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

internal deletion constructs, such as ID1. The fact that YQRL was the only sequence, other than the perimembrane basic residues, common to all constructs showing TGN-like staining suggests that these four residues play a preeminent role in localization of the chimeric proteins to the TGN. These observations do not completely rule out the involvement of sequences located NH2-terminal to YQRL in TGN localization, although no other strict requirement for specific residues was apparent from our deletion analysis. To unequivocally establish the intracellular localization of a deletion mutant having the fewest residues of the TGN38 cytoplasmic tall, the construct ID5 (Fig. 4, top) was stably Figure 5 . Immunoelectron microscopic co-localization of the ID5 chimeric protein and endogenous TGN38 in stably transfected NRK cells. Thin frozen sections of calls were sequentially immunostained with a rabbit antibody to Tac and 15-nm gold-conjugated protein A, follo~ved by a rabbit antibody to endogenous TGN38 and 10-nm gold-conjugated protein A, as described in the Materials and Methods section. (G) Golgi complex. Bar, 0.2 #m. Figure 6 . Identification of cytoplasmic residues critical for TGN localization. Single residues within the cytoplasmic tail of the ID5 construct (see Fig. 4 ) were substituted by alanine or aspartic acid residues, as indicated in the scheme. Numbering corresponds to the published sequence of TGN38 (Luzio et al., 1990) . Mutants were expressed by transient transfection of NRK ceils and localized by indirect immunofluorescenee using a mouse mAb to Tac (70-'7) and rhodamine-conjugated antibodies to mouse IgG. Constructs Y333A, L336A, and R335D were observed at the plasma membrane, whereas Q334A, R335A, and the rest of the constructs (not shown in the figure) showed a TGN localization similar to ID5. Bar, 10 ~m. Internalization of mI-labeled anti-Tat pre-bound to transiently transfected CV-1 ceils was measured as described in Materials and Methods. Each point relnesents the average value derived from two (Tac) or three liD5 and R335D) independent experiments performed in duplicate. Duplicate determinations varied by <10%. No detectable binding or internalization was observed in untransfected CV-1 cells.