Functional Characterization of the Plastidic Phosphate Translocator Gene Family from the Thermo-Acidophilic Red Alga Galdieria sulphuraria Reveals Speciﬁc Adaptations of Primary Carbon Partitioning in Green Plants and Red Algae

In chloroplasts of green plants and algae, CO 2 is assimilated into triose-phosphates (TPs); a large part of these TPs is exported to the cytosol by a TP/phosphate translocator (TPT), whereas some is stored in the plastid as starch. Plastidial phosphate translocators have evolved from transport proteins of the host endomembrane system shortly after the origin of chloroplasts by endosymbiosis. The red microalga Galdieria sulphuraria shares three conserved putative orthologous transport proteins with the distantly related seed plants and green algae. However, red algae, in contrast to green plants, store starch in their cytosol, not inside plastids. Hence, due to the lack of a plastidic starch pool, a larger share of recently assimilated CO 2 needs to be exported to the cytosol. We thus hypothesized that red algal transporters have distinct substrate speciﬁcity in comparison to their green orthologs. This hypothesis was tested by expression of the red algal genes in yeast ( Saccharomyces cerevisiae ) and assessment of their substrate speciﬁcities and kinetic constants. Indeed, two of the three red algal phosphate translocator candidate orthologs have clearly distinct substrate speciﬁcities when compared to their green homologs. GsTPT (for G. sulphuraria TPT) displays very narrow substrate speciﬁcity and high afﬁnity; in contrast to green plant TPTs, 3-phosphoglyceric acid is poorly transported and thus not able to serve as a TP/3-phosphoglyceric acid redox shuttle in vivo. Apparently, the speciﬁc features of red algal primary carbon metabolism promoted the evolution of a highly efﬁcient export system with high afﬁnities for its substrates. The low-afﬁnity TPT of plants maintains TP levels sufﬁcient for starch biosynthesis inside of chloroplasts, whereas the red algal TPT is optimized for efﬁcient export of TP from the chloroplast. form, 200–400 mesh, preequilibrated with 150 m M sodium acetate; Bio-Rad). The ﬂow-through of the anion-exchange columns containing the proteoliposomes with the imported radiolabeled phosphate was quantiﬁed by a liquid scintillation counting. All substrates used for uptake studies were purchased from Sigma-Aldrich. Kinetic constants were determined by measuring the initial velocity of each experiment. Michaelis-Menten constant ( K M ) has been analyzed with at least six external phosphate

In plants, the photosynthetic light reactions provide the energy for major plastid localized pathways, such as CO 2 assimilation, the synthesis of starch, fatty acids, several amino acids, nucleic acids, and the reductive assimilation of inorganic ions like nitrate and sulfate (Weber et al., 2005;Zrenner et al., 2006). To supply the cell and the organism with these primary metabolites, a large number of precursors, end products, and intermediates have to be transported across the organelle envelope membrane and therefore present-day plastids are extensively connected to the cytoplasm by metabolite transporters that reside in the envelope membranes (Tegeder and Weber, 2006;Weber and Fischer, 2007).
Chloroplasts originated approximately 1.6 billion years ago through a single primary endosymbiosis between a nonphotosynthetic primitive mitochondriate eukaryote and a cyanobacterium Bhattacharya et al., 2007;Reyes-Prieto et al., 2007). Within a period of 0.15 billion years, establishment of the plastid and divergence of the three major lineages of the Archaeplastida (Adl et al., 2005), that is the red algae (Rhodophyceae), green algae/land plants (Chloroplastida), and glaucophytes (Glaucophyta), began Yoon et al., 2004). Establishment of the chloroplast within the host cell required massive remodeling of its membrane proteome; novel transport proteins to connect its metabolism with the metabolic network of the host cell had to be acquired Weber and Fischer, 2007). Phylogenetic and phylogenomics analyses recently revealed that a large portion of these plastid-resident transporters is host derived, indicating that integration of the chloroplast with host metabolism was predom-inantly a host-driven process (Tyra et al., 2007). Genes encoding these transporters are not present in extant cyanobacterial genomes but most are conserved throughout the Archaeplastida, indicating they have been established at an early stage during formation of endosymbiosis, likely already at the stage of the protoalga (Tyra et al., 2007).
Of particular importance for establishment of chloroplasts was an efficient and controlled export of photoassimilates from the endosymbiont to its host cell . In the Chloroplastida, a triose-P (TP) translocator (TPT) exports a significant amount of the dihydroxy-acetonephosphate (DHAP) that is generated by the Calvin-Benson cycle to the cytosol; there it mainly serves as precursor for Suc and cell wall biosynthesis (Flü gge, 1999;Reiter, 2002). However, part of the TP is retained in the plastid stroma and used to fuel starch biosynthesis and other intraplastidial biosynthetic pathways (Zeeman et al., 2007). TPT candidate orthologs are highly conserved throughout the photosynthetic eukaryotes , and transport experiments with isolated organelles or reconstituted organellar membranes demonstrated TP transport activity in the red algae and glaucophytes lineages, respectively (Schlichting and Bothe, 1993;Weber et al., 2004). Apparently, TP already served as the main export photoassimilate at the stage of the proto-alga . In the plant model organism Arabidopsis (Arabidopsis thaliana), additional sugar phosphate transporters with specialized functions have been characterized (Flü gge, 1999). Import of phosphoenolpyruvic acid (PEP) from the cytosol by the PEP/phosphate transporter (PPT) drives fatty acid biosynthesis and synthesis of compounds by the shikimate pathway (i.e. aromatic amino acids; Fischer et al., 1997;Knappe et al., 2003b;Voll et al., 2003) and a Glc-6-P/phosphate transporter (GPT) provides Glc-6-P for starch synthesis in heterotrophic plastids (Kammerer et al., 1998;Niewiadomski et al., 2005). Last, a protein closely related to the GPT, the pentose-P/phosphate transporter (XPT), connects the oxidative pentose-P pathways (OPPPs) in cytosol and plastid (Eicks et al., 2002;Flü gge and Gao, 2005).
Phylogenetic analysis showed that candidate orthologs for the PPT and the GPT/XPT type of translocators are present in the genome of the ancient red microalga Galdieria sulphuraria, which is separated from green plants by an evolutionary distance of at least 1 billion years . However, reconstituted membrane fractions from G. sulphuraria showed neither significant PEP nor Glc-6-P transport activity (Weber et al., 2004). This raises the intriguing question as to whether the genetic repertoire present in the last common ancestor of Rhodophyceae and Chloroplastida was functionally maintained or whether it diverged after separation of the two phyla. To address this question, we used G. sulphuraria as a model for the Rhodophyceae. The draft genome of G. sulphuraria is publicly available and the unicellular organism shares core features of carbon metabolism with other red algae (Viola et al., 2001;Barbier et al., 2005a). In contrast to higher plants, starch in red algae is produced in the cytosolic compartment, using UDP-Glc as precursor (Coppin et al., 2005;Patron and Keeling, 2005). Cell wall polysaccharides as well as the major soluble carbohydrate floridoside (a-D-galactopyranosyl-1-2#-glycerol) are produced in the same compartment. Notably, in addition to photoautotrophic growth, G. sulphuraria also is able to grow mixo-or heterotrophically on more than 50 different carbon sources (Gross and Schnarrenberger, 1995;Barbier et al., 2005a). Carbon partitioning thus has to be coordinated not only during alternating light and dark periods but also under continuous heterotrophic growth conditions. It is not known in detail how the carbon allocation between the plastid and the cytosol is accomplished in red algae. Based on phylogenetic data, the plastid phosphate translocators presumably are the major routes for metabolite exchange. They have to balance a high demand for photo-assimilates in the cytosol with maintaining sufficient levels of TPs for rhodoplast-localized pathways.
Here we report the heterologous expression and biochemical characterization of the plastid phosphate transporter family from G. sulphuraria. With regard to higher plants we will discuss an alternative strategy to fine-tune carbon flux across the plastid membrane in photosynthetic eukaryotes.

Molecular Features of the Plastidic Phosphate
Translocator Homologs from G. sulphuraria Three genes encoding proteins with significant similarity to higher plant plastidic phosphate translocators (pPTs) have previously been identified in the genome of the red alga G. sulphuraria . Based on detailed phylogenetic analysis, GsHET39C12 is the candidate ortholog to the plastid TPT and will thus be called GsTPT. GsA14H8 and GsA16F5 cluster with the functionally characterized green plant plastid Glc-6-P/phosphate transport proteins (GPTs) and PEP/phosphate translocator proteins (PPTs), respectively. They were thus assigned with the acronyms GsGPT and GsPPT, respectively. The coding sequences of GsTPT and GsPPT contain 1,224 nucleotides, corresponding to 407 amino acid residues and computed molecular masses of 45.83 kD (TPT) and 44.98 kD (PPT). GsGPT has a calculated molecular mass of 45.45 kD, consisting of 410 amino acid residues that are encoded by 1,233 nucleotides. The N-terminal regions of all three proteins represent putative plastid target sequences that display only low sequence similarity with the pPTs from higher plants (Supplemental Fig. S1) and the red alga Cyanidioschyzon merolae. The G. sulphuraria proteins show an average sequence identity of 37% and a 55% similarity with their Arabidopsis homologs and a slightly higher identity of 48% to the corresponding C. merolae proteins (Supplemental Table S2). Six Lys and two Arg residues that are invariantly embedded in conserved motifs in all functionally characterized pPT proteins to date (Knappe et al., 2003a) are also found in all three G. sulphuraria homologs. Relative to the GsTPT sequence, the positively charged amino acids are at positions K123, K199, R261, K266, K267, K361, R362, and K399 (Supplemental Fig. S1). All three putative transport proteins are highly hydrophobic proteins that are predicted by ConPred II (Arai et al., 2004) to contain nine to 10 membrane spanning a-helices. The GsTPT and GsGPT genes contain one intron that separates a short first exon from the residual coding sequence (Supplemental Table S2). GsPPT has a total of four exons. The last exon covers approximately 50% of the codons. All three sequences have been submitted to GenBank and also can be found at the Michigan State University Galdieria genome database (http://genomics.msu.edu/ galdieria). Accession numbers and annotations are listed in Supplemental Table S2.

Heterologous Expression of GsTPT, GsPPT, and GsGPT
To assess the substrate specificities of the G. sulphuraria phosphate translocators (PTs), we cloned the corresponding cDNAs into the yeast (Saccharomyces cerevisiae) expression vector pYES/NT under control of the Gal-inducible GAL4 promoter. The regions of the cDNAs encoding for the putative target sequences of each protein were removed and instead fused to an N-terminal hexa-His tag. After transformation of the corresponding constructs into the yeast strain INVSc1, all three pPT homologs could be successfully expressed and accumulated in the membrane fraction (Supplemental Fig. S2). Immunoblot analysis with an anti-poly-His tag antibody verified the Gal-inducible accumulation of the pPT proteins (Supplemental Fig.  S2, lane 2-4) compared to controls, which maintained the empty expression vector (Supplemental Fig. S2, lane 1). The calculated molecular masses of the N-terminal His-tagged proteins were 41.6 kD, 45.2 kD, and 41.2 kD for GsTPT, GsPPT, and GsGPT, respectively. The presence of recombinant protein was verified for all biological replicates by western blot before reconstitution.
Functional Analysis of Three Putative Transport Proteins from G. sulphuraria All functionally characterized pPTs of higher plants catalyze, in addition to their characteristic substrates, a strict homo-exchange of orthophosphate (Pi) (Flü gge, 1999). To test whether the yeast-expressed, recombinant GsPTs are functional, we thus first examined their ability to catalyze the signature Pi homo-exchange.To this end, they were reconstituted into liposomes that were preloaded with 30 mM Pi (i.e. the liposomes contained 30 mM Pi inside). Then radiolabeled [ 32 P]phosphate was added to the liposomes and the uptake kinetics were recorded. In GsTPT and GsPPT containing proteoliposomes, protein-mediated isotope equilibrium was reached within 10 min, using an external Pi concentration of 0.25 mM (Fig. 1). GsGPT mediated Pi uptake equilibrated after 15 min with a 4 times higher external [ 32 P]Pi concentration of 1 mM. Pi exchange followed a first-order kinetic with an equilibrium plateau (V max ) of 400 6 13.74, 187.3 6 3.31, and 237.1 6 7.02 nmol mg protein 21 and initial rates of 60.69 6 2.48, 30.94 6 0.43, and 23.23 6 1.27 nmol mg protein 21 min 21 for GsTPT, GsPPT, and GsGPT, respectively. Control membranes from yeast cells transformed with an empty pYES/NT vector or boiled recombinant proteins showed negligible accumulation rates over time (data not shown). Pyridoxal-phosphate and 4,4#-diisothiocyanostilbene-2,2#-disulfonic acid also effectively inhibited GspPT-mediated import of [ 32 P]phosphate (data not shown). Marginal uptake of radiolabeled [ 32 P]phosphate compared to control experiments was measured when vesicles did not contain Pi as a countersubstrate ( Fig. 1). In summary, all putative phosphate transport proteins from G. sulphuraria were functionally expressed in yeast and are able to catalyze the signature Pi homo-exchange activity, similar to their higher plant homologs.

Transport Properties of GsTPT, GsPPT, and GsGPT
To assess the substrate specificity of the recombinant GspPT proteins, vesicles were preloaded with saturating concentrations (i.e. 30 mM) of various counter-  exchange substrates and initial rates of [ 32 P]Pi uptake were determined.

GsPPT
Reconstituted GsPPT protein efficiently used Pi and PEP as a countersubstrate for the import of radiolabeled [ 32 P]phosphate (Fig. 2B). Eighteen additional substrates, as listed above for the GsTPT, showed marginal initial Pi uptake rates of less than 30% of the Pi/Pi homo-exchange rate ( Fig. 2B; data not shown). GsPPT has an apparent K M value of 0.76 6 0.075 mM for Pi and a K i value of 0.36 6 0.04 mM for PEP. Compared to PEP, 11 times higher 3-PGA and 22 times higher DHAP levels, respectively, were needed to inhibit Pi uptake by 50% (Table I). Glc-6-P had no affinity for the Pi binding site.

GsGPT
Vesicles reconstituted with the GPT homolog from G. sulphuraria were preloaded with the same 20 substrates as given above; in all cases the [ 32 P]Pi uptake was much lower than for liposomes preloaded with 30 mM Pi (Fig. 2C). GsGPT has rather low affinity to Pi (K M , 5.07 6 0.4 mM; Table I). Compared to nonpreloaded proteoliposomes (13.03% 6 5.9%), approximately 3-to 4-fold higher uptake rates were measured when liposomes were preloaded with either 3-PGA or DHAP. Importantly, however, even high concentrations of 3-PGA and DHAP did not significantly inhibit Pi homoexchange (Table I). Glc-6-P and PEP are nonrelevant substrates for the GsGPT protein.
Transcript Levels and Protein Activity of the pPT Homologs in G. sulphuraria Cells In land plants, TPT expression is confined to photosynthetic tissues whereas GPT expression is highest in heterotrophic tissues (Kammerer et al., 1998;Flü gge, 1999;Niewiadomski et al., 2005). PPT genes are expressed at low levels in various tissues (Knappe et al., 2003b). To determine the expression pattern of GspPTs, semiquantitative reverse transcription-PCR was used. Figure 3 shows that GsTPT, GsPPT, and GsGPT transcripts are clearly detectable and abundant under both Uptake experiments were initiated with 0.25 mM [ 32 P]Pi (A and B) or 1 mM [ 32 P]Pi (C) and terminated after 1.5 min (GsTPT), 2 min (GsPPT), or 4 min (GsGPT). As negative control, proteoliposomes were preincubated (2 min) with inhibitor stop solution to calculate net uptake activity. Relative uptake activities were compared to the Pi/Pi counter-exchange experiment, which was set to 100%. Data are summarized as the arithmetic mean 6 SE of at least three independent experiments. Glc6P, Glc-6-P; Glc1P, Glc-1-P; Frc6P, Fru-6-P; Gal1P, Gal-1-P; Gly3P, Gly-3-P.
autotrophic and heterotrophic growth conditions. The steady-state transcript levels of all three genes were higher than those of the reference gene actin (Gs15600; Galdieria Genome Browser Build 3.0). Liposomes reconstituted with total membrane fractions exhibit high [ 32 P]Pi import using DHAP as a countersubstrate under both culturing regimes. Exchange rates with PEP and 3-PGA as counter-exchange substrates were much lower as compared to DHAP. Exchange of Glc-6-P with Pi was close to the background values of unloaded vesicles.

DISCUSSION
In this study, we report the molecular and biochemical characteristics of the pPT family from the unicellular red alga G. sulphuraria. All three members of the GsPT gene family were heterologously expressed in yeast, functionally reconstituted into liposomes (Supplemental Fig. S1; Fig. 1), and their substrate specificities and kinetic constants were assessed.
The G. sulphuraria Genome Encodes a High-Affinity TPT with Narrow Substrate Specificity Recombinant G. sulphuraria TPT mediates a strict counterexchange of the TP DHAP with Pi (Fig. 2). Interestingly, the kinetic constants of GsTPT considerably differ from those of the green plant TPT (Table I). The red algal transporter exhibits a 3 times higher affinity for Pi and, as indicated by its low K i value, the substrate DHAP has a 2-fold higher affinity to the Pi binding site. Further, GsTPT is highly specific for DHAP. In contrast to the green plant ortholog, 3-PGA poorly acts as a countersubstrate, even under saturating conditions ( Figs. 2A and 3). In addition, 3-PGA is a weak competitive inhibitor of Pi import with an 8-fold and 15-fold higher K i value as compared to the TPT from plants and the K i (DHAP) constant from GsTPT, respectively. These results argue strongly against 3-PGA as a physiological relevant substrate of GsTPT. PEP and Glc-6-P were even less efficient exchange substrates and inhibitors of Pi transport ( Fig. 2; Table   I). Very likely, the distinct kinetic properties of GsTPT represent an important adaptation to red algal carbon metabolism. In contrast to the Chloroplastida, red algae store an insoluble starch-like polymer called floridean starch and synthesize cell wall building blocks and the main soluble carbohydrate floridoside, composed of UDP-Gal and Gly-3-P moieties in their cytosol (Viola et al., 2001;Collen et al., 2004;Barbier et al., 2005a). Hence, primary carbon partitioning in red algae is predominantly organized within the cytosol and GsTPT enables solely the export of TPs, even at low stromal concentrations, to cope with the massive cytosolic demand of recently fixed carbon (Fig. 4).
On the other hand, GsTPT is highly expressed and active in heterotrophically cultured G. sulphuraria cells (Fig. 3). Under these conditions, GsTPT presumably mediates TP import into the plastid, thus mediating carbon flux between cytosol and nonphotosynthetic rhodoplasts. In contrast, land plants export TP from photosynthetic chloroplasts via TPT and import Glc-6-P into heterotrophic plastids via GPT. The TPT is an integral part of photoassimilate partitioning in plants (Häusler et al., 2000;Smith and Stitt, 2007). Accumulation of Suc and hexose-Ps in the cytosol sequester cytosolic Pi and thus slow down the export of TP. A resulting increased stromal 3-PGA/Pi ratio allosterically activates ADP-Glc pyrophosphorylase, which is the committing step of plastidic starch biosynthesis (Ballicora et al., 2004). Plants have to maintain sufficiently high levels of TP in the plastid stroma to enable both the regeneration of the CO 2 acceptor ribulose-1,5-bisP and starch biosynthesis during the light period (Zeeman et al., 2007). The apparent lower affinity of the plant transporter for Pi and TP thus ensures higher steady-state levels of TP in the stroma to drive these reactions.

G. sulphuraria Has a PPT with Similar Properties as Its Green Plant Ortholog
PEP uptake from the cytosol is required for stromalocalized fatty acid and shikimate biosynthesis (Flü gge, 1999). Previously published results have been ambig- uous about the existence of a putative PPT in G. sulphuraria. Phylogenetic analysis revealed an orthologous candidate gene , although isolated membrane fractions did not exhibit pronounced PEP uptake activity (Weber et al., 2004). Here, we show that EST GsA16F5  represents a highly conserved PPT with almost identical kinetic constants in both the green and red lineages (Table I). Both transporters are highly specific for Pi and PEP and poorly accept 3-PGA, DHAP, or any other of the 16 tested metabolites as a substrate ( Fig. 2; data not shown). In addition, we employed an improved reconstitution method for total membranes from autotrophic and heterotrophic cell cultures that yielded lower background and a weak but clearly detectable PEP/Pi counter-exchange activity (Fig. 3). The low PEP transport capacity (Fig. 3) mirrors published data from chloroplasts of C 3 land plants (Flü gge and Weber, 1994). TPT is by far the most abundant transport protein in C 3 chloroplasts and accounts for a high DHAP exchange capacity (Flü gge, 1999). Fatty acid biosynthesis and the shikimate pathway in red algae are predicted to be similar to those of seed plants and are presumably localized in the plastid, based on phylogenetic analyses and N-terminal target sequence predictions (Weber et al., 2004;Richards et al., 2006).
A question that cannot be conclusively answered at the moment is whether rhodoplasts, analogous to chloroplasts of green plants, also have negligible activities of plastid phosphoglyceromutase and enolase to produce PEP from plastidial 3-PGA (Trimming and Emes, 1993). The G. sulphuraria genome encodes for two phosphoglyceromutase isozymes (Gs04140 and Gs52680) and a single enolase (Gs21490), which are all three computationally predicted as cytosolic enzymes (Nielsen et al., 1997) without any N-terminal targeting sequences, when compared with their homologous genes from Arabidopsis (phosphoglyceromutase: At4g09520, At3g08590, At1g09780; enolase: At2g36530, At2g29560, At1g74030; Friso et al., 2004;Larkin et al., 2007). Alternatively, PEP could be generated from pyruvic acid via the pyruvate Pi dikinase (PPDK) reaction. The annotated PPDK gene Gs42070 is computationally predicted to be localized in the cytosol, exhibiting a high similarity with cyanobacterial PPDK genes, and is not phylogenetically related to the dual targeted enzyme from Arabidopsis (At4g15530; Parsley and Hibberd, 2006). Although computational targeting predictions have to be taken with a grain of salt, bioinformatics analysis indicates that conversion of TP to PEP is not possible in G. sulphuraria, and rhodoplasts thus depend on PEP import from the cytosol to drive PEP-depended reactions in the stroma (Fig. 4).

G. sulphuraria Does Not Possess a Plastidic Hexose-P Importer
Reconstituted recombinant putative GsGPT mediated a strict Pi/Pi homo-exchange that followed a firstorder rate kinetics (Fig. 1). An unusually high K M value of 5 mM for Pi (Table I) indicates that GsGPT under physiological conditions is not able to compete with GsTPT and GsPPT for the common substrate Pi. Although liposomes preloaded with high (i.e. saturating) concentrations of DHAP or 3-PGA (i.e. 30 mM internal substrate concentration) showed Pi uptake rates of approximately 50% of Pi/Pi homo-exchange, it is important to notice that neither 3-PGA nor DHAP significantly inhibited Pi/Pi homo-exchange either at physiological concentrations or at 100-fold excess (Table I). The prime substrate candidate Glc-6-P was poorly accepted under saturating conditions (Fig. 2) Figure 3. Transcript abundance and protein activity of GsTPT, GsPPT, and GsGPT in G. sulphuraria cells. Top, Total RNA from auto-and heterotrophically cultured G. sulphuraria cells was isolated, treated with DNase, and used for first-strand cDNA synthesis. Expression of GsTPT, GsPPT, GsGPT, and actin transcripts was amplified by PCR with 25, 30, or 35 elongation cycles and visualized on an ethidium bromidestained 1.25% (w/v) agarose gel. Bottom, Relative [ 32 P]Pi uptake rates into liposomes reconstituted with total G. sulphuraria membranes isolated from heterotrophically (black bars) or autotrophically (white bars) grown cells, respectively. Membrane fractions were reconstituted into liposomes preloaded with 30 mM various substrates. Experiment was performed as in Figure 2. and it did not inhibit Pi import in the Pi/Pi homoexchange setup (Table I). These results strongly indicate that 3-PGA, DHAP, and Glc-6-P are not physiologically relevant substrates of the putative GsGPT. We also tested various additional metabolites, such as nucleotide sugars, mononucleotides, hexose-Ps (Glc-1-P, Fru-6-P, and Gal-1-P), pentose-Ps, and a precursor of floridoside biosynthesis, Gly-3-P. All of these are poorly exchanged with [ 32 P]Pi and the physiological substrate of GsGPT thus remains unidentified.
In nongreen plastids of plants, Glc-6-P is the preferred precursor for starch synthesis and NADPH generation via the OPPP due to the absence of a Fru-1,6-bisphosphatase (FBPase) activity (Flü gge, 1999). In photosynthetic tissues, plastidic FBPase is inactivated at night and thus hexose-Ps cannot be generated from TPs (and vice versa). In contrast, red algal starch biosynthesis is cytosolic and plastidic FBPase from G. sulphuraria is not subject to a strict redox regulation (Reichert et al., 2003;Oesterhelt et al., 2007). Importing TPs via the GsTPT could thus sustain the production of hexose-P for carbon and NADPH supply in the rhodoplast during the night or prolonged heterotrophic growth conditions, thus bypassing the requirement for a plastidic hexose-P translocator (Fig. 4).
An Operative DHAP/3-PGA Reduction Shuttle Is Unlikely in G. sulphuraria 3-PGA is a major substrate for the TPT in higher plants (Flü gge, 1999). In spinach (Spinacia oleracea) and other model plants, 3-PGA levels in the plastid and the cytosol frequently exceed DHAP levels by severalfold (Gerhardt et al., 1987). However, the affinity of the spinach TPT is identical for both metabolites. It has been proposed that in vivo a significant amount of 3-PGA transport occurs across the envelope membrane; the TPT could thus operate as an NADPH reduction equivalent shuttle between stroma and cytosol (Heineke et al., 1991). The substrate specificity and the kinetic constants of recombinant proteins, as discussed in detail earlier, and total membrane fractions from G. sulphuraria presumably cannot sustain a physiological relevant exchange rate of 3-PGA across the rhodoplasts envelope membrane. Still, DHAP offers a possibility to export reduction equivalent to the cytosol. The nonphosphorylating glyceraldehyde-3-P dehydrogenase (Rius et al., 2006), which is also present in the draft genome sequence of the red alga (Gs00840) or the combined glycolytic reaction sequence of glyceraldehyde-3-P dehydrogenase and 3-PGA kinase (Plaxton, 1996) generates NADPH, or NADH and ATP, respectively. However, in contrast to green plants, not 3-PGA but Pi must be postulated as the alternative countersubstrate for DHAP.

Impact of Subtle Differences of TPT in Chloroplastida and Rhodophyta
TP is the main entry and branching point in the glycolytic network in photosynthetically active cells (Plaxton, 1996). Metabolic flux has to be regulated between generating hexose-Ps via gluconeogenesis and glycolytic "downstream" products, such as 3-PGA, PEP, and pyruvic acid. In higher plants, the redox transfer by a TP/3-PGA shuttle via nonphosphorylating glyceraldehyde-3-P dehydrogenase and 3-PGA kinase is proposed to be marginal and glyceraldehyde-3-P dehydrogenase expression is mainly induced under stress conditions, such as anaerobiosis and heat shock or Suc feeding (Yang et al., 1993;Bustos et al., 2008;Holtgrefe et al., 2008). On the other hand, glycolysis sulphuraria. In photosynthetically active cells, TP is exported via the GsTPT from rhodoplasts to sustain soluble and insoluble carbohydrate synthesis (floridoside and floridean starch) in the cytosol. Under heterotrophic growth conditions the same transporter supplies the rhodoplasts with TP for, e.g. hexose-P synthesis. Rhodoplasts are not capable of producing PEP via partial glycolysis and are thus dependent on the import of PEP from the cytosol via the PPT protein for fatty acid and aromatic amino acid biosynthesis. The red algal plastid envelope does not contain any 3-PGA and Glc-6-P transport activities, which is a specific trait of green plants. Frc-1,6BP, Fru-1,6-bisP; Ru-1,5BP, ribulose-1,5-bisP; Gly3P, Gly-3-P. has to provide adequate amounts of PEP to replenish TCA intermediates used for biosynthesis of, e.g. amino acids, to produce pyruvic acid for mitochondrial metabolism, or to fuel plastid fatty acid biosynthesis (Plaxton, 1996). Because 3-PGA transport is seemingly a specific adaptation in plants, we hypothesize that 3-PGA transport across the chloroplast envelope membrane might provide a bypass of the complex regulation of the cytosolic TP pool that permits supporting cytosolic PEP biosynthesis and serves under stress conditions as an additional redox shuttle system.

CONCLUSION
Although the primary structures of pPTs are highly conserved between green plants and red algae, these proteins have evolved quite distinct biochemical characteristics in the different lineages of the Archaeplastida. Comparative biochemical analysis of candidate orthologs from distantly related organisms thus provides novel insights into alternative modes for regulating "ancient" metabolic pathways. While the TP/ phosphate antiport activity across the plastid envelope membrane is highly conserved across all photosynthetic eukaryotes analyzed to date, the red algal transporters catalyze neither 3-PGA nor Glc-6-P transport (Fig. 4).
G. sulphuraria must have evolved alternative mechanisms to distribute TP into various pools (i.e. floridean starch, floridoside, and fatty acid and amino acid synthesis), such as less stringent redox control of Calvin cycle, OPPP, and glycolysis (Oesterhelt et al., 2007). Plastidial phosphate translocators thus represent crucial components of primary carbon portioning in higher plants and red algae that have evolved to the specific requirements of each lineage through modulation of substrate specificities and kinetic constants.

Growth and Sampling of Galdieria sulphuraria Cells
Galdieria sulphuraria strain 074W (Gross et al., 1999) was cultured autotrophically in a minimal salt medium at 37°C with a light intensity of 80 mmol photons m 22 s 21 . The culture was grown at pH 2.0 in Erlenmeyer flasks under vigorous shaking at ambient air conditions. Heterotrophic growth of the algal cells was performed at 37°C in culture flasks in complete darkness in the identical salt medium containing 25 mM Glc as sole carbon source. Cells were harvested at the late logarithmic phase by centrifugation (3,000g, 5 min, 4°C), washed with 13TE buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA), frozen in liquid nitrogen, and stored at 280°C or directly used for nucleic acid extraction (Barbier et al., 2005b) or total membrane enrichment.
Isolation of Genomic DNA and Total RNA from G. sulphuraria Isolation of genomic DNA was performed as previously described (Barbier et al., 2005b). Briefly, cells were ground in liquid nitrogen and proteins were removed by phenol/chloroform/isoamylalcohol (PCI) extraction. Nucleic acids were precipitated by ethanol and RNA was removed by a DNAse-free RNAse treatment and purified by an additional PCI and ethanol precipitation step. Total RNA was isolated with an acid guanidium isothiocyanate-phenol/ chloroform solution as described by Chomczynski and Sacchi (1987).

Expression of GsPTs in Yeast
The coding sequences of all three GsPT cDNAs were amplified from G. sulphuraria cDNA by PCR (Platinum Pfx polymerase; Invitrogen) using the primer combinations listed in Supplemental Table S1. PCR products were subcloned into the pGEM-T Easy vector system (Promega) and sequenced. Forward and reverse primers for the genes GsHET39C12 and GsA16F5 were designed with BamHI and XhoI restriction recognition sites, respectively. The GsA14H8 specific forward primer had a KpnI and the reverse primer an XhoI restriction site. Each cDNA was ligated in frame with an N-terminal poly-His tag into the yeast (Saccharomyces cerevisiae) expression vector pYES2/NT (Invitrogen). Standard molecular methods were applied for DNA restriction and cloning (Sambrook et al., 1995).
The resulting constructs were transformed into competent yeast INVSc1 cells (Invitrogen). Selection, maintenance of the transformants, and Galinducible expression of the recombinant proteins were done according to the manufacturer's instructions (pYES2/NT expression system; Invitrogen). Preparation of yeast membranes containing heterologously expressed GsHET3912, GsA16F5, and GsA14H8 proteins, respectively, was done as described previously (Bouvier et al., 2006), except that the expression was induced at an OD 600 of 0.6 in the presence of 2% Gal and cells had been cultured for 8 h at 30°C. Presence of recombinant proteins was verified by standard SDS-PAGE and immunoblot analysis (Sambrook et al., 1995) using an anti-penta-His antibody (Qiagen) and a secondary anti-mouse IgG antibody conjugated with an alkaline phosphatase (Promega).

Reconstitution into Liposomes and Transport Activity Assays
Acetone-washed L-a-phosphatidylcholine (Sigma-Aldrich) had been sonicated (5 min on ice, Branson Sonicator 250, output 2, duty cycle 30%) to a final concentration of 2% (w/v) in 120 mM Tricine-KOH (pH 7.5) and 30 mM internal substrate. Yeast membrane fraction (50 mL) was reconstituted with 950 mL liposome suspension by the freeze-thaw procedure (Kasahara and Hinkle, 1977). After thawing, the proteoliposomes were pulsed 20 times on ice to yield unilamellar vesicles. PD10-desalting columns (GE Healthcare) were preequilibrated with 150 mM Tricine-KOH (pH 7.5) and used to remove nonincorporated substrate from the external medium. If not stated otherwise, all transport studies were initiated with radiolabeled [ 32 P]Pi (GE Healthcare) at a final concentration of 0.25 mM. For each data point, 190 mL proteoliposomes were terminated with 10 mL inhibitor stop solution (200 mM pyridoxal-phosphate, 20 mM 4,4#-diisothiocyanostilbene-2,2#-disulfonic acid, 100 mM Tricine-KOH, pH 8.0). Control experiments with membranes from yeast cells harboring the empty vector were performed in parallel. The addition of inhibitor mix before adding radiolabeled substrate was used to monitor unspecific binding. External phosphate was removed by strong anion-exchange chromatography with AG-1 X8 resin (acetate form, 200-400 mesh, preequilibrated with 150 mM sodium acetate; Bio-Rad). The flow-through of the anion-exchange columns containing the proteoliposomes with the imported radiolabeled phosphate was quantified by a liquid scintillation counting. All substrates used for uptake studies were purchased from Sigma-Aldrich. Kinetic constants were determined by measuring the initial velocity of each experiment. Michaelis-Menten constant (K M ) has been analyzed with at least six external phosphate concentrations ranging between 0.05 and 10 mM and competitive inhibition of Pi transport was assessed by the inhibitor constant K i as described by Dixon (1953). GraphPad-Prism software was used for nonlinear regression fitting of all enzyme kinetic data.

Reverse Transcription-PCR Gene Expression Study
DNAse-treated RNA was used for first-strand cDNA synthesis (Super-Script II first-strand synthesis system; Invitrogen). Oligonucleotide primers for expression studies of the G. sulphuraria genes are summarized in Supplemental Table S1.

Other Methods
Protein concentrations were determined with a standard Bradford assay (Bio-Rad). Proteins in the membrane fractions were delipidated by TCAacetone extraction before determination (Shultz et al., 2005).
Sequence data from this article can be found in the GenBank/EMBL data libraries under the following accession numbers: GsTPT, EU853171; GsPPT, EU853172; and GsGPT, EU853173.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S2. Expression of pPT homologs from G. sulphuraria in yeast.
Supplemental Table S1. Oligonucleotide primers designed in this study.