The small GTPase Rap1 is known to regulate morphogenesis in Drosophila melanogaster, but the mechanism by which it does this has remained unclear. Rap1 was originally thought to antagonize Ras1 signalling, but a subsequent study showed that mutations in Rap1 induced abnormal cell shape and disrupted migration, rather than affecting Ras-mediated signalling pathways. Now, reporting in Science, Knox and Brown have found that Rap1 probably regulates these properties by regulating the position of adherens-junction proteins.

Studying the behaviour of epithelial cells during Drosophila wing development, Knox and Brown found that clones of Rap1-mutant cells spread — often in pairs or groups of four — into surrounding wild-type cells instead of staying together as a group. Combined with the fact that Rap1-mutant cells lacked the typical hexagonal shape and had a reduced apical surface compared with wild-type cells, this indicated that Rap1 might regulate apical cell–cell adhesion.

The authors then studied components of adherens junctions — DE-cadherin, α-catenin and β-catenin — on the apical surface of Rap1-mutant cells and noted that they localized predominantly to one side of the cells, forming 'clusters' of adherens-junction proteins with Rap1-mutant, but not wild-type, neighbouring cells. Interestingly, the cytoskeletal linker proteins AF6/canoe — to which activated Rap1 binds — and ZO-1 — which interacts with both AF6/canoe and α-catenin — were also mislocalized, indicating that ZO-1 might link adherens-junction proteins and Rap1. By contrast, loss of Rap1 function had no effect on septate-junction-associated proteins. Furthermore, as DE-cadherin and β-catenin didn't mislocalize along the apical–basal axis of Rap1-mutant cells, Rap seems to affect the distribution of adherens junctions around the periphery of apical cells specifically.

So, could the mislocalization of adhesion molecules such as DE-cadherin be responsible for the dispersal of the Rap1-mutant cells into surrounding cells? As there are other examples of cell-sorting in response to differential adhesion, the authors proposed that small groups of Rap1-mutant cells could effectively be 'drawn in' to surrounding wild-type tissue as a result of adhesion being stronger between mutant and wild-type cells than between Rap1-mutant cells. Also, because adherens-junction proteins showed mislocalization in undispersed cells, this indicates that the dispersal phenotype occurs as a result of mislocalization, rather than vice versa.

Closer inspection, using a green-fluorescent protein (GFP)–Rap1 fusion protein, showed that GFP–Rap1 was highly concentrated at adherens junctions (consistent with a potential link between Rap1, ZO-1 and AF6/canoe). The authors also noticed that while GFP–Rap1 localized around the cell cortex during cell division, when sister cells subsequently formed it was transiently enriched at the junction between them.

Based on these findings, Knox and Brown propose that the localization of Rap1 at adherens junctions might be necessary for maintaining adherens-junction proteins here, too. Without Rap1, the adherens-junction 'ring' of proteins that surrounds a dividing mother cell might not be resealed during cytokinesis, and as a consequence, it could recoil to one side, causing the proteins to 'cluster'. Rap1, therefore, seems to be able to mediate cell-shape changes through its ability to regulate adherens-junction positioning.