Direct evidence for a receptor-ligand interaction between the T-cell surface antigen CD2 and lymphocyte-function-associated antigen 3.