Redirecting Primary Metabolism to Boost Production of Tyrosine-Derived Specialised Metabolites in Planta

L-Tyrosine-derived specialized metabolites perform many important functions in plants, and have valuable applications in human health and nutrition. A necessary step in the overproduction of specialised tyrosine-derived metabolites in planta is the manipulation of primary metabolism to enhance the availability of tyrosine. Here, we utilise a naturally occurring de-regulated isoform of the key enzyme, arogenate dehydrogenase, to re-engineer the interface of primary and specialised metabolism, to boost the production of tyrosine-derived pigments in a heterologous plant host. Through manipulation of tyrosine availability, we report a 7-fold increase in the production of tyrosine-derived betalain pigments, with an upper range of 855 mg·kg−1·FW, which compare favourably to many in vitro and commercial sources of betalain pigments. Since the most common plant pathway for tyrosine synthesis occurs via arogenate, the de-regulated arogenate dehydrogenase isoform is a promising route for enhanced production of tyrosine-derived pharmaceuticals in diverse plant hosts.

SCienTifiC REPORTS | (2018) 8:17256 | DOI: 10.1038/s41598-018-33742-y ( Fig. 1a) 8 . ADHα is the product of a gene duplication event specific to Caryophyllales, and most species in Caryophyllales, therefore, possess at least two copies of ADH -the de-regulated ADHα and a Tyr-sensitive ADHβ 8 . Clearly, the expression of the de-regulated ADHα isoforms in heterologous plant platforms has the potential to enhance the production of Tyr-derived metabolites. Betalains are one such example of Tyr-derived metabolites, and are a class of pigments comprising yellow-orange betaxanthins and red-violet betacyanins, unique to Caryophyllales 11,12 . Betacyanins are the major natural red-colour dyes used in the US food industry 11,12 . The committed biosynthetic pathway for betalains is relatively short, proceeds in three enzyme-mediated steps (Fig. 1b), and has been successfully expressed in both plant and microbial host platforms [13][14][15] . Betalain pigments have been previously employed as an enzyme-coupled bio-sensor to explore flux in the production of Tyr-derived metabolites 13 , offering a powerful tool to visualise the impact of manipulating primary Tyr metabolism. Here, we demonstrate the capacity of de-regulated ADH to, in turn, boost the production of Tyr-derived metabolites in the heterologous plant platform, Nicotiana benthamiana.

Results and Discussion
In our previous work, transient expression of a de-regulated ADHα driven by the CaMV 35S promoter resulted in an approximate ten-fold increase relative to its paralog, the Tyr-sensitive ADHβ 8 . While transient over-expression of the Tyr-sensitive ADHβ alone does not significantly enhance the production of L-Tyr (relative to a GFP expression control), the over-expression of the de-regulated ADHα increases Tyr levels by an order of magnitude 8 . We here examined the subcellular localization of these two paralogs following transient expression in Nicotiana benthamiana utilizing fusion of GFP to the C-terminus of the ADH isoforms. Consistent with previously observed plastid localization of the ADHα and ADHβ proteins in protoplast transformations 8 , we found in whole tissue visualization that both paralogs exhibit similar patterns associated with discrete patches at the plastids 16 (Fig. 2). The location of both paralogs to the plastids emphasizes that the previously observed difference in Tyr levels between the expression of de-regulated ADHα and Tyr-sensitive ADHβ isoforms are not the consequence of localization to different sub-cellular compartments, with differing pools of available arogenate substrate. We then coupled the ADH isoforms to a Tyr-derived biosynthetic pathway producing betalains. Here, we isolated the three essential enzymes of the committed betalain pathway: DODA and CYP76AD1 from Beta vulgaris and cDOPA-5GT from Mirabilis jalapa, and assembled these into a multi-gene vector following a previously published design (See Methods) [13][14][15] . Additionally, either ADHα or ADHβ from Beta vulgaris was incorporated. The firefly luciferase gene was also included to adjust for differences in transformation efficiency between replicates, and for within-leaf variation between infiltration sites (See Methods) (hereafter these multi-gene vectors are referred to as ADHα-BET and ADHβ-BET, Supplementary Fig. 1a,b). Transient expression of ADHα-BET versus ADHβ-BET in N. benthamiana, generated notable pigment production, three-days post-infiltration, which was absent in the luciferase control ( Fig. 3a,b, Supplementary Fig. 1c). Incorporation of ADHα results in noticeably darker and more intense purple pigmentation spots relative to the ADHβ construct (Fig. 3b), suggesting enhanced production of Tyr-derived betalain pigments.
To confirm these qualitative observations, we further quantified the levels of betalains in two ways. First, liquid chromatography-mass spectrometry (LC-MS) analysis was conducted to identify and quantify the betacyanin pigments produced (Fig. 3c). Three main types of betacyanins, betanin, isobetanin, and betanidin ( Fig. 3d), were identified with betanin comprising over 90% of total betacyanin observed ( Supplementary Fig. 2a-e). We then quantified relative betalain content between ADHα-BET and ADHβ-BET infiltration spots, using the predominant compound, betanin, as a proxy for total betalain content. The incorporation of ADHα resulted in an average 7.3 fold increase in betanin production relative to the incorporation of ADHβ (Fig. 2e), indicating that about 70% of our previously described ten-fold increase in Tyr levels 8 are successfully translated to the enhanced accumulation of downstream Tyr-derived metabolites. Second, the spectrophometric absorbance spectra for betacyanins was measured at 540 nm, corrected for the effect of chlorophyll a absorbance, to quantify the actual mass of betalains produced. Here our data indicate that inclusion of ADHα yields an average of 516 mg·kg −1 ·FW, relative to 74.40 mg·kg −1 ·FW with ADHβ, a similar 7-fold increase in betalain production (Fig. 2f). With an upper range of 855 mg·kg −1 ·FW, the ADHα enhanced yields of betalains compare favourably to many in vitro and commercial sources of betalain pigment 17 emphasising the potential of manipulating primary metabolism to enhance the yield of specialised metabolites in planta. Finally, we confirmed that the relative efficacy of ADHα and ADHβ was not due to differences in transformation efficiency, by replicating these experiments normalised to luciferase expression ( Supplementary Fig. 3a,b).
We have demonstrated the efficacy of the de-regulated ADHα to override intrinsic metabolic regulation of Tyr metabolism in a heterologous host plant, to enhance the over-production of the Tyr-derived metabolites, betalain pigments. In particular, our data emphasise the value of utilising naturally occurring de-regulated isoforms, in this case, ADHα, to re-engineer the interface between primary and specialised metabolic pathways 18 . The fact that the most common pathway for Tyr synthesis in plants is ADH-dependent, emphasises the broad utility of this de-regulated ADHα isoform for the enhanced production of Tyr-derived pharmaceuticals in diverse plant hosts 8 . It is notable that a considerable portion (>70%) of the elevated Tyr (relative to the elevated Tyr from ADHβ) is subsequently incorporated into the Tyr-derived betalain pigments, indicating efficient downstream processing in response to increased Tyr availability. However, in this instance, both the de-regulated ADHα and the betalain pathway, are peculiar to the plant order Caryophyllales 8,11 , and it is likely that the betalain pathway evolved in the context of, or in response to, enhanced Tyr metabolism 3 , ensuring that the reconstituted betalain pathway is predisposed to utilise elevated availability of the Tyr precursor.
Future work should explore the utility of relaxed ADH isoforms for other pharmaceutical production in planta. As demonstrated with the optimisation of Tyr-derived metabolic pathways in microbial platforms, additional rate-limiting steps may constrain the gains in Tyr-derived metabolites 10 despite overall increases in Tyr availability. Also a further topic for evaluation is the broader consequence of overriding Tyr regulation for general plant host metabolism, which may limit carbon flow into phenylalanine-derived pathways 8 , and may affect production of physiologically significant compounds such as anthocyanins, flavonoids, and lignin 3 . Further characterisation of the key residues responsible for the de-regulated ADHα isoform, should allow refined gene-editing approaches in planta, to subtly moderate Tyr metabolic regulation yielding enhanced production of Tyr-derived metabolites, whilst limiting pleotropic physiological consequences. RNA extraction and cDNA synthesis. Tissue was snap frozen in liquid nitrogen and stored at −80 °C until needed. Frozen tissue was ground to a fine powder with a mortar and pestle in liquid nitrogen. RNA   benthamiana were performed according to the previously described agroinfiltration method 19 with some modifications. All constructs were transformed into the Agrobacterium tumefaciens GV3101 strain, and grown at 28 °C in LB media supplemented with antibiotics until reaching an OD 600 of approximately 1.5. Cultures were then brought to a final OD 600 of approximately 0.5 in infiltration media (10 mM MgCl2, 0.1 mM acetosyringone, 10 mM MES at pH 5.6). Cultures were left at room temperature for 2-3 h before infiltration. Infiltration was performed on 6-week old N. benthamiana plants and individual plants represented independent biological replicates. Young, expanding leaves were chosen to infiltrate and infiltration was facilitated by first generating a small nick on the adaxial leaf surface. The positioning of infiltration spots was alternated between biological replicates to account for intra-leaf variation 20

Generation of vectors for betalain synthesis.
The construction of multi-gene vectors containing the betalain biosynthetic genes was carried out using Golden Gate cloning 23,24 . Where necessary, gene sequences were domesticated to remove BsaI and BpiI restriction enzyme sites and alternative nucleotides were chosen to ensure no alteration to the amino acid sequence, using codon optimisation for N. benthamiana. Each multigene vector also included the firefly (Photinus pyralis) luciferase gene to adjust for differences in transformation efficiency and within leaf variation 20 . The luciferase gene was obtained from the plasmid pNWA62 provided by Dr Nick Albert (Plant and Food Research, Palmerston North, New Zealand) and had previously been modified to include an intron and codon optimised in order to enhance translation 25  To quantify relative betalain content between ADHα and ADHβ infiltration spots, the predominant betalain component, betanin, was chosen as a proxy for total betalain content.
Total mass quantification of betalains using spectrophotometer. Betalain content was estimated spectrophotometrically using a SANYO SP75 UV-VIS (Sanyo, Osaka, Japan) spectrophotometer, as A538 − (0.46 × A662) , where A538 and A662 are the absorbance values for betacyanins and chlorophyll a at 538 nm and 662 nm respectively. The subtraction of (0.46 × A662) compensated for the small overlap in absorption by extracted chlorophyll and the correction factor was recalculated for this extraction method. Absorbance values were converted to betanin equivalents using the molar extinction coefficient ε = 60 000 l mol −1 cm −1 and molecular weight = 550 g-mol 27 .
Betalain quantification using plate reader to normalise for luciferase. Leaf tissue was sampled three days post-infiltration with a leaf corer (9 mm diameter) and snap frozen in liquid nitrogen in 2 ml tubes with two 3 mm glass beads. Five technical replicates were sampled for each infiltration spot. Sampled leaf tissue was ground frozen using a Tissue Lyser II homogeniser (QIAgen, Hilden, Germany). Homogenised samples were resuspended in 300 μl of SPB extraction buffer (50 mM sodium phosphate buffer, 2 mM dithiothreitol, 10% v/v glycerol, 1% v/v Triton X-100) 28 and mixed by vortexing. Samples were then centrifuged at 12,100 g for 10 min, and 100 μl of each supernatant was transferred to individual wells of a black µCLEAR 96-well microplate (Greiner Bio-One, Kremsmünster, Austria). Luminescence and absorbance were measured for each well with a CLARIOstar microplate reader (BMG Labtech, Aylesbury, UK). To measure luminescence, 100 μl of SteadyGlo Luciferase Assay Substrate (Promega, Madison, WI, US) was added to each well before plate reading and luminescence was measured at 25 °C using standard settings with no filter. To ensure that luminescence levels fell within a linear range, a luciferase standard curve was also produced using five 1:10 serial dilutions of QuantiLum Recombinant Luciferase (Promega, Madison, WI, US) in SPB supplemented with 1 mM bovine serum albumin (BSA). Absorbance values ranging from 400-700 nm were measured with a resolution of 5 nm. Betacyanin relative concentration was calculated as A540 − (0.1 × A660) where A540 and A660 are the absorbance values for betacyanins and chlorophyll a at 540 nm and 660 nm respectively, corrected to the average SPB buffer values for each wavelength. Obtained betacyanin values were then normalised to the luciferase luminescence measured for each well and corrected to the average luminescence units recorded for the SPB buffer alone.