Yeast Kex1p is a Golgi-associated membrane protein: deletions in a cytoplasmic targeting domain result in mislocalization to the vacuolar membrane

An analysis of the oligosaccharide modification of Kexlp in-dicated secretory compartments to which the protein had been exposed, and the likely subcellular location of Kexlp. The modification of the Asn-linkod oligosaccharide cores of Kexlp occurred within the Golgi apparatus and likely involves several sequential steps. Initially, in a pre-sec7 compartment, the core oligosaccharides are partially modified as judged by an increase in apparent molecular mass (Cooper and Bussey, 1989) . Subsequently, in a post-sec7 compartment, the core is further modified in a MNNl-dependent manner. Invertase produced in sec7 cells at the restrictive temperature does not receive c~(1--'3) linked mannose residues to the core or outer chain (Franzusoff and Schekman, 1989) , most likely because of the failure of secretory proteins to reach the compartment containing the cx(1--,3) mannosyltransferase. It is consistent that sec7 and mnnl mutations result in a similar reduction in the glycosyl elaboration of Kexlp, as they would respectively either prevent Kexlp from reaching, or reduce the activity of, the c~(1~3) mannosyltransferase. The Golgi-based modification of Kexlp can thus be explained as a result of two processes. The first occurs before the sec7 block and most likely involves the addition of an ot(l~6) mannose residue followed by the attachment of an oz(1--'2) linked mannose residue (indicated by * in Fig. 1 ). The second step involves the MNN1dependent addition of up to three a ( l ' 3 ) linked mannose residues (indicated by a circle in Fig. 1) to the oligosaccharide (Ballou et al., 1990) . Graham and Emr (1991) provided evidence for at least three functional compartments in the yeast Golgi apparatus that contain, from cis to trans, the following activities: (1) ct(l~6) mannosyltransferase, (2) cx(l~3) mannosyltransferase, and (3) Kex2p endoprotease. Kexlp must reach the Golgi compartment housing the MNNl-dependent o~(1---3) mannosyltransferase as it is modified by this enzyme. In addition, to process killer toxin, Kexlp must reach the proposed third Golgi compartment containing Kex2p as the toxin precursor substrate for Kexlp is created by a Kex2p-mediated endoproteolytic cleavage. No post-Kex2p compartment before secretory vesicles has been observed (Graham and Emr, 1991) and, therefore, it is likely that Kexlp resides with Kex2p in the same late Golgi compartment (Redding et al., 1991) . It is interesting to note that both Kexlp and Kex2p Figure 9 . Model for the retention of Kexlp within the secretory pathway. A model for the retention of Kexlp is presented. Kexlp is retained in a late Golgi compartment because of its cytoplasmically exposed domain interacting with a receptor (Recpt). Kexlp-AS, being soluble, is secreted by default to the cell surface. Kexlp-Hpa is membrane associated and residues in the vacuole because of several possible reasons: (1) Kexlp-Hpa is partially misfolded and is targeted to the vacuole via an unknown system; (2) the membranespanning domain of Kexlp contains a cryptic targeting signal for the vacuole; (3) Kexlp-Hpa is secreted to the plasma membrane where it is endocytosed and delivered to the vacuole; and (4) membrane-associated proteins that leave the Golgi apparatus are targeted to the vacuole by default. Kexlp is delivered to the vacuole when overproduced to a high level, presumably due to saturation of a Golgi retention receptor. The horizontal bar protruding from the Kexlp-WT box represents the membrane-spann