TAT1 and TAT2 tyrosine aminotransferases have both distinct and shared functions in tyrosine metabolism and degradation in Arabidopsis thaliana

Plants produce various l-tyrosine (Tyr)-derived compounds that are critical for plant adaptation and have pharmaceutical or nutritional importance for human health. Tyrosine aminotransferases (TATs) catalyze the reversible reaction between Tyr and 4-hydroxyphenylpyruvate (HPP), representing the entry point in plants for both biosynthesis of various natural products and Tyr degradation in the recycling of energy and nutrients. To better understand the roles of TATs and how Tyr is metabolized in planta, here we characterized single and double loss-of-function mutants of TAT1 (At5g53970) and TAT2 (At5g36160) in the model plant Arabidopsis thaliana. As reported previously, tat1 mutants exhibited elevated and decreased levels of Tyr and tocopherols, respectively. The tat2 mutation alone had no impact on Tyr and tocopherol levels, but a tat1 tat2 double mutant had increased Tyr accumulation and decreased tocopherol levels under high-light stress compared with the tat1 mutant. Relative to WT and the tat2 mutant, the tat1 mutant displayed increased vulnerability to continuous dark treatment, associated with an early drop in respiratory activity and sucrose depletion. During isotope-labeled Tyr feeding in the dark, we observed that the tat1 mutant exhibits much slower 13C incorporation into tocopherols, fumarate, and other tricarboxylic acid (TCA) cycle intermediates than WT and the tat2 mutant. These results indicate that TAT1 and TAT2 function together in tocopherol biosynthesis, with TAT2 having a lesser role, and that TAT1 plays the major role in Tyr degradation in planta. Our study also highlights the importance of Tyr degradation under carbon starvation conditions during dark-induced senescence in plants.

As one of the 20 proteinogenic amino acids, L-tyrosine (Tyr is used to represent L-Tyr throughout) is an essential building block of protein synthesis (1). Plants and microbes are capable of synthesizing Tyr de novo through the shikimate pathway (2-4). In plants, Tyr also serves as the precursor of a diverse array of specialized (secondary) metabolites of paramount interest to the food and pharmaceutical industry (e.g. isoquinoline alkaloids, rosmarinic acid, vitamin E, salidroside, and betalain pigments) (5)(6)(7)(8)(9)(10)(11). The headgroup of isoprenoid benzoquinones, such as plastoquinone and ubiquinone, can also be derived from Tyr in plants, and these compounds function as electron and proton carriers in photosynthetic and respiratory electron transport chains (12,13). Besides metabolism of Tyr to diverse plant natural products, degradation of Tyr likely plays an important role in providing an energy source during periods of carbohydrate shortage, as each mole of Tyr generates the most net ATP among all of the 20 proteinogenic amino acids during their degradation (14,15). A number of studies showed that degradation of some other amino acids (e.g. branched-chain amino acids and lysine) plays a crucial role under energy-limited conditions during dark-induced senescence in plants ( 16 -23); however, a potential contribution of Tyr degradation in the process remains to be examined.
The removal of the amino group of Tyr is the committed step of Tyr degradation and some Tyr metabolic pathways (Fig. 1). The reaction is mediated by Tyr aminotransferases (TATs, 2 EC 2.6.1.5), which catalyze the reversible reaction between Tyr and 4-hydroxyphenylpyruvate (HPP). In most microbes, HPP is the intermediate of the Tyr biosynthetic pathway, and TATs are usually responsible for the final step of Tyr biosynthesis from HPP (24,25). Indeed, knockout mutants of TATs in microbes often exhibit Tyr auxotrophy (26 -28). Only a limited number of plant species (i.e. legumes) have a microbial-like Tyr biosynthetic pathway via the HPP intermediate (29 -31) and likely have TAT(s) that synthesizes Tyr from HPP. Most plants, however, synthesize Tyr through the alternative pathway via the arogenate intermediate (9,32,33). As a result, in plants, TAT  Table S1 and Figs. S1-S4. 1 To whom correspondence should be addressed: Dept. of Botany, University of Wisconsin-Madison, 430 Lincoln Dr., Madison, WI 53706. E-mail: maeda2@wisc.edu.
enzymes are likely responsible for degradation and metabolism, rather than biosynthesis, of Tyr ( Fig. 1). Current knowledge of the Tyr degradation pathway is mainly based on knowledge from microbes and mammals (Fig. 1); Tyr is first converted to HPP by TAT and then to homogentisate by HPP dioxygenase (HPPD). Homogentisate is further converted to maleylacetoacetate and fumarylacetoacetate by homogentisate oxidase (HGO) and maleylacetoacetate isomerase (MAAI), respectively. Finally, fumarylacetoacetate hydrolase (FAH) converts fumarylacetoacetate into acetoacetate and fumarate, which enters the tricarboxylic acid (TCA) cycle (34,35). All five enzyme activities of the Tyr degradation pathway (i.e. TAT, HPPD, HGO, MAAI, and FAH; Fig. 1) were first detected in mammalian liver (36), and their corresponding genes were later identified (37)(38)(39)(40)(41). Mutations in FAH, TAT, HPPD, and HGO genes cause buildup of Tyr and its derivatives and lead to various genetic disorders, including tyrosinemia and alkaptonuria (42). Corresponding genes were isolated in Pseudomonas putida (35), and their simultaneous expression in Escherichia coli facilitated the efficient consumption of Tyr. Knockout mutants of HGO, MAAI, or FAH, on the other hand, could not grow on the minimal medium with Tyr as the sole carbon source (35), indicating that the Tyr degradation pathway is crucial for generating energy and carbon skeletons from Tyr in this microbe.
Corresponding genes and enzymes were also identified in plants (43,44), suggesting that plants have the same Tyr degradation pathway as animals. All plants so far investigated have nonplastidic TAT, HPPD, and HGO enzymes (45)(46)(47), and Arabidopsis MAAI and FAH are also predicted to be cytosolic due to the lack of a plastid-targeting signal (43). Thus, the major degradation pathway of Tyr is likely located in the cytosol (Fig.  1), although some upstream reactions may redundantly operate in the plastids, depending on plant species (45,46). The FAH mutants of Arabidopsis show a short-day-sensitive-cell-death (sscd) phenotype, which can be recovered by simultaneously knocking out the upstream HGO, likely due to elimination of the accumulation of a toxic succinylacetoacetate derived from fumarylacetoacetate (48). In plants, homogentisate, an intermediate of the Tyr degradation pathway, is also used for synthesis of tocochromanols and plastoquinone (49). Although Arabidopsis HGO knockout mutants have no visible phenotype (48), the hgo deletion mutant of soybean accumulates elevated levels of tocopherols and the brown pigment pyomelanin (47), suggesting that the substrate of HGO, homogentisate, was redirected to these natural product pathways. In Arabidopsis, HPPD knockdown mutant showed an albino phenotype due to reduced level of plastoquinone, whereas HPPD overexpression resulted in elevated accumulation of tocopherols (50,51). Early feeding experiments of spinach chloroplasts indicated that [ 14 C]homogentisate is incorporated into tocopherols and plastoquinone as efficiently as [ 14 C]Tyr or [ 14 C]HPP (52), suggesting that all of these compounds serve as precursors of isoprenoid benzoquinones in plants. However, it is not fully understood how TATs contribute to the Tyr degradation pathway in planta.
Virus-induced silencing of TAT in Papaver somniferum led to reduction in the levels of isoquinoline alkaloids derived from both HPP and Tyr (53). Arabidopsis possesses at least two homologous TAT enzymes, TAT1 and TAT2 (At5g53970 and At5g36160, respectively (45,54,55)), although additional aminotransferases having TAT activity are likely present (25). The TAT1 enzyme prefers Tyr as an amino donor substrate, has a low K m toward Tyr, and favors Tyr deamination to form HPP rather than the reverse reaction. TAT1 had relatively broad keto acid acceptor specificity, and the amino group of Tyr could be transferred to ␣-ketoglutarate, phenylpyruvate, and, to a lesser extent, 4-methylthio-2-oxobutanoate, which are then converted to glutamate, phenylalanine, and methionine, respectively (45). Knocking out TAT1 does not induce visible phenotype under normal conditions but led to a decrease in tocopherol levels, indicating that TAT1 is involved in the biosynthesis of tocopherols (55). TAT1 expression was also found to be significantly increased during continuous dark treatment (56), hinting that TAT1 might also play a role in dark-induced senescence, a phenomenon known to activate amino acid degradation (57). The TAT2 enzyme, on the other hand, has a broad substrate specificity and a relatively high K m toward Tyr and prefers HPP transamination to Tyr (45); however, the in vivo function of TAT2 is currently unknown.
Here, to further examine the physiological roles of TATs and how Tyr is metabolized and degraded in plants, we isolated single and double mutants of Arabidopsis tat1 and tat2, characterized their visible and metabolic phenotypes under different conditions, and performed 13 C tracer experiments. Whereas multiple pathways of Tyr biosynthesis and utilization can exist, depending on different plant species ( Fig. 1 (9)), Arabidopsis provides a relatively simple system to study Tyr metabolism and TAT functions as it does not accumulate lineagespecific Tyr-derived metabolites, like rosmarinic acid or isoquinoline alkaloids. Because orthologs of Arabidopsis TAT1 and TAT2 are present in many plant species (25,45), this study provides baseline information to further investigate Tyr metabolism and degradation pathways in other plant species.

The tat1 and tat2 mutants show reduced cytosolic TAT activity
To investigate the roles of Arabidopsis TAT1 and TAT2 in vivo, two independent T-DNA insertion lines of TAT1 and TAT2 (tat1-1, tat1-2, tat2-1, and tat2-2) were isolated from SALK T-DNA knockout collections (58). Sequencing analyses showed that tat1-1 and tat2-1 carried a T-DNA insertion in the second and third exons, respectively, whereas the T-DNA insertions of tat1-2 and tat2-2 were in first intron and 40 nucleotides upstream of the transcription start site, respectively, of the corresponding locus ( Fig. 2A). To test functional redundancy of TAT1 and TAT2, these single mutants were crossed to generate two double mutants, tat1-1 tat2-1 and tat1-2 tat2-2 (see "Experimental procedures" and Fig. 2A). RT-PCR analysis showed that target gene expression was abolished in single mutant lines except for tat1-2. The tat1-1 tat2-1 double mutant eliminated both TAT1 and TAT2 expression, whereas tat1-2 tat2-2 lacked TAT2 expression but still displayed some TAT1 expression ( Fig. 2B and Fig. S1). qPCR analysis further showed tat1-1 and tat1-2 had significantly lower TAT1 transcript level
Total TAT activity (transamination of Tyr to form HPP using ␣-ketoglutarate or phenylpyruvate as a keto acceptor) in the crude extracts of all single and double mutants was quantified and compared with that of WT. The quantification of specific activity was based on the formation of glutamate from ␣-ketoglutarate keto acceptor or phenylalanine from phenylpyruvate ( Fig. 2E), both of which showed similar results overall. All single and double mutants showed significantly lower total TAT activity than WT. No significant difference was observed in total TAT activity between single and double mutants. Compared with WT, tat mutants showed 20 -40% lower TAT activity, whereas the remaining TAT activities in the mutants are likely derived from other TAT enzymes present in Arabidopsis (25,45). These results showed that down-regulation of either TAT1 or TAT2 resulted in significant reduction of total TAT activity.
To further examine the impacts of tat1 and/or tat2 mutations in plastidic and cytosolic fractions of TAT and other related activities, subcellular fractionation experiments were conducted using both WT and tat mutant leaf tissues. A cytosolic marker, phosphoenolpyruvate carboxylase (PEPC) activity, was only detected in the cytosolic fraction and was not different between WT and tat mutants (Fig. 3C), suggesting that the plastid fractions were free from cytosolic contaminations. Similarly, a plastid marker, nitrite reductase (NiR) activity, was enriched in the plastidic fraction and only minor activity was detected in the cytosolic fraction in WT and tat mutants (Fig.  3D). Activity of HPPD, which catalyzes the subsequent step of TAT ( Fig. 1), showed similar patterns as PEPC activity, and, as expected, both WT and tat mutants showed a similar level of HPPD activity in both crude and cytosol (Fig. 3B). TAT activity was detected in both cytosolic and plastidic fractions (Fig. 3A), as reported previously (45). The cytosolic TAT activity was significantly lower in the tat1-1, tat2-1, and tat1-1 tat2-1 mutants than WT, whereas plastidic TAT activity was similar among all genotypes (Fig. 3A) in agreement with their complete or partial deficiency of TAT1 and TAT2 enzymes localized in the cytosol (45). The tat1-1 tat2-1 mutant, which almost completely lacks TAT1 and TAT2 expression (Fig. 2B), still had Ͼ70% TAT activity (Fig. 3A), suggesting that other unknown enzymes with TAT activity likely contribute to the remaining TAT activity in both cytosol and plastids.

TAT1 and TAT2 play major and minor roles, respectively, in tocopherol biosynthesis
None of the tat1 or tat2 mutants grown on soil exhibited visible phenotypic differences from WT under standard growth conditions at 100-E light intensity (Fig. 4A). The root length

Plant tyrosine transaminases function in tyrosine metabolism
was also not different among genotypes grown on MS medium (data not shown). To test the function of TATs in Tyr metabolism, Tyr and tocopherol levels were analyzed in WT and single and double tat mutants. Both tat1 single mutants accumulated more Tyr and fewer tocopherols than WT (Fig. 4B), consistent with a previous report (55). In contrast, the tat2 single mutants had Tyr and tocopherol levels similar to WT (Fig. 4B). Tyr and tocopherol levels of the tat1-1 tat2-1 and tat1-2 tat2-2 double mutants were similar to those of their respective tat1 background (Fig. 4, B and C), with one exception that the Tyr level of tat1-1 tat2-1 was significantly higher than tat1-1 (Fig. 4B). These results suggest that TAT1 plays a major role in Tyr metabolism for tocopherol biosynthesis, whereas TAT2 has a limited role under standard growth conditions.
To further examine TAT functions under enhanced synthesis of tocopherols, WT and tat mutants were subjected to con-

Plant tyrosine transaminases function in tyrosine metabolism
tinuous high-light intensity (ϳ800 E) conditions, known to enhance tocopherol biosynthesis (59,60). When their tocopherol levels were analyzed at 0, 24, and 48 h of treatment, both WT and tat mutants accumulated more tocopherols (Fig. 4D). After 48 h of treatment, the single tat2-1 mutant accumulated significantly fewer tocopherols than WT. Also, the tat1-1 tat2-1 double mutant had its tocopherol level at half of tat1-1 and one-fourth of WT (Fig. 4D). These results suggest that TAT2 also plays a significant role in tocopherol production when the demand of tocopherols is elevated, such as under high-light stress.

The tat1 but not tat2 mutants deplete carbon and energy fast during dark-induced senescence
To test potential roles of TAT1 and TAT2 in Tyr degradation, such as under carbon and energy starvation, we examined the response of WT and tat mutants to dark-induced senescence (56,57). Four-week-old WT plants and tat single and double mutants were subjected to continuous dark treatment for up to 8 days. The tat1 single and tat1 tat2 double mutants exhibited more pale leaf phenotypes, especially in young leaves (11th to 14th leaf), which had significantly lower total chlorophyll contents than WT and tat2 mutants (Fig. 5A). The tat1 knockout mutant (tat1-1) showed more severe reduction in chlorophyll content than the knockdown mutant (tat1-2). Also, the tat1 tat2 double mutants behaved similarly to their respective tat1 backgrounds (Fig. 5A). When dark-treated plants were recovered under low light (ϳ40 E) for another 4 days, mutants with tat1-1 background could not recover, unlike the other genotypes (Fig. 5B).
A prolonged dark treatment was also conducted using 2-week-old seedlings of WT, tat1, and tat2 mutants grown in 1/2 MS medium (Fig. 5C), and the levels of chlorophylls, Tyr, and sucrose were analyzed (Fig. 5D). Again, chlorophyll and Tyr levels were significantly reduced and increased, respectively, in both tat1-1 and tat1-2 mutants as compared with WT (Fig. 5D). The differences in Tyr levels between WT and tat1 were much more pronounced under the dark treatment (Fig. 5D) than under standard growth conditions (Fig. 4B). Interestingly, the levels of sucrose, the major soluble carbon source of plants, were also lower in both tat1 mutants than in WT (Fig. 5D). No significant differences were observed between WT and tat2. These results suggest that tat1 mutants have reduced degradation of Tyr and depleted carbon source during the dark-induced senescence.
To test this hypothesis further, the MS medium was supplemented with different concentrations of sucrose during the dark treatment. When a very low concentration (0.05%, w/v) of sucrose was supplied, the tat1 single mutants still exhibited more chlorotic leaves than WT or tat2 (Fig. 5C) and contained lower levels of sucrose compared with WT and tat2 after 18 days (Fig. 5D). However, when 1% (w/v) sucrose was supplemented to the growth medium, tat1 mutants grew similarly to WT or tat2 and did not exhibit chlorotic leaves during the dark treatment (Fig. 5C). Also, all WT and tat mutants retained similar levels of chlorophylls and sucrose after 18 days of the dark treatment, although tat1 mutants still accumulated more Tyr than WT or tat2 (Fig. 5D). Thus, sucrose could rescue the carbon-depleting phenotype of tat1 mutants during the dark-induced senescence.
To also examine potential impacts on energy metabolism, respiration rate of WT and tat mutants were monitored for 10 days during the dark treatment. All genotypes had similar respiration rates at day 0 and showed sigmoidal trend of reduction over the 10 days of dark treatment as expected (Fig. 6). On day 5, tat1-1 showed a significantly lower respiration rate than WT, whereas other tat mutants did not. On day 10, the minimum respiration rates of both tat1-1 and tat1-2 were significantly lower than that of WT. Overall, tat1 took less time to reach the

TAT1 is involved in degradation of Tyr into the TCA cycle
To further examine the relative contributions of TAT1 and TAT2 in Tyr metabolism and degradation, leaf discs of WT, tat1-1, tat1-2, tat2-1, and tat2-2 were fed with stable isotopelabeled Tyr, and the incorporation of 13 C into different downstream compounds was monitored over time. The feeding experiment was conducted in the dark, which induces active degradation of Tyr (Fig. 5) and lowers the dilution of labeled Tyr by de novo biosynthesis of Tyr (55). The amounts of labeled Tyr taken up by the leaves were not significantly different between WT and tat mutants during the 6-h labeling time course (Fig. 7A). Whereas homogentisate was not detectable under this experimental condition, both ␥-tocopherol and ␣-tocopherol showed a gradual increase in 13 C label enrichment during the time course (Fig. 7, B and C). Both tat1-1 and tat1-2 mutants showed significantly lower 13 C label incorporation into both ␥-tocopherol and ␣-tocopherol (Fig. 7, B and C). The tat2 mutant showed significantly lower 13 C label incorpo-   5). Asterisks, significant lower respiration rates than WT (t test, p Ͻ 0.05).

Plant tyrosine transaminases function in tyrosine metabolism
ration than WT only to ␥-tocopherol at 4 h (Fig. 7B), consistent with the minor role of TAT2 in tocopherol synthesis (Fig. 4).
Incorporation of 13 C into fumarate and succinate, key intermediates of the TCA cycle, was also analyzed from the same labeled samples. Whereas 13 C-labeled tocopherols were detected in up to 50% of the corresponding total pools, only less than 1% of the total fumarate pool was 13 C-labeled after 6 h of [ 13 C]Tyr feeding, mainly due to much larger pool sizes of these TCA cycle intermediates as compared with those of tocopherols ( Fig. 7 and Fig. S2). 13 C-Labeled fumarate and succinate increased during the 6 h of [ 13 C]Tyr feeding in WT and tat2 at similar rates (Fig. 7, D and E). In contrast, the tat1 mutants had less than one-fourth of 13 C-labeled fumarate and succinate as compared with WT at 4 and 6 h of labeling. These labeling results are consistent with the phenotypic and metabolic responses of tat1 and tat2 (Fig. 5) and further indicate that TAT1, but not TAT2, plays the major role in degradation of Tyr into the TCA cycle.

In vivo functions and localization of Arabidopsis TAT1 and TAT2
Arabidopsis has at least two TAT enzymes, TAT1 and TAT2, that have distinct biochemical properties in vitro; TAT2 has much lower catalytic activity and broader substrate specificity than TAT1 (45,54,55). In this study, reduction of TAT1 expression resulted in increased and decreased accumulation of Tyr and tocopherols, respectively (Figs. 2 and 4), consistent with the results of Riewe et al. (55). In contrast, TAT2 deficiency had a much lower impact, resulting in slightly elevated Tyr accumulation in one of the tat1 tat2 double mutants (Fig. 4B) and no changes in tocopherol levels (Fig. 4C) under normal growth conditions. However, the labeling analyses showed that tat2 incorporated significantly less 13 C label from Tyr into ␥-tocopherol than WT (Fig. 7B). Also, the tat2 mutation significantly reduced tocopherol accumulation in both WT and tat1 backgrounds under high-light conditions, where there is a greater demand for tocopherols (Fig. 4D). Indeed, TAT2 expression is induced under various abiotic stress conditions in the aerial tissues of Arabidopsis seedlings based on a publicly available database (Fig. S3) (61). Also, in durum wheat root (Triticum turgidum), cadmium treatment induced contigs having the highest similarity to Arabidopsis TAT1 and TAT2 by 20 -50 and 3 times, respectively (62). Notably, TAT2 expression is 40 times higher than TAT1 in WT under normal light conditions (Fig. 2, C and D). Because TAT2 has ϳ80 times lower V max toward Tyr than TAT1 (45), the high TAT2 expression may partly compensate for the lower catalytic activity of TAT2 than TAT1. Taken together, both in vitro and in vivo characterizations of Arabidopsis TAT1 and TAT2 indicate that TAT1 is the major TAT responsible for both tocopherol biosynthesis and Tyr degradation, whereas TAT2 partially contributes to tocopherol biosynthesis in stress conditions that require high levels of tocopherol production.
Previous biochemical enzyme characterization showed that TAT enzymes have broad substrate specificity and can use amino acid donors and keto acid acceptors other than Tyr or HPP, respectively (45,54,55). Despite the drastic accumulation of Tyr in genotypes carrying the tat1 mutation, relatively little or no genotypic differences were observed in the levels of other amino acids before and after the dark treatment ( Fig. S4 and Table S1), which include phenylalanine and methionine that can be produced by the reactions catalyzed by TAT1 (45). Potential effects of the tat1 mutation on other amino acids are likely either too small to be detected or compensated by other mechanisms of amino acid homeostasis (e.g. regulation of de novo biosynthesis, protein degradation).
A prior subcellular localization study using GFP fusion proteins showed that Arabidopsis TAT1 and TAT2 are nonplastidic (45). The TAT1 ortholog in Petunia hybrida, phenylpyruvate aminotransferase (PhPPY-AT), was also shown to be localized outside of the plastids (63). In this study, subcellular fractionation of WT, tat1, and tat2 mutants demonstrated that the lack of either TAT1 or TAT2 reduces TAT activity in the cytosolic fraction, but not in the plastid fraction (Fig. 3), further providing genetic evidence that both TAT1 and TAT2 are

Plant tyrosine transaminases function in tyrosine metabolism
localized outside of the plastids. The TAT activity detected in the plastic fraction (Fig. 3) (45) was not affected by knocking out TAT1 or TAT2 (Fig. 3). Also, the cytosol fraction of the tat1 tat2 double mutants still retained more than 70% of TAT activity (Fig. 3). Thus, Arabidopsis must have additional aminotransferases with TAT activity both within and outside the plastids.

Tocopherol biosynthesis from Tyr via TAT1 and HPPD during senescence
Tocopherols are lipid-soluble antioxidants that accumulate during senescence (64 -66). Whereas the phytol tails of tocopherols are mainly provided from that of chlorophylls that are actively degraded during senescence (67,68), the polar headgroup of tocopherols is derived from HPP, likely converted from Tyr by some TATs (Fig. 1). 13 C-Labeled ␥-tocopherol and ␣-tocopherol levels increased upon [ 13 C]Tyr feeding to darktreated Arabidopsis leaves as expected, but to a much lesser degree in the tat1 mutants than in WT or tat2 mutants (Fig. 7). Thus, TAT1 is a major player in metabolizing Tyr into tocopherols also during dark senescence. Our current genetic data are consistent with a prior observation that TAT activity and the TAT1 gene were up-regulated together with tocopherol accumulation upon senescence in Arabidopsis (66). TAT1 homologs in Salvia miltiorrhiza are also highly responsive upon abiotic and biotic stress, such as UV-B, methyl jasmonate, and silver ion (69).
The Arabidopsis co-expression database (ATTED-II) shows that the TAT1 gene is strongly co-expressed with the gene encoding HPPD (At1g06570 (45, 70)), which converts HPP into homogentisate (Fig. 1). TAT and HPPD homologs are also coordinately regulated in other plants under senescence and stress conditions; HPPD expression was up-regulated during senescence and ripening in mango (Mangifera indica L.) together with tocopherol accumulation, and its promoter was shown to be responsive to ethylene (71), a critical plant hormone coordinating these processes (72)(73)(74). Similarly, HPPD is up-regulated under senescence and oxidative stress in barley (75,76) and rice (77). In addition to their coordinated expression, TAT1 and HPPD are localized in the same subcellular compartment, the cytosol (45,46,78), although exceptions exist for HPPD in some plants (46,79,80). Thus, at least in Arabidopsis, TAT1 and HPPD function together in the cytosol during senescence to convert Tyr into homogentisate, which either undergoes the Tyr degradation pathway in the cytosol/mitochondria or is used to synthesize isoprenyl quinone derivatives (e.g. tocopherols) within the plastids (Fig. 1).
Arabidopsis has relatively simple Tyr metabolic pathways unlike other plants, such as the order Caryophyllales (e.g. beets) and Ranunculales (e.g. opium poppy), producing diverse lineage-specific Tyr-derived specialized metabolites-betalain pigments and benzylisoquinoline alkaloids, respectively (81,82). A very recent study, however, indentified Arabidopsis HPP reductases (HPPRs) that can convert HPP, partly provided by TAT1, into 4-hydroxyphenyllactic acid (83). Although these HPPR genes are not co-expressed with TAT1, unlike HPPD, and the physiological roles of HPPRs and 4-hydroxyphenyllactic acid are currently unknown, a complex Tyr metabolic network appears to exist in the downstream of TATs that involves different subcellular compartments, even in Arabidopsis. Future studies can explore how different competing pathways of the Tyr metabolic network are coordinately regulated in different plant species with various specialized metabolic pathways.

In planta function of TAT1 in Tyr degradation
Microarray analysis of the Arabidopsis transcriptome upon 5-day continuous dark treatment shows that all genes involved in the Tyr degradation pathway, HPPD (At1g06570), HGO (At5g54080), MAAI (At2g02390), and FAH (At1g12050), as well as TAT1 (but not TAT2), are at least 2-fold up-regulated upon the dark treatment (56). Thus, besides being involved in tocopherol biosynthesis (as discussed above), TAT1 also plays a role in Tyr degradation in plants. The current study further provides biochemical data that support this notion; 13 C-labeled fumarate and succinate, intermediates of the TCA cycle, increased linearly upon [ 13 C]Tyr feeding to Arabidopsis leaves in the dark (Fig. 7, D and E). Whereas tat2 mutants showed very similar labeling patterns to WT, both tat1 mutants showed very little 13 C incorporation into fumarate and succinate (Fig. 7, D and E), despite the fact that all genotypes showed similar 13 C incorporation to the endogenous Tyr (Fig. 7A). Thus, TAT1, but not TAT2, substantially contributes to the catabolism of Tyr into the TCA cycle.
The observed reductions in 13 C incorporation due to tat1 were similar for ␣-tocopherol and fumarate (Fig. 7, C versus D and E), suggesting that TAT1 likely contributes equally to tocopherol biosynthesis and Tyr degradation. The result also implies that overall availability of homogentisate determines the activity of these two pathways that compete for homogentisate and occur within and outside the plastids, respectively (Fig. 1). Thus, this finding raises an additional question about how these two pathways derived from homogentisate are regulated in planta. Loss of function of HGO, which is localized in the cytosol, led to increased accumulation of tocopherols and tocotrienols in soybean (47), suggesting that blocking the Tyr degradation pathway can redirect carbon flux toward tocochromanol biosynthesis. To utilize homogentisate in plastidic tocochromanol biosynthesis, homogentisate has to be imported to the plastids. Thus, some of the outstanding questions for future studies are to experimentally determine the subcellular localization(s) of the remaining Tyr catabolic enzymes (i.e. MAAI and FAH) and to define the contribution and identity of the plastid envelope-localized homogentisate transporter(s).

Physiological and metabolic roles of Tyr degradation during senescence in plants
Senescence is a developmentally and environmentally regulated process, which requires extensive energy and metabolic reorganization (84 -86). One of the major processes that take place during senescence is the turnover of proteins through various mechanisms, including autophagy (87,88), the ubiquitin-proteasome pathway (89), and organelle-localized various protease systems (90, 91). Protein catabolism is followed by amino acid degradation processes, whose biochemical pathways have not been fully defined, and physiological functions and regulations in plants are under active investigation (15,17). Because Tyr is one of the most energetic of the 20 proteinogenic amino acids (15,92), the regulation of Tyr degradation and its coordination with other metabolic pathways are likely critical to maintain energy, carbon, and nitrogen resources during senescence. In this study, the genetic disruption of the first step of Tyr degradation catalyzed by TAT1 led to profound impacts on both physiology and metabolism of Arabidopsis plants under prolonged dark treatment: tat1 showed accelerated chlorosis and lower respiration rate and sucrose contents than WT (Fig. 5), the degrees of which correlated with those of TAT1 down-regulation (Fig. 2C) and Tyr accumulation (Fig. 5D and Fig. S4). None of the tat2 mutations significantly impacted these responses in either WT or tat1 background (Fig. 5), consistent with the minor effects of tat2 in the conversion of Tyr into fumarate and succinate shown by the 13 C-labeling experiment (Fig. 7, D and E). Thus, degradation of Tyr mediated by TAT1 plays a critical role during the process of darkinduced senescence in Arabidopsis.

Plant tyrosine transaminases function in tyrosine metabolism
The accelerated tat1 senescence phenotypes were rescued by external supply of sucrose to the media, even though elevated accumulation of Tyr was still maintained (Fig. 5, C and  D). This result supports the hypothesis that reduced carbon and/or energy supply in tat1 through Tyr degradation is likely responsible for the tat1 senescence phenotypes (Fig.  5), including reduced respiration rate (Fig. 6). The cell death phenotype of the Arabidopsis fah mutant in short-day conditions (48) could also be rescued by sucrose supplementation to the media (93), suggesting that the blockage of Tyr catabolism at a later step of TAT also results in a carbon depletion phenotype, although potential accumulation of toxic succinylacetoacetate derived from fumarylacetoacetate complicated the interpretation of the fah cell death phenotype (48). Previous studies showed that degradation of branched-chain amino acids and lysine serve as alternative electron donors to maintain respiration during dark-induced senescence through the electron-transfer flavoprotein (ETF) and electron-transfer flavoprotein:ubiquinone oxidoreductase (ETFQO) system in Arabidopsis (16 -23). The loss-of-function mutants of these processes also showed enhanced susceptibility to drought stress (94). Interestingly, the etf and etfqo mutants, which cannot transfer electrons derived from amino acid oxidation to ubiquinone, accumulate aromatic amino acids, including Tyr, besides branched-chain amino acids and lysine (21,22). The tat1 mutant has less severe phenotypes than the etf and etfqo mutants (Fig. 5) (21, 22), similar to mutants specifically defective in branched-chain amino acids and lysine catabolism (18,20,23). Therefore, degradation of Tyr may also provide reduced electrons to cellular respiration, in addition to providing TCA cycle substrates (e.g. fumarate), during dark-induced senescence. The tat mutants and other amino acid catabolic mutants (18, 20 -23, 95, 96) will now allow us to investigate relative contributions and potential coordination of different amino acid degradation pathways to metabolic reorganization and maintenance of central carbon metabolism during senescence and under various stress conditions.

Plant materials and growth conditions
After stratifying the seeds at 4°C for 3 days, Arabidopsis thaliana plants of Columbia-0 WT and tat mutants were grown under a 12/12-h ϳ100-E light/dark cycle with 85% air humidity in 6 ϫ 6-cm square pots with the soil (Propagation Mix, Sun Gro, Agawam, MA) supplied with 1ϫ Hoagland solution until 4 weeks old, unless otherwise indicated. Individual pots of different genotypes were randomized in each tray. The same light condition was also used to grow WT and tat mutant seedlings on the agar until 2 weeks old. Seeds were vapor-sterilized using the chlorine gas method (97) and plated on 0.8% (w/v) phytoagar (RPI, Mount Prospect, IL) containing half-strength Murashige and Skoog basal salt mixture (RPI) with 0, 0.05, or 1% (w/v) sucrose supplementation and then stratified for 3 days before being transferred to the above light regime.
For dark treatments, WT and tat mutants were grown on soil for 4 weeks under the above growth condition and transferred at the end of the light cycle to complete darkness in a closed chamber with no illumination, with water being supplied from the bottom of pots. After 8 days of the dark treatment, plants were recovered under ϳ40-E light conditions for 4 days. The same dark treatment was also applied to agar-grown 2-weekold WT and tat mutant plants for 18 days.
High-intensity light treatments were performed according to (98). Four-week-old plants were moved to ϳ800-E from ϳ100-E light intensity and kept under continuous high-light conditions for 48 h, with water being supplied from the bottom of pots.

Isolation of tat1 and tat2 single mutants and tat1 tat2 double mutants
The seeds of T-DNA insertion lines of TAT1 (At5g53970), tat1-1 (SALK_045398) and tat1-2 (SALK_141402C), and TAT2 (At5g36160), tat2-1 (SALK_099352C) and tat2-2 (SALK_052382C), in the WT background Col-0 were obtained from the Arabidopsis Biological Resource Center. Genomic DNAs were extracted from 1-week-old plants of WT and tat mutants as described (99). Homozygous T-DNA insertion lines of tat1 and tat2 were identified using PCR on the genomic DNAs with the gene-specific forward primers 5Ј-AGTTGCA-CCGAAACACTCAAC-3Ј and 5Ј-CCATAAGCCCTACCAA-ACCA-3Ј, respectively, and the reverse primer 5Ј-ATTTTGC-CGATTTCGGAAC-3Ј targeting the first 310 bp of T-DNA insertion. Amplified DNA fragments were also subjected to DNA sequencing to determine the exact locations of T-DNA insertions. Double mutants of tat1-1 tat2-1 and tat1-2 tat2-2 were generated by crossing corresponding single mutants and isolating double homozygous mutants through genotyping of the F 2 and F 3 populations.

Quantification of TAT1 and TAT2 expression in WT and tat mutants
Total RNA was extracted from 2-week-old WT and tat mutants as described (100). cDNA samples were reverse-transcribed from the RNA using a cDNA RT kit (Applied Biosystems) according to the manufacturer's instructions.

Quantification of enzyme activities in crude extracts, cytosol, and plastid fractions
Subcellular fractionations and TAT activity assays of crude, cytosol, and plastid fractions of WT and tat mutant leaves were performed as described (45). Whole rosette leaves were harvested and processed as described (103) with modification not to include the mitochondrial fractions in the assay. One milliliter of crude, cytosol, or plastid fractions was desalted by a 3-ml bed volume of home-packed Sephadex G50 column (GE Healthcare). NiR and PEPC activity assays were performed as described in Ref. 104 and in Refs. 105 and 106 as the marker activity assays for plastid and cytosol fractions, respectively. The TAT activity assays of WT and tat mutants were carried out in the reaction mixtures containing 100 mM sodium phosphate (pH 8), 200 M PLP, 1 mM Tyr, 1 mM ␣-ketoglutarate or phenylpyruvate, and the crude, cytosol, or plastid fraction. The reaction was initiated by adding an enzyme fraction. After incubation at 30°C for 45 min, the reactions were terminated by adding methanol to 66% (v/v) final concentration for HPLCbased detection. The reaction product, glutamate or phenylalanine, was quantified as o-phthaldialdehyde derivatives using HPLC (Agilent 1260 equipped with Eclipse Plus XDB-C18 column (107)) with a 30-min linear gradient of 10 -70% methanol in 0.1% (v/v) ammonium acetate (pH 6.8) at a flow rate of 0.5 ml/min. Standard curves were generated using respective authentic standards for quantification.

[ 13 C]Tyr feeding experiments
For feeding experiments, seventh and eighth leaves of 4-week-old WT, tat1-1, tat1-2, tat2-1, and tat2-2 were harvested, and five leaf disks were obtained from five different leaves of each genotype using a 5-mm cork borer and stacked together to make one biological replicate. Each replicate was placed in each well of a 96-well plate (VWR, Radnor, PA), acclimated in the dark with 92 l of water for 1 h, and then supplemented with 8 l of 0.1 N NaOH containing 400 nmol of [ 13 C 15 N]Tyr ( 13 C 9 , 15 N; Sigma-Aldrich) or nonlabeled Tyr (0.1 M NaOH was to help Tyr to dissolve in the solution). Samples were collected at 0, 1, 2, 3, 4, and 6 h after [ 13 C 15 N]Tyr or nonlabeled Tyr was added.

Extraction of metabolites from Arabidopsis leaves
To quantify metabolite (e.g. Tyr and tocopherols) levels under normal or high-light conditions, several seventh or eighth leaves of 4-week-old plants were punched in the middle by a 5-mm cork borer to obtain ϳ30 mg of tissue as one biological replicate. Leaf disks were collected in a 1.5-ml centrifuge tube containing 400 l of extraction solution (methanol/chloroform at 2:1, 100 M norvaline and 100 M docosanol as recovery standards for polar and nonpolar phase, 100 g/ml butylated hydroxytoluene to avoid oxidation of tocopherols). Metabolites were then extracted by adding 300 l of water and 125 l of chloroform, followed by vigorous vortexing for 5 min and centrifugation at 20,000 ϫ g for 5 min for phase separation. The upper polar phase of 400 l was transferred to a new centrifuge tube and dried under a benchtop speed vacuum (Labconco, Kansas City, MO). The dried polar phase was resuspended in 200 l of methanol containing 100 M 4-chlorobenzoic acid as another recovery standard. The lower nonpolar phase of 200 l was transferred to a new centrifuge tube, also dried down in the benchtop speed vacuum (Labconco), and resuspended in 1 ml of methanol.

Targeted metabolites and quantification of 13 C label incorporation by GC-MS
Half of the nonpolar extract was used for chlorophyll content measurement as described (108). The other half was used for tocopherol quantification by drying down in a 200-l volume glass insert and resuspended in 40 l of pyridine (Acros Organics) by sonication for 10 min and then mixed with N-methyl-N-(TMS)trifluoroacetamide with 1% trimethylchlorosilane (Cerilliant, Round Rock, TX) before incubation at 50°C for 1 h. The polar extract was split into two aliquots in the 200-l volume glass insert each and dried down as above. One of them was used for sucrose and glucose analyses by being resuspended in 40 l of pyridine containing 15 mg/ml methoxyamine, followed by the N-methyl-N-(TMS)trifluoroacetamide derivatization (109) as described above for tocopherols. The other half of the polar sample was used for quantification of Tyr, fumarate, and succinate contents by being resuspended simply in pyridine and then derivatized in 40 l of N-tert-butyldimethylsilyl-Nmethyltrifluoroacetamide with 1% tert-butyldimethylchlorosilane (Cerilliant) (110) followed by incubation at 50°C for 1 h.
One microliter of the derivatized samples was injected into GC-MS (ISQ-LT, Thermo Scientific, MA) equipped with a 5% phenyl phase column (TG-5MS, Thermo Scientific) by 1:10 split mode with a helium flow rate of 1.2 ml/min. The inlet temperature was set at 260°C. For the derivatized polar phase metabolites, the oven program began at 70°C and was held for Plant tyrosine transaminases function in tyrosine metabolism 2 min, before being ramped to 300°C by 5°C/min, and then held at 300°C for 10 min. For the derivatized nonpolar metabolites, the oven program began at 70°C and was held for 2 min, before being ramped to 300°C by 15°C/min and then held at 300°C for 10 min. Ion source and MS transfer line temperatures were both set to 300°C, and the electron energy was 70 eV. Both data of selected-ion monitoring (SIM) and total ion chromatogram were collected during the retention time of 9.3-50 min for TBDMS derivatives, 6 -50 min for TMS derivatives of polar phase metabolites, and 7-27.5 min for TMS derivatives of nonpolar phase metabolites.

Plant respiration rate measurement
For respiration measurements of the day 0 time point, WT and tat mutants were grown under normal light conditions (ϳ100 E) for 4 weeks, and the respiration rate of the seventh and eighth leaf of each biological replicate was measured at 6 p.m. after 1 h of dark adaptation using an LI-6400XT photosynthesis system equipped with the 6400-40 leaf chamber (Lincoln, NE). Afterward, plants were transferred to continuous darkness, and the respiration rates were measured at 6 p.m. in the dark after day 1, 3, 5, 7 and 10 of the dark treatment. Sigmoid curve fitting was performed using the gnuplot-based IC 50