Calcium signals in guard cells enhance the efficiency by which abscisic acid triggers stomatal closure

Summary During drought, abscisic acid (ABA) induces closure of stomata via a signaling pathway that involves the calcium (Ca2+)‐independent protein kinase OST1, as well as Ca2+‐dependent protein kinases. However, the interconnection between OST1 and Ca2+ signaling in ABA‐induced stomatal closure has not been fully resolved. ABA‐induced Ca2+ signals were monitored in intact Arabidopsis leaves, which express the ratiometric Ca2+ reporter R‐GECO1‐mTurquoise and the Ca2+‐dependent activation of S‐type anion channels was recorded with intracellular double‐barreled microelectrodes. ABA triggered Ca2+ signals that occurred during the initiation period, as well as in the acceleration phase of stomatal closure. However, a subset of stomata closed in the absence of Ca2+ signals. On average, stomata closed faster if Ca2+ signals were elicited during the ABA response. Loss of OST1 prevented ABA‐induced stomatal closure and repressed Ca2+ signals, whereas elevation of the cytosolic Ca2+ concentration caused a rapid activation of SLAC1 and SLAH3 anion channels. Our data show that the majority of Ca2+ signals are evoked during the acceleration phase of stomatal closure, which is initiated by OST1. These Ca2+ signals are likely to activate Ca2+‐dependent protein kinases, which enhance the activity of S‐type anion channels and boost stomatal closure.


Introduction
Land plants control gas exchange with the surrounding atmosphere by modulating the aperture of stomatal pores in the leaf surface (Shimazaki et al., 2007;Kim et al., 2010;Kollist et al., 2014). In the light, stomata open and enable CO 2 uptake for photosynthesis, whereas they close during drought to protect plants from desiccation. Several lines of evidence show that the stress hormone abscisic acid (ABA) plays a central role in drought-induced stomatal closure (Roelfsema et al., 2012;Munemasa et al., 2015). An in-depth understanding of the molecular mechanisms that underlie ABA-dependent stomatal closure, therefore, can open new strategies of breeding plants with improved drought tolerance.
Stomata rapidly close after stimulation with extracellular ABA (Guzel Deger et al., 2015), which was taken as an indication that this stress hormone is perceived by a cell surface receptor (Joshi-Saha et al., 2011). However, recent findings point to a rapid uptake of ABA into guard cells (Boursiac et al., 2013;Merilo et al., 2015), which is followed by the perception through cytosolic PYrabactin Resistant/PYrabactin resistant-Like/Regulatory Component of ABA Receptors (PYR/PYL/RCAR; Levchenko et al., 2008;Ma et al., 2009;Park et al., 2009). Loss of multiple of these PYR/PYL/RCAR receptors causes stomata to become ABA insensitive (Gonzalez-Guzman et al., 2012;Merilo et al., 2013), which indicates that these proteins are essential for guard cell ABA perception.
In addition to the calcium (Ca 2+ )-independent OST1 pathway, guard cells are also likely to exhibit a Ca 2+ -dependent signaling chain that activates SLAC1, as well as the homologous channel SLAH3 (Geiger et al., 2011;Brandt et al., 2015;Guzel Deger et al., 2015). Such a Ca 2+ -dependent pathway was already postulated in the pioneering work of De Silva et al. (1985). Later experiments with Ca 2+ -sensitive dyes revealed that ABA can indeed trigger a transient elevation of the cytosolic free Ca 2+ concentration in Commelina communis guard cells (McAinsh et al., 1990;Gilroy et al., 1991). The ABA-dependent rise of the cytosolic Ca 2+ level was postulated to activate plasma membrane anion channels, based on experiments in Vicia faba, Arabidopsis and tobacco (Nicotiana tabacum) guard cells (Schroeder & Hagiwara, 1989;Allen et al., 1999;Chen et al., 2010;Stange et al., 2010). However, the hypothesis was challenged by the contrasting finding that ABA is able to activate S-type anion channels in the absence of cytosolic Ca 2+ signals (Levchenko et al., 2005;Marten et al., 2007). One may thus propose that the guard cell ABA signaling pathway is based on a core Ca 2+ -insensitive (OST1-dependent) chain (Cutler et al., 2010), which is modulated by Ca 2+ -dependent processes. However, the interconnection between these two branches and their individual roles in stomatal closure have not been resolved.
Most of the aforementioned guard cell studies that address Ca 2+ signaling have been carried out with stomata in epidermal peels or epidermal fragments. These isolated tissues offer the advantage that fluorescence signals of guard cells are not disturbed by autofluorescence of mesophyll cells. Moreover, stimuli such as ABA can be easily applied to guard cells in epidermal peels, from the side that faces the leaf interior, whereas the side covered by the cuticle offers a strong barrier for many solutes. However, stomatal movements in epidermal peels are reduced in amplitude and response time, in comparison with the stomatal responses in intact leaves (Willmer & Mansfield, 1969;Roelfsema & Hedrich, 2002). It is thus desirable to work with guard cells in intact leaves, but this approach requires a new generation of reporters that enable cytosolic Ca 2+ measurements in intact tissues.
Newly developed genetically encoded Ca 2+ -reporters, such as GCaMP6 and R-GECO1, display much higher Ca 2+ -dependent changes in fluorescence intensity, compared with Yellow Cameleon 3.6 (Zhao et al., 2011;Chen et al., 2013;Waadt et al., 2017). Recently, R-GECO1 has been fused to mTurquoise to generate a highly sensitive Ca 2+ sensor with an internal reference (Waadt et al., 2017). We therefore used intact Arabidopsis leaves that express R-GECO1-mTurquoise (RG-mT) to study Ca 2+ signals in stomata that were stimulated with ABA via microcapillaries in contact with the guard cell wall.

Plant material and growth conditions
All Arabidopsis thaliana lines were in the Col-0 background, the ost1-3, slac1-3, slah3-1 single mutants, the slac1-3/slah3-1 double mutants, and plants expressing RG-mT have been described previously (Yoshida et al., 2002;Merilo et al., 2013;Guzel Deger et al., 2015;Waadt et al., 2017). The ost1-3 mutant was transformed with the RG-mT construct, as described for wildtype by Waadt et al. (2017), using the floral dip method and the Agrobacterium tumefaciens strain GV3101 (Zhang et al., 2006). Seeds were sown on sterilized soil, and plants were grown in a growth cabinet with 60% relative humidity, a cycle of 12 h : 12 h, light : dark, temperatures of 21°C (light) and 18°C (dark), and a photon flux density of 100 lmol m À2 s À1 . After 12 d, the seedlings were transferred to pots (diameter 6 cm) and grown for another 2-3 wk in the same conditions.
Measurements were carried out on stomata, either in isolated epidermal peels or in intact leaves that were excised with a sharp razor blade from 4-to 5-wk-old plants and gently fixed in a petri dish (diameter 35 mm) with the adaxial side attached to doublesided adhesive tape. The leaves were immersed in the following solution: 10 mM potassium chloride (KCl), 1 mM calcium chloride (CaCl 2 ) and 10 mM potassium citrate, pH 5 and illuminated with white light (100 lmol m À2 s À1 ) for at least 2 h before the start of the experiment.
Epidermal strips were gently peeled with a pair of tweezers from the abaxial side of RG-mT-expressing leaves. The strips were fixed on cover slips (diameter 18 mm) with medical adhesive (Medical Adhesive B; Aromando, D€ usseldorf, Germany) and placed in the following bath solution: 10 mM KCl, 1 mM CaCl 2 , 10 mM Mes-bis-tris propane, pH 6.0.

Microelectrode techniques
Stomata in the abaxial epidermis of intact leaves, or epidermal strips, were visualized with a water immersion objective (W Plan-Apochromat, 639/1.0; Carl Zeiss, Jena, Germany) mounted to an upright microscope (Axioskop 2FS; Zeiss). ABA was applied with single-barreled microelectrodes that were pulled from borosilicate glass capillaries (inner diameter, 0.58 mm; outer diameter, 1.0 mm; Hilgenberg, Malsfeld, Germany; http://www. hilgenberg-gmbh.com) on a horizontal laser puller (P2000; Sutter Instruments Co., Novato, CA, USA). The electrode tips were filled with ABA at the standard concentration of 50 lM, or concentrations ranging from 0.5 to 100 lM to obtain a dose-response curve, and the electrodes then further filled with 300 mM KCl. Control experiments were conducted with 50 lM benzoic acid. The electrodes were connected via silver/silver chloride (Ag/ AgCl) half-cells to a headstage of a custom-made amplifier (input impedance > 10 11 Ω, Ulliclamp01). A glass capillary that was filled with 300 mM KCl and sealed with 300 mM KCl in 2% agarose served as a reference electrode. The microelectrodes were mounted to a piezo-driven micro-manipulator (MM3A; Kleindiek Nanotechnik, Reutlingen, Germany) and slowly moved towards the guard cell wall. The tip potential was monitored during manipulation of the microelectrode, and when the electrode came into contact with the guard cell wall it suddenly changed to values more negative than À15 mV. After establishment of a connection between the guard cell wall and microelectrode, ABA, or benzoic acid as control, was ejected from the electrode with a current of À0.8 nA for a period of 20-30 s. Directly after termination of current ejection, the microelectrode was removed from the guard cell wall.

New Phytologist
The plasma membrane conductance of guard cells was studied in voltage clamp experiments with double-barreled microelectrodes. These microelectrodes were fabricated from two borosilicate capillaries (inner diameter, 0.58 mm; outer diameter, 1.0 mm; Hilgenberg), which were aligned, heated, twisted 360°, and pre-pulled on a vertical puller (L/M-3P-A; Heka, Lambrecht/Pfalz, Germany). Subsequently, the joint capillaries were pulled on a horizontal laser puller (P2000; Sutter Instruments Co.). The double-barreled electrodes were backfilled with 300 mM KCl and had a tip resistance that ranged from 180 to 280 MΩ. Both barrels of the microelectrode were connected by Ag/AgCl half cells to the Ulliclamp01 amplifier, which enables voltage clamp experiments with an internal differential amplifier. Voltage pulses were applied with WINWCP software (Dempster, 1997; University of Strathclyde, https://www.strath.ac.uk) and recorded at 1 kHz, using USB-6002 interfaces (National Instruments, Austin, TX, USA; http://www.ni.com). A dual low-pass Bessel filter (LPF 202A; Warner Instruments Corp., Hamden, CT, USA) was used to low-pass filter the electrical signals at 0.5 kHz.

Quantitative fluorescence microscopy
Fluorescence signals of Ca 2+ -imaging experiments were obtained from regions of interest in the central part of guard cells that included the nucleus. The measurements were carried out with a charge-multiplying charge-coupled device camera (QuantEM; Photometrics; http://www.photometrics.com) that was mounted to a CARV, Crestoptics, Rome, Italy confocal spinning disc unit. Within the CARV unit, three filter wheels were used, while the spinning disc was moved out of the light path. The R-GECO1 and mTurquoise subunits in RG-mT were excited with light of an LED illumination system (pE-4000; CoolLED, Andover, UK) at 435 nm and 580 nm, respectively. The emission signals were passed through dichroic mirrors with cut-off wavelengths of 450 nm (T450 LPXR; Chroma Technology Corp., Bellows Falls, VT, USA) and 590 nm (FF593 BrightLine; Semrock, http:// www.semrock.com) and band filters at 475/28 nm (BrightLine HC; Semrock, Semrock Inc., IDEX Corp.; Lake Forest, IL, USA) and 628/40 nm (BrightLine; Semrock).
Changes in stomatal aperture were monitored during the Ca 2+ -imaging experiments, with light provided by a halogen bulb in the microscope lamp, filtered through a far-red light bandpass filter (713/30 nm). All images were analyzed offline with the IMAGE-J/FIJI software package (Schindelin et al., 2012). The statistical and mathematical analysis of the data was carried out with PRISM 6 and 7 (GraphPad Software, San Diego, CA, USA; https://www.graphpad.com) and ORIGIN PRO 8 (Originlab Corp., Northampton, MA, USA).

ABA ejection evokes movement of single guard cells
In a previous study, rapid stomatal closure was induced by nanoinfusion of ABA-containing solution, which was pressure-injected through open stomata into intact leaves (Guzel Deger et al., 2015). Based on this approach, we found that 20 lM ABA induces closure of stomata within 20 min. However, nanoinfusion alters the optical properties of the leaf surface, which is a disadvantage when it is used in combination with quantitative fluorescence microscopy. We therefore introduced a 'currentejection technique' to stimulate single guard cells with ABA in intact Arabidopsis leaves (Fig. 1a). Single-barreled electrodes were slowly moved towards the abaxial epidermis of an intact leaf until the electrode tip came into contact with the guard cell wall (Fig. 1a). In this configuration, electrically charged molecules, such as ABA À , can be ejected from the glass microcapillary into the guard cell wall by application of current pulses. Stimulation with ABA, by a current of À0.8 nA for 20-30 s, caused a rapid reduction of the stomatal aperture (Supporting Information Videos S1). After a lag time of only 1.44 min (Fig. 1b,c; SE = 0.29 min, n = 9), the stomata closed with a maximal velocity of 0.28 lm min À1 (SE = 0.05 lm min À1 , n = 9). By contrast, application of benzoic acid as control (À0.8 nA, 20-30 s) did not affect the aperture of stomata (Fig. 1b,c). ABA had a strong impact on the guard cell that was closely located to the tip of the current-ejection electrode (asterisk in Fig. 1b, left panel), whereas the guard cell on the other side of the pore remained curved (Fig. 1b).
The asymmetric response of guard cells in a stomatal complex suggests that the current-ejection method only provides ABA to a restricted area of the guard cell wall. This was studied by current ejection of the fluorescent dye Lucifer Yellow CH (LY) using the same conditions as already described for ABA (Fig. S1b,c; Methods S1; Videos S2). Indeed, current ejection of LY resulted in a localized fluorescence signal that decreased exponentially from the tip of the electrode (Fig. S1b,c). Based on the LY experiments, it was estimated that the current-ejection procedure transferred a short dose of ABA with a local concentration of 1.1 lM (Fig. S1d). The dose of ABA that was applied to guard cells could be modulated by changing the ABA concentration in the currentejection electrodes. This revealed that a local ABA concentration of 0.2 lM triggered stomatal closure, with only half of the average magnitude, compared with 1.1 or 1.6 lM ABA (Fig. 1d), whereas guard cells did not respond to 0.02 lM ABA (Fig. 1d).
ABA-induced Ca 2+ signals are associated with initiation and acceleration of stomatal closure Several genetically encoded fluorescent Ca 2+ reporters have become available in recent years, of which RG-mT was chosen, since it exhibits a strong Ca 2+ -dependent change in fluorescence emission ratio and it does not dramatically affect plant growth (Zhao et al., 2011;Waadt et al., 2017). The emission ratio of RG-mT was calibrated to the cytosolic Ca 2+ concentration, using the fluorescent Ca 2+ -reporter dye FURA2 (see Fig. S2; Methods S1; Videos S3), and far-red light was used to monitor closure of the stomatal pore ( Fig. 2a-c; Videos S4-S6). Current ejection of ABA caused closure of the stomatal pore in all experiments (n = 41); however, the impact of ABA on the cytosolic Ca 2+ level split into three guard cell populations (Fig. 2a-c). In a first group By contrast to the experiments with ABA, current ejection of benzoic acid did not cause stomatal closure in any of the 24 experiments. Only in two out of 24 guard cells were transient changes in the cytosolic free Ca 2+ concentration observed, and these Ca 2+ signals had a smaller amplitude than those elicited by ABA (Fig. S3).

New Phytologist
Our data thus indicate that ABA-induced Ca 2+ signals can be clustered into two groups (Fig. 2a,b). In a small group, the Ca 2+ signals precede stomatal closure (Fig. 2b), whereas the Ca 2+ level rises during stomatal closure in the majority of guard cells (Fig. 2a). The occurrence of these groups was tested by fitting the frequency distribution of cells with Ca 2+ signals; the number of cells were plotted against the time interval between stimulation with ABA and occurrence of the peak in the cytosolic Ca 2+ -level (Fig. 2e). According to the corrected Akaike information criterion (Burnham et al., 2011), a model based on the sum of two Gaussian functions (solid line in Fig. 2e, R 2 = 0.99) was 55 times more likely than the model with one Gaussian function (dotted line in Fig. 2e, R 2 = 0.75), whereas a model based on the sums of three Gaussian functions was very unlikely (10 37 less likely as the sum of two Gaussian functions; striped line in Fig. 2e, R 2 = 0.99). This analysis thus strongly supports that ABA-induced Ca 2+ signals in guard cells occur in two time windows; some are elicited early (before the stomata start to close), whereas others are evoked later (during stomatal closure).
Elevated cytosolic Ca 2+ levels rapidly activate S-type anion channels Elevated cytosolic Ca 2+ levels were shown to activate plasma membrane anion channels in guard cells of several species (Allen et al., 1999;Chen et al., 2010;Stange et al., 2010), but it is unknown how fast this response occurs in Arabidopsis. We therefore studied this response in real time, with guard cells expressing www.newphytologist.com RG-mT. Guard cells were impaled with double-barreled electrodes, and cytosolic Ca 2+ concentration changes were evoked with voltage pulses (Grabov & Blatt, 1998;Voss et al., 2018) from À100 mV, stepwise for 10 s, to more negative membrane potentials (Fig. 3a,b). The cytosolic Ca 2+ level of the cell shown in Fig. 3(a,b) hardly changed in response to a 10 s pulse of À180 mV, but a transient rise in the Ca 2+ concentration was triggered by pulses to À200 and À220 mV (Videos S7). During the hyperpolarizing pulses, inward currents are facilitated by K + channels, which are voltage activated and deactivate at À100 mV in c. 0.5 s (Roelfsema & Prins, 1997). After termination of the voltage pulses in which Ca 2+ signals were elicited, an additional conductance was recorded (arrows in Fig. 3a), which transiently reached maximum conductance, at 13.8 s (SE = 0.9 s, n = 27) after the cytosolic Ca 2+ peak (Fig. 3a). It is likely that this slow current is facilitated by Ca 2+ -activated S-type anion channels, as was previously shown for tobacco guard cells (Chen et al., 2010;Stange et al., 2010). In Arabidopsis, guard cell S-type anion channels are encoded by SLAC1 and SLAH3 (Negi et al., 2008;Vahisalu et al., 2008;Guzel Deger et al., 2015), and the voltage responses of the slac1 and slah3 loss-of-function mutants were therefore compared with wild-type (Fig. 3c,d). In wild-type, 10 s pulses from À100 mV, stepwise to À180, À200, and À220 mV, induced inward currents that slowly deactivated after returning to À100 mV (Fig. 3c), just as in guard cells expressing RG-mT (Fig. 3a). In the slah3-1 single mutant, the hyperpolarizing pulses elicited currents that had a similar magnitude as in wild-type (Figs 3d, S4). However, these currents were only detected in six out of 10 slac1-3 guard cells, where, on average, they had a reduced magnitude (Figs 3d, S4). Finally, the loss of both SLAC1 and SLAH3 caused a complete lack of Ca 2+ -activated currents (Fig. 3c,d). These data thus strongly suggest that both SLAC1 and SLAH3 contribute to the Ca 2+ -activated conductance in Arabidopsis guard cells.

Loss of OST1 prevents ABA-induced stomatal closure and alters Ca 2+ signals
The protein kinase OST1 plays a central role in ABA-induced stomatal closure (Mustilli et al., 2002;Merilo et al., 2013;Guzel Deger et al., 2015). However, it is unclear how loss of OST1 affects Ca 2+ signals. Guard cells of ost1-3, expressing RG-mT, were therefore stimulated by current-ejection of ABA. In the majority of guard cells, ABA neither induced stomatal closure nor provoked a change in cytosolic free Ca 2+ level (Fig. 4a, 26 out of 31 cells; Videos S8). Despite the lack of stomatal closure, transient changes of the cytosolic free Ca 2+ concentration were observed in five out of 31 guard cells ( Fig. 4b; Videos S9). ABA Fig. 3 Cytosolic Ca 2+ signals rapidly activate SLAC1 and SLAH3 anion channels in Arabidopsis thaliana guard cells. (a) Arabidopsis guard cells were stimulated with 10 s pulses from a potential of À100 mV, to À180, À200, and À220 mV (upper trace). The voltage pulses to À200 and À220 mV evoked a transient increase of the cytosolic free Ca 2+ concentration (middle trace). The Ca 2+ signals caused activation of inward currents (lower trace) after returning the voltage to À100 mV, as indicated by the arrows below the current trace. (b) Pseudo-color images that represent the cytosolic free Ca 2+ concentration of the same guard cell as in (a) determined from the R-GECO1-mTurquoise signal. The images were acquired before, during, and after the stimulation of the cell with hyperpolarizing voltage pulses. The asterisk marks the position at which the right guard cell was impaled with a double-barreled electrode. Bar, 10 lm. The calibration bar above the images links the color code to [Ca 2+ ] cyt . See also Supporting Information Videos S7. (c) Guard cells were stimulated with voltage pulses from a holding potential of À100 mV, for 10 s to À180, À200, and À220 mV as indicated above the current traces. In wild-type guard cells these voltage pulses caused activation of anion channels (upper trace) that facilitate inward currents after returning the voltage to À100 mV, as indicated by the dotted lines. These anion currents were absent in guard cells of the slac1-3/slah3-1 double mutant (lower trace). (d) Average currents facilitated by S-type anion channels, as measured in (c), evoked by voltage pulses of À180, À200, and À220 mV in Col-0 wild-type (white bars), slah3-1 (gray bars; see also Fig. S4), slac1-3 (black bars; see also Fig. S4), and slac1/slah3 (striped bars, note that bars approximate 0 pA). Errors bars represent + SE (Col-0, n = 9; slah3-1, n = 8; slac1-3, n = 10; slac1/slah3, n = 13).

Research
New Phytologist thus triggered Ca 2+ signals in approximately one out of six ost1-3 guard cells, whereas it evoked Ca 2+ signals in three out of four guard cells of wild-type (Fig. 2). For comparison, current ejection of benzoic acid as control evoked only a Ca 2+ signal in one out of 21 ost1-3 guard cells (Fig. S5).
Cytosolic Ca 2+ signals trigger rapid activation of anion channels in ost1-3 OST1 is important for a variety of stomatal responses (Melotto et al., 2006;Xue et al., 2011;Merilo et al., 2013), but it is unclear to what extent it is necessary for Ca 2+ -dependent responses in guard cells. Guard cells of ost1-3, expressing RG-mT, were therefore stimulated with 10 s hyperpolarization pulses (Fig. 5a,b). Just as in wild-type, these pulses evoked a transient elevation of the cytosolic Ca 2+ level (Videos S10) and activated S-type anion channels, with a similar voltage dependence as in wild-type (Fig. 5c). The cytosolic Ca 2+ concentration changes were plotted against the currents carried by S-type anion channels at À100 mV in Fig. 5(d). For wild-type, a Hill equation was fitted to the data, which revealed that a 90 nM increase of the cytosolic Ca 2+ concentration led to a half-maximal response (Fig. 5d). The Hill equation did not converge to the data of ost1-3, but the number of cells in which large changes of the cytosolic Ca 2+ level occurred was higher in the mutant (Fig. 5d). Combined with the finding that the voltage pulses triggered S-type anion channel currents with a similar magnitude in ost1-3 and wild-type (Fig. 5c), this suggests that ost1-3 guard cells have a slightly lower Ca 2+ responsiveness, as wild-type.

Discussion
ABA evoked stomatal closure in Arabidopsis in the absence of Ca 2+ signals in one out of four stomata, whereas a transient rise in the Ca 2+ level was detected in three out of four experiments. These data are in line with early experiments with C. communis, in which ABA-dependent Ca 2+ signals were detected in eight out of 10 stomata (McAinsh et al., 1990) or in 14 out of 38 stomata (Gilroy et al., 1991). This suggests that ABA-induced Ca 2+ signals are common in guard cells, but not absolutely required for stomatal closure.
Ca 2+ signals occur in two phases of the guard cell ABA response The cytosolic Ca 2+ signals arose in two phases after stimulation of Arabidopsis guard cells with ABA (Fig. 2). In the majority of cells, the cytosolic Ca 2+ concentration increased transiently during the stage in which the stomata were closing. It is feasible that these Ca 2+ signals are provoked by the sudden changes in osmotic content of guard cells, which arise at the start of stomatal closure. Such a mechanism is supported by the finding that fast changes in the osmotic content of tobacco guard cells provoke Ca 2+ release from intracellular stores (Voss et al., 2016). This class of ABA-induced Ca 2+ signals will not occur in the ost1-3 mutant, as its stomata do not close in response to ABA, and thus osmotic changes in the cytosol are not evoked by the hormone. As a result, ABA-induced Ca 2+ signals are impaired in ost1-3 and only five out of 31 ost1-3 stomata showed changes in the cytosolic Ca 2+ level; all of which did not exceed 100 nM (Fig. 4).
ABA can also induce Ca 2+ signals that precede closure of the stomatal pore (Fig. 2b), which suggests that the hormone also stimulates Ca 2+ channels by a mechanism that does not depend on changes in osmotic pressure. This early response may explain why ABA can also trigger repetitive rises in the Ca 2+ concentration of Arabidopsis guard cells in isolated epidermal tissue (Allen et al., 1999(Allen et al., , 2001Kl€ usener et al., 2002;Islam et al., 2010). Note that in isolated epidermal tissues the ABA-induced stomatal closure response is less pronounced than in intact leaves (Islam et al., 2010), and osmotically induced Ca 2+ signals are therefore less likely to occur. This suggests that ABA evokes these early Ca 2+ www.newphytologist.com signals through a mechanism that is not dependent on OST1, but instead through stimulation of nonselective cation channels in the guard cell plasma membrane (Hamilton et al., 2000;Pei et al., 2000;Siegel et al., 2009).

Role of Ca 2+ signals in ABA-induced stomatal closure
ABA-induced stomatal closure is likely to involve a Ca 2+ -independent and -dependent signaling mechanisms. The initial Ca 2+independent step releases the protein kinase OST1 from inhibition (Cutler et al., 2010). In guard cells, OST1 will activate SLAC1, which leads to the release of anions from guard cells and provokes stomatal closure (Geiger et al., 2009;Lee et al., 2009). The Ca 2+ signals that can occur before or during stomatal closure probably enhance the activity of SLAC1 and also activate SLAH3, since these two anion channels are also activated by hyperpolarization-induced Ca 2+ signals (Fig. 3). Owing to a further stimulation of the S-type anion channels in guard cells, Ca 2+ signals seem to speed up stomatal closure (Fig. 2d). The Ca 2+ -dependent response is likely to be provoked by Ca 2+ -dependent protein kinases (CPKs; Harper et al., 1991;Geiger et al., 2010Geiger et al., , 2011Brandt et al., 2015) and calcineurin Blike (CBL)-interacting protein kinases (CIPKs) that bind to CBL proteins (Maierhofer et al., 2014;Kudla et al., 2018). Studies with CPK loss-of-function mutants support the function of these protein kinases in ABA-induced stomatal closure. In cpk8, cpk10, the cpk3/6 double, and the cpk5/6/11/23 quadruple mutants, ABA-induced stomatal closure was impaired in intact leaves that were floated on solution (Mori et al., 2006;Brandt et al., 2015;Zou et al., 2015). However, experiments with the cpk23 and Ca 2+ -dependent activation of S-type anion channels in ost1-3. (a) An Arabidopsis thaliana ost1-3 guard cell was stimulated with 10 s voltage pulses from a potential of À100 mV, to À180, À200 and À220 mV (upper trace). The voltage pulses evoked a transient increase of the cytosolic free Ca 2+ concentration (middle traces), which caused activation of S-type anion channels (lower trace) that facilitate inward currents after returning the voltage to À100 mV (arrows below the current trace). (b) Pseudo-color images that represent the cytosolic free Ca 2+ concentration of the same guard cell as in (a), determined from the R-GECO1-mTurquoise signal. The images were acquired before, during, and after application of the À200 mV voltage pulse. The asterisk marks the position at which the right guard cell was impaled with a double-barreled electrode. Bar, 10 lm. The calibration bar next to the images links the color code to [Ca 2+ ] cyt . See also Supporting Information Videos S10. (c) Average change in S-type anion channel current, recorded at À100 mV and induced by hyperpolarizing pulses to À180, À200, and À220 mV, in wild type (WT, white bars) and ost1-3 (black bars). Data are from experiments shown in Fig. 3(a)

Research
New Phytologist cpk4/5/6/11 mutants put this general role of CPKs in question. ABA could still induce stomatal closure in the cpk4/5/6/11 mutant (Guzel Deger et al., 2015), and loss of CPK23 even caused plants to become more tolerant to drought (Ma & Wu, 2007). Future studies will thus have to disclose which targets are addressed by individual CPKs and CIPKs and how these interactions contribute to the regulation of stomatal movements.
Cytosolic Ca 2+ signals regulate not only plasma membrane ion channels but also the vacuolar two-pore K + channels, which are important for stomatal closure (Gobert et al., 2007;Latz et al., 2013;Wang et al., 2015). As suggested by Wheeler & Brownlee (2008), the Ca 2+ signals may thus serve as a unifying signal that can coordinate transport processes between the plasma membrane and intracellular membranes. Such a coordinated response is likely to be important for rapid stomatal closure, in which osmolytes are first released from the vacuole into the cytosol and finally extruded across the plasma membrane into the apoplast (Wheeler & Brownlee, 2008;Kollist et al., 2014).

Future directions
In addition to the drought hormone ABA, stomata also respond to a variety of other signals, such as CO 2 , microbe-associated molecular patterns, and blue light. Previously, the associated Ca 2+ signals were studied in isolated epidermal tissues (Young et al., 2006;Harada & Shimazaki, 2009;Thor & Peiter, 2014), but new genetically encoded Ca 2+ sensors now enable experiments with intact leaves. Studies with these new sensors can reveal if guard cell Ca 2+ responses are stimulus specific, or if similar Ca 2+ signals are recorded, irrespective of the stimulus that induces stomatal closure.
The newly developed sensors will also be of great advantage to study the nature of Ca 2+ channels that give rise to Ca 2+ signals in guard cells. ABA-induced activation of Ca 2+ -permeable plasma membrane channels in guard cells was reported almost 20 yr ago (Hamilton et al., 2000;Pei et al., 2000), but the genes encoding these channels still need to be uncovered. The osmotically acitavated calcium channels, which are expressed in guard cells, have been associated with osmotically induced Ca 2+ signals (Yuan et al., 2014). These channels are thus good candidates for those that generate Ca 2+ signals during acceleration phase of ABA-induced stomatal closure.

Supporting Information
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.     Methods S1 Procedures to estimate the ABA concentration in the guard cell wall that was imposed by current-ejection and calibration of R-GECO1-mTurquoise with FURA2.
Videos S1 Stomatal closure induced by current-ejection of ABA.
Videos S2 Current-ejection of Lucifer Yellow CH (LY) into the wall of an Arabidopsis guard cell.
Videos S4 ABA-induced rise in the cytosolic Ca 2+ concentration of a guard cell, during stomatal closure.
Videos S5 ABA-induced rise in the cytosolic Ca 2+ concentration of a guard cell, before stomatal closure.
Videos S6 ABA-induced stomatal closure in the absence of a cytosolic Ca 2+ signal in the guard cells.
Videos S7 Voltage-induced Ca 2+ signals in an Arabidopsis guard cell.
Videos S9 ABA-induced Ca 2+ -signals in an ost1-3 stoma that were not linked to stomatal closure.
Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.