In the light of stomatal opening: new insights into ‘the Watergate’


  • M. Rob G. Roelfsema,

    1. Molecular Plant Physiology and Biophysics, Julius-von-Sachs Institute for Biosciences, Biocenter, Würzburg University, Julius-von-Sachs-Platz 2, D-97082 Würzburg, Germany
    Search for more papers by this author
  • Rainer Hedrich

    Corresponding author
    1. Molecular Plant Physiology and Biophysics, Julius-von-Sachs Institute for Biosciences, Biocenter, Würzburg University, Julius-von-Sachs-Platz 2, D-97082 Würzburg, Germany
    Search for more papers by this author



  • Summary 1

  • I. Introduction 1
  • II. The hydrodynamic valve 2
  • III. Regulation of ion transport in guard cells 14
  • IV. Interaction between guard cell signaling pathways 18
  • V. Outlook 20
  • Acknowledgements 21

  • References 21


Stomata can be regarded as hydraulically driven valves in the leaf surface, which open to allow CO2 uptake and close to prevent excessive loss of water. Movement of these ‘Watergates’ is regulated by environmental conditions, such as light, CO2 and humidity. Guard cells can sense environmental conditions and function as motor cells within the stomatal complex. Stomatal movement results from the transport of K+ salts across the guard cell membranes. In this review, we discuss the biophysical principles and mechanisms of stomatal movement and relate these to ion transport at the plasma membrane and vacuolar membrane. Studies with isolated guard cells, combined with recordings on single guard cells in intact plants, revealed that light stimulates stomatal opening via blue light-specific and photosynthetic-active radiation-dependent pathways. In addition, guard cells sense changes in air humidity and the water status of distant tissues via the stress hormone abscisic acid (ABA). Guard cells thus provide an excellent system to study cross-talk, as multiple signaling pathways induce both short- and long-term responses in these sensory cells.

I. Introduction

Stomata can be regarded as hydraulically driven valves that operate in the aerial parts of land plants. These pores enable the diffusion of gases between the plant's interior and the atmosphere. As a result of the swelling and shrinking of two guard cells that border the pore, stomata can open and close and thus form a ‘Watergate’ that controls the loss of water (for uptake and transport of water in plants see review by Zimmerman et al., 2004). The regulation of stomatal movement by endogenous and environmental signals has fascinated plant biologists for centuries (Meidner, 1987). In 1812 the German botanist, Moldenhauer, wrote about the stomata of maize:

… nur am frühen Morgen, wenn die Sonne auf die noch thauigen Blätter schien, oder wenn ich sonst das Blatt, ohne es vom Stengel zu trennen, oder es zu beschädigen, in ein Gefäß mit Wasser bog, und es so dem Sonnenlicht aussetzte, wenn also die Ausdünstung der Blätter zurückgehalten wurde, sah ich sie geöffnet. … ,

Translated, this states that he only observed open stomata in the early morning with the sun shining on still-dewy leaves, or when he put the leaf, without excision from the stem, into a jar with water and let the sun shine on it, in other words when transpiration was prevented. Moldenhauer (1812) recognized three important signals affecting stomatal movement: light, humidity, and time of day. The pathways coupling these signals to stomatal movements have, to date, only been partially resolved and are still the subject of intensive research.

The central function of stomata is to balance the uptake of CO2, which is essential for photosynthesis, with loss of water from the plant. During evolution, land plants developed a cuticle on aerial tissues to prevent dehydration. This gas-impermeable waxy layer, however, also limits the uptake of CO2. To circumvent the latter restriction, the leaf epidermis harbors a large number of pores, named stomata. The degree of opening of these stomata is determined by two guard cells that surround the stomatal pore. Apart from leaves, stomata are, in certain plants, also found on the stem, petioles, primary roots (Christodoulakis et al., 2002) or in nectaries (Horner et al., 2003). In these organs, stomata presumably have functions other than facilitating CO2 uptake, such as the extrusion of nectar from active nectaries (Horner et al., 2003). Little is known about the regulation of stomatal movement in these nonphotosynthetic organs. In this review, we will therefore focus on the control of stomatal movement in photosynthetic-active leaves.

In general, light and drought act in an antagonistic manner on stomatal movement. Light induces the opening of stomata to enhance CO2 uptake, while drought causes stomata to close, thereby limiting water loss through transpiration. In nature, stomata will often receive both signals at the same time, as sunny periods frequently coincide with drought. Under these circumstances, plants will have to make a trade off – either open the stomata for optimal photosynthesis, or close them to save H2O. The fine-tuning of this process probably differs between plant species and will also depend on other growth conditions and the developmental stage of the plant. However, the principal mechanisms used to process these signals are probably conserved within the plant kingdom. An exception to this rule is the group of plants with crassulacean acid (CAM) metabolism. These plants open their stomata at night and close them during the day, which allows them to grow in extremely dry environments. Although the regulation of stomatal movement by light is inversed in CAM plants, the signaling pathways may still have features in common with C3 and C4 plants (Section III, heading 1). For more details on CAM metabolism see previous excellent reviews by Black & Osmond (2003) and Lüttge (2002).

This present review will attempt to couple recent advances in the understanding of guard cell ion-transport with changes in the osmotic pressure, turgor and cell volume. Guard cells have often been studied in isolation from neighboring cells, and the results were seldom tested for significance in intact plants. At the 13th international workshop on Plant Membrane Biology (2004), in Montpellier, this led to the remark that: ‘Guard cells seem to have gained life independent of the plant’. Here, we will attempt to put guard cells back in their natural context and discuss the current knowledge of guard cell physiology in relation to stomatal function in the intact plant. In this context, the interaction of guard cells with neighboring cells, and their responses to signals from more distant cell types, will be addressed.

Although stomata are found on most land plants, a few species have been favored for research on stomata. In this review we will mainly discuss the guard cells of broadbean (Vicia faba L.), as guard cells of this species have been studied by using gas exchange, ion-uptake measurements, microprobe analysis, biochemistry, electrophysiology and the pressure probe. For molecular biology and signal transduction research, guard cells of Arabidopsis thaliana are often the system of choice. The stomata of the latter two species differ considerably in size (Langer et al., 2004), but also have many features in common. The stomatal complexes of these two dicot species consist only of two kidney-shaped guard cells and they are distributed on the leaf surface without an obvious patterning. In addition to these dicot species, two nongrass monocot species –Commelina communis and Trandescantia virigiana– have been important in stomatal research. The epidermis of these species is easily detached from the leaf (Weyers & Travis, 1981) and has been used to study ion uptake (MacRobbie & Lettau, 1980a,b; MacRobbie, 1987), turgor pressure (Franks, 2003) and cytoplasmic Ca2+ signaling (McAinsh et al., 1995). The stomatal complexes of these monocots consist of guard cells that are bordered by subsidiary cells and they are distributed parallel to each other on the leaf surface. Stomata of Gramineae are even more distinct because their guard cells are dumbbell shaped. The mechanics of stomata of these species may differ to some extent; however, the hypothesis put forward by Von Mohl (1856) –‘die Porenzellen erweitern durch ihre Turgeszenz die Spaltöffnung und verengen sie durch Collabiren’– which means that guard cells open the stomatal pore with their turgor and close it by collapsing, seems to hold for stomata of all plant species.

II. The hydrodynamic valve

1. Osmotic motor

Water potential  Like most motions exerted by plants, stomatal movements are hydraulically driven. Stomatal movements, however, differ from many other motions exerted by plants because they are completely reversible. Movements related to growth, like hypocotyl bending, involve both cell division and expansion and therefore are only partially reversible. Water movement into the guard cell is driven by osmosis and can be best understood considering the chemical potential of water. Accumulation of solutes in the guard cell cytoplasm lowers the chemical potential of water (µw) and its derivative, ‘water potential’ (Ψ). The water potential (Eqn 1) is equivalent to the chemical potential of a solution, relative to the chemical potential of pure water (inline image) at the same temperature and atmospheric pressure, divided by the partial mole volume of water, w (Nobel, 1970), as follows:

image((Eqn 1) )

The water potential of pure water thus is 0 and it becomes more negative when solutes accumulate. The role of the water potential in stomatal movement has been reviewed by Raschke et al. (1988) and is based on the relationship (Eqn 2) of the water potential (Ψ), turgor pressure (p) and osmotic pressure (π):

image((Eqn 2) )

This means that an increase in the osmotic potential of the guard cell leads to an initial decrease in the water potential. Given a high hydraulic conductivity of the plasma membrane, water will flow into the guard cell and the Ψ of the guard cell will equilibrate with that of the apoplast. The inflow of water will cause the turgor pressure to rise and the guard cells to swell. The increase in volume of both guard cells causes opening of the stomatal pore.

As a result of the radial arrangement of cellulose fibrils in the cell wall, the expansion of kidney-shaped guard cells occurs mainly along the longitudinal axis (Sharpe et al., 1987; Shope et al., 2003). As a result, the guard cells bend, keeping the length of the stomatal complex virtually unchanged. In addition to bending, the cross section of guard cells changes from a flattened oval to a circular shape during stomatal opening (Fig. 1) (Von Mohl, 1856; Raschke & Dickerson, 1972; Franks et al., 2001). Changes in guard cell shape and volume during stomatal movements were recently quantified for guard cells of V. faba by using a pressure probe and confocal microscopy (Franks et al., 2001; Shope et al., 2003). Shope et al. (2003) found that the volume of guard cells in opened stomata ranges from 5 to 8 pl and decreases, during stomatal closure, to 3.5–4.5 pl. This indicates that guard cells of V. faba lose ≈ 40% of their volume during stomatal closure.

Figure 1.

Changes in the shape of guard cells during stomatal movements. During stomatal opening the guard cell volume increases, which causes them to bend and changes their cross section from an flat oval (right drawing), to a circle (left drawing). Redrawn after Raschke & Dickerson (1972) and Wanner (2004).

Guard cell volume and turgor  The increase in cell size during stomatal opening also leads to an increase of the guard cell surface. These processes were studied with fluorescent dyes that accumulate in the plasma membrane (Shope et al., 2003). Images obtained with a confocal laser-scanning microscope revealed an approximately equal change in surface area and guard cell volume during stomatal movement (Shope et al., 2003). This implies that the guard cells need to expand their plasma membrane surface by ≈ 40% during stomatal opening. Such a large change in surface area is probably accomplished by fusion of vesicles with the plasma membrane, while vesicles are budding from the plasma membrane during stomatal closure. The latter process was supported with fluid-phase endocytosis (Diekmann et al., 1993) and confirmed by emerging fluorescent vesicles derived from the plasma membrane during stomatal closure (Shope et al., 2003; Meckel et al., 2004). Vesicle fusion and fission of plant protoplasts can be studied by using the patch clamp technique, as the capacitance of the protoplast is linearly related to its plasma membrane surface (Homann, 1998). Application of hydrostatic pressure to the guard cell protoplasts reversibly alters the membrane capacitance. In addition, changes in the plasma membrane conductance occur as ion-channels are incorporated into or removed from the plasma membrane (Homann & Thiel, 2002; Hurst et al., 2004). Hydro-passive volume changes in guard cell protoplasts thus provide an elegant system to study vesicle trafficking and proteins involved this process, such as the syntaxins (Leyman et al., 1999).

The turgor pressure necessary to provoke an increase in guard cell volume and surface area has been recorded for intact guard cells by using the pressure probe (Franks et al., 1998; Franks et al., 2001). A pressure of 4.5 MPa was required for a maximal stomatal opening, a value well in agreement with the data of Raschke et al. (1972) and Fischer (1972) (Tables 1 and 2). The latter authors used concentrated sucrose solutions to induce stomatal closure, and achieved complete closure in solutions with an osmotic pressure of 8 MPa (Raschke et al., 1972). Taking into account a decrease in cell volume of 40% during stomatal closure, the osmotic pressure of guard cells in fully opened stomata was 4.8 MPa. These results led to the conclusion that the turgor of V. faba guard cells increases with 0.23 MPa µm−1 increase in stomatal aperture (Table 2). Fischer (1973) determined a very similar value of 0.2 MPa µm−1. Apparently, a turgor increase of ≈ 0.2 MPa is necessary to drive a 1-µm increase in aperture for V. faba stomata in epidermal strips.

Table 1. Biophysical parameters of open and closed Vicia faba stomata in the presence or absence of viable epidermal cells
 Closed stomataOpen stomata
Guard cell volume (disrupted epidermal cells)4 pla6.5 pla
Stomatal aperture (disrupted epidermal cells)2 µmb14 µmb
Guard cell turgor (disrupted epidermal cells)2.0 MPac4.8 MPac
Stomatal aperture (viable epidermal cells)0 µmd8 µmd
Guard cell turgor (viable epidermal cells)1.0 MPae4.5 MPae
K+ content (disrupted and viable epidermal cells)0.3 pmolf2.5 pmolf
Table 2. Estimated changes in biophysical parameters of Vicia faba stomata during stomatal movement, in the presence or absence of viable epidermal cells
Change in:Disrupted epidermal cellsViable epidermal cells
  • a

    Owing to the approximate linear relationship between stomatal volume and turgor pressure (Shope et al., 2003), as well as between turgor pressure and stomatal aperture (Franks et al., 1998), a volume change of 0.21 pl µm−1 can be calculated for stomata in epidermal peels with disrupted epidermal cells. (Based on the data shown in Table 1.)

  • b

    It is unclear how the volume changes of guard cells during stomatal movement is influenced by viable epidermal cells.

  • c

    The data in Table 1 relate to a change in guard cell turgor pressure of 0.23 MPa µm−1 for stomata in epidermal strips with disrupted epidermal cells.

  • d

    Viable epidermal cells reduce the stomatal aperture by ≈ 50% and thus the change in guard cell turgor per µm aperture is twice as large. Franks et al. (1998) determined a change of 0.4 MPa µm−1.

  • e

    Based on the data in Table 1, guard cells accumulate 0.18 pmol of K+ per µm increase in stomatal aperture. Fischer (1972) concluded that guard cells take up 0.21 pmol µm−1, in the absence of viable epidermal cells.

  • f

    In the presence of viable epidermal cells the stomatal aperture is reduced by 50%, indicating that guard cells take up twice as much K+ per µm incease in stomatal aperture.

Guard cell volume0.21 pl µm−1 a b
Guard cell turgor0.23 MPa µm−1 c0.46 MPa µm−1 d
K+-content0.18 pmol µm−1 e0.36 pmol µm−1 f

Interaction of guard cells and epidermal cells  It should be noted that the data of Fischer (1972) and Raschke et al. (1972) were obtained using epidermal peels of V. faba in which the epidermal pavement cells were disrupted. In the latter preparations, the maximal stomatal aperture was ≈ 12 µm. In the experiments of Franks et al. (1998), the epidermal cells remained intact and the maximal aperture was only 8 µm. These differences are in agreement with data of Klein et al. (1996), who compared fusicoccin-induced stomatal opening in epidermal strips of V. faba, with viable and disrupted epidermal pavement cells. The stomatal opening was 8 µm in the presence and 16 µm in the absence of viable epidermal cells. A reduction in the stomatal aperture of ≈ 50% by viable subsidiary cells and epidermal cells, over a whole range of apertures, was also found for T. virgiana (Franks et al., 1998) and for C. communis (MacRobbie & Lettau, 1980a). Apparently, viable epidermal cells provide a back pressure that reduces the stomatal aperture by ≈ 50%. The absence of back pressure in many epidermal strip preparations also explains why stomata often do not completely close, while fully closed stomata are generally observed in dark-adapted intact leaves (Felle et al., 2000).

The interaction of epidermal cells with the stomata was studied in detail by using epidermal strips of T. virgiana (Franks et al., 1998). If the epidermal cells were disrupted, the stomata started to open as soon as turgor pressure was applied. When epidermal cells had a turgor pressure of 0.9 MPa, however, no stomatal opening was observed with guard cell turgor pressures of < 1.25 MPa. At higher pressures, stomata started to open, but viable epidermal cells reduced the final stomatal aperture (Franks et al., 1998; Franks, 2003). The increase of turgor without causing stomatal opening has been described as the ‘Spannungsphase’, by Stålfelt (1929). At higher turgor pressures, guard cells are in the ‘Motorphase’, in which an increase in turgor leads directly to stomatal opening. During the ‘Motorphase’ a virtual linear relationship between the guard cell turgor and the stomatal aperture exists, as long as the aperture is not limited by constrains in the extensibility of the cell wall (Franks et al., 1998).

Ion accumulation  In intact plants, guard cells need to accumulate a large volume of solutes to raise their osmotic potential and force the stomatal pore to open. In the early literature, guard cells were believed to convert starch into sugars, as starch disappears from guard cell chloroplasts during stomatal opening in the light (Strugger & Weber, 1926). However, this is not the case for guard cells of all plants, as A. thaliana guard cells synthesize starch during the light period and degrade this polymer in the dark (Stadler et al., 2003). In this respect, guard cells are not different from mesophyll cells, which accumulate starch during the light period and remobilize carbohydrates in the dark (Zeeman et al., 1998). Apparently, the storage and breakdown of starch in guard cells does not necessarily correlate to stomatal movement.

Later reports showed that stomatal opening depends on the uptake of K+ into guard cells (Fischer, 1968). Guard cells of V. faba take up ≈ 2 pmol of K+ during stomatal opening (MacRobbie, 1987), which relates to a K+ uptake of 0.18 pmol of K+µm−1 (Table 2); Fischer (1972) determined a value of 0.21 pmol of K+µm−1. In the long term, guard cells can only accumulate positively charged ions, when these are counterbalanced by anions. For guard cells in epidermal strips of V. faba, K+ uptake was balanced by the accumulation of equal amounts of Cl and malate (Raschke & Schnabl, 1978). Providing that the latter would hold for guard cells in intact plants, these motor cells accumulate 2 pmol of K+, 1 pmol of Cl and 1 pmol of malate. Malate will probably have only a single negative charge as it is predominantly stored in the guard cell vacuole (Schnabl & Kottmeier, 1984a), which presumably has an acidic pH. The maximum increase in the osmotic pressure can now be calculated (Nobel, 1970) as follows:

image((Eqn 3) )

where πs is the osmotic pressure of all solutes, Σ cj is the sum of the concentration of all solutes and RT has a value of 2.44 l MPa mol−1 at 20°C. Given a volume of 4 pl, the increase in πs would be 2.4 MPa. Considering a turgor pressure of ≈ 1 MPa of guard cells in closed stomata, the final turgor pressure would be ≈ 3.4 MPa. This is less than the 4.5 MPa that would be expected for fully open stomata. In fact, the increase in osmotic pressure has been overestimated above, as the increase in volume of the guard cells during stomatal opening was not taken into account. Furthermore, Eqn 3 only holds for dilute solutions, as it overestimates the osmotic pressure of concentrated solutions (Nobel, 1970). Apparently, the measured accumulation of K+ salts in guard cells does not cause a rise in osmotic pressure large enough to explain the increase in stomatal aperture, a discrepancy also observed by Fischer (1972) and MacRobbie (1987). This indicates that either the accumulation of K+ has been systematically underestimated, or that other solutes accumulate in addition to K+ salts.

Sugar uptake  There are several lines of evidence indicating that guard cells accumulate sugars, in addition to K+ salts, during stomatal opening. The concentration of sucrose in guard cells of V. faba can increase, from 0.2 to 0.7 pmol per cell, during the light period (Talbott & Zeiger, 1996). However, sucrose mainly accumulates during the late light period, when K+ concentrations already decrease again. The accumulated sucrose thus may just replace K+ over the course of the day, instead accelerating stomatal opening. The role of sugars in stomatal movement is supported by the results of Ritte et al. (1999), who found that guard cell protoplasts of Pisum sativum take up significant amounts of glucose, fructose and sucrose. More recently, the A. thaliana hexose transporters AtSTP1 (Stadler et al., 2003) and AtSTP13 (M. Büttner, pers. comm.) were found to be highly expressed in guard cells. However, the transcript level of the AtSTP1 transporter was highest in the dark, which brings its role in stomatal opening into question. Mutant plants lacking AtSTP1 also did not show a reduction in stomatal opening, indicating that either other hexose transporters (such as AtSTP13) or sucrose transporters can act as a backup for AtSTP1, or that sugar uptake is not required for stomatal opening.

Although there is no problem in understanding how the uptake of sugars can contribute to stomatal opening, it is unknown how sugars can disappear rapidly from the guard cell cytoplasm during stomatal closure. The latter process can occur very rapidly, as abscisic acid (ABA) induces stomatal closure within 10 min (Roelfsema et al., 2004). The rate of starch synthesis is regulated by the activity of ADP-glucose pyrophosphorylase, which is ≈ 35 mmol kg−1(dry weight) h−1 in guard cells (Outlaw & Tarczynski, 1984). Based on a dry weight of 3.1 pg per guard cell (Outlaw & Lowry, 1977), this correlates to the conversion of 0.1 pmol of ADP-glucose h−1 per guard cell, which is much too slow to support ABA-induced stomatal closure. It is therefore more likely that sugars are extruded from guard cells during stomatal closure, which also would explain the accumulation of sucrose in the apoplast of guard cells at low humidity (Outlaw & De Vlieghere-He, 2001). The transport proteins that facilitate such an extrusion of sugars, however, have not yet been identified, but it may be that STP- and SUC-transporters can act in the reverse mode (own unpublished data).

Na+-supported stomatal movement  Under physiological conditions, guard cells will mainly accumulate K+ salts during stomatal opening because this cation is available at concentrations of 3–40 mm in the guard cell wall (Felle et al., 2000; see review Roelfsema & Hedrich, 2002). However, stomatal opening is also supported by other monovalent cations such as Rb+, Li+, Cs+ and Na+ (Humble & Hsiao, 1969). Most of these cations will not accumulate in the apoplast at concentrations high enough to compete with K+. An exception is Na+, for plants growing on salt-enriched soils. Na+ supports stomatal opening in V. faba and C. communis, and in C. communis Na+ is even more efficient than K+ (Willmer & Mansfield, 1969). However, stomatal closure is disabled in stomata that were previously opened in the presence of Na+ (Jarvis & Mansfield, 1980; Robinson et al., 1997). Guard cells thus seem to be equipped with a Na+-uptake system, but have limitations considering Na+ release. The inability to extrude Na+ may be caused by Na+ blockage of plasma membrane K+ channels (Thiel & Blatt, 1991). Alternatively, it may be the result of an inability of slow vacuolar (SV) channels to extrude Na+ (Ivashikina & Hedrich, 2005; see also section II, 2). The latter result is well in agreement with the swollen vacuoles observed for guard cells that were opened in the presence of Na+ (MacRobbie, 1983). Halophytes may have developed two mechanisms to overcome this problem: they are either capable of extruding Na+ or they selectively accumulate K+, but not Na+ (Robinson et al., 1997). In the halophyte Aster tripolium, the latter adaptation was associated with a Na+-induced change in the activity of K+-uptake channels (Véry et al., 1998).

2. Guard cell ion transporters

Ion-transport proteins of guard cells have been studied in great detail because their activity is linked directly to stomatal movement. Guard cell ion transport differs from that in most other plant cells because ion transport in guard cells occurs in two directions (to enable stomatal opening as well as closure). By contrast, other plant cells (such as mesophyll- or epidermal pavement cells) grow irreversibly. It is therefore highly likely that the properties of ion transporters, and their regulation by various signals, differ between different cell types. Cell type-specificity of ion-transport genes has rarely been documented, but each cell seems to run a unique set of transport proteins. For instance, the expression rate of the K+-channel gene, GORK, is very high in guard cells, but it is virtually absent from mesophyll cells (Ache et al., 2000).

The regulation of guard cell ion transport by light and humidity will be addressed in section III. Below we will review the present knowledge about the properties of ion-transport proteins in the guard cell plasma- and vacuolar membrane. Because of the large data set available for these transport proteins, it is impossible to cover this in every detail (see review Véry & Sentenac, 2002). Instead, we will focus on the properties of transporters linked to stomatal function (Fig. 2).

Figure 2.

Overview of ion-transport proteins that have been identified or postulated in the plasma membrane (upper scheme) and vacuolar membrane (lower scheme) of guard cells. Transport proteins, for which one or more genes have been isolated, are shown in black, while those for which only physiological evidence exists are in grey. 1. Outward-rectifying K+ channel, GORK. 2. Inward-rectifying K+ channel, KAT1, KAT2, AKT1, AKT2/3 and AtKC. 3. Hyperpolarization-activated Ca2+ channel. 4. Ca2-ATPase, ACA. 5. R-type anion channel. 6. S-type anion channel. 7. H+-ATPase, AHA. 8. NO3 transporter, CHL1. 9. Cl transporter. 10. Slow vacuolar (SV) channel; 11. Fast vacuolar (FV) channel. 12. Vacuolar K+-selective (VK) channel. 13. K+ transporter, NHX. 14. H+-pyrophosphatase. 15. V-type H+-ATPase, V1 and V0-subcomplexes. 16. Ca2+-dependent protein kinase (CDPK)-activated anion channel. 17. Hyperpolarization-activated anion channel. 18. Malate carrier, AttDT. 19. Ca2+-carrier, CAX; ACA. 20. Vacuolar ACA. 21. Voltage-gated Ca2+ channel. 22. Inositol triphosphate (IP3)- and inostitol hexakis-phosphate (IP6)-gated Ca2+ channels. 23. Cyclic ADP ribose-activated Ca2+ channel.

Plasma membrane transport  K + transport The importance of K+ uptake during stomatal opening has been undisputed, ever since Fischer showed, in 1968, that stomatal movement in epidermal strips depends on the uptake of K+ (Fischer, 1968). However, the nature of the proteins facilitating K+ uptake remained unclear until Schroeder et al. (1984) showed the presence of K+-selective ion channels in the plasma membrane of V. faba guard cells. Another breakthrough came 8 yr later with the isolation of the first plant K+-channel genes (Anderson et al., 1992; Sentenac et al., 1992), which was followed by the guard cell-specific expression of the K+ channel encoded by KST1 and KAT1 (Müller-Röber et al., 1995; Nakamura et al., 1995).

Plant plasma membrane K+-channel types can be divided into three groups: outward-rectifying channels, which become more active at positive potentials (Fig. 3c); inward-rectifying channels, activating at more negative potentials (Fig. 3d); and a third type of channel that is largely voltage-independent. The first two types of K+ channels have been recorded in guard cells of various plant species, such as V. faba (Schroeder et al., 1987; Blatt, 1988) A. thaliana (Roelfsema & Prins, 1997; Pei et al., 1997), Nicotiana bethamiana and N. tabacum (Armstrong et al., 1995; Dietrich et al., 1998), Zea mays (Fairley-Grenot & Assmann, 1993) and Populus tremula ×P. tremuloides (Langer et al., 2004). Based on their voltage dependence, outward-rectifying K+ channels are, at physiological conditions, responsible for K+ extrusion, while inward-rectifying channels mediate K+ uptake. Genes encoding voltage-independent K+ channels are only expressed at low rates in guard cells and the activity of these channels has not, so far, been recognized for guard cells (Blatt et al., 1990; Blatt, 1992; Roelfsema & Prins, 1997, 1998).

Figure 3.

K+ channel currents in guard cells of intact plants and through ‘Shaker like’ K+ channels expressed in Xenopus oocytes. (a) Plasma membrane current of an Arabidopsis thaliana var. Landsberg erecta guard cell on the abaxial side of a leaf. The plasma membrane was stepped from a holding potential of −80 mV, to test potentials ranging from −180 mV to 20 mV, at 20-mV increments. Note the activation of inward- and outward-rectifying K+ channels with increasing negative- and positive potentials, respectively. (b) Current–voltage relationship of steady-state currents from the same cell as in (a). Note the activation of inward- and outward-rectifying K+ channels at potentials negative of −120 mV and positive of −60 mV, respectively. (c) Whole-oocyte currents of GORK recorded in a bath solution containing 30 mm KCl, 1 mm CaCl2, 1 mm MgCl2 and 10 mm Tris/Mes, pH 7.5. Currents were elicited from a holding potential of −80 mV to test potentials ranging from −80 to 50 mV, at 10-mV increments. (d) Oocyte currents of KAT1 recorded in 30 mm KCl, 1 mm CaCl2, 1 mm MgCl2 and 10 mm Mes/Tris, pH 5.6. Currents were elicited with test potentials ranging from 20 to −150 mV, at 10-mV increments.

The gating of outward- and inward-rectifying K+ channels differs in a crucial way: activation of outward K+ channels is dependent on the extracellular K+ concentration, while that of inward K+ channels is not. Outward-rectifying K+ channels respond to changes in the extracellular K+ concentration in such a way that the channel always activates at potentials slightly more positive than the Nernst potential for K+ (Blatt, 1988; Roelfsema & Prins, 1997; Gaymard et al., 1998; Ache et al., 2000). In contrast, the activation of inward channels is not affected by K+ (Blatt, 1992; Roelfsema & Prins, 1997), and these channels do not exclusively facilitate K+ uptake, but can mediate K+ extrusion at very low extracellular K+ concentrations (Véry et al., 1995; Brüggemann et al., 1999). In intact plants of V. faba, the K+ concentration in the guard cell wall has been estimated at between 3 and 40 mm (Felle et al., 2000; Roelfsema & Hedrich, 2002). The outward K+ channel of guard cells in intact V. faba becomes active at potentials close to −50 mV, while the inward rectifier activates at potentials negative of −100 mV (Roelfsema et al., 2001). In intact plants of A. thaliana (Fig. 3a,b) and Populus (Langer et al., 2004) the outward K+ channels also activate at −50 mV, but the threshold potential of inward K+ channels is more negative than −100 mV. A species-specific threshold potential for activation was not found by using patch-clamp recordings (Dietrich et al., 1998), indicating that the threshold potential is set by cytoplasmic factors that are lost during the whole-cell configuration.

The first two plant K+ channels were cloned from Arabidopsis, and denoted KAT1 (Anderson et al., 1992) and AKT1 (Sentenac et al., 1992); this was followed by the identification of the first KAT1 homolog, KST1, from potato (Müller-Röber et al., 1995). All these genes encode inward-rectifying K+ channels when expressed in Xenopus oocytes (Schachtman et al., 1992; Müller-Röber et al., 1995; Véry et al., 1995) or yeast cells (Bertl et al., 1995). These and later studies revealed that all members of the ‘Shaker like’ K+ channels are localized in the plasma membrane and function as K+-selective ion channels (Dietrich et al., 2001; Véry & Sentenac, 2002). KST1 and KAT1 are highly expressed in guard cells (Müller-Röber et al., 1995; Nakamura et al., 1995) and therefore soon became known as guard cell K+-uptake channels. Mutants of A. thaliana, with an En-transposon inserted in the open reading frame of KAT1, were therefore expected to lack inward K+ channels. However, recordings with intracellular microelectrodes revealed no difference between inward K+ currents of the KAT1 En-insertion mutant and wild type (Szyroki et al., 2001). Furthermore, these loss-of-function mutants displayed normal K+-uptake kinetics and stomatal opening. This led to a search for other K+-channel genes expressed in guard cells, which could back up for the loss of KAT1. Indeed, guard cells were found to express the K+-channel genes KAT2 (Pilot et al., 2001), AKT1, AKT2/3 and AtKC1, although at lower expression levels than that of KAT1 (Szyroki et al., 2001). Functional K+ channels probably assemble four subunits, encoded by one or more of these genes, to form a homomeric or a hetromeric complex (Dreyer et al., 1997). This might explain why the introduction of a nonfunctional KAT1 gene lowered the inward K+ conductance of guard cells by 70% (Kwak et al., 2001). Nonfunctional KAT1 subunits may form tetrameric channel proteins in combination with all five wild-type genes, thereby creating nonfunctional K+ channels.

The properties of inward K+ channels are thus not encoded by a single gene, but are probably the result of different properties encoded by at least five genes (Szyroki et al., 2001). The subunits KAT1, KAT2 and AKT1 all encode inward-rectifying K+ channels in heterologous expression systems. The fourth gene, AtKC1, probably does not encode a functional K+ channel itself, but its expression can alter the properties of K+ channels formed by the other subunits (Reintanz et al., 2002). The fifth gene expressed in guard cells, AKT2/3, encodes a largely voltage-independent channel when expressed in Xenopus oocytes (Marten et al., 1999; Lacombe et al., 2000). This channel probably plays an important role in the phloem of the shoot (Marten et al., 1999), but not in roots (Birnbaum et al., 2003). Voltage-independent K+ channels have not been recognized for guard cells, indicating that the properties the AKT2/3 subunits may change through interaction with other K+-channel subunits (Ivashikina et al., 2003), or as a result of phosphorylation (Chérel et al., 2002).

In contrast to the inward-rectifying K+ channels, outward-rectifying K+ channels in the guard cells of A. thaliana are encoded by a single gene only, named GORK (Ache et al., 2000). Disruption of the GORK gene causes a complete loss of the outward-rectifying K+ channels in the plasma membrane of guard cells (Hosy et al., 2003). The stomata of the GORK loss-of-function mutant close more slowly after treatment with ABA than do wild-type stomata (Hosy et al., 2003). However, the fact that stomata still close in response to ABA and darkness in gork-1, indicates that guard cells possess an alternative transport system capable of releasing K+ during stomatal closure.

Anion transport   The efflux of K+ from guard cells during stomatal closure is electrically neutralized by a concomitant efflux of anions. Anions pass through the guard cell membrane via channels that conduct a range of small anions (Hedrich & Marten, 1993; Schmidt & Schroeder, 1994). Two types of channels conducting anions have been identified in the guard cell membrane: rapid (R)-type (Keller et al., 1989; Hedrich et al., 1990); and slow (S)-type (Schroeder & Hagiwara, 1989; Linder & Raschke, 1992) anion channels (Fig. 4). The most obvious difference between these channels is a strong decrease in the open probability of R-type channels when cells are clamped to potentials negative of −50 mV and a slow inactivation after returning to more positive potentials (Kolb et al., 1995), while S-type channels exhibit only a weak voltage-dependent deactivation (Linder & Raschke, 1992; Schroeder & Keller, 1992; Dietrich & Hedrich, 1994).

Figure 4.

R- and S-type anion channels in the guard cell plasma membrane. (a) Current traces of R-type anion channels of a far depolarized guard cell in an intact Vicia faba plant (Roelfsema et al., 2001). The currents were elicited from a holding potential of −140 mV to test potentials ranging from −120 to 80 mV at 20-mV increments. (b) Current traces of S-type anion channels of a V. faba guard cell in an epidermal strip bathed in 5 mm CsCl, 1 mm CaCl2 and 1 mm Mes/BTP, pH 6.0. The currents were elicited from a holding potential of 20 mV to test potentials ranging from 0 to −180 mV, at −20-mV increments. Note the slow deactivation of S-type anion channels at the most negative membrane potentials. (c) Current–voltage relationship of the same cell as in (a) sampled during the last 100 ms of the test pulses. Note the maximum conductance at −80 mV. (d) Current–voltage relationship of the same cell as in (b) sampled during the last 500 ms of the test pulses. The reversal potential at 0 mV indicates an incomplete block of outward-rectifying K+ channels under the conditions applied.

Guard cells in intact plants often maintain membrane potentials negative of −100 mV (Roelfsema et al., 2001; 2002) and the difference in gating between S-type and R-type anion channels may thus be of physiological importance. Anion-channel activation will normally depolarize the guard cell plasma membrane (see II. 3). However, in cells with a membrane potential negative of −120 mV, a depolarization cannot be initiated by R-type channels. Indeed, we have observed guard cells in intact plants with a negative membrane potential and a high activity of R-type channels (own unpublished results). In these cells a depolarization can be initiated only through changes in the voltage dependence of R-type channels or through the activation of S-type anion channels. A change in the voltage dependence of R-type anion channels can be caused by anions at the extracellular side of the channels. Organic anions, such as malate, acetate and proprionate, shift the half-maximal activation potential of R-type channels to more negative values (Hedrich & Marten, 1993; Dietrich & Hedrich, 1998) and thus increase the potency of the channel to initiate a depolarization. The effect of malate is probably the most significant, as this anion accumulates at high concentrations in guard cells during stomatal opening (Outlaw & Lowry, 1977; Raschke & Schnabl, 1978). Anion channels are permeable to malate (Hedrich et al., 1994; Schmidt & Schroeder, 1994) and thus will extrude this organic anion during stomatal closure. The apoplastic concentration of malate was found to rise in response to high CO2 concentrations (Hedrich et al., 1994) and during prolonged illumination (Lohaus et al., 2001).

Apart from a difference in gating characteristics, R- and S-type channels also differ in their sensitivity to anion-channel blockers. R-type channels are blocked by stilbene derivates, such as DIDS (4,4′-diisothiocyanatostilbene-2,2′-disulfonic acid; Hedrich & Marten, 1993), while S-type channels are insensitive to DIDS (Schwartz et al., 1995, own unpublished results). S-type channels are blocked by NPPB (5-nitro-2-(3-phenylpropylamio)benzoic acid), 9-AC (anthracene-9-carboxylic acid) and niflumic acid (Schroeder et al., 1993; Schwartz et al., 1995), but these blockers are not specific because they also inhibit R-type anion channels (Marten et al., 1992) and outward-rectifying K+ channels (Garrill et al., 1996).

Apart from the difference in gating properties and their sensitivity to stilbene derivatives, R- and S-type anion channels have many features in common. Their single-channel conductance and ion selectivity are similar, which indicates that both channel types have a common pore structure or are just two gating modes of a single protein (Dietrich & Hedrich, 1994). Both channel types were also found in A. thaliana guard cells (Pei et al., 1997; Pei et al., 2000); however, the genes encoding these plasma-membrane anion channels are still unknown. Several genes homologous to animal ClC channels were found in the Arabidopsis genome, but these genes probably encode anion channels of intracellular membranes (Geelen et al., 2000; Lurin et al., 2000). In agreement with these results, ClC channels were not identified in purified plasma membranes of cell-suspension cultures of A. thaliana (Marmagne et al. 2004). Instead, voltage-dependent anion channel (VDAC) proteins were present in the plasma membrane, an anion channel normally associated with the mitochondria. Future experiments need to provide unequivocal evidence for VDAC proteins as genuine plasma-membrane anion channels that harbor the properties of R- and S-type anion channels.

Compared to the anion-release system of the guard cell membrane, little is known about the anion-uptake transporters. During stomatal opening guard cells accumulate anions, which electrically neutralize the uptake of K+. Anion uptake is a flexible process that depends on the extracellular presence of anions. As mentioned in II. 1, guard cells accumulate Cl and malate in equal concentrations when stomata, in epidermal strips, are supplied with Cl (Raschke & Schnabl, 1978). However, if no plasma membrane-permeable anions are applied, K+ uptake is completely balanced by malate synthesis (Raschke & Schnabl, 1978). Other anions, such as NO3 can also support stomatal opening (Guo et al., 2003). Arabidopsis mutants, with reduced levels of the NO3 transporter, CHL1, displayed smaller stomatal apertures than wild type, when NO3 was supplied to detached leaves (Guo et al., 2003). Stomatal opening, however, was similar to that of wild-type Arabidopsis in the presence of Cl, indicating that the CHL1 transporter enables the uptake of NO3 into guard cells, but does not represent the Cl uptake system. Most likely, guard cells also possess a carrier that couples the uptake of Cl with that of H+, but this transporter still remains to be identified.

Ca2+ transport   The interaction between extracellular K+ and Ca2+ on stomatal movement has been studied extensively (see reviews Raschke, 1975a; MacRobbie, 1987). In general, high extracellular K+ and Ca2+ concentrations act antagonistically, as stomatal opening is stimulated by K+ and inhibited by Ca2+. The sensitivity to extracellular Ca2+, however, differs between species. Whereas the presence of 0.25 mm extracellular Ca2+ reduced stomatal opening in C. communis by > 50% (DeSilva et al., 1985a), it reduced stomatal apertures to < 25% in V. faba (Fischer, 1972) and A. thaliana (Roelfsema & Prins, 1995). The latter data reflect an effect of Ca2+ at an extracellular K+ concentration of 50 mm; at lower K+ concentrations the Ca2+ effects become even more apparent (Fischer, 1972). Extracellular Ca2+ probably inhibits stomatal opening by evoking increases in the intracellular Ca2+ concentration (McAinsh et al., 1995). It has long been thought that high external Ca2+ concentrations would enhance leak Ca2+ currents across the plasma membrane. Such leak currents of Ca2+ would be supported by the high concentration gradient for Ca2+ across the plasma membrane. The apoplastic Ca2+ concentration is ≈ 100 µm (Felle et al., 2000; Roelfsema & Hedrich, 2002), while the cytoplasmic concentration ranges from 100 to 300 nm (Webb et al., 2001; Levchenko et al., 2005). Recently, it was shown that guard cells do not simply leak Ca2+ but instead possess a plasma-membrane Ca2+ sensor (CAS) (Han et al., 2003). The CAS sensor triggers a rise in the intracellular Ca2+ concentration, in response to an increase in the extracellular Ca2+ concentration. Guard cells expressing antisense constructs of CAS were no longer responsive to a rise in the extracellular Ca2+ concentration.

Calcium may enter guard cells via nonselective Ca2+ channels in the plasma membrane. These nonselective channels have been found in a large number of plant cells and are, in general, permeable to monovalent and divalent cations (Demidchik et al., 2002). In guard cell protoplasts, Ca2+-permeable channels were identified that are activated by stretch (Cosgrove & Hedrich, 1991) or ABA (Schroeder & Hagiwara, 1990). Nonselective cation channels that activate upon hyperpolarization were suggested by Lohse & Hedrich (1992) (Fig. 5b,c). The physiological relevance of these channels in guard cells became apparent after Grabov & Blatt (1998) showed hyperpolarization-induced rises in the cytoplasmic Ca2+ concentration. Subsequent studies showed that these channels are stimulated by 50 µm H2O2 in A. thaliana (Pei et al., 2000) and through phosphorylation and ABA in V. faba (Hamilton et al. 2000; Köhler & Blatt, 2002).

Figure 5.

Electrical properties of Vicia faba guard cell protoplasts after elimination of the plasma membrane K+ and anion currents, redrawn after Lohse & Hedrich (1992). (a) Current–voltage relationship of the plasma membrane H+-ATPase measured in the whole-cell patch clamp mode. A voltage ramp was applied from −240 to 10 mV. (b) Voltage relationship of whole-cell currents, with ATP in the pipette and in the absence (white symbols) or presence (black symbols) of 5 mm LaCl3 in the bath solution. (c) Voltage relationship of whole-cell currents remaining after ATP depletion in the absence of LaCl3.

The Ca2+ homeostasis in plant cells is maintained by Ca2+ transporters in the plasma membrane, vacuolar membrane and endoplasmic reticulum (ER) (Sanders et al., 2002). The plasma membrane harbors P-type ATPases (ACAs) that can transport Ca2+ against steep concentration gradients. These Ca2+ ATPases are characterized by an autoinhibitory domain at the N-terminus that binds Ca2+/calmodulin (Sze et al., 2000; Axelsen & Palmgren, 2001). At high cytoplasmic Ca2+ concentrations the N-terminus can be released from the core protein, thereby activating the Ca2+ pump. The latter mechanism provides a negative-feedback system that automatically counteracts a rise in cytoplasmic Ca2+. Ca2+ regulates the ACA pumps also via an additional pathway involving Ca2+-dependent protein kinases (CDPK) (Sze et al., 2000). Genes encoding Ca2+-permeable channels in the guard cell plasma membrane have not yet been cloned (Hetherington & Brownlee, 2004), but they may be homologous to the ionotropic glutamate receptors (Lacombe et al., 2001).

H+ transport   The guard cell membrane is energized by a P-type ATPase that translocates H+ from the cytoplasm to the guard cell wall (Fig. 5a). This proton pump is the only primary active transporter in the plasma membrane, apart from a number of P-type ATPases that transport Ca2+ or heavy metals (Axelsen & Palmgren, 2001). The guard cell H+-ATPase has a low activity at pH 7.5 and becomes more active at acidic pH values (Becker et al., 1993). This pH dependence indicates a role for the proton pump in maintaining H+ homeostasis. Guard cells maintain a cytoplasmic pH of 7.5–7.8 (Blatt & Armstrong, 1993; Grabov & Blatt, 1997); a shift to more acidic values thus automatically activates the plasma membrane H+-ATPase. In addition, H+-ATPases are regulated through an autoinhibitory domain at the C-terminus. This domain is released from the catalytic site, after binding of a 14-3-3 protein, based on phosphorylation-dependent (Würtele et al., 2003), as well as phosphorylation-independent, mechanisms (Fuglsang et al., 2003). Binding of the phosphorylated C-terminus to 14-3-3 proteins is stabilized by fusicoccin, a fungal toxin that hyperstimulates stomatal opening (Braunsgaard et al., 1998; Kinoshita & Shimazaki, 1999).

Vacuolar membrane transport  A proportion of the ions that accumulate in the guard cell during stomatal opening will remain in the cytoplasm, while the rest is sequestered in the vacuoles. Guard cells posses several small vacuoles that may be interconnected to form a reticulum (Palevitz et al., 1981; Faraday et al., 1982). Swollen guard cells contain only few but large vacuoles that shrink and become fragmented during stomatal closure (Diekmann et al., 1993). However, a detailed study of changes in vacuole size during stomatal movement has not been conducted. A vacuolar sequestration is of importance for ions, such as Ca2+, that also act as second messengers. Vacuoles are an important intracellular store of Ca2+ and are probably involved in the release of Ca2+ during signaling events in guard cells (MacRobbie, 1998; Sanders et al., 2002; Hetherington & Brownlee, 2004). Furthermore, sequestration of malate is important, as this anion negatively feeds back on carbon fixation, by inhibition of phospho(enol)pyruvate (PEP) carboxylase (Schnabl & Kottmeier, 1984b; Raschke et al., 1988). Indeed, most of the malate in guard cell protoplasts of V. faba is located in the vacuole (Schnabl & Kottmeier, 1984a).

Although a number of channels, carriers and pumps at the vacuolar membrane have been identified (Fig. 3), the regulation of vacuolar membrane transport is poorly understood. In general, transport activities were studied using isolated vacuoles, devoid of cytoplasmic components that regulate these transporters in intact cells. Recording techniques that measure the activity of ion transporters at more physiological conditions are technically demanding, but will be essential for a better understanding of ion transport regulation in the future. Likewise, little is known about the membrane potential of the guard cell vacuole. A potential difference of 20–50 mV, negative at the cytoplasmic site, is often assumed (MacRobbie, 1998), but no direct recordings of vacuoles in intact guard cells exist. Furthermore, the vacuolar membrane potential presumably changes during stomatal movements, but the extent to which changes in this potential contribute to ion release remains unknown.

K+ transport   Three types of K+-permeable channels have been determined in the vacuolar membrane of guard cells (Allen et al., 1998). Based on differences in gating characteristics, these channels were classified as slow vacuolar (SV), fast vacuolar (FV) (Hedrich et al., 1986; Hedrich & Neher, 1987; Schulz-Lessdorf & Hedrich, 1995; Allen & Sanders, 1996) and vacuolar K+-selective (VK) channels (Ward & Schroeder, 1994; Allen & Sanders, 1996). SV channels represent the most prominent ion conductance in the vacuolar membrane and slowly activate at depolarizing potentials (more positive at the cytoplasmic side), conditions that allow K+ release from the vacuole. However, at large depolarization, the K+ flux reverses and K+ is transported into the vacuole via SV channels (Ivashikina & Hedrich, 2005). FV channels are similar in this respect, as they also activate with depolarization, but their activation is much faster. The Ca2+ dependence of FV and SV channels differs – FV are inhibited by Ca2+ concentrations of > 0.1 µm (Hedrich et al., 1986; Allen & Sanders, 1996), whereas SV channels are inactive at low concentrations of Ca2+, but are activated by Ca2+ concentrations of > 0.5 µm (Hedrich & Neher, 1987). Both channels are cation permeable and thus may enable K+ release, probably under different cytoplasmic conditions. VK channels represent the third type of K+-selective channel in the vacuole (Ward & Schroeder, 1994). These channels are nonrectifying and thus appear as K+-selective leaks that may enable K+ release or uptake at conditions that do not support the activation of SV- or FV channels (Allen et al., 1998).

Vacuolar K+ channels may be encoded by the KCO (TPK) genes, as all of these channels, apart from TPK4, are targeted to the vacuolar membrane (Czempinski et al., 2002; D. Becker et al., unpublished). The KCO channels have been renamed TPK (which stands for two pore domain K+ channels) (Becker et al., 2004), with the exception of KCO3, which has only one pore domain. The pore domains of the TPK channel and KCO3 contain the GYGD motif, suggesting that they encode ion channels with a K+-selective conductance (Czempinski et al., 1999). This was recently proven for TPK4, the sole Arabidopsis TPK channel located in the plasma membrane (Becker et al., 2004). Whether or not this is also the case for vacuolar TPK channels is currently under investigation.

K+ ions are taken up into vacuoles against the vacuolar membrane potential, the transport therefore is probably mediated by secondary active carriers. K+ uptake may be carried out by the Na+/H+ (NHX)-transporters, well known for their ability to transport Na+ (Gaxiola et al., 1999), but which also mediate H+-coupled K+ transport (Venema et al., 2003; Cellier et al., 2004). In addition, less well-characterized transporters are found in the Arabidopsis genome; these transporters have high sequence similarity to known H+/K+ carriers (Sze et al., 2004).

Anion transport   The vacuolar membrane potential supports a flow of anions from the cytoplasm into the vacuole through anion channels. For guard cell vacuoles, only a single publication has reported on anion channels (Pei et al., 1996). This anion channel was reported to be activated by CDPK and to facilitate the flow of Cl and malate2– into vacuoles. For other cell types, voltage-dependent anion channels have been described that open at negative membrane potentials, indicating that they indeed support anion sequestration into vacuoles (Pantoja & Smith, 2002; Hafke et al., 2003). This may be important for the sequestration of malate, as the cytoplasmic concentration of this anion has to be kept low in order to prevent the inhibition of PEPcarboxylase (Raschke et al., 1988; Schnabl & Kottmeier, 1984b). For malate, the uptake process into the vacuole is complex, as malate has two negative charges at neutral pH, but may lose one upon entry into the acidic vacuole. Monovalent malate anions could either be trapped in the vacuole, or flow back into the cytoplasm, depending on the selectivity of anion channels. Alternatively, malate may be transported into the vacuole by carriers, such as the malate carrier, AttDT (Emmerlich et al., 2003).

Anions can also be released via anion channels, but this requires depolarization of the vacuolar membrane. If the membrane potential would approximate 0 mV, anions would flow down their concentration gradient into the cytoplasm. Vacuolar anion channels may be encoded by genes homologous to the bacterial and animal ClC channels. Disruption of the AtClC-a gene resulted in plants that accumulate less NO3 when grown at high levels of NO3 (Geelen et al., 2000). However, the activity of ClC channels, in general, and their role in the vacuolar membrane, in particular, still await confirmation by direct measurements. Patch clamp studies, on isolated vacuoles of plants overexpressing a CLC channel or lacking one, may help to establish the function of these genes.

Ca2+ transport   The large vacuole of guard cells plays an important role in Ca2+ homeostasis (Sanders et al., 2002). Although the Ca2+ concentration of guard cell vacuoles has not been studied in detail, it is generally assumed that it is in the mm range, as recorded for rhizoid cells of Riccia fluitants and root cells of Z. mays (Felle, 1988). Ca2+ accumulates in the vacuoles as a result of the activity of ATP-dependent ACA pumps (Sze et al., 2000; Axelsen & Palmgren, 2001) as well as of the CAX-encoded Ca2+/H+ antiporters (Hirschi et al., 1996). The ACA pumps are regulated by Ca2+/calmodulin, as are their counterparts at the plasma membrane, while the regulation of the CAX carriers seems to be more indirect and involves several CXIP proteins that closely interact with CAX1 and -4 (Cheng & Hirschi, 2003).

In contrast to the limited information available about Ca2+ sequestration into the vacuoles of guard cells, many reports have focused on Ca2+-release channels. Because of the steep concentration gradient, the opening of Ca2+-permeable channels in the vacuolar membrane results in a Ca2+ flux to the cytoplasm. This, in turn, leads to a rise in the cytoplasmic Ca2+ concentration, which may encode a signal for downstream responses (Evans & Hetherington, 2001; Sanders et al., 2002). Based on patch clamp recordings, plant vacuolar membranes seem to harbor a number of ligand-gated Ca2+-permeable channels. In beet roots, vacuolar membrane Ca2+ currents are induced by inositol triphosphate (IP3) and cyclic-ADP ribose (Allen et al., 1995), while guard cell vacuolar membranes have also been shown to be sensitive for cyclic-ADP ribose (Leckie et al., 1998) and inositol hexakis-phosphate (IP6) (Lemtiri-Chlieh et al., 2003). In addition to these ligand-gated Ca2+ channels, a voltage-gated Ca2+ channel (VVCa-Channel) has been described for the guard cell vacuolar membrane (Allen & Sanders, 1994).

The vacuolar SV channel, initially described as a K+ channel (Hedrich et al., 1986), has been suggested to function as a Ca2+-induced Ca2+-release (CICR) channel (Ward & Schroeder, 1994; Allen & Sanders, 1995; Pei et al., 1999; Sanders et al., 2002). The latter role of the SV-channels was investigated by Pottosin et al. (1997), who found that the SV channel was blocked by luminal Ca2+. Later studies of Pottosin et al., 2004) and Ivashikina & Hedrich (2005) also disagree with a role of SV-channels in Ca2+-induced Ca2+ release.

The single-copy gene, AtTPC (Furuichi et al., 2001), is likely to encode one of the Ca2+ channels in the vacuolar membrane, as its gene product was localized to this membrane by using a proteomics approach (Carter et al., 2004). Genes homologous to the animal IP3- and ryanodine-receptors have not been identified in the Arabidopsis genome (Schwacke et al., 2003). Apparently, the genes encoding these receptors in plants have developed independently from the animal counterparts, or have diverged considerably from them during evolution.

H+ transport   Two types of H+-translocating pumps – V-type H+-ATPases (V-ATPases) and pyrophosphatases (PPases) – co-reside in the same vacuolar membrane (Hedrich et al., 1986; 1989; Sze et al., 1999). The V-type ATPase is a protein complex similar to the F1/F0-ATPases and consists of a large number of different subunits (Sze et al., 2002; Kluge et al., 2003). The V1 subcomplex catalyzes ATP hydrolysis and rotates, during H+ translocation, with respect to the V0 subcomplex in the membrane. The stoichiometry of these pumps is variable: values were determined ranging from more than 3 (Davies et al., 1994) to 1 H+ per hydrolyzed ATP (Müller & Taiz, 2002). A low stoichiometry is probably important for very acidic vacuoles because a high proton motive force can inverse the proton flux through V-type ATPases (Gambale et al., 1994). For guard cell vacuoles the stoichiometry may be higher, as these compartments contain high concentrations of K+ and thus may not have such a low pH. V-type ATPases are probably regulated by the redox state as well as through changes in the cytoplasmic nucleotide concentration (Kluge et al., 2003), but no information is available about this type of regulation in guard cells. The importance of these transporters in guard cell physiology is evident from studies on the det3 mutant (Schumacher et al., 1999). These mutants are affected in subunit C of the V-ATPase and display only 70% of the V-ATPase activity compared to wild-type cells. Guard cells of det3 did not close in response to an increase in the extracellular Ca2+ concentration or to the application of H2O2 (Allen et al., 2000). The absence of these responses was therefore linked to an altered Ca2+ homeostasis in det3 guard cells.

In addition to the V-ATPases, PPases in Arabidopsis contribute to the generation of the electrochemical gradient across vacuolar membranes (Drozdowicz & Rea, 2001). These relatively simple proteins use the hydrolysis of PPi to pump H+ into the vacuolar lumen. The activity of these pumps is highest in young growing tissues, in which PPi is available as a by-product of biosynthesis (Maeshima, 2000). Because the synthetic rate of macromolecules is probably less pronounced in mature guard cells, it is unlikely that PPases play a major role in energizing the guard cell vacuolar membrane.

3. Coupling of ion fluxes

Transport of osmotically active compounds across the plasma- and vacuolar membranes is driven by H+-translocating primary transporters. Although both plasma- and vacuolar membranes also posses Ca2+-ATPases, the latter pumps probably serve to compensate for Ca2+ influx. This means that guard cell transport of solutes against their concentration gradient is predominantly fueled by H+-ATPases (Fig. 6). The electrochemical gradient generated (proton motive force) can be decomposed in an electrical potential and in a H+-concentration difference, both of which are used to drive solute flow.

Figure 6.

Coupling of ion-fluxes during stomatal movements. During stomatal opening an electrochemical gradient across the plasma membrane is maintained by active H+-ATPases (1). Based on the hyperpolarized membrane potential (−110 mV), K+ can be taken up via inward-rectifying K+ channels (2). Cl may be taken up from the apoplast by symport with H+ (3), while malate is synthesized from phosphoenolpyruvate and CO2 (4). V-type ATPases maintain the electrochemical gradient across the vacuolar membrane (5). Cl can be transported along the electrical potential (≈ −40 mV) into the vacuole via anion channels (6). A malate carrier may be involved in maintaining low cytoplasmic malate levels (7), while K+ is probably taken up against the electrical potential, by an H+-driven antiporter (8). Stomatal closure is associated with efflux of K+ via channels (9), presumably causing a depolarization of the vacuolar membrane (Em ≈ 0 mV). In the event that the vacuolar membrane depolarizes, Cl may be extruded via an anion channel (10). Activation of plasma-membrane anion channels enables an anion efflux along the electrical potential (11) and causes a depolarization (−50 mV). As a result of the depolarized plasma membrane, K+ can be released via outward-rectifying K+ channels (12).

Stomatal opening  During stomatal opening, the plasma membrane potential of V. faba guard cells in intact plants is ≈−110 mV (Fig. 7a). The reversal potential of K+ is close to −75 mV and the membrane potential thus can drive the uptake of potassium. K+ uptake is facilitated by inward-rectifying K+ channels, which activate at membrane potentials negative of −100 mV (Roelfsema et al., 2001). For each K+ taken up, an H+ ion has to be extruded in order to keep the membrane electrically charged. The H+ gradient can drive the uptake of anions, such as NO3 or Cl, by an H+-coupled symport system (MacRobbie, 1987). The stoichiometry of the latter transporter is not known for guard cells, but has been found to range between 1 and 2 for other cell types (Sanders, 1980; Felle, 1994). In addition to carrier-mediated anion uptake, guard cells synthesize considerable amounts of malate (Raschke et al., 1988). Most malate will be stored in the acidic vacuole and thus will have a single negative charge. Synthesis of one malate anion and its transport to the vacuole will thus result in a surplus of one H+ ion. The H+ ion will be extruded from the guard cell to keep the cytoplasmic pH constant and enable the uptake of one K+ ion. This process is as efficient as K+ uptake in combination with Cl, provided that the Cl/H+ carrier has a stoichiometry of 1. If the stoichiometry of the latter carrier is > 1, uptake of K+ in combination with Cl would thus require higher H+-extrusion rates at the plasma membrane than with malate synthesis.

Figure 7.

Light- and CO2-induced changes in the electrical properties of Vicia faba guard cells in intact plants. (a) Guard cell membrane potential hyperpolarizing in the light and depolarizing in the dark. The bar below the graph indicates the time-points at which the microscope lamp was switched on and off. (b) Guard cell hyperpolarizing in the absence of atmospheric CO2 and depolarizing again with 700 µl l−1 of CO2. The bar below the graph indicates when the CO2 supply was shut off (white area) and on (striped area). From Roelfsema et al. (2002). (c) Activation of an inward current measured in a guard cell clamped constantly to −100 mV after supplying CO2, as indicated by the bar below the graph (white area: 0 µl l−1 CO2, striped area: 700 µl l−1 CO2).

As an alternative to ion uptake, guard cells may accumulate sugars. Guard cell protoplasts of P. sativum take up ≈ 15 fmol of glucose per cell h−1 and, at lower rates, also fructose and glucose (Ritte et al., 1999). These molecules are probably imported via H+-coupled transport by carriers, such as the hexose transporter, STP1 (Stadler et al., 2003). The stoichiometry of the sucrose transporter, SUC1, was found to be one sucrose molecule per H+ (Carpaneto et al., 2005); if this stoichiometry would apply also for other sugar transporters, guard cells can take up one sugar molecule per H+ ion extruded. Sugar uptake thus seems to be less efficient than K+ uptake combined with malate accumulation.

As discussed in section II. 2, most of the solutes taken up during stomatal opening will be sequestered in the guard cell vacuole (Fig. 6). The vacuolar membrane potential is negative at the cytoplasmic side, as for the plasma membrane, but now all solutes are to be transported in the opposite direction. Anions can be transported via ion channels as long as the concentration difference between the cytoplasm and the vacuole is compensated by the membrane potential difference. Provided that the vacuolar membrane potential is −40 mV during stomatal opening, the concentration of monovalent anions in the vacuole should be five times lower than in the cytoplasm. If the anion concentration difference exceeds this value, anion transport into the vacuole is probably coupled to a flow of H+ ions to the cytoplasm. K+ ions are transported into the vacuole against the potential difference and thus are probably sequestered into the vacuole by an H+-coupled antiport mechanism. Finally, sugars may be taken up into the vacuole by a recently identified subfamily of sugar transporters localized in this endomembrane and homologous to ERD6 (M. Büttner, pers. comm.).

Stomatal closure  In V. faba, stomatal closure can occur very rapidly compared to stomatal opening. Whereas the transpiration rate increases with a velocity of 0.7 µmol m−2 s−2 in the light, it decreases with a velocity of 1.9 µmol m−2 s−2 after the application of ABA (Langer et al., 2004). This difference in velocity probably correlates with the transport mechanisms involved in both processes. Whereas opening is based on active transport, closure is based on the release of solutes along their concentration gradients (Fig. 6). Stomatal closure therefore may be dependent entirely on the activation of ion channels in the vacuolar and plasma membranes. K+ ions will flow from the vacuole to the cytoplasm when K+ channels in the vacuolar membrane are activated. The activation of K+ channels will also depolarize the vacuolar membrane and provide a driving force for anion efflux. The anions will thus be extruded from the vacuole after the activation of anion-permeable channels. However, it is unclear whether the events described above really occur during stomatal closure, as neither the activation of K+ and anion channels, nor the concurrent depolarization of the vacuolar membrane, have, to date, been recorded.

During dark-induced stomatal closure, the plasma membrane of V. faba guard cells depolarizes slowly to ≈ −50 mV (Roelfsema et al., 2001; Fig. 7a). This value is positive of the K+-reversal potential and thus supports the efflux of K+ ions via outward-rectifying K+ channels. The dark-induced depolarization is probably brought about by the deactivation of H+ ATPases in combination with the activation of S-type anion channels (Roelfsema et al., 2001; 2002). An efflux of K+ via plasma-membrane channels is thus accompanied by anion extrusion.

It is unclear how sugars are released from the vacuole and cytoplasm of guard cells. In principle this could occur via sugar-permeable porins but, in Arabidopsis, porins are only associated with mitochondrial and plastid membranes (Schwacke et al., 2003; Clausen et al., 2004). Alternatively, sugars may be released by transporters encoded by genes homologous to sugar carriers; such carriers may function in a reverse mode or function in a mode uncoupled from H+ translocation.

III. Regulation of ion transport in guard cells

Because of the close connection between ion transport and changes in the osmotic content of guard cells, stomatal movement can be traced back to guard cell ion transport. Regulation of ion transport in guard cells thus directly affects the leaf conductance, an important determinant for plant growth. Nevertheless, guard cells seem to respond more or less autonomously to environmental signals. This is exemplified by the responses of stomata, in isolated epidermal strips, to signals such as light, ABA and CO2. Although the latter responses mimic the behavior of stomata in intact plants, it has long been recognized that the responses measured in epidermal strips are different from those measured on intact leaves (Willmer & Mansfield, 1969). For instance, light induces an increase in the transpiration rate of 0.7 µmol m−2 s−2 (Langer et al., 2004) in V. faba leaves, which correlates to an increase in the stomatal aperture of ≈ 6.5 µm h−1 (Kappen et al., 1987). A much slower rate of opening, of ≈ 2.5 µm h−1, was found for stomata in epidermal strips (Schwartz & Zeiger, 1984). Likewise, stomatal closure induced by ABA also occurs more slowly in epidermal strips than in intact leaves. Whereas stomata in intact leaves close with a rate of ≈ 18 µm h−1, based on the transpiration measurements (Langer et al., 2004), or with a rate of ≈ 21 µm h−1, as determined by microscopical observations (Roelfsema et al., 2001), a velocity of ≈ 8 µm h−1 was found with epidermal strips (Blatt & Armstrong, 1993). This indicates that guard cells, to some extent, lose their responsiveness after isolation, which may be caused by slow dedifferentiation in the absence of signals derived from viable neighboring cells. Note that the guard cells of tobacco and sugar beet can regenerate complete new plants (Sahgal et al., 1994; Hall et al., 1996). Apparently, guard cells in intact leaves are embedded in a niche, which retains them in a highly specialized state. The way in which mature guard cells are influenced by other cells within the leaf is not known in detail, but signal transduction pathways causing epidermal leaf cells to develop into guard cells are probably involved (Bergmann, 2004; Larkin et al., 2003; Bergmann et al., 2004).

The loss of guard cell responsiveness in isolated epidermal strips has led to the development of techniques to study guard cells in the intact plant (Felle et al., 2000; Roelfsema et al., 2001). As expected, guard cell responses recorded using the latter method differed from those observed with guard cells in isolated epidermal strips and protoplasts thereof. Below we will discuss the current knowledge about guard cell signal transduction and relate these to electrophysiological data and transpiration measurements. As mentioned in the first section of this review, the leaf conductance is mainly influenced by light, humidity and circadian rhythms. The latter regulatory mechanism has been recently reviewed in this series (Webb, 2003) and we will therefore only cover the effect of light and humidity.

1. Light-induced stomatal opening

The nature of the light receptor in guard cells has been long discussed. At the beginning of the 20th century, plant biologists realized that stomata open at low CO2 concentrations and postulated that light acts via a lowered CO2 concentration in the leaf (Stahl, 1920; Heath, 1959). This hypothesis seemed to fit action spectra obtained for stomatal opening, which are very similar to the absorption spectrum of chlorophyll (Kuiper, 1964). A closer examination, however, revealed that blue light was more effective than red light (Sharkey & Raschke, 1981). In fact, blue light can still stimulate stomatal opening when provided on top of saturating intensities of red light (Iino et al., 1985). This response to blue light can now be assigned to the activity of the PHOT1 and PHOT2 blue-light receptors located in the plasma membrane (Kinoshita et al., 2001; Sakamoto & Briggs, 2002). Stomata of the phot1 phot2 double mutant are no longer specifically responsive to blue light; however, they still open in response to photosynthetic-active radiation. Guard cells thus have two light-perception pathways acting parallel to each other: the PHOT1/PHOT2-mediated pathway, which is dependent on blue light, and a second pathway that is dependent on photosynthetic-active radiation (Assmann & Shimazaki, 1999; Roelfsema et al., 2002).

Blue light-dependent pathway  The PHOT receptors are receptor kinases associated with the plasma membrane that are activated by blue light (Briggs & Christie, 2002). The N-terminal half of the protein contains two LOV-domains that bind flavin mononucleotides (Christie et al., 1999) and act as light sensors, while the C-terminal half is a serine/threonine kinase. Blue light triggers autophosphorylation of PHOT receptors, which leads to the binding of 14-3-3 proteins (Kinoshita et al., 2003). Binding of the 14-3-3 protein may help to keep the PHOT receptors in an active state. A similar mechanism is found for the activation of plasma-membrane H+-ATPases, downstream in the blue light signal pathway. Here, the C-terminus forms an autoinhibitory domain that binds the catalytic site of the H+-ATPase (Braunsgaard et al., 1998; Fuglsang et al., 2003). Blue light leads to the phoshorylation of this domain and the binding of 14-3-3 proteins (Kinoshita & Shimazaki, 1999). This releases the C-terminus from the catalytic site and activates the H+-ATPase.

Although it is tempting to speculate that the PHOT receptors directly phosphorylate the H+-ATPase, this has yet to be shown. The signaling pathway seems to be more complicated, as PHOT1-induced stomatal opening depends on a functional RPT2 protein (Inada et al., 2004). Based on homology, the RTP2 protein presumably mediates protein–protein interactions. Furthermore, several lines of evidence indicate the involvement of changes in the cytoplasmic Ca2+ concentration of guard cells (Shimazaki et al., 1999). The blue light response of guard cells is inhibited by verapamil, a Ca2+-channel blocker (Shimazaki et al., 1997), and is stimulated by high concentrations of external Ca2+ (Roelfsema et al., 1998). In seedlings of Arabidopsis, blue light induces a rise in the cytoplasmic Ca2+ level, via a PHOT1-dependent pathway (Baum et al., 1999). Furthermore, PHOT1 and -2 stimulate the activity of hyperpolarization-activated Ca2+ channels in the plasma membrane of mesophyll cells (Stoelzle et al., 2003). Blue light-activated Ca2+ channels, however, have not yet been shown for guard cells, and the effect of blue light on the cytoplasmic Ca2+ concentration of guard cells is still unknown.

The blue light-induced activation of H+-ATPases can be recorded as extracellular acidification by guard cell protoplasts (Shimazaki et al., 1986) or guard cells in epidermal strips (Roelfsema et al., 1998). Alternatively, this process can be measured as a blue light-stimulated outward current across the guard cell plasma membrane (Assmann et al., 1985). The extrusion of H+ drives the uptake of K+ via inward-rectifying K+ channels, at a ratio of 1 : 1 (Section II, 3). In theory, the magnitude of the H+ current thus should be linked to the speed of the stomatal opening induced by blue light. Continuous blue light triggers an increase in the stomatal conductance of V. faba leaves of ≈ 0.08 (mol m−2 s−1) h−1 (Iino et al., 1985). Based on the results of Kappen et al. (1987), such a change in conductance is linked to an increase in stomatal opening of ≈ 4 µm h−1. For stomata in intact leaves this would require an increase in turgor pressure of ≈ 1.8 MPa h−1 (Table 2). In order to support such a change in the turgor pressure, guard cells should accumulate ≈ 1.4 pmol of K+ h−1 (Table 2), which requires an extrusion of ≈ 1.4 pmol of H+ h−1, correlating to an outward current of 36 pA. In guard cell protoplasts measured in the whole-cell configuration, blue light induces an outward current of ≈ 6 pA (Assmann et al., 1985; Schroeder, 1988; Taylor & Assmann, 2001). The current measured in these patch-clamp studies is thus much smaller, as would be expected from the rate of stomatal opening. Loss of cytoplasmic components in the whole-cell mode may explain the relatively small blue light-induced currents, as suggested by the larger current measured in slow-whole-cell recordings (Schroeder, 1988). In the guard cells of intact plants, the blue light-triggered current was even larger (Roelfsema et al., 2001); here, on average, an outward current of 60 pA was recorded 2 min after the onset of blue light. Continuous blue light triggered a transient response; the currents levelled off after reaching a peak value within 3 min (Fig. 8c).

Figure 8.

Abscisic acid (ABA) and blue light-triggered changes in the electrical properties of the plasma membrane of Vicia faba guard cells in intact plants. (a) Depolarization of a guard cell elicited by 10 µm ABA applied to an impaled guard cell via the leaf surface. The black bar below the graph indicates time of ABA application. From Roelfsema et al. (2004). (b) ABA-induced changes in the plasma membrane current of a guard cell constantly clamped to a holding potential of −100 mV. The black bar below the trace indicates the time-point of ABA application. From Roelfsema et al. (2004). Note that ABA triggers inward currents in two phases: in phase 1 a large (but transient) increase occurs, while a small (but steady) inward current is present in phase 2. (c) Change in plasma membrane current of a guard cell clamped to −100 mV, triggered by blue light given on a background of red light. The current trace is interrupted by short intervals, in which current-voltage profiles of the cell were recorded.

Photosynthetic-active radiation-dependent pathway  Apart from the blue light-specific pathway, guard cells are also sensitive to photosynthetic-active radiation (PAR). This is most obvious from the effect of DCMU (3(3,4-dichlorophenyl)-1-1-dimethylurea), an inhibitor of photosystem II, which inhibits red light-induced (Kuiper, 1964), but not blue light-induced, stomatal opening (Sharkey & Raschke, 1981; Schwartz & Zeiger, 1984). This response is often referred to as the red light response, but it can also be activated by blue light at high intensities. For this pathway, the old hypothesis, postulating that intercellular CO2 may be an intermediate signal, seems to apply. We found that guard cells in the intact plant did not respond to red light irradiation, given as a small beam directly on individual guard cells (Roelfsema et al., 2002). However, switching on and off red light over a large area on the leaf did alter the plasma membrane potential of guard cells. The broad beam of red light was found to lower the intercellular CO2 concentration, while a small beam of red light had no effect. CO2 thus probably functions as an intermediate signal in the PAR-dependent pathway.

It has long been debated to what extent guard cells can conduct photosynthesis (Outlaw, 1989). Although the chlorophyll content of guard cells is much lower than that of mesophyll cells, chlorophyll fluorescence recordings indicate that chloroplasts of guard cells and mesophyll cells do not differ much (Goh et al., 1999; Baker et al., 2001). Guard cells thus seem to operate photosynthetic electron transport and CO2 fixation (Shimazaki et al., 1982; Wu & Assmann, 1993), enabling them to lower their own CO2 concentration, to some extent. This may explain why guard cells in epidermal strips open in red light. The diffusion rate of CO2 in bath solutions covering the epidermal strip will be low, and guard cell photosynthesis may therefore have a significant effect on the intracellular CO2 concentration. In intact plants, however, guard cell photosynthesis will be less effective because guard cells have direct access to the atmosphere and the CO2 diffusion rate is much higher.

Guard cells in intact plants respond to high CO2 concentrations or the offset of red light, with the activation of anion channels, which causes depolarization of the plasma membrane (Brearley et al., 1997; Roelfsema et al., 2002; Fig. 7b). As a result of this depolarization, the flux of K+ across the plasma membrane reverses. Instead of K+ uptake via inward-rectifying channels in the absence of CO2, K+ is extruded via outward-rectifying channels in the presence of CO2. The activation of anion channels is thus sufficient to enable both the efflux of anions and K+. As a result of the loss of osmolites from guard cells, the stomata will close.

Various mechanisms have been suggested that couple a rise in the CO2 concentration to changes in the activity of plasma-membrane ion channels (reviewed in Vavasseur & Raghavendra, 2005). Hedrich et al. (1994) found that the apoplastic malate concentration rises at elevated concentrations of CO2 and, in turn, can activate R-type anion channels in guard cells. Similar observations were made by Raschke et al. (2003), using guard cells in epidermal strips. Alterations of ion-channel activity via elevated Ca2+ concentrations have been suggested by Webb et al. (1996), while a role for the intracellular pH could be excluded by Brearley et al. (1997). Finally, Goh et al. (2002) found that both photosynthesis and the activity of inward K+ channels are affected by cytoplasmic ATP (Goh et al., 2002). Photosynthesis and plasma membrane transport may thus be coordinated via changes in the cytoplasmic energy charge of guard cells.

2. Humidity and water status

Drought stress acts antagonistically to light on stomatal movement: it induces stomatal closure or inhibits opening to prevent excessive loss of water via the stomata. Stomata respond directly to changes in humidity and to ABA released from roots upon the onset of drought stress. The latter signal-transduction pathway has been studied extensively, while the direct response of guard cells to humidity has received much less attention. It has been postulated that ABA is released from mesophyll cells during drought and subsequently induces stomatal closure (Hartung et al., 1988). However, the finding that the stomata of ABA-insensitive mutants still respond to humidity (Assmann et al., 2000) points to a more direct response of guard cells to the air water content. Although guard cells are capable of responding to changes in air humidity, they seem poorly constructed to monitor changes in the plant water status directly (Raschke, 1987). Changes in the latter will only cause small changes in the guard cell volume because of the high osmotic pressure in these cells. Lowering the relative humidity of the air will affect the turgor pressure in guard cells, as well as in epidermal pavement cells. Because of the mechanical advantage of epidermal cells (Sharpe et al., 1987), stomata will initially open after a drop in air humidity, but stomatal closure sets in upon prolonged exposure times (Kappen et al., 1987; Mott et al., 1997). Closure of stomata, observed at low air humidity, apparently is not a simple consequence of turgor loss, but is a guard cell-controlled process.

Abscisic acid-induced stomatal closure  Guard cells are located some distance from the root tips, the tissue that will first experience soil drought. The restricted ability of guard cells to monitor such changes in water status is overcome by ABA-dependent signaling. Roots in drying soil release ABA into the xylem stream, which functions as a signal for the stomata to close (Hartung et al., 2002; Wilkinson & Davies, 2002). The guard cell receptor for ABA is unknown and its location within the guard cell is discussed controversially. One data set points to a receptor at the extracellular site of the plasma membrane (Anderson et al., 1994; Yamazaki et al., 2003), while another set of results indicates a localization inside the cell (Allan et al., 1994; Schwartz et al., 1994; Levchenko et al., 2005). Downstream of the presumed ABA receptor, a large number of signal intermediates have been suggested. We will not go into detail about ABA signaling, as the receptor remains a black box and suggested signaling pathways have already been the topic of a number of recent reviews (Hetherington, 2001; Schroeder et al., 2001; Finkelstein et al., 2002; Jones & Assmann, 2004). Instead, we will restrict the discussion to a small number of intermediates that have been linked directly to the transcription-independent regulation of guard cell ion-transporter activity.

ABA-induced rises in cytoplasmic-free Ca2+  The regulation of ion transport in guard cells has been linked for a long time to changes in the cytoplasmic Ca2+ concentration. The first indications of Ca2+ as a second messenger came from experiments with C. communis, which showed that a high level of extracellular Ca2+ enhances the ABA inhibition of stomatal opening (DeSilva et al., 1985a), while Ca2+-channel blockers had the opposite effect (DeSilva et al., 1985b). Further evidence was obtained with Ca2+-reporter dyes, which show that ABA can induce a rise in cytoplasmic Ca2+ (McAinsh et al., 1990; Gilroy et al., 1991). Finally, in guard cells of C. communis that were injected with the Ca2+ buffer, BAPTA (1,2-bis(2-aminophenoxy)ethane-N,N,N’,N’-tetraacetic acid), ABA did not evoke a rise in the cytoplasmic Ca2+ concentration and stomatal closure was prevented (Webb et al., 2001). Despite this evidence for the involvement of Ca2+ in ABA action, a Ca2+-independent ABA signal pathway also seems to exist in guard cells. Several publications have reported a population of stomata, in C. communis, that close in response to ABA without detectable changes in the cytoplasmic Ca2+ concentration (McAinsh et al., 1990; Gilroy et al., 1991; Allan et al., 1994). This indicates that a rise in the cytoplasmic Ca2+ concentration of guard cells is not essential for stomatal closure.

Rises in the cytoplasmic Ca2+ concentration were proposed to be initiated via the ABA activation of hyperpolarization-activated plasma membrane Ca2+ channels (Grabov & Blatt, 1998). These plasma-membrane channels were shown to activate through direct interaction with ABA in V. faba (Hamilton et al. 2000) and through the release of H2O2 in Arabidopsis (Pei et al., 2000). Alternatively, ABA could induce increases in the Ca2+ concentration through the stimulation of Ca2+ channels in intracellular membranes, gated by cADP ribose (Leckie et al., 1998) or inositol phosphates (Staxen et al., 1999; Lemtiri-Chlieh et al., 2003); see also section II, 2. However, electrophoretic injection of these messengers, into guard cells of intact V. faba plants, could not mimic ABA action (Levchenko et al., 2005). Apparently, cADP ribose, IP3 and IP6 cannot replace ABA concerning the activation of anion channels.

ABA not only triggers single rises in the cytoplasmic Ca2+ concentration, it can also induce repetitive Ca2+ oscillations (Staxen et al., 1999; Allen et al., 2000). The oscillation frequency was reported to be signal specific and was suggested to enable guard cells to differentiate between signals leading either to stomatal opening or closure (McAinsh & Hetherington, 1998; Allen et al., 2001; Hetherington & Brownlee, 2004). ABA-induced oscillations in cytoplasmic Ca2+ have an average period of 10.3 min in Arabidopsis (Allen et al., 2001), which is longer than the 5–10 min required for the initiation of stomatal closure by ABA (Roelfsema et al., 2004). This takes the frequency of Ca2+ oscillations as a signal that induces stomatal closure into question. Nevertheless, the Ca2+ signature may play a role in other responses, such as the inhibition of stomatal reopening after removal of ABA (Allen et al., 2001), or ABA-induced changes in gene transcription. Such a role for Ca2+ oscillations has been described for gene regulation in animal cells (Dolmetsch et al., 1998).

Evidence for ABA-induced Ca2+ oscillations was obtained with guard cells in epidermal strips, but guard cells in intact plants did not show ABA-induced changes in cytoplasmic Ca2+ (Levchenko et al., 2005). This might indicate that Ca2+ oscillations reflect the inability of isolated guard cells to maintain a steady Ca2+ concentration (own unpublished data). Note that Ca2+ oscillations in isolated guard cells can also occur spontaneously and that, on occasion, these oscillations are turned off by ABA (Klüsener et al., 2002). Apparently, Ca2+ changes triggered by ABA are heterogeneous in nature, and future research needs to explain how the variability of Ca2+ responses relates to stomatal function.

ABA regulation of plasma-membrane ion transport   The most prominent change in ion-transport activity at the plasma membrane of guard cells is the activation of plasma-membrane anion channels (Thiel et al., 1992; Grabov et al., 1997; Pei et al., 1997; Raschke et al., 2003; Roelfsema et al., 2004). The activation of anion channels creates a pathway for anion efflux and thereby depolarizes the plasma membrane. In addition to anion-channel activation, ABA probably also reduces the activity of plasma-membrane H+-ATPases (Goh et al., 1996; Zhang et al., 2004), which would favor a pronounced depolarization. As a result of plasma-membrane depolarization, K+ ions will be extruded from the guard cells via outward-rectifying channels (Thiel et al., 1992; Roelfsema et al., 2004). Based on the biophysical properties of the plasma membrane, the activation of anion channels is thus sufficient to induce an extrusion of anions as well as K+. Nevertheless, ABA was also found to regulate plasma-membrane K+ channels directly. The maximal conductance of inward K+ channels was found to be reduced by ABA, in guard cells of epidermal strips (Blatt & Armstrong, 1993) as well as in protoplasts thereof (Schwartz et al., 1994). ABA was also reported to stimulate outward-rectifying K+ channels in guard cells of isolated epidermal strips (Blatt et al., 1990), but this could not be confirmed by measurements on guard cells in intact plants (Roelfsema et al., 2004). In the discussion of K+-channel regulation by ABA, it is often forgotten that these channels in the first instance are regulated by the membrane potential. In most cells, ABA will depolarize the plasma membrane (Fig. 8a) to values at which the outward K+ channel becomes active, but at which the inward K+ channel is no longer activated (Roelfsema et al., 2004). In these cells, the reduction in maximum conductance of inward K+ channels will thus have no physiological implications.

In guard cells of intact plants, ABA activates plasma-membrane anion channels in a transient manner; anion currents peak after ≈ 5 min and decay thereafter (Fig. 8b; Roelfsema et al., 2004; Levchenko et al., 2005). However, residual anion-channel activity remains in the presence of ABA and the current only returns to prestimulus values after ABA removal. The first phase of this response may thus lead to rapid stomatal closure, while the second phase will mainly prevent the reopening of stomata. The two phases may explain the differences recognized between ABA induction of stomatal closure and ABA inhibition of stomatal opening.

It has long been postulated that ABA stimulates anion channels through a Ca2+-dependent pathway, based on the following observations: ABA can induce rises in the cytoplasmic Ca2+ concentration, and both R- and S-type anion channels are activated by cytoplasmic Ca2+ (Schroeder & Hagiwara, 1989; Hedrich et al., 1990). Surprisingly, the activation of anion channels in guard cells of intact plants occurred without a concomitant rise in the cytoplasmic Ca2+ concentration (Levchenko et al., 2005). The response was not fully Ca2+ independent, as it was repressed after injection of the Ca2+ buffer, BAPTA. Apparently, Ca2+ signals are not essential for ABA activation of anion channels, in agreement with previous results described for ABA on anion channels, using patch-clamped protoplasts (Schwarz & Schroeder, 1998). A very similar situation exits for ABA inhibition of inward-rectifying K+ channels. These channels display a steep Ca2+ dependence (Grabov & Blatt, 1999), but inhibition of the inward K+ channel by ABA in V. faba protoplasts occurred in a Ca2+-independent manner (Romano et al., 2000). This leads to the conclusion that both anion- and inward K+ channels are sensitive to cytoplasmic Ca2+, but are predominantly regulated by ABA via a Ca2+-independent pathway.

IV. Interaction between guard cell signaling pathways

The signal pathways for light and humidity have been discussed, so far, as linear chains of events that couple environmental changes to a response at the plasma or vacuolar membrane. In nature, however, signals from the environment are seldom so discrete. On a hot summer day, for instance, high light intensities will often coincide with low humidity. The first signal will cause stomata to open, while the second induces stomatal closure. Guard cells therefore must be able to process these signals in such a way that a collision of opposing responses is prevented. The models of guard cell signaling should also explain why stomatal opening is depressed at midday and how stomata of shaded plants in the canopy respond to sudden sunflecks. Several mechanisms may function in guard cells to obtain the optimal response with multiple changes in the environment. Downstream in these pathways, various responses may differ in timing, while upstream signal pathways could converge at certain stages.

1. Timing of guard cell responses

Guard cells in intact plants display responses at the plasma membrane with a specific timing, depending on the stimulus received. This is most obvious for ABA action where, after a lag time of ≈ 2 min, ABA induces an inward anion current that peaks after ≈ 5 min and than drops to a residual level after ≈ 10 min (Roelfsema et al., 2004; Fig. 8b). As discussed in section III. 2, the initial large inward current will cause a rapid closure of stomata, while the residual anion channel activity may prevent stomata from reopening. During the second phase of the ABA response, stomata may become responsive to light again, although the degree of stomatal opening will be smaller, as in the absence of ABA (Leymarie et al., 1998). The specific timing of the ABA response enables plants to respond quickly to drought without completely losing sensitivity to other environmental factors.

Although the timing of ABA responses in guard cells is robust, the magnitude of responses can differ considerably. With repetitive ABA applications, we found that the responsiveness of single guard cells to ABA can change in time (Roelfsema et al., 2004). This heterogeneity in responsiveness could be beneficial for plants under field conditions, as pointed out by Kaiser & Kappen (2001). During drought, the ABA level in the xylem will slowly rise and because of differences in responsiveness it will completely close some stomata, while others will remain open, to some extent. Upon stomatal closure, transpiration will decrease and less ABA will reach the stomatal complex. This may create a simple feedback system that does not involve communication between guard cells and the root. Such a communication would be very difficult in tall trees because of the large distance the signal would need to travel. In favor of such a feedback system is the observation that guard cells monitor the amount of ABA reaching the stomatal complex, instead of the ABA concentration (Trejo et al., 1995).

Another feedback system regulating stomatal movement involves change in the intracellular CO2 concentration (Roelfsema et al., 2002, Fig. 9). High CO2 concentrations induce stomatal closure, which, in turn, lowers CO2 uptake and thus decreases the intracellular CO2 concentration. In contrast to the ABA response, the guard cell response to CO2 sets in slowly and anion channel activation does not reach a maximum within 15 min (Fig. 7c). This may not apply for guard cells in epidermal strips, as, in this preparation, a peak anion channel activity was recorded 2 min after CO2 application (Brearley et al., 1997). The absence of a specific timing of the CO2 response, in intact plants, points to the role of CO2 in a feedback loop. This response may continue until the CO2 concentration in the substomatal cavity decreases to such an extent that the response automatically shuts off (Fig. 9).

Figure 9.

Schematic presentation of signal pathways leading to light-induced stomatal opening. (a) Epidermal pavement cell, (b) guard cell, (c) mesophyll cell. Photosynthetic-active radiation (red light) reduces the CO2 concentration in the leaf, thereby deactivating anion channels in guard cells. Blue light acts on the PHOT1 and 2 receptors in the plasma membrane and activates the H+-ATPase. Both pathways lead to hyperpolarization of guard cells and provoke K+ uptake, causing the guard cell to swell and the stomatal pore to open. Note that opening of the stomatal pore increases the influx of CO2, which, in turn, can activate guard cell anion channels and induce stomatal closure. The pathway involving CO2 thus provides a negative feedback mechanism.

Guard cells rapidly respond to blue light, the membrane potential hyperpolarizes after ≈ 60 s (Roelfsema et al., 2001), but the rise in plasma membrane current starts earlier (Fig. 8c). The outward H+-ATPase current peaks 3 min after the onset of blue light, in agreement with a maximal degree of phosphorylation of plasma membrane H+-ATPases (Kinoshita & Shimazaki, 1999). Just as with ABA, blue light will cause maximal change in stomatal movement several minutes after stimulus onset. After prolonged stimulation with blue light, the outward H+ current decays to a steady-state level, sufficient to cause prolonged stomatal opening with time.

2. Cross talk of signaling pathways

Cross talk of signaling pathways in plant cells is an emerging topic in plant biology (Taylor & McAinsh, 2004) and may explain the ability of sessile organisms to adapt to changing environments. As a result of cross talk, one signal may alter the guard cell's responsiveness to another signal. Such an interaction has been proposed for CO2 and ABA (Raschke, 1975b), ABA and blue light (Goh et al., 1996; Roelfsema et al., 1998), and for blue light and CO2 (Assmann, 1988). High CO2 concentrations can increase the ABA responsiveness, while ABA, in turn, decreases the blue light responsiveness of the stomata. Finally, the blue light responsiveness is enhanced at low intercellular CO2 concentrations. It is not difficult to imagine what sort of complexity emerges if the interaction of multiple signals are taken into account. Instead of testing various signal combinations on different plant species, it may be more fruitful to search for molecular determinants at which several signal pathways converge.

Interaction of signaling may occur downstream in the pathways, as distinct signals can regulate the same ion transporters. For instance, both ABA and CO2 activate anion channels, which may explain their synergistic mode of action concerning stomatal closure (Raschke, 1975b). Cross talk may also occur, to a greater extent, upstream and involve common regulatory proteins or second messengers. The search for such interactions is assisted by the signaling mutants available for Arabidopsis. For instance, the stomata of abscisic acid insensitive (abi1) mutants were found to open more slowly in response to light than those of wild-type plants (Eckert & Kaldenhoff, 2000). This indicates that ABA signaling interacts with the blue light and PAR-response pathways in guard cells.

Many other interactions have also been suggested for the ABA signal pathway. ABA signaling may be linked to salt tolerance as well as to cold acclimation (Chinnusamy et al., 2004). Furthermore, a close correlation was found between the ABA- and sugar-sensing pathways (Finkelstein & Gibson, 2002). Note, however, that these interactions were recognized for whole plants or seedlings, but were not always confirmed for guard cells. The interconnections found may hold for some ABA responses, but not necessarily for all. For instance, the ABI4 gene is important for ABA and sugar signaling in seeds, but the gene does not seem to play a role in the regulation of stomatal movement. Another mutation (abi8) that affects both ABA and sugar signaling does interfere with the ABA inhibition of stomatal opening (Brocard-Gifford et al., 2004). Apparently it is hard to generalize interconnections of pathways. After all, from gene-expression studies we know that different cell types are equipped with different hardware, which makes it unlikely that the same software will run in all cell types. Differences in pathways for specific responses, cell types and species therefore should be taken into account.

V. Outlook

Guard cell research has progressed a long way, from the initial microscopic observations of stomatal apertures by Moldenhauer (1812), to the specific regulation of single ion channels. On this way, however, it was often forgotten to reflect on whether the changes measured in transporter activity had the proper amplitude, direction and kinetics, in view of the dramatic changes in osmotic content of guard cells during stomatal movement. The calculations in this review indicate that blue light-induced changes in ion fluxes, recorded for guard cells in the intact plant, fit changes in the ion content of guard cells during stomatal movement. In general, however, the ion fluxes and changes in solute concentrations recorded are somewhat smaller than the changes calculated in solute concentration. This may point to the involvement of additional transport processes, such as the accumulation of sugars. With the identification of the first sugar transporters in guard cells, these nonionic molecules will come into focus again.

Regulation of stomatal movement by some signals from the environment is partially understood at the molecular level, such as the blue light-dependent activation of plasma-membrane H+-ATPases. For this pathway, both the receptors and the downstream responses have been identified (Kinoshita & Shimazaki, 1999; Kinoshita et al., 2001) and fit the changes in transport activity measured directly on single guard cells (Roelfsema et al., 2001; Taylor & Assmann, 2001). Identification of the signaling components that interconnect the receptor with the H+-ATPase may shed light on the way that the blue light signaling is affected by other signals, such as ABA. Compared to the blue light-dependent pathway, little is known about the PAR (CO2)-dependent pathway. Recently, stomatal movement was shown to be related to ascorbate synthesis (Chen & Gallie, 2004), which indicates that the cytoplasmic redox state could serve as a switch, initiating this response. Compared to the light responses of guard cells, ABA signaling has received overwhelming attention. A large number of mutants have been isolated, which are either hyper- or less sensitive to the stress hormone (Schroeder et al., 2001; Finkelstein et al., 2002). Despite the attempt to put the mutated gene products in hierarchal schemes, a common model for ABA signal transduction has not yet been obtained. The fact that, in some cases, data have been integrated, irrespective of the plant species, cell type or response studied, may have added to the degree of complication. Putting the signal components in place will be a major task for plant cell biology. In this context, fast electrical responses and more slowly emerging changes in gene-expression rates should be decomposed for given cell types. Guard cells will continue to be one of the systems of choice, because of the possibility to measure responses for defined target proteins, within seconds up to hours after stimulus onset. We can thus expect exciting new insights into ‘the Watergate’, soon again.


We wish to thank D. Geiger, V. Levchenko and H. Marten (University of Würzburg) for providing data for the figures, S. Liebig for the drawing in Fig. 1, and I. Marten (University of Würzburg) for help with the preparation of the manuscript. This work was supported by grants from the Deutsche Forschungsgemeinschaft to R.H.