Recent work has indicated that nitric oxide (NO) and its synthesis are important elements of signal cascades in plant–pathogen defence, and are a prerequisite for drought and abscisic acid (ABA) responses in Arabidopsis thaliana and Vicia faba guard cells. NO regulates inward-rectifying K+ channels and Cl− channels of Vicia guard cells via intracellular Ca2+ release. However, its integration with related signals, including the actions of serine–threonine protein kinases, is less well defined. We report here that the elevation of cytosolic-free [Ca2+] ([Ca2+]i) mediated by NO in guard cells is reversibly inhibited by the broad-range protein kinase antagonists staurosporine and K252A, but not by the tyrosine kinase antagonist genistein. The effects of kinase antagonism translate directly to a loss of NO-sensitivity of the inward-rectifying K+ channels and background (Cl− channel) current, and to a parallel loss in sensitivity of the K+ channels to ABA. These results demonstrate that NO-dependent signals can be modulated through protein phosphorylation upstream of intracellular Ca2+ release, and they implicate a target for protein kinase control in ABA signalling that feeds into NO-dependent Ca2+ release.
Abscisic acid (ABA) mediates adaptive responses of plants to adverse environmental conditions, including drought and high salinity (Davies and Zhang, 1991; Willmer and Fricker, 1996). During water-stress conditions, ABA accumulates in leaf tissues, where it triggers the closure of stomata to reduce transpirational water loss. The cascade of events leading from the ABA stimulus to stomatal closure entails changes in cytosolic free [Ca2+] ([Ca2+]i) and pHi that lead to a concerted modulation of at least four distinct ion currents (Blatt, 2000). ABA affects the activities of inward- (IK,in) and outward-rectifying (IK,out) K+ channels and of Cl− channels (ICl) to bias the membrane for KCl efflux and a loss of cell turgor.
A parallel thread of investigations has highlighted other contributions from reactive oxygen species. Among these, recent studies identified mutations in NADPH oxidases that suppress ABA-evoked Ca2+ channel activity (Murata et al., 2001), and a primary requirement for nitric oxide (NO) in ABA stimulus transmission and downstream control of K+ and Cl− channels (Garcia-Mata and Lamattina, 2003; Garcia-Mata et al., 2003; Neill et al., 2002). NO effects on guard cells, gas exchange and transpirational water loss through the stomata of the leaf epidermis are well documented (Desikan et al., 2002; Garcia-Mata and Lamattina, 2001, 2002; Neill et al., 2002). Our recent work (Garcia-Mata et al., 2003) demonstrated that NO is required for ABA-mediated control of IK,in and ICl, and it acts by activating ryanodine-sensitive Ca2+-channels of intercellular Ca2+ stores to elevate [Ca2+]i in Vicia guard cells. NO effects on (Ca2+-insensitive) IK,out, by contrast, occur at higher NO levels and appear to be mediated separately, through direct S-nitrosylation of target proteins associated with the K+ channel (Sokolovski and Blatt, 2004).
Elements of these responses to ABA and NO also parallel the characteristics of guard cell behaviour associated with fungal elicitors (Blatt et al., 1999). Indeed, NO is thought to play a key role in plant–pathogen defence, notably in relation to Ca2+ signalling (Delledonne et al., 1998; Durner et al., 1998; Lamotte et al., 2004; Ligterink et al., 1997; Romeis et al., 1999; Zhang et al., 1998). Because [Ca2+]i elevation in these circumstances – and in ABA (Assmann, 2003; Blatt, 2000; Hetherington, 2001; MacRobbie, 1997; Schroeder et al., 2001) – is closely tied with protein (de-)phosphorylation, we were interested to assess its interplay with NO and ABA signalling, especially in their control of IK,in and ICl. We report here that the NO enhancement of [Ca2+]i increases, but not the basal [Ca2+]i rise evoked by voltage, is blocked by broad-range protein kinase antagonists, and that suppressing NO action on [Ca2+]i increases translates to a loss of response to NO and ABA in the Ca2+-dependent currents IK,in and ICl. These results point to closely related targets for NO and protein phosphorylation associated with intracellular Ca2+ release, and they demonstrate the potential for protein (de-)phosphorylation to modulate the NO signal in guard cells.
Protein kinase antagoism suppresses IK,inandIClcontrol by NO
The inward-rectifying K+ channels of Vicia guard cells are exquisitely sensitive to NO. Adding the NO donor S-nitroso-N-acetyl-penicillamine (SNAP), sufficient to yield NO at rates of 10 nm min−1, leads to a 80–90% inactivation of IK,in within 2 min of exposure; this action is reversible on washing SNAP from the bath or on addition of the NO scavenger cPTIO and the effects of NO are fully repeatable, even on the same guard cell (Garcia-Mata et al., 2003). By contrast, we found that the same NO exposures failed to influence IK,in when guard cells were pre-treated with the broad-range protein kinase antagonists staurosporine and K252A. Figure 1 shows current traces (inset) and steady-state current–voltage (I–V) curves from one Vicia guard cell recorded before and after 6-min exposures to 10 nm min−1 NO, alone and after 10-min pre-treatment with 1 μm staurosporine. Clamp voltage steps from −100 mV to voltages between −120 and −240 mV yielded current typical of IK,in, showing a time-dependent activation and rapid deactivation on returning to −100 mV. NO exposure reduced IK,in to <20% of the control at all clamp voltages in the first instance, but in the presence of staurosporine the current remained virtually unchanged. Similar results were obtained in 10 other experiments and with K252A, and we have pooled these data (see Figure 2). In every case, NO alone suppressed IK,in within 2–3 min of the treatments, while exposures after 5–10 min pre-treatments with staurosporine and K252A were without effect. NO treatment also led to a reversible, 2.2 ± 0.4-fold increase in background current identified with Cl− (anion) channels (ICl) at the plasma membrane (Grabov et al., 1997; Pei et al., 1997), indicating a significant rise in ICl in response to NO. However, in the presence of the kinase antagonists the background current remained unaffected by NO treatments (see Figure 2).
Whereas NO normally affects IK,in gating (Garcia-Mata et al., 2003), we found that this was not the case in the presence of staurosporine or K252A, as evident from two observations. First, no indication could be found for a change in current kinetics. Halftimes for current activation in NO were statistically equivalent when compared with current activation kinetics before NO treatments from the same guard cells. Activation halftimes at −200 mV before treatments were 314 ± 35 ms and during exposures to 10 nm min−1 NO were 335 ± 22 ms (n = 11) in the presence of either staurosporine or K252A. Secondly, fitting the steady-state I–V curves from these same experiments to a Boltzmann function showed no appreciable change in the voltage sensitivity for gating (see also Figure 1). Visually satisfactory and statistically best fittings were obtained with joint fittings (±NO) to the equation:
in which only the maximum conductance (gmax) varied between curves. Here R is the gas constant, T is the temperature (in degrees Kelvin) and F is the Faraday constant. The remaining parameters – the gating charge coefficient (δ), K+ equilibrium voltage (EK) and is the voltage giving half-maximal activation (V1/2) – were held in common between data sets. Joint fittings in which V1/2 also varied between curves gave statistically equivalent results and yielded values for V1/2 of −152 ± 18 mV before, and −148 ± 21 mV during NO treatments in the presence of staurosporine and K252A. Thus, the action of NO on the voltage dependence for activation was suppressed by the protein kinase antagonists.
Staurosporine and K252A suppress voltage-evoked increases in [Ca2+]i
Guard cell ion channels are subject to control by protein (de-)phosphorylation although, with few exceptions, protein phosphatase antagonists have proved most effective on these currents (MacRobbie, 1997). 1/2A-type protein phosphatase antagonists strongly promote the Cl− channels (Grabov et al., 1997) and the abi1 (2C-type) protein phosphatase mutant suppresses ICl activation by ABA (Pei et al., 1997). Both K+ channels, too, are affected by 1/2A-type protein phosphatase antagonists (Li et al., 1994; Thiel and Blatt, 1994), the abi1 protein phosphatase (Armstrong et al., 1995) as well as inhibitors of 2B-type (Ca2+-dependent) protein phosphatases (Horvath et al., 2002; Luan et al., 1993). IK,in and ICl are sensitive to other signalling elements as well, notably to cytosolic-free [Ca2+] ([Ca2+]i). In Vicia guard cells, IK,in activity declines steeply with [Ca2+]i above 200 nm, with an apparent cooperativity coefficient of 4 and Ki near 300 nm (Grabov and Blatt, 1999). Evoked [Ca2+]i increases, too, are sensitive to protein (de-)phosphorylation. Both the 1/2A-type protein phosphatase antagonists okadaic acid and Calyculin A are known to enhance Ca2+ channel activity at the guard cell plasma membrane and to augment [Ca2+]i elevation through effects on internal release and recovery (Köhler and Blatt, 2002).
Because a prominent feature of NO action is its effect in elevating [Ca2+]i (Garcia-Mata et al., 2003), we tested whether the effects of staurosporine and K252A might be mediated through changes in the ability of NO to affect [Ca2+]i. Experiments were carried out using three-barrelled microelectrodes to load guard cells with the Ca2+-sensitive dye Fura2, and the ratio of Fura2 fluorescence on excitation at 340 and 390 nm was used to quantify [Ca2+]i under voltage clamp. Figure 3 shows the results of [Ca2+]i measurements from one guard cell challenged with NO, first in the absence and then in the presence of 1 μm staurosporine following a 5-min pre-treatment with the kinase antagonist. Voltage steps to −200 mV in each case evoked [Ca2+]i increases to values near 400 nm in the absence of NO. In the presence of NO, the same voltage step evoked a [Ca2+]i rise over 750 nm (see also Garcia-Mata et al., 2003), but with antagonist pre-treatment no such enhancement of the [Ca2+]i transient was observed. Similar results were obtained in eight independent experiments (Figure 4), confirming that NO action in elevating [Ca2+]i was suppressed by the protein kinase antagonists.
Genistein fails to suppress NO action on evoked [Ca2+]i transients and IK,in
Previous studies have implicated roles for protein tyrosine (de-)phosphorylation in ABA signalling. MacRobbie (2002) reported an inhibition of K+(86Rb+) efflux from guard cell vacuoles following treatments with phenylarsine oxide, a protein tyrosine phosphatase antagonist. While these effects may be linked to [Ca2+]i release, our own work has highlighted other actions of phenylarsine oxide in direct, peptide dicysteine modification that affects K+ channels at the guard cell plasma membrane (Sokolovski and Blatt, 2004). As an alternative, we carried out tests with the antagonist genistein as a screen for protein tyrosine kinase activity in NO signalling. Again, IK,in and background (ICl) current measurements were carried out under voltage clamp with and without additions of 10 nm min−1 NO and 10 μm genistein; effects on evoked [Ca2+]i increases were recorded by fluorescence ratio imaging after loading guard cells with Fura2.
Figure 5 shows data from one guard cell challenged with NO first in the absence and then in the presence of genistein. Treatment with NO alone resulted in the characteristic, reversible inactivation of IK,in. Adding genistein alone had little effect on the current; but, unlike the experiments with staurosporine and K252A, further addition of NO yielded IK,in inactivation comparable to the control and an increase in background (ICl) current (not shown), even after 8 min continuous superfusion with the protein tyrosine kinase antagonist. Similar results were obtained in seven independent experiments and are summarized in Figure 6 along with parallel measurements of evoked [Ca2+]i increases. These results indicate that genistein had no effect on IK,in or the background (ICl) current, nor on the enhanced [Ca2+]i increases observed with NO, suggesting that protein tyrosine kinase activity is not closely linked to transmission of the NO signal.
Protein kinase antagonism selectively protects IK,infrom inactivation by ABA
Control of IK,in and background (ICl) current by ABA depends on signal transmission through the NO intermediate. Garcia-Mata et al. (2003) showed previously that NO scavenging, as well as cytosolic Ca2+ buffers, prevented ABA-mediated inactivation of IK,in and enhancement of the background current, but had no effect in preventing the enhancement of (Ca2+-independent) IK,out by ABA. Because the protein kinase antagonists staurosporine and K252A, like NO scavenging, suppressed NO action on [Ca2+]i and control of the Ca2+-sensitive currents, we anticipated that the antagonists should also affect ABA control of IK,in and the background current, but not of IK,out.
To test this idea, we carried out measurements under voltage clamp while challenging guard cells with 20 μm ABA either in the absence or presence of 1 μm staurosporine. The currents recorded from one guard cell are shown in Figure 7. Similar results were obtained in five other experiments and with K252A (data not shown). Whereas ABA treatments normally results in roughly a 50–80% reduction in IK,in and two- to three-fold increases in both the background (ICl) current and IK,out within 8–10 min (see Figure 4 of Leyman et al., 1999; Figure 2 of Garcia-Mata et al., 2003), we found that after 4- to 5-min pre-treatments with the antagonist adding ABA gave a similar rise in IK,out but no significant change in IK,in or background current. These results indicate that both ABA- and NO-dependent signalling can be modulated through protein phosphorylation, and they implicate a target for protein kinase control in ABA signalling that feeds into NO-dependent Ca2+ release. We return to these points below.
Our results above now identify protein phosphorylation, sensitive to staurosporine and K252A in vivo, that is essential for NO signal transmission to potentiate [Ca2+]i through intracellular Ca2+ release, and is a prerequisite for ABA-mediated control of IK,in and ICl. We also find that the protein kinase antagonists do not influence ABA action on current through the (Ca2+-insensitive) outward-rectifying K+ channels. Thus, we delimit a major site of action of protein kinase blockade to step(s) associated with NO-enhanced Ca2+ release from intracellular stores and, hence, to a subset of ABA-associated events leading to K+ and Cl− channel regulation.
A key line of evidence is the blockade by protein kinase antagonists of NO-enhanced [Ca2+]i increases (Figures 3 and 4). In its effect on guard cell [Ca2+]i, the primary target for NO appears to be intracellular Ca2+ release rather than Ca2+ entry across the plasma membrane. Garcia-Mata et al. (2003) showed that NO enhanced voltage-evoked [Ca2+]i increases without a significant effect on their voltage dependence, and direct measurements showed no change (or, if anything, a decrease) in the Ca2+ channel activity at the plasma membrane. These data find support also in the sensitivity of NO-enhanced [Ca2+]i elevation to GMP cyclase and cADPR-receptor antagonists (Garcia-Mata et al., 2003). We note, too, that neither staurosporine nor K252A affect Ca2+ channel activity significantly at the guard cell plasma membrane (Köhler and Blatt, 2002). Thus, the blockade of the enhanced, but not the basal [Ca2+]i response, by staurosporine and K252A suggests that the protein kinase antagonists target only a subset of elements that contribute to intracellular Ca2+ release when engaged by NO.
This interpretation is also consistent with the concomitant loss in sensitivity of the inward-rectifying K+ channel and background (Cl− channel) currents to NO (Figures 1 and 2) and to ABA (Figure 7), a loss of response which is not mirrored in IK,out. The [Ca2+]i sensitivities of IK,in and ICl– and [Ca2+]i insensitivity of IK,out– are well documented (Blatt, 2000; Grabov and Blatt, 1999; Hetherington, 2001; Schroeder et al., 2001) and these two channel currents are selectively targeted by NO in a [Ca2+]i-dependent manner (Garcia-Mata et al., 2003). By contrast, the basal activities of both currents have generally been found to be insensitive to any changes effected by broad-range protein kinase antagonists (Grabov et al., 1997; Schulzlessdorf et al., 1996; Thiel and Blatt, 1994), although rundown of guard cell Cl− channels is promoted by kinase antagonism under some conditions when isolated in protoplasts (Schmidt et al., 1995). Indeed, most of the evidence for protein (de-)phosphorylation control of the ion channels in vivo (above) is derived from the actions of protein phosphatase antagonists.
Protein kinase activities have been linked to evoked responses in guard cells, notably to ABA (Sheen, 1997). Activation of a Ca2+-dependent protein kinase is essential to stimulate ICl in ABA (Li and Assmann, 1996; Li et al., 2000; Mori and Muto, 1997) and stomatal closure is known to require at least one other protein kinase in Arabidopsis identified with the ost1 mutant (Mustilli et al., 2002). How the phosphorylation events in these instances are associated with ion transport and its control is more difficult to assess at present. As a case in point, we note that both staurosporine and H7, another broad-range protein kinase antagonist, were found to rescue ABA-evoked changes in IK,in and IK,out as well as stomatal closure in transgenic tobacco carrying the mutant (dominant-negative) abi1 protein phosphatase (Armstrong et al., 1995). It is of interest, too, that similar treatments suppress K+ channel responses to the Avr9 fungal pathogen elicitor without affecting the basal channel activities (Blatt et al., 1999). Protein kinase activation and [Ca2+]i elevation are closely interlinked within the signalling responses behind many plant–pathogen interactions and, for Avr9/Cf9 signalling, these may include rapid expression of novel protein kinase genes (Rowland et al., 2005).
How might protein kinase blockade affect NO signal transmission? Clearly, the pathway from NO to [Ca2+]i elevation leading through GMP cyclase and cADPR suggests a relatively small number of potential foci for interaction. One straightforward interpretation is that phosphorylation of one or more endomembrane Ca2+ channels is required as a ‘priming’ step, permitting enhanced Ca2+ release in the presence of NO. For example, the Ca2+-permeable SV (slow vacuolar) channel is strongly regulated by phosphorylation and its current is sensitive to H7 (Bethke and Jones, 1997) and to 2B-type protein phosphatase antagonists (Allen and Sanders, 1995). In mammalian tissues, a major pathway for NO action is via cGMP-dependent protein kinases (Moreno et al., 2001) and changes [Ca2+]i, for example through phosphorylation of inositol-1,4,5-trisphosphate receptors (Archer et al., 1994). These observations aside, we stress that our data do not rule out parallel actions of protein kinase antagonism independent of changes in [Ca2+]i such as might be mediated directly via the K+ and Cl− channels (for example, see Cherel et al., 2002; Li et al., 1998) or associated regulatory protein subunits.
In conclusion, we find that the protein kinase antagonists staurosporine and K252A protect the background (Cl− channel) current and inward-rectifying K+ channels of Vicia guard cells from the action of NO and ABA. Kinase antagonism also suppresses NO action in promoting evoked [Ca2+]i increases that are a key factor in regulating both channel currents. These results point to protein (de-) phosphorylation control of intracellular Ca2+ release as an important factor in modulating NO-dependent signalling in guard cells.
Plant material and electrophysiology
Epidermal peels were prepared from Vicia faba L. grown under a 16/8 h (l/d) and 21/14°C cycle (Sokolovski and Blatt, 2004). All operations were carried out on a Zeiss Axiovert microscope (Zeiss, Jena, Germany) with 40x and 63x LWD DIC optics. Epidermal peels were fixed in the experimental chamber with an optically clear, pressure-sensitive adhesive (50/50 medical adhesive; Dow Corning, Brussels, Belgium) and were bathed in 5 mm Ca2+-MES, pH 6.1 [MES titrated to its pKa with Ca(OH)2] with 10 mm KCl. Measurements were carried out in continuously flowing solution at 20 chamber volumes per minute.
Recordings were obtained with two-barrelled microelectrodes coated with paraffin wax to reduce electrode capacitance (Blatt and Armstrong, 1993). Current-passing and voltage-recording barrels were filled with 200 mm K+-acetate (pH 7.5) to minimize salt leakage and salt-loading artefacts associated with the Cl− anion without imposing a pH load. In some experiments the electrodes were filled with 200 mm K+-HEPES, pH 7.5, to suppress any changes in cytosolic pH. Connections to amplifier headstages were via 1 m KCl Ag|AgCl halfcells, and a matching halfcell and 1 m KCl-agar bridge served as the reference (bath) electrode.
Electrical and numerical analysis
Mechanical and electrical design have been described previously (Blatt and Armstrong, 1993). Voltage clamp control, data acquisition and analysis were carried out using Henry II software (Y-Science, Glasgow, UK; available for academic use by download at http://www.gla.ac.uk/ibls/BMB/mrb/lppbh.htm). Currents were normally filtered with a low-pass Butterworth filter (cut-off frequency, 1 kHz) and sampled at 2 kHz. IK,out was determined and activation halftimes (t1/2) were taken from two-step voltage clamp protocols after subtracting instantaneous currents at each voltage. Where appropriate, analyses were carried out by non-linear, least-squares fittings using a Marquardt–Levenberg algorithm (Marquardt, 1963). Results are reported as mean ± SE and taken to be significant at P < 0.05.
[Ca2+]i was determined by Fura2 fluorescence ratio imaging as described previously with a GenIV-intensified Pentamax-512 CCD camera (Princeton Instruments, Princeton, NJ, USA) (Köhler and Blatt, 2002). Measurements were corrected for background before loading and analysed with Universal Imaging software (West Chester, PA, USA). Fura2 fluorescence was calibrated in vitro and in vivo after permeabilization, and estimates of loading indicated final Fura2 concentrations <10 μm (Garcia-Mata et al., 2003).
Nitric oxide was generated in solution from SNAP, which spontaneously releases NO in a pseudo first-order reaction with a halftime in solution of approximately 5 h (Hou et al., 1999). NO generation was assayed by the Griess reaction (Zhang et al., 2003) in perfusion buffer and indicated that 10 μm SNAP releases 2.5–3.0 μm NO over 2 h, equivalent to approximately 2 nm NO min−1μm−1 SNAP in standing solution (data not shown). Because the K+ channel measurements were carried out in continuously flowing solution, this figure represents a lower estimate of the rate of NO generation.
Chemicals and solutions
S-nitroso-N-acetyl-penicillamine was dissolved in 1:1 ethanol:H2O, and diluted >1000-fold for use. Ethanol alone at this concentration had no effect (Grabov and Blatt, 1998; Hamilton et al., 2000). All other compounds were used directly. All reagents were from Sigma (Poole, UK) or Calbiochem (Darmstadt, Germany).
This work was supported by BBSRC grants P09640, C10234 and BB/C500595/1 to MRB and by the Fundacion Antorchas to C.G.-M.