Characterisation of a novel gene family of putative cyclic nucleotide- and calmodulin-regulated ion channels in Arabidopsis thaliana


*For correspondence (fax +49 761203 2675;
e-mail ).


In plants, cyclic GMP is involved in signal transduction in response to light and gibberellic acid. For cyclic AMP, a potential role during the plant cell cycle was recently reported. However, cellular targets for cyclic nucleotides in plants are largely unknown. Here we report on the identification and characterisation of a new gene family in Arabidopsis, which share features with cyclic nucleotide-gated channels from animals and inward-rectifying K+ channels from plants. The identified gene family comprises six members (Arabidopsis thaliana cyclic nucleotide-gated channels, AtCNGC1–6) with significant homology among the deduced proteins. Hydrophobicity analysis predicted six membrane-spanning domains flanked by hydrophilic amino and carboxy termini. A putative cyclic nucleotide binding domain (CNBD) which contains several residues that are invariant in other CNBDs was located in the carboxy terminus. This domain overlaps with a predicted calmodulin (CaM) binding site, suggesting interaction between cyclic nucleotide and CaM regulation. We demonstrated interaction of the carboxy termini of AtCNGC1 and AtCNGC2 with CaM in yeast, indicating that the CaM binding sites are functional. Furthermore, it was shown that both AtCNGC1 and AtCNGC2 can partly complement the K+-uptake-deficient yeast mutant CY162. Therefore, we propose that the identified genes constitute a family of plant cyclic nucleotide- and CaM-regulated ion channels.


Cyclic nucleotides (cNMPs) are ubiquitous second messengers in prokaryotes and eukaryotes. In the bacterium Escherichia coli (E. coli), cyclic AMP (cAMP) positively regulates operons in the absence of glucose via binding to the catabolite activator protein (CAP) ( Aiba et al. 1982). The slime mould Dictyostelium uses cAMP as a starvation signal, leading to fruit body formation ( Dormann et al. 1998). In animals, one prominent example of cAMP action is the regulation of glycogenolysis and glycogen synthesis ( Lafontan et al. 1997). Furthermore, cAMP is involved in the transduction of odorant signals, whereas cyclic GMP (cGMP) is a second messenger for visual signals ( Zagotta & Siegelbaum 1996).

In contrast, much less is known about the role of cNMPs in plants. Microinjection and pharmacological studies provide evidence that cGMP can activate signalling cascades in plants. These investigations revealed that cGMP is involved in phytochrome and gibberellic acid signal transduction ( Bowler et al. 1994; Penson et al. 1996). An important role for cAMP in the cell cycle control of tobacco cell culture was recently reported ( Ehsan et al. 1998). Inward-rectifying K+ channels that have been identified in plants contain a cNMP binding domain (CNBD) and thus represent putative targets for cNMPs. For the Arabidopsis inward-rectifying K+ channels KAT1 and AKT1, it could be shown that cGMP can modulate its activity ( Gaymard et al. 1996; Hoshi 1995). Kurosaki et al. (1994) suggest the presence of ion channels that are activated by cAMP, are permeable for Ca2+, and are negatively regulated by calmodulin (CaM). Further support for a functional connection between cNMPs and Ca2+/CaM in plants originates from experiments indicating the requirement of cGMP and Ca2+/CaM for phytochrome signal transduction in the tomato mutant aurea and in soybean ( Bowler et al. 1994). Interestingly, high levels of Ca2+/CaM can inhibit the action of cGMP in these systems, whereas high cGMP concentrations negatively regulate the Ca2+/CaM pathway.

Similarly, crosstalk between cNMPs and Ca2+/CaM is known to occur in rod and olfactory cyclic nucleotide-gated channels (CNG channels) in animals. These channels serve as final targets for cNMPs and a functionally significant feature is their permeability for Ca2+. CaM increases the apparent affinity constant for cNMPs, resulting in a decreased cation influx ( Liu et al. 1994).

In this paper, we report on the identification and characterisation of a novel gene family in Arabidopsis comprising four new members in addition to the previously published genes AtCNGC1 and AtCNGC2 (Köhler & Neuhaus 1998). All members of this family show homology to animal CNG channels and plant inward-rectifying K+ channels. We demonstrate that both AtCNGC1 and AtCNGC2 can partly complement the K+-uptake-deficient yeast mutant CY162, providing evidence that these genes code for functional transport proteins.


Isolation of AtCNGC1–4

The recently identified cDNA sequences of AtCNGC1 and AtCNGC2 (Köhler & Neuhaus 1998) were used to screen the Arabidopsis database and four new sequences were obtained (two of which we analysed in greater detail). One genomic clone (GenBank database accession no. b18973.gbÈgss) included an open reading frame of 75 amino acids that showed 82% amino acid sequence identity to AtCNGC1. The full-length cDNA, which was designated AtCNGC3, was isolated by 5′ and 3′ RACE using a cDNA library from Arabidopsis. The coding sequence of AtCNGC3 encompasses 706 amino acids.

The second genomic clone (GenBank database accession no. AB010695; 28449–32733 bp, reverse orientation) contained the full coding sequence (694 amino acids) of a new gene, which was designated AtCNGC4. The deduced protein has 32% amino acid sequence identity to AtCNGC1. The full-length cDNA of this gene was obtained by RT–PCR from Arabidopsis mRNA. Southern blot analysis using different restriction endonucleases indicate that AtCNGC1–4 are single copy genes (data not shown).

The third and fourth clone found in the GenBank database (accession nos AB013396 and AC005170) showed significant homology to AtCNGC1–4. The corresponding cDNA sequences were predicted by computer analysis but require experimental confirmation. The preliminary cDNAs derived from the genomic clones AB013396 (53879–57199 bp) and AC005170 (67103–64371 bp, reverse orientation) were designated AtCNGC5 (710 amino acids) and AtCNGC6 (747 amino acids), respectively. Two partial cDNAs encoding AtCNGC5 and AtCNGC6 were isolated from an Arabidopsis cDNA library. These results together with Northern analysis for AtCNGC1–4 (data not shown) demonstrate that all six genes are transcribed.

Primary structure analysis

Alignment of the amino acid sequences of AtCNGC1–4 revealed high conservation throughout almost the entire length of the polypeptides ( Fig. 1). In contrast, the N-termini of AtCNGC1–4 were considerably divergent. Table 1 summarises the amino acid sequence identities and similarities among all AtCNGC proteins, including the preliminary sequences of AtCNGC5, AtCNGC6 and the sequence of the recently identified calmodulin-binding transporter HvCBT1 from barley ( Schuurink et al. 1998 ).

Figure 1.

Sequence comparison of AtCNGC1–4.

Regions of sequence identity or similarity among all four aligned sequences are indicated by a black background, amino acid residues conserved in three sequences are highlighted by a grey background. The six putative membrane spanning regions (S1–S6), the pore region (P), the CNBD and the predicted CaM binding site are indicated.

Table 1.  Per cent amino acid sequence identity and similarity among AtCNGC1–6 and HvCBT1 ( Schuurink et al. 1998 )
  1. The numbers in parentheses show the amino acid sequence similarity.

AtCNGC132 (43)60 (69)32 (42)57 (64)56 (65)62 (70)
AtCNGC2 32 (43)46 (56)34 (42)34 (42)32 (43)
AtCNGC3  31 (40)49 (59)49 (59)54 (63)
AtCNGC4   32 (40)32 (40)33 (41)
AtCNGC5    78 (81)52 (62)
AtCNGC6     51 (61)

The C-termini of AtCNGC1–6 have significant sequence similarities to CNG channels, ether-à-go-go (eag) K+ channels, and inward-rectifying K+ channels from plants. In all cases approximately 42% sequence similarity was found. Figure 2(a) shows an alignment of the amino acid sequences of AtCNGC1–4 that are similar to CNBDs of proteins such as eag K+ channels, CNG channels, cAMP- and cGMP-dependent kinases, and CAP of E. coli. The CNBDs of AtCNGC1–4 contain several highly conserved residues that are important determinants for cNMP binding ( Shabb & Corbin 1992). Aspartate replaces a conserved glutamate in AtCNGC2–4 (D600, D548, D565, respectively), and a serine replaces a conserved arginine in AtCNGC1–4 (S575, S618, S564, S583, respectively). However, the same replacements are found in the eag K+ channel from Drosophila which contains a functional CNBD ( Bruggemann et al. 1993 ).

Figure 2.

Analysis of the AtCNGC1–4 polypeptide sequences.

(a) Multiple sequence alignment of CNBDs. Putative CNBDs of AtCNGC1–4 are aligned with CNBDs of the Drosophila eag K+ channel, DmEAG ( Warmke et al. 1991 ); the bovine olfactory CNG-channel, Bolf ( Ludwig et al. 1990 ); the bovine rod photoreceptor CNG-channel, Bret ( Kaupp et al. 1989 ); the bovine protein kinase A, BcAK1 ( Titani et al. 1984 ); the bovine protein kinase G, BcGK1 ( Takio et al. 1984 ); and the catabolite gene activator protein, CAP ( Aiba et al. 1982 ). Regions of sequence similarity among all aligned CNBDs are indicated by a black background. Dark grey or light grey background highlights conserved residues present in at least eight or six of the aligned sequences, respectively. Residues important for ligand binding are marked by arrows. Regions of secondary structure of CAP are underlined.

(b,c) Hydrophobicity analysis of AtCNGC1 (b) and AtCNGC2 (c). The hydrophobicity profile of AtCNGC1 and AtCNGC2 was generated using the method of Kyte & Doolittle (1982) with a window size of 19 amino acids. The x-axis indicates the amino acid position, the y-axis indicates hydrophobicity values. Positive values show hydrophobic regions. The position of putative transmembrane domains (S1–S6) and the pore regions (P) are indicated.

(d,e) Multiple sequence alignment of pore (d) and S4 regions (e). The S4 motifs and the pore regions of AtCNGC1–4 were aligned with S4 motifs and pore regions of the Arabidopsis K+ channel AKT1 ( Sentenac et al. 1992 ), the Drosophila Shaker channel ( Pongs et al. 1988 ), the Drosophila Eag channel, DmEAG ( Warmke et al. 1991 ), and the bovine olfactory channel, Bolf ( Ludwig et al. 1990 ). Positively charged residues in the S4 motifs are boxed. Residues conserved in all pore regions are indicated by a black background. Residues conserved in six of eight aligned sequences are indicated by a dark grey background. Light grey background highlights residues conserved in five of eight sequences. The K+ channel motif GYGD and motifs in aligned positions are boxed.

Hydrophobicity analysis predicted six membrane spanning segments (S1–S6), with a region of lower hydrophobicity between S5 and S6 forming a putative pore ( Fig. 2b,c). Both the C- and N-termini are mainly hydrophilic. The plots for AtCNGC3 and AtCNGC4 are similar to the plots for AtCNGC1 and AtCNGC2, respectively (data not shown).

An alignment of the putative pore domains of AtCNGC1–4 with consensus pore sequences of CNG channels and K+ channels is presented in Fig. 2(d). The conserved K+ channel-motif GYGD is absent in all AtCNGC proteins. Instead, in AtCNGC1 and AtCNGC3 the motif GQNL is present at the respective positions, whereas in AtCNGC2 an ANDL motif and in AtCNGC4 a GN-L motif were found. This amino acid composition in the pore region resembles those of unspecific CNG channels ( Zagotta & Siegelbaum 1996).

The region between S5 and S6 extends over 96 amino acids, which is considerably longer than in CNG channels or K+ channels, where this region is only 40–50 amino acids long.

The S4 motif of AtCNGC1–4 is characterised by only four positively charged amino acids, which is less than that found in voltage-gated channels which have positively charged amino acids at every third position ( Fig. 2e).

Complementation test in the yeast mutant CY162

In the yeast strain CY162 two K+ transporters (TRK1 and TRK2) were deleted ( Ko & Gaber 1991). Cells of this strain show severely limited growth on media containing low K+ concentrations. The expression of inward-rectifying K+ channels, for example KAT1 and AKT1, rescues this mutant ( Anderson et al. 1992 ; Sentenac et al. 1992 ). We tested the ability of AtCNGC1 and AtCNGC2 to complement CY162 by expressing the corresponding cDNAs under the control of a galactose-inducible promoter ( Fig. 3). Yeast cells transformed with the empty vector as a control only grew on media containing high K+-concentrations (100 m m). KAT1, which served as a positive control, enabled growth on media containing K+ as low as 1 m m. Yeast cells expressing either AtCNGC1 or AtCNGC2 also grew on media containing 10 m m K+. No growth was detected for cells transformed with the empty vector at this low K+ concentration.

Figure 3.

Complementation test in the yeast mutant CY162.

Yeast cells of CY162 were transformed with the KAT1, AtCNGC1 and AtCNGC2 cDNAs, and the empty vector pYES2. Equal amounts of cells were dotted on plates with the indicated KCl concentrations.

The C-termini of AtCNGC1 and AtCNGC2 interact with calmodulin

The C-termini of AtCNGC1–4 are likely to be cytoplasmatic and could thus bind to possible interacting partners. Therefore, this domain of AtCNGC1 (amino acids 457–716) was used as a bait in a two-hybrid screen. We screened an Arabidopsis activation domain library and found two interacting clones, one coding for calmodulin2/3/5 and the other for calmodulin4 ( Gawienowski et al. 1993 ) ( Fig. 4). In a subsequent experiment, these clones were also shown to interact with the C-terminus of AtCNGC2 (amino acids 502–726). Figure 4(b) summarises the results of interaction tests in the two-hybrid system. An amino acid motif representing a putative CaM binding site was found in the C-terminus for AtCNGC2 (Y647 to Y660) ( Fig. 1). In AtCNGC1, a less conserved motif is present from F602 to R615. A helical wheel representation indicated that these amino acids can form a basic amphiphilic helix (data not shown) that is typical for CaM binding sites ( O’Neil & DeGrado 1990). AtCNGC3 (Y591-R604) and AtCNGC4 (Y629-Y642) have amino acid motifs representing putative CaM binding sites at similar positions. The predicted CaM binding sites for AtCNGC1–4 are positioned within the last part of the putative CNBD.

Figure 4.

Interaction of AtCNGC1 and AtCNGC2 C-termini with calmodulin 2/3/5 and calmodulin 4 in yeast.

(a) Growth of yeast cells on media lacking histidine is shown in parts A and B. The colony colour of the transformants was determined by the filter lift assay (parts C and D). Numbers indicate interacting partners, which are specified in (b).

(b) Growth and colony colour of transformed yeast cells. Growth of yeast cells was scored as follows: + + + very good growth; + + good growth; ± weak growth. The colony colour was determined by filter lift assay.


We have identified in Arabidopsis a new gene family comprising six genes that encode proteins with homology to K+ channels and CNG channels.

AtCNGC1–4 possess six putative membrane spanning domains and an unusual long pore region spanning 96 amino acids between S5 and S6. The conserved K+ channel pore motif GYGD is not present, making it unlikely that AtCNGC1–4 are monospecific for K+ ions. Considering the amino acid composition of the pore, it is more likely that AtCNGC1–4 are non-selective cation channels similar to CNG channels from animals, which are permeable for Ca2+ and monovalent cations ( Yau & Baylor 1989). AtCNGC1–4 contain a neutral residue in the pore region (glutamine in AtCNGC1 and AtCNGC3 and asparagine in AtCNGC2 and AtCNGC4) in place of the acidic glutamate that is found in the pore of CNG channels. However, replacement of this glutamate by a neutral glutamine has only marginal affects on ion permeabilities ( Eismann et al. 1994). AtCNGC1 and AtCNGC2 enabled growth of the K+-uptake-deficient yeast mutant CY162 at 10 m m K+, whereas the mutant that was transformed with the empty vector as a control did not grow at this low K+ concentration. This suggests that AtCNGC1 and AtCNGC2 can indeed enhance K+ uptake.

The proposed cytoplasmic C-termini of AtCNGC1–4 contain a CNBD that has several amino acids that are invariant in functional cNMP binding proteins. Furthermore, we located a putative CaM binding site that partially overlaps with the CNBD. This is in contrast to animal CNG channels, where a CaM binding site is located in the N-terminus ( Liu et al. 1994). CaM-binding sites of AtCNGC2 and AtCNGC4 exhibit features characteristic of other CaM binding sites, including the presence of two aromatic amino acids separated by 12 residues and the tendency to form a basic amphiphilic α helix ( Ikura et al. 1992; O’Neil & DeGrado 1990). In AtCNGC1 and AtCNGC3, the proposed CaM binding sites possess positively charged residues (R615 and R604, respectively) instead of hydrophobic or aromatic residues. However, this rather unusual feature of CaM binding sites was described by Weitz et al. (1998) for a functional CaM binding site of a rod photoreceptor channel.

We demonstrated interaction of the C-termini of AtCNGC1 and AtCNGC2 with calmodulin2/3/5 and calmodulin4 in yeast, thus confirming that the predicted CaM binding sites are functional. Binding of CaM to CNG channels in animals reduces the apparent affinity for cNMPs and enables a Ca2+ dependent feedback regulation ( Liu et al. 1994). This is achieved by disrupting the interaction between C-terminus and N-terminus by CaM in the presence of Ca2+ ( Varnum & Zagotta 1997). The overlapping of the CaM binding sites in AtCNGC1–4 with the CNBD suggests a direct inhibition of cNMP binding in the presence of Ca2+/CaM enabling a similar regulation as in CNG channels. In plants, the presence of ion channels exhibiting these regulatory properties could explain the crosstalk between cGMP and Ca2+/CaM dependent pathways involved in phytochrome signalling ( Bowler et al. 1994).

In addition to a conserved CNBD, AtCNGC1–4 contain a voltage sensor-like motif (S4), containing fewer positively charged residues as the S4 motif of voltage-gated K+ channels. These features are characteristic of animal CNG channels and their presence in AtCNGC1–4 suggests that they are gated by cNMPs rather than by changes in membrane potential.

Taken together, our investigations revealed the presence of a new gene family in Arabidopsis that shares common structural features with members of the superfamily of voltage-gated channels and, most strikingly, to CNG channels. AtCNGC1 and AtCNGC2 are probably functional channels because these proteins partly complement the yeast mutant CY162, which is deficient in K+ uptake. Based on the structural data presented in this study, we are currently focusing on the further functional characterisation of AtCNGC1–4.

Experimental procedures

Isolation of AtCNGC3–4 clones

The sequence of the genomic fragment b18973.gbÈgss was used to design primers for 5′ and 3′ RACE-PCR using the Arabidopsis Matchmaker library (Clontech) as a template. A C-terminal fragment of 1120 bp was amplified by nested PCR with a primer specific for the 5′ cloning site of the library (Match1 5′-CTATTCGATGATGAAGATACCCCA-3′) and two gene specific primers (RACE C-term1 5′-CAACTACTGTAAGAGTGGAGGAAATGAGAG-3′and RACE C-term2 (5′-GGAGAGATGCAGAGCAATGGATGTCTC-3′). To amplify an N-terminal fragment of 710 bp, the primers Match2 (5′-GATGCACAGTTGAAGTGAACTTGC-3′), RACE N-term1 (5′-GAGCAACTTGAGACAAAGATGGCGTCTAATG-3′) and RACE N-term2 (5′-CTGAGATCTTTTGGAAGACTAGAGAGAAG-3′) were used. Based on the sequence of the 710 bp fragment, the primer RACE N-term3 (5′-GCAGCACCAGCCCAAGCAGTTTCAAG-3′) was designed and was combined with primer Match2 to amplify a 960 bp fragment that contained the start ATG.

RT–PCR for AtCNGC3 was performed with total RNA isolated from 6-week-old plants according to Logemann et al. (1987) . First-strand cDNA was synthesised with reverse transcriptase from GibcoBRL, Life Technologies (Gaithersburg, MD, USA) according to the protocol supplied by the manufacturer and the gene specific primer AtCNGC3 antisense (5′-CTAGGTTTCATCCATAGGAAACTCAGGATCGGC-3′). PCR amplification was performed with primers AtCNGC3 sense (5′-TTGGATCCATGAATCCCCAAAGAAACAAATTCGTAAGG-3′) and AtCNGC3 antisense.

All PCR products were subcloned into pBluescript KS II (Stratagene) and sequenced.

The cDNA of AtCNGC4 was obtained by RT–PCR as described for AtCNGC3 utilising the primers AtCNGC4 sense (5′-TTGGATCCATGATGTTGGGTCGAATACTTGACCC-3′) and AtCNGC4 antisense (5′-TCAATAATCATCAAAATCGTCGGGATTGGG-3′).

Complementation test in CY162

The cDNAs of AtCNGC1, AtCNGC2 and KAT1 were subcloned into the yeast expression vector pYES2 (Invitrogen, San Diego, CA, USA). The Saccharomyces cerevisiae mutant strain CY162 ( Ko & Gaber 1991) was transformed by the PEG–1 method ( Gietz et al. 1992 ). Transformants were first selected on medium consisting of 0.67% (w/v) yeast nitrogen base and 2% agar supplemented with 4% glucose, 100 m m KCl and all amino acids. Uracil was omitted from the medium. Colonies were then plated on LS media ( Rodriguez-Navarro & Ramos 1984) supplemented with 4% galactose and 100 m m KCl. After one overnight incubation, cells were inoculated into liquid medium of the same composition and grown to mid log phase. Cells were washed once with water and equal amounts of cells were dotted on LS agar plates supplemented with 4% galactose and with a KCl concentration of either 1, 5, 10 or 100 m m.

Two hybrid screen and interaction tests

The C-termini of AtCNGC1 (amino acids 457–716) and AtCNGC2 (amino acids 502–726) were cloned into the vectors pGBT9 and pGAD424 (Clontech) giving rise to BD-AtCNGC1/2 C-terminus and AD-AtCNGC1/2 C-terminus, respectively. The bait vector BD-AtCNGC1 C-terminus was co-transformed with approximately 4 × 105 clones of an Arabidopsis Matchmaker library (Clontech) into yeast strain HF7c (Clontech) following the instructions of the manufacturer. Plasmid DNA recovered from two positive clones was sequenced and named AD-Calmodulin2/3/5 and AD–Calmodulin4. Interaction tests between different constructs were performed by co-transformation into HF7c. His+ colonies were assayed for β-galactosidase activity by filter lift assay.

Computer analysis

Analysis of sequencing data and sequence comparisons were carried out with the GCG software package (Genetics Computer Group, Inc., Madison, WI, USA). Graphical representations of sequence alignments were obtained with GENEDOC (www.cris. com/∼ketchup/genedoc/shtml). Hydrophobicity analysis was undertaken according to the algorithm of Kyte & Doolittle (1982).


We are grateful to Dr Richard F. Gaber (Northwestern University, Chicago, IL, USA) for kindly providing yeast strain CY162. We thank Dorothea Haasen and Chris Lundberg for helpful comments on the manuscript. This work was supported by a DFG grant GRK 257 (fellowship to C.K.) and EC B104-CT96-0101.