|Home | About | Journals | Submit | Contact Us | Français|
The prototype foamy virus (PFV) glycoprotein, which is essential for PFV particle release, displays a highly unusual biosynthesis, resulting in posttranslational cleavage of the precursor protein into three particle-associated subunits, i.e., leader peptide (LP), surface (SU), and transmembrane (TM). Glycosidase digestion of metabolically labeled PFV particles revealed the presence of N-linked carbohydrates on all subunits. The differential sensitivity to specific glycosidases indicated that all oligosaccharides on LP and TM are of the high-mannose or hybrid type, whereas most of those attached to SU, which contribute to about 50% of its molecular weight, are of the complex type. Individual inactivation of all 15 potential N-glycosylation sites in PFV Env demonstrated that 14 are used, i.e., 1 out of 2 in LP, 10 in SU, and 3 in TM. Analysis of the individual altered glycoproteins revealed defects in intracellular processing, support of particle release, and infectivity for three mutants, having the evolutionarily conserved glycosylation sites N8 in SU or N13 and N15 in the cysteine-rich central “sheets-and-loops” region of TM inactivated. Examination of alternative mutants with mutations affecting glycosylation or surrounding sequences at these sites indicated that inhibition of glycosylation at N8 and N13 most likely is responsible for the observed replication defects, whereas for N15 surrounding sequences seem to contribute to a temperature-sensitive phenotype. Taken together these data demonstrate that PFV Env and in particular the SU subunit are heavily N glycosylated and suggest that although most carbohydrates are dispensable individually, some evolutionarily conserved sites are important for normal Env function of FV isolates from different species.
The envelope glycoprotein (Env) of retroviruses is synthesized as a precursor molecule in infected cells (reviewed in reference 29). It assembles into oligomeric complexes in the rough endoplasmic reticulum (RER), is extensively modified, and then is cleaved by a cell-encoded protease into surface (SU) and transmembrane (TM) subunits during transport to the cell surface before being incorporated into the budding retroviral particle. Some modifications, such as addition of N-linked sugars, start while the nascent protein is still cotranslationally translocated across the membrane and into the lumen of the RER. The sites of covalent attachment are asparagine (N) residues within the canonical sequence for N-linked carbohydrates, N-X-S/T, where X is any amino acid except proline. During the process of oligomerization of the Env proteins and their intracellular transport to the cell surface, the carbohydrate chains are heavily modified by removal and addition of specific sugar residues. In general there are three types of N-linked carbohydrate chains, i.e., high-mannose, hybrid, and complex, which can be distinguished experimentally by digestion with specific glycosidases. The number and position of N-glycosylation sites vary widely among retroviruses. For example, the Env protein of human immunodeficiency virus (HIV) type 1 has approximately 30 glycosylation sites in the SU and TM subunits (6, 13), whereas the Env protein of murine leukemia virus (MuLV) harbors only 7 to 8 in SU and none in TM (5, 12). At least some of the oligosaccharides must have important roles in the proper folding of Env, because treatment of cells with the glycosidase inhibitor tunicamycin results in unglycosylated Env molecules that are trapped in the ER. However, in some cases (e.g., MuLV and HIV) individual sites and in a few cases (e.g., MuLV and simian immunodeficiency virus) multiple sites of N-linked carbohydrate chains can be eliminated by mutagenesis without affecting the transport of the Env protein (5, 12), whereas in other cases (e.g., HIV and MuLV), mutagenesis of specific sites results in a block only after transport to the RER or Golgi (5-7, 12, 26). In addition to their role in protein folding, carbohydrates of retroviral glycoproteins have also been implicated in other functions such as masking of immunodominant epitopes (e.g., HIV and simian immunodeficiency virus) or regulation of coreceptor usage (e.g., HIV) (23, 24, 26, 28).
Recently retroviruses have been regrouped into two subfamilies, the Orthoretrovirinae and Spumaretrovirinae. As an expression of their unique replication strategy, foamy viruses (FVs) constitute the only genus of the subfamily Spumaretrovirinae (reviewed in reference 27). The particle-associated glycoprotein of FV is unique compared to other retroviral envelope proteins because its coexpression is strictly required for the FV particle release process and its function cannot be replaced by heterologous viral glycoproteins (reviewed in reference 18). The FV envelope precursor protein has a highly unusual biosynthesis for a retroviral glycoprotein. It is translated as a full-length precursor protein into the RER and initially has a type III protein configuration, with both its N and C termini located intracytoplasmically (10, 19). Only during its transport to the cell surface is it posttranslationally processed by cellular, most likely furin-like proteases, and not the signal peptidase complex, into at least three subunits (4, 9). The N-terminal signal or leader peptide (LP) has a type II conformation, whereas the C-terminal TM subunit has a type I conformation. The internal SU subunit presumably associates with extracellular domains of TM on the luminal side (19, 32). Image reconstruction analysis from cryoelectron microscopy pictures of the characteristic prominent Env spike structures on FV particles indicates that the heterotrimeric Env protein complexes form trimers, similar to those reported for other viral glycoproteins (31). For the FV budding process at least two essential interactions between Env and Gag are required (19, 21). One of these is the contact of the N-terminal cytoplasmic region of the FV Env LP, the so-called budding domain, containing an essential conserved WXXW sequence motif, with the N terminus of the FV Gag protein (19, 32). The LP of FV Env is glycosylated, and cleavage products are viral particle associated (19). In the present study we intended to determine the extent of prototype FV (PFV) Env N glycosylation, and we analyzed the requirement of individual N glycosylation sites for PFV particle release and infectivity.
The PFV retroviral vector pMH118 expressing an enhanced green fluorescent protein (EGFP) from an internal spleen focus-forming virus U3 promoter has been described previously (14). The pMH120 vector is a variant thereof having the EGFP marker gene replaced by the LacZ gene. The expression constructs of the individual PFV envelope mutants with mutations affecting the first three potential N-glycosylation sites (ΔN1, pczHFVenvEM058 [N25Q]; ΔN2, pczHFVenvEM077 [N109Q]; and ΔN3, pczHFVenvEM078 [N141Q]) and of the SU/TM cleavage site mutant (pczHFVenvEM020 [R571T]) were described earlier (19, 22). All further N-glycosylation site mutants were generated by recombinant PCR techniques (11) and are based on the prototype foamy virus gp130Env expression construct pczHFVenvEM002 described earlier (21). All PCR-derived fragments were sequenced to confirm the desired mutations and exclude further off-site mutations. Details on the cloning procedures for the individual mutants are available on request. The following new single-site mutants were generated: ΔN4, pczHFVenvEM105 (N183Q); ΔN5, pczHFVenvEM106 (N286Q); ΔN6, pczHFVenvEM107 (N311Q); ΔN7, pczHFVenvEM108 (N346Q); ΔN8, pczHFVenvEM109 (N391Q); ΔN8.1, pczHFVenvEM131 (S393V); ΔN8.2, pczHFVenvEM151 (T392V); ΔN9, pczHFVenvEM110 (N405Q); ΔN10, pczHFVenvEM111 (N423Q); ΔN11, pczHFVenvEM112 (N527Q); ΔN12, pczHFVenvEM113 (N556Q); ΔN13, pczHFVenvEM114 (N782Q); ΔN13.1, pczHFVenvEM132 (S784V); ΔN13.2, pczHFVenvEM152 (S783V); ΔN14, pczHFVenvEM115 (N808Q); ΔN15, pczHFVenvEM116 (N833Q); ΔN15.1, pczHFVenvEM133 (T835A); and ΔN15.2, pczHFVenvEM153 (E834Q). The positions of the potential N-glycosylation sites in the subdomains of the PFV Env protein are shown schematically in Fig. Fig.11.
For biochemical analysis of the PFV Env mutants, the individual expression constructs were cotransfected together with FV vector pMH118 into 293T cells by the calcium phosphate or polyethylene imine method essentially as described previously (19). Western blot analysis was performed as described previously (19). Polyclonal antisera used were specific for PFV Gag (1), the LP domain of PFV Env (amino acids [aa] 1 to 86) (19), or the LP and SU domains of PFV Env (aa 1 to 571) (20). Furthermore, a hybridoma supernatant (clone P3E10) specific for the SU subunit of PFV Env was employed in some experiments (4). Metabolic labeling of transfected cells and viral particles secreted into the supernatant, particle purification, and immunoprecipitation with FV protein-specific antisera were done as described previously (19). For glycosidase treatment of viral proteins, immune complexes were eluted from protein A-Sepharose pellets after the final washing step by boiling in 60 μl 1× denaturation buffer (0.5% sodium dodecyl sulfate [SDS], 1% β-mercaptoethanol) for 10 min at 98°C. Subsequently, the supernatant was split in three equal parts and transferred to fresh tubes, and 4 μl 10× G5 buffer (0.5 M sodium citrate, pH 5.5) or 4 μl 10× G7 buffer (0.5 M sodium phosphate, pH 7.5) and 4 μl 10% NP-40 were added for endoglycosidase H (endo H) or N-glycosidase F (PNGase F) digestion, respectively. Following volume adjustment to 40 μl with H2O and addition of 250 units endo H or PNGase F (New England Biolabs), the tubes and corresponding mock-treated controls lacking the enzymes were incubated for 4 to 6 h at 37°C. Cell surface biotinylation of 293T cells transiently transfected with the individual Env expression constructs was carried out essentially as described previously (22). Briefly, 293T cells were transiently transfected and metabolically labeled with [35S]methionine and [35S]cysteine. At 36 h after the addition of the DNA, cell surface protein was labeled with NHS-biotin (Calbiochem) at 1 mg/ml in phosphate-buffered saline for 30 min. Subsequently, the biotinylation reaction was stopped by adding phosphate-buffered saline containing 100 mM glycine prior to cell lysis in radioimmunoprecipitation assay buffer. Lysates were precipitated with an FV-positive chimpanzee serum as described earlier (8, 20), separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), and blotted onto nitrocellulose membranes (Hybond ECL; Amersham). Envelope protein expression at the cell surface was analyzed using streptavidin conjugated to horseradish peroxidase (Pierce), followed by detection with ECL Plus (Amersham). The chemiluminescent biotin signal was allowed to fade overnight. Thereafter, the blot was exposed to X-ray film, and total cellular envelope expression was detected by autoradiography.
Supernatants containing recombinant viral particles were generated essentially as described earlier (17, 20). Briefly, FV supernatants were produced by cotransfection of 293T cells with the Gag/Pol-expressing vector pMH118 (EGFP) or pMH120 (LacZ) and an Env expression plasmid as indicated. Twenty-four hours posttransfection, sodium butyrate (10 mM final concentration) was added to the growth medium for 8 h. Subsequently the medium was replaced, and viral supernatants were harvested an additional 16 h later. Transductions were performed by infection of 1.5 × 104 cells plated 24 h in advance in 12-well plates for 4 h using 1 ml of viral supernatant or dilutions thereof. The amount of EGFP-positive cells was determined by fluorescence-activated cell sorter analysis 48 to 72 h after infection. All transduction experiments were performed at least three times, and in each independent experiment the values obtained with wild-type (wt) PFV Env were arbitrarily set to 100. Alternatively, LacZ-expressing viral supernatants were titrated by infection of target cells with 10-fold serial dilutions in 12-well plates and subsequent histochemical β-galactosidase staining 48 h postinfection.
For assaying the potential temperature sensitivity of specific mutants, virus was produced in parallel at either 37°C or 30°C, by shifting one plate per construct of duplicate transfections to 30°C during the sodium butyrate induction step and leaving it at this temperature until virus was harvested 24 h later. Subsequently, the HT1080 target cells were incubated with viral supernatant or dilutions thereof for 6 h at 30°C. After removal of the viral supernatant, the target cells were incubated for an additional 16 h at 30°C before they were shifted to 37°C and assayed by fluorescence-activated cell sorting 48 h later.
FV particles contain three different processed glycoprotein subunits, LP, SU, and TM (4, 19). A schematic outline of the PFV gp130Env precursor organization and the locations of putative N-glycosylation sites within the individual subunits are depicted in Fig. Fig.1.1. First we determined which of these subunits contain N-linked carbohydrate chains. Therefore, PFV particles were generated by cotransfection of 293T cells with the PFV Gag/Pol-expressing vector pMH118 and the wild-type PFV Env protein expression construct pczHFVenvEM002 or the PFV Env SU/TM cleavage mutant expression construct pczHFVenvEM020 and subsequently metabolically labeled. Following immunoprecipitation of cell- and PFV particle-associated viral proteins by using a mixture of anti-PFV Env and Gag antisera, the immune complexes were digested with endo H or PNGase F or mock incubated. In Fig. Fig.22 the result of an SDS-PAGE analysis of such samples is shown. All three viral particle-associated PFV Env subunits of the wild-type protein contain N-linked carbohydrates, as their mobility increased after PNGase F treatment (Fig. (Fig.2,2, lane 8) compared to a mock-treated control (Fig. (Fig.2,2, lane 2). Similarly, both Env subunits of a SU/TM cleavage mutant (ΔSU/TM), LP and SU-TM, showed a decrease in molecular weight upon PNGase F treatment (Fig. (Fig.2,2, lane 9). Interestingly, a similar mobility shift of the LP and TM subunits of the wild-type protein was seen in endo H (Fig. (Fig.2,2, lane 5)- and PNGase F (Fig. (Fig.2,2, lane 8)-treated samples. In contrast, the SU subunit showed only a small mobility shift after endo H treatment (Fig. (Fig.2,2, lane 5) but much larger shift after PNGase F treatment (Fig. (Fig.2,2, lane 8). Thus, this analysis indicated that all viral particle-associated PFV Env subunits are N glycosylated. Furthermore, the differential sensitivity of the individual subunits to endo H and PNGase F suggests that that the N-linked carbohydrate chains of the LP and TM subunits are of the high-mannose or hybrid type, whereas at least some if not most of the carbohydrate chains of the SU subunit are of the complex type.
The results presented above demonstrated that all PFV Env subunits are N glycosylated. To identify at which of the 15 potential N-linked glycosylation sites (N1 to N15) found in the PFV Env protein sequence carbohydrate chains are indeed added, 15 PFV Env mutants, termed ΔN1 to ΔN15, were generated, in which one individual potential glycosylation site at a time was inactivated by changing the N-X-S/T consensus sequence to Q-X-S/T. After cotransfection of the different single mutants together with pMH118 into 293T cells, samples of cell lysates were separated by SDS-PAGE and analyzed by Western blotting using a PFV Gag (Fig. (Fig.3A)3A) - or PFV Env LP (Fig. (Fig.3B)-specific3B)-specific antiserum or a PFV SU-specific monoclonal antibody (Fig. (Fig.3C).3C). All mutants were expressed at similar levels intracellularly (Fig. (Fig.3B).3B). Separation of the precursor protein gp130Env by low-percentage SDS-PAGE revealed small but detectable mobility shifts for mutants ΔN2 to ΔN15 (Fig. (Fig.3B,3B, upper panel, lanes 2 to 15) compared to the wild-type protein (Fig. (Fig.3B,3B, upper panel, lane 16). This indicated that carbohydrate chains are added at all potential N-glycosylation sites of the PFV Env protein except the first and confirmed our previous glycosylation analysis of the PFV Env leader peptide gp18LP (19). Furthermore, a mobility shift of the LP cleavage products was observed only for the ΔN2 mutant (Fig. (Fig.3B,3B, middle panel, lane 2), although both ΔN2 and ΔN3 showed a mobility shift of the gp130Env precursor protein (Fig. (Fig.3B,3B, upper panel, lanes 2 and 3). This implies that LP cleavage occurs between PFV Env aa 109 and 141, which we recently confirmed by N-terminal protein sequencing of PFV Env subunits (4). In addition, the Western blot analysis of cell lysates revealed that the individual glycosylation mutants showed a differential processing of the LP. Mutants ΔN13 and ΔN14 (Fig. (Fig.3B,3B, middle panel, lanes 13 and 14) displayed a moderately reduced processing and mutant ΔN15 (Fig. (Fig.3B,3B, middle panel, lane 15) displayed a strongly reduced processing, whereas no processing at all could be detected for ΔN8 (Fig. (Fig.3B,3B, middle panel, lane 8). Cell surface biotinylation analysis of selected mutants revealed intracellular transport defects for mutants ΔN8, ΔN13, and ΔN15 (Fig. (Fig.4).4). For wild-type PFV Env (Fig. (Fig.4A,4A, lane 5) and the control glycosylation mutant ΔN11 (Fig. (Fig.4A,4A, lane 4) significant amounts of gp80SU and gp48TM glycoprotein subunits and small amounts of gp130Env precursor were detected at the cell surface. Mutant ΔN8 displayed the strongest transport defect, since no Env precursor or processed subunits could be detected at the cell surface (Fig. (Fig.4A,4A, lane 1), although total protein expression was only slightly reduced compared to the other samples (Fig. (Fig.4B,4B, lane 1). For mutant ΔN15 only small amounts of unprocessed gp130Env precursor were detectable at the cell surface (Fig. (Fig.4A,4A, lane 3), whereas mutant ΔN13 showed cell surface expression of both precursor and processed subunits (Fig. (Fig.4A,4A, lane 2); however, the relative amount of precursor compared to the processed subunits was significantly larger in comparison to the case for the wild-type protein.
Taken together, these data imply that the PFV Env is N glycosylated at 14 out of 15 sites. Furthermore, they suggest altered intracellular processing and transport upon individual inactivation of glycosylation at N391 (N8), N782 (N13), N808 (N14), and N833 (N15).
Unlike orthoretroviruses, FVs require their cognate glycoprotein for particle release into the supernatant (reviewed in reference 18). To determine whether certain N-glycosylation site mutants fail to support PFV particle release, we biochemically characterized PFV particles purified by ultracentrifugation from supernatants of 293T cells cotransfected with the FV vector pMH118 and the individual Env expression constructs. The protein composition analysis of purified PFV particles by Western blotting using antisera specific for PFV Gag, PFV Env LP, and Env SU subunits is shown in Fig. Fig.3.3. No particle-associated Gag or Env proteins were detected for mutant ΔN8 (Fig. 3A to C, lane 8). In particle preparation of cells cotransfected with mutant ΔN13 or ΔN15, only small amounts of particle-associated Gag and Env proteins were detected (Fig. (Fig.33 A to C, lanes 13 and 15). Mutants ΔN2, ΔN12, and ΔN14 (Fig. 3A to C, lanes 2, 12, and 14) displayed a somewhat reduced physical particle release, whereas all other mutants revealed no obvious phenotype and showed a particle release comparable to that of wild-type PFV Env (Fig. 3A to C, lanes 1, 3 to 7, and 9 to 11). Furthermore, in the Western blot analysis a mobility shift in FV particle-associated LP cleavage products was observed only for the ΔN2 mutant (Fig. (Fig.3B,3B, lower panel, lane 2), whereas a shift in the SU subunit was detectable for mutants ΔN3 to ΔN7 (Fig. (Fig.3C,3C, lanes 3 to 7) and ΔN9 to ΔN12 (Fig. (Fig.3C,3C, lanes 9 to 12). The higher-molecular-weight LP cleavage products gp28LP and gp38LP (Fig. (Fig.3B,3B, lower panel) that we observed previously are generated not by alternative proteolytic processing or differential glycosylation but by additional posttranslational modification of the gp18LP cleavage product (4; N. Stanke and D. Lindemann, unpublished results).
Taken together, these data demonstrate that the PFV LP subunit is N glycosylated at a single site at N109; that the SU subunit is N glycosylated at 10 sites at N141, N183, N286, N311, N346, N391, N405, N423, N527 and N556; and that the TM subunit is N glycosylated at 3 sites at N782, N808, and N833. Furthermore, they show that individual inactivation of glycosylation at a single site, N391 (N8), in SU and at all three sites, N782 (N13), N808 (N14), and N833 (N15), in TM results in the failure of these proteins to efficiently support PFV particle release.
To determine the infectivity of PFV particles containing the mutant glycoproteins, supernatants of the 293T cells cotransfected with pMH118 described above were further analyzed in an EGFP marker gene transfer assay on HT1080 cells. The results of this analysis are summarized in Fig. Fig.5A.5A. All mutants except three showed infectivities within a threefold range of that of wild-type PFV Env (Fig. (Fig.5A).5A). Mutants ΔN13 and ΔN15 showed 65-fold and 400-fold reductions in infectivity, respectively, whereas no infectivity could be detected for mutant ΔN8. To analyze the infectivities of selected mutants in more detail, viral particle preparations of mutants ΔN8, ΔN11, ΔN13, and ΔN15 were titrated on HT1080 cells by using a more sensitive LacZ marker gene transfer assay (Fig. (Fig.5B).5B). Supernatants of mutant ΔN8 showed an infectivity that was at least 4 × 104-fold lower than that of wild-type FV Env-containing samples, whereas the infectivities of mutants ΔN13 and ΔN15 were reduced 600- to 700-fold. Similar to what was observed in the EGFP transfer assay, mutant ΔN11 showed an infectivity comparable to that of the wild type.
Taken together, these data demonstrate a correlation between the physical particle release and infectivity of specific supernatants. This suggests that those N-glycosylation mutants that support PFV particle release display normal envelope functions such as receptor binding, fusion, and entry into target cells.
Next we examined whether the lack of glycosylation at N8, N13, and N15 per se is responsible for the strong phenotypes observed for corresponding mutants or whether the regions surrounding these sites influence Env function, similar to what was observed for Friend murine leukemia virus (12, 15). Therefore, we generated additional mutations, either abolishing glycosylation by altering the third amino acid in the N-X-S/T signal sequence (ΔN8.1, ΔN13.1, and ΔN15.1) or introducing a different amino acid at the variable second position (ΔN8.2, ΔN13.2, and ΔN15.2) that should not interfere with N glycosylation (Fig. (Fig.1B).1B). Mutant ΔN11, lacking one N-glycosylation site but showing no obvious phenotype, was used as an additional control in the following experiments. Western blot analysis revealed a mobility shift of the gp130Env precursors in cell lysates of mutants ΔN8, ΔN8.1, ΔN13, ΔN13.1, ΔN15, ΔN15.1, and ΔN11 (Fig. (Fig.6A,6A, lanes 2, 3, 5, 6, 8, 9, and 11) compared to wild-type protein (Fig. (Fig.6A,6A, lanes 1 and 12). In contrast, no mobility shift was observed for mutants ΔN8.2, ΔN13.2, and ΔN15.2 (Fig. (Fig.6A,6A, lanes 4, 7, and 10), indicating that unlike the other mutants, they still contained N-linked carbohydrate chains at the indicated glycosylation sites. LP processing of mutants ΔN8.1 and ΔN13.1 was similar to that of mutants ΔN8 and ΔN13, respectively (Fig. (Fig.6C,6C, lanes 2, 3, 5, and 6), whereas that of mutant ΔN15.1 was significantly improved compared to that of mutant ΔN15 (Fig. (Fig.6C,6C, lanes 8 and 9). The other mutants, ΔN8.2, ΔN13.2, ΔN15.2, and ΔN11, showed LP processing similar to that of the wild type (Fig. (Fig.6C,6C, lanes 4, 7, 10, 11, and 12). Subsequently, the infectivity of the supernatants of the transfected 293T cells was examined using the EGFP transfer assay (Fig. (Fig.6E).6E). Mutant ΔN8.1 showed a similar infectivity as mutant ΔN8, and mutant ΔN13.1 showed a marginal eightfold increase compared to mutant ΔN13, whereas mutant ΔN15.1 displayed an infectivity that was 70-fold higher than that of mutant ΔN15 but still at only about 4% of the level of the wild type. In contrast, mutants ΔN8.2, ΔN13.2, and ΔN15.2 showed dramatically increased infectivities. Similar to the control mutant ΔN11, mutants ΔN8.2 and ΔN13.2 reached levels of 70 to 80% of wild type, whereas the infectivity of ΔN15.2 was only about 20% of wild type. Analysis of particle release revealed in general a good correlation to the infection analysis (Fig. 6D and E). Particle secretion of mutants ΔN8.2, ΔN13.2, ΔN15.2, and ΔN11 into the supernatant was indistinguishable from that of the wild type (Fig. (Fig.6D,6D, lanes 4, 7, 10, 11, and 12). No particle release could be detected for mutants ΔN8 and ΔN8.1 (Fig. (Fig.6D,6D, lanes 2 and 3), whereas that of mutants ΔN13 and ΔN15 (Fig. (Fig.6D,6D, lanes 5 and 8) was strongly reduced. In contrast, particle release of mutants ΔN13.1 and ΔN15.1 (Fig. (Fig.6D,6D, lanes 6 and 9) was increased but still lower than that of the wild type (Fig. (Fig.6D,6D, lane 12). Interestingly, significantly higher amounts of secreted particles of mutant ΔN13.1 compared to mutant ΔN13 could be detected (Fig. (Fig.6D,6D, lanes 5 and 6), although the infectivities of these mutants were only marginally different (Fig. (Fig.6E6E).
Taken together, these data suggest that the N glycosylation at N391 (N8) in SU and N782 (N13) in TM by itself is important for PFV Env function, whereas at N833 (N15) a combination of N glycosylation and surrounding sequences seems to influence glycoprotein function.
Several viral envelope mutations have been reported to have temperature-sensitive phenotypes (reviewed in reference 2). In particular, mutation analysis of Moloney MuLV N-glycosylation sites identified one mutant that displayed a cell type-specific, temperature-sensitive phenotype (5). Therefore, we examined whether some of the PFV Env N-glycosylation mutants showing a decreased infectivity or no infectivity at 37°C could be rescued at lower temperatures. Mutant viruses were produced in parallel by transient transfection of 293T cells at either 37°C or 30°C. Subsequently, target cells were incubated for 6 h with viral supernatants and for an additional 16 h at the corresponding temperature of virus production before shifting of all cells to 37°C and analysis by flow cytometry 72 h after infection. The analysis (Fig. (Fig.7)7) revealed a general two- to fivefold increase in the relative infectivity of most mutants compared to the wild type, although the sensitivity of the assay was somewhat reduced at 30°C due to generally lower infection efficiencies. Nevertheless, mutant ΔN15 showed a 40-fold increase, and infectivity could be rescued from background levels to 20% of that of the wild type (Fig. (Fig.7).7). The corresponding mutant ΔN15.1, displaying the highest infection efficiency at 37°C of this set of mutants with reduced infectivity, showed an eightfold increase in infectivity, to levels comparable to that of the wild-type protein (Fig. (Fig.7).7). Mutants ΔN13 and ΔN13.1 showed only a slightly higher increase in infectivity compared to the mock-treated sample, whereas both ΔN8 mutants displayed infectivities at the detection limit of the assay. Thus, these data imply that only mutant ΔN15 displays a significant temperature-sensitive phenotype that can partially be rescued at 30°C.
The PFV envelope glycoprotein displays a highly unusual biosynthesis. Similar to other retroviral glycoprotein precursor proteins, it is proteolytically processed into mature subunits during its transport to the cell surface. However, unlike as reported for other retroviral glycoproteins, two cleavage events, which result in the generation of LP, SU, and TM subunits, take place only late during intracellular transport, most likely by furin-like proteases (4). In this study we analyzed the extent of N glycosylation of the different PFV Env subunits and the function of individual N-glycosylation sites for viral replication. We demonstrate that all potential N-glycosylation sites except one that are present in the PFV Env protein sequence are indeed modified. The LP subunit contains only one carbohydrate chain, whereas the TM and SU subunits have oligosaccharides attached at 3 and 10 sites, respectively. Interestingly, the pattern of digestion by different glycosidases suggests that the LP and TM subunits contain only carbohydrate chains of the high-mannose or hybrid type, whereas the majority of the oligosaccharides attached to the SU subunit are of the complex type. Recent results on the binding of HIV and Ebola virus GPs to the C-type lectins DC-SIGN and DC-SIGNR suggest a role for high-mannose-type carbohydrate structures in determining the specificity of the interaction (16). Therefore, it might be possible that the specific carbohydrate composition of individual PFV Env subunits, in particular that of SU, may influence binding to target cells and interaction with cellular receptors.
Sequence comparison with glycoproteins from other FV species revealed the conservation of 2 out of 10 N-glycosylation sites in SU and all three sites in TM (data not shown). Mutation analysis of the individual glycosylation sites demonstrated that all the nonconserved sites are dispensable for PFV envelope function in vitro when inactivated. Interestingly, only mutation of some of the conserved carbohydrate attachment sites resulted in detectable phenotypes in our analysis, suggesting that these sites might be important for glycoprotein function of all FV isolates. Mutation of the first (N13) and third (N15) glycosylation site in the TM subunit, which are located in the unique cysteine-rich, prolonged central “sheets-and-loops” region of unknown function (30), resulted in greatly diminished particle release. In addition, both mutants showed a reduced cell-associated LP processing and cell surface expression, indicating that the intracellular transport of these proteins is less efficient than wild type. Since the extracellular domain of retroviral TM subunits normally harbors the oligomerization domains and correct oligomerization of glycoproteins is essential for proper intracellular transport, the results might indicate that these two mutations affect oligomerization, although this was not addressed directly in our analysis. The results with alternative mutants affected in both glycosylation sites (ΔN13.1 and ΔN15.1) or neighboring sequences (ΔN13.2 and ΔN15.2) suggest that in case of N13 the glycosylation at this position itself is important for Env function. For N15, alternative inhibition of glycosylation (ΔN15.1) led to a significant improvement of particle release and infectivity, and an amino acid change not affecting glycosylation (ΔN15.2) resulted in a slight reduction in the infectivity of the released particles. Furthermore, analysis of the temperature dependence of the phenotypes observed for individual mutants suggests that only the ΔN15 mutant displays a significant temperature-sensitive phenotype. Therefore, in this case the sequence surrounding N15 rather than the attached carbohydrate chain, or a perhaps a mixture of both, seems to be important for normal glycoprotein function. This resembles the case for Friend ecotropic murine leukemia virus Env protein, where only one of eight signals for N-linked glycan attachment is critical for envelope function (12). Subsequently, a more detailed mutational analysis of this MuLV N-linked glycosylation site and surrounding sequences revealed that N glycosylation per se is not required for MuLV Env function but that the region surrounding this glycosylation site mediates envelope folding and the stability of the interaction between SU and TM (15).
In the PFV Env SU subunit, only inactivation of the conserved carbohydrate attachment site N8 impaired Env function. Similar to what was observed for the ΔN13 and ΔN15 mutants, the ΔN8 mutant protein was not properly transported intracellularly, as indicated by the lack of LP cleavage of the mutant precursor protein and cell surface expression. However, unlike the two mutants with mutations in TM, neither ΔN8 nor the alternative glycosylation site mutant ΔN8.1 showed any detectable infectivity or particle release. In contrast, the ΔN8.2 mutant, retaining glycosylation but having the variable second amino acid of the N-X-S/T glycosylation site signal sequence mutated, displayed a wild-type phenotype. This strongly suggests that the carbohydrate chain itself serves an important role in PFV Env function, most likely in folding, as both glycosylation mutants showed no proteolytic processing of the precursor protein. However, since this glycosylation site is conserved in the SU subunits of FV isolates from different species, it is possible that it may be involved in other Env function, such as interaction with the cellular receptor, although this apparently cannot be addressed in the context of the full-length PFV Env protein due to the transport defect of these mutants. However, preliminary data obtained using PFV Env immunoadhesins indicate that glycosylation site N8 or surrounding sequences might be important for receptor interaction (A. Duda and D. Lindemann, unpublished observations). Since the digestion pattern of the SU subunit with different glycosidases suggests that the majority of the carbohydrate chains in SU are of the complex type, it would be interesting to know which sites are of the high-mannose or hybrid type and in particular what type of carbohydrate chain is added to N8.
This work was supported by grants from the DFG (Li621/2-1, Li621/2-3, and Li621/3-1) and the BMBF (01ZZ0102) to D.L.