WT rice (Oryza sativa L) cultivar Nipponbare (Nip), along with OsMADS31 knockout (osmads31) and overexpression (OE) lines generated in the Nip background, were used in this study. Seeds were germinated and cultivated hydroponically following established protocols. Plants were grown in a greenhouse under controlled conditions: 30°C day/28°C night temperature, 70% relative humidity, 14-h light/10-h dark photoperiod, and 600 μmol m^-² s^-¹ light intensity.

The CRISPR/Cas9 system was employed to generate osmads31 knockout mutants. Two 20-bp target sequences (5’-GTGATTGTGTTCTCAGGCAC-3’ and 5’-GCACCCGTTTTGAGGAGATG-3’) were selected from the OsMAS31 coding sequence using established protocols (Miao et al., 2013). The osmads31 knockout plasmids were introduced into the Nip cultivar via Agrobacterium-mediated transformation. Transgenic plants were selected for hygromycin resistance and genotyped by PCR using knockout identification primers (osmads31-F/R, Supplementary Table 1 ). Two independent osmads31 knockout mutant lines (KO1 and KO2) were obtained, and potential off-target effects were evaluated using CRISPR-P (http://cbi.hzau.edu.cn/cgi-bin/CRISPR) (Liu et al., 2017).

To generate OsMADS31 overexpression (OE) lines, the OsMADS31 open reading frame (ORF) was amplified from Nip cDNA using overexpression identification primers (OsMADS31OE-F/R, Supplementary Table 1 ) and then cloned into the pCAMBIA1390-Ubi vector. To generate pOsMADS31::GUS transgenic plants, a 2-kb fragment of the OsMADS31 promoter (upstream of ATG) was amplified from the Nip genomic DNA using GUS identification primers (OsMADS31GUS-F/R, Supplementary Table 1 ) and cloned into pCAMBIA1301-Ubi. All constructs were verified by sequencing (Tsingke Biotechnology) and subsequently introduced into Agrobacterium tumefaciens strain EHA105 by electroporation. The transformed Agrobacterium cells were then used to transfect Nip plants as described previously (Hiei et al., 1994).

Total RNA was extracted from rice cultivar Nip plants using the TRIzol Reagent (Invitrogen, Waltham, MA, USA). To carry out the spatial expression analysis of OsMADS31, total RNA was isolated from root and shoot in Nip three-leaf stage seedlings, and different tissues of Nip heading stage including root, node, leaf, culm, anther, and young spike. The isolated total RNA was reverse-transcribed into cDNA. Subsequently, RT-qPCR was performed on the ABI PRISM 7500 system (Applied Biosystems) using Takara RR420A reagents and gene-specific primers designed with Oligo 7 (Sangon Biotech, Shanghai). OsActin served as the internal control gene.

Seedlings of the WT as well as osmads31 mutant and OsMADS31 overexpression lines were treated with 6‰ NaCl at the three-leaf-stage. RNA was extracted from the salt-treated seedlings using TRIzol Reagent, and cDNA was synthesized using HiScript II QRT SuperMix (+gDNA wiper) (Vazyme R223-01). Then, qPCR was conducted using ChamQ Universal SYBR qPCR Master Mix (Vazyme Q711-02), with OsActin serving as the reference gene. Relative gene expression levels were calculated using the 2^−ΔΔCt method (Livak and Schmittgen, 2001) and presented as mean ± SD. All primers used in this study are listed in Supplementary Table 2 .

GUS staining was performed as previously described (Li et al., 2015). Briefly, tissues of pOsMADS31::GUS transgenic rice plants were incubated in GUS staining solution overnight at 37°C. Plant tissues were subsequently destained using 75% ethanol to remove chlorophyll. Then, the tissues were examined and photographed using the Olympus SZX7 microscope.

The accumulation of O₂ ^•- and H₂O₂ in rice plants was determined by staining with nitroblue tetrazolium chloride (NBT) and 3,3’-diaminobenzidine (DAB), respectively, as described previously (Huang et al., 2019; Jambunathan 2010). Briefly, the second leaves were collected from the KO, OE and WT plants before and after the 6‰ NaCl treatment. To perform NBT staining, 3-cm segments of leaves were prepared, vacuum-infiltrated with the NBT solution (0.1%, pH 7.8) for 30 min, and incubated at room temperature for 24 h. To perform DAB staining, leaf samples were vacuum-infiltrated with the DAB solution (1 mg/mL in 10 mM Na₂HPO₄ containing 0.05% Tween-20, pH 3.8) for 30 min and then incubated at 37°C with shaking for 48 h. Both the NBT- and DAB-stained samples were destained with 95% ethanol and photographed under a stereomicroscope (Nikon XAZ100).

To determine the subcellular localization of OsMADS31, the coding sequence (CDS) of OsMADS31 (excluding the stop codon) was amplified from the Nip genomic DNA using gene-specific primers ( Supplementary Table 1 ). The PCR product was cloned into the pCAMBIA1390-GFP vector. The resulting 35S::OsMADS31-GFP construct was transiently expressed in rice protoplasts via polyethylene glycol (PEG)-mediated transformation (Wang et al., 2021). GFP signal was visualized using the Zeiss LSM880 confocal microscope (Wang et al., 2022a).

After treatment with different concentrations of NaCl, the germination rate of seeds was counted on the seventh day. Transgenic and WT seeds were incubated on Murashige and Skoog (MS) medium containing varying NaCl concentrations for 12 days, and root and shoot lengths were measured subsequently.

Uniform, plump rice seeds were surface-sterilized with 0.3% sodium hypochlorite for 20 min, rinsed three times with distilled water, and placed on moist filter paper in Petri dishes at 37°C. The germinated seeds were transferred to 96-well trays containing culture solution, replaced every 4 days. After growing to the three-leaf stage (~2 weeks) under standard hydroponic conditions, seedlings were exposed to 6‰ NaCl solution for 6 days. Subsequently, the seedlings were allowed to recover for 6 days in normal solution before calculating the survival rate.

SOD, POD, and CAT activities as well as MDA and Pro contents were measured using the respective kits, as described previously (Gao et al., 2023a, Gao et al., 2023b, Gao et al., 2023c). Measurements were taken both pre-treatment and at 24 h post-salt-stress treatment. Three biological replicates were performed, and data were expressed as mean ± SD.

WT and osmads31 mutant transgenic seedlings at the three-leaf stage were sampled pre-treatment and at 24 h after 6‰ NaCl exposure. Three biological replicates were performed per genotype, and the seedlings were flash-frozen in liquid nitrogen and submitted to Genedenovo Biotechnology Co., Ltd (Guangzhou, China) for RNA-seq. The obtained transcriptomic data were analyzed using established methods (Gao et al., 2023b, Gao et al., 2023c; Zhang et al., 2020). Additional data analysis was conducted using the Genedenovo Biotechnology Platform Omicsmart (https://www.omicsmart.com). The cDNA libraries were sequenced on the Illumina sequencing platform by Genedenovo Biotechnology Co., Ltd (Guangzhou, China).

Relative expression levels of differentially expressed genes (DEGs), identified through RNA-seq, were subjected to log-transformation and normalization. Subsequently, the gene expression data were examined using correlation analysis, weighted gene co-expression network analysis (WGCNA)-based network construction, and Cytoscape visualization (Hu et al., 2016). Protein-protein interactions were predicted using STRING (https://cn.string-db.org/).

Data are presented as the mean ± SD of three or more independent replicates. The mean of replicates for each experiment was calculated using the PASW Statistics 18 software. Statistical significance was determined using one-way ANOVA and Student’s t-test at p < 0.05. Bar charts were created using GraphPad Prism 10.3.1. Heatmaps were constructed and drawn using TBtools-II v.2.142 (Chen et al., 2023).

To characterize the conserved features of MIKC-type MADS domains in rice, we performed multiple sequence alignment of their amino acid sequences. The results showed that MADS domains of the DNA binding domain are highly conserved among the 61 MIKC-type MADS-box transcription factors found in rice ( Supplementary Figure 1 ). According to previous studies, MIKC-type MADS domains exist in two lineages (Type I and Type II) of MADS-box genes across plants, animals, and fungi. Notably, the majority of plant MADS-box genes belong to the Type II lineage and contain three additional domains compared with Type I genes: an intervening (I) domain, a keratin-like coiled-coil (K) domain, and a variable-length C-terminal (C) domain; these three domains together form the plant-specific MIKC-type structure (Nam et al., 2003). The conserved nature and distinctive characteristics of MIKC-type MADS domains provide important clues for determining the functional roles of MADS-box transcription factors in plant development and stress responses.

The OsMADS31 CDS comprises 723 base-pairs and is predicted to encode a 240-amino acid protein. The amino acid sequence of OsMADS31 contains both the MADS and K-box domains. The MADS domain is involved in DNA binding and protein dimerization. Notably, MADS-box family proteins typically function as dimers, with their primary DNA-binding element consisting of an antiparallel coiled-coil formed by two amphipathic α-helices from each subunit (Norman et al., 1988). The MADS domain is generally associated with the K-box region. Most plant MADS-box proteins possess the typical MIKC structure, which mediates dimerization and cofactor binding, serving as the primary determinant for DNA binding. The K-box protein interaction domain, which mediates heterodimerization of MIKC-type MADS-box proteins, contains several heptad repeats with hydrophobic amino acids at the first and fourth positions. This suggests that the K-box domain forms three amphipathic α-helices, designated as K1, K2, and K3 (Yang et al., 2003). Analysis of conserved motifs among the 61 highly conserved MIKC-type MADS-box proteins identified in rice ( Supplementary Figure 2 ) revealed that all of these 61 MADS-box proteins contain both the MADS and K-box domains, with sequence logos showing 12 conserved amino acid residue motifs. Phylogenetic analysis demonstrated that the rice gene OsMADS31 is evolutionarily closely related to its orthologs in maize ( Supplementary Figure 3 ).

The expression of OsMADS31 was examined in the root and shoot of WT Nip three-leaf stage seedlings ( Figure 1A ), and different tissues (roots, nodes, leaves, stems, anthers, and young panicles) of WT Nip heading stage ( Figure 1B ). The results of RT-qPCR analysis showed that the expression of OsMADS31 began to increase at 4 h after the salt stress treatment at the three-leaf-stage of WT Nip and peaked at 24 h, with expression levels reaching approximately 10-fold higher than the baseline level ( Figure 1C ). To further examine OsMADS31 expression, a vector containing the OsMADS31 promoter-driven GUS reporter gene was constructed and transformed into Nip. GUS staining of the resulting transgenic plants demonstrated GUS activity in the coleoptile of germinating seeds as well as in whole seedlings, stems, leaves, anthers, and young panicles ( Supplementary Figure 4 ). Quantitative analysis indicated higher expression in leaves and stem nodes, and the cross-sections of stained tissues revealed deeper staining in vascular bundles. These results suggest that OsMADS31 is expressed in multiple tissues and at different developmental stages, implying its functional importance in rice growth.

We also examined the subcellular localization pattern of OsMADS31. Online prediction tools indicated that OsMADS31 is nuclear-localized. These results were experimentally verified ( Figure 2 ).

To further investigate whether OsMADS31 is involved in salt stress tolerance, a vector for knocking-out OsMADS31 expression was constructed using the CRISPR/Cas9 technology and introduced into Nip via Agrobacterium-mediated transformation. The resultant osmads31 knockout mutants were verified by sequencing and off-target cleavage detection ( Supplementary Figure 5A ). Concurrently, OsMADS31 overexpression lines were generated in the Nip background. Finally, two knockout mutants (KO1 and KO2) and two overexpression lines (OE1 and OE2) were selected, based on the expression levels of OsMADS31 determined by RT-qPCR ( Supplementary Figure 5B ). All four genotypes were propagated to the T₂ generation, and homozygous knockout mutant and transgenic plants were used for subsequent experiments.

Next, we investigated the effects of the knockout mutation of OsMADS31 on the phenotype and agronomic traits of homozygous T₂-generation KO and OE plants at maturity, with the WT plants serving as controls. Knockout mutant plants exhibited significant reductions in grain length, grain width, plant height, seed-setting rate, 1000-grain weight, panicle length, and grains per panicle compared with WT plants. Conversely, overexpression lines displayed significant increases in panicle length and grain number per panicle compared with the WT ( Supplementary Figure 6 & 7 ). These results demonstrate that the knockout mutation of OsMADS31 affects rice seed development.

We further investigated the salt stress tolerance of osmads31 knockout mutant and OsMADS31 overexpression plants by performing seed germination assays. Uniformly plump seeds of the WT, osmads31 knockout mutants (KO1 and KO2), and OsMADS31 overexpression lines (OE1 and OE2) were germinated on MS medium containing no NaCl (control; CK) or different NaCl concentrations (6‰, 8‰, 9‰) ( Figure 3A ). Under normal conditions (CK), seeds of all genotypes showed normal germination and growth, with no significant differences. At 6‰ and 8‰ NaCl concentrations, compared with the WT, KO1 and KO2 seedlings exhibited significantly reduced shoot and root lengths, whereas OE1 and OE2 seedlings showed significantly increased shoot and root lengths ( Figure 3B-I ). At 9‰ NaCl concentration, although KO1 and KO2 seeds could germinate, their seedlings displayed nearly no root or shoot growth; by contrast, OE1 and OE2 seedlings were maintained growth of both shoots and roots ( Figure 3 ). These results demonstrate that OsMADS31 plays a crucial role in salt stress tolerance during seed germination.

To further evaluate the salt stress tolerance of the WT, osmads31 knockout mutants, and OsMADS31 overexpression lines, seedlings at the three-leaf stage were subjected to 6‰ NaCl treatment for 6 days, followed by a 6-day recovery period, and then analyzed for phenotypic differences ( Figure 4A-C ). The KO mutant seedlings exhibited wilting and growth retardation under salt stress, with significantly lower survival rates than the WT after recovery; by contrast, OE plants outperformed the WT in growth and showed significantly higher survival rates ( Figure 4D ). These results demonstrate that OsMADS31 enhances salt stress tolerance in rice.

Next, to determine the defense response of osmads31 knockout mutants and OsMADS31 overexpression lines to salt stress, we measured the activities of antioxidant enzymes (CAT, POD, and SOD) as well as the contents of MDA and Pro in WT, KO, and OE plants at the three-leaf stage before and after treatment with 6‰ NaCl for 24 h. Compared with the WT, KO mutants exhibited significantly lower CAT, POD, SOD activities and Pro content ( Figures 5A-D ), but significantly higher MDA content ( Figure 5E ); by contrast, OE mutants showed the opposite trends. We also determined the accumulation of ROS in all three genotypes before and after the salt stress treatment by performing DAB and NBT staining of leaves, followed by destaining. The intensity of NBT and DAB staining was significantly higher in KO mutants than the WT ( Figures 4E, F ). Together, these results indicate that knocking-out OsMADS31 expression reduces antioxidant enzyme activities and disrupts the homeostasis of related physiological parameters under salt stress in rice.

To further elucidate the role of OsMADS31 in salt stress response at the molecular level, we performed Gene Ontology (GO) enrichment analysis of WT and osmads31 knockout mutants before and after salt stress treatment DEGs. Comparative analysis of secondary bar plots revealed that, compared with the pre-treatment control (WT-vs-KO (CK)), salt-stressed samples (WT-vs-KO (Salt)) showed significantly fewer upregulated genes than downregulated genes. Notably, in the Biological Process category, genes involved in ‘response to stimulus’ were downregulated by more than twofold in WT-vs-KO (Salt) compared with WT-vs-KO (CK) ( Supplementary Figures 8A, C ). Volcano plot of differentially expressed genes after under salt stress ( Supplementary Figures 8E ). Comparative GO enrichment circle plots before and after the salt stress treatment demonstrated significant enrichment of genes in the Biological Process category ( Supplementary Figures 8B, D ). Bubble plot analysis further revealed that the WT-vs-KO (Salt) comparison was primarily enriched in GO terms such as ‘response to harmful substances’ and ‘metabolic processes’ ( Supplementary Figures 9A, C ). The directed acyclic graph (DAG) showed predominant enrichment of genes involved in ‘response to stimulus’ ( Supplementary Figures 9B, D ). Subsequent Gene Set Enrichment Analysis (GSEA) of ‘response to stimulus’-related genes after salt treatment highlighted processes related to salt stress response, while regulatory network analysis identified three key functional categories: protein kinases, catalases, and transcription factors ( Supplementary Figures 9E, F ). These results suggest that salt stress likely affects plant physiological responses by modulating specific gene expression patterns in rice.

Among the DEGs, 455 were upregulated and 756 were downregulated genes in KO mutants following salt treatment ( Supplementary Figure 10A, B ), suggesting that these DEGs may be directly associated with salt stress response. KEGG enrichment analysis identified the following top 10 pathways in WT-vs-KO (CK): ‘Biosynthesis of secondary metabolites, Cutin, suberin, and wax biosynthesis, Protein processing in the endoplasmic reticulum, Metabolic pathways, Glutathione metabolism, Carbon fixation in the calvin cycle, Photosynthesis-antenna proteins, Phenylpropanoid biosynthesis, Benzoxazinoid biosynthesis, and Betaine biosynthesis’ ( Supplementary Figure 10C ). Similarly, KEGG enrichment analysis of the WT-vs-KO (Salt) data showed enrichment of the following 10 pathways: ‘Phenylpropanoid biosynthesis, Biosynthesis of secondary metabolites, Plant hormone signal transduction, Pyruvate metabolism, Metabolic pathways, Alanine, aspartate and glutamate metabolism, Glycolysis/Gluconeogenesis, MAPK signaling pathway-plant, Plant-pathogen interaction, and Arginine and proline metabolism’ ( Supplementary Figure 10D ). KEGG hierarchy analysis of the top 20 pathways identified in WT-vs-KO (Salt) revealed carbon metabolism, amino acid biosynthesis and metabolic pathways as the three most affected pathways, with glycolysis/gluconeogenesis and lipid metabolism also potentially impacted. Subsequent GSEA revealed genes involved in plant hormone signal transduction ( Supplementary Figure 10E-G ). Among the genes downregulated under salt stress, several encoded transcription factors belonging to the NAC, MYB, ethylene-responsive factor (ERF), AP2/EREBP, basic helix-loop-helix (bHLH), and bZIP families ( Supplementary Figure 10H ). These results demonstrate that salt stress primarily affects the physiological and metabolic processes of plants by regulating genes involved in secondary metabolite biosynthesis, metabolic pathways, and plant hormone signal transduction.

To further investigate the relationship between OsMADS31 and DEGs, we performed WGCNA. WGCNA is a systematic biological method used to analyze inter-gene relationships in expression data by constructing weighted networks based on co-expression patterns for identifying gene modules associated with specific phenotypes. In this study, WGCNA was conducted using 17,695 genes with relative expression levels > 1. The results revealed 20 distinct modules, with each module containing 3 to 7,836 genes. Among these, four modules (black, brown4, magenta, and pink) showed significant associations ( Supplementary Figure 11 ). The identification of these modules provides crucial insights for understanding the mechanism of OsMADS31-mediated salt stress tolerance in rice.

Gene regulatory network analysis enables the identification of hub genes within the regulatory network and facilitates functional prediction of unknown genes based on known gene functions. GO and KEGG enrichment analyses demonstrated that these modules were primarily involved in biosynthesis of secondary metabolites and metabolic pathways, consistent with previous KEGG enrichment findings. Although each module exhibited distinct KEGG pathways, a substantial number of genes participated in several common pathways including Biosynthesis of secondary metabolites, Plant hormone signal transduction, Metabolic pathways, and Peroxisome-related processes ( Supplementary Figure 12 ). Co-expression network analysis was performed to identify hub (key) genes within each module and their interactions with genes in other modules ( Figure 6 ). The black module contained 200 genes, including four hub genes (Os03g0764600, Os03g0815100, Os04g0519700, and Os01g0863300), encoding an HHO transcription factor, a stress-responsive NAC transcription factor, an auxin response factor, and a MYB family transcription factor, respectively. For example, the core gene Os03g0815100 in the Black module encodes the stress-responsive NAC transcription factor (SNAC1). Studies have shown that overexpressing SNAC1 upregulates the expression of multiple stress-related genes, thereby enhancing salt tolerance in rice (Hu et al., 2006). The brown4 module comprised 100 genes, with three hub genes (Os05g0195200, Os04g0546800, and Os01g0797600) that encode a CCCH tandem zinc finger protein, an ethylene-responsive transcription factor, and an ERF, respectively. For example, the core gene Os01g0797600 (ethylene-responsive transcription factor ERF3/OsAP37) in the Brown4 module, when its encoded protein OsAP37 is overexpressed in transgenic rice plants driven by the OsCc1 promoter, exhibited enhanced salt tolerance during the vegetative growth stage (Oh et al., 2009). The magenta module included 200 genes, with eight hub genes (Os01g0203000, Os06g0127100, Os01g0915600, Os03g0135700, Os09g0468700, Os02g0603600, Os09g0434500, and Os06g0614100), encoding a BZR1 homolog, an AP2/EREBP transcription factor, a bHLH transcription factor, an ERF, a bZIP transcription factor, and two helix-loop-helix DNA-binding proteins. For example, the core gene Os06g0127100 (AP2/EREBP transcription factor) in the Magenta module encodes dehydration-responsive element-binding proteins 1C, 1E, and 1G (DREB1C, DREB1E, DREB1G), which promote tolerance to salt stress in rice (Wang et al., 2022b). Another core gene in the Magenta module, Os09g0434500 (ethylene-responsive factor OsBIERF), belongs to a family whose members OsBIERF1, OsBIERF3, and OsBIERF4 exhibit expression regulated by salt stress (Cao et al., 2006). The pink module consisted of 150 genes, including two hub genes (Os03g0758900 and Os01g0730700), both of which encode WRKY transcription factors. These hub genes play critical regulatory roles in processes such as plant hormone signal transduction, metabolic pathways, peroxisome function, and biosynthesis of secondary derivatives, suggesting their potential involvement in salt stress response mechanisms.

MADS-box proteins are a family of transcription factors characterized by a conserved MADS-box domain of approximately 58–60 amino acids at the N-terminus, which primarily functions in DNA binding (West et al., 1997; Yanofsky et al., 1990). In plants, MADS-box transcription factors are involved in nearly all critical growth and developmental processes (Alvarez-Buylla et al., 2000; Rounsley et al., 1995). Many MADS-box genes regulate plant development by interacting with other MADS-box proteins. Studies have shown that OsMADS29 in rice forms a heterodimer with its paralog OsMADS31 to cooperatively regulate seed development (Theißen and Becker, 2004; Nayar et al., 2014). In Arabidopsis, AGL24 and SVP participate in shoot development and floral transition by interacting with other proteins at different developmental stages (Michaels et al., 2003). A total of 61 MIKC-type MADS genes have been identified in rice, exhibiting highly conserved amino acid residues ( Supplementary Figure 1 ). Conserved gene sequence and domain analyses revealed that all possess MADS and K-box structures, along with sequence signatures of 12 conserved amino acid motifs ( Supplementary Figure 2 ). Comparative phylogenetic analysis indicated that the rice OsMADS31 gene is closely related to homologs in maize ( Supplementary Figure 3 ). Most MIKC-type MADS genes in plants are Type II and exhibit plant-specific features (Nam et al., 2003). The conservation and specificity of the MIKC-type MADS domain provide critical insights for further functional studies in plant development and stress responses. OsMADS31 is localized to the nucleus ( Figure 2 ). RT-qPCR and GUS staining demonstrated its widespread expression across multiple rice tissues. It responds to salt stress, with expression levels under 6‰ NaCl treatment reaching 10-fold higher than baseline after 24 hours ( Figure 1 , Supplementary Figure 4 ). Knockout of osmads31 led to significant reductions in agronomic traits, while overexpression promoted increased panicle length and grain number ( Supplementary Figure 6 & 7 ). Analyses of the conserved specificity of the MIKC-type MADS gene domain and its expression patterns suggest that OsMADS31 acts as a positive regulator of rice development and stress responses.

The seed germination and seedling growth stages of rice are periods during which rice are most sensitive to external environmental factors (Zhang et al., 2014). Therefore, rice seed germination and seedling growth are often used as key indicators for evaluating salt stress tolerance (Yu et al., 2018). The MADS-box transcription factor AGL21 is a negative regulator of seed germination and post-germination growth and thus, plays a role in seed germination under abiotic stress. AGL21-overexpressing plants exhibit hypersensitivity to ABA, salt, and osmotic stress during seed germination and early growth, whereas agl21 mutants show reduced sensitivity (Yu et al., 2017). In a GWAS study on salt stress during rice germination, the candidate gene OsMADS31, a member of the MADS-box transcription factor family, was predicted to contribute to salt tolerance during germination despite its downregulation under salt stress conditions (Yu et al., 2018). Using mRNA in identification of unannotated salinity stress-inducible transcripts in rice (Oryza sativa L.) demonstrated that OsMADS31 expression was upregulated in roots after salt stress (Mizuno et al., 2010). The osmads31 knockout reduced salt stress tolerance, inhibiting shoot and root growth during germination, while osmads31 overexpression promoted growth ( Figure 3 ). These results demonstrate that OsMADS31 significantly stimulates rice seed germination under salt stress. Rice is highly sensitive to salinity during early seedling growth (Nam et al., 2015), making improved salt tolerance crucial for seedling establishment and yield under stress. Under salt stress at the three-leaf stage, osmads31 knockout lines exhibited wilting and reduced survival rates, whereas osmads31overexpression enhanced survival ( Figure 4 ). Increases in reactive oxygen species (ROS) generated by oxidative stress can damage lipids, proteins, nucleic acids, and membranes. Plants counteract salt-induced oxidative stress by deploying antioxidants such as catalase (CAT), peroxidase (POD), and superoxide dismutase (SOD) (Liu et al., 2019). In this study, osmads31 knockout lines showed significantly lower CAT, POD, and SOD activities than the wild-type (WT), which was accompanied by reduced antioxidant enzyme activity and increased oxidative damage. By contrast, osmads31 overexpression lines displayed elevated CAT, POD, and SOD levels ( Figure 5 ), which increased antioxidant capacity and stress-responsive metabolite accumulation, facilitating ROS scavenging and mitigated membrane lipid peroxidation. These findings indicate that OsMADS31 strengthens antioxidant defenses, thereby reducing salt stress damage.

Reactive oxygen species (ROS), including H₂O₂, O₂ ^-, and hydroxyl radicals (•OH), are chemical compounds that can cause cellular membrane damage, lipid peroxidation, and malondialdehyde (MDA) production (D’Autreaux and Toledano, 2007). The MDA level serves as a common indicator of membrane lipid peroxidation and indirectly reflects plant stress resistance (Zhang et al., 2013). Overexpression of SlMBP11 in tomato enhanced salt tolerance, as evidenced by lower relative electrolyte leakage and reduced MDA content in transgenic plants (Guo et al., 2016), suggesting that MADS-box genes may participate in antioxidant systems to confer salt tolerance. Following salt stress treatment, OsMADS31-overexpressing (OE) lines showed significantly lower MDA levels than those of wild-type (WT) and knockout lines ( Figure 5E ), indicating reduced oxidative damage in OE plants under salt stress. Previous studies have demonstrated that salt stress can increase antioxidant enzyme activity while inducing ROS production and accumulation, thereby regulating plant salt tolerance (Miller et al., 2010; Wahid et al., 2007; Yan et al., 2014). Under salt stress, osmads31 knockout lines exhibited more intense DAB and NBT staining than the WT, while osmads31 overexpression lines showed minimal staining ( Figure 4E, F ). Additionally, osmolyte accumulation is a crucial response to salt stress, with proline being the most extensively accumulated compound (Wang et al., 2015). Proline scavenges ROS, stabilizes subcellular structures, and regulates redox homeostasis (Szabados and Savoure, 2010). The osmads31 knockout lines accumulated significantly less proline than the WT under salt stress, while osmads31 overexpression lines showed the opposite trend ( Figure 5D ). Physiological studies on rice salt tolerance have identified proline as a key osmoprotectant (Ozturk et al., 2021) that safeguards cellular proteins and membrane structures from degradation or damage, thereby improving seedling salt tolerance (Li et al., 2018) In this study, OsMADS31-OE lines displayed markedly increased proline accumulation under salt stress at the three-leaf stage ( Figure 5D ), suggesting that enhanced salt tolerance in OE plants may correlate with proline biosynthesis. Transcriptome analysis revealed that salt stress alters specific gene expression patterns involving three regulatory networks: protein kinases, catalases, and transcription factors. KEGG enrichment analysis highlighted carbon metabolism and amino acid biosynthesis/metabolism as major pathways affected by salt stress, together with glycolysis/gluconeogenesis, lipid metabolism, and plant hormone signal transduction ( Supplementary Figure 10 ). WGCNA identified differentially expressed gene modules participating in secondary metabolite biosynthesis/metabolic pathways, where hub genes played pivotal regulatory roles in plant hormone signaling, metabolic pathways, and peroxisome processes ( Figure 6 ), indicating their involvement in salt stress responses. Together, the results provide evidence that OsMADS31 improves salt tolerance by upregulating antioxidant-related genes, activating antioxidant enzymes, and reducing oxidative damage. However, the molecular mechanisms via which it exerts these effects will require further investigation, particularly with respect to how OsMADS31-mediated stress signaling pathways enhance crop salt tolerance.