Introduction

Polyamines (PAs) are small polycationic molecules bearing amino groups. Diamine putrescine (Put), triamine spermidine (Spd), and tetraamine spermine (Spm) are the most common PAs found in higher organisms. Put and Spd are ubiquitous in all living cells, whereas Spm is present mostly in eukaryotes and some bacteria (Pegg and Michael 2010). Chemical or genetic depletion of Put and/or Spd levels is lethal for many organisms, and Spm deficiency can lead to growth malfunction, thus suggesting essential roles in growth and development (Imai et al. 2004a, b; Kusano et al. 2008). The biological functions of PAs were initially associated with their ability to bind anionic macromolecules, and thus they were considered to be polycations with unique structural roles. Later studies showed that PAs also act as regulatory molecules in fundamental cellular processes, including cell division, differentiation, gene expression, DNA and protein synthesis, and apoptosis in many organisms (Seiler and Raul 2005; Childs et al. 2003; Igarashi and Kashiwagi 2010a; Alcázar et al. 2010a). In plants, PAs are implicated in physiological processes, including organogenesis, embryogenesis, floral initiation and development, leaf senescence, pollen tube growth, fruit development and ripening, response to abiotic and biotic stresses. The PA metabolic route is a central pathway for intermediary nitrogen metabolism, and interactions with other metabolites, including stress-protective compounds, hormones, and signaling molecules, have been postulated (Kusano et al. 2008; Alcázar et al. 2010a; Handa and Mattoo 2010). Chemical and/or genetic manipulation of PA metabolism in plants has direct agronomic applications aimed at increasing phytonutrient content, fruit quality, and plant protection against abiotic stresses and various diseases as well as pharmaceutical uses in the formation of Put-derived alkaloids (Walters 2003; Alcázar et al. 2006, 2010a; Kusano et al. 2008). Collectively, the potential applications of PAs are broad, but using them requires understanding the processes by which PAs exert their functions. In this review, we focus on investigations of PAs in development and stress for which a molecular and/or genetic basis has been revealed.

Polyamine metabolism and transport

Before we consider the involvement of PAs in development and stress, it is necessary to introduce how PA homeostasis is achieved by transcriptional, translational and posttranslational control of genes and enzymes involved in the biosynthesis, catabolism, conjugation, and transport of PAs. Here we will focus on essential biochemical and evolutionary aspects of PA metabolism and transport. The integration of PAs in the context of broader metabolic networks can be found in recent reviews (Kusano et al. 2008; Moschou et al. 2008c; Alcázar et al. 2010a; Bitrián et al. 2012).

ADC and ODC pathways

Almost all eukaryotes synthesize Put directly from ornithine in a reaction catalyzed by ornithine decarboxylase (ODC) (Fig. 1). However, plants have an alternative pathway for Put formation, in which Put is synthesized from arginine by the action of arginine decarboxylase (ADC) followed by two successive steps catalyzed by agmatine iminohydrolase (AIH) and N-carbamoyl-Put amidohydrolase (CPA) (Fuell et al. 2010) (Fig. 1). ADC, AIH and CPA genes in plants are thought to derive from the cyanobacterial ancestor of the chloroplast by endosymbiotic gene transfer to the nuclear genome of the plant cell (Illingworth et al. 2003). The plant ADC differs from the bacterial ADC in the presence of an N-terminal transit peptide that directs the protein to the chloroplast. Indeed, ADC protein and activity have been detected in chloroplasts of oat and tobacco (Borrell et al. 1995; Bortolotti et al. 2004) whereas ODC has mainly been found in the nucleus and cytosol (Fuell et al. 2010). ODC genes have been identified in several plant species, in most cases as single copy (Fuell et al. 2010). However, ODC is absent in many species of the Brassicaceae, thus indicating its dispensability for plant survival in the presence of an alternative pathway for Put biosynthesis. A similar result has been found in yeast, in which the co-transformation with plant ADC and bacterial agmatine ureohydrolase (AUH) reconstitutes Put biosynthesis and complements the growth defects of ODC yeast mutants (Klein et al. 1999).

Fig. 1
figure 1

Polyamine metabolic network implicated in development and stress. Connections are shown between the polyamine (PA) biosynthetic pathway (central) with ethylene biosynthesis, PA conjugation to hydroxycinnamic acids, tropane and pyrrolizidine alkaloids biosynthesis, and eiF5A hypusination. The biological functions of each of the branches are indicated. ADC arginine decarboxylase, ACL5 acaulis5, AIH agmatine iminohydrolase, AtTSM1 Arabidopsis thaliana Tapetum-Specific Methyltransferase 1, CPA N-carbamoyl-Put amidohydrolase, CYP98A8/A9 cytochrome P450 98A8/A9, DHS deoxyhypusine synthase, DOHH deoxyhypusine hydroxylase, dSAM decarboxylated SAM, HSS homospermidine synthase, Met methionine, MTA 5′-methylthioadenosine, ODC ornithine decarboxylase, PMT Put N-methyltransferase, SAM S-adenosylmethionine, SAMDC S-adenosylmethionine decarboxylase, SHT Spd hydroxycinnamoyltransferase, SPDS spermidine synthase, SPMS spermine synthase

Full size image

The co-existence of both ADC and ODC pathways in some species may relate to their differential contribution to stress, development and tissue specificity. Generally, the ADC pathway is activated under stress whereas ODC is required for cell proliferation and Put biosynthesis in roots (Carbonell and Blázquez 2009). In plants missing ODC, duplication of ADC genes might be a compensatory mechanism (Hanfrey et al. 2001). Early in the evolution of the Brassicaceae, the duplication of an ancestral ADC generated two paralogs (ADC1 and ADC2) that have been maintained even after the genome shrinkage in species like Arabidopsis thaliana (Galloway et al. 1998). This suggests that conservation of ADC paralogs in the Brassicaceae might underlie some evolutionary benefit.

Studies in animals evidenced the inhibitory effect of high PA levels on ODC activity by promoting ODC enzyme degradation. This feedback regulation is dependent on the synthesis of a non-competitive ODC inhibitor (ODC antizyme, ODC-AZ) that binds ODC and mediates its ubiquitin-independent degradation by the 26S proteasome. The levels of ODC-AZ protein are regulated by ribosomal frameshifting modulated by PAs, in such a way that accumulation of PAs promotes the translation of the full-length ODC-AZ (Kahana 2007). In contrast to mammalian and fungal ODC, plant ODC is insensitive to exogenous application of Spd or Spm, and its degradation is not feedback regulated by PAs (Hiatt et al. 1986; Fuell et al. 2010; Illingworth and Michael 2012). Furthermore, plant ODC cannot bind the human ODC-AZ, which supports the view that this mechanism for ODC regulation is absent in plants (Illingworth and Michael 1998, 2012).

A wide range of abiotic and biotic stresses induce the expression of ADC, leading to Put accumulation (Alcázar et al. 2010a). In addition to this, posttranslational processing of ADC is required for its activity in some species studied, like oat (Malmberg and Cellino 1994). However, this is not always necessary, as the unprocessed carnation ADC retains its activity (Chang et al. 2000).

AIH and CPA plant genes were first identified in Arabidopsis (Piotrowski et al. 2003; Janowitz et al. 2003). The Arabidopsis AIH encodes a 43-kDa protein that forms homodimers (Janowitz et al. 2003). Arabidopsis CPA has the typical domains found in the superfamily of C–N hydrolases and forms a complex of eight monomers with high affinity to N-carbamoylputrescine (Piotrowski et al. 2003). In contrast to ADC, AIH and CPA transcripts are generally not increased under stress. However, overexpression of ADC is sufficient to induce Put accumulation in plants (Alcázar et al. 2005) and thus, ADC is considered the rate-limiting enzyme in Put synthesis (Alcázar et al. 2010a).

Aminopropyltransferases

Triamines (Spd) and tetraamines (Spm and tSpm) are synthesized by aminopropyltransferases (APTs), which transfer aminopropyl residues to amine acceptors (Put or Spd), forming Spd, Spm, or its isomers (Fig. 1). The donor of the aminopropyl groups is decarboxylated S-adenosylmethionine (dSAM). dSAM is formed through the decarboxylation of S-adenosylmethionine (SAM), a universal methyl donor, by SAM decarboxylase (SAMDC). 5′-Methylthioadenosine (MTA) resulting from dSAM is recycled after Spd or Spm biosynthesis through the methionine cycle to regenerate SAM (Shao et al. 2012) (Fig. 1).

SPDS likely originated from a common ancestor prior to the separation of prokaryotes and eukaryotes, followed by the acquisition of a new activity, SPMS, that evolved independently in plants, fungi and animals. Conversely, a change of SPDS to tSPMS activity in archaea and bacteria seems to be the origin of tSPM in plants by horizontal gene transfer (Minguet et al. 2008).

Most characterized APTs form homodimers with the exception of thermophiles, in which APTs form tetramers (Belda-Palazón et al. 2012). In many dicots, such as Arabidopsis, SDPS are encoded by gene paralogs (SPDS1 and SPDS2), and their expression depends on tissue, developmental stage and environment (Alcázar et al. 2010a). Arabidopsis SPDS1, SPDS2 and SPMS are localized in the nucleus and cytosol and can form homo- and heterodimers. Interestingly, heterodimers are only found in the nucleus, thus suggesting a nuclear localization of the metabolic channeling from Put to Spm found in Arabidopsis (Panicot et al. 2002; Alcázar et al. 2011; Belda-Palazón et al. 2012). Consistent with a different evolutionary origin, the Arabidopsis tSPMS (ACAULIS 5, ACL5) does not belong to this complex of APTs (Panicot et al. 2002) and has specific roles in vascular tissue development and stem elongation as discussed in detail below (see “The role of thermospermine in stem and vascular development”). Actually, ACL5 expression in Arabidopsis is induced by auxins, whereas SPMS expression is induced by abscisic acid (ABA) (Hanzawa et al. 2000, 2002). Furthermore, ACL5 expression is restricted to vascular tissue, whereas SPMS does not show a particular cell- or tissue-specific patterns (Clay and Nelson 2005; Muñiz et al. 2008).

S-adenosylmethionine decarboxylase

dSAM is exclusively used in PA production, because no other metabolic reactions are known to require this substrate. This has been the basis for the proposed opposite effect between PAs and ethylene, because both biosynthetic pathways share SAM as substrate (Bitrián et al. 2012) (Fig. 1). dSAM concentrations in the cell are usually low and constitute a rate-limiting substrate in the synthesis of higher PAs. The transcriptional and posttranscriptional regulation of SAMDC has been extensively studied in many organisms, including plants. SAMDC is synthesized as an inactive proenzyme which is autocatalytically cleaved to generate two subunits, α and β. In contrast to mammalian SAMDC, the autocatalytic processing is not inhibited by Put in plants (Xiong et al. 1997). The transcript leader sequence of plant SAMDC contains two conserved upstream open reading frames (uORFs) of 3–4 codons (tiny uORF) and 50–54 codons (small uORF) overlapping in one nucleotide. The poor Kozak sequence found in the tiny uORF suggests that the ribosome scanning would preferentially recognize the translation initiation codon of the small uORF, which is responsible for the translational suppression of the downstream SAMDC ORF. Evidence indicates that it is not the mere presence of the uORF sequence alone that is required for translational repression, as modification of the codon frame is not translationally repressive and most polymorphisms found in the uORF sequences are conservative (Hanfrey et al. 2002).

Arabidopsis has four genes coding for SAMDC (Ge et al. 2006). At the transcriptional level, SAMDC expression is induced by different abiotic stresses, and its overexpression often correlates with enhanced tolerance to different types of stress (Wi et al. 2006). However, in some cases, stress-induced increases in SAMDC expression do not correlate with higher Spd and/or Spm levels, probably owing to tight regulation of the enzyme activity.

Polyamine oxidation

PAs are oxidatively deaminated by amine oxidases (AOs) in enzymatic reactions that produce H2O2 (Fig. 2). Based on the cofactor involved, AOs are classified into copper-containing amine oxidases (CuAOs) and FAD-dependent polyamine oxidases (PAOs) which differ in their substrate preferences, mechanism of substrate oxidation, and subcellular localization.

Fig. 2
figure 2

Polyamine oxidation pathways present in the cytosol, apoplast and peroxisomes. PAs are oxidatively deaminated by copper-containing amine oxidases (CuAOs) and FAD-dependent amine oxidases (PAOs) producing the corresponding aminoaldehydes and H2O2. CuAO oxidation produces 4-aminobutanal from Put and 4-aza-8-amino-octanal or 4,9-diaza-dodecan-1,12 dialdehyde from Spd and Spm, respectively. PAOs catalyze the terminal oxidation of Spd or Spm producing 4-aminobutanal or 3-(aminopropyl)-4-aminobutanal (APAL), respectively, or the back conversion of Spm to Spd and from Spd to Put producing APAL. AMADH aminoaldehyde dehydrogenase, DAP diaminopropane, GABA γ-aminobutyric acid

Full size image

Copper-containing amine oxidases

CuAOs are homodimeric enzymes with high affinity for oxidizing the primary amino groups of Put and cadaverine (Cad) and lower affinity for Spd and Spm (Moschou et al. 2012). Put oxidation in plants produces 4-aminobutanal, which spontaneously cyclizes to Δ1-pyrroline and can be further converted to γ-aminobutyric acid (GABA) by an amino aldehyde dehydrogenase (AMADH) (Petrivalský et al. 2007) (Fig. 2). Plant CuAOs occur at high levels in Fabaceae and are loosely associated with cell walls (Cona et al. 2006). Arabidopsis carries ten putative CuAO-encoding genes, four of which (ATAO1 and AtCuAO1–3) have been characterized. The expression of CuAO genes is differentially modulated during development, wounding, and treatment with hormones or elicitors. CuAO proteins also differ in their localization, being AtCuAO1 and ATAO1 apoplastic, whereas AtCuAO2 and AtCuAO3 peroxisomal (Møller and McPherson 1998; Reumann et al. 2009; Planas-Portell et al. 2013). Tobacco has two CuAO-encoding genes (NtMPO1 and NtMPO2) with high sequence homology to AtCuAO3, although they preferentially oxidize N-methyl-Put, a reaction essential for pyrrolidine alkaloid biosynthesis. Consistently, these genes are specifically expressed in roots, the tissue where alkaloids are synthesized (Heim et al. 2007; Katoh et al. 2007). Based on sequence homology with CuAO genes, NtMPOs are thought to evolve from ancestral CuAOs that acquired novel functionalities in nicotine biosynthesis.

FAD-dependent amine oxidases

PAOs catalyze the oxidation of Spd, Spm, and/or acetylated derivatives at their secondary amino groups (Tavladoraki et al. 2012). They are classified into one of two families depending on whether they terminally oxidize PAs or catalyze PA back conversion. PAOs of the first family are found only in plants and bacteria. They oxidize the carbon at the endo side of the N 4 of Spd and Spm, producing 4-aminobutanal and N-(3-aminopropyl)-4-aminobutanal (APAL), respectively (Moschou et al. 2012) (Fig. 2). These PAOs are present in high quantities in certain tissues of plants belonging to the Gramineae family (Angelini et al. 2010). Maize has three genes (ZmPAO1–3) encoding identical extracellular proteins (Cervelli et al. 2000), and barley contains two isoforms with different cellular localizations: HvPAO1, found in the apoplast, and HvPAO2, found in the vacuole (Cervelli et al. 2000, 2001, 2004).

PAOs catalyzing PA back conversion oxidize the carbon at the exo side of the N 4 of Spd and Spm (and/or their acetylated derivatives) producing Put and Spd, respectively (Fig. 2). Intriguingly, this reaction is inhibited by Put through a yet unknown mechanism. The best-characterized plant PAOs involved in PA back conversion are from A. thaliana. This species has four PAO-encoding genes (AtPAO1–4) as well as a fifth (AtPAO5) that contains conserved functional PAO motifs (Tavladoraki et al. 2006; Kamada-Nobusada et al. 2008; Takahashi et al. 2010; Ahou et al. 2014). In addition, Arabidopsis has four other genes that share considerable homology with FAD-dependent PAOs, whose potential function is discussed below (see “Gene expression regulation”). Tissue- and organ-specific expression studies of various AtPAO genes evidence some overlapping patterns but also remarkable differences. This, together with their contrasted substrate specificity, suggests a functional diversity of AtPAO genes (Takahashi et al. 2010). The different subcellular localizations of AtPAO proteins also support this view: AtPAO2–4 are localized in peroxisomes, whereas AtPAO1 and AtPAO5 are predicted to be cytosolic. Thus, PA catabolism in the Arabidopsis apoplast is mediated predominantly by CuAOs. The genes encoding the peroxisomal proteins AtPAO3 and AtCuAO3 are induced by ABA, salicylic acid (SA), jasmonic acid (JA), and the pathogen-associated molecular pattern flagellin 22, suggesting a cooperation between both catabolic pathways in these organelles to external stimuli (Moschou et al. 2008c; Planas-Portell et al. 2013).

Oxidation of PAs by AOs not only contributes to the regulation of PA homeostasis but also generates catabolic products linked to biological functions (Cona et al. 2006; Angelini et al. 2010). Thus, H2O2 generated in all AO-catalyzed reactions contributes to stomatal opening and programmed cell death, and GABA produced by terminal oxidation of Put and Spd plays a role in stress signaling (see “Abiotic stress and polyamine oxidation”) (Angelini et al. 2010; Shelp et al. 2012). The view of PA catabolism has shifted from a mere mechanism for regulating PA levels to a participative metabolic pathway in response to different types of abiotic, biotic, and developmental signals.

Polyamine conjugation

PAs occur in plants not only as free forms but also as conjugates to hydroxycinnamic acids (HCAs) to form hydroxycinnamic acid amides (HCAAs). In contrast to previous thinking that considered conjugates to be inactive forms, evidence indicates that they are essential for certain developmental processes. A large proportion of PAs in plants are mono-, di-, or tri-substituted with phenolic acids such as coumaric, caffeic, or ferulic acids (Bassard et al. 2010) (Fig. 3). PA conjugates are not rare since hydroxycinnamoyl Put, Spd, and Spm conjugates are found in many plant species. An emerging area of study is the characterization of PA conjugate biosynthetic pathways and the enzymes involved. HCAAs are produced by N-acylation of PAs with coenzyme A (CoA)-activated thioesters in reactions catalyzed by N-acyltransferases (Fuell et al. 2010). The agmatine coumaroyltransferase was the first N-hydroxycinnamoyltransferase identified in plants (Burhenne et al. 2003). Recently, genes encoding several Spd N-acyltransferases have been characterized in Arabidopsis (Fuell et al. 2010). Substantial evidence supports a role for hydroxycinnamoyl-Spd conjugates in pollen development, as discussed below.

Fig. 3
figure 3

Chemical structure of hydroxycinnamic acids (HCA; p-coumaroyl, caffeoyl, dihydrocaffeoyl, feruloyl, hydroxyferuloyl and sinapoyl) to which PAs can be bound by N-acylation to form mono-, di- or tri-conjugated hydroxycinnamic acid amides (HCAA). The putrescine backbone is indicated in red and aminopropyl moieties added to form spermidine, spermine and thermospermine are depicted in blue. R1 and R2 indicate substituents of the different HCA

Full size image

Polyamine transport

The extensive literature on PA transport in Escherichia coli and Saccharomyces cerevisiae (yeast) contrasts with the limited knowledge of mechanisms underlying PA transport obtained from plants, despite early evidence pointing to the occurrence of such mechanisms and the detection of PAs in phloem and xylem sap in different plant species (Pistocchi et al. 1987; Antognoni et al. 1998).

In plants, the first attempts to identify PA transporters have been based on surveys for orthologous sequences of known systems in other organisms. So far, few PA transporters have been identified in plants, mainly in rice and Arabidopsis. An Spd-preferential uptake system (OsPUT1), analog to yeast amino acid permease AGP2, has been characterized in rice using heterologous expression and yeast complementation assays (Mulangi et al. 2012). Except for Spm, PAs (such as Put and Cad) as well as the PA precursors ornithine and agmatine did not compete for Spd uptake mediated by OsPUT1. Consistent with the function of AGP2 as a general amino acid permease (Igarashi and Kashiwagi 2010b), Spd uptake was reduced by competition with some amino acids (Mulangi et al. 2012). However, OsPUT1 could not be localized to the plasma membrane of transiently transformed onion or rice cells. The need for posttranslational modifications might cause the mislocalization of OsPUT1 (Mulangi et al. 2012).

A recent report indicates that the evolution of PA transport might be determined by the acquisition of functional novelties induced by random mutations (Brill et al. 2012). Multidrug transporters are involved in moving toxic compounds outside the cell (in the case of microorganisms) or to subcellular organelles (in plants). EmrE is one of the most well-characterized multidrug transporters in E. coli. A genetic screen for EmrE mutations conferring resistance to the antibiotic erythromycin identified a single amino acid substitution that impaired EmrE functionality but provided PA importer activity (Brill et al. 2012). This highlights the enormous capacity for evolutionary novelties in PA transporters and the potential impact of naturally occurring genetic variation for such genes. Indeed, a recent screen for natural variation in paraquat resistance has identified an Arabidopsis L-type amino acid transporter (LAT), called Resistant to Methylviologen 1 (RMV1), that is responsible for the uptake of paraquat and PAs. Genetic variation at RMV1 provides differential responses in root growth to exogenous paraquat and PA applications. Consistently, RMV1 localizes to the plasma membrane of Arabidopsis plants (Fujita et al. 2012). Essentially, evidence demonstrates that in prokaryotes and eukaryotes, PA transport is not a passive mechanism, and selectivity is achieved through specific recognition and translocation that can be affected by induced mutations and natural genetic variation.

Polyamines and development

Early studies about the involvement of PAs in plant development focused on the phenotypic consequences of exogenous application of PAs and/or their biosynthetic inhibitors. The determination of PA contents at different stages of plant development provided evidence for correlations between PAs and flowering, cell division, and other developmental processes. Both early and more recent works have described the occurrence of crosstalk between PAs and hormones such as gibberellins, auxins, and ethylene, but the intrinsic mechanisms underlying such interactions are not completely established (Kusano et al. 2008; Alcázar et al. 2010a; Bitrián et al. 2012). Genetic approaches in recent years have unraveled mechanisms by which PAs exert their roles in some developmental processes. These include the involvement of Spd in the posttranslational modification of eIF5A, implications of HCAA conjugates in pollen development and the role of tSpm in vascular tissue development that are discussed in the following sections.

Spermidine-dependent hypusination of eIF5A

Seminal works by Tabor et al. reported that S. cerevisiae loss-of-function mutants of ODC (spe1), SAMDC (spe2) and SPDS (spe3) were unable to grow in the absence of exogenous PAs (Cohn et al. 1980; Balasundaram et al. 1991; Hamasaki-Katagiri et al. 1998). Indeed, these mutants show cell cycle arrest in the G1–S phases and abnormal actin polymerization, phenotypes that resemble cell division cycle (cdc) and actin mutants. Growth of spe1, spe2, or spe3 yeast mutants could be restored by exogenous application of either Spd or Spm. However, only Spd has been proved to be essential for cell division, as the growth-stimulating effect of Spm is due to its oxidation to Spd via FMS1 (FENPROPIMORPH-RESISTANCE MULTICOPY SUPPRESSOR 1) (Chattopadhyay et al. 2003). The underlying basis for such a dramatic need of Spd for cell survival has different plausible explanations, associated mainly with the requirement of PAs for essential cellular functions, one of which is the posttranslational activation of the eukaryotic translation initiation factor eIF5A.

eIF5A is essential for cell division in eukaryotes, and although its specific molecular role is not known, it is thought to act as interactor between RNA and proteins essential for the translation machinery or as an initiation factor in a subset of mRNAs (Kang and Hershey 1994). For eIF5A to become an active protein, it must be hypusinated in one specific lysine residue in its N-terminal domain. Hypusination is a specific posttranslational modification that requires the activity of the deoxyhypusine synthase (DHS) enzyme, which catalyzes the NAD-dependent transfer of the aminobutyl moiety of Spd to eIF5A to form a deoxyhypusine residue. This step is followed by the hydroxylation of deoxyhypusine to hypusine, catalyzed by deoxyhypusine hydroxylase (DOHH) (Park et al. 2010) (Fig. 1). Hypusine modification is one of the most specific eukaryotic posttranslational modifications known, as it occurs only in eIF5A, whose activation is absolutely dependent on the availability of a free Spd pool. This observation might underlie the loss of viability in cells depleted of Spd in many organisms. Arabidopsis adc1 adc2, spds1 spds2, and samdc1 samdc4 double loss-of-function mutants, which are impaired in the biosynthesis of Put/Spd/Spm or Spd/Spm, are embryo lethal (Imai et al. 2004b; Urano et al. 2005; Ge et al. 2006). In contrast, the double knockout spms acl5, which is impaired in Spm/tSpm biosynthesis, is fully viable, thus showing that Spm and its isomer tSpm are not essential for survival (Imai et al. 2004a). In contrast to yeast, in which genetic depletion of PAs can be rescued by exogenous uptake, in Arabidopsis it has not been possible to isolate PA auxotrophic mutants.

Despite the essential role of Spd in eIF5A activation (Park et al. 2010), Spd-dependent cell growth in mammalian cells, and possibly those of other organisms, cannot be explained by eIF5A alone. The kinetics of PA levels, eIF5A inactivation and cell growth using different inhibitors suggest that cell division is already inhibited by PA depletion before active eIF5A pools are significantly affected. Thus, Spd or the products of its oxidation or conjugation might have a role in cell growth on its own (Nishimura et al. 2005).

An interesting finding in plants is the capacity of DHS from tobacco and Senecio vernalis to act as homospermidine synthases (HSSs) for the formation of homospermidine from Put, using Spd as an amino butyl donor (Ober and Hartmann 1999a, b) (Figs. 1, 3). Homospermidine is required for the synthesis of pyrrolizidine alkaloids, secondary metabolites distributed in some angiosperms that play a role in plant–herbivore interactions (Hartmann 1999). HSSs in plants are thought to derive from a DHS that lost its main activity but retained HSS functionality. However, it is still intriguing to ask why DHSs from species that do not contain pyrrolizidine alkaloids, such as tobacco, would retain the capacity to synthesize homospermidine (Ober and Hartmann 1999b).

The role of spermidine conjugates in pollen development

HCAAs are abundant phenolic compounds in the seeds and reproductive organs of plants. High HCAA concentrations occur during the transition to flowering and flower development (Bassard et al. 2010). The potential developmental and stress roles of such PA conjugates are largely unknown, mainly because most genes involved in HCAA biosynthesis have not been identified. However, important progress has unraveled the genetic basis for the biosynthesis of some Spd conjugates and their contributions to pollen development.

An Spd hydroxycinnamoyltransferase (SHT) has been characterized in Arabidopsis (Grienenberger et al. 2009). Incubation of recombinant SHT with different PAs (Put, Spd, and Spm) and hydroxycinnamoyl-CoA esters identified a high selectivity for the acylation of Spd. SHT expression is restricted to the tapetum cells of anthers, and the gene’s downregulation or loss-of-function leads to reduced levels of N 1,N 5,N 10-trihydroxyferuloyl-Spd and N 1,N 5-dihydroxyferuloyl-N 10-sinapoyl-Spd in flower buds (Grienenberger et al. 2009). Thus, from the reaction product of Spd acylation by SHT to the final products detected in plants, three hydroxylations and one O-methylation occur (Fig. 1). The Arabidopsis thaliana Tapetum-Specific Methyltransferase 1 (AtTSM1) gene is co-expressed with SHT, and its downregulation via RNA interference (RNAi) depletes the levels of N 1,N 5-dihydroxyferuloyl-N 10-sinapoyl-Spd, as in the sht mutant, and increases the levels of N 1,N 5,N 10-trihydroxyferuloyl-Spd, thus suggesting that it participates in the O-methylation activity described above confirmed in vitro (Fellenberg et al. 2008; Grienenberger et al. 2009).

Interestingly, the expression of SHT is regulated by MALE STERILITY 1, a transcription factor involved in regulating genes required for normal pollen wall and coat development (Ito et al. 2007; Yang et al. 2007; Grienenberger et al. 2009). Indeed, the pollen walls of sht mutants show some irregularities, and the grains lack their typical autofluorescence. However, sht pollen is still viable, at least under greenhouse conditions. In contrast, AtTSM1 RNAi lines show severe defects in silique formation and seed set (Fellenberg et al. 2008), suggesting that mutations in different steps of the pathway have different impacts. From an evolutionary perspective, some of the enzymatic activities in the biosynthesis of phenolamide intermediates seem to be the result of adaptive evolution. The Arabidopsis cytochrome P450 gene paralogs, CYP98A8 and CYP98A9, are derived from a retroposition event subjected to accelerated positive selection. Interestingly, CYP98A8 participates in the metahydroxylation of triferuloyl-Spd, and both CYP98A8 and CYP98A9 act redundantly in the hydroxylation of tricoumaroyl-Spd (Fig. 1) (Matsuno et al. 2009).

Mutation of SPDS1 or SAMDC1 in Arabidopsis leads to a reduction of hydroxycinnamoyl-Spd conjugates, whereas mutation of SHT does not lead to Spd or Put increases (Fellenberg et al. 2012). The effects of SAMDC mutation on the synthesis of hydroxycinnamoyl-Spd conjugates may provide a basis for the observed correlations between loss of pollen viability at high temperature and reduced SAMDC activity in other species (Song et al. 2002). In addition, PAs are thought to be involved in modulating pollen tube growth (Antognoni and Bagni 2008; Rodriguez-Enriquez et al. 2013). Once the pollen lands on the stigma tissue of the pistil, compatible pollen grains hydrate and the pollen tube elongates for the delivery of sperm cells into the female gametophyte. Recently, a mechanism has been proposed that implies increases in cytosolic Ca2+ induced by Spd. Indeed, PAs are known regulators of different ion channels (Alcázar et al. 2010a). However, Spd effects on pollen tube growth are mediated through release of H2O2 derived from oxidation of Spd by PAO activity rather than Spd itself. This is demonstrated by the inhibition of pollen tube growth and reduced seed set in the Arabidopsis pao3 mutant (Wu et al. 2010).

The role of thermospermine in the Arabidopsis stem and vascular development

The acaulis5 (acl5) mutant was originally isolated from an ethyl methanesulfonate (EMS) mutagenesis screen for Arabidopsis plants affected in stem elongation in the Landsberg erecta (Ler) accession (Hanzawa et al. 1997). This phenotype is also present in the Columbia background (Imai et al. 2006), thus making it a background-independent phenotype. Two allelic mutants (acl5-1 and acl5-2) were found that exhibited inhibition of stem elongation after the transition to the reproductive stage. These mutants carried identical causal mutations. The acl5-1 mutant does not show obvious phenotypes before flowering and cannot be rescued by exogenous application of hormones. Microscopic examination of the stem vascular tissue of acl5-1 revealed an overproliferation of xylem elements (Hanzawa et al. 1997). These alterations are in line with the ACL5 expression pattern, which is specific to vascular tissue.

A more detailed analysis revealed that the acl5-1 mutant does not develop xylem fibers and has important defects in secondary cell wall formation, owing mainly to early initiation of cell death in developing xylem vessel elements (Muñiz et al. 2008). Mapping of the acl5-1 mutation identified a nonsynonymous nucleotide substitution on a protein similar to SPDSs and SPMSs (Hanzawa et al. 2000). The mutation was suggested to affect a potential dSAM-binding site. Recombinant ACL5 converted Spd to Spm in bacterial cultures, and thus ACL5 was originally described as an SPMS (Hanzawa et al. 2000). However, later evidence indicated that ACL5 might have another enzymatic activity than Spm biosynthesis. Among the four Arabidopsis APT members (SPDS1, SPDS2, SPDS3/SPMS, and ACL5), SPDS3/SPMS was unequivocally assigned as an SPMS (Panicot et al. 2002). Consistently, spms-1 mutants were almost depleted in Spm levels, but, surprisingly, they did not show any obvious phenotypic change from the wild type. The spms-1 acl5 double mutant reconstituted the acl5 phenotype, but Spm levels in acl5 single mutants were barely affected (Imai et al. 2004a).

Knott et al. (2007) shed light on this controversy by characterizing a tSPMS from the diatom Thalassiosira pseudonana that is a homolog of Arabidopsis ACL5. Indeed, enzymatic assays with recombinant ACL5 identified tSpm as the product of ACL5 reaction. This PA, originally isolated from the thermophilic bacterium Thermus thermophilus, is an isomer of Spm that can be separated from it by classical thin-layer chromatography or benzoyl derivation of PAs by HPLC. The identity of ACL5 as a tSPMS was confirmed in the absence of tSpm in acl5 mutant plants and partial complementation of the acl5 phenotype by exogenous application of tSpm (Kakehi et al. 2008).

Suppression of the striking acl5 phenotype has been used to identify pathways involved in tSpm signaling. EMS-based suppressor screens for the acl5-1 dwarf phenotype identified four suppressor of acaulis 5 mutants (sac51-d to sac54-d), differing in their genetic dominance (Imai et al. 2006). The dominant sac51-d mutation disrupts a short uORF of a putative basic helix-loop-helix transcription factor. Premature termination of such a uORF sequence leads to higher translation of the SAC51 ORF and suppression of the acl5-1 phenotype. Indeed, deletion of the uORF sequence and overexpression of SAC51 partially complement the acl5-1 phenotype, and thus tSpm is proposed to participate in the translational activation of SAC51 (Imai et al. 2006). The causal polymorphism for the semidominant sac52-d mutation lies in a nonsynonymous substitution for a gene coding RIBOSOMAL PROTEIN L10, which is a component of the 60S ribosomal subunit highly conserved in eukaryotes (Imai et al. 2008). The mutated allele might enhance the translation of some transcripts, among them SAC51, which in turn contributes to the suppression of dwarfism in acl5-1. Thus, suppressor screens for acl5-1 indicate that tSpm is an important PA for the translational control of transcripts involved in stem elongation, in line with the concept of PA modulon introduced in bacteria and discussed below (see “Gene expression regulation”).

The Arabidopsis loss-of-function mutant samdc4 (bud2, bushy and dwarf 2) exhibits semidwarfism and vascular tissue alterations that partially resemble the acl5 phenotype (Ge et al. 2006; Cui et al. 2010). This mutant also shows hyposensitivity to auxins and hypersensitivity to cytokinins, which might explain the loss of apical dominance that gives its bushy appearance. The similarities to the acl5 mutant suggest that SAMDC4 is actively involved in tSpm biosynthesis, which contrasts with the absence of a distinctive phenotype in samdc1 mutants.

Polyamines and abiotic stress

Elevated PA levels are one of the most remarkable metabolic hallmarks in plants exposed to stresses such as drought, salinity, chilling, heat, hypoxia, ozone, UV, and heavy metals (Gill and Tuteja 2010; Alcázar et al. 2010a). These changes are produced mainly by alterations of PA biosynthesis, PA oxidation, and/or interactions with other pathways in response to stress.

Abiotic stress and polyamine biosynthesis

Transcriptomic studies performed in Arabidopsis have revealed differential regulation of PA biosynthetic genes by abiotic stress (Alcázar et al. 2006). The characterization of PA loss-of-function mutants in this species has provided evidence for the involvement of PAs in resistance traits (Urano et al. 2004; Yamaguchi et al. 2007; Cuevas et al. 2008). These analyses have been complemented by the evaluation of lines overexpressing PA biosynthetic genes in different plant species (Alcázar et al 2010b). Heterologous overexpression of ODC, ADC, SAMDC, and SPDS from different organisms in rice, tobacco and tomato induced tolerance to many stresses in correlation with the degree of Put and/or Spd and Spm accumulation (Gill and Tuteja 2010; Alcázar et al. 2010a). Compelling evidence thus indicates that PAs have a protective role against abiotic stress in different plant species. However, a current challenge is to elucidate the molecular mechanism(s) by which PAs exert these stress-induced protective roles.

Abiotic stress and polyamine oxidation

Increases in PA levels during stress and ABA treatment are often accompanied by stimulation of PA oxidation mediated by extracellular AOs and transport of PAs to the apoplast, where PAs are usually present in low amounts (Moschou et al. 2008c; Toumi et al. 2010) (Fig. 2). The expression of genes encoding apoplastic AOs, such as ZmPAO and AtCuAO1, is induced by ABA (Xue et al. 2009; Wimalasekera et al. 2011b; Planas-Portell et al. 2013), and Arabidopsis atcuao1 mutants display hyposensitivity to osmotic stress (Wimalasekera et al. 2011b). Altogether, these observations suggest a possible involvement of AOs in plant abiotic stress responses.

Evidence points to the occurrence of a metabolic flow from PAs to GABA and H2O2 during the stress response. As consequence, these reaction products are considered mediators of AO-dependent stress signaling. Indeed, higher CuAO, AMADH, and peroxidase activities and increases in Put, Spd and GABA contents occur in pea by mechanical wounding (Petrivalský et al. 2007). Quantitative correlations are also observed between PA degradation and GABA accumulation in soybean exposed to salinity stress (Xing et al. 2007). The H2O2 produced by apoplastic PA oxidation contributes to cell wall strengthening during wound healing in several plant species (Rea et al. 2002; Angelini et al. 2010). In addition, H2O2 produced through this oxidation is proposed to act as a mediator in ABA signaling during stomatal closure in Vicia faba and in cytosolic antioxidant defense activation in maize (An et al. 2008; Xue et al. 2009). Nevertheless, because the apoplast is not provided with the high levels of antioxidant enzymes present in intracellular compartments, there is a potential risk for PA-derived H2O2, when not properly quenched, to lead to programmed cell death instead of survival (Moschou et al. 2008a). Indeed, salt application to tobacco plants overexpressing ZmPAO shifts the balance to promote cell death rather than resistance (Moschou et al. 2008c).

Collectively, the results indicate that apoplastic AOs involved in terminal PA catabolism contribute to plant abiotic stress responses. However, not much is known concerning the role(s) of intracellular AOs, most of which are localized in peroxisomes. Transcriptomic studies evaluating changes linked to PA catabolism are scarce, but the available data indicate that genes encoding peroxisomal PAOs (AtPAO2–4) are induced by ABA and mechanical wounding (Moschou et al. 2008d), whereas for the genes encoding peroxisomal CuAOs, AtCuAO3 is responsive to ABA and AtCuAO2 is responsive to mechanical wounding (Planas-Portell et al. 2013), suggesting a potential role of the peroxisome PA catabolism in plant responses to stress. However, peroxisomes can efficiently eliminate H2O2 as a result of their high catalase activities. Thus, it is unlikely that H2O2 produced by AOs would be present in sufficient amounts to promote programmed cell death as observed in tobacco (Moschou et al. 2008b). Indeed, increased intracellular PA catabolism in Arabidopsis is not associated with programmed cell death during stress (Moschou et al. 2012). This class of AOs may be rather involved in signaling, as reported for the peroxisomal PA catabolism that increases intracellular Ca2+ levels during pollen tube growth (Wu et al. 2010) (Fig. 2).

Omics approaches

Global systems’ biology approaches are useful for allocating PAs in the context of broader transcriptomic, metabolic and proteomic networks. There have not been a large number of these analyses, but some interesting findings have been reported.

Hierarchical clustering expression analyses in Arabidopsis using public microarray data (Eisen et al. 1998; Hruz et al. 2008) indicate that some, but not all, PA biosynthesis and catabolism gene paralogs share similar expression patterns (Fig. 4). Thus, SPDS1/SPDS2 and most PAOs (except for PAO5) tend to clade together, whereas ADC1/ADC2 and ACL5/SPMS show significant differences, in agreement with their different implications in stress and development. Remarkably, among the 9,848 microarray samples used for the hierarchical clustering analysis (Hruz et al. 2008), ABA application is the treatment that exhibits the most striking changes in the PA transcriptome (Fig. 4). Using the same dataset, we looked for the nature of the top-200 co-expressed genes showing positive correlations with PA metabolism genes, in a wide range of stress and developmental conditions (Fig. 5). The results enabled the categorization, based on gene ontology, of PA metabolic genes as involved in stress responses, development, or both. ADC2, SPMS, PAO1, PAO2 and PAO3 are genes whose expression is positively correlated with other genes induced during stress. Conversely, ADC1, AIH, SPDS1, SPDS2, SAMDC4, PAO5 and ACL5 fall in the category of genes whose expression correlates with other genes induced during development. In agreement with these observations, samdc4, acl5 and the double spds1 spds2 loss-of-function mutants exhibit evident developmental abnormalities (Hanzawa et al. 1997; Imai et al. 2004b; Ge et al. 2006). These analyses, however, could not assign ADC1 in the category of stress-related genes, despite its implications in freezing tolerance (Cuevas et al. 2008). Interestingly, both PA biosynthesis and catabolism genes are found in stress and developmental categories. Hence, both biosynthesis and catabolism seem to play a role in stress and development.

Fig. 4
figure 4

Hierarchical clustering expression analyses of PA metabolism genes and heat map of expression profiles in response to ABA derived from publicly available microarray data (Hruz et al. 2008). Increased or reduced transcription of genes is shown in red and green, respectively

Full size image
Fig. 5
figure 5

Heat map for the enrichment in gene ontology terms found in the top 200 co-expressed genes that show positive correlations with PA biosynthetic (blue) and PA oxidation (orange) genes. The heat map has been generated using publicly available microarray data from 9,848 samples deposited in Genevestigator in a wide range of environmental and developmental conditions (Hruz et al. 2008)

Full size image

Metabolomic approaches provide a comprehensive view of how metabolic networks are regulated in adverse environments and identify metabolic flows that impair homeostasis (Obata and Fernie 2012). Metabolic analyses in response to drought in Arabidopsis evidence the accumulation of Put and GABA (Urano et al. 2009), the latter of which is derived from PA catabolism. Indeed, a common hallmark in Arabidopsis plants subjected to different stresses such as cold, heat, cadmium treatment, and a combination of high light and sulfur depletion is the accumulation of Put (Obata and Fernie 2012), and in some cases this response is mediated by ABA (Urano et al. 2009; Alcázar et al. 2010a). Further metabolic profiling analyses in hormone and stress-signaling mutants should provide information on the regulatory circuits that modulate PA metabolism and their coordinated action with other metabolic pathways.

Global proteomic analyses are still scarce, and the available data are fragmentary. Some proteomic analyses revealed accumulation of SPDS protein by drought in wheat (Hajheidari et al. 2007) and in CPA and SPDS by oxidative stress in tobacco cells (Vannini et al. 2012). SAM synthetase protein also accumulated in response to salt and Spd treatments in cucumber (Li et al. 2013). However, the potential effect of PAs on posttranslational modifications relevant for stress signaling remains to be explored.

Polyamines and biotic stress

Most studies about the involvement of PAs in biotic stress have focused on incompatible plant–pathogen interactions that rely on pathogen recognition and activation of defense, often accompanied by hypersensitive response (HR). Early works using tobacco mosaic virus (TMV) infection reported the accumulation of free and conjugated PAs during HR in tobacco (Walters 2003; Bassard et al. 2010). Thus, PA conjugates may contribute to defense against pathogenic microorganisms, and antimicrobial properties have been proposed (Walters 2003). Despite this, most studies on biotic stress in PAs have focused on PA catabolism. Indeed, part of the TMV-induced HR driven by the N gene in tobacco requires accumulation of Spd or Spm (substrates of PAOs) in the apoplast (Yoda et al. 2003).

A common view from this and other reports in different plants and pathosystems is that H2O2 derived from PA catabolism contributes to reactive oxygen species (ROS) generation during HR (Yoda et al. 2003; Gonzalez et al. 2011), and thus Spd and/or Spm contribute to pathogen resistance. Recently, tSpm catabolism has also been shown to contribute to Arabidopsis disease resistance (Marina et al. 2013). In addition to this effect, Spm induces salicylic acid (SA)-independent accumulation of certain pathogenesis-related proteins in TMV-infected tobacco plants. This might be partly due to the Spm-induced mitochondrial malfunction that triggers activation of some mitogen-activated protein kinase (MAPK) which are inducers of HR (Takahashi et al. 2003). Whether this is a conserved mechanism between species requires to be investigated.

In Arabidopsis, exogenously supplied Spm also induces the expression of a very similar subset of genes mimicking cucumber mosaic virus-induced HR (Mitsuya et al. 2009). Similarly, SPMS overexpressor lines in Arabidopsis exhibit higher expression of defense genes, thus providing increased basal defense that correlates with resistance (Gonzalez et al. 2011). These observations highlight the relevance of studying global transcriptional regulation of genes by PAs, as discussed below (see “Gene expression regulation”).

PAs should be considered as an integral part of the HR that contributes to resistance through ROS generation, although PAs are dispensable for this response, given the occurrence of alternative ROS-generating pathways during HR (Yoda et al. 2003). However, their contribution to resistance to a broad range of pathogens in different species can be highly useful in practical applications.

Potential mechanisms of action

Although pleiotropic effects of PAs complicate efforts to understand the underlying PA functions, recent studies have identified several key areas in which molecular bases are now being actively studied (Igarashi and Kashiwagi 2010a; Pegg and Casero 2011). This section discusses recent advances in these cutting-edge topics.

Epigenetic regulation

The potential role of PAs in the epigenetic control of gene expression is an emerging area of research. In plants, the presence of a well-defined DNA methylation pattern is necessary for normal growth and development. DNA methylation in plants protects the genome from transposable elements and modulates gene expression (Chan et al. 2005). SAM is a common methyl donor in reactions such as DNA methylation and PA biosynthesis. Thus, it has been suggested that changes in PA levels could affect DNA methylation and consequently gene expression (Fraga et al. 2004). A current hypothesis is that increased Spd and Spm biosynthesis can reduce Put and dSAM levels. Given the competitive inhibitory effect of dSAM on DNA methyltransferase, reduced dSAM levels may lead to increased methyltransferase activity, thereby emphasizing the degree of DNA methylation and possibly causing gene silencing. As such, effects on epigenetic modulation by PAs could be explained partly in the context of its metabolic network.

Histone methylation is regulated not only by histone methyltransferases but also by the removal of methyl groups chiefly catalyzed by lysine-specific demethylase 1 (LSD1) (Shi et al. 2004). A key positive chromatin mark associated with promoters of active genes is histone H3 dimethyl-lysine 4 (H3K4me2) (Schneider et al. 2004). LSD1 catalyzes the demethylation of H3K4me2 and is associated with transcriptional repression. Therefore, LSD1 could potentially be a regulator of gene expression through the modulation of chromatin structure (Huang et al. 2007). Remarkably, LSD1 shares considerable homology with FAD-dependent PAOs (Shi et al. 2004). Moreover, inhibition of LSD1 by PA analogs results in reexpression of aberrantly silenced genes in cancer cells (Huang et al. 2007). Arabidopsis has four gene relatives of human LSD1, some of which participate in the repression of FLOWERING LOCUS C (FLC), thus promoting transition to flowering (Jiang et al. 2007). This suggests that PAOs homologous to LSD1 may be among the regulators of gene expression in both animal and plant cells (Krichevsky et al. 2007). In addition, high PA levels alter histone acetylation and histone acetylase and deacetylase activities in animal proliferative cells, which in turn affects gene regulation (Minois et al. 2011). An important recent discovery is that Spd prolongs life span in various organisms by protecting certain classes of genes from strong deacetylation during aging, thereby allowing their transcription (Eisenberg et al. 2009). A similar mechanism could explain the antisenescence effect of PAs in plants (Tiburcio et al. 1994). It seems clear that the role of PAs in the epigenetic control of gene expression in plants is a promising area of research that needs further investigation.

Polyamine modulon

An interesting concept that has contributed to our understanding of PA functions derives from the isolation of a ‘PA modulon’ in bacteria that consists of a set of gene transcripts whose translation is enhanced by PAs (Pegg and Casero 2011). PAs predominantly exist in RNA complexes, thus affecting various steps of translation. PAs actually stimulate the synthesis of some proteins both in vitro and in vivo and increase the fidelity of translation promoting the in vivo assembly of 30S ribosomal subunits (Igarashi and Kashiwagi 2011). The analysis of the proteins whose synthesis is enhanced by PAs at the translational level has been achieved using a PA-requiring E. coli mutant grown with or without exogenous Put. These approaches have revealed that Put enhances the translation of a set of gene transcripts, classified as belonging to a PA modulon (Igarashi and Kashiwagi 2006). Most of its members are transcription factors that enhance the synthesis of several kinds of mRNA, tRNA, and rRNA. In this way, PAs modulate the level of many kinds of proteins to maintain optimal conditions for cell growth (Igarashi and Kashiwagi 2010a). The concept of PA modulon has recently been applied and investigated in yeast and mammalian cells (Nishimura et al. 2009; Uemura et al. 2009). In plants, the effects of PAs in gene expression have been investigated using PA-overproducing lines. Comparison of transcriptome profiles of Arabidopsis ADC2 overexpressors accumulating Put with those of wild-type plants revealed alterations in the expression of genes involved in auxin, ethylene, and ABA biosynthesis and signaling (Marco et al. 2011). Changes in gene expression were also observed for genes responsive to biotic stress (Marco et al. 2011). In contrast, Arabidopsis transgenic lines with elevated Spm levels caused by SAMDC1 overexpression identified alterations in the transcripts of SA and JA-responsive genes, some receptor-like kinases, MAPKs and transcription factors involved in stress signaling. Results also indicate that Put and Spm induce differential transcriptional responses, as only 150 genes (10 % of total differentially expressed genes) were commonly up- or downregulated by both Put and Spm (Marco et al. 2011). Whether or not the concept of PA modulon could be applied to plants requires further investigations.

Polyamines and NO signaling

Studies indicate that PA signaling involves crosstalks with other stress pathways. NO is a signaling molecule involved in a wide range of functions in plants. It participates in posttranslational modifications such as S-nitrosylation (i.e., incorporation of the NO moiety into a cysteine sulfur atom to form an S–NO bond) that affect protein structure and protein–protein interactions (Lounifi et al. 2013). Recent evidence suggests that PAs and NO have some overlapping physiological roles in plants (Yamasaki and Cohen 2006). PAs are actually biochemically related to NO through arginine, a common precursor in their biosynthetic routes, thus suggesting that alteration of PA homeostasis could affect NO bioavailability and vice versa (Filippou et al. 2013). PAs induce rapid synthesis of NO in Arabidopsis, which suggests that NO might contribute to PA signaling during stress (Tun et al. 2006). In agreement with this view, Arabidopsis mutants deficient in CuAO1 display lower NO production and enhanced S-nitrosylation following PA treatments (Wimalasekera et al. 2011a). These observations suggest that the possible relationship between PAs and NO-originated nitrosative signaling could regulate abiotic stress responses in plants (Tanou et al. 2013).

The mode of action by which PAs affect S-nitrosylation can differ among PAs, suggesting that individual PAs may act on specific signaling pathways (Lounifi et al. 2013). Thus, several proteins, such as NADP-dependent malic enzyme, miraculin, and 60S ribosomal protein, are nitrosylated exclusively by Put, whereas proteins involved in sugar metabolism, proteolysis, and terpenoid metabolism are nitrosylated exclusively by Spm. In contrast, Spd nitrosylates various subgroups of proteins, the majority of which are associated with energy, amino acid metabolism, and stress responses. Notably, the identification of Δ1-pyrroline-5-carboxylate Synthase 1 (P5CS1) as a Spd target along with the strong induction of P5CS1 expression by Spd and NO application provides evidence for an NO-dependent functional link between PAs and proline under abiotic stress conditions (Filippou et al. 2013; Tanou et al. 2013). Another interesting finding is that methionine synthase, which catalyzes the production of methionine, is nitrosylated by Put and Spd in salt-stressed plants. Methionine is a precursor for both PAs and ethylene, which prevent and induce plant senescence, respectively. Therefore, nitrosylation of methionine synthases may provide an additional link between PA and ethylene pathways involved in the control of senescence-related stress responses (Tanou et al. 2013). Collectively, the available data indicate that PAs are integrated with NO signaling to provide a coordinated response against different types of abiotic stresses through the occurrence of metabolic interactions, PA-triggered S-nitrosylation of key proteins in the stress response and interactions with ABA and redox (ROS) signaling.