This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 19/3/986    most recent tpc.106.047407v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Duy, D. Articles by Philippar, K. Search for Related Content PubMed PubMed Citation Articles by Duy, D. Articles by Philippar, K. Agricola Articles by Duy, D. Articles by Philippar, K.

PIC1, an Ancient Permease in Arabidopsis Chloroplasts, Mediates Iron Transport^[W]

^a Department für Biologie 1, Botanik, Ludwig-Maximilians-Universität München, D-80638 Munich, Germany
^b Molecular Plant Nutrition, Institute for Plant Nutrition, University of Hohenheim, D-70599 Stuttgart, Germany

¹ To whom correspondence should be addressed. E-mail philippar{at}lrz.uni-muenchen.de; fax 49-89-17861-185.

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
In chloroplasts, the transition metals iron and copper play an essential role in photosynthetic electron transport and act as cofactors for superoxide dismutases. Iron is essential for chlorophyll biosynthesis, and ferritin clusters in plastids store iron during germination, development, and iron stress. Thus, plastidic homeostasis of transition metals, in particular of iron, is crucial for chloroplast as well as plant development. However, very little is known about iron uptake by chloroplasts. Arabidopsis thaliana PERMEASE IN CHLOROPLASTS1 (PIC1), identified in a screen for metal transporters in plastids, contains four predicted -helices, is targeted to the inner envelope, and displays homology with cyanobacterial permease-like proteins. Knockout mutants of PIC1 grew only heterotrophically and were characterized by a chlorotic and dwarfish phenotype reminiscent of iron-deficient plants. Ultrastructural analysis of plastids revealed severely impaired chloroplast development and a striking increase in ferritin clusters. Besides upregulation of ferritin, pic1 mutants showed differential regulation of genes and proteins related to iron stress or transport, photosynthesis, and Fe-S cluster biogenesis. Furthermore, PIC1 and its cyanobacterial homolog mediated iron accumulation in an iron uptake–defective yeast mutant. These observations suggest that PIC1 functions in iron transport across the inner envelope of chloroplasts and hence in cellular metal homeostasis.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Some transition metals, and in particular iron, are essential micronutrients in plants. Thus, to control metal homeostasis, plants have developed specified strategies for metal ion acquisition, distribution to organs and tissues, and subcellular compartmentalization (for overview, see Hall and Williams, 2003; Curie and Briat, 2003; Colangelo and Guerinot, 2006). Dicotyledonous plants such as Arabidopsis thaliana take up ferrous iron [Fe(II)] after reduction of Fe(III) chelates from the soil. This first step is accomplished by the action of the plasmalemma root ferric chelate reductase FERRIC REDUCTASE/OXIDASE2 (Robinson et al., 1999) and the major root metal transporter IRON-REGULATED TRANSPORTER1 (IRT1) (Eide et al., 1996; Henriques et al., 2002; Varotto et al., 2002; Vert et al., 2002), which mediates Fe²⁺ uptake into root epidermis cells. Distribution of iron in the plant is achieved by long-distance transport of Fe chelates in the vasculature. A strong chelator of iron is the aminocarboxylate nicotianamine, and members of the YELLOW STRIPE1-LIKE (YSL) transporter family in Arabidopsis are likely candidates that contribute to iron distribution by loading and unloading Fe-nicotianamine from the vascular tissue (Le Jean et al., 2005; Waters et al., 2006). Within the plant cell, iron has to be compartmentalized into different organelles, such as chloroplasts, mitochondria, and vacuoles. However, to date, only two members of the NRAMP (for natural resistance-associated macrophage protein) family of metal transporters, NRAMP3 and NRAMP4, have been shown to play a role in Fe mobilization from the vacuole during seedling development (Thomine et al., 2003; Lanquar et al., 2005). The iron transport pathway across the envelopes of chloroplasts and mitochondria remains unknown, although chloroplasts in particular represent a major sink for metal ions (see below).

Chloroplasts are organelles enclosed by an outer and an inner envelope membrane and have evolved from the endosymbiosis of free-living cyanobacteria with an ancient eukaryotic cell (for review, see Vothknecht and Soll, 2005). Because chloroplasts are the site of photosynthesis, they provide the basis for life on earth in its present form. However, chloroplasts represent only one type of the plastid organelle family in higher plants (for overview, see Möller, 2005). Proplastids in meristematic tissue and etioplasts in dark-grown plantlets develop into the mature, autotrophic chloroplast of the green leaf. By contrast, storage plastids are heterotrophic organelles that convert photosynthates derived from source tissues into storage compounds. Thus, in addition to photosynthesis, plastids harbor many more vital biosynthetic functions, such as nitrogen and sulfur assimilation or the biosynthesis of fatty acids and aromatic amino acids. In consequence, these functions require an active solute exchange across the outer and inner envelope membranes surrounding the chloroplast stroma. Metal transport proteins in both membrane systems thus provide a bottleneck to the control of metal homeostasis in the chloroplast as well as in the plant cell.

Because of their potential for valency changes, the transition metals Fe, Cu, and Mn play a vital role in photosynthetic electron transport in chloroplasts (Raven et al., 1999). Whereas the photosynthetic apparatus represents one of the most iron-enriched systems in the plant cell (photosystem II, photosystem I, cytochrome b₆-f complex, and ferredoxin), copper ions catalyze electron transfer via plastocyanin and a cluster of Mn atoms is required as the catalytic center in the oxygen-evolving complex. Furthermore, stroma-localized Fe and Cu/Zn superoxide dismutases scavenge reactive oxygen species in the water–water cycle (Kliebenstein et al., 1998; Asada, 1999). In addition, Zn is known to function as a cofactor (RNA polymerase, zinc finger domains) in plastid transcription. During germination, development, and iron stress, ferritin clusters in plastids serve as iron stores (Briat et al., 1999; Connolly and Guerinot, 2002). Furthermore, Fe-S cluster biogenesis in chloroplasts requires the import of iron. Fe-S cluster proteins are essential components of the photosynthetic electron transport chain and are involved in protein import, chlorophyll biosynthesis, and breakdown as well as in nitrogen and sulfur assimilation (for overview of Fe-S biogenesis, see Balk and Lobreaux, 2005; Ye et al., 2006).

Despite these essential functions for metal ions in chloroplasts, very little is known about metal transport proteins in plastid envelopes. To date, the only chloroplast proteins demonstrated to be involved in metal ion transport are the copper-transporting P-type, heavy-metal ATPases PAA1, PAA2, and HMA1 as well as the magnesium transport protein MRS2-11 (Shikanai et al., 2003; Abdel-Ghany et al., 2005; Drummond et al., 2006; Seigneurin-Berny et al., 2006). Whereas PAA1, HMA1, and MRS2-11 have been reported to be localized in the inner chloroplast envelope, PAA2 transports Cu across the thylakoid membrane. Direct measurements of iron transport on isolated pea (Pisum sativum) chloroplasts have shown that iron is transported in the ferrous form across the inner envelope (Shingles et al., 2001, 2002). Fe²⁺ uptake into chloroplasts is most likely energized by a proton gradient and can be inhibited by Zn²⁺, Cu²⁺, or Mn²⁺. The putative metal uptake protein might thus mediate Fe²⁺/H⁺-uniport and be able to transport Zn²⁺, Cu²⁺, and Mn²⁺ as well.

In this study, we characterized PERMEASE IN CHLOROPLASTS1 (PIC1), a transmembrane protein in the inner envelope of Arabidopsis chloroplasts. PIC1 is of cyanobacterial origin and most likely functions in iron permease in the chloroplast envelope. The detailed characterization of pic1 knockout mutants reveals a phenotype reminiscent of iron-deficient chloroplasts. This includes chlorosis, an altered organization of leaf mesophyll cells, and severe defects in chloroplast and thylakoid development. These findings are supported by differential gene expression in pic1 mutants (e.g., downregulation of genes involved in photosynthesis and Fe-S cluster biogenesis). By contrast, the accumulation of ferritin clusters in pic1 mutant plastids as well as the upregulation of stress-related genes indicates an iron-overload reaction in the cytosol and impaired metal homeostasis at the cellular level. Because PIC1 and its cyanobacterial homolog sll1656 were able to complement the growth of an iron uptake–defective yeast mutant and to confer iron uptake, we conclude that the protein is involved in iron transport and homeostasis in chloroplasts.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
A Putative Permease of Cyanobacterial Origin in Chloroplast Envelopes
In our search for potential metal ion transporters in plastid envelopes, we screened the Arabidopsis chloroplast proteome for polypeptides that are hydrophobic, have a basic isoelectric point, contain potential transmembrane domains, and show significant homology with cyanobacterial proteins. The protein At2g15290 was identified in a mixed envelope preparation of Arabidopsis chloroplasts (Froehlich et al., 2003) and according to in silico analyses was classified into the virtual hydrophobic proteome from plastid envelope membranes (Rolland et al., 2003). Furthermore, a screening of publicly available databases revealed outstanding similarities to cyanobacterial proteins (Figure 1A ). To date, the function of all proteins homologous with At2g15290 is annotated as unknown; however, some cyanobacterial proteins were grouped into prokaryotic clusters of orthologous groups of proteins (COGs) (Tatusov et al., 2003). Among these COGs were high-affinity Fe²⁺/Pb²⁺ permeases, permease components of major facilitator proteins and ABC-type (for ATP binding cassette) transport systems or amino acid transporters, indicating that At2g15290 might function as a permease in solute transport across the plastid envelope membranes. Thus, in this report, we refer to At2g15290 as PIC1 (for PERMEASE IN CHLOROPLASTS1).

View larger version (63K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 1. PIC1 Is a Permease of Cyanobacterial Origin with Four -Helical Transmembrane Domains. (A) Phylogenetic tree of the PIC family in eukaryotes and cyanobacteria. In plants, At-PIC1 groups into a family of homologous proteins from monocots such as rice (Os-PIC1 and Os-PIC2) and maize (Zm-PIC1 and Zm-PIC2). The closest relative to At-PIC1 is found in the dicotyledonous plant Lotus japonicus (Lj-PIC1), whereas the homolog from the moss Physcomitrella patens (Pp-PIC1) is more distantly related. In the genomes of the green alga Chlamydomonas reinhardtii as well as the red alga Cyanidioschyzon merolae, we could identify homologs to At-PIC1. Cyanobacterial relatives to At-PIC1 from Synechocystis sp PCC 6803 (sll1656), Anabena sp PCC 7120 (all4113), Thermosynechococcus elongatus BP-1 (tll0396), and Prochlorococcus marinus strain MIT 9313 (PMT0365) are depicted. All proteins were identified by BLAST screening of At-PIC1 against the respective databases (see Methods). The distance tree is based on 148 amino acids of the predicted mature proteins (see Methods and Supplemental Figure 1 online). Bootstrap values for the respective branches are indicated. (B) Predicted topology for PIC1. According to the AM consensus prediction (ARAMEMNON database; Schwacke et al., 2003), four -helical transmembrane domains are proposed for At-PIC1 as well as for the homologs Os-PIC1, Os-PIC2, sll1656, and all4113. CT, C terminus; NT, N terminus. (C) Sequence alignment of At-PIC1 with its closest relatives from plants (Lj-PIC1; 74% amino acid identity of mature proteins) and cyanobacteria (sll1656 and all4113; 24 and 23% amino acid identity of mature proteins, respectively). Amino acids identical in all four proteins are shown as black-boxed letters, and residues conserved in two or three sequences are shown as gray-boxed letters. The predicted four transmembrane domains of At-PIC1 are boxed, and the potential cleavage site of the chloroplast transit peptide is indicated by a black arrowhead.

The amino acid sequence of the mature PIC1, to our surprise, showed significant homology only with proteins of cyanobacterial origin. The protein with the highest similarity (24% amino acid identity) is encoded by the gene sll1656 in Synechocystis sp PCC 6803 (Figure 1C). The sole eukaryotic relatives except in vascular plants were found in the genomes of the green alga Chlamydomonas reinhardtii and the red alga Cyanidioschyzon merolae (Figure 1A). Thus, PIC1 seems to originate directly from the endosymbiosis of a cyanobacterium with a eukaryotic cell, which led to the formation of chloroplasts (Vothknecht and Soll, 2005). Except in the ancient cyanobacterium Gloeobacter violaceus, at least one homolog was present in all sequenced cyanobacterial genomes. Interestingly, Gloeobacter is the only genus lacking a thylakoid membrane system (Nakamura et al., 2003).

PIC1 Is Localized in the Inner Envelope of Arabidopsis Chloroplasts
To verify the predicted chloroplast localization of PIC1, we performed in vitro import experiments into isolated Arabidopsis chloroplasts. The precursor protein PIC1 was imported into isolated intact chloroplasts of 8-d-old seedlings and processed to a mature protein of 22 kD (Figure 2A ). The protein was resistant to protease treatment, confirming import into chloroplasts. In isolated pea chloroplasts, At-PIC1 displayed the same import pattern (data not shown). All import experiments showed an intermediate product of 23.5 kD (Figure 2A) that can be pulse-chased into the inner envelope membrane (E. Firlej-Kwoka, personal communication). Thus, we conclude that PIC1 uses the general import pathway into chloroplasts but that insertion into the inner envelope membrane occurs via an intermediate. Based on ChloroP (Emanuelsson et al., 1999), the precursor protein PIC1 of 31.4 kD (296 amino acids) is predicted to have a chloroplast transit peptide with a processing site after amino acid residue 15. However, in vitro import experiments and immunoblot analysis showed a mature protein of 22 kD (Figures 2A and 2B), requiring the cleavage of a far larger transit peptide. Furthermore, the alignment with all cyanobacterial homologs and the predicted mature proteins of the plant subfamily support this observation (Figure 1A; see Supplemental Figure 1 online). Thus, we conclude that the precursor PIC1 is processed after amino acid residue 81 (Figure 1C), a cleavage site that is supported by ChloroP analysis after removal of the first 15 residues. The resulting mature-sized protein is 216 amino acids long and has a molecular mass of 22.8 kD.

View larger version (70K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 2. Subcellular Localization of PIC1 to the Inner Envelope of Chloroplasts. (A) In vitro import of PIC1 into chloroplasts isolated from cotyledons of 8-d-old Col-0 seedlings. Purified intact chloroplasts equivalent to 10 µg of chlorophyll were incubated for 15 min at 25°C with the in vitro–translated 35S-labeled PIC1 precursor protein (TL). After the import reaction, chloroplasts were either not treated (–) or treated (+) with the protease thermolysin for 30 min at 4°C. Intact chloroplasts were recovered by centrifugation, and import products were analyzed by SDS-PAGE and autoradiography. The precursor protein PIC1 is processed to a mature protein of 22 kD (arrowheads). The mature protein as well as an intermediate form of 23.5 kD (asterisks) are protease-protected, confirming import into intact chloroplasts. The translation mixture sample (TL) represents 10% of the precursor added to the import reaction. (B) Immunoblot analysis of PIC1 localization in chloroplast subfractions. Chloroplasts from 6-week-old Arabidopsis rosette leaves were fractionated into envelopes (env), stroma (str), and thylakoid membranes (thy), as described in Methods. Equal amounts of proteins from the subfractions were separated by SDS-PAGE and subjected to immunoblot analysis using an antibody directed against the C-terminal part of At-PIC1. The -PIC1 stain gave a single band of 22 kD in chloroplast envelopes only. Antisera against the marker proteins Ps-Tic110 (inner envelope), Ps-LSU (stroma), and Ps-LHCP (thylakoid membranes) were used as controls. Numbers at left indicate the molecular mass of proteins in kilodaltons. (C) Subcellular localization of PIC1-GFP fusion proteins. Arabidopsis mesophyll protoplasts were transiently transformed with constructs for PIC1-GFP and OEP7-GFP. Images are shown for GFP fluorescence, chlorophyll autofluorescence, and an overlay of both. PIC1-GFP signals were localized in a spotted pattern to the periphery of chloroplasts. OEP7-GFP, included as a marker for the chloroplast outer envelope, gave more evenly distributed GFP fluorescence around the chloroplasts.

To verify the chloroplast localization in intact cells, full-length PIC1 was fused with the green fluorescent protein (GFP) and Arabidopsis mesophyll protoplasts were transiently transformed with this construct (Figure 2C). Merging the fluorescence images of GFP and chlorophyll indicated an envelope insertion of PIC1. GFP signals were located at the periphery of chloroplasts only and were associated with neither thylakoid membranes nor cytosolic or plasma membrane components of the cell. However, the GFP distribution appeared more spotted than the even signal of the outer envelope protein At-OEP7 (Lee et al., 2001), which was used as a control. Again, this indicates an association of PIC1 with the inner envelope membrane, because similar GFP signal patterns can be found for the inner membrane–linked VIPP1 (for vesicle-inducing protein in plastids) protein (Aseeva et al., 2004) and the P-type ATPase PAA1 (Abdel-Ghany et al., 2005).

PIC1 Is Ubiquitously Expressed in Arabidopsis and Peaks in Green Tissues
When we probed the Affymetrix full genome microarray (ATH1 chip) for the expression of PIC1 in rosette leaves and roots of 4-week-old Arabidopsis (Clausen et al., 2004), it demonstrated that PIC1 mRNA was present in both tissues but that transcript levels in leaves were approximately twofold higher than those in roots (Figure 3A ). This expression pattern could be confirmed by quantitative real-time RT-PCR on cDNA of 7-d-old seedlings (Figure 3B). Here, the relatively high transcript density, 1500 PIC1 transcripts per 10,000 actin transcripts, originated mostly from cotyledons. Screening the array data of the AtGenExpress consortium (Schmid et al., 2005) revealed that PIC1 is ubiquitously expressed throughout all developmental stages of Arabidopsis. Pronounced expression, however, was always linked to green tissues, consistent with a function of the protein in chloroplasts. Whereas mature pollen contained almost no PIC1 mRNA, the highest transcript content was found in cotyledons and the first two leaves of 7-d-old seedlings, in young rosette leaves, and in the shoot apex (Figure 3C). To further analyze the gene expression of PIC1 with respect to abiotic or biotic stimuli, we searched the publicly available databases. However, we did not find PIC1 to be regulated by any treatment or growth conditions. Moreover, PIC1 was not regulated by iron-sufficient or -deficient conditions (M.L. Guerinot and E. Rogers, personal communication), nor did it play a role in Zn tolerance (Talke et al., 2006). Thus, it is evident that PIC1 represents a constitutive and essential gene in Arabidopsis.

View larger version (18K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 3. Gene Expression of PIC1 Peaks in Green Tissues. (A) Expression profile of PIC1 in leaves and roots of 4-week-old Arabidopsis plants. Microarray signals (Affymetrix GeneChip Arabidopsis ATH1) are made comparable by scaling the average overall signal intensity of all probe sets to a target signal of 100 (arbitrary units). The average scaled signals (±SE) of three independent experiments are shown. (B) Quantification of PIC1 mRNA by quantitative real-time RT-PCR in 7-d-old Arabidopsis seedlings. Left, the transcript content in whole seedlings is displayed relative to 10,000 transcripts of actin 2/8 (n = 4; ±SD). Right, when seedlings were separated into cotyledons and roots, the PIC1 mRNA in roots was only 21 ± 9% of that in cotyledons. The transcript content was quantified relative to actin 2/8 mRNA (n = 3; ±SD) and normalized to the amount in cotyledons, which was set to 100%. Please note that the y axis is divided into two different scales for PIC1 transcripts/10,000 actin transcripts (left) and PIC1 transcripts in percentage (right). (C) In silico transcriptional profiling of PIC1 expression (arbitrary units) during plant development. Data used to create the in silico transcriptional profiles were obtained from AtGenExpress experiments (Schmid et al., 2005). Signal intensities were averaged from three technical replicates ± SD. c, cotyledons of 7-d-old plants; l, first and second leaves of 7-d-old plants; rl2 to rl12, rosette leaves 2, 4, 6, 8, 10, and 12 of 17-d-old plants; cl, cauline leaves of 21-d-old plants; sl, senescing leaves of 35-d-old plants; hy, hypocotyls of 7-d-old plants; sa, shoot of apex 7-d-old plants; r7, roots of 7-d-old plants; r17, roots of 17-d-old plants.

View larger version (20K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 4. Characterization of pic1 Mutant Alleles. (A) Diagram of the genomic organization of PIC1 and positions of identified T-DNA insertions. The PIC1 gene is 1704 bp long and comprises four exons (black arrows). The consecutive numbering of the gene starts with the predicted transcriptional start, located behind a potential TATA box (black bar). In consequence, PIC1 has a 5' UTR of 206 bp and an annotated 186-bp 3' UTR (National Center for Biotechnology Information reference sequence NM_127089). In three independent T-DNA insertion lines, the T-DNA has inserted into different parts of exon 1 (pic1-1 and pic1-2) and into the putative promoter region (position –430; pic1-3). ROK2, T-DNA present in the SALK collection (SALK_104852); AC161 and AC106, T-DNA present in the GABI-Kat lines 577D06 and 804F07, respectively. Positions and directions of left border T-DNA sequences (LB) and of gene-specific oligonucleotide primers used for PCR genotyping of pic1 mutants (04fw, 04rev, 52fw, 52rev, 77fw, and 77rev) and for RT-PCR (pic1,2fw, pic1,2rev, LCfw, and LCrev) are depicted. (B) The T-DNA in pic1-1 and pic1-2 interrupts the open reading frame of PIC1. In pic1-1, the T-DNA ROK2 has inserted at position 214 of the PIC1 gene, interrupting the open reading frame behind amino acid 2. In pic1-2, the AC161 T-DNA blocks translation after amino acid 37 of exon 1 (position 320 of PIC1). Please note that behind the T-DNA insertion site, we identified a deletion of 28 bp (boxed letters) in the PIC1 coding region. (C) Homozygous mutations in pic1-1 and pic1-2 result in loss of full-length PIC1 transcripts. Top graph, quantification of PIC1 transcripts in 7-d-old seedlings of Col-0 wild type and homozygous pic1-1, pic1-2, and pic1-3 mutant lines. Transcript density was measured by quantitative real-time RT-PCR with the primer pair LCfw-rev (see [A]) as described in Methods. The PIC1 mRNA content (n = 3; ±SD) was calculated relative to actin 2/8 transcripts and normalized to the amount in Col-0 seedlings, which was set to 100% (arbitrary units). Whereas in pic1-1/pic1-1 and pic1-2/pic1-2 the mRNA level was 10% of wild-type RNA (white bars), pic1-3/pic1-3 seedlings contained 34 ± 8.4% of Col-0 PIC1 mRNA (black bars). Bottom panel, RT-PCR experiments with pic1,2fw-rev primers (see [A]), which flank the T-DNA insertion sites of pic1-1 and pic1-2 mutant alleles. PCR was performed on cDNA of 7-d-old seedlings from Col-0 as well as from homozygous pic1-1, pic1-2, and pic1-3 (ho) and the corresponding wild-type alleles PIC1-1/PIC1-1, PIC1-2/PIC1-2, and PIC1-3/PIC1-3 (wt). Please note that the primer pair pic1,2fw-rev was not able to amplify a specific product (343 bp) on homozygous pic1-1 and pic1-2 mutants, indicating that these alleles lack full-length PIC1 transcripts. In turn, the 10% residual RNA detected by quantitative RT-PCR with C-terminal primers (top graph) most likely resulted from incomplete mRNA.

Whereas the homozygous pic1-3 and heterozygous pic1-1 and pic1-2 mutants showed no obvious phenotype, the appearance of the homozygous, independent knockout alleles pic1-1 and pic1-2 was dramatic (Figure 5 ). The loss of PIC1 function led to dwarfed and chlorotic plants (Figures 5A and 5B). The residual 35% fully transcribed PIC1 mRNA in homozygous pic1-3, however, was sufficient to provide a wild-type appearance to the mutant plants (Figure 5C). Homozygous pic1-1 and pic1-2 were seedling-lethal on soil; therefore, heterotrophic growth of the mutants had to be supplemented with 1% sucrose on Murashige and Skoog (MS) agar medium. Cotyledons in young seedlings were red, most likely because chlorophyll was low or absent and anthocyane pigments accumulated (Figure 5A). The majority of the slowly growing mutants produced almost transparent rosette leaves; however, in approximately one-third of the plants, the leaves appeared white. This subpopulation of mutants was even smaller and did not produce inflorescences (data not shown). Because pic1-1 and pic1-2 represent independently derived mutant alleles, the described phenotype in both lines (Figures 5A and 5B) can be linked to the loss of PIC1 function. Roots, like the entire pic1 mutant plant, were smaller than in wild-type plants. However, we did not observe any significant phenotype in roots of pic1 mutants maintained on sucrose-supplemented medium.

View larger version (95K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 5. Loss of PIC1 Generates Chlorotic and Dwarfed Mutant Plants. The phenotypes of homozygous pic1 mutants (ho) and the corresponding wild-type alleles (wt) are depicted. The age of the plants is given in days (d) or weeks (w). Homozygous plants of pic1-1 and pic1-2 mutants were grown on MS agar medium supplemented with 1% sucrose. If not indicated otherwise, black bars = 0.5 cm and white bars = 5.0 cm. (A) Homozygous pic1-1 mutants are dwarfed and chlorotic. Cotyledons in 7-d-old seedlings of homozygous pic1-1/pic1-1 (ho) are red. The first leaves after 2 weeks are transparent, and so is the fully grown rosette in the dwarfed plants after 4 weeks of growth. Please note that the shoot meristem of pic1-1/pic1-1 is pale green (cf. [B] and [D]). In pic1-1/pic1-1 gPIC1 (T2 generation, 4 weeks old), the loss of PIC1 has been fully complemented by transformation of the mutant with the PIC1 gene (see Methods). (B) pic1-2/pic1-2 plants display the same phenotype described for pic1-1/pic1-1 in (A). (C) The residual 35% of fully transcribed PIC1 mRNA in homozygous pic1-3 mutants (see Figure 3C) is enough to produce mutant plants with wild-type appearance. (D) Close-up of shoot apex and young leaves of a homozygous pic1-2 mutant plant (4 weeks old). Meristem and young leaves of the mutant are pale green, whereas older leaves become chlorotic during development and growth. (E) Close-up of developing inflorescence of a homozygous pic1-1 mutant plant (6 weeks old). Young sepals and the style of the flower are pale green. (F) Fully developed homozygous pic1-1 mutant after 8 weeks of growth. The sepals of the flower have turned white, the styles and developing siliques are pale green, and the rosette leaves and the stem of the plant are red.

Interestingly, the shoot apex and the very young emerging leaves of homozygous pic1-1 and pic1-2 mutants were pale green and their leaves turned chlorotic while growing (Figure 5D). After 6 weeks on sucrose-containing agar, surviving mutant plants were able to produce an inflorescence; again, the young sepals of the flower were pale green and became white subsequently (Figure 5E). Eight-week-old plants were fully grown with a mature size of 1 cm. The style and developing silique of the flower were pale green, whereas leaves and stem accumulated anthocyane pigments (Figure 5F). Because all flowers of homozygous mutants were sterile, pic1-1 and pic1-2 had to be propagated in the heterozygous state.

In summary, the described phenotypes of pic1-1 and pic1-2 mutants suggest an essential function of PIC1 in photosynthesis and plant growth. Obviously, loss of PIC1 function produces chlorotic plants, which are not able to assimilate carbohydrates by photosynthesis. Hence, their growth has to be supplemented with sucrose.

Loss of PIC1 Function Leads to Impaired Chloroplast Development
Next, leaf morphology and ultrastructure were analyzed in the homozygous pic1-1 mutant. Compared with Col-0 wild type (Figure 6A ), mature rosette leaves of pic1-1 were retarded in growth (Figure 6B). Interestingly, secondary and tertiary veins in pic1-1 leaves were thicker than in the wild type, and their diameter was commensurate with that of the primary vein. The organization of leaf mesophyll into palisade and spongy parenchyma cells was lost in pic1-1/pic1-1, and the leaf surface was extremely curled (Figures 6C to 6F). This deformation in the mutant leaf was already visible in cotyledons (Figure 6F). In addition, leaf cross sections revealed that mesophyll cells of homozygous pic1-1 (Figures 6D, 6F, and 6H), in contrast with wild-type plants (Figures 6C, 6E, and 6G), do not contain fully differentiated chloroplasts. Thus, the chlorotic phenotype of pic1-1/pic1-1 plants results from impaired chloroplast development.

View larger version (87K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 6. Leaf Morphology Is Changed in pic1-1 Mutant Plants. (A) and (B) Rosette leaves from 4-week-old Col-0 wild-type (A) and homozygous pic1-1 mutant (B) plants. Leaves of pic1-1/pic1-1 are reduced in size, and the diameter of primary, secondary, and tertiary veins is not graduated, as in the wild type. Before photography, leaves were destained in 70% ethanol. (C) to (F) Semithin cross sections of rosette leaves from 17-d-old Col-0 wild-type (C) and pic1-1/pic1-1 (D) as well as from 7-d-old cotyledons of Col-0 (E) and homozygous pic1-1 (F). In pic1-1/pic1-1 mutants, the organization of the leaf mesophyll into palisade and spongy parenchyma cells is lost (D). Furthermore, the leaf surface is extremely curled, and inside the cells no chloroplasts are visible ([D] and [F]). (G) and (H) Transmission electron micrographs of cortex cells from 7-d-old cotyledons of Col-0 (G) and pic1-1/pic1-1 (H) seedlings. Please note that the cotyledon cells in homozygous pic1-1 plants do not contain correctly developed chloroplasts (H).

View larger version (206K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 7. Chloroplast Development in pic1-1/pic1-1 Is Impaired, and Mutant Plastids Accumulate Ferritin. Chloroplast ultrastructure was monitored by transmission electron microscopy. Plastoglobules were present in plastids of mutant and wild-type plants at all developmental stages. M, mitochondria; S, starch grains; V, vacuoles. (A) to (C) Plastids from 7-d-old cotyledons. (A) shows a control chloroplast (Col-0 wild type) containing grana and stroma thylakoids, starch grains, and plastoglobules. By contrast, plastids of the mutant pic1-1/pic1-1 are about half the size and contain only inchoate ([B], white arrowhead) or no (C) thylakoids. (D) to (F) Chloroplasts in 17-d-old rosette leaves of Col-0 (D) and pic1-1/pic1-1 ([E] and [F]). Two types of plastids were present in rosette leaves of homozygous pic1-1: plastids characterized by the formation of membrane vesicles (asterisks) in the stroma (E), and plastids without thylakoids (F). (G) and (H) Proplastids in the shoot apex of 17-d-old plants. In the differentiating meristem of Col-0 wild-type plants (G), proplastids were characterized by developing thylakoid systems (white arrowhead). The shoot apex of pic1-1/pic1-1 (H) contains proplastids with developing thylakoids (white arrowhead) as well as without thylakoids and less electron-dense matrix (organelle at top right) in the same cell. In pic1-1 mutant plastids, phytoferritin clusters (white circles) are visible. (I) Ferritin cluster typical for plastids of cotyledons from 7-d-old pic1-1/pic1-1.

View larger version (45K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 8. Immunoblot Analysis of pic1 Mutants and Wild-Type Plants. Fifty micrograms of total protein, extracted from leaves of 4-week-old homozygous pic1-1 and pic1-2 mutants and Col-0 wild-type plants, were subjected to immunoblot analysis. Numbers at left indicate the molecular mass in kilodaltons, and asterisks indicate the molecular mass of the respective precursor proteins. Except for RBCS (tobacco [Nicotiana tabacum]), VIPP1 (pea), and PSBO-1 (spinach [Spinacia oleracea]), all antisera are directed against Arabidopsis proteins. Below, the molecular mass in kilodaltons of the respective precursor (pre) and mature (mat) forms of all proteins is indicated, and the chloroplast localization is specified. For Arabidopsis Genome Initiative codes of Arabidopsis proteins, see Table 1. (A) Proteins related to metal homeostasis are differentially regulated, whereas proteins associated with photosynthesis and carbon fixation are absent in pic1 mutants. Antisera against the following proteins were used: FER (ferritin; 28 pre, 23.5 mat, stroma), CSD1 (cytosolic copper superoxide dismutase; 15 kD), CSD2 (plastidic CSD; 22 pre, 18 mat, stroma), FSD1 (plastidic iron superoxide dismutase; 24 pre, 23 mat, stroma), LHCB4 (chlorophyll binding protein B4; 31 pre, 27 mat, attached to thylakoids), RBCS (small subunit of ribulose-1,5-bis-phosphate carboxylase/oxygenase; 20.2 pre, 14 mat, stroma), and OEP16.1 (outer envelope protein in chloroplasts of 16 kD). Please note that the FER antibody detects the ferritin protein family in Arabidopsis (FER1 to FER4) and LHCB4 reacts with the proteins LHCB4.1, -4.2, and -4.3 (see Table 1). (B) Chloroplast-localized proteins are properly imported into pic1 mutant plastids and processed to their mature size. Antisera against the following proteins were used: VIPP1 (vesicle-inducing protein in plastids; 43 pre, 36 mat, inner envelope), PORB (protochlorophyllide oxidoreductase B; 43 pre, 36 mat, stroma), and PSBO-1 (33-kD subunit of the oxygen-evolving complex; 40.5 pre, 33 mat, thylakoid lumen). TL, in vitro translation product of the precursor PORB. Please note that the PORB antibody detects PORB and PORC, both of which are decreased in pic1 mutant leaves (see Table1).

Photosynthetic Capacity Is Lost but Protein Import Is Still Functional in pic1 Plastids
Whereas metal homeostasis–associated proteins in pic1 mutants are differentially regulated, the photosynthetic capacity of mutant plastids was completely lost. This becomes apparent by the heterotrophic growth and chlorosis of homozygous pic1-1 and pic1-2 (Figures 5 and 6), the impaired biogenesis and degradation of thylakoids (Figure 7), and the downregulation of genes associated with photosynthesis and carbon assimilation (Table 1). For example, transcript abundance of chlorophyll binding proteins (e.g., LHCB4), subunits of the oxygen evolving complex (e.g., PSBO-1), proteins involved in chlorophyll biosynthesis (e.g., protochlorophyllide oxidoreductase; POR), and ribulose-1,5-bis-phosphate carboxylase/oxygenase (e.g., RBCS-1A) was decreased dramatically in pic1 mutants. Again, we verified these data by immunoblot analysis, in which LHCB4 and RBCS-1A proteins were not detectable and POR as well as PSBO-1 were decreased in pic1 mutants compared with the wild type (Figures 8A and 8B). It should be noted that all detected plastid-localized proteins were processed to their mature form in pic1 mutants (i.e., imported into the mutant plastids).

To further analyze the protein import capacity of pic1 mutant plastids, we chose the following proteins: VIPP1, which is localized at the inner chloroplast envelope (Aseeva et al., 2004); POR, which is localized in the chloroplast stroma; and PSBO-1, which is localized in the thylakoid lumen (Figure 8B). All three proteins were imported into plastids and processed to their mature size, as shown by immunoblot analysis. We never detected any residual precursor proteins in pic1 mutants, indicating that protein import into plastids is functional in pic1.

In contrast with proteins linked to metal homeostasis or photosynthesis, genes associated with housekeeping functions in the chloroplast stroma, such as aminoacyl tRNA synthetases (e.g., SerRS) or a nucleoside-diphosphate kinase (NDPK), as well as the general protein import pore Toc75 (Toc 75-III) and the amino acid–selective channel OEP16 (OEP16.1) in the outer chloroplast envelope, were not regulated in pic1 mutants (Table 1, Figure 8A).

PIC1 and Its Homolog from Synechocystis Mediate Iron Uptake in Yeast
Several cyanobacterial relatives of At-PIC1 are annotated as potential permeases for metal ions. In addition, the upregulation of ferritin gene expression and the accumulation of ferritin proteins in homozygous pic1-1 and pic1-2 mutants suggest a possible function of PIC1 in iron transport into chloroplasts. Furthermore, the phenotype of pic1 mutants, in particular leaf chlorosis, inhibition of chloroplast development, and lack of palisade parenchyma differentiation, is reminiscent of iron deficiency (Henriques et al., 2002; Varotto et al., 2002).

To investigate a potential role of PIC1 in metal transport across the plastidic inner envelope, we expressed the cDNA of PIC1 and its cyanobacterial homolog sll1656 from Synechocystis in the yeast double mutant fet3 fet4, which is defective in low- and high-affinity Fe uptake (Dix et al., 1994). Because of its reliance on additional and less efficient uptake mechanisms, the fet3 fet4 mutant requires high Fe concentrations in its growth medium. PIC1 and sll1656 cDNAs in the yeast expression vector pFL61 were transformed into fet3 fet4 yeast mutants. The cDNA of the iron transporter IRT1 from Arabidopsis roots (Eide et al., 1996) and the empty vector pFL61 were used as positive and negative controls, respectively. Because the fet3 fet4 mutant cannot grow under iron-limited conditions, the ability of the transformants to grow on minimal medium supplemented with Fe concentrations ranging from 0 to 20 µM FeCl₃ was investigated (Figure 9A ). Whereas yeast cells expressing the empty vector did not grow on 5 µM FeCl₃, expression of the Fe²⁺ transporter gene IRT1 from Arabidopsis efficiently complemented yeast growth. Expression of PIC1 and sll1656 restored the ability of fet3 fet4 to grow when supplemented with 5 µM or greater concentrations of FeCl₃. Similarly, in liquid minimal medium containing 20 µM FeCl₃, fet3 fet4 mutant cells expressing either PIC1 or sll1656 entered the exponential growth phase 2 to 3 h earlier than the vector control (data not shown). However, the complementation was not as efficient as with IRT1.

View larger version (50K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 9. PIC1 and Its Synechocystis Homolog sll1656 Are Able to Complement the Growth of Metal Uptake–Defective Yeast Mutants. The plasmid pFL61 (empty vector control) as well as the cDNA of IRT1, PIC1, and sll1656 in pFL61 were introduced into the fet3 fet4 (A) and ctr1 (B) yeast mutants. Serial dilutions with an OD600 of 10–1, 10–2, and 5 x 10–3 of rapidly growing yeast cells were dotted on SC (–uracil) minimal medium, pH 5.0, supplemented with different concentrations of FeCl3 or EDTA as indicated. The growth of yeast cells was documented after 2 d on the plate. (A) PIC1 and sll1656 are able to restore the growth of fet3 fet4 in the concentration range 5 to 10 µM FeCl3. IRT1-complemented fet3 fet4 cells are able to grow without additional iron, whereas dilutions of the empty vector–containing cells (pFL61) need 20 µM FeCl3 to grow. (B) With 0.1 and 0.5 mM EDTA in the medium, PIC1, sll1656, and IRT1 are still able to restore the growth of the yeast mutant ctr1. By contrast, cells transfected with the empty vector pFL61 grow poorly on 0.1 mM EDTA and fail to grow with 0.5 mM EDTA in the medium.

To show whether complementation of the fet3 fet4 mutant by PIC1 and sll1656 is attributable to iron uptake into the cells, we measured short-term ⁵⁹Fe accumulation in fet3 fet4 cells transformed with the empty vector pDR195 or the cDNAs of IRT1 (positive control), PIC1, and sll1656 in pDR195 (Table 2 ). The expression of either protein, PIC1 or sll1656, significantly increased the Fe uptake rates compared with the vector control. In agreement, with the growth complementation assays, however, Fe uptake rates mediated by PIC1 and sll1656 were severalfold lower than those mediated by IRT1. These results support a direct role of these two proteins in the membrane transport of iron.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Sequence comparison revealed that besides the PIC1 family in vascular plants and green and red algae, the only relatives of PIC1 are of cyanobacterial origin. Thus, it is very likely that PIC1 proteins evolved directly during endosymbiosis between a cyanobacterium and an ancient eukaryotic cell. This evolutionary step led to the formation of chloroplasts and hence to their function in photosynthesis in land plants and algae (Vothknecht and Soll, 2005). Because of the severe phenotype of pic1 knockout plants, it became evident that PIC1 function is essential for thylakoid biogenesis and, in turn, for photosynthesis. The fact that the cyanobacterium G. violaceus contains neither thylakoid membranes (Nakamura et al., 2003) nor PIC1 homologs emphasizes this observation.

PIC1 proteins of cyanobacterial origin belong to COGs (Tatusov et al., 2003), which generally play a role in ion or solute transport across membranes. Several cyanobacterial relatives of PIC1 have been annotated as potential permease components of ABC transporters. Unlike eukaryotic ABC transporters, prokaryotic transporters of this family are composed of independent permease and ABC protein subunits (Higgins, 2001). Here, the permease subunits represent the integral membrane proteins and form the pore, whereas the soluble ABC subunits bind to the permeases and energize the transport by ATP cleavage. Several soluble, potential ABC subunits, which are predicted to be located in plastids, exist in plants and might represent interacting partners of PIC1 (Garcia et al., 2004). Among them, the gene GCN1, which encodes two soluble ABC subunits, is upregulated in pic1-1 knockout mutants (Table 1). Thus, a function of PIC1 in membrane transport processes most likely has been evolutionarily conserved.

pic1 Mutant Plants Display a Severe Chlorotic Phenotype
PIC1 is expressed ubiquitously throughout all organs and developmental stages of Arabidopsis (Figure 3). In agreement with an essential function in chloroplasts, PIC1 transcript abundance peaked in green tissues. However, PIC1 is also expressed in roots and therefore might fulfill a similar function at the inner envelope of root plastids as well.

Two independently derived T-DNA insertion lines of PIC1 (pic1-1 and pic1-2) could be characterized as knockout mutations. Both mutant alleles generated the same phenotype in the homozygous progeny (Figure 5). Furthermore, we were able to complement the phenotype of pic1-1 by the gene PIC1 under the control of its own promoter. Thus, it is evident that the loss of PIC1 was responsible for the described phenotype. Interestingly, 10% residual transcript fragments of PIC1 were detected in pic1-1/pic1-1 and pic1-2/pic1-2 (Figure 4C). However, these transcripts lacked the N-terminal part of the PIC1 coding sequence; therefore, they could not function as templates for the synthesis of functional PIC1 proteins. Most likely, these transcript fragments originated from an ectopic promoter activity of the T-DNA insert.

pic1 knockout mutants developed severe chlorosis and only grew heterotrophically. Hence, PIC1 function is necessary for chlorophyll synthesis and/or accumulation and is a prerequisite for proper photosynthesis as well as carbohydrate assimilation in chloroplasts. Differential regulation of transcript and protein levels of genes related to photosynthesis in pic1 mutants (see below) emphasizes this observation. As demonstrated by transmission electron microscopy, functional thylakoid membrane systems did not develop in the homozygous pic1-1 mutant. Because rudimentary thylakoid membranes were present in plastids of cotyledons and the meristem of pic1-1/pic1-1 but not in mature rosette leaves, we conclude that thylakoid biogenesis in pic1-1/pic1-1 is arrested at a proplastid-like stage. This is consistent with the observation that young leaf organs of pic1 mutants were pale green and contained some chlorophyll but turned chlorotic while growing. As a consequence, vesicle formation inside plastids of rosette leaves suggests that inchoate thylakoids are degraded during maturation of the leaf (Figures 7E and 7F). Furthermore, the fact that mutant leaves contain smaller and fewer plastids than wild-type leaves suggested a parallel organelle decomposition as well (Figure 6H). However, residual but fewer plastids without thylakoids remained in leaf cells, probably necessary for biosynthetic functions apart from photosynthesis (for functions of plastids, see Möller, 2005). This phenotype of pic1 mutants was emphasized by the presence of transcripts and proteins associated with housekeeping functions of the chloroplast (see below).

A Role of PIC1 in Iron Transport across the Chloroplast Inner Envelope
Growth complementation on low iron supply of a yeast mutant defective in iron uptake, and uptake assays using radiolabeled Fe, showed that PIC1 and its homolog sll1656 from Synechocystis are capable of mediating membrane transport of iron. Thus, to our knowledge, PIC1 appears to be the first protein shown to be involved in iron transport across plastid membranes. Other chloroplast proteins with a role in metal ion transport across the chloroplast envelope are the copper-transporting ATPases PAA1 and HMA1 and the magnesium transport protein MRS2-11 (Shikanai et al., 2003; Abdel-Ghany et al., 2005; Drummond et al., 2006; Seigneurin-Berny et al., 2006). As most of the plant transporters involved in metal ion transport across the plasma membrane or the tonoplast, such as ZIP or NRAMP proteins, usually transport a broad range of heavy metal ions (Colangelo and Guerinot, 2006), it cannot be excluded that PIC1 and sll1656 might transport other metal ions besides iron as well. Because PIC1 and sll1656 were able to complement the Cu uptake–deficient yeast mutant ctr1, a function in copper transport might be possible. However, because of the Cu dependence of Fe uptake in yeast (Dancis et al., 1994b), it is not yet clear whether the observed complementation of ctr1 is a result of restored copper or iron transport. Moreover, we think that the phenotypes, differential gene expression, and metal content in the pic1 mutants favor a function in iron rather than copper transport (see discussion below).

In short-term uptake studies with radiolabeled Fe, PIC1 and sll1656 showed iron uptake rates that were much lower than those mediated by IRT1 (Table 2). These lower Fe uptake rates were in agreement with a retarded and weaker growth complementation of yeast cells on iron-limiting medium by PIC1 and sll1656 (Figure 9). A similar difference from IRT1-mediated Fe uptake in yeast has been observed in yeast cells expressing metal transporters of the NRAMP family (Curie et al., 2000; Thomine et al., 2000; Kaiser et al., 2003). Furthermore, because our assay relied on the expression of a plastid protein in yeast, it is most likely that iron uptake was mediated by a certain amount of plasma membrane–localized, mistargeted protein, explaining the low Fe uptake rates compared with the plasma membrane–intrinsic IRT1.

Chlorosis, caused by the inhibition of chlorophyll synthesis and chloroplast development (Briat and Lobreaux, 1997), is a classical symptom of iron deficiency in leaves and particularly in plastids. Furthermore, in iron-deficient leaves, the palisade parenchyma cells do not differentiate properly, and in chloroplasts, grana stacks of thylakoids are reduced drastically (Henriques et al., 2002; Varotto et al., 2002). In the case of pic1 knockout plants, we propose that the transport pathway of iron is blocked at the inner chloroplast envelope, causing the described iron deficiency symptoms: chlorosis, lack of cell differentiation in leaf mesophyll parenchyma, and impaired thylakoid development.

Because pic1 knockout mutants were able to germinate and develop rudimentary thylakoids in plastids of cotyledons and pale green young rosette leaves, an alternative iron uptake pathway seems to exist in Arabidopsis plastids. Possible bypassing transporters could be plastid-localized, iron-transporting proteins, such as NRAMP1 (Curie et al., 2000) and NRAMP6, which contain potential chloroplast transit peptides (for review, see Curie and Briat, 2003; Colangelo and Guerinot, 2006). Interestingly, both of these genes were slightly upregulated in pic1-1 mutants (Table 1). NRAMP1 has a function in Arabidopsis roots (Curie et al., 2000), but according to the data of the AtGenExpress consortium (Schmid et al., 2005), transcripts of NRAMP1 are further present in the vegetative shoot apex and flowers, whereas NRAMP6 is very specifically expressed at later stages in seed development. Thus, it might be possible that these metal transport proteins confer a weak iron supplementation to pic1 mutant plastids in seeds and meristematic tissues.

The described phenotype of pic1 mutants as well as the in vitro iron uptake mediated by PIC1 both suggest an in vivo function in iron import into the plastid. However, impaired metal homeostasis and ferritin accumulation in pic1 mutants might also be caused by disturbed copper transport, a defect in iron export from the plastid, or inefficient iron mobilization for Fe-S cluster assembly in chloroplasts (see below). With regard to the strong sink activity for Fe exerted by the plastidic expression of ferritin (Van Wuytswinkel et al., 1998), a role of PIC1 in iron export becomes unlikely. In conclusion, PIC1 most likely functions in Fe uptake into chloroplasts, but other roles in plastid metal ion homeostasis and transport (e.g., Fe export, Cu import/export) might be possible.

Loss of PIC1 Disturbs Metal Homeostasis in Leaves
In addition to structural changes in pic1 knockout lines that resemble iron deficiency, leaf cells apparently experience iron stress and perturbed metal homeostasis. Unfortunately, we were not able to isolate intact plastids from pic1 knockout plants to determine their iron content. However, when we measured the iron content in a total leaf extract from homozygous pic1-1 plants, it was not altered relative to that of wild-type leaves. Furthermore, we were not able to rescue the pic1 mutant phenotype by the addition of exogenously supplied Fe chelates {Fe(III)-EDTA, Sequestren [Fe(III)-EDDHA]}. When we tested the knockout mutants pic1-1 and pic1-2 grown on agar plates (1% sucrose, 0.5x MS salts) under iron-deficient (addition of 300 µM ferrozine), iron-sufficient, and iron-abundant [addition of 50 µM Fe(III)-EDTA] conditions, we did not detect significant differences in the mutant phenotype. Furthermore, homozygous pic1-3 mutants in this assay did not display a phenotype compared with wild-type plants. In all plants analyzed for plus/minus iron, we monitored PIC1 transcripts by quantitative RT-PCR, which demonstrated no regulation of PIC1 expression by iron (data not shown). In publicly available microarray data, PIC1 was constitutively expressed under all biotic and abiotic impacts described (e.g., light, hormones, stress, Pb and Zn tolerance, and phosphate and potassium deficiency). Moreover, expression of the permease was under the control of neither iron nor zinc (Talke et al., 2006; M.L. Guerinot, E. Rogers, and U. Krämer, personal communication). Thus, these results show that PIC1 represents a constitutive and essential gene in Arabidopsis that seems not to be regulated by exogenous metal availability, like the iron uptake transporter IRT1 in roots (Eide et al., 1996; Connolly et al., 2002; Vert et al., 2003).

But what happens to metal homeostasis of the plant cell when PIC1 function is lost? In pic1 mutant plants, ferritin proteins accumulated in plastids to much higher levels than in wild-type plants. A major role of ferritin is to precipitate Fe in the inner core of the ferritin protein cluster and thus to withdraw iron from an intracellular generation of reactive oxygen species (for overview, see Briat et al., 1999; Connolly and Guerinot, 2002). The unusual accumulation of ferritin in pic1 knockout mutants was documented by electron microscopy (Figures 7H and 7I) at the transcript level (Table 1) as well as at the protein level (Figure 8). Thus, the increase in ferritin protein and the formation of ferritin clusters in mutant plastids was accompanied by the upregulation of FER1 and FER4. Iron overload in plants has been described to cause severe oxidative stress, because free and excess Fe²⁺ ions lead to the formation of hydroxyl radicals via the Fenton reaction (Briat and Lobreaux, 1997). As a consequence, plants upregulate Cu-dependent as well as Fe-dependent superoxide dismutases (CSD and FSD) (Kliebenstein et al., 1998; Asada, 1999, Abdel-Ghany et al., 2005). Interestingly, the Cu content of pic1-1 shoots was twice as great as in wild-type shoots, whereas the Fe levels were the same (see Supplemental Table 1 online). Compared with the wild type, CSD1 (cytosolic) and CSD2 (plastid-localized) increased parallel to the Cu content in leaves of pic1 mutants, whereas transcripts and protein levels of the major plastid-localized iron superoxide dismutase (FSD1) were slightly downregulated, suggesting a shortage of usable iron in plastids. Furthermore, most transcripts that were increased in homozygous pic1-1 are supposed to function in stress-related processes (data not shown), suggesting that pic1 knockout mutants suffer from cytosolic iron stress, plastidic iron shortage, and a general copper overload. In conclusion, we propose that PIC1 function is closely linked to metal homeostasis in plastids and the cytosol.

To understand the integration of PIC1 into the metal homeostasis of the cell, we screened for differentially expressed metal transport proteins in pic1-1 mutants (Table 1). Besides NRAMP1 and NRAMP6 (see above), we found the putative Fe-nicotianamine transporter YSL1 to be upregulated in pic1-1. YSL1 has been shown to be involved in the loading of iron into seeds, and the expression of YSL1 is increased in response to iron excess (Le Jean et al., 2005; Waters et al., 2006). Although YSL1 as its relative YSL2 in Arabidopsis most likely is localized in the plasma membrane, it is discussed that the protein might be involved in Fe detoxification in specific cellular compartments (Le Jean et al., 2005). By contrast, transcripts of the iron uptake transporter IRT1, which is induced by iron deficiency in roots (Eide et al., 1996; Connolly et al., 2002; Vert et al., 2003), are decreased in pic1-1. Because IRT1 signals were at the detection limit of the Affymetrix microarray, we verified this observation by quantitative RT-PCR (see Supplemental Figure 2 online). Interestingly, compared with wild-type signals, IRT1 transcripts in pic1-1 and pic1-2 were decreased under iron-sufficient and iron-abundant conditions but not under iron deficiency, indicating that IRT1 expression was repressed more strongly in pic1 mutants than in the wild type. Thus, the loss of PIC1 might interfere with iron-triggered signal transduction within the cell.

Furthermore, expression of the copper-transporting ATPase PAA1 (inner envelope of chloroplasts) (Abdel-Ghany et al., 2005) and the potential high-affinity Cu uptake protein COPT2 (for overview, see Colangelo and Guerinot, 2006) was downregulated. Thus, expression of genes related to iron stress and metal transport in pic1 mutants reflects iron deficiency in plastids (FSD1, NRAMP1, and NRAMP6), iron overload in the cytosol (FER1, FER4, CSD1, YSL1, and IRT1), and an increased copper content in leaves (CSD2, PAA1, and COPT2). The described phenotype of pic1 knockout plants (ferritin increase, IRT1 decrease, and oxidative stress) and complementation of the copper uptake–deficient yeast