Nucleotide Metabolism in Plants^,

Abstract

Nucleotide metabolism is an essential function in plants.

NUCLEOTIDE METABOLISM IN PLANTS

Nucleotides are essential for life. It is easy to validate this statement—one just needs to recall that nucleotides are the building blocks of DNA and RNA, and that many molecules that are central for metabolism, for example ATP, NADH, Co-A, and UDP-Glc, are nucleotides or contain nucleotide moieties. Generally, a nucleotide is defined as a phosphorylated ribose or deoxyribose linked to a nitrogen-containing heterocyclic group called the nucleobase via a glycosidic bond (Fig. 1). Because of the phosphate groups, nucleotides are negatively charged, whereas at neutral pH, nucleosides and nucleobases are uncharged. The exception is xanthine, which is partially charged as a free base (pKa 7.4) but completely charged at the base in xanthosine (pKa 5.5) or the corresponding nucleotides (Fig. 1; Sigel et al., 2009).

Figure 1.

Structural composition of nucleobases, nucleosides, and nucleotides. For the nucleobases, “R” is simply a proton. For the nucleosides, “R” is a sugar moiety that can be ribose or deoxyribose (carrying a proton instead of a hydroxyl group at the 2′ carbon of the ribose). Nucleotides have up to three phosphate groups esterified to the hydroxyl group of the 5′-carbon of the nucleoside sugar determining the prefix mono-, di-, or tri- in the name of the molecule. The terminal phosphate always carries two charges, irrespective of the number of phosphates present. The pyrimidine nucleobases (upper row) and the purine nucleobases (lower row) are shown with the groups attached to the heterocycles highlighted in red (oxo groups), blue (amino groups), and gray (methyl group). dTMP, deoxy-TMP; dXMP, deoxy-XMP; dIMP, deoxy-IMP.

Many excellent reviews focus on general (Wagner and Backer, 1992; Moffatt and Ashihara, 2002; Stasolla et al., 2003; Zrenner et al., 2006; Zrenner and Ashihara, 2011) or particular aspects (Smith and Atkins, 2002; Kafer et al., 2004; Ashihara et al., 2018) of plant nucleotide metabolism. The aim of this review is to provide an update on how nucleotide metabolism is hardwired, mostly focusing on the cellular level, because our understanding of the organization at the tissue and organ level remains very limited. The presented models are mostly based on results from Arabidopsis (Arabidopsis thaliana). These will often be valid for most plants, but certainly there will be species-dependent variations. We also cover extracellular nucleotide metabolism and review the evidence for overlap between cytokinin metabolism and central nucleotide metabolism. Figure 2 shows a general overview of plant nucleotide metabolism.

Figure 2.

Schematic overview of plant nucleotide metabolism. Nucleotides are synthesized “de novo” from precursor molecules listed in the upper left box. The phosphorylation of nucleoside monophosphates (NMPs) via diphosphates (NDPs) generates nucleoside triphosphates (NTPs), which serve as building blocks for RNA synthesis and as precursors for the biosynthesis of the metabolites shown in the center (SAM, UDP-Glc, and NADH are given as examples). However, the NTPs, in particular ATP and GTP, are not only precursors for other metabolites, but are also essential stores of chemical energy in the phosphoanhydride bonds used in a multitude of energetic coupling reactions, as well as important donors of phosphate in kinase reactions (not shown). NDPs can be reduced to dNDPs, which after phosphorylation to dNTPs serve as precursors for DNA biosynthesis. RNA degradation in the cytosol releases nucleoside monophosphates, whereas nucleosides are produced during vacuolar RNA degradation. Adenosine and adenine are products of biochemical reactions involving SAM. Nonenzymatic decay (depurination) and enzymatic repair reactions result in nucleoside and nucleobase release from DNA. Nucleobases and nucleosides can be recycled to nucleotides in so-called salvage reactions. Plants are also capable of full nucleotide degradation via certain nucleosides and nucleobases releasing the nitrogen of the nucleobases as ammonia.

DE NOVO SYNTHESIS

Purine De Novo Synthesis

Plants possess metabolic pathways for the de novo synthesis of purine nucleotides generating AMP, as well as pyrimidine nucleotides yielding UMP. During de novo biosynthesis, nucleotides are newly synthesized from activated ribose (5-phosphoribosyl-1-pyrophosphate [PRPP]), Gln, Asp, and bicarbonate, as well as specifically for the purine nucleotides Gly and formyl tetrahydrofolate (Fig. 2).

There is strong evidence that AMP biosynthesis occurs entirely in the plastids, because the 11 enzymes (catalyzing 12 reactions; Smith and Atkins, 2002) required for AMP biosynthesis in Arabidopsis all have an N-terminal organelle-targeting peptide, and C-terminal yellow fluorescent protein-fusion proteins of several of these enzymes were observed exclusively in the plastids when they were transiently expressed in Nicotiana benthamiana in our laboratory (Fig. 3A; N. Medina Escobar and C.-P. Witte, unpublished data). In rice (Oryza sativa), the pathway also seems to reside in plastids (Zhang et al., 2018). However, it has been reported that in nodules of the tropical legume cowpea (Vigna unguiculata), purine biosynthesis is targeted to plastids and mitochondria (Atkins et al., 1997; Smith and Atkins, 2002). It may be worthwhile to reconfirm this special localization in nodules using fluorescent tagged proteins.

Figure 3.

Purine and pyrimidine de novo biosynthesis. A, Purine de novo biosynthesis. B, Pyrimidine de novo biosynthesis. Enzymes and transporters include brittle1 (BT1; 1), AMP deaminase (AMPD; 2); IMP dehydrogenase (IMPDH; 3); GMP synthetase (GMPS; 4); asparate transcarbamoylase (ATCase; 5); dihydroorotatase (DHOase; 6); dihydroorotate dehydrogenase (DHODH; 7); UMP synthase (UMPS; 8); and CTP synthetase (CTPS; 9). An anchor symbol denotes an association with the respective membrane.

AMP is exported from the plastids by the adenine nucleotide uniporter BT1 (Fig. 3A, no. 1), which can also transport ADP and ATP (Leroch et al., 2005; Kirchberger et al., 2008; Hu et al., 2017). Interestingly, BT1 from Arabidopsis and maize (Zea mays) was reported to be dually localized to the chloroplast and mitochondria (Bahaji et al., 2011b) and a BT1 mutant with a severe dwarf phenotype could be complemented with a cDNA coding for an N-terminally truncated version of BT1 that is exclusively located in the mitochondria and not in the plastids (Bahaji et al., 2011a). This either indicates that there are certain tissues or developmental stages where purine nucleotide biosynthesis occurs mainly in mitochondria, or that BT1 has an alternative and essential function in this organelle. In the latter case, BT1 might still also be involved in exporting adenylates from plastids, but the abrogation of the plastidic variant would not cause the strong dwarf phenotype, indicating that de novo synthesized AMP has an alternative, BT1-independent way to leave the plastids.

GMP biosynthesis requires inosine 5′-monophosphate (IMP; Fig. 3A), which can either be derived from AMP deamination in the cytosolic compartment catalyzed by AMP deaminase (AMPD; Fig. 3A, no. 2) or from direct export of IMP from the plastids, because IMP is generated there en route to AMP (Fig. 3A). Mutation of AMPD is zygote lethal (Xu et al., 2005), and coformycin, an AMPD inhibitor, is a potent herbicide after its phosphorylation in vivo (Dancer et al., 1997). These phenotypic effects might be caused by hampered GMP biosynthesis, suggesting that AMPD could be required for this process. Consistent with this, AMPD is strongly activated by ATP (Han et al., 2006), and this regulation might balance cellular ATP and GTP concentrations. However, it has also been reported that AMPD inhibition might be detrimental by severely altering the cellular energy charge and that the GTP pool is not altered upon AMPD inhibition (Sabina et al., 2007), implying that GMP biosynthesis is independent of AMPD and that there is an alternative IMP supply from the plastids. The activity of AMPD likely resides in the cytosol, but the protein has an N-terminal transmembrane domain and is clearly attached to a membrane (Han et al., 2006).

The following enzymatic reactions for GMP biosynthesis probably take place in the cytosol: (1) oxidation of IMP to xanthosine 5′-monophosphate (XMP; Fig. 1) by IMP dehydrogenase (IMPDH; Fig. 3A, no. 3); and (2) amination of XMP to GMP by GMP synthetase (GMPS; Fig. 3A, no. 4). Neither of these enzymes has an apparent subcellular targeting peptide, and both were detected in the cytosolic proteome of Arabidopsis (Ito et al., 2011). Also, IMPDH from cowpea nodules was associated with the cytosolic fraction (Shelp and Atkins, 1983). GMP is a quite strong competitive inhibitor of IMPDH (Atkins et al., 1985), maybe resulting in feedback regulation in vivo.

Pyrimidine De Novo Synthesis

The first dedicated reaction of UMP biosynthesis (Fig. 3B) is catalyzed by aspartate transcarbamoylase (ATCase; Fig. 3B, no. 5). The enzyme is located in the plastids. The next enzyme, dihydroorotatase (DHOase; Fig. 3B, no. 6) was associated to plastids in cell fractionation studies (Doremus and Jagendorf, 1985 and references on the SUBA web server; Hooper et al., 2017) but was located in the cytosol when transiently overexpressed in Arabidopsis protoplasts as a GFP-tagged fusion protein (Witz et al., 2012). Maybe DHOase is associated with the chloroplast membrane via a protein-protein interaction, and the interaction partner is overwhelmed by strong overexpression of DHOase. The next enzyme in pyrimidine biosynthesis, dihydroorotate dehydrogenase (DHODH, Fig. 3B, no. 7), is associated with mitochondria (Doremus and Jagendorf, 1985; Witz et al., 2012) and likely located on the outer surface of the inner mitochondrial membrane, as observed for the mammalian orthologs (Ullrich et al., 2002). UMP synthase (UMPS; Fig. 3B, no. 8), the final enzyme in pyrimidine biosynthesis, was associated with the plastids and the cytosol, as shown by cell fractionation of pea (Pisum sativum) leaves (Doremus and Jagendorf, 1985) and was present in the cytosol after overexpression in Arabidopsis protoplasts (Witz et al., 2012). Thus, UMP is generated in the cytosol, while it appears that the responsible enzyme might have some affinity for the chloroplast. Because ATCase (Fig. 3B, no. 5) is feedback regulated by uridylates, in particular UMP (Doremus and Jagendorf, 1985), the cytosolic uridine nucleotide pool must be tightly connected to the plastidic pool.

CTP biosynthesis requires the phosphorylation of UMP to UTP (see section on "The Generation of Nucleoside Triphosphates"), which is the substrate of CTP synthetase (CTPS; Fig. 3B, no. 9). There are five CTPS isoenzymes in Arabidopsis, and all reside in the cytosol. For one isoform (CTPS3), the activity and stimulatory allosteric regulation by UTP and GTP have been recently shown (Daumann et al., 2018). Interestingly, some CTPS isoenzymes form filamentous aggregates, called cytoophidia, inactivating the enzyme. In vitro, these are generated in particular in the presence of CTP, indicating that the enzyme is feedback regulated by this mechanism (Daumann et al., 2018). Knockout mutants for each CTPS were characterized, and except for CTPS2, which showed a complete block of germination, no phenotypes were observed in the single mutants (Daumann et al., 2018), indicating redundancy of the CTPS enzymes in most situations.

The Generation of Nucleoside Triphosphates

The final steps of UMP, GMP, and CTP biosynthesis occur in the cytosol. With the exception of CTP, which is directly synthesized from UTP, purine and pyrimidine nucleoside triphosphate synthesis is achieved by phosphorylation of the respective monophosphates (Fig. 4).

Figure 4.

Synthesis of NTPs and dNTPs. Synthesis of cytidylates (A), uridylates and thymidylates (B), guanylates (C), and adenylates (D). RNR (10), ribonucleotide reductase; dNK (11), deoxynucleoside kinase; UMK (12), UMP kinase; NDPK (13), nucleoside diphosphate kinase; TK (14), thymidine kinase; DHFR-TS (15), dihydrofolate reductase-thymidylate synthase; TMK (16), thymidylate kinase; GMK (17), guanylate kinase; AMK (18), adenylate kinase. The subcellular locations where enzymes with these activities are found are indicated. For TMK, a location in the plastids is only assumed. The mononucleotides (AMP, GMP, UMP, and CMP) may also be derived from salvage reactions (see Figs. 5 and 6).

The plastids and the mitochondria, which possess their own transcription and translation machineries, must be supplied with ribonucleotides and deoxyribonucleotides from the cytosol. Not much is known about (1) the phosphorylation state in which nucleotides are taken up, (2) which transporters are involved, or (3) whether the concentrations of (desoxy) nucleotides differ in the distinct cellular compartments and how this may be regulated—subcellular distributions have been estimated only for the adenylates (Stitt et al., 1982). Describing the subcellular distribution of the enzymes involved in the last two steps of mononucleotide phosphorylation can help in building hypotheses regarding the exact nucleotide species imported into organelles.

The pyrimidine nucleotides, UMP and CMP, are phosphorylated by UMP kinases (UMKs; Fig. 4, A and B, no. 12) Arabidopsis possesses two evolutionarily distinct families of such enzymes: (1) UMKs related to adenylate kinases (AMKs) encoded by four genes; and (2) UMKs related to eubacterial UMP kinases encoded by two genes. The AMK-like UMKs have not yet been characterized, except for a biochemical analysis of UMK3 (At5g26667), which was shown to utilize UMP and CMP as the best substrates (Zhou et al., 1998). These enzymes have been predicted to reside in the cytosol and the mitochondria (Lange et al., 2008). From the eubacterial UMP kinase family, one member, called plastid UMP kinase (PUMKIN; At3g18680), was shown to be located in chloroplasts and to have UMK activity in vitro. Interestingly, the enzyme binds certain plastidic transcripts and is involved in plastid RNA metabolism, which may not require its enzymatic function. Mutants are small and compromised in plastid translation and photosynthetic performance (Schmid et al., 2019). The orthologous enzyme in rice is localized in chloroplasts and participates in RNA metabolism, and the corresponding loss-of-function mutants are pale green (Zhu et al., 2016; Chen et al., 2018a). Additionally, they contain less UDP and more UMP (Dong et al., 2019), suggesting that UMP phosphorylation in the chloroplast is functionally important. The phosphorylation of thymidine 5′-monophosphate (TMP) is not catalyzed by UMKs but by a dedicated TMP kinase (TMK; Fig. 4B, no. 16). In Arabidopsis, a mitochondrial and a cytosolic version are generated from a single gene by alternative splicing. Mutation of TMK leads to early seed abortion at the zygote state (Ronceret et al., 2008).

The phosphorylation of GMP to GDP and dGMP to dGDP (Kumar et al., 2000) is catalyzed by guanylate kinases (GMK; Fig. 4C, no. 17), of which plants have two different types, a cytosolic type (GKc), and an organelle type dually targeted to plastids and mitochondria (GKpm; Fig. 4C). The activity of GKpm in rice, pea, and Arabidopsis is regulated by guanosine 3′, 5′-bisdiphosphate (ppGpp), a bacterial and plastid signaling molecule (Nomura et al., 2014). Suppression of Arabidopsis GKpm (At3g06200) transcripts by RNA interference results in a pale green or albino phenotype (Sugimoto et al., 2007), emphasizing the importance of nucleotide monophosphate transport into organelles. Interestingly, the loss-of-function mutant of the rice GKpm gene is pale green but does not exhibit DNA depletion in the organelles, suggesting that deoxynucleotide and ribonucleotide metabolism are not fully linked (Sugimoto et al., 2007).

Adenylate monophosphate is converted by adenylate kinase (AMK; Fig. 4D, no. 18) into ADP. Arabidopsis has seven AMK isoforms. AMK1, AMK2, and AMK5 are located in the plastids, AMK3 and AMK4 reside in the cytosol, and AMK7 is present in the mitochondria. For AMK1, a mitochondrial localization has also been observed (Carrari et al., 2005; Lange et al., 2008). Because the plastids harbor the de novo synthesis for AMP, they will not require net import of adenylates. Consistently, the loss-of-function mutant for AMK2 has a bleached phenotype (Lange et al., 2008), suggesting that there is no net ADP or ATP import to compensate for compromised adenlyate kinase activity in plastids. A strong reduction of plastidic AMK activity in rice also results in an albino phenotype (Wei et al., 2017). In potato (Solanum tuberosum), a reduction of plastidic AMK activity led to an increase of the adenylate pool (AMP, ADP, ATP, and ADP-Glc) and increased starch synthesis in tubers, but it is still not understood how plastidic AMK and adenylate de novo synthesis are connected (Regierer et al., 2002). Interestingly, it was suggested that the AMKs might contribute to protect RNA from random misincorporation of methyl-6 A marks. AMKs are highly selective for AMP versus N6-methyl AMP released during the degradation of RNA species carrying this abundant A modification. The selectivity of the AMKs possibly prevents the formation of N6-methyl ATP, which is a substrate of RNA polymerase II (Chen et al., 2018b).

Recently, a broad-spectrum mononucleotide kinase only distantly related to the AMKs but with relatively high adenylate kinase activity (At5g60340), was described. This enzyme was localized in the nucleus, and a knockout mutant was affected in stem elongation (Feng et al., 2012).

Plastids and mitochondria possess nucleoside monophosphate kinases for all nucleotides. Thus, nucleoside monophosphates are probably imported into these organelles and are phosphorylated to dinucleotides. Enzymes catalyzing the next step to trinucleotides, the nucleoside diphosphate kinases (NDPKs), should therefore be found in plastids and mitochondria, as well as in the cytosol. This indeed has been observed (Luzarowski et al., 2017). The exact locations of the enzymes have been debated and a detailed phylogenetic analysis suggests the presence of a fourth enzyme type in the endoplasmic reticulum (ER; Dorion and Rivoal, 2015). The NDPKs are multisubstrate enzymes accepting all nucleoside/deoxynucleoside diphosphates (Zrenner et al., 2006), but there is a preference for generating GTP (Kihara et al., 2011), which in the chloroplast may assist in repairing photosystem II (Spetea and Lundin, 2012). Mutation of the gene for the plastidic NDPK in rice results in a pale green phenotype and a lower photosynthetic rate (Ye et al., 2016; Zhou et al., 2017), but since the chloroplast function is partially retained, there also must be nucleoside triphosphate import into this organelle. Interestingly, NDPKs can also have moonlighting activity as modulators of gene expression (Dorion and Rivoal, 2018).

Besides nucleotides, the nucleus and organelles need deoxynucleotides (dNTPs) for DNA synthesis. dNTP synthesis requires the reduction of the hydroxyl moiety on the 2′ carbon of the ribose by an enzyme complex called ribonucleotide reductase (RNR; Fig. 4, no. 10). The RNR complex is comprised of two large regulatory (R1) and two small catalytic (R2) subunits. Mutation of the major R2 subunit gene (tso2) results in lower dNTP concentrations and abnormal plant development, while the additional mutation of a further R2 subunit gene (Arabidopsis has three R2 subunit genes in total) is lethal (Wang and Liu, 2006). The substrates of RNR are the ribonucleotide diphosphates, suggesting that for CTP, a dedicated phosphatase might exist to support dCDP synthesis (Fig. 4A). Alternatively, CDP for dCTP synthesis might be generated from salvage of cytidine (see section on "Purine and Pyrimidine Salvage Metabolism"). Interestingly, RNR is subject to a complex allosteric regulation to adjust the correct dNTP pool sizes (Sauge-Merle et al., 1999). In plants, RNR resides exclusively in the cytosol, with the potential to relocate to the nucleus upon exposure to UV radiation (Lincker et al., 2004). The plastidic DNA replication, especially, seems to rely strongly on sufficient RNR activity, because partially compromising the function of the large RNR subunit by different mutations in the corresponding gene resulted in reduced dNTP levels and impaired chloroplast division in Arabidopsis (Garton et al., 2007). Consistently, chlorophyll biosynthesis in rice is reduced in mutants of the small RNR subunit genes (Chen et al., 2015). All dNDPs can be synthesized directly by RNR except for thymidine 5′ diphosphate, because it has no ribonucleotide counterpart. Instead, RNR catalyzes the formation of deoxy-UDP from UDP (Fig. 4B) and deoxy-UMP is methylated at C5 to TMP catalyzed by thymidilate synthase. In Arabidopsis, three enzymes were recently characterized as thymidylate synthases, which are also dihydrofolate reductases (DHFR-TS; Fig. 4B, no. 15), with only two isoforms displaying thymidylate synthase activity (Gorelova et al., 2017). Interestingly, in roots the subcellular locations of all isoforms depends on the developmental stage of the cells: cytosolic, nuclear, and mitochondrial (but not plastidic) locations have been observed. The two active isoforms seem to be redundant, since only a double mutant of the respective genes is lethal, whereas single-gene loss-of-function mutants are phenotypically inconspicuous (Gorelova et al., 2017). The substrate for dihydrofolate reductase-thymidylate synthase is dUMP (Gorelova et al., 2017), but the RNR provides dUDP. It is unknown which enzyme links these two processes in vivo. An alternative dUMP source in mitochondria is the deamination of dCMP, as shown recently in rice (Xu et al., 2014; Niu et al., 2017).

SALVAGE AND DEGRADATION

Metabolic Sources of Nucleosides and Nucleobases

Nucleosides and nucleobases can be released from nucleotides or nucleic acids during metabolism (Figs. 2, 5, and 6) or can be taken up from the environment (Girke et al., 2014), where they can occur in substantial amounts (Phillips et al., 1997).

Figure 5.

Salvage and degradation of purines. A, Reactions of purine nucleobase and nucleoside salvage, as well as purine nucleotide degradation, which overlaps partially with GMP synthesis. The salvage pathways are highlighted by light gray shading, and the degradation reactions are encircled in dark gray. Metabolites that can only undergo degradation and cannot be salvaged are shown with brown shading. B, Purine ring catabolism. The transport steps for urate and (S)-allantoin are not shown explicitly. APRT (19), adenine phosphoribosyltransferase; ADK (20), adenosine kinase; AMPD (2), AMP deaminase; IMPDH (3), IMP dehydrogenase; GMPS (4), GMP synthetase; HGPRT (21), hypoxanthine guanine phosphoribosyltransferase; IGK (22), inosine guanosine kinase; GMPP (23), GMP phosphatase; GSDA (24), guanosine deaminase; XMPP (25), XMP phosphatase; NSH1 (26), nucleoside hydrolase 1; NSH2 (27), nucleoside hydrolase 2; XDH (28), xanthine dehydrogenase; UOX (29), urate oxidase; ALNS (30), allantoin synthase; ALN (31), allantoinase; AAH (32), allantoate amidohydrolase; UGAH (33), ureido-Gly aminohydrolase; UAH (34), ureidoglycolate amidohydrolase.

Figure 6.

Salvage and degradation of pyrimidines. NC-β-Ala, N-carbamyl-β-Ala; malonate-SA, malonate semialdehyde. UCK (35), uridine cytidine kinase; UCPP (36), UMP CMP phosphatase; CDA (37), cytidine deaminase; NSH1 (26), nucleoside hydrolase 1; PLUTO (38), plastidic nucleobase transporter; UPRT (39), uracil phosphoribosyltransferase; DPYD (40), dihydropyrimidine dehydrogenase; DPYH (41), dihydropyrimidine hydrolase; β-UP (42), β-ureidopropionase; BAAT (43) β-Ala aminotransferase.

The main metabolic source for most nucleosides is probably the turnover of RNA, in particular in the vacuole. Vacuolar RNA degradation, for example of ribosomal RNA after ribophagy (Floyd et al., 2015), generates nucleotides that likely are degraded to nucleosides by vacuolar phosphatases. The details of this process have not been investigated so far. The tonoplast membrane possesses a nucleoside exporter (equilibrative nucleoside transporter 1 [ENT1, At1g70330]; Bernard et al., 2011) for the release of nucleosides into the cytoplasm. For adenosine, the turnover of S-adenosyl Met (SAM), used for methylation reactions, is another important source (Figs. 2 and 5). After transfer of the methyl group from SAM, the resulting S-adenosyl homo-Cys (SAH) is hydrolyzed to homo-Cys and adenosine (Sauter et al., 2013).

There are no strong sources for nucleobases in plant metabolism, except for adenine, which is released during polyamine, nicotianamine, and ethylene biosynthesis (Sauter et al., 2013). In all three pathways SAM is used and 5′-methylthioadenosine is generated, which is hydrolyzed to 5-methylthioribose and adenine (Siu et al., 2008). Also the degradation of cytokinins produces small amounts of adenine (Schmülling et al., 2003). Purine bases are released from nucleic acids by spontaneous depurination (Barbado et al., 2018), resulting in low amounts of adenine and guanine. The metabolic source of hypoxanthine is likely the spontaneous deamination of adenine in DNA, resulting in a hypoxanthine base, which is removed from the DNA by base excision repair (Karran and Lindahl, 1980). Similarly, the nonenzymatic deamination of cytosine in DNA generates uracil, which is removed by base excision repair (Fig. 2).

Purine and Pyrimidine Salvage Metabolism

Nucleosides and nucleobases can be converted into nucleotides, which is called “salvage” (Figs. 2, 5, and 6). In contrast to de novo biosynthesis of nucleotides, which generates nucleotides from basic metabolites (see above), the salvage reactions recycle nucleobases and nucleosides derived from metabolism or uptake (see section on "Metabolic Sources of Nucleosides and Nucleobases") to nucleotides. Nucleobases react with activated phosphoribose (PRPP) to the respective nucleotides—a reaction catalyzed by phosphoribosyltransferases (PRTs). Adenine PRT (APRT; Fig. 5A, no. 19), hypoxanthine guanine PRT (HGPRT; Fig. 5A, no. 21), and uracil PRT (UPRT; Fig. 6, no. 39) are the three types of nucleobase-specific enzymes present in plants. Nucleosides are phosphorylated to nucleoside monophosphates by kinases. Adenosine kinase (ADK; Fig. 5A, no. 20), inosine guanosine kinase (IGK; Fig. 5A, no. 22), and uridine cytidine kinase (UCK; Fig. 6, no. 35) salvage ribonucleotides, whereas thymidine kinase (TK; Fig. 4B, no. 14) and deoxynucleoside kinase (dNK; Fig. 4, A, C, and D, no. 11) phosphorylate thymidine and the other three deoxynucleosides, respectively.

Several salvage enzymes have a critical function for plant metabolism and the mutation of the respective genes has severe consequences. Mutation of the gene for the main APRT (Fig. 5, no. 19) activity, APT1 (At1g27450), results in male sterility (Gaillard et al., 1998), whereas a strong downregulation increases the resistance to oxidative stress (Sukrong et al., 2012). Deletion or strong downregulation of the gene for the main ADK (Fig. 5, no. 20) enzyme, ADK1 (At3g09820), compromises transmethylation reactions, because the accumulating adenosine inhibits SAH hydrolase—an enzyme of the SAM cycle. It has been shown that ADK1 and SAH hydrolase interact and partially reside in the nucleus, probably mediated by nuclear methyltransferases (Lee et al., 2012). Reduced transmethylation causes a range of developmental abnormalities (Moffatt et al., 2002; Young et al., 2006). Guanine and hypoxanthine salvage seems to be less critical, because HGPRT (Fig. 5, no. 21) mutants are phenotypically normal except for a slight delay in germination (Liu et al., 2007; Schroeder et al., 2018). Mutation of HGPRT leads to guanine but not hypoxanthine accumulation in vivo, probably reflecting the fact that guanine can only be salvaged, whereas hypoxanthine can also be degraded (Baccolini and Witte, 2019). IGK has been measured in plant extracts (Fig. 5A, no. 22; Katahira and Ashihara, 2006; Deng and Ashihara, 2010), but the corresponding gene is still unknown. Some evidence has been provided that the activity is associated with the intermembrane space of mitochondria (Combes et al., 1989).

The only uracil salvage activity in Arabidopsis is located in plastids and encoded by UPP (UPRT; Fig. 6, no. 39). Mutation of UPP leads to growth arrest in the seedling stage and an albino phenotype (Mainguet et al., 2009). Interestingly, it was recently shown that this phenotype is unrelated to the lack of UPRT activity in the mutant, but is caused by the absence of the UPP protein per se, demonstrating that uracil salvage does not play such an essential role for Arabidopsis as previously thought (Ohler et al., 2019). Salvage of uridine is more prominent than salvage of uracil in Arabidopsis. Uridine and cytidine salvage are performed by dual-specific UCKs (Ohler et al., 2019). These enzymes also possess a UPRT-like domain, but they do not have UPRT activity (Chen and Thelen, 2011). Simultaneous mutation of UCK1 and UCK2 results in dwarf plants that fail to reach maturity (Chen and Thelen, 2011). Previously, UCK1 and UCK2 were found to localize in plastids (Chen and Thelen, 2011), but Ohler et al. (2019) demonstrated that these enzymes reside in the cytosol, which was also confirmed in our laboratory (M. Chen and C.-P. Witte, unpublished data).

Interestingly, deoxynucleoside-specific salvage enzymes also exist (Fig. 4). Thymidine is salvaged by thymidine kinase (TK; Fig. 4B, no. 14) and the mutation of both TK genes is lethal for Arabidopsis (Clausen et al., 2012). However, it is unclear, why thymidine salvage is of such importance. TK occurs in the cytosol, the mitochondria, and the plastids (Xu et al., 2015) and is of particular importance for chloroplast maintenance when germinating seedlings turn autotrophic (Pedroza-García et al., 2019). The other deoxynucleosides are salvaged by an enzyme with broad deoxynucleoside specificity (dNK; Fig. 4, no. 11; Clausen et al., 2012), potentially associated with mitochondria (Clausen et al., 2014).

Purine Nucleotide Degradation

Instead of being salvaged, nucleobases and nucleosides can also be fully degraded by plants, but for guanine, adenine, and adenosine, a salvage reaction needs to precede degradation. Adenine and adenosine first must be converted to AMP, which can then be deaminated by AMPD (Fig. 5A, no.2) to IMP as a first step in degradation. This is necessary because Arabidopsis and plants in general lack adenosine deaminase (Dancer et al., 1997; Chen et al., 2018b). Interestingly, for N6-methyl AMP, plants as well as many other eukaryotes possess a special deaminase, called N6-methyl-AMP deaminase (MAPDA; Chen et al., 2018b). N6-methylated adenine is the most frequent modification in mRNA, but it is also present in other RNA species (Chen and Witte, 2019). MAPDA is phylogenetically related to adenosine deaminases and hydrolyzes N6-methyl AMP to IMP, removing the aminomethyl group. This example shows that modified nucleotides must also have an access route to general nucleotide degradation.

From IMP, the purine nucleotide degradation pathway cannot be entered directly in Arabidopsis, but conversion to XMP, and apparently even to GMP, is required (Baccolini and Witte, 2019). These recent results show that the route for AMP catabolism and the route for GMP biosynthesis (partially) overlap. Therefore, branch points of both routes must be controlled, but it is not yet clear how this is achieved. GMP dephoshorylation by a yet unknown phosphatase (GMPP; Fig. 5A, no. 23) initiates purine nucleotide catabolism. At the stage of guanosine, salvage back to the nucleotide level via IGK (Fig. 5A, no. 22) is still possible, but the GMPP and IGK reactions might be spatially or temporarily separated to avoid a futile cycle. The deamination of guanosine to xanthosine by guanosine deaminase (GSDA; Fig. 5A, no. 24; Dahncke and Witte, 2013) marks the point of no return, because xanthosine cannot be salvaged and is dedicated for degradation (Yin et al., 2014). Although xanthosine appears to be generated mainly by GSDA, there is strong evidence for an alternative route directly from XMP to xanthosine catalyzed by an XMP phosphatase (XMPP; Fig. 5A, no. 25; Baccolini and Witte, 2019). An XMP-specific phosphatase, which may represent this XMPP, is currently under investigation in our laboratory. In summary, GMP catabolism begins with dephosphorylation and deamination of guanosine, and most AMP is apparently also degraded via GMP, although some AMP might be dephosphorylated already at the stage of XMP.

In clear contrast to purine metabolism in many other organisms, guanine is not an intermediate of purine nucleotide catabolism in Arabidopsis and probably most plants. For degradation, guanine must first be salvaged to GMP (Dahncke and Witte, 2013; Baccolini and Witte, 2019). In addition, in contrast to purine metabolism in many other organisms, inosine and hypoxanthine are not major intermediates of purine nucleotide catabolism in Arabidopsis, because they are derived not from IMP dephosphorylation (Baccolini and Witte, 2019), but possibly from transfer RNA turnover and base excision repair of deaminated adenine in DNA (Fig. 5A). Because guanine and hypoxanthine do not play an important role in purine nucleotide degradation, HGPRT (Fig. 5A, no. 21) is decoupled from purine catabolism in plants (Baccolini and Witte, 2019), which is in stark contrast to humans, where mutation of HGPRT results in accumulation of purine nucleotide breakdown products and severe phenotypic consequences, known as the Lesch-Nyhan Syndrome (Torres and Puig, 2007).

Purine catabolism can lead to the complete disintegration of the purine ring in plants (Fig. 5B; Werner and Witte, 2011) to recycle nitrogen (Soltabayeva et al., 2018), but it is also used to generate the intermediates uric acid and especially allantoin, which counteract stress by reducing reactive oxygen species (Brychkova et al., 2008; Watanabe et al., 2014; Irani and Todd, 2016, 2018; Lescano et al., 2016; Ma et al., 2016; Casartelli et al., 2019; Nourimand and Todd, 2019). Sometimes also the accumulation of allantoate has been observed (Alamillo et al., 2010). In tropical legumes like soybean (Glycine max) or common bean (Phaseolus vulgaris), the ureides allantoin and allantoate are used as long-distance nitrogen transport compounds for the export of fixed nitrogen from the nodules (Tegeder, 2014; Carter and Tegeder, 2016) mediated by the ureide permease (UPS) transporters (Desimone et al., 2002; Schmidt et al., 2006; Collier and Tegeder, 2012). Ureides also function in long-distance transport in non-nodulated legumes (Díaz-Leal et al., 2012; Quiles et al., 2019) and are probably used in many plants for this purpose (Lescano et al., 2016; Redillas et al., 2019).

It has recently been shown that xanthosine hydrolysis to xanthine and ribose is catalyzed by a cytosolic nucleoside hydrolase heteromer consisting of nucleoside hydrolase 1 (NSH1; Fig. 5A, no. 26) and NSH2 (Fig. 5A, no. 27) in vivo (Baccolini and Witte, 2019). NSH1 has only weak xanthosine and inosine but strong uridine hydrolase activity (Jung et al., 2009; Jung et al., 2011; Riegler et al., 2011; Baccolini and Witte, 2019). However, NSH1 is required to activate NSH2, which is the stronger xanthosine and inosine hydrolase in the complex. Nucleoside catabolism is the major metabolic source of ribose, which is recycled to ribose 5-phosphate by ribokinase in the plastids (Riggs et al., 2016; Schroeder et al., 2018). Xanthine and hypoxanthine are catabolized by the same enzyme, xanthine dehydrogenase (XDH; Fig. 5A, no. 28; Urarte et al., 2015), finally to uric acid in the cytosol. Arabidopsis has a second gene encoding XDH (At4g34900) with no apparent XDH activity in vivo (Hauck et al., 2014). Interestingly, XDH has been shown to play a dual role during powdery mildew pathogen attack on Arabidopsis (Ma et al., 2016). In the epidermis, the enzyme is postulated to operate as an NADH oxidase generating superoxide (Zarepour et al., 2010) to prevent fungal entry, whereas in the mesophyll it works as a XDH producing urate, which is suggested to function as a reactive oxygen species scavenger. It was proposed that by this mechanism the reactive oxygen species are confined to the infection site. For further degradation, uric acid must be imported into the peroxisomes, but molecular details about this import are still unknown. In the peroxisome, urate is oxidized by urate oxidase (UOX; Fig. 5B, no. 29) and is hydrolyzed and decarboxylated by allantoin synthase (ALNS; Fig. 5B, #30) to (S)-allantoin (Lamberto et al., 2010; Pessoa et al., 2010). Mutation of UOX leads to strong accumulation of uric acid, which is deleterious for peroxisome maintenance in the embryo, leading to a severe suppression of germination and seedling establishment (Hauck et al., 2014)—surprisingly, accumulation of similar amounts of xanthine in Arabidopsis plants lacking XDH does not lead to strong phenotypic alterations under standard growth conditions (Hauck et al., 2014; Schroeder et al., 2018; Soltabayeva et al., 2018). Allantoin must be transported from the peroxisomes to the ER for further degradation, but it is unclear how this is achieved. One may speculate that allantoin accumulation under certain stress conditions might be caused in part by altered peroxisome-to-ER transport efficiency. However, one apparent reason for allantoin accumulation is an altered catalytic capacity for allantoin generation and degradation under stress (Irani and Todd, 2016; Lescano et al., 2016; Irani and Todd, 2018; Casartelli et al., 2019).

In the ER, allantoin is hydrolyzed by four enzymes (Fig. 5B, nos. 31–34) completely releasing the ring nitrogen as ammonia (Todd and Polacco, 2006; Werner et al., 2008; Serventi et al., 2010; Werner et al., 2010). These enzymes are also responsible for supplying the shoots of tropical legumes with nitrogen exported from the nodules as allantoin and allantoate (Werner et al., 2013; Díaz-Leal et al., 2014).

Pyrimidine Nucleotide Degradation

Pyrimidine nucleotide catabolism is initiated by UMP/CMP phosphatase(s) (UCPP; Fig. 6, no.36), which have not yet been identified. Their activity might be temporarily and/or spatially separated from UCKs (Fig. 6, no. 35; Ohler et al., 2019) to avoid a futile cycle of pyrimidine nucleotide dephosphorylation and pyrimidine nucleoside salvage. Cytidine is deaminated to uridine by a cytosolic cytidine deaminase (CDA; Fig. 6, no. 37). Interestingly, plants can neither degrade nor salvage the free base cytosine (Katahira and Ashihara, 2002). Arabidopsis contains several copies of CDA, but only one copy is functional. The mutation of CDA results in smaller plants, probably because cytidine accumulation is toxic (Chen et al., 2016). Generally, the accumulation of nucleosides can reduce plant performance, as has been shown for GSDA (Fig. 5A, no. 24) mutants (Schroeder et al., 2018). Consistently, transgenic lines with increased vacuolar nucleoside export (Bernard et al., 2011) are smaller than the wild type.

Uridine is hydrolyzed by the cytosolic NSH1 (Fig. 6, no. 26) to uracil and ribose (Jung et al., 2009). NSH2 is not involved in uridine hydrolysis. NSH1 occurs in two forms in vivo, either as a homomer (probably a homodimer; Kopecná et al., 2013) for uridine hydrolysis, or as a heteromer interacting with NSH2 for xanthosine and inosine hydrolysis (Baccolini and Witte, 2019). One should note that these nucleoside hydrolases usually do not hydrolyze cytidine (Jung et al., 2009) or guanosine in vivo unless these compounds accumulate in catabolic mutants (Dahncke and Witte, 2013; Chen et al., 2016; Baccolini and Witte, 2019). Adenosine is a substrate of NSH1 and also of 5′-methylthioadenosine nucleosidase 2 (MTAN2; Siu et al., 2008) in vitro, but both enzymes hydrolyze adenosine with very low catalytic efficiency. The main adenosine hydrolytic activity of Arabidopsis resides probably in the apoplast (see below).

The plastidic nucleobase transporter (PLUTO; Fig. 6, no. 38) reallocates uracil, probably in symport with protons from the cytosol, into the plastids for further metabolic conversion (Witz et al., 2012). PLUTO belongs to the Nucleobase:Cation Symporter 1 family and transports guanine and adenine in addition to uracil, albeit with lower efficiency. There are indications that a thiamine precursor, hydroxymethylpyrimidine, is also a PLUTO substrate (Beaudoin et al., 2018). Interestingly, it was recently reported that PLUTO orthologs from two grasses do not transport uracil, but only adenine and guanine next to a few other substrates (Rapp et al., 2016), indicating that uracil metabolism might be organized differently in these species. However, one should note that definitive evidence for a function of PLUTO in uracil transport into plastids in vivo has not yet been presented for any plant. Other transporters capable of uracil transport have been identified, but they are located in the plasma membrane (Schmidt et al., 2004; Niopek-Witz et al., 2014).

In the plastid, there is a branch point: uracil can either be salvaged by UPRT (Fig. 6, no. 39; see section on "Purine and Pyrimidine Salvage Metabolism") or be degraded. When uracil is applied from outside, strong catabolic activity is usually observed (Ashihara et al., 2001; Katahira and Ashihara, 2002). The first reaction, in which the uracil ring is reduced to dihydrouracil by dihydropyrimidine dehydrogenase (DPYD/PYD1; Fig. 5, no. 40) residing in plastids, was shown to be rate limiting (Tintemann et al., 1985). Compared to mammalian DPYD, the plant enzyme lacks C-terminal domains for cofactor binding, which are involved in electron delivery to the active site. Therefore, the plant enzyme is probably incomplete and might require a so far unknown interaction partner for activity. The loss of activity when the enzyme is expressed in the cytosol instead of the plastid is in agreement with this hypothesis (Cornelius et al., 2011). Mutants of DPYD/PYD1 show delayed germination and a misregulation of abscisic acid-responsive genes, whereas constitutive overexpression results in an increase in growth and seed number (Cornelius et al., 2011). In the next enzymatic steps, the dihydrouracil ring is opened by dihydropyrimidine hydrolase (DPYH/PYD2; Fig. 6, no. 41) and then the carbamino group is hydrolytically released by β-ureidopropionase (β-UP/PYD3; Fig. 6, no. 42; Walsh et al., 2001), generating β-Ala. Not only uracil, but also 5-methyluracil (thymine), is degraded in this pathway (Cornelius et al., 2011), resulting in β-aminoisobutyrate instead of β-Ala. The first reaction (DPYD; Fig. 6, no. 40) is located in the plastids, the second (DPYH; Fig. 6, no. 41) in the ER, and the third (β-UP; Fig. 6, no. 42) in the cytosol, but it is unclear why such a distribution is favorable, and how the metabolites are shuttled to these different locations (Zrenner et al., 2009). Recently, the combination of genome-wide association data with correlation networks built from metabolite and transcriptome data identified an aminotransferase correlated with β-Ala. The corresponding mutants accumulated β-Ala, indicating that this might be the missing β-Ala aminotransferase (BAAT/PYD4; Fig. 6, no. 43) of pyrimidine catabolism in plants (Wu et al., 2016).

Pyrimidine catabolism is induced by nitrogen starvation and in senescence (Zrenner et al., 2009; Cornelius et al., 2011), suggesting that, similar to purine nitrogen, pyrimidine nitrogen is also recycled by plants. When uracil is given as the sole nitrogen source, its degradation can support the growth of Arabidopsis to a limited extent (Zrenner et al., 2009).

eATP

Extracellular ATP (eATP) is a signal molecule that is either actively released upon a stimulus by plant cells via exocytosis or transport or is derived from damaged cells (Fig. 7; Cao et al., 2014). A plasma membrane-based nucleotide transporter belonging to the mitochondrial carrier family (pmANT1; Fig. 7, no. 44) is involved in ATP export with physiological relevance, at least in pollen (Rieder and Neuhaus, 2011). eATP plays a role in stress responses and is perceived by the receptor-like kinase does not respond to nucleotides 1 (DORN1; Fig. 7, no. 50), which recognizes ATP, GTP, and ADP, but not AMP and adenosine (Choi et al., 2014). Interestingly, CTP and NAD are also sensed by plant cells, and a potential receptor for NAD has been identified recently (Wang et al., 2017).

Figure 7.

Excretion, perception, and degradation of extracellular ATP. pmANT1 (44), plasma membrane adenine nucleotide transporter; APY3 (45), apyrase 3; 5′ NT (46), 5′-nucleotidase; ENT3 (47), equilibrative nucleoside transporter 3; NSH3 (48), nucleoside hydrolase 3, AZG1/2 (49), azaguanine resistant 1 / azaguanine resistant 2, DORN1 (50), does not respond to nucleotides 1.

The ATP signal might be quenched by an apoplastic apyrase (Riewe et al., 2008a), an enzyme that hydrolyzes NTPs or NDPs to nNMPs. Seven apyrases are encoded by the Arabidopsis genome (APY1–APY7), and APY1 and APY2 were believed to represent these extracellular enzymes (Lim et al., 2014). However, by scanning the substrate spectra of the apyrases, only APY3 (Fig. 7, no. 45) showed strong activity with ATP (and other NTPs), though APY5 and APY6 also were slightly active with NTPs (Chiu et al., 2015). These apyrases are therefore possible candidates for the apoplastic enzymes in Arabidopsis. However, other secreted phosphatases might also be involved, for example members of the unspecific purple acid phosphatases (Wang et al., 2011; Del Vecchio et al., 2014).

The AMP resulting from ATP dephosphorylation is hydrolyzed in the apoplast to adenosine by a 5′-nucleotidase (Fig. 7, no. 46). An AMP-specific extracellular 5′-nucleotidase associated to the plasma membrane was purified from peanut (Arachis hypogaea; Sharma et al., 1986; Gupta and Sharma, 1996), but the corresponding gene has not been identified. Adenosine can be either taken up via the adenosine proton symporter equilibrative nucleoside transporter 3 (ENT3; Fig. 7, no. 47; Traub et al., 2007; Cornelius et al., 2012) or further hydrolyzed by the apoplastic purine-specific NSH3 (Fig. 7, no. 48; Jung et al., 2011) to adenine and ribose. NSH3 hydrolyzes inosine more efficiently than adenosine, whereas a cell wall-bound nucleoside hydrolase of potato, probably the ortholog of NSH3 in this plant, was highly specific for adenosine and did not hydrolyze inosine (Riewe et al., 2008b).

Simultaneous genetic blockage of nucleoside uptake and hydrolysis leads to an accumulation of adenosine and uridine in the apoplast, a reduction of PSII efficiency, and a higher susceptibility to the necrotrophic fungus Botrytis cinerea, possibly caused by reduced expression of WRKY33 (Daumann et al., 2015), which is known to be essential for Botrytis resistance (Liu et al., 2015). Treatment with eATP increases the resistance to Botrytis (Tripathi et al., 2018), and the expression of WRKY33 and other defense-related genes is reduced in a DORN1 mutant and boosted in a DORN1 overexpression line upon challenge with eATP (Jewell et al., 2019). Taken together, these observations suggest that adenosine accumulation in the apoplast dampens the DORN1-mediated response, indicating that ENT3 and NSH3 are required to remove the breakdown products of eATP signaling. Also the adenine resulting from adenosine hydrolysis by NSH3 is taken up by plant cells. It is not entirely clear which transporters mediate this uptake. Possible candidates are azaguanine resistant 1 (AZG1) and AZG2 (Fig. 7, no. 49), which have been shown to facilitate adenine and guanine uptake in Arabidopsis seedlings (Mansfield et al., 2009), or members of the nucleobase-ascorbate transporter family (Niopek-Witz et al., 2014) as well as of the purine permease (PUP) family (Girke et al., 2014). Interestingly, fungi seem to be able to influence the purinergic signaling in the apoplast by interfering with apoplastic nucleotide metabolism via the excretion of nucleotidases to improve colonization (Nizam et al., 2019).

CONNECTIONS TO CYTOKININ HOMEOSTASIS

The biosynthesis of the cytokinins involves the generation of N6-modified AMP, carrying an isoprenoid group (Sakakibara, 2005). However, cytokinin ribotides or ribosides are inactive—the free modified base is the active hormone binding to the receptors (Yamada et al., 2001; Romanov et al., 2018). The question arises whether the enzymes that are employed for cytokinin homeostasis (activation from ribotides and inactivation to ribosides/ribotides) are the same as for the metabolism of adenine nucleotides.

It was shown that a cytokinin ribotide-specific enzyme (cytokinin riboside 5′-monophosphate phosphoribohydrolase, called lonely guy [LOG]) can release the active cytokinin from the ribotide (Kurakawa et al., 2007; Kuroha et al., 2009). A recent report demonstrated that mutation of the seven genes coding for functional LOGs in Arabidopsis resulted in a phenotype that cannot be attenuated by exogenous cytokinin ribotides, suggesting that the hydrolysis by LOGs is the main pathway of cytokinin activation (Osugi et al., 2017). Therefore, the cytosolic nucleoside hydrolases do not seem to be involved in cytokinin activation, although it could be shown that cytokinin ribosides are substrates in vitro, catalyzed with comparatively low efficiency (Jung et al., 2009; Kopecná et al., 2013). Consistently, cytokinin-related phenotypes were not observed in nucleoside hydrolase mutants (Riegler et al., 2011). However, long-distance transport of cytokinins may involve an activation of cytokinin ribosides/ribotides in the apoplast prior to uptake or perception (Romanov et al., 2018). In Arabidopsis, NSH3, located in the apoplast (see section titled “eATP”), has been shown to hydrolyze adenosine, but cytokinin ribosides have not been assessed (Jung et al., 2011). Interestingly, an apoplastic nucleoside phosphorylase was isolated from potato that converted cytokinin ribosides to cytokinins and ribose-1-phosphate in the presence of phosphate, and that can also work in the synthesis direction of ribosides (Bromley et al., 2014). The enzyme preferred cytokinins/cytokinin ribosides over adenine/adenosine as substrates and is supposedly involved in cytokinin-mediated tuber endodormancy. Close homologs in Arabidopsis (e.g. At4g28940) are predicted to be located in the apoplast (SUBA web server; Hooper et al., 2017), but have not yet been characterized.

The inactivation of cytokinins is inter alia achieved by transferring a phosphoribosyl moiety onto the active cytokinin, resulting in an inactive cytokinin-ribotide. This reaction is at least partially performed by an APRT (Fig. 5A, no. 19), because (1) APT1 mutants convert cytokinins less efficiently to the respective ribotides (Moffatt et al., 1991), (2) APT1 loss-of-function plants contain more active cytokinins (Zhang et al., 2013), and (3) APT1 mutants have cytokinin-related phenotypes (Gaillard et al., 1998a). APT1 is also clearly involved in the salvage of adenine, which is less efficiently converted to AMP in an APT1 mutant (Moffatt et al., 1991) and is slightly more abundant in a mutant plant with reduced APT1 activity (Sukrong et al., 2012). In conclusion, APT1 participates in adenine and cytokinin metabolism and the question arises how one enzyme can serve both distinct metabolic roles adequately. There is also some evidence that ADK (Fig. 5A, no. 20) contributes to cytokinin homeostasis, because plants with reduced ADK activity show cytokinin-related phenotypes and contain more cytokinin ribosides (Schoor et al., 2011). However, ADK is also involved in the maintenance of transmethylation activity; therefore, an indirect impact of reduced ADK activity on cytokinin homeostasis cannot be fully excluded.

CONCLUSIONS

The synthesis, interconversion, and degradation of nucleotides is intrinsically linked with the propagation and reading of genetic information, with energy metabolism including the metabolic activation of many biomolecules, but also with methylation reactions, signal transduction, the recycling of nitrogen, and the modification of oxidative stress. We are getting closer to completing a full inventory of enzymes involved in plant nucleotide metabolism, but we are far from understanding how these enzymes operate together to achieve nucleotide homeostasis. There is only fragmentary information about regulation on all levels, transport processes are incompletely defined, and the organization of nucleotide metabolism at the tissue and organ levels is not well understood (see Outstanding Questions). These issues need to be addressed to allow a better integration of nucleotide metabolism into a molecular model of plant physiology.