Unraveling the spatiotemporal organization of a developmental oscillator

C. elegans expresses ~3,700 genes rhythmically during larval development. These oscillations are the apparent manifestation of a developmental clock that times molting and possibly other developmental events. Expression of each gene peaks once per larval stage, but for each gene at a different time. It is unknown whether this unusually broad peak phase dispersion is an inherent characteristic of this oscillator or an artifact from an analysis of entire worms. Additionally, it is unclear whether oscillating genes follow the same dynamics in each tissue they are expressed in, or tissue-specific differences exist. To distinguish between these alternatives and gain insight into the spatial organization of the oscillator, we used single cell RNA sequencing to create an “oscillation-atlas” of the whole worm. We find rhythmic expression in nine primarily epithelial cell types, further supporting a link to molting. Strikingly, the broad phase dispersion is also seen at the individual tissue level. It is thus a genuine feature of this oscillator, which distinguishes it from other genetic oscillators such as the circadian clock. Further, we identify 2261 additional oscillating genes, which we previously were unable to classify as such in bulk RNA-seq data. Moreover, we find first hints towards peak-phase differences of genes expressed in more than one tissue, which could argue for a certain level of tissue-autonomy. To gain further insight into the oscillator machinery, we investigated chromatin accessibility through an ATAC-seq time course. This revealed an unexpectedly dynamic chromatin landscape, with >5,000 sites exhibiting rhythmically changing chromatin accessibility. These chromatin sites possess the intrinsic capacity to drive oscillatory gene expression with specific spatial and temporal output. Intersection with transcription factor binding motif and ChIP-seq binding data allowed us to link a set of transcription factors to these dynamics, including some that we identified in parallel in a screen for molting genes. Hence, we propose that we have identified likely components of the C. elegans developmental clock