*4.5. Total RNA Extraction and RNA Sequencing*

Total RNA was extracted from each sample using an RNAprep Pure Plant Kit (Tiangen Biotech, China) following the manufacturer's instructions. The quality and quantity of RNA samples were assessed by gel electrophoresis and a Nanodrop (Thermal Fisher, Waltham, MA, USA). The resultant 15 RNA samples for SDWG005 were used for RNA-seq analysis. Upon treatment with DNase I, 1 μg of RNA from each sample was used for sequencing library preparation with a NEB Next Ultra TM RNA Library Prep Kit for Illumina (NEB, Ipswich, MA, USA). Library preparations were sequenced from paired ends to obtain 150 bp reads (PE150) on an Illumina platform.

#### *4.6. Quality Control and Comparative Analysis of RNA-seq Data*

Raw data were processed through in-house Perl scripts to remove low-quality reads (more than 20% of bases presenting a Q value ≤ 20 or ambiguous sequence content ("N") exceeding 5%). Clean reads were then obtained by removing reads containing adapters and poly-N sequences. Clean reads were mapped to the rice reference genome sequence (MSU7.0, http://rice.plantbiology.msu.edu/) with Hisat2 tools [40]. Only reads with a perfect match or one mismatch were further analyzed.

#### *4.7. Di*ff*erential Expression Analysis*

Gene expression levels were quantified as fragments per kilobase of transcript per million fragments mapped (FPKM) values. Pearson's correlation coefficient [41] was used to calculate the correlation coefficients of FPKM of all transcripts among three replications for each sample. Differential expression analysis of the two conditions was performed using DEseq2 [42]. The resulting *p* values were adjusted using the Benjamini and Hochberg approach [43] to control the false discovery rate. Genes with an adjusted *p* < 0.05 and |log2FC| > 1 identified by DEseq2 were assigned as differentially expressed. Hierarchical clustering analysis of DEGs was conducted using cluster *R* package based on lgFPKM of them among three replications for each sample (https://cran.r-project.org/web/packages/cluster). Heatmap was plotted by pheatmap *R* package (https://cran.r-project.org/web/packages/pheatmap), Venn was plotted by Venndiagram [44] *R* package. Expression pattern of genes in different clusters were plotted using our in house scripts based on *R* language (File S1).

### *4.8. Quantitative Real-Time PCR*

Quantitative real-time PCR (qRT-PCR) was applied to quantify the expression levels of 10 selected DEGs in SDWG005 and 9311. RNA for all the samples was reverse-transcribed using a RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific) following the manufacturer's protocol. qRT-PCR was performed using ABI StepOnePlus™ Real-Time System. Relative gene expression levels were calculated using the 2−ΔΔ*C*<sup>t</sup> method, as described by Min et al. [45]. The expression of each replicate was normalized according to the reference gene *OsActin* (LOC\_Os03g50885). For randomly selected genes, specific primers (listed in Table S4) were designed and tested by qRT-PCR. The mean of three replicates represents the relative expression level.

#### *4.9. Gene Functional Annotation and Enrichment Analysis*

Gene functions were annotated based on the following databases: Nr (non-redundant protein sequence database), Pfam (database of homologous protein family), KOG/COG (Clusters of Orthologous Groups of proteins), Swiss-Prot (manually annotated, non-redundant protein sequence database), KO (KEGG Ortholog) and GO (Gene Ontology database). GO enrichment analysis of the DEGs was implemented with clusterProfiler *R* packages [46]. We defined significant enrichment based on an FDR < 0.05.
