Our work explores why the short branches (spikelets) situated towards the base of the wheat inflorescence (spike) produce fewer and smaller grains than those in the centre. The basal spikelets initiate first in development but soon lag behind the central spikelets, ultimately showing higher rates of floret abortion and giving wheat spikes their lanceolate shape. Uncovering the genetic network governing this phenomenon of basal spikelet infertility could facilitate breeding improved spike architecture, with the potential for higher per-spike, and perhaps per-plot, productivity.

To investigate this process, we generated a semi-spatial transcriptomic timecourse of micro-dissected central and basal spike sections collected at five stages of early spike development (Waddington W2-5). We utilised NILs differing at the VRT-A2 locus to allow us to exclude downstream targets of this well-characterised spike architecture gene. We found that even at W2, over 7,000 high confidence gene models, including 547 transcription factors, were differentially expressed between the central and basal sections.

After filtering for expression patterns of interest and against the rice literature, we identified MOF1/MFS2 and SEP1-6 as candidates for further study. Both promote the transition from spikelet meristems to floral meristems in rice and, intriguingly, are more strongly expressed in the centre versus the base of the early wheat spike. Given that floret abortion is determined by floret maturity two weeks pre-anthesis, we hypothesise that accelerating the spikelet-to-floral meristem transition in basal spikelets could reduce abortion and raise their productivity.

To test this, we aimed to boost the expression of our candidate genes specifically in the basal spikelets at W2-2.5. We developed a custom promoter by identifying genes with the desired expression pattern, then selecting putative proximal regulatory regions using tissue-specific ATAC-seq data. This promoter was shown to confer basal spikelet-specific expression to a tdTomato reporter and we are now generating additional transgenics to mis-express our candidate floral transition regulators.

We have also begun exploring spike development with fully spatial transcriptomics. Our semi-spatial dataset was used to inform selection of a 300-gene panel for VizGen MERFISH, a highly multiplexed form of fluorescence in-situ hybridisation providing subcellular resolution of single transcripts. MERFISH transcript counts correlate well with expression values from the semi-spatial RNA-seq (ρ=0.67±0.04 SD), indicating that the data can be used quantitatively as well as qualitatively. In addition to validating findings from our semi-spatial study, the data is proving useful for generating new hypothesises on the genetics of spike development and basal spikelet infertility.