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

A second model entails recognition of the cytoplasmic determinant at a post-TGN location, such as the plasma membrane, with TGN localization being a consequence of retrieval from another compartment. According to this model, TGN38 can traffic normally from the TGN to the plasma membrane where the YQRL motif then functions as an internalization signal. From the endosomal compartment, the protein would be transported back into the TGN by default or by additional sorting information in the localization determinant. This explanation would be analogous to the view of the ER localization motif, KDEL, as a retrieval rather than retention signal (Pelham, 1991) . Consistent with this retrieval model is our observation of a small amount of T-G-G at the cell surface of CV-1 stable transfectants, the ability of the chimeras to internalize bound antibody and the detection of chimeras in early endosomes. In addition, recent observations have suggested the existence of a pool of TGN38 cycling between the TGN and the call surface (Ladinsky, M. S., and K. E. Howell. 1992. Mol. Biol. Cell. 3:309) . To what extent a retrieval mechanism contributes to the localization of TGN38 chimeras to the TGN, however, remains to be established. The failure of the endocytosed R335D mutant to localize to the TGN and the different distributions of transferrin receptor and TGN-localized Tac-TGN38 chimeras suggest that the property of internalization per se is insufficient to specify TGN localization.