Figure 6: Robo receptors control PSC cell clustering by repressing Cdc42 activity.

(a,b) LifeactGFP (green) is expressed in PSC cells under the control of the col driver (col>) (a) and co-expressed with a constitutive form of Cdc42 (cdc42-CA) (b). (c,d) Quantification of PSC cell numbers (c) and PSC cell clustering (d). (e–g) Antp (green) and RFP (col>moeRFP, red) label PSC cells in robo2 KD (e), when a dominant negative form of Cdc42 (cdc42-DN) is expressed (f) or when robo2 KD and cdc42-DN are co-expressed in the PSC (g). While the expression of cdc42-DN has no major effect on PSC cells (f), its co-expression in robo2 KD (g) rescues the robo2 KD PSC defect (e). (h,i) Quantification of PSC cell numbers (h) and PSC cell clustering (i). (j,k) Antp (red) labels PSC cells in vilse¹/+ heterozygous mutant (j) and in vilse¹/+; robo2^Ex33/+ trans-heterozygous mutant (k). (l,m) Quantification of PSC cell number (l) and PSC cell clustering (m). A PSC cell clustering defect is observed in vilse¹/+ heterozygous mutant. (n,o) Antp (red) labels PSC cells in col>vilse (n) and col>vilse>robo KD (o). (p,q) Quantification of PSC cell numbers (p) and PSC cell clustering (q). Averaging the clustering of all col>robo KD>vilse, LGs does not reveal a significant clustering defect (Fig. 6q). However, these LGs (n=10 lobes) can be divided into two classes: clustered (60%) and unclustered (40%), revealing a partial rescue of PSC cell clustering defect. Statistical analysis t-test (Mann–Whitney nonparametric test) was performed using GraphPad Prism 5 software. Nuclei are labelled with Topro (blue). Scale bars, 10 μm.