Identification and characterization of DcUCGalT1, a galactosyltransferase responsible for anthocyanin galactosylation in purple carrot (Daucus carota L.) taproots

Purple carrots (Daucus carota ssp. sativus var. atrorubens Alef.) accumulate large amounts of cyanidin-based anthocyanins in their taproots. Cyanidin can be glycosylated with galactose, xylose, and glucose in sequence by glycosyltransferases resulting in cyanidin 3-xylosyl (glucosyl) galactosides in purple carrots. The first step in the glycosylation of cyanidin is catalysis by UDP-galactose: cyanidin galactosyltransferase (UCGalT) transferring the galactosyl moiety from UDP-galactose to cyanidin. In the present study, a gene from ‘Deep purple’ carrot, DcUCGalT1, was cloned and heterologously expressed in E. coli BL21 (DE3). The recombinant DcUCGalT1 galactosylated cyanidin to produce cyanidin-3-O-galactoside and showed optimal activity for cyanidin at 30 °C and pH 8.6. It showed lower galactosylation activity for peonidin, pelargonidin, kaempferol and quercetin. It accepted only UDP-galactose as a glycosyl donor when cyanidin was used as an aglycone. The expression level of DcUCGalT1 was positively correlated with anthocyanin biosynthesis in carrots. The enzyme extractions from ‘Deep purple’ exhibited galactosylation activity for cyanidin, peonidin and pelargonidin, while those from ‘Kuroda’ (a non-purple cultivar) did not.

In the present study, we cloned a gene of UCGalT from purple carrot (DcUCGalT1) and determined its ability to galactosylate cyanidin, peonidin, pelargonidin, kaempferol and quercetin. We also investigated the relationship between the DcUCGalT1 expression levels and anthocyanin accumulation to understand the importance of the role of DcUCGalT1 in the biosynthesis of anthocyanins in purple carrot taproots.

Results
Nucleotide sequence and deduced amino acid sequence of DcUCGalT1. DcUCGalT1  The deduced amino acid sequence of DcUCGalT1 was compared with that of two anthocyanidin galactosyltransferases from Aralia cordata and Actinidia chinensis, two flavonoid galactosyltransferases from Petunia hybrida and Vigna mungo as well as anthocyanidin glucosyltransferase from Vitis vinifera (VvGT1) using the ESPript3.0 website 25 . Among these UGTs, the amino acid sequence of DcUCGalT1 showed the highest identity with the UCGalT from A. cordata (Fig. 2). Like other UGTs, the UDP-sugar binding PSPG (Putative Secondary Plant Glycosyltransferase) motif also exists in the C-terminal domain of DcUCGalT1 (Residues Trp332-His375). Histidine is specifically conserved as the last amino acid residue of the PSPG motif in galactosyltransferases 23 . However, the last amino acid residue of the PSPG motif in glucosyltransferase, such as VvGT1, is always a glutamine (Fig. 2). Like the other four anthocyanindin galactosyltransferases, the last amino acid of the deduced PSPG motif in the DcUCGalT1 is also histidine. Thus DcUCGalT1 is deduced to be a cyanidin -3-O-galactosyltransferase.

Expression of DcUCGalT1 in E. coli and purification of recombinant DcUCGalT1. DcUCGalT1
was cloned into the pET30 vector, which resulted in addition of His-tag, thrombin, S-tag, and enterokinase sequences to the N-terminus of the protein. The exact sequences of recombinant DcUCGalT1 (rDcUCGalT1) actually being expressed in E. coli are shown in Supplementary Fig. S1. The rDcUCGalT1 was successfully expressed in E. coli BL21(DE3) in the presence of 1.0 mM IPTG for over 12 h at 18 °C. The molecular mass of rDcUCGalT1 was calculated as 54.80 kDa, while SDS-PAGE analysis showed a prominent band of the purified enzyme at about 55 kDa (Fig. 3, white arrow).
UFGT activity of rDcUCGalT1. The glycosylation ability of rDcUCGalT1 towards cyanidin and four other flavonoid substrates was determined. The rDcUCGalT1 glycosylated cyanidin to form a new product by using UDP-galactose as the glycosyl donor (Fig. 4A,B). The new product showed the same retention time as cyanidin 3-O-galactoside standard (Fig. 4D) and showed an ion at m/z 447.1 in the LC-MS spectral data (Fig. 4C), thus confirming that the new product was cyanidin 3-O-galactoside.
The rDcUCGalT1 also conjugated peonidin, pelargonidin, kaempferol and quercetin with UDP-galactose as the glycosyl donor (See Supplementary Fig. S2). However, rDcUCGalT1 showed no glycosylation activity towards cyanidin with UDP-glucose or UDP-xylose as glycosyl donors ( Table 1). The activity of the enzyme for cyanidin was the highest among all the acceptor substrates tested (Table 1). In addition, rDcUCGalT1 accepted only UDP-galactose as a glycosyl donor among the UDP sugars tested in the present study.
Optimal temperature and pH for rDcUCGalT1 activity towards cyanidin. rDcUCGalT1 activity towards cyanidin was determined at temperatures from 10-60 °C with maximum activity being observed at 30 °C (Fig. 5A). rDcUCGalT1 activity was barely detectable when the temperature reached 60 °C. The enzyme activity was also tested over the pH range of 6.8-10.4. rDcUCGalT1 showed maximum activity at pH 8.6 in the pH range tested (Fig. 5B). With an increase or decrease from pH 8.6, rDcUCGalT1 activity gradually decreased, becoming barely detectable when the pH reached 10.4. Glycosylation of anthocyanins with crude enzyme from taproots of carrots. The crude enzyme was extracted from the taproots of 'Deep purple' and 'Kuroda' carrot cultivars (See Supplementary Fig. S3). The glycosylation activity of the crude enzyme was determined using reverse phase HPLC. The crude enzyme from 'Kuroda' did not conjugate cyanidin with UDP-galactose as the glycosyl donor (Fig. 4E). However, that from 'Deep purple' glycosylated cyanidin with the same donor substrate (Fig. 4F). The crude enzyme from 'Deep purple' also showed glycosylation activity towards peonidin and pelargonidin but the crude enzyme from 'Kuroda' did not (See Supplementary Fig. S4).
Ultraviolet (UV) light irradiation significantly enchanced carrot anthocyanin accumulation in previous reports 17,27 . The 'Kuroda' carrot taproots don't accumulate anthocyanin at any time when grown in soil. For 'Deep purple' carrot grown in soil, the taproots don't accumulate any anthocyanin in the first two weeks, although they accumulate rich anthocyanins when they are mature. However roots of both cultivars did accumulate   anthocyanins when exposed to UV light. In this study, the Petri dishes planted with carrots seeds were placed vertically and the carrot seedlings were grown along the surface of the solid medium so that UV-light could stimulate the entire roots of carrots. The roots of 12-day-old seedlings of 'Kuroda' and 'Deep purple' both turned purple after UV-light irradiation, suggesting that UV-light induced anthocyanin accumulation in carrot roots (Fig. 6B). With UV-light irradiation, the roots of 12-day-old seedlings of 'Deep purple' showed deeper purple color than that of 'Kuroda' . However, the roots of 12-day-old seedlings of 'Kuroda' and 'Deep purple' were white when grown in the dark, which suggests that anthocyanins hadn't accumulated in the roots (Fig. 6B). The expression profiles of DcUCGalT1 in the roots of 12-day-old seedlings of 'Kuroda' and 'Deep purple' were also determined. The abundance of DcUCGalT1 transcript was much higher in the roots of 12-day-old seedlings of 'Kuroda' and 'Deep purple' with UV irradiation than that of 'Kuroda' and 'Deep purple' grown in the dark (Fig. 6C). With UV-light irradiation, the expression levels of DcUCGalT1 in the roots of 12-day-old seedlings of 'Deep purple' were higher than that of 'Kuroda' .

Discussion
Purple carrot cultivars accumulate large amounts of anthocyanins in their taproots. The predominant anthocyanins in purple carrot taproots are derived from cyanidin 18 . UCGalT was thought to catalyze the first step in the glycosylation of cyanidin, which transferred the galactosyl moiety from UDP-galactose to the 3-O-position of cyanidin 17 . In the present study, a cDNA clone, namely DcUCGalT1, was obtained and sequenced from a purple  carrot cultivar (Deep purple). Of those analyzed, the identity of the deduced amino acid sequence of DcUCGalT1 was closest to that of UCGalT from A. cordata, which may be because D. carota and A. cordata both belong to the Apiales order. The molecular mass of glycosyltransferases generally ranges from 45 to 60 kDa 23 . That of DcUCGalT1 was calculated to be 49.37 kDa, a little lower than a native UCGalT (52 kDa) from D. carota which was determined using a Sephadex G-75 column 20 . The rDcUCGalT1 activity was monitored and confirmed by measuring the formation of cyanidin-3-O-galactoside using HPLC and LC-MS analyses. These data suggest that rDcUCGalT1 transferred a galactose from UDP-galactose to the 3-OH of cyanidin. The optimal galactosyltransferase activity of rDcUCGalT1 for cyanidin was obtained at 30 °C and at pH 8.6, higher than the optimum temperature of 23 °C and pH 6.8 previously reported for a native UCGalT from D. carota 20 , which may be caused by the addition of a His-tag, thrombin, S-tag, and enterokinase sequences to the N-terminus of the rDcUCGalT1protein. The optimal pH for rDcUCGalT1 activity was slightly alkaline, similar to other UFGTs 28,29 .
Peonidin-and pelargonidin-derived anthocyanins have previously been found in purple carrots 18 . In the present study, rDcUCGalT1 also glycosylated peonidin and pelargonidin. The activity of rDcUCGalT1 for cyanidin was highest among all three endogenous anthocyanins in purple carrots, which may partly explain why purple carrots accumulated much more cyanidin-based anthocyanins than those based on peonidin and pelargonidin. In addition, rDcUCGalT1 accepted only UDP-galactose as a glycosyl donor in the present study.
It has been reported that chalcone synthase 1, chalcone-flavonone isomerase 1, flavanone 3-hydroxylase 1, flavonoid 3′-monooxygenase 1, dihydroflavonol 4-reductase 1, and leucoanthocyanidin dioxygenase 1/leucoanthocyanidin dioxygenase 2 were highly expressed in purple carrot taproots but not or scarcely expressed in non-purple carrot taproots at the 60-day-old stage 26 . The expression of these genes appeared to be the determining steps in the production of anthocyanins in purple carrots 26 . In the present study, using qRT-PCR, we have described the expression patterns of DcUCGalT1 in the taproots of purple and non-purple carrot cultivars at 60 days. We have found that DcUCGalT1 was highly expressed in the taproots of all three 60-day-old purple carrot cultivars. However, the DcUCGalT1 transcript was not or scarcely detectable in the 60-day-old taproots of the six non-purple carrot cultivars, showing a similar pattern to that described in grapevines 30 and kiwifruit 24 . Anthocyanin content increased in 12-day-old roots of 'Deep purple' and 'Kuroda' after UV irradiation. Similar to UCGalT from D. carota reported previously 20 , the expression levels of DcUCGalT1 also increased both in 'Deep purple ' and 'Kuroda' after UV irradiation. These results suggest that DcUCGalT1 was positively correlated with anthocyanin biosynthesis in carrot taproots. The crude enzyme extractions from 'Deep purple' showed galactosylation activity predominantly for cyanidin, as well as peonidin and pelargonidin, while those from 'Kuroda' did not. In purple carrots, galactosyltransferases catalyze the first step in the glycosylation of many different kinds of the anthocyandins 2 . The glycosylation step is critical for stability and water solubility of anthocyanidins, producing the first stable coloured compounds of the anthocyanin biosynthetic pathway 24,31 . Loss of activity of UDP-flavonoid 3-O-galactosyltransferase leads to no accumulation of anthocyanin in kiwifruit 24 .
In some other plant species, transcription factors, especially MYBs, were reported to control the transcription of structural genes for anthocyanin biosynthesis, including UFGTs 32-34 . That DcUCGalT1 was not or was scarcely expressed in non-purple carrots is possibly caused by the lack of activation by transcription factors. In future work, we will focus on indentification of transcription factors which control expression of DcUCGalT1.

Methods
RNA extraction and cDNA synthesis. The total RNA was extracted from the carrots using an RNA Simple Total RNA Kit (Tiangen, Beijing, China) following the manufacturer's instructions. cDNA was synthesized as described previously 26 and was diluted 20-fold for gene cloning and qRT-PCR analysis.

DcUCGalT1 cDNA expression in Escherichia coli. By using the ClonExpress II One
Step Cloning Kit (Vazyme Biotech Co. Ltd., Nanjing, China), the coding sequence of DcUCGalT1 was subcloned into a pET-30a(+ ) expression vector (Novagen, Darmstadt, Germany) between the BamHI/SacI sites using a forward primer: 5′ -GCCATGGCTGATATCGGATCCATGGGGAGTACAAATCTGGAAC-3′ and a reverse primer: 5′ -ACGGAGCTCGAATTCGGATCCTCAGACAGCAATCACTTTTACTAGC-3′ , followed by transformation into E. coli DH5α . The authenticity of the resulting clones was confirmed by PCR analysis and sequencing. The recombinant plasmid pET-30a (+ )-DcUCGalT1 was extracted from the positive clones and then transformed into E. coli BL21(DE3) cells (TransGen, Beijing, China). The BL21(DE3) cells containing recombinant plasmid were grown in 50 mL LB medium containing 50 mg L −1 kanamyacin at 37 °C for 4-5 h while being shaken at 230 rpm until an OD 600 of 0.4-0.6 was attained. The recombinant protein was induced by adding isopropyl β -D-1-thiogalactopyranoside (IPTG) to a final concentration of 1.0 mM to the culture and adjusting the temperature to 18 °C. The culture was incubated for at least 12 h.
Recombinant protein purification. Cells were harvested by centrifugation, re-suspended in 2 mL lysis buffer (pH = 7.5) containing 50 mM NaH 2 PO 4 , 300 mM NaCl, 10% glycerol, 10 mM β -mercaptoethanol and 10 mM imidazole. The cell suspension was then sonicated for 20 min on ice. The cell debris in the sonicated mixture was removed by centrifugation at 12,000 × g for 10 min at 4 °C then filtered using 0.22-μ m microfiltration membranes. The soluble mixture was passed through a column containing Ni-NTA-agarose resin (1.5 mL bed volume) (Qiagen, Hilden, Germany). The resin was washed with 1 mL wash buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10% glycerol, 10 mM β -mercaptoethanol and 50 mM imidazole, pH = 7.5) six times. The 6 × His-tagged recombinant protein was then eluted from the resin with 2 mL elution buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10% glycerol, 10 mM β -mercaptoethanol and 250 mM imidazole, pH = 7.5). The recombinant protein fraction was desalted against a buffer (50 mM NaH 2 PO 4 -Na 2 HPO 4 , pH = 7.5, 0.1 mM EDTA, 1 mM DTT) using a HiTrap desalting column (GE Healthcare Life Sciences, Beijing, China) according to the manufacturer's instructions. Then a protease inhibitor cocktail (Product Number: P8849; Sigma, St. Louis, USA) was added to the purified protein fraction (10 μ L/1 mL). The purified protein concentration in the fractions was quantified using a Bradford assay kit with bovine serum albumin (BSA) as standard and analyzed using 12% SDS-PAGE with Coomassie staining. rDcUCGalT1 activity assay. The rDcUCGalT1 activity was assayed in 50 mM NaH 2 PO 4 -Na 2 HPO 4 (pH = 7.5), 1 mM UDP-galactose,-glucose, or -xylose, 0.2 mM acceptor substrates, cyanidin, peonidin, pelargonidin, kaempferol and quercetin (Sigma), and about 0.1-0.5 μ g of purified rDcUCGalT1 in a final volume of 50 μ L. About 0.1-0.5 μ g of the total soluble protein from BL21(DE3) cells containing the DcUCGalT1 plasmid without induction was used as the control. The reaction was initiated by adding various flavonoids after the mixture was incubated at 30 °C for 3 min. After further incubation at 30 °C for 5 min, the reaction was terminated by adding 10 μ L 12 M HCl, vortexed for 60 s then centrifuged at 12,000 × g for 10 min to precipitate proteins. Twenty μ L of the supernatant was analyzed by HPLC to detect and quantify the glycosylated products. A 1200 series HPLC (Agilent Technologies, Palo Alto, USA) with an Inertsil OPD -SP column (C18, 4.6 mm × 250 mm, particle size 5 μ m, GL Sciences, Inc., USA) was used for separating the reactants and the products. Cyanidin, peonidin and pelargonidin were separated from their glycosides using a linear gradient from 20% solvent A (acetonitrile) and 80% solvent B (0.5% acetic acid in water) to 40% of A and 60% of B in 10 min at 1.0 mL/min, using diode array detection at 530 nm. The conditions used for separating kaempferol and quercetin from their glycosides were as follows: a linear gradient of 20% solvent A and 80% B to 90% of A and 10% of B in 10 min at 1.0 mL/min, 360 nm.
Liquid chromatography-mass spectrometry (LC-MS) was used to determine the glycoside product of cyanidin using 20 μ L of the reaction mixture. The conditions used for elution were as follows: a linear gradient of 20% solvent A and 80% solvent B to 40% of A and 60% of B in 20 min at 0.5 mL/min, 530 nm. The peak corresponding to the product was subjected to negative ion mass spectrometry analysis (MS) with the same other conditions as described previously 36 . Effect of pH and temperature on rUCGalT1 activity. The optimum pH for enzyme activity was determined in 50 mM NaH 2 PO 4 -Na 2 HPO 4 (pH = 6.8-8.0) and 50 mM glycine-NaOH (pH = 8.6-10.4) at 40 °C for 5 min. The effect of temperature (10-60 °C) on enzyme activity was determined at pH 8.6 in 50 mM glycine-NaOH for 3 min.
Kinetic analyses of rUCGalT1 towards anthocyanidins. The initial velocity versus substrate concentration curves for the rUCGalT1 reaction were determined using a fixed anthocyanidin (cyanidin, peonidin, and pelargonidin) concentration of 0.2 mM and varying the concentration of UDP-galactose from 0.25-4 mM, or using a fixed UDP-galactose concentration of 2 mM and varying the concentration of anthocyanidins (0.025-0.2 mM for cyanidin, 0.02-0.4 mM for peonidin and 0.025-0.2 mM for pelargonidin). The reactions took place in 50 mM glycine-NaOH (pH = 8.6) at 30 °C. The apparent K m and V max values were estimated using Hanes-Woolf plots.
Extraction and activity assay of enzyme from the taproots of 'Deep purple' and 'Kuroda'. The 60-day old taproots of 'Deep purple' and 'Kuroda' were grown under the same conditions as previously described 26 and were frozen in liquid nitrogen then ground to a fine powder. The powdered tissue (about 0.2 g) was extracted in 1 mL of extraction buffer (10 mM NaH 2 PO 4 -Na 2 HPO 4 , pH 7.5, 1 mM DTT and 0.1 mM EDTA) by vortexing briefly for mixing. The cell debris in the mixture was removed by centrifugation at 12,000 × g for 10 min at 4 °C then filtration with 0.22-μ m microfiltration membranes. The supernatant was then injected into a HiTrap desalting column (GE) against the extraction buffer to remove the anthocyanins. A protease inhibitor cocktail (10 μ L /mL) (Product Number: P9599; Sigma) was then added to the extracts that were used as the source of enzymes. All steps in enzyme extraction were carried out at 4 °C with the protein concentration being quantified using a Bradford assay kit.
The UCGalT activity was assayed in a mixture (50 μ L) containing 8 μ g of enzyme extract, 50 mM glycine-NaOH (pH = 8.6), 1 mM UDP-galactose, 0.2 mM acceptor substrates (cyanidin, peonidin and pelargonidin) which was incubated at 30 °C for 30 min. Five μ L of 12 M HCl was added to the reaction mixture then centrifuged to precipitate proteins. A total of 20 μ L of the supernatant was analyzed using HPLC using the method outlined above.

Detection of the expression patterns of DcUCGalT1 in taproots of purple and non-purple carrots.
Three purple carrot cultivars (Deep purple, Purple 68 and Tianzi2hao) and six non-purple carrot cultivars (Kuroda, Sanhongliucun, Junchuanhong, Bejo1719, Qitouhuang and Baiyu) were grown in the soil under the same conditions as previously described 26 . The total RNA was extracted from the 60-day-old carrot taproots using the method as previously described 26 . cDNA was synthesized and was diluted 20-fold for detecting gene expression levels by qRT-PCR analysis using the method as previously described 26 .
Detecting the effect of UV-light irradiation on the DcUCGalT1 expression patterns in 'Deep purple' and 'Kuroda'. The seeds of 'Deep purple' and 'Kuroda' were sterilized as follows: soaked in tap water for 12 h, immersion in 70% (v /v ) ethanol for 1 min, soaking in sodium hypochlorite (1.0% available chlorine) containing a drop of Tween ™ 20 for 45 min, rinsing with sterile tap water three times, and transferring onto filter papers to remove excess water. After sterilization, seeds were placed onto solid Murashige and Skoog medium 37 in 90 × 15 -mm Petri dishes, and then were placed vertically at 25 °C in the dark or with continuous UV-containing white light (315-420 nm) 17 irradiation for 12 days. The total RNA was extracted from the 12-day-old roots using the method as previously described 26 . cDNA was synthesized and was diluted 20-fold for detecting gene expression levels by qRT-PCR analysis using the method as previously described 26 .

Quantitative real-time polymerase chain reaction (qRT-PCR) expression analysis.
The forward primer 5′ -ATTCGAGGAACTAGACCCTGACC-3′ and reverse primer 5′ -GCCACAGACTTA GGATTACGCTTG-3′ were designed using Primer 5 for qRT-PCR expression analysis. qRT-PCR was performed as described previously 26 . The DcActin1 gene was used as an internal standard for normalization 26 . The experiments were conducted in triplicate using three biological RNA samples for each carrot cultivar. The relative gene expression was calculated with the 2 −ΔΔCT method 38 . For comparing the expression patterns of DcUCGalT1 among purple and non-purple carrots, the Δ Δ C T was calculated by subtracting Δ C T of 'Sanhongliucun' from Δ C T of all the carrot cultivars. For detecting the effect of UV-light irradiation on the DcUCGalT1 expression patterns in 'Deep purple' and 'Kuroda' , the Δ Δ C T was calculated by subtracting Δ C T of 'Kuroda' under dark from Δ C T of all the samples.