n-glycan structures: recognition and processing in the er

9
N-glycan structures: recognition and processing in the ER Markus Aebi 1 , Riccardo Bernasconi 2 , Simone Clerc 1 and Maurizio Molinari 2, 3 1 Institute of Microbiology, Department of Biology, Eidgeno ¨ ssische Technische Hochschule (ETH) Zu ¨ rich, CH-8093 Zu ¨ rich, Switzerland 2 Institute for Research in Biomedicine, CH-6500 Bellinzona, Switzerland 3 Ecole Polytechnique Fe ´ de ´ rale de Lausanne, School of Life Sciences, CH-1015 Lausanne, Switzerland The processing of N-linked glycans determines secretory protein homeostasis in the eukaryotic cell. Folding and degradation of glycoproteins in the endoplasmic reticu- lum (ER) are regulated by molecular chaperones and enzymes recruited by specific oligosaccharide structures. Recent findings have identified several components of this protein quality control system that specifically modify N-linked glycans, thereby generating oligosac- charide structures recognized by carbohydrate-binding proteins, lectins. In turn, lectins direct newly synthesized polypeptides to the folding, secretion or degradation pathways. The ‘‘glyco-code of the ER’’ displays the folding status of a multitude of cargo proteins. Deciphering this code will be instrumental in understanding protein homeostasis regulation in eukaryotic cells and for inter- vention because such processes can have crucial import- ance for clinical and industrial applications. Biosynthesis of core oligosaccharides and their transfer from a lipid donor onto nascent chains Asparagine (N)-linked protein glycosylation is a covalent protein modification that occurs across the three domains of life: Bacteria, Archaea and Eukarya. In eukaryotic cells, N-linked glycosylation is the most prominent modification of secretory proteins. This complex biosynthetic pathway has been studied best in the fungus Saccharomyces cere- visiae, but it seems to be highly conserved in other fungal, plant and animal species [1]. The pathway initiates at the ER membrane where a lipid carrier, dolichylpyropho- sphate, serves as a membrane anchor for the assembly of an oligosaccharide. The bipartite biosynthesis of the oligosaccharide initiates at the cytoplasmic side of the ER membrane. Nucleotide-activated sugar donors serve as substrates for a series of different glycosyl transferases that lead to a mannose 5 -N-acetylglucosamine 2 oligosac- charide. This lipid-bound oligosaccharide is translocated across the membrane [2] where an additional four mannose and three glucose residues are added. In the ER lumen, dolichylphosphomannose and dolichylphosphoglucose act as substrates for the individual glycosyl transferases. The transfer of the N-glycan precursor onto nascent polypep- tide chains is catalysed by oligosaccharyltransferase (OST), a protein complex consisting of eight different sub- units. OST utilizes the activated glucose 3 -mannose 9 -N- acetylglucosamine 2 (Glc 3 Man 9 GlcNAc 2 ) oligosaccharide donor as a substrate to covalently modify defined aspar- agine side chains within the acceptor sequence N-X-S/T (X cannot be proline) [1]. Most of the eukaryotic species studied transfer this specific oligosaccharide structure to proteins (Figure 1 and Box 1). N-glycosylation is in many respects a unique covalent protein modification, but in the context of this review, the spatio-temporal aspect of the transfer reaction and the physicochemical properties of the oligosaccharide are of central importance. The N-X-S/T sequon is a short and clearly defined acceptor sequence The N-X-S/T sequon can be found frequently in proteins. In contrast to other covalent post-translational modifications such as O-glycosylation or phosphorylation, no additional, large protein domain is required to define an N-linked glycosylation site. However, several studies suggest that the N-X-S/T sequence can be modified only if it lies within a flexible domain of the polypeptide [3], suggesting that the peptide acceptor takes on a defined structure in the course of the modification reaction. Accordingly, the overall num- ber of proteins and sites that are glycosylated can be increased by executing N-glycosylation before the folding process, during or immediately after the translocation of the nascent chain into the ER lumen [4]. Indeed, in many eukaryotic systems, the OST is associated with the trans- locon and with the ribosome [5]. It has been proposed that defined subunits of the OST might act as chaperones or enzymes to modulate, or even prevent, the folding of the target proteins in order to facilitate N-glycosylation [6,7]. A large, hydrophilic and branched oligosaccharide structure is transferred to proteins The hydrophilic structure itself (Figure 1) affects the solu- bility and folding of proteins [1]. More importantly, it can be modified by several ER-localized glycosyl hydrolases and one glucosyl transferase in a cascade of reactions that, owing to the branched structure of the oligosaccharide, can result in a variety of structures that serve as ligands for carbohydrate binding proteins, lectins (Figure 2). These characteristics of N-glycans, the possibility to transfer them to different sites of proteins without the need of an extended primary protein acceptor sequence, the sequential processing of the oligosaccharide and the binding of specific lectins are a prerequisite for the function of N-glycan structures as clearly defined signalling molecules Review Corresponding authors: Aebi, M. ([email protected]); Molinari, M. ([email protected]). 74 0968-0004/$ see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.tibs.2009.10.001 Available online 21 October 2009

Upload: markus-aebi

Post on 13-Sep-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

N-glycan structures: recognition andprocessing in the ERMarkus Aebi1, Riccardo Bernasconi2, Simone Clerc1 and Maurizio Molinari2,3

1 Institute of Microbiology, Department of Biology, Eidgenossische Technische Hochschule (ETH) Zurich, CH-8093 Zurich,

Switzerland2 Institute for Research in Biomedicine, CH-6500 Bellinzona, Switzerland3 Ecole Polytechnique Federale de Lausanne, School of Life Sciences, CH-1015 Lausanne, Switzerland

Review

The processing of N-linked glycans determines secretoryprotein homeostasis in the eukaryotic cell. Folding anddegradation of glycoproteins in the endoplasmic reticu-lum (ER) are regulated by molecular chaperones andenzymes recruited by specific oligosaccharide structures.Recent findings have identified several components ofthis protein quality control system that specificallymodify N-linked glycans, thereby generating oligosac-charide structures recognized by carbohydrate-bindingproteins, lectins. In turn, lectins direct newly synthesizedpolypeptides to the folding, secretion or degradationpathways. The ‘‘glyco-code of theER’’ displays the foldingstatus of a multitude of cargo proteins. Deciphering thiscode will be instrumental in understanding proteinhomeostasis regulation in eukaryotic cells and for inter-vention because such processes can have crucial import-ance for clinical and industrial applications.

Biosynthesis of core oligosaccharides and their transferfrom a lipid donor onto nascent chainsAsparagine (N)-linked protein glycosylation is a covalentprotein modification that occurs across the three domainsof life: Bacteria, Archaea and Eukarya. In eukaryotic cells,N-linked glycosylation is the most prominent modificationof secretory proteins. This complex biosynthetic pathwayhas been studied best in the fungus Saccharomyces cere-visiae, but it seems to be highly conserved in other fungal,plant and animal species [1]. The pathway initiates at theER membrane where a lipid carrier, dolichylpyropho-sphate, serves as a membrane anchor for the assemblyof an oligosaccharide. The bipartite biosynthesis of theoligosaccharide initiates at the cytoplasmic side of theER membrane. Nucleotide-activated sugar donors serveas substrates for a series of different glycosyl transferasesthat lead to a mannose5-N-acetylglucosamine2 oligosac-charide. This lipid-bound oligosaccharide is translocatedacross themembrane [2] where an additional fourmannoseand three glucose residues are added. In the ER lumen,dolichylphosphomannose and dolichylphosphoglucose actas substrates for the individual glycosyl transferases. Thetransfer of the N-glycan precursor onto nascent polypep-tide chains is catalysed by oligosaccharyltransferase(OST), a protein complex consisting of eight different sub-units. OST utilizes the activated glucose3-mannose9-N-acetylglucosamine2 (Glc3Man9GlcNAc2) oligosaccharide

Corresponding authors: Aebi, M. ([email protected]);Molinari, M. ([email protected]).

74 0968-0004/$ – see front matter � 2009 Elsevie

donor as a substrate to covalently modify defined aspar-agine side chains within the acceptor sequence N-X-S/T(X cannot be proline) [1]. Most of the eukaryotic speciesstudied transfer this specific oligosaccharide structure toproteins (Figure 1 and Box 1).

N-glycosylation is in many respects a unique covalentprotein modification, but in the context of this review, thespatio-temporal aspect of the transfer reaction and thephysicochemical properties of the oligosaccharide are ofcentral importance.

The N-X-S/T sequon is a short and clearly defined

acceptor sequence

TheN-X-S/T sequon can be found frequently in proteins. Incontrast to other covalent post-translational modificationssuch as O-glycosylation or phosphorylation, no additional,large protein domain is required to define an N-linkedglycosylation site. However, several studies suggest thattheN-X-S/T sequence can bemodified only if it lies within aflexible domain of the polypeptide [3], suggesting that thepeptide acceptor takes on a defined structure in the courseof the modification reaction. Accordingly, the overall num-ber of proteins and sites that are glycosylated can beincreased by executing N-glycosylation before the foldingprocess, during or immediately after the translocation ofthe nascent chain into the ER lumen [4]. Indeed, in manyeukaryotic systems, the OST is associated with the trans-locon and with the ribosome [5]. It has been proposed thatdefined subunits of the OST might act as chaperones orenzymes to modulate, or even prevent, the folding of thetarget proteins in order to facilitate N-glycosylation [6,7].

A large, hydrophilic and branched oligosaccharide

structure is transferred to proteins

The hydrophilic structure itself (Figure 1) affects the solu-bility and folding of proteins [1]. More importantly, it canbe modified by several ER-localized glycosyl hydrolasesand one glucosyl transferase in a cascade of reactions that,owing to the branched structure of the oligosaccharide, canresult in a variety of structures that serve as ligands forcarbohydrate binding proteins, lectins (Figure 2).

These characteristics of N-glycans, the possibility totransfer them to different sites of proteins without the needof an extended primary protein acceptor sequence, thesequential processing of the oligosaccharide and the bindingof specific lectins are a prerequisite for the function ofN-glycan structures as clearly defined signalling molecules

r Ltd. All rights reserved. doi:10.1016/j.tibs.2009.10.001 Available online 21 October 2009

Figure 1. The N-linked core oligosaccharide structure. The core oligosaccharide is

composed of two N-acetylglucosamine (blue squares), nine mannose (green

circles) and three glucose residues (blue circles). A, B and C define the

oligosaccharide branch. Letters a–n are assigned to each hexose and are used

throughout the text. During biosynthesis, residues a–g are added onto the

cytosolic face of the ER membrane, and residues h–n are added after the

oligosaccharide precursor is flipped across the membrane. The linkage between

individual hexoses is shown.

Box 1. N-linked protein glycosylation occurs in all domains

of life

The Glc3Man9GlcNAc2 composition of the glycan that is transferred

to proteins is conserved in animal, plant and fungal species.

Pathways with reduced complexity have been identified in proto-

zoan species, but it is always a derivative of the above structure that

is transferred to proteins. For example, proteins expressed in

species belonging to the genera Trypanosoma, Leptomonas and

Herpetomona are modified with Man9GlcNAc2 oligosaccharides, in

Crithidia with Man7GlcNAc2, in some Leishmania species and

Blastocrithidia with Man6GlcNAc2 [78–80] and in Tetrahymena

pyroformis with Glc3Man5GlcNAc2 [81]. It has been proposed that

secondary loss of glycosyl transferases has led to the diversification

of N-glycosylation pathways within the eukaryotic domain [82]. In

an alternative model, eukaryotic N-linked protein glycosylation

originates from a prokaryotic (or perhaps archaeal?) process where

the Man5GlcNAc2 oligosaccharide is transferred to protein and

sequential addition of ER-lumen-oriented glycosyl transferases and

extension of OST complexity has led to the Glc3Man9GlcNAc2

producing pathway [1,82,83]. Regardless of the evolutionary path-

way that occurred in eukaryotes, experimental evidence indicates

that N-linked protein glycosylation in the three domains of life

shares common features:

1. Isoprenoid lipids are used as membrane anchors for the

assembly of the oligosaccharide at the plasma membrane/ER

membrane [84].

2. Assembly of the oligosaccharide takes place in the cytoplasm.

Oligosaccharide synthesis continues after translocation of the

lipid-linked oligosaccharide only in eukaryotes [85].

3. Although the oligosaccharide structures transferred to proteins

are highly diverse in the prokaryotic pathways and do not

resemble those found in eukaryotes, prokaryotic OST and the

catalytic subunit of the eukaryotic enzymes share a high level of

sequence similarity [86].

4. The consensus sequence to be N-glycosylated contains the N-X-

S/T sequon. In the one bacterial system studied so far, an

extended sequence, D/E-X1-N-X2-S/T, is required [87].

Review Trends in Biochemical Sciences Vol.35 No.2

in the process of protein folding and quality control.This signal, either alone or in combination with specificproperties of the covalently attached polypeptide, deter-mines the fate of a glycoprotein in the ER. Accordingly,not all N-linked glycans of a given protein are used as suchsignals [8,9].

Sequential processing of the oligosaccharide yieldsspecific N-glycan structures that direct protein folding,export or degradationGlycan-dependent protein folding

Processing of Glc3Man9GlcNAc2 by glucosidase I Thetriglucosylated form of the protein-bound oligosaccharidehas a half-life of a few seconds. The outermost glucoseresidue (glucose n, Figure 1) might in fact be removedimmediately after addition of the oligosaccharide onto thepolypeptide nascent chain by glucosidase I. This a1,2exoglucosidase is a type II membrane protein member

of the glycosyl hydrolase (GH) family 63 [10] that isassociated with the translocon complex in close proximityto OST [11]. This observation supports the hypothesis thatglucosidase I processing contributes to the efficiency ofglycosylation by shifting the direction of the equilibriumof the OST reaction towards the product. So far, theGlc2Man9GlcNAc2 structure generated by glucosidase Ihas attracted little interest; indeed, it was considered tobe a transient trimming intermediate rapidly processed byglucosidase II to themore long-livedmonoglucosylated formof the protein-bound oligosaccharide. However, the recentcharacterization of malectin, a well conserved ER-residenttype I membrane protein as a putative Glc2-high mannose-binding lectin [12], hints at a possible function of thisprocessing intermediate of N-linked glycans in proteinbiogenesis and quality control (Figure 2).

Processing of Glc2Man9GlcNAc2 and of Glc1Man9

GlcNAc2 by glucosidase II: association with and release

from calnexin and calreticulin The glucose residuesm andl (Figure 1) are removed by glucosidase II. This a1,3exoglucosidase is a luminal member of the GH family 31and contains a catalytic a-subunit and a regulatoryb-subunit [10]. The b-subunit contains a mannose6-phosphate receptor homology domain (MRH) [13] andcomprises the conserved C-terminal -XDEL sequence thatretains the holoenzyme in the ER lumen. The a-subunit

75

Figure 2. Several different oligosaccharide structures are displayed on immature cargo polypeptides located in the ER lumen. Most of the oligosaccharide structures, as

well as the glycosyl hydrolases involved in their generation, are conserved in mammals and in S. cerevisiae. Mammalian cells are characterized by the presence of a re-

glucosylating activity (UGT1) and a more extensive de-mannosylation of non-native polypeptides can occur, resulting in the removal of up to four a1,2-bonded mannose

residues by one or more members of the GH family 47 (Box 2). The right-hand column lists the signal encoded by the combination of a given N-glycan structure and the

associated polypeptide. Mammalian-specific features are highlighted with a grey background.

Review Trends in Biochemical Sciences Vol.35 No.2

alone possesses hydrolytic activity toward p-nitrophenyl a-D-glucopyranoside [14,15], but the b-subunit is required forprocessing the core oligosaccharide [14].

The regulation of glucosidase II function is not fullyunderstood. In a fast and possibly concerted action withglucosidase I, glucosidase II first removes glucose m(Figure 1) and generates Glc1Man9GlcNAc2, the ligandof the ER-resident lectin chaperones calnexin and calreti-culin (Figure 2). Calnexin is a type I transmembraneprotein and calreticulin is its soluble homolog. They arecomposed of a globular carbohydrate-binding domain,which folds into a leguminous (L)-type lectin-like b-sand-wich structure, and a proline-rich P-domain that recruitsthe lectin-associated oxidoreductase ERp57 [16,17]. ERp57catalyses the formation of native disulfide bonds in foldingpolypeptides. The different membrane topology of the twoER lectins determines their substrate specificity. Themembrane-anchored calnexin preferentially interacts withpolypeptide-bound monoglucosylated glycans close to themembrane, whereas calreticulin prefers association with

76

more peripheral glycans [18–20]. Reduction of the physicalconstraints upon release from calnexin/calreticulin/ERp57results in a structural collapse that often leads to attain-ment of the native polypeptide structure [21].

Uponrelease of the foldingpolypeptide fromcalnexinandcalreticulin, glucosidase II removes the innermost glucose l(Figure 1), thereby generating the Man9GlcNAc2 structure.Although both reactions catalysed by glucosidase II removean a1,3-linked glucose, the second cleavage event requiresthe transient separation and repositioning of the glucosi-dase II active site [22]. This time window is possiblyexploited by calnexin and calreticulin to associate withthe folding polypeptide [11]. When tested in isolated micro-someswith chains arrested in the translocon, glucosidase IIactivity progresses with high efficiency only when a secondN-linked glycan is available on the same polypeptide chain.It has been suggested that association of the b-subunit withthe 6’-tetramannosyl branch of an oligosaccharide allowsproper positioning of thea-subunit to cleave glucose l fromadistinct oligosaccharide in spatial proximity [11]. However,

Figure 3. Oligosaccharide preferences of ER-resident glycosyl transferase, glycosyl hydrolases, glucosyl transferase and lectins. The figure summarizes the transfer of the

core oligosaccharide on newly synthesized cargo polypeptides (OST), the progressive modification of the N-linked glycan by glycosyl hydrolases (glucosidases and

mannosidases) and a glucosyl transferase (UGT1). Different oligosaccharides act as ligands for a series of ER-localized lectins (malectin, CNX, CRT, cargo receptors, OS-9

and XTP3-B splice variants). A colour code shows arbitrary strength of affinities/activities of the given ER-resident sugar binding/modifying protein for a given

oligosaccharide structure (from white (no affinity/activity) to red (higher affinity/activity)). For cargo receptors and ERAD lectins see also Ref. [52].

Review Trends in Biochemical Sciences Vol.35 No.2

both in vitro [14,23,24] and in vivo [25], the presence of asingleN-linked glycan is sufficient for glucosidase II activityand for substrate associationwith calnexin and calreticulin.The rapid nature of the concerted action of glucosidase I andglucosidase II in living cells allows immediate association ofpolypeptide chains emerging in theER lumenwith calnexin,calreticulin and ERp57, thereby resulting in co-transla-tional formation of native disulfide bonds [26,27].

Processing of Man9GlcNAc2 by UGT1: the calnexin/

calreticulin cycle Glycoproteins can fold properly after asingle association with a lectin chaperone, as observed inmammalian cells [28] and in Schizosaccharomyces pombe[29]. However, propermaturation of the glycoproteinmightrequire more than one association [30,31]. The ER ofmulticellular eukaryotes, of certain fungi (e.g. S. pombeand Mucor rouxii) and of unicellular eukaryotes such asTrypanosoma cruzi [32] contains a folding sensor, theUDP-glucose:glycoprotein glucosyltransferase (UGT1 orUGGT), a member of the glycosyl transferase family 24[10]. UGT1 consists of a large N-terminal domain thatbinds non-native protein structures and a C-terminalcarbohydrate transferase domain [33]. Thisbifunctionality allows UGT1 to ‘‘inspect’’ polypeptidesthat display Man9GlcNAc2 oligosaccharides and toexclusively re-glucosylate the terminal mannose g(Figure 1) in polypeptides that have a pseudo-nativeprotein structure [34,35]. It remains controversialwhether UGT1 activity requires the proximity of theoligosaccharide to be re-glucosylated with a structuraldefect in the immature polypeptide chain [36,37].Regeneration of the Glc1Man9GlcNAc2 oligosaccharide

results in substrate re-association with calnexin/calreticulin/ERp57 for another round of folding-attempts(Figure 2). The repeated removal and re-addition of glucosel by the counteracting actions of glucosidase II and UGT1drives cycles of substrate release and re-association withthe lectin chaperones, the so-called calnexin/calreticulincycle [38]. During the off-phase, N-glycans are exposed toER-resident a1,2-mannosidases belonging to the GHfamily 47 [39] that can remove, one-by-one, the terminala1,2-bonded mannose residues. Progressive N-glycan de-mannosylation renders the associated polypeptide aweaker ligand for calnexin/calreticulin [40], a bettersubstrate for glucosidase II [23] and a suboptimalsubstrate for UGT1 (Figure 3) [41]. Altogether, itfacilitates the interruption of folding attempts anddirects folding-defective polypeptides into the ER-associated degradation (ERAD) pathway [42,43].

Glycan-dependent export of native proteins from the ER

Cargo receptors: VIPL, ERGIC-53 and VIP36

Glycopolypeptides are offered a time window to completetheir folding program in the calnexin chaperone system[42]. Upon attainment of the native structure, the majorityof cargo proteins are exported from the ER in vesiclescoated with cytosolic coatamer protein II (COPII) thatbud at ER exit sites. In mammalian cells, thesetransport vesicles undergo homotypic fusion to generatea stationary ER–Golgi intermediate compartment(ERGIC) from which cargo proteins reach the cis-Golgiin COPI-coated vesicles [44]. By contrast, COPII-coatedcargo vesicles in yeast are delivered directly to the Golgicompartment [45]. Transmembrane proteins can interact

77

Box 2. Oligosaccharide de-mannosylation in vitro and in vivo

Yeast Mns1p efficiently removes mannose i from Man9 oligosacchar-

ides to generate Man8B in vitro [39,88]. Although prolonged incuba-

tion (up to 24 h) with high concentrations of Mns1p results in the

generation of Man7 and Man6 sugars in vitro [88], this does not occur

in living cells where Mns1p generates only Man8B. In fact, further

trimming of protein-bound oligosaccharides in S. cerevisiae requires

Htm1p, the yeast ortholog of EDEM [56].

Mammalian ERManI also efficiently removes mannose i from Man9

oligosaccharides to generate Man8B in vitro [39,88]. In contrast to the

yeast protein, even after 24 hours incubation at non-physiologic

concentrations of enzyme, further oligosaccharide processing to

Man6-7 species by ERManI is negligible [88]. Consistently, over-

expression of ERManI in mammalian cells results in accumulation of

Man8 species [89]. We therefore propose that, like yeast, extensive

substrate de-mannosylation in mammalian cells requires interven-

tion by several members of the GH family 47. However, the identity of

the mannosidases supporting ERManI in extensive substrate de-

mannosylation remains a matter of controversy. Part of the

uncertainty stems from the lack of inhibitors that specifically

inactivate individual subclasses of the GH family 47 and from the

presence of multiple isoforms of EDEM and Golgi mannosidases that

could have redundant activities in the ER lumen, especially when

ectopically expressed. Based on the conservation of the yeast

machineries, EDEM proteins seem good candidates; however, no

in vitro glycanase activity has been detected so far. EDEM proteins

were initially described as mannosidase-like lectins devoid of

enzymatic activity because they lack two cysteines thought to be

conserved in all active a1,2-mannosidases [90–92] and because

substrate de-mannosylation is abolished in S. cerevisiae strains that

lack Mns1p [91]. However, more recent data show that: (i) the two

missing cysteines in EDEM proteins are dispensable for mannosi-

dase activity and are not conserved among a1,2-mannosidases [93];

(ii) the removal of mannose i by Mns1p is a prerequisite for further

oligosaccharide processing by Htm1 [56]; (iii) overexpression of

EDEM1 [65,94] or EDEM3 [89] enhances de-mannosylation of

folding-defective polypeptides in mammalian cells; (iv) EDEM

proteins conserve all catalytic residues required for enzymatic

activity, as well as the architecture of the active site and the residues

required to bind kifunensine, a specific inhibitor of a1,2-mannosi-

dases [39,95].

Review Trends in Biochemical Sciences Vol.35 No.2

directlywith the cytosolicCOPII coat,whereas soluble cargoproteins can require specific receptors for recruitment intoCOPII-coatedvesicles [46].ERexportof certainglycosylatedproteins is facilitated by several leguminous L-type lectinslocated in the ER (VIPL), cycling between ER and ERGIC(ERGIC-53) or between ERGIC and cis-Golgi (VIP36)[47–49]. Emp47p and Emp46p, yeast orthologs of ERGIC-53, have beenproposed to act as cargo receptors between theER and the Golgi in S. cerevisiae [50,51].

It is significant to note that VIP36 and VIPL preferen-tially associate with native proteins displaying high-man-nose glycans with three mannose residues, but not glucose,on the oligosaccharide branch A (i.e. Man9 oligosacchar-ides, Man8 structures lacking the terminal mannose resi-due on branch B or the one on branch C and Man7

structures lacking both terminal mannoses on branchesB and C) (Figure 3) [49]. These cargo export lectins havemuch lower affinity for extensively de-mannosylated gly-cans that, consistently, serve to tag the associated poly-peptide for disposal rather than for export [52]. A thirdcargo receptor, ERGIC-53, binds with lower affinity to abroader range of oligosaccharides, even if they are cappedwith a terminal glucose residue (Figure 3) [49].

Glycan-dependent quality control and ERAD

Folding-defective polypeptides or components that do notincorporate into protein complexes must eventually becleared from the ER folding environment. In both S. cer-evisiae and higher eukaryotes, the importance of mannosetrimming for ERAD of misfolded glycoproteins has beenreported [9,53,54], supporting the suggestion of a mannosetimer mechanism [55]. The slowly acting ER a1,2-manno-sidase I (ERManI in mammals; Mns1p in yeast) wasproposed to operate as such a molecular timer, with theMan8GlcNAc2 isomer B generated upon removal of thea1,2-bonded mannose of the central oligosaccharidebranch acting as the N-glycan degradation signal if dis-played on terminallymisfolded polypeptides (Figure 2, Box2). More recent results suggest that more extensive de-mannosylation of misfolded polypeptides occurs [52] andthat, at least in yeast, different members of the GH family

78

47 of a1,2-mannosidases intervene sequentially to gener-ate the ERAD signal [56]. In mammalian cells, the GHfamily 47 of a1,2-mannosidases comprises ERManI andthe ER degradation enhancer, mannosidase a-like proteinsEDEM1, EDEM2, EDEM3 (Htm1p in yeast) as well as thethree Golgi-resident MAN1A, MAN1B and MAN1C (Box 2and Box 3) [10].

Generating the degradation signal: processing of

Man8GlcNAc2 by members of the GH family 47 Recentanalyses in S. cerevisiae revealed that removal of at leasttwo a1,2-linked mannose residues from the protein-bound oligosaccharide is required for degradation ofglycoproteins. The N-glycan degradation signal is aterminal a1,6-linked mannose, generated by processingthe C branch (removal of mannose k, Figure 1) of aMan8GlcNAc2 glycan [56,57]. This reaction is catalysedby Htm1p (EDEM), which functionally depends on priorprocessing of theN-glycan by glucosidase I and II as well asMns1p (ERManI) [56]. Htm1p was found in a complex withthe oxidoreductase Pdi1p. Similar to the function proposedfor the EDEM1-associated mammalian reductase ERdj5[58], Pdi1p could be recruited to identify non-properlyfolded proteins as substrates for Htm1p or to promoteunfolding of ERAD substrates to facilitate theirdislocation across the ER membrane.

In higher eukaryotes, further processing of Man8-

GlcNAc2 was observed as well, and generation of Man7-

GlcNAc2, Man6GlcNAc2 and Man5GlcNAc2 glycanstructures was reported to precede or elicit degradation(Figure 2) [59–64]. Interestingly, in cell lines characterizedby N-glycosylation with aberrant oligosaccharides lackingthe cleavable mannose residues on branches B and C (e.g.the Chinese hamster ovary mutant lines B3F7 andMadIA214), processing of a1,2-linked mannose residuesis still required for glycoprotein disposal [64,65]. In thesecells, the only cleavable mannose residues are those on thebranch A that, whenmodified with a terminal a1,3-bondedglucose, determine polypeptide retention by the calnexinchaperone system. Extraction from calnexin/calreticulinis possible only by endomannosidase cleavage of branch

Box 3. De-mannosylation as a signal for disposal

In the model system S. cerevisiae, a terminal a1,6-linked mannose

generated by sequential removal of mannose i (by Mns1p) and k (by

Htm1p) (Figure 1) recruits the ERAD lectin Yos9p. Association of

misfolded polypeptides with Yos9p directs them to the site of

dislocation across the ER membrane, thus facilitating polypeptide

disposal [56,57]. Due to the presence of the calnexin/calreticulin cycle

and the existence of a multitude of ER-localized mannosidases, the

role of de-mannosylation and the subsequent recognition of ERAD

substrates in mammalian cells is less clear. Three different models

have been proposed:

Model 1. Removal of mannose i by ERManI generates the ERAD

signal

It was initially proposed that removal of mannose i (Figure 1) by

ERManI was required and sufficient to elicit disposal of folding-

defective polypeptides from the ER lumen. In this model, EDEM

proteins act as Man8-lectins to direct folding-defective polypeptides to

the ERAD pathway (Box 2) [91,92]

Model 2. Extensive substrate de-mannosylation by ERManI generates

the ERAD signal

Several reports show that misfolded polypeptides are subjected to

extensive de-mannosylation in the ER lumen [52]. ERManI shows

exquisite specificity for removing mannose i from oligosaccharides.

However, prolonged incubation with high concentrations of the

purified enzyme might result in more extensive de-mannosylation

(but refer to Box 2) [88]. It has been postulated that such conditions are

achieved in living cells upon enrichment of ERManI in the ER quality

control (ERQC) compartment [96]. In this model, EDEM proteins might

act as lectins and deliver ERAD substrates to the ERQC [52].

Model 3. Extensive substrate de-mannosylation by members of the

GH family 47 generates the ERAD signal

Extensive de-mannosylation of ERAD candidates requires the con-

certed and combined action of several members of the GH family 47;

namely, ERManI [60], and/or EDEM1-3 [65,89], and/or Golgi manno-

sidases MAN1A–C [97]. This scenario is closer to the model proposed

for S. cerevisiae, where the concerted intervention of Mns1p and

Htm1p eventually exposes the a1,6-bonded mannose j, which is a

ligand for the ERAD lectin Yos9p [56,57]. In mammalian cells, an

additional strong signal for disposal is represented by removal of

mannose g from the A branch of the protein-bound oligosaccharide

(Figure 1) [64,65]. As this is the only re-glucosylatable residue of

protein-bound oligosaccharides, its removal results in irreversible

polypeptide extraction from the calnexin cycle, thus interrupting futile

folding attempts.

Review Trends in Biochemical Sciences Vol.35 No.2

A or by preventing entry into the cycle due to exomannosi-dase removal of the re-glucosylatable mannose g (Figure 1).Thesedata highlight the importance of extraction of folding-defective glycopolypeptides from the calnexin/calreticulinchaperone system, which becomes irreversible, only uponde-mannosylation of branch A [66]. Based on the temporalorder of N-glycan processing that is determined by thesubstrate specificity of the different glycosyl hydrolases(Figures 2 and 3), we propose that removal of the terminala1,2-linked mannose of the A branch ensures the exit fromthe calnexin/calreticulin cycle and is necessary, but notsufficient, for glycoprotein degradation. Additional trim-ming of the C branch generates the terminal misfoldingsignal, the a1,6-linked mannose, recognized by the humanER-degradation lectins OS-9 (Yos9p) and XTP3-B [67–69],which target the N-glycoprotein for ERAD (Figure 2).

Interpreting the degradation signal: ER lectins Yos9p,

OS-9 and XTP3-B The finding that the chemical structureof the N-glycan is important in the degradation ofmisfolded glycoproteins led to the proposal that a lectinreceptor is involved in the recognition of ERAD substrates

Box 4. Key unanswered questions

The generation of a glycan signal that displays the folding status of a

protein requires the recognition of unfolded or partially folded protein

domains. It appears that specific sensors (UGT1) or components of

the ER protein folding machinery are instrumental for this purpose.

How this recognition is translated into a covalent modification of a

defined oligosaccharide, however, is not fully understood.

With respect to the identification of central activities required for

the generation of the glycan signals, the latest results from the

Aebi and Weissman laboratories have established the role of yeast

EDEM as an a1,2-mannosidase that specifically removes mannose

k from oligosaccharides [56,57]. By contrast, the role of EDEM

proteins in mammalian ERAD remains a matter of some debate.

Their intervention as active mannosidases is questioned and a

demonstration of their activity in vitro is still awaited. Reports

showing that EDEM1 forms functional complexes with calnexin to

sequester misfolded polypeptides released from this ER lectin [98]

[54,70,71]. Yos9p and mammalian OS-9.1/OS-9.2 andXTP3-B1/XTP3-B2 constitute a group of proteins thathave lectin-like domains with homology to the mannose6-phosphate receptor family and are implicated in therecognition of misfolded glyoproteins for degradation[68,69]. S. cerevisiae Yos9p is part of the Hrd1p complex(named for HMG-CoA reductase degradation) to which it isbound via Hrd3p, a transmembrane protein with a largeluminal domain, where it performs a proof-reading orgating function [72,73]. Hrd3p recruits the misfoldedproteins while Yos9p scans the substrate for the correctN-glycan structure, signalling terminal misfolding.Determination of the N-glycan substrate specificity ofYos9p revealed that it binds N-glycans that have aterminal a1,6-linked mannose [57]. This findingcorroborates the working model in which this N-glycanstructure, generated by Htm1p (EDEM) [56], is the N-glycan signal for glycoprotein degradation in yeast.Mammalian OS-9 and XTP3-B splice variants were foundalso in large complexes containing the HRD1 E3 ubiquitinligase and SEL1L (orthologs of Hrd1p and Hrd3p,respectively) [67–69,74,75]. The components of this

and/or with ERdj5 to unfold terminally misfolded polypeptides [58]

and/or with derlins [99] and/or SEL1L [94] to deliver misfolded

polypeptides to dislocons in the ER membrane and/or with ERManI

[100], thereby stabilizing and activating this short-lived glycosyl

hydrolase are certainly of interest but need to be confirmed by

studying the involvement in such complexes of the endogenous

proteins. The reports showing that putative ERAD lectins, such as

EDEM1 [94] and OS-9 [69], use their lectin sites to form complexes

with components of the dislocation machinery (i.e. SEL1L) rather

than with misfolded proteins are of particular interest because they

open new questions about the actual role of protein-bound

oligosaccharides in ERAD.

Finally, it remains to be shown that the ERAD system is required for

the degradation of misfolded proteins (as used for the experimental

visualisation of the process) and for the maintenance of ER protein

homeostasis in the secretory network of eukaryotic cells.

79

Review Trends in Biochemical Sciences Vol.35 No.2

mammalian complex are similar to those in S. cerevisiae,suggesting evolutionary conservation. The lectin activity ofOS-9 is required for enhancement of glycoprotein ERADand, likeYos9p,OS-9 canspecifically bindN-glycans lackingthe terminal a1,2-linked mannose from the C branch(Figure 3) [67]. However, a study using lectin mutantssuggested that the MRH domains of OS-9 and XTP3-Bmight not be required for binding to ERAD substrates,but rather for interaction with SEL1L [69], which carriesseveral N-glycans (Box 4) [76,77].

Concluding remarksThe unique properties of the N-glycosylation process,namely the transfer of a large oligosaccharide structureto a well defined but short acceptor sequence, are essentialprerequisites for the use of the N-linked glycan as a uni-versal signal. The signal displays the folding status of aprotein and is interpreted by the quality control machineryof protein homeostasis in the ER. In the course of eukar-yotic evolution, a fine-tuned, universal system, based ondifferential processing and recognition of N-linked glycanshas evolved that, in combination with the folding factors inthe ER, ensures proper folding of a multitude of differentproteins and homeostasis of secretory proteins.

Interestingly, we observe a correlation between thephylogenetic origin of the glycan domains and their rolein the glycan-dependent protein-processing in the ER: TheA-branch of the N-linked glycan, phylogenetically old andmade in the cytoplasm, is used as a signal in the primaryprocess of glycan-assisted protein folding. This requiresthat N-glycosylation occurs before folding. The B- and theC-branch of theN-linked glycan (phylogenetically younger)are used for the secondary process of protein degradationand they are the basis for the generation of the ‘‘degra-dation signal’’. In this combination, the quality controlprocess in the ER represents a prime example of a codethat is embedded in the seemingly complex structures of N-linked glycans. Years after the original proposal by AriHelenius that N-linked glycans serve as signals in thequality control of protein folding [55], we are still in theprocess of deciphering this code, how the signals are gener-ated and how they are interpreted. We can expect novelfascinating discoveries along this way.

AcknowledgementsM.M. is supported by grants from the Foundation for Research onNeurodegenerative Diseases, the Fondazione San Salvatore, the SwissNational Center of Competence in Research on Neural Plasticity andRepair, the Swiss National Science Foundation, the Synapsis Foundationand the Bangerter-Rhyner Foundation. Work in the laboratory of M.A. issupported by the Swiss National Science Foundation and the ETHZurich.

References1 Helenius, A. and Aebi, M. (2004) Roles of N-linked glycans in the

endoplasmic reticulum. Annu. Rev. Biochem. 73, 1019–10492 Helenius, J. et al. (2002) Translocation of lipid-linked oligosaccharides

across the ER membrane requires Rft1 protein. Nature 415, 447–4503 Petrescu, A.J. et al. (2004) Statistical analysis of the protein

environment of N-glycosylation sites: implications for occupancy,structure, and folding. Glycobiology 14, 103–114

4 Ruiz-Canada, C. et al. (2009) Cotranslational and posttranslational N-glycosylation of polypeptides by distinct mammalian OST isoforms.Cell 136, 272–283

80

5 Harada, Y. et al. (2009) Oligosaccharyltransferase directly binds toribosome at a location near the translocon-binding site. Proc. Natl.Acad. Sci. U.S.A. 106, 6945–6949

6 Wilson, C.M. et al. (2005) Ribophorin I associates with a subset ofmembrane proteins after their integration at the sec61 translocon. J.Biol. Chem. 280, 4195–4206

7 Schulz, B.L. et al. (2009) Oxidoreductase activity ofoligosaccharyltransferase subunits Ost3p and Ost6p defines site-specific glycosylation efficiency. Proc. Natl. Acad. Sci. U.S.A. 106,11061–11066

8 Spear, E.D. and Ng, D.T. (2005) Single, context-specific glycans cantarget misfolded glycoproteins for ER-associated degradation. J. CellBiol. 169, 73–82

9 Knop, M. et al. (1996) N-glycosylation affects endoplasmic reticulumdegradation of a mutated derivative of carboxypeptidase yscY inyeast. Yeast 12, 1229–1238

10 Cali, T. et al. (2008) The endoplasmic reticulum: crossroads for newlysynthesized polypeptide chains. Prog. Nucleic Acid Res. Mol. Biol. 83,135–179

11 Deprez, P. et al. (2005) More than one glycan is needed for ERglucosidase II to allow entry of glycoproteins into the calnexin/calreticulin cycle. Mol. Cell 19, 183–195

12 Schallus, T. et al. (2008) Malectin: a novel carbohydrate-bindingprotein of the endoplasmic reticulum and a candidate player in theearly steps of protein N-glycosylation. Mol. Biol. Cell 19, 3404–3414

13 Munro, S. (2001) TheMRH domain suggests a shared ancestry for themannose 6-phosphate receptors and other N-glycan-recognisingproteins. Curr. Biol. 11, R499–R501

14 Watanabe, T. et al. (2009) Genetic analysis of glucosidase II {beta}-subunit in trimming of high-mannose-type glycans. Glycobiology 19,834–840

15 Trombetta, E.S. et al. (2001) Quaternary and domain structure ofglycoprotein processing glucosidase II. Biochemistry 40, 10717–10722

16 Kozlov, G. et al. (2006) Crystal structure of the bb’ domains of theprotein disulfide isomerase ERp57. Structure 14, 1331–1339

17 Frickel, E.M. et al. (2002) TROSY-NMR reveals interaction betweenERp57 and the tip of the calreticulin P-domain. Proc. Natl Acad. Sci.U.S.A. 99, 1954–1959

18 Wada, I. et al. (1995) Chaperone function of calreticulin whenexpressed in the endoplasmic reticulum as the membrane-anchoredand soluble forms. J. Biol. Chem. 270, 20298–20304

19 Pieren, M. et al. (2005) The use of calnexin and calreticulin by cellularand viral glycoproteins. J. Biol. Chem. 280, 28265–28271

20 Molinari, M. et al. (2004) Contrasting functions of calreticulin andcalnexin in glycoprotein folding and ER quality control. Mol. Cell 13,125–135

21 Hebert, D.N. et al. (1996) Calnexin and calreticulin promote folding,delay oligomerization and suppress degradation of influenzahemagglutinin in microsomes. EMBO J. 15, 2961–2968

22 Petrescu, A.J. et al. (1997) The solutionNMR structure of glucosylatedN-glycans involved in the early stages of glycoprotein biosynthesisand folding. EMBO J. 16, 4302–4310

23 Totani, K. et al. (2006) Substrate specificity analysis of endoplasmicreticulum glucosidase II using synthetic high-mannose-type glycans.J. Biol. Chem. 281, 31502–31508

24 Wilkinson, B.M. et al. (2006) Yeast GTB1 encodes a subunit ofglucosidase II required for glycoprotein processing in theendoplasmic reticulum. J. Biol. Chem. 281, 6325–6333

25 Vanoni, O. et al. (2008) Consequences of individual N-glycan deletionsand of proteasomal inhibition on secretion of active BACE. Mol. Biol.Cell 19, 4086–4098

26 Chen, W. et al. (1995) Cotranslational folding and calnexin bindingduring glycoprotein synthesis. Proc. Natl. Acad. Sci. U.S.A. 92, 6229–

623327 Molinari, M. and Helenius, A. (1999) Glycoproteins form mixed

disulphides with oxidoreductases during folding in living cells.Nature 402, 90–93

28 Solda, T. et al. (2007) Substrate-specific requirements for UGT1-dependent release from calnexin. Mol. Cell 27, 238–249

29 Fanchiotti, S. et al. (1998) The UDP-Glc:glycoproteinglucosyltransferase is essential for Schizosaccharomyces pombeviability under conditions of extreme endoplasmic reticulum stress.J. Cell Biol. 143, 625–635

Review Trends in Biochemical Sciences Vol.35 No.2

30 Wada, I. et al. (1997) Promotion of transferrin folding by cyclicinteractions with calnexin and calreticulin. EMBO J. 16, 5420–5432

31 VanLeeuwen, J.E. andKearse, K.P. (1997) Reglucosylation ofN-linkedglycans is critical for calnexin assemblywithTcell receptor (TCR)alphaproteins but not TCRbeta proteins. J. Biol. Chem. 272, 4179–4186

32 Trombetta, S.E. et al. (1989) Glucosylation of glycoproteins bymammalian, plant, fungal, and trypanosomatid protozoamicrosomal membranes. Biochemistry 28, 8108–8116

33 Guerin, M. and Parodi, A.J. (2003) The UDP-glucose:glycoproteinglucosyltransferase is organized in at least two tightly bounddomains from yeast to mammals. J. Biol. Chem. 278, 20540–20546

34 Caramelo, J.J. et al. (2004) The endoplasmic reticulumglucosyltransferase recognizes nearly native glycoprotein foldingintermediates. J. Biol. Chem. 279, 46280–46285

35 Pearse, B.R. et al. (2008) A cell-based reglucosylation assaydemonstrates the role of GT1 in the quality control of a maturingglycoprotein. J. Cell Biol. 181, 309–320

36 Taylor, S.C. et al. (2004) The ER protein folding sensor UDP-glucoseglycoprotein-glucosyltransferase modifies substrates distant to localchanges in glycoprotein conformation.Nat. Struct. Mol. Biol. 11, 128–

13437 Ritter, C. et al. (2005) Minor folding defects trigger local modification

of glycoproteins by the ER folding sensor GT. EMBO J. 24, 1730–173838 Caramelo, J.J. and Parodi, A.J. (2008) Getting in and out from

calnexin/calreticulin cycles. J. Biol. Chem. 283, 10221–1022539 Moremen, K.W. and Molinari, M. (2006) N-linked glycan recognition

and processing: the molecular basis of endoplasmic reticulum qualitycontrol. Curr. Opin. Struct. Biol. 16, 592–599

40 Spiro, R.G. et al. (1996) Definition of the lectin-like properties of themolecular chaperone, calreticulin, and demonstration of itscopurification with endomannosidase from rat liver Golgi. J. Biol.Chem. 271, 11588–11594

41 Sousa, M.C. et al. (1992) Recognition of the oligosaccharide andprotein moieties of glycoproteins by the UDP-Glc:glycoproteinglucosyltransferase. Biochemistry 31, 97–105

42 Molinari, M. (2007) N-glycan structure dictates extension of proteinfolding or onset of disposal. Nat. Chem. Biol. 3, 313–320

43 Molinari, M., Calanca, V., Galli, C., Lucca, P. and Paganetti, P. (2003)Role of EDEM in the release of misfolded glycoproteins from thecalnexin cycle. Science 299, 1397–1400

44 Appenzeller-Herzog, C. and Hauri, H.P. (2006) The ER-Golgiintermediate compartment (ERGIC): in search of its identity andfunction. J. Cell Sci. 119, 2173–2183

45 Bonifacino, J.S. and Glick, B.S. (2004) The mechanisms of vesiclebudding and fusion. Cell 116, 153–166

46 Barlowe, C. (2003) Signals for COPII-dependent export from the ER:what’s the ticket out? Trends Cell Biol. 13, 295–300

47 Nyfeler, B. et al. (2008) Identification of ERGIC-53 as an intracellulartransport receptor of alpha1-antitrypsin. J. Cell Biol. 180, 705–712

48 Zhang, Y.C. et al. (2009) A review of ERGIC-53: its structure,functions, regulation and relations with diseases. Histol.Histopathol. 24, 1193–1204

49 Kamiya, Y. et al. (2008) Molecular basis of sugar recognition by thehuman L-type lectins ERGIC-53, VIPL, and VIP36. J. Biol. Chem.283, 1857–1861

50 Schroder, S. et al. (1995) The Golgi-localization of yeast Emp47pdepends on its di-lysine motif but is not affected by the ret1-1mutation in alpha-COP. J. Cell Biol. 131, 895–912

51 Sato, K. and Nakano, A. (2002) Emp47p and its close homologEmp46p have a tyrosine-containing endoplasmic reticulum exitsignal and function in glycoprotein secretion in Saccharomycescerevisiae. Mol. Biol. Cell 13, 2518–2532

52 Lederkremer, G.Z. (2009) Glycoprotein folding, quality control andER-associated degradation. Curr. Opin. Struct. Biol. 19, 1–9

53 Su, K. et al. (1993) Pre-Golgi degradation of yeast prepro-alpha-factorexpressed in a mammalian cell. Influence of cell type-specificoligosaccharide processing on intracellular fate. J. Biol. Chem. 268,14301–14309

54 Jakob, C.A. et al. (1998) Degradation of misfolded endoplasmicreticulum glycoproteins in Saccharomyces cerevisiae is determinedby a specific oligosaccharide structure. J. Cell Biol. 142, 1223–1233

55 Helenius, A. (1994) HowN-linked oligosaccharides affect glycoproteinfolding in the endoplasmic reticulum. Mol. Biol. Cell 5, 253–265

56 Clerc, S. et al. (2009) Htm1 protein generates the N-glycan signal forglycoprotein degradation in the endoplasmic reticulum. J. Cell Biol.184, 159–172

57 Quan, E.M. et al. (2008) Defining the glycan destruction signal forendoplasmic reticulum-associated degradation.Mol. Cell 32, 870–877

58 Ushioda, R. et al. (2008) ERdj5 is required as a disulfide reductase fordegradation of misfolded proteins in the ER. Science 321, 569–572

59 Kitzmuller, C. et al. (2003) Processing of N-linked glycans duringendoplasmic-reticulum-associated degradation of a short-livedvariant of ribophorin I. Biochem. J. 376, 687–696

60 Hosokawa, N. et al. (2003) Enhancement of endoplasmic reticulum(ER) degradation of misfolded Null Hong Kong alpha1-antitrypsin byhuman ER mannosidase I. J. Biol. Chem. 278, 26287–26294

61 Frenkel, Z. et al. (2003) Endoplasmic reticulum-associateddegradation of mammalian glycoproteins involves sugar chaintrimming to Man6-5GlcNAc2. J. Biol. Chem. 278, 34119–34124

62 Foulquier, F. et al. (2004) Endoplasmic reticulum-associateddegradation of glycoproteins bearing Man5GlcNAc2 andMan9GlcNAc2 species in the MI8-5 CHO cell line. Eur. J. Biochem.271, 398–404

63 Foulquier, F. et al. (2002) The unfolded protein response in a dolichylphosphate mannose-deficient Chinese hamster ovary cell line pointsout the key role of a demannosylation step in the quality-controlmechanism of N-glycoproteins. Biochem. J. 362, 491–498

64 Ermonval, M. et al. (2001) N-glycan structure of a short-livedvariant of ribophorin I expressed in the MadIA214 glycosylation-defective cell line reveals the role of a mannosidase that is not ERmannosidase I in the process of glycoprotein degradation.Glycobiology 11, 565–576

65 Olivari, S. et al. (2006) EDEM1 regulates ER-associated degradationby accelerating de-mannosylation of folding-defective polypeptidesand by inhibiting their covalent aggregation. Biochem. Biophys.Res. Commun. 349, 1278–1284

66 Ruddock, L.W. and Molinari, M. (2006) N-glycan processing in ERquality control. J. Cell Sci. 119, 4373–4380

67 Hosokawa, N. et al. (2009) Human OS-9, a lectin required forglycoprotein endoplasmic reticulum-associated degradation,recognizes mannose-trimmed N-glycans. J. Biol. Chem. 284, 17061–

1706868 Bernasconi, R. et al. (2008) A dual task for the Xbp1-responsive OS-9

variants in the mammalian endoplasmic reticulum: inhibitingsecretion of misfolded protein conformers and enhancing theirdisposal. J. Biol. Chem. 283, 16446–16454

69 Christianson, J.C. et al. (2008) OS-9 and GRP94 deliver mutantalpha1-antitrypsin to the Hrd1/SEL1L ubiquitin ligase complex forERAD. Nat. Cell Biol. 10, 272–282

70 Yang, M. et al. (1998) Novel aspects of degradation of T cell receptorsubunits from the ER in T cells: importance of oligosaccharideprocessing, ubiquitination, and proteasome-dependent removalfrom ER membranes. J. Exp. Med. 187, 835–846

71 Liu, Y. et al. (1997) Intracellular disposal of incompletely foldedhuman alpha1-antitrypsin involves release from calnexin and post-translational trimming of asparagine-linked oligosaccharides. J. Biol.Chem. 272, 7946–7951

72 Gauss, R. et al. (2006) A complex of Yos9p and the HRD ligaseintegrates endoplasmic reticulum quality control into thedegradation machinery. Nat. Cell Biol. 8, 849–854

73 Denic, V. et al. (2006) A luminal surveillance complex that selectsmisfolded glycoproteins for ER-associated degradation. Cell 126, 349–

35974 Mueller, B. et al. (2008) SEL1L nucleates a protein complex required

for dislocation of misfolded glycoproteins. Proc. Natl Acad. Sci. USA105, 12325–12330

75 Hosokawa, N. et al. (2008) Human XTP3-B forms an endoplasmicreticulum quality control scaffold with the HRD1-SEL1L ubiquitinligase complex and BiP. J. Biol. Chem. 283, 20914–20924

76 Mueller, B. et al. (2006) SEL1L, the homologue of yeast Hrd3p, isinvolved in protein dislocation from the mammalian ER. J. Cell Biol.175, 261–270

77 Lilley, B.N. and Ploegh, H.L. (2005) Multiprotein complexes that linkdislocation, ubiquitination, and extraction of misfolded proteins fromthe endoplasmic reticulum membrane. Proc. Natl Acad. Sci. U.S.A.102, 14296–14301

81

Review Trends in Biochemical Sciences Vol.35 No.2

78 Parodi, A.J. and Quesada-Allue, L.A. (1982) Protein glycosylation inTrypanosoma cruzi. I. Characterization of dolichol-boundmonosaccharides and oligosaccharides synthesized ‘‘in vivo’’. J.Biol. Chem 257, 7637–7640

79 Parodi, A.J. et al. (1984) Glycoprotein assembly in Leishmaniamexicana. Biochem. Biophys. Res. Commun. 118, 1–7

80 de la Canal, L. and Parodi, A.J. (1987) Synthesis of dolicholderivatives in trypanosomatids. Characterization of enzymaticpatterns. J. Biol. Chem. 262, 11128–11133

81 Taniguchi, T. et al. (1985) Carbohydrates of lysosomal enzymessecreted by Tetrahymena pyriformis. J. Biol. Chem. 260, 13941–

1394682 Banerjee, S. et al. (2007) The evolution of N-glycan-dependent

endoplasmic reticulum quality control factors for glycoproteinfolding and degradation. Proc. Natl. Acad. Sci. U.S.A. 104, 11676–

1168183 Burda, P. and Aebi, M. (1999) The dolichol pathway of N-linked

glycosylation. Biochim. Biophys. Acta 1426, 239–25784 Parodi, A.J. and Leloir, L.F. (1979) The role of lipid intermediates in

the glycosylation of proteins in the eucaryotic cell. Biochim. Biophys.Acta 559, 1–37

85 Helenius, J. and Aebi, M. (2002) Transmembrane movement ofdolichol linked carbohydrates during N-glycoprotein biosynthesis inthe endoplasmic reticulum. Semin. Cell Dev. Biol. 13, 171–178

86 Kelleher, D.J. and Gilmore, R. (2006) An evolving view of theeukaryotic oligosaccharyltransferase. Glycobiology 16, 47R–62R

87 Kowarik, M. et al. (2006) Definition of the bacterial N-glycosylationsite consensus sequence. EMBO J. 25, 1957–1966

88 Herscovics, A. et al. (2002) The specificity of the yeast and human classI ER alpha 1,2-mannosidases involved in ER quality control is not asstrict previously reported. Glycobiology 12, 14G–15G

89 Hirao, K. et al. (2006) EDEM3, a soluble EDEM homolog, enhancesglycoprotein ERAD andmannose trimming. J. Biol. Chem. 281, 9650–

9658

82

90 Lipari, F. andHerscovics, A. (1996) Role of the cysteine residues in thealpha1,2-mannosidase involved in N-glycan biosynthesis inSaccharomyces cerevisiae. The conserved Cys340 and Cys385residues form an essential disulfide bond. J. Biol. Chem 271,27615–27622

91 Jakob, C.A. et al. (2001) Htm1p, a mannosidase-like protein, isinvolved in glycoprotein degradation in yeast. EMBO Rep. 2, 423–430

92 Hosokawa,N. et al. (2001)AnovelER {alpha}-mannosidase-likeproteinaccelerates ER-associated degradation. EMBO Rep. 2, 415–422

93 Tatara, Y. et al. (2005) A single free cysteine residue and disulfidebond contribute to the thermostability of Aspergillus saitoi 1,2-alpha-mannosidase. Biosci. Biotechnol. Biochem. 69, 2101–2108

94 Cormier, J.H. et al. (2009) EDEM1 recognition and delivery ofmisfolded proteins to the SEL1L-containing ERAD complex. Mol.Cell 34, 627–633

95 Karaveg, K. et al. (2005) Mechanism of class 1 (glycosylhydrolasefamily 47) {alpha}-mannosidases involved in N-glycan processing andendoplasmic reticulum quality control. J. Biol. Chem. 280, 16197–

1620796 Avezov, E. et al. (2008) Endoplasmic reticulum (ER) mannosidase I is

compartmentalized and required for N-glycan trimming to Man56GlcNAc2 in glycoprotein ER-associated degradation. Mol. Biol.Cell 19, 216–225

97 Hosokawa, N. et al. (2007) Stimulation of ERAD of misfolded nullHong Kong alpha1-antitrypsin by Golgi alpha1,2-mannosidases.Biochem. Biophys. Res. Commun. 362, 626–632

98 Oda, Y. et al. (2003) EDEM as an acceptor of terminally misfoldedglycoproteins released from calnexin. Science 299, 1394–1397

99 Oda, Y. et al. (2006) Derlin-2 and Derlin-3 are regulated by themammalian unfolded protein response and are required for ER-associated degradation. J. Cell Biol. 172, 383–393

100 Termine, D.J. et al. (2009) The mammalian UPR boosts glycoproteinERAD by suppressing the proteolytic downregulation of ERmannosidase I. J. Cell Sci. 122, 976–984