Stomatal pores formed by a pair of guard cells in the leaf epidermis control gas exchange and transpirational water loss. Stomatal closure is mediated by the release of potassium and anions from guard cells. Anion efflux from guard cells involves slow (S-type) and rapid (R-type) anion channels. Recently the SLAC1 gene has been shown to encode the slow, voltage-independent anion channel component in guard cells. In contrast, the R-type channel still awaits identification. Here, we show that AtALMT12, a member of the aluminum activated malate transporter family in Arabidopsis, represents a guard cell R-type anion channel. AtALMT12 is highly expressed in guard cells and is targeted to the plasma membrane. Plants lacking AtALMT12 are impaired in dark- and CO2-induced stomatal closure, as well as in response to the drought-stress hormone abscisic acid. Patch-clamp studies on guard cell protoplasts isolated from atalmt12 mutants revealed reduced R-type currents compared with wild-type plants when malate is present in the bath media. Following expression of AtALMT12 in Xenopus oocytes, voltage-dependent anion currents reminiscent to R-type channels could be activated. In line with the features of the R-type channel, the activity of heterologously expressed AtALMT12 depends on extracellular malate. Thereby this key metabolite and osmolite of guard cells shifts the threshold for voltage activation of AtALMT12 towards more hyperpolarized potentials. R-Type channels, like voltage-dependent cation channels in nerve cells, are capable of transiently depolarizing guard cells, and thus could trigger membrane potential oscillations, action potentials and initiate long-term anion and K+ efflux via SLAC1 and GORK, respectively.
Aluminum-activated malate transporter (ALMT) channels were first described as plasma membrane located, Al3+-activated malate channels by patch-clamp studies in root cells (Kollmeier et al., 2001; Pineros and Kochian, 2001), and were genetically identified in Triticum aestivum (TaALMT1) (Sasaki et al., 2004) and Arabidopsis thaliana (AtALMT1) (Hoekenga et al., 2006). These channels play a central role in aluminum resistance by releasing malate from the root tip, thereby chelating aluminum in the rhizosphere (Sasaki et al., 2004; Hoekenga et al., 2006). In a later study it has been shown that certain AtALMTs can also reside on the tonoplast acting as vacuolar malate channels (Kovermann et al., 2007). Interestingly, one member of the ALMT protein family, ZmALMT1, however, activates Al3+-independently, and transports inorganic anions such as Cl−, and , rather than malate (Pineros et al., 2008). Very recently an Arabidopsis ALMT protein (AtALMT12) has been described to be strongly expressed in guard cells, and to be permeable for chloride and nitrate (Sasaki et al., 2010). Loss-of-function mutants were impaired in stomatal closure. However, under the conditions used by the Sasaki laboratory, neither the S-type nor the R-type channel activities appeared altered in guard cells. Since the authors observed the GFP fluorescence predominatly in the endoplasmic reticulum (ER), they predicted AtALMT12 to function in the release of inorganic anions from the ER. Thus the role of AtALMT12 for stomatal closure remains elusive.
Based on independent, parallel studies to those of the Sasaki laboratory (2010), we here provide convincing evidence: (i) that AtALMT12 is expressed in the plasma membrane of guard cells; (ii) that guard cells of loss-of-function mutants are impaired in malate-dependent R-type channel activity; and (iii) that AtALMT12 is malate permeable and thus well suited for the release of malate from guard cells, as shown for Vicia faba (Keller et al., 1989; Dietrich and Hedrich, 1994).
Tissue-specific expression analysis of AtALMT12
In order to identify potential candidates for R-type channels, we searched for ALMTs expressed in guard cells. Gene expression data from microarray experiments (http://www.bar.utoronto.ca) indicated strong mRNA accumulation of AtALMT12 in guard cells. To verify the microarray data we transformed Arabidopsis plants with the GUS gene under the control of a 2018-bp promoter region (pAtALMT12) upstream of the genomic sequence of AtALMT12. Strong GUS activity in pAtALMT12::GUS transformants was detected in guard cells of different tissues (Figure 1a, b; Figure S1a, b). Additionally, a signal was also observed in the pollen tissue, as well as in the stele of roots (Figure S1b, c). These observations were consistent with the microarray gene expression data and data presented by Sasaki et al. (2010).
Subcellular localization of an AtALMT12-GFP fusion protein
In order to investigate the subcellular localization of AtALMT12, the green fluorescence protein (GFP) was fused to the C- as well as the N-terminal end of AtALMT12. Transient expression of these constructs in Arabidopsis protoplasts under the control of the 35S promoter revealed that AtALMT12-GFP was targeted to the plasma membrane (Figure 1c–f). In order to verify plasma membrane localization of AtALMT12 in guard cells, where its promoter activity has been detected, we also performed independent confocal laser scanning microscopy analyses of stably transformed Arabidopsis. In general most channels are present only at very low protein numbers, and can therefore hardly be visualized using their own promoter. We thus generated transgenic Arabidopsis plants expressing the AtALMT12-GFP construct under the control of the stomata-specific promoter AtMYB60 (Cominelli et al., 2005; Nagy et al., 2009). Confocal microscopy analysis showed GFP fluorescence along the guard cell periphery, thus confirming plasma membrane localization recognized by transient expression in Arabidopsis protoplasts (Figure 1g, h). Taken together, our studies identified AtALMT12 as a protein of the Arabidopsis guard cell plasma membrane, rather than a protein of the intracellular membrane systems.
Analysis of mutant lines carrying a T-DNA insertion in the AtALMT12 gene
Guard cells are sensitive to environmental and endogenous changes, including light, CO2 and ABA. To unravel the physiological role of AtALMT12 in guard cells we analyzed stomatal movements in two independent atalmt12 mutant lines (atalmt12-1 and atalmt12-2) identified in the JIC T-DNA insertion mutant collection (Tissier et al., 1999). The absence of the AtALMT12 transcripts in the mutants was demonstrated by RT-PCR (Figure S2). As a first step we investigated whether wild-type plants and atalmt12 mutants differ in their reaction to the plant hormone ABA, which is synthesized in response to drought stress. As observed by Sasaki et al. (2010), in the presence of ABA stomata of wild-type guard cells closed efficiently. In contrast, both atalmt12 mutant plants barely responded to ABA, and stomata remained largely open even after 2 h of incubation with ABA (Figure 2). In all experiments the light-stimulated opening of guard cells was similar in wild-type and mutant plants. In order to see whether the difference in ABA response could also be observed in response to other stimuli at the whole-plant level, we monitored the stomatal conductance in atalmt12 mutants and wild-type plants in response to light and CO2. Compared with wild-type plants atalmt12 mutants exhibited a much slower decline of stomatal conductance in response to light–dark transitions (Figure 3a, b). Furthermore, the increase in [CO2] from 365 to 800 ppm caused a rapid stomatal closure in the wild type, which was less pronounced in mutant plants (Figure 3c, d).
Electrophysiological studies on atalmt12 guard cell protoplasts and AtALMT12 expressing Xenopus laevis oocytes
The fact that atalmt12 mutant plants are impaired in stomatal closure and members of this family operate as anion channels (Pineros et al., 2008) directed us to examine the electrical properties of this potential anion channel in guard cells. To study the anion channel transport capacity of Arabidopsis guard cells, we performed patch-clamp studies with protoplasts isolated from wild-type plants and atalmt12-1 mutants. The slac1 mutant exhibits a largely reduced S-type current, whereas R-type currents remained unaffected by the loss of this anion channel function (Vahisalu et al., 2008; Geiger et al., 2009, 2010). Consequently, we tested whether guard cells isolated from the atalmt12 mutants appear altered in plasma membrane R-type anion currents. In contrast to V. faba (Hedrich et al., 1990; Marten et al., 1991; Raschke et al., 2003), R-type channels in Arabidopsis guard cells have not been characterized in detail. Therefore, we first analyzed the R-type properties of wild-type guard cells because knowledge of R-type characteristics represents a basic requirement for understanding the phenotype of the atalmt12 mutants. To resolve R-type-specific anion release currents we performed patch-clamp studies using 75 mm sulfate-based pipette solutions (cf. Vahisalu et al., 2008). With 20 mm sulfate in the extracellular medium, and the plasma membrane clamped to a holding potential of −180 mV, depolarizing voltage pulses elicited inward currents (Figure S3a; cf. Sasaki et al., 2010). Under these conditions anion currents reversed around the Nernst potential of sulfate, and displayed fast activation and deactivation kinetics. Thus Arabidopsis guard cells appear to express plasma membrane anion channels with voltage-dependent properties of the R-type channel GCAC1/QUAC found in V. faba (Kolb et al., 1995; Schulz-Lessdorf et al., 1996; Raschke et al., 2003). However, no difference in anion channel activity could be detected between wild-type and atalmt12 mutant plants under these conditions (Figure S3b). In V. faba malate was shown to activate R-type currents (Hedrich and Marten, 1993; Raschke et al., 2003). We thus challenged Arabidopsis wild-type and almt12-1 mutant guard cell anion channels with malate. In the presence of extracellular malate, R-type anion currents of almt12-1 mutants appeared reduced by 40% when compared with wild-type guard cells (Figure 4a, b). Subtracting wild-type currents from those observed in atalmt12 guard cells resulted in a bell-shaped current–voltage curve. This electrical behaviour points to strong voltage dependence and channel activation upon depolarization, which are both hallmarks of the R-type anion channels. Such differential R-type currents between wild type and atalmt12-1 guard cell protoplasts have not been observed by Sasaki et al. (2010) because their experiments were performed solely under external chloride-based conditions, meaning in the absence of the R-type channel gating modifier malate. Thus, based on the characteristic voltage-dependent features of the malate-dependent currents (Figure 4), the loss-of-function phenotype suggested that AtALMT12 is likely to encode a malate-sensitive component of the R-type anion channel of Arabidopsis guard cells.
To study the malate-sensitive component of the R-type current in guard cells, we expressed AtALMT12 in Xenopus oocytes. Following the injection of AtALMT12 cRNA into oocytes, depolarizing voltages elicited outward currents under chloride-based external solutions (Figure S4a). The amplitudes of these anion uptake currents appeared to depend on the external chloride concentration (Figure S4b; cf. Sasaki et al., 2010). Upon replacement of external chloride by malate, however, voltage pulses elicited both inward and outward currents of up to 7 μA (Figure 5a, b). Interestingly, Sasaki et al. (2010) could only observe outward currents (anion uptake), and no inward currents (anion release) in the presence of external malate and sulfate (Figures 5 and S5). This discrepancy with our results may have arisen from Sasaki et al.’s (2010) use of less physiological external pH conditions. For further characterization we examined the voltage-dependent gating of AtALMT12 with a double voltage pulse protocol under extracellular ionic conditions similar to those used in patch-clamp experiments with guard cell protoplasts. Therein, the pulse to −200 mV was followed by a depolarizing pulse to +60 mV, before applying trains of hyperpolarizing pulses (Figure 5a). After opening AtALMT12 channels at +60 mV, they were forced to deactivate as a function of the subsequent negative voltage steps. In line with the R-type currents in protoplasts (Figure 4), the voltage dependence of these steady-state currents could also be described by a bell-shaped current–voltage curve (Figure 5b). In the presence of comparable external malate concentrations (Figure 4, 20 mm in protoplasts; Figure 5b, 25 mm in oocytes), channel activation already occurred at less depolarized potentials in oocytes than in protoplasts, very probably because of the divergent cytosolic composition (Figures 4 and 5b). Channel gating depended on the concentration of malate (Figure 5b, c). Upon an increase in the malate concentration the half-maximal activation potential shifted towards more negative membrane potentials (Figure 5c; Table S1; Hedrich and Marten, 1993; Raschke et al., 2003). When sulfate was injected into oocytes, AtALMT12-mediated inward currents increased (Figure S5), indicating that sulfate is also the preferred substrate of the anion channel (Roberts, 2006). This is in agreement with the finding that the presence of K2SO4 reduces the impact of Cl− in stomatal action (Schnabl and Raschke, 1980). In contrast to AtALMT1, AtALMT12 currents were not stimulated by extracellular Al3+ treatment (Figure 6a).
Malate is not only a key metabolite and signaling component for guard cells, but was shown to represent a substrate for R-type channels in V. faba guard cells (Keller et al., 1989; Hedrich and Marten, 1993). Following the injection of malate (20 mm final cytosolic concentration) buffered to pH 7.5, inward currents were recorded with moderate negative-going membrane potentials (Figure 6b). Note that malate injection activates pronounced R-type-like currents with AtALMT12 in the absence of malate in the oocyte external medium. The addition of 10 mm malate into the external medium further maximized these inward currents. These experiments show that similarly to the V. faba R-type channel (Hedrich and Marten, 1993), malate functions as a gating modifier as well as a permeating substrate of AtALMT12.
In Arabidopsis the ALMTs constitute a small gene family of 14 members that can be subdivided into three clades (Kovermann et al., 2007). The best characterized AtALMT1, which is located in the root plasma membrane, is a member of clade 1, whereas the vacuolar AtALMT9, mainly expressed in the leaf mesophyll, is assigned to clade 2 (Hoekenga et al., 2006; Kovermann et al., 2007). Microarray data suggest that AtALMT12, a member of clade 3, is strongly expressed in guard cells. Our present study and an independent one recently published by Sasaki et al. (2010) could confirm the predicted predominant expression of AtALMT12 in this cell type. Using two different AtALMT12 loss-of-function mutants we observed an impaired stomatal closure when leaves were exposed to ABA. Similar results were obtained by Sasaki et al. (2010) with independent mutant lines. These authors could also show impaired stomatal closure in the presence of Ca2+ and in the dark. In our study, furthermore, gas-exchange measurements revealed a delayed and not complete stomatal closure in plants lacking the AtALMT12 protein. These results convincingly show that AtALMT12 is required for efficient stomatal closure.
AtALMT12 is a strongly voltage-dependent plasma membrane anion channel
Sasaki et al. (2010) stated that AtALMT12 is an outward rectifying (anion uptake into the cytosol) channel permeable for chloride and nitrate, but not sulfate and malate, as demonstrated in our work. From Sasaki’s studies AtALMT12 function in stomatal movement remained elusive, as the localization for AtALMT12 was observed in both the endoplasmic reticulum and the plasma membrane. This equivocal localization may result from the transient expression used by these authors, whereas in our study a clear plasma membrane localization of AtALMT12 was observed in plants stably transformed with a similar construct. In particular, the predicted direction of the anion movements in Sasaki et al. (2010), makes it difficult to attribute a role in stomatal closure to AtALMT12 located in the plasma membrane. Furthermore, working in the absence of malate the authors could not detect differences in R-type channel activities between guard cells of wild-type and atalmt12 mutant Arabidopsis plants. Therefore, Sasaki et al. (2010) postulated a predominant role of AtALMT12 in the release of inorganic anions from the endoplasmic reticulum into the cytosol. However, as the volume of the endoplasmic reticulum is relatively small it is thus rather unlikely that anions released from this compartment have a major impact in stomatal closure. Furthermore, AtALMT-mediated currents measured on native membranes have so far shown only inward rectification (Kovermann et al., 2007; Zhang et al., 2008). Assuming a slight negative membrane potential between the ER and the cytosol, resulting from the activity of a V-type ATPase, no anion release to the cytosol would be detectable, taking into account the current–voltage curves presented by Sasaki et al. (2010).
In previous studies, ALMTs have been characterized as channels for dicarboxylates such as malate and fumarate, but also for inorganic anions like Cl−, and (Pineros et al., 2008). As the latter substrate specificity was also known for R-type channels (Roberts, 2006), and because AtALMT12 is highly expressed in guard cells, we explored whether AtALMT12 could act as the as yet unidentified plasma membrane R-type channel. Our studies on guard cell protoplasts and AtALMT12-expressing oocytes provide evidence that AtALMT12 represents a malate-sensitive component of the R-type anion channel in guard cells of Arabidopsis. In both systems, protoplasts and oocytes, R-type anion currents showed similar voltage dependence and kinetics. In line with ZmALMT1 from Zea mays, but in contrast to TaALMT1 or AtALMT1, AtALMT12 was not induced by Al3+ (Sasaki et al., 2004; Hoekenga et al., 2006; Pineros et al., 2008; ). We thus suggest renaming it QUAC1 in accordance with the naming of the S-type anion channel of guard cells, SLAC1 (Raschke et al., 2003; Negi et al., 2008; Vahisalu et al., 2008). As in atalmt12/quac1 mutant guard cells, only malate-dependent anion currents appeared to be affected, one would predict that QUAC is based on different channels exhibiting similar current–voltage curves, but possibly with different substrate specificities and malate dependencies. Alternatively, QUAC could be formed by heteromers with AtALMT12 as one subunit responsible for malate sensing. Clade 3 of the AtALMT protein family is constituted by three additional members: AtALMT11, AtALMT13 and AtALMT14 (Kovermann et al., 2007). AtALMT11 is not likely to be a candidate for an additional component of the QUAC channel because the predicted structure is constituted by only two membrane domains, and the protein is weakly expressed in guard cells and other tissues (http://www.bar.utoronto.ca). However, AtALMT13 and AtALMT14 might be candidates for further members of a heteromeric QUAC complex, although no expression data are available so far. Future studies on the guard cell-expressed QUACs thus need to focus on the identification of additional components of the R-type channel.
In former studies it has been shown that besides Cl−, and , the R-type channels also exhibit permeability for malate (Keller et al., 1989; Hedrich and Marten, 1993; Roberts, 2006). Malate in the apoplast has been postulated to play an important role in activating anion release, possibly by inducing a shift in the voltage dependence of R-type anion channels (Hedrich and Marten, 1993; Raschke et al., 2003; Konrad and Hedrich, 2008). As shown in Figure 6b, AtALMT12 is not only sensitive to malate, but is also permeable for this organic ion, and it is therefore tempting to speculate that this component of QUAC might play an important role in this feed-forward stimulus for long-term anion release.
AtALMT12/R-type/QUAC1 and SLAC1 poise guard cells for voltage and volume control
SLAC1, addressed by ABA signaling kinase/phosphatase pairs, seems to be required: (i) to drive the long-term efflux of osmotically active anions from guard cells and stomatal closure; and (ii) thus for an effective decrease in transpiration when soil water is limiting (Negi et al., 2008; Vahisalu et al., 2008; Geiger et al., 2009). QUAC-like channels in V. faba also appear to be addressed by the water stress hormone (Raschke et al., 2003; Roelfsema et al., 2004). In contrast to SLAC1, QUAC1 activation in oocytes neither required the presence of OST1 nor required plant-specific CPKs (cf. Geiger et al., 2009, 2010). These observations strongly suggest that SLAC1 and QUAC1 are likely to represent response elements on different branches of the ABA signaling pathway leading to stomatal closure (Levchenko et al., 2005).
Arabidopsis thaliana wild-type plants (Col-0) and mutant plants were grown in controlled environment chambers in potting soil or on agar medium under a 8-h/16-h light to dark regime (90 μmol m−2 s−1, 21°C, 60% relative humidity). The floral-dip method (Clough and Bent, 1998) was used for obtaining transgenic lines.
Selection of atalmt12 T-DNA mutant lines
T-DNA insertion lines of AtALMT12 were obtained from the JIC collection (Tissier et al., 1999), and plants homozygous for the T-DNA insertion were isolated by PCR genotyping. Two independent T-DNA insertion lines of AtALMT12 were identified (atalmt12-1 and atalmt12-2). Wild-type lines were selected as those plants that genotyped as wild type (WT-like) during homozygous PCR screening of the JIC mutant line seed batch. In all assays of stomatal measurements (Figures 2 and 3) atalmt12-1 and atalmt12-2 mutant lines were compared with the corresponding WT lines (WT-like-1 and -2). For details see Supporting information.
Subcellular localization and tissue-specific expression
Tissue-specific expression of AtALMT12 in Arabidopsis was analysed by amplifying a 2018-bp promoter region fused to the GUS reporter gene in the pGPTV-Bar vector (Becker et al., 1992). C- and N-terminal GFP fusion constructs with AtALMT12 were generated in pUC18-GFP5T-sp (Meyer et al., 2006) and a modified pART7 vector (Endler et al., 2006) for transient expression, and a C-terminal GFP fusion construct was generated in a modified pMDC83 vector for stable transformation (Nagy et al., 2009). For details see Supporting information.
Stomatal aperture and gas exchange measurements
Gas exchange was measured with a portable gas exchange system (LI-6400; LI-COR, http://www.licor.com). For response to ABA, detached rosette leaves were treated with 10 μm ABA after pre-incubation in opening buffer. For details see Supporting information.
Plant growth conditions and isolation of guard cell protoplasts for electrophysiological studies were performed as described previously (Geiger et al., 2009). Using the patch-clamp technique, protoplasts were studied in the whole-cell configuration essentially as described by Wolf et al. (2006). The standard bath solution was composed of 2 mm MgCl2, 0.5 mm LaCl3, 10 mm MES, pH 5.6/Tris and either 20 mm CaGluconate2 + 20 mm Cs2SO4 or 20 CaMalate. The pipette solution consisted of 75 mm Cs2SO4, 2 mm MgCl2, 5 mm Mg-ATP, 10 mm Hepes, pH 7.1/Tris. To obtain a free-Ca2+ concentration of 1 μm, the pipette solution additionally contained 5 mm EGTA plus 4.2 mm CaCl2. The osmolality of the pipette and bath media was adjusted to 440 and 400 mosmol kg−1, respectively, with d-sorbitol. Oocyte measurements were performed using the two-electrode voltage-clamp technique (TEVC). 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 variable concentrations of NaH-malate, NaCl and/or NaGluconate. If necessary, osmolality was adjusted to 220 mosmol kg−1 using d-sorbitol. Injection of 50 nl of a 200 mm Na+-malate solution resulted in a final malate concentration of 18 mm in the oocyte because a mean volume of 500 nl for the spherical oocytes was calculated from the averaged oocyte diameter of 1 mm. Voltage pulse protocols similar to those used for studies of guard cell R-type channels (Hedrich et al., 1990; Marten et al., 1991) were applied from a holding voltage of −20 mV. Details of solutions and pulse protocols are mentioned in the figure legends. For further details of data acquisition and analysis see Supporting information.
We would like to thank Dr Alexis De Angeli for critical discussions and Bo Burla for help with LICOR measurements. This work was supported by the Swiss National Foundation (to EM), by grants GK1342 and FOR964 of the Deutsche Forschungsgemeinschaft to RH, and by a KSU grant to RH and KAR.