Seed Biofortification and Phytic Acid Reduction: A Conflict of Interest for the Plant?

Most of the phosphorus in seeds is accumulated in the form of phytic acid (myo-inositol-1,2,3,4,5,6-hexakisphosphate, InsP6). This molecule is a strong chelator of cations important for nutrition, such as iron, zinc, magnesium, and calcium. For this reason, InsP6 is considered an antinutritional factor. In recent years, efforts to biofortify seeds through the generation of low phytic acid (lpa) mutants have been noteworthy. Moreover, genes involved in the biosynthesis and accumulation of this molecule have been isolated and characterized in different species. Beyond its role in phosphorus storage, phytic acid is a very important signaling molecule involved in different regulatory processes during plant development and responses to different stimuli. Consequently, many lpa mutants show different negative pleitotropic effects. The strength of these pleiotropic effects depends on the specific mutated gene, possible functional redundancy, the nature of the mutation, and the spatio-temporal expression of the gene. Breeding programs or transgenic approaches aimed at development of new lpa mutants must take into consideration these different aspects in order to maximize the utility of these mutants.


Introduction
Phytic acid (myo-inositol-1,2,3,4,5,6-hexakisphosphate, InsP6) is a ubiquitous component of eukaryotic cells which, together with its metabolism, plays a number of regulatory roles [1]. In plants, phytic acid is the most abundant form of phosphorus occurring in seeds (up to 85% of total phosphorus and with amounts even 1000 fold higher than those detected in vegetative tissues) and other plant tissues and organs such as pollen, roots, tubers and turions. During seed development, the synthesis of phytic acid is coordinated with that of other seed storage compounds. Accumulation significantly increases after the "cell division phase", reaching a plateau at the end of the "cell expansion phase" [2,3]. In mature seeds, phytate is organized into spherical inclusions called globoids, which are in turn found within protein bodies. Phytate deposits are also observed to occur transiently in various tissues and subcellular compartments during grain development [4][5][6][7]. Depending upon the species, the amount and distribution of phytic acid in different parts of the seed can be quite variable, with the most striking differences found in cereal grains. In the case of barley, wheat and rice, a large amount (80%) of phytic acid is stored in the aleurone and bran (maternal teguments) and only a limited amount accumulates in the embryo. The distribution of phytate is opposite in maize seeds, where 80% of phytate accumulates in the embryo and scutellum [8]. In the case of legume seeds, such as common bean, more than 95% of seed phytic acid is accumulated in the cotyledons [9], while in the model species Arabidopsis, phytic acid is mostly stored in the embryo [5]. During germination, in order to support seedling growth, phytic acid is then degraded by phytase enzymes to remobilize the phosphorus stored as phytate salts [10].
Due to its chemical structure (highly negatively charged at physiological pH), phytic acid easily precipitates in the form of phytate salts, binding important mineral cations such as iron, zinc, potassium, calcium, and magnesium. Monogastric animals, including humans, lack phytases in their digestive tract and fail to process the phytates present in seeds, thus phytic acid is poorly digested and decreases the nutritional value of the seeds by limiting phosphorus and mineral bioavailability. Poor mineral bioavailability, due to high molar ratios between phytic acid and mineral cations, is ascribed as one of the most important causes of mineral deficiencies (mainly iron and zinc) in populations whose diet is largely based on staple crops [11,12]. On the other hand, InsP6 is largely excreted by nonruminants. A common practice to provide for an animal's nutritional requirement for phosphorus (P) has been the supplementation of feed with nutrient P. However, this procedure increases P concentration in manure, leading to P accumulation in soils, and the consequent risk of P pollution in runoff water [13]. To obviate these problems, a solution is the development of low phytic acid (lpa) crop seeds [14].
The availability and management of P in agriculture is a challenging global problem: reserves of rock P are non-renewable and enhanced uptake and utilization of P would be of value for agricultural production in P-deficient environments throughout the world. It would also contribute to the long-term goal of sustainable and environmentally friendly agricultural production [14,15]. The total P accumulated in seed crops of major grains and legumes represents in sum more than 50% of the total P fertilizer used annually worldwide [16]. Therefore, reducing seed total P, together with a reduction of phytate content, might also contribute to these goals [17]. This has been achieved at least for the barley lpa1-1 mutant, in which seeds show a decrease both in phytic acid and total P [18].
The biological functions of phytic acid and the identification of genetic resources and strategies useful in engineering high-yielding, stress-tolerant low-phytate germplasm have been reviewed by a number of authors [10,11,14,17]. However, in most cases, a strong emphasis was placed on the agronomic aspects of the topic, with only modest efforts to integrate these aspects with the emerging knowledge of the regulatory role of phytic acid and inositol metabolism. The aim of this review is to summarize the most recent results in the literature about relevant aspects of the phytic acid pathway and lpa mutants. Moreover, we discuss some pleiotropic effects of lpa mutants with respect to the reported roles, important for cell signaling and plant processes, of phytic acid and key enzymes and metabolites of this complex pathway.

Biosynthetic Pathways
In plants, it is now generally accepted that InsP6 biosynthesis occurs through two different routes: a "lipid-dependent" pathway, which is ubiquitous in eukaryotic cells, and a "lipid-independent" pathway ( Figure 1). The first operates in all plant tissues, while the second appears to predominanate in seeds. Phytic acid biosynthesis needs the de novo production of myo-inositol (hereafter referred to as "Ins") through a highly conserved reaction, shared by all living organisms, in which the enzyme D-myo-inositol 3-phosphate synthase (MIPS) converts D-glucose-6-phosphate to myo-inositol 3-phosphate (Ins(3)P1) ( Figure 1). Myo-inositol 3-phosphate is then dephosphorylated to free Ins by a specific Mg 2+ -dependent inositol monophosphate phosphatase (IMP). Interestingly, the IMP enzyme has a dual activity: besides Ins(3)P1, it also hydrolyzes L-galactose 1-phosphate (L-Gal 1-P), a precursor of ascorbic acid synthesis [19,20]. The reaction catalyzed by IMP can be reversed by the action of myo-inositol kinase (MIK). Since Ins(3)P1 is produced directly from glucose 6-phosphate by MIPS, it is not clear exactly why MIK activity is important for InsP6 biosynthesis, although the importance of MIK in seed InsP6 metabolism has been demonstrated by a number of mutations in the MIK gene [21-23]. A possible explanation is that MIK could provide more substrate diversity for the generation of inositol bisphosphate to feed the lipid-independent pathway, since it is able to produce multiple inositol monophosphates [21].
The main difference between the "lipid-dependent" and "lipid-independent" routes is the way inositol tri-phosphates (InsP3) are generated. In the "lipid-dependent" pathway, Ins is converted to phosphatidylinositol (PtdIns) by a phosphatidylinositol synthase (PtdIS). Next, the headgroup of PtdIns is sequentially phosphorylated by phosphatidylinositol kinases to produce PtdIns (4,5)P2. This molecule is the substrate of a PtdIns-specific phospholipase C activity that releases Ins(1,4,5)P3, a molecule central to signal transduction [24].
The so-called "lipid-independent" pathway is entirely independent from inositol lipid synthesis and consists of sequential phosphorylation of the Ins ring to InsP6, through the action of a number of specific inositol phosphate kinases. As already mentioned, the first phosphorylation step, consisting in the conversion of Ins to InsP1, is catalyzed by myo-inositol kinase (MIK). The production of InsP2 from InsP1 requires a monophosphate kinase. A good candidate for this step is a homolog of 2-phosphoglycerate kinase (2-PGK), which catalyzes the production of 2,3-6 bisphosphoglycerate from 2-phosphoglycerate in archaea [14]. In rice, a mutation in this gene (OsPGK1) generates an lpa phenotype, while overexpression increases seed InsP6 content, suggesting that OsPGK1 is a key gene for InsP6 synthesis, being involved in (probably the rate-limiting) step from InsP1 to InsP2 [22,25,26]. Further phosphorylation steps, required to convert InsP3 into the more phosphorylated InsP4, InsP5 and InsP6, involve at least three types of inositol kinases: (i) the evolutionarily conserved inositol phosphate kinase 2, also known as inositol polyphosphate multikinase (IPK2/IPMK), which is a dual 6-/3-kinase; (ii) the inositol 1,3,4-trisphosphate 5-/6-kinase (ITPK) belonging to the family of ATP-grasp fold proteins; and (iii) the inositol polyphosphate 2-kinase (IPK1), which specifically phosphorylates InsPs in the 2-position.
Once synthesized, phytic acid is stored as globoids inside the storage vacuoles where it is actively transported by a specific InsP6 transporter, a multidrug-resistance-associated protein (MRP), belonging to the ATP-binding cassette (ABC) family [27]. The involvement of all the above-mentioned proteins in InsP6 synthesis and accumulation is supported by direct enzyme isolation and biochemical characterization and/or by the isolation of corresponding lpa mutants [19,[28][29][30][31][32][33][34] (Table S1).

Genomic Organization and Regulation of Phytic Acid Pathway Genes
Genes involved in phytic acid biosynthesis and transport have been characterized in Arabidopsis thaliana and crop plants such as rice, wheat, soybean and common bean, through forward (through the screening for lpa mutations) and reverse genetics [35][36][37][38][39]. In this section we will describe the main classes of genes coding for enzymes of the pathway, MIPS, IMP, MIK, 2-PGK, IPK2, ITK and IPK1, and MRP genes coding for InsP6 transporters. Most of these are members of small gene families, with the exception of MIK, which is typically encoded by a single copy locus [35]. However, some differences among species have been reported concerning the presence of single or multiple loci for each gene function. Moreover, the tissue specificity of expression can vary among members of a gene family, and may indicate redundancy in vegetative or seed tissues. All these aspects contribute significantly to the establishment of possible lpa mutant phenotypes.

MIPS
The MIPS gene is a prime example of variable loci numbers reported across species. In fact, in the barley genome, the presence of only one MIPS gene has been described, while two are present in rice and common bean, and several are found in Arabidopsis, maize, and soybean [36,37,[40][41][42][43][44]. One of two MIPS genes in rice and in common bean, and one of four in soybean were highly expressed in developing seeds [36,37,40] and their down-regulation in rice and soybean, through mutation or RNAi, caused an lpa phenotype in seeds [43,[45][46][47]. However, a systematic study performed on mutants isolated in Arabidopsis showed that no single mutant in three MIPS genes induces an lpa phenotype in seeds, suggesting a redundant role for these genes [35]. Interestingly, a reduction in InsP6 in leaves was reported for atmips1 and atmips2 mutant plants. Moreover, atmips2 showed increased susceptibility to viruses, fungi and bacteria, while atips1 did not, suggesting a different degree of redundancy between the two genes in vegetative versus seed tissues [48]. Promoter-GUS fusion analyses of the three Arabidopsis genes showed overlapping activity of AtMIPS1, AtMIPS2 and AtMIPS3 promoters at the globular stage of seed development [44]. At later stages, strong GUS staining was present in the entire seed only for AtMIPS1, while AtMIPS2 and AtMIPS3 promoter activity was limited to the seed coat and funiculus [44]. These data suggest an essential role of myo-inositol synthesis at early stages of embryo development, but not at later stages, when the synthesis of myo-inositol mediated by AtMIPS1 alone seems to be sufficient. However, as the complete knock out of AtMIPS1 caused neither an lpa phenotype in the seed [35] nor embryo abnormalities, under normal (non stressed) conditions as observed in double and triple mutant (embryo lethality in this last case) [44], it can be deduced that the myo-inositol synthesized in the endosperm and maternal tissues by AtMIPS2 and AtMIPS3 can be transported into the embryo [44]. Immunolocalization studies, with an antibody against the three AtMIPS isoforms, showed that AtMIPS proteins appear to be specifically located within the endosperm cytosol both at torpedo stage and in mature seed, but not in the embryo, where phytic acid is accumulated [49]. These data on protein localization do not exactly correlate with the expression of AtMIPS1 [44], suggesting that a post-trascriptional regulation of AtMIPS expression is present in the seed. Moreover, the presence of AtMIPS proteins in the endosperm, a seed compartment where InsP6 is not accumulated, implies a complex interaction between endosperm and embryo during the synthesis and subsequent accumulation of InsP6 during Arabidopsis seed development [49]. In contrast, RINO1 (the rice seed MIPS gene) transcript was first detected at the apex of embryos of developing rice seeds [42]. Consequently, its expression colocalizes with InsP6-containing globoids within the scutellum and aleurone layers [4]. In soybean, GmMIPS1 expression was first detected in maternal tissue, and then in the embryo and cotyledons [50]. In both monocots and dicots, MIPS expression starts very early during seed development, a few days before the beginning of accumulation of phytic acid, and then decreases [36,37].

Myo-Inositol Reversible Dephosphsorylation: IMP and MIK
IMP genes form a small family including IMP genes, similar to mammalian ones, and IMPL (IMP-like) genes, coding for chloroplastic proteins more similar to prokaryotic IMP genes [19,36]. A different number of loci coding for IMP enzymes has been reported in different species: one IMP gene in barley and in common bean [37,51], one IMP and one IMPL gene in rice [36], one IMP and two IMPL genes in Arabidopsis [19], and three IMP genes in tomato [52]. The three Arabidopsis genes are expressed in several tissues with AtIMP showing the highest levels of expression, except in seeds, where AtIMPL1 is predominant [53]. Interestingly, a time course analysis of seed development showed IMP and IMPL gene expression correlated with that of SAL1/FRY, which codes for myo-inositol polyphosphate 1-phosphatase. This enzyme is involved in the myo-inositol "salvage pathway" and SAL1/FRY expression only partially overlaps the expression of AtMIPS genes. This expression pattern suggested that the salvage pathway may also be involved in myo-inositol synthesis during seed development [54]. More recently, it was shown that AtIMPL2 functions in the histidine biosynthetic pathway while AtIMP and AtIMPL1 catalyze the hydrolysis of inositol-and galactose-phosphates in Arabidopsis vegetative tissues [53]. During common bean seed development, PvIMP expression is similar to that of MIPS, being very high at early stages, and then declining to undetectable levels before the start of phytic acid accumulation [37].
To date, one MIK gene has been identified in plant genomes [21,23,35,37]. This gene is expressed at high levels during seed development, as shown in Arabidopsis and common bean [35,37].

PGK
Genes coding for 2-PGK have been characterized only in rice and Arabidopsis [22,25,26]. In rice, OsLpa1 (OsPGK1) is expressed in shoot, root, and panicle [23]. In Arabidopsis, two genes of this family were described, At5g60760 and At3g45090. They are present in different organs, however only At5g60760 is highly expressed during silique development and is ncessary for InsP6 synthesis [22]. Among the different Arabidopsis genes involved in phytic acid synthesis in siliques, At5g60760 together with AtMIK is the most highly expressed [35].

IPK2
IPK2 kinase is specific for the lipid-dependent pathway, which is not the major route to phytic acid in the seed [14]. However, a low phytate seed phenotype (reduction of phytic acid content by 35%-48%) in the Arabidopsis atipk2β mutant indicates the lipid-dependent pathway is active in the seed [55]. Despite similar expression in vegetative and reproductive tissues, mutation of AtIPK2α has no effect on levels of phytic acid in the seed, suggesting independent roles for the two genes in seed development [35]. AtIPK2α plays also a role in pollen germination and root growth [56], while AtIPK2β is involved in axillary shoot branching through the auxin signaling pathway [57] and its expression in tobacco leads to improved tolerance to abiotic stresses [58]. Consistent with the role of IPK2 in later steps of phytic acid synthesis, IPK2 transcripts, when assessed during seed development, showed the highest accumulation later than MIPS and IMP.

IPK1
A single IPK1 gene was described in several crop plants. The rice OsIPK1 shows the highest expression in the aleurone between 7 and 10 days after anthesis [36]; common bean PvIPK1 expression does not appear to be significantly regulated during seed development [37]; and wheat TaIPK1 doubles its expression during seed development [38]. Two IPK1 genes were described in maize and in Arabidopsis. The maize ZmIPK1A gene is expressed in a range of tissues including immature ears, seeds at 12 DAF, middle-stage endosperm and maturing embryos, while ZmIPK1B is expressed in roots [33]; the two Arabidopsis IPK1 genes are expressed in different tissues, with only one expressed in siliques that causes an 83% reduction in seed phytate level when knocked down [31,55]. Three IPK1 genes were identified in soybean and only one is highly expressed in the seed [65].

MRP
The first gene coding for an ABCC-type InsP6 transporter was mapped to the maize ZmMRP4 locus [66]. The functional characterization of this gene family was performed with the Arabidopsis homolog AtMRP5 (also referred to as AtABCC5) [67], previously characterized for its involvement in regulating stomatal movements and drought tolerance [68,69]. Phylogenetic analyses of MRP type transporters indicate that InsP6 transporters are represented as single copy genes [70][71][72], as described for Arabidopsis AtMRP5, maize ZmMRP4 and rice OsMRP5 [66,67,73]. ZmMRP4 and OsMRP5 are expressed in different organs including seeds [66,73]. AtMRP5-promoter fusion to a GUS reporter gene showed staining mainly in vascular tissues and in guard cells, with no staining in seeds [74]. However, from publicly available microarray data, it is clear that AtMRP5 is expressed at different stages of seed development [75]. It has been recently shown that soybean and common bean, two closely related legume species, bear a paralog copy (PvMRP2 and GmMRP13) of the genes characterized for their role in phytic acid accumulation (PvMRP1 and GmMRP3 and GmMRP19) [76,77]. In common bean PvMRP1 and PvMRP2 are expressed in almost all tissues at similar levels, with the exception of developing seeds, where PvMRP1 is expressed at consistently higher levels than PvMRP2 (unpublished resultsA similar behavior is found for the corresponding soybean genes [78].

Low Phytic Acid (lpa) Mutants
Several lpa mutants have been isolated in important grain crops, such as barley, maize, rice, wheat, soybean, common bean and pea (Table S1). lpa mutations can be grouped into three classes, depending on the affected step of the biosynthetic pathway or mode of compartmentation/transport: (i) mutations involved the first steps of the pathway (from glucose 6-P to myo-inositol 3-phosphate), commonly indicated as "supply pathway"; (ii) mutations perturbing the end of the pathway (from myo-inositol 3-phosphate to InsP6); and (iii) mutations affecting tissue compartmentation of InsP6 and/or its transport and storage to the vacuole (MRP transporter) ( Figure 1). Mutants belonging to the first and the third classes are generally characterized by decreased InsP6 levels accompanied by a molar equivalent increase in inorganic Pi, but not by accumulation of lower InsPs (inositol phosphates with up to five phosphate residues), a characteristic specific of the second class of mutants ( Table 1).
The similarities between the first and third class of mutants triggered some confusion in the initial characterization of some lpa mutants in the InsP6 transporter genes (MRP). The first hypothesis was that they were affected in the MIPS gene [61,79]. In fact, a reduced expression of the ZmMIPS1S gene was reported in the maize lpa1 mutant. However, mapping of the maize lpa1 locus led to the identification of a defective ZmMRP4 [66]. Available maize genomic data helped to identify that the ZmMIPS1S and the ZmMRP4 genes map very closely on chromosome 1S, thus explaining the incorrect association of the mutation to the ZmMIPS1S gene [61,66,79,80]. Phylogenetic studies identified the AtABCC5/AtMRP5, a high affinity InsP6 ATP-binding cassette transporter, as the closest

Metabolic and Phenotypic Traits of Low Phytic Acid Mutants
To ensure mineral bioavailabity, the reduction of the phytate:mineral cation molar ratio is very important, thus a consistent phytic acid reduction is highly desirable in lpa mutants. Unfortunately, negative agronomic traits, such as low germination rates, reduced seed development and weight, and stunted vegetative growth, have been frequently reported for many lpa mutants, making them of limited value to breeders [61,[86][87][88][89]. For many crops, obtaining lpa mutants with no or very limited pleiotropic agronomic defects still remains a challenging goal, although a few well performing lpa mutants have been reported [90][91][92]. In barley, at least one lpa mutant (Hvlpa1-1) has been shown to be equivalent to or even better than its wild type parent. In addition, some lpa mutations do not severely affect the yields, especially in non-stressful production environments, thus suggesting that at least some lpa mutants show potential for use in breeding [90,93] and a number of lpa varieties have been registered [94,95].
The number and strength of pleiotropic effects on InsP6 biosynthesis in lpa mutants are the result of differences in copy number (functional redundancy) and the spatio-temporal expression of genes and their respective products (protein and metabolite localization). In addition, an increasing number of studies describe tight integration of InsP6 metabolic and signaling pathways. The existence of complex metabolic crosstalk among enzyme activities and the type and abundance of different InsPs in a differentiated cell types and tissues may result in unexpected phenotypes (reviewed by [96][97][98][99][100]).
The most severe pleiotropic defects have been found in lpa mutants carrying defective MIPS or MRP genes, as well as in mutants in other genes that result in phytic acid reductions of more than 70% (Table S1 and references therein). These negative effects on agronomic performance and seed viability are not surprising, given that most of the lpa mutations affect myo-inositol and other metabolites that are important for normal development. Since MIPS activity is the only source of the myo-inositol ring, and since myo-inositol is an essential cellular metabolite important to numerous pathways and functions (Figure 2), perturbing MIPS expression may prove deleterious, if not lethal, even if MIPS suppression is specifically targeted to the seed. For example, mips soybean mutants (L33 and Gm-lpa-TW75-1), with 50% phytic acid decrease, showed reduced field emergence, especially when seeds were produced in a subtropical environment (the so-called "seed-source" effect), while high seed abortion and 95% phytic acid reduction occur in transgenic soybean plants with an almost complete suppression of the MIPS gene by RNAi [45,86,101]. Defects in embryogenesis and embryo lethality have been reported for Arabidopsis mips mutants [102,103]. Maize and rice lpa mutants, with defects in the MRP gene, have very similar phenotypes, and the severity of the phenotype (seedling or embryo death) correlates with the extent of phytic acid reductions [61,73,88,104,105]. Surprisingly, no embryo defects neither seed lethality/abortion have been reported for known mrp mutants of dicot species (soybean, common bean and Arabidopsis), although phytic acid is significantly reduced in their seeds [67,82,85]. Most likely, tissue and cellular compartmentation of the phytic acid pathway significantly contributes to the effects on seed development caused by perturbations due to lpa mutations [14,50]. Altered myo-inositol contents have been reported in many lpa mutants. As expected, mips and mik mutants have seed myo-inositol levels lower and higher than wild type, respectively (Table S1 and references therein). The soybean mutant Gm-lpa-ZC-2, harboring a non-functional IPK1 protein, has increased myo-inositol levels [65]. Contrasting data are reported in mrp mutants of maize and common bean. In these mutants MIPS gene expression is downregulated, thus lower seed myo-inositol content would be expected. This occurs in the bean lpa1 and the maize lpa1-241 mutants, however higher seed myo-inositol has been reported in the maize lpa1-1 mutant [66,76,106]. An inverse relationship between MIPS expression and seed myo-inositol content is also observed in two allelic mrp mutations of rice and in three barley lpa mutants (lpa2-1, lpa3-3 and M955), for which the molecular defects are still unknown [81,107]. Changes in myo-inositol content also affect the synthesis of derived metabolites, such as galactinol and raffinosaccharides (Table S1) [43,76,[107][108][109][110]. Seed myo-inositol content has also been suggested to correlate with response to ABA during seed germination [19,76,111]. The Arabidopsis and common bean lpa (mrp) mutants have opposite myo-inositol seed contents and also show opposite responses to ABA during germination, which is strongly inhibited in common bean lpa1 (Pvmrp1) mutant, while it is unaffected in the Arabidopsis mrp5 seeds [68,76]. Ins(1,4,5)P3 (InsP3) levels are critical for ABA response in germinating seeds, as ABA hypersensitivity is accompanied by an increase in InsP3 catabolism [112][113][114]. Most likely, the flux of myo-inositol used to feed the lipid dependent pathway leading to the production of Ins(1,4,5)P3 or the regulation of the salvage pathway used for myo-inositol catabolism might be affected in these mrp mutants ( Figure 1).
Misregulation (in most cases downregulation) of multiple genes for the synthesis and transport/accumulation of phytic acid has also been reported in a number of lpa mutants. Decreased expression of MIPS and IMP has been shown in lpa2-1, lpa3-3 and M955 barley mutants [107]. A point mutation in the bean phytic acid transporter (Pvmrp1) causes a general transcriptional down regulation of the genes of phytic acid pathway, indicating that the myo-inositol cellular pool and phytic acid biosynthesis are controlled by phytic acid itself and/or other constituents of the pathway [76]. Similar results have been reported in a more extensive study on the Arabidopsis lpa mutants atmik, atitpk1, atitpk4, atipk1, atipk2β, atmrp5 and at2pgk [35]. In these knock-out mutants, the expression of several genes of the InsP6 pathway was affected, confirming that other constituents of InsP6 metabolism also modulate the transcription of genes in the InsP6 pathway. Another interesting outcome of this study was the observation that the lipid dependent pathway, compared to the lipid independent one, has a more relevant regulatory role in mediating stress response. In fact, defects affecting inositol phosphate kinases (atipk1, atipk2β, atitpk1 and atitpk4-1) are more relevant for abiotic stress (NaCl, mannitol and H2O2) sensitivity, than mutations in genes of the early pathway (atmik1 and at2pgk) or in the phytic acid transporter (atmrp5).

Enzymes and Metabolites of the Phytic Acid Pathway Have Regulatory Roles in Cell Signaling and Plant Processes
Despite its importance, the biological role of InsP6 in plants is still poorly understood in both normal and extreme environmental conditions. Myo-inositol synthesis and catabolism impact metabolites involved in many critical plant biochemical pathways, such as (i) the production of compatible solutes, like galactinol, raffinose family oligosaccharides, pinitol and cell wall polysaccharides; (ii) the generation of inositol polyphosphates (InsPs), phytic acid and inositol polyphosphate pyrophosphates (PP-InsPs); and (iii) the synthesis of phosphoinositides and the production of Ins(1,4,5)P3 ( Figure 2). Furthermore, D-glucuronic acid, the primary breakdown product of Ins, is utilized in the synthesis of various cell wall pectic and noncellulosic compounds and ascorbic acid [115][116][117]. A growing body of data is elucidating the roles played by inositol metabolism in diverse plant developmental and physiological processes including signal transduction [113,118], sugar signaling [115], storage and polar transport of auxin [44,119], membrane trafficking [120], abiotic and biotic stress response [48,121], phosphorus homeostasis [55,122], photomorphogenesis [64], chromatin modification and remodeling [122,123], and mRNA nuclear export [124]. Highly phosphorylated inositols (InsP5, InsP6, InsP7, InsP8) have also been shown to serve as ligands of plant hormone receptors [125][126][127]. Moreover, several genes, enzymes and compounds for inositol phosphates and, eventually, phytic acid synthesis are part of cytosolic and nuclear metabolic pools with a central role in cellular metabolism. Thus, any perturbations of the pathway, such as those occurring in lpa mutants, may significantly impact seed and plant development ( Figure 2).

Signal Transduction
Inositol metabolism plays a significant role in a wide range of plant developmental and physiological processes, such as response to diverse stimuli (light, gravitropism, abiotic and biotic stresses), downstream responses to ABA and sugars, and auxin mediated processes, among others, as described below. Key molecules are the well known Ins(1,4,5)P3 (InsP3) and InsP6. Very recently, inositol pyrophosphates (InsP7 and InsP8) have been proposed as unique signaling molecules involved in energy sensing and metabolism [127][128][129][130][131]. These high energy molecules, containing a linear chain of two (PP) or three (PPP) phosphates linked to an InsP6 molecule (Figure 1), have been implicated in determination of the phenotypes of lpa mutants, based on their increased expression in seeds of Arabidopsis and maize mrp mutants (mrp5 and lpa1-1 respectively) [129].
In eukaryotes, the myo-inositol signaling pathway relies on InsP3 as second messenger to trigger intracellular Ca 2+ release from intracellular stores. When exposed to an external stimulus, the cell responds increasing InsP3, which is generated by the action of phospholipase C (PLC). The signal induced by InsP3 can be terminated through a catabolic pathway (the so-called "salvage pathway"), in which myo-inositol polyphosphate phosphatases (Ptases) remove the 5-phosphate to regenerate myo-inositol, or by InsP3 removal through sequential phosphorylation mediated by IPK2 (Figure 1). Thus, InsPs and PtdInsPs are interdependent compounds, as PtdInsPs are used as substrates by PLCs to produce InsPs, and InsPs breakdown produces myo-inositol used as substrate to produce PtdInsPs. The direct role of InsP3 in signaling is in question, as an emerging idea is that InsP3 plays a role simply as a precursor of InsP6 [121]. In fact, in guard cells, InsP6 has been shown to trigger intracellular Ca 2+ release after ABA addition with an efficiency ≈100 times higher than that of InsP3 [118]. Moreover, there is no evidence of a canonical InsP3 receptor in plants, although this holds true also for InsP6. One possibility is that plants do not possess an InsP-regulated calcium channel, and an entirely different and uncharacterized mechanism allows both InsP3 and InsP6 to regulate Ca 2+ release, either simultaneously, or independently [111]. Most likely, both InsP3 and InsP6 have parallel and overlapping functions in plants, and there is a large body of evidence showing that InsPs impact cellular Ca 2+ levels. Undoubtedly a strong correlation exists between these two molecules as demonstrated by the finding that transgenic plants and plant cells that constitutively break down InsP3 or synthesize PtdInsP2 contain changes in InsP6: so changes in InsP3 are mirrored by changes in InsP6 [132][133][134].
Rapid increases in InsP3 have been reported in response to myriad stimuli, such as gravitropism, light, salt stress, gibberellic acid, anoxia, cold, heat, drought, and exposure to plant pathogens and elicitors [111,113,114]. Thus, directly or indirectly, InsP3 is crucial to signal transduction. Intriguingly, conditions that lower InsP3 mediated signals, while revealing predictable phenotypes based on signal transduction paradigms, also showed unexpected phenotypes. For example, Arabidopsis plants overexpressing human type I inositol polyphosphate 5-phosphatase (InsP 5Ptase), have a 2%-5% reduction of InsP3 and even under stimulation these levels do not show any increase [135]. In normal conditions no obvious phenotypic alterations were detected, while, as expected, InsP3-mediated responses, such as gravitropism, were delayed in the InsP 5Ptase overexpressing transgenic plants. Surprisingly, human InsP 5Ptase plants were more drought tolerant [132], an unexpected finding based on the classical InsP3 signaling paradigm that predicts decreased levels of InsP3 and calcium would decrease stomatal closure and therefore drought tolerance. This finding might explain the contrasting response to ABA during seed germination observed in common bean and Arabidopsis lpa mutants (see above).
As already mentioned, elevated InsP3 has been shown to correlate with downstream responses to ABA and sugars [107,108]. A plant inositol polyphosphate 5-phosphatase (5Ptase13; At1g05630) has been shown to be involved in linking myo-inositol signaling to sugar sensing and stress response [136]. The 5Ptase13 protein contains a WD40 repeat region that specifically interacts with a Sucrose non fermenting-1-1Related Kinase (SnRK1.1), which functions as a sensor of energy and stress in plants [137]. When sugars are in limited supply, 5Ptase13 interacts with SnRK1.1 and prevents its proteosomal degradation, resulting in transcriptional induction of genes involved in low nutrients/stress response [110,115]. Several other developmental/signaling defects, altering blue light response, gravitropism, and vesicular trafficking for PIN mediated auxin transport have been reported in 5Ptase13 mutants, indicating that hydrolysis of Ins(1,4,5)P3 or PtdIns(4,5)P2 are critical for plant development and signaling [138,139].
Inositol polyphosphate kinase (IPK2) is a key component for InsP3 turnover, as it phosphorylates Ins (1,4,5)P3 successively at the 6-and 3-positions to generate Ins(1,4,5,6)P4 and Ins(1,3,4,5,6)P5, respectively, and IPK2 has very important regulatory roles, indicated by subcellular localization in both the nucleus and cytoplasm (discussed below). Downregulation of AtIPK2α through antisense inhibition has been shown to result in enhanced root growth and pollen germination [56]. Since application of InsP3 can enhance root growth, most likely AtIPK2α silencing may cause the accumulation of InsP3 in addition to the expected depletion of InsP4 and InsP5.

Vesicle Trafficking and Polar Auxin Transport
A number of lpa mutants displays severe defects in embryo and/or plant development, and this often occurs when MIPS genes are mutated or their expression is greatly altered [44,45,88] (Table S1). In developing seeds MIPS genes are highly expressed, suggesting an important role of myo-inositol in seed/embryo development. Using double and triple mips mutants of Arabidopsis, which display phenotypes resembling those of auxin mutants, it has been shown that MIPS-mediated de novo synthesis of myo-inositol is essential for maintaining the normal function of endomembrane trafficking and for maintaining endomembrane structure. This is critical for correct auxin transport and thus for correct auxin localization during embryo pattern formation [44]. PtdIns contributes 21% of the phospholipids in nonphotosynthetic plant membranes, and the various phosphorylated forms of PtdIns have critical roles in cytoskeletal rearrangements, membrane trafficking, and organelle labeling. Indeed, the finding that Arabidopsis mips1/mips3 double mutants can be rescued by over-expressing the phosphatidyl synthase 2 gene (AtPIS2), controlling the entry point for PtdIns synthesis, supports the hypothesis that PtdIns and phosphatidylinositides are essential for endomembrane structure and trafficking [44]. Another study showed that the atmips1 mutant has reduced levels of PtdIns and altered trafficking of the auxin efflux carrier PIN2 [98, 106], suggesting that lower PtdIns content negatively impacts on vesicular trafficking of PIN2. Interestingly, the atmips1 phenotype was evident only if plants were grown under high light intensity conditions (higher than160 μmol m −2 ·s −1 ) [44]. Although currently not intensely studied, it is likely that not only PIN proteins but also many other membrane proteins will depend on PtdIns to reach their membrane destinations. This may indicate other obvious and broad ramifications of inositol metabolisms in plant function [100]. The critical role of myo-inositol as precursor of PtdIns species is also confirmed by the finding that InsP3 is linked to Ca 2+ signaling and control of directional auxin transport mediated by polar PIN auxin transporters [140]. During a screen for suppressors of PIN1 overexpression, Zhang et al., isolated an Arabidopsis mutant (supo1) affected in an inositol phosphate 1-phosphatase known as SAL1/FRY1 [114]. The mutant had increased InsP3 and cytosolic Ca 2+ levels and was able to restore a wild type phenotype in 35S::PIN1 seedlings. PIN overexpressing plants have short roots, agravitropic root growth, and elevated auxin in root tips, presumably caused by the ectopic presence of basally localized PIN1 in the epidermal cells.
Involvement of InsP metabolism in auxin mediated processes is also provided by the finding that plants over-expressing the AtIPK2β gene show an attenuation of the inhibition of IAA induced primary root elongation. This indicates that AtIPK2β can negatively regulate auxin signaling [57]. At least in part, this regulation is transcriptional, since AtIPK2β over-expressing plants repress the expression of CYP83B1 (involved in auxin biosynthesis), MAX4 and SPS (required for auxin-mediated bud inhibition and outgrowth), and stimulate PIN4 expression [57].

Biotic and Abiotic Stress Response
Limiting PtdIns synthesis and/or InsP6 content may also impact plant defense response. The Arabidopsis atmips1 mutant showed enhanced resistance to pathogens and spontaneous cell death, thus implying that MIPS is a repressor of programmed cell death (PCD) [97]. In this mutant decreased contents of myo-inositol, ascorbic acid and PtdIns correlated with elevated ceramide levels, sphingolipid precursors associated to cell death. The hypothesis is that spontaneous cell death is the result of altered oxidative stress sensitivity, induced by changes in myo-inositol, galactinol, and ascorbic acid, along with elevated ceramides and hydroxyceramides that result from decreased PtdIns availability for sphingolipid production [111]. Myo-inositol metabolism and defense response are also linked to carbohydrate metabolism, as a mutation in the hexokinase 1 (HXK1) enzyme has been shown to suppress mips1 light-dependent PCD, indicating an epistatic relationship between mips1 and hxk1 mutants [141].
Like other enzymes of the InsP6 biosynthetic pathway (see below), MIPS has a double cellular localization: cytosolic and nuclear. Interestingly, in a recent work in Arabidopsis, it has been demonstrated that MIPS1 protein binds directly to its own promoter to stimulate transcription by locally inhibiting the activation of ARABIDOPSIS TRITHORAX-RELATED 5 and 6 (ATXR5 and 6)-dependent heterochromatin marks generated by a transposable element [123]. Upon activation of pathogen response, elicited by bacterial flagellin (flg22) treatment, the inhibitory action of MIPS1 on ATXR5/6 is alleviated and expression of MIPS1 decreases. This dual function of MIPS1 may ensure MIPS1 gene expression under normal growth conditions, and its down-regulation during pathogen attack to induce PCD [123]. It would be very interesting to verify if such regulatory mechanism is specific to Arabidopsis or is present in other plants.
Impaired resistance to wounding and herbivory has also been reported in transgenic plants in which inositol polyphosphates are globally reduced through expressing a human type I InsP 5-Ptase. In this case, plants treated with flg22 show impaired Ca 2+ elevation, accompanied by a decrease in the expression of several defense related genes, suggesting that the inability to propagate an InsP3 signal is the primary basis for the altered defense response observed [142]. Other evidence for the involvement of InsPs in wound signaling comes from work on atipk1-1 mutants, which exhibit a pathogen hypersensitive phenotype and increased defense capability via jasmonate receptor COI1-mediated processes, including wound-induced gene expression, defense against caterpillars or root growth inhibition by jasmonate [48,143].
Perturbations of the InsP6 pathway have been reported to alter plant response to environmental stimuli. For instance, heterologous expression of AtIPK2β in tobacco leads to improved tolerance to diverse abiotic stresses (osmotic, drought, freezing temperature, oxidative stress) [58]. Similar results have been obtained by over-expressing the IbMIPS1 gene in sweet potato (Ipomoea batatas L. Lam.). Transgenic plants showed significantly enhanced salt and drought tolerance, and stem nematode resistance. Following exposure to salt and nematode stresses, transcriptome analysis revealed up-regulation of MIPS and IMP genes, together with genes for phosphatidylinositol and ABA signaling pathways, stress responses and ROS-scavenging [144].

Nuclear Functions and Regulation of Phosphorus Homeostasis
A central issue in InsP metabolism is the cellular localization of InsP pools. Dual localization, cytosolic and nuclear, of inositol phosphates and of their biosynthetic enzymes is well documented in yeast [145] and has been demonstrated in plants for many enzymes of the InsP6 pathway, including a number of phosphatidylinositol kinases, AtMIPS1, AtIPK2α, AtIPK2β, AtIPK1 and AtITPK1 [56,64,98,122,123,146].
In yeast, the transcriptional activation of PHO genes in response to Pi deficiency is coordinated with regulated chromatin remodeling [145]. Furthermore, a number of InsPs species have been shown to participate to transcriptional gene regulation via chromatin remodeling and histone modification [147]. For example, proper expression of INO1, coding inositol 1-phosphate synthase, involves integration of INO80 and SNF, and ISW2, which act as positive or negative regulators of transcription, respectively. InsP4 and InsP5 have also been shown to stimulate nucleosome mobilization by the SWI/SNF complex. On the contrary, InsP6 inhibits nucleosome mobilization by NURF, ISW2 and INO80 complexes [148]. The yeast IPK2 and its products, InsP4 and InsP5, are involved in transcriptional regulation in response to environmental and nutritional stresses. IPK2 has been shown to be allelic to Arg82/ArgRIII, a component of the ArgR-Mcm1 transcriptional complex that regulates gene expression for arginine metabolism [149]. IPK2 is also required for the induction of some phosphate responsive genes (PSR), like PHO5, by modulating the chromatin remodeling complexes SWI/SNF and INO80, under normal Pi supply [150]. Xia and coworkers [146] demonstrated that Arabidopsis AtIPK2β is able to complement a yeast Arg82/ArgRIII mutant lacking a functional ArgR-Mcm1 transcription complex. However, no data are available to support a similar role for AtIPK2β in PSR gene induction, as was shown in yeast.
Plant Pi homeostasis is a highly regulated process, and involves the perception of Pi present in the environment, followed by acquisition, remobilization and recycling of Pi [151]. The Arabidopsis mutant atipk1-1 exhibits an 83% reduction in seed InsP6, along with an increase in InsP4 and InsP5 in seed and vegetative tissues. Furthermore, atipk1-1 plants show longer root hairs and aberrant phosphate sensing. The plants behave as if they were in a phosphorous-limiting environment, indicating a role of IPK1 in the regulation of Pi acquisition machinery [55].
In a recent work, transcriptional analysis of roots of atipk1-1 mutants showed perturbations of a subset of Pi starvation responsive genes, together with increased expression of genes involved in Pi uptake, allocation and remobilization [122]. The authors observed that the transcriptional activation correlate with reduction of the chromatin association of histone variant H2A.Z. In yeast the Pi responsive genes PHO5 and PHO84 require InsP4 and InsP5 for proper remodeling of chromatin structure [150], thus it might be speculated that in plants, specific InsPs species serve as signals and regulate the eviction of H2A.Z from PSR genes.
Perturbation of Pi and sulfate (SO4) homeostasis and signaling have been observed in lpa plants obtained by over-expressing a bacterial phytase in Arabidopsis [152]. These plants (PHY-US417) showed up to 50% and 45% increases in shoot Pi and SO4 concentrations, respectively, and improved plant growth with enhanced root growth capacity in Pi deficiency. These findings were supported by expression analysis of a subset of genes coding for Pi transporters (PHT1,1, PHT1,4, PHO1 and PHO1,H1) and SO4 (SULTR1,2), the expression of which was upregulated in both overexpressing PHY-US417 and atipk1-1 plants. Interestingly, overexpressing PHY-US417 transgenic plants and the atipk1-1 also showed a stronger remobilization of iron during germination.
One of the key points for proper gene expression is the regulation of nuclear export of mRNA, a process that requires the directional translocation of mRNA-ribonucleoprotein particles (mRNPs) through nuclear pore complexes (NPCs). In eukaryotes, Gle1 is a component of the NPC. In yeast, Gle1 and its cofactor InsP6 activate the DEAD-box ATPase, Dbp5, to allow mRNA export at the NPC. Gle1 is also found in the cytosol, where it plays a role in translation initiation and termination in Dbp5-independent and -dependent manners, respectively [145]. In a very recent work, Lee et al. [124] showed that plant Gle1, in conjunction with InsP6, functions as an activator of the ATPase/RNA helicase LOS4 (low expression of osmotically responsive genes 4, homolog of yeast Dbp5), which is involved in mRNA export in plants, supporting the Gle1-InsP6-Dbp5 paradigm proposed in yeast. Interestingly, an ipk1 mutant has been shown to be defective in nuclear mRNA export, however, the ectopic expression of Gle1 variants with enhanced InsP6 sensitivity was able to rescue the mRNA export defect of the ipk1 mutant. Moreover, a significant improvement of vegetative growth, seed yield, and seed performance of the mutant was observed, suggesting that Gle1 is an important factor responsible for mediating InsP6 functions in plant growth and reproduction [124].

Hormonal Signaling
In recent years, it was discovered that two important phytohormone receptors, TIR1 and COI1, required for auxin and jasmonic acid (JA) signaling, respectively, contain InsPs as structural cofactors, thus widening the regulatory roles involving the InsP6 pathway [126,153]. TIR1 is part of the ubiquitin E3 ligase complex SCF TIR1 . Upon auxin binding, it recruits specific transcriptional repressors (the Aux/IAA repressors) for ubiquitination by the SCF complex. This marking process leads to the degradation of the repressors by the proteasome, alleviating repression and leading to expression of specific auxin responsive genes [154]. The determination of the crystal structure of TIR1 revealed the presence of an InsP6 molecule bound in close proximity to the auxin binding pocket [126]. In a similar way, COI1 is the F-box component of a SCF ubiquitin E3 ligase complex that recruits Jasmonate ZIM-domain (JAZ) transcriptional repressors upon binding to the bioactive JA-isoleucin conjugate (JA-Ile). This triggers JAZ polyubiquitination and subsequent proteasomal degradation, and results in de-repression of MYC2 dependent transcription of jasmonate repsponsive genes [154]. Analysis of the crystal structure of the JA receptor revealed a binding pocket comprised of COI1, bound to JAZ, and containing an InsP5 molecule as a structural cofactor [154]. Arabidopsis ipk1 mutants, that have elevated levels of InsP5 and display enhanced wound-induction of various defense genes, were found to be more sensitive towards exogenous methyl-JA, and exhibited increased defensive capacity against caterpillar herbivory [143], supporting the importance of InsP5 contribution to COI1 function. However, very recently, a link between InsP8 and jasmonate-dependent defense has been discovered. The authors do not rule out that other inositol polyphosphates, other than InsP8, may influence assembly of the jasmonate receptor complex [127], but insect larvae feeding on Arabidopsis vih2 mutant plants (unable to synthesize InsP8, Figure 1) showed a significant weight increase compared with larvae feeding on control plants, indicating that VIH2 plays a role in activating defenses that interfere with insect herbivore development. Furthermore, molecular data indicate that vih2 plants are defective in jasmonate perception.

Conclusions
This object of this review is to integrate current knowledge about different aspects of phytic acid pathway and lpa mutants with the most recent literature concerning the regulatory roles of the multiple components of the pathway in cell signaling and plant processes. In the last decade, the range of important crop plants with lpa mutants has expanded rapidly. Known lpa mutants have seed phytic acid reductions ranging from 10% to 90%, and in many cases their improved value in animal and human nutrition has been demonstrated. However, good agronomic performance and yield stability are still challenging for lpa mutants in many crops. In no case were these mutants due to a spontaneous mutation, underlying the important role of InsP6 pathway for the plant. The increasing number of plant processes in which phytic acid and its metabolism have been shown to play a key role clearly indicates that we need to increase our knowledge of the role of InsPs and phytic acid in the integration and functioning of metabolic and hormonal signaling pathways and in response to biotic and abiotic stresses. This knowledge will be fundamental to understand how far we may go to obtain stable and productive lpa mutants in different crops and how to drive their genetic improvement.