Cloning and characterization of a novel Mg2+/H+ exchanger

Orit Shaul, Donald W. Hilgemann, Janice de‐Almeida‐Engler, Marc Van Montagu, Dirk Inzé, Gad Galili

Author Affiliations

  1. Orit Shaul*,1,
  2. Donald W. Hilgemann2,
  3. Janice de‐Almeida‐Engler3,
  4. Marc Van Montagu3,
  5. Dirk Inzé3,4 and
  6. Gad Galili1
  1. 1 Department of Plant Sciences, The Weizmann Institute of Science, Rehovot, 76100, Israel
  2. 2 Department of Physiology, University of Texas Southwestern Medical Center, TX, 75235, USA
  3. 3 Laboratorium voor Genetica, Department of Genetics, Flanders Interuniversity Institute for Biotechnology, Universiteit Gent, B‐9000, Gent, Belgium
  4. 4 Laboratoire Associé de l'Institut National de la Recherche Agronomique, France
  1. *Corresponding author. E-mail: lpshaul{at}


Cellular functions require adequate homeostasis of several divalent metal cations, including Mg2+ and Zn2+. Mg2+, the most abundant free divalent cytoplasmic cation, is essential for many enzymatic reactions, while Zn2+ is a structural constituent of various enzymes. Multicellular organisms have to balance not only the intake of Mg2+ and Zn2+, but also the distribution of these ions to various organs. To date, genes encoding Mg2+ transport proteins have not been cloned from any multicellular organism. We report here the cloning and characterization of an Arabidopsis thaliana transporter, designated AtMHX, which is localized in the vacuolar membrane and functions as an electrogenic exchanger of protons with Mg2+ and Zn2+ ions. Functional homologs of AtMHX have not been cloned from any organism. Ectopic overexpression of AtMHX in transgenic tobacco plants render them sensitive to growth on media containing elevated levels of Mg2+ or Zn2+, but does not affect the total amounts of these minerals in shoots of the transgenic plants. AtMHX mRNA is mainly found at the vascular cylinder, and a large proportion of the mRNA is localized in close association with the xylem tracheary elements. This localization suggests that AtMHX may control the partitioning of Mg2+ and Zn2+ between the various plant organs.


In all living organisms, cellular functions require a fine homeostasis of various ions and nutrients, including Mg2+ and Zn2+. Mg2+ is required for the function of many enzymes (e.g. phosphatases, ATPases and RNA polymerases). Zn2+ plays both a functional (catalytic) and structural role in several enzyme reactions, and is involved in the regulation of gene expression by zinc‐finger proteins. Both Mg2+ and Zn2+ are essential for the structural integrity of ribosomes. In plants, Mg2+ is also an essential component of chlorophyll, and regulates the activity of key chloroplastic enzymes.

Multicellular organisms have to balance not only their Mg2+ and Zn2+ intake and intracellular compartmentalization, but also the distribution of these ions to various organs. The movement of ions through membrane barriers is mediated by specialized proteins in the form of channels, transporters or ATPases. So far, genes encoding Mg2+ transporters have been cloned only from bacteria and yeast. The bacterial MgtA and MgtB Mg2+ transport proteins are P‐type ATPases (Hmiel et al., 1989). Mg2+ is also transported by the bacterial CorA and mgtE proteins (Smith et al., 1993, 1995), and by the yeast ALR homologs of bacterial CorA (MacDiarmid and Gardner, 1998), but the molecular mechanism of Mg2+ mobilization by these proteins is not known. Among the Zn2+ transport proteins whose genes have been cloned, the bacterial ZntA (Rensing et al., 1997) is a P‐type ATPase. Zn2+ is also transported by the yeast ZRT 1,2 (Zhao and Eide, 1996a, b), the Arabidopsis ZIP 1–4 (Grotz et al., 1998) and the mammalian ZnT 1–4 (McMahon and Cousins, 1998) transporters, but the molecular mechanism of Zn2+ transport by these proteins is also unknown. A mammalian protein designated DCT1 (Gunshin et al., 1997), which belongs to the Nramp family of macrophage proteins, was suggested to be a symporter of protons with various divalent metal cations, including Fe2+ and Zn2+, but was not able to symport Mg2+ ions.

Little is known about transport proteins that control Mg2+ and Zn2+ homeostasis in plants. While the proteins mediating Mg2+ uptake into roots are unknown, Zn2+ transporters, which are induced by Zn2+ deficiency and are probably involved in Zn2+ uptake, have recently been cloned from Arabidopsis (Grotz et al., 1998). Ions absorbed into the cytosol of root cells diffuse towards the vascular cylinder through plasmodesmata and reach the xylem parenchyma cell layer, which borders the xylem vessels. The xylem parenchyma cells were suggested to play a key role in ion secretion into the xylem (xylem loading) and in the release of ions from the xylem (unloading) (Marschner, 1995). These processes require transport through the plasma membrane of the xylem parenchyma cells, but the proteins mediating xylem loading and unloading of Mg2+ and Zn2+ are not known. Unloaded Mg2+ and Zn2+ subsequently enter the surrounding cells through unknown transport proteins. The molecular mechanisms of phloem loading and unloading with Mg2+ and Zn2+ have also not been elucidated. Intracellularly, the vacuole is considered the main organelle mediating Mg2+ homeostasis in the cytosol and the chloroplast. Vacuolar Mg2+ is also important for the cation–anion balance and turgor regulation of cells (Marschner, 1995). The activity of a Mg2+/H+ antiporter was identified in lutoid (vacuolar) vesicles of Hevea brasiliensis (Amalou et al., 1992, 1994) and in vacuolar membranes from roots of Zea mays L. (Pfeiffer and Hager, 1993), but cloning of the corresponding genes has not been reported. The H.brasiliensis transporter was indicated to be electroneutral and to be capable of transporting also Zn2+ cations. In Zn2+‐tolerant species, tolerance is achieved mainly by sequestering Zn2+ in the vacuoles (Brookes et al., 1981), but the transport mechanism is not known.

We describe here the cloning and characterization of an Arabidopsis transporter, designated AtMHX, that is localized in the vacuolar membrane and functions as an electrogenic exchanger of protons with Mg2+ and Zn2+ ions. To our knowledge, genes encoding Mg2+/H+ or Zn2+/H+ exchangers have not yet been cloned from any organism. The main site of AtMHX transcription is the vascular cylinder, suggesting that it plays a role in Mg2+ and Zn2+ partitioning between the various plant organs.


Cloning of AtMHX

Using a PCR approach, we isolated an Arabidopsis cDNA fragment whose deduced amino acid sequence showed homology to NCX1, a mammalian Na+/Ca2+ exchanger (Nicoll et al., 1990). In animal cells, NCX1 plays a major role in extrusion of Ca2+ ions to the extracellular space following excitation. Considering that homologs of animal Na+/Ca2+ exchangers have not been characterized from any plant species, we aimed to determine the role of the plant protein, and cloned its entire cDNA and genomic DNA (these sequences have been submitted to the DDBJ/EMBL/GenBank database under accession Nos AF109178 and AF109182, respectively). The deduced protein encoded by this gene, later designated AtMHX, showed 36% identity to NCX1 (Figure 1A). Hydrophobicity plots indicate that the two proteins share structural similarity (Figure 1B). Both AtMHX and NCX1 contain 11 putative transmembrane spans, which parallel each other throughout the whole sequence (Figure 1A). In addition, both proteins have a long non‐membranal loop between transmembrane spans 5 and 6. In NCX1, this loop is 500 amino acids (aa) long and is not essential for the transport function, but has a regulatory role; in AtMHX this loop is much shorter, being only 100 aa in length.

Figure 1.

(A) Comparison between the deduced amino acid sequences of AtMHX (A, upper rows) and NCX1 (N, lower rows). Most of the non‐membranal loop of NCX1 (between amino acids 297–746) is not shown. The transmembrane spans, as predicted by the Eisenberg, Schwarz, Komarony and Wall method (Eisenberg et al., 1984), are underlined. (B) Hydrophobicity plot of AtMHX (upper line) and NCX1 (lower line). The non‐membranal loop of NCX1 is not shown. The numbers on the x‐axis correspond to the amino‐acid sequence of AtMHX. The numbers on the y‐axis correspond to the hydrophobicity of each protein separately.

We used the coding‐sequence of AtMHX as a probe for low‐stringency Southern blot hybridization of genomic Arabidopsis DNA, which has been digested with restriction enzymes whose recognition sequences appear only at the 5′ or 3′ non‐coding regions (Figure 2). The appearance of a single hybridization band suggested that AtMHX is encoded by a single gene in Arabidopsis. Comparison between the cDNA and genomic clones showed that the AtMHX gene includes eight introns (data not shown). The first intron is 414 bp long and resides within the 5′ untranslated region (5′UTR), while the other introns range between 70 and 148 bp. We have used the inverse PCR technique (Triglia et al., 1988) to obtain part of the 5′ upstream region of the AtMHX gene. To identify the site of transcription initiation within this sequence, we used the 5′ RACE system (Clontech) as described previously (Liu and Gorovsky, 1993), with slight modifications. In short, a synthetic single‐stranded anchor primer was ligated at the 3′ position of an Arabidopsis first‐strand cDNA pool. PCR fragments were amplified using a 5′ gene‐specific primer, and at least 20 of the resulting clones were sequenced. The longest clones included an additional G residue at the 5′ terminus of the anchor primer, which did not reside from the sequences of either the anchor primer or the upstream region of the AtMHX gene. This G residue reflected the position of the 5′‐CAP of the cDNA, and indicated that the transcription initiation site is located at 575 bp upstream to the initiating ATG codon (Figure 3). Typical CAAT and TATA boxes are located at positions −74 and −34, respectively, relative to the transcription initiation site (Figure 3). After AtMHX was cloned in our laboratory, its genomic DNA sequence was also found on chromosome II in the frame of the Arabidopsis genome project (DDBJ/EMBL/GenBank accession No. AC002535), and its partial deduced amino acid sequence, translated from an internal methionine, has also been annotated to have similarity to the Na+/Ca2+ exchanger.

Figure 2.

Low‐stringency Southern blot hybridization of genomic Arabidopsis DNA with the AtMHX coding region. The genomic DNA was cut with HincII (H), BamHI (B), or PvuI (P), whose recognition sequence appear only in the 3′ or 5′ non‐coding regions of the genomic DNA clone.

Figure 3.

The 5′ upstream region of the AtMHX gene. The 414 bp intron in the non‐coding region is underlined. Double‐underlined are the transcription initiation site at position +1, the putative CAAT and TATA boxes at positions −74 and −34, respectively, and the initiating ATG codon (MET).

AtMHX is localized in the vacuolar membrane

AtMHX contains 11 putative transmembrane spans (Figure 1), but lacks any special sequences that could suggest in which cellular membrane it is localized. NCX1 is localized in the plasma membrane and includes a cleaved signal peptide, which is, however, not essential for its localization (Sahin‐Toth et al., 1995). Alignment of the deduced amino‐acid sequences of AtMHX and NCX1 (Figure 1A) shows that AtMHX initiates around the cleavage point of the NCX1 signal peptide, located 32 residues downstream from the NCX1 initiating methionine. To identify the cellular localization of AtMHX, we produced polyclonal antibodies against a peptide from the deduced amino acid sequence of its central non‐membranal loop. These antibodies were purified further against the same peptide. Fractionation of Arabidopsis membranes on sucrose gradient, followed by Western blot analysis, showed that AtMHX co‐fractionates with the vacuolar membrane marker γ‐TIP, and not with plasma membrane or ER markers (Figure 4). Similarly, it did not co‐fractionate with mitochondria, plastids or nuclei, as indicated by differential centrifugation (data not shown). AtMHX is thus localized in the vacuolar membrane. This localization was supported by our electrophysiological analyses in membranes of tobacco cells transformed with the AtMHX gene (see below).

Figure 4.

Intracellular localization of AtMHX in wild‐type Arabidopsis plants. Arabidopsis root membranes were extracted and fractionated in sucrose gradient as previously described (Schaller and DeWitt, 1995). The fractions (fraction 1 = 20% sucrose; fraction 13 = 45% sucrose) were subjected to Western blot analyses with the following antibodies: AtMHX, affinity‐purified antibodies against a peptide from AtMHX deduced amino‐acid sequence; VM, antibodies against a vacuolar membrane marker [VM23, a homolog of γ‐TIP from radish (Raphanus sativus), which is a species closely related to Arabidopsis (Maeshima, 1992)]; PM, antibodies against the Arabidopsis plasma membrane marker protein RD‐28 (Yamaguchi‐Shinozaki et al., 1992); ER, antibodies against the endoplasmic reticulum yeast BiP protein, that specifically recognize plant ER BiP (Shimoni et al., 1995).

AtMHX exchanges protons with Mg2+, Zn2+ and Fe2+ ions

To identify the function of AtMHX, we overexpressed it under control of the strong constitutive 35S promoter (Guilley, 1982) and the Ω enhancer of translation (Gallie et al., 1987) in two independently transformed tobacco BY‐2 cell lines, each composed of a heterogeneous mixture of transformed cells (Nagata et al., 1992; Shaul et al., 1996). The two transgenic cell lines produced a protein with the expected mol. wt of 54 kDa, which did not appear in control, non‐transformed cells (data not shown). Similar to wild‐type Arabidopsis plants, AtMHX co‐migrated in sucrose density gradients with the vacuolar membrane marker of the transformed BY‐2 cells (data not shown), indicating that it was transported to the vacuole also in the transformed tobacco cells.

To determine whether AtMHX carries out electrogenic ion transport, we studied ionic currents in giant patches (8–12 μm) (Hilgemann, 1995; Hilgemann and Lu, 1998) from both the plasma membrane and from vacuoles of transformed and non‐transformed cells. Similar results were obtained in the two independently transformed cultures. Tests for Na+/Ca2+ exchange currents and Na+‐ or Ca2+‐activated conductances were entirely negative. As shown in Figure 5, large currents were activated by applying acidic solutions (pH 5.7) to the exposed (intravacuolar) membrane surface of vacuolar patches from transformed cells (Figure 5A). Solutions on both membrane sides contained only N‐methyl‐glucamine (NMG) as monovalent cation. Current activation required the presence of Mg2+, but not Ca2+ in the pipette (Figure 5E), which strongly suggests an electrogenic Mg2+/H+ exchange process. Vacuolar patches from non‐transformed cells (Figure 5B and E) or patches from the plasma membrane of either transformed or non‐transformed cells (Figure 5E gave almost no current response, supporting the localization of AtMHX in the vacuolar membrane. The current activated by low pH decayed substantially over 30 s (Figure 5A), a kinetic property reminiscent of bovine cardiac Na+/Ca2+ exchange inactivation (Hilgemann, 1990). A current of opposite sign was activated transiently when the control solution (pH 7.7) was applied. This might reflect reverse Mg2+/H+ exchange because there is a 6‐fold proton gradient opposing a 4‐fold Mg2+ gradient (pipette solution, 2 mM Mg2+ at pH 7.0; bath solution, 0.5 mM Mg2+ at pH 7.7). The current–voltage relation of the proton‐activated current (subtraction of records 1 and 2 in Figure 5A) showed a [U] shape (Figure 5C), and the reverse current also shows a complex voltage dependence. These wave forms may reflect the presence of two electrogenic ion translocation steps with opposite voltage‐dependence. Current–voltage relations in vacuolar patches from non‐transformed cells were very shallow in comparison and were monotonic (Figure 5D). The outward current activated by protons was larger when 2 mM Ca2+ was also included in the pipette (n = 3), although 2 mM Ca2+ alone supported no current (n = 4; Figure 5E). Activation was also seen with 50 μM Ca2+ in the pipette (data not shown), but no activation was observed when Ca2+ was applied to the exposed (intravacuolar) membrane surface of vacuolar patches. Activation by cytoplasmic Ca2+ is another property reminiscent of the bovine cardiac Na+/Ca2+ exchanger (Hilgemann, 1990). Both Zn2+ (0.2 mM) and Fe2+ (0.2 mM) supported proton‐activated current when they were included in the pipette instead of Mg2+ (Figure 5E), whilst Co2+, Ni2+ and Cu2+ did not support current (data not shown). From these results we conclude that AtMHX is an electrogenic Mg2+/H+ exchanger in the vacuolar membrane, but it may also transport other divalent metals. Those that support current, Zn2+ and Fe2+, are of similar ionic radius to Mg2+. Since under physiological conditions iron is almost exclusively found in complexes in plant cells (Marschner, 1995), Fe2+ ions are unlikely to be transported in significant amounts by AtMHX in vivo.

Figure 5.

Proton‐activated currents associated with AtMHX expression in vacuolar giant patches. (A) Patch from a vacuole of a transformed cell. Currents are activated by switching from a pH 7.7 solution to a pH 5.7 solution and back to pH 7.7. N‐methyl‐glucamine (NMG) is the only monovalent cation in solution. The pipette solution (pH 7.0; cytoplasmic membrane side) contains 2 mM Mg2+ and 2 mM Ca2+; the bath solution contains 0.5 mM Mg2+. (B) Typical current records for the same protocol in a vacuolar patch from a non‐transformed cell. (C) Current–voltage relations for the proton‐activated current in Figure 5A (squo;2‐1‘), whereby records were subtracted just before (1) and after (2) application of the pH 5.7 solution. In addition, the current–voltage relation is given for the reverse current observed on removing protons, whereby the subtracted records were obtained just after returning to pH 7.7 (3) and 30 s later when the current had decayed (4). (D) Current–voltage relation of the 10 times smaller current activated by the same protocol in a giant patch from a non‐transformed cell. (E) Average magnitudes of 2–4 proton activated currents in giant patches from vacuolar membrane (vm) and plasma membranes (pm) from wild‐type (wt) and transformed (t) cells. In vacuolar patches from transformed cells (t/vm) currents are largest with 2 mM Mg2+ and 2 mM Ca2+ in the pipette (’Mg+Ca'); 2 mM Ca2+ alone did not support a current; currents activated with 2 mM Mg2+, 0.2 mM Zn2+ and 0.2 mM Fe2+ in the pipette are similar in magnitude. No significant currents were obtained in plasma membrane patches of transformed cells (t/pm) or in membrane patches of wild‐type cells (wt/pm, wt/vm) with either of these solutions in the pipette (the currents shown for wt/pm, wt/vm and t/pm were measures with 2 mM Mg2+ and 2 mM Ca2+ in the pipette).

AtMHX apparently functions in Mg2+ and Zn2+ transport in planta

Following the electrophysiological analyses, we wished to ascertain whether AtMHX can also function as a Mg2+ and Zn2+ transporter in planta. To address this, we overexpressed AtMHX in transgenic tobacco plants under control of the 35S promoter and the Ω enhancer of translation. Three independently transformed lines which overexpressed a protein with the expected mol. wt (Figure 6) were selected for further study. The overexpressed protein co‐migrated in sucrose density gradients with the vacuolar membrane marker of tobacco plants (data not shown). When grown in tissue culture plates in medium supplemented with excess Mg2+ or Zn2+ ions, the three transgenic tobacco lines exhibited necrotic lesions in their leaves, which were not observed in wild‐type plants grown under the same conditions (an example is shown in Figure 7). Similar lesions did not appear in transgenic tobacco plants grown in the presence of high concentrations of other cations, or equal concentrations of the accompanying anions with different cations (data not shown). These findings suggest that AtMHX is also capable of transporting Mg2+ and Zn2+ in planta.

Figure 6.

Western blot of transformed tobacco plants with anti‐AtMHX antibodies. Twenty micrograms of protein were loaded on each lane. C, a non‐transformed tobacco plant; 1, 2 and 9, three independently transformed tobacco lines.

Figure 7.

The phenotype of transgenic tobacco plants grown on medium containing elevated levels of Mg2+ or Zn2+. (A) Shoots of tobacco plants that were grown in tissue culture in Nitsch medium (Nitsch, 1969) containing 60 mM Mg2SO4 (compared with 0.75 mM in a standard Nitsch medium). Arrows indicate the necrotic lesions in the two independent transgenic tobacco lines 2 (right) and 9 (middle), which do not appear in the non‐transformed plant (left). (B) Tobacco plants grown in a tissue culture plate in Nitsch medium containing 0.175 mM Zn2SO4 (compared with 0.035 mM in a standard Nitsch medium). The plate was divided to three zones: bottom, non‐transformed plants; upper, the two transgenic lines 2 (right) and 9 (left). Arrows indicate the necrotic lesions in leaves of the transgenic plants.

Next, we wished to study whether AtMHX overexpression affects the accumulation of magnesium or zinc in the plants. Interestingly, no difference was observed in the total content of these minerals between shoots of transformed and non‐transformed plants (Figure 8). The amounts of magnesium or zinc increased to similar levels in shoots of transformed and non‐transformed plants upon growth in media containing elevated levels of these minerals (Figure 8).

Figure 8.

Magnesium and zinc content in shoots of plants grown in plates including high levels of these minerals. Each plate included 12 seedlings, whose aerial parts were pooled before analysis. Each column represents the average ±SD of four plates for the wild‐type plants, or two plates for each of the two independently transformed lines 2 and 9. (A) Magnesium content in shoots of plants grown in Nitsch medium (Nitsch, 1969) containing standard Mg2+ levels (0.75 mM Mg2SO4), or with the indicated levels of Mg(NO3)2 (Mg‐Nit) or Mg2SO4 (Mg‐Sul). (B) Zinc content in shoots of plants grown in Nitsch medium containing standard Zn2+ levels (0.035 mM Zn2SO4), or with 0.5 mM Zn2SO4 (Zn‐Sul).

AtMHX mRNA is mainly associated with the xylem tracheary elements of all plant organs

Northern blots showed that AtMHX mRNA is more abundant in Arabidopsis roots, stems and inflorescence than in leaves (Figure 9). In situ hybridization analysis was carried out to delineate the sites of expression (Figure 10). In cross sections of roots and stems, expression was seen in the vascular cylinder (Figure 10A and B). Expression could also be seen in the root epidermis. In longitudinal sections of all plant organs (roots, stems, leaves, inflorescence and siliques), intense staining was observed in close proximity to the xylem tracheary elements (e.g. Figure 10C–H). No signal was observed in this region upon hybridization with sense probes (data not shown). In addition, expression could be observed in meristematic regions and in very young leaves (data not shown). Altogether, we observed that in all mature plant organs, AtMHX mRNA is found mainly at the vascular cylinder, and a large proportion of the mRNA is localized in close association with the xylem tracheary elements. As differentiated tracheary elements are dead cells, the expression should reside at the living cell layers that border the xylem tracheary elements, such as the xylem parenchyma. However, expression of AtMHX in other cell layers of the vascular cylinder cannot be excluded.

Figure 9.

Northern blot analysis of AtMHX mRNA. RNA was extracted from various Arabidopsis tissues and subjected to Northern blot analysis as previously described (Shaul et al., 1996). Twenty micrograms of total RNA were loaded on each lane. (A) Hybridization with the AtMHX coding sequence. (B) Methylene blue staining, showing the levels of RNA loaded. I, inflorescence; S, stems; L, leaves; R, roots.

Figure 10.

In situ hybridization of AtMHX mRNA. The analyses were performed in both Arabidopsis and radish, a closely related species, yielding similar results. (A, C, E, G) Pictures taken in a dark field. The light‐colored dots represent the hybridization signal. (B, D, F and H) The same pictures as (A), (C), (E) and (G), respectively, taken in a bright field, showing the toluidine blue staining; the silver‐grain hybridization signal may appear as dark dots in bright field. (A, B) Cross section of a radish root. (C, D) Longitudinal section of a radish stem. (E, F) Longitudinal section of an Arabidopsis silique. (G, H) Longitudinal section of an Arabidopsis leaf. (cr, cortex; e, embryo; ep, epidermis; vc, vascular cylinder; xy, xylem tracheary elements). Bar = 100 μM.


We report here the characterization of an Arabidopsis transporter, AtMHX, which is the first Mg2+ transporter to be cloned from a multicellular organism, and the first cloned exchanger of protons with Mg2+ and Zn2+ ions. Although AtMHX shows limited sequence homology with animal Na+/Ca2+ exchangers, the two transporter species are functionally distinct. In animals, Na+/K+‐ATPases generate Na+ gradients that provide the driving force for most transport processes, including Ca2+ extrusion from the cytosol to the extracellular space by Na+/Ca2+ exchangers localized in the plasma membrane. In plants, the driving force for most transport processes is the electrochemical H+ gradient, which is generated by H+‐ATPases localized in both the plasma membrane and the vacuolar membrane (Maathuis and Sanders, 1992). Notably, in contrast to animal Na+/Ca2+ exchangers, AtMHX is incapable of transporting Ca2+ ions, but can serve as an electrogenic exchanger of protons with Mg2+, Zn2+ and Fe2+ ions. The transport of Mg2+ and Zn2+ in vitro by AtMHX occurs at concentration ranges found in the plant cytosol (i.e. the millimolar range for Mg2+ and the micromolar range for Zn2+) (Marschner, 1995). In contrast to Mg2+ and Zn2+, iron is almost exclusively found in complexes in plant cells, mostly in the chloroplast (Marschner, 1995), and therefore Fe2+ ions are unlikely to be transported in significant amounts by AtMHX in vivo. AtMHX is localized in the vacuolar membrane, and as vacuolar pH is lower compared with the cytosol, we expect that under physiological conditions AtMHX transports divalent cations into the vacuole. Our observations that transgenic tobacco plants ectopically overexpressing AtMHX are specifically sensitive to high Mg2+ or Zn2+ levels support the in vitro transport studies, and indicate that this protein is apparently also capable of transporting Mg2+ and Zn2+ ions in planta. Still, overexpression of the transporter did not alter the total content of magnesium or zinc in shoots of the transgenic plants. Upon growth in media containing elevated Mg2+ or Zn2+ levels, these minerals increased to similar levels in shoots of transformed and non‐transformed plants. Thus, further study is necessary to determine the direct cause of necrotic lesions in leaves of transformed plants grown under such conditions. One possible explanation could be a localized change in the content of Mg2+ or Zn2+ in the necrotic lesions, which occupy a small part of the total shoot area. An alternative assumption is that, being localized in the internal vacuolar membrane, AtMHX does not change the total amounts of Mg2+ or Zn2+ in the cells, but can affect their intracellular distribution. Accordingly, under conditions of high cellular Mg2+ or Zn2+ levels, extensive exchange of these cations with vacuolar protons may impair the cellular pH or ATP balance (the ATP being needed for re‐building the pH balance by H+‐ATPases).

Little is known about the mechanisms regulating Mg2+ and Zn2+ homeostasis in plants. Imbalance in these ions has severe consequences: Mg2+ deficiency impairs the export of carbohydrates from source to sink sites, increases the photosensitivity of leaves and impairs root growth; at the same time, high cytosolic levels of Mg2+ inhibit photosynthesis (Marschner, 1995). Zn2+ deficiency increases RNA degradation and leads to a drastic reduction in the rate of protein synthesis, while Zn2+ toxicity is characterized by chlorosis, and by inhibition in the rate of photosynthesis and root growth (Marschner, 1995). The significant enrichment of AtMHX mRNA at the vascular cylinder of all Arabidopsis organs suggests that this transporter plays an important role in controlling Mg2+ and Zn2+ distribution between the various plant organs. A large proportion of the mRNA is localized in close association with the xylem tracheary elements. As differentiated tracheary elements are dead cells, the expression should reside at the living cell layers that border the xylem tracheary elements, such as the xylem parenchyma. Our working hypothesis for the potential role of AtMHX in the xylem parenchyma cells is related to the function of these cells in loading and unloading of the xylem vessels with minerals and nutrients, and also in reabsorption of certain ions from the xylem sap along the pathway to the shoot (Marschner, 1995). It was shown that the xylem parenchyma (and stem tissue in general) of certain species can reabsorb some minerals (potassium, sodium and nitrate) from the xylem sap in periods of ample root supply, and release these minerals into the xylem sap in periods of insufficient supply (Marschner, 1995). AtMHX may determine the proportion of Mg2+ and Zn2+ ions to be stored in the vacuoles of the xylem parenchyma cells, and, consequently, their amount available in the cytosol during xylem loading and unloading. According to this hypothesis, AtMHX could function in a buffering capacity, sequestering excess amounts, or creating a vacuolar pool that could be used in periods of deficiency.

Materials and methods

Cloning of AtMHX

A first‐strand cDNA prepared from poly(A)+ mRNA purified from A.thaliana cv. C‐24 was employed in a PCR with the following degenerate primers, originally selected for cloning another gene: 5′‐CA(C/T)GA(G/A)AA(G/A)GTICA(G/A)GGIGG‐3′, and 5′‐GCCCA(G/A)TGIA(G/A)IGCIGT(G/A)TG‐3′. The resulting cDNA clone, 735 bp in length, included an open reading frame that showed homology to animal Na+/Ca2+ exchangers. To clone the 5′ and 3′ regions, the 5′ and 3′ RACE systems of Clontech were employed according to the manufacturer's instructions. To obtain the full‐length cDNA, the first‐strand cDNA described above was used as a template for a PCR with the following primers: 5′‐GGGGGAACGCTTGACCGATTC‐3′ and 5′‐CCGGGCCTCCAAAATCATAGT‐3′. One microliter of the product of this reaction was used as template for a second PCR with the following nested primers: 5′‐CCCGTGATCGGCGTATTGTGA‐3′ and 5′‐GCCAACTGCCTTTGAACTTTG‐3′. The PCR product including the full length cDNA was ligated into the pGEM‐5Zf(+) vector (Promega) that was linearized with EcoRV to obtain plasmid p218.

To obtain the genomic clone, total genomic DNA (prepared from A.thaliana cv. C‐24) was used as a template for two successive PCRs using the same primer‐sets used for obtaining the cDNA clone. One microliter of the second PCR was used as template for a third PCR with the following nested primers: 5′‐ATGCCGCTCACCGAGATATT‐3′ and 5′‐TCTTCTACTCATGGGGTTTTTC‐3′.

Preparation of antibodies

Polyclonal antibodies were raised against a synthetic peptide that was derived from AtMHX sequence: Cys Met Ser Arg Gly Asp Arg Pro Glu Glu Trp Val Pro Glu Glu Ile. The peptide was linked through its initial Cys residue to the high‐molecular‐weight KLH carrier (Calbiochem) as previously described (Harlow and Lane, 1988) and injected to rabbits. The antibodies were affinity purified against this peptide using the SulfoLink Coupling Gel (Pierce) according to manufacturer's instructions.

Expression of AtMHX in tobacco BY‐2 cells and tobacco plants

For construction of a chimeric gene for AtMHX expression, plasmid p218 was used as a template for a PCR using an upper primer introducing an NcoI site at the initiating ATG codon, 5′‐GGGGTTTGAATAAGTTACCATGGCCTCAATTCTTA‐3′, and the lower primer 5′‐TCTTCTATATGACGCCTGAAACT‐3′. The PCR product was cloned into plasmid pJD330, 3′ to the 35S promoter (Guilley, 1982) and to the Ω 5′UTR sequence, previously shown to enhance translation of eukaryotic mRNAs (Gallie et al., 1987), and 5′ to the nopaline synthase transcription termination and polyadenylation signals (Holsters et al., 1980). This chimeric gene was cloned into an Agrobacterium binary vector, immobilized into Agrobacterium tumefaciens C58C1(pMP90), and stably transformed into the tobacco BY‐2 cell line (Nagata et al., 1992) as described previously (Shaul et al., 1996). Tobacco Samsun NN plants were transformed with the same Agrobacterium strain as described previously (Horsch et al., 1985).

Growth of transformed tobacco plants in medium supplemented with high levels of various ions and mineral content analysis

F1 seeds of transformed and non‐transformed tobacco plants were surface‐sterilized and germinated in tissue‐culture plates on Nitsch medium (Nitsch, 1969), including kanamycin as a selective agent. Ten‐day‐old seedlings were transferred with their intact roots to 15‐cm diameter plates containing Nitsch medium supplemented with various minerals, as indicated in the text and the legends to 7 and 8. The plants were grown further for 1 month. For mineral content analyses, all the aerial parts were excised from each plant, washed twice in double‐distilled water, dried for 48 h is a 70°C oven, and crushed into a fine powder. For each sample, 120–250 mg dry powder was weighed into 50 ml polypropylene disposable test tubes, 5 ml of concentrated nitric acid were added, and after 10 min at room temperature the samples were digested in the microwave. Analyses were conducted on portions of these solutions. Magnesium and zinc content were determined by inductively coupled plasma atomic emission spectrometry. An ICP‐AES model [Spectroflame] from Spectro (Kleve, Germany) was used.


Seven‐day‐old BY‐2 cells were treated with 1% cellulase YC and 0.1% pectolyase Y‐23 (both from Seishin Pharmaceutical Co., Japan) in 0.4 M mannitol, 10 mM MES, pH 5.5, for ∼1 h at room temperature, and then washed twice with the same solution excluding the enzymes. Cells were then incubated for 2 h in a medium containing 0.12 M mannitol, 20 mM KCl, 2 mM MgCl2 and 10 mM HEPES pH 6.8. Vacuoles, which appear as large clear membrane bubbles, were extruded spontaneously by the cells upon placing them in hypotonic medium of the same composition without mannitol. Giant membrane patches were formed by the methods described previously (Hilgemann, 1995; Hilgemann and Lu, 1998). Both the pipette and the bath solutions contained 100 mM N‐methyl‐d‐glucamine (NMG), 100 mM MES, 10 mM HEPES, 15 mM Tris, and HCl or NMG to adjust pH. The pipette solution (pH 7.0) contained in addition divalent cations, as indicated in the text and Figure 5, and the bath solution contained in addition 0.5 mM Mg2+. The holding potential was 0 mV; membrane potential indicates the bath (intravacuolar) membrane side with respect to the pipette (cytoplasmic) membrane side. Current–voltage relations were determined by applying 10 ms cumulative voltage steps in the hyperolarizing direction, then in the depolarizing direction, and finally back to 0 mV. Records presented are subtractions of the indicated raw current records.

In situ hybridization

In situ hybridization was performed as described previously (Van der Eycken et al., 1996). The radioactive anti‐sense probes included the whole AtMHX cDNA and 3′ non‐coding sequence. Sense probes of the same sequence were used as a control. Photographs were taken with a Diaplan microscope equipped with bright‐ and dark‐field optics (Leitz, Wetzlar, Germany).


We thank R.De Groodt and S.Feng for expert technical assistance, K.D.Philipson and D.A.Nicoll for testing AtMHX mRNA in Xenopus oocytes, Y.Kapulnik, S.Lev‐Yadun, D.Aviv, H.Rahamimoff and S.Schuldiner for helpful discussions, Y.Avivi for critical reading of the manuscript, M.Maeshima for the kind gift of anti‐radish‐VM23 antibodies, M.J.Chrispeels for a kind gift of anti‐RD28 antibodies, D.R.Gallie for a king gift of plasmid pJD330, and the Tobacco Science Research Laboratory, Japan Tobacco, Inc., for permitting us to use the tobacco BY‐2 cell‐line. D.I. is a Research Director of the Institut National de la Recherche Agronomique (France). G.G. is an incumbent of the Bronfman Chair of Plant Sciences.