Regulator of Awn Elongation 3, an E3 ubiquitin ligase, is responsible for loss of awns during African rice domestication

Significance Selection for a common domestication trait targeted different genes in Asian and African rice. We identify an E3 ubiquitin ligase named Regulator of Awn Elongation 3 (RAE3) that causes awnlessness in African rice and demonstrate its genetic relationship with other genes. Loss of function of RAE3 leads to awnlessness even when other awn genes (RAE1 and RAE2) are functional; that is, RAE3 is a key gene for awn elongation in rice. Diversity analysis shows that while the dysfunctional rae3 allele is fixed across cultivated African rice, it is not found in wild African rice or in Asian rice. The discovery of RAE3 simultaneously deepens our understanding of awn developmental pathways and lends insight into the complex processes underlying crop domestication.

pBluescript vector. DNA fragments containing each gene were transferred into the pCAMBIA1380 vector carrying the ubiquitin promoter upstream of the multiple-cloning site. Only the line carrying the Os06g0695900 construct could complement the awned phenotype (8 positive transgenic lines out of 12 in total transgenic calli) (Fig. 2e). Six lines showed awned phenotype and the quantitative data are found in Fig. S3 and Table   S1. To examine the function of OsRAE3 with or without mutation in RING-H2 domain in rice, the Os06g0695900 CDS was cloned into pCAMBIA1300 carrying 3× FLAG on the  Table S1. To observe RAE3 cell localization, the Os06g0695900 CDS (OsRAE3(WT)) and mutated OsRAE3, OsRAE3(C136S) was cloned into the pENTR/SD/D-topo vector (Invitrogen, Waltham, MA, USA). OsRAE3(C136S)/pENTR was generated through site directed mutagenesis PCR using a specific primer pair (KM154-KM155) with OsRAE3(WT)/pENTR as the template. Then genes cloned into pENTR were transferred into pEG101 by using Gateway LR clonase II (ThermoFisher Scientific, MA, USA) by following manufactured protocol. The constructs in pEG101 were used for transient expression analysis in onion epidermal cells and rice protoplast.
Primers. The primers used in this study are listed in Table S5 (4). To obtain the corresponding sequence of O. barthii, we conducted de novo assembly of two BAC clone sequences (OBART0063E22 and OBART0027J20) that were screened via PCR amplification of sequences around the RAE3 locus and provided by the National U00096.3) and pIndigoBAC-5 (GenBank: EU140754.1), which is the BAC's backbone, using the BWA-MEM algorithm (5). Unmapped paired reads were extracted using SAMtools with the option -f 13 for the view function (6). The unmapped reads were then assembled using ABySS (7) with the setting k = 51, which provided the best result in the tested k values of 31-71. Assembled contigs longer than 5 kb were applied to contig extension using the PRICE tool (8). Parameters for PriceTI were set as follows:fs ./path/to/reads.fastq 200 -icf ./path/to/contigs.fasta 1 1 5 -nc 10 -target 90 0o ./path/to/output.fa. The resultant contigs were further processed using SSPACEstandard for scaffolding (9), with parameter -x set to 1. The longest scaffold, containing 6 Ns in a gap, was confirmed to correspond to the O. sativa subset sequence. BLAST comparison of the obtained sequences was performed using WebACT to create the comparison files required by Artemis Comparison Tools (ACT) (10,11). The local genome sequence of the putative RAE3 location was then visualized using ACT, which was tuned to eliminate line segments connecting sequences of relatively low similarity.
RAE3 Expression data retrieving from RED database. Expression values based on FPKM were retrieved from the Rice Expression Database (RED (12); http://expression.ic4r.org, searched on Jan 4, 2020) which is based on RNA-seq data of O. sativa cv. Nipponbare. The expression values corresponding to the tissues that have "normal" and "WT" without any treatment in experiment name were retrieved and calculated mean value among each tissue. We used the data from the Project IDs; DRP000391, DRP001762, SRP017256, SRP029886, SRP047482, and SRP049102.

RNA isolation and quantitative reverse-transcription (qRT)-PCR.
For qRT-PCR analysis of target genes (RAE1 (Os04g0350700), RAE2 (Os08g0485500) and RAE3 (Os06g0695900)), young panicle tissues (< 1 cm in length) of Koshihikari (O. sativa japonica), GLSL14, GLSL20 and GLSL26 (in which segments of chromosomes 4, 6, and 8 were individually substituted into O. glaberrima from the Koshihikari genetic background, as indicated in Fig. 1) were used as samples. Total RNA was extracted using the RNeasy Plant Mini Kit (QIAGEN, Hilden, Germany), and first-strand cDNA synthesis was performed using the Omniscript RT Kit (QIAGEN, Hilden, Germany). The StepOne Real-Time PCR system (Applied Biosystems, Waltham, MA, USA) was used to analyze the relative expression levels of target genes. Expression levels of target genes were normalized to the endogenous ubiquitin transcript level (Os01g0328400). The comparative cycle threshold (△△CT) method was used to calculate relative expression levels of target genes. Primers used in this study are listed in Table S5.
Alignment of the RAE3 sequence and RAE3 phylogenetic tree. After visualizing the genome rearrangement around the RAE3 location, the CDS sequence of Os06g0695900, which corresponds to OsRAE3, and the 3-kb upstream and 1-kb downstream sequences, were retrieved from the O. sativa genome as the promoter and terminator regions. µl of 99% ethanol and vortex vigorously. Spin the sample at 10,000 g in a microcentrifuge for 1 min and remove the supernatant. Add 50 µl of 99% ethanol and vortex vigorously. Onion bulbs (Allium cepa) were purchased from the supermarket (Nagoya, Japan). The onion was kept at room temperature before being used for particle bombardment. The onion bulb scale leaves were cut into strips of approximately 2 cm × 3 cm, and a strip of scale leaf was placed on a stack of wet Kimwipes in a 9 cm Petri dish.
Then, the strips of onion scale leaves were subjected to particle bombardment using the biolistic PDS1000/He Particle Delivery System (Bio-Rad). Bombardment was performed with a 1100 psi rupture disc (#1652329, Bio-Rad) under the condition of 28inchHg (vacuum level in chamber), 1100 psi helium pressure, 590 MPa pressure. After bombardment with gold particles, samples were incubated at 28ºC for 16 h in the dark.
The epidermal layer was peeled off, soaked in liquid MS medium with or without 2 µM transferred to a 24-well plate and incubated at 22°C in the dark for 12-16h. Protoplasts were observed using a confocal laser scanning microscope (LSM700; Zeiss).

E3 ubiquitin ligase assay in yeast
Yeast strain, media, and reagents. The yeast strain used in this work is YTK2812 (MATa leu2 trp1 his3 ade2 can1 pdr5::Hyg) which is constructed in this study. Cells were grown in synthetic medium (0.69% yeast nitrogen base without amino acids, 2% Dglucose, appropriate amino acids and nucleic acids) at 30˚C. To initiate degradation of was generated through PCR using specific primers (KM154-KM155), with FLAG-Δ RAE3(WT)-TIR as the template. The plasmid which is FLAG-ΔRAE3(WT)-TIR was named pOK832 and FLAG-ΔRAE3(C136S)-TIR was named pOK833. AtIAA17 (At1g04250) was cloned using the pOK521 plasmid (16) as a template and added the 3xHA tag to the 5′ side to produce 3× HA-IAA17-p415ADH construct used as a substrate for FLAG-ΔRAE3-TIR. The constructs for observing cell localization of chimeric OsRAE3 protein fused with GFP were made by using NEBuilder as following manufacture protocol. GFP sequence was amplified with specific primers (KBU71-KBU72) using pMDC111 as a template. PCR products of GFP fused with 19 to 20 bp of complementary sequence of vector plasmid were transferred into pOK832 or pOK833 after EcoRI treatment using the NEBuilder Hifi assembly kit (New England BioLabs, Ipswich, MA, USA) with 50 degrees C for 20 min. Immunodetection was performed using the Chemi-Lumi One L system (Nacalai Tesque, Kyoto, Japan) with a bioanalyzer (LAS4000 mini; GE Healthcare Biosciences, Piscataway, NJ) or X-ray films. calls were filtered out as were SNPs with greater than 0.05% missingness. Nucleotide diversity was computed in 10-kb bins for all 10-kb windows with a bin-average SNP missing data rate < 20%. RAE3 nucleotide diversity ratio (πglaberrima/ πbarthii) was computed using 50-SNP sliding windows with a 2-SNP step size.        (Fig. 4b).