By analyzing the trafficking of HRP–P-selectin chimeras in which the lumenal domain of P-selectin was replaced with horseradish peroxidase, we determined the sequences needed for targeting to synaptic-like microvesicles (SLMV), dense core granules (DCG), and lysosomes in neuroendocrine PC12 cells. Within the cytoplasmic domain of P-selectin, Tyr777 is needed for the appearance of P-selectin in immature and mature DCG, as well as for targeting to SLMV. The latter destination also requires additional sequences (Leu768 and 786DPSP789) which are responsible for movement through endosomes en route to the SLMV. Leu768 also mediates transfer from early transferrin (Trn)-positive endosomes to the lysosomes; i.e., operates as a lysosomal targeting signal. Furthermore, SLMV targeting of HRP–P-selectin chimeras, but not the endogenous SLMV protein synaptophysin/p38, previously shown to be delivered to SLMV directly from the plasma membrane, is a Brefeldin A–sensitive process. Together, these data are consistent with a model of SLMV biogenesis which involves an endosomal intermediate in PC12 cells. In addition, we have discovered that impairment of SLMV or DCG targeting results in a concomitant increase in lysosomal delivery, illustrating the entwined relationships between routes leading to regulated secretory organelles (RSO) and to lysosomes.

One pathway in forming synaptic-like microvesicles (SLMV) involves direct budding from the plasma membrane, requires adaptor protein 2 (AP2) and is brefeldin A (BFA) resistant. A second route leads from the plasma membrane to an endosomal intermediate from which SLMV bud in a BFA-sensitive, AP3-dependent manner. Because AP3 has been shown to bind to a di-leucine targeting signal in vitro, we have investigated whether this major class of targeting signals is capable of directing protein traffic to SLMV in vivo. We have found that a di-leucine signal within the cytoplasmic tail of human tyrosinase is responsible for the majority of the targeting of HRP-tyrosinase chimeras to SLMV in PC12 cells. Furthermore, we have discovered that a Met-Leu di-hydrophobic motif within the extreme C terminus of synaptotagmin I supports 20% of the SLMV targeting of a CD4-synaptotagmin chimera. All of the traffic to the SLMV mediated by either di-Leu or Met-Leu is BFA sensitive, strongly suggesting a role for AP3 and possibly for an endosomal intermediate in this process. The differential reduction in SLMV targeting for HRPtyrosinase and CD4-synaptotagmin chimeras by di-alanine substitutions or BFA treatment implies that different proteins use the two routes to the SLMV to differing extents. INTRODUCTIONThe efficient sorting of many transmembrane proteins to a variety of post-Golgi destinations is controlled by short specific sequences located within their cytoplasmic domains, sorting signals (for review, see Trowbridge et al., 1993;Sandoval and Bakke, 1994). At present, two major groups of sorting signals have been identified. The first group comprises tyrosine-based signals, which usually conform to the consensus YXXØ (where X is any amino acid, and Ø is a strong hydrophobic amino acid) or FXNPXY. The second group of sorting signals contains di-leucine/di-hydrophobic signals, in which one of the leucines can be substituted by isoleucine, methionine, or valine without loss of function (Letourner and Klausner, 1992;Bremnes et al., 1994;Sandoval and Bakke, 1994;Pond et al., 1995). Sorting signals falling outside these groups include the amphipathic ␣-helixes, which can adopt a supercoiled conformation and were found in the cytoplasmic domains of vesicle-associated membrane protein II (VAMPII) and the ␤-chain of the interleukin-1 receptor (Grote et al., 1995;Subtil et al., 1997). In addition, clusters of acidic residues in the context of a casein kinase II recognition site were shown to facilitate intracellular sorting of both furin and the mannose-6-phosphate receptor (Schafer et al., 1995;Voorhees et al., 1995;Mauxion et al., 1996). The functioning of sorting signals requires their direct (or possibly indirect) interaction with adaptor protein (AP) complexes, such as AP1, AP2, and AP3, which assemble with clathrin during vesicular budding (for review, see Kirchhausen et al., 1997;Odorizzi et al., 1998), or with arrestins, which function as adaptors for G-protein-coupled receptors (Ferguson et al., 1996;Goodman et al., 1996).Whereas t...