Plant cells maintain high Ca2+ concentration gradients between the cytosol and the extracellular matrix, as well as intracellular compartments. During evolution, the regulatory mechanisms, maintaining low cytosolic free Ca2+ concentrations, most likely provided the backbone for the development of Ca2+-dependent signalling pathways. In this review, the current understanding of molecular mechanisms involved in Ca2+ homeostasis of plants cells is evaluated. The question is addressed to which extent the mechanisms, controlling the cytosolic Ca2+ concentration, are linked to Ca2+-based signalling. A large number of environmental stimuli can evoke Ca2+ signals, but the Ca2+-induced responses are likely to differ depending on the stimulus applied. Two mechanisms are put forward to explain signal specificity of Ca2+-dependent responses. A signal may evoke a specific Ca2+ signature that is recognized by downstream signalling components. Alternatively, Ca2+ signals are accompanied by Ca2+-independent signalling events that determine the specificity of the response. The existence of such parallel-acting pathways explains why guard cell responses to abscisic acid (ABA) can occur in the absence, as well as in the presence, of Ca2+ signals. Future research may shed new light on the relation between parallel acting Ca2+-dependent and -independent events, and may provide insights in their evolutionary origin.
The physiological function of calcium ions is normally assigned to their role in signalling mechanisms, and a large number of publications have provided evidence for this role of Ca2+ in animal and plant cells (Sanders et al. 2002; Berridge, Bootman & Roderick 2003; Hetherington & Brownlee 2004). Because of the overwhelming evidence for the role of Ca2+ in cellular signalling pathways, its role as a structural component of the extracellular matrix in plants is sometimes overlooked. Plant cells maintain free Ca2+ concentrations of about 10−4m in the cell wall and vacuole, whereas the basal cytosolic concentration approximates 10−7m (Miller & Sanders 1987; Felle 1988; Bethmann et al. 1995; Felle & Hepler 1997; Felle et al. 2000). These large concentration gradients imply that cells are equipped with an efficient machinery to transport Ca2+ from the cytoplasm to extra- or intracellular storage compartments. This review attempts to give an overview of molecular mechanisms giving rise to Ca2+ homeostasis in plant cells. The signalling pathways controlling the cytosolic free Ca2+ concentration are probably the most ancient Ca2+-dependent processes in cells (Case et al. 2007). We therefore explore the possibility that these pathways have given rise to the regulatory mechanisms of other ion transporters by Ca2+-dependent mechanisms. Finally, we discuss several models explaining the mode of action of Ca2+ signals in plant cells.
Maintenance of low cytosolic Ca2+ concentrations is essential for all living organisms, because elevated concentrations of Ca2+ will lead to the aggregation of proteins and nucleic acids, as well as the precipitation of calcium phosphate (Jaiswal 2001). A tight control of the intracellular Ca2+ concentration therefore must have developed very early in evolution. Indeed, Ca2+-dependent signalling mechanisms are common in all organisms including prokaryotes (Shemarova & Nesterov 2005). Because of the widespread function of Ca2+ in signalling, it may also have served as a unifying signal during the establishment of endosymbiotic relations (Bothwell & Ng 2005). During the development of eukaryotic cells, the common Ca2+ signal could have enabled communication between the newly adapted organelles. Furthermore, the presence of intracellular storage compartments allowed the generation of Ca2+ signals independent of extracellular Ca2+. It is likely that based on these properties, the role of Ca2+ has been expanded to a central element in signal transduction of eukaryotes, in the course of evolution (Case et al. 2007).
The Ca2+ transport mechanisms and Ca2+-dependent signalling mechanisms thus are likely to have a common origin in all eukaryotes. However, even though plants, fungi and animals may share these basic elements, many details will differ, because the kingdoms have branched early at the evolutionary tree (Baldauf 2003). For this reason, Ca2+ signalling models, based on well-studied animal systems, may not always apply for plant cells (Wheeler & Brownlee 2008). In order to understand Ca2+ signals in plants, we thus must be able to discriminate between conserved mechanisms and those that have diverged during evolution.
Ca2+ EXPORT FROM THE CYTOSOL
Root hair cells of Sinapsis alba are able to keep the cytosolic free Ca2+ concentration as low as 300 nm in the presence of 10 mm Ca2+ in the extracellular medium (Felle, Tretyn & Wagner 1992). In order to maintain such a low cytosolic Ca2+ concentration, Ca2+ ions thus have to be transported against a steep concentration gradient. In addition, the positively charged molecules are often transported against a very negative membrane potential, contributing to a large electrochemical gradient for Ca2+ ions. In Arabidopsis thaliana, this function appears to be carried out by several members of the Arabidopsis calcium ATPase (ACA)-encoded P-type ATPases. These ATP-driven pumps are members of a large protein family that are characterized by a phosphorylated (P) intermediate state of the protein during the pump cycle (Sze et al. 2000; Axelsen & Palmgren 2001). Plasma membrane Ca2+-ATPases belong to the group of P-type 2B proteins that share high homology to the plasma membrane calcium ATPase (PMCA) pumps from animals (Axelsen & Palmgren 2001). However, animal PMCA pumps are localized exclusively in the plasma membrane, whereas several P-type 2B proteins localize to intracellular membranes in Arabidopsis (Hong et al. 1999; Geisler et al. 2000; Lee et al. 2007). The 2B class of P-type ATPases is characterized by the presence of an N-terminus that acts as an auto-inhibitory domain. This N-terminal auto-inhibitory domain is released from the catalytic side of the ATPase after binding calmodulin in the presence of Ca2+ (Baekgaard, Fuglsang & Palmgren 2005; Baekgaard et al. 2006). Because of this mechanism, the type 2B pumps are likely to become active at high intracellular Ca2+ concentrations, and deactivate at low Ca2+ levels.
A second group of Ca2+-ATPases, classified as 2A P-type, lacks the auto-inhibitory domain that is present at the N-terminus of type 2B pumps (Sze et al. 2000). The founding member of the plant 2A P-type ATPases localizes to the endoplasmic reticulum and was named endoplasmic reticulum-type Ca2+ ATPase (ECA1) (Liang et al. 1997). However, other members of this gene family may reside in other membranes, as ECA3 was found in the Golgi apparatus (Mills et al. 2008). The genes encoding the type 2A Ca2+-ATPases show high sequence similarity to endo- and sarcoplasmic reticulum pumps of animal cells (Sze et al. 2000). The most prominent member of these pumps is the sarcoplasmic reticulum calcium (SERCA) pump operating in muscle cells. Various states in the pump cycle of SERCA proteins have recently been crystallized, which revealed that these pumps possess a binding pocket for two Ca2+ ions in the core of the transmembrane spanning domain (Olesen et al. 2007). Hydrolysis of ATP causes conformational changes that enable entry of Ca2+ to the binding pocket from the cytoplasm, followed by the extrusion into the sarcoplasmic reticulum. The stoichiometry of two Ca2+ ions transported per hydrolysed ATP molecule theoretically enables this pump to build up a Ca2+ concentration gradient of 104 (Olesen et al. 2007). In fact, transport of Ca2+ will be influenced by counter-transport of two H+ ions per cycle (Yu et al. 1993). In vacuoles, the pH of the lumen is normally acidic, whereas the cytosol has a pH approximating 7.5. Given that vacuolar P-type ATPases also display a stoichiometry of two Ca2+ ions per ATP, the counter-transport of two H+ ions will enable these pumps to easily overcome the Ca2+ concentration gradient in the range of 103–104, and a vacuolar membrane potential of approximately 15 mV, negative at the cytosolic side (Lew 2004); expressed as Vm = Vcytosol – Vnon-cytosol according to Bertl et al. (1992). The latter values are based on vacuolar Ca2+ concentrations ranging from 0.1 to 2 mm, and cytosolic free Ca2+ concentrations of 200–500 nm (Miller & Sanders 1987; Felle 1988; Bethmann et al. 1995). P-type ATPases thus are likely to play an essential role in maintaining low cytosolic, as well as high vacuolar, Ca2+ concentrations.
In addition to the P-type Ca2+-ATPases, a group of high capacity cation/H+ exchangers (CAX) contributes to Ca2+ homeostasis in plant cells (Shigaki & Hirschi 2006). The Arabidopsis CAX gene family exhibits six members that can be separated into two groups based on sequence homology. It is likely that proteins encoded by members of both groups can transport Ca2+, because CAX1 and CAX2 are in different groups, but both can rescue a yeast mutant lacking endogenous Ca2+ transporters (Hirschi et al. 1996). The so-far characterized CAX-encoded transporters localize to the vacuolar membrane (Shigaki & Hirschi 2006), and thus are likely to sequester Ca2+ into the vacuole. In addition to the vacuole, the plasma membrane also seems to harbour Ca2+/H+ exchange proteins (Kasai & Muto 1990), but the genes encoding these transporters have not been identified yet.
Based on their affinity and capacity, different roles for Ca2+-ATPases and Ca2+/H+ exchange proteins have been postulated. The P-type ATPases in general have a high affinity for Ca2+ (Km 0.1–2 µm; Sze et al. 2000), whereas CAX1 was found to exhibit a relative low affinity (Km 10–15 µm; Hirschi et al. 1996) and high capacity. For this reason, it can be assumed that the Ca2+-ATPases are responsible for regulating the low basal cytosolic free Ca2+ levels, while CAX transporters may enable the removal of excess Ca2+ after a Ca2+ burst. However, physiological evidence that supports the assumed functional difference between both types of Ca2+ transporters is still awaited.
Ca2+ ENTRY INTO THE CYTOSOL
Apart from the extracellular compartment and endoplasmic reticulum (Klüsener et al. 1995) that are utilized as Ca2+ stores by animal cells, plant cells possess additional stores from which Ca2+ can enter the cytosol. In plant cells, the central vacuole most likely plays a major role in Ca2+ homeostasis. Furthermore, plastids probably also function in generating cytoplasmic Ca2+ signals (Nomura et al. 2008; Weinl et al. 2008). Ca2+ will normally enter the cytoplasm through Ca2+-permeable channels that mainly have been studied for the plasma membrane and the vacuolar membranes. Here, an overview will be presented of the knowledge obtained for channels within these two membranes and the genes that may encode them.
A large number of studies have been reported on cation channels, in the plasma membrane of plant cells, that are permeable for Ca2+. Most of these channels show little selectivity for the type of cations conducted. An exception are the stretch-activated channels in guard cells (Cosgrove & Hedrich 1991), as well as onion epidermal cells (Ding & Pickard 1993), which displayed selectivity for Ca2+. Recently, stretch-activated channels in the plasma membrane of root cells were found to be encoded by MSL genes [mechano-sensitive channel of small conductance-like (MscS)] (Haswell et al. 2008). However, the 2 out of 10 characterized members of the MSL gene family displayed a higher conductivity for Cl- than to Ca2+.
Plant cells appear to differ fundamentally from animal cells with respect to the voltage dependence of Ca2+-conducting channels in the plasma membrane. Whereas depolarization-activated Ca2+ channels dominate in most animal cells (Dolphin 2005), the Ca2+ conductance of plant plasma membranes normally is enhanced upon hyperpolarization. The existence of an inward rectifying Ca2+-permeable conductance was already proposed in a study on plasma membrane H+-ATPases (Lohse & Hedrich 1992). Later studies confirmed the presence of these channels in tomato suspension cultured cells (Gelli & Blumwald 1997), guard cells (Hamilton et al. 2000; Pei et al. 2000), mesophyll cells (Stoelzle et al. 2003) and root epidermal cells (Kiegle et al. 2000; Véry & Davies 2000). The ability of these channels to trigger a hyperpolarization-induced increase of the cytosolic Ca2+ concentration was shown for intact guard cells, loaded with the Ca2+ indicator dye FURA-2 (Grabov & Blatt 1998; Levchenko et al. 2005). In later experiments, this particular voltage dependence was used to trigger cytosolic Ca2+ signals through hyperpolarization, using low extracellular K+ concentrations (Allen et al. 2001; Konrad & Hedrich 2008).
In addition to their voltage-dependent regulation, a number of signals have been identified that enhance the activity of hyperpolarization-activated Ca2+ channels. In general, these channels seem to be stimulated by reactive oxygen species (ROS), because H2O2 enhanced channel activity in guard cells (Pei et al. 2000), mesophyll cells (Dietrich, unpublished results), as well as root epidermal cells (Demidchik, Shabala & Davies 2007). Furthermore, specific signals stimulate hyperpolarization-activated channels in a cell type-specific manner. The fungal elicitor IF4 was able to stimulate hyperpolarization-activated channels in tomato suspension cells (Gelli, Higgins & Blumwald 1997), blue light activated them in mesophyll cells (Stoelzle et al. 2003) and abscisic acid (ABA) enhanced their activity in guard cells (Hamilton et al. 2000). These data suggest that various signals act in a cell-specific manner through modulation of the activity of hyperpolarization-activated Ca2+-permeable channels.
Just as with hyperpolarization-activated channels in tomato suspension cells (Gelli et al. 1997), the fungal elicitor Pep-13 activates voltage-independent Ca2+-permeable channels in parsley suspension cells (Zimmermann et al. 1997). This suggests overlapping functions for hyperpolarization-activated and voltage-independent channels with respect to pathogen defence responses. Both pathways for Ca2+ entry, however, may differ with respect to the process that switches off Ca2+ influx. Ca2+ entry through hyperpolarization-activated channels will be terminated upon depolarization of the cell, whereas voltage-independent channels require other feedback signals for deactivation.
Voltage-independent Ca2+-permeable channels were also found in the plasma membrane of wheat, as well as Arabidopsis, root cells (Demidchik et al. 2002a; White & Davenport 2002). It is therefore likely that these channels serve in Ca2+ uptake by roots from the soil, but the way in which these channels are regulated is unknown. Voltage-independent channels normally should be deactivated in intact cells, because a constitutive activity of Ca2+ channels will be lethal. A low basal activity indeed was found for the weakly voltage-dependent, cold-activated Ca2+ channels in mesophyll cells (Fig. 1) (Carpaneto et al. 2007).
In Arabidopsis, SV channels are most likely encoded by the two-pore channel 1 (TPC1) gene (Peiter et al. 2005). Loss of the TPC1 gene in tpc1-2 mutants resulted in a decreased sensitivity for ABA during seed germination, and a reduced stomatal closure induced by high extracellular Ca2+ (Peiter et al. 2005). These data suggest that TPC1 is involved in Ca2+ signalling, even though various other Ca2+-dependent responses were found to be unaffected in the tpc1-2 loss of function mutant (Ranf et al. 2008). A role for TPC1 in cation homeostasis was strengthened by the phenotype of fatty acid oxygenation up-regulated 2 (fou2), a mutant with a point mutation in a domain of TPC1 directed towards the vacuolar lumen (Bonaventure et al. 2007). The SV channels of fou2 display a reduced sensitivity to luminal Ca2+, and their vacuoles accumulate higher Ca2+ and lower K+ levels than wild type (Fig. 2) (Beyhl et al. 2009). It is likely that the higher luminal Ca2+ concentration of fou2 vacuoles causes increased Ca2+ signal strength, in responses that involve activation of Ca2+ channels other than the SV channels.
Even though it is likely that vacuolar Ca2+ release occurs in plant cells (Knight, Trewavas & Knight 1996; Knight & Knight 2000), it is unclear how extracellular signals trigger the activation of Ca2+-permeable channels in vacuolar membranes. Intracellular messengers that are important for Ca2+ signalling of animal cells, such as inositol[1,4,5]triphosphate (IP3) and cyclic ADP-ribose (cADPR), were shown to activate cation channels in vacuoles (Sanders et al. 2002; McAinsh & Pittman 2009). However, no homologous genes for the respective animal channels are present in the Arabidopsis genome (Wheeler & Brownlee 2008), and the molecular nature of ligand-gated channels in the vacuole remains unknown.
Genes encoding Ca2+-permeable channels
Cyclic nucleotide gated channels
The Arabidopsis genome harbours several gene families that are likely to encode Ca2+-permeable channels. The genes with high homology to the animal cyclic nucleotide-gated channels are the best candidates in this respect (Talke et al. 2003). Several of these genes can act as negative regulators (Clough et al. 2000; Balague et al. 2003; Jurkowski et al. 2004) or positive regulators (Yoshioka et al. 2006) of pathogen defence responses, which in turn have been associated with cyclic nucleotide- and Ca2+-dependent signalling (Durner, Wendehenne & Klessig 1998). Additional evidence for their role in Ca2+ signalling was obtained for Arabidopsis pollen lacking the AtCNGC18 (cyclic nucleotide gated channel 18) encoded channel. Loss of AtCNGC18 resulted in the inability of cngc18-1 pollen grains to develop pollen tubes, the growth of which is known to depend on a Ca2+ influx at the pollen tip (Frietsch et al. 2007). Furthermore, the expression of AtCNGC18 in Escherichia coli caused enhanced Ca2+ accumulation. Despite of the indications that at least some of the AtCNGC genes encode Ca2+-permeable channels, the biophysical analysis of these channels in heterologous expression systems has encountered only limited success (Leng et al. 2002; Demidchik & Maathuis 2007). Recent studies with guard cell protoplasts, however, have confirmed that the AtCNGC2 channel is permeable for Ca2+ (Ali et al. 2007).
Plant glutamate receptor-like channels
The second family of genes, which is likely to encode Ca2+-permeable channels based on sequence homology, are the ionotropic glutamate receptor-like (GLR) genes (Lacombe et al. 2001). The role of glutamate receptor-like channels in Ca2+ signalling is supported by the ability of glutamate and several amino acids to induce a Ca2+ influx in root cells (Dennison & Spalding 2000), as well as in mesophyll cells (Meyerhoff et al. 2005). It is likely that the root cell response is mainly exerted by AtGLR3.3, because loss of AtGLR3.3 in the glr3.3-1 and glr3.3-2 mutants leads to smaller glutamate-induced depolarization of the plasma membrane, as well as loss of the glutamate-triggered rise of the cytosolic free Ca2+ concentration (Qi, Stephens & Spalding 2006). Glutamate-induced increases in the cytosolic Ca2+ concentration and the associated depolarization of the membrane potential are transient in nature (Meyerhoff et al. 2005; Qi et al. 2006). This feature is likely to result from a desensitization of the channels during a prolonged exposure to agonists. The potency to desensitize the channels, however, can vary for different amino acids (Stephens, Qi & Spalding 2008).
Just as with the cyclic nucleotide-gated channels, limited success with heterologous expression has hampered a detailed biophysical analysis of glutamate-activated channels. Recent experiments with chimerical proteins constructed from animal ionotropic glutamate receptor and homologous plant genes, however, could confirm a Ca2+ permeability for the pore domains derived from the AtGLR1.1 and AtGLR1.4 plant glutamate receptor-like genes (Tapken & Hollmann 2008). At least some of the plant glutamate receptor-like proteins are therefore likely to encode voltage-independent Ca2+-permeable channels in the plasma membrane. Although it is generally assumed that plant cyclic nucleotide-gated and glutamate receptor-like channels reside in the plasma membrane, localization in the vacuole or organelles of some members of these gene families cannot be ruled out.
In addition to the genes that display sequence homology to animal Ca2+ channels, plant genomes may harbour unexpected genes encoding Ca2+ channels. Channels and co-transporters can have a high amino acid sequence similarity, as has become evident from studies with the Cl- channel (CLC) and slow anion channel-associated 1 (SLAC1) encoded anion channels. The CLC genes were first described to encode Cl--selective channels (Jentsch, Steinmeyer & Schwarz 1990), but later some of the CLC gene products were shown to facilitate Cl-/H+ antiport (De Angeli et al. 2006; Miller 2006). The opposite was true for the SLAC1 and SLAC1 homologs (SLAH) genes, which were annotated as dicarboxylate transporters because of sequence similarity with the yeast transporter malate permease (MAE1) (Grobler et al. 1995). Instead, it turns out that SLAC1 and SLAH genes most likely encode plasma membrane anion channels (Negi et al. 2008; Vahisalu et al. 2008). Recent data obtained by functional expression of SLAC1 in Xenopus oocytes actually show that SLAC1 is not permeable for malate (Geiger et al. 2009a). Furthermore, microbial-type rhodopsins can encode light-driven proton pumps (bacteriorhodopsins), as well as cation-permeable light-gated channels (channelrhodopsins) (Nagel et al. 2002, 2003). Apparently, it is not always possible to predict if a gene encodes an ion channel, co-transporter or pump, based on the amino acid sequence. For this reason, it is very well possible that some genes annotated as cation transporters or pumps in Arabidopsis, in fact act as Ca2+-permeable channels.
KEEPING STEADY Ca2+ LEVELS
Ca2+-dependent feedback regulation
The basic function of the Ca2+ transport proteins discussed earlier is probably to keep stable cytosolic free Ca2+ levels. This implicates that the influx of Ca2+ has to be precisely balanced by its extrusion to extracellular or intracellular compartments. Coupling of Ca2+ efflux and influx is accomplished by linking the activity of these transporters to concentration of free Ca2+ in the cytosol. This type of feedback regulation is most obvious for the P-type 2B Ca2+-ATPases that are regulated by the auto-inhibitory domain binding Ca2+-calmodulin (Baekgaard et al. 2005). At high intracellular free Ca2+ concentrations, the auto-inhibitory domain is released from the catalytic side, and the Ca2+ pump becomes active.
The P-type 2A Ca2+-ATPases lack the auto-inhibitory domain and thus cannot be directly regulated by Ca2+. Instead, these Ca2+-ATPases seem to be regulated through a more indirect mechanism, which involves two small membrane proteins, phospholamban and sarcolipin, in the homologous SERCA pumps (Gramolini et al. 2006). The P-type 2A ATPases thus seem to represent constitutive active Ca2+ pumps, whereas P-type 2B proteins are activated at elevated cytosolic Ca2+ concentrations.
Regulation through Ca2+-dependent protein kinases
Instead of a direct regulation by Ca2+ or Ca2+ calmodulin, transport proteins can also be regulated by Ca2+-dependent protein kinases. For instance, Ca2+-dependent protein kinases were shown to activate an anion channel in the vacuole membrane of V. faba guard cells (Pei et al. 1996). Likewise, ABA activation of the hyperpolarization-activated Ca2+-permeable channels in the plasma membrane of guard cells is affected by the loss of CPK3 and CPK6 function (Mori et al. 2006), which represent calcium-dependent protein kinases (CDPKs). The CDPKs belong to the CDPK–SNF-related kinases (SnRK) superfamily of protein kinases, which harbour a large number of Ca2+-regulated serine–threonine protein kinases (Hrabak et al. 2003). Many of these protein kinases have an auto-inhibitory junction domain, acting as a pseudosubstrate, which connects the N-terminal kinase domain with the C-terminus. In CDPKs, the auto-inhibitory domain is released from the catalytic site after binding of Ca2+ by a calmodulin-like domain in the C-terminus (Hrabak et al. 2003).
In plants, no genes have been identified with high homology to the animal calmodulin-activated protein kinases (Hrabak et al. 2003). Instead, calcium calmodulin protein kinases (CCaMKs) are present in several plant species, but not in Arabidopsis. These CCaMKs have Ca2+-binding EF-hands in addition to a Ca2+–calmodulin binding site and require Ca2+, as well as Ca2+–calmodulin for activation (Hrabak et al. 2003). Two homologous CCaMK proteins are involved in the Ca2+-dependent formation of infection structures during nodulation (Levy et al. 2004; Tirichine et al. 2006).
The group of SnRK3 protein kinases also belongs to the CDPK–SnRK superfamily. These proteins do not directly bind Ca2+ or Ca2+–calmodulin, but instead are regulated through interaction with calcineurin B-like (CBL) proteins that in turn bind calcium (Albrecht et al. 2003; Gong et al. 2004; Hedrich & Kudla 2006; Waadt et al. 2008). Because of this interaction, the SnRK3 proteins also have been named CBL-interacting protein kinases (CIPKs). The vacuolar CAX1 transporter is activated by the CIPK24/SOS2 (salt overly sensitive 2) protein kinase (Cheng et al. 2004), thus providing a potential link between Ca2+ and the activity of H+/Ca2+ transporters. CIPKs seem to play a general role in the regulation of ion transporters, as CIPK24/SOS2 activates the H+/Na+-antiporter SOS1 (Quintero et al. 2002; Gong et al. 2004), and the activity of the AKT1 K+ channel depends on CIPK23 and CBL1/9 (Xu et al. 2006; Geiger et al. 2009b).
Voltage-dependent feedback regulation
In animal cells, voltage-dependent Ca2+ channels are activated during the depolarizing phase of action potentials (Hille 1992; Dolphin 2005). The activity of these channels is reduced again during a subsequent hyperpolarizing phase that is to some extent caused by the Ca2+-dependent activation of K+ channels (Sanguinetti & Tristani-Firouzi 2006). The activation of K+ channels by Ca2+ thus represents a negative feedback loop that may have evolved early in the evolution of prokaryotes (Case et al. 2007). Apart from preventing a potential harmful excessive influx of Ca2+, the delayed feedback also causes a transient increase in the Ca2+ concentration representing a cytosolic signal. A similar feedback system is likely to operate in vascular plants, but here plasma membrane Ca2+ channels are normally activated upon hyperpolarization (Gelli & Blumwald 1997; Hamilton et al. 2000; Pei et al. 2000). A hyperpolarization-induced Ca2+ signal can be turned off during a subsequent depolarization caused by a Ca2+-dependent activation of anion channels (Schroeder & Hagiwara 1989; Hedrich et al. 1990). Three types of Ca2+-activated plasma membrane anion channels in plants may perform the latter function, which are characterized by either a strong deactivation at negative potentials (Keller et al. 1989; Kolb et al. 1995), a weaker deactivation (Schroeder & Hagiwara 1989; Linder & Raschke 1992) or instead activation (Elzenga & VanVolkenburgh 1997). The slowly deactivating (S-type) channels are encoded by SLAC1-like genes (Negi et al. 2008; Vahisalu et al. 2008). Functional expression studies in Xenopus oocytes have shown that SLAC1 encodes bonafide S-type anion channels (Geiger et al. 2009a), whereas the genes encoding the two other anion channels still have to be identified.
In analogy to animal cells, the feedback mechanism of plant cells, described earlier, may facilitate action potentials in plant tissues. In the Venus flytrap, for instance, Ca2+-dependent action potentials can be initiated by electrical currents passed across the leaf (Trebacz et al. 1996), which causes a transient depolarization. For vascular plants, the Ca2+ changes occurring during action potentials have not yet been directly coupled to changes in ion channel activity. However, for cells of the algae Chara corallina, the onset of the action potential was found to coincide with an increase of the cytosolic Ca2+ concentration (Kikuyama & Tazawa 1998), and it involves the activation of two types of anion channels (Homann & Thiel 1994).
ENCODING Ca2+ SIGNALS
Provided that Ca2+-dependent signalling pathways initially have developed to maintain steady cytosolic Ca2+concentrations, it is likely that cells learned to recognize a disturbance of the Ca2+ homeostasis as a potential harmful event. Strong mechanical, osmotic or cold stimuli may cause a temporal leakage of the plasma membrane, and trigger a rise of the cytoplasmic Ca2+ concentration. During evolution, cells are likely to have evolved specialized proteins that recognize harmful conditions and in turn provoke a Ca2+ signal. Such a signalling mechanism enables the cells to respond long before the unfavourable conditions lead to cell damage. Indeed, several environmental stress conditions were found to be potent stimuli for triggering Ca2+ signals. Wind-, cold- or touch-induced transient cytosolic Ca2+ elevations could be recorded in seedlings transformed with the Ca2+ reporter protein aequorin (Knight et al. 1991; Knight, Smith & Trewavas 1992). Mechanical stimuli can also provoke cytosolic Ca2+ waves in single guard cells (Fig. 3), whereas hypo-osmotic conditions trigger rises of the cytosolic Ca2+ level of tobacco suspension cells (Takahashi et al. 1997).
The capability of the stress hormone ABA to trigger cytosolic Ca2+ signals (McAinsh, Brownlee & Hetherington 1990; Gilroy et al. 1991) may have evolved from the general ability of stress conditions to induce Ca2+ signals. Hostile environmental conditions like drought or cold sensed only by a subset of cells, like epidermal pavement cells, can be signalled by ABA to neighbouring cells, such as guard cells. Within these adjacent cells, ABA may trigger similar Ca2+ responses as in the cells that directly sense the hostile environment.
Given the plethora of stimuli that induce Ca2+ signals, the question arises if all these signals lead to similar responses in plant cells. This question has been intensively discussed with respect to guard cells, because a variety of stimuli cause Ca2+ signals in this cell type. Transient increases of the cytosolic Ca2+ concentration can be induced by ABA (McAinsh et al. 1990; Gilroy et al. 1991; Allen et al. 1999; Marten et al. 2007), elevated CO2 concentrations (Webb et al. 1996; Marten et al. 2008), as well as blue light (Harada & Shimazaki 2009). However, the guard cell responses to these stimuli must differ, because CO2 and ABA cause stomatal closure, whereas blue light induces opening (Roelfsema & Hedrich 2005). Recent data obtained with aequorin expressing Arabidopsis plants have shown that the blue light-induced changes in the cytosolic Ca2+ concentration indeed differ from those triggered by CO2 or ABA (Harada & Shimazaki 2009). Light triggers a fast transient Ca2+ increase (peak at 7 s) followed by a slow transient (peak at 142 s), whereas CO2 and ABA trigger only slow Ca2+ elevations. The two blue light-induced Ca2+ rises originate from different processes, because the fast Ca2+ peak depends on photosynthesis, whereas the slow Ca2+ transient is triggered by phototropins. Furthermore, the slow Ca2+ signal is only observed at very low extracellular K+ concentrations that probably lead to a large blue light-induced hyperpolarization of the plasma membrane (Harada & Shimazaki 2009). At physiologically relevant extracellular K+ concentrations, of approximately 5 mm (Felle et al. 2000), guard cells are unlikely to display the blue light-specific slow Ca2+ signal. In contrast, ABA and CO2 are able to trigger slow Ca2+ changes at physiological K+ concentrations, as well as under voltage clamp conditions (Marten et al. 2007, 2008). Under physiological conditions, Ca2+-dependent signalling in guard cells thus seems to be predominantly associated with stomatal closure, which fits an evolutionary origin from stress responses.
Even though Ca2+ signalling primarily seems to be associated with stomatal closure, the question remains to what extent Ca2+ signals encode signal-specific information. CO2 and low air humidity both lead to stomatal closure, but the signalling pathways activated by both signals are likely to differ. The response of Arabidopsis stomata in intact plants to high atmospheric CO2 levels was found to strictly depend on the SLAC1-encoded anion channel (Vahisalu et al. 2008). SLAC1 loss of function mutants also did not display rapid stomatal closure after a transfer from high to low air humidity. However, with the latter stimulus, slow closure of stomata was still observed (Vahisalu et al. 2008). Provided that responses to low air humidity involve ABA signals (Xie et al. 2006), the signalling events to ABA and CO2 must differ. CO2 exclusively induces stomatal closure through activation of plasma membrane anion channels (Brearley, Venis & Blatt 1997; Roelfsema et al. 2002), whereas ABA also inhibits H+-ATPases (Shimazaki, Iino & Zeiger 1986; Merlot et al. 2007). Two lines of reasoning have been put forward that explain how Ca2+-dependent responses can encode signal-specific information. (1) The nature of the Ca2+ signal may encode specific information; or (2) a second pathway acting parallel to the Ca2+ signal may influence the final response (Fig. 4). In certain responses, both mechanisms may even act in concert.
Stimulus-induced changes in the cytoplasmic Ca2+ concentration may not be limited to a single rise in the cytoplasmic Ca2+ concentration, but instead may lead to repetitive Ca2+ elevations. These repetitive Ca2+ spikes were first found to be induced by high extracellular Ca2+ concentrations in guard cells of Commelina communis (McAinsh et al. 1995), but later also were recognized in response to ABA (Staxen et al. 1999). Similar results were obtained in experiments with guard cells of Arabidopsis (Allen et al. 1999, 2000). Because of variations in shape and frequency of the repetitive Ca2+ elevations at different signal strength (McAinsh et al. 1995), it was postulated that the sequential rises in the cytosolic Ca2+ level encode a signal specific ‘Ca2+ signature’ (McAinsh & Hetherington 1998; McAinsh & Pittman 2009). Recognition of such a Ca2+ signature by Ca2+-dependent proteins could lead to a specific downstream response (Fig. 4).
The importance of repetitive Ca2+ rises was backed up by experiments with the de-etiolated 3 (det3) mutant, lacking a subunit of the vacuolar H+-ATPase (Allen et al. 2000). In this mutant, extracellular Ca2+ triggered only a single increase of the cytosolic Ca2+ concentration and did not lead to stomatal closure. The data suggest that sequential rises in the cytosolic Ca2+ level encode information that is essential for stomatal closure. This hypothesis was tested in a later study, in which Ca2+ signals were provoked with bath solutions that were likely to lead to either hyper- or depolarization of the plasma membrane (Allen et al. 2001). Indeed, the period and duration of exposure to both bath solutions could be linked to the degree of induction or sustainability of stomatal closure.
Apart from stimulus-induced Ca2+ signals, Arabidopsis guard cells also can show spontaneous repetitive changes in the cytosolic Ca2+ level (Klüsener et al. 2002). ABA suppressed these spontaneous Ca2+ changes in half of the cells tested. Likewise, the frequency of spontaneous Ca2+ elevations was lowered after exposing guard cells to increasing concentrations of CO2 (Young et al. 2006).
Spontaneous Ca2+ elevations may be linked to repetitive changes in the membrane potential, as found for V. faba guard cells in epidermal strips (Thiel, Macrobbie & Blatt 1992; Blatt & Armstrong 1993) or protoplasts thereof (Fig. 5) (Konrad & Hedrich 2008). During the hyperpolarized phase of such repetitive membrane potential changes, Ca2+ may enter the guard cell through hyperpolarization-activated cation channels (Grabov & Blatt 1998), whereas the cytoplasmic Ca2+ concentration subsequently will decrease again during the depolarized phase (Fig. 5) (Levchenko et al. 2008). However, the connection between spontaneous changes in the membrane potential and repetitive elevations of cytosolic Ca2+ has not been documented by direct measurements yet.
Parallel acting Ca2+-dependent and -independent pathways
Alternatively to the information encoded by the Ca2+ signature, stimulus-specific information also may be provided through Ca2+-independent signalling pathways that act parallel to the Ca2+-dependent events (Fig. 4). The existence of such Ca2+-independent signalling pathways was evident from experiments with V. faba guard cells in intact plants (Levchenko et al. 2005). Even though ABA was able to activate plasma membrane anion channels in these cells, no changes in the cytoplasmic free Ca2+ concentration could be detected. The inability to detect Ca2+ signals was unlikely caused by technical limitations, because hyperpolarizing voltage steps did provoke elevation of the cytosolic free Ca2+ level (Levchenko et al. 2005). Lowering the cytosolic free Ca2+ concentration in guard cells with the Ca2+ buffer BAPTA, does inhibit ABA-induced activation of anion channels (Levchenko et al. 2005) and stomatal closure (Webb et al. 2001; Siegel et al. 2009). Apparently, a certain threshold cytosolic Ca2+ level is required to enable ABA responses in guard cells.
Alike the data obtained by Levchenko et al. (2005), Romano et al. (2000) found that ABA inhibits inward rectifying K+ channels, in V. faba guard cells, in the absence of Ca2+ signals. The occurrence of ABA-induced Ca2+ signals, however, appears to be species dependent because ABA does trigger transient changes in the cytosolic Ca2+ level of guard cells in Nicotiana tabacum plants (Marten et al. 2007). Likewise, elevations of the cytosolic Ca2+ concentration were found in response to darkness, in N. tabacum guard cells (Marten et al. 2008). For both stimuli, approximately half of the cells activate plasma membrane anion channels in the presence of a transient rise of the Ca2+ concentration, whereas this response occurred in the absence of Ca2+ signals in the remaining cells (Fig. 6).
The activation of anion channels in guard cells in the presence or absence of Ca2+ signals puts the role of the cytoplasmic Ca2+ concentration for stomatal closure into question. What is the role of cytosolic Ca2+ elevations, if anion channels activate (Levchenko et al. 2005) or stomata close (Gilroy et al. 1991) even in the absence of such Ca2+ signals? It could be the evolutionary origin of these responses that brings clarity into this controversial topic. Whereas stomatal closure may have been strictly linked to cytosolic Ca2+ signals in early land plants, parallel acting pathways could have developed to obtain signal specificity in the course of evolution. For certain responses, the Ca2+ signal retained a traffic light function, because a full response only occurs if it is accompanied by a simultaneous rise in the cytosolic Ca2+ concentration. Ca2+ signals seem to execute this function in the response of N. tabacum guard cells to darkness, because the highest degree of anion channel activation is found in the presence of cytosolic Ca2+ elevations (Marten et al. 2008). For other responses, depending on the species, the original function of Ca2+ signals may have been overtaken by Ca2+-independent signalling pathways. Such a transformation from Ca2+-dependent to Ca2+-independent pathways seems feasible for ABA, because the Ca2+-dependent CDPK encoded protein kinases are closely related to the Ca2+-independent SnRK2 protein kinases [including open stomata 1 (OST1)] that are essential for ABA signalling (Mustilli et al. 2002; Hrabak et al. 2003; Umezawa et al. 2004; Yoshida et al. 2006).
Spatial distribution of Ca2+ signals
In this review, Ca2+-dependent signalling responses have been discussed either for plant cells in general, or with a focus on guard cells. However, given the large number of cell types with very different functions, it is likely that Ca2+-dependent signalling events will vary considerably between different cell types. Using a cell-specific expression system for aequorin, Kiegle et al. (2000) were able to show that drought or NaCl indeed triggered variable Ca2+ responses, depending on the cell type that was studied. However, even if time-dependent Ca2+ changes would be similar, it is likely that they would cause different responses in different cell types, because of the cell type-specific variations in expression of genes encoding Ca2+-dependent proteins. To some extent, variations in responses may also be caused by differences in the subcellular localization of Ca2+ signals. With common methods used for monitoring Ca2+ signals, subcellular differences in Ca2+ changes are not resolved. However, membrane-targeted aequorin or confocal imaging techniques revealed that nuclear Ca2+ changes can differ considerably from signals derived from the whole cytosol (Pauly et al. 2001; Bothwell et al. 2008). In addition to these three-dimensional differences in Ca2+ concentration changes, another level of complication is provided by the fourth dimension, because circadian rhythms are interconnected with changes in the cytosolic Ca2+ concentration (Love, Dodd & Webb 2004). These circadian changes in the cytosolic Ca2+ level indicate that plant cells alter the activity of certain Ca2+ transport proteins in the course of the day, and thus are likely to display variable responses to a certain stimulus depending on the time of day (Dodd et al. 2006).
Even though the importance of the cytosolic Ca2+ concentration in signalling systems of plants is undisputed, its exact role in guard cell signalling still remains unclear. The observation that responses involving changes in the cytoplasmic Ca2+ concentration often also are induced by Ca2+-independent mechanisms has led to confusion. The occurrence of both signalling pathways may be understood as products of evolution, the Ca2+-dependent mechanisms possibly being more ancient than the Ca2+-independent components. So far, however, guard cell signalling has only been studied with few unrelated vascular plants, and it therefore is hard to draw conclusions with respect to evolutionary origins from the available data. At this time, it is therefore difficult to asses if an exchange between Ca2+-dependent and -independent signalling mechanisms involving either SnRK2 or CDPK- and CIPK-encoded protein kinases, has occurred in evolution. The progress with respect to obtaining whole genome sequences of evolutionary distant species may shed light on this issue.
The genome sequence data that have become available so far indicate that vascular plants do not possess genes with homology to inositol[1,4,5]triphosphate (IP3) or ryanodine receptors from animals (Wheeler & Brownlee 2008). The presence of these channel types in chlorophyte algae indicates that these genes have been lost in land plants in the course of evolution (Wheeler & Brownlee 2008). In contrast, genes with homology to glutamate receptor and cyclic nucleotide gated channels have been conserved in land plants. However, the properties of these plant channels seem to differ from their animal counterparts, because the expression of full-length glutamate receptor-like channel genes from Arabidopsis in Xenopus oocytes did not yield functional channels (Tapken & Hollmann 2008). Heterologous expression of AtCNGC1 and from 2 Arabidopsis indicated a high conductance for K+ (Leng et al. 2002), but later studies showed that the CNGC2 channel is permeable for Ca2+ (Ali et al. 2007). Future studies may be directed to search for proteins that modulate these channels, either by co-expression in the heterologous expression systems or by expression studies with plant cells. New candidates for plant Ca2+ channels are likely to be uncovered through genome studies, as well as with screens for signalling mutants. The search for mutants defective in nodule formation, for instance, has led to the identification of cation channels that are involved in the generation of nuclear Ca2+ signals (Ane et al. 2004; Imaizumi-Anraku et al. 2005; Parniske 2008).
Apart from the identification of new Ca2+ signalling components, technical advances that enable visualization of signalling events at high resolution, like two photon-, total internal refraction fluorescence (TIRF) and stimulated emission depletion (STED) microscopy (Donnert et al. 2006), are likely to provide new insights in the role of localized cytosolic Ca2+ signals. These techniques enable the recording of Ca2+ concentration changes in different structures within single plants cells, especially in combination with targeted expression of the improved yellow cameleon-encoded fluorescent Ca2+ indicator proteins (Nagai et al. 2004; Monshausen, Messerli & Gilroy 2008; Weinl et al. 2008). Future studies, however, should not only be directed to study Ca2+ concentration changes in further detail, but new techniques will also enable the manipulation of Ca2+ signals in intact tissues. In neurobiology, the targeted expression of channelrhodopsin has enabled the light-induced excitation of selected nerve cells in intact organisms (Nagel et al. 2005; Fiala et al. 2006). These channels are likely to provide a tool for triggering Ca2+ signals in plant cells as well. Using these light-gated channels, it may become possible to study the impact of Ca2+ signals, at a defined intensity or frequency, on responses such as stomatal movement, in intact plants. Finally, the recent results with guard cells in intact plants revealed that Ca2+-dependent signalling is often backed up by Ca2+-independent signalling pathways. The full impact of Ca2+ signalling therefore can probably only be understood, if the role of such parallel acting Ca2+-independent signalling mechanisms are also taken into account.
The authors would like to thank A. Carpaneto, Istituto di Biofisica, Genova, Italy; V. Levchenko, Belarusian State University, Minsk, Belarus; Petra Dietrich, University of Erlangen-Nürnberg, Germany; K. Konrad, D. Beyhl and I. Marten, University Würzburg, Germany for providing figures, and G. Nagel and A. Stange, University Würzburg, Germany for help with the preparation of the manuscript. This work was supported by the GK1342 and FG964 grants of the Deutsche Forschungsgemeinschaft to R.H., and the SFB567/A8 grant to M.R.G.R.