The barley inflorescence (spike) comprises a multi-noded central stalk (rachis) with tri-partite clusters of uni-floretted spikelets attached alternately along its length. Relative fertility of lateral spikelets within each cluster leads to spikes with two or six rows of grain, or an intermediate morphology. Understanding the mechanisms controlling this key developmental step could provide novel solutions to enhanced grain yield. Classical genetic studies identified five major SIX-ROWED SPIKE (VRS) genes, with four now known to encode transcription factors. Here we identify and characterise the remaining major VRS gene, VRS3, as encoding a putative Jumonji C-type H3K9me2/me3 demethylase, a regulator of chromatin state. Exploring the expression network modulated by VRS3 reveals specific interactions, both with other VRS genes and genes involved in stress, hormone and sugar metabolism. We show that combining a vrs3 mutant allele with natural six-rowed alleles of VRS1 and VRS5 leads to increased lateral grain size and greater grain uniformity.

BackgroundThe time required to analyse RNA-seq data varies considerably, due to discrete steps for computational assembly, quantification of gene expression and splicing analysis. Recent fast non-alignment tools such as Kallisto and Salmon overcome these problems, but these tools require a high quality, comprehensive reference transcripts dataset (RTD), which are rarely available in plants.ResultsA high-quality, non-redundant barley gene RTD and database (Barley Reference Transcripts – BaRTv1.0) has been generated. BaRTv1.0, was constructed from a range of tissues, cultivars and abiotic treatments and transcripts assembled and aligned to the barley cv. Morex reference genome (Mascher et al. Nature; 544: 427–433, 2017). Full-length cDNAs from the barley variety Haruna nijo (Matsumoto et al. Plant Physiol; 156: 20–28, 2011) determined transcript coverage, and high-resolution RT-PCR validated alternatively spliced (AS) transcripts of 86 genes in five different organs and tissue. These methods were used as benchmarks to select an optimal barley RTD. BaRTv1.0-Quantification of Alternatively Spliced Isoforms (QUASI) was also made to overcome inaccurate quantification due to variation in 5′ and 3′ UTR ends of transcripts. BaRTv1.0-QUASI was used for accurate transcript quantification of RNA-seq data of five barley organs/tissues. This analysis identified 20,972 significant differentially expressed genes, 2791 differentially alternatively spliced genes and 2768 transcripts with differential transcript usage.ConclusionA high confidence barley reference transcript dataset consisting of 60,444 genes with 177,240 transcripts has been generated. Compared to current barley transcripts, BaRTv1.0 transcripts are generally longer, have less fragmentation and improved gene models that are well supported by splice junction reads. Precise transcript quantification using BaRTv1.0 allows routine analysis of gene expression and AS.

Summary Hordeum species develop a central spikelet flanked by two lateral spikelets at each inflorescence node. In ‘two‐rowed’ spikes, the central spikelet alone is fertile and sets grain, while in ‘six‐rowed’ spikes, lateral spikelets can also produce grain. Induced loss‐of‐function alleles of any of five Six‐rowed spike ( VRS ) genes ( VRS 1‐5 ) cause complete to intermediate gains of lateral spikelet fertility. Current six‐row cultivars contain natural defective vrs1 and vrs5 alleles. Little information is known about VRS mechanism(s). We used comparative developmental, expression and genetic analyses on single and double vrs mutants to learn more about how VRS genes control development and assess their agronomic potential. We show that all VRS genes repress fertility at carpel and awn emergence in developing lateral spikelets. VRS 4, VRS 3 and VRS 5 work through VRS 1 to suppress fertility, probably by inducing VRS 1 expression. Pairing vrs3 , vrs4 or vrs5 alleles increased lateral spikelet fertility, despite the presence of a functional VRS 1 allele. The vrs3 allele caused loss of spikelet identity and determinacy, improved grain homogeneity and increased tillering in a vrs4 background, while with vrs5 , decreased tiller number and increased grain weight. Interactions amongst VRS genes control spikelet infertility, determinacy and outgrowth, and novel routes to improving six‐row grain.