
431ER quality control:towards an understanding at themolecularlevelLarsEllgaard*andAriHeleniustsubject to conformation-based screening by members ofTheprocessof'gualitycontrol'intheendoplasmicreticulum(ER) involves a varietyof mechanisms that collectivelyensuremajor molecular chaperone families.These chaperonesthatonlycorrectlyfolded,assembledandmodifiedproteinsarehave the capacity to recognize properties common to non-transportedalongthesecretorypathway.Incontrast,non-nativeproteinssuchasexposedhydrophobic areas.Theynativeproteinsareretainedandeventuallytargetedforselectively associate with proteins that display such fea-degradation.Recent work provides the first structural insightstures and in doing so promote folding and assemblyAslong as they are engaged in interactions with the substrateintotheprocessofglycoproteinfoldingintheERinvolvingthelectin chaperones calnexinand calreticulin.Underlyingproteins they also prevent export from the ER.principles governingthechoiceofchaperone systemengagedAnother level of quality control is specific for individuaby different proteins have also been discovered.protein species or protein families [3-5]. This levelincludes specialized folding factors such asHSP47.whichAddressesInstitute of Biochemistry,TH Zurich, Universitatstrasse 16,CH-8092only interacts with collagens [6]. It also includes specialZurich,Switzerlandescort proteins, such as the receptor associated protein*e-mail: lars.ellgaard@bc.biol.ethz.ch(RAP),whichaccompaniesLDL(lowdensitylipoprotein)te-mail:ari.helenius@bc.biol.ethz.chreceptor family members from the ER to the Golgi com-Current Opinion in Cell Biology 2001, 13:431437plex [7]. Moreover, there are also retention factors, such asTAP (transporter associated with antigen processing)and0955-0674/01/s-seefrontmattertapasin for MHC Class I antigens, that restrict transport to2001ElsevierScienceLtd.Allrightsreserveda limited setof conformers of specific substrateproteins[8]AbbreviationsCRTcalreticulinThe quality control process involves a complex sorting sys-CNXcalnexinERendoplasmic reticulumtem that separates proteins according to their folding andERADER-associated degradationmaturation status.The folding and assembly process isPDIprotein disulfideisomerasethus functionally coupled to export by vesicular transport.3Dthree-dimensionalFor certain proteins, transport relies on the exposure ofUPRunfolded protein responseUGGTUDP-glucose:glycoprotein glucosytransferasespecific signal sequences.Such signals, like the DXEsequences in the cytosolic domains of many membraneIntroductionproteins [9-11], may accelerate transport or, like RXRThe endoplasmic reticulum (ER)plays an essential role insequences in ion channels,they mayprevent prematurethefolding and maturation of newly synthesized proteins intransport [12,13]. Selective retrieval of misfolded proteinsthe secretorypathway.It provides an environment opti-bound to chaperonesfrom theintermediatecompartmentmized for folding,oxidation and oligomeric assembly ofor the cis-Golgi by retrograde transport to the ER has alsoproteins translocated into the lumen or inserted into thebeen reported [14].membrane. Folding in the ER is assisted by a large varietyWhereas folded proteins rapidly move via ER exit sites andof folding enzymes, molecular chaperones and folding sen-sors [1]. Many of these associate with growing nascentthe intermediate compartment to the cis-Golgi and beyond,chains and continue to assist folding until a protein haspersistently misfolded or unassembled proteins eitheraggregate or become degraded. Degradation is importantacquired its native structure.To ensurethe fidelityof thematuration process, exit from the ER is regulated by a strin-because the folding process is far from quantitative evengent quality control system that inhibits the secretion ofundernormal cellular growth conditions.In most casesincompletely folded or misfolded proteins [2]. In addition toER-associated degradation (ERAD) of misfolded proteinssecuring extended exposure of proteins to the foldinginvolves their retrotranslocation to the cytosol, ubiquitina-machinery,qualitycontrolpreventsdeplovmentofpoten-tion anddegradationbyproteasomes[15-17l.Nottially malfunctioning proteins that could be detrimental tosurprisingly,a number of diseases such as cystic fibrosis,i-antitrypsin deficiency and familial hypercholestero-the cell and the organism.For many proteins in the ERproper folding and maturation depends on co-and post-laemia are associated withER retention and degradation oftranslational modifications. Here we primarily focus on thefolding-defective mutant proteins [4,18].ER quality control system that is in place for proteinscontaining N-linked glycans.It is important to note that the accumulation of misfoldedproteins in the ER, especially observed under conditionsQualitycontrolmechanismsof stress,triggers activation ofa wide range ofgenes encod-ER quality control operates at several levels and by multi-ing for proteins of the secretory pathway. This is theple mechanisms [3]. At a general level, all proteins areso-called unfolded protein response (UPR)[19-21]
431 The process of ‘quality control’ in the endoplasmic reticulum (ER) involves a variety of mechanisms that collectively ensure that only correctly folded, assembled and modified proteins are transported along the secretory pathway. In contrast, nonnative proteins are retained and eventually targeted for degradation. Recent work provides the first structural insights into the process of glycoprotein folding in the ER involving the lectin chaperones calnexin and calreticulin. Underlying principles governing the choice of chaperone system engaged by different proteins have also been discovered. Addresses Institute of Biochemistry, ETH Zürich, Universitätstrasse 16, CH - 8092 Zürich, Switzerland *e-mail: lars.ellgaard@bc.biol.ethz.ch †e-mail: ari.helenius@bc.biol.ethz.ch Current Opinion in Cell Biology 2001, 13:431–437 0955-0674/01/$ — see front matter © 2001 Elsevier Science Ltd. All rights reserved. Abbreviations CRT calreticulin CNX calnexin ER endoplasmic reticulum ERAD ER-associated degradation PDI protein disulfide isomerase 3D three-dimensional UPR unfolded protein response UGGT UDP-glucose:glycoprotein glucosyltransferase Introduction The endoplasmic reticulum (ER) plays an essential role in the folding and maturation of newly synthesized proteins in the secretory pathway. It provides an environment optimized for folding, oxidation and oligomeric assembly of proteins translocated into the lumen or inserted into the membrane. Folding in the ER is assisted by a large variety of folding enzymes, molecular chaperones and folding sensors [1]. Many of these associate with growing nascent chains and continue to assist folding until a protein has acquired its native structure. To ensure the fidelity of the maturation process, exit from the ER is regulated by a stringent quality control system that inhibits the secretion of incompletely folded or misfolded proteins [2]. In addition to securing extended exposure of proteins to the folding machinery, quality control prevents deployment of potentially malfunctioning proteins that could be detrimental to the cell and the organism. For many proteins in the ER, proper folding and maturation depends on co- and posttranslational modifications. Here we primarily focus on the ER quality control system that is in place for proteins containing N-linked glycans. Quality control mechanisms ER quality control operates at several levels and by multiple mechanisms [3]. At a general level, all proteins are subject to conformation-based screening by members of major molecular chaperone families. These chaperones have the capacity to recognize properties common to nonnative proteins such as exposed hydrophobic areas. They selectively associate with proteins that display such features and in doing so promote folding and assembly. As long as they are engaged in interactions with the substrate proteins they also prevent export from the ER. Another level of quality control is specific for individual protein species or protein families [3–5]. This level includes specialized folding factors such as HSP47, which only interacts with collagens [6]. It also includes special escort proteins, such as the receptor associated protein (RAP), which accompanies LDL (low density lipoprotein) receptor family members from the ER to the Golgi complex [7]. Moreover, there are also retention factors, such as TAP (transporter associated with antigen processing) and tapasin for MHC Class I antigens, that restrict transport to a limited set of conformers of specific substrate proteins [8]. The quality control process involves a complex sorting system that separates proteins according to their folding and maturation status. The folding and assembly process is thus functionally coupled to export by vesicular transport. For certain proteins, transport relies on the exposure of specific signal sequences. Such signals, like the DXE sequences in the cytosolic domains of many membrane proteins [9–11], may accelerate transport or, like RXR sequences in ion channels, they may prevent premature transport [12,13]. Selective retrieval of misfolded proteins bound to chaperones from the intermediate compartment or the cis-Golgi by retrograde transport to the ER has also been reported [14]. Whereas folded proteins rapidly move via ER exit sites and the intermediate compartment to the cis-Golgi and beyond, persistently misfolded or unassembled proteins either aggregate or become degraded. Degradation is important because the folding process is far from quantitative even under normal cellular growth conditions. In most cases, ER-associated degradation (ERAD) of misfolded proteins involves their retrotranslocation to the cytosol, ubiquitination and degradation by proteasomes [15–17]. Not surprisingly, a number of diseases such as cystic fibrosis, α1-antitrypsin deficiency and familial hypercholesterolaemia are associated with ER retention and degradation of folding-defective mutant proteins [4,18]. It is important to note that the accumulation of misfolded proteins in the ER, especially observed under conditions of stress, triggers activation of a wide range of genes encoding for proteins of the secretory pathway. This is the so-called unfolded protein response (UPR) [19–21]. ER quality control: towards an understanding at the molecular level Lars Ellgaard* and Ari Helenius†

432Membranes and sortingFigure1Schematic representation of calnexin (CNX)P-domainTMDand calreticulin (CRT). (a) CNX is a type 1membrane protein of 570 residues, whereasCalnexin570(a)A111222CCRT is a soluble,lumenal protein of 400residues.The transmembranedomain (TMD)ofCNX is depicted in black. Regions A, B and Cshow 50-55% sequence identity [63]. TheP-domaincentral P-domain contains two sequence repeat111222A400CalreticulinOtypes, designated 1 and 2, each repeated fourtimes in CNX and three times in CRT. (b) Acartoon of the CRT P-domain NMR structure is(b)shown (reproduced with permission from[36]). Type 1 repeats are indicated in yellowand type 2 repeats in white. The three β-sheetsandang-helicaltumaredrawnasribbonsResidues of three hydrophobic clusters are65drawn as stick models.Several of the UPR-induced proteins are involved in pro-interaction with CRT and CNX is thus ensured. Overall,CRT and CNX together with glucosidase II and UGGTtein folding and glycosylation in the ER, in ERAD, in lipidmetabolism and in vesicular transport [22*]. Thus, qualitycooperateto increase thefoldingefficiency,to preventpre-control, ERAD and UPR are tightly coordinated processesmature oligomeric assemblyand topreventthe export of(recently reviewed in [23]).misfolded glycoproteins from the ER.The main featuresof the CNX/CRT cycleare now well established, and aThecalnexin/calreticulincyclethorough review has recently been published [26']. BelowOneof themostcommonmodifications of proteinstranslo-we address some detailed aspects of the cycle and discusscated into the ER is the addition of N-linked glycans. Forunresolved questions.glycoproteins, a particularly well studied ER quality con-Chaperone selectionintheendoplasmictrol systemis in place,involving two homologous lectins.calnexin (CNX)and calreticulin (CRT)(Figure 1).ThereticulumIn addition to CNX and CRT, the ER contains a large col-process of N-linked glycosylation occurs through the trans-fer of a triglucosylated, branched core oligosaccharidelection of other molecular chaperones and folding factors(GlcsMangGlcNAc2)tothenascentpolypeptidechain as itwith different properties and functions [27]. Each newlyenters theER lumen.Soon after transfer,trimming of thesynthesized protein makes use of only a few of the availablecore oligosaccharide by the successive action of ERchaperones. What are the parameters that determine whichglucosidases I and II to the monoglucosylated form,chaperones a protein will engage and in which order?Glc,MangGlcNAc2,allows theglycopolypeptideto interactwith CRT and CNX.Being lectins they specificallyRecent work shows that for glycoproteins the choice ofinteract with glycoproteins,but only if these havechaperone depends,in part, on the position of the glycansmonoglucosylated N-glycans [24]. Interaction with CNXin the sequence [28']. Growing nascent chains that haveand CRT exposes the folding glycoprotein to the associat-N-linked glycans within the first _50 residues from theed co-chaperone,ERp57,a thiol oxidoreductase of theamino terminus preferentially interact with CRT andproteindisulfideisomerase(PDI)family[25]CNX. In contrast, glycoproteins in which the glycans occurlater in the sequence first interact with Bi, an abundantThe association between substrate glycoprotein and lectinERchaperone of the Hsp70 family that binds to hydropho-is terminated by glucosidase II, which removes thebic peptide sequences,and later during post-translationalfolding withCRT and CNX.remaining glucose from the glycan. If, at this point, theglycoprotein has reached its native conformation, it is noWith respect to CRT and CNX, it is clear that despite theirlongerretainedin theERandcanbetransportedtotheGolgi complex.If not, re-addition of a glucose to thesequence similarity and identical oligosaccharide specifici-N-linked glycan occurs by the action of the UDP-ty in vitro [29-31], these two lectins exhibit only partiallyglucose:glycoproteinglucosyltransferase(UGGT),a lume-overlapping substrate specificities in vivo [32-34]nal enzyme that acts as a folding sensor [26'].TheMoreover, when theybind to the same protein, theyhaveglycoprotein is thereby ‘tagged'for renewed interactionin some cases been shown to interact with different gly-with CRT and CNX.Thepossibility of multiple rounds ofcans [35]. At the molecular level, the main differences are
Several of the UPR-induced proteins are involved in protein folding and glycosylation in the ER, in ERAD, in lipid metabolism and in vesicular transport [22••]. Thus, quality control, ERAD and UPR are tightly coordinated processes (recently reviewed in [23]). The calnexin/calreticulin cycle One of the most common modifications of proteins translocated into the ER is the addition of N-linked glycans. For glycoproteins, a particularly well studied ER quality control system is in place, involving two homologous lectins, calnexin (CNX) and calreticulin (CRT) (Figure 1). The process of N-linked glycosylation occurs through the transfer of a triglucosylated, branched core oligosaccharide (Glc3Man9GlcNAc2) to the nascent polypeptide chain as it enters the ER lumen. Soon after transfer, trimming of the core oligosaccharide by the successive action of ER glucosidases I and II to the monoglucosylated form, Glc1Man9GlcNAc2, allows the glycopolypeptide to interact with CRT and CNX. Being lectins they specifically interact with glycoproteins, but only if these have monoglucosylated N-glycans [24]. Interaction with CNX and CRT exposes the folding glycoprotein to the associated co-chaperone, ERp57, a thiol oxidoreductase of the protein disulfide isomerase (PDI) family [25]. The association between substrate glycoprotein and lectin is terminated by glucosidase II, which removes the remaining glucose from the glycan. If, at this point, the glycoprotein has reached its native conformation, it is no longer retained in the ER and can be transported to the Golgi complex. If not, re-addition of a glucose to the N-linked glycan occurs by the action of the UDPglucose:glycoprotein glucosyltransferase (UGGT), a lumenal enzyme that acts as a folding sensor [26•]. The glycoprotein is thereby ‘tagged’ for renewed interaction with CRT and CNX. The possibility of multiple rounds of interaction with CRT and CNX is thus ensured. Overall, CRT and CNX together with glucosidase II and UGGT cooperate to increase the folding efficiency, to prevent premature oligomeric assembly and to prevent the export of misfolded glycoproteins from the ER. The main features of the CNX/CRT cycle are now well established, and a thorough review has recently been published [26•]. Below we address some detailed aspects of the cycle and discuss unresolved questions. Chaperone selection in the endoplasmic reticulum In addition to CNX and CRT, the ER contains a large collection of other molecular chaperones and folding factors with different properties and functions [27]. Each newly synthesized protein makes use of only a few of the available chaperones. What are the parameters that determine which chaperones a protein will engage and in which order? Recent work shows that for glycoproteins the choice of chaperone depends, in part, on the position of the glycans in the sequence [28•]. Growing nascent chains that have N-linked glycans within the first ~50 residues from the amino terminus preferentially interact with CRT and CNX. In contrast, glycoproteins in which the glycans occur later in the sequence first interact with BiP, an abundant ER chaperone of the Hsp70 family that binds to hydrophobic peptide sequences, and later during post-translational folding with CRT and CNX. With respect to CRT and CNX, it is clear that despite their sequence similarity and identical oligosaccharide specificity in vitro [29–31], these two lectins exhibit only partially overlapping substrate specificities in vivo [32–34]. Moreover, when they bind to the same protein, they have in some cases been shown to interact with different glycans [35]. At the molecular level, the main differences are 432 Membranes and sorting Figure 1 Schematic representation of calnexin (CNX) and calreticulin (CRT). (a) CNX is a type 1 membrane protein of 570 residues, whereas CRT is a soluble, lumenal protein of 400 residues. The transmembrane domain (TMD) of CNX is depicted in black. Regions A, B and C show 50–55% sequence identity [63]. The central P-domain contains two sequence repeat types, designated 1 and 2, each repeated four times in CNX and three times in CRT. (b) A cartoon of the CRT P-domain NMR structure is shown (reproduced with permission from [36••]). Type 1 repeats are indicated in yellow and type 2 repeats in white. The three β-sheets and an α-helical turn are drawn as ribbons. Residues of three hydrophobic clusters are drawn as stick models. 1 570 1 400 Calnexin Calreticulin (a) (b) A B 1111 2 1112 2 2 222 C A B C P-domain P-domain TMD N C

ERqualitycontrol:towardsanunderstandingatthemolecularlevelEllgaardandHelenius433thatthe extendedP-domainarm in CRT(Figure 1)isThree main models can be proposed to explain howshorter than the corresponding arm in CNX [36] and thatUGGT recognizes its substrates. First, exposed hydropho-CRT is a soluble, lumenal protein, whereas CNX is abic peptide elements in the glycoprotein substrate couldtypeI membrane protein of the ER membrane.As previ-be recognized by the enzyme. It has been shown thatUGGT binds to immobilized hydrophobic peptides andously suggested[32,33,35],the latter differenceclearlyplays a role in substrate selection.Analyzing the substratesthat this interaction can be inhibited by denatured glyco-proteins [46]. Second, recognition could involve thebound to a soluble,anchor-free mutant of CNX and amembrane-bound version of CRT,Danilczyk and cowork-innermost GlcNAc unit of the oligosaccharide.In foldeders [37] recentlyobserved that the substrate specificities ofproteins this sugar interacts with neighboring amino acidthe two chaperones were essentially inverted.residues, an interaction which may be lost upon denatura-tion[26'].Finally,the enzymemayrecognizethedynamicERp57-catalyzeddisulfidebondformationproperties of the polypeptide moiety.In other words, itThe oxidizing environment and the presence of severalmay be sensitive to the mobility or deformability of thedifferentthiol oxidoreductasesallowformationofdisulfideprotein to which the glycan is connected.bonds in the ER.One of the oxidoreductases,ERp57,functionsasaco-chaperone withCRTandCNX[25,38°]UGGTstudies havebeenhampered bythelack of recom-It most probablyforms one-to-one complexes with bothbinant enzyme, the tendency of substrates to aggregateCRT and CNX [38'] and has been shown to accelerateand the heterogeneity of substrate glycoforms.With theoxidative refolding of monoglucosylated RNaseB in therecent expression of UGGT in insect cells and the use ofpresenceof CRTorCNXinvitro[39].Theformationofnon-aggregating, homogenous substrates such as glycopep-transientintermoleculardisulfidebondsbetweenERp57tides, yeast acid phosphatase and RNase B,more rigorousand newlysynthesized glycoproteins have,moreover, beenanalysis of this interesting and important enzyme shouldobserved in living cells [40]. When the association ofbepossible.CRT and CNX withglycoproteinswasblocked,formationThreedimensionalstructuresof calreticulinof mixed disulfides with ERp57 was also inhibited, indi-andcalnexincating that lectin binding is a prerequisite for substrateFor several years,3D structure determination of CRT andrecognition by this thiol oxidoreductase.CNX has been pursued by several groups.Now, as a firstSubstraterecognitionbyUDP-glucose:step towards a more detailed understanding of their func-glycoproteinglucosyltransferasetion at the molecular level theNMR structure of theUGGT, the folding sensor in the CNX/CRT cycle, is aCRTP-domain has been solved[36",50].In addition,thelarge, soluble, lumenal enzyme[26'].Its catalytic domaincrystal structure of a CNX ectodomain fragment,fordisplays a conserved 300 amino acid sequence at thewhich crystallization conditions have previously beencarboxyl terminus of the protein with homology to glyco-reported [51], has recently been solved but not yetsyltransferases of family8[41].UDP-glucose,transportedpublished [23].into the ER lumen from the cytosol [42], is the glucoseThe NMR structure of the CRT P-domain (residuesdonor, whereas the acceptors are glucose-free highmannoseoligosaccharides attached to incompletelyfolded189-288)showsan extendedhairpinfold comprisingtheglvcoproteins. UGGT is present throughout the ERentire polypeptide chain with amino and carboxyl terminiincludingthetransitional ER elements[43]in close spacial proximity ([36]; Figure 1).This unusualstructure constitutes a new fold. It is stabilized by threeThe exact mechanism by whichUGGT distinguishes foldedshort anti-parallel β-sheets as well as by three smallfrom non-native glycoproteins is not known. Given the largehydrophobic clusters each involving two highly conservedvariety of unrelated glycoproteins that serve as substrates, ittryptophyl residues, one from each strand of the hairpinis likely that the enzyme resembles classical molecular chap-The three-fold repetition of both the β-sheets and theerones in that it recognizes features shared by incompletelyhydrophobic clusters reflects the repetitive nature of thefolded proteins.The enzyme is specific forglycoproteins asP-domain sequence,which contains two sets of amino acidit uses neither glycans nor short glycopeptides as substratessequences each repeated three times (Figure 1).The[44]. Misfolded non-glycosylated proteins do not inhibittopology and the elongated shape of the P-domain suggestUGGT[44,45];however,a misfolded glycoprotein contain-thatitconstitutesanextended,somewhatcurved protru-ing only the innermost GlcNAc unit of the oligosaccharidesion from the CRT core domain [36]. This is incan inhibit it [46]. Furthermore, although UGGT does notagreement with the recent finding by gel filtration andrecognize glycoproteins in a random coil conformation,itsedimentation analysis that full-length CRT is an elongat-efficiently reglucosylates a variety of partially folded con-ed molecule [52'].formers [47].This is consistent with the finding that itsfunction in cells coincides with later stages of folding[48].InA brief mention of the unpublished ectodomain crystalglycoproteins with multiple domains, UGGT selectivelystructure byChevet andcoworkers[23]describes therecognizes glycans in the misfolded domains [49].CNX ectodomain as containing‘a lectin domain, as well
that the extended P-domain arm in CRT (Figure 1) is shorter than the corresponding arm in CNX [36••] and that CRT is a soluble, lumenal protein, whereas CNX is a type I membrane protein of the ER membrane. As previously suggested [32,33,35], the latter difference clearly plays a role in substrate selection. Analyzing the substrates bound to a soluble, anchor-free mutant of CNX and a membrane-bound version of CRT, Danilczyk and coworkers [37] recently observed that the substrate specificities of the two chaperones were essentially inverted. ERp57-catalyzed disulfide bond formation The oxidizing environment and the presence of several different thiol oxidoreductases allow formation of disulfide bonds in the ER. One of the oxidoreductases, ERp57, functions as a ‘co-chaperone’ with CRT and CNX [25,38•]. It most probably forms one-to-one complexes with both CRT and CNX [38•] and has been shown to accelerate oxidative refolding of monoglucosylated RNaseB in the presence of CRT or CNX in vitro [39••]. The formation of transient intermolecular disulfide bonds between ERp57 and newly synthesized glycoproteins have, moreover, been observed in living cells [40••]. When the association of CRT and CNX with glycoproteins was blocked, formation of mixed disulfides with ERp57 was also inhibited, indicating that lectin binding is a prerequisite for substrate recognition by this thiol oxidoreductase. Substrate recognition by UDP-glucose: glycoprotein glucosyltransferase UGGT, the folding sensor in the CNX/CRT cycle, is a large, soluble, lumenal enzyme [26•]. Its catalytic domain displays a conserved 300 amino acid sequence at the carboxyl terminus of the protein with homology to glycosyltransferases of family 8 [41]. UDP-glucose, transported into the ER lumen from the cytosol [42], is the glucose donor, whereas the acceptors are glucose-free high mannose oligosaccharides attached to incompletely folded glycoproteins. UGGT is present throughout the ER including the transitional ER elements [43]. The exact mechanism by which UGGT distinguishes folded from non-native glycoproteins is not known. Given the large variety of unrelated glycoproteins that serve as substrates, it is likely that the enzyme resembles classical molecular chaperones in that it recognizes features shared by incompletely folded proteins. The enzyme is specific for glycoproteins as it uses neither glycans nor short glycopeptides as substrates [44]. Misfolded non-glycosylated proteins do not inhibit UGGT [44,45]; however, a misfolded glycoprotein containing only the innermost GlcNAc unit of the oligosaccharide can inhibit it [46]. Furthermore, although UGGT does not recognize glycoproteins in a random coil conformation, it efficiently reglucosylates a variety of partially folded conformers [47]. This is consistent with the finding that its function in cells coincides with later stages of folding [48]. In glycoproteins with multiple domains, UGGT selectively recognizes glycans in the misfolded domains [49]. Three main models can be proposed to explain how UGGT recognizes its substrates. First, exposed hydrophobic peptide elements in the glycoprotein substrate could be recognized by the enzyme. It has been shown that UGGT binds to immobilized hydrophobic peptides and that this interaction can be inhibited by denatured glycoproteins [46]. Second, recognition could involve the innermost GlcNAc unit of the oligosaccharide. In folded proteins this sugar interacts with neighboring amino acid residues, an interaction which may be lost upon denaturation [26•]. Finally, the enzyme may recognize the dynamic properties of the polypeptide moiety. In other words, it may be sensitive to the mobility or deformability of the protein to which the glycan is connected. UGGT studies have been hampered by the lack of recombinant enzyme, the tendency of substrates to aggregate and the heterogeneity of substrate glycoforms. With the recent expression of UGGT in insect cells and the use of non-aggregating, homogenous substrates such as glycopeptides, yeast acid phosphatase and RNase B, more rigorous analysis of this interesting and important enzyme should be possible. Three dimensional structures of calreticulin and calnexin For several years, 3D structure determination of CRT and CNX has been pursued by several groups. Now, as a first step towards a more detailed understanding of their function at the molecular level, the NMR structure of the CRT P-domain has been solved [36••,50]. In addition, the crystal structure of a CNX ectodomain fragment, for which crystallization conditions have previously been reported [51], has recently been solved but not yet published [23]. The NMR structure of the CRT P-domain (residues 189–288) shows an extended hairpin fold comprising the entire polypeptide chain with amino and carboxyl termini in close spacial proximity ([36••]; Figure 1). This unusual structure constitutes a new fold. It is stabilized by three short anti-parallel β-sheets as well as by three small hydrophobic clusters each involving two highly conserved tryptophyl residues, one from each strand of the hairpin. The three-fold repetition of both the β-sheets and the hydrophobic clusters reflects the repetitive nature of the P-domain sequence, which contains two sets of amino acid sequences each repeated three times (Figure 1). The topology and the elongated shape of the P-domain suggest that it constitutes an extended, somewhat curved protrusion from the CRT core domain [36••]. This is in agreement with the recent finding by gel filtration and sedimentation analysis that full-length CRT is an elongated molecule [52•]. A brief mention of the unpublished ectodomain crystal structure by Chevet and coworkers [23] describes the CNX ectodomain as containing ‘a lectin domain, as well ER quality control: towards an understanding at the molecular level Ellgaard and Helenius 433

434Membranes and sortingFigure 2CRT show weak sequence similarity to legume lectins suchas pea lectin, which in turn shows structural homology togalectins and pentraxins [54].These proteins are character-ized by a β-sandwich structure containing two opposingCRTβ-sheets each of six or seven β-strands. Secondary structureERp57prediction of CRT and CNX using the PHDsec algorithmG55lshowsthatbothproteinsare.infact,likelytocontain10or 11β-strands outside the P-domain region.Therefore, thepossibility exists that the CRT/CNX lectin domain is char-Lacterized by a similar fold to legume lectins. However, incontrast to the well characterized plant lectins, which are2multivalent,CRT binding to monoglucosylated IgG ismonovalent[56].Interestingly,ERGIC-53andVIP36,twoCNXmannose specific lectins, shuttling between the ER and theERGolgi complex also both show sequence similarityto theleguminous lectins [57,58].Structural insights intoglycoproteinfoldingWith the oligosaccharide binding function of CRT andCNX mapped to a distinct lectin domain, the functions ofCytosolthe P-domain become all the more intriguing. A priori, theP-domain could be a site for direct interaction with unfold-edproteins.Two recent papersdescribein vitroRibosomeexperiments that suggest such a function for CRT andCNX [59°,60].Both proteins were found to bind to unfold-Current Opinion in CellBiologyed, non-glycosylated proteins but not to native conformers.A model of CNX, CRT and ERp57 interacting with a growingIn addition, they suppressed thermal denaturation andnascent chain (brown)of a glycoprotein in the ER.The growingaggregation and kept substrates in a folding-competentnascent chain of the glycoprotein is cotranslationally translocatedstate.Generally,interactions between classical molecularintothelumenoftheERthroughthetransloconcomplexwhichcontains a large, lumenal protrusion [64].Phosphorylation of thechaperones such as those belonging to the Hsp60, Hsp70cytosolic tail of CNx leads to increased association with theand Hsp90 families and their non-native substrates areribosome[65].CNX,CRTandERp57areknowntointeractco-andmediated by hydrophobic contacts. However, the highlypost-translationally with glycoprotein chains in the ER of live cells.charged surface of the CRT P-domain[36] provides noEarlyassociationwithCNXandCRTispossiblebecause,afteraddition of the core glycans by the oligosaccharyl tranferase enzyme,obvious sitesforprotein-proteininteractions of this sort.two of the glucoses are rapidly trimmed to generate theWhether the results described in [59',60] affect proteinmonoglucosylated form of the glycans (the glucose is represented byfoldingin vivo remains tobe seen.the blue circles labeled G'). The glycans bind to the lectin domainsof CNX and CRT. The P-domain, which forms a long, slightly curvedAlthough a role for the P-domain in binding to unfoldedarmextendingfromthelectindomain,is likelytogenerateapartiallyclosed space within which folding of the glycopolypeptide can occurprotein cannot be ruled out, it seems likely that it partici-inaprotected environment.It isalsopossible,asshownhereforpates in other protein-protein interactions. The topologyCRT,that theP-domainsof CNXandCRT interact withERp57of the P-domain places the tip of the hairpin loop at aallowingthisthiol oxidoreductasetointeractoptimallywith cysteinediscrete distance from the lectin domain.Likewise, thein the glycoprotein substrate and thereby promote proper formationofdisulfidebondsprotein moiety of a bound glycoprotein substrate wouldbe placed at a distance from the lectin domain due to thepresence of the glycan. Thus, the tip of the hairpin loopas a distinct loop', indicating a structure similar to CRT.could constitute a protein-binding site, with the mostTaken together, a uniform picture of the 3D structure ofobviousligandbeingtheco-chaperoneERp57.TheNMRCRT and CNX is emerging, where the P-domain consti-data show indications of slow conformational exchange intutes a finger-like extension from the globular corethe central region of the P-domain, suggesting a certaindegree of plasticity [36].This plasticity could endow thestructurethatisresponsiblefortheinteraction with theoligosaccharide of the substrate glycoprotein. That the P-boundERp57withsomefreedomofmovementthatdomaindoesnot containthecarbohydratebindingsiteiswouldallowittoadoptdifferentpositionsinrespecttotheconsistent with the lack of structural homology withglycoprotein substrates bound to the glycan-binding siteknown lectins and the experimental observation that theand thus allow access to cysteines at different positionsP-domain alone does not bind to glycoproteins [53].The intrinsic flexibility of CRT was recently deducedfrom its hydrodynamic properties and was suggested tobeAt present, we can only speculate about the structure of theof potential importance for the protein's function as alectin domain.However, outside the P-domain, CNX andmolecular chaperone [52']
as a distinct loop’, indicating a structure similar to CRT. Taken together, a uniform picture of the 3D structure of CRT and CNX is emerging, where the P-domain constitutes a finger-like extension from the globular core structure that is responsible for the interaction with the oligosaccharide of the substrate glycoprotein. That the Pdomain does not contain the carbohydrate binding site is consistent with the lack of structural homology with known lectins and the experimental observation that the P-domain alone does not bind to glycoproteins [53]. At present, we can only speculate about the structure of the lectin domain. However, outside the P-domain, CNX and CRT show weak sequence similarity to legume lectins such as pea lectin, which in turn shows structural homology to galectins and pentraxins [54]. These proteins are characterized by a β-sandwich structure containing two opposing β-sheets each of six or seven β-strands. Secondary structure prediction of CRT and CNX using the PHDsec algorithm [55] shows that both proteins are, in fact, likely to contain 10 or 11 β-strands outside the P-domain region. Therefore, the possibility exists that the CRT/CNX lectin domain is characterized by a similar fold to legume lectins. However, in contrast to the well characterized plant lectins, which are multivalent, CRT binding to monoglucosylated IgG is monovalent [56•]. Interestingly, ERGIC-53 and VIP36, two mannose specific lectins, shuttling between the ER and the Golgi complex also both show sequence similarity to the leguminous lectins [57,58]. Structural insights into glycoprotein folding With the oligosaccharide binding function of CRT and CNX mapped to a distinct lectin domain, the functions of the P-domain become all the more intriguing. A priori, the P-domain could be a site for direct interaction with unfolded proteins. Two recent papers describe in vitro experiments that suggest such a function for CRT and CNX [59•,60]. Both proteins were found to bind to unfolded, non-glycosylated proteins but not to native conformers. In addition, they suppressed thermal denaturation and aggregation and kept substrates in a folding-competent state. Generally, interactions between classical molecular chaperones such as those belonging to the Hsp60, Hsp70 and Hsp90 families and their non-native substrates are mediated by hydrophobic contacts. However, the highly charged surface of the CRT P-domain [36••] provides no obvious sites for protein–protein interactions of this sort. Whether the results described in [59•,60] affect protein folding in vivo remains to be seen. Although a role for the P-domain in binding to unfolded protein cannot be ruled out, it seems likely that it participates in other protein–protein interactions. The topology of the P-domain places the tip of the hairpin loop at a discrete distance from the lectin domain. Likewise, the protein moiety of a bound glycoprotein substrate would be placed at a distance from the lectin domain due to the presence of the glycan. Thus, the tip of the hairpin loop could constitute a protein-binding site, with the most obvious ligand being the co-chaperone ERp57. The NMR data show indications of slow conformational exchange in the central region of the P-domain, suggesting a certain degree of plasticity [36••]. This plasticity could endow the bound ERp57 with some freedom of movement that would allow it to adopt different positions in respect to the glycoprotein substrates bound to the glycan-binding site and thus allow access to cysteines at different positions. The intrinsic flexibility of CRT was recently deduced from its hydrodynamic properties and was suggested to be of potential importance for the protein’s function as a molecular chaperone [52•]. 434 Membranes and sorting Figure 2 A model of CNX, CRT and ERp57 interacting with a growing nascent chain (brown) of a glycoprotein in the ER. The growing nascent chain of the glycoprotein is cotranslationally translocated into the lumen of the ER through the translocon complex which contains a large, lumenal protrusion [64]. Phosphorylation of the cytosolic tail of CNX leads to increased association with the ribosome [65]. CNX, CRT and ERp57 are known to interact co- and post-translationally with glycoprotein chains in the ER of live cells. Early association with CNX and CRT is possible because, after addition of the core glycans by the oligosaccharyl tranferase enzyme, two of the glucoses are rapidly trimmed to generate the monoglucosylated form of the glycans (the glucose is represented by the blue circles labeled ‘G’). The glycans bind to the lectin domains of CNX and CRT. The P-domain, which forms a long, slightly curved arm extending from the lectin domain, is likely to generate a partially closed space within which folding of the glycopolypeptide can occur in a protected environment. It is also possible, as shown here for CRT, that the P-domains of CNX and CRT interact with ERp57 allowing this thiol oxidoreductase to interact optimally with cysteines in the glycoprotein substrate and thereby promote proper formation of disulfide bonds. CRT CNX ERp57 Translocon Translocon Ribosome Cytosol ER G G S S Current Opinion in Cell Biology

ER quality control: towards an understanding at the molecular level Ellgard and Helenius435Hurtley SM, Helenius A: Protein oligomerization in theAlthough speculative at present, an additional function ofendoplasmicreticulum.AnnuRevCellBiol1989,5:277-307the P-domain could be to act as a'diffusion barrier for gly-Ellgaard L, Molinari M, Helenius A: Setting the standards: quality3coproteins upon dissociation from the glycan-binding site.control in the secretory pathway.Science 1999,286:1882-1888As pointed out recently,the relativelylow affinity of CRTAridor M, Balch WE: Integration of endoplasmic reticulumfor monoglucosylated IgG of -105 M-1 implies rapid roundssignaling in health and disease. Nat Med 1999, 5:745-751.of association and dissociation [56'].A steric constraint onHerrmannJM,MalkusP,SchekmanR:OutoftheER-outfitters5the diffusion of the glycoprotein away from the chaperoneescortsandguides.Trends Cell Biol 1999,9:5-7.would ensurea higherlocal glycan concentration inthe6.NagataK:Hsp47:acollagenspecificmolecularchaperone.Trendsvicinity of the lectin domain and therebyfacilitate renewedBiochemSci1996,21:22-26.association.A similar effect would be achieved with multi-Bu G, Geuze HJ, Strous GJ, Schwartz AL: 39kDa receptor-associated protein is an ER resident proteins and molecularple glycan chains, which are known to stabilize thechaperone for LDL receptor related proteins.EMBOJ1995,complexes [61].These ideas havebeen incorporated into14:2269-2280themodel shown inFigure2whereCNX and CRT bind to8Ortmann B,Copeman J,LehnerPJ,SadasivanB,HerbergJAthe N-linked glycans of the growing nascent chain throughGrandea AG, Riddell SR, Tampe R, Spies T, Trowsdale J et al.:Acritical rolefortapasin in theassembly andfunction ofthe lectin domain. The P-domain of CRT, containing amultimeric MHC class I-TAP complexes.Science1997,bound molecule of ERp57,wraps around the chainto cre-277:1306-1309.ate a protective barrier around the folding chain and to9.Balch WE, McCaffery JM, Plutner H, Farquhar MG: Vesicularposition ERp57 for disulfide oxidation.As already menstomatitis virus is sorted and concentrated during export from theendoplasmicreticulum.Cell1994.76:841-852tioned, CNX and CRT are known to interact concurrently10.Nishimura N, Balch WE: A di-acidic signal required for selectivewith influenza hemagglutinin and other proteins cotransla-exportfromthe endoplasmic reticulum.Science1997,tionally and binding occurs to distinct sugars [35,62].277:556-558.11.Ma D, Zerangue N, Lin YF, Collins A, Yu M, Jan YN, Jan LY: Role ofConclusionsERexportsignalsincontrollingsurfacepotassiumchannelnumbers.Science2001,291:316-319.The quality control mechanisms in the ER ensure the struc-12.Zerangue N, Schwappach B, Jan YN, Jan LY: A new ER traffickingtural integrity of proteins delivered to the organelles of thesignalregulatesthesubunitstoichiometryofplasmamembranesecretory and endocytic pathways and the extracellular space.K(ATP)channels.Neuron1999,22:537-548.Recent progress has provided a better understanding of oxida-13.Zerangue N, Malan MJ, Fried SR, Dazin PF, Jan YN, Jan LY,tive folding of glycoproteins through the cooperation of ERp57Schwappach B:Analysis of endoplasmic reticulum traffickingand lectin chaperones CRT or CNX, of the basis for recogni-signals by combinatorial screening in mammalian cells. Proc NatlAcadSciUSA2001,98:2431-2436tion of unfolded substrate glycoprotein by the UGGT, as well14.Hammond C, Helenius A: Quality control in the secretory pathway:as of the structure ofCRT and CNX. Some of the'rules'under-retention of a misfolded viral membrane glycoprotein involveslying chaperone selection have also been deduced. In thiscycling between the ER,intermediate compartment and Golgicontext, it has to be kept in mind, however, that the chaperoneapparatus.JCell Bio/1994,126:41-52system is quite flexible, with chaperones able to substitute for15.WernerED,BrodskyJL,McCrackenAA:Proteasome-dependentendoplasmicreticulum-associatedproteindegradation:aneach other, cooperate in different ways and respond differentlyunconventionalroutetoafamiliarfate.ProcNatlAcadSciUSAto physiological changes and cellular stress.1996,93:13797-1380116.Hiller MM, Finger A, Schweiger M, Wolf DH: ER degradation of aWithregards to glycoprotein folding in theER,futurework ismisfolded luminal proteinbythecytosolicubiquitin-proteasomepathway.Science1996,273:1725-1728likely to focus on the interplay between the different compo-17.Bonifacino JS, Weissman AM: Ubiquitin and the control of proteinnents of the CNX/CRT cycle at the molecular level.fate in the secretory and endocytic pathways.Annu Rev Cell DevImportantunresolvedquestionsincludeifandhowconfor-Biol1998,14:19-57changes occur upon binding of substratemationalKopito RR,Ron D:Conformational disease.Nat Cell Biol 2000,18.glycoprotein by CRT and CNX,howthe co-chaperone2:E207-209ERp57isrecruitedbyCRTandCNX,howitmanagesto19.Mori K, Sant A, Kohno K, Normington K, Gething MU, Sambrook JF:interact productively with the large number of different sub-A 22 bp cis-acting element is necessary and sufficient for theinduction of the yeast KAR2 (BiP) gene by unfolded proteins.strates,and finally to what extent CRT and CNX are involvedEMBOJ1992,11:2583-2593.inprotein-proteininteractionswithunfoldedproteins?20.Chapman R, Sidrauski C, Walter P: Intracellular signaling from theendoplasmicreticulumtothenucleus.AnnuRevCellDevBiolAcknowledgements1998,14:459-485We are thankful to Rolf Moser for help with the figures and to Anna21.KaufmanRJ:Stress signalingfromthelumen ofthe endoplasmicMezzacasa and Christiane Ritter for critical reading of the manuscript Thereticulum:coordination of gene transcriptionaland translationalwork was supported by the Swiss National Science Foundation.controls.GenesDev1999,13:1211-1233.22.Travers KJ,Patil CK,Wodicka L,LockhartDJ,Weissman JS,WalterPReferencesandrecommendedreadingFunctional and genomic analysesreveal an essential coordination.Papers of particular interest,published within the annual period of review.betweentheunfoldedproteinresponseand ER-associatedhave been highlighted as:degradation.Cell2000,101:249-258.Using DNA microarrays, the authors investigate which genes are induced by: of special interestthe UPR in S. cerevisiae. Approximately 400 UPR targets were identified"*of outstanding interestrepresentinga widerangeofproteins of the secretorypathway.The demon-Gething MJ, Sambrook J: Protein folding in the cell. Nature 1992,stration ofcoordinated responses of UPR and ERAD for thedisposalofmisA355:33-45folded ER proteins links the two processes mechanistically
Although speculative at present, an additional function of the P-domain could be to act as a ‘diffusion barrier’ for glycoproteins upon dissociation from the glycan-binding site. As pointed out recently, the relatively low affinity of CRT for monoglucosylated IgG of ~105 M–1 implies rapid rounds of association and dissociation [56•]. A steric constraint on the diffusion of the glycoprotein away from the chaperone would ensure a higher local glycan concentration in the vicinity of the lectin domain and thereby facilitate renewed association. A similar effect would be achieved with multiple glycan chains, which are known to stabilize the complexes [61]. These ideas have been incorporated into the model shown in Figure 2 where CNX and CRT bind to the N-linked glycans of the growing nascent chain through the lectin domain. The P-domain of CRT, containing a bound molecule of ERp57, wraps around the chain to create a protective barrier around the folding chain and to position ERp57 for disulfide oxidation. As already mentioned, CNX and CRT are known to interact concurrently with influenza hemagglutinin and other proteins cotranslationally and binding occurs to distinct sugars [35,62]. Conclusions The quality control mechanisms in the ER ensure the structural integrity of proteins delivered to the organelles of the secretory and endocytic pathways and the extracellular space. Recent progress has provided a better understanding of oxidative folding of glycoproteins through the cooperation of ERp57 and lectin chaperones CRT or CNX, of the basis for recognition of unfolded substrate glycoprotein by the UGGT, as well as of the structure of CRT and CNX. Some of the ‘rules’ underlying chaperone selection have also been deduced. In this context, it has to be kept in mind, however, that the chaperone system is quite flexible, with chaperones able to substitute for each other, cooperate in different ways and respond differently to physiological changes and cellular stress. With regards to glycoprotein folding in the ER, future work is likely to focus on the interplay between the different components of the CNX/CRT cycle at the molecular level. Important unresolved questions include if and how conformational changes occur upon binding of substrate glycoprotein by CRT and CNX, how the co-chaperone ERp57 is recruited by CRT and CNX, how it manages to interact productively with the large number of different substrates, and finally to what extent CRT and CNX are involved in protein–protein interactions with unfolded proteins? Acknowledgements We are thankful to Rolf Moser for help with the figures and to Anna Mezzacasa and Christiane Ritter for critical reading of the manuscript. The work was supported by the Swiss National Science Foundation. References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: • of special interest ••of outstanding interest 1. Gething MJ, Sambrook J: Protein folding in the cell. Nature 1992, 355:33-45. 2. Hurtley SM, Helenius A: Protein oligomerization in the endoplasmic reticulum. Annu Rev Cell Biol 1989, 5:277-307. 3. Ellgaard L, Molinari M, Helenius A: Setting the standards: quality control in the secretory pathway. Science 1999, 286:1882-1888. 4. Aridor M, Balch WE: Integration of endoplasmic reticulum signaling in health and disease. Nat Med 1999, 5:745-751. 5. Herrmann JM, Malkus P, Schekman R: Out of the ER — outfitters, escorts and guides. Trends Cell Biol 1999, 9:5-7. 6. Nagata K: Hsp47: a collagen specific molecular chaperone. Trends Biochem Sci 1996, 21:22-26. 7. Bu G, Geuze HJ, Strous GJ, Schwartz AL: 39kDa receptorassociated protein is an ER resident proteins and molecular chaperone for LDL receptor related proteins. EMBO J 1995, 14:2269-2280. 8. Ortmann B, Copeman J, Lehner PJ, Sadasivan B, Herberg JA, Grandea AG, Riddell SR, Tampe R, Spies T, Trowsdale J et al.: A critical role for tapasin in the assembly and function of multimeric MHC class I-TAP complexes. Science 1997, 277:1306-1309. 9. Balch WE, McCaffery JM, Plutner H, Farquhar MG: Vesicular stomatitis virus is sorted and concentrated during export from the endoplasmic reticulum. Cell 1994, 76:841-852. 10. Nishimura N, Balch WE: A di-acidic signal required for selective export from the endoplasmic reticulum. Science 1997, 277:556-558. 11. Ma D, Zerangue N, Lin YF, Collins A, Yu M, Jan YN, Jan LY: Role of ER export signals in controlling surface potassium channel numbers. Science 2001, 291:316-319. 12. Zerangue N, Schwappach B, Jan YN, Jan LY: A new ER trafficking signal regulates the subunit stoichiometry of plasma membrane K(ATP) channels. Neuron 1999, 22:537-548. 13. Zerangue N, Malan MJ, Fried SR, Dazin PF, Jan YN, Jan LY, Schwappach B: Analysis of endoplasmic reticulum trafficking signals by combinatorial screening in mammalian cells. Proc Natl Acad Sci USA 2001, 98:2431-2436. 14. Hammond C, Helenius A: Quality control in the secretory pathway: retention of a misfolded viral membrane glycoprotein involves cycling between the ER, intermediate compartment and Golgi apparatus. J Cell Biol 1994, 126:41-52. 15. Werner ED, Brodsky JL, McCracken AA: Proteasome-dependent endoplasmic reticulum-associated protein degradation: an unconventional route to a familiar fate. Proc Natl Acad Sci USA 1996, 93:13797-13801. 16. Hiller MM, Finger A, Schweiger M, Wolf DH: ER degradation of a misfolded luminal protein by the cytosolic ubiquitin-proteasome pathway. Science 1996, 273:1725-1728. 17. Bonifacino JS, Weissman AM: Ubiquitin and the control of protein fate in the secretory and endocytic pathways. Annu Rev Cell Dev Biol 1998, 14:19-57. 18. Kopito RR, Ron D: Conformational disease. Nat Cell Biol 2000, 2:E207-209. 19. Mori K, Sant A, Kohno K, Normington K, Gething MJ, Sambrook JF: A 22 bp cis-acting element is necessary and sufficient for the induction of the yeast KAR2 (BiP) gene by unfolded proteins. EMBO J 1992, 11:2583-2593. 20. Chapman R, Sidrauski C, Walter P: Intracellular signaling from the endoplasmic reticulum to the nucleus. Annu Rev Cell Dev Biol 1998, 14:459-485. 21. Kaufman RJ: Stress signaling from the lumen of the endoplasmic reticulum: coordination of gene transcriptional and translational controls. Genes Dev 1999, 13:1211-1233. 22. Travers KJ, Patil CK, Wodicka L, Lockhart DJ, Weissman JS, Walter P: •• Functional and genomic analyses reveal an essential coordination between the unfolded protein response and ER-associated degradation. Cell 2000, 101:249-258. Using DNA microarrays, the authors investigate which genes are induced by the UPR in S. cerevisiae. Approximately 400 UPR targets were identified representing a wide range of proteins of the secretory pathway. The demonstration of coordinated responses of UPR and ERAD for the disposal of misfolded ER proteins links the two processes mechanistically. ER quality control: towards an understanding at the molecular level Ellgaard and Helenius 435

436Membranes and sorting23.Chevet E, Cameron PH, Pelletier MF,Thomas DY,Bergeron JJ:The39.Zapun A, Darby NJ, Tessier DC, Michalak M, Bergeron JJ, Thomas DYendoplasmic reticulum: integration ofprotein folding,quality.Enhanced catalysis of ribonuclease B folding by the interaction ofcontrol, signaling and degradation.Curr Opin Struct Biol 2001calnexin or calreticulin with ERp57.JBiol Chem 1998,11:120-124273:6009-6012Using purified components, it was clearly demonstrated that ERp57 accel-24.Hammond C, Braakman I, Helenius A:Role of N-linkederates the in vitro, oxidative refolding of monoglucosylated RNaseB only inoligosaccharides,glucose trimming and calnexin duringthepresenceof CRTorCNX.Thepaperprovides thefirstdirectevidenceglycoprotein folding in the endoplasmic reticulum.Proc Natl Acadthat the disulfide isomerase activity of ERp57 towards glycoproteins isSciUSA1994,91:913-917.enhanced through associationwith CRT and CNX.25.Oliver JD, van der Wal FJ, Bulleid NJ, High S: Interaction of the thiol-40.Molinari M, Helenius A: Glycoproteins form mixed disulphides withdependentreductaseERp57withnascentglycoproteins.Scienceoxidoreductases during folding in living cells. Nature 1999,1997,275:86-88402:90-93.26.Parodi A: Protein glucosylation and its role in protein folding. AnnuThe first demonstration that transient mixed disulfides are formed in vivoRevBiochem2000,69:69-93between thiol oxidoreductases PDI and ERp57 and their respective sub-An outstanding review on the role of N-linked glycosylation in protein folding.strate proteins.Furthermore, ERp57 was found to be the oxidoreductase ofchoice for CRT- and CNX-bound substrate glycoproteins.27.Ellgaard L, Helenius A: Folding of influenza hemagglutinin in theliving cellIn Mechanisms of Protein Folding:Frontiers in Molecular41.Tessier DC, Dignard D, Zapun A, Radominska-Pandya A, Parodi AJ,Biology,edn 2. Edited by Pain R. Oxford University Press;Bergeron J, Thomas DY: Cloning and characterization of2000:352-363.mammalian UDP-glucose glycoprotein: glucosyltransferase andthedevelopmentofaspecificsubstrateforthisenzyme.28.MolinariM,HeleniusA:Chaperoneselection during glyoproteinGlycobiology 2000,10:403-412.translocation into the endoplasmic reticulum.Science 2000,288:331-333.42.Vanstapel F,Blanckaert N: Carrier-mediated translocation ofUsing virus-infected cells, chaperone selection for viral glycoproteins duringuridinediphosphateqlucoseintothelumenofendoplasmictransiocation into the ER was investigated. The results show that if glyco-reticulum-derivedvesiclesfromratliver.JClinInvest1988,proteins contain an N-linked glycan within the first -50 residues from the82:1113-1122amino terminus they interact cotranslationally with CRT and CNX. If the gly-can appears later in the sequence, the initial chaperone interaction occurs43.Cannon KSl, Helenius A:Trimmingand readdition ofglucosetothrough BiP.N-Linked oligosaccharides determines calnexin association of asubstrate glycoprotein in living cells.JBiol Chem1999,29.WareFE,VassilakosA,PetersonPA,JacksonMR,LehrmanMAWilliamsDB:Themolecularchaperonecalnexinbinds274:7537-7544.Glc,MangGlcNAc2 oligosaccharides as an initial step in44.Sousa MC,Ferrero-Garcia MA,Parodi AJ:Recognition of therecognizing unfoided glycoproteins.J Biol Chem 1995.oligosaccharide and protein moieties of glycoproteins by the270:4697-4704.UDP-Glc: glycoprotein glucosyltransferase.Biochemistry1992,30. Spiro RG, Zhu Q, Bhoyroo V, Soling HD: Definition of the lectin-like31:97-105properties of the molecular chaperone,calreticulin,and45.Trombetta S,Bosch M,Parodi AJ:Glucosylation ofglycoproteins bydemonstration of its copurification withendomannosidase frommammlian,plant,fungal and trypanosomatid protozoaratliverGolgi.JBiolChem1996,271:11588-11594.microsomalmembranes.Biochemistry1989,28:8108-811631.Vassilakos A, Michalak M, Lehrman MA, Williams DB:Oligosaccharidebinding characteristics of themolecular46.SousaM,ParodiAJ:Themolecularbasisfortherecoqnitionofchaperonescalnexinandcalreticulin.Biochemistry1998misfolded glycoproteins by the UDP-Glc:glycoprotein37:3480-3490.glucosyltransferase.EMBOJ1995,14:4196-4203.32.PetersonJR,OraA,NquyenVanPHeleniusA:Transient,lectin-like47.Trombetta ES,Helenius A:Conformational requirements forassociationofcalreticulinwithfoldingintermediatesof cellularglycoprotein reglucosylation in the endoplasmic reticulum.J Celland viral glycoproteins. Mol Biol Cell 1995, 6:1173-1184.Biol 2000,148:1123-1130.Wada l, Imai S, Kai M,Sakane F,Kanoh H:Chaperone function of33.48.Labriola C, Cazzulo JJ, Parodi AJ: Trypanosoma cruzi calreticulin is acalreticulin when expressed in the endoplasmic reticulum as thelectin that binds monoglucosylated oligosaccharides but notmembrane-anchoredandsolubleforms.JBiolChem1995proteinmoietiesofglycoproteins.MolBiolCell1999270:20298-20304.10:1381-1394.34.Van Leeuwen JEM, KearseKP:The relatedmolecular chaperones49. Ritter C, Helenius A: Recognition of local misfolding by the ERcalnexinandcalreticulindifferentiallyassociatewithnascentfolding sensor UDP-glucose:glycoprotein glucosyltransferase. NatTcell antigen receptorproteinswithintheendoplasmic reticulum.Struct Biol 2000,7:278-280.JBiolChem1996,271:25345-25349.50.Ellgaard L, Riek R, Braun D, Herrmann T, Helenius A, Wuthrich K35.Hebert DN,Zhang JX, Chen W,Foellmer B, Helenius A:The numberThree-dimensional structure topology of the calreticulin P-domainand location of glycans on influenza hemagglutinin determinebased on NMR assignment.FEBS Lelt 2001,488:69-73.folding and association with calnexin and calreticulin.J Cell Biol1997,139:613-62351.Hahn M, Borisova S, Schrag JD, Tessier DC, Zapun A, Tom R.Kamen AA, Bergeron JJM, Thomas DY, Cygler M: Identification and36.Ellgaard L, Riek R, Herrmann T, Guntert P, Braun D, Helenius Acrystallization of a protease-resistant core of calnexin that retainsWuthrich K:NMRstructure of the calreticulin P-domain.Proc Natlbiological activity.J Struct Biol 1998,123:260-264.AcadSciUSA2001,98:3133-3138The three-dimensional structure of the CRT P-domain solved byNMR spec52.Bouvier M, Stafford WF: Probing the three-dimensional structuretroscopy reveals a novelhairpin fold formed bythe entire polypeptide chain.of human calreticulin.Biochemistry2000,39:14950-14959.The elongated structure is stabilized by three short anti-parallei β-sheets andStructural properties of CRT were characterized bythoroughbiochemicalthree small hydrophobic clusters. The structure provides new clues for theand biophysical analysis. Analytical ultracentrifugation and gel filtration indi-function of CRT as a molecular chaperonecate that the protein is a highly assymmetric molecule. Circular dichroism37.Danilczyk UG, Cohen-Doyle MF, Williams DB: Functionalmeasurements shows a relatively low thermal denaturation transition mid-relationship between calreticulin,calnexin,and the endoplasmicpoint of 42.5'C, whereas limited proteolysis identifies stable core domains.reticulumluminaldomainofcalnexin.JBiolChem2000,with one corresponding to the P-domain.275:13089-13097.53.Peterson JR,Helenius A:In vitro reconstitution of calreticulin-38.Oliver JD, Roderick HL, Llewellyn DH, High S: ERp57 functions as asubstrate interactions.J Cell Sci 1999,112:2775-2784.subunitofspecificcomplexesformedwiththeERlectinscalreticulin and calnexin. Mol Biol Cell 1999,10:2573-258254.VijayanM,ChandraN:Lectins.CurrOpinStructBiol1999Demonstration that both CRT and CNX can be isolated in discrete com-9:707-714.plexes with ERp57, but not PDI, from, for instance, semi-permeabilized cells55.Rost B, Sander C: Prediction of protein secondary structure atand microsomes. The interaction was shown to be independent of the pres-better than 70% accuracy.J Moi Biol1993,232:584-599.enceof substrateglycoprotein
23. Chevet E, Cameron PH, Pelletier MF, Thomas DY, Bergeron JJ: The endoplasmic reticulum: integration of protein folding, quality control, signaling and degradation. Curr Opin Struct Biol 2001, 11:120-124. 24. Hammond C, Braakman I, Helenius A: Role of N-linked oligosaccharides, glucose trimming and calnexin during glycoprotein folding in the endoplasmic reticulum. Proc Natl Acad Sci USA 1994, 91:913-917. 25. Oliver JD, van der Wal FJ, Bulleid NJ, High S: Interaction of the thioldependent reductase ERp57 with nascent glycoproteins. Science 1997, 275:86-88. 26. Parodi A: Protein glucosylation and its role in protein folding. Annu • Rev Biochem 2000, 69:69-93. An outstanding review on the role of N-linked glycosylation in protein folding. 27. Ellgaard L, Helenius A: Folding of influenza hemagglutinin in the living cell. In Mechanisms of Protein Folding: Frontiers in Molecular Biology, edn 2. Edited by Pain R. Oxford University Press; 2000:352-363. 28. Molinari M, Helenius A: Chaperone selection during glycoprotein • translocation into the endoplasmic reticulum. Science 2000, 288:331-333. Using virus-infected cells, chaperone selection for viral glycoproteins during translocation into the ER was investigated. The results show that if glycoproteins contain an N-linked glycan within the first ~50 residues from the amino terminus they interact cotranslationally with CRT and CNX. If the glycan appears later in the sequence, the initial chaperone interaction occurs through BiP. 29. Ware FE, Vassilakos A, Peterson PA, Jackson MR, Lehrman MA, Williams DB: The molecular chaperone calnexin binds Glc1Man9GlcNAc2 oligosaccharides as an initial step in recognizing unfolded glycoproteins. J Biol Chem 1995, 270:4697-4704. 30. Spiro RG, Zhu Q, Bhoyroo V, Söling HD: Definition of the lectin-like properties of the molecular chaperone, calreticulin, and demonstration of its copurification with endomannosidase from rat liver Golgi. J Biol Chem 1996, 271:11588-11594. 31. Vassilakos A, Michalak M, Lehrman MA, Williams DB: Oligosaccharide binding characteristics of the molecular chaperones calnexin and calreticulin. Biochemistry 1998, 37:3480-3490. 32. Peterson JR, Ora A, Nguyen Van P, Helenius A: Transient, lectin-like association of calreticulin with folding intermediates of cellular and viral glycoproteins. Mol Biol Cell 1995, 6:1173-1184. 33. Wada I, Imai S, Kai M, Sakane F, Kanoh H: Chaperone function of calreticulin when expressed in the endoplasmic reticulum as the membrane-anchored and soluble forms. J Biol Chem 1995, 270:20298-20304. 34. Van Leeuwen JEM, Kearse KP: The related molecular chaperones calnexin and calreticulin differentially associate with nascent T cell antigen receptor proteins within the endoplasmic reticulum. J Biol Chem 1996, 271:25345-25349. 35. Hebert DN, Zhang JX, Chen W, Foellmer B, Helenius A: The number and location of glycans on influenza hemagglutinin determine folding and association with calnexin and calreticulin. J Cell Biol 1997, 139:613-623. 36. Ellgaard L, Riek R, Herrmann T, Guntert P, Braun D, Helenius A, •• Wuthrich K: NMR structure of the calreticulin P-domain. Proc Natl Acad Sci USA 2001, 98:3133-3138. The three-dimensional structure of the CRT P-domain solved by NMR spectroscopy reveals a novel hairpin fold formed by the entire polypeptide chain. The elongated structure is stabilized by three short anti-parallel β-sheets and three small hydrophobic clusters. The structure provides new clues for the function of CRT as a molecular chaperone. 37. Danilczyk UG, Cohen-Doyle MF, Williams DB: Functional relationship between calreticulin, calnexin, and the endoplasmic reticulum luminal domain of calnexin. J Biol Chem 2000, 275:13089-13097. 38. Oliver JD, Roderick HL, Llewellyn DH, High S: ERp57 functions as a • subunit of specific complexes formed with the ER lectins calreticulin and calnexin. Mol Biol Cell 1999, 10:2573-2582. Demonstration that both CRT and CNX can be isolated in discrete complexes with ERp57, but not PDI, from, for instance, semi-permeabilized cells and microsomes. The interaction was shown to be independent of the presence of substrate glycoprotein. 39. Zapun A, Darby NJ, Tessier DC, Michalak M, Bergeron JJ, Thomas DY: •• Enhanced catalysis of ribonuclease B folding by the interaction of calnexin or calreticulin with ERp57. J Biol Chem 1998, 273:6009-6012. Using purified components, it was clearly demonstrated that ERp57 accelerates the in vitro, oxidative refolding of monoglucosylated RNaseB only in the presence of CRT or CNX. The paper provides the first direct evidence that the disulfide isomerase activity of ERp57 towards glycoproteins is enhanced through association with CRT and CNX. 40. Molinari M, Helenius A: Glycoproteins form mixed disulphides with •• oxidoreductases during folding in living cells. Nature 1999, 402:90-93. The first demonstration that transient mixed disulfides are formed in vivo between thiol oxidoreductases PDI and ERp57 and their respective substrate proteins. Furthermore, ERp57 was found to be the oxidoreductase of choice for CRT- and CNX-bound substrate glycoproteins. 41. Tessier DC, Dignard D, Zapun A, Radominska-Pandya A, Parodi AJ, Bergeron JJ, Thomas DY: Cloning and characterization of mammalian UDP-glucose glycoprotein: glucosyltransferase and the development of a specific substrate for this enzyme. Glycobiology 2000, 10:403-412. 42. Vanstapel F, Blanckaert N: Carrier-mediated translocation of uridine diphosphate glucose into the lumen of endoplasmic reticulum-derived vesicles from rat liver. J Clin Invest 1988, 82:1113-1122. 43. Cannon KSl, Helenius A: Trimming and readdition of glucose to N-Linked oligosaccharides determines calnexin association of a substrate glycoprotein in living cells. J Biol Chem 1999, 274:7537-7544. 44. Sousa MC, Ferrero-Garcia MA, Parodi AJ: Recognition of the oligosaccharide and protein moieties of glycoproteins by the UDP-Glc: glycoprotein glucosyltransferase. Biochemistry 1992, 31:97-105. 45. Trombetta S, Bosch M, Parodi AJ: Glucosylation of glycoproteins by mammlian, plant, fungal and trypanosomatid protozoa microsomal membranes. Biochemistry 1989, 28:8108-8116. 46. Sousa M, Parodi AJ: The molecular basis for the recognition of misfolded glycoproteins by the UDP-Glc:glycoprotein glucosyltransferase. EMBO J 1995, 14:4196-4203. 47. Trombetta ES, Helenius A: Conformational requirements for glycoprotein reglucosylation in the endoplasmic reticulum. J Cell Biol 2000, 148:1123-1130. 48. Labriola C, Cazzulo JJ, Parodi AJ: Trypanosoma cruzi calreticulin is a lectin that binds monoglucosylated oligosaccharides but not protein moieties of glycoproteins. Mol Biol Cell 1999, 10:1381-1394. 49. Ritter C, Helenius A: Recognition of local misfolding by the ER folding sensor UDP-glucose:glycoprotein glucosyltransferase. Nat Struct Biol 2000, 7:278-280. 50. Ellgaard L, Riek R, Braun D, Herrmann T, Helenius A, Wuthrich K: Three-dimensional structure topology of the calreticulin P-domain based on NMR assignment. FEBS Lett 2001, 488:69-73. 51. Hahn M, Borisova S, Schrag JD, Tessier DC, Zapun A, Tom R, Kamen AA, Bergeron JJM, Thomas DY, Cygler M: Identification and crystallization of a protease-resistant core of calnexin that retains biological activity. J Struct Biol 1998, 123:260-264. 52. Bouvier M, Stafford WF: Probing the three-dimensional structure • of human calreticulin. Biochemistry 2000, 39:14950-14959. Structural properties of CRT were characterized by thorough biochemical and biophysical analysis. Analytical ultracentrifugation and gel filtration indicate that the protein is a highly assymmetric molecule. Circular dichroism measurements shows a relatively low thermal denaturation transition midpoint of 42.5°C, whereas limited proteolysis identifies stable core domains, with one corresponding to the P-domain. 53. Peterson JR, Helenius A: In vitro reconstitution of calreticulinsubstrate interactions. J Cell Sci 1999, 112:2775-2784. 54. Vijayan M, Chandra N: Lectins. Curr Opin Struct Biol 1999, 9:707-714. 55. Rost B, Sander C: Prediction of protein secondary structure at better than 70% accuracy. J Mol Biol 1993, 232:584-599. 436 Membranes and sorting

ER quality control: towards an understanding at the molecular level Ellgaard and Helenius43756.Patil AR,Thomas CJ, Surolia A:Kinetics and themechanism of60. Ihara Y, Cohen-Doyle MF, Saito Y, Wiliams DB: Calnexininteraction of the endoplasmic reticulum chaperone,calreticulin,discriminates between protein conformational states andwith monoglucosylated (Glc,Man,GlcNAc2) substrate.J Biolfunctions as a molecular chaperone in vitro. Mol Cell 1999Chem2000,275:24348-24356.4:331-341.This paper contains a convincing demonstration that CRT interacts only withRodan AR, Simons J, Trombetta ES, Helenius A: N-linked61.the monoglucosylated (Glc,MangGlcNAc2) form of chicken IgG in vitro, thatoligosaccharidesarenecessaryandsufficientforassociationofthe binding is monovalent and that it does not involve any protein-proteinRNase B with calnexin and calreticulin. EMBOJ1996.interaction. A moderate affinity of CRT for its substrate in the range of15:6921-6930105 M-1 was determined by surface plasmon resonance.62.Hebert DN, Zhang JX, Helenius A:Proteinfolding and maturation in57.Hauri H, Appenzeller C, Kuhn F, Nufer O: Lectins and traffic in thea cell-free system. Biochem Cell Biol 1998, 76:867-873.secretory pathway.FEBS Lett 2000, 476:32-37.David V,Hochstenbach F,Rajagopalan S,Brenner MB:Interaction63.Fiedler K, Simons K: A putative novel class of animal lectins in the58.with newly synthesized and retained proteins in the endoplasmicsecretory pathway homologous to leguminous lectins. Cell 1994,reticulum suggests a chaperone function forhuman integral77:625-626membraneprotein IP9o(calnexin).JBiolChem1993,268:9585-959259.Saito Y hara Y, Leach MR, Cohen-Doyle MF, Williams DB:Calreticulin functions in vitro as a molecular chaperone for bothMenetret J, Neuhof A, Morgan DG, Plath K, Radermacher M, Rapoport TA64.glycosylated and non-glycosylated proteins.EMBO J1999,Akey CW:The structure of ribosome-channel complexes engaged in18:6718-6729protein translocation. Mol Cell 2000, 6:1219-1232.Using in vitro assays, CRT was shown to bind unfolded, glycosylated as wellas non-glycosylated proteins, prevent their aggregation and keep them in a65.Chevet E, Wong HN,Gerber D, Cochet C, Fazel A, Cameron PH,folding competent state. The paper indicates a potential role for CRT as aGushue JN,Thomas DY, Bergeron J:Phosphorylation by CK2 andclassical molecular chaperone and advocates the idea of a 'dual-bindingMAPKenhances calnexinassociationwith ribosomes.EMBOJmode of CRT to substrate glycoproteins.1999,18:3655-3666
56. Patil AR, Thomas CJ, Surolia A: Kinetics and the mechanism of • interaction of the endoplasmic reticulum chaperone, calreticulin, with monoglucosylated (Glc1Man9GlcNAc2) substrate. J Biol Chem 2000, 275:24348-24356. This paper contains a convincing demonstration that CRT interacts only with the monoglucosylated (Glc1Man9GlcNAc2) form of chicken IgG in vitro, that the binding is monovalent and that it does not involve any protein–protein interaction. A moderate affinity of CRT for its substrate in the range of ~105 M–1 was determined by surface plasmon resonance. 57. Hauri H, Appenzeller C, Kuhn F, Nufer O: Lectins and traffic in the secretory pathway. FEBS Lett 2000, 476:32-37. 58. Fiedler K, Simons K: A putative novel class of animal lectins in the secretory pathway homologous to leguminous lectins. Cell 1994, 77:625-626. 59. Saito Y, Ihara Y, Leach MR, Cohen-Doyle MF, Williams DB: • Calreticulin functions in vitro as a molecular chaperone for both glycosylated and non-glycosylated proteins. EMBO J 1999, 18:6718-6729. Using in vitro assays, CRT was shown to bind unfolded, glycosylated as well as non-glycosylated proteins, prevent their aggregation and keep them in a folding competent state. The paper indicates a potential role for CRT as a classical molecular chaperone and advocates the idea of a ‘dual-binding’ mode of CRT to substrate glycoproteins. 60. Ihara Y, Cohen-Doyle MF, Saito Y, Williams DB: Calnexin discriminates between protein conformational states and functions as a molecular chaperone in vitro. Mol Cell 1999, 4:331-341. 61. Rodan AR, Simons JF, Trombetta ES, Helenius A: N-linked oligosaccharides are necessary and sufficient for association of RNase B with calnexin and calreticulin. EMBO J 1996, 15:6921-6930. 62. Hebert DN, Zhang JX, Helenius A: Protein folding and maturation in a cell-free system. Biochem Cell Biol 1998, 76:867-873. 63. David V, Hochstenbach F, Rajagopalan S, Brenner MB: Interaction with newly synthesized and retained proteins in the endoplasmic reticulum suggests a chaperone function for human integral membrane protein IP90 (calnexin). J Biol Chem 1993, 268:9585-9592. 64. Menetret J, Neuhof A, Morgan DG, Plath K, Radermacher M, Rapoport TA, Akey CW: The structure of ribosome-channel complexes engaged in protein translocation. Mol Cell 2000, 6:1219-1232. 65. Chevet E, Wong HN, Gerber D, Cochet C, Fazel A, Cameron PH, Gushue JN, Thomas DY, Bergeron JJ: Phosphorylation by CK2 and MAPK enhances calnexin association with ribosomes. EMBO J 1999, 18:3655-3666. ER quality control: towards an understanding at the molecular level Ellgaard and Helenius 437