The group IV-A cyclic nucleotide-gated channels, CNGC19 and CNGC20, localize to the vacuole membrane in Arabidopsis thaliana

The cyclic nucleotide-gated channels, CNGC19 and CNGC20, are the sole members of the highly isolated evolutionary group IV-A in Arabidopsis plants. Prior studies have shown that the expression of both CNGC19 and CNGC20 genes are induced by salinity and biotic stress. In this report, CNGC19 and CNGC20 were determined to localize to the vacuolar membrane. Co-expression of CNGC19 and CNGC20 increased the efficiency of vacuolar localization. CNGC19 and CNGC20 are, therefore, vacuolar membrane channels that are hypothesized to mediate plant response to salinity and biotic stress.


Introduction
Plants have evolved several distinct classes of transporters and channels to facilitate the movement of cations across cellular membranes (Mä ser et al. 2001). The members of the plant cyclic nucleotide-gated channel (CNGC) family are evolutionarily and structurally related to Shaker-type K + channels, but are typically permeable to a range of monovalent cations (Leng et al. 2002;Balagué et al. 2003;Hua et al. 2003;Gobert et al. 2006;Christopher et al. 2007). Several CNGCs have also been shown to translocate divalent cations such as Ca 2+ and Mg 2+ (Leng et al. 1999;Ali et al. 2006;Frietsch et al. 2007;Urquhart et al. 2007;Guo et al. 2010), and there is a growing body of evidence that the molecular function of some CNGCs is to mediate Ca 2+ signalling (Talke et al. 2003;Ma et al. 2009). Cyclic nucleotide-gated channels are oppositely regulated by two distinct second messengers, cyclic nucleotides and Ca 2+ /calmodulin, through partially overlapping ligand-binding domains located near the C-terminus (Leng et al. 1999(Leng et al. , 2002Arazi et al. 2000;Kö hler and Neuhaus 2000;Balagué et al. 2003;Li et al. 2005;Yoshioka et al. 2006). In general, cyclic nucleotides activate CNGCs, whereas Ca 2+ /calmodulin inhibit them. The phenotypic characterization of cngc mutants and antisense lines implicates members of this channel family in the uptake and distribution of monovalent and divalent cations (Sunkar et al. 2000;Li et al. 2005;Gobert et al. 2006;Ma et al. 2006;Christopher et al. 2007;Guo et al. 2008Guo et al. , 2010Yuen and Christopher 2010), plant defence responses (Yu et al. 1998;Clough et al. 2000;Balagué et al. 2003;Yoshioka et al. 2006;Moeder et al. 2011), gravitropism (Ma et al. 2006;Borsics et al. 2007) and pollen tube elongation (Chang et al. 2007;Frietsch et al. 2007).
Plant CNGCs are divided into five phylogenetic subfamilies, designated as groups I, II, III, IV-A and IV-B (Mä ser et al. 2001). Of the 20 CNGCs in Arabidopsis, only two genes, CNGC19 and CNGC20, constitute group IV-A. These tandem genes are closely spaced on chromosome 3. Whereas CNGC19 is expressed within the vasculature of roots and leaves, CNGC20 is expressed in the root cortex, the carpel and sepals of flowers, guard cells and in leaf mesophyll cells proximal to veins (Kugler et al. 2009). Although salt stress influences the expression of both genes (Maathuis 2006;Kugler et al. 2009;Yuen and Christopher 2010), knockout mutations in either gene do not result in hypersensitivity to Na + , or altered accumulation of Na + within plant tissues (Maathuis 2006;Kugler et al. 2009). Null mutants of CNGC19 and CNGC20 exhibit reduced resistance to avirulent pathogen infection, indicating a role for these genes in plant defence responses (Moeder et al. 2011). Furthermore, CNGC19 is induced by bacterial infection and the bacterial elicitor, flg22, and T-DNA insertion mutants of CNGC19 display increased susceptibility to the fungal pathogen, Botrytis cinerea (Moeder et al. 2011).
To understand how individual CNGCs contribute to cation fluxes in plants, it is crucial to define their distribution patterns within cells. In plants, CNGCs are present at both the plasma membrane (PM) and the membrane of vacuoles. The subcellular localization of CNGCs has primarily been accomplished through the use of chimeric fluorescent reporter proteins. Fluorescent protein fusions indicate that at least four of the six CNGCs belonging to group I are targeted to the PM: CNGC3 (Gobert et al. 2006), CNGC10 ) and CNGC11 and CNGC12 (Baxter et al. 2008). CNGC18, a pollen-specific member of group III, was shown to localize to a region of the PM near the tip of expanding pollen tubes (Chang et al. 2007;Frietsch et al. 2007). The group II members, CNGC7 and CNGC8, are also expressed exclusively in pollen, but localize to the tonoplast (Chang et al. 2007). The subcellular distribution of CNGCs has been studied in a few cases by immunolocalization. Using paralogue-specific antibodies, CNGC5 (group II) was detected at the PM of leaf protoplasts by immunofluorescence microscopy, and CNGC10 was shown by high-resolution immunoelectron microscopy to pass through the secretory pathway and localize to the PM ). In contrast, no experimental information is available on the subcellular localizations of CNGC19 or CNGC20. Protein sorting algorithms predict that CNGC20 resides in the chloroplast (Sherman and Fromm 2009). However, no CNGC-related sequences were identified in two comprehensive, experimentally derived datasets of the Arabidopsis chloroplast proteome, AT_CHLORO (Ferro et al. 2010) and the Plastid Proteome DataBase (van Wijk and Baginsky 2011).
In this report, we utilized fluorescent reporter protein fusions and immunolelectron microscopy to determine the subcellular destinations of CNGC19 and CNGC20. We demonstrated that both channels are targeted to vacuolar membranes. Our findings suggest that group IV-A CNGCs mediate plant responses to salinity and pathogen infection by facilitating the movement of cations between the central vacuole and the cytosol.

Transient expression in leaf protoplasts
Protoplasts were isolated from 3-to 4-week-old Arabidopsis leaves and transfected with plasmid DNA using the protocol described by He et al. (2006). For standard transfections involving only one reporter construct, 20 mL of 1 mg mL 21 plasmid DNA were added per 200 mL of protoplasts suspended in MMg solution (0.4 M mannitol, 15 mM MgCl 2 , 4 mM MES, pH 5.7). In cases where protoplasts were simultaneously co-transfected with two reporter constructs, a 20-mL mixture containing 10 mg of each plasmid was utilized. After transfection, the protoplasts were incubated in W5 solution (154 nM NaCl, 125 mM CaCl 2 , 5 mM KCl, 2 mM MES, pH 5.7). Samples were analysed 18+ hours post-transfection with an Olympus FV-1000 laser scanning confocal microscope. For mitochondria staining, the protoplasts were incubated in W5 solution containing 0.5 mM MitoTracker Orange CMTMRos (Life Technologies, Grand Island, NY, USA) 30 min prior to analysis. The excitation/emission filters utilized for fluorescence detection were 488/505-525 nm for GFP(S65T), 543/585-615 nm for mCherry and MitoTracker Orange, and 633/650 nm for chlorophyll autofluorescence. Confocal microscopy was performed at the Biological Electron Microscope Facility (University of Hawaii, Manoa, HI, USA).

Immunoelectron microscopy
An affinity-purified rabbit polyclonal antibody was raised against a unique region of CNGC20 (residues 664 -678, sequence Ac-CLERSSVNPDGTRIR-amide) by New England Peptide (Gardner, MA, USA). For immunogold labelling, developing roots and leaves were preserved by highpressure freezing/freeze-substitution techniques as described previously (Andème Ondzighi et al. 2008). The resin-embedded sections were placed on formvar-coated nickel slot grids and blocked for 30 min with 3 % (w/v) non-fat dried milk solution in 0.01 M phosphate-buffered saline pH 7.2 containing 0.1 % Tween-20 (PBST). The sections were washed in PBST and then incubated with a 10-fold dilution of the CNGC20 primary antiserum or the rabbit pre-immune serum (negative control). After washing, a 25-fold dilution of secondary antibody, goat anti-rabbit IgG conjugated to 15 nm gold particles (Ted Pella, Inc.), was added for 2 h at room temperature. Sections were washed and stained with uranyl acetate and lead citrate. All observations were performed using a Philips CM10 microscope (Philips, Hillsboro, OR, USA). The CNGC20 antiserum was tested by immunoblot analysis [as shown in Supporting Information] by the method described in Neuteboom et al. (2009).

Results
The hydrophilic N-terminus of group IV-A CNGCs is novel and conserved CNGC19 and CNGC20 polypeptides possess the major structural features common to members of the plant CNGC family. These include a central hydrophobic core region consisting of six membrane-spanning a-helices (S1-S6), and partially overlapping cyclic nucleotidebinding and calmodulin-binding domains situated near their C-termini ( Fig. 1). Interestingly, the N-terminal hydrophilic ends of CNGC19 and CNGC20 [171 and 205 amino acids (aa) in length, respectively] are substantially longer than the N-termini of other Arabidopsis CNGC paralogues (30 -126 aa; Table 1). The hydrophilic N-termini of CNGC19 and CNGC20 share 59.6 % protein sequence identity, but lack significant sequence similarity to the N-termini of other Arabidopsis CNGCs. According to the neural network subcellular prediction program, TargetP, both CNGC19 and CNGC20 potentially harbour chloroplast localization signals (Table 1), with hypothetical transit peptide cleavage sites after positions L48 of CNGC19 and R25 of CNGC20.
We compared the N-terminal segments of CNGC19 and CNGC20 with those belonging to group IV-A CNGCs encoded by the sequenced genomes of poplar (Populus trichocarpa), rice (Oryza sativa L. ssp. japonica) and moss (Physcomitrella patens ssp. patens). The rice and poplar orthologues have N-terminal hydrophilic regions that are comparable in length to CNGC19 and CNGC20 (183 -209 aa), while the N-terminus of the P. patens orthologue is substantially shorter (135 aa; Table 1). Figure 1. Domain arrangement of CNGC19 and CNGC20. Shaded boxes indicate the positions of the N-terminal 19-aa motif, the six transmembrane segments (S1 -S6) of the hydrdophobic core region and the C-terminal overlapping cyclic nucleotide-binding (CNBD) and calmodulin-binding domains (CaMBD). FL corresponds to the full-length versions of CNGC19 and CNGC20, and N1 and N2 represent portions of the N-terminal region utilized in GFP fusion constructs.
Protein sequence alignment revealed several strictly conserved residues within the hydrophilic N-terminal regions of group IV-A CNGCs, including a 19-aa interval with the conserved sequence LxxSGxLGxCxDPxCxxCP (Fig. 2). Unlike CNGC19 and CNGC20, the group IV-A CNGCs of rice, poplar and moss were not predicted to localize to chloroplasts (Table 1).

GFP fusions with CNGC19 or CNGC20 are not targeted to chloroplasts
To determine whether CNGC19 and CNGC20 truly possess N-terminal chloroplast sorting signals, constructs were developed for the expression of the first 88 aa of CNGC19 or the first 117 aa of CNGC20, translationally fused to the N-terminus of the green fluorescent protein variant, GFP(S65T). The corresponding chimeric proteins, designated as CNGC19 N1 -GFP and CNGC20 N1 -GFP, were transiently expressed in Arabidopsis leaf protoplasts under the CaMV 35S promoter. Despite harbouring the predicted chloroplast transit peptide sequences of their respective CNGC paralogues (Fig. 1), expression of CNGC19 N1 -GFP and CNGC20 N1 -GFP in protoplasts resulted in diffuse cytoplasmic fluorescence ( Fig. 3B 3A). The CNGC19 N1 and CNGC20 N1 protein segments span from the initiator Met residue to just prior to the conserved N-terminal 19-aa motif of group IV-A CNGCs. To address the question of whether a longer interval of the Nterminus was required for chloroplast targeting, a second set of GFP fusions was generated using nearly the entire N-terminal region (CNGC19 N2 : aa 1 -166; CNGC20 N2 : aa 1 -200). Unlike their shorter counterparts, the longer CNGC19 N2 -GFP and CNGC20 N2 -GFP fusions displayed punctate GFP fluorescence patterns. However, these punctate signals did not overlap with chlorophyll autofluorescence ( Fig. 3D and E), indicating that the CNGC19 N2 and CNGC20 N2 sequences directed the chimeric proteins to intracellular structures which were distinct from chloroplasts. In addition to the truncated CNGC-GFP fusions, the localization patterns of fusions containing the fulllength sequences of CNGC19 (CNGC19 FL -GFP) or CNGC20 (CNGC20 FL -GFP) were examined. Several distinct localization patterns were observed in protoplasts transfected with the full-length CNGC19 and CNGC20 fusion constructs. The CNGC19 FL -GFP fusion labelled elliptical structures ( Fig. 4A), short non-spherical endomembranes (Fig. 4B), as well as intracellular structures possessing an extended endomembrane morphology (Fig. 4C). The CNGC20 FL -GFP fusion typically exhibited punctate labelling ( Fig. 4D), with occasional weak labelling of extended endomembranes (Fig. 4E). The endomembrane localization patterns of CNGC19 FL -GFP and CNGC20 FL -GFP were not consistent with labelling of the chloroplast envelope, as contiguous endomembranes labelled by the full-length CNGC fusions were sometimes observed wrapping around two or more chloroplasts (arrows in Fig. 4C and E). A marker for the chloroplast envelope inner membrane was generated by fusing A. thaliana plastidic nucleotide transporter 2 (NTT2) to the N-terminus of the monomeric red fluorescent protein (RFP) variant, mCherry. Protoplasts simultaneously expressing NTT2-mCherry in combination with either CNGC19 FL -GFP or CNGC20 FL -GFP did not display significant co-localizing GFP and RFP fluorescence (Fig. 5). The NTT2-mCherry marker consistently labelled the perimeter of individual chloroplasts, indicating that the trafficking of proteins targeted to the chloroplast envelope was not impaired in the leaf protoplast transient expression system. Figure 2. Sequence alignment of group IV-A CNGCs from Arabidopsis, poplar, rice and moss. Three putative members of group IV-A are present in poplar (GenBank: EEE95941.1, EEF07751.1 and EEF07752.1), one in rice (GenBank: BAD16877.1) and one in P. patens (COSMOSS v1.6: Pp1s204_120V6). Residues matching the consensus (five or more sequences) are shaded in black, and similar residues are shaded in grey. Amino acids strictly conserved among all sequences are denoted by an asterisk below the alignment. The 19-aa conserved interval (LxxSGxLGxCxDPxCxxCP) and first transmembrane domain (S1) are boxed in red. The indicated position of S1 is based on TMHMM 1.0 predictions for both CNGC19 and CNGC20. CNGC19 FL -GFP and CNGC20 FL -GFP co-localize with vacuole membrane markers aTIP-and gTIP-mCherry To address the possibility that CNGC19 and CNGC20 are targeted to vacuolar membranes, co-transfection experiments were performed with the vacuole membrane markers, a-tonoplast intrinsic protein (aTIP)-mCherry and gTIP-mCherry. Elliptical structures labelled by CNGC19 FL -GFP were co-labelled by aTIP-mCherry (Fig. 6A) and gTIP-mCherry (Fig. 6D). The CNGC19 FL -GFP fusion also co-localized with aTIP-and gTIP-mCherry within nonelliptical membranes (Fig. 6B, C and E). Although these membranes occasionally resembled the limiting membrane of the central vacuole (Fig. 6E), in many instances CNGC19 FL -GFP co-localized with short regions of intense vacuole marker fluorescence ( Fig. 6B and C).
Comparison of the signal pattern of CNGC20 FL -GFP with that of aTIP-or gTIP-mCherry revealed that CNGC20 FL -GFP co-localized with the vacuole markers in both extended endomembranes (Fig. 7A-C) and ring-like structures ( Fig. 7A and B). However, neither vacuole marker co-localized with the punctate CNGC20 FL -GFP signals ( Fig. 7A -C), implying that a substantial proportion of CNGC20 FL -GFP fusion protein was not incorporated into vacuolar membranes.
A CNGC19 FL -mCherry fusion was generated to directly compare the subcellular localization patterns of fulllength CNGC19 and CNGC20 within the same cell. Protoplasts co-transfected with CNGC19 FL -mCherry and CNGC20 FL -GFP displayed completely co-localizing patterns of GFP and RFP fluorescence (Fig. 8). In addition, the presence of the frequent punctate structures derived from CNGC20 FL -GFP was greatly diminished when it was co-expressed with CNGC19 FL -mCherry.
Punctate CNGC20 N2 -GFP and CNGC20 FL -GFP signals co-localize with the Golgi marker Man49-mCherry Co-transfection experiments with additional organelle markers were performed to investigate the punctate labelling pattern of CNGC20 FL -GFP. The CNGC20 FL -GFP punctate signals did not overlap with BiP1-mCherry-HDEL [Supporting Information], an endoplasmic reticulum (ER) marker generated by fusing A. thaliana ER luminal binding protein 1 (BiP1) to a modified version Figure 4. Subcellular localization pattern of GFP fused to full-length CNGC19 or CNGC20. Confocal laser scanning microscope images of leaf protoplasts transiently expressing (A -C) CNGC19 FL -GFP or (D, E) CNGC20 FL -GFP. The arrows indicate contiguous membranes labelled by CNGC19 FL -GFP (C) or CNGC20 FL -GFP (E) that wrap around multiple chloroplasts. Column 1, GFP signal (green); column 2, chlorophyll autofluorescence (red); column 3, merged GFP and chlorophyll signals. Scale bars represent 5 mm. of mCherry harbouring the C-terminal ER retention motif, HDEL. We also did not observe co-localization between CNGC20 FL -GFP and markers for the mitochondria [Supporting Information] or peroxisomes [Supporting Information]. However, co-expression with the Golgi marker Man49-mCherry (Nelson et al. 2007) revealed that some (but not all) of the punctate CNGC20 FL -GFP signals overlapped with Man49-mCherry fluorescence (Fig. 9A).

Localization of native CNGC20 at the membrane of vacuoles by immunoelectron microscopy
To verify that CNGC20 is targeted to vacuolar membranes, immunolocalization experiments were performed using an affinity-purified rabbit polyclonal antiserum raised against a peptide unique to CNGC20. Arabidopsis root sections immunolabelled with the CNGC20 antibody were examined by transmission electron microscopy (TEM). Anti-CNGC20 labelling was observed at the membranes of vacuoles (Fig. 10A, C and D). Labelling was also detected at the ER ( Fig. 10A and B), and on darker-stained areas within the cytoplasm (Fig. 10A and D), which may represent CNGC20 at various stages along the secretory pathway. We did not detect labelling in control samples challenged with the rabbit pre-immune serum in place of the anti-CNGC20 antibody [Supporting Information] or in leaf sections containing chloroplasts incubated with the anti-CNGC20 antibody [Supporting Information]. Unlike the over-expression that occurs when CNGC20 FL -GFP is transiently expressed in protoplasts under the CaMV 35S promoter, immunolocalization allows for the detection of CNGC20 under its normal expression pattern in plants. Thus, immunoelectron microscopy confirmed the presence of native CNGC20 in planta at the membrane of vacuoles. Figure 7. Comparison of CNGC20 FL -GFP localization with markers for the tonoplast. Confocal laser scanning microscope images of leaf protoplasts co-transfected with CNGC20 FL -GFP and (A, B) aTIP-mCherry or (C) gTIP-mCherry. Column 1, GFP signal (green); column 2, RFP signal (red); column 3, merged GFP and RFP signals; column 4, merged GFP and RFP signals with chlorophyll autofluorescence (blue). Scale bars represent 5 mm. Figure 8. Simultaneous expression of CNGC19 FL -mCherry and CNGC20 FL -GFP. (A, B) Confocal laser scanning microscope images of leaf protoplasts co-transfected with CNGC19 FL -mCherry and CNGC20 FL -GFP. Column 1, GFP signal (green); column 2, RFP signal (red); column 3, merged GFP and RFP signals; column 4, merged GFP and RFP signals with chlorophyll autofluorescence (blue). Scale bars represent 5 mm.

Discussion
In Arabidopsis, the CNGC phylogenetic tree consists of 20 members, of which CNGC19 and CNGC20 are the sole members of subgroup IV-A. They are distinguished among CNGCs by their long and novel conserved N-termini, yet retain key features common to all CNGCs such as the overlapping cyclic nucleotide-and calmodulin-binding domains of the C-terminus, six predicted membrane spanning domains and a close resemblance to Shaker-type K + channels. To better understand the roles that CNGC19 and CNGC20 play in regulating cation fluxes in plants, we investigated their subcellular locations using the leaf mesophyll protoplast transient expression system. Although the conserved hydrophilic N-termini of CNGC19 and CNGC20 are predicted to contain chloroplast transit peptides, translational fusions of GFP with CNGC19 or CNGC20 did not localize to chloroplasts. We instead found that CNGC19 FL -GFP and CNGC20 FL -GFP co-localized with markers for the vacuole membrane (aTIP-mCherry; gTIP-mCherry) when simultaneously expressed in protoplasts. Under our experimental conditions, aTIP-and gTIP-mCherry typically did not appear to label the limiting membrane of the large central vacuole. This is consistent with previous studies showing that epitope-tagged or fluorescent protein-tagged versions of aTIP, dTIP and gTIP strongly label relatively small elliptical structures when transiently expressed in protoplasts or cultured cells (Park et al. 2004;Kim et al. 2006;Saito et al. 2011). These structures may represent either small vacuoles or ring-like extensions of the central vacuole membrane that are referred to as vacuole 'bulbs'. Vacuolar bulbs were first identified in Arabidopsis transgenic plants stably expressing gTIP-GFP under the CaMV 35S promoter (Saito et al. 2002). The intensity of gTIP-GFP fluorescence in bulbs is 3-fold higher than that in the limiting membrane of the central vacuole (Saito et al. 2002), and their formation is correlated with the fusion of small vacuoles (Saito et al. 2011). It is speculated that vacuolar bulbs may serve as a reservoir of membranes to facilitate the rapid expansion or transformation of vacuoles or plant response to salt stress (Saito et al. 2002;Boursiac et al. 2005), or as specialized subregions of the vacuole where hydrolytic activities are localized (Saito et al. 2002;Zheng and Staehelin 2011). In addition to labelling bulblike structures, CNGC19 FL -GFP and CNGC20 FL -GFP occasionally co-localized with short, non-elliptical membranes of intense vacuole marker fluorescence. We hypothesize that these co-labelled membranes correspond to regions of the tonoplast where vacuole bulbs were unfolded to expand the limiting membrane of the central vacuole.
Protoplasts expressing CNGC20 FL -GFP frequently exhibited punctate GFP fluorescence that partially co-localized with the Golgi marker Man49-mCherry. Thus, CNGC20 FL -GFP may have a lower efficiency in trafficking to the vacuole than CNGC19 FL -GFP under our experimental conditions, resulting in the accumulation of CNGC20 FL -GFP in Golgi and possibly other undefined punctate subcellular structures. We observed a decrease in punctate labelling when CNGC20 N2 -GFP was co-expressed with CNGC19 FL -mCherry, with 70 -80 % more of the cells exhibiting some vacuolar labelling, possibly indicating an increased efficiency in CNGC20 FL -GFP trafficking to the vacuolar Figure 9. Comparison of CNGC19 N2 -GFP and CNGC20 N2 -GFP with a marker for Golgi. Confocal laser scanning microscope images of leaf protoplasts co-transfected with the Golgi marker Man49-mCherry and (A) CNGC20 FL -GFP, (B) CNGC19 N2 -GFP or (C) CNGC20 N2 -GFP. Column 1, GFP signal (green); column 2, RFP signal (red); column 3, merged GFP and RFP signals; column 4, merged GFP and RFP signals with chlorophyll autofluorescence (blue). Scale bars represent 5 mm. membrane in the presence of co-expressed CNGC19 FL -mCherry. Since plant CNGC polypeptides most likely assemble as tetramers to form functional cation channels (Hua et al. 2003), we speculate that the simultaneous expression of CNGC19 FL -mCherry and CNGC20 FL -GFP causes the chimeric proteins interacting with each other to form CNGC19/CNGC20 hetero-multimers, which are transported together to the vacuole membrane.
Protoplasts expressing GFP fused to the entire hydrophilic N-terminal region of CNGC19 or CNGC20 (CNGC19 N2 -GFP; CNGC20 N2 -GFP) do not display diffuse cytosolic labelling, but instead exhibit punctate labelling. This suggests that the N-terminal regions of CNGC19 and CNGC20 contain information that influences protein sorting, but are not sufficient to direct proteins to the vacuole. Interestingly, the N-terminal regions of group IV-A CNGCs contain a novel 19-aa conserved motif (LxxSGxLGxCxDPx CxxCP). Since GFP fusions harbouring shortened versions of the N-terminal region truncated just prior to this motif (CNGC19 N1 -GFP; CNGC20 N1 -GFP) exhibit diffuse labelling of the cytosol, we speculate that the 19-aa interval may play a role in protein sorting. Interestingly, in protoplast co-transfection experiments, the Golgi marker Man49-mCherry co-localized with CNGC20 N2 -GFP, but not with CNGC19 N2 -GFP. This suggests that differences exist between the protein sorting information contained within the N-terminal hydrophilic regions of CNGC19 and CNGC20. These differences may contribute, at least in part, to the dissimilar localization patterns of our fulllength CNGC19 FL -GFP and CNGC20 FL -GFP fusions with regard to the absence or presence of punctate labelling (respectively).
The presence of CNGC19 and CNGC20 at vacuolar membranes suggests that these channels serve as pathways for the passive transport of cations between the vacuole and cytosol. Since the genes encoding CNGC19 and CNGC20 are upregulated by salt stress (Maathuis, 2006;Kugler et al. 2009;Yuen and Christopher 2010), one possible function of these channel proteins is to ameliorate the effects of deleterious levels of Na + in the cytosol by facilitating Na + redistribution between the cytosol and vacuole. It is unlikely, however, that CNGC19 and CNGC20 play a direct role in the sequestration of Na + to the central vacuole since Na + must be actively transported against its electrochemical gradient, a function performed by vacuolar Na + /H + (NHX) antiporters (Apse et al. 1999(Apse et al. , 2003. An alternative possibility is that CNGC19 and CNGC20 facilitate the plant's response to salinity by mediating Ca 2+ signalling. Salt stress induces a rapid rise in cGMP levels, and a transient increase in free cytosolic [Ca 2+ ]; at moderate salt concentrations (50 mM), suppressing cGMP accumulation with an inhibitor of guanylyl cyclases also diminishes the Ca 2+ spike (Donaldson et al. 2004). Cyclic nucleotidegated channels, which are activated by cyclic nucleotides (Li et al. 2005), could function as a link between cGMP accumulation and the influx of Ca 2+ into the cytosol (Donaldson et al. 2004). The potential involvement of vacuolar CNGCs in Ca 2+ signalling is circumstantially supported by experiments in tobacco protoplasts that have demonstrated that cAMP and cGMP can trigger the influx of Ca 2+ to the cytosol from both intracellular and extracellular Ca 2+ stores (Volotovski et al. 1998). In addition to being upregulated by salt stress, AoB PLANTS www.aobplants.oxfordjournals.org CNGC19 and CNGC20 have been implicated in the Arabidopsis response to infection by bacterial and fungal pathogens (Moeder et al. 2011). If CNGC19 and CNGC20 are indeed Ca 2+ channels, they may serve a similar molecular function in the response to both abiotic and biotic stress by mediating calcium signalling through the release of vacuolar Ca 2+ into the cytosol.

Conclusions
CNGC19 and CNGC20 are components of vacuole membranes. Under the protoplast transient expression system, CNGC20 is weakly trafficked to the vacuole.
However, co-expression of CNGC19 and CNGC20 results in efficient transport of CNGC20 to the vacuolar membrane, possibly due to the formation of heteromultimeric channels. How CNGC19 and CNGC20 influence cation fluxes within plant cells in response to salt stress and biotic stress remains unclear. Future experiments defining the permeability of these channels to various cations will clarify whether CNGC19 and CNGC20 are directly involved in the subcellular redistribution of Na + or function as mediators of Ca 2+ signalling.

Conflict of Interest Statement
None declared.

Supporting Information
The following Supporting Information is available in the online version of this article.
File 4: Figure. Pre-immune serum staining of roots. Immunolabelling of Arabidopsis cryo-fixed thin root tissue sections with rabbit pre-immmune serum and anti-rabbit 15 nm gold-conjugated secondary antiserum as a negative control. Samples were viewed via TEM as described in the Methods. No labelling was observed. G, Golgi apparatus, Cy, cytoplasm; Vc, vacuole; ER, rough endoplasmic reticulum; Pm, plasma membrane; M, mitochondrion.
File 5: Figure. Immunolabelling of leaves. Immunolabelling of Arabidopsis cryo-fixed thin leaf tissue sections with anti-CNGC20 antiserum and anti-rabbit 15 nm gold-conjugated secondary antiserum. Samples were viewed via TEM as described in the Methods. No labelling was observed in the chloroplasts (CT); some labelling was observed in the edges of vacuoles (V).
File 6: Figure. Immunoblot analysis of CNGC20 in Arabidopsis total cellular proteins. Immunoblot analysis was performed using the anti-CNGC20-specific peptide antiserum on 40 mg of total seedling proteins (14-day-old) from wild type (WT) and the CNGC20 T-DNA mutant (MUT). The homozygous T-DNA insert (SALK_129133.22.05) is in the fourth exon of the CNGC20 locus (At3g17700). The antiserum detects a single band of 84 kDa in the wild-type protein sample, which is the predicted size of CNGC20, whereas no CNGC20 protein is detected in the mutant. This indicates that the antiserum is specific to CNGC20. Coomassie-stained proteins (COOM) are shown from a duplicate gel.