Identification of cyclic GMP-activated nonselective Ca2+-permeable cation channels and associated CNGC5 and CNGC6 genes in Arabidopsis guard cells.

Cyclic GMP activates Ca2+-permeable cation channels in the plasma membrane of Arabidopsis guard cells. Cytosolic Ca2+ in guard cells plays an important role in stomatal movement responses to environmental stimuli. These cytosolic Ca2+ increases result from Ca2+ influx through Ca2+-permeable channels in the plasma membrane and Ca2+ release from intracellular organelles in guard cells. However, the genes encoding defined plasma membrane Ca2+-permeable channel activity remain unknown in guard cells and, with some exceptions, largely unknown in higher plant cells. Here, we report the identification of two Arabidopsis (Arabidopsis thaliana) cation channel genes, CNGC5 and CNGC6, that are highly expressed in guard cells. Cytosolic application of cyclic GMP (cGMP) and extracellularly applied membrane-permeable 8-Bromoguanosine 3′,5′-cyclic monophosphate-cGMP both activated hyperpolarization-induced inward-conducting currents in wild-type guard cells using Mg2+ as the main charge carrier. The cGMP-activated currents were strongly blocked by lanthanum and gadolinium and also conducted Ba2+, Ca2+, and Na+ ions. cngc5 cngc6 double mutant guard cells exhibited dramatically impaired cGMP-activated currents. In contrast, mutations in CNGC1, CNGC2, and CNGC20 did not disrupt these cGMP-activated currents. The yellow fluorescent protein-CNGC5 and yellow fluorescent protein-CNGC6 proteins localize in the cell periphery. Cyclic AMP activated modest inward currents in both wild-type and cngc5cngc6 mutant guard cells. Moreover, cngc5 cngc6 double mutant guard cells exhibited functional abscisic acid (ABA)-activated hyperpolarization-dependent Ca2+-permeable cation channel currents, intact ABA-induced stomatal closing responses, and whole-plant stomatal conductance responses to darkness and changes in CO2 concentration. Furthermore, cGMP-activated currents remained intact in the growth controlled by abscisic acid2 and abscisic acid insensitive1 mutants. This research demonstrates that the CNGC5 and CNGC6 genes encode unique cGMP-activated nonselective Ca2+-permeable cation channels in the plasma membrane of Arabidopsis guard cells.


8Br-cGMP Activates Inward Nonselective Cation Currents in Arabidopsis Guard Cell Protoplasts
Cyclic nucleotide-activated ion channels function as a type of cyclic nucleotide-gated Ca 2+ -permeable channel in mammalian cells, and some Arabidopsis CNGCs have been reported to encode functional cyclic nucleotidegated Ca 2+ -permeable channels (Leng et al., 2002;Ali et al., 2007;Gao et al., 2012). We performed experiments to test whether cyclic nucleotides could trigger currents in Arabidopsis guard cells. We found that inward currents were activated in Columbia wild-type guard cells upon extracellular application of 500 mM membranepermeable 8-Bromoguanosine 39,59-cyclic monophosphate (8Br)-cGMP (Fig. 1). We used Mg 2+ as the main divalent cation in the bath solution because it has been established that Ca 2+ -permeable channels in plants cells are often permeable to multiple divalent cations, including Ca 2+ , Ba 2+ , and Mg 2+ , and Mg 2+ would be unlikely to interfere with Ca 2+ -dependent responses (Thuleau et al., 1994;Pei et al., 2000;Véry and Davies, 2000;Wang et al., 2004;Finka et al., 2012).
Since Mg 2+ and Cl 2 were the main ions in the bath and pipette solutions under the imposed conditions, it was possible that the 8Br-cGMP-activated inward currents result from either the influx of Mg 2+ ions or the efflux of Figure 1. Whole-cell currents recorded in Columbia wild-type guard cell protoplasts of Arabidopsis. A, Typical whole-cell recording in a Columbia wild-type guard cell protoplast before (2) and after (+) application of the membranepermeable cGMP analog 8Br-cGMP (500 mM) to the bath solution. B, Average current-voltage curves of steady-state whole-cell currents recorded in Columbia wild-type guard cell protoplasts before (2) and after (+) application of 500 mM 8Br-cGMP (n = 12). Values depict means 6 SE.
Cl 2 across the plasma membrane of Arabidopsis guard cells. Therefore, we analyzed the reversal potential of the 8Br-cGMP-activated currents. Due to the voltage dependence of the 8Br-cGMP-activated currents, the reversal potential could be approximately extrapolated as the voltage where the average 8Br-cGMP-activated currents merged with the average control currents in wild-type Arabidopsis guard cells, which was approximately +21 mV (Fig. 1B) after correction of the liquid junction potential of 24 mV measured as described previously . Under the imposed conditions, equilibrium potentials of Mg 2+ and Cl 2 were +23 and 281 mV after correction of ionic activities, respectively. The reversal potential of the 8Br-cGMP-activated currents at +21 mV was close to the equilibrium potential of Mg 2+ (+23 mV) but far from that of Cl 2 (281 mV). These data show a Mg 2+ permeability of the 8Br-cGMP-activated currents. We tested the effects of La 3+ and Gd 3+ , two well-known plant Ca 2+ channel blockers (Allen and Sanders, 1994;Grabov and Blatt, 1999;Lemtiri-Chlieh and Berkowitz, 2004;Wang et al., 2004;Gao et al., 2012), on the 8Br-cGMP-activated inward currents in guard cells. We found that the 8Br-cGMP-activated inward currents were strongly inhibited by 100 mM La 3+ (n = 3) or 100 mM Gd 3+ (n = 3; Fig. 2, A and B). We further tested the selectivity of 8Br-cGMP-induced currents to cations by replacing Mg 2+ with the large cation, N-methyl-D-glucamine (NMDG), in the bath solution and found that the 8Br-cGMP-activated inward currents were abolished (Fig. 2C). In contrast, similar current amplitudes were observed after MgCl 2 was replaced by BaCl 2 at the same concentration in the bath solution (Fig. 2D), showing that the 8Br-cGMPactivated inward currents recorded in guard cells were carried mainly by these divalent cations. We also analyzed the 8Br-cGMP-activated currents using simple BaCl 2 solutions, as described previously (Pei et al., 2000), and found the 8Br-cGMP-activated Ba 2+ currents (Fig. 2E). The reversal potential of the recorded 8Br-cGMP-activated currents was +19 mV (Fig. 2E), close to the equilibrium potential of Ba 2+ (+20 mV) but distant from the equilibrium potential of Cl 2 (253 mV). 8Br-cGMP-activated Ba 2+ currents were also recorded using step-wise voltage pulses, indicating that the currents are hyperpolarization activated, but the activation is not time dependent (Supplemental Fig. S1). The "spiky" nature of these 8Br-cGMP-activated currents is similar to Ca 2+ -permeable cation channel currents recorded previously in guard cells (Schroeder and Hagiwara, 1990;Hamilton et al., 2000;Pei et al., 2000). These inward currents were observed after BaCl 2 was replaced by CaCl 2 in the bath solution with BaCl 2 in the pipette solution (n = 9) and the reversal potential was +17 mV, far from the equilibrium potential of Cl 2 (253 mV; Fig. 2E). The permeability ratio for Ba 2+ relative to Ca 2+ was 1.46, as determined according to the Goldman-Hodgkin-Katz equation (Hille, 1992). These results show that 8Br-cGMP activates nonselective Ca 2+ -permeable Figure 2. Ion selectivity and blocker sensitivity of 500 mM 8Br-cGMP-activated channels recorded in Arabidopsis Columbia wild-type guard cell protoplasts. A, Whole-cell recording in a Columbia wild-type guard cell protoplast showing the inhibition of 8Br-cGMP (500 mM)-activated currents by the Ca 2+ channel blocker La 3+ (100 mM). B, Whole-cell recordings in a Columbia wild-type guard cell protoplast showing 8Br-cGMP-activated channel inhibition by the Ca 2+ channel blocker Gd 3+ (100 mM). Two overlapping traces after Gd 3+ application are depicted. C, Whole-cell recording of a Columbia wild-type guard cell protoplast showing the inhibition of 8Br-cGMP-activated currents by the replacement of Mg 2+ in the bath solution by 100 mM NMDG-Cl. Two overlapping traces before 8Br-cGMP application and after replacement of Mg 2+ in the bath solution by 100 mM NMDG-Cl are depicted. D, Whole-cell recording of 8Br-cGMPactivated currents in a Columbia wild-type guard cell protoplast showing similar current amplitudes when MgGlu and MgCl 2 in the bath solution were replaced with 100 mM BaCl 2 . E, 8Br-cGMP-activated currents in a Columbia wild-type protoplast recorded using BaCl 2 pipette and bath solutions without NADPH in the pipette solution (see "Materials and Methods"). 8Br-cGMP-activated currents were recorded in a 100 mM BaCl 2 bath solution first, and then 100 mM BaCl 2 in the bath solution was replaced with 100 mM CaCl 2 . cation currents. Further experiments showed that the 8Br-cGMP-activated currents can also be carried by the monovalent cation Na + (Supplemental Fig. S2), similar to ABA-activated Ca 2+ -permeable cation (I Ca ) channels in guard cells (Kwak et al., 2003). Taken together, these results showed that cGMP activates nonselective Ca 2+ -permeable cation channels in the plasma membrane of guard cells. Therefore, we named the 8Br-cGMP-activated nonselective Ca 2+ -permeable cation currents I cat-cGMP .

Guard Cell-Expressed Putative CNGC Genes
Genes encoding defined plasma membrane Ca 2+permeable channels remain poorly understood in guard cells. To identify putative Ca 2+ -permeable channel genes underlying I cat-cGMP , we analyzed guard cell-specific microarray sets of Arabidopsis (Yang et al., 2008;Pandey et al., 2010) and found that CNGC1 (At5g53130), CNGC2 (At5g15410), CNGC5 (At5g57940), CNGC6 (At2g23980), CNGC15 (At2g28260), and CNGC20 (At3g17700) genes were highly expressed in guard cells (Supplemental Table  S1). We obtained cngc transfer DNA (T-DNA) insertion mutant lines for these CNGC genes. As CNGC5 and CNGC6 are among the two most closely related of the 20 CNGC genes, we generated cngc5 cngc6 double mutants by crossing the single mutants for further experiments (Fig. 3). Reverse transcription PCR and quantitative real-time (RT)-PCR analyses showed that the mRNA levels of CNGC5 and CNGC6 were greatly reduced or abolished in cngc5-1cngc6-1 (Fig. 3, B and C).

Mutations in CNGC5 and CNGC6 Impaired I cat-cGMP
To pursue the identification of genes encoding ion channels that mediate I cat-cGMP in Arabidopsis guard cells, we performed patch-clamp experiments on cngc5-1 and cngc6-1 single mutant and cngc5-1 cngc6-1 double mutant guard cells as well as on cngc2 and cngc1 cngc20 double mutants. Patch-clamp experiments showed that average current magnitudes of I cat-cGMP were only partially reduced in the two single mutants, including an insignificant average reduction in the cngc6-1 single mutant (Fig. 4, B and C; P # 0.005 for cngc5-1 versus the wild type, P = 0.585 for cngc6-1 versus the wild type at 2184 mV). Note that all data were included in these analyses, and the data from cngc6-1 guard cells included one guard cell with a large leak-like current after 8Br-cGMP addition, thus adding to a larger degree of error ( Fig. 4C; for an independent cngc6-1 data set, see Fig. 7C below). We analyzed cngc5-1 cngc6-1 double mutant guard cells, which showed strongly impaired 8Br-cGMP-activated currents (Fig. 4, A and D), compared with the Columbia wild type ( Fig. 1; P , 0.001 at -184 mV). These data indicate that CNGC5 and CNGC6 contribute to the activity of the I cat-cGMP channel currents in Arabidopsis guard cells.
To further test the effects of mutations in CNGC5 and CNGC6 on I cat-cGMP , we generated a second double mutant allele, cngc5-2 cngc6-2, in the Wassilewskija (Ws) accession. CNGC6 mRNA levels in the cncg5-2 cngc6-2 mutant were greatly reduced (Supplemental Fig. S3). The CNGC5 transcript has two splice isoforms in Ws (Supplemental Fig. S3). We thus amplified several independent CNGC5 transcripts in cngc5-2. The 59 untranslated region insertion in cngc5-2 showed either greatly reduced transcripts for two different primer pairs, including primers that amplified transcript starting 490 bp 59 of the CNGC start site (Supplemental Fig. S3), or an increased transcript level with primers amplifying from the predicted start (ATG) site of the CNGC5 transcript (Supplemental Fig. S3, B and C). These data indicate that the CNGC6 gene was clearly reduced, whereas the CNGC5 gene was less clearly affected in the cngc5-2 cngc6-2 double mutant. We confirmed significant activation of I cat-cGMP by 20 mM cGMP applied in the pipette solution in Ws wild-type guard cells (Supplemental Fig. S4). In contrast, impaired activation was observed in cngc6-2 mutant guard cells (cngc5-2 cngc6-2 double mutant allele; Supplemental Fig. S4), providing evidence for an important function of CNGC6 in mediating the cGMP-activated current in the Ws accession. The effect of the T-DNA insertion in CNGC5 and aberrant CNGC5 transcripts amplified in cngc5-2 cngc6-2 cannot, at present, unequivocally define or exclude a contribution of CNGC5 to this phenotype in the Ws accession. showing that no clear CNGC5 transcripts were detected in the cngc5-1 cngc6-1 mutant, but a possible low level of CNGC6 transcripts was detected in the cngc5-1 cngc6-1 mutant after 35 PCR cycles of amplification. C, Quantitative RT-PCR analysis of the cngc5-1 cngc6-1 double mutant shows that the transcript levels of CNGC5 and CNGC6 were greatly reduced.
Microarray data sets suggest that several other CNGC genes are expressed in guard cells, including CNGC1, CNGC2, CNGC15, and CNGC20 (Leonhardt et al., 2004;Yang et al., 2008;Pandey et al., 2010;Bauer et al., 2013;Supplemental Table S1), and it has been reported that CNGC2 functions as a cAMP-activated Ca 2+ -permeable channel in Arabidopsis guard cells . Therefore, we isolated cngc1, cngc2, and cngc20 T-DNA insertion mutants and generated a cngc1 cngc20 double mutant line in the Ws ecotype background. The cngc15 insertion mutant line (SALK_017995) had a T-DNA insertion in the 59 untranslated region of CNGC15 and did not exhibit a reduction in CNGC15 transcript level; thus, CNGC15 could not be analyzed in this study. We performed further patch-clamp experiments on cngc2 and cngc1 cngc20 mutant guard cells and found that there was no significant difference of I cat-cGMP in cngc2 mutant guard cells (Fig. 5B) and its Columbia wild type ( Fig. 5A; P = 0.474 at 2184 mV) as well as between the cngc1 cngc20 double mutant (Fig. 5D) and its Ws wild type ( Fig. 5C; P = 0.245 at 2184 mV). These results indicate that 8Br-cGMP mainly activated CNGC5 and CNGC6 in the wild type relative to CNGC1, CNGC2, and CNGC20 in Arabidopsis guard cells under the imposed conditions. cAMP has been shown to activate Ca 2+ -permeable currents in Arabidopsis guard cells using Ba 2+ as the main divalent cation in both bath and pipette solutions (Lemtiri-Chlieh and Berkowitz, 2004). As reported above, 8Br-cGMP-activated inward currents could be carried by Ba 2+ (Fig. 2, D and E). We analyzed cAMP-activated inward currents using Mg 2+ as the main charge carrier. The results showed that only modest cAMP-activated inward currents were observed in both Columbia wild-type (Supplemental Fig. S5A) and cngc5-1 cngc6-1 double mutant (Supplemental Fig. S5B) guard cells with 100 mM cAMP applied into the pipette solution. There was no significant difference between Columbia wild-type and cngc5-1 cngc6-1 double mutant guard cells (Supplemental Fig. S5; P = 0.151 at 2184 mV). These data and the lack of impairment of cGMP activation in cngc2 guard cells (Fig. 5B), compared with cAMP response impairment in cngc2 , suggest that the cGMP-activated currents in the Mg 2+ solutions analyzed here differ from the cAMP-activated currents.

YFP-CNGC5 and YFP-CNGC6 Localize to the Periphery of Nicotiana benthamiana Protoplasts
To gain insight into the cellular targeting of yellow fluorescent protein (YFP)-tagged CNGC5 and CNGC6 proteins, we transiently expressed CNGC5 and CNGC6 fused to enhanced-YFP (EYFP) in N. benthamiana protoplasts and conducted EYFP fluorescence imaging experiments using confocal microscopy. The results showed localization of YFP-CNGC5 in microdomains in the periphery of N. benthamiana protoplasts (Fig. 6). Similar microdomain localizations of plant GFP-CNGC fusions have been observed previously, including for  CNGC1, CNGC11, and CNGC12 (Ali et al., 2006;Urquhart et al., 2007). YFP-CNGC6 also localized to the periphery of N. benthamiana protoplasts. YFP-CNGC6 did not show a strong concentration in microdomains (Fig. 6) and overlapped partially with the plasma membrane stain FM4-64 (Bolte et al., 2004;Fig. 6). In contrast, control YFP-expressing protoplasts showed more broadly distributed YFP fluorescence signals, including in the cytoplasm and nuclei (Fig. 6). These results provide initial evidence that CNGC5 and CNGC6 are targeted to the periphery of plant cells at the plasma membrane, which correlates with the strong reductions in I cat-cGMP activity in the plasma membrane of cngc5-1 cngc6-1 double mutant guard cells and is consistent with recent YFP fusion localization analysis of CNGC6 (Gao et al., 2012).

Application of Intracellular cGMP Activates Currents in Wild-Type Arabidopsis Guard Cells
We used membrane-permeable 8Br-cGMP to activate large currents in guard cells, allowing facile analyses before and after application (Figs. 1, 2, 4, and 5). To test whether intracellular cGMP can activate I cat-cGMP , we conducted further patch-clamp experiments using cGMP at a lower concentration (20 mM) added to the pipette solution in whole-cell recordings. We observed obvious I cat-cGMP currents in Columbia wild-type guard cells (Fig. 7A). I cat-cGMP was reduced in cngc5-1 (Fig. 7B), cngc6-1 (Fig. 7C), and cngc5-1 cngc6-1 (Fig. 7D) mutant guard cells compared with the Columbia wild type ( Fig. 7A; P = 0.013, 0.044, and 0.032 for cngc5-1, cngc6-1, and cngc5-1 cngc6-1 mutants, respectively, compared with the wild type at 2184 mV), indicating that intracellular cGMP is capable of activating I cat-cGMP and CNGC5 and CNGC6 function in the intracellular cGMP response. Note that we cannot exclude that an additional CNGC gene(s) may be up-regulated in these cngc insertion mutant lines or that additional CNGC genes contribute to the small average residual currents in cncg5 cngc6 double mutant guard cells. However, clearly, both CNGC5 and CNGC6 function in the establishment of the cGMP-activated currents. We also tested the effects of 20 mM cAMP added in the pipette solution. We did not observe obvious activation of inward currents in Columbia wild-type guard cells (Supplemental Fig. S6), indicating that cAMP activation may occur through other channels or at higher cAMP concentrations.
Similar to I cat-cGMP , ABA-activated I Ca channels are permeable to Ba 2+ (Pei et al., 2000;Murata et al., 2001), Na + (Kwak et al., 2003), and Mg 2+ (Supplemental Fig. S7). Therefore, we performed patch-clamp experiments to test whether CNGC5 and CNGC6 function as ABA-activated I Ca channels using Ba 2+ as the main divalent cation in both bath and pipette solutions, as reported previously (Pei et al., 2000;Murata et al., 2001). Genotypeblind patch-clamp experiments showed that measurable ABA activation of I Ca currents was observed in cngc5-1 cngc6-1 guard cells (P , 0.037; Fig. 8, C and D) as well as in Columbia wild-type guard cells (P , 0.018; Fig. 8,  A and B). A small effect of the cngc5 cngc6 double mutation on ABA-activated I Ca channel currents could not be excluded, based on slightly smaller average currents and an apparent slight shift in the activation potential. The ABA-insensitive mutants growth controlled by abscisic acid2 (gca2) and abscisic acid insensitive1 (abi1-1) show impaired ABA activation of I Ca channel currents (Pei et al., 2000;Murata et al., 2001). We next analyzed I cat-cGMP in gca2 and abi1-1 mutant guard cells as well as in Landsberg erecta wild-type guard cells and found that 8Br-cGMP-activated I cat-cGMP currents were not disrupted in gca2 (Fig. 9B) and abi1-1 (Fig. 9C) guard cells compared with the Landsberg erecta wild type (Fig. 9A). These observations together strongly suggest that CNGC5 and CNGC6 alone are not essential for ABA-activated I Ca channels in Arabidopsis guard cells. To further test whether CNGC5 and CNGC6 function in ABA signaling, we pursued ABA-induced stomatal closure analyses and found that both the Columbia wild type and the cngc5-1 cngc6-1 double mutant showed functional ABA-induced responses (Fig. 10), in line with patch-clamp analyses (Figs. 8 and 9). Moreover, cytosolic Ca 2+ has also been shown to play a role in CO 2 -induced stomatal movements (Schwartz et al., 1988;Webb et al., 1996;Young et al., 2006;Xue et al., 2011). Therefore we also addressed CO 2 -induced changes in whole-plant stomatal conductance of CNGC5 and CNGC6 double mutants. The results show that mutation of CNGC5 and CNGC6 did not disrupt CO 2 -induced stomatal closure and opening (Fig. 11). Moreover, light/dark transitions showed similar intact plant stomatal conductances in wild-type and cngc5-1 cngc6-1 double mutant plants (Supplemental Fig. S8).

DISCUSSION
Intracellular Ca 2+ plays an essential role in the regulation of guard cell ion channels and stomatal movements (Schroeder and Hagiwara, 1989;McAinsh et al., 1990;Webb et al., 1996;Grabov and Blatt, 1998;   , 2000;Fan et al., 2004;Mori et al., 2006;Young et al., 2006;Marten et al., 2007;Siegel et al., 2009;Chen et al., 2010). Stimulus-triggered Ca 2+ influx across the plasma membrane of plant cells plays essential roles in many Ca 2+ signaling responses (Dodd et al., 2010). Two gene families have been proposed to encode Ca 2+ channels in plants, the Glu receptor-like gene family (Lacombe et al., 2001) and the CNGC gene family (Kaplan et al., 2007). However, there are only a few studies that have shown disruption of a defined Ca 2+ -permeable channel activity in mutant plants in defined candidate plant Ca 2+ channel genes Gao et al., 2012;Laohavisit et al., 2012).

MacRobbie
Activation mechanisms of plasma membrane Ca 2+permeable I Ca currents have been characterized in Arabidopsis guard cells (Hamilton et al., 2000;Pei et al., 2000;Murata et al., 2001;Köhler and Blatt, 2002;Kwak et al., 2003;Miao et al., 2006;Mori et al., 2006;Hua et al., 2012). However, the identities of genes encoding defined plasma membrane Ca 2+ channels in guard cells remain unknown. CNGCs have been found to be key players as cation channels in mammalian cells, and the Arabidopsis genome includes 20 CNGC genes (Kaplan et al., 2007;Ward et al., 2009). Some of these Arabidopsis CNGCs were found to function in pollen tube growth, plant immune responses, and heat responses (Yu et al., 1998;Frietsch et al., 2007;Ma and Berkowitz, 2011;Finka et al., 2012;Mach, 2012;Abdel-Hamid et al., 2013;Fischer et al., 2013;Tunc-Ozdemir et al., 2013a, 2013b. But direct voltage-clamp recordings of a cngc mutant in planta has been reported for two cngc mutants to date Gao et al., 2012). In this research, we identified CNGC5 and CNGC6, which are required for cGMP-activated nonselective Ca 2+ -permeable cation channel activity in Arabidopsis guard cells (Figs. 2, 4,  and 7). T-DNA insertion mutants in these CNGC genes show that CNGC5 and CNGC6 encode a new type of nonselective cation channel in the plasma membrane of Arabidopsis guard cells.
A previous study showed cAMP-activated Ca 2+permeable currents in Arabidopsis guard cells using Ba 2+ in the bath and pipette solutions by adding membranepermeable 1 mM dibutylyl-cAMP to the bath solution (Lemtiri-Chlieh and Berkowitz, 2004;Ali et al., 2007). In addition, it was recently reported that 50 mM dibutylyl-cAMP together with the animal phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine activated significant inward Ca 2+ currents in Arabidopsis root protoplasts using Ca 2+ as the main charge carrier, but effects of cGMP were not analyzed (Gao et al., 2012). Under our experimental conditions, we added cyclic nucleotides in the pipette solution at a lower concentration (20 and 100 mM for cAMP, 20 mM for cGMP) and observed limited cAMP activation of cation currents in guard cells (Supplemental Figs. S5 and S6) but a significant activation of the cation currents by 20 mM cGMP (Fig. 7) using solutions in which Mg 2+ was the main divalent cation. Our findings do not exclude that higher cAMP concentrations may activate such channels in a Mg 2+containing bath solution. In addition, cAMP might have weaker effects on these CNGCs compared with cGMP in guard cells. Indeed, it has been reported that mammalian CNGCs expressed in X. laevis oocytes could be activated by cGMP approximately at a 100-fold lower concentration than cAMP (Gordon and Zagotta, 1995). Further research will be needed to determine whether different interacting proteins, including possible different CNGC heteromeric complexes, or conditions affect channel activation properties in these different plant cell types, and a direct comparison of cAMP-or cGMP-activated currents in the same type of cells (root cells or guard cells) would require further experiments. Figure 10. ABA-induced stomatal closure in the Columbia wild type and the cngc5-1 cngc6-1 double mutant. A, Percentage changes in stomatal apertures relative to the stomatal apertures prior to 1 mM ABA incubation. B, Stomatal apertures from the same experiments shown in micrometers. Time-course experiments were pursued for ABA-induced stomatal closing in the Columbia (Col) wild type and the cngc5-1 cngc6-1 double mutant (genotype-blind experiments). Stomatal apertures were individually mapped, and images were captured and measured before and after the addition of 1 mM ABA (Siegel et al., 2009). Average stomatal apertures at time 215 min were 5.0 6 0.2 mm (Columbia wild type) and 4.83 6 0.18 mm (cngc5-1 cngc6-1); n = 24 individually mapped stomata each for the Columbia wild type and the cngc5-1 cngc6-1 double mutant. Values depict means 6 SE. Figure 11. Time-resolved patterns of intact whole-plant rosette stomatal conductances in response to elevated (A) and reduced (B) CO 2 concentration in cngc5-1 cngc6-1 double mutant and Columbia (Col) wild-type plants. To analyze elevated CO 2 -induced stomatal closure, the ambient CO 2 concentration was increased from 400 to 800 mL L 21 at time zero (A). To analyze low-CO 2 -induced stomatal opening, CO 2 was decreased from 400 to 0 mL L 21 at time zero (B). Data represent averages of 12 individual plants 6 SE.
Note that, although cGMP and cAMP can activate plant ion channels, it remains unknown whether these small molecules act as second messengers in cells of terrestrial plants. Indeed, whether cAMP and cGMP are produced in response to stimuli in higher plants is controversial and remains a matter of debate. Nevertheless, the ability of cGMP to activate CNGC5and CNGC6-dependent ion channel activity in guard cells, with little effect of cAMP, indicates a preference in nucleotide activation mechanisms for CNGC5 and CNGC6. Furthermore, proteins with possible guanylate cyclase activity in plants have been proposed and debated (Ludidi and Gehring, 2003;Qi et al., 2010;Ashton, 2011;Berkowitz et al., 2011). Further research is needed to determine whether this animal paradigm can be applied to plant ion channel regulation. It is conceivable that CNGC5 and CNGC6 are activated in vivo in guard cells by other natural stimuli than cGMP, and more research will be needed to investigate this question. For example, ABA activates Ca 2+ -permeable channels with properties very similar to these currents, via distinct signaling mechanisms (Murata et al., 2001;Pei et al., 2000;Köhler and Blatt, 2002;Kwak et al., 2003;Miao et al., 2006;Mori et al., 2006;Hua et al., 2012). These data indicate that other or higher order cngc mutants might well function as the ABA-activated I Ca channels.
In previous studies of reactive oxygen species and ABA activation of Ca 2+ -permeable channels, no ATP was added in the pipette solutions for electrophysiological analyses (Pei et al., 2000;Mori et al., 2006). In this study, we used ATP-free solutions containing 0.1 mM dithiothreitol (DTT) for both ABA-activated I Ca and cGMP-activated I cat-cGMP recordings, indicating that ATP was not strictly required for the activation of these channels. NADPH was included for ABA activation. Similarly, ATP was not required for ligand-activated Ca 2+ -permeable cation channels in pollen (Wu et al., 2011). Note, however, that these results do not exclude additional regulation of Ca 2+ -permeable channels by ATP-dependent mechanisms or protein kinases, as demonstrated previously for guard cell I Ca channels (Köhler and Blatt, 2002;Mori et al., 2006), given that conditions prior to whole-cell patch clamping did not preclude ATP-dependent reactions; thus, the phosphorylation state of these channels may be preset prior to patch clamping. For example, ABA activation of I Ca channels is disrupted in the Ca 2+ -dependent protein kinase cpk6 and cpk3 mutant guard cells (Mori et al., 2006).
This study and a recent study (Gao et al., 2012) provide evidence that YFP fusions of CNGC6 are localized in the vicinity of the plasma membrane, consistent with patch-clamp analyses of T-DNA insertion mutants. YFP fusions of the close homolog YFP-CNGC5 showed localization in microdomains in the cell periphery. Similar results were reported previously for other plant membrane proteins, including CNGC1, CNGC11, and CNGC12 channels (Brady et al., 2004;Ali et al., 2006;Sutter et al., 2006;Urquhart et al., 2007;Gutierrez et al., 2010). Further research is needed to determine the relevance of such microdomain accumulation of membrane proteins.
Gene chip data show that several CNGCs are expressed in Arabidopsis guard cells, including CNGC1, CNGC2, CNGC5, CNGC6, CNGC15, and CNGC20, and the expression level of CNGC5 and CNGC6 is relatively high compared with other homologs (Yang et al., 2008;Supplemental Table S1). Patch-clamp results show that mutations in CNGC5 and CNGC6 impaired 8Br-cGMP-activated currents, but T-DNA insertion mutations in CNGC1, CNGC2, and CNGC20 did not, indicating that CNGC5 and CNGC6 are important targets for guard cell cGMPactivated channel activity in the plasma membrane of Arabidopsis guard cells under the imposed conditions. Additional CNGCs may be involved in the overlapping, partially redundant formation of regulated ion channels in guard cells. Therefore, our findings do not exclude that other CNGCs in guard cells, in addition to or together with CNGC5 and CNGC6, form regulated ion channels in vivo.
ABA-activated I Ca channel activity was not disrupted by mutations in CNGC5 and CNGC6 (Fig. 8). In addition, we found that the ABA-insensitive gca2 and abi1-1 mutants, for which ABA activation of I Ca channels is impaired (Pei et al., 2000;Murata et al., 2001;Miao et al., 2006), showed I cat-cGMP similar to the wild type (Fig. 9). Taken together, we conclude that CNGC5-and CNGC6-mediated ion channels alone are not essential for these ABA responses. It has been suggested that cGMP is required, but not sufficient, for ABA-and nitric oxide-induced stomatal closure (Dubovskaya et al., 2011). Analyses of additional guard cell-expressed cGMP-dependent proteins, in particular CNGCs, in addition to CNGC5 and CNGC6 are needed for understanding a proposed link between cGMP and ABA signaling. This includes analyses of higher order cngc mutants, including expansion of the cngc5-1 cngc6-1 double mutants analyzed here, as higher order (partial) redundancy cannot be excluded at this time.
In this study, we identify two genes, CNGC5 and CNGC6, that are required for the function of a new type of nonselective Ca 2+ -permeable cationpermeable channel activity in the plasma membrane of plants. CNGC5 and CNGC6 are required for the cGMP activation of I cat-cGMP in the guard cell plasma membrane.

Plant Materials and Growth Conditions
Arabidopsis (Arabidopsis thaliana) plants (Columbia, Landsberg erecta, and Ws ecotypes) were grown in soil (Sunrise) in a growth chamber (Conviron) under a 16-h-light/8-h-dark cycle at a photon fluence rate of approximately 75 mmol m 22 s 21 during the day, a humidity of approximately 75%, and a temperature of 21°C 6 0.5°C. The cngc2 (SALK_129133), cngc5-1 (SALK_149893), and cngc6-1 (SALK_042207) mutants were obtained from the Arabidopsis Biological Resource Center. cngc5-2 (FLAG_295E04) and cngc6-2 (FLAG418_D11) were in the Ws background and obtained from the INRA. cngc1 and cngc20 were in the Ws background and kindly provided by Dr. Hillel Fromm (INRA) and the Arabidopsis knockout facility at the University of Wisconsin Biotechnology Center (Krysan et al., 1999;Sunkar et al., 2000).

Patch-Clamp Experiments
Guard cell protoplasts of Arabidopsis were isolated enzymatically as described previously (Vahisalu et al., 2008). For patch-clamp experiments of I cat-cGMP , the bath solution contained 92.5 mM MgGlu, 7.5 mM MgCl 2 , and 10 mM MES-HCl (5 mM HCl), pH 5.6, and osmolarity was adjusted to 485 mmol L 21 using D-sorbitol. The pipette solutions contained 9.75 mM MgGlu, 0.25 mM MgCl 2 , 4 mM EGTA, and 10 mM HEPES-Tris, pH 7.1, and osmolarity was adjusted to 500 mmol L 21 . Each day, 0.1 mM DTT was freshly added in the bath and pipette solutions, and no NADPH was added in the pipette solution for all cGMP-dependent I cat-cGMP recordings. Control experiments performed in the absence of DTT showed that 8Br-cGMP continued to activate I cat-cGMP (n = 4 guard cells). A voltage ramp protocol from 2180 to +20 mV (holding potential, 0 mV; ramp speed, 200 mV s 21 ) was applied for I cat-cGMP recordings. Whole-cell currents were recorded every minute for 10 min after accessing whole-cell configurations with patch-clamp seal resistances of no less than 10 GV. Data prior to 8Br-cGMP exposure were used as preexposure baseline control conditions. Subsequently, cyclic nucleotide-activated currents were recorded 30 times per minute after cyclic nucleotide (8Br-cGMP) was added to the bath solution. For experiments analyzing the effects of cAMP and cGMP added to the pipette solution, the first trace recorded after accessing the whole-cell configuration was used as a baseline control, and the traces recorded subsequently were analyzed for effects of cyclic nucleotides. Liquid junction potential was 24 6 1 mV, measured as described previously  and corrected in Figure  1B. No leak subtraction was applied to the depicted data. A 1 M KCl agar bridge was used as a bath electrode to stabilize bath electrode potentials, which is needed in particular when bath Cl 2 concentrations are changed during recordings.
Whole-cell patch-clamp recordings of ABA activation of I Ca were performed as described previously (Munemasa et al., 2007;Vahisalu et al., 2008). The pipette solution contained 10 mM BaCl 2 , 4 mM EGTA, and 10 mM HEPES-Tris, pH 7.1, osmolarity was adjusted to 500 mmol L 21 using D-sorbitol, and 0.1 mM DTT and 5 mM NADPH were freshly added each day. The bath solution contained 100 mM BaCl 2 and 10 mM MES-Tris, pH 5.6, osmolarity was adjusted to 485 mmol L 21 using D-sorbitol, and 0.1 mM DTT was freshly added each day. A ramp voltage protocol from +20 to 2180 mV (holding potential, 0 mV; ramp speed, 200 mV s 21 ) was used for I Ca recordings. Initial control whole-cell currents were recorded once each minute for 16 times starting 1 to 3 min after accessing the whole-cell configurations. The seal resistance was no less than 10 GV. Then, ABA was added to the bath solution by perfusion, and guard cell protoplasts were exposed to ABA in the bath solution for 3 min. Subsequently, I Ca was recorded for another 16 min. Liquid junction potential of 218 mV was not corrected in the figures, and leak currents were not subtracted.

RNA Isolation, RT-PCR, and Quantitative RT-PCR Experiments
Total RNA was isolated with an RNeasy Plant Mini Kit (Qiagen). After treatment with RNase-free DNase I (Qiagen), first-strand complementary DNA (cDNA) was synthesized from 1 mg of total RNA using a first-strand cDNA synthesis kit (GE Healthcare) according to the manufacturer's instructions. DNA fragments for CNGC5 and CNGC6 were amplified by 35 PCR cycles using specific primers (Supplemental Table S2). The EF-1a transcript was amplified for 22 PCR cycles.
For real-time quantitative RT-PCR analysis, total RNA was extracted from 3-week-old plants using Trizol reagent (Invitrogen). cDNA was synthesized from RNA using Moloney murine leukemia virus reverse transcriptase (Promega) with oligo(dT) 15 primers (Promega). Real-time quantitative PCR was performed using TransStart Green qPCR SuperMix on a Bio-Rad CFX Connect Real Time PCR system according to the manufacturer's protocols with specific primers (Supplemental Table S2). Quantification of relative gene expression was achieved by normalization to 18S ribosomal RNA.

YFP Fusion Protein Expression Analyses
The coding regions of the CNGC5 and CNGC6 cDNAs were cloned into pENTR/D-TOPO vector using specific primers (Supplemental Table S2), sequenced, and then transferred to the N-terminal fusions to YFP destination vector pH35YG by Gateway LR recombination reaction (Invitrogen). The Agrobacterium tumefaciens strain GV3101 carrying the gene of interest was used and infiltrated at an optical density at 600 nm of 0.5 together with the p19 strain in Nicotiana benthamiana. Mesophyll protoplasts were isolated from the leaves after 5 d of infiltration according to instructions (Asai et al., 2002;Cheng et al., 2002) and then treated with FM4-64 (Invitrogen). Fluorescence imaging was analyzed by confocal microscopy (Nikon Eclipse TE2000-U) using 488-nm excitation and 500-to 550-nm emission filters for YFP or 568-nm excitation and 580-to 650-nm emission filters for FM4-64.

Whole-Rosette Stomatal Conductance Measurements
To analyze CO 2 -and light/dark transition-induced changes in whole-plant stomatal conductance, we used 25-to 28-d-old plants and a custom-made gasexchange device. The device and plant growth conditions have been described previously (Kollist et al., 2007;Vahisalu et al., 2008).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Barium currents activated by 8Br-cGMP in guard cells.
Supplemental Figure S2. Na+ permeability of the 8Br-cGMP-activated currents in guard cells.
Supplemental Figure S5. Modest inward currents activated by 100 mM cAMP.
Supplemental Figure S6. Twenty millimeters of cAMP failed to activate obvious inward currents in Columbia wild-type guard cells.
Supplemental Figure S8. Whole plant stomatal conductances in response to light/dark transition.
Supplemental Table S1. Transcriptome-derived raw expression data of CNGC transcripts in guard cells.
Supplemental Table S2. Oligonucleotides used in this work.

ACKNOWLEDGMENTS
We thank A. Boisson-Dernier for technical advice and T. Demura for the pH35YG plasmid.

AUTHOR CONTRIBUTIONS
The cGMP-activation of ionic currents in guard cells was initially found by I.M. and impairment in these currents in cngc5, cngc6 and cngc5 cngc6 mutants were found by Y.