Golgi Glycosyltransferases Research Articles

ST6Gal-I expression in ovarian cancer cells promotes an invasive phenotype by altering integrin glycosylation and function

BackgroundOvarian adenocarcinoma is not generally discovered in patients until there has been widespread intraperitoneal dissemination, which is why ovarian cancer is the deadliest gynecologic malignancy. Though incompletely understood, the mechanism of peritoneal metastasis relies on primary tumor cells being able to detach themselves from the tumor, escape normal apoptotic pathways while free floating, and adhere to, and eventually invade through, the peritoneal surface. Our laboratory has previously shown that the Golgi glycosyltransferase, ST6Gal-I, mediates the hypersialylation of β1 integrins in colon adenocarcinoma, which leads to a more metastatic tumor cell phenotype. Interestingly, ST6Gal-I mRNA is known to be upregulated in metastatic ovarian cancer, therefore the goal of the present study was to determine whether ST6Gal-I confers a similarly aggressive phenotype to ovarian tumor cells.MethodsThree ovarian carcinoma cell lines were screened for ST6Gal-I expression, and two of these, PA-1 and SKOV3, were found to produce ST6Gal-I protein. The third cell line, OV4, lacked endogenous ST6Gal-I. In order to understand the effects of ST6Gal-I on cell behavior, OV4 cells were stably-transduced with ST6Gal-I using a lentiviral vector, and integrin-mediated responses were compared in parental and ST6Gal-I-expressing cells.ResultsForced expression of ST6Gal-I in OV4 cells, resulting in sialylation of β1 integrins, induced greater cell adhesion to, and migration toward, collagen I. Similarly, ST6Gal-I expressing cells were more invasive through Matrigel.ConclusionST6Gal-I mediated sialylation of β1 integrins in ovarian cancer cells may contribute to peritoneal metastasis by altering tumor cell adhesion and migration through extracellular matrix.

Signal-Mediated Dynamic Retention of Glycosyltransferases in the Golgi

Golgi-resident glycosyltransferases are a family of enzymes that sequentially modify glycoproteins in a subcompartment-specific manner. These type II integral membrane proteins are characterized by a short cytoplasmically exposed amino-terminal tail and a luminal enzymatic domain. The cytoplasmic tails play a role in the localization of glycosyltransferases, and coat protein complex I (COPI) vesicle-mediated retrograde transport is also involved in their Golgi localization. However, the tails of these enzymes lack known COPI-binding motifs. Here, we found that Vps74p bound to a pentameric motif present in the cytoplasmic tails of the majority of yeast Golgi-localized glycosyltransferases, as well as to COPI. We propose that Vps74p maintains the steady-state localization of Golgi glycosyltransferases dynamically, by promoting their incorporation into COPI-coated vesicles.

Calsenilin and CALP interact with the cytoplasmic tail of UDP-Gal:GA2/GM2/GD2 β-1,3-galactosyltransferase

GalT2 (UDP-Gal:GA2/GM2/GD2 beta-1,3-galactosyltransferase) is a Golgi-resident type II membrane protein that participates in the synthesis of glycosphingolipids. The molecular determinants for traffic and localization of this and other glycosyltransferases are still poorly characterized. Considering the possibility that interactions with other proteins may influence these processes, in the present study we carried out a yeast two-hybrid screening using elements of the N-terminal domain of GalT2 as bait. In this screening, we identified calsenilin and its close homologue CALP (calsenilin-like protein), both members of the recoverin-NCS (neuronal calcium sensor) family of calcium-binding proteins. In vitro, GalT2 binds to immobilized recombinant CALP, and CALP binds to immobilized peptides with the GalT2 cytoplasmic tail sequence. GalT2 and calsenilin interact physically when co-expressed in CHO (Chinese-hamster ovary)-K1 cells. The expression of CALP or calsenilin affect Golgi localization of GalT2, and of two other glycosyltransferases, SialT2 (CMP-NeuAc:GM3 sialyltransferase) and GalNAcT (UDP-GalNAc:lactosylceramide/GM3/GD3 beta1-4 N-acetylgalactosaminyltransferase), by redistributing them from the Golgi to the ER (endoplasmic reticulum), whereas the localization of the VSV-G (G-protein of the vesicular stomatitis virus) or the Golgin GM130 was essentially unaffected. Conversely, the expression of GalT2 affects the localization of calsenilin and CALP by shifting a fraction of the molecules from being mostly diffuse in the cytosol, to clustered structures in the perinuclear region. These combined in vivo and in vitro results suggest that CALP and calsenilin are involved in the trafficking of Golgi glycosyltransferases.

Golgi Localization of Glycosyltransferases Requires a Vps74p Oligomer

The mechanism of glycosyltransferase localization to the Golgi apparatus is a long-standing question in secretory cell biology. All Golgi glycosyltransferases are type II membrane proteins with small cytosolic domains that contribute to Golgi localization. To date, no protein has been identified that recognizes the cytosolic domains of Golgi enzymes and contributes to their localization. Here, we report that yeast Vps74p directly binds to the cytosolic domains of cis and medial Golgi mannosyltransferases and that loss of this interaction correlates with loss of Golgi localization of these enzymes. We have solved the X-ray crystal structure of Vps74p and find that it forms a tetramer, which we also observe in solution. Deletion of a critical structural motif disrupts tetramer formation and results in loss of Vps74p localization and function. Vps74p is highly homologous to the human GMx33 Golgi matrix proteins, suggesting a conserved function for these proteins in the Golgi enzyme localization machinery.

Length and Composition Analysis of the Cytoplasmic, Transmembrane and Stem Regions of Human Golgi Glycosyltransferases

A dataset of experimentally characterized, human Golgi GlyTs with type II membrane topology was created. Based on the experimentally observed acceptor substrate preferences, the GlyTs were classified into five functional categories: biosynthesis of blood group antigens, glycolipids, N-glycans, O-glycans and glycosaminoglycans. The cytoplasmic, transmembrane and stem (CTS) regions were predicted and their length and composition were analyzed. The stem region of GlyTs involved in the biosynthesis of glycolipids and blood group antigens appear to have a shorter stem region compared to those GlyTs which participate in the biosynthesis of N- and O-linked glycans and glycosaminoglycans. The stem regions of all the GlyTs, irrespective of the functional category to which they belong, were found to be rich in disorder-promoting amino acid residues. Thus, the stem region is largely devoid of any regular secondary structure thereby facilitating its tethering role. A higher frequency of occurrence of basic amino acids is observed towards the N-terminus of the transmembrane domain and this is suggested to be important for topogenesis of these enzymes.

Catalytic Mechanism of Glycosyltransferases: Hybrid Quantum Mechanical/Molecular Mechanical Study of the Inverting N-Acetylglucosaminyltransferase I

The Golgi glycosyltransferase, N-acetylglucosaminyltransferase I (GnT-I), catalyzes the transfer of a GlcNAc residue from the donor UDP-GlcNAc to the C2-hydroxyl group of a mannose residue in the trimannosyl core of the Man5GlcNAc2-Asn-X oligosaccharide. The catalytic mechanism of GnT-I was investigated using a hybrid quantum mechanical/molecular mechanical (QM/MM) method with a QM part containing 88 atoms treated with density functional theory (DFT) at the BP/TZP level. The remaining parts of a GnT-I complex, altogether 5633 atoms, were modeled using the AMBER molecular force field. A theoretical model of a Michaelis complex was built using the X-ray structure of GnT-I in complex with the donor having geometrical features consistent with kinetic studies. The QM(DFT)/MM model identified a concerted SN2-type of transition state with D291 as the catalytic base for the reaction in the enzyme active site. The TS model features nearly simultaneous nucleophilic addition and dissociation steps accompanied by the transfer of the nucleophile proton Hb2 to the catalytic base D291. The structure of the TS model is characterized by the Ob2-C1 and C1-O1 bond distances of 1.912 and 2.542 A, respectively. The activation energy for the proposed reaction mechanism was estimated to be approximately 19 kcal mol-1. The calculated alpha-deuterium kinetic isotope effect of 1.060 is consistent with the proposed reaction mechanism. Theoretical results also identified interactions between the Hb6 and beta-phosphate oxygen of the UDP and a low-barrier hydrogen bond between the nucleophile and the catalytic base D291. It is proposed that these interactions contribute to a stabilization of TS. This modeling study provided detailed insight into the mechanism of the GlcNAc transfer catalyzed by GnT-I, which is the first step in the conversion of high mannose oligosaccharides to complex and hybrid N-glycan structures.

Capacity of the Golgi Apparatus for Cargo Transport Prior to Complete Assembly

In yeast, particular emphasis has been given to endoplasmic reticulum (ER)-derived, cisternal maturation models of Golgi assembly while in mammalian cells more emphasis has been given to golgins as a potentially stable assembly framework. In the case of de novo Golgi formation from the ER after brefeldin A/H89 washout in HeLa cells, we found that scattered, golgin-enriched, structures formed early and contained golgins including giantin, ranging across the entire cis to trans spectrum of the Golgi apparatus. These structures were incompetent in VSV-G cargo transport. Second, we compared Golgi competence in cargo transport to the kinetics of addition of various glycosyltransferases and glycosidases into nascent, golgin-enriched structures after drug washout. Enzyme accumulation was sequential with trans and then medial glycosyltransferases/glycosidases found in the scattered, nascent Golgi. Involvement in cargo transport preceded full accumulation of enzymes or GPP130 into nascent Golgi. Third, during mitosis, we found that the formation of a golgin-positive acceptor compartment in early telophase preceded the accumulation of a Golgi glycosyltransferase in nascent Golgi structures. We conclude that during mammalian Golgi assembly components fit into a dynamic, first-formed, multigolgin-enriched framework that is initially cargo transport incompetent. Resumption of cargo transport precedes full Golgi assembly.

The synthesis of novel N‐glycans in Entamoeba histolytica and the quality control, secretion and cell surface arrangement of glycoproteins

Entamoeba histolytica is a protozoan parasite, which causes ulcerative colitis and liver disease in vast regions of the underdeveloped world. Glycoconjugates in this cytotoxic parasite are of great interest due to their role in adhesion, colonization and cell lysis. The Entamoeba genome predicts 1) a truncated Alg pathway, 2) protein reglucosylation, and 3) two Golgi glycosyltransferases. We determined that the trophozoite invading stage has a functional set of enzymatic activities able to synthesize, fold and properly export glycoproteins carrying a variety of N-glycans. We describe the trimming and elongation processes that lead to the formation of complex type N-glycans with novel extensions. The structures of high mannose N-glycans and at least 13 complex-n-glycans were fully characterized, with others partially described. The activity of novel processing enzymes was detected as well. With a limited repertoire of glycosyltransferases and seemingly simple secretory organelles, the trophozoites are able to synthesize and export a multiplicity of N- and O-glycans. Compartmentalization, thus, is highly regulated. The results demonstrate that N-glycan biosynthesis is relevant not only to the stability of secreted proteins, but also to the makeup of the parasite’s surface and to the interaction with the host tissues.

COG Complex‐Mediated Recycling of Golgi Glycosyltransferases is Essential for Normal Protein Glycosylation

Defects in conserved oligomeric Golgi (COG) complex result in multiple deficiencies in protein glycosylation. On the other hand, acute knock-down (KD) of Cog3p (COG3 KD) causes accumulation of intra-Golgi COG complex-dependent (CCD) vesicles. Here, we analyzed cellular phenotypes at different stages of COG3 KD to uncover the molecular link between COG function and glycosylation disorders. For the first time, we demonstrated that medial-Golgi enzymes are transiently relocated into CCD vesicles in COG3 KD cells. As a result, Golgi modifications of both plasma membrane (CD44) and lysosomal (Lamp2) glycoproteins are distorted. Localization of these proteins is not altered, indicating that the COG complex is not required for anterograde trafficking and accurate sorting. COG7 KD and double COG3/COG7 KD caused similar defects with respect to both Golgi traffic and glycosylation, suggesting that the entire COG complex orchestrates recycling of medial-Golgi-resident proteins. COG complex-dependent docking of isolated CCD vesicles was reconstituted in vitro, supporting their role as functional trafficking intermediates. Altogether, the data suggest that constantly cycling medial-Golgi enzymes are transported from distal compartments in CCD vesicles. Dysfunction of COG complex leads to separation of glycosyltransferases from anterograde cargo molecules passing along secretory pathway, thus affecting normal protein glycosylation.

The Steady‐State Distribution of Glycosyltransferases Between the Golgi Apparatus and the Endoplasmic Reticulum is Approximately 90:10

Several lines of evidence support a novel model for Golgi protein residency in which these proteins cycle between the Golgi apparatus and the endoplasmic reticulum (ER). However, to preserve the functional distinction between the two organelles, this pool of ER-resident Golgi enzymes must be small. We quantified the distribution for two Golgi glycosyltransferases in HeLa cells to test this prediction. We reasoned that best-practice, quantitative solutions would come from treating images as data arrays rather than pictures. Using deconvolution and computer calculated organellar boundaries, the Golgi fraction for both endogenous beta1,4-galactosyltransferase and UDP-N-acetylgalactosamine:polypeptide N-acetylgalactosaminyltransferase 2 fused with green fluorescent protein (GFP) was 91% by fluorescence microscopy. Immunogold labeling followed by electron microscopy and model analysis yielded a similar value. Values reflect steady-state conditions, as inclusion of a protein synthesis inhibitor had no effect. These data strongly suggest that the fluorescence of a GFP chimera with an organellar protein can be a valid indicator of protein distribution and more generally that fluorescent microscopy can provide a valid, rapid approach for protein quantification. In conclusion, we find the ER pool of cycling Golgi glycosyltransferases is small and approximately 1/100 the concentration found in the Golgi apparatus.

Modification of Plant N‐Glycans Processing: The Future of Producing Therapeutic Protein by Transgenic Plants

AbstractFor Abstract see ChemInform Abstract in Full Text.

Endoplasmic reticulum retention of the large splice variant of the UDP-galactose transporter is caused by a dilysine motif

Nucleotide-sugar transporters supply mainly the Golgi glycosyltransferases with substrates. Some glycosyltransferases in the endoplasmic reticulum (ER), however, also use activated sugars. Recent studies have demonstrated that UDP-galactose (UDP-Gal) is the substrate for the ER resident ceramide-galactosyltransferase (cer-GalT) and cells expressing cer-GalT are able to retain the UDP-Gal transporter (UGT) by physical contacts formed between the two proteins. Here, we describe a second active mechanism for ER localization of the UGT. The UGT is produced in two splice forms UGT1 and UGT2. The proteins vary only at their extreme C-termini but show strikingly different intracellular distribution. Although N-terminally epitope tagged forms of UGT1 localize exclusively to the Golgi, similar constructs of UGT2 show both ER and Golgi localization. The dilysine motif KVKGS contained in UGT2 can be demonstrated to be responsible for the dual localization because: (1) disturbance of the signal via site specific mutation or C-terminal extension completely shifts the transporter to the Golgi, (2) transfer of the dilysine motif is sufficient to redistribute the Golgi CMP-sialic acid transporter to the ER, and (3) replacement of KVKGS by the strong ER retention signal KKNT is sufficient to completely retain UGT2 in the ER.

Hypersialylation of β1 Integrins, Observed in Colon Adenocarcinoma, May Contribute to Cancer Progression by Up-regulating Cell Motility

Colon adenocarcinomas are known to express elevated levels of alpha2-6 sialylation and increased activity of ST6Gal-I, the Golgi glycosyltransferase that creates alpha2-6 linkages. Elevated ST6Gal-I positively correlates with metastasis and poor survival, and therefore ST6Gal-I-mediated hypersialylation likely plays a role in colorectal tumor invasion. Previously we found that oncogenic ras (present in roughly 50% of colon adenocarcinomas) up-regulates ST6Gal-I and, in turn, increases sialylation of beta1 integrin adhesion receptors in colon epithelial cells. However, we wanted to know if this pattern held true in vivo and, if so, how beta1 hypersialylation might contribute to colon tumor progression. In the present study, we find that beta1 integrins from colon adenocarcinomas consistently carry higher levels of alpha2-6 sialic acid. To explore the effects of increased alpha2-6 sialylation on beta1-integrin function, we stably expressed ST6Gal-I in a colon epithelial cell line lacking endogenous ST6Gal-I. ST6Gal-I expressors (with alpha2-6 sialylated beta1 integrins) exhibited up-regulated attachment to collagen I and laminin and increased haptotactic migration toward collagen I, relative to parental cells (with completely unsialylated beta1 integrins). Blockade of ST6Gal-I expression with short interfering RNA reversed collagen binding back to the level of ST6Gal-I nonexpressors, confirming that alpha2-6 sialylation regulates beta1 integrin function. Finally, we show that beta1 integrins from ST6Gal-I expressors have increased association with talin, a marker for integrin activation. Collectively, these findings suggest that beta1 hypersialylation may augment colon tumor progression by altering cell preference for certain extracellular matrix milieus, as well as by stimulating cell migration.

Cisternal Rab Proteins Regulate Golgi Apparatus Redistribution in Response to Hypotonic Stress

We show that a physiological role of the extensively studied cisternal Golgi rab protein, rab6, is modulation of Golgi apparatus response to stress. Taking exposure of cells to hypotonic media as the best-known example of mammalian Golgi stress response, we found that hypotonic-induced tubule extension from the Golgi apparatus was sensitive to GDP-rab6a expression. Similarly, we found that Golgi tubulation induced by brefeldin A, a known microtubule-dependent process, was inhibited by GDP-restricted rab6a, rab6a', and rab33b, the most commonly studied cisternal rab proteins. These GDP-rab levels were sufficient to inhibit rab-induced redistribution of Golgi glycosyltransferases into the endoplasmic reticulum (ER), also a microtubule-dependent process, and to depress Golgi membrane association of the GTP-conformer of rab6. Nocodazole-induced Golgi scattering, a microtubule-independent process, also was inhibited by GDP-rab6a expression. In comparison, we found similar GDP-rab expression levels had little inhibitory effect on another microtubule-independent process, constitutive recycling of Golgi resident proteins to the ER. We conclude that Golgi cisternal rabs, and in particular rab6a, are regulators of the Golgi response to stress and presumably the molecular targets of stress-activated signaling pathway(s). Moreover, we conclude that rab6a can regulate select microtubule-independent processes as well as microtubule-dependent processes.

Structural features of the lysosomal hydrolase mannose 6-phosphate uncovering enzyme

The uncovering enzyme (UCE) removes N-acetylglucosamine from lysosomal enzymes to uncover the mannose 6-phosphate (Man-6-P) determinant necessary for targeting these enzymes to lysosomes. Failure to create the Man-6-P determinant is one cause of lysosomal storage diseases. Despite its medical importance, little structural information about UCE is available. In this report we have developed a model for the membrane proximal portion of the lumenal domain of UCE based on the structure of the EFG-3 and -4 domains of the extracellular segment of the beta chain of integrin alpha Vbeta 3. In this model the EGF-like domains of UCE (residues 285-345) are predicted to form a rod-shaped stalk region, similar to the stem region in Golgi glycosyltransferases. This stalk causes the proposed catalytic domain (residues 1-277) to be extended away from the Golgi membrane. A portion of the proposed catalytic domain (residues 85-256) resides in Cluster of Orthologous Group (COG) 4632 with four bacterial proteins but is not homologous to any known eukaryotic proteins. Thus, UCE may have evolved from the fusion of a unique catalytic domain with a common EGF-like stalk domain. We have determined by mass spectrometry that the four disulfide bonds of the proposed catalytic domain are located between Cys(2)-Cys(172), Cys(66)-Cys(99), Cys(83)-Cys(274), and Cys(258)-Cys(265). Finally, we determined that four of the six potential N-linked glycosylation sites are glycosylated (Asn 159, Asn 165, Asn 247, and Asn 317) in COS cells.

Modification of plant N-glycans processing: The future of producing therapeutic protein by transgenic plants

Transgenic plants are regarded as one of the most promising systems for the production of human therapeutic proteins. The number of therapeutic proteins successfully produced in plants is steadily arising. However, the glycoproteins normally produced from plants are not the same as native therapeutic proteins produced from mammals or humans. In addition to in vitro enzymatic modeling glycoproteins, there are two gene manipulation strategies to humanize plant N-glycans connected to the glycoproteins. One is retaining the recombinant glycoproteins in endoplasmic reticulum (ER), the site where few specific modifications of N-glycans occurs. The other is inhibiting the plant endogenous Golgi glycosyltransferase and/or adding new glycosyltransferase from mammalians. In this review, the biosynthesis of N-glycans in plants, the modification of the plant N-glycans processing will be discussed.

Organization of Golgi glycosyltransferases in membranes: complexity via complexes.

Organization of Golgi glycosyltransferases in membranes: complexity via complexes.

Persistence of Golgi matrix distribution exhibits the same dependence on Sar1p activity as a Golgi glycosyltransferase.

We investigated the relative distributional persistence of Golgi "matrix" proteins and glycosyltransferases to an endoplasmic reticulum exit block induced by expression of a GDP-restricted Sar1p. HeLa cells were microinjected with plasmid encoding the GDP-restricted mutant (T39N) of Sar1p to block endoplasmic reticulum exit and then scored for the distribution of GM130 (Golgi matrix protein of 130 kDa), a cis located golgin; p27, a member of the p24 family of proteins; giantin, a protein that interacts indirectly with GM130; and the Golgi glycosyltransferase, N-acetylgalactosaminyltransferase-2 (GalNAcT2). All of these proteins lost their compact, juxtanuclear distribution and displayed characteristics of endoplasmic reticulum/cytoplasmic accumulation with the same dependence on plasmid concentration. The kinetics of redistribution of GM130 and GalNAcT2 were identical. Expression of Sar1pT39N displaced the COPII coat protein Sec13p from endoplasmic reticulum exit sites consistent with disruption of these sites. This occurred without disturbing the overall distribution of endoplasmic reticulum membrane. Furthermore, the reassembly of a juxtanuclear Golgi matrix as assayed by the distribution of GM130 following washout of the Golgi disrupting drug, brefeldin A, was blocked by microinjected Sar1pT39N plasmids. We conclude that the persistence, i.e. stability and maintenance, of Golgi matrix distribution and its reassembly following drug disruption are exquisitely dependent on Sar1p activity.

An improved chemical fixation method suitable for immunogold localization of green fluorescent protein in the Golgi apparatus of tobacco Bright Yellow (BY-2) cells.

In plant systems, the green fluorescent protein (GFP) is increasingly used as a marker to study dynamics of the secretory apparatus using fluorescence microscopy. The purpose of this study was to immunogold localize the GFP, at the electron microscopic level, in a line of tobacco BY-2-cultured cells, expressing a GFP-tagged Golgi glycosyltransferase. To this end we have developed a simple, one-step chemical fixation method that allow good structural preservation and specific labeling with anti-GFP antibodies. Using this method, we have been able to show that an N-glycan GFP-tagged xylosyltransferase is specifically associated with Golgi stacks of BY-2 transformed cells and is preferentially located in medial cisternae. As an alternative to cryofixation methods, such as high-pressure freezing, which requires specialized and expensive equipment not available in most laboratories, this method offers researchers the opportunity to investigate GFP-tagged proteins of the endomembrane system in tobacco BY-2 cells.

Different glycosyltransferases are differentially processed for secretion, dimerization, and autoglycosylation.

Modification of Golgi glycosyltransferases, such as formation of disulfide-bonded dimers and proteolytical release from cells as a soluble form, are important processes to regulate the activity of glycosyltransferases. To better understand these processes, six glycosyltransferases were selected on the basis of the donor sugars, including two N-acetylglucosaminyltransferases, core 1 beta1,3-N-acetylglucosaminyltransferase (C1-beta3GnT) and core 2 beta1,6-N-acetylglucosaminyltransferase (C2GnT-I); two fucosyltransferases, alpha1,2-fucosyltransferase-I (FucT-I) and alpha1,3-fucosyltransferase-VII (FucT-VII); and two sialyltransferases, alpha2,3-sialyltransferase-I (ST3Gal-I) and alpha2,6-sialyltransferase-I (ST6Gal-I). These enzymes were fused with enhanced green fluorescence protein and stably expressed in Chinese hamster ovary cells. Spectrofluorimetric detection and immunoblotting analyses showed that all of these glycosyltransferases except FucT-VII were secreted in the medium. By examining dimers formed in cells and culture media, we found that all of the enzymes, except ST3Gal-I, form a combination of monomers and dimers in cells, whereas the molecules released in the media are either exclusively monomers (C2GnT-I and ST6Gal-I), dimers (FucT-I) or a mixture of both (C1-beta3GnT). These results indicate that dimerization does not always lead to Golgi retention. Analysis of the N-glycosylation status of the enzymes revealed that the secreted proteins are generally more heavily N-glycosylated and sialylated than their membrane-associated counterparts, suggesting that the proteolytic cleavage occurs before the glycosylation is completed. Using FucT-I and ST6Gal-I as a model, we also show that these glycosyltransferases are able to perform autoglycosylation in the dimeric forms. These results indicate that different glycosyltranferases differ significantly in dimerization, proteolytic digestion and secretion, and autoglycosylation. These results strongly suggest that disulfide-bonded dimerization and secretion differentially plays a role in the processing and function of different glycosyltransferases in the Golgi apparatus.