Ethylene-induced stomatal closure in Arabidopsis occurs via AtrbohF-mediated hydrogen peroxide synthesis

Summary

Ethylene is a plant hormone that regulates many aspects of growth and development. Despite the well-known association between ethylene and stress signalling, its effects on stomatal movements are largely unexplored. Here, genetic and physiological data are provided that position ethylene into the Arabidopsis guard cell signalling network, and demonstrate a functional link between ethylene and hydrogen peroxide (H₂O₂). In wild-type leaves, ethylene induces stomatal closure that is dependent on H₂O₂ production in guard cells, generated by the nicotinamide adenine dinucleotide phosphate hydrogen (NADPH) oxidase AtrbohF. Ethylene-induced closure is inhibited by the ethylene antagonists 1-MCP and silver. The ethylene receptor mutants etr1–1 and etr1–3 are insensitive to ethylene in terms of stomatal closure and H₂O₂ production. Stomata of the ethylene signalling ein2–1 and arr2 mutants do not close in response to either ethylene or H₂O₂ but do generate H₂O₂ following ethylene challenge. Thus, the data indicate that ethylene and H₂O₂ signalling in guard cells are mediated by ETR1 via EIN2 and ARR2-dependent pathway(s), and identify AtrbohF as a key mediator of stomatal responses to ethylene.

Introduction

The plant hormone ethylene influences many aspects of plant growth and development, including fruit ripening, senescence, seed germination, seedling growth and abscission, as well as mediating stress and pathogen defence responses (Guo and Ecker, 2004; Schaller and Kieber, 2002). A characteristic response to ethylene is the triple response of etiolated seedlings, involving inhibition of hypocotyl and root cell elongation, radial swelling of the hypocotyl, and exaggerated curvature of the apical hook. The triple response has provided the basis for many genetic studies that have identified various ethylene-signalling mutants impaired in ethylene perception and subsequent downstream signalling processes (Guo and Ecker, 2004).

In Arabidopsis, ethylene perception is mediated via a group of five histidine kinase-like receptors that are members of the two-component signalling protein family (Grefen and Harter, 2004; Hwang et al., 2002). Genetic and biochemical analyses indicate that ethylene signalling includes both positive and negative regulators (Alonso and Stepanova, 2004). In the absence of ethylene, a downstream protein kinase, CTR1, actively represses signalling components further down the pathway. Binding of ethylene to its receptors results in inactivation of CTR1, thereby relieving this repression of the ethylene response pathway. The positions of various gene products participating in ethylene signalling have been determined mostly by phenotypic analysis of the ethylene-induced triple response of mutant seedlings.

The involvement of ethylene in regulating stomatal aperture is not clear. In some species, ethylene mediates auxin-induced stomatal opening (Levitt et al., 1987; Merritt et al., 2001), whereas in others it induces closure (Madhavan et al., 1983; Pallas and Kays, 1982). A recent paper has reported that ethylene can modulate abscisic acid (ABA)-induced stomatal closure in Arabidopsis thaliana (Tanaka et al., 2005), but there are no reports of the effects of ethylene alone on Arabidopsis stomata. Moreover, nothing is known about the potential signalling pathways that might be regulated by ethylene in guard cells.

Guard cell signalling is highly complex and modulated by many stimuli, most notably ABA (Fan et al., 2004; Hetherington and Woodward, 2003). Among the many molecules participating in ABA-activated signalling pathways in guard cells, recent additions include nitric oxide (NO) and hydrogen peroxide (H₂O₂) (see Desikan et al., 2004a). H₂O₂ is a signalling molecule of widespread importance in plant responses to various biotic and abiotic stimuli that include pathogen challenge, drought stress, exposure to atmospheric pollutants, extremes of temperatures, gravitropism, hormones, cell development and senescence (Apel and Hirt, 2004; Laloi et al., 2004; Neill et al., 2002).

Recent data indicate that H₂O₂-induced stomatal closure requires the ethylene receptor ETR1, implying a link between H₂O₂ and ethylene signal transduction. Stomata of the ethylene-insensitive etr1–1 mutant of Arabidopsis, in which the ETR1 protein contains a Cys65Tyr mutation, are insensitive to H₂O₂, whereas those of etr1–3, with an Ala31Val mutation, respond to H₂O₂ (Desikan et al., 2005). These data suggest that ETR1, and the Cys65 residue of ETR1 in particular, are important for H₂O₂ signalling in guard cells. Furthermore, H₂O₂ signalling via ETR1 is mechanistically different to that of the triple response to ethylene, because the latter requires the presence of the histidine kinase domain in ETR1 (Gamble et al., 2002; Qu and Schaller, 2004) but the former does not (Desikan et al., 2005).

Such functional interaction between ethylene and H₂O₂ is not completely unexpected. For example, H₂O₂ and ethylene interact in the induction of ethylene-responsive genes (Desikan et al., 2001; Vandenabeele et al., 2003), and ethylene synthesis is required for H₂O₂ generation during ozone-induced cell death (Moeder et al., 2002).

In this paper, the induction by ethylene of stomatal closure in Arabidopsis is reported. The failure of ethylene to induce stomatal closure in plants mutated in the ethylene receptor ETR1 (in which ethylene binding by ETR1 is blocked) demonstrates that ETR1 mediates ethylene-induced stomatal closure. Moreover, such closure requires the synthesis and action of H₂O₂ that is generated via AtrbohF, an NADPH oxidase homologue present in guard cells (Kwak et al., 2003). Data are also presented to show that both ethylene and H₂O₂-induced stomatal closure are mediated by EIN2- and ARR2- dependent pathway(s). These observations indicate that ethylene and H₂O₂ cross-talk occurs during stomatal closure in Arabidopsis, and provide a new paradigm for ethylene signalling.

Results

Ethylene induces stomatal closure in Arabidopsis

To determine the effect of ethylene on Arabidopsis stomata, wild-type leaves were incubated in the ethylene-releasing compound ethephon or in ACC (1-aminocyclopropane-1-carboxylic acid), the immediate precursor of ethylene. Stomatal apertures were then measured in epidermal leaf fragments prepared from these leaves. Both ethephon and ACC caused stomatal closure in a dose-dependent manner (Figure 1a), and ethephon-induced closure was initiated within 30 min of exposure (Figure 1a, inset). Furthermore, application of ethephon to either soil- or hydroponically grown plants reduced transpiration, as determined gravimetrically (data not shown). To confirm that it was ethylene per se that was responsible for stomatal closure, leaves were incubated in a sealed environment to which were added various amounts of ethylene gas. Again, ethylene induced stomatal closure in a dose-dependent manner (Figure 1c). Thus, Arabidopsis stomata closed in response to ethylene irrespective of its source.

Figure 1.  Ethylene induces stomatal closure in Arabidopsis leaves. (a) Wild-type Arabidopsis (Col-0) leaves were floated on solutions containing various doses of ethephon, and stomatal apertures measured after 2.5 h. Inset: time course for stomatal closure induced by ethephon (100 μm). Solid lines, control; dotted lines, ethephon. Data are from 3–5 independent experiments (mean ± SE, n = 60–100). *P < 0.01 (one-way anova) versus control. (b) Arabidopsis leaves were floated on solutions containing various doses of the ethylene precursor 1-aminocyclopropane-1-carboxylic acid, and stomatal apertures measured after 2.5 h. Data are from three independent experiments (mean ± SE, n = 60). *P < 0.001 (one-way anova) versus control. (c) Arabidopsis leaves floating on buffer were exposed to ethylene gas in a sealed environment, and stomatal apertures measured after 2.5 h. Data are from three independent experiments (mean ± SE, n = 60). *P < 0.001 (one-way anova) versus control. (d) Effect of 1-MCP on ethylene-induced closure. Arabidopsis leaves were first floated on MES/KCl buffer for 2.5 h to open stomata. The leaves were then exposed in Petri dishes with buffer, either to air or 1-MCP gas (300 ppb) in a sealed environment for 4 h. Incubation was then continued with 1-MCP alone (M), 10 μl l⁻¹ ethylene (E) alone, or 1-MCP in the presence of ethylene (M + E) for a further 2.5 h, after which stomatal apertures were measured. Data are from three independent experiments (mean ± SE, n = 50–200). *P < 0.001 (one-way anova) versus ethylene alone. (e) Effects of silver and the copper chelator diethyldithiocarbamate (DDC) on ethylene- and H₂O₂-induced stomatal closure. Wild-type Col-0 leaves were exposed to either buffer (con), silver nitrate (Ag), DDC or 100 μm ethephon alone (E), or ethephon in the presence of 100 μm silver nitrate (E + Ag) or 100 μm DDC (E + DDC), and apertures measured after 2.5 h. Alternatively, leaves were exposed to 100 μm H₂O₂ (H) alone or H₂O₂ in the presence of 100 μm silver nitrate (H + Ag) or 100 μm DDC (H + DDC), and apertures measured. As a control treatment, 100 μm copper acetate (Cu) was included. Data are from 3–6 independent experiments (mean ± SE, n = 60–120). *P < 0.001 (one-way anova) versus ethephon alone.

Ethylene-induced stomatal closure is inhibited by 1-MCP and silver, and requires copper

1-methylcyclopropene (1-MCP) is a competitive inhibitor of the ethylene receptor and has been used widely to block various effects of ethylene (Sisler and Serek, 1997). Pre-treatment of leaves with 1-MCP for 4-h blocked ethylene-induced stomatal closure (Figure 1d), indicating that ethylene effects on stomata occur through an ethylene receptor. Silver is a well-known inhibitor of ethylene effects such as the triple response and flower senescence (Beyer, 1976), with a mechanism of action likely to be via displacement of copper from an ethylene receptor (Rodriguez et al., 1999). Pre-treatment of leaves with silver nitrate inhibited ethephon-induced stomatal closure substantially, but had no significant effect on closure induced by a different stimulus, in this case H₂O₂ (Figure 1e). Ethylene binding to its receptor is mediated by a copper co-factor, co-ordinated by the Cys65 and His69 residues in ETR1, and a mutation of this receptor that prevents copper binding, such as in etr1–1, also prevents ethylene binding (Rodriguez et al., 1999). A requirement for copper during ethylene-induced stomatal closure was thus investigated. Incubation of leaves with copper acetate or the copper chelator diethyldithiocarbamate (DDC) alone had no significant effect on the stomatal response (Figure 1e). However, pre-treatment of leaves with DDC inhibited ethephon-induced stomatal closure substantially (Figure 1e). In contrast, H₂O₂-induced closure was not significantly affected by DDC (Figure 1e).

Ethylene induces H₂O₂ generation in guard cells via AtrbohF

Hydrogen peroxide (H₂O₂) is an important signalling molecule in guard cells (see Desikan et al., 2004a), and a role for H₂O₂ has previously been established in ABA-induced stomatal closure (Murata et al., 2001; Mustilli et al., 2002; Pei et al., 2000; Zhang et al., 2001). Furthermore, previous work has shown that H₂O₂-induced stomatal closure requires ETR1 (Desikan et al., 2005). Thus, ethylene-induced stomatal closure was monitored in the presence of the antioxidant N-acetyl cysteine (NAC), or diphenylene iodonium (DPI), an inhibitor of H₂O₂-generating NADPH oxidases. Pre-treatment of leaves with either NAC or DPI inhibited ethephon-induced stomatal closure (Figure 2a), suggesting that H₂O₂ synthesis is required for ethylene-induced stomatal closure. DPI or NAC alone had no significant effect on stomatal apertures compared to buffer controls (P > 0.05; data not shown). Ethylene-induced H₂O₂ synthesis in guard cells was monitored using confocal microscopy and the H₂O₂-sensitive fluorescent dye DCFH₂-DA (Desikan et al., 2004b; Pei et al., 2000; Zhang et al., 2001). Treatment with ethylene, supplied either directly or via ethephon, increased H₂O₂-mediated fluorescence within 30 min, a response reduced by NAC or DPI (Figure 2a,b). The correlation between ethylene-induced stomatal closure and accumulation of H₂O₂ indicates that H₂O₂ production is an essential signal in the ethylene response, and the inhibition by DPI implicates an NADPH oxidase as a source of this H₂O₂. Importantly, AVG (l-α-(2-aminoethoxyvinyl)glycine hydrochloride), an inhibitor of ethylene biosynthesis, did not inhibit H₂O₂-induced closure (apertures for control = 2.3 μm ± 0.1; H₂O₂ = 1.5 μm ± 0.06; H₂O₂ +AVG = 1.4 μm ± 0.09; mean ± SE, n = 60–100 guard cells), indicating that ethylene biosynthesis is not required for H₂O₂-induced closure.

Figure 2.  Ethylene induces H₂O₂ generation in guard cells via AtrbohF. (a) Ethylene-induced H₂O₂ synthesis and stomatal closure. Epidermal fragments from wild-type leaves were exposed for 30 min to either buffer alone (control), 100 μl l⁻¹ ethylene gas (ethylene), 100 μm ethephon (E), or ethephon pre-treated with 1 mm NAC (E + NAC) or 10 μm DPI (E + DPI) for 10 min. H₂O₂ was visualized using confocal microscopy. Data for H₂O₂ production (black bars) are represented as H₂O₂ fluorescence (%) relative to average levels of fluorescence from control treatments (mean ± SE, n = 100–350 guard cells). Stomatal apertures (white bars) were measured following the same treatments of leaves after 2.5 h. Data are from at least three independent experiments (mean ± SE, n = 60). *P < 0.001 (one-way anova) versus ethephon alone. (b) Confocal images of H₂O₂ fluorescence. Scale bar = 15 μm. (c) Epidermal fragments isolated from either wild-type Col-0 (wt), atrbohD, atrbohF or atrbohD/F leaves were treated with either buffer alone (black bars) or 100 μm ethephon (white bars) for 30 min and H₂O₂ fluorescence detected via confocal microscopy. Data are represented as percentage response relative to H₂O₂ fluorescence in control treatments. Data are from at least three independent experiments (mean ± SE, n = 100–125 guard cells). *P < 0.001 (one-way anova) versus appropriate controls comprising buffer alone. (d) Leaves of either wild-type (wt), atrbohD, atrbohF or atrbohD/F plants were incubated in buffer alone (black bars) or 100 μm ethephon (white bars) for 2.5 h, and stomatal apertures measured. Data are from at least five independent experiments (mean ± SE, n = 100 guard cells). *P < 0.001 (one-way anova) versus appropriate controls comprising buffer alone.

Recent work has shown that respiratory burst oxidase homologue (Rboh) NADPH oxidases generate H₂O₂ during stomatal closure in response to ABA in Arabidopsis, and that AtrbohD and AtrbohF are highly expressed in guard cells (Kwak et al., 2003). Consequently, single and double atrbohD and atrbohF mutant Arabidopsis lines were investigated for their guard cell responses to ethylene. Ethephon induced an increase in H₂O₂ fluorescence in guard cells of the atrbohD mutant (Figure 2c). However, guard cells of neither the atrbohF single mutant nor the atrbohD/F double mutant responded to ethephon with increased H₂O₂ production (Figure 2c), indicating that a functional AtrbohF is required for ethylene-induced H₂O₂ synthesis in guard cells. To provide the functional link between ethylene-induced H₂O₂ synthesis via Atrboh and stomatal closure, stomatal apertures of the atrboh mutants were measured following exposure to ethephon. atrbohD stomata responded by closing, whereas stomata of both atrbohF and atrbohD/F mutants did not (Figure 2d). Although the initial stomatal apertures for both atrbohF and atrbohD/F are smaller than for wild-type, atrbohF and atrbohD/F stomata still close in response to ABA (partly) or H₂O₂ or NO (fully), indicating that further closure from control values is still possible (Bright et al., 2006, and unpubl. data). In summary, these data suggest that a functional AtrbohF protein is required for the induction by ethylene of both H₂O₂ synthesis and stomatal closure.

Ethylene-induced stomatal closure and H₂O₂ synthesis are mediated by the ethylene receptor ETR1

Extensive genetic analyses in Arabidopsis have shown that there are five ethylene receptors, with functional redundancy between them (Chen et al., 2005). To begin to determine the input of these receptors into guard cell ethylene responses, the role of the best-characterized receptor ETR1 was analysed. Stomata of the dominant gain-of-function ethylene-insensitive etr1–1 mutant showed little response to ethephon or ethylene gas (Figure 3a). Similarly, etr1–1 guard cells failed to generate H₂O₂ in response to ethephon or ethylene gas (Figure 3b). To demonstrate that there was not a total impairment of stomatal function in etr1–1, etr1–1 leaves were treated with ABA, which induced both stomatal closure and H₂O₂ synthesis (Figure 3a,b) indicating that the guard cells of the etr1–1 mutant are still functional. As ABA induces H₂O₂ synthesis in etr1–1 guard cells, but etr1–1 stomata are insensitive to H₂O₂ (Desikan et al., 2005), presumably other signals activated by ABA mediate the stomatal closure response. Stomatal responses were also assayed in the etr1–3 mutant that is similarly ethylene-insensitive in the triple response (Hall et al., 1999). As with etr1–1, etr1–3 stomata were insensitive to ethephon and ethylene with respect to stomatal closure; however, they did close in response to ABA (Figure 3a). Ethephon-induced H₂O₂ production was also impaired in etr1–3 guard cells (Figure 3b). Together, these data are in agreement with the inhibition of ethylene-induced stomatal closure by 1-MCP and silver (Figure 1), and demonstrate that ethylene signalling in guard cells operates through ETR1.

Figure 3.  Ethylene-induced stomatal closure and H₂O₂ synthesis are mediated by the ethylene receptor ETR1. (a) Stomatal responses. Arabidopsis leaves from etr1–1 or etr1–3 plants were exposed to buffer alone (black bars), 100 μm ethephon (grey bars), 100 μl l⁻¹ ethylene gas (white bars) or 10 μm ABA (striped bars), and stomatal apertures measured after 2.5 h. Data are from at least three independent experiments (mean ± SE, n = 60). *P < 0.01 (one-way anova) versus appropriate control comprising buffer alone. (b) H₂O₂ generation in etr1–1 and etr1–3 guard cells. H₂O₂ fluorescence was determined after treatment of etr1–1 or etr1–3 guard cells with either buffer (control, black bars), ethephon (100 μm, grey bars), ethylene (100 μl l⁻¹, white bars) or ABA (10 μm, striped bar) for 30 min. Data are represented as a percentage of the response in control treatments. Data are from at least three independent experiments (mean ± SE, n = 100–150 guard cells). *P < 0.001 (one-way anova) versus appropriate control comprising buffer alone.

Other ethylene pathway-related proteins are active in the guard cell ethylene-signalling network

To outline a potential ethylene-signalling network in guard cells, guard cell responses to ethylene were determined for other ethylene perception and signalling mutants. The RAN1 protein is thought to mediate loading of copper into ethylene receptors, and mutations in the RAN1 gene affect ethylene responses (Hirayama et al., 1999). Stomata of the ran1–1 mutant did not respond to ethephon, although they did close in response to H₂O₂ and ABA at the indicated concentrations (Figure 4a). In keeping with these responses, ran1–1 guard cells did not generate H₂O₂ in response to ethephon (Figure 4b).

Figure 4.  Ethylene-signalling components are active in the guard cell ethylene-signalling network. (a) Arabidopsis leaves from Col-0, ran1–1, ein2–1, Ler or arr2 plants were exposed to 100 μm ethephon (white bars), 100 μm H₂O₂ (grey bars) or 10 μm ABA (striped bars), and stomatal apertures measured after 2.5 h. Data are from at least three independent experiments (mean ± SE, n = 60). *P < 0.001 (one-way anova) versus appropriate controls (black bars). (b) Response of Col-0, ran1–1, ein2–1, Ler or arr2 guard cells to ethephon. H₂O₂ fluorescence was determined after treatment of Col-0, ran1–1, ein2–1, Ler or arr2 guard cells with either buffer (control, black bars) or ethephon (100 μm, white bars) for 30 min. Data are represented as a percentage of the response in control treatments. Data are from at least three independent experiments (mean ± SE, n = 88–197 guard cells). *P < 0.001 (one-way anova) versus appropriate controls (black bars).

EIN2 is an natural resistance-associated macrophage (N-RAMP)-like protein that is active in the ethylene response pathway. Although its actual function remains unknown (Chen et al., 2005), it is crucial for ethylene signalling, as mutations in EIN2 result in an ethylene-insensitive, highly pleiotropic phenotype (Alonso et al., 1999). Consequently, ethylene- and H₂O₂-induced stomatal closure were examined in the ein2–1 mutant. Neither ethephon nor H₂O₂ caused stomatal closure in ein2–1 at the indicated concentrations (Figure 4a), indicating that EIN2 function is required for both ethylene and H₂O₂ signalling in guard cells. However, ein2–1 guard cells did generate H₂O₂ in response to ethephon (Figure 4b), indicating that the failure of ein2–1 stomata to close in response to ethephon is linked to their insensitivity to H₂O₂. On the other hand, ABA did induce stomatal closure, indicating that ein2–1 stomata still retain some functionality (Figure 4a).

The two-component response regulator protein ARR2 has recently been implicated as a component of ethylene signalling (Hass et al., 2004). In order to explore the potential involvement of ARR2 in guard cell ethylene signalling, the arr2 loss-of-function mutant was utilized. Compared to wild-type, arr2 stomata did not close in response to either ethephon or H₂O₂ (Figure 4a). However, although the control apertures for arr2 are smaller than those of wild-type, arr2 stomata do close in response to ABA (Figure 4a), indicating that, under these conditions, both ethylene and H₂O₂ require a functional ARR2 protein for the induction of stomatal closure. However, ARR2 is not required for H₂O₂ generation, as arr2 guard cells still made H₂O₂ in response to ethephon (Figure 4b).

Stomatal responses were also determined in the ctr1–1 mutant (Kieber et al., 1993). CTR1 is a Raf-like protein kinase that potentially acts at the head of a putative MAPK signalling module in the ethylene pathway (Guo and Ecker, 2004). CTR1 is a negative regulator of ethylene signalling, and the ctr1–1 mutant displays constitutive ethylene responses (Kieber et al., 1993). ctr1–1 stomata were not open more widely than those of wild-type, and responded normally to ethylene, H₂O₂ or ABA (data not shown).

Discussion

The data described here demonstrate convincingly that ethylene can induce stomatal closure in Arabidopsis and is thus another player in the expanding orchestra of signals that can elicit stomatal movements. Stomatal closure is induced in wild-type leaves by ethylene, irrespective of whether the ethylene is supplied directly as a gas, generated from ethephon, or in planta via biosynthesis from ACC, and this response has clear parallels with other ethylene effects. Ethylene-induced stomatal closure is inhibited by both 1-MCP and silver, demonstrating that ethylene interaction with at least one of its receptors is essential. The dominant gain-of-function mutants of ETR1, etr1–1 and etr1–3, are ethylene-insensitive in terms of stomatal closure, indicating that the ETR1 ethylene receptor is critical, as for other ethylene responses such as the triple response. The effects of the copper chelator DDC, together with the ethylene insensitivity of the ran1–1 copper transport mutant, indicate that copper is required for ethylene-induced stomatal closure, potentially as a component of the ETR1 receptor. No phenotype was reported for the ran1–1 allele in the absence of an inhibitor of ethylene biosynthesis (Hirayama et al., 1999), so the failure of ethylene-induced stomatal closure in the ran1–1 mutant may indicate differences in ethylene perception and/or signal transduction between guard cells and etiolated seedlings.

Despite varied reports in the literature on the effects of ethylene on stomatal movements in different species, the data presented here show clearly that ethylene does cause stomatal closure in intact Arabidopsis leaves. Tanaka et al. (2005) reported recently that ethylene inhibited ABA-induced stomatal closure, in seeming contrast to the data reported here. However, Tanaka et al. (2005) did not report the effects of ethylene alone in their system. Moreover, they used epidermal peels, whereas the studies described here used intact leaves. We have confirmed and extended the studies of Tanaka et al. (2005) in our laboratory. Importantly, although ethylene gas induces stomatal closure in leaves, it does not do so in epidermal peels. However, ABA does, and ethylene does inhibit ABA-induced closure in peels (apertures for control, ABA, ethylene and ethylene +ABA are 1.5 ± 0.05 μm, 1 ± 0.06 μm, 1.4 ± 0.05 μm and 1.5 ± 0.05 μm, respectively) and leaves (apertures for control, ABA, ethylene and ethylene +ABA are 2.1 ± 0.05 μm, 1.1 ± 0.06 μm, 1.1 ± 0.04 μm and 1.6 ± 0.05 μm, respectively), indicating that ethylene-induced closure requires some cell-to-cell communication between guard cells and mesophyll cells that ABA signalling does not. Alternatively, the competence of guard cells in epidermal peels may have been compromised by the peeling process. Together, these data indicate that, in the absence of ABA, ethylene can induce stomatal closure in leaves, but in the presence of ABA has the opposite effect, in that it inhibits ABA-induced closure. Thus there are likely to be complex interactions between ABA and ethylene, for example in dehydrated or senescent tissue, where the synthesis of both ABA and ethylene is elevated. Stomatal movements are indeed governed via a complex network of signalling components (Fan et al., 2004; Hetherington and Woodward, 2003), and it may be that some components dominate whilst others are only influential in specific physiological or developmental contexts. Determining the variable contributions of each component will be a major task. Preliminary experiments using thermographic imaging indicate that this technique may be a valuable tool to analyse in planta effects of various hormones on transpiration in wild-type and mutant backgrounds (Tagliavia and Desikan, unpubl. data). It should be noted that our data indicate that ethylene might be a significant factor inducing stomatal closure during hypoxic waterlogged conditions, when rapid stomatal closure cannot be attributed to ABA synthesis (Jackson, 2002). That Arabidopsis stomata close in response to ethylene in an intact leaf but not in isolated epidermal peels may also be significant, as bioassays aimed at identifying xylem-mobile substances transported from roots to shoots typically use leaf peels.

The data presented here demonstrate a complex interaction between ethylene and H₂O₂ in which ETR1 is a point of convergence. Ethylene perceived by ETR1 induces H₂O₂ generation in stomatal guard cells. H₂O₂ then subsequently induces closure via a mechanism that requires H₂O₂ perception by sensing machinery that includes ETR1. H₂O₂ was first shown to induce stomatal closure by McAinsh et al. (1996), and since then several studies have combined to place H₂O₂ as a central mediator of stomatal closure. Oligosaccharide elicitors induce stomatal closure via H₂O₂ (Lee et al., 1999), as do ABA (Pei et al., 2000; Zhang et al., 2001) and darkness (Desikan et al., 2004b). H₂O₂ generation induced by ABA is mediated by the D and F isoforms of the NADPH oxidase enzyme Rboh (Kwak et al., 2003). Other scientists have used the atrbohD/F double mutant to show that methyl jasmonate and ozone also induce guard cell H₂O₂ generation via AtrbohD/F (Joo et al., 2005; Suhita et al., 2004). Our data showing that ethylene induces H₂O₂ generation that is required for stomatal closure again emphasize H₂O₂ as a key component of the guard cell signalling network. Although the data of Kwak et al. (2003) suggest some overlap in the functions of AtrbohF and D, this might not be the case in the ethylene-regulated stomatal response, as only the F isoform appears essential for H₂O₂ production. It is likely that regulation of AtrbohF by different stimuli occurs via interaction with other protein complexes in discrete sub-cellular locations. Whether ethylene-induced H₂O₂ production is unique to guard cells or is present in other cell types such as roots, where both ethylene and H₂O₂ play a role in cell growth (Foreman et al., 2003), is not yet known.

The functional link between ETR1 and H₂O₂ in ethylene-induced stomatal closure is indicated by the observations that etr1–1 and etr1–3 stomata do not generate H₂O₂, nor do they close in response to ethylene, demonstrating that ethylene perception in guard cells requires ETR1. etr1–1 stomata are also insensitive to exogenous H₂O₂, whereas etr1–3 stomata respond normally (Desikan et al., 2005). This important difference between etr1–1 and etr1–3 mutants may reflect the different nature of the etr1–1 and etr1–3 mutations. In the etr1–1 mutant, the Cys65 of the ETR1 protein is mutated to Tyr. In the etr1–3 mutant, this Cys65 is retained, the mutation being Ala31Val (Chang et al., 1993). Both mutations result in reduced ethylene binding to ETR1 (Hall et al., 1999). A potential explanation for the H₂O₂ insensitivity of etr1–1, but the retention of H₂O₂ responsiveness in etr1–3, may reside in Cys65 – a potential thiol target for oxidation by H₂O₂. It should be noted that, although the etr1–1 mutant has an altered leaf phenotype in mature plants, its stomatal size is not obviously aberrant and they do retain functionality, by closing in response to ABA (Figure 3) or increased CO₂ (Desikan and Tagliavia, unpubl. data). Furthermore, H₂O₂ synthesis in response to ABA is not impaired. This indicates, that although ABA and H₂O₂ might share some signalling components in the pathway leading to stomatal closure (e.g. Pei et al., 2000), there are others, such as ETR1, that are unique. Differences between ABA- and H₂O₂-mediated regulation of ion channels have been demonstrated elsewhere (Kohler et al., 2003). ABA induction of H₂O₂ synthesis and stomatal closure in the H₂O₂-insensitive etr1–1 mutant indicates that ABA signalling via H₂O₂-dependent but ETR1-independent routes is operational.

The role of ethylene-signalling elements acting downstream of ETR1 in ethylene- and H₂O₂-induced stomatal closure was investigated by analysing the ein2–1, arr2 and ctr1–1 Arabidopsis mutants. CTR1 is a Raf-like protein kinase that interacts physically with the C-terminus of ETR1 and is a negative regulator of the ethylene response pathway in the triple response (Clark et al., 1998; Guo and Ecker, 2004). Here, ctr1–1 guard cells responded like the wild-type to ethylene and to H₂O₂ in the stomatal closure response.

EIN2 is an N-RAMP-like protein acting downstream of ETR1 and functioning in a wide range of ethylene responses in Arabidopsis and other plant species (Guo and Ecker, 2004; Shibuya et al., 2004). However, a role for EIN2 in regulating stomatal function has not been previously reported. Here, neither ethylene nor H₂O₂ caused stomatal closure in the ein2–1 mutant, but ethylene did induce H₂O₂ generation in ein2–1 guard cells. This positions EIN2 downstream of H₂O₂ and demonstrates a requirement for a functioning EIN2 protein for ethylene-induced stomatal closure, as well as indicating a point of convergence for ethylene and H₂O₂ signalling. The details of EIN2 action are not known, but an effect on ion concentrations, in particular calcium, has been suggested (Chen et al., 2005).

The two-component response regulator protein ARR2 has recently been proposed as a new component of ethylene signalling (Hass et al., 2004). In etiolated Arabidopsis seedlings, the loss of ARR2 function results in reduced ethylene sensitivity in the hypocotyl growth response, whereas ectopic expression of a dominant active version of ARR2 causes a triple response-like phenotype (Hass et al., 2004). Furthermore, ARR2 regulates the transcription of an ERF1::LUC reporter gene in a phosphorylation and ethylene-dependent manner in Arabidopsis protoplasts (Hass et al., 2004). The positioning of ARR2 in ethylene signalling has recently been criticized (Chen et al., 2005); however, reduced ethylene sensitivity is only pronounced under certain growth conditions (C. Hass, V. Mira-Rodado, J. Kilian and K. Harter, University of Tübingen, Germany, unpublished data). Our data show that, under the conditions used in the experiments here, guard cells of the arr2 loss-of-function mutant do not close in response to either ethylene or H₂O₂, although they do respond to ABA. These observations confirm a role for ARR2 in ethylene signalling (Hass et al., 2004), and position ARR2 downstream of H₂O₂ in the ethylene-signalling pathway in guard cells.

ABA is central to a complex signalling network within guard cells, and it seems increasingly likely that removal of one hub of the ABA-signalling network can be compensated for (Fan et al., 2004; Hetherington and Woodward, 2003). On the other hand, the ethylene- and H₂O₂-signalling pathway in guard cells may be much more linearly restricted, in that removal of a component such as EIN2 or ARR2 blocks the pathway leading to stomatal closure.

In summary, the data presented herein show that ethylene causes stomatal closure in Arabidopsis via ethylene-induced H₂O₂ synthesis, mediated by the NADPH oxidase isoform AtrbohF. Both ethylene and H₂O₂ signalling are initiated by the ethylene receptor ETR1. These data confirm the crucial role of H₂O₂ as a closure signal elicited by several stimuli in the guard cell signalling network (Hetherington and Woodward, 2003), and highlight ETR1 as a point of convergence for ethylene and H₂O₂ signalling (Figure 5). The data also show that ethylene signalling in guard cells involves proteins such as RAN1 and EIN2 that are active in ethylene signalling in other systems, and is also dependent on a functioning ARR2 pathway. The Arabidopsis stomatal system offers an attractive tool with which to dissect novel aspects of ethylene signalling.

Figure 5.  Potential model of the ethylene signalling pathway in guard cells. Binding of ethylene to ETR1 leads to H₂O₂ production mediated by AtrbohF, and H₂O₂ also signals through ETR1. Downstream of ETR1, both EIN2 and ARR2 are essential but their order and interactions are as yet unknown.

Experimental procedures

Plant material

Wild-type and mutant seeds of A. thaliana ecotype Columbia (Col-0) were sown on Levington's F2 compost and grown under a 16-h photoperiod (60–100 μE m⁻² sec⁻¹), at 22°C and 80% relative humidity in controlled-environment growth chambers (Sanyo Gallenkamp, Loughborough, UK). For the arr2 experiments, both wild-type Landsberg erecta (Ler) or arr2 knock-outs were grown under short-day conditions (8-h photoperiod) with the same light intensity, temperature and relative humidity as above. atrboh seeds were obtained from J. Jones (Sainsbury Laboratory, Norwich, UK) and etr1–1, etr1–3, ein2–1 and ran1–1 seeds were obtained from the Nottingham Arabidopsis Stock Centre (Nottingham, UK).

Stomatal bioassays

Stomatal assays were performed on leaves essentially as described by Desikan et al. (2005). Leaves were floated for 3 h under continuous illumination (60–100 μE m⁻² sec⁻¹) in MES/KCl buffer (5 mm KCl/10 mm MES/50 μm CaCl₂, pH 6.15). Once the stomata were fully open, leaves were treated with various compounds for a further 2.5 h. When inhibitors (DDC, silver nitrate, NAC or DPI) were used, they were added to the MES/KCl buffer on which leaves were floated 15 min prior to other treatments. Control treatments involved addition of buffer or appropriate solvents used with inhibitors. The leaves were subsequently homogenized individually in a Waring blender (Christison Scientific, Gateshead, UK) for 30 sec, and the epidermal fragments collected on a 100 μm nylon mesh (SpectraMesh, BDH-Merck, Nottingham, UK). Stomatal apertures from epidermal fragments were then measured using a calibrated light microscope attached to an imaging system (leica qwin software, Leica, Milton Keynes, UK).

Ethylene and 1-MCP gas treatment

For treatment of leaves with ethylene gas, detached leaves were floated on stomatal opening buffer in open Petri dishes in a gas-impermeable sealed 1 l Kilner jar and placed under continuous light. After 2.5 h, either air or ethylene (99.9%, Fluka, Dorset, UK) was injected at various concentrations into the sealed jar, and leaves incubated for a further 2.5 h. For 1-MCP treatment, 180 mg of Ethylbloc was weighed into a beaker with a stirrer kept in a 1 l Kilner jar (provided by J. Vahala, University of Helsinki, Finland), and 1.8 ml of 0.9% NaOH was injected through the sealed top to dissolve the 1-MCP. From this stock of 3 × 10⁵ ppb, 1 ml of gas was syringed out and injected into a new Kilner jar that had the leaves floating in buffer solution as described above. Following incubation of leaves with 300 ppb 1-MCP for 4 h under light and constant stirring, ethylene gas was injected into the jar containing 1-MCP and incubations continued for a further 2.5 h. Following incubations, leaves were homogenized and stomatal apertures measured as described above.

Measurement of H₂O₂ using confocal microscopy

Epidermal fragments from mature leaves, prepared as described above, were incubated in MES/KCl buffer for 2–3 h. Following this, the fragments were loaded by incubation in 50 μm of the H₂O₂-sensitive fluorescent dye 2′,7′-dichlorodihydrofluorescein diacetate (H₂DCFDA, Molecular Probes, Leiden, The Netherlands) for 10 min. After washing in fresh buffer for a further 20 min, the fragments were challenged with various compounds as indicated in the figure legends. For ethylene gas treatments, fragments were loaded with dye, washed, and subsequently exposed to ethylene gas or air for 30 min in a sealed Kilner jar. Controls included the addition of the appropriate solvents. Confocal laser scanning microscopy was used to visualize fluorescence, using an excitation wavelength of 488 nm and an emission wavelength of 515–560 nm (Nikon PCM2000; Nikon Europe BV, Badhoewvedorp, the Netherlands). Images were acquired and analysed using scion image software (Scion Corp., Frederick, MA, USA) to measure the relative fluorescence intensities in the cells following various treatments. Data represent fluorescence intensities expressed as a percentage of the control values, from several guard cells analysed in different experiments.

Statistical analysis

Where appropriate, data between different batches of experiments were normalized to account for differences in wild-type control (i.e. buffer alone) values between experiments. Means, associated standard errors and P-values were derived from one-way independent factorial anovas followed by non-parametric (Games–Howell) post hoc tests. All statistical analyses were performed using spss for Windows, version 12.0 (SPSS Inc., Chicago, IL, USA).

Acknowledgements

This work was funded by the Biotechnology and Biological Sciences Research Council, The Leverhulme Trust (R.D.), the Stress Imaging EU project (C.T.) and by the Deutsche Forschungsgemeinschaft (K.H.). We would like to thank J. Vahala (University of Helsinki, Finland) for kindly providing 1-MCP, J. Jones (Sainsbury Laboratory, John Innes Centre, Norwich, UK) for the rboh mutant seeds, A. Hetherington (Lancaster University, UK) for use of thermography facilities, and M. Jackson (University of Bristol, UK) for advice.