The T-cell–antigen-presenting-cell contact site is a highly organized complex of cell-surface receptors and associated intracellular signalling proteins that is known as the immunological synapse. Although the components of the immunological synapse are well defined, the mechanisms by which it forms are poorly understood. Now, a report published in Cell indicates that protein–protein interactions can direct the formation of discrete microdomains of receptors and signalling molecules after T-cell activation.

Previous studies have implicated both lipid rafts and actin and myosin cytoskeletal networks as mediators of protein clustering at the immunological synapse. So, to further investigate the mechanisms by which microdomains of signalling molecules form, Douglass and Vale used both conventional confocal microscopy and single-molecule tracking to analyse the overall membrane distribution and single-molecule behaviour of green fluorescent protein (GFP)-tagged T-cell-signalling molecules, after the activation of Jurkat T cells by immobilized T-cell receptor (TCR)-specific antibodies.

In the absence of T-cell activation, individual molecules sometimes moved rapidly and at other times were static. However, the different signalling molecules showed distinct patterns of mobility: CD2 molecules were mostly immobile, with only occasional periods of rapid mobility, whereas both a marker of lipid rafts (the ten N-terminal amino acids of LCK (LCK10) fused to GFP) and a non-raft-associated protein (CD45) spent little time immobile. These patterns of diffusion indicate that immobilization was not due to lipid-raft formation. Consistent with this, after TCR crosslinking both raft-associated wild-type LAT (linker for activation of T cells) and a LAT mutant that cannot associate with lipid rafts but can mediate protein–protein interactions (LAT(C-S)) showed reduced mobility.

After TCR crosslinking, CD2 molecules were observed to cluster and to colocalize with LCK and LAT but not CD45. Colocalization of LCK and CD2 was not a result of LCK being a raft-associated protein, because the lipid-raft marker LCK10 did not colocalize with CD2. Furthermore, a raft-associated LAT mutant that cannot be phosphorylated (LAT(Y-F)) did not colocalize with CD2, which indicates that protein–protein interactions mediated by tyrosine-phosphorylated LAT mediate the colocalization of LAT and CD2. Consistent with this, CD2 clustering after TCR crosslinking was impaired in a cell line derived from Jurkat T cells that has severely reduced LAT expression, and this clustering could be restored by transfecting the cells with LAT but not LAT(Y-F).

Further single-molecule analysis indicated that CD45 was excluded from CD2-containing microdomains, whereas LCK, LAT and LAT(C-S) were often associated with these microdomains and, when associated, showed little mobility. Together, these studies indicate that LCK and LAT are preferentially trapped in the CD2-containing microdomains through protein–protein interactions and that the association of these proteins with lipid rafts does not regulate this process. This led the authors to suggest that protein–protein interactions should be considered an early event in T-cell signalling and a mechanism of immunological-synapse formation.