Formation of multicellular rosettes in the developing neuroepithelium. (A) Schematic indicates the embryonic stage, and the red box shows imaging location in the neural plate. Aligned rows of cells (blue boxes) form in the relatively flat neural plate and contract their shared mediolateral junctions within 2 h, concomitant with bending along the midline (x,z views). (B) Insets show aligned cells indicated in A. (C and D) Computational segmentation of areas shown in A reveals an increasing number and complexity of rosettes over time, quantified in D. Bar graphs show mean ± SEM, n = 827 rosettes from 4 embryos. **P < 0.01, *P < 0.05, Student’s t test. (E) The intensity of Lifeact-EGFP increases as the shared mediolateral junctions contract (normalized to junction area). Graph shows mean (orange lines) ± SEM (shaded areas) from three embryos. ***P < 0.001, Wilcoxon Signed Rank Test, t: 0 m versus t: 30 m. (F–I) FRAP of the actin at the central vertex shows a higher immobile fraction and a faster recovery time in eight-cell rosettes (n = 17 rosettes, 6 embryos) compared with five-cell rosettes (n = 17 rosettes, 6 embryos), inset shows first 5 s of fluorescence recovery. Graphs show mean ± SEM *P < 0.05, Student’s t test. (J and K) Schematic indicates embryonic stage and red box shows imaging location in the neural plate. Treatment with Latrunculin A causes relaxation of shared mediolateral junctions and flattening of the neural plate. (L and M) Computational segmentation of areas shown in J reveals that Latrunculin A reduces the number and complexity of rosettes. Bar graphs show mean ± SEM, n = 2,044 rosettes from 4 embryos. **P < 0.01, *P < 0.05, two-sided Student’s t test. (N) The intensity of Lifeact-EGFP localized to shared mediolateral junctions is decreased by Latrunculin A treatment (normalized to junction area). Graph shows mean (orange lines) ± SEM (shaded areas) from three embryos. ***P < 0.001, Wilcoxon Signed Rank Test, t: 0 m versus t: 30 m.