Zinc-Dependent Global Transcriptional Control, Transcriptional Deregulation, and Higher Gene Copy Number for Genes in Metal Homeostasis of the Hyperaccumulator Arabidopsis halleri

Establishment of Experimental Conditions

To compare the effect of external Zn supply on Zn accumulation and partitioning in A. thaliana and A. halleri, plants were cultivated hydroponically and supplied with different subtoxic Zn concentrations for 3 weeks (Fig. 1, A and B ). In A. halleri, the ratio of shoot Zn concentrations to root Zn concentrations was above unity in plants cultivated in a Zn-deficient hydroponic solution containing no added Zn for 3 weeks (Fig. 1C). Continuous supply of Zn concentrations of between 1 and 5 μm led to an increase in the shoot-to-root ratio of Zn concentrations, with maximum ratios approximately tripling the ratio determined under Zn-deficient conditions. By contrast, A. thaliana partitioned into the shoots only a small proportion of the Zn concentration accumulated in the roots (Fig. 1, A and B). This was reflected in shoot-to-root Zn concentration ratios substantially below unity, which decreased further with increasing Zn supply (Fig. 1C). These results indicated that A. halleri accumulates Zn predominantly in the shoot even at very low Zn supply, whereas A. thaliana primarily immobilizes Zn in the roots irrespective of the Zn supply. When A. halleri was cultivated in media containing high Zn concentrations of 30 or 300 μm Zn, shoot-to-root ratios of Zn concentrations were below unity and only about one-quarter of the ratios measured at 0.3 μm Zn. In addition to the dramatic interspecies differences in Zn partitioning between roots and shoots, the data suggest a species-specific modulation of Zn partitioning dependent on the Zn supply.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Long-term Zn accumulation in A. halleri and A. thaliana roots and shoots. Hydroponically grown 4.5-week-old plants were supplied with different Zn concentrations in the culture medium for 3 weeks before harvest. The Zn concentrations in root (A) and shoot (B) tissues were determined by inductively coupled plasma optical emission spectroscopy. White circles, A. thaliana; black circles: A. halleri. C illustrates the ratio of Zn concentrations in shoot versus root tissues. Note that at 10 μm ZnSO₄, the Zn concentration in A. thaliana roots exceeds that of A. halleri roots by more than 30-fold. In preliminary experiments, exposure of hydroponically grown A. thaliana to 30 μm ZnSO₄ for a period of 3 weeks led to the onset of mild toxicity symptoms; therefore, this Zn condition was excluded. Data are from one experiment representative of two independent experiments. Values represent mean ± se, calculated from n = 3 culture vessels per Zn condition. For C, data were log₂ transformed. Data were plotted with Origin-Pro 7 SR4 (v 7.0552; OriginLab Corporation) and logarithmic fits to the data were performed for A and B, a cubic B-spline connection was used for the smooth curves in C.

To select suitable conditions and time points for transcript profiling, transcript levels of three metal homeostasis genes, ZIP4, ZIP9, and NAS2, were determined in preliminary experiments (Fig. 2 ). These genes are known to be Zn responsive at the transcript level in A. thaliana (Grotz et al., 1998; Wintz et al., 2003) and they have previously been identified as candidate genes for Zn accumulation and tolerance in A. halleri (Becher et al., 2004; Weber et al., 2004).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Expression analysis of known Zn-regulated marker genes. Real-time RT-PCR analysis of expression of ZIP4, ZIP9, and NAS2 genes in roots (top section) and shoots (bottom section) of hydroponically grown A. halleri (black bars) and A. thaliana (light gray bars). Two different basal Zn conditions upon which additional Zn was supplied were assessed: Short Zn oversupply denotes a short-term high Zn supply to plants precultivated at control Zn concentrations; for A. thaliana, 30 μm Zn, and for A. halleri, 300 μm Zn were added to the culture medium for 2 and 8 h before harvest. Zn deficiency/resupply denotes a supply of 5 μm Zn to Zn-deficient plants for 2, 8, and 24 h before harvest. Note that substantial changes in root transcript levels occur as soon as 2 h after the change of Zn concentration in the culture medium whereas shoot transcript levels reflect the change in external Zn supply only at a later time point. Data shown are transcript levels relative to EF1α from one experiment representative of two independent biological experiments. Tissues from at least three culture vessels containing three plants each were pooled for each condition. Genes were concluded to be not expressed (n.e.) when C_T values were above a threshold of 35 and reaction efficiencies (REs) with the respective cDNA were more than 5% below the mean RE for the respective primer pair (see also “Material and Methods”).

Plants of two different Zn steady states were subjected to short-term increases in external Zn supply. In one experiment, plants grown under control (1 μm) Zn conditions were exposed to excess Zn concentrations of 30 μm Zn for A. thaliana and 300 μm Zn for A. halleri, for 2 and 8 h. The choice of concentrations reflects naturally selected Zn tolerance in A. halleri, whereas A. thaliana possesses only basic Zn tolerance (Clemens, 2001; Becher et al., 2004). (In preliminary experiments, the chosen concentrations were the maximum Zn concentrations that did not cause a reduction in biomass gain in A. thaliana and A. halleri, respectively, during 4 d of exposure [M. Becher, I.N. Talke, A.N. Chardonnens, and U. Krämer, unpublished data].) In an additional experiment, Zn-deficient plants cultivated in a medium lacking added Zn for 3 weeks were resupplied with a Zn concentration of 5 μm (Fig. 2).

In roots, Zn-induced changes in transcript levels were observed as early as 2 h after increasing the Zn concentrations in the hydroponic medium (Fig. 2, top row: ZIP4, ZIP9). In shoots, transcriptional responses were observed 2 or 8 h after the increase in Zn concentrations in the hydroponic medium (Fig. 2, bottom row: ZIP4, ZIP9, and NAS2). Transcripts of ZIP9 encoding a transporter of the ZIP family and NAS2 encoding a nicotianamine synthase were not detectable in shoots of either Arabidopsis species grown in the presence of sufficient (control) Zn, but were strongly induced upon Zn deficiency (Fig. 2, ZIP9, NAS2). This suggested that neither of the two species experienced Zn deficiency under the chosen control conditions.

Comparison of A. halleri and A. thaliana Transcript Profiles with ATH1 GeneChips

According to the results from the preliminary experiments, microarrays were hybridized with labeled cRNAs prepared from roots and shoots of A. halleri and A. thaliana grown at 1 μm Zn and after 2 and 8 h of exposure to excess Zn for roots and shoots, respectively (see “Material and Methods”). Under control conditions, higher signal intensities were detected for a total of 628 probe sets in roots, and for 1,739 probe sets in shoots of A. halleri compared to A. thaliana (as outlined in “Materials and Methods;” for visualization see all colored, nongray datapoints in Fig. 3, A and B ; complete gene lists with annotations in Supplemental Tables I and II). We also identified genes that responded at the transcript level to excess Zn in A. halleri (see “Materials and Methods;” Supplemental Tables III and IV). To group the identified genes according to functional classes, a genome-wide annotation list was assembled; this list includes the class of metal homeostasis-related genes, which contained all functionally characterized A. thaliana genes encoding proteins involved in metal transport, chelation, binding, and trafficking and all other members of the respective protein families, and genes encoding members of protein families known to participate in metal homeostasis in other organisms (Supplemental Table XI: class “Metal ion homeostasis;” Fig. 3, see “Material and Methods”). The intersection of the gene list of metal homeostasis-related genes and of the genes that are either more highly expressed (Supplemental Tables I and II) or Zn regulated in A. halleri (Supplemental Tables III and IV) contained a total of 17 genes in the roots and 19 genes in the shoots, with seven genes in common between both roots and shoots (Fig. 3; Table I ). We considered these genes as an initial set of major candidate genes for an involvement in naturally selected metal tolerance and hyperaccumulation in A. halleri.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

The relation between ATH1 microarray signals of GeneChips hybridized with cRNA from A. halleri and from A. thaliana grown under control Zn conditions. For each of the 22,810 probe sets on the ATH1 array, the mean signal derived from the two replicate arrays of roots (A) and shoots (B) of A. halleri was plotted on the y axis and the respective mean signal of A. thaliana was plotted on the x axis. Signal values represent mean signal intensities calculated from MAS 5.0-scaled raw signals of replicate arrays, without further normalization. Symbols and colors reflect the data filtering which was applied to identify candidate genes more highly expressed in A. halleri compared to A. thaliana and involved in metal ion homeostasis or related processes. Gray diamonds represent all probe sets that did not pass any filter. Yellow squares represent probe sets that exhibited a 4-fold higher signal intensity in A. halleri than in A. thaliana and passed the value filter (for details see “Materials and Methods”). Blue triangles represent probe sets that, in addition, passed a first annotation filter (I) of genes annotated as metal ion (Zn, Fe, Cu, and Mn) binding, membrane and transport associated, oxidative stress protection associated, pathogen response related, and RNA metabolism associated (Supplemental Table XI). Red circles represent probe sets that passed a further annotation filter (II) that included only genes with a predicted function in metal ion homeostasis. Candidate genes that passed the value filter and either annotation filter and that were chosen for further analysis by real-time RT-PCR are represented by green triangles (passed annotation filter I) or circles (passed annotation filter II) with a black outline (for details see Table I).

The metal homeostasis genes newly identified here as candidate genes encode proteins encompassing both cellular functions known to be of primary importance in metal homeostasis: membrane transport of metals and metal binding/chelation. The 12 new candidate genes encoding proteins that are likely to be involved in the transport of Zn or other metals across membranes were Zn-regulated transporter/iron-regulated transporter-like protein 10 (ZIP10; roots, 111-fold), metal tolerance protein 8 (MTP8; shoots, 65-fold), heavy metal-associated domain-containing protein 4 or heavy metal ATPase 4 (HMA4; shoots, 53-fold), ferric reductase defective 3 (FRD3; roots, 45-fold), phosphate transporter 1;4 (PHT1;4; roots, 19-fold), cation/H⁺ exchanger 18 (CHX18; roots, 6.5-fold), MTP11 (shoots, 4.7-fold), P-type ATPase from Arabidopsis 1 (PAA1/HMA6; shoots, 4.6-fold), iron-regulated transporter 2 (IREG2; roots, 4.5-fold), iron-regulated transporter 3 (IRT3; roots, 4.5-fold), yellow-stripe 1-like protein 6 (YSL6; shoots, 4.4-fold), and ZIP3 (down-regulated 2.7-fold in roots following exposure to excess Zn; Fristedt et al., 1999; Curie et al., 2001; Mäser et al., 2001; Rogers and Guerinot, 2002; Blaudez et al., 2003; Jensen et al., 2003; Lahner et al., 2003; Shikanai et al., 2003; Green and Rogers, 2004; Hussain et al., 2004; Shin et al., 2004). A total of five new candidate genes encoding metal-binding proteins or proteins involved in the biosynthesis of metal chelators included ferritin 2 (FER2; shoots, 11-fold), S-adenosyl-Met synthetase 1 (SAMS1; shoots, 10-fold), SAMS2 (shoots, 8.5-fold), FER1 (shoots, 5.1-fold), and SAMS3 (roots and shoots, around 5-fold; Peleman et al., 1989; Petit et al., 2001). The compound S-adenosyl-Met is the substrate for nicotianamine synthase, and the A. halleri nicotianamine synthase genes NAS2 and NAS3 have previously been implicated in Zn tolerance of A. halleri (Becher et al., 2004; Weber et al., 2004; see also Table I).

Two additional, closely related genes encoding the protein disulfide isomerases 1 and 2 were included among the set of candidate genes analyzed here because of their interesting regulation and potential function in metal detoxification (PDI1: shoots, 39-fold higher in A. halleri; PDI2: shoots, 17-fold; Table I; Rensing et al., 1997; Narindrasorasak et al., 2003). These proteins are the closest sequence relatives of a previously characterized endoplasmic reticulum-localized protein from castor bean (Ricinus communis), which catalyzes the isomerization of disulfide bridges within proteins. Enhanced PDI activity is thought to protect plants from stress that results in the modification of thiol groups of endoplasmic reticulum proteins (Houston et al., 2005).

Microarray Analysis of Zn-Dependent Transcriptional Regulation

Among candidate genes, the microarray data suggested a trend toward decreasing transcript levels in response to excess Zn in roots of both A. thaliana and A. halleri (Table I). Conversely, in A. halleri roots transcript levels of ZIP and NAS genes were increased under Zn deficiency, when compared to control conditions (Table I). In shoots of A. thaliana, transcript levels of several candidate genes were increased in response to excess Zn, namely for PDI and SAMS genes, and for YSL6 (Table I). In A. halleri, transcript levels of these genes were constitutively high and unchanged in response to excess Zn (Table I). Our data suggested that, especially in shoots, A. halleri responds differently to excess Zn than A. thaliana at the transcript level.

We investigated whether the species-specific patterns observed in metal-dependent transcriptional regulation of candidate genes are also reflected globally across the entire panel of genes represented on the ATH1 microarray. In roots of A. halleri and A. thaliana, exposure to excess Zn resulted in increased transcript levels for 0.5% and 0.9%, respectively, of all expressed genes, and in decreased transcript levels for only 0.09% and 0.07%, respectively, of all expressed genes (Supplemental Tables III and V, P < 0.1). Thus, the predominant directions of transcriptional responses to excess Zn were identical in roots of A. halleri and A. thaliana, with a global trend toward up-regulation opposing a trend toward down-regulation among our set of candidate genes in both species. In shoots of A. thaliana, among all expressed genes, transcriptional up-regulation was observed in response to excess Zn (0% down, 3.2% up, P < 0.1; Supplemental Table VI). By comparison, in shoots of A. halleri a smaller proportion of expressed genes responded, and transcriptional down-regulation was predominant (0.8% down, 0.2% up, P < 0.1; Supplemental Table IV). This indicated that primarily in the shoots, the two species respond differently to excess Zn also at the whole transcriptome level.

Confirmation of Candidate Genes by Real-Time RT-PCR

To assess the reliability of the cross species expression differences between A. halleri and A. thaliana detected using the microarrays, we determined transcript levels by an alternative method, quantitative real-time RT-PCR, for a subset of 18 genes (Table II ). Overall, there was good qualitative agreement between the ratios of microarray signals in A. halleri relative to A. thaliana and the ratios of transcript levels as determined by real-time RT-PCR (Table II). The false negatives according to microarray hybridization, ZIP3, and HMA4 (in roots), could be attributed to a high divergence of the A. halleri sequence from the A. thaliana sequence in the region covered by the oligonucleotide probes on the microarray, in combination with a modest magnitude of the A. halleri to A. thaliana ratio of transcript levels. For several other genes, quantitative discrepancies between microarray and real-time RT-PCR data are likely to be attributable to cross hybridization between high transcript levels of a gene in A. halleri and the probes for a closely related member of the same gene family (NAS2 on NAS4 in roots, NAS3 on NAS2 in shoots, NAS2 on NAS3 in shoots under Zn deficiency, and PDI1 on PDI2). It was reported earlier that very low signals (for example, for ZIP10, MTP8, and ZIP9 in A. thaliana) are often quantified imprecisely on the ATH1 chip (Czechowski et al., 2004), and for these genes in particular, the real-time RT-PCR results are likely to be more accurate.

Functional Analysis of the Novel Candidate Protein AhHMA4

Out of all candidate genes identified here, absolute transcript levels were highest for A. halleri HMA4 (see Table II), which encodes a membrane transport protein of the P_1B transition metal pump family of the P-type ATPases (Axelsen and Palmgren, 2001). To determine the cellular function of this candidate gene, which had not previously been implicated in metal hyperaccumulation or metal tolerance in A. halleri, we expressed AhHMA4 and AtHMA4 in metal hypersensitive mutants of budding yeast (Saccharomyces cerevisiae; Fig. 4 ). Both the A. thaliana and the A. halleri HMA4 proteins were similarly able to complement Zn hypersensitivity of the zrc1 cot1 double mutant and Cd hypersensitivity of the ycf1 mutant of yeast (Fig. 4, A and B). When the transport functions of AtHMA4 and AhHMA4 were disrupted by converting the conserved Asp residues (D401) of the phosphorylation motifs (Axelsen and Palmgren, 2001) into Alas, respectively, complementation was no longer observed. Consequently, Zn and Cd are substrates of both Arabidopsis HMA4 proteins in yeast, and the observed complementation is attributable to the transport functions of the proteins and not to the known ability of the cytoplasmic C termini of AtHMA4 and related proteins to bind metal ions (Bernard et al., 2004; Papoyan and Kochian, 2004).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Functional analysis of AtHMA4 and AhHMA4 cDNAs in yeast mutant strains. Expression of Arabidopsis HMA4 cDNAs in the Zn-hypersensitive zrc1 cot1 double mutant (A) and in the Cd-hypersensitive ycf1 mutant (B) is shown. Wild-type cells were transformed with pFL38H-GW (ev), and mutant cells were transformed with pFL38H-GW (ev) and with wild-type and mutated (D401A) cDNAs of both AtHMA4 and AhHMA4 in pFL38H-GW. The phosphorylation of the Asp residue (here D401) of the conserved phosphorylation motif is an indispensable step of the reaction cycle in all P-type ATPases. Serial dilutions of transformants were spotted on LSP plates containing the indicated concentrations of ZnSO₄ (A) or CdSO₄ (B). The 10⁰ dilution corresponds to an OD₆₀₀ of 0.5. Plates were incubated at 30°C for 4 d.

Gene Copy Number of Highly Expressed Candidate Genes in A. halleri

Previously, high transcript levels of the Zn tolerance gene MTP1 were partly attributed to the presence of several MTP1 gene copies in the genome of A. halleri, whereas the genome of the closely related nontolerant species A. thaliana contains only a single MTP1 gene copy (Dräger et al., 2004). To estimate the genomic copy number for other candidate genes in A. halleri, we generated genomic DNA blots for 12 of these genes (Kim et al., 1998; Marquardt et al., 2000; Small and Wendel, 2000; Ishizaki et al., 2002). Complex band patterns suggested that more than one gene copy is present in the A. halleri genome for the genes ZIP3, ZIP6, ZIP9, and HMA4 (Fig. 5, A to D ; Table III ; Supplemental Fig. 1; for data evaluation see also “Materials and Methods”). For comparison, the maximum expression of these genes was 8-, 10-, 13-, and 30-fold higher in A. halleri than in A. thaliana, respectively, as determined by real-time RT-PCR (Table II). By contrast, the simple band patterns of the genomic DNA blots indicated that a number of other candidate genes are most likely to be present as single copies in the A. halleri genome. These include FRD3, ZIP4, MTP8, ZIP10, IRT3, PHT1;4, and PDI1, the maximum expression of which was 16-, 12-, 11-, 6-, 5-, 5-, and 4-fold in A. halleri compared to A. thaliana, respectively (Tables II and III; Fig. 5). Among the group of candidate genes for which DNA gel-blot analysis was performed, more than single gene copies were detected for each of the three candidate genes exhibiting the highest transcript levels in A. halleri (normalized to EF1α transcript levels), i.e. HMA4 and ZIP9 and ZIP3 (compare Table II, left section). Based on sequence data and restriction analysis of genomic PCR products, it is unlikely that additional bands on the DNA gel blots originated from uncharacterized restriction sites rather than additional gene copies in the A. halleri genome (see Table III; M. Hanikenne and U. Krämer, unpublished data).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Genomic DNA gel blot for ZIP9 (A), HMA4 (B), FRD3 (C), and MTP8 (D) in A. thaliana and in A. halleri. Blots were performed using genomic DNA extracted from A. thaliana (Col accession) and from two A. halleri individuals (Lan 3-1 and Lan 5 of the accession Langelsheim). A to D, Five micrograms of genomic DNA were digested with EcoRI, HindIII, or NcoI, resolved on a 0.9% (w/v) agarose gel, blotted, and hybridized with radiolabeled A. halleri-derived cDNA probes. A, An EcoRI restriction site is present in the region spanned by the ZIP9 probe in A. thaliana only, a HindIII restriction site is present in the intron spanned by the probe in A. halleri only, and an NcoI restriction site is present in an exon spanned by the probe in both species. B, A HindIII restriction site is present in one of the introns spanned by the HMA4 probe in A. halleri. C, A HindIII restriction site is present in the region spanned by the FRD3 probe in both species. D, After restriction with NcoI, a predicted 20.9 kb fragment in A. thaliana was not detected by the MTP8 probe, probably because of the large size of this fragment (see also Table III). Note that for clarity, lanes were rearranged in the ZIP9 blot.

Regulation of Candidate Gene Expression

Transcriptional Zn responses were further investigated for a subset of 18 candidate genes using real-time RT-PCR (Fig. 6 ). In the roots of A. halleri, short-term exposure to 300 μm Zn for 2 and 8 h resulted in the down-regulation of transcript levels of a group of genes, which encode several ZIP family Zn transporters with likely functions in Zn influx into the cytoplasm (Fig. 6A). Down-regulation of transcript levels of this group of genes was clearly less pronounced in the roots of A. thaliana after short-term exposure to 30 μm Zn. It has to be kept in mind that in roots of A. halleri the Zn-induced decrease in transcript levels occurred from very high transcript levels under control conditions. Consequently, despite a more pronounced down-regulation in A. halleri, final root transcript levels 8 h after supply of excess Zn were generally higher in A. halleri than in A. thaliana (compare Fig. 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Transcriptional responses of selected candidate genes to Zn and other ions. Transcript levels relative to EF1α were assessed in roots (A) and shoots (B) of hydroponically grown plants subjected to short-term treatment with high Zn, long-term Zn deficiency, or short-term treatment with Cd, Cu, or Na (see “Materials and Methods” for details). RTLs within a treatment were normalized to the RTL of the control treatment in the respective experiment (“RTL normalized to control”). To determine transcriptional responses for each treatment and each gene, the mean RTL normalized to control was calculated from the respective values from two independent biological experiments. Bold letters indicate genes that exhibit higher transcript levels in A. halleri compared to A. thaliana after cultivation in the presence of control Zn concentrations. Note that some genes were not expressed in some conditions, see Figures 2 and 7 for details.

Different from roots, shoot transcript levels of most ZIP genes were generally as low in A. halleri as in A. thaliana under control conditions (see Fig. 2; Table II). In shoots of A. halleri, there was virtually no transcriptional response to a short-term oversupply of Zn, whereas A. thaliana shoots responded by a marked down-regulation of several genes including primarily ZIP genes and PHT1;4 (Fig. 6B). To test whether the shoots of A. halleri were able to respond transcriptionally to changes in Zn supply, we investigated gene expression under Zn deficiency (Fig. 6). Zn deficiency resulted in a very pronounced increase in shoot transcript levels of genes encoding ZIP family transporters and nicotianamine synthase proteins (Fig. 6B; see Table I). In roots of A. halleri, Zn deficiency resulted in a moderate increase in transcript levels of these genes, when compared to the high transcript levels observed under control conditions (Fig. 6A; for comparison with A. thaliana, see Figs. 2 and 7 ).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Zn-dependent transcriptional regulation of selected candidate genes. Real-time RT-PCR analysis of expression of HMA4, ZIP3, and IRT3 genes in roots (top section) and shoots (bottom section) of hydroponically grown A. halleri (black bars) and A. thaliana (light gray bars). Two different basal Zn conditions upon which additional Zn was supplied were assessed: Short Zn oversupply and Zn deficiency/resupply. For a detailed description, see Figure 2 and “Materials and Methods.” Data shown are transcript levels relative to EF1α from one experiment representative of two independent biological experiments. Tissues from at least three culture vessels with three plants each were pooled for each condition. Genes were concluded to be not expressed (n.e.) when C_T values were above a threshold of 35 and reaction efficiencies with the respective cDNA were more than 5% below the mean RE for the respective primer pair. Note that ZIP3 was not expressed in shoots only for the condition A. thaliana, 30 μm Zn supply, 8 h. At all other conditions, ZIP3 transcripts were detectable but RTLs were very low (A. halleri: 4.3 [control], 3.5 [2 h], and 2.6 [8 h]; A. thaliana: 1.7 [control], 1.7 [2 h], and 3.9 [control Zn deficiency]), and therefore do not appear as visible bars in the diagram.

To analyze the specificity of transcriptional Zn responses, candidate gene regulation in A. halleri was also investigated following exposure to moderate excesses of Cd (30 μm) or copper (Cu; 5 μm) and NaCl (100 mm) for 2 and 8 h (Fig. 6). The results suggested that overall, Zn responses were rather specific. Some of the Zn-responsive genes, primarily NAS and ZIP genes, were also down-regulated in response to excess Cu, as for excess Zn. The transcriptional responses to excess Zn and excess Cd, however, were distinct. Furthermore, there was a second group of genes exhibiting a pattern of regulation that was clearly different from the Zn- and Cu-repressed genes. Expression of these genes was induced in response to exposure to 100 mm NaCl and included MTP8, PHT1;4, FRD3, ZIP6, and OASA2. Transcript levels of subsets of these genes were also increased following exposure to excess Cu or Cd, supporting their responsiveness to abiotic stress. PDI transcript levels showed few changes over the entire range of conditions, with an increase in root PDI2 transcript levels in response to excess Cu in A. halleri and an increase in PDI1 transcript levels in response to Zn in shoots of A. thaliana.

Transcript levels of HMA4 appeared to be constitutively very high in the roots of A. halleri across all treatments, and only slightly induced under Zn deficiency in A. halleri shoots (Fig. 6, A and B). A detailed analysis confirmed the extremely high expression of HMA4 in roots and shoots of A. halleri (Fig. 7). Under Zn deficiency, shoot HMA4 transcript levels were slightly increased, and they were reduced to control levels shortly after readdition of Zn to the medium (Fig. 7).

We investigated transcript levels of the additional newly identified candidate genes ZIP3 and IRT3, under different conditions of Zn deficiency, resupply, and oversupply (Fig. 7). There were overall similarities in short-term Zn responses of plants maintained under control (1 μm added Zn) and Zn-deficient (0 μm added Zn) conditions (Fig. 7; see also Fig. 2). In response to resupply of 5 μm Zn to Zn-deficient plants, there was a decrease in transcript levels for these Zn-deficiency-induced genes in both A. halleri and A. thaliana. This demonstrated clearly that there is no general defect in the magnitude or time scale of transcriptional Zn responses in A. halleri (Fig. 7; see also Fig. 2: ZIP4 and ZIP9 in controls and at time points 8 and 24 h). In roots of A. halleri, after the initial down-regulation, which was strongest 2 h after resupply of Zn, a general trend of a gradual increase in transcript levels was observed at later time points (Fig. 7).

Plant Material, Growth Conditions, and Experimental Treatments

Plants of Arabidopsis thaliana (L. Heynhold, accession Columbia [Col]) and Arabidopsis halleri (L. O'Kane and Al-Shehbaz subsp. halleri, accession Langelsheim) were grown from seeds, which were the progeny of a single mother plant in a pool of six individuals grown in an open greenhouse from seeds collected at Langelsheim. Plants were cultivated in hydroponic culture as described in Becher et al. (2004), with a photoperiod of 11 h light, 13 h dark, at a photon flux density 145 μmol m⁻² s⁻¹ during the day, and day and night temperatures of 20°C and 18°C, respectively. In all experiments, each culture vessel of 400 mL hydroponic solution contained three individual plants, and all treatments were harvested at the same time. Experiments for subsequent RNA extraction were harvested by carefully separating roots and shoots of plants, pooling tissues according to treatment, and freezing in liquid nitrogen. Harvested tissues were stored at −80°C until further processing.

Short-term high Zn, Cd, Cu, and Na treatments were initiated when plants were 6.5 weeks old. For short-term high Zn supply, the medium was replaced with fresh control medium (1 μm ZnSO₄) 8 h before harvest. Then nothing (controls) or 30 μm ZnSO₄ and 300 μm ZnSO₄ for A. thaliana and A. halleri, respectively, was added to the hydroponic solution 8 or 2 h before harvest. Three biologically independent experiments, each comprising five culture vessels per treatment (15 plants per treatment) were conducted. In experiments with Cd, Cu, and Na treatments of A. halleri, media were exchanged (1 μm ZnSO₄; Cd: none; Cu: 0.5 μm CuSO₄; Na: 0.1 μm Na₂MO₄) 3 d before harvest; 30 μm CdCl₂, 5 μm CuSO₄, or 100 mm NaCl were added as treatments 8 or 2 h before harvest. Two biologically independent experiments were conducted, with at least two culture vessels (at least six plants in total) for each treatment.

For Zn deficiency experiments with Zn resupply, the treatment was initiated when plants were 4.5 weeks old and lasted for 3 weeks. For deficiency, ZnSO₄ was omitted from the hydroponic solution; the sufficient control solution contained 5 μm ZnSO₄. The last change of medium was done 3 d before harvest. For Zn resupply, 5 μm ZnSO₄ was added to the hydroponic solution of Zn-deficient plants 24, 8, and 2 h before harvest. Two biologically independent experiments were conducted, each comprising at least three culture vessels (at least nine plants) per treatment.

In the experiments conducted to determine root and shoot Zn concentrations after long-term treatment with a range of different Zn supplies, plants were 4.5 weeks old when treatments were initiated. For A. halleri, the hydroponic solution contained 0, 0.3, 1, 5, 10, 30, or 300 μm ZnSO₄, for A. thaliana 0, 1, 5, or 10 μm ZnSO₄. Upon harvest after 3 weeks of treatment, shoots were briefly rinsed in ultrapure water and blotted dry, and roots from three individuals of the same species and hydroponic culture vessel were desorbed together in 150 mL of an ice-cold solution of 5 mm CaCl₂ and 1 mm MES-KOH (pH 5.7) for 20 min, with a replacement of the solution after 5 min, followed by two washes in 150 mL ice-cold water, each for 3 min. Tissues were pooled by culture vessel and dried at 60°C for 3 d. Two biologically independent experiments were conducted, each comprising three culture vessels (nine plants) per treatment.

Determination of Zn Concentrations

Dried root and shoot material was homogenized, and for each sample, 35 to 70 mg were processed and elemental contents analyzed by inductively coupled plasma optical emission spectroscopy using an IRIS Advantage Duo ER/S (Thermo Jarrell Ash) as described previously (Becher et al., 2004).

RNA Extraction

For microarray expression profiling, total RNA was extracted with Trizol (Invitrogen Life Technologies) according to the manufacturer's instructions. RNeasy spin columns and the RNase-free DNase set for on-column DNase digest (Qiagen) were used according to the manufacturer's instructions for subsequent RNA purification and digestion of genomic DNA, and as an alternative method of total RNA isolation when transcript levels were determined by real-time RT-PCR only (RNeasy plant mini kit and RNase-free DNase set; Qiagen). Quality and quantity of RNA was checked visually by denaturing gel electrophoresis and by photometric analysis (A₂₆₀ and A₂₈₀).

Microarray Expression Profiling and Data Analysis

Synthesis of cDNA, cRNA labeling, and the hybridization on the GeneChip Arabidopsis ATH1 genome array were done as recommended by the manufacturer (Affymetrix; manual 701025 rev 1, https://www.affymetrix.com/support/downloads/manuals/expression_s2_manual.pdf). In short, 20 μg of total RNA were used for first-strand cDNA synthesis with 100 pmol T7(dT)₂₄ primers and SuperScript II RT (Invitrogen), second-strand cDNA synthesis was carried out with Escherichia coli DNA ligase, DNA polymerase I, and RNase H (all Invitrogen). Biotin-labeled cRNAs were synthesized by in vitro transcription using the Enzo BioArray HighYield RNA transcript labeling kit (Enzo Diagnostics). Labeled cRNAs were quantified according to Affymetrix instructions, and a corrected amount of 25 μg cRNA were fragmented in fragmentation buffer (200 mm Tris-acetate, pH 8.1, 500 mm KOAc, 150 mm MgOAc). Before hybridization on ATH1 arrays, sample quality was assessed by hybridization on the test3-array (Affymetrix) for examination of 3′ to 5′ intensity ratios of houskeeping genes. ATH1 GeneChips were hybridized with cRNA from the following experiments: of the short-term Zn supply experiments, two for each treatment, tissue, and species; of the Zn deficiency experiments, one for each tissue from the 5 μm Zn (sufficient) and 0 μm Zn (deficient) treatment, respectively (Dr. F. Wagner, Resource Center and Primary Database, Berlin).

The microarray suite software package (MAS 5.0, Affymetrix) was used to generate probe set signals of the scanned ATH1 arrays. At a target intensity of 100, scaling factors for arrays hybridized with cRNA from A. halleri were on average 2-fold higher than scaling factors for A. thaliana arrays (A. halleri: 0.943 ± 0.196; A. thaliana: 0.424 ± 0.093). The arithmetic means of signal intensities over all probe sets were 146.9 ± 8.3 for A. halleri and 135.4 ± 9.6 for A. thaliana. For A. thaliana an average of 13,907 (61.0%) and 11,646 (51.1%) of the total number of 22,810 probe sets were assigned a “present call” by the MAS 5.0 software in roots and shoots, respectively. On average 9,750 (42.8%) and 9,389 (41.2%) probe sets were assigned a “present call” in roots and shoots, respectively, for A. halleri. Pivot data of each hybridized ATH1 GeneChip containing raw signal values and present, marginal, and absent calls (flags) were imported from MAS 5.0 into GeneSpring GX (v 7.2; Agilent Technologies, http://www.chem.agilent.com/). Within GeneSpring, the following normalization steps were carried out for each chip: first, signals below 0.01 were set to 0.01; second, all signals within a dataset were normalized to the 50th percentile of the measurements of the chip, using only measurements with a raw signal of at least 12.5; third, in a per gene normalization, the signal for each probe set was normalized to the respective signal of the control chip (from the same experiment; chips from A. thaliana for interspecies comparisons, control condition for within-species comparisons). Calculation of mean values for raw and normalized signals, statistical analysis, and data filtering were performed in GeneSpring. For A. halleri versus A. thaliana comparisons, the following filters were applied sequentially to identify genes more highly expressed in A. halleri than in A. thaliana (value filter). First, to avoid the identification of false positives in a cross species evaluation, probe sets were selected that exhibited an at least 4-fold higher mean normalized signal in A. halleri than in A. thaliana. Second, among these probe sets, we selected probe sets for which the mean raw signal value was above 25 in A. halleri. Third, we selected only those probe sets for which in a one sample Student's t test the P value was below 0.05, with false discovery rate control adjusted at 5% (Benjamini and Hochberg, 1995). Fourth, only those probe sets were retained that were assigned at least one “present call” for A. halleri among the chips under comparison. For within-species comparisons, to identify genes up-regulated and/or down-regulated upon exposure to Zn, the following selection criteria were employed sequentially: For up-regulation/down-regulation in the experimental Zn condition (2 or 8 h high Zn supply; Zn deficiency), probe sets with (1) a mean normalized signal at least 2-fold higher/lower than in the control condition, (2) a mean raw signal above a threshold value (25 for A. halleri, 12.5 for A. thaliana), and (3) present calls in all biological replicates of the experimental condition (up-regulation)/control condition (down-regulation). To obtain an estimate of the percentage of genes regulated in within-species comparisons (as reported in the “Results”), a P-value cutoff of 0.1 was applied. The same cutoff was used for the gene lists shown in Supplemental Tables III, IV, and V. The P-value cutoffs were 0.1 for up-regulation and 0.2 for down-regulation for the gene lists shown in Supplemental Table VI, and 0.05 for up-regulation and 0.4 for down-regulation for Supplemental Tables VII and VIII.

Lists of candidate probe sets and their respective signal data were exported from GeneSpring. Tables for assigning Affymetrix ATH1 probe set identifiers to Arabidopsis Genome Initiative locus identifiers were obtained from The Arabidopsis Information Resource (TAIR; http://www.arabidopsis.org) and Genomanalyse im biologischen System Pflanze (http://gabi.rzpd.de/services/Affymetrix.shtml). In cases where an ATH1 probe set could not be assigned to one single locus identifier, all locus identifiers represented by this probe set and its associated signal data are given. Annotation and classification of genes as metal ion homeostasis related, metal ion (Zn, Fe, Cu, and Mn) binding, membrane and transport associated, oxidative stress protection associated, pathogen response related, and RNA metabolism associated was done through literature review, keyword searches in TAIR, use of the PlantsT database (http://plantst.genomics.purdue.edu/), and using MapMan annotation based on TAIR5 (Usadel et al., 2005; http://gabi.rzpd.de/projects/MapMan/). Annotation of genes was obtained from TAIR. Table associations were done with Microsoft Access. In all Tables and Supplemental Tables, P values are given as exported from the program GeneSpring GX, i.e. from a one sample Student's t test and not adjusted for false discovery rate.

Real-Time RT-PCR

Synthesis of cDNA, primer design, and quality control were done as described in Becher et al. (2004). Primer sequences are given in Supplemental Table IX.

Real-time RT-PCR reactions were performed in 384-well plates with an Applied Biosystems ABI Prism 7900HT sequence detection system (Applied Biosystems; http://www.appliedbiosystems.com/) using SYBR Green to monitor cDNA amplification. Equal amounts of cDNA, corresponding to approximately 0.06 ng of mRNA were used in each reaction. In addition, a reaction contained 5 μL of SYBR Green PCR Master Mix (Applied Biosystems) and 2.5 pmol of forward and reverse primers (Eurogentec) in a total volume of 10 μL. The following standard thermal profile was used: 2 min at 50°C, 10 min at 95°C, 40 repeats of 15 s at 95°C and 60 s at 60°C, and a final stage of 15 s at 95°C, 15 s at 60°C, and 15 s at 95°C to determine dissociation curves of the amplified products. Data were analyzed using 7900 HT sequence detection system software (v 2.2.1, Applied Biosystems). Threshold cycle (C_T) values were determined for each reaction at a threshold value of the normalized reporter R_n of 0.2. Furthermore, the reaction efficiency (RE) was determined for each PCR reaction with LinRegPCR v7.5 (Ramakers et al., 2003). Real-time RT-PCR was performed on material from at least two independent biological experiments. For confirmation of microarray data we used material from one experiment used for microarray hybdridization and from at least one additional independent experiment not used for microarray hybridization. At least two technical repeats were done for each combination of cDNA and primer pair, and the quality of the PCR reactions was checked through analysis of the dissociation and amplification curves. Only those reactions were used for data evaluation for which transcripts were reliably detectable (C_T values below 35), and RE was above 1.5. Mean reaction efficiencies (RE_m) were determined for each primer pair and for each Arabidopsis species from all reactions passing the quality control (at least 40 reactions; Supplemental Table IX). Mean C_T values (C_T,m) were calculated from all quality-checked technical repeats of a cDNA-primer pair combination.

Over all cDNAs used, the average C_T value for the constitutively expressed control gene, elongation factor EF1α (At5g60390, Becher et al., 2004), was 18.68 with a sd of 1.03 (n = 76). Transcript levels of genes of interest (GI) within a cDNA were normalized to the respective transcript level of EF1α using the following formula:

For clarity, values shown in Tables and Figures are RTL × 10³.

Cloning

All cloning and molecular biology procedures were carried out according to standard protocols unless indicated otherwise (Sambrook and Russel, 2001). For the candidate genes, A. halleri cDNA fragments were amplified with Red-Taq DNA polymerase (Sigma) using primers designed on the basis of the corresponding A. thaliana sequences (Supplemental Table X). The standard PCR program was as follows: 3 min at 95°C, followed by 35 cycles of 30 s at 95°C, 30 s at 55°C, 1 min per kb at 72°C, and a final extension step of 7 min at 72°C. The PCR products were cloned into the pCR2.1 TOPO vector (Invitrogen) and the inserts of at least three independent plasmids were sequenced. Unless stated otherwise, A. halleri-cDNA-derived probes for DNA gel-blot hybridizations were prepared from pCR2.1 TOPO clones by excision of fragments using EcoRI. For ZIP3, ZIP6, PHT1;4, and MTP8, a 472 bp HindIII-SalI fragment, a 380 bp EcoRI-NdeI fragment, a 550 bp EcoRI fragment, and a 594 bp HindIII fragment were used, respectively.

Full-length cDNAs for AtHMA4 and AhHMA4 were obtained as follows. cDNA libraries were prepared from mRNA of A. thaliana (Col-0 accession) inflorescence tissues and from root tissues of several individuals of A. halleri (Langelsheim accession), respectively, using the SMART RACE cDNA amplification kit according to the manufacturer's instructions (BD Biosciences). The open reading frames of AtHMA4 and AhHMA4 were subsequently amplified using a proofreading polymerase (Pfu Turbo; Stratagene) and the following PCR program: 3 min at 95°C, followed by 10 cycles of 30 s at 95°C, 30 s at 58°C, 8 min at 68°C, 20 cycles of 30 s at 95°C, 30 s at 55°C, 8 min at 68°C, and a final extension step of 7 min at 68°C. The products were cloned directionally into the GATEWAY entry vector pENTR TOPO (Invitrogen), and later subcloned into the yeast (Saccharomyces cerevisiae) centromeric expression vector pFL38H-GW by site-directed recombination according to the manufacturer's instructions (Invitrogen). pFL38H-GW was constructed by inserting a SmaI-BglII fragment from the vector pFL61-AKT1 (Sentenac et al., 1992) into the SmaI-BglII-digested vector pFL38 (Bonneaud et al., 1991), followed by replacement of the URA3 marker with the HIS3 marker, which was obtained from YDpH using BamHI, and inserted into the pFL38-derived vector following digestion with BglII. Subsequently, AKT1 was excised from the resulting vector using NotI, the linearized vector blunt ended using the Klenow fragment of DNA polymerase I, and a blunt Gateway cassette inserted according to the manufacturer's instructions (Invitrogen). To construct the AtHMA4 D401A and AhHMA4 D401A mutant cDNAs, linear amplification was performed of the AtHMA4 and AhHMA4 in the yeast expression vector pFL38H-GW in a PCR reaction using mutagenic primers (Supplemental Table X; 12 cycles of 95°C for 30 s, 55°C for 60 s, and 68°C for 16 min), followed by digestion of the parent DNA with DpnI, and subsequent transformation of E. coli. Constructs were verified by sequencing.

Yeast Complementation Assays

The following yeast strains were used in the assay: zrc1 cot1 (Mat a, zrc1∷natMX3, cot1∷kan-MX4, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0; Becher et al., 2004) and its parental strain BY4741 (Mat a, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0), ycf1 (Mat α, his3Δ1, leu2Δ0, lys2Δ0, ura3Δ0, YDR135c∷kanMX4), and its parental strain BY4742 (Mat α, his3Δ1, leu2Δ0, lys2Δ0, ura3Δ0; http://web.uni-frankfurt.de/fb15/mikro/euroscarf/index.html). Preparation of competent cells and transformations were carried out following the polyethylene glycol method (Dohmen et al., 1991). Both yeast mutant strains were transformed with pFL38H-GW empty vector (ev) or pFL38H containing AtHMA4 and AhHMA4, AtHMA4 D401A, and AhHMA4 D401A cDNAs. As control, the corresponding wild-type strains were transformed with pFL38H-GW ev. Transformants were selected and maintained on synthetic complete (SC) medium lacking His and with d-Glc as a carbon source. For complementation assays, transformants were grown overnight in 5 mL SC −His media to early stationary phase (OD₆₀₀ approximately 1.0). Yeast cells were then washed twice with modified low sulfate/phosphate (LSP) medium (Conklin et al., 1992), which contained 80 mm NH₄Cl, 0.5 mm KH₂PO₄, 2 mm MgSO₄, 0.1 mm CaCl₂, 2 mm NaCl, 10 mm KCl, trace elements, vitamins, and supplements −His as in SC, 2% [w/v] d-Glc, and serially diluted to identical OD₆₀₀. Five microliters of each serial dilution were spotted onto LSP −His plates (1.5% agarose, Seakem, BMA) containing various concentrations of ZnSO₄ (2 μm for controls and between 100 and 400 μm for experimental treatments) or CdSO₄ (15–60 μm). Results are from one transformant representative of at least three independent transformants for each experiment and two independent experiments.

DNA Gel-Blot Analysis

Genomic DNA of A. thaliana (Col-0 accession) and clones of two A. halleri individuals (Lan 3-1 and Lan 5) of the Langelsheim accession (Ernst, 1974) were isolated as described previously (Dräger et al., 2004). Five micrograms of genomic DNA were digested with 50 units EcoRI, HindIII, or NcoI in the appropriate buffer (Roche), supplemented with 100 μg mL⁻¹ bovine casein (Sigma) overnight at 37°C, and separated on a 0.9% (w/v) TRIS-acetate-EDTA agarose gel. The DNA was transferred onto an uncharged nylon membrane (Hybond N; Amersham Biosciences) and cross-linked with UV light at 120 mJ (UV Stratalinker 1800; Stratagene). The probes were radiolabeled by random priming using [α³²P]dCTP (Hartmann Analytics) according to the manufacturer's instructions (ReadyPrime labeling kit; Amersham Biosciences). Stringent hybridization and washing conditions were used to minimize cross hybridization within gene families. The membranes were hybridized overnight at 60°C in the following buffer: 0.25 m Na phosphate (pH 7.2), 1 mm Na₂EDTA, 6.7% (w/v) SDS, and 1% (w/v) bovine serum albumin. Blots were washed twice in 2× SSC containing 0.1% (w/v) SDS for 10 min at room temperature and once in 2× SSC containing 0.1% (w/v) SDS for 10 min at 60°C. X-ray films were exposed for between 5 and 30 d. As all blots were done using clones of the two A. halleri individuals named 3-1 and 5, there can be a minimum of one and a maximum of two alleles per gene copy per lane. When no restriction site was present in the region spanned by the probe, the presence of more than two bands in any one lane was interpreted as an indication for a gene copy number above one. When one restriction site was present in the region spanned by the probe, the presence of more than four bands (two per allele) in a lane was interpreted as an indication for a gene copy number above one. For one 200-bp NcoI fragment of ZIP9 located entirely within a single exon (see Table III), we expected no length polymorphisms between alleles of one gene copy, based on the sequences of multiple cloned A. halleri cDNAs (I.N. Talke, M. Hanikenne, U. Krämer, unpublished data).