HAMSTER UDP-N-ACETYLGLUCOSAMINE-DOLICHOL-P N-ACETYLGLUCOSAMINE-1-P TRANSFERASE HAS MULTIPLE TRANSMEMBRANE SPANS AND A CRITICAL CYTOSOLIC LOOP

UDP-GlcNAc:dolichol-P GlcNAc-1-P transferase (GPT) is an endoplasmic r eticulum (ER) enzyme responsible for synthesis of GlcNAc-P-P-dolichol, the committed step of dolichol-P-P-oligosaccharide synthesis. The seq uence of hamster GPT predicted multiple transmembrane segments (Zhu, X , and Lehrman, M. A. (1990) J. Biol. Chem. 265, 14250-14255). GPT has also been predicted to act on the cytosolic face of the ER membrane, b ased on topological studies of its substrates and products. In this re port we test these predictions by: (i) immunofluorescence microscopy w ith antibodies specific for native GPT sequences or epitope tags inser ted into GPT, after selective permeabilization of the plasma membrane with digitonin; (ii) insertion of Factor Xa cleavage sites; (iii) in v itro translation of GPT; and (iv) site-directed mutagenesis, The loops between the Ist and 2nd and between the 9th and 10th predicted transm embrane spans of GPT were found to be cytosolic. In contrast, the loop between the 6th and 7th transmembrane spans, as well as the carboxyl terminus, were lumenal. Thus, hamster GPT must cross the ER membrane a t least three times, consistent with previous computer-assisted predic tions. There was no apparent N-glycosylation or signal sequence cleava ge detected by in vitro translation. The cytosolic loop between the 9t h and 10th transmembrane spans is the largest hydrophilic segment in G PT and, as judged by site-directed mutagenesis, has a number of conser ved residues essential for activity. Hence, these results directly sup port the hypothesis that dolichol-PP-oligosaccharide assembly is initi ated in the cytosol and that a downstream intermediate must translocat e to the lumenal face of the ER membrane.