling was performed. Therefore, the amino acid sequence of UGT93Y1 was submitted to IntFOLD web server (https://www.reading.ac.uk/bioinf/ IntFOLD/; accessed on 8 September 2021). Among the top five 3D models, model Int-FOLD6HHpredTS5 was chosen due to its high GMQS of 0.6541 and confidence p-value of 2.293 × 10 −14 . Scores greater than 0.4 are characteristic for more complete and confident models; thus, the predicted model is very close to the native protein 3D structure [45]. With the exception of the first 10 amino acids at the N-terminus, no residue exceeded the disorder/order probability score of 0.5 (Supplementary material Figure S6), while the conserved regions in UGTs (catalytic His29, activating Asp134 and the PSPG box from Trp342 to Gln385) are located in highly ordered regions of the protein. Accordingly, these residues and regions also show low per-residue errors (Supplementary material Figure S7). UGT93Y1 is a typical GT1 family member; it adopts the GT-B fold, and it should follow an inverting mechanism ( Figure 5).

Homology Modeling and Ligand Docking
To better understand the structural requirements of enzymatic glucosylation of racemic menthol, homology modelling was performed. Therefore, the amino acid sequence of UGT93Y1 was submitted to IntFOLD web server (https://www.reading.ac.uk/bioinf/IntFOLD/; accessed on 9 September 2021). Among the top five 3D models, model IntFOLD6HHpredTS5 was chosen due to its high GMQS of 0.6541 and confidence p-value of 2.293 × 10 −14 . Scores greater than 0.4 are characteristic for more complete and confident models; thus, the predicted model is very close to the native protein 3D structure [45]. With the exception of the first 10 amino acids at the N-terminus, no residue exceeded the disorder/order probability score of 0.5 (Supplementary material Figure S6), while the conserved regions in UGTs (catalytic His29, activating Asp134 and the PSPG box from Trp342 to Gln385) are located in highly ordered regions of the protein. Accordingly, these residues and regions also show low per-residue errors (Supplementary material Figure S7). UGT93Y1 is a typical GT1 family member; it adopts the GT-B fold, and it should follow an inverting mechanism ( Figure 5). Ligand ((−)-menthol and (+)-menthol) docking was performed on the predicted 3D model of UGT93Y1 by using the Autodock Vina tool implemented in UCSF Chimera and the protein 3D structure 2vce (https://www.rcsb.org/; accessed on 8 September 2021) with the acceptor and donor ligands trichlorophenol and UDP-glucose, respectively, as template. The ligand (−)-menthol was installed in chair conformation with equatorial arranged substituents in a plane with trichlorophenol in the active site ( Figure 5). The bond lengths (2.7 and 3.6 Å) and angle (92 • ) of the arrangement His29/(−)-menthol-OH/C1-UDP-glucose correspond well to the lengths (1.8 and 3.8 Å) and angle (108 • ) of the catalytic conformation in the crystal structure of 2vce. In the case of the (+)-menthol enantiomer, the structure is rotated horizontally by 180 • (the methyl group now facing the catalytically active His29) and tilted a few degrees to the side. This extends the bond lengths (3.9 and 3.2 Å) and reduces the angle to 77 • and might explain the slight preference of UGT93Y1 for (−)-menthol. A closer look at the active site shows that it is lined with nonpolar amino acids (Leu101, Leu136, Pro157, Ile158, Phe164, Ile195 and Met199), which presumably interact polar (van der Waals and carbon hydrogen) and nonpolar (alkyl and Pi-alkyl) with terpene alcohol (Figure 5) similar to the diterpene steviol glucoside in UGT76G1 [46].

Discussion
In this study, we identified five UDP-glucose-dependent glycosyltransferases (UGT93Y1, UGT93Y2, UGT85K11, UGT72B27 and UGT73B24) by screening 57 members of a UGT whole-cell library that were capable of glucosylating (±)-menthol in vivo and in vitro (Supplementary material Figures S4 and 1). Two new plant biocatalysts from C. sinensis (UGT93Y1 and UGT93Y2) showed outstanding enzymatic activities compared to the other catalytically active proteins. These can be used for the development and improvement of biotechnological processes for the production of small molecule glucosides, which are widely used for industrial applications in pharmaceutical, food or cosmetics industries. Due to its minty smell, cooling and refreshing properties, menthol is widely applied as a flavor and fragrance additive in different industries [14]. Glycosylation using enzymes that have major benefits compared to chemical synthesis can increase the low water solubility of 435.5 mg L −1 at 25 • C [2] of this physiologically active compound, which impedes various applications [31]. Recombinant plant glycosyltransferases from Arabidopsis thaliana have already been described to glucosylate monoterpenoids such as geraniol, nerol, linalool and terpineol including (±)-menthol. [47]. However, only glycosyltransferases of the UGT73C clade (C1, C3, C5 and C6) were identified that could convert (±)-menthol. Here, we show that UGTs of the 93, 85 and 72 clades are also able to metabolize the monoterpenol in addition to UGT73B24. A number of UGT85 members have already been identified as monoterpene UGTs [48], while for the UGT72 group no enzyme has been detected so far that converts monoterpene alcohols. Thus, it is comprehensible that, in our study, the UGT72B27 protein showed the lowest activity. UGT93Y1 and UGT93Y2 are the new members of the 93 group and form a new subgroup of their own. UGT93Y1 shows an extraordinary menthol glycosylation activity, which is unusual, as the plant from which the corresponding UGT93Y1 gene was isolated from does not produce menthol. This result confirms once more that some UGTs involved in secondary plant metabolism show broad substrate tolerance [48], which is underpinned by the activity towards other menthol isomers such as (+)-isomenthol and (+)-neomenthol and the bicyclic (±)-fenchyl alcohol ( Figure 2).
In addition to its menthol glucosyltransferase activity, UGT93Y1 exhibits UDP-α-Dglucose glucohydrolase activity ( Figure 3A), which has already been demonstrated for a number of glycosyltransferases [49][50][51]. This means that the enzyme is able to transfer the glucose molecule from UDP-glucose to the acceptor water in the absence of other acceptor substrates. This results in the release of UDP from the sugar donor, which can be detected by the UDP Glo™ glycosyltransferase assay. Fortunately, this background reaction has been previously observed with many UGTs and can be exploited to determine sugar donor specificity [50]. However, the biological relevance is unknown.
Even though UGT93Y1 and UGT93Y2 show high sequence identity (74.3%), both enzymes differ significantly in their ability to metabolize (±)-menthol (Figure 1, Supplementary material Figures S4 and S5). Inspection of the amino acids lining the active site ( Figure 5) shows that three of them (I158T; I195V and M199F; Supplementary material Figure S2) are exchanged in the catalytic center of UGT93Y2 compared with UGT93Y1. While position 158 is located in a region of higher disorder (Supplementary material Figure S6) and higher structural inaccuracy (Supplementary material Figure S7), positions 195 and 199 are located in regions of lower disorder and higher prediction accuracy. These positions are likely to be essential for activity toward (±)-menthol and could be targeted by site-directed mutagenesis [52] or error-prone PCR [53] to find variants with increased catalytic efficiency.
UGT93Y1 shows non-Michaelis-Menten kinetics as the enzyme activity is inhibited at high substrate concentrations ( Figure 3B). Substrate inhibition is a well-known feature of human glucuronosyltransferases [54] and has also been described for plant glycosyltransferases [55]. The significance of such "atypical" kinetics in vivo is not yet understood but is observed for many enzymes [56]. In recent years, it has become apparent that the mechanisms of substrate inhibition are very complex. Multiple (at least two) binding sites within the enzyme, the formation of a ternary dead-end enzyme complex and/or ligand-induced changes in enzyme conformation, among others, are discussed as causes of substrate inhibition [54]. Substrate inhibition is undesirable in biotechnological processes, but studies have shown that inhibition of the catalytic activity of enzymes by an excess of substrate can be overcome by replacing individual amino acids in the protein [57].
According to the NMR and literature data, the products formed by UGT93Y1 during the whole-cell biotransformation were established to be (±)-menthyl β-D-glucopyranosides. Thus, UGT93Y1 isolated from C. sinensis does not stereospecifically glucosylate racemic menthol, which could be one reason for its high activity. Plants appear to be a rich source of promiscuous UGTs that not only glycosylate substrates of their native host but also non-native acceptor molecules. Similarly, Eucalyptus perriniana cells metabolize (−)-and (+)-menthol, a non-native metabolite of this plant, to the respective monoterpenyl 3-O-β-D-gentiobioside and 3-O-β-D-triglucoside (2,6-di-O-(β-D-glucopyranosyl)-β-D-glucopyranoside), respectively [34,58]. Moreover, in this plant, the first two glucose transfers appear to be nonspecific, whereas the third glucose unit is transferred only to the (+)-menthyl derivative.
In contrast to plant cell cultures, whole-cell biotransformation using E. coli cells represents a simple but highly efficient alternative for the production and purification of glucosides [59][60][61]. With minor improvements in cultivation parameters, an economical large-scale bioprocess can be developed, replacing expensive chemical synthesis that requires protection and deprotection reactions, resulting in low specificities and yields. Various terpene glycosides occur naturally and are obtained in small amounts from plant extracts [48,62,63]. Due to their high economic importance, glycosylation using enzyme biocatalysts becomes very attractive for the synthesis of glycosides and exhibits high efficiency, stereoselectivity and regioselectivity in an environmentally friendly manner.

Chemicals
All chemicals and reagents were purchased in analytical scale from Sigma Aldrich, Steinheim, Germany, unless otherwise mentioned.

RNA Isolation, cDNA Synthesis and Cloning of UGT93Y1 and UGT93Y2
Total RNA form tea leaves of C. sinensis var. sinensis were isolated according to the CTAB method [64] with minor modifications. The extraction buffer was prepared without spermidine. After precipitation overnight at 4 • C, the RNA pellet was solved in 600 µL RLT Plus Buffer of the RNeasy ® Plant Mini Kit from Qiagen. The following steps were performed according to manufacturer's instructions. The RNA concentration was determined using the CLARIOstar LVis plate (BMG Labtech, Ortenberg, Germany). The integrity of the RNA preparation was confirmed by agarose gel electrophoresis.
First strand cDNA synthesis using the M-MLV Reverse Transcriptase was performed according to the manufacturer's instructions (Promega, Mannheim, Germany).

Recombinant Protein Production and Protein Purification
Protein expression was performed using E. coli BL21(DE3)pLysS cells containing pGEX-4T-1 UGT93Y1 and pGEX-4T-1 UGT93Y2 according to [65], with minor modifications. After growing the pre-cultures overnight at 37 • C and 150 rpm in Luria-Bertani medium containing 100 µg/mL ampicillin and 34 µg/mL chloramphenicol, 1 L main culture and containing the corresponding antibiotics, it was inoculated with 10 mL of pre-culture and further incubated at 37 • C and 120 rpm until OD 600 reached 1 in a chicane flask. Gene expression was induced with 1 mM isopropyl-β-D-thiogalactopyranoside, and cultures were incubated over night at 18 • C at 150 rpm. Cells were harvested via centrifugation, and the pellets were stored at −80 • C.
Recombinant fusion proteins with an N-terminal GST tag were purified by Novagen ® GST Bind™ Resin following the manufacturer's instructions. The cells were resuspended in the binding buffer and disrupted by sonication. The crude protein extract was incubated over night at 4 • C with the resin to bind GST fusion protein. The recombinant protein was eluted with GST elution buffer containing reduced glutathione. The quality and quantity of the purified proteins was verified using SDS-PAGE, and the protein concentration was determined with Roti ® Nanoquant (Carl Roth, Karlsruhe, Germany).

Qualitative Enzyme Assay Using LC-MS Screening
For the initial substrate screening, the glycosyltransferase reaction was performed according to [39] with minor modifications. The enzyme reaction with a final volume of 100 µL containing 5 µg of purified recombinant protein in 100 mM TRIS-HCl buffer (pH 7.5), 1 mM UDP-glucose and 600 µM substrate dissolved in dimethyl sulphoxide (DMSO). The reaction was incubated at 30 • C, with constant shaking at 400 rpm overnight. LC-MS analysis was performed according to literature [39].

Quantitative UDP Glo™ Glycosyltransferase Screening
Positive substrates that were determined via LC-MS were further screened via UDP Glo™ Glycosyltransferase Assay (Promega, Mannheim, Germany) according to the manufacturer's instruction in order to determine the kinetic parameters of the UGTs. The enzymatic reaction was performed according to [66] with minor modifications. After starting the reaction by adding UDP-glucose, samples were incubated for 30 min at 30 • C and 400 rpm. The reaction was stopped by adding 12.5 µL 0.6 M HCl and further neutralization with 1 M TRIZMA base. Five µL of the GT reaction was pipetted to a 384 well plate [40]. The calculation of kinetic data was performed with KaleidaGraph (https://www.synergy.com/; v4.5.4; accessed on 8 September 2021). The corresponding genes were cloned into pGEX-4T-1 vector, and E. coli Waksman cells were transformed with the respective plasmid. The enzyme activity of various UGTs towards (±)-menthol was investigated in a small-scale biotransformation process according to [37]. The in vivo biotransformation begun by adding 0.5 mg/mL (±)-menthol at 18 • C. The samples were analyzed in triplicates, and the production of menthyl glucoside was investigated 6, 24 and 48 h after substrate addition. The culture supernatant was centrifuged twice (13.500 rpm, RT and 10 min). Menthyl glucoside formation was analyzed via LC-UV-MS. For relative quantification, signals for m/z 363 (M+HCOO-; formic acid adduct) were integrated. The highest value was set to 100%, and the relative biotransformation rate was calculated.

Production and Purification of Menthyl Glucoside from Large Scale Biotransformation
Menthyl glucoside was produced by whole-cell biotransformation on a large-scale in a 5 L fermenter batch according to [37]. In vivo biotransformation was induced by adding (±)-menthol (Sigma Aldrich, Steinheim, Germany) to the bacterial culture. Purification of the glucoside was performed as described in [37].

NMR Analysis
Thirty-five mg of purified glucoside was solved in 200 µL methanol-d 4 (Sigma-Aldrich, Steinheim, Germany), centrifuged at maximum speed for 2 min and evaporated. The residue was solved in 400 µL methanol-d 4 (Sigma-Aldrich, Steinheim, Germany), transferred to an NMR tube and filled up to a final volume of 600 µL. NMR spectra were recorded with a Bruker DRX 500 spectrometer (Bruker, Karlsruhe, Germany). The chemical shifts were referred to the solvent signal. The one-dimensional and two-dimensional COSY, HMQC and HMBC spectra were acquired and processed with standard Bruker software (XWIN-NMR) and MestreNova software (mestrelab.com; accessed on 8 September 2021).

Homology Modelling of UGT93Y1 and Molecular Docking
The protein sequence of UGT93Y1 was submitted to IntFOLD web server (https:// www.reading.ac.uk/bioinf/IntFOLD/; accessed on 8 September 2021) in order to develop a model with sufficient query sequence coverage and sequence identity [45]. The most reliable 3D structure was selected based on the Global Model Quality Score (GMQS 0.6541) and confidence p-value (2.293 × 10 −14 ). The GMQS ranged between 0 and 1, and scores greater than 0.4 generally indicate more complete and confident models, which are highly similar to the native structure. The consistency of the GMQS allows calculating a p-value, which quantifies the probability that a model is incorrect. The p-value = 2.293 × 10 −14 means that 1/2.293 E14 incorrect models will have higher scores. Templates used for homology modelling were 2vch_A, 2acv_A, 2c1x_A, 2pq6_A and 5tme_A (https://www. rcsb.org/; accessed on 8 September 2021). The docking of (−)-menthol and (+)-menthol in the active site of UGT93AY1 was performed with the Autodock/Vina tool [67] implemented in UCSF Chimera (https://www.cgl.ucsf.edu/chimera/download.html; accessed on 8 September 2021).

Supplementary Materials:
The following are available online. Figure S1: SDS-PAGE (A) and Western blot (B) using a primary monoclonal Anti-Glutathione-S-Transferase (GST) antibody produced in mouse (Sigma Aldrich) of recombinant UGT93Y1 (1) and UGT93Y2 (2) in purified protein elution fractions. Blue and green arrows show the UGTs and GST proteins, respectively. Marker proteins (M), Figure S2: Amino acid sequences of UGT93Y1 and UGT93Y2 from C. sinensis var. sinensis. The catalytically active His and the activating Asp are marked with red and blue arrows, respectively. The red box represents the conserved PSPG box, Figure S3: LC-MS analysis of (±)-menthyl glucoside Extracted ion chromatogram m/z 363 [M + HCOO] − (left panel). Mass spectrum in negative mode (-MS) and product ion mass spectrum (-MS2) of m/z 363 (right panels), Figure S4: Screening of a library of family 1 UGTs containing enzymes from diverse plant species with (±)-menthol by whole-cell biotransformation. The strain that produced the highest amount of the glucoside product was set to 100%, Figure S5: Glucosylation of (±)-menthol catalyzed by UGT93Y1 (A) and UGT93Y2