Glycobiology Advance Access originally published online on May 7, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Glycobiology, 2003, Vol. 13, No. 9 77R-91R
© 2003 Oxford University Press
REVIEW |
The contribution of N-glycans and their processing in the endoplasmic reticulum to glycoprotein biosynthesis
Department of Cell Biology, Yale University School of Medicine, 333 Cedar Street, PO Box 208002, New Haven, CT 06520
accepted on April 24, 2003
 |
Abstract |
|---|
Top
Abstract
IntroductionThe attachment of N-glycans to nascent glycoproteins in the endoplasmic reticulum (ER) is intimately related to glycoprotein biogenesis. Processing of N-linked oligosaccharides begins in the ER and participates in glycoprotein folding and assembly. The elucidation of N-glycan processing mechanisms in the ER is uncovering their role in glycoprotein biosynthesis.
Key words: degradation / folding / glycoprotein / processing / transport
|
Introduction |
|---|
The most common theme in protein glycosylation is diversity, both in the composition and function of the carbohydrate moieties (Kornfeld and Kornfeld, 1976
; Varki, 1993). The plethora of structures generated reflects the multiple elongation reactions of protein-bound glycans in the Golgi apparatus. Such remarkable variety contrasts with the absolute conservation of the dolichol-bound oligosaccharide (Glc3Man9GlcNAc2) used to initiate protein N-glycosylation in the endoplasmic reticulum (ER). Dolichol moieties differ in length in some eukaryotes (Parodi and Leloir, 1979; Schenk et al., 2001), but no variations have been described in the structure of the precursor oligosaccharide. The only exceptions were found in certain protozoa that utilize truncations of the canonical structure (Glc0–3Man5–9 GlcNAc2) (Parodi, 1993). Over the past several years, these conserved N-glycans were found to contribute to glycoprotein biogenesis in the ER before being modified to serve other functions in their final destinations. |
A conserved precursor for N-glycosylation |
|---|
The robust specificity of the glycosyltransferases in the ER contributes to the conservation of the precursor oligosaccharide, in which two portions can be distinguished structurally and by their biosynthetic origin (Figure 1). The proximal domain containing the residues that lay near the protein backbone is assembled on the cytosolic leaflet of the ER from soluble nucleotide-sugar donors (residues a–g in Figure 1). This proximal domain, conserved in trypanosomatids and most mutant cells with glycosylation defects, may contribute to the stability of acceptor glycoproteins. The stabilizing role of small oligosaccharides (5–10 monosaccharides) is also seen in bacterial glycoproteins, although the structures and linkages are more variable than in eukaryotes (Lechner and Wieland, 1989
; Moens and Vanderleyden, 1997). Supporting the notion that these proximal residues may serve a structural role, they are often well defined in crystal structures, sometimes interacting with the surrounding polypeptide chain, and they are usually inaccessible to endoglycosydases without denaturation.
|
Although N-linked glycans can be dispensable for the function of many glycoproteins, they were shown in some cases to affect protein conformation, stability, and function, with the sugar residues closest to the polypeptide making the most significant contribution (for recent examples see Gala and Morrison, 2002
; Mimura et al., 2001; Wyss et al., 1995). The role of the proximal residues on glycoprotein function so far has been documented in too few cases to be generalized (Rademacher et al., 1988; Rudd and Dwek, 1997), but few are also the instances in which this was carefully analyzed.
The distal domain of the dolichol-bound precursor is completed in the lumen of the ER, using Man and Glc residues transferred from dol-P-Man and dol-P-Glc, which are synthesized on the cytosolic face and then flipped into the lumen of the ER (residues h–n in Figure 1). This distal portion is highly flexible (Petrescu et al., 1997; Wooten et al., 1990) and most often not seen in crystal structures. These residues, typically remodeled in the Golgi, are less likely to interact directly with or contribute to the conformation and stability of the underlying polypeptide. In contrast to the proximal portion, which can be sometimes inaccessible to glycosydases and glycosyltransferases remaining as high-mannose structures on mature glycoproteins, the residues in the distal domain are processed by glycosidases in the ER.
|
Processing of N-glycans in the ER |
|---|
Glucose removal
Given its remarkable capacity for glycan processing, the Golgi apparatus could conceivably have incorporated the glycosidases required to dismantle the protein-bound Glc3Man9GlcNAc2. Instead, glucose and mannose removal from the N-glycans begins in the ER immediately after being transferred to protein. The terminal glucose residue n is removed cotranslationally by glucosidase I, a membrane-bound enzyme in the ER (Kalz-Fuller et al., 1995
; Romero et al., 1997; Simons et al., 1998). Glucosidase II then removes the two remaining Glc residues (m and l, Figure 1). Unlike typical glycosidases in the secretory pathway, which are type II membrane proteins (Herscovics, 1999), glucosidase II is a soluble enzyme resident in the ER. It is composed of one catalytic subunit (
) and one accessory subunit (ß) assembled into a highly asymmetric heterodimer (Trombetta et al., 1996, 2001). The catalytic subunit is fully active in vitro in the absence of the ß subunit (Trombetta et al., 2001), but both subunits are required for glucosidase II activity in vivo (D'Alessio et al., 1999; Pelletier et al., 2000; Treml et al., 2000). Alternatively spliced forms of both chains have been described (Arendt et al., 1999; Treml et al., 2000; Ziak et al., 2001), but the catalytic core appears conserved in all of them. Although glucosidases are confined to the ER, some tissues express an endomannosidase that allows Glc removal in the cis Golgi (Dong et al., 2000; Zuber et al., 2000). The endomannosidase, absent in lower eukaryotes (Dairaku and Spiro, 1997), cleaves after one of the terminal mannose residue that is usually not hydrolyzed by -mannosidases in the ER (Figure 1, residue g).
Mannosidases in the ER
Mannose residue i can be removed in the ER by the ubiquitous -mannosidase I. Saccharomyces pombe was reported to lack mannosidase I activity (Ziegler et al., 1994), but a homolog of mannosidase I can be identified in its genome. Also, although Man8 oligosaccharides are barely detectable in S. pombe, a small proportion of Man8GlcNAc2 can be consistently detected in pulse chase experiments (D'Alessio et al., 1999; Fernandez et al., 1994), suggesting that ER–mannosidase I activity is present in fission yeast. In higher eukaryotes, further mannose trimming can occur in the ER by other -mannosidases (Daniel et al., 1994; Herscovics, 2001; Moremen et al., 1994).
Processing in the ER does not simply prepare N-glycans for further trimming once they reach the Golgi. In yeast, for example, trimming by ER glycosidases is not a requisite for subsequent elongation reactions in the Golgi, which simply extend the structures that egress from the ER to mannans without further processing (Esmon et al., 1984; Puccia et al., 1993). In higher eukaryotes, mannose trimming also occurs in the Golgi, where -mannosidases are capable of hydrolyzing six of the nine mannosyl residues in Man9GlcNAc2 forming a Man3GlcNAc2core in complex type N-glycans, being able to act on any structure that escapes mannose trimming in the ER. Mannose removal in the ER is therefore unnecessary for the synthesis of complex glycans and may be considered a timer for the residence of glycoproteins in the ER.
Glycoprotein reglucosylation
Deglucosylated glycoproteins can be transiently reglucosylated on mannose residue g by a soluble glucosyltransferase (GT) in the ER (Trombetta et al., 1991; Trombetta and Parodi, 1992). As opposed to the membrane-bound glucosyltransferases that utilize dol-P-Glc to add Glc residues to the dol-P-P oligosaccharide precursor, the soluble reglucosylating enzyme utilizes UDP-Glc as Glc donor, imported into the ER from the cytosol (Perez and Hirschberg, 1986; Vanstapel and Blanckaert, 1988). Most glycoproteins are reglucosylated (Cacan et al., 2001; Gañan et al., 1991; Gotz et al., 1991), re-creating the same trimming intermediate generated by glucosidase II (Figure 2). The added Glc is also removed by glucosidase II, and this seems to be the only function of this enzyme in trypanosomatids, because they do not synthesize Glc containing dol-P-P-oligosaccharide precursors (Bosch et al., 1988; Trombetta et al., 1989). Such a cycle of deglucosylation and reglucosylation can occur repeatedly on glycoproteins during their biosynthesis (Suh et al., 1989) and is now recognized as a central part of a quality control and chaperone pathway in the ER lumen (Figure 2). The reglucosylating enzyme recognizes as substrates only incompletely folded glycoproteins and does not reglucosylate native structures or free oligosaccharides (Parodi, 2000). This unique capacity of GT to distinguish between native and nonnative conformations of its glycoprotein substrates was discovered and characterized in vitro, but a similar preference has been observed in living cells (Cannon and Helenius, 1999; Labriola et al., 1999; Suh et al., 1989; Wada et al., 1997). GT seems to prefer partially structured conformations rather than fully denatured polypeptides, suggesting that it may act at late stages during glycoprotein folding (Labriola et al., 1999; Sousa and Parodi, 1995; Trombetta and Helenius, 2000). Substrate conformation is evaluated in small domains (Ritter and Helenius, 2000) where the N-glycans have to be present (Sousa and Parodi, 1995). Interestingly, mannosidases in the ER do not remove the mannnose to which Glc units are added by reglucosylation (Figure 1, residue g), maximizing the chances for reglucosylation before exiting the ER. However, this Man residue can be removed in some cell types by the Golgi endomannosidase, terminating reglucosylation cycles.
|
|
N-glycosylation and glycoprotein folding |
|---|
Attachment of oligosaccharides to nascent polypeptides is an essential process in eukaryotes. Some glycoproteins are functional when their N-glycans are removed enzymatically or by site-directed mutagenesis, but a complete lack of N-glycosylation is lethal, as shown by mutant cells or by cells treated with N-glycosylation inhibitors (such as tunicamycin) (Elbein, 1987
; Kukuruzinska et al., 1987). Mutants have been isolated that transfer truncated oligosaccharides to protein or with impaired N-glycan processing in the ER, indicating that the presence of N-glycans rather than their precise structure is required for cell viability (Kukuruzinska et al., 1987; Stanley, 1984; Stanley and Ioffe, 1995). N-glycans are added cotranslationally to nascent polypeptides (Bergman and Kuehl, 1977; Kiely et al., 1976; Rothman and Lodish, 1977; Sefton, 1977) before they acquire their native structure, and perhaps their most basic function in the ER is to facilitate the folding of glycoproteins. They appear to do this in a rather nonspecific way, providing bulky hydrophilic moieties that reduce aggregation and promote solubility of folding intermediates (Chu et al., 1985; Edge et al., 1993; Kern et al., 1992, 1993; Paulson, 1989; Schulke and Schmid, 1988; Wang et al., 1996). Such requirement of carbohydrate lubrication for glycoprotein folding may be necessary in the conditions prevalent in lumen of the ER, because it is not observed for protein folding in the cytosol or mitochondria. This is illustrated by invertase, an enzyme that is produced in two forms, one cytosolic, and another secreted, differing only by a signal sequence. Unlike cytosolic invertase, the secreted form is critically dependent on addition of N-glycans for folding but not on their processing. Invertase secretion is blocked by tunicamycin, resulting in aggregation and retention in the ER (Bergh et al., 1987; Feldman et al., 1987; Ferro-Novick et al., 1984; Kuo and Lampen, 1974; Roitsch and Lehle, 1989) but not by lack of Glc or Man removal (Esmon et al., 1984; Puccia et al., 1993). The effect of tunicamycin can be alleviated by low temperature (Bergh et al., 1987; Ferro-Novick et al., 1984), indicating that N-glycans are required for proper folding of invertase in the ER.
Many glycoproteins need the N-glycans for folding and transport out of the ER but not for their biological function (Helenius, 1994). When N-glycosylation is prevented, proteins usually misfold, aggregate, and are retained in the ER by quality control mechanisms that prevent transport of proteins that have not acquired a native conformation (Ellgaard et al., 1999). Misfolded proteins that do not pass quality control are translocated back to the cytosol for proteasomal degradation (a process termed ERAD, for ER associated degradation) (Bonifacino and Weissman, 1998; Brodsky and McCracken, 1999; Cabral et al., 2001; Plemper and Wolf, 1999) after deglycosylation by a cytosolic N-glycanase (Suzuki et al., 2002a). A ubiquitin ligase selective for glycosylated polypeptides has been recently proposed to participate in proteasomal glycoprotein degradation (Yoshida et al., 2002).
Increased levels of misfolded proteins in the ER induce the expression of chaperones and folding factors resident in the ER. This process, known as the unfolded protein response (UPR), can be triggered in cultured cells by treatments that induce protein misfolding, such as heat shock, reduction of disulfide bonds, mobilization of calcium out of the ER, or by preventing protein glycosylation (Chapman et al., 1998; Kaufman et al., 2002). Several genes involved in protein N-glycosylation (Casagrande et al., 2000; Helenius et al., 2002; Ng et al., 2000; Travers et al., 2000), as well as mannosidase I in yeast (Travers et al., 2000), GT (Arnold et al., 2000; Fernandez et al., 1996), calnexin (CNX) (Parlati et al., 1995), and a Man8b-specific lectin (see later discussion) (Hosokawa et al., 2001) are among the transcripts induced during the UPR, emphasizing the role of N-glycans and their processing in promoting glycoprotein folding in the ER. Moreover, synthesis of the Glc3Man9GlcNAc2 precursors is also augmented during UPR (Doerrler and Lehrman, 1999; Shang et al., 2002).
The deficiency in glycoprotein folding seen in cells that transfer truncations of the canonical Glc3Man9GlcNAc2 is in part due to the loss of the outermost Glc residue (residue n, Figure 1). Because this residue is key for recognition of the dolichol-P-P oligosaccharide precursor by the oligosaccharyltransferase, its absence results in underglycosylation of proteins (Silberstein and Gilmore, 1996). Smaller glycans could conceivably support folding for the biophysical reasons considered, if the oligosaccharyltransferase would utilize them efficiently and glycosylation sites were occupied, as in trypanosomatids. If shorter oligosaccharides could potentially be functional, what is therefore special about Glc3Man9GlcNAc2? Why is this particular structure so strictly conserved? During the past several years, different lines of research revealed that before being remodeled in the Golgi, N-glycans at different stages of processing link the underlying polypeptide chains, often incompletely folded, with lectins assisting glycoprotein assembly, transport, and degradation.
|
Lectins in the ER |
|---|
Lectins as molecular chaperones
Partially deglucosylated or reglucosylated structures (Glc1Man5–9GlcNAc2) are specifically recognized by two lectins in the ER, CNX and calreticulin (CRT). Binding to CNX/CRT exposes glycoprotein ligands to the thiol oxidoreductase Erp57 (Farmery et al., 2000
; Kang and Cresswell, 2002; Oliver et al., 1999; Van der Wal et al., 1998). In that way, disulfide-containing substrates can form mixed disulfide intermediates with Erp57, facilitating correct disulfide bond formation (Antoniou et al., 2002; Lindquist et al., 2001; Molinari et al., 2002; Molinari and Helenius, 1999). The binding of CNX/CRT to transient and dynamic monoglucosylated structures has been conceptualized in a glycoprotein specific chaperone cycle (Figure 2) (Hammond et al., 1993). Recent reviews summarize different aspects of this cycle that promote glycoprotein folding, providing chaperone and quality control function in the ER (Bergeron et al., 1994; Cabral et al., 2001; Chevet et al., 1999; Ellgaard and Helenius, 2001; Helenius and Aebi, 2001; High et al., 2000; Jakob et al., 2001b; Lehrman, 2001; Parodi, 2000; Schrag et al., 2003; Spiro, 2000; Trombetta and Helenius, 1998). Unlike other chaperone systems, which directly interact with polypeptide chains, this cycle is centered around monoglucosylated oligosaccharides and relies on the concerted action of carbohydrate-modifying enzymes and lectins. When deglucosylation of the protein-bound Glc3Man9GlcNAc2 is prevented by glucosidase inhibitors or in cells defective in glucosidase I or II, the interaction of glycoproteins with CNX/CRT is in most cases abolished, confirming the requirement of the single Glc residue for binding. Direct protein–protein interactions between CNX/CRT and substrate glycoproteins are also proposed (Danilczyk and Williams, 2001; Ihara et al., 1999; Leach et al., 2002; Mancino et al., 2002; Saito et al., 1999; Stronge et al., 2001), although a consensus supports the requirement for an initial contact mediated by monoglucosylated oligosaccharides.
Both CNX and CRT appear to be monomeric (Bouvier and Stafford, 2000), with low affinities but high specificity for monoglucosylated oligosaccharides (
50 µM) (Patil et al., 2000), which may facilitate spontaneous dissociation of substrates. The two or three Man residues extending from the Glc-Man linkage contribute to the binding (Kapoor et al., 2002; Vassilakos et al., 1998). From the structure of the lectin domain of CNX (highly conserved in CRT) it appears that the Glc-Man portion of the oligosaccharide bound to CNX/CRT would be inaccessible to glucosidase II (Schrag et al., 2001), suggesting that glycoproteins are deglucosylated after a transient, low-affinity interaction with CNX/CRT. A series of conspicuous repeats in CNX and CRT (Proline-rich or P domain) protrudes from the carbohydrate binding domain adopting a highly extended conformation (Ellgaard et al., 2001; Schrag et al., 2001). The tip of the P domain binds to ERp57 (Frickel et al., 2002), providing a direct link between the monoglycosylated glycoproteins bound by CNX/CRT and the disulfide isomerase functions of Erp57 (Figure 2). It would be interesting to establish whether ERp57 possess unfoldase activity, like its close homolog protein disulfide isomerase (Gilbert, 1997; Gillece et al., 1999; Orlandi, 1997; Tsai et al., 2001), as suggested by their interaction with substrates lacking Cys residues (Gillece et al., 1999; Oliver et al., 1997).
Lectins facilitating glycoprotein degradation
With the observation that the extent of mannose trimming in the ER-influenced ERAD of misfolded CPY (CPY*) it was proposed that a lectin would recognize the demannosylated oligosaccharide resulting form mannosidase I action (Man8b, Figure 2) (Jakob et al., 1998; Knop et al., 1996; Su et al., 1993). As opposed to CNX/CRT, this lectin seems to target glycoproteins for degradation instead of promoting their folding. A candidate for the Man8b-specific lectin (M8 lectin), termed Mnl1, Htm1, or EDEM, was identified as a homolog of mannosidase I lacking key amino acids essential for mannosidase activity (Hosokawa et al., 2001; Jakob et al., 2001a; Nakatsukasa et al., 2001). Yeast strains lacking the M8 lectin degrade misfolded glycoproteins less efficiently (Jakob et al., 2001a; Nakatsukasa et al., 2001), and its overexpression accelerates their degradation (Hosokawa et al., 2001; Molinari et al., 2003; Oda et al., 2003), supporting a role for this lectin in promoting the degradation glycoproteins bearing Man8b oligosaccharides in the ER. These carbohydrate structures alone cannot be a signal for degradation, because long lived ER-resident glycoproteins carry partially demannosylated oligosaccharides. In fact, the presence of Man8b oligosaccharides does not accelerate ERAD of native CPY, suggesting that to promote glycoprotein degradation the Man8b structures have to be attached to nonnative protein moieties. This could reflect a certain ability of the M8 lectin to distinguish native from nonnative glycoproteins, or perhaps more likely, it may interact with or be coupled to the action of other ER proteins. The carbohydrate-binding properties have not been directly studied for the M8 lectin. It would be interesting to establish whether ligands bind reversibly or if glycoprotein ligation by M8 leads irreversibly to ERAD, and whether it can bind Glc1Man8GlcNAc2 and thus compete with CNX/CRT for monoglucosylated glycoproteins.
Lectins in intracellular sorting and trafficking
Another lectin, ERGIC-53, is also present in the ER but accumulates in the intermediate compartment, cycling between the ER and the cis Golgi (Hauri et al., 2000a,b). ERGIC-53 was identified by different groups based on its reactivity with antibodies and by its mannose binding properties. It appears to function as a cargo receptor for selective export of certain glycoproteins from the ER, and mutations in ERGIC-53 cause combined deficiency of coagulation factors V and VIII (Nichols et al., 1998). The carbohydrate-binding domain of ERGIC-53 has a similar structure to the one in CNX/CRT (Velloso et al., 2002) but is arranged as a homo oligomer, whereas CNX and CRT are monomeric (Bouvier and Stafford, 2000). A homolog of ERGIC-53 with restricted tissue distribution was recently identified (Yerushalmi et al., 2001). Another lectin, VIP36, has an overlapping distribution with ERGIC-53 (Fullekrug et al., 1999) and also recognizes high-mannose N-glycans (Hara-Kuge et al., 1999). VIP36 is thought to participate in intracellular transport of glycoproteins (Fiedler et al., 1994; Fiedler and Simons, 1996; Hara-Kuge et al., 2002; Yamashita et al., 1999).
|
A role for N-glycan processing in glycoprotein folding and degradation |
|---|
Not only the oligosaccharide utilized to initiate N-glycosylation but also its processing in the ER have been highly conserved in eukaryotes. However, unlike its essential presence on glycoproteins, the processing of N-glycans in the ER is not essential for viability of isolated cells (Table I). The lack of N-glycan processing does not have major consequences in lower eukaryotes or mammalian cells in culture, most of which have virtually no discernable pheno-type. The lack of glucosidase I, II, or
-mannosidases in the ER due to genetic deficiency or due to growth in the presence of inhibitors does not compromise cell viability.
|
Mannose trimming in the ER seems to have little effect on glycoprotein folding and secretion. Eukaryotic cells can be cultured for days in the presence of high concentrations of mannosidase inhibitors and the secretion of most glycoproteins analyzed is rarely affected. However, mannosidase inhibitors are toxic to mammals, due to their inhibition of lysosomal mannosidases causing storage defects and by interfering with complex oligosaccharide formation (Elbein, 1987
).
Inhibition of glucosidases in the ER has often been observed to affect the secretion of numerous glycoproteins (Helenius, 1994; Lodish and Kong, 1984). Glycoproteins from enveloped viruses are particularly susceptible to inhibition of ER glucosidases (Doms et al., 1993), which has lead to the consideration of glucosidase inhibitors as potential antiviral agents (Gruters et al., 1987; Mehta et al., 1998). Although the lack of Glc removal in the ER is well tolerated by yeast or cultured mammalian cells, it has deleterious effects in multicellular organisms. Glucosidase II deficiency in Dictiostelium (Freeze et al., 1997) and Arabidopsis (Burn et al., 2002) and glucosidase I deficiency in Arabidopsis (Boisson et al., 2001; Gillmor et al., 2002) lead to abnormal development. Interestingly, the block of complex N-glycans has virtually no consequences in Arabidopsis (von Schaewen et al., 1993), suggesting that glucose trimming is required for the maturation of some glycoproteins that seem to be involved in cellulose biosynthesis (Gillmor et al., 2002). A homozygous null patient for glucosidase I died at 10 weeks of age (De Praeter et al., 2000). Mutations in the ß-subunit of glucosidase II are associated with autosomal dominant plycystic liver disease (Drenth et al., 2003; Li et al., 2003). Deglucosylation in mammals may not only cause a generalized block of the CNX/CRT pathway but probably affects mostly intercellular interactions by interfering with the essential formation of complex oligosaccharides (Marquardt and Freeze, 2001; Stanley and Ioffe, 1995).
The viability of yeasts and cultured mammalian cells with deglucosylation defects implies that not only glycan processing but also the CNX/CRT cycle is not essential for these cells. The CNX/CRT pathway can be blocked without major consequences for cell viability in S. pombe by deletion of glucosidase II or GT (D'Alessio et al., 1999; Fernandez et al., 1996), although glycoprotein reglucosylation is essential for viability under stress conditions (Fanchiotti et al., 1998). Moreover, the CNX/CRT pathway is absent in Saccharomyces cerevisiae, which lacks most of its components. CNX- and CRT-deficient cell lines have also been described, although the corresponding null mice are embryonic lethal (Denzel et al., 2002; Mesaeli et al., 1999; Rauch et al., 2000). It can be argued that the lethal phenotypes arise from the defective folding of numerous glycoproteins but may also reflect that CNX and/or CRT serve essential function(s) other than binding monoglucosylated oligosaccharides. For instance, though the viability of strains lacking gls1, gls2, and GT indicate that the CNX pathway is not essential in S. pombe (D'Alessio et al., 1999; Esmon et al., 1984; Fanchiotti et al., 1998; Fernandez et al., 1996; Trombetta et al., 1996) the calnexin gene is essential (Parlati et al., 1995). Also, CRT is involved in heart function and Ca2+ homeostasis in the ER (Arnaudeau et al., 2002; Li et al., 2002; Mesaeli et al., 2001; Nakamura et al., 2001), and CNX and CRT have been implicated in phagocytosis (Muller-Taubenberger et al., 2001) and necrotic cell death (Xu et al., 2001). The advent of structural and biochemical data will allow the design of experiments to dissect the contribution of carbohydrate binding to CNX/CRT function. Mannosidase I activity is not essential for S. cerevisiae, and mannan elongation in the Golgi does not require Man trimming in the ER (Puccia et al., 1993). The M8 lectin is also not essential in yeast (Jakob et al., 2001a; Nakatsukasa et al., 2001).
|
Variable fates for glycoproteins |
|---|
Although not essential for the viability of individual cells, the N-glycan structures and processing reactions have most likely been conserved to confer some advantage for glycoprotein maturation. The enzymes and lectins involved in N-glycan processing in the ER and their functionality are beginning to be understood, but it is still difficult to predict the contribution of oligosaccharide processing to the maturation of individual glycoproteins (Table II). For instance, the effect of preventing N-glycosylation with tunicamycin on the expression of glycoproteins, can be cell type–dependent (Sibley and Wagner, 1981
). Also, inhibition of glucosidases can accelerate ERAD for some glycoproteins but not for others (Ayalon-Soffer et al., 1999; Fagioli and Sitia, 2001; Moore and Spiro, 1993). Similarly, inhibition of ER-mannosidases does not always stabilize glycoproteins against ERAD (Cabral et al., 2000). Moreover, misfolded glycoproteins are degraded at different rates despite being exposed to the same N-glycan processing machinery in a given cell type, indicating that the carbohydrate structures are not dominant. The variable fate followed by glycoproteins probably reflects a balance between chaperone and degradative mechanisms, such as the seemingly opposing action of monoglucosylated structures driving the CNX/CRT pathway promoting folding and the M8 lectin facilitating glycoprotein degradation.
|
Glycoproteins not only vary by their dependence on their N-glycans for correct maturation. There is evidence for redundancy between carbohydrate dependant and noncarbohydrate-mediated chaperone mechanisms. For example, incorrectly folded proteins are still subject to ERAD in yeast strains lacking the M8 lectin or mannosidase I or in the presence of tunicamycin. When access to CNX/CRT is impaired, glycoproteins can bind to other chaperones like BiP (Zhang et al., 1997
), showing how other factors in the ER can compensate the lack of N-glycan mediated pathways. Such plasticity implies that the use of the cellular machinery, at least for some glycoproteins, is not only dictated by its primary sequence but can be adapted to the chaperone capacity available in the ER. This principle is illustrated by glycoproteins that exhibit different fates depending on the cellular environment (Table II). |
Factors affecting the processing of N-glycans in the ER |
|---|
Except for the sequential action of glucosidases I and II, the other carbohydrate processing and lectin binding reactions in the ER (Figure 2) are not sequential. It is not clear what determines whether a Man9GlcNAc2 oligosaccharide on a nascent glycoprotein is reglucosylated or demannosylated; whether a Glc1Man9GlcNAc2 structure will be processed by glucosidase II or
-mannosidases; whether Glc1Man8GlcNAc2 will bind to CNX/CRT or to the M8 lectin, ERGIC-53, or VIP36; or why ERGIC-53, a mannose-binding lectin, affects the transport of only certain glycoproteins.
The prevalence of particular oligosaccharide structures and their binding to different lectins may influence the partition of individual glycoproteins into chaperone or degradative pathways. These alternative pathways appear to monitor discrete domains of the polypeptide backbone in glycoproteins, because CNX, CRT, and probably the M8 lectin appear to be monomeric, interacting with one oligosaccharide at a time. Reglucosylation is also restricted to individual glycans within small domains in glycoproteins. Consequently, because glycoproteins generally have more than one oligosaccharide, N-glycosylation sites on different domains of the same glycoprotein could simultaneously promote opposite fates. It appears that to understand glycoprotein biosynthesis, it will be necessary to understand the factors that influence the processing of their oligosaccharides. For example, the susceptibility of VSV-G glycoprotein to endomannosidase action does not always correlate with the levels of enzymatic activity in various cells (Karaivanova et al., 1998). Therefore, aside from expression levels of enzymes and lectins, other factors that may affect the prevalence of N-glycan processing intermediates may need to be considered.
Localization
One of the factors that may affect the structure of particular N-glycans is their precise subcellular localization during biosynthesis. During their maturation, some glycoproteins do not appear to leave the ER (Duvet et al., 1998; Shenkman et al., 1997), whereas others accumulate in the ER but venture into the intermediate compartment to different extents (Hammond and Helenius, 1994; Raposo et al., 1995). Moreover, the distribution of the enzymes and lectins that act on them does not seem to be uniform, and therefore glycoproteins may be differentially exposed to N-glycan processing enzymes and lectins depending on where they accumulate before leaving the ER (or being degraded).
The subcellular distribution of some enzymes and lectins, like ER -mannosidase glucosidase II, GT, and CRT has been analyzed by immunoelectron microscopy by Roth and colleagues. They are enriched in the ER, but a substantial proportion can also be found in transitional elements beyond the ER (Cannon and Helenuis, 1999; Lucocq et al., 1986; Zuber et al., 2000, 2001). In hepatocytes, glucosidase II is largely restricted to the ER (Lucocq et al., 1986), but GT is not uniformly distributed along the ER and is enriched in post-ER pre-Golgi compartments (Zuber et al., 2001). Similarly, the distribution of ER -mannosidase and glucosidase II is not completely overlapping (Roth et al., 1990). The subcellular localization of CNX and the M8 lectin have not been studied in detail, but CNX is found in the ER by fluorescence microscopy in most cells analyzed. In S. cerevisiae, mannosidase I accumulates in the ER at steady state, but the structure of its oligosaccharides reveals it is exposed to processing enzymes in the cis Golgi (Massaad et al., 1999).
Aside from variations introduced by splice isoforms of glucosidase II or tissue-specific homologs of CNX (Watanabe et al., 1994) and ERGIC-53 (Yerushalmi et al., 2001), variations may also be encountered in the distribution of enzymes and lectins between different cell types. For instance, the enrichment of GT in the ER was found to be cell type–dependent (Cabral et al., 2002). Glucosidase II also accumulated in post-ER compartments in kidney tubular cells (Brada et al., 1990). This alternative localization could be due to tissue-specific factors, such as differential expression of the ß-subunit or splice variants of the -subunit (Ziak et al., 2001). The localization of CRT, GT, and glucosidase II may also be influenced by their variable KDEL-related C-terminal signals. The mistargeting of some glycoproteins when the recycling of ERGIC-53 is inhibited (Vollenweider et al., 1998) may indeed illustrate how the precise localization of lecins and enzymes has the potential to influence the fate of its substrates/ligands.
Substrate availability
Calcium is required for the activity of most enzymes and lectins that act on glycans in the ER, including GT, CNX, CRT, ERGIC-53, mannosidase I, and apparently VIP36 (and probably the M8 lectin by its homology with mannosidase I). Such a cation requirement seems well fit for the ER, which, unlike other organelles, accumulates high concentrations of Ca2+. They are also highly dynamic, and fluctuations in Ca2+ concentrations could potentially affect glycan processing and recognition. Indeed, perturbation of Ca2+ homeostasis by ionophores affects the retention of proteins in the ER (Booth and Koch, 1989; Sambrook, 1990). Apart from a direct role in catalysis, Ca2+ may indirectly affect glucosidase II and mannosidase I, which have an EF-hand motif, not required for enzymatic activity in vitro but likely to mediate interaction of these enzymes with other components in the ER. The role of Ca2+ in the processing of carbohydrates in the ER contrasts with the N-glycan processing in the Golgi, where most glycosidases and glycosyltransferases require Mg2+ or Mn2+.
The supply of UDP-Glc into the ER lumen may also affect the efficiency of glycoprotein reglucosylation. Such variations can arise from differences in the activity of the UDP-Glc transporter and/or in the cytosolic concentration of UDP-Glc in response to cellular metabolism. That substrate availability in vivo can affect glycosyltransferases was shown by the impact of small reductions (about fourfold) in cytosolic UDP-Glc concentration, leading to resistance to the glucosyltransferase toxin B by a factor of 10,000 (Chaves-Olarte et al., 1996; Flores-Diaz et al., 1998).
Also, the cyclic nature of the reglucosylation process consumes several UDP-Glc molecules per oligosaccharide, and the UDP generated as a reaction product by GT may cause product inhibition of GT and affect the antiporter mechanism of UDP-Glc transport (Hirschberg et al., 1998). Enzymes capable of hydrolyzing UDP to UMP in the ER (Failer et al., 2002; Trombetta and Helenius, 1999) may serve a similar role to those described in the Golgi, generating nucleoside monophosphates to alleviate product inhibition of the glycosyltransferases and to facilitate nucleotide sugar transport (Hirschberg et al., 1998). Although the role of the ER-UDPases in glycoprotein reglucosylation is so far only speculative, even a modest UDPase activity has been shown to affect UDP-GlcNAc transport and utilization in the Golgi of Kluyveromyces lactis (Lopez-Avalos et al., 2001).
Oligosaccharide specificity
In contrast to the Golgi endomannosidase, which acts preferentially on demannosylated, monoglucosylated oligosaccharides (Lubas and Spiro, 1988), N-glycan processing enzymes in the ER prefer fully mannosylated substrates. Glucosidase II activity seems to decrease concomitantly with the size of the substrate oligosaccharide (Grinna and Robbins, 1980), implying that as substrates are degraded by mannosidases in the ER, glucose removal would be increasingly less efficient, extending the participation of monoglucosylated glycoproteins in the CNX/CRT pathway. Reglucosylation also appears to be inefficient on mannose-trimmed oligosaccharides (Sousa et al., 1992), in agreement with the preferential accumulation of Glc1Man9GlcNAc2 in vivo. However, the oligosaccharide specificity of GT may not be the only factor determining the abundance of reglucosylated products. Structures smaller than Man9GlcNAc2 may be less abundant due to low -mannosidase activity or due to selective degradation of demannosylated glycoproteins. Alternatively, these structures may be less accessible to GT if they are bound to lectins in the ER. ER -mannosidase (Bause et al., 1992) and yeast mannosidase I (Cipollo and Trimble, 2002; Scaman et al., 1996) also seem to prefer the complete oligosaccharides as substrates.
It is not clear whether the specificities established in vitro for these enzymes reflect their mode of action in vivo. For example, whereas GT prefers Man9GlcNAc2 as a substrate in vitro, glycoproteins with small oligosaccharides are readily reglucosylated in vivo (Ermonval et al., 1997, 2001; Labriola et al., 1999; Parodi, 1993). Also, the specificity observed in vitro for CNX and CRT does not account for their substrate selectivity in vivo. The oligosaccharide specificity for CNX and CRT is very similar when analyzed with isolated oligosaccharides (Peterson and Helenius, 1999; Spiro et al., 1996; Vassilakos et al., 1998; Ware et al., 1995), partly explained by their highly conserved lectin domain. However, the spectrum of glycoproteins bound to these lectins in living cells is not completely overlapping (Danilczyk et al., 2000; Peterson et al., 1995). They can even bind to the same glycoprotein at different stages of maturation or to different glycosylation sites (Hebert et al., 1997). These differences appear to be due to the membrane anchorage of CNX compared to the soluble nature of CRT and can be reversed by inverting their topology (Danilczyk et al., 2000). However, topology does not seem to account for the selective interaction of CRT with some glycoproteins (Gao et al., 2002).
Free oligosaccharides in the ER
The oligosaccharide specificity of the various enzymes and lectins dictates not only their substrate preferences but also the extent to which they can be inhibited by free oligosaccharides present in the lumen of the ER. These apparently arise by hydrolysis of dol-P-P precursors, and their potential to interfere with glycan processing is reduced by exporting them out of the ER for subsequent degradation in lysosomes (Cacan and Verbert, 2000; Moore, 1999). Interestingly, part of these free oligosaccharides are believed to arise from an N-glycanase in the lumen of the ER (Weng and Spiro, 1997). Such activity has not been isolated, and its presence in the ER lumen is based on subcellular fractionation experiments. More recently, the study of the release of free oligosaccharides from the VSV-G protein further supported protein N-deglycosylation in the ER (Spiro and Spiro, 2001). The existence of an ER resident N-glycanase (Suzuki et al., 1997) would have interesting consequences for quality control, because deglycosylating glycoproteins would lead irreversibly to their misfolding and subsequent degradation. Although the endoglycosidases identified so far are cytosolic (Hirsch et al., 2003; Suzuki et al., 2002a,b), endoglycosidases that enter the secretory pathway have been described (Chang et al., 2000; Ftouhi-Paquin et al., 1997; Seko et al., 1999; Sheldon et al., 1998).
|
Outlook |
|---|
Like other functions of oligosaccharides, unraveling their role in glycoprotein biogenesis has proved to be a challenging and exciting task. Most of the enzymes and lectins that participate in glycoprotein synthesis in the ER have now been identified. Their functionality is beginning to be understood, but the behavior of glycoproteins is still difficult to predict. Understanding the factors that affect the fate of glycoproteins in vivo is important, because at least some of the published observations are likely to be glycoprotein or cell type–specific. Such specific aspects may turn out to be most valuable, because understanding the details underlying glycoprotein biosynthesis in normal and pathological conditions is likely to provide insights of medical significance, as exemplified by the role of ERGIC-53 in coagulation disorders. The challenge will be to integrate the forthcoming structural, cell biological, and genetic information to understand how ER glycoforms contribute to glycoprotein biogenesis.
1 To whom correspondence should be addressed; e-mail: sergio.trombetta{at}yale.edu 
|
Abbreviations |
|---|
CNX, calnexin; CRT, clareticulin; ER, endoplasmic reticulum; ERAD, ER-associated degradation; ERGIC, ER-Golgi intermediate compartment; GT, UDP-Glc: glycoprotein glucosyltransferase; UPR, unfolded protein response
|
References |
|---|
Antoniou, A.N., Ford, S., Alphey, M., Osborne, A., Elliott, T., and Powis, S.J. (2002) The oxidoreductase ERp57 efficiently reduces partially folded in preference to fully folded MHC class I molecules. EMBO J., 21, 2655–2663.[CrossRef][ISI][Medline]
Arendt, C.W., Dawicki, W., and Ostergaard, H.L. (1999) Alternative splicing of transcripts encoding the alpha- and beta-subunits of mouse glucosidase II in T lymphocytes. Glycobiology, 9, 277–283.
Arnaudeau, S., Frieden, M., Nakamura, K., Castelbou, C., Michalak, M., and Demaurex, N. (2002) Calreticulin differentially modulates calcium uptake and release in the endoplasmic reticulum and mitochondria. J. Biol. Chem., 24, 24.
Arnold, S.M., Fessler, L.I., Fessler, J.H., and Kaufman, R.J. (2000) Two homologues encoding human UDP-glucose:glycoprotein glucosyltransferase differ in mRNA expression and enzymatic activity. Biochemistry, 39, 2149–2163.[CrossRef][Medline]
Ayalon-Soffer, M., Shenkman, M., and Lederkremer, G.Z. (1999) Differential role of mannose and glucose trimming in the ER degradation of asialoglycoprotein receptor subunits. J. Cell Sci., 112, 3309–3318.[Abstract]
Bause, E., Breuer, W., Schweden, J., Roeser, R., and Geyer, R. (1992) Effect of substrate structure on the activity of Man9-mannosidase from pig liver involved in N-linked oligosaccharide processing. Eur. J. Biochem., 208, 451–457.[ISI][Medline]