Elevated expression of metal transporter genes in three accessions of the metal hyperaccumulator Thlaspi caerulescens


  • *Present address: Academisch Ziekenhuis Utrecht, Department of Haematology, Heidelberglaan 100, 3508 GA Utrecht.

Correspondence: Dr M.G.M. Aarts. Fax: + 31 317418094; e-mail: m.g.m.aarts@plant.wag-ur.nl


Heavy metal hyperaccumulation in plants is an intriguing and poorly understood phenomenon. Transmembrane metal transporters are assumed to play a key role in this process. We describe the cloning and isolation of three zinc transporter cDNAs from the Zn hyperaccumulator Thlaspi caerulescens. The ZTP1 gene is highly similar to the Arabidopsis ZAT gene. Of the other two, one is most probably an allele of the recently cloned ZNT1 gene from T. caerulescens (Pence et al; Proceedings of the National Academy of Science USA 97, 4956–4960, 2000). The second, called ZNT2, is a close homologue of ZNT1. All three zinc transporter genes show increased expression in T. caerulescens compared with the non-hyperaccumulator congener T. arvense, suggesting an important role in heavy metal hyperaccumulation. ZNT1 and ZNT2 are predominantly expressed in roots and ZTP1 is mainly expressed in leaves but also in roots. In T. arvense, ZNT1 and ZNT2 are exclusively expressed under conditions of Zn deficiency. Their expression in T. caerulescens is barely Zn-responsive, suggesting that Zn hyperaccumulation might rely on a decreased Zn-induced transcriptional downregulation of these genes. ZTP1 expression was higher in plants from calamine soil than in plants from serpentine or normal soil. The calamine plants were also the most Zn tolerant, suggesting that high ZTP1 expression might contribute to Zn tolerance.


A limited number of plant species, called hyperaccumulators, accumulate certain heavy metals to extremely high, normally severely toxic concentrations in their shoots (Ernst 1968; Baker & Brooks 1989; Reeves 1992). Most of these species are more or less restricted to strongly metal-enriched soil types but some of them are also commonly found on non-metalliferous soil (Ingrouille & Smirnoff 1986; Baker & Proctor 1990; Schat, Llugany & Bernhard 1999). There is considerable debate concerning the ultimate evolutionary explanation of the hyperaccumulation trait. The herbivore defence hypothesis, which states that metal hyperaccumulation is a way to reduce damage by herbivory and parasitism (Boyd & Martens 1992) is presently favoured by most authors and supported by circumstantial experimental evidence (Boyd & Martens 1994; Martens & Boyd 1994; Boyd, Shaw & Martens 1994; Pollard & Baker 1997; Davis & Boyd, 2000; Ghaderian, Lyon & Baker 2000).

In comparison with normal plants, hyperaccumulators are characterized by strongly enhanced rates of uptake, tolerance and root-to-shoot transport of the metals in question (Lasat, Baker & Kochian 1996; Schat et al. 1999). The underlying mechanisms and their precise inter-relationships are largely unknown yet. In F2 crosses between the Zn hyperaccumulator Arabidopsis halleri and the non-hyperaccumulator Arabidopsis petraea, the Zn hyperaccumulation and tolerance traits segregated independently (Macnair et al. 1999). Additionally, hyperaccumulation, tolerance and root-to-shoot transport of Zn varied independently between Thlaspi caerulescens accessions from different soil types (Schat et al. 1999). These results suggest that hyperaccumulation might be a complex trait, with uptake, internal transport and tolerance being at least partly under independent genetic control. Transmembrane metal transporters may be decisively involved in uptake, xylem loading and unloading (Lasat, Baker & Kochian 1998) and vacuolar sequestration of heavy metals, particularly in the leaf epidermal cells (Vázquez et al. 1994; Küpper, Zhao & McGrath 1999), trichomes (Krämer et al. 1997), or stomatal guard cells (Heath, Southworth & Dallura 1997). The molecular basis of Zn uptake and transport in plants is, as yet, largely unexplored.

Grotz et al. (1998) have isolated and functionally characterized three Zn transporter genes from Arabidopsis, called ZIP1, ZIP2 and ZIP3 (ZIP: ZRT, IRT-like protein), by functional complementation of a yeast mutant defective in Zn uptake. They also identified a related genomic DNA sequence predicted to encode the ZIP4 protein. ZIP genes belong to a growing family of putative metal transporter genes with members in the fungal, plant and animal kingdom (Grotz et al. 1998; Eng et al. 1998). The proteins encoded by ZIP genes have a high degree of similarity with the yeast ZRT1 and ZRT2 proteins that are involved in the high and low-affinity Zn uptake system (Zhao & Eide 1996a,b), and with the Arabidopsis IRT1 transporter that mediates Fe uptake (Eide et al. 1996). ZIP4 is more closely related to Arabidopsis IRT1 than to ZIP1, ZIP2 or ZIP3 (Eng et al. 1998). ZIP gene expression is Zn-regulated. ZIP1 and ZIP3 are induced in roots and ZIP4 in both roots and shoots of Zn-limited plants. The ZIP1 and ZIP3 proteins, which are presumably plasma membrane located, are suggested to play a role in the uptake of Zn from the rhizosphere, whereas ZIP4, which contains a potential chloroplast targeting sequence, was suggested to mediate transport of Zn into plastids (Grotz et al. 1998).

Recently, using the complementation strategy applied by Grotz et al. (1998), a Zn transporter cDNA was isolated from T. caerulescens by Pence et al. (2000). This transporter gene named ZNT1, is also a member of the ZIP gene family and is highly homologous to Arabidopsis ZIP4 (Grotz et al. 1998). ZNT1 is highly expressed in roots and shoots of T. caerulescens, both under conditions of Zn deficiency and at normal nutritional Zn supply. In the related non-hyperaccumulator species, Thlaspi arvense, it is expressed under Zn deficient conditions, but shows strong downregulation at normal Zn supply. The high expression of Zn transporters in T. caerulescens, irrespective of Zn availability, has been suggested to be the major reason for the enhanced Zn uptake of this species. In general, alterations of the patterns of Zn-responsive transcriptional regulation of Zn transporters might play a pivotal role in Zn hyperaccumulation (Lasat et al. 2000; Pence et al. 2000).

The ZAT gene encodes another Zn transporter known in plants. The ZAT cDNA was isolated from Arabidopsis (Van der Zaal et al. 1999) and is homologous to the mammalian Zn transporter genes ZnT2 (Palmiter, Cole & Findley 1996) and ZnT3 (Wenzel et al. 1997), which are involved in Zn vesicular sequestration, and ZnT4 (Huang & Gitschier 1997), which is involved in Zn transport into milk. Transgenic Arabidopsis which overexpressed the ZAT gene exhibited enhanced Zn resistance and an increased Zn content in roots under high Zn exposure suggesting that the ZAT protein is involved in the plant-internal compartmentation of this metal (Van der Zaal et al. 1999).

The present study aims to identify additional Zn transporters in T. caerulescens and to characterize the variation in metal preference patterns with respect to uptake, root-to-shoot transport and tolerance among T. caerulescens accessions from contrasting soil types (calamine, serpentine and non-metalliferous soil) in relation to the expression of Zn transporter genes. We identified three cDNAs putatively encoding Zn transporter proteins and tried to establish their role in T. caerulescens Zn hyperaccumulation.


Plant material and plant culture

Seeds were collected from three T. caerulescens accessions. Accession La Calamine (LC) originated from a calamine ore waste, enriched in Zn, Cd and Pb, at La Calamine, Belgium. Accession Monte Prinzera (MP) originated from Ni-enriched serpentine soil at Monte Prinzera, Italy and accession Lellingen (LE) originated from a non-metalliferous soil, at Lellingen, Luxembourg. The T. arvense non-hyperaccumulator reference accession originated from a roadside in Amsterdam, the Netherlands. Arabidopsis thaliana, ecotype Columbia, was obtained from the Nottingham Arabidopsis Stock Centre, UK.

To grow plants, seeds of T. caerulescens and T. arvense accessions were sown on moist peat. Three-week-old seedlings were transferred to 600 mL polyethylene pots (one plant per pot), filled with modified half-strength Hoagland's nutrient solution (Schat et al. 1996), supplemented with ZnSO4 and/or NiSO4 at the desired concentrations. The solutions were replaced twice a week. Germination and plant culture were performed in a climate chamber (20/15 °C day/night; 250 μmol m−2 s−1 at plant level, 14 h d−1; 75% relative humidity).

Zn and Ni uptake, root-to-shoot transport and tolerance assays

For the measurement of Zn and Ni uptake, plants of each of the T. caerulescens and T. arvense accessions were exposed to nutrient solution supplemented with 1, 10 and 100 μM of ZnSO4 or 0, 1, 10 and 100 μM of NiSO4, supplied alone or together in factorial combinations. Five plants were used per treatment. After 3 weeks of growth, the plants were harvested, after desorbing the roots systems with ice-cold 5 mM PbNO3 (1 h). Roots and shoots were dried at 80 °C, wet-ashed in a 4 : 1 mixture of HNO3 (65%) and HCl (37%), in Teflon bombs at 140 °C for 7 h and analysed for Zn and Ni, using flame atomic absorption spectrometry (Perkin Elmer 1100B; Perkin Elmes, Norwalk, CT, USA). Total uptake was calculated on a total plant dry weight basis. Shoot-to-root metal concentration ratios were used as an estimate of root-to-shoot transport.

To measure tolerance, plants of each of the T. caerulescens and T. arvense accessions were exposed to nutrient solution supplemented with a series of Zn or Ni concentrations. The tolerance was inferred from the presence or absence of chlorosis after 3 weeks of growth under metal exposure and represented by the first concentration of an increasing series at which chlorosis was observed.

Library construction and screening

A cDNA library was prepared from mRNA extracted from roots of T. caerulescens, accession LC, grown hydroponically in a solution containing 10 μM ZnSO4. The cDNA library with a primary titre of 1·9 × 107 pfu μL−1 was constructed in a phagemid vector (pAD-GAL4–2·1) of the HybriZap-2·1 two-hybrid cDNA cloning system (Stratagene, La Jolla, CA, USA). About 9.5 × 104 plaques of the amplified library were screened. An Arabidopsis ZAT1 (Van der Zaal et al. 1999) partial cDNA clone (EST 143A21, GenBank accession no. 46380; Newman et al. 1994) was used as probe, together with cloned DNA fragments obtained by polymerase chain reaction (PCR) using degenerate primers ZTPfor (5′-TTY GCI GCI GGI GTI ATH CTN GCN AC-3′) and ZTPrev (5′-GCI AGI ARR TCI ACN AGN GCC ATR TA-3′). Degenerate primers were based on conserved DNA sequences in the Arabidopsis ZIP genes (Grotz et al. 1998) and six additional Arabidopsis homologues. Plaques were lifted and blotted onto a nylon membrane (Amersham Pharmacia Biotech, Uppsala, Sweden), according to the recommended procedures and hybridized with the 32P-labelled DNA probes (random-primed DNA labelling system, Amersham Pharmacia Biotech, Uppsala, Sweden). Pre-hybridization and hybridization were performed in hybridization solution [10% dextran sulphate, 1 M NaCl, 1% sodium dodecyl sulphate (SDS)] supplemented with denatured salmon sperm DNA (100 μg mL−1). After an overnight incubation at 65 °C the membranes were rinsed twice in 2 × SSC (300 mM NaCl, 30 mM Na Citrate, pH 7.0) for 2 min at room temperature and twice in 2 × SSC, 1% SDS for 20 min at 65 °C. Positive plaques were purified and the phagemid vector was extracted by in vivo excision according to the instruction provided by the manufacturer.

DNA and predicted amino acid sequence analysis

DNA sequences of both strands of each cDNA were determined by automated sequencing. DNA homology searches and sequence analyses were performed using the Basic Local Alignment Search Tool (BLAST) (Altschul et al. 1997). Multiple sequence alignments were performed by using the CLUSTAL program (DNASTAR, Madison, WI, USA). Potential protein targeting signals were predicted by PSORT (Klein, Kanehisa & DeLisi 1985), potential transmembrane sequences were predicted using TMHMM (Sonnhammer, von Heijne & Krogh 1998).

RNA isolation and RNA blot analysis

Total RNA was extracted using the RNeasy Extraction Kit (Qiagen GmbH, Hilden, Germany) from leaves and roots of T. caerulescens, accessions LC and MP and Arabidopsis ecotype Col (all grown on normal potting soil) and from leaves and roots of hydroponically grown T. arvense and T. caerulescens accessions LC, MP and LE, exposed to 0, 2 and 10 μM Zn. Ten micrograms of total RNA was separated by gel-electrophoresis using a 1% agarose gel and capillary blotted onto Hybond N+ nylon membrane (Amersham), according to standard procedures.

Genomic DNA fragments representing ZIP2, ZIP3 and ZIP4 (obtained by PCR) and the insert of EST clone 143A21 (Acc. no. T46380) representing ZAT1, as well as DNA fragments representing the full cDNAs (ZTP1, ZNT1 and ZNT2) were used as probes for RNA blot hybridization. A PCR fragment of 0·6 kb representing the T. arvense ZTP1 homologue (ZTP1-arvense) was also used as probe for RNA blot hybridization. The forward primer used in this PCR reaction was 5′-GGC AGA CTT ACG GGT TCT TCA GG-3′ and the reverse primer was 5′-GTG AGA ACG GAA AAG CCA ATA GC-3′. Membranes were pre-hybridized, hybridized and washed as described above except that additional washes were performed at high stringency (in 0·1 × SSC, 1% SDS for 20 min at 65 °C). Membranes were stripped using a solution of 0·1% SDS, 2 mM Tris-HCl (pH 8·0), 1 mM EDTA for 10 min at 65 °C. The membranes were checked for removal of the probe before re-probing. The hybridization signals were scanned using a Fuji Phosphor Imager (BAS 2000; Fuji Photo Film Co., Tokyo, Japan) and, if necessary, quantified using the TINA programme provided by the manufacturer.

DNA isolation and DNA blot analysis

DNA was extracted from leaves of T. arvense, T. caerulescens, accessions LC, MP and LE and Arabidopsis as described by Aarts, Koncz & Pereira (2000). One microgram of Thlaspi and 0·5 μg of Arabidopsis genomic DNA, separately digested with HindIII, was separated by gel-electrophoresis using a 1% agarose-TAE (Tris-Acetate 40 mM, EDTA 1 mM) gel and vacuum blotted onto Hybond N+ nylon membrane (Amersham) according to standard procedures. DNA blot hybridization was performed as the RNA blot hybridization described before.


Physiological characterization of T. caerulescens accessions

The Zn and Ni uptake, shoot-to-root concentration ratio and tolerance characteristics were determined for hydroponically grown T. caerulescens and non-hyperaccumulator T. arvense plants (Tables 1 and 2). In comparison with T. arvense, the T. caerulescens accessions La Calamine (LC), Monte Prinzera (MP) and Lellingen (LE) were all characterized by a strongly enhanced Zn uptake, Zn shoot-to-root concentration ratio and Zn tolerance. However, there were pronounced differences between these T. caerulescens accessions. As far as tolerance is concerned, the LC plants maintained normal growth and leaf pigmentation at 1000 μM Zn in the nutrient solution. The MP and LE plants already showed chlorosis at 100 and 50 μM Zn, respectively (Table 1). The analysis of Zn uptake showed that LC plants accumulated significantly less Zn (P < 0·01) than MP and LE plants, both at low and high supply levels (Table 2a.). Root-to-shoot transport of Zn, as estimated from the ratio between shoot and root Zn concentration, was significantly higher (P < 0·01) in all T. caerulescens accessions than in T. arvense and, among accessions, higher in LE than in LC and MP plants (Table 2c).

Table 1.  Tolerance to Zn and Ni of Thlaspi arvense and T. caerulescens plants. Tolerance was inferred from the presence or absence of chlorosis after 3 weeks of growth under metal exposure. The figures represent the first concentration of an increasing series at which chlorosis was observed
 Zn (μM)Ni (μM)
  1. T. caerulescens LE

  2. 50

  3. 100

T. arvense< 25< 25
T. caerulescens LC1000100
T. caerulescens MP100250
Table 2.  Uptake (a), shoot concentration (b), and shoot-to-root concentration ratio (c) of Zn and Ni in the non-hyperaccumulator Thlaspi arvense (Ta) and accessions (acc.) LC, MP and LE of the hyperaccumulator T. caerulescens (Tc), after 3 weeks of growth in a nutrient solution with factorial combinations of Zn and Ni concentrations. Uptake is expressed on a whole plant dry weight basis (μmol metal g−1 total plant dry weight), shoot concentration is given as μmol metal g−1 shoot dry weight. nt = not tested. Standard errors varied between 3 and 20% of the means
  Zn supply (μM)
Species (acc.)Ni supply (μM)110100110100
(a) Uptake of Zn and Ni
  Zn uptakeNi uptake
Tc (LC)04·68·135·7ntntnt
Tc (MP)013·253·9118·7ntntnt
Tc (LE)06·942·1ntntntnt
(b) Shoot concentration of Zn and Ni
  Zn shoot concentrationNi shoot concentration
Tc (LC)05·64·442·2ntntnt
Tc (MP)014·151·598·6ntntnt
Tc (LE)08·552·5ntntntnt
(c) Shoot-to-root concentration ratio of Zn and Ni
  Zn shoot-to-root ratioNi shoot-to-root ratio
  1. 100

  2. nt

  3. nt

  4. nt

  5. nt

  6. nt

  7. nt

Tc (LC)07·14·14·5ntntnt
Tc (MP)01·50·80·5ntntnt
Tc (LE)025·062·5ntntntnt

The Ni tolerance was higher in T. caerulescens than in T. arvense, particularly in the MP plants originating from serpentine soils (Table 1). The Ni accumulation varied strongly between the accessions (P < 0·01). The MP plants and, to a lower degree also the LE plants, hyperaccumulated Ni, although only at low external Zn concentrations (1 μM). Higher Zn concentrations strongly inhibited Ni uptake whereas Ni had no effect on Zn uptake (Table 2a). The LC plants did not show any Ni hyperaccumulation at all, irrespective of the concentration of Zn in the nutrient solution (Table 2a). Remarkably, there was no inhibitory effect of Zn on Ni uptake in these plants. Transport of Ni was consistently higher in T. caerulescens than in T. arvense with the highest shoot-to-root concentration ratios for LE and the lowest for LC. The MP and LE plants exhibited a pronounced inhibitory effect of Ni on Zn transport and vice versa. This was not apparent in LC plants (Table 2c).

Identification of Zn transporter genes in T. caerulescens

To assess whether any of the known Arabidopsis Zn transporter genes might be differentially expressed in T. caerulescens, Northern blot experiments were performed using genomic DNA fragments representing the ZIP2, ZIP3 and ZIP4 genes and a partial cDNA clone representing ZAT1 as probes. The blots contained RNA from leaves and roots of T. caerulescens accessions LC and MP and Arabidopsis ecotype Col, all grown on normal potting soil. ZAT1 was clearly overexpressed in T. caerulescens leaves compared to roots, or to Arabidopsis leaves and roots (Fig. 1). Transcription of ZIP2 and ZIP3 was not detected in any of the samples, but weak ZIP4 transcription was found for all samples (data not shown). The ZIP2 and ZIP3 probes hybridized strongly to a T. caerulescens DNA-blot (data not shown), proving that the inability to detect ZIP2 and ZIP3 transcripts was not due to lack of homology.

Figure 1.

Northern blot analysis of Arabidopsis ZAT gene expression in leaves and roots of T. caerulescens, accessions LC and MP (TcLC and TcMP) and Arabidopsis thaliana ecotype Columbia (AtCol), grown in normal potting soil. The second row represents the hybridization of the blot with 16S rRNA used as a loading control. The blot was washed under low-stringency conditions.

In order to obtain Thlaspi-specific probes for ZIP homologues, a degenerate PCR approach was chosen. From database searches it became apparent that in Arabidopsis at least 10 ZIP-like gene sequences are present, including the known ZIP and IRT genes. Degenerate primers were designed on two conserved DNA sequences found within this gene family (see ‘Materials and Methods’). In total, PCR fragments for five different ZIP-like genes were isolated from T. caerulescens accession LC, which were homologous to Arabidopsis genes ZIP1, ZIP3 and ZIP4 (data not shown).

Isolation of T. caerulescens Zn transporter cDNAs

The Arabidopsis partial ZAT1 cDNA insert and DNA fragments L13 and L3, homologous to, respectively, ZIP1 and ZIP4, were used as probes to screen a T. caerulescens cDNA library. The library was made with root RNA from LC plants grown hydroponically on medium containing 10 μM Zn. With probe L13 no positive clones were found. With each of the other two probes about 40 positive clones were obtained. After sequencing the 5′ ends of the longest inserts, three different full-length cDNA sequences were identified, one from a ZAT1-specific clone and two from L3-specific clones. The complete DNA sequence for each full-length clone was determined and compared to sequences in the GenBank database. The cDNA fragment obtained with the Arabidopsis ZAT1 partial cDNA probe, is 1340 base pairs (bp) long and encodes a predicted protein of 393 amino acids. The cDNA has 90% DNA identity and 75% amino acid identity with Arabidopsis ZAT1 cDNA (Van der Zaal et al. 1999). We called this gene ZTP1 for Zn TransPorter 1 (Fig. 2). The ZTP1 protein is predicted to be an integral membrane protein with six potential transmembrane domains (Fig. 2), just like the Arabidopsis ZAT protein of which it is most likely the T. caerulescens orthologue. The subcellular location of ZTP1 is unclear. Although previous studies with the Arabidopsis ZAT gene suggested targeting of the protein to the vacuole (Van der Zaal et al. 1999), an obvious vacuolar targeting signal was not detected.

Figure 2.

Amino acid sequence alignment of the T. caerulescens ZTP1 zinc transporter with the A. thaliana ZAT zinc transporter (GenBank accession no. AF072858). The sequences were aligned using the CLUSTAL method (DNAstar). The putative transmembrane domains, predicted by TMHMM (Sonnhammer et al. 1998), are overlined and numbered. Identical residues are shaded.

The cDNA clones obtained with the L3 probe are 1375 and 1520 bp long and encode predicted proteins of 409 and 423 amino acids, respectively. They share 90 and 83% DNA identity and 76 and 65% amino acid identity with the predicted Arabidopsis ZIP4 DNA and protein sequences (Grotz et al. 1998). By database search we found the first clone to be nearly identical (99% DNA identity) to the ZNT1 Zn transporter gene recently cloned from T. caerulescens accession Prayon (ZNT1-PR;Pence et al. 2000) and we believe it to be the LC allele of ZNT1 (ZNT1-LC). The second clone, which is a close homologue of ZNT1-LC with 80% DNA identity, was labelled ZNT2 (Fig. 3). One recently deposited EST sequence representing the ZIP4 gene was found in the GenBank database (Acc. No. AV441840), which starts at nearly the same position as the ZNT1-LC cDNA clone and downstream of the 5′ end of the ZNT2 cDNA. After comparison of this EST sequence to the genomic DNA sequence of ZIP4 in Arabidopsis, we detected an in-frame stop codon only 3 bases upstream of the EST 5′ end. As no putative intron–exon boundary sequences were found between the stop codon and the EST sequence start, we concluded that the predicted start codon of ZIP4 corresponds to the first ATG codon we identified in both the ZNT1-LC and the ZNT2 cDNA clones, which is thus most likely the protein translation start codon. With this new ZIP4 cDNA sequence, the predicted ZIP4 protein sequence is extended at the N-terminus by 34 amino acids compared with earlier publications (Grotz et al. 1998; Pence et al. 2000). This start codon is not present in the ZNT1-PR cDNA sequence reported by Pence et al. (2000) and consequently the predicted amino acid sequence of ZNT1-LC is 30 amino acids longer at its N-terminus compared to ZNT1-PR. The ZNT1-LC and ZNT2 proteins are predicted to have a N-terminal signal sequence (Fig. 3), as well as the eight potential transmembrane domains previously reported for ZNT1-PR by Pence et al. (2000). ZNT1-LC and ZNT2 are likely to be targeted to the plasma membrane.

Figure 3.

Amino acid sequence alignment of ZNT1 (LC) and ZNT2 with ZNT1 (Prayon) (GenBank accession no. AF133267) and ZIP4, predicted from genomic and partial cDNA sequences (GenBank accessions no. ATU95973 and AV441840). The sequences were aligned using the CLUSTAL method included in the DNAstar programme (DNAstar). The putative transmembrane domains, predicted by TMHMM (Sonnhammer et al. 1998), are numbered and overlined. Identical residues are shaded.

Expression of Zn transporter genes in different T. caerulescens accessions and in T. arvense

To determine the expression of the identified Zn transporter genes, low stringency hybridizations were performed using the ZTP1, ZNT1-LC and ZNT2 cDNA inserts as probes on blots containing RNA from the three different T. caerulescens accessions and the non-hyperaccumulator T. arvense. Compared to T. arvense, ZTP1 is higher expressed in both roots and leaves of T. caerulescens accessions LC, MP and LE, grown hydroponically at 0, 2 and 10 μM Zn (Fig. 4). The expression in T. caerulescens LC was slightly higher than in the other two T. caerulescens accessions. In MP and LE the difference in expression between roots and leaves was more pronounced, with enhanced mRNA levels in leaves. To confirm that the overexpression of ZTP1 in LC was not due to preferential hybridization of the LC probe to the LC RNA, an additional hybridization was performed using a 0·6 kb PCR-fragment of the ZTP1 homologue from T. arvense as a probe (ZTP1-arvense). The T. arvense probe preferentially hybridized to T. arvense RNA compared to T. caerulescens RNA. However, as with the ZTP1 probe, the ZTP1-arvense probe hybridized stronger to T. caerulescens LC RNA than to T. arvense RNA. After image analysis, we estimated that the ZTP1 mRNA hybridization signal was about five times higher in LC than in T. arvense.

Figure 4.

Northern blot analysis of ZTP1 (A) and ZTP1-arvense (B) expression in leaves and roots of T. arvense (Ta) and T. caerulescens, accessions LC, MP and LE (TcLC, TcMP and TcLE), grown at 0, 2 and 10 μM Zn. Approximately equal loading of the RNAs in blots A and B is shown in C and D, respectively, after hybridization with a 16S rRNA-specific probe. The blots were washed under low-stringency conditions.

The ZNT1 expression levels in root and leaf of all accessions are very similar to the ZNT2 expression levels. In comparison with T. arvense, both genes are highly expressed in all three T. caerulescens accessions, at all tested Zn concentrations (Fig. 5). The expression is higher in roots than in leaves, which is most pronounced for ZNT1. The T. arvense ZNT1 transcript was only detected in roots and leaves of plants grown at 0 μM Zn. Also ZNT2 is expressed only at 0 μM Zn, although it is barely detectable after hybridization (Fig. 5). Under these conditions, and especially in roots, the expression is much lower in T. arvense than in T. caerulescens. To establish whether ZNT1 is downregulated by high zinc concentrations, we hybridized a RNA-blot, containing root and leaf RNA of LC plants sampled in a 48 h time period after transfer from 0 μM Zn to 1 mM Zn, with a ZNT1-LC-specific probe (Fig. 6). There was no apparent reduction in hybridization signal in this 48 h period, either in roots or in shoots. To determine the level of cross hybridization of the T. caerulescens LC probes to MP, LE and T. arvense DNA and RNA, low and high stringency DNA blot hybridizations were performed. Under low stringency conditions, the ZTP1 probe detected the corresponding ZTP1 genomic fragment as well as one cross-hybridizing ZTP1-homologous sequence, present in both Thlaspi species and in Arabidopsis (Fig. 7). Upon high stringency washing only the ZTP1-containing fragments were observed. The Southern analysis with ZNT1 and ZNT2 probes showed the respective homologues in T. arvense, Arabidopsis and all T. caerulescens accessions. Under reduced stringency conditions each of the two probes cross-hybridized to their respective DNA fragments, but not to any additional DNA fragments (Fig. 7), suggesting that these two genes are the only closely related ZIP4 homologues present in T. caerulescens and T. arvense.

Figure 5.

Northern blot analysis of ZNT1 (A) and ZNT2 (B) expression in leaves and roots of T. arvense (Ta) and T. caerulescens accessions LC, MP and LE (TcLC, TcMP and TcLE), grown at 0, 2 and 10 μM Zn. Approximately equal loading of the RNAs in blots A and B is shown in C and D, respectively, after hybridization with a 16S rRNA-specific probe. The blots were washed under low-stringency conditions.

Figure 6.

Northern blot analysis of ZNT1 expression in roots (R) and leaves (L) of T. caerulescens accessions LC grown at 1 mM Zn for a 48 h period. The blot was washed under low-stringency conditions.

Figure 7.

Southern blot analysis of ZTP1 (A), ZNT1 (B) and ZNT2 (C).Genomic DNA from T. arvense (Ta), T. caerulescens accessions LC (TcLC), MP (TcMP), LE (TcLE) and A. thaliana ecotype Columbia (At) was digested with HindIII. The figure represents low-stringency (A and C) and high-stringency (B) washings. Under low stringency washing, the ZTP1 (A) and ZNT2 (C) probes cross-hybridize to only one other homologous gene copy (*). For the ZNT2 probe, this is the ZNT1 gene, as the weaker hybridization signals in (C) are the very strong hybridizing signals in (B).


The physiological analysis of three Zn hyperaccumulator T. caerulescens accessions and a non-hyperaccumulator T. arvense accession confirms that the Zn hyperaccumulator species is characterized by a much higher Zn uptake, shoot-to-root concentration ratio and tolerance than the related non-hyperaccumulator species. Additionally, we have observed a high and independent inter-accession variability for these physiological properties in T. caerulescens.

The metallicolous T. caerulescens accessions LC and MP are specifically adapted to their native soils, since the observed metal tolerance characteristics correspond well with the soil metal composition at the sites of seed collection. The LC and MP plants exhibited elevated tolerance to, respectively, Zn and Ni. The high level of Zn tolerance in LC plants is associated with decreased uptake and transport of this metal, compared to the non-metallicolous accession LE. Reduced Zn accumulation and transport in T. caerulescens and Arabidopsis halleri accessions from calamine soil, as compared to accessions from non-metalliferous soil, has been reported previously (Meerts & Van Isacker 1997; Bert et al. 2000; Escarréet al. 2000), indicating that Zn tolerance and accumulation in these species are independent traits. Schat et al. (1999) however, reported a very low Zn and Cd shoot-to-root concentration ratio in a non-metallicolous accession originating from another site near Lellingen, about 4 km distant from the site of origin of the accession LE used in the present study. More extensive comparisons of non-metallicolous and metallicolous accessions have shown that the low transport in the former accession from Lellingen is probably highly exceptional (data not shown).

In contrast to the observations for Zn uptake by the Zn-tolerant LC accession, the highly Ni-tolerant MP plants showed an increased rather than decreased uptake of Ni and Zn, compared to the non-metallicolous accession LE. This indicates once more that metal tolerance and metal uptake are independent traits.

The observed metal uptake, transport and tolerance characteristics suggest an important role for transmembrane metal transporters in the metal hyperaccumulation mechanism. Thorough genetic and physiological analysis of the inter-accession variability in respect to the expression of T. caerulescens Zn transporter genes could give more insight into the function of this genes and the mechanism of metal hyperaccumulation. In this work it was shown that T. caerulescens contains at least three different expressed genes with strong homology to Zn transporters. ZTP1 is the closest homologue of the Arabidopsis ZAT Zn transporter gene and most likely the Thlaspi orthologue. Based on Southern analysis there appears to be one other ZAT/ZTP1 homologous DNA sequence present in both Thlaspi and Arabidopsis. A corresponding cDNA has not been found and expression has thus not been tested. Overexpressing the ZAT gene in Arabidopsis caused an increased Zn content in roots as well as enhanced Zn tolerance. This suggests that the ZAT protein is involved in the internal compartmentation of this metal (Van der Zaal et al. 1999). Based on the Northern analysis, we conclude that ZTP1 is clearly overexpressed in T. caerulescens compared to T. arvense, with the highest expression in LC. LC is also the accession with the highest tolerance to Zn. Together with its predominant expression in the leaves this emphasizes the proposed role for ZTP1/ZAT-like transporters in Zn compartmentation and also suggests an important contribution of ZTP1 expression to Zn tolerance.

ZNT1 and ZNT2 resemble the Arabidopsis ZIP4 gene, suggested to encode a Zn transporter (Grotz et al. 1998). Experimental evidence supporting this view was recently provided by Pence et al. (2000), also mentioned by Lasat et al. (2000), who described the cloning of a ZNT1 partial cDNA from T. caerulescens accession Prayon, by functional complementation of a Zn uptake deficient yeast mutant. We have been unable to show any functional complementation using the same zinc uptake-deficient yeast mutant zhy3 (Zhao & Eide 1996b) transformed with overexpression constructs containing either the ZTP1, ZNT1 or ZNT2 cDNA sequences. It may be that the presence of the plant N-terminal signal sequence interfered with proper intracellular localization of the heterologous protein. The presence of an N-terminal signal sequence may be also the reason that the ZNT2 cDNA was not identified in the functional complementation experiment that yielded the ZNT1 cDNA (Pence et al. 2000; Lasat et al. 2000), although the ZNT2 mRNA is not considerably less abundant than the ZNT1 mRNA. On the other hand, the presence of a signal sequence did not disturb the functional complementation of the yeast mutant by the Arabidopsis ZIP1, ZIP2 and ZIP3 genes (Grotz et al. 1998), which belong to the same ZIP-like gene family, although not as closely related to ZNT1 and ZNT2 as ZIP4. As for ZNT2 in T. caerulescens, neither a full-length nor a partial ZIP4 cDNA clone was picked up from the Arabidopsis seedling cDNA-expression library that yielded the ZIP1, ZIP2 and ZIP3 cDNAs (Grotz et al. 1998), although its transcript should not be less abundant.

Initially the ZIP4 protein was predicted to contain a potential chloroplast targeting sequence. Based on the recently deposited ZIP4 partial cDNA sequence (acc. no. AV441840) the predicted protein contains an N-terminal signal sequence and is most likely targeted to the plasma membrane (PSORT, Klein et al. 1985). This is more in accordance with expression in both roots and shoots.

The ZNT1 and ZNT2 genes are clearly part of the ZIP-like gene family. First of all they encode proteins with eight predicted transmembrane domains, as found in the Arabidopsis ZIP4 protein and other related ZIP proteins. The predicted proteins contain a histidine-rich region between transmembrane domains III and IV (Fig. 3), which is proposed to be the heavy metal binding sequence (Eng et al. 1998). Finally, the proteins contain the conserved ZIP signature sequence in transmembrane domain IV. This sequence is fully conserved among all members of the ZIP family (Grotz et al. 1998; Eng et al. 1998) and suggested to play a role in substrate transport over a membrane.

In comparison with T. arvense, ZNT1 and ZNT2 are highly expressed in at least four, respectively, three, T. caerulescens accessions. The high expression concerns predominantly root tissue, which is a strong indication of a role for these genes in enhanced Zn uptake from the soil. This is in line with the strongly enhanced Zn uptake in T. caerulescens compared with T. arvense (Table 2). The differences in ZNT1/ZNT2 expression among T. caerulescens accessions are less pronounced and also much less associated with differences in Zn uptake between these accessions. It is clear that expression of ZNT1 and ZNT2 does not account for all of the observed inter-accession differences in Zn uptake, transport to the shoots, or in Zn tolerance. Of course it would be very unlikely that zinc uptake is solely controlled by the expression of two zinc transporter genes. Additional zinc transporters, like the non-metal-specific Nramp transporters (Thomine et al. 2000) or other ZIP-like proteins are also likely to be involved in zinc uptake. Alternatively, zinc transporter regulation may act on the post-transcriptional level as was reported for the yeast ZRT1 zinc transporter (Gitan et al. 1998; Gitan & Eide, 2000).

The Ni hyperaccumulation in MP and, to a lower degree, in LE and the complete absence of this phenomenon in LC, suggests the presence of at least two different uptake systems involved in Zn/Ni hyperaccumulation in T. caerulescens: a Zn-specific high-affinity uptake system and a low-affinity Zn/Ni uptake system, which prefers Zn over Ni. The latter seems to be suppressed in the LC accession, but overexpressed in the MP accession. The relatively small difference in ZNT1 and ZNT2 expression between the different accessions suggests that other genes are responsible for the enhanced Ni accumulation in MP and LE.

All three Zn transporters are highly expressed in T. caerulescens at all Zn concentrations tested. The highest concentration of 10 μM approaches the Zn concentration which is available as water-soluble Zn in heavily Zn-contaminated soils (Ernst & Nelissen, 2000). Pence et al. (2000) observed that ZNT1 expression is downregulated only after prolonged exposure to 50 μM Zn, whereas in T. arvense it is downregulated at 1 μM. In addition, we observed that the transcriptional downregulation of ZNT1 was not yet obvious after 48 h exposure to 1 mM Zn. The apparent decrease in Zn-imposed downregulation of zinc transporter genes in T. caerulescens has been conserved among at least four accessions and may well be the first evolutionary step that led to the ability to accumulate and tolerate high Zn levels in this hyperaccumulator species.


We thank Martijn Fiers for making the cDNA library, Richard Immink and Marco Busscher for their support in performing the automatic sequence reactions, Dr David Eide for providing the zinc uptake-deficient yeast strain, Dr Arle Kruckeberg for providing plasmids and technical support and Paul Koevoets and Professor Dr Wilfried Ernst for critical reading of the manuscript.

Part of this work was supported by the Portuguese Foundation for Science and Technology, programme PRAXIS XXI (grant no. BD/16152/98).