Abscisic acid regulation of guard-cell K⁺ and anion channels in Gβ- and RGS-deficient Arabidopsis lines

Abstract

In mammals, basal currents through G protein-coupled inwardly rectifying K⁺ (GIRK) channels are repressed by Gα_i/oGDP, and the channels are activated by direct binding of free Gβγ subunits released upon stimulation of Gα_i/o-coupled receptors. However, essentially all information on G protein regulation of GIRK electrophysiology has been gained on the basis of coexpression studies in heterologous systems. A major advantage of the model organism, Arabidopsis thaliana, is the ease with which knockout mutants can be obtained. We evaluated plants harboring mutations in the sole Arabidopsis Gα (AtGPA1), Gβ (AGB1), and Regulator of G protein Signaling (AtRGS1) genes for impacts on ion channel regulation. In guard cells, where K⁺ fluxes are integral to cellular regulation of stomatal apertures, inhibition of inward K⁺ (K_in) currents and stomatal opening by the phytohormone abscisic acid (ABA) was equally impaired in Atgpa1 and agb1 single mutants and the Atgpa1 agb1 double mutant. AGB1 overexpressing lines maintained a wild-type phenotype. The Atrgs1 mutation did not affect K_in current magnitude or ABA sensitivity, but K_in voltage-activation kinetics were altered. Thus, Arabidopsis cells differ from mammalian cells in that they uniquely use the Gα subunit or regulation of the heterotrimer to mediate K_in channel modulation after ligand perception. In contrast, outwardly rectifying (K_out) currents were unaltered in the mutants, and ABA activation of slow anion currents was conditionally disrupted in conjunction with cytosolic pH clamp. Our studies highlight unique aspects of ion channel regulation by heterotrimeric G proteins and relate these aspects to stomatal aperture control, a key determinant of plant biomass acquisition and drought tolerance.

G protein-coupled inwardly rectifying potassium or “GIRK” channels (also known as Kir3 channels) comprise important targets of heterotrimeric G protein regulation in mammals (1, 2). GIRK channels mediate signals from muscarinic, adrenergic, opioid, dopaminergic, and GABA_B receptors (3). Basal activity of GIRK channels is repressed by their direct binding of Gα_i/oGDP (4, 5) within a macromolecular complex that includes Gβγ (4–7). Upon activation of G_i/o-coupled G protein-coupled receptors (GPCRs), formation of the GTP-bound form of Gα_i/o both alleviates Gα-mediated repression and releases βγ dimers that independently interact with the channel (7, 8). Gβ_1-4γ binding strengthens GIRK interaction with phosphatidylinositol 4,5-bisphosphate (PtdInsP₂), thereby promoting conformational changes that increase channel open time (3, 9–11). Conversely, Gα_q-based activation of phospholipase C opposes GIRK activity via both depletion of PtdInsP₂ and activation of PKC-based phosphorylation events (12).

Numerous studies analyzing GIRK activity in Xenopus oocytes and cultured mammalian cells have led to the beautifully intricate model described above. However, studies analyzing GIRK activity in the appropriate native cell context upon genetic depletion of Gβ subunits are lacking. Arabidopsis has single genes, AtGPA1 (henceforth referred to as GPA1) and AGB1, encoding canonical α and β subunits, two identified genes, AGG1 and AGG2, encoding γ subunits, and a single Regulator of G protein Signaling (RGS) gene, AtRGS1 (henceforth referred to as RGS1), encoding RGS1 which accelerates GTPase activity of GPA1 (13–15). The present study takes advantage of knockout mutants (13, 16) in the model plant system, Arabidopsis thaliana, to investigate G protein regulation of inward K⁺ (K_in) currents, outwardly rectifying K⁺ (K_out) currents, and slow anion currents in their native condition and to assess the roles of G protein-based pathways in cellular function.

Specialized guard cells residing in pairs in the leaf surface regulate the apertures of microscopic pores, “stomata,” through which plants both take up the CO₂ required for photosynthesis and, inevitably, lose water vapor (17–20). Guard cell responses to the plant hormone abscisic acid (ABA) play a vital role in plant resistance to drought, a major cause of crop loss (18). ABA inhibits guard-cell K_in channels and activates Ca²⁺-permeable channels and anion channels through which anion efflux occurs (17–20). Osmotically driven guard cell inflation is thereby inhibited and guard cell deflation is promoted, resulting in inhibition of stomatal opening, promotion of stomatal closure, and reduced plant water loss. Guard-cell K_in currents share with GIRK channels the properties of inward rectification, activation by ATP, activation by PtdInsP₂, and regulation by cellular redox status (9, 10, 20–23). We previously showed that genetic depletion of GPA1 relieves ABA inhibition of guard-cell K_in currents (24, 25). These studies led to a number of additional questions. (i) As in mammalian cells, do G protein subunits regulate basal levels of K_in current observed in the absence of agonist? (ii) Is loss of an inhibitory regulator, Gα, or gain of a stimulatory regulator, free Gβγ (as in mammals), responsible for the gpa1 phenotype? (iii) Do RGS proteins modulate K⁺ currents in plants as in mammalian cells? (iv) Is there G protein regulation of K_out channels? (v) Given our previous observations that plant Gα subunits also regulate slow anion channels (24, 25), do plant Gβ subunits participate in this regulation? (vi) What are the effects of altered expression of G protein components on integrated guard cell responses to ABA, as reflected in regulation of stomatal apertures?

Results

Basal Levels and ABA Inhibition of K_in Current Do Not Differ Among agb1 and gpa1 Single and gpa1 agb1 Double Mutants.

We applied patch-clamp whole-cell recording techniques to evaluate basal K_in current magnitude, i.e., in the absence of ABA application, in two independent agb1 mutants. Fig. 1 shows that basal (Control) K⁺ channel activity does not differ among the agb1 mutant lines and their isogenic wild type, Col-0. We next made a side-by-side comparison of Gα and Gβ single mutants and the double mutant derived from them. Because the Gβ mutants are in the Columbia (Col) ecotypic background, the Col gpa1 null mutant, gpa1-4, which has not been previously evaluated for its ion channel characteristics, was used for comparison instead of Ws (Wassilewskija)-based gpa1-1 or gpa1-2 (24, 25). None of the Gα or Gβ mutations affected basal K_in currents.

Fig. 1.

ABA sensitivity of guard cell K_in current regulation is abrogated in agb1-1, agb1-2, and gpa1-4 single mutants and in agb1-2 gpa1-4 double mutants. (A) Representative whole-cell recordings of K_in currents (Col, agb1-1, agb1-2, gpa1-4, gpa1 agb1 double mutant) with or without 50 μM ABA. Whole-cell currents were recorded from a holding potential of −79 mV with 3.9-s voltage steps from −219 to −59 mV in +20-mV increments, 10 min after achieving the whole-cell configuration. Time and current scales are shown in A. (B) Current–voltage (I–V) curves (mean ± SE) of time-activated whole-cell K_in currents. Time-activated currents were calculated by subtracting the instantaneous current at 20 ms from the average steady-state current between 3.55 and 3.87 s. n = 13, 15 cells for control and ABA treatment of Col; n = 11, 11 cells for agb1-1; n = 10, 12 cells for agb1-2. n = 15, 10 cells for gpa1-4; and n = 10, 10 cells for gpa1-4 agb1-2.

ABA inhibition of K_in current was attenuated in the agb1-1 and agb1-2 mutants. Identical attenuation was also seen in gpa1-4 and in the gpa1-4 agb1-2 double mutant (Fig. 1); i.e., no additive or synergistic effects were observed (Fig. 1).

ABA Inhibition of K_in Currents Is Unaltered in AGB1-Overexpression (OX) Lines.

If, as in mammalian cells, Gβ subunit release from the heterotrimer promotes K⁺ channel activity, then overexpression of AGB1 would be expected to stimulate K_in currents. We evaluated two independent lines in which AGB1 overexpression in a wild-type background was clearly observed by immunoblot analysis [supporting information (SI) Fig. S1]. However, the K_in currents of these lines showed wild-type sensitivity to inhibition by ABA, as did the control empty vector lines (Fig. 2). A wild-type phenotype was maintained even at subsaturating ABA concentrations which might be expected to more readily reveal any ABA-hypersensitive phenotype (Fig. S2). In addition, agb1-2 lines complemented with an AGB1 construct and exhibiting AGB1 overexpression showed wild-type ABA responses (Fig. S3), whereas gpa1-4 agb1-2 double mutants complemented with this construct retained ABA hyposensitivity (Fig. S3).

Fig. 2.

ABA inhibits K_in currents similarly in wild-type and AGB1 overexpressing lines. (A) Whole-cell recordings of K_in currents with or without 50 μM ABA from Col, two AGB1 overexpressing lines (OX5, OX6), and two empty vector lines (EV1, EV3) as controls for the OX lines. (B) I–V curves (mean ± SE) of time-activated whole-cell K_in currents as recorded in A. n = 7, 8 cells for control and ABA treatment of Col; n = 9, 10 cells for OX5; n = 13, 11 cells for OX6; n = 10, 10 cells for EV1; and n = 8, 9 cells for EV3.

rgs1 Null Mutation Does Not Affect K⁺ Current Amplitude or ABA Response but Affects Kinetics of Voltage Activation.

There are no previous studies on RGS regulation of plant ion channel activity or guard cell function. We found that RGS1 is expressed in guard cells (Fig. S4), but lack of RGS1 did not affect the magnitude of basal K⁺ current, its voltage dependency, or its inhibition by ABA (Fig. S5). However, accelerated kinetics of K_in current response after voltage activation were observed in the rgs1 mutants under control conditions (Fig. S5). This finding indicates that RGS1 is important in the dynamics of voltage-dependent activation of K_in current.

Gα and Gβ Subunit Interaction.

To confirm interaction between GPA1 and AGB1, a myc epitope-tagged AGB1 construct was transformed into Arabidopsis suspension cells by Agrobacterium-mediated transformation (Fig. 3A). Protein extracts were subjected to immunoprecipitation by using anti-GPA1 antibodies (26) (Fig. S6) or preimmune serum, and anti-myc antibody was used in subsequent immunoblot analysis. Fig. 3B shows the coimmunoprecipitation of AGB1 with GPA1.

Fig. 3.

AGB1 interacts with GPA1 but does not influence GPA1 expression levels. (A) Myc-AGB1 expression in Arabidopsis suspension cells transformed with 35S::myc epitope-tagged AGB1, detected by using anti-myc antibody. Immunoblot with anti-GPA1 antibody illustrates equal loading. (B) AGB1 coimmunoprecipitates with GPA1. Total protein extracts were coimmunoprecipitated (IP) by anti-GPA1 (α-GPA1) or by preimmune serum, then immunoblotted (IB) with the anti-myc (α-myc) antibody. (C) Immunoblot of GPA1 in Col and two agb1 mutants. GPA1 is expressed at a similar total level among Col wild-type and agb1 mutant lines, and membrane vs. soluble localization of GPA1 is not affected in agb1 mutants. Tot, total fraction; Mem, crude membrane fraction; Sol, soluble protein fraction. (D) Confocal laser scanning microscopy images of guard and epidermal cells showing membrane localization of GPA1-CFP in the agb1-2 mutant background. Shown are confocal (Left), differential interference contrast microscopy (Center), and merged (Right) images of guard cells (Upper) and epidermal cells (Lower). (Scale bars: Upper, 5 μm; Lower, 10 μm.)

One explanation for the similarity of the K_in phenotypes of agb1 and gpa1 plants would be that the agb1-2 mutation affects GPA1 expression. However, reverse-transcriptase PCR (RT-PCR) and quantitative reverse-transcriptase real-time PCR (Q-PCR) analyses indicate that mutations in agb1 have no effect on expression of GPA1, nor does the gpa1 null mutation affect AGB1 expression (Fig. S7). In addition, immunoblot analyses performed on total, crude membrane and soluble protein fractions from rosette leaves showed that the presence or absence of AGB1 did not affect either total GPA1 levels or GPA1 partitioning to the microsomal fraction (Fig. 3C). To confirm the uniform membrane-delimited localization of GPA1 in agb1-2 mutants, confocal laser scanning microscopy was performed on epidermes of agb1-2 seedlings expressing a GPA1-CFP fusion protein. GPA1-CFP fluorescence in both leaf epidermal cells and guard cells was observed at the plasma membrane (Fig. 3D); this fluorescence receded from the cell wall upon guard cell plasmolysis, indicating that the fluorescence was indeed associated with the membrane as opposed to the cell wall (data not shown).

K_out Currents Are Unaltered in Gβ Mutants, Whereas Anion Currents Show Conditional Alteration.

In plant cells, inward and outward K⁺ currents are mediated by molecularly distinct channel proteins (20, 27). Wild-type Col did not show ABA regulation of K_out currents (24, 28). In addition, K_out currents exhibited wild-type behavior in agb1 mutants, the gpa1-4 mutant, the gpa1 agb1 double mutant, and AGB1-OX lines (Fig. 4 A–D).

Fig. 4.

ABA does not affect K_out currents of either wild-type or G protein mutants and affects anion currents in a pH- and G protein-dependent manner. (A) Typical whole-cell recordings of K_out currents (Col, agb1-1, agb1-2, gpa1-4, gpa1 agb1 double mutant) with or without 50 μM ABA. Currents were recorded from a holding potential of −79 mV with 3.9-s voltage steps from −59 to 61 mV in +20-mV increments, 10 min after achieving the whole-cell configuration. Time and current scales are shown below B. (B) Typical whole-cell recordings of K_out currents from Col, AGB1 overexpression lines (OX5 and OX6) and pGWB42 empty vector control lines (EV1 and EV3) with and without 50 μM ABA treatments. (C) I–V curves (mean ± SE) of time-activated whole-cell K_out currents, calculated by subtracting the instantaneous current at 20 ms from the average steady-state current between 3.55 and 3.87 s. n = 13, 15 cells for control (○), ABA treatment of Col (●); n = 11, 11 cells for control (□), ABA treatment of agb1-1 (■); and n = 10, 12 cells for control (▵), ABA treatment of agb1-2 (▴). n = 15, 10 cells for control (▿), ABA treatment of gpa1-4 (▾); and n = 10, 10 cells for control (◇), ABA treatment of gpa1-4 agb1-2 (♦). (D) I–V curve (mean ± SE) of time-activated whole-cell K_out currents. n = 7, 8 cells for control (○) and ABA treatment of Col (●); n = 9, 10 for OX5 (□, ■); n = 13, 11 for OX6 (▵, ▴); n = 10, 10 for EV1 (▿, ▾); n = 8, 9 for EV3 (◇, ♦). (E) Typical whole-cell recordings of slow anion currents (Col, agb1-1, agb1-2) with or without 50 μM ABA under strong cytosolic pH buffering (10 mM Hepes-Tris). Whole-cell anion currents were recorded 12 min after achieving the whole-cell configuration. The holding potential was +30 mV and voltage steps were from −145 to +35 mV in +30-mV increments. (F) Typical whole-cell recordings of anion currents (Col, agb1-1, agb1-2) plus or minus 50 μM ABA under weak cytosolic pH buffering (0.1 mM Hepes-Tris). (G) I–V curves (mean ± SE) of steady-state whole-cell anion currents as recorded in E. Steady-state currents were acquired by subtracting the basal currents at a holding potential of +30 mV from the average currents between 42.5 and 50.0 s. n = 12, 13 cells for control, ABA treatment of Col; n = 12, 12 cells for agb1-1; and n = 13, 13 cells for agb1-2. (H) I–V curves (mean ± SE) of steady-state whole-cell anion currents as recorded in F. Steady-state currents were acquired as in G. n = 19, 23 cells for control, ABA treatment of Col; n = 10, 10 for agb1-1; and n = 17, 21 for agb1-2.

Slow anion currents of guard cells are enhanced by ABA (29). gpa1 null mutants exhibited wild-type anion channel response to ABA in the absence of cytosolic pH clamp and loss of this response in the presence of a cytosolic pH clamp (24). This same interaction with cytosolic pH status was observed for the agb1 mutants (Fig. 4 E–H).

ABA-Inhibition of Stomatal Opening Is Altered in Gβ Mutants and in GαGβ Double Mutants.

To assess a correlation between ion channel regulation and cellular function, assays of stomatal responses were conducted. ABA inhibition of stomatal opening (Fig. 5A) was attenuated in the two agb1 mutants compared to wild type. Just as for ion channel regulation, the double gpa1 agb1 mutant phenotype was identical to that of the single mutants (Fig. 5A). In the AGB1-OX lines, ABA inhibition of stomatal opening was retained at wild-type levels (Fig. 5B and Fig. S3), consistent with the wild-type response of these lines for ABA modulation of K_in current. rgs1 null mutants also showed wild-type ABA inhibition of stomatal opening (Fig. S8), consistent with the wild-type response of these mutants with regard to steady-state K⁺ current magnitude after ABA application. All genotypes exhibited wild-type ABA induction of stomatal closure (Fig. 5 C and D).

Fig. 5.

ABA inhibition of stomatal opening is impaired in agb1 and gpa1 single and double mutants and is unaltered in AGB1-OX lines. (A) Reduced ABA (20 μM) inhibition of stomatal opening (mean ± SE) in agb1-1, agb1-2, gpa1-4, and gpa1-4 agb1-2 double mutants as compared to wild type. ABA significantly inhibited stomatal opening in Col (P ≤ 0.001) but had no significant effect in any of the other mutant lines (P > 0.05). Inset illustrates two Col guard cells defining an open stomatal pore. (Scale bar: 10 μm.) Data are mean ± SE from five independent replicates. In each replicate, >150 stomata were measured for each genotype × treatment combination. (B) Significant ABA (20 μM) inhibition of stomatal opening in AGB1-OX lines (OX5, OX6), empty vector controls (EV1, EV3), and wild type (Col). Mean ± SE from four replicates. (C) Significant ABA (20 μM) promotion of stomatal closure (mean ± SE) in agb1-1, agb1-2, gpa1-4, and gpa1-4 agb1-2 double mutants. Mean ± SE from four replicates. (D) Significant ABA (20 μM) promotion of stomatal closure in AGB1-OX lines (OX5, OX6) as compared to empty vector controls (EV1, EV3). Mean ± SE from three replicates.

Discussion

Regulation of Plant K_in Channels by G Protein Complex Components.

The plant hormone ABA inhibits K_in currents of wild-type guard cells (17) and null mutation of the Arabidopsis G protein α subunit gene, GPA1, results in loss of this response (24, 25). A primary goal of the present report was to determine the roles of Gα vs. Gβγ in this phenomenon and to determine whether there was evidence that the heterotrimeric state of the Arabidopsis G protein complex could play a regulatory role.

In mammalian systems, Gα, within the heterotrimeric complex, suppresses basal GIRK current levels in the absence of agonist, and upon agonist perception, freed Gβγ dimer acts to increase the open probability of GIRK channels (5, 7). Three key findings of our study are that, in Arabidopsis: (i) loss of either the Gα or Gβ subunits has no effect on basal (−ABA) K⁺ current; (ii) loss of Gα or Gβ or both causes identical hyposensitivity to ABA-inhibition of K_in currents and stomatal opening; and (iii) overexpression of the Gβ subunit has no effect on K_in current, either in the absence or in the presence of ABA. These observations lead to our first conclusion: the Gβγ dimer does not operate on K⁺ channels in the same manner in plant and mammalian cells. Specifically, the mammalian mechanism in which free Gβγ activates the channels is not supported by our AGB1 knockout mutant and overexpression data in Arabidopsis. The interpretation of the wild-type nature of the AGB1 overexpression phenotype is tempered by the possibility that active Gβγ is limited in plant cells, for example by a fixed pool of Gγ subunits. However, this is unlikely in that AGB1 overexpression confers clear scorable phenotypes (30, 31).

Coimmunoprecipitation of GPA1 and AGB1 from Arabidopsis cells (Fig. 3 A and B) supports previous observations of their interaction in heterologous systems (30, 32) as well as biochemical evidence for interaction in rice and pea (33, 34). Interaction between AGG1 or AGG2 and AGB1 is seen in yeast two-hybrid and in vitro binding assays and by FRET (15, 32). Collectively, these data support the notion that plant α, β, and γ subunits form heterotrimers. Given the existence of a heterotrimer, three hypotheses are consistent with the identical phenotypes of the agb1 and gpa1 null mutants: (i) the independent action of free Gα and free Gβγ subunits sum to mediate ABA-inhibition of the K_in channels, and loss of either Gα or Gβ is sufficient to cause ABA hyposensitivity; (ii) the Gα subunit mediates ABA inhibition of the K_in channels; and (iii) ABA regulation of the K_in channels is mediated by the G protein heterotrimer.

If the first hypothesis were correct, we might anticipate that the double gpa1 agb1 mutant would exhibit greater ABA hyposensitivity than gpa1 or agb1 single mutants, e.g., in stomatal opening assays, and that overexpression of AGB1 would result in ABA hypersensitivity, but these phenotypes were not observed (Figs. 1 and 2 and Fig. S2). Accordingly, we discuss the latter two hypotheses below.

In mammalian systems, Gβγ dimers are important for maintenance of appropriate signaling through Gαs, and thus Gα-dependent signaling (hypothesis ii) can also be disrupted in Gβ mutants, as we observed. In mammalian cells, Gβ knockdown also can affect Gα expression or targeting (35–37); however, we saw no evidence for these phenomena in Arabidopsis (Fig. 3 C and D and Fig. S7). Instead, AGB1 may have more direct regulatory effects on GPA1. For some mammalian Gαs, Gβγ may act as a lever to promote conformational change of the Gα subunit upon GPCR activation, thus promoting GDP release (38, 39) (but also see ref. 40). In addition, segments of Gβ as well as Gγ interact with GPCRs (41), and Gγs (and Gβ5) appear to play important roles in GPCR-Gα coupling specificity (41), and disruption of such coupling may be occurring in the Arabidopsis agb1 mutants.

In plants, data from previous patch-clamp experiments employing pharmacological G protein modulators implicated regulation of K_in channels by both cytosol-mediated and membrane-delimited G protein-based pathways (13, 42, 43). Indirect regulation of Arabidopsis K_in channels via a signaling cascade would parallel the indirect inhibition of GIRK channels by Gα_q, via activation of PLC and PKC (12). Of interest in this regard are the observations that ABA elevates PLC activity in guard cells (44) and that the guard-cell K_in channels are activated by PtdInsP₂ (23). PLC-mediated hydrolysis of PtdInsP₂ may reduce availability of this K_in channel activator in the guard-cell plasma membrane. In guard cells, PLC and PLD appear to operate in the same ABA-signaling cascade (45), and phosphatidic acid (PA), the product of PLD activity, is another attractive candidate regulator: PA inhibits guard-cell K_in currents (45), and PLD is ABA-activated (45) and regulated by Gα in guard cells (46). In addition, one class of Shaker-like channels, the KCNQ channels, whose mutation is associated with genetic diseases including long QT syndrome and atrial arrhythmias (47), displays modulation via a G_s-dependent cascade, including activation by PtdInsP₂ (47, 48). On a sequence homology basis, the guard cell K_in channels, despite evincing inward rectification, are similar to metazoan Shaker K⁺ channels (27).

Alternatively, the ABA hyposensitive phenotypes reported here may be accounted for by disruption of signaling via a nondissociated heterotrimer (hypothesis iii), as occurs in certain yeast and mammalian G protein signaling cascades (49–51). Heterotrimer-dependent signaling would be expected to be perturbed equally in agb1 single mutants, gpa1 single mutants, and gpa1 agb1 double mutants, i.e., the phenotypes we observe. Our data do not directly speak to the question of whether such a heterotrimer would contain GDP-GPA1 or GTP-GPA1: one FRET study on plant cells indicates that a mutant, constitutively active (GTP-bound) form of GPA1 can still exhibit FRET with AGB1, consistent with retention of a GTP-GPA1 subunit in a heterotrimer (32, 50, 52). Alternatively, ABA might stimulate activity of an as yet unidentified GDI or RGS protein (other than RGS1) and thus shift GDP-Gα into the heterotrimeric complex.

The basal state in animals is reduced K_in channel activity in the absence of agonist. In contrast, in Arabidopsis, the reduced activity level occurs when the agonist (ABA) is present. If the above scenario proves to be correct, Gα may inhibit GIRK-like current in a mechanistically similar manner in Arabidopsis cells and mammalian cells and this may have been the ancestral action of Gα on K⁺ channel activity. Consistent with the idea of mechanistic similarities, during regulation of mammalian GIRK channels by RGS, alteration in current kinetics but not steady-state current–voltage relationships is commonly observed (e.g., 6, 53–55), and this feature is also shown here for Arabidopsis K_in channels (Fig. S5).

Integrated Guard Cell Responses in G Protein Complex Mutants.

K_out channels of guard cells are Shaker-like channels that mediate K⁺ efflux during stomatal closure (20, 27). As reported previously (24, 28), we did not observe ABA activation of K_out channels in wild-type plants. Although G proteins regulate some mammalian Shaker-type channels (47, 48), we found no evidence for G protein involvement in modulation of either basal K_out currents or their ABA responsiveness (Fig. 4), consonant with our observation that wild-type ABA induction of stomatal closure occurs in the G protein mutants and transgenics (Fig. 5).

ABA activation of ion channels other than K_out channels may be more central to ABA promotion of stomatal closure. Anion loss through ABA-activated slow anion channels (29) decreases anion content in the guard cell and promotes membrane depolarization which drives K⁺ efflux, resulting in water efflux, guard cell deflation, and stomatal closure. Our previous research indicated that Gα-dependent and pH_i-dependent cascades provide redundant pathways for ABA-activation of anion channels in guard cells (24). We similarly observed a wild-type activation of anion channels by ABA in the agb1-1 and agb1-2 mutants and, as for gpa1 mutants, this ABA-activation was eliminated in guard cells subjected to cytosolic pH clamp (Fig. 4 E–H). These results further support the conclusion that ABA-related guard cell phenotypes of Gβ mutants recapitulate those of Gα mutants. Intriguingly, for other ABA-related processes (seed germination, root growth, seedling gene expression), gpa1 and agb1 mutants exhibit ABA hypersensitivity (56), suggesting unexplored richness in the mechanisms of hormonal signaling through plant heterotrimeric G proteins.

Materials and Methods

Plant Material and Growth Conditions.

All transgenics were in the Col accession of Arabidopsis thaliana. An ethyl methanesulfonate-generated mutant line (agb1-1), transfer (T)-DNA insertional mutant lines (agb1-2, gpa1-4, rgs1-1, rgs1-2), and gpa1-4 agb1-2 double-mutant lines have been described (16, 26, 31).

To generate AGB1-overexpressing lines (AGB1-OX), AGB1 cDNA was cloned into the vector pGWB1:35S:YFP; corresponding empty vector lines (EV) were also isolated as controls. The GPA1-CFP construct was as described in ref. 26. Constructs were transformed into plants by Agrobacterium-mediated transformation. The myc epitope-tagged AGB1 was cloned into Gateway plant destination vector pGWB21 and transformed into Arabidopsis suspension cells by Agrobacterium-mediated transformation.

For electrophysiological and physiological assays, all lines were grown under 0.120 mmol m⁻² s⁻¹ light (8 h/16 h day/night cycle) with ≈80% relative humidity, and 22°C/20°C day/night temperatures. For each experiment, wild-type Col plants were grown and assessed simultaneously with the mutant and/or transgenic lines.

Patch-Clamp Analyses.

Guard cell protoplast isolation and K⁺ current recording were performed as described (24, 25), with minor modifications (see Fig. 1 legend and SI Text). Anion current recording was performed according to Pei et al. (29) and Wang et al. (24) with some modifications (see Fig. 4 legend and SI Text). Data were compared by using the Student t test. Results with P ≤ 0.05 were considered statistically significant.

GPA1 Coimmunoprecipitation and Immunoblot Analysis.

Total protein extracts of Arabidopsis suspension cells were coimmunoprecipitated by anti-GPA1 antibodies or by preimmune serum as described (26). Protein lysates from young fully expanded rosette leaves of 4- to 5-week-old plants were prepared as described (25). Proteins from total, microsomal, and soluble fractions were separated on 10% SDS-polyacrylamide gels, electroblotted onto nitrocellulose membrane, and immunoblotted (56) (SI Text).

Confocal Microscopy.

Confocal imaging was performed by using an inverted Zeiss LSM510 Confocal microscope with a Plan-Neofluar 40×/1.3 oil differential interference contrast microscopy objective. For CFP, an excitation wavelength of 458 nm was used, and fluorescence was detected by using a 475–525-nm band-pass filter. Postacquisition image processing was done with the LSM 5 Image Browser (Zeiss) and Adobe Photoshop.

Stomatal Bioassay.

Stomatal aperture bioassays were conducted as described (24), with minor modifications (see SI Text). Values are means ± SE from at least three independent replicates, with at least 150 stomatal apertures measured per each replicate. Stomatal aperture measurements (Fig. 5) were performed blind.