Effects of GTPase knockdown on collective cell migration. (A) Example of a wound-healing experiment. (insets) Velocity vectors showing that front cells begin to migrate before deeper cells. Bar, 100 µm. (B) Immunofluorescent staining of E-cadherin and ZO1 before scratching the monolayer. Bar, 20 µm. (C) Monolayer edge evolution over 500 min. (inset top) Edge evolution during the first 125 min. (inset bottom) Increase of wound-healing rate during the first 125 min. (D) Construction of a kymograph of a wound-healing experiment: mean speed of cells at different distances from the monolayer edge over time. (E) Speed kymographs for control cells and under depletion of Cdc42, Rac1, and RhoA. All kymographs are averages of four locations in a well (similar to Kim et al., 2013). (F) Western blots of control (CTRL) and RhoA-, Rac1-, and Cdc42-depleted cells. (G) Monolayer velocity over time (left to right) for control (top) and RhoA-depleted cells (bottom). Bars, 100 µm. (H) Wound-healing rate. Each point was calculated as the difference between the migration rate in one location and the mean of the same day’s control experiment. N = 6–9 d, n = 24–47 locations (CDC42: N = 9, n = 37; RAC1: N = 6, n = 24; RHOA: N = 9, n = 47). Statistics via Wilcoxon signed rank test: ***, P ≤ 0.0001. The right panel shows the migration rate in one location as a function of the mean daily control for the two RhoA hairpins.