The ABC transporter ABCG36 is required for cadmium tolerance in rice.

Cadmium (Cd) is a highly toxic heavy metal in nature, which causes severe damage to plant growth. The molecular mechanisms for Cd detoxification are poorly understood. Here, we report that a G-type ATP-binding cassette transporter, OsABCG36, is involved in Cd tolerance in rice. OsABCG36 was expressed in both roots and shoots at a low level, but expression in the roots rather than the shoots was greatly up-regulated by a short exposure to Cd. A spatial expression analysis showed that Cd-induced expression of OsABCG36 was found in both the root tip and the mature root region. Transient expression of OsABCG36 in rice protoplast cells showed that it was localized to the plasma membrane. Immunostaining showed that OsABCG36 was localized in all root cells except the epidermal cells. Knockout of OsABCG36 resulted in increased Cd accumulation in root cell sap and enhanced Cd sensitivity, but did not affect tolerance to other metals including Al, Zn, Cu, and Pb. The concentration of Cd in the shoots was similar between the knockout lines and wild-type rice. Heterologous expression of OsABCG36 in yeast showed an efflux activity for Cd, but not for Zn. Taken together, our results indicate that OsABCG36 is not involved in Cd accumulation in the shoots, but is required for Cd tolerance by exporting Cd or Cd conjugates from the root cells in rice.

Introduction

Cadmium (Cd) is one of the naturally occurring heavy metals that is highly toxic to all organisms, including plants. Cd excess in plants can cause disruption of nutrient homeostasis, dysfunction of proteins, DNA and membrane damage, and the accumulation of toxic reactive oxygen species, resulting in significant reductions of crop growth and production (Sandalio ; Hall, 2002; Boominathan and Doran, 2003). More seriously, Cd accumulation in the edible parts of crops potentially threatens human and animal health.

To adapt to Cd-contaminated environments, plants have evolved diverse mechanisms to cope with Cd stress, such as Cd extrusion, chelation, and sequestration (Hall, 2002; Weber ; Lin and Aarts, 2012). Several studies have shown that ATP-binding cassette (ABC) transporters may play important roles in Cd tolerance in plants (Kim , 2007; Oda ; Park ; Brunetti ). For example, AtABCG36/AtPDR8 was reported to be essential for Cd resistance in Arabidopsis (Kim ). This protein is localized at the plasma membrane of root cells and functions as an efflux transporter for Cd or Cd conjugates. On the other hand, two ABCC-type transporters, AtABCC1 and AtABCC2, are involved in vacuolar sequestration of Cd in Arabidopsis (Park ). Knockout of these genes resulted in hypersensitivity to Cd. Recently, another ABCC-type transporter, AtABCC3, was reported to be involved in Cd tolerance by transporting phytochelatin–Cd complexes into the vacuole in Arabidopsis (Brunetti ). In rice, there are at least 120 ABC transporters (Garcia ), but only few of them have been characterized in terms of Cd tolerance. One of them, OsABCG43, is a Cd-inducible transporter (Oda ). Its expression in yeast increased Cd tolerance (Oda ), but its exact role in rice is unknown. Another ABCG-type transporter gene, OsABCG36/OsPDR9, was reported to be induced by Cd and was suggested to play a possible role in heavy metal stress (Moons, 2003, 2008). However, the role of OsABCG36/OsPDR9 in Cd tolerance has not been investigated. In the present study, we functionally characterized OsABCG36 in terms of its gene expression, cellular and subcellular localization, and transport activity. We also obtained two independent mutant lines of OsABCG36 by using the CRISPR/Cas9 technique and compared their Cd tolerance and accumulation with that of wild-type rice. Our results showed that OsABCG36 is involved in Cd tolerance by transporting Cd out of the root cells.

Materials and methods

Generation of OsABCG36 knockout lines

To create the OsABCG36 knockout lines, the CRISPR/Cas9 genome-targeting system was used. The pCRISPR-OsABCG36 plasmid with two OsABCG36-specific target sites was constructed as described by Ma . Briefly, two specific target sequences (CGCTCGGCATTCTGCCCAAC and GACCTACAACGGGCACGGCA) within OsABCG36 were selected by a BLAST search (http://blast.ncbi.nlm.nih.gov/Blast.cgi) of the target sequences, including protospacer adjacent motif (PAM) sequence, against the rice genome sequence. These two sequences should have a difference of at least two bases compared with similar non-target sequences within the PAM or PAM-proximal region and have more than five base mismatches in the PAM distal region to non-target sequences based on off-target analysis (http://skl.scau.edu.cn/offtarget/). Then, target sequences were introduced into sgRNA expression cassettes by overlapping PCR, producing pU6a-OsABCG36-sgRNA or pU3-OsABCG36-sgRNA fragments. Using restriction–ligation reactions containing BsaI and T4 DNA ligase, these fragments were cloned into pYLCRISPR/Cas9Pubi, resulting in the pCRISPR-OsABCG36 constructs. These constructs were introduced into Agrobacterium tumefaciens strain EHA101 and transformed into the wild-type rice (Oryza sativa cv. Nipponbare). Mutation detection was carried out using primer pairs flanking the OsABCG36-specific target sites. The PCR products were sequenced directly using OsABCG36-specific primers. Homozygous mutants of OsABCG36 were selected for further phenotypic analysis as described below.

Plant materials and growth conditions

The wild-type rice (cv. Nipponbare) and two independent OsABCG36 knockout lines (osabcg36-1, osabcg36-2) were used in this study. Rice seeds were germinated in deionized water in a growth chamber for 2 days in darkness at 28 °C and then placed on a net floating on a solution containing 0.5 mM CaCl2. Seedlings were grown for 2 days at 28 °C and used for various experiments.

RNA isolation and gene expression analysis

To examine the expression pattern of OsABCG36, wild-type rice seedlings (7 days old) were exposed to different Cd concentrations (0–20 μM) for different periods of time. Root fragments (0–1 cm and 1–2 cm from the root tip) and shoots were sampled for RNA extraction, with three replicates for each sample. Total RNA was extracted using the TRIzol reagent kit (Life Technologies) according to the manufacturer’s instructions. Total RNA (1 μg) was used for first-strand cDNA synthesis using a Hiscript II Q RT SuperMix Kit (Vazyme). Quantitative reverse transcription–PCR (qRT–PCR) was performed with ChanQTM SYBR Color qPCR Master Mix (Vazyme) on a StepOnePlus Real-Time PCR System (Analytik Jena AG). The primers for gene expression analysis of OsABCG36 were 5′-ATTCTAGCAAGAGAGCAAGTG-3′ and 5′-GGTCTCATTGGAGGCAGAG-3′. The primers for gene expression analysis of OsABCG37 were 5′-AACACCGTCGAGGCAATCGG-3′ and 5′-TCCCGTGTCCATTGTAGGTC-3′. The primers for gene expression analysis of OsABCG44 were 5′-CCTTCGATGAGCTGTTCCTG-3′ and 5′-TCCTGTGCCAGTGTGGTTAC-3′. Histone H3 was used as an internal standard, with the primers 5′-GGTCAACTTGTTGATTCCCCTCT-3′ and 5′-AACCGCAAAATCCAAAGAACG-3′.

Subcellular localization of OsABCG36

To construct the GFP-OsABCG36 fusion gene, OsABCG36 cDNA was amplified from the Nipponbare cDNA by PCR using the OsABCG36-specific primers 5′-AAGCTTCGATGGACGCGGCGGGGGAGATCCAGAA-3′ (HindIII site in italic text) and 5′-GGATCCTCATCTCTTCTGGAAGTTGAACTT-3′ (BamHI site in italic text). The amplified cDNA fragment was cloned into the pYL322-GFP vector after the GFP coding region (Ma ), producing the GFP-OsABCG36 construct.

To construct a plasma membrane-localized fluorescence marker protein, mCherry-OsRac3 (Chen ), we amplified the full-length coding sequence of OsRac3 from the Nipponbare cDNA. The primer sequences 5′-gcatggacgagctgtacaagATGGCGTCC AGCGCCTCCCGGTTC-3′ and 5′-GAAATTCGAGCTTCTCGA GTTAttaGGATTTGAAGCATGAC-3′ were used for amplification (lower-case text indicates the bases that can anneal to the flanking sequences of the insertion site on the target plasmid). The resulting fragment was then cloned in frame after the mCherry coding region into the p35S-mCherry-NosT vector by the Ω-PCR strategy (Chen ).

The plasmid DNA GFP-OsABCG36 or GFP together with a plasma membrane-localized marker, mCherry-OsRac3, or a tonoplast-localized marker, mCherry-AtTPK (Zeng ), was introduced into rice leaf protoplasts by polyethylene glycol-mediated transformation as previously described by Chen . After transformation, cells were incubated in the dark at room temperature for 12–15 h. Fluorescence images were captured using a confocal laser scanning microscope (TCS SP8; Leica).

Cell and tissue specificity of OsABCG36 expression

To investigate the cell and tissue specificity of OsABCG36 expression, we introduced the transformation vector carrying ProOsABCG36-GFP fusion into rice (cv. Nipponbare). The promoter (2.026 kb) of OsABCG36 was amplified by PCR from Nipponbare genomic DNA using the primers 5′-AAGCTTCAAAATGAG GATCCAATGGATT-3′ (HindIII site in italic text) and 5′-GTCGACCTCCCACC ACCACCAAAACC-3′ (SalI site in italic text). Using HindIII and Sal, the amplified fragment was cloned into the pCAMBIA1300-GFP vector carrying the GFP gene and the terminator of the nopaline synthase gene, producing the ProOsABCG36-GFP construct. The resulting construct was transformed into A. tumefaciens strain EHA101, which was introduced into the wild-type rice (cv. Nipponbare) by A. tumefaciens-mediated transformation.

The roots of the wild-type rice and the transgenic line carrying ProABCG36-GFP were sampled for immunostaining. After immunostaining using an antibody to GFP as described by Yamaji and Ma (2007), the GFP signal was observed by confocal laser scanning microscopy (TCS SP8; Leica).

Yeast assay

The entire open reading frame of OsABCG36 was amplified from the wild-type rice cDNA by PCR using the primers 5′-AAGCTTAAAATGGACG CGGCGGGGGAGATCCA-3′ (HindIII site in italic text) and 5′-GGATCCTCATCTC TTCTGGAAGTTGAACT-3’ (BamHI site in italic text). Using HindIII and BamHI, the amplified fragment was cloned into the pYES2 vector (Invitrogen), producing the pYES2-OsABCG36 construct.

The yeast strain used in this study was Δycf1 (MATalpha; Δtrp1; Δhis3; Δleu2; Δura3; Δycf::TRY1). For evaluation of Cd tolerance, pYES2-OsABCG36 and empty vector pYES2 were transformed into the Δycf1 mutant strain. Transformants carrying the plasmid were spotted on to solid SD-uracil medium (2% galactose) with 0, 5, 10, or 15 μM CdSO4. The plates were photographed after incubation at 30 °C for 3 days. For evaluation of Cd efflux in yeast liquid culture, transformants were selected on uracil-deficient medium and grown in liquid SD-uracil medium (2% glucose). Cells at mid-exponential phase were harvested and transferred to SD-uracil medium (2% galactose) at pH 4.5. After cells had been cultured for 4 h, the OD600 of each strain was adjusted to 4.0. CdSO4 was then added to the medium to a final concentration of 20 µM. After 4 h of incubation with shaking, cells were collected and washed once with deionized water. Cells (OD600=4.0) were suspended in fresh SD-uracil medium (2% galactose) without Cd and grown at 30 °C for another 4 h. The cells were harvested every 2 h and digested with 2 M HCl. The concentration of Cd in the digest solution was measured as described below. All experiments were independently conducted at least three times.

For evaluation of Zn tolerance, the growth of yeast strain zrt1cot1 (MATa; his3Δ1; leu2Δ0; met15Δ0; ura3Δ0; zrc1::natMX; cot1::kanMX4) at different Zn concentrations was tested on a synthetic solid medium containing 2% (w/v) galactose, 0.67% yeast nitrogen base without metal, 0.2% appropriate amino acids, and 2% agar at pH 6.0 with 50 mM MES. Transformants carrying OsABCG36 or the empty vector pYES2 were spotted on to the medium supplemented with 0, 60, 80, or 100 μM ZnSO4. The plates were photographed after incubation at 30 °C for 3 days.

To examine the subcellular localization of OsABCG36 in yeast, the GFP-OsABCG36 fragment from the pYL322-GFP-OsABCG36 construct described above was cloned into the pYES2 vector, producing the pYES2-GFP-OsABCG36 construct. The resulting construct was introduced into the Δycf1 mutant strain according to the manufacturer’s protocols (S.c.easy comp transformation kit; Invitrogen). The plasma membrane of the yeast cells was stained by CellMaskTM Deep Red plasma membrane stain (Thermo Fisher Scientific) (Wang ). The fluorescence signal was observed by a confocal laser scanning microscope (TCS SP8; Leica).

Phenotypic analysis of OsABCG36 knockout lines

To compare the tolerance of the knockout line to different metals with that of wild-type rice, 5-day-old seedlings of the wild-type rice and two OsABCG36 knockout lines were exposed to a solution containing 30 μM Cd, 0.5 μM Cu, 100 μM Zn, 30 μM Al, or 10 μM Pb at pH 4.5 for 24 h. The root length of each seedling was measured with a ruler before and after the treatments, with five seedlings measured per metal treatment.

To determine Cd accumulation in the root cell sap, 5-day-old seedlings of wild-type rice and two OsABCG36 knockout lines were exposed to 20 μM Cd (pH 4.5) in 0.5 mM CaCl2 solution. After 8 h, root segments (0–1 and 1–2 cm from the root tip) were sampled and placed in a ultrafree-MC centrifugal filter unit (Millipore, Billerica, MA, USA). The units were centrifuged at 3000 g for 10 min and then immediately frozen at –80 °C overnight. The cell sap was collected by centrifugation at 20 400 g for 10 min after thawing the samples at room temperature. The Cd concentration in the solution was determined as described below.

To assess the uptake of Cd, 25-day-old seedlings were exposed to a 1/2 Kimura B solution containing 5 μM Cd in a 5 litre pot for 12 h. The roots were washed with 5 mM CaCl2 three times and sampled by cutting with a sharp blade. The samples were dried in an oven for 3 days at 70 °C. The samples were then digested and Cd concentration was determined as described below. Four biological replicates were made for this experiment.

To compare Cd accumulation in different organs, 15-day-old seedlings of wild-type rice and two OsABCG36 knockout lines were exposed to a 1/2 Kimura B nutrient solution (pH 5.6) containing 0, 0.1, 1, or 5 µM Cd as CdSO4. This solution was renewed every 2 days. Plants were grown in a glasshouse at 25–30 °C under natural conditions. After 14 days, the roots and shoots were harvested after washing with 5 mM CaCl2 and were subjected to determination of mineral elements as described below. Each treatment was conducted with four biological replicates.

Cd concentration in xylem sap

Hydroponically grown rice seedlings (25 days old) were exposed to 5 μM Cd for 3, 6, or 9 days. The seedlings’ shoots were then cut off at 2 cm above the roots by using a razor blade. The first drop of xylem sap was discarded. Xylem sap was collected for 1 h using a micropipette and subjected to determination of mineral elements as described below.

Determination of mineral elements

All samples were dried at 70 °C in an oven and then digested with 65% HNO3. The metal concentration in the root cell sap, xylem sap, and the digested solution was determined by ICP-MS (Plasma Quant MS; Analytik Jena AG).

Results

Sequence and phylogenetic analysis of OsABCG36

The full-length open reading frame of OsABCG36 (LOC_Os01g42380.1) was cloned based on information available in a public database (http://rice.plantbiology.msu.edu/). The cloned sequence was identical to that in the database. OsABCG36 consists of 21 exons and 20 introns, encoding a protein of 1457 amino acids (see Supplementary Fig. S1A, B at JXB online). It belongs to the G subfamily of rice ABC transporters. Phylogenetic analysis identified homologs of OsABCG36 in Arabidopsis and rice, which are divided into half-size and full-size ABC transporters (Supplementary Fig. S1C). OsABCG36 belongs to the full-size group and shares 57% identity with AtABCG36 and OsABCG43, which are known to confer Cd tolerance in Arabidopsis and yeast, respectively (Kim ; Oda ).

Expression pattern of OsABCG36

The expression of OsABCG36 was detected in both the roots and shoots at a low level (Fig. 1A; Supplementary Fig. S2A). The accumulation of OsABCG36 mRNA was significantly enhanced by Cd in the root, but not in the shoot (Fig. 1A; Supplementary Fig. S2A). Furthermore, expression in roots was induced by Cd in a dose-dependent manner (Fig. 1B). The expression level was similar in the root tip (0–1 cm) and the mature zone (1–2 cm) in both the presence and absence of Cd (Fig. 1C). A time-course analysis of Cd-treated roots showed that the expression of OsABCG36 was up-regulated rapidly by Cd and reached its maximum at 3 h (Fig. 1D). These results suggested that OsABCG36 is regulated by Cd.

Fig. 1.

Expression pattern of OsABCG36 in rice. (A) Expression of OsABCG36 in roots and shoots. Rice seedlings were exposed to a solution containing 10 μM Cd for 6 h. (B) Dose-dependent expression of OsABCG36 in the roots. Rice seedlings were exposed to a solution containing different Cd concentrations for 3 h. (C) Expression of OsABCG36 in different root regions. The roots were exposed to 0.5 mM CaCl2 solution (pH 4.5) containing 10 μM Cd for 3 h. RNA was extracted from the root tips (0–1 cm) and basal root region (1–2 cm). (D) Time-dependent expression of OsABCG36 in the roots. Rice seedlings were exposed to a solution containing 10 μM Cd for different times. The expression of OsABCG36 was determined by qRT–PCR. Histone H3 was used as an internal standard. Expression relative to seedlings not treated with Cd is shown. Data are means ±SD of three biological replicates. Asterisks indicate significant differences between the Cd-untreated sample and the Cd treatments (Student’s t-test; **P<0.01).

We also tested the response of other five OsABCG36 homologs (OsABCG32, OsABCG34, OsABCG35, OsABCG37, and OsABCG44) to Cd in rice. Among these, only OsABCG37 and OsABCG44 were expressed in roots, and their expression was not up-regulated by Cd (Supplementary Fig. S2B–D).

Tissue and cell specificity of OsABCG36 expression

To examine the tissue specificity of OsABCG36 expression in rice root, we generated transgenic rice lines expressing the OsABCG36 pro::GFP construct. Immunostaining with an antibody to GFP showed that the signal was detected in the root cells (Fig. 2B). Furthermore, the signal in all root cells except the epidermal cells was significantly enhanced by exposure to Cd (Fig. 2C). No signal was detected in the wild-type root (Fig. 2A), indicating that the anti-GFP antibody is specific.

Fig. 2.

Tissue and cell specificity of OsABCG36 expression. Roots of wild-type rice (A) and transgenic rice carrying the OsABCG36 promoter fused with GFP (B, C) were used for immunostaining after exposure to 0 (A, B) or 10 μM Cd (C) for 24 h. The fluorescent signal of anti-GFP antibody is shown. Scale bar=100 μm. (This figure is available in colour at JXB online.)

Based on WoLF PSORT (https://www.genscript.com/tools/wolf-psort) analysis, OsABCG36 was predicted to be localized on the plasma membrane. To assess the subcellular localization of OsABCG36 in plant cells, we co-expressed GFP empty vector or the GFP-OsABCG36 fusion with the plasma membrane-localized marker mCherry-OsRac3 (Chen ) and the tonoplast-localized marker mCherry-AtTPK (Zeng ) in rice protoplasts. In contrast to GFP alone, which was widely distributed in the cytoplasm and nucleus (Fig. 3), the GFP-OsABCG36 signal was mainly co-localized with the fluorescence of mCherry-OsRac3 (Fig. 3), while it was separated from the fluorescence of mCherry-AtTPK (Supplementary Fig. S3), confirming that OsABCG36 is a plasma-membrane-localized protein.

Fig. 3.

Subcellular localization of OsABCG36. GFP alone or the GFP-OsABCG36 fusion was co-expressed with mCherry-OsRac3, a plasma-membrane-localized marker, in rice leaf protoplast cells. The upper panels show the localization of GFP and mCherry-OsRac3 as a control. The lower panels show the localization of GFP-OsABCG36 and mCherry-OsRac3. GFP, mCherry, and merged GFP and mCherry fluorescence are shown. Scale bar=10 μm. (This figure is available in colour at JXB online.)

Transport activity for Cd in yeast

To examine whether OsABCG36 has transport activity for Cd, we expressed it in the Cd-sensitive yeast mutant Δycf1. On SD-Gal medium without Cd, yeast cells carrying the vector control or OsABCG36 showed similar growth (Fig. 4A). However, on SD-Gal medium containing Cd, the yeast cells expressing OsABCG36 exhibited better growth than those expressing the vector control (Fig. 4A). Furthermore, we pretreated yeast cells carrying OsABCG36 or the vector control with Cd for 4 h, then transferred them to liquid medium without Cd. During the Cd pretreatment period, the OsABCG36-expressing yeast cells showed similar Cd content to those carrying the empty vector; however, the OsABCG36-expressing cells showed a significant decrease of Cd content compared with the vector control during the Cd release period (Fig. 4B). The fluorescent signal of GFP-OsABCG36 was partially colocalized with the signal of the plasma membrane marker (Supplementary Fig. S4), suggesting that at least part of OsABCG36 was localized to the plasma membrane in yeast cells. These results indicated that OsABCG36 functions as an efflux transporter for Cd in yeast.

Fig. 4.

Transport activity of OsABCG36 for Cd in yeast. (A) Effect of OsABCG36 expression on Cd tolerance. Yeast cells (Δycf1) carrying empty vector pYES2 or OsABCG36 were spotted on to SD-uracil medium with or without Cd at different dilutions. The plates were incubated at 30 °C for 3 d. (B) Transport activity of OsABCG36 for Cd. Yeast cells carrying empty vector pYES2 or OsABCG36 were exposed to 20 μM Cd at pH 4.5 for 4 h, followed by exposure to a solution without Cd. Cells were harvested at different time points and subjected to Cd determination by ICP-MS after digestion with 2 M HCl. Asterisks represent significant differences from the cells carrying empty vector (*P<0.05, **P<0.01; Tukey’s test)). Data are presented as means ±SD (n = 3).

In addition, we expressed OsABCG36 in a Zn-hypersensitive yeast mutant, Δzrt1cot1. The growth of the yeast cells expressing OsABCG36 was similar to that of cells carrying the empty vector control under treatment with Zn (Supplementary Fig. S5), suggesting that OsABCG36 likely has no transport activity for Zn in yeast.

Cd tolerance of OsABCG36 knockout lines

To examine the role of OsABCG36 in Cd tolerance in rice, we generated two independent knockout transgenic lines of OsABCG36 using CRISPR/Cas9 technology and compared their Cd tolerance. These mutant lines carried a deletion or insertion, which led to a frame shift and the premature termination of OsABCG36, respectively (Fig. 5A, B). In the absence of Cd, the root and shoot growth of the two OsABCG36 knockout lines was similar to that of the wild-type rice (Fig. 5C; Supplementary Fig. S6). However, the root growth of the knockout lines was more inhibited than that of the wild-type rice in the presence of 2 µM Cd (Fig. 5D, E). We also evaluated the tolerance of the knockout lines to other metals, including Cu, Zn, Al, and Pb. There were no differences in tolerance to these metals between the knockout lines and the wild-type rice (Fig. 5F; Supplementary Fig. S7). These results suggested that the sensitivity of the OsABCG36 knockout lines to Cd is specific.

Fig. 5.

Sensitivity analysis of OsABCG36 knockout lines to toxic metals. (A) The two target sites of OsABCG36 used in the CRISPR/Cas9 targeting system. Black boxes represent exons and the lines between boxes represent introns. The triangles above the second and fourth exons represent the target sites in the CRISPR/Cas9 system. (B) OsABCG36 sequence of two independent CRISPR-mutated lines. The dashed boxes indicated the mutation sites. Underlines indicate the PAM sequences. WT, wild type. (C, D) Cd sensitivity of the WT and two OsABCG36 knockout lines. Germinated seeds were exposed to 0.5 mM CaCl2 solution (pH 4.5) containing 0 or 2 μM Cd for 5 days. Scale bar=5 cm. (E) Effect of Cd on root and shoot length. (F) Effect of toxic metals on root elongation. Seedlings (5 days old) were exposed to 0.5 mM CaCl2 solution (pH 4.5) alone or containing 30 μM Cd, 30 μM Al, 100 μM Zn, or 0.5 μM Cu for 24 h. Relative root elongation refers to (root elongation with metals)/(root elongation without metals) ×100. Data are means ±SD (n = 5). (G) Cd concentration in the root cell sap in different root segments. Roots of the WT and two OsABCG36 mutant lines were exposed to 0.5 mM CaCl2 (pH 4.5) containing 20 μM Cd for 8 h; root tips (0–1 cm) and basal roots (1–2 cm) were excised and Cd concentration in the root cell sap was determined by ICP-MS. (H) Root Cd concentration of WT and OsABCG36 mutant lines. Seedlings (25 days old) were exposed to a nutrient solution containing 5 μM CdSO4 for 12 h. Asterisks indicate significant differences between the WT and OsABCG36 knockout lines (*P<0.05, **P<0.01; Tukey’s test). (This figure is available in colour at JXB online.)

We further compared the Cd concentration of cell sap in the root tips (0–1 cm) and mature regions (1–2 cm) in the knockout lines and wild-type plants. The OsABCG36 knockout lines showed higher Cd concentration in the cell sap in both root segments than wild-type plants (Fig. 5G). Furthermore, we conducted a Cd uptake experiment with the whole roots of the knockout lines and wild-type plants. The Cd content in the knockout lines was higher than that in the wild-type plants (Fig. 5H). These results indicated that OsABCG36 is able to transport Cd out of the cell for Cd detoxification.

Role of OsABCG36 in Cd accumulation in rice

To test whether OsABCG36 is also involved in Cd accumulation, we grew two knockout lines and wild-type plants under hydroponic conditions with three different Cd concentrations for 2 weeks. In the absence of Cd or in the presence of a low Cd concentration (0.1 or 1 μM), the growth of the wild-type and knockout lines was similar (Fig. 6A, C, E; Supplementary Fig. S8). There were no differences between wild-type and knockout lines in the concentration of Cd in the roots and shoots (Fig. 6D, F). However, at a higher Cd concentration (5 μM), the root length of the knockout lines was shorter than that of the wild-type rice (Fig. 6B). The dry weight of the roots was also lower than that of the wild-type (Fig. 6C). Mineral analysis showed that the concentration of Cd in roots was significantly higher in the knockout lines than in the wild-type (Fig. 6D), but the Cd concentration in shoots was similar in the knockout lines and wild-type plants (Fig. 6F). The concentrations of Zn, Cu, Fe, and Mn in the roots and shoots were also similar in all lines (Supplementary Fig. S9). Furthermore, the Cd concentration in the xylem sap was similar in all lines under long-term Cd treatment (Supplementary Fig. S10). These data indicated that the increased root Cd concentration of the knockout lines in treatments with high Cd may be attributed to failure to transport Cd out of the cell, resulting in greater Cd accumulation in the cytosol, and that OsABCG36 does not function in Cd accumulation in the aerial parts of rice.

Fig. 6.

Growth and Cd concentrations of OsABCG36 knockout lines. (A, B) Growth of wild-type (WT) rice and two knockout lines grown in a nutrient solution containing 0 or 5 μM CdSO4 for 2 weeks. Scale bar=20 cm. (C, E) Dry weight of the roots (C) and the shoots (E). (D, F) Concentration of Cd in the roots (D) and the shoots (F). The Cd concentration in the roots and shoots was determined by ICP-MS. Data are means ±SD of three biological replicates. Asterisks indicate significant differences between the WT and OsABCG36 knockout lines (*P<0.05; Student’s t-test). (This figure is available in colour at JXB online.)

Discussion

ABC transporters constitute a large superfamily of proteins in bacteria, animals, and plants (Guidotti, 1996). ABC proteins are able to transport various substrates, such as hormones, lipids, glutathione conjugates, inorganic acids, xenomolecules, and heavy metals (Rea, 2007; Verrier ). There are more than 120 ABC members in both Arabidopsis and rice, which are classified into seven subfamilies (ABCA–ABCG) (Garcia ). Several ABC transporters, including AtABCC1, AtABCC2, AtABCC3, and AtABCG36, have been reported to play a role in Cd tolerance by Cd extrusion or vacuolar sequestration (Kim ; Park ; Brunetti ). However, the exact functions of most ABC transporters in plants are still unknown. In the present study, we functionally characterized an ABCG-type transporter gene, OsABCG36, which encodes a plasma-membrane-localized protein (Fig. 3). OsABCG36 is expressed in both roots and shoots, but only expression in the roots was rapidly and greatly up-regulated in response to Cd (Fig. 1A, D). Knockout of OsABCG36 resulted in decreased tolerance to Cd, but did not alter tolerance to other metals, including Zn, Cu, Al, and Pb (Fig. 5F; Supplementary Fig. S7). Furthermore, loss of function of OsABCG36 resulted in greater Cd accumulation in the roots and root cell sap (Fig. 5G). Although the cell sap data alone could not be used for a quantitative analysis of symplastic versus apoplastic Cd, because weakly bound Cd or Cd conjugates released from the cytosol and vacuoles may be able to make contact with the cell walls and partially bind there after the root cells are lysed by the freeze–thaw procedure, the difference was also found in the total Cd concentration of the roots (Fig. 5H). Given that OsABCG36 showed an efflux transport activity for Cd in yeast (Fig. 4B), the higher Cd accumulation in the root cell sap of the mutants is the result of loss of Cd efflux from the root cells. This may be associated with enhanced Cd toxicity in the mutants (Fig. 5D, F, 6B). The transport substrate of OsABCG36 remains to be identified in the future, but it could be ionic Cd or Cd conjugates.

Similar to most full-size G-type ABC transporters, OsABCG36 consists of two nucleotide-binding domains and two ABC2 transmembrane domains (Supplementary Fig. S1B). OsABCG36 is not the closest homolog of AtABCG36 (Supplementary Fig. S1C), but OsABCG36 functions similarly to AtABCG36 (Kim ). Both of these proteins are efflux transporters of Cd or Cd conjugates that contribute to Cd tolerance in plants. However, OsABCG36 showed some differences from AtABCG36 in terms of expression pattern and cellular localization. For example, in the absence of Cd, the expression level of OsABCG36 was low in the root and shoot (Fig. 1A), whereas that of AtABCG36 was high (Kim ). In the presence of Cd, the expression of OsABCG36 is rapidly and greatly induced only in the roots (Fig. 1A). Furthermore, the expression level increased with increasing external Cd concentrations (Fig. 1B), whereas that of AtABCG36 was slightly enhanced by Cd in both the roots and shoots of Arabidopsis (Kim ). Under normal conditions, OsABCG36 was expressed in the root cells at a low level, whereas AtABCG36 was strongly expressed in the root hair and epidermis. These differences suggest that some functions of OsABCG36 and AtABCG36 diverged during evolution. Additionally, AtABCG36 has also been suggested to play a role in Pb tolerance (Kim ). However, we found that knockout of OsABCG36 did not affect tolerance to other metals, including Pb (Fig. 5F; Supplementary Fig. S7). Similar differences between rice and Arabidopsis have also been reported for other ABC transporters. For example, AtABCC1/2 is involved in both Cd and As tolerance in Arabidopsis (Song ; Park ), whereas OsABCC1 is involved only in As tolerance and not Cd tolerance (Song ).

Detoxification of toxic metals can be achieved by a variety of mechanisms, including compartmentalization (vacuolar sequestration), chelation by metal ligands such as phytochelatins and metallothioneins, and efflux of toxic metals from the cells (Hall, 2002). The present work provides evidence that OsABCG36 is involved in transporting Cd out of the root cells (i.e. efflux). However, the effect of knockout of OsABCG36 on Cd tolerance was observed only at relatively high Cd concentrations (Fig. 5D–F, 6B). Furthermore, root elongation in the mutants was not completely inhibited in the presence of Cd (Fig. 5D–F, 6B). These results suggest that OsABCG36-mediated efflux of Cd is one of the components of Cd tolerance in rice and that its contribution is not very large.

Although knockout of OsABCG36 decreased tolerance to Cd (Fig. 5D–F, 6B), it did not affect Cd accumulation in the shoots (Fig. 6F), suggesting that OsABCG36 is not involved in Cd accumulation. In rice, several types of transporters have been shown to be involved in the uptake and distribution of Cd (Clemens and Ma, 2016; Wang ). OsNramp5, a Nramp (natural resistance-associated macrophage protein) member, was identified as a major pathway for Cd uptake in rice (Ishimaru ; Sasaki ). Two P-type ATPase transporters, OsHMA3 and OsHMA2, have been reported to be responsible for the vacuolar sequestration of Cd in the root cell and root-to-shoot translocation of Cd, respectively (Ueno ; Miyadate ; Satoh-Nagasawa ; Takahashi ; Yamaji ; Shao ). At the reproductive stage, OsLCT1, a low-affinity cation transporter1, was proposed to transport Cd into rice grains (Uraguchi et al., 2011, 2014). OsHMA2 expressed in the rice nodes is also involved in Cd distribution to the grain (Yamaji ). Among these transporters, only OsHMA3 was demonstrated to play an important role in Cd tolerance, through vacuolar sequestration of Cd in the roots (Sasaki ). Therefore, there are at least three pathways for Cd in the root cells taken up by OsNramp5; vacuolar sequestration mediated by OsHMA3, efflux from the root cells by OsABCG36, and root-to-shoot translocation by OsHMA2. The lack of difference in shoot Cd accumulation between the mutants and wild-type rice in this study suggests that the contribution of OsABCG36 to shoot Cd accumulation is very small. One possible explanation for this is the presence of redundant genes with similar functions to OsABCG36 in rice. Phylogenetic analysis showed that OsABCG36 was closely related to OsABCG32, OsABCG34, OsABCG35, OsABCG37, and OsABCG44, and shared high identity (76–85%) with them at the amino acid level (Supplementary Fig. S1C). Among these, OsABCG37 and OsABCG44 are expressed in the roots. Recently, a comparative RNA-seq-based analysis showed that OsABCG37 and OsABCG44 were up-regulated in roots by high-Cd stress (Tan ). These data suggested that OsABCG37 and OsABCG44 possibly play a similar function to OsABCG36 in roots although the exact roles of these genes in Cd tolerance and accumulation remain to be examined in future.

In conclusion, our results indicate that OsABCG36 localized at the plasma membrane is involved in Cd tolerance in rice through efflux of Cd out of the root cells.

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. Analysis of OsABCG36.

Fig. S2. Expression analysis of OsABCG36 and its homolog genes.

Fig. S3. Co-localization of OsABCG36 and AtTPK.

Fig. S4. Subcellular localization of OsABCG36 in yeast cells.

Fig. S5. The growth of zrt1cot1 transformed with empty vector pYES2 or OsABCG36.

Fig. S6. Root length of wild-type and two OsABCG36 knockout lines without Cd.

Fig. S7. Pb sensitivity of the wild-type and two OsABCG36 knockout lines.

Fig. S8. Growth of OsABCG36 knockout lines.

Fig. S9. Concentration of essential metals in the roots and shoots.

Fig. S10. Cd concentration in the xylem sap.

Click here for additional data file.

Author contributions

JFM and JX designed the experiments; SF performed most of the experiments; YL, XZ, GY, DC, ZW, MS, JC, DYC, and RL participated in the research; SF and JX analyzed the data; JX wrote the paper; JFM and JX revised the paper.