INTRODUCTION

Phosphorus (P) is an essential macronutrient for all living organisms. It serves various basic biological functions as a structural element in nucleic acids and phospholipids, in energy metabolism, in the activation of metabolic intermediates, as a component in signal transduction cascades, and the regulation of enzymes.

Of the major nutrients, P is the most dilute and the least mobile in soil. High sorbing capacity for P in the soil (e.g. sorbtion to metal oxides), P mineralization (e.g. calcium phosphates such as apatite), and/or fixation of P in organic soil matter (by converting soluble P into organic molecules) result in low availability of this macronutrient for uptake into plants (Marschner, 1995). P is absorbed by plants as orthophosphate (Pi, inorganic phosphate). Pi concentration in the soil solution hardly reaches 10 µM and may even drop to submicromolar levels at the root/soil interface, where Pi uptake by plants and root surface-colonizing microorganisms leads to the generation of a zone of Pi depletion around the root cylinder that is maintained due to slow diffusion of Pi from regions distant to the root surface (Figure 1).

In industrialized countries, low P availability in agricultural soils is compensated by a high input of P fertilizer to guarantee high crop productivity and yield. Water run-off, soil erosion and leakage in highly fertilized agricultural soils may cause environmental problems such as eutrophication of lakes and rivers. As forecasted by Tilman et al. (2001), during the next 50 years, which is likely to be the final period of rapid agricultural expansion, demand for food by global population will be a major driver of global environmental change. Conversion of natural ecosystems to agriculture by 2050 will be accompanied by an approximate 2.5-fold increase in nitrogen- and P- driven eutrophication of terrestrial, freshwater, and near-shore marine ecosystems. Modern agricultural soils are almost universally maintained at high fertilization. Selection of new cultivars is usually made under such conditions and will not normally distinguish between plants varying in nutrient efficiency (Stevens and Rick, 1986). To alleviate the forecasted adverse negative effects of agricultural expansion, scientists have started to use classical breeding strategies and biotechnology to improve crop plants, based on the current knowledge and aiming at an improved crop yield with a lower input of fertilizer, thus protecting the environment.

In contrast, in many developing tropical countries, subsistence farmers can not buy enough fertilizer due to limited financial capacities or poor infrastructure (Sanchez et al., 1997). As a consequence, P deprivation dramatically limits crop yield and is one of the reasons for poverty and malnutrition. In the future, agriculture from both developed as well as developing countries could thus benefit from modern crop varieties with enhanced P efficiency, thus leading to improved fertilizer management and increased crop yield on low-P soils. Thorough knowledge of the plant's response to P deprivation stress will contribute to the rational and targeted breeding of P efficient crop plants. Therefore, the authors of this chapter focus on summarizing the current state of research covering physiology, biochemistry, and molecular genetics of P acquisition and allocation, and P homeostasis within the plant. Although this review will mainly focus on knowledge acquired on Arabidopsis thaliana, some specific results obtained with other plant species will also be included in this work. For example, formation of mycorrhizae, which is observed in most vascular plants and strongly contributes to plant P nutrition, does not occur in Brassicaceae and therefore Arabidopsis is not a suitable model for mycorrhizae studies.

PHOSPHATE TRANSPORT ACROSS MEMBRANES

The textbook tells us that the development of a barrier such as the cell membrane was a prerequisite for the development of life, and enabled single cells to support metabolic, reproductive, and developmental activities under stable physico-chemical conditions (Buchanan et al., 2000). Much later during the evolution of life, the development of an endomembrane system in eukaryotic cells allowed the formation of organelles, each with its own chemical ‘inner world’ leading to a compartmentalization of solutes within the cell. It is the hydrophobic nature of the lipid bilayer that allows sequestration of hydrophilic compounds, such as most nutrients and metabolites, on either side of the membrane. This generation of different (bio)chemical environments enabled cells to establish a division and separation of biosynthetic, catabolic and storage functions, allowing metabolic flexibility and efficiency. Integration of these functions in the network of the cell and of the entire organism needs a selective transport of compounds across the membrane barriers. This selective permeability is mediated by membrane-spanning proteins within the lipid bilayer. These so-called transport or carrier systems are often highly specific for a certain compound (e.g. a carbohydrate such as sucrose) or a group of similar substances (e.g. divalent cations).

It was the pioneering work of Emmanuel Epstein (Epstein and Hagen, 1952; Epstein et al., 1963) which demonstrated that ion uptake processes across the plasma membrane follow Michaelis-Menten kinetics comparable to those of enzymatic processes known from biochemistry, thus allowing calculations of functional parameters such as pH optimum, Michaelis constant K_m, uptake velocity V_max, and minimal concentration of the ion at which transport occurs, i.e. C_min, Analysis of nutrient uptake kinetics into plant roots using a radiotracer medium-depletion method (Drew et al., 1984) revealed that Pi uptake kinetics in plants are generally hyperbolic and monophasic at low Pi concentrations (µM range) in the medium and biphasic when extended to high Pi concentrations (mM range), thus suggesting Pi uptake being mediated by high- and low-affinity transport mechanisms, respectively (Schachtman et al., 1998). However, in suspension-cultured tobacco cells, only one kind of Michaelis-Menten-type Pi transport system exhibiting a high-affinity for Pi has been described with no evidence for a low-affinity Pi transport (Shimogawara and Usuda, 1995). Moreover, concentration-dependent Pi influx of barley mesophyll protoplasts shows a combination of one Michaelis-Menten-type kinetics at low Pi concentrations and linear increase at higher Pi concentrations (Mimura et al., 1990). At Pi concentrations in the µM range, which corresponds to the natural conditions in cultivated soils, high-affinity transport occurs (Cogliatti and Clarkson, 1983; Drew et al., 1984). As there are micro- or even submicromolar concentrations of available Pi at the root-soil interface, but millimolar concentrations of intracellular Pi, Pi uptake across the cell boundary must be effected against a steep concentration gradient. This is accomplished by transport of the anion across the membrane coupled to the transport of protons (H⁺-symport). Plants and fungi use a proton P-type ATPase pump to generate an electrochemical gradient across the plasma membrane at the expense of ATP (Figure 2). This leads to the formation of a large membrane potential with a negative potential of the cytoplasm (⁻150 to ⁻200 mV). Consequently, the transport of Pi and other anions is usually coupled to protons in a secondary transport process. Thus, the driving force for Pi influx is the proton gradient generated by the P-type H⁺-ATPase (Schachtman et al., 1998; Sze et al., 1999; Thibaud et al., 1988; Ullrich-Eberius et al., 1984).

Pi is transported from the root to the shoot via the xylem, which consists of dead cells and thus reflects the extracellular space. During leaf senescence or Pi starvation conditions, Pi is redistributed from old source leaves to young sink organs or to the starved root. Transport steps involved in Pi acquisition and distribution within the plant include: Pi uptake at the root periphery, secretion from xylem parenchyma cells prior to xylem loading, loading of shoot tissues with Pi released from the xylem. Subsequent processes, such as Pi uptake into the sieve element/companion cell complex of the vascular tissue during Pi redistribution from source leaves, or Pi release from phloem cells in sink organs and subsequent Pi transfer to surrounding cells, are still poorly understood. Once inside the cell, Pi passes the membranes of cell organelles, such as plastids (e.g. chloroplasts and amyloplasts), mitochondria, and vacuoles, often in exchange with other solutes or protons. Metabolites that are exchanged between the plastids and the cytosol have to be transported across the two envelope membranes. Whereas the outer envelope allows the passage of small molecules, the inner envelope membrane is the actual permeability barrier and the site of Pi and metabolite transport (Flügge, 2001). The solutes to be transported, in exchange with inorganic Pi, are triose phosphates, pentose-phosphates, hexose-phosphates, or phosphoenolpyruvate (PEP). The stoichiometry of these exchanges is important, because it ensures that every molecule of phosphoester transported across either side of the plastid membrane is counterbalanced by the translocation of a Pi molecule to the opposite site (cytosol or plastid).

Mitochondria are the major sites of energy transduction in eukaryotes and uptake of Pi into the mitochondrial matrix is essential for the oxidative phosphorylation of ADP to ATP. Despite this central importance in metabolism, only very little is known about plant Pi transport across the mitochondrial inner membrane (see Kiiskinen et al., 1997 and references therein). Plant mitochondrial Pi transport systems probably catalyze Pi influx as a homodimer either via a Pi/H⁺ symport, a Pi/OH⁻ antiport mechanism, or via Pi/Pi exchange.

Pi homeostasis (from the Greek words for “same” and “steady”), that is the processes used to actively maintain fairly stable Pi conditions in the cell, is essential for cell integrity and includes the tight regulation of Pi concentrations and transport in cellular compartments, i.e. the cytosol, plastids, mitochondria, and vacuoles (Mimura, 1999). In vacuolated cells of higher plants, the vacuole acts as a storage pool, or ‘non-metabolic pool’, of Pi, and at adequate Pi supply of the plants about 85 – 95% of the total Pi is located in the vacuole (Bieleski, 1973). In contrast, in leaves of Pi-deficient plants, virtually all Pi is localized to the cytosol and chloroplasts, thus representing the ‘metabolic pool’ of Pi in the plant (Bieleski, 1973; Marschner, 1995). Therefore, vacuolar Pi is used to buffer the cytoplasmic Pi level against fluctuations caused by variable external Pi supply or metabolic activities. To date, only a few reports investigating Pi transport across the tonoplast have been published. Pi uptake measurements with intact isolated vacuoles and in situ using ³¹P-nuclear magnetic resonance (NMR) analysis suggest the existence of Pi transport systems in the tonoplast with ATP being important for Pi influx (Massonneau et al., 2000; Mimura, 1999).

Generally, the understanding of the molecular and biochemical mechanisms involved in Pi uptake at root interfaces (including mycorrhizal interfaces) and subsequent allocation of Pi to different plant tissues and organelles is central for the understanding of Pi homeostasis and is just beginning to be better understood on the molecular physiological level.

PHOSPHATE TRANSPORTERS

The model plant Arabidopsis performs the same functions and contains essentially the same genes as other flowering plants. With completion of its genome sequence (The Arabidopsis Initiative, 2000), Arabidopsis is the main model system for laboratory studies in basic plant biology. Understanding Arabidopsis nutrient transport will assist in understanding and improving nutrient acquisition efficiency in more familiar crop plants which could finally lead to the development of an environmentally sound agriculture with low input of fertilizer. However, it has to be noted that some biological phenomena can not be studied in this model plant, such as e.g. some species-specific plant-pathogen interactions, mycorrhizal symbiosis or nitrogen fixation.

Comparison of membrane transport processes between Arabidopsis, animals, fungi and prokaryotes has identified over 600 predicted membrane transport systems in Arabidopsis( http://www.biology.ucsd.edu/~ipaul-sen/transport/), a similar number to that of the worm Caenorhabditis elegans (∼700 transporters) and over twofold greater than either baker's yeast (Saccharomyces cerevisiae) or the bacterium Escherichia coli (∼300 transporters) (The Arabidopsis Initiative, 2000). Recently, rapid progress has been made in the molecular biology of Pi uptake. Using sequence information of Arabidopsis, expressed sequence tag (EST) clones (accession numbers 134M11T7, 178H14T7, and ATTS2854) (Mitsukawa et al., 1997; Muchhal et al., 1996; Newman et al., 1994; Smith et al., 1997), which exhibited sequence similarity to the yeast high-affinity Pi transporter PHO84 (Bun-Ya et al., 1991), full-length cDNAs and their genomic clones were isolated from DNA libraries of Arabidopsis. An Arabidopsis Information Resource (TAIR) Gene Search at  http://arabidopsis.org/info/genefamily/genefamily.html reveals the existence of nine genes in the haploid Arabidopsis genome coding for inorganic Pi transporters that exhibit high sequence similarity to each other. According to the rules recommended by the Commission on Plant Gene Nomenclature ( http://mbclserver.rutgers.edu/CPGN/) this plant gene family was named the Pht1 family (Bucher et al., 2001; table 1). The plant Pht1 family together with the yeast and fungal homologs belong to the Pi:H⁺ symporter (PHS) family which is a member of the large major facilitator superfamily (MFS)(Pao et al., 1998). These nine Pht1 genes are located on chromosomes 1, 2, 3, and 5, respectively (Figure 3a). Interestingly, Pht1;1, Pht1;2, Pht1;3, and Pht1;6 are clustered within 24,200 bp on chromosome 5, suggesting the occurrence of several gene duplication events during the molecular evolution of the Pht1 family in the respective region of the Arabidopsis genome to give rise to four individual Pht1 genes on chromosome 5. To give a visual impression of the similarities between the deduced amino acid sequences, an unrooted phylogenetic tree diagram of the Arabidopsis Pht1 family members is presented in Figure 4. The close relatedness of Pht1;1, Pht1;2, and Pht1;3 sequences furthermore suggests that the three genes may share a common ancestral gene and are evolutionary younger than Pht1;6. This hypothesis is supported by the presence of an intron in the first three Pht1 members whereas no intron is contained in the Pht1;6 genomic region (Figure 3b). The Arabidopsis Pht1 genes encode proteins of approximately 520 amino acids with comparable predicted secondary structures derived from computational analysis, each is characterized by twelve hydrophobic domains presumably spanning the plasma membrane, a hydrophilic N and C terminus localized in the cytoplasm and a large hydrophilic loop between transmembrane spanning domains 6 and 7 (Figure 5). Comparison of Arabidopsis and other plant species Pht1 polypeptide sequences with those of Pi transporters from distantly related species, such as yeast, Neurospora crassa and Glomus versiforme, reveals a high degree of conservation at three positions which could be the target for posttranscriptional processing, such as N-glycosylation and protein phosphorylation (Rausch et al., 2001). Such protein modifications may be of structural significance or of functional importance for transport activity. To date, functionality of Pht1 Pi transporters can only be analyzed in a hydrophobic plasma membrane environment of either yeast or plant cells. Yeast mutants which are defective in either one or two genes encoding high-affinity Pi transporters, respectively, were used for functional complementation to analyze the biochemistry of Pht1 protein-mediated Pi uptake into yeast cells (Daram et al., 1998; Leggewie et al., 1997; Liu et al., 1998b; Muchhal et al., 1996; Rausch et al., 2001). The cDNAs encoding the transporters were cloned into a yeast expression vector behind a strong yeast promoter. Growth of complemented yeast strains and net uptake rates of Pi into the cells indicated that Pi transport of the yeast mutants had been restored by the expression of the plant genes. Expression of Arabidopsis Pht1;1 gene at high levels in tobacco-cultured cells increased the rate of Pi uptake and cell growth under Pi-limited conditions (Mitsukawa et al., 1997). The K_m values determined from the Pi absorption curves of either complemented yeast cells or transformed tobacco cells were in the micromolar range. Maximal uptake velocity was at around pH 5 and was reduced by inhibitors of H⁺-ATPase activity and by the action of protonophores. This is consistent with a proton symport mechanism operating in the plasma membrane. It can therefore be concluded that the Pht1 gene family encodes high-affinity proton/Pi cotransporters that mediate Pi uptake across the plasma membrane. Analysis of Pi transport kinetics of more Pht1 family members will reveal whether this holds true for all Pht1 proteins or whether there are some members, for example, displaying low-affinity Pi uptake. RNA gel blot analysis was used to show that Arabidopsis Pht1 genes are predominantly expressed in roots (Smith et al., 1997; Mucchal et al., 1996). Pht1;4 is, although at low levels, also expressed in Arabidopsis leaves (M. Bucher, unpublished data). More detailled investigations of gene expression in tomato and the legume plant Medicago truncatula using mRNA in situ hybridization and immunolocalization have shown that at least two members of the Pht1 family are predominantly expressed in cells of the root epidermis (Figure 6) including root hairs and in the root cap cells (Chiou et al., 2001; Daram et al., 1998; Liu et al., 1998a). During Pi deprivation conditions, tomato Pht1:1 (LePT1) gene transcripts accumulated in the stelar tissue suggesting a role for the respective protein as a Pi-retrieval system in the stele, reloading apoplastic Pi, which did not readily enter the xylem, in xylem parenchyma cells (Daram et al., 1998). It was shown that Pht1 proteins localize to the plasma membrane and that their abundance is predominantly regulated at the transcriptional level (Muchhal and Raghothama, 1999). Therefore, a primary role in Pi uptake at the root-soil interface can be assigned to this gene family. Moreover, expression analyses in mycorrhized plant species has shown that Pht1 members and homologous proteins are expressed in cells and tissues at three different interfaces involved in Pi acquisition, namely in the hyphal network of mycorrhizal fungi at the soil-fungus interface and in the rhizodermal cells at the soil-root interface, both to allow Pi uptake from the soil solution into the fungal hyphae or the root, respectively, and in root cortex cells colonized with mycorrhizal fungi at the fungus-root interface to allow Pi transfer from the fungus to the host plant (see Rausch and Bucher, 2002 for a review and references therein). The recent completion of the Arabidopsis thaliana genome has allowed expression analysis of all Pht1 members in this species. Promoter-reporter gene fusions for all nine Pht1 genes have been generated using either the green fluorescent protein (GFP) or the β-glucuronidase (GUS) reporter genes (Mudge et al., 2002). Similarly, Karthikeyan et al. (2002) have fused the Pht1;1 and Pht1;4 promoters to the GUS or the luciferase reporter genes, respectively. Four of these promoters were shown to direct reporter gene expression to the root epidermis, outer cortex cells, endodermis and stele, and were induced under low Pi conditions in a manner similar to the previously described Pht1 genes. Other promoters directed reporter gene expression to non-root tissues, such as flowers, leaf vascular tissue, hydathodes of cotyledons, shoot buds, and pollen. Although promoter-reporter gene studies do not necessarily always reflect the expression profile of a gene in vivo, the presented work clearly suggests that Pht1 proteins do not only mediate Pi uptake from the soil solution, but play important roles in Pi translocation within the plant, thus significantly expanding our view of the role of Pht1 Pi transporters in plant nutrition.

Recently, the Arabidopsis cDNA ARAth;Pht2;1 (Pht2;1), which encodes a novel Pi transporter, the first member of the Pht2 Pi transporter family in vascular plants, has been cloned (Table 1; Daram et al., 1999). The Pht2;1 cDNA encodes a 61-kD protein which is structurally similar but distinct from the proteins of the Pht1 family in having a large hydrophilic loop between the transmembrane helices 8 and 9 and a long hydrophilic N terminus. Computational modeling suggested extracellular localization of both the N and C termini. Two boxes of eight and nine amino acids, respectively, which are located in the N- and C-terminal domains of the Pht2;1 protein are highly conserved among species across all kingdoms (eubacteria, archeae, fungi, plants, and animals), including sodium-dependent Pi transporters from yeast and mammals. These conserved domains may be of importance for the Pi transport activity or for the structure of the protein. The Pht2;1 gene is predominantly expressed in green tissue. Although the protein sequence is highly similar to that of eukaryotic sodium-dependent Pi transporters, functional analysis of the Pht2;1 protein in mutant yeast cells indicated that it is a H⁺/Pi symporter. Its fairly high apparent K_m for Pi (0.4 mM) and high mRNA abundance around the vascular tissue in leaves, based on RNA in situ hybridization studies, suggested a role for Pi loading of shoot organs (Daram et al., 1999). The independent efforts of three groups subsequently showed that Pht2;1 is localized to the chloroplast. Fusions of the entire Arabidopsis Pht2;1 protein or the N-terminal part of the transporter, respectively, to GFP indicated transport of the fusion proteins to the chloroplast (Versaw and Harrison, 2002; Bucher et al., unpublished). A subcellular proteomics approach, which was developed to identify the most hydrophobic chloroplast membrane proteins in spinach, in combination with immunolocalization and GFP fusion studies allowed Ferro et al. (2002) to identify the spinach Pht2;1 protein as an integral protein of the inner envelope membrane of chloroplasts. Pht2;1 thus differs from members of the Pht1 Pi transporter family in primary structure, affinity for Pi, subcellular localization and hence presumed function.

Arabidopsis Pht3 genes encode a small family of mitochondrial Pi transporters (Table 1; see  http://mbclser-ver.rutgers.edu/CPGN/IonXWeb/Pht.group.html and  http://arabidopsis.org/info/genefamily/Chloroplast.html). Pht3;1 (NCBI accession number BAB08283; Nakamura et al., 1997), Pht3;2 (PIR Entry T49281), and Pht3;3 (PIR Entry B84550) encode proteins of 309 to 375 amino acids in length which contain 4 to 6 possible transmembrane helices as is suggested by TMpred, a software for the prediction of transmembrane regions (accessible at  http://www.ch.embnet.org/software/TMPRED_form.html; Hofmann and Stoffel, 1993). The protein sequences are similar and are about 50% identical in their amino acid sequence. Pht3 genes show homology to the first plant mitochondrial Pi translocator cDNA (Mpt1) cloned from birch (Betula pendula Roth) in a screening for ozone-inducible genes (Kiiskinen et al., 1997b). The Mpt1 cDNA encodes a 364 amino acid polypeptide which is 66% similar to bovine mitochondrial Pi translocator protein isoform B. Birch Mpt1 transcript abundance remains low during leaf development and is lower in roots and leaves when compared to young shoots undergoing wood formation and lignification.

The envelope membrane of plastids contains specific translocators that are involved in transport processes of photosynthetic intermediates. Different classes of inner membrane plastidic Pi translocators that mediate the transport of phosphorylated compounds in exchange with inorganic Pi have recently been characterized (Table 1; Flügge, 1999). The members of the Pi translocator family exhibit partially overlapping substrate specificities. This allows the efficient uptake of individual phosphorylated substrates even in the presence of high concentrations of other phosphorylated metabolites. Moreover, this multiplicity of transporters relates to specific plastid types and differences are seen even within the same plastid population, reflecting the flexibility of plastid metabolism (Bowsher and Tobin, 2001). Arabidopsis TPT is a nuclear gene encoding the chloroplast triose-phosphate/Pi translocator precursor. The mature protein is involved in the export of carbon fixed during the day from the chloroplasts into the cytosol in the form of triose phosphates (Flügge, 1999). The need for phosphoenolpyruvate, as an immediate precursor for the synthesis of secondary products via the shikimic acid pathway or as a precursor for fatty acid or aromatic amino acid biosynthesis, necessitates a plastidic phosphoenolpyruvate/Pi translocator encoded by the PPT gene (Fischer et al., 1997). The PPT protein exhibits approximately 33% identity to the TPT. The corresponding gene is expressed in both photosynthetically active tissues and in non-green tissues, although transcripts were more abundant in non-green tissues. Expression of the coding region in transformed yeast cells and subsequent transport measurements of the purified recombinant translocator showed that the protein mediates transport of Pi in exchange with C3 compounds phosphorylated at C-atom 2, particularly phosphoenolpyruvate. Forward genetics analysis of the previously characterized cue1 Arabidopsis mutant (Li et al., 1995) revealed that PPT is required for synthesis of aromatic amino acids from the shikimic acid pathway, as well as palisade cell development (Streatfield et al., 1999). Non-photosynthetic plastids must import carbon in the form of hexose phosphates via the glucose 6-phosphate Pi translocator, named GPT. The Arabidopsis GPT exhibits 38 and 36% identity with the TPT and PPT proteins, respectively. Thus, GPT proteins represent a third group of plastidic Pi antiporters. Transport experiments with Arabidopsis GPT purified from transformed yeast cultures demonstrated that the GPT protein mediates a 1:1 exchange of glucose 6-phosphate mainly with inorganic Pi, released during starch biosynthesis, and triose phosphates, generated from the oxidative pentose pathway (Kammerer et al., 1998). The fourth type of plastidic Pi translocator is encoded by the pentose phosphate/Pi translocator gene from Arabidopsis, named XPT due to its preference for the substrate xylulose 5-phosphate (Xul-5-P) that is transported in exchange with inorganic Pi or triose phosphates, respectively (Eicks et al., 2002). The XPT protein shares 35, 34, and 47% identity with TPT, PPT, and GPT proteins, respectively. XPT transcripts are present in all organs. In the Arabidopsis genome, a total of eight genes are grouped into the Arabidopsis TPT translocator family containing the four types described above ( http://arabidopsis.org/info/genefamily/Antiporters.html). Characterization of the other four family members is needed to unequivocally assign a biological function to each gene.

An essential step in Pi uptake into the shoot is the loading of the xylem with Pi absorbed by the roots. The PHO1 gene, isolated by positional cloning from a mutant deficient in loading Pi to the xylem, is likely to encode the Pi efflux carrier involved in this process (Hamburger et al., 2002). The pho1 mutant and its corresponding gene will be described in further detail in a later section.

Although multiple cDNA clones encoding Pi transporters have been isolated and characterized from several plants, a vacuolar Pi transporter has not been identified yet.

RESPONSES TO PI DEFICIENCY

Plants have evolved a series of morphological and metabolic adaptations that are triggered by phosphate deficiency. These adaptations are aimed at increasing the acquisition of this vital but poorly available nutrient from the soil as well as sustain plant growth and survival under low P availability. At the morphological level, the architecture of the root system is modified, with an increase in root/shoot ratio, increase in the length and density of root hairs, as well as proliferation of lateral roots. In some plants, there is also a reduction in the gravitropism of roots. All these changes improve the capacity of the plant root system to better explore and mine the soil for phosphate. Under Pi deprivation stress, the roots enhance secretion of protons or organic acids in order to enhance the solubilization of insoluble inorganic phosphate complexes. Roots also release phosphatases and nucleases to acquire phosphate from organic sources. The kinetics of Pi transport into the root and across the plant is modified through changes in transporter abundance and affinity for Pi. Biochemical pathways less dependant on intracellular Pi level are activated, such as the replacement of phospholipids by sulfolipids. Although several of these responses to Pi deprivation stress have been documented for a number of agriculturally important plants, the use of Arabidopsis offers the opportunity to get a better understanding of the genetic and biochemical basis of Pi-stress responses.

MUTANTS AFFECTED IN PHOSPHATE ACQUISITION AND HOMEOSTASIS

The screening and analysis of Arabidopsis mutants affected in mineral nutrition has been a fruitful approach for the study of genes involved in the uptake and metabolism of ions, such as iron and nitrate (Crawford, 1994). To date, at least 6 mutants have been characterized in Arabidopsis that affect Pi uptake and metabolism. These mutants can be classified in two major classes: mutants primarily affected in components of Pi acquisition and distribution in plants, and mutants affected either in the perception or the response to Pi deprivation stress.

The first genetic screen for mutants affected in Pi acquisition was done by Poirier et al. (1991). Analysis of total P present in leaves of 2000 EMS-mutagenized plants enabled the isolation of a single plant having 3–4 times less total P than wild type. The mutant, named pho1, showed classical phenotypes of Pi deficiency, including reduced growth and accumulation of anthocyanins (Figure 15). The amount of free Pi present in leaves of the mutant grown in soil was 20 times lower than wild type, while the amount of P present in lipids, nucleic acid and other esters were largely unchanged. In plants grown in either a hydroponic culture system or in agar, the amount of Pi in leaves was also strongly reduced in the mutant, while Pi in roots was similar to wild type (Delhaize and Randall, 1995; Poirier et al., 1991). The quantity of other ions was mainly unchanged in the pho1 mutant, with the exception of a relatively small reduction in potassium content in leaves of the mutant. The rate of phosphate uptake into the roots was similar between mutant and wild type over a wide range of external phosphate concentration (0.32 to 1000 µM). In contrast, transfer of the Pi from the roots to the shoot was reduced to 3–10% of the wild type levels in plants grown in media with less than 200 µM external Pi. However, Pi that was delivered to the xylem via the hypocotyl (thus bypassing the root system) was transferred to the shoot at a normal rate. The defect in phosphate transfer from the roots to the shoots could be overcome by providing higher levels of Pi in the medium. Transfer of sulfate from the roots to the shoot was essentially normal in the pho1 mutant. Together, these results indicated that the pho1 mutant was impaired in a protein involved in the loading of Pi to the xylem in the root. This mutant provided a genetic evidence for the involvement of distinct protein(s) involved in the uptake of Pi from the external solution into the root and the subsequent transfer of the Pi to the xylem vessel for its transport to the shoot.

The PHO1 gene has recently been identified by a mapped-based positional cloning strategy (Hamburger et al., 2002). The protein is divided in two domains, the N-terminal half is mainly hydrophilic while the C-terminal half has 6 potential membrane-spanning domains. Although the presence of membrane-spanning domains is a feature expected for an integral membrane protein, such as an ion transporter, the structure of PHO1 appears quite distinct from members of the major facilitator superfamily, which typically have twelve membrane-spanning domains grouped in two clusters of six (Pao et al., 1998). Strikingly, PHO1 shows no homology to characterized solute transporters, including the family of Pht1 H⁺/Pi co-transporters identified in plants and fungi (see paragraph phosphate transporters). PHO1 shows, however, highest homology to the Rcm1 mammalian receptor for xenotropic murine leukemia retrovirus (Battini et al., 1999). Although the function of Rcm1 is still unknown, it is interesting to note that several receptors for mammalian retroviruses have been found to be solute transporters, such as transporters for amino acids and Na⁺-Pi cotransporters (Kavanaugh and Kabat, 1996; Rasko et al., 1999). PHO1 also shows significant homology to the Saccharomyces cerevisiae Syg1 protein involved in the mating pheromone signal transduction pathway (Spain et al., 1995). PHO1 was found to be mainly expressed in the roots and is only weakly up-regulated by Pi deprivation stress (Hamburger et al. 2002). Promoter-GUS fusion experiments revealed predominant expression of the PHO1 promoter in the stelar cells of the root and the lower part of the hypocotyl (Figure 16). This pattern is consistent with the role of PHO1 in loading Pi to the xylem (Poirier et al., 1991). Expression of PHO1 in yeast and Xenopus oocytes failed to reveal a Pi transport activity associated with protein. It is possible that the heterologous systems used to express PHO1 do not allow a functional reconstitution of transporter activity. Alternatively, it is possible that PHO1 represents only one subunit of a multi-subunit transporter and that the presence of other unidentified components is essential for Pi transport activity. Finally, it is also possible that PHO1 may not directly be involved in Pi transport but that it indirectly influences the activity of a plasma membrane Pi transporter. Clearly further studies are required to reveal the mode of action of PHO1 in Pi loading to the xylem. Interestingly, the Arabidopsis genome contains 10 additional genes showing homology to PHO1 (Hamburger et al., 2002). It is likely that several members of this family may also play a role in Pi homeostasis.

Delhaize and Randall (1995) have isolated a second mutant affected in Pi acquisition. Approximately 100 000 EMS mutagenized plants were grown in soil and screened visually for phenotypes consistent with nutrient toxicity or deficiency. Approximately 2000 plants identified in this first screen were subsequently analyzed by X-ray fluorescence spectrometry (Delhaize et al., 1993). From this screen a novel allele of pho1 was identified as well as two alleles of a novel mutant named pho2. This mutant was characterized by a 2- to 4-fold increase in the amount of Pi in leaves compared to wild type while the level of Pi in roots was unchanged (Delhaize and Randall, 1995). There was also an increase of Pi in stems, siliques, and seeds. The Pi toxicity symptoms of the pho2 mutant were apparently affected by the transpiration rate. Under high transpiration conditions in soil-grown plants, the margins of the pho2 leaves became necrotic, and under extreme conditions, the whole leaf senesced. Pho2 mutant plants were approximately 50% smaller (by fresh weight) compared to wild type. Although toxicity symptoms were influenced by transpiration rates, the higher Pi level in leaves of pho2 were observed even in plants grown submerged in liquid solution, indicating that the higher accumulation of Pi in leaves was independent of transpiration.

Analysis of the kinetics of Pi uptake into whole seedlings revealed that the pho2 mutant had about a twofold greater Pi uptake rate than wild type plants under Pi-sufficient conditions. Furthermore, a greater proportion of the acquired Pi was located in the shoot in pho2 (Dong et al., 1998). This difference in the Pi uptake rate between pho2 and wild type seedlings was abolished when shoots were removed, clearly indicating that the mutation primarily affected shoot Pi acquisition. Although pho2 mutants could recycle Pi from the shoots to the roots through the phloem, the proportion of shoot Pi translocated to the roots was less than half of that found for the wild type plants. Two mechanisms have been proposed to explain the phenotype of the pho2 mutant. One model is that pho2 has a specific impairment in a leaf Pi transporter that would influence the phloem transport of Pi between shoots and roots. In this context, it is interesting to note that the pho2 locus does not map close to any of the PHO1 homologues or of the low affinity shoot-specific Pi transporter Pht2;1 gene, which has been hypothesized to play a role in Pi loading of shoot organs (Daram et al., 1999). Alternatively, pho2 could have a defect in a protein involved in sensing leaf Pi concentration, which in turn regulates the activity or expression of Pi transporters in the shoot, leading to uncontrolled accumulation of Pi under Pi-sufficient conditions (Dong et al., 1998). A clearer picture of the role of PHO2 in Pi homeostasis will require the eventual cloning and characterization of the gene.

Four Arabidopsis mutants have been isolated with genetic screens based on the induction of Pi-scavenging proteins and genes, such as phosphatases and nucleases, in response to Pi deprivation stress. For example, the pup1 and pho3 mutants were isolated by screening for an alteration in the induction of root acid phosphatases in plant grown in Pi-deficient media. Acid phosphatases are detected through the conversion of the colorless substrate 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) to a blue compound following cleavage of the phosphate group. Screening of mutagenized plants growing in Pi-deficient agar medium containing BCIP lead to the isolation of three mutants showing reduced blue staining of the roots. Of those mutants, the pup1 (phosphatase underproducer) mutant has been described (Trull and Deikman, 1998). The mutant was shown to be missing only one (160 kD) of the multiple acid phosphatase found in Arabidopsis. The mutant was otherwise normal in most of the classical responses to Pi deprivation stress, including up-regulation of phosphatase activity, accumulation of anthocyanins, change in root growth and morphology, as well as changes in Pi distribution between shoot and roots (Trull and Deikman, 1998). The overall growth pattern of the mutant was largely indistinguishable from wild type. It is likely that the PUP1 gene codes for the 160 kD acid phosphatase.

In a separate screen based on overlaying roots of Pi-deficient plants with an agar layer containing BCIP, a mutant named pho3 was isolated (Zakhleniuk et al., 2001). The mutant had lower level of acid phosphatase activity in roots of Pi-deficient and Pi-sufficient plants compared to wild type, while the shoot acid phosphatase activity was higher in the mutant under the same conditions. The changes in phosphatase activity were not correlated to a particular isoform. The pho3 mutant showed a typical Pi-stress response when grown in soil, including poor growth, production of anthocyanins, and increased accumulation of starch. In the rosette leaves of mature soil-grown plants, the total P amount in the mutant was 50% of wild type levels. In plants grown in agar media containing 0.63 mM Pi, the total P level was reported to be 25% lower than the wild type level in the shoot and about 65% lower in the root. This decrease in total P is, however, not reflected by a decrease in level of free Pi in roots (Lloyd et al., 2001). It is thus unclear at this point which of the various Pi-containing components of the plant cells is reduced in the mutant. The pho3 mutant shares several of the characteristics of the pho1 mutant, including overall reduced growth, lower P level and accumulation of anthocyanins. However, in contrast to pho1, the roots of pho3 mutant appear to show a reduced total P content compared to wild type. As no complementation test has been done between pho1 and pho3, it is still unclear whether the PHO3 gene is distinct from PHO1.

Nucleases are some of the enzymes that are induced under Pi deprivation stress and that are involved in scavenging Pi from the environment. Arabidopsis has been shown to be able to grow in media containing DNA as the main Pi source, indicating that the release of nucleases and phosphatases is sufficient to acquire Pi from extracellular DNA (Chen et al., 2000). A genetic screen has been based on this ability of Arabidopsis to acquire Pi from exogenous DNA. A screen of 50 000 mutagenized seedlings lead to the isolation of 22 mutant lines that showed severe growth defect in media containing DNA as the only Pi source, but which recovered on medium containing soluble inorganic Pi. Further analysis of nine of these lines revealed different levels of growth retardation of roots in media containing either DNA, RNA or inorganic Pi. One line, named psr1 (phosphate starvation response) showed a significant reduction in the level of several isoforms of acid phosphatases as well as of both RNS1 and RNS2 ribonucleases in plants grown in Pi-deficient medium, indicating that the mutant was affected in the production of several enzymes normally induced by Pi deprivation stress (Chen et al., 2000). Genetic analysis indicated that psr1 behaved as a single recessive mutation. These data are consistent with the hypothesis that psr1 may be a mutant in a component of the Pi starvation response pathway in higher plant.

The phr1 (phosphate starvation response) mutant was isolated by screening a mutagenized population containing the GUS reporter gene under the control of the promoter of the AtIPS1 gene (AtIPS1::GUS) (Rubio et al., 2001). In wild type plants transformed with the AtIPS1::GUS construct, GUS expression is strongly induced in Pi-deficient plants (Martin et al., 2000). The phr1 mutant displayed a strong reduction in GUS expression following Pi deprivation stress, as well as a reduction in the expression of several genes normally induced by Pi starvation, such as AtIPS1 and At4 genes, the high affinity Pi transporter Pht1;1 (AtPT1), the acid phosphatase AtACP5, the ribonuclease RNS1, and AtIPS3, a gene encoding a protein of unknown function. Furthermore, the phr1 mutant was deficient in the accumulation of anthocyanins normally induced by Pi deficiency, although anthocyanins accumulation induced by jasmonic acid and abscisic acid was normal. Also normal in the mutant was the increase in density and length of root hairs induced by Pi deprivation stress (Rubio et al., 2001). The PHR1 gene has been cloned and was shown to be a transcription factor (see below for further details). PHR1 thus represents the first identified element of the phosphate starvation signaling in higher plants (Rubio et al., 2001).

In addition to the forward genetic approach of mutant screening and gene isolation, the cloning and characterization of numerous Pi transporters opens the way for the analysis of mutants in these transporters by reverse genetic approaches using T-DNA or transposon tagging. A mutant in the PHT2;1 gene encoding a low affinity Pi transporter located in the chloroplast has recently been described (Versaw and Harrison, 2002). Although only a single copy of the PHT2;1 gene is present in the Arabidopsis genome, the phenotype of the null mutant was subtle. Mutants grown in high Pi media had a slightly reduced rosette size (70–80% of wild type ) and reduced Pi content (74% of wild type). Under low-Pi conditions, no difference were observed in fresh weight of mutants and wild type. However, root Pi content of the mutant was 50–70% that of the wild type, while leaf Pi content of the mutant was 130–150% that of the wild type. Redistribution of Pi from old leaves towards young leaves in plants that were switched from high Pi media to low Pi media was reduced in mutants. Together, these results indicate a deficiency in Pi retranslocation under Pi-deprivation stress in the mutant. Expression analysis of several genes known to be up-regulated by Pi-deprivation stress indicated that the response was stronger in the mutant compared to wild type.

ANALYSIS OF PHOSPHATE ACQUISITION USING ARABIDOPSIS ACCESSIONS

In addition to the uses of mutants in a forward genetic approach using physical, chemical or biological (T-DNA or transposon) mutagenesis, exploitation of the natural variations present among wild type accessions of Arabidopsis can be used to gain insight into the genetic components of Pi acquisition (Alonso-Blanco and Koornneef, 2000). Krannitz et al. (1991a, b) measured a number of parameters associated with Pi acquisition in plants that were grown on Pi-depleted media for 16 days. These parameters included initial Pi uptake rates, the external Pi concentration at which there is no net Pi uptake (referred as C_min), root length and mass, as well as root/shoot ratio. Highly significant differences were recorded among the accessions for all measured parameters.

Recently, Narang et al. (2000) have used the natural variations found in accessions to analyze the morphological and physiological parameters which have the greatest impact on the ability of Arabidopsis to acquire Pi from a sparingly soluble source. An initial set of 36 accessions were grown on 0.4% agarose-solidified media with Pi supplied either as soluble potassium phosphate (KH₂PO₄) or as the sparingly soluble hydroxylapatite (Ca₅(PO₄)₃OH). The phosphate acquisition efficiency (PAE) was measured as the mean of the (phosphate content in shoot for plants grown in media containing 0.5 g/L hydroxylapatite / phosphate content in shoot for plants grown in media containing 1 mM potassium phosphate). The accessions C24, Co and Cal exhibited the highest PAE while Col-0 and Te exhibited the lowest PAE. Using these five accessions, significant differences were recorded for root morphology (root length and mass, root hair length and density), phosphate uptake kinetics, rhizosphere acidification, organic acid release, and the ability of the roots to penetrate substrates of different density (agarose concentrations of 0.15% and 0.4%). The high PAE of accessions C24 and Co was mediated by a root system having long and dense root hairs, high Pi uptake per unit root length and high substrate penetration ability. In contrast, the accessions Col-0 and Te having a low PAE produced roots that had a low substrate penetration, short and sparse root hairs, and low uptake efficiency per unit root length. The accession Cal, was intermediate for these parameters, but displayed in addition an enhanced rhizosphere acidification capability. This study highlighted the importance of the architecture of the root system, and in particular of the root hairs, in the acquisition of Pi from sparingly soluble source.

F1 hybrids resulting from a cross between accessions C24 and Col-0 showed superior ability to acquire Pi from hydroxylapatite when compared to either parent, representing a classical case of heterosis or hybrid vigor (Narang and Altmann, 2001). The hybrid inherited the long root hair length and high phosphate transporter expression from C24 while it inherited the long root length trait of Col-0. The root hair density of the hybrid was intermediate between the parents. The data suggests that the superiority of the F1 hybrid is due to the accumulation of favorable dominant genes at numerous loci.

THE PHOSPHATE REGULON IN PLANTS

As described above, plants have evolved a complex series of adaptive responses to Pi deprivation, which includes secretion of Pi-mobilizing enzymes, such as phosphatases, and changes in the Pi transporters. Similar adaptive changes to Pi deprivation also occur in bacteria and fungi. The mechanisms that control the adaptation of Escherichia coli and S. cerevisiae to Pi limitation have been extensively studied and numerous components of the regulatory networks linking Pi levels to gene activation have been described. Genetic screens aimed at understanding how Pi deprivation leads to activation of secreted phosphatases in both E. coli and S. cerevisiae have been central to this research.

In E. coli, the PHO regulon comprises a set of at least 15 genes involved in the acquisition of Pi (Figure 17) (Torriani, 1990). These genes are controlled by a two-component regulatory system encoded by the phoB-phoR operon. PhoR is a histidine protein kinase located in the inner membrane and acts as a Pi sensor. Under Pi deprivation, phoR autophosphorylates itself, which in turn leads to phosphorylation of phoB, the positive regulatory protein. The phosphorylated phoB binds to a specific region (the Pho box) of the promoter of each gene of the Pho regulon and activates transcription.

In S. cerevisiae, the Pho regulon comprises a set of at least 22 genes (Ogawa et al., 2000). The proteins Pho80, Pho81, Pho85 and Pho4 are the key factors controlling expression of the genes under the Pho regulon (Figure 18) (Lenburg and O'Shea, 1996). Pho4 is the transcription factor that binds a consensus nucleotide (Pho4 box) found in the promoters of genes forming the regulon, which includes the Pi transporters Pho84 and Pho 89 as well as the acid phosphatase Pho5. The level of phosphorylation controls activation of Pho4. Whereby the hyperphosphorylated Pho4 is mainly located in the cytoplasm, the hypophosphorylated form is found predominantly in the nucleus where it can act to activate transcription. Phosphorylation of Pho4 is controlled by the Pho80-Pho85 cyclin-cyclin dependent protein kinase complex (CDK), which is itself controlled by the Pho81 CDK inhibitor. Under low Pi conditions, the Pho81 protein inhibits the Pho80-Pho85 complex, preventing phosphorylation of Pho4, and thus activating transcription of the regulon. Alternatively, under high Pi availability, the Pho80-Pho85 kinase phosphorylates Pho4. Pho4 interacts with Pho2 which itself activates transcription of only a subset of Pho4-dependent genes.

By analogy to E. coli and S. cerevisiae, elucidation of the signaling components linking Pi deprivation to the regulation of genes involved in metabolic and developmental adaptation of plants to nutrient deficiency has been tackled by a genetic approach. As described above, genetic screens have been done in Arabidopsis to isolate mutants that are affected in the response of plant to Pi deficiency. The pup1 and pho3 mutants were isolated based on a reduced activation of extracellular root acid phosphatases following Pi deficiency (Trull and Deikman, 1998; Zakhleniuk et al., 2001). Whereas pup1 is likely to be defective in a gene encoding one particular acid phosphatase, the defect in pho3 is unclear and could involve a Pi transporter. In contrast, the Arabidopsis psr1 mutant, which was found defective in the production of both nucleases and acid phosphatases under Pi deficiency, is likely to encode a gene involved directly in the Pho signal transduction pathway (Chen et al., 2000). The Arabidopsis PSR1 gene has not yet been isolated.

The first gene involved in the Pi-stress signal induction pathway in photosynthetic eukaryotes has been identified in Chlamydomonas reinhardtii (Wykoff et al., 1999). The C. reinhardtii mutant psr1 was isolated from a screen based both on the survival of cells after exposure to high concentration of radioactive ³²Pi as well as on the reduced activation of acid phosphatase following Pi deprivation (Shimogawara et al., 1999). The mutant also fails to increase the rate of Pi transport upon exposure to Pi limitation. The C. reinhardtii PSR1 gene was cloned and shown to encode a protein containing at the N-terminal half a region similar to the MYB DNA-binding domain, and at the C-terminal half a glutamine-rich sequence characteristic of transcriptional activators (Wykoff et al., 1999). The level of PSR1 mRNA transiently increases up to 13-fold 8 hours after a shift to Pi-free media. The level of PSR1 protein increases at least 10-fold one day after Pi starvation. Immunolocalization of the protein demonstrated that it was located in the nucleus in both Pi-sufficient and Pi-deficient cells. Together, these data support the conclusion that PSR1 is a transcriptional factor and that under Pi deprivation stress the increased accumulation of PSR1 in the nucleus leads to activation of genes forming the Pho regulon of C. reinhardtii.

A functional homologue of the C. reinhardtii PSR1 gene has been identified in Arabidopsis (Rubio et al., 2001). The Arabidopsis phr1 mutant was isolated based on the lack of induction of the promoter of the AtIPS1 gene under Pi deprivation stress, as detected by a AtIPS1::GUS construct. The mutant also fails to activate a number of Pi starvation-induced genes and metabolic responses, including accumulation of anthocyanin. The Arabidopsis PHR1 protein shares with the C. reinhardtii PSR1 the presence of a MYB DNA-binding domain and a predicted coiled-coil domain, indicating potential dimerization of the protein with itself or with other proteins. Analysis of the binding of PHR1 to sequences of the AtIPS1 promoter indeed revealed that PHR1 binds as a dimer to an imperfect palindromic sequence. This PHR1-binding sequence is present in the promoter of several genes that are up-regulated following Pi deprivation stress, such as the promoter of the acid phosphatase AtACP5, the ribonuclease RNS1 and the Pi transporter Pht1;1 (AtPT1) (Rubio et al., 2001). PHR1 was found located in the nucleus in both Pi-deficient and Pi-sufficient cells. However, in contrast to the C. reinhardtii PSR1 gene, the PHR1 gene is only weakly up-regulated by Pi deprivation stress. It is concluded that PHR1 is a transcription factor participating in the Pho regulon system of higher plants. Interestingly, the phr1 mutant is not defective in the increase in root hair density and length in response to Pi deprivation stress. Thus, only a subset of Pi starvation responses is mediated via PHR1. At least 15 other proteins are present in Arabidopsis that shares features of the MYB-coiled-coil protein family, raising the possibility that other members of this family may also contribute to the response of plants to Pi deficiency.

Cytokinins have been shown to repress the induction of several genes in roots by Pi deficiency, including AtIPS1 and At4 (Martin et al., 2000). Screening for Arabidopsis plants which have high AtIPS1 gene expression when grown in Pi-deficient media containing kinetin identified several mutants which were allelic to the cre1 mutant (Franco-Zorrilla et al., 2002). CRE1 encodes for a cytokinin receptor that may trigger a cascade of phosphorelay reactions upon perception of the hormone. In addition to AtIPS1, several Pi-starvation response genes showed reduced cytokinin sensitivity in the cre1 mutants, such as AtACP5 and AtPT1. CRE1 was down-regulated by Pi starvation and induced by cytokinins. These results indicate an important contribution and integration of the cytokinin two-component signaling circuitry with the regulation of Pi starvation responses in roots.

Analysis and identification of the promoter sequences and DNA binding proteins that are involved in the regulation of genes induced by Pi deprivation stress is still at an early stage in plants. Proteins binding to the promoter region of the tomato TPSI1 gene and of the Arabidopsis Pi transporter Pht1;4 (AtPT2) were analyzed by DNA mobility-shift assay (Mukatira et al., 2001). These two genes are specifically up-regulated by Pi deprivation. Two regions of these promoters specifically bound nuclear protein factors only from Pi-sufficient plants and not from Pi deprivation-stressed plants. Similar findings have also been found for the B. nigra ß-glucosidase gene induced under Pi deprivation stress (Malboobi et al., 1998). These results indicate that induction of genes by Pi deprivation may be, at least in part, mediated by the decreased interaction of repressor proteins with the promoters of these genes.

Phosphite (also referred as phosphorous acid or phosphonate, HPO₃^2-) is an isotere of the Pi anion in which one oxygen is replaced by a hydrogen. Phosphite is rapidly absorbed by plant cells and competes with Pi, indicating that the same transporters are involved in the entry of Pi and phosphite (MacDonald et al., 2001). However, in contrast to Pi, phosphite cannot be covalently assimilated into organic compounds, as enzymes catalyzing the transfer of Pi do not accept phosphite. Phosphite has been shown to efficiently suppress many of the Pi starvation response observed in Arabidopsis and Brassica nigra. Thus, growth of plant on media without Pi but containing phosphite do not respond to Pi deprivation with changes in root/shoot ratio, root hair formation, anthocyanin accumulation, activities of phosphate-scavenging enzymes such as RNAse, DNAse and phosphatase, as well as changes in the expression of several genes normally induced by Pi-deprivation stress, including AtACP5, AtPT1, AtPT2 and At4 (Ticconi et al., 2001, Varadarajan et al., 2002). These experiments indicate that the cell sensors and/or signal transduction components involved the coordinated response to Pi-deprivation stress actually perceive phosphite as Pi even though the former compound is not metabolized, thus depriving the plants of the mechanisms needed to adapt to this stress. Phosphite may be used as a useful tool for the search of genes involved in the Pi starvation signal transduction pathway (Abel et al., 2002).

CONCLUSIONS

Although progress on the knowledge of the genes and proteins implicated in Pi transport and homeostasis has accelerated in recent years, still much remains to be learned. For example, although the whole genome of Arabidopsis has now been sequenced, we still do not know any gene that is involved in the transport of Pi across the vacuole, an important aspect of Pi homeostasis. Combination of both classical forward genetics along with reverse genetics and all the tools associated with functional genomics should allow to gain new insights into the structure and regulation of Pi transport across all membranes, as well as on the signal transduction pathways and biochemical regulation involved in Pi deprivation stress response and Pi homeostasis. Ultimately, it is hoped that our knowledge on Pi transport and metabolism in Arabidopsis and other plants will permit the development of crops capable of high yield with reduced additional input of Pi in soil.