INTRODUCTION

TOP INTRODUCTION H2O2 MIMICS ABA ACTIVATION... H2O2 DOES NOT MIMIC... LITERATURE CITED

Recent evidence has implicated the action of reactive oxygen species (ROS), notably hydrogen peroxide (H₂O₂), in abscisic acid (ABA) signaling of guard cells. ABA is known to evoke increases in cytosolic-free [Ca²⁺] ([Ca²⁺]_i), dependent on flux through Ca²⁺ channels in the plasma membrane and release from intracellular Ca²⁺ stores (Grabov and Blatt, 1998; Hamilton et al., 2000; Pei et al., 2000), which inactivates inward-rectifying K⁺ channels (I_K,in) and activates anion channels to bias the plasma membrane for solute efflux and stomatal closure (MacRobbie, 1997; Blatt, 2000; Schroeder et al., 2001). ABA also activates outward-rectifying K⁺ channels (I_K,out) through a parallel rise in cytosolic pH (see Blatt, 2000, and refs. therein). H₂O₂ was suggested as an intermediate early in ABA signal transduction because when added externally it, too, triggers stomatal closure and is known to activate Ca²⁺ channels and elevate [Ca²⁺]_i in many plant cells (Price et al., 1994; Pei et al., 2000; Murata et al., 2001; Schroeder et al., 2001; Zhang et al., 2001b). ROS production is augmented by exogenous ABA and its block by diphenylene iodonium and the abi1 mutant (dominant-negative) protein phosphatase suppresses stomatal closure in Arabidopsis (Pei et al., 2000; Murata et al., 2001; Zhang et al., 2001b).

These observations aside, little attention has focused on the K⁺ channels that ultimately mediate the solute flux to drive stomatal closure. We expected H₂O₂ to trigger the same pattern of response, activating the Ca²⁺ channels, inactivating I_K,in, and activating I_K,out, assuming that it transmitted the ABA signal. However, our results underscored both qualitative and quantitative differences between H₂O₂ and ABA actions, leading us to question the validity of arguments for H₂O₂ as a second messenger in this case.

	H2O2 MIMICS ABA ACTIVATION OF Ca2+ CHANNELS

TOP INTRODUCTION H2O2 MIMICS ABA ACTIVATION... H2O2 DOES NOT MIMIC... LITERATURE CITED

The hyperpolarization-activated Ca²⁺ channel in the plasma membrane of Vicia faba guard cells was activated by micromolar concentrations of H₂O₂ (Fig. 1) under the same conditions we used previously to characterize its response to ABA and protein phosphorylation (Hamilton et al., 2000; Köhler and Blatt, 2002). At a voltage of 150 mV, activation by H₂O₂ occurred both in the cell-attached configuration (Fig. 1, A and B) and with isolated inside-out patches (not shown). One hundred micromolar H₂O₂ enhanced channel activity (NP_o) more than 100-fold (Fig. 1C), with NP_o rising from 0.002 ± 0.001 to 0.14 ± 0.08 in cell attached, and from 0.0001 ± 0.00001 to 0.15 ± 0.06 in inside-out recordings. H₂O₂ is freely permeable across biological membranes (Heldt and Fluegge, 1992), so the similar effects on the Ca²⁺ channel in attached and isolated patches suggests the dominant site of action is on, or closely associated with the channel protein itself.

View larger version (31K): [in this window] [in a new window]   Figure 1.   H2O2 activates guard cell Ca2+ channels. A, Current recorded from a cell-attached patch with 30 mM Ba2+-HEPES (bath and pipette). Clamp voltage, 150 mV; arrow, 10 µM H2O2 addition. At least 10 channels are evident in the presence of H2O2 (open levels, right). Scale: vertical, 5 pA; horizontal, 30 s. B, Channel activity (NPo = apparent channel no. X open probability) calculated from overlapping 5-s segments for the data in A. C, NPo increases with H2O2 concentration. Results pooled from nine independent (cell-attached and inside-out) experiments. One millimolar ATP was added to the bath solution for inside-out recordings (Köhler and Blatt, 2002). Data normalized as NPo ratio [=NPo(+H2O2)/NPo(H2O2)] are fitted to a simple Michaelian function (solid curve [K1/2] = 76 ± 28 µM). D. H2O2 does not change the single-channel amplitude. Amplitude histogram derived from a representative experiment with an inside-out patch in 10 µM H2O2. Clamp voltage, 150 mV; single-channel amplitude, 1.9 ± 0.4 pA. Gaussian fittings to amplitudes before (lower curve, amplitudes not shown for clarity) and after H2O2 addition. For details of experimental materials and methods, see Köhler and Blatt (2002).

Qualitatively, H₂O₂ action on the Ca²⁺ channel was comparable with that of ABA and phosphatase antagonists (Hamilton et al., 2000; Köhler and Blatt, 2002), arising from increases in open probability and a recruitment of "cryptic" channels (compare with Köhler and Blatt, 2002). In the presence of 500 µM H₂O₂, the open probability (P_o) increased from 0.0004 ± 0.0002 to 0.03 ± 0.01 and the number of channels rose 6 ± 2-fold (n = 6). Analysis of open and closed lifetime distributions from isolated patches with a single channel in H₂O₂ indicated open (_o, 0.8 ± 0.03 and 3.5 ± 0.4 ms) and closed (_c, 1.1 ± 0.3, 6 ± 3, and 223 ± 39 ms) lifetimes comparable with those obtained previously in ABA (Hamilton et al., 2001; Köhler and Blatt, 2002). Like ABA, H₂O₂ had no measurable effect on single channel amplitude (Hamilton et al., 2000; see also Fig. 1D). These results and the channel characteristics indicate that H₂O₂ and ABA activate the same Ca²⁺ channel and in a similar manner.

Although differences between species cannot be ruled out, the characteristics of the Ca²⁺ channels in V. faba and Arabidopsis are similar (Hamilton et al., 2000, 2001; Pei et al., 2000), thus implying that the responses to H₂O₂ and ABA may be general phenomena. The Ca²⁺ channels of both species are strongly voltage dependent, activating negative of 100 mV, and both are permeable to Ba²⁺ as well as Ca²⁺. We found that H₂O₂ activated the V. faba Ca²⁺ channel with a similar concentration dependence (K_1/2 = 76 ± 28 µM H₂O₂) and with a delay (2 ± 0.5 min, n = 9) that was independent of the H₂O₂ concentration between 10 and 500 µM both in cell-attached and inside-out configurations (Fig. 1; compare with Pei et al., 2000).

	H2O2 DOES NOT MIMIC ABA ACTIVATION OF IK,out

TOP INTRODUCTION H2O2 MIMICS ABA ACTIVATION... H2O2 DOES NOT MIMIC... LITERATURE CITED

Zhang et al. (2001a) reported that I_K,in in V. faba guard cells is suppressed by exogenous H₂O₂. Significantly, they worked at a concentration of 10 µM H₂O₂well below the K_1/2 for the Ca²⁺ channelbut did not pursue the observation further. To quantify the effects of H₂O₂ on I_K,in and I_K,out, we carried out voltage-clamp experiments with intact guard cells as described previously (Blatt and Armstrong, 1993; Grabov and Blatt, 1998, 1999). We found that (Fig. 2), like ABA, H₂O₂ treatments suppressed I_K,in, shifting its activation to more negative voltages. However, unlike ABA, H₂O₂ also depressed I_K,out and the effect on both K⁺ channels was irreversible. The response of I_K,out and I_K,in to H₂O₂ occurred with halftimes of 6 ± 2 and 4 ± 0.5 min, respectively, for concentrations from 1 to 50 µM (Fig. 3, A and B). Furthermore, we observed quantitatively equivalent results, even when exposures were restricted to 30 to 60 s and H₂O₂ was then washed from the bath (Fig. 3, A and B). One micromolar H₂O₂ was sufficient for near-maximal effect on both K⁺ channels (Fig. 3C). Finally, H₂O₂ did not have a significant effect on the halftimes for activation at any concentration tested (I_K,out, Fig. 3A, inset; I_K,in, not shown), suggesting an effect mediated by a change in the number of functional channels rather than by alterations in their gating kinetics.

View larger version (18K): [in this window] [in a new window]   Figure 2.   H2O2 suppresses both IK,in and IK,out. Data from an intact V. faba guard cell under a two-electrode voltage clamp bathed in 10 mM KCl and 5 mM Ca2+-MES, pH 6.1. A, Current response 2 min before and 11 min after 1-min exposure to 10 µM H2O2. Three-second clamp voltage steps (12) to voltages between +30 and 250 mV from a holding voltage of 100 mV. Scale: horizontal, 1 s; vertical, 100 µA cm2. B, Current-voltage curves derived from A and additional data of the same cell before () and 1 (), 3 (), 6 (), and 11 () min after adding H2O2. K+ channel currents obtained by subtracting instantaneous from steady-state current at each voltage. Data for IK,in and IK,out fitted jointly to common Boltzmann functions (solid curves). For details, see Grabov and Blatt (1999) and Blatt and Armstrong (1993).

View larger version (16K): [in this window] [in a new window]   Figure 3.   V. faba guard cell K+ channels are roughly 100-fold more sensitive to H2O2 than the Ca2+ channel. A, Time for block of IK,out independent of H2O2 concentrations above 1 µM. Current at 0 mV determined as in Figure 2 and plotted as time after adding 1 (n = 3, ), 10 (n = 4, ), and 50 (n = 3, ) µM H2O2 for 2 min. Solid curve, Fitting to single exponential decay (t1/2, 6 ± 2 min). Inset, Halftimes for IK,out activation in the presence of 1 (n = 3, ), 10 (n = 4, ), and 50 (n = 3, ) µM H2O2. Data fitted empirically to a single exponential decay function. B, Time for block of IK,in recorded at 200 mV, as in A. Solid curve, Fitting to single exponential decay (t1/2, 4 ± 0.5 min). C, Block by H2O2 of IK,in () at 200 mV and IK,out () at 0 mV fitted to Michaelian functions. K1/2:IK,in, 0.1 ± 0.4 µM; IK,out, 0.3 ± 0.2 µM.

It is not surprising that the guard cell K⁺ channels are sensitive to ROS because, like many proteins, they can be expected to harbor reactive groups (e.g. sulfhydryl bonds; for KAT1 of Arabidopsis, see Anderson et al., 1992). Previous data have shown effects of O₃ on V. faba guard cell K⁺ channels (Torsethaugen et al., 1999) and ROS action in vivo (Wang et al., 1997) and after heterologous expression (Duprat et al., 1995) is known for other K⁺ channels. In fact, a direct action of H₂O₂ to render the K⁺ channels nonfunctional seems the simplest explanation in this case because the effects were complete without change in activation kinetics and at concentrations roughly 100-fold lower than were effective in activating the Ca²⁺ channel. Although at present we cannot rule out a rise in [Ca²⁺]_i at these very low concentrations, H₂O₂ action solely through [Ca²⁺]_i is inconsistent with the response of I_K,out, which is known to be Ca²⁺ insensitive (Hosoi et al., 1988; Blatt and Armstrong, 1993; Lemtiri-Chlieh and MacRobbie, 1994; Grabov and Blatt, 1999). At first sight, it is surprising that H₂O₂ should suppress I_K,out because H₂O₂ induces stomatal closure in epidermal strips of Arabidopsis and V. faba (Pei et al., 2000; Zhang et al., 2001b) and, therefore, might be expected to stimulate K⁺ loss from the cells. However, other pathways for K⁺ efflux have been reported (Thiel et al., 1992; Pei et al., 1998) and their response to H₂O₂ is unknown (see also Duprat et al., 1995).

Most important, the finding that H₂O₂ inhibited the K⁺ outward rectifier in guard cells shows that H₂O₂ does not mimic ABA action on guard cell ion channels as it acts on the K⁺ outward rectifier in a manner entirely contrary to that of ABA. This observation brings into question previous evidence based on exogenous applications for a role of H₂O₂ as a second messenger in ABA signaling of stomata. We pose this question now also in a general sense. Although it may be argued that H₂O₂ could act differently inside and outside the guard cell, the ROS is freely permeable across biological membranes (Heldt and Fluegge, 1992). Therefore, any such differences in action would require localized and exceedingly tight coupling between the sites for H₂O₂ generation and action at the inner surface of the plasma membrane. Other differences in action between ABA and H₂O₂ are known. For example, Allen et al. (2000) reported that H₂O₂ triggered Ca²⁺ oscillations with a "Ca²⁺ fingerprint" but not an "ABA fingerprint" in the Arabidopsis det3 mutant, suggesting that the signaling cascades are different, although they might share components. No doubt, ABA and oxidative stress responses are linked (Guan et al., 2000; Pei et al., 2000; Zhang et al., 2001b), but through a network of signaling pathways that have evolved to deal with the common situation of combined stress inputs (Knight and Knight, 2001).

In conclusion, we question the role of H₂O₂ as a critical second messenger regulating guard cell ion channels in response to ABA. The Ca²⁺ channel is a target for both ABA and H₂O₂ signal processing, but in our view it serves as a focal point integrating signal transduction pathways and, thus, links these several pathways, among others, to membrane voltage (Gradmann et al., 1993; Grabov and Blatt, 1998, 1999), NAD(P) H and the cellular redox state (Murata et al., 2001), and protein phosphorylation (Köhler and Blatt, 2002). Our data suggest that the ABA and H₂O₂ pathways diverge further downstream in their actions on the K⁺ channels and, thus, on stomatal control.

Received October 10, 2002; returned for revision October 22, 2002; accepted October 22, 2002.