Figure 4. Twi expression is required for the temporal progression of contractile states.

(a) Wild-type pulses progressively increase in magnitude. Probability density functions of the timing of pulses of increasing amplitude-bins are shown. (b) Wild-type pulses become increasingly frequent. The time interval between consecutive pulses is shown with respect to developmental time. Line shows best-fit. (c) Wild-type pulses transition from unconstricting and unratcheted pulses to ratcheted pulses. Probability density functions of the timing of different pulse behaviour classes are shown. (Inset) Cumulative distribution function. Dotted lines demarcate the respective mean timing. (d,g) twi-RNAi disrupts the temporal ordering from weak to strong pulses. Probability density functions of the timing (x axis) of pulses of different amplitude-bins (y axis) for five control (d) and five twi-RNAi embryos (g). (e,h) Period between consecutive pulses decreases more in control embryos (e) than in twi-RNAi embryos (h). Lines show best-fit lines. (f,i) Probability density functions of the timing of pulses of different behaviours show ratcheted pulses occurring after unratcheted and unconstricting pulses in control embryos (f), but co-occurring with them in twi-RNAi embryos (i). (Insets) Cumulative distribution functions. Dotted lines show the respective mean timing. (j) s.d.'s in pulse timing show that twi-RNAi pulses of similar amplitude-bin occur over a broader time period than corresponding control pulses. (k) twi-RNAi disrupts the temporal separation of pulses with different amplitudes. Jensen–Shannon divergence (see Methods) quantifies the dissimilarity of timing between pulses from different amplitude-bins. The lowest- and highest-amplitude pulses from control cells (upper right corner) display more divergent timing than those from twi-RNAi cells (lower left corner), reflecting greater separation and ordering in timings of pulses with different magnitudes in control compared with twi-RNAi tissues. (l,m) Twi expression promotes individual cells to transition into the ratcheted state. The probability of a cell transitioning from having a pulse of a given behaviour class (left columns) into having a subsequent pulse of another behaviour class (right columns) are shown. Wild-type cells (l) but not twi-RNAi cells (m) show biased transitions to the ratcheted state. Both the colours and the widths of the arrows represent the probability of transition.