Fig. 3: Knirps repression is rapidly reversible and memoryless.

A Testing the reversibility of Knirps repression using a step-like optogenetic perturbation. Upon removal of Knirps repressor from the nucleus, transcriptional activity can remain repressed or recover, depending on whether repression is irreversible or reversible. B Snapshots from a movie before (top) and after (bottom) the optogenetic export of Knirps protein. Nuclei whose transcription was originally repressed by Knirps fully reactivate after 4 min of illumination. C Heatmap of single-cell reactivation trajectories sorted by response times. Response time is defined as the interval between the perturbation time and when the MS2 spots reappear. D Average repressor concentration (green) and the fraction of actively transcribing cells (magenta) before and after blue light illumination. We find that Knirps repression is rapidly reversible within 4 min. (n = 229 nuclei from 4 embryos, averaged over a -2% to 2% window along the anterior-posterior axis centered on the Knirps concentration peak). E Fast reactivation occurs with an average of 2.5 min. The reactivation response time is calculated as the interval between the perturbation and when a locus is first observed to resume transcription. (n = 139 nuclei from 4 embryos). Inset panel describes the cumulative distribution of reactivation times. To exclude gene loci that were transiently OFF due to transcriptional bursting or missed detections, we focused this analysis on gene loci that were silent for at least 2 min before perturbation. F Knirps repression is memoryless. Plot showing the reactivation response time of individual loci as a function of the time spent in the repressed state before optogenetic reactivation. Fitting with a linear regression model (gray dotted line) results in p-value = 0.495, which confirms that the reactivation response time is independent of the repressed duration of the locus. Red dots represent the means of the binned data. (Error bars in D and F indicate the bootstrap estimate of the standard error. p-value in F is for the F-test on the regression model, which tests whether the model fits significantly better than a degenerate model consisting of only a constant term).