A cDNA encoding Pht2;1 from potato, a new member of the plant Pht2 gene family of low-affinity orthophosphate (Pi) transporters, was isolated. The expression pattern of the corresponding gene as well as its ortholog from Arabidopsis was analyzed and the encoded proteins were localized in the two plants. Pht2;1 expression is strongly upregulated by light in potato and Arabidopsis leaf tissue. RNA gel blot analysis, reverse transcription-polymerase chain reaction (RT-PCR), promoter/GUS, and protein/green fluorescent protein (GFP) fusion studies, respectively, indicate that the gene is expressed in both auto- and heterotrophic tissues and its encoded protein is localized to the plastids. The similar patterns of Pht2;1 gene regulation in potato and Arabidopsis prompted us to screen publicly available gene expression data from 228 Arabidopsis oligonucleotide microarrays covering 83 different experimental conditions. Modulation of Pht2;1 transcript levels was overall moderate, except for a limited number of experimental conditions where Pht2;1 mRNA concentrations varied between 2- and 3.7-fold. Overall, these analyses suggest involvement of the Pht2;1 protein in cell wall metabolism in young, rapidly growing tissues, independent of other Pi transporters such as the high-affinity Solanum tuberosum Pi transporter 1 (StPT1). Cluster analysis allowed identification of colinear or antiparallel expression profiles of a small set of genes involved in post-translational regulation, and photosynthetic carbon metabolism. These data give clues about the possible biological function of Pht2;1 and shed light on the complex web of interactions in which Pht2;1 could play a role.
The initial step in the utilization of the macronutrient phosphorus in cellular metabolism is its uptake into the cell as orthophosphate (Pi). Pi transport across membranes is mediated by integral membrane proteins, which are specific for the Pi anion (Daram et al., 1998, 1999; Leggewie et al., 1997). A screen for cDNAs encoding Pi transporters in potato had previously resulted in the identification of three Pi transporters, i.e. Solanum tuberosum Pi transporter 1 (StPT1), StPT2, and StPT3 (Leggewie et al., 1997; Rausch et al., 2001), highly homologous to the members of the Pi transporter Pht1 family and high-affinity Pi transporters of fungal origin (Mudge et al., 2002; Rausch and Bucher, 2002). While StPT1 transcripts are detectable in both root and shoot at high abundance during both high and low Pi conditions, the StPT2 gene is induced exclusively in roots at limiting Pi concentrations, while StPT3 transcripts are abundant in roots colonized by arbuscular-mycorrhizal fungi. Based on expression patterns, the potato Pht1 members appear to be predominantly involved in Pi uptake at the root–soil interface, or at the root–fungus interface in case of the arbuscular-mycorrhizal symbiosis. After uptake into roots, Pi is mainly translocated symplastically to the xylem parenchyma cells, and secretion into the xylem for long-distance translocation to the above-ground organs is mediated by another type of transporter-like protein (Hamburger et al., 2002; Marschner, 1995). In Pi-sufficient plants, most of the Pi absorbed by the root is transported in the xylem to growing leaves, which are strong sinks for this anion. In Pi-deficient plants, however, the restricted supply of Pi from the roots to the shoot is augmented by increased mobilization of stored Pi in the older leaves and re-translocation to both younger leaves and growing roots, from where Pi can again be recycled to the shoot (Jeschke et al., 1997). StPT1 expression in aerial parts at relatively high levels, and the differential pattern of Pht1 gene expression in above-ground tissues of Arabidopsis thaliana (Karthikeyan et al., 2002; Mudge et al., 2002), potato (Leggewie et al., 1997), and tomato (Daram et al., 1998; Liu et al., 1998) provides support for the involvement of Pht1 family proteins in Pi allocation in the shoot.
To maintain cellular Pi homeostasis, a tight control of the Pi concentration in each compartment of the cell is required (Mimura, 1999). Photosynthesis and carbon partitioning in the light–dark cycle are strongly affected by cytosolic and chloroplastidic Pi concentrations (Marschner, 1995). The role of Pi in carbon partitioning has been clearly demonstrated with isolated chloroplasts (Flügge, 1999; Heldt, 1977). Both processes require Pi transport across the chloroplast inner envelope membrane. A variety of inner membrane Pi translocators mediate the exchange of carbon compounds with Pi or triose-phosphates, respectively (Flügge, 1999). Pi can be transported into the chloroplast via the triose-phosphate/phosphate translocator (TPT). The TPT is expressed almost exclusively in photosynthetic tissue. Three additional groups of phosphate translocator (PT) proteins have been identified in the inner envelope membrane of plastids, and all of them function as antiport systems using Pi or phosphorylated C3 and C6 compounds as counter-substrates (Eicks et al., 2002). The phosphoenolpyruvate (PEP)/PT (PPT) mediates the transport of mainly PEP into the organelles, where it is used as a substrate for the shikimate pathway. In contrast to the TPT, the PPT transports Pi out of the chloroplasts into the cytosol, as does the glucose 6-phosphate (Glc-6-P)/PT (GPT), which imports Glc-6-P into non-photosynthetic plastids in exchange for either triose-phosphate or Pi (Flügge, 2002). The fourth member of the PT protein family translocates xylulose 5-phosphate (Xul-5-P) in exchange for triose-phosphate or Pi into plastids. Xul-5-P/PT (XPT) is assumed to provide the plastidic pentose phosphate pathway with carbon skeletons when the demand for intermediates of the cycle is high (Eicks et al., 2002).
Arabidopsis thaliana Pht2;1 (ARAth;Pht2;1) was the first member of the Pht2 family of plant Pi transporters to be identified (Daram et al., 1999). Despite the homologies to fungal and mammalian Na+-dependent Pi transporters, functional analysis of the Pht2;1 protein in mutant yeast cells indicated that it is an H+/Pi symporter dependent on the electrochemical gradient across the yeast plasma membrane. Here, we report on the characterization of the Arabidopsis Pht2;1 ortholog from potato, designated accordingly Solanum tuberosum Pi transporter 1 (SOLtu;Pht2;1). Evidence is provided for developmental and light-dependent control of the Pht2;1 gene and for localization of Pht2;1 to both photoautotrophic and heterotrophic plastids in potato and Arabidopsis. Our data and recently published work (Ferro et al., 2002; Versaw and Harrison, 2002) provide molecular evidence for light-regulated unidirectional Pi transport into plastids mediated by Pht2;1. Possible roles of Pht2;1 deduced from expression analysis in both potato and Arabidopsis and combined with genome-wide expression analysis in Arabidopsis are discussed.
Source–sink interactions for phosphate in potato
The export of Pi from leaves of different age and Pi re-allocation on a whole-plant level was investigated using 32Pi as a tracer (Figure 1). When loaded onto the youngest accessible leaf, 32Pi remained in the apical part of the shoot during the 6-h duration of the experiment (Figure 1a), reflecting net import of Pi, while 32Pi loaded to an older leaf was re-allocated to younger leaves and the root, in particular the root tips. Weaker radioactive signals were detected in the oldest leaves. This indicates Pi efflux from mature leaves and influx into roots and predominantly the younger leaves, both representing strong Pi sinks.
In addition to 32Pi translocation on the whole plant level, Pi uptake and distribution in detached leaves was investigated (Figure 1b). Leaf 1 indicates the youngest detachable leaf, while leaf 8 indicates a fully expanded, mature leaf from the same plant. Leaves in between are successively numbered leaves 2–7. The signal on the X-ray film indicated that the detached leaves 2 and 3 accumulated the largest amounts of 32Pi, while leaf 1 from the shoot apex and older leaves absorbed less of the tracer. Thus, in potato, similar to other plant species, young leaves and root tips act as strong sinks for Pi, while old tissues represent a Pi source.
Cloning of Pht2;1 from potato and computational sequence analysis
The SOLtu;Pht2;1 cDNA (Pht2;1; AY603690) was cloned by screening of a potato leaf cDNA library using a 690-bp fragment from the 3′-coding region of Arabidopsis Pht2;1, which shared similarity with the 3′ regions of type III Pi transporter genes from animals (Daram et al., 1999). A cDNA of 2122 bp in length encoding a putative protein of 581 amino acids was finally isolated and sequenced. blast searches yielded several expressed sequence tags (ESTs) from tomato (Accession numbers: BG126474; BE433007; AWO39864; AWO39344) and potato (BE472326; BI176349) identical or highly homologous, respectively, to Pht2;1. Significant homology to full-length cDNA sequences from plants was found for four cDNAs, i.e. a putative Pi permease from Mesembryanthemum crystallinum (U84890), ARAth;Pht2;1 (AJ302645), PHT2-1 from Medicago truncatula (AF533081), and Phtc from Spinacia oleracea (AJ491150), both published during the course of this work. The predicted 61-kDa potato Pht2;1 protein has a long hydrophilic N-terminus and 12 transmembrane domains (TMs) interrupted by a large hydrophilic loop between TM8 and TM9 (Figure 2). The isoelectric point of the protein was calculated to be 9.25. Protein sequence alignments show that the five plant members of the Pht2 family are highly conserved among each other (Figure 2). The highest degree of divergence can be found along the hydrophilic N-terminus and the hydrophilic loop between TM8 and TM9. Within the N-terminus, three sites for serine/threonine (Ser/Thr) phosphorylation or N-linked glycosylation are conserved between all five plant Pht2 Pi transporters (Figure 2). Although experimental evidence is presently lacking, this suggests a possible post-translational modification of Pht2;1.
Pi transporter expression in various potato organs
Genomic DNA gel blot analysis indicated that Pht2;1 is a single or low-copy number gene in the haploid potato genome, similar to Pht2;1 from Arabidopsis, containing at least one intron (data not shown). RNA gel blot analysis revealed the presence of the Pht2;1 transcript in above ground organs at varying abundance, while transcripts corresponding to the previously described Pi transporter StPT1 (Leggewie et al., 1997) was present in all plant organs analyzed, including roots and tubers (Figure 3a). Pht2;1 is preferentially expressed in young leaves and floral buds. In contrast to Pht2;1, StPT1 transcripts are hardly detectable in young leaves, while mRNA abundance is higher in mature leaves. The presence of Pht2;1 and StPT1 transcripts was additionally analyzed in different organs of mature flowers (Figure 3b). Although the expression patterns of the two Pi transporters differ on the whole plant level, parallel, strong expression is found in the flower organs. Gene-specific reverse transcription-polymerase chain reaction (RT-PCR) analysis clearly demonstrated the presence of Pht2;1 transcripts in developing and mature tubers and roots (Figure 3c). In the same experiment, two bands were detected on the gel when Pht2;1 cDNA and potato genomic DNA, respectively, were used as templates, indicating the presence of an intron sequence of approximately 1.2 kbp in length in the amplified region of the potato Pht2;1 gene. Accordingly, StPT1 was also expressed in tubers and roots (Figure 3c). Similar results for Pht2;1 were obtained when RT-PCR was performed with Arabidopsis and primers specific for ARAth;Pht2;1 (Figure 3d). Signals were obtained with RNA extracted from both leaves and roots of soil-grown plants and plants cultivated at low and high P conditions, respectively, while no cDNA was amplified with reverse-transcribed RNA originating from a Pht2;1 knock-out mutant (mutant described elsewhere). The use of genomic DNA as a template for PCR confirmed the presence of an intron in the respective genomic region. Thus, Pht2;1 is expressed in both auto- and heterotrophic tissues in potato as well as Arabidopsis.
Histochemical localization of Arabidopsis Pht2;1 expression
To allow a more thorough study of Pht2;1 gene expression during plant development, we amplified the promoter region of ARAth;Pht2;1 from genomic DNA following a PCR approach (see Experimental procedures). The use of two primer sets resulted in the cloning of promoter fragments of 2 and 1 kbp in length, respectively. The 1- and 2-kbp promoter fusions to the GUS reporter gene were introduced into Arabidopsis plants, and progeny of the transgenic lines were assayed for GUS activity. The 2-kbp promoter of the Arabidopsis Pht2;1 gene (AY606273), as well as the 1-kbp Pht2;1 promoter fragment (data not shown), directed strong expression of GUS in stem and leaf tissue (Figure 4). In leaves, GUS activity was detectable in all cells, with the highest concentration of the stain in the vasculature (Figure 4a,b), thus corroborating previous data obtained by in situ mRNA hybridization (Daram et al., 1999). In open flowers, GUS activity was high in sepals and stamens (Figure 4c), in the inflorescence stem adjacent to the silique, and at both ends of a young developing silique (Figure 4f). GUS expression was normally low in petals and absent in stigmatic papillae (Figure 4c,e) and in stamens of flower buds (data not shown), whereas stamens in open flowers developed GUS staining (Figure 4d). During progression of silique development, GUS activity became apparent in the entire silique wall, but not in seed coats (Figure 4f,g). Isolated seeds showed GUS activity at the connection site to the funiculus (Figure 4g). In young seedlings, GUS staining was well detectable in cotyledons, including trichomes, and in the hypoctoyl adjacent to the root (Figure 4h). Roots of transgenic seedlings exhibited GUS activity in the stele, including the vascular system and the pericycle, mainly at the sites where lateral roots emerged (Figure 4i,j). During the progression from the mature embryo to the young germinating seedling, GUS activity was high in cotyledons and became more and more restricted to the vascular system in roots and hypocotyl, the latter being most obvious in dark-grown seedlings (Figure 4k–n).
Differential regulation of Pht2;1 and StPT1 expression in leaves
To correlate source–sink interactions for Pi with low- and high-affinity Pi transporter mRNA levels, RNA gel blot analysis was performed with total RNA isolated from leaves of different ages similar to those used for the 32Pi measurements (Figure 1b). Similar to the observation in Figure 3(a), the two Pi transporters Pht2;1 and StPT1 are regulated in a complementary way on the mRNA level with high Pht2;1 transcript levels in young versus old leaves and vice versa with low StPT1 mRNA levels in young versus old leaves, respectively (Figure 5a). The profiles of the respective mRNA levels differ from those for the mRNA profile of the carbohydrate source marker cytosolic fructose-1,6-bisphosphatase (cyFBPase; Zrenner et al., 1996; Figure 5a). Thus, the regulation of Pht2;1 and StPT1 gene expression is not related to the carbohydrate source capacity of leaves, but it rather specifically reflects the Pi sink/source relationship in leaves of different ages (Figure 1). Comparably high Pi concentrations in leaves 1–5, followed by steadily decreasing concentrations of Pi in leaves 6–9, respectively, were measured (Figure 5b). The values determined for leaves 7–9 were significantly different from those of leaf 1.
To further determine whether Pht2;1 and StPT1 expression is regulated by Pi concentrations in leaves, RNA gel blot analysis was performed with total RNA isolated from leaf 2 of potato plants cultivated with half-strength Hoagland solution containing either 5 µm or 1 mm Pi, respectively. After 6–8 weeks of cultivation under these conditions, the plants kept under low Pi conditions showed severe symptoms of nutrient deficiency (Figure 5c). Pht2;1 transcript levels were not increased during Pi deprivation, thus confirming the results previously obtained with ARAth;Pht2;1 (Daram et al., 1999). In contrast, signal strength of Pht2;1 on the blot was moderately increased under high Pi conditions, while StPT1 expression was increased under Pi deprivation (Figure 5d).
Subcellular localization of Pht2;1/GFP fusion proteins
The long hydrophilic N-terminus was predicted by protein-sorting algorithms to mediate either chloroplastidic or mitochondrial targeting of the Pht2;1 protein (data not shown). In a first attempt to determine the subcellular localization of the mature polypeptide, radioactively labeled Pht2;1 protein from Arabidopsis was used for in vitro chloroplast import studies using in vitro translated ARAth;Pht2;1 cRNA and chloroplasts from Pisum sativum (data not shown). This approach failed to demonstrate uptake of the labeled protein into chloroplasts, which however does not exclude the possibility that Pht2;1 is imported into chloroplasts in vivo. Next, partial cDNAs of ARAth;Pht2;1 and SOLtu;Pht2;1, respectively, both encoding the N-terminus were fused to the green fluorescent protein (GFP) reporter gene and were subsequently used to transiently transform Arabidopsis or tobacco leaf protoplasts, respectively. An overlay of chlorophyll-derived red fluorescence and green fluorescence originating from GFP in protoplasts transiently expressing the chimeric genes resulted in yellow fluorescence, thus indicating chloroplast targeting of the fusion proteins containing the signal peptide (SP) for plastid targeting and the first TM1 in the case of potato Pht2;1, and SP with or without TM1, respectively, in the case of Arabidopsis Pht2;1 (Figure 6). Non-fused GFP localized to the cytosol.
Regulation of Pht2;1 expression by light
Abundance of Pht2;1 transcript was the highest in sink tissues of the aerial part of the plant, such as floral buds, open flowers, and young leaves (Figure 3). To further investigate the possible influence of sink strength on Pht2;1 transcript abundance, source leaves were artificially converted into sink organs by wrapping them in aluminum foil for 48 h (Eschrich and Eschrich, 1987), while the remaining parts of the plant were exposed to light. The physiological conversion of the leaf from a carbohydrate source to a sink did not increase expression of Pht2;1 (Figure 7a). Similar to cyFBPase, respective mRNA levels were rather drastically reduced, suggesting direct regulation of Pht2;1 gene expression by light.
Subsequently, leaf 3 was harvested from individual plants at defined time-points throughout a 16-h light/8-h dark cycle and during a prolonged dark period, thus allowing the detection of a possible diurnal rhythmicity of Pht2;1 and StPT1 gene expression (Figure 7b). Although there is no evidence for a circadian control of Pi transporter gene expression (data not shown), it was confirmed that, unlike StPT1, Pht2;1 mRNA concentrations were strongly reduced in the leaves after 48 h of dark treatment. Within 6 h after re-exposure to light, high Pht2;1 transcript levels were again reached (Figure 7b), suggesting regulation of gene expression by light or photoassimilates. To analyze the degree of conservation of Pht2;1 regulation between potato and Arabidopsis, a similar time course experiment was performed using Arabidopsis seedlings. The data indicated that ARAth;Pht2;1 is also a light-regulated gene (Figure 7c), exhibiting a rapid increase of transcript levels within 2 h of exposure to light. This increase early during the time course was more pronounced with Pht2;1 than with ribulose bisphosphate carboxylase (Rbcs) encoding the small subunit of Rubisco. Subsequently, both Pht2;1 and Rbcs mRNA levels gradually decreased to very low levels within 24 h in darkness. To evaluate a possible regulation of Pht2;1 gene expression by photoassimilates, transgenic plants exhibiting an alteration in photosynthetic activity and hence carbohydrate partitioning were analyzed with respect to Pht2;1 mRNA levels. To this end, potato plants expressing the plastidic fructose-1,6-bisphosphatase (cpFBPase; Kossmann et al., 1994) or the sucrose transporter gene (StSUT1; Riesmeier et al., 1994), respectively, at strongly reduced levels because of antisense repression were grown in the greenhouse, and subsequently Pht2;1 mRNA levels were analyzed in leaves and compared to wild-type controls. Although the leaves of the respective transgenic lines exhibited the documented phenotype, indicating strongly altered concentrations of soluble sugars, Pht2;1 transcript levels remained unchanged in both young and old leaves (data not shown), thus making Pht2;1 regulation by carbohydrates unlikely. Similarly, by using photosynthetic inhibitors, such as atrazine (Figure 7d) and DCMU (data not shown), it was shown that photosynthetic activity per se does not regulate Pht2;1 gene expression in Arabidopsis seedlings grown in liquid medium. However, the increase in Pht2;1 transcript levels is reduced in presence of ABA and light induction of Pht2;1 is inhibited by cycloheximide, indicating requirement of protein synthesis for its activation (Figure 7d).
Alteration of Pht2;1 mRNA levels in transgenic plants
To analyze a possible regulatory function of plastidic Pht2;1 transport activity in whole-plant Pi allocation, transgenic potato plants were generated in which the expression of Pht2;1 was either strongly reduced or increased by introducing a chimeric construct containing the coding region of Pht2;1 linked in antisense or sense orientation, respectively, to the 35S CaMV constitutive promoter. From numerous kanamycin-resistant plants, strong Pht2;1 overexpressors and repressors, respectively, were raised and analyzed for Pi transporter expression in leaves (Figure 5e). In contrast to our expectation (Figures 3a and 5a), StPT1 and Pht2;1 were not coordinately expressed in the transgenic lines. While Pht2;1 transcript levels varied substantially in both Pht2;1 repressor and overexpressor lines as compared to the wild type, StPT1 mRNA levels remained unchanged in respective leaves. Additionally, compared to controls, transgenic lines exhibited no differences on the physiological level, i.e. in leaf Pi content (data not shown), 32Pi distribution pattern (Figure 1a), 32Pi uptake into leaves (Figure 1b), tuber yield (data not shown), or specific gravity of the tubers (reflecting starch content; data not shown). The data indicate that, although Pht2;1 and StPT1 mRNA profiles are complementary in the leaves (Figures 3a and 5a), the two genes are independently regulated by different mechanisms. A strong increase or repression of Pht2;1 mRNA accumulation did not significantly change Pi allocation and starch metabolism. Similarly, no alterations in chloroplast functioning could be demonstrated by measuring both the light response of gas exchange and chlorophyll a fluorescence in leaves of the different lines (data not shown).
In silico Pht2;1expression analysis
The observed similarities in the regulation of the expression of SOLtu;Pht2;1 and ARAth;Pht2;1 prompted us to investigate in more detail the expression of Arabidopsis Pht2;1 under different experimental conditions. To this end, a compilation of relative Pht2;1 expression levels from Affymetrix AG and ATH1 genome arrays retrieved from the publicly accessible Nottingham Arabidopsis Stock Centre (NASC) microarray database was established. The comparative analysis of 228 GeneChips®, including replications, yielded signal log ratios of gene expression covering 85 experimental conditions (Figure 7e; Tables S1 and S2 for a description of the different experiments). Interestingly, Pht2;1 expression was induced ≥twofold under only six different experimental conditions, including treatment of cell suspension cultures with the herbicide isoxaben, which is the cellulose synthesis inhibitor 2,6-dichlorobenzonitrile (3.7-fold; NASC chip name: A2-willa-ISOX-REP2), in senescing cell cultures (3.6-fold; Swidzinski Senescence AGA; Swidzinski et al., 2002), during exposure of roots to the heavy metal lead (3.5-fold; Gawronski_6ASWG50PbR), and in shoot apices of a mutant that is affected in meristem activity and that exhibits prolonged transition from vegetative to reproductive growth and altered leaf morphology (2.2-fold; PD001_AG_HEA21). Interestingly, in contrast to the situation in roots, lead-inducible Pht2;1 expression was not observed in leaves (Tables S1 and S2). Similar to our RNA gel blot data, P deficiency failed to affect Pht2;1 transcript levels, neither did absence of N or K (Tables S1 and S2). In contrast, downregulation of Pht2;1 of ≥twofold occurred under only four conditions, including ozone treatment of whole plants (2.7-fold; Ozone_with; Clayton et al., 1999), Agrobacterium infiltration of inflorescence stalks (2.6-fold; Agro_tum_rep2), in the snakeskin (sks) mutant (2.5-fold; MB001_AG_SKS), and heat shock of cultured cells (2.3-fold; Pcd_heat_rep3). The sks mutant peels off its epidermal layer probably because of cell wall alterations (Malcolm Bennett, Nottingham University, Nottingham, UK, personal communication).
Table S1. Information about the microarray data and the corresponding experiments reported in this paper
Table S2. Data from a cluster of Pht2;1 and coexpressed genes from 84 chips from the ATH set of arrays from NASC
Subsequently, hierarchical cluster analysis of expression data from the ATH1 set of arrays revealed several genes in the Arabidopsis genome, which are co-expressed with Pht2;1 within the set of chosen experiments (Figure 7f; Table S2). These putatively Pht2;1 co-regulated genes code for A. thaliana pentatricopeptide (PPR) repeat-containing protein (At5g25630) mRNA, multidrug and toxic compound extrusion (MATE) efflux protein-related (At4g22790) mRNA, thioredoxin family (At1g07700) mRNA, jacalin lectin family (At1g19715) mRNA, unknown protein (At2g21960) mRNA, and plastidic malate dehydrogenase (NADP) protein, putative (At5g58330) mRNA, the differential expression of the latter matching closest to Pht2;1 expression. Two genes were expressed in a manner opposite to that for Pht2;1, namely expressed protein (At5g67350 and At1g70420) mRNAs. The program psort prediction (Bannai et al., 2002) at http://psort.ims.u-tokyo.ac.jp/form.html indicated chloroplastidic localization for the MATE efflux protein-related protein, the thioredoxin family protein, and the NADP-dependent malate dehydrogenase, respectively, whereas indication for chloroplast targeting exists additionally for PPR repeat-containing protein and unknown protein (At2g21960) as predicted by targetp at http://www.cbs.dtu.dk/services/TargetP/.
Intracellular Pht2;1 protein targeting
The long hydrophilic N-terminus is unique for the plant Pht2 proteins (Figure 2) in comparison with related Pi transporters of bacterial (archaea and eubacteria) and eukaryotic (human, rat, nematode, yeast, and fungal) origin (Daram et al., 1999). It was previously hypothesized that the Pht2;1 protein is a plasma membrane protein involved in Pi loading of shoot organs (Daram et al., 1999). In the light of chloroplastidic localization of Pht2;1 presented in this work and elsewhere (Figure 6; Ferro et al., 2002; Versaw and Harrison, 2002; Zhao et al., 2003), this hypothesis, which was based on mRNA expression data, yeast complementation analyses, as well as the failure of observing uptake of in vitro-synthesized Arabidopsis Pht2;1 protein into pea chloroplasts (our own unpublished results) must now be abandoned. Moreover, immunolocalization studies identified Pht2;1 from spinach as an integral protein of the chloroplast inner envelope membrane (Ferro et al., 2002). Overall, this suggests a role of Pht2;1 activity in chloroplastidic Pi uptake.
Control of Pht2;1 gene expression
Young tissue with dividing and rapidly expanding cells is a strong sink for Pi with the synthesis of nucleic acids and phospholipids probably being major metabolic loads (Figure 1; Biddulph et al., 1958) and was shown to be the site of highest Pht2;1 expression (Figures 3, 4, and 5a,b). In contrast, neither Pi supply to the plant nor carbohydrate source–sink interactions regulate Pht2;1 transcript levels (Figure 5).
Similar to Arabidopsis Pht2;1, the potato ortholog is predominantly expressed in aerial parts of the potato plant (Figures 3 and 4). However, detection of Pht2;1 transcripts in flower organs, in roots and in tubers of potato as well as in roots of Arabidopsis (Figure 3) indicated that Pht2;1 is also targeted to non-green plastids in heterotrophic tissue. High expression levels of both plastidic Pht2;1 and the plasma membrane Pht1 transporter StPT1 in flowers (Figure 3a,b) suggest high rates of Pi influx and high demand for intracellular Pi in the respective tissues.
It was recently shown that a large number of genes, including Pht2;1, are regulated by phyA in response to a continuous far-red light signal (Tepperman et al., 2001). The temporal profile of Pht2;1 regulation is similar in potato and Arabidopsis and Pht2;1 transcript levels strongly decrease during dark treatment, and subsequently increase to high levels upon re-exposure to light (Figure 7b,c). In contrast, no light regulation was detectable for StPT1 gene expression, suggesting a ‘house-keeping’ role in Pi transport for the Pht1 transporter, while Pht2;1 expression probably is interconnected with and responds to changes in plant metabolism. The results presented in this work thus suggest that SOLtu;Pht2;1 expression is under the control of a network of signaling pathways, including light and developmental signals as well as Pi sink strength in individual plant organs.
Possible physiological function of Pht2;1 activity
It is well known that Pi concentrations in the cytosol and the chloroplast affect photosynthesis and carbon partitioning in the light–dark cycle (Flügge, 1999; Heldt, 1977). Pi transport across the chloroplast inner membrane is mainly mediated by the TPT protein with a passive counter-exchange mechanism, whereas unidirectional transport occurs at much lower rates (Loddenkotter et al., 1993; Stitt, 1997). Arabidopsis Pht2;1 activity was assayed in complemented yeast mutants and was shown to mediate H+-coupled low-affinity Pi transport into cells (Daram et al., 1999). Recently, similar results were shown for Pht2;1 from M. truncatula (Zhao et al., 2003), suggesting that Pht2;1 from potato exhibits similar kinetics, although no experimental data are presently available. Unidirectional Pi transport across the inner envelope membrane of chloroplasts could be required during periods when the Pi concentration within the chloroplast is too low and consequently ATP synthesis is inhibited. Such a situation may occur when removal of triose-phosphate and conversion to sucrose in the cytosol is too slow, and as a consequence, Pi supply from the cytosol to the chloroplast via the TPT is inadequate (Stitt, 1997). Thus, Pht2;1 may be required in photosynthetic tissues to maintain levels of Pi within the stroma, which allow rapid photosynthesis to continue. However, under our experimental conditions, the absence of significant effects of down- or upregulation of Pht2;1 expression on gas exchange and chlorophyll fluorescence and thus chloroplast functioning (data not shown) would seem to rule out an involvement of Pht2;1 activity in the regulation of photosynthesis activity. Alternatively, Pi imported via Pht2;1 into plastids could counter-balance ADP/ATP exchange via the ADP/ATP translocator to maintain plastidic Pi homeostasis in both green and non-green tissue.
High StPT1 mRNA levels correlated with the Pi source capacity of old leaves containing the least amounts of Pi (Figures 1 and 5a,b). In addition, StPT1 mRNA abundance is increased in leaves of Pi-starved plants (Figure 5d), suggesting a Pi retrieval function of StPT1 in old leaves comparable to the proposed function of the ortholog Lycopersicon esculentum Pi transporter 1 (LePT1) from tomato during Pi starvation in root stelar tissue (Daram et al., 1998).
Modification of Pht2;1 expression in transgenic plants
There is compelling evidence for transcriptional regulation of Pht1 Pi transporter expression in plants probably regulated by the internal Pi concentration of the tissue (Karthikeyan et al., 2002; Liu et al., 1998; Muchhal and Raghothama, 1999). It was therefore tempting to speculate, that up- or downregulation of Pht2;1, respectively, could have an influence on StPT1 expression because of altered Pi concentrations in the plastids affecting Pi metabolism. In fact, recent data obtained with the Arabidopsis Pht2;1 null mutant pht2;1-1 supported this hypothesis (Versaw and Harrison, 2002). It was shown that pht2;1-1 exhibited a trend for increased transcript levels of phosphate-responsive genes, including the Pht1-type transporter gene AtPT2, under low Pi conditions. Our results on the overexpression, co-suppression, and antisense repression of Pht2;1 in potato do not support the existence of a similar mechanism in potato leaves, as StPT1 expression was unaffected by changes in Pht2;1 expression (Figure 5e), neither was there a correlation between the Pht2;1 expression level in leaves and Pi content or distribution, respectively (Figure 1; data not shown). In contrast, the mutant pht2;1-1 had altered levels of Pi in leaves dependent on the external Pi concentration (Versaw and Harrison, 2002). These discrepancies may be explained by differences in the experimental conditions, by residual amounts of Pht2;1 protein or by a different regulatory role of Pht2;1 in plastidic Pi import in potato and Arabidopsis, respectively.
In silico analysis of Pht2;1 expression
The meta-analysis of gene chip data, including 83 experimental conditions, revealed that Pht2;1 transcript levels were not subject to frequent alterations (Figure 7e). Suspension cultured cells compensate for the almost complete loss of cellulose from their cell walls as a result of inhibition of cellulose synthesis by constructing walls made predominantly of pectin (Shedletzky et al., 1992). Under these conditions, Pht2;1 expression was induced 3.7-fold (Figure 7e; Tables S1 and S2). Interestingly, Pht2;1 was shown to be one of nearly 500 genes the expression of which significantly fluctuated in Arabidopsis cells progressing through the cell cycle (Menges et al., 2002). Among these potential cell cycle-regulated genes, Pht2;1 expression peaked in M phase together with mitotic cyclins, CDKs, and a putative auxin-regulated protein. During M phase, the cell plate is formed at the phragmoplast, dividing a mother cell into two daughters, which requires the synthesis and transport of new membrane and wall materials (Vincken et al., 2003). It is thus tempting to speculate that transcriptional regulation of Pht2;1 is part of an essential response to altered cellulose and/or pectin biosynthesis. Experimental verification of this hypothesis is needed.
Cluster analysis suggests that Pht2;1 may be part of a functional module consisting of seven genes, most of which encode (putative) plastidic proteins (see Results). PPR proteins constitute a large family of plant proteins corresponding to >200 genes in the complete Arabidopsis genome (Small and Peeters, 2000). The few PPR motif-containing proteins characterized so far are involved in RNA metabolism and/or translation (see, e.g. Meierhoff et al., 2003). The MATE family in Arabidopsis consists of at least 56 members and encodes putative secondary transporters sharing homology with bacterial efflux transporters (Diener et al., 2001). The thioredoxin system is involved in numerous functions in the chloroplast such as light activation of chloroplast enzymes, including ATP synthase and NADP–malate dehydrogenase (Buchanan et al., 2000). The predicted functional module thus provides a possible link between Pht2;1-mediated Pi transport and photosynthetic carbon metabolism. In the future, genetic and biochemical studies will be performed to test the hypothesis that Pht2;1 plays a role during periods of high demand for intermediates of light-activated plastidic carbon metabolism.
Strains and plasmids
Escherichia coli DH5α (Invitrogen, Basel, Switzerland) (supE44 hsdR17 recA1 endA1 gyrA96 thi-1 relAl; Bethesda Research Laboratories, Bethesda, MD, USA), SOLR™, and XL1-Blue MRF' (Stratagene, Amsterdam, the Netherlands) were cultured according to standard techniques (Sambrook et al., 1989). The potato leaf cDNA library was described previously by Kossmann et al. (1992). Recombinant phages (5 × 105) were screened using the BglII fragment from the 3′ end of the ARAth;Pht2;1-coding region as a radioactive probe (Daram et al., 1999) according to the manufacturer's protocol (Stratagene). Protran BA85 filters (Schleicher and Schuell Inc., Rieden, Switzerland) were hybridized overnight at room temperature according to standard conditions (Sambrook et al., 1989). The ARAth;Pht2;1 BglII fragment was additionally used as a radioactive probe on RNA gel blots. StPT1 and StPT2 cDNAs (Leggewie et al., 1997), UBQ encoding ubiquitin, and cyFBPase (Zrenner et al., 1996) were used as radioactive probes on RNA gel blots. For potato transformation, the SOLtu;Pht2;1 cDNA was introduced in sense or antisense orientation, respectively, in the binary vector pBinAR-Kan containing the 35S cauliflower mosaic virus promoter and the T-DNA octopine synthase gene terminator in the binary vector Bin19 (Bevan, 1984; Franck et al., 1980; Gielen et al., 1984). The primers P7540 (5′-GAGAGAGCTCGATCAATATATGATCTCCCAGG-3′) and P8540 (5′-GAGACCATGGCTGAAAAGAAGAAGAAG-3′), and the primers P6540 (5′-GAGAGAGCTCGTACTCAGAGTAGAGGTGCG-3′) and P8540 were used to amplify the 1- and 2-kbp ARAth;Pht2;1 promoter fragments, respectively, from genomic DNA. The PCR products were subsequently digested with SacI and NcoI and ligated into the pBluescript vector opened with the same enzymes and containing the GUS reporter gene and the nopaline synthase (nos) terminator. The entire promoter-GUS reporter gene cassettes were finally introduced into the binary vector pCAMBIA3300 for Agrobacterium-mediated transformation of A. thaliana. For each chimeric reporter gene, four independent transgenic lines were analyzed to determine tissue and cell specificity of GUS activity.
The control plasmid mAv5-GFP3 for cytoplasmic localization of GFP was kindly provided by T. Merkle (Institute of Biology II, Albert-Ludwigs University of Freiburg, Freiburg, Germany). mAv5-GFP3 is based on pUC19 and contains the (S65T) GFP-encoding sequence under the control of the 35S promoter and the nos terminator (Haasen et al., 1999).
Plant material and growth conditions
The plant material was from Solanum tuberosum cv. Désirée and A. thaliana ecotype Columbia (Col)-2. Potato plants were grown in either soil or a quartz-sand/soil mixture (10 : 1) with constant irrigation with half-strength Hoagland medium (Hoagland and Broyer, 1936) containing either 5 µm or 1 mm Pi, respectively. Arabidopsis and potato plants were grown in a conventional greenhouse with a day and night cycle of 14 and 10 h, respectively, and temperature settings of 18–21°C during the day and 15–18°C at night at 40–80% relative humidity. Various organs from soil-grown plants were harvested for RNA gel blot analysis. For light/dark experiments, potato plants were cultivated in a climate chamber, and leaf 3 at the apex of individual plants was harvested to perform RNA gel blot analysis; leaf 1 indicates the youngest detachable leaf (length c. 0.5 cm). In a second experiment, leaves 7 and 8 were wrapped in aluminum foil for 48 h on an intact plant grown in the greenhouse before harvest. Potato transformation was performed using Agrobacterium-directed gene transfer essentially as described by Rocha-Sosa et al. (1989). Transformation of Arabidopsis was carried out using the floral dip procedure (Clough and Bent, 1998) and transgenic seedlings were selected by treatment with glufosinate (BASTA).
For −P and +P treatments, Arabidopsis seedlings were first grown on 0.5× MS medium (Murashige and Skoog, 1962) containing 1% sucrose for c. 2 weeks and subsequently transferred to the greenhouse in quartz-sand irrigated with 1/8-strength Hoagland medium containing either 1 or 125 µm Pi, respectively. Arabidopsis seedlings used for histochemical analysis or light induction experiments, respectively, were grown in continuous light or in the dark where indicated, on 0.5× MS agar plates containing 1% sucrose after initial stratification at 4°C in the dark for 4 days. Arabidopsis plants grown in liquid culture were used for treatment of seedlings with atrazine, ABA (both at 10 µm final concentration) or cycloheximide (50 µm final concentration). Eight 1-week-old A. thaliana seedlings were transferred from agar plates to Erlenmeyer flasks containing 0.5× MS medium and 0.5% sucrose. Plants were cultivated for 1 week in a day/night cycle of 16 and 8 h, respectively, on a rotary shaker at 60 r.p.m. Finally, plants were incubated for 24 h in darkness, with atrazine, ABA, or cycloheximide, respectively, added 1 h before transfer to light. Light induction was subsequently performed in white light for 2 and 4 h, respectively.
The translocation of 32Pi in plants was basically investigated according to Dong et al. (1998). Uniform wild-type and transgenic potato plants from sterile culture were first adapted to greenhouse conditions in quartz-sand culture and fertilized with quarter strength Hoagland solution (containing 250 µm Pi in the form of NH4H2PO4). After at least 1 week of adaptation, the plants were transferred to 50 ml quarter strength Hoagland solution (250 µm Pi) for 2 h under illumination to adjust to the experimental conditions. Finally, a 2-µl pipette tip was used to prick a hole in one defined leaf blade (sink or source leaf) and 2 × 2 µl of labeled KH2PO4 (approximately 37 × 103 Bq ml−1) were applied. Twenty minutes later, 2 × 2 µl distilled water was applied to the same site. After 6 h, the leaves fed with 32Pi were detached and discarded. The plants were then carefully spreaded and wrapped in plastic foil. The distribution of radioactivity was subsequently visualized on X-ray film after overnight exposure at −80°C.
For 32Pi uptake and distribution experiments in detached leaves, wild-type and transgenic plants, respectively, were incubated in quarter strength Hoagland's solution with 2 mm Pi for the first day and 5 mm Pi for the second day. Next, the leaves were cut off from the stem and placed individually into 1.5-ml Eppendorf tubes with uptake medium containing 5 mm EDTA, pH 6.0, 1 µmPO4 (74 kBq per 1.5 ml), and 1 mm KH2PO4 (final specific activity in the medium, 50 kBq ml−1, 0.8 MBq mol−1). The leaves were fed for 1 h under illumination. Next, the leaf petioles, which had been in contact with the radioactive solution, were removed; the laminae were placed in plastic foil and exposed to X-ray film overnight at −80°C.
Pi measurements were essentially performed as described by Hurry et al. (2000). Briefly, potato leaves were ground to a fine powder in liquid N2 and, depending on the leaf weight, extracted in 1–5 ml of 3% HClO4. The thawed homogenate was immediately centrifuged for 5 min at 14 000 g to remove cell debris and the supernatant (200–600 µl) was subsequently assayed for Pi concentration with 400 µl of ferrous sulphate-ammonium molybdate reagent as described by Tausky and Shorr (1953). After a 10-min incubation, the OD720 was measured and the values were related to a standard curve. The reliability of the assay was tested by recovery experiments, in which additional KH2PO4 solution was added to the frozen plant material. An average recovery of 103.5 ± 14.8% was reached (n = 6).
RNA extraction and RT-PCR
RNA extraction and RNA gel blot analysis were performed as described previously by Rausch et al. (2001). For RT-PCR studies, 1 µg of total RNA isolated from different Arabidopsis and potato organs was reverse transcribed using Superscript II RT (Gibco, Basel, Switzerland). A 502-bp ARAth;Pht2;1 fragment was amplified by PCR from the cDNA library using the gene-specific primers 5′-CAAGGACTGGCTCTCAAGAC-3′ and 5′-TCCTGATGCATTTGTAGACTAG-3′. Use of the same primer combination with Arabidopsis genomic DNA resulted in an amplification product of 632 bp, allowing to distinguish between fragments amplified from the cDNA library or from genomic DNA in samples. As a control, a 407-bp ACT2 partial cDNA was amplified using the gene-specific primers 5′-CTTCCCTCAGCACATTCCAG-3′ and 5′-AACATTGCAAAGAGTTTCAAGGT-3′ (mainly according to Sorrell et al., 2002). Amplification on genomic DNA with the same primers resulted in a DNA fragment of 496 bp. For PCR amplification of a 504-bp SOLtu;Pht2;1 cDNA fragment, the gene-specific primers 5′-GGTCAAGGATTGGGATTGAAG-3′ and 5′-CCGTATGCATTTGTACACGAG-3′ were used. The same primer combination amplified a DNA product of c. 1.7 kbp on genomic potato DNA. Gene-specific primers 5′-AATGAATTTGGTTTGTTCAGTAAGG-3′ and 5′-AAACTTAAACAGGACTGTCCTTCC-3′ for amplification of a 755-bp partial StPT1 cDNA were used for a control. All PCRs were performed with 35 or 40 cycles for potato or Arabidopsis DNA, respectively. Accession numbers are AJ302645 for ARAth;Pht2;1, U41998 for ACT2, and AY603690 for SOLtu;Pht2;1.
Plant material was incubated in 0.1% X-Gluc (Biosynth, Staad, Switzerland) and 0.1% Triton X-100 (Fluka Chemie AG, Buchs, Switzerland) in 0.05 m sodium phosphate buffer, pH 7.2, at 37°C. The stained material was fixed in ethanol:acetic acid (3 : 1) overnight at 4°C. Complete removal of chlorophyll from the tissue was performed in 100% ethanol. Samples for sectioning were embedded in Technovit 7100 (Kulzer, Wehrheim, Germany) and 8-µm-thick sections were mounted on glass slides. The indigo precipitate was visualized using a stereomicroscope (Olympus SZX12, Olympus Optical Schweiz AG, Volketswil, Switzerland) or by light microscopy (Olympus AZ70) in combination with Nomarski optics (GUS activity appears indigo blue).
Subcellular localization of GFP
The primer pairs AtPep1 (5′-GAGAGGATCCATGACTCTTCCTTATCGTTTCTCTTCC-3′) and AtPep2 (5′-GAGACCCGGGGTTCTGTATCGGTTGATCAGC-3′), and AtPep1 and AtPepTM (5′-GAGACCCGGGTCCAGATCCAACAGAAGTCC-3′), respectively, were used to amplify by PCR the nucleotide sequence encoding the 88 and 188 N-terminal amino acids, respectively, of ARAth;Pht2;1. Similarly, the primers StTM1 (5′-GAGAGGATCCATGACTTCTTCCTACTCTTTATCTTC-3′) and StTM2 (5′-GAGACCCGGGTCCCAATCCTTGACCCAAAGAC-3′) were used to amplify the nucleotide sequence encoding the 143 N-terminal amino acids of SOLtu;Pht2;1. The PCR products were digested with the restriction enzymes BamHI and SmaI, and subsequently ligated in frame between the 35S promoter and the GFP cDNA in the correspondingly digested plasmid mAv5-GFP3 to generate ARAth;Pht2;1 SP-GFP, ARAth;Pht2;1 SP-TM1-GFP and SOLtu;Pht2;1 SP-TM1-GFP. Sequences of the PCR products were verified by sequencing.
For transient expression of the fusion proteins in plant cells, ARAth;Pht2;1 SP-GFP, ARAth;Pht2;1 SP-TM1-GFP, and SOLtu;Pht2;1 SP-TM1-GFP chimeric genes were transformed into Arabidopsis or tobacco mesophyll protoplasts, respectively. Arabidopsis protoplast isolation and transformation was essentially carried out as described by Abel and Theologis (1994). However, after transfection and the stepwise dilution with W5 solution over a period of 20 min, protoplasts were collected by centrifugation for 5 min at 50 g, re-suspended in 4 ml of W5 solution and incubated for 48 h in Petri dishes in the dark at RT before analysis by confocal laser scanning microscopy. Nicotiana tabacum SR1 plants (6–8 weeks old) grown under sterile conditions in glass pots with MS1 medium were used for protoplast isolation and transformation, generally following the optimized transformation protocol for SR1 tobacco protoplasts provided by Negrutiu et al. (1987), with a final PEG CMS 4000 concentration of 20%. After incubation for 20 min at RT, the protoplasts were re-suspended in 5 vol of W5, centrifuged for 5 min at 30–50 g, and finally cultured for 24–48 h in 5 ml of W5 medium before analysis.
Arabidopsis and tobacco protoplasts were subsequently analyzed with a confocal laser scanning microscope (Leica DM IRBE and Leica TCS SP laser; Leica, Unterentfelden, Switzerland) under oil with a 40× objective at excitation wavelengths of 476 + 488 nm (GFP) and 568 nm (chlorophyll auto-fluorescence), using the Leica nt program.
Sequence analysis, prediction of transmembrane regions and protein orientation, scanning of the protein sequences for the occurrence of patterns stored in the PROSITE database, and multiple sequence alignment were essentially carried out as described previously by Daram et al. (1999). blast searches were performed at The Arabidopsis Information Resource site (TAIR; http://arabidopsis.org).
In silico Pht2;1expression analysis
Gene expression data from a total of 228 oligonucleotide microarrays were downloaded from the Affymetrix AG and ATH1 array collections in the NASC microarray database (status July 2003). Relative expression levels were calculated for each treatment (see Tables S1 and S2), and treatments with replicates were averaged to single values, resulting in signal log ratios covering 85 individual experimental conditions. A cluster analysis of relative expression values from 26 conditions from the ATH1 set of arrays was performed using TIGR multiexperimentviewer 2.1. (MeV) by means of the Self Organizing Tree Algorithm (SOTA; Dopazo and Carazo, 1997; Herrero et al., 2001) followed by K-Means Clustering (Soukas et al., 2000). To identify genes that were oppositely expressed relative to Pht2;1, the values for Pht2;1 were changed to the opposite sign and clustering was performed as described above. The cluster data were graphically displayed using hierarchical clustering (Eisen et al., 1998) and the construction of support trees (Graur and Li, 2000) features of MeV 2.1. The data are interpreted based on the annotation given for the ATH1 chip by TIGR on 7 January 2004 and on blast analysis of the encoded protein sequences.
We thank Drs Jens Kossmann and Babette Regierer for the potato cDNA library, cpFBPase and StSUT1 potato plants; Dr Uwe Sonnewald for the cytosolic fructose-1,6-bisphosphatase cDNA, Dr Jörg Leipner for his support of the photosynthetic fluorescence measurements; Drs Felix Kessler and Pierre Daram for protein import studies; Silvia Brunner for her help with Figure 7(c); Dr Mark Stitt for a fruitful discussion. We especially thank the contributors of the NASC microarray database (NASCArrays) for providing the data. This work has been partially supported by the Swiss National Science Foundation (Grant no. 31–52934.97) and the Swiss Federal Institute of Technology (ETH) Zurich.