Rabies virus glycoprotein is important in the biology and pathogenesis of neurotropic rabies virus infection. This transmembrane glycoprotein is the only viral protein on the surface of virus particles, is the viral attachment protein that facilitates virus uptake by the infected cell, and is the target of the host humoral immune response to infection. The extracellular domain of this glycoprotein has N-glycosylation sequons at Asn37, Asn247, and Asn319. Appropriate glycosylation of these sequons is important in the expression of the glycoprotein. Soluble forms of rabies virus glycoprotein were constructed by insertion of a stop codon just external to the transmembrane domain. Using site-directed mutagenesis and expression in transfected eukaryotic cells, it was possible to compare the effects of site-specific glycosylation on the cell-surface expression and secretion of transmembrane and soluble forms, respectively, of the same glycoprotein. These studies yielded the surprising finding that although any of the three sequons permitted cell surface expression of full-length rabies virus glycoprotein, only the N-glycan at Asn319 permitted secretion of soluble rabies virus glycoprotein. Despite its biological and medical importance, it has not yet been possible to determine the crystal structure of the full-length transmembrane form of rabies virus glycoprotein which contains heterogeneous oligosaccharides. The current studies demonstrate that a soluble form of rabies virus glycoprotein containing only one sequon at Asn319 is efficiently secreted in the presence of the N-glycan processing inhibitor 1-deoxymannojirimycin. Thus, it is possible to purify a conformationally relevant form of rabies virus glycoprotein that contains only one N-glycan with a substantial reduction in its microheterogeneity. This form of the glycoprotein may be particularly useful for future studies aimed at elucidating the three-dimensional structure of this important glycoprotein.
Key words: glycoproteins/N-linked glycosylation/rabies virus/sequon
Rabies virus is a medically important neurotropic virus (for review, see Fishbein and Robinson, 1993). It has a negative-strand RNA genome that encodes five proteins, including one glycoprotein. Rabies virus glycoprotein (RGP) of the ERA strain is a 505 amino acid type I membrane glycoprotein that oligomerizes into trimers (Whitt et al., 1991; Gaudin et al., 1992) and that contains three Asn-X-Ser/Thr potential N-glycosylation sites (sequons) in the extracellular domain (Anilionis et al., 1981). RGP is important in the biology and pathogenesis of rabies virus infection. It is the only viral protein on the surface of virus particles anchored in the viral envelope, it is the viral attachment protein that facilitates virus uptake via the host cell receptor, and it is the target of the host humoral immune response (for review, see Coll, 1995). In addition, RGP itself is efficacious as a vaccine (Kieny et al., 1984; Prehaud et al., 1989; Xiang et al., 1994).
Appropriate glycosylation of RGP is important for its proper expression and function. For example, nonglycosylated RGP is not expressed at the cell surface (Burger et al., 1991; Shakin-Eshleman et al., 1992) and is not effective as a vaccine (Yelverton et al., 1983; Lathe et al., 1984). In addition, it is critical that one of its three sequons is glycosylated for RGP to be expressed at the cell surface (Shakin-Eshleman et al., 1992). Finally, for monoglycosylated forms of RGP, the level of cell surface expression correlates with the efficiency of glycosylation at the corresponding sequon (Shakin-Eshleman et al., 1992; Kasturi et al., 1995).
We previously constructed a soluble, truncated form of RGP (denoted RGP(WT)T434) by inserting a stop codon amino-terminal to the transmembrane domain (Shakin-Eshleman et al., 1993). This soluble form of RGP was appropriately glycosylated, assembled, and secreted by transfected cells (Wojczyk et al., 1995). It was also appropriately antigenic and protective as a vaccine (Wojczyk et al., 1995; Xiang et al., 1995). These results led to the current studies which compare the effects of site-specific glycosylation on the cell-surface expression and secretion of transmembrane and soluble forms, respectively, of the same glycoprotein. This investigation yielded the surprising finding that although any of the three sequons permitted cell surface expression of full-length RGP, only the N-glycan at Asn319 permitted secretion of soluble RGP.
In addition, despite its biological and medical importance, it has not yet been possible to structurally characterize RGP in detail. Crystallization studies have been hampered by the need for detergents to solubilize the full-length transmembrane glycoprotein and by the microheterogeneity of the oligosaccharides on mature RGP. The current studies use site-directed mutagenesis followed by eukaryotic expression in the presence of N-glycan processing inhibitors to demonstrate the possibility of purifying a simplified yet conformationally relevant form of RGP, that is, a soluble form of recombinant RGP that contains only one N-glycan with a substantial reduction in its microheterogeneity. This form of RGP may be particularly useful for future studies aimed at elucidating the three-dimensional structure of this important glycoprotein.
Construction of RGP termination mutants
Wild-type RGP of the ERA strain, denoted RGP(WT), is a 505 amino acid, type I membrane glycoprotein containing a 22 amino acid transmembrane domain, a 44 amino acid cytoplasmic domain, and three Asn-X-Ser/Thr sequons within the extracellular domain at Asn37 (position 1), Asn247 (position 2), and Asn319 (position 3) (Anilionis et al., 1981; Figure 1 ). The construction of plasmids encoding glycosylation mutants of RGP(WT): pRGP(1--), pRGP(-2-), and pRGP(--3), was previously described (Shakin-Eshleman et al., 1992). Termination mutants encoding truncated forms of RGP were generated for each glycosylation mutant by introducing a termination codon at amino acid position 434 (Figure 1 ; Shakin-Eshleman et al., 1993). The termination mutants pRGP(1--)T434, pRGP(-2-) T434, and pRGP(--3)T434 each encode a glycoprotein with a single glycosylation sequon; pRGP(WT)T434 encodes a glycoprotein with all three sequons. All termination mutants lack the transmembrane and cytoplasmic domains; therefore, during their synthesis, they are all able to be completely translocated across the membrane of the endoplasmic reticulum and released into its lumen.
Isolation of cell lines expressing truncated, soluble forms of RGP
We previously showed that RGP(WT)T434 is one of the major soluble proteins secreted by pRGP(WT)T434-transfected CHO (Chinese hamster ovary) cells (Wojczyk et al., 1995). To evaluate the importance of N-glycosylation at particular sequons on expression and secretion of soluble forms of RGP, plasmids encoding RGP(1--)T434, RGP(-2-)T434, and RGP(--3)T434 were stably cotransfected with pSV2neo into CHO cells. G418-resistant colonies were isolated and, after expansion, the cells were metabolically labeled and screened for RGP expression by immunoprecipitation of cell lysates with rabbit polyclonal antiserum followed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis. Colonies demonstrating the highest level of RGP expression were cloned by a single round of limiting dilution.
Characterization of expression, glycosylation and antigenicity of cell-associated, soluble, truncated forms of RGP
Clonal cell lines expressing each of the glycosylation mutants were metabolically labeled and RGP was immunoprecipitated from cell lysates and conditioned media using a rabbit polyclonal anti-rabies antiserum; the immune complexes were analyzed by SDS-PAGE and autoradiography (Figure 2 ). RGP(1--)T434, RGP(-2-)T434, and RGP(--3)T434 each contain only one of the three sequons normally present in the viral glycoprotein. Examination of the detergent lysate of CHO cells expressing RGP(1--)T434 revealed a major species (Mr = 48.5 kDa) consistent with unglycosylated, truncated RGP and a minor species with slightly higher apparent Mr (51 kDa), consistent with the addition of a single N-glycan (Figure 2 , lane 1). This result suggests inefficient, incomplete core glycosylation of Asn37, consistent with previous studies of glycosylation mutants of full-length RGP (Shakin-Eshleman et al., 1992). In contrast, lysates of cells expressing RGP(-2-)T434 and RGP(--3)T434 yielded a single major species consistent with the addition of a single N-linked oligosaccharide (Figure 2 , lanes 3 and 5, respectively; also see Figure 3 , lanes 5 and 7, respectively). Recombinant RGP(WT)T434 isolated from the cell lysate also had a major band, consistent with the addition of two N-glycans, and a faint band with slightly larger apparent Mr, consistent with a triply glycosylated form of soluble RGP (Figure 2 , lane 7; also see Wojczyk et al., 1995). Taken together, these results suggest efficient, complete core glycosylation at both Asn247 and Asn319, and inefficient core glycosylation at Asn37 (also see Shakin-Eshleman et al., 1992, 1993).
Influence of site-specific glycosylation on secretion of soluble RGP
Interestingly, when secretion of these truncated glycosylation mutants was compared, only RGP(--3)T434 and RGP(WT)T434 were secreted (Figure 2 , lanes 6 and 8, respectively). Based on phosphorimaging results, transfected clonal cell lines secreted at least one-half as much RGP(--3)T434 as compared with RGP(WT)T434 (data not shown). In contrast, RGP(1--)T434 and RGP(-2-)T434 were not secreted (Figure 2 , lanes 2, 4). This demonstrates that core glycosylation of Asn319 by itself can support secretion of RGP, whereas glycosylation at either only Asn37 or Asn247 can not. This is in contrast to previous results with full-length RGP glycosylation mutants, where glycosylation at any one of the three sequons was able to support cell surface expression of recombinant RGP (Shakin-Eshleman et al., 1992). In addition, the efficiency of glycosylation at individual sequons is not the only factor determining secretion of soluble forms of RGP, because RGP(-2-)T434 and RGP(--3)T434 are both efficiently glycosylated, but only the latter is secreted.
Studies with the endoglycosidases Endo H and Peptide N-glycosidase F (PNGase F) demonstrated that the N-glycan on RGP(--3)T434 secreted by wild-type CHO cells was Endo H-resistant and PNGase F-sensitive (Figure 4 , lanes 4-6), suggesting that the N-glycan on secreted RGP(--3)T434 was of the complex type. These results are similar to those found with RGP(WT)T434 (Figure 4 , lanes 1--3; see also Wojczyk et al., 1995).
Figure Influence of glycosylation inhibitors on RGP(--3)T434 secretion
Since the N-glycan on the secreted form of RGP(--3)T434 was of the complex type, it was important to examine the effect of N-glycan processing on RGP(--3)T434 secretion. Therefore, transfected cells were metabolically labeled in the presence of three glycosylation inhibitors. Radiolabeled RGP(--3)T434 was then immunoprecipitated from conditioned media and cell lysates and analyzed by SDS-PAGE.
To allow core glycosylation but then block the initial step in N-glycan processing, glucose removal, metabolic labeling was performed in the presence of castanospermine, an alpha-glucosidase inhibitor (Elbein, 1991). In the presence of castanospermine, secretion of RGP(--3)T434 was significantly decreased (Figure 5 , bottom panel, lane 4) when compared to the control (Figure 5 , bottom panel, lane 1). These results are similar to those seen with RGP(WT)T434 (Figure 5 , top panel, lanes 1,4; see also Wojczyk et al., 1995). RGP(--3)T434 secreted in the presence of castanospermine was partially Endo H resistant (Figure 5 , bottom panel, lane 5), indicating that this inhibitor was not completely efficacious in blocking processing of the glucose residues. This [rdquor]leakiness" of the castanospermine block has been seen previously by us and others (Wojczyk et al., 1995).
Figure
Since castanospermine did not result in the production of a homogeneous form of secreted RGP(--3)T434 (i.e., partial Endo H resistance), and since castanospermine partially blocked RGP(--3)T434 secretion, this inhibitor would not be useful for preparing homogenous, purified RGP(--3)T434 for structural studies. Therefore, swainsonine, an alpha-mannosidase II inhibitor, and 1-deoxymannojirimycin, an alpha-mannosidase I inhibitor, were used to examine the influence of other steps in N-glycan processing on RGP(--3)T434 secretion. Both inhibitors did not inhibit secretion of RGP(--3)T434 (Figure 5 , bottom panel; compare lane 1 with lanes 6 and 8, respectively). In addition, the secreted protein was completely Endo H sensitive in each case (Figure 5 , bottom panel, lanes 7 and 9, respectively). The effects of these latter two inhibitors on RGP(--3)T434 secretion were very similar to those for RGP(WT)T434 (Figure 5 , top panel, compare lanes 6-9) which were also reported previously (Wojczyk et al., 1995). Taken together, these results suggest that removal of glucose residues from the N-glycan of RGP(--3)T434 is a necessary and sufficient condition for secretion of this glycoprotein. In addition, further processing from a high mannose to a complex type N-glycan is not necessary for RGP(--3)T434 secretion.
RGP(--3)T434 is secreted as an oligomer
Both the transmembrane form of RGP (Whitt et al., 1991; Gaudin et al., 1992) and secreted RGP(WT)T434 (Wojczyk et al., 1995) are assembled into trimers. Therefore, it was important to determine whether secreted RGP(--3)T434 is oligomeric. To this end, chemical cross-linking studies were performed. Metabolically labeled conditioned medium was incubated with the chemical cross-linker ethylene glycolbis (succinimidylsuccinate) (EGS) and then immunoprecipitated with rabbit polyclonal anti-rabies antiserum. The immunoprecipitates were analyzed by SDS-PAGE and autoradiography. This approach identified cross-linked species having mobilities consistent with those expected for dimers and higher oligomers of RGP(--3)T434 (Figure 6 , lanes 8-10). Only the monomer form was observed when conditioned medium was incubated in the absence of the cross-linker (Figure 6 , lane 7). Virtually identical results were obtained when RGP(WT)T434 was examined by chemical cross-linking (Figure 6 , lanes 2-5; see also Wojczyk et al., 1995). Similar results were also obtained using a cleavable cross-linker, dithiobis (succinimidylpropionate) (DSP); under nonreducing conditions the pattern of monomers and higher oligomers was seen, under reducing conditions only RGP monomers were seen (data not shown). Taken together, these findings demonstrate that glycosylation of Asn319 alone, which is necessary and sufficient for supporting secretion of soluble RGP, is also sufficient for its appropriate assembly into (homo-)oligomers.
Figure Influence of site-specific glycosylation of RGP on the kinetics of its cell-associated expression
The lack of secretion of RGP(-2-)T434 was surprising and unexplained. To begin to explore the mechanism for this finding, the kinetics of disappearance of truncated forms of RGP from transfected cells due to degradation and/or secretion were analyzed by pulse-chase experiments. Transfected cells were pulse-labeled for 20 min and chased for defined times. Radiolabeled RGP was immunoprecipitated from cell lysates, analyzed by SDS-PAGE and autoradiography, and quantified by phosphorimaging. Both RGP(WT)T434 and RGP(--3)T434 have very similar kinetics of disappearance from the cell with t1/2 = ~2.5 h from the time of maximal expression (Figure 7 ); this is primarily due to their secretion (Figure 2 , lanes 6, 8). However, for RGP(1--)T434 and RGP(-2-)T434, which are not secreted into the conditioned medium (Figure 2 , lanes 2,4), the kinetics were different, with a prolonged rate of disappearance from the cell with t1/2 = ~5 h from the time of maximal expression (Figure 7 ). For comparison, pulse-chase experiments with full-length, transmembrane forms of RGP demonstrated that RGP(WT) and RGP(-2-), which are well expressed on the plasma membrane, disappeared from cells with a t1/2 = 18-19 h from the time of maximal expression (data not shown). These results suggest that, in contrast to full-length forms of RGP, nonsecreted RGP(1--)T434 and RGP(-2-)T434 are rapidly degraded intracellularly.
Figure Influence of site-specific glycosylation on the interaction of RGP with the cellular chaperone grp78/BiP
To begin to investigate possible mechanisms for retention, and subsequent degradation, of nonsecreted, truncated forms of RGP, radiolabeled RGP was immunoprecipitated from cell lysates using anti-grp78/BiP and anti-rabies antibodies. The immunoprecipitates were then analyzed by SDS-PAGE and autoradiography (Figure 8 ). grp78/BiP is a soluble protein in the endoplasmic reticulum that acts as a chaperone by associating transiently with, and facilitating appropriate folding and assembly of, newly synthesized glycoproteins (Bole et al., 1986). The loss of N-glycosylation, by site-directed mutagenesis or tunicamycin treatment, may cause conformational changes in proteins leading to surface exposure of hydrophobic regions resulting in formation of aggregates with grp78/BiP (Hurtley et al., 1989).
Figure
As was found in Figure 7 , when cell lysates from transfected wild-type CHO cells were immunoprecipitated with anti-rabies antiserum, the relevant truncated and transmembrane forms of RGP were observed (Figure 8 , lanes 2, 4, 6, 9, 12). When the lysates were incubated with anti-grp78/BiP, a ~78 kDa band corresponding to grp78/BiP was seen (Figure 8 , lanes 3, 5, 7, 10, 11, 13, 14). As positive controls, RGP(WT)T434 and RGP(WT) expressing cells were treated with tunicamycin before immunoprecipitation with anti-grp78/BiP (Figure 8 , lanes 11 and 14, respectively). In these latter cases, nonglycosylated RGP(WT) and RGP(WT)T434 are not expressed on the cell surface or secreted, respectively, by transfected cells (data not shown; see also Shakin-Eshleman et al., 1992, and Wojczyk et al., 1995). In addition, in the presence of tunicamycin, these nonglycosylated forms of RGP, along with many other nonglycosylated cellular (glyco)proteins, strongly coimmunoprecipitate with grp78/BiP (Figure 8 , lanes 11, 14).
When tunicamycin was not used and cell lysates were incubated with anti-grp78/BiP, RGP(1--)T434 was strongly coimmunoprecipitated along with grp78/BiP (Figure 8 , lane 3). In addition, anti-rabies antiserum coimmunoprecipitated grp78/BiP along with RGP(1--)T434 (Figure 8 , lane 2). These results correspond well with the finding that RGP(1--)T434 is inefficiently glycosylated (Figures 2 , 3) which then leads to the strong interaction of nonglycosylated RGP with grp78/BiP. In contrast, incubation with anti-grp78/BiP resulted in slight coimmunoprecipitation of RGP(-2-)T434 and RGP(--3)T434 (Figure 8 , lanes 5 and 7, respectively) and no coimmunoprecipitation of RGP(WT)T434 or RGP(WT) (Figure 8 , lanes 10,13, respectively); this suggests that these efficiently glycosylated forms of RGP were reasonably well-folded.
Interestingly, in the absence of tunicamycin, anti-rabies antiserum did coimmunoprecipitate grp78/BiP along with the RGP(-2-)T434, RGP(--3)T434, and RGP(WT)T434 truncated forms of RGP (Figure 8 , lanes 4, 6, 9), but not along with the full length, transmembrane form, RGP(WT) (Figure 8 , lane 12). This suggests that, under these conditions, a small portion of these truncated forms of RGP are bound to grp78/BiP either because they are not completely glycosylated, because they are subtly misfolded, or because they fold more slowly.
To determine the kinetics of the association of truncated forms of RGP with grp78/BiP, cells were pulse-labeled for 20 min and chased for defined times. The cell lysates were immunoprecipitated with anti-grp78/BiP antibody, separated by SDS-PAGE, visualized by autoradiography, and quantified by a phosphorimaging method (data not shown). These studies showed that maximal binding of RGP(1--)T434 to grp78/BiP peaked at 2 h of chase and then decayed with half-maximal binding at 4.5 h after chase. In similar studies, no interaction between grp78/BiP and either RGP(WT)T434, RGP(-2-)T434, or RGP(--3)T434 was detected.
Taken together, these results suggest that RGP(1--)T434, which is predominantly not glycosylated, is not secreted by cells because of its significant interaction with grp78/BiP followed by eventual intracellular proteolytic degradation. Although RGP(-2-)T434 binds somewhat to grp78/BiP (Figure 8 , lanes 4, 5), this does not explain its lack of secretion by the transfected cells (Figure 2 ), because a similar level of interaction with grp78/BiP was found with RGP(--3)T434 (Figure 8 , lanes 6, 7), which is similarly efficiently glycosylated but is also well-secreted (Figures 2 , 3).
Influence of site-specific glycosylation on the interaction of RGP with other cellular chaperones
Since binding to grp78/BiP does not explain the lack of secretion of RGP(-2-)T434, the interaction of RGP with another chaperone, calnexin (Ou et al., 1993), was investigated. When cell lysates were incubated with either anti-calnexin antibody or anti-rabies antiserum, no coimmunoprecipitation of calnexin with either full-length or truncated forms of RGP was found (data not shown). Additional attempts to coimmunoprecipitate any form of RGP with calnexin were tried including the use of other anti-calnexin antibodies, different conditions for cell lysis, and various glycosylation inhibitors. Interactions between calnexin and RGP were not detected in any of these experiments (data not shown).
Similar coimmunoprecipitation experiments with another chaperone, calreticulin (Nauseef et al., 1995; Sontheimer et al., 1995), were performed. Again, no interaction was detected between this chaperone and the various forms of RGP (data not shown).
Additional coimmunoprecipitation experiments were performed with an anti-KDEL antibody. The KDEL (Lys-Asp-Glu-Leu) sequence is present at the carboxyl terminus of several soluble proteins important in protein folding, including grp78/BiP, GRP 94, and protein disulfide isomerase (Freedman et al., 1994); there is no KDEL sequence in RGP of the ERA strain of rabies virus (Anilionis et al., 1981). Although RGP(1--)T434 was coimmunoprecipitated with anti-KDEL, confirming the results with anti-grp78/BiP (Figure 8 ), no other form of RGP tested was coimmunoprecipitated by this antibody (data not shown). Taken together, these results suggest that the lack of secretion of RGP(-2-)T434 is not due to binding to any cellular protein containing the KDEL sequence.
Since similar results were obtained with either RGP(-2-)T434, RGP(--3)T434, and RGP(WT)T434 in coimmunoprecipitation experiments with various known chaperones, we attempted to identify a novel cellular protein which differentially bound to RGP(-2-)T434. To this end, cell lysates were prepared in the presence of the cleavable cross-linker DTSSP and then immunoprecipitated with anti-rabies antiserum. Following separation by SDS-PAGE under nonreducing conditions, proteins were then separated in a second dimension by SDS-PAGE under reducing conditions. These studies did not identify any novel radiolabeled band which specifically associated with RGP(-2-)T434 (data not shown).
RGP(-2-)T4343 does not form large intracellular aggregates
Finally, to determine whether RGP(-2-)T434 was not secreted by transfected cells due to its assembly into large intracellular aggregates, various forms of cell-associated, metabolically labeled RGP were centrifuged on sucrose gradients and then immunoprecipitates were analyzed by SDS-PAGE. In no cases were large aggregates detected (data not shown).
Both cell surface and secreted proteins traverse the secretory pathway in order to reach their final destinations. Surprisingly, site-specific N-glycosylation has a different role in facilitating either the cell surface expression of full-length, transmembrane forms of RGP or the secretion of truncated, soluble forms of RGP. For example, N-glycosylation of any one of three glycosylation sequons is a necessary and sufficient condition for cell surface expression of full-length RGP (Shakin-Eshleman et al., 1992). In addition, the level of cell surface expression corresponds with the efficiency of glycosylation at particular sequons (Shakin-Eshleman et al., 1992; Kasturi et al., 1995). In contrast, the present study demonstrates that N-glycosylation of at least Asn319 is required for secretion of truncated forms of RGP; that is, RGP(WT)T434 and RGP(--3)T434 are secreted, but RGP(1--)T434 and RGP(-2-)T434 are not. In addition, in the presence of tunicamycin, nonglycosylated RGP(WT)T434 is not secreted.
The lack of secretion of RGP(1--)T434 can primarily be explained by its inefficient glycosylation and the strong interaction of the nonglycosylated form with grp78/BiP. As has been described for many other nonglycosylated proteins (Hurtley et al., 1989), this type of interaction with grp78/BiP can prevent secretion or cell surface expression and can lead to intracellular proteolysis of the misfolded or incompletely folded protein. This is presumably the same mechanism responsible for the low cell surface expression of full-length RGP(1--) (Shakin-Eshleman et al., 1992). Nonetheless, even though small amounts of RGP(1--)T434 are glycosylated, no immunoreactive protein was detected in the conditioned medium of the transfected cells, because glycosylated RGP(1--)T434 either is not secreted or is unstable in conditioned medium following secretion.
In contrast to RGP(1--)T434, RGP(-2-)T434 was efficiently and appropriately glycosylated, yet was still not secreted by transfected cells. The recognition of cell-associated RGP(-2-)T434 by three conformation-dependent monoclonal antibodies suggests that it folds into a reasonably appropriate conformation. However, the mechanism preventing its secretion has not yet been identified. Although by coimmunoprecipitation experiments it did not bind strongly to any chaperone tested (grp78/BiP, calnexin, calreticulin, etc.), or to any other identifiable protein, it was degraded intracellularly with a half-life similar to that of RGP(1--)T434. In addition, since proper oligomerization is an important pre-requisite for exit from the endoplasmic reticulum, experiments aimed at assessing whether RGP(-2-)T434 could correctly oligomerize were conducted. Since normal RGP trimers are not stable on sucrose gradients (Whitt et al., 1991), chemical cross-linking of intracellular forms of soluble RGP, with subsequent analysis by sucrose gradients or SDS-PAGE, was attempted. However, this approach did not yield satisfactory, interpretable results. Nonetheless, based on studies using sucrose gradients in the absence of chemical cross-linkers, RGP(-2-)T434 did not form insoluble, disulfide-linked aggregates inside the endoplasmic reticulum or other cellular compartments. This lack of aggregation is in contrast, for example, with previous results reported with variant forms of vesicular stomatitis virus G protein (Machamer and Rose, 1988).
Despite the minor binding of RGP(--3)T434 to grp78/BiP in coimmunoprecipitation experiments, suggesting that the recombinant glycoprotein may be slightly misfolded, it is nonetheless efficiently secreted by transfected cells. It is intriguing that, in addition to its role in cell surface expression of RGP(--3), the N-glycan at Asn319 is of particular importance for secretion of RGP(--3)T434. The region near Asn319, especially the Arg333 residue, is important for the virulence of rabies virus (Seif et al., 1985; Tuffereau et al., 1989) and for virus-induced fusion and syncitium formation (Morimoto et al., 1992). In addition, the nearby putative leucine zipper domain (residues 378-399 in the ERA strain; Conzelmann et al., 1990) may play a role in RGP oligomerization. It is possible that, in the absence of a transmembrane domain, the N-glycan at Asn319 is particularly critical in stabilizing the local conformation of this important region, thus allowing proper folding and oligomerization of soluble forms of RGP.
There have been few other studies comparing the role of site-specific N-glycosylation in supporting cell surface expression and secretion of membrane-bound and soluble forms of the same glycoprotein. For example, using site-directed mutagenesis, Tifft et al. determined that recombinant human CD4 is expressed at the cell surface if either or both of the two N-glycosylation sequons are glycosylated (Tifft et al., 1992); studies using site-directed mutagenesis or tunicamycin demonstrated that there was no cell surface expression of the nonglycosylated protein (Konig et al., 1988; Tifft et al., 1992). In contrast, a soluble, truncated form of human CD4 was efficiently secreted by transfected cells in the absence of N-glycosylation at either site (Davis et al., 1990). Another study examined variant forms of growth hormone, which is normally efficiently secreted as a nonglycosylated protein (Guan et al., 1985). When a chimeric, membrane-bound form of growth hormone was constructed by linking it to the transmembrane and cytoplasmic domains of vesicular stomatitis virus G protein, the resulting chimera was not expressed at the cell surface unless novel N-glycosylation sequons were inserted, allowing N-glycans to be attached to the extracytoplasmic, growth hormone domain. The same group also examined analogous chimeras creating membrane-bound forms of the alpha-subunit of human chorionic gonadotropin (hCG; Guan et al., 1988), normally a secreted glycoprotein with two efficiently N-glycosylated sequons. In this case, nonglycosylated forms of soluble hCG were efficiently secreted, whereas nonglycosylated membrane-bound forms of hCG were not expressed at the cell surface. Although, the N-glycan at Asn78 of hCG initially seemed to be particularly important for cell surface expression of the chimeric protein, further studies suggested that the local amino acid sequence at Asn78, and not the N-glycan itself, was important in maintaining the stability and conformation of this protein. Taken together, these studies suggested that secreted proteins might be more tolerant than membrane-bound proteins to differences in conformation or solubility due to the absence of N-glycans (Guan et al., 1988). However, these findings contrast significantly with the current results where N-glycosylation specifically at Asn319 is required for secretion of soluble RGP, whereas glycosylation at either Asn37, Asn247, or Asn319 permits cell-surface expression of full-length RGP.
Since only the N-glycan at Asn319 is required for secretion of truncated RGP, the effects of processing of this N-glycan on secretion of RGP(--3)T434 were examined. As found previously with RGP(WT)T434, castanospermine inhibited secretion of the recombinant glycoprotein. Inhibition of the function of the trimming alpha-glucosidase was found to similarly inhibit secretion of other glycoproteins (e.g., lipoprotein lipase; Masuno et al., 1992); in contrast, some proteins are efficiently secreted in the presence of inhibitor (e.g., IgM; Peyrieras et al., 1983). In analogy to RGP(WT)T434, RGP(--3)T434 was efficiently secreted in the presence of either the alpha-mannosidase I inhibitor 1-deoxymannojirimycin or the alpha-mannosidase II inhibitor swainsonine.
Efficient secretion of RGP(--3)T434 in the presence of 1-deoxymannojirimycin will make it possible to purify from CHO cells (Wojczyk et al., 1996) a form of this recombinant glycoprotein with a substantial reduction in the microheterogeneity of its N-glycan. The N-glycan secreted, soluble RGP(--3)T434 will have, at most, a mixture of the Man9GlcNAc2 and Man8GlcNAc2 sequences (Bischoff et al., 1986). However, even this limited degree of N-glycan heterogeneity in the presence of 1-deoxymannojirimycin can be protein and glycosylation site specific (Bischoff et al., 1986), and preliminary studies with RGP(WT)T434 suggest that its N-glycans are quite homogeneous under these conditions (data not shown). Due to its appropriate oligomerization and antigenicity, appropriate immunogenicity as a DNA vaccine (unpublished results), solubility, and limited heterogeneity, RGP(--3)T434 is a biologically relevant form of RGP that may be particularly amenable to crystallization. This would then allow the determination of the three-dimensional structure of this medically important glycoprotein.
Construction of RGP termination mutants
The construction of plasmid pRGP(WT), which contains the cDNA sequence for the full-length, wild-type RGP from the ERA strain of rabies virus (Kieny et al., 1984), inserted into the BglII site of the pSG5 eukaryotic expression vector (Green et al., 1988), was described previously (Burger et al., 1991).
The construction of plasmids encoding mono-glycosylated forms of full-length RGP: pRGP(1--), pRGP(-2-), and pRGP(--3), was described previously (Shakin-Eshleman et al., 1992). The numbers in parentheses indicate the presence of glycosylation sequons 1, 2, or 3, at Asn37, Asn247, or Asn319, respectively; the symbol (-) indicates the deletion of the corresponding sequon by site-directed mutagenesis (Figure 1 ).
Termination mutants containing one or more of the glycosylation sequons of RGP were generated by inserting the stop linker CTAGCTAGCTAG (Pharmacia, Piscataway, NJ) at the HincII restriction site in pRGP(WT), pRGP(1--), pRGP(-2-), and pRGP(--3), as described previously (Shakin-Eshleman et al., 1993). These plasmids, denoted pRGP(WT)T434, pRGP(1--) T434, pRGP(-2-)T434, and pRGP(--3)T434, encode truncated forms of RGP that lack the transmembrane and cytoplasmic domains of the full-length protein. The soluble protein, RGP(WT)T434, is secreted by transfected cells (Wojczyk et al., 1995).
Cell lines and tissue culture
The Pro-5 wild-type CHO cell line (Stanley et al., 1975) was obtained from the American Type Culture Collection (Rockville, MD). The transfected CHO cell lines expressing either full-length or soluble wild-type RGP (RGP(WT) and RGP(WT)T434, respectively) were described previously (Burger et al., 1991; Shakin-Eshleman et al., 1992; Wojczyk et al., 1995). Cells were routinely cultured in complete medium: alpha-modified minimal essential medium (alpha-MEM) supplemented with 10% fetal calf serum, 2 mM glutamine, 100 IU/ml of penicillin, 100 µg/ml of streptomycin, and 0.25 µg/ml of amphotericin B.
Monoclonal and polyclonal antibodies
Rabbit polyclonal anti-rabies virus antiserum was produced by immunizing New Zealand White rabbits (East Acres Biologicals; Southbridge, MA) intramuscularly with 100 µg of [beta]-propiolactone inactivated rabies virus (ERA strain) vaccine in complete Freund's adjuvant using a previously described protocol (Wiktor et al., 1973). Four booster intramuscular immunizations were performed at weekly intervals in incomplete Freund's adjuvant using the same concentration of vaccine. Antiserum collected 1 week following the last booster immunization was titered using immunoprecipitation and SDS-PAGE (see below).
The 523-11 and 509-6 mouse monoclonal RGP-specific antibodies were used from ascites, were produced at The Wistar Institute, and were described previously (Flamand et al., 1980; Lafon et al., 1983). The hybridoma cell line producing the 62-80-6 anti-RGP mouse monoclonal antibody (Smith et al., 1984) was provided by J. Smith (Centers for Disease Control; Atlanta, GA); ascites containing 62-80-6 was used in the studies described below. Rabbit polyclonal anti-KDEL and anti-calreticulin antibodies were purchased from Affinity Bioreagents (Neshanic Station, NJ). Mouse monoclonal anti-grp78/BiP and anti-protein disulfide isomerase antibodies, and rabbit polyclonal anti-grp78/BiP, antibodies recognizing the carboxyl terminus or the amino terminus of calnexin were purchased from Stress Gen (Victoria, BC, Canada). In addition, a rabbit polyclonal anti-calnexin antibody (Ou et al., 1993) was kindly provided by Dr. John J. M. Bergeron (McGill University; Montreal, Canada).
Stable transfection
CHO cells were cotransfected with 10 µg of a plasmid encoding a truncated form of RGP and 1 µg of pSV2neo (Southern and Berg, 1982) with a 2 min glycerol shock at 4 h (Burger et al., 1991). After 48 h, and every 2 days thereafter, the medium was replaced with fresh complete alpha-MEM containing 1 mg/ml of active G418 (Gibco, Grand Island, NY). G418-resistant colonies were isolated, amplified, and screened for secretion and/or cell-associated expression of metabolically labeled RGP (see below). Clonal cell lines were isolated by repetitive subcloning by limiting dilution and maintained in complete alpha-MEM containing 0.5 mg/ml of G418.
Metabolic labeling
Petri dishes (100 mm) of 80% confluent transfected cells were washed twice with Hanks' balanced salt solution (HBSS; Sigma, St. Louis, MO). Five milliliters of methionine-free complete Dulbecco's modified Eagle's medium (DMEM), containing 50 µCi of (35S)methionine (Amersham Corp., Arlington Heights, IL) and the same supplements as complete alpha-MEM, were added.
In some experiments, the effect of inhibiting N-glycosylation on the expression of RGP was examined. In these cases, the cells were preincubated with one of the following soluble glycosylation inhibitors (for review, see Elbein, 1991) in 1 ml of complete alpha-MEM: 50 µg/ml castanospermine, 5 µg/ml tunicamycin, 5 µg/ml swainsonine, or 1 mM 1-deoxymannojirimycin. Following incubation at 37°C for 2 hr, the cells were washed twice with HBSS. One milliliter of methionine-free complete DMEM was then added, containing both 18 µCi of [35S]methionine and the appropriate glycosylation inhibitor at the concentration described above.
Following a 4 h incubation at 37°C, the conditioned medium was collected. The cells were washed twice with ice-cold phosphate-buffered saline (PBS; 100 mM Na2HPO4/NaH2PO4, 150 mM NaCl, pH 7.4) containing 200 µg/ml phenylmethylsulfonyl fluoride (PMSF) (PBS/PMSF), scraped, and transferred to 1.5 ml Eppendorf tubes on ice. Cells were washed once with ice-cold PBS/PMSF, resuspended in 100 µl of lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, pH 7.4, containing 0.5% Nonidet P-40 and 200 µg/ml PMSF), incubated on ice for 20 min, and clarified by centrifugation at 12,000 × g for 20 min at 4°C.
Pulse-chase experiments
Sixty millimeter dishes of transfected cells were washed twice with HBSS. One milliliter of methionine-free complete DMEM was then added, containing 18 µCi of [35S]methionine. After labeling at 37°C for a 20 min pulse, the cells were washed twice with HBSS. Then 1 ml of methionine-free complete DMEM, containing an excess of cold methionine (1 mM), was added. At chase times of 0, 1, 2, 4, and 20 h after the pulse, the conditioned media were removed. The remaining cells were washed once with ice-cold PBS/PMSF, resuspended in 100 µl of lysis buffer, incubated on ice for 20 min, and clarified by centrifugation at 12,000 × g for 20 min at 4°C (cell lysate).
Immunoprecipitation
From one 100 mm dish of transfected cells expressing radiolabeled RGP, 800 µl of conditioned medium diluted with 200 µl of 5× RIPA Buffer (100 mM Tris pH 7.5, 1% Triton X-100, 1% deoxycholate, 0.3 M NaCl, 0.1% SDS, 1 mM PMSF) and 100 µl of cell lysate were prepared. Rabbit polyclonal anti-rabies virus antiserum (1:100 final dilution) or one of the RGP-specific mouse monoclonal antibodies (1:50 final dilution) was added to the cell lysate (for the cell-associated form of RGP) or conditioned medium (for the secreted forms of RGP) and incubated with gentle agitation for 16 h at 4°C. Alternatively, one of the anti-grp78/BiP, anti-calnexin, anti-proline disulfide isomerase, anti-calreticulin, or anti-KDEL antibodies was added to the cell lysate at a final dilution of 1:50. Immune complexes were isolated with either Protein A beads or Protein G beads (Gibco BRL, Gaithersburg, MD) depending on the type of antibody. Following a 3 h incubation at 4°C with gentle agitation with either 20 µl or 40 µl of a 50% slurry of beads (for the cell lysate or conditioned medium, respectively), the beads were washed three times for 10 min each at 4°C with gentle agitation in wash buffer (15 mM Tris pH 7.5, 0.5 M NaCl, 5 mM EDTA, 1% Nonidet P-40). Immunoprecipitated RGP was eluted from the beads by boiling for 5 min in 40 µl of either sample buffer (62 mM Tris pH 6.8, 2% SDS, 10% glycerol) containing 5% [beta]-mercaptoethanol or in an endoglycosidase buffer (see below), and analyzed directly by SDS-PAGE or subjected to endoglycosidase digestion, respectively.
Oligomerization analysis
To each 800 µl aliquot of conditioned medium from metabolically labeled cells, 10, 30, or 40 µl of a stock solution of the covalent cross-linker EGS (Pierce Chemical Co; Rockford, IL) dissolved in dimethyl sulfoxide (DMSO; 18.4 mg EGS/250 µl DMSO) was added and incubated for 30 min at room temperature (Doms et al., 1987). The final concentrations of EGS were 2, 6, and 8 mM. The reaction was quenched by incubation with 30 mM glycine for 30 min at room temperature. Cross-linked, secreted forms of RGP were immunoprecipitated and analyzed by SDS-PAGE on 9% reducing gels. In some experiments, the cleavable cross-linker DSP (Pierce) was used (1--10 mM) and the immunoprecipitates were analyzed by SDS-PAGE under either reducing or nonreducing conditions (Wojczyk et al., 1995).
For sedimentation analysis (Doms et al., 1987; Wojczyk et al., 1995), 14 ml linear sucrose gradients were prepared from stocks of 5% and 20% sucrose (w/w) in MNT buffer (20 mM 2-(N-morpholino)ethanesulfonic acid, 100 mM NaCl, and 30 mM Tris at pH 7.4). Radiolabeled conditioned medium (200 µl) diluted with 200 µl MNT buffer was loaded onto the prepared gradients and centrifuged at 40,000 r.p.m. for 16 h at 4°C using an SW 40Ti rotor (Beckman Instruments, Inc.; Palo Alto, CA). Fractions (800 µl) were collected from the bottom of each tube, mixed with 200 µl of 5× RIPA buffer, immunoprecipitated, and analyzed by SDS-PAGE.
Endoglycosidase digestion
Immunoprecipitated, radiolabeled, full-length, or truncated RGP were eluted from Protein A beads into either Endo H buffer (60 mM Na2HPO4/NaH2PO4, pH 5.5, containing 1% SDS, 200 µg/ml PMSF) or Peptide N-glycosidase F (PNGase F) Buffer (30 mM Na2HPO4/NaH2PO4, pH 7.2, 20 mM EDTA). Samples were divided, supplemented with 4 µl of Endo H (4 mU; Boehringer-Mannheim; Indianapolis, IN) or PNGase F (0.8 U; Boehringer-Mannheim) in the corresponding buffer, or with the same volume of buffer alone, and incubated for 24 h at 37°C. An additional volume (4 µl) of Endo H or PNGase F was added after 16 h of incubation. The samples were then mixed with 12 µl of 5× sample buffer, boiled, and analyzed by SDS-PAGE.
Analysis of radiolabeled proteins by SDS-PAGE
Proteins suspended in sample buffer were reduced with 5% [beta]-mercaptoethanol, boiled for 5 min, and separated by SDS-PAGE (10% gels). After electrophoresis, gels were fixed, incubated with Amplify (Amersham Corp.), dried, and exposed to Kodak XAR-5 film (Eastman Kodak; Rochester, NY). Gels were also analyzed directly using a phosphorimager (Molecular Dynamics; Sunnyvale, CA) and Image-quant version 3.22 software. The relative molecular masses of radiolabeled proteins were determined using the following [14C]-labeled markers (Gibco-BRL): myosin H-chain (200 kDa), phosphorylase B (97.4 kDa), bovine serum albumin (68 kDa), ovalbumin (43 kDa), carbonic anhydrase (29 kDa), or [beta]-lactoglobulin (18 kDa).
Two-dimensional SDS-PAGE analysis
In an effort to identify novel chaperones interacting with RGP, experiments using two-dimensional SDS-PAGE were performed. Cells (80% confluent) in 100 mm Petri dishes were labeled for 4 h with [35S]methionine. Following two washes with PBS/PMSF, the cells from each dish were lysed with 200 µl of 1% Triton X-100 in PBS containing 100 µg/ml of the cleavable cross-linker DTSSP (Pierce). Following a 30 min incubation at room temperature, 10 µl of 1 M Tris, pH 7.5 were added. The lysate was incubated on ice for 5 min and then centrifuged at 12,000 × g for 20 min at 4°C. Lysates were immunoprecipitated with rabbit anti-rabies antiserum and Protein A-agarose, as above. Following elution of the proteins into sample buffer lacking [beta]-mercaptoethanol, the samples were separated by SDS-PAGE on 9% gels under nonreducing conditions. The individual lanes containing the separated proteins were cut from the gel, soaked in sample buffer containing 5% [beta]-mercaptoethanol for 10 min, and then polymerized to the top of a second 9% gel. Following this second electrophoresis under reducing conditions, the proteins were visualized by autoradiography.
We thank R. Doms, P. Curtis, and J. Bergeron for helpful discussions and J. Bergeron and J. Smith for providing antibodies and hybridoma cell lines.
RGP, rabies virus glycoprotein; CHO, Chinese hamster ovary; alpha-MEM, alpha-modified minimal essential medium; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; HBSS, Hanks' balanced salt solution; DMEM: Dulbecco's modified Eagle's medium; PBS, phosphate-buffered saline; PMSF, phenylmethylsulfonyl fluoride; EGS, ethylene glycolbis (succinimidylsuccinate); DMSO, dimethyl sulfoxide; DSP, dithiobis (succinimidylpropionate); DTSSP, 3,3'-dithiobis(sulfosuccinimidylpropionate); Endo H, Endoglycosidase H; PNGase F, Peptide N-glycosidase F; hCG, human chorionic gonadotropin; EDTA, ethylenediaminetetraacetic acid.
5To whom correspondence should be addressed at: Department of Pathology and Laboratory Medicine, 220 John Morgan Building, University of Pennsylvania, Philadelphia, PA 19104, USA
Glycobiology
Pages
©
Introduction
Results
Discussion
Materials and methods
Acknowledgments
Abbreviations
References
Introduction
Results
Discussion
Acknowledgments
Abbreviations
References
6Present address: Department of Pathology, Johns Hopkins University, 720 Rutland Avenue, Baltimore, MD 21287, USA
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 27 Jan 1998
Copyright© Oxford University Press, 1998.
![]()
CiteULike
Connotea
Del.icio.us What's this?
This article has been cited by other articles:
![]() |
L. Sissoeff, M. Mousli, P. England, and C. Tuffereau Stable trimerization of recombinant rabies virus glycoprotein ectodomain is required for interaction with the p75NTR receptor J. Gen. Virol., September 1, 2005; 86(9): 2543 - 2552. [Abstract] [Full Text] [PDF] |
||||
![]() |
B. S. Wojczyk, N. Takahashi, M. T. Levy, D. W. Andrews, W. R. Abrams, W. H. Wunner, and S. L. Spitalnik N-glycosylation at one rabies virus glycoprotein sequon influences N-glycan processing at a distant sequon on the same molecule Glycobiology, June 1, 2005; 15(6): 655 - 666. [Abstract] [Full Text] [PDF] |
||||
![]() |
K. Nakatsukasa, S. Okada, K. Umebayashi, R. Fukuda, S.-i. Nishikawa, and T. Endo Roles of O-Mannosylation of Aberrant Proteins in Reduction of the Load for Endoplasmic Reticulum Chaperones in Yeast J. Biol. Chem., November 26, 2004; 279(48): 49762 - 49772. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. N. R. Weber, M. A. Morse, and N. J. Gay Four N-linked Glycosylation Sites in Human Toll-like Receptor 2 Cooperate to Direct Efficient Biosynthesis and Secretion J. Biol. Chem., August 13, 2004; 279(33): 34589 - 34594. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. A. Cooper, H. J. Joshi, M. J. Harrison, M. R. Wilkins, and N. H. Packer GlycoSuiteDB: a curated relational database of glycoprotein glycan structures and their biological sources. 2003 update Nucleic Acids Res., January 1, 2003; 31(1): 511 - 513. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. P. Maillard and Y. Gaudin Rabies virus glycoprotein can fold in two alternative, antigenically distinct conformations depending on membrane-anchor type J. Gen. Virol., June 1, 2002; 83(6): 1465 - 1476. [Abstract] [Full Text] [PDF] |
||||
![]() |
M. Flamand, F. Megret, M. Mathieu, J. Lepault, F. A. Rey, and V. Deubel Dengue Virus Type 1 Nonstructural Glycoprotein NS1 Is Secreted from Mammalian Cells as a Soluble Hexamer in a Glycosylation-Dependent Fashion J. Virol., July 1, 1999; 73(7): 6104 - 6110. [Abstract] [Full Text] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||






