Localization of TGN38 to the trans-Golgi network: involvement of a cytoplasmic tyrosine-containing sequence

The ability of a tyrosine-containing cytoplasmic sequence to cause localization to the TGN can be explained by at least two different, although not mutually exclusive models. The first model assumes that the cytoplasmic determinant is directly recognized at the level of the TGN, resulting in retention or slow exit from this compartment. By analogy with tyrosine-based internalization signals (Pearse, 1985 (Pearse, , 1988 Glickman et al., 1989; Beltzer and Spiess, 1991) , the TGN localization determinant characterized here may bind to TGN-specific clathrin-associated adaptor molecules, such as 3,-adaptin (Robinson, 1990) . Interestingly, a large fraction of the yeast Golgi peptidase kex2p, whose cytoplasmic tail is necessary for Golgi localization, is missorted to the plasma membrane in yeast mutants deficient in clathrin heavy chains (Payne and Schekman, 1989; Seeger and Payne, 1992) . Interaction with components of non-clathrin coats of the TGN is also possible. For any of these potential interactions to result in TGN retention, however, association of cytoplasmic sequences with coated structures would have to be a relatively stable event. The specificity and kinetics of the interaction would be dictated by amino acid residues in or around the tyrosine-based motif, such that single amino acid substitutions could perturb the interaction.