A novel intracellular K+/H+ antiporter related to Na+/H+ antiporters is important for K+ ion homeostasis in plants.

In this study we have identified the first plant K+/H+ exchanger, LeNHX2 from tomato (Lycopersicon esculentum Mill. cv. Moneymaker), which is a member of the intracellular NHX exchanger protein family. The LeNHX2 protein, belonging to a subfamily of plant NHX proteins closely related to the yeast NHX1 protein, is abundant in roots and stems and is induced in leaves by short term salt or abscisic acid treatment. LeNHX2 complements the salt- and hygromycin-sensitive phenotype caused by NHX1 gene disruption in yeast, but affects accumulation of K+ and not Na+ in intracellular compartments. The LeNHX2 protein co-localizes with Prevacuolar and Golgi markers in a linear sucrose gradient in both yeast and plants. A histidine-tagged version of this protein could be purified and was shown to catalyze K+/H+ exchange but only minor Na+/H+ exchange in vitro. These data indicate that proper functioning of the endomembrane system relies on the regulation of K+ and H+ homeostasis by K+/H+ exchangers.

With concentrations between 0.1 and 0.2 M, potassium is the most abundant cation in plant cells. The main pool of potassium inside the plant cell is in the vacuole. The function of potassium in this organelle is thought to be purely biophysical; to generate cell turgor to drive cell expansion (1). Under conditions that limit K ϩ availability, the role of K ϩ in the vacuole can be replaced by other ions like Na ϩ , as has been reported to occur under conditions of salt stress (2). Vacuolar concentrations thus vary from high (200 mM) to low (20 mM), suggesting the existence of active K ϩ import and export mechanisms at the vacuolar membrane (3).
The role and concentration of K ϩ in other endomembrane organelles in plants are largely unknown. Apart from a specific K ϩ requirement, secondary ion transporters might rely on K ϩ for pH regulation in these organelles. It was shown that apart from the vacuole, V-type H ϩ -transporting ATPase is found in various membranes of the secretory system where vesicle acidification is important for ligand-receptor binding and protein modification, trafficking, and sorting (4,5). In animal cells, the pH gradient from neutral to acidic along organelles from both the secretory and endocytic pathways is suggested to be under tight control by the operation of secondary ion transporters providing proton leak pathways (6,7,8). Second, solute uptake energized by the pH gradient might be required to generate the osmotic pressure needed for vesicle fusion (5). In view of the abundance of K ϩ in the cell, K ϩ /H ϩ exchangers are likely candidates to be involved in pH and osmoregulation of intracellular compartments as well as active uptake of K ϩ into vacuoles (3) although the biochemical evidence for the existence of such antiporters is scarce (9,10).
In contrast to the limited information available on K ϩ /H ϩ antiporters, many reports have indicated the existence of vacuolar Na ϩ /H ϩ antiporters in plants (11). The first vacuolar Na ϩ /H ϩ exchanger AtNHX1 was identified recently (12), and it was shown that overexpression of this gene in plants enhances salt tolerance (13,14,15). A family of six genes was identified in Arabidopsis (AtNHX1 to AtNHX6) that shows sequence homology to mammalian and yeast NHE or NHX exchangers (16,17). It was demonstrated however that AtNHX1 could catalyze both Na ϩ /H ϩ and K ϩ /H ϩ exchange (14,18). A similar ion specificity was reported for the human NHE7 isoform. It was shown that this isoform is expressed in the trans-Golgi network, indicating that regulation of pH and ionic composition of intracellular compartments by K ϩ /H ϩ or Na ϩ /H ϩ exchange is an important task of these antiporters (7). This notion was strengthened by the observation that the NHX1 protein is essential for osmotolerance and endosomal protein trafficking in yeast (19,20). Indeed, plant NHX genes were shown not only to be involved in salinity tolerance, but also in vacuolar pH regulation (21,22), and to be induced by NaCl, KCl, and osmotic stress (12,16,(23)(24)(25)(26), or even heat shock (27).
We have identified the first NHX proteins in tomato. One isoform, LeNHX2 is relatively distant from the AtNHX1 protein. In this study we describe this isoform, and analyze its function by heterologous expression in yeast. The LeNHX2 protein turns out to be the first intracellular K ϩ /H ϩ antiporter protein in plants, which enables the maintenance of higher K ϩ concentrations in intracellular compartments under conditions of salt stress when expressed in yeast. These data show that plants contain different NHX genes with different ionic specificities regulating K ϩ , Na ϩ , and H ϩ ion homeostasis in intracellular compartments.

EXPERIMENTAL PROCEDURES
Isolation of LeNHX1 and LeNHX2-The Institute for Genomic Research (Rockville, MD) tomato EST data base (www.tigr.org) was searched in order to identify putative tomato genes with homology to NHE-or NHX-like proteins. Two ESTs (260098 and 254764) were identified and obtained from Clemson University Genomics Institute. Elongation of 5Ј-and 3Ј-ends was performed by nested PCR with two pairs of nested primers and a tomato (Lycopersicon esculentum Mill. cv. * This work was supported by Grants PB97-1266 from the Spanish Dirección General de Enseñ anza Superior e Investigación, FD97-0496-C03-02 from the Fondo Europeo de Desarrollo Regional and from the Plan Andaluz de Investigación (PAI) from the Junta de Andalucía (Spain) (to J. P. D.), and by Grant BIO2002-00552 and a Ramón y Cajal fellowship from the Ministerio de Ciencia y Tecnología (to K. V.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Moneymaker) root hair cDNA library (28) as template. One pair of primers was deduced from the polylinker sequence, and a pair of reversed gene-specific primers was derived from the EST sequences. Fragments were subcloned into pSTBlue-1, using Novagen's perfectly blunt cloning kit (Novagen, Madison, WI), and sequenced. The entire LeNHX2 coding sequence was amplified by nested PCR using a mixture of 7:1 Klenterm and Accuterm DNA polymerases (Labgen Molecular Biology) and cloned into the yeast expression vector pRS699, between the yeast PMA1 promoter and terminator (29), giving rise to plasmid pRS-LeNHX2. A different construct, containing a sequence coding for a RGSH 10 epitope tag in the C-terminal end was amplified from the cDNA library and cloned in yeast expression vector pYes (Invitrogen), containing the GAL1 promoter, giving rise to plasmid pY-LeNHX2:H 10 . The coding region was sequenced in both directions to confirm the fidelity of the constructs. The complete cDNA sequences were deposited in the EMBL nucleotide sequence data base (LeNHX1 accession no. AJ306630, LeNHX2 accession no. AJ306631).
Plant Growth Conditions-Tomato seeds (L. esculentum Mill, cv. Pera) were surface-sterilized and then germinated and grown for 5 weeks in sterile vermiculite under a light irradiance of 150 mol m Ϫ2 s Ϫ1 (16-h photoperiod) at 26°C and 60 -65% relative humidity. Onetenth strength Hoagland's nutrient solution (30) was applied from emergence of the first leaf and raised to quarter-strength (2 weeks after sowing) thereafter, every 3 days. Plants were treated for 0, 1, 6, and 24 h with 130 mM NaCl or for 1 h with 100 M abscisic acid, before harvesting root, stem, and leaf tissues.
RNA Blot Analysis-A DNA probe for LeNHX2 was made by amplification of a 415-bp fragment of the C-terminal coding region using gene-specific forward and reverse primers. Total RNA from different tissues was isolated and purified according to Logemann et al. (31). RNA for all blots (15 g) was run on 1.2% denaturing formaldehyde agarose gels and blotted onto nylon membranes (Hybond TM N ϩ , Amersham Biosciences) (32). The DNA probes were radioactively labeled with [␣-32 P]dCTP by random priming using the Rediprime TM II kit (Amersham Biosciences). Nylon filters were prehybridized and hybridized at 65°C in hybridization buffer containing 7% (w/v) SDS, 300 mM sodium phosphate, pH 7.0, and 1 mM EDTA (33). Blots were washed twice in 4ϫ SSC, 0.1 (w/v) SDS, and once in 0.4ϫ SSC, 0.1 (w/v) SDS, at 65°C for 15 min each. Membranes were rehybridized with a tomato 18 S rDNA gene probe to monitor RNA loading quantity (a gift from Dr. N. Ferrol, Est. Exp. Zaidín, CSIC, Granada, Spain). Nylon filters were exposed to a Phosphorimager screen (BioRad Molecular Imager System) and/or exposed to X-Omat AR5 film (Kodak), where hybridization signal was recorded with a Phosphorimager analyzer (BioRad Molecular Imager System) or digitized with a BioRad imaging analyzer, respectively. Transcript amount was determined by quantification of band signal intensities of digitized images and relativized to the respective ribosomal signals using Scion Image software (version 3.62 a ; www.scioncorp.com).
Determination of Total Intracellular Ion Content-Cells were grown in liquid APG medium (APD medium with galactose instead of glucose) supplemented or not with 20 mM NaCl as indicated. 50-ml fractions were taken when the cells reached an OD 660 nm of 0.25 Ϯ 0.01 and centrifuged (2 min at 3000 ϫ g). Cells were washed twice with 10 ml of ice-cold 10 mM MgCl 2 , 10 mM CaCl 2 , 1 mM HEPES. Washed cells were resuspended in 2 ml of the same buffer and the cell density determined from the optical density at 660 nm. Separately the relationship between OD 660 nm and yeast dry weight was determined. Intracellular ions were extracted by addition of HCl to a final concentration of 0.4% and incubation for 20 min at 95°C. After removal of cell debris by centrifugation, potassium and sodium ion content was determined with an atomic absorption spectrometer.
Determination of Vacuolar and Cytoplasmic Ion Content-The vacuolar and cytoplasmic Na ϩ and K ϩ content was determined by treating the cells with cytochrome c that selectively permeabilizes the plasma membrane (35,36). Cells were grown and washed as above and resuspended in 50 l of 2% cytochrome c, 18 g/ml antimycin, 1 mM HEPES, 10 mM MgSO 4 , 10 mM CaCl 2 and 5 mM 2-deoxy D-glucose. After incubation for 20 min at room temperature, the cells were centrifuged and washed three times with 2 ml of the same solution without cytochrome c. The supernatants containing the ions released by the cytochrome c treatment were pooled to determine cytoplasmic ion content. The remaining ions, corresponding to the vacuolar fraction, were extracted by addition of HCl to a final concentration of 0.4% in a total volume of 2 ml and incubation for 20 min at 95°C. Potassium and sodium ion content was determined with an atomic absorption spectrometer as above.
Yeast Membrane Fractionation-Microsomal membranes were isolated as described (18) The microsomal membrane pellet (2 ml) was layered on a linear (28 ml) 20 -55% sucrose gradient prepared in 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, and 30 l of protease inhibitor mixture (Sigma), and centrifuged 18 h at 100,000 ϫ g in a Beckman SW28 swinging bucket rotor. 2-ml fractions were collected, and sucrose density and protein content determined.
Tomato Cell Membrane Fractionation-Microsomal membranes were obtained from Tomato calli (L. esculentum Mill. cv. Pera) as previously described (37). Microsomal membranes were layered on a linear (14 ml) 20 -40% sucrose gradient and centrifuged as described above. 1-ml fractions were collected, and sucrose density and protein content determined.
Marker Enzyme Analysis-In tomato cell membrane fractions latent UDPase activity (Golgi) (38) Antimycin A-insensitive NADH-cytochrome c reductase (ER) 1 (39) and nitrate-sensitive ATPase (Tonoplast) (40) was measured. The LeNHX2 protein was detected by immunoblot analysis using the affinity-purified polyclonal antibody raised against the LeNHX2 c-terminal peptide EPIMHSSRRAGYDGH (Sigma-Genosys). In yeast membrane fractions marker enzyme proteins were detected by immunoblot analysis. The distribution of ER membranes was assayed by determination of NADPH-cytochrome c oxidoreductase activity (41). SDS-PAGE and Western blotting were performed as described (18) using a monoclonal antibody raised against the 100-kDa subunit of the vacuolar proton ATPase VPH1, a monoclonal antibody against the late Golgi protein vps10p, and a monoclonal antibody against the endosomal/prevacuolar protein pep12p (Molecular Probes). The recombinant LeNHX2 protein in yeast was detected using a monoclonal antibody raised against the RGSH 4 epitope (Qiagen).
Reconstitution of LeNHX2:H 10 Protein and Measurement of Cation/H ϩ Exchange in Vitro-Reconstitution was performed by elimination of detergent using Sephadex G-50 (fine; Amersham Biosciences) spin columns, and Biobeads (SM-2; Bio-Rad) as previously described (18) using 5 g of protein and in the presence of ammonium sulfate. In order to encapsulate pyranine during the reconstitution, it was essential to preload the G-50 spin column with 200 l of 2.5 mM pyranine (BTP salt) in reconstitution buffer, as the pyranine present in the liposome-protein-detergent mixture was eliminated by Ͼ99% at the top of the column. This shows that the reconstitution takes place along the passage of the sample through the column, allowing also an efficient elimination of imidazole or traces of KCl and NaCl in the protein sample. An inside acid pH gradient was created by diluting the proteoliposomes 50-fold in (NH 4 ) 2 SO 4 -free medium. Cation/H ϩ exchange activity was monitored by measuring the rise in fluorescence of encapsulated pyranine in the presence of various monovalent cations as previously described (18). For measurement of antiporter activity in the absence of a preimposed pH gradient, ammonium sulfate in the reconstitution was replaced by Cs 2 SO 4 , pH during reconstitution was adjusted to 7.2, and measurements were made in the same buffer containing equilibrium Cs 2 SO 4 concentration.
Affinity Purification of the LeNHX2 Polyclonal Antibody-A rabbit polyclonal antibody was raised against a peptide corresponding to the C-terminal 15 amino acids of the LeNHX2 protein (Sigma-Genosys). The antibody was purified from the serum by affinity purification using the LeNHX2 protein. 10 g of purified protein was separated by SDSgel electrophoresis and transferred to a polyvinylidene difluoride membrane. The band containing the protein was cut from the membrane, blocked, and incubated for 3 h with the serum at a 1:20 dilution in 1.5 ml of 20 mM K-phosphate, pH 7.5, 150 mM NaCl, 0.1% bovine serum albumin. After washing, the antibody was eluted with 750 l of 0.2 M glycine, pH 2.7, 1 mM EGTA and neutralized with an equal volume of 0.2 M Tris-HCl, pH 8.0, 0.2% bovine serum albumin, 0.02% NaN 3 . The purified antibody solution was used at a 1:10 dilution in Western blot experiments.
Protein Determination-Protein was determined by the method of Schaffner and Weissmann (42) with bovine serum albumin as standard.

Isolation and Molecular Description of LeNHX1 and
LeNHX2-To clone antiporters of the NHX family from tomato we aimed at identifying homologues of the Arabidopsis thaliana vacuolar (Na ϩ ,K ϩ )/H ϩ antiporter AtNHX1. Two ESTs were identified and the complete corresponding cDNAs were obtained based on PCR elongation of 3Ј-and 5Ј-coding sequences. The open reading frames of LeNHX1 and LeNHX2 encode for proteins of 531 and 534 amino acids, respectively. Both predicted proteins, which are only 31% identical, are closely related to the family of intracellular Na ϩ /H ϩ antiporters found in animals, plants, and fungi ( Figs. 1 and 2) (16,17,43,44). In Arabidopsis, 6 members of this family were identified that could be subdivided into 2 relatively distant groups (16). All NHX proteins described from other plant species fall within the same group as the AtNHX1 protein constituting a group of closely related plant NHX sequences. As the proteins from the second group are much less known, we focused this study on a description of the functioning of the LeNHX2 protein.
Expression Studies in Plants-Plant and yeast NHX genes are proposed to be involved in salinity or osmotolerance (13,16,23,45). To test the involvement of the LeNHX2 protein in these processes we subjected tomato plants of 5-week-old grown in vermiculite to a shock of 130 mM NaCl. The LeNHX2 gene showed a rapid induction by salt stress or the stress hormone abscisic acid in leaves, but was constitutively expressed in stems and roots (Fig. 3), indicating a role of this protein in the salt or osmotic stress response.
Yeast Complementation Studies-Disruption of the NHX1 gene in S. cerevisiae confers a strong hygromycin B sensitivity to the cells (Fig. 4C), as has been reported by others (12, 23, 46). LeNHX2 could efficiently complement this sensitivity. The ef-fect on salt sensitivity of NHX1 disruption and complementation by the plant gene was studied using a yeast strain with additional disruptions in the plasma membrane Na ϩ efflux systems ENA1-4 and NHA1 (23). An effect of the NHX1 disruption on NaCl sensitivity was observed already at 20 mM NaCl (Fig. 4B). The tomato gene could partly compensate for NHX1 disruption in this strain. We could not demonstrate any effect of LeNHX2 expression on Li ϩ sensitivity (data not shown).
Effect of LeNHX2 Expression on Intracellular Na ϩ and K ϩ Content in Yeast-In normal growth conditions no effect of NHX1 disruption or complementation by the LeNHX2 gene on intracellular K ϩ concentrations could be observed (Fig. 5A). K ϩ content was however much lower in NHX1-disrupted cells as compared with the wild-type cells or cells expressing LeNHX2 when intracellular K ϩ content was compromised by growing the cells in 20 mM NaCl (Fig. 5B). Intracellular Na ϩ content was similar for all strains tested in these conditions, indicating that the NHX1 or LeNHX2 protein does not contribute to Na ϩ accumulation (Fig. 5B). To determine whether the loss in K ϩ was due to a diminution of the cytoplasmic or vacuolar K ϩ concentration, we treated the cells with cytochrome c, which selectively permeabilizes the plasma membrane (35,36). Cytoplasmic K ϩ content was very similar for all strains, but the control strain or the strain expressing LeNHX2 had 2-3-fold higher vacuolar K ϩ content (Fig. 5C).
Detection of LeNHX2 in Membrane Fractions of Yeast Cells-A sequence encoding a C-terminal RGSH 10 epitope tag was added to LeNHX2 after which it was inserted in vector pYes (Invitrogen) harboring the GAL1 promoter. The presence of the tagged LeNHX2 protein in AXT3 cells conferred resistance to hygromycin and NaCl when cells were grown on galactose-containing media and still resulted in higher levels of intracellular K ϩ , but not Na ϩ (data not shown), indicating that the protein was functional. The tagged LeNHX2 protein was readily detected in microsomal membranes isolated from galactose-grown cells using a monoclonal antibody raised against the RGSH 4 epitope. In a linear sucrose gradient, the LeNHX2 protein equilibrated in low density sucrose fractions corre- sponding to vacuolar or endomembranes, and clearly separated from higher density mitochondrial or plasma membrane markers (data not shown). In more detail, the LeNHX2 protein had a distribution pattern similar to the late Golgi marker vps10p and, to a lesser extent, the endosomal/prevacuolar marker pep12p. The vacuolar marker VPH1 equilibrated at slightly higher sucrose densities (17) (Fig. 6).
Detection of LeNHX2 in Membrane Fractions of Tomato Cells-To detect the LeNHX2 protein in plant membrane fractions, we obtained a polyclonal antibody raised using a peptide corresponding to the extreme C-terminal 15 amino acids (Sigma-Genosys). In yeast, this antibody detected a 50-kDa protein only in cells expressing the LeNHX2 protein (data not shown).
After fractionation of tomato cell microsomes on a linear sucrose gradient, the LeNHX2 protein comigrated with the markers for Golgi and vacuolar membranes (Fig. 7).
Purification of His-tagged LeNHX2 Protein and Cation/H ϩ Antiport Activity-The tagged LeNHX2 protein was purified by Ni 2ϩ affinity chromatography (Fig. 8A) and reconstituted in proteoliposomes containing the pH indicator pyranine. After creation of an inside acid pH gradient, the capacity of monovalent cations to dissipate the pH gradient was assayed (Fig. 8B). In the presence of K ϩ a significant internal alkalinization could be observed, especially at concentrations above 100 mM. The exchange reaction was much less efficient in the presence of Na ϩ . In the presence of Li ϩ or Cs ϩ only a residual activity was FIG. 2. Comparison of the primary structure of NHX sequences from different organisms. The amino acid sequence of NHX proteins representative of the different groups shown in Fig. 1 were aligned using ClustalX 1.5b. Positions containing conserved amino acids are shaded black, while positions containing identical amino acids in all sequences are marked with an asterisk. The membrane topology of the LeNHX1 and LeNHX2 proteins, as predicted at genome.cbs.dtu.dk (56) is shown above the sequence by a filled (transmembrane segment for both proteins) or open (transmembrane segment for one of the two proteins) bar. Also indicated are segments 6 and 7 that were slightly below the threshold in this prediction for both proteins. The numbering of the transmembrane segments is based on the topology prediction for the human NHE1 protein (57). The first transmembrane segment is absent from all the plant sequences, while the transmembrane segment between TM9 and TM10 (not numbered) is proposed to be extracellular in the NHE1 model. Amino acids in the human NHE1 isoform that have been shown to affect the inhibition of antiporter activity by amiloride derivatives (52,53,54) are indicated below the sequence at their corresponding position in the alignment of the NHX sequences.
observed. The exchange reaction showed sigmoidal kinetics as a function of substrate concentration, indicating allosteric effects (Fig. 8C). The same ionic specificity was observed when the exchange reaction was assayed in the absence of a preimposed pH gradient although the exchanger was much less efficient in these conditions (Fig. 9A). The exchange reaction was only slightly inhibited by the amiloride analogs EIPA and Benzamil (Fig. 9B). DISCUSSION In this study we cloned two tomato genes of the NHX family. In plants, the AtNHX1 gene was studied extensively and a role for this protein in salt tolerance has been demonstrated (13,14,15). As the LeNHX2 gene is relatively distant from AtNHX1 ( Figs. 1 and 2) it was characterized in more detail. The identification of the LeNHX2 protein as an intracellular K ϩ /H ϩ exchanger highlights the importance of the maintenance of proper K ϩ ion homeostasis during salt stress.
Expression Studies in Plants-In Arabidopsis, 6 isoforms of the NHX family were described (16). The LeNHX1 protein is closely related to proteins AtNHX1 to AtNHX3 while the Microsomal membrane extracts from AXT3 cells expressing LeNHX2:H 10 were fractionated on a linear sucrose gradient (20 -55%, w/w). Samples (25 g of protein) were resolved by SDS-PAGE and blotted. Distribution of LeNHX2 was assayed with a monoclonal antibody raised against the RGSH 4 epitope (closed symbols), and compared with distribution of VPH1 (A) as marker for the vacuolar membrane, vps10p (C) as marker for the late Golgi membrane and pep12p (B) as marker for prevacuolar/ endosomal membranes. As a marker for ER membranes, NADPH cytochrome c oxidase activity was assayed using 8 g of protein for each sucrose gradient fraction (D). The amount of marker enzyme in Western blots was determined by quantification of the bands using Scion Image software (version 3.62a; www.scioncorp.com). The data represent the means of three blots using the same gradient.

FIG. 7. Subcellular fractionation of LeNHX2 in tomato cells.
Microsomal membranes were prepared from tomato callus as indicated, and layered on a linear sucrose gradient (20 -45% w/w, A). Samples (25 g of protein) were resolved by SDS-PAGE and blotted. Distribution of LeNHX2 was assayed with a polyclonal antibody raised against the C-terminal 15 amino acids of the protein (B). Distribution of Golgi (), ER (‚), and tonoplast (E) was assayed with marker enzyme activity as described, using NADPH cytochrome c oxidase activity as a marker for the ER, latent UDPase activity as a marker for Golgi membranes, and nitrate-sensitive ATPase activity as a marker for tonoplast membranes (C).
LeNHX2 protein is closely related to a different subgroup of plant NHX proteins and shows highest sequence homology to the AtNHX5 protein (Fig. 1). The LeNHX2 gene is ubiquitously expressed but we could show a modest increase in transcript levels in aerial parts of tomato plants by short term NaCl stress or treatment with the plant stress hormone abscisic acid, indicating a role for this protein in osmotic stress in plants (Fig. 3). The AtNHX5 gene was also shown to be induced by NaCl stress in Arabidopsis seedlings (16), but not by abscisic acid treatment. This would indicate that similar isoforms are under control of differential regulatory pathways in different plant species. Such discrepancies were also suggested from studies comparing the expression of isoforms from halophytic species like Mesembryanthemum crystallinum and Atriplex gmelini with those of Arabidopsis and rice (13,(23)(24)(25)(26).
Yeast Complementation Studies-In yeast NHX1 is the only member of the intracellular antiporter family (16). A role for NHX1 in yeast salinity tolerance was demonstrated, but for unknown reasons the protein is also involved in conferring resistance to other toxic cations like hygromycin B (12,45). We show that the plant LeNHX2 gene can complement the salt-and hygromycin-sensitive phenotypes caused by NHX1 disruption (Fig. 4).
Yeast achieves salt tolerance mainly through exclusion of toxic sodium ions at the plasma membrane through the operation of a plasma membrane Na ϩ pump and Na ϩ /H ϩ antiporter (23,47). A disruption of these plasma membrane Na ϩ efflux systems results in a dramatic increase in the intracellular Na ϩ /K ϩ ratio when the cells are grown in the presence of NaCl (48,49). An effect of NHX1 gene disruption on salt sensitivity is most clearly demonstrated in this yeast strains. However, even in these conditions a role of NHX1 in salt tolerance can only be observed when cells are grown at low K ϩ concentrations (23), indicating that potassium ion homeostasis is important for the observed phenotype.
We now show that under salt stress conditions in this strain, intracellular K ϩ concentrations are strongly affected by the disruption of the NHX1 gene, which can be complemented by the tomato antiporter LeNHX2 (Fig. 5B). Under the mild NaCl stress conditions used we could not detect any effect on intracellular Na ϩ concentrations (Fig. 5B). Similar results were obtained recently studying the effects of AtNHX1, 2 and 5 in yeast (16). A minor increase in Na ϩ concentration could be observed only in yeast cells expressing AtNHX1, while no increases were observed in cells expressing AtNHX5. On the contrary, all isoforms caused a very significant increase of intracellular K ϩ concentrations as compared with the NHX1disrupted strain, the most efficient being AtNHX5. By treating the cells with cytochrome c, we could show that in salt stress conditions the NHX1 disruption and complementation by LeNHX2 affects K ϩ concentrations in vacuolar or intracellular compartments, while cytoplasmic K ϩ concentrations remain similar (Fig. 5C). This suggests that the accumulation of K ϩ in intracellular compartments by NHX1 or LeNHX2 plays an important role for growth at limiting cytoplasmic K ϩ concentrations. On the other hand, we did not obtain evidence for the involvement of LeNHX2 in the active accumulation of K ϩ inside vacuoles in the absence of salt stress (Fig. 5A).
It has been proposed that in plants Na ϩ ions can replace K ϩ ions for osmotic adjustment in the vacuolar pool upon conditions of salt stress (1). Also in yeast, the accumulation of salt in vacuoles by the ScNHX1 protein would reduce the concentration of toxic Na ϩ ions in the cytoplasm, thereby improving salt tolerance (12,17,23). Our data show that K ϩ homeostasis of intracellular compartments is an additional toxicity target of salt stress in plants and yeast. Essential cellular processes like regulation of pH in organelles of the secretory pathway might rely on the presence of K ϩ /H ϩ antiporters at these organelle using the readily available K ϩ , while vacuolar pH and Na ϩ concentrations are controlled by distinct Na ϩ /H ϩ antiporters. It was indeed shown that disruption of NHX1 in yeast does not abolish ⌬pH-dependent Na ϩ /H ϩ antiport activity in vacuolar vesicles (50). In summary, our data indicate that the physiologically significant activity of NHX1 and LeNHX2 is K ϩ /H ϩ and not Na ϩ /H ϩ antiport.
Biochemical Characterization of LeNHX2-Accumulation of K ϩ in intracellular compartments is consistent with the observed distribution of the LeNHX2 protein in a linear sucrose gradient (Figs. 6 and 7). Furthermore, we demonstrated that the purified LeNHX2 protein specifically transports K ϩ , exhibiting very low activities with other monovalent cations (Figs. 8B and 9A). Previously, we showed that the AtNHX1 protein is a less specific monovalent cation/H ϩ antiporter, with similar affinities for K ϩ and Na ϩ (18). These data are consistent with the effect of AtNHX1, but not LeNHX2 or AtNHX5 on intracellular Na ϩ concentrations in yeast (16). In contrast to the AtNHX1 protein, that exhibited simple saturation kinetics FIG. 8. Purification of LeNHX2: H 10 protein and assay of cation/H ؉ exchange activity. LeNHX2:H 10 protein was purified as described under "Experimental Procedures." 100 l of the purified protein fraction containing LeNHX2:H 10 was trichloroacetic acid-precipitated and subjected to SDS-PAGE on a 10% acrylamide gel and stained with Coomassie Brillant Blue (A). An acid-inside pH gradient was created in proteoliposomes containing purified protein by ammonium sulfate dilution as described under "Experimental Procedures" (B). The cation/H ϩ exchange reaction was started by the addition of 150 mM of chloride salts of the indicated cations (arrow). The rate of the exchange reaction as a function of K ϩ (E), Na ϩ (q), Li ϩ (‚), and Cs ϩ (ƒ) concentration was estimated from the initial rate of fluorescence change (C). with K m values for K ϩ or Na ϩ of around 40 mM, the LeNHX2 protein shows sigmoidal kinetics as a function of K ϩ concentration suggesting more than one K ϩ binding site that show positive cooperativity (Fig. 8C). This results in a strong dependence of the activity on K ϩ concentrations in the range from 50 to 150 mM K ϩ , which is in accordance with the concentrations normally found in the cytoplasm of plant cells (3). The lower K m values obtained for the AtNHX1 protein strengthen a role for this enzyme in vacuolar Na ϩ accumulation, as cytoplasmic Na ϩ concentrations are suggested to be much lower. Allosteric kinetics with respect to Na ϩ were only reported for the mammalian isoform NHE4, while the majority of mammalian isoforms function at maximal turnover rates with respect to Na ϩ , but exhibit allosteric kinetics with respect to H ϩ resulting in a strong activation by a decrease in cytoplasmic pH (51). At the same time, a variety of effector molecules or posttranslational modifications such a phosphorylation or dephosphorylation could modify the K ϩ activation profile. In the absence of a preimposed pH gradient, addition of KCl also induced the rapid formation of proton efflux from the vesicles (Fig. 9A). The LeNHX2 protein contains the typical motives that are suggested to be involved in amiloride binding and inhibition ( Fig.  2) (52)(53)(54). Nevertheless, unlike our previous results obtained with the reconstituted AtNHX1 protein (18), the LeNHX2 protein was only slightly inhibited by amiloride analogs (Fig. 9B). It has to be noted that in the reconstituted system we could not assay higher concentrations of inhibitors, as these compounds interfere with membrane permeability (18).
In conclusion, we have cloned two genes from the NHX antiporter family in tomato. The LeNHX2 gene encodes the first intracellular K ϩ /H ϩ antiporter in plants, and our data indicate that the control of K ϩ concentrations and pH in intracellular compartments by this enzyme is important for ion homeostasis and salt tolerance in plants and yeast. Finally, the resemblance of the phenotype of cells expressing ScNHX1 with that of cells expressing LeNHX2 suggests that LeNHX2 might be the true functional homologue of the ScNHX1 gene.