Restricted spatial expression of a high-affinity phosphate transporter in potato roots

Plants respond to phosphorus (P) depletion by increasing the rate of P uptake by roots (Schachtman et al.,1998), and an upregulation of the synthesis of a carrier system is believed to contribute to the plant's observed enhancement of P absorption. The kinetics of P uptake by whole plants is complex, requiring a dual mechanism; a high-affinity system operating in the low micromolar range and a low-affinity system operating at higher concentrations has been characterised in Catharanthus roseus suspension cultured protoplasts(Furihata et al., 1992). Genes encoding high-affinity phosphate (Pi) transporters have now been identified in several species, including Arabidopsis(Muchhal et al., 1996; Okumura et al., 1998; Smith et al., 1997), rose(Kai et al., 1997), Medicago truncatula (Lui et al., 1998b), tomato(Daram et al., 1998) and potato (Leggewie et al., 1997; Rausch et al., 2001). The corresponding genes have been grouped into the Pht1 family of proton-Pi cotransporters (Bucher et al.,2001), which are energised by the plasma membrane proton ATPase(Schachtman et al., 1998). With the exception of rose, at least two different members of the Pht1 family have been found in each of these plant species. Most can be grouped together into three structurally related subfamilies, and members of the same group may share common functional properties.

Studies of the regulation and tissue and cellular distribution of members of the Pht1 family are commencing and this has indicated that many of the genes are expressed in the roots in response to P deprivation. In some cases transcripts or proteins have been localised to the root epidermis and even root hairs, strongly suggesting a role in Pi capture and uptake(Chiou et al., 2001; Daram et al., 1998; Karthikeyan et al., 2002; Mudge et al., 2002; Liu et al., 1998a). The detection of transporters that are more widely distributed throughout plant tissues and that are less tightly regulated by P depletion provides evidence that some members of the Pht1 family may be involved in the internal distribution of Pi (Daram et al.,1998; Leggewie et al.,1997; Liu et al.,1998a; Mudge et al.,2002; Rosewarne et al.,1999). Such transporters may be responsible for loading or unloading from the xylem or phloem, deposition into the seed and storage tissue or remobilisation of Pi from senescing leaves. Two Arabidopsismutants, pho1 and pho2, exhibit defects in xylem or phloem loading that may be due to mutations in the Pi transport system. pho1lacks the ability to load Pi into the xylem, causing shoot Pi concentrations to be reduced (Poirier et al.,1991). However, the gene involved, PHO1, may be regulatory, rather than directly involved in Pi efflux from stellar cells(Hamburger et al., 2002). pho2 accumulates high levels of Pi in leaves even during P deprivation. A phloem loading defect has been proposed(Dong et al., 1998), but shoot cell regulation of internal Pi concentrations may be responsible.

In addition to the high-affinity system, Pht2;1, a member of the Pht2 low-affinity transporter family (K_m for Pi, 0.4 mM), has been identified in Arabidopsis(Daram et al., 1999). Although this protein is structurally related to members of a mammalian Na/Pi transporter family, its activity in a yeast expression system is dependent on the proton gradient across the yeast plasma membrane but independent of the sodium gradient. The mammalian counterpart is present in many cell types and is believed to be responsible for the absorption of Pi for normal cellular function. Pht2;1 is constitutively expressed in green tissue, suggesting that it possesses a comparable `house keeping' function, but it has recently been shown, by green fluorescent protein (GFP) promoter fusion analysis, to be associated with the chloroplast and to have a very specific transport function(Versaw and Harrison, 2002). The subcellular distribution of members of the Pht1 family has so far exclusively indicated localisation at the plasma membrane(Chiou et al., 2001; Muchhal and Raghothama,1999).

In potato, three high-affinity transporters, StPT1, StPT2 and StPT3, have been cloned and sequenced (Leggewie et al., 1997; Rausch et al.,2001). Northern blot analysis has shown that StPT1 mRNA is present not only in root tissue but also to a lesser extent in the source leaves,flowers and tubers. Although amounts of StPT1 mRNA in the root increase in response to P depletion, a low level of constitutive expression can be detected throughout the plant. StPT2 mRNA, however, is only detected by northern blots in roots after P depletion and StPT3 is only present in roots colonised by mycorrhiza (Rausch et al.,2001).

In the present work, we have used an antibody to a unique sequence from StPT2 that is not present in StPT1 or StPT3, or any other known plant protein,to study the tissue, cellular and subcellular distribution of the transporter in both hydroponically and quartz grown plants. Our findings suggest that StPT2 is expressed most strongly near the root tip in the zone of elongation behind the meristematic region and in the root hairs. Immunolocalisation using confocal microscopy shows that in those cells that express StPT2, the transporter is present only in the apical plasma membrane.

Potato plants (Solanum tuberosum cv Desiree) were propagated in sterile magenta boxes in 0.4% agar gel containing 0.441% Murashige and Skoog medium (ICN, Basingstoke, UK) and 2% sucrose, adjusted pH to 5.7 with KOH. The plants were grown in a tissue culture room maintained at 22°C under constant light. Once root growth was established the plants were transferred either to hydroponic culture or to quartz in 0.5× Hoagland's medium(Röhm and Werner, 1987)in a growth cabinet set at 23°C with a 16 hour day length. The plants were harvested after 3 weeks. For P-depletion experiments, plants were supplied with Hoagland's medium from which 0.5 mM KH₂PO₄ had been omitted 5 days before harvest.

Two peptides, VEIESEEAKIEQISRDETC, corresponding to a nonconserved region from the central loop of StPT2, and CLIEEEEGIND, from a conserved region of the vacuolar pyrophosphatase (VPPase), were synthesised by Sigma Genosys(Pampisford, UK). The peptides (5 mg) were conjugated to 5 mg of maleimide activated keyhole limpet haemocyanin or bovine serum albumen (Sigma Genosys)in 1 ml of deionised water overnight at room temperature and purified on a sephadex G25 column. Rabbits were immunised with 1 mg of conjugate with 0.5 ml Freunds complete adjuvant, followed by a boost one month later. After a further 2 weeks, repeated at 3 weekly intervals, bleeds were collected and IgG was purified from the sera using an Econo-Pac serum IgG purification kit(Bio-Rad, Hemel Hempstead, UK) according to the manufacturer's instructions. IgG was prepared in the same manner from pre-immune sera. The VPPase antibody detected a protein band on western blots with an apparent molecular weight of 66 kDa in tonoplast vesicle preparations from wheat and mung bean (R.G.-W.,unpublished), and no bands were detected in plasma membrane preparations(Theodoulou et al., 2003).

Rat monoclonal antibodies to α-tubulin were purchased from Serotec(Oxford, UK). Rat monoclonal antibodies to plant cell wall methyl esterified pectin (JM 7) were a kind gift from Maxwell S. Bush, John Innes Centre,Norwich, UK (Bush and McCann,1999) and plasma membrane ATPase antibodies were a gift from Mark Boutry, Université Catholique de Louvain(Morsomme et al., 1998). Fluorescent labelled secondary antibodies (Alexa Fluor™ 568 goat anti-rabbit and Alexa Fluor™ 488 goat anti-rat) were purchased from Molecular Probes (Cambridge, UK).

Whole extracts were prepared from fresh tissue. Segments of root were excised, weighed and transferred to a chilled glass homogeniser. Gel electrophoresis loading buffer (10% glycerol, 2% sodium dodecyl sulphate(SDS), 5% mercaptoethanol, 0.00125% bromophenol blue and 62.5 mM Tris HCl, pH 6.8) was added in the ratio of 10 μl per μg of tissue. The tissue was homogenised and the homogenate was centrifuged for 2 minutes in a microfuge at 50,000 g to remove the cell debris. Aliquots of the supernatant (50 μl per lane) were loaded immediately onto SDS polyacrylamide gel electrophoresis (PAGE).

Plasma membrane vesicles were purified from whole root tissue and leaves as described by Larsson et al. (Larsson et al., 1994) with minor modifications. Briefly, tissue was harvested on ice and homogenised for in 4×15 minutes in a Waring blender in homogenisation buffer (500 mM sorbitol, 50 mM HEPES, 5 mM EGTA, 1 mM dithiothreitol (DTT), 1 mM PMSF, pH 7.5) at a ratio of 2:1 (volume/fresh weight). The mixture was filtered and centrifuged for 12 minutes at 1000 g. The supernatant was collected and centrifuged for a further 45 minutes at 50,000 g and the pellet was resuspended in 1/25 the original volume of resuspension buffer (330 mM sorbitol, 5 mM KCl, 1 mM DTT, 5 mM K-phosphate, pH 7.8) and homogenised in a glass homogeniser. 4.5 ml of the mixture was loaded onto 13.5 g gradients comprising 6.6% dextran T500, 6.6% polyethylene glycol 3350, 330 mM sorbitol, 5 mM KCl, 1 mM DTT, 5 mM K-phosphate, pH 7.8, mixed and incubated on ice for 5 minutes. The gradients were centrifuged for 10 minutes at 1000 g and the upper phases were removed to fresh lower phases, prepared by centrifuging 13.5 g gradients and removing the upper phases, which were then used to wash the lower phases. The phases were centrifuged again and the phase partitioning was repeated before the upper phases, enriched in plasma membrane vesicles, were pooled and centrifuged for 60 minutes at 100,000 g. The resulting pellet was resuspended in 330 mM sorbitol, 5 mM Tris-maleic acid, pH 7.8 and the protein concentration assayed by a modification of the method of Appleroth and Angsten(Appleroth and Angsten, 1987)with bovine serum albumen as a standard. For western blotting 20 μg of protein was loaded onto each lane.

The purity of the plasma membrane fractions was determined using marker enzymes (Gordon-Weeks et al.,1996; Hodges and Leonard,1974) and the activities were compared to the activities of the enzymes in microsomal and tonoplast (Rea et al., 1992) fractions. In a typical experiment the specific activity of the plasma membrane marker (vanadate-sensitive ATPase) in the plasma membrane fraction was 14.20 μmol Pi released per mg protein per hour, whereas the specific activity in the microsomal fraction was 5.8 and in the tonoplast, 1.73. The VPPase activity was 1.34 μmol pyrophosphate hydrolysed per mg protein per hour, compared with 10.60 (microsomes) and 27.28(tonoplast). Azide sensitive ATPase activity (mitochondrial contamination) was 0.88 μmol Pi released per mg protein per hour (1.60 in the microsomes) and NADH-dependent cytochrome c reductase (endoplasmic reticulum contamination)was 2.61 μmol cytochrome c reduced per mg protein per hour (11.84 in the microsomes).

Protein samples were separated on 12.5% SDS PAGE gels and transferred overnight to nitrocellulose hybond c membranes (Amersham Biosciences, Little Chalfont, UK) in transfer buffer containing 48 mM Tris, 39 mM glycine, 0.03%SDS and 10% methanol. Membranes were blocked for 1 hour in Tris buffered saline (TBS; 20 mM Tris, 2 mM NaCl, pH 7.5) containing 5% dried milk. Blots were probed with purified IgG antibodies for 4 hours at room temperature at a 1:1000 dilution in TBS containing 1% dried milk. After washing for 1 hour in TBS with 1% milk, the blots were exposed to secondary antibody (donkey anti-rabbit purchased from Sigma Genosys) conjugated to horse radish peroxidase for 45 minutes at a 1:10,000 dilution. The blots were washed three times in TBS and the immunoreactive bands were visualised using enhanced chemiluminescence kits (Amersham Biosciences) according to the manufacturer's instructions.

For total RNA isolation, root tissue was ground to a powder under liquid nitrogen. 0.2 g were removed to a 2 ml tube containing 1.5 ml Trizol (Sigma,Poole, UK) and the contents vortexed for 30 seconds. After a 5 minute incubation at room temperature the suspension was centrifuged at 12,000 g for 10 minutes and the supernatant removed to a clean 2 ml tube. An equal volume of chloroform was added and the phases vortexed for 15 seconds and incubated for 5 minutes at room temperature. The phases were separated by centrifugation at 4°C for 15 minutes at 12,000 g and the aqueous layer removed to a clean 2 ml tube. An equal volume of chloroform-isoamylalcohol was added and the aqueous layer separated as before. A one-tenth volume of 3.0 M sodium acetate and one-sixth volume of isopropanol were added, and after 10 minutes incubation at room temperature the RNA was pelleted by centrifugation at 12,000 g at 4°C for 10 minutes. The pellet was washed twice with 70% alcohol and air dried.

PCR reactions used cDNA prepared from total RNA isolated from root tissue using 18S RNA as a loading control(Burleigh, 2001). For amplification of StPT2, 5 pmol of Oligo (dT)_12-18 was used as the antisense primer for first strand cDNA and synthesis of 18S RNA used 5 pmol of the antisense primer 5′-CAC TTC ACC GGA CCA TTC AAT CG-3′. Reactions used 0.5 μg RNA and Superscript II™ RT (GibcoBRL, Paisley,UK), following the manufacturer's recommended protocol. PCR was done initially using 18S cDNA with one-twentieth volume of the first strand cDNA as template,2.5U Taq DNA polymerase (Promega, Southampton, UK), PCR buffer supplemented with 1.5 mM MgCl₂ (MBI-Fermentas, St Leon-Rot,Germany), and nucleotide (Amersham Biosciences) concentrations as recommended by the supplier in a 50 μl reaction volume. Sense (5′-GAG GGA CTA TGG CCG TTT AGG-3′) and antisense primers were used at a final concentration of 200 pmol. Reactions were performed on an Omnigene Thermal Cycler with a heated lid (Hybaid, Ashford, UK) programmed to give a temperature profile of 2 minutes at 94°C followed by 40 cycles of 30 seconds at 94°C, 30 seconds at 55°C, 1 minute 30 seconds at 72°C and a final 5 minute extension at 72°C. PCR products were analysed on 1% (w/v) Tris acetate EDTA-agarose gels, using a GeneRuler™ 1 kb DNA ladder (MBI Fermentas). Gels were visualised using an Eagle-eye II system (Stratagene, La Jolla, CA). Template amounts were adjusted to achieve equal loading and the corresponding amounts of oligo (dT)_12-18 cDNA were used for StPT2 PCR reactions. GCT CGC GTC GGC CTC CGT CAC was used as the forward primer and CCA ATA CGG TTG GCC TCC AAT G as the reverse primer sequence. The PCR reaction conditions were as described above except that the cycle number was reduced to 30.

Analysis of root tissue for confocal microscopy followed the pre-embedding staining method of Wick et al. (Wick et al., 1981). Root tissue was fixed for 3 hours with 2% formaldehyde in phosphate buffered saline (PBS) comprising 2.7 mM KCl, 1.47 mM KH₂PO₄, 0.13 M NaCl and 8 mM Na₂HPO₄ at pH 7.3. Tissue was then incubated for 30 minutes at room temperature in PBS containing 0.5% cellulase and washed repeatedly with PBS. The tissue was blocked overnight in blocking buffer (5%(v/v) normal horse serum, 5% (v/v) normal goat serum and 50 mM L-lysine in PBS, pH 7.2) containing 2% Triton X-100 before the addition of primary antibodies. Rabbit antipeptide IgGs and JM7 were added at a concentration of 1:50 in blocking buffer and rat anti-α-tubulin at a concentration of 1:3. After 12 hours the tissue was taken out of the antibody solution, washed in PBS and incubated for a further 12 hours with fluorescent labelled secondary antibody raised in goat (Serotec) at a concentration of 1:50 in blocking buffer. Alexa Fluor 488 gave green labelling and Alexa Fluor 568,red. Tissue was stored in PBS with 1 mM sodium azide at 4°C.

Immunolabelled roots for sectioning were embedded in gelatin (0.4% w/v)containing albumen (30% w/v) and sucrose (20% w/v) polymerised by the addition of glutaraldehyde (2.5% v/v). Sections (50 μm) were cut with a vibrating microtome (Camden Instruments, Camden, UK).

Root tissue labelled by immunofluorescence was viewed with a Leica TCS confocal microscope equipped with Argon, Krypton and HeNe lasers. Cells were imaged with 10×/0.3 or 20×/0.75 PL Fluotar objectives or 40×/1.0 or 63×/1.32 PLANAPO oil-immersion objectives and recorded at 1024×1024 pixels per image. Switching off the appropriate laser line using the acousto-optical transmission filter (AOTF) in the confocal microscope showed that there was negligible `bleed-through' between channels. Fluorescent images in TIFF format were manipulated using Adobe PhotoShop and analysed using Leica TCS software.

On request, all novel material described in this publication will be made available in a timely manner for noncommercial research purposes. No restrictions or conditions will be placed on the use of any materials described in this paper that would limit their use for noncommercial research purposes.

Western blots of plasma membranes prepared from potato roots and leaves from P-fed and P-depleted potato plants were probed with an antibody raised to a peptide sequence, VEIESEEAKIEQISRDETC, corresponding to an nonconserved region of StPT2. Plants grown both hydroponically and in quartz media were studied.

Fig. 1 shows a typical result from one of three replica experiments using quartz-fed plants. Blots from both fed and depleted roots contained two broad bands of approximately 52 kDa and 30 kDa, and both bands were reduced in intensity in membranes isolated from P-fed root tissue. No bands were present in plasma membrane preparations from leaves. Similar results were obtained when hydroponically grow plants were analysed, except that the 52 kDa and 30 kDa bands were weaker in the P-fed tissue (data not shown).

The true molecular mass of StPT2, deduced from the cDNA sequence, is 58 kDa, and although an antibody to the Pi transporter from tomato, LePT1,detected a protein of the expected size on SDS-PAGE(Muchhal and Raghothama,1999), membrane proteins often behave as smaller molecules on such gel systems. Chiou et al. (Chiou et al.,2001) raised an antibody to a C-terminal sequence from MtPT1, a 59 kDa protein, that detected a diffuse protein band with an apparent molecular mass of 45 kDa on western blots of extracts from Pichia pastoristransformed with MtPT1. The antibody immunoprecipitated two ³⁵S-methionine-labelled proteins of 45 kDa and 33 kDa (believed to be an N-terminally truncated product) that had been synthesised in an in vitro transcription/translation coupled system. We therefore assume that the 52 kDa band corresponds to the transporter.

Whole tissue extracts were then made from potato roots from plants grown hydroponically both in the absence and in the presence of P. Separate extracts were made from three different parts of the root: the tip region (1 cm segments), the middle region (4 cm segments above the tip section) and 2 cm segments from the upper root region. The extracts were analysed by western blotting probed with the StPT2 antibody. The experiment was performed three times and Fig. 2 (upper panel)shows a typical result. Two bands (52 kDa and 30 kDa) were detected by the antibody in the root tip region and both bands were more intense in tissue from P-depleted plants. A further faint band of approximately 55 kDa was present in the extracts from P-depleted roots. The intensity of the 52 kDa band decreased and that of the 30 kDa band increased if the sample was incubated at 30°C for 2 hours before loading (data not shown), indicating that the 30 kDa band could be a degradation product. Furthermore, the reduced intensity of the 30 kDa band relative to the 52 kDa band in whole extracts compared with that found in plasma membrane extracts suggests that degradation of the protein could have taken place during the plasma membrane isolation procedure. We were not able to identify the 55 kDa band, but StPT2 contains two conserved phosphorylation sites(Leggewie et al., 1997) and its molecular mass relative to that of the major immunoreactive band suggests that it could be a phosphorylated form of the protein.

In extracts from the middle root region only the 52 kDa band was detected,and although the band was stronger in P-depleted tissue it was much fainter than in the tip region. In the upper root region the 52 kDa band was only visible in the extract from P-depleted tissue and it was very weak.

Total RNA was extracted from the tip, middle and upper regions of P-fed and P-depleted hydroponically grown roots using the same segments that were used for western blot analysis. The experiment was performed three times, and Fig. 2 (middle panel) shows a typical result. The RNA was standardised using cDNA generated with an 18S reverse primer as template amplified with the 18S primers as described in Materials and Methods (lower panel). Using the corresponding amounts of oligo(dT) cDNA as template, PCR reactions using the StPT2 primers generated a product of the expected size (1 kb) in each of the samples (middle panel). Expression was highest in the root tips and in the middle and upper root regions the expression was higher in P-depleted than in P-fed roots.

The results obtained from western blotting and RQRT-PCR indicated that the expression of the StPT2 in potato roots is predominantly confined to a region close to the tip. We therefore examined this area of the root using immunofluorescence and a confocal microscope to study the distribution of the transporter in more detail.

Root tissue was double labelled with StPT2, VPPase or plasma membrane ATPase antibodies and α-tubulin antibody. Plants grown both in hydroponic and in quartz media were studied in the presence and absence of P. Although it is possible to manipulate the nutrient status of hydroponically grown plants accurately, this method of culture affects the root morphology– in particular, root hair development. Hence, a comparison of StPT2 labelling in roots subjected to both treatments was performed to confirm that the localisation pattern was not due to growth conditions. In all cases at least three root tips from three separate experiments were analysed.

Fig. 3A shows a low-power image of a root tip from a P-depleted plant grown in hydroponic culture immunolabelled with StPT2 (red) and α-tubulin (green). The figure shows StPT2 labelling confined to a region just behind the root tip, between 250μm and 1200 μm from the root cap, whereas α-tubulin labelling extends the full length of the root. Quartz-grown plants showed identical staining patterns (not shown). In all P-supplied roots examined, labelling for StPT2 appeared to be confined to a slightly more restricted area, further from the tip (Fig. 3B). IgG from pre-immune serum did not label the tissue(Fig. 3C); similarly, omitting the primary antibody gave no labelling (not shown). The full length of the root was immunolabelled with the VPPase antibodies(Fig. 3D) and the ATPase antibodies (Fig. 3E). Yellow indicates colocalisation of the two secondary antibodies, and there was a higher proportion of yellow labelling in the VPPase-labelled roots and in the ATPase-labelled roots away from the tip.

To examine the transverse distribution of StPT2 at the root tip, an optical section was taken at 100 μm beneath the root surface, the maximum depth that the confocal microscope can detect fluorescence(Fig. 3F). This showed that theα-tubulin antibody labelling extends across the entire root width,whereas the red StPT2 labelling is confined to the periphery, presumably in the epidermis.

The expression of Pi transporters in root hairs has been reported in tomato, Arabidopsis and Medicargo truncatula(Chiou et al., 2001; Daram et al., 1998; Karthikeyan et al., 2002; Mudge et al., 2002). To study the expression of StPT2 in root hairs, root parts from P-depleted quartz-grown potato plants that contained hairs were labelled with the antibody. However,many of the root hairs appeared to have burst during the cell wall digestion and blocking procedures used to examine the whole roots, and there was evidence of bacterial contamination inside the tissue. Milder conditions were required (cellulase digestion reduced to 10 minutes and blocking buffer Triton X-100 concentration to 0.5%) to examine the root hairs. Fig. 3G shows a projected image of the region above the extension zone that contains root hairs(Marschner, 1995b) prepared under these conditions. StPT2 labelling (red) is present at the ends of the root hairs but not in the main part of the root beneath.

Western blotting detected StPT2 immunoreactive bands in purified plasma membrane preparations from potato roots, and tissue was examined under high magnification to further examine the subcellular localisation of the StPT2 labelling. Optical sections from tissue labelled with StPT2, plasma membrane ATPase, VPPase, α-tubulin and cell wall antibodies were compared. The cell wall provides a barrier to molecules greater than 3-5 nm, which would impede the passage of antibodies (Carpita et al., 1979), and preparation of root material included partial enzymatic digestion of the tissue in an attempt to permealise the cell wall. The peptide to which the StPT2 antibody is raised is on the central loop of the protein, which is believed to be cytoplasmically orientated, and the ATPase antibody is also raised to a cytoplasmic-facing domain(Morsomme et al., 1998). Therefore, apart from the cell wall antibody, each of the antibodies must pass through not only the cell wall, but also the plasma membrane to label their targets. The roots were treated with high concentrations (2%) of Triton X-100(Dyer and Mullen, 2001) to solubilise both the plasma membrane and the internal membranes of the cells. The complete effectiveness of this procedure could be confirmed by labelling the tissue with the VPPase because the epitope recognised by this antibody is on part of the molecule that protrudes into the vacuole(Kim et al., 1995).

Tissue 500 μm from the tip at approximately 10 μm depth from roots from P depleted plants, grown both in hydroponic and quartz media, was examined to view epidermal cells. All labelling experiments were performed at least three times on roots excised from different plants, and each gave similar labelling patterns. Fig. 4A shows a root from a hydroponically grown plant with cross-sectional views of some cells showing StPT2 labelling (red) peripheral to the α-tubulin (green) (see dotted arrow). This pattern is consistent with the localisation of StPT2 in the plasma membrane. The cells in this plane are not uniformly labelled with StPT2, however, and in some cells the labelling covers the α-tubulin labelling (solid arrow), suggesting a superficial view of the cells covered with labelled plasma membrane; in others regions, however, the cells are only labelled with α-tubulin (double arrow). The plasma membrane ATPase (red) labelled the epidermal cells more uniformly in this plane than those in Fig. 4A, with all cells showing ATPase labelling peripheral toα-tubulin (Fig. 4B) and only a few showing red covering the α-tubulin labelling (see solid arrow).

VPPase labelling (red) in double-labelled images with α-tubulin(green) at this magnification also suggests labelling associated with a membrane (Fig. 4C), with the higher proportion of yellow indicating overlapping of the labelling. To confirm that StPT2 antibodies do not label the cell wall, labelling of roots with cell wall and StPT2 antibodies was compared(Fig. 4D), and this showed that StPT2 labelling (red) is inside the green cell wall labelling.

Optical sections suggest that StPT2 labelling is on the outer surface of the root tip, in the epidermal cells and on part of the root cap(Fig. 3F). The width of roots examined near the tip ranged between 200 μm and 300 μm, which is outside the range of the confocal microscope. Therefore, to obtain a more complete picture of the lateral distribution of StPT2 labelling in whole mounted tissue we examined transverse sections. Sections of approximately 50 μm in width were taken from between 1000 and 250 μm from the root tip of hydroponically grown P-depleted plants labelled with StPT2 and α-tubulin antibodies before sectioning. Low-power images of all sections showed labelling with StPT2 only at the periphery (Fig. 5A,C) but α-tubulin labelling present across each section(Fig. 5B,C). At 300 μm from the tip (Fig. 5C) all the cells were labelled with α-tubulin, but where the root was more differentiated(700 μm, Fig. 5B), theα-tubulin became weaker in the cortical cells than in the epidermis and endodermis. StPT2 labelling was most intense at 500 μm from the tip and at 300 μm labelling began to decrease. Cross-sections of roots 700 μm from the tip labelled with StPT2 (Fig. 5D) and plasma membrane ATPase(Fig. 5E) (red) both show labelling of the epidermis in this region, and ATPase labelling is also detectable in the plasma membrane of the outer cortex. However, the ATPase labelling is present all around the epidermal cells, whereas the StPT2 labelling is confined to the apical surface and the disto-lateral regions of the cells, but is virtually absent from the basal surface. Fig. 5F,G shows dual labelling of the same sections with α-tubulin (green) (F and G) and StPT2 (F) and plasma membrane ATPase (G), red. StPT2 labelling is only present on the apical surface of the cells surrounding the α-tubulin labelling (see double dotted arrow). A small amount of ATPase labelling appears to be associated with cytoplasmic components, possibly membrane vesicles, in the epidermis(arrow head), and a superficial view of other cells has been captured in this plane (double arrow).

This distribution pattern is confirmed by measurement of the pixel density across labelled cells (Fig. 6),which shows similar intensities in the basal and apical plasma membrane in ATPase-labelled cells, but only in the apical membrane in StPT2-labelled cells.

In sections taken 250 μm from the tip ATPase labelling is found only in the epidermis and not the cortex. Flattened cells outside the epidermis are present in this region that may be `root boarder' cells from the root cap(Hawes et al., 1990), and these are labelled by both ATPase and StPT2 antibodies (not shown). Although these cells are partly detached from the root they have been shown to be viable and may play a role in plant defence against pathogens.

The presence of at least three members of the Pht1 family in potato, that differ both in their regulation and spatial expression patterns, implies that each plays a specific and distinctive role in the process of P acquisition(Leggewie et al., 1997; Rausch et al., 2001). Although an unequivocal functional analysis of the transporters can only be achieved by experiments based on antisense or knockout plants, progress towards ascertaining their physiological role can be made by accurately establishing their tissue, cellular and subcellular distribution. We have shown biochemically that StPT2 is present in root plasma membrane fractions, and western blotting and RQRT-PCR both indicate that its expression is concentrated at the root tip. By examining immunolabelled root whole mount preparations both before and after transverse sectioning we have confirmed these findings and have only found StPT2 labelling in root hairs, root cap cells and in the root epidermis, where it is confined to the apical plasma membrane. The latter observation is an important finding as in plant tissues,few plasma membrane proteins – and none believed to be directly involved in nutrient acquisition – have been shown to have polar localisation. This pattern is fully consistent with a role for the transporter in capture and uptake of Pi from the soil environment.

Concentration of StPT2 expression at the root tip may reflect the fact that the most active zone in absorption is the elongation zone near the tip, where an elevation in metabolic activity is required to sustain the degree of cell growth that occurs. This must place a high demand on cells for mineral translocation and could result in a localised upregulation of the expression of certain plasma membrane transporters. In rice, for example, maximum NH₄ uptake has been shown to occur at a distance of 1 mm behind the root tip (Colmar et al., 1998). The root tip also plays an important role in nutrient acquisition because it penetrates into fresh areas of the soil likely to be richer in nutrients than the soil around the more mature parts of the root, which forms the depletion zone(Clarkson, 1991; Marschner, 1995a).

In some plant species the whole root has been shown to be capable of absorbing Pi (Ferguson and Clarkson,1975) and of responding to P deprivation by upregulation of the rate of Pi uptake (Clarkson et al.,1978). Accordingly, the expression of some members of the Pth1 family in roots is not restricted to the tip. In tomato, LePT1, an StPT1 homologue, was found to be expressed in all parts of the root by western blot analysis, apart from the apex, and most weakly in younger tissue(Muchhal and Raghothama,1999). StPT1 and LePT1 are members of the subfamily 1(Bucher et al., 2001) and both transporters are constitutively expressed and more widely distributed in plant tissues than StPT2 (Leggewie et al.,1997; Liu et al.,1998a), which is a member of the subfamily 2, and is confined to roots. However, in another study, in situ hybridisation analysis of tomato roots showed that LePT1 expression is strongest in the differentiation zone and root cap (Daram et al.,1998). The longitudinal distribution of LePT2, also a member of subfamily 2 and more likely to resemble the pattern of StPT2, has not yet been investigated. However, MtPT1, a further member of subgroup 2 from Medicargo truncatula, is present in all root parts(Chiou et al., 2001). In Arabidopsis members of subfamilies 1 and 3 show different expression patterns within the root, with the expression of some being absent from the root tip but others being expressed in all root parts including the tip region(Karthikeyan et al., 2002; Mudge et al., 2002).

We were able to measure a detectable but much reduced level of StPT2 expression in upper root parts compared with the tip by western blotting and RQRT-PCR. However, no StPT2 labelling was found on the main root under the confocal microscope away from the tip, possibly as a result of damage during fixation to the outer surface of the epidermis in the mature tissue, where any traces of StPT2 labelling would be located. By contrast, VPPase, plasma membrane ATPase and α-tubulin labelling extended up the root.

Members of subgroups 1 and 3 of the Pth1 family are found in the epidermis,cortex and steele (particularly behind the cap in response to P depletion)(Daram et al., 1998; Liu et al., 1998a), whereas both LePT2 (Liu et al., 1998a)and MtPT1 (Chiou et al., 2001),members of subgroup 2, are found predominantly in the root epidermis. In Arabidopsis expression of Pi transporters from groups 1 and 3 has been detected predominantly in the epidermis, and also in other tissues(Karthikeyan et al., 2002; Mudge et al., 2002).

Strong expression of Pi transporters has been detected in root hairs(Chiou et al., 2001; Daram et al., 1998; Karthikeyan et al., 2002; Mudge et al., 2002). Hairs form above the root tip (Marschner,1995b), when plants are grown in solid media or aeroponic culture. The low levels of the potato Pi transporter protein in upper root parts that we detected by western blotting, compared with the levels of LePT1 reported for aeroponically grown tomato roots(Muchhal and Raghothama,1999), may have been due to a greater number of hairs on the upper roots of the tomato plants. Our quartz-grown plants had relatively few hairs,partly because of damage to the roots on removal from the solid medium. Alternatively, StPT2 may not be as highly expressed in hairs as LePT1.

GFP-promoter fusion analysis has suggested that expression of one Pi transporter in Arabidopsis roots occurs only in the trichoblasts(Mudge et al., 2002). In this plant trichoblasts form from cells overlying the cortical cell junctions(Gilroy and Jones, 2000), but we found that all epidermal cells appear to express StPT2 equally(Fig. 5F), although it may be significant that these particular roots did not form hairs. In some hair-bearing regions of Medicago truncatula roots all epidermal cells are labelled with MtPT1 (Chiou et al.,2001), but we were not able to detect labelling in the epidermal cells of the underlying root in the root hair regions.

We have shown by immunolocalisation of potato root tissue that StPT2 is only present in the plasma membrane. All members of the Pth1 family have so far been shown to be similarly localised, but this has been determined by western blotting of purified membrane fractions and heterologous transient expression (Muchhal and Raghothama,1999; Chiou et al.,2001), as opposed to the direct examination of plant tissue that we have used here.

Cytoplasmic Pi homeostasis is partly regulated at the tonoplast(Sakano et al., 1995), but no transporters involved in the bidirection movement of Pi across the vacuolar membrane have been identified. Members of the Pi transporter families could be present on the tonoplast, or possibly other endomembranes. The cytoplasmic volume is 4% of the total cell volume, causing the plasma membrane and tonoplast to be separated from each other by a narrow band of cytoplasm and not easily distinguishable under the confocal microscope. StPT2-(plasma membrane) and VPPase-(tonoplast) labelled tissue did appear different,however, as both low- and high-power images labelled with the VPPase antibody contained more yellow (colocalisation) than those labelled with StPT2. This may be because the tonoplast has more folds or projections into the cytoplasm than the plasma membrane or it may be less well fixed, causing the membrane-associated proteins to merge with cytoplasmic components. The presence of a small amount of VPPase immunoreactivity present in plasma membrane fractions has been a matter of some dispute, and a low level of targeting of the VPPase to the plasma membrane could also affect the labelling pattern (Williams et al.,1990).

Immunolocalisation has shown that the Medicago truncatulatransporter MtPT4 is in close proximity to mycorrhizal arbuscules, suggesting that it is present on the periarbuscular membrane and providing an essential step towards understanding the complex process of nutrient exchange at the interface during symbiosis (Harrison et al., 2002). It would be interesting now to examine immunolabelled potato root tissue colonised with mycorrhiza using antibodies raised to StPT3 to establish whether the transporters are similarly localised in potato.

Our high-power images of transverse sections indicate that StPT2 is restricted to the apical surface of the plasma membrane, showing cellular polarity. This also explains the absence of StPT2 label in some cells in the optical section shown in Fig. 4A, where the section must pass through these cells beneath the superficial labelling. However, labelling only on the external surface of the epidermal cells could indicate that the StPT2 antibodies can not penetrate the cell wall (Carpita et al.,1979) or that they bind nonspecifically to adhesive components present in the cell wall or on the cell surface(Clarke et al., 1979) –mucigel is present only on the surface of younger unthickened root tissue(Hay et al., 1986). However,pre-immune serum did not label the cell surface, cell wall labelling is outside that of StPT2 and membrane ATPase antibody is able to label the whole cell periphery. In all species analysed the plasma membrane ATPase is encoded by a multigene family that displays complex differential expression patterns within plant tissues. One isoform has been shown to be asymmetrically distributed in the plasma membrane of epidermal and cortical cells of the maize root apex (Jahn et al.,1998). The ATPase antibody used here, raised to 110 residues from a tobacco ATPase sequence (Morsomme et al., 1998), is likely to crossreact with more than one potato isoform.

The concentration of StPT2 at the tips of root hairs also suggests cellular polarity, and this distribution pattern is also observed for MtPT1(Chiou et al., 2001). However,it may reflect the fact that the tip region is more accessible to antibodies. The use of transformed plants expressing GFP-fusion proteins may provide more insight, although the question of the correct targeting of the GFP-tagged proteins remains.

Polar localisation of a membrane protein has recently been observed for an auxin transporter, AtP1N1. The transporter is only present at the basal end of auxin-transport competent cells (Palme and Galweiler, 1999), which facilitates the unidirectional flow of the hormone. The mechanisms responsible for the targeting of the AtP1N1 to the restricted area of the cell are currently being investigated and are believed to involve rapid actin-dependent cycling of the protein between the plasma membrane and endosomal compartments(Geldner et al., 2001). It would be interesting to establish whether actin filaments are also involved in the targeting of StPT2 to the apical plasma membrane.

Northern blots have shown that StPT2 mRNA was not detectable in roots of P-fed plants (Leggewie et al.,1997), and we found significant upregulation in response to P depletion in the upper parts of the root using western blots and RQRT-PCR. By contrast, we found that expression at the tip was still detectable, in P-fed plants. However, not only are northern blots less sensitive than RQRT-PCR but also localised StPT2 expression could be diluted beyond detectable levels in whole root extracts because the tip region represents a very small proportion of the whole root volume. Our findings suggest that it is necessary for the plant to maintain a high level of transport activity at the root tip under both circumstances in order to sustain growth. In addition, differences between transcript and protein levels could reflect a delay in protein turnover, and the existence of a number of conserved putative phosphorylation sites present within the Pht1 family suggests that regulation of the transporters at the post-translational level may also occur.

Rothamsted Research receives grant-aided support from the Biotechnology and Biological Research Council of the United Kingdom. Support was also received in the form of an Academic Links with China Scheme from the British Council. The authors would like to thank Jens Kossmann for support and encouragement and for financial contribution towards the preparation of the antisera, and Phillip R. Gordon-Weeks for guidance with the operation of the confocal microscope.