AtMRP6/AtABCC6, an ATP-Binding Cassette transporter gene expressed during early steps of seedling development and up-regulated by cadmium in Arabidopsis thaliana

Background ABC proteins constitute one of the largest families of transporters found in all living organisms. In Arabidopsis thaliana, 120 genes encoding ABC transporters have been identified. Here, the characterization of one member of the MRP subclass, AtMRP6, is described. Results This gene, located on chromosome 3, is bordered by AtMRP3 and AtMRP7. Using real-time quantitative PCR (RT-Q-PCR) and the GUS reporter gene, we found that this gene is essentially expressed during early seedling development, in the apical meristem and at initiation point of secondary roots, especially in xylem-opposite pericycle cells where lateral roots initiate. The level of expression of AtMRP6 in response to various stresses was explored and a significant up-regulation after cadmium (Cd) treatment was detected. Among the three T-DNA insertion lines available from the Salk Institute library, two knock-out mutants, Atmrp6.1 and Atmrp6.2 were invalidated for the AtMRP6 gene. In the presence of Cd, development of leaves was more affected in the mutants than wild-type plants, whereas root elongation and ramification was comparable. Conclusion The position of AtMRP6 on chromosome 3, flanked by two other MRP genes, (all of which being induced by Cd) suggests that AtMRP6 is part of a cluster involved in metal tolerance, although additional functions in planta cannot be discarded.


Background
Contamination of soil by agronomical and industrial activities, notably heavy metals, is a major problem for human health. In the past years, decontamination by plants (phyto-remediation) has been the subject of intensive research. Some heavy metals such as copper, iron and zinc are oligo-elements essential for plant development, however they can become toxic at higher concentrations. Conversely, non-nutrient metals such as cadmium (Cd), lead and mercury are potentially toxic even at very low doses. Nonetheless, their toxicity varies between plant species. For example, metal-tolerant plants are able to grow in highly contaminated soils. Mechanisms responsible for the uptake and storage of heavy metals in plants began to be understood [1]. First after mobilization of metal ions from soils, uptake of heavy metals occurs into root cells through more or less specific channels and/or transporters [2][3][4]. In a second phase occuring in the cytoplasm metal ions are associated with amino acids, organic acids, glutathione or longer glutathione-derived peptide, phytochelatins (PCs). When plants are exposed to Cd, an increase in PCs synthesis occurs and these PCs participate in the root to shoot translocation of Cd [5]. In a third phase, glutathione and PCs-Cd complexes are excluded from the cytosol into vacuolar or extra-cellular compartments by various transporters, among which are ABC transporters [6,7].
Fifteen members of the Arabidopsis ABC transporters belong to the multidrug resistance-associated protein (MRP) subfamily [32]. MRP proteins display two hydrophobic domains (TMD) containing six membrane spans and two hydrophilic, cytosolic, nucleotide binding domains (NBD) which are organized in pairs. In most of MRP proteins, an additional hydrophobic domain (TMD 0 , including 3 to 5 transmembrane spans) is present at the N-terminal part of the transporter. In most ABC transporters, the binding and subsequent hydrolysis of ATP at their NBD provides the energy required for substrate translocation across the membrane. Structurally, each NBD exhibits one 'Walker A' and one 'Walker B' motif which is endowed by all ABC members, as well as by other ATP-binding proteins, and a highly conserved C motif or ABC transporter signature, being located between both Walker sequences, which is specific to ABC transporters. Until now, five members of this subclass (AtMRP1 to AtMRP5) have been characterized and AtMRP1, AtMRP2 and AtMRP3 have been found to exhibit glutathione S-conjugate transport activity [19,33]. In the case of AtMRP2 and AtMRP3, an additive chloro-phyll catabolites transport activity was reported [19,20]. Interestingly, AtMRP3 is also able to complement the loss of YCF1, which is an ABC transporter involved in Cd detoxification in yeast [20]. In planta, AtMRP3 is up-regulated by a Cd treatment [28,34], but the evidence that AtMRP3 is a Cd-transporter has not yet been obtained and to our knowledge there is no description of any Atmrp3 mutant in the literature till now. In addition, AtMRP4 and AtMRP5 are involved in the control of stomatal movements. More precisely AtMRP5 participates in the control of water loss via the regulation of anion and calcium channels [30,31,[35][36][37]. Here, we report the expression pattern of AtMRP6 which is part of a cluster of three MRP genes co-regulated by Cd. Two T-DNA insertion mutants were isolated, and an increased sensitivity to Cd during early stages of development was observed in these two lines.

cDNA isolation and protein organization
AtMRP6 (according to the nomenclature proposed by Martinoia and col. [32]) was directly cloned by RT-PCR using MR06-NotStart and MR06R-StopNot oligonucleotide primers (table 1) and a full-length cDNA of 4398 bp was obtained (GenBank AY052368). Alignment of this cDNA with the genomic sequence (5200 bp) from chromosome III allowed us to deduce the genomic organization of the gene. AtMRP6 extends on a 5.2 kb fragment and is spaced out into 9 exons. ( figure 1A). This cDNA was unstable in Escherichia coli, requiring a growth of the bacteria at 30°C in order to avoid mutations. Two other members of the MRP sub-family, AtMRP7 and AtMRP3, flank the AtMRP6 gene at its 5'-and 3'-end, respectively. All are oriented in the same transcription direction. AtMRP7 and AtMRP3 are the closest related genes to AtMRP6 and this cluster probably results from two successive gene duplications [38]. Mean percentage amino acid identities of AtMRP6 compared to AtMRP7 and AtMRP3 were 79.5% and 64.0%, respectively. The AtMRP6 cDNA contains an open reading frame, which encodes a 1466aminoacids polypeptide with a predicted molecular weight of 164.4 kDa. Based on a Kyte and Doolittle hydropathy plot using ProtScale and depending on the software used for transmembrane spans prediction, AtMRP6 exhibits 11 (PredTmr algorithm) to 16 (PHDhtm algorithm) transmembrane helixes. When using Aramemnon, 16 different algorithms are compared and a consensus sequence is proposed with 12 transmembrane spans. However, in this prediction, downstream from the first nucleotide-binding domain, the second half of the protein exhibits only 4 transmembrane helixes whereas 6 transmembrane spans are usually found. HMMTop_V2 (very well-known and suitable for the analysis of ABC transporters) as well as Phobius, proposed a model with 15 transmembrane helixes. Taking into account the fact i) ABC transporters should have an internal symmetry; ii) Gene structure and protein topology    the two NBD should be accessible to the cytosol; iii) the two NBD should not overlap the transmembrane region, we consider that the most probable model is the one presented in figure 1B, with at least 15 transmembrane helixes, two-halves of 6 transmembrane helixes and a TMD 0 of at least 3 transmembrane spans.

AtMRP6 can be expressed in mammalian cells but not in yeast
In order to investigate the ability of AtMRP6 to transport classical substrates of MRPs proteins, heterologous expression of the cDNA was realized in both yeast and mammalian cells (HEK-293 cells).
EGFP was fused at the C-terminal part of AtMRP6 to localize its expression in both expression systems. Particular attention was dedicated to the integrity of plasmids due to the instability of AtMRP6 in E. coli. As shown in figure 2, a weak expression of the full size transporter was observed in HEK-293 cells. In yeast, the plasmid was intact but the protein underwent a maturation step, leading to a truncated version of the transporter ( figure 2A). In these conditions, no complementation of the ∆ycf1 mutant by AtMRP6-GFP was observed (data not shown). In HEK-293 cells, AtMRP6-GFP was fully translated (figure 2B) but its expression level was low due to a weak yield of transfection and cellular expression (figure 2C), compared for instance to the GFP control (data not shown). Cell survival experiments conducted in the presence of exogenous Heterologous expression of AtMRP6 in yeast and mammalian cells  Cd in the culture medium did not allow us to distinguish vector-transformed cells from cells expressing AtMRP6 (data not shown).

AtMRP6 promoter-GUS fusion is essentially expressed in seedlings
AtMRP6 gene expression was determined by RT-Q-PCR in different tissues. As shown in figure 3A, AtMRP6 transcripts were principally detected in seedlings but at a very low level compared to the actin-2 gene. Expression was also found in roots, leaves and flowers but was absent from stems. This data was confirmed by analysis of independent homozygous transgenic lines expressing the βglucuronidase reporter gene under the control of two different promoter regions of AtMRP6, one corresponding to the intergenic region (687 pb), the other corresponding to a 2.5 kb promoter region overlapping the ORF of AtMRP7. Transgenic plants expressing both constructions exhibited the same expression pattern. The GUS reporter gene was observed in germinating seeds (figure 3B), in young seedlings essentially in cotyledons (figure 3C), in more developed seedlings at the base of leaves and in the apical meristem ( figure 3D). Expression was also detected in lateral root primordia (figure 3E), restricted to pericycle cells, which are found opposite the xylem pole on the side where lateral roots initiate (figure 3F).

AtMRP6 is up-regulated by H 2 O 2 and Cd exposure
In order to determine in which process AtMRP6 could be involved, its expression level in response to numerous stresses was investigated by RT-Q-PCR in Arabidopsis plantlets. A significant variation of AtMRP6 expression level was observed after hydrogen peroxide treatment but not in response to hormones (brassinosteroid, abscisic acid and analogous-compounds, gibberillic acid or methyl jasmonate, figure 4) or to salt or cold stress (data not shown). Concomitantly by a transcriptomic analysis of genes regulated by Cd [39], we observed that AtMRP6 was one of the most induced ABC transporter genes. Such an up-regulation by Cd was confirmed by RT-Q-PCR, AtMRP6 being up-regulated in roots after a 30-hr exposition to 5 µM Cd (figure 4).    , hormones (brassinosteroid, 1-naphtaleneacetic acid, abscissic acid, salicylic acid), continuous light or darkness, or in the presence of calcium channels inhibitors known to interfere with Cd entry into the plant (data not shown, [4]). In hydroponic conditions, wild type Columbia ecotype (Col-0), Atmrp6.1 and Atmrp6.2 KO mutant plants were exposed to 5 or 50 µM CdSO 4 , conditions that triggered an up-regulation of AtMRP6 ( figure 4). For all plant genotypes, similar Cd contents were found by ICP-AES analysis in roots and leaves as well as similar GSH, γ-EC and phytochelatin contents determined by HPLC. Finally, all genotypes exhibited an equivalent resistance to Cd in terms of root growth and development (data not shown). Since the expression of AtMRP6 was essentially pronounced in seedlings (figure 3C-D), investigation of Cd effects was evaluated in Atmrp6.1 and Atmrp6.2 seedlings when seeds were directly sown on a Cd-contaminated medium. Three weeks after germination, root elongation and ramification in the absence or presence of 1-5 µM CdSO 4 were equivalent in all plant genotypes. However, Atmrp6.1 seedlings were more affected than control plants, notably at shoot level (figure 5B). In the absence of Cd, the fresh weight of Atmrp6.1, Atmrp6.2 and wild type rosette-leaves from seedlings were similar (20.4 ± 5.1 mg, 19.5 ± 2.9 mg and 19.6 ± 5.0 mg, respectively). Conversely, after Cd treatment, the fresh weight of Atmrp6.1 and Atmrp6.2 seedlings were significantly lower compared to wild-type (3.7 ± 1.2, 4.3 ± 0.8, and 6.9 ± 1.6, respectively) (figure 5C; mean of 4 independent experiments, 2 replicates per experiment). This reduction in fresh weight of the mutants was not accompanied by a change in Cd, GSH, γ-EC or phytochelatin content.

Isolation and characterization of Atmrp6 knockout plants
Thus, it can be concluded that invalidation of AtMRP6 increases Cd-sensitivity of seedlings. The possibility of an eventual functional redundancy within the AtMRP3/ AtMRP6/AtMRP7 cluster was investigated. Since it had already been demonstrated that AtMRP3 is induced by Cd [28], we examined comparatively in wild type plants the expression levels of the three MRP genes belonging to the cluster, together with AtMRP1 as a control. As shown in figure 5D, the expression of the three genes was up-regulated by Cd in plant roots, whereas the expression level of AtMRP1 remained unchanged. The likely gene duplication at the basis of the AtMRP3/AtMRP6/AtMRP7 cluster [38] led us to investigate the expression level of AtMRP3 and AtMRP7 in the Atmrp6.1 mutant genetic background at the seedling stage of development. Whatever the presence or absence of Cd, no significant difference in AtMRP3 and AtMRP7 expression levels was observed. Therefore, invalidation of AtMRP6 was not correlated with an overexpression of AtMRP3 or AtMRP7.

Discussion
ABC transporters, especially from the MRP subfamily, are frequently involved in the detoxification of various xenobiotics, among which, heavy metals are found. Here, we tried to decipher the function of a previously uncharacterized A. thaliana gene, AtMRP6, which is flanked by two other MRPs gene on chromosome III, AtMRP3 and AtMRP7.
Analysis of AtMRP6 gene expression by RT-Q-PCR as well as by promoter GUS analysis, demonstrated that this gene is weakly expressed and has a restricted pattern of expression, mainly in germinating seeds and seedlings. Subcellular localization of AtMRP6 in planta was attempted through two different approaches. First, CaMV35s transgenic plants expressing AtMRP6-GFP were generated. Strikingly, whereas empty vector and AtMRP6 antisens plants were easily obtained, it was never the case for the sense construction, probably indicating a toxicity of this gene product under over-expressing conditions. As an alternative way to address the localization of the transporter, mesophyll cell protoplasts were transfected with AtMRP6-GFP by the classical polyethylene glycol method. No fluorescence could be observed in these conditions whereas, in control cells expressing the GFP alone, fluorescence was detected in the cytoplasm and in the nucleus. The subcellular localization of AtMRP6 could not be determined however, our experiments highlighted the difficulties when working with this gene. In addition, heterologous expression of transporters in yeast constitutes an elegant approach to screening for complementation of various mutants and also to perform flux experiments with radiolabelled compounds. In the case of AtMRP6, no complementation of the ∆ycf1 mutant could be obtained in this study: AtMRP6 being truncated ( figure 2A). We assume that this truncation of the protein was probably due to a toxicity of the transporter for the host. The development of such host toxicity is also consistent with an almost systematic mutation of the corresponding plasmid that occurred in bacteria at 37°C. When looking for an alternative expression system for AtMRP6, HEK-293 cells were transfected. As shown in figure 2B-C, AtMRP6 expression was successfully obtained. However, despite many efforts (assays with various plasmids such as pCi, pcDNA6 or pEGFP, optimization of the Kozak sequence, use of different cationic lipid transfection reagents), the Isolation, phenotypic characterization of AtMRP6 knock-out plants and co-regulation of the AtMRP3, 6, 7 genes cluster  Results obtained in this study by RT-Q-PCR (figure 5D) and within a previous transcriptomic analysis [39], demonstrate that AtMRP6 expression is up-regulated in roots within 30-hr by 5 µM Cd. Interestingly, not only AtMRP6, but the three members of the gene cluster were also upregulated by after Cd exposition. These results are in accordance with an enhanced level of both AtMRP3 and AtMRP6 transcripts, reported previously in cDNA microarray experiments [34]. It has already been reported that AtMRP3 can be important in Cd detoxification since its heterologous expression in the yeast strain deprived of ycf1 restores Cd tolerance [20]. However, in Arabidopsis, despite the fact that Cd-related induction of AtMRP3 is correlated with Cd uptake after a short metal exposure [28], whether AtMRP3 is involved in Cd transport or in the detoxification of toxic compounds produced after the metal stress awaits future studies. In the case of AtMRP7, very little data is available about its tissue expression [38] and its function is still unknown. A fourth gene, located upstream of the MRP cluster, is also up-regulated in roots by Cd treatment: it encodes a mitochondrial-localized serine acetyl-transferase, SAT3 or serat2.2 (At3g13110; [40]). This enzyme catalyzes the formation of O-acetyl-Ser from L-Ser and acetyl-CoA, which is used in cysteine synthesis, an important component of glutathione. Expression of the bacterial enzyme in tobacco led to an increase in cysteine and glutathione contents [41]. Moreover, the high activity of SAT is associated with nickel tolerance in Thlaspi nickel hyper-accumulators [42] suggesting a major role of SAT in heavy metal resistance. Recently, expression of SAT3 has been achieved in tobacco; however no experiments have been performed in relation to Cd [43]. All these results suggest that these four genes (AtMRP3, AtMRP6, AtMRP7 and SAT3), oriented in the same transcription direction on chromosome III, are members of a Cd-responding cluster. This hypothesis is also supported by the fact that all these genes are up-regulated by a Cd treatment into the same organ (roots) and in the same time scale (24-hr for SAT3, [40]; 30-hr for the three MRP genes). Identification of such Cd-responsive elements would be useful in the context of phytoremediation strategies either to drive the expression of cadmium-transporter or reporter genes that might be used as biosensors of contaminated soils.
At the sight of the expression pattern of this gene (figure 3), a phenotype was expected at root level in T-DNA KO lines. One cannot exclude that the neighboring MRP genes might complement the deletion of AtMRP6. For this reason, the expression levels of AtMRP3 and AtMRP7 were compared in wild type plants and in Atmrp6 genetic backgrounds. No significant difference in their expression lev-els was detected in the presence or in the absence of cadmium (data not shown). Thus, it is possible that if a mechanism of gene compensation is taking place in Atmrp6 KO plants, it involves (an)other gene(s) than AtMRP3 and AtMRP7 or that the basal levels of expression of AtMRP3/7 are sufficient to compensate for the absence of AtMRP6. Alternatively, these two genes could be up-regulated in the few cells expressing AtMRP6 in roots without significantly affecting their global root-expression level. The screening of several dozen conditions to observe a phenotype for Atmrp6 KO plants allowed us to show that, in the presence of Cd, the deletion of AtMRP6 has a small but significant impact on the development of primary leaves whereas roots elongation and ramification were unaffected. This phenotype was lost in 3-to 5-week-old plant, probably because at this developmental stage, Cd translocation from root to shoot is much lower, as already reported for AtMRP3 [34].

Conclusion
We have shown that AtMRP6, AtMRP3 and AtMRP7, as well as SAT3, are part of a Cd-regulated gene cluster. The narrow expression profile of the AtMRP6 gene in the plant, essentially during the first step of seedling development might explain the discrete phenotype observed in T-DNA KO lines and is more consistent with a function of this transporter in plant growth/development rather than in Cd detoxification. If our results demonstrate that AtMRP6 is part of a cluster involved in metal tolerance, and that invalidation of this gene leads to a higher susceptibility of young seedlings, the precise function of this transporter in the plant will remain to be determined.

cDNAisolation and subcloning in expression systems
Total RNAs from Arabidopsis plantlets were extracted by the Trizol™ method. Complementary DNAs were synthe-sized by using the First-Strand cDNA Synthesis Kit according to the manufactor's instructions (Amersham). PCR were realized using a high fidelity Taq polymerase with different primers MR06-NotStart and MR06R-StopNot showed in table 1. The NotI-flanked PCR product was cloned in the pCR-XL-Topo from Invitrogen ® and sequenced. The AtMRP6 cDNA sequence has been deposited in GenBank under the accession number AY052368. In order to localize AtMRP6, the C-terminal part of the cDNA was epitope-tagged with GFP. The plasmids pEGFP-N2 (from BD Biosciences ® ) and pCR-XL-AtMRP6 were used to generate the AtMRP6-EGFP-N2 fusion by the "splicing by overlap extension" technique as already described [44]. For this purpose, primers used were AtMRP6-GFP_A, AtMRP6-GFP_C, AtMRP6-GFP_B, and Rev_fin_GFP+NotI (table 1). The different sub-clonings from the pCR-XL-Topo AtMRP6-GFP to the yeast vector pYES2 (Invitrogen ® ) and the mammalian expression vector pCI (Promega ® ) were realized by a single restriction with NotI.

Generation of AtMRP6::GUS lines
Two AtMRP6 promoters, corresponding to the intergenic region (687 bp) and to a 2511 bp sequence upstream of the start codon, were amplified on genomic DNA from Col-0 using specific primers (table 1) inserting SbfI and XmaI restriction sites and with PyroBest taq polymerase (Takara). PCR products were cloned in pGEM-T easy vector and verified by sequencing. SbfI-XmaI fragments were then inserted in pBI101 plant vector opened with the same enzymes. Arabidopsis thaliana Col-0 plants were transformed using Agrobacterium tumefaciens. Seedlings were selected on 30 µM kanamycin plates and six independent lines for each construction exhibiting a similar GUS pattern were selected.

Identification of Atmrp6 knockout mutants
Homozygous T-DNA insertion knockout mutants of AtMRP6 (At3g13090) were identified from SALK #110544 (Atmrp6.1), SALK #084905 (Atmrp6.2) and SALK #091430 (Atmrp6.3) seeds were obtained from the NASC (Nottingham, GB). A corresponding wild-type for each mutant was identified in the lineage of heterozygous T-DNA insertion mutants and were designated as Col-0 in the following. The T-DNA insertion site was confirmed by DNA sequencing. The presence of only one T-DNA insertion site was determined by Southern-blot as well as by segregation analysis of plantlets on 30 µM kanamycin.

Real-Time quantitative RT-PCR
Total RNA was extracted from leaves, roots, stems, flowers, seedlings and germinating seeds, using Trizol ® according to the manufacturer's instruction (Invitrogen). Genomic DNA was removed from the samples using Dnase I (Ambion). Reverse transcription was performed using the First Strand cDNA Synthesis kit (Amersham) and an oligo-dT primer. PCRs were carried out using the SYBR Green Mix (Takara) in an optical 96-wells plate with the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems). Specific primers for each gene were designed using the LightCycler Probe Design Software (Roche). The presence of a single amplicon in each PCR reaction was confirmed by dissociation curves and by loading on agarose gel. Standard curves were derived from reactions with actin-2 (At5g09810) specific primers, and a dilutions' series of cDNA templates. Relative quantity of transcripts analysed in each RNA sample was normalized to the standard curve and the mean value was calculated from three to four independent replicates.

Cd treatment
For early Cd exposure, seeds were sown directly on agar plates containing 1 or 5 µM CdSO 4 . A longer vernalisation period of 4 days was used and seedlings were grown in a 14-hr light, 21°C, 10-hr dark, 18°C cycle for 21 days. Leaves were harvested and fresh weights were determined. Cd and thiol contents were measured by ICP-AES and by HPLC, respectively.
For late Cd treatment, 3-4 weeks old plants grown on sand were transfered in hydroponic conditions in a similar light/dark period at 23°C/19°C respectively, 250 µmol.m -2 .s -1 and 75% relative humidity. Cd treatments were carried out by adding 5 or 50 µM CdSO 4 in nutrient solution for 6, 24 or 30 h as previously described [39]. Shoots and roots were harvested separately and supplied for Cd quantification by ICP-AES (6-hr and 30-hr) or for thiols measurement by HPLC (30-hr).

Determination of Cd content
Fresh leaves, roots and seedlings from Cd-treated and untreated plants were dried 72-hr minimum at 50°C and mineralized in 70% HNO 3 at 210°C for 10 min. The Cd concentration in the solution was determined using inductively coupled plasma optical emission spectroscopy