Phosphatidylinositol (4,5)bisphosphate inhibits K+-efflux channel activity in NT1 tobacco cultured cells.

In the animal world, the regulation of ion channels by phosphoinositides (PIs) has been investigated extensively, demonstrating a wide range of channels controlled by phosphatidylinositol (4,5)bisphosphate (PtdInsP2). To understand PI regulation of plant ion channels, we examined the in planta effect of PtdInsP2 on the K+-efflux channel of tobacco (Nicotiana tabacum), NtORK (outward-rectifying K channel). We applied a patch clamp in the whole-cell configuration (with fixed "cytosolic" Ca2+ concentration and pH) to protoplasts isolated from cultured tobacco cells with genetically manipulated plasma membrane levels of PtdInsP2 and cellular inositol (1,4,5)trisphosphate: "Low PIs" had depressed levels of these PIs, and "High PIs" had elevated levels relative to controls. In all of these cells, K channel activity, reflected in the net, steady-state outward K+ currents (IK), was inversely related to the plasma membrane PtdInsP2 level. Consistent with this, short-term manipulations decreasing PtdInsP2 levels in the High PIs, such as pretreatment with the phytohormone abscisic acid (25 microM) or neutralizing the bath solution from pH 5.6 to pH 7, increased IK (i.e. NtORK activity). Moreover, increasing PtdInsP2 levels in controls or in abscisic acid-treated high-PI cells, using the specific PI-phospholipase C inhibitor U73122 (2.5-4 microM), decreased NtORK activity. In all cases, IK decreases stemmed largely from decreased maximum attainable NtORK channel conductance and partly from shifted voltage dependence of channel gating to more positive potentials, making it more difficult to activate the channels. These results are consistent with NtORK inhibition by the negatively charged PtdInsP2 in the internal plasma membrane leaflet. Such effects are likely to underlie PI signaling in intact plant cells.

Many environmental and internal signals induce increased metabolism of phosphoinositides (PIs) in both animal and plant cells. PIs are involved in numerous cellular processes important for cell develop-ment and growth, including secretion of metabolites, vesicular transport, organization of the cytoskeleton, and regulation of ion channels and transporters (for review, see Hilgemann, 2003;Meijer and Munnik, 2003;Suh and Hille, 2005;Huang, 2007). In many signaling cascades, phosphatidylinositol (4,5)bisphosphate (PtdInsP 2 ) undergoes cleavage by phospholipase C (PLC), producing diacylglycerol and inositol (1,4,5)trisphosphate (InsP 3 ). While diacylglycerol operates within the plane of the membrane, production of InsP 3 is practically synonymous with further signaling through mobilization of Ca 2+ from internal stores (Berridge and Irvine, 1984;Berridge et al., 1998).
In the last decade, signaling via PtdInsP 2 itself, rather than via its cleavage products, has become the focus of intense study in animal cells. Detailed descriptions have accumulated about PtdInsP 2 interactions with membrane ion channels and other ion transporters, highlighting its important roles in conveying information about physical and chemical stimuli and in maintaining cellular ionic homeostasis (for review, see Suh and Hille, 2005;Huang, 2007).
In the plant world, this topic has been evolving at a much slower pace. Over two decades of study focusing on plant signaling via the PI pathway and Ca 2+ mobilization have linked signals that cause cell shrink-ing with the cleavage of PtdInsP 2 , for example, in the case of the phytohormone abscisic acid (ABA) in Vicia faba guard cells (Lee et al., 1996) or blue light in the leaf motor cells, such as "flexors" of Samanea saman (Kim et al., 1996) and "extensors" of Phaseolus coccineus (Mayer et al., 1997). Moreover, a strong positive correlation has been revealed between these "shrinking signals" and the activation of anion-and K + -release channels in the plasma membrane of the shrinking cells (Blatt, 2000;Suh et al., 2000;Hetherington, 2001;Schroeder et al., 2001;Moran, 2007aMoran, , 2007bPandey et al., 2007). Accumulating evidence dissociated the activation of these channels from obligatory Ca 2+ mobilization, suggesting alternative signaling pathways: for example, Marten et al. (2007), using fluorescent Ca 2+ -reporter dye, demonstrated ABA activation of anion channels in V. faba and tobacco (Nicotiana tabacum) guard cells in the absence of or prior to an observable rise in cytosolic [Ca 2+ ]. Similarly, Lemtiri-Chlieh and MacRobbie (1994) found in patch-clamp experiments with Ca 2+ -buffered cytosolic milieu that ABA activation of K + -release channels in V. faba guard cells was independent of cytosolic Ca 2+ . Importantly, and consistent with the latter report, direct examination of K + -release channels in excised membrane patches from V. faba guard cells (Hosoi et al., 1988) or from S. saman motor cells (Moshelion and Moran, 2000) revealed that their activity did not require Ca 2+ at their cytosolic surface. Cytosolic alkalinization or membrane depolarization resulting from H + pump inhibition and anion channel activation have been proposed as alternative mediators of ABA activation of guard cell K + -release channels (Irving et al., 1992;Macrobbie, 1992;Blatt and Armstrong, 1993; for a recent review, see Pandey et al., 2007). However, depolarization accounted only partially for the stimulation of K + -release channel activity by a "shrinking" blue light in the case of S. saman leaf motor cells (Suh et al., 2000).
Could PtdInsP 2 itself modulate K + -release channels? To address this question, Liu et al. (2005) examined three plant K channels, two K + influx channels, LKT1, a tomato (Solanum lycopersicum) homolog of the Arabidopsis (Arabidopsis thaliana) AKT1, the Arabidopsis KAT1, and the Arabidopsis K + -efflux channel, SKOR. Applying a patch clamp to frog oocytes expressing these channels, they demonstrated Ca 2+ -independent regulation by PI lipids. Isomers of PtdInsP, PtdInsP 2 , or PtdInsP 3 applied to the cytosolic side of excised membrane patches prevented the rundown and enhanced the activity of these channels (Liu et al., 2005). These results, resembling direct activation of various types of animal ion channels by PtdIns(4,5)P 2 (for review, see Hilgemann, 2004;Suh and Hille, 2005;Huang, 2007), suggested a direct correlation between PtdInsP 2 level and SKOR activity in oocytes (Liu et al., 2005).
In contrast, interpretation of in planta experiments leads to a direct relation between the up-regulation of the K + -release channel and PtdInsP 2 cleavage (i.e. to an inverse correlation between PtdInsP 2 level in the membrane and channel activity). This can be seen in the case of SKOR-like channels in guard cells (Blatt, 1990;Blatt et al., 1990;Lemtiri-Chlieh and MacRobbie, 1994;Lemtiri-Chlieh, 1996) or in S. saman flexors (Kim et al., 1996;Suh et al., 2000).
To reconcile this apparent discrepancy, we undertook to examine this relationship more explicitly in planta. We investigated NtORK in tobacco cultured cells (NT1), the previously described representative of the SKOR-related K + -release channel family (Kasukabe et al., 2006;Sano et al., 2007Sano et al., , 2008. We used NT1 with genetically manipulated basal levels of PtdInsP 2 and InsP 3 (Perera et al., 2002;Im et al., 2007). Three types of cell lines were studied: "Low PIs," cells with lowered (relative to controls) levels of PtdInsP 2 and InsP 3 ; "High PIs," cells with elevated levels of both PIs; and control cell lines transformed with "empty" vectors and the wild type. Using a patch clamp in a whole-cell configuration, with the cytosolic concentrations of Ca 2+ and protons tightly controlled by appropriate buffers, we established in these cells a negative correlation between the basal level of PtdInsP 2 and NtORK activity, as expected from the reported plant motor cell findings. Further manipulations of the PtdInsP 2 membrane levels, causing either its cleavage by the plant hormone ABA and by elevation of bath pH or its accumulation by PLC inhibition, strengthened this conclusion. Taken together, our results strongly suggest the inhibition of NtORK by PtdInsP 2 . In addition, the initiation of a signaling cascade by ABA in the NT1 cells, which is characteristic of guard cells, should further encourage the use of this model system in studies of signal transduction.

Channel Identification through Comparison with Previous Work
In whole-cell patch-clamp configuration, increasingly depolarizing voltage pulses evoked ion channel activity. This is evident in all protoplasts of the three types of NT1 cell lines, the Low PIs, the High PIs, and the various controls ( Fig. 1A; see "Materials and Methods"), as time-dependent outward currents. The sigmoidal time and voltage dependencies of these currents resembled those of outward K + currents, flowing through K channels, described already in the similar tobacco cell line BY2 (Stoeckel and Takeda, 2002;Sano et al., 2007).

Channel Identification by Blockers
We identified the channels as K + -conducting channels by following treatments with the classical K channel blockers: (1) internal Cs + to block the outward current (Supplemental Fig. S1A), (2) external Cs + to block the inward "tail" current (Supplemental Fig. S1B), and (3) external Ba 2+ to block both the outward and the inward currents (Supplemental Fig. S1C). Furthermore, in all of the cell lines, the reversal potentials of these currents were only slightly more depolarized than the calculated K + equilibrium (Nernst) potential of 281 mV (Supplemental Fig. S2), providing additional support for identifying these channels as K + -selective K + -release channels, NtORK channels (Sano et al., 2007). Actually, although two genes, NtORK1 and NtORK2, have been mentioned in the literature, only NtORK1 has been characterized to some extent so far (Langer et al., 2002;Kasukabe et al., 2006;Sano et al., 2007). Because our measurements do not indicate clearly two separate types of channels, we use the inclusive name, NtORK.
Correlating Channel Activity with PtdInsP 2 Level At external pH 5.6, the NtORK channels were most active in the Low PIs, least active in the High PIs, and they displayed intermediate activity in the control protoplasts, as evident from the net, steady-state K + currents normalized to cell capacitance, I# K ( Fig. 1; Supplemental Fig. S3,A and B;see "Materials and Methods"). The effect of the InsP 3 5-phosphatase with which the Low PIs were generated was evident in the cells where the 5-phosphatase was targeted to the plasma membrane (I2-8 and I4-2; Perera et al., 2002;Supplemental Fig. S4). When the 5-phosphatase was inactive (C348S) or missing the C terminus and therefore no longer bound to the plasma membrane (DC), the channels behaved similarly to the wild-type or "empty-vector" (C5, GFP) controls (Supplemental Fig.  S3A). These data indicate that plasma membrane targeting, not just the ability to hydrolyze InsP 3 (Supplemental Fig. S5), is essential. It is most likely that the InsP 3 5-phosphatase needs to be in close proximity to the channel to have an effect.
In contrast to I# K , the baseline "leaks" (normalized instantaneous currents), I# O , were larger in the High PIs than in all other cells, but in all cell types, leak conductance was constant over the whole voltage range and its ion selectivity was poor. We base this latter conclusion on the linearity of the I# O -E M (currentmembrane potential) curves and their nearly null intersection with the abscissa, which was an intermediate value between all equilibrium potentials calcu-

Components of Channel Inhibition
In an attempt to resolve the possible effects of PtdInsP 2 on NtORK into effects on properties of the open channel versus effects on channel gating (i.e. opening and closing), we examined the voltage dependence of its chord conductance, G# K , extracted from I# K (Eq. 1 in "Materials and Methods"). Three characteristic Boltzmann parameters served to compare the cell lines: G# max , the maximum attainable conductance (normalized to cell capacitance); E 1/2 , the voltage at which half of G# max is attained; and z, the effective charge of the gating subunits (Eq. 2 in "Materials and Methods"). As could be deduced from the current measurements, cells with higher PtdInsP 2 levels had lower G# max , and vice versa (Fig. 2, A and B). This analysis revealed also a mean difference of about 23 mV between E 1/2 values of the High PIs and the rest of the cell lines. This difference is concealed in the I# K -E M plots (Supplemental Fig. S3, A and B) and even in the G# K -E M plots ( Fig. 2A) and is clearly visible only in the plots of the probability of opening, P O , versus membrane potential (Eq. 3 in "Materials and Methods"). Thus, the P O -voltage relationships (P O -E M s) of the High PIs, reflecting the voltage dependence of the channel-gating subunits, were shifted significantly to more depolarizing membrane potentials (rightward) relative to Low PIs and controls (Fig. 2, B and C). z values, reflected in the "slopes" of the G# K -E M curves, did not vary among the three cell types (High PIs, Low PIs, and controls; Fig. 2B).

Effects of ABA on Plasma Membrane PtdInsP 2 Content
If indeed NtORK activity depends inversely on the membrane PtdInsP 2 content, lowering the PtdInsP 2 level should increase NtORK activity. Therefore, we examined whether we can lower the PtdInsP 2 in the high-PI NT1 cells utilizing the plant hormone ABA, which is known to activate PLC in guard cells (Lee et al., 1996). Indeed, a 15-min incubation of protoplasts isolated from the high-PI cells with 25 mM ABA significantly lowered the PtdInsP 2 levels (Fig. 3). The mean effects of ABA on the already low levels of PtdInsP 2 present in wild-type cells were minute, and in the GFP cells they were insignificant.

Effects of ABA on NtORK
In the patch-clamp experiments, 10-to 20-min preincubation of protoplast with 25 mM ABA increased I# K in the High PIs and had no effect on I# K in the control cell lines (Fig. 4A). This correlated with changes in the PtdInsP 2 levels. Boltzmann analysis of NtORK G# K revealed that in the High PIs, but not in the controls, ABA pretreatment increased G# max considerably and shifted E 1/2 by about 15 mV in a hyperpolarizing (leftward) direction (Fig. 5). z was not affected by ABA (Fig. 5B).

Effects of pH on the Content of Plasma Membrane PtdInsP 2
Because a small alkalinization of the apoplast was correlated with stomatal closure in Vicia faba and potato (Solanum tuberosum) leaves (Hedrich et al., 2001), we hypothesized that pH would have an ABA-mimicking (although ABA-independent) effect. Incubating the wild-type cells for about 15 min in external solution buffered to pH 7.0 (rather than to pH 5.5) decreased the plasma membrane PtdInsP 2 levels only by roughly 10 units (pmol min 21 mg 21 plasma membrane protein), and in GFP cells the mean PtdInsP 2 level was not affected (Fig. 6). In contrast, in high-PI protoplasts, the pH shift decreased the PtdInsP 2 levels by about 100 to 440 units ( Fig. 6), resembling the effect of ABA (Fig. 3).

Effects of pH on NtORK
In correlation to the effect of pH on PtdInsP 2 levels, 10-to 20-min incubation of protoplasts in the external solution buffered to pH 7 prior to attaining a wholecell patch-clamp configuration increased I# K in the High PIs but did not affect I# K in the control cell line (Fig. 7A). Boltzmann analysis of NtORK G# K -E M relationships ( Fig. 8) revealed an increase of the maximum conductance, G# max (Fig. 8, A and B) and a hyperpolarizing (leftward) shift of about 36 mV of the P O -E M relationship (Fig. 8, B and C). In contrast to enhancing I# K (i.e. NtORK activity), pH 7 significantly decreased the leak currents in both High PIs, down to its levels in the control cells (Fig. 7B).

Effects of PLC Inhibitor
If PtdInsP 2 inhibits NtORK activation, then preventing its cleavage by inhibiting PLC, consequently leading to PtdInsP 2 accumulation, should decrease NtORK conductance. In ABA-pretreated protoplasts from High PIs (to promote their NtORK activity) or in wild-type protoplasts, a bath exposure to U73122 (the commonly used PI-PLC inhibitor; Bleasdale et al., 1990) decreased I# K progressively within a few minutes (Fig. 9). Adding the same volume of the external solution to the bath with the High PIs or adding U73343, the inactive analog of U73122 (Bleasdale et al., 1990), to wild-type protoplasts had no immediate discernible effect on I# K (Fig. 9, B and D). Longer term incubation of control cell protoplasts with U73122 (including several minutes prior to attaining the whole-cell configuration) resulted, on average, in about 65% inhibition of I# K ; however, the same treatment with U73343 inhibited I# K by only about 20% (Fig. 10A). U73122 also decreased I# O , while U73343 did not affect it (Fig. 10B).
Boltzmann analysis of the G# K -E M relationship under these treatments in steady-state conditions revealed a roughly 65% inhibition of G# max by U73122 (Fig. 10, C and D) and, notably, also a shift of the activation curve in the depolarizing (rightward) direction (Fig. 10, D and E). Two molecular mechanisms can explain the diminished NtORK activity accompanying the increased levels of PtdInsP 2 in the High PIs. A major effect can be attributed to PtdInsP 2 inhibition of the maximum chord conductance, G# max (which is G# K at its satura- ; and of the number of NtORK channels in the plasma membrane (actually, in a unit area of it) attainable for voltage activation, N#. Thus, PtdInsP 2 could decrease N#, or P max , or g S , or any combination of them. Which of the three factors is the more likely target for PtdInsP 2 effects? g S In different channels, these effects seem to vary. Binding of an anionic lipid palmitoyloleoylphosphatidylglycerol to the two-transmembrane-domain bacterial K channel, KcsA, in a lipid bilayer increased g S (Marius et al., 2008). But PtdInsP 2 did not seem to affect g S , neither while inhibiting the animal olfactory cyclic nucleotide-gated channels (Zhainazarov et al., 2004) nor while activating the two-transmembranedomain KIR channels (inward-rectifying K channels; Yang et al., 2000), or the Shaker-like KCNQ channels (voltage-gated KQT-like K channels) in mammalian cells , or the plant Shaker-like SKOR (K + -release) channel in frog oocytes (Liu et al., 2005).

P max
Palmitoyloleoylphosphatidylglycerol also increased the open probability of KcsA channels (Marius et al., 2008), and increased open probability was also a hallmark of PtdInsP 2 activation of several other different animal channels. For example, dissociation of PtdInsP 2 from the epithelial sodium channel, ENaC, decreased its open probability without affecting its plasma membrane abundance (Mace et al., 2008; for review, see Suh and Hille, 2008, and references therein). PtdInsP 2 also increased the open probability of the plant SKOR channels in excised oocyte membrane patches (Liu et al., 2005). A word of caution: a change of open probability at a single membrane potential can actually mean a shift of the voltage activation curve rather than a change in the P max ; we have not found this distinction in many of the published reports.

N#
Regulation of the activity of some ion channels, such as TRPV5 and ENaC channels in animal cells, has been  shown to involve recycling between the plasma membrane and endomembranes (Lambers et al., 2007;Mace et al., 2008). For example, treatments causing the dissociation of PtdInsP 2 from the ENaC subunits initiated the retrieval of ENaC protein from the plasma membrane (Mace et al., 2008). In tobacco plants, ABA treatment, known to enhance the metabolism of PtdInsP 2 , caused the endocytosis of the K + -influx channel, KAT1-GFP, in epidermal and guard cells within 10 to 20 min. This effect was selective, since the H + -ATPase (PMA2-GFP) in the same cells was not affected even after a 120-min exposure (Sutter et al., 2007). Endocytosis and membrane recycling have already been linked with PtdInsP 2 , for example in the growth of a pollen tube (Dowd et al., 2006;Helling et al., 2006). This complex link has been clarified recently in yeast. There, sites of endocytosis needed to be enriched in PtdInsP 2 , but for endocytosis to be completed PtdInsP 2 had to be removed (Sun et al., 2007). This is consistent with enhanced turnover of PtdInsP 2 being capable of promoting endocytotic internalization of ion channels. Based on this, we may expect enhanced endocytosis in the plasma membrane of the High PIs, where PtdInsP 2 is metabolized at an enhanced rate (Im et al., 2007), and consequently NtORK removal from the plasma membrane is reflected in a decreased N#. Thus, we deem it most likely that a combined decrease in P max . N# underlies the inhibitory effect of PtdInsP 2 on NtORK G# max .

G# K -E M Shift
The other inhibitory effect of the high PtdInsP 2 levels in the High PIs consists of shifting the voltage dependence of NtORK gating (the P O -E M relationship, which is the G# K -E M relationship normalized to G# max ) by over 25 mV to more positive values. This means that it is more difficult to activate the channel in the High PIs, since in order to activate the same fraction of channels as in controls the High PIs require an additional depolarization of 25 mV (Fig. 3C). Such a shift has been observed also in animal cells under similar conditions of increased PtdInsP 2 levels (Yaradanakul et al., 2007; for review, see Suh and Hille, 2008). The direction of this shift is consistent with the NtORK channel protein sensing an increased density of negative surface charges at the internal surface of the plasma membrane, which is consistent, in turn, with the enrichment of the internal leaflet with PtdInsP 2 .
The general importance of electrostatic interactions between the membrane and proteins (Mulgrew-Nesbitt et al., 2006) and, in particular, the importance of the negative surface charges of PtdInsP 2 for channel activity has been highlighted elegantly (in the case of the animal KCNQ potassium channel) by increasing the cytosolic concentration of Mg 2+ and applying a series of organic polycations with increasing valency. This, in quantitative agreement with neutralization of the internal negative charges, diminished channel activity (Suh and Hille, 2007;Lundbaek, 2008; see discussion by Suh and Hille, 2008). Interestingly, quantitative modeling of PtdInsP 2 -scavenging polycations predicted also a lack of the screening effect of polycations  when PtdInsP 2 concentrations in the membrane are increased (Suh and Hille, 2008). Such may be the explanation of a reported insensitivity of a plant K + -release channel to cytosolic polycations in V. faba guard cells (Liu et al., 2000). The concurrent inhibition by polycations of a K + -influx channel in these guard cells, in a whole-cell configuration but not in excised patches, was interpreted as due to a different specific mechanism rather than to PtdInsP 2 charge screening (Liu et al., 2000).
In contrast to the G# K -E M shift, the sensitivity to voltage of the gating process (i.e. the slope of the P O -E M relationship, embodied in the parameter z) was not affected by PtdInsP 2 levels. This is consistent with unchanged properties of the channel-gating subunits themselves.

Two Mechanisms of Enhancing NtORK Activity in High PIs by ABA Degradation of PtdInsP 2
Applying ABA as a means to hydrolyze PtdInsP 2 through PLC activation according to the V. faba guard cell paradigm indeed resulted in considerable lowering of the PtdInsP 2 level in the High PIs (Fig. 3). Alteration of NtORK activity had the same underlying biophysical components as did genetic manipulation: lowering of the PtdInsP 2 level increased maximum attainable NtORK conductance (per unit area), G# max , i.e. increased P max . N# . g S , or, perhaps, as suggested earlier, P max . N#. Could ABA increase N# by transcriptional activation of NtORK? It is quite unlikely, since in intact Arabidopsis suspension cells a short-term ABA treatment induced GORK mRNA, which was noticeable only after about 4 h (Becker et al., 2003). Rather, an increase of N# could occur through removal (by PLC-mediated hydrolysis) of the PtdInsP 2 from a direct inhibitory interaction with channels already in the membrane, or by recruitment of channels to the membrane through tipping the balance between endocytosis and exocytosis toward the latter when the PtdInsP 2 levels declined.
Additionally, the ABA-induced decrease in the PtdInsP 2 level led to a reversal of the voltage shift of the High PIs channel gating back to the control range, in line with the presumably diminished negative charge density at the internal plasma membrane surface. Resembling our results, in patch-clamp experiments conducted on protoplasts of V. faba guard cells, ABA not only increased the outward K + currents but, under certain conditions, also caused a similar, hyperpolarizing shift of the activation curve of the K + -release channels (Lemtiri-Chlieh, 1996). These specific conditions were cytosolic K + concentrations below 150 mM, which we interpret as cytosolic solutions of ionic strength low enough to not mask the negative surface charges on the inside of the plasma membrane.
In the controls, the minute ABA-induced changes in PtdInsP 2 levels (even if significant in the case of the wild type) did not affect the activity of NtORK. This effective lack of modification of the already low PtdInsP 2 level can be attributed to a simultaneous ABA activation of phospholipase D (PLD; Ritchie and Gilroy, 1998;Jacob et al., 1999;Hallouin et al., 2002;Zhang et al., 2004). Phosphatidic acid, released from the abundant phosphatidylcholine by PLD, is expected to activate the endogenous PtdIns (4) Figure 7. Data of both high-PI cell lines were pooled together for clarity. Lines were calculated using Equation 2 and the averaged best-fit Boltzmann parameters (G# max , E 1/2 , z; shown in B) obtained from analyses of individual G# K -E M s. B, Mean 6 SE best-fit Boltzmann parameters. The parameter values differ significantly if they are denoted by different letters (a or b). Numbers inside columns are the numbers of cells contributing to the averages. C, Symbols show mean 6 SE P O -E M relationships. Data of both high-PI cell lines were pooled together for clarity. Lines are as in A, except for using Equation 3 here. Vertical dashed lines and arrows denote E 1/2 values at pH 5.6 (right) and at pH 7 (left). Note the pH neutralization-induced hyperpolarizing shift of P O -E M of the High PIs. Wang et al., 2006) and thereby enhance the production of PtdInsP 2 . PtdInsP 2 promotes the PLD activity, in turn producing more PtdInsP 2 in a positive feedback (Huang, 2007). We are tempted to speculate that it is quite likely that in the control NT1 cell lines the ABAstimulated opposing effects of the two phospholipases, PLC and PLD, are sufficiently well balanced to keep the PtdInsP 2 level practically constant (as demonstrated in Fig. 3) and, consequently, the NtORK activity unchanged.
The lack of short-term effects of ABA on GORK channel activity in Arabidopsis guard cells (Wang et al., 2001) resembles our results with the control cell lines, but in Arabidopsis it could be due to technical reasons (the application of ABA after attaining a whole-cell configuration, i.e. on the background of altered cytoplasmic milieu and disrupted signaling cascade). In contrast, ABA enhanced within minutes K + -release channels (identified as GORK; Becker et al., 2003) in Arabidopsis suspension cells (Jeannette et al., 1999). In this case, the cells were assayed with an impaling electrode, which presumably minimized the perturbation of the internal milieu. These results resemble our own results with the High PIs, and it would be interesting to assay the effect of ABA on the PI levels in the membranes of the Arabidopsis suspension cells.

Two Mechanisms of Enhancing NtORK Activity in High PIs by Degradation of PtdInsP 2 through the Removal of External Protons
In the experiments presented here, in which protoplasts were incubated at pH 7 for several minutes prior to attaining whole-cell configuration (i.e. before introducing the strongly pH-buffered patch-pipette solution to the cytosolic milieu), it is very reasonable to assume that external alkalinization of an intact proto- Figure 10. Steady-state effect of U73122 on NtORK conductance. Protoplasts of the control line C-5 (see "Materials and Methods") were exposed to U73122 or U73343 in the bath (2.5 mM) starting about 3 min prior to attaining the whole-cell configuration. A depolarizing pulse of +85 mV was applied repeatedly every 45 s from a holding potential of  plast led immediately to a partial, even if only transient, cytosolic alkalinization (Heppner et al., 2002;Boron, 2004). Thus, the effects of external alkalinization on NtORK could be mediated by cytosolic alkalinization.
In patch-clamp experiments on excised guard cell plasma membrane patches, cytosolic alkalinization turned out to increase the number of K + -release channels available for activation. Moreover, this process was membrane delimited (i.e. independent of soluble cytosolic components, and, in particular, independent of cytosolic [Ca 2+ ], at least at 50 nM and 1 mM; Miedema and Assmann, 1996). This seems to suggest direct pH sensing located on the interior of the membrane. Alternatively, intracellular alkalinization, which is likely to deprotonate Ca 2+ -binding sites, could increase their availability for Ca 2+ binding, activating PLC even at a Ca 2+ concentration as low as 50 nM (Hunt et al., 2004).
Interestingly, external alkalinization enhanced the activity of TRPV5 channels in animal cells, with patchclamp experiments excluding pH effects through the cell interior. Excluding also a direct effect on the channel protein, these experiments suggested the mediation of an external pH sensor other than the channel itself (Lambers et al., 2007).
Why were control NT1 cells unaffected by the higher pH? The minute changes in the originally low levels of membrane PtdInsP 2 in the controls could be due to a balance in the activities of PLD and PLC in these cells, as argued earlier for ABA stimulation. Consequently, this could result in unchanged NtORK activity. Future experiments should address this hypothesis.

Two Mechanisms of Inhibiting NtORK Activity in High PIs and Controls by Accumulation of PtdInsP 2 through PLC Inhibition
Targeting the PLC for inhibition by U73122 was expected to shift the dynamic balance between PtdInsP 2 production and its cleavage (Staxen et al., 1999;Perera et al., 2001;De Jong et al., 2004;Parre et al., 2007) in both the High PIs and the control cells, wild type and C5, leading to PtdInsP 2 accumulation and, consequently, to NtORK inhibition. Indeed, the inhibition of NtORK by U73122 included two elements, resembling the effect of the genetically elevated PtdInsP 2 levels: depression of G# max (likely via decreasing P max . N#) and a shift of the voltage activation range to depolarizing potentials, consistent with PtdInsP 2 accumulation in the internal leaflet of the membrane.

Physiological Significance
Taken together, our data strongly suggest that PtdInsP 2 inhibits NtORK. PtdInsP 2 also inhibited an anion channel in guard cells of Arabidopsis and V. faba (Lee et al., 2007). While preventing stomatal closure can be achieved, in principle, by inhibiting only one of these two ion-releasing channels, no osmotically significant loss of ions (K + and Cl 2 ) or stomatal closure would occur without both channels operating simultaneously. We propose that lowering of PtdInsP 2 levels in guard cell plasma membrane is what activates both channels or, at least, predisposes them for activation, eventually leading to stomatal closure. The physiological initiating signal may be ABA. This type of signaling may mediate ABA-induced stomatal closure even in the absence of an observable rise in the concentration of cytosolic Ca 2+ (Levchenko et al., 2005;Marten et al., 2007). PtdInsP 2 hydrolysis might not elevate cytosolic Ca 2+ , for example, if the resulting InsP 3 fails to convert to InsP 6 , a proposed physiological agent of Ca 2+ rise in guard cells (Lemtiri-Chlieh et al., 2000, 2003. Lowering of PtdInsP 2 levels may also be the underlying mechanism of the induction of stomatal closure by intracellular or extracellular alkalinization. Notably, transgenic Arabidopsis plants with lowered PtdInsP 2 and InsP 3 (expressing the same gene of human 5-phosphatase as in the tobacco I2-8 and I4-2 Low PIs) were more resistant to drought, losing less water through their leaves and exhibiting more efficient stomata closure (Perera et al., 2008), as would be expected from the hypothesized relief of PtdInsP 2 inhibition of the ion-release channels in their guard cells.

Inverse or Direct
Interestingly, a reversible rundown of channel activity in the absence of cytosolic components required for membrane-delimited phosphorylation, hydrolyzable ATP and Mg 2+ , was documented for the SKORrelated K + -release channel of S. saman (presumed to be SPORK; Moran, 1996). This resembles the rundown of SKOR activity in the oocytes (Liu et al., 2005). Such rundown has been highlighted as symptomatic of a PtdInsP 2 requirement for channel activity (Suh and Hille, 2008). If, indeed, SPORK requires elevated PtdInsP 2 for its activity, the current models of motor cell signaling in moving leaves (Kim et al., 1996;Moran, 2007aMoran, , 2007b will have to be altered. We posit that NtORK behavior represents the physiological relationship between K + release channel activity and PtdInsP 2 in all plant motor cells. The fact that the PtdInsP 2 level in the High PIs was elevated constitutively could not be the reason for depressed NtORK activity in these cells, since even a short-term PtdInsP 2 accumulation caused by U73122 had the same inhibitory effect on the channel. The different responses of SKOR (Liu et al., 2005) and NtORK to increased PtdInsP 2 level (activation versus inhibition, respectively) may be due, at least partially, to the following reasons: (1) heterologous versus homologous expression of the channels: different interactions with the surrounding milieu (types of membrane lipids, enzymes, cystoskeleton, other proteins) or different protein modifications of the channels themselves; (2) different mode of lipid application: exogenous versus in vivo; (3) different channel sequences: in spite of close homology, there could be differences, in particular at the termini and the cytosolic loops, dictating different three-dimensional structures, leading to different interactions with PtdInsP 2 or with an intervening protein (Suh and Hille, 2008).
The last possibility seems the most likely. For example, SKOR, the Arabidopsis xylem-loading channel (Gaymard et al., 1998), may indeed be stimulated by elevating PtdInsP 2 even in planta. Thus, a K + -release channel of maize (Zea mays) root stele, and therefore, a likely SKOR ortholog, was inhibited by pretreating the plants with ABA (Roberts, 1998), as was K + xylem loading in barley (Hordeum vulgare; Cram and Pitman, 1972). This was quite opposite from the stimulating effect of ABA on the (unresolved) K + -release channels in V. faba guard cells (Lemtiri-Chlieh and MacRobbie, 1994;Lemtiri-Chlieh, 1996) or on GORK in Arabidopsis suspension cells (Jeannette et al., 1999).
Is NtORK like SKOR or like GORK? Based on comparison of the amino acid sequences, NtORK1 (the sole gene of the K + -release channel in the tobacco suspension cells; Sano et al., 2008) is more similar to SKOR than to GORK (72% versus 66% identical; Sano et al., 2008). However, because NtORK1's dominant distribution in tobacco plant leaves (Sano et al., 2008) resembles that of GORK in Arabidopsis (Becker et al., 2003), NtORK is likely a GORK ortholog (Sano et al., 2008). Our present results support this view.

CONCLUSION
The previously observed correlations between the phospholipid level and GORK (and, very likely, SPORK) activity in plant motor cells were extended here into a causal relationship between the PtdInsP 2 level and the activity of a homologous channel, NtORK. This was achieved by manipulating the endogenous levels of PtdInsP 2 in the plant cell membrane, on a long-or a short-term scale, while monitoring the in situ NtORK activity, as prescribed by a recent critical review: "The more ways you can change the PIP 2 [PtdInsP 2 ], the stronger the evidence becomes" (Suh and Hille, 2008). Our findings are consistent with NtORK inhibition by the negatively charged PtdInsP 2 in the internal plasma membrane leaflet. Biophysical analysis of NtORK whole-cell outward K + currents provided an insight into a possible mechanism underlying this causality. Thus, NtORK activity was diminished mainly through decreased maximum available conductance via the channels (G max ), and, to a somewhat lesser extent, by altering the voltage dependence of channel activation and making the channels more difficult to open. Such effects are likely to underlie PI signaling in intact plant cells.
We are aware that our observations are only the beginning of a prolonged exploration of the possible interactions of PIs with plant ion channels and with other proteins (Suh and Hille, 2008). Although PIs are much less abundant in the plant plasma membrane than in the animal plasma membrane, the wealth of reports on plant PI signaling suggests that channel-PtdInsP 2 interactions in the plant cell will not be much rarer. Because ion channels are crucial for signaling and osmotic homeostasis, understanding these interactions will provide a handle for manipulating plant water relations.

Cell Lines
One wild-type and eight transgenic lines of tobacco (Nicotiana tabacum) cultured cells, NT1, were used in this work: two High PIs, cell lines expressing the GFP-fused human phosphatidylinositol (4)phosphate 5-kinase, type Ia, HsPIPKIa-2 and HsPIPKIa-3, and a control GFP line, transformed with the same vector but with GFP alone (Im et al., 2007); two Low PIs, cell lines expressing the human-type InsP 5-phosphatase, I2-8 (Perera et al., 2002) and I4-2, both with membrane-associated phosphatase; three controls for the Low PIs: C-5, an empty vector control, C348S, an inactive mutant phosphatase, and DC, a truncated soluble phosphatase localized in the cytosol. The soluble form of the InsP 5-phosphatase DC lacks the C-terminal isoprenylation site (the last four amino acids; De Smedt et al., 1997). The inactive mutant C348S has a single amino acid substitution (Cys-348 to Ser) in the catalytic domain (Communi and Erneux, 1996). The three InsP 5-phosphatase constructs (I4-2, C348S, and DC) were subcloned into the Gateway binary vector pK2GW7 containing a cauliflower mosaic virus 35S promoter (Functional Genomics Division, Department of Plant Systems Biology, University of Gent, Belgium), and the DNA sequence was verified by sequencing. Tobacco cell transformation and selection of transgenic tobacco lines were as described previously (Perera et al., 2002).

Tissue Culture
Tobacco ('Bright Yellow 2') cultured cells (NT1) were maintained in 16 mL of liquid culture medium (see "Solutions" below). The 100-mL Erlenmeyer flasks with cells were agitated at about 125 rpm in the dark at about 28°C. Cells were subcultured every 7 d at a 1:16 (v/v) dilution with fresh medium at room temperature (Perera et al., 2002).

Protoplast Isolation
Protoplasts were isolated from the tobacco cells at 4 d (4 3 24 h) after subculture. A 0.5-mL suspension was centrifuged at 36g for 5 min, and the upper medium was poured off and quickly replaced with 2.5 mL of cell wall digestion solution (see "Solutions" below). Subsequently, the cells were agitated on a rotary shaker at about 60 rpm for 1.5 h at 30°C. The enzymatic reaction was stopped by washing the cells with 5 mL of isotonic solution (see below) through a nylon filter (50-mm mesh) into a 12-to 15-mL test tube. The test tube was centrifuged at 36g for 5 min. The supernatant was discarded, and the protoplasts were resuspended in 300 mL of isotonic solution in the same test tube. The isolation procedure was conducted at room temperature. To prevent regrowth of cell walls and reproduction of bacteria, the protoplasts were placed on ice for the duration of the experiment (up to 9 h).

Procedure
Patch-clamp experiments were performed at room temperature (20°C-22°C) using a Digidata 3122A interface and the pClamp 9 or 10 program suite from Axon Instruments, which was used both for running the experiment and analysis. Fire-polished borosilicate glass patch-clamp pipettes (Sutter Instru-ments; catalog no. BF150-86-10) had resistance of 50 to 100 M in the external solution (see "Solutions" below). A drop with protoplasts was added to the external solution in the bath, and the protoplasts were allowed to settle for several minutes or the bath was filled with the external solution after the drop with protoplasts was placed on the chamber bottom (the order did not influence the results). Whole-cell configuration was attained usually spontaneously without obtaining a giga-seal. Recording started several minutes after decrease and stabilization of the baseline current at the holding potential of 270 mV, usually within about 10 min. Outward K + currents were recorded from the protoplasts applying 1-to 3-s-long voltage pulses between +85 and 2110 mV in 215-mV voltage steps at 45-s intervals. All experiments were performed in voltage-clamp mode. The error in the voltage clamping of the whole-cell membrane, largely due to the series resistance of the patch pipette, was compensated at approximately 80% by analog circuitry of the amplifier. Mean series resistance was 15.8 6 5.4 M (6SD, n = 106). Currents were filtered at 500 Hz and sampled usually at 1 to 2 kHz.

Analysis
Prior to averaging, a normalized net steady-state current, I# K , was obtained by subtracting the instantaneous current from the total steady-state current, then dividing by the cell's capacitance, as read off the amplifier dial. The normalized instantaneous current, I# O , was obtained similarly. Mean capacitance was 32.9 6 9.4 pF (n = 106). The mean diameter of protoplasts selected for these experiments was 45.4 6 6.3 mm (n = 106). The average specific capacitance was 0.51 6 0.10 mF cm 22 (n = 106).
The normalized chord conductance, G# K , was extracted from I# K according to Equation 1: where E M is the membrane potential and E rev is the current reversal potential, determined separately for each cell line (as described in Supplemental Fig. S2). Individual G# K -E M relationships were fitted with the Boltzmann equation (Hille, 2001) using Origin (version 7.0220; Origin Lab Co.), according to Equation 2: where G# max is the maximum conductance (normalized to capacitance), E 1/2 is the voltage at which half of G# max is attained, and z is the effective charge of the gating subunits. The individual best-fit parameter values (G# max , E 1/2 , and z) were then averaged for each cell line. Channel open probability, P O , was obtained by normalizing G# K to G# max , according to Equation 3: Other details were as published previously (Yu et al., 2001). Differences between means were deemed significant if, using a two-sided Student's t test, P , 0.05.
ABA (Sigma; catalog no. 862169) 10 mM stock solution was prepared by dissolving 13 mg of powder in approximately 100 mL of 1 M KOH and completing the volume with bath solution to 5 mL. After adjusting pH to 5.6 by MES, the stock solution was stored at 220°C. U73122 (Calbiochem; catalog no. 662035) and U73343 (Calbiochem; catalog no. 662041) 5 mM stock solutions were prepared in dimethyl sulfoxide (Fluka; catalog no. 41650).
K 4 -BAPTA was from Molecular Probes, and other chemicals were from Sigma or Merck and were of analytical grade.

PtdIns(4,5)P 2 Mass Measurements
Protoplasts were harvested by centrifugation, plasma membranes were isolated by aqueous two-phase partitioning, lipids were extracted, and PtdIns (4,5)P 2 mass measurements were carried out as described .

InsP 3 Determination
Either the membrane lipid hydrolysate (as above) was assayed, or, to measure total InsP 3 level, cells were harvested by filtration and immediately frozen in liquid N 2 , ground to a fine powder, and precipitated with cold 10% (v/v) perchloric acid. InsP 3 assays were carried out using the TRK1000 InsP 3 assay kit (GE Healthcare Life Sciences) as described previously (Perera et al., 1999(Perera et al., , 2002.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Identification of K channel currents using K channel blockers.
Supplemental Figure S2. Determination of the reversal potential, E rev .
Supplemental Figure S3. The effect of PtdInsP 2 on whole-cell currents.
Supplemental Figure S4. Subcellular localization of the 5-phosphatase in the Low PIs.
Supplemental Figure S5. InsP 3 content in the InsP 5-phosphataseharboring transgenic cells and their controls.