view - journal of virology - american society for microbiology

JOURNAL OF VIROLOGY,0022-538X/97/$04.0010

May 1997, p. 3742–3750 Vol. 71, No. 5

Copyright q 1997, American Society for Microbiology

Folding of Rabies Virus Glycoprotein: Epitope Acquisition andInteraction with Endoplasmic Reticulum Chaperones

YVES GAUDIN*

Laboratoire de genetique des virus du CNRS, 91198 Gif sur Yvette Cedex, France

Received 1 October 1996/Accepted 3 February 1997

Four well-characterized monoclonal antibodies (MAbs) directed against rabies virus glycoprotein (G) wereused to study G folding in vivo. Two of the MAbs were able to immunoprecipitate incompletely oxidized foldingintermediates. The two others recognized G only after folding was completed. By using these MAbs, the abilityof G to undergo low-pH-induced conformational changes during folding was also investigated. It appeared thatsome domains acquire this ability before folding is completed. In addition, interactions between unfolded Gand some of the molecular chaperones were analyzed. Unfolded G was associated with BiP and calnexin.Association with BiP was maximal immediately after the pulse, whereas association with calnexin was maximalafter 5 to 10 min of chase. The effects of tunicamycin and castanospermine on chaperone binding and foldingwere also studied. In the presence of both drugs, calnexin binding was reduced, consistent with the view thatcalnexin specifically recognizes monoglucosylated oligosaccharides, but some residual binding was still ob-served, indicating that calnexin also recognizes the polypeptide chain. In the presence of both drugs, associ-ation with BiP was increased and prolonged and folding was impaired. However, the global effects of the drugswere different, since folding was much more efficient in the presence of castanospermine than in the presenceof tunicamycin. Taken together, these results provide the basis to draw a schematic view of rabies virusglycoprotein folding.

Most glycoproteins are translated on membrane-bound ri-bosomes and inserted cotranslationally into the endoplasmicreticulum (ER) in an unfolded form. The ER contains a highconcentration of calcium and has an oxidizing potential redoxwhich allows disulfide bond formation. Like other cellularcompartments, it contains many folding enzymes and molecu-lar chaperones. Folding enzymes catalyze rate-limiting steps infolding such as proline isomerization and disulfide bond for-mation. The primary function of chaperones is to bind tonascent polypeptide chains in order to prevent aggregation andto limit potentially dead-end pathways (reviewed in reference18).In recent years, viral glycoprotein folding has been exten-

sively studied (reviewed in reference 10). Several viral glyco-proteins have been shown to begin to fold cotranslationally.For the hemagglutinin (HA) of influenza virus, it has beendemonstrated that conformational epitopes and some of thenative intrachain disulfide bonds are present before completesynthesis of the polypeptide chain (6). Folding continues post-translationally and can be monitored in pulse-chase experi-ments by observing the acquisition of conformational epitopes,disulfide bond formation, and enzymatic or receptor functions.Only correctly and completely folded molecules are trans-ported out of the ER. Unfolded viral glycoproteins are foundassociated with molecular chaperones, among which the beststudied are BiP and calnexin. BiP is a soluble member of theheat shock protein 70 (HSP 70) family of chaperones (38)which has been shown to associate transiently with foldingintermediates of many viral membrane proteins including in-fluenza virus HA (17, 24) and vesicular stomatitis virus (VSV)G (32). Calnexin is a phosphorylated transmembrane lectinwhich specifically interacts with partially trimmed, monoglu-cosylated N-linked oligosaccharides (19, 22, 41, 53). Monoglu-

cosylated oligosaccharides are generated by trimming the twooutermost glucose residues from the core oligosaccharides(27). This is achieved by glucosidase I, which removes the firstof the three glucoses, and glucosidase II, which removes thesecond (and eventually the third). Monoglucosylated oligo-saccharides can also be generated by reglucosylation ofhigh-mannose glycans through the action of UDP-glucose:gly-coprotein glucosyltransferase, which has the property of dis-tinguishing between folded and unfolded glycoproteins andreglucosylates only the latter (48, 49). The three enzymes (glu-cosidase I and II and the glucosyltransferase) are responsiblefor the deglucosylation-reglucosylation cycle that allows bind-ing of glycoproteins to and their release from calnexin (21, 22).This explains why inhibition of N-linked glycosylation oftenleads to defects at the level of folding. However, some resultssuggest that monoglucosylated N-linked oligosaccharides arenot the only determinants involved in calnexin binding to un-folded glycoproteins (53). It has been reported that some un-glycosylated proteins are also able to bind calnexin (5, 26, 31,43, 52) and that the transmembrane domain of glycoproteinsmay also play a role in this interaction (33).Rabies virus glycoprotein (G) is a type I membrane glyco-

protein. It is a trimer (3 3 65 kDa) that forms a spike extend-ing 8.3 nm from the viral membrane (14). The complete ma-ture glycoprotein molecule is 505 amino acids long (1, 57). Itsamino acid sequence indicates that it has three potential N-linked oligosaccharide acceptor sites, of which only one or twoare glycosylated, depending on the strain (2, 8, 9, 56). Theectodomain contains 14 cysteines forming disulfide bridges.G is responsible for adsorption of virus onto the host cell

and therefore determines the tissue tropism of the virus. Afterattachment and internalization of the virus via the endocyticpathway, G mediates low-pH-induced fusion of the viral enve-lope with the endosomal membrane (16, 37, 54). The pHthreshold for fusion is 6.3 6 0.1 and preincubation in theabsence of a target membrane below pH 6.75 leads to inhibi-tion of viral fusion properties. However, loss of fusion prop-

* Phone: 33 1 69 82 38 37. Fax: 33 1 69 82 43 08. E-mail: [email protected].

3742

on April 10, 2019 by guest

http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

erties can be reversed by readjusting the pH to above 7 (13, 15,16). This behavior is different from that observed with thebest-characterized fusogenic virus, influenza virus, for whichlow-pH-induced fusion inactivation is irreversible.It has been demonstrated that G can assume at least three

different states: the “native” (N) state detected at the viralsurface above pH 7; the activated hydrophobic state (A), whichinteracts with the target membrane as a first step of the fusionprocess; and the fusion-inactive conformation (I). There is apH-dependent equilibrium between these states which isshifted toward the I state at low pH (13, 15, 44). The differentconformations have been characterized by different biochem-ical and biophysical techniques, and it has been shown that theI conformation is 3 nm longer than the N one, from which it isalso antigenically distinct (13). We have proposed that G istransported through the Golgi apparatus in an I-like confor-mation to avoid nonspecific fusion during its transport in theacidic Golgi vesicles (12, 15). However, as yet nothing is knownabout rabies virus glycoprotein folding, and all the informationwe have is extrapolated from folding of the glycoprotein ofVSV (5, 19, 20, 32, 34), another rhabdovirus. This extrapola-tion is probably not valid: despite 60% homology between theglycoproteins of the Indiana and New Jersey serotypes, thegeneral features of the folding of both glycoproteins are nota-bly different (34).Monoclonal antibodies (MAbs) have been widely used to

study the intracellular transport, folding, and processing ofmany viral glycoproteins (40, 50, 58). Our laboratory possessesa collection of several hundred MAbs directed against G. Mostof them are neutralizing and therefore have allowed the selec-tion of mutants escaping neutralization (3, 42, 44, 46). Fromthis approach, two major and two minor antigenic sites weredefined at the G surface. More recently, the binding site ofnonneutralizing MAbs recognizing sodium dodecyl sulfate(SDS)-denatured G were localized by a method involving pep-tides (28).In this investigation, I used four well-characterized repre-

sentative MAbs to study rabies virus G folding in vivo. Theability of these MAbs to bind G after various periods of chasefollowing pulse-labeling was investigated and compared. Twoof the MAbs were able to immunoprecipitate incompletelyoxidized folding intermediates, whereas the other two recog-nized G only after its folding was complete. In addition, inter-actions between unfolded G and the molecular chaperonesBiP, calnexin, and GRP 94 were analyzed. The effect of glyco-sylation inhibitors such as tunicamycin and castanospermineon chaperone binding and folding was also studied. Takentogether, these results provide the basis for a cinematic view ofrabies virus G folding.

MATERIALS AND METHODS

MAbs.MAbs 30AA5, 18B5, 29EC2, and 17D2 have been described previously(12, 13, 28, 42, 44, 46). They were used as mouse ascites preparations. Anti-calnexin polyclonal antibody (SPA 860) and anti-BiP polyclonal antibody (SPA826, used in Western blot experiments) were purchased from Stress Gen. Anti-BiP polyclonal antibody (PA1-014, used in immunoprecipitation experiments)and anti-GRP94 rat MAb (MA3-016) were supplied by Affinity Bioreagents.Pulse-chase experiments and immunoprecipitations. BSR cells (a clone of

BHK 21 cells) grown in 30-mm-diameter petri dishes (about 5 3 105 cells) wereinfected with the PV strain at a multiplicity of infection of 5. At 21 h postinfec-tion, the culture medium was replaced with methionine- and cysteine-free me-dium and the culture was incubated for 1.5 h. The cells were then labeled with250 ml of prewarmed medium containing 100 mCi of [35S]methionine and[35S]cysteine (Promix; Amersham) for 3 min (in some cases, the labeling timewas longer; this is indicated in the figure legends). Prewarmed normal growthmedium containing 5 mM cold methionine and 1 mM cold cysteine was thenadded for the chase period. The cells were washed in cold TD buffer (137 mMNaCl, 5 mM KCl, 0.7 mM Na2HPO4, 25 mM Tris-HCl [pH 7.5]) containing 20mM N-ethylmaleimide (NEM) and lysed on ice in 500 ml of TD buffer containing

1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), 20mM NEM, and an antiprotease cocktail (2 mg of leupeptin per ml, 2 mg ofantipain per ml, 2 mg pepstatin per ml, 2 mg of chymostatin per ml, 16 mg ofaprotinin per ml). The lysates were centrifuged in an Eppendorf centrifuge for 5min at 12,000 3 g and 48C. The supernatant was incubated with anti-G oranti-chaperone antibodies for 1 h at 48C. Protein A-Sepharose was then added,and the mixture was incubated for 45 min at 48C. The resulting immune com-plexes were centrifuged and washed three times with TD buffer containing 1%CHAPS and the antiprotease cocktail. The complexes were boiled for 5 min inLaemmli buffer (under nonreducing conditions; b-mercaptoethanol was omit-ted) before analysis by SDS-polyacrylamide gel electrophoresis (PAGE) (8%polyacrylamide) and autoradiography. Quantification of radioactivity was per-formed with a PhosphorImager (Molecular Dynamics).To detect interactions between BiP and G, TD buffer was replaced by TB

buffer (20 mM NaCl, 5 mM CaCl2, 25 mM Tris-HCl [pH 7.5]). During cell lysis,the TB buffer was supplemented with apyrase (30 U/ml) to deplete the mediumof ATP. The other adjuvants (NEM, CHAPS, and antiproteases) were present asin other immunoprecipitations.To investigate the ability of labeled G to undergo the low-pH-induced con-

formational changes, the supernatants were split into two 250-ml portions afterremoval of insoluble debris. One portion was diluted twofold in TD buffercontaining 1% CHAPS, and immunoprecipitations were performed as describedabove. The other portion was diluted twofold in 31 mM citric acid–138 mMdibasic sodium phosphate (pH 6.4). Incubation with MAbs and protein A-Sepharose and all the washing steps were performed at pH 6.4 (in the presenceof 1% CHAPS and the antiprotease cocktail).Western blot analysis. BSR cells grown in 30-mm petri dishes were infected

with the PV strain at a multiplicity of infection of 5. At 21 h postinfection, thecells were washed with cold TD buffer containing 20 mM NEM. They were thenlysed in the same buffer (or TB buffer supplemented with 30 U of apyrase per mlto detect BiP) containing 1% CHAPS and the antiprotease cocktail, and G wasimmunoprecipitated with the different MAbs as described in the previous sec-tion. Proteins in the immunoprecipitates were separated by SDS-PAGE (10%polyacrylamide) and electrophoretically transferred to a nitrocellulose mem-brane. The membrane was then incubated with anti-chaperone antibodies. Thebinding was revealed by the use of anti-rabbit or anti-rat antibodies coupled toperoxidase. The peroxidase activity was detected with the light-based enhancedchemiluminescence system as described by the manufacturer (Amersham).Endo H treatment. Immune complexes associated with protein A-Sepharose

were incubated for 5 min at 1008C in phosphate-citrate buffer (pH 5.6) contain-ing 1% SDS. The supernatant was split into two 30-ml portions; half of thesample was then treated with 8 mU of endo-b-N-acetylglucosaminidase H (endoH) for 14 h at 378C. The remaining half was incubated at 378C in the absence ofendo H and used as a control.Tunicamycin and castanospermine treatment. Tunicamycin was used at 5

mg/ml and was present during the methionine and cysteine deprivation, the pulse,and the chase. Castanospermine (1 mM) was added only 45 min before the pulseand was present in the pulse medium and the chase medium.

RESULTS

Acquisition of epitopes during folding of G. The kinetics ofacquisition of different epitopes of rabies virus G during fold-ing in the ER were investigated by using four previously char-acterized antibodies. All of them were able to bind native G(Table 1). 17D2 is directed against a peptide located betweenamino acids 255 and 272 (28) and recognized SDS-denaturedG in a Western blot. The epitope of 29EC2 is located in theamino-terminal part of G and contains amino acids 10, 13, and15 (12, 44). 30AA5 and 18B5 are directed against antigenicsites II and III respectively (42, 46). Rabies virus PV waschosen for this study because only one band of G with only onecarbohydrate chain is detected in Coomassie blue-stained gels(2). However, a minority species with two carbohydrate chainswas also detected in pulse-chase experiments (see below).Rabies virus-infected BSR cells were labeled with [35S]me-

thionine and [35S]cysteine for 3 min and chased for periods ofup to 80 min. At the end of each chase period, the cells weretreated with NEM to block sulfhydryl groups and trap foldingintermediates. The cells were then lysed and immunoprecipi-tated with the different antibodies. The immunoprecipitateswere analyzed by SDS-PAGE under both reducing and non-reducing conditions (Fig. 1). SDS-PAGE under nonreducingconditions has been widely used to monitor disulfide bondformation (4, 6, 20, 50). The protein mobility increases as the

VOL. 71, 1997 RABIES VIRUS GLYCOPROTEIN FOLDING IN THE ER 3743


http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

disulfide bridges are formed, forcing the protein to migrate ina more compact conformation.Both 17D2 and 29EC2 were able to recognize G immedi-

ately after the pulse. At this time, two bands, labeled G1 andG2, were seen in reduced gels (Fig. 1, MAbs 17D2 and29EC2). G1 corresponds to the untrimmed glycoproteins hav-ing one carbohydrate chain. G2 most probably corresponds toan untrimmed minor species with two carbohydrate chains.The amount of radioactivity in both G1 and G2 bands in-creased between 0 and 5 min of chase (Fig. 1 and 2). This isprobably caused by completion of nascent chains that had beenlabeled but not terminated during the pulse (4). However, it isdifficult to exclude the possibility that the antibodies do noteasily recognize very early-folding intermediates. During thechase, the untrimmed glycoproteins move to the Golgi com-plex, where the high-mannose carbohydrates are trimmed andwhere further addition of sugars occurs, leading to conversionof the glycoproteins into GT1 and GT2. However even after 80

min, at least one-third of the immunoprecipitated G still mi-grated as G1. After immunoprecipitation by 17D2 in a nonre-duced gel, four bands were seen immediately after the pulse.The faster-migrating band corresponds to fully oxidized, un-trimmed G (called NT) with only one carbohydrate chain. Theothers (labeled IT) correspond to incompletely or incorrectlyoxidized forms (the incompletely oxidized forms of G with twocarbohydrate chains are not seen, probably because they arepresent in insufficient amounts). The half-life of these foldingintermediates was less than 10 min (Fig. 1, 17D2). Aggregatesmigrating as diffuse bands at the top of the gel were alsodetected. They probably correspond to multimers formed byaberrant interchain disulfide bridges. Although 29EC2 wasable to immunoprecipitate the folding intermediates, largeamounts of antibodies and long exposure of the gel for auto-radiography were necessary to detect them, indicating that29EC2 has a higher affinity for the fully oxidized form.Neither the folding intermediates nor the aggregates were

immunoprecipitated by MAbs 30AA5 and 18B5: these anti-bodies were not able to recognize G immediately after thepulse (Fig. 1). For both MAbs, the amount of G immunopre-cipitated increased during the chase and reached a maximumabout 40 min after the pulse (Fig. 2). Therefore, the form of Grecognized by MAbs 30AA5 and 18B5 is a more completelyfolded form than the one recognized by 17D2 and 29EC2.Acquisition of the ability to undergo low-pH-induced con-

formational changes. To identify the moment when G or someof its domains were able to undergo low-pH-induced confor-mational changes, experiments were performed as in the pre-vious section, except that the cell lysates were incubated at pH6.4 before immunoprecipitation. Under these conditions, ma-ture G undergoes low-pH-induced conformational changes(13, 15).The results were identical at pH 6.4 and 7.4 for MAb 29EC2

(data not shown), consistent with the fact that this antibody isable to recognize both the native and the low-pH conformation

FIG. 1. Recognition of pulse-labeled G by MAbs 17D2, 29EC2, 30AA5, and 18B5. Infected BSR cells were pulse-labeled with [35S]methionine for 3 min and thenchased in cold medium for the times indicated before lysis in TD–1% CHAPS–20 mMNEM (see Materials and Methods). G in the lysates was then immunoprecipitatedwith indicated MAbs and analyzed by SDS-PAGE (8% polyacrylamide) under reduced (R) or nonreduced (NR) conditions followed by autoradiography. Abbreviations:agg, aggregates; IT, incompletely or incorrectly oxidized forms; NT, fully oxidized untrimmed G; GT1 and GT2; fully trimmed glycoprotein having, respectively, oneand two carbohydrate chains; G1 and G2, precursor (not fully trimmed) of G having, respectively, one and two carbohydrate chains.

TABLE 1. Characteristics of the MAbs used to studythe G-folding pathway

MAb Epitopelocation

Recognition of:

SDS-treatedGa

Native G (N) Fusion-inactive G (I)

On thevirus

After solubi-lization

On thevirus

After solubi-lization

17D2b 255–270 1 6 1 2 229EC2c 10–15 2 2 1 1 130AA5d Site II 2 1 1 2 218B5e Site III 2 1 1 6 2

a As judged in a Western blot assay (28).b Characterized in references 28 and 45.c Characterized in references 12, 15, and 44.d Characterized in references 13, 15, and 42.e Characterized in references 45 and 46.

3744 GAUDIN J. VIROL.


http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

of G after its solubilization (15). In contrast, MAbs 18B5 and30AA5 were unable to immunoprecipitate G when the cellswere lysed at pH 6.4 (data not shown). As site II and, to alesser extent, site III are modified during the structural tran-sition from native to low-pH-inactivated G (13, 16), this resultindicates that the forms of G immunoprecipitated by 30AA5and 18B5 at pH 7.4 are able to undergo low-pH-induced con-formational changes.The results obtained with MAb 17D2 were particularly in-

teresting: when the cells were lysed at pH 6.4, the amount of Gimmunoprecipitated decreased as the chase period increased(Fig. 3). This is consistent with the fact that MAb 17D2 doesnot recognize the low-pH fusion-inactive conformation of G(45). Therefore, only unfolded G that is unable to undergo thelow-pH conformational change is immunoprecipitated. How-ever, this low-pH sensitivity precedes G recognition by MAbs18B5 and 30AA5 (compare Fig. 2 and 3), suggesting that somedomains of G are able to undergo the low-pH-induced struc-tural transition before folding is complete.Acquisition of endo H resistance. Acquisition of endo H

resistance by the sugar chains indicates the arrival of glycopro-teins in the medial cisternae of the Golgi (11). Rabies virus-infected cells were pulse-labeled for 5 min and then chased forup to 2 h. The proteins were immunoprecipitated and incu-bated in the presence or absence of endo H. The samples werethen analyzed by SDS-PAGE under reduced conditions. Fig-ure 4 shows that half of the G immunoprecipitated by 17D2had acquired endo H-resistant sugars after about 50 min, inagreement with results obtained for other rabies virus strain(35, 54). However, after 2 h of chase, a significant fraction of G(about 25%) still contained endo H-sensitive sugar chains.Interaction of unfolded G with BiP and calnexin. The inter-

actions between G and some of the chaperones of the ER werethen tested. For this purpose, infected-cell lysates were immu-noprecipitated with 17D2, 29EC2, 30AA5, or 18B5. The im-munoprecipitates were subjected to SDS-PAGE and trans-ferred to a nitrocellulose membrane. The membranes werethen incubated with a rabbit polyclonal serum directed againstBiP or calnexin or with a rat MAb directed against GRP 94.The binding reaction was revealed with the enhanced lumines-cence kit by using anti-rabbit or anti-rat antibodies coupled toperoxidase.

Calnexin was associated with the glycoproteins immunopre-cipitated by 17D2 and 29EC2 but was detected in only smallamounts in 30AA5 and 18B5 immunoprecipitates (Fig. 5A).Low ionic strength and ATP depletion from the lysates (seeMaterials and Methods) were necessary for BiP to be detectedin the 17D2 immunoprecipitates. Traces of this chaperonewere also present after immunoprecipitation by 29EC2. It wasnot found in 30AA5 and 18B5 immunoprecipitates (Fig. 5B).GRP 94 was never found associated with G after immunopre-cipitations, whatever the experimental conditions (data notshown), suggesting that there was no interaction between theseproteins or that this interaction was very weak.To identify the folding intermediates associated with the

chaperones and to investigate the kinetics of these associa-tions, cells were labeled for 3 min and immunoprecipitated atthe end of the chase period with a rabbit polyclonal serumdirected against BiP or calnexin. When the lysates were immu-noprecipitated with anti-BiP, a labeled BiP band at 78 kDa, aswell as G bands corresponding to IT and NT forms, could beobserved (Fig. 6A). The association was maximal immediatelyafter the pulse and then decreased rapidly with time.When the lysates were precipitated with anti-calnexin, a

labeled calnexin band at 90 kDa was observed (Fig. 6B). TheIT and NT forms of G were also seen. Although the associationwas detected immediately after the pulse, it was maximal after5 to 10 min of chase (at this time, virtually all G protein wascalnexin bound) and decreased thereafter. The association be-tween G and both chaperones was therefore transient.Effect of tunicamycin on the association of G with chaper-

ones and folding. Previous studies have shown that calnexinbinds monoglucosylated N-linked oligosaccharides (19, 22, 53).However, it has also been reported that some proteins were

FIG. 2. Kinetics of G recognition by MAbs. The amount of radioactivitycontained in the bands in the reduced panels of Fig. 1 was quantified with aPhosphorImager. For each MAb, the maximum amount of immunoprecipitatedG by this MAb was defined as 100% and the other values were calculated aspercentages thereof.

FIG. 3. Ability of pulse-labeled G to undergo low-pH-induced conforma-tional changes. Infected BSR cells were pulse-labeled with [35S]methionine for 3min and then chased in cold medium for the times indicated before lysis inTD–1% CHAPS–20 mM NEM. After elimination of insoluble debris, the lysateswere split in two. Half was incubated at pH 7.4 (lanes 7) and the other half wasincubated at pH 6.4 (lanes 6) before immunoprecipitation by MAb 17D2. Theimmunoprecipitates were analyzed by SDS-PAGE (8% polyacrylamide) underreduced conditions and autoradiography. The radioactivity in the bands wasquantified with a PhosphorImager, and the percentage of G immunoprecipitatedat pH 6.4 was calculated. The amount of G immunoprecipitated by MAb 17D2at pH 7.4 after the same chase period was defined as 100%.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

able to form a stable complex with calnexin in spite of theabsence of N-linked oligosaccharides (5, 26, 31, 43, 52). Tuni-camycin blocks core glycosylation of nascent glycoprotein pre-cursors. Therefore, we studied the effect of tunicamycin on the

formation of G-calnexin complexes and folding. Nascent G intunicamycin-treated cells (5 mg of culture medium per ml) didnot acquire N-linked oligosaccharides, and its molecularweight was reduced (Fig. 7). In the tunicamycin-treated cells,the association of G with calnexin was strongly inhibited. How-ever, PhosphorImager quantification revealed that about 13%of labeled unglycosylated G (UG) was associated with thechaperone after 10 min of chase (in contrast to more than 75%in the absence of the drug) (Fig. 7A). As already shown inother systems (5), an important proportion of the unglycosy-lated G associated with calnexin was aggregated after analysisin gels under nonreducing conditions (data not shown).In the presence of tunicamycin, G folding was impaired:

immunoprecipitations of pulse-labeled infected cells by 17D2and analysis of immunoprecipitates revealed that most of theprotein was associated in aggregates (Fig. 7B). Incompletely orincorrectly oxidized monomeric forms were also detected evenafter 1 h of chase. As previously observed for other glycopro-teins, incubation with tunicamycin also resulted in a massiveand prolonged interaction with BiP (see Fig. 9). Finally, Gsynthesized in tunicamycin-treated cells was poorly recognizedby 18B5 and 30 AA5 (see Fig. 10) even after 40 min of chase,indicating that G was unable to fold correctly under such con-ditions.Effect of castanospermine on the association of G with chap-

erones and folding. Castanospermine is an a-glucosidase in-hibitor that prevents glucose trimming. As such, it inhibitsglycoprotein binding to calnexin (19, 22). G was detected as amolecule migrating slightly more slowly in castanospermine-treated cells than in the untreated cells (Fig. 8B). As in tuni-

FIG. 4. Kinetics of acquisition of endo H resistance. Infected cells werepulse-labeled for 5 min and then chased for the times indicated. The proteinswere immunoprecipitated by MAbs 17D2 and then incubated in the presence(1) or absence (2) of endo H followed by SDS-PAGE (10% polyacrylamide)(see Materials and Methods). The graph represents the percentage of endoH-resistant oligosaccharides determined by a PhosphorImager.

FIG. 5. Coprecipitation of G with calnexin and BiP. Infected cells were lysed,and G was immunoprecipitated with the indicated MAbs. Immunoprecipitateswere revealed by Western blotting with anti-calxexin MAbs (A) or anti-BiP (B)(see Materials and Methods). CNX, Calnexin; H, immunoglobulin heavy chain ofanti-G MAbs (recognized by anti-rabbit antibodies coupled to peroxidase).

FIG. 6. Kinetics of association of pulse-labeled rabies virus G with BiP (A)and calnexin (B). Infected BSR cells were pulse-labeled with [35S]methionine for3 min and then chased in cold medium for the times indicated before lysis. Afterelimination of insoluble debris, the supernatant was split in two; half was immu-noprecipitated with 17D2 (lanes 17), and half was immunoprecipitated with theanti-chaperone antibody (Ab) (lanes B for anti-BiP and C for anti-calnexin). Theimmunoprecipitates were analyzed under nonreducing conditions by SDS-PAGE(8% polyacrylamide). CNX, calnexin; IT, incompletely or incorrectly oxidizedforms; NT, fully oxidized untrimmed G.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

camycin-treated cells, the association of G with calnexin waslargely inhibited (Fig. 8A). However, again, about 12% oflabeled G was associated with calnexin after 10 min of chase.This calnexin-associated G had not undergone glucose trim-ming as judged by its migration in SDS-PAGE (Fig. 8A), but,as in tunicamycin-treated cells, an important part of G associ-ated with calnexin in the presence of castanospermine wasaggregated in gels under nonreducing conditions (data notshown). G folding in castanospermine-treated cells was slowedand less efficient than in untreated cells: incompletely or in-correctly oxidized glycoproteins were still detectable after 40min of chase, and large aggregates were also present at the topof nonreduced gels (Fig. 8B). This slower folding resulted in aprolonged interaction with BiP (Fig. 9). However, a significantfraction of G was recognized by both 18B5 and 30AA5 (Fig.10), indicating that folding did occur in the presence of cas-tanospermine.The total amount of G recovered (from the cells and the

supernatant) was then quantified for chase periods up to 4 h. Inthe absence of castanospermine, about 80% of the G immu-noprecipitated after 30 min of chase was recovered after 4 hwhereas less than 50% was recovered in the presence of thedrug (data not shown). Therefore, G degradation was en-hanced in the presence of castanospermine.

DISCUSSION

This work provides a detailed picture of rabies virus G fold-ing in the ER. The folding steps identified on the pathway fromthe unfolded nascent chain to the folded G are described inFig. 11. The folding pathway of rabies virus G shows manysimilarities to those of VSV G and influenza virus HA. How-

ever, the folding is slower: the t1/2 for folding is about 20 minfor rabies virus G whereas it is less than 5 min for VSV G andHA. The folding efficiency is in the 70 to 80% range, which israther high (10).Almost immediately after the pulse, unfolded G is recog-

nized by MAbs 17D2 and 29EC2. The epitopes recognized bythese MAbs are not (29EC2) or poorly (17D2) accessible on

FIG. 7. Effect of tunicamycin (Tun) on pulse-labeled G association withcalnexin and folding. Infected BSR cells were incubated in the presence orabsence of 5 mg of tunicamycin per ml, pulse-labeled with [35S]methionine for 3min, and then chased in cold medium for the times indicated before lysis. (A)After elimination of insoluble debris, the supernatant was split in two; half wasimmunoprecipitated with 17D2 (lanes 17), and half was immunoprecipitatedwith the anti-calnexin antibody (Ab) (lanes C). The immunoprecipitates wereanalyzed by SDS-PAGE (10% polyacrylamide) under reducing conditions. (B)After elimination of insoluble debris, G was immunoprecipitated with 17D2 andthe immunoprecipitates were analyzed by SDS-PAGE (10% polyacrylamide)under nonreducing conditions. agg, aggregates; UG, unglycosylated G.

FIG. 8. Effect of castanospermine (CST) on pulse-labeled G association withcalnexin and folding. Infected BSR cells were incubated in the presence orabsence of 1 mM castanospermine, pulse-labeled with [35S]methionine for 3 min,and then chased in cold medium for the times indicated before lysis. (A) Afterelimination of insoluble debris, the supernatant was split in two; half was immu-noprecipitated with 17D2 (lanes 17), and the other half was immunoprecipitatedwith the anti-calnexin antibody (Ab) (lanes C). The immunoprecipitates wereanalyzed by SDS-PAGE (8% polyacrylamide) under reducing conditions. (B)After elimination of insoluble debris, G was immunoprecipitated with 17D2 andthe immunoprecipitates were analyzed by SDS-PAGE (10% polyacrylamide)under nonreducing conditions. agg, aggregates; CNX, calnexin.

FIG. 9. Prolonged association of pulse-labeled G with BiP in the presence of5 mg of tunicamycin (Tun) per ml or 1 mM castanospermine (CST). InfectedBSR cells were incubated in the presence or absence of the drugs, pulse-labeledwith [35S]methionine for 3 min, and then chased in cold medium for the timesindicated before lysis. After elimination of insoluble debris, the supernatant wassplit in two; half was immunoprecipitated with 17D2 (lanes 17), and half wasimmunoprecipitated with the anti-BiP antibody (Ab) (lanes B). The immuno-precipitates were analyzed by SDS-PAGE (8% polyacrylamide) under reducingconditions. UG, unglycosylated G.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

the native molecule on the viral surface (28, 44). However,once G is solubilized with 1% CHAPS under conditions con-serving its structural integrity (14), these MAbs bind native G(15) (see above). This suggests that both epitopes are locatedunderneath the head of G (14) and that steric hindrance due tohigh concentrations of G in the viral membrane inhibits theinteraction of MAbs 17D2 and 29EC2 with native G at the viralsurface.

When MAbs 17D2 and 29EC2 were used, incompletely ox-idized folding intermediates (IT) were detected. Their half-lives were less than 10 min, and they were associated withcalnexin and, to a much lesser extent, with BiP. Both chaper-ones were also associated with the unfolded fully oxidizedform. As for other glycoproteins (20, 26), the association withBiP was maximal immediately after the pulse whereas theassociation with calnexin was maximal after 5 to 10 min ofchase. Inhibition of G binding to calnexin by using tunicamycinor castanospermine resulted in an increased and prolongedassociation with BiP, suggesting a precursor-product relation-ship, such as that already observed in other systems (20, 26).No interaction between G and GRP 94 has been detected. It ispossible that the association of GRP 94 and folding interme-diates is weak and that cross-linking is required to detect suchan interaction, as is the case during the folding of the immu-noglobulin light chain (36).After dissociation from calnexin, G is recognized by MAbs

30AA5 and 18B5, which are directed against antigenic sites IIand III, respectively. At this stage, folding is most probablycomplete. Antigenic sites II and III have been involved inrabies virus pathogenicity: some mutations located in thesesites attenuate or abolish virulence (7, 42, 46, 51). These re-gions have been suggested to interact with the viral receptor(s)(29, 30). Therefore, by analogy to influenza virus HA (55), itmay be speculated that these sites are located in the previouslydescribed head of G (14). The fact that antigenic sites II andIII fold with the same kinetics suggests either that these sitesare very close to each other in the three-dimensional structureof G or that the folding of the head is a global and cooperativephenomenon. Taken together with the fact that epitopes rec-ognized by MAbs 17D2 and 29EC2 (which acquire their nativestructure rapidly) are most probably located underneath thehead of G, this may suggest that the head of G folds after itstail. This would be different from the situation in HA, forwhich it has been shown that folding progresses from the top tothe bottom of the molecule (50), i.e., from the head to the tail.Another result of this study is that G is able to undergo

low-pH-induced structural transitions before complete folding.This suggests that some domains of G are able to act as in thenative structure before undergoing complete folding. We haverecently shown that mutations in position 392 and 396 affectthe low-pH-induced structural transitions and that this regionof the glycoprotein controls the conformational change from Nto I. We have proposed that this region is a part of the tail ofG (12). The results obtained here may suggest that this region

FIG. 10. Folding of G in the presence of castanospermine (CST) and tuni-camycin (Tun). (A) Infected BSR cells were incubated in the presence or ab-sence of the drugs, pulse-labeled with [35S]methionine for 3 min, and then chasedin cold medium for 40 min before lysis (see Materials and Methods). Afterelimination of insoluble debris, the supernatant was split into three; one-thirdeach was immunoprecipitated with 17D2 (lanes 17), 18B5 (lanes 18), and 30AA5(lanes 30). The immunoprecipitates were analyzed by SDS-PAGE (8% poly-acrylamide) under reducing conditions. UG, unglycosylated G; Ab, antibody. (B)The bands corresponding to those shown in Fig. 11A were quantified with aPhosphorImager, and the percentage of G immunoprecipitated with MAbs 18B5and 30AA5 was calculated. The amount of G immunoprecipitated with MAb17D2 was defined as 100%.

FIG. 11. Schematic view of rabies virus G folding. The halflives for epitopes formation, complete oxidation of G, and acquisition of endo H resistance are indicated.The duration of G association with BiP and calnexin is also shown (in this case, the end of the bar indicates the t1/2 for dissociation determined from Fig. 6).



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

acquires its native structure early in the folding pathway. Thisreinforces the idea that the tail of G is folded before its head.The exact role of BiP and calnexin in rabies virus G folding

is unknown. The results shown in Fig. 9 and 10 indicate thatdespite increased interaction with BiP, G cannot fold properlyin tunicamycin-treated cells. Therefore, as for VSV Indiana G(32), interaction with BiP is not sufficient for unglycosylated Gfolding. It also appears that although castanospermine inhibitsmost of G binding to calnexin, folding does occur in the pres-ence of the drug, suggesting that calnexin binding is not nec-essary for G folding. However, in the presence of castanosper-mine, folding is less efficient and is slowed, which is consistentwith a chaperone role for calnexin. Furthermore, G degrada-tion increases in the presence of this drug, most probablybecause binding to calnexin protects G from degradation in theER. This feature has already been observed for influenza virusHA (23) and for free heavy chains of major histocompatibilitycomplex class I (25).As mentioned above, tunicamycin and castanospermine in-

hibit the G-calnexin association. This is consistent with theview that calnexin acts as a lectin and recognizes specificallythe Glc1Man9GlcNAc2 structure on glycoproteins (19, 22, 39,53). However, in the presence of both drugs, residual bindingis still detected. In the presence of tunicamycin, the calnexin-associated glycoproteins are unglycosylated, and in presence ofcastanospermine, they migrate slightly slower than G in un-treated cells. Therefore, in both cases, the residual binding tocalnexin is not due to residual monoglucosylated N-linked oli-gosaccharides. This suggests that calnexin also recognizes thepolypeptide chain of unfolded proteins in an oligosaccharide-independent manner, as already proposed by other groups(26, 31, 43, 52). The nature and location of these polypeptidedomains recognized by calnexin are unknown. Some studiessuggest that the transmembrane domain of glycoproteins isimportant for interaction with calnexin (33). Whether hydro-phobic regions normally buried in the native ectodomain butexposed in the unfolded G are involved remains to be estab-lished. Alternatively, as in castanospermine- and tunicamycin-treated cells, an important proportion of G associated withcalnexin is aggregated (in agreement with the results of Can-non et al. [5] for VSV G); it is conceivable that there is nodirect interaction between G and calnexin when glycosylationand glycan trimming are abrogated.It has been shown that deletion of all the glycosylation sites

completely blocks rabies virus G surface expression (47). Thiscould be due to a low level of interaction with calnexin. How-ever, the fact that folding is much less efficient in the presenceof tunicamycin than in the presence of castanospermine de-spite the same level of interaction with calnexin suggests thatglycosylation not only is needed for G-calnexin interactions butalso probably facilitates G folding by rendering folding inter-mediates more soluble.Additional studies are needed for a better description and

understanding of G folding; the results presented here are onlya first step in this direction. However, it is hoped that byidentifying the steps in the G-folding pathway, we will be ableto define and express independent folding domains at highlevels to perform their structural analysis.

ACKNOWLEDGMENTS

I thank Anne Flamand, Christine Tuffereau, and Rob Ruigrok forhelpful discussions and careful reading of the manuscript.This work is supported by CNRS UPR 9053.

REFERENCES1. Anilionis, A., W. H. Wunner, and P. J. Curtis. 1981. Structure of the glyco-protein gene in rabies virus. Nature (London) 294:275–278.

2. Arita, M., and P. Atanasiu. 1980. Etude comparative du poids moleculaire deplusieurs souches de virus rabique par electrophorese sur gel de polyacryl-amide. Ann. Virol. (Paris) 131E:201–215.

3. Benmansour, A., H. Leblois, P. Coulon, C. Tuffereau, Y. Gaudin, A. Fla-mand, and F. Lafay. 1991. Antigenicity of rabies virus glycoprotein. J. Virol.65:4198–4203.

4. Braakman, I., H. Hoover-Litty, K. R. Wagner, and A. Helenius. 1991. Fold-ing of influenza hemagglutinin in the endoplasmic reticulum. J. Cell Biol.114:401–411.

5. Cannon, K. S., D. N. Hebert, and A. Helenius. 1996. Glycan-dependent and-independent association of vesicular stomatitis virus G protein with cal-nexin. J. Biol. Chem. 271:14280–14284.

6. Chen, W., J. Helenius, I. Braakman, and A. Helenius. 1995. Cotranslationalfolding and calnexin binding during glycoproteins synthesis. Proc. Natl.Acad. Sci. USA 92:6229–6233.

7. Coulon, P., P. E. Rollin, and A. Flamand. 1983. Molecular basis of rabiesvirus virulence. II. Identification of a site on the CVS glycoprotein associatedwith virulence. J. Gen. Virol. 64:693–696.

8. Dietzschold, B. 1977. Oligosaccharides of the glycoprotein of rabies virus.J. Virol. 23:286–293.

9. Dietzschold, B., J. H. Cox, and L. G. Schneider. 1979. Rabies virus strains: acomparison study by polypeptide analysis of vaccine strains with differentpathogenic patterns. Virology 98:63–75.

10. Doms, R. W., R. A. Lamb, J. K. Rose, and A. Helenius. 1993. Folding andassembly of viral membrane proteins. Virology 193:545–562.

11. Dunphy, W. G., and J. E. Rothman. 1985. Compartimental organization ofthe Golgi stack. Cell 42:13–21.

12. Gaudin, Y., H. Raux, A. Flamand, and R. W. H. Ruigrok. 1996. Identificationof amino acids controlling the low-pH-induced conformational change ofrabies virus glycoprotein. J. Virol. 70:7371–7378.

13. Gaudin, Y., R. W. H. Ruigrok, M. Knossow, and A. Flamand. 1993. Low-pHconformational changes of rabies virus glycoprotein and their role in mem-brane fusion. J. Virol. 67:1365–1372.

14. Gaudin, Y., R. W. H. Ruigrok, C. Tuffereau, M. Knossow, and A. Flamand.1992. Rabies virus glycoprotein is a trimer. Virology 187:627–632.

15. Gaudin, Y., C. Tuffereau, P. Durrer, A. Flamand, and R. W. H. Ruigrok.1995. Biological function of the low-pH, fusion-inactive conformation ofrabies virus glycoprotein (G): G is transported in a fusion-inactive state-likeconformation. J. Virol. 69:5528–5534.

16. Gaudin, Y., C. Tuffereau, D. Segretain, M. Knossow, and A. Flamand. 1991.Reversible conformational changes and fusion activity of rabies virus glyco-protein. J. Virol. 65:4853–4859.

17. Gething, M. J., K. McCammon, and J. Sambrook. 1986. Expression ofwild-type and mutant forms of influenza hemagglutinin: the role of folding inintracellular transport. Cell 46:939–950.

18. Gething, M. J., and J. Sambrook. 1992. Protein folding in the cell. Nature(London) 355:33–45.

19. Hammond, C., I. Braakman, and A. Helenius. 1994. Role of N-linked oli-gosaccharide recognition, glucose trimming, and calnexin in glycoproteinfolding and quality control. Proc. Natl. Acad. Sci. USA 91:913–917.

20. Hammond, C., and A. Helenius. 1994. Folding of VSV G protein: sequentialinteraction with BiP and calnexin. Science 266:456–458.

21. Hammond, C., and A. Helenius. 1995. Quality control in the secretory path-way. Curr. Opin. Cell Biol. 7:523–529.

22. Hebert, D. N., B. Foellmer, and A. Helenius. 1995. Glucose trimming andreglucosylation determine glycoprotein association with calnexin in the en-doplasmic reticulum. Cell 81:425–433.

23. Hebert, D. N., B. Foellmer, and A. Helenius. 1996. Calnexin and calreticulinpromote folding, delay oligomerization and suppress degradation of influ-enza hemagglutinin in microsomes. EMBO J. 15:2961–2968.

24. Hurtley, S. M., D. G. Bole, H. Hoover-Litty, A. Helenius, and C. S. Copeland.1989. Interactions of misfolded influenza virus hemagglutinin with bindingprotein (BiP). J. Cell Biol. 108:2117–2126.

25. Jackson, M. R., M. F. Cohen-Doyle, P. A. Peterson, and D. B. Williams. 1994.Regulation of MHC class I transport by the molecular chaperone, calnexin(p88, IP90). Science 263:384–397.

26. Kim, P. S., and P. Arvan. 1995. Calnexin and BiP act as sequential molecularchaperones during thyroglobulin folding in the endoplasmic reticulum. J.Cell Biol. 128:29–38.

27. Kornfeld, R., and S. Kornfeld. 1985. Assembly of asparagine-linked oligo-saccharides. Annu. Rev. Biochem. 54:631–664.

28. Lafay, F., A. Benmansour, K. Chebli, and A. Flamand. 1996. Immunodom-inant epitopes defined by a yeast-expressed library of random fragments ofthe rabies virus glycoprotein map outside major antigenic sites. J. Gen. Virol.77:339–346.

29. Lafay, F., P. Coulon, L. Astic, D. Saucier, D. Riche, A. Holley, and A.Flamand. 1991. spread of the CVS strain of rabies virus and of the avirulentmutant AvO1 along the olfactory pathways of the mouse after intranasalinoculation. Virology 183:320–330.

30. Lentz, T. L., P. T. Wilson, E. Hawrot, and D. W. Speicher. 1984. Amino acidsequence similarity between rabies virus glycoprotein and snake venom cu-raremimetic neurotoxins. Science 226:847–848.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

31. Loo, T. W., and D. M. Clarke. 1994. Prolonged association of temperature-sensitive mutants of human P-glycoprotein with calnexin during biogenesis.J. Biol. Chem. 269:28683–28689.

32. Machamer, C. E., R. W. Doms, D. G. Bole, A. Helenius, and J. K. Rose. 1990.Heavy chain binding protein recognizes incompletely disulfide-bonded formsof vesicular stomatitis virus G protein. J. Biol. Chem. 265:6879–6883.

33. Margolese, L., G. L. Waneck, C. K. Suzuki, E. Degen, R. A. Flavell, and D. B.Williams. 1993. Identification of the region on the class I histocompatibilitymolecule that interacts with the molecular chaperone, p88 (Calnexin, IP90).J. Biol. Chem. 268:17959–17966.

34. Mathieu, M. E., P. R. Grigera, A. Helenius, and R. R. Wagner. 1996. Folding,unfolding, and refolding of the vesicular stomatitis virus glycoprotein. Bio-chemistry 35:4084–4093.

35. Mebatsion, T., M. J. Schnell, and K. K. Conzelmann. 1995. Mokola virusglycoprotein and chimeric proteins can replace rabies virus glycoprotein inthe rescue of infectious defective rabies virus particles. J. Virol. 69:1444–1451.

36. Melnick, J. M., J. L. Dul, and Y. Argon. 1994. Sequential interaction of thechaperones BiP and GRP94 with immunoglobulin chains in the endoplasmicreticulum. Nature (London) 370:373–375.

37. Mifune, K., M. Ohuchi, and K. Mannen. 1982. Hemolysis and cell fusion byrhabdoviruses. FEBS Lett. 137:293–297.

38. Munro, S., and H. R. B. Pelham. 1986. An HSP70-like protein in the ER:identity with the 78 kd glucose-regulated protein and immunoglobulin heavychain binding protein. Cell 46:291–300.

39. Ora, A., and A. Helenius. 1995. Calnexin fails to associate with substrateproteins in glucosidase-deficient cell lines. J. Biol. Chem. 270:26060–26062.

40. Otteken, A., P. L. Earl, and B. Moss. 1996. Folding, assembly, and intracel-lular trafficking of the human immunodeficiency virus type 1 envelope gly-coprotein analyzed with monoclonal antibodies recognizing maturationalintermediates. J. Virol. 70:3407–3415.

41. Ou, W. J., P. H. Cameron, D. Y. Thomas, and J. J. M. Bergeron. 1993.Association of folding intermediates of glycoproteins with calnexin duringprotein maturation. Nature (London) 364:771–776.

42. Prehaud, C., P. Coulon, F. Lafay, C. Thiers, and A. Flamand. 1988. Anti-genic site II of the rabies virus glycoprotein: structure and role in viralvirulence. J. Virol. 62:1–7.

43. Rajagopalan, S., Y. Xu, and M. B. Brenner. 1994. Retention of unassembledcomponents of integral membrane proteins by calnexin. Science 263:387–390.

44. Raux, H., P. Coulon, F. Lafay, and A. Flamand. 1995. Monoclonal antibodieswhich recognize the acidic configuration of the rabies glycoprotein at thesurface of the virion can be neutralizing. Virology 210:400–408.

45. Raux, H., Y. Gaudin, and A. Flamand. Unpublished results.46. Seif, I., P. Coulon, P. E. Rollin, and A. Flamand. 1985. Rabies virulence:

effect on pathogenicity and sequence characterization of rabies virus muta-tions affecting antigenic site III of the glycoprotein. J. Virol. 51:505–514.

47. Shakin-Eshleman, S. H., A. T. Remaley, J. R. Eshleman, W. H. Wunner, andS. L. Spitalnik. 1992. N-linked glycosylation of rabies virus glycoprotein.Individual sequons differ in their glycosylation efficiencies and influence oncell surface expression. J. Biol. Chem. 267:10690–10698.

48. Sousa, M., and A. J. Parodi. 1995. The molecular basis for the recognition ofmisfolded glycoproteins by the UDP-Glc:glycoprotein glucosyltransferase.EMBO J. 14:4196–4203.

49. Sousa, M. C., M. A. Ferrero-Garcia, and A. J. Parodi. 1992. Recognition ofthe oligosaccharide and protein moieties of glycoproteins by the UDP-Glc:glycoprotein glucosyltransferase. Biochemistry 31:97–105.

50. Tatu, U., C. Hammond, and A. Helenius. 1995. Folding and oligomerizationof influenza hemagglutinin in the ER and the intermediate compartment.EMBO J. 14:1340–1348.

51. Tuffereau, C., H. Leblois, J. Benejean, P. Coulon, F. Lafay, and A. Flamand.1989. Arginine or lysine in position 333 of ERA and CVS glycoprotein isnecessary for rabies virulence in adult mice. Virology 172:206–212.

52. Van Leeuwen, J. E. M., and K. P. Kearse. 1996. Calnexin associates exclu-sively with individual CD3d and T cell antigen receptor (TCR) a proteinscontaining incompletely trimmed glycans that are not assembled into mul-tisubunit TCR complexes. J. Biol. Chem. 271:9660–9665.

53. Ware, F. E., A. Vassilakos, P. A. Peterson, M. R. Jackson, M. A. Lehrman,and D. B. Williams. 1995. The molecular chaperone calnexin bindsGlc1Man9GlcNAc2 oligosaccharide as an initial step in recognizing un-folded glycoproteins. J. Biol. Chem. 270:4697–4704.

54. Whitt, M. A., L. Buonocore, C. Prehaud, and J. K. Rose. 1991. Membranefusion activity, oligomerization, and assembly of the rabies virus glycopro-tein. Virology 185:681–688.

55. Wilson, I. A., J. J. Skehel, and D. C. Wiley. 1981. Structure of the hemag-glutinin membrane glycoprotein of influenza virus at 3 Å resolution. Nature(London) 289:368–373.

56. Wunner, W. H., B. Dietzschold, C. L. Smith, M. Lafon, and E. Golub. 1985.Antigenic variants of CVS rabies virus with altered glycosylation sites. Vi-rology 140:1–12.

57. Yelverton, E., S. Norton, J. F. Obijeski, and D. V. Goeddel. 1983. Rabiesvirus glycoprotein analogs: biosynthesis in E. coli. Science 219:614–620.

58. Yewdell, J. W., A. Yellin, and T. Bachi. 1988. Monoclonal antibodies localizeevents in the folding, assembly, and intracellular transport of the influenzavirus hemagglutinin glycoprotein. Cell 52:843–852.



http://jvi.asm.org/

Dow

nloaded from

http://jvi.asm.org/

view - journal of virology - american society for microbiology

Documents