CeHMT-1, a Putative Phytochelatin Transporter, Is Required for Cadmium Tolerance in Caenorhabditis elegans*

Phytochelatins (PCs), (γ-Glu-Cys)n Gly polymers that were formerly considered to be restricted to plants and some fungal systems, are now known to play a critical role in heavy metal (notably Cd2+) detoxification in Caenorhabditis elegans. In view of the functional equivalence of the gene encoding C. elegans PC synthase 1, ce-pcs-1, to its homologs from plant and fungal sources, we have gone on to explore processes downstream of PC fabrication in this organism. Here we describe the identification of a half-molecule ATP-binding cassette transporter, CeHMT-1, from C. elegans with an equivalent topology to that of the putative PC transporter SpHMT-1 from Schizosaccharomyces pombe. At one level, CeHMT-1 satisfies the requirements of a Cd2+ tolerance factor involved in the sequestration and/or elimination of Cd·PC complexes. Heterologous expression of cehmt-1 in S. pombe alleviates the Cd2+-hypersensitivity of hmt– mutants concomitant with the localization of CeHMT-1 to the vacuolar membrane. Suppression of the expression of ce-hmt-1 in intact worms by RNA interference (RNAi) confers a Cd2+-hypersensitive phenotype similar to but more pronounced than that exhibited by ce-pcs-1 RNAi worms. At another level, it is evident from comparisons of the cell morphology of ce-hmt-1 and cepcs-1 single and double RNAi mutants that CeHMT-1 also contributes to Cd2+ tolerance in other ways. Whereas the intestinal epithelial cells of ce-pcs-1 RNAi worms undergo necrosis upon exposure to toxic levels of Cd2+, the corresponding cells of ce-hmt-1 RNAi worms instead elaborate punctate refractive inclusions within the vicinity of the nucleus. Moreover, a deficiency in CeHMT-1 does not interfere with the phenotype associated with CePCS-1 deficiency and vice versa. Double ce-hmt-1; ce-pcs-1 RNAi mutants exhibit both cell morphologies when exposed to Cd2+. These results and those from our previous investigations of the requirement for PC synthase for heavy metal tolerance in C. elegans demonstrate PC-dependent, HMT-1-mediated heavy metal detoxification not only in S. pombe but also in some invertebrates while at the same time indicating that the action of CeHMT-1 does not depend exclusively on PC synthesis.

The toxicity of supraoptimal levels of the essential heavy metals copper and zinc and of trace or higher levels of the nonessential heavy metals cadmium, arsenic, mercury, and lead is thought to result from the displacement of endogenous co-factors from their cellular binding sites, thiol-capping of essential proteins and peptides, and promotion of the formation of reactive oxygen species (1). In humans, the repercussions of heavy metal exposure can range from acute poisoning to progressive kidney, liver, and lung dysfunction and, in some instances, cancer. Although its true clinical and epidemiological significance remains to be determined, chronic exposure to heavy metals is often associated with muscular and neurological degenerative conditions reminiscent of muscular dystrophy, multiple sclerosis, Alzheimer disease, and Parkinson disease (2)(3)(4).
Three classes of heavy metal-binding peptides are known to participate in the homeostasis and detoxification of heavy metals in most animals. These are the ubiquitous thiol tripeptide glutathione (GSH), a family of small (4 -8 kDa) cysteine-rich proteins termed metallothioneins, and several higher molecular mass albeit sequence-unrelated metal-binding proteins. In all of the plants studied and in some fungi (as exemplified by Schizosaccharomyces pombe and Candida glabrata) and some animals (as exemplified by the model nematode Caenorhabditis elegans), a third class of cysteine-rich peptides termed phytochelatins (PCs) 1 has also been shown to be involved in heavy metal detoxification.
In those systems that have been studied in sufficient detail, namely the fission yeast S. pombe and plants, PCs thiol coor-dinate and chelate heavy metals to promote their removal from the cytosol by vacuolar sequestration (1,13,14). In the most thoroughly characterized of these systems, S. pombe, the vacuolar sequestration of heavy metal-PC complexes has been inferred to be catalyzed by a vacuolar membrane-localized "half-molecule" ATP-binding cassette (ABC) transporter, SpHMT1 (S. pombe heavy metal tolerance factor 1) (15,16).
Canonical, "full-molecule" ABC proteins consist of two transmembrane domains (TMDs), each of which contains 4 -6 transmembrane ␣-helices, two nucleotide binding folds (NBFs), each of which contains the Walker A and B boxes, and the ABC signature motif. So-called half-molecule ABC transporters, by contrast, contain only one TMD and one NBF (17). In addition to these core domains, some members of both types of ABC transporters possess a hydrophobic N-terminal extension (NTE) and a linker (L0) sequence contiguous with the TMD or the NBF. In the case of SpHMT-1, the mature 90.5-kDa polypeptide species contains an NTE containing five hydrophilicity minima and an L0 linker sequence contiguous with one TMD and one NBF (15).
The precise mechanism of action of SpHMT1 is unknown. On the one hand, it has been demonstrated that S. pombe hmt1 Ϫ mutants are hypersensitive to Cd 2ϩ in the growth medium (15) and that vacuoles isolated from S. pombe overexpressing plasmid-borne SpHMT1 mediate the MgATP-dependent, vanadateinhibitable uptake of Cd⅐PC complexes and apo-PCs (16). On the other hand, the relevance of what has been inferred from investigations of S. pombe to the equivalent processes in plants is contentious. HMT-1 homologs have yet to be isolated from or even detected in the genomes of vascular plants (18), although an MgATP-energized, vanadate-inhibitable transport pathway for PCs, analogous to that identified in S. pombe, has been characterized in vacuolar membrane vesicles purified from oat roots (19).
Here we report the identification and functional characterization of a half-molecule ABC transporter from C. elegans that is crucial for Cd 2ϩ tolerance in the intact organism. In so doing, we establish that an HMT-1-dependent pathway for the detoxification of Cd 2ϩ is not restricted to fungi and plants but is also operative in some animals. That said, the results of these investigations also reveal a versatility for CeHMT-1 in the intact worm and possibly its equivalents in other organisms that has not been considered previously. In particular, whereas the basic structural and functional properties of CeHMT-1 are consistent with it being an ortholog of SpHMT1, comparisons of the phenotypes of ce-hmt-1 and ce-pcs-1 single RNAi worms and ce-hmt-1;ce-pcs-1 double RNAi worms yield findings that invoke auxiliary roles for CeHMT-1 in Cd 2ϩ detoxification that do not depend only on upstream PC synthesis. Isolation and Heterologous Expression of ce-hmt-1-The cDNA corresponding to the coding sequence for CeHMT-1 was amplified from C. elegans N2 strain RNA by reverse transcription (RT) PCR. Total RNA was extracted in TRIzol R Reagent (Invitrogen), according to the manufacturer's recommendations, and the first strand was synthesized by standard procedures using a SuperScript preamplification system (Invitrogen). The primers for amplification of the 2.4-kb ce-hmt-1 open reading frame (5Ј-CAAATTCCGCTCGAGATGGGCTTTTCACCA-TTT-3Ј and 5Ј-CTACGGAAGCTCCTCGCCCGGGATTGGCAAG-3Ј) were designed to generate XhoI and SmaI restriction sites at the 5Ј-and 3Ј-termini, respectively, of the amplification product. For heterologous expression, the ce-hmt-1 cDNA (GenBank TM accession numbers AF497513 and AF497514) was subcloned into the XhoI and SmaI restriction sites of the S. pombe-Escherichia coli shuttle vector pREP4X (American Type Culture Collection) to place it under the control of the thiamine-repressible promoter of the S. pombe nmt-1 ("no message in thiamine") gene. The resulting pREP4X-ce-hmt-1 construct was expressed in LK100 S. pombe cells. As a control, pREP4X vector lacking the ce-hmt-1 insert was transformed into Sp223 and LK100 yeast cells. In all cases, S. pombe cells were transformed using the lithium acetate procedure (20), and the transformants were selected for uracil prototrophy in EMM.

Yeast Strains and Growth
RT-PCR Analysis of Expression of ce-hmt-1 in C. elegans-For RT-PCR analysis of the expression of ce-hmt-1, N2 worms were cultured on solid nematode growth medium (NGM) for 3 days using standard procedures (21). Mixed stage worms collected from the plates were then inoculated into liquid S-medium supplemented with E. coli OP50 and CdCl 2 and cultured with gentle shaking at 20°C. After 6 and 24 h of incubation, the worms were collected, washed free of bacteria in M9 buffer, and frozen in liquid nitrogen. Total RNA was extracted using TRIzol R Reagent (Invitrogen) as described above. The first cDNA strand was synthesized from total RNA (1 g) using an Advantage TM RT-for-PCR kit according to the manufacturer's recommendations (Clontech). Amplification was by PCR using the ce-hmt-1-specific oligonucleotide primer pair 5Ј-TGAGCGGTGTGTAGAGTTGG-3Ј and 3Ј-TTCACTGGCTTTTGCCTTCT-5Ј to yield a 1097-bp amplification product. The primers were designed to preclude amplification from any contaminating genomic DNA that might have been present in the cDNA preparations. To verify that equivalent amounts of RNA had been amplified, the same RNA samples were also subjected to RT-PCR using an oligonucleotide primer pair (5Ј-CTTTCACCACCACCGCT-GAGCGTG-3Ј and 3Ј-CCGATCCAGTCGGAGTACTTGCG-5Ј) specific for the C. elegans actin ce-act-4 gene. In all cases, PCR was performed for 18, 25, 30, or 35 cycles of 30 s at 94°C, 30 s at 55°C, and 1 min at 72°C. The PCR products were resolved by electrophoresis on 1.2% (w/v) agarose gels.
Double-stranded RNA Synthesis-For the in vitro synthesis of cehmt-1 RNA, 958 bp of the coding sequence corresponding to the NTE of the CeHMT-1 polypeptide was PCR-amplified from the ce-hmt-1 cDNA using the primer pairs 5Ј-CCAAAAAATCCGGCGGCCGCATGGG-CTTTTCACCATTTCTCG-3Ј and 3Ј-GTGTGGCGAGCTCGTCGACAA-TCCATTTG-5Ј and subcloned into pBluescript vector (Invitrogen). RNA was transcribed from the T3 and T7 promoters of the construct using an in vitro RNA transcription kit (Stratagene). After DNase digestion of the template, the sense and antisense preparations were gel-purified, mixed in an equimolar ratio in 1 mM EDTA buffered to pH 7.5 with Tris-HCl, heated for 1 min in boiling water, and annealed at room temperature as described for ce-pcs-1 (12). To ensure that the ce-hmt-1 dsRNA preparations from the annealing step were devoid of residual single-stranded RNA, the mixtures were treated with RNase T1 (22). The dsce-hmt-1 RNA product was purified by phenol/chloroform extraction, reconstituted in water, diluted to a concentration of 500 ng/l with injection buffer (23), and stored at Ϫ80°C for injection.
dsRNA Microinjection and Phenotypic Screens-The ce-hmt-1 RNAi experiments and phenotypic screens were performed as described for ce-pcs-1 (12). In all cases, the dsce-hmt-1 RNA-or control (injection buffer alone)-injected worms were first cultured for 12-18 h on NGM plates seeded with E. coli OP50 before transfer to plates supplemented with Cd 2ϩ . Heavy metal sensitivity was screened in two ways. In the first procedure, individual worms were transferred sequentially onto NGM plates supplemented with 0, 25, 50, and 100 M CdCl 2 . Egg laying was allowed to proceed for 6 h at each Cd 2ϩ concentration, after which time the worms were transferred to the next higher Cd 2ϩ concentration for another round of egg laying. In this way, it was determined that 25 M CdCl 2 was sufficient to cause severe morphological and developmental changes in the CeHMT-1-deficient worms but not in the controls. In the second screen, individual injected worms were transferred to NGM plates supplemented with 0, 5, 10, 25, 50, or 100 M CdCl 2 and allowed to lay eggs for 6 h, after which time they were transferred for another 6 h onto plates containing 25 M CdCl 2 . Only the progeny of worms that exhibited the CeHMT-1-deficient phenotype upon transfer to plates containing 25 M Cd 2ϩ (100% of the progeny from worms injected with dsce-hmt-1 RNA) were scored at the other concentrations. The phenotypes of the worms were examined 4 -6 days after hatching.
Subcellular Localization of Heterologously Expressed CeHMT-1 in S. pombe-To determine the subcellular localization of heterologously expressed CeHMT-1 in S. pombe, a construct was generated containing the full-length ce-hmt-1 cDNA fused in-frame with egfp, the gene encoding the red-shifted variant of the 33-kDa green fluorescence protein (GFP). This construct pREP4X-ce-hmt-1::egfp, was used to transform LK100 cells. Vacuoles were visualized by incubating the transformants with acridine orange (0.05 g/ml), a fluorescent lipophilic weak base that accumulates in acidic compartments. Nuclei were visualized after incubating the transformants with 4Ј,6-diamidino-2-phenylindole (DAPI, 100 ng/ml) (Molecular Probes) in Tris-buffered saline for 30 min before being subjected to three cycles of washing in the same buffer minus the fluorescent indicator. Log phase LK100/pREP4Xce-hmt-1::egfp cells and LK100/pREP4X cells stained with acridine orange or DAPI were examined at 100ϫ magnification under a Leica DAS microscope equipped with a DMR camera, UV light source, and GFP-, rhodamine-and DAPI-specific filter sets.
Chemicals-All of the general reagents were obtained from Fisher, Research Organics, Inc., and Sigma

RESULTS
Identification and Cloning of ce-hmt-1-The discovery of a functional PCS gene, ce-pcs-1 in C. elegans (11,12), and the results of earlier investigations suggesting that in S. pombe its homolog SpPCS acts upstream of SpHMT1 (15, 16) prompted a systematic search of worm sequence databases for HMT-1 homologs. Knowing that the C. elegans genome contains some 60 open reading frames for ABC proteins, of which 29 fall into the half-molecule transporter category (24), and that even ABC transporters with disparate capabilities can have a relatively high degree of sequence similarity (18), putative C. elegans HMT-1 homologs were identified according to two structural criteria in addition to overall sequence similarity. The first criterion was the possession of a forward orientation, halfmolecule ABC transporter structure consisting of a single TMD contiguous with a single NBF, in that order. The second was the presence of an NTE containing five hydrophilicity minima.
Application of these criteria disclosed only one gene whose putative translation product satisfies the structural requirements of an SpHMT1 equivalent. Designated ce-hmt-1 for C. elegans heavy metal tolerance factor 1 (alias haf-5 according to the nomenclature of Ref. 24), this gene encodes a 90.7-kDa polypeptide, CeHMT-1, that is 51% sequence similar (34% sequence identical) to SpHMT1. CeHMT-1 consists of a single TMD containing six putative transmembrane spans and an NBF containing a Walker A box motif (GSSGSGKST), an ABC signature motif (LSGGEKQRVAIARTI), and a Walker B box motif (LILVLD) (Fig. 1). As inferred from application of the TMpred and TopPred II computer algorithms, CeHMT-1 possesses a 223-amino acid residue NTE in tandem with the TMD and NBF consisting of a 189-amino acid residue membranespanning domain (MSD0) encompassing five hydrophilicity minima (25) and a 34-amino acid residue linker region (L0) (Fig. 1).
The cDNA corresponding to ce-hmt-1 was isolated by an RT-PCR of total RNA extracted from mixed stage worms using primers designed to span the entire coding region. After confirming the fidelity of the 2.4-kb amplification product (Gen-Bank TM accession number AAM33381) by sequencing, it was used for the experiments described below and for refining the exon-intron structure of the genomic sequence in the C. elegans genomic databases (GenBank TM accession number AF497514).
Expression of ce-hmt-1 RNA Is Constitutive and Not Induced by Cd 2ϩ -Although the expression of some of the transporters implicated in the transport of heavy metals is subject to transcriptional modulation (26 -28), this does not appear to be the case for ce-hmt-1. The steady state levels of ce-hmt-1 transcripts before and after the exposure of mixed stage N2 worms to 10 or 25 M CdCl 2 for 6 or 24 h when assessed by semiquantitative RT-PCR using ce-hmt-1-specific primers are the same (data not shown).
ce-hmt-1 Suppresses the Cd 2ϩ Hypersensitivity of S. pombe hmt1 Ϫ Mutants-If ce-hmt-1 encodes a protein that is equivalent to or has significant functional overlap with SpHMT-1, its heterologous expression in hmt1 Ϫ mutant S. pombe cells should alleviate Cd 2ϩ hypersensitivity. Therefore, to gain an initial indication of its functional capabilities, ce-hmt-1 was subcloned into the E. coli-S. pombe shuttle vector pREP4X under control of the thiamine-repressible plasmid-borne promoter of the nmt-1 gene and transformed into S. pombe hmt1 Ϫ mutant strain LK100; thiamine-repressible suppression of the Cd 2ϩ -hypersensitive hmt1 Ϫ phenotype was then assayed. These experiments demonstrated that expression of the pREP4X-ce-hmt-1 construct confers increased Cd 2ϩ tolerance versus empty vector-transformed LK100 controls regardless of whether growth is monitored as colony formation by serially diluted inocula on solid media containing Cd 2ϩ after 8 days ( Fig. 2A) or as cell density after 72 h of growth in liquid media containing Cd 2ϩ (Fig. 2B). As determined from the concentrations of Cd 2ϩ required to inhibit growth in liquid medium by 50%, plasmid-borne ce-hmt-1 increases the tolerance of LK100 cells by at least 10-fold (Fig. 2B). The effects seen are specifically attributable to expression of the ce-hmt-1 gene insert in that repression of the nmt-1 promoter when the growth media are supplemented with thiamine abolishes the Cd 2ϩ tolerance that is otherwise conferred by pREP4X-ce-hmt-1 (Fig. 2C).
Heterologously Expressed CeHMT-1 Localizes to the Vacuolar Membrane of S. pombe-SpHMT1 localizes to the vacuolar membrane of S. pombe and is considered to catalyze the vacuolar sequestration of Cd⅐PC complexes (16). Having determined that heterologously expressed CeHMT-1 is able to suppress the Cd 2ϩ hypersensitivity of SpHMT1-deficient S. pombe mutants, the intracellular distribution of CeHMT-1 in this system was examined. For this purpose, the full-length coding sequence of ce-hmt-1 was fused in-frame with the coding sequence for the red-shifted variant of GFP, enhanced GFP (EGFP), and cloned behind the nmt-1 promoter of pREP4X.
When transformed into S. pombe LK100, pREPX4ce-hmt-1::egfp yields cells in which the bulk of the green fluorescence associated with the fusion product localizes to the vacuole periphery (Fig. 3A). Fluorescence microscopy reveals intense fluorescent green circular structures in the transfor-mants that coincide with the membranes bounding the vacuoles as determined by differential interference contrast microscopy (Fig. 3A) and the distribution of acridine orange, a fluorescent pH gradient indicator that undergoes preferential accumulation in acidic compartments (Fig. 3B). Neither of these fluorescence signals derives from the nucleus, because DAPI staining of the same cells yields a punctate distribution distinct from those of the other two probes (Fig. 3C).
ce-hmt-1 Is Required for Cadmium Tolerance in C. elegans-To determine whether CeHMT-1 is required for heavy metal tolerance in the intact organism, CeHMT-1-deficient worms were generated by RNAi and screened for growth on standard NGM plates supplemented with different concentrations of CdCl 2 .
The results of these manipulations were unequivocal. It was demonstrated that whereas the progeny of the control injection buffer-injected worms and the CeHMT-1-deficient worms in media devoid of heavy metals are indistinguishable and develop into normal-sized, gravid, fertile adults, the latter are markedly more sensitive than the former to exposure to Cd 2ϩ . The progeny of the control injection buffer-injected worms are Aliquots of the cell suspensions were then serially diluted 2-, 5-, 10-, and 20-fold and spotted onto solid EMM supplemented with glucose, adenine, leucine, and the indicated concentrations of CdCl 2 . Colonies were visualized after incubating the plates for 8 days at 30°C. B, S. pombe wild-type and hmt1 Ϫ mutant cells transformed with empty pREP4X vector (E and •, respectively), and hmt1 Ϫ mutant cells transformed with pREP4X containing the ce-hmt-1 insert (Ⅺ) grown in liquid EMM lacking thiamine. C, S. pombe wild-type and hmt1 Ϫ mutant cells transformed with empty pREP4X vector (E and •, respectively) and hmt1 Ϫ mutant cells transformed with pREP4X containing the ce-hmt-1 insert (Ⅺ) grown in liquid EMM supplemented with thiamine (5 g/ml) to repress expression of ce-hmt-1 from the nmt-1 promoter of the construct. One hundred microliter aliquots from standard overnight cultures were inoculated into 2 ml of the same medium but containing CdCl 2 at the concentrations indicated. OD 600 nm was measured after growth at 30°C for 72 h. Nomarski images and GFP, acridine orange, and DAPI fluorescence images were collected using a Leica DAS microscope equipped with a UV light source and a DMR camera. Cells were examined at 100ϫ magnification and fluorescence images were collected using GFP-, rhodamine-, and DAPI-specific filter sets. DIC, differential interference contrast.
comparable with those grown on medium lacking heavy metal when grown on media containing low (5, 10, or 25 M) concentrations of CdCl 2 (Fig. 4). Indeed, at the highest concentrations of CdCl 2 screened (50 and 100 M), control worms reach adulthood, albeit slowly, and lay eggs after ϳ5 days (data not shown). By contrast, ce-hmt-1-deficient worms are extremely sensitive to Cd 2ϩ in the growth medium, such that even at the lowest CdCl 2 concentrations tested (5 M) they arrest in the early larval stages and eventually die. The effects seen are irreversible in that the arrested Cd 2ϩ -treated CeHMT-1-deficient larvae do not resume growth after transfer to standard NGM plates.
Disruption of the Two Closest Homologs of CeHMT-1 Does Not Influence Cd 2ϩ Tolerance-As a control to determine whether CeHMT-1-dependent Cd 2ϩ tolerance is specific to this gene product and not a property shared by other ABC transporters, RNAi was performed against the two closest CeHMT-1 homologs, Haf-7 and Haf-9 (24). Haf-7 and Haf-9 were chosen because both are half-molecule ABC transporters, which, like CeHMT-1 but unlike other C. elegans half-molecule ABC transporters, possess a hydrophobic NTE.
As is evident from the results shown in Fig. 4, the acute sensitivity of ce-hmt-1 RNAi worms is specific for this halfmolecule ABC transporter in that targeted suppression of haf-7 and haf-9 does not confer Cd 2ϩ hypersensitivity. The progeny of dshaf-7-, dshaf-9-, and injection buffer-injected worms develop into normal-sized, gravid adults at all of the CdCl 2 concentrations tested and are indistinguishable from those grown on media devoid of heavy metal.
CeHMT-1-deficient Worms Are More Sensitive to Cd 2ϩ and Morphologically Distinguishable from CePCS-1-deficient Worms-Knowing from previous investigations that CePCS-1deficient worms, like CeHMT-1-deficient worms, are hypersensitive to Cd 2ϩ (12), it was considered important to determine whether the sensitivities and morphologies of ce-pcs-1 and cehmt-1 RNAi worms when screened under identical conditions are consistent with disruption of a common metal detoxification pathway.
The results of these analyses were instructive because they revealed that CeHMT-1-deficient worms are more sensitive to Cd 2ϩ and exhibit Cd 2ϩ -elicited changes in cellular morphology distinct from those of CePCS-1-deficient worms. When scored as the concentration of Cd 2ϩ required for a 50% decrease in the number of individuals that reach adulthood, the progeny of ce-hmt-1 RNAi worms are at least 4-fold more sensitive to Cd 2ϩ than the progeny of ce-pcs-1 RNAi worms (Figs. 4 and 5). Furthermore, the progeny of ce-hmt-1 and ce-pcs-1 RNAi worms undergo distinguishable morphological changes upon exposure to heavy metals (Fig. 6). As exemplified by worms hatched and cultured for 4 days on media containing toxic concentrations of CdCl 2 , the intestinal cells of the progeny of dsce-pcs-1-injected worms undergo necrosis as noted previously (12). In contrast, the same cells in the progeny of dsce-hmt-1injected worms fabricate spherical refractive inclusions of ϳ20 m in diameter in the immediate vicinity of the nucleus (Fig.  6). These inclusions are absent from the larvae immediately after hatching on media containing Cd 2ϩ but are apparent after ϳ24 h and persist throughout the worm's lifetime.
During the latter stages of preparing this manuscript for publication, two knock-out lines, VC252 and VC287, were obtained from the C. elegans Knock-out Consortium, each of which contains a partial deletion of the ce-hmt-1 gene. Both of these lines are of similar Cd 2ϩ hypersensitivity and undergo the same morphological changes upon exposure to Cd 2ϩ as the dsce-hmt-1 RNAi worms described here (data not shown), confirming that the Cd 2ϩ hypersensitivity of and the morphological changes seen in CeHMT-1-deficient worms are the same whether CeHMT-1 function is disrupted by gene deletion or RNAi.

dsce-pcs-1 RNAi Does Not Influence the Cd 2ϩ Hypersensitivity of dsce-hmt-1 RNAi Worms or the Assembly of Intracellular
Inclusions-If CeHMT-1 acts downstream of PC synthesis and is solely responsible for the intracellular sequestration and/or cellular elimination of Cd⅐PC complexes, a deficiency in Ce-HMT-1 activity would be expected to cause the accumulation of Cd⅐PC complexes and/or their derivatives in the cytosol. Accumulation of such complexes in the cytosol might then provide the framework for and/or elicit formation of the refractive inclusions seen in Cd 2ϩ -treated CeHMT-1-deficient worms. Cytosolic Cd⅐PC accumulation would, in turn, likely preclude salvage of PC constituents from their usual site of delivery, acidic vacuolysosomal compartments, and thereby exhaust cellular reserves of the PC precursor GSH. Because GSH is not only involved in metal chelation but also in alleviating some of the secondary oxidative consequences of heavy metal toxicity, this could account for why dsce-hmt-1 RNAi worms are more sensitive to Cd 2ϩ than dsce-pcs-1 RNAi worms. A specific prediction therefore follows. Namely, if the extreme Cd 2ϩ hypersensitivity of CeHMT-1-deficient worms and their facility for the elaboration of cytosolic inclusions depends exclusively on the synthesis of PCs, coincident suppression of ce-pcs-1 expression should abolish the ce-hmt-1 RNAi-specific phenotype.
In testing this hypothesis and refuting it through the generation of doubly CePCS-1-and CeHMT-1-deficient worms, results were obtained demonstrating that the morphological changes associated with the ce-hmt-1 RNAi phenotype neither depend obligatorily on a functional copy of ce-pcs-1 nor are alleviated by its suppression. Targeted suppression of the expression of both ce-pcs-1 and ce-hmt-1 neither diminishes the Cd 2ϩ hypersensitivity associated with the loss of CeHMT-1 function nor interferes with assembly of the intracellular inclusions formed upon exposure to Cd 2ϩ . Worms that are deficient in both CePCS-1 and CeHMT-1, like worms that are singly deficient in CeHMT-1, undergo developmental arrest and form nuclearly associated inclusions when grown on media containing Cd 2ϩ (25 M) (Fig. 6) but are indistinguishable from injection buffer-injected controls when grown on media lacking heavy metals (data not shown). Indeed, the intestinal cells of FIG. 4. Cd 2؉ hypersensitivity of ce-hmt-1 RNAi worms. The progeny of injection buffer-injected, dshaf-7 RNA-injected, or dshaf-9 RNA-injected worms (E), dsce-hmt-1 RNA-injected worms (•), and dscepcs-1 RNA-injected worms (f) were cultured on NGM plates supplemented with the indicated concentrations of CdCl 2 . Shown are the percentages of worms that had reached adulthood 4 days after hatching. The same symbol (E) is used for the progeny of the injection buffer-injected, dshaf-7 RNA-injected, and dshaf-9 RNA-injected worms because their survivorship profiles were indistinguishable. For the injection buffer controls and individuals injected with dsce-hmt-1 RNA, at least nine worms were injected and 200 progeny scored; for dshaf-7 and dshaf-5 RNA, at least five worms were injected and at least 100 progeny scored; for dsce-pcs-1 RNA, at least three worms were injected and at least 60 progeny scored.
doubly deficient ce-pcs-1;ce-hmt-1 RNAi worms not only form inclusions but also undergo necrosis with the same time dependence as singly deficient ce-pcs-1 RNAi worms (Fig. 6). Apparently, CeHMT-1 does not exclusively participate in PCdependent Cd 2ϩ detoxification but in addition participates in another pathway or pathways that also contribute to the alleviation of heavy metal toxicity. DISCUSSION The results of these studies establish that the forward orientation half-molecule ABC transporter CeHMT-1 plays a critical role in Cd 2ϩ tolerance in C. elegans. It bears a structural resemblance to another forward orientation half-molecule ABC transporter, SpHMT-1, which in S. pombe localizes to the vacuolar membrane and participates in PC-dependent heavy metal detoxification. When heterologously expressed in Cd 2ϩhypersensitive S. pombe hmt1 Ϫ mutants, CeHMT-1 localizes to the vacuolar membrane and partially restores Cd 2ϩ tolerance. Also, as would be expected if this ABC transporter plays a pivotal role in the detoxification of heavy metals in the intact organism, CeHMT-1-deficient ce-hmt-1 RNAi worms are acutely sensitive to exposure to Cd 2ϩ .
CeHMT-1 has a special status in this respect. Its nearest equivalents, Haf-7 and Haf-9 from the same organism, do not contribute to heavy metal tolerance. Moreover, although a few full-molecule ABC transporters, namely one member of the multidrug resistance-associated protein (MRP-1) and two members of the P-glycoprotein subfamily (PGP-1 and PGP-3), have been shown to contribute to heavy metal tolerance in C. elegans (29), their phenotypic impact is low by comparison to that of CeHMT-1. C. elegans triple mrp-1;pgp-1;pgp-3 mutants, for instance, exhibit a markedly less pronounced Cd 2ϩhypersensitive phenotype (29) than their ce-hmt-1 RNAi counterparts.
From investigations of the molecular basis of heavy metal detoxification in S. pombe it has been inferred that SpHMT1 acts downstream of PC synthase by catalyzing the transport of Cd⅐PC complexes into the vacuole (15,16). Our finding that CeHMT-1-deficient worms, like CePCS-1-deficient worms, are Cd 2ϩ -hypersensitive is consistent with the general applicability of such a scheme to other organisms that harbor genes for HMT-1-type ABC transporters. This being the case, the phenotypes of CeHMT-1-deficient and CePCS-1-deficient worms are nonetheless sufficiently different to necessitate auxiliary roles. Specifically, CeHMT-1-deficient worms are markedly more sensitive to Cd 2ϩ than CePCS-1-deficient worms, and, whereas the intestinal cells of the former elaborate refractive inclusions upon exposure to Cd 2ϩ , the same cells of the latter do not, but instead undergo necrosis. Accordingly, closer inspection of the genetic interactions between ce-hmt-1 and ce-pcs-1 does not demonstrate a strict dependence of either on the other, as would be predicted for genes whose translation products are exclusively restricted to a common pathway. Suppression of ce-pcs-1 expression in conjunction with ce-hmt-1 does not interfere with formation of the intestinal inclusions, nor does it decrease Cd 2ϩ sensitivity to the level observed in single cepcs-1 RNAi worms. Rather, the intestinal cells of double ce-pcs-1;ce-hmt-1 RNAi worms manifest both phenotypes, namely necrosis and the formation of refractive inclusions after exposure to Cd 2ϩ .
The most straightforward explanation for these observations is that CeHMT-1 is required for Cd 2ϩ detoxification, but not only in a PC-dependent manner. It is probably also involved in the transport of molecules other than PCs. For example, Ce-HMT-1 may contribute to the cellular sequestration and/or elimination of heavy metals coordinated with other ligands such as GSH and/or metallothioneins as well as PCs and/or is involved in alleviating the toxic effects of the reactive oxygen species formed when cells are exposed to heavy metals.
A corollary of the conclusion that CeHMT-1 has auxiliary functions is that the same may apply to other HMT-1-like transporters. Domain comparisons among the half-molecule ABC transporters in the protein databases of other animals disclose transporters that are HMT-1-like (for instance, FIG. 5. Gross morphology of progeny of injection buffer-injected worms (Control), dsce-hmt-1 RNA-(ce-hmt-1), and dsce-pcs-1 RNA-injected (ce-pcs-1) worms. The progeny of injected worms were cultured for 4 days on NGM plates supplemented with 5 M CdCl 2 as described in the Fig. 4 legend before inspection and Nomarski microscopy. The worms shown are representative of the populations scored in Fig. 4.

FIG. 6. Morphology of intestinal cells of progeny of injection buffer-injected worms (Control)
, dsce-hmt-1 RNA-(ce-hmt-1), dsce-pcs-1 (ce-pcs-1) RNA-, and dsce-hmt-1;dsce-pcs-1 RNA-injected worms (ce-hmt-1;ce-pcs-1). Shown are images from the progeny of RNAi worms injected with dsce-hmt-1 (ce-hmt-1), dsce-pcs-1 (ce-pcs-1), or a mixture of dsce-pcs-1 and dsce-hmt-1 (pcs-1;hmt-1) RNAs after 4 days on NGM plates supplemented with 25 M CdCl 2 . White arrows indicate refractive inclusions, and black arrows indicate necrotic regions. For the acquisition of these high resolution images, the worms were immersed in buffer containing 5 mM sodium azide and photographed within 5 min of immobilization. CG4225 from Drosophila and mammalian mitochondrial ABC protein 3 (MTABC3) from humans). This would not be expected from what is known of SpHMT1, because Drosophila and humans do not deploy a PC-dependent pathway for heavy metal detoxification (30). Three explanations capable of reconciling what ostensibly is a contradiction might follow from these findings and from the results of studies implicating mammalian mitochondrial ABC protein 3 in iron homeostasis (31). The first explanation is that the HMT-1s have undergone divergence such that representatives in different organisms serve different roles. The second is that, in addition to heavy metal detoxification in C. elegans or iron homeostasis in higher animals, these transporters are involved in other cellular processes. The third possibility is that the mammalian mitochondrial ABC protein 3 of humans and CG4225 from Drosophila, in addition to iron homeostasis, are involved in detoxification of heavy metals and/or the by-products of heavy metal action.