Dynamic regulation of guard cell anion channels by cytosolic free Ca2+ concentration and protein phosphorylation


  • Zhong-Hua Chen,

    1. Laboratory of Plant Physiology and Biophysics, Plant Sciences Research Group, Faculty of Biomedical and Life Sciences, Bower Building, Glasgow G12 8QQ, UK
    Search for more papers by this author
  • Adrian Hills,

    1. Laboratory of Plant Physiology and Biophysics, Plant Sciences Research Group, Faculty of Biomedical and Life Sciences, Bower Building, Glasgow G12 8QQ, UK
    Search for more papers by this author
  • Choon K. Lim,

    1. Laboratory of Plant Physiology and Biophysics, Plant Sciences Research Group, Faculty of Biomedical and Life Sciences, Bower Building, Glasgow G12 8QQ, UK
    Search for more papers by this author
  • Michael R. Blatt

    Corresponding author
      For correspondence (fax +44(0) 141 330 4447; e-mail m.blatt@bio.gla.ac.uk).
    Search for more papers by this author

For correspondence (fax +44(0) 141 330 4447; e-mail m.blatt@bio.gla.ac.uk).


In guard cells, activation of anion channels (Ianion) is an early event leading to stomatal closure. Activation of Ianion has been associated with abscisic acid (ABA) and its elevation of the cytosolic free Ca2+ concentration ([Ca2+]i). However, the dynamics of the action of [Ca2+]i on Ianion has never been established, despite its importance for understanding the mechanics of stomatal adaptation to stress. We have quantified the [Ca2+]i dynamics of Ianion in Vicia faba guard cells, measuring channel current under a voltage clamp while manipulating and recording [Ca2+]i using Fura-2 fluorescence imaging. We found that Ianion rises with [Ca2+]i only at concentrations substantially above the mean resting value of 125 ± 13 nm, yielding an apparent Kd of 720 ± 65 nm and a Hill coefficient consistent with the binding of three to four Ca2+ ions to activate the channels. Approximately 30% of guard cells exhibited a baseline of Ianion activity, but without a dependence of the current on [Ca2+]i. The protein phosphatase antagonist okadaic acid increased this current baseline over twofold. Additionally, okadaic acid altered the [Ca2+]i sensitivity of Ianion, displacing the apparent Kd for [Ca2+]i to 573 ± 38 nm. These findings support previous evidence for different modes of regulation for Ianion, only one of which depends on [Ca2+]i, and they underscore an independence of [Ca2+]i from protein (de-)phosphorylation in controlling Ianion. Most importantly, our results demonstrate a significant displacement of Ianion sensitivity to higher [Ca2+]i compared with that of the guard cell K+ channels, implying a capacity for variable dynamics between net osmotic solute uptake and loss.


The cytosolic free concentration of Ca2+ ([Ca2+]i) plays an important role in many signalling events in plants, and in guard cells is a central component in response to the plant stress hormone abscisic acid (ABA). In ABA, elevation of [Ca2+]i is initiated by changes in Ca2+ channel activity which promotes influx of Ca2+ through the plasma membrane and, in turn, triggers Ca2+-induced Ca2+ release from intracellular stores (Blatt, 2000; Hetherington and Brownlee, 2004; McAinsh and Pittman, 2009). The consequent rise in [Ca2+]i affects membrane transport, altering the balance of osmotic solute flux to favour a loss in cell turgour and initiate stomatal closure (Blatt, 2000; Hetherington and Brownlee, 2004; Schroeder et al., 2001).

The key role played by [Ca2+]i in coordinating transport is evident in its actions on K+ and anion fluxes. Elevating [Ca2+]i leads to an inactivation current carried by inwardly rectifying K+ channels (IK,in) and to the activation of inward current (outward flux) through anion channels (Ianion), primarily carried by so-called slow (S-type) anion channels (Grabov et al., 1997; Pei et al., 1997). This transition between the two inward currents – one carried by IK,in and the second by Ianion– is crucial for driving membrane depolarization (Blatt, 2000). Once positive of the K+ equilibrium voltage, the flux of Cl and other anions is balanced by a parallel flux of K+ through outward-rectifying K+ channels (IK,out) and results in a net loss of osmotically active solutes. In Arabidopsis, Ianion was recently identified with the slac1 mutant in which the current is greatly impaired, rendering the slac1 mutant virtually insensitive to ABA (Negi et al., 2008; Vahisalu et al., 2008). These results underscore the importance of SLAC1 and of Ianion in the mechanics of stomatal closure.

Other signals also contribute to control of Ianion in guard cells. Notably, several lines of evidence point to [Ca2+]i-independent pathways (Blatt, 1999; Levchenko et al., 2005) as well as both [Ca2+]i-dependent and -independent protein (de-)phosphorylation (Grabov et al., 1997; Li et al., 2000; Mori et al., 2006). Nonetheless, how these pathways are integrated to regulate Ianion has yet to be explored in any detail. One problem is the virtual absence of quantitative kinetic information: although we know much about the [Ca2+]i dynamics of guard cells, its coupling to ABA and its various targets we remain ignorant of the dynamic characteristics for Ianion and its [Ca2+]i sensitivity as well as its interaction with protein (de-)phosphorylation in regulating Ianionin vivo. Indeed, whereas IK,in has been shown to respond to an apparent cooperative binding of four Ca2+ ions per channel, and with a Kd near 300 nm (Grabov and Blatt, 1999), our understanding of Ianion has been described in terms of limiting [Ca2+]i only. Because its dynamics are a key to understanding the balance of Cl and other anion fluxes as well as its regulation by other factors, knowing the [Ca2+]i dynamics of Ianion regulation is essential for any quantitative elucidation of allied controls on the current.

We have now addressed this gap in knowledge, making use of concurrent measurements of Ianion and [Ca2+]iin vivo and employing the voltage clamp to experimentally manipulate [Ca2+]i in Vicia faba guard cells. We find that Ianion is strongly dependent on [Ca2+]i and that this Ca2+-based control is modulated secondarily by okadaic acid (OA)-sensitive protein dephosphorylation. Additional evidence points to a [Ca2+]i-independent component of Ianion, and hence to at least three modes of control affecting Ianion. Intriguingly, the [Ca2+]i dependence of Ianion is offset to more elevated [Ca2+]i by comparison with IK,in, indicating that at intermediate [Ca2+]i the guard cell membrane need not be biased either for net solute uptake or for its loss.


We recorded Ianion from V. faba guard cells bathed in 15 mm CsCl and 15 mm tetraethylammonium chloride (TEA-Cl), using microelectrodes filled with 300 mm CsCl to eliminate the background of IK.in and IK,out. Under these conditions, the membrane was dominated by Ianion with characteristics similar to those reported by Grabov et al. (1997) and exhibited all the hallmarks of S-type anion currents (Linder and Raschke, 1992; Schroeder and Keller, 1992). As illustrated in Figure 1, the current activated slowly at positive voltages and, once activated, stepping the membrane from +50 mV to voltages negative of approximately 0 mV led to a deactivation of the current over a period of seconds. A substantial instantaneous current was observed on negative voltage steps (near −100 μA cm−2 at −200 mV) but the steady-state current declined in amplitude at voltages negative of −80 mV (Figure 1 and insets).

Figure 1.

 Steady-state and instantaneous current–voltage curves for the anion channel current (Ianion) of guard cells from Vicia faba L.
The guard cells were bathed in 5 mm Ca2+-2-(N-morpholino)propanesulfonic acid (MES), pH 6.1 with 15 mm tetraethylammonium chloride (TEA-Cl) and 15 mm CsCl. Instantaneous (closed symbols) and steady-state currents (open symbols) from one guard cell derived from voltage clamp cycles are shown (bottom inset, symbols indicate corresponding times at which current–voltage curves were derived) with conditioning voltage step to +50 mV (10 sec) followed by test steps (8 sec) to voltages between +50 and −240 mV. Cell parameters: volume, 4850 μm3; surface area, 1600 μm2; aperture, 10 μm. Scale: vertical, 250 mV or 100 μA cm−2; horizontal, 5 sec. Note that the upper set of current traces were obtained during the first 5 min after impalement; the lower set of traces were from the same guard cell 30 min after impalement. Top inset: Mean instantaneous and steady-state current–voltage curves from 16 guard cells recorded 20–30 min after impalement.

Ianion comprises both [Ca2+]i-dependent and -independent components

A common feature of Ianion was its enhancement, especially following prolonged voltage-clamp protocols (Figure 1). Ianion has been mooted to exhibit long-lived open states which are favoured by positive voltage steps and protein phosphorylation (Grabov et al., 1997), and thus might account for such characteristics. However, clamp steps to voltages negative of −150 mV are known to activate Ca2+ channels and promote a rise in [Ca2+]i (Grabov and Blatt, 1998; Hamilton et al., 2000, 2001; Pei et al., 2000) which could also affect Ianion. To distinguish between these alternatives, we made use of a four-step protocol cycle incorporating common positive (activating, Vact) and negative (test, Vtest) voltage steps, and compared the effects of an intervening conditioning voltage step (Vcond) on the amplitude and relaxation kinetics of Ianion during test voltage steps. Using trains of such clamp cycles with all conditioning steps held at −120 mV yielded Ianion without significant change in the current characteristics (Figure 2a). As expected, stepping the conditioning voltage to increasingly negative values (Figure 2b) promoted Ianion during these steps, consistent with the increased electrochemical driving force; significantly, however, subsequent test voltage steps to the common value of −150 mV also showed a corresponding enhancement of Ianion (Figure 2b). This increase in Ianion was observed despite the near complete deactivation of the current during conditioning voltage steps, it was seen when conditioning steps were extended twofold to 10 sec, and it was augmented by increasing extracellular [Ca2+] from 1 to 10 mm (not shown). Similar results were obtained in 18 other experiments, and are summarised in Figure 2d.

Figure 2.

 Negative voltages enhance subsequent anion channel current (Ianion) activity in Vicia faba guard cells.
A four-step protocol was used to introduce a 5-sec conditioning voltage step (Vcond) during 30-sec clamp steps (Vact) to +50 mV used to activate Ianion. After a further 5 or 10 sec, the voltage was stepped to a test voltage (Vtest) of −150 mV to assay the instantaneous Ianion amplitude. Guard cells were bathed in 10 mm Ca2+-2-(N-morpholino)propanesulfonic acid (MES), pH 6.1 with 15 mm tetraethylammonium chloride (TEA-Cl) and 15 mm CsCl.
(a) Voltage protocol (top) and current traces (bottom) from one guard cell clamped over eight cycles with a single Vcond of −120 mV repeated for each cycle showed little or no change in Ianion amplitude when subsequently clamped to Vtest.
(b) Voltage protocol (top) and current traces (bottom) from one of 18 guard cells which showed both the expected increase in Ianion with Vcond and a subsequent rise in current at the common Vtest of −150 mV when clamped over eight cycles with Vcond stepped from −120 to −240 mV.
(c) Current trace from one of eight guard cells which showed little change in Ianion during Vtest following increasing negative Vcond steps from −120 to −240 mV. Scale: vertical, 300 mV or 50 μA cm−2; horizontal, 10 sec.
(d) Mean Ianion ± SE plotted as a function of cycle time (top) with a single Vcond of −120 mV (open bars, n = 19) and with increasing negative Vcond (filled bars, voltages indicated; n = 42). Of the latter, 16 guard cells showed no effect of the changes in Vcond (grey bars) and 26 showed a pronounced increase in current amplitude (black bars).

Intriguingly, we found that a fraction of the guard cells were insensitive to the history of conditioning voltage steps. Out of 42 guard cells, 16 failed to show an increase in Ianion at −150 mV following the more negative conditioning steps. Instead, these cells were characterised at the −150-mV test voltage by a baseline of current, with a mean near −30 μA cm−2 (Figure 2c,d) that was unaffected by increasing extracellular [Ca2+] between 1 and 10 mm (not shown) and, hence, the electrochemical potential driving force for Ca2+ entry across the plasma membrane. Thus, the characteristics for Ianion yielded two distinct patterns, with the larger fraction displaying a ‘memory’ of the conditioning voltage. Because this characteristic was associated with negative voltages – and, hence, conditions favouring channel closure and Ianion deactivation rather than with channel opening – the findings suggested that a rise in [Ca2+]i was behind the enhanced current.

[Ca2+]i affects Ianion when displaced well away from resting values

To test the idea of a rise in [Ca2+]i enhancing Ianion at negative voltages, we combined recordings with concurrent measurements of [Ca2+]i using fluorescence ratio imaging after pre-loading guard cells with the fluorescent dye Fura-2. Fluorescence image pairs were collected at 1-sec intervals using a 535-nm interference filter (±20 nm) after excitation with light at 340 and 390 nm, and were corrected for background fluorescence collected prior to dye injections. Fluorescence ratio (f340/f390) images were calibrated against known standards (Garcia-Mata et al., 2003; Grabov and Blatt, 1998) and were used to estimate [Ca2+]i near the plasma membrane during voltage clamp cycles. Figure 3a shows data recorded from one of 20 independent experiments (cells) in which the same four-step protocol (see Figure 2b) was used to pre-condition the guard cell. Mean f340/f390 values were calculated at each time point from a band of pixels equivalent to a 2-μm wide region around the guard cell periphery to estimate the [Ca2+]i near the plasma membrane, and are plotted (Figure 3a, top) together with the Ianion recorded during test voltage steps (Figure 3a, bottom) at the end of each clamp cycle. As expected, conditioning steps elicited a rise in [Ca2+]i which was most evident with increasing negative voltages. With conditioning steps negative of −150 mV, [Ca2+]i recovered only partially to the initial resting value of 130 ± 18 nm, consistent with the relatively slow rate of export of Ca2+ from the cytosol and the increased Ca2+ load (Hamilton et al., 2000; Köhler and Blatt, 2002; Sokolovski et al., 2005). Coincident with the increase in [Ca2+]i, the amplitude of Ianion at the start of each test voltage step also increased. Similar voltage-evoked increases in [Ca2+]i were obtained in each of 19 other guard cells, yielding a mean resting [Ca2+]i of 125 ± 13 nm and a maximum [Ca2+]i during the final conditioning voltage step of 1.3 ± 0.1 μm (Figure 3c). Of these, 13 guard cells showed a concurrent increase in Ianion while, as before, the remainder showed no significant rise in current in parallel with the conditioning voltage and [Ca2+]i.

Figure 3.

 Negative voltage steps enhance the anion channel current (Ianion) by elevating cytosolic-free [Ca2+] ([Ca2+]i).
Data from guard cells bathed in 10 mm Ca2+-2-(N-morpholino)propanesulfonic acid (MES), pH 6.1 with 15 mm tetraethylammonium chloride (TEA-Cl) and 15 mm CsCl.
(a) [Ca2+]i (red trace, top) and Ianion (blue trace, bottom) recorded during a four-step clamp protocol as in Figure 2(b) and plotted as a function of time during the full protocol. The [Ca2+]i trace was determined from Fura-2 fluorescence ratio images recorded at 1-sec intervals and averaged from a 2 μm wide band around the cell periphery. Ianion traces below show only current recorded at Vact and Vtest during the final 15 sec of each cycle. Fluorescence ratio images (upper insets, a to d) taken at times indicated along the [Ca2+]i trace (pseudocolour scale indicated). Bars above the [Ca2+]i trace indicate periods used to estimate the mean [Ca2+]i during Vtest steps for analysis (see below and Figures 4 and 6).
(b) [Ca2+]i and Ianion traces as in (a) from one of eight guard cells pre-injected with the Ca2+ buffer 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA).
(c) Mean [Ca2+]i (triangles, top) and Ianion (circles, bottom) during Vtest steps from 13 guard cells showing current enhanced by[Ca2+]i elevation (open symbols) and from all eight guard cells pre-injected with BAPTA (filled symbols).

We carried out similar experiments after additionally loading guard cells with the Ca2+ buffer 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA) to suppress changes in [Ca2+]i. BAPTA shows a high specific affinity for Ca2+ with a dissociation constant near 200 nm (Hyrc et al., 2000), and thus has the capacity to restrict [Ca2+]i following stimuli that would otherwise increase its elevation substantially above this range. Guard cells were impaled with electrodes filled with 50 mm BAPTA and were loaded by electrophoretic microinjection before starting the recordings. In each of 19 experiments with BAPTA-loaded guard cells, conditioning voltages negative of −150 mV failed to yield increases in [Ca2+]i beyond approximately 300 nm, even with conditioning voltage steps to −240 mV, and test voltage steps showed a corresponding lack of increase in the amplitude of Ianion. Figure 3b shows the results from one guard cell, and the data from all 19 experiments (cells) are pooled in Figure 3c. These findings clearly identify the requirement for elevation of [Ca2+]i during the conditioning voltage steps to promote Ianion.

To quantify the [Ca2+]i dependence of the current we used Ianion at the start of the test voltage steps – that is, the current amplitude prior to its deactivation at −150 mV – and the time-averaged [Ca2+]i during the test voltage steps (see Figure 3a, solid bars above trace). Figure 4 includes the data from Figure 3a,b and shows that enhancement of Ianion was most pronounced with elevation of [Ca2+]i above approximately 400 nm. We analysed these data making use of nonlinear, least-squares fitting to a Hill equation (Hill, 1910) of the form

Figure 4.

 Anion current (Ianion) shows a steep dependence on elevated [Ca2+]i in Vicia faba guard cells.
(a) Mean Ianion amplitude ± SE (n = 13) derived from four-step protocol recordings and Fura-2 fluorescence ratio image analysis as shown in Figure 3(a). The solid curve is the result of fitting by non-linear least-squares to a Hill function and offset (Eqn 1) with an apparent Kd of 720 ± 65 nm and Hill coefficient of 3.5 ± 0.4.
(b) Data and fitting from (a) replotted after subtracting the predicted offset of 33 ± 2 μA cm−2.

where Io and Imax correspond to Ianion at zero and saturating [Ca2+]i, Ks is the apparent component binding constant for Ca2+ and n is the Hill coefficient. Best fittings (Figure 4a, solid curve) were obtained with a baseline Ianion near 30 μA cm−2 at resting [Ca2+]i and a rise in the current with elevated [Ca2+]i consistent with a Hill coefficient >3 and an apparent Kd near 700 nm. Analyses calculated on a cell-by-cell basis gave statistically equivalent results, with a Hill coefficient of 3.5 ± 0.4 and apparent Kd of 720 ± 65 nm, thus indicating a probable binding of four Ca2+ ions and a substantial elevation of [Ca2+]i above resting values needed to facilitate Ianion.

Ianion is enhanced by okadaic acid

In part, the action of ABA on Ianion has been linked to ABA- and Ca2+-dependent protein kinases (Li et al., 2000; Assmann, 2003). Indeed, previous studies identified a pronounced sensitivity of Ianion to protein phosphatase antagonists and its dependence on several protein kinases and phosphatases for activation by ABA (Armstrong et al., 1995; Grabov et al., 1997; Pei et al., 1997; Kwak et al., 2002; Sokolovski et al., 2005). These observations raise the question whether the Ca2+ dynamics of Ianion might reflect an intermediate step mediated through protein (de-)phosphorylation. To provide a quantitative answer, we used a similar strategy to that outlined before, combining Fura-2 fluorescence and voltage clamp measurements and treatments with the PP1/2A antagonist okadaic acid (OA). Ianion, and the background (or ‘leak’) current which it dominates, is known to be promoted in the presence of OA (Thiel and Blatt, 1994; Schmidt et al., 1995; Schulzlessdorf et al., 1996), but its impact on the [Ca2+]i dependence of the current is unknown.

For these experiments, we carried out measurements both before and after pre-treating guard cells with additions of 1 μm OA to the bath. Figure 5 shows an analysis of currents recorded at the −150 mV test voltage after three conditioning voltage steps between −120 and −240 mV with and without OA pre-treatment. Included are similar measurements from guard cells pre-injected with the Ca2+ buffer BAPTA. Current relaxations showed a weak dependence of the time constant on conditioning voltage in the absence of OA (Figure 5, open squares and inset lower right) and were not significantly affected by the Ca2+ buffer (not shown). By contrast, after treatments with the phosphatase antagonist Ianion showed an additional, slow-relaxing component to the current kinetics clearly visible in the inset (Figure 5, circles and inset top left), much as was previously reported (Grabov et al., 1997). Analysis of the current amplitudes showed that OA affected Ianion consistently with actions on both the [Ca2+]i-dependent and -independent current characteristics. At resting [Ca2+]i, Ianion in all 14 guard cells challenged with OA was enhanced roughly twofold to give a baseline of current near −80 μA cm−2 with test clamp steps to −150 mV (Figure 6a). Of these, Ianion in eight guard cells also showed a pronounced rise with increasing [Ca2+]i following negative conditioning voltage steps. Analysis of these current amplitudes yielded a sigmoid dependence of the mean Ianion on [Ca2+]i that was well fitted to the Hill equation (Eqn. 1) with a Hill coefficient of 3.2 ± 0.5 and a Kd of 573 ± 38 nm (Figure 6a,b, solid curves). These data are replotted in Figure 6b after subtracting the baseline of [Ca2+]i-independent Ianion. Comparison with the fittings from Figure 4 (dashed curves) illustrates the shift in the apparent Kd for Ca2+ with the phosphatase antagonist.

Figure 5.

 The protein phosphatase antagonist okadaic acid slows anion channel current (Ianion) relaxations.
Time constants (τ) for current relaxation at −150 mV on stepping from +50 mV are plotted as a function of the preceding conditioning voltage steps (see Figure 2b) at −120, −190 and −240 mV. Relaxations are fitted to a sum of two exponentials with a common fast-relaxing component. Data are means ± SE of analyses from six independent experiments. Control, open circles; 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA) pre-injection, open squares; okadaic acid pre-treatment, filled circles. Insets: representative current relaxations in the control (bottom right) and with okadaic acid pre-treatment (bottom left).

Figure 6.

 The protein phosphatase antagonist okadaic acid acts primarily to enhance a [Ca2+]i-independent component of the anion current (Ianion) in Vicia faba guard cells.
(a) Mean Ianion amplitude ± SE (n = 8) derived from four-step protocol recordings and Fura-2 fluorescence ratio image analysis as shown in Figure 3(a). The solid curve is the result of fitting by non-linear least-squares to a Hill function and offset (Eqn 1) with an apparent Kd of 573 ± 38 nm and Hill coefficient of 3.2 ± 0.5.
(b) Data and fitting from (a) replotted after subtracting the predicted offset of 79 ± 3 μA cm−2. Dashed curves are fittings from Figure 4 and are included for visual comparison.

We also examined the effect of phosphatase antagonism on Ca2+-induced stomatal closure, taking advantage of time-lapse cinemicrography to record stomatal apertures. Measurements were carried out raising external [Ca2+] after equilibrating epidermal peels in buffer with different Ca2+ concentrations with or without the addition of 1 μm OA. In the absence of direct measurements of [Ca2+]i under these conditions, we assumed a progressive effect of external [Ca2+] in elevating [Ca2+]i (De Silva et al., 1985; MacRobbie, 1990; McAinsh et al., 1995). Figure 7 summarises the results of four independent experiments (80 stomata) under each set of conditions. We found that increasing external [Ca2+] to 50 mm, and to a lesser extent to 10 mm, affected the mean aperture when individual stomata were normalised to the starting apertures (Figure 7a), but this effect was independent of treatments with OA (Figure 7b). The phosphatase antagonist accelerated the rates of closure, but this effect was essentially independent of [Ca2+] outside (Figure 7c).

Figure 7.

 External Ca2+ and the protein phosphatase antagonist okadaic acid independently promote stomatal closure.
Mean stomatal apertures ± SE recorded from four independent experiments (80 stomata) following treatments with increasing external [Ca2+] and with 1 μm okadaic acid.
(a) Stomatal apertures without (open bars) and with (filled bars) okadaic acid pre-treatment following transfer to external [Ca2+] of 0.1, 1, 10 and 50 mm.
(b) Apertures with okadaic acid pre-treatment normalised to the corresponding values without okadaic acid from (a) as a function of external [Ca2+].
(c) Time constants (τ) for stomatal closure on transfer to increasing external [Ca2+] with/without okadaic acid pre-treatment. Aperture as a function of time fitted to single exponentials to derive values for τ at each external [Ca2+] and fitted by non-linear least-squares to a simple Michaelian function with a common Kd of 3.4 ± 0.4 mm Ca2+. Note that okadaic acid pre-treatment decreases the value of τ at low [Ca2+] but has no obvious effect on stomatal sensitivity to external [Ca2+].


Coordinated regulation of anion and K+ fluxes at the guard cell plasma membrane is vital for controlling stomatal aperture and, hence, for gas exchange and the efficiency of water use by plants. A number of signals factor into this regulation and have been shown to interact at several levels (Blatt, 2000; Sokolovski and Blatt, 2007; Wang and Song, 2008). Among these, [Ca2+]i is a major contributor in controlling the two inward-directed currents which predominate at the guard cell plasma membrane (Blatt, 2000; Hetherington and Brownlee, 2004). One of these currents, IK,in carried by inward-rectifying K+ channels, facilitates stomatal opening and inactivates with elevated [Ca2+]i; the second current, Ianion carried by an efflux Cl and other anions, is known to activate with an increase in [Ca2+]i during stomatal closure (Blatt, 2000; Pandey et al., 2007). Thus, [Ca2+]i appears to lie at a pivot of the scales balancing the membrane between states favouring solute uptake and stomatal opening or solute loss and stomatal closure. Remarkably, virtually no information has been forthcoming about the dynamics for [Ca2+]i control of Ianion, despite its essential role in depolarizing the membrane to engage net solute loss (Schroeder et al., 2001; Negi et al., 2008; Vahisalu et al., 2008). Our results now address this gap in knowledge, demonstrating a pronounced bias to [Ca2+]i above 400–500 nm needed to activate Ianion as well as an unexpected complexity to the interaction of protein phosphorylation with [Ca2+]i in its regulation. We find that: (i) Ianion comprises both [Ca2+]i-dependent and -independent components, the Ca2+-sensitive component showing a relatively high apparent Kd for Ca2+; (ii) both current components are sensitive to the protein phosphatase antagonist OA; and (iii) phosphatase antagonism displaces the apparent Kd for Ca2+ of Ianion without altering its intrinsic dynamics. These results underscore a complementarity in [Ca2+]i-dependent control of Ianion and of IK,in, with a minimum of overlap in the transition between osmotic solute influx and efflux. They also indicate a functional separation between [Ca2+]i and OA-sensitive protein phosphorylation, thus implicating three pathways of control for Ianion.

Ca2+-dependent and -independent modulation of Ianion

A key finding was the observation that Ianion is activated by [Ca2+]i, but only when raised substantially above the resting value. Fura-2 fluorescence measurements indicated a mean resting [Ca2+]i of 125 ± 13 nm, similar to values reported previously under several different experimental conditions (Bothwell et al., 2006; Garcia-Mata et al., 2003; Gilroy et al., 1991; Grabov and Blatt, 1997, 1998, 1999; McAinsh et al., 1990; Sokolovski et al., 2005; Webb et al., 1996). We observed that transient increases in [Ca2+]i above resting promoted Ianion, but the effect was evident principally when [Ca2+]i rose to values near and above 500 nm. A detailed analysis of Ianion showed an apparent Kd for [Ca2+]i near 720 nm and a Hill coefficient between 3 and 4 (Figure 4). It indicated a substantial degree of cooperativity –‘gearing’ the [Ca2+]i dependence of Ianion enhancement to small changes in [Ca2+]i– but within a dynamic range that is roughly four- to sixfold above the resting [Ca2+]i. These findings are supported by studies in which Ca2+ buffering of the cytosol was elevated under experimental control. Much as was reported by Hedrich et al. (1990) using the Ca2+ buffer EGTA, in our hands introducing BAPTA into the cytosol suppressed voltage-evoked [Ca2+]i transients and the concurrent increases in Ianion. Significantly, we can now put these observations within a quantitative framework: concurrent Fura-2 fluorescence recordings showed that [Ca2+]i increases were constrained to a 300-nm ceiling near the Kd for Ca2+ of BAPTA (Hyrc et al., 2000) and, thus, substantially below the dynamic range of Ianion for [Ca2+]i. An important conclusion to be drawn from this analysis is that [Ca2+]i-mediated enhancement of Ianion only occurs at [Ca2+]i near the highest range of values reported in the presence of 10–50 μm ABA (McAinsh et al., 1990; Gilroy et al., 1991), nitric oxide (Garcia-Mata et al., 2003), reactive oxygen species (Kwak et al., 2003) and CO2 (Webb et al., 1996). In other words, stimuli which lead to modest increases in [Ca2+]i are likely to act on Ianion via other means.

It is of interest, too, that a baseline of Ianion activity was evident even at resting [Ca2+]i. At a clamp voltage of −150 mV this component yielded an amplitude near −30 μA cm−2, or approximately 25% of the mean current recorded at saturating [Ca2+]i. A similar current amplitude was observed in guard cells which failed to show a rise in Ianion when [Ca2+]i was elevated (Figures 2 and 3), and the juxtaposition suggests the concurrence of two independent modes for Ianion gating or two distinct subpopulations of slow-activating anion channels at the membrane, one subject to [Ca2+]i activation and the other not. Indeed, the persistence of a [Ca2+]i-insensitive component to Ianion is consistent with past evidence for a second [Ca2+]i-independent mode of Ianion and its activation by ABA, both in V. faba (Grabov et al., 1997; Schwarz and Schroeder, 1998; Sokolovski et al., 2005) and in tobacco (Levchenko et al., 2005; Marten et al., 2007). Significantly, this component, like the [Ca2+]i-sensitive current, showed a substantial elevation with protein phosphatase antagonism (Figure 5), leading us to the conclusion (below) that Ianion encompasses three distinct patterns of control.

Phosphorylation enhances the Ca2+ sensitivity of Ianion and elevates the Ca2+-independent Ianion component

In addition to its action on the baseline of Ianion, we found that protein phosphatase antagonism modified the dynamics for the [Ca2+]i sensitivity of the current. This action was purely on the apparent Kd for [Ca2+]i (Figure 6) rather than on the degree of cooperativity in current enhancement. It showed that favouring protein phosphorylation with OA treatments displaced the effective range of [Ca2+]i action on Ianion but not the intrinsic sensitivity of the current to a change in [Ca2+]i. How might such an action be interpreted? On the basis of [Ca2+]i kinetics alone our data do not discriminate between a direct and indirect effect: either one in which Ca2+ binding and protein (de-)phosphorylation events occur sequentially within a signal cascade or in which these events operate in parallel. The sequential model is clearly relevant to Ca2+-dependent protein kinases which may be countered by OA-sensitive protein dephosphorylation. It is consistent with evidence of altered stomatal function in cpk mutants (Mori et al., 2006; Cheong et al., 2007; Ma and Wu, 2007), although phenotypic analyses of these mutants lend themselves to other interpretations as well. The parallel model would embody an effect of OA on the predisposition for [Ca2+]i modulation of an element in the signal cascade, but without the (de-)phosphorylation event itself being a prerequisite for signal transmission. Such facilitation has been cited to explain the synergy between the actions of ABA and protein phosphatase antagonists on Ianion (Grabov et al., 1997), among others. Both models are consistent with an increase in apparent Kd for Ca2+ and can accommodate the change in Kd without apparent change in cooperativity.

It is much more difficult to explain the differences in current kinetics in the context of a sequential model, and these observations therefore argue for parallel and independent control of the current. For example, assuming that current activation occurs via a Ca2+-dependent protein kinase and OA suppresses dephosphorylation of its target, then the consequence for Ianion should have been the same. Yet we found that only OA treatments introduced a slow-relaxing component to the current (Figure 6), much as was reported previously (Grabov et al., 1997). Of course, the action of OA is unlikely to be so targeted. These differences highlight three distinct regulatory modes for Ianion: the first associated with [Ca2+]i-independent current, much as proposed before (Grabov et al., 1997; Levchenko et al., 2005; Schwarz and Schroeder, 1998), the second related to the additive effect of the [Ca2+]i-dependent component and the third related to its modulation by protein (de-)phosphorylation. Whether these same patterns apply to other species is an open question, although it does seem a reasonable assumption. Nonetheless, we are aware that phosphorylation-dependent control of Ianion, or at least its sensitivity to protein phosphatase antagonists, may differ between guard cells in V. faba (Grabov et al., 1997) and Arabidopsis (Pei et al., 1997). In fact, the interactions between protein phosphorylation and [Ca2+]i are likely to be more complicated still, as some effects of Ca2+-dependent protein kinase mutations have been associated with long-term adaptive changes in stomatal behaviour (Mori et al., 2006).

Physiological implications

The regulatory characteristics of Ianion lead us to draw two important conclusions. First, we note that the dynamic range of its [Ca2+]i sensitivity is displaced to substantially higher free concentrations than those of the other major inward current in the guard cell. Current carried by inward-rectifying K+ channels in V. faba guard cells is inactivated with increasing [Ca2+]i, but shows an apparent Kd for [Ca2+]i near 300 nm and a Hill coefficient consistent with the binding of four Ca2+ ions (Grabov and Blatt, 1999). By contrast, elevated [Ca2+]i becomes important in controlling Ianion only near and above 500 nm as full inactivation of IK,in is approached (Figures 3 and 4). This separation in the functional [Ca2+]i dynamics of the two currents is significant: it ensures the minimum of overlap between conditions favouring solute uptake and solute loss, and it highlights the importance of [Ca2+]i in transitions between these two states of the membrane. In short, the [Ca2+]i dynamics of these two currents serve as landmarks for the breadth of the physiological [Ca2+]i variation – roughly between 200 and 900 nm– relevant to control of ion transport at the guard cell plasma membrane and, hence, stomatal regulation. Second, we find the effects of [Ca2+]i and of protein phosphorylation on Ianion to be predominantly additive, rather than synergistic. Again, this conclusion draws on our quantitative analysis of [Ca2+]i and Ianion which sets a precedent in defining the boundaries of the current and its sensitivity to [Ca2+]i. The additive nature of the controls reflects the substantial baseline component of the current which is independent of [Ca2+]i and strongly enhanced under phosphatase antagonism, and it is at least superficially consistent with our observations that Ca2+-induced stomatal closure is accelerated by OA but without a substantial change in the sensitivity to external [Ca2+] (Figure 7). By contrast, a synergism is evident in a modest shift of the Kd for [Ca2+]i action (Figure 6), but it is less obvious how this more subtle effect contributes to stomatal control. A significant challenge now lies in associating the different kinetic interactions between the two signals with the specific targets for (de-)phosphorylation.

Experimental procedures

Plant material and experimental protocols

Vicia faba L. cv. Bunyard Exhibition was grown on potting mix and perlite (70:30) at 22°C and 60% relative humidity with 200 μmol m−2 sec−1 photosynthetic photon flux density under a 16/8 h day/night cycle. Plants were watered daily, with Hoagland’s solution every other day. Epidermal strips were prepared from newly expanded leaves of 2–4-week-old plants as described previously (Blatt, 1992). The surface area and volume of impaled guard cells were calculated assuming a spheroid geometry using Henry EP software (Glasgow University, http://www.psrg.org.uk/Software/henry.htm). Epidermal peels were affixed to the glass bottom of the experimental chamber after coating the chamber surface with an optically clear and pressure-sensitive silicone adhesive and all operations were carried out on an Axiovert S100TV microscope (Zeiss, http://www.zeiss.com/) fitted with Nomarski differential interference contrast optics. Measurements were conducted in continuous flowing solutions controlled by a gravity-fed system at a rate of 20 chamber volumes min−1. The standard perfusion medium contained 15 mm TEA-Cl, 15 mm CsCl and 5 mm 2-(N-morpholino)propanesulphonic acid (MES), titrated with Ca(OH)2 to a pKa at 6.1. All measurements were carried out at an ambient temperature of 22 ± 1°C. Chemicals were from Sigma (http://www.sigmaaldrich.com/) unless otherwise specified.


Recordings were obtained using double- and triple-barrelled microelectrodes coated with paraffin to reduce capacitance (Blatt and Armstrong, 1993; Grabov and Blatt, 1998). Current and voltage recording barrels were filled with 300 mm CsCl (pH 7.5) to block K+ current and enhance Cl current. Microelectrodes were connected to amplifier headstages via 1 m KCl/Ag–AgCl half-cells, and a 1 m KCl agar bridge served as the reference electrode. In some experiments, [Ca2+]i was buffered by loading cells with BAPTA. In this case, microelectrodes included 50 mm BAPTA, pH adjusted to 7.5 with CsOH, in addition to 300 mm CsCl. Protein phosphorylation was manipulated by pre-incubating epidermal strips in 1 μm OA for 20 min before impalements.

Cytosolic Ca2+ recording

The pentapotassium salt of the Ca2+-sensitive fluorescent dye Fura-2 was prepared as a 10 mm stock in dimethyl sulphoxide (DMSO) and was diluted to 20 μm in 50 mm CsCl before filling the third barrel of three-barrelled microelectrodes. Guard cells were injected electrophoretically using a μPion module (μP system, Y-Science, http://www.psrg.org.uk) while clamping the membrane to −50 mV, thus ensuring that the injection current did not cross the membrane. The final Fura-2 concentrations estimated by integrating injection currents were <10 μm. Dye loading was judged to have been successful by visual checks for cytosolic dye distribution and by stabilization of the fluorescence ratio signal (Grabov and Blatt, 1998).

Measurement of [Ca2+]i was carried out and calibrated as described previously (Grabov and Blatt, 1998). Fura-2 was excited using 340 and 390 nm light (10 nm slit width; Polychrom II, TILL Photonics, http://www.till-photonics.com/) and the fluorescence collected after passage through a 535 nm bandpass (±20 nm) filter (Schott, http://www.schott.com/). Images were recorded at 1-, 2- and 5-sec intervals using a cooled Pentamax-FT512 CCD camera and GenIV intensifier (Princeton Instruments, http://www.princetoninstruments.com/). Measurements were corrected for background using the average of three images collected before dye loading and were analysed using MetaFluor and MetaMorph software (http://www.moleculardevices.com/pages/software/metamorph.html).

Stomatal aperture measurements

Stomatal apertures were recorded in epidermal peels and measurements were carried out under continuous superfusion in Ca2+-MES buffer, pH 6.1, with additions as described. In any one experiment, 20–30 stomata were recorded by time-lapse photomicrography (AxioCam, Zeiss) at 2-min intervals. Stomatal apertures and their response to experimental challenges were determined directly from image sets using Image J software (v. 1.38p; NIH, http://rsbweb.nih.gov/ij/).

Numerical analyses

Where appropriate data are presented as means ± SE of n observations, and differences validated by Student’s t-test or anova. Curve fittings were by non-linear least-squares using a Marquardt–Levenberg algorithm (Marquardt, 1963).


We thank Amparo Ruiz-Prado for preparing plant materials and Jian-Wen Wang (Suzhou University, China) for helpful discussion during the preparation of the manuscript. This work was supported by research grants from the UK Biotechnology and Biological Sciences Research Council (BBSRC grants BB/F001673/1 and BB/F001630/1).