Calcium-sensing receptor regulates stomatal closure through hydrogen peroxide and nitric oxide in response to extracellular calcium in Arabidopsis

Abstract

The Arabidopsis calcium-sensing receptor CAS is a crucial regulator of extracellular calcium-induced stomatal closure. Free cytosolic Ca²⁺ (Ca²⁺_i) increases in response to a high extracellular calcium (Ca²⁺_o) level through a CAS signalling pathway and finally leads to stomatal closure. Multidisciplinary approaches including histochemical, pharmacological, fluorescent, electrochemical, and molecular biological methods were used to discuss the relationship of hydrogen peroxide (H₂O₂) and nitric oxide (NO) signalling in the CAS signalling pathway in guard cells in response to Ca²⁺_o. Here it is shown that Ca²⁺_o could induce H₂O₂ and NO production from guard cells but only H₂O₂ from chloroplasts, leading to stomatal closure. In addition, the CASas mutant, the atrbohD/F double mutant, and the Atnoa1 mutant were all insensitive to Ca²⁺_o-stimulated stomatal closure, as well as H₂O₂ and NO elevation in the case of CASas. Furthermore, it was found that the antioxidant system might function as a mediator in Ca²⁺_o and H₂O₂ signalling in guard cells. The results suggest a hypothetical model whereby Ca²⁺_o induces H₂O₂ and NO accumulation in guard cells through the CAS signalling pathway, which further triggers Ca²⁺_i transients and finally stomatal closure. The possible cross-talk of Ca²⁺_o and abscisic acid signalling as well as the antioxidant system are discussed.

Introduction

Extracellular calcium (Ca²⁺_o) has long been known to promote free cytosolic Ca²⁺ (Ca²⁺_i) increase and stomatal closure (Schwartz, 1985; MacRobbie, 1992; McAinsh et al., 1995). The calcium-sensing receptor (CAS) was then identified and proven to be involved in this Ca²⁺_o-induced Ca²⁺_i increase (CICI) and stomatal closure (Han et al., 2003; Tang et al., 2007). Furthermore, CAS was reported to localize in the chloroplast thylakoid membrane (Nomura et al., 2008). However, how the guard cells convert the Ca²⁺_o signal through CAS and finally lead to CICI was not well understood.

Hydrogen peroxide (H₂O₂) was shown to be a second messenger in response to biotic and abiotic perturbations (Fukao and Bailey-Serres, 2004; Laloi et al., 2004). H₂O₂ also functions as a signalling molecule in abscisic acid (ABA)-induced stomatal movements (Pei et al., 2000). The AtrbohD and AtrbohF (Arabidopsis respiratory burst oxidase homologues D and F) NADPH oxidases double mutant atrbohD/F abolishes the ABA-driven Ca²⁺_i increase and stomatal closure (Kwak et al., 2003). Meanwhile, H₂O₂ has been found to stimulate Ca²⁺_i transients in regulation of stomatal movement (Allen et al., 2000; Pei et al., 2000). Extracellular calmodulin (ExtCaM), a Ca²⁺_o-activated protein, induces H₂O₂ generation and the increased H₂O₂ further elevates stomatal Ca²⁺_i concentration (Chen et al., 2004), suggesting that Ca²⁺_o signalling, H₂O₂ production, and Ca²⁺_i transients may be orchestrated in guard cells.

Nitric oxide (NO), another key intermediate in ABA and multiple stress responses (Garcia-Mata and Lamattina, 2001, 2002; Neill et al., 2002a), operates as a downstream player of H₂O₂ signalling in guard cells (Lum et al., 2002; Bright et al., 2006). Interestingly, the Ca²⁺_o-driven NO accumulation in guard cells was reported in recent experiments in Arabidopsis (Li et al., 2009), suggesting the possibility of Ca²⁺_o-driven NO accumulation through H₂O₂ generation. Like H₂O₂, an NO-induced Ca²⁺_i increase and stomatal closure have also been found in guard cells of Arabidopsis and Vicia faba (Neill et al., 2002a; Garcia-Mata et al., 2003). Notably, NO was shown to contribute to Ca²⁺_i transients in ABA-induced stomatal closure (Desikan et al., 2002). However, the underlying mechanisms by which NO contributes to the Ca²⁺_o-induced Ca²⁺_i increase and stomatal closure are not clear.

Chloroplasts are a major source of H₂O₂ and NO production in plant cells (Asada and Takahashi, 1987; Jasid et al., 2006) and also a sensor of environmental stresses through chloroplast redox signalling (Pfannschmidt, 2003). The application of exogenous ABA resulted in H₂O₂ generation and stomatal closure, which occurs in chloroplasts earlier than within the other regions of guard cells (Zhang et al., 2001). A recent study revealed that chloroplasts play a critical role in CAS-mediated CICI and subsequent stomatal closure in Arabidopsis (Nomura et al., 2008). These results suggest that stomatal movements responding to environmental signals are probably regulated by H₂O₂ from the chloroplasts. Ca²⁺_o-induced stomatal closure may involve cross-talk between Ca²⁺_o and chloroplast-localized CAS as well as H₂O₂ in chloroplasts.

Besides what is known about guard cell signalling pathways, other findings also suggested the correlation among Ca²⁺_o, H₂O₂, NO, and chloroplasts. Ca²⁺_o can increase the activity of NADPH oxidase (NOX) and trigger H₂O₂ production in maize leaves (Sagi and Fluhr, 2001; Hu et al., 2007; Potocký et al., 2007). Mitogen-activated protein kinase 6 (MPK6) and the prohibitin gene PHB3 were recently identified to function in H₂O₂-mediated NO synthesis during lateral root formation and in abiotic stress responses (Wang et al., 2010; Wang et al., 2010). Consistently, NO synthesis was shown to be up-regulated by H₂O₂ in other plant species (Lum et al., 2002; Li et al., 2009). In addition, NO also promoted Ca²⁺_i transients in plant cells in response to biotic and abiotic stresses (Gould et al., 2003; Lamotte et al., 2004, 2006; Vandelle et al., 2006). Similarly to H₂O₂, using NO-sensitive fluorescent probes, NO synthesis was also evident first in chloroplasts of the mesophyll and across all epidermal cell types including guard cells in response to environmental stimuli (Foissner et al., 2000; Gould et al., 2003; Arnaud et al., 2006).

In this report, the Apollo 4000 system with H₂O₂- and NO-selective electrochemical sensors was used to detect H₂O₂ and NO production in epidermis or chloroplasts. Different tools were also utilized to characterize functionally the role of H₂O₂ and NO as well as the contribution of the CAS signalling pathway in Ca²⁺_o-induced Ca²⁺_i increase and stomatal closure, including CAS antisense lines (CASas) and Arabidopsis mutants defective in H₂O₂ or NO synthesis (e.g. the H₂O₂ synthetic enzymatic mutant atrbohD/F and a mutant indirectly impaired in NO synthesis Atnoa1), a histochemical technique, and H₂O₂-and NO-specific fluorescent dyes. The transcriptional activities of Arabidopsis cytosolic antioxidant enzymes such as copper/zinc superoxide dismutase 1 (CSD1), ascorbate peroxidase 1 (APX1), and glutathione-disulphide reductase (ATGR1) in leaves were also investigated for the involvement of the antioxidant system in Ca²⁺_o-induced H₂O₂ generation. A functional relationship among CAS, H₂O₂, and NO as well as the antioxidant system was established in Ca²⁺_o-dependent guard cell signalling.

Materials and methods

Plant materials and growth conditions

Arabidopsis plants of the wild type and various mutants were grown in mixture matrix (turves:vermiculite=1:1) with a 16 h light and 8 h dark cycle under a photon flux rate of 200 μM m⁻² s⁻¹ at 22 °C, 70% relative humidity. Fully expanded Arabidopsis leaves of ∼5 weeks old were harvested for immediate use. Seeds of CASas, atrbohD/F, and Atnoa1 mutants (background Col-0) were all obtained from Duke University. These mutants were further confirmed by reverse transcription-PCR analysis with specific primers on total RNA extracted from leaves (Supplementary Fig. S1A, C available at JXB online). Primer sequences used for PCR are shown in Supplementary Table S1. CAS antisense lines were identified by western blot analysis using anti-CAS antibody as described (Han et al., 2003) (Supplementary Fig. S1B).

Stomatal bioassay

Stomatal assays were performed essentially as described by Desikan et al. (2002). Abaxial epidermal strips from similar rosette leaves were floated in 10 mM MES buffer (pH 6.15) containing 50 mM KCl and 50 μM CaCl₂ for 2 h under light conditions to open the stomata before the addition of various compounds. Following this, 2 mM CaCl₂, 10 μM H₂O₂, or 60 μM sodium nitroprusside (SNP) was added to the buffer and incubated for another 2 h to assay stomatal closure. To study the effect of catalase (CAT), 2-(4-carboxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl 3-oxide potassium salt (cPTIO), or neomycin on stomatal closure, epidermal strips were transferred to and incubated in MES buffer containing 2 mM CaCl₂ plus 100 U ml⁻¹ CAT, 100 μM cPTIO, or 100 μM neomycin for 2 h.

For time-resolved measurements of Ca²⁺_o-, H₂O₂-, or SNP-induced stomatal closure, abaxial epidermal strips with open stomata were incubated in various buffers for different times as indicated in the figure legends. Stomatal apertures were determined as the ratio of width to length using image analysis computer software (SigmaScan Pro5) and were presented as the percentage with respect to untreated control or zero time as the standard.

Fluorescent imaging by microscopy

The variations of H₂O₂ and NO in guard cells were examined by loading 2′,7′-dichlorofluorescein diacetate (H₂DCFDA) and 3-amino,4-aminomethyl-2′,7′-difluorofluorescein diacetate (DAF-FM DA) as described (Suhita et al., 2004; Gonugunta et al., 2008). Abaxial epidermal strips with open stomata were transferred to MES buffer with 50 μM H₂DCFDA or 10 μM DAF-FM DA for 15 min in darkness at room temperature and then rinsed with MES buffer three times before various treatments. Fluorescence was observed under a Motic AE31 fluorescence microscope (Speed Fair Co., Ltd, Hong Kong, excitation filter, 488 nm and emission, 535 nm) with a digital video camera (Motic MHG-100B, Speed Fair Co., Ltd, Hong Kong). Images were acquired from fluorescence between different samples, genotypes, and during time-courses at identical illumination intensity, camera gain, and exposure time in this experiment. The fluorescent pixel intensities of subtracting the basal signal from guard cells were analysed using MetaMorph 7.5 software. H₂DCF fluorescence and chlorophyll autofluorescence in guard cells were further visualized using a laser scanning confocal microscope (LEICA TCS SP2) at an excitation wavelength of 488 nm. H₂DCF fluorescence was detected from 495 nm to 535 nm, whereas chlorophyll autofluorescence was detected between 630 nm and 730 nm.

Subcellular localization of H₂O₂ in guard cells

The histochemical localization of H₂O₂ based on generation of electron-dense cerium perhydroxide precipitates was performed as described (Bestwick et al., 1997; Pellinen et al., 1999). Tissue fragments (2×5 mm²) from leaves incubated in MES buffer with or without 2 mM CaCl₂ or 10 μM ABA were excised and kept in 50 mM MOPS buffer containing 5 mM CeCl₃ at pH 7.2 under vacuum for 20 min. After being double fixed with 2.5% glutaraldehyde and 1.0% OsO₄, the CeCl₃-treated sections were dehydrated in an ascending ethanol series and embedded with the abaxial epidermis clinging to the surface of the embedding block. Blocks were progressively dehydrated in a graded acetone series and polymerized. Blocks of abaxial epidermis were sliced into 70–90 nm sections on a ultramicrotome, and mounted on uncoated copper grids. Sections of the guard cells were observed using transmission electron microscopy (TEM; JEM-2100HC, JEOL, Japan).

Preparations of leaf epidermis and chloroplasts

Arabidopsis leaf epidermis was extracted as described by Pandey et al. (2002). The epidermis peels, which were pale green and ∼1 mm² in size, were ready for H₂O₂ or NO monitoring using a four-channel free radical Apollo 4000 analyser (World Precision Instruments, Sarasota, FL, USA). The intactness of epidermal cells in the epidermal fragments was evaluated using Evans blue dye as described (Mergemann and Sauter, 2000).

Arabidopsis chloroplasts were isolated using the method of Aronsson and Jarvis (2002). The integrity of the isolated chloroplasts was estimated to be >85% (Supplementary Fig. S2 at JXB online). The chloroplast pellets were finally resuspended in HMS buffer containing 50 mM HEPES, 3 mM MgSO₄, 0.3 M sorbitol, pH 7.2. The chloroplast counts were determined microscopically using a haemocytometer.

H₂O₂ and NO measurement by the Apollo 4000 system

The H₂O₂- and NO-selective electrochemical sensors ISO-HPO and ISO-NOP (WPI, Sarasota, FL, USA), widely used for animals, plants, microalga, and subcellular organelles (Bouchard and Yamasaki, 2008; de Oliveira et al., 2008; D’Agostino et al., 2009; Pandolfi et al., 2010), were respectively connected to the Apollo 4000 system (World Precision Instruments) and adapted for real-time detection of H₂O₂ and NO from Arabidopsis leaf epidermis or chloroplast. Both sensors consist of a combination of an internal H₂O₂- or NO-sensing pair of working and counter (reference) electrodes. Each sensor fits inside a replaceable stainless steel membrane sleeve filled with an electrolyte solution and is separated from the outside environment by an H₂O₂- or NO-selective membrane that covers the end of the stainless steel. The poise voltage of each sensor was set to 400 mV for H₂O₂ or 865 mV for NO detection. Direct current from each sensor presents the environmental H₂O₂ or NO concentration and the data are recorded on a PC connected to the Apollo 4000 system. The H₂O₂ or NO sensors were inserted into a water-jacketed chamber containing 3 ml of MES buffer with epidermal peels from 0.5 g of Arabidopsis rosette leaves or 1 ml of HMS buffer. A circulating bath was used to keep the temperature constant at 30 °C and the sample was mixed at a low rate using a magnetic bar controlled by a magnetic stirrer. To observe Ca²⁺_o-induced H₂O₂ and NO production from the epidermis, 10 mM or 50 μM CaCl₂ (in MES buffer) was supplemented when the current signal became stable. For H₂O₂ detection from chloroplasts, a chloroplast suspension of 3×10⁶ individuals was transferred to HMS buffer with or without 10 mM CaCl₂. The H₂O₂ sensor was calibrated in a set of known H₂O₂ standards (0–10 μM range) while the NO sensor was calibrated with S-nitroso-N-acetyl-penicillamine (SNAP) in 0.1 M CuCl₂ (0–50 nM range of NO) as described by the manufacturer (Fig. 1A, B).

Fig. 1.

Electrochemical detection of H₂O₂ and NO production from Arabidopsis epidermis in response to Ca²⁺_o. (A, B) Calibration of the H₂O₂- and NO-selective electrochemical sensors ISO-HPO (A) and ISO-NOP (B). The current output jumped rapidly after each addition of H₂O₂ of known concentration or SNAP in CuCl₂ solution that produced a known concentration of NO. The calibration curve is shown in the inset. (C, D) Trace obtained upon supplementation with 10 mM (upper trace) or 50 μM (lower trace) CaCl₂ in epidermal peel suspensions using ISO-HOP (C) and ISO-NOP (D). Each trace represents the average of three samples (±SE). The arrow shows the time point at which CaCl₂ was applied.

Fluorometric detection of H₂O₂ and NO production from chloroplasts

The fluorometric method used for the detection of NO from barley root mitochondria suspensions, as described recently by Gupta and Kaiser (2010), was adapted for measuring H₂O₂ and NO production from chloroplasts. For all experiments, chloroplasts were first incubated in HMS buffer with 50 μM H₂DCFDA or 10 μM DAF-FM DA for 1 h in darkness. Chloroplasts were then rinsed twice by centrifugation at 2500 g for 4 min each, resuspended with HMS buffer, and kept in darkness until analysed. To study concentration-dependent Ca²⁺_o induction of H₂O₂ and NO production, H₂DCF- or DAF-FM-pre-loaded chloroplasts were incubated in HMS buffer with 0–2 mM CaCl₂ for 10 min. For time-related analysis, 2 mM CaCl₂ was applied for the time points ranging from 0 min to 10 min. All experiments were performed at a controlled temperature of 30 °C. The fluorescence of 5×10⁴ ml⁻¹ chloroplasts was detected using a fluorescence spectrophotometer (Cary Eclipse, Varian) with 495±5 nm and 515±5 nm as the excitation and emission wavelengths, respectively.

Real-time quantitative PCR analysis

Detached leaves were incubated in MES buffer containing 50 μM or 2 mM CaCl₂ for 0–24 h under light at 22 °C. Total RNA was extracted from leaves collected at various time points given in the figure legends using the TRIZOL Reagent (Invitrogen Inc., CA, USA). For real-time quantitative PCR, first-strand cDNA was synthesized using M-MLV reverse transcriptase (Takara Bio Inc., Japan) with an oligo d(T)₁₈ primer. The resulting cDNAs were used as templates for subsequent PCRs which were performed on the Rotor-Gene™ 6000 real-time analyser (Corbett Research, Mortlake, Australia) in standard mode with FastStart Universal SYBR Green (ROX, Roche Ltd., Mannheim, Germany) according to the manufacturer’s protocol. All cycling conditions were as follows: 10 min at 94 °C; 40 cycles of 30 s at 94 °C, 30 s at 60 °C, and 30 s at 72 °C, followed by a melting curve program (55–99 °C, with a 5 s hold at each temperature). The primers were designed according to known sequences of Arabidopsis CSD1, APX1, and ATGR1 genes (GenBank accession nos AT1G08830, AT1G07890, and AT3G24170, respectively) acquired from NCBI. The primers used for amplification are listed in Supplementary Table S2 at JXB online, and the products were checked by melting curve analysis. Amplified products were cloned into PMD-18T vector (Takara) and subjected to sequencing analyses. The mean mRNA expression level was normalized using the ΔΔCt method described by Livak and Schmittgen (2001) with Actin2 as the internal control.

Results

Ca²⁺_o induces H₂O₂ and NO production in guard cells of Arabidopsis epidermis

To assess the effects of Ca²⁺_o on promoting H₂O₂ and NO production from Arabidopsis epidermis, electrochemical sensors ISO-HPO and ISO-NOP were used to detect H₂O₂ and NO, respectively, in epidermal peels. Calibrations of both sensors were performed, and changes in the concentration of H₂O₂ and NO were then calculated based on a linear relationship between H₂O₂ or NO and the corresponding current over the current ranges 0–250 pA or 0–100 pA (Fig. 1A, B). The specificity of both sensors was also confirmed by further experiments showing that CAT or cPTIO could, respectively, remove the H₂O₂ or NO signal, while CAT or cPTIO themselves could not affect this signal (Supplementary Fig. S3 at JXB online). To initiate the reaction, a high or low concentration of CaCl₂ was added to the suspension of epidermal peels. Upon high CaCl₂ (10 mM) supplementation, H₂O₂ production from the epidermis monitored by a H₂O₂ electrochemical sensor rapidly increased to nearly 7 μM within 100 s and then reached a plateau (Fig. 1C). While high CaCl₂ caused a gradual increase of NO in epidermis, no significant change occurred after adding a low CaCl₂ concentration to a suspension of epidermal peels (Fig. 1D). In addition, NO synthesis seems to lag behind that of H₂O₂ when the epidermis is exposed to Ca²⁺_o signal.

In order to determine whether the detected production of H₂O₂ and NO was mainly released from guard cells, the H₂O₂ and NO targets in Arabidopsis epidermal peels in response to Ca²⁺_o were investigated. The endogenous H₂O₂ and NO in 10 mM CaCl₂-pre-incubated epidermal peels were determined by using the H₂O₂ and NO fluorescent probes H₂DCFDA and DAF-FM DA. Intensive green fluorescence for both probes was observed in guard cells rather than in epidermal cells (Supplementary Fig. S4A, B at JXB online). However, the fact that no fluorescence was observed from epidermal cells was not due to death during the preparation of epidermal fragments, because ∼80% of epidermal cells were still alive when estimated using Evans blue dye (Supplementary Fig. S4C). These results indicated that the detected H₂O₂ and NO were mainly produced from guard cells of extracted epidermis when exposed to Ca²⁺_o.

Chloroplasts are the cellular source for H₂O₂ production in Ca²⁺_o-exposed guard cells

It is widely accepted that chloroplasts in guard cells of most higher plants are potential sources of H₂O₂ (Wang and Song, 2008). In the present experiments, chloroplasts of guard cells on abaxial epidermal sections were selected for study of Ca²⁺_o-induced H₂O₂ production using the CeCl₃ staining method. After exposure to 2 mM CaCl₂, H₂O₂ accumulation was visible using TEM as black precipitate spots on thylakoids and in stroma of chloroplasts (Fig. 2A–D), similar to the results seen for methyl viologen-induced H₂O₂ generation in spongy chloroplasts (Pellinen et al., 1999). Interestingly, intensive green fluorescence of H₂DCF from chloroplasts was also detected in 2 mM CaCl₂-exposed guard cells (Fig. 2E) compared with the control. ABA was suggested to lead to the formation of H₂O₂ in chloroplasts and stomatal closure (Zhang et al., 2001). Therefore, experiments were carried out to examine whether there is downstream signalling cross-talk between Ca²⁺_o and ABA. Consistently, ABA-driven H₂O₂ accumulation in chloroplasts of guard cells was also observed (Supplementary Fig. S5 at JXB online).

Fig. 2.

Ca²⁺_o-induced H₂O₂ accumulation in guard cell chloroplasts from Arabidopsis. (A, B) TEM images of chloroplasts in guard cells with CeCl₃ staining from a leaf section incubated in 50 μM (A) or 2 mM (B) CaCl₂. (C) Further detail of A. (D) Further details of B. The black spots in the TEM images represent H₂O₂ forming electron-dense cerium perhydroxide precipitates. Examples of individual precipitates are shown by arrows. s, stroma; t, thylakoids. Bar=200 nm. (E) Confocal images of changes in H₂DCF fluorescence intensity from guard cell chloroplasts of the wild type in response to 50 μM (control) and 2 mM CaCl₂. H₂DCF, chlorophyll, and overlay fluorescence images from guard cell are shown. Bar=10 μm. (This figure is available in colour at JXB online.)

It was then investigated whether Ca²⁺ could directly cause H₂O₂ generation from chloroplasts of guard cells in vitro. Purified chloroplasts were transferred to HMS buffer with or without (control) 10 mM CaCl₂, and the H₂O₂ signal was detected by the H₂O₂ sensor ISO-HPO. Here, chloroplasts in suspension buffer containing no CaCl₂ can also be detected to produce H₂O₂ (Fig. 3A). This is because transferring the stable chloroplasts to a new suspension buffer will also cause H₂O₂ generation without any stimulation until they are stable again. However, the chloroplasts incubated in HMS with 10 mM CaCl₂ responded much more strongly than the control within 100 s (Fig. 3A), suggesting that Ca²⁺ enhanced H₂O₂ production in chloroplasts. Using a fluorescence spectrophotometer, a Hill curve could be fitted to the data of CaCl₂ dose-dependent H₂O₂ production (Fig. 3B) with a Hill coefficient of ∼1.1. The activation time course of 2 mM CaCl₂ could be described by a simple sigmoid equation, with an activation time constant of 6 min (Fig. 3C). However, no pronounced increase in NO from chloroplasts was observed when exposed to 0–10 mM CaCl₂ (Supplementary Fig. S6 at JXB online), indicating no correlation between Ca²⁺ and NO production in chloroplasts. It is well known that chloroplasts are the major source of H₂O₂ in plants, and H₂O₂ production in chloroplasts from guard cells or mesophyll cells in response to various stimuli has been well documented (Pellinen et al., 1999; Zhang et al., 2001), suggesting that the chloroplasts from guard cells and mesophyll cells are identical in their function to produce H₂O₂. Although total chloroplasts were used to study Ca²⁺-induced H₂O₂ production, the results may also represent Ca²⁺-induced H₂O₂ production in chloroplasts from guard cells. These data demonstrated that H₂O₂, but not NO, was produced in chloroplasts in response to a lower concentration of Ca²⁺ (∼1.1 mM) within a few minutes.

Fig. 3.

Effect of Ca²⁺ on H₂O₂ generation in isolated Arabidopsis chloroplasts. (A) The upper trace represents the rapid H₂O₂ production in chloroplasts incubated with 10 mM CaCl₂, and the lower trace represents less production of H₂O₂ in chloroplasts incubated without CaCl₂ (control). Data are averages of three samples ±SE. (B) Averaged increases in relative H₂DCF fluorescence in chloroplasts plotted as a function of applied CaCl₂ for 10 min (n=8, ±SE). Data were fitted to the Hill equation. (C) The trace shows the time-course of H₂O₂ production in chloroplasts upon 2 mM CaCl₂ supplementation. H₂DCF fluorescence was measured from the time at which CaCl₂ was provided (n=8, ±SE).

H₂O₂ and NO generation are essential for Ca²⁺_o-induced stomatal closure

The effects of Ca²⁺_o, H₂O₂, or the NO donor SNP on stomatal closure have been demonstrated before (Pei et al., 2000; Neill et al., 2002a; Han et al., 2003). Therefore, the question was asked as to whether H₂O₂ and NO generation are required for Ca²⁺_o-induced stomatal closure. The fluorescent probes of H₂O₂ and NO enabled determination of the kinetics of H₂O₂ and NO changes in guard cells upon exposure to Ca²⁺_o (Fig. 4A, B, D, E). According to Allen et al. (2001), 1–10 mM Ca²⁺_o can lead to stomatal closure. To prolong the response of guard cells to Ca²⁺_o, a low concentration (2 mM) of Ca²⁺_o was used in this experiment. In the presence of Ca²⁺_o, an alteration in H₂O₂ or NO production in the wild type was visible by 15 min, reaching saturation at 25 min, and then slowly decreased, compared with the untreated control. These results were analogous to ABA-driven H₂O₂ and NO formation in guard cells (Suhita et al., 2004; Gonugunta et al., 2008). Here, time-resolved measurements of Ca²⁺_o-, H₂O₂-, and SNP-induced stomatal closure were also performed (Fig. 4C). Exposure to H₂O₂ or SNP induced a striking decrease in stomatal aperture after 20 min. Comparably, with Ca²⁺_o treatment, stomatal aperture decrease steeply after 30 min, which was a 15 min lag behind Ca²⁺_o-induced H₂O₂ and NO production. Thus, H₂O₂ or SNP appeared to result in a more rapid decrease in stomatal aperture than Ca²⁺_o, suggesting the possibility that H₂O₂ and NO generation are required for stomatal closure in response to Ca²⁺_o.

Fig. 4.

H₂O₂ and NO productions occur before stomatal closure induced by Ca²⁺_o in Arabidopsis. (A, B) Kinetics of the increase in H₂DCF (A) and DAF-FM (B) fluorescence intensities in guard cells of both wild type and CASas in response to 2 mM Ca²⁺_o (n=50, ±SE). The relative changes in H₂O₂ and NO production were expressed using the fluorescence at time zero as the standard (100%). (D, E) Examples of H₂DCF (D) and DAF-FM (E) fluorescence images of guard cells from A and B, respectively. Bar=50 μm. (C) Kinetics of 2 mM Ca²⁺_o-, 10 μM H₂O₂-, and 60 μM SNP-induced stomatal closure of both the wild type and CASas (n=100, ±SE). Stomatal apertures (width to length) were measured as the percentage using zero time as the standard (100%). (This figure is available in colour at JXB online.)

The H₂O₂ scavenger CAT and NO scavenger cPTIO were further used to study their effects on Ca²⁺_o-induced H₂O₂ and NO production as well as stomatal closure. Ca²⁺_o-induced NO production in guard cells has been shown by using NO-specific fluorescent dyes (Garcia-Mata and Lamattina, 2007; Li et al., 2009). Here a significant increase in H₂DCF or DAF-FM fluorescence was also observed in guard cells in Ca²⁺_o-treated epidermal peels (P <0.001), demonstrating the Ca²⁺_o-induced H₂O₂ and NO production in guard cells (Fig. 5). However, pre-treatment of guard cells with CAT or cPTIO significantly suppressed H₂O₂ or NO generation and stomatal responses to Ca²⁺_o in the wild-type plants (Col-0) (P <0.001) (Fig. 5; Table 1). In line with this result, the guard cells from atrbohD/F and Atnoa1 were less sensitive to Ca²⁺_o (P <0.001) (Table 1), indicating a requirement for H₂O₂ and NO synthesis for Ca²⁺_o-induced stomatal closure. Furthermore, pre-treated epidermal peels with H₂O₂ or SNP can induce stomatal closure in the wild type (P <0.001); however, both atrbohD/F and Atnoa1 stomata did close in response to SNP (P <0.001) while Atnoa1 stomata were less responsive to H₂O₂ compared with the wild type and atrbohD/F (P <0.001) (Table 1). Since H₂O₂ unidirectionally caused NO generation within 60 s in guard cells while NO had no effect on H₂O₂ production (Bright et al., 2006; Li et al., 2009), it is assumed that Ca²⁺_o-induced NO generation is H₂O₂ dependent.

Fig. 5.

CAS regulates Ca²⁺_o-induced H₂O₂ and NO production in guard cells of Arabidopsis. (A, B) Guard cells pre-loaded with 50 μM H₂DCFDA in the wild type and CASas were incubated in MES buffer (control), 2 mM CaCl₂, or 2 mM CaCl₂ plus 100 U ml⁻¹ CAT for 20 min in darkness. H₂DCF fluorescence images and intensities of the guard cells were recorded in A and B, respectively. (C, D) Guard cells pre-loaded with 10 μM DAF-FM DA in the wild type and CASas were incubated in MES buffer (control), 2 mM CaCl₂, or 2 mM CaCl₂ plus 100 μM cPTIO for 20 min in darkness. DAF-FM fluorescence images and intensities of the guard cells were recorded in C and D, respectively. The data are expressed as the average ±SE (n=100). Bar=50 μm. (This figure is available in colour at JXB online.)

Table 1.

The effects of CAT or cPTIO on Ca²⁺-, H₂O₂-, or SNP-induced stomatal closure in various Arabidopsis genotypes

Treatment	Stomatal aperture of each genotypea
	Col-0	CASas	atrbohD/F	Atnoa1
None (control)	100.0±1.0	102.3±1.2	100.7±1.2	101.3±1.1
2 mM Ca2+	58.9±0.7	103.2±1.3	83.5±1.2	84.6±1.1
2 mM Ca2++100 U ml−1 CAT	86.2±1.1	101.1±1.4	NDb	ND
2 mM Ca2++100 μM cPTIO	96.5±1.0	101.7±1.2	ND	ND
10 μM H2O2	66.0±0.7	71.3±1.0	59.9±0.8	87.7±1.2
60 μM SNP	67.4±0.8	74.0±1.0	65.4±0.8	66.2±0.8

^a Stomatal aperture from Col-0 without any treatment is taken as 100%. Results are averages ±SE derived from analyses of stomatal apertures from 100 guard cells of at least three plants for each genotype.

^bND, not determined.

CAS mediates H₂O₂ and NO generation during Ca²⁺_o-induced stomatal closure

Earlier studies have shown that stomatal movement is modulated by the Ca²⁺_o level through CAS while Ca²⁺_o-induced stomatal closure is abolished in CASas lines (Han et al., 2003; Tang et al., 2007). Consistently, the results demonstrate that Ca²⁺_o fails to induce stomatal closure in the CASas mutants (P <0.001), and CASas plants pre-treated with CAT or cPTIO remain impaired in Ca²⁺_o-induced guard cell responses (P >0.05) (Table 1). However, the stomatal bioassay showed that exogenous H₂O₂ or SNP induced stomatal closure in both Col-0 and CASas plants (P <0.001) (Table 1). Further, the involvement of CAS in Ca²⁺_o-induced H₂O₂ and NO production was investigated. It was found that H₂DCF or DAF-FM fluorescence intensity was impaired in CASas plants after Ca²⁺_o treatment (P <0.001) (Figs 4A, B, D, E, 5). Consistently, fewer black precipitate spots of H₂O₂ were seen in CASas guard cell chloroplast (Fig. 6B) than in the wild type (Fig. 6A) in response to Ca²⁺_o. However, no precipitate spots of H₂O₂ were found in the cytoplasm, while some were detected on the plasma membrane of both wild-type and CASas guard cells in response to Ca²⁺_o (Supplementary Fig. S7 at JXB online), which was due to the activation of plasma membrane NOX by Ca²⁺_o (Sagi and Fluhr, 2001; Potocký et al., 2007). Although a small amount of NOX-driven H₂O₂ production somewhat promotes Ca²⁺_o-induced stomatal closure, H₂O₂ production in chloroplasts is more important for this process because chloroplasts and CAS are crucial for Ca²⁺_o-induced stomatal closure (Han et al., 2003; Nomura et al., 2008) and because Ca²⁺_o-induced H₂O₂ production was mainly seen in chloroplast (Fig. 2E). Meanwhile, H₂DCF fluorescence intensity was impaired in CASas guard cell chloroplasts after the exposure to Ca²⁺_o (Fig. 6C). These results indicate that CAS affects events upstream of H₂O₂-dependent NO generation, thereby triggering stomatal closure.

Fig. 6.

Ca²⁺_o-induced H₂O₂ accumulation in Arabidopsis guard cell chloroplasts is impaired in CASas. (A) TEM images showing the 2 mM CaCl₂-induced H₂O₂ accumulation in guard cell chloroplasts with CeCl₃ staining in the wild type. (B) Limited black precipitate spots in chloroplast of CASas guard cell with CeCl₃ staining exposed to 2 mM CaCl₂. The black spots in the TEM images represent H₂O₂ forming electron-dense cerium perhydroxide precipitates. Examples of individual precipitates are shown by arrows. s, stroma; t, thylakoids. Bar=200 nm. (C) Confocal images of H₂DCF fluorescence intensity from guard cell chloroplasts of the wild type and CASas in response to 2 mM CaCl₂. H₂DCF, chlorophyll, and overlay fluorescence images from guard cells are shown. The scale bar represents 10 μm. (This figure is available in colour at JXB online.)

Ca²⁺_o induces changes in transcript levels of cytosolic antioxidant enzymes in Arabidopsis

In this study, transient H₂O₂ production induced by Ca²⁺_o as well as stomatal closure induced by Ca²⁺_o within 2 h have been demonstrated. Previous work showed that ABA, CaCl₂, H₂O₂, or SNP could increase the gene expression of cytosolic antioxidant enzymes such as superoxide dismutase, CAT, APX, and glutathione reductase in maize plants (Hu et al., 2005, 2007; Zhang et al., 2006; Sang et al., 2008). Based on these findings, it was decided to determine whether the cytosolic antioxidant system functions in H₂O₂ metabolism of Ca²⁺_o stimulation. The expression of CSD1, which encodes a cytosolic superoxide dismutase that catalyses the conversion of O₂- to H₂O₂ (Kliebenstein et al., 1998), started to increase at 4 h of 2 mM CaCl₂ treatment and reached a maximum at 20 h (Fig. 7A). APX1 and ATGR1, which encode cytosolic ascorbate peroxidase and glutathione-disulphide reductase, respectively, in the metabolism of H₂O₂ (Neill et al., 2002b), were expressed in leaves after 2 mM CaCl₂ treatment, with a decrease during the first 4–8 h, followed by an increase after 8 h (Fig. 7B, C). For the control, the expression of CSD1 and APX1 remained unchanged during the 24 h period without CaCl₂ treatment, while ATGR1 transcription declined gradually in MES buffer under light. Although the expression of antioxidant genes in stomatal cells is hard to detect, the cytosolic antioxidant system might function as a mediator in Ca²⁺_o and H₂O₂ signalling in guard cells.

Fig. 7.

The transcription level of three cytosolic antioxidant enzymes in Arabidopsis leaves challenged with Ca²⁺_o. qRT-PCR analysis of time-resolved relative expression levels of CSD1, APX1, and ATGR1 in leaves incubated with MES buffer (control) or with 2 mM CaCl₂ during 24 h is shown in (A), (B), and (C), respectively. The transcript levels at 0 h of incubation were set at 1. Error bars indicate the SE from three independent repeats.

Discussion

In this study, the focus was on Ca²⁺_o-induced stomatal movement and the involvement of CAS, H₂O₂, NO, and the antioxidant system in the process. By combining the present data with results from other studiess, a model was built describing the Ca²⁺_o signal transduction pathway, which cross-talks with ABA signalling, leading to stomatal closure (Fig. 8). This model allowed the systematic understanding of the molecular mechanism of guard cells in sensing environmental cues.

Fig. 8.

Schematic representation of the possible signalling cascade leading to the stomatal closure by Ca²⁺_o and ABA. This model integrates the present data as well as results from previous studies described in the ‘Discussion’. The cascades for which the evidence is either ambiguous or still lacking are indicated by dotted arrows, while the well-established events (directly or indirectly) are represented by solid arrows.

Chloroplasts are the potential sites of CAS-mediated H₂O₂ generation in guard cells in response to Ca²⁺_o stimulation

Chloroplasts were postulated to regulate stomatal movement as a sensor of environmental stresses (Wang and Song, 2008). Further evidence was provided for the chloroplast control of Ca²⁺_o-induced Ca²⁺_i transients and stomatal closure (Nomura et al., 2008). H₂O₂ generation in chloroplasts occurred much earlier than within the other regions of guard cells after the application of exogenous ABA (Zhang et al., 2001). Consistent with this, TEM and confocal data indicate that ABA- and Ca²⁺_o-induced stomatal closure were both accompanied by H₂O₂ accumulation in chloroplasts (Fig. 2; Supplementary Fig. S5 at JXB online), suggesting a common route for Ca²⁺_o and ABA signalling mediated by H₂O₂ generation in chloroplasts. In the present experiments, it was found that a low concentration of Ca²⁺ could directly cause H₂O₂ generation in vitro within a few minutes (Fig. 3). Therefore, it is possible that Ca²⁺_o-induced H₂O₂ accumulation in chloroplasts requires an early limited increase of Ca²⁺_i which might be induced by the activation of membrane calcium channels.

CAS, a putative Ca²⁺-binding protein localized in chloroplast thylakoid membranes (Han et al., 2003; Nomura et al., 2008), elicits a cascade of intracellular signalling events including the phospholipase C–inositol 1,4,5-triphosphate (PLC–IP₃) pathway, and mediates CICI (Tang et al., 2007). This signalling cascade is probably activated by a Ca²⁺ increase in chloroplast stroma. It was noted that a large Ca²⁺ increase in chloroplast stroma followed by Ca²⁺_i transients was detected when tobacco leaves were transferred from light to darkness (Sai and Johnson, 2002). However, Ca²⁺_o-induced Ca²⁺_i transients were disrupted in CASas plants (Han et al., 2003; Nomura et al., 2008; Weinl et al., 2008). These results suggested that elevation of the stroma Ca²⁺ level might be required for the CAS-mediated CICI signalling pathway. An early limited increase in Ca²⁺_i regulated by the membrane calcium channel in response to Ca²⁺_o may meet the need for the stromal Ca²⁺ increase required for CAS activation in chloroplasts. The observation that Ca²⁺_o-induced H₂O₂ accumulation in chloroplast was impaired in CASas suggests that chloroplast H₂O₂ may function downstream of CAS in the Ca²⁺_o signalling pathway (Fig. 5A, B). In the studies of Han et al. (2003) and Nomura et al. (2008), Ca²⁺_o induced a rapid (the first step) but impaired cytoplasmic Ca²⁺ increase in CASas, and the prolonged cytoplasmic Ca²⁺ oscillations (the second step) were significantly disrupted compared with the wild type. However, they are normal in terms of ABA-driven stomatal closure (Nomura et al., 2008). Based on these findings, it is proposed that Ca²⁺_o causes biphasic Ca²⁺_i transients. The first step is an early increase of Ca²⁺_i that activates CAS in chloroplasts, followed by the second step for H₂O₂ generation and long-term Ca²⁺_i transients that also act downstream of ABA signalling through a different pathway such as cytosolic alkalinization (Suhita et al., 2004) and ABA receptors. In contrast to H₂O₂, no firm evidence for Ca²⁺-induced NO production in chloroplasts has been obtained (Supplementary Fig. S6 at JXB online). Therefore, it is possible that NO is produced around the chloroplasts through H₂O₂ diffusion, considering the requirement for H₂O₂ in ABA-induced NO generation (Bright et al., 2006).

Temporal sequence of CAS-mediated CICI and stomatal closure

As described (Allen et al., 1999, 2000, 2001; Han et al., 2003; Weinl et al., 2008), 1–10 mM Ca²⁺_o can stimulate an initial Ca²⁺_i spiking within 30 s, followed by a long-term Ca²⁺_i transient which was blocked in CASas. However, 1–2 mM Ca²⁺_o-induced stomatal closure was visible by ∼30 min as described by Weinl et al. (2008) and in the present work (Fig. 4C), while 10 mM Ca²⁺_o caused a much more rapid reduction in stomatal aperture (Allen et al., 2001). According to these reports, 2 mM or 10 mM Ca²⁺_o were also used as an experimental stimulus to probe the guard cell signalling pathway in this study. The long-term Ca²⁺_i transients regulated by CAS seem more likely to lead to the final stomatal closure. What happens to the guard cells is still unknown from the initial Ca²⁺_i spiking at 30 s to the stomatal closure at 30 min except for long-term Ca²⁺_i transients when exposed to 1–2 mM Ca²⁺_o, and how the guard cells maintain the long-term Ca²⁺_i transients during this 30 min and perform closure that is regulated by CAS remained unclear before. H₂O₂ at 100 μM was shown to cause one or two separate transients of Ca²⁺_i, leading to stomatal closure (Allen et al., 2000). Therefore, there is enough time (within 30 min) for guard cells to perform CAS-mediated H₂O₂ production to maintain Ca²⁺_i transients.

The data indicate that Ca²⁺_o-driven H₂O₂ and NO generation, which were inhibited in CASas, could be detected within 50–100 s (Fig. 1C, D) and increased steeply at ∼15 min in guard cells, earlier than Ca²⁺_o-induced stomatal closure (Fig. 4) but later than Ca²⁺_o-driven H₂O₂ production from chloroplasts at low concentration (Fig. 3). Interestingly, H₂O₂ and SNP had more rapid effects than Ca²⁺_o on stomatal closure (Fig. 4C), supporting the speculation that Ca²⁺_o induces Ca²⁺_i transients and stomatal closure through CAS regulation as well as H₂O₂ and NO generation. As for the order of H₂O₂ and NO generation, it was shown by Bright et al. (2006) that H₂O₂→NO seemed to be more acceptable. In the present work, the electrochemical sensors were used for H₂O₂ and NO detection from the epidermis and chloroplasts. Ca²⁺_o-induced H₂O₂ and NO production was observed from guard cells or chloroplasts (Figs 1, 3A). Moreover, more rapid generation of H₂O₂ than NO suggests that NO may act downstream of H₂O₂. However, the time point when Ca²⁺_o induced visible H₂O₂ and NO production was not basically in line when comparing the two approaches. The exact reasons for such a discrepancy in real-time H₂O₂ and NO detection from guard cells are not sure clear. This could be due to the varied sensitivity of each method, which, however, also resulted in an identical conclusion supporting the Ca²⁺_o signal transduction model.

ABA was shown to promote Ca²⁺_i transients, leading to stomatal closure (Allen et al., 2000). However, ABA-induced cytosolic alkalinization was visible by 10 min and ABA-induced H₂O₂ and NO production were detected by ∼15 min (Suhita et al., 2004; Gonugunta et al., 2008). These results regarding time are almost consistent with ABA-induced initial Ca²⁺_i transients at >10 min. It is well known that cytosolic alkalinization, and H₂O₂ and NO production are all required for ABA-induced stomatal closure and act upstream of long-term Ca²⁺_i transients (Pei et al., 2000; Suhita et al., 2004; Bright et al, 2006; Gonugunta et al., 2008). Similar to ABA-induced H₂O₂ and NO production, both of them are also needed to maintain Ca²⁺_i transients in response to Ca²⁺_o, leading to stomatal closure.

Pharmacological analysis further supports the Ca²⁺_o signalling transduction model

More evidence for the Ca²⁺_o signalling model was provided by studying the effects of CAT and cPTIO on Ca²⁺_o-induced H₂O₂ and NO as well as stomatal closure in various Arabidopsis genotypes. The use of these two compounds in CASas, atrbohD/F, and Atnoa1 plants reversed Ca²⁺_o-induced stomatal closure and blocked Ca²⁺_o-triggered H₂O₂ and NO production in guard cells. NOX is plasma membrane located; however, a report showed that ABA induced H₂O₂ synthesis not only via a plasma membrane NOX but also in the chloroplast, which occurred in chloroplasts earlier than within the other regions of guard cells (Zhang et al., 2001). In addition, a recent study revealed that chloroplasts played a critical role in CAS-mediated CICI and subsequent stomatal closure in Arabidopsis (Nomura et al., 2008). As shown in Table 1, compared with the wild type, atrbohD/F did respond to Ca²⁺_o, but Ca²⁺_o-induced stomatal closure was partially inhibited, while CAT, the H₂O₂ scavenger, showed more efficiency in this inhibition. Furthermore, compared with the wild type and atrbohD/F, Ca²⁺_o could not induce stomatal closure in CASas, and Ca²⁺_o-induced H₂O₂ production was largely impaired in guard cells and chloroplasts in CASas. The present data combined with those of other studies suggest that the CAS-mediated H₂O₂ synthesis in chloroplast is more critical for Ca²⁺_o-induced stomatal closure than of plasma membrane NOX. The relationship between CAS, H₂O₂, and NO was further established by the observations that H₂O₂ and SNP caused stomatal closure in both wild-type and CASas plants. If CAS mediates Ca²⁺_o-induced stomatal closure directly by Ca²⁺_i transients and this does not require H₂O₂ and NO production, it could not be understood why Ca²⁺_o-induced stomatal closure was also inhibited by the two compounds as well as in atrbohD/F and Atnoa1 plants. Further evidence suggests that H₂O₂ is required for NO generation because H₂O₂ could not induce stomatal closure in Atnoa1 plants. A similar conclusion has been drawn in a previous study (Bright et al., 2006). The finding that H₂O₂ and NO are involved in the CAS-mediated Ca²⁺_o signalling in guard cells raises the questions of how the signals from CAS to H₂O₂ are transducted and what are the signalling components involved in this process.

The putative Ca²⁺_o signal transduction pathway cross-talks with ABA and the antioxidant system

ABA was shown to promote stomatal closure in CASas plants (Nomura et al., 2008), suggesting that there is a signal converging site downstream of CAS during ABA and Ca²⁺_o signalling. In addition, ABA signalling has been known to be involved in H₂O₂-dependent NO generation (Bright et al., 2006), which were proven to act as downstream signal molecules of CAS in this study. Furthermore, findings of both Ca²⁺_o- and ABA-induced H₂O₂ generation in chloroplasts indicate that H₂O₂ may function as a common downstream component between ABA and CAS signalling (Fig. 6; Supplementary Fig. S5 at JXB online).

Interestingly, the antioxidant system seems to allow the plants to acclimatize to Ca²⁺_o-induced H₂O₂ production. Antioxidant gene transcription was analysed after a short and long period of Ca²⁺_o stimulation (Fig. 7). When exposed to Ca²⁺_o, elevation of the CSD1 expression level would promote cytosolic H₂O₂ production, while biphasic responses of APX1 or ATGR1 expression could contribute to H₂O₂ production at the beginning and H₂O₂ scavenging after saturation. This ‘H₂O₂ buffer’ can protect the plant from oxidative damage due to a long period of Ca²⁺_o signalling. A similar phenomenon was observed in ABA-treated maize plants, suggesting that the antioxidant system could also function as the convergence point between ABA and Ca²⁺_o signalling.

In summary, by using pharmacological, biochemical, and genetic approaches, this study provided comprehensive supporting evidence for a putative signalling cascade during Ca²⁺_o-induced stomatal closure. Briefly, CAS is probably activated in chloroplast stroma by a Ca²⁺_o-induced early period of Ca²⁺ transients, which initiates an intracellular signalling cascade that involves H₂O₂ and NO production, Ca²⁺_i transients, and subsequent stomatal closure in Arabidopsis guard cells. Ca²⁺_o-induced H₂O₂ and NO production and Ca²⁺_i transients as well as antioxidant enzymes transcriptional changes were also found in ABA-induced stomatal closure, which suggest a signalling interaction between Ca²⁺_o, ABA, and the antioxidant enzymes system in stomatal movement.

Supplementary data

Supplementary data are available at JXB online.

Figure S1. Verifying the mutants using PCR and western blot.

Figure S2. The integrity of isolated chloroplasts was estimated by H₂DCF fluorescent dye.

Figure S3. The effects of CAT or cPTIO on H₂O₂ or NO current detected by a H₂O₂- or NO-selective electrochemical sensor.

Figure S4. Localization of H₂O₂ and NO on Arabidopsis epidermal peels in response to Ca²⁺_o.

Figure S5. ABA-induced H₂O₂ accumulation in Arabidopsis guard cell chloroplasts from the wild type with CeCl₃ staining viewed using TEM.

Figure S6. Effect of Ca²⁺ on NO generation in isolated Arabidopsis chloroplasts.

Figure S7. TEM imgages of plasma membrane and cytoplasm in Arabidopsis guard cell with CeCl₃ staining in response to Ca²⁺_o.

Table S1. Primer sequences used for PCR analysis.

Table S2. Primer sequences used for RT-PCR analysis.