To ensure the success of the new generation in annual species, the mother plant transfers a large proportion of the nutrients it has accumulated during its vegetative life to the next generation through its seeds. Iron (Fe) is required in large amounts to provide the energy and redox power to sustain seedling growth. However, free Fe is highly toxic as it leads to the generation of reactive oxygen species. Fe must, therefore, be tightly bound to chelating molecules to allow seed survival for long periods of time without oxidative damage. Nevertheless, when conditions are favorable, the seed's Fe stores have to be readily remobilized to achieve the transition toward active photosynthesis before the seedling becomes able to take up Fe from the environment. This is likely critical for the vigor of the young plant. Seeds constitute an important dietary source of Fe, which is essential for human health. Understanding the mechanisms of Fe storage in seeds is a key to improve their Fe content and availability in order to fight Fe deficiency. Seed longevity, germination efficiency and seedling vigor are also important traits that may be affected by the chemical form under which Fe is stored. In this review, we summarize the current knowledge on seed Fe loading during development, long-term storage and remobilization upon germination. We highlight how this knowledge may help seed Fe biofortification and discuss how Fe storage may affect the seed quality and germination efficiency.
The seed is a very special stage in the life of a plant. It is the basis for the next generation and the survival of the species. To ensure the success of the new generation, the mother plant transfers a large proportion of nutrients it has accumulated during its vegetative life to its seeds. Most work on seed nutrient storage focuses on carbohydrate, lipids and proteins, which provide the bulk energy, carbon and nitrogen required to initiate development. Depending on the species, 30–90% of plant nitrogen is allocated to seeds at the end of the plant life cycle . However, mineral stores are also very important to sustain seedling growth. Especially, iron (Fe) is required in large amount in mitochondria, which provide the energy and redox power during the initial stage of the seedling development. For example, in the case of Arabidopsis (Arabidopsis thaliana), 50–55% of Fe is allocated to seeds at the end of the life cycle [2,3]. Moreover, the nutrients need to be stored in a compact and stable form. Seed storage proteins form pseudo crystalline aggregates in protein bodies within seed embryo cells. Free Fe is highly toxic as it leads to the generation of reactive oxygen species (ROS) through the Fenton reaction. In cells, Fe must therefore be tightly bound to chelating molecules to prevent the formation of ROS. This is probably even more important in seeds than in other organs, as seeds represent resistance forms designed to survive for long periods of time, often several years, and through stresses. Nevertheless, when conditions are favorable, the nutrients stored in seeds have to be readily remobilized to sustain the early development of the seedling before it acquires photosynthetic capacity and becomes able to take up nutrients from the environment. Fe is a major component of the photosystem I, and its availability to the young seedling is necessary to achieve the transition toward active photosynthesis. This is likely critical for the vigor of the young plant.
From the point of view of animals and especially human beings, seeds are a major food resource. They provide not only carbohydrates, proteins and fat but also micronutrients such as Fe and zinc, which are essential for human health. About two billion human beings suffer from Fe deficiency, which leads to asthenia, anemia and in extreme cases to death. The prevalence of Fe deficiency is highest in populations that rely mostly on a plant-based diet because plants, and seeds, in particular, do not provide sufficient Fe . Understanding the mechanisms of Fe storage in seeds is a key to improve their Fe content and availability in order to fight Fe deficiency using biofortification. From the agricultural perspective, seed longevity, germination efficiency and seedling vigor are also very important traits that breeders and farmers seek to improve. Here also, seed nutrient content and especially the content and chemical form of Fe may be critical. In this review, we will summarize the current knowledge on seed Fe loading during development, long-term storage and remobilization upon germination, updating the last review on the topic . We will highlight how this knowledge may help to improve seed Fe biofortification, longevity, germination efficiency and stand establishment.
Iron uptake and translocation to the shoots
Plants take up Fe using distinct strategies according to species [6,7]. Most plant species use strategy I, which is based on the reduction of Fe in the rhizosphere followed by its uptake by a transporter for Fe(II). In contrast, graminaceous species, which include the major grain crops such as rice, wheat and barley, have evolved strategy II. Strategy II is based on the release of small Fe(III) chelating molecules called phytosiderophores in the rhizosphere and subsequent uptake of siderophore–Fe(III) complexes. Under Fe deficiency, strategy I plants activate the expression of the genes involved in Fe solubilization and uptake. The molecular players in this response have been identified in Arabidopsis but are essentially conserved in other species. The Arabidopsis plasma membrane H+-ATPase 2 (proton pump) drives rhizosphere acidification, which is important to solubilize Fe . The coumarin transporter pleiotropic drug resistance 9 (PDR9) allows the secretion of coumarins with Fe binding and reducing properties in the rhizosphere. Enzymes of the coumarin biosynthetic pathway are induced along with PDR9 [9,10]. The membrane-bound ferric chelate reductase FRO2 reduces chelate-bound Fe(III) to Fe(II), which is taken up into root cells by the divalent metal iron regulated transporter 1 (IRT1) [11,12]. IRT1 is not specific for Fe(II) and drives the uptake of a range of other divalent metal cations including Zn, Mn, Co, Cd and Ni [13–15]. A sophisticated mechanism enables the removal of IRT1 from the plasma membrane when intracellular concentrations of non-Fe metals increase . In addition, several genes involved in the sequestration of these metal cations are activated under Fe deficiency . In graminaceous species such as rice, maize and barley, the components that are upregulated for Fe acquisition under Fe deficiency include the phytosiderophore efflux and Fe-siderophore influx transporters as well as the biosynthetic pathway of phytosiderophores of the mugineic acid family . The S-adenosyl methionine cycle is activated to synthesize methionine, which is used by the nicotianamine synthase (NAS) to produce nicotianamine (NA), the precursor of mugineic acid (MA). Although NA is also synthetized in strategy I species, mugineic acids are unique to graminaceous species. NA is converted to mugineic acid by the nicotianamine aminotransferase (NAAT) and to deoxymugineic acid (DMA) by the deoxymugineic acid synthase (DMAS). MA and DMA are secreted in the rhizosphere by an efflux transporter named transporter of mugineic acid 1 (TOM1) in rice (Oryza sativa) or yellow stripe 3 (YS3) in maize (Zea mais) [18,19]. Fe-siderophore complexes are taken up by specific influx transporters, yellow stripe 1 (YS1) in maize and YS1-Like 15 (YSL15) in rice [20–22]. Despite the difference in their uptake strategies, the control of Fe deficiency responses involves a similar network of negatively and positively regulating bHLH (basic Helix Loop Helix) in graminaceous plants and the other species [23,24]. In addition, hemerythrin domain containing RING proteins that mediate the Fe-dependent degradation of specific bHLH, are also found in both strategy I and strategy II plants [25–27].
To be translocated to the shoots, Fe has to go through the endodermis cell layer and to be released into the xylem sap together with ligands that maintain it in a soluble form. Recently, suberin deposition around the endodermis has been shown to provide a control point for Fe translocation. Under Fe deficiency, the suberin layer is degraded allowing the transfer of Fe from the root cortex to the central cylinder where it can be loaded into the xylem for translocation to the aerial parts [28,29]. Based on its expression pattern and plasma membrane localization, the Fe efflux transporter ferroportin (FPN1)/iron regulated 1 (IREG1) has been proposed to mediate Fe loading in the xylem in Arabidopsis . The MATE family citrate efflux transporters, ferric reductase deficient 3 (FRD3) in Arabidopsis and FRD3-Like 1(FRDL1) in rice, play an important role in the translocation of Fe and other metals by loading citrate that chelates Fe and prevents its precipitation in the xylem sap [31–35].
Remobilization from other organs to the seeds
At the vegetative stage, most of the Fe translocated to aerial parts is used for photosynthesis in chloroplasts in mesophyll cells. Photosystem I and Ferredoxin, which are very abundant in chloroplasts, contain numerous FeS clusters. This assumption based on biochemical needs is supported by cell fractionation experiments and Fe imaging in leaves [36–38]. Upon transition to the reproductive stage, a large part of the Fe present in vegetative tissues is transferred to the seeds. In Arabidopsis, 50–55% of total plant Fe ends up in the seeds at the end of the life cycle (the iron harvest index, Figure 1) [2,3]. A recent study based on Fe isotope labeling indicated that most of the Fe stored in seeds originates from vegetative tissue, implying a major remobilization of Fe .
Fe uptake and remobilization to seeds in
Prior to its movement from the vegetative tissues to the seeds, Fe has to be made available in the senescing tissues. Autophagy was proposed to play a key role in nutrient remobilization from vegetative tissues to seeds . This was based on the finding that autophagy is involved in nitrogen remobilization to seeds . Recently, the finding that a mutation in Autophagy 5 (ATG5), which strongly impairs autophagy, decreases the proportion of Fe allocated to seeds from 50–55% to 15–20% highlighted the importance of autophagy for Fe remobilization . Similar results were obtained for other micronutrients such as Zn and Mn. Thus, the role of autophagy is clearly not limited to Fe. Recently, several pathways have been implicated in the degradation of chloroplasts, which contain most of the cellular Fe, during leaf senescence . They include (i) the senescence-associated vacuoles that are important for the degradation of stromal proteins such as RuBisCo and glutamine synthase, (ii) chlorophagy in which whole chloroplasts are encapsulated in autophagosomes , (iii) a vesicular pathway depending on the ATG8 interacting protein that degrades envelope and thylakoid membranes in addition to stromal proteins  and (iv) CV-containing vesicles (CCVs) that contain the CV protein (chloroplast vesiculation) responsible for chloroplast destabilization . In contrast with the others, the CCV pathway does not require autophagy. It will be interesting to determine the respective importance of these pathways for Fe remobilization during vegetative organ senescence, as they could allow to target some nutrients more specifically than autophagy.
In addition to autophagy, the timing of senescence is also an important factor controlling remobilization. If senescence occurs too fast, the time window for remobilization is reduced and a lower amount of nutrients is allocated to seeds. Hence, NAC transcription factors, which control the onset of senescence, also play a major role in the control of nutrient remobilization from leaves to seeds [45–47]. More work will be required to understand how the interplay between autophagy and senescence controls Fe allocation to seeds in different species.
After it has been made available in vegetative tissue, Fe has to be transported to developing seeds through the phloem. Many lines of evidence have pointed to a key role of NA in this process. Complete loss of NA synthesis leads to sterility [48–50]. This has initially prevented a direct analysis of the role of NA in Fe transport to seeds, as the mutants did not set seeds. The first evidence that NA is important for Fe transport to the seeds came from the study of mutants impaired in YSL genes. The strong expression of OsYSL2 in phloem tissues of aerial parts of rice and developing seeds and the ability of this transporter to take up NA–Fe and NA–Mn complexes when expressed in Xenopus oocytes provided the first indication that NA is important for Fe transport through the phloem to the seeds . The analysis of rice lines in which OsYSL2 expression was silenced further supported this hypothesis . In these lines, Fe transport to the shoots and the seeds was strongly decreased. Conversely, increased OsYSL2 expression under a phloem-specific promoter led to a strong increase in seed endosperm Fe concentration . More recently, another YSL transporter from rice, OsYSL9, was implicated in Fe delivery to developing rice grains. OsYSL9 expression is upregulated by Fe deficiency in roots but downregulated in leaves under same conditions . The role of YSL in metal transport to seeds is not restricted to rice. Similar conclusions were obtained from the analysis of Arabidopsis knockout mutants targeting AtYSL1 gene or double mutants affecting AtYSL1 and AtYSL3 [54–56]. Like OsYSL2, AtYSL1 is expressed in phloem tissues of leaves and in developing siliques. Similar to OsYSL9, it is downregulated under Fe deficiency. Loss of function mutants in AtYSL1 accumulated higher levels of NA in shoots but much lower NA and Fe concentrations in mature seeds . Combining a mutation in AtYSL1 with a mutation in another Arabidopsis YSL, which is strongly expressed in phloem, AtYSL3, leads to even more severe phenotypes: the ysl1ysl3 mutants display symptoms of strong Fe deficiency with interveinal chlorosis as well as decreased fertility. Like ysl1, they also accumulate lower concentrations of Fe in their seeds when provided with a high Fe concentration in the medium, which alleviates the defect in fertility and upregulate YSL1 expression [55,56]. Further inflorescence grafting experiments showed that expression of AtYSL1 and AtYSL3 in the rosette was sufficient to restore fertility but not seed Fe content . Furthermore, the identification of a quadruple mutant combining knockout in three of the four genes encoding NAS in Arabidopsis and a knockdown in the fourth one allowed to obtain fertile plants with drastically decreased levels of NA . In this quadruple nas mutant, Fe and NA levels are lower in seeds, similar to the phenotype observed in ysl1. Interestingly, in NA biosynthesis mutants, Fe accumulates in phloem cells, indicating that NA is required to retrieve Fe from the phloem and provide it to sink tissues, such as seeds . This role may be shared between NA and mugineic acids in graminaceous plants. On the other hand, all YSL transporters characterized so far using expression in yeast or in Xenopus oocytes mediate the influx of NA–Fe or MA–Fe complexes into cells. The finding that NA and YSL are important for the phloem transport of Fe does not imply that Fe is transported as NA–Fe complex in the phloem.
Another important player in Fe transport from vegetative organs to the seeds is the oligopeptide transporter OligoPeptide Transporter 3 (OPT3). Full loss of function of OPT3 leads to embryo lethality, suggesting that the substrate of OPT3 is an essential cellular metabolite . Even though the closest homolog of OPT3, BjGT1, is a glutathione transporter , transport assays in yeast failed to show OPT3 ability to transport GSH . Instead, the expression of OPT3 could complement the yeast fet3fet4 mutant impaired in Fe uptake, suggesting that it transports Fe complexes. The exact substrate of OPT3 remains to be identified. The identification of a weak allele of opt3, opt3-2, allowed investigating its function at the reproductive stage . In opt3-2 mutants, Fe concentration is increased in vegetative organs but decreased in seeds, indicating a role in Fe transfer from vegetative organs to seeds. More recently, OPT3 was shown to be targeted to the plasma membrane and mainly expressed in phloem cells [60,62]. In opt3-2 mutants, Fe concentration was strongly increased in the xylem sap and decreased in the phloem sap, pointing to the role of OPT3 in Fe transfer from the xylem to the phloem . Combined with its expression in phloem cells, this suggests that OPT3 is involved in Fe loading into the phloem. Hence, both OPT3 and NA play critical roles in the long distance transport of Fe in the phloem to sink organs and especially seeds: OPT3 would be necessary for loading Fe into the phloem, while NA would be required for Fe retrieval from the phloem [57,60]. Interestingly, OPT3 is also important for Fe deficiency signaling, and opt3-2 upregulated Fe deficiency responsive gene expression even under Fe sufficient conditions [60,62]. In rice, OsOPT7, a homolog of OPT3, was characterized . OsOPT7 could not transport GSH or Fe when expressed in Xenopus oocytes and yeast. Like AtOPT3, OsOPT7 is induced under Fe deficiency, but its expression is more widespread than that of OPT3 [61,63]. Moreover, the phenotype of opt7 mutant does not resemble that of opt3-2. Whether graminaceous plant genomes carry orthologs of OPT3 remains to be determined.
Iron loading into seed tissues
There is no continuity between the vasculature of the mother plant and the provasculature of the embryo. Therefore, Fe, as other nutrients, has to be released from the phloem and re-absorbed by the embryo. In Arabidopsis and other dicotyledonous plants, nutrients are released at the level of the chalaza and the nucellus into the embryo sac fluid from which the embryo takes up nutrients (Figure 2). Accordingly, AtYSL1 expression is detected in the funiculus and the chalazal endosperm . Analyses of Fe speciation in the embryo sac fluid of developing pea seeds revealed that Fe is present as ferric Fe bound to citrate and malate in this extracellular compartment . In agreement with this finding, the citrate efflux transporter FRD3 is expressed in the peripheral cell layer of the embryo and the cell layer of the tegument facing the embryo sac during the seed development (Figure 2) . This suggests that FRD3 secretes citrate in the embryo sac to maintain Fe solubility and availability for uptake by the embryo. In dicotyledonous plants, Fe is taken up as ferrous Fe. This implies that Fe must be reduced before uptake by the embryo. However, genetic analyses in Arabidopsis failed to identify a membrane-bound ferric chelate reductase that is important for Fe acquisition by the embryo among FRO2 homologs. Instead, further analysis of the embryo sac fluid in pea showed the presence of a high concentration of ascorbate, which is sufficient to reduce Fe(III) to Fe(II) prior to its uptake by the embryo . Consistently, in Arabidopsis vtc (vitamin c) mutants deficient in ascorbate biosynthesis, the Fe content of the seed is decreased. The speciation of Fe in the embryo sac is thus similar to that encountered in other extracellular compartments of the plant, such as the xylem sap. In contrast, it is striking that ascorbate is used for Fe reduction, whereas membrane-bound ferric chelate reductases are used for uptake in roots and leaves as well as in intracellular organelles such as mitochondria and plastids [11,65,66]. The transporters responsible for secreting ascorbate and Fe in the embryo sac remain to be identified. In the case of Zn, the heavy metal pumping P-type ATPases, HMA2 and HMA4, release Zn from mother tissues for subsequent uptake by the embryo . In the case of Fe, the plasma membrane Fe efflux transporter IREG1/FPN1 may be involved in this process but so far no defect in Fe supply to the embryo has been reported for ireg1/fpn1 mutant .
Fe transfer from the maternal to embryo tissues.
In wheat and barley, the nutrients, including Fe, are provided to the grain by a single vascular strand along the ventral crease . Fe is mostly provided by the phloem and moves through several specialized cell layers: the crease vascular parenchyma, the pigment strand and the nucellar projection. Fe ends up in the transfer cells that are facing the embryo. From there, Fe has to be released in the extracellular space separating the mother plant from the embryo (Figure 3). The transfer cells, as the modified aleurone cells facing them on the side of the embryo, have highly invaginated plasma membranes favoring nutrient release and reabsorption . Fe accumulation and speciation in these structures show marked contrasts in the mature wheat grain. Fe is highly concentrated in the nucellar projection and co-localizes with sulfur. Fe accumulates to a lesser extend in the modified aleurone, from where it is probably distributed to other aleurone cells, the embryo and the endosperm . X-ray absorption spectra indicate that in the nucellar projection, Fe is mostly associated with NA, whereas in the modified aleurone, it is associated with phytate . In agreement with the speciation, barley homologs of AtVIT1, which drives Fe influx into the vacuole where phytate is localized, are strongly expressed in the aleurone [68,70–72]. In contrast, the genes encoding NAAT, NAS and YSL, that favor Fe mobility, are expressed at high levels in transfer cells [68,73] (Figure 3). In the future, it will be important to determine the specific expression pattern and the role of each of the genes involved in Fe transport in this complex structure. This should allow the identification of key transporters that release Fe to the extracellular space from the transfer cells as well as those responsible for taking it up into the embryo in the modified aleurone cells.
Fe distribution in a mature cereal grain (e.g. wheat grain).
Iron storage in seeds
The localization of Fe in seeds differs according to species and seed developmental stage. The localization and subcellular localization are strongly associated with Fe speciation, i.e., the nature of the ligand that binds Fe and determines it bioavailability . For example, Fe phytate complexes that are stored in vacuoles are notoriously poorly bioavailable . In contrast, ferritin Fe stored in plastids constitutes a highly bioavailable source of Fe .
Iron and other metals are not distributed evenly in seed tissues. In contrast, Fe distribution follows striking patterns. For example, in Arabidopsis mature embryo, Fe is highly concentrated around vascular tissues [71,77–81]. Interestingly, the pattern of Fe distribution is distinct from that of other metals: Mn is concentrated in the subepidermal cell layers of Arabidopsis cotyledons in mature seeds, while Zn is evenly distributed in Arabidopsis embryo [71,77,78,81]. The patterns of metal localization were initially discovered using synchrotron X-ray fluorescence imaging of intact seeds and were then confirmed and refined using additional approaches. The use of Perls staining intensified by diaminobenzidine (Perls DAB) allowed identification of the precise cell type accumulating Fe as the proto endodermis of the embryo . The use of microparticle-induced X-ray emission allowed quantification of the pattern: even though it is concentrated in proto-endodermal cells, Fe is present in other cells, albeit at much lower concentrations, and the Fe concentrated around provascular tissues accounts for 50% of the Fe content of the total seed . Finally, the use of electron-dispersive X-ray imaging coupled with transmission electron microscopy (TEM EDX) provided higher spatial resolution allowing the determination of the subcellular localization of Fe in the globoids of vacuoles of proto-endodermal cells (Figure 4) [79,82]. This subcellular localization is in agreement with the finding that mutations in AtVIT1 (vacuolar iron transporter 1) disrupt the pattern of Fe distribution [71,79]. In vit1 mutants, Fe is not concentrated around vascular tissues but instead co-localizes with Mn in subepidermal cells [71,79]. Nevertheless, Fe is still localized in vacuoles in vit1 mutants . Recently, the vacuolar Mn transporter AtMTP8 (metal tolerance protein 8) was shown to be responsible for concentrating Mn in subepidermal cells in wild-type seeds and Fe localization in these cells in the vit1 mutant background [77,78]. In the mtp8 knockout mutant seeds, Mn is concentrated around the vascular tissues together with Fe. In a double mutant combining vit1 and mtp8 mutations, Mn and Fe are evenly distributed in Arabidopsis embryo . Therefore, by creating strong sinks for Fe or Mn, vacuolar metal transporters determine the pattern of metal distribution in mature Arabidopsis embryo. The phenotype of vit1 mutant, which survives poorly when germinated under Fe-deficient conditions, indicates that adequate tissue localization of Fe in the embryo is crucial [71,79]. Little information is available on how the Fe distribution pattern is generated during embryo development. Consistent with a strong expression of VIT1 during seed development, Fe patterning is already apparent at the torpedo stage at the center of the cotyledons where the provascular tissue will differentiate [71,80]. However, in Brassica napus, high-resolution Perls DAB imaging detects Fe in the nuclei of all embryo cells at the torpedo stage . Further during development, at the bent cotyledon stage, Fe is detected in cytoplasmic vesicles and finally ends up in the vacuoles of the cells surrounding the vasculature in the mature embryo .
Fe distribution in
Arabidopsis thaliana mature embryo.
The observation of Fe in the nucleus at the torpedo stage in B. napus is consistent with the detection of high Fe concentrations in the nucleus of embryo cells in the developing pea (Pisum sativum) seeds . Therefore, Fe subcellular localization evolves during embryo development. These changes in localization are most likely paralleled by changes in speciation and therefore of the Fe bioavailability. The analysis of Fe bioavailability in pea seeds at different developmental stages shows that Fe bioavailability in immature peas is higher than in mature pea seeds .
Whether the pattern of Fe distribution in the vacuoles of cells surrounding the vasculature described in Arabidopsis is conserved among angiosperm has been investigated. In beans (Phaseolus vulgaris), Fe stores also concentrates around vascular tissues . Recently, this observation was extended to Brassicales and even Rosids, which represent a wide group including a third of the angiosperms [87,88]. However, in many cases, the number of cell layers that accumulate Fe is increased compared to Arabidopsis: Brassicales accumulate Fe in the endodermis and one cortical cell layer and other rosids in even more cortical cell layers [83,87,88]. While screening representatives of different groups of angiosperm, interesting exceptions were identified in Chenopodium quinoa and Carica papaya [87,88]. There are certainly more diverse patterns to be discovered.
In grains, the seeds of graminaceous species, the distribution of Fe is very different from that observed in Brassicales. In grains, in contrast with Arabidopsis, the embryo represents only a small volume of the seed, while the major part is constituted by the starchy endosperm. The endosperm is surrounded by the so-called aleurone cell layer. During germination, the aleurone layer is activated by gibberellins, releases enzymes that digest carbohydrates stored in the endosperm and undergoes programmed cell death (PCD) [89,90]. The embryo is equipped with an absorptive structure called the scutellum, somewhat equivalent to human placenta, which take up nutrients from the endosperm. In grains, a large part of the Fe is concentrated in the aleurone layer and another pool is found in the embryo [69,91,92] (Figure 3). In wheat (Triticum durum), the highest concentrations of Fe are found in the aleurone layer and in the scutellum . Separate measurements of wheat flour and bran indicated that almost 60% of grain Fe is in the bran, which includes the aleurone layer and therefore represents the major Fe pool in grains . Within the aleurone layer, Fe is concentrated together with phosphorus and other minerals in globoids inside the protein storage vacuole, similar to the subcellular localization observed in Arabidopsis . In situ X-ray absorption spectroscopy analyses have indicated that in the aleurone layer, most Fe is bound to phytate, either as Fe(II) or Fe(III) in good agreement with its subcellular localization [69,92]. Another pool is bound to citrate, which may be involved in the transport from modified aleurone cells present in the crease to aleurone cells . In rice, Fe accumulates in the aleurone layer together with phytate and in the scutellum, as observed in wheat [91,94]. The exact speciation of Fe in the embryo and in the endosperm has not yet been determined in these tissues that contain lower Fe concentration, probably because of limitations in the sensitivity of X-ray absorption spectroscopy. Based on co-occurrence of elements, it has been proposed that Fe is bound to phosphate (possibly as phytate) in the scutellum in wheat and to other ligands in the other parts of the embryo and in the endosperm [69,92]. These ligands could include proteins or smaller molecules such as NA, which is present in grains as in seeds of non-graminaceous species [55,95]. In wheat flour that corresponds to the starchy endosperm, NA is the main ligand for Fe .
Little is known on the mechanisms that control the pattern of Fe localization in grains. A developmental analysis showed that throughout rice grain development, Fe co-localizes with phosphorus in the aleurone cell layer, indicating that Fe is most likely bound to phytate as soon as it is stored . Mutations in OsVIT1, OsVIT2 and mitochondrial iron transporter have been shown to perturb Fe localization within the embryo [97,98]. OsYSL9 is strongly expressed in the scutellum and participates in Fe storage in the embryo . The phytosiderophore efflux transporter TOM2 is expressed in the dorsal vascular bundle, epithelium and the scutellum of the embryo and may also participate in the distribution of Fe . Silencing TOM2 gene expression did not alter total Fe content of the seed. The analysis of Fe distribution using sXRF, Perls staining or seed dissection could provide more insights in its function in developing rice grains.
The abundance, the localization and the speciation of Fe are equally important parameters to be taken into account in attempts to increase Fe content and bioavailability in grains, which are two of the major targets of biofortification projects. Having the major Fe pool stored as Fe phytate in the aleurone layer is probably the worst combination for human nutrition, as phytate has an extremely high affinity for metals, which drastically limits Fe absorption in the intestine . Moreover, the aleurone layer is often discarded during grain processing, such polishing for rice or preparation of white flour for wheat [93,96,99]. Hence, biofortification strategies using targeted molecular approaches have aimed at changing both the localization and the speciation of Fe in grains. Several reviews have been recently published on the attempts made to biofortify wheat and rice grains [99–101]. This review will, therefore, not extensively cover the topic of biofortification and rather just highlight a few examples. The initial attempt to biofortify rice grains through biotechnology used the gene encoding ferritin, which provides Fe under a highly bioavailable form, driven by an endosperm-specific promoter . Although the work provided a critical proof of concept, the increase in Fe content was modest and variable . Probably the most efficient strategy for biofortification by manipulating a single gene has been the activation or the overexpression of NA synthase in rice grains . This resulted not only in a strong increase in grain Fe concentration but also in Fe bioavailability, as NA appears to favor Fe intestinal absorption . More recently, the expression of a Fe vacuolar transporter TaVIT2 in wheat endosperm was also shown to be a relevant approach to increase Fe content and bioavailability . Probably, the expression of TaVIT2 in the endosperm vacuoles creates a strong Fe sink during development. Subsequently, Fe becomes available when endosperm cells die and lose their internal structure. Current efforts use pyramiding of several constructs to combine the benefits of multiple transgenes. This way, by combining the expression of transporters that limit vacuolar storage with elevated ferritin and NA synthesis, an increase in Fe concentration in grains by up to sixfold has been obtained .
A role for iron in the regulation of seed germination?
Seed germination starts with water uptake by the dry seed and ends with radicle protrusion and initiation of cell division . This is the critical phase of plant emergence because it is tightly regulated by water availability, temperature, oxygen and light conditions. Germination is also controlled by endogenous plant hormones, such as abscisic acid, gibberellins and ethylene, that play a major role in regulating early seed germination. These signals regulate germination through the process of dormancy, which is an endogenous block to the completion of the germination of a mature seed . Germination is followed by seedling growth during which mobilization of reserves and nutrient stored within the seed sustain active cell metabolism till the acquisition of autotrophy. Thus, stand establishment, the first critical component of crop yield, results from the successful and fast completion of both germination and early seedling growth.
Although the role of Fe in postgerminative events is quite well documented (see below), its possible involvement in the regulation of germination, i.e. the sum of molecular events that allow radicle protrusion, is rather unknown. The effect of Fe on seed germination has mostly been documented in the context of soil toxicity since high Fe concentrations are toxic and inhibit germination [107,108], but this is a common feature to many metals. An increase in phytate-degrading enzyme activity has been reported to occur during seed germination, with a concomitant decline in phytate [109–111], thus increasing Fe bioavailability . Müller et al.  showed that hydroxyl radicals were likely to play a role in cell wall loosening during radicle elongation and weakening of the endosperm of cress seeds. Although Fenton reaction is a natural candidate mechanism for explaining hydroxyl radical production, they did not provide any evidence about its possible involvement in vivo. Thus, whether Fe availability participates in the numerous biochemical and molecular processes involved in the regulation of radicle protrusion is actually not known.
Murgia and Morandini  demonstrated that, when seeds developed on the mother plant under Fe deficiency, they were more dormant at harvest, thus suggesting that seed Fe availability can regulate sensu stricto germination. It is highly likely that Fe could regulate germination by controlling ROS homeostasis through its involvement in the Fenton reaction. ROS indeed have been proposed to be key players in seed germination and dormancy [114–116]. Seed dormancy alleviation requires controlled ROS generation  but excessive levels of ROS trigger oxidative stress and prevent germination. This is occurring when seeds are germinated in unappropriated environmental conditions or when they have aged . With regards to the relationship between Fe and ROS metabolism, we can thus hypothesize that Fe homeostasis, as controlled by sequestration and release by vacuoles, is likely to participate in the ability of seeds to germinate and to play a role in the regulation of dormancy, by buffering ROS homeostasis. However, whether Fe speciation (and localization) is important for seed longevity and control of dormancy and germination remains to be addressed.
Iron remobilization after germination
Besides human nutrition, the main role of seed Fe stores is to sustain the early development of seedlings after germination. In dicots in which most of the Fe is probably stored as Fe phytate in vacuoles, retrieving Fe stores likely requires first to break down the phytate, then to transport Fe out of the vacuole and out of the cells. In Arabidopsis, Perls DAB staining allowed monitoring the fate of Fe stores after germination: the high Fe concentration around vascular tissues observed in dry seeds disappeared within 4 days, indicating that Fe was rapidly redistributed to the growing organs of the seedling [79,80]. In Arabidopsis, this process requires two redundant vacuolar transporters AtNRAMP3 and AtNRAMP4 (Natural Resistance Associated Macrophage Protein) . AtNRAMP3 and AtNRAMP4 are highly expressed after radicle protrusion. In the case of loss of AtNRAMP3 and AtNRAMP4 function mutants, seedling growth is arrested before the onset of photosynthesis . In the nramp3nramp4 double mutant, Fe remains blocked inside the endodermal vacuoles after germination [79,80,82]. The phenotype of nramp3nramp4 is partially suppressed by mutations in AtVIT1 by redirecting Fe storage in vacuoles of cortical cells . The phenotype of nramp3nramp4 is also rescued by Fe supplementation. However, even in the presence of Fe in the medium, the mutant activates Fe deficiency responses . Interestingly, chloroplast functions and especially Fe-requiring plastidial enzymes are repressed in the mutant, while the mitochondrial function is maintained . This suggests that Fe is prioritized to mitochondria or that these organelles rely on an independent pool of Fe. AtNRAMP3 and AtNRAMP4 encode divalent cation transporters. As Fe is most likely stored as Fe(III) in the vacuole, it means that Fe reduction is probably required prior to Fe export by AtNRAMP3 and AtNRAMP4. However, the Fe reduction system active in vacuoles of germinating cells remains to be identified. The phytase allowing Fe release prior to its reduction and export also remains to be identified. The importance of such activity is highlighted by the finding that expression of the bacterial phytase US417 or mutation in the phytate biosynthetic enzyme IPK1 (inositol phosphate kinase 1) in Arabidopsis significantly accelerates Fe remobilization . After its export from the vacuole, Fe needs to be exported from storage cells (the endodermis in Arabidopsis). However, the transporters involved in Fe efflux from cells during germination have not been identified. One candidate could be IREG1/FPN1, the plant homolog of the transporter involved in the release of Fe into the blood flow from intestine epithelial cells . Once it is exported from the cells, Fe needs to remain in a soluble form to diffuse to other cells. The citrate efflux transporter FRD3 is expressed in young seedlings and could be involved in the release of citrate to form soluble Fe complexes in the apoplast. This hypothesis is supported by the finding that frd3 mutants are chlorotic and exhibit slow root growth after germination . This phenotype can be rescued by supplementation of Fe or citrate in the cytosol.
In relation with their distinct organization, Fe remobilization mechanisms in seeds of graminaceous species are likely very different. In grains, most of the Fe is in the aleurone layer, outside of the embryo. During germination, under the positive control of gibberellins, aleurone cells secrete amylases that digest the starch in the endosperm and eventually undergo PCD . Membranes are destroyed in this process, and it is likely that the content of globoids including Fe bound to phytate is released in the endosperm. Iron efflux from vacuoles is probably not a necessary step for mobilization. Accordingly, no orthologs of AtNRAMP3 or AtNRAMP4 have been reported in graminaceous plants. In the endosperm, phytate is probably hydrolyzed by phytase and the released Fe bound to NA. As in dicotyledonous species, the precise phytase involved have not yet been identified. Then, the main question is how is Fe taken up from the endosperm into the embryo and further distributed to growing organs within the embryo. Time course analysis of gene expression and Fe localization during rice germination has highlighted several candidates [94,120]. X-ray fluorescence imaging revealed that after 24 h of germination, Fe accumulates in the epithelium and scutellum and the Fe level then decreases in the scutellum and increases in the coleoptile. Iron is already visible in the root tip after 36 h . Genes encoding NA and phytosiderophore biosynthesis were upregulated during early seedling development . Interestingly, NA synthase expression was prominent in the endosperm, while NA amino transferase (NAAT) was expressed in the embryo. Accordingly, several members of the YSL NA and phytosiderophore transporters, YSL2, 6, 10, 12 and 14, are upregulated . The expression of TOM2, a NA and phytosiderophore efflux transporter, and of OsYSL2 increased markedly in the scutellum during early seedling development [19,120]. These data indicate that Fe is transported as a complex with NA into the embryo or a complex with phytosiderophores within the embryo. Several transporters for free divalent Fe were also upregulated during germination: OsIRT1, several OsNramp genes and a Ferroportin/IREG [94,120]. As in dicotyledonous species, the citrate efflux transporter OsFRDL1 is upregulated, indicating that citrate is also secreted to maintain Fe solubility during cell-to-cell transport [34,35]. Finally, the rice homolog of the chloroplast Fe influx system in Arabidopsis PIC1 is upregulated after 3 days probably to sustain the differentiation of plastids into chloroplasts during early seedling development . Many transporters of Fe2+ or Fe complexes with NA or siderophores are apparently involved in Fe remobilization after grain germination. However, the individual contribution of each of these transporters remains to be analyzed to identify the key steps in Fe remobilization in graminaceous species.
With its uptake into chloroplasts of the developing seedling, Fe completes its journey from the senescing leaves of the mother plant to the expanding photosynthetic apparatus of its progeny, via the seed.
Conclusions and perspectives
Many transporters and ligands for Fe have been characterized, and this knowledge has been used to design targeted biofortification strategies. However, many key steps remain to be elucidated at the molecular level. The mechanisms that make Fe available for reallocation to the seeds during leaf senescence need to be explored in more detail. It would be important to determine which of the different pathways that contribute to chloroplast degradation are most important for Fe mobilization. Important questions also still need to be answered concerning phloem transport: the ligand for Fe in the phloem is still not unequivocally identified. Moreover, although OPT3 plays a key role in Fe transport to seeds, the specific substrate of this transporter is still unknown. The transfer of Fe from the mother plant to the embryo has been under investigated so far and probably deserves more attention (Figure 2). The pioneering discovery that ascorbic acid is responsible for Fe reduction in the extracellular space separating the mother tissues from the embryo raises important questions. How general this mechanism is remains to be determined. For example, the relevance of this finding for grain staples needs to be addressed. There is also a need to identify molecular players involved in ascorbate secretion and regeneration. Moreover, even though FRD3 has been proposed to be responsible for citrate efflux in the intracellular space separating the maternal tissues and the embryo, this hypothesis would need to be substantiated and other transporters potentially involved in citrate and malate efflux need to be identified. Last but not least, the transport mechanisms responsible for Fe efflux from mother tissues and reuptake into the embryo remain completely unknown. Tissue-specific transcriptomic analyses have provided candidates for mediating these steps in barley . The role of each candidate has now to be analyzed in detail, and similar analyses are lacking in Arabidopsis or other dicots. Manipulating these steps would certainly open new perspectives for biofortification. Besides the effectors of Fe transport and complexation during Fe loading into seeds, it would also be important to get insights into the master regulators that control Fe storage in seeds. How Fe storage is connected to the signaling pathways and transcription factors that regulate seed development, such as leafy cotyledon 2 or WRINCKLED, needs to be unraveled .
When considering Fe remobilization after germination, many questions are still open. In Arabidopsis and in dicots in general, several steps upstream and downstream of the action of vacuolar efflux by NRAMP3/4 remain to be investigated. The molecular identity of the phytases that make Fe available for transport has to be determined. Whether Fe needs to be reduced before transport by vacuolar NRAMP and by which mechanism are still open questions. Downstream of vacuolar efflux the plasma membrane transporters that allow Fe efflux from endodermal cells to other cell types, as well as the mechanisms that allow Fe distribution to plastids, are still to be identified. Similar questions arise concerning the use of Fe stored in aleurone cells after the germination of grains of monocots. Finally, the master regulators that orchestrate Fe remobilization in response to hormonal cues, such as gibberellins, also need to be defined.
In conclusion, even though many players involved in Fe storage and remobilization from seeds have been discovered in the last few decades, we still lack a molecular understanding of several key steps. In addition to improving our basic knowledge of how plants transfer nutrients from one generation to the next, unraveling these steps holds great promises for seed Fe biofortification and possibly for improving seed quality and stand establishment.
basic Helix Loop Helix
ferric reductase deficient 3
Iron REGulated 1
OligoPeptide Transporter 3
pleiotropic drug resistance 9
yellow stripe 1
The authors wish to thank Quentin Mari (@qntnmr) for his artistic contribution to the drawing of figure 3. The work on seed iron storage in S.T., S.M. and C.B. laboratories is supported by the CNRS, the INRA and Paris Sorbonne University, as well as by the collaborative ANR grant ISISTOR (ANR-16-CE20-0019-02).
The authors declare that there are no competing interests associated with the manuscript.