A tyrosine aminotransferase involved in rosmarinic acid biosynthesis in Prunella vulgaris L

Rosmarinic acid (RA) and its derivants are medicinal compounds that comprise the active components of several therapeutics. We isolated and characterised a tyrosine aminotransferase of Prunella vulgaris (PvTAT). Deduced PvTAT was markedly homologous to other known/putative plant TATs. Cytoplasmic localisation of PvTAT was observed in tobacco protoplasts. Recombinantly expressed and purified PvTAT had substrates preference for L-tyrosine and phenylpyruvate, with apparent K m of 0.40 and 0.48 mM, and favoured the conversion of tyrosine to 4-hydroxyphenylpyruvate. In vivo activity was confirmed by functional restoration of the Escherichia coli tyrosine auxotrophic mutant DL39. Agrobacterium rhizogenes-mediated antisense/sense expression of PvTAT in hairy roots was used to evaluate the contribution of PvTAT to RA synthesis. PvTAT were reduced by 46–95% and RA were decreased by 36–91% with low catalytic activity in antisense transgenic hairy root lines; furthermore, PvTAT were increased 0.77–2.6-fold with increased 1.3–1.8-fold RA and strong catalytic activity in sense transgenic hairy root lines compared with wild-type counterparts. The comprehensive physiological and catalytic evidence fills in the gap in RA-producing plants which didn’t provide evidence for TAT expression and catalytic activities in vitro and in vivo. That also highlights RA biosynthesis pathway in P. vulgaris and provides useful information to engineer natural products.

E. coli strain DL39. Despite enhanced RA accumulation has been demonstrated by overexpressing TAT in S. miltiorrhiza hairy roots and transgenic P. frutescens plants 52,53 . However, biochemical confirmation of their enzymatic activity has not been shown in major RA-producing plants. In plants, tocopherols and RA have been defined roles in free radical scavenging and in defence against abiotic and biotic stresses 11,54 , and BIA are used by humans as stimulants, narcotics, and therapeutic agents 50,55 . These natural products increase plant fitness and are also linked to potential benefits for human health.
Although much work has been done to identify the genes responsible for RA metabolism, the enzymes involved in the supply of precursors are still poorly defined in P. vulgaris. Here, we isolated and characterised tyrosine aminotransferase cDNA (PvTAT) using homology-based cloning, determined its subcellular localisation, characterised the heterologously overexpressed protein, and showed its functional complementation for tyrosine auxotrophic mutant DL39. Moreover, antisense/sense-based expression showed PvTAT is involved in RA biosynthesis in planta.

Results
Identification of a tyrosine aminotransferase as a candidate rosmarinic acid biosynthetic enzyme. A deep transcriptome database was generated by 454 GS-FLX Titanium pyrosequencing using a cDNA library prepared from P. vulgaris root (http://www.ncbi.nlm.nih.gov/sra/SRX130243). In search of the transcriptome database of P. vulgaris using S. miltiorrhiza TAT (KM575934.1) and P. frutescens TAT (AJ458993.1) as the query sequence revealed the presence of highly homologous unigenes (similarity > 80%). We used homologous cloning and obtained a cDNA belonging to the PLP-dependent Asp aminotransferase superfamily (AATlike proteins) and also belonging to the tyrosine aminotransferase (TAT) subfamily according to the Panther classification system (http://www.pantherdb.org/). The cDNA contained a 1,236-bp ORF (Supplemental Fig. S1, lane 1) and encoded a predicted translation product of 411 amino acids with a molecular mass of 45 kDa and an isoelectric point of 5.97. The gene had a sequence of 2,634-bp (Supplemental Fig. S1, lane 2) with seven extrons and six introns and was identical to the cDNA except for three nucleotides. These differences did not result in amino acid substitutions.
Expression analysis of PvTAT. We measured the relative transcript expression level of PvTAT in different organs (roots, stems, leaves, and spikes) of P. vulgaris at complete flowering stage by qRT-PCR. PvTAT involved in RA biosynthesis was expressed in all the tissues but possessed distinct expression patterns. The highest transcript abundances were observed in root tissues, followed by leaves. The relative expression levels in roots and leaves were 4.6-and 2.8-fold higher than in the stems, while the transcript level in the spikes was set to 1 (Fig. 2c).
Cytoplasmic localisation of PvTAT. To experimentally determine the subcellular localisation of the PvTAT in planta, we expressed a PvTAT-GFP fusion protein in tobacco protoplasts. An engineered green-shifted green fluorescent protein (GFP), by itself or fused in frame to the C-terminus of TAT, was driven by the cauliflower mosaic virus (CaMV) 35S promoter. The resulting constructs or GFP alone were transformed into tobacco leaves. As shown in Fig. 2d, when GFP alone was expressed, fluorescence was observed almost throughout the cytoplasm and nucleus. Protoplasts with PvTAT-GFP exhibited a different pattern, only the cytoplasm spread fluorescence. Together, the difference between the PvTAT-GFP and GFP free fluorescence pattern suggests that PvTAT is localised in the cytoplasm, which is consistent with the prediction made by PSORT Prediction (http:// psort.hgc.jp/form.html).
PvTAT possesses tyrosine aminotransferase activity. To functionally determine the in vitro biochemical roles of the putative PvTAT activity, the pET32a-TAT plasmid or empty vector was expressed in the E. coli strain Rosetta (DE3) and His-Tag purified to near homogeneity (Supplemental Fig. S2, lane 3). The purified recombinant TAT showed evidence of tyrosine aminotransferase activity (Supplemental Fig. S3a). Tyrosine aminotransferase activity in the forward direction forming 4-HPP and glutamate was found to be highest at pH 8.0 and 30 °C (Supplemental Fig. S3b,c), a pH optimum similar to that of found for the identified Arabidopsis At5g36160 46 . The substrate specificity of purified recombinant PvTAT was assessed using three aromatic amino acids as amino donors, together with several potential oxoacid acceptors as potential co-substrates. The activity of PvTAT indicates that it could use L-Tyr, L-Phe, and L-tryptophan (L-Trp) to varying degrees as amino donors and use α-ketoglutarate, pyruvate, oxaloacetate, and phenylpyruvate to diverse degrees as amino acceptors. With α-ketoglutarate as the amino acceptor, PvTAT shows higher preference for L-Tyr amino donor with an apparent K m of 0.40 mM than for L-Phe (K m of 10.2 mM) and L-Trp (K m of 8.14 mM) ( Table 1; Supplemental Fig. S4a-c). The catalytic efficiency (k cat /K m ) for L-Tyr was 20-and 23-fold greater than that of L-Phe and L-Trp. With L-Tyr as the amino donor, PvTAT preferentially utilises aromatic oxoacid phenylpyruvate with a K m of 3.90 mM as the amino acceptor, other than aliphatic oxoacid (α-ketoglutarate, pyruvate, and oxaloacetate with a K m of 10.80, 52.06, and 4.72 mM, respectively) ( Table 1; Supplemental Fig. S4d-g). In the reverse direction with L-glutamate (L-Glu) as amino donor, PvTAT favours 4-HPP as oxoacid acceptor with a K m of 0.48 mM (Table 1; Supplemental Fig. S4h-j). To confirm TAT enzyme activity, the native enzyme was compared to heat-inactivated enzyme as a negative control. We determined that PvTAT preferentially catalyses the transamination of L-Tyr to form 4-hpp, which is used as substrate in RA biosynthesis. Unexpectedly, PvTAT has a preference for phenylpyruvate as the preferred amino acceptor rather than α-ketoglutarate. It is likely that PvTAT would catalyse the formation of phenylalanine from tyrosine and phenylpyruvate and that 4-HPP would be produced that in vivo would be channeled toward RA biosynthesis.  Table S1. (b) Phylogenetic relationship of and related proteins. An unrooted neighbour-joining tree was constructed using MEGA 5.05 software. Numbers refer to the percentage bootstrap test support for each node over 1,000 bootstrap iterations. Numbers in parentheses show the percentage amino acid identity of each protein compared with TAT from P. vulgaris. The identifiers of the proteins are provided in Supplemental Table S1. (c) Relative expression of PvTAT in various organs of P. vulgaris. Gene relative expression was determined by qRT-PCR and is shown relative to the expression level of the housekeeping gene β-actin. Relative expression was calculated using the 2 −ΔΔt method.Transcripts of PvTAT in spikes were set to 1. Data are means ± SD of three replicates. *P < 0.05 and **P < 0.01 as determined by Tukey HSD test. (d) Subcellular localisation of PvTAT. The PvTAT-GFP or free GFP construct was expressed in tobacco protoplasts and the fluorescence was observed under a confocal microscopy. PvTAT-GFP fusion protein localises in the cytoplasm in tobacco protoplasts. Free GFP served as the control. GFP fluorescence is in green and chloroplasts fluorescence is in red.

TAT functionally complements the E. coli tyrosine auxotrophic mutant DL39. To examine
PvTAT activity in vivo, the recombinant pBAD33-TAT plasmid or pBAD33 57 was transformed into the E. coli triple mutant DL39 58 for an initial functional complementation assay. This strain has mutations in tyrB (tyrosine aminotransferase), aspC (aspartate aminotransferase), and ilvE (branched-chain aminotransferase), leading to auxotrophy for tyrosine, phenylalanine, aspartate, valine, isoleucine and, leucine. As shown in Fig. 3a, all transformants could grow in M9 medium supplemented with tyrosine, phenylalanine, leucine, isoleucine and valine, whereas only clones transformed with PvTAT were able to grow on M9 medium devoid of tyrosine and phenylalanine. Figure 3b displays the growth curve of the E. coli mutant transformed with the native plasmid or the plasmid expressing PvTAT. Bacteria transformed with PvTAT grew continuously, while the bacteria transformed with the native plasmid showed little growth over 20 hours. The complementation was only specific to the tyrosine and phenylalanine auxotrophy, because growth did not occur without aspartate or the branched chain amino acids leucine, isoleucine and valine in combination (Fig. 3c). These results show that the putative PvTAT functionally complements the auxotrophic E. coli mutant and does not participate in the anabolic pathways for aspartate, leucine, isoleucine, and valine.
PvTAT is involved in RA synthesis in P. vulgaris. To determine whether PvTAT is involved in RA biosynthesis in planta, we used an antisense/sense expression-based approach to develop transgenic hairy root lines with regulated levels of PvTAT transcription. Wild-type hairy root lines with untransformed A. rhizogenes ATCC 15834 and hairy root lines transformed with empty vector pCAMBIA2300 were used as wide-type control and vector control, respectively. PCR amplification of rolB/C gene, nptII, and PvTAT confirmed the transformed lines (Supplemental Fig. S5). Despite the relatively low transgenic rate, four antisense and five sense PvTAT-expressing hairy root lines were developed that showed 46-95% decreases and 0.77-2.6-fold increases, respectively, in transcription levels relative to the wild-type lines (Fig. 4a). To investigate the degree of decrease/ increase in total TAT activity due to the antisense/sense expression of PvTAT, we assayed TAT activity in the hairy root lines. Total activity was reduced by 23-67% in antisense-expressing hairy root lines and increased by 43-60% in sense-expressing hairy root lines (Fig. 4b), consistent with the expectation that the antisense/sense expression of PvTAT leads to loss/gain of TAT activity. Reduced PvTAT expression and activity resulted in a decrease in RA levels by 36-92% in all downregulated lines. Increased PvTAT expression and activity led to an increase in RA accumulation by 31-83% in all overexpressing lines (Fig. 4c). The positive relationship between PvTAT expression, TAT activity, and RA accumulation suggests that PvTAT serves an important role in the production of RA in P. vulgaris (Fig. 4d). Together, these results provide in vivo evidence of the major role of PvTAT in RA biosynthesis in P. vulgaris.

Discussion
In plants, many substances are derived from tyrosine, such as RA, homogentisate, tocopherol, and BIA, which contribute to human health. RA and its numerous derivatives have many potential nutraceutical and pharmaceutical applications 11 . Since RA was first discovered and isolated in Rosmarinus officinasis 59 , researchers have been studying its biogenesis. Because of its biological and pharmaceutical importance, RA has been studied in many medicinal plant species. TAT genes involved in RA biosynthesis have been cloned from C. blumei, S. miltiorrhiza 15 , and P. frutescens 53 . Aminotransferase activity are also implicated in the biosynthesis of aromatic volatiles, tocopherol, BIA, and tropane alkaloid and have been characterised in melons, Arabidopsis, P. somniferum, and A. belladonna 34, 42-44, 46, 50, 51 , respectively. Here, we cloned the entry-point enzyme TAT from P. vulgaris, a member of the Labiatae. Through the isolation and characterisation of TAT from P. vulgaris, we provide biochemical and physiological evidence for the involvement of 4-HPP as an intermediate in RA biosynthesis (Fig. 1). Based on a phylogenetic tree of sequence similarity, plant, mammalian, and bacterial TATs all differ from each other despite being grouped in the same subdivision within the same subgroup of the class I aminotransferases 60,61 . Conserved amino acid residues responsible for catalysis and binding PLP are found in all proteins (Fig. 2a). Among them, Lys residue is the catalytic sites interacting with the pyridine nitrogen of PLP. Lys-246 may perform this catalytic function in PvTAT. Arg in the highly conserved attachment site is one of the residues responsible for the interaction of Tyr with TAT 40 , and Arg-254 may play this role in PvTAT (Fig. 2a). Although PvTAT shares only 15% amino acid identity with E. coli TAT, it was able to complement the E. coli mutant DL39 58 (Fig. 3a). This was also the case for Arabidopsis At5g36160, At5g53970, and PhPPY-AT 33,42,45 . None these are involved in the anabolic pathways for leucine, valine, isoleucine, and aspartate (Fig. 3c).
In plants, despite L-Phe and L-Tyr are synthesised in plasmids from prephenate, which is derived from chorismate formed through the shikimate pathway 62 . Secondary metabolites, like phenolic acid, flavonoids, volatiles, alkaloids, tocopherols, and cyanogenic glycoside and proteins are derived from the aromatic amino acids L-Phe and L-Tyr 63 . However, plants have retained multiple TATs that likely utilise L-Phe and L-Tyr as substrates and are not plastid localised, but rather cytosol localised. A petunia PhPPY-AT localised in the cytosol was able to directly catalyse the formation of L-Phe from phenylpyruvate using L-Tyr as the amino donor, yielding 4-HPP as a by-product 45 . Moreover, an A. belladonna ArAT4, which is not localised to any organelle, was able to directly catalyse the formation of L-Tyr from 4-HPP using L-Phe as the amino donor, yielding phenylpyruvate as a by-product 50 . The two reactions are mutually inverse and are analogous to the formation of L-Phe and L-Tyr in bacteria, demonstrating the existence of a second, direct, extraplastidic route to L-Phe and L-Tyr biosynthesis in plants and leading to a proposed role for PhPPY-AT and Ab-ArAT4 in modulating aromatic amino acid homeostasis 45,50 . Arabidopsis TAT genes (At2g24850, At5g53970) were both shown to encode the enzyme capable of interconverting L-Tyr and 4-HPP, as well as L-Phe and phenylpyruvate, catalysing the first step in tocopherol biosynthesis 45,64 . Another Arabidopsis TAT gene (At5g36160) was also reported to be capable of interconverting L-Tyr and 4-HPP, as well as L-Phe and phenylpyruvate 46 . More recently, Wang, et al. 48 reported that Arabidopsis At5g53970 favoured the direction of Tyr deamination to 4-HPP but At5g36160 preferred the reverse reaction, both were localised outside of the plastids. The opium poppy TAT was as an aminotransferase capable of interconverting L-Tyr and 4-HPP, as well as L-Phe and phenylpyruvate, and contributed to BIA biosynthesis 34 . Additionally, an E. sinica aromatic aminotransferases (EsAroAT1) was showed to function as opium poppy TAT, but is unlikely to contribute to ephedrine alkaloid biosynthesis through the catabolism of L-Phe to phenylpyruvate as a precursor to benzaldehyde 51 . Herein, PvTAT was also confirmed to localise in the cytoplasm and is an aminotransferase capable of interconverting L-Tyr and 4-HPP, interconverting L-Phe and phenylpyruvate  Table 1). Antisense expression of PvTAT led to dramatically decreased RA accumulation (Fig. 4c), confirming the involvement of this enzyme in the formation of RA in P. vulgaris.
Although we cannot role out that PvTAT may have amino donor substrates other than L-Tyr, L-Phe, and L-Trp, we demonstrated its strong preference for L-Tyr over the structurally related amino acid L-Phe and L-Trp (Table 1), similar behavior with A. officinalis TATs 14 , Arabidopsis At5g36160 and At5g53970 45, 46, 48 , opium poppy TAT 34 , and E. sinica AroAT1 51 . A recombinant TAT from T. cruzi was shown to have both tyrosine aminotransferase and alanine aminotransferase activities, displaying an extended substrate enzyme activity 65 . Traditionally, TAT enzymes have been characterised with aliphatic keto acid (α-ketoglutarate, oxaloacetate, and pyruvate) as acceptors and α-ketoglutarate or oxaloacetate was considered as the preferred keto acid acceptor 14,33,34,46 . However, this perception has been changed since the characterisation of PhPPY-AT and CmArAT1 showed a preference for phenylpyruvate rather than α-ketoglutarate, and interconnected Tyr and Phe metabolism in petunia and melon 45 , more and more aminotransferases were found with similar behavior. Ab-ArAT4 has been characterised to prefer the conversion of 4-HPP to Tyr using Phe amino donor, also interconnecting Tyr and Phe metabolism in A. belladonna 50 . What's more, Arabidopsis At5g53970 was showed phenylpyruvate preference and may contribute to the cytosolic formation of Phe, like for phenylpropanoid biosynthesis, similar to its homolog in petunia 48 . Our work shows that PvTAT shows a substrate preference toward Tyr and phenylpyruvate (Table 1). However, the K m (approx. 4 mM) for the best acceptor substrate (phenylpyruvate) is high and there may be more preferred substrates in vivo. Based on the presented data, PvTAT appears to be not an enzyme with high efficiency when compared with Arabidopsis At5g53970 and Ab-ArAT4 but possesses similar catalytic efficiencies to Arabidopsis At5g36160 and PhPPY-AT.
Many attempts have been made to engineer increased RA production by diverting more primary metabolites toward the production of 4-hydroxyphenylpyruvate or by the regulation of biotic/abiotic factors. Co-expression Data are means ± SD of three replicates. *P < 0.05 and **P < 0.01 as determined by Tukey HSD test. WT, the untransformed hairy roots lines (wild-type control); VC, empty vector transformed hairy roots lines (vector control); Anti-TAT, antisense-expressed TAT hairy root lines; Sense-TAT, sense-expressed TAT hairy root lines.
of SmTAT and SmHPPR in S. miltiorrhiza increased RA by 4.3-fold, although overexpression of SmTAT alone did not increase RA content 52 . Overexpression of the PfTAT gene from P. frutescens led to RA levels to be 1.5-2.0 folds higher than controls 53 . By combining the genetic manipulation of the ectopic expression of AtPAP1 and co-suppression of the cinnamoyl-CoA reductase and caffeic acid O-methyltransferase genes in S. miltiorrhiza, a 2.49-fold increase of RA was obtained 66 . Our data showed that overexpression of the PvTAT in hairy roots increased TAT expression, followed by increased enzyme activity, and and 31-83% increases in RA accumulation (Fig. 4c). In contrast, antisense expression of PvTAT successfully suppressed TAT expression, followed by decreased enzyme activity and RA accumulation, especially in antisense line 3 and 4, with 90% and 92% decreases in RA accumulation (Fig. 4c). The positive relationship between PvTAT mRNA levels, total TAT activity, and RA accumulation suggest that PvTAT has an important role in the production of this phenolic acid in P. vulgaris. The biochemical characterisation of a recombinant TAT in vitro, coupled with the physiological evaluation of function in planta, supports the role of 4-HPP as an intermediate in the formation of RA precursors.
Together, we presented the isolation and characterisation a tyrosine aminotransferases cDNA (PvTAT) from P. vulgaris, commonly known as "heal-all". We have provided comprehensive information for PvTAT expression and test for the catalytic activities in vitro and in vivo, which fill in the gap in PA-producing plants. What's more, the physiological evidence that PvTAT important for the synthesis of rosmarinic acid (RA) in P. vulgaris was confirmed by modest correlation between transcript level, TAT activity, and RA accumulation. These results highlight the RA biosynthesis pathway in P. vulgaris, and provide useful information to engineer natural products. These observations open an opportunity to modulate the level of this biotechnologically and pharmaceutically important phenolic acid through the modulation of PvTAT expression.

Materials and Methods
P. vulgaris RNA and DNA isolation and cDNA synthesis. Total RNA from P. vulgaris leaves was isolated using the RNAprep pure Plant Kit (TIANGEN, Beijing, China) according to the manufacturer's protocol and reverse transcribed to generate cDNA using the PrimeScript RT reagent kit (Takara, Japan). Genomic DNA was isolated from P. vulgaris leaves using a DNA isolation kit (Omega, USA) following the manufacturer's protocol.
Amplification of PvTAT cDNA. Homologous cloning was used to obtain the ORF sequence of P. vulgaris. The same Labiatae family members of S. miltiorrhiza (KM575934.1) and C. blumei (AJ458993.1) TAT sequences were used as the query sequence to locally blast against P. vulgaris transcriptome database (Genebank SRX130243) with BioEdit (v7.0.9). Unigenes with more than 80% similarity were chose and assembled using SeqMan of DNAStar. A joint and predicted PvTAT ORF sequence was then obtained. A PCR reaction was carried out using the forward primer (5′-ATGGAGTTGCAGAGCTCGGCGCAG-3′) and the reverse primer (5′-TTAGTAGGAGTGTCGTTCGCAGAAC-3′) to amplify PvTAT ORF sequence. Using the same primer pairs, the DNA sequence was amplified using the DAN as a template.
Multiple sequence alignment and phylogenetic analysis. Amino acid alignments were performed using DNAMAN 7.0. An unrooted neighbour-joining phylogenetic tree was constructed using MEGA version 5.05 with the Poisson model and default parameters. A bootstrap test of 1000 replicates was used to value the reliability of the phylogeny. The proteins used for sequence alignments and phylogenetic tree construction are available in Supplemental Table S1.
Transcription analysis of PvTAT. Plant materials were described in Ru, et al. 67 . At the reproductive stage, leaves, stems, roots, and spikes were separately sampled and kept at −80 °C before use. Extraction and reverse-transcription of total RNA were carried out as mentioned above. Gene expression was determined by qRT-PCR following our previous method 68 and is shown relative to the expression level of the housekeeping gene β-actin. Relative expression was calculated using the 2 −ΔΔt method. Transcript levels of related genes in the spike were set to 1. Data are means ± SD of three replicates of three independent individuals. Tukey HSD test was used to compare the difference among tissues. Primers were listed in Supplementary Table S2.
Subcelluar localisation analysis of PvTAT. The full-length ORF sequence introduced XhoI and NcoI restriction sites was subcloned into the PC0390-pA7-GFP plasmid to produce TAT-GFP fusion proteins. Construction of recombinant plasmid and transient transformation of tobacco were referenced to our previous method 67 . Tobacco protoplasts and leaves were both observed for the determination of PvTAT locolisation. Images were acquired using a confocal microscope (Nikon, Japan) and software of NIS-Elements Viewer 4.0. The PC0390-pA7-GFP plasmid was used as a marker. GFP was excited at 488 nm and collected over a 500-550 nm bandpass. Chlorophyll fluorescence was excited at 638 nm and collected over a 662-737 nm bandpass.
Expression and purification of recombinant proteins. The pET32a-TAT plasmid encoding PvTAT was constructed by PCR amplification of the PvTAT ORF sequence that introduced EcoRI and HindIII restriction sites. The PCR fragments were subcloned into the PMD19T Simple cloning vector (Takara) and checked by sequencing. The PMD19T-TAT plasmid and the pET32a (NEB, England) vector were digested by EcoRI-HindIII restriction enzymes to produce the fusion protein with an N-terminal His tag. The resulting plasmid was termed pET32a-TAT and transformed into the E. coli strain Rosetta (DE3).
Escherichia coli strain Rosetta (DE3) cells carrying the pET32a-TAT plasmids or the pET32a empty plasmids were grown at 37 °C in 500 ml of Terrific-Broth supplemented with 34 mg l −1 chloramphenicol and 100 mg l −1 carbenicillin. When the culture reached an OD 600 of ~0.6, isopropyl-β-1-D-thiogalactoside (1 mM) was added to a final concentration of 0.3 mM to induce protein expression and the culture was incubated for 6 h. The cells were harvested by centrifugation at 10,000 g for 10 min at 4 °C. The pellet was washed twice with PBS buffer, suspended in 200 mM Tris, 100 mM KCl, and 10% glycerol (w/v), pH 8.0, and lysed by high-pressure broken using a cell disrupter (JN-02C, JNBIO, Guangzhou, China). The crude extract was centrifuged at 12,000 g for 20 min at 4 °C to yield a cell-free supernatant. The supernatant was used for affinity purification over His-Tag purification resin (Solaibao, Beijing, China). Proteins eluted with 250 mM imidazole were analysed by SDS-PAGE on a 12% (w/v) acrylamide gel and visualised using Coomassie Brilliant Blue R-250 staining. Protein concentrations were determined using a Bradford Protein Assay Kit (Takara).
Western blotting analysis was determined as Wang, et al. 69 described previously, except the protein was transferred to NC membrane (Solarbio). The membrane was incubated with anti-His antibodies at a dilution of 1:3000 in blocking buffer for 1 h, after washing with TBST buffer for 10 min, the membrane was then incubated with goat anti-rabbit IgG alkaline phosphatase-conjugated as secondary antibodies (1:3000 dilution) for 1 h. Antigen-antibody complex was visualised using the BCIP/NBT kit (Beyotime, China).
Enzyme assay. To measure TAT activity in P. vulgaris hairy root lines, proteins were extracted and determined as Lopukhina, et al. 42 described previously with minor modification. Hairy roots (0.2 g fresh weight) were extracted with 3 ml of ice-cold buffer, containing 100 mM potassium phosphate buffer, 0.1 mM EDTA, 8.0 mM α-ketoglutarate, 0.2 mM pyridoxal-5-phosphate (PLP), and 1 mM dithiothreitol, pH 8.0. After centrifugation at 12,000 g at 4 °C for 10 min, the supernatant was incubated on ice and used as a crude extract enzyme. The reaction mixture containing 125 mM KH 2 PO 4 /K 2 HPO 4 buffer, 5.5 mM L-Tyr, 0.75 mM EDTA, 0.1 mM PLP, and 10 mM α-ketoglutarate, pH 8.0, was pre-incubated at 30 °C for 30 min. The reaction was initiated by adding 200 μl of crude protein and incubated at 30 °C in a total volume of 3 ml. After 30 min the reaction was terminated with 1 ml of 10 M KOH and extinction of 4-hydroxybenzaldehyde (product derivative) was measured exactly 30 min later at 331 nm using an ultraviolet-visible spectrophotometer (UV-1700, SHIMADZU, Japan). The extinction coefficient used for 4-hydroxybenzaldehyde was 24 900 L mol −1 cm −1 33 .
To estimate the pH optimum for the recombinant protein, pH values were varied between 6 and 10 and 5 μl (0.2 μg) of purified protein was assayed as described above in a total volume of 1 ml. Enzyme kinetics were obtained by monitoring the absorbance of various transamination reaction products 4-HPP at 331 nm, phenylpyruvate at 320 nm, and indole-3-pyruvate at 328 nm corresponding to the substrates L-Tyr, L-Phe, and L-Trp. The assay contained 125 mM KH 2 PO 4 /K 2 HPO 4 buffer, 0.75 mM EDTA, 0.1 mM PLP, and various substrate concentrations, pH 8.0, and incubated at 30 °C in a total volume of 1 ml. When α-ketoglutarate (10 mM) was the amino acceptor, 0-4 mM L-Tyr, 0-50 mM L-Phe, and 0-50 mM L-Trp were used as the amino donor. When L-Tyr (5.5 mM) was the amino donor, 0-100 mM α-ketoglutarate, 0-200 mM pyruvate, 0-300 mM oxaloacetate, and 0-40 mM phenylpyruvate were used as the amino acceptor.
V max and K m values were calculated according to a nonlinear regression of the Michaelis-Menten equation, where V = (V max S)/(K m + S) 70 . For the calculation of k cat , a molecular weight of 65 kDa was assumed for the His-tagged TAT protein. All enzyme assays were performed at an appropriate enzyme concentration so that reaction velocity was linear and proportional to enzyme concentration during the incubation time period. Kinetic data were evaluated by curveExpert 1.4. At least triplicate assays were performed for all data points.

Antisense and sense expression of PvTAT in planta.
To construct pCAMBIA2300-TAT sense and antisense expression vectors, the PvTAT ORF, introduced XbaI and KpnI restriction sites, was inserted into the pCAMBIA2300 vector in sense and antisense orientations. Then, the pCAMBIA2300-TAT and pCAMBIA2300 plasmids were transformed into an A. rhizogenes strain ATCC 15834 using the heat shock method. Positive clones were confirmed by PCR and restriction enzymes digestion. Plant transformation was carried out according to our previous method 68 , except that the transformants were screened with cefotaxim solution and kanamycin together.
Identification of transgenic hairy roots by PCR. DNA was isolated from wild-type and transgenic hairy roots using a genomic DNA isolation kit (Omega) according to the manufacturer's protocol. Four primer pairs, rolB/C, nptII, 35S forward primer 1(35S F1), and STAT reverse primer (STAT R) were designed for the identification of sense-expressing hairy root lines. 35SF was located at the 35S promoter of the pCAMBIA2300 vector and STATR was from the PvTAT gene. 35S forward primer 2 (35SF2) and ATAT reverse primer (ATATR) were designed for the identification of antisense-expressing hairy root lines. ATATR was also from the PvTAT gene. Primers were shown in Supplemental Table S2.
Scientific RepoRts | 7: 4892 | DOI:10.1038/s41598-017-05290-4 HPLC assay for RA content. RA accumulation in wild-type hairy roots and transgenetic hairy root lines was assayed using HPLC according to our previous method 68 . Analysis for PvTAT gene expression. qRT-PCR analysis for PvTAT gene expression in wild-type hairy roots and transgenetic hairy root lines was determined as mentioned above. Transcripts of PvTAT in the wild-type hairy root lines were set to 1.

Data analysis.
One-way analysis of variance (ANOVA) was performed to highlight differences in wide-type hairy root lines, hairy root lines transformed with empty vector, and transgenic hairy root lines followed by 'Tukey HSD' post-hoc multiple comparison tests. Normality and homogeneity of variance of data were assessed prior to analysis. All the data analyses were accomplished using the "Statistical Package" for Social Sciences program (SPSS 16.0, SPSS Inc., USA). OriginPro8.0 (OriginLab Corporation, Northampton, MA) was used for graphical presentation. All the data were expressed as means ± SD with triplicate replicates.