Ozone-triggered rapid stomatal response involves the production of reactive oxygen species, and is controlled by SLAC1 and OST1


For correspondence (fax +37 2737 4900; e-mail hannes.kollist@ut.ee).


The air pollutant ozone can be used as a tool to unravel in planta processes induced by reactive oxygen species (ROS). Here, we have utilized ozone to study ROS-dependent stomatal signaling. We show that the ozone-triggered rapid transient decrease (RTD) in stomatal conductance coincided with a burst of ROS in guard cells. RTD was present in 11 different Arabidopsis ecotypes, suggesting that it is a genetically robust response. To study which signaling components or ion channels were involved in RTD, we tested 44 mutants deficient in various aspects of stomatal function. This revealed that the SLAC1 protein, essential for guard cell plasma membrane S-type anion channel function, and the protein kinase OST1 were required for the ROS-induced fast stomatal closure. We showed a physical interaction between OST1 and SLAC1, and provide evidence that SLAC1 is phosphorylated by OST1. Phosphoproteomic experiments indicated that OST1 phosphorylated multiple amino acids in the N terminus of SLAC1. Using TILLING we identified three new slac1 alleles where predicted phosphosites were mutated. The lack of RTD in two of them, slac1-7 (S120F) and slac1-8 (S146F), suggested that these serine residues were important for the activation of SLAC1. Mass-spectrometry analysis combined with site-directed mutagenesis and phosphorylation assays, however, showed that only S120 was a specific phosphorylation site for OST1. The absence of the RTD in the dominant-negative mutants abi1-1 and abi2-1 also suggested a regulatory role for the protein phosphatases ABI1 and ABI2 in the ROS-induced activation of the S-type anion channel.


Stomata, small pores on the aerial parts of plants, control CO2 influx for photosynthesis and water vapor loss. They also restrict the entry of ozone (O3) – a major air pollutant with an increasingly negative impact on crop yields, global carbon fixation (Hopkin, 2007) and climate change (Sitch et al., 2007). Ozone degrades immediately to reactive oxygen species (ROS) in the apoplastic space of plant cells, and has therefore been used as a tool to study the signaling role of the apoplastic ROS (Kangasjärvi et al., 2005; Wrzaczek et al., 2009). ROS are involved in the regulation of abscisic acid (ABA)- (Lee et al., 1999; Pei et al., 2000), ethylene- (Desikan et al., 2006), methyl jasmonate- (Munemasa et al., 2007) and salicylic acid-mediated (Mori et al., 2001) stomatal signaling. The rapid induction of ROS during CO2-induced stomatal closure has also been shown (Kolla et al., 2007). Collectively, ROS are central intermediate signaling components in plant guard cells, and it is likely that ozone is a useful tool for the study of ROS-dependent stomatal signaling.

Stomatal guard cells are among the most studied and best understood plant signaling systems, yet there remain considerable gaps in the understanding of the signaling that leads to stomatal movements in response to different stimuli. For example, the importance of guard cell anion channels as central regulators of stomatal closure was demonstrated 20 years ago (Keller et al., 1989; Schroeder and Hagiwara, 1989), but the protein essential for guard cell anion channel functioning, SLAC1, was identified only very recently (Negi et al., 2008; Vahisalu et al., 2008). SLAC1 is essential for stomatal closure in response to ABA, CO2, O3, light–dark transitions and humidity change, and by secondary messengers Ca2+, H2O2 and NO. However, the signaling cascades upstream of SLAC1, which require the capturing of very early and probably transient responses, are as yet unexplored.

Most experiments addressing the molecular details of guard cell signaling have been performed with epidermal peels or isolated guard cells. The preparation procedure(s) are likely to introduce unwanted effects, including an elevated production of ROS as a result of damage. Thus, using intact plants and minimum handling of the plant would help to define the function of ROS in stomatal regulation more clearly. The simple application of ozone to intact plants offers the possibility to control the concentration and duration of the exposure precisely. We have constructed a gas-exchange system where eight soil-grown Arabidopsis plants can be enclosed in individual flow-through exposure vessels, in a non-invasive manner (Kollist et al., 2007). Using this system we have shown that as little as 150 nl l−1 of O3 triggers a rapid transient decrease (RTD) in stomatal conductance (Kollist et al., 2007). The decrease was induced within a few minutes of O3 exposure, but the stomata reopened again despite the continuous presence of ozone. The recovery suggests that the closure was not a result of physical ozone damage, but instead reflects the biological action of ROS formed from ozone breakdown in the apoplast, transduced through a signaling cascade.

The lack of ozone-triggered RTD in the ABA-insensitive mutant abi2-1 (Kollist et al., 2007), carrying a dominant-negative mutation in the type-2C protein phosphatase ABI2, suggests a role for protein phosphorylation in O3/ROS-induced stomatal signaling. Murata et al. (2001) have also shown that H2O2-induced stomatal closure was impaired in the abi2-1 mutant. H2O2-induced stomatal closure was also disrupted in the recessive ABA-insensitive mutant gca2 (Pei et al., 2000). On the contrary, in abi1-1, another ABA-insensitive dominant-negative mutant of the protein phosphatase ABI1 (Murata et al., 2001), and in mutants of the protein kinase OPEN STOMATA 1 (OST1) (also referred to as SRK2E and Snf1-related protein kinase 2.6, SnRK2.6), a positive regulator of ABA-induced stomatal closure (Mustilli et al., 2002), H2O2-induced stomatal closure was not disrupted, suggesting a role for these proteins between ABA perception and ROS production.

We have shown that activation of S-type anion channels is required for ROS-induced stomatal signaling, as ozone-triggered RTD was absent in the S-type anion channel mutant slac1 (Vahisalu et al., 2008). However, other structural and signaling components involved in relaying the ROS signal from apoplast to stomatal movements have not been identified. During recent years, the molecular identities of many other guard cell transport proteins have been established (for a review see Pandey et al., 2007; Ward et al., 2008). Testing the characteristics of ozone-triggered RTD in plant lines carrying mutated versions of proteins involved in stomatal regulation could help to understand their role in ROS-dependent processes.

Here we have explored the ozone-triggered RTD further. We show the time and concentration dependence of the process, and provide evidence that stomatal closure coincides with the elevated burst of ROS in guard cells. This suggests that RTD is induced by the ROS triggered by the application of ozone. By analyzing RTD in several mutants carrying mutations in proteins shown to be involved in stomatal regulation, we show that OST1, ABI1, ABI2 and SLAC1 are regulators of the ROS-induced rapid stomatal closure. We demonstrate physical interaction between SLAC1 and OST1, and provide evidence that SLAC1 is phosphorylated by OST1.


The ozone-triggered RTD is dependent on exposure time and ozone concentration

Applying 250 nl l−1 of ozone induced a 40% decrease in stomatal conductance in wild-type (WT) Arabidopsis Col-0 plants within 5–10 min of exposure, followed by reopening to the pre-exposure level within the next 40 min (Figure 1a). The commonly observed sustained ozone-induced decrease in stomatal conductance (Ahlfors et al., 2004) was visible 90 min after ozone onset. The same stomatal behavior was seen in the 10 other ecotypes tested (Table S1). To further elucidate the relationship between the duration of ozone exposure and the decrease in stomatal conductance, Arabidopsis Col-0 plants were treated with 250 nl l−1 of ozone for 30, 70, 180, 360 and 720 s. Already a 30-s pulse of 250 nl l−1 ozone caused a clearly detectable decrease in stomatal conductance (Figure 1b). The decrease reached a maximum with 180 s of ozone, and longer exposures did not decrease the conductance any further. To address the effect of ozone concentration, we applied 3-min ozone pulses with concentrations ranging from 50 to 600 nl l−1 (Figure 1c). The response can be separated into three segments: ‘no response’, ‘response’ and ‘saturation’. In the ‘no response’ segment, essentially no decrease in stomatal conductance was observed. The threshold for the ‘response’ segment was 80 nl l−1 of ozone (95% confidence interval, shown by the dashed lines). After the threshold, the decrease in stomatal conductance increased approximately linearly, by 0.15% per additional nl l−1 of ozone, with 95% confidence intervals of ±0.04% per nl l−1. The ‘saturated’ segment, where the decrease in stomatal conductance reached its maximum, was obtained with concentrations higher than 434 nl l−1 of ozone.

Figure 1.

 Ozone-triggered rapid transient decrease (RTD) in stomatal conductance is dependent on the time and concentration of ozone applied.
(a) Time course of stomatal conductance (n = 3, ±SEM) of Col-0 plants after the onset of 250 nl l−1 ozone exposure, indicated by the gray bar.
(b) Ozone-triggered decrease in stomatal conductance after the application of 250 nl l−1 ozone for 30, 70, 180, 360 and 720 s (n = 3, ±SEM).
(c) Decrease in stomatal conductance in response to increasing ozone concentrations. Each point represents an independent experiment. The response is divided into three segments, separated by two vertical solid lines: ‘no response’, ‘response’ and ‘saturation’. The solid curve shows a bootstrap aggregate fit of data using 1000 bootstrap samples, each fitted with a logistic function. Dashed curves mark the 95% confidence intervals. The segment borders are identified by computing the maxima of the third derivative (i.e. the maximum of the change of acceleration).

To test the responsiveness of guard cells to ozone during the recovery period (20–60 min after the first ozone pulse), we applied four additional 3-min (250 nl l−1) pulses of ozone with 9-min intervals after the initial pulse (Figure 2a). These successive pulses had no effect on stomatal conductance, indicating that the induction of the RTD was blocked during the recovery period. The responsiveness to ozone reappeared when a second pulse of ozone was applied at 100 min, when stomatal conductance had fully recovered to the pre-exposure value (Figure 2b). However, applying nine additional 3-min ozone pulses with 12-min intervals throughout the recovery phase did not induce RTD (Figure 2c), suggesting the importance of a resting period for the guard cells to sense and respond to ozone again.

Figure 2.

 Guard cells are temporarily desensitized to ozone by a primary 3-min ozone pulse.
Time courses of stomatal conductance of Col-0 plants after the onset of 250 nl l−1 ozone shared between: (a) five 3-min pulses with 9-min intervals, (b) two 3-min pulses with a 90-min interval, (c) ten 3-min pulses with 12-min intervals. The experiments were repeated between three and five times, with similar results.

Ozone-triggered RTD involves a rapid burst of ROS

To address whether ozone induces an intrinsic ROS burst in guard cells, we analyzed early ROS production using fluorescence dye and confocal microscopy. The duration of ozone application was chosen according to the time course of stomatal conductance upon ozone exposure (Figure 1a): when RTD was induced (3 min); when stomata had reopened to the pre-exposure level (45 min); and when the sustained decrease in stomatal conductance was initiated (90 min). In addition to the Col-0 WT, we used various mutants: slac1-1, where ozone-triggered RTD is absent (Vahisalu et al., 2008); ost1-3 (also referred to as srk2e; Yoshida et al., 2002); and atrbohD and atrbohD/F, with mutations in NADPH oxidase catalytic subunit genes, previously shown to regulate ROS production in guard cells (Torres et al., 1998; Kwak et al., 2003). At the times indicated, plants were removed from the O3 treatment, epidermal peels were isolated, stained with 100 μm H2DCFDA and visualized by confocal microscopy 8 min after the removal of the plants from the treatment. Images were processed and ROS production in guard cells was quantified as fluorescence brightness.

A low background level of ROS was visible in the untreated control plants (Figure 3a, 0 min). In all plants studied, ozone exposure caused a bi-phasic ROS accumulation: elevated ROS signal was detected after 3 min of ozone exposure, after 45 min ROS levels were close to those of the untreated plants, and after 90 min of ozone exposure a second increase in ROS accumulation was evident. In all the mutants studied the first burst was lower than in the Col-0 plants. In the ost1-3 and the atrbohD mutants, the second increase in ROS accumulation (90 min) was higher than the first increase (Figure 3a; Tukey’s honestly significant difference test results for the data presented in Figure 3a are shown in Table S2). Untreated atrbohD and atrbohD/F mutants showed lower ROS levels than Col-0 WT, and in atrbohD/F, the second ROS peak (90 min) was lower than in all the other lines.

Figure 3.

 Ozone induces an intrinsic burst of reactive oxygen species (ROS) in Arabidopsis guard cells.
(a) Col-0, slac1-1, ost1-3, atrbohD and atrbohD/F Arabidopsis plants were exposed to 350 nl l−1 of ozone for 0, 3, 45, 90 min, and epidermal peels were isolated and stained with 100 μm H2DCFDA, and visualized by confocal microscopy. ROS production in guard cells, indicated in brightness units, was quantified by ImageQuaNT software (n = 12, ±SEM).
(b) Col-0 plants were exposed to to 350 nl l−1 of ozone for 0, 3, 12, 45 and 90 min, epidermal peels were isolated and treated as described in (a).

To localize the subcellular sites of ROS production during ozone-triggered RTD, we analyzed the confocal microscopy images in more detail (Figure 3b). At the earliest time point, 3 min from the beginning of ozone exposure, ROS accumulation was visible in the chloroplasts, from where ROS seemed to diffuse to the other parts of the cell. A more diffused, transient cytoplasmic ROS accumulation was visible at 12 min, which disappeared by 45 min, when ROS accumulation was again at low levels, and only found in chloroplasts. After 90 min of ozone exposure, a second peak of ROS accumulation, again spatially co-localized with chloroplasts, was visible.

Rapid ozone-triggered RTD is a specific process

A 3-min pulse of ozone triggered RTD in all 11 Arabidopsis ecotypes tested (Table S1), suggesting that RTD is a genetically robust response. Many mutants deficient in ABA signaling and ROS production in guard cells, in response to different stimuli, have been identified (Li et al., 2006). We tested several of them to elucidate the role of these proteins in the apoplastic ROS-induced RTD (Table S1).

The α and β subunits of the heterotrimeric G protein have been shown to be necessary for the ozone-induced oxidative burst in guard cells (Joo et al., 2005). The α subunit is also involved in regulating stomatal closure in response to ABA (Wang et al., 2001). Stomata of the Gα mutant gpa1-4, the Gβ mutant agb1-2 and the double mutant agb1/gpa1 responded to ozone like the WT (Table S1), suggesting that heterotrimeric G proteins were not involved in the signaling from apoplastic ROS to the activation of anion fluxes under our conditions.

The AtrbohD, AtrbohE and AtrbohF NADPH oxidase subunits have a role in guard cell ABA signal transduction (Kwak et al., 2003). However, atrbohD, atrbohE and atrbohF single mutants, and all double mutant combinations, responded to ozone pulse like the WT (Table S1), suggesting that the O3-derived apoplastic ROS formation can functionally mimic the apoplastic ROS production by NADPH oxidases in guard cells.

The plant stress hormones ethylene, jasmonic acid, salicylic acid and ABA are important regulators of ozone responses (Kangasjärvi et al., 2005), and have also been shown to be regulators of guard cell signaling (Lee et al., 1999; Pei et al., 2000; Mori et al., 2001; Desikan et al., 2006; Munemasa et al., 2007). We tested mutants deficient in biosynthesis and/or essential signal components for each of these hormones – none of them were required for RTD (Table S1). Exceptionally, some components of ABA signaling, but not ABA biosynthesis, were required for RTD (see below).

A mutant deficient in the protein kinase HT1, known to control stomatal movements in response to CO2 (Hashimoto et al., 2006), responded to ozone like the WT (Table S1). The mutant of the plasma membrane-localized ATP binding cassette transporter AtMRP5, shown to have impaired Ca2+ activation of guard cell anion channels (Suh et al., 2007), had a normal response to ozone (Table S1), as did the cpk3 and cpk6 (calcium-dependent protein kinase 3 and 6) mutants required for the activation of S-type anion currents by ABA and calcium (Mori et al., 2006) (Table S1). The ABA-insensitive mutant gca2 (Pei et al., 2000) also responded to ozone like the WT (Table S1), indicating that the activity of GCA2 is not required for the apoplastic ROS-induced stomatal closure. These results suggest that the apoplastic ROS-induced stomatal movements did not operate through the same set of regulatory components through which CO2, and Ca2+-dependent signaling act.

There were, however, mutants that did not display the O3-triggered RTD. We have previously shown that the ABA-insensitive protein phosphatase type-2C mutant abi2-1 completely lacked the ozone-triggered RTD (Kollist et al., 2007). Here (Table S1), we show that, in addition to abi2-1, the ozone-triggered RTD was also absent in abi1-1, ost1-1 and ost1-3, and in two K+-channel mutants the kinetics of RTD was altered.

Ozone-triggered RTD is modulated by K+ channels, and requires functional SLAC1, OST1, ABI1 and ABI2

Stomatal movements are facilitated by the activity of ion channels and transporters in the plasma membrane and vacuolar membrane of guard cells (Pandey et al., 2007). In addition to the S-type ion channel mutant slac1-1, where the ozone-triggered RTD was absent (Figure 4a), the patterns of RTD were also different in two potassium channel mutants: gork-1, where the guard cell plasma membrane K+ outward rectifying channel GORK activity is fully suppressed (Hosy et al., 2003), and kincless, where the inward rectifying K+ current is abolished (Lebaudy et al., 2008) (Figure 4b). In the Ws-2 WT, ozone triggered an RTD of 30% within 4 min of exposure. The response was strongly delayed in gork-1, where the RTD was only 5% within 4 min, and a maximal decrease of 23% was only achieved 16 min after the beginning of the exposure (Figure 4b). In kincless, ozone caused a 22% RTD within 4 min, but reopening was almost completely absent; after 40 min, conductance had fully recovered in the WT, whereas in kincless no recovery was seen (Figure 4b). These results suggest that, in addition to initial anion currents (SLAC1), subsequent GORK-mediated K+ flux is required for the rapid decrease in stomatal conductance. During the recovery period, additional ozone pulses had no effect (Figure 2a). This implies that the activity of the inward-rectifying potassium channel was not directly affected by the apoplastic ROS induced by O3.

Figure 4.

 Ozone-triggered rapid transient decrease (RTD) in stomatal conductance is absent in slac1-1, and is altered in gork-1 and kincless.
Time courses of stomatal conductance (n = 3, ±SEM) after the onset of a 3-min 250 nl l−1 ozone pulse, indicated by the gray bar. (a) Col-0 and slac1-1. (b) Ws, kincless and gork-1.

In order to elucidate the role of protein phosphorylation in the ozone-triggered RTD upstream of SLAC1, we analyzed RTD in abi1-1, abi2-1 and ost1-1 in more detail (Figure 5). Whereas a 3-min O3 pulse triggered RTD in both Col-0 and Ler WTs, stomata of all three protein phosphorylation mutants were insensitive to the O3 pulse, suggesting that the protein kinase OST1 and the phosphatases ABI1/ABI2 are required for the SLAC1-dependent RTD.

Figure 5.

 Ozone-triggered rapid transient decrease (RTD) in stomatal conductance is absent in abi1-1, abi2-1 and ost1-1.
Time courses of stomatal conductance (n = 3, ±SEM) in Ler, abi1-1, abi2-1 and ost1-1, after the onset of a 3-min 250 nl l−1 ozone pulse, indicated by the gray bar.

Protein interaction, genetic and phosphoproteomic analysis suggest that SLAC1 is regulated by OST1

The absence of ozone-triggered RTD in ost1 mutants immediately suggested a model where SLAC1 is regulated by OST1. Therefore, we first addressed whether SLAC1 and OST1 interact in a split-ubiquitin membrane yeast two-hybrid system (Johnsson and Varshavsky, 1994). SLAC1 was fused to the C terminal of ubiquitin (Cub) and the LexA-VP16 transcription factor as bait, and the OST1 was fused to the N terminal of ubiquitin with an I3G mutation (NubG) as prey (Figure S1a). Growth on selective media, and strong activation of the lacZ reporter gene in yeast co-transformed with the bait and prey (Figure S1b), indicated an interaction between SLAC1 and OST1. To confirm this result in planta, we used a bimolecular fluorescence complementation (BiFC) assay. Co-infiltration of 35S:SLAC1-YFPC and 35S:OST1-YFPN in Nicotiana benthamina leaves yielded YFP signals in the plasma membrane and nucleus, with the latter possibly being a result of high expression levels from the 35S promoter (Figure 6a). No fluorescence signal was observed when 35S:YFPC and 35S:YFPN, 35S:SLAC1-YFPC and 35S:YFPN, or 35S:YFPC and 35S:OST1-YFPN were co-infiltrated. The data supports the split-ubiquitin yeast two-hybrid analysis, and provides evidence for a physical interaction between SLAC1 and OST1.

Figure 6.

 Interaction, phosphoproteomic and mutant data suggest that SLAC1 is regulated by OST1 kinase.
(a) Bimolecular fluorescence complementation assays were performed with Nicotiana benthamiana leaves infiltrated with 35:SLAC1-YFPC and 35S:OST1-YFPN. YFP signal (YFP), chlorophyll autofluorescence (Chl) and overlay of YFP and chlorophyll autofluorescence (Merge) are shown. A similar result was observed in ∼20 individual cells in three independent repeats.
(b) Schematic diagram of the SLAC1 protein, indicating the positions of the mutations in slac1-6, slac1-7 and slac1-8.
(c) Time courses of stomatal conductance (n = 5, ±SEM) in Col-er and alleles of slac1 after the onset of a 3-min 250 nl l−1 ozone pulse (indicated by the gray bar).
(d) Recombinant protein kinase OST1 has autophosphorylation activity (left panel), phosphorylates SLAC11–186 (middle panel) and a generic substrate histone III (right panel).
(e) Different SLAC1–186 fragments phosphorylated by OST1 and separated on SDS-PAGE gel supplemented with Mn2+-Phos-tag. The bands visible represent the degree of phosphorylation by OST1.
(f) SLAC1–186 phosphorylated by OST1 was analyzed with nanoLC-MS/MS system. The upper table shows the sequence of identified peptides phosphorylated by OST1. The lower table shows the protein sequence of SLAC11–186. Phosphorylated serines are indicated in red. The results were confirmed through several replications.

The prediction programs NetPhos and Scansite (http://www.cbs.dtu.dk/services/NetPhos and http://scansite.mit.edu) suggested several putative phosphorylation sites in the N-terminal tail of SLAC1. To analyze the importance of these sites for the regulation of SLAC1 in vivo, we used TILLING (Till et al., 2003), and identified mutants where three of the predicted phosphorylation sites were mutated: S38F (slac1-6), S120F (slac1-7) and S146F (slac1-8). Two of these mutants, slac1-7 and slac1-8 were deficient in RTD, similar to slac1-1, whereas slac1-6 responded to ozone like the WT (Figure 6c).

Finally, we performed in vitro phosphoproteomic experiments to examine whether SLAC1 is a substrate of OST1. As shown earlier (Belin et al., 2006), recombinant 6xHis N-terminal-tagged OST1 protein produced in Escherichia coli displayed autophosphorylating activity, and effectively phosphorylated a generic substrate, such as histone III (Figure 6d). To address phosphorylation of SLAC1 by OST1, we used a SLAC1 N-terminal fragment (SLAC1–186) as a substrate. The phosphorylation of SLAC11–186 by OST1 was rapid and efficient (Figure 6d). Prediction programs and experiments with slac1-7 and slac1-8 suggested the presence of several functionally important phosphorylation sites in SLAC1. To address this, we separated the OST1 phosphorylated SLAC1–186 on SDS-PAGE gel, supplemented with Mn2+-Phos-tag, which incorporates a phosphate-binding compound that retards the mobility of phosphorylated proteins in proportion to their degree of phosphorylation (Kinoshita et al., 2006). Multiple differentially migrating bands in Phos-tag gels indicated that several amino acids in SLAC1–186 were phosphorylated by OST1 (Figure 6e). To address whether S120 and S146 of SLAC1 are targets of OST1, we used SLAC1–186 fragments where S120 and S146 were mutated to alanine. Clearly, the different multiphosphorylation pattern of S120A, compared with the WT SLAC1–186, indicates that S120 was phosphorylated by OST1 (Figure 6e). On the contrary, S146 was not phosphorylated by OST1, as the SLAC11–186 fragment containing the mutation S146A, or the combined mutations of S120A and S146A, revealed similar phosphorylation patterns as the WT SLAC1–186 and SLAC1–186 containing S120A, respectively. Additionally, we analyzed the SLAC11–186 fragment phosphorylated by OST1 with mass spectrometry. This indicated that Ser59, Ser86, Ser113 and Ser120, but not Ser146, were phosphorylated by OST1 (Figure 6f and Figure S2).

Taken together, these data indicate that regulation of SLAC1 by OST1 may involve the phosphorylation of multiple amino acids, and suggests that Ser120 of SLAC1 is one of the functionally important phosphosites.


Reactive oxygen species have a demonstrated role in guard cell signaling in response to various external and internal factors (Pei et al., 2000; Kolla et al., 2007). Apoplastic ROS, formed as a result of the activity of the NADPH oxidases (Kwak et al., 2003), are essential components in guard cell ABA signaling. Ozone is known to degrade to various ROS in the apoplast, thus it is likely that the ozone-triggered RTD addressed in this study is a result of the action of the apoplastic ROS. This ROS from ozone degradation would act in a similar manner as the ROS produced by NADPH oxidase activity in ABA- or methyl jasmonate-induced stomatal closure (Kwak et al., 2003; Munemasa et al., 2007). Hence, ozone can be used as a tool to simplify the very complex regulatory network in guard cells, and enables the study of the role of ROS alone. The fast kinetics of RTD (Figure 1b) implies that either the ROS are perceived directly in the apoplast, followed by a rapid signal transmission to guard cells, or that the ROS formed in the apoplast translocate (most likely after dismutation to H2O2, as inline image is impermeable through a biological membrane) to the inside of guard cells, where they immediately elicit the response.

Ozone triggered a biphasic ROS accumulation in guard cells, where chloroplasts were the major source for ROS formation (Figure 3). Previously, it has been shown that ozone-induced ROS production was initiated from guard cell chloroplasts, followed by ROS production in guard cell membranes, which required NADPH oxidases encoded by the AtrbohD and AtrbohF genes (Joo et al., 2005). The time point studied with atrbohD and atrbohF by Joo et al. (2005) was 1 h. In our experiments, the first phase of O3-induced ROS accumulation (detected after 3 min) was significantly reduced, but not abolished, in atrbohD and atrbohD/F mutants (Figure 3a). This suggests the presence of several sources for initial ROS production, for example, cell wall peroxidases, in addition to the NADPH oxidases. Recently, it has been shown that OST1 can phosphorylate AtrbohF, and possibly regulates its activity (Sirichandra et al., 2009). The ost1-3 mutant had lower initial ROS production (Figure 3a), suggesting that regulation of Atrboh-mediated ROS production by OST1 could be functionally relevant during initial ozone responses. Interestingly, the second ROS peak was lower in the atrbohD/F double mutant (Figure 3a), which suggests that the membrane-bound NADPH oxidases have an influence on the second peak of ROS production. It is noteworthy that the timing of the first and second peak of ROS accumulation (Figure 3) coincided with the fast and slow decrease in stomatal conductance triggered by ozone (Figure 1a). The decline in ROS production to control levels at 45 min, despite the plant being continuously exposed to ozone, favors a model for enzymatic control of ROS production. However, whether this temporal coincidence also has a mechanistic grounding needs to be studied further, as well as the role of chloroplastic ROS in guard cell signaling.

After the perception of the apoplastic ROS, the signal is rapidly transduced to the guard cell anion channel SLAC1, as RTD was absent in the slac1 mutant. However, RTD is not a result of the activity of the plasma membrane anion channel only: in the K+ efflux channel mutant gork-1, the ozone response was delayed, suggesting that after the activation of the anion fluxes, K+ fluxes through GORK were also required for the rapid decrease in stomatal conductance (Figure 4b). Thus, apparently the signal from apoplastic ozone/ROS is first passed to SLAC1, and the subsequent activation of anion fluxes and membrane depolarization are required for GORK activation. The lack of recovery in stomatal conductance in the kincless mutant indicates a central role for K+ inward-rectifying channels during the recovery period (Figure 4b).

An intriguing feature of the ozone-triggered RTD is that during the recovery period further ozone pulses had no effect (Figure 2b). This suggests that the primary ozone pulse changes the status of the voltage-dependent channel assembly in a manner that temporarily desensitizes stomata to further ozone pulses. Blocking and desensitization have been shown to occur for several ion channels under different treatments (Roelfsema and Prins, 1997; Raschke et al., 2003). Additionally, blue light treatment rapidly induces Ca2+ transients (Baum et al., 1999), and a 30–120-min recovery period is required to elicit the response again. Similarly, both H2O2 (Price et al., 1994) and ozone (Clayton et al., 1999; Evans et al., 2005) induced rapid transient increases in cytosolic Ca2+, and a recovery period of a few hours was required before the response could be elicited again. One of the possible causes of desensitization may be channel inactivation (Hedrich et al., 1990; Pei et al., 1998), or if Ca2+ is part of the RTD, it might need to be returned to its original resting state concentration and/or localization.

Several different cellular regulatory and signaling processes in stomatal guard cells appear to converge at the activation of anion fluxes, where SLAC1 is a central component. ROS also play a role in several of these regulatory cascades. The absence or presence of RTD in various guard cell signaling and ion channel mutants (Table S1) allowed us to dissect components that are required for apoplastic ROS responses in guard cells. Collectively, our results suggest that the signal pathway from apoplastic ROS to guard cell anion channel activation is at least partly independent from previously described CO2 and ABA signaling pathways.

The use of protein kinase and phosphatase inhibitors has previously shown that protein phosphorylation is involved in the activation of anion channels (Schmidt et al., 1995; Grabov et al., 1997; MacRobbie, 1998; Pei et al., 2000). Some protein kinases involved in these responses have been identified. The protein kinase HT1, a central component in the regulation of stomatal movements by CO2, was not required for the ROS activation of SLAC1. The activity of the Ca2+-dependent protein kinases CPK3 and CPK6, which have been shown to be involved in the Ca2+-dependent ABA activation of S-type anion currents (Mori et al., 2006), showed a normal RTD response. Thus, these protein kinases, and other proteins, such as GCA2, previously shown to be involved in CO2- or ABA- and Ca2+-dependent stomatal regulation, are not components required for the apoplastic ROS-induced regulation of stomatal movement. The OXI1 kinase, necessary for some ROS signaling processes (Rentel et al., 2004), was not required for RTD either (Table S1). In addition, ABA itself is not part of the signal cascade, as the ABA-deficient aba1-3 mutant showed WT responses.

However, we have previously shown that in addition to slac1 (Vahisalu et al., 2008), the ABA-insensitive protein phosphatase type 2C mutant, abi2-1, completely lacked the ozone-induced decrease in stomatal conductance (Kollist et al., 2007). The ozone-triggered RTD was also missing in a second dominant-negative mutant of the type-2C protein phosphatases, ABI1, and in two mutant alleles of the OST1, demonstrating that these proteins are essential in the apoplastic ROS-induced RTD (Figure 5). It has been shown recently that ABI1, ABI2 and OST1 interact physically both in yeast two-hybrid and in planta assays, and that these and other phosphatases act as negative regulators of OST1 via PYR/PYL/RCAR proteins (Fujii et al., 2009; Umezawa et al., 2009; Vlad et al., 2009). Similarly, our genetic data suggested that ABI1 and ABI2 could be regulators of OST1 in ROS-induced stomatal closure through dephosphorylation (Figures 5 and 7). Using the split-ubiquitin yeast two-hybrid and BiFC assays, we showed a physical interaction between OST1 and SLAC1 (Figure S1b and 6a). The physical interaction between SLAC1 and OST1, and the requirement for OST1-dependent phosphorylation for the activation of SLAC1, was also very recently demonstrated by two other groups (Geiger et al., 2009; Lee et al., 2009).

Figure 7.

 Schematic model of events during ozone-triggered rapid transient decrease (RTD) in stomatal conductance.
Ozone induces reactive oxygen species (ROS) production in the guard cells (1). This leads to the activation of OST1 (2). The lack of RTD in abi1-1 and abi2-1 suggests that the protein phosphatases ABI1 and ABI2 regulate OST1 activity (Umezawa et al., 2009). Our data suggest that several amino acid residues of the N-terminal tail of SLAC1 are phosphorylated by OST1, and highlight the functional importance of S120 (3). Phosphorylation leads to the activation of S-type anion channels and anion efflux (A out) from guard cells (4), which in turn causes plasma membrane depolarization, and activation of outward-rectifying K+ channel GORK and K+ efflux from guard cells (5). The overall efflux of anions and K+ contributes to the loss of guard cell turgor, leading to stomatal closure. For reopening, the activity of K+ uptake channels are required (6).

Our phosphoproteomic experiments proved that OST1 was able to phosphorylate multiple amino acids of SLAC1, including Ser120, but not Ser146 (Figure 6d–f), further supporting that OST1 is responsible for SLAC1 activation. This also suggests that Ser146 could be the target for a different protein kinase. An alternative explanation for the absence of RTD in slac1-8 (Figure 6c) could be a conformational change of SLAC1 protein, as in the slac1-8 mutant, Ser146 is substituted with phenylalanine, which is a hydrophobic and considerably larger amino acid. SLAC1 has been shown to be involved in the regulation of stomatal closure in response to many factors, such as CO2, darkness, humidity, ABA and Ca2+ (Vahisalu et al., 2008). Thus, it would be of great interest to study the stomatal responses of slac1-7 and slac1-8 to other stimuli, in order to address whether different phosphorylation patterns of SLAC1 exist in response to different stimuli.

Collectively, our data suggest a model for RTD (Figure 7) where OST1 activates SLAC1 via phosphorylation, and suggests the importance of S120 for this regulation. ABI1 and ABI2 keep OST1 inactive by dephosphorylation (Umezawa et al., 2009; Vlad et al., 2009). Possibly, the protein phosphatase activity of ABI1 and ABI2 could also be directed towards SLAC1: this should be addressed in further studies. Recently, OST1 has been shown to phosphorylate and reduce the activity of the inward K+ channel KAT1 (Sato et al., 2009). Thus, OST1 would activate SLAC1 and at the same time inactivate KAT1, leading to faster stomatal closure.

Using ozone we show that the last steps of the ROS-induced signaling cascade, leading to the activation of SLAC1 and the induction of stomatal closure, could be as simple as the recently described minimal signaling pathway for regulating ABA-induced gene expression (Fujii et al., 2009). Furthermore, the two pathways share several common components: ABI1, ABI2 and OST1. Our next challenge is to find out how the cells perceive ROS and translate this into activation of OST1.

Experimental procedures

Plant material and growth conditions

For the whole-plant gas-exchange experiments, 24–26-day-old plants were used. Plants were grown as described previously (Kollist et al., 2007). The ost1-1 mutant is in the Ler background (Mustilli et al., 2002). The ost1-3 is a T-DNA knock-out of OST1 kinase in the Col-0 background (originally referred to as srk2e; Yoshida et al., 2002). For clarity, we refer to srk2e as ost1-3 throughout this report. The source and identity of other mutants used in the study are given in Table S1.

Whole-plant stomatal conductance measurements and fluorescence microscopy

The Arabidopsis whole-rosette gas-exchange measurement details were described previously (Kollist et al., 2007). Prior to ozone exposure, plants were acclimated in the measuring cuvettes for at least 1 h. Plants were exposed to 350 nl l−1 ozone for 3, 12, 45 or 90 min. Abaxial epidermal peels were isolated and loaded with 100 μm H2DCFDA in 10 mm Tris–HCl, pH 7.2, for 5 min in darkness, and were washed with 10 mm Tris–HCl, pH 7.2. Clean-air control peels were isolated and loaded after acclimation in gas-exchange cuvettes.

ROS production was visualized by Nikon TE2000-U C1 confocal microscope (Nikon, http://www.nikon.com) using excitation at 488 nm and emission at 530 nm. Images were processed using Nikon EZ-C1 FreeViewer software (gold version 3.30; Nikon). Brightness values of individual guard cell pairs were obtained after correcting for the brightness of epidermal cells with ImageQuaNT v4.2a (Molecular Dynamics, now part of GE Healthcare, http://www.gelifesciences.com).

Isolation of TILLING lines

New ethyl methanesulphonate mutants slac1-6 (S38F), slac1-7 (S120F) and slac1-8 (S146F) were identified through TILLING (Till et al., 2003; http://tilling.fhcrc.org). Details are described in Appendix S1.

Split-ubiquitin membrane yeast two-hybrid assay

The split-ubiquitin yeast two-hybrid assay was conducted using the DUALmembrane kit 3 (Dualsystems Biotech, http://www.dualsystems.com). Details are described in Appendix S1.

BiFC interaction experiments

The cDNA of SLAC1 and OST1 was cloned into the pSPYNE and pSPYCE vectors. BiFC experiments were performed using transient transfection of N. benthamiana leaves with Agrobacterium tumefaciens, as described by Voinnet et al. (2003), images were acquired 48–72 h after transfection by confocal microscopy (also see Appendix S1).

Protein expression and purification

OST1 and SLAC1 N-terminal fragments encoding amino acids 1–186 (SLAC11–186) cDNAs were cloned into the pQE-30 UA (Qiagen, http://www.qiagen.com) and pET28a (Novagen, now part of Merck, http://www.merck4biosciences.com) vectors, respectively. 6xHIS-OST1 was expressed using the XL-1 blue E. coli strain (Stratagene, http://www.stratagene.com). 6xHIS-SLAC11–186 variants were expressed in Rosetta (DE3) pLysS cells (Novagen). Recombinant proteins were purified using a Chelating Sepharose™ Fast Flow (Amersham, now part of GE Healthcare, http://www.gelifesciences.com) column chelated with CoCl2 (for details, see Appendix S1).

In vitro kinase assays and mass spectrometry

Proteins were separated by SDS-PAGE using a 12% (w/v) acrylamide gel or 10% (w/v) acrylamide gel supplemented with Phos-Tag (Kinoshita et al., 2006). Gels were stained with Coomassie brilliant blue R-250 (Sigma-Aldrich, http://www.sigmaaldrich.com), and incorporation of 32P to the proteins was detected and visualized by autoradiography (for further details see Appendix S1). NanoLC-MS/MS analysis for the mapping of OST1 phosphorylation sites in SLAC11–186 was carried out by using LTQ-Orbitrap (Thermo Fisher Scientific, http://www.thermofisher.com), equipped with a nanospray source (Proxeon, http://www.proxeon.com) and a 1200 Series nano-LC system (Agilent Technologies, http://www.agilent.com) (for details, see Appendix S1).


We acknowledge several labs (listed in Table S1) for sharing their mutants. Work in the labs of HK and ML was funded by the Estonian Science Foundation (grants 7763, 6766, 7869, 7361 and EMP24) and Estonian Ministry of Education and Research (theme SF0180071s07). Work in the JK’s lab was supported by the Academy of Finland Centre of Excellence program (2006–2011), MB was supported by an Academy of Finland Post-Doctoral grant (decision# 108760), and TV was supported by the Finnish Graduate School in Plant Biology.