Eph receptors and their ephrin ligands are membrane bound, yet they generally mediate repulsive responses. As cell–cell contact must occur to facilitate ligand–receptor interaction, how is this followed by rapid cell repulsion? Two reports in Nature Cell Biology now provide a very attractive answer — Eph–ephrin complexes are removed from the cell surface by rapid, localized endocytosis after cell–cell contact.

Zimmer et al. observed the accumulation of internalized ligand–receptor complexes after interactions between EphB2- and ephrinB1-expressing cells (which signal bidirectionally). Full-length EphB2 or ephrinB1 clustered at cell–cell contact sites and were trans-endocytosed — the receptor was internalized by ligand-expressing cells, and vice versa. This trans-endocytosis required intact Eph and ephrin cytoplasmic tails. For example, truncation of ephrinB1 in the recipient cell impaired internalization into this cell and favoured internalization into the EphB2-expressing stimulator cell. Furthermore, trans-endocytosis of Eph–ephrin complexes into receptor-expressing cells required Eph kinase activity.

During in vitro co-culture assays of cells expressing ephrinB1 or EphB2, rapid clustering was followed by bidirectional trans-endocytosis and cell retraction. Blocking ephrinB1 endocytosis caused receptor–ligand clusters to grow, be engulfed by the EphB2 cell and induce strong retraction. But truncation of both cytoplasmic tails resulted in strong cell–cell adhesion.

In the other study, Marston et al. used microinjection experiments to study EphB4–ephrinB2 interactions. At regions of ligand–receptor contact, EphB4 phosphorylation was followed by membrane protrusions and ruffles, and the cells subsequently separated. Similar to the results of Zimmer et al., receptor- and ligand-positive internalized vesicles were seen, but in this case, trans-endocytosis into EphB4-expressing cells predominated. The vesicles contained membrane derived from the neighbouring cell surface, but probably didn't arise from clathrin-coat- or caveolae-mediated mechanisms.

So, the localized ruffles and extensions prompted Marston et al. to look at actin-dependent internalization. Indeed, blocking actin polymerization inhibited EphB4 internalization and subsequent cell retraction. Rac-mediated membrane ruffling, in particular, was required for internalization.

Both groups verified the physiological relevance of these results using cell-culture models — of growth-cone collapse and venous–arterial boundary delineation — in which they confimed their findings that trans-endocytosis of Eph–ephrin complexes is needed for cells to kiss and break up.