Abscisic acid and CO₂ signalling via calcium sensitivity priming in guard cells, new CDPK mutant phenotypes and a method for improved resolution of stomatal stimulus–response analyses

Abstract

Background

Stomatal guard cells are the regulators of gas exchange between plants and the atmosphere. Ca²⁺-dependent and Ca²⁺-independent mechanisms function in these responses. Key stomatal regulation mechanisms, including plasma membrane and vacuolar ion channels have been identified and are regulated by the free cytosolic Ca²⁺ concentration ([Ca²⁺]_cyt).

Scope

Here we show that CO₂-induced stomatal closing is strongly impaired under conditions that prevent intracellular Ca²⁺ elevations. Moreover, Ca²⁺ oscillation-induced stomatal closing is partially impaired in knock-out mutations in several guard cell-expressed Ca²⁺-dependent protein kinases (CDPKs) here, including the cpk4cpk11 double and cpk10 mutants; however, abscisic acid-regulated stomatal movements remain relatively intact in the cpk4cpk11 and cpk10 mutants. We further discuss diverse studies of Ca²⁺ signalling in guard cells, discuss apparent peculiarities, and pose novel open questions. The recently proposed Ca²⁺ sensitivity priming model could account for many of the findings in the field. Recent research shows that the stomatal closing stimuli abscisic acid and CO₂ enhance the sensitivity of stomatal closing mechanisms to intracellular Ca²⁺, which has been termed ‘calcium sensitivity priming’. The genome of the reference plant Arabidopsis thaliana encodes for over 250 Ca²⁺-sensing proteins, giving rise to the question, how can specificity in Ca²⁺ responses be achieved? Calcium sensitivity priming could provide a key mechanism contributing to specificity in eukaryotic Ca²⁺ signal transduction, a topic of central interest in cell signalling research. In this article we further propose an individual stomatal tracking method for improved analyses of stimulus-regulated stomatal movements in Arabidopsis guard cells that reduces noise and increases fidelity in stimulus-regulated stomatal aperture responses ( Box 1). This method is recommended for stomatal response research, in parallel to previously adopted blind analyses, due to the relatively small and diverse sizes of stomatal apertures in the reference plant Arabidopsis thaliana.

INTRODUCTION

Stomatal guard cells play a critical role in the regulation of plant gas exchange and water use efficiency (Kim et al., 2010). Several key physiological stimuli regulate stomatal aperture, including abscisic acid (ABA), CO₂ and pathogenic elicitors (Raschke et al., 1988; Pandey et al., 2007; Joshi-Saha et al., 2010; Kim et al., 2011). One signalling component that links these diverse stimuli is the cytosolic free-calcium concentration ([Ca²⁺]_cyt), which induces stomatal closure (DeSilva et al., 1985; Gilroy et al., 1990; McAinsh et al., 1990; Webb et al., 1996; Grabov and Blatt, 1998; Allen et al., 1999a; Staxén et al., 1999; Klüsener et al., 2002; Young et al., 2006). Several independent mechanisms have been identified that are activated by [Ca²⁺]_cyt in guard cells, and contribute to ABA-induced stomatal closing, including [Ca²⁺]_cyt activation of S-type anion channels (Schroeder and Hagiwara, 1989; Mori et al., 2006; Vahisalu et al., 2008; Siegel et al., 2009; Chen et al., 2010), [Ca²⁺]_cyt activation of R-type anion channels (Hedrich et al., 1990; Meyer et al., 2010), [Ca²⁺]_cyt down-regulation of plasma membrane proton pumps (Kinoshita et al., 1995), [Ca²⁺]_cyt down-regulation of K⁺ influx channels (Schroeder and Hagiwara, 1989; Kelly et al., 1995; Grabov and Blatt, 1997; Kwak et al., 2001) and Ca²⁺-activation of vacuolar K⁺ release (VK) channels (Ward and Schroeder, 1994; Allen and Sanders, 1996; Gobert et al., 2007).

These studies and subsequent studies have identified mechanisms through which plasma membrane and tonoplast ion channels and ATPases function as key mediators in a network that controls stomatal closing and inhibits stomatal opening (reviewed in MacRobbie, 1998; Schroeder et al., 2001; Pandey et al., 2007). The importance of [Ca²⁺]_cyt, relative to Ca²⁺-elevation-independent control of stomatal closure has recently been quantified in Arabidopsis showing that the Ca²⁺-dependent response is responsible for approx. 70 % of the ABA response (Siegel et al., 2009). This research is consistent with previous studies that demonstrated a requirement for Ca²⁺ in ABA-induced closure (DeSilva et al., 1985; Schwartz, 1985; McAinsh et al., 1990; Grabov and Blatt, 1998; MacRobbie, 2000). Recent ABA signalling models (Cutler et al., 2010; Hubbard et al., 2010; Melcher et al., 2010) have not yet incorporated a clear role for [Ca²⁺]_cyt. [Ca²⁺]_cyt elevation-independent ABA-induced stomatal closure accounted for approx. 30 % of the response (Siegel et al., 2009). Ca²⁺-independent signalling mechanisms function in ABA- and CO₂-induced stomatal closing and have been reviewed in depth elsewhere (Levchenko et al., 2005; Israelsson et al., 2006; Sirichandra et al., 2009; Kim et al., 2010; Hubbard et al., 2010; Raghavendra et al., 2010). In this study, we examine and review studies that have identified and analysed Ca²⁺-dependent stomatal closure and provide an overview of the mechanisms that have been established during [Ca²⁺]_cyt signalling in guard cells, and consider models of how specificity and plasticity in Ca²⁺-signalling networks may be achieved.

DYNAMICS IN ‘RESTING’ [Ca²⁺]_cyt AND SPONTANEOUS Ca²⁺ TRANSIENTS

Due to the cytotoxicity of large Ca²⁺ concentrations inside cells and diverse Ca²⁺-binding proteins expressed in cells, the free cytosolic Ca²⁺ concentration is maintained within the 150 nm range in unstimulated cells through the action of Ca²⁺-ATPases and Ca²⁺–H⁺ ion exchangers at the plasma membrane and membranes of internal stores. However, a number of studies have described fluctuations in [Ca²⁺]_cyt in guard cells, which had not yet been challenged with ABA or known stimuli (with the caveat that Ca²⁺ imaging experiments are performed under specific conditions which may inadvertently include physiological stimuli). For example, in Vicia faba guard cells, spontaneous increases in [Ca²⁺]_cyt were observed before the application of ABA (Grabov and Blatt, 1998). Similarly, spontaneous transients were noted in a subset of unstimulated Commelina communis cells (Staxén et al., 1999), and spontaneous [Ca²⁺]_cyt transients were observed in Arabidopsis guard cells expressing the non-invasive FRET reporter Yellow Cameleon with no exposure to ABA (Allen et al., 1999b; Klüsener et al., 2002). As with stimulus-induced [Ca²⁺]_cyt increases (see below), spontaneous [Ca²⁺]_cyt transients are not necessarily synchronized between guard cells of the same pair (Allen et al., 1999b). It is possible that [Ca²⁺]_cyt increases happen not at the level of the cell but at the level of [Ca²⁺]_cyt microdomains, as observed in mammalian tissue (Berridge, 2006). This may contribute to the highly stochastic nature of spontaneous [Ca²⁺]_cyt transients. Spontaneous Ca²⁺ transients have also been observed in guard cells of intact leaves of Arabidopsis plants (Yang et al., 2008). These Ca²⁺ transients can be inhibited by loading cells with high concentrations of Ca²⁺ chelators including the Ca²⁺ chelating reporter fura-2 and its derivative BAPTA (Young et al., 2006). Interestingly, these spontaneous Ca²⁺ transients can be inhibited by removal of extracellular Ca²⁺ or by buffering the extracellular free Ca²⁺ concentration to 200 nm (Young et al., 2006; Siegel et al., 2009) and can be dampened or removed by more positive (depolarized) membrane potentials in guard cells (Grabov and Blatt, 1998; Staxén et al., 1999; Klüsener et al., 2002; Young et al., 2006). Spontaneous [Ca²⁺]_cyt transients have been observed in guard cells from several species and using different [Ca²⁺]_cyt imaging techniques indicating that spontaneous transients in [Ca²⁺]_cyt are a conserved feature of guard cell [Ca²⁺]_cyt dynamics that are sensitive to external Ca²⁺ and cytoplasmically loaded Ca²⁺ buffer concentrations.

While the nature and function of spontaneous Ca²⁺ transients is currently poorly understood, it is likely that they include a contribution from Ca²⁺-permeable cation channels in the plasma membrane that are activated at increasingly negative (hyperpolarized) membrane voltages (Aharon et al., 1998; Hamilton et al., 2000; Pei et al., 2000). Driving membrane voltage to more negative (hyperpolarized) membrane potentials increased the amplitude of [Ca²⁺]_cyt transients (Grabov and Blatt, 1998) and increases the open probability of Ca²⁺-permeable cation channels in the plasma membrane of guard cells (Hamilton et al., 2000; Pei et al., 2000). Consistent with these findings, the proportion of guard cells with transients and the frequency of transients is dependent on the incubation buffer used. Low [K⁺]_ext (0·1 mm) results in more negative (hyperpolarized) membrane potentials and concomitant rapid transients in the majority of cells, while high [K⁺]_ext (100 mm) causes more positive (depolarized) membrane potentials and abolishes the Ca²⁺ transients (Klüsener et al., 2002). That some cells exhibit spontaneous transients while others in the same epidermal preparation have constant resting [Ca²⁺]_cyt levels may reflect a contribution from variation in the membrane potentials of the population of cells, with only those passing a certain threshold of negative voltages exhibiting spontaneous transients.

CO₂-INDUCED STOMATAL CLOSING IS STRONGLY IMPAIRED IN THE ABSENCE OF Ca²⁺ ELEVATIONS

The significance of [Ca²⁺]_cyt-dependent ABA-induced stomatal closure, relative to Ca²⁺-elevation-independent control of stomatal closure, has recently been quantified in Arabidopsis using the above described approaches that inhibit spontaneous and ABA-induced [Ca²⁺]_cyt increases. Guard cells where increases in [Ca²⁺]_cyt were prevented using the Ca²⁺-chelator BAPTA showed only 30 % of the ABA-induced stomatal closure response compared with guard cells where [Ca²⁺]_cyt was allowed to increase (Siegel et al., 2009). Under the same experimental conditions that inhibit [Ca²⁺]_cyt elevations (i.e. in the presence of BAPTA, buffering free Ca²⁺ concentration in the extracellular medium to 200 nm; Siegel et al., 2009), we analysed stomatal closing induced by elevation in the CO₂ concentration here. Interestingly, under these conditions, CO₂-induced stomatal closing was strongly inhibited (Fig. 1B), whereas addition of 50 µm Ca²⁺ to the bath medium was sufficient to ensure a robust CO₂-induced stomatal closing response (Fig. 1A). Thus CO₂-induced stomatal closing is strongly Ca²⁺ dependent in Arabidopsis, consistent with previous findings in Commelina guard cells (Schwartz, 1985; Webb et al., 1996). The Ca²⁺ requirement for CO₂-induced stomatal closing (Fig. 1) is particularly interesting because recent research has revealed that the Ca²⁺-independent protein kinase, OST1, is required for CO₂-induced stomatal closure in Arabidopsis (Xue et al., 2011). These data indicate that present models strictly separating Ca²⁺-dependent and Ca²⁺-independent stomatal closing may be oversimplified, and that these mechanisms may interact in an unknown manner and will require further research.

Fig. 1.

CO₂-induced stomatal closing is strongly inhibited under conditions that prevent [Ca²⁺]_cyt elevations in guard cells. (A) Robust CO₂-induced stomatal closing was observed in response to CO₂ elevation with 50 µm Ca²⁺ added to the bathing medium (10 mm KCl, 7·5 mm iminodeactic acid, 10 mm MES, pH 6·2). Stomatal closing of individual stomata was tracked (see Box 1 for Methods) in response to elevations in the extracellular CO₂ concentration from ambient (approx. 390 ppm) to 800 ppm (added at time = 0) with 50 µm Ca²⁺ in the bathing medium. (B) Buffering the extracellular free Ca²⁺ concentration to a low level of 200 nm using the Ca²⁺ chelator BAPTA strongly inhibited CO₂-induced stomatal closing. Extracellular bathing media and free Ca²⁺ buffering to 200 nm was performed using identical solutions to those reported in Siegel et al. (2009). Experiments were conducted on 4- to 5-week-old plants according to the protocols described in (Siegel et al., 2009). CO₂-induced stomatal closing experiments were performed under identical Ca²⁺ solutions to those published for ABA-induced stomatal closing experiments with 200 nm free Ca²⁺ or 50 µM Ca²⁺ in the bath medium (Siegel et al., 2009). Two independent experiments are shown for each condition (open and closed symbols); errors are presented as ± s.e. of the mean, n = 30.

STIMULUS-INDUCED CHANGES IN [Ca²⁺]_cyt

Several distinct stimuli are known to cause an increase in guard cell [Ca²⁺]_cyt, including cold (Allen et al., 2000; Wood et al., 2000), extracellular Ca²⁺ ([Ca²⁺]_ext) (McAinsh et al., 1995), ABA (McAinsh et al., 1990; Schroeder and Hagiwara, 1990; Grabov and Blatt, 1998; Allen et al., 1999a, b; Staxén et al., 1999; MacRobbie, 2000; Jung et al., 2002; Marten et al., 2007), CO₂ (Webb et al., 1996; Young et al., 2006) and certain pathogenic elicitors (Klüsener et al., 2002). In cells that show constant resting Ca²⁺ levels, upon application of a given stimulus, [Ca²⁺]_cyt can increase from approx. 100–150 nm to ≥350 nm. The pattern of [Ca²⁺]_cyt increases is related to the stimulus applied and the buffer in which cells are incubated. For example, for cells incubated in a high [K⁺] buffer (50 mm) a cold shock results in a single large transient (Fig. 2A), whereas exposure to high [Ca²⁺]_ext induces [Ca²⁺]_cyt oscillations (Fig. 2B).

Fig. 2.

Stimulus-induced increases in guard cell [Ca²⁺]_cyt. Plants transformed with the FRET-based Ca²⁺ sensor YC3·6, targeted to guard cells with the pGC1 guard cell promoter, were imaged (Allen et al., 1999b). Cells were initially incubated in a stomatal-opening buffer and then after 10 min of recording were challenged with (A) a cold shock (4 °C; n = 15), (B) 10 mm external CaCl₂ (n = 26) or (C–E) 10 µm ABA (n = 31) as indicated. External [Ca²⁺] elevation induces oscillations in [Ca²⁺]_cyt consistent with previous studies (McAinsh et al., 1995; Allen et al., 2000). The response to ABA was variable, with some cells showing no changes in [Ca²⁺]_cyt and others exhibiting oscillations in [Ca²⁺]_cyt in response to ABA under the imposed conditions (Allen et al., 1999b; Hugouvieux et al., 2001; Jung et al., 2002; Kwak et al., 2003), as discussed in the text. All traces are expressed as the ratio of recorded fluorescence emitted at 535 nm divided by fluorescence emitted at 480 nm after excitation at 440 nm as described in Allen et al. (1999b). Daily misting of plants with water was used to reduce endogenous ABA concentrations and increased the percentage of guard cells that showed ABA-induced [Ca²⁺]_cyt elevation, as reported previously (Allen et al., 1999a,b). Epidermal peels were prepared from 3- to 5-week-old plants stably transformed with the cameleon construct according to the method described in Allen et al. (1999b). Samples were incubated in either a high-[K⁺] buffer (50 mm K⁺, 10 mm MES-Tris, pH 6·2) for cold and the indicated CaCl₂ treatment (Fig. 2A, B) or stomatal-opening buffer (5 mm K⁺, 50 µm CaCl₂, 10 mm MES-Tris, pH 6·2; ABA) for 3 h before imaging (C–E). Imaging was performed as described in Allen et al. (1999b). Cells were treated with either a cold shock (4 °C; n = 15), 10 mm CaCl₂ (n = 26) or 10 µm ABA (dissolved in methanol). All traces were corrected for YFP bleaching through subtraction of the overall gradual decline in fluorescence in each channel from each data point (Allen et al., 1999b; Klüsener et al., 2002).

Physiological stimuli can also alter the pattern of spontaneous [Ca²⁺]_cyt transients; for example, transfer from a low to a high [CO₂] buffer reduces the rate of spontaneous Ca²⁺ transients in wild-type Arabidopsis (Landsberg erecta) guard cells (Young et al., 2006). This elevated CO₂-induced dampening of Ca²⁺ transients was proposed to occur due to depolarization of the guard cell plasma membrane mediated by CO₂ signalling (Young et al., 2006). Both types of responses (stimulation and dampening) have been associated with exposure of cells to ABA, with some studies showing ABA-induced [Ca²⁺]_cyt transients (Schroeder and Hagiwara, 1990; Grabov and Blatt, 1998; Allen et al., 1999a, 2000, 2002; Staxén et al., 1999; Islam et al., 2010). However, in cells that exhibit spontaneous Ca²⁺ transients, prior to ABA application, ABA-dependent dampening and slowing of spontaneous [Ca²⁺]_cyt transients was found (Grabov and Blatt, 1998; Staxén et al., 1999; Klüsener et al., 2002). A number of studies have noted a variety of ABA-induced responses ranging from repetitive oscillations to no measureable ABA-dependent change in [Ca²⁺]_cyt (Schroeder and Hagiwara, 1990; Gilroy et al., 1991; Allen et al., 1999a; Hugouvieux et al., 2001; Jung et al., 2002; Kwak et al., 2002, 2003; Levchenko et al., 2005). These studies and data in Fig. 2 indicate that amongst a population of guard cells different microenvironments or signalling states may exist resulting in either lack of ABA-induced [Ca²⁺]_cyt increases (Fig. 2C) or ABA-induced [Ca²⁺]_cyt increases (Fig. 2D, E). Note that plants analysed in Fig. 2C–D were grown at high humidity, and were misted daily with water, similar to previous reports (Allen et al., 1999a; see methods in the figure captions). Daily misting of plant leaves with water may serve to reduce endogenous ABA levels in guard cells, and appears to enhance mechanisms mediating ABA-induced [Ca²⁺]_cyt elevations.

Mechanistically, evidence suggests that [Ca²⁺]_cyt increases induced by separate stimuli are likely generated by a different combination of mechanisms. [Ca²⁺]_cyt oscillations recorded in response to [Ca²⁺]_ext, cold and ABA have distinct patterns (Allen et al., 2000), indicating their generation is differently regulated. To generate a specific [Ca²⁺]_cyt pattern, such as a repetitive oscillation, requires the co-ordinated action of both Ca²⁺ influx channels and active Ca²⁺ efflux transporters. Sorting out the specific genes encoding candidate Ca²⁺-permeable ion channels and efflux transporters that mediate specific [Ca²⁺]_cyt patterns will require further research. This question is likely to be complicated by the circumstances that (a) large families of candidate Ca²⁺-permeable ion channels exist in plants with overlapping gene functions (Lacombe et al., 2001; Kaplan et al., 2007) and (b) more than one Ca²⁺ channel type is likely to contribute to any given stimulus-induced Ca²⁺ elevation pattern.

[Ca²⁺]_cyt-REACTIVE MECHANISMS

There are a large number of [Ca²⁺]_cyt-dependent mechanisms that have been described within the guard-cell signalling network, most of which function in mediation of stomatal closing, leading to the proposal that [Ca²⁺]_cyt functions as a ‘hub’ within the stomatal signalling network (Hetherington and Woodward, 2003). The diversity of [Ca²⁺]_cyt-dependent processes is illustrated by the large number (approx. 250) of proteins containing Ca²⁺-binding ‘EF-hand’ motifs encoded within the Arabidopsis thaliana genome (Day et al., 2002). This is likely to be an underestimate of the total number of Ca²⁺-binding proteins present, as some proteins such as the 14-3-3 protein GF14 have been described as Ca²⁺-binding (Lu et al., 1994; Athwal and Huber, 2002) but do not contain an EF hand (Day et al., 2002). EF-hand-containing proteins include several families of signalling components commonly referred to as Ca²⁺-sensors which include calmodulins (CaM) and CaM-like (CML) proteins, calcineurin-B-like (CBL) proteins and Ca²⁺-dependent protein kinases (CPKs); these protein families have been extensively reviewed elsewhere (Dodd et al., 2010; Kudla et al., 2010) and so will not be discussed in detail here. Moreover, additional ‘target proteins’ including ion channels contain EF-hand(s), and can be directly regulated by [Ca²⁺]_cyt. For example, the K⁺- and Ca²⁺-permeable two-pore channel 1 (TPC1) protein which encodes the slow vacuolar (SV) channel (Peiter et al., 2005) contains two EF hands and is Ca²⁺ activated (Hedrich and Neher, 1987; Pei et al., 1999).

The literature might in a few cases over-interpret findings as mainly supporting a model in which only unique stimulus-specific patterns in [Ca²⁺]_cyt elevations, rather than Ca²⁺ elevations above thresholds in concentrations and durations, are required in plant cells for Ca²⁺ to elicit an output. (We have avoided the attractive term ‘Ca²⁺ signature’ in our previous studies, as this can have several definitions. Nevertheless, further research in plant model cell systems is needed to elucidate the underlying cell-signalling and biochemical mechanisms, which would represent important breakthroughs in plant Ca²⁺ signalling.) Experimental imposition of [Ca²⁺]_cyt transients through repetitive buffer swap experiments showed that any Ca²⁺ elevation above a threshold value could trigger a rapid ‘Ca²⁺-reactive’ stomatal closing response [see supplemental data for Allen et al. (2001)]. Thus the Ca²⁺-reactive stomatal closing response was shown to be largely independent of the [Ca²⁺]_cyt elevation frequency, number or duration of transients (Allen et al., 2001). Nevertheless, the rate and the final degree of Ca²⁺-reactive stomatal closing depends on the number and integrated amplitude of Ca²⁺ elevations, which per se could also be modelled by a threshold-type mechanism. S-type anion channels in the plasma membrane of guard cells play a central role in controlling stomatal closure (Schroeder and Hagiwara, 1989; Schroeder, 1995). The rapid Ca²⁺-reactive phase of stomatal closing in response to imposed Ca²⁺ transients is reduced in the Ca²⁺-dependent protein kinase double mutant cpk3cpk6 (Mori et al., 2006; see Fig. 4A). Moreover, mutations that disrupt the gene encoding the S-type anion channel, SLAC1 (Negi et al., 2008; Vahisalu et al., 2008), show a dramatically impaired Ca²⁺-reactive stomatal closure (Vahisalu et al., 2008). These findings are consistent with the model that Ca²⁺-dependent protein kinases activate S-type anion channels, thus mediating stomatal closing (Mori et al., 2006; Geiger et al., 2010).

In contrast to Ca²⁺-reactive stomatal closing, imposed Ca²⁺ oscillations of particular frequencies and durations resulted in inhibition of stomatal re-opening, after Ca²⁺ transients had been terminated (Allen et al., 2001; Li et al., 2004). This latter Ca²⁺-dependent inhibition of stomatal re-opening has been coined long-term ‘Ca²⁺-programmed’ stomatal closure (Allen et al., 2001; Yang et al., 2003; Li et al., 2004). The gca2 mutant exhibits more ‘spiky’ Ca²⁺ oscillations during ABA and elevated CO₂ exposures and gca2 exhibits impaired stomatal closure responses (Allen et al., 2001; Young et al., 2006). In the gca2 mutant, experimental imposition of strong Ca²⁺ oscillations restored 65 % of stomatal closure compared with wild-type guard cells (Allen et al., 2001). Further research suggested that the increased rate of [Ca²⁺]_cyt transients observed in gca2 mutant guard cells (Allen et al., 2001) could result from the more negative (hyperpolarized) membrane potential of this mutant (Young et al., 2006). Indeed membrane potential hyperpolarization enhances the rate of [Ca²⁺]_cyt transients in guard cells, as discussed earlier (Grabov and Blatt, 1998; Klüsener et al., 2002).

CALCIUM SENSITIVITY PRIMING

Given that both spontaneous transients and stimulus-specific changes in [Ca²⁺]_cyt have been reported in guard cells, the question remains; how are guard cells able to respond in a Ca²⁺-dependent manner? In other words why can Ca²⁺ alter, for example, anion channel activity in response to the appropriate stimulus, but not in response to spontaneous Ca²⁺ transients? There is increasing evidence to suggest that the prior conditions that guard cells have experienced modulate the sensitivity of the signalling network to increases in [Ca²⁺]_cyt, thus determining the final response of the cell. For example, [Ca²⁺]_cyt-induced activation of the S-type anion channels was found not to occur in Arabidopsis guard cells under non-stimulated conditions (see fig. 3 in Allen et al., 2002). Exposure to high extracellular Ca²⁺ concentrations, and even after a subsequent return to more physiological extracellular Ca²⁺ levels, was found to alter the signalling state of guard cells, such that [Ca²⁺]_cyt activation of anion channels was restored (Allen et al., 2002; Mori et al., 2006; Suh et al., 2007). These data along with the resolution of spontaneous Ca²⁺ transients in guard cells provided first evidence that the [Ca²⁺]_cyt sensitivity of stomatal closing mechanisms can be up- and down-regulated (i.e. primed or de-primed).

Ensuing studies showed that [Ca²⁺]_cyt-activation of S-type anion channels is primed by pre-exposure of guard cells to ABA (Siegel et al., 2009; Chen et al., 2010). Similarly, K⁺-influx channels in Arabidopsis guard cells are rendered more [Ca²⁺]_cyt sensitive by pre-exposure of guard cells to ABA as well (Siegel et al., 2009). These findings and Ca²⁺ imaging studies led to the proposal of the ‘Ca²⁺-sensitivity priming’ hypothesis (Young et al., 2006; Siegel et al., 2009), which postulates that a given physiological stimulus renders a component(s) of the signalling network more [Ca²⁺]_cyt responsive (Fig. 3). Thus, for example, in the primed state a lower amplitude Ca²⁺ transient would cause a stronger activation of anion channels (Chen et al., 2010). This model resolves some of the potential paradoxes found in the literature.

Fig. 3.

Ca²⁺-sensitivity priming. In the Ca²⁺-sensitivity priming model, the response of a cell to a given stimulus depends on the conditions the cell has previously experienced. A generic model is presented, where the magnitude of Ca²⁺ increases and ABA-dependent priming combine to determine the eventual response (Siegel et al., 2009; Chen et al., 2010). Without a specific stimulus only spontaneous [Ca²⁺]_cyt transients occur. In the presence of a stimulus (X) the amplitude of transients increases, but in the un-primed state this is insufficient to trigger the downstream response (e.g. ion-channel activation). A previously encountered condition (e.g. ABA) may modulate a Ca²⁺-response pathway, making it more Ca²⁺ responsive or ‘primed’. When stimulus X is perceived under the primed conditions the sensor is more Ca²⁺ responsive and therefore downstream responses are triggered. Note, however, that [Ca²⁺]_cyt sensitivity enhancement could theoretically occur at any point in the network downstream of or parallel to [Ca²⁺]_cyt. See text for details.

An important question for Ca²⁺ sensitivity priming remains, whether ABA or CO₂ can sensitize guard cells to typical baseline resting [Ca²⁺]_cyt concentrations of 150 nm in vivo. Some circumstantial evidence exists that this may be the case. For example, in Vicia faba and other species, guard cells have been previously shown to exhibit ABA-induced Ca²⁺ increases in only a minority of guard cells (Schroeder and Hagiwara, 1990; Gilroy et al., 1991; McAinsh et al., 1992). Furthermore, loading the Ca²⁺ reporter fura-2 into Vicia faba guard cells was shown to enable ABA-induced stomatal closing at resting [Ca²⁺]_cyt levels in the absence of measurable [Ca²⁺]_cyt elevations (Levchenko et al., 2005). However, in the same experiments loading a higher concentration of the fura-2-analogous Ca²⁺ chelator BAPTA into guard cells reduced resting Ca²⁺ levels to below typical baseline levels and then inhibited ABA-induced stomatal closing (Levchenko et al., 2005). Thus, ABA may enhance (prime) the sensitivity of stomatal closing mechanisms even to resting Ca²⁺ levels. This gives rise to the question whether Ca²⁺-independent signalling in vivo (Fig. 2C) is truly Ca²⁺ independent. More research is needed to investigate this question.

There are many potential (non-mutually exclusive) mechanisms through which Ca²⁺-sensitivity priming may occur, including post-translational modification of signalling proteins, sub-cellular re-localization, protein–protein interactions, coincidence detection of Ca²⁺ and a second parallel signal, interaction of parallel Ca²⁺-dependent and -independent signalling pathways and transcriptional reprogramming (for reviews and discussion of putative mechanisms, see Hubbard et al., 2010; Kim et al., 2010). These events may occur at the level of the Ca²⁺-sensors, but may potentially play a role at any point of the signalling network downstream of or parallel to [Ca²⁺]_cyt responses.

Recent research has shown that elevation of the intracellular CO₂ and bicarbonate concentrations in Arabidopsis guard cell protoplasts rapidly enhances the ability of [Ca²⁺]_cyt to activate S-type anion channels, within approx. 3–5 min of whole-cell patch clamp initiation (Xue et al., 2011). These findings demonstrate that Ca²⁺-sensitivity priming can occur rapidly in guard cells, and may therefore not underlie a slower transcriptional regulation mechanism (Xue et al., 2011). The Ca²⁺-dependent protein kinase CPK1 of Mesembryanthemum crystallinum is dynamically re-localized within cells depending on growth conditions; upon transfer to low humidity the kinase moves from the plasma membrane to soluble and nuclear cell fractions, where it is presumably able to activate downstream signalling processes (Chehab et al., 2004). Further research is needed to elucidate the unknown biochemical and cell biological mechanisms by which the stomatal closing stimuli CO₂ and ABA enhance the sensitivity of Ca²⁺-dependent stomatal closing (Young et al., 2006; Siegel et al., 2009; Chen et al., 2010; Xue et al., 2011).

MODULAR BASIS OF [Ca²⁺]_cyt SIGNALLING

Plants contain a diverse array of proteins associated with Ca²⁺ signalling. For [Ca²⁺]_cyt-signals to be generated and interpreted with specificity, a specific subset of these components must be employed at a given time and location. The concept of the ‘signalosome’ has previously been proposed for mammalian Ca²⁺ signalling (Berridge et al., 2003), and is equally applicable for plant systems, particularly given the large number of Ca²⁺-binding site-containing proteins encoded in plant genomes. The components of the signalosome may be tissue-, cell- or subcellular location-dependent, and may also vary with time, either in response to endogenous cues such as those provided by the circadian clock, or in response to particular stimuli. On a technical level, variation in growth conditions, age and physiological conditions of individual plants may influence the composition of the guard cell signalosome and therefore the sensitivity of stomata to different stimuli.

Evidence for specific control of subcellular location of Ca²⁺-signalling components has been elegantly demonstrated through transient expression analysis for the CBL–CIPK network, where particular CBL–CIPK pairs are restricted to defined sub-cellular locations (Waadt et al., 2008; Batistič et al., 2010). CIPK14 is located at the tonoplast when in complex with CBL2 or CBL3, but is found at the plasma membrane when co-expressed with CBL8 (Batistič et al., 2010). It would therefore be predicted for the intact plant that expression of the individual Ca²⁺-sensor would affect downstream signalling specificity.

Fig. 4.

Imposed Ca²⁺ oscillation-induced stomatal closure is impaired in cpk3,cpk6, cpk10, cpk4,cpk11 and cpk32,cpk7,cpk8 mutants compared with wild-type stomata. Samples were pre-incubated in depolarizing stomatal-opening buffer for 3 h (50 mm KCl and 10 mm MES-Tris at pH = 5·6). Stomatal apertures of individually tracked stomata (see Box 1) were measured periodically at the indicated times. Starting at time = 0 and for the first 40 min the sample was exposed to hyperpolarizing (1 mm KCl, 1 mm CaCl₂ and 10 mm MES-Tris at pH 5·6) solution four times for 5 min each (Allen et al., 2001; Mori et al., 2006). For experiments illustrated in Figs 4–6 plants were tented using transparent plastic wrap to maintain a high humidity (>95 %) to reduce endogenous ABA levels. After 2 weeks of growth, slits were cut in the plastic tent to allow enhanced air circulation. Twenty-four hours before each experiment, the old plastic was replaced with a new plastic wrap for a high humidity treatment. Stomatal movement analyses were performed following the imposed Ca²⁺ oscillation protocol and solutions described in Mori et al. (2006). The lower epidermis of Arabidopsis leaves was attached to glass cover slides using Hollister medical adhesive 9 to attach a leaf abaxial side down onto a coverslip. A razor blade was then used to carefully remove mesophyll layers of the leaf until only the epidermal layer remained. The coverslip was then incubated in 4 mL depolarizing buffer (50 mm KCl and 10 mm MES-Tris at pH 5·6) in a Petri dish (35 × 10 mm) for 3 h under white light (200 µE) to open stomata. The coverslip with epidermal layer attached was then sealed to a glass slide with a 2-cm hole drilled in the middle to create a 200 µL perfusion well. Depolarizing buffer was alternated with hyperpolarizing buffer (1 mm KCl, 1 mm CaCl₂, and 10 mm MES-Tris at pH 5·6) using a bath perfusion system. Four 5-min extracellular Ca²⁺ pulses were applied in 10-min intervals in the first 40 min, as described previously (Mori et al., 2006). In between each transient and after the final transient depolarizing buffer was continuously applied. Stomatal apertures were measured using an inverted microscope at ×40. Both mutant and wild-type plants could be placed side by side on one coverslip during a single experiment. Eight individual stomata per condition were measured and tracked using Scion Image Software. Experiments were repeated at least three times for each mutant and parallel wild-type controls. Seed sources were cpk3-1 cpk6-1 (Mori et al., 2006), cpk10 (Salk_032021) and cpk32 (GABI 824E2). For information on cpk4 and cpk11 sources see Fig. 7. Experiments were performed single-blinded with the genotype identity unknown. Data are from Valerio (2007).

ANALYSIS OF Ca²⁺-DEPENDENT PROTEIN KINASE FUNCTIONS IN STOMATAL REGULATION

Microarray analysis provides evidence that Ca²⁺-signalling components are expressed in guard cells, e.g. CPK3, 6, 4, 10 and 11. Additional CPKs are expressed in guard cells, whereas CPK2, 18 and 34 (for example) are only very weakly expressed in this cell type (Leonhardt et al., 2004; Mori et al., 2006; Zhu et al., 2007; Yang et al., 2008; Zou et al., 2010). Interestingly, most of the above-mentioned Ca²⁺-dependent protein kinases have been reported to contribute to ABA-induced stomatal closing (Mori et al., 2006; Zhu et al., 2007; Zou et al., 2010). Here, we show that individual mutants in these same genes show partial insensitivities in the rapid Ca²⁺-reactive stomatal closing response, as shown for the cpk3cpk6, cpk10, cpk4cpk11 and cpk7cpk8cpk32 mutants (Fig. 4; Valerio, 2007). However, mutants in these Ca²⁺-dependent protein kinases (CDPKs), including cpk10, and even cpk4cpk11 double-mutant stomata and cpk7cpk8cpk32 triple-mutant stomata did not show a clear ABA-insensitive phenotype in our analyses of ABA-induced stomatal closing (Fig. 5) and ABA-inhibition of stomatal opening (Fig. 6; Valerio, 2007). These findings with a cpk4cpk11 double mutant (Figs 5 and 7; Valerio, 2007) do not correlate with a very strong ABA insensitivity reported for cpk4cpk11 (Zhu et al., 2007). Note that the cpk4 allele (cpk4-2; SALK_000685) in the double-mutant plants used for these experiments differs from the ones used in a previous study (Zhu et al., 2007; cpk4-1; SALK_081860), whereas the same cpk11-2 allele was used. The T-DNA insertion of the cpk4-2 allele analysed here was identified as lying in the 6th exon and RT-PCR analysis revealed that plants harbouring cpk4-2 and cpk11-2 alleles do not express mature transcripts (Fig. 7). These findings showing statistically functional ABA responses but slightly impaired Ca²⁺-reactive stomatal closing responses in the analysed cpk mutants (Valerio, 2007; Figs 4–6) point to the hypothesis that combinations of these mutants may lead to a stronger ABA insensitivity, as found for the partial ABA insensitivity of cpk3cpk6 double mutants (Mori et al., 2006). Furthermore, CPK23 is also expressed in guard cells, but functions genetically as a negative regulator in drought stress experiments as well as control of stomatal aperture (Ma and Wu, 2007). Contrary to the phenotype of cpk23 mutants, CPK23 on the other hand activates the SLAC1 anion channel in Xenopus oocytes and S-type anion currents in guard cell protoplasts of cpk23 mutants are reduced (Geiger et al., 2010). The in vivo mechanisms mediating these opposing responses will require further analysis. The closest homologue to CPK23, CPK21, can also mediate activation of the SLAH3 anion channel and to a smaller extent SLAC1 anion channels in Xenopus oocytes (Geiger et al., 2011). SLAH3 shows a relative nitrate permeability similar to S-type anion currents in Vicia faba guard cells (Schmidt and Schroeder, 1994; Geiger et al., 2011).

Fig. 5.

ABA-induced stomatal closure was not dramatically impaired in cpk10, cpk4cpk11 and cpk32cpk7cpk8 compared with Col-1 wild type. Whole leaves were pre-incubated in stomatal-opening buffer (10 mm KCl, 7·5 mm iminodeactic acid, 10 mm MES, pH 6·2) for 2 h and then incubated for an additional hour with 0, 1 or 10 µm ABA. Stomatal aperture (A) and ratio (width : length) (B) were then measured. In (C) the average normalized stomatal apertures from (A) are shown. Experiments were performed double blind with both genotype and ABA concentration unknown; n = 3 experiments with 30 stomata analysed per condition in each experiment (90 stomatal apertures analysed per bar in the graphs). Error bars represent ± s.e. of the mean relative to n = 3 experiments. Plants were grown at approx. 95 % humidity and were subjected to 24 h of high (>98 %) humidity prior to experimentation. ABA-dependent stomatal movement imaging analyses (Valerio, 2007) were performed following the protocol from Kwak et al. (2002). Rosette leaves from 4- to 5-week-old plants were excised and the curvature of each leaf was gently inverted using a finger so that the abaxial side did not form any air pockets when floating on solution during incubation. Using forceps the whole leaf was placed with the abaxial side down in 4 mL of stomatal-opening buffer (10 mm KCl, 7·5 mm iminodeactic acid, 10 mm MES, pH 6·2) in a Petri dish (35 × 10 mm). Leaves were incubated in stomatal-opening buffer for 2 h under white light (200 µE). Then, the stomatal-opening buffer was removed via a pipette and replaced with 4 mL stomatal-opening buffer containing 0 µm, 1 µm or 10 µm ABA and the leaves were incubated under white light for one more hour. The abaxial epidermis of each leaf was peeled using forceps and placed on a glass coverslip with approx. 200 µL of the treatment buffer. Experiments were performed double-blinded. Both genotype identity and ABA concentrations were unknown to the experimenter. Data are from Valerio (2007).

Fig. 7.

Characterization of cpk4-2 cpk11-2 mutant plants. (A) Cartoon showing genomic DNA of CPK4 and CPK11 and the T-DNA insertion sites of the corresponding mutant alleles. White boxes represent exons. Arrowheads show primer locations used for RT-PCR experiments shown in (B). The T-DNA insertion site of the cpk4-2 allele was localized in the 6th exon at nucleotide 2184 after ATG by sequencing. In the cpk4-2 mutant, 11 nucleotides (2185–2196) are deleted. The T-DNA insertion of cpk11-2 in the first exon corresponds to that published by Zhu et al. (2007) and was confirmed. (B) RT-PCR analysis of complementary DNA prepared from mutant and wild-type plant RNA revealed that in both, cpk4-2 and the cpk11-2 mutant plants no transcripts were detected (35 PCR cycles). Loading control was Actin2 (23 PCR cycles).

Fig. 6.

ABA inhibition of stomatal opening is not affected in cpk10, cpk4cpk11 and cpk32cpk7cpk8 compared with Col-1 wild type. Stomatal apertures (A) and ratios (width : length) (B) were measured. In (C) the averages of normalized apertures are shown. Experiments were performed double blind with both genotype and ABA concentration unknown to the experimenter; n = 3 experiments with 30 stomata per condition per experiment (90 stomatal apertures analysed per bar in graph). Error bars represent ± standard error of the mean for n = 3 experiments. For analyses of ABA inhibition of stomatal opening, plants were subjected for 24 h to darkness to reduce stomatal apertures. Whole leaves were then pre-incubated in stomatal-opening buffer with 0, 1 or 10 µm ABA concentrations for 2 h. The epidermis was then peeled off and attached onto a glass slide. Stomatal apertures were measured using a microscope at ×40. Thirty stomata per condition were measured in each epidermal peel using Scion Image Software. Data are from Valerio (2007).

An aspect of signalosome models is that protein complexes are dynamically remodelled; and this could include a potentially [Ca²⁺]_cyt-dependent remodelling (Berridge et al., 2003). For example, adding a continuous high extracellular Ca²⁺ concentration to Arabidopsis leaf epidermis causes [Ca²⁺]_cyt elevation (McAinsh et al., 1995; Allen et al., 1999a), but under the appropriate conditions does not cause stomatal closing, in spite of the [Ca²⁺]_cyt elevation, indicating a non-primed state (K. E. Hubbard et al., unpubl. res.). However, repetitively pulsing extracellular Ca²⁺ to the same leaf epidermis causes a clear Ca²⁺-reactive stomatal closing response (Allen et al., 2001; e.g. Fig. 4). These findings suggest that this later Ca²⁺ treatment of repetitive extracellular Ca²⁺ pulses causes a Ca²⁺-primed state compared with simple continuous elevation in extracellular Ca²⁺.

Such remodelling can establish feedback mechanisms (Berridge et al., 2003). It is entirely possible that an analogous process contributes to [Ca²⁺]_cyt sensitivity priming in guard cell signalling. Many Arabidopsis genes are transcriptionally regulated in response to [Ca²⁺]_cyt, which include transcripts associated with Ca²⁺-signalling such as CML24/TCH2 (Braam, 1992; Kaplan et al., 2006) and CPK32 (Kaplan et al., 2006). Similarly, many Ca²⁺-signalling components are regulated at the level of transcript abundance by ABA treatment (Hoth et al., 2002; Sánchez et al., 2004; Zeller et al., 2009). Therefore the potential exists for similar [Ca²⁺]_cyt-dependent and conditional remodelling of the guard cell signalosome, which should be further investigated. Note, however, that CO₂/HCO₃-induced priming of Ca²⁺-dependent activation of anion channels occurs within <3 to 5 min in whole patch clamp experiments, demonstrating that relatively rapid events function in priming Ca²⁺sensitivity (Xue et al., 2011). Similarly the repetitive Ca²⁺ pulse-induced Ca²⁺-reactive stomatal closure is measurable within approx. 5 min of Ca²⁺ pulsing as resolved using tracking of individual stomatal apertures (see Box 1, overleaf).

Box 1. Improved resolution of stimulus-induced stomatal movements in guard cells by tracking of individual stomatal apertures

Arabidopsis guard cells have become a prime model system for analysing signal transduction, since early research combining genetic and ion channel analyses in this system (Ichida et al., 1997; Pei et al., 1997, 1998; Roelfsema and Prins, 1997). Arabidopsis stomata are small relative to other stomatal model systems and stomatal apertures of various plant types including Arabidopsis are known to show variability in the size of individual stomatal complexes and also variability in the opening apertures of stomata of similar size in a given leaf (Gorton et al., 1988; Mott and Buckley, 2000; Mott and Peak, 2007). Thus stomatal aperture measurements are expected to show a clear degree of statistical variation. Use of blind experiments, in which the genotype and, when possible, the stimulus being applied to guard cells is unknown to the experimenter (Murata et al., 2001) has been employed by several laboratories, has become a standard in the field and has aided in addressing the above limitations of the range of stomatal aperture sizes found under any given condition.

Research in our laboratory has shown that a major additional improvement in experiments can be made, by adding imaging of the same individual stomatal apertures over time (Allen et al., 2001; Mori et al., 2006; Vahisalu et al., 2008; Siegel et al., 2009), while performing blind experiments. In such ‘stomatal tracking’ experiments the lower side of a leaf is attached to a glass coverslip in an extracellular incubation medium (Webb et al., 2001; Young et al., 2006). The mesophyll and upper leaf epidermis are removed surgically for better optical resolution of stomatal apertures in the intact lower leaf epidermis (Young et al., 2006). For stimulus-induced stomatal closing analyses, a field of well-opened stomata is located and images are captured (e.g. using Scion Image software) for later analyses and data storage. The bottom (dry side) of coverslips can be marked with colour marker pens to label grids in the regions where apertures where imaged, for finding these same stomata subsequently if needed. Images of the same stomatal apertures are taken over time and can be stored for later analyses of individual stomatal apertures and for deposition of image files. While this approach has been used as a standard for imposed Ca²⁺ oscillation studies (Allen et al., 2001; Mori et al., 2006; Vahisalu et al., 2008; Fig. 4), we have found that this method also substantially improves stomatal movement response analyses to any given stimulus (Siegel et al., 2009; see Figs 1 and 4 and, Box Fig. 1). For example, while individual stomata are known to have diverse apertures (e.g. Box Fig. 1C), the relative responses of wide open stomata and smaller stomatal apertures to ABA or to CO₂ were comparable (Fig. 1 and Box Fig. 1; Siegel et al., 2009). Note that this method has previously been proposed and used in Vicia faba (Gorton et al., 1988), for which stomata exhibit relatively weak ABA and CO₂ responses, compared with, for example, Arabidopsis. We propose that this simple image-capturing approach, together with blind analyses, be used as a standard for stomatal response research in arabidopsis. Our research experience with this method shows that this approach will aid in greatly improving resolution and robustness and in defining the functions of individual Ca²⁺-independent and Ca²⁺-dependent components and mechanisms in stomatal response analyses.

Box Fig. 1.

ABA-induced stomatal closing of individually tracked stomatal apertures. (A) Average individually tracked stomatal apertures in the presence of 50 µm Ca²⁺ (open triangles) and in the presence of 200 nm free Ca²⁺ (open squares) in the bath solution from three experiments are shown and were normalized to the stomatal apertures at time = 0. (B, C) ABA-induced stomatal closing in the presence of 50 µm Ca²⁺ in five individually tracked stomatal apertures. In (A; open triangles) normalized stomatal apertures of the same stomata depicted in (B) and (C) are shown. Methods used in these experiments tracking individual stomatal apertures are described in Siegel et al. (2009). ABA-induced stomatal closing experiments are reproduced from Siegel et al. (2009) with permission of the publisher.

CONCLUSIONS

Ca²⁺ functions as a second messenger in guard cell signalling and stomatal movements, as was originally described over 20 years ago (DeSilva et al., 1985; Schwartz, 1985; Schroeder and Hagiwara, 1989; McAinsh et al., 1990). The recently derived Ca²⁺-sensitivity priming model and the concomitant signalosome model provide a mechanism for specificity in plant [Ca²⁺]_cyt signalling. Further research at the single-cell type specific level should shed light on the underlying molecular and protein-mediated Ca²⁺ signalling mechanisms. Guard cells provide a potent system to pursue combined single-cell time-resolved small molecule and protein imaging, electrophysiological ion-channel regulation, cell biological, and whole-leaf and -plant response analyses which undoubtedly will spring new surprises and lead to new levels of understanding of plant-cell signalling dynamics.