Reconstructing the transcriptional ontogeny of maize and sorghum supports an inverse hourglass model of inflorescence development

Abstract

Assembling meaningful comparisons between species is a major limitation in studying the evolution of organismal form. To understand development in maize and sorghum, closely-related species with architecturally distinct inflorescences, we collected RNAseq profiles encompassing inflorescence body plan specification in both species. We reconstructed molecular ontogenies from 40 B73 maize tassels and 47 BT×623 sorghum panicles and separated them into transcriptional stages. To discover new markers of inflorescence development, we used random forest machine learning to determine stage by RNAseq. We used two descriptions of transcriptional conservation to identify hourglass-like developmental stages. Despite short evolutionary ancestry of 12 million years, we found maize and sorghum inflorescences are most different during their hourglass-like stages of development, following an ‘inverse-hourglass’ model of development. We discuss if agricultural selection may account for the rapid divergence signatures in these species and the observed separation of evolutionary pressure and developmental reprogramming.HighlightsTranscript dynamics identify maize tassel and sorghum panicle developmental stagesRandom forest predicts developmental age by gene expression, providing molecular markers and an in silico staging applicationMaize and sorghum inflorescences are most similar when committing stem cells to a determinant fateExpression conservation identifies hourglass-like stage, but transcriptomes diverge, similar to ‘inverse hourglass’ observations in cross-phyla animal embryo comparisons

Conserved, divergent and heterochronic gene expression during Brachypodium and Arabidopsis embryo development

“…To relate developmental progression between Arabidopsis and Brachypodium embryogenesis, we calculated a pseudotime metric (Leiboff and Hake 2019 ), namely developmental time units (DTUs), to reconstruct the molecular ontogenies for each species using the expression trajectory information in their respective PCA plots (Fig. 3 C, D).…”

“…Sample pseudotime indexing was performed as described in previous studies with some modifications (Leiboff and Hake 2019 ). Specifically, all of the differentially expressed genes were used to separate samples on a PCA plot for each species.…”

“…To generate a phasigram (Leiboff and Hake 2019 ; Levin et al 2016 ), we perform a PCA analysis for separating genes based on the standardized expression data. The first two components, component 1 and component 2, were used to draw the PCA plot.…”

Comparative transcriptomics reveals desynchronisation of gene expression during the floral transition between Arabidopsis andBrassica rapacultivars

“…ese closely related species move through a similar morphological sequence of developmental stages, so one might expect their transcriptomes to progress along a path of similar gene expression states. Under this assumption, we would expect to see that plants at similar morphological developmental stages exhibit similar transcriptomes (Leiboff & Hake, 2019).…”

“…Transcriptome comparison between Arabidopsis and B. rapa by alignment of gene expression profiles using curve registration (Leiboff & Hake, 2019;Ramsay & Silverman, 2005) suggests that there is not one, but many different 'developmental progressions' of gene expression running at different speeds relative to each other. ere is, therefore, no single common 'developmental time' based on gene expression in these closely related plants.…”