Quantitative imaging of transcription in living Drosophila embryos links polymerase activity to patterning

Spatiotemporal patterns of gene expression are fundamental to every developmental program. The resulting macroscopic domains have been mainly characterized by their levels of gene products. However, the establishment of such patterns results from differences in the dynamics of microscopic events in individual cells such as transcription. It is unclear how these microscopic decisions lead to macroscopic patterns, as measurements in fixed tissue cannot access the underlying transcriptional dynamics. In vivo transcriptional dynamics have long been approached in single-celled organisms, but never in a multicellular developmental context. Here, we directly address how boundaries of gene expression emerge in the Drosophila embryo by measuring the absolute number of actively transcribing polymerases in real time in individual nuclei. Specifically, we show that the formation of a boundary cannot be quantitatively explained by the rate of mRNA production in each cell, but instead requires amplification of the dynamic range of the expression boundary. This amplification is accomplished by nuclei randomly adopting active or inactive states of transcription, leading to a collective effect where the fraction of active nuclei is modulated in space. Thus, developmental patterns are not just the consequence of reproducible transcriptional dynamics in individual nuclei, but are the result of averaging expression over space and time.