A Common Highly Conserved Cadmium Detoxification Mechanism from Bacteria to Humans

Cadmium poses a significant threat to human health due to its toxicity. In mammals and in bakers' yeast, cadmium is detoxified by ATP-binding cassette transporters after conjugation to glutathione. In fission yeast, phytochelatins constitute the co-substrate with cadmium for the transporter SpHMT1. In plants, a detoxification mechanism similar to the one in fission yeast is supposed, but the molecular nature of the transporter is still lacking. To investigate further the relationship between SpHMT1 and its co-substrate, we overexpressed the transporter in a Schizosaccharomyces pombe strain deleted for the phytochelatin synthase gene and heterologously in Saccharomyces cerevisiae and in Escherichia coli. In all organisms, overexpression of SpHMT1 conferred a markedly enhanced tolerance to cadmium but not to Sb(III), AgNO3, As(III), As(V), CuSO4, or HgCl2. Abolishment of the catalytic activity by expression of SpHMT1K623M mutant suppressed the cadmium tolerance phenotype independently of the presence of phytochelatins. Depletion of the glutathione pool inhibited the SpHMT1 activity but not that of AtHMA4, a P-type ATPase, indicating that GSH is necessary for the SpHMT1-mediated cadmium resistance. In E. coli, SpHMT1 was targeted to the periplasmic membrane and led to an increased amount of cadmium in the periplasm. These results demonstrate that SpHMT1 confers cadmium tolerance in the absence of phytochelatins but depending on the presence of GSH and ATP. Our results challenge the dogma of the two separate cadmium detoxification pathways and demonstrate that a common highly conserved mechanism has been selected during the evolution from bacteria to humans.

Cadmium poses a significant threat to human health due to its toxicity. In mammals and in bakers' yeast, cadmium is detoxified by ATP-binding cassette transporters after conjugation to glutathione. In fission yeast, phytochelatins constitute the cosubstrate with cadmium for the transporter SpHMT1. In plants, a detoxification mechanism similar to the one in fission yeast is supposed, but the molecular nature of the transporter is still lacking. To investigate further the relationship between SpHMT1 and its co-substrate, we overexpressed the transporter in a Schizosaccharomyces pombe strain deleted for the phytochelatin synthase gene and heterologously in Saccharomyces cerevisiae and in Escherichia coli. In all organisms, overexpression of SpHMT1 conferred a markedly enhanced tolerance to cadmium but not to Sb(III), AgNO 3 , As(III), As(V), CuSO 4 , or HgCl 2 . Abolishment of the catalytic activity by expression of SpHMT1 K623M mutant suppressed the cadmium tolerance phenotype independently of the presence of phytochelatins. Depletion of the glutathione pool inhibited the SpHMT1 activity but not that of AtHMA4, a P-type ATPase, indicating that GSH is necessary for the SpHMT1-mediated cadmium resistance. In E. coli, SpHMT1 was targeted to the periplasmic membrane and led to an increased amount of cadmium in the periplasm. These results demonstrate that SpHMT1 confers cadmium tolerance in the absence of phytochelatins but depending on the presence of GSH and ATP. Our results challenge the dogma of the two separate cadmium detoxification pathways and demonstrate that a common highly conserved mechanism has been selected during the evolution from bacteria to humans.
Cadmium is a trace element, the presence of which in the environment is essentially due to human activities. It is a highly toxic non-biological heavy metal able to enter living cells via transporters usually used for the uptake of essential cations such as calcium, iron, zinc, and so forth (1). The reactivity of cadmium with thiol groups and its ability to displace essential biological metals result in oxidative stress and eventually cell death (2). To cope with cadmium toxicity, living organisms have developed different strategies.
In animals, as in the bakers' yeast cytoplasmic cadmium is complexed with the thiol tripeptide glutathione, a general redox regulator (3,4). Bis(glutathionato)-cadmium complexes (Cd-GS 2 ) 4 are then driven from the cytoplasm to lesser sensitive cellular compartments by dedicated transporters. The prototypical transporter of Cd-GS 2 is the GS-X pump, ScYCF1, in Saccharomyces cerevisiae (5) and, even if still controversial, to a lesser extent HsMRP1 in humans (6). HsMRP1 probably acts as an efflux pump at the plasma membrane, delivering cadmium in the extracellular medium, whereas ScYCF1 allows sequestration of cadmium into the vacuole (5). A study of a deficient Scycf1 strain has shown that it was extremely cadmium-sensitive, pointing to a major role of ScYCF1 in cadmium tolerance and detoxification (5). Additionally, ScYCF1 was also found involved in arsenic, antimony, mercury, and lead detoxification (references cited within Ref. 7).
A second strategy of cadmium detoxification is found in plants (8), with the exception of mosses, and in the fission yeast Schizosaccharomyces pombe (9). In that case, cytoplasmic cadmium is chelated by enzymatically synthesized thiol peptides, phytochelatins of the structure (␥-Glu-Cys) n -Gly, where n equals 2-11 (10). These peptides are produced upon cadmium exposure by the constitutively expressed phytochelatin synthase enzyme (11), the structure of which has been recently resolved (12). The complex phytochelatin-cadmium (PC-Cd) is then transferred to the vacuole by undetermined transporter(s) in plants and by the ABC half-transporter, SpHMT1, in S. pombe (13,14). It is interesting to note that in S. pombe, a mutation of the phytochelatin synthase enzyme, SpPCS, or of the SpHMT1 transporter led to similar cadmium-hypersensitive phenotypes (15). This indicates that the transfer of PC-Cd complexes from the cytoplasm to the vacuole is essential in cadmium resistance in this organism.
In mammals, besides GSH, cadmium can be detoxified after association with metallothioneins, a superfamily of ubiquitous cysteine-rich low molecular weight proteins. These biomolecules, discovered as cadmium-containing proteins in horse kidney, have extremely high metal and sulfur contents (up to 10% w/w) (16).
More recently, it appears that cadmium can be taken up, as a free metallic cation, by energized transporters from the P 1b -ATPases family. For instance, CadA and AtHMA4 are efficient cadmium transporters, respectively, in Listeria monocytogenes and in Arabidopsis thaliana (17,18). However, if the detoxification activity of ScYCF1 (or HsMRP1 in humans) is GSH-dependent, cadmium transporters from the P 1b -ATPases family are functional in the absence of GSH.
This scheme, discriminating different strategies engaged by plants and animals in cadmium detoxification, has recently been ruled out by the discovery of a functional phytochelatin synthase enzyme in the worm Caenorhabditis elegans that is able to complement the cadmium sensitivity of an S. pombe PCS knock-out strain (19). CePCS1-deficient worms were found markedly more sensitive to cadmium intoxication, leading to the first demonstration of the PCmediated detoxification of cadmium in an animal (19). In fact, potential phytochelatin synthase orthologs have been found in a large list of eukaryotes outside the plant kingdom (20). The identification of an ortholog of the ABC transporter SpHMT1 in C. elegans, CeHMT-1, has completed the homology between the S. pombe and C. elegans pathways for cadmium detoxification (15,19). Strikingly, CeHMT-1-deficient worms were found strongly more sensitive to cadmium than CePCS1-deficient worms (19), suggesting that the role of CeHMT-1 was not limited to the transport of PC-Cd in cadmium tolerance.
In the present study, we show that SpHMT1 overexpression is able to confer cadmium tolerance in organisms devoid of phytochelatins, such as S. cerevisiae and Escherichia coli. This function requires the presence of glutathione in the cell and a functional ATP-binding domain in the protein.
The S. pombe strains were the previously described the SpHMT1-deficient ⌬hmt1 strain LK100 and the corresponding S. pombe wild-type strain Sp223 (h Ϫ ade6-216, leu1-32, ura4-294) (13) as well as the phytochelatin synthase-deficient ⌬pcs strain Sp27 and the corresponding S. pombe wild-type strain FY254 (h Ϫ ade6-M210 leu1-32 ura4-⌬18 can1-1) (21). Transformation of S. pombe was performed as described previously (22). Cells were routinely grown at 30°C in complete (YPD medium or Edinburgh's minimal medium (EMM) supplemented appropriately (i.e. without Leu (EMMϪLeu) as described previously (23). Medium containing different concentrations of CdCl 2 was prepared immediately prior to the growth experiment. Plates used for the cadmium spot test were prepared by adding the indicated concentration of CdCl 2 to the minimal plate medium.
Plasmid constructions and gene expression were performed in the E. coli strains Top10 (Invitrogen). Bacteria cells were grown at 37°C in liquid Luria-Bertani (LB) medium supplemented with appropriate antibiotics.
Growth Inhibition by Cadmium and Determination of Metal Content-To examine growth in liquid medium, yeasts grown to saturation in EMMϪLeu medium were subcultured at a starting A 600 of 0.1 into the same medium containing different concentrations of CdCl 2 , and the extent of growth after 48 h was determined by measuring the A 600 . The growth rate of each culture was also determined by taking A 600 readings at multiple intervals throughout the entire 48-h period. To examine growth on plates, cells were grown overnight to saturation in EMMϪLeu or S-URA 2% raffinose medium. This overnight culture was diluted in EMMϪLeu or S-URA 2% galactose medium to an A 600 of 0.4, which in turn was diluted in 10-fold increments. Two microliters of each 10-fold dilution was spotted onto EMMϪLeu or S-URA 2% galactose plates containing different concentrations of CdCl 2 and incubated for 7 or 3 days at 30°C. In the case of E. coli, single colonies of bacteria transformed with pUC19 or pUC-SpHMT1-GFP were grown to saturation in LB medium (37°C, 180 rpm) and were subcultured at a starting A 600 of 0.05 into the same medium containing different concentrations of CdCl 2 . The growth rate of each culture was determined by taking A 600 readings at multiple intervals throughout the entire 6-h period. Determination of cadmium content was realized by induced coupled plasma experiments. After centrifugation, pellets were washed three times with 10 mM EDTA, dried for 48 h at 50°C, and mineralized. The metal content was determined using ICP (ICP-OES Vista-MPX, Varian). Metal resistance assays in the presence of BSO were carried out as described previously (7).
Localization of SpHMT1-The localization of the SpHMT1-EGFP-N2 fusion protein was examined in LK100 S. pombe and in the Y04069 S. cerevisiae transformed strains. After overnight culture, yeast cells (0.8 OD) were resuspended in 1 ml of selective medium with galactose containing 8 M FM4-64 (Red Synaptracer 3-2 Interchim FP-41109A) After 15 min at 30°C, cells were centrifuged and washed for 2 h at 30°C under agitation with 1 ml of selective medium containing galactose. Yeast cells were washed two times in phosphate-buffered saline and resuspended in 1 ml of water before observation on a glass slide. The observations were done with a confocal laser scanning microscope (Leica TCS SP2 AOBS) fitted with a krypton/argon laser at ϫ100 magnification. Excitations were performed at 488 or 568 nm for EGFP-N2 (excitation peak at 488 nm and emission peak at 507 nm) and FM4-64 (excitation peak at 515 nm, emission peak at 640 nm), respectively. The fluorescence was collected through 510 -570 and 660 -800 nm for EGFP and FM4-64, respectively. The localization of SpHMT1 in bacteria was conducted with log-phase cells expressing GFP-tagged SpHMT1. Cells were examined at 100-fold magnification on a poly-lysine-coated slide by using a Nikon Optiphot-2 fluorescence microscope. The fluorescence was collected through a band-pass filter at 510 -570 nm (EGFP-N2 excitation peak at 488 nm, emission peak at 507 nm). Images were captured with a Zeiss AxioCam camera and its dedicated software.
Electron Microscopy-Bacteria transformed with pUC19 or pUC-SpHMT1-GFP were fixed overnight in 4% paraformaldehyde and 0.1% glutaraldehyde in phosphate-buffered saline and extensively washed in phosphate-buffered saline. Cell pellets were infiltrated in a 2% agar gel and embedded in Unicryl. Eighty nm-tick sections were cut, collected on EM grids, preincubated in 20 mM Tris buffer (pH 7.5) containing 0.1% bovine serum albumin, 0.1% fish gelatin, 0.05% Tween 20 (buffer-T), and followed by a 90-min incubation in the same buffer-T containing a 1:90 dilution of anti-GFP polyclonal antibodies (AbCam antibody ab290). Sections were washed three times in buffer-T and then incubated in a 1:25 dilution of 10-nm goldconjugated secondary antibodies for 45 min (Amersham Biosciences). After washing in Tris, sections were stained with uranyl acetate and lead citrate and photographed on a FEI CM12 microscope (FEI, Eindhoven, The Netherlands). For statistical analysis, pictures were randomly taken of each of the two samples and of the gold particles located along the plasma membrane, and cytoplasmic background labeling was counted in ϳ80 different cells from each sample.
Membrane Extraction and Western Blot Analysis-Transformed bacteria were grown in 5 ml of LB ampicillin and subcultured in 1 liter of LB for 4 h at 37°C. The culture was stopped when the A 600 reached 1.5. Bacteria were collected by low speed centrifugation (4000 ϫ g, 15 min at 4°C). All following steps were carried at 4°C. Pellets were resuspended in the buffer 50 mM Tris-HCl, 5 mM MgCl 2 , 1 mM dithiothreitol, 5 M leupeptin, 5 M pepstatin A, and 1 mM phenylmethylsulfonyl fluoride. Bacteria were lysed by two successive passages through a French press (18,000 p.s.i.). EDTA was then added at 10 mM. Unbroken bacteria and membrane residues were removed by a 30-min centrifugation at 15,000 ϫ g. Membranes were collected by centrifugation at 100,000 ϫ g and washed in 15 ml of the buffer 50 mM Tris-HCl, pH 8, 5 M leupeptin, 5 M pepstatin A, and 1 mM phenylmethylsulfonyl fluoride and centrifuged again at 100,000 ϫ g. The pellet was finally suspended in 50 mM Tris-MES, pH 8, 300 mM sucrose. Membrane extracts were denaturated at room temperature and subjected to SDS-PAGE electrophoresis. After transfer onto polyvinylidene difluoride membrane, proteins were detected with an anti-GFP antibody (monoclonal antibody JL-8, BD Biosciences; dilution 1:5000) as already described (7).
Periplasmic Extraction-The periplasm extraction was done as already described (24). Briefly, bacteria transformed with pUC19 or pUC-SpHMT1-GFP grown to saturation in LB medium supplemented with 100 M ampicillin were subcultured at a starting A 600 of 0.1 into 50 ml of LB with 50 M cadmium and grown overnight. The culture was stopped when the A 600 reached 2.6. Bacteria were collected by low speed centrifugation (4000 ϫ g, 15 min at 4°C). Pellets were resuspended in 1.5 ml of Tris-HCl, pH 8, by briefly vortexing, and 600 l of CHCl 3 was added. After brief vortexing, the tubes were maintained at room temperature for 15 min, and then 5 ml of 10 mM Tris-HCl, pH 8, was added. Intact cells were separated by centrifugation (6000 ϫ g, 20 min). The supernatant containing the SpHMT1 Requires Glutathione but Not Phytochelatins periplasm was collected, and 1 ml was used to determine the cadmium periplasm content by ICP.

RESULTS AND DISCUSSION
Overexpression of SpHMT1 in S. pombe Rescues Cadmium Tolerance in a Sphmt1-deleted Strain-As already reported by Ortiz et al. (13,14), an S. pombe mutant having a defect in the Sphmt1 gene and unable to accumulate PC-Cd complexes in the vacuole was found to be cadmium-sensitive. This defect was rescued by expression of the wild-type SpHMT1 protein in the mutant (Fig. 1A). Three independent clones of S. pombe strains expressing a fusion of SpHMT1 with the green fluorescent protein also exhibited an enhanced cadmium tolerance when compared with the Sphmt1 deletion strain (Fig. 1A). Thus, the C-terminal fusion of GFP did not impair the protein function as already observed for other ABC transporters such as HsMRP1 or ScYCF1 (7,25). The fusion protein co-localized with the vacuolar marker FM4-64 (Fig. 1B), in accordance with the location of the transporter at the vacuolar membrane previously deduced from membrane fractionation (13). A similar vacuolar location has been observed for CeHMT-1 in S. pombe using a similar GFP fusion strategy (19). The protein was detected in Western blots using anti-GFP antibodies at the 112 kDa expected molecular mass (Fig. 1C). We recently reported that ScYCF1 confers a tolerance to other metal(loid)s besides cadmium (7). Surprisingly, among the metal(loid)s tested, SpHMT1 was able to confer a tolerance to cadmium but not to Sb(III), AgNO 3 , As(III), As(V), CuSO 4 , or HgCl 2 (supplemental Fig. 1).
Overexpression of SpHMT1 Complements Cadmium Tolerance of an S. pombe Strain Devoid of Phytochelatins-To determine whether SpHMT1 is an exclusive PC-Cd transporter, we used the Sp254 S. pombe strain deleted for the PCS gene, Sp27. This strain is unable to synthesize phytochelatins from glutathione and is highly sensitive to cadmium (21). In this genetic background, overexpression of SpHMT1 was able to completely restore cell tolerance to cadmium in the drop test as well as in liquid media experiments (Fig. 1, D and E). The cadmium resistance was not restored when the ⌬pcs strain was transformed with the empty vector pART (Fig. 1E). These results indicate that the lack of phytochelatins can be compensated by SpHMT1 overexpression and strongly suggest that SpHMT1 can contribute to cadmium tolerance in S. pombe independently of any phytochelatin synthesis. To confirm this hypothesis, expression of SpHMT1 was investigated in organisms naturally devoid of PC.
SpHMT1 Rescues Cadmium Tolerance in the Hypersensitive ⌬ycf1 Strain of S. cerevisiae-In S. cerevisiae, the cadmium detoxification strategy differs from the one reported in S. pombe (5,13,14). There is no PCS gene ortholog in the genome of S. cerevisiae, and cadmium detoxification occurs  FEBRUARY 20, 2009 • VOLUME 284 • NUMBER 8 mainly by the formation of bis(glutathionato)-cadmium complexes in the cytoplasm followed by their transfer into the vacuole by the full-sized ABC transporter ScYCF1 (5). In accordance, overexpression of ScYCF1 in the S. cerevisiae wild-type context allowed an increase in cadmium tolerance ( Fig. 2A). Remarkably, overexpression of SpHMT1 was even more efficient, allowing a yeast cadmium tolerance up to 200 M. The ability of SpHMT1 to confer cadmium tolerance was even more pronounced in the cadmium-hypersensitive Scycf1 mutant context ( Fig. 2A). In this strain, overexpression of SpHMT1 was able to induce a cadmium resistance greater than the one observed after overexpression of the homologous transporter, ScYCF1. A Western blot analysis revealed that the fusion protein SpHMT1-GFP was reproducibly and largely more detected than ScYCF1-GFP based on a similar quantity of proteins loaded on gel (Fig. 2B). This result might argue with the great efficiency of SpHMT1 in cadmium detoxification. In contrast with its location at the vacuolar membrane in S. pombe, in S. cerevisiae, SpHMT1-GFP was located in vesicles surrounding the main vacuole (Fig. 2C). Experiments conducted here, with the two major yeast models that diverged ϳ400 million years ago (26), S. cerevisiae using GSH to complex cadmium and S. pombe using PC, confirm that SpHMT1 activity in cadmium tolerance is independent of PC. Altogether these results demonstrate that SpHMT1 can use another substrate than PC for the cadmium resistance activity. To investigate further possible substrates coming from the thiol pathway, a S. cerevisiae yeast strain depleted of GSH was used.

SpHMT1 Requires Glutathione but Not Phytochelatins
Cadmium Tolerance Conferred by SpHMT1 Depends on Glutathione Synthesis-The comparison of the action mechanism of ScYCF1, the activity of which is strictly coupled to the presence of GSH and of SpHMT1, depending on the presence of PC, lead us to evaluate the hypothesis that GSH might be a candidate as a co-substrate substitute for PC in cadmium transport by SpHMT1. This hypothesis was tested by a direct application in the culture medium of DL-BSO, an inhibitor of ␥-glutamyl cysteine synthetase (27). In S. cerevisiae, after application of 5 mM BSO, the GSH pool was reduced to 32% of its initial content after 6 h and to 50% after 24 h (7). In the absence of metal, application of 5 mM BSO had no effect on growth of the different yeast strains studied (Fig. 2D). In the presence of 5 mM BSO combined with 50 or 100 M cadmium, the growth of wild-type strain was diminished by about 30 or 50%, respectively, as already reported (7). Similarly, in the presence of BSO, the cadmium tolerance of Scycf1 strains expressing ScYCF1-GFP or SpHMT1-GFP was diminished. In contrast, cadmium tolerance resulting from an overexpression of the P 1B -ATPase AtHMA4, (the plasma membrane Arabidopsis Cd-ATPase) was unaffected in the presence of BSO (Fig. 2D) (18). Thus, it is FIGURE 2. SpHMT1-GFP confers cadmium (Cd) tolerance to the S. cerevisiae ⌬ycf1 yeast strain. A, heterologous overexpression of SpHMT1-GFP conferred a higher cadmium tolerance to a wild-type S. cerevisiae yeast strain (Y00) than overexpression of the endogenous transporter ScYCF1. GFP was expressed as a control. Similar results were obtained in the ⌬ycf1 context (Y04 strain) B, SDS-PAGE electrophoresis followed by a Western blot analysis using anti-GFP antibodies on cell lysate of S. cerevisiae yeast strains overexpressing SpHMT1-GFP or ScYCF1-GFP. The expression level of SpHMT1-GFP was found to be strongly higher than that of ScYCF1-GFP. C, in confocal microscopy, SpHMT1-GFP fluorescence in S. cerevisiae cells localized in vesicles surrounding the vacuoles. D, sensitivity to the glutathione inhibitor DL-BSO of the different yeast strains overexpressing either the GFP as a control or ScYCF1-GFP, SpHMT1-GFP, or AtHMA4-GFP. Experiments were done at least in triplicate, and a typical result is shown. likely that SpHMT1, similar to ScYCF1, is able to use glutathione as a co-substrate in cadmium transport.
SpHMT1 Detoxification Activity Depends on ATP Hydrolysis and Multimerization-The transport activity of ABC transporters is linked to the binding of ATP and its hydrolysis at the level of nucleotide-binding domains (28). To ensure that cadmium tolerance conferred by SpHMT1 overexpression was due to a transport activity, a SpHMT1 K623M variant of the protein, unable to bind ATP, was produced. Overexpression of this variant did not confer cadmium tolerance to Sphmt1 or Scycf1 deletion strains (Fig. 3A), demonstrating that the binding of ATP by the protein is crucial in the cadmium resistance process. Using Western blot, we confirmed that the SpHMT1 K623M variant was properly expressed at a level comparable with the wild-type version of SpHMT1 (Fig. 3B). Moreover, it is interesting to note that SpHMT1 K623M had a dominant negative effect when overexpressed in the wild-type S. pombe strain (Fig.  3A). ABC transporters are generally multimeric proteins, and the simplest explanation of this dominant negative effect is that this non-functional polypeptide invalidates the transport activity of the multimeric protein. The fact that both S. cerevisiae and S. pombe yeast cells transformed by the SpHMT1 K623M variant plasmid were sensitive to cadmium intoxication strongly suggests that the resistance given by the wild-type SpHMT1 is not due to a chelation process but rather due to an active transport mechanism.
SpHMT1 Is Expressed at the Periplasm of E. coli Cells-Because a recent study has shown that a few amounts of PC were synthesized in S. cerevisiae (29), SpHMT1-GFP was also expressed in a prokaryote, E. coli DH10B, fully devoid of PC. Under fluorescence microscopy, only cells transformed with pUC-SpHMT1-GFP were exhibiting a green fluorescent signal (Fig. 4A). Proteins from the soluble and microsomal fractions were extracted, separated by SDS-PAGE, blotted, and immunodetected with an anti-GFP antibody (Fig. 4B). A 112-kDa apparent polypeptide was detected in the SpHMT1-GFP microsomal fraction that is consistent with the predicted molecular mass of SpHMT1-GFP. A putative dimeric complex was detected at 250 kDa as well as bands at around 65 kDa that could be proteolytic fragments from SpHMT1, as already described in the natural host (13). Finally, taking advantage of the C-terminal EGFP tag and 10-nm gold-conjugated secondary antibodies, immunocytolocalization of the transporter was investigated by electronic microscopy in E. coli (Fig. 4C). Although a mean of 1.6 Ϯ 0.4 plasma membrane gold particles per m 2 was observed in 86 independent cells expressing the empty vector, this number was significantly increased to 3.84 Ϯ 0.4 (p ϭ 2.4 10 Ϫ4 ) in 78 bacteria expressing pUC-HMT1-GFP. These data demonstrate that in a heterologous system such as E. coli, without any eukaryotic signal peptide, HMT1-GFP can be expressed, preferentially in the periplasmic membrane.
SpHMT1 Enhances E. coli Cadmium Tolerance-In DH10B E. coli cells transformed by pUC19, the growth was drastically affected when cadmium concentration was over 25 M (Fig.  4D). Conversely, overexpression of SpHMT1-GFP induced cell tolerance up to 100 M cadmium (Fig. 4, D and E). Similar results were obtained using other E. coli strains, DH5 and JC7623 (data not shown). Because E. coli does not naturally produce PC, these results confirm that SpHMT1 can contribute to cadmium tolerance by another way than PC-Cd transport. In yeast, cadmium tolerance conferred by SpHMT1 results from cadmium sequestration into the vacuole. In E. coli, the cadmium content in the periplasmic space of SpHMT1-GFP-overexpressing bacteria was found 3-fold the one in control bacteria (Fig. 4F), suggesting that in E. coli, SpHMT1 conferred cadmium tolerance through cadmium sequestration into the periplasmic space. Analysis by exclusion chromatography of cadmium complexes in periplasmic spaces of pUC SpHMT1-GFP-transformed bacteria and pUC19-transformed bacteria revealed a similar profile, the only difference being a higher level of the different complexes in SpHMT1-GFP-expressing bacteria. GSH-Cd complexes were resolved but did not form the major cadmium ligand. In the two genetic backgrounds, the majority of cadmium was associated to a large peak revealing a high molecular mass complex (7 kDa), the nature of which was not resolved.
Conclusion-We have shown that SpHMT1 is able to rescue the cadmium-sensitive phenotype of an S. pombe mutant deficient for the PCS gene, suggesting that PC-Cd is not the only substrate of SpHMT1. When heterogously overexpressed in the Scycf1 deletion mutant of S. cerevisiae, which is cadmiumhypersensitive and devoid of PCS, SpHMT1 also allowed recovery of a wild-type phenotype. The complementation of the Scycf1 strain was found BSO-sensitive, suggesting that cadmium conjugated to glutathione is a substrate of SpHMT1. Overexpression of a SpHMT1 K623M variant exhibiting a point mutation in the nucleotide-binding domain led to a suppression of the cadmium tolerance, disclosing a possible cadmium chelation by the overexpressed protein. This mutated form had a dominant negative effect, suggesting that SpHMT1 functions as a multimeric protein. Because S. cerevisiae has been shown to synthesize a few amounts of phytochelatins (29), SpHMT1 was also ovexpressed in E. coli. Even in the context of this prokaryote, fully devoid of PC, SpHMT1 was able to confer a strong tolerance to cadmium, accompanied by a different partitioning of cadmium in the bacteria. These results demonstrate that SpHMT1 is not an exclusive PC-Cd transporter but that it can likely take in charge cadmium-glutathione conjugates. This gives an explanation for the stronger phenotype of the CeHMT-1-deficient mutant when compared with the CePCS1-deficient mutant under cadmium stress. The absence of phytochelatin would notably affect free cadmium chelation in the cytoplasm, thus conferring cadmium sensitivity. Invalidation of CeHMT-1 has a higher impact on cadmium resistance through the inval-idation of cadmium sequestration in the vacuole. Thus, invalidation of this transporter results in critical levels of cadmium in the cytoplasm, leading to a strong toxicity.
Altogether these results demonstrate that SpHMT1 is a polyvalent transporter that can take in charge different forms of cadmium complexes, as illustrated on Fig. 5. It is the first study showing that an ABC transporter (HMT1) can accommodate glutathione, as well as phytochelatin-cadmium complexes. This information would help identify the transporter in charge of cadmium detoxification in plants, a protein that has long been sought after by different groups for many years based on its ability to use phytochelatin-cadmium complexes. Our results challenge the dogma of the two separate cadmium detoxification pathways and demonstrate that a common highly conserved cadmium detoxification mechanism has been selected during the evolution from bacteria, including plants and yeast, to humans. Moreover, besides the fact that cadmium is implicated in cancer in humans (in liver and kidney), the nature of the human transporter responsible for cadmium detoxification remains an open question. Our results would lead to building new models of heavy metal detoxification to prevent/cure diseases linked to the exposition of humans to this toxin.