The Pht1;9 and Pht1;8 transporters mediate inorganic phosphate acquisition by the Arabidopsis thaliana root during phosphorus starvation


  • E. Remy,

    1. Instituto Gulbenkian de Ciência, Rua da Quinta Grande, 6, 2780-156 Oeiras, Portugal
    Search for more papers by this author
  • T. R. Cabrito,

    1. Institute for Biotechnology and BioEngineering (IBB), Center for Biological and Chemical Engineering, Department of Bioengineering, Instituto Superior Técnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisbon, Portugal
    Search for more papers by this author
  • R. A. Batista,

    1. Instituto Gulbenkian de Ciência, Rua da Quinta Grande, 6, 2780-156 Oeiras, Portugal
    Search for more papers by this author
  • M. C. Teixeira,

    1. Institute for Biotechnology and BioEngineering (IBB), Center for Biological and Chemical Engineering, Department of Bioengineering, Instituto Superior Técnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisbon, Portugal
    Search for more papers by this author
  • I. Sá-Correia,

    1. Institute for Biotechnology and BioEngineering (IBB), Center for Biological and Chemical Engineering, Department of Bioengineering, Instituto Superior Técnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisbon, Portugal
    Search for more papers by this author
  • P. Duque

    1. Instituto Gulbenkian de Ciência, Rua da Quinta Grande, 6, 2780-156 Oeiras, Portugal
    Search for more papers by this author

Author for correspondence:
Paula Duque
Tel: +351 21 4464630


  • The activation of high-affinity root transport systems is the best-conserved strategy employed by plants to cope with low inorganic phosphate (Pi) availability, a role traditionally assigned to Pi transporters of the Pht1 family, whose respective contributions to Pi acquisition remain unclear.
  • To characterize the Arabidopsis thaliana Pht1;9 transporter, we combined heterologous functional expression in yeast with expression/subcellular localization studies and reverse genetics approaches in planta. Double Pht1;9/Pht1;8 silencing lines were also generated to gain insight into the role of the closest Pht1;9 homolog.
  • Pht1;9 encodes a functional plasma membrane-localized transporter that mediates high-affinity Pi/H+ symport activity in yeast and is highly induced in Pi-starved Arabidopsis roots. Null pht1;9 alleles exhibit exacerbated responses to prolonged Pi limitation and enhanced tolerance to arsenate exposure, whereas Pht1;9 overexpression induces the opposite phenotypes. Strikingly, Pht1;9/Pht1;8 silencing lines display more pronounced defects than the pht1;9 mutants.
  • Pi and arsenic plant content analyses confirmed a role of Pht1;9 in Pi acquisition during Pi starvation and arsenate uptake at the root–soil interface. Although not affecting plant internal Pi repartition, Pht1;9 activity influences the overall Arabidopsis Pi status. Finally, our results indicate that both the Pht1;9 and Pht1;8 transporters function in sustaining plant Pi supply on environmental Pi depletion.


In plants, phosphorus (P) is an essential macronutrient that plays a pivotal role in many basic physiological and metabolic processes. It constitutes a structural element of key molecules, such as phospholipids and nucleic acids, a component of energy transfer reactions and signal transduction events, as well as an activator of metabolic intermediates and a regulator of enzyme activity (reviewed in Marschner, 1995). Plants acquire P exclusively in its ionic, fully oxidized form as inorganic phosphate (Pi) at the root–soil interface. Once imported into the root epidermis and cortex, Pi is subsequently loaded into the xylem for transfer, via the transpiration stream and root pressure, to the shoot, where Pi is ultimately allocated to the aerial portions of the plant (Poirier et al., 1991; Delhaize & Randall, 1995). Despite its widespread occurrence in the environment, Pi is unevenly distributed and relatively immobile in the soil, mainly because of its adsorption onto organic substances and precipitation with cations (Lopez-Bucio et al., 2000; Hirsch et al., 2006), rendering Pi often limiting to sustain continuous plant growth and development (reviewed in Raghothama, 1999).

To overcome low Pi availability, plants have evolved a complex array of tightly controlled adaptive mechanisms intended to maintain P homeostasis. In addition to metabolic conversion, inadequate Pi supply triggers the remobilization of internal Pi within the plant coordinated with an enhanced acquisition of external Pi (reviewed in Lin et al., 2009; Rouached et al., 2010). Recycling of internal Pi is facilitated by the alteration of partitioning between cell compartments, such as the export of stored Pi from the vacuole to preserve a relatively stable cytoplasmic Pi concentration (Mimura et al., 1990). At the whole-plant level, Pi deficiency is accompanied by Pi translocation from the shoot to root tissues and from senescent to young expanding leaves, both processes requiring transfer via phloem vessels (Versaw & Harrison, 2002; Aung et al., 2006; Chiou et al., 2006; Nagarajan et al., 2011). Together with more specialized mechanisms, such as symbiosis with mycorrhizal fungi (reviewed in Karandashov & Bucher, 2005), plants have developed three main mechanisms to maximize the ability of the root to absorb Pi from the soil. First, plants respond to Pi deficiency by allocating more carbon to the root (reviewed in Hermans et al., 2006) and by remodeling the spatial configuration of the root system (reviewed in Peret et al., 2011). The plant Arabidopsis thaliana, particularly the Columbia (Col-0) ecotype, has proven to be a useful model to dissect the alteration of root morphology and architecture triggered by Pi deficiency (Williamson et al., 2001; Sanchez-Calderon et al., 2005). Low Pi availability in Arabidopsis promotes the development of a highly branched root system to the detriment of the primary root (PR), characterized by the stimulation of lateral root (LR) primordia formation and LR emergence (Linkohr et al., 2002; Lopez-Bucio et al., 2002; Perez-Torres et al., 2008). Moreover, PR growth inhibition is accompanied by a drastic increase in root hair density and elongation (Bates & Lynch, 1996; Williamson et al., 2001). Together, these responses enable the optimization of soil exploration and of the absorptive surface area of the root (reviewed in Lopez-Bucio et al., 2003). Another strategy developed by plants to increase Pi uptake relies on the secretion of Pi-solubilizing root exudates, including organic acids and enzymes, such as phosphatases and nucleases, in order to liberate Pi from inaccessible soil complexes (del Pozo et al., 1999; Haran et al., 2000; Diatloff et al., 2004). Finally, enhanced Pi acquisition is also achieved by the induction of a high-affinity transport activity that determines the Pi import capacity of the root cells, an adaptive response well conserved among the plant kingdom (reviewed in Rausch & Bucher, 2002; Smith et al., 2003).

All steps in plant Pi assimilation, from its initial uptake by the root to its subsequent allocation among different tissues, partitioning between subcellular compartments and mobilization within the plant, require the concerted action of multiple membrane Pi transport systems. The Arabidopsis genome contains at least five phylogenetically distinct classes of integral membrane proteins possessing Pi transport activity (Muchhal et al., 1996; Daram et al., 1999; Takabatake et al., 1999; Knappe et al., 2003; Guo et al., 2008). The 16 transporters belonging to the plastidic Phosphate Translocator (pPT) group function as antiport systems using Pi and phosphorylated C3, C5 or C6 compounds as counter substrates (Knappe et al., 2003), whereas the four other phosphate transporter families have been named Pht1, Pht2, Pht3 and Pht4. The sole member of the Pht2 family, Pht2;1, shares similarity with the PHO-4 transporter of Neurospora crassa and the closely related Pho89 transporter of Saccharomyces cerevisiae, both of which possess Na+/Pi symport activity (Versaw & Metzenberg, 1995; Martinez & Persson, 1998). In Arabidopsis, the Pht2;1 gene encodes a low-affinity Pi transporter located in the chloroplast inner envelope membrane that mediates Pi translocation within the aerial parts of the plant and influences whole-plant Pi allocation (Daram et al., 1999; Versaw & Harrison, 2002). Members of the Pht3 family are predicted to encode mitochondrial transporters assumed to catalyze Pi exchange between the matrix and the cytosol via Pi/H+ symport or Pi/OH antiport activities (Takabatake et al., 1999), but none of the three Pht3 transporters identified in Arabidopsis has been functionally characterized. The Pht4 family includes six transporters that share similarity with the mammalian SLC17 type I Pi transporters. Systematic characterization of all six members by heterologous expression in yeast, as well as expression and subcellular localization studies, has suggested that Pht4 transporters mediate low-affinity Pi transport between the cytosol and chloroplasts or heterotrophic plastids (Guo et al., 2008). Only one member, Pht4;6, localizes to the Golgi apparatus, where it determines salt tolerance by facilitating the export of Pi generated as a byproduct of glycosylation processes (Cubero et al., 2009).

As all Pi transporters described above are thought to participate mainly in Pi transfer across internal cellular membranes, they are unlikely to mediate external Pi uptake, which represents the primary and crucial step in plant Pi acquisition. This function has been traditionally attributed to the Pht1 family, which belongs to the Major Facilitator Superfamily (MFS) and groups nine closely related members in Arabidopsis, Pht1;1–Pht1;9 (reviewed in Pao et al., 1998; Nussaume et al., 2011). Plant Pht1 transporters share extensive homology with each other and have been identified on the basis of their similarity with the S. cerevisiae high-affinity Pi transporter Pho84, which mediates Pi import across the plasma membrane and catalyzes Pi/H+ symport activity (Bun-Ya et al., 1991). Like their yeast counterpart, the Pht1 genes characterized so far also encode high-affinity, plasma membrane-localized Pi transporters in Arabidopsis (Muchhal et al., 1996; Mitsukawa et al., 1997; González et al., 2005; Bayle et al., 2011). Genome-wide transcriptional analyses indicate that all Pht1 genes, except for Pht1;6, are predominantly expressed in root tissues and highly and reversibly induced by Pi starvation (Misson et al., 2005; Morcuende et al., 2007; Thibaud et al., 2010), in agreement with more detailed expression analyses involving promoter–reporter gene fusion experiments (Karthikeyan et al., 2002; Mudge et al., 2002). This transcriptional activation in response to Pi deficiency strongly suggests the involvement of the corresponding transporters in Pi acquisition from the environment, a function that has been validated for the closely related members Pht1;1 and Pht1;4. In roots, both genes are strongly expressed in the epidermal cell layer (Mudge et al., 2002), and mutations in the Pht1;1 gene cause a decrease in Pi uptake capacity of the root that results in lower Pi content of shoot tissues, whereas Pi-starved pht1;4 mutant plants display a significant reduction in Pi uptake that does not affect the shoot Pi content (Misson et al., 2004; Shin et al., 2004). Nevertheless, the double mutant exhibits a significant decrease in Pi uptake and shoot accumulation under both Pi-sufficient and Pi-deficient conditions, indicating compensatory effects between the two transporters (Shin et al., 2004). Together with specific expression in Pi-starved roots, some members of the Pht1 family are also significantly expressed in a diverse range of tissues, suggesting their potential involvement also in internal Pi distribution within the plant (Karthikeyan et al., 2002; Mudge et al., 2002). Indeed, the third Pht1 member functionally characterized so far, Pht1;5, plays a critical role in the mobilization of Pi from P source to sink organs in accordance with the P status of the plant (Nagarajan et al., 2011).

Here, we report the functional characterization of the Arabidopsis Pht1;9 transporter and show that it mediates root Pi acquisition under Pi-deprived conditions, contributing to the overall plant Pi status. Insights into the function of the closest Pht1;9 homolog, Pht1;8, are also presented.

Materials and Methods

Plant materials and growth conditions

The Arabidopsis thaliana (L.) Heynh. ecotype Colombia (Col-0) was used in all experiments. Seeds of the T-DNA insertion mutants pht1;9-1 (SALK_050730) and pht1;9-2 (SALK_013457) were obtained from the Nottingham Arabidopsis Stock Centre (NASC) (Nottingham, UK). The exact T-DNA insertion sites were confirmed using gene-specific primers (Supporting Information Table S1) and primers annealing at the T-DNA borders, which also allowed PCR-based genotyping to identify homozygous lines.

Seeds were surface sterilized and sown on Pi-free Murashige & Skoog (1962) medium (MSP11; Caisson Labs, North Logan, UT, USA) supplemented with 0.1 g l−1 myo-inositol and 0.5 g l−1 2-(N-morpholino)ethanesulfonic acid (MES), adjusted to pH 5.7 and solidified by 0.8% UltraPure Agar (A046; Caisson Labs). The Pi concentration was adjusted to either 1.25 mM (control Pi medium) or 50 μM (low Pi medium) with KH2PO4. Whenever indicated, control Pi medium was supplemented with arsenate (As(V); NaH2AsO4). After 3 d of stratification, seeds were placed in a growth chamber or in a climate-controlled growth cabinet, and seedlings were transferred to soil (‘Pot plant’ Shamrock, Bord Na Móna, Newbridge, Ireland; 90 mg l−1 P2O5) after 2–3 wk. Plants were cultivated under long-day conditions (16-h light, 22°C : 8-h dark, 18°C; 60% relative humidity).

Gene expression analysis

Total RNA was extracted from seedlings or tissues harvested from 6-wk-old plants using TRIzol® reagent (Invitrogen, New York, USA) or the innuPREP Plant RNA kit (Analytik Jena, Jena, Germany) according to the manufacturers’ instructions. DNAse-treated total RNA (600 ng) was reverse transcribed using the M-MLV reverse transcriptase (Promega, Madison, WI, USA), and first-strand cDNA (2 μl) was used as template for semi-quantitative PCR amplification with gene-specific primers (Table S1). Except for Pht1;9, the primers used to detect Pht1 gene expression were as described by Shin et al. (2004). The housekeeping ROC1 and UBQ10 genes were used as loading controls for tissue and seedling analyses, respectively. PCR cycle numbers were 25 and 35 for housekeeping and Pht1 genes, respectively, except for the amplification of Pht1;9 in the overexpression lines, where 25 cycles were used. Band intensities were quantified using ImageJ software ( after separation of the DNA fragments on agarose gels, and the expression levels of the different genes were normalized against UBQ10 expression. Each semi-quantitative reverse transcription-polymerase chain reaction (RT-PCR) experiment was replicated at least three times.

Yeast manipulations

The parental S. cerevisiae strain BY4741 and the derived deletion mutant BY4741_Δpho84 (Bun-Ya et al., 1991) were used. The Arabidopsis Pht1;9 full-length coding sequence was cloned into the pGREG576 vector (Jansen et al., 2005) as described in Cabrito et al. (2009). Strains and vector were acquired from the Euroscarf collection (Frankfurt, Germany). Expression of the corresponding green fluorescent protein (GFP)-Pht1;9 fusion protein was tested by western blotting and fluorescence microscopy analyses (Cabrito et al., 2009). Pi limitation and arsenate susceptibility were assessed in ammonium phosphate basal medium (KNA) (Vargas et al., 2007) and MMB-U liquid medium (Cabrito et al., 2009), supplemented or not with the indicated concentrations of (NH4)2HPO4 or As(V), respectively. Growth media were inoculated with nonadapted exponential phase cells (OD600 nm = 0.4), and growth curves were followed by measuring culture OD600 nm. To measure the Pi intracellular content, cells grown to OD600 nm = 0.4 ± 0.04 in liquid KNA medium supplemented with 0.1 mM of (NH4)2HPO4 were harvested and analyzed using the phosphomolybdate colorimetric assay described by Ames (1966). 32Pi uptake assays were performed as described by Daram et al. (1999). Uptake values were normalized to the OD600 nm of the cell suspension, and kinetic data were analyzed by the Lineweaver–Burk method to determine the Michaelis–Menten constants. To compare the in vivo active import of protons, the external medium pH was monitored as in Vargas et al. (2007) using liquid KNA medium supplemented with 0.1 mM of (NH4)2HPO4.

Phenotypical assays

PR elongation assays were carried out by transferring 5-d-old seedlings grown on vertically oriented control Pi medium plates to plates containing the tested medium and growing them for an additional 8 d. Sixteen seedlings were analyzed per genotype per condition, and root tip positions were scored every 2–3 d. To evaluate LR parameters, seedlings were allowed to grow for an additional 12 d. PR and LR length, together with LR number, were measured on scanned images using ImageJ. Shoot biomass was determined on 3-wk-old seedlings by measuring the fresh weight of four pooled plant shoots eight times per genotype per condition. For all phenotypical assays, similar results were obtained in at least three independent experiments, and data representative from a single experiment are presented.

Generation of transgenic lines

Plant transformation was achieved by the floral dip method (Clough & Bent, 1998) using Agrobacterium tumefaciens strain EHA105.

For genomic complementation, a 4426-bp fragment encompassing the entire Pht1;9 gene was PCR amplified (Table S1) and inserted into the promoterless version of the pBA002 vector via the HindIII/XbaI restriction sites. After introduction into pht1;9 mutant plants, three complementation lines were recovered for each mutant (pht1;9-1C1–3 and pht1;9-2C1–3). Similar results were obtained after phenotypical analysis of both sets of complementation lines, and representative results for two lines for each mutant are shown.

To generate the Pht1;9 overexpression constructs, PCR fragments corresponding to the Pht1;9 transcript (5’ untranslated region (UTR) + coding sequence) and including or not the stop codon (Table S1) were inserted under the control of the 35S promoter into the native or yellow fluorescent protein (YFP) versions of the pBA002 vector, using the XhoI/AscI or XhoI/PacI restriction sites, respectively. After transformation of wild-type plants, six transgenic lines overexpressing the Pht1;9 transcript were selected. Extensive phenotypical analysis of five of these lines yielded similar results, and representative results for two lines (Pht1;9OX1 and 2) are presented. Twenty-eight lines overexpressing the Pht1;9-YFP fusion transcript were recovered and analyzed by confocal microscopy. Phenotypical analysis results for the line exhibiting the strongest YFP signal (p35S:Pht1;9-YFP) are shown.

To generate Pht1;9 RNAi lines, a 489-bp fragment encompassing nucleotides 85–573 of the Pht1;9 coding sequence was PCR amplified (Table S1) and sequentially cloned in opposite orientation via the XhoI/HindIII and EcoRI/SpeI restriction sites into the intermediate vector pSKint (Guo et al., 2003). This cassette was subsequently subcloned into the pBA002 vector under the control of the 35S promoter. After wild-type plant transformation, 12 lines were recovered and analyzed. Representative results for nine of these lines are presented.

Protoplast transfection and confocal microscopy

For protoplast transfection, the YFP gene in the pBA002:Pht1;9-YFP and pBA002:YFP plasmids was replaced with the GFP sequence. Arabidopsis protoplasts were generated as described by Yoo et al. (2007), transfected by polyethylene glycol transformation (Abel & Theologis, 1994) and analyzed by confocal microscopy.

Confocal microscopic analyses were performed with an LSM 510 Meta laser scanning microscope (Zeiss, Jena, Germany). Excitation/detection wavelengths used to monitor fluorescence were 488 nm/500–550 nm for GFP, 514 nm/535–590 nm for YFP and 458 nm/> 560 nm for autofluorescence.

Pi and arsenic content analyses

Shoot and root Pi contents were measured as described by Versaw & Harrison (2002) in pools of three and eight plants, respectively. Results presented are from two independent biological repetitions, each performed in triplicate.

To determine total arsenic content, pooled shoot tissue from 3-wk-old seedlings grown on 250 μM As(V) was placed in an oven until completely dry. Weighed material (5–10 mg) was then powdered and digested in 1 ml of 65% HNO3 (Fluka 02650; Sigma-Aldrich, St Louis, MO, USA) overnight at 80°C and diluted by the addition of 3 ml of H2O (Fluka 14211; Sigma-Aldrich). The arsenic concentration of the digests was quantified using the Atomic Absorption–Hydride Generation system (Atomic Absorption Spectrometer, Perkin-Elmer, Waltham, Massachusetts, USA, 3100, coupled to a FIAS100 Flow Injection System, Perkin-Elmer) at the Laboratório de Análises, Instituto Superior Técnico (Lisbon, Portugal) according to method 3114 described by Eaton et al. (2005). Arsenic standards for analytical calibration were from Merck KGaA (Darmstadt, Germany), and four independent samples were processed per genotype.


The Arabidopsis Pht1;9 gene is highly expressed in Pi-starved roots and encodes a plasma membrane-localized transporter

To initiate the characterization of the Arabidopsis Pht1;9 gene (At1g76430), we examined its tissue-specific expression pattern by RT-PCR. As shown in Fig. 1(a), the gene was expressed predominantly in root tissues, with transcript levels being low in young seedlings and siliques, and barely detectable in flowers. No Pht1;9 expression was observed in stems or leaves, except in senescent leaf tissues where the gene was markedly induced.

Figure 1.

Expression pattern and subcellular localization of the Pht1;9 transporter in Arabidopsis. (a) Reverse transcription-polymerase chain reaction (RT-PCR) profile of Pht1;9 expression in different tissues. Expression of the CYCLOPHILIN (ROC1) gene was used as a loading control. (b) RT-PCR analysis of Pht1;9 and Pht1;4 transcript levels in the shoot and root of 14-d-old seedlings grown on control inorganic phosphate (+Pi) or low Pi (−Pi) medium. Expression of the UBIQUITIN10 (UBQ10) gene is shown as a loading control. (c) Confocal laser scanning microscopy images of wild-type mesophyll protoplasts transiently expressing either the cytoplasmic p35S:GFP or the p35S:Pht1;9-GFP construct. (d) Confocal laser scanning microscopy images of wild-type root tips stably expressing either the cytoplasmic p35S:YFP or the p35S:Pht1;9-YFP construct. Bars, 25 μm. The YFP/GFP signals and chloroplast autofluorescence are visualized by green and red coloration, respectively. GFP, green fluorescent protein; YFP, yellow fluorescent protein.

Previous studies have demonstrated the Pi starvation responsiveness of Pht1;9 expression in root tissues (Mudge et al., 2002; Shin et al., 2004; Misson et al., 2005; Morcuende et al., 2007). To validate the experimental conditions for the analysis of the effect of Pi availability on Arabidopsis growth, seedlings were exposed to either control (1.25 mM Pi) or Pi-deprived (0 mM Pi) conditions (Fig. 1b). As expected, limited Pi supply considerably induced the expression of the Pht1;4 gene, used as a marker of Pi status in both roots and shoots (Shin et al., 2004). Similarly, this treatment was sufficient to strongly enhance Pht1;9 expression in roots, whereas no transcript was detected in shoot tissues, even under Pi starvation (Fig. 1b).

We next investigated the subcellular localization of the Pht1;9 transporter using C-terminal protein fusions with either GFP or YFP placed under the control of the strong 35S promoter. Transient expression in Arabidopsis protoplasts revealed that the Pht1;9-GFP fusion is targeted to the plasma membrane (Fig. 1c). Confocal microscopy analysis of transgenic Arabidopsis root tips stably expressing the Pht1;9-YFP fusion further confirmed the plasma membrane localization of the Pht1;9 transporter, despite the low level of YFP signal in comparison with the cytoplasmic control (Fig. 1d). Exposure of the transgenic roots to limiting Pi conditions did not allow better visualization of the fluorescence signal, as already observed by González et al. (2005) for a p35S:Pht1;1-GFP construct.

The Arabidopsis Pht1;9 transporter functions as a high-affinity Pi/H+ symporter in yeast

The Pi transport properties of the Arabidopsis Pht1;9 transporter were then analyzed in a heterologous system. To this end, a GFP-Pht1;9 fusion was introduced into the wild-type and high-affinity phosphate transporter mutant Δpho84 (Bun-Ya et al., 1991) strains of S. cerevisiae. In agreement with the plasma membrane localization observed in plant cells (see Fig. 1c,d), the GFP signal was clearly detected at the periphery of the yeast cell (Fig. 2a). Correct expression of the GFP-Pht1;9 fusion protein in yeast was also confirmed by western blotting (data not shown). These results indicate that the full-length Pht1;9 protein is targeted to the yeast plasma membrane, a prerequisite to study Pht1;9 transport properties in the Δpho84 background.

Figure 2.

Functional properties of the Arabidopsis Pht1;9 transporter in Saccharomyces cerevisiae. (a) Fluorescence microscopy images of exponential-phase wild-type yeast cells harboring either the cloning vector pGREG576 or the pGREG576_Pht1;9 plasmid after induction of recombinant protein production. (b) Comparison of the growth curves of nonadapted wild-type or Δpho84 yeast cells in inorganic phosphate (Pi)-free liquid medium unsupplemented or supplemented with 0.1 mM Pi or 1 mM Pi, harboring either the cloning vector pGREG576 or the pGREG576_Pht1;9 plasmid. Results shown are representative of three independent experiments. (c) Pi content of wild-type or Δpho84 yeast cells harboring either the cloning vector pGREG576 or the pGREG576_Pht1;9 plasmid grown in 0.1 mM Pi liquid medium (means ± SD, = 4). Statistically significant differences: ***, < 0.001; Student’s t-test. (d) Pi uptake rate of Δpho84 yeast cells harboring either the cloning vector pGREG576 or the pGREG576_Pht1;9 plasmid as a function of external Pi concentration (means ± SD, = 3). (e) External medium alkalinization promoted by energized wild-type or Δpho84 yeast cells harboring either the cloning vector pGREG576 or the pGREG576_Pht1;9 plasmid. Results shown are representative of three independent experiments. (f) Comparison of the growth curves of nonadapted wild-type or Δpho84 yeast cells in 1 mM Pi liquid medium unsupplemented or supplemented with 100 μM As(V), harboring either the cloning vector pGREG576 or the pGREG576_Pht1;9 plasmid. Results shown are representative of three independent experiments. GFP, green fluorescent protein; WT, wild type.

We then tested the ability of Pht1;9 to complement the growth defect of the yeast Δpho84 mutant (Fig. 2b). As expected, Δpho84 mutant cells carrying the empty vector grew as the wild-type strain under conditions of high Pi supply (1 mM Pi), whereas their growth rate was restricted when the Pi concentration in the culture medium was reduced to 0.1 mM or less. The growth rate of wild-type cells expressing GFP-Pht1;9 was unaffected in comparison with cells carrying the empty vector, regardless of whether the medium Pi concentration was limited or optimal. Nevertheless, the expression of GFP-Pht1;9 fully restored wild-type growth rates of Δpho84 mutant cells at micromolar Pi concentrations (Fig. 2b). To correlate this increased growth with an enhanced Pi uptake capacity of the yeast cells, we measured the free Pi content of the different strains grown under Pi-deficient conditions (Fig. 2c). Importantly, the Pi content of Δpho84 mutant cells carrying the empty vector was about half that of wild-type cells, whereas the Pi levels in Δpho84 mutant cells expressing GFP-Pht1;9 were not significantly different from those of the wild type. We next monitored direct Pi uptake by the Pht1;9 transporter in the Δpho84 mutant background. As shown in Fig. 2(d), expression of the GFP-Pht1;9 fusion significantly enhanced the rate of Pi uptake when compared with mutant cells carrying the empty vector. GFP-Pht1;9-mediated Pi uptake followed Michaelis–Menten kinetics, exhibiting apparent Km and Vmax values of 23.6 ± 3.2 μM and 4950 ± 455 pmol Pi min−1, respectively (Fig. S1). Collectively, data stemming from heterologous expression in yeast indicate that the Arabidopsis Pht1;9 transporter is able to mediate high-affinity Pi acquisition.

In S. cerevisiae, the regulation of intracellular pH is essentially sustained by the action of the plasma membrane H+-ATPase Pma1 (Serrano, 1978). The activity of this proton pump and the passive proton influx through the plasma membrane can be estimated by monitoring the pH of the external medium (Vargas et al., 2007). To substantiate the proton dependence of the Pi transport catalyzed by Pht1;9, we compared the alkalinization curves of yeast cells expressing or not the plant transporter under limiting Pi conditions. Figure 2(e) shows that cells lacking Pho84 displayed a reduced rate of extracellular medium alkalinization when compared with wild-type strains. However, and consistent with the observations for yeast growth and Pi content, expression of the GFP-Pht1;9 fusion protein was found to raise the rate of H+ efflux from Δpho84 mutant cells to a level similar to that of wild-type cells. This clearly demonstrates that the Arabidopsis Pht1;9 transporter functions as a Pi/H+ symporter, like its yeast and plant counterparts (Bun-Ya et al., 1991; Pao et al., 1998).

Arsenate (As(V)), the oxyanion of arsenic, possesses similar chemical properties to Pi, allowing this toxic ion to enter plant cells through the high-affinity Pi uptake systems (Meharg & Macnair, 1992; Shin et al., 2004; Catarecha et al., 2007). We therefore determined whether the Pht1;9 transporter influences As(V) delivery to the yeast cell. As shown in Fig. 2(f), expression of the GFP-Pht1;9 fusion induced a pronounced growth defect in yeast, as reflected by the lower biomass and increased lag phase exhibited by both wild-type and Δpho84 mutant cells in the presence of 100 μM As(V). This finding further confirms that the Arabidopsis Pht1;9 possesses Pi transport activity.

Loss of Pht1;9 function exacerbates Pi starvation responses and enhances arsenate tolerance in Arabidopsis

To gain insight into the potential role of Pht1;9 in Pi transport in Arabidopsis, we isolated two mutant alleles from the SALK collection carrying a T-DNA insertion within the Pht1;9 gene, named pht1;9-1 and pht1;9-2 (Fig. 3a). Sequence analysis of the T-DNA/genomic DNA junctions confirmed that the insertions are located 75 bp upstream of the start codon in pht1;9-1 and within the sole intron of the Pht1;9 gene in pht1;9-2. RT-PCR analysis of Pht1;9 expression in pht1;9-1 and pht1;9-2 homozygous plants revealed that both mutants lack the full-length Pht1;9 transcript, most probably resulting in null alleles (Fig. 3b).

Figure 3.

Phenotype of Arabidopsis Pht1;9 loss-of-function mutants. (a) Schematic representation of the Pht1;9 gene showing the T-DNA insertion sites in the pht1;9-1 (SALK_050730) and pht1;9-2 (SALK_013457) mutants. Boxes represent exons with untranslated regions (UTRs) in white, the line between boxes indicates the intron, and triangles depict the sites of the T-DNA insertions. Bar, 200 bp. (b) Reverse transcription-polymerase chain reaction (RT-PCR) analysis of full-length Pht1;9 transcript levels in wild-type (Col-0) and pht1;9 mutant (pht1;9-1 and pht1;9-2) 14-d-old seedlings grown on low inorganic phosphate (Pi) medium. Expression of the UBIQUITIN10 (UBQ10) gene is shown as a loading control. (c) Effect of Pi starvation on shoot biomass (FW, fresh weight; left panel) and primary root (PR) elongation (right panel) of wild-type (Col-0) and pht1;9 mutant (pht1;9-1 and pht1;9-2) seedlings (means ± SD, = 8 for shoot biomass; = 16 for PR elongation). (d) Effect of Pi starvation on lateral root (LR) development of wild-type (Col-0) and pht1;9 mutant (pht1;9-1 and pht1;9-2) seedlings. Both LR density (left panel) and total LR length (right panel) are shown (means ± SD, = 8). (e) Effect of As(V) toxicity on shoot biomass (FW, fresh weight; left panel) and PR elongation (right panel) of wild-type (Col-0) and pht1;9 mutant (pht1;9-1 and pht1;9-2) seedlings (means ± SD, = 8 for shoot biomass; = 16 for PR elongation). Statistically significant differences between pht1;9 mutants and the wild type under each condition: ***, < 0.001; **, < 0.01; Student’s t-test.

Under standard growth conditions, wild-type and pht1;9 mutant plants were indistinguishable (Fig. 3c–e and data not shown). Given the observed high Pht1;9 expression in Pi-deprived roots (see Fig. 1b) and the heterologous expression results obtained (see Fig. 2), we evaluated the response of the pht1;9 mutants to Pi starvation. When compared with seedlings grown under control Pi conditions, the shoot biomass of wild-type seedlings subjected to low Pi supply was reduced by 25%, whereas, in the same conditions, the shoot biomass of the pht1;9-1 and pht1;9-2 mutants was reduced by c. 45% (Fig. 3c). We then examined pht1;9 mutant PR growth response to low Pi availability. After 7 d of exposure to low Pi (50 μM Pi) conditions, the PRs of wild-type and pht1;9 mutant seedlings were c. 30% and 50% shorter, respectively, than those of seedlings grown under Pi-sufficient conditions (Fig. 3c). Similar results were obtained when no supplemental Pi was added to the external medium (Fig. S2a). We next sought to investigate the effects of Pi limitation on remodeling of the root system architecture in the pht1;9-1 and pht1;9-2 mutants. As expected, after 7 d of exposure to low Pi (50 μM Pi) conditions, wild-type seedlings harbored a highly branched root system, with a more than two-fold increase in emerged LR density when compared with seedlings grown under control Pi conditions (Fig. 3d). This stimulatory effect was greatly amplified in the pht1;9 mutants, which developed about twice as many visible LRs than the wild type when supplied with low amounts of Pi. Furthermore, subsequent elongation of the LRs was enhanced by more than four-fold in pht1;9 mutant seedlings and by only 2.5-fold in wild-type seedlings (Fig. 3d). Thus, the deleterious effects of Pi starvation on Arabidopsis seedling development are exacerbated by loss of Pht1;9 function.

Prompted by the observed effect of As(V) in yeast, we also checked whether Pht1;9 function affects the response to this toxic ion in vivo. Because As(V) competes with plant Pi influx and Pi transport activity is induced by Pi starvation, As(V) toxicity is highly correlated with the availability of Pi in the external medium (Meharg & Macnair, 1992; Shin et al., 2004; Catarecha et al., 2007). Hence, to minimize the deleterious effects of Pi starvation on pht1;9 mutant plants, we analyzed As(V) toxicity effects on seedlings grown under Pi-sufficient conditions (Fig. 3e). Typical symptoms of As(V) intoxication in Arabidopsis include, among others, inhibition of above-ground and root tissue growth (Shin et al., 2004; Catarecha et al., 2007), as clearly observed after 7 d of exposure to the metalloid (Figs 3e, S2b). Shoot biomass and PR elongation in wild-type seedlings were decreased by 60% and 40%, respectively, whereas both pht1;9 mutants exhibited only c. 35% and 20% reduction of the same growth parameters. These findings indicate that mutations in the Pht1;9 gene cause enhanced As(V) tolerance in Arabidopsis.

The striking phenotypical similarity displayed by the pht1;9-1 and pht1;9-2 mutant alleles strongly suggests that the observed defects result from loss of Pht1;9 function. To exclude the possibility of the effect of a mutation in another gene, we transformed both mutants with a DNA fragment spanning the Pht1;9 gene, including the 729 bp immediately upstream of the start codon. The corresponding transgenic complementation lines exhibited complete restoration of shoot growth and PR elongation wild-type sensitivity to low Pi supply and As(V) challenge (Fig. S2a,b). In addition, genomic complementation of the pht1;9 mutants fully suppressed the defects observed in LR formation and elongation induced by Pi starvation (Fig. S2c). We thus confirmed that disruption of the Pht1;9 gene is responsible for the identified Pi deficiency mutant phenotypes.

Overexpression of Pht1;9 results in enhanced tolerance to Pi starvation and hypersensitivity to arsenate in Arabidopsis

To obtain further clues on the functional analysis of the Pht1;9 gene, we generated transgenic Arabidopsis lines expressing the Pht1;9 transcript under the control of the 35S promoter in the wild-type background. Two independent lines containing one single integration event and noticeably overexpressing the Pht1;9 transcript, even in Pi-starved roots (Fig. S3), Pht1;9OX1 and Pht1;9OX2, were selected for further phenotypical characterization, together with the overexpression p35S:Pht1;9-YFP line described in Fig 1(d).

No phenotypical differences were observed between wild-type and Pht1;9-overexpressing plants grown on soil or under control Pi conditions. However, Pht1;9 overexpression conferred enhanced and reduced tolerance to low Pi and high As(V) supply, respectively, as assessed after quantification of the seedling shoot biomass (Fig. 4a). Moreover, the reduction in PR growth observed in wild-type seedlings under low Pi conditions was greatly attenuated in Pht1;9-overexpressing lines, even under severe Pi medium depletion (Fig. 4b). By contrast, and as expected for As(V)-hypersensitive lines, the reduction in PR elongation induced by As(V) toxicity was more pronounced in Pht1;9-overexpressing lines than in wild-type plants (Fig. 4b). Finally, although Pi starvation induced a marked increase in LR density and elongation in wild-type plants, Pht1;9-overexpressing plants showed only a slight increase in PR branching under the same conditions (Fig. 4c). Together with the results described above, this demonstrates that deletion and overexpression of Pht1;9 confer exact opposite phenotypes during Pi deficiency in Arabidopsis. In addition, these findings confirm that the Pht1;9-YFP fusion used to determine plasma membrane localization encodes a functional Pht1;9 transporter.

Figure 4.

Phenotype of transgenic Arabidopsis lines overexpressing Pht1;9. (a) Effect of inorganic phosphate (Pi) starvation and As(V) toxicity on shoot biomass (FW, fresh weight) of wild-type (Col-0) and three Pht1;9-overexpressing (Pht1;9OX1, Pht1;9OX2 and p35S:Pht1;9-YFP) seedlings (means ± SD, = 8). (b) Effect of Pi starvation (top panel) and As(V) toxicity (bottom panel) on primary root (PR) elongation of wild-type (Col-0) and three Pht1;9-overexpressing (Pht1;9OX1, Pht1;9OX2 and p35S:Pht1;9-YFP) seedlings (means ± SD, = 16). (c) Effect of Pi starvation on lateral root (LR) development of wild-type (Col-0) and three Pht1;9-overexpressing (Pht1;9OX1, Pht1;9OX2 and p35S:Pht1;9-YFP) seedlings (means ± SD, = 8). Both LR density (left panel) and total LR length (right panel) are shown (means ± SD, = 8). Statistically significant differences between Pht1;9-overexpressing lines and the wild type under each condition: *, < 0.05; **, < 0.01; ***, < 0.001; Student’s t-test. YFP, yellow fluorescent protein.

The Arabidopsis Pht1;9 and Pht1;8 transporters play similar roles during Pi deficiency and arsenate exposure

Phylogenetic analysis of the Arabidopsis Pht1 family revealed two main clusters, one of which includes all Pht1 members, except the highly similar Pht1;9 and Pht1;8 (At1g20860) which appear to be more distantly related to the others (Shin et al., 2004). Indeed, the coding sequences of these two genes share 78% nucleotide identity and encode proteins that are 79% identical. Interestingly, Pht1;8 induction has also been shown in Pi-starved roots (Mudge et al., 2002; Misson et al., 2005; Morcuende et al., 2007; Thibaud et al., 2010). To uncover the potential biological significance behind this clustering, we generated Arabidopsis RNAi lines by introducing a Pht1;9 hairpin construct under the control of the 35S promoter in wild-type plants. The selected fragment shares an overall identity of 81% with the Pht1;8 transcript, whereas only low homologies could be detected with small portions of the Pht1;6 and Pht1;7 transcripts. Accordingly, nine transgenic lines, named Pht1;9silA–I, displaying intermediate levels of Pht1;9 expression between wild-type and pht1;9 mutant plants and a concomitant level of Pht1;8 silencing in Pi-starved roots, were isolated (Figs 5, S4) and selected for phenotypical analysis. In particular, we identified two lines, Pht1;9silD and Pht1;9silG, exhibiting barely detectable expression levels of both genes. Relative to the wild type, no significant differences were observed in the steady-state transcript levels of the other Pht1 genes in either of the pht1;9 mutants or any of the Pht1;9 RNAi lines (Fig. 5), demonstrating the target specificity of our construct and suggesting the absence of compensatory mechanisms, at least at the transcriptional level, between the Pht1;9 and Pht1;8 transporters and other Pht1 members.

Figure 5.

Pht1 gene expression in Arabidopsis pht1;9 mutants and Pht1;9 RNAi transgenic lines. Reverse transcription-polymerase chain reaction (RT-PCR) analysis of Pht1;9 and Pht1;8 transcript levels in roots of 14-d-old seedlings of the wild type (Col-0), the pht1;9 mutants (pht1;9-1 and pht1;9-2) and nine Pht1;9 RNAi lines (Pht1;9silA-I) grown on low inorganic phosphate (Pi) medium. Expression of the UBIQUITIN10 (UBQ10) gene is shown as a loading control.

The response of these RNAi lines to low Pi availability and As(V) challenge was evaluated on the basis of the shoot biomass, PR elongation and LR development parameters described above (Table 1). Consistent with wild-type Pht1;8 and Pht1;9 expression levels, the Pht1;9silF line behaved as wild-type plants, whereas the Pht1;9silA and Pht1;9silC transgenic lines, exhibiting an approximate two-fold reduction in the expression of both genes, displayed intermediate phenotypes between the wild type and pht1;9 mutants. However, more pronounced defects than for the pht1;9 mutants were observed for the remaining six lines (Pht1;9silB, D, E, G, H and I), which showed < 40% of wild-type Pht1;8 and Pht1;9 transcript levels. Remarkably, the two lines exhibiting nearly total depletion of both genes, Pht1;9silD and Pht1;9silG, were the most sensitive and resistant to low Pi supply and As(V) toxicity, respectively (Table 1). Taken together, these findings open the way to the elucidation of the in vivo role of the Pht1;8 transporter, which appears to act analogously to Pht1;9 in the Arabidopsis Pi-deprived root.

Table 1.   Phenotype of Arabidopsis Pht1;9 RNAi transgenic lines
 50 μM Pi1.25 mM Pi + 250 μM As(V)
SB (% control)aPRE (% control)bLRD (LR number cm−1)aTLRL (cm)aSB (% control)aPRE (% control)b
  1. Effect of inorganic phosphate (Pi) starvation and As(V) toxicity on shoot biomass (SB), primary root elongation (PRE) and lateral root development (LRD, lateral root density; TLRL, total lateral root length) of seedlings of the wild type (Col-0), the pht1;9 mutants (pht1;9-1 and pht1;9-2) and nine Pht1;9 RNAi lines (Pht1;9silA–I) grown on either low Pi (50 μM Pi) or control (1.25 mM Pi) medium supplemented with 250 μM As(V). Different letters indicate statistically significant differences between means (< 0.05; Student’s t-test). aMeans ± SD, = 8; bMeans ± SD, = 16.

Col-071.82 ± 2.98 (a)68.64 ± 4.89 (a)2.84 ± 0.58 (a)5.37 ± 1.42 (a)62.23 ± 4.47 (a)73.01 ± 5.31 (a)
pht1;9-156.31 ± 3.45 (c)52.93 ± 5.11 (c)4.85 ± 1.27 (b)8.85 ± 1.12 (b)73.49 ± 5.88 (c)81.20 ± 3.99 (c)
pht1;9-255.25 ± 3.70 (c)53.80 ± 4.14 (c)5.11 ± 0.75 (b)8.41 ± 0.72 (b)71.80 ± 4.34(c)83.39 ± 3.84 (c)
Pht1;9silA62.84 ± 3.51 (b)58.01 ± 4.67 (b)3.40 ± 1.19 (ab)7.32 ± 2.62 (ab)67.66 ± 3.87 (bc)76.74 ± 4.58 (b)
Pht1;9silB49.23 ± 3.32 (d)47.93 ± 4.88 (d)5.33 ± 1.80 (b)8.52 ± 1.49 (b)77.90 ± 4.36 (c)86.34 ± 3.82 (d)
Pht1;9silC64.94 ± 2.68 (b)57.64 ± 4.66 (b)3.63 ± 0.91 (ab)7.44 ± 2.46 (ab)68.44 ± 5.79 (ac)76.81 ± 4.21 (b)
Pht1;9silD44.31 ± 3.14 (e)45.73 ± 4.76 (de)5.83 ± 0.93 (b)10.55 ± 1.26 (c)83.20 ± 4.76 (d)88.17 ± 4.30 (de)
Pht1;9silE50.19 ± 3.11 (d)48.45 ± 4.94 (d)5.09 ± 0.65 (b)7.83 ± 1.91 (b)76.67 ± 4.39 (c)85.37 ± 3.89 (c)
Pht1;9silF69.53 ± 4.14 (a)65.85 ± 4.71 (a)3.15 ± 0.55 (a)5.30 ± 0.88 (a)59.08 ± 4.24 (a)75.42 ± 5.07 (a)
Pht1;9silG43.89 ± 1.74 (e)46.03 ± 4.12 (de)5.69 ± 0.87 (b)11.50 ± 1.38 (c)84.85 ± 2.84 (d)87.29 ± 5.24 (d)
Pht1;9silH49.29 ± 4.11 (d)49.23 ± 3.22 (c)4.72 ± 1.01 (b)8.17 ± 2.62 (b)78.83 ± 5.76 (c)86.51 ± 4.75 (d)
Pht1;9silI50.07 ± 2.30 (d)47.89 ± 5.23 (d)5.34 ± 1.21 (b)8.23 ± 2.72 (b)75.76 ± 4.13 (c)85.38 ± 3.08 (c)

The Arabidopsis Pht1;9 and Pht1;8 transporters participate in Pi acquisition and arsenate uptake

The data gathered here point to a role of the Pht1;9 transporter in root Pi acquisition when Arabidopsis seedlings experience Pi starvation. To evaluate the contribution of Pht1;9 to total plant Pi uptake, we quantified the free Pi content of the shoots and roots of seedlings exposed to either a high or low Pi regime (Fig. 6a). No significant differences were observed between the wild type, the two pht1;9 mutants and the three Pht1;9-overexpressing lines in either Pi contents when seedlings were grown at control Pi levels, suggesting that Pht1;9 does not play a role in Pi acquisition under these conditions. Moreover, the shoot-to-root Pi content ratio remained nearly constant among all lines, indicating that Pi allocation from the root to the aerial parts of the plant is not affected by Pht1;9 function. Under low Pi supply, the Pi content of both tissues was greatly reduced in wild-type seedlings, with an average shoot and root Pi content of c. 35% of that of seedlings grown under control Pi conditions (Fig. 6a). In addition, the pht1;9 mutants and overexpression lines accumulated c. 25% less and 20% more Pi, respectively, than the wild type in both shoot and root tissues, yet the shoot-to-root ratios were equivalent between the genotypes. These data indicate that, although the Pi distribution process among the shoot and the root is not targeted by Pht1;9 function, this transporter contributes significantly to root Pi acquisition during Pi depletion, subsequently influencing the whole-plant Pi status.

Figure 6.

Inorganic phosphate (Pi) and arsenic content of Arabidopsis wild-type, pht1;9 mutant and Pht1;9 transgenic plants. (a) Pi concentration in the shoot (closed bars) and roots (open bars) of seedlings of the wild type (Col-0), the pht1;9 mutants (pht1;9-1 and pht1;9-2), two Pht1;9 RNAi lines (Pht1;9silD and Pht1;9silG) and three Pht1;9-overexpressing lines (Pht1;9OX1, Pht1;9OX2 and p35S:Pht1;9-YFP) grown under control Pi (top panel) or low Pi (bottom panel) medium (means ± SD, = 6). Shoot-to-root ratios for each genotype are indicated above the corresponding bars. (b) Total arsenic accumulation in the shoot of seedlings of the wild type (Col-0), the pht1;9 mutants (pht1;9-1 and pht1;9-2), two Pht1;9 RNAi lines (Pht1;9silD and Pht1;9silG) and three Pht1;9-overexpressing lines (Pht1;9OX1, Pht1;9OX2 and p35S:Pht1;9-YFP) grown under control Pi medium supplemented with 250 μM As(V). Bars represent means ± SD, = 4. Statistically significant differences from the wild-type: *, < 0.05; **, < 0.01; ***, < 0.001; Student’s t-test. DW, dry weight; FW, fresh weight; YFP, yellow fluorescent protein.

To establish whether the differences in As(V) tolerance correlate with a change in plant arsenic accumulation, we measured the amount of arsenic present in the shoots of the seedlings after As(V) exposure. As shown in Fig. 6(b), and consistent with their As(V)-tolerant phenotype, pht1;9 mutant seedlings accumulated significantly less arsenic than the wild type. By contrast, the shoots of Pht1;9-overexpressing lines accumulated up to 1.5 times the arsenic detected for the wild type under the same conditions. These results strongly suggest that the Pht1;9 transporter mediates a non-negligible fraction of As(V) uptake in Arabidopsis.

Finally, we examined whether the Pht1;8 transporter also participates in Pi and As(V) uptake in Arabidopsis by analyzing the free Pi and arsenic contents of the Pht1;9silD and Pht1;9silG silencing lines (see Table 1). Consistent with the enhanced response to Pi deprivation exhibited by these two lines when compared with the pht1;9 mutants, the shoot and root Pi contents of Pht1;9silD- and Pht1;9silG Pi-starved seedlings were significantly lower than those of pht1;9-1 and pht1;9-2, whereas they remained unaffected under high Pi conditions when compared with the wild type (Fig. 6a). Similarly, in comparison with wild-type seedlings, both RNAi lines exhibited an even more reduced arsenic content than the pht1;9 mutants on As(V) challenge (Fig. 6b). Thus, our results suggest that the Pht1;8 transporter acts as Pht1;9 to sustain Pi acquisition under Pi starvation conditions, also being able to participate in As(V) uptake.


Low P availability represents a major environmental stress for plants, which limits plant growth and hence has a negative impact on crop production in many agricultural ecosystems (reviewed in Bieleski, 1973; Raghothama, 1999). To sustain continuous development, plants respond to Pi limitation by a genome-wide transcriptional reprogramming that triggers an array of metabolic and developmental processes ultimately aimed at maintaining P homeostasis. A common strategy involves the activation of high-affinity transport systems in order to enhance Pi acquisition at the soil–root interface (reviewed in Schachtman et al., 1998). Here, we provide evidence that two members of the Pht1 family, Pht1;9 and Pht1;8, mediate Pi acquisition when Arabidopsis experiences Pi starvation.

The first indication that the Arabidopsis Pht1;9 gene encodes a functional Pi transporter came from the heterologous expression study performed in S. cerevisiae. The yeast Δpho84 mutant lacks a functional copy of a high-affinity Pi transporter gene and thus displays deficient growth under limiting Pi conditions (Bun-Ya et al., 1991). Complementation of this mutant’s growth defect at micromolar Pi concentrations, supported by Pi content analysis and a Pi uptake assay, demonstrates that the plant Pht1;9 protein is indeed capable of mediating high-affinity Pi transport. The apparent Km value for Pi uptake of 23.6 μM determined here for Pht1;9 is within the range of affinities (Km values of 3–31 μM) previously reported for other Pht1 transporters from higher plants (Mitsukawa et al., 1997; Daram et al., 1998; Rae et al., 2003; Liu et al., 2008). The higher affinity reported by Mitsukawa et al. (1997) for the Arabidopsis Pht1;1 transporter (Km of 3.1 μM) was assessed in plant cell cultures, and estimates from heterologous systems may not accurately reflect actual in planta parameters, as the requirement for additional endogenous factors cannot be ruled out. However, the extracellular alkalinization assay performed in the same yeast mutant background indicates that Pht1;9 catalyzes the influx of protons when importing Pi into the cell, pointing to H+-coupled Pi symport activity. Plants must acquire Pi from their environment against a step concentration gradient, as the Pi concentration in the soil seldom exceeds 10 μM, whereas intracellular Pi levels reach the millimolar range (reviewed in Bieleski, 1973; Schachtman et al., 1998). Moreover, because plants acquire P exclusively in its anionic form, transport systems must overcome the negative membrane potential of the plant cell. Root Pi uptake is therefore a dynamic process that requires the activity of secondary active transporters using the proton motive force to transport Pi. Accordingly, the pH dependence of Pi transport activity has been shown for numerous plant Pht1 transporters (Leggewie et al., 1997; Mitsukawa et al., 1997; Daram et al., 1998; Liu et al., 2008).

The second line of evidence for Pht1;9 Pi transport activity stemmed from our As(V) challenge assays. Indeed, Pht1;9 expression in yeast enhances the susceptibility to the As(V) inhibitory growth effects, although the reported growth advantage of the Δpho84 mutation to yeast cells against As(V) toxicity (Wykoff et al., 2007) was not detected under the low concentration employed here. Results obtained with the Pht1;9 loss-of-function and Pht1;9-overexpressing Arabidopsis lines exposed to toxic As(V) concentrations confirmed that Pht1;9 contributes significantly to plant As(V) uptake. Such a contribution has already been reported for the Arabidopsis Pht1;1, Pht1;4 and Pht1;5 transporters (Shin et al., 2004; Catarecha et al., 2007; Nagarajan et al., 2011). In particular, a semi-dominant mutant allele of Pht1;1, pht1;1-3, confers As(V) resistance to Arabidopsis seedlings, which is mediated by an enhanced ability to accumulate arsenic whilst reducing As(V) uptake rates (Catarecha et al., 2007). However, unlike the pht1;1-3 mutant, pht1;9 mutants accumulate less arsenic than the wild type, suggesting that this reduced accumulation is caused by less As(V) entering plant cells through the Pht1;9 transporter.

Importantly, the present work shows that Arabidopsis pht1;9 mutants exhibit exacerbated responses to Pi starvation, including amplified root system architectural modifications, with pronounced PR elongation slowdown and concomitant massive LR proliferation, both correlating with an enhanced reduction of shoot biomass when compared with the wild type. By contrast, Pht1;9 overexpression enhances the resistance to Pi starvation. Because root branching substantially determines the plant’s ability to effectively explore soil resources and thus absorb nutrients, it is not surprising that Pi availability alters so markedly the root system architecture. In parallel with PR growth attenuation, these adjustments are believed to allow optimized soil foraging, as Pi becomes more limiting with increasing soil depth (reviewed in Lopez-Bucio et al., 2003). Sanchez-Calderon et al. (2005) have reported that Pi deprivation promotes, in the Arabidopsis PR, an irreversible shift from an indeterminate to a determinate developmental program, involving the progressive loss of root apical meristem activity as a result of reduced cell elongation/proliferation and premature cell differentiation. This genetically controlled process is promoted by a sensing mechanism at the root tip and can ultimately lead to PR growth arrest (Ticconi et al., 2004; Sanchez-Calderon et al., 2005, 2006; Svistoonoff et al., 2007). We did not observe such a dramatic phenomenon under our Pi-limiting conditions, which were probably not sufficiently drastic, even when no Pi supplement was added to the culture medium (0 mM Pi). Indeed, no special measures were undertaken to Pi deplete the gelling agent (Jain et al., 2009). Alternatively, other elements that influence the developmental root response to Pi limitation, such as metals, may not have been present in standard amounts in our culture media. In particular, cellular iron toxicity has been shown to potentiate the inhibition of root apical meristem activity, mainly as a consequence of its improved bioavailability during Pi deficiency (Svistoonoff et al., 2007; Ward et al., 2008; Zheng et al., 2009). Nevertheless, exposing the wild type to no supplemental Pi phenocopied the pht1;9 mutant response observed under 50 μM Pi. Collectively, results from our phenotypical analyses strongly suggest that Pht1;9 functions primarily in root Pi acquisition during growth under low Pi supply, a role consistent with the root plasma membrane subcellular localization detected for the Pht1;9 transporter. However, unlike Pht1;1 and Pht1;4 (Shin et al., 2004), Pht1;9 does not appear to be involved in Pi uptake under Pi-sufficient conditions, as indicated by the indistinguishable phenotypes displayed by wild-type and pht1;9 mutant Pi-replete plants, and consistent with the very low Pht1;9 expression levels detected in roots under such conditions (Mudge et al., 2002; Misson et al., 2005; Morcuende et al., 2007). In addition, Pht1;9 is unlikely to participate in Pi allocation to the shoot, mainly because no substantial differences were observed between the shoot-to-root ratios of wild-type, mutant or transgenic plants grown under varying Pi supplies. Rather, Pi content analysis of these different genotypes indicates that Pht1;9 function affects the accumulation of free Pi to the same extent in shoot and root tissues, therefore influencing the overall plant Pi status.

We also investigated whether the close homolog Pht1;8 shares a role with Pht1;9. Indeed, double Pht1;8/Pht1;9 silencing lines display a stronger Pi starvation-related phenotype than the pht1;9 loss-of-function alleles. These results pave the way to the elucidation of the in vivo function of the Arabidopsis Pht1;8 transporter, which, like Pht1;9, appears to play a crucial role in Pi acquisition solely during Pi deficiency, as supported by both the unchanged Pi content of our Pht1;8/Pht1;9 silencing lines and the low Pht1;8 expression levels under sufficient Pi supply (Misson et al., 2005; Morcuende et al., 2007; Thibaud et al., 2010). Interestingly, no significant changes in Pht1;8 transcript levels were observed in the pht1;9 mutants, suggesting the absence of transcriptional compensation between the two genes. Nevertheless, despite a similar lack of compensating Pht1;1 and Pht1;4 transcript levels in the respective pht1;4 and pht1;1 mutants, both of these transporters work together to enable the plant to take advantage of a high Pi supply following a period of Pi starvation (Shin et al., 2004). More generally, Pi-starved root transcript levels of all Pht1 family members – except Pht1;6 whose expression is undetectable under such conditions – were not significantly altered in either the pht1;9 mutant or the Pht1;8/Pht1;9 silencing line backgrounds, as also observed in the pht1;1/pht1;4 double mutant described by Shin et al. (2004), and despite an altered plant Pi status compared with wild-type plants. This apparent discrepancy finds its explanation in a recent study by Thibaud et al. (2010), demonstrating that it is the external Pi status, rather than the internal plant Pi content, that modulates, together with the root growth response (Svistoonoff et al., 2007), the transcriptional response to low Pi availability.

The Pi acquisition Pht1;9 function during Pi starvation is consistent with its specific Pi-deprived root expression pattern. Indeed, we detected strong induction of Pht1;9 expression under prolonged Pi starvation in the Arabidopsis root, in agreement with previously observed sustained Pht1;9 induction on medium- and long-term Pi deprivation (Misson et al., 2005), which is shortly repressed after Pi resupply (Morcuende et al., 2007). Interestingly, similar expression regulation was reported for the Pht1;8 gene and, indeed, Pht1;8 and Pht1;9 are the most strongly induced Pht1 genes in Pi-starved roots (Misson et al., 2005; Morcuende et al., 2007). Recent computer-based predictions identified cis-regulatory elements conserved among plant promoters of Pi starvation-induced genes, several of which have been validated experimentally as sufficient to drive highly specific Pi-starved root expression (Misson et al., 2005; Tittarelli et al., 2007; Karthikeyan et al., 2009; Thibaud et al., 2010), such as the P1BS (PHR1 Binding Site) element (Rubio et al., 2001; Bustos et al., 2010). The MYB transcription factor PHR1 (Phosphate Starvation Response 1) and its close homolog PHL1 (PHR1-Like 1) play a central regulatory role in the global transcriptional responsiveness to Pi starvation in Arabidopsis (Rubio et al., 2001; Bustos et al., 2010). The presence of P1BS boxes in the promoter of genes induced systemically by Pi starvation emphasizes the role of PHR1 as a major element of the Pi long-distance signal transduction pathway (Thibaud et al., 2010). Such P1BS elements are present in the proximal promoters of Pht1;8 and Pht1;9, and, indeed, up-regulation of both genes is greatly attenuated in phr1 and phr1/phl1 double-mutant Pi-starved seedlings (Bustos et al., 2010). Promoter–reporter gene fusion experiments represent a useful tool to assess gene expression cell specificity, particularly in plant root tissues. However, Mudge et al. (2002) failed to observe Pht1;9 promoter activity using the β-glucuronidase reporter gene, even under Pi deprivation, indicating that Pht1;9 expression levels are below the detection threshold for this type of assay. We were therefore unable to determine which of the root cell layers specifically expresses Pht1;9. In addition, no clear information on cell specificity can be expected from publicly available microarray data, as these are gathered from Pi-replete tissues (; Zimmermann et al., 2004). We were nevertheless also able to detect Pht1;9 expression in developing siliques. Interestingly, the Pht1;9 gene has been identified as a marker associated with seed mineral P concentration after quantitative trait locus (QTL) analysis of two Arabidopsis recombinant inbred populations (Waters & Grusak, 2008), but we were unable to detect any germination defects for the pht1;9 mutants under various Pi supplies (data not shown). Finally, we detected strong Pht1;9 induction in senescent leaves. Although this induction was not detected by Mudge et al. (2002), van der Graaff et al. (2006) subsequently found clear Pht1;9 up-regulation at the particular leaf senescence stage tested here. This suggests a role for Pht1;9 in Pi redistribution from older to young leaves under limiting Pi conditions, as has been attributed to the Pht1;5 and Pht2;1 transporters (Versaw & Harrison, 2002; Nagarajan et al., 2011).

In addition to transcriptional regulation, additional Pi starvation signaling mechanisms may operate to tightly control the activity of the Pht1;9 and Pht1;8 transporters during Pi deficiency. Importantly, Pht1;9 and Pht1;8 transcript levels are concomitantly up-regulated in pho2 mutant and miRNA399-overexpressing plants, suggesting that both genes are downstream components of the PHO2 regulatory system (Aung et al., 2006; Rouached et al., 2010). In Arabidopsis, PHO2 encodes the ubiquitin-conjugating E2 enzyme UBC24, which is targeted by the miRNA399 microRNA, a component of the shoot-to-root Pi deficiency signaling pathway (Lin et al., 2008). The early miR399 induction in shoots by Pi starvation and its subsequent long-distance movement in phloem vessels signal the down-regulation of PHO2 expression in roots, which, in turn, activates the high-affinity Pi transport system and ultimately regulates whole-plant Pi homeostasis (Pant et al., 2008). Thus, pho2 mutant and miRNA399-overexpressing plants accumulate excessive Pi in their shoots as a result of enhanced root Pi uptake (Aung et al., 2006). By contrast, a mutation in the PHF1 (Phosphate Transporter Traffic Facilitator 1) gene leads to an 80% reduction in net Pi uptake relative to the wild type under low Pi conditions (González et al., 2005). The plant-specific SEC12-related protein PHF1 is an endoplasmic reticulum (ER)-associated factor, involved in the formation of COPII vesicles for the export of newly synthesized proteins to the Golgi apparatus, essential for efficient intracellular trafficking and therefore plasma membrane targeting of the Pht1;1 transporter (González et al., 2005). Recently, loss of PHF1 function has also been shown to impair appropriate plasma membrane targeting of two other Pht1 members, Pht1;2 and Pht1;4, suggesting that PHF1 plays a broader role in regulating Pht1 transporter transit through the ER (González et al., 2005; Bayle et al., 2011). Using elegant phosphorylation-mimicking mutagenesis of Pht1;1, Bayle et al. (2011) also demonstrated that phosphorylation of a serine residue at the C-terminal extremity of Pht1;1 is sufficient to prevent its exit from the ER when internal Pi is high. Interestingly, phosphorylation events in serine residues have been reported to occur at the C-terminal extremity of Pht1;9, as well as of Pht1;5 and Pht1;7 (Bayle et al., 2011). Further functional and molecular characterization of the Pht1;9 and Pht1;8 transporters holds much promise in uncovering the full extent of their role in the regulation of plant P homeostasis.


We thank N.-H. Chua for the pBA002 vectors and V. Nunes and M. Baião for technical assistance. This work was funded by Fundação para a Ciência e a Tecnologia (Grants PTDC/AGR-AAM/67858/2006 and PTDC/AGR-AAM/102967/2008, as well as PostDoctoral Fellowships SFRH/BPD/44640/2008 and SFRH/BPD/81221/2011 awarded to E.R. and T.R.C., respectively). P.D. is supported by Programa Ciência (QREN/MCTES).