Phosphatidylinositol 4,5-bisphosphate is important for stomatal opening

Transcription

1 The Plant Journal (27) 52, doi: /j X x Phosphatidylinositol 4,5-bisphosphate is important for stomatal opening Yuree Lee 1, Yong-Woo Kim 2, Byeong Wook Jeon 1, Ki-Youb Park 1, Su Jeoung Suh 3, Jiyoung Seo 1, June M. Kwak 4, Enrico Martinoia 1,3, Inhwan Hwang 2 and Youngsook Lee 1, * 1 POSTECH-UZH Global Research Lab., Division of Molecular Life Sciences, POSTECH, Pohang, , Korea, 2 Center for Plant Intracellular Trafficking, POSTECH, Pohang, , Korea, 3 Institut für Pflanzenbiologie, Universität Zürich, 88 Zurich, Switzerland, and 4 Department of Cell Biology and Molecular Genetics, University of Maryland, College Park, MD 2742, USA Received 2 June 27; revised 21 July 27; accepted 25 July 27. *For correspondence (fax ; ylee@postech.ac.kr). Correction added after online publication, 31 October 27: correction to author s name. Summary Previously, we demonstrated that a protein that binds phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P 2 ] inhibits both light-induced stomatal opening and ABA-induced stomatal closing. The latter effect is due to a reduction in free PtdIns(4,5)P 2, decreasing production of inositol 1,4,5-trisphosphate and phosphatidic acid by phospholipases C and D. However, it is less clear how PtdIns(4,5)P 2 modulates stomatal opening. We found that in response to white light irradiation, the PtdIns(4,5)P 2 -binding domain GFP:PLCd1PH translocated from the cytosol into the plasma membrane. This suggests that the level of PtdIns(4,5)P 2 increases at the plasma membrane upon illumination. Exogenously administered PtdIns(4,5)P 2 substituted for light stimuli, inducing stomatal opening and swelling of guard cell protoplasts. To identify PtdIns(4,5)P 2 targets we performed patchclamp experiments, and found that anion channel activity was inhibited by PtdIns(4,5)P 2. Genetic analyses using an Arabidopsis PIP5K4 mutant further supported the role of PtdIns(4,5)P 2 in stomatal opening. The reduced stomatal opening movements exhibited by a mutant of Arabidopsis PIP5K4 (At3g5696) was countered by exogenous application of PtdIns(4,5)P 2. The phenotype of reduced stomatal opening in the pip5k4 mutant was recovered in lines complemented with the full-length PIP5K4. Together, these data suggest that PIP5K4 produces PtdIns(4,5)P 2 in irradiated guard cells, inhibiting anion channels to allow full stomatal opening. Keywords: PtdIns(4,5)P 2, anion channel, PIP kinase, phospholipase C, stomatal opening, guard cells. Introduction Guard cells sense environmental and physiological stimuli, and tightly regulate the stomatal aperture by responding sensitively to a wide variety of exogenous and internal stimuli such as light, temperature, internal CO 2 concentration and ABA. ABA-induced stomatal closure (Hetherington, 21; Schroeder et al., 21; Fan et al., 24) involves changes in reactive oxygen species (Pei et al., 2; Zhang et al., 21), phosphatidylinositol 3-kinase activity (Park et al., 23), calcium oscillations (McAinsh et al., 199; Allen et al., 2) and actin organization (Eun and Lee, 1997). Phospholipases C (PLC) and D (PLD) participate in the ABAinduced stomatal closure response by producing the calcium-mobilizing secondary messenger inositol 1,4,5- trisphosphate [Ins(1,4,5)P 3 ; Hunt et al., 23] and phosphatidic acid (PA). Phosphatidic acid binds to ABI1, a negative regulator of ABA responses (Leung et al., 1997), decreasing its PP2C-type phosphatase activity (Zhang et al., 24; Mishra et al., 26). The ultimate targets of many signal mediators are ion channels and pumps, which are responsible for ion influx and efflux and the resulting changes in osmotic potential that lead to stomatal opening and closure. There has been far less investigation into the stomatal opening process than into stomatal closure (Dietrich et al., 21). Light, which is a potent stimulus for inducing stomatal opening, activates the plasma membrane H + -ATPase by phosphorylation of its C-terminus (Kinoshita and Shimazaki, 1999), allowing binding of a protein and activation of the proton pump (Emi et al., 21; Kinoshita and Shimazaki, 83 Journal compilation ª 27 Blackwell Publishing Ltd

2 84 Yuree Lee et al. 22). Activation of the plasma membrane H + -ATPase is a prerequisite for stomatal opening as it leads to hyperpolarization of the membrane potential, which catalyzes opening of inward-rectifying K + channels (Schroeder et al., 1987) and provides the driving force for K + influx into guard cells. The positive charges of K + ions are counterbalanced by malate synthesis within the guard cells, and by Cl ions which enter by proton co-transport (Roelfsema and Hedrich, 25). Although the role of anion channels in ABA-induced stomatal closure is better known, they may also be involved in the regulation of stomatal opening. Slow anion channels are activated by depolarization and increasing cytosolic Ca 2+ levels, releasing Cl and other anions (Hedrich et al., 199; Schroeder and Keller, 1992). Together with the outwardrectifying K + channels, which also open in response to depolarization of the membrane potential (Schroeder et al., 1987), anion channel opening results in a decline in osmotic potential, with consequent water efflux and stomatal closure. As various anion channel inhibitors induce stomatal opening, it was suggested that these channels also play a role in the opening process (Schroeder et al., 1993; Schwartz et al., 1995; Leonhardt et al., 1999). The slow anion channels remain activated at hyperpolarized membrane potentials, often as negative as )2 mv (Linder and Raschke, 1992), and supply a background flux of anions that generate a small shunt-like pathway, controlling against further hyperpolarization and over-opening of the stomata. Phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P 2 ] is an important signal molecule that is involved in various processes such as pollen tube growth (Kost et al., 1999; Monteiro et al., 25), salt and osmotic stress (DeWald et al., 21), vesicle trafficking (Martin, 21), actin organization (Janmey, 1994; Caroni, 21), modulation of the plasma membrane vanadate-sensitive H + -ATPase (Memon and Boss, 199), ion channel activity (Hilgemann et al., 21; Liu et al., 25) and guard cell movements (Jung et al., 22). Guard cells have been shown to contain PtdIns(4,5)P 2 (Parmar and Brearley, 1993) and in Vicia faba guard cells, PtdIns(4,5)P 2 levels transiently decrease following application of ABA, suggesting a role in the ABA signaling cascade for stomatal closure (Lee et al., 1996). Furthermore, the PLC inhibitor 1-[6-[((17b)-3-methoxyestra-1,3,5[1]-trien-17-yl)amino]hexyl]-1H-pyrrole-2,5-dione (U-73122) inhibited ABAinduced calcium oscillations in guard cells and stomatal closure, providing supporting evidence for the importance of PtdIns(4,5)P 2 hydrolysis by PLC in the ABA-induced stomatal closure process (Staxén et al., 1999). In addition, PtdIns(4,5)P 2 activates PLD (Qin et al., 1997), and following ABA application the transient increase in PLD activity releases PA, which has an inhibitory effect on the inward K + channel (Jacob et al., 1999). However, PtdIns(4,5)P 2 also appears to be involved in stomatal opening. This was demonstrated using the PtdIns(4,5)P 2 -binding protein GFP:PLCd1PH, which inhibited not only ABA-induced stomatal closure, but also light-induced stomatal opening when expressed in guard cells (Jung et al., 22). Phosphatidylinositol 4,5-bisphosphate is generated from phosphatidylinositol 4-phosphate (PtdIns(4)P) or phosphatidylinositol 5-phosphate (PtdIns(5)P) by phosphatidylinositol phosphate kinase (PIP kinase). In Arabidopsis, although there are 11 type I/II PIP kinases that are predicted to produce PtdIns(4,5)P 2 from either PtdIns(4)P or PtdIns(5)P (Mueller- Roeber and Pical, 22), this activity has only been confirmed for PIP5K1 and PIP5K1 (Mikami et al., 1998; Perera et al., 25). The PIP kinase PIP5K1 belongs to the B subfamily, which contains putative membrane occupation and recognition nexus (MORN) repeats, and it is expressed strongly in procambial cells (Elge et al., 21). In Arabidopsis, PIP5K1 expression is induced rapidly by drought, salt and ABA (Mikami et al., 1998) and is regulated by a soluble protein kinase (Westergren et al., 21). The PIP kinase PIP5K1 belongs to the A subfamily, which lacks MORN repeats, and is most abundant in inflorescence stalks and flowers; its V max is 1-fold lower than PIP5K1 (Perera et al., 25). Although the presence and absence of MORN repeats suggests membrane and non-membrane localizations for PIP5K1 and PIP5K1, respectively, their cellular localizations and physiological functions remain undetermined. In this paper we confirm that PtdIns(4,5)P 2 promotes stomatal opening and identify a mechanism of its action: it inhibits anion current activation. Moreover, we describe a gene encoding a PIP5K that is expressed in guard cells, and show that this lipid kinase generates PtdIns(4,5)P 2 in vitro. We present a number of lines of evidence that support a role for this gene in light-induced stomatal opening. Results PtdIns(4,5)P 2 binding domain GFP:PLCd1PH translocates to the plasma membrane in response to white light irradiation GFP:PLCd1PH (phospholipase Cd1 pleckstrin homology domain) binds PtdIns(4,5)P 2 and is widely used as a specific biosensor for the lipid (Stauffer et al., 1998). It can be used to visualize the minute amounts of this lipid that exist in plant cells (Stauffer et al., 1998; Kost et al., 1999). Previously, we reported that overexpression of GFP:PLCd1PH in guard cells inhibited light-induced stomatal opening, probably by interfering with the normal interactions between PtdIns(4,5)P 2 and other molecules (Jung et al., 22). Therefore, this result suggests that PtdIns(4,5)P 2 is important for light-induced stomatal opening. To test whether illumination leads to increased PtdIns(4,5)P 2 content, we overexpressed GFP:PLCd1PH (Figure 1a) in V. faba guard cells and observed the localization of fluorescence before and after 3 h of irradiation with 17 lmol m )2 sec )1 white light (Figure 1b). Translocation was quantified by measuring the green fluorescence intensity of GFP from microscopic

3 Roles of PtdIns(4,5)P 2 in stomatal opening 85 Figure 1. GFP:PLCd1PH translocates from cytosol to plasma membrane in response to illumination in Vicia faba guard cells. (a) Diagrams showing the GFP:PLCd1PH and GFP:MT2a fusion constructs in 326GFP-3 G vector. NOS; terminator derived from the nopaline synthase. (b) Fluorescence images of guard cells expressing GFP:PLCd1PH or GFP:MT2a in darkness or after 3 h illumination. Bars = 1 lm. (c) Measurement of fluorescence intensity in the plasma membrane and cytosol. Guard cell fluorescence images were scanned along two lines (white bar) drawn at right angles to the long axis of the cells, at about 25% of the distance from both ends (left). From the resulting intensity profiles (right), the average peak pixel intensities of the cell boundary (black bar) and the cell interior (white bar) were obtained. (d) Relative pixel intensity of plasma membranes from guard cells transformed with GFP:MT2a and GFP:PLCd1PH in darkness or after 3 h of illumination. Means SE from 6 1 cells are shown. (e) Light- and dark-induced changes in the fluorescence ratio of GFP:PLCd1PH at the plasma membrane (PM) versus GFP:PLCd1PH in the cytosol. GFP:PLCd1PH fluorescence was visualized using time-lapse confocal microscopy for 1 h each of light and dark conditions as indicated by white and black bars at the bottom. Means SE from 17 cells are shown. (a) (b) (c) (d) GFP: PLCδ1PH GFP : MT2a Fluorescence intensity at PM / cytosol S GFP PLCδ1PH NOS 35S GFP MT2a NOS Dark Light 3 h PM region Cytosol Initial After 3 h PLCδ1PH MT2a PLCδ1PH Light Darkness (e) Fluorescence intensity at PM / cytosol Time (min) images. Fluorescence images of guard cells were scanned along two lines drawn at right angles to the long axis of the cells, at about 25% of the distance from both ends (Figure 1c, left). From the resulting intensity profiles (Figure 1c, right), the average peak pixel intensities of the cell boundary (which should include the plasma membrane) and the cell interior were obtained. The ratios of the two values were compared before and after irradiation. Initially, the intensity of fluorescence at the cell boundary was similar to that of the cytosol (mean SE = 1.9.2%, P >.5; Figure 1d, the first white bar). However, following 3 h of irradiation with white light, the fluorescence intensity was higher at the cell boundary than in the cytosol (1.37.3%, P <.1), indicating translocation of GFP:PLCd1PH from the cytosol to the plasma membrane. Although GFP:PLCd1PH can bind Ins(1,4,5)P 3 as well, it is unlikely that the increase in the fluorescence ratio was caused by a decrease in the Ins(1,4,5)P 3 level in the cytosol, as GFP:PLCd1PH was expressed at a high level in the cytosol using the 35S promoter, and its fluorescence is independent of whether it is in the bound or free state. In order to control for circadian clock-dependent translocation during the 3-h experiment, we also measured the fluorescence changes in darkness. We observed that fluorescence at the cell boundary increased slightly during the experimental period (1.16.2%, P <.5) compared with that of the cytosol. However, under light irradiation, the extent of increase in fluorescence at the cell boundary was significantly higher than that in the dark (P <.1). During stomatal opening the vacuole swells. As a result, the cytosol moves close to the nuclear area or to the periphery of the cell, a process that may resemble translocation of the protein to the nucleus or plasma membrane. To assess the extent of this effect, we constructed a fusion of GFP and the cytosolic Arabidopsis protein metallothionein 2a (MT2a; Lee et al., 24) as a negative control for translocation (Figure 1b,d). Initially, the fluorescence intensity of GFP:MT2a at the cell boundary was 1.6.3% of that in the cytosol (P >.1). However, after 3 h of irradiation with

4 86 Yuree Lee et al. white light, this had increased to % (P <.5), relative to the cytosol. This value was similar to that observed for GFP:PLCd1PH after 3 h in the dark (P >.1), but different from that following 3 h of irradiation (P <.1, Figure 1d). Therefore, we conclude that light induces translocation of GFP:PLCd1PH from the cytosol to the plasma membrane independently of the circadian clock. The translocation of GFP:PLCd1PH was partially reversed upon transfer of the cells to darkness after the light treatment (Figure 1e, n=17), further supporting the light dependency of the process. The plasma membrane is a major target in the guard cell signaling cascade, and the light-dependent translocation of GFP:PLCd1PH to this membrane suggests a function for PtdIns(4,5)P 2 in the cellular light signaling process. Stomatal opening is induced by PtdIns(4,5)P 2 The results described above suggest that PtdIns(4,5)P 2 is a factor that mediates stomatal opening. Therefore, we tested whether or not application of exogenous PtdIns(4,5)P 2 can induce stomatal opening. Vicia faba guard cells were incubated in a medium containing PtdIns(4,5)P 2 mixed with shuttle carriers (Ozaki et al., 2) that assist in intracellular delivery of PtdIns(4,5)P 2, after which their stomatal apertures were measured. Under darkness, treatment of epidermal tissues with 1 lm PtdIns(4,5)P 2 significantly enhanced circadian clock-dependent stomatal opening (P <.1). In contrast, when PtdIns(4,5)P 2 was replaced by PtdIns(4)P, no significant difference could be observed between the experimental and control stomata (P >.1, Figure 2a). The specificity of PtdIns(4,5)P 2 -induced stomatal movement was further tested using other phosphoinositides, including PtdIns(3)P, PtdIns(5)P, PtdIns(3,4)P 2 and PtdIns(3,5)P 2. Only PtdIns(3,4)P 2 slightly increased the stomatal aperture. None of the other lipids tested showed a significant effect (P >.1, Figure 2a,b). The effect of PtdIns(4,5)P 2 on stomatal opening was concentration dependent between 1 and 3 lm (Figure 2c). In Commelina communis, a similar and statistically significant effect was observed on stomatal opening following a 2-h application of PtdIns(4,5)P 2 (P <.1, data not shown). We speculated that if exogenously applied PtdIns(4,5)P 2 induced stomatal opening by increasing PtdIns(4,5)P 2 levels at the plasma membrane, then it should also have induced translocation of GFP:PLCd1PH to the plasma membrane. Indeed, a significant increase in GFP:PLCd1PH fluorescence at the cell boundary was observed at 6 min after application of PtdIns(4,5)P 2 (P <.1, Figure 2d and Supplementary Figure S1a; n=13), whereas no such translocation was observed after application of PtdIns(4)P (P >.1, Figure 2d and Supplementary Figure S1b; n=9). As PtdIns(4,5)P 2 is cleaved by PLC, it is possible that PLC inhibition may represent a mechanism for increasing (a) Stomatal aperture (μm) (c) Stomatal aperture (μm) (e) (f) Stomatal aperture (µm) GFP:PLCδ1PH Fluorescence intensity at PM / cytosol Control 1 µm PIP 2 1 µm PIP 2 2 µm PIP 2 3 µm PIP Control PI3P PI4P PI5P PI45P Control U U U U Time (min) GFP:PLCδ1PH Fluorescence intensity at PM / cytosol PtdIns(4,5)P 2 levels, and consequently stomatal opening. This hypothesis was tested by investigating the effect of U (a specific inhibitor of PLC in guard cells, as reported by Staxén et al., 1999) on stomatal opening. The guard cells (b) Stomatal aperture (μm) (d) (g) Protoplast volume (% of initial) Control PI34P 2 PI35P 2 PI45P 2 PI45P 2 PI4P Time (min) 9 Control PI4P U73122 PI45P 2 Control Darkness Light Figure 2. Phosphatidylinositol 4,5 bis-phosphate [PtdIns(4,5)P 2 ] enhances stomatal opening in darkness and induces swelling of guard cell protoplasts of Vicia faba. (a, b) Stomatal aperture of guard cells treated with 1 lm of various kinds of phosphoinositides, including PtdIns(4,5)P 2 and phosphatidylinositol 4-phosphate [PtdIns(4)P]. The epidermal peels were maintained in darkness for the entire experiment, which began.5 h prior to the photoperiod and ended at 2.5 h. During this time the stomata exhibited circadian clock-driven opening movements. Values represent the means SE from (a) and (b) stomata. (c) Stomatal apertures of guard cells treated with various concentrations of PtdIns(4,5)P 2 in darkness. Values represent the means SE of stomata. (d) PtdIns(4,5)P 2 -induced changes in the localization of GFP:PLCd1PH fluorescence of guard cells in darkness. Before and after treatment with 2 lm PtdIns(4,5)P 2 or PtdIns(4)P, GFP:PLCd1PH fluorescence was analyzed following the protocol described in Figure 1. n=13 for PtdIns(4,5)P 2 and n=9 for PtdIns(4)P. (e) Stomatal aperture of guard cells treated with.1 lm 1-[6-[((17b)-3-methoxyestra-1,3,5[1]-trien-17-yl)amino]hexyl]-1H-pyrrole-2,5-dione (U-73122) or its inactiveanalog1-[6-[((17b)-3-methoxyestra-1,3,5[1]-trien-17-yl)amino]hexyl]- 2,5-pyrrolidinedione (U-73343) inthedark. Valuesrepresentthemeans SEof stomata. (f) U inducedchanges in the localization of GFP:PLCd1PH fluorescence of guard cells in darkness. Before and after treatment with.5 lm U or U-73343,GFP:PLCd1PHfluorescencewasanalyzed.n =17forU-73122andn =6 for U (g) Effect of PtdIns(4,5)P 2 or U on the volume of guard cell protoplasts. Values represent the means SE from protoplasts.

5 Roles of PtdIns(4,5)P 2 in stomatal opening 87 (a) Current (pa) 12 (c) 1 Current (pa) mv b a c 12 mv +3 mv Current (pa) 1 1 a: Initial 2 b: NPPB 3 c: Washout Time (sec) I T1 I T (b) (d) ΔI/I T (%) I T I T Time (sec) * Time (sec) Time Control (22) PI4P (2) PI45P 2 (28) PI34P 2 (7) DAG (11) Figure 3. Phosphatidylinositol 4,5 bis-phosphate [PtdIns(4,5)P 2 ] inhibits the slow anion current activated by a depolarizing voltage stimulus applied to Vicia faba guard cell protoplasts. (a) Identification of S-type anion currents. a, Whole-cell patch-clamp recordings showing typical slow anion currents. The membrane potential was held at +3 mv to activate the S-type channel, then hyperpolarized to )12 mv for 6 sec. b, External application of 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB) resulted in inhibition of slow anion currents within 5 min. c, Following removal of the inhibitor by perfusion with control bath medium, the slow anion current recovered within 1 min. (b) Slow anion current increases with time when the membrane potential is kept depolarized at +3 mv. After establishing the whole cell configuration, the membrane potential was held at +3 mv for 3 min, after which the voltage was stepped to )12 mv (I T ) for 6 sec. The membrane potential was held at +3 mv for the next 1 min, after which the same voltage step to )12 mv was repeated (I T1 ). (c) Slow anion current of guard cell protoplasts treated with PtdIns(4,5)P 2. Starting 5 min after the first recordings (I T ), 1 lm PtdIns(4,5)P 2 was applied to protoplasts for 5 min, after which the second recordings (I T1 ) were made. (d) The effect of phosphoinositides on the time-dependent (1 min at +3 mv) increase in anion current. PtdIns(4,5)P 2 inhibited the time-dependent anion current increase, whereas the control and phosphatidylinositol 4-phosphate [PtdIns(4)P] did not (numbers in the parenthesis indicate the number of cells tested). The star indicates a significant difference in the I/I T values of protoplasts treated with 1 lm PtdIns(4,5)P 2, compared with the non-treated time control (P <.5). treated with U showed statistically significant increases in stomatal opening compared with the control (P <.1), whereas those treated with its inactive analog, 1-[6-[((17b)-3-methoxyestra-1,3,5[1]-trien-17-yl) amino]hexyl]-2,5-pyrrolidinedione (U-73343), did not (P >.1, Figure 2e). After exposure to U the stomatal apertures reached the maximum after 2 h and remained in that state for 5 h (data not shown). This effect of U on stomatal opening can be attributed to increased levels of PtdIns(4,5)P 2 at the plasma membrane, as evidenced by the translocation of GFP:PLCd1PH fluorescence to the plasma membrane 6 min after U treatment (P <.1, Figure 2f and Supplementary Figure S1c; n=17). Guard cells treated with inactive U did not show any noticeable translocation of GFP:PLCd1PH fluorescence (P >.1, Figure 2f and Supplementary Figure S1d; n=6). To confirm the role played by PtdIns(4,5)P 2 in stomatal opening, we tested whether PtdIns(4,5)P 2 could substitute for light in inducing protoplast swelling via an increase in osmotic pressure (Zeiger and Hepler, 1977; Amodeo et al., 1992). We observed a similar degree of swelling in guard cell protoplasts that were treated with either 1 lm PtdIns(4,5)P 2 or irradiated with white light for 2 min (P >.5, Figure 2g). There was no significant change in the volume of protoplasts incubated in darkness without PtdIns(4,5)P 2 or in the presence of PtdIns(4)P (P >.5, Figure 2g). In addition, the volume of guard cell protoplasts treated with U increased more than that of the controls (P <.1, Figure 2g). These results provide additional support for the suggestion that PtdIns(4,5)P 2 can substitute for light in inducing stomatal opening. Slow anion current is inhibited by PtdIns(4,5)P 2 Stomatal opening requires the coordinated and balanced activities of many ion channels and transporters. To examine whether or not PtdIns(4,5)P 2 induces stomatal opening via alteration of ion channel activities we performed wholecell patch clamping of V. faba guard cell protoplasts and analyzed K + and anion channel activities before and after

6 88 Yuree Lee et al. application of PtdIns(4,5)P 2. Inward (n =14) and outward (n =14) K + channel activities were unaltered by 1 lm PtdIns(4,5)P 2 (data not shown). As anion channels inhibit stomatal opening (Schwartz et al., 1995; Leonhardt et al., 1999), inhibition of their activities may represent a mechanism for enhancing this process. In order to measure anion currents, we used a pipette solution containing.3 lm free Ca 2+ and 2 lm guanosine 5 -triphosphate (GTP), which have been shown to enhance anion currents across the plasma membrane of guard cells (Hedrich et al., 199). S-type anion currents were identified by their typical time dependence and sensitivity to 5 lm 5-nitro-2-(3-phenylpropylamino) benzoic acid (NPPB; Figure 3a). The current magnitude measured after a 1-min exposure to +3 mv (I T1 ) increased in comparison to initial currents (I T ), a result that was expected as depolarization activates anion channels (Figure 3b; Schroeder and Keller, 1992). Treatment with PtdIns(4,5)P 2 inhibited this timedependent increase in anion currents (Figure 3c). To quantify these effects and to test whether or not 1 lm PtdIns(4,5)P 2 specifically inhibits the current, we compared the magnitude of steady-state anion currents at the end of 6 sec hyperpolarizing voltage steps applied before and after treatment with various lipids. The magnitude of current change relative to initial current (DI/I T ) (%) = [(I T1 I T )/I T ] 1 (relative current increase) was about 16 56% in untreated control cells. Protoplasts treated with 1 lm PtdIns(4)P showed a magnitude of DI/I T similar to the time control (164 5%). In contrast, anion currents from protoplasts treated with 1 lm PtdIns(4,5)P 2 showed DI/I T of 4 2%, significantly lower than the time control or PtdIns(4)P (P <.5, Figure 3d). We tested the effects of phosphatidylinositol 3,4-bisphosphate (PtdIns(3,4)P 2 ) and 1-palmitoyl-2-oleoyl-sn-glycerol (DAG) on anion current, as PtdIns(3,4)P 2 has been reported in the guard cells of C. communis (Parmar and Brearley, 1993), and DAG is a product of PtdIns(4,5)P 2 hydrolysis, as well as an inducer of stomatal opening (Lee and Assmann, 1991). Protoplasts treated with 1 lm PtdIns(3,4)P 2 and DAG exhibited slightly reduced DI/I T values, but these effects were not statistically significant (PtdIns(3,4)P 2,92 52%; DAG, 78 2%). Therefore, we conclude that of the lipids tested, PtdIns(4,5)P 2 was the most effective at inhibiting development of an anion current. Arabidopsis PIP5K4 mutants exhibit reduced stomatal opening To test whether PtdIns(4,5)P 2 is important for stomatal opening in vivo we used a genetic approach. The enzyme that produces PtdIns(4,5)P 2 is PIP kinase (PI4P5K and PI5P4K), and 11 different PIP kinases have been identified in Arabidopsis (Mueller-Roeber and Pical, 22). We obtained Arabidopsis mutants from the SALK T-DNA insertion populations deficient for these genes and tested their stomatal opening. We observed altered stomatal opening in a mutant deficient in PIP5K4 (At3g5696); the T-DNA insertion in pip5k4 was confirmed by polymerase chain reaction (PCR) of genomic DNA. The T-DNA was inserted into the first exon of PIP5K4, 1192 nucleotides downstream of the initiation codon (Figure 4a). To confirm that the pip5k4 mutant does not generate a PIP5K4 transcript, reverse transcriptase (RT)- PCR was performed using total RNA. As expected, the PIP5K4 transcript was not amplified from pip5k4, whereas it was amplified from wild-type (WT) Arabidopsis (Figure 4b). Under natural light, the pip5k4 mutant exhibited delayed stomatal opening (data not shown), and this phenotype was confirmed by performing a stomatal opening test in the dark or under white light irradiation (Figure 4c). At the beginning of the photoperiod (T = h), the apertures of pip5k4 stomata did not differ significantly from WT (P >.5), whereas after 3 h of illumination with 17 lmol m )2 sec )1 white light, the mean aperture size of pip5k4 stomata ( lm) was significantly smaller than that of WT (4.2.3 lm, P <.1). If the reduced stomatal opening in pip5k4 was due to decreased production of PtdIns(4,5)P 2, replenishment of PtdIns(4,5)P 2 should enable recovery of normal movement. We tested this idea by treating epidermal strips of pip5k4 plants with exogenous PtdIns(4,5)P 2. These strips were incubated in medium containing 1 lm PtdIns(4,5)P 2 and irradiated with white light, after which stomatal apertures were measured (Figure 4d). We observed a reduction in the light-induced opening of peeled epidermis compared with that in detached whole leaves. Stomatal apertures reached maximal opening after 4 h of illumination. The stomatal apertures of PtdIns(4,5)P 2 -treated pip5k4 ( lm) were similar to those of WT (3.2.6 lm, P >.5), and significantly larger than those of pip5k4 without treatment ( lm, P <.1; Figure 4d). The stomatal apertures of PtdIns(4)P- and PtdIns(3,4)P 2 - treated pip5k4 plants were not significantly different from those of untreated pip5k4 plants (Figure 4e). These results indicate that the reduced stomatal opening observed for pip5k4 mutants is most likely due to a reduced level of PtdIns(4,5)P 2. We tested whether U differentially affects stomatal responses in WT and pip5k4 mutant plants. Epidermal layers of WT and pip5k4 leaves were peeled off and incubated in a solution containing.1 lm U or U under darkness. In WT plants, the guard cells treated with U showed statistically significant increases in stomatal opening compared with the control (P <.1), whereas those treated with its inactive analog U did not (P >.1). Similar responses to the two drugs were observed in the pip5k4 plants (Figure 4f). To ensure that the reduced stomatal opening in pip5k4 was indeed due to the deficiency in PtdIns(4,5)P 2 caused by

7 Roles of PtdIns(4,5)P 2 in stomatal opening 89 (a) ATG prok (b) wt pip5k4 PIP5K4 Tubulin (c) Stomatal aperture (µm) wt - light wt - dark pip5k4 - light pip5k4 - dark (d) Stomatal aperture (µm) wt pip5k4 pip5k4 + PI45P (e) Stomatal aperture ( μ m) pipk4 pipk4 + PI34P 2 pipk4 + PI45P 2 pipk4 + PI4P (f) Stomatal aperture (µm) WT-control pipk4 -control WT-U73122 pipk4 -U73122 WT-U73343 pipk4 -U Figure 4. Mutation of Arabidopsis PIP5K4 (At3g5696) results in delayed stomatal opening. (a) Genetic structure of PIP5K4 and site of the T-DNA insertion. Boxes represent exons. ROK2, T-DNA present in the SALK Arabidopsis mutants. (b) reverse transcriptase-polymerase chain reaction amplification of PIP5K4 mrna. PIP5K4 transcript was amplified from wild-type (WT), but not PIP5K4 knockout plants (pip5k4). TUBULIN was amplified as a positive control. (c) Stomatal apertures of WT and pip5k4 plants. Results shown are from four independent experiments (mean SE). n (dark) = 8 97, n (light) = (d) Effects of Phosphatidylinositol 4,5 bis-phosphate [PtdIns(4,5)P 2 ] on stomatal apertures of WT and pip5k4 plants. Epidermal strips from wild-type and pip5k4 leaves were peeled and incubated on medium with or without 1 lm PtdIns(4,5)P 2 under 17 lmol m 2 sec 1 white light. Values represent the means SE from stomata. (e) No effect of phosphatidylinositol 4-phosphate [PtdIns(4)P] or phosphatidylinositol 3,4-bisphosphate [PtdIns(3,4)P 2 ] on stomatal opening movement in pip5k4 plants under 17 lmol m )2 sec )1 white light. Values represent the means SE from stomata. (f) Stomatal aperture of guard cells treated with.1 lm of 1-[6-[((17b)-3-methoxyestra-1,3,5[1]-trien-17-yl)amino]hexyl]-1H-pyrrole-2,5-dione (U-73122) or its inactive analog 1-[6-[((17b)-3-methoxyestra-1,3,5[1]-trien-17-yl)amino]hexyl]-2,5-pyrrolidinedione (U-73343). Epidermal strips from wild type and pip5k4 leaves were peeled and incubated on medium with.1 lm U or U in the dark. Values represent the means SE of stomata. mutation of PIP5K4, we transformed pip5k4 plants with a construct expressing the full-length cdna of PIP5K4 driven by its own promoter. The complemented lines expressing PIP5K4 exhibited similar stomatal opening to WT (Figure 5b). Thus, the phenotype of reduced stomatal opening in the pip5k4 mutant was recovered in complemented lines (Figure 5b). PIP5K4 is expressed in guard cells and localized to the plasma membrane To verify that PIP5K4 is expressed in guard cells, we performed RT-PCR using the same total RNA preparations of Arabidopsis guard cell and mesophyll cell protoplasts as those described by Mori et al. (26). The guard cell preparation showed little contamination with mesophyll cells (Figure 1a of Mori et al., 26). We determined that PIP5K4 was expressed in both cell types (Figure 6a). If PIP5K4 is important for light-induced stomatal opening and is responsible for light-dependent production of PtdIns(4,5)P 2 at the plasma membrane (as suggested by the results shown in Figure 1), it should localize to the plasma membrane. We investigated the localization of PIP5K4 using V. faba guard cells that had been transformed by biolistic bombardment with vector expressing GFP:PIP5K4. The fluorescence was localized to the plasma membrane (Figure 6b and Supplementary Figure S2b), and this localization did not alter in a light-dependent manner (data not shown). Free GFP was localized to the cytosol regardless of the light condition (Figure 6b and Supplementary Figure S2a).

8 81 Yuree Lee et al. (a) wt pip5k4 PIP5K4 1 PIP5K4 2 PIP5K4 Tubulin (a) (b) GC MC GFP alone PIP5K4 UBQ GFP:PIP5K4 Stomatal aperture (µm) wt pip5k4 PIP5K4-1 PIP5K4-2 (b) Figure 5. The reduced stomatal opening of pip5k4 is recovered by complementation in pip5k4 lines expressing PIP5K4 from its own promoter. (a) reverse transcriptase-polymerase chain reaction amplification of PIP5K4 mrna. PIP5K4 transcript was amplified from wild-type (WT) and complemented lines, but not from pip5k4. Tubulin was amplified as a positive control. (b) Stomatal apertures of WT, pip5k4 and complemented pip5k4 lines 1 and 2 (PIP5K4-1 and PIP5K4-2, respectively) under 17 lmol m )2 sec )1 white light. The complemented lines were produced by transforming pip5k4 plants with a construct expressing the full-length cdna of PIP5K4 under its own promoter. Results shown are from three independent experiments (mean SE, n= ). PIP5K4 has PIP kinase activity PIP5K4 comprises a conserved PIP kinase catalytic domain, a dimerization domain and the repeated MORN motif (Mueller-Roeber and Pical, 22). To test whether PIP5K4 exhibits PIP kinase activity, we purified the entire kinase protein or the catalytic domain of PIP5K4 without MORN repeats (D1 388) fused to glutathione-s-transferase (GST). Full-length proteins were less stable than the catalytic domain lacking the MORN repeats; thus we added twice as much of the full-length protein (full-length protein, 1 lg; catalytic domain, 5 lg) to the kinase assay. We determined the kinase activity of the purified fusion proteins and GST alone (as a negative control) using exogenous PtdIns(4)P as the substrate. Both GST PIP5K4 fusion proteins (with or without MORN repeats) exhibited PIP kinase activity when supplied with PtdIns(4)P, whereas GST alone did not (Figure 6c). GST PIP5K4 did not show phosphatidylinositol (PI) kinase activity when supplied with PtdIns as a substrate (Figure 6c). These results indicate that PIP5K4 encodes an active PIP kinase, which can take PtdIns(4)P, the major PtdInsP in the cell, as a substrate and produce PtdIns(4,5)P 2. Discussion In this paper we provide evidence that PtdIns(4,5)P 2 is an important signal mediator for stomatal opening, and we (c) PI4P PI PI4P PI PI4P GST GST-PIP5K4 Δ1-388 GST-PIP5K4 PI45P 2 Origin Substrate Protein Figure 6. Characterization of PIP5K4. (a) Detection of PIP5K4 mrna expression in guard cell (GC) and mesophyll (MC) protoplasts using reverse transcriptase-polymerase chain reaction. UBQ, encoding the ubiquitin-conjugating enzyme E2, was amplified as a positive control. (b) Fluorescence (upper panels) and corresponding bright-field images (bottom panels) of guard cells expressing GFP alone or GFP:PIP5K4. Bars = 1 lm. (c) Phosphatidylinositol phosphate (PIP) kinase activity of bacterially expressed glutathione-s-transferase (GST)-PIP5K4 in vitro. The purified GST-PIP5K4 fusion proteins without membrane occupation and recognition nexus repeats (D1 388) or the full-length enzyme were assayed for PIP kinase activity, as described in Experimental procedures. The whole plate is shown in a box on the left. The bottom region of the same plate is enlarged on the right. Correction added after online publication, 31 October 27: GC label in (a) corrected. identify PIP5K4 as an enzyme that synthesizes PtdIns(4,5)P 2. We demonstrate that the fluorescence intensity of a PtdIns(4,5)P 2 -binding peptide (GFP:PLCd1PH; Stauffer et al., 1998) is stronger at the plasma membrane than in the cytosol of guard cells irradiated with white light, but not in those in darkness (Figure 1b), suggesting a light-dependent increase in PtdIns(4,5)P 2 levels at the plasma membrane of guard cells. The increase in PtdIns(4,5)P 2 levels could be due to a light-induced increase in synthesis and/or a decrease in hydrolysis of PtdIns(4,5)P 2. Regardless of this, the lightdependent appearance of PtdIns(4,5)P 2 at the plasma membrane is consistent with the suggestion that it plays a

9 Roles of PtdIns(4,5)P 2 in stomatal opening 811 role in light signal transduction. Further support for this idea comes from the observation that exogenous application of PtdIns(4,5)P 2 induced stomatal opening and swelling of guard cell protoplast in darkness (Figure 2). Phosphatidylinositol 4-phosphate, a metabolite and precursor of PtdIns(4,5)P 2, is not responsible for the stomatal opening effect of PtdIns(4,5)P 2, as shown in Figure 2. This result is consistent with the previous observation that PtdIns(4)Pbinding protein, which presumably reduces the free PtdIns(4)P levels, exerts an effect opposite to that of PtdIns(4,5)P 2 -binding protein in stomatal opening movement (Jung et al., 22). The result is also consistent with the recent observation that PtdIns(4,5)P 2 synthesis, but not PtdIns(4)P synthesis, is the rate-limiting step in the plant phosphoinositides pathway (Im et al., 27). In addition, anion channel activity was altered by PtdIns(4,5)P 2, but not by PtdIns(4)P (Figure 3), which may explain, at least partly, why they have different effects on stomatal opening movement. Another hydrolysis product of PtdIns(4,5)P 2, Ins(1,4,5)P 3, is well known for its effect on stomatal closing (Blatt et al., 199; Gilroy et al., 199), which would appear to preclude an effect on stomatal opening. Moreover, U-73122, which inhibits the production of Ins(1,4,5)P 3, showed effects similar to that of PtdIns(4,5)P 2. Possible targets for the action of PtdIns(4,5)P 2 at the guard cell plasma membrane include ion pumps and channels. Recent studies in animals have shown that PtdIns(4,5)P 2 regulates a variety of ion transporters and channels, activating Na + /Ca + exchangers (Hilgemann and Ball, 1996), inwardly rectifying potassium channels (Huang et al., 1998) and the epithelial sodium channel (Yue et al., 22). In addition, PtdIns(4,5)P 2 regulates the cystic fibrosis transmembrane conductance regulator (CFTR), which functions as an anion channel enabling the passage of chloride or other anions across an electrochemical gradient (Himmel and Nagel, 24). Among the many potential target transporters in guard cells, we tested anion channel activities, and observed that those were inhibited by PtdIns(4,5)P 2 (Figure 3). The slow anion channel can play a role as a negative regulator of stomatal opening. An anion channel, when activated, releases anions, and thus depolarizes membrane potential in plant cells. Interestingly, it retains a significant opening at strongly hyperpolarized potentials, as low as )2 mv (Linder and Raschke, 1992; Schroeder and Keller, 1992; Schroeder et al., 1993), acting as a leak pathway that inhibits further hyperpolarization of membrane potential. Therefore, activation of an anion channel inhibits overactivation of inward K + channels that are responsible for the K + uptake necessary for stomatal opening. Supporting the idea of anion channels as negative regulators of stomatal opening, various studies have demonstrated that anion channel inhibitors enhance opening (Schroeder et al., 1993; Schwartz et al., 1995; Leonhardt et al., 1999). It is noteworthy that the anion channel CFTR responds differently to PtdIns(4,5)P 2 depending on its phosphorylation status: application of PtdIns(4,5)P 2 to non-phosphorylated CFTR activates a chloride current, whereas phosphorylated CFTR is inhibited. In most cases, PtdIns(4,5)P 2 regulates channel activity via direct binding. It would be interesting to elucidate the mechanism by which PtdIns(4,5)P 2 modulates anion channel activity in guard cells. However, this awaits molecular identification of the anion channels at the plasma membrane of these cells. In addition to the slow anion channel, PtdIns(4,5)P 2 can modulate other channels or pumps that are important for stomatal opening, and a candidate might be the inward K + channel, which plays an important role in stomatal opening. However, we could not find any effect of PtdIns(4,5)P 2 on K + channel activity. In contrast to our result, recently published data have demonstrated that PtdIns(4,5)P 2 restores activity of shaker-type K + channels run down following patch excision (Liu et al., 25). Although these different results may be due to cell type, they are most likely a PtdIns(4,5)P 2 concentration effect. We used 1 lm PtdIns(4,5)P 2, whereas Liu et al. (25) used concentrations up to 5 lm, which are unlikely to represent true physiological values. At relatively low concentrations of PtdIns(4,5)P 2 (2 lm), they were unable to detect any significant change in K + current from the giant patch, a result that is consistent with our data. In addition, while we used V. faba guard cells, they used oocyte cells expressing the gene encoding the K + channel. The increase in fluorescence of the PtdIns(4,5)P 2 indicator at the plasma membrane of irradiated cells is indirect evidence for de novo synthesis of PtdIns(4,5)P 2 at this site. Recently, PtdIns(4,5)P 2 synthesis, but not PtdIns(4)P synthesis, was shown to be the rate-limiting step in the plant phosphoinositides pathway. In these experiments, the ratio of PtdIns(4)P to PtdIns(4,5)P 2 in WT tobacco cells was found to be 1:1, whereas in tobacco cells expressing human PIPKIa, a 1-fold increase in plasma membrane PtdIns(4,5)P 2 was observed without any change in the PtdIns(4)P level (Im et al., 27). To investigate possible changes in PtdIns(4,5)P 2 synthesis in response to light, we measured PIP kinase activity in guard cell extracts. However, we were unable to obtain consistent results, most likely because of a very low level of enzyme activity in this cell type. As an alternative approach to test the importance of PtdIns(4,5)P 2 synthesis in light signal transduction leading to stomatal opening, we screened PIP kinase knockout mutants for altered stomatal opening. Among the two knockout mutants tested, pip5k4, which contains a mutation in PIP5K4 (At3g5696), exhibited a smaller stomatal aperture under light (Figure 4c), while pip5k3, which contains a mutation in PIP5K3 (At3g5696), did not differ from WT with respect to stomatal movement (data not shown). Normal stomatal opening movements were recovered in the pip5k4 mutant by application of PtdIns(4,5)P 2 (Figure 4d), and complementation using lines expressing PIP5K4 under its own promoter

10 812 Yuree Lee et al. (Figure 5b). These results suggest that for normal stomatal opening sufficient PtdIns(4,5)P 2 must be present in the plasma membrane and that PIP5K4 contributes to the synthesis of PtdIns(4,5)P 2. The possibility that PIP5K4 affects stomatal movement via some of its other functions is remote, although it still remains to be shown that an inactive kinase mutant of PIP5K4 cannot complement stomatal movement. Consistent with this explanation, we observed localization of this enzyme at the plasma membrane (Figure 6b). Other PIP5Ks may also participate in this pathway, as light-induced stomatal opening was not completely inhibited in pip5k4 plants (Figure 4c), and most PIP5Ks except for PIP5K3, 6, and 1 are present in guard cells, although the expression of no single gene predominates in this cell type (Leonhardt et al., 24). Under darkness, the PtdIns(4,5)P 2 level may not differ much between WT and knock-out guard cells, as the PLC inhibitor enhanced stomatal opening to similar extents in the two genotypes of plants under darkness. It is possible that PIP5K4 is mainly responsible for the light-dependent increase of PtdIns(4,5)P 2 production and that other PIP5Ks produce PtdIns(4,5)P 2 in the dark. How is PIP5K activity regulated in guard cells? With the exception of one report, which showed that its activity is reduced by phosphorylation (Westergren et al., 21), little is known about the regulation of PIP5K activity in plants. In animal cells, the activity of PIP5K I isoforms is often regulated by small Rho GTP-binding proteins such as RhoA, Rac1 and Cdc42 (Chong et al., 1994; Weernink et al., 24). Plants contain a unique subfamily of Rho GTPases, the Rop GTPases (for Rho-related proteins from plants; Li et al., 1998; Yang, 22) that are most similar to the mammalian RAC GTPases. Rop GTPases play roles in guard cell signaling (Lemichez et al., 21; B.W. Jeon, J.-U. Hwang, J.M. Kwak, Z. Yang and Y. Lee, our unpublished results), as well as in many other processes including pollen tube growth and actin organization, in which PtdIns(4,5)P 2 was also found to be important (Lin and Yang, 1997; Kost et al., 1999; Fu et al., 22). Further investigation will be required to determine whether or not Rop GTPases act as upstream regulators of PIP5Ks. Light-dependent increases in PtdIns(4,5)P 2 levels at the plasma membrane can be caused not only by increased synthesis but also by a reduction in PtdIns(4,5)P 2 hydrolysis by PLC. The Arabidopsis genome contains nine putative phosphatidylinositol-specific phospholipase C (PI-PLC) isoforms (Mueller-Roeber and Pical, 22; Hunt et al., 24), of which only PLC1 and PLC2 have been characterized (Hirayama et al., 1995, 1997). Expression of PLC1 is induced under environmental stress (Hirayama et al., 1995) and decreased expression using antisense PLC1 reduces the inhibitory effect of ABA on germination and downregulates the expression of many drought/cold-inducible genes (Sanchez and Chua, 21). Previous physiological experiments have suggested that PI-PLCs are also important for ABA signal transduction in guard cells (Staxén et al., 1999; Hunt et al., 23; Mills et al., 24). At concentrations that also inhibit recombinant PI-PLC activity, the PLC inhibitor U inhibits stomatal guard cell responses to ABA and cytosolic Ca 2+ oscillations (Staxén et al., 1999). In addition, it has been observed that reducing the level of PI-PLC in tobacco guard cells partially interferes with ABA inhibition of stomatal opening (Hunt et al., 23; Mills et al., 24). However, little is known about the function of PI-PLCs in light-induced stomatal opening. We investigated the involvement of PLC on stomatal opening using U The specificity of U as an inhibitor of PLC in guard cells was rigorously shown by Staxén et al. (1999), who showed that U-73122, but not its inactive analog U-73343, reduced activity in a recombinant plant PI-PLC, stomatal closing movement and Ca 2+ oscillation. We observed that guard cells treated with U had an accelerated circadian clock-induced stomatal opening in darkness. Guard cells treated with the inactive analog, U-73343, were no different from control cells with respect to stomatal opening (Figure 2e). Treatment with U also induced swelling of guard cell protoplast in the dark (Figure 2g). Therefore, it is possible that PLCs are also involved in the regulation of light-induced stomatal opening. An interesting question that remains to be answered is whether or not the PLCs involved in light signaling are the same as those in ABA signaling. In summary, our results demonstrate that PtdIns(4,5)P 2 is an important factor for light-induced stomatal opening and that PIP5K4 is at least partially responsible for the production of PtdIns(4,5)P 2 in guard cells. For a better understanding of the stomatal opening process, we will need to elucidate how PIP5K is regulated, what other enzymes are able to synthesize PtdIns(4,5)P 2 in guard cells, and how PtdIns(4,5)P 2 regulates the anion channel activity. Experimental procedures Plant materials and chemicals Vicia faba, C. communis and A. thaliana plants were grown for 3, 5 and 5 weeks, respectively, in a greenhouse at 22 2 C with light/ dark cycles of 16/8 h. For the delivery of bisphosphorylated phosphoinositides into the cells, Shuttle PIP TM Carrier-1, histone H1 (Molecular Probes, was used, and PtdIns(4,5)P 2 delivery was confirmed using BODIPY tetramethylrhodamine-x C 6 -PtdIns(4,5)P 2 (Molecular Probes). The fluorescence was evenly distributed at the plasma membrane and displayed a punctuate staining pattern inside the cell. For stomatal movement assays, synthetic PtdIns(4,5)P 2 (L-a-phosphatidylinositol 4,5- diphosphate), PtdIns(3,4)P 2, PtdIns(3,5)P 2, PtdIns(3)P, PtdIns(4)P, and PtdIns(5)P with dioctanoyl acyl chains were used (Sigma- Aldrich, For electrophysiological recordings, sodium salts of PtdIns(4)P and PtdIns(4,5)P 2 were purchased from Sigma-Aldrich, and 1-palmitoyl-2-oleoyl-sn-glycerol (DAG) from Avanti Polar Lipids ( and the lipid solutions were sonicated immediately before treatment. U and U were purchased from Sigma-Aldrich and dissolved in dimethyl sulfoxide.