Arabidopsis ANTR1 Is a Thylakoid Na+-dependent Phosphate Transporter

In this study, the putative anion transporter 1 (ANTR1) from Arabidopsis thaliana was shown to be localized to the chloroplast thylakoid membrane by Western blotting with two different peptide-specific antibodies. ANTR1 is homologous to the type I of mammalian Na+-dependent inorganic phosphate (Pi) transporters. The function of ANTR1 as a Na+-dependent Pi transporter was demonstrated by heterologous expression and uptake of radioactive Pi into Escherichia coli cells. The expression of ANTR1 conferred increased growth rates to the transformed cells and stimulated Pi uptake in a pH- and Na+-dependent manner as compared with the control cells. Among various tested effectors, Pi was the preferred substrate. Although it competed with the uptake of Pi, glutamate was not transported by ANTR1 into E. coli. In relation to its function as a Pi transporter, several physiological roles for ANTR1 in the thylakoid membrane are proposed, such as export of Pi produced during nucleotide metabolism in the thylakoid lumen back to the chloroplast stroma and balance of the trans-thylakoid H+ electrochemical gradient storage.

Solute and metabolite transporters play essential roles in physiological processes, including nutrient uptake, cell homeostasis, signal transduction, growth, and stress responses in every living organism. In the plant chloroplast, the photosynthetic organelle, most transporters have been identified and biochemically characterized from the envelope membrane (1,2). Among them, there are several translocators for inorganic phosphate (P i ), all functioning as antiport systems using P i or phosphorylated C3 and C6 compounds as counter substrates (3,4). Much less information is available for transport processes across the chloroplast thylakoid membrane, which is mostly studied as the site of light-driven photosynthetic reactions coupled to ATP synthesis. Only a few thylakoid transporters have been identified and functionally characterized. Examples are ATP transport across spinach thylakoid membrane into the lumenal space and a thylakoid ATP/ADP carrier identified and characterized in Arabidopsis thaliana (5,6). An active nucleotide metabolism in the thylakoid lumen (5) implies the existence of additional, yet unidentified, transporters, such as those recycling P i to the soluble stroma.
A few Na ϩ -coupled P i transporters (NaP i ) 6 have in recent years been reported in green algae and vascular plants (7,8). NaP i systems are known to be mostly active in mammals, whereas H ϩ -coupled P i transport is dominant in plants (4). Three NaP i types have been described in eukaryotes, NaP i -II and III being the main ones in mammals (9 -11). NaP i -II type plays the role of an intracellular P i accumulation system, whereas NaP i -III type has the characteristics of a housekeeping system (9). The molecular mechanisms controlling the NaP i -II and III uptake systems have recently been reviewed (11). NaP i -I represents a group of proteins for which the endogenous substrate, ionic coupling, and physiological function are still under debate. Heterologous expression of the rabbit renal NaP i -1 and the human orthologue NPT1 in Xenopus laevis oocytes have implicated type I in Na ϩ -dependent P i transport as well as in a channel-like conductance of organic and inorganic anions (10). The first identified vesicular glutamate transporters, VGLUT1 and VGLUT2, were initially also characterized as NaP i -I type.
Most recently, VGLUTs have been shown to transport glutamate as well as P i by two independent mechanisms (12).
In A. thaliana, there are six genes encoding anion transporters (ANTR1-6), sharing homology with the NaP i -I members. The ANTR1 and ANTR2 proteins have been localized to the chloroplast and the latter precisely assigned to the inner envelope using proteomics and immunodetection with a peptidespecific antibody (1,13). The subcellular location of ANTR3- 6 has not yet been addressed experimentally. Moreover, no function of the ANTRs has yet been determined.
In the present study, we show that the recombinantly expressed Arabidopsis ANTR1 facilitates Na ϩ -dependent P i transport into Escherichia coli. This bacterium contains two H ϩ -coupled P i uptake systems, namely the low affinity system Pit and the high affinity system Pst (14,15). There are no known NaP i transporters in E. coli, which makes this organism suitable for heterologous expression and functional characterization of ANTR1. Moreover, by the use of peptide-specific antibodies we demonstrate that Arabidopsis ANTR1 is a thylakoid membrane transporter.

EXPERIMENTAL PROCEDURES
Organisms-Arabidopsis (A. thaliana cv. Columbia) plants were grown hydroponically at 120 mol photons m Ϫ2 s Ϫ1 and 22°C with 8-h light/16-h dark cycles. Total protein extracts from various tissues were prepared according to Ref. 13. Intact chloroplasts and pure thylakoid and envelope membranes were prepared as described (6). For heterologous expression experiments, E. coli strain TOP10F' (Invitrogen) carrying the lacIq repressor for inducible expression using isopropyl 1-thio-␤,Dgalactopyranoside (IPTG) was employed.
In Silico Analyses-Prediction of subcellular location was performed using TargetP (17). Membrane topology was analyzed using a package of programs at the ARAMEMNON Web site and another one at PRODIV-TMHMM (18,19). Sequence alignments were performed using ClustalW software (20).
Cloning and Heterologous Expression of His 6 -Xpress-ANTR1-FLAG Fusion Protein in E. coli-The Arabidopsis At2g29650 (ANTR1 gene) was PCR-amplified from the RAFL09-06-K07 cDNA clone (RIKEN BioResource Center, Ref. 16) using a sense primer (5Ј-GAGAGACTGCAGAACGCGAGAGCTCTTCT-TTGC-3Ј) flanking the PstI site (underlined) and an antisense primer (5Ј-GAGAGAAAGCTTTCATTTATCGTCATCGT-CTTTATAATCATCGATTATCTTCTCTCCGGT-3Ј) harboring the HindIII site (underlined) and the sequence encoding the FLAG peptide, DYKDDDDK (boldface). The PCR product was inserted between the PstI and HindIII sites in the pTrcHisC vector (Invitrogen) and transformed into E. coli TOP10F' cells. Such a fusion construct has recently been successfully employed to express another thylakoid transporter (6). The authenticity of the vector construct was confirmed by DNA sequencing.
Transformed (EANTR1) and control (carrying an empty plasmid) E. coli cells were precultivated overnight in Terrific Broth medium containing 70 mM potassium phosphate buffer, pH 7.5 (TB-P i ), supplemented with 50 g/ml ampicillin, and re-inoculated to an A 600 of 0.6 in fresh TB-P i medium, supplemented with 50 g/ml ampicillin and 100 M IPTG for induction. Cultures were grown aerobically at 200 rpm and 25°C, and aliquots were harvested by centrifugation (8000 ϫ g, 5 min) every 2 h during an 8-h induction period. Harvested cells were stored at Ϫ80°C prior to electrophoretic analyses.
Assay of P i and Glutamate Uptake into E. coli Cells-Transformed and control cells were induced with IPTG for 4 h, harvested by centrifugation, resuspended to an A 600 of 0.3 in TB medium (no added P i ), and starved for 2 h. Cells were harvested by centrifugation and washed with 25 mM Tris-succinate buffer, pH 6.5. P i uptake was assayed by the addition of 1 l of [ 32 P]orthophosphate (50 mCi/mmol; 1 mCi ϭ 37 MBq; GE Healthcare) to a final concentration of 100 M in 20-l aliquots, each containing 2 mg (wet weight) of cells in Tris-succinate buffer, pH 6.5, supplemented with 3% glucose, in the absence or presence of 25 mM NaCl. The suspension was incubated at 25°C for the indicated time periods. P i transport was terminated by the addition of 1 ml of ice-cold 25 mM Tris-succinate buffer, pH 6.5, and rapid filtration under vacuum as previously described (6). Following three washes with the same ice-cold buffer, the radioactivity retained on the filter was determined by liquid scintillation spectrometry. Similar assays were carried out in 25 mM Tris-succinate buffer, pH 5.5 and 7.5. For determination of the transport affinity (K m ) and maximal rate (V max ), uptake of 32 P i was carried out using a range of concentrations of 0 to 500 M. When the effect of Na ϩ concentration on P i uptake was measured, the amount of P i was kept constant (100 M) and the amount of NaCl added to the reaction mixture was varied (0 -50 mM). Choline chloride was used to replace NaCl in some experiments to verify the importance of Na ϩ versus Cl Ϫ . Where indicated, the uptake experiments were carried out using 100 M 32 P i in competition with 2.5 mM nonlabeled effectors. For uptake inhibition experiments, E. coli cells were preincubated for 2 min with the Na ϩ ionophore monensin (100 M) before the addition of 100 M 32 Pi.
For glutamate transport experiments, control and EANTR1 cells were first incubated for 30 min in 25 mM Tris-succinate buffer, pH 6.5, followed by washing with the same buffer. The uptake was assayed as described above for P i but using 100 M L- [3, H]glutamate (500 mCi/mmol; 1 mCi ϭ 37 MBq; PerkinElmer Life Sciences) in 25 mM Tris-succinate buffer, pH 6.5, supplemented with 3% glucose and 25 mM NaCl. The data throughout this study represent the mean Ϯ S.D. of at least three independent experiments with three replicates.
Protein Analysis-Proteins were separated by electrophoresis in 14% (w/v) acrylamide-SDS gels and analyzed by Western blotting using antibodies directed toward epitopes within the protein or the fused tags. An ANTR1-specific antibody was produced in rabbit against a peptide corresponding to the 15 residues within the N terminus of the protein (73-88, CEGDKVS-GNNDVVSDSP) and purified by affinity chromatography (Innovagen, Lund, Sweden). Another antibody was produced in rabbit against the peptide 427-443 (CSQGTDAFSQSGLYSN), common for ANTR1 and ANTR2. Where indicated, anti-FLAG M2 monoclonal (Sigma) and anti-Xpress (Invitrogen) antibodies were also used. To verify the purity of the chloroplast fractions, antibodies against the envelope TIC110 protein (gift from Prof. J. Soll, Munich University) and the thylakoid LHCII protein (Agrisera, Umeå, Sweden) were employed.

Structural Analyses of the Arabidopsis ANTR1 Protein-
Analysis of the 512-amino acid sequence of the ANTR1 protein from Arabidopsis (UniProtKB O82390) revealed the presence of a putative chloroplast transit peptide (TargetP score 0.99) with a cleavage site between amino acids 59 -60 (Fig. 1A). The theoretical molecular masses for the full-length and processed forms are 56.5 and 50.7 kDa, respectively.
ANTR2 is the closest homologue among the ANTRs, showing 70% sequence identity (80% similarity) to ANTR1 (data extracted from ARAMEMNON). Although both proteins have been localized to the chloroplast (13), the transit peptide is the least conserved region between them, implying a distinct intrachloroplast location and/or import pathway. ANTR1 ortho-logues are present in other sequenced plant species such as rice (86%) and Popullus (88%) (data extracted from ARAMEM-NON). No sequence homology was found to other known P i transport systems from plant chloroplasts or E. coli.
The topology of NaP i -I has never been addressed experimentally. The NaP i -I proteins are predicted to contain 12 transmembrane domains (TMDs) (9, 10) and an extended hydrophilic loop between TMD VI and TMD VII, as in the characteristic 6 ϩ 6 configuration of many major facilitator superfamily members. Prediction of ANTR1 topology at ARAMEMNON (18) revealed the presence of 8 -12 putative TMDs and an "in" (i.e. stroma) orientation for both the N and C termini. The most controversial hydrophobic regions (score Ͻ 0.4) correspond to TMD IV, TMD IX, and TMD X (Fig. 1, A  and B). In addition, another set of topology programs at PRODIV-TMHMM (19) indicated the presence of 12 putative TMDs. The most plausible reason for the large variability between various tested software is the presence of a significant number of charged residues as well as prolines and glycines within the putative TMDs, as in the case of the thylakoid ATP/ ADP carrier (6). A 12-TMD topology was resolved in the crystal structure of four different major facilitator superfamily proteins (21). Assuming that members of the same superfamily of transporters have a similar structure regardless of the type of substrate, we propose a 12-TMD model for ANTR1 in the thylakoid membrane in which both the termini and the long central loop (46 residues) connecting the two six-helix halves are exposed to the stromal side of the membrane (Fig. 1B).
A 5-residue consensus sequence has been proposed for various Na ϩ -dependent transport systems (22), which is consistent with G342/A381X 4 LX 3 PR391 in rabbit NaPi-1 (23) and also present in rat VGLUT2 (Fig. 1C). This consensus sequence was found partially conserved in ANTR1, i.e. the leucine (non-polar) residue is changed for a glutamine (polar but uncharged) residue (Fig. 1C). Nevertheless, despite this non-conservative change the Na ϩ dependence of ANTR1 transport activity has been validated experimentally as described below.
Expression and Localization of the ANTR1 Protein in Arabidopsis-Western blot analysis of total protein extracts from various Arabidopsis tissues was performed using an ANTR1-specific antibody (see "Experimental Procedures"). A single cross-reacting band with M r of 45 was detected and corresponds to a protein mainly expressed in photosynthetic tissues (e.g. mature leaves and flower buds), less in senescent leaves, and absent in roots ( Fig. 2A), in line with microarray expression data available at Genevestigator data base (24).
The predicted chloroplast location of ANTR1 was experimentally confirmed using transient expression of a green fluorescent protein fusion construct (13) but not precisely assigned to any of the chloroplast membrane compartments. In this study, chloroplasts isolated from Arabidopsis leaves as well as thylakoid and envelope membrane subfractions, purified by sucrose density gradient centrifugation (6), were analyzed by Western blotting using the anti-ANTR1 antibody. The 45-kDa protein band was detected in chloroplasts, enriched in the thylakoid subfraction, but not found in the envelope (Fig. 2B, upper  panel). No cross-reacting products were detected when the corresponding rabbit preimmune serum was used (Fig. 2, A and  B, upper panel), supporting the specificity of the anti-ANTR1 antibody. The 45-kDa thylakoid protein is indicated as the ANTR1 protein. Control experiments using antibodies for envelope (TIC110) and thylakoid (LHCII) markers verified the high purity of the tested fractions (Fig. 2B, lower panel). When using another antibody raised against a common peptide for ANTR1 and ANTR2 (see "Experimental Procedures"), two distinct cross-reacting protein bands with similar M r of ϳ45 were detected in the chloroplast membrane subfractions. The upper band corresponds to the thylakoid protein, also detected with FIGURE 1. Sequence analyses of the Arabidopsis ANTR1 protein. A, the amino acid sequence of the full-length ANTR1 protein from Arabidopsis (Uni-ProtKB O82390). The predicted transit peptide cleavage site is indicated by an arrow. The anion:cation symporter consensus sequence is marked by asterisks. TMDs predicted with high scores (Ͼ0.8) are highlighted, whereas TMDs predicted with low scores (Ͻ0.4) are boxed. B, topology model based on TMD prediction as in A. C, conservation of the consensus sequence for Na ϩ -dependent transport systems in rat VGLUT2, rabbit NaP i -1 and Arabidopsis ANTR1. The consensus sequence and the identical and conserved residues are indicated below the alignment. the ANTR1-specific antibody (Fig. 1C). The lower band was detected exclusively in the envelope and corresponds most likely to ANTR2, in line with previous observations (13).
Expression and Functional Characterization of ANTR1 in E. coli Cells-P i transporter mutants of yeast are widely used for functional analysis of plant P i transporters. However, previous work (13) and our own attempts showed that, although expressed, the ANTR1 protein was inactive with respect to anion transport. Instead, the bacterial system has proven successful for functional characterization of another thylakoid transporter (6) as well as of a prokaryotic Na ϩ -dependent P i transporter (25). Therefore, here we have expressed the fulllength ANTR1 protein and studied its functional properties in E. coli cells as a recombinant His 6 -Xpress-ANTR1-FLAG fusion construct (Fig. 3A). A cross-reacting band with M r of 56 was detected using the anti-FLAG antibody in the IPTG-induced transformed (EANTR1), but not in control, cells (Fig.  3B). The 56-kDa protein band was readily detected after 2 h of induction, and its level remained constant for up to 6 -8 h. It corresponds to a membrane protein, as revealed by subcellular fractionation (data not shown). The 56-kDa protein was also immunodetected by the anti-ANTR1 antibodies as well as by the Xpress peptide antibody (data not shown). The preserved N terminus harboring the Xpress tag (Fig. 3A) and the size of 56 kDa for the recombinant full-length protein indicate that the E. coli system is unable to process the transit peptide (see Fig. 1A). During 8 h of IPTG induction in TB-P i medium, the EANTR1 cells grew considerably faster than the control cells (Fig. 3C), suggesting an advantage conferred by the activity of the heterologously expressed Arabidopsis protein.
Arabidopsis ANTR1 renders the transformed strain able to take up the highest levels of P i in the presence of NaCl, whereas a 2-to 3-fold reduction in the uptake activity was determined in the absence of NaCl (Fig. 4A). P i uptake in control cells both in the absence and presence of NaCl yielded ϳ20% of the maximal level of accumulated P i in the EANTR1 cells. The uptake activity of EANTR1 cells in the absence of NaCl was only slightly higher than that of control cells under the same conditions, a phenomenon probably due to presence of trace amounts of NaCl in the cell suspension. The specificity for Na ϩ versus Cl Ϫ was verified by the 10-fold reduction in transport activity when NaCl was replaced by choline hydrochloride in the assay reactions (Fig. 4B). Preincubation with a Na ϩ ionophore, monensin, reduced the P i accumulation into the EANTR1 cells in the presence of NaCl to the level observed in control cells (data not shown), most likely due to its inability to inhibit the intrinsic H ϩ -coupled P i transport of E. coli (Ref. 15 and references therein). Taken together, these data indicate that, in contrast to the bacterial systems, ANTR1 can efficiently utilize the Na ϩ gradient created across cytoplasmic membrane for P i uptake, in line with the presence in ANTR1 of the consensus sequence for Na ϩ -dependent transport systems (Fig. 1C).
The induced expression of the cloned ANTR1 gene by the addition of IPTG does not appear to repress the expression of other host-encoded P i uptake mechanisms that result in the activity in the uptake of P i in non-induced cells. In the noninduced EANTR1 cells the activity was low and Na ϩ -independ-  ent (data not shown). After induction, the Na ϩ -dependent component was dramatically increased and showed saturation behavior, suggesting an induction of ANTR1 activity in addition to that of bacterial-encoded Na ϩ -independent P i uptake systems. These observations suggest the advantage conferred by the ANTR1 (in the background of bacterial-derived activity) in taking up more P i in a Na ϩ -dependent transport mechanism (see Figs. 3C and 4A). In all the forthcoming experiments, the transport activity of ANTR1 was characterized by subtracting the values in EANTR1 cells by those for the control strain.
Next, we investigated the sensitivity of ANTR1 transport activity to pH in the presence or absence of NaCl. The activity of ANTR1 in the absence of added NaCl was low, in the pH range 5.5 to 7.5, although 3-fold higher at pH 5.5 than at 6.5, and gradually increased with increasing pH only in the presence of NaCl (Fig. 4C). Thus, at pH 6.5 the activity was 15-fold higher than that observed at pH 5.5 and only 2-fold further increased at pH 7.5. A severe reduction was observed in the uptake at pH 8.5 (by 60%) as compared with 7.5, as also reported for the mammalian Na ϩ -dependent Pit transporters (26). The activity in control cells was at its optimum at pH 5.5 and showed no stimulation by the addition of NaCl (data not shown). Increased P i transport by ANTR1 and the stimulation by NaCl at higher pH may be explained as follows. At pH below 6.0 most P i is present as H 2 PO 4 Ϫ and the transport does not seem to require Na ϩ ions. The activity increases dramatically with pH when the HPO 4 2Ϫ species becomes dominant and requires the presence of Na ϩ ions.
Among the alkali ions tested, maximum (15-fold) stimulation of P i uptake by ANTR1 at pH 6.5 was achieved with Na ϩ followed by K ϩ (5-fold) (Fig. 4D). On the other hand, Li ϩ rather inhibited the uptake in E. coli, most likely due to indirect effects on bacterial metabolism.
The ANTR1 contribution to the P i accumulation in the EANTR1 strain showed a NaCl concentration dependence in the range of 0 to 5 mM, with a maximal (saturation) level at 15 to 25 mM (Fig. 5A). The level of P i uptake in control cells was virtually not changed under these conditions, and at higher NaCl concentrations (50 mM) P i uptake was inhibited in both strains (data not shown). The apparent K m and V max with respect to Na ϩ were 1.17 Ϯ 0.36 mM and 99.15 Ϯ 5.17 nmol mg Ϫ1 protein h Ϫ1 , respectively.
Next, we have studied the effect of P i concentration on the 32 P-labeled uptake in transformed and control cells under inducing conditions. The P i concentration dependence showed a hyperbolic behavior (Fig. 5B) with an apparent K m of 78.7 Ϯ 34 M and a V max of 161 Ϯ 28 nmol mg Ϫ1 protein h Ϫ1 , values that define ANTR1 as a high affinity P i transporter.
To investigate whether ANTR1 specifically transports P i and/or other anions, we measured the effect of various nonlabeled effectors on the P i uptake in EANTR1 and control cells (Table 1). Most efficient competitors in EANTR1 cells were nonlabeled P i and its analogue, methylphosphonate. Replacement of a hydroxyl group by a methyl group in methylphosphonate altered by ϳ80% the uptake levels of 32 P i . The hexose or the aromatic ring in glucose-6-phosphate and phenyl-nitrophosphate affected the uptake only by 40 and 30%, respectively, indicating a relatively rigid site for substrate binding. Other compounds such as pyrophosphate and vanadate were less efficient. Sulfate was found to be the least efficient competitor, showing no substantial influence on the radioactive P i uptake, which confirms P i -related compounds as the main type of substrates. The fact that glutamate competed by 56% the P i uptake in EANTR1 cells indicates that ANTR1 can bind, but not obligatorily transport, this organic anion. When comparing the values obtained in control and transformed cells, it is obvious that the expression of ANTR1 contributes significantly to the competition by both P i and glutamate of the assayed P i uptake.
To verify whether ANTR1 also transports glutamate, we have performed uptake of L-[3,4-3 H]glutamate in control and EANTR1 cells. As shown in Fig. 6A, the uptake proceeded with similar time course and extent in both types of cells, notably 60-fold lower than in the case of P i transport (see Fig. 4A). Taken together, the data from Fig. 6A and Table 1 indicate that glutamate can bind, but is not transported, by ANTR1. Based on the sequence alignment of VGLUT2, NaP i -1 and ANTR1, among the five TMD-located charged residues proposed to be responsible for glutamate but not for P i transport in VGLUTs (12), only the 2 residues in the putative TMD4 (corresponding to Arg-184 and Glu-191 in VGLUT2) are conserved in ANTR1 (Fig. 6B). Poor conservation is also observed in the rabbit NaP i -1, i.e. only the residues corresponding to Arg-88 (TMD1) and Arg-184 (TMD4) in VGLUT2, which adds to the fact that there are no available indications in the literature for glutamate transport mediated by this protein. Because the structural selectivity for P i binding and transport by NaP i are not known, we are unable at the present stage to explain using the sequence information why P i , but not glutamate, is transported by ANTR1 as experimentally demonstrated in this work.

DISCUSSION
Animal NaP i -I proteins were initially characterized as Na ϩdependent phosphate transporters, but more recent studies point to organic as well as inorganic anions as substrates involved in various processes ranging from vesicular storage of the neurotransmitter glutamate to the degradation and metabolism of glycoproteins (10). Arabidopsis ANTRs show 30% identity to the members of the NaP i -I family, but the existence of a selectivity filter for P i is not known and the one for other potential substrates such as glutamate is far from conserved. Tris-succinate buffer, pH 6.5, at the indicated concentrations of NaCl. B, the uptake was carried out as in A but using 25 mM NaCl and the indicated concentrations of 32 Pi. In both panels, the values obtained for EANTR1 cells were subtracted by those for the control strain. Data represent the mean Ϯ S.D. of at least three independent experiments with three replicates. of at least three independent experiments with three replicates. B, sequence alignment of VGLUT2, NaP i -1, and ANTR1 in the TMD1, TMD2, TMD4, and TMD7 regions. The 5 residues important for glutamate uptake activity of VGLUT2 (12) are indicated above the alignment, whereas the identical and conserved residues are indicated below the alignment.

TABLE 1 Effect of various compounds on phosphate transport into control and EANTR1 cells
Uptake of 100 M ͓ 32 P͔orthophosphate was carried out for 3 min in the absence (None) or presence of 2.5 mM nonlabeled effectors (added 2 min before 32 P i ) and stopped by rapid filtration. All assays were performed at pH 6.5. Data are the mean of three independent experiments with three replicates each. S.E. is Ͻ10%. The data in this work demonstrate that the chloroplast thylakoid membrane of A. thaliana contains a major facilitator superfamily/anion:cation symporter member, namely the previously annotated ANTR1 (13). We show that the recombinant ANTR1 is specific for P i transport across E. coli membrane in a Na ϩ -dependent manner. ANTR1 is predicted as a chloroplast protein with a 59-residue transit peptide (Fig. 1A). This protein was exclusively detected in photosynthetic tissues and localized to the chloroplast thylakoid membrane. In Western blotting experiments with two different peptide-specific antibodies, a M r of 45 was consistently indicated (Fig. 2, B and C), which differs significantly from its theoretical size (453 residues, 50.7 kDa). Similar migration shift was observed for the envelope ANTR2 protein (Fig. 2C and Ref. 13), which has a theoretical mass of 49.9 kDa. One possible explanation could be an incorrectly predicted cleavage site. However, the M r of 56 for the immunodetected band corresponding to the recombinant His 6 -Xpress-ANTR1-FLAG protein is also different from its theoretical size (554 residues, 61.2 kDa). Therefore, the most likely reason for the significant shift in the molecular weight of both the recombinant and processed forms of ANTR1 is the high content (60%) of charged and hydrophilic residues, common for this type of solute transporters.

Effector
Available information on NaP i transporters (10,12,26) could be compared with the uptake data for recombinant ANTR1 (Figs. 4 -6). The obtained K m value for P i (78 M) is 10-to 100-fold lower than the values determined in the case of the rabbit NaP i -1 (1 mM, Ref. 10), the human NPT1 (0.29 mM, Ref. 10), and the rat VGLUT2 (10 mM, Ref. 12), indicating a high affinity in addition to the strict transport of P i by ANTR1. The transport kinetics of ANTR1 as studied in E. coli have revealed a dependence of P i transport on external Na ϩ as the driving force with an apparent affinity of 1 mM, i.e. much lower than the determined K m for P i . This may indicate a stoichiometry of Ͼ1 Na ϩ :P i , which is in agreement with the transport mechanism for other NaP i members (26).
An important question is whether the obtained K m values of ANTR1 for P i and Na ϩ in E. coli are physiologically relevant in chloroplast thylakoids. The answer is affirmative and explained below. (i) The K m value in thylakoids, at least in the case of P i , may be the same, if not much lower, than the one obtained in the E. coli system, as in the case of the thylakoid ATP/ADP carrier (6). (ii) To accurately determine the P i concentration in organelles such as chloroplast (stroma) seems to be difficult; thus, values ranging between Յ1 and ϳ10 mM have been reported (27)(28)(29). (iii) When it comes to Na ϩ , its physiological concentration in the stroma is ϳ100 mM (30). (iv) We know very little about the ionic strength of the thylakoid membrane, and the expected concentration of permeable ions (most likely K ϩ , Mg 2ϩ , and Cl Ϫ ) may be Ͻ10 mM (31).
The P i transport activity of mammalian NaP i transporters is dependent on pH (10). The pH pattern obtained for the Na ϩdependent ANTR1 activity in E. coli closely resembles the one reported for other NaP i -I members such as the human NPT1 (32). In our experimental conditions, when the uptake was assayed at pH 5.5, although low, the Na ϩ -independent P i transport system was dominant, and the contribution of a Na ϩ -de-pendent system to the cellular P i uptake activity was progressively increased with increasing pH, reaching its maximum at pH 7.5 (Fig. 4C).
The mammalian NaP i transporters use the inwardly created Na ϩ electrochemical gradient created by the Na ϩ ,K ϩ -ATPase to drive P i import into cells (10). Both H ϩ and Na ϩ are used as the major coupling ions in energy transduction processes in bacteria (33). Across the chloroplast thylakoid membrane an electrochemical gradient of H ϩ is generated as a result of photosynthetic electron transfer reactions. This trans-thylakoid H ϩ gradient not only powers the synthesis of ATP but also acts as a feedback regulatory component in photosynthetic events (Ref. 31 and references therein). The physiological pH in the thylakoid lumen is tightly regulated to a narrow range, 6.5 to 7.0 in darkness and 5.8 to 6.5 under growth light (34). Assuming that the obtained pH profile combined with Na ϩ dependence of ANTR1 in E. coli (Fig. 4C) is also valid in chloroplast thylakoids, then ANTR1 would be a H ϩ -driven P i transporter under acidic (light) conditions, whereas at pH Ն6.5 (darkness) P i transport should be essentially Na ϩ -dependent. The possibility that K ϩ ions can also drive P i transport across thylakoids, as they do in E. coli although with lower efficiency than Na ϩ , cannot be excluded either. The direction of transport does not obligatorily depend on the orientation in the membrane (as predicted in Fig. 1B) but also on the trans-thylakoid H ϩ , Na ϩ , K ϩ , and P i electrochemical gradients. The reason for using a certain cation (H ϩ , Na ϩ , or K ϩ ) must be in the physiological status of the plant or in the conditions it had been experiencing in earlier stages of evolution. The physiological needs of the plant to keep the lumenal pH in the narrow range (34) and the concentration of lumenal ions low (31) should be taken into consideration as well as the availability of metabolic energy.
Indications about the physiological role of ANTR1 in planta are provided by data extracted from the Genevestigator microarray data base (24). ANTR1 expression in Arabidopsis is 7-to 10-fold up-regulated by light of various qualities, visible light of high intensities, and programmed cell death, indicating a putative role in thylakoid biogenesis and turnover, as in the case of the thylakoid ATP/ADP carrier (6). For comparison, the envelope ANTR2 protein is not significantly affected by the above-mentioned stress conditions (data extracted from Genevestigator) and may represent a more ubiquitous plastid housekeeping (P i ) transport system. As for the physiological role(s) of the thylakoid ANTR1 in relation to a P i transport function, there are several possibilities: (i) a pathway to recycle P i produced during nucleotide metabolism in the lumen (5) and to power the chloroplast ATPase; (ii) a pathway to get P i into the lumen, thus lowering its levels in the stroma. ANTR1 activity in thylakoids would be yet another mechanism, in addition to envelope P i transport and sequestration in metabolic intermediate, for P i depletion of stroma, which has been proposed and recently demonstrated to modulate the conductivity of ATPase to protons, thus maintaining a low pH in the lumen and down-regulating photosynthetic light capture (29,35). (iii) Balance of the trans-thylakoid H ϩ electrochemical gradient storage upon change in physiological status of the plant by interacting with cations and thus participating in ion homeostatic mechanisms (34).