The ddg1 mutant that exhibited dwarfism and delayed leaf senescence was isolated from the mutant library of the japonica rice variety Kitaake by 1% ethyl methanesulfonate (EMS) in 2017. For EMS treatment, we soaked dry seeds of Kitaake in water for 24 h, added 1% EMS to soak for 10 h, and soaked in water for 24 h, then sowed seeds in the field. The ddg1 mutant was crossed with the indica variety 93–11 to develop the gene mapping population. 521 extreme individuals selected from the F₂ population were used to map ddg1 for dwarfism and delayed leaf senescence. The RGA1^KO mutant used in this study was kindly provided by Professor Yidan Ouyang and Doctor Shengyuan Sun [17]. All the plants were grown in paddy fields in Huai’an (Jiangsu Province) and Lingshui (Hainan Province), China.

For salt treatment, we germinated seeds of both wild-type (WT, Kitaake) and ddg1 in water for 48 h at 37°C in darkness, and then transferred them to a Yoshida rice nutrient salt mixture (Coolaber, NSP1040, Beijing, China) for 2 weeks, exchanging fresh solution every 3 days. Next, the 2-week-old seedlings were transferred to a hydroponic culture solution supplemented with 150 mM NaCl for 7 d. Finally, the rice seedlings were grown in climate chambers under 14/10 h light/dark and 30°C/25°C day/night conditions.

At the heading stage, the first internodes of the WT and the ddg1 mutant were collected and fixed with 2.5% glutaraldehyde, and the processed samples were observed with a scanning electron microscope (JSM-840, JEOL).

Chlorophyll was extracted from leaves of wild-type and ddg1 at maturity. 0.1 g fresh leaves was cut and soaked in anhydrous ethanol for 48 h in darkness. The absorbance values at 649, and 665-nm wavelength were measured using a spectrophotometer. Chlorophyll contents were determined according to as previously described method [18].

For GA treatment, we grew seeds of wild-type and ddg1 in hydroponic media with different concentrations of GA for two weeks, and the seedling height of the WT and ddg1 were measured. For BR sensitivity test, leaves with partial leaf blades and sheathes were cut from the 2-week-old seedlings of the wild type and ddg1. Leaves were put into a solution with different concentrations of 24-Epibrassinolide (E1641, Sigma) or control solution, and then incubated in the light for 3 d. Finally, lamina joint angles were measured using IMAGEJ software.

For map-based cloning of DDG1, we chose 521 recessive individual plants showing dwarfism and delayed leaf senescence similar to the phenotype of the ddg1 mutant from an F₂ population derived from a cross between ddg1(♀) and the indica variety 93-11(♂). At heading stage, the mutant anthers were emasculated and subjected to the 9311 pollen. Simple sequence repeat (SSR) markers and the developed insertion/deletion (InDel) markers with Primer Premier 5 on chromosome 7 were used for fine mapping. Next, we amplified the corresponding fragments from ddg1 and the WT, and compared sequences using the BioXM 2.6 software.

To verify whether the Os07g0616000 mutation was responsible for the phenotype of ddg1, For constructing the Os07g0616000 complementary vector, the full-length coding sequence (CDS) was amplified and inserted into the p1390Ubi vector. Approximately 2 kb promoter sequence of Os07g0616000 was amplified and substituted with ubiquitin promoter region of p1390Ubi vector. The recombinant was transformed into the calluses from the ddg1 seeds by agrobacterium-infection methods. T₂ generation transgenic plants were used for phenotype analysis. The primers used in the map-based cloning and complementary vector construction were listed in Table S1.

Total RNAs were extracted using the TRIzol reagent (Invitrogen) from various tissues of the WT and ddg1, and cDNAs were synthesized using a HiScript^® II 1st strand cDNA synthesis kit (#R211; Vazyme, Nanjing, China). qRT-PCR experiments were performed using a real-time PCR detection system (BIO-RAD, CFX96) with three biological repetitions using the rice Ubiquitin (LOC_Os03g13170) gene as the internal control. The primers used for qRT-PCR were given in Table S1.

The full-length CDS of DDG1 and RGA1 (primers listed in Table S1) were amplified and inserted into the pGBKT7 vector and the pGADT7 vector, respectively. Different combinations of plasmids were also transformed into Saccharomyces cerevisiae strain AH109 on the SD/-Trp/-Leu/-His/-Ade medium. The specific method of yeast transformation can be found in the manufacturer’s instructions, which we followed carefully (http://www.clontech.com/).

For BiFC assays, we inserted the full length CDS of DDG1 and RGA1 (lacking stop codon) into the p2YN and p2YC at PacI and SpeI sites, respectively [19]. We next transformed the DDG1-nYFP and RGA1-cYFP plasmids or corresponding empty vectors into Agrobacterium strain EHA105, and then infiltrated this into leaves of N. benthamiana for 48 h. Fluorescences were then observed by laser confocal microscopy (Zeiss). The primers used for BiFC were listed in Table S1.

For LCI assays, the full-length CDS of DDG1 and RGA1 were amplified and fused to the pCAMBIA1300-nLUC and pCAMBIA1300-cLUC vector, respectively. These recombinants were co-transformed into N. benthamiana leaves by Agrobacterium infection. Empty pCAMBIA 1300-nLUC and pCAMBIA1300-cLUC vectors were used as negative controls. Primers used for LCI assays were listed in Table S1.

We isolated a dwarf and dark-green leaves mutant, ddg1, from an ethyl methanesulfonate (EMS)-derived mutagenesis population in a japonica rice variety, Kitaake. Compared with WT, ddg1 exhibited dwarfism and short panicles at maturity (Figs. 1A–1C). Each internode of ddg1 was almost shortened (Fig. 1B). The seeds of ddg1 were small and round (Fig. 1D).The plant height and panicle length of the ddg1 mutant decreased by 30% and 27.8% (Figs. 1E, 1F). We observed no difference in the tiller number between the wild type and ddg1 (Fig. 1G). However, ddg1 showed delayed leaf senescence and had higher chlorophyll content than that of the WT at maturity (Figs. 2A and 2B). Similarly, two senescence-associated genes, SGR and Osh36, were significantly repressed in ddg1 (Figs. 2C and 2D). These results suggest that DDG1 regulates plant height and senescence.

Normal cell proliferation and cell expansion are responsible for plant organ size. Hence, in order to examine the dwarfism of ddg1 in more detail, we examined the first internodes of the WT and ddg1 at the heading stage using histological analysis. Our SEM observations showed that cell length and width were similar between the wild type and the ddg1 mutant (Fig. S1), which suggests that the DDG1 mutation does not influence cell expansion.

Both GA and BR are essential for plant growth and developmental processes, including stem development, grain size, senescence, and seed germination [10]. In order to verify whether DDG1 is involved in the GA and BR signaling pathways, we tested the sensitivity of ddg1 to GA and 24-epibrassinolide. First, we cultured the WT and ddg1 seedlings with different concentrations of GA and measured the seedlings’ heights after 2 weeks of growth. As shown in Figs. 3A and 3B, the average seedling height of the wild type and ddg1 increased significantly with an increase in GA concentration, but the rate of increase of seedling height for the ddg1 mutant was significantly lower than that of the wild type, indicating that ddg1 was less sensitive to GA. Secondly, for BR sensitivity test, we used 0, 0.1, or 1 μM of 24-epiBL to treat excised leaf segments of the 2-week-old WT and ddg1 mutant plants and then measured the lamina joint angle. Compared with the wild type, the rate of increase in the lamina joint angle in ddg1 was remarkably reduced (Figs. 3C–3F). The above results indicate strongly that DDG1 has a direct and critical involvement in the GA and BR signaling pathways.

To isolate the DDG1 gene responsible for the ddg1 phenotype, we crossed ddg1 with indica var 93–11 to construct the mapping population. Using 521 extreme individuals selected from the F₂ population, we mapped DDG1 to a 140-kb region on rice chromosome 7 (Fig. 4A). We found that there were 18 genes in the 140-kb region (http://rice.plantbiology.msu.edu/) (Fig. 4A), and among these 18 genes, one was the DEP2 (Os07g0616000) gene, which regulates panicle architecture and grain size [20,21]. We then sequenced the promoter and coding region of DEP2 between the wild type and ddg1 via Sanger sequencing [22] and found that there was a one-nucleotide deletion of DEP2 in ddg1, compared with the wild type (Fig. 4B). One-nucleotide deletion of DEP2 in ddg1 led to premature termination of the DEP2 protein translation (Fig. S2). Next, in order to study further whether the DEP2 mutation was responsible for the defects in ddg1, we conducted a genetic complementation test. Approximately 2 kb promoter and full-length CDS of DEP2 were transformed into the calluses from the ddg1 mutant and found that the positive complementary plants were similar to the wild type (Fig. 4C), compared to the negative control plants (Fig. S3). Importantly, we found the specific mutant DDG1 gene to be DEP2/Os07g0616000.

To examine the expression pattern of DDG1, we first learned that DDG1 was expressed in both vegetative and reproductive tissues by using the RiceXPro database (Fig. S4; https://ricexpro.dna.affrc.go.jp/). In addition, we performed qRT-PCR to analysis the expression pattern of DDG1. As shown in Fig. 5A, DDG1 was expressed in roots, stem, leaves, panicles, and seeds. Interestingly, we observed that the expression level of DDG1 was significantly repressed in ddg1 (Fig. 5B).

DDG1 encodes a plant-specific protein without any known functional domain and that has 10 exons. To analyze the evolutionary relationship of the DDG1 protein, we performed a BLAST search in order to identify DDG1 homologs and found that DDG1 showed high amino acid sequence similarity to proteins in other species, including Oryza brachyantha (88%), Brachypodium distachyon (73%), Triticum dicoccoides (73%), and Panicum virgatum (68%; Fig. S5).

Since the DDG1 protein has no known functional domain, we performed Y2H screening to identify its interacting proteins. To accomplish this, we fused the full-length CDS of DDG1 to the pGBKT7 vector (DDG1-BD) and used it as bait to screen the interacting proteins against a rice leaf yeast library of 2-week-old Kitaake, and we were able to identify one candidate segment belonging to part of the RGA1 gene. After this result, we fused the full-length CDS of RGA1 to the pGADT7 vector (RGA1-AD). Next, DDG1-BD and RGA1-AD were co-transformed into the yeast cells, which grew well on our quadruple dropout media (Fig. 6A; Fig. S6). We then performed bimolecular fluorescent complementation assays to detect the interaction. As shown in Fig. 6B, the fluorescence signal was observed when the fusion constructs p2YN-DDG1 and p2YC-RGA1 were present, but not in the corresponding control. Using a LCI assay, we detected obvious LUC activity signal when DDG1-nLUC was co-expressed with RGA1-cLUC in N. benthamiana (Fig. 6C). Thus, we conclude that DDG1 interacts with RGA1. In addition, compared with the wild type, knock-out of RGA1 showed dwarfism, short panicles, and darker-green leaves, which were similar to the phenotype of ddg1 (Fig. 7).

Microarray analysis showed that 889 up-regulated and 841 down-regulated genes related-salt stress between control and the RGA1 mutant d1, indicating that RGA1 may be involved in the regulation of salt stress in rice [23]. Therefore, we speculated that DDG1 might also be involved in the regulation of salt stress response in rice. To test this hypothesis, we treated the ddg1 mutant and wild type seedlings with 150 mM NaCl. Compared with the wild type, ddg1 seedlings were more sensitive to salt stress, and the survival frequencies of were approximately 22% and 62% under salt stress conditions in ddg1 and WT plants, respectively (Figs. 8A–8C). Consistent with the salt sensitivity of ddg1, knock-out of RGA1 resulted in a lower survival rate than that of wild-type plants (Fig. S7). Compared with ZH11, the expression level of RGA1 in knock-out of RGA1 plants was decreased (Fig. S8). These results showed that DDG1 and RGA1 both play important roles in rice salt tolerance.