Guard-cell signalling for hydrogen peroxide and abscisic acid


Author for correspondence:
Chun-Peng Song
Tel: +86 378 3880007
Fax: +86 378 3880006

I. Introduction

Guard cells integrate signals on water availability and CO2 levels, and those from hormones, light and other environmental conditions, to regulate the size of the stomatal aperture. The phytohormone abscisic acid (ABA) is an endogenous antitranspirant that reduces the rate of water loss through stomatal pores in the leaf epidermis. Enhanced biosynthesis of ABA occurs in response to water deficiency, which causes the intracellular redistribution and accumulation of ABA in guard cells. Increased ABA content results in the efflux of ions, loss of turgor of the guard cells, and closure of the stomatal pores (Bray, 1997; Schroeder et al., 2001a).

Although many ABA-signalling intermediates have been identified in guard cells (Schroeder et al., 2001a, 2001b), there are still large gaps in our understanding of the molecular mechanism of stomatal movement. Over the past several years, considerable progress has been made towards understanding the essential role of hydrogen peroxide (H2O2) in ABA-induced stomatal closure. There is considerable evidence to indicate that H2O2 regulates stomatal movement, and that this regulatory action is very similar to that of ABA-regulated stomatal movement. Here we discuss the role of H2O2 as a signalling molecule in plants, particularly in guard cells. We focus specific attention on crosstalk between the ABA-mediated and reactive oxygen species (ROS)-mediated signalling pathways by highlight the points at which crosstalk can occur in the guard-cell signalling network. We also present some personal perspectives that we feel are useful to enhance our current understanding of guard-cell signalling.

II. H2O2 generation and signalling in guard cells

1. Chloroplasts are potential sources of H2O2 in guard cells

Compared with epidermal cells, chloroplasts are a prominent feature of guard cells in most species of higher plants. Guard cell chloroplasts are postulated to have organ-specific functions with respect to the regulation of stomatal movement, and their role in regulating stomatal movement has been a subject of extensive study.

Chloroplasts in guard cells are smaller than those found in mesophyll cells, and have only a few granal stacks of thylakoids in their stroma (Sack, 1987; Willmer & Fricker, 1996). The number of chloroplasts in guard cells is about one-third the number found in mesophyll cells (Allaway & Setterfield, 1972). Further, their chlorophyll content per cell is between 1 and 4% of that in mesophyll cell chloroplasts (Zemel & Gepstein, 1985; Birkenhead & Willmer, 1986; Reckmann et al., 1990). Biochemical studies on the stroma of chloroplasts have detected the presence of the principal enzymes of the Calvin cycle (Shimazaki & Zeiger, 1985; Zemel & Gepstein, 1985; Shimazaki et al., 1989; Willmer & Fricker, 1996). However, the activity of the key enzyme in the Calvin cycle, Rubisco, in guard cells is negligible (Outlaw et al., 1979; Vaughn, 1987; Reckmann et al., 1990; Kopka et al., 1997). If one uses the highest reported values of Rubisco activity (Shimazaki & Zeiger, 1987; Shimazaki et al., 1989), taking into account the low chlorophyll content in guard cell chloroplasts, CO2 fixation by the Calvin cycle should be no greater than 2–4% of that found in mesophyll cells (Outlaw & De Vlieghere-He, 2001). The finding that there is no relationship between photosynthetic capacity and stomatal conductance at high light in transgenic tobacco plants that have reduced amounts of Rubisco and impaired photosynthesis (anti-Rubisco plants) suggests that there may not be a direct mechanism linking the two processes (von Caemmerer et al., 2004). However, although guard cell chloroplasts are generally smaller, less numerous, and have fewer grana than mesophyll chloroplasts (Sack, 1987; Willmer & Fricker, 1996), photophosphorylation, on a chlorophyll basis, has been reported to be as high as 80% of that in mesophyll cells (Shimazaki & Zeiger, 1985). It is suggested that the excess excitation energy from the conversion of light energy to chemical energy is a source of generation of ROS in guard cells. Therefore the chloroplast not only supplies energy, but also is a sensor of environmental information, and the chloroplast redox signalling allows the plant to acclimatize to the environmental stresses (Pfannschmidt, 2003).

2. H2O2 generation in guard cells

The ability of guard cells to generate H2O2 has been demonstrated in tobacco, tomato and Commelina communis (Allan & Fluhr, 1997; Lee et al., 1999). The fungal elicitor cryptogein induced H2O2 generation in guard cells and the surrounding epidermal cells (Allan & Fluhr, 1997). When H2O2 is generated by an oligogalacturonide elicitor or chitosan, stomatal closure occurs in the leaves of tomato and C. communis (Lee et al., 1999). Treatment of epidermis strip with catalase (CAT), an H2O2 scavenger, reduced both stomatal closure and H2O2 generation. These findings confirm that H2O2 is required to initiate stomatal closure.

The application of exogenous ABA resulting in rapid H2O2 generation was first reported in the guard cells of Vicia faba (Miao et al., 2000) and Arabidopsis (Pei et al., 2000). Subsequently, Zhang and colleagues conducted a detailed study of the generation of H2O2 in response to exogenous application of ABA to the guard cells of V. faba (Zhang et al., 2001c). The results of the time-course experiments with single-cell assays using the fluorescent probe dichlorofluorescein showed that generation of H2O2 was dependent on ABA concentration, and that the increase in fluorescence intensity of chloroplasts occurred significantly earlier than within the other regions of guard cells. When ABA was microinjected into the guard cells, marked H2O2 production resulted and this response preceded stomatal closure (Fig. 1). Therefore it appears that the generation of H2O2 is essential for ABA-induced stomatal closure in plants.

Figure 1.

Abscisic acid (ABA)-induced hydrogen peroxide (H2O2) generation and its signal transduction pathways in guard cells. Above, confocal imaging of microinjection ABA into guard cells induced H2O2 generation visualized with the H2O2-sensitive dye DCFH-2DA (2′, 7′-dichlorofluorescein diacetate) (Zhang et al., 2001c). (a) Guard cells loaded with DCFH-2DA before microinjection of 1 µm ABA; (b) the same cells as in (a) after microinjection of 1 µm ABA at 120 s; (c) bright-light images of the fluorescence images in (b); scale bar in (c) for all three images. Below, model for H2O2 signal transduction in guard cells (modified from Cheng & Song, 2006; Neill et al., 2002c). Hydrogen peroxide can be generated from various sources under different environmental conditions. H2O2 might diffuse to cells from the apoplast or generated sources (Zhang et al., 2001c; Neill et al., 2002c), and in turn be redistributed into various subcellular compartments. H2O2 can be sensed by ATGPX3 (Arabidopsis glutathione peroxidase 3) (Miao et al., 2006), and modulate the activities of many components that contribute to cell signalling, such as protein phosphatases (e.g. ABI1 and ABI2) (Murata et al., 2001), protein kinases OST1 (see Section IV; Mustilli et al., 2002), transcription factors (Stone, 2004), and calcium and potassium channels (Pei et al., 2000; Zhang et al., 2001b). R, receptor; red dashed arrows indicate that exact details of the signal pathways are still unknown and need to be addressed. See text for details.

In discussing the potential mechanisms by which H2O2 might be generated in guard cells, two primary sources of H2O2 in guard cells should be considered. First, as mentioned above, chloroplasts are considered to be the principal source of ROS in plant cells (Foyer & Harbinson, 1994). Under normal photosynthesis, chloroplasts generate approx. 150–250 µmol of H2O2 mg−1 chlorophyll h−1. Although ROS are generated during normal cell metabolism, and prevention of their excess production or scavenging is an integral feature of cellular metabolism (Asada, 1992; Foyer et al., 1997), oxidative damage to the plant often occurs under biotic and abiotic stresses. Many environmental variables, such as light, temperature, water or nutrient availability, affect the efficiency of photosynthetic electron transport chain and the redox state of chloroplasts (Foyer & Noctor, 2003; Pfannschmidt, 2003). For instance, singlet oxygen (1O2) is generated in chloroplasts at high light intensities when the absorption of light energy exceeds the capacity for CO2 assimilation. Excited triplet chlorophyll molecules in the photosystem II complex interact with oxygen to generate 1O2 (Fryer et al., 2002; Hideg et al., 2002). At the same time, the hyperreduction of the photosystem II complex favours the direct reduction of O2 by photosystem I and the subsequent production of superoxide anions (inline image), H2O2 and the hydroxyl radical (·OH) (Foyer & Noctor, 2003; Pfannschmidt, 2003).

A second proposed mechanism for generating H2O2 involves NADPH oxidase, which is located in the cell membrane (Pei et al., 2000; Zhang et al., 2001c). Enzymatic sources of H2O2 also result from those reactions that are catalysed by cell wall peroxidases, amine oxidases and other flavin-containing enzymes (Neill et al., 2002b, 2002c). Allan & Fluhr (1997) suggested that H2O2 was generated via intracellular flavin-containing enzymes, apoplastic peroxidases and amine oxidase-type enzymes in guard cells and epidermal cells of tobacco in response to elicitor challenge. A pH-dependent cell wall peroxidase can also generate H2O2 (Peng & Kuc, 1992).

In Arabidopsis, ABA-induced stomatal closure was inhibited by diphenylene iodinium (DPI) (Cross & Jones, 1986). Given that H2O2-induced calcium channel activation is dependent on NADPH, this finding suggests a role for an NADPH oxidase-like enzyme mediating H2O2 formation in response to ABA in Arabidopsis guard cells (Murata et al., 2001).

In the ABA insensitive1 (abi1) and ABA insensitive2 (abi2) point mutant plants with strongly reduced phosphatase activities, it was shown that ABA was unable to generate ROS in the abi1 mutant plants, but ABA could still induce ROS production in the abi2 mutant plants (Murata et al., 2001). These data suggest that the abi2-1 mutation impairs ABA signalling downstream of ROS production (Murata et al., 2001).

3.  H2O2 signalling in stomatal movements

Until relatively recently, H2O2 was viewed mainly as a toxic cellular metabolite. H2O2 is continually generated from various sources during normal metabolism. Generally, ROS have very short half-lives. For example, the life span of 1O2 is almost 4 µs in water and 100 µs in a nonpolar environment (Foyer & Harbinson, 1994). H2O2 is moderately reactive, is a relatively long-lived molecule (half-life of 1 ms) and can diffuse some distances from its production site. Plant cells can generate rapidly and inactivate H2O2 in response to external stimuli. Therefore H2O2 satisfies all the important criteria for being an intracellular messenger. Indeed, there are now many reports in which H2O2 has been demonstrated as a ubiquitous intracellular messenger under physiological conditions (Rhee, 1999).

There are early reports that exposure to methyl viologen, an inline image generator, or exogenous H2O2 has a remarkable effect on the size of the stomatal aperture (Price, 1990), and that inline image is involved in the signal transduction pathway(s) of stomatal movement by altering membrane fluidity (Purohit et al., 1994). Subsequently, a detailed study on C. communis further confirmed the role of ROS in mediating stomatal closure (McAinsh et al., 1996).

It is well known that ABA plays a central role in regulating stomatal closure. This event is closely related to the cellular responses to H2O2 in guard cells (Pei et al., 2000; Zhang et al., 2001b, 2001c). H2O2 can inhibit ABA-induced stomatal closure, and this effect was reversed by ascorbic acid at concentrations lower than 10−5 m. Further, ABA-induced stomatal closure was also abolished partly by the addition of exogenous CAT and DPI. It has been clearly demonstrated that ABA-induced inhibition of stomatal opening is a distinct process from ABA-induced stomatal closure (Mishra et al., 2006), and that ROS and nitric oxide (NO) are also involved in ABA-induced inhibition of stomatal opening (Yan et al., 2007).

The results of whole-cell patch-clamp analysis showed that H2O2 can mimic ABA in its ability to inhibit the inward K+ current, an effect that can be reversed by the addition of ascorbic acid. In addition, the intracellular application of CAT or DPI partly abolished the inhibition of ABA-mediated inward K+ current across the plasma membrane of guard cells (Song et al., 1997; An et al., 2000; Zhang et al., 2001b). These results suggested that H2O2 mediates ABA-induced stomatal movement by targeting inward K+ channels in the plasma membrane.

It has been suggested that H2O2 is an intercellular messenger because it is active in the cell in which it is generated or in neighbouring cells (Rhee, 1999). Cytosolically generated ROS can enter other cells via an apoplastic route (Allan & Fluhr, 1997). Consistent with this conclusion is the observation that unexposed stomatal guard cells, which have no symplastic connection to their epidermal neighbours, also showed increases in intracellular ROS when nearby rose bengal-loaded cells were activated to produce ROS. This characteristic of H2O2 may be important to stomatal response, because autonomous guard cells through this signal can integrate and process multiple complex signals from the environment and respond by opening and closing stomata in order to adapt to the environmental signal.

Although the exact details are still not known, an imperfect picture may be drawn of the role of H2O2 as a signalling molecule mediating stomatal movement in response to ABA and drought. This role is presented schematically in Fig. 1. However, many questions still remain unanswered: What are the concentrations of H2O2 in the different subcellular compartments? Are specific H2O2 responses induced by different stimuli? Can the H2O2 that is generated in a particular cell have an effect in neighbouring cells?

III. Gene expression regulated by H2O2 and ABA

1. Global gene expression profiles

To gain insight into the molecular mechanisms involved in the crosstalk between the ABA and ROS-signalling pathways, large-scale gene-expression analyses have been undertaken by several laboratories in response to ROS and/or ABA (Desikan et al., 2001; Seki et al., 2002; Leonhardt et al., 2004; Takahashi et al., 2004; Vandenabeele et al., 2004). Guard cell expression profiles were compared with those of mesophyll cells, resulting in identification of 64 transcripts expressed preferentially in guard cells. These genes encode transcription factors, signal transduction proteins such as protein kinases, receptor protein kinases, and metabolic pathway proteins. Interestingly, expression profiling reveals ABA modulation of many known guard cell ABA-signalling components at the transcription level (Leonhardt et al., 2004).

Further studies focused on the relationship between ABA- and H2O2-regulated gene expression in Arabidopsis seedlings (Wang et al., 2006). The expression patterns of genes in response to ABA and H2O2 among the 24 000 genes are shown in Fig. 2. Further analysis indicated that 143 of the upregulated genes and 75 of the downregulated genes were responsive to both ABA and H2O2 (Fig. 2). Therefore these results suggest that there is an overlap between ABA- and H2O2-induced transcription of genes, and these two signalling molecules regulated many downstream genes in a coordinated manner (Wang et al., 2006).

Figure 2.

Diagrams showing the classification of genes up- or downregulated in response to hydrogen peroxide (H2O2) and abscisic acid (ABA) treatment (Wang et al., 2006). (a) 143 genes were upregulated by both H2O2 and ABA; (b) 75 genes were repressed by both H2O2 and ABA.

These data are consistent with previous reports that the responses to ABA and oxidative stress are linked (Guan et al., 2000; Murata et al., 2001; Zhang et al., 2001c; Kwak et al., 2003). For example, genes highly induced by ABA, including KIN1, COR47 and rd29A, were also observed in all microarray data. Phenylalanine ammonia lyase 1 (PAL1; At2g37040), DREEB2A and rbohD (a subunit of NADPH oxidase) also showed increased expression levels in two studies (Desikan et al., 2001; Wang et al., 2006). It is also important to identify the functions of previously uncharacterized genes that responded to the application of exogenous ABA and H2O2. However, the functions of many of these genes and their products remain to be determined, and represent areas of future investigation.

2. Transcription factors

Several transcription factors that are redox-controlled have been identified. For example, OxyR is a bacterial H2O2 sensor, first identified as a transcription factor in Escherichia coli and Salmonella species (Stone, 2004; D’Autreaux & Toledano, 2007). This transcription factor is activated by H2O2 via the formation of a disulfide bond that induces significant structural alterations in the protein (Stone, 2004).

In yeast, the transcription factor Yap1 activates the expression of antioxidant genes in response to oxidative stress. The regulation of Yap1 involves its nuclear accumulation: it is activated by oxidation and its reduction is mediated by thioredoxins. However, oxidation of the transcription factor is not done directly by the hydroperoxide, but is mediated by the glutathione peroxidase (GPX)-like enzyme GPX3. The activation of Yap1 is achieved by the formation of a disulphide bond within the Yap1 molecules. Therefore, in this case, the enzyme GPX3 is acting as an H2O2 receptor or sensor, and functions as a redox-transducer to transduce the oxidative signal to the transcription factor (Delaunay et al., 2000, 2002).

Alterations in gene expression during systemic acquired resistance in plants are often modulated by the protein NPR1 (nonexpressor of pathogenesis-related genes 1) and the TGA family of transcription factors. The latter belong to the family of bZIP (basic leucine zipper)-type transcription factors. In the nonactivated state, NPR1 exists as an oligomer that is maintained by intermolecular disulphide bonds. On activation, NPR1 is reduced to a monomeric form, which then accumulates in the nucleus and alters gene expression (Mou et al., 2003). Furthermore, reduction of the cysteine residues in TGA1 and TGA4 allows them to interact with NPR1, which stimulates the binding of TGA1 to DNA (Fobert & Despres, 2005). However, the exact identity of redox-perception proteins in plants is not yet known, although it has been suggested that thioredoxins are likely candidates (Fobert & Despres, 2005).

Recently, AtMYB60 and AtMYB61, two of the R2R3-MYB family of transcription factors in Arabidopsis, were both reported to be specifically expressed in guard cells (Cominelli et al., 2005; Liang et al., 2005). Stomata from wild-type (WT) and atmyb60-1 plants closed to the same extent in the dark. When exposed to light, mutant leaves displayed a significant reduction in the opening of stomatal pores compared with WT. These data suggest that a functional AtMYB60 gene is required for light-induced opening. Wild-type and atmyb60-1 stomatal pores displayed the same degree of reduction in their aperture under ABA treatment, indicating that the loss of AtMYB60 gene function does not result in ABA-hypersensitive stomatal regulation (Cominelli et al., 2005). Similarly, the role of AtMYB61 in regulating stomatal aperture appeared to be independent of ABA, but is involved in the light-to-dark transition in stomatal aperture alteration (Liang et al., 2005). Whether or not the two transcription factors function in H2O2 signalling in guard cell remains unclear.

IV. Points of connection between ABA and H2O2-signalling cascade in guard cells

Abscisic acid and H2O2 are known to regulate stomatal movement via various common components that target a number of cell-signalling regulators. In recent reports, it has been shown that protein phosphorylation, NADPH oxidase, the cytosolic free Ca2+ concentration ([Ca2+]cyt), nitric oxide, cytosolic pH and the cellular redox states are involved in the crosstalk between ABA and H2O2 in regulating guard-cell signalling. These findings may have profound implications for identifying as-yet unknown additional regulatory functions in which these two apparently independent signal molecules might be involved.

1. Protein kinases and protein phosphatases

Members of the type 2C Ser/Thr phosphatase (PP2C) superfamily and the serine/threonine protein kinase OST1 (open stomata 1) dephosphorylation have been shown to play an important role in ABA or H2O2 signal transduction. In plants, several protein phosphatases that can inactivate mitogen-activated protein kinases (MAPKs) in vitro have been characterized. These enzymes include members of the protein phosphatase 2C family, protein tyrosine phosphatase (PTP), and dual-specificity PTP (Tena et al., 2001; Gupta & Luan, 2003).

Two members of the PP2C superfamily, ABI1 and ABI2, are key regulators of ABA-mediated responses (Leung et al., 1997; Gosti et al., 1999; Merlot et al., 2001). Dominant abi1-1 and abi2-1 mutations result in Arabidopsis plants becoming insensitive to ABA signalling for stomatal closure, seed germination and gene expression (Koornneef et al., 1989; Leung et al., 1997; Allen et al., 1999; Hoth et al., 2002). Abscisic acid treatment is known to activate OST1, which is an Arabidopsis orthologue of V. faba AAPK and a positive regulator in ABA-mediating responses in guard cells (Mustilli et al., 2002; Assmann, 2003). Abscisic acid-induced ROS production does not occur in ost1 plants, although the stomata in ost1 plants still close in response to H2O2. In addition, knockout mutant plants for two other members of the PP2C superfamily, PP2C-HAB and PP2CA, display strong hypersensitivity to ABA, and also establish that PP2C-HAB and PP2CA are negative regulators of ABA signalling (Leonhardt et al., 2004; Saez et al., 2004; Kuhn et al., 2006; Yoshida et al., 2006b). Thus these four members of the PP2C superfamily may be prime candidates as targets for ROS in mediating responses to ABA.

A recent study revealed that ABI1 interacts with OST1 and disrupts ABA-mediated activation of OST1 protein kinase activity in the abi1, but not abi2, mutation (Yoshida et al., 2006a). These data, and the relatively strong ABA insensitivity of abi1-1 and ost1-1, indicate that ABI1 and OST1 may function upstream in the ABA-signalling network from the proposed branch points. Several key steps involved in guard-cell sensing of abiotic stresses, which include ABA synthesis, NO production and OST1 kinase, are required for pathogen-associated molecular pattern-induced stomatal closure, which is part of the plant's innate immune response in order to restrict bacterial invasion. This finding suggests that stomata play an important role in host defence (Melotto et al., 2006). An unresolved question regarding the H2O2-signalling pathway is the mechanism whereby the oxidative signal transduces to or interacts with the ABA-signalling pathway. The mechanism or manner in which OST1 regulates ROS production directly via the NADPH oxidase also remains unknown.

Protein tyrphosphatase (PTP)  In barley aleurone protoplasts, it was reported that the specific PTP inhibitor phenylarsine oxide (PAO) blocked the activation of MAPK by ABA (Knetsch et al., 1996). This result suggests that PTP is a positive regulator of MAPK activation. However, in vitro experiments have shown that the Arabidopsis PTP1 (AtPTP1) is reversibly inactivated by H2O2, and the inactivation of AtPTP1 is strongly associated with the activation of Arabidopsis MAP kinase 6 (AtMPK6) (Gupta & Luan, 2003). These data suggest that AtPTP1 may be a primary target for ROS signalling, and is an activator of a MAPK cascade.

Using two specific PTP inhibitors, PAO and 3,4 dephosphatin (3,4 DP) in a leaf epidermis bioassay, MacRobbie (2002) suggested that PTP activity is essential for stomatal closure induced by four different factors: ABA, H2O2, elevated intracellular calcium concentration, and darkness. Pretreatment of the epidermis with PAO inhibited ABA-induced H2O2 production and reversed ABA-induced stomatal closure (Shi et al., 2004). Phenylarsine oxide also enhanced the sensitivity of the inhibitory action of ABA on seed germination (Reyes et al., 2006). At 10 µm ABA, prevention of MAP kinase activation by PD98059 partially inhibited stomatal closure and reduced the ion efflux transient (MacRobbie & Kurup, 2007). However, the effects of PAO and 3,4 DP on the activation of MAPKs were not examined in these studies. It is still not known whether PTP is a negative regulator of MAPK activation in vivo. Furthermore, there is also no information on the role of PTP in ABA-induced antioxidant defence responses in plants.

Mitogen activated protein kinases (MAPK)  The results of several studies strongly suggest that MAPKs are involved in stomatal responses, as well as other plant responses to ABA and H2O2. Burnett et al. (2000) identified a MAPK that is activated by ABA in pea guard cells (Pisum sativum). Tobacco MPK4 (NtMPK4) is also expressed preferentially in the epidermis, and the enhanced sensitivity to ozone in NtMPK4-silenced plants was caused by an abnormal regulation of stomatal closure in an ABA-independent manner (Gomi et al., 2005). MAPK modules affect not only the stomatal aperture, but also the density of stomata. For example, null mutations in the Arabidopsis mitogen-activated protein kinase kinase kinase YODA lead to excess stomata, whereas constitutive activation of YODA eliminated stomata (Bergmann et al., 2004).

It has been found that ABA and H2O2 can activate the same MAPK (Lu et al., 2002; Desikan et al., 2004), and that MAPK mediates both ABA- and H2O2-induced stomatal closure (Desikan et al., 2004). This suggests that ABA and H2O2 may converge on those MAPK-signalling pathways that are involved in regulating stomatal closure. It has been demonstrated that the specific MEK1/2 inhibitor PD98059 abolished the ABA-induced H2O2 generation and ABA-mediated stomatal closure in epidermal peels of V. faba (Jiang et al., 2003, 2008). This finding further reveals that MAPKs are involved in ABA- or H2O2 signal transduction pathways in guard cells. However, no specific MEK1/2 gene has yet been cloned to confirm that it controls stomatal behaviour.

Recently, it has been reported that ROS acts upstream of the MAPK cascade in the ABA-induced antioxidant defence response in maize (Zhang et al., 2006). Time-course analysis of ROS production and MAPK activation showed that the accumulation of H2O2 preceded the activation of MAPK following activation of the ABA-signalling pathway. Abscisic acid-induced activation of MAPK was almost fully arrested by pretreatment with inhibitors of ROS production such as DPI and imidazole, both of which inhibit NADPH oxidase, and tiron and dimethylthiourea, which are scavengers for inline image and H2O2, respectively. Collectively, these data clearly suggest that ABA-induced H2O2 production activates MAPK, which in turn leads to the upregulation of antioxidant defence systems in plants. Together, the data suggest that the MAPK cascade-dependent increase in ABA-induced H2O2 production could be an amplification loop in ABA signalling. The possible existence of positive amplification loops in ROS signalling has previously been reported in plants in response to elicitors (Yoshioka et al., 2003) and oxidative stress (Rizhsky et al., 2004). It has also been demonstrated that MPK3 antisense plants are less sensitive to exogenous H2O2 with respect to the inhibition of stomatal opening and the promotion of stomatal closure. This finding suggests that MPK3 is involved in the H2O2 signal transduction pathway in guard cells of Arabidopsis. Because ABA-induced H2O2 generation was normal in these plants, this result indicates that MPK3 probably acts downstream in the cell-signalling pathway of H2O2 (Gudesblat et al., 2007).

Thus several types of protein kinase have been shown to be activated in the presence of H2O2. However, whether these kinases act in stand-alone manner, or via the formation of macromolecular complexes, is not known.

2. NADPH oxidase

The two partially redundant Arabidopsis NADPH oxidase catalytic subunit genes AtrbohD and AtrbohF function in ABA signal transduction in guard cells (Kwak et al., 2003). Interestingly, compared with the wild type, the AtrbohD/F double mutant plants produce less ABA-induced ROS, and ABA-activated plasma membrane Ca2+-permeable channels and the ABA-induced stomata closure are also disrupted (Kwak et al., 2003). The application of exogenous H2O2 restored the ability of the Ca2+ channel activation and stomatal closure in response to ABA stimulation in these double mutant plants. AtrbohD phosphorylation of S343/347 appears to be necessary, but not sufficient for full activation of RbohD, and phosphorylation sites S343/347 of RbohD are conserved in RbohF (Nühse et al., 2007). Therefore, in guard cells, the direct phosphoregulation of RbohD by CDPKs may reflect a new paradigm for regulation of these plant NADPH oxidase enzymes. These data provide direct molecular genetic and cell biological evidence that ROS are rate-limiting second messengers in ABA signalling, and that the AtrbohD and AtrbohF NADPH oxidases function in guard cell ABA signal transduction. These mutants will no doubt be an important research tool to dissect downstream responses to ABA and other signalling molecules likely to interact with ROS.

Keller et al. (1998) reported that Arabidopsis plasma membranes contain an enzyme (RbohA) that is closely related to the neutrophil NADPH oxidase, both of which have a large hydrophilic N-terminal domain with two EF hand motifs. The strong calcium binding by the EF hand motifs in the RbohA N-terminal region provides a potential mechanism for direct regulation by calcium. In addition, NADPH oxidases extracted from tomato (Solanum lycopersicum) membranes can be regulated in vitro by calcium (Sagi & Fluhr, 2001). Excitingly, two potato calcium-dependent protein kinases, St CDPK4 and St CDPK5, phosphorylated only Ser-82 and Ser-97 in the N terminus of NADPH oxidase in a calcium-dependent manner and regulated the oxidative burst (Kobayashi et al., 2007). However, further experimental evidence is still needed to show that this process does occur in guard cells. Furthermore, ROPs (Rho-related GTPases from plants) are a class of plant proteins that are closely related to the mammalian Rac family (Agrawal et al., 2003). Very recently, using yeast two-hybrid assay, pulldown and fluorescence resonance energy transfer (FRET) microscopy, Wong et al. (2007) found that interaction between Rac GTPases and the N-terminal extension was ubiquitous in plants, and the substantial part of N-terminal region of Rboh, including the two EF-hand motifs, was required for the interaction. The FRET analysis also indicates that cytosolic Ca2+ concentration may regulate Rac–Rboh interaction in a dynamic manner. Taken together, the results suggest that cytosolic Ca2+ concentration may modulate NADPH oxidase activity by regulating the interaction between Rac GTPase and Rboh.

As mentioned above, the activation of MAPKs can amplify ROS signals by regulating NADPH oxidase activity directly, or activating transcription factors to enhance the expression of NADPH oxidase genes in ROS signal transduction (Mittler et al., 2004). ABA can enhance the gene expression (Kwak et al., 2003) and activity (Jiang & Zhang, 2002) of NADPH oxidase. Therefore the ABA-activated MAPK might enhance ROS signals via the activity of NADPH oxidase.

3. Cytosolic free Ca2+ concentration

The [Ca2+]cyt is involved in guard-cell ABA signalling (Roelfsema & Prins, 1995) and H2O2 signalling (McAinsh et al., 1996). Changes in [Ca2+]cyt have been associated with stimuli that lead to stomatal closure. These changes in [Ca2+]cyt depend on calcium release from intracellular stores and on calcium influx across the plasma membrane. First, a calcium channel that is activated when the cell membrane is hyperpolarized, and that is regulated by ABA and H2O2, has been identified in the guard cell plasma membrane (Hamilton et al., 2000; Pei et al., 2000). Treatment with low (< 10−5 m) concentrations of methyl viologen and H2O2 can activate the calcium channel to elevate [Ca2+]cyt (McAinsh et al., 1996). The rise in [Ca2+]cyt activates the intracellular calcium sensor protein, calmodulin, which in turn communicates the signal to activate a downstream target CAT that then lowers the intracellular concentrations of H2O2. Collectively, these results provide evidence that calcium has dual functions in the regulation of cellular H2O2 homeostasis in plants, which in turn influences redox signalling in response to environmental signals (Hung et al., 2005).

It has been found that H2O2-induced activation of calcium-permeable (ICa) channels and H2O2-induced stomatal closing are abolished in ABA-insensitive mutant gca2 plants (Himmelbach et al., 1998), and this observation provides genetic evidence for the role of ROS and ICa channels in ABA signalling (Murata et al., 2001). In the abi1-1 mutant, ABA activation of ICa channels was disrupted, but H2O2 activation of ICa channels and H2O2-induced stomatal closing were not disrupted. The abi2-1 mutation, also disrupted ABA activation of ICa; however, in contrast to abi1-1, abi2-1 impaired both H2O2 activation of ICa and H2O2-induced stomatal closing (Murata et al., 2001).

The recent findings of Mori and colleagues show that two Arabidopsis guard cell-expressed CDPK genes, CPK3 and CPK6, have important functions in the regulation of guard-cell ion channels, and provide genetic evidence for the existence of calcium sensors that transduce stomatal ABA signalling (Mori et al., 2006). ABA and calcium activation of slow-type anion channels and, interestingly, ABA activation of plasma membrane calcium channels were impaired in the guard cells of independent alleles of single and double cpk3cpk6 mutant plants. Furthermore, ABA- and calcium-induced stomatal closing was partially impaired in these cpk3cpk6 mutant alleles. However, rapid-type anion channel current activity was not affected, and this finding is consistent with the partial stomatal closing response in double mutants via a proposed branched signalling network. Impairment in ROS signalling partially impairs, rather than abolishes, ABA-induced stomatal movement (Kwak et al., 2003). The present findings are consistent with a model in which there exists an additional signalling pathway in the ABA signal transduction network, and which runs in parallel to the ABA → ROS → ICa-signalling pathway (Köhler et al., 2003; Mori et al., 2006).

Are there any other differences between ABA- and H2O2-induced increases in [Ca2+]cyt? The observation that H2O2 triggers calcium oscillations with a ‘calcium fingerprint’ but not an ‘ABA fingerprint’ in the Arabidopsis det3 mutant (Allen et al., 2000) suggests that the two signalling cascades are different, although they might have shared components. Köhler et al. (2003) reported that ROS may not be a critical second messenger for ABA signalling in guard cells. Their study compared the responses of K+ channels to exogenous H2O2 and ABA, and they found that like ABA, H2O2 treatments suppressed inline image, shifting its activation to more negative voltages. However, unlike ABA, H2O2 also depressed inline image, and the effect on both K+ channels was irreversible. The results of previous studies have shown that exogenous H2O2 application causes partial closure of the stomata when compared with ABA, which causes complete closure (Pei et al., 2000), and that the stomatal responses to ABA are different to those obtained following the application of exogenous H2O2 or ozone (Torsethaugen et al., 1999; Allen et al., 2000).

4. Nitric oxide

Many documents have demonstrated that NO, like H2O2, is also essential in ABA promoting stomatal closure (Neill et al., 2002a, 2002b). Application of PTIO (or cPTIO), an NO scavenger, inhibited ABA-induced stomatal closure, indicating the involvement of endogenous NO in this stomatal response to ABA (Desikan et al., 2002; Garcia-Mata & Lamattina, 2002; Neill et al., 2002a). By contrast, different NO donors induce stomatal closure in pea, Arabidopsis, tomato, barley and wheat (Neill et al., 2002a). Further investigations showed that ABA-mediated NO generation is in fact dependent on ABA-induced H2O2 production in V. faba and Arabidopsis guard cells (et al., 2005; Bright et al., 2006). The genetic data demonstrated that nitrate reductase, but not nitric oxide synthase (NOS), was the major source of NO in guard cells in response to ABA-mediated H2O2 synthesis (Bright et al., 2006). However, in the abi1-1 and abi2-1 mutants, NO synthesis still occurs in response to ABA. Treatment with the NO donor sodium nitropusside did not induce stomatal closure in these mutants, indicating that the action of both phosphatase enzymes occurs downstream of NO synthesis (Desikan et al., 2002). NO and ROS were also found in the ABA inhibition of stomatal opening by Yan et al. (2007); their results showed that NO scavenger or NOS activity inhibition not only reversed the ABA inhibition of stomatal opening, but also reversed the process in the H2O2 inhibition of stomatal opening. This result indicates that NO is synthesized by a NOS-like enzyme and downstream of ROS in the signalling of preventing stomatal opening. Clearly, both H2O2 and NO are synthesized, perhaps in parallel, in response to ABA. Therefore to discern the spatial and temporal coordination of these events during stomatal movements, it is necessary to develop new approaches to monitor simultaneously the generation of H2O2 and NO in response to ABA. Various ABA-signalling mutants defective in stomatal responses will also provide useful tools to position H2O2 and NO in the signalling scheme in guard cells.

5. Ethylene, methyl jasmonate and salicylic acid

One of the members of the family of ethylene receptors, ETR1, is a well characterized hybrid histidine kinase (HK) in Arabidopsis, the molecular genetics and physiological and biochemical functions of which have been analysed extensively (Guo & Ecker, 2004). The results of a focused study on ETR1 from Arabidopsis showed that it is essential for H2O2 perception and its stimulation results in stomatal closure (Desikan et al., 2005). The etr1-1 mutant plant, which contains a Cys65Tyr mutation, has an incomplete stomatal closure in response to H2O2, and this result suggests that the thiol of Cys65 is pivotal in the responses to H2O2 in Arabidopsis guard cells. The results of experiments using mutant plants that lacked either HK activity or the complete HK domain of ETR1 indicated that the kinase domain was not required for H2O2 signalling. In its other role as an ethylene receptor, it appears that the presence of the HK domain of ETR1 is required (Gamble et al., 2002), suggesting that the signalling through ETR1 invoked by H2O2 is different from that invoked by the presence of ethylene. Although the exact mechanisms of ETR1 signalling in both yeast and Arabidopsis have yet to be determined, it appears that thiol modification of the protein may be crucial. Furthermore, it is not known whether this modification was a direct effect of H2O2, or was mediated by another H2O2-sensing protein. However, a controversial result indicated that ABA-induced stomatal closure could be inhibited by application of ethylene or the ethylene precursor 1-aminocyclopropane-1-carboxylic acid (ACC) (Tanaka et al., 2005).

Methyl jasmonate (MeJA) elicits stomatal closing similar to ABA. The Arabidopsis MeJA-insensitive coronatine-insensitive 1 (coi1) mutation impaired MeJA-induced but not ABA-induced stomatal closing (Munemasa et al., 2007). MeJA-induced production of ROS and NO was blocked in coi1 guard cells. Meanwhile, in coi1 guard cell protoplasts, ABA could activate both slow anion currents and Ca2+-permeable channels, while MeJA did not elicit either slow anion currents or Ca2+-permeable cation currents. Also, MeJA did not induce stomatal closing, but stimulated production of ROS and NO in the abi2-1 mutant. These results suggest that MeJA triggers stomatal closing via a receptor distinct from the ABA receptor, and that the coi1 mutation disrupts MeJA signalling upstream of the blanch point of ABA signalling and MeJA signalling in Arabidopsis guard cells.

Many plant pathogens can penetrate leaf tissues through stomatal opening, so narrowing stomatal apertures may be advantageous for plant defence (Melotto et al., 2006). Work in our laboratory (Dong et al., 2001) provided the evidence that H2O2 may function as an intermediate in SA signalling in guard cells; the effect of stomatal closure induced by SA evidently could be reversed by CAT or ascorbic acid, in contrast to previous observations that application of SA to DCFH-loaded epidermal tissue caused little or no detectable change in fluorescence (Allan & Fluhr, 1997). Similarly, application of SA to the epidermal peels evoked an elevation of inline image; in addition, inline image-specific scavengers superoxide dismutase and 4,5-dihydroxy-1,3-benzenedisulfonic acid (Tiron), or salicylhydroxamic acid (an inhibitor of guaiacol peroxidase), suppressed the SA-induced stomatal closure (Mori et al., 2001). These results support the concept of involvement of ROS in signal transduction in SA-induced stomatal closure.

6. G Proteins

Recent studies have shown that the Arabidopsis G protein α-subunit GPA1 is also a vital regulator of plant growth and development as well as signal transduction in guard cells. T-DNA insertion mutants of GPA1 exhibit greater rates of water loss than their wild-type counterparts (Wang et al., 2001). GPA1 has been demonstrated to be involved in the regulation of inward K+ channels and pH-independent ABA-activation of slow anion channels. These results suggest that G proteins are involved in the ABA-signalling network in guard cells. Interestingly, a modest early O3 increase in ROS production by guard cell chloroplasts is observed in gpa1-4 mutant plants, but not in the agb1-2 mutant plants (a T-DNA knockout insertion mutation Gβ in subunits) (Joo et al., 2005). The gpa1-4 mutant lacking the Gα protein exhibits no membrane fluorescence at any time during the O3 stress response, suggesting that the Gα is absolutely required for activation of the DPI-inhibitable membrane-bound ROS-generating system. However, full activity of Gβ protein (or possibly its stability, or that of the Gβγ complex) depends on the presence of the Gα protein. Therefore the α- and β-subunits of the Arabidopsis heterotrimeric G protein act both synergistically and separately in activating different intracellular ROS-generating systems.

Recently, a G protein-coupled receptor (GCR2) has been identified as a plasma membrane ABA receptor (Liu et al., 2007). Lui and colleagues demonstrated that GCR2 interacts with the G protein subunit GPA1 genetically and physically to mediate all known ABA responses in Arabidopsis. They also reported that this receptor binds ABA with high affinity at physiological concentrations with expected kinetics and stereospecificity. Overexpression of GCR2 results in an ABA-hypersensitive phenotype. However, two recent studies argue that GCR2 is not likely to be either a transmembrane protein or a G protein-coupled receptor (Gao et al., 2007; Johnston et al., 2007). Both studies suggest that GCR2 is probably a plant homologue of bacterial lanthionine synthetases. Thus the exact function of GCR2 in plants requires more detailed studies.

In addition, Chen et al. (2004) reported that extracellular calmodulin (ExtCaM) stimulates a cascade of intracellular signalling events to regulate stomatal movement. Following an analysis of the changes in [Ca2+]cyt and H2O2 in V. faba guard cells and those obtained using the epidermal strip bioassay, Chen and colleagues suggested that ExtCaM induces an increase in both H2O2 levels and [Ca2+]cyt, and results in a size reduction of the stomatal aperture. The results of pharmacological studies have implicated the heterotrimeric G protein in the transmission of the ExtCaM signal, which acts at a site upstream from the sites at which increases in [Ca2+]cyt and H2O2 generation occur in guard cell responses (Chen et al., 2004). In addition, the guard cells of gpa1 mutants have impaired ExtCaM induction of H2O2 generation. Collectively, Chen and colleagues’ results strongly suggest that ExtCaM activates an intracellular signalling pathway that involves the activation of a heterotrimeric G protein, H2O2 generation, and changes in [Ca2+]cyt in the regulation of stomatal movements.

7. Phosphatidylinositol (PI) signalling system

Phosphoinositide metabolism has been shown to play important roles in ABA-induced changes of [Ca2+]cyt and stomatal closing (Gilroy et al., 1990; Lee et al., 1996; Staxén et al., 1999). Phosphatidylinositol (PI) kinases include PI 3-kinase (PI3K) and PI 4-kinase (PI4K), both of which synthesize PI 3-phosphate (PI3P) and PI 4-phosphate (PI4P). The results of previous studies have shown that PI3P and PI4P exist in the guard cells of C. communis, and that they could be involved in guard-cell signalling (Parmar & Brearley, 1993, 1995; Jung et al., 2002). Inhibition of either phospholipase C (PLC) or phospholipase D (PLD) partially reverses ABA-induced stomatal closure. However, the simultaneous application of both inhibitors (U-73122, an inhibitor of PLC; 1-butanol, a selective inhibitor of phosphatidic acid production by PLD) does not result in enhanced attenuation of ABA-induced stomatal closure. These results suggest that PLD seems to function in the same pathway as PLC (Jacob et al., 1999). However, the relative location of the two phospholipases in the signalling chain remains unknown.

The PI signalling system is also involved in methyl viologen-induced and H2O2-regulated stomatal movements (Zhou et al., 2002). H2O2-induced stomatal closure can be partially reversed by neomycin (an inhibitor of phosphoinositide PLC), heparin (a competitive inhibitor of inositol 1,4,5-triphosphate, IP3) and lithium chloride (an inhibitor of inositol monophosphatase). Moreover, using [3H] phosphatidylinositol 4,5-bisphosphate ([3H]PIP2) as substrate of PLC, we observed that the external application of H2O2 or ABA could activate PLC in the protoplasts of V. faba (Zhou et al., 2002).

Zhang et al. (2003) reported that PLD is activated by H2O2 in Arabidopsis protoplasts. Yamaguchi et al. (2004) showed that PLD activation by H2O2 enhances phytoalexin biosynthesis in rice cells, and this result illustrates an important link between ROS generation and the PLD-mediated signalling cascade. They also showed that PLD activation by H2O2 involves a protein tyrosine kinase (PTK), which is a critical new pathway that leads to PLD activation. PLD-mediated phosphatidic acid (PA) production has been shown to promote ABA-induced stomatal closure and gene expression, as well as oxidative stress, by multiple mechanisms (Zhang et al., 2005).

In animal cells, ROS generation in neutrophils is activated by phosphatidylinositol 3-phosphate (PI3P). Considering that guard cells display PI3P and PI 3-kinase activity, Park et al. (2003) investigated the role of PI3P in ROS generation in guard cells exposed to ABA. They found that the PI3-kinase inhibitors wortmannin or LY294002 inhibited ABA-induced ROS generation and stomatal closing. The endosome-binding domain of human EEA1 (early endosomal antigen 1), which specifically binds to PI3P, also inhibits ABA-induced ROS generation and stomatal closing when overexpressed in guard cells. The application of exogenous H2O2 partially reversed the effects of wortmannin or LY294002 on ABA-induced stomatal closing (Park et al., 2003). These results support a role for PI3P in ABA-induced ROS generation and stomatal closing. In neutrophils, PI3P regulates H2O2 production by binding to the noncatalytic component p40phox of the NADPH oxidase (Ellson et al., 2001). However, a plant homologue of p40phox has not been reported. Therefore the detailed mechanism of action of PI3P during ROS generation in plant cells awaits further investigation.

8. pH and redox states

ABA-induced alkalization of the cytosol is the trigger for activation of outward K+ channel in the plasma membrane of guard cells (Blatt & Armstrong, 1993). MacRobbie (1997) suggests that pH may serve as a second messenger for ABA in a calcium-independent pathway. Consistent with the notion that ABA inhibits IKin in guard cells, IKin is activated by acidification of cytosol (Grabov & Blatt, 1997). The results of studies on cloned K+ channels provide a molecular basis for pH modulation of inward K+ channels such as KAT1 and KST1, both of which are expressed in guard cells (Hoth et al., 1997; Hoth & Hedrich, 1999; Tang et al., 2000). The fact that the activity of these channels is regulated in a membrane-delimited manner indicates that pH may directly modify the channel protein (Miedema & Assmann, 1996; Hoth et al., 1997).

Analysis of confocal pH imaging using the fluorescent pH-sensitive probe 5-(and-6)-carboxy seminaphthorhodafluor-1-acetoxymethylester (SNARF-1-AM) revealed that the external application of H2O2 leads to rapid alkalization in the cytoplasm and acidification of the vacuoles of guard cells of V. faba. This effect could be abolished by pretreatment with butyric acid (Zhang et al., 2001a). These results suggest that the alkalization of cytoplasm via efflux of cytosolic protons into the vacuoles in guard cells following an H2O2 challenge is an important event in the early stage of activation of the signal cascade that leads to stomatal closure (Zhang et al., 2001a). In another study, analysis of the kinetics of changes in cytosolic pH and ROS production show that the alkalization of cytoplasm preceded ROS production during the stomatal response to both ABA and methyl jasmonate (Suhita et al., 2004). The finding that cytosolic pH changes before and after ROS production may reflect the parallel and branched nature of the ABA-signalling network in guard cells (Leung & Giraudat, 1998; Schroeder et al., 2001a; Hetherington & Woodward, 2003; Fan et al., 2004).

Several researchers have suggested that ABA/ROS signalling in guard cells is controlled by the ascorbic acid redox state (Pnueli et al., 2003; Chen & Gallie, 2004; Davletova et al., 2005). A knockout mutation in ascorbate peroxidase 1 (APX1) in Arabidopsis resulted in the accumulation of H2O2. In addition, the transcript levels of catalase and chloroplast superoxide dismutase were reduced, but there were no significant changes in the transcript levels of glutathione peroxidase (Pnueli et al., 2003). Furthermore, the stomatal apertures of the apx1 knockout plants did not close in response to darkness or open in response to light (Pnueli et al., 2003). APX1 has also been shown to play a role in protecting the chloroplast from oxidative damage (Davletova et al., 2005). Interestingly, Arabidopsis plants with increased dehydroascorbate reductase (DHAR), the enzyme that generates ascorbate from dehydroascorbate, have an increased ascorbate redox state and reducted H2O2 levels in the guard cells (Pnueli et al., 2003). Chen & Gallie (2004) reported that guard cells with an increase in the ascorbate redox state were less responsive to H2O2 or abscisic acid stimulation. Furthermore, they found that these plants exhibited increased water loss under drought conditions, whereas suppressing DHAR expression conferred increased drought tolerance.

9. Peroxisensor

Generally, intracellular H2O2 levels appear to vary between 1 and 700 nm (Stone, 2004). When the intracellular levels are > 1 nm, H2O2 is toxic to cells and cell death responses are initiated (Stone, 2004). For proteins to perceive the presence of ROS, such as H2O2, in order to act as signal intermediaries, there needs to be specific recognition of H2O2 by the protein, or a direct chemical interaction that leads to signal propagation. The former is unlikely because of the small molecular size of H2O2. However, the latter is likely because the oxidizing nature of H2O2 will allow it to modify thiol residues in proteins directly. Furthermore, the chemistry of H2O2 sensing dictates that the proteins will have a commonality, such as active thiol groups becoming potential ROS targets (Cooper et al., 2002; Vranováet al., 2002; Foyer & Noctor, 2005; D’Autreaux & Toledano, 2007).

There are redox groups within proteins that potentially can toggle between oxidation and reduction states in a rapid and ROS dose-dependent manner; in doing so, the structures of the proteins will be altered and such proteins may partake in H2O2-mediated signalling. It is well known that glutathione peroxidases (GPX) are considered to be key enzymes involved in scavenging oxyradicals in animals (Arthur, 2000). Interestingly, Delaunay et al. (2000, 2002) demonstrated that GPX3 functions in both hydroperoxide sensing and scavenging in budding yeast.

Several cDNAs encoding proteins that are similar to animal GPXs have been isolated from diverse plant sources. It has also been demonstrated that most plant GPX proteins have a primary structure that is similar to that of animal phospholipid hydroperoxide glutathione peroxidase (PHGPX) (Churin et al., 1999). However, GPXs in higher plants have lower activity than those in animals, because they contain a cysteine (Cys) residue at the putative catalytic site instead of the selenocysteine (Sec), which is found in animal GPXs. This low activity has made it difficult to clarify the potential physiological role of GPX in higher plants.

Previous reports showed that Arabidopsis glutathione peroxidase (ATGPX) might be a potential detoxicant of H2O2, using GSH directly as a reducing regent (Eshdat et al., 1997). Recent study indicated that ATGPX3 mutations have impaired ABA and drought stress responses, and are sensitive to and produce more H2O2 in their guard cells under stress conditions. The strong expression of ATGPX3 in guard cells suggests that ATGPX3 functions in the stomata to control water loss (Miao et al., 2006).

How do these findings relate to H2O2 perception and signal relaying? Biochemical data indicate that transient inactivation of ABI1 and ABI2 phosphatase by H2O2 would allow or enhance the ABA-dependent signalling process (Meinhard & Grill, 2001; Meinhard et al., 2002). ATGPX3 interacted strongly with ABI2 and relatively weakly with ABI1. Interestingly, the redox states of both ATGPX3 and ABI2 were altered after exposure to H2O2, and the changes in redox state of ATGPX3 and ABI2 were coupled in vitro. These results suggest that ABI2 represents a likely target for redox regulation by the oxidized form of ATGPX3 in the ABA-signalling pathway (Miao et al., 2006). Therefore it is also possible that ATGPX is a conserved enzyme in plants, with a function similar to that of the yeast H2O2 sensor or transducer, whereby ABA and H2O2 directly regulate the gene transcription through a GPX-mediated perception of H2O2. However, the biochemical and structural basis of ATGPX-like enzymes as a hydroperoxide sensor or redox transducer, as shown for the Yap-1 system in yeast (Delaunay et al., 2002), remains to be established in plant cells.

V. Future challenges

There are numerous points of interaction between ROS and ABA in guard-cell signalling. These points of interaction are located at the core of the fine control of stomatal movement in response to changes in environmental conditions (Fig. 3). Therefore elucidating the details of the signalling networks in guard cells for ABA and ROS will be a challenging and fascinating task.

Figure 3.

Simplified working model showing the genetic interactions and putative functions and locations within the reactive oxygen species (ROS)-mediated abscisic acid (ABA)-signalling network in guard cells. Positive and negative regulators are coloured green and red, respectively. Dashed arrows indicate that exact details of the signal pathway are still unknown. Carotenoid, the precursor from which ABA is synthesized; PA, phosphatidic acid; PLC, phospholipase C; PLD, phospholipase D; PIP2, phosphatidylinositol 4,5-bisphosphate; IP3, trisphosphate inositol; GCR2, G protein-coupled receptor; PI3K, phosphatidylinositol-3 kinase; PI4K, phosphatidylinositol-4 kinase; ATGPX3, Arabidopsis glutathione peroxidase 3; GCA2, growth controlled by abscisic acid 2; ABI1, ABA insensitive1; ABI2, ABA insensitive2; [Ca2+]cyt, cytosolic free Ca2+ concentration; OST1, open stomata1; Asa/redox, ascorbic acid redox state; TRX, thioredoxin; GRX, glutaredoxin; GPA1, the Arabidopsisα-subunit of the trimeric G protein; TF, transcription factor; MAPK, mitogen-activated protein kinases; CDPK, calcium-dependent protein kinases; ETR1, ethylene response 1; ROS scavengers include both enzymatic and nonenzymatic antioxidants such as glutathione, vitamin E, superoxide dismutase, glutathione reductase, peroxidases, ascorbate peroxidase and catalase.

During the past decades, much progress has been made toward understanding ABA signalling in guard cells using molecular genetics approaches. Clearly, H2O2 plays a central role in the guard cell ABA-signalling network. Recently, a high-throughput genetic screen for Arabidopsis-deficient mutants in guard-cell signalling using thermal imaging has been developed (Merlot et al., 2002). Using this system, many mutant plants with altered stomatal responses to H2O2 have been identified (Song et al., 2006). This system will be used eventually to identify other genetic loci that may play major roles in the mechanism of H2O2-induced stomatal closure. Furthermore, this system could be useful for resolving the question of how these new components interact with known important molecules in the ABA signal transduction pathway.

The guard cell is an elegant system that is unique, and that can be used for studying signal transduction and dissecting an intricate network of signalling pathways. H2O2 is generated in response to ABA, and causes a reduction in the size of the stomatal aperture. We are now entering an exciting period in the study of ROS signalling in plants. It is time to fit the pieces of the puzzle into place. The localized and temporal production of ROS is likely to be extremely critical in the cellular and intracellular transduction of ROS signals. Although the highly compartmentalized nature of the enzymes involved in ROS scavenging in cells is well defined, we still have much to discover about the initiation of ROS signalling, the sensing and response mechanisms, and the regulatory mechanism(s) of their production and inactivation. To this end, we need to answer many questions, including: What is the H2O2-specific sensor or receptor? What pathways manage the level of ABA and ROS in guard cells? At the same time, we do not know what is the critical process or what are the rate-limiting steps when H2O2 is perceived by the cells. We also do not know which genes are exclusive or indispensable for H2O2 function, and whether the expression of these genes is tissue-specific. Another question might be: How do ROS-mediated pathways integrate with the known signal transduction networks involved in development, growth and stress responses? To answer such questions will require combining methods of functional genomics with those required for genetic and biochemical analysis, and then applying them to guard cells.


We thank Pengcheng Wang for help in preparing the manuscript. Research in the laboratory of C.-P. Song is supported by grants from the National Natural Sciences Foundation of China (30530430 and 30625005) and from the National Key Basic Special Funds (2003CB114305).



 Summary 1 
I.Introduction 1 
II.H2O2 generation and signalling in guard cells 2 
III.Gene expression regulated by H2O2 and ABA 4 
IV.Points of connection between ABA and H2O2-signalling cascade in guard cells 5 
V.Future challenges 11 
 Acknowledgements 12 
 References 12