Terrestrial plants rely on stomata, small pores in the leaf surface, for photosynthetic gas exchange and transpiration of water. The stomata, formed by a pair of guard cells, dynamically increase and decrease their volume to control the pore size in response to environmental cues. Stresses can trigger similar or opposing movements: for example, drought induces closure of stomata, whereas many pathogens exploit stomata and cause them to open to facilitate entry into plant tissues. The latter is an active process as stomatal closure is part of the plant's immune response. Stomatal research has contributed much to clarify the signalling pathways of abiotic stress, but guard cell signalling in response to microbes is a relatively new area of research. In this article, we discuss present knowledge of stomatal regulation in response to microbes and highlight common points of convergence, and differences, compared to stomatal regulation by abiotic stresses. We also expand on the mechanisms by which pathogens manipulate these processes to promote disease, for example by delivering effectors to inhibit closure or trigger opening of stomata. The study of pathogen effectors in stomatal manipulation will aid our understanding of guard cell signalling.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Stomata are pores on the surface of leaves, formed by a pair of guard cells that open during the day to allow transpiration and CO2 entry for photosynthesis. The pore aperture is actively adjusted in response to changing environmental conditions such as light intensities, CO2 and drought. A wide range of pathogens takes advantage of these pores, as natural openings on the leaf surface, to gain access to the leaf interior. The importance of stomata as an entry point is only now becoming fully realised (Melotto et al., 2008). Some pathogenic fungi possess the ability to directly penetrate the leaf surface, but others, such as the rusts, specialise in using stomata as an entry route (Grimmer et al., 2012). On nonhost species, rust pathogens do not accurately locate stomata, and so cannot penetrate the leaf and cause infection. It has been suggested that this is due to inappropriate biochemical or thigmotrophic signals from the nonhost plant (Ayliffe et al., 2011; Cheng et al., 2012, 2013). Additionally, some fungi and oomycetes utilise stomata as a means of dispersing spores after successful colonisation of the leaf (Grimmer et al., 2012).
Bacteria are unable to penetrate the cuticle by themselves and require pre-existing openings such as the stomatal pores, hydathodes or wounds (Fig. 1). Far from opportunistically relying on finding an open pore, bacteria actively seek out open stomata (possibly by chemotaxis in response to photosynthetic products; Kroupitski et al., 2009). They also release compounds, called effectors, that promote opening of the pores (Melotto et al., 2008). Gaining access to the leaf interior, therefore, does not simply depend on a chance encounter with a stomate, but in fact involves a sophisticated and specific interaction between the pathogen and its host.
Clearly, guard cells have to integrate multiple, diverse, even conflicting signals and translate them into a physical response, making these highly specialised cells an interesting and challenging area of current research. In this review we will discuss current knowledge of stomatal responses to microbial stresses and compare them with those triggered by abiotic stress.
II. Stimulus-induced closure
Stomatal pores are under rapid, dynamic control, allowing opening in favourable conditions and closure in less favourable conditions. During closure, activation of guard cell outward anion channels leads to membrane depolarisation and so opening of the GUARD CELL OUTWARD RECTIFYING K+ channel (GORK). This decrease in cytosolic [K+] reduces the osmotic potential, water moves out and, as a result, turgor pressure decreases and the two cells deflate, coming together and closing the pore (Roelfsema & Hedrich, 2005; Hedrich, 2012). Changes in aperture are not produced solely through the actions of ion channels and alterations in turgor, but also involve membrane trafficking and reorganisation of the cytoskeleton (see Section IV 'Membrane dynamics during stomatal responses'), as well as alterations to metabolism (reviewed in Vavasseur & Raghavendra, 2005) and transcriptional reprogramming (Cominelli et al., 2005; Liang et al., 2005).
With the advances of Arabidopsis genetics, much has been learned of the molecular components and pathways underlying stomatal regulation. However, most methods to monitor stomatal behaviour, either directly or indirectly, are not suited to screening large populations of mutants (Table 1). For mutant screens, thermal imaging was particularly successful and allowed identification of some of the most pivotal components of guard cell signalling, including OPEN STOMATA 1 (OST1, a kinase involved in initiating downstream signalling), ARABIDOPSIS H+ ATPASE 1 (AHA1/OST2; Merlot et al., 2002, 2007), SLOW ANION CHANNEL 1 (SLAC1; Negi et al., 2008) and BLUE LIGHT SIGNALLING 1 (BLUS1; Takemiya et al., 2013). One of the most intensively studied pathways in guard cells is closure in response to abscisic acid (ABA), the hormone that mediates drought stress (Fig. 2; Cutler et al., 2010; Kim et al., 2010), providing the currently best understood genetic framework of stomatal regulation.
|Equipment||Material||Parameter measured||Selected results||References|
|Thermal camera|| |
|Leaf temperature||Identification of OST1, AHA1, SLAC1-2, BLUS1||Merlot et al. (2002, 2007), Negi et al. (2008), Takemiya et al. (2013)|
|Gas analyser|| |
|Plant/leaf conductance||Integration of stomatal responses with photosynthesis and climate change||Streit et al. (2014)|
|Weight||Development of a simple phenotyping method||Hopper et al. (2014)|
|Pressure probe|| |
|Leaf turgor pressure||Noninvasive monitoring of water relations||Zimmermann et al. (2008)|
|Leaf conductance||Demonstration of circadian rhythms in leaf conductance||Martin & Heidner (1971)|
|Light microscope|| |
Cell contents (e.g. starch and K+)
|Demonstration of circadian rhythms in isolated stomata||Gorton et al. (1989)|
|High-throughput screening system|| |
|Simultaneous screening of complete functional groups||G. Bourdais et al. (unpublished)|
|Fluorescence/confocal microscope|| |
Guard cell protoplasts
Demonstration of cytosolic calcium oscillations
Demonstration of vacuolar changes
|McAinsh et al. (1990), Tanaka et al. (2007)|
|Patch clamp/microelectrode|| |
Guard cell protoplasts
|Ion fluxes||Identification of stretch-activated channels||Cosgrove & Hedrich (1991)|
1. ABA-mediated stomatal closure
ABA is sensed by a family of 14 cytosolic proteins known as PYRABACTIN RESISTANCE/PYRABACTIN RESISTANCE-LIKE/REGULATORY COMPONENTS OF ABA RECEPTORS (PYR/PYL/RCAR; Ma et al., 2009; Park et al., 2009) and a sextuple knockout mutant is necessary to completely inhibit stomatal closure in response to ABA (Gonzalez-Guzman et al., 2012). After binding of ABA, the PYR/PYL/RCAR receptors bind to and repress protein phosphatase 2Cs (PP2Cs), such as ABA INSENSITIVE 1 (ABI1), which are constitutive negative regulators of SUCROSE NON-FERMENTING 1 (SNF1)-related protein kinases (SnRKs, such as OST1). This allows the autophosphorylation of the SnRKs and consequent downstream signalling events (Cutler et al., 2010; Kim et al., 2010). At the whole plant level, a triple SnRK knockout is needed to predominantly eliminate ABA responses (Wang et al., 2013).
Unusually for guard cells, a single SnRK knockout – OST1/SnRK2.6, appears to be sufficient to prevent ABA-induced stomatal closure. OST1 is known to physically interact with many downstream target proteins, including the ion channels POTASSIUM CHANNEL IN ARABIDOPSIS THALIANA 1 (KAT1), SLOW ANION CHANNEL-ASSOCIATED 1 (SLAC1; S-type) and QUICK-ACTIVATING ANION CHANNEL 1 (QUAC1; R-type), and the NADPH oxidases RESPIRATORY BURST OXIDASE HOMOLOGUES D and F (RBOHD/F; Geiger et al., 2009; Sato et al., 2009; Acharya et al., 2013; Imes et al., 2013). Thus, when activated, OST1 not only promotes stomatal closure (activation of the outward anion channels SLAC1 and QUAC1), but also prevents stomatal opening (inhibition of the K+in channel KAT1). It was shown that ost1 mutants did not activate downstream events such as Ca2+ uptake channels or reactive oxygen species (ROS) production, leading Acharya et al. (2013) to conclude that OST1 is a critical limiting component in ABA regulation of stomatal aperture, ion channels and NADPH oxidases.
The PYR/PYL/RCARs-PP2C-SnRK receptor-phosphatase-kinase chain was shown to be the minimum pathway required, in vitro, to recapitulate ABA signalling in plant protoplasts (Fujii et al., 2009). However, the ABA signalling network in guard cells includes many more players, possibly because reducing the pore aperture is not simply activation of ion channels to promote closing, but also inhibition of channels involved in opening, and multiple (even conflicting) signals from different stimuli have to be integrated. Many of the upstream components of ABA signalling also have roles in the response to other stresses. PYR/PYL/RCARs function in darkness, ozone and, to a lesser extent, CO2-induced stomatal closure (Merilo et al., 2013), with the activation of OST1 and S-type anion channels contributing to darkness, ozone, low humidity and CO2-induced closure (Xue et al., 2011; Merilo et al., 2013). The finding that OST1 was more important in CO2-induced closure than PYR/PYL/RCARS led the authors to conclude that there may be another, thus far unidentified, signalling pathway activating OST1 (Merilo et al., 2013).
A possible mechanism for OST1 activation without ABA and ABA receptors could be direct inhibition of ABI1 by ROS (Meinhard & Grill, 2001). ROS production by RBOHD/F has previously been implicated in bicarbonate-induced closure, although it was suggested that the timing and magnitude may differ from that seen with ABA (Kolla et al., 2007). This all suggests a common pathway of stomatal regulation in response to abiotic stress and a convergence point that would appear to be at, or shortly downstream of ABA perception. Interestingly, as OST1 interacts with RBOHs, the inhibition of ABI1 by ROS provides for a positive feedback loop in this section of the signal transduction pathway. Furthermore, as H2O2 can activate plasma membrane calcium channels in guard cells (Pei et al., 2000), and Ca2+ has been shown to activate RBOHD (Ogasawara et al., 2008), there may be another positive feedback loop immediately downstream of this.
2. Hormone interactions in stomatal regulation
Compared with ABA, the effects of the other hormones are less well understood. One of the reasons for this may be that variable stomatal responses have been noted, leading to the conclusion that concentration and relative ratios of the different hormones are important (Acharya & Assmann, 2009). Furthermore, ABA can alter the amounts of other hormones in Arabidopsis and, as the responses of hormones in guard cells are distinct from mesophyll cells, this suggests that these hormone alterations are specifically required for guard cell function (Jin et al., 2013).
Jasmonic acid (JA) was upregulated 30 min after ABA application in guard cells (Jin et al., 2013). JA promotes closure in stomata, as does methyl jasmonate (MeJA), which is readily converted to JA and subsequently jasmonoyl isoleucine (JA-Ile, the active form of JA). MeJA-induced stomatal closure has also been associated with many downstream events common to ABA-induced closure such as cytosolic pH increases, ROS production (through RBOHD/F), NO production, K+out activation and S-type anion channel activation (Gehring et al., 1997; Suhita et al., 2004; Munemasa et al., 2007). JA-Ile is sensed by a protein complex containing CORONATINE INSENSITIVE 1 (COI1), and consequently, MeJA- but not ABA-induced closing is impaired in coi1 (Munemasa et al., 2007). This appeared to be due to diminished activation of S-type anion channels and decreased ROS and NO in coi1 mutants (Munemasa et al., 2007, 2011).
Although the components of the signalling pathway between COI1 and second messenger production/channel activation are unknown, closing occurs within the same timescale as ABA-induced closure (Munemasa et al., 2007). Stomata of JASMONOYL ISOLEUCINE CONJUGATE SYNTHASE 1 (JAR1, catalyst of the last step of JA-Ile synthesis) mutants do not close in response to MeJA, and have reduced closure to ABA (Suhita et al., 2004). Additionally, ost1 has reduced closure to MeJA, whereas rbohD/F and gork1 do not close, indicating shared components at many levels in the two pathways and a possible role for jasmonates in ABA-induced closure (Suhita et al., 2004). Conversely, endogenous ABA is required for JA signalling, as the ARABIDOPSIS THALIANA ABA DEFICIENT 2 (ABA2) mutant has impaired closure, reduced gene induction and suppressed [Ca2+]cyt elevations in response to MeJA, but not to ABA (Hossain et al., 2011). In addition to increasing JA levels, ABA also decreases auxin levels (Jin et al., 2013). Auxin can partially prevent ABA inhibition of opening, so this represents another mechanism whereby ABA promotes closure (increasing JA) and inhibits opening (decreasing auxin).
Salicylic acid (SA) promotes closure and ROS production in stomata, but this appears to be through peroxidases and not NADPH oxidases (Acharya & Assmann, 2009; Khokon et al., 2011). SA-deficient mutants can close in response to ABA, but ABA-deficient (aba2) plants did not close stomata in response to SA (Zeng & He, 2010). More recently, it has been suggested that aba2 plants can close in response to SA (Issak et al., 2013). As there was more than an order of magnitude difference in SA concentrations between the two studies, these results may not be as contradictory as they first seem. Regardless, the interplay between SA and ABA signalling needs further investigation.
Ethylene (ET) has been linked to both opening and closure. In experiments where exogenous application promoted closure (Desikan et al., 2006), this closure required ROS production by RBOHF, with no apparent involvement of RBOHD. Interestingly, the ETHYLENE OVERPRODUCER 1 mutant (eto1) showed reduced closure to ABA. Although both ABA and ET induced closure, when applied together closure was inhibited (Tanaka et al., 2005), indicating an antagonistic relationship between ET and ABA. This fits well with the results of Merritt et al. (2001), where it was suggested that auxin-induced opening is mediated by auxin-induced ethylene production, and points to a more complex picture of hormone interactions in guard cell signalling.
Cytokinins (CKs) have been linked to stomatal opening in planta, and synthesis is inhibited during drought stress. However, exogenously applied CKs show varied responses depending on concentration and CK species, although in general they tend to inhibit ABA-induced closure (Daszkowska-Golec & Szarejko, 2013). Similarly to auxin, CKs have been suggested to act through modulation of ethylene biosynthesis (Tanaka et al., 2006).
ABA, JA, SA, ET, CKs and auxins are all known to function in whole plant immune signalling (Pieterse et al., 2012). CKs can synergistically modulate SA signalling, whereas SA and JA are known to work antagonistically in defence gene expression. Therefore, it would be interesting to see how stomata respond to the combined stimuli of SA and JA. Both induce closure in stomata individually, so antagonism in combination would seem unlikely. As noted above for ET and ABA, however, this is not necessarily the case, and the complex effects of antagonistic abiotic stimuli on stomatal conductance have recently been shown (Merilo et al., 2014). Relative ratios of hormones can be important in stomatal signalling and components linking hormone receptors to second messenger production remain to be elucidated and are possibly where pathways converge. However, Jin et al. (2013) have recently suggested that crosstalk may not be through shared downstream components but rather that ABA operates upstream to control the concentrations of other hormones. Hormone responses in guard cells appear to be the result of interdependent signalling pathways that we only have a superficial understanding of. It is worth noting that as hormones are important regulators of whole plant physiology, so direct effects need to be distinguished from pleiotropic phenotypes (Cao et al., 2011).
3. Microbe-triggered stomatal closure
It is now well established that many plants close their stomata upon perception of microbes, and that this can reduce the severity of infections (Melotto et al., 2006). The importance of this mechanism in plant immunity was further demonstrated in grapevines, whereby plants with stomata that had been pre-closed by ABA treatment showed reduced infection rates (Allègre et al., 2009). Additionally, decreased stomatal numbers have been correlated with decreased susceptibility to downy mildew (Plasmopara viticola) in grapevine cultivars (Alonso-Villaverde et al., 2011), demonstrating the significance of stomatal patterning as well as functionality in plant immunity.
Plants detect the presence of microbes on and in their leaves by responding to molecular signatures termed microbe- or pathogen- associated molecular patterns – MAMPs or PAMPs. MAMPs are sensed by a class of plasma membrane-localised receptors, collectively referred to as pattern recognition receptors (PRRs), which are specific for a particular MAMP and absolutely required for the response. Downstream events of MAMP perception include increases in ROS (from RBOHD alone), NO and Ca2+, activation of G-proteins and anion channels (Fig. 2; Mersmann et al., 2010; Jeworutzki et al., 2010; Macho et al., 2012; Lee et al., 2013). Thus, many downstream components of MAMP signalling are shared with the ABA pathway and, obviously, the final response of stomatal closure is the same (Pei & Kuchitsu, 2005; Acharya & Assmann, 2009; Cutler et al., 2010; Kim et al., 2010). The question of whether ABA is required for MAMP-induced closure of stomata has still not been fully resolved. The finding that ABA DEFICIENT 3 (ABA3) mutants do not close in response to MAMP treatment would seem to support an absolute requirement (Melotto et al., 2006), although recent studies show only partial impairment (Montillet et al., 2013). All these observations raise important questions about the commonalities between ABA and MAMP pathways leading to stomatal closure.
Most research has focussed on stomatal closure in response to bacteria and the bacterial MAMP flg22, the elicitor-active epitope of bacterial flagellin (Boller & Felix, 2009). This is detected by the FLAGELLIN SENSING 2 (FLS2) receptor, which forms a flg22-dependent complex with BRI1-ASSOCIATED RECEPTOR KINASE 1 (BAK1). Cytoplasmic kinases, including BOTRYTIS-INDUCED KINASE 1 (BIK1), associate at the plasma membrane with FLS2 and BAK1 and are phosphorylated upon ligand binding (Dou & Zhou, 2012). Interestingly, BIK1 has recently been shown to directly interact with and activate RBOHD (Kadota et al., 2014; Li et al., 2014), which could implicate RBOHs as a convergence point for MAMP and ABA signalling pathways. The finding that RBOHD can form a complex with FLS2, and that not all BIK1-phosphorylated residues are conserved between RBOHD and RBOHF, may explain why RBOHF is less important for MAMP signalling than for ABA signalling (Mersmann et al., 2010; Kadota et al., 2014).
There is also evidence that the pathways of stomatal regulation may converge at OST1, whereby PP2Cs are inhibited either by PYL/PYR/RCARs (abiotic stress) or ROS (biotic stress). Melotto et al. (2006) were the first to show that ost1 mutants are severely impaired in stomatal closure in response to flg22 and lipopolysaccharides (LPS; another bacterial MAMP). At the whole leaf level, this impairment translated to increased bacterial growth in ost1 after surface inoculation. However, it has recently been suggested that OST1 may not be a main component of microbe-induced closure. An oxylipin signalling pathway is instead postulated, as LIPOXIGENASE1 (LOX1) deficient mutants were more susceptible to Pseudomonas infection and had reduced ability to close stomata in response to bacteria and flg22 (Montillet et al., 2013). Oxylipins have previously been implicated in stomatal opening (Ohashi et al., 2005), suggesting a greater role in guard cell signalling that remains to be fully investigated. Montillet et al. (2013) also showed a lack of OST1 activation upon flg22 treatment, although unfortunately this was at concentrations below those that induce stomatal closure. It will be interesting to see the effect of flg22 on OST1 activity at more guard cell-relevant concentrations.
MAMP-induced closure also shares components with other hormone signalling pathways. Mutants deficient in ETHYLENE RECEPTOR 1 (ETR1) and the downstream signalling component ETHYLENE INSENSITIVE 2 (EIN2) did not close in response to ET, flg22 or bacteria (Desikan et al., 2006; Mersmann et al., 2010). A possible mode of action is the reduced FLS2 transcription in ET mutants (Boutrot et al., 2010). Similarly, SA-deficient mutants have reduced closure in response to flg22 (Zeng & He, 2010), implying that not only ET, but also SA may be required for full flg22-induced closure.
Analogous to the ABA response, where inhibition of opening and promotion of closure were observed, flg22 and yeast elicitor (YEL) both inhibit K+in channels and activate S-type channels (Zhang et al., 2008; Ye et al., 2013). Similar mechanisms are known to operate in the grasses. Chitosan, a form of the fungal MAMP chitin, induces stomatal closure in barley, and light-induced stomatal opening was inhibited by the barley powdery mildew fungal pathogen Blumeria graminis, with both of these stimuli activating S-type anion channels (Koers et al., 2011). This suggests that the general stomatal responses to pathogens are evolutionarily conserved between dicots and monocots. However, less is known about the signal transduction pathways that relay the information from the receptors to the channels and many of these remain to be identified in the grasses.
Important questions include how is OST1 activated during immune signalling and how important is it to overall MAMP-induced closure? Do pathways converge at RBOH and OST1? Are downstream components of RBOH and OST1 the same for all stimuli or is there differential activation? As guard cell signalling has proved to consist of a network rather than a pathway, it seems plausible that there is more than one convergence point.
4. Downstream kinases – CPKs and MAPKs
There has been much discussion as to whether ABA is able to bypass OST1 and initiate closure through CALCIUM DEPENDENT PROTEIN KINASEs (CPKs; Mori et al., 2006; Geiger et al., 2010, 2011). It was shown in oocytes that ABA could activate S-type anion channels independently of OST1 via ABI1 and CPKs (Brandt et al., 2012). However, OST1 is absolutely required for S-type channel activation in planta (Acharya et al., 2013). Nevertheless, CPKs do have roles in both ABA- and microbe-induced closure (Fig. 2). The cpk3/cpk6 double mutant showed greatly reduced stomatal closure in response to ABA (Mori et al., 2006), whereas knockout of CPK6 alone was sufficient to disrupt YEL-induced stomatal closure, inhibition of opening (Ye et al., 2013) and MeJA activation of S-type anion channels (Munemasa et al., 2007, 2011). Flg22 activates CPKs 4, 5, 6 and 11 in mesophyll cells (Boudsocq et al., 2010), but it remains to be seen which CPKs mediate flg22 signalling in guard cells. In addition, SLAC1 and SLOW ANION CHANNEL HOMOLOGUE 3 (SLAH 3) can be stimulated by CPKs 21 and 23, but a stomatal phenotype has yet to be found (Geiger et al., 2010, 2011; Scherzer et al., 2012). Therefore, it would seem that CPK6 is broadly involved in mediating stomatal closure but that further CPKs also play roles. It will be interesting to see whether there proves to be signalling specificity at the level of CPKs.
MITOGEN ACTIVATED PROTEIN KINASES (MAPKs) have been implicated in stomatal closure in response to ABA and MAMPs, although no signalling cascades have been fully established to date. The mpk9 and mpk12 single mutants closed normally, but the double mutant showed reduced closure in response to ABA and exogenously applied H2O2 (Jammes et al., 2009; Montillet et al., 2013). More recently, it has been suggested that an amino acid substitution in MPK12 alone may be enough to impair the ABA inhibition of opening (Des Marais et al., 2014). Both kinases act downstream of ROS production and [Ca2+] increases in YEL-induced stomatal closure (Salam et al., 2013), however, there was no alteration to flg22-induced closure in the double mutant (Montillet et al., 2013). Mutant mpk9/12 plants were more susceptible to P. syringae infection (Jammes et al., 2011), providing some evidence of a role for MPK9 and 12 in MAMP-regulation of stomatal responses.
The predominant MAPKs activated by MAMP elicitation are MPK3, 4 and 6. The mpk3 and mpk6 single mutants both showed impaired closure in response to flg22, although ABA closure was normal. MPK3 antisense plants had reduced closure in response to bacteria and LPS, whereas closure to ABA was normal (Gudesblat et al., 2007, 2009). Interestingly, ABA inhibition of opening was impaired, as were H2O2 responses (Gudesblat et al., 2007). In contrast to MPK3 and 6, which are positive regulators of MAMP-triggered responses, MPK4 acts as a negative regulator of this pathway. Mutant mpk4 plants exhibited enhanced flg22-induced ROS burst (Xu et al., 2014), however, stomata in MPK4 overexpressing plants responded normally to bacteria and flg22 (Berriri et al., 2012).
Although there is convincing evidence for some CPKs and MAPKs in the regulation of stomatal closure in response to MAMPs and ABA, there is considerable redundancy in these signalling networks, as can be seen from the necessity of double mutants to produce a phenotype. Further research is needed to fully elucidate the role of these kinases in stimulus-induced closure of stomata. As previously mentioned, knockout mutants can produce pleiotropic phenotypes and it has been shown that mpk4 has high endogenous amounts of SA (Petersen et al., 2000), which will both affect guard cell responses and produce systemic effects.
5. Integrating guard cell signalling
Spatio-temporal dynamics of signalling components are central to guard cell signalling and have to be considered if we are attempting to integrate distinct pathways. The ABA receptors are cytoplasmic, as is OST1. The channels that need to be activated in order to control turgor, however, are localised at the plasma membrane and the tonoplast, requiring OST1 and CPKs to be present at least transiently at these locations also. This contrasts with the PRRs, which are plasma membrane-localised but can also reach the vacuole upon ligand activation (Beck et al., 2012; Spallek et al., 2013). Spatially distinct receptors could allow sharing of soluble downstream components whilst still allowing for different targets. Conversely, it may mean that a shared target (i.e. the plasma membrane anion channel) has to be activated by a different series of events if mobility of components is a limiting factor.
Most of the well-studied components of guard cell signalling have been resolved spatially, however much less is known about the temporal dynamics. ABA produces peak anion channel currents at 5 min and most of the closure takes place 10–20 min after application (Roelfsema et al., 2004). It was suggested that moderate stomatal responses were mediated by S-type channels whereas responses of greater magnitude involved R-type channels as well, which provide an initial boost, with residual anion currents helping to prevent reopening. Interestingly, it has been proposed that this form of R-type/S-type interplay could ensure a quick response to drought while helping to maintaining sensitivity to other stimuli (Roelfsema & Hedrich, 2005).
Although R-type channels have not yet been implicated in closure in response to microbes, S-type anion channels are involved, and closure takes place over the same 10–20 min timescale as ABA (Koers et al., 2011). It is also worth considering that different branches of the signalling pathway may be producing temporally distinct responses. Thus, there is anion channel activation within minutes, possibly induced by OST1 interacting directly with plasma membrane channels. Subsequently, long-term maintenance of closure involving metabolic and transcriptional changes may be mediated by [Ca2+]cyt and MAPK cascades. The duration of the stomatal response remains to be fully investigated, but it is interesting to note that epidermal peels from plants which had been sprayed with ABA the previous day, will not open to the same extent as untreated plants (Pantin et al., 2013). The recent evidence of ABA affecting the amounts of other hormones (Jin et al., 2013), such as JA increases after 30 min, also implicates this hormone in contributing to the long-term maintenance of ABA-induced closure.
III. Pathogen interference with stomatal control
Stomata of most plants open in the morning in response to the light stimulus. The blue light receptors, PHOTOTROPIN 1 and 2 (PHOT1/2), are located at the plasma membrane and autophosphorylate, activating proton pumps such as AHA1 via 14-3-3 proteins (Shimazaki et al., 2007). This leads to membrane hyperpolarisation and opening of the voltage-gated inward potassium channels KAT1, KAT2 and ARABIDOPSIS K+ TRANSPORTER 1 and 2 (AKT1 and 2), increasing the osmotic potential of the cell and so prompting water to move in, thus increasing the turgor pressure and volume. The cells bow outwards, increasing the pore aperture (Fig. 1). Sucrose is also used as an osmoticum, especially later in the day, and malate accumulates as a counterion to K+ (Talbott & Zeiger, 1996).
Gene regulation is involved in opening and it was recently shown that a bHLH transcription factor binds to the KAT1 promoter, and that this binding was reduced in the presence of ABA (Takahashi et al., 2013). Intriguingly, stomata of slac1 mutants open less in response to light, low humidity and low CO2. This is due to reduced K+in activity, possibly to compensate for the impaired closure response (Laanemets et al., 2013). As described in the previous section, perception of microbes can trigger closure in light-opened stomata, and thus it would seem reasonable that pathogens which could overcome this host defence response would increase their chances of successful colonisation. This could come about in two ways – either by preventing stomatal closure in the first place or by re-opening already closed stomata.
1. Pathogen inhibition of stomatal closure
To date, there are no confirmed instances of effectors directly inhibiting pathways promoting closure of stomata. However, the exact mechanism of many effectors remains to be elucidated. For instance, the bacterial effector HopM1, previously known to trigger proteasome-dependent degradation of host proteins such as ADP ribosylation factor (ARF) guanine nucleotide exchange factor (GEF) AtMIN7/BEN1/BIG5 (Nomura et al., 2006, 2011), was recently shown to affect stomatal closure (Lozano-Durán et al., 2014). In this case, a 14-3-3 protein is the target of HopM1, consistent with these proteins being involved in the inhibition of stomatal closure. Interestingly, it has previously been shown that 14-3-3 proteins can regulate RBOHs (Elmayan et al., 2007), and indeed the microbe-induced ROS burst was reduced in plants expressing HopM1 (Lozano-Durán et al., 2014). Alternatively, HopM1 could indirectly affect FLS2 responses through negative crosstalk with brassinosteroid signalling (Albrecht et al., 2012; Belkhadir et al., 2012), which is upregulated by degradation of a 14-3-3 protein (Lozano-Durán et al., 2013, 2014). Another candidate for effector-triggered inhibition of stomatal closure comes from Xanthomonas campestris, where an uncharacterised effector capable of reversing bacterial, LPS- and ABA-induced closure of stomata has been shown to target the MPK3 pathway (Fig. 3; Gudesblat et al., 2009).
2. Re-opening of stomata by pathogens
Many plant pathogens appear to be able to force stomata to open, to aid either entry or exit. Stomatal opening is promoted by a toxin called coronatine (COR), which is produced by some strains of P. syringae, and interacts with COI1 (Melotto et al., 2006). During infection with a COR-producing bacterial strain, stomata will initially close but later re-open (Melotto et al., 2006). This re-opening is not seen with COR− bacterial strains. As mentioned previously, flg22 can inhibit K+in channels; however, when COR is applied at the same time as flg22, this inhibition is reversed, K+in channels have normal activity and stomata stay open (Melotto et al., 2006; Zhang et al., 2008). This effect is not exclusive to MAMP signalling as COR also prevents ABA-induced closure (Zheng et al., 2012). COR acts through COI1 and NAC transcription factors, without which it is unable to re-open stomata or to prevent-ABA induced closure.
It would seem counter-intuitive that a JA-Ile mimic such as COR would promote stomatal opening when JA has been widely shown to promote closure (e.g. Suhita et al., 2004; Munemasa et al., 2007, 2011). Several suggestions as to the reasons for this have been raised. There is growing evidence of bimodal effects and complex crosstalk between pathways in hormone-induced stomatal signalling. It had previously been suggested that JA might have a bimodal effect (Melotto et al., 2008) in the same way that auxin does (Blatt & Thiel, 1994). However, both Munemasa et al. and Suhita et al. studied MeJA responses over a wide range of concentrations and all induced closure.
Antagonism with other hormones such as ABA or SA has also been discussed (Melotto et al., 2008) and relative hormone concentrations do seem to be very important (Section II 'Stimulus-induced closure'). In an effort to address these conflicting reports of JA and COR responses, Okada et al. (2009) demonstrated dose-dependent opening of Ipomea stomata when using naturally occurring (+)COR in the range 1–100 μM. Naturally occurring (−)jasmonyl isoleucine produced a similar dose-dependent opening response, but racemic (±)jasmonic acid did not. This could imply that some of the contradictory results are due to exogenously applied JA and MeJA not being perceived as intended in vivo. More recently, it has been proposed that MeJA might not act through COI1 (Montillet et al., 2013). This is consistent with COR producing effects in coi1 mutants, in other tissues, leading to the suggestion that it may have multiple targets (Geng et al., 2012).
RESISTANCE TO P. SYRINGAE PV MACULICOLA 1 (RPM1)-INTERACTING PROTEIN 4 (RIN4; Liu et al., 2009a) is necessary for re-opening of stomata by COR. RIN4 is a plasma membrane-associated negative regulator of immune signalling that interacts with AHA1 and AHA2 and is targeted by several P. syringae effectors (Liu et al., 2009b). At a whole plant level, RIN4 associates with the proteins RESISTANCE TO P. SYRINGAE (RPS2) and RESISTANCE TO P. SYRINGAE PV MACULICOLA (RPM1). RIN4 cleavage by the effector AvrRpt2 activates RPS2, whereas RIN4 phosphorylation by the effectors AvrB or AvrRPM1 activates RPM1. When activated, RPS2 and RPM1 initiate defence signalling pathways. The exact mechanisms of RIN4 interaction with AHA1 and AHA2, and the effects of MAMP perception remain to be fully elucidated (Liu et al., 2009a; Elmore & Coaker, 2011).
Fungal pathogens also target the plasma membrane proton pump, the most well studied of these being fusicoccin, produced by the necrotroph Fusicoccin amygdale. This irreversibly activates the pump and is commonly used as a tool in stomatal research (Assmann & Schwartz, 1992). Fusicoccin acts to stabilise the H+-ATPase:14-3-3 complex (Oecking et al., 1997; Baunsgaard et al., 1998; de Boer & de Vries-van Leeuwen, 2012). As plasma membrane proton pumps are one of the first steps towards increasing cell turgor for stomatal opening, they may be a common target for effector proteins (Fig. 3). Other effectors known to reopen stomata include the P. syringae effector syringolin A, whose mode of action is proteasome inhibition (Schellenberg et al., 2010). Together with HopM1 degrading its host targets, this implicates proteasome-related processes as playing a role in both stomatal opening and closure.
Pseudomonas syringae pathovars can secrete 20–30 different effectors during infection (Chang et al., 2005) and this includes molecules that can suppress MAMP-triggered immunity at all levels of the pathway (Dou & Zhou, 2012). Inspection of a plant–pathogen effector interaction network showed no new guard cell targets (Mukhtar et al., 2011). However, this subset of proteins was chosen to reflect whole plant immune responses and further investigations that specifically look at guard cell signalling components are necessary. Given that stomata provide a major entry route for pathogens to access plant tissues, and that pathogens secrete a multitude of effectors, it is surprising that more effectors inhibiting MAMP-induced stomatal closure are not known to date.
Observations of lock-open of stomata in crop species suggest that there are many more effectors still to be investigated, in particular from fungal and oomycete pathogens. Oxalate production by the fungal pathogen Sclerotinia sclerotiorum during infection of Vicia increases stomatal conductance, alters guard cell osmoregulation and interferes with ABA-induced closure (Guimarães & Stotz, 2004). Stomata remained open even in the dark and allowed the pathogen to exit the leaf in regions removed from the initial inoculation site. Plants were less susceptible to oxalate-deficient mutants. The authors also discussed the possibility of oxalate interference with calcium homeostasis as a potential mechanism of action. This is an interesting idea, as Commelina is well documented to have large deposits of calcium oxalate crystals in its subsidiary cells and these have been implicated in calcium homeostasis, allowing the plant to control the amount of calcium in the vicinity of the guard cells (Ruiz & Mansfield, 1994). Potentially, this could be an example of the same molecule being used by both plant and pathogen to regulate the same process.
Lock-open has also been well documented in grape vines. The stomata of plants infected with the oomycete Plasmopara viticola remained open in darkness and during drought stress, even before visible symptoms of infection, whereas uninfected areas produced normal responses. Stomatal lock-open was not related to mechanical obstruction by the pathogen in the sub-stomatal cavity and epidermal peels, once washed, regained the ability to close in response to ABA (Allègre et al., 2007). Previously, in barley, CKs had been implicated in the failure of stomata to close at night when infected with the hemibiotrophic fungus Rhynchosporium secalis (Ayres & Jones, 1975). More recently, the R. secalis secreted toxin NIP1 has been shown to stimulate H+-ATPases, so could potentially promote stomatal opening (van't Slot et al., 2007). It is important to differentiate these specific manipulations of stomatal aperture by pathogens from nonspecific aperture changes. For instance, pathogen-induced cell death in neighbouring cells can cause stomata to lock open due to the decrease in turgor pressure in the surrounding cells (Prats et al., 2006).
Curiously, effectors can work synergistically with ABA to enhance stomatal closure (Goel et al., 2008). HopAM1, from P. syringae, has been suggested to protect developing colonies from osmotic stress and this is supported by observations of pathogen-infected plants showing enhanced closure in the field (Grimmer et al., 2012). These findings indicate that pathogens may have greater control over the regulation of stomata than previously envisaged, whereby they may induce opening or closure depending on their current infection-stage requirements.
IV. Membrane dynamics during stomatal responses
The turgor changes in guard cells during opening and closing of stomatal pores are associated with large and reversible volume changes that coincide linearly with surface area changes of up to 40% (Franks et al., 2001; Shope et al., 2003). Using whole-cell patch-clamp techniques and confocal imaging, fusion and fission of vesicles that resemble endocytic structures have been observed during the swelling and shrinking of guard cell protoplasts and intact guard cells (Homann & Thiel, 1999, 2002; Shope et al., 2003; Meckel et al., 2005, 2007). The exchange of membrane between the plasma membrane and internal compartments, such as the vacuole, during guard cell movement allows the large surface area changes that otherwise exceed the elasticity of bio-membranes (Wolfe et al., 1986). However, endosomal trafficking is not simply for the removal and provision of membrane lipids, but also regulates the delivery, recycling and degradation of plasma membrane proteins. In guard cells, this has been most intensively studied with respect to K+ channels. Sutter et al. (2007) found that ABA triggered selective internalisation of KAT1 into endosomal membrane pools that could later be recycled back to the plasma membrane, indicating that stimulus–response mechanisms include dynamic control of plasma membrane K+ channel density.
Endosomal compartments in plants comprise the trans-Golgi network (TGN), that also functions as the early endosome (EE), and late endosomes (LE)/multi-vesicular compartments (MVBs; Reyes et al., 2011). The nature of KAT1-GFP labelled endosomes remains elusive, as no co-localisation with TGN/EE nor LE/MVB marker proteins could be observed (Sutter et al., 2007). Also, only a few molecular regulators of this guard cell-specific and stimulus-dependent endosomal trafficking have been identified so far. In a previous study, a SOLUBLE N-ETHYLMALEIMIDE-SENSITIVE FACTOR ATTACHMENT PROTEIN RECEPTOR (SNARE) protein from Nicotiana tabacum has been implicated in guard cell responses to ABA (Leyman, 1999) and its homologue in Arabidopsis, SYNTAXIN OF PLANTS 121 (SYP121), functions in the recycling of KAT1 and its distribution to plasma membrane microdomains (Sutter et al., 2006; Eisenach et al., 2012). Similarly, the PROTON ATPASE TRANSLOCATION CONTROL 1 (PATROL1), which belongs to the family of Munc13 proteins that are involved in SNARE complex-mediated exocytotic processes in animal cells, is required for the translocation of AHA1 to the plasma membrane (Hashimoto-Sugimoto et al., 2013). Plants that are affected in SYP121 or PATROL1 function show defects in guard cell opening (Eisenach et al., 2012; Hashimoto-Sugimoto et al., 2013), indicating the importance of SNARE-mediated vesicle fusion processes during stomatal opening.
In general, plant syntaxins are located at acceptor membranes, facilitating membrane fusion events in concert with other SNARE proteins during vesicle-associated trafficking (Lipka et al., 2007). Some syntaxins have specific functions during resistance responses and are involved in the secretion of defence compounds to the sites of pathogen attack. For example, the plasma membrane localised SYP121, also known as PENETRATION RESISTANCE 1 (PEN1), confers pre-invasion resistance to Arabidopsis against powdery mildews via the release of ill-defined antimicrobial compounds (Collins et al., 2003; Kwon et al., 2008). Together with its closest homologue SYP122, it can also function as a negative regulator of diverse defence responses (Z. Zhang et al., 2007). With respect to microbe-induced guard cell responses, it would be interesting to test if the localisation and/or activity of SYP121 is modulated upon pathogen infection either as part of the innate immune response or even by effector molecules targetting KAT1-trafficking. Interestingly, mutants that are affected in FLS2 late endosomal sorting to the vacuole are specifically impaired in flg22-induced stomatal closure (Spallek et al., 2013). This might indicate that, during microbe-induced stomata responses, late endosomal trafficking is of importance in relaying signals from plasma membrane resident receptors to targets further downstream and might be involved in the regulation of ion channels and SNARE trafficking.
1. Cytoskeleton dynamics
Actin has recently been implicated in vacuolar changes during stomatal opening (Li et al., 2013). Mutants with aberrant actin filament reorganisation and dynamics had delayed light-induced opening and increased numbers of small, unfused vacuoles. Open stomata have actin filaments in a radial pattern around the pore, whereas closed stomata have randomly oriented actin fragments (Eun & Lee, 1997). ABA causes disintegration of the radial pattern of actin but has little effect on microtubules (Eun & Lee, 1997; Lemichez et al., 2001). Several components of actin remodelling in ABA-induced closure have been identified, such as the ROP GTPase RAS-RELATED C3 BOTULINUM TOXIN SUBSTRATE1/RHO-RELATED PROTEIN FROM PLANTS6 (Rac1/ROP6) complex (Lemichez et al., 2001; Jiang et al., 2012). Additionally, actin regulates plasma membrane channels involved in both opening (K+in channels) and closing (stretch-activated Ca2+ channels) of stomata (Hwang et al., 1997; W. Zhang et al., 2007). Actin plays an important role in guard cell dynamics and ion channel regulation, but our knowledge of how this integrates with stimulus signalling pathways is limited.
Endomembrane and organelle movement predominantly depend on the actin cytoskeleton and myosin motor proteins, but microtubules have also been implicated in endocytic transport processes. Dynamic changes of microtubules in guard cells in response to ABA, light, CO2 and humidity have been described (Yu et al., 2001; Lahav, 2004; Eisinger et al., 2012a,b; Higaki et al., 2012). Microtubules can function as anchors to slow down the motility of organelles close to the plasma membrane (Crowell et al., 2009) and endosomes (Ambrose et al., 2013) to allow exo- and endocytotic processes. Therefore, a similar role for microtubules in tethering vesicles to the cell cortex to allow an exchange of membrane portions and cargo with the plasma membrane during guard cell movements may occur. Chemical and genetic interference with microtubule function impaired guard cell movement (Fukuda et al., 1998; Marcus et al., 2001; Eisinger et al., 2012a), although there are conflicting reports (Assmann & Baskin, 1998). Nevertheless, a recent publication describes a signalling pathway leading to ABA-induced microtubule depolymerisation during stomatal closure (Jiang et al., 2014), providing a further link between microtubule dynamics and guard cell movement. Thus, it seems reasonable that microtubules do have functions during guard cell movement, possibly by regulating membrane dynamics and cell wall properties.
Microtubules and the actin cytoskeleton both function in defence responses and undergo rapid re-arrangements during biotic interactions (Hardham et al., 2007, 2008; Hardham, 2013). Perception of chitin, elf26 (the elicitor-active epitope of bacterial elongation factor Tu) and flg22 induces a transient increase in actin filament abundance, which is thought to be the result of increased actin polymerisation and/or decreased actin filament disassembly (Henty-Ridilla et al., 2013, 2014). Interestingly, ACTIN DEPOLYMERISING FACTOR4 (ADF4) is required for actin remodelling in hypocotyl cells in response to elf26, but not chitin, indicating specificity in MAMP-induced actin dynamics (Henty-Ridilla et al., 2014). An intact actin cytoskeleton is necessary for correct subcellular trafficking of FLS2 after ligand-induced internalisation (Beck et al., 2012), but it remains to be seen whether flg22 induces more global changes in actin filament abundance in guard cells and how this fits in with the depolymerisation necessary for stomatal closure (Eun & Lee, 1997; Gao et al., 2009).
In order to overcome resistance, pathogens have employed strategies to interfere with these responses. Henty-Ridilla et al. (2013) observed that, although the MAMP-induced increase in actin filaments is related to early immune responses, a successful infection by virulent bacteria is associated with a decrease in actin filament arrays and increased actin filament bundling. To date, only a few effectors are known that specifically manipulate cytoskeleton function. One example is the bacterial effector HopZ1a, which disrupts microtubule networks and secretory trafficking through acetylation of tubulin (Lee et al., 2012). Similarly, harpin, one of the first effectors identified (Wei et al., 1992), induced microtubule disruption and actin microfilament bundling in Vitis and tobacco suspension cultures (Qiao et al., 2010; Guan et al., 2013). Interestingly, flg22 was also shown to disrupt microtubule networks and actin filaments in Vitis cells (Chang & Nick, 2012).
It remains to be seen whether effectors that interfere with cytoskeleton and vesicle trafficking also act in guard cells or whether their secretion and/or activity is spatio-temporally regulated, so that they are produced at later time points when bacteria have already accessed the mesophyll. Temporally controlled effector secretion has been described for some hemibiotrophic pathogens (Kelley et al., 2010; O'Connell et al., 2012). As vesicle trafficking and cytoskeleton dynamics are important during guard cell responses, it seems possible that pathogens that rely on stomatal pores to gain access to the plant interior could target these processes to interfere with guard cell-specific defences.
V. Conclusions and outlook
It is becoming clear that OST1, as well as RBOHs, plays a central role as a common regulator of many guard cell signal transduction pathways. In addition to shared components, we are also starting to discover areas of specificity. For instance, MAPKs may be specific to certain pathways. It is also apparent that the redundancy seen in ABA signalling, that necessitates the generation of double mutants, is not so evident in pathogen signalling. It will be exciting to discover which components are specific to the individual pathways and what they contribute to these pathways.
The many interactions between stomata and microbes seem to make a difference to whether a pathogen is able to produce infection in the host plant. Common targets for effectors reversing stomatal immunity are coming to light, such as H+-ATPases, 14-3-3 proteins and proteasome-related processes. The full repertoire and mechanisms by which pathogens manipulate stomata remain to be determined. However, a common theme of tailoring stomatal apertures to the pathogen lifestyle seems to be emerging.
A recurrent topic of stomatal research is efforts to model stomatal responses, ranging from climate models down to the subcellular level. Subcellular level models include pressure changes within guard cells and neighbouring cells (Cooke et al., 1976), the ABA signalling network (Li et al., 2006) and ion channels (Hills et al., 2012). Models such as these can be used in conjunction with experiments to help determine the mechanisms involved in stimulus-induced responses. Recent technological advances such as high-throughput imaging have allowed the generation of large datasets at subcellular resolution in intact plants (Salomon et al., 2010), and are suitable for identifying and quantifying stomatal phenotypes (Spallek et al., 2013; Lozano-Durán et al., 2014). The accelerated progress of imaging techniques and analyses makes gathering large, high-quality datasets easier and will hopefully lead to increased understanding of guard cell dynamics at both the cellular and subcellular levels. These data will provide a useful basis for producing and testing stomatal models. This, in turn, will aid our understanding of the full extent of microbe-regulated stomatal responses and the interplay with the environment.
We would like to thank members of the Robatzek laboratory for fruitful discussions, and Richard Morris (JIC) for critically reading the manuscript. Research in S.R.'s laboratory is supported by the Gatsby Charitable Foundation and the European Research Council (ERC STORM).