ABA depolarizes guard cells in intact plants, through a transient activation of R- and S-type anion channels

Authors


For correspondence (fax +49 931 8886157; e-mail hedrich@botanik.uni-wuerzburg.de).

Both authors contributed equally to this study.

Summary

During drought, the plant hormone abscisic acid (ABA) induces rapid stomatal closure and in turn reduces transpiration. Stomatal closure is accompanied by large ion fluxes across the plasma membrane, carried by K+ and anion channels. We recorded changes in the activity of these channels induced by ABA, for guard cells of intact Vicia faba plants. Guard cells in their natural environment were impaled with double-barrelled electrodes, and ABA was applied via the leaf surface. In 45 out of 85 cells tested, ABA triggered a transient depolarization of the plasma membrane. In these cells, the membrane potential partially recovered in the presence of ABA; however, a full recovery of the membrane potentials was only observed after removal of ABA. Repetitive ABA responses could be evoked in single cells, but the magnitude of the response varied from one hormone application to the other. The transient depolarization correlated with the activation of anion channels, which peaked 5 min after introduction of the stimulus. In guard cells with a moderate increase in plasma membrane conductance (ΔG < 5 nS), ABA predominantly activated voltage-independent (slow (S)-type) anion channels. During strong responses (ΔG > 5 nS), however, ABA activated voltage-dependent (rapid (R)-type) in addition to S-type anion channels. We conclude that the combined activation of these two channel types leads to the transient depolarization of guard cells. The nature of this ABA response correlates with the transient extrusion of Cl from guard cells and a rapid but confined reduction in stomatal aperture.

Introduction

During periods of limiting water supply, plants reduce their stomatal conductance, to decrease transpiration (Assmann and Shimazaki, 1999; Schroeder et al., 2001). Stomatal closure, however, also reduces CO2 uptake and thereby limits carbon assimilation. To optimize the uptake of CO2 and the evaporation of H2O, plants have developed signalling pathways, which tightly control stomatal opening. These pathways regulate the activity of ion transport proteins at the plasma membrane and tonoplast (Assmann and Shimazaki, 1999; MacRobbie, 1998; Schroeder et al., 2001). Changes in the activity of these transporters alter the ionic composition of guard cells and thereby provide the driving force for stomatal movement.

Recently, we developed a method to study plasma membrane transport of single guard cells in the intact plant (Roelfsema et al., 2001). These cells responded to light and CO2, with large membrane potential changes that altered the direction of the K+ flux across the plasma membrane (Roelfsema et al., 2001, 2002). Membrane potential changes of this magnitude, induced by light and CO2, had not been recorded with guard cells in epidermal strips or protoplasts thereof. Probably, the mechanical stress during isolation of epidermal strips or protoplasts, in combination with loss of the natural environment, alters the responsiveness of guard cells to these signals (Roelfsema and Hedrich, 2002).

In addition to the light- and CO2-signalling pathways previously studied, plants are equipped with a drought-sensing system. Here, the phytohormone abscisic acid (ABA) plays a central role (Assmann and Shimazaki, 1999; Schroeder et al., 2001), as mutants insensitive to ABA (abi) tend to wilt (Koornneef et al., 1984; Roelfsema and Prins, 1995). These abi mutants lack ABA-induced plasma membrane responses recognized for wild-type guard cells (Armstrong et al., 1995; Pei et al., 1997). In wild-type guard cells, an effect of ABA on at least three types of ion channels has been reported. ABA enhances the activity of outward K+ channels (Blatt and Armstrong, 1993) and slow (S)-type anion channels (Grabov et al., 1997; Pei et al., 1997), while that of inward K+ channels is reduced (Blatt and Armstrong, 1993; Lemtiri-Chlieh and MacRobbie, 1994; Schwartz et al., 1994).

The sensitivity of stomata to ABA depends on other signals, such as CO2 and indole-3-acetic acid (IAA) (Raschke, 1987). This explains, to some extent, why all guard cells do not display the same response to the stress hormone. In guard cells, ABA can induce rises in cytoplasmic Ca2+, but only a limited number of the cells display this response (Allen et al., 1999; Gilroy et al., 1991; McAinsh et al., 1990). In some experiments, ABA was even found to turn spontaneous Ca2+ oscillations off (Klüsener et al., 2002).

Although targets of ABA regulation have been recognized for guard cells, the sequence of events leading to stomatal closure still awaits a detailed analysis. Recordings with ion-selective miniature electrodes at the guard cell wall showed an increase of Cl extrusion starting a few minutes after ABA application (Felle et al., 2000). The Cl concentration peaks 15 min after stimulus onset and returns to pre-stimulus values within an hour. Tracer flux experiments with epidermal peels also showed that ABA temporarily increases the efflux of anions (MacRobbie, 1987). Altogether, these data point to an ABA-induced transient activation of anion channels in guard cells. So far, however, such a transient response has not been measured for guard cells in epidermal strips or protoplasts thereof.

Here, we attempt to bridge the gap between data from transpiration measurements, ion flux recordings and changes in ion channel activity, by monitoring the ABA response of single guard cells in their natural environment: the intact plant. In line with ion flux measurements, guard cells transiently depolarized in response to ABA, because of the activation of both rapid (R)-type and S-type anion channels.

Results

ABA-induced stomatal closure and membrane depolarization

Guard cells surround the stomatal pore and therefore can be easily ABA stimulated via the leaf surface. The responsiveness of guard cells to ABA, applied via leaf surface perfusion, was tested microscopically. Figure 1(a) depicts the abaxial epidermis of a Vicia faba leaf, with an ABA-responsive stoma adjacent to an ABA-insensitive one. The upper, ABA-responsive stoma started to close 8 min after the introduction of the stress hormone and reached its maximal closure within 20 min (Figure 1a,b). Out of 37 stomata tested, 14 closed upon exposure to 10 µm ABA. The other 23 stomata did not respond to ABA, just as the lower stoma in Figure 1(a). Based on this heterogeneity, it can be assumed that at least 30% of the guard cells will show an ABA-induced change in the electrical properties of the plasma membrane.

Figure 1.

Effect of 10 µm ABA, applied via leaf cuticle perfusion, on stomatal opening.

(a) Abaxial side of a V. faba leaf, an experimental solution (5 mm KCl, 5 mm potassium citrate (pH 5.0), 0.1 mm CaCl2 and 0.1 mm MgCl2), ran between the objective and the leaf surface (volume 0.3 ml) at flow rate of 1.5 ml min−1. The leaf was illuminated with white light at a photon flux density of 500 µmol m−2 sec−1, projected from the adaxial side at an area of 20 mm diameter. At t = 10 min (left figure), the experimental solution was changed to one containing 10 µm ABA. Stomatal closure occurred during the following 20 min (right figure) for the upper stoma, but not for the lower stoma.

(b) Changes in stomatal aperture plotted against time, from the same stomata as in (a) (○, upper stoma; ●, lower stoma). At t = 10 min, the solution on the cuticle was changed for one containing 10 µm ABA.

Plasma membrane responses to ABA were monitored for guard cells that were impaled with double-barrelled electrodes. Upon impalement, the membrane potential of these cells transiently depolarizes and reaches a new stable value in 2–4 min (Roelfsema et al., 2001). After reaching a stable membrane potential, 71 out of 85 cells could be classified as ‘depolarized’; these cells had an average membrane potential of −74 mV (SD 10). The other 14 cells were classified as ‘hyperpolarized’ and had an average membrane potential of −112 mV (SD 16). Note, however, that the recorded membrane potentials may be affected by electrical leaks caused by the impalement. These potential leaks are small for hyperpolarized cells (see Experimental procedures), but may have an impact on the membrane potential of depolarized cells.

In line with the observation that some stomata do not close in response to ABA, an ABA-induced change in the membrane potential could not be observed in 40 cells, 2 of which were hyperpolarized before ABA application (Figure 2a). In the other 45 cells, ABA depolarized the plasma membrane (Figure 2b–d). The amplitude of the ABA-induced depolarization was variable and depended on the membrane potential before application of the stress hormone. Cells that could be classified as ‘depolarized’, further depolarized upon exposure to ABA (Figure 2b,c). The depolarization reached its maximal value (ΔEm = 22 mV, SD 10) 5 min (SD 1) after application of the stimulus. Following this transient, the membrane potential partially recovered to a value 12 mV (SD 8) more positive than that before application of ABA. A subsequent removal of ABA from the perfusion solution caused a recovery of the membrane potential to a pre-stimulus value. Before reaching the pre-stimulus value, however, a large percentage of the ‘depolarized’ cells showed an overshoot response (Figure 2c). Guard cells that could be classified as ‘hyperpolarized’ (Roelfsema et al., 2001) had an average membrane potential of −112 mV (SD 16) and strongly depolarized upon exposure to ABA (ΔEm = 57 mV, SD 19, Figure 2d). Again, the cells spontaneously re-polarized in the presence of the hormone; however, a complete recovery of the membrane potential occurred only after the removal of ABA. The plasma membrane responses thus split into four groups: (i) depolarized or hyperpolarized non-responsive cells; (ii) depolarized cells further depolarizing with ABA; (iii) depolarized cells with an overshoot after removal of ABA; and (iv) hyperpolarized cells with a strong ABA-induced depolarization.

Figure 2.

ABA-induced changes in the electrical properties of the plasma membrane of guard cells in intact V. faba plants.

(a–d) Guard cell membrane potentials before, during and after the application of 10 µm ABA, as indicated in the bar below the graphs. Four types of responses could be distinguished: (a) Cells not responding to ABA; (b) depolarized cells, further depolarizing after application of ABA; (c) depolarized cells, further depolarizing with ABA and transiently hyperpolarizing after washing out ABA; and (d) hyperpolarized cells becoming depolarized in the presence of ABA.

(e) Plasma membrane current of a guard cell clamped continuously at −100 mV. The guard cell was exposed to 10 µm ABA, as indicated by the bar below the graph. ABA induced first a transient increase of inward current, followed by a steady increase of inward current at lower amplitude.

All ABA-responsive cells shared an initial transient depolarization. The most positive potential reached during this depolarization was independent of the membrane potential before hormone application (Figure 2b–d). The most depolarized potential was −52 mV (SD 12) and −55 mV (SD 11) for initially depolarized or hyperpolarized cells, respectively. The underlying change in ion channel activity was explored through voltage clamp studies. Guard cells were exposed to the hormone, while their plasma membrane was constantly clamped at −100 mV. At this potential, the plasma membrane ion conductance was small, as voltage-dependent K+ channels were not active (Roelfsema et al., 2001). ABA triggered a transient increase in inward current, which reached a peak value approximately 5 min after introduction of the hormone (Figure 2e). During prolonged ABA stimulation, the inward current dropped again, but remained larger than that before stimulus onset. This response could be elicited by the stress hormone at concentrations as low as 1 µm (data not shown).

Transient changes in the activity of plasma membrane ion channels

To study the ionic basis of the ABA-induced increase in plasma membrane conductance, the membrane potential of guard cells was clamped at regular intervals to a range of test potentials (diamonds in Figure 3a). Before exposure to ABA, the guard cell membrane conductance was dominated by inward and outward rectifying K+-selective channels (Figure 3b, open circle; Roelfsema et al., 2001). In contact with ABA, a dramatic change in the plasma membrane conductance properties was observed (Figure 3b, closed circle). In this cell, ABA-stimulated ion channels that activate instantaneously (Figure 3b,c) and inhibited time-dependent outward rectifying K+ channels (Figure 3b,d). Upon prolonged hormone application, these conductance changes were reversed (Figure 3b–d, closed squares), in line with the transient depolarization (Figure 3a). After the removal of ABA, the cell became transiently hyperpolarized (Figure 3a), which correlated with an increase of outward current recorded at membrane potentials negative of −60 mV (Figure 3c, open squares).

Figure 3.

ABA induced changes in plasma membrane conductance.

(a) Membrane potential trace of a guard cell exposed to 10 µm ABA (dark area of bar below the graph). Diamonds indicate the time points at which voltage clamp protocols were applied.

(b) Current traces from the same cell as in (a), symbols correlate. Arrows indicate the 0-nA level. Cells were clamped from a holding potential of −100 mV to potentials ranging from −180 to 0 mV with 20-mV increments. Note that the introduction of ABA (●) caused a dramatic increase of currents measured directly after the capacity compensation peak; concurrently, time-activated outward currents are reduced. The currents virtually recovered to pre-stimulus values after a prolonged exposure to ABA (▪).

(c) Instantaneous currents, sampled directly after termination of the capacity compensation peak, plotted against the clamp voltage (symbols correspond to (a)). Note that ABA triggered an increase of inward current at test potentials negative of −40 mV (●). The effect of ABA was reversed after prolonged exposure (▪). The removal of ABA resulted in a further decrease of inward currents, at potentials negative of −60 mV (□).

(d) Steady-state currents, sampled at the end of the 2-sec test pulses, plotted against the clamp voltage (symbols correspond to (a)). Note that ABA caused a large shift of the zero current potential to more positive values.

The effect of ABA on outward rectifying K+ channels varied; in 14 out of 27 cells, the stress hormone decreased the conductance of these channels, as in Figure 3. However, in 10 other cells depolarizing in response to ABA (Figure 4a), a less than 10% change in current, mediated by outward K+ channels, was found (Figure 4b). In contrast to the variable effect of ABA on outward K+ channels, the stress hormone stimulated instantaneously activating channels in all transiently depolarizing cells (Figures 3b,c and 4b). In Figure 4(b), these channels slowly deactivate at potentials ranging from −120 to −160 mV, a feature reminiscent of S-type anion channels (Linder and Raschke, 1992; Schroeder and Hagiwara, 1989).

Figure 4.

ABA-responsive guard cell, in which the conductance of outward rectifying channels remains unchanged.

(a) Membrane potential trace of a guard cell exposed to 10 µm ABA (dark area of bar below the graph). Diamonds indicate the time points at which voltage clamp protocols were applied.

(b) Current traces from the same cell as in (a), symbols correlate. Arrows indicate the 0-nA level. Cells were clamped from a holding potential of −100 mV to potentials ranging from −180 to 0 mV with 20-mV increments. Note that ABA does not affect the time-dependent outward currents. ABA transiently increases instantaneous inward currents (●) that slowly deactivate at potentials from −120 to −160 mV. The increase in instantaneous current levels off (▪) after prolonged exposure to ABA.

Based on the voltage-dependent activation and deactivation, the inward current triggered by ABA is most likely conducted by anion channels. However, non-selective cation channels (Demidchik et al., 2002) or voltage-independent K+ channels (Marten et al., 1999) could also contribute to the inward conductance. To exclude the latter possibility, we filled electrodes with 300 mm CsCl and used a perfusion solution containing BaCl2. The presence of these K+-channel blockers eliminated currents carried by inward and outward rectifying K+ channels (Figure 5a). Under these conditions, ABA still elicited a transient inward current at a holding potential of −100 mV as well as at −40 mV (Figure 5b). This behaviour, together with the Nernst potential of K+≈−73 mV, indicates that the inward current is not carried by K+-selective channels. The current may thus be conducted by anion- or non-selective channels, which are not affected by Cs+ and Ba2+.

Figure 5.

Ionic nature of ABA-induced inward current.

(a) Current traces of a guard cell impaled with a double-barrelled electrode containing 300 mm CsCl and exposed to Ba2+ via the perfusion solution flowing over the leaf. The guard cell was clamped stepwise from a holding potential of −100 mV to test potentials ranging from −180 to 0 mV with 20-mV increments. Arrow indicates the 0-nA level. Note that the time-dependent outward as well as the inward rectifying K+ channels are blocked.

(b) Current traces of the same guard cell as in (a), exposed to a short pulse (40 sec) of 10 µm ABA as indicated by the dark area in the bar below the graphs. ABA induced a transient inward current at a holding potential of −100 mV (left graph) as well as at a holding potential of −40 mV (right graph).

(c) Current trace of a guard cell impaled with an electrode containing 300 mm CsCl, exposed to a perfusion solution with 5 mm KCl and clamped to −100 mV. In contact with 10 µm ABA (dark area in bar below the graph), the inward current across the plasma membrane transiently increased.

(d) Current–voltage relations of the same guard cells as in (c), determined with a ramp of 2 sec, from −180 to 60 mV, before and during the ABA response (Symbols correlate to (c)). The clamp protocol was applied from a holding potential of −100 mV. Note that during the ABA response the reversal potential shifts from −30 to 20 mV.

The reversal potential of ABA-stimulated channels was determined using electrodes filled with 300 mm CsCl and a perfusion solution with KCl. Although a low activity of inward K+ channels was still recorded at these conditions (data not shown), it ensured that K+ is the main cation in the guard cell wall (Roelfsema and Hedrich, 2002). At a holding potential of −100 mV, ABA supply induced a transient inward current (Figure 5c). Before and after ABA application, the reversal potential was determined with fast (2 sec) voltage ramps from −180 to 60 mV (Figure 5d). During the ABA response, the reversal potential shifted from −30 to 20 mV, the latter potential being close to the Nernst potential of Cl. If the ABA-induced current was carried by non-selective K+-conducting (Demidchik et al., 2002) cation channels, a shift of the reversal potential to values negative of −30 mV would have been expected. Instead, the reversal potential shifted to more positive values, showing that the ABA-induced current is predominantly carried by anion channels.

ABA activates both R- and S-type anion channels

ABA activation of anion channels was further studied, using a BaCl2-containing perfusion solution and KCl in the microelectrode. Ba2+ blocked the inward rectifying K+ channels and reduced the current carried by outward rectifying K+ channels (Figure 6b, closed circle; Schroeder et al., 1987). Ba2+ did not affect the time course of anion channel activation, as ABA stimulation peaked 5 min (SD 1, n = 21) after stimulus onset, just as in the absence of Ba2+. Under both conditions, guard cell anion channels could be repetitively activated by ABA, each exposure resulting in an increase of inward current (Figure 6a). Although ABA repetitively activated inward currents with a similar time course, the amplitude levelled off with each exposure. The increase of inward current was accompanied by an increase in conductance of instantaneously activating channels (Figure 6b, closed circle and square). The properties of these channels differed from one ABA exposure to another. During the first exposure, the instantaneous activating channels had a peak conductance at −100 mV (Figure 6c), a feature characteristic for R-type anion channels (Dietrich and Hedrich, 1998; Hedrich et al., 1990; Keller et al., 1989; Kolb et al., 1995; Schroeder and Keller, 1992). During the second (data not shown) and third (Figure 6d) exposure to ABA, the peak conductance at −100 mV decreased. The remaining current was reminiscent of S-type anion channels (Linder and Raschke, 1992; Schroeder and Hagiwara, 1989; Schroeder and Keller, 1992), as it displayed a more linear current voltage relation and slowly deactivates at −180 mV (Figure 6b, open square).

Figure 6.

Repetitive activation of anion channels by ABA.

(a) Plasma membrane current of a guard cell clamped at −100 mV and exposed three times to 10 µm ABA (dark areas in the bar below the graph). The membrane was clamped to a range of defined test potentials at regular intervals, indicated by diamonds. The recording was carried out with an experimental solution containing 5 mm BaCl2 to inhibit outward and inward rectifying K+ channels.

(b) Current traces of the same cell as in (a), symbols correlate. Arrows indicate the 0-nA level. The plasma membrane was clamped from −100 mV to test potentials ranging from −180 to 0 mV with 20-mV increments; for clarity, only traces of −180, −120, −60 and 0 mV are shown. Note that the presence of Ba2+ reduced the currents carried by inward and outward rectifying K+ channels (○). The introduction of ABA (● and ▪) caused an increase of instantaneously activating currents.

(c) Instantaneous currents, sampled after termination of the capacity compensation peak, plotted against the clamp voltage. Currents were measured before (○) and during (●) the first application of ABA.

(d) Same as in (c), but now for current sampled before (□) and during (▪) the third ABA application. Note that the inward current has a peak value at −100 mV during the first ABA application, which was absent during the third application.

The relative contribution of R- and S-type anion channels to the ABA-induced increase in conductance was estimated. Instantaneous current–voltage relations were fitted with an equation (see Experimental procedures), which sums the current carried by R-type and S-type anion channels. This analysis revealed that in 6 out of 13 responses, R-type anion channels dominated the anion conductance. In the latter experiments, the entrance of Ba2+, via Ca2+- and Ba2+-permeable channels (Demidchik et al., 2002; Hamilton et al., 2000; Pei et al., 2000) cannot be excluded, and thus Ba2+ may have altered the activation state of the two anion channel types (Hedrich et al., 1990; Schroeder and Hagiwara, 1989). The ABA effect was therefore analysed in the absence of Ba2+ as well. Under these conditions, the distinct properties of both anion channel types were apparent in response to voltage steps from −100 to −180 mV (Figure 7a,c). In cells dominated by S-type anion channels, the current instantaneously increased at −180 mV (Figure 7a), while an instantaneous decrease was observed for cells with R-type anion channels (Figure 7c). The latter current change is because of a rapid deactivation of anion channels (within 40 msec) (Kolb et al., 1995), after stepping to −180 mV. Because of the distinct gating properties of both channel types, cells dominated by S-type anion channels displayed a linear instantaneous current–voltage relation (Figure 7b), while cells dominated by R-type anion channels exhibited a pronounced peak current (Figure 7d). The relative contribution of both channel types to the conductance increase induced by ABA was resolved using Equation 1 (see Experimental procedures). The anion channel type dominating the ABA-induced conductance increase depended on the magnitude of the response, which was defined as the total increase in anion conductance. In cells with a small response (ΔGinst < 2.5 nS), S-type channels formed the dominant anion conductance (Figure 7e). At intermediate responses (2.5 nS < ΔGinst < 5 nS), S-type channels were dominant in the majority of cells, but in other cells, R-type channels formed the dominant anion conductance (Figure 7e). Finally, an equal number of cells with R- or S-type anion channels were found for large ABA responses (ΔG > 5 nS, Figure 7e).

Figure 7.

Contribution of R- and S-type anion channels to the ABA-induced increase in plasma membrane conductance.

(a) Current trace of guard cell exposed to ABA in an experimental solution containing 5 mm KCl. The plasma membrane was clamped from a holding potential of −100 mV to a test potential of −180 mV, which caused an instantaneous increase of inward current, followed by a time-dependent activation of inward rectifying K+ channels.

(b) ABA-induced change in plasma membrane current, plotted against the clamp voltage for the same cell as in (a). The linear current–voltage relation indicates that the instantaneous current is carried by S-type anion channels.

(c) As in (a), but now the inward current instantaneously decreased at −180 mV followed by a slow activation of inward rectifying K+ channels.

(d) ABA-induced change in plasma membrane current, plotted against the clamp voltage for the same cell as in (c). Inward currents peak around −120 mV, indicating that the current in this cell was mainly carried by R-type anion channels. The instantaneous current–voltage plots of (b) and (d) were fitted with the following equation: inline image, where Gslow and Grapid represent the conductivity of S- and R-type anion channels, Vhalf and δ are the half maximal activation potential and the gating charge of rapid anion channels, and F, R and T have their usual meaning.

(e) Number of cells, which had either S-type (Gslow) or R-type anion channels (Grapid) as the dominant anion channel, binned for five classes of ABA-induced increases in plasma membrane conductance (Gtotal). The values of Gslow and Grapid were determined from 34 experiments of which the instantaneous current–voltage plots were fitted with the equation above. Note that S-type anion channels are dominant in cells with a small response to ABA, while cells with a larger ABA response also can have R-type anion channels as the dominant anion conductance.

Guard cell responses to short of ABA pulses

To test if the activation of anion channels required the continuous presence of ABA, we applied pulses of the phytohormone shorter than 150 sec. Figure 8(a) depicts the current trace of a cell clamped at −100 mV and exposed to 50 µm ABA for 60 sec. Following the ABA pulse, the inward current transiently increased, reaching a peak value approximately 4.5 min after the onset of hormone application. The same time course was determined when responses to short ABA pulses were averaged (Figure 8b). The chain of events triggered by ABA pulses thus showed kinetics similar to those induced by continuous ABA supply (Figure 2e). Upon application of an ABA pulse, however, a small outward current followed the initial transient in inward current (Figure 8b), while an inward current remained with prolonged ABA supply (Figure 2e).

Figure 8.

Guard cells responses to short pulses of ABA.

(a) Plasma membrane current of a guard cell constantly clamped to −100 mV in a perfusion solution containing BaCl2; 50 µm ABA was applied for 60 sec, as indicated by the bar below the trace. The pulse of ABA induced a transient increase in inward current peaking 4.5 min after start of the ABA application.

(b) Average change in the plasma membrane current of guard cells exposed to short (30–150 sec) pulses of ABA (10–50 µm). The average was calculated for eight experiments, in which ABA was applied at t = 0; error bars represent SE. The membrane current was normalized for the peak current (Im, peak = −1). ABA induced a transient change in current, which peaked between 4 and 5 min after start of hormone application. Note that average current change becomes positive relative to the pre-stimulus value 10 min after the ABA application.

(c) Plasma membrane current of a single guard cell (different from the cell shown in (a)) exposed two times to 50 µm ABA for 30 sec, as indicated by the bars below the trace. The current trace is interrupted by 16 min. Note that both ABA pulses triggered current changes with a similar time coarse, but with an increased amplitude during the second application.

Finally, we examined if successive ABA pulses could reproducibly induce current transients in guard cells. The cell displayed in Figure 8(c) was exposed to 50 µm ABA for 30 sec; this pulsed ABA treatment was repeated after 30 min. Both ABA applications resulted in current transients with a similar time course; however, the amplitude increased during the second ABA application (Figure 8c). The ‘run up’ in the latter experiment, together with the ‘run down’ in Figure 6, shows that the responsiveness of a single guard cell can change in time.

Discussion

Heterogeneity in guard cell responsiveness to ABA

Exogenously applied ABA induced stomatal closure, but stomata within a leaf did not respond uniformly. Even neighbouring stomata, sometimes, were found to differ in their sensitivity towards the stress hormone (Figure 1). A variable ABA sensitivity has already been reported for stomata in epidermal strips (Raschke, 1987). In our studies, the variation in responsiveness did not correlate with the localization of the stomata within the leaf, indicating that this feature is not strictly coupled to the well-documented patchiness in stomatal opening (Mott and Buckley, 2000). Mutant analysis showed that the sensitivity to ABA can change more than 10-fold with a single gene mutation (Hugouvieux et al., 2001; Klein et al., 2003; Pei et al., 1998; Roelfsema and Prins, 1995). Natural variation in the transcription rate of such genes or in the phosphorylation state of the encoded proteins may underlie variations in guard cell sensitivity to ABA.

In agreement with the variations in hormone sensitivity of stomata, some guard cells depolarized upon exposure to ABA, while the membrane potential of others remained unchanged. Similar results have been obtained for ABA-induced Ca2+ signals in guard cells (Allen et al., 1999; Gilroy et al., 1991; McAinsh et al., 1990). A variable hormone sensitivity of guard cells could be beneficial for plants, as it provides a mechanism for fine-tuning the ABA-mediated drought response. At a moderate increase in the ABA level, only a small population of stomata will close, while higher hormone concentrations sequentially will close other populations. This all-or-none response may be beneficial for the plant, as it could prevent oscillations in stomatal aperture, which hamper an optimal CO2 supply for photosynthesis. Such oscillations occur at small stomatal apertures (Kaiser and Kappen, 2001) and thus would be inevitable if all stomata reduce their aperture in response to ABA. When instead some stomata close while others remain open, oscillations will be prevented and stomata can still respond to CO2 even at high concentrations of ABA (Leymarie et al., 1998).

Activation of R- and S-type anion channels

In all ABA-responsive cells, the stress hormone increased the activity of ion channels that activated instantaneously. Based on the changes of the instantaneous current–voltage relation, the channels activated by ABA were identified as R- and S-type anion channels. Both channel types facilitate the efflux of anions and depolarize the plasma membrane (Keller et al., 1989; Schroeder and Hagiwara, 1989). Previous reports on guard cells in isolated epidermal strips and protoplasts thereof revealed that ABA increases the conductance of S-type anion channels in guard cells of Arabidopsis thaliana (Pei et al., 1997) and Nicotiana benthamiana (Grabov et al., 1997). An effect of ABA on R-type anion channels, however, had not been recognized before.

R- and S-type anion channels have been described for guard cells of V. faba (Keller et al., 1989; Schroeder and Hagiwara, 1989; Schroeder and Keller, 1992), as well as for A. thaliana (Pei et al., 1997, 2000), indicating that both channel types are conserved within the plant kingdom. In V. faba, the properties of R- and S-type anion channels are similar with respect to their selectivity and single-channel conductance, which led to the hypothesis that both channels represent different gating modes of a single protein (Dietrich and Hedrich, 1994). Both anion channel types however display an obvious difference in gating characteristics and voltage dependence (Linder and Raschke, 1992; Schroeder and Keller, 1992). The open probability of R-type anion channels strongly decreases at membrane potentials negative of −100 mV, a feature not found for S-type channels. Guard cells in intact plants can have membrane potentials more negative than −100 mV (Roelfsema et al., 2001). In the absence of gating modifiers, such as extracellular anions (Dietrich and Hedrich, 1998), R-type channels, therefore, are not capable of inducing an initial depolarization. However, once a guard cell becomes depolarized, the activation of R-type channels can boost stomatal closure. We found that guard cells displaying a moderate response to ABA predominantly activated S-type anion channels, while cells with a large response showed the activity of R-type anion channels, too. This suggests a role for R-type anion channels in fast ABA-induced stomatal closure.

Regulation of voltage-dependent K+ channels

Guard cells in intact plants apparently respond to ABA with the transient activation of two types of anion channels. As a result, the plasma membrane depolarizes and the activity of K+ channels is altered through voltage-dependent regulation. In addition, ABA can alter the maximum conductance of voltage-dependent K+ channels. Previously, ABA has been shown to increase the activity of outward rectifying K+ channels (Blatt and Armstrong, 1993; Lemtiri-Chlieh and MacRobbie, 1994), but this response was not observed in patch clamp experiments by others (Schwartz et al., 1994). In intact plants, only few cells showed a conductance increase of outward K+ channels, while in the majority of cells, the conductance remained unchanged or decreased. The ABA stimulation of outward K+ channels in epidermal strips depends on an alkalinization of the cytoplasm (Blatt and Armstrong, 1993). In intact plants, such an ABA-dependent cytoplasmic pH-change may thus not occur.

ABA also inhibits inward rectifying K+ channels of guard cells in epidermal strips (Blatt and Armstrong, 1993) and protoplasts thereof (Lemtiri-Chlieh and MacRobbie, 1994; Schwartz et al., 1994). For guard cells in intact plants, the effect of ABA on these channels could not be precisely determined because of overlap with currents carried by S-type anion channels. However, if ABA would have inhibited inward K+ channels, hyperpolarized cells would have become more hyperpolarized (Roelfsema and Prins, 1998). For guard cells in intact plants (Figure 2d), this was not observed; instead, ABA depolarized guard cells to potentials where inward K+ channels are inactive.

Mutants of A. thaliana lacking the voltage-dependent inward K+ channel Arabidopsis thaliana 1 (KAT1; Szyroki et al., 2001), or the outward K+ channel guard cell outward rectifying K+ channel (GORK; Hosy et al., 2003) display stomatal movements similar to those of wild-type plants. The mutants lacking GORK are completely devoid of time-activated outward K+ channels, but still close in response to darkness and ABA, although the latter response is slowed down to some extent (Hosy et al., 2003). Apparently, changes in the conductance of voltage-dependent K+ channels only have a small effect on stomatal movement. The activation of anion channels however depolarizes guard cells and can alter the direction of K+ transport across the plasma membrane. This leads to the conclusion that anion channels represent a prime target for ABA action in guard cells.

Transient efflux of Cl

The transient ABA activation of anion channels is well in agreement with ion-selective electrode recordings of the apoplastic Cl concentration in intact leaves (Felle et al., 2000). Furthermore, a temporarily increase of the guard cell Cl conductance was already predicted based on tracer flux experiments with epidermal strips (MacRobbie, 1981). Apparently, the stress hormone causes a large, but transient, efflux of anions followed by a small but steady extrusion. This will cause rapid reduction in the stomatal aperture (as shown in Figure 1b), but does not cause complete stomatal closure. The following low, but persisting, activation of anion channels may prevent re-opening of the stomata. An incomplete stomatal closure allows a rapid re-opening of the stomata when the ABA level in the leaf drops again (Cummins et al., 1971). In contrast, a large and steady activation of anion channels would have resulted in a complete loss of guard cell turgor and in turn complete stomatal closure. In the latter situation, re-opening of stomata would be a tardy process, as the stomata would have to overcome the ‘Spannungsphase’ before they start opening again (Sharpe et al., 1987).

Experimental procedures

Guard cell measurements on intact plants

Broad bean (V. faba L. cv. Grünkernige Hangdown, Gebag, Hannover, Germany) plants were grown in a green house. A leaf of a 4–6-week-old plant was mounted with the adaxial side to a Plexiglas holder in the focal plane of an upright microscope (Axioskop 2FS, Carl Zeiss, Göttingen, Germany). The cells were visualized with a water immersion objective (Achroplan 40×/0.80 W, Carl Zeiss). The solution between the objective and the leaf surface (volume 0.3 ml) was constantly exchanged at flow rate of 1.5 ml min−1. The standard experimental solution contained 5 mm KCl, 5 mm potassium citrate (pH 5.0), 0.1 mm CaCl2 and 0.1 mm MgCl2; where indicated 5 mm BaCl2 was added instead of 5 mm KCl, (±) ABA was obtained from Lancaster (Newgate, UK) and added at a concentrations ranging from 1 to 50 µm. The leaves were illuminated, at an area with a diameter of 20 mm, by the microscope lamp (HAL 12 V/100 W, Carl Zeiss) at a photon flux density of 500 µmol m−2 sec−1 at the adaxial side, which corresponds to a density of 20 µmol m−2 sec−1 at the abaxial side of the leaf.

Guard cells were impaled with double-barreled electrodes pulled from borosilicate glass capillaries and filled with 300 mm KCl or 300 mm CsC1, as described previously by Roelfsema et al. (2001). The reference electrode (300 mm KCl agarose bridge) was placed in the solution on the leaf surface. A potential difference may exist between the guard cell wall and perfusion solution; this potential was recorded with blunt electrodes brought in contact with the guard cell wall (Roelfsema et al., 2001). An average surface potential of −4 mV was measured; this potential was not affected by ABA. Both barrels of the intracellular electrode were connected via Ag/AgCl half cells to a microelectrode amplifier (VF-102, Bio-Logic, Claix, France); the membrane potential was clamped to test voltages using a differential amplifier (CA-100, Bio-Logic). The data were filtered at 300 Hz and sampled at 1 kHz using the same system and software as described earlier by Roelfsema et al. (2001).

Although the approach used here is far less invasive than the measurements carried out previously with epidermal strips or turgor-free protoplasts, the impalement of a single cell introduces electrical leaks that may affect the membrane potential. First, microelectrode amplifiers only have a confined resistance and conduct a leak current. The amplifiers used here have a resistance of 100 GΩ and will cause a leak current of −2 pA at −100 mV. The depolarization resulting from this current will depend on the resistance of the plasma membrane. For V. faba guard cells in intact plants, the highest resistance is found in the voltage range from −100 to −70 mV; here, a value of approximately 5 GΩ was determined (Roelfsema et al., 2001). A leak current of −2 pA through the amplifiers, thus, will cause a maximum depolarization of −10 mV. Second, a leak current may be conducted because of an imperfect connection between glass electrode and the plasma membrane. Such a leak conductance will be non-selective and conduct current over the whole voltage range. It, therefore, cannot be distinguished from voltage-independent ion channels and will depolarize the membrane. Note, however, that a large percentage of cells became hyperpolarized after the removal of ABA (Figure 2c,d). The conductance of the voltage-independent channels and leaks is small in hyperpolarized cells (approximately 200 pS, Roelfsema et al., 2001), we therefore conclude that, in these cells, the connection between the electrode and plasma membrane was tight. However, a number of cells remained depolarized after ABA washout; for the latter cells, the possibility of an impalement-induced leak conductance cannot be excluded.

Numerical analysis

Instantaneous current–voltage relations were fitted with sigmaplot 2000 (SPSS Science, Chicago, IL, USA) using the following equation:

image(1)

where Im is the membrane current, Vm is the membrane potential, Vrev is the reversal potential, Gslow is the voltage-independent conductance, Grapid is the voltage-dependent conductance, δ is the gating charge, F is the Faraday constant, Vhalf is the half maximal activation potential, R is the gas constant and T is the temperature. Current–voltage relations were fitted, assuming that Erev = 0 mV and using the following constraints: Gslow > 0; Grapid > 0; −150 mV < Vhalf < −50 mV and 1 < δ < 5. Only instantaneous current–voltage relations, which could be fitted with a regression coefficient larger than 0.90, were used for further analysis.

Acknowledgements

We thank R. Steinmeyer (University of Würzburg) for help with the analysis, P. Dietrich (University of Würzburg) and D. Sanders (University of York) for their helpful discussions. This research was supported by grants of the Deutsche Forschungsgemeinschaft and the Körber award to R.H.

Notes added in proof

During the reviewing process, a paper has appeared, which confirms part of our observations: Raschke, K., Shabahang, M. and Wolf, R. (2003) The slow and the quick anion conductance in whole guard cells: their voltage-dependent alternation, and the modulation of their activities by abscisic acid and CO2. Planta, 217, 639–650.

This paper is dedicated to Hidde B.A. Prins.

Ancillary