Visualization of phosphatidylinositol 4,5-bisphosphate in the plasma membrane of suspension-cultured tobacco BY-2 cells and whole Arabidopsis seedlings

Authors

  • Wessel Van Leeuwen,

    1. Section Plant Physiology, University of Amsterdam, Swammerdam Institute for Life Sciences, Kruislaan 318, NL-1098 SM, Amsterdam, The Netherlands
    Search for more papers by this author
    • These authors contributed equally to this work.

    • Present address: Laboratory of Genetics, Wageningen University, Arboretumlaan 4, NL-6703 BD, Wageningen, The Netherlands.

  • Joop E.M. Vermeer,

    1. Section Plant Physiology, University of Amsterdam, Swammerdam Institute for Life Sciences, Kruislaan 318, NL-1098 SM, Amsterdam, The Netherlands
    2. Section Molecular Cytology, University of Amsterdam, Swammerdam Institute for Life Sciences, Kruislaan 318, NL-1098 SM, Amsterdam, The Netherlands
    3. Centre for Advanced Microscopy, University of Amsterdam, Swammerdam Institute for Life Sciences, Kruislaan 318, NL-1098 SM, Amsterdam, The Netherlands
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Theodorus W.J. Gadella Jr,

    1. Section Molecular Cytology, University of Amsterdam, Swammerdam Institute for Life Sciences, Kruislaan 318, NL-1098 SM, Amsterdam, The Netherlands
    2. Centre for Advanced Microscopy, University of Amsterdam, Swammerdam Institute for Life Sciences, Kruislaan 318, NL-1098 SM, Amsterdam, The Netherlands
    Search for more papers by this author
  • Teun Munnik

    Corresponding author
    1. Section Plant Physiology, University of Amsterdam, Swammerdam Institute for Life Sciences, Kruislaan 318, NL-1098 SM, Amsterdam, The Netherlands
      *(fax +31 205257934; e-mail munnik@science.uva.nl).
    Search for more papers by this author

*(fax +31 205257934; e-mail munnik@science.uva.nl).

Summary

Phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P2] is an important signalling lipid in mammalian cells, where it functions as a second-messenger precursor in response to agonist-dependent activation of phospholipase C (PLC) but also operates as a signalling molecule on its own. Much of the recent knowledge about it has come from a new technique to visualize PtdIns(4,5)P2in vivo, by expressing a green or yellow fluorescent protein (GFP or YFP) fused to the pleckstrin homology (PH) domain of human PLCδ1 that specifically binds PtdIns(4,5)P2. In this way, YFP-PHPLCδ1 has been shown to predominantly label the plasma membrane and to transiently translocate into the cytoplasm upon PLC activation in a variety of mammalian cell systems. In plants, biochemical studies have shown that PtdIns(4,5)P2 is present in very small quantities, but knowledge of its localization and function is still very limited. In this study, we have used YFP-PHPLCδ1 to try monitoring PtdIns(4,5)P2/PLC signalling in stably-transformed tobacco Bright Yellow-2 (BY-2) cells and Arabidopsis seedlings. In both plant systems, no detrimental effects were observed, indicating that overexpression of the biosensor did not interfere with the function of PtdIns(4,5)P2. Confocal imaging revealed that most of the YFP-PHPLCδ1 fluorescence was present in the cytoplasm, and not in the plasma membrane as in mammalian cells. Nonetheless, four conditions were found in which YFP-PHPLCδ1 was concentrated at the plasma membrane: (i) upon treatment with the PLC inhibitor U73122; (ii) in response to salt stress; (iii) as a gradient at the tip of growing root hairs; (iv) during the final stage of a BY-2 cell division. We conclude that PtdIns(4,5)P2, as in animals, is present in the plasma membrane of plants, but that its concentration in most cells is too low to be detected by YFP-PHPLCδ1. Hence, the reporter remains unbound in the cytosol, making it unsuitable to monitor PLC signalling. Nonetheless, YFP-PHPLCδ1 is a valuable plant PtdIns(4,5)P2 reporter, for it highlights specific cells and conditions where this lipid becomes abnormally concentrated in membranes, raising the question of what it is doing there. New roles for PtdIns(4,5)P2 in plant cell signalling are discussed.

Introduction

Phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P2] is a minor lipid that is thought to be in the plasma membranes of most, if not all, eukaryotic cells, where it plays various roles in cell signalling. It was first discovered as a substrate for phospholipase C (PLC) and phosphatidylinositide 3-kinase (PI3K) to generate second messengers such as inositol 1,4,5-trisphosphate (InsP3)/diacylglycerol (DAG) and PtdIns(3,4,5)P3, respectively. Later, it was found to be a signal in its own right, binding proteins involved in signalling and regulating their activity. PtdIns(4,5)P2 binds proteins with defined domains, such as PH-, ENTH- and FERM-domains, but also binds other proteins of which the three-dimensional structures complement the head group of this lipid (Carlton and Cullen, 2005; Heo et al., 2006; Yu and Lemmon, 2003). Upon binding PtdIns(4,5)P2 protein targets can become activated, for example small G-proteins, PLDs and certain ion channels (Suh and Hille, 2005; Toker, 2002). Alternatively, PtdIns(4,5)P2-rich membrane locations can ‘assemble’ complexes of signalling proteins that, via their juxtaposition, activate signalling or cellular reorganization. In this way PtdIns(4,5)P2 has been postulated to regulate membrane trafficking and actin polymerization (Balla, 2007; Carlton and Cullen, 2005; De Matteis et al., 2005; Downes et al., 2005; Halstead et al., 2005; Haucke, 2005; Hilpela et al., 2004; Varnai et al., 2006; Zoncu et al., 2007).

Most of our understanding of PtdIns(4,5)P2 as a signal and signal precursor was established for animal cells, but similar roles may exist in plants. The first plant proteins that bind and/or are activated by this lipid are now being characterized (Allwood et al., 2002; Monks et al., 2001; Pappan et al., 1997; Peterman et al., 2004; Qin et al., 2002), and more are predicted, based on PH-, ENTH- and FERM-domain sequences, in plant genomes (van Leeuwen et al., 2004; Meijer and Munnik, 2003). Similarly, several plant stress treatments have been claimed to induce PtdIns(4,5)P2 hydrolysis (Cho et al., 1993, 1995; DeWald et al., 2001; van der Luit et al., 2000; Perera et al., 1999, 2006), whereas the different lipid kinases, phosphatases and PLCs that are required for its metabolism seem abundantly present (Hunt et al., 2004; Mueller-Roeber and Pical, 2002; Torabinejad and Gillaspy, 2006). PtdIns(4,5)P2 has also been reported to rapidly accumulate: i.e. during salt stress in Arabidopsis seedlings and suspension cultures, where levels increased from 8- to 20-fold within 15–30 min (DeWald et al., 2001; Pical et al., 1999).

Not all PtdIns(4,5)P2 signalling roles seem duplicated. For example, plant genomes lack type-I and type-II PI3-kinases and the mammalian lipid–second messenger, PtdIns(3,4,5)P3, has indeed never been found in plant lipid extracts (Meijer and Munnik, 2003).

To study PtdIns(4,5)P2 in plants it has been convenient to radio-label plant material with 32P orthophosphate and to subsequently monitor treatment-induced changes in its radioactivity after extraction and thin layer chromatography. A limitation of that technique is that it only presents an average response. We do not know which cells respond or in which membrane(s) the response is located. In animal cells, a powerful new technique has been developed that provides such information. By transforming cells with a gene coding for a pleckstrin homology (PH) domain that specifically binds PtdIns(4,5)P2, and is fused to the GFP gene, the expressed GFP-PH chimera was shown to bind PtdIns(4,5)P2 in the plasma membrane in vivo. Moreover, upon PLC activation, the PtdIns(4,5)P2 was shown to be rapidly metabolized, with the fluorescence dissipating into the cytosol until the lipid was resynthesised (Stauffer et al., 1998; Varnai and Balla, 1998). The PH domain of human PLCδ1 was used because it is specific for PtdIns(4,5)P2 compared with other phosphoinositides.

In the plant field, there are some preliminary studies that have used the PHPLCδ1-GFP approach, i.e. in pollen tubes of Arabidopsis and tobacco (Dowd et al., 2006; Kost et al., 1999) and in Arabidopsis root hairs (Vincent et al., 2005). However, expression in pollen was only transient, and its use in plants in general has never been validated: i.e. it was never demonstrated to monitor intracellular PtdIns(4,5)P2 and does not cause detrimental effects on lipid signalling. In this paper, we have stably expressed a YFP-PHPLCδ1 chimera in tobacco Bright Yellow-2 (BY-2) cells and Arabidopsis plants, and have shown that it monitors increases in PtdIns(4,5)P2 in the plasma membrane but is unsuitable to report the activation PLC.

Results

Verification of the phospholipid-binding specificity of PHPLCδ1

Prior to using the PH domain of human PLCδ1 (PHPLCδ1) as a PtdIns(4,5)P2 reporter, we first wanted to confirm its lipid-binding specificity. To this end a glutathione-S-transferase (GST) fusion was made and the chimera was overexpressed in Escherichia coli. Proteins were isolated and the purified GST-PHPLCδ1 was tested in a protein–lipid overlay assay (Dowler et al., 2000, 2002). Of all seven naturally occurring polyphosphoinositides, only PtdIns(4,5)P2 bound PHPLCδ1 in a dose-dependent manner, with 0.8 pmol still being detected (Figure 1). GST itself did not bind any of these lipids (data not shown). These results confirm that PHPLCδ1 binds PtdIns(4,5)P2 with relatively high specificity and affinity, as previously documented (Dowler et al., 2000, 2002).

Figure 1.

 Lipid-binding specificity of PHPLCδ1. A protein–lipid overlay assay was conducted by spotting all naturally occuring polyphosphoinositide (PPI) isomers (indicated on the left) on a Hybond-C extra membrane, ranging from 100 to 0.8 pmol (indicated on the top). The blot was incubated overnight with purified GST-PHPLCδ1 protein (0.5 μg ml−1), and subsequent binding was detected by an anti-GST antibody and enhanced chemiluminescence.

Validation of the use of the PHPLCδ1 domain in tobacco BY-2 cells

To test the ability of the domain to function as a PtdIns(4,5)P2 biosensor in living plant cells, we fused the PHPLCδ1 domain to an enhanced version of the yellow fluorescent protein (eYFP) and stably expressed it in tobacco BY-2 cells via Agrobacterium-mediated transformation. As a control, a cell line was created expressing a construct without the PHPLCδ1 domain (i.e. only YFP).

After 5–6 weeks, stably growing cell suspensions were obtained that were subcultured every week. No growth differences between yfp, yfp-ph or wild-type (wt) BY-2 cells were ever observed over a period of ∼2 years (data not shown), indicating that there was no deleterious effect from overexpressing YFP or YFP-PHPLCδ1.

Imaging both transgenic cell lines under the confocal microscope revealed, to our surprise, very similar patterns (Figure 2a–f): the fluorescence was predominantly localized in the cytosol and there was no apparent membrane localization of YFP-PHPLCδ1. At first glance, fluorescence appears to be in the plasma membrane, but that is because the cytosol is squeezed against the plasma membrane by the huge vacuole. A true plasma membrane localization gives a homogenous, tight, fluorescent line, whereas a squeezed vacuole against the plasma membrane is much more heterogenous, i.e. diffuse. This can be checked by co-staining the cells with a membrane dye such as FM4-64, and this indeed confirmed our idea that YFP-PHPLCδ1, just like YFP alone, is predominantly localized in the cytosol (Figures S1 and S2), in contrast to mammalian cells.

Figure 2.

 Fluorescence patterns of tobacco BY-2 cells stably expressing YFP or YFP-PHPLCδ1. (a–c) Control ‘yfp’ cells, expressing YFP; (d–f) ‘yfp-ph’ cells expressing YFP-PHPLCδ1. Panels (a) and (d) show the differential interference contrast (DIC) image of panels (b) and (e), respectively. The scale bar in panels (b) and (e) is 50 μm; the scale bar in panels (c) and (f) is 10 μm.

As plant cells are notorious for their low PtdIns(4,5)P2 levels (Meijer and Munnik, 2003; Zonia and Munnik, 2006), we reasoned that the level of PtdIns(4,5)P2 in the membrane might be too low for detection.

Effect of the PLC inhibitor U73122 on the YFP-PHPLCδ1 localization in tobacco BY-2 cells

To test this idea, the effect of the PLC inhibitor U73122 was studied. The inactive analogue, U73343, was used a positive control. As shown in Figure 3, U73122 induced two striking effects. The first was an accumulation of YFP-PHPLCδ1 at the plasma membrane over a 2-h incubation period (Figure 3a–f). The second was that the fluorescence in the cytosol became more evenly dispersed: fewer cytoplasmic strands were visible and the size of the vacuoles was reduced, appearing as smooth, circular black holes in the fluorescent cytosol (Figure 3c,f). None of these effects were found when the inactive U73343 was used (Figure 3b,e). In yfp cells U73122 did not induce YFP translocation to the membrane, but it did reveal the second phenomenon (Figures 3c and S3).

Figure 3.

 Effect of the phospholipase C (PLC) inhibitor U73122 on the localization of YFP-PHPLCδ1. Confocal scanning laser microscope (CLSM) imaging of yfp (a–c) or yfp-ph cells (d–i) that were treated for 2 h with buffer (control, a, d), 10 μm of the PLC inhibitor U73122 (c, f–l) or its inactive analogue U73343 (10 μm; b, e). Panels (g–l) show the in vivo co-localization of YFP-PHPLCδ1 fluorescence with the membrane marker FM4-64 in U73122-treated yfp-ph cells. Whole cells are shown in panels (g–i), whereas part of a cell is shown in panels (j–l). Green shows YFP-PHPLCδ1 fluorescence (g, j) and red shows FM4-64 fluorescence (h, k), whereas yellow shows their co-localization (i, l). FRAP analyses on yfp (m, n) and yfp-ph cells (o, p) were performed. Fluorescence in a 6.6-μm-wide area (containing plasma membrane and cytosol) was bleached and subsequently analysed for fluorescence recovery in cells treated for 2 h either with (n, p) or without 10 μm U73122 (m, o). The average fluorescence is shown for five independently measured cells, relative to the level of fluorescence prior to bleaching, measured for 5 s. The grey area surrounding the curve indicates the standard deviation. The grey rectangle between 0.3 and 0.9 s indicates the time that the area was bleached. Scale bars: (a–c), 20 μm; (d–i) 50 μm; (j–l) 10 μm.

To investigate whether the increased fluorescence was occurring at, and not just near, the plasma membrane, two additional types of experiments were performed. In the first, the red-fluorescent lipophylic membrane marker FM4-64 was used. As shown in Figure 3(g–i), co-incubation of the yfp-ph cells with 10 μm U73122 and 2 μm FM4-64 revealed that YFP-PHPLCδ1 co-localized with FM4-64 fluorescence at the plasma membrane, whereas there was no co-localization observed in control cells (data not shown). In the second type of experiments, a FRAP (fluorescence recovery after photobleaching) approach was used. With FRAP analyses membrane-bound YFP-PHPLCδ1 molecules can be distinguished from free cytosolic YFP-PHPLCδ1 molecules, because the latter have a much higher diffusion rate. In short, a small region of membrane and cytosol, not too close to cytoplasmic strands, was bleached for 0.6 sec after which fluorescence recovery was measured with 100-msec intervals. Accordingly, FRAP analyses were performed 2 h after cells were treated either with or without 10 μm U73122. The results from a typical experiment are represented in Figure 3(m–p), showing the average fluorescence recovery (solid black line) measured in five different cells, with the surrounding grey area indicating the standard deviation. For the yfp cells, U73122 treatment had no significant effect on the fluorescence recovery (Figure 3m,n). In contrast, for yfp-ph cells the effect was dramatic (P < 0.0001; Figure 3o,p), revealing a strongly decreased diffusion rate, indicating that many of the YFP-PHPLCδ1 molecules were bound to the plasma membrane.

Together, these results suggest that YFP-PHPLCδ1, stably expressed in tobacco BY-2 cells, is a biosensor for PtdIns(4,5)P2, but that the concentration of the lipid in the membrane is normally too low for detection. Hence, most of the YFP-PHPLCδ1 is freely moving in the cytosol without bound PtdIns(4,5)P2. The U73122 data suggests a role for PLC in maintaining these low PtdIns(4,5)P2 levels, but also implies a role for PLC/PtdIns(4,5)P2 in the maintenance of cytoplasmic strands.

The effect of salt stress on the formation and localization of PtdIns(4,5)P2 in tobacco BY-2 cells

Salt stress has been shown to rapidly increase PtdIns(4,5)P2 levels in Arabidopsis seedlings and cell suspensions (DeWald et al., 2001; Pical et al., 1999). Where this PtdIns(4,5)P2 accumulates is unknown. We reasoned that salt stress could be a good system to validate whether PtdIns(4,5)P2 responses can be monitored with YFP-PHPLCδ1 and if so, where in the cell this would be. First, we checked whether our tobacco cells indeed responded to salt stress with an increase in PtdIns(4,5)P2. Therefore, cells were pre-labelled with 32Pi for 3 h and were then treated with 125 mm NaCl for different times (0–20 min). Lipids were extracted, separated by thin layer chromatography, and their radioactivity was visualized and quantified by phosphoimaging. A typical response is shown in Figure 4(a–d). PtdIns(4,5)P2 levels quickly rose by between six- and eightfold within 20 min of adding salt. In general, the 32P responses of the yfp and yfp-ph cells were always very similar (Figure 4a,b; not shown) and were indistinguishable from that of wild-type BY-2 cells (not shown).

Figure 4.

 Effect of salt stress on the formation and localization of PtdIns(4,5)P2 in tobacco BY-2 cells. The change in 32P-labelled PtdIns(4,5)P2 on an alkaline TLC is shown in response to 125 mm NaCl after 0, 5, 10 and 20 min, and for 20 min of vehicle without salt (control). The response of yfp (a) and yfp-ph cells (c) is shown. Arrowheads indicate PtdIns(4,5)P2. Panels (b) and (d) show the quantification of PtdIns(4,5)P2 in the yfp and yfp-ph cells, respectively, plotted relative to the control PtdIns(4,5)P2 levels at t = 0 and t = 20 min (control). Panels (e–l) show the in vivo pattern of YFP-PHPLCδ1 fluorescence in BY-2 cells imaged with a confocal scanning laser microscope (CLSM) after 1, 2, 3, 4, 5, 6, 7 and 8 min of treatment, respectively, with 125 mm NaCl. The arrows in panel (g) indicate membranes of which the fluorescence levels are increased (see also Video Clip S1). The scale bar in panel (e) is 20 μm. Panels (m–n) show a FRAP analysis of yfp (m) or yfp-ph (n) cells, respectively, 5 min after treatment with 125 mm NaCl. Fluorescence in a 6.6-μm-wide area (containing plasma membrane and cytosol) was bleached and subsequently analysed for fluorescence recovery. The average fluorescence is shown for 5 s, measured in five independent cells, plotted relative to the level of fluorescence prior to bleaching. The grey area surrounding the curve indicates the average level ± the standard deviation. The grey rectangle between 0.3 and 0.9 s indicates the time that the area was bleached. Bar = 10 μm.

To image the cells with a confocal microscope an eight-well Nunc chamber was used, allowing the addition of salt in a controlled manner, i.e. by adding an equal volume of cell-free medium either with or without salt. Although the final concentration of salt (125 mm) was lower than that frequently used by others, extensive plasmolysis was already visible within a minute, and continued for up to 10 min (Figure 4e–l). Because of the movement after the salt addition, it was difficult to image cells within the first minute and so images were routinely captured ∼1 min after salt addition. In yfp cells, the cytosolic localization of fluorescence never changed upon salt treatment (data not shown). In yfp-ph cells, however, plasma membrane fluorescence increased between 2 and 5 min (Figure 4e–i, arrows; Video Clip S1), revealing a local increase in the concentration of PtdIns(4,5)P2. This increased fluorescence coincided with the process of plasmolysis. After maximum shrinkage and during slow recovery (building up turgor, after 5–10 min), the fluorescence of the plasma membrane became indistinguishable from the fluorescence in the cytosol again (Figure 4l).

Not every yfp-ph cell that we imaged revealed a clear increase in plasma membrane fluorescence. We therefore also analysed cells using the FRAP technique. A typical result is presented in Figure 4(m,n), showing FRAP analyses for five independent cells 5 min after adding 125 mm NaCl to yfp and yfp-ph cells, respectively. The average fluorescence recovery is again depicted as a black line with the grey area indicating the standard deviation. FRAP analysis clearly revealed a significantly slower recovery of fluorescence than in yfp cells (P < 0.0001).

Together, these results suggest that YFP-PHPLCδ1 can monitor rapid changes in PtdIns(4,5)P2 concentrations in plant membranes, and provides evidence that the salt-induced PtdIns(4,5)P2 response actually occurs at the plasma membrane.

Expression of YFP-PHPLCδ1 in Arabidopsis thaliana

To determine whether the PtdIns(4,5)P2 biosensor could also be used in whole plants, transgenic Arabidopsis plants were generated, stably expressing YFP-PHPLCδ1 behind a 35S promoter. Homozygous T3 lines grew normally and were indistinguishable from YFP-transformed or untransformed plants (data not shown).

As shown in Figure 5, YFP-PHPLCδ1 was expressed throughout the plant, as can be expected from 35S promoter-driven gene expression. Analyzing root and leaf tissue in more detail revealed that YFP-PHPLCδ1 was predominantly localized in the cytosol again (Figure 5a–m), and was clearly excluded from the nucleus (e.g. Figure 5b, indicated by *). Overall, fluorescence images resembled those observed in the BY-2 cells expressing YFP-PHPLCδ1 (the yfp-ph cells).

Figure 5.

 YFP-PHPLCδ1 localization in Arabidopsis plants. Confocal fluorescence images taken from different cell types from Arabidopsis plants stably expressing YFP-PHPLCδ1. (a) Maximal projection of an image stack taken from a root tip, (b, c) root cortex cells, (d, e) root epidermal cells, (f, g) growing root hair, (h, i) older, not growing root hair, (j, k) leaf epidermal cells and (l, m) a guard cell. Bars: (a) 50 μm; (b–m) 10 μm. Images (a, b, d, f, h, j and l) show YFP fluorescence, whereas (c, e, g, i, k and l) show differential interference contrast (DIC) images (see also Video Clips S2–S4).

Interestingly, when analysing growing root hairs, YFP-PHPLCδ1 clearly labelled the plasma membrane at the tip (Figure 5f,g; Video Clip S2). In contrast, root hairs that had stopped growing failed to display such a tip-labelling gradient (Figure 5h,i). In leaf epidermal and guard cells YFP-PHPLCδ1 appears to be in the plasma membrane, but it is the cytosol again that is squeezed into a thin layer against the plasma membrane by the vacuole. In Video Clips S3 and S4, a 3D stack of leaf epidermal and guard cells is shown, clearly revealing that the fluorescence is located in the cytosol (note the presence of cytoplasmic strands).

As salt stress induced the translocation of YFP-PHPLCδ1 to the plasma membrane of tobacco BY-2 cells (Figure 4e–l), its effect on Arabidopsis seedlings was analysed. 32Pi labelling of 6-day-old seedlings gave a clear PtdIns(4,5)P2 increase, essentially as described by DeWald et al. (2001; data not shown). Under the confocal microscope, a clear translocation of YFP-PHPLCδ1 to the plasma membrane was observed in both root epidermal and cortex cells within 20 min of salt application (Figure 6).

Figure 6.

 YFP-PHPLCδ1 translocation in Arabidopsis roots in response to salt stress. Confocal images of YFP-PHPLCδ1 distribution in root cortex cells of 7-day-old seedlings after 15 min of incubation either without (a–d) or with 250 mm NaCl. Panels (e–p) show a confocal time-lapse following root epidermal cells after 250 mm NaCl with the time (min) indicated at the bottom right (see also Video Clip S5). Bar = 10 μm; * nuclei.

Epidermal cells first showed a strong plasmolysis, but after ∼15 min YFP-PHPLCδ1 fluorescence was visible at the plasmolysed membrane (Figure 6k–o, arrowhead; Video Clip S5). Root cortex cells, on the other hand, did not exhibit plasmolysis, but still displayed an accumulation of YFP-PHPLCδ1 fluorescence at the plasma membrane (Figure 6c,d). These results provide further evidence that the PtdIns(4,5)P2 response to salt stress occurs at the plasma membrane.

YFP-PHPLCδ1 dynamics during tobacco BY-2 cell division

While analysing numerous cells in the yfp and yfp-ph BY-2 cell suspensions, a striking phenomenon was observed when a cell was caught during the process of cell division. On examining more dividing cells a picture started to emerge, of which a typical example is represented in Figure 7, which shows a time-lapse series of a dividing yfp-ph cell over a period of 50 min. To follow the formation of the new cell membrane (the cell plate), we co-incubated cells with the red fluorescent lipophylic dye FM4-64 (lower panels, in red; Bolte et al., 2004; Dhonukshe et al., 2006). From time t = 1–31 min (Figure 7) the YFP-PHPLCδ1 fluorescence pattern was more or less identical to the YFP images in the dividing yfp cells (not shown), reflecting the distribution of non-bound YFP-PHPLCδ1 throughout the cytosol. However, at the final stage of cytokinesis, when the new growing membrane approached the original plasma membrane (Figure 7, t = 41–51 min), YFP-PHPLCδ1 started to accumulate on the edge of the cell plate; this was never observed for YFP in yfp cells (data not shown).

Figure 7.

 YFP-PHPLCδ1 dynamics during cell division of BY-2 cells. Upper panel shows YFP-PHPLCδ1 fluorescence and lower panel shows FM4-64 fluorescence (red). Time is indicated in min (also see Video Clip S6). Arrow at t = 21 min indicates cell plate-labelling with FM4-64, whereas the arrow at t = 51 min points to local elevated PtdIns(4,5)P2 levels at the final stage of cytokinesis. Bar = 20 μm.

These results suggest that the cell plate becomes rich in PtdIns(4,5)P2 just prior to fusing with the ‘old’ plasma membrane, indicating a role for this lipid in membrane fusion and cell division. Such transient, local, high concentrations of a specific lipid represent new information for plant cells, and provide a striking example of how useful this lipid biosensor can be. For a movie of this cell division, see Video Clip S6.

Discussion

YFP-PHPLCδ1, a non-invasive PtdIns(4,5)P2 reporter

In this paper a YFP fusion with the PH domain of human PLCδ1 has been used to visualize PtdIns(4,5)P2 in living plant cells. Overexpression of the construct did not cause any deleterious effects in either stably-transformed BY-2 cells or whole Arabidopsis plants: both grew and developed normally. This is important, as it is unlikely that overexpression produces abnormal images of the distribution of PtdIns(4,5)P2.

Plant cells contain large vacuoles, which squeeze the cytoplasm into a thin layer against the plasma membrane, making it difficult to discriminate between cytosolic- and plasma membrane-fluorescence. However, in this case it was clearly localized in the cytosol and not in the plasma membrane, as in animal cells (Stauffer et al., 1998; Varnai and Balla, 1998). This does not mean that plant cells contain their PtdIns(4,5)P2 in the cytosol. On the contrary, it simply reflects the fact that there is not enough PtdIns(4,5)P2 in the plant plasma membrane for YFP-PHPLCδ1 to bind to. Compared with animal cells and green algae, higher plant cells contain very low levels of PtdIns(4,5)P2 (Meijer and Munnik, 2003; Munnik et al., 1994b, 1998b; Vermeer, 2006; Zonia and Munnik, 2006). Overexpressing YFP-PHPLCδ1 may have saturated the PtdIns(4,5)P2 in the membrane, with the excess suffusing the cytoplasm with fluorescence, thereby masking the fluorescence in the membrane. Using FRAP analysis and co-labelling with FM4-64 we were able to shown that small quantities of PtdIns(4,5)P2 were indeed present in the plasma membrane. Using the PLC inhibitor U73122, a strong, although slow (1–2 h), accumulation of YFP-PHPLCδ1 at the plasma membrane was found, suggesting that these low PtdIns(4,5)P2 levels, at least in part, may be caused by a constitutive active PLC system that constantly hydrolyses PtdIns(4,5)P2. This would not only result in lower PtdIns(4,5)P2 levels but could also generate high levels of Ins(1,4,5)P3 in the cytosol. In theory, cytosolic Ins(1,4,5)P3 could also bind the PH domain because Ins(1,4,5)P3 is nothing more than the head group of the lipid. However, in plants, Ins(1,4,5)P3 levels are also extremely low (DeWald et al., 2001; Stevenson-Paulik et al., 2005; Williams et al., 2005), and we know that the affinity of PHPLCδ1 for the lipid head group is much higher than for the water-soluble Ins(1,4,5)P3 (van der Wal et al., 2001). Independently we confirmed this by expressing a tandem construct of PHPLCδ1 (i.e. YFP-2xPHPLCδ1) in cowpea protoplasts, and showed that this increased the localization of the sensor at the plasma membrane, rather than enhancing its ‘stay’ in the cytosol (Vermeer, 2006).

This study shows that PtdIns(4,5)P2 is in the plasma membrane, but that the levels are very low. Extra proof of plasma membrane-localized PtdIns(4,5)P2 came from the salt stress experiments. Treatment of BY-2 cells with 125 mm NaCl increased PtdIns(4,5)P2 levels up to ∼sevenfold within 20 min, and visually increased the concentration of YFP-PHPLCδ1 at the plasma membrane, which was confirmed with FRAP experiments (Figure 4). Similarly, the level of PtdIns(4,5)P2 was increased in Arabidopsis seedlings (not shown; DeWald et al., 2001), and was shown to translocate YFP-PHPLCδ1 to the plasma membrane in root epidermal and cortex cells. Increased plasma membrane localization was also found in growing root hairs (but not in non-growing root hairs) and during cell division, when old and new plasma membranes were about to fuse.

The importance of validation studies such as these and by Vermeer et al. (2006) is illustrated by a recent research paper in which GFP-PHPLCδ1 was wrongly used as an Ins(1,4,5)P3 reporter in plant cells (Tang et al., 2007). It has been clearly demonstrated by van der Wal et al. (2001), and by our own studies (Vermeer, 2006), that GFP-PHPLCδ1 detects PtdIns(4,5)P2 and not Ins(1,4,5)P3. Moreover, Tang et al., (2007) claim that GFP-PHPLCδ1 was localized to the plasma membrane, which is clearly not the case if one studies their figure carefully and takes into consideration the results presented here (Figure 5). They misinterpreted cytosolic fluorescence, squeezed against the plasma membrane by the large vacuole, as a plasma membrane signal, and overlooked the cytoplasmic strands disappearing from the cell periphery into the cell.

Is PtdIns(4,5)P2 the prime PLC substrate in plants?

In mammalian cells, YFP-PHPLCδ1 primarily labels the plasma membrane. Upon PLC activation, the biosensor can then be seen to translocate transiently into the cytosol, after which it returns back to the membrane. In this way, PLC signalling was visualized for the first time in a living cell (Stauffer et al., 1998; Varnai and Balla, 1998). In plants, very little PtdIns(4,5)P2 is present in the membrane; hence, most of the YFP-PHPLCδ1 is already in the cytosol. Although this demonstrates that the biosensor cannot be used as a PLC signalling reporter in plant cells, it also emphasizes that there is a big difference between the function of PtdIns(4,5)P2 in plants and animals (van Leeuwen et al., 2004; Testerink and Munnik, 2005; Zonia and Munnik, 2006).

The low concentration of PtdIns(4,5)P2 is also not supportive of a major role in PLC signalling in plants. In the literature, the rapid turnover of PtdIns(4,5)P2 has always been claimed to be able to compensate for this. However, if one looks in more detail at the 32Pi incorporation studies into the different phospholipid pools of various plant systems, i.e. suspension cultured tomato, Medicago and Arabidopsis cells, tobacco pollen, and Medicago and Arabidopsis seedlings (DeWald et al., 2001; den Hartog et al., 2001, 2003; van der Luit et al., 2000; Pical et al., 1999; Ruelland et al., 2002; Zonia and Munnik, 2004), then the PtdIns(4,5)P2 levels are always extremely low, and its turnover characteristics are by far not as striking as in mammalian cells or green algae (Munnik et al., 1998b). In animal cells and green algae, PtdIns(4,5)P2 and PtdInsP levels are more or less the same (∼1:1), whereas in higher plants the PtdInsP:PtdIns(4,5)P2 ratio ranges from 10:1 to 100:1 (Meijer and Munnik, 2003; Munnik et al., 1994a,b, 1998a,b). PtdInsP quantities seem normal in higher plants, although 32P-labelling studies tend to overestimate their levels because of rapid 32Pi uptake, the high turnover of ATP and PtdInsP, and because each lipid contains two radioactive phosphates (Munnik et al., 1994a; Vermeer, 2006).

In vitro, plant PLCs hydrolyse PtdIns4P as efficiently as PtdIns(4,5)P2, whereas PtdIns is only hydrolysed when millimolar Ca2+ concentrations are present (reviewed in Munnik et al., 1998a). As plants have ‘normal’ PtdInsP levels, perhaps this lipid is a much more realistic precursor for the production of InsP3, InsP6 and PtdOH. Recently, Dowd et al. (2006) provided evidence that in specialized cells, i.e. tip-growing pollen tubes, a correlation between PLC and the maintenance of a tip-focussed PtdIns(4,5)P2 gradient can be found. Using GFP-PHPLCδ1 and GFP-PetPLC1 they showed that GFP-PHPLCδ1 only accumulated at the plasma membrane where GFP-PetPLC1 was lacking, namely at the plasma membrane of the growing tip. Although the authors concluded that this must be PLC hydrolysing PtdIns(4,5)P2 (Dowd et al., 2006), the data may also reflect the hydrolysis of PtdIns4P limiting its phosphorylation to PtdIns(4,5)P2, which has functions on itself (see below). Using a PtdIns4P- specific biosensor (YFP-PHFAPP1) we observed a similar gradient as for PtdIns(4,5)P2 (J. E. M. Vermeer, T. W. J. Gadella, T. Munnik, unpublished data). In vivo evidence for a PtdInsP-specific PLC was recently provided in Arabidopsis leafs upon exposure to a bacterial effector protein (Andersson et al., 2006).

In summary, it seems unlikely that PtdIns(4,5)P2 simply fulfils the classic substrate role in plant PLC signalling. Interestingly, the unicellular green algae Chlamydomonas and Dunaliella do contain PtdInsP and PtdIns(4,5)P2 levels and turnover characteristics that resemble those of mammalian systems. Moreover, PLC signalling can be readily activated (Brederoo et al., 1991; Einspahr et al., 1988; Irvine et al., 1992; Munnik et al., 1998b; Musgrave et al., 1992, 1993; Quarmby et al., 1992), and in a more recent proteomic screen of Chlamydomonas flagellar proteins, a putative Ins(1,4,5)P3 receptor has been identified (Pazour et al., 2005). Putative Ins(1,4,5)P3 receptor homologues are still missing from higher plant genomes. This may represent an interesting divergence in PLC signalling between algae and higher plants (see also Zonia and Munnik, 2006).

PtdIns(4,5)P2 function in plants

What other roles could PtdIns(4,5)P2 possibly fulfil in the membrane of a higher plant cell? In animal cells it is clear that PtdIns(4,5)P2 is not only a substrate for PLC (or PI3K), but is also emerging as a signalling molecule itself, regulating the activity and/or localization of a number of target proteins, affecting various cell biological processes. A few examples, with special emphasis to the observations made in this study, have been summarized below, to increase our understanding of what might be going on.

(i) Ion channel regulation. PtdIns(4,5)P2 is increasingly being recognized as a key regulator of ion channel activity: especially K+ channels (Suh and Hille, 2005). The first plant report suggesting a role for PtdIns(4,5)P2 in K+-channel regulation has recently appeared (Liu et al., 2005). In principal, the salt stress-induced PtdIns(4,5)P2 accumulation in plant plasma membranes could reflect such a function.

(ii) Membrane fusion and fission. During mammalian cell division, PtdIns(4,5)P2 is predominantly localized in the furrow membrane and is required for the proper completion of cytokinesis (Brill et al., 2000; Emoto et al., 2005; Saul et al., 2004; Stock et al., 1999; Wong et al., 2005; Zhang et al., 1999). It is thought to provide an active zone in the cleavage furrow for efficient membrane fusion, as well as fission. The highly localized PtdIns(4,5)P2 formation during plant cell division might reflect a similar role. Interestingly, Patellin1, a novel Sec14-like protein, was recently shown to bind PtdIns(4,5)P2 and to localize to the cell plate of dividing Arabidopsis cells (Peterman et al., 2004).

(iii) Organization of the cytoskeleton. An additional role of the assembly of PtdIns(4,5)P2 at the cleavage furrow is to regulate cytoskeletal dynamics. PtdIns(4,5)P2 binds directly to a variety of actin-regulatory proteins, and can modulate their function (Carlton and Cullen, 2005; Emoto et al., 2005; Hilpela et al., 2004). Alternatively, its local increase may anchor proteins to the plasma membrane via specific PtdIns(4,5)P2-binding modules. Besides the PH domain, these include ENTH, FERM, PDZ, Tubby and WASP (Balla, 2005; Carlton and Cullen, 2005). Plants are predicted to contain several of these domains, including proteins involved in the organization of the cytoskeleton (van Leeuwen et al., 2004; Meijer and Munnik, 2003). The low PtdIns(4,5)P2 levels in plants predict that these binding domains must display a higher affinity for this lipid than their mammalian counterparts, or that additional protein domains displaying affinity for other plasma membrane components are required for efficient plasma membrane tethering. The responses observed during cell division and salt stress, but also the U71322-induced changes on the cytoplasmic strands and vacuole morphology, may be signs of such functions. In growing pollen tubes, PtdIns(4,5)P2 has also been suggested to play a role in the regulation of the actin cytoskeleton (Dowd et al., 2006).

(iv) Membrane trafficking. In mammalian cells, many polyphosphoinositides (PPIs) are involved in endo- and/or exocytosis (Carlton and Cullen, 2005; De Matteis et al., 2005; Haucke, 2005). Although we found no evidence for PtdIns(4,5)P2 accumulation in small vesicular structures, as we did for PtdIns3P and PtdIns4P (Vermeer, 2006; Vermeer et al., 2006), the polar gradient of PtdIns(4,5)P2 in growing root hairs may reflect a membrane trafficking event, as has been proposed earlier (Dowd et al., 2006; Vincent et al., 2005). Of course, ion channel regulation and organization of the cytoskeleton are other interesting possibilities that may reflect this gradient.

Summarizing, we conclude that the PtdIns(4,5)P2 biosensor is a useful tool. Although higher plants clearly exhibit a different PLC signalling mechanism than that found in animals (i), minor levels of PtdIns(4,5)P2 are present in the plasma membrane of plants (ii), which was discovered to accumulate in the newly formed membrane of dividing cells (iii) and to accumulate in the plasma membrane upon salt stress (iv).

Experimental procedures

Plant material

Tobacco (Nicotiana tabacum) BY-2 cells were grown in Murashige and Skoog (MS) medium, supplemented with Gamborg B5 vitamins, 3% (w/v) sucrose and 1 mg l−1 2,4-dichlorophenoxy-acetic acid (2,4-D; Duchefa, http://www.duchefa.com). Cells were cultured in the dark at 24°C, and were shaken in 300-ml Erlenmeyer flasks containing 50 ml MS medium at 125 rpm.

Seedlings of A. thaliana cv. Columbia were cultivated on 0.5x MS plates in a growth chamber at 21°C with a 16-h light period. Mature plants were grown on soil in the greenhouse following the same regime.

Constructs and generation of transgenic BY-2 cells and Arabidopsis plants

The human PLCδ1-PH domain (amino acids 1–175) was fused to the C-terminus of eYFP (Clontech, http://www.clontech.com), which had an additive Q70K mutation to enhance its stability. The resulting eYFP-PHPLCδ1 sequence was XbaI-BamHI cloned into the pGreen-1K vector, containing a double CaMV 35S promoter (from –343 to –90 and from –343 to +8) and the NPTII gene as a kanamycin resistance marker. An eYFP-Q70K was cloned for control experiments.

Vectors were transformed into Agrobacterium tumefaciens strain EHA105 and subsequently into BY-2 cells. In short, 200 μl of an overnight (O/N) bacterial culture was mixed with 8 ml of (5-days old) BY-2 cells, supplemented with 0.2 mm acetosyringone. Cells were kept for 3 days in a sterile Petri dish, at 24°C in the dark, and were subsequently transferred to fresh MS plates (with 2,4-D, 3% sucrose and 0.8% agar), containing carbenicillin (250 mg l−1) and increasing concentrations of kanamycin, every 3 days: two times 40 mg l−1, two times 100 mg l−1 and two times 200 mg l−1 kanamycin, respectively. Calli were maintained on 200 mg l−1 kanamycin plates and were transferred to fresh plates every 3 weeks. Transgenic calli were checked for YFP fluorescence using a fluorescence stereomicroscope, and were selected for inoculation into liquid medium. Afer 3–6 weeks a stably growing cell suspension was obtained, which was subcultured every week and used after 4–6 days for both microscopic and biochemical analysis.

Arabidopsis thaliana cv. Columbia plants were transformed using either the eYFP or the eYFP-PHPLCδ1 constructs by standard floral-dip transformation (Clough and Bent, 1998). Kanamycin-resistant T2 plants were selected for high fluorescence levels using a fluorescence stereomicroscope. Homozygous T3 plants were used for further analysis. Seeds were germinated on 0.5x MS plates with 0.8% agar after 48 h of stratification, and were grown at 21°C as described above.

Overexpression and purification of GST fusion proteins

Both GST and GST-PHPLCδ1 constructs were cloned into the pGEX-KG vector using PCR and pfu polymerase, and were used to transform into the Escherichia coli strain BL21 (DE3). Expression was induced with 1 mm isopropyl-beta-D-thiogalactopyranoside (IPTG) for 9 h at 20°C, after which total protein was extracted using lysozyme, freeze/thaw and sonication. GST fusion proteins were further purified using 1-ml GSTrap FF columns (Amersham, http://www.amersham.com), after which the pure protein was quantified and stored at −20°C until use in the protein lipid overlay assay.

Protein lipid overlay assay

Serial dilutions of all seven known PPI isomers (CellSignals, Inc., Lexington, KY, USA) were spotted onto a Hybond-C extra membrane (Amersham), ranging from 0.8 to 100 pmol. The membrane was dried, washed in binding buffer and incubated with the purified GST-PHPLCδ1 fusion protein, essentially as described by Dowler et al. (2000, 2002). Binding was detected using an anti-GST antibody (Santa Cruz Biotechnology, Inc., http://www.scbt.com) followed by ECL.

Confocal imaging

Routinely, 4–6-days-old weekly subcultured BY-2 cells were imaged in eight-well Nunc chambers (Nunc, Inc., http://www.nuncbrand.com). Cells (150 μl), at a standard concentration of 20 mg fresh weight ml−1 medium, were treated by adding an equal volume of cell-free medium either with or without 20 μm U73122 (1-[6-((17β-3-methoxyestra-1,3,5(10)-trien-17-yl)amino)-hexyl]-1H-pyrrole-2,5-dione; 20 μm U73343 (1-[6-((17β-3-methoxyestra-1,3,5(10)-trien-17-yl)amino)-hexyl]-2,5-pyrrolidinedione) (Calbiochem, http://www.merckbiosciences.co.uk/html/CBC/home.html) or with 250 mm NaCl. Confocal images were acquired using a confocal scanning laser microscope (CLSM; Zeiss LSM510; Zeiss, http://www.zeiss.com) with a Zeiss c-apochromat 40× (NA 1.2) water-immersion objective and a pinhole corresponding to 1–2 Airy units. YFP was excited using a 514-nm argon ion laser line, and fluorescence was detected using a 540/20-nm bandpass filter. FM4-64 (N-(3-triethylammonium-propyl)-4-(6-(4-(diethylamino)phenyl) hexatrienyl) pyridinium dibromide; Invitrogen, http://www.invitrogen.com) was added to a final concentration of 2 μm, and its fluorescence was detected using the same laser line and a 650-nm long-pass filter. Each scan was the result of averaging eight time-frames to produce a high signal-to-noise ratio.

32Pi phospholipid labelling, extraction and analysis

BY-2 cells were labelled in their own growth medium using 100 μCi ml−1 32PO43– (carrier-free; Amersham). Labelling took place in 2-ml Eppendorf ‘safelock’ tubes, containing 85-μl aliquots. After 3 h, cells were treated with an equal volume of cell-free medium either with or without 20 μm U73122, 20 μm U73343 or 250 mm NaCl (to give half the final concentration) for the times indicated. Incubations were stopped by adding 20 μl 50% perchloric acid. Lipids were extracted by adding 3.75 vol. CHCl3/MeOH/HCl (50:100:1 by vol.). A two-phase system was induced by adding 3.75 vol. CHCl3 and 1 vol. NaCl (0.9% w/v), and was processed as described previously (Munnik et al., 1996).

Lipids were separated on Silica-60 TLC plates (Merck, http://www.merck.com) employing an alkaline TLC solvent composed of CHCl3/MeOH/25% NH4OH/H2O (90:70:4:16, by vol.; Munnik et al., 1994a). Radioactivity was visualized by autoradiography and quantified by phospho-imaging (Storm; Molecular Dynamics, http://www.gelifesciences.com). The radioactivity in PtdIns(4,5)P2 was calculated by dividing it by the radioactivity incorporated in the total lipid fraction. The fold increase was calculated by dividing the radioactivity in the stimulated cells by the radioactivity in the corresponding control. Autoradiographs shown represent general phenomena, representative for 3–6 independent experiments.

Acknowledgements

We thank Tobias Meijer for providing the PH domain from HsPLCδ1, and Alan Musgrave for exciting ideas and discussion. The work was primarily supported by the Netherlands organization for scientific research [NWO), research council Earth and Life Sciences (ALW); project numbers 810-66.012 (JEMV, TWJG) and 810.66.011 (WvL, TM)]. TWJG’s lab was additionally supported by an EU Integrated project on Molecular Imaging (LSHG-CT-2003-503259) and TM’s lab was supported by NWO-ALW (813.06.0039, 863.04.004 and 864.05.001), NWO-CW (810-36.005; 700.56.007), the EU (HPRN-CT-2000-00093; HPRN-CT-2002-00251) and the Royal Netherlands Academy of Arts and Sciences (KNAW).

Ancillary