The use of calnexin and calreticulin by cellular and viral glycoproteins.

Calnexin and calreticulin are homologous lectin chaperones that assist maturation of cellular and viral glycoproteins in the mammalian endoplasmic reticulum. Calnexin and calreticulin share the same specificity for monoglucosylated protein-bound N-glycans but associate with a distinct set of newly synthesized polypeptides. We report here that most calnexin substrates do not associate with calreticulin even upon selective calnexin inactivation, while BiP associates more abundantly with nascent polypeptides under these conditions. Calreticulin associated more abundantly with orphan calnexin substrates only in infected cells and preferentially with polypeptides of viral origin, showing stronger dependence of model viral glycoproteins on endoplasmic reticulum lectins. This may explain why inactivation of the calnexin cycle affects viral replication and infectivity but not viability of mammalian cells.

Calnexin and calreticulin are homologous lectin chaperones that assist maturation of cellular and viral glycoproteins in the mammalian endoplasmic reticulum. Calnexin and calreticulin share the same specificity for monoglucosylated protein-bound N-glycans but associate with a distinct set of newly synthesized polypeptides. We report here that most calnexin substrates do not associate with calreticulin even upon selective calnexin inactivation, while BiP associates more abundantly with nascent polypeptides under these condi-

tions. Calreticulin associated more abundantly with orphan calnexin substrates only in infected cells and preferentially with polypeptides of viral origin, showing stronger dependence of model viral glycoproteins on endoplasmic reticulum lectins. This may explain why inactivation of the calnexin cycle affects viral replication and infectivity but not viability of mammalian cells.
About 20% of mammalian proteins are predicted to be secretory proteins (1). Most of them are N-glycosylated during transit through the endoplasmic reticulum (ER) 1 translocon. The preassembled glycan added to nascent chains is composed of 2 N-acetyl glucosamines, 9 mannoses, and 3 terminal glucose residues. Two glucose residues are rapidly trimmed by ERresident glucosidases, and the generated monoglucosylated Nglycan directs association of the nascent polypeptide with the lectin chaperones calnexin (Cnx) and calreticulin (Crt) and with the glycoprotein-specific oxidoreductase ERp57 (2)(3)(4)(5)(6). For most glycoproteins, transient association with Cnx, Crt, and ERp57 is not essential, but contributes to the efficiency of the maturation process. Although Cnx and Crt share the same ligand specificity, they bind a distinct set of polypeptides, possibly because the first is membrane-bound whereas the latter is soluble in the ER lumen. It has been shown in fact that Cnx binds to membrane-proximal glycans whereas Crt binds to glycans that emerge deeper into the lumen (7)(8)(9)(10)(11)(12)(13)(14)(15).
Recently, Cnx-knockout mice were established (16). An unexpected recombination event resulted in the generation of a second mouse strain expressing a shorter version of Cnx (sCnx). Cnx-deficient mice and mice expressing sCnx were phenotypically identical. Directly after birth, their size, respiration, and feeding were normal. At 8 -10 days after birth, they were one-third lighter than their littermates, and they progressively developed severe motor abnormalities. Most of the mice either died or had to be sacrificed within 4 weeks (16), showing the incapacity of Crt to fully replace Cnx activity. Because of severe breeding difficulties of the original Cnx-deficient strain, only the strain expressing sCnx was continued (16). Here we made use of embryonic fibroblasts derived from these mice to better characterize the function of Cnx and Crt during protein folding.
We first established that sCnx originated from an alternative splicing of the Cnx gene that removed several functional residues, thus selectively depleting cells of Cnx activity. No evidence of compensatory unfolded protein response (UPR) was found, indicating the lack of substantial protein-folding defects at the cellular level upon eradication of Cnx activity. Despite the fact that Crt and Cnx share the same specificity for monoglucosylated N-glycans, Crt did not normally associate with orphan Cnx substrates upon selective depletion of Cnx activity. Rather, enhanced association of nascent chains with BiP was observed.
Three cellular Cnx substrates, the folding-competent BACE501 and the folding-defective BACE457 and BACE457⌬ were expressed in cells selectively depleted of Cnx activity to determine whether the membrane-bound versus soluble status, the number of protein-bound N-glycans, or the substrate folding competence affected the capacity of Crt to associate with orphan Cnx substrates. None of these specific Cnx substrates associated with Crt upon eradication of Cnx activity. Crt did associate with orphan Cnx substrates only in virus-infected cells and, preferentially, with orphan Cnx substrates of viral origin. These results showed an intriguing diversity among cellular protein and viral proteins expressed in infected cells in the use of and in the dependence on the Cnx/Crt chaperone system.
For infections, cells were plated in 6-cm Petri dishes to 70 -80% confluence and incubated for 30 -60 min at room temperature with the selected virus diluted to 30 plaque-forming units per cell in RPMI/0.2% bovine serum albumin, pH 6.8. The inoculum was replaced with cell culture medium, and incubation proceeded for 4 h. Cells were then metabolically labeled. For transfections, cells plated at 90% confluence were transfected with BACE expression plasmids using Lipofectamine * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Both authors contributed equally to this work. S]Met/Cys in 1 ml of starvation medium/dish and chased for the times indicated in the figures with Dulbecco's modified Eagle's medium supplemented with 5 mM cold Met/Cys. Postnuclear supernatants were prepared by solubilization of cells in 800 l/dish of ice-cold 2% CHAPS in Hepes buffer saline, pH 6.8, containing 20 mM ice-cold N-ethylmaleimide, protease inhibitors (HBS), and 10 units of apyrase/dish (Sigma) to rapidly deplete cellular ATP for anti-BiP immunoprecipitations. Cell extracts were prepared by 10 min of centrifugation at 10,000 ϫ g and analyzed by reducing SDS-PAGE.
Immunoprecipitations were performed by adding protein A beads (Sigma, 1:10 w/v swollen in HBS) and the selected antibody to the cell extracts. Incubations were for 1-4 h in a cold room.
The immunoprecipitates were washed three times with HBS/0.5% CHAPS and resuspended in sample buffer for SDS-PAGE. Gels were exposed to BioMax (Kodak) films and scanned with an Agfa scanner. The influenza virus strain X-31 was a kind gift from J. Brunner. Antibodies were kind gifts from A. Helenius (calnexin and HA), I. Braakman (PDI), T. Wileman (ERp57), J. C. Perriard (VSV G protein), W. Nauseef (calreticulin), and P. Paganetti (BACE). Antibody to BiP is commercially available (StressGen). The plasmids for BACE and for VSV G protein expression were kind gifts from P. Paganetti and K. Ueberla, respectively.
For EndoH treatment, the selected glycoprotein was immunoisolated from the cell extracts with specific antibody and denatured before incubation for 1 h at 37°C with 1 milliunit of EndoH (Roche Applied Science) in HBS-0.5% CHAPS.
Immunoblotting-Cells were washed once in ice-cold phosphate-buffered saline and lysed in 2% CHAPS. Equal amounts of proteins were separated by reducing SDS-PAGE and transferred to polyvinylidene difluoride membranes. Membranes were decorated with antibodies to selected ER chaperones and stained with horseradish peroxidase-conjugated anti-rabbit IgG. Immune complexes were visualized using enhanced chemiluminescence (LAS system).
Indirect Immunofluorescence Microscopy-For indirect immunofluorescence, MEF were plated 24 h before the experiment on alcian bluetreated glass cover slips, washed twice with phosphate-buffered saline, and fixed at room temperature for 20 min in serum-free medium (SFM)-Hepes containing 3.7% formaldehyde. Cells were washed twice in SFM-Hepes and twice with phosphate-buffered saline and permeabilized with 0.05% saponin (PS) for 15 min. Cells were incubated with primary antibody diluted in PS for 45 min, washed 15 min with PS, and then incubated with conjugated secondary antibody diluted in PS for 30 min. Cells were rinsed with PS, then with water, and mounted in Mowiol.
Reverse Transcription-Polymerase Chain Reaction (RT-PCR)-1.2 ϫ 10 6 wt or S-CNX MEF were plated in 6-cm dishes. After 12 h, cells were lysed in TRIzol reagent (Invitrogen) and RNA isolated according to the instructions of the manufacturer. Two micrograms of RNA were used for cDNA production by SuperScriptII reverse transcriptase (Invitrogen) primed by oligo(dT) (Invitrogen). RT-PCR was performed using TaqDNA polymerase (Invitrogen) with Cnx-specific primers: Forw (atggaagggaagtggttactgtgt) and Rev (cccggacat gtgagtcggaggcttccatttgccctt). Gene sequencing was performed on an ABI Prism 310 instrument (Applied Biosystems).

RESULTS
Characterization of sCnx-We first assessed intracellular expression and localization of sCnx by indirect immunofluorescence of semipermeabilized cells. sCnx conserved the N-terminal ER-targeting signal sequence and the C-terminal ER retention motif of full-length Cnx, because it co-localized with the ER marker BiP (Fig. 1A). Replacement of Cnx with sCnx did not affect cell viability and did not trigger a compensatory up-regulation of other ER-resident proteins such as BiP, PDI, ERp57, or Crt ( Fig. 1B) showing that MEF expressing sCnx (S-CNX MEF) were not under debilitating ER stress.
Analysis of sCnx transcripts (see "Experimental Procedures") revealed a deletion of 417 base pairs (Fig. 1C), from base 307 to 725, because of excision of the Cnx exons four, five, and six (Fig. 1D). The gene product was characterized by the substitution G103V, and by the deletion of 139 residues (Lys 104 -Leu 242 ) ( Fig. 1E) resulting in faster electrophoretic mobility (Fig. 1B). Among the residues deleted are those involved in glucose binding that mediates association of Cnx with substrates (Tyr 165 , Lys 167 , Tyr 186 , Met 189 , and Asp 217 ) (18), two cysteines involved in a surface-exposed disulfide bond required for substrate binding (Cys 161 and Cys 195 ), and the conserved Asp 118 involved in the coordination of a calcium ion. Thus, deletion of the Cnx glycan binding site and of other residues possibly involved in maintenance of the native Cnx architecture did not affect cell viability, ER morphology, and did not trigger a UPR in cultured cells (Fig. 1). However, it did cause the same phenotypic alteration at the organism level as deletion of Cnx (16).
sCnx Does Not Associate with Newly Synthesized Glycoproteins-We next determined whether sCnx binds to newly synthesized proteins in the ER. wt and S-CNX cells were pulsed for 10 min with [ 35 S]methionine and [ 35 S]cysteine and chased for 0 -60 min with unlabeled amino acids. Detergent extracts were immunoprecipitated with antibody raised to the cytosolic tail of Cnx and of sCnx ( Fig. 2A).
Several labeled cellular proteins coprecipitated with Cnx. Most of these proteins were chased away in 60 min, indicating that they were Cnx substrates that associated transiently with the ER lectin ( Fig. 2A, wt). Deletion of the 139 residues upon alternative splicing of the Cnx gene abolished association of labeled cellular polypeptides with sCnx ( Fig. 2A, S-CNX).
Calreticulin Does Not Normally Associate with Orphan Calnexin Substrates-Cnx and Crt associate with a wide array of substrates because they bind monoglucosylated N-glycans, which are transiently exposed by all glycosylated polypeptides expressed in the ER (19 -24). Unexpectedly, however, they were found to associate with a distinct set of polypeptides in living cells (7)(8)(9)(10)(11)(12)(13)(14). To verify if Cnx inactivation resulted in changes in Crt specificity, acquiring novel substrates, the de-

FIG. 1. Expression of ER-resident chaperones in wt and in S-CNX cells. A, indirect immunofluorescence with anti-BiP (left) and
anti-Cnx (right) antibody in permeabilized wt (upper panels) and S-CNX MEF (lower panels). B, detergent extracts of wt and S-CNX cells (5 g each) were separated on SDS-PAGE gels. Proteins were blotted on polyvinylidene difluoride membranes and probed with specific antibodies. The cytosolic protein p65 is shown as a loading control. C, Cnx gene was amplified from a Cnx-containing expression plasmid (pCNX) or from cDNA produced by RT-PCR of total transcripts prepared from wt and from S-CNX MEF. The shorter PCR product was purified from the agarose gel and sequenced. D, deletion in the Cnx gene is shown (base numbering). E, deletion in the Cnx protein (amino acid numbering). Shown are the residues important for the lectin function of Cnx. tergent extracts of wt and of S-CNX cells were immunoprecipitated with Crt-specific antibodies (Fig. 2B). The arrowhead (also shown in C) shows the only Cnx substrate that clearly associates more abundantly with Crt upon deletion of Cnx activity.
Thus, analysis of the immunoprecipitates did not reveal significant differences in Crt-associated polypeptides in wt versus S-CNX cells. More specifically, there was no indication that Crt acquired new substrates upon selective depletion of Cnx activity (Fig. 2B).
To better analyze this, wt and S-CNX MEF were pulselabeled, chased for 3 min, detergent-solubilized, and Cnx and Crt immunoprecipitates were analyzed in the same gel (Fig.  2C). In wt cells, the pattern of cellular proteins associating with Cnx (Fig. 2C, lane 1) and Crt (lane 2) was only partially overlapping. Cnx associates with newly synthesized proteins of all sizes, whereas Crt associates preferentially with larger polypeptides. The black lines in Fig. 2C show a selection of 14 labeled polypeptides of molecular masses between 30 and 100 kDa that associate with Cnx but not with Crt in wt cells. None of these Cnx substrates were found to associate with Crt upon Cnx inactivation (lane 3). Thus, most cellular Cnx substrates remain inaccessible to Crt upon selective eradication of Cnx activity.
Lanes 4 -6 in Fig. 2C are controls showing that deletion of Cnx activity prevents substrate association (lane 4) and that cell incubation with the glucosidase inhibitor castanospermine, which prevents trimming of N-glycans to the monoglucosylated form, substantially inhibits association of newly synthesized polypeptides with Cnx and Crt (lanes 5 and 6, respectively).
Association of newly synthesized polypeptides with Crt was unaffected by Cnx inactivation. However, this was not true for substrate association with the other abundant ER chaperone BiP/GRP78. This was significantly increased, especially at early chase times (0 and 5 min of chase, Fig. 2D), indicating increased association of BiP with nascent chains upon Cnx inactivation. Overall, substrate association with BiP and with Crt remained transient in S-CNX cells. Together with the findings reported above (depletion of Cnx did not trigger UPR and viability of S-CNX cells was normal), this was a further indication that (most) newly synthesized cellular proteins re-mained folding-competent upon selective deletion of Cnx activity.
Association with ER Chaperones and Maturation of ␤-Secretase in wt Cells and in Cells Expressing sCnx-Unbiased analysis of labeled cellular proteins (Fig. 2) did not show the dramatic increase of Crt-associated radioactivity that would be expected if Crt replaces Cnx and would associate with orphan Cnx substrates upon Cnx inactivation. We next specifically tested the competence of Crt to associate with selected cellular Cnx substrates and analyzed whether membrane-bound versus soluble status or extent of N-glycosylation made a difference. To this end, three variants of human BACE were individually transfected in the two cell lines. BACE457⌬ is a soluble, diglycosylated protein, and BACE457 and BACE501 are type I membrane proteins with 2 and 4 N-glycans, respectively. BACE457⌬ and BACE457 are folding-incompetent and are eventually degraded from the ER (25)(26)(27), whereas BACE501 folding is productive, and the native glycoprotein is transported along the secretory pathway to the plasma membrane (28,29).
As is shown for endogenous glycoproteins (Fig. 2), sCnx did not associate with the recombinantly expressed BACE variants (Fig. 3, A-C, lane 4) and Crt did not associate with the BACE variants even after deletion of Cnx activity (Fig. 3, A-C, compare lane 5 with 6). Instead, normalization for the slightly lower expression level of the recombinant proteins in S-CNX MEF revealed a moderate increase of BiP binding.
Altogether, the data shown in Figs. 2 and 3 confirmed the clear distinction between Cnx and Crt substrates. Furthermore, the data show that orphan Cnx substrates remain inaccessible to Crt even upon eradication of Cnx activity, no matter whether the substrate protein is membrane-bound (BACE457 and BACE501), soluble (BACE457⌬), weakly glycosylated (BACE457 and BACE457⌬) or more heavily glycosylated (BACE501), folding-competent (BACE501), or folding-incompetent (BACE457 and BACE457⌬).

FIG. 2. Cellular substrates of Cnx, Crt, and BiP. A, Cnx and sCnx (arrows)
were pulled-down with a Cnx-specific antibody from detergent extracts of metabolically labeled wt and S-CNX MEF chased for 0 -60 min. No Cnx substrate does coprecipitate with sCnx, confirming that sCnx is not functional as a lectin chaperone. B, same as described in A, for Crt. C, samples immunoprecipitated with Cnx (X) or Crt (R) antibody have been loaded side-by-side to compare co-precipitatinglabeled cellular proteins. Chase time is 3 min. Asterisks show Cnx in lanes 1 and 5,  Crt in lanes 2, 3, and 6, and sCnx in lane 4. D, same as described in A, for BiP. Comparison of labeled polypeptides coprecipitating with BiP after a 0-and a 60-min chase allows identification of cross-reacting, unspecific bands (ϩ). In all gels, ϩ shows labeled bands that crossreact with protein A beads.
We next determined whether selective depletion of Cnx activity affected maturation of BACE501. Native BACE501 is released from the ER and is subject to complex glycosylation in the Golgi that decreases electrophoretic mobility and renders BACE501 N-glycans EndoH-resistant (28,29). Both the molecular mass increase and the acquisition of EndoH resistance, as a further estimate of the rate of BACE501 export from the ER, occurred with similar efficiency and kinetics in wt and S-CNX MEF (Fig. 3D).
SFV Glycoproteins Associate More Abundantly with Calreticulin in S-CNX Cells-Analysis of cellular polypeptides showed that the substrate specificity of Crt did not change upon Cnx depletion. We next analyzed the fate of four viral envelope glycoproteins that served as models in the past years to elucidate functioning of the ER folding machinery: the G protein of vesicular stomatitis virus (VSV, Ref. Cells were infected with SFV 4 h before pulse-chase analysis (see "Experimental Procedures"). Folding of SFV glycoproteins E1 and p62 involves transient interaction with BiP, Cnx, Crt, PDI, and with the oxidoreductase ERp57 (33). Native E1/p62 heterodimers leave the ER and transit through the Golgi complex to the plasma membrane. Infection of wt and S-CNX MEF with SFV shuts down synthesis of cellular proteins. During the pulse, therefore, radioactivity is only incorporated in viral polypeptides, and analysis of cell lysates in nonreducing SDSpolyacrylamide gels only shows the viral glycoproteins E1 and p62 expressed in the ER, E2 generated by a convertase-mediated cleavage of p62 in the Golgi, and the viral capsid (C) expressed in the host cell cytosol (Fig. 4A). Conversion of p62 in E2 determines the rate of SFV glycoprotein maturation and was slightly accelerated in S-CNX cells (Fig. 4A). In wt MEF SFV glycoproteins transiently associated with Cnx (Fig. 4B) and Crt (Fig. 4C). The ERp57-containing mixed disulfides, formed as transient intermediates of the oxidative folding process (33), were equally distributed in the Cnx-and Crt-bound fractions, because ERp57 forms functional complexes with both ER lectins (wt in Fig. 4, B and C, respectively).
In S-CNX cells, viral glycoproteins and mixed disulfides did not co-precipitate with sCnx and associated more abundantly, but still transiently, with Crt (S-CNX, Fig. 4, B and C, respectively). After a 72-min chase, in wt cells, a significant amount of viral proteins was still associated with Cnx and Crt; upon depletion of Cnx activity, release from Crt was slightly faster. Thus, maturation of SFV glycoproteins progressed normally without Cnx. Unlike the cellular proteins previously investigated, SFV glycoproteins did associate more abundantly with Crt in the absence of functional Cnx.

Association with ER Chaperones and Maturation of VSV G Protein in wt Cells and in Cells Expressing Non-functional
Calnexin-We next analyzed the behavior of a glycoprotein considered a classic Cnx substrate with undetectable association with Crt in wt cells, the VSV G protein (9,30).
wt and S-CNX MEF were infected with VSV 4 h before pulse-chase analysis. Oxidative folding of the VSV G protein is sequentially assisted by BiP and Cnx (30). Acquirement of the native structure is monitored by following the formation of the native B2 epitope with the specific monoclonal antibody I14 (34). Here we made use of I14 to compare the kinetics of G protein maturation in the two cell lines by pulling down from detergent extracts prepared after several chase times the fraction of native G protein.
The amount of I14 reactive-labeled G protein increased at the same rate in both cell lines to about 20 min of chase, and then weakly decreased, again with similar kinetics after the 40 min chase, upon release from cells of viral particles containing labeled G protein (Fig. 5A and data not shown). Consistently, there was no difference, if any a slight acceleration in S-CNX cells, in the rate at which the G protein acquired EndoHresistant glycans (Fig. 5B). Thus, as for the proteins tested above, maturation of VSV G protein was not affected by Cnx depletion, even though in wt cells, Cnx is the lectin chaperone of choice for the G protein. We next determined whether in the case of the G protein Crt did intervene during folding in S-CNX MEF.
Calreticulin Associates with VSV G Protein in S-CNX Cells-To identify the chaperones assisting maturation of the  E1, E2, and p62. B, viral proteins associated with Cnx were pulled-down with specific antibody. Note that E2, generated in the Golgi, and C, a cytosolic protein, do not co-precipitate with Cnx. Viral glycoprotein:ERp57-containing mixed disulfides are shown (33). C, same as described in B, for Crt. G protein, co-precipitations with BiP, Cnx, and Crt were analyzed with reducing SDS-PAGE.
In wt cells, the G protein abundantly associated with BiP and with Cnx. As previously shown in other cell lines (30), BiP association (3-min chase, Fig. 5, C and F)) preceded the association of G protein with Cnx (10-min chase, Fig. 5, D and F). Labeled G protein did not co-precipitate with Crt (Fig. 5E), confirming preferential association of this glycoprotein with Cnx (9). sCNX did not associate with G protein (Fig. 5D). Lack of Cnx caused a clear change in the kinetics of G protein association with BiP (Fig. 5, C and F) and resulted in a significant transient association of G with Crt (Fig. 5, E and F). The release of G protein from Crt in S-CNX cells (t1 ⁄2 ϭ 11 min) was faster compared with G protein release from Cnx in wt cells (t1 ⁄2 ϭ 18 min, Fig. 5F). This could possibly account for the slightly faster acquirement of EndoH-resistant glycans (Fig. 5B) and faster release of viral particles from S-CNX cells (data not shown). Thus, similar to SFV glycoproteins and unlike glycoproteins of cellular origin, G protein associated with Crt when Cnx was not available. Importantly, this was a peculiarity of the G protein expressed in infected cells. In fact if the G protein was transiently expressed upon cell transfection with a VSV G protein expression plasmid, the G protein behaved like cellular glycoproteins that associated more abundantly with BiP, but not with Crt upon Cnx inactivation (Fig. 5G).
Calreticulin Retains Folding-incompetent HA in the ER-In wt MEF, maturation of HA is characterized by the stepwise conversion of the newly synthesized glycoprotein from intermediate oxidation states (IT, Ref. 35) into fully oxidized, native form (NT) that can be followed in nonreducing gels (Fig. 6A). During oxidative folding in the ER, HA transiently associates with Cnx and with Crt (Fig. 6C).
Influenza HA is strongly dependent on Cnx for maturation (14). Consistently, analysis of HA folding in S-CNX MEF showed that about 50% of the labeled HA progressively entered in disulfide-bonded (DBA in the nonreducing gel, Fig. 6B) and in non-covalent aggregates; the latter persisting upon sample reduction (Fig. 6B). In the absence of Cnx, misfolded disulfide-bonded HA did persistently associate with Crt (Fig. 6D). Thus, folding-competent and folding-incompetent cellular Cnx substrates did not associate with Crt upon depletion of Cnx. All model viral glycoproteins tested in this study did so, in infected cells, independently of their folding competence. DISCUSSION We characterized and made use of embryonic fibroblasts derived from mice expressing a shorter, inactive Cnx variant to better characterize chaperone involvement during protein maturation in the mammalian ER. Cnx deletion did not result in significant protein misfolding, shown by the lack of a compensatory UPR in S-CNX MEF and by analysis of several cellular and viral model proteins.
Unbiased analysis of labeled cellular proteins showed that despite equal specificity for monoglucosylated N-glycans exposed by nascent polypeptides, Crt and Cnx assist a different subset of proteins during maturation. Importantly, Crt remained inaccessible for most Cnx substrates even upon selective inactivation of Cnx. Rather, BiP associated more abundantly with nascent chains in the absence of functional Cnx.
Only upon viral infection, Crt associated with orphan Cnx substrates of viral origin. A saturation of the Crt binding capacity that precludes association of orphan Cnx substrates in non-infected cells is one possible explanation of our data. However, even though VSV and SFV infections shutdown host protein biosynthesis, thereby setting Crt free, influenza virus did not. Rather it similarly activates the "surrogate" activity of Crt.
Significantly, Crt failed to associate with orphan Cnx substrates of viral origin when these were expressed in transfected rather than in infected cells. Thus, cell infection (at least with the three viruses used in this study) activates the surrogate function of Crt, consistent with a stronger dependence of viral glycoproteins from Cnx and Crt. This is actually an Achilles heel for viruses that has been exploited for designing antiviral therapies. In fact, the Cnx/Crt cycle is required for replication and infectivity of many human pathogens (36 -38), but not for viability of cultured mammalian cells. Iminosugar derivatives FIG. 5. Maturation of VSV G protein in wt and in S-CNX cells. A, native form of labeled G protein is specifically pulleddown from detergent extract with the specific monoclonal antibody I14. B, release of the G protein in the Golgi has been determined by measuring acquirement of EndoH-resistant glycans. C, analysis of G protein associated with BiP. D, same as described in C, for Cnx. E, same as described in B, for Crt. F, kinetics of G protein release from Cnx and BiP in wt cells and from Crt and BiP in S-CNX cells. G, wt (W) and S-CNX cells (S) were transiently transfected for expression of the VSV G protein (lanes 1 and 8). Anti-Cnx (lanes 2 and 3), Crt (lanes 4 and 5), and BiP (lanes 6 and 7)  have proved useful in reducing the viral level in an animal model of chronic hepatitis B infection (39), of bovine diarrhea virus (40), and mortality in the case of Japanese encephalitis virus infection (41).
Is Cnx a dedicated chaperone? Cnx associates transiently with numerous glycoproteins. If Cnx assistance would be essential for their folding, dramatic consequences should emerge upon Cnx eradication or inactivation. This is obviously not the case, at least for cultured cells and during embryogenesis, and hints at a role for Cnx as an obligate chaperone only for a restricted subset of mammalian polypeptides. Several ER chaperones are dedicated to specific substrates (a comprehensive list is presented in Ref. 2). Calmegin, the testis-specific homologue of Cnx involved in spermatogenesis is an important example for this discussion. Similar to Cnx, calmegin also associates transiently with a large number of nascent, monoglucosylated glycoproteins in the sperm, but its depletion only prevents maturation of one of them, namely fertilin-␤ (42,43). Consistent with the lack of effects on other glycoproteins; maturation, sperm number, viability, or motility are unaffected in calmegin-deficient mice, whereas the fertilin-␤-regulated sperm's capacity to bind to the egg zona pellucida is lost; thus, leading to male sterility (42,44). Analogies with Cnx are striking, because Cnx is dispensable for maturation of most glycoproteins. In mice, clear signs of progressive pathology, eventually causing early postnatal death and motor disorders in the survivors (16), show however that Cnx may be required for a very restricted set of proteins, possibly expressed in tissuespecific fashion. Characterization of the physiologic Cnx substrate(s) is the next challenge.
How do we explain that deletion of Cnx, an abundant ER chaperone transiently associating with a large number of newly synthesized polypeptides, does not result in substantial protein misfolding with activation of the UPR? In contrast with cytosolic proteins, polypeptides expressed in the ER benefit from being singularly injected through individual translocons that offer a fair spacing between aggregation-prone, unstructured nascent chains, thus per se offering protection from aggregation (45). Moreover, co-translational addition of N-glycans increases the hydrophilicity of the unstructured chains and association of BiP/GRP78 with exposed hydrophobic patches (46,47), further preventing aggregation, and may ex-pose orphan Cnx substrates in replacing folding factors, as shown in this study by the increased association of nascent chains with this abundant luminal chaperone in S-CNX cells.
A final remark is that upon analysis of maturation, surface transport, and activity of several viral and cellular proteins (14, 48 -53, and this work), influenza virus HA remains the only one substantially suffering lack of CNX assistance. Thus, HA could represent a model for those yet-to-be-identified cellular proteins requiring Cnx for maturation. What is so special in HA? Influenza HA has a peculiar head-to-tail structure, with the N-terminal Cys 14 covalently paired with the C-terminal Cys 466 (54). With an average synthesis rate of 4 residues per second, oxidation of Cys 14 has to be delayed by about 100 s to assure efficient maturation of the newly synthesized protein. For this specific protein, the fraction of Cnx strategically positioned at the luminal exit of the translocon complex (55,56) and associating co-translationally with nascent glycopolypeptides (32,57,58) clearly plays an essential role in determining productive folding. Protection of cysteines from premature oxidation to facilitate formation of native intramolecular subdomains or intermolecular functional complexes may be one of the physiologic tasks for which Cnx is dedicated.

FIG. 6. Maturation of influenza virus HA in wt and in S-CNX cells. A,
influenza virus HA has been immunoisolated from detergent extracts of metabolically labeled and chased wt MEF with specific antibody; the antibody also pulls down the viral nuclear protein (NP). B, same as described in A in S-CNX MEF. DBA are HA-containing aggregates. C, detergent extracts of wt MEF were immunoprecipitated with antibody to Cnx (left) or Crt (right), and then subjected to a second immunoprecipitation with antibody to HA. In the anti-Cnx panel, the last lane shows the lack of HA interaction with sCnx. D, detergent extracts of S-CNX cells were subjected to a double immunoprecipitation, first with anti-Crt, then with anti-HA antibody.