Characterization of Glycosylated Gag Expressed ... - Journal of Virology

JOURNAL OF VIROLOGY, July 1997, p. 5355–5360 0022-538X/97/$04.0010 Copyright © 1997, American Society for Microbiology

Vol. 71, No. 7

Characterization of Glycosylated Gag Expressed by a Neurovirulent Murine Leukemia Virus: Identification of Differences in Processing In Vitro and In Vivo RYUICHI FUJISAWA, FRANK J. MCATEE, JOSEPH H. ZIRBEL,

AND

JOHN L. PORTIS*

Laboratory of Persistent Viral Diseases, Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, Hamilton, Montana 59840 Received 2 December 1996/Accepted 24 March 1997

The neuroinvasiveness of a chimeric murine retrovirus, CasFrKP (KP), is dependent on the expression of glycosylated Gag (gp85gag). This viral protein is the product of alternate translation initiation 88 codons upstream of and in frame with the initiation codon of pr65gag, the precursor of the viral core proteins. Although expression of glycosylated Gag affects virus spread in the spleen, it appears not to affect virus spread in vitro in fibroblast cell lines (J. L. Portis et al., J. Virol. 68:3879–3887, 1994). The differential effects of this protein in vitro and in vivo have not been explained, and its function is unknown. We have here compared the in vitro processing of this molecule with that expressed in spleens of infected mice. In vitro, gp85gag was cleaved near the middle of the molecule, releasing the C-terminal half (containing capsid and nucleocapsid domains of pr65gag) as a secreted glycoprotein. The N-terminal half of the protein was associated with the plasma membrane as a ;55-kDa glycoprotein bearing the matrix domain of pr65gag as well as the N-terminal 88 residue L domain. This processing scheme was also observed in vivo, although two differences were seen. There were differences in N-linked glycosylation of the secreted form of the protein expressed in the spleen. In addition, whereas the membrane-associated species assumed the orientation of a type II integral membrane protein (Ncyto Cexo) in fibroblasts in vitro, a subpopulation of spleen cells was detected in which the N terminus of the protein was exposed at the cell surface. These results suggest that the differential effects of glycosylated Gag expression in vivo and in vitro may be related to differences in posttranslational processing of the protein. Unlike pr65gag, which is a cytosolic protein, pr75gag is synthesized into the vesicular compartment, where it becomes glycosylated, cleaved by a cellular protease, and displayed at the plasma membrane. The protein is not thought to be incorporated into virions (2, 12, 33). Based on computer predictions (29) as well as immunofluorescence analysis (33), glycosylated Gag is a type II integral membrane protein (Ncyto Cexo) with a signal/anchor sequence located near the junction of the L and MA domains. In addition to cell surface species, secreted forms of glycosylated Gag have been found in culture fluid in vitro (12, 30) as well as in the serum in vivo (1). In the case of Moloney MuLV, the secreted species appear to have an affinity for extracellular matrices (14). Although glycosylated Gag is highly immunogenic and appears to contain important target epitopes for antiviral immunity in mice (5, 19), its function in the virus life cycle undoubtedly accounts for its conservation in replication competent MuLV and feline leukemia virus. Interruption of expression of glycosylated Gag has demonstrated that the protein facilitates virus spread (8, 36). This effect, in our hands, is demonstrable only in certain circumstances. Working with the neurovirulent MuLV CasFrKP (KP), a derivative of the wild mouse virus CasBrE, we observed that the kinetics of spread in vitro in fibroblast cell lines appeared unaffected by the presence or absence of glycosylated Gag (36). In contrast, mutants lacking the protein spread more slowly than wild-type virus in vivo at least in part due to slower spread within the spleen (8). Furthermore, unlike wild-type virus, these mutants did not spread to the brain and failed to induce neurologic disease. We have shown previously that neuroinvasiveness is strongly influenced by the kinetics of spread of virus in peripheral nonneuronal tissues during the first few days after neonatal inoculation (10). Thus, the role of glycosylated Gag in neuroinvasiveness could

Murine leukemia viruses (MuLVs) are type C retroviruses which consist of enveloped particles with a nucleoprotein core. The genetic information for these structural elements is contained in the three retroviral genes (gag, pol, and env). MuLVs are considered to be simple retroviruses which, unlike the lentiviruses, lack additional regulatory and accessory proteins. Accessory proteins are dispensable for virus replication but modulate functions at virtually every stage of the virus life cycle and are often required for expression of virulence. One protein of MuLV, glycosylated Gag, probably should be considered to be an accessory protein since it also is dispensable for virus replication (8, 17, 36, 41), but it is an important virulence determinant (8, 36). Glycosylated Gag was found originally at the surface of murine leukemia cells infected with Gross leukemia virus and has been shown to contain the Gross cell surface antigen (22, 44). It is now apparent that this protein is a universal feature of MuLV-infected cells (12, 16, 40) as well as feline leukemia virus-infected cells (30). The glycosylated Gag polyprotein is synthesized and processed independently of pr65gag, which is the precursor polyprotein of the internal virion proteins. In MuLV, the components of pr65gag are organized in the order NH2-matrix (MA)-pp12gag-capsid (CA)-nucleocapsid (NC) and are cleaved from the precursor by the viral protease during virus assembly. Translation of glycosylated Gag is initiated upstream of and in frame with that for pr65gag (38), yielding a protein (pr75gag) with a unique N terminus of 8 to 10 kDa (called the L domain) (12, 13, 39) but also carrying all of the components of pr65gag. * Corresponding author. Mailing address: Rocky Mountain Laboratories, 903 S. 4th St., Hamilton, MT 59840. Phone: (406) 363-9339. Fax: (406) 363-9204. E-mail: [email protected]. 5355

5356

FUJISAWA ET AL.

be an indirect consequence of its effect on virus spread in the periphery, although a more direct effect cannot be ruled out at this time. The differential effects of glycosylated Gag in vivo and in vitro suggested the possibility that expression and/or posttranslational modification of the protein may be different in these two environments. To date, however, the analyses of glycosylated Gag expression have been carried out on either fibroblastic or leukemia cell lines in vitro. Here we have compared the processing of the protein expressed in fibroblasts with that expressed in the spleens of infected mice. Expression of the glycosylated Gag precursor as well as its proteolytic cleavage appeared similar in vitro and in vivo, although differences in glycosylation were noted. However, whereas the L domain exhibited strictly a cytosolic orientation in fibroblasts in vitro, it was displayed at the cell surface on a subpopulation of spleen cells in vivo, suggesting an alternate orientation in the membrane. MATERIALS AND METHODS Mice, viruses, and inoculations. All mice used in this study were IRW (inbred Rocky Mountain White) mice bred and raised at Rocky Mountain Laboratories. These mice are highly susceptible to central nervous system disease induced by a variety of neurovirulent retroviruses (34). Mice were inoculated intraperitoneally as neonates (1 to 2 days postnatally) with 30 ml of the virus stocks containing 6 3 104 to 9 3 104 focus-forming units of infectivity. Virus stocks were assayed for infectivity by using a focal immunoassay described previously (9). At 12 to 14 days postinoculation, mice were killed by exsanguination under methoxyflurane inhalation anesthesia and spleens were removed for analysis (see below). Two viruses were used in this study. CasFrKP (KP) is a chimeric virus consisting of the viral genome of CasBrE clone 15-1 (34) with the ;0.5-kb segment of the leader sequence between KpnI in the R region of the long terminal repeat and a PstI site upstream of the start codon for pr65gag from Friend MuLV clone FB29 (43). CasFrKP[2314] (KPgg2) is identical to KP except for two point mutations in the optimal consensus sequence described by Kozak (20) around the CTG initiation codon for glycosylated Gag, effectively knocking out expression of this protein (35, 36). KP induces neurologic disease with an incubation period of 18 to 23 days, whereas KPgg2 is nonneurovirulent (36). The virus stocks were prepared as described previously (34). KP and KPgg2 virus stocks contained 2.1 3 106 and 3.1 3 106 focus-forming units/ml, respectively. Sample preparation for immunoblot analysis. Mus dunni cells were infected with KP or KPgg2 virus stock at a multiplicity of infection of ;20 in the presence of 8 mg of Polybrene per ml. The cells were grown in RPMI 1640 plus 10% fetal calf serum (FCS) and were passaged at least three times at a ratio of 1/10 to ensure confluent infection before each experiment was carried out. Spleens were harvested from mice at 12 to 14 days postinoculation, and cell suspensions were prepared by gentle teasing. The technique used for detection of cell surface proteins was essentially as described by Le Bivic et al. (21), using NHS-SS-biotin (Pierce). Briefly, M. dunni cells were labeled in situ after having reached near confluence in 25-cm2 flasks. Spleen cells were labeled in suspension. Cells were first washed with Dulbecco’s phosphate-buffered saline (DPBS) without added protein and incubated 30 min with 0.5 mg of NHS-SS-biotin per ml in DPBS on ice. Cells were then washed once with DPBS, once with Dulbecco’s minimal essential medium, and once again with DPBS, after which they were lysed with 0.01 M Tris-Cl–0.15 M NaCl–0.001 M EDTA containing 0.5% Nonidet P-40 (NP-40) (lysis buffer). To half of the lysate was added sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis sample buffer containing SDS and 2-mercaptoethanol (final concentrations, 2 and 5% respectively), and the samples were boiled (designated total lysate). The other half was incubated with 50 ml of packed immobilized Streptavidin (Ultralink; Pierce) to purify biotinylated proteins. Tubes were rotated at 4°C for 1 h. After washing with lysis buffer three times, biotinylated proteins were released by boiling in sample buffer (surface fraction). For detection of secreted species of glycosylated Gag, culture supernatants from M. dunni cells or spleen cells were analyzed. Twenty-four-hour supernatants were collected from confluent monolayers of M. dunni cells or spleen cells cultured at a concentration of 2.5 3 107/ml in RPMI 1640 plus 10% FCS. Supernatants were centrifuged at low speed to remove insoluble material. Then 0.5 ml of each sample was centrifuged at 3,000 3 g at 4°C for 30 min in a Centricon 100 (Amicon) to filter out virions. Samples were then boiled in sample buffer. PNGase F digestions. N-glycosidase F (PNGase F) digestions were performed on the samples prepared as described above. PNGase F (500 U; New England Biolabs) was added to 5-ml aliquots of each sample in a final volume of 20 ml containing 0.1% NP-40 and 0.05 M NaPO4 (pH 7.4) (final concentrations), and digestion was performed at 37°C for 2 h. After incubation, sample buffer was again added and the samples boiled prior to electrophoresis.

J. VIROL. Immunoblot analysis. The procedure was based on that described in detail by Lynch and Portis (26). Samples (;105 cell equivalents per track) were analyzed by electrophoresis in SDS–9% polyacrylamide gels and transferred to Immobilon-P membranes (Millipore). After blocking with 10% skim milk at 4°C overnight, membranes were incubated with one of the following antibodies: rabbit anti-CA antiserum (26), monoclonal antibody 690 (anti-MA) (27), and rabbit anti-peptide 4210 antiserum (anti-L; specific for the L domain of glycosylated Gag) (35). The blots were developed with species specific horseradish peroxidase-conjugated antibodies (Bio-Rad) and visualized with the ECL system (Amersham) as previously described (11). Immunofluorescent staining with anti-L antiserum. M. dunni cells grown on coverslips, either live or after fixation with 80% ice-cold acetone for 10 min, were washed with phosphate-buffered balanced salt solution (34) plus 2% FCS and incubated at 4°C for 1 h with rabbit anti-L antiserum. Cells were then washed with phosphate-buffered balanced salt solution plus 2% FCS and incubated with a 1/500 dilution of fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit immunoglobulin (Ig) (Cappel) at 4°C for 30 min. Coverslips were examined with a Nikon Microphot fluorescence microscope, and photomicrographs taken with Fuji Provia 400 film. Peptide absorption of anti-L antiserum. Peptide 4210 consists of 18 residues at the extreme N terminus of the L domain of glycosylated Gag and was used as the immunogen to produce the rabbit anti-L-domain antiserum (35). One milligram of peptide 4210 per ml of 50 mM Tris–5 mM EDTA (pH 8.5) was coupled to 1 ml of SulfoLink coupling gel (Pierce) through the sulfhydryl group of its amino-terminal cysteine residue as instructed by the manufacturer. The coupling efficiency was 96%, estimated by using Ellman’s reagent (Pierce). One milliliter of neat anti-L antiserum was applied to the column and eluted with ;1 column volume of phosphate-buffered saline. The protein concentration of the eluate was determined (Bio-Rad protein assay reagent) and equilibrated with that of the starting material. The success of the absorption was assessed by showing that the absorbed antiserum no longer stained acetone-fixed KP-infected M. dunni cells. Flow cytometry. Flow cytometry was carried out on both spleen cell suspensions and suspensions of M. dunni cells. The latter were grown in tissue culture flasks, from which they were removed with saline-trypsin-EDTA. Infected and uninfected live cells were stained as described previously (27), using the following primary antibodies: biotinylated monoclonal antibody 667 (anti-SU) (27), biotinylated monoclonal antibody 34 (anti-Gag matrix protein) (7), rat monoclonal antibody R18-7 (anti-capsid protein) (6), and rabbit anti-peptide 4210 (anti-L domain of glycosylated Gag) (35). Biotinylated primary antibodies were detected with FITC-streptavidin (GIBCO), the rat monoclonal antibody was detected with mouse IgG-absorbed FITC-conjugated anti-rat Ig (Biomeda), and the rabbit antibody was detected with FITC-conjugated anti-rabbit Ig (Sigma). Dead cells were stained with propidium iodide and gated out. Controls for specificity included staining uninfected cells and infected cells in the absence of primary antibodies. In addition, peptide 4210-absorbed anti-L antiserum (see above) was used as a control for the specificity of the anti-L antiserum. Cells were analyzed with a FACStar fluorescence-activated cell sorter (Becton Dickinson), and the data were collected in the log mode. Ten thousand cells were analyzed per virus, and the experiments were repeated at least five times.

RESULTS Characterization of cell-associated species of glycosylated Gag expressed by M. dunni fibroblasts and spleen cells. Glycosylated Gag is expressed as an 85-kDa precursor glycoprotein, the processing of which leads to cell surface expression as well as secreted forms. There is, however, no general agreement on the nature of these species, although it appears that both the full-length protein as well as proteolytic cleavage products may be displayed at the cell surface. M. dunni fibroblasts and spleen cells infected with KP were incubated with NHS-SS-biotin to label cell surface proteins. The cells were lysed with NP-40 (total lysate), and the biotinylated proteins were isolated with immobilized streptavidin (surface fraction). Samples were analyzed by immunoblotting with anti-Gag antibodies specific for the capsid protein and the L domain of glycosylated Gag (Fig. 1). Both uninfected cells and cells infected with the KPgg2 virus were used as controls (Fig. 1). Anti-CA and anti-L reacted with the gp85gag precursor in both M. dunni and spleen cells; this protein, however, was not detected at the cell surface of either M. dunni or spleen cells (Fig. 1). Instead a major species of 50 to 55 kDa was detected along with several minor species of ,40 kDa. These proteins were reactive only with anti-L, not with anti-CA. No reactivity of cell surface species with anti-CA was observed with either cell

VOL. 71, 1997

PROCESSING OF GLYCOSYLATED Gag

5357

terize the secreted component(s) of glycosylated Gag, we analyzed culture supernatants. We first present the results for M. dunni fibroblasts (Fig. 2). A protein of 45 to 50 kDa was detected with anti-CA in the culture supernatant of KP-infected cells but not in the supernatant from KPgg2-infected cells (Fig. 2A). This protein was nonreactive with antibody specific for the L domain of glycosylated Gag (Fig. 2A). The size of the secreted form of glycosylated Gag (;45 kDa) approximated that of the cell surface species (Fig. 1). Indeed, after PNGase F treatment to remove N-linked sugars, the protein backbones of the cell surface and secreted forms were comparable in molecular size (Fig. 2B). Together these results suggested that the precursor gp85gag was cleaved near the middle of the molecule, releasing the C-terminal portion into the secretory pathway and retaining the N-terminal portion in the plasma membrane (shown schematically in Fig. 2C). The secreted protein identified in the supernatant from KPinfected spleen cells was of a higher molecular size than that secreted by M. dunni cells (Fig. 3). However, after removal of N-linked sugars with PNGase F, the sizes of the respective protein backbones were similar, indicating that the size difference was due to differential glycosylation. Because of the similarity in size of the secreted as well as the cell surface species (Fig. 3) expressed by M. dunni and spleen cells, it can be inferred that the proteolytic processing of the precursor of glycosylated Gag, gp85gag, was also similar. Exposure of the L domain of glycosylated Gag at the surface in a subpopulation of spleen cells. Computer predictions (29) indicate that glycosylated Gag is a type II integral membrane protein with its N-terminal L domain comprising the cytoplasmic tail (Ncyto Cexo). This model was tested by flow cytometry FIG. 1. Immunoblot analysis of Gag proteins produced by M. dunni and spleen cells infected with KP. Uninfected and KPgg2-infected cells were included as controls. Chronically infected M. dunni cells or spleen cell suspensions prepared 12 days after neonatal intraperitoneal inoculation of KP or KPgg2 were surface labeled with NHS-SS-biotin (Pierce). The cells were lysed with NP-40 (Total), and the biotinylated proteins were isolated with immobilized streptavidin (Surface). Samples were separated on 9% polyacrylamide gels, electroblotted, and probed with either rabbit anti-CA (p30gag) antiserum or rabbit anti-L antiserum (anti-peptide 4210 [35]). Note that both anti-CA and anti-L detected gp85gag in the total lysate of KP-infected M. dunni and spleen cells. In the surface fraction of either M. dunni or spleen cells, only anti-L detected glycosylated Gag species, and these ranged from 50 to 55 kDa, with several minor species of smaller size (denoted by brackets). No glycosylated Gag forms reactive with anti-CA were detected in the surface fraction. The minimal reactivity of anti-CA with pr65gag and pr55gag in the surface fractions of spleen cells could represent limited access of the NHS-SS-biotin to cytosolic proteins, likely a consequence of some level of cell death.

sample (Fig. 1). We also performed immunoblotting with an anti-Gag monoclonal antibody 690 (27) specific for MA protein; the blots were similar in appearance to the anti-L blots (not shown). Collectively these results indicated that the glycosylated Gag precursor, gp85gag, was cleaved intracellularly, the N-terminal portion containing the L domain as well as MA determinants being displayed at the plasma membrane. Fibroblasts and spleen cells appeared to generate similar cell surface forms of glycosylated Gag (Fig. 1; see also Fig. 3). It should be noted that a ;100-kDa protein was detected with anti-CA in the total lysate of KP-infected M. dunni and spleen cells but was not observed in the biotinylated surface preparations (Fig. 1). This protein was not seen in the total lysates of KPgg2infected cells, indicating that it represented a form of glycosylated Gag. Though not seen in Fig. 1, longer exposures clearly showed that this band also reacted with anti-L. Characterization of secreted species of glycosylated Gag produced by M. dunni fibroblasts and spleen cells. To charac-

FIG. 2. (A) Immunoblot analyses of Gag proteins secreted into culture medium from M. dunni fibroblasts chronically infected with KP. A supernatant from cells chronically infected with KPgg2 was included as a control. Supernatants were passed through Centricon 100 filters (Amicon) to remove virus particles. Blots were probed with anti-CA or anti-L antiserum. Anti-CA detected a 45- to 50-kDa band in the KP supernatant which was not seen in the KPgg2 sample. This protein was not reactive with anti-L. The band at ;70 kDa was also detected in the KPgg2 supernatant (not shown), indicating that it did not represent glycosylated Gag. (B) Comparison of the protein backbones of the secreted and cell surface species of glycosylated Gag expressed by KP-infected M. dunni cells. The secreted sample was the same sample of supernatant fluid from KP-infected M. dunni cells shown in panel A. Cell surface species were biotinylated proteins isolated with immobilized streptavidin (i.e., same as surface fraction in Fig. 1). Samples were either treated or not treated with PNGase F to remove N-linked sugars (see Materials and Methods) and were separated on 9% polyacrylamide gels. Blots were probed with a mixture of anti-CA and anti-L antisera so as to detect both secreted and cell surface species in the same blot. Note that after PNGase F treatment, secreted and cell surface proteins exhibited identical molecular sizes (;40 kDa). (C) Based on these immunoblot results, a schematic showing the putative location of the proteolytic cleavage site is shown. Consensus sequences for N-linked glycosylation (18) are indicated by tree-like symbols. SA, signal/anchor sequence; cyto, cytosol; exo, lumen of the endoplasmic reticulum/ Golgi complex and external surface of the plasma membrane.

5358

FUJISAWA ET AL.

FIG. 3. Comparison of the secreted and cell surface species of glycosylated Gag expressed by M. dunni and spleen cells infected with KP, before and after removal of N-linked sugars with PNGase F. The blots were probed with a mixture of anti-CA and anti-L antisera so as to detect both secreted and cell surface species in the same blot. (Top) Prior to PNGase F treatment (2PNGaseF), the secreted species expressed by spleen cells was slightly larger than that expressed by M. dunni cells. This difference was abolished by removal of N-linked sugars (1PNGaseF). (Bottom) Both before and after PNGase F treatment, the cell surface species of glycosylated Gag expressed by M. dunni and spleen cells were similar in size. The doublets noted most strikingly in the secreted proteins after PNGase F treatment could be a consequence of incomplete deglycosylation, although it is also possible that the upper bands were derived from the ;100-kDa glycosylated Gag species detected in the total lysates in Fig. 1 (see Discussion).

on chronically infected M. dunni fibroblasts and spleen cells, using antibodies directed to the L domain as well as anti-MA, which is predicted to be oriented extracellularly (Fig. 2C). In M. dunni cells infected with KP, MA was clearly demonstrable

J. VIROL.

at the cell surface, and as predicted, there was no evidence for exposure of the L domain (Fig. 4A). That the L domain was demonstrable intracellularly is shown in Fig. 4B, in which staining was observed only after acetone fixation. As with M. dunni cells, anti-MA stained a major population of spleen cells infected with KP (Fig. 4A). Consistent with the immunoblot analyses of cell surface proteins (Fig. 1), there was no evidence of anti-CA staining (not shown). Surprisingly, however, a subpopulation of spleen cells stained with anti-L, suggesting that the L domain was exposed at the plasma membrane. It should be noted that only live cells were analyzed, since dead cells were gated out after propidium iodide uptake. However, this does not rule out the possibility that the anti-L staining was nonspecific. Although KPgg2-infected spleen cells are an appropriate negative control for artifacts which might arise simply as a function of virus infection, we also absorbed the anti-L antiserum by affinity chromatography with its cognate peptide. The capacity of the antiserum to stain live KPinfected spleen cells was completely absorbed by this procedure (not shown). These results, therefore, strongly suggest that in a subpopulation of spleen cells, glycosylated Gag assumed an alternate orientation with the L domain exposed at the cell surface. DISCUSSION While glycosylated Gag has been studied extensively in the past and a variety of molecular forms have been identified, there appears to be no general consensus on the nature of the

FIG. 4. (A) The L domain of glycosylated Gag was exposed extracellularly on a subpopulation of KP-infected spleen cells but had a cytosolic orientation in KP-infected M. dunni cells. M. dunni cells were chronically infected with either KP or KPgg2 or were uninfected (N). Spleen cells were harvested from mice 12 days after neonatal intraperitoneal inoculation of KP or KPgg2 and from uninoculated age-matched mice. Cells were stained in suspension with anti-MA, anti-L-domain, and anti-CA (not shown). Dead cells were stained with propidium iodide and gated out (10,000 events). In M. dunni, the frequency of infection of both KP and KPgg2, as assessed by staining with anti-SU antibodies, was 100%. For spleen cells, the frequencies of infection were 84.1% for KP and 71.1% for KPgg2. In M. dunni cells, there was no evidence of extracellular exposure of the L domain at the cell surface (compare mean fluorescence [M.Fl.] for KP, KPgg2, and uninfected M. dunni cells). In contrast, anti-L revealed a population of cells in the KP-infected spleen with the L domain exposed extracellularly (P , 0.01 by unpaired Student’s t test of at least 27 mice per group, comparing mean fluorescence of KP with that of either KPgg2 or normal). Staining with anti-CA was consistently negative (not shown). (B) Although anti-L antiserum failed to detect the L domain of glycosylated Gag at the surface of live KP-infected M. dunni cells, staining was detectable in these cells in a paranuclear localization consistent with the Golgi complex after fixation with acetone. Thus, the L domain was accessible to antibody only when the cells were permeabilized.

VOL. 71, 1997

cell surface and secreted species expressed by MuLV-infected cells. For instance, Ledbetter et al. (23) found two glycosylated Gag species of 85 and 95 kDa expressed at the cell surface, while Pillemer et al. (33) found an additional 55-kDa species. In these studies, secreted forms were not observed. Both of these studies were carried out on AKR leukemia cells, which are clonal. Edwards and Fan (12), on the other hand, found two secreted forms of glycosylated Gag of 40 and 55 kDa in Moloney MuLV-infected NIH 3T3 cells, but in that study the nature of the cell surface species was not investigated. The focus of the present study was to gain some understanding of the fate of each of the proteolytic cleavage products of the glycosylated Gag precursor protein during the course of a virus infection. In view of the fact that glycosylated Gag functions to enhance virus spread in the mouse but appears not to function in mouse fibroblasts in vitro (36), we compared the processing of the protein in these two venues. The virus that we studied is a neurovirulent MuLV, KP, the neuropathogenesis of which is dependent on the expression of glycosylated Gag (35). Both in vitro and in vivo, glycosylated Gag was expressed as an 85-kDa precursor protein which was cleaved, presumably by a cellular protease, into two major species. The C-terminal fragment containing the CA domain of pr65gag was released as a secreted form into the culture medium. Based on the size of its protein backbone (;40 kDa), this protein likely also contained the C-terminal nucleocapsid domain of pr65gag. The N-terminal fragment remained cell associated and carried the L-domain and MA determinants. After enzymatic removal of the N-linked sugars, the two fragments were found to be nearly identical in size (;40 kDa), indicating that the cleavage site was near the middle of gp85gag precursor protein, probably within the N-terminal portion of the CA domain. Thus, for the virus KP, processing of the glycosylated Gag precursor protein gp85gag appeared to involve essentially one major cleavage event, and the cleavage site appeared to be identical whether the protein was expressed in fibroblasts or spleen cells. The fates of these products, however, appeared to be somewhat different depending on the host cell. Since it is not known which of these cleavage fragments is required for function of the protein, each will be discussed separately. The secreted C-terminal cleavage product produced by KPinfected spleen cells was slightly larger than that expressed by fibroblasts, the difference being a consequence of N-linked glycosylation. This is not surprising, as proteins expressed by different cell types can be differentially glycosylated (15, 42). Although the potential role of sugar moieties in protein folding and stabilization is clear (31), there are several situations in which the structures of the glycans themselves have been shown to affect function. Perhaps the best examples are the carbohydrate ligands for mammalian lectins (45), which are involved in cell adherence and homing. There is evidence, for instance, that differential glycosylation of one of the ligands of L-selectin, CD34, may regulate the avidity of the adhesive interaction (3). Although the ligand specificity of the selectins is generally dependent on O-linked sugars, N-linked glycans have been shown to affect the adhesive interaction with Pselectin (24) as well as with neural cell adhesion molecules (28). It is of interest in this regard that a secreted component of glycosylated Gag appears to bind avidly to the extracellular matrix (14), an observation which we have confirmed for the virus KP (not shown). Thus, it is conceivable that the nature of the sugar residues expressed on the secreted form of glycosylated Gag could, by virtue of its interaction with the extracellular matrix, affect the migration of virus-infected cells. Such effects on cell migration might influence the dissemination of

PROCESSING OF GLYCOSYLATED Gag

5359

virus-infected cells both within the spleen as well as out of the spleen and into the tissues. Differences were also observed in the N-terminal cleavage fragment expressed by infected fibroblasts and spleen cells. Although no demonstrable differences in N-linked glycosylation were observed (Fig. 3), there appeared to be differences in the topology of the protein in the plasma membrane (Fig. 4). An epitope located at the extreme N terminus of the L domain of protein (35) was expressed at the cell surface in the spleen. This epitope was detected only intracellularly in M. dunni fibroblasts. The extracellular exposure of the L domain on spleen cells, however, does not prove conclusively that the protein had assumed a type I orientation (Nexo Ccyto). For instance, viral proteins released from infected cells might have been nonspecifically bound to the cell surface. As a control for nonspecific adsorption, however, we looked for the presence of viral capsid protein at the cell surface with anti-CA antiserum. There was no evidence for accessibility of this protein to either biotin (Fig. 1) or antibody by flow cytometry (not shown). Cumulatively, these results are most consistent with the L domain of the protein entering the lumen of the vesicular compartment in a subpopulation of spleen cells. This could occur cotranslationally, as was shown for mutants of the HN protein of paramyxoviruses (32), or posttranslationally, as has been demonstrated for the pre-S envelope protein of hepatitis B virus (4, 37) and the PrP protein (25). Posttranslational alterations in membrane orientation might exhibit cell type specificity, since it requires participation of cytosolic factors which, at least in the case of PrP, appear to be expressed in a cell-type specific fashion. In addition to the 85-kDa precursor depicted in Fig. 1, a ;100-kDa form of glycosylated Gag was also detected by immunoblot analysis in total cell lysates of KP-infected cells. The larger size of this protein appeared not to be a consequence of differential glycosylation, since it did not disappear after PNGase F digestion (not shown). Since this protein also expressed the L domain, it most likely represented a readthrough product of glycosylated Gag with a C-terminal extension into the coding sequence of the pol gene. Like the gp85gag species, this protein was not detected at the cell surface. Whether it contributed to the processed forms of glycosylated Gag reported here is not clear. We did observe a doublet after PNGase F digestion of the secreted proteins from both M. dunni and spleen cells (Fig. 3), the upper band of which would be consistent with a secreted form of glycosylated Gag carrying a C-terminal extension. It is of interest that previous mapping studies (36) identified a highly charged proline rich sequence near the C-terminal end of the L domain of glycosylated Gag, which influenced the kinetics of virus spread in vivo. That observation might suggest that it is the cell-associated component of glycosylated Gag which is required for function. Furthermore, the possibility now exists that this sequence is exposed at the cell surface in the spleen, which might allow for interaction with receptors on neighboring cells. Such interactions could, for example, enhance cell-to-cell spread of virus either directly or perhaps indirectly through induction of cell proliferation. The studies described herein were initially motivated by the observation that glycosylated Gag enhanced virus spread in vivo but not in vitro. We have found that the gp85gag precursor is cleaved into two fragments and have observed differences in the processing of each fragment during virus infection of spleen cells and cultured fibroblasts. Whether these differences are relevant to function must await further understanding of the functional domains of the protein.

5360

FUJISAWA ET AL. ACKNOWLEDGMENTS

We thank Diane Brooks for assistance with flow cytometry. We also thank Gary Hettrick and Robert Evans of the RML Graphics Department for the figure reproductions. REFERENCES 1. Aoki, T., E. A. Boyse, and L. J. Old. 1968. Wild-type Gross leukemia virus. I. Soluble antigen (GSA) in the plasma and tissues of infected mice. J. Natl. Cancer Inst. 41:89–96. 2. Aoki, T., E. A. Boyse, L. J. Old, E. DeHarven, U. Hammerling, and H. A. Wood. 1970. G (Gross) and H-2 cell-surface antigens: location on Gross leukemia cells by electron microscopy with visually labeled antibody. Proc. Natl. Acad. Sci. USA 65:569–576. 3. Baumhueter, S., N. Dybdal, C. Kyle, and L. A. Lasky. 1994. Global vascular expression of murine CD34, a sialomucin-like endothelial ligand for Lselectin. Blood 84:2554–2565. 4. Bruss, V., X. Lu, R. Thomssen, and W. H. Gerlich. 1994. Post-translational alterations in transmembrane topology of the hepatitis B virus large envelope protein. EMBO J. 13:2273–2279. 5. Chen, W., H. Qin, B. Chesebro, and M. A. Cheever. 1996. Identification of a gag-encoded cytotoxic T-lymphocyte epitope from FBL-3 leukemia shared by Friend, Moloney, and Rauscher murine leukemia virus-induced tumors. J. Virol. 70:7773–7782. 6. Chesebro, B., W. Britt, L. Evans, K. Wehrly, J. Nishio, and M. Cloyd. 1983. Characterization of monoclonal antibodies reactive with murine leukemia viruses: use in analysis of strains of Friend MCF and Friend ecotropic murine leukemia virus. Virology 127:134–148. 7. Chesebro, B., K. Wehrly, M. Cloyd, W. Britt, J. Portis, J. Collins, and J. Nishio. 1981. Characterization of mouse monoclonal antibodies specific for Friend murine leukemia virus-induced erythroleukemia cells: Friend-specific and FMR-specific antigens. Virology 112:131–144. 8. Corbin, A., A. C. Prats, J. L. Darlix, and M. Sitbon. 1994. A nonstructural gag-encoded glycoprotein precursor is necessary for efficient spreading and pathogenesis of murine leukemia viruses. J. Virol. 68:3857–3867. 9. Czub, M., S. Czub, F. J. McAtee, and J. L. Portis. 1991. Age-dependent resistance to murine retrovirus-induced spongiform neurodegeneration results from central nervous system-specific restriction of virus replication. J. Virol. 65:2539–2544. 10. Czub, M., F. J. McAtee, and J. L. Portis. 1992. Murine retrovirus-induced spongiform encephalomyelopathy: host and viral factors which determine the length of the incubation period. J. Virol. 66:3298–3305. 11. Czub, S., W. P. Lynch, M. Czub, and J. L. Portis. 1994. Kinetic analysis of spongiform neurodegenerative disease induced by a highly virulent murine retrovirus. Lab. Invest. 70:711–723. 12. Edwards, S. A., and H. Fan. 1979. Gag-related polyproteins of Moloney murine leukemia virus: evidence for independent synthesis of glycosylated and unglycosylated forms. J. Virol. 30:551–563. 13. Edwards, S. A., and H. Fan. 1980. Sequence relationship of glycosylated and unglycosylated Gag polyproteins of Moloney murine leukemia virus. J. Virol. 35:41–51. 14. Edwards, S. A., Y. C. Lin, and H. Fan. 1982. Association of murine leukemia virus Gag antigen with extracellular matrices in productively infected mouse cells. Virology 116:306–317. 15. Elbein, A. D. 1987. Inhibitors of the biosynthesis and processing of N-linked oligo saccharide chains. Annu. Rev. Biochem. 56:497–534. 16. Evans, L. H., S. Dressler, and D. Kabat. 1977. Synthesis and glycosylation of polyprotein precursors to the internal core protein of Friend murine leukemia virus. J. Virol. 24:865–874. 17. Fan, H., H. Chute, E. Chao, and M. Feuerman. 1983. Construction and characterization of Moloney murine leukemia virus mutants unable to synthesize glycosylated Gag polyprotein. Proc. Natl. Acad. Sci. USA 80:5965– 5969. 18. Gavel, Y., and G. von Heijne. 1990. Sequence differences between glycosylated and non-glycosylated Asn-X-Thr/Ser acceptor sites: implications for protein engineering. Protein Eng. 3:433–442. 19. Kondo, T., H. Uenishi, T. Shimizu, T. Hirama, M. Iwashiro, K. Kuribayashi, H. Tamamura, N. Fujii, R. Fujisawa, M. Miyazawa, et al. 1995. A single retroviral Gag precursor signal peptide recognized by FBL-3 tumor-specific cytotoxic T lymphocytes. J. Virol. 69:6735–6741. 20. Kozak, M. 1989. The scanning model for translation: an update. J. Cell Biol. 108:229–241. 21. Le Bivic, A., F. X. Real, and E. Rodriguez-Boulan. 1989. Vectorial targeting of apical and basolateral plasma membrane proteins in a human adenocar-

J. VIROL. cinoma epithelial cell line. Proc. Natl. Acad. Sci. USA 86:9313–9317. 22. Ledbetter, J., and R. C. Nowinski. 1977. Identification of the Gross cell surface antigen associated with murine leukemia virus-infected cells. J. Virol. 23:315–322. 23. Ledbetter, J., R. C. Nowinski, and S. Emery. 1977. Viral proteins expressed on the surface of murine leukemia cells. J. Virol. 22:65–73. 24. Lenter, M., A. Levinovitz, S. Isenmann, and D. Vestweber. 1994. Monospecific and common glycoprotein ligands for E- and P-selectin on myeloid cells. J. Cell Biol. 125:471–481. 25. Lopez, C. D., C. S. Yost, S. B. Prusiner, R. M. Myers, and V. R. Lingappa. 1990. Unusual topogenic sequence directs prion protein biogenesis. Science 248:226–229. 26. Lynch, W. P., and J. L. Portis. 1993. Murine retrovirus-induced spongiform encephalopathy: disease expression is dependent on postnatal development of the central nervous system. J. Virol. 67:2601–2610. 27. McAtee, F. J., and J. L. Portis. 1985. Monoclonal antibodies specific for wild mouse neurotropic retrovirus: detection of comparable levels of virus replication in mouse strains susceptible and resistant to paralytic disease. J. Virol. 56:1018–1022. 28. Milev, P., B. Meyer-Puttlitz, R. K. Margolis, and R. U. Margolis. 1995. Complex-type asparagine-linked oligosaccharides on phosphacan and protein-tyrosine phosphatase-zeta/beta mediate their binding to neural cell adhesion molecules and tenascin. J. Biol. Chem. 270:24650–24653. 29. Nakai, K., and M. Kanehisa. 1992. A knowledge base for predicting protein localization sites in eukaryotic cells. Genomics 14:897–911. 30. Neil, J. C., J. E. Smart, M. J. Hayman, and O. Jarrett. 1980. Polypeptides of feline leukemia virus: a glycosylated Gag-related protein is released into culture fluids. Virology 105:250–253. 31. Olden, K., B. A. Bernand, M. J. Humphries, T.-K. Yeo, S. L. White, S. A. Newton, H. C. Bauer, and J. B. Parent. 1985. Functions of glycoprotein glycans. Trends Biochem. Sci. 3:78–82. 32. Parks, G. D., and R. A. Lamb. 1993. Role of NH2-terminal positively charged residues in establishing membrane protein topology. J. Biol. Chem. 268: 19101–19109. 33. Pillemer, E. A., D. A. Kooistra, O. N. Witte, and I. L. Weissman. 1986. Monoclonal antibody to the amino-terminal L sequence of murine leukemia virus glycosylated Gag polyproteins demonstrates their unusual orientation in the cell membrane. J. Virol. 57:413–421. 34. Portis, J. L., S. Czub, C. F. Garon, and F. J. McAtee. 1990. Neurodegenerative disease induced by the wild mouse ecotropic retrovirus is markedly accelerated by long terminal repeat and gag-pol sequences from nondefective Friend murine leukemia virus. J. Virol. 64:1648–1656. 35. Portis, J. L., R. Fujisawa, and F. J. McAtee. 1996. The glycosylated Gag protein of MuLV is a determinant of neuroinvasiveness: analysis of second site revertants of a mutant MuLV lacking expression of this protein. Virology 226:384–392. 36. Portis, J. L., G. J. Spangrude, and F. J. McAtee. 1994. Identification of a sequence in the unique 59 open reading frame of the gene encoding glycosylated Gag which influences the incubation period of neurodegenerative disease induced by a murine retrovirus. J. Virol. 68:3879–3887. 37. Prange, R., and R. E. Streeck. 1995. Novel transmembrane topology of the hepatitis B virus envelope proteins. EMBO J. 14:247–256. 38. Prats, A. C., G. De Billy, P. Wang, and J. L. Darlix. 1989. CUG initiation codon used for the synthesis of a cell surface antigen coded by the murine leukemia virus. J. Mol. Biol. 205:363–372. 39. Schultz, A., and S. Oroszlan. 1978. Murine leukemia virus Gag polyproteins: the peptide chain unique to Pr80 is located at the amino terminus. Virology 91:481–486. 40. Schultz, A. M., E. H. Rabin, and S. Oroszlan. 1979. Posttranslational modification of Rauscher leukemia virus precursor polyproteins encoded by the gag gene. J. Virol. 30:255–266. 41. Schwartzberg, P., J. Colicelli, and S. P. Goff. 1983. Deletion mutants of Moloney murine leukemia virus which lack glycosylated Gag protein are replication competent. J. Virol. 46:538–546. 42. Schwarz, R. T., and R. Datema. 1984. Inhibitors of trimming: new tools in glycoprotein research. Trends Biochem. Sci. 9:32–34. 43. Sitbon, M., B. Sola, L. Evans, J. Nishio, S. F. Hayes, K. Nathanson, C. F. Garon, and B. Chesebro. 1986. Hemolytic anemia and erythroleukemia, two distinct pathogenic effects of Friend MuLV: mapping of the effects to different regions of the viral genome. Cell 47:851–859. 44. Tung, J. S., T. Yoshiki, and E. Fleissner. 1976. A core polyprotein of murine leukemia virus on the surface of mouse leukemia cells. Cell 9:573–578. 45. Varki, A. 1994. Selectin ligands. Proc. Natl. Acad. Sci. USA 91:7390–7397.