Article pubs.acs.org/JAFC
Glucosylation of Smoke-Derived Volatiles in Grapevine (Vitis vinifera) is Catalyzed by a Promiscuous Resveratrol/Guaiacol Glucosyltransferase Katja Har̈ tl,† Fong-Chin Huang,† Ashok P. Giri,†,‡ Katrin Franz-Oberdorf,† Johanna Frotscher,§ Yang Shao,† Thomas Hoffmann,† and Wilfried Schwab*,† †
Biotechnology of Natural Products, Technische Universität München, Liesel-Beckmann-Strasse 1, 85354 Freising, Germany Plant Molecular Biology Unit, Division of Biochemical Sciences, CSIR-National Chemical Laboratory, Pune 411 008 Maharashtra, India § Geisenheim University, Department of Grapevine Breeding, Von-Lade-Strasse 1, 65366 Geisenheim, Germany ‡
S Supporting Information *
ABSTRACT: Vinification of grapes (Vitis vinifera) exposed to forest fire smoke can yield unpalatable wine due to the presence of taint compounds from smoke and the release of smoke derived volatiles from their respective glycosides during the fermentation process or in-mouth during consumption. To identify glycosyltransferases (GTs) involved in the formation of glycosidically bound smoke-derived volatiles we performed gene expression analysis of candidate GTs in different grapevine tissues. Second, substrates derived from bushfire smoke or naturally occurring in grapes were screened with the candidate recombinant GTs. A resveratrol GT (UGT72B27) gene, highly expressed in grapevine leaves and berries was identified to be responsible for the production of the phenolic glucosides. UGT72B27 converted the stilbene trans-resveratrol mainly to the 3-Oglucoside. Kinetic analyses yielded specificity constants (kcat/KM) of 114, 17, 9, 8, and 2 mM−1 s−1 for guaiacol, trans-resveratrol, syringol, methylsyringol, and methylguaiacol, respectively. This knowledge will help to design strategies for managing the risk of producing smoke-affected wines. KEYWORDS: glucosyltransferase, guaiacol, resveratrol, smoke-derived volatiles, wine, smoke taint, Vitis vinifera
■
nasal “ash” perception.14 Meanwhile, in-mouth hydrolysis of monoglucosides and disaccharide glycosides of smoke-derived volatiles by microbial hydrolases and/or enzymes of human saliva has been demonstrated.15 In nature, the transfer of a glycosyl moiety from an activated sugar donor to an acceptor molecule is generally catalyzed by glycosyltransferases (GTs). GT sequences have been classified into more than 90 families (Carbohydrate-Active Enzyme database, CAZy, http://www.cazy.org/) through sequence homology comparison and on biochemical properties. GTs, which glycosylate small secondary metabolites and anthropogenic compounds belong to the GT1 family.16 In the CAZy collection, GT1 is one of the biggest GT families and reflects the diversity of their substrates such as molecules involved in interaction between plants and other organisms, and in protection against oxidative stress.17 Most plant GT1 members carry a conserved signature motif of 44−46 residues in their Cterminal amino acid sequences (plant secondary product GT box, PSPG box), which can be used to locate GT1 genes in different plant genomes.18 Although it has been suggested that volatile phenols found in smoke can be absorbed directly via the fruit cuticle and via the
INTRODUCTION Increasing frequencies of forest and bush fires in the proximity of grape (Vitis vinifera) growing areas have caused concern among winemakers.1 Wines produced from grapes affected by smoke in vineyards exhibit aromas such as “ash”, “burnt”, and “smoky” with “an excessively drying” back-palate and a retronasal “ash” perception, which results in low quality wines and reduced market value.2 Since guaiacol and 4-methylguaiacol are generally found in smoke from wood fire3,4 they are used as diagnostic markers for smoke exposed grapes (Figure 1).5−7 The primary mode of entry of volatile phenolic compounds into berries is suggested to take place directly via the fruit cuticle rather than through leaves.8 However, even when the level of guaiacol and 4-methylguaiacol was insignificant in grape juice the resulting wines showed smoke-related characteristics, pointing toward the occurrence of precursors of guaiacol in fruit affected by smoke.9,10 A screening for potential precursor molecules uncovered glycosidically bound metabolites of guaiacol and other phenolics upon grapevine smoke exposure.11,12 Analysis by stable isotope tracer technique demonstrated that the guaiacol conjugates were translocated between leaves and fruit, but only to a limited extent.12 Glycosylation of xenobiotic molecules is one of the main detoxification mechanisms in higher plants.13 Interestingly, some phenolic glucosides showed a flavor effect when tasted by a sensory panel in model wine indicating an in-mouth hydrolysis of the glucosides, which might explain the retro© 2017 American Chemical Society
Received: Revised: Accepted: Published: 5681
April 24, 2017 June 26, 2017 June 28, 2017 June 28, 2017 DOI: 10.1021/acs.jafc.7b01886 J. Agric. Food Chem. 2017, 65, 5681−5689
Article
Journal of Agricultural and Food Chemistry
Sigma-Aldrich (Steinheim, Germany), or Fluka (Steinheim, Germany) unless otherwise stated. RNA Extraction and cDNA Synthesis. Plant material (1 g) was ground to fine powder in liquid nitrogen using a mortar and pestle. Total RNA was extracted by the cetyltrimethylammonium bromidemethod19 adapted for the extraction of RNA from grape berries.20 DNA digest and cleanup were performed using the High Pure RNA Isolation kit (Roche, Mannheim, Germany). RNA integrity was confirmed by agarose gel electrophoresis. RNA concentration and the ratio of absorptions at 260 nm vs 280 nm were determined using the Nanodrop 1000 (Thermo Scientific, Dreieich, Germany). Complementary DNA was synthesized from 150 ng of total RNA using the iScript cDNA Synthesis system (Bio-Rad Laboratories, Hercules, U.S.A.) according to the manufacturer’s instructions. Expression Analysis. Expression of three UGT genes (UGT88F12, UGT72B27, and UGT92G6) and two reference genes VvActin and VvSAND (actin, GenBank Accession EC969944; SAND family protein, GenBank Accession XM_002285134) was quantified by quantitative real-time PCR in three biological and two technical replicates for each tissue (berry skin, inflorescence, leave, and root), variety and developmental stage. Primers were designed using PrimerBLAST (http://www.ncbi.nlm.nih.gov/tools/primer-blast/21), except those for VvActin and VvSAND, which were adopted as previously published.20,22 Primer sequences are given in Table S2. Quantitative real-time PCR was carried out using iQ SYBR Green Supermix (BioRad Laboratories) following the manufacturer’s instructions with one hundredth of the reaction consisting of cDNA and a final primer concentration of 250 nM in a final volume of 23 μL. Reactions were run on the iQ5 Real Time PCR Detection System (Bio-Rad Laboratories). Thermocycling conditions were 95 °C for 3 min followed by 40 cycles of 95 °C for 30 s, 55 °C (60 °C for UGT88F12) for 45 s and 72 °C for 30 s. The PCR efficiencies were determined employing a serial dilution series of a pool of all cDNAs prior to the relative quantification of single RNA samples. The PCR efficiency measurement yielded the linear range of the Ct-values. Values outside the linear range were designated as o.o.r. (out of range) samples. Relative normalized quantities were calculated from quantification cycle values using the qbasePLUS software (www.biogazelle.com) applying a modified ΔΔ-Ct method to take multiple reference genes and gene specific amplification efficiencies into account.23 Inter-run calibration was used to correct for run-to-run differences.23 Means and standard deviations were calculated. Cloning of GTs in Expression Vectors. Open reading frames (ORFs) were obtained from the publicly available V. vinifera genome.24,25 The UGT72B27 (GSVIVT01027064001, X M _00 228 088 7.4 ), U GT 8 8 F 1 2 (GSVIV T010 260 540 01, XM_002271522.4), and UGT92G6 (GSVIVT01031678001, XM_010652122.2) full-length ORFs were amplified from berry skin cDNA of V. vinifera cv. White Riesling 239−34 Gm. Specific primers were designed introducing the restriction sites XmaI and NotI for UGT88F12, SmaI and NotI for UGT72B27, and BamHI and NotI for UGT92G6 (Table S3). The genes were cloned into the pGEM-T easy vector (Promega, Madison, WI, U.S.A.), and subsequently subcloned into the pGEX-4T-1 expression vector (Amersham Bioscience, Freiburg, Germany). The sequence identities were confirmed by sequencing (Eurofins MWG Operon, Ebersberg, Germany). Expression of Heterologous Proteins. For expression of the recombinant proteins, the vector constructs were transformed into the E. coli strain BL21 (DE3) pLysS (Novagen, Schwalbach, Germany). The cells were grown at 37 °C, 5000g in 1 L of LB medium supplemented with ampicillin (100 μg mL−1) and chloramphenicol (23 μg mL−1) until OD600 reached 0.5 to 0.7 (exponential phase). To induce protein expression, 1 mM of isopropyl-β-D-thiogalactopyranoside (IPTG) was added, and the cells were cultivated further at 16−18 °C overnight. Subsequently, the cells were harvested by centrifugation (10 min, 5000g) and stored at −80 °C. Purification of Recombinant GTs. Cells were redissolved in binding buffer containing 10 μM of proteinase inhibitor phenylmethylsulfonyl fluoride (PMSF), and disrupted by sonication. The GST-fusion proteins were purified with a GST Bind resin (Novagen)
Figure 1. Smoke-derived volatile phenols and natural constituents of grapevine.
leaves epidermis the prevailing pathway and the location of glycosylation within the vine remain unknown.7 As the ratio of free phenols to their glycosidically bound metabolites can vary distinctly, it is hypothesized that the glycosylation step might be subject to unknown regulation and potentially could be manipulated by vineyard management. In order to devise strategies for managing and reducing the risk of producing smoke-affected wines, we screened for GTs highly expressed in leaves and berries of V. vinifera. Further, we performed gene expression analyses and biochemically characterized the recombinant candidate GTs for their substrate specificity. A resveratrol GT efficiently catalyzes the glucosylation of phenols, predominantly found in smoke from wood fire.
■
MATERIALS AND METHODS
Plant Material and Chemicals. All plant material was obtained from the Institute of Grapevine Breeding of Geisenheim University (Geisenheim, Germany). Field grown vines were planted in 2002 in deep sandy loam in a north−south row orientation and a vine spacing of 1.8 × 1.3 m2. All field grown vines were nonirrigated and grafted onto the rootstock variety Börner due to presence of phylloxera. Therefore, roots from ungrafted plants of grapevine (V. vinifera) cv. Gewurztraminer 11−18 Gm (Geisenheim) were collected in a phylloxera-free glasshouse experiment from cuttings. The roots were sampled from three individual plants. In the field, samples were collected from a block of 60 vines of the same clone during vintage 2011. Berries were sampled at six dates (6 to 17 weeks post flowering), leaves at three dates (approximately ages of 1, 3, and 5 weeks) and inflorescences at three dates (developmental stages 55, 57, and 65 according to the BBCH-scale for grapevine16). At each sampling time, three biological replicates were collected consisting of ten inflorescences, berries, or leaves sampled at random from a plot of 20 vines each. Additionally, berries of cv. White Riesling 239−34 Gm were sampled at three dates (9, 13, and 17 weeks post flowering) for cloning of the UGTs. All berry samples were peeled in the vineyard and berry skins were immediately frozen in liquid nitrogen. Subsequently, only skins were used for further analysis. For maturity measurements of berries see Table S1 of the Supporting Information (SI). Chemicals were purchased from Roth (Karlsruhe, Germany), 5682
DOI: 10.1021/acs.jafc.7b01886 J. Agric. Food Chem. 2017, 65, 5681−5689
Article
Journal of Agricultural and Food Chemistry while following the manufacturer’s instructions. After 2 h the proteins were eluted and quantified by the method of Bradford.26 The presence of the recombinant proteins was verified by SDS-PAGE and Western blot. Radioenzymatic Assay for Substrate Screening. The enzyme activity was determined as previously described.27 In short, each reaction (200 μL) consisted of Tris-HCl buffer (100 mM, pH 8), 500 μM substrate and a mixture of unlabeled and 14C-labeled UDP-glucose (24.95 nmol unlabeled, 33 pmol labeled equals 0.01 μCi UDP[14C]glucose). The aglycon screens were performed under optimized conditions that were determined for each enzyme, respectively (UGT88F12:37 °C, 4 μg of recombinant protein; UGT72B27:21 °C, 5 μg of protein; UGT92G6:30 °C, 10 μg protein). Finally, all samples were incubated for 24 h, and 1 mL of water-saturated 1butanol was added to stop the reaction. The organic phase (800 μL) was mixed with 2 mL Pro Flow P+ cocktail (Meridian Biotechnologies Ltd., Epsom, U.K.), and the radioactivity was determined by liquid scintillation counting (LSC, Tri-Carb 2800TR, PerkinElmer, Waltham MA, U.S.A.). Product Identification Using LC−MS. The standard assay (200 μL) contained 50 μL of crude protein extract, 100 mM Tris-HCl buffer (pH 7.5), 500 μM UDP-glucose, and 600 μM substrate. The reaction was initiated by the addition of UDP-glucose, incubated at 30 °C with constant shaking for 60 min, and was stopped by heating at 75 °C for 10 min. The product mixtures were centrifuged at 10 000g for 5 min twice to remove the protein residue and analyzed by LC−MS. Products were monitored using diagnostic ions listed in Table S4. The LC system consisted of a quaternary pump and a variable wavelength detector, all from Agilent 1100 (Bruker Daltonics, Bremen, Germany). The column was a LUNA C18 100A 150 × 2 mm2 (Phenomenex, Aschaffenburg, Germany). LC was performed with the following binary gradient system: solvent A, water with 0.1% formic acid and solvent B, 100% methanol with 0.1% formic acid. The gradient program used was as follows: 0−7 min, 90% A/10% B to 50% A/50% B; 7−10 min, 50% A/50% B to 100% B, held for 5 min; 15−20 min, 100% B to 90% A/10% B, held for 10 min. The flow rate was 0.2 mL/min. Attached to the LC was a Bruker esquire 3000plus mass spectrometer with an ESI interface that was used to record the mass spectra. The ionization voltage of the capillary was 4000 V, and the end plate was set to −500 V. Determination of Kinetic Parameters Using a pH-Sensitive Colorimetric Assay. A calibration curve was established as previously described.28 Briefly, in a 2 mM sodium phosphate buffer (pH 8, 120 μL) containing 0.01 mM phenol red, 0.1 mM MnCl2, 10 mM substrate, and 5 μg purified UGT72B27, different amounts of 10 mM hydrochloric acid were added to final concentrations of 0.1, 0.2, 0.3, 0.4, 0.5, and 0.6 mM, and the OD557 was recorded. A quantitative linear relationship between proton concentration and absorbance was established. For the determination of kinetic parameters, 0.01 mM phenol red, 0.1 mM MnCl2, 10 mM substrate, and 5 μg purified UGT72B27 were mixed with phosphate buffer (2 mM, pH 8). The reaction was started by adding UDP-glucose to a final concentration of 2 mM, and the absorbance at 557 nm was recorded for each sample at 10 s intervals for a total of 3 min. The total volume was 120 μL. The activities of the enzymes were calculated from the calibration curve. All measurements were done in triplicates and the values were averaged. Finally, KM and vmax were calculated by nonlinear regression of the Michaelis−Menten equation using the Microsoft Excel Solver.
(GSVIVT01027064001, XM_002280887.4), and UGT92G6 (GSVIVT01031678001) due to their high transcript levels in leaves and berries. Next we analyzed the expression profiles in more detail in V. vinifera cv. Gewurztraminer berries, leaves, inflorescences, and roots to survey transcript abundance during plant and fruit development (Figure S1). UGT88F12 transcript levels were high in unripe berries (6 weeks post flowering), decreased during ripening, and reached a second peak in ripe fruit (17 weeks post flowering). In developing flowers and leaves (I1−I2 and L1-L3) the expression was highest, reaching values that exceeded the upper limit of the applied standard curve (out of range, oor). Only in flowers at full bloom UGT88F12 mRNA abundance was lower. In berry tissues, UGT72B27 expression was highest in unripe fruit and decreased toward the ripe stage. Expression levels in inflorescences were comparable to unripe berries. In inflorescences (I1−I2) the levels slightly increased and dropped again at full bloom (I3). In contrast, UGT72B27 transcript levels increased during leaf development peaking in five-weekold tissues (L3). UGT92G6 expression increased from unripe to medium ripe berry (B6−B11) then dropped to show moderate and constant levels in later stages (B13−B17). UGT92G6 expression in developing flowers first decreased but then increased again slightly at full bloom. During leaf development, UGT92G6 transcripts decreased. Overall, expression of the three putative GTs was low in root tissue, and exhibited distinct changes during berry, flower, and leaf development. Substrate Specificity of GTs toward Various Naturally Occurring Grapevine Leaf and Fruit Secondary Metabolites. The substrate specificity of UGT88F12, UGT72B27, and UGT92G6 was assessed toward a set of 17 naturally occurring grapevine metabolites30−34 and structurally related compounds comprising anthocyanidins (malvidin and pelargonidin), flavonols (quercetin and kaempferol), monoterpenes (thymol, β-citronellol, carvacrol, geraniol, linalool, and farnesol), monoalcohols (octanol, decanol, hexanol, and 2phenylethanol), the phenylpropene eugenol, the phenolic aldehyde vanillin, and the furanone furaneol (Table 1). Vanillin was found to be the preferred substrate of UGT88F12. In addition, the flavonols quercetin and kaempferol were glucosylated with a comparably high catalytic activity. Moderate activity was detected toward the monoterpene carvacrol and the phenylpropene eugenol. The UGT72B27 converted thymol and carvacrol most efficiently with a relative enzymatic activity of 100 and 82%, respectively. Furthermore, it showed moderate affinity toward vanillin and eugenol. In contrast to UGT88F12, the flavonol substrates quercetin and kaempferol were poor substrates. Clearly, UGT92G6 preferred the monoterpene carvacrol above all other substrates. However, moderate affinity could be asserted for the flavonols. The overall activity toward monoalcohols, anthocyanidins, and furaneol was low. UGT72B27 Converts Smoke Derived Phenolics to the Corresponding Glucose Conjugates. As the three UGTs preferred phenols as substrates, the smoke-derived phenolic xenobiotics guaiacol, 4-methylguaiacol, syringol, 4-methylsyringol, m-cresol, and o-cresol (Figure 1) were tested employing LC−MS analysis (Figure 2). Not only was activity determined against these substrates, but also the identity of the glucoside products was verified. UGT72B27 converted all the tested smoke-derived phenolics to the corresponding glucoseconjugates with high efficiency. 4-Methylguaiacol, m-cresol, ocresol, and guaiacol were almost completely converted to their
■
RESULTS Differential Expression of Candidate GTs in above Ground Tissues Suggests Their Potential Role in Retaining and Controlling the Release of Smoky Volatiles. In order to find enzymes that mediate the glucosylation of smoke-derived phenolic compounds, we screened our collection of highly expressed V. vinifera GT sequences22,29 and selected three candidate genes UGT88F12 (GSVIVT01026054001, XM_002271522.4), UGT72B27 5683
DOI: 10.1021/acs.jafc.7b01886 J. Agric. Food Chem. 2017, 65, 5681−5689
Article
Journal of Agricultural and Food Chemistry
grapevine hydroxystilbene with health-promoting properties35,36 was tested as a potential substrate by LC−MS analysis. UGT72B27 showed high activity for the stilbene resulting in the formation of two isomeric products, namely transresveratrol-3-O-glucoside (piceid) and trans-resveratrol-4′-Oglucoside (resveratroloside; Figure 3A). The identity of both glucosides was confirmed by tandem mass spectrometry due to a neutral loss of 162 Da indicative of a glucose moiety and reference data.37,38 UGT88F12 displayed only minor glucosylation activity of trans-resveratrol, as it yielded low levels of only one product corresponding to trans-resveratrol-3-O-glucoside. UGT92G6 did not show any activity on trans-resveratrol. Kinetic Properties of Recombinant UGT72B27 Suggest 6-Resveratrol and Biosimilars as the Natural Substrates in Grapevine Tissues. Building upon the assessed substrate specificity, biochemical characterization of recombinant UGT72B27, the most active of the three candidate GT enzymes, was performed to identify the affinity of the enzyme, the conversion rates and the preferred substrates (Table 2). Optimal conditions (pH 8 and 5 μg of recombinant protein) were employed, together with a selection of compounds including smoke-derived xenobiotics (guaiacol, syringol, 4-methylguaiacol, 4-methylsyringol, o-cresol, and mcresol). Guaiacol and the putative natural substrate transresveratrol were efficiently converted to their respective glucosides with specificity constants (kcat/KM) of 114.41 and 16.97 mM−1 s−1, respectively owing to low KM values of 11.24 and 36.15 μM. Additionally, thymol was efficiently converted (kcat/KM of 13.52, KM of 52.79 mM−1 s−1). In contrast, the structurally related metabolites 4-methylsyringol, phenol, and Furaneol showed specificity constants of 1.93, 1.76, and 0.99 mM−1 s−1, respectively (Table 2). Consequently, it can be concluded that trans-resveratrol might act as a natural UGT72B27 substrate.
Table 1. Relative Enzymatic Activities of UGT88F12, UGT72B27, and UGT92G6 Towards a Set of Aglycon Substratesa substrates
UGT88F12
UGT72B27
UGT92G6
malvidin pelargonidin quercetin kaempferol thymol β-citronellol carvacrol geraniol linalool farnesol octanol decanol hexanol 2-phenylethanol eugenol vanillin furaneol
11 0 80 73 12 4 45 6 6 5 6 6 5 4 51 100 7
3 0 10 11 100 6 82 6 4 6 5 3 6 6 65 50 15
0 1 51 43 10 2 100 11 4 2 5 3 10 7 22 15 4
a
The relative enzymatic activities [%] were determined by radiochemical analysis with UDP-[14C]Glc. They refer to the highest level of extractable product measured for every enzyme (UGT88F12:100% = 0.25 nkat mg −1 ; UGT72B27:100% = 0.7 nkat mg −1 ; UGT92G6:100% = 0.3 nkat mg−1). Empty vector control was always
