Just like W. E. Hill's famous optical illusion (see picture), ephrins have mastered the art of being simultaneously attractive and repulsive. Three papers, in Nature, EMBO Journal and the Journal of Biological Chemistry provide clues as to how these signalling molecules lead their double life.

Eph receptor tyrosine kinases and their membrane-tethered ligands, the ephrins, provide guidance cues for developing neurons and blood vessels. The classical view of ephrin signalling is that it mediates repulsion between ephrin-expressing and Eph-expressing cells, but this relationship doesn't always hold true. Johan Holmberg and colleagues found such a case when they knocked out the gene for ephrin-A5 in mice: some of the mice lacked brains because the neural tube had failed to close at the cranial end — probably due to failure of an adhesive, rather than a repulsive, signal.

In situ hybridization in wild-type mouse embryos revealed that ephrin-A5 and three splice variants of its receptor, EphA7, are expressed at the edges of the cranial neural folds as the neural tube closes, and cells dissected from them adhered more tightly to EphA7-coated surfaces if they expressed ephrin-A5. But do all the EphA7 splice variants behave in the same way? In chemotactic assays, cells expressing full-length EphA7 actually repelled ephrin-A5, but when expression of a truncated splice form of EphA7 (EphA7-T1) was switched on, the repulsive effect was blocked. But there's more to this response than just inhibition of repulsion, because when cells expressing the two EphA7 variants were plated out on confluent layers of a mixture of ephrin-A5+ and ephrin-A5− cells, the cells expressing the EphA7-T1 grew preferentially on the ephrin-A5+ cells, whereas those expressing the full-length receptor preferred ephrin-A5− cells. The most likely mechanism is that the truncated receptor acts in a dominant-negative manner, a model supported by the finding that EphA7-T1 expression reduces tyrosine phosphorylation of the full-length receptor.

Another peculiarity of ephrin signalling is that it's bidirectional: engagement of Eph receptors transmits signals to the ligand-expressing cells, as well as the receptor-expressing cells. This is all the more intriguing for the A-type ephrins, which are tethered by a glycosylphosphatidylinositol anchor. How can this transmit signals to the cell's interior? Davy and Robbins find that ephrin-A5, activated by the extracellular domain of the EphA5 receptor, stimulates adhesion to fibronectin and laminin, with subsequent morphological changes. These effects are blocked by an antibody against β1-integrin. Experiments with inhibitors implicate mitogen-activated protein kinases and Src-family protein kinases in this response. Huai and Drescher have used a similar system to fish for molecules downstream of ephrin-A activation, and find a mysterious 120 kDa protein that becomes phosphorylated on a tyrosine residue before integrin-mediated adhesion occurs.

We're beginning to paint — albeit in broad brushstrokes — a picture of how ephrins and Eph receptors communicate their mixed messages, but now we must focus in on the fine detail. Are the adhesive effects of truncated Eph receptors due to 'reverse' signalling to ephrin-A-expressing cells? What's the order of events upon ephrin activation, and what's the identity of the 120 kDa stranger lurking in the shadows?