Barley mildew and its elicitor chitosan promote closed stomata by stimulating guard-cell S-type anion channels

Summary

Stomatal closure is known to be associated with early defence responses of plant cells triggered by microbe-associated molecular patterns (MAMPs). However, the molecular mechanisms underlying these guard-cell responses have not yet been elucidated. We therefore studied pathogen-induced changes in ion channel activity in Hordeum vulgare guard cells. Barley mildew (Blumeria graminis) hyphae growing on leaves inhibited light-induced stomatal opening, starting at 9 h after inoculation, when appressoria had developed. Alternatively, stomatal closure was induced by nano-infusion of chitosan via open stomata into the sub-stomatal cavity. Experiments using intracellular double-barreled micro-electrodes revealed that mildew stimulated S-type (slow) anion channels in guard cells. These channels enable the efflux of anions from guard cells and also promote K⁺ extrusion by altering the plasma membrane potential. Stimulation of S-type anion channels was also provoked by nano-infusion of chitosan. These data suggest that MAMPs of mildew hyphae penetrating the cuticle provoke activation of S-type anion channels in guard cells. In response, guard cells extrude K⁺ salts, resulting in stomatal closure. Plasma membrane anion channels probably represent general targets of MAMP signaling in plants, as these elicitors depolarize the plasma membrane of various cell types.

Introduction

At least two types of defense responses against pathogenic micro-organisms can be distinguished in plant cells (Jones and Dangl, 2006; Boller and Felix, 2009; Schulze-Lefert and Panstruga, 2011). A first line of defense is induced by microbe-associated molecular patterns (MAMPs) that are recognized by plasma membrane receptors. In most cases, MAMPs represent well-conserved, microbe-specific, molecular structures on the surface of micro-organisms, such as flagella proteins (Felix et al., 1999) or chitin fragments (Felix et al., 1993). Upon detection of such MAMPs, plant cells execute a number of non-specific defense responses that provide basal resistance (Nürnberger and Kemmerling, 2006). In addition to the basal resistance, plants also possess additional defense mechanisms that are induced by so-called ‘effector proteins’ that are extruded by pathogenic micro-organisms into the cytosol of plant cells (Jones and Dangl, 2006; Boller and Felix, 2009).

In the past decade, multiple pattern recognition receptors (PRR) have been identified that are able to recognize certain MAMPs (Boller and Felix, 2009). Most of these receptors belong to a large family of leucine-rich repeat protein kinase (LRR-PK) receptors, which comprise an extracellular binding domain for MAMPs and a cytosolic protein kinase domain. Other MAMP receptors have been found to lack the intracellular protein kinase domain (Ron and Avni, 2004) or possess an extracellular lysine motif (Kaku et al., 2006) instead of an LRR domain. It is likely that binding of a MAMP to these receptors leads to oligomerization with other plasma membrane receptors (Chinchilla et al., 2007; Boller and Felix, 2009), which in turn causes interaction of the cytosolic domains and the initiation of basal defense responses.

Among the earliest events induced by MAMP receptors are changes in ion transport activity at the plasma membrane of host cells (El-Maarouf et al., 2001; Wendehenne et al., 2002; Colcombet et al., 2009; Jeworutzki et al., 2010). In general, MAMPs trigger depolarization of the plasma membrane potential (Pelissier et al., 1986; Kuchitsu et al., 1997; Felle et al., 2000; Jeworutzki et al., 2010), an increase in the extracellular pH (Felix et al., 1993, 1999; Felle et al., 2004) and elevation of the cytosolic free Ca²⁺ concentration (Blume et al., 2000; Lecourieux et al., 2005; Ranf et al., 2008; Jeworutzki et al., 2010). All these responses were recorded within 10 min after application of MAMPs, but their inter-relationship is still poorly understood.

In mesophyll cells of Arabidopsis thaliana, depolarization triggered by the MAMP flg22 is accompanied by the efflux of K⁺ and Cl⁻, as well as an influx of H⁺. It was postulated that elicitors such as flg22 activate plasma membrane anion channels through a Ca²⁺-dependent mechanism (Jeworutzki et al., 2010; Krol et al., 2010). However, activation of anion channels by MAMPs has not yet been demonstrated by direct measurements. Lack of direct proof for stimulation of plasma membrane anion channels by MAMPs in patch-clamp studies is probably related to the protoplast isolation procedure. The cell wall-degrading enzymes used to obtain protoplasts, such as cellulase and pectinase, induce MAMP responses and thus are likely to interfere with the ability of cells to respond to the elicitor under investigation (Carden and Felle, 2003).

Guard cells offer the unique possibility of recording ion channel activity without applying fungal enzymes, by using intracellular multi-barreled micro-electrodes. This technique does not require any cell preparation procedures, as it can be used in intact plants (Roelfsema et al., 2001; Roelfsema and Hedrich, 2005). Ion uptake into guard cells closely correlates with stomatal opening, as it leads to osmotic swelling that forces adjacent guard cells to bend away from each other (MacRobbie, 1987; Raschke et al., 1988; Roelfsema and Hedrich, 2005). During stomatal opening, K⁺ ions are transported by K⁺ uptake channels, a process that is driven by the electrical charge difference (membrane potential) across the guard-cell plasma membrane. The required negative membrane potential is generated by proton pumps, which also create a pH gradient. Co-transporters in the guard-cell membrane use this pH gradient for anion uptake. During stomatal closure, the release of K⁺ salts is strongly dependent on S- (slow) and R-type (rapid) anion channels, whose names are based on differences in the velocity of voltage-dependent activation (Schroeder and Hagiwara, 1989; Hedrich et al., 1990; Linder and Raschke, 1992; Schroeder and Keller, 1992). Active anion channels extrude anions and thereby depolarize (degrade) the membrane potential, which in turns leads to the release of K⁺ via K⁺ efflux channels (Roelfsema and Hedrich, 2005).

Recent studies have provided evidence that micro-organisms, or MAMPs, can trigger stomatal closure (Lee et al., 1999; Melotto et al., 2006; Liu et al., 2009). As these micro-organisms use open stomata to enter the leaf, MAMP recognition, and in turn stomatal closure, can prevent further infestation (Melotto et al., 2008). Stomatal closure is not only provoked by bacteria that pass through open stomata, but also by filamentous fungi that breach the cuticle using appressoria. During early phases of infection, stomatal opening is inhibited by barley mildew (Blumeria graminis), irrespective of the resistance properties of the host plant (Prats et al., 2006). At later stages of infection, stomata are locked open in plants displaying effector-triggered hypersensitive immune responses (Prats et al., 2006, 2010). Here, we show that inhibition of stomatal opening represents a very local response that correlates with the maturation of appressoria. Infection with barley mildew and nano-infusion of its elicitor chitosan (Maffi et al., 1998) lead to enhanced activity of S-type anion channels in guard cells. This response explains why stomata in the proximity of barley mildew appressoria remain closed in the light.

Results

Inhibition of stomatal opening

Infection with mildew is known to inhibit stomatal opening (Prats et al., 2006), but the stage of infection at which stomata start to recognize the pathogen has not yet been determined. We therefore fixed leaves of Hordeum vulgare cv. Ingrid with their adaxial side to the microscope table, i.e. with the abaxial leaf surface facing the objective. Using a micro-manipulator, 8–20 mildew conidia were placed in close proximity to stomatal complexes. The ability of stomata to open in the light was tested by applying white light from the microscope lamp at an intensity of 300 μmol m⁻² sec⁻¹ (Figure 1a,b). After monitoring stomatal opening for 30 min, the microscope lamp was turned off, as the development of mildew is retarded at high light intensities (Carver et al. 1994). Light-induced stomatal opening was tested again after an interval of 3 h. Although infection did not affect stomatal opening 3 and 6 h after infection, a reduced ability to open was observed at 9, 12 and 24 h after infection (Figure 1a,b).

Figure 1.  Inhibition of light-induced stomatal opening by barley mildew. Hordeum vulgare cv. Ingrid leaves were inoculated with conidiospores of Blumeria graminis, and the ability of stomata to open in light (300 μmol m⁻² sec⁻¹) was tested every 3 h. (a) Apertures of stomata were determined every 5 min during 30 min of illumination. Between illumination periods, the leaves were kept in darkness for 2.5 h, as indicated by the bar below the graph. Data for stomata located in close proximity to germinating conidia are shown in gray, and data for uninfected leaves are given in black (errors bars represent SE, n = 15). Note that light fails to trigger opening of stomata in infected leaves at 9, 12 and 24 h after infection. (b) Transmission light microscopy images of stomata (upper panels) and fluorescent microscopy images of fluorescein diacetate-loaded mildew hyphae (lower panels), obtained at 3, 9 and 24 h after infection. Images were obtained after exposing stomata to light for at least 30 min. The arrowheads in the lower panels indicate the primary germ tube (3 h after infection) and secondary germ tubes (9 and 24 h after infection) growing from the conidia.

The time point at which mildew infection starts to inhibit stomatal opening correlates with the completion of appressoria, as observed by microscopy (Figure 1b) (Both et al., 2005; Zhang et al., 2005). This suggests that guard cells recognize MAMPs associated with the penetration peg that breaches the cuticle. The ability of potential MAMPs to diffuse within the apoplast was studied by comparing stomata adjacent to the appressorium with those located further away (Figure 2a,b). Infection not only inhibited opening of the stoma directly neighboring the appressorium, but also those next to it, located in the same file (Figure 2a). However, opening of stomata located further away (Figure 2a) or in neighboring files (Figure 2b) was not affected. The impact of infection on stomata thus appears to be constrained to a region with a length of approximately 200 μm.

Figure 2.  Spatial influence of barley mildew on stomatal opening. The influence of mildew on stomatal movement was determined for stomata close to the appressoria and further away, in the same file (a) or in neighboring files (b). Mildew conidia were placed on the abaxial cuticle and allowed to form appressoria overnight. The next morning (16–18 h after infection), light-induced stomatal opening was measured. Data were averaged for the stomatal complex in closest proximity to the appressoria (downward arrowheads) and each subsequent stomatal complex (error bars represent SE, n = 8). The stomatal aperture was plotted against the mean distance between stomata in a single file (a) or between neighboring files (b). Note that mildew appressoria inhibit light-dependent stomatal opening only in the three closest stomata within a file.

Mildew infection retards light-induced hyperpolarization of guard cells

Infection with mildew probably alters guard-cell ion transport, as accumulation of K⁺ salts in these motor cells drives the osmotic swelling that underlies stomatal opening (Raschke et al., 1988; Roelfsema and Hedrich, 2005). In the stomatal complexes of grasses, K⁺ salts are shuttled between the guard cells and subsidiary cells (Raschke and Fellows, 1971; Buchsenschutz et al., 2005), and ion transport in these stomata may therefore differ from that of dicots, which lack subsidiary cells. For this reason, we used intracellular double-barreled micro-electrodes to study the ion transport properties of guard cells in the monocot species barley.

Barley guard cells, with a free-running membrane potential that was negative of −100 mV in the light, strongly depolarized after transition to darkness (Figure 3a). Changes in ion channel activity associated with this response were tested by clamping the membrane potential from −100 mV stepwise to test voltages ranging from −160 to 20 mV (Figure 3b). The time-dependent activation of inward current at −140 and −160 mV is reminiscent of the properties of K⁺-selective uptake channels (Fairleygrenot and Assmann, 1992; Szyroki et al., 2001; Very and Sentenac, 2003), whereas the sigmoidal current activation kinetics at −40 and −20 mV are typical of K⁺ efflux channels (Fairleygrenot and Assmann, 1992; Ache et al., 2000; Very and Sentenac, 2003). Light altered the activity of K⁺ channels in individual cells (Figure 3b,c), but, on average, there was no significant change in the conductance of K⁺ uptake channels (t-test, P > 0.1).

Figure 3.  Darkness-induced depolarization of barley guard cells. A barley guard cell was impaled using a double-barreled micro-electrode to record the membrane potential and perform voltage clamp measurements. (a) Membrane potential trace of a guard cell transferred from the light (300 μmol m⁻² sec⁻¹) to darkness and vice versa, as indicated by the bar below the trace. Note that the voltage trace was interrupted by short periods during which ion channel activity was tested using voltage clamp protocols. (b) Superposition of current traces obtained from the same guard cell as in (a) (symbols correspond to those in (a)). The cell was clamped from −100 mV stepwise to voltages ranging from −160 (lower current trace) to −20 mV (upper current trace), in light (open circle) or in darkness (closed circle). (c) Current–voltage relationship determined at the end of the 2 sec test pulse as shown in (b) (symbols correspond). Inset: magnification of the current–voltage relationship for potentials ranging from −120 to −60 mV. Note that darkness caused an increase in the inward current amplitude at membrane potentials negative of −40 mV.

In the voltage range from −100 to −60 mV, light consistently causes a shift of the current amplitude to more positive values (Figure 3c, inset). Because of these current changes, the point of interception of the current–voltage curve with the zero current axis (x axis) shifts from −112 to −60 mV (Figure 3c, inset). This strongly suggests that these current changes underlie changes in the light-induced membrane potential in barley guard cells. Guard cell responses of monocots are thus similar to those of dicots, despite of structural differences in their stomatal complexes. In the stomata of Vicia faba and Nicotiana tabacum (tobacco), light was shown to stimulate plasma membrane H⁺-ATPases and deactivate S-type anion channels (Roelfsema and Hedrich, 2005; Shimazaki et al., 2007).

In uninfected leaves, approximately half of the guard cells (23 of 39) hyperpolarized upon application of light, but the remaining cells did not (Figure 4a). Infection with barley mildew increased the number of cells that did not hyperpolarize in the light to 28 of 34 (Figure 4b). The reduced likeliness of light-induced hyperpolarization correlated with enhanced inward current amplitudes at potentials between −120 and −80 mV (Figure 4c,d). This suggests that mildew infection stimulates plasma membrane anion channels in guard cells.

Figure 4.  Mildew-dependent inhibition of light-induced hyperpolarization in guard cells. Membrane potential changes recorded for guard cells in uninfected leaves (a) or in close proximity to mildew appressoria (b). (a) In control leaves, 16 of 39 guard cells displayed a steady depolarized membrane potential (upper traces), irrespective of transfer from the light (300 μmol m⁻² sec⁻¹) to darkness and vice versa, as indicated by the bar below the trace. The remaining 23 of 39 guard cells depolarized after switching off the microscope lamp and hyperpolarized again in the light (lower trace). (b) The development of mildew appressoria close to the stomatal complex increased the number of guard cells that failed to display large light-dependent membrane potential changes to 28 of 34. (c, d) Averaged current–voltage relationship of guard cells in control leaves (c) or guard cells located close to the appressoria of mildew (d). Current–voltage relationships are shown for cells in the light (open symbols) and darkness (closed symbols). Insets: magnification of the current–voltage relationship for potentials ranging from −120 to −60 mV. Error bars represent SE (n = 11–15).

Mildew infection stimulates S-type anion channels

The potential impact of barley mildew on S-type anion channels was examined by using micro-electrodes filled with CsCl. Cytoplasmic Cs⁺ inhibited K⁺ efflux channels (compare Figures 3b and 5a), and thus improves the recording of anion channels. Plasma membrane currents were elicited by clamping the plasma membrane from −100 mV stepwise to test potentials ranging from −160 to 0 mV. Instantaneous inward currents, measured directly after stepping to the test potential of −100 mV, were stimulated by mildew infection (Figure 5b). In line with the activation of anion channels by mildew, linear extrapolation of the current–voltage relationship revealed a reversal potential that was positive of 0 mV (Figure 5b, inset). Moreover, averaged traces of barley guard cells recorded by stepping from −100 to −60 or −40 mV show slowly activating inward currents (Figure 5c) reminiscent of those carried by S-type anion channels (Schroeder and Hagiwara, 1989; Linder and Raschke, 1992; Vahisalu et al., 2008; Geiger et al., 2009). These data thus strongly suggest that infection with mildew enhances the activity of S-type anion channels (Figure 5c).

Figure 5.  Stimulation of S-type anion channels by mildew. (a) Superposition of mean current traces of barley guard cells impaled using double-barreled micro-electrodes filled with 300 mm CsCl. The cells were clamped from a holding potential of −100 mV stepwise to potentials ranging from −160 (lower trace) to 0 mV (upper trace). Error bars represent SE (n = 8 or 9). Guard cells were located in uninfected leaves (left traces) or close to mildew appressoria (right traces). Note that K⁺ efflux channels were blocked by Cs⁺. (b) Averaged current–voltage relationship for guard cells at the start of the test pulse (instantaneous currents). Data were obtained for either guard cells located close to mildew appressoria (open triangles) or control cells (open circle). Inset: linear interpolation of both instantaneous current–voltage relationships revealed an intercept at 43 mV. (c) Magnification of mean current traces shown in (a), obtained by clamping cells from −100 to −60 mV (upper traces) and −40 mV (lower traces). Note the enhanced slowly activating inward currents in guard cells in close proximity to mildew appressoria (open triangle) compared to the control (open circle).

Stomatal closure induced by the fungal elicitor chitosan

Components of fungal cell walls, such as chitin or chitosan fragments, represent potent MAMPs that are recognized by cells of a variety of plant species (Ryan, 1987; Maffi et al., 1998; Lee et al., 1999). The ability of these MAMPs to affect stomatal movement in barley was tested using the nano-infusion technique (Hanstein and Felle, 2004). Micro-capillaries were positioned in open stomata, and injection of the elicitor was monitored on the basis of the spread of a fluorescent marker in the apoplast (Figure 6a). Injection of chitosan induced rapid closure of stomata, with an average half-time of 10 min (Figure 6b,c). During the period from 9 am to 3 pm, stimulation with chitosan triggered the closure of 17 of 19 stomata, whereas the injection of control solution only affected five of 16 stomata (Figure 6b). However, the sensitivity of stomata to injection of control solution was enhanced later in the day (Figure 6c).

Figure 6.  Stomatal closure induced by nano-infusion of chitosan. (a) Overlay of transmitted light images and fluorescence images obtained before (upper panel), directly after (middle panel), and 15 min after (lower panel) injection of chitosan solution into the sub-stomatal cavity. A micro-capillary with a broken tip (diameter approximately 2 μm) was inserted into the leaf via the open stoma on the left. Nano-infusion of solution containing 10 mg L⁻¹ chitosan and 1 mg L⁻¹ lucifer yellow was monitored by the spread of fluorescence below the leaf surface (middle panel). Note that application of chitosan caused closure of the stoma on the right (lower panel). (b, c) Normalized data for stomata stimulated by nano-infusion of chitosan (open triangle) or control solution (open circle) as shown in (a). Data were obtained from stomata next to those through which the nano-infusion capillary was inserted; arrows indicate the start of nano-infusion. Means were calculated for experiments performed between 9 am and 3 pm (b) and 3 and 6 pm (c). Error bars represent SE (n = 11–19).

The ion conductance changes underlying the chitosan-triggered stomatal closure were studied using guard cells impaled with intracellular micro-electrodes. Chitosan was applied using a nano-infusion pipette positioned in a neighboring stoma. In guard cells clamped to −100 mV, application of chitosan triggered a repetitive increase of inward current (Figure 7a), with an average amplitude of −163 pA at 7.5 min after nano-infusion (Figure 7a–c). In general, chitosan did not alter the activity of K⁺ uptake channels, but enhanced the inward current amplitude at potentials ranging from −120 to −60 mV (Figure 7b,c). Much smaller changes in current were observed after infusion of control solution (Figure 7c). Slowly activating inward currents, recorded after stepping the membrane potential from −100 to −60 mV, were enhanced after stimulation with chitosan (Figure 7d). Apparently, chitosan stimulates S-type anion channels, just as infection with barley mildew does.

Figure 7.  Nano-infusion of chitosan stimulates S-type anion channels. (a) Plasma membrane current trace of a barley guard cell impaled using a double-barreled electrode and clamped to −100 mV. The cell was stimulated by nano-infusion of chitosan through a neighboring stoma (arrow indicates the start of infusion). Note that chitosan triggered a repetitive increase in inward current. The current trace is interrupted at time points during which voltage step protocols were applied. (b) Superimposed current traces obtained by clamping the plasma membrane from −100 mV stepwise to potentials ranging from −180 mV (lower trace) to 20 mV (upper trace). Data are from the same cell as in (a), recorded before (open diamond) or after (open triangle) nano-infusion of chitosan. Note that infusion of chitosan stimulates K⁺ efflux channels and S-type anion channels, but inhibits K⁺ uptake channels. (c) Averaged current–voltage relationship of barley guard cells stimulated by nano-infusion of chitosan (open triangles) or control solution (open circles). Inset: magnification of the average current–voltage relationship from −120 to −60 mV. Error bars represent SE (n = 8). Note that infusion of chitosan enhances the inward current amplitude in the voltage range from −120 to −60 mV. (d) Mean current traces of cells clamped stepwise from −100 to −60 mV for cells before (open diamonds) or after (open triangles) nano-infusion of chitosan. Note that chitosan stimulates the time-dependent activation of S-type anion channels, as is evident from the slow increase in inward current at −60 mV. Error bars represent SE (n = 8).

Discussion

Plant cells are equipped with a large number of plasma membrane receptors that recognize the molecular patterns of a variety of micro-organisms (Boller and Felix, 2009). It is thus very likely that MAMP receptors are involved in early responses of barley leaf cells to mildew infection. During the first hours of infection, growth of primary germ tubes on the cuticle induces accumulation of ROS (Hückelhoven et al., 1999) and provokes changes in the apoplastic pH (Felle et al., 2004). Apparently, ordinary epidermis cells already recognize growth of primary germ tubes at this stage of infection, but this does not yet influence stomatal movements. Light-induced stomatal opening is first inhibited 9 h after infection, when hyphae start to penetrate the cuticle (Figure 1).

Apparently, guard cells receive signals from maturating appressoria that cause inhibition of stomatal opening. These signals probably spread into a restricted area of approximately 200 μm around the appressoria, as light-induced stomatal opening is unaffected further away (Figure 2). Guard cells are not known to establish a susceptible interaction with mildew (Lin and Edwards, 1974), but mildew hyphae can penetrate several other cell types within the epidermis. It is feasible that such interactions provoke a release of MAMPs, which are in turn perceived by guard cells. Alternatively, the guard cells may simply recognize ubiquitous MAMPs of fungal hyphae, such as chitosan.

Influence of barley mildew on guard-cell ion transport

Despite the importance of grasses in agriculture, little is known about regulation of ion transport in their stomatal complexes. In contrast to dicots, guard cells in grasses are dumbbell-shaped and flanked by subsidiary cells. During stomatal opening and closure, K⁺ ions are shuttled between guard cells and subsidiary cells, which implies that K⁺ transport occurs in opposite directions in both cell types. Light provokes large changes in the guard-cell membrane potential of barley (Figures 3 and 4), similar to those observed in V. faba and N. tabacum (Roelfsema et al., 2001; Marten et al., 2008). This suggests that guard-cell responses to light do not fundamentally differ between monocots and dicots.

Infection with barley mildew stimulates the activity of S-type anion channels (Figure 5). The enhanced activity of anion channels explains why stomata in infected leaves fail to open in the light. S-type anion channels release Cl⁻ and NO₃⁻ (Schmidt and Schroeder, 1994; Geiger et al., 2009, 2011), and, due to the negative charge of these ions, the plasma membrane depolarizes (Roelfsema and Hedrich, 2005). Active S-type anion channels therefore prevent hyperpolarization of the plasma membrane, which is a prerequisite for ion channel-mediated uptake of K⁺. In infected leaves, guard cells fail to hyperpolarize in the light because of the enhanced activity of S-type anion channels, and therefore the stomata remain closed.

MAMP-triggered stomatal closure

The influence of a major fungal MAMP on stomatal movement was studied by nano-infusion of chitosan into the sub-stomatal cavity. The receptor for chitosan may be related to the receptor complex of chitin elicitor binding protein (CEBiP) and Oryza sativa chitin elicitor receptor kinase (OsCERK) in rice (Oryza sativa), which is essential for recognizing chitin fragments (Shimizu et al., 2010). Within this complex, the CEBiP proteins bind chitin fragments (Kaku et al., 2006), and OsCERK is a receptor kinase that most likely initiates downstream events (Miya et al., 2007; Shimizu et al., 2010).

During early times of the day, the chitosan triggered fast stomatal closure, but infusion itself did not affect the stomatal aperture (Figure 6). This fast response to chitosan correlates with increased activity of S-type anion channels (Figure 7). In Arabidopsis, the guard-cell S-type anion channel is encoded by SLAC1 (Vahisalu et al., 2008; Geiger et al., 2009), and SLAC1 probably represents an early target of MAMP-triggered responses. Future studies are required to determine whether MAMP signals alter the phosphorylation status of SLAC1, as previous studies have shown that SLAC1 is regulated by protein kinases (Geiger et al., 2009; Lee et al., 2009; Vahisalu et al., 2010).

The barley response differs from that previously reported for transgenic tobacco guard cells (Blatt et al., 1999). In tobacco guard cells expressing the tomato Cf-9 receptor, stimulation with the Avr-9 elicitor altered the activity of K⁺ channels. However, plasma membrane K⁺ channels were not affected by chitosan in barley (Figure 7c), suggesting different downstream signaling pathways for Cf-9 in tobacco and chitosan in barley.

Guard cell signal transduction chain

Guard cell responses to MAMPs and abscisic acid (ABA) show similarities, as both agonists stimulate S-type anion channels and induce rapid stomatal closure (Pei et al., 1997; Roelfsema et al., 2004). However, ABA and MAMP signaling differ at the receptor level, as ABA is bound by cytosolic receptors (Ma et al., 2009; Park et al., 2009), whereas MAMP receptors have extracellular binding sites (Boller and Felix, 2009). Despite these differences, the signaling chains downstream of the receptors appear to share components. In Arabidopsis, activation of the guard cell-specific protein kinase OST1/SRK2E is essential for ABA- as well as flg22-induced stomatal closure (Mustilli et al., 2002; Melotto et al., 2006). OST1 was recently shown to activate the S-type anion channel SLAC1 (Geiger et al., 2009; Lee et al., 2009), suggesting that OST1 can also mediate flg22-induced activation of SLAC1.

Alternatively, MAMP receptors in guard cells may trigger an increase in the cytosolic free Ca²⁺ concentration, as shown for various cell types (Blume et al., 2000; Lecourieux et al., 2005; Felle et al., 2008; Ranf et al., 2008; Jeworutzki et al., 2010). In barley leaves infected with mildew, two phases of Ca²⁺ responses could be distinguished, the first at 2 h after infection and the second at 8 h after infection (Felle et al., 2004). The second phase thus correlates with the time frame during which mildew infection starts to inhibit stomatal opening. This suggests that, during the period of 8–9 h after infection, Ca²⁺ signals are triggered by MAMPs in several cell types, including guard cells. An increase in the cytosolic free Ca²⁺ concentration will in turn lead to stimulation of anion channels (Schroeder and Hagiwara, 1989; Chen et al., 2010; Stange et al., 2010), mediated by calcium-dependent protein kinases (Geiger et al., 2010). However, MAMP-induced increases in the cytosolic free Ca²⁺ concentration remain to be confirmed for guard cells by direct measurements.

MAMP responses of plant cells

In a pioneering study, Pelissier et al. (1986) showed that fungal elicitors induce depolarization of melon (Cucumis melo) and tobacco root cells. Similar responses have been observed with several other species and cell types (Kuchitsu et al., 1997; Felle et al., 2000; Mithofer et al., 2005; Jeworutzki et al., 2010). It is likely that MAMP-triggered activation of anion channels is also the cause of depolarization in these cell types. However, the plasma membrane anion channels targeted by MAMP signals may differ between cell types. Experiments with anion channel blockers suggest that R-type anion channels, instead of S-type, are important for MAMP responses in Arabidopsis suspension cells (Colcombet et al., 2009).

In line with the activation of plasma membrane anion channels, MAMP-stimulated cells extrude anions as well as K⁺ (Felle et al., 2000; Wendehenne et al., 2002; Jeworutzki et al., 2010), but start to take up H⁺ (Felix et al., 1993; Felle et al., 2000; Jeworutzki et al., 2010). The loss of K⁺ salts seems to be a very general response after recognition of micro-organisms by plant cells. This suggests that plants cells reduce their turgor during the early phase of an interaction with microbes, which may prevent cell rupture if the cell becomes damaged.

Whereas the loss of osmolytes presumably has a limited impact on large cells, it results in large turgor changes in small guard cells. Because of the reduced turgor in guard cells, the stomata close, which can protect leaves against further invasion by micro-organisms (Melotto et al., 2008). However, many pathogenic fungi overcome this hurdle by developing appressoria that enable penetration of closed stomata (Read et al., 1997) or the cuticle. In interactions with these pathogens, stomatal closure may represent a rudimentary response, or it may play a role that has not yet been identified. Studies with mutants whose stomata fail to close upon infection may help to uncover such roles. In Arabidopsis, stomata of the ost1 and ost2 mutants were shown to be impaired in responses to flg22 (Melotto et al., 2006; Liu et al., 2009). However, the phenotype of these mutants is not specific for MAMPs, as they are also impaired in ABA responses (Mustilli et al., 2002; Merlot et al., 2007). Future studies may lead to identification of signal transduction components in guard cells that are specific for MAMP responses, and thus enable a more detailed assessment of the impact of anion channel activity on basal resistance.

Experimental Procedures

Plant material and cultivation of fungus

Hordeum vulgare (L.) cv. Ingrid plants were grown in a growth cabinet equipped with L36W/25 fluorescent tubes (Osram, http://www.osram.de), providing light at a photon flux density of 150 μmol m⁻² sec⁻¹. The climate cabinet simulated a day/night cycle of 12/12 h, with respective temperatures of 22/16°C. The first developed leaves of 1-week-old plants were used for the measurements. Seeds were kindly provided by R. Panstruga (Max Planck Institute for Plant Breeding Research, Cologne, Germany).

The fungus Blumeria graminis DC Speer f. sp. hordei strain CC1 (originally obtained from T. Carver, Institute of Grassland and Environmental Research, Aberystwyth, UK) was cultivated in a climate cabinet with a day/night cycle of 16/8 h (22/18°C) and a photon flux density of 300 μmol m⁻² sec⁻¹. The fungus was grown on Hordeum vulgare (L.) cv. Bonus plants until distinct white colonies were visible. For each experiment, 8–20 spores were transferred from the donor leaves to the abaxial side of an uninfected leaf, under a microscope, as described below, using micro-capillaries and an MM3A micro-manipulator (Kleindiek Nanotechnik, http://www.nanotechnik. com).

Microscopy set-up

Intact plants were placed on a table of an upright microscope (Axioskop 2FS; Zeiss, http://www.zeiss.com). The adaxial side of a leaf was fixed onto an acrylic glass holder using double-sided adhesive tape. A flow of humidified, CO₂-free air (400 ml min⁻¹) was directed onto the leaf surface by a nozzle (14 × 2 mm), and white light was provided by the microscope lamp (HAL 12/100 W; Zeiss) at a photon flux density of 300 μmol m⁻² sec⁻¹. Stomata on the abaxial side were visualized using a long-distance objective (Epiplan LD 50×; Zeiss), and images were obtained using an IMAG-K4 CCD camera (Walz, http://www.walz.com) at intervals of 10–30 sec using Kappa Cameras software (Kappa Camera Control, http://www.kappa.de). Stomatal apertures were determined using Metamorph software (Molecular Devices, http://www.moleculardevices.com) or ImageJ software (http://rsb.info.nih.gov/ij).

High-quality images (as shown in Figure 1b) were obtained using a water-immersion objective (W-Plant Apochromat; Zeiss). Hyphae of mildew were stained by incubation with a solution containing 2.4 μm fluorescein diacetate. Fluorescein was excited using light from a mercury lamp (LQ-HXP 120-5; Leistungselektronik Jena GmbH, http://www-lej.de) passed through a bandpass filter (470/40 nm), and reflected onto the specimen using a dichroic mirror (500 nm). Emission light was passed through a bandpass filter (500/40 nm) and detected using a cooled CCD EM camera (QuantEM: 512SC, Photometrics, http://www.photometrics.com) using Visiview software (Visitron, http://www.visitron.de).

Micro-electrode measurements

Leaves were fixed to the microscope table using double-sided adhesive tape in the evening, and measurements were performed the next day. For micro-electrode measurements, guard cells were visualized using a water immersion objective in contact with a drop of solution (5 mm KCl, 5 mm potassium citrate pH 6.0, 0.1 mm CaCl₂) on the leaf surface. The leaf was illuminated using the microscope lamp at a photon flux density of 300 μmol m⁻² sec⁻¹.

Guard cell impalement was performed using a piezo-driven MM3A micro-manipulator (Kleindiek Nanotechnik) holding a double-barreled micro-electrode. Micro-electrodes were pulled from borosilicate glass capillaries (inner diameter 0.58 mm, outer diameter 1 mm; Hilgenberg, http://www.hilgenberg-gmbh.de), as previously described (Roelfsema et al., 2001). The electrodes were backfilled with 300 mm KCl or 300 mm CsCl and had a tip resistance of approximately 140 MΩ. Both barrels of the electrodes were connected using Ag/AgCl half cells to head stages (HS-2A × 0.01, input resistance approximately 10¹³ Ω; Axon Instruments, Molecular Devices) connected to a two-electrode voltage clamp amplifier (Geneclamp 500; Axon Instruments). The reference electrode, a 300 mm KCl/2% agarose salt bridge, connected to an Ag/AgCl half cell, was placed in a drop of bath solution between the objective and the cuticle.

Voltage clamp protocols were applied via an ITC-16 interface (HEKA, http://www.heka.com), controlled by pulse software (HEKA). Data were filtered at 200 Hz using a low-pass filter, and sampled at 1 kHz for short pulses, whereas long recordings were sampled at 33 Hz.

Nano-infusion

Uninfected leaves were fixed to the table of an upright microscope as described for micro-electrode measurements, and the same experimental conditions were applied. Single-barreled capillaries were pulled on a horizontal electrode puller (P2000; Sutter Instruments, http://www.sutter.com), and the tip was broken to a diameter of approximately 2 μm. The tip of the capillary was back-filled using a solution containing 10 mm KMes pH 6.0, 1 mg L⁻¹ lucifer yellow and 10 mg L⁻¹ chitosan. Chitosan solution was obtained by dissolving 1 mg ml⁻¹ crab shell chitosan (Sigma-Aldrich, http://www.sigmaaldrich.com) in 1 m acetate. After stirring overnight, insoluble material was removed by centrifugation at 24 000 g for 20 min. The control solution for Nanoinfusion was supplemented with 10 mM K acetate, in order to probe for effects of acid loading of guard cells.

The tip of the nano-infusion capillary was moved into an open stoma using an MHW-3 micro-manipulator (Narishige, http://narishige-group.com). Solutions were injected from the capillary into the sub-stomatal cavity using a pneumatic drug injector (PEDS-02; NPI Electronics, http://www.npielectronic.com) using a pressure of 140 kPa for 2–4 sec. Successful infusion was evident from the spread of fluorescent solution below the abaxial epidermis. Changes in the stomatal aperture of an adjacent stoma, or the ion channel activity of its guard cells, were recorded as described above.

Acknowledgements

We thank Ralph Panstruga (Max Planck Institute for Plant Breeding Research, Cologne, Germany) for helping us get started with the project and for preparation of the manuscript, Ulrich Hildebrand (Lehrstuhl für Botanik II, University of Würzburg, Germany) for useful discussion and supplying barley mildew conidia, and Hubert Felle (Botanisches Institut I, University of Gießen, Germany) for introducing us to the nano-infusion technique. This work was supported by grants from the Deutsche Forschungsgemeinschaft to M.R.G.R. and I.M. (grant number SFB 567) and to S.K. (grant number GK 1342) and from the Deutscher Akademischer Austauschdienst to A.G.-D.