The role of calcium in ABA-induced gene expression and stomatal movements

Authors

  • Alex A. R. Webb,

    Corresponding author
      For correspondence (fax +44 1223 333953; e-mail aarw2@cam.ac.uk).
    Search for more papers by this author
    • Current address: Department of Plant Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EA, UK.

  • Mark G. Larman,

    Search for more papers by this author
    • Current address: Department of Anatomy and Developmental Biology, University College London, Gower Street, London WC1E 6BT, UK.

  • Lucy T. Montgomery,

    1. Department of Biological Sciences, Institute of Environmental and Natural Sciences, Lancaster University, Bailrigg, Lancaster LA1 4YQ, UK
    Search for more papers by this author
  • Jane E. Taylor,

    1. Department of Biological Sciences, Institute of Environmental and Natural Sciences, Lancaster University, Bailrigg, Lancaster LA1 4YQ, UK
    Search for more papers by this author
  • Alistair M. Hetherington

    1. Department of Biological Sciences, Institute of Environmental and Natural Sciences, Lancaster University, Bailrigg, Lancaster LA1 4YQ, UK
    Search for more papers by this author

For correspondence (fax +44 1223 333953; e-mail aarw2@cam.ac.uk).

Summary

There is much interest in the transduction pathways by which abscisic acid (ABA) regulates stomatal movements (ABA-turgor signalling) and by which this phytohormone regulates the pattern of gene expression in plant cells (ABA-nuclear signalling). A number of second messengers have been identified in both the ABA-turgor and ABA-nuclear signalling pathways. A major challenge is to understand the architecture of ABA-signalling pathways and to determine how the ABA signal is coupled to the appropriate response. We have investigated whether separate Ca2+-dependent and -independent ABA-signalling pathways are present in guard cells. Our data suggest that increases in [Ca2+]i are a common component of the guard cell ABA-turgor and ABA-nuclear signalling pathways. The effects of Ca2+ antagonists on ABA-induced stomatal closure and the ABA-responsive CDeT6-19 gene promoter suggest that Ca2+ is involved in both ABA-turgor signalling and ABA-nuclear signalling in guard cells. However, the sensitivity of these pathways to alterations in the external calcium concentration differ, suggesting that the ABA-nuclear and ABA-turgor signalling pathways are not completely convergent. Our data suggest that whilst Ca2+-independent signalling elements are present in the guard cell, they do not form a completely separate Ca2+-independent ABA-signalling pathway.

Introduction

There is good evidence that the calcium ion is a second messenger in the transduction pathways by which ABA regulates the size of the stomatal pore (Leung and Giraudat, 1998; McAinsh and Hetherington, 1998). This ABA-turgor signalling pathway is well characterized, and probably includes increases in cytosolic free calcium, [Ca2+]i (Allan et al., 1994; Gilroy et al., 1990; Grabov and Blatt, 1998; Irving et al., 1992; McAinsh et al., 1990, McAinsh et al., 1992; Schroeder and Hagiwara, 1990); increases in pH (Blatt and Thiel, 1993); inositol (1,4,5) trisphosphate (Ins(1,4,5)P3) (Lee et al., 1996); cyclic adenosine 5′-diphosphoribose (cADPR) (Leckie et al., 1998); a farnesyltransferase (Pei et al., 1998); phospholipase D (Jacob et al., 1999); inositol hexakisphosphate (Lemtiri-Chlieh et al., 2000); and the activity of protein kinases (Hey et al., 1997; Pei et al., 1996; Li et al., 2000) and protein phosphatases (Gosti et al., 1999; Hey et al., 1997).

The ABA signal can also be relayed to the nucleus of guard cells to bring about alterations in the pattern of gene expression (Taylor et al., 1995). Stimulation of the ABA-nuclear signalling pathway by exogenous ABA increased expression of dehydrin-like transcripts in the guard cells of both Vicia faba (Shen et al., 1995) and pea (Hey et al., 1997). In Arabidopsis guard cells, ABA increased the expression of a water-transport protein encoded by the AthH2 gene (Kaldenhoff et al., 1995). Similarly dehydration, which elevates endogenous ABA levels, has been demonstrated to alter the levels of expression of genes encoding proteins involved in carbon metabolism and ion transport in guard cells of potato (Kopka et al., 1997).

The nature of the guard cell ABA-nuclear signalling pathways are less well defined than the ABA-turgor transduction pathways. Reversible protein phosphorylation has been implicated in transduction of the ABA signal to both the nucleus and the membrane bound ion transporters controlling stomatal movements (Hey et al., 1997). However, despite the central role for increases in [Ca2+]i in the control of stomatal movements, there is no evidence that alterations in [Ca2+]i are a component of the guard cell ABA-nuclear transduction pathway (Shen et al., 1995).

Notable advances in dissecting the role of second messengers in the ABA-nuclear signalling pathway in other cell types have been obtained by assaying the activity of promoter–reporter gene fusions in heterologous expression systems (Sheen, 1996; Wu et al., 1997). These investigations have demonstrated that calcium-dependent protein kinases are probably involved in control of the ABA-responsive barley HVA1 gene promoter (Sheen, 1996), and that regulation of the kin2 and rd29A gene promoters by ABA may involve increases in [Ca2+]i mediated by Ins(1,4,5)P3 and cADPR (Wu et al., 1997). Furthermore, reversible protein phosphorylation and dephosphorylation appear to be common features of ABA-regulated HVA1, kin2 and rd29A promoter activity (Sheen, 1996, Sheen, 1998; Wu et al., 1997).

In this study we have investigated the involvement of Ca2+ in the guard cell ABA-turgor and ABA-nuclear signalling pathways. These studies aimed to determine whether separate Ca2+-dependent and Ca2+-independent ABA signal-transduction pathways operate in stomatal guard cells (Allan et al., 1994; Romano et al., 2000).

Results

ABA-induced reductions in guard cell turgor in Cammelina communis require increased [Ca2+]i

Cytosolic BAPTA prevents ABA-induced increases in [Ca2+]i and inhibits guard cell movements

The resting [Ca2+]i of Commelina communis guard cells that had been loaded with fura-2 alone was approximately 80–150 nm(Figure 1a; n = 11). Challenge of these guard cells with 1 µm ABA caused increases in [Ca2+]i ranging from approximately 50–500 nm above ‘resting’ values in seven of the 11 cells tested (Figure 1a). Both sustained and transient increases in [Ca2+]i were observed. The transient increases in [Ca2+]i lasted 3–20 min before returning to the values observed before treatment with ABA. The ABA-induced increases in [Ca2+]i were not homogeneous across the cell. A representative example of the spatial analysis of ABA-induced increases in [Ca2+]i is shown in Figure 2(a). The mean whole-cell increase in [Ca2+]i in this cell 10 min after addition of ABA was approximately 250 nm above resting. However, in some areas of the cell increases of approximately 1000 nm or more were observed (Figure 2a). These responses to ABA are similar to our previous experiments using the indicator indo-1 (McAinsh et al., 1992).

Figure 1.

Effect of cytosolic loading with fura-2 or fura-2 and BAPTA on the ABA-turgor signalling pathway in guard cells of C. communis.

Single guard cells of closed stomatal pairs of C. communis were micro-injected and fura-2 (a) or fura-2 and BAPTA (b) was loaded into the cytosol. The upper trace of each graph represents the half-stomatal aperture of the loaded (●) and unloaded (○) guard cells of the same stomatal pair. The lower trace of each graph is the [Ca2+]i of the loaded cell recorded by ratio photometry of fura-2 fluorescence. The stomata were irrigated with CO2-free MES-KCL (open boxes), MES-KCL and 1 µm ABA (closed boxes), and subsequently MES-KCL and 1 mm CaCl2 (hatched boxes). Bar, 10 min. Each graph shows representative examples from 11 experiments.

Figure 2.

Digital ratio imaging of the effects of 1 µm ABA on [Ca2+]i in guard cells of C. communis loaded with fura-2 alone or fura-2 and BAPTA.

Single guard cells of stomatal pairs of C. communis were micro-injected, and fura-2 alone (a) or fura-2 and BAPTA (b) was loaded into the cytosol. Images were pseudocolour-coded according to a predetermined in vitro calibration curve. n = 10.

The presence of 1,2,-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid (BAPTA) in the cytosol resulted in a completely different pattern of responses to ABA. The range of the resting [Ca2+]i of cells that had been co-loaded with fura-2 and BAPTA was <10–100 nm(Figure 1b; n = 11). Importantly, the whole-cell mean [Ca2+]i was unaltered by ABA in any of the cells tested (n = 11; Figure 1b). Furthermore, detectable localized increases in [Ca2+]i in response to ABA were also prevented by the presence of BAPTA in the cytosol (Figure 2b).

Cytosolic loading of fura-2 had no significant effect on the ability of ABA to induce stomatal closure. 30 min after the addition of 1 µm ABA, the half stomatal apertures of the fura-2-loaded and unloaded cells were 2.6 ± 0.6 and 1.8 ± 0.7 µm, respectively (±SEM; n = 11, P = 0.28). In all the fura-2-loaded cells where an increase in [Ca2+]i was measured, the loaded cell closed at the same rate as the unloaded cell (0.25 ± 0.06 µm min−1 during the first 5 min after addition of ABA; n = 7). Four of the fura-2-loaded cells showed no increase in [Ca2+]i in response to ABA. Two of these failed to close completely; the other two closed more slowly (0.08 ± 0.06 µm min−1 in the first 5 min after addition of ABA) than the unloaded cells (0.315 ± 0.04 µm min−1 in the first 5 min) (n = 2). Importantly, there appeared to be a correlation between an increase in [Ca2+]i and closure. Those cells in which a detectable increase in [Ca2+]i was measured closed at the same rate as the unloaded cells; those fura-2-loaded cells in which there was not a detectable increase in [Ca2+]i failed to close, or closed more slowly.

Figures 1(b) and 3 demonstrate that the presence of BAPTA in the cytosol inhibited ABA-induced reductions in guard-cell turgor. Inhibition of ABA-induced reductions in guard-cell turgor by cytosolic BAPTA was observed in all the cells tested (n = 11). In nine of the cells tested, BAPTA completely prevented the reduction in guard-cell turgor in response to ABA. The remaining two BAPTA-loaded cells closed more slowly (0.1 ± 0.01 µm min−1 during the first 5 min after addition of ABA) than the unloaded cells (0.21 ± 0.06 µm min−1 during the first 5 min after addition of ABA) (n = 2) in response to ABA. The unloaded cell acted as a powerful control, as individual guard cells of a stomatal pair tend to respond symmetrically to stimuli.

Figure 3.

Effect of BAPTA on the rate of half-stomatal aperture closure in C. communis following the addition of 1 µm ABA.

Single guard cells of stomatal pairs of C. communis were micro-injected, and fura-2 and BAPTA was loaded into the cytosol. The rate of closure following addition of 1 µm ABA was measured in the BAPTA-loaded (●) and unloaded (○) cells from the same stomatal pair. Bars, ±SEM; n = 11.

BAPTA significantly reduced the rate of closure and the final aperture achieved (Figure 3). 30 min after the addition of 1 µm ABA, the mean half stomatal apertures of BAPTA-loaded and unloaded cells were 3.4 ± 0.3 and 1.7 ± 0.3 µm, respectively (P < 0.001).

The effect of BAPTA was reversed by elevated [Ca2+]e

By increasing the extracellular calcium concentration [Ca2+]e to 1 mm, it was possible to partially overcome the effect of BAPTA and induce reductions in guard-cell turgor (Figure 1b). Elevations in [Ca2+]e caused reductions in turgor only in those cells where a subsequent elevation of [Ca2+]i was detected. Elevating [Ca2+]e caused either sustained or transient increases in [Ca2+]i in five of the eight (62.5%) BAPTA-loaded guard cells tested. These increases ranged from 50 to 500 nm above resting, and the transient increases lasted 5–10 min. The presence of BAPTA in the cytosol prevented oscillations in [Ca2+]i in response to elevations in [Ca2+]e in all cells tested (Figure 1b). This contrasted markedly with the response of the cells that did not contain BAPTA in the cytosol. In all cells loaded with fura-2 alone, increases in [Ca2+]e caused elevations in [Ca2+]i that were oscillatory, and the peak height of [Ca2+]i increase ranged from 0.5–1 µm(Figure 1a).

Role of Ca2+ in ABA signalling in guard cells of Arabidopsis

Stimulus-induced increases in [Ca2+]i in guard cells

Elevation of [Ca2+]e to 1 mm increased [Ca2+]i as reported by either fluorescence photometry of Ca green-1 dextran (n = 3) (Figure 4), or digital ratio imaging and photometry of fura-2 fluorescence (n = 5) (Figure 5), in all cells tested. These increases in [Ca2+]i were up to 1 µm(Figure 5), and were reversible by returning [Ca2+]e to 0.01 mm (data not shown). An irregular pattern of oscillations in [Ca2+]i in response to 1 mm CaCl2 were observed in four of the eight cells tested (Figure 4). Imaging of a 1 mm CaCl2-induced oscillation in [Ca2+]i revealed that [Ca2+]e-induced [Ca2+]i oscillations were spatially heterogeneous (Figure 5). In the cell shown in Figure 5, [Ca2+]i peaked 1 min earlier in the lower region of the cell than in the upper region. In those cells in which oscillations were not detected, 1 mm CaCl2 caused sustained increases in [Ca2+]i of up to 1 µm. These increases were maintained until the 1 mm CaCl2 was removed (data not shown).

Figure 4.

Stimulus-induced oscillations in [Ca2+]i in guard cells of A. thaliana.

Changes in [Ca2+]i in response to 1 mm CaCl2 were measured using fluorescence photometry of Ca green-1 dextran (10 000 MW). Increases in [Ca2+]i resulted in an increase in fluorescence emission of the Ca-green-1.

Figure 5.

Digital ratio imaging of an oscillation in [Ca2+]i in a guard cell of A. thaliana (Col) using fura as the calcium indicator.

Images were captured every minute and are displayed left to right, top to bottom. Guard cells were challenged with 1 mm CaCl2 at the arrow. The fluorescence ratio images were pseudocolour-coded according to a predetermined in vitro calibration curve.

Treatment with 10 µm ABA caused detectable increases in [Ca2+]i in three of the five cells tested (Figure 6). ABA caused sustained increases in [Ca2+]i ranging from approximately 200–400 nm above resting. These increases were sustained for at least 5 min after addition of ABA.

Figure 6.

ABA-induced increase in [Ca2+]i in a single guard cell of A. thaliana (Col).

Changes in [Ca2+]i in response to 10 µm ABA were recorded by capturing digital ratio images of fura fluorescence and calculating the mean whole cell [Ca2+]i. Cells were irrigated in MES-KCL (open box) or MES-KCL and 10 µm ABA (black box) (n = 5).

Role of Ca2+ in ABA-turgor signalling

ABA, CO2, darkness and exogenous Ca inhibited the opening of stomata of A. thaliana (Col) transformed with cDet6-19::GUS (Figure 7a). These data demonstrate that ABA and CaCl2 both inhibited stomatal opening (P > 0.05), and that the ABA-turgor pathway was functional under conditions similar to those used to perform the histochemical GUS assay. This suggests that the CDeT6-19::GUS insert did not interfere with normal signalling in the guard cell.

Figure 7.

Stomatal movements in detached epidermis of A. thaliana (Col) transformed with cDeT6–19::GUS.

Epidermal pieces with initially closed stomata were incubated in CO2-free MES-KCL in the light at 23°C for 3 h. At the end of the incubation period stomatal apertures were measured. Values represent means of 120 measurements. Each experiment was repeated three times. Bars indicate SEM (a). In (b) the least significant difference (t = 0.05) was 0.1 µm, which is smaller than the symbols and therefore is not shown.

(a) CO2-free MES-KCL with 700 µl l−1 CO2, 1 mm CaCl2, 10 µm ABA or darkness.

(b) CO2-free MES-KCL plus (▾) or minus (●) 2 mm BAPTA and a range of ABA concentrations.

To investigate whether Ca2+ was critical in ABA-turgor signalling in guard cells of A. thaliana (Col), we determined the effect of 2 mm BAPTA in the external medium on ABA-induced inhibition of stomatal opening. BAPTA completely abolished ABA-induced inhibition of stomatal opening in detached epidermis of transformed A. thaliana(Figure 7b). This effect was apparent at all the concentrations of ABA investigated (10−7−10−4 m), and was statistically significant (P < 0.001).

Ca2+ is involved in ABA-nuclear signalling in guard cells

GUS expression driven by the CDeT6-19 gene promoter was increased in the guard cells of epidermal strips that had been incubated in 10 µm ABA for 3 h (Figure 8a,b). This ABA-induced CDeT6-19 promoter activity was inhibited by compounds that may limit stimulus-induced increases in [Ca2+]i. For example, incubating epidermal strips in the presence of 100 µm of the putative Ca2+-channel blockers LaCl3 and verapamil inhibited the ability of ABA to induce GUS activity under the control of the CDeT6-19 promoter (Figure 8e,g). Chelating external calcium using 2 mm BAPTA also inhibited ABA-induced GUS expression driven by the CDeT6-19 promoter (Figure 8c).

Figure 8.

Figure 8.

Regulation of CDeT6-19 promoter activity in guard cells of A. thaliana (Col).

GUS activity under the control of the CDeT6-19 promoter was visualized in epidermal peels using a histochemical stain (a–k). Prior to histochemical staining for GUS activity, the detached epidermis was incubated for 3 h in MES-KCL (a) , containing either 10 µm ABA (b) , 10 µm ABA and 2 mm BAPTA (c) , 2 mm BAPTA (d) , 10 µm ABA and 100 µm verapamil (e) , 100 µm verapamil (f) , 10 µm ABA and 100 µm LaCl3(g) , 100 µm LaCl3(h) , 1 mm CaCl2(i) , 10 mm CaCl2(j) , 700 µl l−1 CO2(k). Images were obtained at ×10 or ×20. The effect of 5 mm BAPTA on activity of the GUS protein was investigated using a fluorometric assay (l). Proteins extracted from leaves of A. thaliana transformed with a 35SGUS chimeric gene construct were incubated in the absence (○) or presence (●) of 5 mm BAPTA at 37°C in the presence of the MUG substrate. GUS activity caused hydrolysis of MUG to MU, resulting in an increase in fluorescence (Ex, 350; Em, 455). The data shown here are from one experiment. This experiment was repeated three times with identical results. Similar results were obtained when the effect of 5 mm BAPTA on the activity of commercial GUS was assayed (data not shown).

Figure 8.

Figure 8.

Regulation of CDeT6-19 promoter activity in guard cells of A. thaliana (Col).

GUS activity under the control of the CDeT6-19 promoter was visualized in epidermal peels using a histochemical stain (a–k). Prior to histochemical staining for GUS activity, the detached epidermis was incubated for 3 h in MES-KCL (a) , containing either 10 µm ABA (b) , 10 µm ABA and 2 mm BAPTA (c) , 2 mm BAPTA (d) , 10 µm ABA and 100 µm verapamil (e) , 100 µm verapamil (f) , 10 µm ABA and 100 µm LaCl3(g) , 100 µm LaCl3(h) , 1 mm CaCl2(i) , 10 mm CaCl2(j) , 700 µl l−1 CO2(k). Images were obtained at ×10 or ×20. The effect of 5 mm BAPTA on activity of the GUS protein was investigated using a fluorometric assay (l). Proteins extracted from leaves of A. thaliana transformed with a 35SGUS chimeric gene construct were incubated in the absence (○) or presence (●) of 5 mm BAPTA at 37°C in the presence of the MUG substrate. GUS activity caused hydrolysis of MUG to MU, resulting in an increase in fluorescence (Ex, 350; Em, 455). The data shown here are from one experiment. This experiment was repeated three times with identical results. Similar results were obtained when the effect of 5 mm BAPTA on the activity of commercial GUS was assayed (data not shown).

Incubation of epidermal peels from CDeT6-19GUS transformed plants with 1 mm CaCl2(Figure 8i), in the absence of exogenous ABA, failed to induce detectable GUS activity in guard cells. Treatment with 10 mm CaCl2 caused weak induction of GUS activity in guard cells (Figure 8j). However, even in the presence of the ionophore A23187, [Ca2+]e was unable to drive the high levels of GUS expression observed in the presence of ABA (data not shown). Therefore it appears that an increase in [Ca2+]e was sufficient to induce weak activity of the CDeT6-19 promoter, but strong induction of the promoter required the presence of exogenous ABA.

In guard cells of C. communis and Vicia faba, both ABA and CO2 utilize calcium-based transduction chains (Brearley et al., 1997; Webb et al., 1996) to bring about reductions in stomatal pore size. Therefore we investigated whether CO2 can induce the activity of the CDeT6-19 promoter in guard cells of A. thaliana. Neither incubation of epidermal strips for 3 h with aeration by air containing 700 µl l−1 CO2(Figure 8k), nor growth of plants for 3 days in air containing 700 µl l−1 CO2, induced GUS activity under the control of the CDeT6-19 promoter (data not shown).

The calcium antagonists and agonists used in this investigation did not interfere with the ability of the GUS gene to report alterations in the activity of the CDeT6-19 gene promoter. This was demonstrated in two ways. First, none of the treatments described above altered GUS expression in the guard cells of 35SGUS transformants of A. thaliana (data not shown). Second, 5 mm BAPTA had no effect on the hydrolysis of 4-methylumbelliferyl-β-d-glucoronide (MUG) to 4-methylumbelliferone (MU), either by recombinant GUS protein extracted from 35SGUS transformants or by commercial β-glucoronidase (Figure 8l; data not shown).

Discussion

[Ca2+]i is required for the full guard-cell turgor response to ABA in C. communis

The presence of BAPTA in the cytosol inhibited ABA-induced reductions in turgor in all cells tested. In nine of 11 cells, BAPTA completely prevented the ABA-induced reduction in turgor. We conclude that an increase in [Ca2+]i is critical for the turgor response to ABA, because none of the cells loaded with BAPTA closed as much as the unloaded cell from the same stomatal pair. Therefore none of the BAPTA-loaded cells had the same rate or extent of closure in response to ABA as the control cells.

BAPTA loading tended to lower resting [Ca2+]i compared to cells loaded with fura-2 alone. We can exclude the possibility that the presence of BAPTA in the cytosol reduced [Ca2+]i below a level required for normal functioning of the guard cell, because BAPTA-loaded cells opened to the same aperture as the unloaded cell from the same stomatal pair (Figure 1).

The effects of cytosolic BAPTA on guard cell turgor are consistent with the documented effects of buffering [Ca2+]i to resting values on the ABA-induced inhibition of IKin (Kelly et al., 1995; Lemtiri-Chlieh and MacRobbie, 1994; Schroeder and Hagiwara, 1989; but see also Romano et al., 2000); activation of anion efflux (Hedrich et al., 1990; Linder and Raschke, 1992; Schroeder and Hagiwara, 1989; Schwarz and Schroeder, 1998); and inhibition of the H+-ATPase (Kinoshita et al., 1995).

Potentially, Ca2+-independent processes also contribute to stomatal closure. For example, the activation of IKout, which is responsible for allowing K+-efflux across the plasma membrane and is thus an essential step in the closure of stomata, is regarded as a Ca2+-independent process (Lemtiri-Chlieh and MacRobbie, 1994). Thus, whilst interacting Ca2+-dependent and -independent ABA-signalling pathways co-reside within the guard cell, our data suggest that at least one Ca2+-dependent event is a prerequisite for complete stomatal closure in response to ABA.

ABA-induced reductions in turgor in the absence of a detectable increase in [Ca2+]i

In four of 22 cells (two BAPTA-loaded and two fura-2-loaded), ABA caused a small reduction in half-aperture without a detectable increase in [Ca2+]i. ABA-induced reductions in turgor in the absence of a detectable increase in [Ca2+]i have been interpreted as evidence for the presence of Ca2+-independent and -dependent signalling pathways (Allan et al., 1994). However, in the present study it was demonstrated that the presence of BAPTA in the cytosol completely or partially inhibited the response to ABA in all cells studied. We conclude that a failure to observe an ABA-induced increase in [Ca2+]i is not evidence for the presence of a totally separate Ca2+-independent ABA signal-transduction pathway. If a completely separate Ca2+-independent pathway were present, BAPTA would not have inhibited reductions in guard-cell turgor. Our data do not exclude the possibility that Ca2+-independent processes can bring about a partial turgor response to ABA.

Romano et al. (2000) observed ABA-induced inhibition of IKin in the absence of a detectable increase in [Ca2+]i. Romano et al. (2000) concluded that the regulation of IKin can occur in a Ca2+-dependent and Ca2+-independent manner, and favoured a model in which these pathways are redundant. Importantly, Romano et al. (2000) detected an ABA-induced inhibition of IKin when 5 mm BAPTA was present in the cytosol. However, this observation may not support the hypothesis that IKin can be regulated in a Ca2+-independent manner, because other researchers have demonstrated that ABA is unable to regulate IKin when higher levels of Ca2+ buffer are present in the cytosol (e.g. 44 mm EGTA or 20 mm BAPTA; Lemtiri-Chlieh and MacRobbie, 1994). Taken together, these data suggest that low concentrations of buffer are able to prevent detectable global increases in [Ca2+]i but are unable to prevent ABA-induced regulation of IKin. However, when higher concentrations of buffer are used, it is possible that large, localized increases in [Ca2+]i, possibly near the plasma membrane, are prevented and thus ABA can no longer regulate IKin

A similar explanation may underlie our observation that ABA evoked a partial response in the absence of a detectable increase in [Ca2+]i in two BAPTA-loaded cells. The buffering due to BAPTA may have been incomplete in the two BAPTA-loaded cells that showed a partial response to ABA, particularly near the tonoplast and plasma membrane where [Ca2+]i transients can be larger and more rapid than the global increase in [Ca2+]i(Figures 1a and 2a). The imaging of near membrane increases in Ca2+ is technically demanding and prone to artefacts (Romano et al., 2000), but there is one report that ABA-induced global increases in [Ca2+]i are preceded by a large increase in Ca2+ near the plasma membrane of guard cells (Grabov and Blatt, 1998). Alternatively, Ca2+-independent signalling components may have evoked the partial response to ABA in the BAPTA-loaded cells. Whatever the explanation, it is clear that BAPTA limited the response to ABA in all cells tested, demonstrating that any Ca2+-independent pathway(s) that may be present are unlikely to be totally separate from the Ca2+-dependent pathway, and probably cannot bring about stomatal closure in the absence of an increase in [Ca2+]i. These data suggest that the Ca2+-dependent pathway is not redundant.

Involvement of Ca2+ in ABA signalling in guard cells of Arabidopsis

Ca2+ is critical for ABA-induced inhibition of stomatal opening

We have provided direct evidence that an increase in [Ca2+]i is a component in the signalling pathways by which ABA regulates stomatal movements in A. thaliana. ABA increased [Ca2+]i in single guard cells of A. thaliana(Figure 6), and BAPTA in the extracellular medium completely abolished the ABA-induced inhibition of stomatal opening (Figure 7b). BAPTA may have limited influx of Ca2+ across the plasma membrane via an ABA- and hyperpolarization-regulated Ca2+ channel (Grabov and Blatt, 1999; Hamilton et al., 2000). However, it is likely that internal stores are also central to ABA-turgor signalling in guard cells (Grabov and Blatt, 1999). It was not possible to test further the role of elevations in [Ca2+]i in ABA signalling in A. thaliana by micro-injecting BAPTA into the cytosol, due to the relatively small changes in half stomatal aperture in this species.

The aperture responses of the stomata of the Columbia ecotype of A. thaliana to ABA, Ca2+ and other stimuli are quantitatively similar to those reported previously for the Landsberg erecta ecotype (Allen et al., 1999a; Allen et al., 1999b; Leymarie et al., 1998; Roelfsema and Prins, 1995; Webb and Hetherington, 1997). We observed ABA-induced and [Ca2+]e-induced increases of [Ca2+]i in the Columbia ecotype of similar magnitudes and forms to those reported previously in the stomata of L. erecta (Allen et al., 1999a; Allen et al., 1999b).

The higher quantum efficiency of fura compared to indo-1 and the cameleon Ca2+-indicator allowed imaging of Arabidopsis guard cells [Ca2+]i with greater spatial resolution than has previously been possible, partially because pixel binning was not required in this study (Allen et al., 1999a; Allen et al., 1999b). We were able to observe that complex spatial–temporal patterns underlie oscillations in [Ca2+]i in guard cells of Arabidopsis(Figure 5). However, the technical difficulties associated with micro-injecting the guard cells of this species precluded obtaining sufficient loaded cells to perform a detailed analysis of these patterns. For the same reasons it was not possible to analyse the effects of calcium agonists on Ca2+ signals in the guard cells of A. thaliana. Such analysis awaits the development of new variants of calcium cameleon with a sufficient quantum efficiency to allow detailed spatial analysis of Arabidopsis guard cell [Ca2+]i, or improvements in micro-injection techniques.

The temporal patterns of the [Ca2+]e-induced oscillations in [Ca2+]i in A. thaliana guard cells were not identical to those we have previously observed in C. communis (McAinsh et al., 1995). In A. thaliana, 1 mm CaCl2 induced irregular oscillations (Figure 4), whereas in C. communis 1 mm CaCl2 induced oscillations that had a characteristic biphasic, ‘spike-shoulder’ pattern. The reasons for these differences are unclear. However, a number of factors are known to affect the pattern of stimulus-induced oscillations in guard cell [Ca2+]i. These include the plasma membrane potential (Grabov and Blatt, 1998; Hetherington et al., 1998; Staxén et al., 1999); ABA (Hetherington et al., 1998; Staxén et al., 1999); [Ca2+]e (McAinsh et al., 1995); and osmotic shock (Hetherington et al., 1998).

Ca2+ is involved in ABA-induced gene expression in guard cells of A. thaliana

We have also provided evidence that elevations in [Ca2+]i form a component of the transduction chain by which ABA regulates the CDeT6-19 promoter. We have shown that ABA increases [Ca2+]i in guard cells of A. thaliana(Figure 6), and that activation of the CDeT6-19 promoter by ABA was inhibited by Ca2+ antagonists (Figure 8). Despite the semi-quantitative nature of the GUS reporter, the effects of the Ca2+ antagonists on cDeT6-19 promoter activity were marked, suggesting that Ca2+ is an important component in the ABA-nuclear signalling pathway(s). LaCl3, verapamil and BAPTA probably inhibited ABA-induced increases in [Ca2+]i by inhibiting Ca2+-influx across the plasma membrane (Grabov and Blatt, 1999; Hamilton et al., 2000). We were unable to confirm whether LaCl3, verapamil and BAPTA prevented ABA-induced increases in [Ca2+]i in A. thaliana. Such experiments would have required larger data sets than we could obtain by micro-injection of fluorescent indicators in to guard cells of A. thaliana.

The effects of the Ca2+-antagonists on ABA-induced gene expression in the guard cell are strikingly similar to the effects of these agents upon the ABA-turgor signalling pathway. For example, BAPTA inhibited both ABA-induced CDeT6-19 activity and ABA-induced inhibition of stomatal opening in A. thaliana(Figures 7 and 8). Similarly, the putative Ca2+ channel blockers, LaCl3 and verapamil both inhibited ABA-induced CDeT6-19 promoter activity (Figure 8), and have previously been shown to prevent ABA-induced inhibition of stomatal opening in C. communis (DeSilva et al., 1985). Taken together, these data suggest that an increase in [Ca2+]i is a common component of the transduction pathways by which ABA regulates CDeT6-19 promoter activity in guard cells of A. thaliana and induces reductions in guard cell turgor.

The histochemical GUS assay is semi-quantitative and does not usefully report small changes in gene expression, therefore it was not possible to determine whether there is an absolute requirement for Ca2+ during ABA-nuclear signalling. More quantitative analysis of the activity of the cDet6-19 promoter might have been achieved using a fluorescent substrate in cell homogenates. However, the techniques for purifying guard cells involve a cold shock during blending or an osmotic shock during protoplasting. The cold and osmotic signalling pathways closely overlap the ABA signalling pathways. Thus purification of the guard cells to produce homogenates would have prevented the ABA signalling pathway being studied in isolation. Our previous studies demonstrated that cold induces cDeT6-19 promoter activity (J.E.T. et al., unpublished observations). We did attempt a quantitative measure of GUS activity in situ using the ImaGene green fluorescent substrate (Molecular Probes; see Experimental procedures) and photometry. This proved unsuccessful because the substrate was unable to enter guard cells. We observed fluorescent signal only from guard cells that had previously been damaged. We concluded that the cell wall prevented entry of the substrate. Other researchers have found that this fluorescent substrate is suited to measuring only very high levels of expression in vivo, and have been unable to quantify changes in gene expression in planta (Fleming et al., 1996).

There is convincing evidence that increases in [Ca2+]i also form a component of the ABA-nuclear signalling pathways in maize leaf protoplasts (Sheen, 1996); etiolated hypocotyls of the phytochrome-deficient tomato mutant aurea (Wu et al., 1997); wheat aleurone (Napier et al., 1989); and chickpea seeds (Colorado et al., 1991). Increases in [Ca2+]i have been implicated in the regulation of a number of ABA-inducible gene promoters including rab18 (Knight et al., 1997); lti78 (Knight et al., 1997; Wu et al., 1997); HVA1 (Sheen, 1996); and kin2 (Wu et al., 1997). Therefore increases in [Ca2+]i appear to act as a component of several of the ABA-nuclear signalling pathways operating in plant cells.

ABA was more effective in activating the CDeT6-19 promoter than increased [Ca2+]e in the absence of exogenous ABA (Figure 8). This represents a difference in the ABA-nuclear signalling pathway compared to the ABA-turgor signalling pathway, because increases in [Ca2+]e and ABA both reduced guard cell turgor in Arabidopsis(Figure 7a; Roelfsema and Prins, 1995; Webb and Hetherington, 1997) and C. communis (McAinsh et al., 1995). Therefore increases in [Ca2+]e, probably by increasing [Ca2+]i(Figures 4 and 5), appear to be able to activate sufficient downstream signalling elements to strongly activate the ABA-turgor signalling pathway, but the same increases in [Ca2+]e appear to only weakly stimulate the ABA-nuclear signalling pathway. These data suggest that there are differences in the architecture of the ABA-turgor and ABA-nuclear signalling pathways operating in guard cells.

There is conflicting evidence concerning the ability of increases in [Ca2+]i to substitute for ABA in the ABA-nuclear pathways operating in other plant cell types. Increases in [Ca2+]i induced by micro-injection of Ca2+ into the cytosol (Wu et al., 1997), or by treatment with the ionophore A23187 (Sheen, 1996), resulted in increased activity of ABA-responsive reporters in heterologous expression assays in the absence of ABA. Similarly, increases in [Ca2+]e in the absence of exogenous ABA have been demonstrated to induce the appearance of ABA-responsive polypeptides (Colarado et al., 1991; Napier et al., 1989). In the present study we were able to induce only weak GUS expression driven by the ABA-responsive CDeT6-19 promoter in guard cells in the absence of ABA. Similarly, Sheen (1996) observed that the HVA1 promoter was not activated in response to 1 mm[Ca2+]e.

In animal cells, both the spatial (Hardingham et al., 1997) and temporal pattern of stimulus-induced increases in [Ca2+]i (Dolmetsch et al., 1997, Dolmetsch et al., 1998; Li et al., 1998) affect the pattern of alteration in gene expression. Particularly striking is evidence that the activation of a specific subset of pro-inflammatory transcription factors in Jurkat T cells depends on the frequency of oscillations in [Ca2+]i (Dolmetsch et al., 1998). One possibility is that elevated [Ca2+]i may be able to substitute for ABA in the ABA-nuclear signalling pathway only if the appropriate ‘Ca2+-signature’ is generated (McAinsh and Hetherington, 1998). A recent report suggests that the correct pattern of oscillations of [Ca2+]i may have a central role in driving reductions in guard-cell turgor (Allen et al., 2000).

Whilst the appropriate Ca2+-signature may be very important for the correct regulation of gene expression in a number of cell types, the failure of elevated [Ca2+]e to induce strong activity of an ABA-responsive promoter in guard cells in this and a previous study (Shen et al., 1995) could have another explanation. There may be Ca2+-independent component(s) of the ABA signal-transduction pathways that are required for full activity of the CDeT6-19 promoter. These Ca2+-independent signalling elements may be activated in response to ABA, but not activated in response to elevated [Ca2+]e or [Ca2+]i. Preliminary data suggesting that a mitogen-activated protein kinase cascade may contribute to ABA signalling in guard cells (Burnett et al., 2000) present a tantalizing clue to the identity of the Ca2+-independent signalling elements.

Experimental procedures

Plant material

Commelina communis plants were grown and mounted for micro-injection as described by Webb et al. (1996). Arabidopsis thaliana (Col), transformed with a chimeric construct containing the sequence between −889 and 114 of the CDeT6-19 gene translationally fused to the Escheria coliβ-glucoronidase A (GUS) gene followed by the NOS termination region, were grown as described by Taylor et al. (1995). A. thaliana (Col) containing a chimeric 35SCaMV::gusA (35SGUS) (a generous gift of Professor D. Bartels, Bonn, Germany) were grown under the same conditions.

Micro-injection of guard cells of C. communis

Iontopheretic micro-injection of fura-2 or fura-2 with BAPTA was performed as described by McAinsh et al. (1995). The electrode tips contained either 10 mm fura-2 pentapotassium salt (Molecular Probes, Cambridge BioScience, Cambridge, UK) (kd = 145 nm) or 10 mm fura-2, 100 mm BAPTA (Sigma, Poole, Dorset, UK) (kd = 210 nm) and 10 mm CaCl2. The fura-2 was used to measure [Ca2+]i and to act as a fluorescent tracer for BAPTA. The rate of iontophoretic flow of BAPTA out of the micro-electrode is uncertain, and thus it was difficult to estimate the final concentration of BAPTA delivered into the cytosol (Jürgens et al., 1994). However, it can be assumed that the concentration of BAPTA in the cytosol exceeded that of the fura-2 (approximately 10−4 m; McAinsh et al., 1990). Typically, cells were iontophoretically loaded with fura-2 for 1 min or fura-2 and BAPTA for 2 min with negative current pulses of 2 nA. Inclusion of 10 mm CaCl2 in the fura-2 with BAPTA injection solution increased the post-injection viability of the cells.

Following micro-injection, the epidermal strip was maintained under conditions that promote stomatal opening (McAinsh et al., 1995) and irrigated with CO2-free 10 mm MES-KOH, 50 mm KCl pH 6.2 (MES-KCL) for approximately 1 h. Fluorescence and aperture measurements were made only on stomata that met all the criteria for estimating viability described by McAinsh et al. (1995).

Micro-injection of fluorescent dyes into guard cells of A. thaliana

Leaves of 4–5-week-old A. thaliana (Col) were mounted abaxial surface down on cover slips that had been coated with a thin layer of medical adhesive (Hollister Inc., Libertyville, IL, USA). The mesophyll was removed as described by Webb and Hetherington (1997). Surrounding the epidermis with a perspex ring made a perfusion chamber which was sealed with petroleum jelly. The epidermis was irrigated with CO2-free 10 mm MES pH 6.2, 10 mm KCl or CO2-free 10 mm MES pH 6.2, 50 mm KCl at 22°C. Stomata irrigated with 10 mm KCl were initially closed (0–2 µm), whilst stomata irrigated with 50 mm KCl were initially open (4–6 µm).

Single guard cells were impaled with filamented quartz micropipettes (outside diameter 1 mm; internal diameter 0.6 mm) pulled on a P2000 laser puller (Sutter Instruments Co., Novato, CA, USA). The tips of the injection pipette were filled with 50 mm KCl containing either 0.5 mm fura pentapotassium salt or 0.1 mm calcium green-1 dextran, 10 000 MW (Molecular Probes). The fluorescent indicators were loaded into the cytosol of guard cells using an oil-filled pressure injector by applying 0.1–0.4 MPa (Leckie et al., 1998).

To assess guard-cell viability following micro-injection, initially closed stomata were irrigated with CO2-free MES-KCl in the light at 22°C for 1 h to promote stomatal opening. Alternatively, guard cells that were injected when stomata were open were irrigated with the same buffer for 15 min. Only those cells that retained the fluorescent indicator in the cytosol and were open to the same degree as the uninjected cell of the stomatal pair were used for further experimentation.

Cells in which the resting [Ca2+]i, before challenge with an external stimulus, was greater than approximately 200 nm were excluded, as our investigations with C. communis have indicated that high resting [Ca2+]i is usually indicative of cell damage (unpublished observations).

Ratio photometry and digital ratio imaging

Digital ratio imaging and ratio photometry of fura-2 fluorescence and single wavelength photometry of Ca-green-1 10000 MW dextran were performed as described by McAinsh et al. (1995). Short breaks in the ratio photometric traces are due to closing the photomultiplier shutter to allow bright-field illumination of the stomatal pair for half-stomatal aperture measurements. BAPTA and the solutions used in this investigation were non-fluorescent and did not interfere with the measurement of fura-2 fluorescence.

The distribution of calcium indicator dyes, and therefore fluorescence intensity, was uneven in the cells because much of the dye-containing cytoplasm was associated with the perinuclear region (McAinsh et al., 1990). To obtain the strongest possible signal from the rest of the cell cytoplasm, high intensifier and video gains were used, which caused saturation of the signal in the perinuclear cytoplasm. Consequently, during ratio imaging changes in [Ca2+]i were not accurately measured in the perinuclear region of the cells. Therefore in the digital ratio images the perinuclear regions of the cells have been digitally masked and the [Ca2+]i has been shown at resting values.

Measurement of guard cell movements

Alterations in the half stomatal aperture of the fura-2 or BAPTA-loaded guard cell and the unloaded guard cell of a stomatal pair were measured at 800× magnification using an eyepiece micrometer at 5 min intervals.

Stomatal movements in the epidermis of leaves of A. thaliana (Col) transformed with the CDeT6–19-GUS chimeric fusion were measured as described by Webb and Hetherington (1997). The inhibition of stomatal opening was measured because the conditions and protocols for this assay were similar to those used for the histochemical GUS assay, allowing direct comparison of the ABA-turgor and ABA-nuclear signalling pathways in A. thaliana.

Histochemical GUS assay

Epidermal peels were removed, by hand from the abaxial surface of the four and five leaves (measured from the floral apex) of 4–5-week-old plants of A. thaliana (Col) transformed with the CDeT6–19GUS or 35SGUS constructs. The epidermal pieces were floated cuticle uppermost on 10 mm MES pH 6.2 at 20°C until required (<1 h). To investigate regulation of the activity of the CDeT6-19 promoter, the epidermal pieces were floated for 3 h on CO2-free MES pH 6.2 in the absence or presence of 10−5 m ABA at 23°C, and were illuminated at 150 µmol m−2 sec−1. Calcium agonists and antagonists were dissolved to the appropriate final concentration in the incubation buffer. CO2-free air was obtained by passing laboratory air through a column of soda lime (Carbosorb Brand, Merck, Poole, Dorset, UK). Air containing 700 µl l−1 CO2 was supplied from a pressurized gas cylinder (BOC Special Gases, London, UK). GUS activity was visualized as described by Taylor et al. (1995). All experiments were repeated at least three times on lines derived from at least two independent transformants. Due to the semi-quantitative nature of the histochemical GUS assay, several precautions were taken to ensure objective interpretation of the data: (i) all experiments were repeated independently by three experimenters; (ii) many of the experiments were performed without experimenter knowledge of the treatment; and (iii) the images of GUS staining were recorded on videotape and viewed by a second experimenter to confirm the first experimenter's observations. The lines chosen for this study were those identified by a previous investigation (Taylor et al., 1995) to have high GUS activity in the guard cells when challenged by ABA.

Fluorescent GUS Assay

The effect of 5 mm BAPTA on the activity of β-glucoronidase in crude protein extracts from 35SGUS transformants and commercial β-glucoronidase type IX-A (Sigma) was assayed in vitro by performing a fluorometric GUS assay (Gallagher, 1992). Each experiment was repeated three times.

Acknowledgements

The research described here was supported by grants from the European Commission and the UK Biotechnology and Biological Sciences Research Council. Alex Webb is grateful to the Royal Society for the award of University Research Fellowship. The authors wish to thank Professor E.A.C. MacRobbie for critical reading of the manuscript.

Ancillary