Modification of the cytoplasmic domain affects the subcellular localization of Golgi glycosyl-transferases.

Our goal was to engineer a Golgi glycosyltransferase epitope-tagged on its cytoplasmically exposed, short, N-terminal domain that gave normal subcellular localization. Partial replacement of the cytoplasmic tail of human alpha-2,6-sialyltransferase (SialylT) with the negatively charged myc or FLAG epitope resulted in almost complete mislocalization of the chimera expressed in Vero cells. A granular cytoplasmic staining pattern was seen by immunofluorescence. Spacing the negatively charged residues progressively outward from the negative N-terminus resulted in increasingly more normal localization of myc or FLAG-tagged protein to a juxtanuclear Golgi-like distribution. Substitution of a neutrally charged VSV-G sequence for these tags resulted in normal localization of the chimera to the juxtanuclear Golgi region. Insertion of the myc epitope within the N-terminal domain of the short form of bovine beta-1,4-galactosyltransferase (GalT) gave a chimeric protein that mislocalized in BHK cells. No signal was detected with a monoclonal anti-epitope antibody indicating that the myc epitope was masked. Placement of myc or FLAG epitopes at the NH2-terminus of human N-acetylglucosaminyltransferase I (GlcNAc-T) resulted in chimeric proteins that in Vero cells displayed little Golgi localization. We conclude that positioning of negative charge, in particular, close to the membrane, typically produces a failure of type II Golgi glycosyltransferases to exit the ER/CGN, presumably due to quality control mechanisms. These proteins may be successfully epitope-tagged on their N-terminal domain either using a neutral or positively charged sequence or spacing any negatively charged sequence out from the membrane.