Effect of Erbb2 and EGFR inhibition on the hai1a phenotype. (A and B) Control and hai1a morphant periderm-nuclear-EGFP embryos with or without injection of erbb2 MO, treatment with the EGFR inhibitor PD168393 or par2b deficiency were imaged and analyzed as in Fig. 3. Percentage of cells undergoing extrusion without (A) and with (B) obvious nuclear fragmentation (mean ± SEM; n = 10) is shown. In A, the hai1a MO, hai1a MO plus PD168393, and hai1a erbb2 MO groups were different from control but not from each other by one-way ANOVA and Bonferroni posttest (*, P ≤ 0.05). No other groups were different from control. In B, the hai1a MO PD168393 group was different from control (*, P < 0.0001). (C–G) Control or hai1a morphant basal-layer LifeActGFP embryos without or with PD168393 treatment were imaged and analyzed as in Fig. 4. Percentage of cells in the field imaged that maintained stable cell–cell contact (mean ± SEM; n = 10) is shown. Two-way ANOVA and Tukey posttest were used to analyze C–G. In C, the hai1a MO group was different from control and from the other groups (*, P < 0.0001). Control and hai1a morphant periderm-nuclear-EGFP embryos without or with PD168393 treatment (D and F) or erbb2 MO treatment (E and G) were labeled with BrdU and analyzed as in Fig. 6. The percentage of cells labeled with BrdU (mean ± SEM; n = 3) in periderm (D and E) and basal layer (F and G) is shown. In D, the hai1a MO and hai1a MO par2b^−/− groups were different from control and from each other (*, P < 0.05). PD168393 treatment had a significant effect overall, and the hai1a MO par2b^−/− ± PD168393 groups were different (*, P < 0.0001). The same result was obtained in E with erbb2 MO. In F and G, the hai1a group was different from all other groups (*, P < 0.05 in F; *, P < 0.0001 in G). NS, not significant.