The Oryza sativa cultivars Nipponbare (NPB), Dongjing (DJ), and SSBM were used in this study. The rice T-DNA insertion seed pool used to isolate the osarm1 mutant (PFG_3A-12233.R) was obtained from Rice T-DNA Insertion Sequence Database (developed by Dr. Gynheung An, Department of Plant Systems Biotech, Kyung Hee University, Republic of Korea) (Jeon et al., 2000; Jeong et al., 2006). The T-DNA insertion site in the osarm1 mutant was identified using OsARM1 gene-specific primers XS830/XS831 paired with the T-DNA right border primer PGVRB. The primers used in this article are listed in Table S1. Generation of the OsARM1-KO mutants by CRISPR/Cas9 editing is described below.

All rice seeds were surface sterilized with 75% ethanol for 40 s, followed by 20% NaClO for 20 min. After being rinsed ~five times with sterile distilled water, the seeds were germinated on half-strength Murashige and Skoog (1/2 MS) medium at 25°C under a 12-h light/12-h dark photoperiod.

Two-week-old rice seedlings were transferred to Kimura B nutrient solution with or without 2, 5, 25, or 40 μM As(III) for arsenite tolerance analysis. After 7 or 14 days of treatment, the phenotypes were recorded by measuring the root lengths and heights of the seedlings.

For As(III) treatment of Arabidopsis, seeds of wild-type (Col-0) and transgenic plants expressing OsARM1-GFP were germinated on 1/2 MS medium for 1 week under normal growth conditions. The seedlings were subsequently transferred to 1/2 MS medium or 1/2 MS medium containing 10 or 20 μM As(III) for further growth. Photographs were taken and dry weights were recorded at 2 weeks after treatment.

The expression level of OsARM1 was tested in total RNA extracted from various organs (roots, stems, leaf blades, panicles, and grains) of 14-week-old soil-grown plants with the TRIzol RNA extraction kit (Invitrogen, Thermo Fisher Scientific) following the manufacturer's instructions and RT-PCR using the primers XS147 and XS148 with OsACTIN1 (XS1043 and XS1044) as a reference gene.

To test the expression of OsARM1 and transporter genes in response to As, total RNA was extracted from various tissues of 2-week-old seedlings. The isolated RNA was reverse transcribed to obtain first-strand cDNA using PrimeScript RT reagent with gDNA eraser kit (Takara). The qRT-PCR was performed using gene-specific primers (Table S1) and SYBR Premix Ex Taq II (Takara) on a StepOne Plus real-time PCR system (Applied Biosystems). OsACTIN1 (Wang et al., 2012) or OsGAPDH (Pabuayon et al., 2016) was used as an internal reference.

To generate the OsARM1pro::GUS plasmid, about 1.5 kb of sequence in the promoter region of OsARM1, including 19 bp downstream of the start codon (ATG) of OsARM1 was amplified from rice genomic DNA by high-fidelity PCR using primer pairs XS489 and XS490 (Table S1). This fragment contained 1.5 kb of the promoter region of OsARM1, 19 bp from the first exon of OsARM1, and 8 bp from the vector, which was translationally in-frame with the GUS fusion. This PCR fragment was then cloned into the XcmI site of the pCXGUS-P vector (Chen et al., 2009). The insert was validated by sequencing and the confirmed plasmid was introduced into Agrobacterium tumefaciens strain LBA4404 and transformed into the wild-type rice cultivar SSBM using a rapid and efficient Agrobacterium-mediated transformation method (Toki, 1997). Putative transformants selected on 1/2 MS medium containing hygromycin (50 mg·L⁻¹) were verified by PCR using a promoter-specific forward primer and the GUS-specific reverse primer GUS-SEQ (Table S1). Subsequently, the T₃ homozygous lines were sown on 1/2 MS medium containing hygromycin and the resistant plants were used for further analysis.

Homozygous transgenic lines expressing OsARM1_pro::GUS were stained to detect GUS activity as described previously (Zheng et al., 2012; Chen et al., 2013). The samples were immersed in GUS staining solution [50 mM K₄Fe(CN)₆·3H₂O, 50 mM K₃[Fe(CN)₆], 0.2 M NaH₂PO₄·2 H₂O, 0.2 M Na₂HPO₄·12 H₂O, 10% Triton X-100, 100 mg·mL⁻¹ X-Gluc] and vacuum infiltrated for 20 min, followed by 8 h to overnight incubation at 37°C, depending on the desired staining intensity. After staining, the tissues were rinsed several times with 75% ethanol until the chlorophyll was removed. GUS-stained seedling tissues were observed under a fluorescence stereo-microscope (SteREO Lumar. V12). The tissues were then cleared with HCG solution (chloral hydrate/water/glycerol 8:3:1) and observed and photographed under a fluorescence microscope (Leica DM5000B).

The full-length OsARM1 cDNA was amplified using primer pairs XS260 and XS261 (Table S1) and fused to the N terminus of green fluorescent protein (GFP) in the pBI-eGFP vector (Xiao et al., 2008). For the transient expression analyses, the empty vector (which expresses GFP) and the OsARM1-GFP construct were transiently expressed in rice protoplasts isolated from 8-day-old green seedlings as described previously (Zhang et al., 2011). The AtARF4-RFP plasmid (Piya et al., 2014) was used as a nuclear marker for co-expression with GFP or OsARM1-GFP. To generate stable transgenic lines, the OsARM1-GFP construct was introduced into A. tumefaciens LBA4404 and transformed into wild-type Arabidopsis Col-0 by the floral dip method (Clough and Bent, 1998). To further confirm the localization of OsARM1-GFP in the nuclei, the leaves of 1-week-old transgenic Arabidopsis expressing OsARM1-GFP and the GFP control were infiltrated with phosphate-buffered saline (PBS, pH 7.4) containing 2 ng·μL⁻¹ 4′, 6′-diamidino-2-phenylindole (DAPI) for 10 min and washed several times with PBS (Wang et al., 2016). The fluorescence was detected by confocal laser scanning microscopy (Zeiss 7 DUO NLO).

To generate OsARM1-KO mutants by CRISPR/Cas9, 19, or 20 bp of conserved sequence from the OsARM1 regions encoding the MYB-type HTH DNA-binding domains were selected as targets (Table S1). The sequences were ligated to the pYLCRISPR/gRNA vector, followed by ligation to pYLCRISPR/Cas9-MTmono vector after dual-nested PCR as previously described (Ma et al., 2015). The plasmids were introduced into A. tumefaciens strain LBA4404 and transformed into wild-type rice NPB using the traditional rice transformation method (Toki, 1997). The genome editing of OsARM1-KO transgenic rice lines was confirmed by sequencing PCR products amplified with primer pairs M4T1-F/M4T1-R and M4T2-F/M4T2-R (Table S1).

To generate OsARM1 overexpression (OsARM1-OE) lines, the entire cDNA sequence of OsARM1 was amplified using primer pairs XS491 and XS492 and cloned into the XcmI site of the binary vector pCXSN-Myc (Chen et al., 2009). The plasmids (constructed as described above) were also introduced into A. tumefaciens strain LBA4404 and transformed into wild-type rice NPB using the traditional rice transformation method (Toki, 1997). The expression levels of OsARM1 in the OsARM1-OE transgenic lines were determined by RT-PCR analyses.

Two-week-old NPB, OsARM1-KOs, and OsARM1-OEs seedlings were treated with Kimura B solution contained 2 or 25 μM As(III) for 7 days, and root and shoot samples were collected separately. To remove the adsorbed As from the root surface, the roots were washed three times with distilled water. Then the roots and shoots were collected separately and dried at 65°C for 2 days. To determine the As levels in various aboveground organs, NPB, OsARM1-KOs, and OsARM1-OEs rice plants were grown in As-containing soil (20 mg As per 1 kg soil) until the maturity. Samples were harvested from various organs as previously described (Yamaji and Ma, 2014) and dried at 65°C for 2 days. After digestion with 10 mL of 4:1 HNO₃/HClO₄ at 180°C for 8 h (Chen et al., 2005), the total As contents were determined using atomic fluorescence spectrometry (AFS-8220).

ChIP assays were performed as described previously (Yamaguchi et al., 2014) using 2-week-old OsARM1-OE transgenic rice seedlings and Arabidopsis seedlings expressing the OsARM1-GFP fusion protein (GFP-9). After coating with anti-Myc (Abiocode) or anti-GFP (Abiocode) antibodies (Cell Signaling Technology), the protein/DNA complexes were immunoprecipitated with Dynabeads Protein G (Invitrogen) for at least 4 h at 4°C. The precipitated DNA was purified using a DNA purification kit (Qiagen), and the enriched DNA fragments were subjected to qPCR using the specific primers listed in Table S1. The OsACTIN1 and AtACTIN2 promoters were used as negative controls. All ChIP assays were repeated three times with similar results.

The expression pattern of OsARM1 was first investigated by quantitative reverse-transcription PCR (qRT-PCR) analysis, in different organs of 14-week-old wild-type rice (O. sativa L. japonica. cv. Nipponbare, NPB) grown in soils. The OsARM1 transcript was widely distributed in all organs with high levels in the rachis and spikelets (Figure 1A).

To investigate the potential role of OsARM1 in the plant response to As(III) stress, we next tested whether As could induce OsARM1 expression. To this end, we subjected 2-week-old wild-type rice seedlings (NPB) to treatment with 50 μM As(III) and investigated the expression patterns of OsARM1 by qRT-PCR in various tissues including leaves, stems, basal regions, and roots. OsARM1 expression was rapidly induced after As(III) treatment at various time points, peaking at 1-h of As(III) exposure (Figure 1B). In particular, OsARM1 transcript levels increased 37.1-, 88.8-, 71.9-, and 111.3-fold in As(III)-treated leaves, stems, basal regions, and roots, respectively, compared with the untreated controls (Figure 1B). The As(III)-inducible expression of OsARM1 is consistent with our previous findings (Yu et al., 2012). In addition, OsARM1 expression levels were significantly reduced after 3–48 h of As(III) treatment (Figure 1B).

To further elucidate the expression pattern of OsARM1, we generated transgenic rice plants expressing OsARM1 promoter fusions with the β-glucuronidase (GUS) reporter (OsARM1_pro::GUS). Histochemical GUS staining showed that, under normal growth conditions, OsARM1_pro::GUS was expressed primarily in the enlarged vascular bundles of node I and node II, the vessel tissues of the husk, and the anther at the reproductive growth stage (Figure 2A, upper images). However, OsARM1_pro::GUS expression was considerably lower in leaves, stems, basal regions, and roots at the vegetative growth stage (Figure 2B, panels a–d). Given that the expression of OsARM1pro::GUS is extremely low under normal growth conditions (Figure 2B, panels a–d), we investigated the induction of OsARM1 promoter under As(III) stress (Figure 2B, panels e–h). After exposure to 50 μM As(III) for 6 h, OsARM1_pro::GUS expression was activated in the above-mentioned tissues (Figure 2B, panels e–h). On the other hands, no GUS signals in the tested tissues of control samples were observed under normal conditions (Figure 2A, lower images and Figure 2B, panels i–l) or under As(III) stress conditions (Figure 2B, panels m–p).

We further investigated the cell specificity of OsARM1 expression in roots, leaves, and stems by observing tissue slices by fluorescence microscopy. GUS activity was detected in the xylem transfer cells of roots (Figure S1A), the vessels of leaves (Figures S1B,D) and the tracheids of stems (Figure S1C). Together, our results reveal that OsARM1 expression is induced by As(III) treatment and that OsARM1 transcript accumulates specifically in vascular tissues.

To determine the subcellular localization of OsARM1, we cloned OsARM1 in the pBI-eGFP vector (Xiao et al., 2008) fused to the N-terminus of the gene encoding enhanced green fluorescent protein (eGFP). We then transiently co-expressed the empty vector pBI-eGFP (GFP) or the construct expressing OsARM1-GFP together with the nuclear marker AtARF4-RFP in isolated rice protoplasts and detected the fluorescence by confocal laser scanning microscopy. In the rice protoplasts expressing just GFP, the green fluorescent signals were observed throughout the cell; by contrast, the green fluorescent signals of OsARM1-GFP were primarily detected in the nucleus, with weak signals in the cytosol (Figure 3A). The nuclear localization of OsARM1-GFP was further confirmed its co-localization with the red fluorescent signal of the nuclear marker AtARF4-RFP (Figure 3A; Piya et al., 2014). To confirm this observation, we also transformed the OsARM1-GFP construct into wild-type Arabidopsis (Col-0) to generate transgenic lines. In contrast to transgenic lines expressing GFP from the empty vector, the OsARM1-GFP fusion protein predominately localized to the nucleus in the Arabidopsis leaf epidermal cells (Figure 3B). To further confirm this observation, we stained leaf cells of OsARM1-GFP plants with 4′,6-diamidino-2-phenylindole (DAPI) and found that OsARM1-GFP co-localized with DAPI, which confirmed the nuclear localization of OsARM1 in Arabidopsis cells (Figure 3B; bottom images).

To assess the role of OsARM1 in the As(III) response, we identified a rice T-DNA insertion mutant, osarm1 (PFG_3A-12233.R), from the Rice T-DNA Insertion Sequence Database (Jeon et al., 2000; Jeong et al., 2006). Genotyping and sequence analyses demonstrated that a T-DNA insertion is located in the promoter region (-449) of OsARM1 in this mutant (Figures S2A,B). RT-PCR analysis revealed that the expression of OsARM1 was clearly reduced in the osarm1 homozygous mutant in various tissues compared to the wild-type control (Figure S2C). When the seedlings were exposed to 40 μM As(III) for 7 days, the root length and plant height of DJ were significantly reduced (Figures 4A–C). By contrast, the osarm1 mutant showed increased tolerance to As(III) stress, including increased plant height and root length, compared to the wild-type control (Figure 4A; Figure S10). The enhanced As(III) tolerance of osarm1 mutants was further confirmed by calculating the relative root elongation and plant height of rice plants before and after As(III) treatment (Figures 4B,C).

Phenotypically, young osarm1 seedlings showed no significant morphological changes; however, mature osarm1 plants were semi-dwarf and partially sterile compared to wild-type (DJ) plants grown under the same conditions (Figures S3A,B). As a result, the seed setting rate of osarm1 was only approximately 30% that of DJ (Figure S3C). To test whether the growth inhibition and reduced fertility of the osarm1 mutant during late development resulted from the lesion in OsARM1 (Figure S3), we used CRISPR-Cas9 to generate rice lines with knockout mutations of OsARM1 (designated OsARM1-KOs). PCR and sequencing identified two alleles: a 1-bp insertion in target I and a 1-bp deletion in target II, among the transgenic lines (Figure S4). Such mutations would cause a frameshift in the ORF, leading to early termination of translation of OsARM1 protein. Further, the OsARM1 protein in the OsARM1-KO mutants was potentially to lack the intact structure and normal transactivation activity. The growth and fertility of both OsARM1-KO lines (KO-1 and KO-2) were not obviously different from wild type (Figure 4D; Figures S3D–F), indicating that these phenotypes of osarm1 likely result from a separate mutation, possibly caused by an unlinked T-DNA insertion. However, when 2-week-old plants were exposed to 40 μM As(III) for 7 days, OsARM1-KOs were more resistant to As(III) that the wild-type (NPB) control (Figure 4D). This phenotype is consistent with that of the osarm1 T-DNA insertion mutant (Figure 4A). We calculated the relative root length and plant height after As(III) treatment, finding that the values for the OsARM1-KO plants were significantly higher than those of wild type (Figures 4E,F). In addition, when we reduced the concentration of As(III) to 2 and 5 μM, we observed no obvious inhibition of the root growth of osarm1 and OsARM1-KO plants. However, the relative root lengths of wild type DJ and NPB were shorter than those of osarm1 and OsARM1-KO plants (Figures S5A,B,D,E). Under such low As(III) concentrations, the relative heights of wild-type plants and the OsARM1 mutants showed no significant differences (Figures S5C,F).

To further evaluate the role of OsARM1 in plant As(III) tolerance, we generated transgenic lines overexpressing OsARM1 via Agrobacterium-mediated transformation of wild-type plants (NPB). To this end, we cloned OsARM1 into the binary vector pCXSN-Myc, under the control of the 35S promoter. RT-PCR analyses showed that OsARM1 was overexpressed in the three independent OsARM1-OE T₀ transgenic lines obtained (Figure 5A).

As shown in Figure 5, the shoot and root growth in lines OE-2 and OE-3 were not significantly different from that of wild type under normal growth conditions. After treatment with 25 μM As(III) for 7 days, the growth of both OE lines (OE-2 and OE-3) lagged behind that of wild-type (NPB) seedlings (Figure 5B), indicating their increased sensitivity to As(III)-induced stress compared to NPB. Statistical analysis (Figure 5C) revealed that the root lengths of the two OE lines (OE-2 and OE-3) grown on As(III)-containing liquid medium were 73.7 ± 7.0 and 78.6 ± 5.8%, respectively, of the lengths of OE roots not treated with As, and were significantly shorter than the NPB roots on As(III) medium (90.7 ± 11.1%). In the presence of 25 μM As(III) on day 7, the relative shoot heights of both OE lines (OE-2 and OE-3) were significantly reduced compared to that of NPB (Figure 5D). Similarly, the root growth of the OsARM1-OE lines was suppressed by growth on 2 and 5 μM As(III) for 14 days (Figure S6). These findings imply that OsARM1 is involved in the regulation of the response to As(III) stress in rice.

The increased and attenuated As(III)-tolerant phenotypes in OsARM1-KOs and OsARM1-OEs, respectively, may be due to altered uptake or translocation of As. To investigate this, we measured the As contents in wild-type (NPB), OsARM1-KOs (KO-1 and KO-2), and OsARM1-OEs (OE-2 and OE-3) plants. After exposure to 2 μM As(III) for 7 days, the root and shoot samples were harvested separately for As measurements. As shown in Figure 6A, even though the OsARM1-OEs and NPB showed no significant differences in As contents of their roots, the As levels in roots of OsARM1-KOs were lower than the wild-type NPB, and the As levels in shoots of OsARM1-KOs were higher than NPB, while the As content in shoots of OsARM1-OEs were a little lower than NPB. Moreover, in plants grown on 25 μM As(III) for 7 days, the roots of OE-2 and OE-3 had As levels that were 1.5- and 1.6-fold higher, respectively, than that of wild type. In contrast to the results in roots, the As contents in OsARM1-OE shoots were lower than in NPB and As contents in OsARM1-KO shoots (KO-1 and KO-2) were 1.7- and 1.9-fold higher, respectively, than that of NPB. No significant differences of As contents were detected in root tissues between the OsARM1-KO lines and NPB (Figure 6B), but the absolute values of As concentrations were lower in the OsARM1-KOs (317.3 ± 16.4 and 323.4 ± 15.7 mg/kg DW) in the roots in comparison with wild type (334.1 ± 12.7 mg/kg DW). These results indicate that OsARM1 is likely involved in regulating uptake and root-to-shoot translocation of As in rice.

To assess the distribution of As in shoots, we measured the As concentrations in various organs above node II in NPB, OsARM1-KOs, and OsARM1-OEs plants. All plants were grown in 20 mg As per kg soil until the ripening stage. Compared with wild-type NPB, the OsARM1-KO lines contained significantly higher levels of As in node I (Figure 6C). By contrast, the As contents were reduced in nodes I and II of the OsARM1-OE lines compared to wild type (Figure 6C). As a result, the OsARM1-OEs showed slightly reduced levels of As in the peduncle, rachis, and husk (Figure 6C). Taken together, the increased and reduced accumulation of As in node I of OsARM1-KOs and OsARM1-OEs, respectively, relative to wild type suggest that OsARM1 may function in regulating the expression of key arsenic transporter genes in this tissue, a vital site for controlling As translocation to the grain.

To identify differentially expressed genes involved in As translocation under As(III) stress, we extracted total RNA from roots of 2-week-old wild-type NPB, OsARM1-KO, and OsARM1-OE seedlings under normal conditions or seedlings treated for 6 h with Kimura B nutrient solution containing 50 μM As(III). We analyzed the expression of representative As-responsive transporter genes, OsLsi1, OsLsi2, and OsLsi6, by qRT-PCR. As expected, under As(III) stress, the transcripts of OsLsi1, OsLsi2, and OsLsi6 were upregulated in OsARM1-KO lines but downregulated in OsARM1-OE lines compared with wild-type (NPB) plants (Figure 7A).

The altered expression pattern of OsLsi1, OsLsi2, and OsLsi6 in the OsARM1-KO and OsARM1-OE lines suggested that OsARM1 regulates the plant response to As(III) by modulating the transcript levels of these transporters. To investigate the direct regulation of these As transporter genes by the OsARM1 transcription factor, we used the OsARM1-OE lines (which express an ARM1-Myc fusion protein) for further chromatin immunoprecipitation-quantitative PCR (ChIP–qPCR) analysis. Given that R2R3-MYB transcription factors bind to the AC-I element [ACC(A/T)A(A/C)] in the promoters and genomes of target genes to directly regulate their transcription (Zhong and Ye, 2012), we first analyzed the promoter sequences of OsLsi1, OsLsi2, and OsLsi6. Five (P1, P2, P3, P4, and P5) AC-I elements were detected in the OsLsi1, OsLsi2, and OsLsi6 sequences (Figure 7B). ChIP-qPCR data revealed that for each OsLsi gene, only one DNA fragment, P5 from OsLsi1, P3 from OsLsi2, and P1 from OsLsi6, was enriched in the DNA immunoprecipitates produced using anti-Myc antibody (Figure 7C). To further determine the interaction between OsARM1 and As transporter genes, we took advantage of the two available transgenic Arabidopsis lines expressing OsARM1-GFP, GFP-2, and GFP-9 (Figure S7A). Phenotypic analyses (Figures S7B,C) showed that both transgenic lines were highly sensitive to As(III) treatment compared to wild type (Col-0), with responses resembling that of the OsARM1-OE lines (Figure 5), suggesting that OsARM1-GFP is functional in Arabidopsis.

We also analyzed the expression of AtNIP1;1, AtNIP3;1, and AtNIP5;1, three well-known Arabidopsis genes homologous to aquaporin, under As(III) stress (Ali et al., 2009; Kamiya et al., 2009; Xu et al., 2015). The expression of AtNIP3;1 and AtNIP5;1 declined after treatment with 40 μM As(III) for 6 h and their expression levels were higher in GFP-9 than in wild-type Col-0. At the same time, the expression of AtNIP1;1 increased in GFP-9 after As(III) treatment (Figure S8A). Promoter sequence analysis identified three AC-I elements (P1, P2, and P3) in the promoter regions of AtNIP1;1, AtNIP3;1, and AtNIP5;1 (Figure S8B). ChIP-qPCR data showed that one or two DNA fragments from each promoter, i.e., P1 in AtNIP1;1, P3 in AtNIP3;1, and P1 and P2 in AtNIP5;1, were enriched in the DNA immunoprecipitates produced using anti-GFP antibody (Figure S8C). These results confirm that OsARM1 can directly bind to the promoter regions of OsLsi1, OsLsi2, and OsLsi6 in rice as well as AtNIP1;1, AtNIP3;1, and AtNIP5;1 in Arabidopsis.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.