1 Introduction

Although N fertilization is one of the major components in the success of the green revolution and plant growth and productivity, the basis of the profound remodeling induced by nitrogen in nitrogen (N), sulfur (S), and carbon (C) allocation between shoots and roots remains always elusive (Good et al. 2004; Hirel et al. 2007; Masclaux-Daubresse et al. 2010). Thus, it is well known that N deficiency induces proliferation of roots at the expense of shoots whereas excess of N induced the opposite situation (Scheible et al. 1997). However, the basis of these physiological and molecular switches is still not understood (Scheible et al. 1997; Le Ny et al. 2013). For example, conventional approaches to study root functional and structural responses based on nitrate availability under homogeneous and heterogeneous N supply conditions are hampered by the fact that there is a very strong coordination and integration of absorption and assimilation of N with C fixation at primary metabolism level (Wang et al. 2003; Scheible et al. 2004; Bi et al. 2007; Nero et al. 2009; Bussell et al. 2013). NO3 − uptake and its utilization by plants to match the N demand are sensitively regulated by internal N-sensing mechanisms that control functional and structural responses via a myriad of signaling molecules. Although many candidate systems for N detection have been proposed such as TOR (Target Of Rapamycin) and GCN2 (General Control Non-derepressible 2) kinases as well as GLR (Glutamate Like Receptor) receptors, most approaches for testing these N sensory systems ignore the involvement of hormones such as ethylene and IAA (Lam et al. 2006; Castilho et al. 2014; Xiong and Sheen 2014). However, compared to nitrate, ethylene and IAA play a major role in root growth and development because they act at very low concentrations (nM and μM) and at very short-term durations (min to hours) on root and shoot development (Robinson 2005; Le Deunff et al. 2016). Moreover, their interactions are mostly involved in the primary root and lateral root (LR) development as well as root hair growth (Stepanova et al. 2007; Muday et al. 2012; Hu et al. 2017). Interestingly, plant hormones such as ethylene, indole-3-acetic acid (IAA), and phenylacetic acid (PAA) are derived from L-methionine (Met), L-tryptophan (Trp), and L-phenylalanine (Phe) amino acids, respectively (Giovanelli et al. 1985, Stepanova et al. 2008; Sugawara et al. 2015). This confers to these plant hormones a specific place in N, S, and C primary metabolism in relation with growth during the plant life cycle. Among these hormones, ethylene occupies a special place since it is a volatile molecule involved in alarm, stress, and senescence (Abeles et al. 1992), but is also a powerful natural anesthetic (Baluska et al. 2016). Moreover, the first precursor of ethylene, methionine, is an essential amino acid at the intersection of the metabolism of N, S, and C. Its second precursor, S-adenosyl-L-methionine (AdoMet), is a fundamental cofactor involved in plant transmethylation reactions that is used by folates (THF) as a relay molecule to extend their capacities for methylation (Cossins 2000; Lu 2000; Gerelova et al. 2017). Finally, the direct precursor of ethylene, ACC, is a non-proteinogenic amino acid whose production is exquisitely regulated (Wang et al. 2002; Stepanova et al. 2007) and has a potential role as an agonist on plant glutamate receptors (Le Deunff and Lecourt 2016). Finally, ethylene acts on specific receptors and some components of its signaling pathway such as the transcription factors ethylene insensitive 3 (EIN3) and its homologue EIN3-LIKE 1 (EIL1) play a major role in the regulation of auxin biosynthetic genes (He et al. 2011) and responses to glucose through the hexokinase1 (HXK1) C sensor (Yanagisawa et al. 2003; Yoo et al. 2008).

So far, emphasis has been placed on the regulation of biosynthesis and signaling cascades of both ethylene and auxins and their interactions on root and shoots growth (Muday et al. 2012; Hu et al. 2017). However, a major challenge for the next few years is to address the precise molecular mechanisms underlying the complex signaling network that governs regulation of the metabolic interconnections between mineral nutrition, amino acid biosynthesis, and the production of plant growth regulators. In this aim, an alternative approach consists of stating that the biosynthesis of hormones such as ethylene and IAA is highly regulated through the aminotransferases network in primary metabolism (Le Deunff et al. 2016). Indeed, aminotransferases are the key enzymes that ensure N, S, and C shuttlings but are also essential in the biosynthesis of ethylene and auxin precursors Met, Trp, and Phe, and in the production of major cofactors such as vitamins B9 (folates) and B6.

This review focuses on the metabolic interconnections between ethylene biosynthesis and nitrogen metabolism and, to a lesser extent, sulfur metabolism. Because the focus is mainly on the network of PLP-dependent aminotransferases during ethylene and methionine biosynthesis, ethylene signaling is less discussed.

2 A Highly Regulated Pathway for the Biosynthesis of Methionine and S-Adenosylmethionine (AdoMet): The Ethylene Precursors

In plants, AA biosynthetic pathways can be organized depending on the origin of their carbon skeleton coming from glycolysis, the citric acid cycle, and the oxidative pentose phosphate pathway (Fig. 1; Coruzzi and Last 2000; Stitt et al. 2002). The carbon skeleton of methionine is mainly synthesized via a branch of aspartate pathway, but its sulfur atom is derived from cysteine and its methyl group from the β-carbon of serine. Other amino acids such as Thr, Ile, and Lys also originate from this aspartate-derived metabolic pathway (Azevedo et al. 2006; Jander and Joshi 2009). Due to the importance in human diets of these four amino acids (Met, Thr, Ile, and Lys), the biochemical regulatory mechanisms involved in the biosynthesis of aspartate-derived amino acids have been intensively studied. Thus, allosteric regulations of the branch-point enzymes by the pathway products have been determined by genetic and biochemical approaches and are summarized in Fig. 2. More details and explanations can be found in excellent reviews (Coruzzi and Last 2000; Azevedo et al. 2006; Jander and Joshi 2009; Galili 2011). Here we only focus on methionine biosynthesis that competes with threonine biosynthesis because these two branches of the aspartate pathway use initially a common substrate: O-phospho-homoserine (OPHS). It is now clearly established that in the methionine branch of the aspartate pathway all enzymes are present in the chloroplast. Therefore, chloroplasts are autonomous for de novo Met biosynthesis (Mills et al. 1980; Ravanel et al. 2004). In this pathway, OPHS is converted to cystathionine via the cystathionine γ-synthase (CgS) whereas in the threonine branch pathway OPHS is converted to L-threonine via threonine synthase (TS, Fig. 2). Cystathionine γ-synthase and threonine synthase are rate-limiting steps in Met and Thr biosynthesis (Ravanel et al. 1998; Hesse et al. 2004). The activity of threonine synthase is allosterically activated by S-adenosyl-L-methionine (AdoMet), the end product of the methionine pathway (Curien et al. 1996, 1998; Laber et al. 1999; Zeh et al. 2002). In contrast, cystathionine γ-synthase is negatively regulated at posttranscriptional level by AdoMet that binds to the N-terminus of nascent cystathionine γ-synthase during translation (Fig. 2), reduces mRNA stability, and stops its translational activity (Chiba et al. 1999). Therefore, AdoMet levels exerted a feedback control on its own production that in turn partly modulates Met biosynthesis. AdoMet acts also in concert with lysine as an allosteric inhibitor on the activity of three monofunctional aspartate kinases (AK) involved in the first step of the aspartate-derived amino acid pathway (Fig. 2). Cysteine, used as substrate by CgS enzyme for providing the sulfur atom to methionine, also regulates the pathway by stimulating the aspartate kinase with homoserine dehydrogenase (AK-HSDH) activity (Jander and Joshi 2009). Indeed, allosteric co-regulations of CgS and AK-HSDH1 activities by AdoMet and cysteine, respectively, direct the flux from aspartate into Thr, Ile, and Met production.

Fig. 1
figure 1

Simplified overview of amino acid biosynthesis in plants. Asterisk indicates aminotransferases targeted by AVG and number represents PLP-dependent aminotransferases in the different biosynthetic pathways: 1, tryptophan aminotransferase of Arabidopsis (TAA); 2, 3-phosphoserine aminotransferase (PSAT) and ala-hydroxypyruvate aminotransferase (HA-AT); 3, alanine-hydroxypyruvate aminotransferase (AH-AT); 4, Alanine aminotransferase (AlaAT); 5, β-alanine-pyruvate aminotransferase (BAPAT); 6, branched-chain aminotransferase (BCAT); 7 and 8, eukaryotic aspartate aminotransferase (ET-AAT) and prokaryotic type aspartate/prephenate aminotransferase (PT-AAT/PAT); 9, asparagine synthase (ASN); 10, cystathionine β-lyase (CBL); 11, 1-aminocyclopropane-1-carboxylate synthase (ACS); 12, glutamate decarboxylase (GAD); 13, Gaba transaminases using either α-ketoglutarate (GABA-TK) or pyruvate (GABA-TP); 14, histidinol-phosphate aminotransferase (HPA). Abbreviations used for metabolites are: 5′-PRFAR phosphoribosyl formimino-5-aminoimidazole-′-carboxamide ribonucleotide, IGP imidazoleglycerolphosphate, IAP imidazoleacetolphosphate, AICAR 5-aminoimidazole-4-carboxamideribonucleotide, α-KG α-ketoglutarate, IAA indole-3-acetic acid, PAA phenylacetic acid, SAM S-Adenosylmethionine, PEPCK phosphoenolpyruvate carboxykinase. Green boxes indicate plastidic location of enzymes. ? indicates putative degradation pathway of histidine by tryptophan aminotransferase according to Brunke et al. (2014) in C. glabrata and S. cerevisiae (adapted from Coruzzi and Last 2000)

Full size image
Fig. 2
figure 2

Aspartate-derived amino acid pathway and allosteric regulations of the pathway by final products or intermediates. Positive allosteric regulation is indicated by a dashed arrow line and inhibition by a dashed T line. Enzymes submitted to allosteric regulation are indicated in red. AAT aspartate aminotransferase, AAT/PAT aspartate/prephenate aminotransferase, AK aspartate kinase, AK-HDKS aspartate kinase with homoserine dehydrogenase (HDKS) activity, DHDPS dihydrodipicolinate synthase, TS threonine synthase, CgS cystathionine γ-synthase, SAMS S-adenosylmethionine synthase (adapted from Coruzzi 2003; Jander and Joshi 2009; Sauter et al. 2013)

Full size image

2.1 Glutamine Involvement in the Methionine Pathway Is Hidden Behind the PLP Coenzyme Used by Aminotransferases

The three steps of methionine biosynthesis in the plastids from O-phospho-homoserine require the action of cystathionine γ-synthase (CgS), cystathionine β-lyase (CbL), and methionine synthase (MS), respectively (Fig. 3). Cystathionine γ-synthase and β-lyase are two enzymes that belong to the γ- and β-lyase subfamilies of the α family of aminotransferases (Christen and Mehta 2001). For their catalytic activity, they need pyridoxal-5′-phosphate (PLP), a derivative form of vitamin B6 that is a versatile coenzyme. De novo PLP biosynthesis in the cytosol is catalyzed by pyridoxine synthase enzymes (PDX1 and PDX2) from glutamine, produced by the GS/GOGAT cycle after reduction of nitrate and glyceraldehyde 3-phosphate (G3P) and ribose 5-phosphate (5RP) provided by the glycolysis and pentose phosphate pathways, respectively (Tambasco-Studart et al. 2005; Fitzpatrick 2011; Colinas et al. 2016). PLP is also synthesized in chloroplasts from pyridoxamine 5′-phosphate (PMP) or pyridoxine 5′-phosphate (PNP) by pyridoxamine 5′-phosphate oxidase enzyme (Fitzpatrick 2011). It is not known how de novo PLP synthesis in cytosol modulates the PLP pool in the plastid and which plastidial carriers are involved in the transport of different forms of vitamin B6 (Gerdes et al. 2012). Two olefinic compounds, DL-propargylglycine (PAG) and L-aminoethoxyvinylglycine (AVG), that act as suicide inhibitors of PLP-enzymes of the α family of aminotransferases (Lieberman 1979; Satoh and Yang 1989a) can inhibit in vitro and in vivo activities of CgS and CbL enzymes, respectively (Ravanel et al. 1998).

Fig. 3
figure 3

Interconnections in biosynthetic pathways of methionine, polyamines, ethylene, and AdoMet and methionine salvage cycles in plants. Asterisk indicates aminotransferases targeted by AVG: 1, eukaryotic aspartate aminotransferase (ET-AAT) and prokaryotic type aspartate/prephenate aminotransferase (PT-AAT/PAT); 2, O-acetylserine(thiol)lyase (OASTL); 3, cystathionine γ-lyase (CgL); 4, cystathionine β-lyase (CbL); 5, aminotransferase SAV1/ISS1; 6, 1-aminocyclo-propane-1-carboxylate synthase (ACS). Abbreviations used for metabolites are: AdoMet S-adenosylmethionine, dcAdoMet decarboxylated S-adenosylmethionine, SAH S-adenosylhomocysteine, MTA 5′-methylthioadénosine, MTR 5′-methylthioribose, MTR-P 5′-methylthioribose-phosphate, MTRu-P 5′-methylthioribulose-phosphate, DHKMP 1,2-dihydro-3-oxo-5-methylthiopentene, KMTB 2-oxo-4-methylthiobutyrate; Green boxes indicate plastidic location of enzymes. ? indicates putative aspartate aminotransferase according to Pommerrenig et al. (2011) (adapted from Ravanel et al. 2004; Pommerrenig et al. 2011; Sauter et al. 2013)

Full size image

2.2 Serine Is Doubly Implicated in Methionine Biosynthesis

Serine is directly involved in the methionine pathway through cysteine biosynthesis (Romero et al. 2014). Cysteine provides a sulfur atom to methionine through the action of cystathionine γ-synthase (CgS) and comes from the reduction and assimilatory pathway of inorganic sulfate (SO4 2−). The biosynthesis of cysteine requires serine as the amino acid skeleton-donor and bisulfide from sulfate reduction as a sulfur donor to form cysteine (Takahashi et al. 2011; Romero et al. 2014). In plants, the reduction of sulfate takes place in the chloroplast and produces sulfide, an inorganic anion of sulfur with the chemical formula S2 −. However, in aqueous solution most sulfide ions are neutralized under the conjugate acid form of bisulfide: SH−. The two consecutive reactions catalyzed by serine acyltransferase (SAT) and O-acetylserine(thiol)lyase (OSATL) produce cysteine (Fig. 3). SAT catalyzes the conversion of serine and acetyl-CoA into O-acetylserine (OAS) and acetyl-CoA-SH whereas OSATL catalyzes the final PLP-dependent conversion of bisulfide and OAS into cysteine and acetate (Bonner et al. 2005). These two enzymes can assemble in a hetero-oligomeric cysteine synthase complex formed by one SAT hexamer and two OASTL dimers requiring four molecules of PLP for functioning (Droux et al. 1998). This complex, first found in bacteria, plays a major role in plants for modulation of cysteine biosynthesis via bisulfide and OAS. Some authors consider that this complex could be a cellular sulfur sensor (Yi et al. 2010; Gerelova et al. 2017).

Serine is also indirectly involved in the last step of methionine biosynthesis through the use of a polyglutamate form of tetrahydrofolate (5-methyl-THF: 5-CH3H4PteGlun with n > 3) as a methyl-group donor (Eichel et al. 1995). Thus, homocysteine (Hcy) methylation to form methionine is catalyzed by methionine synthase (MS). The methyl group given by 5-methyl-THF is provided by the β-carbon of serine after two consecutive reactions catalyzed in the plastids by serine hydroxymethyltransferase (THFT) and 5,10-methylenetetrahydrofolate reductase (MTHFR). THFT enzyme uses serine as a methylene group donor to convert tetrahydrofolate (H4PteGlun) into methylene-THF (5,10-CH2H4PteGlun) and glycine (Fig. 3). Then, methylene-THF is converted to its reduced form 5-methyl-THF (5-CH3H4PteGlun) by MTHFR. The one-carbon group of 5-methyl-THF is used by methionine synthase (MS) to convert Hcy to methionine (Shah and Cossins 1970; Gorelova et al. 2017). Polyglutamate forms of THF are synthetized by folylpolyglutamate synthase (FPGS), an enzyme localized in mitochondria, cytosol, and plastids that catalyzes the attachment of glutamate tail to a THF molecule (Ravanel et al. 2001). In Arabidopsis, mutation of the AtDFB gene encoding FPGS enzyme showed a disruption in primary root growth by an alteration of the quiescent center of root meristem, suggesting that THF metabolism plays a major role in root cell proliferation (Srivastava et al. 2011; Reyes-Hernández et al. 2014). Therefore, serine is doubly required for methionine biosynthesis via the cysteine biosynthesis pathway and production of the polyglutamate form of 5-methyl-THF.

2.3 AdoMet Synthase Is a Fundamental Control Point of the Methionine Pathway

The last step of the methionine pathway is catalyzed by methionine adenosyltransferase (SAM or MAT), a cytosolic enzyme that catalyzes conversion of methionine and ATP into AdoMet. In the Arabidopsis genome, SAM is encoded by three genes (SAM1, SAM2, and SAM3) and their expression has been shown to be highly regulated at transcriptional (Peleman et al. 1989a, b; Boerjan et al. 1994; Chen et al. 2016) and posttranscriptional levels (Mao et al. 2015; Jin et al. 2017) in response to hormones, biotic and abiotic stress.

Recently, it has been shown in Arabidopsis that FERONIA (FER), a plasma membrane receptor-like kinase, may negatively regulate SAM (Mao et al. 2015). FERONIA receptor belongs to the CrRLK (Catharanthus roseus Receptor Like Kinase) family and is involved in the elongation of root hairs and leaf cells (Guo et al. 2009; Deslauriers and Larsen 2010; Duan et al. 2010). FER receptor ligand is the RALF peptide (Rapid Alkalinization Factor) that is able to bind to the FER receptor and trigger phosphorylation of many cytosolic proteins such as H+-ATPase AHA2 involved in cell elongation (Haruta et al. 2014). Surprisingly, fer mutants display a dwarf phenotype and a significant increase in ethylene production and AdoMet contents. The dwarf phenotype of fer mutants was mimicked in transgenic plants over-expressing SAM whereas sam1sam2 double mutant showed a wild-type phenotype suggesting that FER receptor might be involved in inhibition of SAM activity thereby reducing levels of AdoMet and ethylene. Further work is needed to determine mechanisms by which FER receptor inhibits SAM1 and SAM2 activities in responses to RALF peptide hormone and ethylene crosstalk.

SAM isoforms are also regulated at the posttranslational level by CPK28, a calcium-dependent protein kinase, that interacts with phosphorylated forms of SAM1, SAM2, and SAM3 to induce their degradation through the ubiquitin/26S proteasome pathway (Jin et al. 2017). At the biochemical level, cpk8 mutants display an increase in AdoMet contents and SAM protein levels, whereas at the phenotypic level they present short hypocotyls and an increase in lignification caused by ethylene overproduction. Inhibiting ACC synthetase (ACS) activity with AVG treatment could restore the wild-type phenotype. Taken together these results demonstrate that the last step of the methionine biosynthetic pathway is highly regulated at the translational and posttranscriptional levels because AdoMet is a regulatory node molecule involved in plant development through transmethylation reactions, crosstalk between ethylene and polyamine biosynthesis, and stress signaling responses.

3 Because of Its Vital Cellular Functions AdoMet Requires Two Salvage Pathways

In Lemna paucicostata, production of AdoMet consumes 80% of Met produced by the methionine pathway whereas 20% of Met is involved for protein synthesis (Giovanelli et al. 1985). AdoMet is one of the major cofactors used in nature with ATP and ensures vital biochemical functions in plant and animal cells (Lu 2000). Indeed, AdoMet is the methyl group donor during transmethylation reactions of lignins, chlorophylls, proteins, phospholipids, and nucleic acids biosynthesis, the precursor for biosynthesis of ethylene, polyamine, nicotianamine (siderophore), and biotin, and it is also involved in many transsulfuration reactions (Lu 2000). Since AdoMet is the end product of the methionine pathway and one of the major cofactors used in nature, the lack of recycling of the AdoMet methylthio-group and adenosine moiety reduces biosynthesis of ethylene and polyamines by a restriction in sulfur availability (Baur and Yang 1972; Bürstenbinder et al. 2007). From a literature survey, it appears that the cellular AdoMet homeostasis is controlled at several levels through: (1) allosteric and transcriptional regulations of the amino acid pathway derived from aspartate (Fig. 2), (2) translational and posttranscriptional regulation of methionine adenosyltransferase, (3) transmethylation reactions, (4) regeneration via the methyl cycle and Yang cycle, and (5) elevated shuttling of AdoMet into several biosynthetic pathway such as ethylene and polyamines (Bürstenbinder et al. 2007; Van de Poel et al. 2012). Because AdoMet is synthesized in the cytosol, its availability for plastidial reactions therefore requires a specific carrier, which is another regulatory point for AdoMet homeostasis (Fig. 3). In this respect, in Arabidopsis a chloroplastic AdoMet carrier was able to catalyze the unidirectional import of cytosolic AdoMet as well as the exchange between a cytosolic AdoMet and a chloroplastic AdoMet or S-adenosylhomocysteine (Ravanel et al. 2004). Therefore, because of its cellular importance, AdoMet requires two salvage pathways to regenerate its adenosine moiety, methylthio-group and methyl group used in methylation reactions.

3.1 A Methyl Cycle Serves to Regenerate the Methyl Group Issued from AdoMet

The S-adenosylhomocysteine (AdoHcy) issued from methylation reactions with AdoMet can be recycled to methionine from the homocysteine (Hcy) intermediate through the methyl or AdoMet cycle (Ravanel et al. 1998). AdoHcy issued from transmethylation reactions is converted into Hcy by AdoHcy hydrolase and Hcy is then regenerated to methionine by Met synthase (Fig. 3). Conversion of homocysteine into methionine can occur in mitochondria, chloroplasts, and cytosol (Clandinin and Cossins 1974; Shah and Cossins 1970; Eichel et al. 1995). The last two steps of this cycle are assumed to occur mainly in the cytosol (Hanson and Roje 2001; Roje et al. 2002a), since an AdoMet carrier is involved to ensure the exchanges between cytosolic and chloroplastic AdoMet or SAH (Ravanel et al. 2004). Met synthase catalyzes methionine formation and simultaneously allows regeneration of the methyl group of AdoMet via the use of a folic compound: 5-CH3H4PteGlu(n) (Cossins 1987; Roje et al. 2002b). The carbon of the methyl group provided by the folic compound originates from the β-carbon of serine (Fig. 3). The methyl cycle is fine-tuned by SAH levels, inhibiting folate compound biosynthesis through MTHFR (Jencks and Mathews 1987; Roje et al. 2002b) and by high levels of AdoMet that reduce methionine production (Crider et al. 2012). In summary, the methyl-group transfer catalyzed by AdoMet in methylation reactions is directly provided from serine through folates (vitamins B9) that serve as donors or acceptors in one-carbon (C1 metabolism) transfer reactions. In other words, AdoMet is used by folates as a relay molecule to extend their capacities for methylation, probably because the methylthio-group of AdoMet is more reactive (Fig. 3). Therefore, the biosynthesis of methionine strongly depends on serine metabolism. In this respect, recent studies have demonstrated that the mutants of genes involved in the plastidial phosphorylated pathway of serine biosynthesis (PPSB) displayed arrested root development and exhibited a strong impairment of carbon and nitrogen metabolisms (Muños-Berthomeu et al. 2009, 2010; Cascales-Miñana et al. 2013). Furthermore, at the root morphological level, recent results in Arabidopsis showed that folic acid is essential in root organogenesis via root cell proliferation in meristems (Reyes-Hernández et al. 2014) and maturation of lateral root primordia upstream of auxin and independently of TOR kinase signaling (Ayala-Rodriguez et al. 2017; Li et al. 2017).

3.2 Ethylene and Polyamines Are By-Products Resulting from the Recycling of the Methylthio-Group of MTA Issued from AdoMet

5′-Methylthioadenosine (MTA), the by-product of the ACC synthase (ACS) and spermidine synthase (SPDS) reactions, is salvaged for the regeneration of the methylthio-group and adenine moiety through the methionine cycle, also known as the Yang or MTA cycle or methionine salvage pathway (Miyazaki and Yang 1987; Sekowska et al. 2004; Pommerrenig et al. 2011; Sauter et al. 2013). In plants, this cycle has been recently revised (Pommerrenig et al. 2011), and it is now composed of six reaction steps (Fig. 3) that involve successively: (1) the conversion of MTA into 5-methylthioribose (MTR) and adenine by MTA nucleosidase (MTN; Adams and Yang 1977; Wang et al. 1982; Rzewuski et al. 2007), (2) the phosphorylation of MTR into 5-methylthioribose-1-phosphate (MTR-1P) by the MTR kinase (MTK; Kushad et al. 1982; Sauter et al. 2004), (3) the isomerization of MTR-1P to 5-methylthioribulose-1-phosphate (MTRu-1P) by MTR-P isomerase (MTI; Pommerrenig et al. 2011), (4) then dehydratase-enolase-phosphatase (DEP) ensures the conversion of MTRu-1P into 1,2-dihydro-3-keto-5-methylthiopentene (DHKMP; Pommerrenig et al. 2011), (5) acireductone dioxygenase (ARD) in the presence of dioxygen catalyzes the conversion of DHKMP into 2-keto-4-methylthio butyrate (KMTB; Sauter et al. 2005; Bürstenbinder et al. 2007), and (6) finally transamination of KMTB to methionine is catalyzed by unknown aminotransferases (Kushad et al. 1983; Pommerrenig et al. 2011). In plants and many microorganisms, different unknown plastidial or cytosolic aspartate aminotransferases (AAT) are presumed to be involved in the last steps of the methionine regeneration pathway (Berger et al. 2003; Sekowska et al. 2004; Pommerrenig et al. 2011).

4 ACS and SAMDC Are Key Enzymes in Ethylene and Polyamine Biosynthetic Pathways and MTA Recycling

The biosynthesis of ethylene and polyamines are linked through the substrate AdoMet, which can be converted into decarboxylated S-adenosyl-L-methionine (dcAdoMet) and 1-aminocyclopropane-1-carboxylic acid (ACC), the ethylene precursor, through the action of S-adenosylmethionine decarboxylase (SAMDC) and 1-aminocyclopropane-1-carboxylate synthase (ACS), respectively (Fig. 3). In several physiological processes, ethylene and polyamines compete for their biosynthesis because these two metabolic pathways use initially a common substrate: AdoMet. Despite several studies that have revealed the competitive interaction between these two pathways during the processes of fruit ripening (Tassoni et al. 2000; Harpaz-Saad et al. 2012), root elongation (Kumar et al. 1996; Locke et al. 2000; Tassoni et al. 2000; Hummel et al. 2002), senescence (Pandey et al. 2000), and fruit or leaf abscission (Saftner 1989; Gil-Amado and Gomez-Jimenez 2012; Harpaz-Saad et al. 2012), little is known about the cross regulations of these two pathways. In fact, the partitioning of the AdoMet flux between these two pathways has never been precisely estimated because of the difficulties in measuring the contents of the different conjugated forms of ACC and polyamines that lead to underestimation of the flux in one or the other of these pathways.

4.1 SAMDC Is the Rate-Limiting Step for Polyamine Biosynthesis

Polyamine homeostasis is tightly regulated and achieved by the modulation of polyamine biosynthesis, conjugation, transport, and catabolism (Tiburcio et al. 2014). In the polyamine pathway AdoMet is converted to decarboxylated S-adenosyl-L-methionine (dcSAM) and MTA via SAMDC (Fig. 3). In Arabidopsis SAMDC genes belong to a small family of four genes that are not functionally redundant. Single and double mutants of SAMDC4 (bud2-1) and SAMDC1 (samdc1) increase root proliferation and growth by alteration of polyamine homeostasis (Kumar et al. 1996; Ge et al. 2006). Knockout of the SMDC4 gene affects growth and development, whereas alteration of the SMDC1 gene displays a more severe phenotype (Ge et al. 2006). The double mutant SMDC4-SMDC1 (bud2-1-samdc1) exhibited embryo lethality, probably through spermidine deficiency (Imai et al. 2004). SAMDC genes are regulated at the transcriptional level and induced by many abiotic stresses such as drought, chilling, heat, hypoxia, ozone, UV, and heavy metals (Alcázar et al. 2006). During tomato fruit ripening, it seems that no competition for AdoMet substrate occurs between SAMDC, ACS, and transmethylation reactions, suggesting that these three pathways can operate simultaneously (Van de Poel et al. 2012). This confirms studies where ethylene production was increased in transgenic plants over-expressing SAMDC (Mehta et al. 2002).

4.2 ACS Aminotransferase Is the Rate-Limiting Step for Ethylene Biosynthesis

In the ethylene pathway, AdoMet is converted by ACC synthase (ACS) into MTA and ACC, a non-proteinogenic amino acid, precursor of ethylene (Figs. 3 and 4). ACS is a PLP-dependent enzyme that belongs to the subgroup I of the α family of aminotransferases (Mehta et al. 1993; Christen and Mehta 2001). This subgroup contains major enzymes involved in carbon, nitrogen, and sulfur assimilation and shuttling such as ACC, aspartate, alanine, tyrosine, and aromatic aminotransferases (Mehta et al. 1993). X-ray structure analysis of the ACS protein revealed that amino acids surrounding the catalytic site share the same function as the aspartate aminotransferase (AAT) counterpart (Capitani et al. 1999; Yamagami et al. 2003). The ACS isoforms form homodimers and heterodimers and the active site of the dimers results from the interaction of the two subunits with each other (Tarun and Theologis 1998; Tsuchisaka and Theologis 2004a). In Arabidopsis, there exist 12 genes encoding ACS but among them only eight genes (ACS2, 4–9, and 11) encode functional ACS (Bleecker and Kende 2000; Yamagami et al. 2003). Indeed, complementation of the E. coli aminotransferase mutant DL39 showed that ACS10 and ACS12 are aminotransferases with broad specificity for aspartate and aromatic amino acids whereas ACS1 forms inactive homodimers and ACS3 was a pseudogene (Tsuchisaka and Theologis 2004a). Based on protein sequences comparison, three subtypes of the eight functional ACS have been defined mainly based on the number of phosphorylation sites in their C-terminal sequence (Chae and Kieber 2005). These sites are involved in posttranslational regulation and modulate the stability and degradation of ACS proteins (Wang et al. 2004; Argueso et al. 2007; Hansen et al. 2009). Thus, the phosphorylation of ACS-type 1 (ACS1, ACS2, ACS6) are controlled by mitogen activated protein kinase (Tatsuki and Mori 2001; Liu and Zhang 2004) while the phosphorylation of ACS-type 2 (ACS4, ACS5, ACS8, ACS9, ACS11) are regulated by casein kinase (Tan and Xue 2014). The ACS-type 3 (ACS7) is a more stable protein because it has no C-terminal phosphorylation site but ubiquitin regulated its degradation via proteasome (Lyzenga et al. 2012).

Fig. 4
figure 4

Interconversion between aminotransferases belonging to the subgroup I of the α family involved in the metabolic pathways of auxins, ethylene, and aromatic amino acids (Phe, Tyr, Trp). 1, phenylpyruvate aminotransferase (PPY-AT); 2, tryptophan aminotransferase of Arabidopsis (TAA/TAR); 3, IPyA/PPA aminotransferase SAV1/ISS1; 4, eukaryotic aspartate aminotransferase (ET-AAT) and prokaryotic type aspartate/prephenate aminotransferase (PT-AAT/PAT). Abbreviations used for metabolite are: KTMB, 2-keto-4-methylthio butyrate; 2-KG, α-ketoglutarate (adapted from Zheng et al. 2013a; Yoo et al. 2013; Le Deunff et al. 2016; Pieck et al. 2015)

Full size image

Although ACS genes are strongly regulated at the transcriptional level depending on many developmental and environmental factors (Van der Straeten et al. 1992; Tsuchisaka and Theologis 2004b), to our knowledge only a few studies have compared the promoter sequences of ACS genes and search for common or specific responsive elements to transcription factors (Rodrigues-Pousada et al. 1993; Wand et al. 2005). As with many aminotransferases, it is likely that transcription of ACS genes is also controlled by more general mechanisms that affect the nutritional and energy status of the cell such as GCN2 and TOR kinase signaling systems (Xiong and Sheen 2015). Briefly, in plants GCN2 (General Control Non-derepressible 2) kinase is a cellular sensing system that allows overcoming amino acid starvation and stress conditions during growth (Hey et al. 2010; Zhang et al. 2003; Lageix et al. 2008; Castilho et al. 2014), whereas TOR (Target Of Rapamycin) kinase is a key regulator conserved in eukaryotic cells that controls nutrients and energy signaling topromote cell proliferation and growth (Loewith and Hall 2011; Xiong and Sheen 2014).

4.3 Cytosolic Levels of ACC Can Be Modulated in Many Ways

ACC produced by activity of the ACS isoforms is a neutral and non-proteinogenic amino acid that displays a structural analogy and molecular mass close to other signaling amino acids molecules such as GABA, α- and β-aminobutyric acids (α-ABA and β-ABA). If ACS is a rate-limiting step in ACC biosynthesis (Fig. 4), there are at least four different ways to modulate the excessive levels of ACC produced by ACS activation.

ACC Is Conjugated Under Different Forms as Many Signal Molecules

The cellular levels of ACC can be decreased by conversion under different conjugated forms. Thus, ACC can be converted in tomato fruits by N-malonyltransferase (AMT) into 1-malonyl-ACC (MACC, Martin et al. 1995) and in Arabidopsis vegetative tissues by JA-aminosynthetase (JAR1, Jasmonic acid resistance 1) into jasmonyl-ACC (JA-ACC; Staswick and Tiryaki 2004). ACC can also be conjugated to the reduced glutathione (GSH). GSH is a very abundant cellular tripeptide (γ-Glu-Cys-Gly) that serves as an essential antioxidant by scavenging reactive oxygen species during plant stresses (Noctor et al. 2011). γ-glutamyltranspeptidase (GGT) converted GSH and ACC into 1-(γ-glutamylamino)-ACC (GACC) in plant tissues (Amrhein et al. 1981).

ACC Can Be Stored in the Vacuole or Transported Long-Distance

Another strategy developed by the plant for buffering an excess of ACC synthesis and ethylene production is ACC storage into the vacuole or long-distance translocation through the xylem (Bradford and Yang 1980; O’Neill 1997; Finlayson et al. 1999) and the phloem (Voesenek et al. 1990). Although ACC compartmentalization and storage into the vacuole have been studied for a long time, the characterization of genes encoding transporters of influx and efflux is not yet achieved (Bouzayen et al. 1991; Tophof et al. 1989; Saftner and Martin 1993). However, it has been recently demonstrated that ACC transport across the plasma membrane is provided by the lysine histidine transporter1 (LHT1). Indeed, neutral amino acids such as alanine and glycine have a competitive effect on the ACC-induced triple response. Likewise, are2 mutants (ACC-resistant 2) defective in LHT1 exhibit a dose-dependent response to ACC (Shin et al. 2015). Therefore, long-distance circulation and short-distance accumulation of ACC are dependent of amino acids transporters that have not been yet completely characterized.

ACC Is Catabolized by a PLP-Dependent Deaminase

A recent study has shown that there exists in the Arabidopsis genome an ACC deaminase gene (AtACD1) encoding a protein homolog to bacterial and fungal ACD and previously identified as a D-cysteine desulfhydrase (D-CDes) (Riemenschneider et al. 2005). ACD1 is a PLP-dependent enzyme that belongs to the β family of aminotransferases (Christen and Mehta 2001) and converts ACC into α-ketobutyrate and ammonia, two important molecules involved in the GS/GOGAT cycle during N assimilation (McDonnell et al. 2009). The transgenic lines over-expressing and under-expressing the AtACD1 gene showed a significant modulation in the triple response of Arabidopsis etiolated seedlings. Likewise, antisense lines exhibited 70% reduction in ACD activity and a significant increase in ethylene production (McDonnell et al. 2009). These results strongly suggest that ACD is responsible for maintenance and control of ACC pools in vivo but the underlying regulatory mechanisms at the transcriptional and posttranscriptional levels of this gene remain to be discovered.

ACC Treatment Reduces Aspartate Levels by an Unknown Mechanism

Seedling treatment by exogenous ACC induces a rapid decrease in primary and lateral root growth within minutes after application (Le et al. 2001; Swarup et al. 2007; Leblanc et al. 2008). Ultra performance liquid chromatography (UPLC) analyses of amino acid contents in root and shoot tissues of B. napus seedlings treated with increasing concentrations of ACC (from 0.1 to 10 μM) show that among amino acids, aspartate level is the most impaired and is positively correlated with changes in the root length and shoot surface area (Lemaire et al. 2013). In roots, a decrease in aspartate levels is also associated with a reduction in methionine (Fig. 5). Because ethylene biosynthesis depends directly on the aspartate-derived amino acid pathway, it is likely that ethylene can exert a feedback control on this pathway. Furthermore, aspartate issued from glycolysis and tricarboxylic acid (TCA) cycle plays a central role in C/N balance (Coruzzi 2003) and auxin catabolism (Ludwig-Müller 2009). How ACC or ethylene controls aspartate levels is a serious challenge in our understanding of the interaction between N, S, and C metabolisms. A combination of mutants of ethylene signaling and biosynthesis associated with pharmacological studies could be used to unravel this major mechanism involved in plant growth.

Fig. 5
figure 5

(a) Relationships between the exploratory root length and shoot surface area with the aspartate contents in the roots and shoots of Brassica napus seedlings treated over 5 days on agar plates under homogeneous feeding of 1 mM K15NO3. (b) Relationship between aspartate and methionine contents in the roots. Values are mean ± SE of three or four repeats of four seedlings each (adapted from Lemaire et al. 2013)

Full size image

4.4 Are ACC and Polyamines Potential Ligands or Modulators of Plant Glutamate Receptors?

Because of their role as neurotransmitters or modulators in animal cells, it is not excluded that ACC, polyamines, GSH, and conjugated forms of ACC may also act on plant glutamate receptors. Assuming ACC acts as a ligand or modulator on the plant glutamate receptors (GLR), this would explain why levels of ACC are so controlled in plant cells (Le Deunff and Lecourt 2016).

The Arabidopsis genome encodes for 20 GLR genes with coding sequences close to the ionotropic glutamate receptors (iGluR) first identified and characterized in mammals (Price et al. 2012; Weiland et al. 2016). Analyses of DNA sequences of GLR have grouped these proteins into three clades (Davenport 2002; Chiu et al. 2002). Expression studies have shown that GLR genes were mainly expressed in roots but also in leaves and reproductive organs (Chiu et al. 2002). The structure of GLRs is organized into four different units comprised of an amino terminal domain (ATD), a ligand binding domain (LBD), a trans-membrane domain (TMD) formed by three complete trans-membrane domains (M1, M3, and M4) with one re-entrant loop (M2) that forms the ion channel, and a C-terminal tail. Plant GLRs probably form tetrameric channels selective for Na+, K+, and Ca2+ as mammalian iGLuR counterparts (Tapken and Hollmann 2008; Price et al. 2012). Heterologous expression in transfected human embryonic kidney and Xenopus oocytes associated with patch-clamp studies have shown that AtGLR3.2, AtGLR3.4, and AtGLR1.4 can be gated by a broad spectrum of amino acids such as Asn, Ser, Gly, Met, Trp, Phe, Leu, Tyr, and Thr. Among these amino acids, Met was the most effective in AtGLR1.4 receptor whereas Arg was the most effective antagonist of Met effect (Vincill et al. 2012, 2013; Tapken et al. 2013). Moreover, the Cys and GSH tripeptides were the most effective agonists for activating AtGLR3.3 to suppress growth of the tomato bacterial pathogen Pseudomonas syringae in Arabidopsis leaves, suggesting that the methylthio-group common for both ligands could specifically interact with the receptor during the innate immunity response (Li et al. 2013).

If plants GLR are compared from an analogical perspective with their mammal’s counterparts, very surprising results can be envisaged. Indeed, in mammals, the ethylene precursor ACC is a synthetic molecule called ACPC and frequently used as a partial agonist of iGluR on glycine binding sites (Nahum-Levy et al. 1999; Inanobe et al. 2005). Similarly, polyamines such as spermine and spermidine stabilize the formation of the LDB dimer in a tetrameric receptor and facilitate the attachment of ATD lobes by “gluing” them together (Mony et al. 2009). Because polyamines and ACC-ACP are directly produced downstream from AdoMet biosynthesis, further investigations are required because ACC and polyamines are involved in many plant responses to biotic and abiotic stresses (Le Deunff and Lecourt 2016). GLR are located in the plasma membrane but also in membranes of plastids and mitochondria (Weiland et al. 2016). This suggests that they can react to a broad spectrum of amino acids present in cytosol or circulating in the apoplast. In this regard, plant GLRs could act as an internal and external N sensor system and play a major role in the regulation of amino acids biosynthesis or their compartmentalization in different organelles in relation to the extracellular flows of amino acids.

5 Relationship Between Ethylene and N Metabolism: The Central Role of Aminotransferases

A close examination of Fig. 3 indicates that no less than six PLP-dependent aminotransferases, namely AAT/PAT, OASTL, CgS, CbL, ACS, and an unknown AAT, are required for ethylene biosynthesis and methionine salvage pathways (Fig. 3). Therefore, among the different plant amino acid pathways (Fig. 1), the methionine biosynthetic pathway needs the highest involvement of aminotransferases for its biosynthesis and is highly dependent on the availability in PLP cofactor (vitamin B6). This ascertainment is exemplified in recent studies of pyridoxine synthase1 enzyme involved in PLP biosynthesis and encoded by two paralogous genes, namely PDX1.1 and PDX1.3 in Arabidopsis (Chen and Xiong 2009a, b; Boycheva et al. 2015). Disruption of either of these genes in single mutants results in vitamin B6 deficiency and a differential reduction of ethylene production and auxin levels while disruption of both genes in double mutants is lethal for seedlings. At the phenotypic level, these mutants displayed impairment in root growth, reduction of meristem size, and altered root cell division and elongation (Chen and Xiong 2009a, b; Titiz et al. 2006). However, the single pdx1.3 mutant is more impaired in root development than the pdx1.1 mutant. These phenotypic differences are mainly explained by the presence of distinct regulatory elements in the upstream region of both genes that lead to a deficit in ethylene production and/or signaling. Thus, PDX1.1 expression is repressed by sucrose and its promoter possesses a sugar response element whereas PDX1.3 promoter contains ethylene and auxin response elements. Several evidences indicate that some components of the ethylene signaling pathway act upstream of the effect of auxin biosynthesis and transport (Růžička et al. 2007; Boycheva et al. 2015). Indeed, the root phenotype of the pdx1.3 mutant is partially restored by ACC treatment but not by auxin treatment (Chen and Xiong 2009a, b; Boycheva et al. 2015). These results perfectly demonstrate that PLP availability plays a central role in ethylene biosynthesis and probably the methionine pathway as well as in phytohormone homeostasis. However, as previously mentioned it is not known how de novo PLP synthesis in cytosol or PLP salvage pathway modulate PLP pool in the plastid and which plastidial carriers are involved in the transport of different forms of vitamin B6 (Gerdes et al. 2012).

5.1 Methionine and Tryptophan Interconversion by VAS1 and TAA Aminotransferases

For a long time, it was assumed that the unique role of methionine as a molecule in the cell was to produce proteins via its utilization in methionyl-tRNA formation as a substrate of aminoacyl-tRNA synthetase. However, it was demonstrated recently that the cytosolic PLP-dependent aminotransferase called VAS1/ISS1 (for reversal of s av3 phenotype and indole severe sensitive1) catalyzes the conversion of methionine as amino acid donor and indole-3-pyruvic acid (IPA) as amino acceptor into L-tryptophan and 2-keto-4-methythiobutyrate (Zheng et al. 2013a; Pieck et al. 2015). IPA is the precursor of IAA (Fig. 4) that is involved in the most important route of plant IAA biosynthesis belonging to the tryptophan-dependent pathways (Stepanova et al. 2008, 2011; Mashiguchi et al. 2011; Won et al. 2011; Zhao 2014). Indeed, among 19 of naturally occurring amino acids examined, methionine was the most catalytically preferred amino acid donor followed by Phe with only 21% specificity relative to Met (Pieck et al. 2015). Therefore, VAS1/ISS1 catalyzes the opposing reaction of tryptophan aminotransferases TAA1 (Fig. 4) and tryptophan aminotransferases-related TAR1 and TAR2 (Stepanova et al. 2008). TAA PLP-dependent enzymes convert Trp by using α-ketoglutarate or pyruvate as organic acid donors, into IPA and glutamate or alanine, respectively (Fig. 4). In Arabidopsis, a vas1-2 single mutant and vas1-2sav3 double mutant showed a significant increase in IAA levels and a fivefold increase in the level of the ethylene precursor ACC (Zheng et al. 2013a). However, free L-methionine levels did not differ between mutant and WT, demonstrating that regulation of the methionine biosynthesis is controlled downstream by methionine synthase (MS) and methionine adenosyltransferase (SAM) (Bürstenbinder et al. 2007). Since the VAS1/ISS1 and TAA aminotransferases catalyze opposite reactions, it is assumed that VAS1 could counteract the production of IPA and methionine by subtracting a part of the IPA and methionine from IAA and methionine pathways, respectively (Zheng et al. 2013a). However, some evidence suggests that SAV1/ISS1 could be mainly involved either in Trp catabolism or in metabolism of Phe and Tyr (Pieck et al. 2015). Indeed, protein sequence analyses of SAV1/ISS1 with other plant aromatic aminotransferases (AroAT) showed that SAV1/ISS1 is a new subgroup of aromatic PLP-dependent enzymes belonging to the Iα aspartate family. This subgroup is distinct from Trp ATs and Tyr ATs previously described and could be involved in Phe and/or Tyr biosynthesis (Pieck et al. 2015). Unfortunately, these studies did not test whether SAV1/ISS1 could also catalyze the conversion of Met and phenylpyruvate into KMTB and Phe, or the conversion of Met and 4-hydroxyphenyl pyruvate into KTMB and Tyr (Fig. 4). The latter reactions would explain the role of SAV1/ISS1 in Phe and Tyr biosynthesis from methionine.

5.2 Methionine and Phenylalanine Interconversion by PPY-AT in the New Phenylpyruvate Pathway

Until now, the arogenate pathway was considered as the exclusive route for Tyr and Phe biosynthesis in plants (Cho et al. 2007; Maeda and Dudareva 2012). However, it was recently demonstrated that Phe could also be synthesized in cytosol via phenylpyruvate alternative pathway (Yoo et al. 2013; El-Azaz et al. 2016). This functional phenylpyruvate pathway was similar to most microorganisms and required two reaction steps. In the first reaction step, prephenate is converted to phenylpyruvate in plastids by arogenate dehydratases (ADT) that retain prephenate dehydratase activity (PDT) from a 22-amino acid region called PAC domain (for PDT activity conferring Domain) conferring PDT activity to ADTs (El-Azaz et al. 2016). In the second reaction step, a cytosolic phenylpyruvate-aminotransferase (PPY-AT) catalyzes the interconversion of phenylpyruvate and tyrosine to 4-hydroxyphenylpyruvate and Phe (Yoo et al. 2013). As demonstrated PPY-AT could also convert phenylpyruvate and Met into Phe and 2-keto-4-methylthiobutyrate (Fig. 4) since PPY-AT is able to use Met as amino donor with only 13% specificity relative to tyrosine (Yoo et al. 2013). Therefore, the discovery of phenylpyruvate pathway links biosynthesis of phenylalanine with catabolism of tyrosine but also increases the complexity of the transaminases network involved in the regulation of aromatic amino acids and methionine (Fig. 4).

5.3 In Search of the Last Missing Aminotransferase to Complete the Yang Cycle

In plants and many microorganisms, due to their promiscuity, various aspartate aminotransferases (AAT) are presumed to be involved in the last step of the Yang cycle catalyzing the transamination of KMTB into methionine (Berger et al. 2003; Sekowska et al. 2004; Pommerrenig et al. 2011). In Arabidopsis, plastids contain two different AAT: a prokaryotic-type (PT-pAAT) and a eukaryotic-type (ET-pAAT) that are involved in the first step of the aspartate-derived amino acid pathway (de la Torre et al. 2006, 2014a, b). It is strongly suspected from analysis sequences that ET-pATT (ASP5) and PT-AAT could be implied in the last step of the Yang cycle (Pommerrenig et al. 2011). Moreover, it was recently discovered that plant PT-pAAT is bispecific since it also displays a prephenate aminotransferase activity (PAT) and catalyzes the conversion of glutamate and prephenate into arogenate and 2-oxoglutarate (Graindorge et al. 2010; Maeda et al. 2011). Therefore, it is not excluded that methylthio-group salvage of AdoMet by the Yang cycle in plastids could compete in its last step with the biosynthesis of Phe and Tyr (lignin and phenylpropanoids precursors) in the arogenate pathway catalyzed by the PAT activity of PT-pAAT (Figs. 1 and 4). In this respect, the PT-pAAT/PAT silenced plants show a severe reduction in growth as well as a decrease in chlorophyll and lignin biosynthesis (de la Torre et al. 2014a, b). Unfortunately, AdoMet and Met levels have not been measured so that it is difficult to evaluate the impact of this enzyme on the last step of the methionine pathway. Likewise, it is likely that the cytosolic aspartate aminotransferases ASP2 and ASP3 could also catalyze transamination of KMTB to methionine when the last step of Yang cycle occurs in the cytosol (Miesak and Coruzzi 2002). In summary, the functioning of AAT/PAT, SAV1/ISS1, and PPY-AT aminotransferases could be a major metabolic crossroad interconnecting the metabolism of methionine and aromatic amino acids (Trp, Phe, and Tyr) with biosynthesis of ethylene, IAA, PAA, and phenylpropanoids.

5.4 Is Tryptophan Aminotransferase Also Involved in Histidine Catabolism?

Although histidine catabolism is well established in animal cells, this pathway has not yet been established in plant cells (Hildebrandt et al. 2015). A surprising result has been recently obtained with the human fungal pathogen Candida glabrata and Saccharomyces cerevisiae that initialize His degradation via the aromatic aminotransferase ARO8. In yeast, ARO8 is known to be an AroAT that in vitro converts Phe, Tyr, and Trp as amino acid donor to phenylpyruvate, α-ketoglutarate, and pyruvate as amino acid acceptors (Iraqui et al. 1998; Urrestarazu et al. 1998). However, in both fungi ARO8 is ten-fold up-regulated by exogenous His and is also able to transfer the donor amino group of His to the α-ketoglutarate acceptor to produce imidazol-5-yl-pyruvate and glutamate (Brunke et al. 2014). Furthermore, heterologous expression of SAV1/ISS1 or TAA1 Arabidopsis genes can rescue the yeast mutant Aro8Aro9 auxotroph for Phe and Tyr (Pieck et al. 2015). In B. napus seedlings, in vivo inhibition of TAA and ACS activities by root treatment with 10 μM AVG (Soeno et al. 2010) induced a 2.5-fold and 8-fold increase of His levels in the roots and shoots, respectively (Fig. 6). However, treatment with 1 mM glutamate was able to restore control levels of amino acids in roots and shoots with the exception of His (Le Deunff et al. 2018). In yeast, His starvation imposed by 3-aminotriazole induced the bZIP transcription factor GCN4 that initiates transcription of 539 genes such as genes encoding amino acids transporters and biosynthetic enzymes as well as genes involved in PLP biosynthesis (Niederberger et al. 1981; Natarajan et al. 2001; Hinnebusch 2005). In Arabidopsis, His homeostasis is crucial for root development since the hap1 mutant of plastidial histidinol phosphate aminotransferases (HPA), the eighth enzyme of His biosynthesis, displays a defect in root meristem maintenance and reduction in root development (Mo et al. 2006). Taken together, these results suggest that TAA could be involved in His catabolism in plants and play a major role in the regulation of amino acid biosynthesis. Therefore, TAA/SAV1/ISS1 interconversion system could play a major role as a metabolic crossroad in the amino acid imbalance and changes in C/N ratio acting on N sensory systems such as the GCN4, GCN2, and TOR kinases signaling involved in plant nutrition.

Fig. 6
figure 6

Changes in free amino acid contents from aspartate-derived amino acid pathway induced by 10 μM AVG and 10 μM AVG + 1 mM Glu treatments in the root (a) and shoot tissues (b) of Brassica napus seedlings treated over 5 days on agar plates under homogeneous feeding of 1 mM KNO3. Values are the average (±SE) of three agar plates (n = 3) of four seedlings each. Significant differences between control (1 mM KNO3) and treatments are given for *p < 0.05; **p < 0.01; ***p < 0.005; (t-test) (adapted from Le Deunff et al. 2018, submitted)

Full size image

6 Use of Olefinic Glycine Inhibitors in Ethylene Biosynthesis: The Devious Trap

Since the 1970s, olefinic glycine analogues have been used as small bioactive molecules to inhibit specifically in vitro and in vivo PLP-dependent aminotransferases involved in the methionine biosynthetic pathway and ethylene biosynthesis (Rando 1974a, b; Berkowitz et al. 2006). In fact, this family of compounds acts as terminal inhibitors of subgroup I of aminotransferases belonging to the α family (Christen and Mehta 2001). The subgroup I contains ACC, aspartate, alanine, and aromatic (His, Phe, Tyr, Trp) aminotransferases (Christen and Mehta 2001). Recent findings have highlighted the non-specificity of these compounds in ACS and TAA/TARs enzymes (Soeno et al. 2010). Due to the structural analogies of this family of inhibitors (Table 1) and the promiscuity of aminotransferases belonging to subgroup I, caution is now required when using these inhibitors to validate molecular studies with mutational approaches or transgenic plants on ethylene and IAA biosynthesis and signaling (Le Deunff and Lecourt 2016; Le Deunff et al. 2016). Moreover, cellular compartmentalization of their target enzymes in the peroxisomes, plastids, mitochondria, and cytosol probably modulate their effect in term of specificity, permeability, and stability during in vivo treatments. However the broad spectrum of action of these compounds remains a powerful tool (Le Deunff and Lecourt 2016). Indeed, they create an imbalance in the amino acid levels which makes it possible to test potential candidates for N sensory systems such as the TOR and GCN2 signaling pathways or to discover new targets involved in N detection and C/N ratio changes in primary metabolism (Le Deunff et al. 2018, submitted).

Table 1 Origin and structure of amino acid analogs belonging to enol ether family able to inhibit catalysis reactions of PLP-dependent aminotransferases (adapted from Lieberman 1979; Berkowitz et al. 2006)
Full size table

6.1 In Search for Highly Specific Inhibitors of ACS and TAA Aminotransferases

Because ethylene and auxin biosynthetic enzymes such as ACS and TAA are encoded by redundant genes in the Arabidopsis genome, the mutational approaches on single or multiple genes by reverse and forward genetics are often associated with plant lethality, poor growth, and sterility (Tsuchisaka et al. 2009; Stepanova et al. 2008). Chemical genetic or genomic approaches can overcome these limitations by selecting for small bioactive molecules able to inhibit specifically protein activity in any tissue and at any time during plant development and avoiding side effects (Zheng and Chan 2002; Blackwell and Zhao 2003; Robert et al. 2009). Recently, these chemical strategies have been used in Arabidopsis to find out new specific inhibitors of ACC synthase (Lin et al. 2010) and TAA enzymes (Soeno et al. 2010; Nakamura et al. 2016). Thus, more specific inhibitors of the ACS enzyme formed from a quinazolinone backbone (called compounds 9393, 9370, and 7303 compounds) have been discovered from their ability to suppress the constitutive triple response of ethylene overproducer mutant eto1-4 (Lin et al. 2010). Similarly, 41 compounds derived from the 2-(aminooxy)-3-(naphthalen-2-yl)propanoic acid (KOK1169/AONP) backbone were able, in vivo and in vitro, to reduce TAA aminotransferase more specifically than L-α-aminooxy-phenylpropionic acid (AOPP) (Nakamura et al. 2016). Today, these compounds form a new class of specific TAA inhibitors designated as “pyruvamine.” Therefore, it is reasonable to assume that the use of these specific and less specific inhibitors in differential transcriptomic studies should lead to the discovery of metabolic hubs in N and C metabolism or N sensory systems involved in the regulation of the root morphogenetic program.

7 Are Aminotransferases Potential Targets for the Improvement of Nitrogen Use Efficiency (NUE)?

An alternative approach to address the metabolic interconnections between mineral nutrition and the production of plant growth regulators consists of proposing that the biosynthesis of hormones such as ethylene and IAA is highly regulated through the aminotransferase network in primary metabolism (Le Deunff et al. 2016; Le Deunff et al. 2018 submitted). Indeed, aminotransferases are the key enzymes that ensure N, S, and C shuttling through production of amino and organic acids but also biosynthesis of ethylene and auxins via Met, Trp, and Phe. Therefore, using mutants or nonspecific inhibitors of aminotransferases involved in ethylene and auxins biosynthesis can modulate the aminotransferase network and impact nitrogen use efficiency (NUE). It is likely that such approaches may disrupt N sensory systems such as TOR and GNC2 kinase signaling, the plant GCN4 transcription factor homolog, or GLR receptors. We present here some recent results in favor of this assumption.

7.1 Over-expression of the Tryptophan Aminotransferase Genes Improves Nitrogen Use Efficiency

In Arabidopsis the members of the Trp aminotransferases namely TAA1 (also known as WEI8, SAV3 and TIR2) and their related proteins TAR1 and TAR2 play critical role in plant development in response to stresses (Stepanova et al. 2008; Ma et al. 2014). Thus, the Arabidopsis double mutant wei8tar2 displays a dwarf and bushy plant phenotype with reduced vasculature and abnormal flowers with complete sterility, whereas the triple mutant wei8tar2tar1 is rootless, seedling-lethal, and impaired in embryo patterning (Stepanova et al. 2008). These mutants are insensitive to ethylene and ACC treatments because they lack the triple response. A recent study has shown that TAA1 is highly transcriptionally regulated by ARR transcription factors (Arapidopsis response regulator) that can bind after dimerization at two different sites. One site is present in the TAA1 promoter region whereas the other is found in the second intron of the gene sequence (Yan et al. 2017). Thus, transcriptional responses are modulated by a subset of ARRs implicated in binding on site 1 or 2. The ARRs biosynthesis is regulated by cytokinin, ethylene, light, and developmental signals. Especially, ethylene response is mediated by the EIN3 transcription factor, which is able to bind to the ARR12 protein and modulate TAA gene expression (Yan et al. 2017). These results explain previous data indicating that TAA1 transcription is activated by EIN3 (He et al. 2011). The major role of TAA1 aminotransferase in plant development was also demonstrated in maize (Zea mays) and rice (Oryza sativa) where the loss of function of two co-ortholog genes ZmVT2 (Vanishing Tassel2) and OsFIB (Fish Bone) caused dramatic effects on vegetative and reproductive development associated with a reduction in IAA levels (Phillips et al. 2011; Yoshikawa et al. 2014).

Furthermore, it was found in Arabidopsis that under nitrogen limitation, aminotransferase TAR2 controls LR proliferation, suggesting that TAR2 is required for the reprogramming of root architecture in response to low nitrogen availability (Ma et al. 2014). Thus, transgenic plants over-expressing the TAR2 gene showed a significant increase in the number of LRs under low and high nitrogen conditions. In order to improve NUE in crop species, over-expression of the TAR2 gene has been recently engineered in wheat (Triticum aestivum). After a selection among the 12 TAR2 alleles present in the wheat genome, the best candidate TaTAR2.1 gene expressed in different organs was over-expressed by using a constitutive promoter (Shao et al. 2017). The engineered plants grown under controlled-environment and field conditions showed an increase in LR-growth under low N supply conditions at the plantlet stage. At maturity and whatever the N supply levels applied, plants also displayed a significant increase in the biomass, plant height, spike number, and grain yield (Shao et al. 2017). Taken together, these results indicate that TAR2 genes show a potential for engineering crop plants for improving gain of yield under nitrogen-limiting conditions.

7.2 Over-expression of Alanine Aminotransferase Genes Improves Nitrogen Use Efficiency

Alanine aminotransferase (AlaAT) is a PLP-dependent aminotransferase that catalyzes the amino group transfer of alanine to α-ketoglutarate to form glutamate and pyruvate. This aminotransferase was used to engineer nitrogen use efficiency in maize, rice, and canola (Good et al. 2007; Shrawat et al. 2008). Over-expression of AlaAT from barley (Hordeum vulgare, HvAlaAT) under the control of root-specific and constitutive promoters OsAnt1 and btg26 in rice and canola, respectively, induces a significant increase in NUE and root biomass in these crop species (Good et al. 2007; Shrawat et al. 2008). Especially, under N-limiting conditions, canola and rice plants exhibit increased biomass and yields. In field trials with 40% less applied N, canola maintains yields. The physiological reasons for this improvement remain elusive. It is assumed that the decrease in some amino acids in shoots such as Gln and Glu could alleviate the repression of these AA on nitrate uptake transporters and increase N uptake (Good et al. 2007). However, it is not excluded that the overproduction of pyruvate can play a major role in both gluconeogenesis and auxins biosynthesis (IAA and PAA). Indeed, auxins production needs pyruvate for TAA/TAR activities since pyruvate is one of the 2-oxoacid acceptors used for their functioning (Le Deunff et al. 2016). Similarly, phosphoenolpyruvate carboxykinase PEPCK activity (catalysing the reversible conversion of oxaloacetate to phosphoenolpyryvate, see Fig. 1) is required for sink tissues metabolically active such as root, stem, and leaves (Malone et al. 2007). Unfortunately, no study of transcriptional regulation of AlaAT or PEPCK promoters has been done as with the TAA1 promoter in Arabidopsis. Therefore, it is difficult to know which endogenous and exogenous signals are involved in the regulation of AlaAT and PEPCK expression.

7.3 Could the Under-expression of ACC Synthase Genes Improve Nitrogen Use Efficiency?

Because a multigene family in Arabidopsis encodes ACS aminotransferases, acs mutants of this specie, mutated at multiple loci, has been engineered to understand the function and regulatory roles of these proteins (Tsuchisaka et al. 2009). Analyses of pentuple, hexuple, and octuple acs mutants demonstrated that in normal conditions ethylene acts as growth repressor in dark- or light-grown plants since pentuple and hexuple mutants are bushier and display significant greater height with the progressive decrease in ethylene production. Indeed, ethylene is known to down-regulate photosynthetic genes (Van Zhong et al. 2003). These acs mutants also exhibited a delayed flowering time, a less response to gravity, and enhanced susceptibility to the necrotrophic pathogen Botrytis cinerea. Among the mutants, growth of the octuple acs mutant is delayed during the initial stage of development and becomes taller and less bushy after 50 days of growth. Moreover, transcription analyses of acs mutants confirmed that there exists a relationship between expression of light signaling genes and ethylene biosynthesis. This study clearly highlights the importance of spatial and temporal combination between ACS isoforms in multiple ethylene-mediated physiological processes during growth and development. The multiple-locus acs mutants appear to be a valuable tool for deciphering how ethylene biosynthesis interacts with nitrogen metabolism in response to different levels of nitrate availability and how decrease in ethylene production affects the levels of Asp, Met, AdoMet, and PAs. In this respect, the crosstalk between nitrogen nutrition and the ethylene plant hormone in Col-0 and 20 natural accessions of Arabidopsis seedlings have been recently investigated (De Gernier et al. 2016). Comparison of the behavior of the 20 accessions behavior revealed that changes in root biomass and ethylene production were negatively correlated at 1 mM but positively correlated at 10 mM nitrate. Greater ethylene release and root biomass production under nitrogen limitation were mainly due to higher transcription levels in the roots of ACS6 and of ACO2 and ACO4 genes, respectively. Taken together these studies indicate that ethylene modulates plant morphology and biomass allocation probably in relation to nitrate availability.

This conclusion is in line with recent findings demonstrating that endogenous glucose signals increase growth by promoting auxin signaling and by antagonizing ethylene signaling through the glucose sensor HXK1 (Moore et al. 2003). Indeed, glucose enhances the degradation of EIN3 and EIL1 (EIN3-Like1) components of ethylene signaling through a proteasome-dependent mechanism controlled by HXK1 (Yanagisawa et al. 2003; Yoo et al. 2008). During primary root growth in Arabidopsis, EIN3 is known to exert a positive feedback loop on the expression of anthranilate synthase and TAA1/TAR2 genes involved in auxin biosynthesis whereas IAA enhances EIN3 stability by repressing EBF1/2(EIN3 Binding F-Box protein1)-mediated degradation of EIN3 (He et al. 2011). Although HXK1 and TOR kinase signaling systems in response to glucose seem to be mostly uncoupled more studies are required to understand possible interactions between both sensory systems (Sheen 2014).

Furthermore this conclusion is also corroborated by the overaccumulation of anthocyanin in ethylene signaling mutants etr1, ein2, ein3/eil1, and rdh3 (root hair defective3) subjected to nitrogen deficiency, whereas exogenous application of ACC almost completely suppresses anthocyanin accumulation in Col-0 WT plants and ACC effect was attenuated in the ethylene signaling mutants (Wang et al. 2015). Anthocyanin is accumulated under low nitrate levels because nitrate relieves the repression of lateral organ boundary (LDB) transcription factors on anthocyanin biosynthetic genes (Rubin et al. 2009). The LDB genes also repress many other N-responsive genes such as nitrate reductase genes NIA2 and NIA2 and nitrate transporters AtNRT1.1 (NPF6.3), AtNRT2.1, and AtNRT2.5. Regulation by ethylene of nitrate responsive genes such as nitrate transporters is also demonstrated during N-induced nutritional stress. Indeed, rapid changes in nitrate supply such as transfers from high to low concentration and vice versa induce a burst of ethylene production in the roots associated with a differential expression of AtNRT2.1 and AtNRT1.1 (NPF6.3) nitrate transporters genes (Tian et al. 2009; Zheng et al. 2013b). In response to nitrate limitation, induction of AtNRT2.1 plays a positive role on ethylene biosynthesis and signaling pathway but the ethylene signaling components such as EIN3/EIL1 induce in turn the repression of AtNRT2.2 (Zheng et al. 2013b). In response to nitrate excess, expression of AtNRT1.1 (NPF6.3) and AtNRT2.1 genes is respectively up- and down-regulated (Tian et al. 2009). However, these responses are abolished in etr1-3 and ein2-1 ethylene receptor mutants, again demonstrating that components of ethylene signaling are probably involved in nitrate nutritional responses.

8 Conclusion

Ethylene biosynthesis from the amino acid aspartate needs at least seven different aminotransferases belonging to the subfamilies I and γ of the α family: AAT, AAT/PAT, CgS, CbL, Vas1/ISS1, ACS and an unknown AAT aminotransferase to complete definitively the Yang cycle. Five of these aminotransferases play a fundamental role in N, S and C shuttling in plant cells (namely CgS, CbL, ACS, AAT/PAT, VAS1/ISS1) and three of them (VAS1/ISS1, AAT/PAT, and an unknown AAT) connect the aspartate/methionine metabolism to aromatic amino acid biosynthesis and catabolism in a unique and major network of aminotransferases in the plastids.. Because of the promiscuous nature of the aminotransferases for their substrates, it is likely that different aminotransferases can be involved in the last step of the Yang cycle within plastids or the cytosol. The functioning of these aminotransferases depends directly on the availability of PLP that is synthesized from glutamine, the first amino acid produced by the glutamine synthetase in the nitrate reduction/assimilation pathway. Understanding how the genes encoding these aminotransferases and their cofactor are transcriptionally and translationally regulated is a major challenge in unraveling the interconnection between Trp, Met, and Phe amino acid biosynthesis, hormone production, and nitrogen metabolism. Involvement of N, S, and C sensory systems in the regulation of these aminotransferases such as hexokinase, glutamate like receptors, TOR, and GCN2 kinases seems highly probable.