Vacuolar membrane localization of the Arabidopsis‘two-pore’ K+ channel KCO1

Authors

  • Katrin Czempinski,

    Corresponding author
    1. Universität Potsdam, Institut für Biochemie und Biologie, Karl-Liebknecht-Str. 24–25, Haus 20, D-14476 Golm, Germany, and Cooperative Research Group of the Max-Planck-Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476 Golm, Germany,
    Search for more papers by this author
  • Jean-Marie Frachisse,

    1. Institut des Sciences du Végétal, CNRS UPR 2355, Avenue de la Terrasse, 91198 Gif-Sur-Yvette Cedex, France, and
    Search for more papers by this author
  • Christophe Maurel,

    1. Biochimie et Physiologie Moléculaire des Plantes, Agro-M/INRA/CNRS/UM2 UMR 5004, 2, Place Viala, F-34060 Montpellier Cedex 1, France
    Search for more papers by this author
  • Helene Barbier-Brygoo,

    1. Institut des Sciences du Végétal, CNRS UPR 2355, Avenue de la Terrasse, 91198 Gif-Sur-Yvette Cedex, France, and
    Search for more papers by this author
  • Bernd Mueller-Roeber

    1. Universität Potsdam, Institut für Biochemie und Biologie, Karl-Liebknecht-Str. 24–25, Haus 20, D-14476 Golm, Germany, and Cooperative Research Group of the Max-Planck-Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476 Golm, Germany,
    Search for more papers by this author

Summary

Potassium (K+) channels play multiple roles in higher plants, and have been characterized electrophysiologically in various subcellular membranes. The K+ channel AtKCO1 from Arabidopsis thaliana is the prototype of a new family of plant K+ channels. In a previous study the protein has been functionally characterized after heterologous expression in Baculovirus-infected insect cells. In order to obtain further information on the physiological function of AtKCO1, the gene expression pattern and subcellular localization of the protein in plants were investigated. The regulatory function of the 5′ region of the AtKCO1 gene was examined in transgenic A. thaliana plants carrying β-glucuronidase (GUS) fusion constructs. Our analysis demonstrates that the AtKCO1 promoter is active in various tissues and cell types, and the highest GUS activity could be detected in mitotically active tissues of the plant. Promoter activity was strongly dependent on the presence of a 5′ leader intron. The same overall structure was identified in two genes encoding AtKCO1-like K+ channels from Solanum tuberosum (StKCO1α and StKCO1β). To investigate the subcellular localization of AtKCO1, the channel protein, as well as a fusion protein of AtKCO1 with green fluorescence protein (GFP), were expressed in transgenic tobacco BY2 cells. In sucrose density gradients, both proteins co-fractionate with tonoplast markers (Nt-TIPa, vATPase). In fluorescence images from transgenic AtKCO1–GFP BY2 cells fluorescence was exclusively detected in the tonoplast. Thus AtKCO1 is the first cloned K+ channel demonstrated to be a vacuolar K+ channel.

Introduction

Potassium (K+) channels contribute to a variety of processes in higher plants, including leaf movements, opening and closing of stomatal pores, ion uptake in roots, and K+ translocation between cells. Two molecular classes of K+-selective channels have been described in plants so far. The unique common feature of all K+ channel subunits identified to date is the presence of a conserved pore region (P-domain) which contributes to K+ conductivity. Functional K+ channels of the first class are multimers of four pore-forming α-subunits, each subunit containing a hydrophobic core with six transmembrane domains (TM) and a single P-domain (Daram et al., 1997; Isacoff et al., 1990; Jan and Jan, 1994). Members of the second class of K+ channels, initially described in yeast, human and Drosophila, contain two P-domains within one α-subunit (Goldstein et al., 1996; Ketchum et al., 1995; Lesage et al., 1996a), suggesting that the association of only two α-subunits is sufficient to constitute a K+ selective pore (Lesage et al., 1996b). We have previously described the cloning and electrophysiological characterization of AtKCO1 from Arabidopsis thaliana, the first plant member of this so-called ‘two-pore’ K+ channel class (Czempinski et al., 1997). AtKCO1 contains a short, highly basic region (KR-motif) of unknown function at the N-terminus, four transmembrane domains, several putative phosphorylation sites in cytosolic parts, and two EF-hand motifs, i.e. potential Ca2+-binding sites at the C-terminus. Ca2+-binding properties of this part of the protein could be demonstrated in vitro by a Ca2+ mobility-shift assay (Czempinski et al., 1999). AtKCO1 elicits outwardly directed potassium currents which are dependent on physiologically relevant, elevated [Ca2+]cyt, as demonstrated for recombinant channel protein in Baculovirus-infected insect cells (Czempinski et al., 1997). With the complete sequence of the Arabidopsis genome being available, additional genes encoding novel K+ channels with homology to AtKCO1 could be identified by searching in the A. thaliana databases (Czempinski et al., 1999; Maeser et al., 2001; K.C. and B.M.-R., unpublished results). Furthermore, proteins homologous to AtKCO1 have been cloned from Samanea saman (SPOCK1; Moshelon et al., 2002) and Solanum tuberosum (StKCO1α and StKCO1β; this report).

The functional characterization of ion channels in vitro in heterologous systems is usually not sufficient to provide an understanding of their function. The expression patterns of ion channel proteins and their corresponding genes in plant tissues and cell types, and the subcellular localization of the proteins, can provide useful complementary information. Data have been accumulated on the expression patterns of Shaker-type K+ channels in plants (summarized by Czempinski et al., 1999), and provide a basis for proposing a role in xylem or phloem loading or unloading (e.g. SKOR, Gaymard et al., 1998; AKT2, Lacombe et al., 2000). In contrast, few expression data are available for ‘two-pore’ K+ channels. AtKCO1 transcripts have been found by RT–PCR in all A. thaliana tissues investigated, including flowers, leaves, roots and stem (Czempinski et al., 1997).

More features of the physiological functions of channel proteins will be found by investigating the subcellular localization and target membrane. So far, conclusions for target membranes where different K+ channels expose their activity have been based on comparisons of electrophysiological properties of heterologously expressed proteins with native K+ currents detectable in corresponding plant tissues and cell types; or, more recently, via the analysis of knock-out mutant plants (Gaymard et al., 1998; Hirsch et al., 1998; Mueller-Roeber et al., 1995; Szyroki et al., 2001).

Here we report on the cellular expression pattern of AtKCO1 in A. thaliana investigated in transgenic plants carrying promoter–reporter (β-glucuronidase) gene constructs. In addition, we demonstrate that AtKCO1 is localized in the vacuolar membrane of plant cells. Electrophysiological studies have shown that K+ channels in the tonoplast play a central role in cell osmoregulation. The localization of AtKCO1 in the vacuolar membrane opens up molecular perspectives for studying the role of this channel in vacuolar storage of ions and osmoregulation.

Results

AtKCO1-like genes have a conserved structure

Genomic fragments encompassing the 5′ upstream and coding sequence of the AtKCO1 gene were isolated and compared with the AtKCO1 cDNA (Czempinski et al., 1997). The gene contains three exons interrupted by two introns, as illustrated in Figure 1(a). A 271 bp 5′ leader intron was identified 32 bp upstream of the ATG start codon. A small intron was found within the coding region, interrupting the first EF-hand motif of the corresponding protein (Czempinski et al., 1997). According to the longest cDNA clone identified, transcription starts at 313 bp upstream of the ATG, resulting in a 42-nucleotide 5′ mRNA leader in the mature mRNA.

Figure 1.

Structure of KCO1-like genes and schematic presentation of promoter–reporter gene fusions for analysis of AtKCO1 transcriptional activity.

(a) Sequence analysis of AtKCO1-like genes from Arabidopsis thaliana and Solanum tuberosum revealed conserved positions of 5′ leader introns (intron 0) and introns within the coding regions (intron 1). Exons are denoted by boxes; 5′ and 3′ flanking regions and introns by solid lines. The 5′ untranslated exon is labelled L; translated exons are labelled with numbers (exons 1 and 2). Numbers below the gene structure indicate sizes (bp) of the corresponding fragments.

(b) KCO1–GUS1 and KCO1–GUS2 constructs carry 5′ regions of AtKCO1 of different lengths and include the 5′ leader intron, whereas KCO1–GUS3 and KCO1–GUS4 constructs were generated by deleting intron 0 or the entire region downstream of the 5′ splice site of the intron, respectively. The 5′ leader intron is indicated by a shaded box. Numbers indicate sizes (kb) of corresponding 5′ regions.

As part of our effort to analyse AtKCO1-like sequences from other plant species, we have isolated two cDNA clones, as well as corresponding genomic sequences from S. tuberosum. These genes encode AtKCO1-like proteins, StKCO1α and StKCO1β, which themselves share 71% identical amino acids and which are about 75% similar to AtKCO1 (56% identity) (Figure 2). According to hydropathy plots and sequence alignments, both StKCO1α and StKCO1β exhibit the same structural arrangement with four predicted transmembrane segments, two P-domains, and cytosolic N- and C-termini (4TM-2P structure). In addition, the N-terminal KR motif, several phosphorylation sites in putative cytosolic regions, and tandem EF-hand motifs at the C-termini are conserved between the A. thaliana and potato proteins. The potato StKCO1α/β and A. thaliana AtKCO1 genes share the same overall structure (Figure 1a): introns were present at conserved positions in both the coding region and the 5′ leader. However, leader introns of the potato genes were much larger (≈2 kbp) than in AtKCO1 (data not shown).

Figure 2.

Comparison of AtKCO1-like proteins from Arabidopsis thaliana, Solanum tuberosum and Samanea saman.

Predicted amino acid sequences of StKCO1α and StKCO1β and alignment with AtKCO1 and SPOCK1. The KR-rich sequence, putative transmembrane segments M1–M4, P-domains P1 and P2, as well as EF-hand motifs EF1 and EF2 are underlined (cf. Czempinski et al., 1997).

AtKCO1 promoter activity in transgenic A. thaliana plants

RT–PCR-based analysis in A. thaliana indicated that AtKCO1 is expressed in various tissues (Czempinski et al., 1997). In order to study the cellular specificity of AtKCO1 promoter activity throughout plant development, 5′ upstream sequences of various lengths of the AtKCO1 gene were fused to the E. coliβ-glucuronidase (GUS) reporter gene (Figure 1b), and transferred into transgenic A. thaliana. For each construct, more than eight independent transformants from the T1 and T2 progenies were analysed.

Transgenic plants harbouring fusion constructs KCO1–GUS1 and KCO1–GUS2, which both contained the 5′ leader intron but differed in lengths of the 5′ upstream region of the AtKCO1 gene, exhibited GUS activity in the same tissues and cell types (shown for KCO1–GUS2 in Figure 3). Mature seeds were collected from transgenic plants and kept for 2–3 days in a moist environment prior to GUS staining. GUS was active in the hilum, where the funiculus was attached during fruit maturation (not shown). When embryos were expelled from seeds, blue staining of the entire embryo was observed (Figure 3a). Small seedlings exhibited prominent GUS activity in the primary root, with strongest activity in the root tip and elongation zone, as well as in the root-hypocoty transition zone (Figure 3b). Seedlings from lines with a lower overall GUS level showed GUS activity only in the root tip and transition zone (not shown). Prolonged staining revealed reporter gene activity also in cotyledons (Figure 3c). In leaves of mature plants, GUS activity was detected in mesophyll cells, guard cells and vascular tissues (Figure 3e). Peeling off the epidermis prior to incubation in GUS staining solution more clearly uncovered AtKCO1 promoter activity in guard cells (Figure 3f). In flowers, GUS staining was observed in sepals (not shown), in anthers and guard cells of the filament (Figure 3g), and pollen grains. No GUS activity could be detected in petals (not shown). Like seedlings, mature plants showed prominent GUS activity in the zone of the root–shoot junction (not shown), in root tips (Figure 3i), and in the central cylinder (Figure 3k). Cross-sections (Figure 3k) revealed staining of conducting tissues and the pericycle. Strong GUS activity was also visible in adventitious lateral roots that started to develop when seedlings were kept on auxin-containing medium (not shown).

Figure 3.

Histochemical localization of GUS activity in transgenic Arabidopsis thaliana plants transformed with the KCO1–GUS2 or the KCO1–GUS3 construct.

Tissues from KCO1–GUS2 plants are shown in (a–c,e–g,i,k); tissues from KCO1–GUS3 plants are shown in (d,h,l).

(a) Embryo expelled from the seed; (b) young seedlings stained for 3.5 h and (c,d) overnight; (e) leaf cross-section, gc = guard cell, v = vascular tissue; (f) epidermal strip of a leaf showing GUS activity in guard cells; (g) anther with GUS-stained filament and guard cells; (h) anthers with GUS-stained pollen grains; (i) longitudinal view and (k) cross-section of a root, c= cortex; en = endodermis; p = pericycle; (l) root from a representative KCO1–GUS3 line.

AtKCO1 leader intron is required for enhanced expression level

To test whether the 5′ leader intron of AtKCO1 contains regulatory elements that influence the expression level of the gene, two deletion constructs were generated lacking either the intron sequence or the entire 3′ region downstream of the 5′ splice site of the intron, leading to constructs KCO1–GUS3 and KCO1–GUS4, respectively (Figure 1b). In transgenic plants harbouring either one of these constructs, GUS activity was strongly reduced and restricted to only a few cell types. No difference in the pattern of GUS staining could be observed between plants carrying the KCO1–GUS3 or KCO1–GUS4 construct (shown for KCO1–GUS3 in Figure 3), indicating that the effect on reporter gene activity can be ascribed to the absence of the leader intron sequence.

Seedlings did not show any GUS activity, even after prolonged overnight staining (Figure 3d). The same was true for mature plants, where hardly any GUS activity could be detected. In some cases the root tips or root caps and the root–shoot transition zone exhibited GUS activity, although staining intensity was strongly reduced compared to plants harbouring the KCO1–GUS1 or KCO1–GUS2 constructs (Figure 3l). In contrast, almost all the 22 KCO1–GUS3 or KCO1–GUS4 transgenic lines analysed in detail showed GUS activity in pollen grains, as was the case for KCO1–GUS1 and KCO1–GUS2 transgenic plants (Figure 3h). Altogether, these data suggest that the leader intron primarily enhances AtKCO1 gene transcription, but does not interfere with its tissue-specific expression.

Fluorescence imaging of GFP fusion protein demonstrates tonoplast localization of AtKCO1

To facilitate the investigation of the subcellular localization of AtKCO1, we prepared transgenic plant material expressing the wild-type AtKCO1 protein, or an AtKCO1–GFP fusion, under the control of the cauliflower mosaic virus (CaMV) 35S promoter. Transgenic A. thaliana plants with an enhanced level of AtKCO1 or expressing the AtKCO1–GFP protein could not be generated from three independent transformations (K.C. and B.M.-R., unpublished results), indicating a negative effect on plant development. In contrast, kanamycin-resistant tobacco BY2 cell lines could be obtained via Agrobacterium-mediated transformation, and were screened for the expression of AtKCO1 and AtKCO1–GFP via immunoblot analysis or direct imaging of GFP fluorescence in living cells. Of 33 independent transgenic BY2 cell lines expressing AtKCO1 or AtKCO1–GFP, all displayed normal growth behaviour and viability, indicating that AtKCO1 expression did not affect cellular function unfavourably.

Fluorescent images obtained by conventional and confocal laser scanning microscopy were evaluated from hundreds of cells of different transgenic lines exhibiting both high and low levels of expression of AtKCO1–GFP. The images shown here represent the typical distribution of GFP fluorescence in transgenic BY2 cells. No detectable signal was present in non-transgenic tobacco cells (data not shown). The confocal images shown in Figure 4 indicate that signal from AtKCO1–GFP was excluded from the nuclear compartment, but was present as a single fluorescent line surrounding the large central vacuole. Frequently, individual cells contained multiple small vacuoles labelled by GFP fluorescence. The presence of multiple vacuoles is a common feature of BY2 cells (J.-M. Neuhaus, personal communication). Additionally, in some cells containing a single large vacuole, a network of connecting strands most similar to transvacuolar strands (Shimmen et al., 1995) was found to exhibit GFP fluorescence (not shown). GFP fluorescence was totally excluded from the plasma membrane, as illustrated in Figure 4(a,b), where bright-field and fluorescence images are shown for the same cell.

Figure 4.

Imaging of AtKCO1–GFP fluorescence in transgenic tobacco BY2 cells and isolated vacuoles.

Fluorescence (a) and corresponding bright-field (b) confocal images of BY2 cells expressing AtKCO1–GFP fusion protein. Arrows in (a,b) indicate the position of the plasma membrane (nu = nucleus, v = large central vacuole). Green fluorescence (c) and bright-field (d) images of a vacuole isolated from a transgenic AtKCO1–GFP-expressing BY2 cell obtained by conventional microscopy.

To further prove association of AtKCO1–GFP fluorescence with the tonoplast, vacuoles were released from protoplasts of transgenic BY2 cells and analysed using conventional fluorescence microscopy. Strong GFP fluorescence was observed in the vacuolar membrane (Figure 4c,d). The images also show a slight GFP fluorescence within the vacuoles, probably resulting from emission of additional optical layers which are not excluded in conventional microscopy.

AtKCO1–GFP and AtKCO1 co-fractionate with tonoplast marker proteins

To provide further evidence for vacuolar membrane localization of AtKCO1 protein, we performed biochemical analyses of fractionated membrane proteins. As a first approach, polyclonal antibodies were raised against a peptide representing a 15 amino acid stretch of the C-terminus of AtKCO1 (see Experimental procedures). In microsomal fractions from different tissues of A. thaliana, including leaf, root, flower and complete young seedling, using up to 50 µg total microsomal protein, no protein band could be detected with the anti-AtKCO1 antiserum, indicating either a low level of AtKCO1 protein in these tissues, or that the protein was unstable under the experimental conditions used. In transgenic BY2 cells expressing AtKCO1, the antiserum detected a single band of ≈40 kDa (not shown) which corresponded to the calculated molecular mass of the AtKCO1 protein (40.7 kDa; Czempinski et al., 1997). The antibodies appeared to be highly specific for AtKCO1, as they did not recognize any cross-reacting protein in wild-type BY2 cells, or even in insect cells expressing StKCO1α or StKCO1β protein (data not shown). In a first set of experiments, total microsomal membranes from AtKCO1–GFP-expressing cells were fractionated on sucrose gradients and subsequently characterized by Western blot analysis (Figure 5a,b). AtKCO1–GFP was additionally detected in these gradients using fluorescence spectroscopy (Figure 5a). AtKCO1–GFP co-fractionated with two tonoplast marker proteins, aquaporin Νt-TIPa and vATPase, with a peak fraction at 23–27% sucrose. Both anti-AtKCO1 antiserum and anti-GFP antibodies detected a single protein band at 55 kDa (Figure 5b). The size of the protein did not closely match the calculated molecular mass of the AtKCO1–GFP fusion protein (66 kDa); however, abnormal electrophoretic mobility has been reported for many membrane proteins (e.g. Lurin et al., 2000). The fractionation did not effectively separate internal membranes from plasma membranes, as indicated by the absence of a peak fraction of the plasma membrane proteins (anti-pmATPase). However, unlike AtKCO1–GFP, pmATPase protein was also present in the 37–48% sucrose fractions (Figure 5b).

Figure 5.

AtKCO1–GFP and AtKCO1 co-fractionate with tonoplast marker proteins.

Total microsomal membranes (T) of AtKCO1–GFP- (a,b) and AtKCO1- (c) expressing cells were generally fractionated through a 16–55% sucrose gradient or enriched for plasma membrane proteins (PM).

(a) Sucrose concentration of the corresponding fractions (percentage, w/v, ▪) and fractionation profile of AtKCO1–GFP as detected by GFP fluorescence (◆). Fluorescence was measured at 508 nm with excitation at 488 nm. The values shown were derived by subtracting the background fluorescence detected in parallel gradients from a wild-type control.

(b) Western blot analysis of membrane fractions probed with antibodies raised against AtKCO1, GFP, and marker proteins Nt-TIPa (tonoplast), pmATPase (plasma membrane), E37 (plastid envelope) and vATPase (tonoplast).

(c) Western blot analysis of membrane fractions probed with antibodies raised against AtKCO1 and marker proteins Nt-TIPa (tonoplast), pmATPase (plasma membrane). Samples (10 µl) from each sucrose gradient fraction were separated by SDS–PAGE, blotted onto PVDF membrane and probed with antibodies. Equal amounts of proteins (10 µg) of plasma membrane-enriched fractions (PM) and total microsomes (T) were loaded. A minor amount of tonoplast protein was still present in the PM-enriched fraction, as detected with the Nt-TIPa antibody.

To exclude the possibility that GFP interfered with endogenous KCO1-targeting signals, we investigated as an additional control the localization of unmodified AtKCO1 protein in tobacco BY2 cells transformed with the CaMV 35S::AtKCO1 construct. Immunodetection of AtKCO1 in membrane fractions separated by sucrose gradient centrifugation showed a distribution pattern similar to that of AtKCO1–GFP, and AtKCO1 was not enriched in purified plasma membrane fractions (Figure 5c), demonstrating that the GFP fusion did not affect targeting of AtKCO1.

Taken together, both fluorescence imaging and biochemical analyses of membranes of transgenic BY2 cells demonstrate that AtKCO1 is located at the tonoplast.

Discussion

The subfamily of KCO1-like K+ channels in higher plants is characterized by a similar molecular structure of the channel proteins and a conserved intron/exon organization of corresponding genes. Both AtKCO1 and StKCO1α/β have gene structures with two introns interrupting the mRNA leader sequence and the coding region at similar positions. Other genes in A. thaliana homologues to AtKCO1 also contain conserved introns in their coding regions (K.C. and B.M.-R., unpublished results).

Expression pattern of the AtKCO1 gene

In the present work we have characterized the expression pattern of AtKCO1 by transforming into A. thaliana promoter–GUS constructs containing different 5′ regulatory regions of the AtKCO1 gene. Although these constructs are likely to reflect mainly the transcriptional activity of the AtKCO1 gene, the expression observed may also be influenced by post-transcriptional control elements residing in the mRNA leader, the 5′ leader intron, or translation signals adjacent to the initiation codon. In general, we observed pronounced GUS activity in actively dividing zones of the root; in floral tissues; and especially in growing and expanding tissues in germinating seedlings, and in adventitious lateral roots in seedlings which have been kept on auxin-containing medium. Removing the 5′ leader intron strongly reduced the level of GUS expression throughout transgenic plants. GUS activity could be detected only in those regions that showed the highest GUS activity with a reporter gene carrying the 5′ leader intron, i.e. root tips, the root–shoot transition zone, or pollen grains. Although the mechanisms by which 5′ leader introns modulate gene activity are not understood in any detail (Callis et al., 1987; Fu et al., 1995a; Fu et al., 1995b; Rethmeier et al., 1997; Rethmeier et al., 1998; Sieburth and Meyerowitz, 1997), the presence of 5′ leader introns in the potato genes StKCO1α and StKCO1β suggests that they also play a functional role in enhancing gene activity of KCO1-like genes in other plant species.

Vacuolar localization of AtKCO1 channel protein

To clarify the subcellular localization of AtKCO1 we analysed transgenic BY2 cell lines expressing GFP-tagged AtKCO1, or the wild-type AtKCO1 protein. Both biochemical analysis of microsomal proteins of cells expressing AtKCO1–GFP or AtKCO1, and fluorescence microscopy of AtKCO1–GFP-expressing cells and isolated vacuoles, revealed localization of the two-pore K+ channel in the tonoplast. This makes AtKCO1 distinct from other cloned K+ channels affiliated with plasma membrane ion channels, based on comparative electrophysiological characterization of channel activity in plant cells and heterologous expression systems. Moreover, the fluorescent images clearly indicated the absence of AtKCO1 protein from the plasma membrane in transgenic plant cells.

At present, little is known about the targeting signals directing membrane proteins to the vacuolar membrane (Jiang and Rogers, 1999). Studies on bean α-TIP (tonoplast intrinsic protein) in transgenic tobacco have demonstrated that the C-terminal transmembrane domain and parts of the cytosolic tail are sufficient for vacuolar targeting (Höfte and Chrispeels, 1992). In addition, the C-terminus was found to stabilize the protein. The AtKCO1 channel provides an additional tool for the molecular analysis of protein targeting to the tonoplast, although it might be difficult to mutagenize transmembrane domains without disrupting folding and structure of the protein.

Distinct vacuolar compartments can be distinguished between tissues or within a single plant cell, depending on the nature of stored soluble proteins, i.e. seed-type storage proteins, vegetative storage proteins or lytic enzymes (Hoh et al., 1995; Paris et al., 1996), or depending on the presence of specific aquaporin isoforms in the tonoplast. Thus reactivity to antibodies raised against α-TIP, γ-TIP and δ-TIP homologues provided a basis to distinguish between at least three vacuolar subtypes (Jauh et al., 1998; Jauh et al., 1999; Paris et al., 1996). In the experiments described here, we used an antibody raised against tobacco tonoplast aquaporin Nt-TIPa which, together with its close A. thaliana homologue ε-TIP (recently renamed TIP4;1) (Johanson et al., 2001), defines still another TIP subclass. Nt-TIPa was used as a marker protein in our studies as we knew this protein to be strongly expressed in the tonoplast of tobacco suspension cells, where it accounts for most of the small neutral solute permeability (Gerbeau et al., 1999). Thus a firm conclusion as to whether the AtKCO1 channel might be associated with a certain type of vacuoles can not be drawn from these experiments at the present stage. In order to precisely characterize the function of the channel in plant cells, it will be necessary to perform a more detailed analysis of the target organelle based on growing information about the biogenesis and maintenance of the different types of vacuoles (Bethke and Jones, 2000).

Vacuolar K+ channels have mostly been studied in Vicia faba guard cells, and have been electrophysiologically characterized as fast vacuolar (FV), slow vacuolar (SV) and vacuolar K+-selective (VK) channels. FV channels are permeable for monovalent cations (Brüggemann et al., 1999) and inhibited by elevated concentrations of cytosolic Ca2+[Ca2+(cyt)] (Tikhonova et al., 1997). SV channels are permeable for mono- and divalent cations (Pottosin et al., 1997; Ward and Schroeder, 1994) and activated by increasing [Ca2+(cyt)] (Hedrich and Neher, 1987; Reifarth et al., 1994). VK channels are also activated by increasing [Ca2+(cyt)] (Allen and Sanders, 1996; Ward and Schroeder, 1994). K+ channels have also been described in the tonoplast of Chara australis (Laver and Walker, 1991) as being regulated by the pH at the vacuolar membrane face and by Ca2+ at the cytosolic side. However, vacuolar K+ channels characterized so far do not represent the electrophysiological properties that have been detected for AtKCO1 heterologously expressed in insect cells (Czempinski et al., 1997). Both the presence of AtKCO1 in the tonoplast and previous electrophysiological characterization (Czempinski et al., 1997) provide hints for the physiological role of AtKCO1. Activation of the channel at potential exceeding the trans-tonoplast equilibrium potential for K+ (Ek) and for [Ca2+(cyt)] exceeding 200 nm will lead to a vacuolar potassium influx. Such a channel could play a role in either K+ uptake, K+ homeostasis or osmoregulation, by allowing K+ vacuolar storage driven by the trans-tonoplast potential.

The strong expression of AtKCO1 in mitotically active tissues suggests that the protein might be involved in K+ transport during cell elongation and development, which is associated with K+ uptake and osmoregulation (Lew, 1991). AtKCO1 promoter activity is also found throughout mature plants, suggesting an additional role of the K+ channel in other osmoregulatory processes and/or signalling cascades. Detailed electrophysiological analysis of the channel protein in its native membrane will contribute to the characterization of vacuolar K+ conductances and the characterization of structure–function relations at the molecular level.

Experimental procedures

Plant material

Arabidopsis thaliana cv. C24 wild-type and transgenic plants (see below) were grown in half-concentrated MS medium supplemented with 1% sucrose and solidified with 0.7% agarose under a 16 h day (140 µE, 22°C)/8 h (22°C) night regime, or in soil (Einheitserde Type GS90, Gebrüder Patzer, Simtal-Jossa, Germany; 16 h fluorescent light, 60, 120, or 180 µE, 20°C, 60% relative humidity/8 h dark, 16°C, 75% relative humidity). Transgenic A. thaliana plants were generated by vacuum infiltration (Bechthold et al., 1993) with Agrobacterium tumefaciens strain GV3101. Kanamycin-resistant plants (T0) were identified and grown from seeds. Experiments were conducted with T1 or T2 plants and two independent lines each of the T2 generation were selected for detailed analysis of GUS activity.

Suspension cultures of tobacco BY2 cells (Nicotiana tabacum L. cv. BY2) were maintained in LS medium containing 0.2 mg l−1 2,4-dichloro-phenoxyacetic acid (2,4-D) in the dark at 26°C, and transferred to fresh medium every 7 days. For protein preparation, exponentially growing cells were used 4–5 days after transfer. Transformation of BY2 cells with plasmids p35S-KCO1 and p35S-KCO1-GFP (see below) was done as described using A. tumefaciens strain GV2260 (An, 1985). Kanamycin-resistant colonies appeared after 3–4 weeks incubation at 26°C.

General procedures

Standard DNA techniques including DNA cloning and preparation of plasmid DNA were performed as described by Sambrook et al. (1989). Plasmid DNA required for sequencing purposes was prepared using Qiagen columns (Qiagen, Hilden, Germany). Sequence determinations were performed by MWG-Biotech (Ebersberg, Germany) and Replicon (Berlin, Germany). For sequence analysis the blast server at the National Center of Biological Information (NCBI, Bethesda, USA), or the University of Wisconsin gcg software package, version 8 (Devereux et al., 1984) was used. Either Pfu polymerase (Stratagene, Heidelberg, Germany) or Taq polymerase (Gibco-BRL, Eggenstein, Germany) was employed for PCR. All PCR-derived fragments were sequenced to ensure the absence of amplification errors.

Isolation of the AtKCO1 gene and homologous genes from S. tuberosum

AtKCO1 genomic clones were isolated, using the complete AtKCO1 cDNA as hybridization probe, from an A. thaliana (C24 ecotype) genomic library (kindly provided by U. Uwer, PlantTec Biotechnologie GmbH, Potsdam, Germany). Fragments corresponding to the AtKCO1 gene were subcloned into pLitmus28 (New England BioLabs, Schwalbach, Germany) giving rise to plasmids pEN12 and pSN24. These subclones were obtained using a 1.2 kb EcoRI/NcoI restriction fragment (pEN12) or a 5.2 kb SalI/NcoI restriction fragment (pSN24), respectively. Both fragments contain 5′ upstream regions of the AtKCO1 gene, including the first 57 bp of the coding region. In addition, a 2.4 kb SpeI fragment containing the entire AtKCO1 gene was subcloned. Sequence determination was done for all clones as described above. The complete AtKCO1 gene sequence was deposited in GenBank/EMBL/DDBJ databases under accession No. Y07825.

Degenerate primers corresponding to highly conserved P-domains of AtKCO1 (PKC1: GGNTAYGGNGAYTTRGTNCC; PKC7: AARCTYTTRTCNCCRTANCCYAA) were used to amplify partial cDNAs from a Solanum tuberosum L. cv. Desirée flower cDNA library (M. Klein and B.M.-R., unpublished results). cDNA fragments representing two different potato genes, designated StKCO1α and StKCO1β, were recovered. The cDNA fragments were subsequently used to isolate the corresponding genomic fragments from an S. tuberosum cv. Desirée genomic library (kindly provided by M. Ebneth, Sungene, Gatersleben, Germany). Fragments of StKCO1α and STKCO1β genes were subcloned into Bluescript and sequenced. Sequences were deposited in GenBank/EMBL/DDBJ databases under accession numbers Y13048and AJ308597, respectively.

Plasmids and constructs

Standard PCR reactions and subcloning procedures were used to engineer AtKCO1 sequences into clones described below. AtKCO1–GUS fusion constructs (KCO1–GUS1 and KCO1–GUS2) were generated using plant expression vector pGPTV-HPT (Becker et al., 1992). Two regions (2.5 and 0.87 kb, respectively), both containing the promoter, the mRNA leader exon, the 5′ leader intron, and portions of the first translated exon of the AtKCO1 gene, were transcriptionally fused to the GUS gene using the XbaI site of the vector (Figure 1b). In addition, two deletion constructs (KCO1–GUS3 and KCO1–GUS4), excluding the 5′ leader intron, were generated (details on cloning procedures are available on request). All recombinant clones were checked by sequence analysis. Plasmid p35S-KCO1 was generated by inserting the AtKCO1 cDNA (accession no. X97323) into SmaI/SalI sites of pBinAR-Kan (Höfgen and Willmitzer, 1988). A C-terminal GFP fusion construct was created using pEGFP-N1(Clontech, Heidelberg, Germany). The AtKCO1 cDNA was fused to EGFP coding sequence using the BamHI site of pEGFP-N1 to generate pKCO1-GFP. The DNA fragment coding for the fusion protein, which possesses a six-amino-acid linker between the KCO1and EGFP sequence, was further subcloned into pBinAR-Kan to generate P35S-KCO1-GFP.

Analysis of GUS expression in transgenic plants

T 1 or T2 seedlings or plants of transgenic A. thaliana lines (10–40 independent lines per construct) were initially used for histochemical analysis. Staining for GUS activity was done as described by incubation in staining solution at 37°C for 2–24 h (Plesch et al., 2000).

Membrane purification

Proteins were quantified according to the method of Bradford (1976), using bovine serum albumin as a standard. Microsomal membranes were prepared according to Mathieu et al. (1996). After washing with 20 mm KCl, 5 mm EDTA pH 5.5, BY2 cells were ground under liquid nitrogen to a fine powder and further homogenized in extraction buffer (50 mm Tris–Mes pH 8.0, 0.5 m sucrose, 10% glycerol, 1 mm MgCl2, 10 mm EDTA, 10 mm EGTA, 10 mm ascorbic acid, 0.5 µg ml−1 leupeptin, 5 mm DTT). Homogenates were centrifuged at 10 000 g for 10 min to remove cell debris and intact organelles. Supernatants were filtered through a 207 µm cheesecloth and spun at 125 000 g for 1 h to pellet microsomal membranes. Membrane pellets were resuspended in a suitable volume of resuspension buffer (330 mm sucrose, 2 mm DTT, 5 mm K (P) pH 7.8, 0.5 µg ml−1 leupeptin) and stored at −80°C until use.

A plasma membrane-enriched fraction was purified from microsomes by aqueous two-phase partitioning using a 6.6% PEG/Dextran, 5 mm KCl system as described by Ephritikhine et al. (1993). Microsomal proteins (500 µg) were applied to step sucrose gradients (generally 16–55%). Gradients were buffered in resuspension buffer. The step sucrose gradients were prepared in 12 ml Beckman ultraclear thin tubes, using stock solutions containing 55, 40, 33.5, 26.5 and 16% (w/v) sucrose in the above-mentioned buffer, and were prepared by sequential layering of the following stocks: 0.75 ml 55% sucrose solution; three 0.97 ml aliquots 40% sucrose solution; three 0.77 ml aliquots 33.5% sucrose solution; two 1 ml aliquots 26.5% sucrose solution; and two 0.75 ml aliquots 16% sucrose solution. 500 µg microsomal proteins were layered on top of the sucrose gradient. Gradients were centrifuged for 16 h (4°C) at 150 000 g (SW 40Ti rotor; Beckman, Germany). Fractions of 1 ml were collected from the top and the sucrose concentration was determined by measurement of the refractive index using a refractometer (A. Krüss Optronic, Hamburg, Germany). Aliquots of fractions were used for detection of GFP fluorescence at 488 nm excitation and 508 nm emission using a SFM25 spectrofluorometer (Kontron Instruments, Munich, Germany). Membrane fractions were diluted with resuspension buffer and collected at 100 000 g for 1 h. Microsomal proteins of each fraction were resuspended in 40–50 µl resuspension buffer.

SDS–PAGE and immunoblotting

Protein molecular weight standards used as markers were purchased from Pharmacia (Freiburg, Germany). Proteins (5–20 µg per lane) were separated via 12% Laemmli SDS–PAGE, and transferred onto a nylon membrane (PVDF membrane, Millipore, Bedford, MA, USA), USA). Blots were blocked with TBST (250 mm NaCl, 0.1% Tween 20, 50 mm Tris pH 7.6) containing 5% fat-free milk powder, incubated overnight at 4°C in TBST with the appropriate dilutions of antiserum or purified antibodies, washed, and reacted with secondary antibody conjugated to alkaline phosphatase (Promega, Mannheim, Germany), or alternatively conjugated to horseradish peroxidase (Pierce, Rockford, IL, USA). Immune complexes were detected by colour assay using nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (Sigma, St. Louis, MO, USA) or luminol (SuperSignal West Dura substrate, Pierce) as substrate.

Antibodies

An anti-KCO1 antiserum was raised against a synthetic peptide (NH2-NDLEAADLDEDGVVG-COOH) located at the C-terminus of AtKCO1 and provided by FZB Biotechnik (Berlin, Germany). Affinity-purified rabbit polyclonal antiserum was used in all experiments. Anti-GFP antibodies were purchased from Clontech. Polyclonal antibodies raised against tobacco Nt-TIPa (tonoplast) have been described previously (Gerbeau et al., 1999). Rabbit antiserum anti-E37 was kindly provided by J. Joyard (CEA, Grenoble, France). Polyclonal antibodies directed against tobacco plasma membrane H+-ATPase and the E subunit of barley vacuolar H+-ATPase were gifts from M. Boutry (University of Louvain-la-Neuve, Belgium) and K.-J. Dietz (University of Bielefeld, Germany), respectively. Antisera were used with the following dilutions: anti-AtKCO1 (1 : 1000); anti-GFP (1 : 1000); anti-Nt-TIPa (1 : 5000); anti-pmATPase (1 : 3000); anti-E37 (1 : 1000); anti-vATPase (1 : 2000).

Isolation of vacuoles from tobacco protoplasts

Protoplasts were prepared from exponentially growing BY2 cells as described by Negrutiu et al. (1987). Protoplasts were stored in W5 medium on ice for 1 h and subsequently used for isolation of vacuoles essentially as described by Sansebastiano et al. (1998). Protoplasts of 10 ml culture volume were pelleted and resuspended in 5 ml prewarmed (42°C) lysis buffer (0.2 m mannitol, 10% Ficoll 400, 20 mm EDTA, 2 mm DTT, 5 mm HEPES pH 8.0, 150 µg ml−1 bovine serum albumin). Release of vacuoles was observed using a microscope.

Microscopy

Cells were settled onto clean glass slides covered with MS medium under glass cover slips. Images were obtained by conventional microscopy using an AX-70 microscope and HQ480/40 filter for excitation and HQ535/50 filter for emission (Olympus, Hamburg, Germany) to visualize GFP, or confocal laser scanning microscopy (Leica DM IRBE Inverse microscope with Leica TCS SP Laser scanning unit) using a 40× Plan-Neofluar oil objective lens (NA 1.3; Zeiss, Jena, Germany). Images (Figure 4) were arranged using Adobe photoshop (Adobe Systems, Mountain View, CA).

Note added in proof

The functional analysis of KC01 in Arabidopsis has been published during the review process of our manuscript [Schönknecht, G., Spoormaker, P., Steinmeyer, R., Brüggemann, L., Ache, P., Dutta, R., Reintanz, B., Godde, M., Hedrich R. and Palme, K. (2001) KC01 is a component of the slow-vascuolar (SV) ion channel. FEBS Letters511, 28--32].

Acknowledgements

We thank Kerstin Zander, Antje Schneider and Eike Kamann for expert technical assistance during biochemical analyses. We are grateful to M. Boutry, K.-J. Dietz and J. Joyard for providing antisera against membrane proteins, and U. Uwer and M. Ebneth for providing A. thaliana and potato genomic libraries, respectively. B. Mueller-Roeber thanks the Max-Planck-Society for previous support within the framework of a Junior Research Group. This work was supported by a grant of the Deutsche Forschungsgemeinschaft (MU1199/2-2) to B. Mueller-Roeber.

GenBank accession numbers: AtKCO1 (Y07825), StKCO1α (Y13048), StKCO1β (AJ308597).

Ancillary