- I. Introduction 2
- II. Guard cell metabolism 2
- III. Guard cell CO2 sensing 8
- IV. Prospects in guard cell metabolism and CO2 sensing 12
In this review we concentrate on guard cell metabolism and CO2 sensing. Although a matter of some controversy, it is generally accepted that the Calvin cycle plays a minor role in stomatal movements. Recent data emphasise the importance of guard cell starch degradation and of carbon import from the guard cell apoplast in promoting and maintaining stomatal opening. Chloroplast maltose and glucose transporters appear to be crucial to the export of carbon from both guard and mesophyll cells. The way guard cells sense CO2 remains an unresolved question. However, a better understanding of the cellular events downstream from CO2 sensing is emerging. We now recognise that there are common as well as unique steps in abscisic acid (ABA) and CO2 signalling pathways. For example, while ABA and CO2 both trigger increases in cytoplasmic free calcium, unlike ABA, CO2 does not promote a cytoplasmic pH change. Future advances in this area are likely to result from the increased use of techniques and resources, such as, reverse genetics, novel mutants, confocal imaging, and microarray analyses of the guard cell transcriptome.
cytosolic free calcium concentration
guard cell protoplasts
mesophyll cell protoplasts
photosynthetic carbon reduction pathway
the small subunit of ribulose-1,5 bisphosphate carboxylase oxygenase
In higher plants, water loss and CO2 uptake are tightly regulated by stomata on the leaf epidermis. Rapid osmolyte accumulation (or loss) and consequent increase (or decrease) in the turgor of guard cells determine the extent of stomatal aperture. Under continuously changing environmental conditions stomata optimise gas exchange between the interior of the plant and the atmosphere. Guard cells possess complex signal transduction networks and modified metabolic pathways. These features allow rapid modulations in guard cell turgor and stomatal conductance, in response to internal and environmental signals (light intensity and quality, humidity, carbon dioxide partial pressure, water availability, hydric and developmental state). A general description of the basic features of stomatal function and regulation can be found in Willmer & Fricker (1996).
In this review, we will first focus on guard cell metabolism. Guard cells are highly specialised for solute accumulation and are well equipped to generate the energy required for the uptake of ions (K+, Cl−), synthesis of anions (particularly malate2–) and accumulation of osmotically active sugars (mainly sucrose). It would appear that guard cell metabolism is modified to meet these needs rather than accomplishing the typical tasks of photosynthetic carbon fixation. The relative contribution of the three processes: photosynthetic carbon reduction pathway (PCRP), the PEPC pathway and carbohydrate import, in guard cell osmo-regulation is still a matter of debate. Guard cell metabolism is highly plastic and is dependent on the energetic state of guard cells and environmental parameters (Outlaw, 2003).
The physiological role of stomata is to prevent water loss and to facilitate CO2 diffusion to mesophyll cells. Much progress has been made in our understanding of the guard cell response to stimuli such as light and water stress (Assmann, 1993, 1999; MacRobbie, 1998; Blatt, 2000; Assmann & Wang, 2001; Hetherington, 2001; Schroeder et al., 2001; Hetherington & Woodward, 2003). But very few studies have discussed the way guard cells sense CO2. One possible reason for this is the technical difficulty of regulating and recording the CO2 partial pressure during an experiment. Despite these limitations, significant progress has been made in the understanding of the guard cell CO2 signalling pathway, since the last reviews in this area (Mansfield et al., 1990; Assmann, 1999).
After presenting an overview of guard cell metabolism and CO2 sensing, this review focuses on short-term responses to CO2 of guard cells. The long-term effects of CO2 on stomatal development have been recently reviewed elsewhere (Hetherington & Woodward 2003).
In early studies of guard cell function, the importance of starch–sugar interconversion in regulating stomatal aperture received considerable attention. Later, the discovery of the large changes in guard cell potassium content during stomatal opening shifted the attention to the mechanisms underlying monovalent cation influx (Willmer & Fricker, 1996). During the last decade, new evidence has again pointed to the significance of carbohydrates, in addition to potassium and anions, during the build-up of the guard cell turgor. Thus, dual processes, involving synthesis and influx of osmotica, coexist in guard cell turgor modulation; however, at present, their respective contributions to the overall control of stomatal turgor is still a matter of debate.
In the green tissues of plants, chloroplasts and mitochondria are the two potential sources of energy, providing ATP and reducing power. Aspects of guard cell bioenergetice have been reviewed previously (Assmann & Zeiger, 1987; Raghavendra & Vani, 1989; Assmann 1993; Parvathi & Raghavendra, 1997; Willmer & Fricker, 1996). The prevailing view is that guard cells possess a high respiratory rate together with limited photosynthetic capability.
1.1. Respiration and photosynthesis As expected for cells having a high metabolic activity, guard cells contain numerous mitochondria (Willmer & Fricker, 1996). The abundance of mitochondria, along with high respiration rates, suggests that oxidative phosphorylation is an important source of ATP to fuel the guard cell machinery (Raghavendra & Vani, 1989; Parvathi & Raghavendra, 1997). The literature suggests that the guard cell mitochondria utilise both cytochrome and alternative pathways of oxidative electron transport (Mawson, 1993; Vani & Raghavendra, 1994).
By contrast to mitochondria, guard cells contain few chloroplasts (Willmer & Fricker, 1996), about one-third of the number present in mesophyll cells (Allaway & Setterfield, 1972). Further, the guard cell chloroplasts are smaller than those found in mesophyll cells (Fig. 1a,b) with limited thylakoid structures and a few granal stacks (Sack, 1987). Their chlorophyll content represents a small fraction of that in mesophyll cell chloroplasts (1–4% on a cellular basis, Zemel & Gepstein, 1985, Shimazaki et al., 1983, Reckmann et al., 1990, Gautier et al., 1991, Birkenhead & Willmer, 1986). On a chlorophyll basis, guard cell cyclic and noncyclic photophosphorylations were estimated to be about 80% of those found in mesophyll cells (Shimazaki & Zeiger, 1985). In the light, the reducing power produced by electron transport in chloroplasts can feed the Calvin cycle (Schwartz & Zeiger, 1984; Shimazaki & Zeiger, 1985). Biochemical studies have detected the main Calvin cycle enzymes (Shimazaki & Zeiger, 1985; Zemel & Gepstein, 1985; Shimazaki et al., 1989; Willmer & Fricker, 1996), but highlighted the very low level of Rubisco present in guard cells (Outlaw et al., 1979; Vaughn, 1987; Reckman et al., 1990; Gautier et al., 1991; Kopka et al., 1997).
High-resolution chlorophyll a fluorescence imaging suggests that the PCRP is active, albeit at low levels in guard cells (Lawson et al., 2003). Even using the highest values of Rubisco reported (Shimazaki & Zeiger, 1987; Shimazaki, 1989) and taking into account the low chlorophyll content in guard cell chloroplasts, CO2 fixation via the Calvin cycle should be limited to only 2%–4% of that found in mesophyll cells (Outlaw & De Vlieghere-He, 2001). Gautier et al. (1991) used mass spec-trometry to compare the unidirectional fluxes of O2 and CO2 during a dark/light transition in GCPs and MCPs. In accordance with other studies, they found that GCPs display a high respiration rate (Fig. 2a,b). The major differences between the two cell types were found at the level of CO2 fluxes (Fig. 2c,d). CO2 exchanges from MCPs appear typical of C3 plants. By contrast, in guard cells CO2 fixation displays a long induction period and stays significantly lower than O2 evolution. Such kinetics are consistent with a major participation of the PEPC pathway in guard cell CO2 fixation, with a time-lag due to the transfer of energy between different cellular compartments. Although the importance of guard cell photosynthesis to stomatal movements is still not totally resolved, the most recent data obtained from transgenic anti-Rubisco tobacco plants (von Caemmerer et al., 2004) argue for a minor role. These authors show that even a large decrease in the quantum yield of PSII in guard cells does not affect the rate of stomatal opening, steady-state stomatal conductance, or the response of stomatal conductance to ambient CO2 concentration.
1.2. Starch storage and mobilisation In the light, mesophyll chloroplasts accumulate starch and lose it in the dark. By contrast, starch is present in darkness in almost all guard cell chloroplasts (Lloyd, 1908; Robinson & Preiss, 1985, fig. 2a). Despite Stadler et al. (2003) recent report of starch accumulation under light in Arabidopsis guard cells, which contrasts with the results of another study (Lascève et al., 1997), starch content is generally inversely correlated to the degree of stomatal aperture (Outlaw & Manchester, 1979). Starch-to-sugar conversion was proposed as an osmotic motor to drive changes in guard cell turgor in the early 20th century (Lloyd, 1908; Scarth, 1927). This hypothesis was widely accepted by most physiologists until the 1960s. Later, the essential role of K+ accumulation in the build up of the osmotic potential driving stomatal movements was revealed (Immamura, 1943; Yamashita, 1952; Fujino, 1967; Fischer, 1968). However, even if potassium is now currently recognised as the major osmoticum, organic anions such as malate2– are likely candidates to balance the positive charges due to K+ accumulation while starch degradation could provide the carbon precursors needed for malate synthesis in the cytosol. Two main pathways allow carbon exchanges between mesophyll cell chloroplasts and the cytosol. Under light, the triose-phosphate/phosphate translocator of the inner envelope membrane of chloroplasts represents the major interface for the distribution of photoassimilates between the chloroplast and the cytosol. At night, maltose and, to a lesser extent, glucose are the major forms of carbon exported from mesophyll cell chloroplasts (Weise et al., 2004) and a maltose transporter, MEX1, located at the chloroplast membrane, has been recently identified (Niittyläet al., 2004). The phosphate translocator from guard cell chloroplasts of Pisum sativum has been characterised (Overlach et al., 1993). Interestingly, the guard-cell phosphate translocator differs from the mesophyll cell one in that it possesses a high affinity for Glc-6-P (as high as that determined for pea-root amyloplasts (Borchert et al., 1989)). This ability to exchange Glc-6-P provides a way to import reduced carbon that could be temporarily stored as starch in the guard cell chloroplast. By contrast with mesophyll cell chloroplasts, recent biochemical analysis of carbon export from illuminated guard cell chloroplasts by Ritte & Raschke (2003) indicated that starch breakdown results in substantial glucose and maltose export besides triose phosphates. This observation points to a specific adaptation/regulation of the guard cell chloroplast to allow starch breakdown under light and the release of precursors of malate and sucrose to the cytoplasm to sustain stomatal opening (Fig. 3). Interestingly, this last study reports that most of the carbon exported by guard cell chloroplasts originated predominantly from starch breakdown, reinforcing the general consensus of a low PCRP in guard cells.
Lascève et al. (1997) took advantage of an Arabidopsis mutant devoid of starch, a phosphoglucomutase mutant (pgm, Caspar et al., 1985), to test the importance of starch in stomatal movements. Microscopic observations confirmed that the guard cell chloroplasts from the pgm mutant plants were starch depleted, while starch was observed at dusk in wild-type plants. In whole plant experiments, such an absence of starch in guard cells in pgm plants did not affect stomatal behaviour under white light and slightly reduced stomatal response to red light. By contrast, stomatal opening under blue light was severely impaired. Interestingly, a wild-type stomatal response to blue light was restored in epidermal strips of pgm plants at high chloride concentration. These data strongly argue that malate synthesis accompanying K+-uptake under blue light is supported by starch breakdown. They also demonstrate the flexibility of guard cells about the nature of the osmoticum accumulated in response to a specific stimulus.
1.3. Guard cell PEPC pathway Most investigations conclude that Cl− influx during stomatal opening cannot compensate for the positive charge resulting from K+ accumulation (Willmer & Fricker, 1996). Thus, it has been proposed that other(s) counterion(s) such as organic anions, and mainly malate2– are involved. Indeed, a good correlation has been observed between stomatal opening and accumulation of malate in guard cells (Allaway, 1973; Pearson, 1973). Malate synthesis is highly dependent on phosphoenolpyruvate carboxylase (PEPC) activity. As the C4 enzyme, the guard cell PEPC is regulated by cytoplasmic pH, Glucose-6P (Glc-6P) and triose-6P (activators) and l-malate (feed-back inhibitor) (Tarczynski & Outlaw, 1990, 1993). However, a high sensitivity of guard cell PEPC to malate would be in contradiction with the large increase in malate content observed during stomatal opening. Indeed, in CAM, C4 and some C3 plants, PEPC is strongly regulated through phosphorylation (Nimmo et al., 1995; Chollet et al., 1996). The phosphorylated enzyme has an increased activity and is considerably less sensitive to retroinhibition by malate (Jiao & Chollet, 1990, 1991). Phosphorylation of the guard cell enzyme results in a 50% increase in the Vmax and in a large reduction in l-malate retroinhibition (Cotelle et al., 1999).
Guard cell PEPC from open stomata was found to be less sensitive to l-malate than the enzyme from closed stomata (Zhang et al., 1994) and later studies demonstrated that the phosphorylation state of the guard cell PEPC correlates with stomatal aperture. Stomatal opening triggered by fusicoccin promotes phosphorylation of the guard cell PEPC while abscisic acid (ABA) results in dephosphorylation (Du et al., 1997). Meinhard & Schnabl (2001) observed that PEPC phosphorylation under light is up-regulated by K+ and suppressed by inhibitors of the proton pump. In a recent work, Outlaw et al. (2002) showed that phosphorylation of the guard cell enzyme after fusicoccin treatment is much lower in the presence of chloride. Thus, guard cell PEPC activation through phosphorylation would not be a primary process but a response to cation influx in the cytosol. By contrast with the C4 enzyme, cytoplasmic alkalinization does not cause PEPC phosphorylation (Outlaw et al., 2002). Conversely, cytoplasmic acidification leads to guard cell PEPC activation, suggesting that cytosolic pH acts as a signal in guard cell PEPC regulation (Meinhard et al., 2001). All these results point to a specific regulation of the guard cell PEPC resulting in malate synthesis during stomatal opening.
In guard cells, the Calvin cycle and β-carboxylation pathways play complementary and redundant roles as shown by inhibitor studies. Stomatal opening is restricted in the presence of 3,3-dichloro-dihydroxyphosphinoyl-methyl-2-propenoate (DCDP), an inhibitor of PEPC. However, ribulose-5-phosphate or 3-PGA could relieve significantly the inhibition of stomatal opening by DCDP, indicating that the Calvin cycle may become significant when PEPC is restricted (Parvathi & Raghavendra, 1997). Such results were confirmed in a recent study by Asai et al. (2000). Thus both Calvin cycle and β-carboxylation pathways are beneficial for stomatal opening, particularly when either of these pathways are restricted.
1.4. Carbohydrate transporters By contrast with the limited guard cell PCRP, recent observations in planta point to a significant participation of sucrose in guard cell turgor under light (Talbott & Zeiger, 1996, 1998). Alternatively, carbohydrates could be imported from the guard cell apoplast as suggested by earlier studies. Epidermal strip experiments have suggested that the guard cell is able to import 14C-glucose and 14C-sucrose (Dittrich & Raschke, 1977; Reddy & Das, 1986). Recently, two distinct sugar import processes have been described in guard cells from Pisum sativum (Ritte et al., 1999). The first was characterised as a saturable hexose proton symporter, the activity of which depends on the membrane potential. The second has the characteristics of a sucrose transporter that could contribute to sucrose import at high apoplastic sucrose contents (> 4 mm). Interestingly, the measurement of sucrose content in the guard cell apoplast in Vicia faba plants gives evidence for just such high concentrations (Lu et al., 1995, 1997; Outlaw & De Vlieghere-He, 2001; Outlaw, 2003). Under high photosynthesis and transpiration, the sucrose content in the guard cell apoplast increased 7-fold, reaching values up to 100 mm. This sucrose accumulation in the apoplast is paralleled by an elevation of sucrose in the guard cell symplast. These results support the observations from Zeiger's lab that in planta a two-phase mechanism contributes to guard cell swelling during the day (Talbott & Zeiger, 1996). In the morning phase, opening is mostly correlated with K+ uptake in guard cells, while during the afternoon phase K+ content declines and sucrose becomes the dominant osmoticum. In such a scheme, sugar transport between the guard cell apoplast and symplast should play a crucial role (Fig. 3).
Work is still in progress concerning the characterisation of guard cell carbohydrate transporters. In Arabidopsis, AtSTP1 has been identified as a monosaccharide-H+ symporter by functional analyses in yeast (Sauer et al., 1990). Its substrate specificity is close to the hexose proton symporter observed in guard cells of Pisum sativum (Ritte et al. (1999). Recent work from Stadler et al. (2003) demonstrates that AtSTP1 gene expression is guard cell specific and displays a strong nycthemeral regulation. AtSTP1 expression is reduced during the light period and quickly up-regulated at the onset of dusk. The authors suggest that AtSTP1 could participate in guard cell import of apoplastic glucose delivered by starch breakdown in mesophyll cells at night. Additionally, it could also participate to some extent in carbohydrate import during the day. An Atstp1 T-DNA null-mutant was analysed but did not present any obvious guard cell phenotype (Stadler et al., 2003). Taking account of the multiplicity of sugar transporters in plants, other carbohydrate carriers supporting redundant or complementary functions with AtSTP1 could account for this absence of phenotype. RT-PCR performed on guard cell RNA preparations (Stadler et al., 2003), and studies of promoter-dependent GFP fluorescence (Meyer et al., 2004) revealed the expression of AtSUC2 and AtSUC3 in guard cells, guard cell specific expression of AtSUC3 being limited to very young leaves. Further analyses of carbohydrate carriers using reverse genetics should bring more information on the role of these transporters in guard cell osmoregulation.
The contribution by K+ or sucrose to guard cell osmoticum depends not only on the time of the day but also on the external stimuli (described in the following section). Thus, it is now necessary to characterise in detail the features of carbohydrate influx/efflux and the distribution between the apoplast and symplast of guard cells. While the import of glucose/sucrose into the guard cells during stomatal opening is well demonstrated, what happens to these carbohydrates during stomatal closure is not clear. Sucrose or glucose can either be exported from or metabolised in the guard cells. Further experiments are needed to examine the fate of glucose/sucrose during stomatal closure.
Guard cell carbon metabolism exhibits specific responses to different stimuli. For example, blue light triggers starch mobilisation, malate synthesis, activates the plasma membrane proton pump and K+ accumulation. Whereas, in red light, it is carbohydrate import and to a limited extent sugars, synthesised by the limited PCRP, which support guard cell turgor. However, under CO2 free-air, the response to red light becomes close to the one observed under blue light (K+ uptake, malate synthesis), highlighting the flexibility of the osmoticum accumulated according to the stimulus.
2.1. Blue/red light Stomatal responses to light are strictly wavelength dependent with blue light more efficient (2–20 fold, Willmer & Fricker, 1996) than red light in most species. These observations suggest that there must be at least two photoreceptors. As the red light response was found to be DCMU sensitive, it was inferred that it depends on chlorophyll and electron transport in guard cell chloroplasts (Tominaga et al., 2001). By contrast, the blue-light response was found to be DCMU stimulated and rotenone sensitive, suggesting an essential role for oxidative phosphorylation (Agbariah & Roth-Bejerano, 1990) in this pathway. Excitation of the blue light photoreceptor triggers an activation of the electrogenic proton pump at the plasma membrane of guard cells (Assmann et al., 1985). Recently the blue light receptor was determined (Kinoshita et al., 2001). In Arabidopsis PHOT1 and PHOT2 (phototropins) are blue light receptors exhibiting serine/threonine kinases activity (Huala et al., 1997). PHOT1 and PHOT2, which are apparently functionally redundant, mediate blue light response in guard cells. These photoreceptors undergo autophosphorylation under blue light irradiation leading to proton pump phosphorylation and interaction with 14-3-3 protein(s) (Kinoshita & Shimazaki, 2002, 2003; Kinoshita & Shimazaki, 2002) which stabilise and activate the proton pump (Maudoux et al., 2000; Emi et al., 2001). It would be interesting to know whether 14-3-3 proteins control other guard cell signalling pathways since 14-3-3 proteins regulate multiple metabolic key enzymes (Cotelle et al., 2000). However, very recent data suggest that there may be a second blue-light signalling pathway that is independent of PHOT1 and PHOT2 (Talbott et al., 2003a).
Histochemical and biochemical analyses in epidermal peels of Vicia faba showed that red light-dependent stomatal opening at ambient CO2 concentrations was largely independent of starch breakdown and K+ uptake (Tallman & Zeiger, 1988; Talbott & Zeiger, 1993; Olsen et al., 2002). This would suggest that opening under red light depends on sucrose synthesis and/or import (Poffenroth et al., 1992; Talbott & Zeiger, 1993). When epidermal peels were submitted to red light illumination under low CO2 conditions, stomatal opening was accompanied by a net increase in K+ content in guard cell and by starch breakdown (Olsen et al., 2002). These features, starch breakdown, malate synthesis and K+ uptake are reminiscent of the type of osmoticum accumulated during stomatal opening under blue light (Hsiao et al., 1973; Ogawa et al., 1978; Talbott & Zeiger, 1993). Blue light-induced starch degradation can be observed in GCPs (Fig. 1c–e. H. Gautier & A. Vavasseur, unpublished data). As discussed above, blue light activates the proton pump, hyperpolarizing the membrane potential, which drives K+ uptake through inward K+ channels. Additionally, the apoplastic acidification resulting from the proton pump activation could power sugar carriers (mainly sugar/H+ symporters) at the plasma membrane.
Interestingly, Talbott & Zeiger (1996) observed a change in the nature of the guard cell osmoticum along the course of the day. During the ‘morning’ phase, stomatal opening correlates with K+ accumulation, while in the ‘afternoon phase’ K+ content declines and sucrose becomes the dominant osmoticum. Such a shift in the nature of the osmoticum is suggestive of a transition from blue light associated osmoticum (K+) during the morning phase to a red light one (sucrose) in the afternoon. Such a change in the osmoticum may explain some discrepancies between observations of stomatal behaviour in planta and in epidermal strips as exemplified in the case of the gork-1 mutant. GORK is an outward rectifying K+ channel from the Shaker family expressed in guard cells (Ache et al., 2000). It locates to the plasma membrane and its activation through membrane depolarisation is proposed to allow K+ efflux from the guard cell cytoplasm during stomatal closure. Since K+ is a major osmoticum, GORK disruption was predicted to greatly affect the ability of stomata to close. Recently, the stomatal phenotype of the gork-1 knockout mutant has been characterised (Hosy et al., 2003). It displays a higher transpiration rate and a lower rate of stomatal closure than the wild-type plant, in accordance with a defect in K+ efflux. However, the gork-1 phenotype is much more pronounced in epidermal strips experiments than in whole plant experiments. The observation that in photosynthesising plants, guard cell K+ is with time replaced by sucrose delivered by mesophyll cells could explain such a discrepancy. Osmoregulation in epidermal peels is likely to be essentially based on K+ exchange with the bathing medium, which in the majority of experiments only contains KCl, explaining the strong phenotype of the mutant in these conditions. By contrast, in whole plant experiments, the accumulation of sucrose in guard cells would lead to a less pronounced phenotype. Indeed, in gork-1 plants, K+ efflux during stomatal closure would not be the main limiting factor. This again raises the question of the fate of sucrose during stomatal closure. It could be consumed at the mitochondrial level, reconverted to starch albeit at a low rate (Outlaw, 2003) or exported to the apoplast (Figs 3b and 5b). It is clear that much information is needed before a clear picture of the role of sucrose can emerge.
2.2. Water stress Under water stress, guard cells display a short-term response based on osmoregulation and a long-term response involving modification of major metabolic enzymes due to alterations in guard cell gene expression. The short-term response is primarily controlled by ABA, which reduces ion uptake and promotes ion efflux. This involves changes in cytoplasmic Ca2+ and pH (Assmann & Shimazaki, 1999; Hetherington, 2001; Schroeder et al., 2001). Only a few studies have investigated the long-term effects of drought stress or ABA on guard cell metabolism. Early studies, using epidermal strips, did not find any metabolic regulation in guard cells under drought stress (Grantz & Schwartz, 1988). However, subsequent studies revealed dramatic changes in guard cell expression profile of key metabolic enzymes during a short drought stress. In Solanum tuberosum, Kopka et al. (1997) observed an up-regulation of the mRNA levels of sucrose synthase and sucrose-phosphate synthase. By contrast, the expressions of KST1 (guard cell inward K+ channel), and of PHA2 (plasma membrane H ± ATPase) were reduced together with vacuolar invertase, UDP-glucose pyrophosphorylase, ADP-glucose pyrophosphorylase (large subunit), cytosolic glyceraldehyde-3-phosphate dehydrogenase, a sucrose/H+ cotransporter and an isoform of PEPC. Interestingly, PEPC, vacuolar invertase, and cytosolic glyceraldehyde-3-phosphate dehydrogenase were regulated specifically in guard cells. These changes in transcript levels were complete before any observation of a decrease in leaf water potential, which suggests the involvement of ABA (Gowing et al., 1993).
Using microarrays covering one-third of the Arabidopsis genome, Leonhardt et al. (2004) compared guard cell expression profiles with those of mesophyll cells. They observed an ABA-modulation of many known guard cell ABA signalling components at the transcript level. Apart from modulating the expression of signalling elements, key enzymes involved in guard cell carbon metabolism were also ABA-repressed. The expression level of RBCs was severely repressed by ABA in guard cells and, to a lesser extent, in mesophyll cells. Four isoforms of PEPC are encoded within the Arabidopsis genome (AtPPC1-4, Sánchez & Cejudo, 2003). Data from microarrays indicate that at least two isoforms of PEPC are expressed in guard cells and mesophyll cells of Arabidopsis, AtPPC2 and AtPPC3. AtPPC2 is the most expressed in both cell types but its level of expression is far more elevated in guard cells. When plants were sprayed with ABA (Leonhardt et al., 2004), a strong decrease in AtPPC2 expression level was observed in both cell types after 4 h of treatment. These observations of a down-regulation of PEPC transcripts by ABA are in good agreement with those of Kopka et al. (1997), with the exception that in S. tuberosum the strong inhibition of PEPC expression under drought stress was guard cell specific. These data highlight the fact that more research is needed into the contribution that metabolic regulation makes to adaptation to reduced water availability stress.
In this review, only the short-term effects of CO2 on stomatal behaviour will be considered. Long-term responses to elevated carbon dioxide have been reviewed elsewhere (Woodward, 1987; Morison, 1998; Assmann, 1999; Gray et al., 2000; Woodward et al., 2001; Hetherington & Woodward, 2003). To optimise the water use efficiency, guard cells must monitor the plant water status and the carbon dioxide demand from the mesophyll. To perform such regulation, CO2 sensing in guard cells is required. Freudenberger (1940) and Heath (1948) were the first to describe the stomatal response to elevated CO2 (reduction in aperture), which was then found to be ubiquitous in higher plants (Morison, 1985; Mansfield et al., 1990).
Studies conducted with epidermal strips or GCPs revealed that CO2 sensing is an intrinsic property of guard cells (Fitzsimons & Weyers, 1986). Later, Mott (1988) observed that in planta guard cells respond to the intercellular CO2 concentration (Ci), which is determined by atmospheric CO2 (Ca) and by the mesophyll assimilation rate. Such sensitivity to Ci allows a tight coupling between stomatal conductance and photosynthesis. In well-watered plants, the stomatal response to CO2 is generally limited, a doubling of Ca (350–700 ppm) resulting in reductions of approximately 40% of stomatal conductance (Morison, 1987).
The response to CO2 is generally curvilinear and more important below 300 ppm than at higher CO2 concentrations (Morison, 1987). In some studies maximal stomatal opening was higher at 100 ppm than in CO2-free air, suggesting that a low CO2 concentration may have a positive effect on stomatal opening (Raschke, 1976; Dubbe et al., 1978). However, whether CO2 may have a positive effect on stomatal opening remains an open question. Moreover, the amplitude of the stomatal response to changes in CO2 partial pressure is tightly dependent on many parameters such as lighting conditions and the water status of the plant.
Is there an interaction between CO2 and ABA?. Some of the early studies indicated independence (Mansfield, 1976; Mansfield & Wilson, 1981; Wilson, 1981), whereas others point to a strong interaction. Raschke (1975) was the first to report such strong interaction; working with Xanthium strumarium plants he observed that stomata did not close in response to elevated, CO2 concentrations unless the leaves had been treated with ABA. Some studies suggested that auxin could be involved in such interaction (Davies & Mansfield, 1987). Raschke (1975) also observed that stomatal responses to ABA were weak in CO2-free air (Raschke, 1975) and similar results have also more recently been reported in Arabidopsis (Leymarie et al., 1998a). As illustrated in Fig. 4, CO2 removal is able to fully abolish the inhibition of transpiration induced by 10 µm ABA in the nutrient solution. Additionally, stomatal sensitivity to CO2 was enhanced when an osmotic stress was applied to the roots (Leymarie et al., 1999). These observations point to a very strong interaction between ABA and CO2 signalling pathways. To test this relationship further, studies were performed using ABA insensitive Arabidopsis mutants (Koornneef et al., 1984). Two Arabidopsis mutants, abi1–1 and abi2–1 exhibit a wilty phenotype (Leung et al., 1994, 1997) indicating a loss of ABA control of the transpiration rate resulting from ABA-insensitive guard cells (Roelfsema & Prins, 1995). These ABA-insensitive mutants provide a good model to test the interactions between ABA and CO2 signalling. Webb & Hetherington (1997) observed that abi1–1 and abi2–1 mutants fail to respond to CO2 and extracellular calcium. From these results they inferred that the signal transduction pathways for ABA, CO2 and Ca2+ converge on, or close to, the ABI1 and ABI2 gene products. Another study using epidermal bioassays by Leymarie et al. (1998b) showed that, in the abi1–1 and abi2–1 contexts, a partial stomatal response to CO2 was observed when the K+ concentration in the bathing medium was decreased. These authors proposed that, according to the osmoticum, ABA and CO2 do not share the same signalling pathways but interact in a synergistic manner and that the ABI1 and ABI2 gene products are involved in this interaction. However, the results of these studies must be treated with caution since the abi1–1 and abi2–1 mutations are dominant. Further characterisation of recessive alleles and molecular studies have shown that ABI1 and ABI2 genes encode two protein phosphatases 2C sharing redundant functions and acting in a negative feedback regulatory loop of the abscisic acid signalling pathway (Gosti et al., 1999; Merlot et al., 2001). Accordingly, double abi1-abi2 mutant plants are ABA hypersensitive at the level of stomatal response to ABA (Merlot et al., 2001). A detailed study of CO2 sensing in these mutants is still awaited before a definitive conclusion can be drawn about the role of these redundant PP2C in integrating ABA and CO2 signalling.
The conditions under which a plant was grown have a major effect on the extent of the stomatal response to CO2. Recently, Frechilla et al. (2002) observed that stomata of growth chamber-grown Vicia faba leaves have an enhanced CO2 response compared with stomata of glasshouse-grown leaves. Complementary studies on the parameters driving this response led Talbott et al. (2003b) to propose air relative humidity as a key factor in modulating stomatal sensitivity to CO2 with elevation of relative humidity resulting in an enhanced CO2 response. The authors also suggested that humidity could function as a signal for leaves inside dense foliage canopies that promotes stomatal opening under low light and low CO2 conditions. Detailed studies of stomatal responses to air relative humidity have resulted in the conclusion that guard cells do not directly sense RH but instead respond to transpiration rate (Mott & Parkhurst, 1991). Accumulation of sucrose in the guard cell apoplast under high transpiration level has been suggested to mediate stomatal response to RH (Outlaw & De Vlieghere-He, 2001). In any case, an up-regulation of guard cell CO2 sensing by RH is difficult to reconcile with the synergistic effect of ABA and CO2 in promoting stomatal closure (Raschke, 1975; Leymarie et al., 1998a,b). Indeed, water stress and associated ABA synthesis are more likely to occur at low RH. However, a stomatal response to humidity is still observed in ABA-deficient and ABA-insensitive Arabidopsis mutants, which would suggest that ABA is not the prime mediator of the guard cell response to RH (Assmann et al., 2000). It is clear that more research is needed to clarify the relationship between CO2 sensing and plant water status.
In epidermal peels and in planta, stomatal response to CO2-free air is generally reduced under darkness and greatly enhanced in the presence of low blue light or strong red light illumination (Assmann, 1988; Vavasseur et al., 1990a, 1990b; Willmer & Fricker, 1996). Accordingly, the increase in stomatal opening triggered by light under CO2-free air is accompanied by a large increase in K+ and Cl− accumulation in guard cells (Lascève et al., 1987).
Recent studies have shown that cytosolic ATP is essential for maintaining the activity of K+-uptake channels in guard cells (Goh et al., 1999, 2002). ATP depletion results in an inhibition of inward K+ currents and photosynthetic electron transport, while addition of ADP together with orthophosphate prevents the inhibitory effect of these treatment. These results suggest that cytoplasmic ATP provides a coupling mechanism between guard cell chloroplasts, mitochondria, and ion transport. As discussed above, low blue light illumination promotes a rapid decrease in starch content (Fig. 1c–e), which could supply carbon skeletons for malate−2 synthesis (Ogawa et al., 1978) providing negative charges to balance the K+ influx and additional substrates for oxidative phosphorylation (Agbariah & Roth-Bejerano, 1990).
Under red light illumination, a significant part of the ATP produced by photophosphorylation is used for H+ pumping (Tominaga et al., 2001). This is consistent with the observation that red light triggers an electrogenic current sensitive to DCMU in GCPs (Serrano et al., 1988). In the absence of CO2, the ATP sink represented by the Calvin cycle should be limited, which should allow an increased transfer of ATP to the cytosol for proton pumping. However, subsequent studies did not confirm the activation of the proton pump under red light (Taylor & Assmann, 2001; Roelfsema et al., 2002). Recent in planta studies combining voltage-clamp and recording of CO2 partial pressure in substomatal cavity led to similar results (Goh et al., 2001, 2002). These authors failed to observe any hyperpolarization of the plasma membrane upon red light illumination when the light beam was limited to the guard cell area. Conversely, when the red light beam was extended to mesophyll cells, a lowering of substomatal CO2 partial pressure was observed accompanied by an hyperpolarization of the guard cell plasma membrane. From these data it can be concluded that the guard cell response is not primarily linked to red light illumination, but more likely to a CO2 lowering driven by photosynthesis in neighbouring mesophyll cells.
The involvement of calcium ions in ABA signalling has recently been comprehensively reviewed (Assmann & Shimazaki, 1999; Hetherington, 2001; and Schroeder et al., 2001). ABA binding to still unidentified receptors activates a transfer of Ca2+ from the guard cell apoplast and the vacuole to the cytosol. The increased Ca2+ in the cytosol ([Ca2+]cyt) inhibits the H+ pump depolarising the membrane, activates outward anion channels in the plasma membrane and blocks K+ uptake through inward K+ channels. Depolarisation and cytoplasmic alkalinization activate outward K+ channels. These events conduct the loss of solutes and stomatal closure. By contrast, much less is known about the intracellular second messengers involved in stomatal response to CO2.
4.1. Cytoplasmic free calcium By contrast to the situation with ABA, there have been relatively few investigations of the role of Ca2+ in guard cell CO2 signalling. Schwartz et al. (1988) demonstrated that, in epidermal strips, external application of calcium chelator (EGTA) results in diminished stomatal response to CO2. Later, Webb et al. (1996) used fluorescence ratio-photometry to measure [Ca2+]cyt in response to changes in CO2. Elevated CO2 (700 ppm) induced increases in guard cell [Ca2+]cyt which were similar to those observed in response to ABA (McAinsh et al., 1990, 1992). These increases in [Ca2+]cyt were reversible upon removal of CO2 and repeated application of CO2 resulted in an additional increase in [Ca2+]cyt. Importantly, removal of extracellular calcium both prevented the CO2-induced increase in [Ca2+]cyt and inhibited the associated reduction in stomatal aperture (Webb et al., 1996). The results of a pharmacological study by Cousson (2000) indicates that the CO2 signal is transduced through depolarisation-mediated activation of plasma membrane voltage-gated l-type Ca2+ channels, which would activate slow anion channels (Schroeder et al., 2001). However, until now l-type Ca2+ channels have not been identified in plants. In any case, the bulk of results underlines the potential importance of [Ca2+]cyt in CO2 signalling, as is the case for most of the effectors of stomatal responses. The transgenic lines expressing the calcium indicator yellow cameleon 2.1 (Allen et al., 1999), developed to monitor Ca2+ signalling in plant cell represent a promising approach for further studies. Despite recent debate (Köhler et al., 2003), work from the Schroeder lab suggests convincingly that generating ABA-evoked [Ca2+]cyt increases involves a reactive oxygen species regulated, voltage-dependent inward Ca2+ channels at the plasma membrane (Kwak et al., 2003). Whether such mechanism also participates in CO2 signalling remains to be determined.
4.2. Apoplastic and cytoplasmic pH Cytoplasmic and apoplastic pH are important factors, which impact on the regulation of key guard cell enzymes, for example PEPC (Cotelle et al., 1999), and ionic channels at the plasma membrane (Schroeder et al., 2001). It is now well recognised that alkalinization of guard cell cytoplasmic pH is an integral component of ABA signalling, with one of the major effects being to activate Ca2+-insensitive pH-dependent outward K+ channels (Schroeder et al., 2001). As CO2 will form carbonic acid in water it might be predicted that CO2 would induce acidification. However, recent studies fail to support this suggestion. Felle & Hanstein (2002) tracked the apoplastic pH of the substomatal cavity using pH sensitive microelectrodes. They found that application of fusicoccin, which activates the proton pump led, as expected, to a strong acidification (0.5 pH unit). An 800–0 ppm CO2 transition or illumination both resulted in an acidification of the apoplastic pH by 0.2–0.3 unit. This pH change was fully reversible when the initial conditions were restored. Brearley et al. (1997) used BCECF and ratio fluorescence microphotometry to measure cytoplasmic pH and did not record any significant pH change during a transition from 0 to 1000 ppm CO2. These observations suggest that, unlike ABA signalling, cytosolic pH changes are not an essential component in CO2 signalling. and highlight the involvement of different downstream elements in both signalling pathways.
4.3. Ionic channels and membrane potential ABA and CO2 signalling share many similarities in the way they alter the membrane potential and the main conductances at the guard cell plasma membrane. Both trigger membrane depolarisation, an inhibition of inward K+-channels, and an activation of outward anion and K+ currents (Brearley et al., 1997; Schroeder et al., 2001). A subset of these responses could be attributed to Ca2+-signalling. As described above, ABA and CO2 promote cytoplasmic calcium increases (McAinsh et al., 1992; Webb et al., 1996) that could drive inhibition of the proton pump (Kinoshita et al., 1995), deactivate inward K+ channels (Schroeder & Hagiwara, 1989; Lemtiri-Chlieh & MacRobbie, 1994) and activate anion channels (Schroeder & Hagiwara, 1989). While ABA and methyl jasmonate (Suhita et al., 2004) promote an alkalinization of the guard cell cytoplasm believed to drive K+ efflux through outward-rectifying K+ channels (Blatt, 1992), such alkalinization seems absent in CO2 signalling. How elevated CO2 results in a rapid increase in the magnitude of current carried by outward-rectifying K+ channels (Brearley et al., 1997) is currently unresolved.
Anion channel activation plays a crucial role in driving membrane potential towards K+ efflux. Hanstein & Felle (2002) studied transients in apoplastic Cl− in intact leaves during changes in substomatal CO2. They noticed that after a fast rise in substomatal CO2 from 150 to 800 ppm, it took several minutes before they recorded a significant increase in apoplastic Cl−. This delay is considerably longer than the one reported by Brearley et al. (1997) on epidermal strips using a higher CO2 partial pressure. Interestingly, the extent of CO2–induced Cl− efflux was the same in darkness and under light. By contrast, light-on, light-off transitions induced rapid variations in apoplastic Cl− when substomatal CO2 was clamped (Hanstein & Felle, 2002). These observations point to specific control of anion channels by light and CO2. Additionally, exposure to CO2-free air induced a ‘desensitisation’ of CO2-triggered Cl− efflux. These results strongly argue for an indirect effect of CO2 in the regulation of anion channels that mediate the Cl− efflux and suggest that an intermediate effector has to accumulate in response to CO2 to induce the full response. Malate has been proposed as such intermediary link between CO2 and anion channel regulation (Hedrich & Marten, 1993; Hedrich et al., 1994). Two anion conductances coexist at the guard cell plasma membrane. Slow-activating (S-type/SLAC) and fast-activating (R-type/QUACK) anion channels have been distinguished. A major role has been ascribed to S-type anion channels in ABA signalling on the basis of anion channels blockers (Schwartz et al., 1995). However, recent studies in intact plants do not exclude a participation of R-type anion channels in this response (Roelfsema et al., 2004). In response to extracellular malate, R-type anion channels display a shift in their activation potential to more negative values (Hedrich & Marten, 1993), increasing their opening probability in open guard cells. Such sensing of apoplastic malate delivered by photosynthesising tissues would provide guard cells with a feedback sensor of CO2 availability. However, this hypothesis was challenged by further observations. First, the extent of Cl− efflux in relation with CO2 has been found to be the same in darkness and under light (Hanstein & Felle, 2002), which suggests that guard cell CO2 sensing is independent of photosynthesis. Second, studies by Esser et al. (1997) and Cousson (2000) exclude CO2 sensing as primarily resulting from feedback stimulation of anion efflux via malate-sensitive anion channel since nonphysiological concentrations of malate need to be applied to observe an inhibition of stomatal opening. Additionally, work from Hedrich et al. (2001), suggests that malate2– and CO2 could act in concert as suggested by their additive effects in stomatal closing. Raschke (2003) and Raschke et al. (2003) propose that malate could participate in a conversion of R-type anion currents into S-type. They observed that CO2 variations between 0 and 700 ppm caused rapid and reversible increases in the activity of S-type, while R-type anion currents responded to CO2 in an unpredictable manner. CO2-sensitive instantaneous background currents, likely driven by anion channels (Pei et al., 1998), are also candidates in transducing the CO2 signal (Roelfsema et al., 2002). While the molecular identity of the main K+ conductances at the guard cell plasma membrane had been identified for years, the nature of proteins driving anions efflux and influx, is still unknown. Progresses in this area would be a considerable aid in understanding the way ABA and CO2 drive anion exchanges at the guard cell plasma membrane.
4.4. Redox regulation The presence of a redox system located at the guard cell plasma membrane and regulating the activity of the proton pump has been proposed based on the observation that NAD(P)H was able to drive proton efflux from guard cells (Vani & Raghavendra, 1989; Raghavendra, 1990; Gautier et al., 1992; Vavasseur et al., 1995). Such redox regulation of the proton pump could link the membrane potential to CO2 metabolism through a modulation of the pool of reducing power. However, Taylor & Assmann (2001) and Roelfsema et al. (2002) failed to confirm the presence of such a redox system in their recent patch-clamp studies.
Zeiger and collaborators (Zeiger & Zhu, 1998; Zhu et al., 1998) proposed that zeaxanthin formation in guard cell chloroplasts could be a mediator of light–CO2 interactions. They observed that stomatal aperture and zeaxanthin content in guard cell chloroplasts were linearly related over a wide range of Ca. That such a relation was absent in darkness pointed to a relation that was light dependent. Dithiothreitol, an inhibitor of zeaxanthin formation, inhibited the CO2 response in the light but not in the dark. These observations have led the authors to propose separate CO2-sensing mechanisms in guard cells in darkness and under light. However, this interesting proposal still needs to be confirmed. Dithiothreitol, besides inhibiting zeaxanthin formation, has a wide range of cellular effects. As already proposed by Assmann (1999), a better understanding of the situation could be gained by characterising stomatal CO2-sensing in the npq1 mutant (Niyogi et al., 1998), which is affected in the xanthophyll cycle and cannot de-epoxidise violaxanthin to zeaxanthin.
4.5. Protein (de)phosphorylation Regulation of proteins through (de)phosphorylation plays a major role in plant development and adaptation (Xiong et al., 2002; Luan, 2003). Pharmacological studies have shown the importance of such regulation in stomatal movements (Cousson et al., 1995; Cotelle et al., 1996; Esser et al., 1997; Suhita et al., 2003). The potential involvement of ABI1 and ABI2, two type 2C protein phosphatases, in integrating ABA and CO2 responses has been discussed above. Other essential protein kinases and protein phosphatases have been recently identified in the ABA signalling pathway leading to stomatal closure. In Arabidopsis, OST1, a calcium-independent protein kinase (Mustilli et al., 2002), which is an orthologue of AAPK, the guard cell-specific ABA-activated serine-threonine protein kinase in Vicia (Li & Assmann, 1996; Li et al., 2000), is an essential element in guard cell ABA signalling but does not participate in CO2 sensing (Mustilli et al., 2002). Again in Arabidopsis, disruption of RCN1, encoding a protein phosphatase 2A, results in guard cell ABA insensitivity (Kwak et al., 2002). Very recently, a type 2C protein phosphatase (AtP2C-HA, Leonhardt et al., 2004) has been identified as acting in an ABA regulatory feed-back loop. It is striking that the type 2C protein phosphatases presently identified are all involved in feedback regulatory loop(s) in the ABA response. To the best of our knowledge, with the exception of OST1 (Mustilli et al., 2002), none of the mutants for these different protein kinases and protein phosphatases have been used in investigations of guard cell CO2 signalling.
Among the new information gained in recent years, the potential role of of carbohydrates in maintaining guard cell turgor during the course of the day is particularly exciting. It could explain some discrepancies in the results according to the level of investigation (whole plant, epidermal strips, protoplasts), since guard cells in planta also depend on the surrounding cells. However there are important questions to address. First, is this process general? Until now it has just been described in Vicia and needs to be validated in a number of species. Second, what happens to sucrose and the other carbohydrates accumulated during opening when stomata close? A precise metabolic profile of the guard cell content during the course of the day would help to understand the interplay between organic and inorganic osmoticum involved in stomatal regulation. Figure 5 proposes a schematic representation of the interactions between guard cell metabolic pathways and membrane transport during light-induced stomatal opening or ABA- and CO2-induced stomatal closure.
Besides the classical biochemical and fluorescence approaches to study the role of the guard cell chloroplast other approaches can now be used. The effects of a lack of guard cell chloroplasts have been addressed in Paphiopedilum species. However, these orchids could have evolved compensatory mechanisms as the result of a long evolution and might not represent an actual ‘disruption’ of the chloroplastic pathway. The role of the guard cell chloroplast could be addressed by manipulating the number of chloroplasts using either pharmacological tools (Izumi et al., 2003) or Arabidopsis mutants affected in chloroplast division (Robertson et al., 1995; Larkin et al., 1997). Modulation of the expression level of key guard cell metabolic enzymes is another way to decipher the respective roles of different metabolic pathways. This is illustrated by the work of von Caemmerer et al. (2004) working with anti-Rubisco plants and in Gehlen et al. (1996) who observed that stomatal opening was delayed in PEPC antisense S. tuberosum plants and accelerated in plants overexpressing PEPC from Corynebacterium glutamicum. Such approaches can now be undertaken specifically at the guard cell level aim to the increasing knowledge about guard cell specific promoters (Taylor et al., 1995; Plesch et al., 2000; Plesch et al., 2001).
Following the pioneering work of Leonhardt et al. (2004), using genechips holding about one-third of the full Arabidopsis genome, the recent availability of full genome microarrays opens the way for exhaustive expression profiling in guard cells. This powerful tool will help in understanding guard cell global gene expression changes modulated by environmental signals. Screening and characterisation of null mutants has been and continues to be a powerful tool in the identification of components involved in ABA signalling. Infrared thermography (Merlot et al., 2002) is an ideal approach for the identification of mutants with altered stomatal response to CO2.
In the course of this review we underlined the strong interaction between ABA and CO2 sensing. It is striking that, while a legion of messenger systems are now clearly identified upstream of [Ca2+]cyt in ABA signalling, for example cyclic ADP-ribose, inositol 1,4,5 triphosphate, active oxygen species, nitric oxide, phospholipase C, phospholipase D (see Schroeder et al., 2001; Garcia-Mata et al., 2003; Hunt et al., 2003), their participation in CO2 sensing has not been investigated. The authors would like to stress that all the molecular tools developed in the course of these studies are potentially applicable to investigations of CO2. For example, few of the numerous mutants affected in their stomatal response to ABA have been studied at the level of their CO2 response. Such studies could determine common and independent elements in the respective signalling pathways. Additionally, they would allow us to ask whether some components are specifically involved in CO2-induced stomatal closure or CO2-inhibition of stomatal opening as already demonstrated for ABA (Li et al., 2000). The recent identification of the ost1 ABA-insensitive mutant is of particular interest. While this mutant is fully impaired in the guard cell ABA response it displays a wild-type response to CO2. Thus, ost1 is an ideal tool to study calcium signalling in relation with CO2 without side-effects from ABA signalling. Comparison of [Ca2+]cyt transient triggered by elevated CO2 in ost1 and wild type plants could reveal key elements in the interaction between CO2 and ABA.
The authors had to make a difficult selection among the myriad of papers addressing guard cell physiology and apologise for work that has not been included. We thank several colleagues who helped by sending reprints of their publications. The authors wish to thank members of their laboratories for helpful discussions. Recent work on stomatal signal transduction in our laboratories was supported by a grant (to AV & ASR) from Indo French Centre for the Promotion of Advanced Research (No. 2203–1), New-Delhi. Special thanks to Professor Bill Outlaw, for his kind permission to use one of his figures as the basis of our model of guard cell CO2 metabolism.