Identification and characterization of low-sulfur-inducible genes and pathways in roots of upland and lowland rice grown under hydroponic and soil conditions Authors: Kyonoshin Maruyama, Yasuhiro Tsujimoto, Miho Kishimoto, Kyouko Yoshiwara, Hiroaki Sakai, Tetsuya Sakurai, Shingo Sakamoto, Nobutaka Mitsuda, Akiko Maruyama-Nakashita, Stanislav Kopriva Published: 2026 ABSTRACT Under low-sulfur conditions in both upland and lowland rice, EIL4-mediated gene regulation plays a key role, with SULTR1;1 contributing to increased root elongation.Keywords: cis-acting promoter elements, EIL4, genome-wide analysis, low-sulfur response, lowland rice, root elongation, SULTR1, 1, transcriptome, upland rice Introduction Sulfur plays a crucial role in living organisms, contributing to the synthesis of essential biomolecules such as amino acids, proteins, and antioxidants (Takahashi et al., 2011; Kopriva et al., 2019). In plants, inorganic sulfur is taken up from the soil as sulfate (SO42–) through the roots and subsequently transported via vascular tissues to other organs, where it is assimilated into cysteine in plastids. Low-sulfur (LS) conditions have become increasingly prevalent worldwide, including in temperate agricultural regions, mainly due to reduced atmospheric sulfur deposition and the decreased use of sulfur-containing fertilizers (Scherer, 2009; Feinberg et al., 2021). LS soils are also common in tropical regions with sandy, low-organic-carbon soils and in the iron-rich, highly weathered landscapes of West Africa. In the Volta River Basin of Ghana, sulfur fertilization has been shown to significantly increase rice yields (Tsujimoto et al., 2013, 2017). Sulfur assimilation involves multiple sulfate transporters that mediate the uptake and internal translocation of SO42– in plants. These transporters are expressed not only in the root epidermis and cortical cells but also in vascular tissues, enabling coordinated sulfate distribution within the plant (Yoshimoto et al., 2007; Takahashi et al., 2011). Once inside, sulfur is assimilated via ATP sulfurylase, adenosine 5′-phosphosulfate (APS) reductase, and sulfite reductase to produce cysteine, a precursor for essential biomolecules such as glutathione and thioredoxin that function in redox regulation and stress responses (Davidian and Kopriva, 2010; Kopriva et al., 2019). In Arabidopsis, genes encoding sulfate transporters and APS reductase, which catalyze the rate-limiting steps in sulfur assimilation, show elevated transcript levels under LS conditions (Takahashi et al., 1997). However, while the LS response has been well characterized in Arabidopsis, its molecular basis in rice remains less understood. Microarray and RNA sequencing analyses have revealed that many LS-responsive genes encode transporters, metabolic enzymes, and detoxification enzymes in Arabidopsis, Chlamydomonas, and tomato (Hirai et al., 2003; Maruyama-Nakashita et al., 2003; Nikiforova et al., 2003; González-Ballester et al., 2010; Canales et al., 2020). Sulfur limitation 1 (SLIM1)/Ethylene-Insensitive 3-like 3 (EIL3), a member of the Arabidopsis EIL transcription factor family, is responsible for the induction of LS-responsive genes such as SULTR1;1 in roots (Maruyama-Nakashita et al., 2006). Although EIL transcription factors have been studied in rice in relation to ethylene signaling and abiotic stress responses (Guo and Ecker, 2004; Dietzen et al., 2020), their roles in LS response remain largely unexplored. In Arabidopsis, several cis-acting promoter elements mediating transcriptional responses to LS conditions have been identified. The auxin response element GAGACA and its reverse complement TGTCTC have been identified as a sulfur-responsive element of SULTR1;1 (SURE11) involved in the LS response in Arabidopsis (Maruyama-Nakashita et al., 2005). The UPE-box consensus sequence (AGRTWCATTGAAYCTRGACR; R=A/G, W=A/T, Y=C/T), found in the promoters of LSU (response to low sulfur) genes, is transcriptionally activated by SLIM1/EIL3 (Wawrzyńska et al., 2010; Wawrzyńska and Sirko, 2016). More recently, the SLIM1/EIL3-binding sequence (SEBS; YGTCYAGRTNCW; N=A/C/G/T) and its reverse complement (WGNAYCTRGACR) were identified through DNA affinity purification sequencing (DAP-seq) (O’Malley et al., 2016). Notably, SEBS is located at the 3′ end of the UPE-box consensus sequence. Despite these findings, the specific cis-acting promoter elements involved in the LS response in rice remain unclear. To date, rice research has focused primarily on Nipponbare (NB), a lowland rice variety whose genome has been fully sequenced. However, upland rice varieties exhibit distinct physiological traits that enable adaptation to water-limited, nutrient-deficient soils. DJ123, an upland rice variety from Bangladesh, is renowned for its resilience and adaptability to challenging growing conditions. It has been widely studied for its genetic contributions to rice improvement, particularly in enhancing tolerance to low-phosphorus soils and boosting yield potential (Nestler et al., 2016). Upland rice varieties, including DJ123, are thought to possess more efficient water and mineral uptake systems than lowland rice, although the underlying mechanisms remain poorly understood. Such physiological traits may also influence sulfur uptake and assimilation under LS conditions. Because LS soils constrain rice productivity in the Volta River Basin of Ghana, we used DJ123—an upland rice variety adapted to nutrient-poor soils elsewhere—as a model to investigate LS responses; its agronomic suitability for this region remains to be determined. Therefore, we established a genome and transcriptome analysis platform for DJ123 to elucidate the molecular basis of LS responses in both upland and lowland rice. Using these platforms, we comprehensively identified LS-responsive genes in NB and DJ123 grown under hydroponic and soil conditions, and examined conserved promoter sequences of representative LS-inducible genes. We further analyzed the function of the LS-inducible transcription factor EIL4, a homolog of SLIM1/EIL3, and generated transgenic rice plants overexpressing SULTR1;1 to evaluate their molecular responses and potential utility for breeding in LS soils, particularly in the Volta River Basin of Ghana. Materials and methods Plant materials and growth conditions Two Oryza sativa L. cultivars, NB (lowland variety) and DJ123 (upland variety), were used in this study. A chimeric construct was generated by placing the ORF of SULTR1;1 from DJ123 under the control of the maize ubiquitin promoter in the sense orientation. The construct was introduced into NB via Agrobacterium tumefaciens-mediated transformation, and transgenic lines overexpressing SULTR1;1 (hereafter referred to as Stox) were selected for further analysis. In addition, a transgenic control line (T-cont) was generated by introducing the empty binary vector into NB using the same transformation procedure. This line was used to distinguish the effects of SULTR1;1 overexpression from those of the transformation process itself. LS soil was collected from the Volta River Basin in Ghana and sieved through a 2 mm mesh. The physicochemical properties of the soil were determined as summarized in Supplementary Table S1. The soil had a pH of 6.1, a cation exchange capacity (CEC) of 11.2 cmol(+) kg−1, mineralizable N of 49.6 mg kg−1, available P of 6.2 mg kg−1, and exchangeable S of 4.7 mg kg−1. Exchangeable cations were quantified as follows: Ca2+, 2.27 cmol kg−1; Mg2+, 0.97 cmol kg−1; K+, 0.15 cmol kg−1; and Na+, 0.20 cmol kg−1. Total C and total N were measured using an NC analyzer (Sumigraph NC-220F, SCAS, Japan). Exchangeable cations were extracted with 1 M NH4OAc (pH 7.0) and measured using an inductively coupled plasma emission spectrometer (ICPE-9000, Shimadzu, Japan). Mineralizable N was determined after a 4 week anaerobic incubation followed by extraction with 10% KCl. Available P was quantified using the Bray II method (Bray and Kurtz, 1945), and exchangeable S was analyzed using Ca(H2PO4)2 extraction containing 500 ppm P (Fox et al., 1964). CEC was determined using the ammonium acetate extraction method at pH 7.0. To equalize macronutrient availability other than sulfur, N, P2O5, and K2O were supplemented to both LS and control treatments at 0.1 g per pot using urea, sodium phosphate, and potassium chloride. For the control treatment, sulfur was supplied at 0.02 g per pot as sodium sulfate, whereas no sulfur fertilizer was added in the LS treatment. This fertilization regime, together with the moderate mineralizable N concentration of the soil, ensured that sulfur was the sole limiting nutrient under LS conditions. Plants were grown in 200 ml plastic pots under flooded conditions for 2 weeks in a growth chamber (LH-411PFQDT-SP, Nippon Medical & Chemical Instruments Co., Ltd) with a 13 h light (28 °C)/11 h dark (25 °C) cycle. Each experimental set included NB, DJ123, and Stox, with four biological replicates per treatment. For hydroponic culture, NB and DJ123 plants were grown under control (0.75 mM sulfur) or LS (0.01 mM sulfur) conditions in modified Yoshida solution (Yoshida et al., 1971). Sulfur-free conditions were established by replacing sulfur-containing salts (K2SO4, MgSO4, ZnSO4, and CuSO4) with their corresponding chloride forms (KCl, MgCl2, ZnCl2, and CuCl2). Sulfur concentration analysis At the end of the growing period, the rice shoots and roots were collected and oven-dried at 80 °C for >48 h, followed by determination of their biomass. Sulfur concentration was determined using an atomic emission spectrometer (ICPE-9000, Shimadzu) after samples were dry-ashed at 550 °C for 2 h and digested with 0.5 M HCl. The reproducibility of the analysis was evaluated using four biological replicates per experiment. Genome sequencing and assembly High-molecular-weight genomic DNA was size-selected using a BluePippin system (Sage Science, USA) and used for PacBio library preparation. Indexed libraries were sequenced on the Sequel platform (Pacific Biosciences, USA). Raw subreads were assembled de novo using Canu version 1.7 (Koren et al., 2017). Redundant contigs showing >98% sequence identity and coverage were removed based on BLASTN searches. The resulting contigs were polished using Quiver (Genomic Consensus version 2.3.2), followed by four iterative rounds of error correction with Pilon version 1.22 (Walker et al., 2014) using both long- and short-read sequences. Chromosome-level scaffolding Chromosome-level scaffolding was performed using RACA version 0.9.1.1 (Kim et al., 2013), with the IRGSP-1.0 genome sequence as the reference assembly. Gene prediction and annotation To support gene prediction, total RNA was isolated from DJ123 roots, stems, leaves, and panicles, and used for RNA sequencing. Libraries were prepared using the NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs, USA) and sequenced on a NextSeq 2000 platform (Illumina, USA) to generate 150 bp paired-end reads. Raw reads were trimmed using Trimmomatic version 0.39 (Bolger et al., 2014) and aligned to the NB reference genome (IRGSP-1.0) using HISAT2 version 2.1.0 (Kim et al., 2019). Gene prediction was performed by integrating ab initio, protein-based, and transcript-based methods. Ab initio gene models were generated using BRAKER version 2.1.0 (Brůna et al., 2021) with RNA-seq evidence and protein sequences from the UniProt plant division, together with reference annotations from the Rice Annotation Project Database (RAP-DB) and the Rice Genome Annotation Project version 7 (RGAP7). Protein-based gene prediction was conducted using GenomeThreader version 1.7.1 (Gremme et al., 2005). Transcript-based gene structures were predicted using both genome-guided and de novo assembly approaches with Trinity version 2.8.5 (Grabherr et al., 2011) and PASA version 2.3.3 (Haas et al., 2003; Rhind et al., 2011). ORFs were predicted using Trinotate version 3.2.0 (Bryant et al., 2017). The gene models obtained from the three approaches were merged using Evidence Modeler version 1.1.1 (Haas et al., 2008) and updated with PASA. Gene structures containing excessively long introns or misassembled regions were manually curated. Transposable elements (TEs) were annotated using Extensive de novo TE Annotator (EDTA) with default parameters (Ou et al., 2019). Completeness of the annotated gene set was assessed using BUSCO version 4 (Manni et al., 2021). Oligo-microarray and statistical analysis of low-sulfur-responsive genes Oligo-microarray analysis was performed as described previously (Maruyama et al., 2006, 2012, 2020), with minor modifications. A custom 44 K Agilent oligonucleotide microarray (Agilent Technologies, USA) designed for rice gene expression profiling was used. Total RNA was labeled using the Agilent Low RNA Input Linear Amplification/Labeling Kit and hybridized to the microarray slides. Four biological replicates were used for each treatment. Hybridized slides were scanned using an Agilent G2505C scanner, and raw signal intensities were extracted with Feature Extraction software version 10.10.1.1 (Agilent Technologies). Expression log ratios and adjusted P-values were calculated using GeneSpring GX version 14.9.1 (Agilent Technologies) with Benjamini–Hochberg false discovery rate (FDR) correction. Principal component analysis (PCA) was conducted using SIMCA version 13 (Umetrics, Sweden) based on log2-transformed expression ratios to visualize transcriptional variation among treatments (LS versus control), growth systems (hydroponic versus soil), and genotypes (NB versus DJ123). Reverse transcription–quantitative PCR analysis Reverse transcription–quantitative PCR (RT–qPCR) analysis was performed as described previously (Maruyama et al., 2020), with minor modifications. Total RNA was extracted using RNAiso (TaKaRa, Japan), and first-strand cDNA was synthesized using SuperScript III Reverse Transcriptase (Invitrogen, USA). RT–qPCR was conducted on a Mic qPCR Cycler (Bio Molecular Systems, Australia) using TB Green Premix Ex Taq II (TaKaRa, Japan). Four biological replicates were used for each treatment. Gene expression values were normalized using two reference genes, Os03G0784400/DJ.03G395000 and Os07G0501900/DJ.07G175100, which were identified in our transcriptome dataset as having highly stable expression under both LS and control conditions in NB and DJ123. All RT–qPCR values were corrected using the geometric mean of these two reference genes. Primer sequences for all target and reference genes are listed in Supplementary Table S2. Transient expression assay using rice protoplasts Transient expression analysis in rice protoplasts was performed as described previously (Kitomi et al., 2020; Sakamoto et al., 2020, 2022), with minor modifications. Etiolated seedlings grown in the dark for 7 d were cut into 1 cm segments and digested with cellulase solution to isolate protoplasts. The isolated protoplasts were transfected with effector, reporter, and reference plasmids in a 96-well plate and incubated for 16–18 h. Luciferase activity was measured using the Dual-Luciferase® Reporter Assay System (Promega, USA) with a multiplate reader (Infinite 200 PRO M Plex, Tecan, Switzerland). Transcriptional activation activity was determined using a dual-luciferase assay. Firefly luciferase (FLUC) activity from the reporter construct was normalized to Renilla luciferase (RLUC) activity derived from the internal control plasmid (FLUC/RLUC ratio). Fold induction was calculated by normalizing the mean FLUC/RLUC value of protoplasts transfected with the green fluorescent protein (GFP)-expressing construct (negative control), which was set to 1. Six independent biological replicates were analyzed for each treatment. TASUKE-based visualization of subspecies polymorphisms The full-length ORF of SULTR1;1 from NB was inspected using TASUKE (Kumagai et al., 2013) to visualize single nucleotide polymorphisms (SNPs) across rice subspecies. SNP information was obtained from the Rice Core Collection database (https://ricegenome-corecollection.dna.affrc.go.jp/), which includes representative accessions from Japonica, Tropical Japonica, Indica, and Aus groups. The NB SULTR1;1 ORF sequence was used as the reference, and the corresponding ORF regions of all available accessions in the database were aligned and compared. Subspecies-wide conservation and divergence at polymorphic sites were assessed based on the allele distribution visualized in TASUKE. Results Establishment of the genomic and transcriptomic analysis platform for DJ123 To enable genomic analysis of upland rice, we performed de novo sequencing of DJ123 using the PacBio platform. The assembly yielded 229 contigs with a total length of 385.75 Mbp and an N50 length of 6.89 Mbp. Among these contigs, 85 were aligned to the 12 chromosomes of the NB reference genome (International Rice Genome Sequencing Project, 2005), enabling construction of a pseudomolecule assembly spanning 377.54 Mbp (101.76% of the estimated genome size of 371 Mbp; Supplementary Table S3). Comparative genome analysis showed overall synteny between DJ123 and NB, but a large inversion (∼5 Mbp) was detected in the centromeric region of chromosome 6 (Fig. 1A, B). This inversion was validated by Sanger sequencing. Repetitive elements accounted for 48.60% (187.46 Mbp) of the DJ123 assembly, slightly higher than in NB (46.77%), with Gypsy-type long terminal repeat (LTR) retrotransposons (16.72%) and helitrons (11.67%) being the most abundant categories (Fig. 1C). Pseudomolecular assembly of DJ123. The pseudomolecular assembly of DJ123 was compared with the Nipponbare (NB) reference genome using a dot plot (A), revealing an inversion on chromosome 6 (B). The circled region in (A) indicates the region shown in (B). The distributions of LTR retrotransposons (Copia, Gypsy, and unknown) (upper tracks) and DNA transposons (Helitron, CACTA, Mutator, PIF_Harbinger, Tc1_Mariner, and hAT) (lower tracks) in DJ123 and NB are shown in (C). To establish a transcriptomic annotation platform, RNA sequencing was performed using roots, stems, leaves, and panicles of DJ123. A total of 40 831 protein-coding genes were predicted using integrated ab initio, transcript-based, and protein-based gene models, and unique locus identifiers were assigned. BUSCO analysis using the embryophyta_odb10 dataset indicated a completeness score of 97.52%. Among the predicted protein-coding genes, 6212 were identical at the amino acid sequence level to their corresponding genes in NB. Sulfur concentration in NB and DJ123 plants grown hydroponically and in soil under low-sulfur and control conditions To verify the effectiveness of the LS treatment, sulfur concentration was quantified in shoots and roots of NB and DJ123 plants grown under hydroponic and soil systems. Under LS conditions, sulfur concentrations were reduced to nearly 1/10th of those observed under control conditions (Tukey’s HSD, P<0.05) (Fig. 2), confirming that a strong sulfur-limited state had been successfully established. Because roots are the primary site of sulfate uptake from the environment, they serve as a critical tissue for characterizing transcriptional responses to sulfur limitation; therefore, transcriptomic analysis was conducted on root samples from NB and DJ123 grown under both hydroponic and soil systems to identify LS-responsive genes in upland (DJ123) and lowland (NB) rice varieties. Low-sulfur (LS) response of NB and DJ123 grown under hydroponic (HP) or soil conditions. After 2 weeks of growth, shoots and roots were collected for sulfur concentration analysis. Data represent means ±SDs of four biological replicates. Sulfur concentrations in plants grown under LS-HP and LS-Soil conditions were significantly lower than those under the corresponding Ctrl-HP and Ctrl-Soil conditions (Tukey’s HSD, P<0.05). Identification of low-sulfur-responsive genes in the roots of NB and DJ123 plants grown under hydroponic and soil conditions To identify LS-responsive genes in the roots of NB and DJ123 under hydroponic and soil conditions, we conducted transcriptomic analysis using our custom 44 K Agilent oligonucleotide microarray. In hydroponically grown NB, 388 genes were significantly up-regulated [FDR<0.05, fold change (FC) >2], whereas 478 were significantly down-regulated (FDR<0.05, FC<0.5) in response to LS (Fig. 3A; Supplementary Tables S4, S5). In soil-grown NB, LS treatment led to the up-regulation of 4664 genes and the down-regulation of 944 genes according to the same statistical criteria (Fig. 3B; Supplementary Tables S6, S7). Similarly, in hydroponically grown DJ123, 127 genes were significantly up-regulated and 325 were significantly down-regulated in response to LS (Fig. 3C; Supplementary Tables S8, S9). In soil-grown DJ123, 3194 genes were up-regulated, whereas 1041 were down-regulated (FDR<0.05, FC>2 or <0.5) (Fig. 3D; Supplementary Tables S10, S11). LS-responsive gene expression in NB and DJ123 roots under hydroponic and soil conditions. This figure compares LS-responsive genes in NB and DJ123 under different growth conditions. (A–D) Genes whose expression significantly increased or decreased in hydroponic- and soil-grown plants. (E–H) Correlation analyses of LS-responsive genes between NB and DJ123, as well as between hydroponic and soil conditions. Log2 FC values for gene expression are displayed in scatter plots, and the top 10 LS-inducible genes (a–j) were identified through correlation analysis. The expression levels of these genes are visualized in a heatmap (I), where each square indicates the relative FC as shown by the color scale. Gene symbols and annotations in (I) are based on the Rice Annotation Project Database (RAP-DB; https://rapdb.dna.affrc.go.jp/). Gene symbols (b), (e), and (g) were not available in the RAP-DB annotation. We compared LS-inducible genes in NB and DJ123 under both hydroponic and soil conditions. The genes showing the strongest induction under LS conditions were Os03G0165900 in NB and DJ.03G050100 in DJ123 (Fig. 3E–H). The genomic position of Os03G0165900 in NB corresponded to that of DJ.03G050100 in DJ123, and both genes encode Sulfur deficiency induced 1 (SDI1) (Aarabi et al., 2016). Furthermore, the top 10 most strongly LS-induced genes were largely shared between NB and DJ123 (Fig. 3E–H). These included genes encoding the sulfate transporter SULTR1;1 (Os03G0195800 and DJ.03G072500; Yoshimoto et al., 2007; Takahashi et al., 2011), the phosphoenolpyruvate/phosphate translocator PPT3 (Os01G0172100 and DJ.01G054400; Knappe et al., 2003), and an isoflavone reductase-like protein IRL (Os01G0106400 and DJ.01G004800; Kim et al., 2010), among others (Fig. 3I). Comparison of the number of low-sulfur-responsive genes in roots of NB and DJ123 plants grown under hydroponic and soil conditions We compared the number of LS-responsive genes in the roots of NB and DJ123 under hydroponic and soil conditions. In NB, 4858 genes showed increased expression under either LS hydroponic or LS soil conditions, with 194 up-regulated under both (Fig. 4A). In DJ123, 3222 genes were up-regulated under at least one of the two conditions, with 99 up-regulated under both (Fig. 4B). Conversely, 1288 genes in NB and 1223 in DJ123 showed decreased expression under at least one condition, with 134 and 143 genes, respectively, down-regulated under both (Fig. 4C, D). Under hydroponic conditions, 434 genes were up-regulated in either NB or DJ123, with 81 up-regulated in both (Fig. 4E). Under soil conditions, 6068 genes showed increased expression in either NB or DJ123, with 1790 up-regulated in both (Fig. 4F). Similarly, 646 genes under hydroponic conditions and 1464 under soil conditions were down-regulated in either NB or DJ123, with 157 and 521 genes, respectively, down-regulated in both (Fig. 4G, H). Additionally, 64 genes were consistently up-regulated in both NB and DJ123 under hydroponic and soil conditions (Fig. 4I; Supplementary Fig. S1), whereas 55 genes were consistently down-regulated (Fig. 4J; Supplementary Fig. S2). A total of 15 and 16 genes were specifically up-regulated in NB and DJ123, respectively (Fig. 4I), whereas 30 and 28 genes were specifically down-regulated in NB and DJ123, respectively (Fig. 4J). However, the expression levels of LS-responsive genes unique to NB or DJ123 were not highly elevated (Supplementary Fig. S3A–D). Comparison of LS-responsive gene counts in the roots of NB and DJ123. This figure compares the numbers of genes whose expression increased or decreased in response to LS in NB and DJ123 under hydroponic and soil conditions. (A–D) Within-variety comparisons. (E–H) Comparison of LS-responsive gene counts between NB and DJ123. (I and J) Summary of LS-induced gene expression changes common or specific to both conditions. Characterization of low-sulfur-responsive genes in roots of NB and DJ123 plants grown under hydroponic and soil conditions To further characterize the transcriptional responses to LS conditions in roots, a PCA was performed under both hydroponic and soil conditions. In Fig. 5A–D, PCA results comparing LS-responsive genes between NB and DJ123 are presented, whereas Fig. 5E–H shows within-variety comparisons, highlighting differences in LS responses between hydroponic and soil-grown plants for each genotype. Under hydroponic conditions, PC1 values for LS-treated samples were lower than those of control samples in both NB and DJ123 (Fig. 5A). The lowest PC1 loading corresponded to Os01G0268600/DJ.01G124000, followed by Os03G0195800/DJ.03G072500 (SULTR1;1) and Os03G0165900/DJ.03G050100 (SDI1) (Fig. 5B). Os01G0268600/DJ.01G124000 encodes a glycoside hydrolase family 31 protein. Similarly, under soil conditions, LS-treated samples in both NB and DJ123 showed lower PC1 values than their respective controls (Fig. 5C). The lowest PC1 loading corresponded to Os03G0165900/DJ.03G050100 (SDI1), followed by Os11G0437600/DJ.11G142900 (DUF50610 family) and Os01G0268600/DJ.01G124000 (Fig. 5D). Os11G0437600/DJ.11G142900 belongs to the plant-specific DUF506 protein family. In PC2, LS-treated NB samples had higher scores than controls (Fig. 5E). The highest PC2 loading was observed for Os03G0165900/DJ.03G050100 (SDI1), followed by Os11G0437600/DJ.11G142900 (DUF50610) and Os08G0379400/DJ.08G166300 (RNG3), which is associated with disease resistance (Fig. 5F). In DJ123 roots, PC2 values were also higher under LS conditions (Fig. 5G), with the highest PC2 loading corresponding to Os03G0618300/DJ.03G281800 (2ODD22), followed by Os03G0195800/DJ.03G072500 (SULTR1;1) and Os11G0437600/DJ.11G142900 (Fig. 5H). Os03G0618300/DJ.03G281800 encodes a member of the 2-oxoglutarate (2OG)/Fe(II)-dependent oxygenase superfamily homologous to Arabidopsis JAO2, which is involved in detoxification of polycyclic aromatic hydrocarbons (Smirnova et al., 2017). Overall, PCA consistently identified SDI1, SULTR1;1, DUF506-family genes, RNG3, 2ODD22, and Os01G0268600/DJ.01G124000, which encodes a glycoside hydrolase family 31 protein, as representative LS-responsive genes across conditions. RT–qPCR analysis confirmed that all six genes were significantly up-regulated in LS-treated NB and DJ123 roots under both hydroponic and soil conditions (Tukey’s HSD, P<0.05) (Fig. 5I). Principal component analysis (PCA) of gene expression in NB and DJ123 roots. PCA was conducted for gene expression data from NB and DJ123 under hydroponic and soil conditions. (A–D) Comparison of LS-responsive genes between the two varieties. (E–H) Within-variety comparisons. In (A, C, E, and G), the y- and x-axes represent PC1 and PC2 scores, respectively. In (B, D, F, and H), the corresponding PC1 and PC2 loadings are shown, where six PCA-selected representative LS-responsive genes are labeled as α–ζ. These six genes (SDI1, SULTR1;1, DUF506-family gene, RNG3, 2ODD22, and Os01G0268600/DJ.01G124000) were subsequently validated by RT–qPCR in (I). (I) The expression levels of these PCA-selected representative genes, all of which exhibited significantly higher expression under LS than under control conditions (Tukey’s HSD, P<0.05). Data represent means ±SDs of four biological replicates. Low-sulfur-responsive genes involved in the sulfur assimilation pathway Genes encoding sulfate transporters (SULTRs) and APS reductases, which are involved in the sulfur assimilation pathway in Arabidopsis, are induced under LS conditions (Takahashi et al., 1997, 2011; Takahashi, 2010, 2019). To identify LS-responsive genes involved in the sulfur assimilation pathway in rice, we analyzed the expression of these genes in NB and DJ123 plants grown under hydroponic and soil conditions. The SULTRs in both NB and DJ123 comprise 12 members, mirroring the number found in Arabidopsis, where they are classified into four major groups. To classify rice SULTRs, a phylogenetic tree was constructed using SULTR amino acid sequences from rice (NB and DJ123), green foxtail (Setaria viridis), maize, soybean, and Arabidopsis. Rice SULTRs were categorized into four groups (Supplementary Fig. S4), with three and two members in the SULTR1 and SULTR2 groups, respectively, as in Arabidopsis. In contrast, the SULTR3 and SULTR4 groups consist of six members and one member, respectively, representing one more and one fewer than in Arabidopsis (Takahashi et al., 2011). Among the rice SULTRs, genes encoding SULTR1;1 and SULTR4;1 were strongly induced under LS conditions in both NB and DJ123 under hydroponic and soil conditions (Fig. 6A). Among other genes involved in sulfur assimilation, APRL1 (encoding APS reductase) and OASTLA2 [encoding O-acetylserine (thiol) lyase] were induced under LS conditions in both NB and DJ123 grown under hydroponic and soil conditions (Fig. 6B). SLIM1/EIL3 is a key transcription factor in the Arabidopsis transcriptional response to LS, and both SLIM1/EIL3 and its rice counterparts have been shown to complement Arabidopsis slim1 mutants (Maruyama-Nakashita et al., 2006). The EIL transcription factor family consists of seven members in NB and six in DJ123, similar to those in Arabidopsis, which has six members (Fig. 6C). In NB, the nucleotide sequences of EIL1 and EIL1A are highly similar, whereas EIL1A is absent from the DJ123 genome. Among the EIL transcription factor genes, EIL4 was the only gene strongly induced under LS conditions in both NB and DJ123 under hydroponic and soil conditions (Fig. 6C). RT–qPCR analysis further confirmed that SULTR4;1, APRL1, OASTLA2, and EIL4 were significantly up-regulated in LS-treated NB and DJ123 roots under both hydroponic and soil conditions (Tukey’s HSD, P<0.05) (Fig. 6D). Heatmap and RT–qPCR validation of LS–inducible gene expression in NB and DJ123 roots. This figure shows log2 expression ratios of key LS–inducible genes in NB and DJ123 roots. Each square indicates the relative FC as shown by the scale bar, and dark gray squares indicate no significant differences. (A–C) Expression patterns of sulfate transporters (A), sulfur assimilation enzymes (B), and EIL transcription factors (C). In (C), EIL1A is absent from the DJ123 genome. H and S represent hydroponic and soil conditions, respectively. Gene symbols follow the RAP–DB nomenclature. (D) RT–qPCR validation of SULTR4;1, APRL1, OASTLA2, and EIL4. Data represent means ± SDs of four biological replicates. Conserved promoter sequences in 64 low-sulfur-inducible promoters of NB and DJ123 and the transactivation activity of EIL4 To identify conserved promoter sequences associated with LS inducibility in rice, we analyzed the promoter regions of 64 genes that were consistently up-regulated in both NB and DJ123 under LS hydroponic and LS soil conditions. Using our genome-wide promoter analysis tool (Maruyama et al., 2012), we searched for conserved hexamer motifs within these promoters in both varieties (Supplementary Tables S12, S13). Z-scores for all possible hexamers (46=4096) were visualized as scatter plots (Fig. 7A). Among these, TCTAGA and AGATTC exhibited the highest and second-highest Z-scores, respectively; these sequences correspond to the core regions of the UPE-box consensus sequence and the SEBS motif. We further focused on SEBS and analyzed its genome-wide distribution. In the NB genome, SEBS was detected in 475 of 22 143 analyzable promoters (Fig. 7B), whereas it was present in 22 of the 64 LS-inducible promoters (Fig. 7C). Similarly, in DJ123, SEBS was found in 577 of 27 889 analyzable promoters (Fig. 7D) and in 23 of the 64 LS-inducible promoters (Fig. 7E). Consensus SEBS sequences were visualized as sequence logos for all analyzable promoters and for the 64 LS-inducible promoters in NB (Fig. 7F, G) and DJ123 (Fig. 7H, I). The reverse complementary motifs were also analyzed (Fig. 7J–M). In the forward SEBS consensus, the 10th base was strongly conserved as T in analyzable promoters (Fig. 7F, H), whereas LS-inducible promoters exhibited either T or G at this position (Fig. 7G, I). In the reverse complementary consensus, the first base varied (T or A) in analyzable promoters (Fig. 7J, L) but was exclusively T in LS-inducible promoters (Fig. 7K, M). Conserved promoter sequences in LS-inducible genes and transactivation by EIL4. Hexamer sequences in 1 kb promoter regions were analyzed using a promoter research tool (Maruyama et al., 2012), and positive Z-scores were plotted for NB (x-axis) and DJ123 (y-axis) (A). Pie charts show the proportion of promoters containing the canonical SLIM1/EIL3-binding sequence (SEBS) or lacking SEBS in NB and DJ123 (B-E). Forward SEBS consensus motifs (Fwd) were visualized as sequence logos for all analyzable promoters (NB, F; DJ123, H) and for the 64 LS-inducible promoters (NB, G; DJ123, I) (F-I). Reverse complementary SEBS consensus motifs (RCs) were visualized for the corresponding promoter sets (NB, J and K; DJ123, L and M) (J-M). The effector construct used for transient expression assays is shown in (N), in which EIL4 was expressed under the maize ubiquitin promoter. Reporter constructs are shown in (O-Q): (O) illustrates native promoter reporters (IRLpro and S1;1pro), (P) illustrates reporters carrying canonical SEBS tandem repeats (SEBS×2 and SEBS×4), and (Q) illustrates reporters carrying SEBS-like tandem repeats (SEBS-l×2 and SEBS-l×4). In these schematic diagrams, the TATA box, canonical SEBS motifs, and SEBS-like motifs are indicated as shown in the figure. (R-T) Relative LUC activity of the six reporter constructs (IRLpro, S1;1pro, SEBS×2, SEBS×4, SEBS-l×2, and SEBS-l×4) co-transfected with the EIL4 effector or the empty vector control (Ctrl) in rice protoplasts. Fold induction was calculated relative to the Ctrl treatment. Different letters indicate significant differences (Tukey's HSD, P<0.05). Data represent means ± SD of six biological replicates. To assess whether EIL4 can activate transcription through SEBS or SEBS-like elements, we conducted transient effector–reporter assays in rice protoplasts and deliberately selected two LS-inducible promoters with contrasting motif contexts (Fig. 7N–T). First, IRL was selected because it is one of the most strongly LS-induced genes in both NB and DJ123 and contains canonical SEBS motifs, thereby serving as a representative SEBS-dependent promoter. Second, SULTR1;1 was chosen because it is a representative LS-inducible sulfate transporter gene that lacks canonical SEBS but harbors a SEBS-like sequence (5′-TGAATCTAGAGCA-3′), allowing us to test whether EIL4 activation extends to SEBS-like variants. The IRL and SULTR1;1 promoters were fused to the LUC reporter gene (IRLpro–LUC and SULTR1;1pro–LUC; Fig. 7O). EIL4 significantly enhanced the expression of both reporters (Tukey’s HSD, P<0.05) (Fig. 7R), suggesting that EIL4 can act through canonical SEBS as well as SEBS-like sequences. Synthetic promoters containing repeated SEBS or SEBS-like motifs (×2 and ×4) (Fig. 7P, Q; Supplementary Table S14) were also activated by EIL4 (Fig. 7S, T), with ×4 constructs exhibiting approximately double the activity of ×2 constructs, consistent with a motif copy number-dependent response. Transgenic rice plants overexpressing SULTR1;1 grown in low-sulfur and control soils SULTR1;1 in both NB and DJ123 is a representative LS-inducible gene involved in sulfur assimilation (Figs 3I, 5I, 6A). Both alleles encode polypeptides of 662 amino acids. Sequence comparison of the full-length ORF identified five SNPs at nucleotide positions 590, 808, 820, 1131, and 1977 relative to the start codon (Supplementary Fig. S5), three of which (590, 808, and 820) were non-synonymous (Supplementary Fig. S6). To assess whether these polymorphisms exhibited subspecies-wide patterns, the SULTR1;1 ORF sequence from NB was used as a reference to inspect nucleotide variation across Japonica, Tropical Japonica, Indica, and Aus accessions in the Rice Core Collection using TASUKE (Kumagai et al., 2013). This analysis revealed highly conserved, subspecies-specific allele distributions at all five positions—590 (Aus/Indica=T; Japonica/Tropical Japonica=C), 808 (Aus=A; Indica/Japonica/Tropical Japonica=G), 820 (Aus/Indica=A; Japonica/Tropical Japonica=G), 1131 (Aus/Indica=C; Japonica/Tropical Japonica=T), and 1977 (Aus/Indica=A; Japonica/Tropical Japonica=G) (Supplementary Fig. S7). These conserved patterns indicate lineage-specific divergence at the SULTR1;1 locus among rice subspecies. To evaluate the agricultural utility of SULTR1;1 from DJ123 in LS soils of the Volta River Basin (Ghana), we generated transgenic NB plants overexpressing the DJ123 SULTR1;1 allele (hereafter referred to as Stox1 and Stox2) (Fig. 8A). Stox and transgenic control (T-cont) plants were grown under LS and control soil conditions, and their growth characteristics were assessed. The expression levels of SULTR1;1 in Stox plants were markedly higher than those in T-cont plants under control conditions, but only slightly higher under LS conditions (Tukey’s HSD, P<0.05) (Fig. 8B). This suggests that overexpression caused a substantial elevation in transcript abundance under control conditions, whereas under LS conditions, endogenous SULTR1;1 expression was already strongly induced. Under control conditions, shoot and root sulfur concentrations were significantly higher in Stox than in T-cont plants, whereas under LS conditions, no significant differences were observed (Tukey’s HSD, P<0.05) (Fig. 8C), indicating that SULTR1;1 overexpression did not further enhance sulfur accumulation under sulfur limitation. The shoot and root biomasses (DW) of Stox plants were significantly lower than those of T-cont plants under both soil conditions (Tukey’s HSD, P<0.05) (Fig. 8D, E). Root length did not differ significantly between Stox and T-cont plants under control conditions, but was significantly greater in Stox plants under LS conditions (Tukey’s HSD, P<0.05) (Fig. 8F, G), where the roots were visibly longer and thinner, suggesting that SULTR1;1 overexpression promotes root elongation specifically under sulfur limitation. Consistent with this, both NB and DJ123 plants grown under LS hydroponic conditions also showed significantly longer roots than those under control conditions (Tukey’s HSD, P<0.05) (Supplementary Fig. S8). Together, these findings suggest that SULTR1;1 enhances root elongation under LS conditions, potentially contributing to improved root foraging capacity in sulfur-limited soils. Growth of rice plants overexpressing SULTR1;1 in low-sulfur (LS) soil. Overexpression lines (Stox1 and Stox2) and their transgenic controls (T-cont) were grown under LS and control soil conditions (A). SULTR1;1 expression levels (B), sulfur concentrations in shoots (C) and roots (D), shoot DW (E), root DW (F), and root length (G) are shown as bar graphs. Data represent means ±SD (n=4 biological replicates; Tukey’s HSD, P<0.05 and P<0.01). Discussion The genome of NB, a lowland rice variety, has been fully sequenced (International Rice Genome Sequencing Project, 2005), and its genome sequence and transcript data have been compiled in the RAP-DB (Sakai et al., 2013). To investigate the LS response in both upland and lowland rice, we established a genome and transcriptome analysis platform for the upland variety DJ123 (Fig. 1). Comparative genomic analysis revealed an inversion near the centromeric region of chromosome 6, suggesting divergence of ancestral lineages between the two varieties. We identified LS-responsive genes in the roots of NB and DJ123 plants grown under both hydroponic and soil conditions, and compared their expression profiles across these environments. The number of LS-responsive genes was markedly higher in soil-grown plants than in those grown hydroponically; however, the most strongly induced genes were largely shared between NB and DJ123 regardless of growth conditions (Fig. 3). The Ghanaian soil used in this study contained a low level of exchangeable sulfur (4.7 mg kg−1; Supplementary Table S1), indicating that sulfate availability to rice plants was limited. In addition, interactions between sulfate ions and soil solid phases or other physicochemical properties could have reduced effective sulfate activity in the rhizosphere (Parfitt, 1978; Tanikawa et al., 1999; Fukushi et al., 2007), potentially enhancing LS responses compared with hydroponic conditions. Therefore, genes that responded consistently to LS in both varieties and growth systems were defined as LS-responsive and were used for subsequent comparative analyses. Consistent with this, PCA showed similar overall transcriptional trends between growth systems, supporting the conclusion that key LS-responsive genes were commonly regulated in both NB and DJ123 (Fig. 4). This integrated approach provided a comprehensive view of how rice roots respond to sulfur limitation in both controlled and soil-based environments. Among all LS-inducible genes, SDI1 showed the greatest increase in expression under LS conditions (Fig. 3). In Arabidopsis, SDI1 forms a complex with MYB28 to suppress aliphatic glucosinolate (GSL) biosynthesis (Aarabi et al., 2016, 2021). Interestingly, SDI1 was also strongly induced under LS in rice, a species that lacks GSLs, suggesting that SDI1 may regulate sulfur metabolism through a distinct mechanism. The transcriptional responses observed in rice differ from those reported in Arabidopsis, where SLIM1/EIL3, the central regulator of the LS response, is not transcriptionally induced under sulfur limitation (Maruyama-Nakashita et al., 2006). In contrast, in rice, EIL4—a homolog of SLIM1/EIL3—showed a marked increase in transcript levels under LS conditions. Although the regulatory basis for this difference remains unclear, it may represent a species-specific adaptation in which transcriptional induction of EIL4 enables rapid reprogramming of sulfur-responsive genes in roots. This additional transcriptional layer suggests that rice may employ a more flexible regulatory mechanism for maintaining sulfur homeostasis than Arabidopsis, possibly reflecting adaptation to more fluctuating soil nutrient conditions. Previous LS response studies in rice have been conducted using the varieties Kitaake and Zhonghua 11 (Wang et al., 2022; Zenzen et al., 2024). Their genome sequences are publicly available, with 35 594 protein-coding genes annotated in Kitaake (Phytozome). Although a previous report suggested that SULTR1;1 and SULTR1;2 were absent in Kitaake (Zenzen et al., 2024), our re-examination of Phytozome data confirmed that Kitaake possesses 12 SULTR members—comparable with those in NB and DJ123—and that their coding sequences are identical between Kitaake and NB (Supplementary Table S15). In Zhonghua 11, SULTR1;1 and OASTLA2 were up-regulated under LS in roots (Wang et al., 2022), consistent with responses in NB and DJ123. In contrast, SULTR4;1 and EIL4 were not significantly induced in Zhonghua 11, possibly reflecting differences in hydroponic culture conditions. Zenzen et al. (2024) reported that EIL4 (OsKitaake08G207100) was strongly induced by LS in Kitaake shoots, yet their figures also indicated increased root expression. Because tissue origins were not clearly indicated in their Supplementary data, we reanalyzed the publicly available RNA-seq dataset (Supplementary Tables S16–S19). Our reanalysis confirmed that EIL4 is up-regulated by LS in both shoots and roots of Kitaake. SULTR1;1 (OsKitaake03G073800) was likewise LS inducible in both tissues. Despite minor promoter differences between NB and DJ123, EIL4 promoter sequences were identical among NB, Kitaake, and Zhonghua 11, and were highly conserved within subspecies (Supplementary Fig. S9), suggesting that the regulatory framework controlling EIL4 expression under LS is conserved across rice varieties. We conducted a genome-wide analysis of LS-inducible promoters in both NB and DJ123, and found that the SEBS—previously identified as a binding site for SLIM1/EIL3 in Arabidopsis—is highly conserved among LS-inducible promoters in rice (Fig. 7). However, neither the NB nor the DJ123 SULTR1;1 promoter contained a canonical SEBS motif, although both possessed SEBS-like sequences that differed by one base. To determine whether EIL4 could activate transcription through these motifs, we performed transient expression assays using rice protoplasts. The results showed that EIL4 activated transcription from both SEBS and SEBS-like promoters (Fig. 7), suggesting that EIL4 mediates LS-responsive gene expression via recognition of both canonical and SEBS-like motifs. Notably, the SEBS was conserved in LS-inducible promoters of both NB and DJ123, whereas SURE11—the key cis-element in the Arabidopsis LS pathway (Maruyama-Nakashita et al., 2005)—did not show comparable conservation in rice. This suggests that Arabidopsis and rice may rely on distinct cis-regulatory architectures for LS signaling. A subset of LS-inducible promoters lacked both canonical SEBS and recognizable SEBS-like sequences, indicating the presence of additional regulatory modes. One possibility is that EIL4 or other EIL-family proteins may bind sequences with partial similarity to SEBS, even when these motifs fall below the threshold used in our motif search. In addition, our promoter analysis identified the ATTACT hexamer—conserved in LS-inducible promoters of both varieties (Supplementary Tables S12, S13)—which resembles a motif implicated in abscisic acid (ABA)-dependent autoregulation of MYB41 in Arabidopsis (Gao et al., 2024). The enrichment of this motif raises the possibility that, in addition to SEBS, other promoter motifs may also contribute to the regulation of LS-inducible genes, potentially linking LS signaling with ABA-responsive pathways. Finally, abiotic stresses are known to influence transcript turnover in plants (Maruyama et al., 1999; Chantarachot and Bailey-Serres, 2018). Given that many LS-inducible genes exhibit only modest changes in transcript abundance, increased mRNA stability may also contribute to their accumulation. Thus, LS-responsive gene expression in rice is likely to be shaped by an integration of transcriptional regulation and general stress-associated post-transcriptional mechanisms. Together, these findings indicate that rice employs both SEBS-dependent and SEBS-independent strategies to orchestrate LS-responsive transcription, integrating multiple promoter motifs and broader stress-responsive regulatory layers within a framework distinct from that of Arabidopsis. In addition to the induction of sulfur assimilation genes, several genes associated with sulfur transfer and sulfur-rich secondary metabolism were markedly repressed under LS. Nicotianamine synthase 1 and 2 (NAS1 and NAS2) showed the strongest decreases in transcript abundance in both varieties (Supplementary Fig. S2; Higuchi et al., 1999; Kobayashi and Nishizawa, 2005). Similarly, Sulfurtransferase 7 (STR7) and 9 (STR9) were down-regulated under both hydroponic and soil conditions (Supplementary Fig. S2; Hatzfeld et al., 2000; Papenbrock and Schmidt, 2000). Because STR enzymes catalyze sulfur transfer and participate in secondary sulfur metabolism and detoxification, their repression probably reflects a reallocation of sulfur toward primary assimilation. Moreover, Thionin 34 (THI34) and Chitinase 9 (CHT9), both encoding sulfur-rich defense proteins, were also strongly repressed (Supplementary Fig. S2; Collinge et al., 1993; Thatayaone et al., 2015). Collectively, these patterns indicate that rice activates a sulfur-saving program under LS stress—limiting sulfur use in secondary and defense metabolism to maintain essential amino acid and protein synthesis—analogous to the repression of GLS biosynthesis and activation of sulfur recycling observed in Arabidopsis (Takahashi et al., 2011; Maruyama-Nakashita, 2017). Previous work reported that transgenic tobacco plants overexpressing GmSULTR1;2b, a soybean homolog of Arabidopsis SULTR1;1, exhibited increased sulfur concentration and slightly enhanced biomass under reduced sulfur supply (Ding et al., 2016). However, whether such effects translate into improved tolerance under severely LS soil conditions remains unclear. To evaluate the agricultural potential of SULTR-mediated enhancement of sulfur uptake in rice, we generated transgenic NB plants overexpressing SULTR1;1 from DJ123 and evaluated their performance in LS and control soils of the Volta River Basin (Fig. 8A). SULTR1;1 expression was markedly higher in transgenic plants (Stox) than in controls under normal sulfur conditions, and modestly higher under LS (Fig. 8B). Correspondingly, sulfur concentration was greater in Stox than in controls under control conditions, but similar under LS (Fig. 8C). These results suggest that soil sulfur levels in the Volta Basin are so low that even enhanced transporter expression cannot increase uptake. Although SULTR1;1 was driven by a constitutive ubiquitin promoter, development of a strong, root-specific, LS-inducible promoter may improve sulfur uptake efficiency under such conditions. However, our synthetic SEBS-based promoters exhibited lower activity than the native IRL promoter in transient assays, highlighting the technical challenges of constructing strong yet finely tuned LS-inducible promoters. Under current technological and varietal constraints, sulfur fertilization remains indispensable for stable rice production in LS soils. Root length was significantly greater in Stox plants than in controls on LS soils, consistent with LS-induced elongation observed in NB and DJ123 under hydroponic LS conditions (Supplementary Fig. S8). Similarly, SULTR1;1 expression and root elongation are linked under LS in Arabidopsis (Zhao et al., 2014; Joshi et al., 2019). SLIM1/EIL3 and GGCT2;1 have been implicated in this response, connecting sulfate uptake and glutathione catabolism to adaptive root elongation (Maruyama-Nakashita et al., 2006; Barberon et al., 2008; Paulose et al., 2013). Our results, together with previous findings, suggest that enhanced root elongation is a conserved adaptive response to LS stress and that SULTR1;1 contributes to this process in rice. In summary, by combining genomic, transcriptomic, and functional analyses of upland and lowland rice varieties, this study provides an integrated understanding of LS-responsive gene regulation in rice roots. Our findings highlight both conserved and species-specific aspects of sulfur signaling, identify EIL4 as a central regulator acting through SEBS-related promoter motifs, and suggest that SULTR1;1 contributes to root adaptive responses under sulfur limitation. This work establishes a foundation for future research aimed at improving sulfur-use efficiency and rice productivity in sulfur-deficient soils. Supplementary Material Acknowledgements The soil samples were legally imported into Japan for pot experiments with permission from the Ministry of Agriculture, Forestry and Fisheries of Japan (27Y289, 1Y2224). We would like to express our deepest gratitude to Dr Kazuko Yamaguchi-Shinozaki and Dr Kazuo Shinozaki. Abbreviations adenosine 5′-phosphosulfate Ethylene-insensitive 3-like Benjamini and Hochberg false discovery rate fold change low-sulfur luciferase Nipponbare overexpressing DJ123 SULTR1;1 principal component analysis reverse transcription–quantitative PCR SLIM1/EIL3-binding sequence Sulfur deficiency-induced 1 Sulfur limitation 1 Sulfur-responsive element Sulfate transporter transgenic control Tukey’s honestly significant difference test Contributor Information Kyonoshin Maruyama, Biological Resources and Post-Harvest Division, Japan International Research Center for Agricultural Sciences, 1-1 Ohwashi, Tsukuba, Ibaraki 305-8686, Japan; Institute of Life and Environmental Sciences, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8572, Japan. Yasuhiro Tsujimoto, Crop, Livestock and Environment Division, Japan International Research Center for Agricultural Sciences, 1-1 Ohwashi, Tsukuba, Ibaraki 305-8686, Japan. Miho Kishimoto, Biological Resources and Post-Harvest Division, Japan International Research Center for Agricultural Sciences, 1-1 Ohwashi, Tsukuba, Ibaraki 305-8686, Japan. Kyouko Yoshiwara, Biological Resources and Post-Harvest Division, Japan International Research Center for Agricultural Sciences, 1-1 Ohwashi, Tsukuba, Ibaraki 305-8686, Japan. Hiroaki Sakai, Bioinformatics Unit, Research Center for Advanced Analysis, Core Technology Research Headquarters, National Agriculture and Food Research Organization, 2-1-2 Kannondai, Tsukuba, Ibaraki 305-8518, Japan. Tetsuya Sakurai, Faculty of Agriculture and Marine Science, Kochi University, 200 Monobe, Nankoku, Kochi 783-8502, Japan. Shingo Sakamoto, Biomanufacturing Process Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Central 6, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8566, Japan. Nobutaka Mitsuda, Biomanufacturing Process Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Central 6, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8566, Japan. Akiko Maruyama-Nakashita, Department of Bioscience and Biotechnology, Faculty of Agriculture, Kyushu University, 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan. Stanislav Kopriva, Universität zu Köln, Germany. Supplementary data The following supplementary data are available at JXB online. Fig. S1. LS-induced genes commonly up-regulated in NB and DJ123. Fig. S2. LS-induced genes commonly down-regulated in NB and DJ123. Fig. S3. LS-responsive genes specifically regulated in NB or DJ123. Fig. S4. Phylogenetic analysis of SULTRs in rice, green foxtail, maize, soybean, and Arabidopsis. Fig. S5. Nucleotide sequence alignment of SULTR1;1 in NB and DJ123. Fig. S6. Amino acid sequence alignment of SULTR1;1 in NB and DJ123. Fig. S7. Single nucleotide polymorphisms (SNPs) in SULTR1;1 across rice subspecies. Fig. S8. LS response of NB and DJ123 rice varieties grown in culture medium. Fig. S9. Promoter sequence alignment of EIL4 in NB, Kitaake, Zhonghua11, and DJ123. Table S1 Physicochemical properties of the soil used for the experiments Table S2. Primers for RT–qPCR Table S3. Genome statistics of the 12 rice chromosomes Table S4. Up-regulated genes in NB under LS hydroponic conditions Table S5. Down-regulated genes in NB under LS hydroponic conditions Table S6. Up-regulated genes in NB under LS soil conditions Table S7. Down-regulated genes in NB under LS soil conditions Table S8. Up-regulated genes in DJ123 under LS hydroponic conditions Table S9. Down-regulated genes in DJ123 under LS hydroponic conditions Table S10. Up-regulated genes in DJ123 under LS soil conditions Table S11. Down-regulated genes in DJ123 under LS soil conditions Table S12. Conservation of all hexamers in NB promoters Table S13. Conservation of all hexamers in DJ123 promoters Table S14. Promoter sequences of SULTR1;1, IRL SEBS, and SEBS-like in Nipponbare and DJ123. Table S15. Key genes involved in sulfur metabolism in Nipponbare, DJ123, and Kitaake. Table S16. Up-regulated genes in shoots of Kitaake under LS conditions Table S17. Down-regulated genes in shoots of Kitaake under LS conditions Table S18. Up-regulated genes in roots of Kitaake under LS conditions Table S19. Down-regulated genes in roots of Kitaake under LS conditions Author contributions KM: designed the research; KM, YT, MK, KY, SS, and NM: performed the research; KM, YT, HS, and TS: analyzed the data; KM, YT, HS, TS. and AMN: wrote the paper. Funding This work was supported by grants from the Ministry of Agriculture, Forestry and Fisheries (MAFF); the Japan Society for the Promotion of Science (JSPS; 15H05257, 22H02229); and the Science and Technology Research Partnership for Sustainable Development (SATREPS) of the Japan Science and Technology Agency (JST)/Japan International Cooperation Agency (JICA; JPMJSA1608). Data availability The raw sequence data generated in this study have been deposited in the Sequence Read Archive (SRA) database of the DNA Data Bank of Japan (DDBJ) under the BioProject ID PRJDB16314. The oligo microarray design and data were deposited at ArrayExpress (E-MTAB-13217 and E-MTAB-15013). References Associated Data Supplementary Materials Data Availability Statement The raw sequence data generated in this study have been deposited in the Sequence Read Archive (SRA) database of the DNA Data Bank of Japan (DDBJ) under the BioProject ID PRJDB16314. The oligo microarray design and data were deposited at ArrayExpress (E-MTAB-13217 and E-MTAB-15013).