Endosperm

We used the number of ESTs representing specific transcripts isolated from 19 rice tissue sources (i.e. callus, suspension cells, seedling, leaf, shoot, root, stem, sheath, phloem, panicle, flower, anther, pistil, endosperm, immature seed, mixed tissues, mature seed, whole plant, and unknown samples) to estimate gene expression levels in the different tissues. The EST evidence was analyzed using the Program to Assemble Spliced Alignments (PASA) software, which utilizes a number of alignment programs to maximally align transcripts to the genome as introduced by Haas et al. [83] (link).

Read counts per gene were obtained from the RNA‐seq alignments using HTSeq (v0.9.1) (Anders et al., 2015) in unstranded union mode. Differentially expressed genes were identified from the gene read counts using DESeq2 (Love et al., 2014) with an alpha level of 0.01. For organ‐specific gene expression, an organ was defined similarly to Sekhon et al. (2011) with some modifications (Sekhon et al., 2011). The endosperm and embryo were grouped into a single organ (seed) and the cob, silk, tassel, and anthers were grouped into one reproductive organ. The SAM was also considered a separate organ establishing six organs for analyses: internode, leaf, reproductive, root, seed, and SAM. Each organ from the developmental gene atlas was individually contrasted against the other five organs retaining genes with an adjusted P < 0.01 and a log₂ fold change >2. A gene was considered organ‐specific if it met these criteria in the organ of interest when compared against all other organs. Tissue‐specific gene expression was similarly characterized for seed and reproductive tissues. Using read counts for genes previously identified as seed‐ or reproductive‐specific, the seed and reproductive tissues were contrasted against the other tissues in its respective organ. A gene was considered tissue‐specific if it had an adjusted P < 0.01 and a log₂ fold change >2 in all tissue comparisons. For stress‐related differential expression, each treatment was contrasted against its respective experimental control, retaining genes with an adjusted P < 0.01 and a log₂ fold change <−2 or a log₂ fold change >2. A gene was considered DE under both abiotic and biotic stress if it met these criteria in at least one treatment from both stress types. The organ‐specific heatmap was generated with the R package ‘gplots’ (Alexa and Rahnenfuhrer, 2016) ‘heatmap.2’ function and all other plots were generated with the R packages ‘ggplot2’ (Wickham, 2009) and ‘RColorBrewer’ (Neuwirth, 2014).

Histochemical GUS staining assay was performed using a substrate buffer: 100 mM sodium phosphate (pH 7.0), 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 1 mM EDTA, 1% Triton-X, 1 mg/mL X-Gluc for embryo and 5 mg/mL for endosperm samples. Samples were incubated at 37 °C for 16 h and washed after the staining 3 times with 75% ethanol.

PIF3 recombinant proteins (full-length) were prepared using PIF3-his DNA (pET28C, Novagen) provided by Ferenc Nagy and induced and purified using commercial kit (HisTrap Fast Flow, GE17-5319-01). Polyclonal anti-PIF3 was obtained from rabbits immunized with PIF3 recombinant protein (Cocalico Biologicals Inc., USA). PIF3 antibodies were further affinity-purified using PIF3 recombinant protein immobilized on nitrocellulose filters as described^{6 (link)}. Endosperms and embryos were homogenized in presence of protein extraction buffer (50 mM Tris-HCl pH 6.8, 2% SDS, 10% glycerol, 100 mM DTT, 0.01% bromophenol blue) and protein extract corresponding to 150 endosperms or 15 embryos was loaded per lane. Proteins were separated on 10% SDS-PAGE gel and transferred to a PVDF membrane (Amersham). Membranes were incubated in 5% milk powder in Tris-buffered saline (TBS) containing PIF3 antibody (1:200 dilution), RGL2 antibody (1:250 dilution)^{6 (link)} or ABI5 antibody (1:2000 dilution)^{19 (link)} for 16 h at 4 °C. Anti- UGPase antibody (Agrisera) was used at 1:10,000 dilution and anti-PHYB antibody and used at 1:1,000 dilution for 16 hours at 4 °C. Anti-rabbit (for PIF3, RGL2, ABI5 and UGPase) or anti-mouse (for PHYB) IgG HRP-linked whole antibody (GE healthcare) in a 1:10,000 dilution was used as secondary antibody for 2 h at RT. Membranes were washed with TBS + 0.05% Tween 3 times for 10 min after the first and second antibody incubation and the immune complexes were detected using the ECL kit (Amersham). All the protein accumulation data shown in main figures and supplementary figures arise from the same blots. Protein signals were quantified digitally using the ImageJ software (v 2.0) and using UGPase signal as a normalization factor. For each blot all values are relative to the strongest signal found in the blot which is arbitrarily set to 1. Western blot repetitions can be found in Supplementary Fig. 8. Where portions of blots have been presented in the main figures, the full blots are shown in the Supplementary Fig. 9.