Over-expression of PHO1 in Arabidopsis leaves reveals its role in mediating phosphate efflux

Authors

  • Aleksandra Stefanovic,

    1. Département de Biologie Moléculaire Végétale, Biophore, Université de Lausanne, CH-1015 Lausanne, Switzerland
    Search for more papers by this author
    • Present address: Laboratory of Sensory Neuroscience, Rockefeller University, 1230 York Avenue, Campus Box 314, New York, NY 10065, USA.

  • A. Bulak Arpat,

    1. Département de Biologie Moléculaire Végétale, Biophore, Université de Lausanne, CH-1015 Lausanne, Switzerland
    Search for more papers by this author
  • Richard Bligny,

    1. Laboratoire de Physiologie Cellulaire Végétale, Centre d’Energie Atomique, 17 rue des Martyrs, F-38054 Grenoble Cedex 9, France
    Search for more papers by this author
  • Elisabeth Gout,

    1. Laboratoire de Physiologie Cellulaire Végétale, Centre d’Energie Atomique, 17 rue des Martyrs, F-38054 Grenoble Cedex 9, France
    Search for more papers by this author
  • Charles Vidoudez,

    1. Département de Biologie Moléculaire Végétale, Biophore, Université de Lausanne, CH-1015 Lausanne, Switzerland
    Search for more papers by this author
  • Michaël Bensimon,

    1. Laboratoire de géologie de l’ingénieur et de l’environnement, Bâtiment GC, Ecole Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzerland
    Search for more papers by this author
  • Yves Poirier

    Corresponding author
    1. Département de Biologie Moléculaire Végétale, Biophore, Université de Lausanne, CH-1015 Lausanne, Switzerland
      (fax +41 21 692 4195; e-mail yves.poirier@unil.ch).
    Search for more papers by this author

(fax +41 21 692 4195; e-mail yves.poirier@unil.ch).

Summary

Inorganic phosphate (Pi) homeostasis in multi-cellular eukaryotes depends not only on Pi influx into cells, but also on Pi efflux. Examples in plants for which Pi efflux is crucial are transfer of Pi into the xylem of roots and release of Pi at the peri-arbuscular interface of mycorrhizal roots. Despite its importance, no protein has been identified that specifically mediates phosphate efflux either in animals or plants. The Arabidopsis thaliana PHO1 gene is expressed in roots, and was previously shown to be involved in long-distance transfer of Pi from the root to the shoot. Here we show that PHO1 over-expression in the shoot of A. thaliana led to a two- to threefold increase in shoot Pi content and a severe reduction in shoot growth. 31P-NMR in vivo showed a normal initial distribution of intracellular Pi between the cytoplasm and the vacuole in leaves over-expressing PHO1, followed by a large efflux of Pi into the infiltration medium, leading to a rapid reduction of the vacuolar Pi pool. Furthermore, the Pi concentration in leaf xylem exudates from intact plants was more than 100-fold higher in PHO1 over-expressing plants compared to wild-type. Together, these results show that PHO1 over-expression in leaves leads to a dramatic efflux of Pi out of cells and into the xylem vessel, revealing a crucial role for PHO1 in Pi efflux.

Introduction

Inorganic phosphate (Pi) is an essential nutrient for all living organisms, and its homeostasis in multi-cellular eukaryotes is dependent on the acquisition and distribution of Pi across various organs. Such movement of Pi depends on Pi influx into cells and controlled Pi efflux out of cells. Examples in which Pi efflux is crucial are release of Pi from the basolateral side of the kidney proximal tubule and small intestine epithelial cells for transfer to the blood stream, transfer of Pi into the xylem of plants, and release of Pi at the peri-arbuscular interface of mycorrhizal roots (Beck and Silve, 2001; Bonfante and Genre, 2010; Bucher, 2006; Marks et al., 2010). Despite its importance, no gene has been identified that specifically mediates phosphate efflux in eukaryotic cells.

Plants acquire Pi by active uptake into the epidermal and cortical cells of the root via proton–Pi symporters of the PHT1 family (Poirier and Bucher, 2002; Raghothama, 2000). Once in root cortical cells, Pi must eventually be loaded into the apoplastic space of the xylem, and transported to the shoot by the transpiration stream and root pressure (Clarkson, 1993). The root endodermis contains a hydrophobic Casparian band that forms a barrier to apoplastic flow of water and ions towards the root vascular cylinder, making movement of ions across the endodermal cells a key step in transfer of ions to the root xylem. A few genes have been identified that participate in transfer of ions into the root xylem. Some of the corresponding proteins have been shown to mediate ion efflux, such as BOR1 and SKOR1 for boron and potassium efflux, respectively (Gaymard et al., 1998; Takano et al., 2002). Influx carriers are also involved in the transfer of ions into xylem vessels, such as the low-affinity sulfate transporters SULTR2;1 and SULTR3;5, the low-affinity Pi transporter PHT1;2, and the bi-directional nitrate transporter NTR1:5 (Ai et al., 2009; Kataoka et al., 2004; Lin et al., 2008; Takahashi et al., 2000). Interestingly, all these genes are expressed primarily in root pericycle and xylem parenchyma cells. A multitude of mechanisms and transporters mediating influx and efflux are thus involved in directing nutrients from the root cortical cells into xylem vessels.

The Arabidopsis thaliana pho1 mutant is defective in the transfer of Pi from roots to shoots, resulting in strong Pi deficiency in above-ground tissues (Poirier et al., 1991). The PHO1 gene was identified by map-based cloning, and the PHO1 gene family has been characterized in Arabidopsis, rice and the moss Physcomitrella patens (Hamburger et al., 2002; Secco et al., 2010; Stefanovic et al., 2007; Wang et al., 2008, 2004). PHO1 is expressed primarily in the vascular tissue of the root. The corresponding protein contains six potential transmembrane spanning domains and a long N-terminal hydrophilic domain (Hamburger et al., 2002). PHO1 shows no homology to characterized solute transporters, including the family of plant PHT1 H+/Pi co-transporters, but contains two domains, named SPX and EXS, that have been identified in some Saccharomyces cerevisiae proteins that are involved in phosphate transport or sensing, and in protein sorting to endomembranes (Wang et al., 2004). Proteins homologous to PHO1 are also present in a large spectrum of eukaryotes, from Caenorhabditis elegans and Drosophila to mammals (Hamburger et al., 2002). The pho1 phenotype could result from a block in uptake of Pi into endodermal cells, either via active uptake of Pi from the apoplast or its transfer via plasmodesmata. Support for this hypothesis comes from specific expression of the promoter of the Pi deficiency-responsive gene IPS1 in the endodermal cells of pho1 roots, indicating that endodermal cells are the primary Pi-deficient cells in the root of the pho1 mutant (Martin et al., 2000). Alternatively, the pho1 mutant could be compromised in acquisition of Pi into the pericycle and/or xylem parenchyma cells of the root, or in the release of Pi into the apoplast of the root vascular cylinder. These latter hypotheses are supported by expression of the PHO1 promoter in the cells of the root vascular cylinder, including the pericycle (Hamburger et al., 2002). However, neither Pi influx or efflux activity has yet been demonstrated for PHO1 when expressed in heterologous systems (including Xenopus oocytes, yeast and liposomes), raising questions about its mode of action in long-distance Pi transport (Hamburger et al., 2002).

In a set of experiments aimed at studying the role of PHO1 in Pi transport, it was discovered that PHO1 over-expression in leaves was associated with high shoot Pi level and reduced shoot growth. Examination of the Pi transport dynamics in these plants revealed that PHO1 over-expression in leaves led to a dramatic export of Pi into the xylem vessel and out of leaves. Altogether, these results reveal a role for PHO1 as an important mediator of Pi efflux.

Results

Isolation of transgenic plants over-expressing PHO1

Initial attempts to over-express PHO1 in transgenic A. thaliana using a vector containing the PHO1 gene under the control of the constitutive CaMV 35S promoter failed to identify a single plant, among more than 50 independent transformants, that showed increased PHO1 expression (data not shown). In contrast, transformation of wild-type (WT) plants with a T-DNA vector containing the PHO1 gene under the control of its own promoter yielded four independent transgenic lines that showed strong PHO1 over-expression in whole seedlings (Figure 1a). All these lines had a distinct phenotype, with rosettes leaves being small, dark-green and slightly curled (Figure 1b). Two independent lines from this group, named L3 and SL5, were selected for further analyses. Northern blot analysis revealed a large increase in PHO1 expression in both roots and shoots of the L3 line compared with WT, for plants grown in media containing low or high Pi (Figure 1c). The increase in PHO1 expression was particularly strong for shoots, as PHO1 is typically weakly expressed in WT shoots compared to roots. Western blot analysis confirmed higher PHO1 protein levels in the roots of WT plants, which were further increased upon Pi-deficiency stress (Figure 1d). PHO1 remained below the detection level in the WT shoot, even in Pi-deficient plants. The SL5 and L3 transgenic lines had strikingly higher PHO1 levels than WT, especially in the shoots.

Figure 1.

 Transgenic plants over-expressing PHO1.
(a) Northern blot analysis of PHO1 expression in WT compared to four independent transgenic lines showing PHO1 over-expression. RNA was extracted from whole plants grown for 2 weeks on agar-solidified medium. The lower panel shows the RNA gel before transfer.
(b) Phenotype of plants over-expressing PHO1. WT and two PHO1 over-expressing lines L3 and SL5 were grown for 8 weeks under 8 h light/16 h dark conditions.
(c) Northern blot analyses of roots (R) and shoot (S) of WT and PHO1 over-expressing line L3. Plants were grown in agar-solidified medium supplemented with 1 mm Pi (+) or 10 μm Pi (−). The lower panel shows the RNA gel before transfer.
(d) Western blot analyses of protein extracts from roots and shoot of WT and the lines SL5 and L3. Plants were grown in agar-solidified medium supplemented with 1 mm Pi (+) or 10 μm Pi (−). For each genotype, the upper panel shows the location of PHO1 on the Western blot (arrowhead) and the lower panel shows a region of the Ponceau-stained filter (Pc) as a control for protein loading.
(e) Western blot analyses of WT and L3 leaves from plants grown in soil. The first two lanes are protein extracts from entire leaves, the second two lanes are protein extracts from leaf mesophyll protoplasts, and the last two lanes are protein extracts of leaf fragments enriched in vascular elements obtained after protoplasting. The upper panel shows the location of PHO1 on the Western blot (arrowhead), and the lower panel shows a region of the Ponceau-stained filter (Pc) as a control for protein loading.

In order to assess whether over-expression of PHO1 in the shoot was associated more with the vascular tissue or the mesophyll cells, a Western blot was performed to compare the PHO1 protein level in leaf mesophyll protoplasts and the leaf fragments enriched in vascular elements that remained after protoplasting (Funk et al., 2002). While Western blot from the mesophyll protoplasts derived from L3 leaves did not reveal PHO1 expression, in agreement with the expression data of Yang et al. (2008), cells associated with the remaining leaf fragments enriched in vascular elements clearly expressed PHO1 (Figure 1e). These results indicate that PHO1 over-expression is primarily associated with the vascular system in leaves.

PHO1 over-expression in leaves leads to reduced shoot growth and increased Pi content

PHO1 over-expressing plants had notably reduced rosette mass compared to the WT, with differences in growth becoming visible after the four-leaf stage (Figure 2a). The shoot Pi contents in L3 and SL5 were two- to threefold higher than in WT plants throughout the monitored growth period, with levels in the L3 line being consistently higher than for the SL5 line (Figure 2b). Elemental analysis of leaves revealed 2.9- and 2.2-fold more phosphorus, respectively, in the L3 and SL5 lines compared to WT (Figure 3a). A significant difference was also observed for iron, with L3 and SL5 lines having 2- and 2.3-fold more iron than WT, respectively. No differences were observed for any other element. Although the phenotype of soil-grown PHO1 over-expressing lines is somewhat reminiscent of the Pi-deficient pho1 null mutant, expression of the Pi starvation-inducible markers Pht1;4, SPX1 and mir399d was not increased in shoots of the L3 line (Figure 3b).

Figure 2.

 Growth and phosphate content of PHO1 over-expressing lines.
(a) Plants were grown in soil in a 12 h day/12 h night cycle, and the shoot fresh weight was measured over 32 days.
(b) Pi content in the shoots quantified in parallel with the growth measurements shown in (a). Errors bars represent standard errors ( 4).

Figure 3.

 Elemental analysis, gene expression and grafting of PHO1 over-expressing lines.
(a) Elemental analyses (Ca, K, P, S, Mg, Na, Fe and Mn) of leaves were performed by inductively coupled plasma atomic emission spectroscopy. Leaves were from 4-week-old WT (white bars), L3 (gray bars) and SL5 (black bars) plants grown in soil. Elements were quantified as milligram per gram dry weight of rosette. Error bars represent standard errors ( 4).
(b) Level of expression of PHO1, mir399d, PHT1;4 and SPX1 in leaves of soil-grown WT plants, the pho1 null mutant and the L3 PHO1-over-expressing line. Transcripts are expressed relative to UBQ10. Error bars represent standard deviation (= 4).
(c) Grafts were made between shoots and roots of wild-type (S WT/R WT), shoots and roots of the L3 line (S L3/R L3), shoot of wild-type and root of L3 (S WT/R L3) and shoot of L3 and root of wild-type (S L3/R WT). Following grafting, plants were grown in soil for 8 weeks.
(d) Inorganic phosphate content in the shoots of the various grafts described in (c) after 8 weeks of growth in soil. Error bars represent standard error (= 10).

To determine whether higher expression of PHO1 in the root of L3 is associated with a higher rate of transfer of Pi from the root to the shoot, we measured the uptake of 33Pi into the root and its translocation to the shoot in WT and L3 plants grown on two external Pi concentrations (Table 1). The uptake rate into the root in L3 was statistically lower than in WT in plants grown on low Pi (10 μm), but no difference was observed for the higher Pi concentration (100 μm). The translocation rate of Pi from the root to the shoot was only slightly increased in the L3 line compared to WT for plants grown at 100 μm Pi, and no statistical difference between WT and L3 was measured for plants grown at 10 μm.

Table 1.   Root phosphate uptake and transfer to the shoot
Pi concentration in the mediumRoot uptake (nmol Pi g FW−1 h−1)Transfer to shoot (%)
  1. *Statistically different from WT control (t test, = 3,  0.02).

100 μm
 WT946 ± 12712.8 ± 0.63
 L3959 ± 10516.8 ± 0.91*
10 μm
 WT513 ± 6015.4 ± 1.7
 L3288 ± 20*13.4 ± 1.7

Given the over-expression of PHO1 in both roots and shoots, we performed micro-grafting experiments to determine whether the higher shoot Pi and reduced rosette growth phenotypes found in the L3 lines are due to high expression of PHO1 in the shoots rather than the roots. When a WT shoot was grafted onto a L3 root, no change in the rosette appearance was observed during development compared to self-grafted WT plants (Figure 3c). By contrast, when a L3 shoot was grafted onto a WT root, the phenotype of mature rosettes was identical to that of self-grafted L3 plants. The shoot Pi content was higher in both L3 self-grafted control and L3 shoot/WT root grafts compared to WT self-grafted control and WT shoot/L3 root grafts (Figure 3d). These results showed that over-expression of PHO1 in shoots leads to reduced rosette growth and increased Pi content in the shoot.

Leaves over-expressing PHO1 rapidly excrete Pi into the extracellular medium

Pi pools in the leaves of WT and PHO1 over-expressing plants grown in soil were analyzed by in vivo31P-NMR to test whether PHO1 over-expression could lead to improper accumulation of Pi in the cytoplasm, which could explain the strong negative effect on shoot growth. Initial measurement (time = 0) revealed a similar distribution of Pi in the vacuole and the cytoplasm in both WT and PHO1 over-expressing plants, with the vacuolar Pi content representing approximately 98 and 91% of the total cellular Pi pool in WT and the L3 line, respectively (Figure 4a). Following these initial measurements, cellular Pi distribution in leaves infiltrated with a solution containing 50 μm Pi was recorded after 12 h. In WT leaves, the cytoplasmic Pi peak remained relatively constant, but the vacuolar Pi signal increased by 50%, indicating net uptake of Pi from the solution. In contrast, leaf cells from the L3 line showed a strong decline of the vacuolar Pi peak, reaching a level equivalent to the cytoplasmic Pi peak. Furthermore, an additional small Pi signal appeared between the vacuolar and cytoplasmic Pi peaks in the L3 line, which was identified as an increase in the external Pi concentration. Thus, while leaves of WT plants acquire Pi from medium with low Pi content (50 μm) and store it in the vacuole, leaves from the L3 transgenic line rapidly release Pi to the medium, with a consequent loss of the vacuolar Pi pool. Overall, the in vivo31P-NMR experiments demonstrated that the deleterious effects of PHO1 over-expression in L3 plants were not due to accumulation of Pi in the cytoplasm. Instead, they revealed rapid efflux of Pi from the leaves of L3, in contrast to the net Pi uptake by WT leaves.

Figure 4.

 Export of inorganic phosphate from leaves in PHO1 over-expressing plants.
(a) In vivo31P-NMR spectra from leaves of WT or the L3 transgenic line at time zero (t0) and after 12 h in a perfusion solution containing 50 μm Pi (+Pi 12 h). Peaks from right to left are assigned to vacuolar Pi (vac-Pi), external Pi (ext-Pi), cytoplasmic Pi (cyt-Pi), phosphatidyl choline (P-cho) and glucose-6-phosphate (glc-6-P).
(b) Leaves from WT, L3 and SL5 transgenic lines were cut into small pieces and immersed in a solution containing 50 μm Pi for 16 h. The phosphate concentration in the bathing solution was quantified every hour over 16 h, and expressed as micromoles Pi released per gram fresh weight of leaves.

In an independent set of experiments, the flux of Pi out of leaves and into the perfusion medium was measured over 16 h. Wild-type leaves showed minimal loss of Pi into the medium, but a large loss of Pi from leaves of the L3 and SL5 transgenic lines was observed, corresponding to 27- and 34-fold increases of Pi in the external medium for SL5 and L3, respectively, compared to WT (Figure 4b). In order to check the specificity of the ion efflux, nitrate was quantified in the external medium together with phosphate. No change in external nitrate content was observed between the 1 and 12 h time points for WT, and only a twofold increase was observed for L3, suggesting specificity of the efflux mechanism for phosphate over nitrate (Table 2).

Table 2.   Quantification of nitrate efflux from leaf pieces (μmol g−1 FW of leaves)
 Time = 1 hTime = 12 h
WT0.60 ± 0.110.68 ± 0.14
L30.64 ± 0.131.36 ± 0.23

The rapid export of Pi out of infiltrated leaves from L3 plants provided an explanation for the lower Pi level measured by in vivo31P-NMR compared to a direct Pi assay on leaves (Figure 4a). Preparation of plant material for in vivo31P-NMR involves immersing small leaf pieces in a low Pi solution for at least 90 min before the first NMR spectrum is recorded. A reconstruction experiment revealed that export of Pi into the perfusion medium in L3 plants is reflected by a corresponding decrease in intracellular Pi, such that the intracellular Pi content in the L3 line decreased by approximately twofold in the first 90–120 min of perfusion.

The export of Pi out of leaves and into the infiltration medium was further examined in WT and PHO1 over-expressing L3 plants that were first grown under a different Pi supply. Plants were grown for 28 days in a root hydroculture system containing either 100 m or 2000 μm Pi, or grown in medium containing 2000 μm Pi for 23 days followed by 5 days in Pi-free medium. Shoots were then harvested, leaves were immersed in a Pi-free solution, and the amount of Pi released into the medium was measured after 12 h (Table 3). For all growth conditions tested, shoot fresh weight was significantly higher for WT compared to L3, and shoot Pi content was two- to fourfold higher in L3 compared to WT. For plants grown in medium containing 100 μm Pi, export of Pi from leaves immersed in solution for 12 h was below the detection limit for WT but reached 29% of the leaf intracellular Pi content for L3. For plants grown continuously at 2000 μm Pi, export of Pi out of WT leaves was detectable but low, at 2.4% of the initial leaf Pi content, but reached 60% for L3 leaves. A small but significant reduction in Pi export was observed for WT and L3 leaves from plants that were starved for Pi for the last 5 days of growth, with Pi export decreasing to 0.9 and 52% of the initial leaf Pi content in WT and L3, respectively (Table 3).

Table 3.   Export of Pi from PHO1 over-expressing leaves of plants grown in various media
GenotypeTreatmentaShoot weight (mg)Shoot Pib (μmol g−1)Media Pic (μmol g−1)Exportd (%)
  1. aPlants were grown for 28 days in hydroculture medium with either 100 or 2000 μm Pi, or for 23 days in medium with 2000 μm Pi followed by 5 days in medium with no Pi (2000→0).

  2. bShoot Pi content was measured at day 28.

  3. cPi exported into the medium was measured after immersing leaves in Pi-free export medium for 12 h.

  4. dPi export was measured as the fraction of shoot Pi that is exported into the medium after 12 h.

  5. ND, not detectable (below 0.01 μmol g−1).

Col100 μm49 ± 1911.9 ± 2.5ND 
L3100 μm22 ± 520.1 ± 3.25.9 ± 1.628.9 ± 4.2
Col2000 μm632 ± 15914.9 ± 0.30.35 ± 0.182.4 ± 1.2
L32000 μm224 ± 1662.6 ± 2.637.5 ± 2.460.1 ± 5.1
Col2000→0 μm414 ± 1618.0 ± 1.60.08 ± 0.070.9 ± 0.8
L32000→0 μm127 ± 4530.7 ± 5.215.9 ± 3.251.6 ± 2.9

Large increase of Pi in xylem exudates from PHO1 over-expressing leaves

The concentration of Pi, as well as of sodium, magnesium, calcium and potassium, was quantified in three compartments – the apoplastic fluid, phloem and xylem exudates – from leaves of WT and the SL5 and L3 transgenic lines (Figure 5a). Compared to WT, the quantity of the various ions in the apoplast and phloem exudate was increased by three- and eightfold in the SL5 line, and by 10- to 15-fold in the L3 line. In both phloem exudate and apoplast, the concentration of Pi did not change significantly more than most other ions measured. However, a striking difference was seen in the xylem exudate of L3 and SL5 plants, with Pi concentrations 100- to 140-fold higher than in WT plants. The concentrations of the other ions increased by less than tenfold. Similar results were obtained for two independent sample sets.

Figure 5.

 Element content in apoplast, guttation fluid and phloem exudates.
(a) Apoplast, guttation fluid and phloem exudates from leaves of soil-grown plants were compared for WT and PHO1 over-expressing lines L3 and SL5 with respect to the content of the elements Na, Mg, P, Ca and K. Values on the y axis represent the values in L3 (or SL5) divided by the values found in WT. Similar results were obtained in two independent experiments.
(b) Distribution of Pi in plants grown in hydroculture containing 33Pi for 12 h as visualized by phosphoimaging.

The distribution of Pi in leaves that were grown for 12 h in medium containing 33Pi followed by 48 h in non-radioactive Pi was analyzed by phosphoimaging. Figure 5(b) shows that, while Pi was evenly distributed across the leaves in WT plants, foci of strong accumulation of Pi were observed at the edge of leaves in the L3 plants, corresponding to the hydathodes, the location of exudation of the xylem vessel content.

Discussion

The present study work demonstrates that PHO1 mediates specific Pi export into the apoplast. Although PHO1 over-expression occurred in both roots and shoots, grafting experiments showed that both the high Pi level in leaves and the reduced shoot growth were only dependent on PHO1 over-expression in shoots. Furthermore, PHO1 expression in leaves remained associated with the vascular system, indicating that, despite being over-expressed, the specificity of the PHO1 promoter for the vascular system is maintained in these plants (Hamburger et al., 2002). In agreement with the pho1 mutant being primarily deficient in Pi (Poirier et al., 1991), PHO1 over-expression in leaves resulted in a two- to threefold increase in the amount of phosphorus, with most other elements remaining stable, with the exception of a small but significant increase in the amount of iron. In nutrient solutions, Pi readily precipitates with iron, resulting in reductions in the availability of both nutrients (Dalton et al., 1983). The slight increase of iron in PHO1 over-expressing plants could potentially be linked to precipitation of iron in planta due to the high Pi content in leaves. However, the most remarkable effect of PHO1 over-expression in leaves, a tissue whose vascular system lacks a Casparian band and endodermal cells, was the dramatic increase in export of Pi into the apoplast. When leaves were immersed in a nutrient solution, the continuum formed between the apoplast and the external medium led to a large export of Pi out of leaves and into the nutrient solution (Figure 4b and Table 3). This resulted in rapid depletion of Pi from the leaves, reaching 60% of the leaf Pi content after 12 h of infiltration (Table 3). The extensive export of Pi from leaves was also reflected in loss of the Pi vacuolar pool (Figure 4a). For rosette plants grown in soil, although Pi concentrations were increased in both apoplast and phloem exudates, the largest increase (>100-fold) was observed in the guttation fluid, indicative of high efflux of Pi into the xylem and its accumulation at the hydathodes, the site of interface between the xylem vessel and the atmosphere.

It is interesting to note that PHO1 was more over-expressed in the L3 line compared to SL5 (Figure 1a,d), and that this correlated with higher levels of Pi and phosphorus in shoots (Figures 2b and 3a), of Pi release from leaves into the infiltration medium (Figure 4b), and of Pi content in guttation fluid (Figure 5a) in the L3 line compared to SL5. These results indicate that a higher level of PHO1 expression in leaves leads to higher Pi export activity. However, this correlation is not found when leaf Pi export is compared between L3 plants grown in Pi-sufficient and Pi-deficient media. While Western analysis showed that PHO1 expression in L3 leaves is further increased in plants grown under Pi-deficient conditions compared to Pi-sufficient conditions (Figure 1d), the level of leaf Pi export in L3 plants grown under Pi-deficient conditions decreased slightly relative to L3 plants grown under Pi-sufficient conditions (Table 3). This decrease in net Pi export is probably due to activation of the Pi salvage pathway triggered by Pi deficiency, leading to activation of the H+/Pi co-transporter of the PHT family and re-acquisition of the exported Pi.

In contrast to the large effect of PHO1 over-expression in leaves, over-expression of PHO1 in roots was associated with only a minor or no change in the rate of transfer of phosphate from root to shoot (Table 1). This difference between the effect of PHO1 over-expression in shoots and roots could be due to mechanisms of post-translational control of PHO1 activity that may be present in roots, a tissue that normally expresses PHO1 to high level, but is absent in shoots. Alternatively, other factors could contribute to limiting the efflux of Pi into the xylem of the root despite PHO1 over-expression, such as the rate of Pi transfer across the endodermal cells. The reasons behind the small negative effect of PHO1 over-expression on the Pi uptake capacity of L3 roots grown in medium with low external Pi are unknown (Table 1).

In vivo NMR spectra revealed normal vacuolar and cytoplasmic Pi levels in leaves of soil-grown plants over-expressing PHO1, indicating that the high level of Pi efflux was largely compensated for by Pi import. This was also supported by the observation that PHO1 over-expressing plants did not exhibit the responses typical of Pi deficiency at the gene expression level. This implies the presence, at some stage of growth, of a substantial futile cycle whereby Pi efflux and influx apparently balance each other in order to maintain adequate intracellular Pi. Futile cycling of ions such as Pi is energetically costly for the plant, and could have negative consequences. Futile cycling of ammonium ions, with an associated increase in respiration, has been observed for plants grown in the presence of high external ammonium, and the energy cost of such a futile cycle was hypothesized to be the basis of ammonium toxicity (Britto et al., 2001).

In addition to its main effect on Pi efflux and growth reduction, PHO1 over-expression also led to an increase concentration of sodium, magnesium, calcium and potassium in the apoplast, xylem exudate and phloem sap, indicating disruption in the transport and homeostasis of numerous ions. The pho1 mutant was shown to be deficient only in Pi homeostasis (Poirier et al., 1991). It is thus unlikely that PHO1 is directly responsible for the efflux of both monovalent and divalent cations in addition to Pi. Changes in cation metabolism are most likely secondary effects of the large Pi efflux. It is possible that the energy cost of futile cycling of Pi may have an indirect effect on the transport of other ions that are also dependent on membrane potential and proton ATPase. The high ion concentration observed in the apoplast could thus have an effect on thermodynamic parameters important for the management of nutrients and water flow. Growth defects could also be caused by the increase in apoplastic ion concentration, as this would decrease the water potential differential across the plasma membrane, and in turn negatively affect the force driving cell-wall expansion and growth (Cosgrove, 2005).

While over-expression of plant membrane proteins is often attempted as an alternative to expression in heterologous systems to assess functionality, there are relatively few reports describing large effects of over-expressing genes encoding transporters on ion homeostasis. Extensive regulation of ion transporters through post-transcriptional and/or post-translational mechanisms is likely to be one important reason behind this observation. Interestingly, in the present work, transgenic plants over-expressing PHO1 were not recovered when a constitutive promoter was used, but only when expression was driven by the native PHO1 promoter, which allows PHO1 expression mainly to the vascular elements. Given the large Pi efflux and the strong deleterious phenotype observed in plants over-expressing PHO1 from its native promoter, it is likely that constitutive expression of PHO1 is lethal to the developing embryo or prevents the successful establishment of viable seedlings.

The role of PHO1 in mediating Pi export, and its predominant expression in root stellar cells, are both in agreement with the phenotype of the pho1 mutant, which was shown to be primarily defective in transfer of Pi to the root xylem (Poirier et al., 1991). PHO1 thus acts in transfer of Pi to the root xylem via export of Pi out of the cells of the stele and into the xylem vessel apoplast. As such, the role of PHO1 is analogous to that of SKOR and BOR1, which mediate the efflux of potassium and boron, respectively, to the root xylem vessel (Gaymard et al., 1998; Takano et al., 2002). However, the action of PHO1 is distinct from that of the rice low-affinity H+/Pi symporter PHT1;2 and the Arabidopsis low-affinity H+/sulfate symporters SULTR2;1 and SULTR3;5, which probably contribute to root-to-shoot ion transfer by promoting re-acquisition of ions into the stelar cells after their passage through the endoderm (Ai et al., 2009; Kataoka et al., 2004; Takahashi et al., 2000). Although the PHO1 gene is weakly expressed in leaves, no function has yet been assigned for such expression, and its physiological relevance to leaf Pi homeostasis is unknown. However, PHO1 expression has recently been shown to influence the response of shoots to Pi-deficient conditions, indicating that PHO1 mediates not only Pi export, but also plays an important role in control of the Pi-deficiency response (Rouached et al., 2011).

While proteins homologous to the plant PHO1 are found in most multi-cellular eukaryotes, including mammals, fungi and insects, the primary biological role of these non-plant homologs is unknown. The human and mouse PHO1 homologs, named XPR1 and RCM1, respectively, have been shown to act as cell-surface receptors for retroviruses (Battini et al., 1999; Tailor et al., 1999; Yang et al., 1999). Although several receptors for retroviruses mediate the transport of various solutes when expressed in Xenopus oocytes, including amino acids and phosphate (Kavanaugh et al., 1994; Kim et al., 1991; Wang et al., 1991), expression of either Arabidopsis PHO1 or mammalian RCM1/XPR1 in Xenopus oocytes failed to elicit Pi transport activity (D. Kabat, Department of Biochemistry and Molecular Biology, Oregon Health and Science University, personal communication; D. Hamburger, M. Jabnoune and Y. Poirier, Department of Plant Molecular Biology, University of Lausanne, unpublished data). Nevertheless, Pi export is an important process that occurs in numerous multi-cellular organisms. While acquisition of Pi by the mycorrhizal fungi Glomus versiforme is known to be mediated by the H+/Pi symporter GvPT (Harrison and van Buuren, 1995), the protein mediating export of Pi out of the fungi towards the plant apoplastic space, an essential step in the transfer of Pi from the fungus to the plant, remains unknown (Bonfante and Genre, 2010). Interestingly, a homolog of the PHO1 protein is present in the ectomycorrhizal fungus Tuber melanosporum (XP_002842097). Given the role of Arabidopsis PHO1 in Pi export, an attractive hypothesis that requires further testing is that non-plant homologs of PHO1 could also be involved in Pi export.

Experimental procedures

Plant material, growth conditions and Pi assay

Wild-type A. thaliana (L.) and transgenic plants over-expressing PHO1 were all in the Columbia ecotype background. Plants grown on agar-solidified medium were placed in growth rooms at 22–23°C under a continuous light intensity of 100 μmol m−2 sec−1. Medium containing half-strength Murashige-Schoog (MS), 1% sucrose and 0.7% agar was used to grow plants on solid medium. For experiments where Pi concentrations were varied, the medium contained 1% sucrose, 0.7% agar, 2.5 mm KNO3, 100 μm Ca(NO3)2, 1 mm MgSO4, 25 μm Fe-EDTA, 14 μm MnCl2, 0.5 μm CuSO4, 1 μm ZnSO4, 70 μm H3BO3, 0.2 μm NaMoO4 and 10 μm NaCl (final pH 5.7). A variable amount of Pi was provided as KH2PO4. Plants in soil were grown either under continuous light at 22°C, or under day/night cycles of 12 h/12 h or 8 h/16 h, with day and night temperatures of 22 and 18°C, respectively. Plants were also grown in a hydroculture system in pots containing washed clay pellets (Seramis, http://www.seramis.de) and drenched with nutrient solution. The basal nutrient solution for all hydroculture experiments included 0.5 mm KNO3, 1 mm MgSO4, 1 mm KH2PO4, 0.25 mm Ca(NO3)2, 100 μm Fe-EDTA, 30 μm H3BO3, 10 μm MnCl2, 1 μm CuCl2, 1 μm ZnCl2, 0.1 μm (NH4)6Mo7O24 and 50 μm KCl. The basal medium contained 1 mm KH2PO4 as the Pi source. In media with lower Pi concentrations, equimolar amounts of KCl were substituted for the KH2PO4.

DNA construct

The construct used to over-express PHO1 has been described previously (Stefanovic et al., 2007). Briefly, a region of 2.1 kb comprising the native PHO1 promoter and the complete PHO1 genomic region from the start to the stop codon was cloned into the pCAMBIA1380 binary vector. The clone was confirmed by sequencing, introduced into Agrobacterium tumefaciens pGV3101, and used for transformation of A. thaliana by the floral-dip method (Clough and Bent, 1998).

Pi quantity and transport assay

Determination of Pi levels in tissues was performed by releasing the cellular content of cells into water by repeated freeze–thaw cycles or by incubation for 30 min at 70°C, and quantifying Pi by the molybdate assay (Ames, 1966). Measurement of the rate of Pi acquisition into the roots and transfer from root to shoot was measured in plants grown in agar-solidified medium essentially as previously described (Poirier et al., 1991). For pulse–chase analysis in intact plants, roots were immersed for 12 h in a modified Hoagland’s solution containing 10 μm Pi and 0.5 μCi ml−1 33Pi. Roots were then washed extensively with a 10 μm Pi solution, and plants were re-potted in a hydroculture system containing Hoagland’s medium for 48 h. Distribution of radioactivity in leaves was analyzed by phosphoimaging (Storm 820; GE Healthcare, http://www.gehealthcare.com).

Total RNA extraction, Northern analyses, and quantitative PCR

RNA was prepared using a LiCl protocol (Sambrook and Russell, 2001). Samples of 20 μg RNA were separated on 1% agarose formaldehyde gels and blotted onto Hybond N+ membranes (Amersham, http://www5.amershambiosciences.com/). Hydridization at high stringency with a PHO1-specific fragment containing 32P-labeled nucleotides was performed according to standard procedures (Sambrook and Russell, 2001). For RT-PCR and real-time quantitative RT-PCR, RNA samples were treated with RNase-free DNase I (Qiagen, http://www.qiagen.com/) to remove any residual genomic DNA. Total RNA was quantified using an ND-1000 spectrophometer (Nanodrop Technologies, http://www.nanodrop.com), and first-strand complementary DNA (cDNA) was synthesized from 1 μg total RNA using M-MLV reverse transcriptase (Promega, http://www.promega.com/) according to the manufacturer’s instructions. Real-time quantitative RT-PCR analysis was performed using a Stratagene Mpx3000 instrument (http://www.stratagene.com/) according to the standard curve method (Rutledge and Cote, 2003). All reactions were prepared in triplicate in optical tubes with caps (Stratagene) using SyberGreen mix (Abgene, http://www.abgene.com) and the reference dye ROX (Abgene). Amplifications were performed with the following thermal cycle: an initial incubation at 95°C for 15 min, followed by 40 cycles of 95°C for 15 sec, 60°C for 30 sec and 72°C for 30 sec. Calculated expression values were normalized against expression levels of ubiquitin. Specific primers for SPX1 (At5g20150), PHT1;4 (At2g38940) and micro399d were as previously described (Bari et al., 2006; Morcuende et al., 2007). PHO1 was amplified using the oligonucleotides 5′-ACACCATTCCAGGCATCCTCCTC-3′ and 5′-ACGGTGAGCAAACAATCTTCCGC-3′.

Protein extraction and Western analyses

Total protein extracts were prepared from homogenized frozen leaf and root material by extraction in 80 mm Tris pH 8.0, 1 mm EDTA, 5 mm DTT, 1% v/v Triton X-100, 10% v/v glycerol and 1 mm PMSF. Protease inhibitor cocktail (5 μl) for plant cell extracts (Sigma, http://www.sigmaaldrich.com/) was added to each sample. Samples were centrifuged at 13 000 g for 5 min, and the protein concentration of the soluble fraction was determined using Bradford reagent (Bio-Rad, http://www.bio-rad.com/) with BSA as the standard. For Western blotting analysis, 50 μg of total protein were subjected to SDS–PAGE according to standard procedures (Sambrook and Russell, 2001). Proteins were electroblotted onto 0.2 μm nitrocellulose membranes (Bio-Rad), and blocked for 2 h at room temperature in PBS-Tween containing 5% w/v non-fat dried milk. Incubation with antibodies was performed in PBS-Tween containing 1% w/v non-fat dried milk using PHO1 antiserum (1:100) or horseradish peroxidase-conjugated anti-protein A (1:500; Sigma). Blots were developed using the ECL Western blotting detection kit (Amersham).

NMR analyses

Ten gram samples of leaves from 3-week-old plants grown under continuous light were cut into 4–5 mm2 pieces, vacuum-infiltrated in perfusion solution, and placed in a 25 mm glass tube under constant perfusion. Details of this assembly and its operation have been described previously (Roby et al., 1987). The perfusion solution contained 5 mm glucose, 10 mm KNO3, 0.5 mm Ca(NO3)2, 1 mm KCl, 0.5 mm MgSO4 and 50 μm K2PO4 at pH 6.0. In vivo31P-NMR spectra were recorded on a spectrometer (AMX 400, Bruker, http://www.bruker.com) as described previously (Aubert et al., 1998). Assignments of cytoplasmic and vacuolar pools of Pi and phosphate esters to specific peaks were made in accordance with those given previously (Bligny et al., 1989; Roby et al., 1987).

Phosphate and nitrate efflux experiments

Pi efflux from leaves was quantified for plants grown in soil or in hydroculture media containing various amount of Pi (see above) for 4 weeks. Rosette leaves were cu,t and the major vein was removed using a razor blade. The leaf pieces were placed in a Falcon tube (BD Biosciences, http://www.bdbiosciences.com) containing 5 ml of ice-cold medium with 5 mm glucose, 10 mm KNO3, 0.5 mm Ca(NO3)2, 1 mm KCl, 0.5 mm MgSO4 and either 50 μm K2PO4 or no Pi, and infiltrated under a mild vacuum over 30 min on ice. After infiltration, the cold medium was replaced with medium at room temperature and samples were placed on the shaker (40 rpm) at 22°C. Aliquots were taken for Pi quantification at various times. For nitrate export quantification, leaf pieces were placed in a Falcon tube containing 5 ml ice-cold medium containing 5 mm glucose, 10 mm K2PO4, 0.05 mm Ca(NO3)2, 1 mm KCl, 0.5 mm MgSO4 and 0.4 mm CaCl2 at pH 6.0. The remainder of the experiment was performed as described for the phosphate export experiment. Phosphate was measured by the molybdate assay (Ames, 1966). The phosphate content was calculated as micromolar Pi per mg fresh weight. Nitrate was quantified using sulfamic acid (Carvalho et al., 1998). Samples were mixed with an equal volume of 0.05 m sulfamic acid, and absorbance was measured at 220 nm.

Guttation fluid, apoplast and phloem content isolation

Guttation fluid was collected from 6 to 7-week-old plants growing under 8 h light/16 h night conditions. Guttation was induced by lowering the temperature to 15°C during the night, while maintaining high humidity by covering the water-filled tray. Apoplast was isolated from leaves of 8- and 10-week-old plants grown in soil under 8 h light/16 h dark conditions to increase the leaf area (Muhling and Lauchli, 2000). Fully expanded leaves were gently infiltrated with ice-cold nanopure water (17 MΩ) (Thermo Scientific Barnstead, http://www.thermo.com) using a syringe and blotted dry on the absorbing paper. The main vein was removed with a razor blade, and the leaf tissue was centrifuged immediately at 75 g and 4°C for 5 min. The ion concentrations measured in the apoplast were multiplied by a dilution factor estimated for WT and PHO1 over-expressors. The dilution was calculated for WT, L3 and SL5 leaves by infiltrating a known concentration of sodium tungstate into the leaf and measuring the tungstate concentration in the collected apoplast by capillary ion electrophoresis. The difference between infiltrated and apoplastic tungstate concentrations represented the volume occupied by the apoplast. Phloem exudates were collected by the EDTA method (King and Zeevaart, 1974), scaled down for use with A. thaliana. Seven mature leaves were collected by excision at the petiole bases. All leaves collected from one plant were placed in a 500 μl microcentrifuge tube with their petioles immersed in 400 μl of 10 mm EDTA (pH 8.5). The tubes were placed in airtight transparent plastic containers in which the atmosphere was water-saturated to prevent uptake of the EDTA solution by the leaves. Exudation lasted 8 h. Due to the large difference in leaf mass of WT and PHO1 over-expressing plants, ion concentrations measured in the phloem exudates were corrected for the mass of leaves used for exudations (per gram fresh weight). Apoplast, guttation fluid and phloem exudates were collected from several plants of each line (WT, SL5 and L3) and pooled.

Elemental analyses

Leaves of 4-week-old plants grown in soil in constant light were dried at 80°C and digested in nitric acid at 100°C. Samples were diluted and analyzed by inductively coupled plasma (ICP) atomic emission spectroscopy for the P, K, Ca, S, Na, Mg, Mn and Fe content as described previously (Lahner et al., 2003).

Samples from guttation fluid, apoplast or phloem sap were diluted and analyzed by sector-field inductively coupled plasma mass spectrometry (SF-ICP-MS, Element2, Thermo Finnigan, http://www.thermo.com). Nitric acid (65%, Suprapur® grade; Merck, http://www.merck.com) was used for dilution of samples and for preparation of standards. Ultrapure water was produced using a Direct-Q™ ultrapure water system (Millipore, http://www.millipore.com). Calibration standards were prepared through successive dilutions in Teflon bottles of 1 g L−1 ICP-MS stock solutions (Bernd Kraft Duisburg, http://www.bkraft.de). The isotopes 23Na, 24Mg, 31P, 44Ca, and 39K were used to quantify these elements in the diluted samples. All isotopes were monitored at medium-resolution mode (resolution power, = 4000), except for 39K isotope which was monitored at high-resolution mode (= 10 000) in order to eliminate potential interferences due the matrix and oxide formation.

Micrografting experiments

Grafting experiments were performed as previously described (Turnbull et al., 2002). Seeds were kept at 22°C under continuous light for the first 2 days, and were transferred to 27°C and a 8 h light/16 h night photoperiod for the next 3 days. Grafting was performed on 5–7-day-old plants. After grafting, plants were returned to 27°C for an additional 3 days. Graft unions were examined, and successfully grafted plants were transferred to soil, where they were grown for another 6 weeks. Prior to phenotype analyses, graft unions were re-examined to ensure that no adventitious roots grew from the shoot scion.

Isolation of mesophyll protoplasts

Protoplasts were prepared from 4-week-old Arabidopsis leaves as described previously (Abel and Theologis, 1994) with slight modifications. The leaves were rinsed four times with sterile distilled water, blotted dry on filter paper, and cut with a razor blade (approximate final size 9 mm2). The leaf tissue was then incubated in 0.5 m mannitol as a pre-plasmolysis step. After 1 h at room temperature, the mannitol solution was replaced with 30 ml protoplasting solution (0.4 m mannitol, 1% w/v cellulase R-10, 0.5% w/v macerozyme (Serva, http://www.serva.de), 0.5% w/v BSA, 30 mm CaCl2, 5 mm 2-mercaptoethanol and 5 mm MES, pH 5.7). The cut leaf tissue was vacuum-infiltrated with protoplasting buffer and was incubated at room temperature with gentle agitation (100 rpm) for 3–4 h. The digested material was filtered through a 50 μm screen to separate the protoplasts from the leaf tissue. Filtered protoplasts were centrifuged at 100 g for 5 min at 4°C, washed once for 1 min with 0.5 m mannitol, 5 mm MES (pH 6.0) and 1 mm CaCl2, and counted on a hemocytometer. The protoplasts were mainly derived from mesophyll cells due to the higher accessibility of these cells to the cell-wall digestion enzymes. Vascular tissue elements are tightly packed and are thus relatively untouched during the digestion procedure (Funk et al., 2002). They were collected as debris left after the digestion and subjected to protein extraction.

Acknowledgements

This research was funded by an Fonds National Suisse grant (3100A0-122493) to Y.P., as well as by the University of Lausanne.

Ancillary