Golgi complex is a multi-compartmental organelle involved in posttranslational modification and sorting of secretory proteins. We have characterized the distribution in Vero cells of three Golgi cisternal membrane proteins, Sialyltransferase (SialylT, tans- Golgi/TGN), galactosyltransferase(GalT, trans-Golgi/TGN), and Nacetylglucosametransferase- 1 (GlcNAcT-1, medial-Golgi), during the process of Golgi disassembly and reassembly following the addition or removal of drugs. After 1 h nocodazole (microtubule depolymerizer) treatment, SialylT and GalT were found in scattered punctate structures that increased in number over time as less and less of these proteins were found perinuclearly. Initially these punctate structures were often negative for GlcNAcT-1. Over a 2 to 3-fold slower time course, GICNAcT-1 co-localized with SialylT and GalT in the scattered punctate structures. Cis-Golgi network marker was found in a separate set of scattered punctate structures from that of cisternae even at 4 h following nocodazole addition. Following nocodazole removal, all the cisternal markers accumulated perinuclearly into a reassembled Golgi at the same rate. After brefeldin A treatment (coat protein inhibitor), all the cisternae markers dispersed to ER with similar kinetics, albeit, in some cases by different tubular extensions of the Golgi. GlicCNAcT-1 and GalT showed similar kinetics of Golgi reassembly following BFA removal while SialylT lagged somewhat behind. Our data suggest that CGN, medial-Golgi and trans-Golgi/TGN are distinct subcompartments that can be separated one from the other by drug treatment; any exchange of components between the Golgi subcompartments must be slow with respect to the observed kinetics of Golgi disassembly. An epitope tagging approach was used to delineate the importance of the above Golgi protein's cytoplasmic tail domain in Golgi targeting and retention. we found that the cytoplasmic tail could be lengthened considerably (3-4 fold) and SialyIT and GalT still accumulated in a perinuclear, Golgi-like distribution with little ER background. One construct, VSV-SialylT, localized essentially exclusively to the Golgi complex. For SialylT, the longest constructs (40-42 amino acids) located relatively well to the Golgi complex while for GalT, the longest constructs (32-34 amino acids) located exclusively to the ER. Surprisingly, the epitope tags of several different GalT constructs was inaccessible to antibody in fixed cells. Any lengthening of the cytoplasmic domain of GIcNAcT-1 resulted in considerable to exclusive accumulation of the chimeric proteins in ER. No cell surface accumulation of any of the chimeric proteins was detected. The specific sequence of the epitope tag was important; the neutral to positively charged VSV epitope tag was preferred over negatively charged myc or FLAG tags. Depending on the exact tail alteration, we found that all three transferases accumulated in the ER with no detectable Golgi or cell surface accumulation. In some cases accumulation in the intermediate compartment or CGN was observed. Most surprisingly, in the one homologous case studied, expression of chimeric human GlcNAcT-1 in HeLa cells, ER accumulation of GlcNAcT-1 led to disruption of pre-existing Golgi. Based on these results, we propose, as the simplest explanation of the data, that alteration of the cytoplasmic tail of Golgi resident proteins can decrease their rate of exit from ER, presumably due to premature oligomerization, and may lead to capture of Golgi proteins in ER. / Ph. D.