1. Introduction
Proteins may carry glycan chains linked by O- or N-glycosidic bonds (see Article 64, Structure/function of N -glycans, Volume 6) to the protein backbone. The biosynthesis and processing for each of these two types of glycan is quite different, reflecting discrete regulation of structure and biological function. The synthesis and processing of O – and N -glycans occurs in the endoplasmic reticulum and the Golgi through the controlled action of a series of glycosyltransferase and glycosidase reactions. The N-glycan chains are assembled through lipid (dolichol) intermediates and further processed by a series of specific glycosidases (see Article 64, Structure/function of N -glycans, Volume 6). This is not the case for the O-glycans that do not rely on lipid intermediates.
The formation and processing of N- and O-glycans is regulated at the level of gene expression, mRNA, and enzyme protein activity. Additional control exists through substrate and cofactor concentrations at the subcellular site of synthesis. These processes are integrated to yield the huge variety of glycan structures utilized by different species, tissue types, cell types, and observed in different states of development and differentiation.
This short review focuses on the formation and processing of the Ser (Thr)-GalNAc class of O-glycans and those biological processing events for specific proteins that depend on the occurrence and manipulation of their O-linked oligosaccharides. Further examples of O-glycan processing events can be found for N-acetylglucosamine linked to serine or threonine (O-GlcNAc) on cytosolic and nuclear proteins (see Article 72, O-linked N -acetylglucosamine (O-GlcNAc), Volume 6) and fucose and glucose O-linked to serine or threonine in the epidermal growth factor domains in a variety of proteins.
2. Processing of O-glycan chains
The assembly of the Ser (Thr)-GalNAc class of O -glycans is achieved by the action of a series of biosynthetic pathways that are located in regions of the endoplasmic reticulum and Golgi apparatus (see Article 65, Structure/function of O-glycans, Volume 6). These pathways involve largely glycosyltransferases, but also include O-glycan specific sulfotransferases and O-acetyltransferases, leading to the formation of defined oligosaccharide sequences, which have been confirmed by structural analysis (Brockhausen, 1999). The pathways lead to predictable structures with core, backbone, and peripheral units that are found linked to specific proteins. The subcellular organization of the enzymes, in particular the glycosyltransferases, into functional units acting on individual proteins is believed to deliver the specific glycosylation patterns found for each protein (Brockhausen, 1999). The examination of the biological function of individual proteins and the status of their glycans has revealed that processing of the oligosaccharide chains frequently plays a role (Van den Steen et al., 1998). The processing of O-glycans to generate biologically active glycoproteins may arise in several ways. Firstly, the attachment of O -glycans due to the sequential action of glycosyltransferases by the known pathways may deliver completed structures through the action of all enzymes in the pathways. However, in some cases, “incomplete” or truncated structures are formed, which have biological function. These glycans arise because of the action of a reduced number of glycosyltransferases in the biosynthetic units. The addition of backbone and/or peripheral monosaccharides does not occur and shortened or “incomplete” oligosaccharides result. This is the case for IgA hinge region O-glycans (Novak etal., 2000). Alternatively, the processing of completed O -glycan chains may be made by the action of catabolic enzymes including glycosidases, sulfatases, or esterases. Thus, O-glycans can be processed while attached to proteins in a manner analogous to the normal catabolic pathways responsible for the degradation and recycling of glycoproteins. These events usually involve removal of terminal residues such as sialic acids, fucose, or sulfate but can also include the total removal of O -glycan chains through the combined action of several exoglycosidases such as sialidases, fucosidase, and galactosidases together with endoglycosidases and O -glycanases responsible for the internal cleavage of the oligosaccharide chains or their removal for the peptide backbone respectively. The literature describing such events suggests that a variety of glycoproteins are associated with O-glycan processing phenomena and some examples are given below.
3. The role of O-linked glycosylation in protein expression and processing
Many O-glycosylated glycoproteins that are known to be biologically processed rely on their O-glycans to achieve these events. This posttranslational modification is an integral step enabling the function of the mature glycoprotein in vivo. Examples have been reported in several contexts and are grouped below to represent a selection from the current literature. Some of these aspects can also be found summarized in Article 65, Structure/function of O-glycans, Volume 6. In addition, glycoproteins with both O – and N-glycans are common and differential roles for the glycan chains relating to processing and other biological functions have been reported.
3.1. Protein synthesis and expression
The expression of proteins can be modulated at different stages by the presence of glycan chains. The synthesis of mature glycophorin A has been found to be dependent on the presence of O -glycans when expressed in glycosylation-deficient CHO cells. In this case, the N-glycans normally present in glycophorin A are not necessary for expression (Remaley et al., 1991). Further investigation showed that expression of glycophorin A without O -glycans could be achieved provided specific N-glycan structures were present. Thus, independent expression of O- and N-glycans can enable glycophorin A expression.
3.2. Proteolytic processing
A number of proteins depend on the addition of O -glycans at specific sites in order to prevent proteolytic cleavage destroying biological activity or preventing stable residence of the protein at its required subcellular location.
Insulin-like growth factor-2 (IGF-2) is expressed in most embryonic tissues and is required for normal development during gestation but is only expressed in adults during tumorigenesis. Two forms occur in serum – a big IGF-2 and a normal form. Proteolytic cleavage occurs to convert big to normal form, and this is mediated at Thr75 by the presence of an O-glycan. Big IGF-2 has no O-glycan and is not clipped (Duguay et al., 1998).
The expression of LDL receptors at the cell surface relies on the presence of O -glycan chains, which prevent proteolytic cleavage of the extracellular domains of these proteins (Kozarsky et al., 1988). A further example is the transferrin receptor (TfR). TfR is normally transported to the cell surface and then cycles between the cell surface and the endosomes. A soluble form, sTfR, is formed by proteolytic cleavage in the endosomes. The protease responsible is sensitive to the presence of an O -glycan at Thr104. Desialylation also abolishes the protection of the O -glycans against proteolytic activity. (Rutledge and Enns, 1996).
Human meprin^, a member of the astacin family of zinc metalloendopeptidases, shows similar properties of proteolytic release from the cell surface. The O-linked carbohydrate side chains are clustered around a 13 amino acid sequence that contains the main cleavage site for proteolytic processing of the subunit. Prevention of O-glycosylation by specific inhibitors leads to enhanced proteolytic processing and the consequence is an increased release of the enzyme (Leuenberger et al., 2003).
Further examples of proteolytic processing are found for the blood clotting factors X and Xa.
The activation of the glycoproteins X and Xa occurs through proteolytic cleavage, by factor IXa and its cofactor VIIIa in the intrinsic cascade and by factor VIIIa and tissue factor in the extrinsic cascade, ultimately leading to thrombin formation. The activation of factor X has been reported to depend on the presence of sialic acid in O-glycans as asialo forms show reduced activation. This remains controversial, as other reports have not observed this relationship. However, the removal of the
O -glycans from factor X results in a significant reduction in the kcat for the activation reaction (Inoue and Morita, 1993).
An example where O-glycosylation promotes proteolytic action in processing is found for CD44 in cells of melanocytic lineage. The proteolytic cleavage and shedding is dependent on the presence of partial or complete O-glycosylation of four serine-glycine motifs localized in the membrane-proximal CD44 ectodomain. Mutation of these serine residues, as well as extensive metabolic O-deglycosylation, block spontaneous CD44 shedding (Gasbarri et al., 2003).
3.3. Subcellular localization of proteins
The transport of proteins to the Golgi apparatus from the endoplasmic reticulum is closely linked with O-glycosylation. The biosynthesis, maturation, and intracellular transport of galactosyltransferase, an established trans-Golgi enzyme, are disrupted by brefeldin A that induces a microtubule-dependent backflow of Golgi components to the endoplasmic reticulum. The targeting of the galactosyltransferase for the Golgi apparatus is dependent on O-glycosylation (Bosshart et al., 1991). Brefeldin A has also been used to demonstrate O-glycosylation of Ribophorin I, a type I transmembrane glycoprotein specific to the rough endoplasmic reticulum. Brefeldin A treatment leads to O-glycosylation of Ribophorin I by glycosyltransferases that are redistributed from the Golgi apparatus to the endoplasmic reticulum (Ivessa etal., 1992).
Human neurotrophin receptor (p75 NTR) contains a cluster of O-glycans in the ectodomain located close to the membrane. The transport and sorting of p75 NTR in Caco-2 cells requires these O-glycans. After initial synthesis, the receptor is targeted to the basolateral membrane and subsequently migrates to the apical membrane. Localization to the apical surface is controlled jointly by the O-glycans together with a membrane anchor domain (Breuza et al., 1999).
3.4. Developmental expression
The expression of O-glycans during developmental processes associated with specific biological activity has been reported. Two examples illustrate this phenomenon.
The CD8aj coreceptor possesses an O-glycosylated polypeptide stalk connecting the glycoprotein to the thymocyte surface. Immature CD4(+) CD8(+) double-positive thymocytes bind major histocompatibility complex class I (MHCI) tetramers more avidly than mature CD8 single-positive thymocytes. This differential binding is regulated by a developmentally expressed sialyltransferase (ST3Gal-I) leading to O-glycan modification. The appearance of core 1 sialic acid linked to CD8j on mature thymocytes decreases CD8aj -MHCI avidity and thus modulates the binding of dimeric CD8 to MHCI (Moody et al., 2001).
The brush border proteins pro-sucrase-isomaltase (pro-SI) and dipeptidyl peptidase IV (DPPIV) are sorted for apical location during the differentiation of the crypt cells. These glycoproteins have both N- and O-glycans. Full O-glycosylation is mediated by the N-glycans and the subsequent polarized sorting of these proteins to the apical membrane requires intact O-glycans. This example demonstrates interrelated but distinct roles for N – and O -glycans and their processing in brush border enzyme expression during intestinal cell differentiation. (Naim et al., 1999).
4. Immune cell expression and response
A regulatory function for T-lymphocyte Core 2 j-1,6-N-acetylglucosaminyltrans-ferase, (C2GnT) in the peripheral immune system has been identified.
CD43 is the major cell-surface glycoprotein and an activation antigen expressed on both CD4 and CD8 single-positive T lymphocytes and carries core 2 O-glycan structures. Downregulation of the activation antigen occurs in thymic positive selection implying that the C2GnT modulated expression of CD43 glycoforms is involved in thymic selection events. (Ellies et al., 1996).
Additional support for this type of role for the Core 2 j-1,6-N-acetylgluco-saminyltransferase comes from overexpression of the enzyme in transgenic mice. Isolation of T lymphocytes from these mice showed a reduced immune response due to impaired cell-cell interaction (Tsuboi and Fukuda, 1997).
