Introduction

In living organisms, a variety of enzymatic and electron transport processes employ iron due to its property as a transition metal to change in between different redox states. Iron deficiency leads to severe biochemical and developmental alterations and may cause death. In order to cope with their iron demand, all organisms have developed efficient mechanisms for acquiring iron from the environment. One mechanism is based on the reduction of ferric iron (Fe3+) to ferrous iron (Fe2+), which is then imported into the cell through the action of bivalent metal transporters. This mechanism, termed Strategy I, is well studied in dicotyledonous flowering plants and especially the model plant Arabidopsis (Arabidopsis thaliana). Recent gene expression, mutant analysis and functional complementation data suggest that Strategy I is employed throughout the green lineage including green algae, bryophytes, dicot and monocot flowering plants (Ishimaru et al. 2006; Lee and An 2009; Lo et al. 2016; Urzica et al. 2012).

Central to this reduction-based strategy is the function of the ZRT, IRT-like Protein (ZIP)-family transporters (Eide et al. 1996; Vert et al. 2002). They are predicted to have eight transmembrane domains and to translocate ions, such as Mn2+, Fe2+, Cd2+, Zn2+ and others, across membranes in the direction of the cytoplasm (Eide et al. 1996; Korshunova et al. 1999; Rogers et al. 2000). The founding member of this family is the Arabidopsis IRON-REGULATED TRANSPORTER1 (AtIRT1), whose main function is the import of rhizosphere iron (Eide et al. 1996; Henriques et al. 2002; Varotto et al. 2002; Vert et al. 2002). IRT proteins have been identified in other organisms, such as Chlamydomonas (Chlamydomonas reinhardtii), rice (Oryza sativa), tomato (Solanum lycopersicum), apple (Malus xiaojinensis), and others, based on the consistent upregulation of their genes in response to iron deprivation and sequence homology to AtIRT1 (Ishimaru et al. 2006; Li et al. 2006; Urzica et al. 2012) (summarized in Table 1). Studies on AtIRT1 have revealed three potential metal coordination sites in the structure of the protein. The loop between transmembrane domains II and III is one of these, predicted to be oriented towards the extracellular space. It was shown to coordinate Zn2+, and mutations in it change the transporter substrate specificity (Potocki et al. 2013; Rogers et al. 2000). Transmembrane domains IV and V contain histidine residues which, together with the polar residues in the vicinity may act as intramembrane metal binding sites (Eng et al. 1998). Indeed, mutating any of these amino acids abolishes all AtIRT1 transport activity (Rogers et al. 2000). In the cytoplasmically-exposed part of AtIRT1, the region between transmembrane domains III and IV, referred to as variable region, contains a histidine-rich sequence that is characteristic to all but a few ZIP family members across all kingdoms (Eng et al. 1998). The function of this region in metal transport is at present not clear. Arabidopsis AtIRT1 loss-of-function plants or fet3fet4 iron-deficient yeast can be fully complemented by an AtIRT1 protein with mutated histidines (Kerkeb et al. 2008), while another study using the fet3fet4 yeast only, reported that the MxIRT1 protein from Malus xiaojinensis is only fully functional with all histidines intact (Zhang et al. 2013).

Table 1 Known iron-regulated ZIP-family proteins from Chlamydomonas and plants
Full size table

AtIRT1 gene expression was shown to be suppressed by the presence of sufficient iron in the rhizosphere and its expression can be modulated in response to a great variety of stimuli (Blum et al. 2014; Brumbarova et al. 2015; Eide et al. 1996). Under iron limitation, AtIRT1 is strongly upregulated in a tissue- and cell-specific manner and this correlates with the enhanced AtIRT1 protein abundance and iron uptake (Blum et al. 2014; Eide et al. 1996; Marques-Bueno et al. 2016; Vert et al. 2002). At the same time, the protein is subjected to strict posttranslational regulation. It was found to localize at the plasma membrane, as well as in the early endosomes/trans-Golgi Network as a result of active processes of secretion and clathrin-mediated endocytosis (Barberon et al. 2011). The AtIRT1 localization at the plasma membrane of root epidermis cells depends on the availability of zinc, manganese and cobalt. In their absence, AtIRT1 localizes predominantly to the outer domain facing the rhizosphere (Barberon et al. 2014). Endocytosed AtIRT1 is retrieved from the degradation pathway and recycled to the plasma membrane. Two proteins, FYVE1/FREE1, a plant-specific ESCRT complex subunit (Gao et al. 2014), and SORTING NEXIN1 (SNX1), a stress-responsive trafficking regulator (Brumbarova and Ivanov 2016), have thus far been implicated in this recycling process (Barberon et al. 2014; Ivanov et al. 2014). The variable region of IRT1 contains two lysine residues that are targets for ubiquitination by at least one E3 ubiquitin ligase, IDF1 (Barberon et al. 2011; Kerkeb et al. 2008; Shin et al. 2013). The classical degradation pathway through the multivesicular body/prevacuolar compartment might not be the only way for degrading iron transporters. Studies in yeast grown under excessive iron showed that heterologously expressed MxIRT1 can be a target for autophagy (Li et al. 2015).

Additional iron-acquisition mechanisms are found in the green lineage. The green alga Chlamydomonas employs a reduction-oxidation-based mechanism that is similar to that of budding yeast (Saccharomyces cerevisiae). There, iron is reduced by a FRE-family ferric reductase, which is followed by oxidation by the multicopper oxidase FOX1 and import into the cell as Fe3+ by the high-affinity FTR1 transporter (Allen et al. 2007; Glaesener et al. 2013; Herbik et al. 2002). Plants of the Poaceae family employ a chelation-based strategy, also known as Strategy II, where potent iron chelators of the mugineic acid family are secreted from the root through the action of the TRANSPORTER OF MUGINEIC ACID FAMILY SIDEROPHORES1 (TOM1). The mugineic acid-iron complexes are then taken up by YELLOW STRIPE (YS)-family transporters (Kobayashi and Nishizawa 2012). Iron chelation may be important to sustain the Strategy I as well. Species-specific substances, such as flavins and phenolic compounds, secreted by dicotyledonous plants were found critical for the acquisition of iron (Fourcroy et al. 2014; Fourcroy et al. 2016; Jin et al. 2007; Rodriguez-Celma et al. 2011; Schmid et al. 2014). The role of these chelators is not entirely clear, however it was proposed that they maintain iron in an accessible form, might have weak iron reduction activity and may affect the growth of beneficial siderophore-producing rhizosphere microflora (Mladenka et al. 2010; Romheld and Marschner 1983; Siso-Terraza et al. 2016).

Since iron import is tightly controlled to balance iron requirements and iron toxicity, we assumed that different types of iron acquisition mechanisms should be controlled in a similar manner. To test this, we investigated gene regulation interaction between Strategy I and alternative iron-acquisition mechanisms and IRT1 protein structure analysis in Arabidopsis, rice and Chlamydomonas. Our data suggest that alternative strategies for iron acquisition in Chlamydomonas and plants are closely coordinated and protein modifications, such as phosphorylation, might play a role in the regulation of the iron import process.

Materials and methods

Phylogenetic analysis

Phylogenetic analysis was performed on the Phylogeny.fr platform (Dereeper et al. 2008). Sequences were aligned with MUSCLE (v3.7) with highest accuracy settings. Alignments of the large intracellular loop were visualized in Jalview (Waterhouse et al. 2009). The phylogenetic trees were constructed in the PhyML program (v3.0) applying the maximum likelihood method and default substitution model. The bootstrapping method with 100 replicates was used for evaluating the branch confidence. Values higher than 70% are indicated on the trees. Graphical representation and edition of the phylogenetic tree were performed with NJplot software.

Accession numbers of proteins/genes used in this study

(A list of accession numbers is also present in a table form in Supplementary table S1).

Arabidopsis thaliana (AGI number): AtIRT1 (At4g19690), AtIRT2 (At4g19680), AtIRT3 (At1g60960), AtZIP1 (At3g12750), AtZIP2 (At5g59520), AtZIP3 (At2g32270), AtZIP4 (At1g10970),AtZIP5 (At1g05300), AtZIP6 (At2g30080), AtZIP7 (At2g04032), AtZIP8 (At5g45105),AtZIP9 (At4g33020), AtZIP10 (At1g31260), AtZIP11 (At1g55910), AtZIP12 (At5g62160), AtZIP13 (At3g08650), AtZTP29 (At3g20870), AtIAR1 (At1g68100);

Chlamydomonas reinhardtii (Chlamydomonas reinhardtii v 5.5 gene identifier): CrZIP1 (Cre07.g355100), CrZIP2 (Cre13.g576050), CrZIP3 (Cre03.g189550), CrZIP4 (Cre09.g392060), CrZIP6 (Cre06.g299600), CrZIP7 (Cre06.g281900), CrZIP10 (CrIRT2, Cre12.g530350), CrZIP11 (CrIRT1, Cre12.g530400), CrZIP14 (Cre02.g087400), CrZRT1 (Cre07.g351950), CrZRT2 (Cre01.g000150), CrZRT3 (Cre13.g573950), CrZRT4 (ZIP9, Cre01.g066187), CrZRT5 (Cre07.g355150);

Cucumis sativus (GenBank identifier): CsIRT1 (AY590764);

Hordeum vulgare (GenBank identifier): HvIRT1 (ACD71460);

Malus xiaojinensis (GenBank identifier): MxIRT1 (AY193886);

Marchantia polymorpha (GenBank identifies): MpZIP3 (KJ146967), MpZIP5 (KJ146969);

Oryza sativa ssp. japonica (NCBI accession number, rice annotation project database http://rapdb.dna.affrc.go.jp gene identifier, MSU rice database http://rice.plantbiology.msu.edu/index.shtml gene identifier):

OsIAR1 (NP_001062003, Os08g0467400, LOC_Os08g36420), OsIRT1 (BAB85123, Os03g0667500, LOC_Os03g46470), OsIRT2 (BAD18964, Os03g0667300, LOC_Os03g46454), OsZIP1 (AAP59425, Os01g0972200, LOC_Os01g74110), OsZIP2 (AAP59426, Os03g0411800, LOC_Os03g29850), OsZIP3 (AY323915, Os04g0613000, LOC_Os04g52310), OsZIP4 (AAP85537, Os08g0207500, LOC_Os08g10630), OsZIP5 (BAD18965, Os05g0472700, LOC_Os05g39560), OsZIP6 (BAD18966, Os05g0164800, LOC_Os05g07210), OsZIP7 (BAD18968, Os05g0198400, LOC_Os05g10940), OsZIP8 (AAP88588, Os07g0232800, LOC_Os07g12890), OsZIP9 (Q0DHE3, Os05g0472400, LOC_Os05g39540), OsZIP10 (Q5Z653, Os06g0566300, LOC_Os06g37010), OsZIP10 (Q5Z653, Os06g0566300, LOC_Os06g37010), OsZIP11 (BAF08104, Os02g0196000, LOC_Os02g10230), OsZIP12 (BAT03391, Os08g0100200, LOC_Os08g01030), OsZIP13 (BAF17087, Os05g0316100, LOC_Os05g25194);

Physcomitrella patens (GenBank identifiers, Phytozome protein identifiers, Phytozome gene identifier/alias): PpZIP1 (EDQ80562, ZIP4, Phypa_67428, Phypa_67429, Phpat.006G078700/ Pp1s14_371V6), PpZIP2 (EDQ63212, Phypa_190331, Phpat.016G005000/ Pp1s144_110V6), PpZIP3 (EDQ79478, no annotation, Phpat.018G030700/ Pp1s19_181V6), PpZIP4 (EDQ62174, Phypa_139457, Phpat.003G110800/ Pp1s157_40V6), PpZIP5 (EDQ63192, Phypa_110147, Phpat.016G001900/Pp1s144_44V6 9), PpZIP6 (EDQ79796, Phypa_116705, Phpat.018G006600/ Pp1s17_293V6), PpZIP7 (EDQ57465, Phypa_145277, Phpat.022G074600/ Pp1s225_20V6), PpZIP8 (EDQ64964, Phypa_84433, no annotation, no annotation/Pp1s123_113V6), PpZIP9 (EDQ59386, Phypa_193317, Phpat.016G066100/Pp1s197_69V6), PpZIP10 (EDQ80497), PpIAR1 (EDQ51556, Phypa_98954, Phpat.012G086700/ Pp1s372_57V6);

Picea sitchensis (GenBank identifiers): PsZIP1 (ABK23592), PsZIP2 (ABK25966), PsZIP3 (ACN40483), PsZIP4 (ADE77858), PsZIP5 (ABK23604), PsZIP6 (ABK21613), PsZIP7 (ABK26545);

Selaginella moelendorfii (GenBank identifiers, gene identifier in Phytozome): SmZIP1 (EFJ29347, 91,749), SmZIP2 (EFJ11937, no annotation available), SmZIP3 (EFJ29045, 92,505), SmZIP4 (EFJ27536, 230,231), SmZIP5 (EFJ10861, 231,997), SmZIP6 (EFJ16604), SmZIP7 (EFJ11228), SmIAR1 (EFJ19982, 418,658);

Solanum lycopersicum (http://ensemblgenomes.org gene identifier): SlIRT1 (Solyc02g069200);

Zea mays (GenBank identifier): ZmIRT1 (ZmZIP10, NP_001152110).

Gene expression and coexpression data analysis

Gene expression data were obtained from the following databases: Oryza sativa - http://www.ricearray.org/expression/expression.php and Zheng et al. (2009); Arabidopsis thaliana – Genevestigator database (www.genevestigator.com).

The following gene coexpression datasets were used for the analysis: Chlamydomonas reinhardtii: Romero-Campero et al. (2013); Arabidopsis thaliana: ATTED-II version 8.0 (http://atted.jp/top_draw.shtml#NetworkDrawer) (Aoki et al. 2016); Oryza sativa: RiceFREND: http://ricefrend.dna.affrc.go.jp/.

Protein structure prediction

Transmembrane domains and the limits of the variable regions of ZIP transporters or the C-terminal cytoplasmic domain of NRAMP1 were calculated using the full-length protein sequences in the TMHMM Server v. 2.0 (http://www.cbs.dtu.dk/services/TMHMM/). Template-based modelling of protein structure was perfomed by the Modelling of the Protein Homology/analogY Recognition Engine V 2.0 (Phyre2, http://www.sbg.bio.ic.ac.uk/phyre2, (Kelley et al. 2015).

Prediction of amino acid modifications

Predictions of ubiquitination within the variable region were made on the UbPred server (http://www.ubpred.org/) (Radivojac et al. 2010). Phosphorylation was predicted on the NetPhos 3.1 server (http://www.cbs.dtu.dk/services/NetPhos/) (Blom et al. 1999).

Results

Phylogenetic analysis of plant and algal ZIP-family transporters

We first wanted to identify the phylogenetic relationship among the ZIP transporters in green algae and land plants. For this purpose, we collected the sequences of all proteins carrying the ZIP signature in the green alga Chlamydomonas, the bryophyte Physcomitrella (Physcomitrella patens), the lycophyte Selaginella (Selaginella moellendorffii), the gymnosperm Sitka spruce (Picea sitchensis), the monocot rice (Oryza sativa) and the eudicot Arabidopsis. The complete list (available in Materials and Methods and in Supplemental Fig. 1) was compiled from three independent collections, resulting from public database searches, literature searches and homology searches based on the sequences of the Arabidopsis ZIP proteins. The phylogenetic tree was generated with the help of the online Phylogeny.fr platform (Dereeper et al. 2008). We could distinguish six different subgroups of ZIP proteins (Fig. 1). One of them, containing the rice and Arabidopsis IRT1 proteins, included exclusively transporters from seed plants. PsZIP3, one of the seven identified ZIP proteins in Sitka spruce is also a part of this cluster. There was one group containing ZIP proteins from all the plant species used. Some algal proteins tended to cluster separately, an effect already observed previously and interpreted as an indication of the independent expansion of the algal and plant ZIP families (Hanikenne et al. 2005). The group of proteins homologous to the Arabidopsis IAA-ALANINE RESISTANT 1 (IAR1) formed a separate cluster. Two groups contained proteins from all investigated species. Among the more diverse one, we could find CrIRT1 and CrIRT2, which are the most likely candidates for Chlamydomonas Strategy I iron transporters (Urzica et al. 2012). These results suggest that the transporters for iron acquisition might have specialized separately in plants and algae and a major separation event might have occurred early in the evolution of seed plants.

Fig. 1
figure 1

Phylogenetic analysis of ZIP-family transporter amino acid sequences. The tree was generated in the Phylogeny.fr platform. The tree was rooted to the Arabidopsis AtNRAMP1 transporter. The size bar corresponds to one substitution per amino-acid position. Bootstrap values above 70% confidence are presented at the branches. ZIP/IRT sequences analyzed were from the green alga Chlamydomonas reinhardtii (Cr), the moss Physcomitrella patens (Pp), the vascular spore plant Selaginella moelendorfii, the seed plants Picea sitchensis (Ps), Arabidopsis thaliana (At), Oryza sativa (Os)

Full size image

Coexpression analysis of iron deficiency-induced ZIP genes

In order to understand the regulation of Strategy I iron acquisition and its interaction with alternative iron import mechanisms, we first concentrated on transcriptional regulation. We screened gene expression databases and primary literature in order to identify among the ZIP family members the known iron-regulated ZIP genes in plants and Chlamydomonas. Evidence for the upregulation of 15 plant and two Chlamydomonas ZIP genes could be obtained (Cohen et al. 1998; Donnini et al. 2010; Eckhardt et al. 2001; Eide et al. 1996; Ishimaru et al. 2006; Li et al. 2006; Li et al. 2013; Lo et al. 2016; Pedas et al. 2008; Vert et al. 2001; Wintz et al. 2003) (Table 1). We concentrated our analysis on three organisms, Chlamydomonas, rice and Arabidopsis, since they represent three cases of combining different iron acquisition mechanisms. For this, we analyzed the immediate coexpression environment of iron importer-encoding genes from Chlamydomonas (CrIRT1 and CrIRT2), Arabidopsis (AtIRT1 and AtIRT2) and rice (OsIRT1 and OsIRT2).

In Chlamydomonas, CrIRT1 was found to be a part of a large regulon (Fig. 2a). Included in it were many genes known to be involved in the maintenance of iron homeostasis. These included genes encoding NATURAL RESISTANCE-ASSOCIATED MACROPHAGE PROTEIN (NRAMP) and VACUOLAR IRON TRANSPORTER (VIT) family transporters, as well as a FERRITIN. Interestingly, the Arabidopsis NRAMP4 gene, a homolog of the here-identified Chlamydomonas NRAMP, is a member of the Arabidopsis root iron homeostasis coexpression network, however it shows a different type of Fe regulation than AtIRT1 (Ivanov et al. 2012). Interestingly, the CrIRT1 network included also the gene encoding the Fe3+ transporter CrFTR1 and CrFRE1 that encodes the ferric reductase common for the two Chlamydomonas iron acquisition strategies, involving FOX-FTR or IRT1. Another indicative gene in the CrIRT1 network is CrFEA2, which encodes an extracellular protein facilitating iron uptake (Allen et al. 2007). FEA proteins are algae-specific and might also be common for the two strategies. CrFEA1 was shown to complement the iron uptake-defective irt1 Arabidopsis mutant (Narayanan et al. 2011). Despite the absence in this network of the multicopper oxidase FOX1, also prominently upregulated under iron deficiency (Urzica et al. 2012), the data suggests that the two iron uptake strategies are tightly coregulated in Chlamydomonas.

Fig. 2
figure 2

Coexpression analysis of IRT-encoding genes from Chlamydomonas (a), rice (b) and Arabidopsis (c). Yellow shapes represent the input genes, green shapes represent genes coexpressed with the input genes. Coexpression data was visualized in the Cytoscape software (version 3.2.1)

Full size image

The much smaller CrIRT2 network contains only three genes. One of these, however, encoding a kelch-repeat protein is of high interest. While the function of the protein is not known, the homologous gene, At3g07720, is a stable iron-deficiency marker at gene and protein level in Arabidopsis, irrespective of growth conditions and developmental stage (Ivanov et al. 2012; Mai and Bauer 2016; Mai et al. 2015).

In rice, the coexpression networks of OsIRT1 and OsIRT2 contained a similar number of genes (Fig. 2b). The OsIRT1 network contained no obvious iron deficiency-related genes. Even though an oxygenase, At3g12900, is a prominent member of the Arabidopsis root iron deficiency coexpression network (Ivanov et al. 2012), BLAST searches revealed that the gene Os03g0856000 coexpressed with OsIRT1, is not its closest homolog. The OsIRT2 network contained genes encoding OPT and NRAMP-family transporters, which are also different from the ones present in the Arabidopsis root network. It needs to be noted however that the Arabidopsis NRAMP1, the closest homolog to the product of the Os07g0258400 gene, can function as an iron transporter and partially rescue the irt1 mutant (Cailliatte et al. 2010). Intriguingly, OsIRT2 was coexpressed with OsIRO2 gene. The OsIRO2 protein is a bHLH transcription factor that regulates, either directly or through downstream transcription factors, the expression of key Strategy II genes, including TOM1 and genes required for the synthesis of the mugineic acid-family iron chelators. OsIRO2 is however not involved in the transcriptional regulation of OsIRT1 (Ogo et al. 2007). This coexpression data suggests that in rice Strategy I and Strategy II show some degree of transcriptional coregulation and may be responding to the activity of common master regulators.

The Arabidopsis AtIRT1 gene showed coregulation with putative iron regulation genes (Fig. 2c). This includes At1g74770, encoding a haemerythrin domain-containing Zn finger protein that is homologous to the E3-ubiquitin ligase BRUTUS (BTS), a regulator of iron uptake in rice and Arabidopsis (Kobayashi et al. 2013; Long et al. 2010). AtIRT2 is one of the stable iron deficiency markers (Ivanov et al. 2012). Its immediate coexpression neighbors include genes like RHS18, predicted to encode a heme-containing peroxidase, and two genes encoding cell wall organization-related enzymes. These genes are all predominantly expressed in the root hair cells (Bruex et al. 2012), consistent with the suggested function of these cells in iron acquisition (Jakoby et al. 2004; Marques-Bueno et al. 2016; Vert et al. 2002). These data show the interaction of Arabidopsis iron acquisition with other processes, such as root development under stress. Also, they emphasize the interconnection of iron acquisition and further iron redistribution and homeostasis, which is a point common to all three analyzed organisms.

Structure of the cytoplasmically-exposed variable region of AtIRT1

As the variable region between transmembrane domains III and IV has a great importance for the regulation of AtIRT1 protein stability, we analyzed its predicted structure (Fig. 3). The loop region contains three helices, two of them bordering the transmembrane domains, and two disordered regions (Fig. 3a). The histidine-rich sequence is a part of the large disordered region (DR2, Fig. 3a) as is lysine K179, which is exposed towards the cytoplasm. K154 on the other hand is within a helix and is close to the membrane. A prediction of phosphorylated amino acids on the NetPhos 3.1 server revealed that both lysines K154 and K179 in the variable region of AIRT1 neighbor potentially phosphorylatable amino acids (Figs. 3 and 4). Therefore phosphorylation may play a role in the regulation of IRT1 together with the known ubiquitination events.

Fig. 3
figure 3

Predicted structure of the variable region of AtIRT1. a Sequence of the variable region. The predicted secondary structure features are indicated. The amino acids noted in the structure in b are underlined and color-coded (red – ubiquitination target; blue – potential phosphorylation target). DR stands for disordered region. b and c Predicted 3D structure of the AtIRT1 protein. The image in c represents the same structure as in b after a 90° clockwise rotation around the vertical axis

Full size image
Fig. 4
figure 4

Analysis of the variable regions of the iron-regulated ZIP proteins form Chlamydomonas, rice and Arabidopsis. a Phylogenetic analysis. The tree was generated in the Phylogeny.fr platform and rooted to the C-terminal cytoplasmic region of the Arabidopsis AtNRAMP1 transporter. The size bar corresponds to 0.2 substitutions per amino acid position. Bootstrap values above 70% confidence are presented above the branches. b Sequence alignment of variable regions. Amino acids predicted to be targets for modification are indicated in color (red – potential ubiquitination target; blue – potential phosphorylation target). The two known ubiquitination targets in AtIRT1 are underlined

Full size image

Potential regulatory sites in the variable region of IRT proteins

We aimed to understand if there is a conservation of potential modification-prone signatures among IRT transporters. For this we analyzed the phylogenetic relationships between the variable regions in the six IRT proteins, together with the existence of amino acids that are potential targets for covalent modifications.

Phylogenetic analysis of the variable regions was performed as described above for the full-length proteins. The C-terminal cytoplasmic region of AtNRAMP1 was used to root the tree. Surprisingly, in the resulting tree (seen on Fig. 4a) the sequences were differently arranged, compared to the tree of the full proteins in Fig. 1. The Arabidopsis AtIRT1 and AtIRT2 regions clustered together, as expected. However, unlike them each of the Chlamydomonas and rice sequences formed distinct branches. Still, the rice regions grouped closer to the Arabidopsis ones, compared to the two Chlamydomonas sequences.

We then scanned the variable regions for amino acids that represent potential targets for covalent modifications. We used the UbPred and NetPhos 3.1 server applications to predict potential ubiquitination and phosphorylation sites, respectively (Fig. 4b). The two known ubiquitinated lysine residues in AtIRT1, K154 and K179 (Barberon et al. 2011; Kerkeb et al. 2008), were also identified by the prediction. With the exception of CrIRT1, the relative position of the first lysine is conserved in the plant IRT proteins, however it was not predicted as a modification target in all cases. The second lysine was less conserved. In OsIRT2, for example, it can be found at a different position relative to the transmembrane domain, and in AtIRT2 and OsIRT1 it has no phosphorylatable amino acids in the vicinity. The variable regions of the two Chlamydomonas IRT proteins are altogether very different from the four plant ones. They are larger, with multiple predicted modification sites. Also, in CrIRT1 and CrIRT2 the histidine-rich regions, potentially involved in metal coordination, are rather obscure, if present at all. The observed differences might reflect organism-specific strategies for post-translational regulation. Alternatively, they may suggest different functions of the transporters in Strategy I of iron uptake. An example of this is AtIRT2 which is not involved in the actual import of rhizosphere iron but probably prevents the overaccumulation of iron resulting from the AtIRT1 activity (Vert et al. 2009). However interesting, we find this data insufficient to draw conclusions on the relationship in the regulation of Strategy I and species-specific parallel iron acquisition strategies.

Discussion

Here, we investigated IRT and ZIP divalent metal transporter sequences from a green alga, spore plants and seed plants and coexpression networks to deduce potential regulatory specificities evolved in the green lineage. Phylogenetic analysis showed that the rice and Arabidopsis IRT transporters group together with only ZIP proteins from seed plants. The inclusion of the PsZIP3 protein in the group suggests that it might be the spruce iron transporter. Chlamydomonas IRT proteins were part of a separate group, which reflects the evolutionary divergence between plants and algae and that the different environment might have required different adaptations for iron acquisition. In the green alga Chlamydomonas and in the land plant rice, the IRT genes were found to be coregulated with components of the alternative available iron uptake systems, FOX1-FTR1 in the alga and Strategy II in the grass, respectively. Using the Arabidopsis AtIRT1 as a model, we identified potential sites for post-translational protein modifications within the variable region between transmembrane domains III and IV.

We found that in cases of coexistence of reduction-based iron uptake with another iron acquisition system, the genes involved in the different iron uptake strategies exhibit coregulation at transcriptional level. In the case of the green alga Chlamydomonas, the regulation was very tight, which also reflects the fact that the two strategies partially employ the same physical effectors, such as the ferric reductase (Allen et al. 2007; Glaesener et al. 2013; Herbik et al. 2002). In addition, the Chlamydomonas CrIRT1 was found coregulated with a large number of other genes using the above-described method, a total of 24. This is a case unlike any of the other here-investigated IRT genes. In rice, the interconnection between the reduction- and chelation-based strategies is less well pronounced. This can however also reflect the complexity of the organisms. In Chlamydomonas, all processes of acquisition, redistribution and utilization occur in the same cell and the regulation is tightly coordinated on subcellular level. On the other side, in the roots of multicellular organisms, there is a high level of compartmentalization and the homeostasis depends on the interaction of different cell types and tissues within the organ (Blum et al. 2014; Jakoby et al. 2004; Marques-Bueno et al. 2016; Seguela et al. 2008; Vert et al. 2002).

Interestingly, CrIRT2, which is itself iron-regulated (Urzica et al. 2012), clusters separately from CrIRT1 and it is in a cluster with only three other genes. The lack of tight coexpression between two closely related IRT genes was a tendency found in each of the three organisms studied. In Arabidopsis, it is known that AtIRT1 is the essential component for iron acquisition, while AtIRT2 cooperates with the uptake system within intracellular compartments (Henriques et al. 2002; Varotto et al. 2002; Vert et al. 2001; Vert et al. 2002). This situation might be true for other organisms as well and therefore be reflected in the coexpression behavior of the two genes.

Ubiquitination of membrane proteins has emerged as a way of regulating their intracellular targeting and stability. In Arabidopsis, plasma membrane-localized receptors such as FLS2 and BRI1, as well as transporters, such as PIN2 and BOR1 are regulated through this post-translational modification (Gohre et al. 2008; Leitner et al. 2012; Martins et al. 2015; Takano et al. 2010). Ubiquitination of ZIP transporters was first described in yeast (Gitan and Eide 2000). The type of ubiquitination observed in all these cases differs substantially. BRI1 receptor and the auxin transporter PIN2 are modified through the addition of K63-linked ubiquitin chains (Martins et al. 2015; Yin et al. 2007), while BOR1 was shown to be mono or diubiquitinated (Kasai et al. 2011; Takano et al. 2010). In case of the Arabidopsis AtIRT1, data showed that lysine residues are modified by the addition of single ubiquitin units (monoubiquitination) at multiple positions (Barberon et al. 2011). Based on a predominant AtIRT1 form whose mobility suggests a molecular weight of approximately 70 kDa, Shin et al. (2013) argue that the majority of AtIRT1 is tetraubiquitinated. Currently only two lysine residues, K154 and K179, both lying within the variable region of AtIRT1, have been shown to undergo ubiquitination (Barberon et al. 2011; Kerkeb et al. 2008). Therefore, additional lysine residues outside the variable region may be targets for this modification. The predicted structure of AtIRT1 contains four such residues, at positions 253, 254, 318 and 324, however the role of these is yet to be determined. The variable regions of all investigated IRT proteins contain lysine residues. Within Arabidopsis and rice, the conservation of the two lysines found originally in AtIRT1 suggests that these might be ubiquitination targets in AtIRT2 and the rice proteins as well.

An interesting feature of the AtIRT1 variable region is the existence of potential phosphorylation sites in the vicinity of the two lysines. Phosphorylation is known to affect the localization and polarity of PIN-family transporters (Friml et al. 2004; Michniewicz et al. 2007; Ding et al. 2011; Rakusova et al. 2011). The effect of such modifications have not been experimentally addressed for AtIRT1, however they suggest new possible regulatory mechanisms. For example the ubiquitination of the FLS2 receptor is dependent on its activation, and therefore phosphorylation (Gohre et al. 2008). The possibility exists therefore, that the ubiquitination events on AtIRT1 are dependent on prior signals encoded in the variable region in the form of amino acid phosphorylation.

In conclusion, at transcriptional level, the regulation of IRT transporters is closely connected to alternative existing iron acquisition strategies. In terms of post-transcriptional regulation, the structure and predicted covalent modifications within the variable region of IRT proteins suggests that more than one modification is involved in regulating the activity, localization and stability of the Fe (II) transporters.