CDPKs CPK6 and CPK3 Function in ABA Regulation of Guard Cell S-Type Anion- and Ca2+- Permeable Channels and Stomatal Closure

Abscisic acid (ABA) signal transduction has been proposed to utilize cytosolic Ca2+ in guard cell ion channel regulation. However, genetic mutants in Ca2+ sensors that impair guard cell or plant ion channel signaling responses have not been identified, and whether Ca2+-independent ABA signaling mechanisms suffice for a full response remains unclear. Calcium-dependent protein kinases (CDPKs) have been proposed to contribute to central signal transduction responses in plants. However, no Arabidopsis CDPK gene disruption mutant phenotype has been reported to date, likely due to overlapping redundancies in CDPKs. Two Arabidopsis guard cell–expressed CDPK genes, CPK3 and CPK6, showed gene disruption phenotypes. ABA and Ca2+ activation of slow-type anion channels and, interestingly, ABA activation of plasma membrane Ca2+-permeable channels were impaired in independent alleles of single and double cpk3cpk6 mutant guard cells. Furthermore, ABA- and Ca2+-induced stomatal closing were partially impaired in these cpk3cpk6 mutant alleles. However, rapid-type anion channel current activity was not affected, consistent with the partial stomatal closing response in double mutants via a proposed branched signaling network. Imposed Ca2+ oscillation experiments revealed that Ca2+-reactive stomatal closure was reduced in CDPK double mutant plants. However, long-lasting Ca2+-programmed stomatal closure was not impaired, providing genetic evidence for a functional separation of these two modes of Ca2+-induced stomatal closing. Our findings show important functions of the CPK6 and CPK3 CDPKs in guard cell ion channel regulation and provide genetic evidence for calcium sensors that transduce stomatal ABA signaling.


Introduction
Stomatal pores in the epidermis of aerial parts of plants facilitate gas exchange between plants and the atmosphere. Stomatal pores are surrounded by pairs of guard cells that mediate stomatal pore opening and closing. Guard cells respond to diverse stimuli, including blue light, CO 2 concentrations, drought, pathogen attack, and plant hormones, including abscisic acid (ABA) [1][2][3]. Stomatal movements are mediated by ion transport across the plasma membrane and vacuolar membrane of guard cells and by organic solute content changes [1,4]. Guard cell ion channels and proton pumps are regulated by the cytosolic free Ca 2þ concentration ([Ca 2þ ] cyt ) such that [Ca 2þ ] cyt elevation activates stomatal closing mechanisms [5][6][7][8][9]. These findings correlate with the Ca 2þ dependence of ABA-induced stomatal closing [10,11].
ABA is a drought-inducible plant hormone. ABA regulates [Ca 2þ ] cyt elevations in guard cells [12][13][14][15][16][17][18] and other cells [19,20]. [Ca 2þ ] cyt elevation activates slow-type (S-type) anion efflux channels and down-regulates inward-rectifying K þ channels and proton pumps in the plasma membrane of guard cells [5,7,16,21]. The concomitant efflux of anions and K þ from guard cells results in turgor reduction and stomatal closure. Thus, Ca 2þ elevation shifts ionic conductance properties in guard cells toward a stomatal closing favoring mode. Other studies indicate a possible role for Ca 2þ in mediating stomatal opening as well [22][23][24]. It is noteworthy that ABA signal transduction in guard cells consists of not only a Ca 2þ -dependent response but also a Ca 2þ -independent possibly parallel pathway [25] (for reviews: [3,4], see Discussion). Whether Ca 2þ -independent mechanisms suffice for mediating full physiological stomatal ABA responses remains unclear. Genes encoding signal transduction proteins in both Ca 2þ -dependent and -independent pathways are still largely unknown and molecular genetic evidence for a rate-limiting role of Ca 2þ as a positive transducer of ABA signaling is lacking. Repression of the SOS3-like calcium binding protein 5 (SCaBP5)/Calcineurin B-like protein 1 (CBL1) and CBL9 genes result in ABA hypersensitivity, suggesting that CBL1 and CBL9 function as negative regulators of an ABA signal transduction pathway [26,27]. However, no genes encoding Ca 2þ -sensing proteins that function as positive transducers of ABA signaling and of ion channel regulation in guard cells have been identified.
The CDPK gene family includes 34 members in Arabidopsis alone [35,46]. Redundancies in CDPK genes are likely to hamper molecular genetic analyses of CDPK functions. Recent studies have shown that relatively few signal transduction genes are expressed in a strictly cell-specific manner in roots, guard cells, and mesophyll cells [51][52][53][54]. However, signal transduction genes that are highly expressed in a given cell type, such as guard cells, are candidates for being incorporated into signal transduction networks within those cells. Furthermore, cell-specific signal transduction assays may allow resolution of phenotypes of single knockout mutants that global whole plant phenotypic analyses would not resolve in reverse genetic studies. Therefore, in the present study we used single cell-type gene expression analysis [54] to determine which CDPK genes are expressed in guard cells. We report phenotypes of loss-of-function mutants in two of the guard cell-expressed CDPKs and characterize functions of these CDPKs in Ca 2þ activation of anion channels, in ABA activation of anion channels, and unexpectedly also in ABA regulation of Ca 2þ channels as well as in Ca 2þ -reactive and ABA-induced stomatal closing. The presented results provide molecular genetic and cell biological evidence for Ca 2þ sensors that function as positive transducers in plant ion channel regulation and ABA-and Ca 2þ -dependent signal transduction in guard cells.

Expression of CDPK Genes
There are 34 CDPK genes in the Arabidopsis genome [35,46]. To investigate whether and, if so, where CDPKs may function in ion channel regulation and guard cell signal transduction branches, we first identified CDPK genes expressed in guard cells using a guard cell-enriched cDNA library and RT-PCR with degenerate oligomers [55]. Two of the guard cellexpressed CDPK genes, CPK3 (AGI No.: At4g23650) and CPK6 (AGI No.: At2g17290), showed initial insertion mutant phenotypes and were therefore further analyzed. The expression of CPK3 and CPK6 in isolated guard cell protoplasts (GCPs) was further analyzed by RT-PCR with gene-specific primers ( Figure 1A) and independently later by cell type-specific genomic scale expression analyses using Affymetrix (Santa Clara, California, United States) GeneChip assays [54]. RT-PCR analysis showed that CPK3 and CPK6 are expressed in both guard cells and mesophyll cells ( Figure 1A). The purity of GCPs was analyzed by RT-PCR with specific primers for the guard cell-expressed potassium channel gene, KAT1 ( Figure 1A) [56]. Guard cell preparations were further examined for contamination of mesophyll cells by analyzing mRNA abundance of a putative calmodulin-binding protein (CBP) (AGI No.: At4g33050), which was identified as being highly expressed in mesophyll cells but absent in guard cells [54]. No CBP mRNA was detected in guard cell preparations ( Figure 1A), indicating that the GCP preparations had no or very little contamination. In addition to CPK3 and CPK6, several other CDPK genes were identified in guard cells by microarray experiments with guard cell RNA (Supplemental Table I in [54]). In this study, we focus on functional dissection of the guard cell-expressed CPK3 and CPK6 genes.
To genetically analyze functions of CPK3 and CPK6 in guard cell signal transduction, we identified T-DNA insertion mutations in CPK3 and CPK6 from the Salk Institute Genomic Analysis Laboratory database [57]. Homozygous T-DNA insertion mutant lines were isolated and genomic sequences of the cpk3-1 (SALK_107620), cpk3-2 (SALK_022862), cpk6-1 (SALK_093308), and cpk6-2 (SALK_033392) insertion mutants were determined. The T-DNA insertions in cpk3-1 and cpk3-2 are localized in the first exon and in the first intron, respectively ( Figure 1B). The insertion in cpk6-1 is localized in the second exon, 60 basepairs downstream of the translation initiation codon, and cpk6-2 is in the first intron ( Figure 1B). Southern blot analyses of homozygous plants indicated only a single band in each line, suggesting a single T-DNA insertion in these mutants (data not shown). Transcripts of CPK3 or CPK6 were not detected in cpk3-1 and cpk6-1 as demonstrated by RT-PCR utilizing whole leaf RNA extracts ( Figure 1C). No RT-PCR band was observed for cpk3-2 (data not shown). For cpk6-2, no full-length cDNA was detected ( Figure 1C). A faint band for transcript downstream of the T-DNA insertion was observed after 35 cycles of amplification showing substantially reduced mRNA levels (8% or less intensity compared to wild-type level, n ¼ 2) ( Figure 1C). Using a primer set in the first exon and the eighth exon, no RT-PCR amplification was observed ( Figure 1B and 1C), showing that the first exon is missing in cpk6-2. We performed RT-PCR with CPK3 primers in cpk6-1 and with CPK6 primers in cpk3-1 to examine whether a compensatory expression occurs. No compensation in the wild-type transcript levels was observed (data not shown).
Activation of S-Type Anion Channels by Cytosolic Ca 2þ Is Impaired in cpk3cpk6 Mutants S-type anion efflux channels have been proposed to play an important role as targets of ABA signal transduction in guard cells and to be regulated by upstream phosphorylation events [5,9,16,32,55,[58][59][60][61][62]. To determine whether CDPKs function in the activation of S-type anion channels in guard cells, we examined Ca 2þ activation of S-type anion channels in wildtype, cpk3, and cpk6 single mutant and double mutant guard cells.

Impairment in ABA Activation of S-Type Anion Channels in cpk3cpk6 Mutants
Previous studies have shown that guard cell signal transduction is mediated by a network of events, which includes parallel Ca 2þ -dependent and -independent signaling branches (see Discussion; for reviews: [3,4]). Therefore, experiments were pursued to analyze ABA activation of Stype anion channels. These experiments were performed under different conditions than those shown in Figure 2, such that cytosolic Ca 2þ elevation alone would not fully activate Stype anion currents (see Materials and Methods) [9]. As shown in Figure 3, with preincubation of guard cells in low extracellular Ca 2þ , 2 lM cytosolic Ca 2þ in the pipette activated S-type anion currents in guard cells of only intermediate amplitudes ( Figure 3A). Under these conditions, ABA up-regulated S-type anion current activities in wild-type protoplasts ( Figure 3A, 3B, and 3E; n ¼ 8). In cpk3-1cpk6-1 (n ¼ 7) and cpk3-2cpk6-2 (n ¼ 7) double mutant guard cells, ABA regulation of S-type anion currents was impaired ( Figure 3C, 3D, and 3E; p , 0.01). These data show that, despite the complex network of ion channel regulation mechanisms in guard cells [3], CDPKs mediate an important Ca 2þ -decoding transduction step in ABA regulation of S-type anion channels (Figures 2 and 3). ABA activation in wild-type guard cells and the impairment in ABA activation of S-type anion channels at 2 lM cytosolic Ca 2þ in cpk3cpk6 mutants ( Figure 3) further provide evidence for a recently proposed hypothesis in which stomatal closing signals (i.e., ABA) mediate priming of guard cell Ca 2þ sensors, such that they can respond to elevated cytosolic Ca 2þ levels [63].

Impairment in ABA activation of I Ca Channels in cpk3 cpk6 Mutants
ABA activates plasma membrane Ca 2þ -permeable (I Ca ) channels [64][65][66]. Combined physiological, molecular genetic, and cell biological analyses have shown that I Ca channels function in the guard cell ABA signal transduction network at hyperpolarized voltages [9,53,[64][65][66]. We examined whether CPK3 and CPK6 function in the regulation of I Ca channels.
Typical I Ca currents were activated by extracellular application of ABA to patch-clamped wild-type guard cells ( Figure 4A and 4B, n ¼ 13). Unexpectedly, ABA activation of I Ca channels was not observed in cpk3-1cpk6-1 and cpk3-2cpk6-2 double mutant guard cells (Figure 4C-4E; n ¼ 14, p ¼ 0.56 for cpk3-1cpk6-1; n ¼ 8, p ¼ 0.47 for cpk3-2cpk6-2 when comparing before and after ABA treatment). Blind patch-clamp experiments in which the genotype of protoplasts was unknown (n ¼ 2 for wild-type, n ¼ 2 for cpk3-1cpk6-1), and similar findings by Y.M., I.C.M., Y.W., and S.M. in this study, further confirmed the impairment of ABA activation of I Ca channels in cpk3cpk6 mutant guard cells. Next we analyzed whether only one of the two CDPKs might affect ABA activation of I Ca channels. Defects in the ABA activation of I Ca channels were observed in the cpk3-1, cpk3-2, cpk6-1, and cpk6-2 single mutants when comparing before and after ABA-treatment). Together these data show that CPK3 and CPK6 function in ABA regulation of I Ca channels ( Figure 4) and Ca 2þ and ABA activation of S-type anion channels (Figures 2 and 3).
To gain insight into the question whether activation of phosphorylation events are required before or after ABA application and during ABA activation of I Ca channels in patch-clamped Arabidopsis guard cells, wild-type guard cells were pretreated with the broad serine/threonine kinase inhibitor, K252a [(8R*,9S*,11S*)-(À)-9-hydroxy-9-methoxycarbonyl-8-methyl-2,3,9,10-tetrahydro-8,11,epoxy-1H,8H,11H-2,7b,11a-triazadibenzo[a,g]cyclo-octa[c,d,e]-trinden-1-one], 20 min prior to, and continuously during, patch clamping. Moreover, the cytoplasm of whole cells was dialyzed in the absence of ATP in whole cell recordings for longer than 15 min prior to extracellular ABA exposure. Ionic currents were recorded in the same guard cells prior to and after exposure to ABA. In negative controls, treatment with K252a alone did not cause constitutive activation of I Ca channel currents without ABA treatment (n ¼ 3). Interestingly, pretreatment with K252a (2 lM) did not disrupt ABA activation of I Ca channels (Figure 4J and 4K; n ¼ 6, p , 0.04 at À180 mV). These findings indicate that the effect of the cpk3 and cpk6 mutations on ABA activation of I Ca channel currents are most likely not caused by direct ABA-induced upstream CDPK activation in the NADPH-dependent activation branch of I Ca channels [66], but possibly by prior CDPK action (see Discussion).
To further evaluate Ca 2þ regulation of stomatal closure, we examined the effect of experimentally imposed [Ca 2þ ] cyt oscillations on stomatal closure in cpk3cpk6 double mutant plants [29][30][31]68]. A Ca 2þ oscillation pattern was applied to guard cells with a similar time pattern to those that cause long-term programmed stomatal closure (i.e., inhibition of stomatal reopening) in Arabidopsis [29]. Hypothesizing a contribution of additional CDPKs or other Ca 2þ transducers in these experiments, we applied a hyperpolarizing buffer with lower extracellular Ca 2þ concentrations (1 mM) and higher KCl concentrations (1 mM) than those used in previous research [29] with the goal of experimentally imposing weaker cytosolic Ca 2þ transients in guard cells (see Materials and Methods). Ca 2þ imaging experiments of guard cells showed that this protocol causes cytosolic Ca 2þ oscillations in cpk3-1cpk6-1 double mutant and in wild-type guard cells (Figure 7, inset). The average amplitudes of imposed [Ca 2þ ] cyt transients were similar in wild-type and cpk3-1cpk6-1 double mutant guard cells (p ¼ 0.25; wild-type: 0.725 ratio units [RU] 6 0.043 SEM, n ¼ 16 wild-type cells; cpk3-1cpk6-1: 0.766 RU 6 0.04 SEM, n ¼ 24 cpk3-1cpk6-1 guard cells). The average integrated total [Ca 2þ ] cyt ratio increase per period was also determined and found to be similar for wild-type and cpk3-1cpk6-2 guard cells (p ¼ 0.46; wild-type: 8.535 RU * 0.1 min 6 0.497, n ¼ 16 wild-type  cells; cpk3-1cpk6-1: 8.456 RU * 0.1 min 6 0.517, n ¼ 24 cpk3-1cpk6-1 cells). These data are consistent with findings that external Ca 2þ -induced [Ca 2þ ] cyt elevations in guard cells include intracellular Ca 2þ release from guard cell organelles [67,69]. Four Ca 2þ transients with a 10-min period, which induce long-lasting ''Ca-programmed'' stomatal closure [29], were applied to wild-type and cpk3cpk6 double mutant stomata. Wild-type stomata started closing immediately after the first Ca 2þ elevation was imposed and continued to show progressive Ca 2þ -reactive closing for longer than 40 min (Figure 7; n ¼ 6 experiments, 47% closure). In contrast, Ca 2þ transient-induced closure of cpk3-1cpk6-1 and cpk3-2cpk6-2 double mutant stomata was reduced (14% and 22 % closure, respectively) (Figure 7; n ¼ 6 experiments, p , 0.01 for cpk3-1cpk6-1 versus wild-type and n ¼ 3 experiments, p , 0.05 for cpk3-2cpk6-2 versus wild-type, p . 0.60 cpk3-1cpk6-1 versus cpk3-2cpk6-2). Stomata of both wild-type and cpk3cpk6 double mutant guard cells remained closed during the ensuing 2 h and 20 min measurements, even though cells were extracellularly bathed in a typical ''stomatal opening'' solution containing 50 mM KCl and 0  mM CaCl 2 and exposed to white light (Figure 7, from 40 to 180 min). Interestingly, the partial stomatal closure of cpk3cpk6 double mutants was also maintained during the ensuing 2 h and 20 min after Ca 2þ transients were terminated, i.e. a significant stomatal closure was observed at 180 min when compared with 0 min in both cpk3-1cpk6-1 and cpk3-2cpk6-2 (p , 0.01 for both mutants). Thus, the rapid Ca 2þ -reactive stomatal closure response is clearly impaired in the cpk3cpk6 double mutants, whereas the longterm Ca 2þ -programmed stomatal closure response [29][30][31] appears to be functional (Figure 7).
Seed germination analyses were also pursued with wildtype and the cpk3-1cpk6-1 and cpk3-2cpk6-2 double mutants. No significant difference in ABA inhibition of seed germination in wild-type and the cpk3cpk6 double mutant alleles was observed in the presence of 0, 0.3, 1, and 5 lM ABA after 3, 5, 7, and 11 d (unpublished data).

Rapid-Type Anion Channel Activity Is Not Greatly Altered in cpk3cpk6 Double Mutant
ABA also regulates a second class or mode of anion channels in guard cells, the rapid-type (R-type) anion channels [62,[70][71][72][73]. Therefore, we also compared R-type anion channel current properties in wild-type and cpk3cpk6 guard cells (Figure 8). The selectivity for anions over cations of these ion currents ( Figure 8A) in Arabidopsis guard cells was further analyzed. Replacing Ba 2þ with the impermeable cation tetraethylamonium in the bath solution did not affect these R-type inward currents, as previously shown [65]. Moreover, use of the impermeable anion gluconate in the pipette solution abolished the current (n ¼ 3, data not shown), confirming that the recorded currents are R-type anion currents. Interestingly, in contrast to S-type anion channels (Figures 2 and 3), no significant difference was observed in the rapid anion channel activity between wild-type and cpk3-1cpk6-1 double mutant guard cells ( Figure 8A and 8B; wildtype, n ¼ 7; cpk3-1cpk6-1, n ¼ 7; p ¼ 0.21 at peak current). Thus, cpk3cpk6 mutant guard cells did not significantly impair Rtype anion currents, which correlated with the partial ABAinduced stomatal closing in cpk3cpk6 mutant guard cells ( Figure 6, Table 1).

Discussion
CDPKs have been predicted to function in response to cytoplasmic Ca 2þ elevations in many physiological processes in plants [33,34]. Many biochemical studies have suggested roles of CDPKs in plant biology [36,40,41,43,45,[48][49][50]74]. A dominant mutant study [37] and a biochemical study of CPK32 [49] have provided evidence for functions of CDPKs in ABA signaling, and gene-silencing/antisense cdpk mutant phenotypes were demonstrated for a plant defense response in tobacco [36] and for nodule formation in Medicago truncatula root hairs [39]. The relative dearth of genetic or reverse genetic CDPK loss-of-function mutation phenotypes may be attributed to partial redundancies in the functions of the large CDPK gene family. Furthermore, cell-specific and mechanistic protein regulation (e.g., ion channel) analyses allow resolution of quantitative single cpk gene disruption mutant phenotypes, as demonstrated here.
Our findings do not exclude the possibility that the CDPKs analyzed here may function in other pathways in other plant tissues. Single cell-type microarray studies have shown that relatively few genes are expressed in a strictly cell-specific manner in Arabidopsis root cell types, guard cells, and mesophyll cells [52,54]. Combinatorial usage of a single protein in different signaling pathways was initially documented in yeast [75]. Furthermore, studies of mitogenactivated protein kinase and protein phosphatase genes in plants have suggested that individual mitogen-activated protein kinases and protein phosphatases function in different signal transduction cascades in different tissues and under different conditions [55,[76][77][78][79] (for review: [80]), and therefore the present study does not exclude that CPK3 and  CPK6 have additional functions in other tissues. The present study demonstrates important functions of the CPK3 and CPK6 CDPKs in control of stomatal movements and in ABA and Ca 2þ regulation of guard cell S-type anion channels and unexpectedly also in ABA regulation of Ca 2þ -permeable I Ca channels. Although many studies have shown Ca 2þ -dependent steps in guard cell signal transduction and in plant ion channel regulation [81], Ca 2þ sensor-encoding genes that function as positive transducers in these responses have remained unknown. A calcineurin-like Ca 2þ sensor, CBL1/SCaBP5, has been reported to function as a negatively regulating Ca 2þ sensor in ABA signaling in guard cells ( [26], but see [82]). Knock-out mutation of another CBL, CBL9, resulted in ABA hypersensitive phenotypes in seed germination, root and shoot growth, and ABA-inducible gene expression in Arabidopsis, indicating its function as a negative regulator of ABA signaling [27]. In the present study, guard cell expression analyses of CDPK genes aided in narrowing the number of candidate genes that may function in this system. Furthermore, the guard cell system has also allowed us to pursue detailed mechanistic and cellular analyses revealing functions of specific CDPKs in planta.

CDPKs Function in Ca 2þ -Reactive Stomatal Closure and S-Type Anion Channel Activation
[Ca 2þ ] cyt elevation in guard cells causes a rapid ''Ca 2þreactive'' stomatal closure response, irrespective of the above threshold Ca 2þ elevation pattern [29]. In contrast, a long-term ''Ca 2þ -programmed'' stomatal closure is modulated by imposed Ca 2þ transient parameters [29][30][31]. Ca 2þ -dependent stomatal movements were analyzed in response to imposed Ca 2þ transients. In cpk3cpk6 double mutant plants, Ca 2þreactive stomatal closure was significantly reduced compared to wild-type ( Figure 7). However, the partial Ca 2þ transientinduced stomatal closing response of cpk3cpk6 double mutant plants was maintained in white light more than 2 h after Ca 2þ transients were terminated. Thus, the imposed long-term Ca 2þ -programmed stomatal closure response [29] appears to be functional in the cpk3cpk6 double mutant, whereas the rapid Ca 2þ -reactive stomatal closure response is clearly impaired (Figure 7). The presented data provide genetic evidence for a mechanistic separation of Ca 2þ -reactive and Ca 2þ -programmed stomatal closure. Several other CDPK transcripts are expressed in guard cells as determined by oligonucleotide-based microarray experiments [54]. It is possible that other guard cell-expressed CDPKs function in Ca 2þ -programmed stomatal closure.
Previous studies have provided evidence that phosphorylation events function in Ca 2þ activation of S-type anion channels [16,32] (see Introduction). The cytosolic Ca 2þ activation of S-type anion channels requires the presence of hydrolyzable ATP in the patch-clamp electrode, and this channel activation is abolished by the general serine/ threonine protein kinase inhibitors K252a and staurosporine [16]. The finding that Ca 2þ activation of S-type anion channels is impaired in cpk3 and cpk6 gene disruption mutants provides molecular genetic evidence for this model and suggests that CDPKs function in [Ca 2þ ] cyt perception upstream of S-type anion channel activation (Figure 9). The cpk6-1 and cpk6-2 alleles exhibit similar defects in all of the phenotypes that were examined, suggesting that both alleles are strong alleles. Although both alleles show no expression of full-length cDNA, a faint band representing residual transcript of the CPK6 open reading frame was amplified in the cpk6-2 allele (8% or less of wild-type level). However, the cpk6-2 transcript is lacking the wild-type origin of transcription. Furthermore, this transcript also lacks the first exon of the wild-type cDNA, and 59 UTRs have been shown to play roles in enhancing translational efficiency [83]. Together, these data indicate that the greatly reduced level of cpk6-2 transcripts and the absence of the first exon are sufficient to cause the strong phenotypes observed in this study. Future direct investigation of the model that CDPKs phosphorylate important targets during S-type anion channel activation will require identification of the molecular components that encode these ion channels or upstream regulators. CPK3 and CPK6 may have distinct functions in the upstream ion channel regulation pathways. The present findings point to a model in which Ca 2þ activation of S-type anion channels functions in the rapid Ca 2þ -reactive stomatal closure response (Figures 2, 7, and 9) [29].

CDPKs, R-Type Anion Channels, and Ca 2þ -Independent Signaling
A Ca 2þ -independent ABA signal transduction branch has been implicated in guard cells, based on data in which ABAinduced cytosolic Ca 2þ increases were not observed in guard cells [13,16,25,28,[84][85][86] (for reviews: [2][3][4]87]). Ca 2þ -dependent or -independent ABA signaling pathways have been reported to be emphasized depending on physiological conditions in Commelina communis [25]. The abolishment of ABA-induced stomatal closing by injection of the Ca 2þ chelator BAPTA into guard cells [86,88] indicates that a Ca 2þ -independent branch within the ABA signaling network would interact with Ca 2þ -dependent mechanisms. The present study provides direct molecular genetic evidence that Ca 2þ sensors function within the Arabidopsis guard cell ABA signaling network (Figures 3, 4, and 6A) and provides a genetic basis to analyze interactions with a Ca 2þ -independent pathway.
Several studies have indicated a role for R-type anion channels in stomatal closure [62,[70][71][72][73]89], in addition to Stype anion channels. ABA activation of R-type anion channels was recently directly demonstrated in a subset of Vicia faba guard cells [62,73]. Direct cytosolic Ca 2þ activation of R-type channels has not yet been shown, although R-type anion channels in V. faba guard cells have been proposed to be Ca 2þ activated based on external CaCl 2 activation [71]. In cpk3cpk6 double mutant guard cells, R-type anion channel activity was not greatly impaired under the imposed conditions ( Figure  8). The partial closure of stomata in response to ABA ( Figure  6A, Table 1) may thus be explained by overlapping physiological functions of R-type and S-type anion channels ( Figure 9). In the present study and in a recent study, R-type anion currents were recorded in patch-clamped Arabidopsis ( Figure 8) and V. faba [86] guard cells at resting [Ca 2þ ] cyt and were proposed to be activated in a Ca 2þ -independent manner in V. faba. However, in both studies, guard cells were preincubated in a solution containing 40 mM CaCl 2 or more prior to patch clamping which would elevate Ca 2þ in guard cells [16,67,69], and therefore Ca 2þ -independent R-type anion channel activation (Figure 9) will require further investigation. The presented cpk3cpk6 mutant data provide first genetic evidence that parallel pathways function in regulation of R-type and S-type anion channels (Figures 2, 3, and 8), even though these anion channels may share molecular components [89] (Figure 9).
Residual small S-type anion channel currents were consistently observed in cpk3cpk6 double mutant guard cells ( Figures 2B, 2C, 3C and 3D), which should also contribute to partial ABA-and Ca 2þ -induced stomatal closing. The rate of anion efflux from Arabidopsis guard cells can be estimated, to predict whether the residual anion currents in cpk3cpk6 double mutants can close stomata within a physiological response. Cell volume of Arabidopsis GCPs was estimated at 0.11 pL according to the average diameter of protoplasts (6 lm). An estimated Cl À concentration decrease from 400 mM to 100 mM during ABA-induced stomatal closure would correspond to Cl À efflux of 33.9 fmol/cell. The membrane potential of guard cells treated with ABA was approximately À50 mV [73]. Cl À efflux currents in wild-type and cpk3cpk6 double mutant guard cells at À50 mV were approximately À60 pA and À25 pA, respectively ( Figure 3E). These S-type anion currents correspond to Cl À efflux rates of 37.3 fmol/min for wild-type and 15.54 fmol/min for the double mutant. This estimate shows that during the slow onset of drought that occurs over several days in intact plants, the residual S-type and R-type anion currents in cpk3 cpk6 double mutants should be sufficient for a physiological stomatal closing response. Similarly, previous estimates have predicted that ion channel activities are over 10-fold higher in biological membranes than required for a typical response, due to their high transport rate [90]. This estimate also highlights the power of analyzing individual ion channel targets for characterizing CDPK functions in plants.
It is possible that additional CDPKs contribute to activation of the residual S-type anion current activity. The partial Ca 2þ -induced stomatal closure in the cpk mutants ( Figures 6B  and 7) can be explained by Ca 2þ activation of residual S-type anion channel currents. In summary, the lack of R-type anion channel activity alteration and the residual S-type anion current in cpk3cpk6 mutant guard cells (Figures 2, 3, and 8) correlate with the partial ABA-and Ca 2þ -induced stomatal closing responses found here (Figures 6, 7, and 9) and may also reflect a contribution of parallel Ca 2þ -independent and/ or pH-dependent [91,92] signaling mechanisms.

CDPK Mutants Impair ABA Activation of Ca 2þ -Permeable Channels
Unexpectedly, ABA activation of I Ca channel currents was impaired in cpk3cpk6 double mutant guard cells (Figure 4). Impairment in ROS signaling partially impairs rather than abolishes ABA-induced stomatal movement responses, consistent with their activity mainly at hyperpolarized voltages [53]. The present findings are consistent with a model in which additional signaling branches function in the ABA signal transduction network parallel to the ABA ! ROS ! I Ca signaling branch (Figure 9) [65]. Several parallel pathways function in the ABA signaling network, including intracellular Ca 2þ -release mechanisms (Figure 9) [15,17,18,81].
Interestingly, in wild-type guard cells, ABA could activate I Ca channels even when cells were preexposed to the general serine/threonine kinase inhibitor K252a and the cytosol of guard cells was dialyzed with an ATP-free solution for longer than 15 min in whole-cell recordings prior to ABA exposure ( Figure 4J and 4K). K252a inhibits CDPK activities in diverse plants [45,[93][94][95] and abolishes Ca 2þ activation of S-type anion channels in guard cells [16,32]. These data indicate the following possible mechanisms by which the cpk3cpk6 mutant . Simplified Models for Early ABA Signal Transduction Network and CPK3 and CPK6 Functions in Guard Cells Plasma membrane I Ca channel and intracellular Ca 2þ -release mechanisms cause cytosolic Ca 2þ elevations. CPK3 and CPK6 mediate [Ca 2þ ] cyt activation of S-type anion channels. CPK3 and CPK6 regulate I Ca Ca 2þ channels by a proposed feedback (A) or parallel pathway (B) or by transcriptional/ translational regulation mechanism (see Discussion). R-type anion channels are regulated in a parallel CPK3-CPK6-independent signal transduction pathway. The arrow connecting R-type and S-type anion channels indicates that these channels may share molecular components [89] alleles may impair ABA activation of I Ca channels in guard cells ( Figure 4): 1. CPK3 and CPK6 may not be directly activated in response to exogenous ABA within the ABA ! NADPH oxidase ! ROS ! I Ca channel signaling branch (Figure 9). But these CDPKs are necessary for maintaining I Ca channel activation in a parallel regulation pathway of I Ca channels [96] that also regulates S-type anion channels ( Figure 9B). In this model, ABA regulation of S-type anion channels is not strictly downstream of I Ca channels ( Figure 9B).
2. CDPKs may regulate I Ca channels in a feedback loop by which cytosolic Ca 2þ elevations regulate I Ca channels ( Figure  9A).
3. The present findings can also be explained by an alternative model in which cpk3cpk6 mutation affects gene expression in such a manner that ABA activation of I Ca channels is impaired. Similarly, the effects of the dominant abi1-1 and abi2-1 protein phosphatase 2C (PP2C) mutants on ABA responses could in principle result from indirect effects of these mutations on these ABA signal components, because these are dominant mutations and no pharmacological PP2C inhibitors are available to test effects of short-term PP2C impairment on channel regulation, as previously discussed [66]. This model in which cpk3cpk6 may affect transcriptional or translational responses would be consistent with the activation of I Ca currents in patch-clamped wild-type Arabidopsis guard cells without added cytoplasmic ATP and in the simultaneous presence of K252a ( Figure 4J and 4K).
Experiments in V. faba guard cells show different responses than wild-type Arabidopsis guard cells that may shed light on the impairment in I Ca channel activation in cpk3cpk6 guard cells. In V. faba guard cells, K252a inhibits ABA activation of I Ca currents and the protein phosphatase inhibitors calyculin A and okadaic acid activate I Ca channel currents, suggesting that phosphorylation/dephosphorylation events regulate I Ca channels in a parallel pathway in V. faba [96] (Figure 9). The present findings in Arabidopsis guard cells would predict a requirement for CDPK-dependent phosphorylation, prior to ABA activation of the ABA ! NADPH oxidase ! ROS ! I Ca channel regulation branch (Figures 4  and 9B). Previous analyses have indicated that Arabidopsis and Vicia guard cells emphasize different components of the guard cell signaling network, including differential phosphorylation-dependent responses [32,55,59,86,97,98]. In this respect, further comparative analyses among species should be useful in illuminating new signaling network mechanisms and branches.
Further studies will be necessary to distinguish the above models. The present study, however, provides a first characterization of a genetic mutation that impairs both Ca 2þ activation of S-type anion channels and ABA activation of Ca 2þ -permeable I Ca channels, indicating closer interactions among mechanisms that regulate these two classes of ion channels in guard cells than previously modeled.

Conclusions
In summary, in this study, we provide direct molecular genetic evidence for Ca 2þ sensors that function as positive transducers in stomatal ABA signaling and demonstrate functions of CPK6 and CPK3 in plant ion channel regulation in defined guard cell signal transduction elements. We functionally characterize disruption mutations in the two guard cell-expressed CDPK genes, CPK3 and CPK6. These CDPKs function in ABA and Ca 2þ activation of S-type anion channels and in ABA regulation of Ca 2þ -permeable I Ca cation channels but not in R-type anion channel activity in guard cells. Furthermore, cpk3cpk6 double mutants partially impair ABA-induced stomatal closure and Ca 2þ -reactive stomatal closure but not long-term Ca 2þ -programmed stomatal closure. Partial stomatal closing responses and differential regulation of R-and S-type anion current activities in cpk3cpk6 mutant guard cells are consistent with models proposing parallel signaling mechanisms in a branched guard cell signal transduction network. Future cell-type-specific analyses of CDPKs may illuminate further cellular functions of this protein kinase family in plants.

Materials and Methods
Plant growth. Arabidopsis thaliana plants (Columbia ecotype) were grown in soil (Sungro Special blend Professional Growing Mix; Seba Beach, Alberta, Canada) in a growth room under a 16-h-light/8-h-dark cycle at a photon fluence rate of 75 lmol m À2 s À1 and a temperature of 20 8C.
Identification of CDPK genes expressed in guard cells. Degenerate oligomer-based RT-PCR was used to identify guard cell-expressed CDPK genes from guard cell-enriched cDNA libraries [55]. Degenerate oligomers were designed from two highly conserved regions that were selected from aligned CDPK peptide sequences: HRDLKPENF and DG(K/R)I(D/N)(Y/F)(E/S)EF. The degenerate oligomers used to amplify guard cell-expressed CDPK genes are as follows: 59-caymgigayytiaarccigaraaytt-39 and 59-aaytcitciwaityiatiykiccrtc-39. Total RNA was extracted from guard cell-enriched epidermal strips as described [55] with the TriZOL reagent (Invitrogen, Carlsbad, California, United States). cDNA was synthesized using the First-Strand cDNA Synthesis Kit (Amersham-Pharmacia Biotech, Little Chalfont, United Kingdom). PCR was performed as described [55]. PCR products were cloned into the pGEM-T Easy vector (Promega, Madison, Wisconsin, United States). The sequences of the cloned PCR products were determined by regular sequencing reactions. Homozygous cpk3 and cpk6 single and double mutants were identified using gene-specific primers and primers that match T-DNA sequences as described previously [53,55].
Cell-type-specific expression analyses and RT-PCR. GCPs and mesophyll cell protoplasts were isolated by enzymatic digestion from 4-wk-old Arabidopsis plants as described in [54]. Total RNA was extracted from GCPs and mesophyll cell protoplasts with TriZOL reagent. Gene-specific primers for CPK3 and CPK6 were localized downstream of the DNA insertion sites as indicated ( Figure 1B). For analyses of KAT1 (AGI No.: At5g46240), CBP (AGI No.: At4g33050), and ACTIN2 (AGI No.: At3g18780) mRNAs, gene-specific primers were used for amplifications.
Patch-clamp analyses. S-and R-type anion and I Ca channel currents in Arabidopsis guard cells were recorded as published previously and described below (S-type: [9], R-type and I Ca : [65]). Patch-clamp data were recorded using an Axopatch 200 amplifier and pClamp software and analyzed with Axograph software (Axon Instruments, Union City, California, United States).
To measure whole-cell S-type anion currents, the pipette solution contained 150 mM CsCl, 2 mM MgCl 2 , 6.7 mM EGTA, 5 mM MgATP, 10 mM HEPES-Tris (pH 7.1), and CaCl 2 to result in 2 lM free Ca 2þ . The bath solution contained 30 mM CsCl, 2 mM MgCl 2 , 1 mM CaCl 2 , and 10 mM MES-Tris (pH 5.6). Guard cells were extracellularly preincubated in the same solution with 40 mM CaCl 2 added prior to patch-clamping (Figure 2), as previously described [9]; except for in ABA regulation analyses in which guard cells were preincubated in a solution containing 1 mM Ca 2þ (Figure 3). Pretreatment of protoplasts with 40 mM CaCl 2 allows analyses of Ca 2þ activation of S-type anion channel currents in Arabidopsis (see [9] for details), and pretreatment with 1 mM extracellular Ca 2þ allows analyses of ABA regulation of S-type anion channels while cytosolic Ca 2þ is elevated, as previously shown [55]. Osmolarity of the solutions was adjusted with sorbitol to 500 and 485 mmol kg À1 , respectively. The liquid junction potential was 0.5 mV. The membrane voltage was stepped from the holding potential of þ35 mV to À145 mV for 40 s in À30-mV decrements [5]. The interpulse period was 12 s. No leak subtraction was performed. Recordings were made 7 to 10 min after access to the whole-cell configuration. For ABA activation of S-type channels, ABA was added to the bath solution prior to patchclamping and seal formation. To test the robustness of ABA insensitivity in cpk mutant guard cells, 50 lM ABA was applied to guard cells.
To measure I Ca channel currents, the pipette solution contained 10 mM BaCl 2 , 0.1 mM DTT, 4 mM EGTA, and 10 mM HEPES-Tris (pH 7.1), and the bath solution contained 100 mM BaCl 2 , 0.1 mM DTT, and 10 mM MES-Tris (pH 5.6). Guard cells were not preincubated in 40 mM CaCl 2 in these experiments. Osmolarity of the solutions was adjusted with sorbitol to 500 and 485 mmol kg À1 , respectively. NADPH (5 mM) was added to the pipette for the measurement of ABA-activated I Ca currents. ABA was added extracellularly 16 min after whole-cell recordings began. In typical experiments, activation of I Ca was observed within 1 min of ABA addition and continued for longer than 30 min. The voltage was ramped from À18 to À198 mV (after liquid junction potential correction) with a ramp speed of 180 mV s À1 . In a standard measurement, the ramp protocol was applied 16 times to obtain an average current for a cell. The interpulse period was 1 min. When H 2 O 2 (5 mM) activation of I Ca was measured, the same procedures, analysis of average currents, and timing of application were used as for ABA regulation experiments.
To measure R-type anion channel currents, the pipette solution contained 75 mM K 2 SO 4 [99], 2 mM MgCl 2 , 5 mM EGTA, 2.5 mM CaCl 2 , and 10 mM HEPES-Tris (pH 7.2) and the osmolarity was adjusted to 500 mmol kg À1 . The bath solution contained 50 mM CaCl 2 , 2 mM MgCl 2 , and 10 mM MES-Tris (pH 5.6), and the osmolarity was adjusted to 485 mmol kg À1 with sorbitol. The voltage was ramped from the holding potential of 0 mV to À200 mV with a ramp speed of À20 mV s À1 . Significance of differences between data sets was assessed by noncoupled double-tailed Student's t-test analysis.
Stomatal movement analyses. To measure changes in stomatal aperture, excised rosette leaves from 3-to 4.5-wk-old Arabidopsis plants were floated on 3 ml of opening buffer (5 mM KCl, 50 lM CaCl 2 , and 10 mM MES-Tris [pH 5.6]) for 2.5 h at 20 8C under light (photon fluency rate of 125 lmol m À2 s À1 ), so that the opening solution was taken up through petioles. After 2.5 h, ABA or CaCl 2 was added to the opening buffer. For Ca 2þ -induced stomatal closing assays, opening solution without added CaCl 2 was used. Leaves were incubated for an additional 3 h, and in the case of cpk3-2cpk6-2, for 3 h and for 45 min (Figure 6), after addition of Ca 2þ . Similar impairment in Ca 2þ -induced stomatal closure was found in cpk3-2cpk6-2 double mutant plants compared to wild-type after both 45 min. and 3 h. Furthermore, similar stomatal responses were observed in cpk3-2cpk6-2 plants 45 min. and 3 h after stimulation in parallel experiments (p . 0.50 at 0 lM CaCl 2 , p . 0.50 at 10 lM CaCl 2 , p . 0.45 at 100 lM CaCl 2 (45 min versus 3 h cpk3-2cpk6-2 data). Leaves were blended in opening solution in a Waring commercial blender (Waring Commercial, Torrington, Connecticut, United States) for 30 s. The blended material was filtered on a 100lm nylon mesh, and epidermal strips were placed onto a microscope slide with a glass coverslip. The ratio of stomatal width to length was measured with an inverted microscope. Width was measured as the distance between the inner walls of the stomata, and length was measured as the length of the guard cells [100,101]. Data from stomatal ratios mirrored stomatal width data ( Figure 6). For each sample, 20 to 30 stomata were measured. Blind experiments were also conducted, in which the genotype of leaves (wildtype or cpk3 cpk6 mutant and the added stimuli) was unknown to the experimenter.
Imposed Ca 2þ oscillation responses. To measure imposed Ca 2þ oscillation-induced stomatal closure, epidermal peels were prepared by using Hollister medical adhesive (Stock No. 7730; Hollister Inc., Libertyville, Illinois, United States) to attach a leaf abaxial side down onto a coverslip. A razor blade was used to carefully remove the cuticle and mesophyll layers of the leaf, leaving the lower leaf epidermal layer containing stomatal complexes intact. The coverslip was then sealed with grease to the bottom of a custom microscope slide with a 2-cm-diameter hole cut in the center, creating a 200-ll well. The well was then filled with depolarizing buffer ( [68]. For each experiment, four separate transients with an approximately 10-min period were imposed. Stomatal apertures of several individually mapped stomata were measured at the indicated time points for 3 h after the start of imposed oscillations. Cytosolic Ca 2þ concentration changes during imposed [Ca 2þ ] cyt transient experiments were confirmed by using the improved Yellow Cameleon 3.6 (YC3.6) fluorescence reporter [102]. Col-0 and cpk3-1cpk6-1 plants were transformed with YC3.6. Based on calibration measurements, the imposed Ca 2þ transients using YC3.6 corresponded to average cytoplasmic Ca 2þ baseline levels of approximately 0.15 lM and average peak levels of approximately 0.6 lM. The imposed oscillation protocol that was applied subjects cells to large changes in both the external Ca 2þ and Cl À concentrations. Earlier versions of cameleon were sensitive to the Cl À concentration, as Cl À affects the pH sensitivity of the yellow fluorescent protein chromophore [103] and therefore previous studies were performed with yellow cameleon YC2.1 which has important amino acid substitutions (V68L and Q69K) that substantially reduce sensitivity to the interacting Cl À and H þ ions [104,105]. In the present study, YC3.6 was used, which also has important amino acid substitutions that substantially reduce sensitivity to the interacting Cl À and H þ ions and a greatly enhanced ratio changes in response to physiological [Ca 2þ ] cyt changes [102]. Previous work with YC2.1 showed that the ratio changes observed during imposed oscillations are due to the changes in calcium, not changes in Cl À , as imposing the same chloride changes while removing extracellular Ca 2þ does not produce any YC2.1 ratio changes (supplemental data to [29]; http://www.nature.com/nature/journal/v411/n6841/extref/4111053aa. html). In these experiments, extracellular Ca 2þ was chelated with 10 mM EGTA, which lowers external Ca 2þ to nanomolar levels, to physically inhibit Ca 2þ influx that would otherwise occur at strongly hyperpolarized potentials. In further controls, replacing CaCl 2 with MgCl 2 in the bath did not produce cameleon ratio changes [106]. Further control experiments were performed with the new reporter YC3.6 in which Ca 2þ changes were imposed but not Cl À changes, by using the impermeant counteranion imminodiacetate (Figure 10). Stomata were extracellularly perfused alternately with solutions containing either 100 mM K þ imminodiacetate (no added Ca 2þ ) (pH 6.15) or 1 mM K þ imminodiacetate and 1 mM Ca 2þ imminodiacetate (pH 6.15). These controls showed YC3.6 ratio changes in the absence of imposed Cl À changes, demonstrating that Ca 2þ changes are being reported (Figure 10), as previously shown for YC2.1 [29,104,105].
Germination assays. Germination assays were performed as described previously [107]. Mutant and wild-type seeds were harvested at the same time and then sterilized using chlorine. Sterilized seeds were plated on 1/2 MS medium containing MES (0.5 g/ L) (pH 5.7), supplemented with 0, 0.3, 1, and 5 lM ABA. Plates were Figure 10. Alterations in the Extracellular Calcium Concentration in the Absence of Changes in the Extracellular Cl À Concentration Cause Large Ratio Changes in YC3.6 in Arabidopsis Guard Cells Illustrating that YC3.6, Like YC2.1, Reports Cytosolic Ca 2þ Changes and Not Cl À Changes during Imposed Transients See Materials and Methods. The YFP/CFP emission ratio is monitored in a guard cell (wild-type) expressing YC3.6. Ratio increases accompany exposure to high extracellular calcium using the membrane-impermeable counteranion, imminodiacetate. DOI: 10.1371/journal.pbio.0040327.g010 kept in the cold room for 2 d and then transferred to a growth room. Seed germination was scored on days 3, 5, 7, and 11.