General Characteristics

In mammals, there are more than 250 GTs that reside in the Golgi where they catalyze the transfer of one sugar to another sugar on a glycan acceptor, generally covalently attached to protein or lipid. GTs are grouped into families of related catalytic or sequence activity in the Carbohydrate Active Enzymes database (CAZy) (Cantarel et al. 2009). There are, for example, 20 mammalian sialyltransferases (Takashima 2008). By contrast, Drosophila has a single sialyltransferase (Koles et al. 2004), and yeast have none. Most GTs transfer a single sugar to an acceptor but there are a few GTs that catalyze the transfer of two different sugars, typically to generate a polymer of repeating units, as in proteoglycans (Fig. 1). The Golgi GT is a type II transmembrane protein with a short amino-terminal cytoplasmic tail in the cytosol, a transmembrane domain, a stalk-like stem region, and a globular catalytic domain in the Golgi lumen (Fig. 2). Some GTs contain two functional domains. For example, several polypeptide GalNAcTs have a lectin domain that binds to GalNAc and a catalytic domain that transfers GalNAc (Hassan et al. 2000; Fritz et al. 2006); a family of N-deacetylase/N-sulfotransferases are important in heparan sulfate biosynthesis (Aikawa et al. 2001); and the putative glycosyltransferase Large has two glycosyltransferase domains (Longman et al. 2003; Aguilan et al. 2009). GTs are often glycosylated by other GTs or, in some cases, by autocatalytic transferase activity. The glycans on some GTs are required for their correct folding during synthesis, and these GTs often lack catalytic activity when generated in bacteria. Interestingly, many GTs are cleaved in their stem region by proteases in the Golgi and secreted from the cell (Shifley and Cole 2008; Kitazume et al. 2009). This may be a mechanism to regulate their activity in the Golgi, or to allow them to act at the cell surface or in the extracellular environment. The latter would require nucleotide sugars to be available outside the cell. Intriguingly, nucleotide sugar transporters of the Golgi have been found to facilitate nucleotide sugar release from the cell (Sesma et al. 2009).

Open in a separate window

Figure 2.

Glycosyltransferases and nucleotide sugar transporters of the Golgi. The diagram depicts a variety of nucleotide sugar transporters that transfer a nucleotide sugar from the cytoplasm into the Golgi lumen in exchange for a nucleotide monophosphate generated by hydrolysis of the nucleotide diphosphate released after transfer of the sugar to an acceptor. A typical type II transmembrane glycosyltransferase (a sialyltransferase; gray) with its short cytoplasmic tail, transmembrane domain, extended stem region, and globular catalytic domain is shown binding CMP-Sia from which it transfers Sia to Gal on a complex N-glycan on a glycoprotein (green). Abbreviations are defined in the legend to Figure 1. (Modified from Figure 4.4 in Essentials of glycobiology, with permission from Freeze and Elbein 2009.)

N-Glycosylation

Glycoproteins that are N-glycosylated arrive in the cis-Golgi carrying N-glycans added cotranslationally in the ER (Kelleher and Gilmore 2006) to a proportion of their Asn-X (not Pro)-Ser/Thr (rarely Cys) sites (Zielinska et al. 2010). The consensus motif for N-glycosylation is necessary but not sufficient for N-glycosylation to occur. The N-glycans on glycoproteins in the cis-Golgi are of the high mannose type and usually contain eight or nine mannose (Man) residues (Stanley et al. 2009). These may remain unchanged during passage through the Golgi and be present on cell surface or secreted glycoproteins (Fig. 1). However, N-glycans are often processed in the Golgi, initially in the cis-Golgi by a set of α-mannosidases that remove Man residues to generate the Man₅GlcNAc₂Asn intermediate which is the substrate of the medial Golgi GT GlcNAcT-I. The transfer of GlcNAc to Man₅GlcNAc₂Asn by GlcNAcT-I initiates the synthesis of hybrid and complex N-glycans. Hybrid N-glycans keep the five Man residues and extend the arm that received GlcNAc by adding Gal and sialic acid and/or other sugars. To become complex, N-glycans lose the terminal two of the five Man residues and acquire a second GlcNAc to form a biantennary, complex N-glycan (Fig. 1). This may be further branched up to six times and each branch may be elongated by the addition of different sugars including Gal, GlcNAc, GalNAc, Fuc, Sia, and disaccharide units (Stanley et al. 2009). Even a simple CHO cell has complex N-glycans that may carry 60 or more sugar residues (North et al. 2010).

Drosophila glycoproteins carry mainly high mannose N-glycans sensitive to cleavage by endoglycosidase H (Endo H), but may also carry complex N-glycans (Aoki et al. 2007; Koles et al. 2007). The complex N-glycans have few sugars. Some neural glycoproteins have N-glycans with terminal sialic acid, and other glycoproteins contain glucuronic acid in antennae and fucose in the core region of N-glycans (Aoki and Tiemeyer 2010; Rendic et al. 2010). C. elegans has high mannose and complex N-glycans with some specialized features such as the incorporation of phosphorylcholine on GlcNAc (Cipollo et al. 2005). Yeast glycoproteins enter the cis-Golgi with mainly Man₈GlcNAc₂Asn at N-glycan sites, the same as Drosophila and mammals. However, these N-glycans quickly acquire additional Man residues and ultimately form yeast mannan with long chains of Man that may be substituted with phosphate (Munro 2001).

Regulation of Golgi Glycosylation

Glycosylation in the Golgi is controlled by the spectrum of GTs, nucleotide sugar synthases, and transporters expressed by a cell, as well as factors that affect the lumenal Golgi environment, and the structure and organization of Golgi membranes. Changes of Golgi pH have long been known to alter glycosylation, and an anion channel has now been identified that regulates Golgi pH (Maeda et al. 2008; Maeda and Kinoshita 2010). Cells lacking this ion channel transport glycoproteins slowly through the secretory pathway, and the glycosylation of both glycoproteins and glycolipids is truncated in these cells. Interestingly, passage through the Golgi is also slowed if the CMP-Sia or GDP-fucose transporters or both are knocked down using RNAi, or by CMP-Sia transporter mutation in Lec2 CHO cells (Xu et al. 2010). In cells with reduced nucleotide sugar transport, glycoproteins with truncated glycans accumulate in the Golgi rather than proceeding to the plasma membrane or to secretion. In addition, ER stress pathways are activated in cells in which nucleotide sugar transport is inhibited (Xu et al. 2010). In Drosophila cells in culture, separate knockdown of two different ppGalNAcTs causes a slowdown in secretion and an alteration in Golgi organization, as well as reduced transfer of GalNAc to mucin glycoproteins (Zhang and Ten Hagen 2010). Thus, Golgi resident proteins appear to contribute to the overall integrity and function of the Golgi.

Other factors in the Golgi lumen that are important for glycosylation are the nucleotide sugar pyrophosphorylases that hydrolyse released nucleotide diphosphates, as their number and activity will affect nucleotide sugar import (Berninsone and Hirschberg 2000; Fig. 2). Chaperones important for certain glycosyltransferases to leave the ER may also function in the Golgi. For example, COSMC is necessary for T-synthase to move along the secretory pathway and to be active in the Golgi (Wang et al. 2010). In another example, the putative glycosyltansferase Large must physically associate with α-dystroglycan to be functional in modifying this substrate in the Golgi (Kanagawa et al. 2004). Recently, inhibitors of glycosyltransferase activities have been discovered. One of these inhibits the activity of GlcNAcT-I, the transferase that initiates the synthesis of complex and hybrid N-glycans (Huang and Stanley 2010). It is a testis-specific glycoprotein that binds to GlcNAcT-I as well as some other medial Golgi enzymes, but appears to inactivate only GlcNAcT-I. It is termed GlcNAcT-I Inhibitory Protein (GnT1IP) and its expression is tightly regulated during spermatogenesis. Another example is a class of proteins of the transmembrane BAX inhibitor motif-containing (TMBIM) family that inhibit Gb3 synthase that generates the glycolipid Gb3, the receptor of Shiga toxin (Yamaji et al. 2010). Yet another mechanism of glycosylation regulation is to change the Golgi content of a glycosyltransferase by relocation. This has been recently described for several ppGalNAcTs that transfer GalNAc to Ser or Thr and initiate mucin glycan synthesis (Gill et al. 2010). Growth factor stimulation of Src tyrosine kinase activity, or injection of activated Src kinase, causes a large proportion of all ppGalNAcTs tested to shift from the Golgi to the ER, which results in an increase in total GalNAc addition to proteins. This is a COP1-mediated process and is dependent on the Arf1 GTPase. Medial Golgi enzymes tested, or the Golgi tethering protein giantin, were not redistributed by Src activation.

Mutations in proteins of the conserved oligomeric Golgi (COG) complex that provides a scaffold important for Golgi membrane structure and tethering of retrograde vesicles, also cause alterations in glycosylation (Smith and Lupashin 2008). The first member of the eight subunits of the COG complex was discovered in a CHO mutant ldlC that has defects in N- and O-glycan and glycolipid synthesis (Podos et al. 1994). Several COG subunits have now been shown to be mutated and to give rise to glycosylation defects in patients with congenital diseases of glycosylation (Zeevaert et al. 2008; Foulquier 2009; Lubbehusen et al. 2010). The mechanism by which COG defects alter multiple glycosylation pathways appears to be caused by partial relocation and degradation of Golgi glycosyltransferases and other glycosylation activities when COG is dysfunctional (see Freeze and Elbein 2009).