Previously, we demonstrated that a protein that binds phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P2] inhibits both light-induced stomatal opening and ABA-induced stomatal closing. The latter effect is due to a reduction in free PtdIns(4,5)P2, 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)P2 modulates stomatal opening. We found that in response to white light irradiation, the PtdIns(4,5)P2-binding domain GFP:PLCδ1PH translocated from the cytosol into the plasma membrane. This suggests that the level of PtdIns(4,5)P2 increases at the plasma membrane upon illumination. Exogenously administered PtdIns(4,5)P2 substituted for light stimuli, inducing stomatal opening and swelling of guard cell protoplasts. To identify PtdIns(4,5)P2 targets we performed patch-clamp experiments, and found that anion channel activity was inhibited by PtdIns(4,5)P2. Genetic analyses using an Arabidopsis PIP5K4 mutant further supported the role of PtdIns(4,5)P2 in stomatal opening. The reduced stomatal opening movements exhibited by a mutant of Arabidopsis PIP5K4 (At3g56960) was countered by exogenous application of PtdIns(4,5)P2. 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)P2 in irradiated guard cells, inhibiting anion channels to allow full stomatal opening.
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 CO2 concentration and ABA. ABA-induced stomatal closure (Hetherington, 2001;Schroeder et al., 2001; Fan et al., 2004) involves changes in reactive oxygen species (Pei et al., 2000; Zhang et al., 2001), phosphatidylinositol 3-kinase activity (Park et al., 2003), calcium oscillations (McAinsh et al., 1990; Allen et al., 2000) and actin organization (Eun and Lee, 1997). Phospholipases C (PLC) and D (PLD) participate in the ABA-induced stomatal closure response by producing the calcium-mobilizing secondary messenger inositol 1,4,5-trisphosphate [Ins(1,4,5)P3; Hunt et al., 2003] 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., 2004; Mishra et al., 2006). 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., 2001). 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 14-3-3 protein and activation of the proton pump (Emi et al., 2001; Kinoshita and Shimazaki, 2002). 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, 2005). 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 Ca2+ levels, releasing Cl– and other anions (Hedrich et al., 1990; Schroeder and Keller, 1992). Together with the outward-rectifying 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 −200 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)P2] is an important signal molecule that is involved in various processes such as pollen tube growth (Kost et al., 1999; Monteiro et al., 2005), salt and osmotic stress (DeWald et al., 2001), vesicle trafficking (Martin, 2001), actin organization (Janmey, 1994; Caroni, 2001), modulation of the plasma membrane vanadate-sensitive H+-ATPase (Memon and Boss, 1990), ion channel activity (Hilgemann et al., 2001; Liu et al., 2005) and guard cell movements (Jung et al., 2002). Guard cells have been shown to contain PtdIns(4,5)P2 (Parmar and Brearley, 1993) and in Vicia faba guard cells, PtdIns(4,5)P2 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-[((17β)-3-methoxyestra-1,3,5-trien-17-yl)amino]hexyl]-1H-pyrrole-2,5-dione (U-73122) inhibited ABA-induced calcium oscillations in guard cells and stomatal closure, providing supporting evidence for the importance of PtdIns(4,5)P2 hydrolysis by PLC in the ABA-induced stomatal closure process (Staxén et al., 1999). In addition, PtdIns(4,5)P2 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)P2 also appears to be involved in stomatal opening. This was demonstrated using the PtdIns(4,5)P2-binding protein GFP:PLCδ1PH, which inhibited not only ABA-induced stomatal closure, but also light-induced stomatal opening when expressed in guard cells (Jung et al., 2002).
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)P2 from either PtdIns(4)P or PtdIns(5)P (Mueller-Roeber and Pical, 2002), this activity has only been confirmed for PIP5K1 and PIP5K10 (Mikami et al., 1998; Perera et al., 2005). 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., 2001). 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., 2001). The PIP kinase PIP5K10 belongs to the A subfamily, which lacks MORN repeats, and is most abundant in inflorescence stalks and flowers; its Vmax is 10-fold lower than PIP5K1 (Perera et al., 2005). Although the presence and absence of MORN repeats suggests membrane and non-membrane localizations for PIP5K1 and PIP5K10, respectively, their cellular localizations and physiological functions remain undetermined.
In this paper we confirm that PtdIns(4,5)P2 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)P2in vitro. We present a number of lines of evidence that support a role for this gene in light-induced stomatal opening.
PtdIns(4,5)P2 binding domain GFP:PLCδ1PH translocates to the plasma membrane in response to white light irradiation
GFP:PLCδ1PH (phospholipase Cδ1 pleckstrin homology domain) binds PtdIns(4,5)P2 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:PLCδ1PH in guard cells inhibited light-induced stomatal opening, probably by interfering with the normal interactions between PtdIns(4,5)P2 and other molecules (Jung et al., 2002). Therefore, this result suggests that PtdIns(4,5)P2 is important for light-induced stomatal opening. To test whether illumination leads to increased PtdIns(4,5)P2 content, we overexpressed GFP:PLCδ1PH (Figure 1a) in V. faba guard cells and observed the localization of fluorescence before and after 3 h of irradiation with 170 μmol m−2 sec−1 white light (Figure 1b). Translocation was quantified by measuring the green fluorescence intensity of GFP from microscopic 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 = 1.09 ± 0.02%, P >0.05; 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 ± 0.03%, P <0.001), indicating translocation of GFP:PLCδ1PH from the cytosol to the plasma membrane. Although GFP:PLCδ1PH can bind Ins(1,4,5)P3 as well, it is unlikely that the increase in the fluorescence ratio was caused by a decrease in the Ins(1,4,5)P3 level in the cytosol, as GFP:PLCδ1PH 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 ± 0.02%, P <0.05) 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 <0.001).
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., 2004) as a negative control for translocation (Figure 1b,d). Initially, the fluorescence intensity of GFP:MT2a at the cell boundary was 1.06 ± 0.03% of that in the cytosol (P >0.1). However, after 3 h of irradiation with white light, this had increased to 1.15 ± 0.02% (P <0.05), relative to the cytosol. This value was similar to that observed for GFP:PLCδ1PH after 3 h in the dark (P >0.1), but different from that following 3 h of irradiation (P <0.001, Figure 1d). Therefore, we conclude that light induces translocation of GFP:PLCδ1PH from the cytosol to the plasma membrane independently of the circadian clock. The translocation of GFP:PLCδ1PH 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:PLCδ1PH to this membrane suggests a function for PtdIns(4,5)P2 in the cellular light signaling process.
Stomatal opening is induced by PtdIns(4,5)P2
The results described above suggest that PtdIns(4,5)P2 is a factor that mediates stomatal opening. Therefore, we tested whether or not application of exogenous PtdIns(4,5)P2 can induce stomatal opening. Vicia faba guard cells were incubated in a medium containing PtdIns(4,5)P2 mixed with shuttle carriers (Ozaki et al., 2000) that assist in intracellular delivery of PtdIns(4,5)P2, after which their stomatal apertures were measured. Under darkness, treatment of epidermal tissues with 10 μm PtdIns(4,5)P2 significantly enhanced circadian clock-dependent stomatal opening (P < 0.01). In contrast, when PtdIns(4,5)P2 was replaced by PtdIns(4)P, no significant difference could be observed between the experimental and control stomata (P >0.1, Figure 2a). The specificity of PtdIns(4,5)P2-induced stomatal movement was further tested using other phosphoinositides, including PtdIns(3)P, PtdIns(5)P, PtdIns(3,4)P2 and PtdIns(3,5)P2. Only PtdIns(3,4)P2 slightly increased the stomatal aperture. None of the other lipids tested showed a significant effect (P >0.1, Figure 2a,b). The effect of PtdIns(4,5)P2 on stomatal opening was concentration dependent between 1 and 30 μm (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)P2 (P <0.01, data not shown). We speculated that if exogenously applied PtdIns(4,5)P2 induced stomatal opening by increasing PtdIns(4,5)P2 levels at the plasma membrane, then it should also have induced translocation of GFP:PLCδ1PH to the plasma membrane. Indeed, a significant increase in GFP:PLCδ1PH fluorescence at the cell boundary was observed at 60 min after application of PtdIns(4,5)P2 (P < 0.01, Figure 2d and Supplementary Figure S1a; n = 13), whereas no such translocation was observed after application of PtdIns(4)P (P >0.1, Figure 2d and Supplementary Figure S1b; n = 9).
As PtdIns(4,5)P2 is cleaved by PLC, it is possible that PLC inhibition may represent a mechanism for increasing PtdIns(4,5)P2 levels, and consequently stomatal opening. This hypothesis was tested by investigating the effect of U-73122 (a specific inhibitor of PLC in guard cells, as reported by Staxén et al., 1999) on stomatal opening. The guard cells treated with U-73122 showed statistically significant increases in stomatal opening compared with the control (P < 0.001), whereas those treated with its inactive analog, 1-[6-[((17β)-3-methoxyestra-1,3,5-trien-17-yl) amino]hexyl]-2,5-pyrrolidinedione (U-73343), did not (P > 0.1, Figure 2e). After exposure to U-73122 the stomatal apertures reached the maximum after 2 h and remained in that state for 5 h (data not shown). This effect of U-73122 on stomatal opening can be attributed to increased levels of PtdIns(4,5)P2 at the plasma membrane, as evidenced by the translocation of GFP:PLCδ1PH fluorescence to the plasma membrane 60 min after U-73122 treatment (P < 0.01, Figure 2f and Supplementary Figure S1c; n = 17). Guard cells treated with inactive U-73343 did not show any noticeable translocation of GFP:PLCδ1PH fluorescence (P > 0.1, Figure 2f and Supplementary Figure S1d; n = 6).
To confirm the role played by PtdIns(4,5)P2 in stomatal opening, we tested whether PtdIns(4,5)P2 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 10 μm PtdIns(4,5)P2 or irradiated with white light for 20 min (P >0.05, Figure 2g). There was no significant change in the volume of protoplasts incubated in darkness without PtdIns(4,5)P2 or in the presence of PtdIns(4)P (P >0.05, Figure 2g). In addition, the volume of guard cell protoplasts treated with U-73122 increased more than that of the controls (P <0.01, Figure 2g). These results provide additional support for the suggestion that PtdIns(4,5)P2 can substitute for light in inducing stomatal opening.
Slow anion current is inhibited by PtdIns(4,5)P2
Stomatal opening requires the coordinated and balanced activities of many ion channels and transporters. To examine whether or not PtdIns(4,5)P2 induces stomatal opening via alteration of ion channel activities we performed whole-cell patch clamping of V. faba guard cell protoplasts and analyzed K+ and anion channel activities before and after application of PtdIns(4,5)P2. Inward (n = 14) and outward (n = 14) K+ channel activities were unaltered by 10 μm PtdIns(4,5)P2 (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 0.3 μm free Ca2+ and 200 μm guanosine 5′-triphosphate (GTP), which have been shown to enhance anion currents across the plasma membrane of guard cells (Hedrich et al., 1990). S-type anion currents were identified by their typical time dependence and sensitivity to 50 μm 5-nitro-2-(3-phenylpropylamino) benzoic acid (NPPB; Figure 3a). The current magnitude measured after a 10-min exposure to +30 mV (IT10) increased in comparison to initial currents (IT0), a result that was expected as depolarization activates anion channels (Figure 3b; Schroeder and Keller, 1992). Treatment with PtdIns(4,5)P2 inhibited this time-dependent increase in anion currents (Figure 3c). To quantify these effects and to test whether or not 10 μm PtdIns(4,5)P2 specifically inhibits the current, we compared the magnitude of steady-state anion currents at the end of 60 sec hyperpolarizing voltage steps applied before and after treatment with various lipids. The magnitude of current change relative to initial current (ΔI/IT0) (%) = [(IT10–IT0)/IT0] × 100 (relative current increase) was about 160 ± 56% in untreated control cells. Protoplasts treated with 10 μm PtdIns(4)P showed a magnitude of ΔI/IT0 similar to the time control (164 ± 50%). In contrast, anion currents from protoplasts treated with 10 μm PtdIns(4,5)P2 showed ΔI/IT0 of 40 ± 20%, significantly lower than the time control or PtdIns(4)P (P <0.05, Figure 3d). We tested the effects of phosphatidylinositol 3,4-bisphosphate (PtdIns(3,4)P2) and 1-palmitoyl-2-oleoyl-sn-glycerol (DAG) on anion current, as PtdIns(3,4)P2 has been reported in the guard cells of C. communis (Parmar and Brearley, 1993), and DAG is a product of PtdIns(4,5)P2 hydrolysis, as well as an inducer of stomatal opening (Lee and Assmann, 1991). Protoplasts treated with 10 μm PtdIns(3,4)P2 and DAG exhibited slightly reduced ΔI/IT0 values, but these effects were not statistically significant (PtdIns(3,4)P2, 92 ± 52%; DAG, 78 ± 20%). Therefore, we conclude that of the lipids tested, PtdIns(4,5)P2 was the most effective at inhibiting development of an anion current.
To test whether PtdIns(4,5)P2 is important for stomatal opening in vivo we used a genetic approach. The enzyme that produces PtdIns(4,5)P2 is PIP kinase (PI4P5K and PI5P4K), and 11 different PIP kinases have been identified in Arabidopsis (Mueller-Roeber and Pical, 2002). 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 (At3g56960); 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 = 0 h), the apertures of pip5k4 stomata did not differ significantly from WT (P >0.05), whereas after 3 h of illumination with 170 μmol m−2 sec−1 white light, the mean aperture size of pip5k4 stomata (2.78 ± 0.03 μm) was significantly smaller than that of WT (4.20 ± 0.03 μm, P <0.001).
If the reduced stomatal opening in pip5k4 was due to decreased production of PtdIns(4,5)P2, replenishment of PtdIns(4,5)P2 should enable recovery of normal movement. We tested this idea by treating epidermal strips of pip5k4 plants with exogenous PtdIns(4,5)P2. These strips were incubated in medium containing 10 μm PtdIns(4,5)P2 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)P2-treated pip5k4 (2.94 ± 0.05 μm) were similar to those of WT (3.02 ± 0.06 μm, P > 0.5), and significantly larger than those of pip5k4 without treatment (2.32 ± 0.06 μm, P <0.01; Figure 4d). The stomatal apertures of PtdIns(4)P- and PtdIns(3,4)P2-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)P2.
We tested whether U-73122 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 0.1 μm U-73122 or U-73343 under darkness. In WT plants, the guard cells treated with U-73122 showed statistically significant increases in stomatal opening compared with the control (P <0.001), whereas those treated with its inactive analog U-73343 did not (P > 0.1). 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)P2 caused by 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. (2006). The guard cell preparation showed little contamination with mesophyll cells (Figure 1a of Mori et al., 2006). 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)P2 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).
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, 2002). To test whether PIP5K4 exhibits PIP kinase activity, we purified the entire kinase protein or the catalytic domain of PIP5K4 without MORN repeats (Δ1–388) 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, 10 μg; catalytic domain, 5 μg) 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)P2.
In this paper we provide evidence that PtdIns(4,5)P2 is an important signal mediator for stomatal opening, and we identify PIP5K4 as an enzyme that synthesizes PtdIns(4,5)P2. We demonstrate that the fluorescence intensity of a PtdIns(4,5)P2-binding peptide (GFP:PLCδ1PH; 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)P2 levels at the plasma membrane of guard cells. The increase in PtdIns(4,5)P2 levels could be due to a light-induced increase in synthesis and/or a decrease in hydrolysis of PtdIns(4,5)P2. Regardless of this, the light-dependent appearance of PtdIns(4,5)P2 at the plasma membrane is consistent with the suggestion that it plays a role in light signal transduction. Further support for this idea comes from the observation that exogenous application of PtdIns(4,5)P2 induced stomatal opening and swelling of guard cell protoplast in darkness (Figure 2). Phosphatidylinositol 4-phosphate, a metabolite and precursor of PtdIns(4,5)P2, is not responsible for the stomatal opening effect of PtdIns(4,5)P2, as shown in Figure 2. This result is consistent with the previous observation that PtdIns(4)P-binding protein, which presumably reduces the free PtdIns(4)P levels, exerts an effect opposite to that of PtdIns(4,5)P2-binding protein in stomatal opening movement (Jung et al., 2002). The result is also consistent with the recent observation that PtdIns(4,5)P2 synthesis, but not PtdIns(4)P synthesis, is the rate-limiting step in the plant phosphoinositides pathway (Im et al., 2007). In addition, anion channel activity was altered by PtdIns(4,5)P2, 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)P2, Ins(1,4,5)P3, is well known for its effect on stomatal closing (Blatt et al., 1990; Gilroy et al., 1990), which would appear to preclude an effect on stomatal opening. Moreover, U-73122, which inhibits the production of Ins(1,4,5)P3, showed effects similar to that of PtdIns(4,5)P2.
Possible targets for the action of PtdIns(4,5)P2 at the guard cell plasma membrane include ion pumps and channels. Recent studies in animals have shown that PtdIns(4,5)P2 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., 2002). In addition, PtdIns(4,5)P2 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, 2004). Among the many potential target transporters in guard cells, we tested anion channel activities, and observed that those were inhibited by PtdIns(4,5)P2 (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 −200 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 over-activation 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)P2 depending on its phosphorylation status: application of PtdIns(4,5)P2 to non-phosphorylated CFTR activates a chloride current, whereas phosphorylated CFTR is inhibited. In most cases, PtdIns(4,5)P2 regulates channel activity via direct binding. It would be interesting to elucidate the mechanism by which PtdIns(4,5)P2 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)P2 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)P2 on K+ channel activity. In contrast to our result, recently published data have demonstrated that PtdIns(4,5)P2 restores activity of shaker-type K+ channels run down following patch excision (Liu et al., 2005). Although these different results may be due to cell type, they are most likely a PtdIns(4,5)P2 concentration effect. We used 10 μm PtdIns(4,5)P2, whereas Liu et al. (2005) used concentrations up to 500 μm, which are unlikely to represent true physiological values. At relatively low concentrations of PtdIns(4,5)P2 (20 μm), 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)P2 indicator at the plasma membrane of irradiated cells is indirect evidence for de novo synthesis of PtdIns(4,5)P2 at this site. Recently, PtdIns(4,5)P2 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)P2 in WT tobacco cells was found to be ≥10:1, whereas in tobacco cells expressing human PIPKIα, a 100-fold increase in plasma membrane PtdIns(4,5)P2 was observed without any change in the PtdIns(4)P level (Im et al., 2007). To investigate possible changes in PtdIns(4,5)P2 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)P2 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 (At3g56960), exhibited a smaller stomatal aperture under light (Figure 4c), while pip5k3, which contains a mutation in PIP5K3 (At3g56960), 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)P2 (Figure 4d), and complementation using lines expressing PIP5K4 under its own promoter (Figure 5b). These results suggest that for normal stomatal opening sufficient PtdIns(4,5)P2 must be present in the plasma membrane and that PIP5K4 contributes to the synthesis of PtdIns(4,5)P2. 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 10 are present in guard cells, although the expression of no single gene predominates in this cell type (Leonhardt et al., 2004). Under darkness, the PtdIns(4,5)P2 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)P2 production and that other PIP5Ks produce PtdIns(4,5)P2 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., 2001), 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., 2004). Plants contain a unique subfamily of Rho GTPases, the Rop GTPases (for Rho-related proteins from plants; Li et al., 1998; Yang, 2002) that are most similar to the mammalian RAC GTPases. Rop GTPases play roles in guard cell signaling (Lemichez et al., 2001; B.W. Jeon, J.-V. 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)P2 was also found to be important (Lin and Yang, 1997; Kost et al., 1999; Fu et al., 2002). 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)P2 levels at the plasma membrane can be caused not only by increased synthesis but also by a reduction in PtdIns(4,5)P2 hydrolysis by PLC. The Arabidopsis genome contains nine putative phosphatidylinositol-specific phospholipase C (PI-PLC) isoforms (Mueller-Roeber and Pical, 2002; Hunt et al., 2004), 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, 2001). 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., 2003; Mills et al., 2004). At concentrations that also inhibit recombinant PI-PLC activity, the PLC inhibitor U-73122 inhibits stomatal guard cell responses to ABA and cytosolic Ca2+ 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., 2003; Mills et al., 2004). 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-73122. The specificity of U-73122 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 Ca2+ oscillation. We observed that guard cells treated with U-73122 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-73122 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)P2 is an important factor for light-induced stomatal opening and that PIP5K4 is at least partially responsible for the production of PtdIns(4,5)P2 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)P2 in guard cells, and how PtdIns(4,5)P2 regulates the anion channel activity.
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 PIPTM Carrier-1, histone H1 (Molecular Probes, http://probes.invitrogen.com. was used, and PtdIns(4,5)P2 delivery was confirmed using BODIPY tetramethylrhodamine-X C6-PtdIns(4,5)P2 (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)P2 (l-α-phosphatidylinositol 4,5-diphosphate), PtdIns(3,4)P2, PtdIns(3,5)P2, PtdIns(3)P, PtdIns(4)P, and PtdIns(5)P with dioctanoyl acyl chains were used (Sigma-Aldrich, http://www.sigmaaldrich.com/). For electrophysiological recordings, sodium salts of PtdIns(4)P and PtdIns(4,5)P2 were purchased from Sigma-Aldrich, and 1-palmitoyl-2-oleoyl-sn-glycerol (DAG) from Avanti Polar Lipids (http://www.avantilipids.com/), and the lipid solutions were sonicated immediately before treatment. U-73122 and U-73343 were purchased from Sigma-Aldrich and dissolved in dimethyl sulfoxide.
Measurement of stomatal apertures
Abaxial epidermal layers of V. faba or Arabidopsis leaves were first peeled and then incubated in a solution containing 10 mm KCl and 10 mm 2-(N-morpholine)-ethanesulfonic acid (MES)-KOH (pH 6.1) with or without drugs or phosphoinositides. In order to supply phosphoinositides, Shuttle PIPTM Carrier-1, histone H1 (Molecular Probes) and phosphoinositides were mixed immediately before the experiments, allowed to equilibrate for 5 min, and added to the buffer solution to generate a working concentration of phosphoinositides. A control solution was prepared from an equivalent amount of Shuttle Carrier-1 without phosphoinositides. In some Arabidopsis experiments that did not necessitate treatment with drugs or lipids, whole leaves were incubated in buffer solution, and then the epidermal layer was peeled off just before observation. The samples were incubated in darkness 0.5 h prior to the beginning of the photoperiod, observed at 1 h intervals with bright field microscopy, Axioskop 2 (Carl Zeiss, http://www.zeiss.com/) and photographed using a CCD camera, Axio Cam (Carl Zeiss). Aperture size was measured from photographs using the Interactive Measurement software package, axiovision 3.0.6 (Carl Zeiss). In all experiments treatment with light or lipids began immediately after the initial measurement.
Measurement of the volume of guard cell protoplasts
Guard cell protoplasts of V. faba were isolated following a procedure modified from Kruse et al. (1989). The youngest fully expanded leaves of 3- to 5-week-old plants were homogenized in a Waring blender 7010 (http://www.waringproducts.com) for 40–50 sec at 10000 g to remove epidermal and mesophyll cells. After washing with tap water, the epidermal fragments were collected on a 220 μm nylon mesh (Small Parts, http://www.smallparts.com/), then incubated in enzyme solution for 30 min at 21°C, with rotation at 100 rpm. The enzyme solution comprised 4.5 parts distilled water and 5.5 parts basic solution (0.5 mm CaCl2, 0.5 mm MgCl2, 10 μm KH2PO4, 5 mm K+-MES [pH 5.5] and 0.45 m d-mannitol) containing 1% (w/v) cellulysin (Calbiochem, http://www.calbiochem.com), 0.3% (w/v) BSA, 0.1% (w/v) polyvinylpyrrolidone (PVP) and 1 mm ascorbic acid. The partially digested epidermal fragments were then incubated in basic solution containing 1.5% (w/v) cellulase RS (Yakult Honsha, http://www.yakult.co.jp), 0.3% (w/v) BSA, 0.02% pectolyase Y-23 (Yakult Honsha) and 1 mm ascorbic acid (pH 5.5), at 21°C, with rotation at 60 rpm. After 40 to 50 min, protoplasts were collected and resuspended in a solution containing 0.35 m d-mannitol, 1 mm CaCl2, 10 mm KCl, 1 mm MES (pH 6.2) and 1 mm ascorbic acid.
To measure change in protoplast volume, the protoplasts were incubated in darkness for 30 min in media with or without lipids and then photographed using bright field microscopy (Axioskop 2, Carl Zeiss) with a CCD camera (Axio Cam, Carl Zeiss). Diameters of protoplasts were measured from photographs using the Interactive Measurement software package axiovision 3.0.6 (Carl Zeiss) and the volumes were calculated by the equation 4/3*π*r.
Patch electrodes were pulled from glass capillaries (Kimax-51, Kimble, http://www.kimble.com) using a two-stage puller (PP-83, Narishige, http://www.narishige.co.jp/) and filled with an intracellular solution comprising 150 mm tetraethylammonium chloride (TEA-Cl), 10 mm HEPES, 2 mm MgCl2, 4 mm MgATP, 200 μm Na2GTP, 6.7 mm EGTA, 3.35 mm CaCl2, and adjusted to pH 7.2 with 2-amino-2-(hydroxymethyl)-1,3-propanediol (TRIS)–HCl and to 400 mmol kg−1 with d-mannitol. The free Ca2+ concentration in the pipette solution was about 0.3 μm (Schroeder and Keller, 1992). The bath solution contained 40 mm CaCl2, 2 mm MgCl2, 10 mm MES (adjusted to pH 5.6 with TRIS-HCl and to 400 mmol kg−1 with d-mannitol). The pipette resistance was ≈10 MΩ and current data were obtained using an Axon Instruments Axopatch 200A amplifier (Axon Instruments, http://www.axon.com/). pclamp software (Axon Instruments) was used for voltage pulse stimulation, online data acquisition and data analyses. The current response of the protoplast in the whole-cell configuration was recorded after the membrane potential had been held at +30 mV for 3 min to activate the anion channels (Schroeder et al., 1993). Current amplitudes were compared for the same cell before and at 5 min following drug treatments. Only cells that maintained >1 GΩ whole seal resistance throughout the experiment were included in the analyses.
For application of lipid to the patch-clamped cells, the lipid was N2-dried, then added to the bath solution. The mixture was sonicated for 2.5–3 min immediately prior to application to the whole-cell patches and released from a micropipette positioned about 150 μm from the target cell. The inner diameter of the drug pipette was nearly double that of the protoplast and the solution flow rate was kept extremely low to reduce perturbation of the patched cell.
Biolistic gene bombardment into V. faba guard cells
Vectors expressing GFP and GFP:PLCδ1PH were introduced into V. faba guard cells using a bombardment technique (Particle Delivery System-1000/He; Bio-Rad, http://www.bio-rad.com/) described previously (Park et al., 2003). The bombarded leaves were then placed onto a Petri dish lined with wet filter paper and kept in the dark at 22 ± 2°C. Between 15 and 20 h after bombardment the leaves were transferred to a solution containing 10 mm KCl and 10 mm MES-KOH (pH 6.1), then either exposed to white light (0.2 mmol m−2 sec−1) or kept in the dark. Epidermal peels were removed from the leaves and observed using an Axioskop 2 fluorescence microscope (Carl Zeiss). For time-lapse recording of single cells, a confocal microscope, FLUOVIEW FV1000 (Olympus, http://www.olympus-global.com/) was used.
Assay for translocation of GFP:PLCδ1PH between the plasma membrane and cytoplasm
Vector expressing GFP:PLCδ1PH was introduced into guard cells by biolistic bombardment. Using the image edit tool of KS Lite version 2.0 (Kontron, http://www.kontron.com/), fluorescence images of guard cells were scanned along two lines drawn at right angles to the long axis of the cells, at about a 25% distance from both ends. The lines seldom crossed the nucleus, which is usually located at the center of guard cells. From the resulting intensity profiles, the average peak pixel intensities of the cell boundary and interior were obtained. The ratios of these two values were compared before and after irradiation.
Verification of AtPIP5K4 knockout plants and generation of complemented lines of pip5k4
A T-DNA insertion line of PIP5K4 was obtained from the Salk Institute Genomic Analysis Laboratory (http://signal.salk.edu/cgi-bin/tdnaexpress, stock number: SALK_001138). Homozygote pip5k4 plants were selected by PCR using the genomic DNA as the template using the T-DNA specific primer 5′-GCGTGGACCGCTTGCTGCAACT-3′ (LBb1 primer, see http://signal.salk.edu/tdna_FAQs.html) and the PIP5K4 specific primers 5′-GACGGGAGTCCTGAATGGGAT-3′ and 5′-GCAGCTACATATTTTTCATCTTGTC-3′. To confirm that the pip5k4 mutant does not generate a PIP5K4 transcript, RT-PCR was performed. Total RNA was extracted from whole leaves and RT-PCR was performed using primers 5′-GAAATGATGAGACTAGAGGCTGAAGGGTTC-3′ and 5′-GAGACTTGTTTCAATTATCCTCAGTGAAGAC-3′.
To generate complemented lines of pip5k4, pip5K4 plants were transformed with a pCAMBIA1302 (BIOS) vector containing a 1.3 kb fragment upstream of the 5′ of the PIP5K4 coding sequence fused to the PIP5 K4 cDNA. The cording sequence of PIP5K4 was amplified from cDNA generated from total RNA by PCR using primers containing SphI and PmlI restriction sites (5′-GCATGCATCAGCAAGGAAACAAAGCTGTGTTC-3′ and 5′-CACGTGTCAATTATCCTCAGTGAAGACCTTG-3′) and the 1.3 kb promoter region was amplified from genomic DNA using a forward primer containing XmaI (5′-CCCGGGTTTTCGATTCCAACGATGAGAACCAA-3′) and a reverse primer containing SphI (5′-GCATGCCTTCTTAAACTAATAAAACTTTTCTCTAAGATACC-3′).
PIP kinase activity assay
Escherichia coli strain Rosetta (DE3) was used for the expression of recombinant PIP5K4 proteins fused to GST. An overnight culture of Rosetta expressing GST-PIP5K4 was diluted 1:100 with fresh culture medium and grown at 37°C with shaking at 200 rpm until an OD600 of 0.6–0.8 was reached, at which point isopropyl-d-thiogalactoside (IPTG) was added to a final concentration of 30 μm. Cells were incubated for an additional 12 h at 16°C and collected by centrifugation and kept frozen at −70°C until required for the purification of the GST-PIP5K4 protein.
The kinase activity of GST-PIP5K4 was measured as described by Lee et al. (1996). Phosphorylation of PtdInsP was undertaken at room temperature for 10 min in a 50 μl mixture containing 50 mm HEPES-KOH (pH 7.4), 3 mm MgCl2, 10 mm 2-glycerophosphate, 2 mm dithiothreitol, 240 mm NaCl, 10 μg PI or PtdIns(4)P (Sigma-Aldrich) and 10 μCi of γ32P-ATP (Amersham-Pharmacia Biotech, http://www5.amershambiosciences.com/) and the purified GST-PIP5K4 fusion protein. For control experiments, purified GST was used in place of the fusion protein. For lipid extraction, 300 μl of chloroform:methanol:0.7 m HCL (8:4:3, v/v) was added to the sample and vortexed for 20 sec. Phase separation was facilitated by centrifugation at 1600 gfor 1 min in a tabletop centrifuge. The upper phase was removed and the lower chloroform phase was washed once more with a fresh upper phase. Divalent cations, which bind to PtdInsP and retard its mobility during TLC, were removed by vigorous mixing with 150 μl of chloroform and 150 μl of an aqueous solution containing 2 m KC1 and 2 mm EDTA. The aqueous phase was discarded, and the solvent was evaporated under a stream of N2 and dissolved in 30 μl of chloroform. Lipids were spotted onto silica gel 60 thin layer chromatography (TLC) plates (Merck, http://www.merck.com/) impregnated with 1% potassium oxalate and 2 mm EGTA, and separated using chloroform:methanol:4 n ammoniumhydroxide (90:70:20, v/v). Plates were autoradiographed using a phosphorimager, FLA-2000R (Fujifilm, http://www.fujifilm.com/). To assist identification of the PtdIns(4,5)P2 band, cold phospholipid standards were run in parallel lanes on the same TLC plate and visualized by exposing the plate to iodine vapor in a sealed tank.
RT-PCR analyses of guard cell and mesophyll cell protoplasts
We used the same total RNA preparations from highly pure guard cell and mesophyll cell protoplasts as those reported previously (Mori et al., 2006). Equal amounts of each cDNA were used for a 25 μl PCR reaction containing 400 nm primers, 1× reaction buffer, 400 μm each nucleotide (dNTP), 2.5 U Ex Taq polymerase (Takara, http://www.takara-bio.com/) and 1 μl each cDNA. The PCR mixtures were denatured at 94°C for 2.5 min, followed by 37 (PIP5K4) or 35 (UBQ, ubiquitin-conjugating enzyme E2, At5 g25760) cycles of amplification (94°C, 30 sec; 54°C for PIP5 K4 or 60°C for UBQ, for 30 sec; 72°C for 1 min 10 sec). Each PCR reaction was repeated twice. We used the following primers for the PCR reactions: PIPK-F, 5′-TAAAGTGCTTCTGAGGATGCTTGCAGC-3′; PIPK-R, 5′-GAAATCACGGAAACGTCTCGAGTACAG-3′; UBQ-F, 5′-TAGAGATGCAGGCATCAAGAGCGCGACT-3′; and UBQ-R, 5′-GCGGCGGAGGCGTGTATACATTTGTGCCA-3′.
We thank Dr Nava Moran for a critical reading of the manuscript. This work was supported by grants awarded to YL from the Crop Functional Genomics Center of Korea (grant no. CG1-1-23), to IH from Biogreen21(Korea), and to JMK from NSF (grant no. MCB-0614203).