Elevated expression of TcHMA3 plays a key role in the extreme Cd tolerance in a Cd-hyperaccumulating ecotype of Thlaspi caerulescens


(fax +1 607 255 2459; e-mail lvk1@cornell.edu or +81 86 434 1209; maj@rib.okayama-u.ac.jp).


Cadmium (Cd) is a highly toxic heavy metal for plants, but several unique Cd-hyperaccumulating plant species are able to accumulate this metal to extraordinary concentrations in the aboveground tissues without showing any toxic symptoms. However, the molecular mechanisms underlying this hypertolerance to Cd are poorly understood. Here we have isolated and functionally characterized an allelic gene, TcHMA3 (heavy metal ATPase 3) from two ecotypes (Ganges and Prayon) of Thlaspi caerulescens contrasting in Cd accumulation and tolerance. The TcHMA3 alleles from the higher (Ganges) and lower Cd-accumulating ecotype (Prayon) share 97.8% identity, and encode a P1B-type ATPase. There were no differences in the expression pattern, cell-specificity of protein localization and transport substrate-specificity of TcHMA3 between the two ecotypes. Both alleles were characterized by constitutive expression in the shoot and root, a tonoplast localization of the protein in all leaf cells and specific transport activity for Cd. The only difference between the two ecotypes was the expression level of TcHMA3: Ganges showed a sevenfold higher expression than Prayon, partly caused by a higher copy number. Furthermore, the expression level and localization of TcHMA3 were different from AtHMA3 expression in Arabidopsis. Overexpression of TcHMA3 in Arabidopsis significantly enhanced tolerance to Cd and slightly increased tolerance to Zn, but did not change Co or Pb tolerance. These results indicate that TcHMA3 is a tonoplast-localized transporter highly specific for Cd, which is responsible for sequestration of Cd into the leaf vacuoles, and that a higher expression of this gene is required for Cd hypertolerance in the Cd-hyperaccumulating ecotype of T. caerulescens.


Cadmium (Cd) is a heavy metal that is toxic to all organisms, including plants. Because of the chemical similarity of Cd to certain essential mineral nutrients, including Zn, Fe and Ca, Cd toxicity involves displacement of these essential elements from a number of metalloproteins (Verbruggen et al., 2009). Because of its high affinity for sulfur, Cd also binds to sulfhydryl residues of constituent proteins or enzymes, resulting in the dysfunction of these proteins. However, a few unique plant species, known as Cd hyperaccumulators, are able to accumulate extraordinarily high concentrations of Cd in the aboveground tissues without showing toxicity symptoms. For example, an ecotype of Thlaspi caerulescens, Ganges, from southern France, can accumulate 10 000 mg Cd kg−1 of shoot dry weight (Lombi et al., 2000), in contrast to 0.1–10 mg Cd kg−1 in normal plant species. So far, in addition to T. caerulescens, three other species (Thlaspi praecox, Arabidopsis halleri and Sedum alfredii) have been identified as Cd hyperaccumulators (McGrath et al., 1993; Dahmani-Muller et al., 2000; Yang et al., 2002; Vogel-Mikušet al., 2005).

Several processes leading to the hyperaccumulation of Cd in these species have been suggested to be necessary. These include enhanced Cd uptake from soil into root cells, efficient xylem loading and transport from the roots to the shoots, and effective detoxification of Cd in the aboveground tissues (Milner and Kochian, 2008; Verbruggen et al., 2009; Zhao and McGrath, 2009). Lombi et al. (2001) compared Cd uptake between two ecotypes (Ganges and Prayon) of T. caerulescens contrasting in Cd accumulation, but showing similar Zn accumulation, and found that there was an approximately fivefold larger Vmax for root Cd uptake in Ganges versus Prayon, but that the Km value for root Zn uptake was similar between the two ecotypes. They postulated the existence of a high-affinity Cd transporter involved in the enhanced uptake of Cd into Ganges roots. Pence et al. (2000) identified the first metal transporter (ZNT1) from T. caerulescens (ecotype Prayon) that is a member of the ZIP family (ZRT/IRT-like protein) of micronutrient and heavy metal transporters. However, they found that TcZNT1 was a high-affinity Zn uptake transporter with a low affinity for Cd uptake when expressed in yeast, which cannot explain the high-affinity root Cd uptake phenotype for the Ganges ecotype. A physiological examination of Cd transport in the Ganges and Prayon ecotypes showed that Fe-deficiency treatment resulted in enhanced Cd accumulation in Ganges, but not in Prayon, suggesting that IRT1, the root Fe2+ uptake transporter, may be involved in the high-affinity Cd uptake in Ganges roots (Lombi et al., 2002). However, TcIRT1 isolated from Ganges did not exhibit Cd transport activity in yeast (Plaza et al., 2007). Therefore, the transporter responsible for high-affinity uptake of Cd in the roots of Ganges remains to be identified.

An efficient root-to-shoot translocation of Cd is also required for the hyperaccumulation of this metal. In Arabidopsis, it is clear that HMA2 and HMA4, belonging to the P1B ATPase subfamily, are essential for the root-to-shoot translocation of Zn and Cd (Hussain et al., 2004; Wong and Cobbett, 2009). They are localized to cells of the root pericycle and pump Cd into xylem vessels for transport to the shoot. In the Cd-hyperaccumulating species T. caerulescens and A. halleri, the expression of AtHMA4 homologs (TcHMA4 and AhHMA4) showed higher expression in both the roots and shoots compared with related non-accumulators (Becher et al., 2004; Bernard et al., 2004; Papoyan and Kochian, 2004; Hammond et al., 2006; Talke et al., 2006; Courbot et al., 2007). Recently, Hanikenne et al. (2008) showed that knockdown of AhHMA4 in A. halleri by RNA interference resulted in the decreased accumulation of Zn and Cd in the shoots, with increased accumulation of these metals in the roots. These findings indicate that high HMA4 transcript levels are required for efficient xylem loading of Cd in hyperaccumulating plant species.

Most of the Cd accumulates in the aboveground tissues, and is primarily stored in leaves. Therefore an efficient and high-capacity Cd detoxification system is required in the leaves of Cd-hyperaccumulating species. When AhHMA4 from A. halleri was expressed in Arabidopsis thaliana under the control of the AhHMA4 promoter, the translocation of Cd to the shoots was increased, but the sensitivity of the transformants to Cd was also increased because of the absence of a shoot detoxification mechanism in A. thaliana (Hanikenne et al., 2008). Furthermore, when a non-accumulator, Thlaspi perfoliatum, was grafted onto T. caerulescens root stock, the accumulation of Zn and Cd was enhanced in the leaves, but the leaves showed significant symptoms of Zn toxicity (Cd toxicity was not examined; de Guimarães et al., 2009). Such studies clearly demonstrate that Cd hypertolerance in the shoot is essential for the hyperaccumulation of Cd. In leaves of T. caerulescens, vacuolar sequestration is the main mechanism for Cd/Zn tolerance (Küpper et al., 1999, 2004; Cosio et al., 2004; Ma et al., 2005; Ueno et al., 2005; Ebbs et al., 2009). The transporter encoded by TcMTP1 (ZTP1) has been proposed to be a major contributor to Zn accumulation in the vacuole of cells in the leaf, and is therefore involved in the Zn hyperaccumulation demonstrated by this species (Assunção et al., 2001). TcMTP1 is expressed mainly in the shoots of T. caerulescens and shows a low level of expression in root tissue. It also appears that there may be more than one copy of TcMTP1 in the Thlaspi genome (Assunção et al., 2001). To date no tonoplast-targeted protein(s) have been shown to transport Cd in Thlaspi.

Herein, we report the isolation and functional characterization of a gene (TcHMA3) from the Cd-hyperaccumulating ecotype of T. caerulescens that contributes to Cd sequestration and hypertolerance. We found that this gene encodes a tonoplast-localized transporter specific for Cd, and that high expression of this gene is required for Cd hypertolerance in the Cd-hyperaccumulating ecotype of T. caerulescens, Ganges.


Comparative transcriptome analysis

Ganges and Prayon are two ecotypes of T. caerulescens that show similar Zn hyperaccumulation, but a pronounced difference in Cd accumulation (Lombi et al., 2002; Zhao et al., 2002; Cosio et al., 2004). Under the same growth conditions, Ganges accumulated 7.2-fold Cd higher than Prayon in the shoot (Lombi et al., 2002). The initial objective of this study was to identify transporter genes that were more highly expressed in the Ganges ecotype, to identify candidate Cd transporter genes that play a role in differential Cd accumulation. Therefore, we performed a microarray analysis with roots of Ganges and Prayon plants grown on Cd, using the Affymetrix A. thaliana GeneChip. T. caerulescens and A. thaliana diverged ca. 20 million years ago, and share about 88% nucleotide identity in coding regions (Peer et al., 2003; Van De Mortel et al., 2008). From this microarray study, we found that 28 genes showed higher expression (greater than twofold) in Ganges than in Prayon (Table S1). Among them, we selected TcHMA3 for further study based on current literature suggesting a role for AtHMA3 in Cd transport at the tonoplast (Gravot et al., 2004; Morel et al., 2009).

Isolation and phylogenetic analysis of TcHMA3

The full-length cDNA of T. caerulescens HMA3 (TcHMA3) was first isolated from Ganges with 5′ and 3′ rapid amplification of cDNA ends (RACE), and then isolated from Prayon by amplifying with a primer set designed from 5′ to 3′ untranslated region (UTR) sequences of Ganges TcHMA3. The alleles isolated from Ganges to Prayon, designated as TcHMA3g and TcHMA3p, respectively, encode polypeptides of 756 amino acids in length (Figure S1). TcHMA3g showed 98.7% identity with TcHMA3p (Figures S1 and S2). The homologous genes in A. thaliana and A. halleri, AtHMA3 and AhHMA3, encode polypeptides of 760 and 757 amino acids, respectively (Figure S1). The percentage identity at the amino acid level between TcHMA3g, and AtHMA3 and AhHMA3, was 86.9 and 88.0%, respectively. TcHMA3 belongs to the P1B subfamily of ATPases and contains the predicted eight transmembrane domains found in other subfamily members (Figure S1). The key identifying structures of P1B-type ATPases were fully conserved in the HMA3s isolated from T. caerulescens as well as A. thaliana and A. halleri, suggesting similar function in the three species (Baxter et al., 2003; Williams and Mills, 2005; Figure S2).

Expression patterns of TcHMA3

To compare transcript levels of the TcHMA3s between the two ecotypes, the absolute quantification method was employed based on quantitative real-time RT-PCR. TcHMA3 genes were expressed at similar levels in both the roots and shoots, and the expression showed no significant changes in response to the levels of Zn or Cd in the nutrient solution for either ecotype (Figure 1a). The expression levels were consistently between three- and sixfold higher in the shoots, and between five and sevenfold higher in the roots of Ganges, compared with Prayon. The expression level was further compared in different leaf tissues among Ganges, Prayon and A. thaliana (ecotype, Wassilewskija). Transcription of TcHMA3 was detected in the mesophyll, epidermis and isolated leaf mesophyll protoplasts of both Ganges and Prayon (Figure 1b), but the expression levels in each leaf tissue were again between four and six times higher in Ganges than in Prayon (< 0.05). In the leaf protoplasts, the HMA3 expression level in Ganges was 8270-fold higher than that in A. thaliana (< 0.01; Figure 1b,c).

Figure 1.

 Expression patterns for TcHMA3 in two ecotypes of Thlaspi caerulescens contrasting in Cd accumulation (Ganges, high Cd accumulator; Prayon, low Cd accumulator), determined using quantitative real-time PCR.
(a) Expression level of TcHMA3 in different tissues and the effect of Zn deficiency and high Zn and Cd exposure on expression. Seedlings of both Ganges (Ga) and Prayon (Pr) were grown in the control solution (10 μm Zn), under an excess of Zn (100 μm), under Zn-deficient conditions (0 μm), under low-Cd conditions (1 μm) and under high-Cd conditions (10 μm) for 1 week.
(b, c) HMA3 expression in different leaf tissues of the two T. caerulescens ecotypes (b) and Arabidopsis thaliana (c). Epidermal cells were peeled with tweezers and protoplasts were enzymatically isolated. The absolute expression level of HMA3 in different species was determined with quantitative real-time RT-PCR, and the data were normalized using the CT value.
(d, e) Genomic copy number of HMA3 in two ecotypes of T. caerulescens and A. thaliana normalized with CT values for SHR (d) and two expressed sequence tags (ESTs), RR11nr025 and RR4nr003 (e).
Error bars represent ± SDs (n = 4 for a; n = 3 for b–e). Significant differences from Prayon determined by Tukey’s test: **< 0.01; *< 0.05.

To dissect the factors controlling TcHMA3 mRNA expression between ecotypes, the copy number of TcHMA3 in genomic DNA was estimated with quantitative real-time PCR. When the data were normalized using the cycle threshold (CT) value of SHR (short root; At4g37650), which is a single-copy gene in A. thaliana DNA, there were five times more copies of TcHMA3 in Ganges than in Prayon and A. thaliana (Figure 1d). We also normalized the data with the CT value of two indel markers designed from expressed sequence tags (ESTs) (RR11nr025 and RR4nr003) of T. caerulescens (Deniau et al., 2006). The genes in Arabidopsis for the ESTs RR11nr025 and RR4nr003 correspond to At3g26520 and At3g19820, respectively, which encode tonoplast intrinsic protein and calmodulin protein. Using these standards, Ganges again had between six and seven times more HMA3 copy numbers than Prayon (< 0.01; Figure 1e), based on either marker. These results suggest that one reason for the higher expression level of HMA3 in Ganges is attributed to multiple copy numbers of this gene in the genome.

Tissue and subcellular localization of TcHMA3

The cellular localization of TcHMA3 protein was examined with immunostaining using an antibody against TcHMA3. TcHMA3 was localized to all mesophyll and epidermal cells of the leaves of Ganges (Figure 2a). This localization is consistent with the Cd distribution in the epidermis and mesophyll of Ganges (Ma et al., 2005). In the Ganges roots, TcHMA3 was localized only to pericycle cells (Figure 2b). A similar localization was also observed in Prayon (Figure 2c,d), but the signal was weaker than in Ganges.

Figure 2.

 Localization of TcHMA3 protein in the two ecotypes of T. caerulescens contrasting in Cd accumulation. Immunohistological co-staining with anti-TcHMA3 antibody (red) and 4′,6-diamidino-2-phenylindole (DAPI; cyan) was performed.
(a, b) Leaf (a) and root (b) of Ganges.
(c, d) Leaf (c) and root (d) of Prayon.
Scale bars: 50 μm.

The subcellular localization of TcHMA3 was first investigated by western blot analysis. Fractionation of microsomes isolated from Ganges roots with a sucrose density gradient showed that TcHMA3 was detected in the same fraction as γ-TIP, a tonoplast marker protein (Figure 3a). To confirm this subcellular localization, we isolated vacuoles from leaves, which were determined using both microscopic observation (Figure S3) and cytochrome c oxidase activity, a mitochondrial marker enzyme, to be a very pure vacuolar preparation. The tonoplast fraction prepared from pure isolated vacuoles (Figure S3) yielded a single band that was the size of TcHMA3 in both Prayon and Ganges (Figure 3b), but the abundance was much higher in Ganges than in Prayon. The single band also indicates the high specificity of the TcHMA3 antibody.

Figure 3.

 Subcellular localization of TcHMA3.
(a) Western blot analysis of TcHMA3 in Ganges roots. Membrane fractions were isolated from the Ganges roots and fractionated using a sucrose density gradient. Anti-γ-TIP antibody was used as a tonoplast marker protein.
(b) Western blot analysis of TcHMA3 in the tonoplast fraction. Tonoplast fractions were prepared from pure vacuoles isolated from the leaves of Ganges and Prayon. SDS-PAGE and western blot analyses were performed using the anti-TcHMA3 antibody.
(c–f) Localization of TcHMA3:eGFP fusion protein transiently expressed in Arabidopsis mesophyll protoplasts.
Expression was monitored 12 h after transfection of protoplasts. (c) eGFP signal; (d) autoflorescence of the chloroplast; (e) bright-field image; (f) overlay of (c–f). Scale bars: 100 μm.

The subcellular localization of TcHMA3 was further confirmed by transiently expressing the C- and N-terminal GFP fusion proteins of TcHMA3 in Arabidopsis mesophyll protoplasts and onion epidermal cells. TcHMA3 from both ecotypes was found to be localized to the tonoplast (Figures 3c–f, S4, S5).

Functional analysis of TcHMA3 in yeast and Arabidopsis

The transport specificity of TcHMA3 from both ecotypes was investigated and compared with that of AtHMA3w in yeast. Yeast expressing TcHMA3g, TcHMA3p or AtHMA3w all showed higher Cd uptake than the vector control (Figure 4a), indicating that all proteins encoded by these genes have Cd transport activity. On the other hand, expression of AtHMA3 in yeast resulted in a highly significant (≤ 0.0005) increase in Pb and Zn accumulation (Figure 4b,c). By contrast, the orthologs from the two Thlaspi ecotypes were not effective in transporting these two metals (Figure 4b,c).

Figure 4.

 HMA3 transport activity in yeast.
Yeast cells (DY1457) expressing TcHMA3 from the two ecotypes of Thlaspi caerulescens, AtHMA3 or the empty vector were exposed to a solution containing 10 μm Cd (a), Pb (b) or Zn (c) for 12 h. The concentration of each metal was determined by atomic absorption spectrometry.
Data are means ± SDs (= 5). Significant differences determined using Tukey’s test: **< 0.01; *< 0.05.

To investigate the role of TcHMA3 in planta, we overexpressed TcHMA3g in the Col-0 ecotype of Arabidopsis under the control of the constitutive promoter CaMV35S. AtHMA3 in Col-0 is a pseudogene resulting from a single base-pair deletion (Morel et al., 2009; Wong and Cobbett, 2009). Two independent transgenic lines showed higher expression of TcHMA3g (Figure S6). Root elongation measurements showed that Cd tolerance was enhanced in the overexpression lines (79% inhibition in Col-0 versus 19% in 2-3 or 3-11) (< 0.01; Figure 5a). Within certain TcHMA3g overexpression lines, the concentration of Cd increased by 31% in the shoots and 42% in the roots (Figure 5b,c). The TcHMA3g overexpressing lines also showed a slight increase in Zn tolerance (< 0.01) (Figure 5a); however, there were no differences in tolerance to Co and Pb between the wild-type and the overexpression lines (Figure 5a). No difference was found in the concentration of Zn, Co and Pb in the roots and shoots between overexpression transgenic lines and Col-0 plants (Figure S6).

Figure 5.

 Effect of overexpression of TcHMA3g on the Cd tolerance and accumulation of Cd.
(a) Cd tolerance in transgenic Arabidopsis seedlings overexpressing TcHMA3g.
(b, c) Cd accumulation in the shoots (b) and roots (c) of transgenic lines overexpressing TcHMA3g. Two independent transgenic lines overexpressing TcHMA3g and wild-type Arabidopsis (ecotype Col-0) were grown in nutrient solution containing 10 μm Cd, 40 μm Zn, 20 μm Pb or 15 μm Co for 1 week. The root length was measured before and after exposure. The Cd concentration was determined by atomic absorption spectrometry.
Error bars represent ± SDs (n = 20 for a; n = 5 for b and c). Significant differences from Col-0 determined by Tukey’s test: **< 0.01; *< 0.05.

Comparison of Cd tolerance between Ganges and Prayon

To compare the Cd tolerance between the Ganges and Prayon ecotypes, we grew plants of both ecotypes on a range of Cd concentrations. The Cd concentration in both the roots and shoots of Ganges was much higher than Prayon when grown on 1 or 10 μm Cd for 7 days (Figure 6), which is consistent with previous studies (Lombi et al., 2002; Zhao et al., 2002). When the Cd concentrations were plotted against relative growth, it is clear that Ganges showed a higher tolerance to Cd than Prayon (Figure 6). In Prayon, the relative growth of the roots and shoots in Cd were reduced to 55 and 76% of non-treated Prayon. This led to Cd accumulation of 3.4 and 2.4 mg per gram dry weight in the roots and shoots, respectively. By contrast, the growth of both the roots and shoots in Ganges plants on Cd were not significantly different from untreated plants, even as Cd concentrations in the roots and shoots reached to 8.4 and 7.4 mg per gram dry weight, respectively (Figure 6).

Figure 6.

 Effect of Cd accumulation on the growth of shoots and roots of the two ecotypes of T. caerulescens contrasting in Cd accumulation. Seedlings of Ganges (Ga, circle) and Prayon (Pr, square) ecotypes of Thlaspi caerulescens were exposed to a nutrient solution containing 0, 10 or 50 μm Cd for 14 days. Open symbol, shoot; closed symbol, root. Error bars represent ±SDs (= 4).


Comparative global transcriptomic analysis led to the identification of TcHMA3 as a putative Cd transporter that is much more highly expressed in the higher Cd-hyperaccumulating ecotype (Ganges) of T. caerulescens (Figures 1a,b and S1). Here we showed that TcHMA3 encodes a tonoplast-localized transporter specific for Cd (Figures 3 and S5). Overexpression of TcHMA3 significantly enhanced the tolerance and accumulation of Cd in A. thaliana (Figure 5), indicating that TcHMA3 is involved in the detoxification of Cd by sequestering Cd into the vacuole.

TcHMA3 belongs to the P1B-type ATPase subfamily. Members of this group transport various heavy metal ions from the cytosol out of the cell or to an organelle (Williams and Mills, 2005). In A. thaliana there are eight P1B-type ATPase members, which have been subdivided into two major subgroups (Baxter et al., 2003). The proteins from AtHMA1 to AtHMA4 are classified into the Zn/Cd/Co/Pb transporting subgroup, and from AtHMA5 to AtHMA8 are in the Cu/Ag transporting subgroup, although AtHMA1 has also been shown to transport Zn, Cu and Ca (Axelsen and Palmgren, 1998; Moreno et al., 2008; Kim et al., 2009). TcHMA3 shares 86.9% identity with AtHMA3 at the amino acid level (Figure S1), and both are tonoplast-localized transporters (Figure 3; Morel et al., 2009). However, TcHMA3 and AtHMA3 differ in their expression patterns, cell-specificity of localization and metal-substrate specificity.

Both TcHMA3 and AtHMA3 are constitutively expressed in the roots and shoots (Figure 1a; Morel et al., 2009), but the expression level of the Ganges TcHMA3 homolog is more than 8000 times higher than that seen for AtHMA3 in Arabidopsis (Figure 1b,c). Furthermore, AtHMA3 is mainly expressed in the hydathodes, guard cells and vascular tissue of Arabidopsis leaves, which are located at the end of the transpiration stream (Morel et al., 2009). However, TcHMA3 expression is localized to the epidermal and all mesophyll cells of the leaves (Figure 2), which is the primary location of heavy metal storage in leaves of T. caerulescens (Küpper et al., 2000; Ma et al., 2005). These results indicate that TcHMA3 and AtHMA3 function in different cells and play different roles for dealing with Cd tolerance in the hyperaccumulating and non-accumulating plant species. In Cd-hyperaccumulating ecotypes, Cd transported from roots to shoots via the xylem is transported across the leaf epidermal and mesophyll cell plasma membrane (Ma et al., 2005) via a currently unidentified transporter. From this study, it appears that TcHMA3 may be the major transporter involved in Cd sequestration in leaf cells by transporting the Cd into the leaf cell vacuole. By contrast, in species that do not hyperaccumulate Cd, such as A. thaliana, Cd is transported to the shoot cells such as hydathodes and guard cells along the transpiration stream, probably as a result of an absence of transporters for accumulation in other leaf cells; therefore, AtHMA3 is only expressed in these specific cells.

In the roots, TcHMA3 is predominantly expressed in the pericycle cells (Figure 2b,d), which is similar to that of AtHMA3 (Morel et al., 2009). This may help the plant tolerate the high levels of Cd taken up by the root before it can be transported via the xylem to the aboveground tissues and stored in the vacuole. This is similar to what has been proposed for the role of AtHMA3, which is assumed to act as a ‘filter’ by decreasing the cytoplasmic concentration of toxic metals, which is achieved by sequestering them into the vacuoles of root pericycle cells, thereby limiting these metals to the xylem (Morel et al., 2009). However, in Cd-accumulating species, efficient xylem loading is a hallmark of the ability to accumulate large quantities of various metals, and it is unlikely that TcHMA3 acts as a long-term storage ‘filter’, like AtHMA3. As well as Zn, Cd seems to be remobilized from the vacuole and transported to the aboveground tissues via the xylem (Lasat et al., 1998).

One distinct difference between TcHMA3 and AtHMA3 is the transport substrate specificity. When AtHMA3 was expressed in either yeast or Arabidopsis, it was able to facilitate the transport of Cd, Pb and Zn (Figure 4; Gravot et al., 2004; Morel et al., 2009). By contrast, when TcHMA3 was expressed in yeast, only transport activity for Cd was found (Figure 4), and no transport or accumulation for Pb and Zn could be seen. Similar results were obtained in Arabidopsis overexpressing TcHMA3. Overexpression of TcHMA3 only enhanced the tolerance and accumulation of Cd, but not of Co and Pb, although a slight increase in Zn tolerance was observed (Figure 5 and Figure S6). A homolog of TcHMA3 in A. halleri was reported to function as a Zn transporter, but not as a Cd transporter, based on yeast complementation studies. Although cells expressing AtHMA3 exhibited Zn tolerance, the tolerance was marginal at best and came nowhere close to the growth rates seen by wild-type yeast on media containing high levels of Zn (Becher et al., 2004). Recently, an Oryza sativa (rice) HMA3 (OsHMA3) was identified (Ueno et al., 2010), which is mainly localized at the tonoplast of all root cells and functions to sequester Cd in the vacuoles. OsHMA3 and TcHMA3 only share 39.5% similarity with each other, but OsHMA3 also shows high specificity for Cd. These findings indicate that different homologs of HMA3 have different transport substrate specificities, depending on the species. The transport substrate specificity in the P1B-ATPase subfamily has been suggested to be determined by specific motifs in TM6 [CPC(x)4SxP], TM7 [N(x)7K] and TM8 (DxG) (Williams and Mills, 2005). However, these motifs are identical in the TcHMA3, AtHMA3 and AhHMA3 proteins (Figure S1). This is a functional feature that needs to be examined in the future.

Comparison of allelic HMA3 genes from the two ecotypes contrasting in Cd accumulation showed that they have very high similarity (98.8% identity) (Figure S1), and the same functions in terms of cellular and subcellular localization and metal transport specificity (Figures 1–5). However, there is a specific difference between two homologous genes in their expression levels. Ganges showed a between three- and sixfold higher expression of TcHMA3 than Prayon in both the roots and shoots (Figure 1). This higher expression of HMA3 in T. caerulescens was also found in previous microarray studies (Hammond et al., 2006; Van De Mortel et al., 2008). Higher expression of HMA3 relative to A. thaliana was also observed in another Cd-hyperaccumulating species, A. halleri (Becher et al., 2004; Filatov et al., 2006). These findings indicate that high expression of HMA3 may be required for the detoxification of Cd in Cd-hyperaccumulating species, although the AhHMA3 protein seems not to transport Cd. In fact, at similar high internal Cd concentrations, the growth of Ganges was unaffected, but that of Prayon was reduced, probably as a result of a lower capacity to sequester Cd into the vacuoles (Figure 6).

The expression level of a gene is regulated by a number of factors. Recently, the high expression of AhHMA4 was reported to be attributable to a combination of modified cis-regulatory sequences and the expansion of the gene copy number (Hanikenne et al., 2008). We also found that the copy number of TcHMA3 was between five and seven times greater in Ganges than in Prayon (Figure 2d,e). These results suggest that copy number at least plays part of the role in the high expression regulation of TcHMA3, and this will also be the subject of future research.

Our results indicate that plants use different strategies for the detoxification of Cd and Zn. Vacuolar sequestration of Zn is thought to be mediated by MTP1 (METAL TOLERANCE PROTEIN 1) in both non-Zn accumulating (A. thaliana) and Zn-accumulating (A. halleri) species (Dräger et al., 2004; Krämer, 2005; Willems et al., 2007). MTP1 is a member of the CDF (cation diffusion facilitator) family of proteins, and shows higher expression in the Zn-hyperaccumulating species (Assunção et al., 2001; Becher et al., 2004). In our studies, we did not observe a difference in MTP1 expression between Ganges and Prayon (Table S1). This is not surprising because both ecotypes hyperaccumulate Zn to similar levels.

In conclusion, TcHMA3 is a tonoplast-localized transporter highly specific for Cd. High expression of this transporter is required for sequestering high levels of Cd into the vacuoles of all leaf cells, thereby detoxifying Cd in Cd-hyperaccumulating plants.

Experimental Procedures

Plant materials and growth condition

Ganges and Prayon ecotypes of T. caerulescens J. & C. Presl (Brassicaceae), and Wassilewskija and Columbia ecotypes of A. thaliana were used. Seeds of T. caerulescens were germinated on moistened filter paper in a Petri dish and incubated at 22°C in the dark for a week. After that, the seedlings were transferred to a 3.5-l plastic pot containing an aerated 1/5 strength Hoagland nutrient solution. Seeds of A. thaliana were soaked in deionized water and stored at 4°C. After 2 days, the seeds were germinated on a net floated on a nutrient solution at pH 5.8. The plants were cultured in a temperature-controlled growth chamber at 22°C (light intensity 180 μmol m−2 s−1, 14-h day/10-h night, 60% relative humidity). The nutrient solution was renewed every 2 days during cultivation.

To compare Cd tolerance between ecotypes, seedlings of Ganges (38-days old) and Prayon (30-days old) pre-cultured as described above were exposed to 10 or 50 μm Cd. After 14 days, the roots were washed with deionized water twice, and then the plants were divided into the shoots and the roots. The samples were dried at 70°C for 2 days and weighed. The concentrations of Cd in the samples were determined after digestion with concentrated nitric acid (60%), as described below.

Microarray analysis

To compare the gene expression profiles between Ganges and Prayon, total RNA was extracted from the roots of the five plants exposed to 10 μm Cd for 5 days using an RNeasy plant mini kit (Qiagen, http://www.qiagen.com). The RNA was converted to first- and second-strand cDNA according to the Affymetrix manual. Biotinylated cRNA was transcribed in vitro with the MEGAscript™ T7 kit (Ambion, http://www.ambion.com), and subjected to microarray analysis with an A. thaliana Affymetrix ATH1 GeneChip®, which contains approximately 24 000 genes, following the manufacturer’s protocol (Affymetrix, http://www.affymetrix.com). The obtained data was normalized with genechip v4.0 (Affymetrix) using the default settings. An expression intensity of above 20 was defined as ‘present’. Fold increases in gene expression in the Ganges ecotype in comparison with the Prayon ecotype were selected using gene spring v5.0. Two biological replicates were made.

Cloning of HMA3s

To clone TcHMA3 from each ecotype, total RNA was extracted with an RNeasy plant mini kit (Qiagen, http://www.qiagen.com). The full-length TcHMA3g cDNA was generated by the RACE method (SMART™ RACE cDNA amplification kit; Clontech, http://www.clontech.com). The primers used were designed according to the sequence of AtHMA3 (At4g30120). The open reading frame (ORF) of TcHMA3p cDNA was amplified by RT-PCR using primers 5′-ATCTCCTCGAAGTTTGGTCCGAATAC-3′ and 5′-AGTAACCGAAAATTGTCGAGCGACT-3′, designed according to the sequence of 5′- and 3′-UTRs of TcHMA3g. The ORF of AtHMA3 cDNA was cloned from the Wassilewskija accession of A. thaliana by RT-PCR using primers 5′-AAGATCTAAAAATGGCGGAAGGTG-3′ and 5′-AAGATCTTCACTTTTGTTGATTGTC-3′, designed according to the sequence AY055217. The entire cDNAs were subcloned into the pGEM-T Easy vector (Promega, http://www.promega.com) and sequenced using a Big-Dye sequencing kit (Applied Biosystems, http://www.appliedbiosystems.com) with gene-specific primers on an Applied Biosystems 3130 Genetic Analyzer (Applied Biosystems).

Expression patterns of HMA3

To investigate the effect of Zn and Cd exposure on the expression of TcHMA3, seedlings of Ganges (14-days old) and Prayon (11-days old) were treated with various concentrations of Zn (0, 10 or 100 μm) or Cd (1 or 10 μm) for 7 days. The plants were divided into shoots and roots, frozen with liquid nitrogen and subjected to RNA extraction, as described below.

To examine the expression of HMA3 in different leaf tissue, the leaves of both Ganges and Prayon were divided into epidermal and mesophyll tissues with tweezers. The mesophyll tissue was cut into halves along the midribs, and one half was used for protoplast isolation (Ma et al., 2005). The epidermal tissue, the remaining half of the mesophyll tissue and isolated protoplasts were subjected to RNA extraction as described below. The leaves of A. thaliana were cut into halves along the midribs, and half was used for protoplast isolation. The remaining half and isolated protoplast were subjected to RNA extraction.

Real-time RT-PCR analysis

The total RNA from various tissues were treated with DNase I (Invitrogen, http://www.invitrogen.com), and then converted to cDNA using the protocol attached to SuperScript II (Invitrogen). Expression of the HMA3s were determined by quantitative RT-PCR with the SYBR Green I reagent (SYBR Premix Ex Taq; Takara, http://www.takara-bio.com) on a Prism 7500 real-time PCR System (Applied Biosystems). A 10-ng portion of the first-strand cDNA was used for the template. The primers used for all HMA3s were 5′-TTAAAGCTGGAGAAAGTATACCGA-3′ and 5′-GCTAGAGCTGTAGTTTTCACCT-3′. To generate standard curves for the absolute quantification for each HMA3 copy number, a series of dilutions (from 1 × 10−1 to 1 × 10−6 ng) of plasmids were made and then subjected to real-time PCR. Amplification efficiency was calculated to be 91 (TcHMA3g), 95 (TcHMA3p) and 92% (AtHMA3w). The CT values for each sample were converted into absolute copy numbers using the standard curves. To compare expression between wild-type and overexpression lines of A. thaliana, the data were normalized to the expression level of the Actin gene. The primers used for Actin were 5′-GAGACTTTCAATGCCCCTGC-3′ and 5′-CCATCTCCAGAGTCGAGCACA-3′.

Construction and transient expression analysis of a GFP–TcHMA3g fusion

The TcHMA3 open reading frame from both the Ganges and Prayon ecotypes were cloned into pSAT6, in frame on the C terminus of the HMA3 proteins, and transiently expressed into freshly isolated protoplasts from 3 to 4-week-old Arabidopsis seedlings, according to the method described by Sheen (2001). Confocal images were taken 15–18 h post-transfection of the Arabidopsis protoplasts. Images were taken on a Leica TCS-SP5 confocal microscope (Leica Microsystems, http://www.leica-microsystems.com) with a 63× magnification, numerical aperture 1.2, water immersion objective. The eGFP was excited with the blue argon ion laser (488 nm), and the emitted fluorescence was collected from 505 to 545 nm.

Estimation of HMA3 copy number in genomic DNA

To investigate HMA3 copy number in the genomic DNA of Ganges and Prayon, a DNA fragment of this gene was amplified from genomic DNA. Primers and PCR conditions were the same as described in the section ‘Real-time RT-PCR analysis’ above, except 40 ng of genomic DNA was used as the template, instead of cDNA for each reaction. Genomic DNA of A. thaliana, which has a single copy of HMA3, was used as the control. The data obtained was normalized based on the CT value for SHR (short root; At4g37650), which is a single copy in A. thaliana genomic DNA, or InDel markers designed from two ESTs, RR11nr025 (GenBank DN925409) and RR4nr003 (GenBank DN923929), of T. caerulescens (Deniau et al., 2006). SHR is involved in radial organization of the root and shoot axial organs, and is thus not related to metal homeostasis (Helariutta et al., 2000). The SHR fragment was amplified using primers 5′-CACGTGCGCAACAAATCCTATG-3′ and 5′-TCATGCGGTTGAAGAGAGCTTG-3′. The fragments of two ESTs, RR11nr025 and RR4nr003, were amplified using primers 5′-GTGGTAACATCACTCTCCTCCGTGG-3′ and 5′-AAGCATTTAGCACTCCTACTCCGGC-3′, as well as 5′-GTGGTAACATCACTCTCCTCCGTGG-3′ and 5′-TAGAGAAACAGAGAATCGAAAATAC-3′, respectively (Deniau et al., 2006).

Membrane protein preparation and western blot analysis

Preparation of tonoplast from the leaf protoplasts was performed according to Shimaoka et al. (2004) with some modifications. Briefly, the purified protoplasts were suspended with medium A (30 mm HEPES, 30 mm K-gluconate, 2 mm MgCl2, 2 mm EGTA, adjusted to pH 7.2 with Tris) containing 0.3 m sucrose and incubated on ice for 5 min. A mixture of protoplasts, vacuoles and lysate was mixed with 12 times the volume of medium C (medium A plus 0.5 m sucrose), and a gradient was formed by overlaying medium D (medium A plus 0.3 m sucrose and 0.2 m mannitol) and medium E (medium A plus 0.5 m mannitol). After centrifugation (900 g for 10 min), purified vacuoles were recovered from the interface between medium E and medium D. The purity of isolated vacuoles was checked with microscopic analysis as well as via the determination of cytochrome c oxidase activity, a mitochondrial marker enzyme, according to the method described by Ma et al. (2005). Isolated vacuoles were photographed and then treated with 50 mm KCl and 0.05% deoxycholate to remove surface proteins from the tonoplast. After freezing at −80°C, the sample was thawed at 25°C and ultracentrifuged at 120 000 g for 75 min to yield a pellet of pure tonoplast.

Membrane protein from the roots was prepared according to the method of Sugiyama et al. (2007). Fifteen grams of roots were used for the extraction of membrane proteins. Equal quantities of samples extracted were mixed with the same volume of sample buffer containing 250 mm Tris–HCl, pH 6.8, 8% (w/v) SDS, 40% (w/v) glycerol, 0.01% (w/v) bromophenol blue and 200 mmβ-mercaptoethanol. The mixture was incubated at 65°C for 10 min and SDS-PAGE was run using 5–20% gradient polyacrylamide gels (ATTO, http://www.atto.co.jp). The transfer to polyvinylidene difluoride membrane was performed with a semi-dry blotting system, and the membrane was treated with the purified primary rabbit anti-TcHMA3 (100-times dilution) and anti-γ-TIP (1000-times dilution) polyclonal antibodies. ECL peroxidase-labeled anti-rabbit antibody (10 000-times dilution; GE Healthcare, http://www.gehealthcare.com) was used as a secondary antibody, and an ECL Plus western blotting detection system (GE Healthcare) was used for detection via chemiluminescence.

Immunohistological staining

Antibody against both TcHMA3s was prepared by immunizing rabbits with synthetic peptide C-GLLQKSEETSKKS (positions 732–744 of TcHMA3). The obtained antiserum was purified through a peptide affinity column before use. The roots and rosette leaves of both Ganges and Prayon were used for immunostaining of TcHMA3 protein with a 1:300 dilution, as described previously (Yamaji and Ma, 2007). Fluorescence of secondary antibody (Alexa Fluor 555 goat anti-rabbit IgG; Molecular Probes, now Invitrogen, http://www.invitrogen.com) was observed with a confocal laser scanning microscope (LSM700; Carl Zeiss, http://www.zeiss.com).

Yeast expression analysis

Each ortholog of TcHMA3 was digested from the pGEM easy T vector used to verify the sequence with NotI, and was cloned into a linearized pFL61 opened with NotI. The correct orientation of the insert was verified by PCR using the primer pair 5′-TCTCGCTTCTTGCCATACTATTGCTTTTGA-3′ and 5′-TTAAAATACGCTGAACCCGAACATAGAAAT-3′. The ORF of AtHMA3 was amplified using the primer pair 5′-CAAGCTCAACGATGGCGGAAGGTG-3′ and 5′-CAGAAGAAGGTTTTCACTTTTG-3′, from a plasmid containing AtHMA3, and cloned into pGEM Easy T vector. This vector was digested with NotI, and the fragment obtained was cloned into pFL61 at the NotI restriction site. The primer pair 5′-CAAGCTCAACGATGGCGGAAGGTG-3′ and 5′-TTAAAATACGCTGAACCCGAACATAGAAAT-3′ was used to identify colonies bearing this construct, and sequencing was conducted to confirm that the sequence was inserted in the correct orientation. Saccharomyces cerevisiae strain DY1457 (MATαade6 can1 his3 leu2 trp1 ura3) was cultured on YPD plates. Transformation of this strain with the target construct (pFL61 vector and the pFL61 vector with the gene of interest) was performed using the lithium acetate/single-stranded carrier DNA/PEG method (Gietz and Schiestl, 2007). To obtain cells for transformation, a single colony was streaked on a fresh YPD plate and incubated for approximately 2 days at 30°C. Cells scraped from this plate were suspended in 1.0 ml of sterile deionized water and then pelleted by centrifugation (13 000 g for 30 sec). The supernatant was discarded and the following, in sequence, were layered over the pellet: 240 μl PEG 3350 (50% w/v), 36 μl 1.0 m lithium acetate, 10 μl single-stranded carrier DNA (10 mg ml−1, herring sperm DNA boiled for 5 min to cause denaturation) and plasmid DNA (0.5–1 μg), and sufficient sterile deionized water to provide a final volume of 360 μl. The mixture was vigorously vortexed for up to 1 min and subjected to a heat shock at 42°C for 20 min. For some constructs, the transformation mix was held overnight at room temperature to enhance the transformation efficiency (Gietz and Schiestl, 2007). The transformation mix was then centrifuged at 13 000 g for 30 sec to pellet the cells. After the supernatant was decanted, the cells were resuspended in 1.0 ml of sterile deionized water. Aliquots of the resuspended cells were plated onto a synthetic complete uracil drop-out selection media (referred to hereafter as SC-URA media). Plates were incubated for 3–5 days at 30°C until transformants were observed. Single colonies were picked from each transformant plate and established on fresh SC-URA plates.

For the determination of metal uptake by yeast transformants, single colonies from SC-URA plates were cultured with shaking in liquid SC-URA media at 30°C until the cells reached an OD600 of approximately 1.0. An aliquot of this primary culture was used to establish a subculture with fresh SC-URA media at an OD600 of 0.1. The media was supplemented with CdCl2, ZnSO4 or Pb(CH3COO)2 at a final concentration of 10 μm. Each metal treatment was replicated five times, and each experiment was performed at least twice to confirm the results. The subcultures were cultured with shaking at 30°C for 12 h. The OD600 was recorded and the cell number in the culture media was determined. The cells were pelleted by centrifugation at 3000 g and then washed three times with sterile deionized water. After the final wash the cells were resuspended in approximately 2.0 ml of sterile deionized water and stored at −20°C. The concentration of the metal of interest in the lysate after digest was determined using a 220FS atomic absorption spectrometer (Varian Inc., http://www.chem.agilent.com), and the results were normalized to the cell number. For each metal, the comparison of the means used a one-way anova in spss v13.0 for windows with Tukey’s test for post hoc analysis.

Heterogeneous expression of TcHMA3g in A. thaliana

To generate a construct carrying CaMV 35S promoter, TcHMA3g and NOS terminator, TcHMA3g cDNA fragments were excised from pGEM-T Easy-TcHMA3g plasmid at the BglII site. The fragment was inserted into the BamHI sites of an Agrobacterium-mediated transformation vector pPZP2Ha3(+) (Fuse et al., 2001). Agrobacterium tumefaciens (strain EHA101) was transformed with the construct and used to transform wild-type Col-0 using the floral-dip method (Clough and Bent, 1998). Transgenic T3 lines were selected by hygromicin resistance with plating the seeds on MS medium. To investigate tolerance to heavy metals in transgenic lines, seeds were sown on nets floated on the nutrient solution containing 10 μm Cd, 40 μm Zn, 15 μm Co or 20 μm Pb. After 7 days, the root length was measured with a ruler. To examine heavy metal accumulation in transgenic lines, seedlings (31-days old) were exposed to the nutrient solution containing heavy metals at the same concentration as above, and grown for 7 days. After the roots were rinsed with deionized water for 10 min twice, the shoots and roots were harvested, dried, weighed and subjected to determination of heavy metal concentrations, as described below.

Determination of heavy metal concentrations

Plant samples were dried in an oven at 70°C and then digested with concentrated nitric acid (60%) by heating to 140°C. The concentrations of Cd, Zn, Co and Pb in the digested solutions were determined by flame or flame-less atomic absorption spectrometry (model Z-2000; Hitachi, http://www.hitachi.com).


We thank Pierre Richaud group for kindly giving us the AtHMA3w cDNA. This research was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (no. 21248009 and 22119002 to JFM), by funding from the USDA Agricultural Research Service (to LVK) and by the National Science Foundation (no. 0346276 to SE).