Functional analysis of the heavy metal binding domains of the Zn/Cd-transporting ATPase, HMA2, in Arabidopsis thaliana


Author for correspondence:
Christopher S. Cobbett
Tel: +61 3 8344 6246
Fax: +61 3 8344 5139


  • • The Zn/Cd-transporting ATPase, HMA2, has N- and C-terminal domains that can bind Zn ions with high affinity. Mutant derivatives were generated to determine the significance of these domains to HMA2 function in planta.
  • • Mutant derivatives, with and without a C-terminal GFP tag, were expressed from the HMA2 promoter in transgenic hma2,hma4, Zn-deficient, plants to test for functionality.
  • • A deletion mutant lacking the C-terminal 244 amino acids rescued most of the hma2,hma4 Zn-deficiency phenotypes with the exception of embryo or seed development. Root-to-shoot Cd translocation was fully rescued. The GFP-tagged derivative was partially mis-localized in the root pericycle cells in which it was expressed. Deletion derivatives lacking the C-terminal 121 and 21 amino acids rescued all phenotypes and localized normally. N-terminal domain mutants localized normally but failed to complement the hma2,hma4 phenotypes.
  • • These observations suggest that the N-terminal domain of HMA2 is essential for function in planta while the C-terminal domain, although not essential for function, may contain a signal important for the subcellular localization of the protein.


Members of the P1B subfamily of the ATPases transport heavy metal ions. There are eight P1B ATPases (or HMAs) in Arabidopsis, of which HMA2 and HMA4 are essential for Zn homeostasis (Hussain et al., 2004). hma2,hma4 double mutants have a severe shoot Zn-deficiency phenotype and accumulate Zn in roots. Both genes are expressed in the root vasculature and are required for the root-to-shoot translocation of Zn via xylem loading (Hussain et al., 2004). Various heterologous expression studies have shown HMA2 (D. Hussain & C. Cobbett, unpublished) and HMA4 efflux Zn from cells (Mills et al., 2005; Verret et al., 2005).

One of the characteristic features of P1B-ATPases is the presence, in most members of this family, of a cytosolic N-terminal region with one or more metal-binding domains (N-MBDs) which generally contain the core consensus motif, GMxCxxC. Structural analysis of N-MBDs of P1B-ATPases from both the monovalent and divalent cation transporting subgroups showed that they have a characteristic βαββαβ-fold structure (Gitschier et al., 1998; Banci et al., 2001, 2002, 2006). Biochemical and functional analyses have shown that, in general, N-MBD of P1B-ATPases bind heavy metals with high affinity (DiDonato et al., 1997; Lutsenko et al., 1997; Jensen et al., 1999; Liu et al., 2005), influence the enzymatic turnover rate (Voskoboinik et al., 1999, 2001; Mitra & Sharma, 2001; Fan & Rosen, 2002; Mana-Capelli et al., 2003; Mandal & Arguello, 2003) and, in many cases, are essential to the in vivo function of the protein, for example in functional complementation (Iida et al., 1998; Payne & Gitlin, 1998; Forbes et al., 1999; Fan et al., 2001; Fan & Rosen, 2002).

In Arabidopsis, HMA2, HMA3 and HMA4 have a variant GxCCxxE motif within a similar βαββαβ-fold domain (Eren et al., 2007). The HMA2 N-MBD can bind Zn2+ (Kd 0.18 µm) and Cd2+ (Kd 0.27 µm) with high affinity with the two Cys residues of the GxCCxxE motif having an important role in Zn and Cd coordination (Eren et al., 2007). Truncation of the N-MBD or mutation of the two Cys residues caused a 50% reduction in HMA2 ATPase activity when expressed in yeast membrane vesicles (Eren et al., 2007). For HMA4, mutation of each of the two Cys residues in the GxCCxxE motif abolished its ability to complement the Zn hypersensitivity of zrc1 and the Cd hypersensitivity of ycf1 mutant yeast strains (Verret et al., 2005). A similar in vivo functional analysis has not been reported for HMA2.

Compared with other P1B-ATPases, both HMA2 and HMA4 have unusually long C-terminal regions containing numerous Cys and His residues. After the predicted last TM domain, the cytosolic C-terminal regions of HMA2 and HMA4 have c. 250 and 470 amino acids, respectively. The C-terminal region of HMA2 binds three Zn2+ ions with high affinity (Kd = 16 ± 3 nm) and truncation of the C-terminal decreased the ATPase activity about twofold compared with the wild-type when expressed in yeast membrane vesicles (Eren et al., 2006). Functional analysis of the HMA4 C-terminal region, however, has produced inconsistent results. Mills et al. (2005) reported that truncating the entire C-terminal of HMA4 did not prevent the mutant protein from complementing the Cd hypersensitivity of the yeast ycf1 mutant. In contrast, Verret et al. (2005) observed that by deleting the 16 His-rich amino acid residues at the C-terminus, the mutant HMA4 protein could no longer complement the Cd and Zn hypersensitivity of ycf1 and zrc1 mutants, respectively. Thus, the importance of the C-terminal region to the in vivo function of HMA4 and HMA2 remains to be resolved.

In this study we have tested the effect of mutations of the N- and C-terminal regions of HMA2 on its function in planta. Mutant constructs were expressed from the native HMA2 promoter and assessed for their ability to complement the Zn-deficiency phenotypes of the hma2,hma4 double mutant and to restore root-to-shoot Cd translocation. We conclude that the N-terminal domain is essential for function in planta while deletion of the C-terminal domain had relatively little effect.

Materials and Methods

Plant materials

The Arabidopsis thaliana (L.) Heynh strains used in this study were wild-type (Col0), hma2-4,hma4-2 double mutant (Hussain et al., 2004) and transgenic HMA2p-HMA2-GFP (Sinclair et al., 2007). Plants were grown in agar and soil as previously described (Howden et al., 1995). The mutant HMA2 constructs generated in this study were transformed into the hma2,hma4 double mutant using Agrobacterium tumefaciens (Clough & Bent, 1998). Transgenic plants were screened for BASTA resistance, and for each construct at least six independent lines homozygous for the transgene at a single locus were obtained.

Creation of mutant HMA2 derivatives

The N-terminal truncated HMA2 (ΔN-MBD) was amplified by PCR using a forward primer 5′-CTGGTCACTTCGATTTTTCCCAAAG and a mutagenic reverse primer 5′-TCTAGAGGTTACCATCTTGTTTAAGGATTCTGC carrying a BstEII site that binds to the start codon of HMA2. The PCR product was digested with SwaI/BstEII and used to replace the corresponding wild-type fragments in both the HMA2p-HMA2 and HMA2p-HMA2-GFP constructs.

Both GIA17A and A391KTG HMA2 mutants were generated by overlapping PCR (Ho et al., 1989), using four primers for each construct. For GIA17A the two outer primers were (forward) 5′-CTGGTCACTTCGATTTTTCCCAAAG and (reverse) 5′-GCATCTGAGAGATTGAAAACACC, and the two complementary mutagenic inner primers were (forward) 5′-CTTGGATTGCCGCTACATCGGAGG and (reverse) 5′-CCTCCGATGTACAGCAAATTCCAAG. For A391KTG the two outer primers were (forward) 5′-ATCCTCTCCACACCAGTAGCCA and (reverse) 5′-GAGCTCCTATTCAATCACAATCTC, and the two complementary mutagenic inner primers were (forward) 5′-GATTGTTGCTTTTGCTAAAACCGGGAC and (reverse) 5′-GTCCCGGTTTTAGCAAAAGCAACAATC. In the first PCR, the outer forward primer was used in combination with the mutagenic reverse primer and, separately, the mutagenic forward primer with the outer reverse primer. After mixing the two PCR products, the outer forward and reverse primers were used in a further PCR reaction to produce a mutant HMA2 carrying the desired mutation/s. Both A391KTG and GIA17A mutant products were cloned into pGEM-T Easy vector (Promega, Melbourne, Australia) and sequences were verified using Big Dye Terminator V3.1 Cycle Sequencing Kit (Applied Biosystems, Melbourne, Australia). After ApaLI/SacI digestion, the mutant A391KTG fragment was substituted for the wild-type fragment in the HMA2p-HMA2 construct, while the mutant GIA17A fragment replaced corresponding fragments from both the HMA2p-HMA2 and HMA2p-HMA2-GFP constructs.

For the three ΔC-terminal HMA2 mutant constructs, ΔC-707, ΔC-830 and ΔC-930, the mutagenic reverse primers (which all carried a premature stop codon followed by a SacI site) used were 5′-GGTTACCCTCCCTATAACATTTGT, 5′-GGTTACCACAACACCCTGAGTCCA and 5′-GGTTACCGCAACTCTCTTTAGCGT, respectively. Each of these primers was used in combination with a common forward primer 5′-ATCCTCTCCACACCAGTAGCCA, and the three ΔC-terminal fragments, digested with ApaLI and SacI, used to replace the corresponding wild-type fragment from the HMA2p-HMA2-GFP construct. To obtain a GFP-tagged version of each ΔC-terminal HMA2 mutant, the mutagenic reverse primers used were 5′-CTCCCTATAACATTTGTTTCCAG, 5′-ACAACACCCTGAGTCCAG and 5′-GCAACTCTCTTTAGCGTAGCTCC. These primers did not introduce a premature stop codon or SacI site following the truncation end-point. These three ΔC-terminal fragments were ligated with a GFP fragment produced using forward primer 5′-ATGGTGAGCAAGGGCGAGG and reverse primer 5′-TGCCAAATGTTTGAACGATCG, and the ligated products, digested with ApaLI and SacI, were used to replace a corresponding wild-type fragment from the HMA2p-HMA2-GFP construct.

Quantitative real-time reverse-transcription PCR

Total RNA was extracted from shoot tissue of 16-d-old plants grown on agar medium using an RNeasy Plant Mini Kit (Qiagen, Melbourne, Australia) and DNase-treated using a DNA-free kit (Ambion, Austin, TX, USA) according to manufacturers’ protocols. cDNA was transcribed from 3 µg total RNA using the SuperScriptIII First Strand Synthesis system for RT-PCR (Invitrogen, Melbourne, Australia) and 1 µl of samples (diluted 1 : 10) was used for real-time RT-PCR analysis. The reactions for real-time RT-PCR were prepared using a SensiMix dT kit (Quantace, Sydney, Australia) and analysed using a Rotor Gene 3000 (Corbett Research, Melbourne, Australia). Each sample was amplified using ACT2 as a control (with primers, 5′-GGTAACATTGTGCTCAGTGGTGG and 5′-CTCGGCCTTGGAGATCCACATC, respectively) and HMA2 (with primers, 5′-GCTGAGGATTGCGTGGTT and 5′-GATAAGTCCACAAGCACAAGCAC), and the expression of transgenes was measured as a relative fold-difference from Col0 wild-type using the comparative quantification method by Rotor Gene 5.0 software (Corbett Research). In comparative quantification, the transcript expression levels of both ACT2 and HMA2 in each transgenic plant line were calibrated against those in Col0 wild-type plants. The calibrated transgene expression level was then normalized to the calibrated ACT2 expression level. A fold-change value of 1 indicates that a transgenic line had similar transgene expression level as endogenous HMA2 in Col0 wild-type. For each independent transgenic line, RNA from three separately grown biological samples was measured in quadruplicate.

Zn content and Cd translocation assays

The Zn content in shoot tissues was determined using inductively coupled plasma atomic emission spectroscopy, as previously described by Hussain et al. (2004). Cd translocation was measured as in Wong & Cobbett (2008). Percentage Cd translocation represents the proportion of total Cd taken up by plants that was translocated to shoots.

Confocal microscopy

Samples were mounted on to slides (in water) for visualization of GFP expression in root tissues. Images were collected using a confocal laser-scanning microscope (Olympus Fluoview FV1000), at excitation 488 nm, attached to an inverted microscope.

Statistical analysis

All data evaluation and statistics were carried out using Microsoft Excel 2004 for Mac. Quantitative values are presented as mean ± standard error of the mean (SE). Significance values were adjusted for multiple comparisons using the Bonferroni correction.


Constructs for assessing the function of HMA2 mutant derivatives in planta

A construct containing an HMA2 cDNA-GFP fusion expressed from the HMA2 promoter region has been described previously (Sinclair et al., 2007). This construct consists of DNA extending from 5920 bp upstream of the ATG codon of HMA2 fused to a full-length HMA2 cDNA, which was itself fused to GFP at the 3′ end (HMA2p-HMA2-GFP). Previous work has shown that the same promoter region expressed a β-glucuronidase reporter gene in vascular tissue and in developing anthers and paralleled expression of an HMA4p-GUS reporter construct, suggesting a basis for the apparent redundancy in function of the two genes (Hussain et al., 2004). In addition, the same HMA2-GFP fusion used here, when previously expressed from the CaMV35S promoter in transgenic plants, localized to the plasma membrane (Hussain et al., 2004). The HMA2p-HMA2-GFP construct was able to phenotypically complement the severe Zn-deficiency phenotypes of the hma2, hma4 mutant, and in roots the HMA2p-HMA2-GFP construct was expressed in pericycle cells (Sinclair et al., 2007), consistent with the HMA2p-GUS expression pattern, and the HMA2-GFP fusion was localized to the plasma membrane. These observations suggest that the tissue specificity of expression of the HMA2p-HMA2-GFP construct and the subcellular localization of the HMA2-GFP fusion reflect the expression pattern of the endogenous gene and the subcellular localization and activity of wild-type HMA2.

We have used this construct to express mutant derivatives of HMA2 in the hma2,hma4 double mutant in order to investigate the role of the N- and C-terminal regions of HMA2 in its function in planta. Specific amino acid substitution or deletion mutations were generated using PCR and, depending on the position of the mutation, particular unique restriction sites in the HMA2p-HMA2-GFP construct were used to substitute the mutant region for the wild-type (Fig. 1). The presence of the GFP was used to confirm that the pattern of expression in root pericycle cells of each construct and the subcellular localization of the mutant protein to the plasma membrane reflected that of the wild-type.

Figure 1.

Cloning of mutant HMA2 derivatives into HMA2p-HMA2-GFP plant expression vector. Numbers indicate amino acid position in the corresponding HMA2 protein. All mutant derivatives were obtained using PCR amplification. The ΔN-MBD and GIAA mutant derivatives were also cloned into an otherwise identical HMA2p-HMA2 vector (lacking the GFP-tag). Cloning of the AKTG and the three ΔC-terminal mutant derivatives into the HMA2p-HMA2-GFP vector displaced the GFP-tag from the expression vector. Separately the three ΔC-terminal mutant derivatives were fused to an amplified GFP PCR product and cloned into the HMA2p-HMA2 plant expression vector. The notation for the ΔC-terminal mutations indicates the last intact HMA2 amino acid in each, respectively.

The same mutations were introduced into a second construct in which only the HMA2 cDNA was expressed from the same HMA2 promoter region (HMA2p-HMA2). Although in the absence of GFP there was no direct means of demonstrating these mutant derivatives were expressed and localized in the same way as the wild-type, these constructs could be used to demonstrate that the presence of the GFP fused to the C-terminus of HMA2 did not obviously influence the capacity of the mutant derivative to function.

The single hma2 mutant has no apparent phenotype and no significant differences from the wild-type in Zn concentrations in roots and shoots (Hussain et al., 2004). Thus, to assess their function, HMA2 mutant constructs were transformed into the hma2, hma4 double mutant, which exhibits a severe shoot Zn-deficiency phenotype, including stunted growth, chlorosis and infertility, when grown in soil. In addition, the hma2,hma4 mutations decrease the ability of plants to translocate Cd from root to shoot to < 3% of wild-type values (Wong & Cobbett, 2008). The single hma4 mutant and consequently the hma2,hma4 mutant complemented with the wild-type HMA2p-HMA2-GFP or HMA2p-HMA2 constructs are phenotypically indistinguishable from the wild-type. However, in these lines shoot Zn concentration is lower than the wild-type (Hussain et al., 2004) and root-to-shoot translocation of Cd is intermediate between the hma2,hma4 double mutant and the wild-type (Wong & Cobbett, 2008). Transgenic lines expressing the HMA2 mutant derivatives were assessed for their growth and fertility, and in some cases shoot Zn concentrations, compared with the hma4 and hma2,hma4 mutants and lines fully complemented by the wild-type constructs. Cd translocation provided a further measure of the function of HMA2 mutant derivatives when transformed into the hma2,hma4 mutant background.

Constructs were transformed into the hma2,hma4 double mutant and BASTA-resistant transgenic lines were selected. These were grown in the presence of added exogenous Zn in order to ensure the rescue of plants in which complementation had not occurred. For each construct, four to six transgenic lines, each segregating for a single transgenic locus according to the segregation of BASTA resistance, were identified and used to generate lines homozygous for the transgene. Two or three lines of each were characterized in detail.

The N-terminal metal-binding domain is essential for HMA2 function in planta

In this study we deleted codons 2–74, inclusive, of HMA2 (Fig. 2). This deletion encompasses the entire N-MBD, including the conserved GICCTSE motif, but does not extend into the predicted first transmembrane domain (TM1). In a separate construct we mutated both the Cys residues within the conserved GIC17C18TSE motif to Ala. These were expressed as both GFP-tagged and nontagged derivatives. As a control the conserved Asp residue of the highly conserved D391KTGT motif that is phosphorylated during the transport reaction cycle was also mutated to Ala. This was expected to abolish function of the protein but not expected to affect its expression or localization. Three transgenic lines expressing each of these constructs were isolated for detailed analysis.

Figure 2.

Schematic diagram of HMA2 showing the mutations generated in this study. The residues in bold, C17C18 and D391, were substituted with Ala residues. Truncations made to the N- and C-terminal regions are indicated by ‘|’. The residues predicted to be part of the first or last transmembrane domains are indicated by a dashed underline. The CysCys-dipeptides and HisXxxHis repeats within the C-terminal region proposed by Eren et al. (2006) to participate in the co-ordination of Zn binding are underlined.

Quantitative RT-PCR of the HMA2 transcript in the GFP-tagged lines (Fig. 3a) demonstrated that each expressed the HMA2 transgene at a level approximately equal to that in the wild-type. Similar results were obtained for the nontagged derivatives (not shown). Each GFP-tagged line was examined by confocal microscopy to examine the expression pattern and subcellular localization of the HMA2-GFP. These observations are shown in Fig. 4 and Supporting Information, Fig. S1. These studies demonstrated that the mutant derivatives were expressed and localized in a manner comparable to the wild-type parental construct.

Figure 3.

Effect of mutations in the N-terminal metal binding domain. (a) The relative transcript expression of the HMA2-GFP, GIAA-GFP and ΔN-MBD-GFP constructs in transgenic lines compared with endogenous HMA2 expression in wild-type (WT) Arabidopsis thaliana plants. Values are means ± SE (n = 3). Significant differences from endogenous HMA2 expression in wild-type as determined by Student's t test are shown (*P < 0.05). (b) Twenty-eight-day-old GIAA-GFP and ΔN-MBD-GFP transgenic plants grown in soil without exogenous supply of ZnSO4. (c) Root-to-shoot Cd translocation of seedlings exposed to 0.2 µm 109CdSO4 for 2 d. Percentage Cd translocation is the proportion of total Cd taken up by plants that was translocated to shoots. Values are means ± SE (n = 6). Significant differences from the hma2,hma4 control as determined by Student's t test are shown (**P < 0.01). In (a), (b) and (c), # refers to the same independent transgenic lines for each construct indicated.

Figure 4.

Expression and localization of wild-type HMA2-GFP and mutant HMA2-GFP derivatives in Arabidopsis thaliana root tissue. Images in the X-Z sections show that all mutant transgenic-GFP fusion proteins were expressed in a similar manner as wild-type HMA2-GFP (in the pericycle cells). Images in the X-Y sections show that all transgenic-GFP fusion proteins were localized to the plasma membrane, with ΔC-707-GFP showing additional intracellular fluorescence. Bars, 20 µm.

Each of the transgenic lines carrying the N-MBD mutant constructs was grown in soil alongside wild-type, hma4 and hma2,hma4 control plants and control hma2,hma4 transgenic lines expressing the wild-type HMA2p-HMA2-GFP or HMA2p-HMA2 constructs or a CaMV35S promoter-GFP construct lacking HMA2 (Fig. 3b). The hma4 control and hma2,hma4 transgenic lines expressing wild-type HMA2 or HMA2-GFP were phenotypically indistinguishable from the wild-type, as expected (not shown). Transgenic lines expressing the (D391A)KTGT mutant were indistinguishable from the hma2,hma4 control (not shown), indicating this mutant was nonfunctional, also as expected. The transgenic lines expressing the N-MBD mutants, both with and without the GFP tag, were also phenotypically indistinguishable from the hma2,hma4 control (Fig. 3b), indicating that there was little, if any, rescue of root-to-shoot Zn translocation in these lines. For each of the N-MBD mutant or wild-type GFP-tagged derivatives, two independent transgenic lines were also assayed for root-to-shoot Cd translocation (Fig. 3c). In all lines tested, the proportion of total Cd taken up by plants that was translocated to shoots (% Cd translocation) (Wong & Cobbett, 2008) was not significantly different from the hma2,hma4 control. Thus, these mutant derivatives, although expressed and localized correctly, appear to be nonfunctional in planta.

The C-terminal domain of HMA2 is not essential for function in planta

To test the role of the C-terminal domain in HMA2 function, we constructed three C-terminal deletions extending from and including codons 708, 831 and 931 through to the end of the protein. These are referred to as ΔC-707, ΔC-830 and ΔC-930, respectively, indicating the last intact amino acid in the mutant protein. The truncated cDNAs were used to replace the full-length HMA2 cDNA in both the HMA2p-HMA2-GFP and HMA2p-HMA2 vectors. Thus, in the former, the truncated HMA2 C-terminus was fused to GFP, while in the latter, the HMA2 derivatives terminated prematurely with a stop codon. The ΔC-707 truncation is identical to the truncated HMA2 derivative used by Eren et al. (2006). This deletion is not expected to extend into the eighth and last predicted transmembrane domain (TM8).

The deletion constructs were transformed into the hma2,hma4 double mutant and three independent transgenic lines for each construct were isolated for further characterization. Each GFP-tagged line was examined by confocal microscopy to determine the expression pattern and subcellular localization of the mutant derivatives ΔC-terminal-GFP in roots. These observations are shown in Fig. 4 and Fig. S1. These studies demonstrated that the mutant derivatives ΔC-830 (Fig. 4) and ΔC-930 (Fig. S1) were expressed and localized in a manner comparable to the wild-type parental construct. In contrast, while the ΔC-707 construct was expressed as expected in the pericycle cells of roots, the subcellular localization was different from the wild-type (Fig. 4). In the ΔC-707 transgenic lines, in addition to the expected plasma membrane localization, green fluorescence was also observed throughout the cells, with a distinct circular nonfluorescing region. Staining of DNA in root cells with DAPI indicated the nonfluorescing region corresponded to the nucleus (not shown). The diffuse green fluorescence observed is likely to be in the cytoplasm (and possibly in some other organelles) and may be a form of GFP from degraded ΔC-707-GFP that has failed to integrate into a membrane. Alternatively, the intact ΔC-707-GFP fusion protein may be mis-localized to internal membranes in the cells.

The nontagged lines were characterized in greater detail. Quantitative RT-PCR of the HMA2 transcript demonstrated that in most cases HMA2 was expressed at a level equal to or greater than the wild-type (Fig. 5a). Although each line contained only a single transgenic locus, it is possible that multiple copies of the T-DNA were integrated at that locus. When grown in soil, the transgenic hma2,hma4 lines expressing the ΔC-830 and ΔC-930 mutant derivatives were indistinguishable from the wild-type or hma4 controls (not shown). In particular, they were fully fertile and set seed as effectively as the wild-type. For lines expressing the ΔC-707 mutant, rosette growth was indistinguishable from the wild-type and plants bolted and self-fertilized effectively. However, in contrast to the other lines, embryo or seed development frequently aborted in these plants. Less than 3% of seeds obtained from these plants were able to germinate. Identical observations were obtained for all independent lines tested (> six) for both the GFP-tagged and nontagged constructs. For all these lines, seed development could be fully rescued by the application of ZnSO4 to the soil. These observations suggest that Zn translocation to shoots, while sufficient for leaf growth and shoot development, was at least partly deficient in the shoot tissues or possibly only in the developing siliques or embryos.

Figure 5.

Effect of mutations to the C-terminal metal binding domain. (a) The relative transcript expression of the HMA2p-HMA2 and the three ΔC-terminal mutant transgenes in transgenic lines compared with endogenous HMA2 expression in wild-type (WT) Arabidopsis thaliana plants. Values are means ± SE (n = 3). Significant differences from endogenous HMA2 expression in wild-type as determined by Student's t test are shown (*P < 0.05). (b) Shoot Zn concentrations in 19-d-old seedlings grown on MMS agar medium. Values are means of five samples of tissues pooled from five plants per sample ± SE. (c) Root-to-shoot Cd translocation of seedlings exposed to 0.2 µm 109CdSO4 for 4 d. Percentage Cd translocation is the proportion of total Cd taken up by plants that was translocated to shoots. Values are means ± SE (n = 12). In (b) and (c) significant differences were determined by Student's t-test (a, different from hma2,hma4; b, different from hma4; bold and nonbold letters indicate P < 0.01 and P < 0.05, respectively). In (a), (b) and (c), # refers to the same independent transgenic lines for each construct indicated.

One of the phenotypes associated with the hma2,hma4 double mutant is a failure of pollen to develop. Although the ΔC-707 transgenic lines appeared to self-fertilize effectively, reciprocal cross-pollinations with wild-type plants were also tested. When wild-type pollen was used to fertilize ΔC-707 transgenic plants, the developing embryos (> 100 embryos from five independent crosses) all aborted. In contrast, pollen from a ΔC-707 line was able to effectively cross-pollinate the wild-type (> 100 viable seeds from a total of three independent crosses).

We assayed the Zn content of the entire aerial portions of 19-d-old transgenic plants expressing the nontagged ΔC-terminal mutant constructs (Fig. 5b). Apart from one of the two ΔC-707 lines, the Zn concentrations in these ΔC-terminal transgenic lines were largely not significantly different from the hma4 control or hma2,hma4 lines transformed with wild-type HMA2p-HMA2. In addition, we measured root-to-shoot Cd translocation in these lines and found that in all lines tested, the % total Cd translocated to shoots was equal to or greater than the hma4 control (Fig. 5c). In general, lines in which Cd translocation was significantly greater than the control corresponded to those in which expression of the transgene was highest. Together these observations suggest that the ΔC-707 mutant is sufficiently active in roots to allow effective translocation of both Zn and Cd to shoots. However, it may not be sufficiently active in some part of the bolt or silique, resulting in the failure of embryos to develop.


Here we describe a useful system for the expression of mutant derivatives of HMA2 in planta in order to assess their function. We believe the promoter region used drives expression of the constructs with a specificity that reflects the expression of the endogenous gene. Previous analysis showed the HMA2p-GUS construct was expressed in developing anthers of a particular stage and this was confirmed by RT-PCR of the endogenous HMA2 transcript (Hussain et al., 2004). In addition, expression was observed in pericycle cells in roots, which are the cells in which Zn accumulates in the hma2,hma4 double mutant, suggesting that is the endogenous site of expression in roots (Hussain et al., 2004; Sinclair et al., 2007). Previous studies have demonstrated that HMA2 itself and the HMA2-GFP fusion protein used here are located at the plasma membrane of cells (Hussain et al., 2004). The wild-type transgenic constructs used here were able to fully complement both the hma2,hma4 Zn-deficiency phenotypes and the inability of the hma2,hma4 double mutant to translocate Cd from root to shoot, indicating that the expression of the constructs and the subcellular localization and activity of their HMA2 or HMA2-GFP proteins reflect that of the wild-type gene and gene product.

Levels of expression of the transgene were generally up to fourfold higher than that of the endogenous HMA2. This may reflect differences in the copy number of the integrated construct in independent transgenic lines. Since similar observations were made for multiple independent lines, we believe the effects of the HMA2 mutations on function are largely independent of transgene copy number. For example, all HMA2 ΔC-707 transgenic lines exhibited the same failure of embryo/seed development and all N-MBD mutant transgenic lines failed to rescue the Zn-deficiency phenotypes. The observations that the wild-type constructs restored shoot Zn content and root-to-shoot Cd translocation to values comparable to the single hma4 mutant suggest that, in general, the levels of expression and activity of HMA2 are similar to those in the wild-type and are neither significantly higher, thereby compensating for the absence of HMA4 as well, nor significantly lower than in the control. Exceptions were some lines in which Cd translocation was significantly higher than in the hma4 control but still intermediate between hma4 and the wild-type and in which HMA2 expression was generally higher.

The C-terminal metal-binding domain of HMA2 is not essential for activity

The deletion of most or some parts of the C-terminal region beyond the last transmembrane domain did not abolish function of the protein. The two smaller truncations appeared to have no effect on function in planta, while the ΔC-707 truncation derivative expressed in transgenic hma2,hma4 plants was unable to fully complement all aspects of the Zn-deficiency phenotype. The ΔC-707 transgenic lines showed normal growth but poor seed set. Shoot Zn content in one of the two lines tested was significantly greater than the hma2,hma4 control, while the other line was no different. However, the difference in shoot Zn concentrations between hma4 and hma2,hma4 controls is < twofold, so a significant difference is difficult to detect. In the Cd translocation assay, where the difference between hma4 and hma2,hma4 is c. 10-fold, the root-to-shoot Cd translocation in the same two lines was equal to or slightly greater than the hma4 control. It is also important to note that Zn shoot concentrations and Cd translocation were measured in plants before bolting and thus give no indication of a possible effect in inflorescences, as might be expected from the fertility phenotype.

Using the GFP-tagged lines, the ΔC-707-GFP derivative appeared to be mis-localized to some extent, unlike the ΔC-830-GFP and ΔC-930-GFP derivatives, which were indistinguishable from the wild-type. It was possible that for the ΔC-707 truncation the fusion of the GFP close to the eighth transmembrane domain may have interfered with the activity or localization of the protein. However, the parallel derivative lacking the GFP tag was also unable to fully complement transgenic plants, resulting in the same failure of embryo development or seed maturation, suggesting the effect was more likely the result of the deletion itself. Some of the GFP fluorescence appeared to be localized at the plasma membrane in the ΔC-707 lines, consistent with the ability of this mutant to complement most aspects of the hma2,hma4 Zn-deficiency phenotype and the component of Cd translocation that can be ascribed to HMA2 activity. It is possible that the ΔC-707-GFP was mis-localized to other membranes in the pericycle cells, although fluorescence appeared rather diffuse. Alternatively, it may be that mis-localized fusion protein was degraded to give rise to a GFP derivative similar in size and location to unfused GFP. Mis-localization may be the primary reason this derivative was unable to fully complement the mutant phenotypes, although an effect on the intrinsic activity of the protein cannot be excluded.

The reason for the mis-localization of the ΔC-707 mutant derivative is not known. There may be signals in the C-terminal region between positions 707 and 830 that are important in directing the localization of the protein. The Cu-transporting ATPases, ATP7A and ATP7B, in mammals are able to traffic from the trans-Golgi to the plasma membrane under conditions of high intracellular Cu. The sequences essential for this trafficking process lie in the C-terminal cytoplasmic domain and include di-leucine motifs (Petris et al., 1998; Francis et al., 1999). These are believed to be the targets for adaptor proteins involved in the targeting of transport proteins to particular membranes within the cell. While there is no evidence for trafficking of P-type ATPases in plant cells, there are a number of di- or tri-leucine sequences in the C-terminal region of HMA2. One of these, L731L, is deleted by the ΔC-707 truncation. In addition, the L684L sequence is expected to lie within TM8, while the L692LL sequence may be situated within TM8 or immediately adjacent to TM8 in the cytoplasmic domain. It is possible both L692LL and L731L contribute to the correct localization of HMA2, and deletion of the latter interferes to some extent with this process. Site-directed mutagenesis of both sites within the wild-type construct would test his hypothesis. Alternatively, an entirely different signal in this region of HMA2 may be involved in its correct localization.

A previous study has characterized the ATPase kinetics and metal dependence of a truncated HMA2 identical to HMA2 ΔC-707 after heterologous expression in yeast (Eren et al., 2006). The ΔC-707-deletion led to a 50% decrease in enzyme turnover rate but had no apparent effect on the relative activation by Zn2+ and Cd2+, suggesting that the C-terminal domain does not control enzyme selectivity. These authors concluded that the C-terminal domain is not essential for enzyme activity but is required for full activity. Our observations are consistent with this hypothesis although the partial loss of function of the ΔC-707 derivative in planta may be more the result of mis-localization than a relatively small effect on enzyme activity.

The N-terminal metal-binding domain is essential for HMA2 activity

A previous study has also examined the role of the N-terminal metal-binding domain of HMA2 (Eren et al., 2007). The purified domain binds Zn and Cd with similar affinities. Deletion of the first 75 amino acids or individual mutations of Cys17 and Cys18 resulted in a c. 50% reduction in activity but did not affect the metal dependence for activity. The authors concluded the N-terminal domain is required for maximum enzyme turnover but is not essential for activity. Our observations suggest the N-terminal domain has a more important role in planta. Although expression and localization were indistinguishable from the wild-type, the N-terminal mutant derivatives were unable either to complement the Zn-deficiency phenotype or increase Cd translocation in the hma2,hma4 double mutant, suggesting they were essentially nonfunctional. Mutation of the N-terminal Cys residues of HMA4 also affected function when expressed in yeast. These mutant derivatives were unable to complement the Cd or Zn sensitivity of ycf1 and zrc1 yeast mutants, respectively (Verret et al., 2005). Our data suggest that the N-terminal domain, unlike the C-terminal domain, is essential for function in planta. If HMA2 were to parallel HMA4, this may also be true for heterologous expression in yeast. In the in vitro assay using HMA2 expressed in yeast membrane vesicles, only a 50% decrease in activity was observed (Eren & Arguello, 2004). Since activity in this assay is dependent on millimolar concentrations of Cys, it may not truly reflect function in vivo. Overall, data suggest an important role for the N-terminal metal binding domain in HMA function.