Open stomata 1 (OST1) kinase controls R–type anion channel QUAC1 in Arabidopsis guard cells

Authors


  • Accession numbers: AtOST1, At4g33950; AtABI1, At4 g26080; AtALMT12/AtQUAC1, At4 g17970; DmQUAC1, KC285588; and DmACT1, KC285589.

For correspondence (e-mail geiger@botanik.uni-wuerzburg.de).

Summary

Under drought stress, the stress hormone ABA addresses the SnR kinase OST1 via its cytosolic receptor and the protein phosphatase ABI1. Upon activation, OST1 phosphorylates the guard cell S–type anion channel SLAC1. Arabidopsis ABI1 and OST1 loss-of-function mutants are characterized by an extreme wilting 'open stomata′ phenotype. Given the fact that guard cells express both SLAC- and R–/QUAC-type anion channels, we questioned whether OST1, besides SLAC1, also controls the QUAC1 channel. In other words, are ABI1/OST1 defects preventing both of the guard cell anion channel types from operating properly in terms of stomatal closure? The activation of the R–/QUAC-type anion channel by ABA signaling kinase OST1 and phosphatase ABI1 was analyzed in two experimental systems: Arabidopsis guard cells and the plant cell-free background of Xenopus oocytes. Patch-clamp studies on guard cells show that ABA activates R–/QUAC-type currents of wild-type plants, but to a much lesser extent in those of abi1–1 and ost1–2 mutants. In the oocyte system the co-expression of QUAC1 and OST1 resulted in a pronounced activation of the R–type anion channel. These studies indicate that OST1 is addressing both S–/SLAC- and R–/QUAC-type guard cell anion channels, and explain why the ost1–2 mutant is much more sensitive to drought than single slac1 or quac1 mutants.

Introduction

Stomata represent turgor-operated valves in the epidermis of leaves, controlling CO2 uptake assimilation and loss of water vapour transpiration. As soil dries out, a rise in the stress hormone abscisic acid (ABA) signals a change in the availability of root water to the shoot. On the level of the stomata a pair of guard cells, which regulate the aperture of the stomata, represents the key target in transpiration control (Blatt, 2000; Hetherington, 2001). Plants that are unable to close their stomata eventually wilt upon drought.

The pioneering studies of Sarah Assmann and co-workers identified an ABA-dependent guard cell-expressed protein kinase, AAPK, in Vicia faba (broad bean) that facilitates the ABA activation of anion channels and thus stomatal closure (Li and Assmann, 1996; Li et al., 2000). In the model plant Arabidopsis thaliana, an orthologous kinase was identified and named SRK2E (SNF1-related protein kinase 2E; Yoshida et al., 2002) or open stomata 1 (OST1; Mustilli et al., 2002) for the inability of the loss-of-function mutant to close its stomata. Besides ost1, abi1–1 was isolated as another major open-stomata mutant (Leung et al., 1994; Meyer et al., 1994). A dominant, single-site mutation deregulates protein 2C-type phosphatase causing an ABA-insensitive mutant phenotype. In contrast to this ABA kinase–phosphatase pair, the ABA receptors (RCARs/PYRs/PYLs) have been isolated in mutant screens by a chemical genetic approach, and by ABI1-based protein–protein interaction assays (Ma et al., 2009; Park et al., 2009). Recently, it was shown that ABA-dependent gene expression and phosphorylation-activation of anion channel proteins share major signal transduction elements (Fujii et al., 2009; Geiger et al., 2009; Lee et al., 2009). In the absence of ABA the phosphatase ABI1 prevents the activity of OST1. When ABA binds to its receptor, ABI1 becomes inactivated, and OST1 auto-phosphorylates and regulates downstream targets such as bZip-transcription factors (Furihata et al., 2006), NADPH oxidase AtrBohF (Sirichandra et al., 2009) and, via the trans-phosphorylation guard cell channel, SLAC1 (Geiger et al., 2009). The opening of S–type anion channels, such as SLAC1, results in anion release and thereby in depolarization of the guard cell plasma membrane. As a result, the depolarization-activated guard cell outward rectifying K+ channel (GORK) mediates the release of potassium ions (Ache et al., 2000). Finally, the massive loss in potassium salts causes a drop in guard cell turgor and volume, and in turn stomatal closure (Hosy et al., 2003). This ABA signaling cascade is significantly disturbed when OST1 is scant (ost1–2) or ABI1 protein (abi1–1) is hyperactive because of the dominant mutation G180D. As a result, the deregulated PP2C phosphatase is no longer under ABA-dependent control of the PYR/PYL/RCAR-like ABA receptor (Miyazono et al., 2009; Yin et al., 2009; Umezawa et al., 2010). In turn, OST1 remains inactive, even in the presence of ABA, mimicking an OST1 loss-of-function phenotype (Joshi-Saha et al., 2011).

Compared with ost1–2 mutants, the loss of SLAC1 results in a less pronounced stomata phenotype, indicating that besides SLAC1, other guard cell anion channels contribute to stomatal closure (Geiger et al., 2011). The osmotic-driven guard cell movements depend on K+ salts and the anions chloride, nitrate, malate and even sulfate. In addition to SLAC1, the SLAC homolog SLAH3 is expressed in guard cells too (Geiger et al., 2011). In contrast to the chloride and nitrate conducting SLAC1, SLAH3 is predominantly permeable to nitrate (Geiger et al., 2011). Furthermore, this second guard cell S–type channel is addressed by ABA via ABI1 and calcium-dependent protein kinase CPK21/23, rather than OST1.

Early patch-clamp studies have shown that voltage-independent S–type anion channels localize side-by-side with voltage-dependent quickly activating anion channels of the R–/QUAC type in the plasma membrane of guard cells (Keller et al., 1989; Schroeder and Hagiwara, 1989; Hedrich et al., 1990; Linder and Raschke, 1992). In search for the molecular nature of the R–/QUAC-type channel, member 12 of the ALMT family was identified (Meyer et al., 2010). Although the first identified member ALMT1 was characterized as an aluminum-activated malate transporter (Sasaki et al., 2004), other family members, including ALMT12, were not sensitive to aluminum and exhibited a channel-like gating behavior (Sasaki et al., 2004, 2010; Pineros et al., 2008; Meyer et al., 2010). As ALMT12 shares the hallmark feature with quickly-activating anion channels of the R–/QUAC type, in analogy with the nomenclature of the first S–type or slow anion channel SLAC1, ALMT12 was named QUAC1 (Meyer et al., 2010). First studies of the structure–function relationship of QUAC1 revealed the high impact of the C terminus in voltage-dependent channel gating (Mumm et al., 2013). QUAC1 conducting malate thus provides a channel for the movement of the major organic anion for proper stomatal action. Interestingly, R–/QUAC-type channels in guard cells – just like the S–type – are addressed by ABA (Levchenko et al., 2005). Given the fact that ABA activates both anion channel types with SLAC1 carrying chloride and nitrate, and with QUAC1 mediating malate transport, it is tempting to speculate that the release of major guard cell anionic osmotica for coordinated stomatal closure is under the control of the same signaling elements.

Thus, we asked whether, in response to ABA, OST1 is activating both SLAC1 and QUAC1. To find the reason why a defect in OST1 causes a more pronounced open-stomata phenotype than the sole lack of either SLAC1 or QUAC1, we studied R–/QUAC-type channels in Arabidopsis guard cells of both open-stomata mutants ost1–2 and abi1–1. In a complimentary approach the OST1 sensitivity of QUAC1 was analyzed in the plant cell-free background of the heterologous expression system Xenopus laevis oocytes.

Results

ABA-induced stimulation of QUAC1-type currents in guard cells

In previous studies we analyzed the selectivity and voltage-dependent gating of R–/QUAC-type channels in the plasma membrane of guard cells and QUAC1-expressing oocytes (Meyer et al., 2010). Using the same biological systems, we characterized the OST1 sensitivity of the guard cell QUAC1 anion channels. To study the role of this ABA signaling SnR kinase in R–/QUAC-type channel regulation in its natural membrane surrounding, we isolated guard cell protoplasts from Arabidopsis wild-type and mutant plants. In the whole-cell patch-clamp configuration, R–/QUAC-type macroscopic anion currents were monitored with 75 mm sulfate-based pipette solutions and 20 mm malate in the extracellular medium (Figures 1 and 2; cf. Meyer et al., 2010). As the presence of extracellular malate specifically promotes the activation of QUAC1 channels (Meyer et al., 2010), this ‘malate state’ of the guard cell currents served as reference for the capacity of ABA to stimulate QUAC1-type currents. ABA was shown to activate the S–type anion channels via ABI1/CPKs in a calcium-dependent manner (Mori et al., 2006; Geiger et al., 2010, 2011), and via ABI1/OST1 independent of the signaling cation Ca2+ (Levchenko et al., 2005; Geiger et al., 2009; Lee et al., 2009; Chen et al., 2010). Therefore, in order to provide conditions suited to resolve ABA-dependent but calcium-independent anion channel activation, the free Ca2+ concentration in the patch pipette solution, which is in equilibrium with the guard cell cytosol in the whole-cell configuration, was buffered to 110 nm. Active S–type anion channels exhibit no pronounced voltage dependence in response to voltage stimulation, and respond very much like ohmic resistors (Geiger et al., 2009). In contrast, R–/QUAC-type channels in guard cells and AtQUAC1 in oocytes show a fast voltage-gated relaxation behavior (Hedrich et al., 1990; Meyer et al., 2010). When the plasma membrane was clamped to a holding voltage of −180 mV, in the absence of ABA, depolarizing 1–s voltage steps from −180 mV to +80 mV elicited typical but small depolarization-activated R–/QUAC-type anion currents (Figure 1). When ABA (25 μm) was applied to the pipette medium, i.e. to the cytosolic side of the plasma membrane, a rise in the R–/QUAC-type currents was observed at the peak current potential of −100 mV (Figure 1b). For quantification of this stimulatory ABA effect, the steady-state current densities (Iss/Cm) were plotted against the applied membrane voltages (Figure 1c). The derived current density–voltage curves [(Iss/Cm)/V] indicate that the presence of cytosolic ABA caused more than a doubling of the R–/QUAC-type anion currents at negative voltages (Figure 1c). This behavior is very much in line with microelectrode impalement recordings from Vicia faba guard cells in their natural environment of the intact bean plant (Levchenko et al., 2005), and confirms that in addition to SLAC-type anion channels, ABA addresses R–/QUAC-type channels in Arabidopsis guard cells too.

Figure 1.

Effect of ABA on the R–/QUAC-type anion currents of wild-type guard cell protoplasts. (a) Fast activation of macroscopic anion currents upon voltage pulses, as indicated. (b) Representative whole-cell current responses to a voltage pulse of −100 mV in the absence and presence of 25 μm cytosolic ABA. (c) Steady-state current densities as a function of the voltage in the absence and presence of 25 μm cytosolic ABA. Data points represent means ± SEs for n = 5 runs under both sets of experimental conditions. The peak currents at −100 mV significantly differ (P < 0.05, Student's t–test). (a, b) The dashed line indicates the zero-current level. (a–c) The pipette medium was buffered to 110 nm free Ca2+.

Figure 2.

R–/QUAC-type anion currents of guard cell protoplasts from the wild type and ABA signaling mutants. (a) Steady-state current densities shown as a function of the clamped voltage recorded from guard cell protoplasts of wild-type (WT) plants (open symbols, n = 5) and abi1–1 mutants (closed symbols, n = 5). (b) Steady-state current densities shown as a function of the clamped voltage recorded from guard cell protoplasts of wild-type plants (open symbols, n = 5) and ost1–2 mutants (closed symbols, n = 6). (a, b) Data points represent means ± SEs. The pipette medium contained 25 μm ABA and was adjusted to 110 nm free Ca2+. Peak currents at −100 mV significantly differ (P < 0.05, Student's t–test).

ABI1 and OST1 represent regulatory elements for the ABA-dependent activation of QUAC1-type guard cell anion currents

To further answer questions about the ABA signaling pathway, we focused on guard cells of the Arabidopsis open-stomata mutants abi1–1 and ost1–2, which are impaired in the ABA activation of SLAC1-type currents (Pei et al., 1997; Siegel et al., 2009; Geiger et al., 2010; Roelfsema et al., 2012 for review). Note that the loss of OST1 function in ost1–2 mutant plants and the hyperactive PP2C phosphatase in abi1–1 plants cause a similar stomata phenotype. When comparing the ABA-activated R–/QUAC-type currents in the 'malate state′ of the mutant abi1–1 with those of the wild type, R–/QUAC-type anion channel activity in guard cells was reduced by more than 50% (Figure 2a). A similar reduction in R–/QUAC-type currents was monitored with ost1–2 guard cells lacking the activity of the open-stomata SnR kinase (Figure 2b). To obtain further evidence of whether the loss of either of the regulatory ABA elements leads to a similar effect on the R–/QUAC-type channel activity, the macroscopic channel current kinetics were analyzed (Table S1) and the relative voltage-dependent open probabilities (rel. Po) were calculated under the given experimental conditions (Figure S1). Inspection of whole-cell current relaxation in response to voltage pulses revealed the activation and deactivation time constants at −100 mV and −160 mV, respectively, for both mutants, which did not significantly differ from their corresponding wild-type ecotypes (Table S1). When the relative open probabilities were plotted against the applied membrane voltages (Figure S1) and fitted in terms of a Boltzmann distribution, the derived half-maximal activation voltage V1/2 could be used for the quantification of voltage-dependent gating behavior. Thereby, we found that R–/QUAC-type channels in both mutants were characterized by half-maximal activation voltages that were well in line with the voltages measured in the respective wild types. Thus, the fact that the abi1–1 mutant, which constitutively inhibits OST1 activation, and the OST1-deficient mutant seem to impair R–/QUAC-type function in a similar manner indicates that the ABA signaling phosphatase–kinase pair addresses R–/QUAC-type channels.

Direct interaction between OST1 and QUAC1

OST1 co-expression with SLAC1 demonstrates that physical interaction between SnRK and the channel is required to elicit S–type anion currents (Geiger et al., 2009; Lee et al., 2009). To test for bimolecular fluorescence complementation (BiFC)-based protein–protein interaction, the AtQUAC1 protein and the SnR kinase were each fused to one half of the yellow fluorescence protein (YFP; Figure 3). The YFP half was linked to the N terminus of AtQUAC1, whereas the complementary half was attached to the C–terminus of AtOST1. When only the YFPNT::AtQUAC1 or the AtOST1::YFPCT construct was expressed in X. laevis oocytes, no YFP-related fluorescence was detected (Figures 3a and S2A); however, YFP fluorescence was observed upon co-expression of the YFPNT::AtQUAC1 and AtOST::YFPCT constructs in X. laevis oocytes, documenting direct channel–kinase interaction (Figure 3b,c).

Figure 3.

QUAC1-OST1 protein–protein interaction validated via bimolecular fluorescence complementation (BiFC). (a) Image from an oocyte expressing AtQUAC1 channels with its N terminus fused to the N–terminal part of YFP (YFPNT::AtQUAC1). (b, c) Images from an oocyte expressing AtQUAC1 and AtOST1 fused with the N terminus-respective C terminus to the N- and C-terminal part of a YFP, respectively (YFPNT::AtQUAC1 + AtOST1::YFPCT). (c) A magnified image from an enlarged section of the oocyte shown in (B) demonstrates that the YFP signal is distributed on the surface of the membrane. Scale bar: a, b, 500 μm; c, 100 μm.

OST1-dependent stimulation of QUAC1 channels

To study whether QUAC1 gating is OST1-sensitive in the plant cell-free background system, we applied the two-electrode voltage-clamp technique to Xenopus oocytes expressing the Arabidopsis guard cell R–/QUAC-type anion channel in the absence and presence of Arabidopsis SnRK OST1. The membrane potential was clamped to +60 mV for 100 ms for QUAC1 channel pre-activation, followed by consecutive 200 ms pulses in the range from +60 to −200 mV in 10 mV decrements. With the latter voltage pulses, current amplitudes changed in a voltage-dependent fashion because of AtQUAC1 channel deactivation, with and without external malate stimulation (Figure 4). As the concentration of malate and other organic anions is quite low in the cytosol of oocytes, at the holding and the pre-pulse voltage of +60 mV malate equilibrates with the unstirred subplasma membrane layer of the oocyte. Thus, pronounced QUAC1 inward currents were only observed in the presence of malate (Figure 4d,e; cf. Meyer et al., 2010). The sole expression of AtOST1 did not induce any macroscopic anion currents independent from the presence of chloride or malate in the bath chamber (Figure 4c,f). When QUAC1 was co-expressed with OST1 (Figure 4b,e), however, voltage pulses elicited increased steady-state inward and outward currents (Figures 4 and 5a,c). The stimulatory effect of the SnR kinase on the QUAC1 currents was monitored in the absence of malate (Figures 4a,b and 5a) as well as with QUAC1 in the 'malate state′ (Figures 4d,e and 5c), and was related to an altered voltage threshold for AtQUAC1 channel activation (Figure 5b,d). Under chloride- and malate-based conditions the half-activation voltage was shifted by 22 and 31 mV, respectively, towards more negative membrane voltages (Figure 5b,d). This finding is well in line with the fact that in guard cells of the ost1–2 as well as abi1–1 mutant, the SnRK-inducible currents were abolished (Figure 2a,b).

Figure 4.

Stimulation of QUAC1 via interaction with OST1. (a–c) Representative voltage-induced current responses of either YFPNT::AtQUAC1-expressing, YFPNT::AtQUAC1 and AtOST1::YFPCT co-expressing or AtOST1::YFPCT-expressing oocytes monitored in chloride-based medium. (d–f) Representative voltage-induced current responses of either YFPNT::AtQUAC1-expressing, YFPNT::AtQUAC1 and AtOST1::YFPCT co-expressing or AtOST1::YFPCT-expressing oocytes monitored in malate-based medium. (a–f) For clearer presentation, current responses to test voltages from +60 mV to −170 mV in 30–mV steps are shown. The dashed line indicates the level at zero current.

Figure 5.

OST1-induced change in voltage-dependent QUAC1 gating. (a, c) Steady-state current–voltage curves determined from oocytes expressing YFPNT::AtQUAC1 in the presence (closed symbols) and absence (half-open symbols) of AtOST1::YFPCT. (b, d) Voltage-dependent relative open probability (rel. Po) determined upon tail current recordings from oocytes expressing YFPNT::AtQUAC1 in the presence (closed symbols) and absence (half-open symbols) of AtOST1::YFPCT. In chloride-based media (b) the half-maximal activation voltage was V1/2 = −55.0 ± 0.3 mV for YFPNT::AtQUAC1 and V1/2 = −77.0 ± 0.2 mV for YFPNT::AtQUAC1+AtOST1::YFPCT. In malate-based media (d) the half-maximal activation voltage was V1/2 = −118.0 ± 0.7 mV for YFPNT::AtQUAC1 and V1/2 = −149.0 ± 2.4 mV for YFPNT::AtQUAC1+AtOST1::YFPCT. (e) Steady-state current–voltage relationship, determined with oocytes co-expressing YFPNT::AtQUAC1 and AtOST1::YFPCT either in the presence or absence of ABI1. ABI1 inhibited AtQUAC1-mediated anion currents. (a, b) Experiments were performed with chloride-based solute conditions. (c, d, e) Experiments were performed with malate-based solute conditions. (a–e) Data points represent means ± SEs; n = 12 or 13 for (a–d); n = 3 for (e).

To prove the specificity and conservation of AtOST1-induced AtQUAC1 activation across species, we cloned and expressed the orthologous QUAC1 anion channel from the more distantly related carnivorous plant Dionaea muscipula (DmQUAC1). Sequence alignments between AtQUAC1 and DmQUAC1 revealed an overall similarity of 67% (similar residues or conservative substitutions) at the amino acid level, with 52% identical residues (Figure S3). Similar to the guard cell-specific expression pattern of AtQUAC1, qRT-PCR experiments revealed that DmQUAC1 expression was restricted to the guard cell-enriched outer epidermis of Dionaea muscipula (Venus flytrap; Figure 6a). Similar to the situation with AtQUAC1 and AtOST1, YFP fluorescence was observed upon co-expression of the YFPNT::DmQUAC1 and AtOST1::YFPCT constructs in X. laevis oocytes, documenting a direct interaction between the anion channel and the SnR kinase (Figure 6b). When DmQUAC1 was expressed in Xenopus oocytes using similar voltage protocols as for AtQUAC1 (cf. Figures 4 and 5), only small anion currents could be observed in malate-based buffers (Figure 6c). Co-expression of DmQUAC1 together with AtOST1, however, resulted in macroscopic anion currents 13-fold higher at −120 mV than with DmQUAC1 expressed alone (Figure 6c). After pre-activation of DmQUAC1/AtOST1 by voltage pulses to +60 mV, consecutive hyperpolarizing voltage pulses resulted in fast voltage-dependent deactivation kinetics, relaxing to steady-state currents with peak amplitudes at −80 mV (Figure 6d). Application of voltage ramps ranging from +40 mV to −160 mV documents malate stimulation of DmQUAC1/AtOST1 currents (Figure 6e). Whereas in chloride-containing external solutions current amplitudes changed in a voltage-dependent fashion (DmQUAC1 channel deactivation), external malate stimulation shifted the peak current amplitude to membrane potentials negative to the applied voltage range. Relative open probability calculations under chloride- and malate-based conditions showed that the half-activation voltage was shifted by 77 mV towards more negative membrane voltages in the presence of external malate (Figure 6f). Taken together our findings indicate that DmQUAC1, just like AtQUAC1, represents an R–type anion channel. Thus, the activation of QUAC-type anion channels by OST1 seems to reflect a species-independent feature (Hetherington and Bardwell, 2011; Dreyer et al., 2012).

Figure 6.

Dionaea muscipula (Venus flytrap) DmQUAC1 represents an R–type anion channel activated by AtOST1. (a) Quantification of DmQUAC1 transcripts in different tissues of the Venus flytrap. Transcripts were normalized to 10.000 molecules of DmActin (n ≥ 6, mean ± SE). To enrich guard cells, the blender method was used with epidermal strips as described in Bauer et al. (2013). Guard cell-enriched (GC-enriched) samples contained eight times more DmQUAC1 transcripts than the untreated epidermis strips. (b) Bimolecular fluorescence complementation (BiFC) in Xenopus oocytes documented the direct interaction between DmQUAC1 and AtOST1. YFPNT::DmQUAC1-expressing oocytes did not show any fluorescence. Scale bars: 100 μm. (c) The steady-state currents (ISS) of YFPNT::DmQUAC1-expressing (open symbols, n = 4) and YFPNT::DmQUAC1 and AtOST1::YFPCT co-expressing oocytes (closed symbols, n = 3) were plotted against the membrane voltage. Oocytes were perfused with a solution containing 10 mm NaH-malate. Data points represent means ± SEs. (d) Macroscopic current responses of YFPNT::DmQUAC1- and AtOST1::YFPCT-expressing oocytes in 25 mm chloride-based medium. Following a pre-activating voltage pulse to +60 mV, 500–ms test pulses ranging from +60 mV to −180 mV in 40–mV decrements were applied. The zero-current level is indicated by the dashed line. (e) Whole-cell currents of DmQUAC1 and AtOST1 co-expressing oocytes were either recorded in standard bath solution with 25 mm chloride or 25 mm malate (upper panel). The current traces in gray were monitored during solution exchange. Voltage ramps ranged from +40 mV to −200 mV (lower panel). The zero-current and voltage levels are indicated by dashed lines. (f) Voltage-dependent relative open probability (rel. Po) was determined upon tail current recordings from oocytes expressing YFPNT::DmQUAC1 together with OST1::YFPCT in the presence and absence of 10 mm malate. In chloride-based media the half-maximal activation voltage (V1/2) was −55.5 ± 11.7 mV, whereas V1/2 was −132.9 ± 15.3 mV in malate-based media. Solid lines represent best fits with a single Boltzmann equation. Data points represent means ± SEs for n = 3 experiments.

Discussion

The dual function of the dicarbonate malate, acting as substrate and extracellular gating modifier, was first observed for the R–/QUAC-type anion channels from V. faba guard cells (Dietrich and Hedrich, 1998; Ache et al., 2010). Thereby, the peak current amplitude and potential varied with the ionic conditions not only for malate but also for other permeable anions on the extracellular side of the plasma membrane. Likewise, in root cells of A. thaliana increasing concentrations of the permeable anion sulfate at the extracellular membrane side stimulated R–/QUAC-type anion channels via shifting the threshold potential for channel activation towards more negative voltages (Frachisse et al., 1999; Diatloff et al., 2004). External malate also altered the threshold potential for activation of the A. thaliana guard cell anion channel QUAC1 when heterologously expressed in Xenopus oocytes (Meyer et al., 2010). Interestingly, malate did not affect the voltage dependence of the R–/QUAC-type anion currents within A. thaliana wild-type guard cells (Meyer et al., 2010). Furthermore, in contrast to A. thaliana hypocotyl cells (Thomine et al., 1995; Frachisse et al., 1999), under chloride-based conditions AtQUAC1-expressing oocytes (Meyer et al., 2010) and A. thaliana guard cell protoplasts (Figure S4) exhibited no AtQUAC1-type inward currents. Under sulfate-based internal solute conditions, however, A. thaliana unitary anion-channel conductances monitored from guard cells (Mumm et al., 2013), hypocotyl cells (Frachisse et al., 1999) and root cells (Diatloff et al., 2004) were comparable. Besides ions, nucleotides can influence the opening of R–/QUAC-type anion channels as well. The regulation via nucleotides have been related either to phosphorylation/dephosphorylation events in tobacco suspension cells or to nucleotide binding in V. faba and A. thaliana (Zimmermann et al., 1994; Thomine et al., 1995; Schulz-Lessdorf et al., 1996; Colcombet et al., 2001). Thus, some features of R–/QUAC-type anion currents appear to be related to the species and/or to the cell type-dependent context. Given this situation one would suggest that most likely several distinct R–/QUAC-type channel subunits and/or certain regulatory factors are specifically present within each cell type.

AtQUAC1 belongs to clade III of the ALMT family, whereas the tonoplast-localized ALMTs and Al3+-activated malate transporters (ALMTs) are assigned to clades II and I, respectively (Barbier-Brygoo et al., 2011). The biophysical properties of AtQUAC1 expressed in oocytes and R–/QUAC-type currents recorded in Arabidopsis guard cells strongly deviate from the members of the ALMT family clades I and II. Thus, it is tempting to speculate that the residual R–/QUAC-type anion currents in guard cells just like AtQUAC1 originate from clade–III members of the ALMT family. Members of the clade–I ALMT family (Sasaki et al., 2004; Hoekenga et al., 2006; Ligaba et al., 2006) have been shown to detoxify soluble Al3+ in acidic soils by the extrusion of malate for chelating the toxic cation (Ryan et al., 1997; Kollmeier et al., 2001; Ligaba et al., 2006; Delhaize et al., 2007), and to be targeted by protein kinases (Ligaba et al., 2009). Pre-incubation of TaALMT1-expressing oocytes with the protein kinase inhibitors K252 and staurosporine inhibited Al3+-activated currents (Ligaba et al., 2009). Using site-directed mutagenesis and expression in oocytes, Ser384 in the C terminus of TaALMT1 was identified as a key residue. We could now show that the QUAC1 from Arabidopsis and Venus flytrap are both interacting with SnRK kinase AtOST1, leading to channel activation, most probably via phosphorylation. As both QUACs from distantly related species originate from the ALMT family, it is tempting to speculate that the regulatory phosphorylation site of OST1 might be located in the C terminus as well.

Recently, we and others found that the activity of the guard cell-expressed S–type anion channels SLAC1 and SLAH3 is controlled by a fast ABA signaling pathway (Figure 7; Mori et al., 2006; Geiger et al., 2009; Lee et al., 2009; Geiger et al., 2010, 2011; Brandt et al., 2012; Scherzer et al., 2012). In the absence of ABA the PP2C protein phosphatase ABI1 inhibits the activity of OST1 and CPKs, and thus SLAC1 and SLAH3 anion channels are kept silent. Upon perception of ABA by RCAR/PYR/PYL ABA receptors (Ma et al., 2009; Park et al., 2009), ABI1 is bound and inactivated by the cytosolic ABA receptors. In turn, the inhibition of OST1 and CPK protein kinases is relieved, and SLAH3 is activated by calcium-dependent CPKs, whereas SLAC1 is activated by both calcium-independent OST1 and calcium-dependent CPKs (Geiger et al., 2009, 2010, 2011; Brandt et al., 2012; Scherzer et al., 2012). Based on our present findings, QUAC1 can be integrated into our current model of ABA-dependent anion channel activation (Figure 7). Just like SLAC1, the Ca2+-independent OST1 branch of the ABA signaling pathway controls QUAC1. Thus, the loss of OST1 actually mimics the functional knock-out/down of both channel types, leading to a more pronounced open-stomata phenotype in the ost1–2 mutant than in the channel mutants lacking solely SLAC1 or QUAC1. Chloride and nitrate efflux through SLAH3 and SLAC1 as well, as malate release by QUAC1, leads to depolarization of the guard cell membrane potential. In turn, outward-rectifying potassium channels release potassium salts, finally leading to guard cell turgor loss and stomatal closure. Future studies must elucidate whether QUAC1 is addressed not only by OST1-dependent but also by calcium-dependent ABA signaling.

Figure 7.

Model of QUAC1/SLAC1 regulation by components of the ABA signaling pathway. ABA perception via the cytosolic ABA receptor PYR/PYL/RCAR (here abbreviated as RABA) leads to the inactivation of the PP2C phosphatase ABI1. As a result the SnRK kinase, OST1 is released from ABI1 inhibition. Thus, in the presence of ABA, OST1 is able to phosphorylate/activate the plant anion channels QUAC1 and SLAC1, finally leading to stomatal closure. SLAC1 and SLAH3 can be activated by an alternative Ca2+-dependent pathway via phosphorylation by CPK23 and CPK21. Whether QUAC1 is also activated via the CPK pathway must be proven in future experiments.

Experimental Procedures

Patch-clamp experiments on guard cell protoplasts

Wild-type (Col–0, Ler) and mutant plants of A. thaliana were grown in controlled environment chambers at 21°C and 60% humidity, under an 8–h light/16–h dark cycle. Plant growth conditions and isolation of guard cell protoplasts for electrophysiology studies were as described previously (Meyer et al., 2010). Protoplasts were electrophysiologically studied in the whole-cell patch-clamp configuration, essentially as described elsewhere (Mumm et al., 2011). Anion currents were measured 7 min after whole-cell access and finally normalized offline to the membrane capacitance (Cm) of the respective protoplast. Clamped voltages were corrected offline for the liquid junction potential (Neher, 1992).

The standard bath solution was composed of (in mm): 2 MgCl2, 0.5 LaCl3, 10 2–(N–morpholino)ethanesulfonic acid (MES), pH 5.6/Tris and 20 CaMalate. The pipette solution consisted of (in mm): 75 Cs2SO4, 2 MgCl2, 5 Mg-ATP and 10 Hepes, pH 7.1/Tris. To obtain a free Ca2+ concentration of 110 nm, the pipette solution additionally contained 10 mm EGTA plus 3 mm CaCl2. The osmolality of the pipette and bath media was adjusted to 440 and 400 mOsmol kg−1, respectively, with d–sorbitol. Then (+/−) cis,trans-ABA was added to the pipette medium (to a final concentration 25 μm) from a 25 mm stock solution that was prepared with isopropanol and stored at −18°C.

Two-electrode voltage-clamp experiments on Xenopus oocytes

For functional analyses, cRNA of AtQUAC1, DmQUAC1 and AtOST1 was prepared using the AmpliCap-Max™T7 High Yield Message Maker Kits (Epicentre, http://www.epibio.com). Oocyte preparation and cRNA injection have been described elsewhere (Becker et al., 1996). Oocytes were injected with 50 nl cRNA 500 ng μl−1 of QUAC1 and 250 ng μl−1 OST1. After 3–4 days of expression at 16°C in ND96 solution, whole-oocyte currents were recorded using the two-electrode voltage-clamp technique. The tail currents were normalized to the saturation value of the calculated Boltzmann distribution and fitted with a single Boltzmann equation to derive the half-maximal activation potential (V1/2). Oocytes were perfused with a standard solution containing 10 mm MES/Tris, pH 5.6, 1 mm CaGluconate2, 1 mm MgGluconate2, 1 mm LaCl3 and 25 mm NaH-malate (referred to as malate-based buffer) or 25 mm NaCl (referred as chloride-based buffer), if not otherwise indicated in the figure legends. The osmolality of media was adjusted to 220 mOsmol kg−1 using d–sorbitol.

Voltage protocols

In order to monitor the activation of R–type guard cell anion channels (Figure 1) in the whole-cell patch-clamp configuration, voltage pulses of 1–s duration were applied in the range from −180 mV to +80 mV in 20–mV increments. For this, a holding membrane voltage of −180 mV was chosen to lead the channel population entirely into the closed state before applying the series of depolarizing voltage pulses. Steady-state currents (ISS) were derived upon measuring the current response at the end of the respective voltage pulse (Figures 1 and 2). Deactivation kinetics of R–type guard cell anion channels were extracted from current responses to deactivating hyperpolarizing voltage pulses after pre-activation with a voltage pulse of +60 mV.

In oocyte experiments, we focused on the deactivation process (Figures 4 and 6d). From a holding potential of +60 mV the following voltage pulse protocol was used. QUAC1 channels were pre-activated by a 200–ms voltage pulse of +60 mV to monitor the deactivation kinetics of QUAC1 during the subsequent test pulses (from +60 mV to −200 mV in 10- or 20–mV steps, if not otherwise indicated in the figure legends). Steady-state current amplitudes (Iss) were determined at the end of the test pulses. The relative voltage-dependent open probability Po was determined using the instantaneous tail currents of the final voltage pulse to −180 mV (DmQUAC1) or −200 mV (AtQUAC1).

YFP fluorescence imaging in oocytes

Images were taken from X. laevis oocytes kept in ND96 at pH 5.6 and expressing either AtQUAC1 or DmQUAC1 fused to the N–terminal part of the YFP half (YFPNT::QUAC1) alone, OST1 fused to the C–terminal part of the YFP half (OST1::YFPCT) alone or both constructs together. After expression for 3–4 days, fluorescence pictures were taken with a confocal laser scanning microscope (TCS SP5; Leica, http://www.leica-microsystems.com/). YFP was excited with an Argon laser at 514 nm and emission of YFP fluorescence was recorded between 525 and 600 nm.

Quantification of DmQUAC1 transcripts

Dionaea muscipula plants were purchased from Cresco b.v. (http://www.crescovof.nl) and grown in plastic pots at 26°C with a 12-h light/12-h dark photoperiod. Material for expression analyses was harvested as follows: isolated traps, glands, trigger hairs, epidermal stripes, buds and roots were immediately frozen in liquid nitrogen. The enrichment of guard cells from epidermal strips of the outer surface of Dionaea traps was performed using the ‘blender’ technique, as described by Bauer et al. (2013). Inspection of the available expressed sequence tag (EST) data from D. muscipula (Schulze et al., 2012) revealed the existence of an AtQUAC1 homolog. Using the SMARTer RACE kit (Clontech, http://www.clontech.com), a cDNA was generated from D. muscipula trap RNA. The cDNA of DmQUAC1 was then amplified using gene-specific oligonucleotide primers directed towards the 5′ and 3′ regions of DmQUAC1 (GenBank accession no. KC285588). Amplification with the Advantage2 cDNA polymerase mix (Clontech) revealed a sequence identical to that expected from the EST data. The full-length cDNA of DmQUAC1 was subsequently cloned into oocyte (BiFC-) expression vectors (based on pGEM vectors) by an advanced uracil excision-based cloning technique, as described by Nour-Eldin et al. (2006). Cloning primers: DmQUAC1 user fwd 5′–GGCTTAAUATGGAGGCTGCAGATGG–3′ and DmQUAC1 user rev 5′–GGTTTAAUTTATTCAGGTCCATGAGAAGG–3′.

Expression analyses

RNA was separately isolated from each sample and transcribed into cDNA using M–MLV reverse transcriptase (Promega, http://www.promega.com). The quantification of DmACT1 (GenBank accession no. KC285589) and DmQUAC1 transcripts was performed by real-time PCR, as described elsewhere (Escalante-Perez et al., 2012). Transcripts were each normalized to 10.000 molecules of DmActin. Primers used: Act vft fwd, 5′–TCTTTGATTGGGATGGAAGC–3′; Act vft rev, 5′–GCAATGCCAGGGAACATAGT–3′; DmQUAC1 LCfwd, 5′–TCTAGCGGGTCGTTAAAG–3′; DmQUAC1 LCrev, 5′–GGGCGATGCTGCTAAGTT–3′.

Acknowledgements

This work was supported by grants of the Deutsche Forschungsgemeinschaft (DFG) within GK1342 ‘lipid signaling’ to RH and DG, within the DFG research group FOR964 to RH, and within the DFG grant HE1640/28–1 to RH. RH was also funded by the European community within the ERC 7th Framework Program (grant EU: 250194). RH, DG and KAR were supported by grants of the Strategic Technologies Program – National Science, Technology and Innovation Plan, Saudi Arabia project no. 10–ENV1181–02. We thank Kristina Frohn for two-electrode voltage-clamp measurements for DmQUAC1, and María Escalante-Pérez for RNA extraction and qRT-PCR experiments with Dionaea muscipula.

Author Contributions

DG, IM, KAR and RH designed the research; DI, JB and PM performed research and analyzed data. All authors contributed to writing the article.

Ancillary