The Arabidopsis vacuolar malate channel is a member of the ALMT family


  • Peter Kovermann,

    1. Institute for Plant Biology, University of Zürich, CH-8008 Zürich, Switzerland
    Search for more papers by this author
    • These authors contributed equally to this study.

    • Present address: Institute of Neurophysiology, Hannover Medical School, D-30625 Hannover, Germany.

  • Stefan Meyer,

    Corresponding author
    1. Institute for Plant Biology, University of Zürich, CH-8008 Zürich, Switzerland
      (fax +41 4463 48286; e-mail
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Stefan Hörtensteiner,

    1. Institute for Plant Biology, University of Zürich, CH-8008 Zürich, Switzerland
    Search for more papers by this author
  • Cristiana Picco,

    1. Institute of Biophysics, National Research Council, I-16149 Genoa, Italy
    Search for more papers by this author
  • Joachim Scholz-Starke,

    1. Institute of Biophysics, National Research Council, I-16149 Genoa, Italy
    Search for more papers by this author
  • Silvia Ravera,

    1. Institute of Physiology, University of Zürich, CH-8057 Zürich, Switzerland
    Search for more papers by this author
  • Youngsook Lee,

    1. POSTECH-UZH Cooperative Laboratory, Division of Molecular Life Sciences, Pohang University of Science and Technology, Pohang, 790-784, Korea
    Search for more papers by this author
  • Enrico Martinoia

    1. Institute for Plant Biology, University of Zürich, CH-8008 Zürich, Switzerland
    2. POSTECH-UZH Cooperative Laboratory, Division of Molecular Life Sciences, Pohang University of Science and Technology, Pohang, 790-784, Korea
    Search for more papers by this author

(fax +41 4463 48286; e-mail


In plants, malate is a central metabolite and fulfills a large number of functions. Vacuolar malate may reach very high concentrations and fluctuate rapidly, whereas cytosolic malate is kept at a constant level allowing optimal metabolism. Recently, a vacuolar malate transporter (Arabidopsis thaliana tonoplast dicarboxylate transporter, AttDT) was identified that did not correspond to the well-characterized vacuolar malate channel. We therefore hypothesized that a member of the aluminum-activated malate transporter (ALMT) gene family could code for a vacuolar malate channel. Using GFP fusion constructs, we could show that AtALMT9 (A. thaliana ALMT9) is targeted to the vacuole. Promoter-GUS fusion constructs demonstrated that this gene is expressed in all organs, but is cell-type specific as GUS activity in leaves was detected nearly exclusively in mesophyll cells. Patch-clamp analysis of an Atalmt9 T-DNA insertion mutant exhibited strongly reduced vacuolar malate channel activity. In order to functionally characterize AtALMT9 as a malate channel, we heterologously expressed this gene in tobacco and in oocytes. Overexpression of AtALMT9-GFP in Nicotiana benthamiana leaves strongly enhanced the malate current densities across the mesophyll tonoplasts. Functional expression of AtALMT9 in Xenopus oocytes induced anion currents, which were clearly distinguishable from endogenous oocyte currents. Our results demonstrate that AtALMT9 is a vacuolar malate channel. Deletion mutants for AtALMT9 exhibit only slightly reduced malate content in mesophyll protoplasts and no visible phenotype, indicating that AttDT and the residual malate channel activity are sufficient to sustain the transport activity necessary to regulate the cytosolic malate homeostasis.


Malate is implicated in a large number of metabolic pathways in all living organisms. In plants it plays a central role in a multitude of functions. As a metabolite of the Krebs cycle it is involved in the production of ATP. In the glyoxylate cycle malate is closely linked to the β-oxidation of fatty acids and the production of NADH. Malate also serves as a temporary carbon store and provides reduction equivalents in C4 and Crassulacean acid metabolism (CAM) plants. It must therefore be present in the cytosol, chloroplasts, mitochondria, peroxisomes, glyoxysomes and vacuole. Furthermore, malate plays a role in pH regulation, is an important osmoticum and acts as a major anion compensating the positive charges of potassium and sodium. A metabolite that is implicated in such a complex network has to be tightly controlled. Using non-aqueous fractionation and in vivo NMR it has been shown that cytosolic malate concentrations are kept very constant, whereas vacuolar malate contents fluctuate diurnally or in response to environmental changes (Gerhardt et al., 1987; Gout et al., 1993). In roots, malate is excreted into the apoplast and may complex and detoxify aluminum or release rock-bound phosphate (Neumann and Martinoia, 2002; Ryan et al., 2001). Transport processes for malate have been investigated extensively, which reflects the interest in this metabolite. Several chloroplast and mitochondrial carboxylate transporters have been identified, with most of them exchanging malate with another carboxylate (Menzlaff and Flügge, 1993; Palmieri et al., 1993). Vacuolar malate transport has also been investigated in detail using flux analysis, membrane potential- and pH-dependent fluorescence probes, and the patch-clamp technique (Cerana et al., 1995; Hafke et al., 2003; Martinoia et al., 1985; Pantoja and Smith, 2002; Pei et al., 1996; Ratajczak et al., 1994). According to these studies, malate uptake is driven by the electrochemical potential difference between the cytosol and the vacuole generated by the vacuolar proton pumps (Maeshima, 2001; Rea and Sanders, 1987). Observations of malate currents across the tonoplast indicate that they are strongly inward-rectifying, thus favoring the movement of malate from the cytosol into the vacuole (Cerana et al., 1995; Epimashko et al., 2004; Hafke et al., 2003; Hurth et al., 2005; Pantoja and Smith, 2002). It was shown that macroscopic currents observed on Kalanchoë daigremontiana vacuoles can be attributed to the activity of a small 3-pS malate-selective channel, and that channel density and open probability suffice for the required nocturnal malate transport (Hafke et al., 2003). Recently, a vacuolar malate transporter from Arabidopsis has been identified at the molecular level (Arabidopsis thaliana tonoplast dicarboxylate transporter, AttDT, At5g47560; Emmerlich et al., 2003). This carrier is an ortholog of the renal Na+/dicarboxylate transporter present in the proximal tubulus of mammalian kidney. Homozygous T-DNA insertional knock-out mutants lacking a functional AttDT did not show an obvious phenotype, but contained less malate in leaves. Leaf malate contents were reduced to 25–50% of the wild-type (WT) contents in Attdt deletion mutants, whereas the residual vacuolar malate transport activity in the mutants was reduced to about 30% of that observed for vacuoles isolated from WT plants. Furthermore, the respiratory coefficient was increased in the deletion mutants, indicating a higher consumption of carboxylates in the absence of the malate transporter. Surprisingly, further investigations using the patch-clamp method revealed that Attdt mutants still exhibited the well-described vacuolar malate channel, as well as citrate transport activity (Hurth et al., 2005). Hence, vacuolar malate transport is catalyzed by a transporter and at least one channel. However, the molecular nature of the vacuolar malate channel remained to be elucidated. Former vacuolar proteomic studies identified a large number of putative vacuolar membrane proteins (Carter et al., 2004; Endler et al., 2006; Jaquinod et al., 2007); however, no putative candidate emerged from these data. Instead, malate channels localized in the plasma membrane have recently been described in wheat (TaALMT1, Triticum aestivum aluminum-activated malate transporter 1; Sasaki et al., 2004), Arabidopsis (AtALMT1; Hoekenga et al., 2006) and rape (BnALMT1 and BnALMT2, Brassica napus AMLT1 and AMLT2; Ligaba et al., 2006). These malate channels confer aluminum tolerance by extruding malate from root epidermal cells into the surrounding soil. In Arabidopsis these AtALMTs form a small protein family of 14 members (Hoekenga et al., 2006). Recent data have clearly demonstrated that gene products of the same protein family are often targeted to different membranes (Becker et al., 2004; Chen, 2005; Czempinski et al., 2002; Endler et al., 2006). This led us to hypothesize that one or several members of the AtALMT protein family could be targeted to the tonoplast, where they function as malate channels.

We describe in this work that an AtALMT9-GFP fusion protein localizes to the tonoplast of Arabidopsis and onion epidermal cells, as well as to the tonoplast of Vicia faba guard cells. Patch-clamp experiments using mesophyll vacuoles isolated from Atalmt9 deletion mutants revealed that the current density was attenuated by approximately 70%, whereas the malate concentrations in the protoplasts were slightly decreased. Furthermore, control patch-clamp measurements on vacuoles derived from AtALTM9-GFP overexpressing tobacco cells showed strongly enhanced malate channel activities. In addition, the functional expression of AtALMT9 in Xenopus oocytes further confirmed the identity of AtALMT9 as a bona fide malate channel. With the exception of a strongly reduced malate channel activity and a slightly reduced vacuolar malate concentration, we could not observe any obvious phenotype for Atalmt9, suggesting possible functional redundancy of the vacuolar malate transporter AttDT and vacuolar ALMTs in Arabidopsis.

Results and discussion

Arabidopsis ALMTs are hydrophobic proteins with slight differential topologies

A dendrogram based on an amino acid sequence alignment (Chenna et al., 2003) showed high similarities at the amino acid level within the AtALMT protein family and TaALMT1 (65.8 ± 1.3% average pair distance, Figure S1). Furthermore, the gene structure is largely conserved within all members with respect to the number of introns (ARAMEMNON plant membrane protein database,; Schwacke et al., 2003). The dendrogram based on amino acid sequence similarity indicated that the AtALMT family is grouped into three distinct clades (Figure 1a). TaALMT1 (AB081803) belongs to clade 1, which includes AtALMT1, 2, 7, 8 and 10 (At1g08430, At1g08440, At2g27240, At3g11680 and At4g00910), whereas clade 2 includes AtALMT3, 4, 5, 6 and 9 (At1g18420, At1g25480, At1g68600, At2g17470 and At3g18440). The protein family members AtALMT11, 12, 13 and 14 (At4g17585, At4g17970, At5g46600, At5g46610) belong to clade 3. A comparison of the hydrophobicity profiles of the well-described AtALMT1 and TaALMT1 with members of clade 2, e.g. AtALMT5 and AtALMT9, indicated that all these proteins are hydrophobic with two transmembrane domains: one of which is localized to the N-terminus, consisting of between five and seven transmembrane α-helices, and the other is localized to the C-terminus, spanning the tonoplast only once (Figure 1b). This prediction was achieved by consulting consensus scores, which represent percentage accordances between different algorithms for the elucidation of transmembrane topologies, which are implemented in the ARAMEMNON database (Schwacke et al., 2003), whereas the consensus scoring strongly enhances the reliability of topology predictions (Nilsson et al., 2000). A graphical overlay of an alignment based on the amino acid sequences of TaALMT1, AtALMT1, AtALMT5 and AtALMT9 by putative transmembrane regions showed that the protein members of clades 1 and 2 differ in the C-terminal transmembrane topology, with respect to number and position (see Figure S1); whereby a possible role of this difference in function and localization can only be assumed.

Figure 1.

 Dendogram and hydrophobicity analysis of the aluminum-activated malate transporter (ALMT) protein family.
(a) The dendrogram of 15 members of the ALMT protein family and the Arabidopsis thaliana tonoplast dicarboxylate transporter (AttDT), based on an amino acid sequence alignment with ClustalW (Chenna et al., 2003), shows that AtALMTs cluster in three main groups, whereas the plasma membrane localized proteins TaALMT1 and AtALMT1 both share the same clade. Branch lengths are proportional to the level of interfered evolutionary change, and are given in relative units.
(b) Hydrophobicity analysis of AtALMT5, AtALMT9, AtALMT1 and TaALMT1 shows slight differences in transmembrane topology. Hydrophobicity values were calculated according to the Kyte and Doolittle method (window size, 11 amino acids; Kyte and Doolittle, 1982) and were plotted against the amino acid position. Positive values indicate hydrophobic regions. Putative transmembrane segments are indicated by horizontal bars that are colored according to the consensus scores, which represent a scale for the accordance between different algorithms for transmembrane structure prediction provided by the ARAMEMNON plant membrane protein database (Schwacke et al., 2003).

Hoekenga et al. (2006) previously characterized the clade-1 member AtALMT1 as an aluminum-activated malate channel localized in the plasma membrane of Arabidopsis roots. Because of the high sequence homology within the ALMT gene family, it seemed possible that all protein members fulfil analogous functions as malate channels within the plant cell. Moreover, taking the fact that gene products of the same protein family may be targeted to different membranes, in this study we investigated AtALMT5 and AtALMT9 as members of clade 2, which is slightly more similar to the previously characterized tonoplast transporter AttDT, compared with the other two clades (Figure 1a).

AtALMT9-GFP localizes in the tonoplast

In order to verify our hypothesis that members of the AtALMT family are targeted to membranes of different organelles, we undertook subcellular localization experiments with two members of the second clade: AtALMT5 and AtAMLT9 (Figure 1a). GFP was fused in frame to the C-terminal end of AtALMT5 and AtALMT9. The transient expression of these constructs in Arabidopsis and onion epidermal cells by particle bombardment demonstrated that AtALMT9 was targeted to the tonoplast in both cases (Figure 2a–e), whereas fluorescence for AtALMT5-GFP was observed exclusively in the endoplasmic reticulum (ER; Figure S2). To preclude the possibility that the fusion proteins were mistargeted because of the lack of chloroplasts in epidermal cells, we conducted the same experiments in chloroplast-containing guard cells of V. faba. The GFP fluorescence pattern of these cells confirmed our observations that AtALMT9 was targeted to the tonoplast (Figure 2f,g), and that AtALMT5 was targeted to the ER (Figure S2). In contrast to epidermal cells, guard cells contain a far more complex vacuolar membrane system, consisting of a large number of invaginations, playing an important role in rapid changes of vacuolar volume during stomatal movement (Gao et al., 2005). Because of this structure, it was easier to visualize the tonoplastic localization of AtALMT9-GFP by a fluorescence signal from the chloroplasts, which are located in the cytoplasm. Taken together these observations clearly demonstrate that in contrast to AtALMT1, AtALMT9 is a tonoplast protein. It is tempting to speculate that the difference in the subcellular localization between these two proteins relies on differences in transmembrane topology within the C-terminal region, as described above. The fact that AtALMT5, also a member of the second clade (Figure 1a), is not localized in the vacuolar membrane may result from slight differences in the predicted structure between AtALMT5 and AtALMT9 (Figure 1b, S1).

Figure 2.

 Subcellular localization of AtALMT9 by transient expression of the AtALMT9-GFP fusion construct.
In Arabidopsis epidermal cells (a,b) as well as in onion epidermal cells (c–e) fluorescence of the AtALMT9-GFP fusion protein is targeted to the tonoplast, which separates the vacuolar lumen from the cytoplasm containing the nucleus (marked by an arrow). The transmission picture (c) is the same onion epidermal cell as in the fluorescent image (d). GFP clearly surrounds the nucleus of the Arabidopsis and onion cells (b and e). Cell walls and nuclei of Arabidopsis epidermal cells were stained in red with propidium iodide, whereas the nucleus in the onion epidermal cell is visible by an overlay of (c) and (d). In chloroplast-containing guard cells of Vicia faba the fluorescence of the AtALMT9-GFP fusion protein (f, g) is also visible in the tonoplast (chloroplasts are shown by red fluorescence in the transmission picture, f). The GFP fluorescence surrounds the chloroplast (marked by an arrow) located in the cytoplasm between the vacuole and the plasma membrane of the cell. Scale bars: (a) 25 μm; (c) 20 μm; (f) 10 μm.

Analysis of AtALMT9 expression by AtALMT9 promoter:GUS plants

Detailed tissue expression pattern analysis is a prerequisite to understanding the function of a given gene product. We thus investigated the tissue specificity of AtALMT9 using plants transformed with a β-glucuronidase gene (GUS) under the control of a 1785 bp promoter region upstream of AtALMT9. Analysis of these transgenic plants revealed GUS activity in the hypocotyl of young seedlings (Figure 3a), and strong activity in the leaves of young and older plants (Figure 3b,c). GUS activity was also detected in the roots of younger plants (Figure 3b). This activity was higher in the later developmental stages (not shown). In the flower tissue, GUS staining was found in both the sepals and the stamina (Figure S3). To investigate the expression profile in leaves in more detail, leaves were embedded and cross-sectioned (Figure 3d). Microscopical analysis of these sections indicated that GUS activity was concentrated in the mesophyll tissue of leaves, whereas only weak activity was visible in the upper and lower epidermal cell layers. Our observations concur with gene expression data from microarray experiments (average of at least 231 chip experiments) available in the Genevestigator A. thaliana microarray database (http://, which indicated that AtALMT9 mRNA accumulated similarly in all tissues of the plant, with the highest levels in flowers, roots and leaves. Summarizing our localization data, we clearly demonstrate that AtALMT9 is a vacuolar protein expressed in nearly all organs of the plant, with a cell-type specificity in leaves, as GUS staining was observed nearly exclusively in mesophyll cell layers. Therefore, we decided to focus our interest on the vacuole of the leaf mesophyll cells for further functional studies.

Figure 3.

 Analysis of AtALMT9 expression by AtALMT9 promoter:GUS reporter plants.
(a) GUS-histochemical staining of young Arabidopsis seedlings (4 days after germination) showing GUS activity in the hypocotyls. GUS-histochemical staining of young (b, 8 days after germination) and older (c, 20 days after germination) Arabidopsis plants, showing GUS expression in the leaves and the stem, and also along the root. (d) Cross-section of a rosette leaf showing GUS expression mainly in the mesophyll cells, and weak expression in the upper and lower epidermal cell layers. Scale bars: (a) 1 mm; (b, c) 0.1 cm; (d) 100 μm.

Malate currents are decreased in Atalmt9 knock-out plants

To reveal the physiological role of the tonoplast-localized AtALMT9 protein, we obtained a SALK T-DNA insertion mutant line for AtALMT9 from the Nottingham Arabidopsis stock center (NASC,; N590362; Scholl et al., 2000), carrying a T-DNA insertion in the first exon of the AtALMT9 gene (SALK_090362, Figure 4a). Homozygous knock-out plants were identified by PCR using appropriate primers (not shown). The absence of the AtALMT9 transcript in these mutant plants was demonstrated by RT-PCR analysis with primers amplifying the entire coding region (1797 bp, Figure 4b). This homozygous line was used to perform electrophysiological analyses, and to investigate whether differences in malate channel activity could be observed compared with vacuoles isolated from WT plants.

Figure 4.

 Analysis of an AtALMT9 T-DNA insertion mutant plant.
(a) The genomic structure of AtALMT9, and the position and orientation of the T-DNA insertion in the AtALMT9 gene of the mutant line SALK_590362. The gene has six exons (black arrows) and the insertion is located at position +19 bp downstream of the start codon. The orientation of the left border (LB) is indicated. (b) Semiquantitative RT-PCR analysis of the abundance of the AtALMT9 transcript demonstrated the lack of AtALMT9 transcript in the homozygous knock-out line.

The presence of a vacuolar malate import channel in the mesophyll of WT plants was already shown by electrophysiological investigations on CAM and C3 plants. Thus far the available data indicate that malate currents across the tonoplast of different plant species share common characteristics, such as inward rectification, selectivity for anions over cations and activation at high membrane potentials (Epimashko et al., 2004; Pantoja and Smith, 2002).

In our study we quantified the malate channel activity at the mesophyll tonoplast with the patch-clamp technique in the whole-vacuole mode. We clamped Arabidopsis WT vacuoles to test voltages of 3-s duration between +100 and −140 mV. The test voltages follow the sign convention proposed by Bertl et al. (1992). To ensure that currents observed at pH 7.5 (bath) were principally caused by movements of the malate2− ions, we buffered malic acid with an impermeable cation according to the method described by Hafke et al. (2003) and Hurth et al. (2005). Under asymmetric ionic conditions (100 mm malate2− out/10 mm malate2− in), large inward currents were observed at negative test voltages. These currents consisted of an instantaneous element (Iinst) superimposed at negative voltages by a slow time-dependent element (Istd, Figure 5a). The current–voltage (IV) relationships for Iinst (data not shown) and Istd (Figure 5c, bsl00001) showed that both elements were inward-rectifying, which is in agreement with the previously described properties of tonoplastic malate currents (Epimashko et al., 2004; Hafke et al., 2003; Pantoja and Smith, 2002). At a test voltage of −140 mV the vacuoles exhibited a mean current density (Istd) of 13.6 ± 1.9 pA pF−1 (n = 5, Figure 5c, bsl00001). This fits well with previously observed values obtained under the same ionic conditions with Arabidopsis WT mesophyll vacuoles (Hurth et al., 2005). In order to further characterize these currents, we performed a tail current analysis under asymmetric ionic conditions (100 mm malate2− out/10 mm malate2− in, data not shown). A reversal potential of +12 mV could be calculated. This reversal potential is closer to the theoretical potential for malate2− (+23 mV) than to the Nernst potentials for other ions in the solution (EBPTH+ = −63 mV, ECl- = 0 mV, ECa2+ = 0 mV and EMg2+ =0 mV; Hafke et al., 2003), which strongly suggests that Istd originates from inward-rectifying malate channels. Similar observations have already been described for Arabidopsis and other plants (Cerana et al., 1995; Hafke et al., 2003; Hurth et al., 2005; Martinoia et al., 1985; Pantoja and Smith, 2002; Pei et al., 1996).

Figure 5.

 Patch-clamp analysis of vacuoles obtained from Arabidopsis Atalmt9 knock-out plants compared with wild-type (WT) vacuoles.
(a) Whole-vacuole malate current density traces of an Arabidopsis WT vacuole, obtained by dividing the raw currents by the tonoplast capacitance under asymmetric ionic conditions (100 mm malate2− out/10 mm malate2− in). (b) Whole-vacuole current density traces of a vacuole isolated from Atalmt9 knock-out plants under the same ionic conditions as in (a) showed strongly diminished malate current densities. (c) Corresponding current density plots derived from whole-vacuole malate currents across the tonoplast from WT plants (bsl00001, n = 5), and from Atalmt9 knock-out plants from the T3 (bsl00066, n = 7) and the T4 (•, n = 6) progenies.

Compared with WT vacuoles, vacuoles isolated from Atalmt9 deletion mutants exhibited strongly diminished inward currents compared with WT vacuoles (Figure 5b,c). At a test voltage of −140 mV we observed a strong current reduction compared with WT plants (T3 progeny, Istd = 3.13 ± 0.96 pA pF−1, n = 7; T4 progeny, Istd = 4.1 ± 0.89 pA pF−1, n = 5; Figure 5c). Despite this difference, the mutant plants grown under normal conditions were phenotypically indistinguishable. Determination of malate content in isolated mesophyll protoplasts showed that deletion mutants contained 19.5 ± 7.4% and 20.0 ± 6.0% (±SE) less malate, compared with the WT obtained from the seed batch of the SALK mutant line and the normal Col-0 WT plants, respectively.

In Attdt deletion mutants, leaf malate content was reduced by 50–75%, and vacuoles isolated from Attdt mutants still contained 30% of the cellular malate in their vacuoles, indicating that under normal conditions AttDT is the major malate importer (Hurth et al., 2005). The high capacity of AtALMT9 at high potential differences indicates that this channel is important under specific conditions when the vacuole is hyperpolarized. However, it also has to be kept in mind that Attdt deletion mutants exhibited only a slight phenotype, which could be observed under certain stress conditions. In addition, our results support the presence of an additional vacuolar malate channel, which together with AttDT can ensure the import of malate in Atalmt9 deletion mutants. In summary, the observations presented above suggest that AtALMT9 is either a vacuolar malate channel or is at least involved in a regulatory function on the channel.

Overexpression of AtALMT9-GFP in tobacco leaves enhances malate currents across the tonoplast

To obtain further proof for the channel activity of AtALMT9 we transiently overexpressed the AtALMT9-GFP fusion construct in leaves of Nicotiana benthamiana under the control of the CaMV 35S promoter. Vacuoles isolated from these leaves revealed clearly detectable fluorescence in the tonoplast, thereby confirming the results of our transient localization studies (Figure 6c). Patch-clamp measurements on these fluorescent vacuoles and tobacco WT vacuoles demonstrated that the overexpression of AtALMT9-GFP enhanced malate current density across the tobacco mesophyll tonoplast. Whereas the AtALMT9-GFP-mediated malate currents exhibited the well-described characteristics, consisting of an instantaneous and a time-dependent component, background currents in WT tobacco vacuoles lacked a pronounced time dependence (Figure 6a,b). The comparison of malate currents from WT vacuoles and AtALMT9-overexpressing vacuoles was achieved by the determination of total current amplitudes. Overall, a 7.4-fold increase of total malate current density was observed in vacuoles derived from overexpressing plants at −120 mV (Figure 6d; WT, 14.2 ± 1.5 pA pF−1, n = 5; AtALMT9-GFP, 104.9 ± 23.9 pA pF−1, n = 8). The relatively large variability results from different levels of transgene expression. Strongly fluorescent vacuoles exhibited a high malate current density, whereas the malate current densities in non-fluorescent vacuoles from the same sample behaved as WT vacuoles (Figure 6d). Furthermore, to exclude the possibility that the high expression of a vacuolar membrane protein fused to GFP could unspecifically induce malate currents, we also performed control experiments by overexpressing the vacuolar sucrose transporter construct SUT4-GFP (Endler et al., 2006). No increase in malate currents could be detected (data not shown).

Figure 6.

 Patch-clamp analysis of AtALMT9-GFP heterologously expressed in Nicotiana benthamiana.
(a) Whole-vacuole malate current density traces of a tobacco wild-type (WT) vacuole obtained under the same conditions as in Figure 5a). (b) Whole-vacuole malate current density traces of a strongly fluorescent vacuole isolated from AtALMT9-GFP-overexpressing tobacco (same conditions as in Figure 5a) exhibited largely increased malate current densities. (c) Transmission and fluorescence picture of a vacuole released from a tobacco mesophyll protoplast overexpressing AtALMT9-GFP. (d) Bar graphs representing current densities at −120 mV obtained by the patch-clamping of WT tobacco vacuoles and vacuoles derived from tobacco plants overexpressing (OE)AtALMT9-GFP under asymmetric ionic conditions (100 mm anions out/10 mm anions in). Scale bar: (c) 15 μm.

To obtain more information about the substrate specificity, we also measured fumarate- and chloride-mediated total currents in vacuoles isolated from the AtALMT9-GFP-expressing tobacco plants (Figure 6d). The increase in fumarate current densities at 120 mV was 3.7-fold (WT, 22.0 ± 1.6 pA pF−1, n = 5; AtALMT9-GFP, 81.8 ± 12.3 pA pF−1, = 4) compared with WT vacuoles. Former studies on K. daigremontiana have shown that mesophyll vacuoles exhibit higher fumarate current densities than malate current densities (Hafke et al., 2003). This is apparently also true for N. benthamiana WT vacuoles (Figure 6d). In vacuoles from AtALMT9-GFP-expressing tobacco plants the increases in fumarate current densities as well as the absolute values are lower compared with those for malate, which indicates that AtALMT9 has a higher selectivity for malate than for fumarate. However, because of the high variability of the fluorescence of the vacuoles, reflecting the expression level of ALMT9, an exact permeability ratio for malate and fumarate can not be deduced. Lower current densities compared with the dicarboxylates could be detected for chloride in tobacco WT vacuoles. The AtALMT9-GFP-mediated increase in chloride currents was by a factor of 2.2 at 120 mV (WT, 10.6 ± 2.2 pA pF−1, n = 6; AtALMT9-GFP, 23.5 ± 1.5 pA pF−1, n = 4). This result indicates that AtALMT9 also exhibits a weak chloride conductance. The AtALMT9-GFP mediated currents can also be calculated by substracting the WT current densities, and further underlines the results presented above, even assuming that two charges are transferred in the case of the dicarboxylates and only one charge is transferred for chloride (malate, 90.7 pA pF−1; fumarate, 59.8 pA pF−1; chloride, 12.9 pA pF−1).

AtALMT9-expressing oocytes show malate currents

The expression of AtALMT9 in Xenopus oocytes served to confirm that the protein indeed has channel-forming capability. Plasma membrane localized ALMT proteins from wheat, Arabidopsis and rape mediated malate fluxes when the respective genes were heterologously expressed in Xenopus oocytes (Hoekenga et al., 2006; Ligaba et al., 2006; Sasaki et al., 2004). Furthermore, it has been shown that plant tonoplast proteins can be targeted to the plasma membrane of oocytes (Maurel et al., 1993; Desbrosses-Fonrouge et al., 2005).

When oocytes injected with AtALMT9 cRNA were challenged with a series of voltage steps from +20 to −160 mV, instantaneously activating currents were recorded (Figure 7a). In contrast to the observations in the tonoplast system (Figures 5 and 6), no currents were activated in a time-dependent manner in response to 3-s voltage pulses (data not shown). The membrane environment or missing post-translational modifications in the heterologous oocyte system (Stühmer and Parekh, 1995) may change the characteristics of the AtALMT9 protein. Current amplitudes increased when the external malate concentration was changed from 0 to 10 mm (Figure 7a,b). This behavior is surprising for the range of negative test potentials (as it implies that more malate moves out the cell in the presence of an oppositely directed gradient), but could indicate a regulatory role of malate in channel functioning. For example, such a type of regulation has been found for some voltage-gated potassium channels (Pardo et al., 1992; Wood and Korn, 2000). Current–voltage relationships for all conditions tested are shown in Figure 7c. Compared with AtALMT9-expressing oocytes, current amplitudes in control oocytes were much smaller and did not significantly increase upon the addition of 10 mm malate. Upon stepping from 0 to 10 mm external malate, currents in cRNA-injected oocytes increased about 1.9-fold at −160 mV, and shifted towards more negative reversal potentials (ΔErev = −11 ± 2.4 mV, n = 4). This behaviour is consistent with the activation of anion-selective inward currents in AtALMT9-expressing oocytes.

Figure 7.

 Voltage-clamp experiments on Xenopus laevis oocytes expressing AtALMT9.
(a, b) Representative current traces recorded in response to voltage pulses from +20 to −160 mV in −20-mV steps. The holding and tail potential was −60 mV. AtALMT9 cRNA-injected oocytes were challenged with external solutions containing 0 mm (a) or 10 mm malate (b).
(c) Mean current–voltage relationships constructed from recordings like the ones shown in (a) and (b). Currents derive from AtALMT9 cRNA-injected oocytes (circles) and control oocytes (squares) in the presence of 0 mm malate (open symbols) or 10 mm malate (filled symbols). Values are the means of at least four experiments; error bars represent the SE. Reversal potentials for currents from cRNA-injected oocytes in 0 mm (1) and 10 mm malate (2) are indicated by arrows.

We used lanthanum (La3+) to exclude the possibility that the observed differences were caused by endogenous oocyte currents appearing at negative potentials (Tokimasa and North, 1996;Picco et al., 2007). The addition of 1 mm LaCl3 to the bath solution containing 10 mm malate did not alter instantaneous current amplitudes, but effectively blocked the typical time-dependent endogenous currents appearing at high negative potentials (data not shown). A common feature shared by ALMT proteins localized at the plasma membrane of plant roots is the slow activation by external aluminum (Al3+; Sasaki et al., 2004; Ligaba et al., 2006; Hoekenga et al., 2006). When AtALMT9-expressing oocytes were challenged with 0.1 mm Al2(SO4)3 in the bath solution without malate, negative currents at −120 mV slowly increased, before reaching a maximum value (1.5-fold ± 0.1; n = 4) within 10 min. By contrast, there was no change in current amplitude in control oocytes (0.98 ± 0.05, n = 4; data not shown).


A large number of studies have shown that malate plays an essential role in the metabolism of the plant cell, and is implicated in ion homeostasis and maintenance of cell turgor. To fulfil these functions malate must be accumulated within the large central vacuole. In this work we have identified AtALMT9, a homolog of TaALMT1 and AtALMT1, as a tonoplastic malate channel. In Arabidopsis knock-out mutants, the vacuolar malate currents were strongly reduced, and the malate concentration was slightly diminished. We confirmed that AtALMT9 was a bona fide malate channel by functional expression of the channel in N. benthamiana and in Xenopus oocytes. The fact that no obvious phenotype could be observed is very probably caused by malate transport activity of AttDT and the residual channel activity observed in knock-out plants. Furthermore, the observation that AtALMT9 is mainly expressed in the mesophyll argues that an epidermis-specific malate channel is also present. Therefore, we are presently investigating other members of the AtALMT family in Arabidopsis to determine if any others localize to the tonoplast. Functional characterization of these additional malate channels, and the generation of double and triple knock-out plants, also in combination with the Attdt deletion mutant, will allow us to elucidate in detail the role of the vacuole in cytosolic malate homeostasis. Furthermore, our study will also allow us to identify and characterize the vacuolar malate channels of CAM plants, and to investigate whether this channel is a prerequisite for CAM metabolism, as is often assumed.

Experimental procedures

Hydrophobicity analysis and comparison of primary structures

Amino acids sequence alignments were performed with the ClustalW algorithm using default parameters (; Higgins et al., 1994). Hydrophobicity profiles were calculated according to the method described by Kyte and Doolittle (1982). Structural consensus scores and the nomenclature of the AtALMT protein family was adopted from the ARAMEMNON plant membrane protein database (; Schwacke et al., 2003). Dendrograms were constructed using phylodraw (Graphics Application Laboratory,

Strains and growth conditions

Escherichia coli (DH5α; Hanahan, 1983) was used for cloning. A. thaliana (Col-0) plants were grown in controlled environment chambers or on agar medium (8-h light//16-h dark, 22°C, 55% relative humidity). N. benthamiana plants were grown in potting soil (16-h light//8-h dark, 22°C, 55% relative humidity). The transformation of Arabidopsis was performed with Agrobacterium tumefaciens (GV3101; Holsters et al., 1980).

Tissue-specific expression and subcellular localization of AtALMT9 in Arabidopsis

A 1785 bp promoter region upstream of AtALMT9 was amplified from genomic DNA of Arabidopsis (Col-0) by high-fidelity PCR using the primers At3g18440g-1847f (5′-GGGGACAAGTTTGTACAAAAAAGCAGGCT-GTTTCTCTCTGTGCCTGAGTTTG-3′) and At3g18440g-30r (5′-GGGGACCACTTTGTACAAGAAAGCTGGGTGACGGATTCTCAAAGAGAATTAAGC-3′). This region was cloned into the GATEWAY entry vector pDONR207 (Invitrogen, before recombination cloning into pMDC163 (Curtis and Grossniklaus, 2003). The vector construct was transformed into Arabidopsis using the Agrobacterium-mediated floral-dipping method (Clough and Bent, 1998). T2 progeny of hygromycin-resistant transformants were GUS-stained at various developmental stages. Embedding of GUS-stained leaves was performed in Technovit (Heraeus Kulzer GmbH, To localize AtALMT9 at the subcellular level, the AtALMT9 cDNA (1797 bp) was amplified from RIKEN clone pda08640 (cDNA clone RAFL09-66-G16; Sakurai et al., 2005; Seki et al., 1998, 2002; Yamada et al., 2003) with the primers Mc8forw (5′-GGTACCATGGCGGCGAAGCAAGGTTCCTTC-3′) and Mc8backw (5′-GGTACCCATCCCAAAACACCTACGAATCTT-3′), and then were ligated at the KpnI site into the pGFP2 vector (Haseloff and Amos, 1995) to create a constitutively expressed AtALMT9-GFP fusion protein. The resulting AtALMT9-GFP construct was transiently expressed in Arabidopsis and onion (Allium cepa) epidermal cells, as well as in guard cells of V. faba using a Helium Biolistic Particle Delivery system (Bio-Rad Laboratories,

Selection of Atalmt9 knock-out lines

Seeds of Atalmt9 knock-out lines (stock number SALK_090362; Salk Institute Genomic Analysis Laboratory, were obtained from the Nottingham Arabidopsis stock center (N590362; NASC, Genomic DNA was extracted from 4-week-old soil-grown plants, and the T-DNA insertion (+19 bp downstream of the ATG) was verified by PCR with the T-DNA-specific primer LBb1 (5′-GCGTGGACCGCTTGCTGCAACT-3′) and the primer At3g18440-TDNA-LB (5′-GTCACCGAATAAAGTGGAAAGC-3′) binding 110-bp upstream of the start ATG. Lines with homozygous T-DNA insertions were identified by genomic PCR with a set of AtALMT9-specific primers (At3g18440-TDNA-LB and At3g18440-TDNA-RB: 5′-AGGTCCACCACCACTTCATAAC-3′). The abundance of the AtALMT9 transcript in homozygous knock-out lines and in WT plants was assayed by isolation of total RNA from whole leaves followed by RT-PCR (Amersham kit, using the AtALMT9-specific primers AtALMT9forw1 (5′-GGTACCATGGCGGCGAAGCAAGGTTCCTTC-3′) and AtALMT9backw1 (5′-GGTACCCATCCCAAAACACCTACGAATCTT-3′) and the control primers Actin-forw (5′-GGAACAGTGTGACTCACACCATC-3′) and Actin-backw (5′-AAGCTGTTCTTTCCCTCTACGC-3′).

Quantification of malate concentrations

For quantification experiments, α-mannosidase and malate were determined in the protoplasts according to the method described by Hurth et al. (2005).

Expression of AtALMT9-GFP in N. benthamiana

For transient overexpression of an AtALMT9-GFP construct in tobacco leaves, the AtALMT9 cDNA (1797 bp) was cloned under the control of the CaMV 35S promoter using the pART7/pART27 cloning/expression system (Gleave, 1992). The Agrobacterium-mediated infiltration of N. benthamiana leaves was conducted as described by Yang et al. (2001), with slight modifications. After agroinfiltration, tobacco plants were grown in the greenhouse at 22°C under 16 h of light. Isolation of vacuoles and patch-clamp experiments were performed 48 h after infiltration.

Isolation of vacuoles

Protoplasts were isolated from leaf mesophyll as previously described by Song et al. (2003), with minor modifications. After incubation at 30°C (45–60 min) protoplasts were liberated by gentle agitation, and a small aliquot (20 μl) of protoplast suspension was transferred into a patch-clamp chamber filled with 200 μl of a lysis solution consisting of patch-clamp bath solution plus 8 mm EDTA. After 4 min the lysis buffer was replaced by the bath solution.

Patch-clamp measurements

Whole-vacuole malate currents were recorded using the standard patch-clamp technique according to the method described by Hamill et al. (1987), in whole-vacuole configuration using an EPC-10 amplifier (HEKA Electronics, Data acquisition and analysis were conducted with pulse (HEKA Electronics) and Origin (OriginLab,, in combination with the pulsefit software (HEKA Electronics). Patch pipettes were prepared from borosilicate glass capillaries (Harvard Apparatus, with a DMZ-Universal puller (Carl Zeiss, Inc., The bath chamber was mounted on an inverted fluorescence microscope (Eclipse TE2000-U; Nikon Instruments, For whole-cell experiments, vacuoles with a diameter of 25–40 μm were selected. The applied voltages refer to the cytoplasmic side of the vacuole, whereas the vacuolar side was at the ground (Bertl et al., 1992). The vacuolar surface areas were determined from capacitance currents measured in response to short (10-ms) voltage steps of 10-mV amplitude (Gillis, 1995). All measurements were made at room temperature (20–22°C). Osmolality of all solutions was calibrated to 440 mosmol kg−1 with mannitol. In all solutions malic acid was buffered with Bis-Tris Propane (BTP), a relatively impermeable cation, to ensure that the currents observed were principally attributable to movements of the malate2− ions (Hafke et al., 2003). The standard bath solution contained 100 mm malic acid, 1 mm CaCl2, 1 mm EDTA and 3 mm MgCl2, pH 7.5, adjusted with BTP. The pipette solution contained 10 mm malic acid, 1 mm CaCl2 and 3 mm MgCl2, pH 5.5, adjusted with BTP. For patch-clamp experiments on substrate specificity, malic acid was replaced by fumaric acid and hydrochloric acid. Whole-cell configuration was made by a short bipolar voltage pulse (± 900 mV, 600 μs each). Current density plots (pA pF−1) of Arabidopsis vacuoles were obtained by plotting the isochronal current amplitude differences between the first and last 20 ms of the voltage step, normalized by the tonoplast capacitance, against the applied test voltage. The comparison of WT and AtALMT9-overexpressing Nicotiana vacuoles was performed by the determination of total steady-state current during the last 20 ms of the voltage step.

Two-electrode voltage clamp (TEVC) on Xenopus oocytes

The AtALMT9 cDNA was amplified from the RIKEN clone pda08640 (cDNA clone RAFL09-66-G16; Sakurai et al., 2005; Seki et al., 1998, 2002; Yamada et al., 2003) with the primers Mc8forwCF3 (5′-GGGAATTCGCGGCCGCATGGCGGCGAAGCAAGGTTCCTTC-3′) and Mc8CF3-backw-1 (5′-TATCAAATCATATGTTACATCCCAAAACAC-3′), and was subcloned into the pCF3 expression vector (at NotI and NdeI restriction sites) for efficient expression in oocytes, as described previously (Preston et al., 1992; Shitan et al., 2003). The plasmid was linearized using the unique site AscI, and was used as a template for the synthesis of capped cRNA using a Message Machine T7 kit (Ambion, Stage V–VI defolliculated oocytes from Xenopus were isolated and maintained as described previously (Virkki et al., 2006). TEVC experiments on oocytes expressing AtALMT9 were conducted according to the method described by Baumgartner et al. (1999). Oocytes were injected with 50 nl cRNA (0.2 μg μl−1) encoding AtALMT9. Control oocytes were injected with 50 nl of double-distilled water. After injection, the oocytes were incubated at 18°C in modified Barth’s solution containing 88 mm NaCl, 1 mm KCl, 0.41 mm CaCl2, 0.82 mm MgSO4, 2.5 mm NaHCO3, 2 mm Ca(NO3)2 and 7.5 mm HEPES, pH 7.5, adjusted with 2-amino-2-(hydroxymethyl)-1,3-propanediol (TRIS), supplemented with penicillin (5 mg ml−1) and spectomycin (5 mg ml−1). Electrophysiological experiments were performed 4–5 days after injection. TEVC was made using the Geneclamp 500E Amplifier (Molecular Devices Corporation, The voltage clamp was controlled, and data were acquired using a computer running pclamp8 software (Molecular Devices Corporation), which also controlled the valves for solution switching. Oocytes were initially superfused with ND-100 solution (100 mm NaCl, 2 mm KCl, 1.8 mm CaCl2, 1 mm MgCl2 and 10 mm HEPES, pH 7.4) before switching to experimental solutions. All experiments were performed at room temperature under continuous flow of experimental solutions. Solutions in the recording chamber were changed at a rate of 5 ml min−1. Bath solutions were chosen in order to reveal malate-dependent currents: 0 or 10 mm malic acid, 0.3 mm CaCl2, buffered with BTP to pH 7.5, adjusted to 220 mosmol kg−1 with mannitol. To test for aluminum-dependent activation, 0.1 mm Al2(SO4)3 was added to a bath solution containing 0 mm malate and 0.1 mm LaCl3.

Data analysis was performed using the clampfit software (Molecular Devices Corporation) and graphpad software (GraphPad Software, Current–voltage curves (I–V) were constructed by plotting the isochronal instantaneous currents at 75 ms of the voltage steps versus the test voltages. Each data set was obtained from at least two batches of oocytes from two different donor frogs.


We thank Franco Gambale for advice and discussion on the oocyte experiments, S. W. Peters for careful reading of the manuscript and I. C. Forster for help with experiments on oocytes. This work was supported by the Alexander von Humboldt Stiftung (PK, 1116390gadodin77), the Deutsche Forschungsgemeinschaft (SM, ME 1955/2-1; JS-S, SCHO 1238/1-1), the Swiss National Foundation, the Roche Research Foundation (PK, 2006/101), the EU-Project ‘VaTEP’ (EM) and Global Research program of the Ministry of Science and Technology of Korea (grant no. 4.0001795.01) (YL, EM). We thank the Salk Institute Genomic Analysis Laboratory for providing the sequence-indexed Arabidopsis T-DNA insertion mutants. The authors wish to thank NASC for providing seeds.