Arsenic tolerance in Arabidopsis is mediated by two ABCC-type phytochelatin transporters

Abstract

Arsenic is an extremely toxic metalloid causing serious health problems. In Southeast Asia, aquifers providing drinking and agricultural water for tens of millions of people are contaminated with arsenic. To reduce nutritional arsenic intake through the consumption of contaminated plants, identification of the mechanisms for arsenic accumulation and detoxification in plants is a prerequisite. Phytochelatins (PCs) are glutathione-derived peptides that chelate heavy metals and metalloids such as arsenic, thereby functioning as the first step in their detoxification. Plant vacuoles act as final detoxification stores for heavy metals and arsenic. The essential PC–metal(loid) transporters that sequester toxic metal(loid)s in plant vacuoles have long been sought but remain unidentified in plants. Here we show that in the absence of two ABCC-type transporters, AtABCC1 and AtABCC2, Arabidopsis thaliana is extremely sensitive to arsenic and arsenic-based herbicides. Heterologous expression of these ABCC transporters in phytochelatin-producing Saccharomyces cerevisiae enhanced arsenic tolerance and accumulation. Furthermore, membrane vesicles isolated from these yeasts exhibited a pronounced arsenite [As(III)]–PC₂ transport activity. Vacuoles isolated from atabcc1 atabcc2 double knockout plants exhibited a very low residual As(III)–PC₂ transport activity, and interestingly, less PC was produced in mutant plants when exposed to arsenic. Overexpression of AtPCS1 and AtABCC1 resulted in plants exhibiting increased arsenic tolerance. Our findings demonstrate that AtABCC1 and AtABCC2 are the long-sought and major vacuolar PC transporters. Modulation of vacuolar PC transporters in other plants may allow engineering of plants suited either for phytoremediation or reduced accumulation of arsenic in edible organs.

Arsenic has been widely used in medicine, industry, and agriculture. In medicine, it was used to treat diseases such as syphilis, trypanosomiasis, or amoebic dysentery. In agriculture, arsenic-based herbicides, such as disodium methanearsonate (DSMA), continue to be applied for weed and pest control. However, arsenic is a highly toxic environmental pollutant that causes global health problems. In Southeast Asia as well as in regions with extensive mining, such as China, Thailand, and the United States, arsenic concentrations in water can be far above the World Health Organization limit of 10 μg/L (133 nM), concentrations that are known to cause health problems. People are exposed to arsenic poisoning both by drinking contaminated water and by ingesting crops cultivated in soils irrigated with polluted water. It has been shown that arsenic in rice paddy fields is readily taken up by rice plants and translocated to the grains, constituting a danger for populations that rely mostly on rice for their diet (1, 2). In Bangladesh alone an estimated 25 million people are exposed to water contaminated with arsenic exceeding 50 μg/L, the Bangladesh government standard, and more than 2 million people are estimated to face a risk of death from cancer caused by arsenic (3–5).

To reduce nutritional arsenic intake, identification of the mechanisms implicated in arsenic accumulation and detoxification of plants is a prerequisite. Considerable progress has been made during the last years in the identification of the molecular mechanisms of arsenic (As) uptake, metabolism, and translocation (for reviews see refs. 6–8). After entering into roots through high-affinity phosphate transporters, arsenate [As(V)] is readily reduced to arsenite [As(III)]. Alternatively, under reducing conditions, As(III) is taken up by membrane proteins belonging to the aquaporin family (9, 10). In rice, it has been shown that As(III) is exported subsequently to the xylem by the silicon efflux transporter Lsi2, resulting in root-to-shoot transport of arsenic (11). The final step of arsenic detoxification in the cell is the sequestration into vacuoles, which depends mainly on glutathione (GSH) and phytochelatins (PCs). PCs are GSH-derived metal-binding peptides whose synthesis is induced by arsenic as well as other toxic metals (12, 13). PCs chelate As(III) and heavy metals with their thiol groups, and the metal(loid)–PC complexes are sequestered into vacuoles, protecting cellular components from the reactive metal(loid)s.

PC-mediated detoxification is unique to plants and a few other PC-producing organisms but different from that of budding yeast or humans, in which PCs or PC synthase (PCS) do not exist. The transporters responsible for active transport of PC-conjugated As(III) as well as for transport of PC-conjugated heavy metals into plant vacuoles (14–17) are yet to be identified but have been proposed to be ABC transporters, as is the case with glutathione (GS)-conjugated As(III) (As(GS)₃) in other organisms (18, 19). For example, in Saccharomyces cerevisiae it has been shown that arsenic is detoxified by YCF1, an ABC protein transporting As(GS)₃ into vacuoles (18). In humans, it has been shown that HsABCC1 and HsABCC2 are involved in arsenic detoxification by transporting As(III) conjugated to GSH (19). Another ABC transporter, HMT1, has been proposed to act as a vacuolar PC transporter in Schizosaccharomyces pombe, although an hmt1 deletion mutant continued to exhibit PC accumulation in vacuoles, and HMT1 did not confer arsenic resistance (20, 21). In plants, no ortholog for HMT1 has been identified.

Here we report the identification of two ABC transporters required for arsenic detoxification. These transporters are plant PC transporters that have been sought since the discovery of PCs (12, 22). Thus this finding provides a key to understanding the detoxification of many xenobiotic molecules, heavy metals, and metalloids, including arsenic, that are conjugated with PCs for detoxification.

Results

atabcc1 atabcc2 Double Knockout Mutant Is Hypersensitive to Arsenic and Arsenic-Based Herbicide.

To identify candidate transporters likely involved in arsenic detoxification in plants, we focused on screening the ABCC subfamily of ABC transporters, also known as the multidrug resistance-associated proteins (MRPs). This subfamily includes ABC transporters implicated in heavy metal resistance, such as ScYCF1, the yeast vacuolar As(GS)₃ transporter (18), and the two human arsenic-detoxifying ABC transporters (19). Members of this family have also been shown to transport GSH conjugates and to confer cadmium resistance in plants and humans (23–26). In addition, for many ABCC proteins in Arabidopsis, vacuolar localization has been demonstrated or proposed on the basis of data obtained using either Western blotting of membrane fractions, GFP fusion proteins, or vacuolar proteomics approaches (27–30). We have therefore isolated homozygous knockout lines for all 15 Arabidopsis ABCC genes and have grown these mutants in the presence of arsenic and arsenic-containing herbicides. Two different forms of arsenic were used because they have different entry pathways to the cell, as well as differential metabolism. Whereas As(V) is taken up by the high-affinity phosphate transporter (31, 32), DSMA is much more hydrophobic and is rapidly absorbed by plant roots. Furthermore, As(V) is reduced to As(III) in the cell, whereas in DSMA arsenic is already present in the As(III) form. Although no As(V)-sensitive atabcc knockout mutant was found, the growth of two deletion mutants, atabcc1/atmrp1 and atabcc2/atmrp2, was impaired in the presence of the arsenic herbicide DSMA, compared with that of the wild type (Fig. 1 and Fig. S1). To determine whether the relatively moderate arsenic sensitivity could be due to overlapping functions of the two ABC transporters, we generated a double knockout atabcc1 atabcc2 mutant. AtABCC1 and AtABCC2, which belong to clade I of the ABCC subfamily, share a high amino acid sequence similarity. Furthermore, both have been localized to the vacuolar membrane in former studies (27, 28). Whereas under control conditions no difference in growth was observed between the wild type and the double knockout mutant, growth of the atabcc1 atabcc2 double knockout line was dramatically impaired on plates containing DSMA as well as low level of As(V) compared with the wild type (Fig. 1 A and B). Quantification of the fresh weight and root length of plants grown in As(V)-containing media confirmed that a significant difference could only be observed for the double knockout, which exhibited a strongly reduced fresh weight even at low As(V) concentrations (Fig. 1 C and D). In the presence of 30 μM As(V), root growth was reduced by more than 60%. The sensitivity of the atabcc1 atabcc2 knockout line was comparable to that observed for cad1-3, an Arabidopsis mutant impaired in PC synthesis (Fig. S2) (33). To verify whether AtABCC1 and AtABCC2 confer arsenic tolerance under natural conditions, we grew atabcc1 and atabcc2 single and double knockout lines, and the corresponding wild type on soil for 3 wk and then irrigated the plants with 133.5 μM As(V) for an additional 8 d. The atabcc1 atabcc2 double mutants were extremely sensitive to As(V) and did not survive this treatment, whereas the wild-type and single knockout plants were not visibly affected (Fig. 1E).

Fig. 1.

Hypersensitivity of atabcc1 atabcc2 double knockout mutants to arsenic [As(V)] and DSMA, an arsenic-based herbicide. (A and B) Wild-type, AtABCC1 knockout (abcc1-3), AtABCC2 knockout (abcc2-2), and double knockout (abcc1 abcc2) seedlings grown in 100 mg/L of DSMA (A) and 50 μM of As(V) (B) containing half-strength MS media for 16 d. Note that single knockout mutants of AtABCC1 or AtABCC2 are also sensitive to the arsenic herbicide (A) but not to As(V) (B). (Scale bar, 1 cm.) (C and D) Fresh weight (C) and root length (D) of each genotype of plants grown on 0–50 μM As(V)-containing 1/2 MS media. Mean ± SEM (for fresh weights, n = 16, from four independent experiments; for root lengths, n = 332–336, from three independent experiments). *P < 0.04, **P < 0.01 (Student's t test). (E) Arsenic-sensitive phenotype of atabcc1 atabcc2 double knockout plants grown in soil. Three-week-old plants were watered with 133.5 μM As(V). Four representative plants out of 12; each was photographed 8 d after arsenic treatment.

AtABCC1 and AtABCC2 Enhance Arsenic Tolerance and Accumulation in the Presence of TaPCS1 in Budding Yeasts.

To test whether AtABCC1 and AtABCC2 confer arsenic resistance by compartmentalization of a GS–As complex into the vacuole, we expressed the transporters in SM4, a budding yeast mutant strain lacking YCF1 and the three other vacuolar ABCC-type ABC transporters, YHL035c, YLL015w, and YLL048c. AtABCC1- or AtABCC2-expressing SM4 yeast lines showed only a marginal increase in arsenic resistance, compared with the empty vector control (Fig. 2A). These results indicate that in contrast to Arabidopsis, AtABCC1 and AtABCC2 cannot mediate arsenic tolerance in budding yeast, which does not synthesize PCs. These experiments also suggest that AtABCC1 and AtABCC2 are not efficient GS–As complex transporters and thus cannot complement ycf1.

Fig. 2.

AtABCC1 and AtABCC2 enhance arsenic tolerance and accumulation in PCS-expressing budding yeasts. (A) Expression of AtABCC1 (ABCC1) or AtABCC2 (ABCC2) in PC-producing yeast (SM7) enhanced tolerance to 100 μM of As(III). EV, empty vector control. In SM4 strain without PCS expression, no growth difference between genotypes was apparent. Cells were grown in half-strength SD-Ura (synthetic dextrose lacking uracil) media supplemented with 100 μM As(III) for 2 d. O.D.₆₀₀, optical density at 600 nm of the yeast suspension that was initially spotted on the plates. (B) Arsenic content of SM7 yeast cells expressing AtABCC1 or AtABCC2 was higher than that of cells transformed with the empty vector. Cells were grown in half-strength SD-Ura media supplemented with 100 μM As(III) for 12 h. Mean ± SEM (n = 21, from seven independent experiments). *P < 0.03, **P < 0.005 (Student's t test).

In contrast to S. cerevisiae and humans, which depend on GSH for arsenic detoxification, in plants detoxification of arsenic has been shown to mainly depend on PCs (8, 16, 33, 34). We therefore tested whether AtABCC1 and AtABCC2 confer arsenic resistance by transporting PC–As complexes. As a first approach, we generated a yeast mutant expressing the wheat PCS TaPCS1 (35, 36) in the SM4 background and named it SM7. This yeast strain was then used to express either AtABCC1 or AtABCC2. When grown on arsenic-free control medium, no large difference could be observed between the SM7 cells transformed with the empty vector or expressing either of the ABC transporters. However, when the yeast cells were exposed to 100 μM As(III), the lines expressing AtABCC1 or AtABCC2 grew dramatically better than controls (Fig. 2A). In this condition, yeast cells expressing the ABC transporters accumulated more arsenic than the corresponding control (Fig. 2B). Furthermore, in the presence of arsenic, AtABCC1- and AtABCC2-expressing SM7 strains contained much higher PC levels compared with SM7 strains transformed with the empty vector (Fig. S3), indicating that efficient PC synthesis is only achieved when PCs are transported into the vacuole.

Together, these results indicate that AtABCC1 and AtABCC2 can mediate vacuolar PC–As sequestration, allowing the PC-producing yeast strains to cope with arsenic. These heterologous reconstruction assays also suggest that chelation of arsenic by PC alone cannot detoxify arsenic, but transporters of PC–As complex are required, because expression of PCS in the SM4 background (SM7) did not increase arsenic tolerance (Fig. 2A).

AtABCC1 and AtABCC2 Mediate PC Transport.

To directly determine whether AtABCC1 and AtABCC2 mediate PC–As transport, we performed transport experiments using microsomal vesicles isolated from SM7 yeast transformed with the empty vector or expressing AtABCC1 and AtABCC2. As shown in Fig. 3A, vesicles isolated from yeast cells expressing AtABCC1 or AtABCC2 efficiently took up As(III)–PC₂ in a time-dependent manner, whereas in vesicles isolated from the empty vector control the uptake was negligible. As expected for an ABC-type transporter, As(III)–PC₂ uptake into yeast vesicles was not inhibited by dissipating the proton gradient with NH₄⁺, whereas in the presence of vanadate the transport activity was completely inhibited (Fig. 3B). Comparison of the uptake activity of apoPC₂ and As(III)–PC₂ revealed that the As(III)–PC₂ complex is transported more efficiently than apoPC₂ by both AtABCC1 and AtABCC2 (Fig. 3B). The ratio of the transport activity between apoPC₂ and As(III)–PC₂ varied between the different batches of PC₂ produced, probably owing to variable stability of the PC₂–As complex. The discrimination between apoPC₂ and As(III)–PC₂ was more pronounced for AtABCC2 than for AtABCC1. As shown for As(III)–PC₂, apoPC₂ uptake was also inhibited by vanadate but not by NH₄⁺ (Fig. S4A). We further tested whether AtABCC2 can transport arsenic in a non-chelated form using microsomal vesicles of the SM7 cells expressing the transporter. Arsenic content measured using inductively coupled plasma mass spectrometry (ICP-MS) showed that without PCs no arsenic was transported via AtABCC2 (Fig. 3C). Furthermore, by comparison of concentrations of PC₂ and arsenic transported into these vesicles, we could estimate that AtABCC2 mediates the transport of two molecules of PC₂ per molecule of arsenic (Fig. 3C).

Fig. 3.

AtABCC1 and AtABCC2 mediate PC transport in microsomes isolated from yeast transformed with AtABCC1 and AtABCC2 transporters. (A) Time-dependent uptake of PC₂–As in yeast microsomes isolated from S. cerevisiae transformed with the empty vector (EV), AtABCC1, and AtABCC2. PC concentration was 25 μM. (B) Comparison of transport activities for apoPC₂ and PC₂–As, and inhibition of PC₂–As transport by NH₄⁺ and vanadate. PC concentration was 25 μM. (C) Relationship between arsenic (Left) and PC (Right) transport activity. PC₂ (500 μM) with 1 mM As(III) or 1 mM As(III) alone was used, and the concentrations of transported PC₂ and arsenic were measured. (D) Concentration dependence of PC₂–As (solid line) and apoPC₂ (dotted line) transport activity of vesicles isolated from AtABCC1 (filled squares) and AtABCC2 (filled triangles) expressing SM7 S. cerevisiae. Inset: Enlarged version at low concentrations of the substrate. Mean ± SEM (from three independent experiments with four replicates each).

Surprisingly, the concentration-dependent As(III)–PC₂ uptake did not follow classic saturation kinetics, but exhibited an apparent sigmoid curve for both AtABCC1 and AtABCC2 (Fig. 3D). At PC₂ concentrations below 10 μM, As(III)–PC₂ uptake activity was negligible but was strongly stimulated by increasing PC concentrations. The PC concentration for half-maximal uptake activity was estimated to be 350 μM. This high concentration suggests a rather low affinity of AtABCC1 and AtABCC2 for PCs. Previous reports demonstrated that PCs can accumulate to concentrations exceeding 1,000 μM under metal(loid) stress (37), in good agreement with the concentrations that exhibited optimal uptake in these experiments (Fig. 3D). Interestingly, the concentration dependence for apoPC₂ was linear and did not reveal a saturation curve up to 1,000 μM apoPC₂ (Fig. 3D).

AtABCC1 and AtABCC2 Are the Major As(III)–PC₂ Transporters in Plant Vacuoles.

To determine whether AtABCC1 and AtABCC2 indeed encode the long-sought functional PC transporters in planta and to determine their contribution to vacuolar PC uptake, we compared the transport activities for As(III)–PC₂ and apoPC₂ of intact vacuoles isolated from the atabcc1 atabcc2 double knockout and wild-type plants. As shown in Fig. 4A, vacuolar uptake activities of both As(III)–PC₂ and apoPC₂ were dramatically reduced by 85% and 95%, respectively, in the mutant plant, indicating that these two ABC transporters are by far the most important PC and PC-conjugate transporters in Arabidopsis. AtABCC11 and AtABCC12, very close homologs of AtABCC1 and AtABCC2, could be responsible for the very low residual PC uptake activity. Comparison of the transport rates of apoPC₂ and As(III)–PC₂ in vacuoles isolated from wild-type Arabidopsis showed that the differences in transport rates for these two forms of PC were less pronounced in Arabidopsis vacuoles than in yeast vesicles. This might be due to the more than twofold higher expression level of AtABCC1 compared with AtABCC2 in fully expanded Arabidopsis leaves, the tissue used for vacuole isolation (Genenvestigator or http://www.bar.utoronto.ca/). In yeast experiments, the selectivity of AtABCC1 for As(III)–PC₂ was much lower than that of AtABCC2. An alternative explanation could be that the different lipid environment in plant and yeast vesicles also affects the selectivity. As observed for yeasts, vacuolar PC transport in Arabidopsis was not inhibited by NH₄⁺ (residual transport activity 98% ± 4%) but strongly inhibited by vanadate (residual transport activity 25% ± 7%) (Fig. S4B), which are typical features of ABC transporters.

Fig. 4.

Vacuoles isolated from the atabcc1 atabcc2 double knockout exhibit only a very low residual PC transport activity, and atabcc1 atabcc2 double knockouts show decreased content of PCs at the seedling stage. (A) Vacuoles were isolated from wild-type and the atabcc1 atabcc2 double knockout plants and tested for transport activities for PC₂–As and apoPC₂. As a negative control, ATP was replaced by ADP. PC concentration was 250 μM. Mean ± SEM (from two independent experiments with four replicates each). (B) Twelve-day-old seedlings were exposed to different concentrations of potassium arsenate for 96 h, and the integrated accumulation of PCs was determined by HPLC-MS. Mean of two replicates (0 μM of arsenic) or three replicates ± SEM (50 and 100 μM of arsenic).

PC contents, integrated over 2 d of continuous arsenic exposure, were reduced in atabcc1 atabcc2 double knockout plants, when total plant PC contents, reflecting the vacuolar and cytoplasmic pools, were measured (Fig. 4B). The difference between wild-type plants and the mutants become more pronounced at high arsenic concentrations, which increased production of PCs. This observation is in line with the sigmoid kinetics of the vacuolar transport observed in vesicles isolated from the yeasts expressing the two AtABCC transporters and again supports the necessity of transporting PCs into the vacuole for continuous PC synthesis. Surprisingly, double mutant plants took up slightly more arsenic, but the transfer of arsenic to the shoot was reduced in these plants (Fig. S5). The same result was obtained in plants grown on plates as well as under hydroponic conditions.

Overexpression of AtABCC1 Increases Arsenic Tolerance Only When Coexpressed with PCS.

To determine whether arsenic tolerance can be increased by expressing one of the PC transporters, we transformed wild-type plants with a construct in which AtABCC1 is driven by the 35S promoter. We could identify only plants in which the transcript levels were maximally increased by 20–40%. However, these lines did not confer increased arsenic tolerance (Fig. S6), which may be explained by the circumstances that the overexpression was not strong enough or the PC transport activity was not the limiting factor in conferring arsenic tolerance. Therefore, we generated AtPCS1 single and AtPCS1 AtABCC1 co-overexpressing lines. No differences in arsenic tolerance could be detected between the wild type and seven AtPCS1 single overexpressing lines, despite transcript levels of AtPCS1 exceeding that of the wild type by 860–3,260% in those lines (Fig. S7). In contrast, the two plants co-transformed with AtPCS1 and AtABCC1 (MP-1 and MP-2) exhibited a consistent increase in arsenic tolerance (Fig. 5). Besides the less-chlorotic phenotype, shoot fresh weight was also increased in the plants overexpressing both genes. When exposed to 70 μM As(V), the shoot weight of these two lines corresponds to 161% and 147% of those of the wild-type and AtPCS1-overexpressing plants (Fig. 5C). Even after normalization with the growth in control medium, to take into account the slightly better growth of the double transformant in normal medium, the two AtPCS1 AtABCC1 double transformants exhibited an increased arsenic tolerance (124% and 114% of the wild type, respectively). These results show that coexpression of an ABC-type PC transporter and PCS can result in plants with enhanced arsenic tolerance.

Fig. 5.

Co-overexpression of PC synthase (AtPCS1) and AtABCC1 results in increased arsenic tolerance. (A) Wild-type plants, plants of a representative AtPCS1-overexpressing line (PCS1), and plants overexpressing AtPCS1 and AtABCC1 (MP-1 and MP-2) grown either on 110 μM As(V)-containing or control half-strength MS plates. (B) Transcript levels of AtABCC1 and AtPCS1 in the plants used in A, determined by qPCR. (C) Shoot fresh weight of wild-type and transgenic plants grown under control conditions or in the presence of 70 μM As(V). Mean ± SEM (from two independent experiments with four replicates each). *P < 0.02 by Student's t test.

Discussion

Cooperation of PC Transport and Synthesis in Arsenic Detoxification.

Previous studies have demonstrated that PCs are required for arsenic tolerance in higher plants (33, 36). PC-conjugated substrates have been suggested to be sequestered into vacuoles. However, until now no plant transporter for PCs has been identified. Our results demonstrate that AtABCC1 and AtABCC2 are both vacuolar PC transporters required for arsenic resistance in A. thaliana. The similar phenotypes of the atabcc1 atabcc2 double knockout and the cad1 mutant, which lacks PC synthesis, suggest that the sequestration of PC–As complexes into vacuoles is as critical for the detoxification of arsenic as the synthesis of PCs. A similar mechanism may work for other metal(loid)s, or alternatively, they may form bis–GS complex and be sequestered for detoxification. The main transporters for the GS-metal(loid) complexes in plants remain to be identified.

The AtABCC1 and AtABCC2 transporters and PCS may function in a concerted way in the arsenic detoxification pathway. The PCS enzyme, although present even when plants are not exposed to heavy metals or metalloids, remains inactive under normal conditions and is rapidly activated when plants take up arsenic or other toxic metal(loid)s (12, 38). Similarly, expression of AtABCC1 and AtABCC2 is not induced by arsenic in Arabidopsis, nor does the absence of one of the ABCCs result in increased transcription of the other (Fig. S8B) (39). AtABCC1 and AtABCC2 are not synthesized de novo but constitutively present in a plant cell to rapidly respond to toxic metal(loid)s and xenobiotic stresses in a way similar to PCS1. Furthermore, the synthesis of PCs seems to be positively correlated with the presence of the PC transporters (Fig. 4B and Fig. S3). In both yeasts and plants, PCs are increased when cells also express PC transporters. This observation suggests that PC synthesis undergoes a feedback regulation and that removal of PCs from the cytosol leads to an increase of its synthesis. A puzzling result was the observation in the double knockout that arsenic concentration was higher in roots and that the root-to-shoot transfer was diminished (Fig. S5), which is in contrast to the observations made in plants lacking GSH or PCs (34). This cannot be due to an alternative PC transporter, because the atabcc1 atabcc2 double mutant is very susceptible to arsenic exposure. Even at lower concentrations the atabcc1 atabcc2 knockout plants still produce PCs (Fig. 4B); however, they accumulate these peptides in the cytosol. As(V) exhibits a complex metabolism in plants, and apparently accumulation of PC–As complexes in the cytosol induces mechanisms that reduce the transfer of arsenic to the shoot, either by inducing genes that may be responsible for the translocation of phosphate from the root to the shoot or by inducing a toxicity response that has the same effect. Consistent with this possibility, atabcc1 atabcc2 plants grown hydroponically in the presence of 5 μM As(V) were reduced in phosphate contents by 16% (P = 0.05) in shoots, whereas in roots their phosphate content was similar to that of the wild type (Fig. S5).

Under our experimental conditions, overexpressing AtPCS1 alone did not result in an increased arsenic tolerance (Fig. S7). Inconsistent results have been reported using a similar approach. Li et al. (40) showed a pronounced increase when AtPCS1 was expressed in Arabidopsis, whereas Guo et al. (41) could only observe an increased tolerance when both AsPCS1 and ScGSH1 were coexpressed, indicating that thus-far-unknown parameters may play a role when PCS1s are overexpressed in Arabidopsis. We also could not observe any increase in tolerance when AtABCC1 was overexpressed alone. Only transgenic plants overexpressing both AtPCS1 and AtABCC1 exhibited an increased arsenic tolerance. Therefore, it has to be assumed that efficient increase in arsenic tolerance requires both PCS and the PC transporter. Modification of ABCC transporter and PCS expression in plants may therefore enable future engineering of reduction in the accumulation of arsenic in edible organs of plants. A similar case was reported recently for reduction of cadmium content in rice grain by overexpression of OsHMA3, a P-type ATPase (42). It will also be interesting to learn whether these two different transport systems can give a synergistic effect.

Biochemical Characteristics of AtABCC1 and AtABCC2.

Both AtABCC1 and AtABCC2 transported apoPC₂ as well as As(III)–PC₂ (Fig. 3B), although As(III)–PC₂ was the substrate more efficiently transported. The transport of apoPC₂ by the ABC transporters may supply the vacuolar apoPC₂ required for stabilization of toxic metal(loid)s transported into the vacuole by other transporters that carry unbound forms of metals and metalloids. Alternatively, or additionally, apoPC₂ might also be used to produce high-molecular-weight thiol–As complexes, which possess high binding capacity for arsenic.

Kinetic studies of the As(III)–PC₂ uptake mediated by AtABCC1 and AtABCC2 revealed that the transport activity was low at low concentrations of As(III)–PC₂ and increased steadily as the substrate concentration was raised (Fig. 3D). The minimal transport activity at low PC concentrations may allow detoxification of toxic metals more efficiently. PCs are synthesized in response to the presence of arsenic or heavy metals by activation of PCS (12, 38). It is likely that at the beginning of this reaction, PCs are accumulated, which subsequently complex arsenic or other heavy metals. If apoPC was efficiently transported into the vacuole even at very low concentrations, fewer PCs would be available in the cytosol for immediate chelation of arsenic when plants are exposed to arsenic. However, after As(III)–PC₂ is accumulated in the cytosol, it has presumably to be transported rapidly into the vacuole to avoid inhibition of PC synthesis by a feedback regulation mechanism (Fig. 4B and Fig. S3). Biochemically, the sigmoidal concentration dependence indicates allosteric regulation by multiple substrate binding sites or interaction with other proteins.

In addition to their central role as PC transporters, AtABCC1 and AtABCC2 have been reported to be involved in a broad range of detoxification processes. Both proteins transport GS conjugates (23, 26), which are formed by the attack of the GS thiolate anion to an electrophilic and potentially toxic substrate. Furthermore, AtABCC2 has been shown to transport chlorophyll catabolites, which are potentially toxic, because they can absorb light and produce harmful reactive oxygen species (43). This versatile and important role in plant detoxification requires that AtABCC1 and AtABCC2 are expressed in all tissues in a constitutive way.

Materials and Methods

Plant Materials, Growth Conditions, and Arsenical Treatments.

Arabidopsis seeds of wild-type and transgenic plants (ecotype Columbia-0) were grown on half-strength Murashige-Skoog (MS) agar plates with 1.5% sucrose in the absence or presence of the indicated concentrations of As(V) (Na₂HAsO₄) or DSMA, an arsenic herbicide (Supleco), in a controlled environment with a 16-h light/8-h dark cycle at 22 °C/18 °C for the indicated times. All ABCC/MRP knockout mutants were obtained from the Salk Genomic Analysis Laboratory, except abcc2-2, abcc11-1, and abcc12-2, which were provided by Dr. Markus Klein. (Philip Morris International, Switzerland).

Isolation of Intact Vacuoles from Arabidopsis Protoplasts and Transport Assays.

Vacuoles were prepared from Arabidopsis mesophyll protoplasts as described previously (44, 45). Transport studies with Arabidopsis mesophyll vacuoles were performed as described for barley (44, 45) but using silicone oil poly(dimethylsiloxane-co-methylphenylsiloxane) 550 (Aldrich) instead of AR200. The assay for apoPC₂ contained 200 μM PC₂ and 200 μM DTT; that for As(III)–PC₂ contained 200 μM PC_2, 400 μM As(III), and 200 μM DTT. ³⁵S-Labeled PC₂ [50 nCi (1,840 Bq)] and ³H₂O [50 nCi (1,840 Bq)] were used as tracer in every assay. The volume of vacuoles was determined using ³H₂O. Transport rates were calculated by subtracting the radioactivity taken up after 20 min from that after 2.5 min incubation.

For other methods, see SI Materials and Methods.

Note Added in Proof.

A report describing the independent identification of the related Abc2 as a vacuolar phytochelatin uptake transporter in S. pombe is in press (46).