To screen for genes involved in the regulation of leaf angle in rice, we created a library of ⁶⁰Co-γ irradiation-induced mutants from which we isolated a mutant with semidwarf and leaf erectness (Figures 1A,B), reduced panicle and leaf length phenotypes (Supplementary Figures 1B,C), based on which the mutant was named regulator of leaf angle (rela). We crossed the rela mutant with the wild-type and found that the ratio of wild-type to mutant in the segregating population was 415:134, close to 3:1, indicating that rela is a recessive mutation controlled by a single nuclear gene. To further reveal the reason for the reduced leaf angle of the rela mutant, we performed a transverse section of the lamina joint and found that the number of sclerenchyma cell layers and the cell area on the abaxial side of the rela mutant were increased compared to those of the wild-type (Figures 1C,F,G). Longitudinal section revealed that the length of parenchyma cells on the adaxial side of the rela mutant was shorter than that of the wild-type (Figures 1D,E). In addition, longitudinal section of the uppermost internode revealed that the cell length in the rela mutant was significantly shorter than that of the wild-type (Supplementary Figures 1A,D–F). These results suggest that OsRELA promotes the elongation of cells in the culm and parenchyma cells on the adaxial side and inhibits the division and expansion of sclerenchyma cells on the abaxial side of the lamina joint, thus positively regulating the leaf angle and plant height.

To clone the gene encoding OsRELA, we crossed the rela mutant with Dular and obtained a segregating population of the F₂ generation. By linkage analysis, OsRELA was narrowed between two genetic markers, Indel 7–10 and Indel 7–12, located on the long arm of chromosome 7. By further fine mapping using 2871 F₂ generation plants, we targeted the OsRELA gene between SNP 25.338 M, and Indel 25.402 M, and six genes were contained in this 64-kb interval: Os07g0615200, Os07g0615400, Os07g0615500, Os07g0615800, Os07g0616000, and Os07g0616200 (Figure 2A). Further genome sequencing revealed a 28-bp deletion in the seventh exon of the Os07g0616000 gene, which shifted the reading frame and eventually led to a premature stop codon, while the rest of the genes were sequenced without mutations (Figure 2B). Our immunoblotting experiments using a specific polyclonal antibody that recognizes the protein encoded by Os07g0616000 showed that the expression of the protein encoded by Os07g0616000 was not detected in the rela mutant (Figure 2C), indicating that the phenotypes of the rela mutant are most likely due to a 28-bp deletion in the Os07g0616000 gene. Coincidentally, OsRELA is located at the same gene locus as the previously reported DENSE AND ERECT PANICLE 2. DEP2 was revealed to be associated with panicle development (Li et al., 2010), and we also found this phenotype in the rela mutant; however, the biological function of OsRELA involved in the regulation of leaf angle has not been elucidated. Phylogenetic tree analysis revealed that OsRELA belongs to a Spermatophyta-specific gene family, and its homologs are distributed in Gymnosperm, Amborella, Eudicots, and Monocots. However, the gene family has fewer members in Eudicots, and there are three copies in Arabidopsis: AT3G14172, AT1G72410, and AT1G17360. Most of the family members are distributed in the Monocots, and there is only one copy in rice (Supplementary Figure 2).

To further clarify the function of OsRELA in regulating leaf inclination, we constructed the pOsRELA:OsRELA vector and transformed it into the rela mutant. The leaf angle of transgenic lines expressing OsRELA was restored to the wild-type state (Figure 2D). We also obtained genome-edited lines of Os07g0616000 using the clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR associated protein 9 (Cas9) genome editing tool (Supplementary Figure 3). In all genome-edited homozygous lines obtained, the leaf angle was significantly smaller than that of the wild-type line, consistent with the rela mutant phenotype (Figure 2E). These genetic results suggest that OsRELA is the key gene that regulates leaf angle in rice.

To determine the localization of OsRELA in rice cells, we transformed the 35S:OsRELA-GFP vector into rice protoplasts for transient expression. Confocal laser scanning microscopy revealed that the fluorescence of the OsRELA-GFP fusion protein merged with the nuclear marker mCherry-NLS signal (Figure 3A). Consistently, GFP fluorescence of the GFP-OsRELA fusion protein was colocalized with the nuclear stain DAPI in the rela/pOsRELA:GFP-OsRELA lines (Figure 3B), suggesting that OsRELA functions mainly in the nucleus.

To further analyze the expression pattern of OsRELA, we performed real-time quantitative PCR (RT–qPCR) to measure the OsRELA expression level, and the results showed that OsRELA was expressed in various tissues of rice (Figure 3C). GUS staining experiments performed on pOsRELA:GUS transgenic plants revealed strong GUS staining in the lamina joint, young florets, stem, vascular bundles, leaf sheath and root (Figure 3D). All these tissues were associated with the phenotypes of the rela mutant.

To analyze OsRELA-regulated genes that participate in the regulation of leaf angle, we performed RNA-seq analysis on the rela mutant and identified 285 genes with significantly different expression levels compared to the wild-type, including 156 upregulated genes and 129 downregulated genes (Figure 4A). We next performed a hierarchical clustering analysis of all genes that were significantly differentially expressed, and the representative top 50 most significant genes are shown (Figure 4B). In addition, gene ontology (GO) enrichment analysis demonstrated that the molecular functions of the differentially expressed genes were mainly involved in DNA binding (GO:0043566), suggesting that OsRELA is likely to be involved in transcriptional regulation by regulating DNA binding (Figure 4C). We then examined the genes with significant expression changes in the wild type and rela by RT–qPCR. Consistent with the RNA-seq results, the expression of OsEXPA5, OsEXPB6, and OsEXPB11 was significantly upregulated, while the expression of OsmiR408, OsmiR528, ILA1 interacting protein 3, OsOFP14, OsILI1, OsILI4 (OsBU1), and OsILI5 (OsBUL1) was significantly downregulated in the rela mutant (Figure 4D), and the expression of seven genes related to cell wall synthesis was also significantly suppressed in rela (Figure 4E). It is worth noting that ILA1 interacting protein 3 (OsIIP3), OsILI1, OsILI4 (OsBU1), and OsILI5 (OsBUL1) were reported to positively regulate leaf angle (Tanaka et al., 2009; Zhang L.Y. et al., 2009; Ning et al., 2011; Jang et al., 2017). Expansin family genes and cell wall synthesis genes were involved in the process of cell wall loosening and cell extension (Cho and Kende, 1997; Kuluev et al., 2014). The phenotypes of the rela mutant are similar to those of BR-deficient and BR-insensitive mutants (Yamamuro et al., 2000; Hong et al., 2003), and both Expansins and ILI family genes have been reported to be downstream of BR signaling and involved in the regulation of leaf angle (Wang et al., 2018), implying that OsRELA may intersect with the BR signaling pathway to regulate leaf inclination in rice.

To further reveal the relationship between OsRELA and the BR signaling pathway, we obtained OsLIC from a rice cDNA library by a yeast two-hybrid approach using full-length OsRELA as bait (Figure 5A). OsLIC negatively regulates leaf angle and acts as an antagonistic transcription factor of OsBZR1 (Zhang et al., 2012). However, OsRELA does not interact with other BR signaling components, such as OsGSK2, DLT, OsBZR1, and RLA1, in yeast (Supplementary Figure 4). Bimolecular fluorescence complementation (BiFC) assays showed that OsRELA and OsLIC interact in rice cells (Figure 5B), and the semi-in vivo pull-down assay results also indicated that the OsLIC-GST fusion protein interacts with OsRELA (Figure 5C).

To further clarify the genetic relationship between OsRELA and OsLIC, using the CRISPR/Cas9 system, a lic-2 mutant line was obtained by targeting the second exon of the OsLIC gene, and DNA sequencing showed that a 1 bp insertion in the second exon of OsLIC resulted in premature termination of translation (Supplementary Figure 5). The lic-2 mutant showed an increased leaf and tiller angle (Wang et al., 2008). We then crossed rela with the lic-2 mutant and found that the phenotypes of the rela/lic-2 double mutant were similar to those of the rela mutant (Figure 5D), implying that the function of OsLIC is dependent on OsRELA. Quantitative real-time PCR results showed that the expression level of OsILI1, the direct downstream target of OsLIC, was reduced in both the rela and rela/lic-2 mutants (Figure 5E). The elevated expression level of OsILI1 and the increased leaf angle in the lic-2 mutant are consistent with what has been reported previously, suggesting that the regulation of OsILI1 expression by OsLIC is dependent on OsRELA.

To clarify the minimal domain of OsRELA that interacts with OsLIC, we performed sequence and bioinformatics analysis on OsRELA. We found a speckle-type BTB/POZ domain at the N-terminus of OsRELA by Bioinformatics Toolkit analysis, which is a multifunctional protein–protein interaction motif (Bardwell and Treisman, 1994). Two COILED-COIL domains were predicted by the ExPASy tool kit: 697–724 aa and 798–849 aa. The COILED-COIL domain plays an important role in mediating protein–protein interactions (Burkhard et al., 2001). A nuclear localization sequence was found in its conserved C-terminal domain by SeqNLS software: 1240–1256 aa (Supplementary Figures 6, 7).

We next created different truncated forms of OsRELA attached to the GAL4-BD domain named BD-OsRELA (1–1365 aa), BD-OsRELA-T1 (1–199 aa), BD-OsRELA-T2 (200–643 aa), BD-OsRELA-T3 (1173–1365 aa) and BD-OsRELA-T4 (1–1172 aa) (Figure 6A). After cotransforming yeast with the OsLIC-AD vector, we found that both OsRELA and OsRELA-T3 interacted strongly with OsLIC, while OsRELA-T1, OsRELA-T2, and OsRELA-T4 did not interact with OsLIC (Figure 6B). The above results suggest that OsRELA interacts with OsLIC through its conserved C-terminal domain. We further used the promoter of OsILI1 to drive luciferase (LUC) as a reporter and OsRELA-FLAG, OsRELAΔC-FLAG, OsLIC-FLAG, and FLAG as effectors for the transient expression assay (Figure 6C). The results showed that OsLIC represses OsILI1 expression, and when OsRELA-FLAG was cotransformed with OsLIC-FLAG into rice protoplasts, the expression level of OsILI1 was elevated. When OsRELAΔC-FLAG was cotransformed with OsLIC-FLAG, OsILI1 expression was repressed (Figure 6D). These results indicate that OsRELA interacts with OsLIC through its conserved C-terminal domain and thus abolishes the transcriptional repression of OsLIC on OsILI1.

Finally, we genetically analyzed the effect of the C-terminal domain of the OsRELA protein on its function and found that deletion of the C-terminal domain could not rescue the phenotypes of the rela mutant (Figure 6E). Moreover, quantitative real-time PCR results revealed that the expression level of OsILI1 was not recovered in the CO-ΔC lines (Figure 6F), indicating that the regulation of OsRELA on OsILI1 expression through OsLIC also depends on its C-terminal domain.

In this study, Nipponbare rice (Oryza sativa subsp. japonica) and Dular rice (Oryza sativa L. ssp. indica) were used as wild-type controls, and the rela mutant was obtained from EMS-induced mutant populations. The F₂ segregation population used for genetic segregation analysis was obtained by crossing the rela mutant with Nipponbare. The F₂ segregant population used for map-based cloning was obtained by crossing the rela mutant with Dular. The rice materials used in this study were grown in the fields of the Chinese Academy of Agricultural Sciences under natural conditions.

After the mature lamina joints of the wild-type and rela mutant were separated, they were rapidly fixed in formaldehyde-acetic acid-alcohol (FAA) at 4°C for 72 h. Subsequently, the samples were dehydrated through a gradient ethanol solution. The dehydrated samples were subjected to a xylene:ethanol clear solution of 1:2, 1:1, and 2:1 and finally transitioned to xylene. The samples were then incubated in a xylene:Paraplast Plus solution (Sigma–Aldrich) solution of 1:1 at 37°C for 48 h. The samples were embedded in Paraplast Plus for 3 days. The samples were then sectioned into 8-μm-thick sections by an RM2245 rotary microtome (Leica). After the removal of Paraplast Plus with xylene:ethanol ratios of 2:1, 1:1, and 1:2, the sections were rehydrated with gradient ethanol, followed by safranin-fixed green staining. Images were taken with an ICC50 HD microscope (Leica), and the cell length and cell area were subsequently measured using ImageJ.

A total of 154 primer pairs exhibiting polymorphisms between Nipponbare and Dular were identified from our primer library. We targeted OsRELA between two genetic markers, Indel 7–10 and Indel 7–12, located on the long arm of chromosome 7. By further fine-mapping using the F₂ generation population of 2871 progeny, we targeted the OsRELA gene between two newly developed genetic markers. We then performed PCR amplification and sequencing of the genomic sequences of all genes within this 64-kb interval in wild-type and rela mutants.

For complementation assays, the putative 2.2-kb promoter region of OsRELA was amplified using the OsRELA-Pro primer pairs, and the coding region was amplified using the OsRELA-CDS and OsRELA-CDSΔC primers and inserted into the pCAMBIA1300 binary vector using EcoRI/HindIII restriction sites.

To knock out the OsRELA and OsLIC genes, we selected 20-bp gene-specific spacer sequences in the coding region and cloned them into the sgRNA-Cas9 vector to obtain OsRELA-CRISPR and OsLIC-CRISPR vectors.

To obtain OsRELA overexpression lines, the promoter of OsRELA was amplified by primer OsRELA-Pro, and the full-length CDS was amplified by using primer OsRELA-GFP and inserted into the pCAMBIA1300 binary vector using EcoRI/HindIII restriction sites to obtain proOsRELA:GFP-OsRELA.

To determine the expression pattern of OsRELA, a 2.2-kb region upstream of the ATG start codon was amplified using OsRELA-GUS primers, and the PCR product was cloned into the pCAMBIA1391z binary vector using HindIII/EcoRI restriction sites to obtain the proOsRELA:GUS vector.

The above vectors were subsequently transformed into wild-type and rela mutant via Agrobacterium-mediated transformation.

An in vitro purified OsRELA polypeptide (Os07g0616000, 500–740 aa) fragment was injected into rabbits to generate the corresponding polyclonal antibody at ABclonal Technology. The specificity of the polyclonal antibody against OsRELA was subsequently verified by immunoblotting assays using wild-type rice seedlings. Samples were well ground in liquid nitrogen and then incubated with protein extraction buffer (50 mM Tris-HCl at pH 7.5, 150 mM NaCl, 10 mM MgCl₂, 1 mM EDTA, 10% glycerol, and protease inhibitor cocktail) at 4°C for 15 min. After centrifugation at 4°C for 10 min, the supernatant was boiled for 10 min and separated on a 6% SDS–PAGE gel. The anti-OsRELA polyclonal antibody was diluted at a concentration of 1:2000 and detected by chemiluminescence reagent (GE Health care), and the Rubisco large subunit was used as the endogenous control.

To determine the subcellular localization of the OsRELA protein, the 4095-bp coding region was cloned into the pAN580 vector using the XbaI restriction site. The 35S:OsRELA-GFP vector was subsequently introduced into rice protoplasts as previously described (Zhang et al., 2011), and the fluorescence of GFP was visualized using a confocal laser scanning microscope (LSM 980; Zeiss).

Different tissues of proOsRELA:GUS transgenic plants were incubated in GUS staining buffer (1 M sodium phosphate, pH 7.0, 0.5 M EDTA, pH 8.0, 10% Triton X-100, 50 mM K3Fe(CN)₆, 0.1 M X-Gluc) at 37°C after isolation. After 12–16 h, the samples were decolorized using 75% ethanol. The pictures were photographed using an ICC50 HD microscope (Leica).

The sequences of OsRELA and OsLIC were obtained from The Rice Annotation Project¹. Homologous of OsRELA in plants were obtained from the NCBI BLAST server². Multiple sequence alignment was performed using DNAMAN9 with default settings. A neighbor-joining tree of OsRELA and its homologs was generated by MEGA7 with 1000 bootstrap replicates. The protein structure analysis was performed using the following software:

HHpred: https://toolkit.tuebingen.mpg.de/tools/hhpred

COILS: https://embnet.vital-it.ch/software/COILS_form.html

SeqNLS: http://mleg.cse.sc.edu/seqNLS/

Total RNA was extracted from lamina joints of rice seedlings using TRIzol reagent (Invitrogen). First-strand cDNA synthesis was performed using the PrimeScript 1st strand cDNA Synthesis Kit (6110A). RT–qPCR was performed as previously described (Zhou et al., 2008), and the rice ubiquitin gene was used as an internal control.

The differential gene expression analysis of 2-week-old rice seedlings of wild-type and rela mutant were performed by Allwegene Technology Inc³.

The full-length coding region of OsRELA was cloned into the pGBKT7 vector as bait, and then yeast two-hybrid screening was performed with a rice cDNA library. The positive clones were subsequently identified by sequencing. The full-length coding regions of OsRELA, OsLIC, OsBZR1, RLA1, DLT, and OsGSK2 were cloned into the corresponding pGBKT7 and pGADT7 vectors using EcoRI/BamHI restriction sites. The corresponding bait and prey constructs were cotransformed into the yeast strain AH109 (Clontech) and grown on the selective medium SD-Trp/-Leu. The positive clones were then transferred to the selective medium SD-Trp/-Leu/-His/-Ade.

The full-length coding region of OsRELA was amplified and cloned into the pSPYNE vector using EcoRI/SalI to obtain NY-OsRELA, and the full-length coding region of OsLIC was amplified and cloned into the pSPYCE vector using EcoRI/SalI to obtain OsLIC-CY. OsRELA-NY/OsLIC-CY, OsRELA-NY/pSPYCE, and pSPYNE/OsLIC-CY were cotransformed into rice protoplasts by PEG-mediated transformation. The fluorescence of YFP was visualized using a confocal laser scanning microscope (LSM 980; Zeiss).

To test the interaction between OsRELA and OsLIC, the full-length coding region of OsLIC was amplified and cloned into pGEX-4T-1 using EcoRI/XhoI, and the vector was subsequently transformed into E. coli strain BL21 to obtain the OsLIC-GST fusion protein. Two-week-old wild-type seedlings were ground in liquid nitrogen and solubilized with 2X protein extraction buffer [100 mM Tris-HCl, pH 7.5, 300 mM NaCl, 2 mM EDTA, pH 8.0, 1% TrionX-100, 10% (v/v) glycerol, and protease inhibitor mixtures (M307; AMRESCO)]. The semi-in vivo pull-down assay was performed as previously described (Qiao et al., 2017).

In the dual-luciferase assay, the 2.2-kb promoter region of OsILI1 was cloned upstream of the firefly luciferase gene (LUC) in pGreenII 0800-LUC to generate the OsILI1-LUC reporter vector. OsLIC-FLAG, OsRELA-FLAG, OsRELAΔC-FLAG, and FLAG were cloned into pGreenII 62-SK as effectors using XbaI/HindIII restriction sites. The combined reporter and effector vectors were cotransformed into rice protoplasts by PEG-mediated transformation. Renilla reniformis (Ren) driven by the 35S promoter was used as an internal control. The LUC activity was quantified with a Dual-Luciferase Assay Kit (Promega) following the manufacturer’s recommendations, and the relative LUC activity was calculated as the ratio of LUC/Ren.