Effects of Brefeldin A on the Processing of Viral Envelope Glycoproteins in Murine Erythroleukemia Cells*

This paper documents the effects of brefeldin A (BFA) on the processing and transport of viral envelope glycoproteins in a retrovirus-transformed murine erythroleukemia (MEL) cell line. BFA is a fungal me-tabolite that disrupts intracellular membrane traffic at the endoplasmic reticulum (ER)-Golgi complex junc-tion. In MEL cells, BFA inhibited the processing of the newly synthesized precursor, gPr90env, of the murine leukemia virus envelope protein, gp70, and curtailed the budding of virions into the culture medium by blocking the transport of this protein out of the ER. The block resulted in the intracellular accumulation of gPr90'"' and two putative products of its processing (78 and 66 kDa). The results of endoglycosidase (endo) H and D digestion of the viral glycoproteins in the presence and absence of BFA indicated that (i) there was no glycoprotein processing during the first -2 h of the BFA block; (ii) active Golgi enzymes relocated to the ER in -2 h during BFA treatment, resulting in the production of partially endo H-resistant forms of the spleen focus-forming virus glycoprotein, gp55 (in controls, this glycoprotein was generally retained in the ER as an endo H-sensitive entity); and (iii) proteolytic processing of gPr9O""' to gp70 occurred prior to the acquisition of endo H resistance and at approximately the same time spectrometry Beckman L5-2000 counter. of into MEL densito-metrically or scintillation spectrometry gel previously

. The MuLV enu gene encodes a product (gPr9O""') that yields two coat proteins, p15 and gp70, by proteolytic cleavage thought to occur in a post-endoplasmic reticulum (ER) compartment (Witte and Wirth, 1979;Fitting and Kabat, 1982). A structurally related recombinant form of gp70 (gp55) is encoded in the SFFV genome (Wolff et al., 1983) and is also expressed by MEL cells. However, because of improper disulfide bonding only a minority of the gp55 molecules is processed and transported out of the ER to the cell surface (Gliniak and Kabat, 1989;Kilpatrick et al., 1989). These viral envelope proteins can be used as test objects for a general study of the processing and transport of glycoproteins under normal and perturbed conditions, since they are synthesized and processed by the same cellular machinery as endogenous MEL cell proteins.
The fungal metabolite, brefeldin A (BFA) (Harri et al., 1963), blocks protein traffic from the ER to the Golgi complex (Misumi et al., 1986;Oda et al., 1987;Perkel et al., 1988;Kat0 et al., 1989). In addition, it induces a rapid and reversible disassembly of Golgi complexes (Fujiwara et al., 1988;Palade, 1989a, 1991) and concomitantly a relocation of Golgi markers and enzyme activities to the ER (Lippincott-Schwartz et al., 1989Doms et al., 1989;Ulmer and Palade, 1989a). Because of these activities, BFA has previously been used to ascertain the kinetics and topology of the posttranslational modifications of MEL cell glycophorins (UImer and Palade, 1989a). It has also been used to study the synthesis and processing of class I and I1 major histocompatibility molecules and their association with antigens for presentation to T cells (Nuchtern et al., 1989;Yewdell et al., 1989;Adorini et al., 1990;St.-Pierre and Watts, 1990). In this study, we have used BFA as a tool in an investigation of the processing of an additional set of proteins, i e . the viral envelope glycoproteins produced by MEL cells.

EXPERIMENTAL PROCEDURES
Materials-BFA was a generous gift from J. Lippincott-Schwartz and R. Klausner (National Institutes of Health (NIH)), and polyclonal goat anti-MuLV gp70 serum was kindly provided by s. Ruscetti (NIH). Recombinant endo-(3-N-acetylglucosaminidase H and Diploccocus pneumoniae endo-P-N-acetylglucosaminidase D were from Boehringer Mannheim, and ["S]methionine (1200 Ci/mmol) was from Amersham C o p .
Metabolic Labeling-For continuous metabolic labeling, cells were collected by centrifugat.ion (10 min a t 2,000 X g), washed three times by resuspension-sedimentation in methionine-free RPMI 1640 medium, then incubated in the same medium (supplemented as mentioned above) containing 25 pCi/ml [35S]methionine in the absence or presence of BFA a t 1 pg/ml. In pulse-chase labeling experiments, 9173 the cells were labeled as above, and an effective chase was obtained by adding a 750,000-fold excess of unlabeled methionine (final concentration of 1.5 mg/ml). In cases where recovery of MEL cells from BFA was investigated, the cells were removed from the culture medium containing BFA and ["S]methionine by centrifugation (10 min a t 2,000 X g), then resuspended in complete RPMI 1640 supplemented as before, and further incubated for selected time periods. T o end the labeling, the cells were collected by centrifugation and washed three times with ice-cold phosphate-buffered saline.
Isolation of Viruses-Metabolically labeled MEL cells were separated from the culture medium by low speed centrifugation (10 min at 2,000 X g). The ensuing supernatant was recovered and subjected to a second low speed spin to remove remaining cells or cell fragments, and then to high speed centrifugation (60 min a t 40,000 X g ) to recover shed virions. The high speed supernatants were discarded, the pelleted viruses were solubilized in 0.5% Nonidet P-40, 2% SDS, 10 mM dithiothreitol in 50 mM Tris-HC1, p H 7.5, and the ensuing lysates were processed through immunoprecipitation.
Immunoprecipitation and Gel Electrophoresis-Washed MEL cells (5 X lO'/aliquot) were lysed for 15 min a t 4 'C in 0.5 ml of 1% Nonidet P-40 in 50 mM Tris-HC1, pH 7.5. Insoluble residues were removed by centrifugation (5 min at 13,000 X g). The resulting supernatants were mixed with 0.5 ml of 4% SDS, 20 mM dithiothreitol in 50 mM Tris-HC1, pH 7.5, and boiled for 10 min. After cooling to room temperature, the samples were processed for immunoprecipitation by the successive addition of (i) 50 p1 of 1 M iodoacetamide for 30 min a t 37 "C, (ii) 0.5 ml of 20% Triton X-100 for 15 min a t 4 "C, (iii) 3-5 p1 of polyclonal goat anti-MuLV gp70 serum (-1:300 to -1:500 dilution) for 15 min a t 4 "C, and (iv) 10 mg of Protein A-Sepharose CL-4B for 45 min a t 4 "C (with gentle agitation) to bind and recover antigen-antibody complexes. The beads were washed with 150 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40 in 50 mM Tris-HC1 buffer, pH 7.5, (3 X 1 min), then with the same buffer containing 500 mM NaCl (1 X 1 min), and finally with distilled water (1 X 1 min). Immune complexes were released from the beads by boiling for 5 min with 3% SDS, 1.5% 0-mercaptoethanol, 2 M urea, 2 mM EDTA in 62.5 mM Tris-HC1 buffer, pH 6.8. Samples of the ensuing lysates were adjusted to 5% glycerol and 10 pg/ml Pyronin Y and processed through sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) by Laemmli's procedure (1970). One-dimensional peptide mapping of the viral glycoproteins was performed using Stuphyloccocus aureus V-8 protease and trypsin (0.05 to 5 pg/well) by our modification (Ulmer and Palade, 1989b) of the procedure of Cleveland et al. (1977).
Determination of Radioactiuity-To assay [35S]methionine incorporation into total MEL cell proteins, aliquots (2 pl) of culture medium or cells in suspension were spotted onto Whatman 3MM filter paper, dried, and incubated in 10% trichloroacetic acid (total volume of 10 ml/filter) for 1 h a t 4 "C. The filters were rinsed in 5% trichloroacetic acid (3 X 15 min) followed by ethanol (2 X 15 min) a t 4 "C and then dried. Radioactivity was determined by liquid scintillation spectrometry in a Beckman L5-2000 scintillation counter. Incorporation of radioactivity into specific MEL cell proteins was determined by fluorography of SDS-PAGE (as in Bonner and Laskey, 1974). Radioactivity in individual proteins was estimated densitometrically using a Kodak Visage 2000 image analyzer or by liquid scintillation spectrometry of gel slices as previously described by us (Ulmer and Palade, 1989b).
Glycosidase Digestion-Immunoprecipitated MuLV and SFFV glycoproteins were released from the Protein A-Sepharose beads by boiling for 10 min with 0.1% SDS and 1 mM phenylmethylsulfonyl fluoride in 100 mM sodium citrate, p H 5.6, for endo-P-N-acetylglucosaminidase H (endo H) digestion, and with 0.1% SDS, 0.7% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 5 mM EDTA in 50 mM sodium citrate, pH 6.5, for treatment with endo-6-N-acetylglucosaminidase D (endo D). Samples (25 pl) were incubated with or without 2 milliunits of endo H or 2.5 milliunits of endo D for 16 h at 37 "C and then solubilized as previously described, for subsequent SDS-PAGE and fluorography.

Effects of BFA on Protein Synthesis and Release into the
Culture Medium In MEL cells, [3sS]methionine labeling of total acid-precipitable material was progressively stimulated by BFA so that, by 6 h, cell radioactivity was -3-fold higher than in controls (Fig. 1A). This result, which may reflect intracellular accumulation of proteins as well as an increase in the rate of protein synthesis, is a departure from results obtained with other cell types (Misumi et al., 1986;Kat0 et al., 1989) in which protein synthesis was not found to be affected by BFA.
The total amount of acid-precipitable radioactivity released into the culture medium increased with time but remained relatively small and not strikingly different from controls (Fig. 1B). However, given the difference in [35S]methionine incorporation already mentioned, it amounted to -10% of cell radioactivity in BFA-treated cells past 2 h, as opposed to -27% in corresponding controls (Fig. 1C). Removal of BFA led to prompt increases in amounts (Fig. 1B) and percentage ( Fig. IC) of released acid-precipitable radioactivity. On a percentage basis, the latter reached initial levels over a 2-4-h period of recovery (Fig. IC). Therefore, BFA leads to accumulation of newly synthesized proteins in MEL cells and decreases, but does not abolish, the release of such proteins into the culture medium.
Release of virions by MEL cells was investigated by virus isolation from culture media and immunoprecipitation of viral glycoproteins therefrom. In controls, radiolabeled gp70 was detected in virions isolated from the culture medium by 2 h of continuous labeling (Fig. 2). In contrast, labeled gp70 was not detected in equivalent preparations from BFA-treated cells even after 6 h of continuous labeling, indicating that budding of virions containing newly synthesized coat proteins was abolished by the drug.

Effect of BFA o n M u L V Glycoprotein
Transport and Processing gPr9F"""The MuLV env gene encodes a precursor protein, gPrSO""", that gives rise to two components of the virus coat (gp70 and p15) by proteolytic cleavage assumed to occur late in the ER or early in the Golgi complex (Witte and Wirth, 1979;Fitting and Kabat, 1982). In control cells, cleavage of gPr90'"' to yield gp70 was first detected by 30 min of continuous labeling (Fig. 3) and reached substantial levels past 2 h (Fig. 4). In a pulse-chase labeling experiment, gPr9O'"' was mostly converted to gp70 by 1-2 h of chase following a 1-h pulse ( Fig. 5A; see also Figs. 6 and 7). The level of labeled gp70 in the cells reached a maximum at -1 h of chase and decreased gradually thereafter, suggesting that at this time protein was being shed by the cells, probably as assembled virions: the decrease coincides in time with the presence of detectable amounts of gp70 in released virions (see Fig. 2). As expected for a protein present in the ER, gPr90'"' accumulated in the presence of BFA, with no apparent conversion to mature gp70 ( 78 kDa-A protein of -78 kDa was also immunoprecipitated from labeled MEL cells with the gp70 antiserum (see Fig. 3 and open arrowhead on lane 1 , Fig. 4). The 78-kDa protein displayed the following properties: (i) it is antigenically related to gPr9O' "' and gp70 (it is immunoprecipitated by gp70 antiserum); (ii) it was already metabolically labeled before 15 min, therefore prior to the appearance of labeled gp70 and at the same time or later than the detection of labeled gPr90'"' (Fig. 3); (iii) its radioactivity decreased with time during pulse-chase labeling in parallel with that of were resuspended in complete medium without RFA and further incubated for up to 4 h of chase/recovery (0). Acid-insolul)le radioactivity was determined for cells in suspension and culture supernatant hy trichloroacetic acid precipitation on filter paper (as described under "Experimental Procedures"), and plotted as total radioactivity in the cells (panel A ) , culture medium ( p o n d H ) . and percent of total radioactivity (cells plus medium) recovered in the culture medium (panc4 C ) . gp70 and 66 kDa-The appearance of labeled m70 was completely inhibited by BFA, over the entire duration (6 h) of the experiments (Fig. 4). By 2-4 h in BFA, however, a protein of lower apparent mass than gp70 (-66 kDa) was immunoprecipitated from the cells (see lower open arrowhead on lane 8, Fig. 4). A similar protein was not detected in the absence of BFA in any experiments. This 66-kDa protein could be (i) a modified form of g p F j . 5 ; (ii) an experimental art.ifact (ie. a degradation product of ~7 0 , gPrSO""', or the 78-kDa protein); or (iii) a proteolytically processed form of The oprn arrowhrad on lane I denotes a 78-kDa protein and those in lane X indicate proteins of 78 and 66 kDa. In order to obtain adequate resolution of gp70 and the 78-kDa protein, MuLV pl5 was recovered on a separate gel (lorwrpanel). gPr9O'"' or the 78-kDa protein. The latter possibility is the most likely explanation, based on the following observations: (i) the 66-kDa protein is more closely related in structure to gPrSO""', gp70, and the 78-kDa protein than to gp55, as suggested by their susceptibility to trypsin and V-8 protease assessed by one-dimensional peptide mapping (not shown); (ii) there was apparently a precursor-product relationship between the 78-and 66-kDa proteins in BFA-treated cells (Fig. 5R); and (iii) labeled p15 appeared in BFA-treated cells at the same time as the 66-kDa protein (see lower panel, Fig.  4). In control cells, p15 appeared concomitantly with gp70, as expected, since it is the cleaved C-terminal segment of the MuLV gPr9O'"' gene product. These results dem0nstrat.e that the proteolytic processing of gPr9O'"' that yields gp70 is completely inhibited in the presence of BFA. Yet, a proteolytic event involving the MuLV envelope glycoproteins does occur after prolonged exposure to BFA (past 2 h), yielding a 66-kDa protein and p15. Under normal circumstances, the 66-kDa protein may exist only transiently or not at all, depending upon the temporal sequence of proteolytic cleavage and oligosaccharide processing (eg. the 66-kDa protein may exist transiently if cleavage takes place before final oligosaccharide processing). In any case, these data demonstrate that proteolytic processing of virally encoded proteins does occur in the ER during BFA treatment, albeit slowly and incompletely, presumably due to relocation of Golgi protease(s) to the ER.

Effect of RFA on the Glycosylation of MuLV
Envelope Proteins During glycoprotein biosynthesis, asparagine-linked oligosaccharides are susceptible to digestion by endo H until converted from high mannose to hybrid or complex type by the action of enzymes in the middle and trans Golgi cisternal compartments (for review see Kornfeld and Kornfeld, 1985). In control cells, the MuLV-encoded glycoproteins, i.e. gPrSO"", gp70, and the putative 78-kDa intermediate, were all endo H sensitive (Fig. 6A), indicating that proteolytic processing of gPr9O"" to gp70 occurs prior to the acquisition of endo H resistance (i.e. before middle to trans Golgi). Even after 5 h of chase, intracellular gp70 was mostly sensitive to endo H (Fig. 6 A ) . Yet some gp70 molecules found in budded virions were completely endo H resistant (Geyer et al., 1984; see Fig. 6A), which demonstrates that up to 5 h a substantial portion of the newly synthesized gp70 molecules is retained in compartments prior to the middle Golgi and suggests that transit from the latter to the plasmalemma and virus budding therefrom are not the rate-limiting steps in the appearance of  Fig. 6R). Endo D specifically removes N-linked oligosaccharides containing an cul-3-linked mannose residue unsubstituted in the 2 position (Mizouchi et al., 1984). Glycoproteins become endo D sensitive by the action of nl-2 mannosidase I in the cis Golgi compartment to yield the ''Man5GlcNAc? structure; subsequent addition of GlcNAc residues by GlcNAc transferase I in the middle Golgi compartment renders such proteins resistant to the enzyme once more (for review, see Kornfeld and Kornfeld, 1985).' In both control and BFA-treated MEL cells, gPr9O""' and the 78-kDa protein were not discernably sensitive to endo Under normal conditions a glycoprotein with N-linked glycans is endo H-sensitive while in the ER and cis Golgi compartment; it hecomes endo H-resistant upon acquiring terminal sugars in the middle and trans Golgi compartments. A similar or the same protein can hecome transiently endo D sensitive while in the cis Golgi compartment when its glycans acquire a Man:,GlcNAc: structure and is converted to endo D-resistant form upon the addition o f a GlcNAc residue in the middle Golgi compartment. D a t any time during pulse-chase experiment,s (Fig. 7, A and  R ) . In cont,rast, gp70 and a portion of the 66-kDa protein were reduced by -2 kDa in apparent molecular mass by the glycosidase (see open arrowheads, Fig. 7, A and E ) . These results demonstrate that proteolytic processing of gPr9O'"" to gp70 and the acquisition of endo D sensitivity by t.he latter were not resolved in time under our experimental conditions. Both apparently occur in a cis Golgi compartment. The endo D sensitivity of the 66-kDa protein probably reflects the redistribution of pertinent protease(s) and glycosyltransferases from the Golgi complex to the ER during RFA treatment.

Effect of RFA on SFFV gp.55 Processing and Turnover
The structurally related gp55 is, for the most part, retained in the ER; only a small fraction is properly disulfide-bonded and transported to the cell surface (Gliniak and Kabat, 1989;Kilpatrick et al., 1989). In our experiments, gp55 accumulated in control MEL cells (no BFA) during continuous labeling experiments, likely due to its retention in the ER (Figs. 3 and  4). In pulse-chase experiments, however, labeled gp55 was lost with a half-life of -5 h (Fig. 5 C ) . This decrease is not likely to be due to processing and maturation, since the mature form of gp55, a 65-kDa protein (Ruta and Kabat, 1980), did not accumulate to a significant degree in these cells. A 65-kDa protein was immunoprecipitated from the cells after 4-6 h of labeling, but in negligible amounts by comparison with gp55 ( see lanes 3 and 4, Fig. 4). It is more probable that this decrease in gp55 radioactivity represents protein degradation. Limited processing of gp55 occurred both in control and RFAtreated cells, as evidenced by a small but progressive decrease (up to -1 kDa) in its mass during the chase period (see Figs. 3, 4, and 7), probably on account of trimming of glucose and some mannose residues in the ER (for review, see Kornfeld and Kornfeld, 1985). It is interesting to note that in the presence of BFA the amount of radioactivity in gp55 decreased only slightly during the chase period (tli.! = -20 h) (Fig. Tic), suggesting that the degradation of this protein is inhibited by BFA in the ER or requires its export out of the ER (e.g. to lysosomes).
Endoglycosidase digestion was performed on control and BFA-treated MEL cells to follow the redistribution of Golgi enzymes to the ER during BFA treatment; gp.55, known to be retained in the ER, was used as a test object. I n control cells, gp55 was completely sensitive to endo H for at least 5 h after its synthesis, resulting in a truncated form of -45 kDa (see open arrowhead, Fig. 8A), which indicates that gp55, as expected, has only high mannose-tyDe oligosaccharide chains. In BFA-treated cells, however, gp55 gradually became partially resistant to endo H and, past 1 to 2 h of chase in the presence of BFA, incompletely digested forms of -47, -49, and -51 kDa were observed (see open arrowheads, Fig. 8B). The presence of three intermediate forms of gp55 revealed by endo H digestion agrees in general with previous reports that this protein has 4 N-linked oligosaccharide chains (Polonoff et al., 1982;Srinivas and Compans, 1983).

DISCUSSION
The viral envelope glycoproteins synthesized in MEL cells provide a useful system to study intracellular protein transport and processing, since they are comprised by two structurally related gene products (encoded by two different viruses) that have distinct biogenetic fates. The first, the SFFV gp55 protein is, for the most part, not exported out of the ER, thereby providing an ER biochemical marker. The second, the MuLV envelope glycoprotein, is transported to the cell surface and in transit undergoes proteolytic and oligosaccharide processing steps that can be used to define the kinetics of these steps, to follow the progress of the protein through the ER and Golgi complex and the redistribution of Golgi enzymes to the ER. Finally, intracellular virus particles and budded virions can be readily detected by electron microscopy. Hence, the effects of BFA can be easily and extensively monitored in this system.

Effects of BFA on M E L Cell Protein Transport and Metab-
olism-As with proteins in other cells (Takatsuki and Tamura, 1985;Misumi et al., 1986;Kat0 et al., 1989;Oda et d., 1990), BFA curtailed without abolishing the release of newly synthesized MEL cell proteins into the culture medium. This continued appearance of labeled proteins in the culture medium in the presence of BFA was not due to an ineffective block in exocytosis by the drug, based on morphological and biochemical data (not shown). Rather, this may represent cytolysis or some other process that remains to be elucidated. In contrast, budding of virus containing newly synthesized coat proteins was completely inhibited by BFA. Yet, virions apparently continued to bud in the presence 'of the drug, as observed by electron microscopy (Ulmer and Palade, 1991); their budding probably involved complete virus particles that were already past the Golgi complex at the time of BFA administration and/or viruses containing only the gag gene products (Shields et al., 1978). Further work using probes to other MuLV proteins will be necessary to determine the nature of these virus particles. These slow budding viruses are, however, a small minority, since the rate-limiting step in virus maturation seems to be in the Golgi complex or earlier, as evidenced by the accumulation of endo H-and D-sensitive forms of MuLV gp70 in control cells. Moreover, previous work on this system (Fitting and Kabat, 1982) and others (Williams et al., 1985) has demonstrated that exit of viral proteins from the ER seems to be the rate-limiting step in protein transport to the cell surface.
General Effects of BFA on Viral Glycoproteins i n M E L Cells-BFA blocked transport of newly synthesized viral envelope proteins out of the ER to the Golgi complex in MEL cells and in addition caused a redistribution of Golgi enzymes to the ER. These changes in traffic patterns were documented primarily biochemically as the absence of Golgi-type modifications during the first 2 h and the late appearance of such modifications past 2 h. In our case, we had the advantage of an ER marker, gp55, that in controls remained permanently endo H sensitive, whereas under BFA it acquired endo H resistance past 2 h. In work done in other laboratories, the redistribution of Golgi antigens was also demonstrated by immunocytochemical tests (Lippincott-Schwartz et al., 1989;Doms et al., 1989). Our results are in agreement with data already reported on murine glycophorins in MEL cells (Ulmer and Palade, 1989a), the T cell antigen receptor in murine T cell hybridomas (Lippincott-Schwartz et al., 1989), and G proteins in vesicular stomatitis virus-infected baby hamster kidney cells (Takatsuki and Tamura, 1985), Chinese hamster ovary cells (Doms et al., 1989), and hepatocytes (Oda et al., 1990).

Specific Effects of BFA on Viral Glycoproteins in MEL
Cells-Besides these general changes, applicable to all cases studied to date, our results have revealed additional modifications that so far apply only to the viral proteins produced in MEL cells. First, the SFFV glycoprotein gp55, which turned over with a half-life of -5 h in control cells, accumulated and turned over at a considerably slower rate (tll2 = -20 h) in the presence of BFA. Therefore, either gp55 requires transport out of the ER for degradation or BFA inhibits its degradation within the ER. Further experiments will be required to determine whether gp55 is degraded in the lysosomes or in the ER. The latter site has recently been proposed in the disposal of excess subunits of the T cell antigen receptor (Lippincott-Schwartz et al., 1988) as well as the asialoglycoprotein receptor (Amara et al., 1989). Second, two proteins have been detected that are antigenically related to gPr9O'"" and gp70, the envelope proteins of MuLV. The first, a -78-kDa protein, seen transiently under normal conditions, is apparently a product of gPr9O'"' processing. This conversion, the nature of which is presently not known, likely occurs in the ER, since the 78-kDa protein was metabolically labeled within 15 min (preceding the labeling of gp70 and p15) and BFA did not inhibit or delay its appearance. One interesting possibility is that the 78-kDa protein is produced by proteolytic cleavage of gPr9O'"" in the ER in a step distinct from the already established proteolytic cleavage of the same protein that produces gp70 and p15 in a post-ER compartment (Fitting and Kabat, 1982). In support of this notion is the fact that endo H reduced the apparent mass of gPr90""" and the 78-kDaprotein by approximately the same amount (-15-18 kDa) (see Fig. 6B), suggesting that the difference in their structure lies, at least in part, in the protein backbone. It seems, however, that only a small fraction of gPr9O'"' undergoes this type of processing, since (i) at any given time, gPr9O'"" is much more abundant than the 78-kDa protein, and (ii) during pulse-chase experiments in the presence of BFA, radioactivity was not effectively chased from gPr9W"' into the 78-kDa protein (see Fig. 6B). The reasons for this alternative processing (compartmentalization of gPr9O""' or isoforms with different protease sensitivities?) remain to be established. If the proposed ER cleavage occurs in the C terminus of gPrSO""', then the subsequent action of the Golgi protease that normally generates gp70 would still produce gp70, but only a truncated version of p15. Conversely, if the ER cleavage occurs in the N terminus of gPr90e"', then the Golgi protease would generate p15 and a truncated gp70. Since p15 from BFA-treated cells was identical in apparent mass to that from control cells (see Fig. 4), the most likely explanation is that the proposed cleavage of gPr9O'"' in the ER occurs in the N terminus of the protein. Accordingly, the second related protein detected, a 66-kDa protein, is probably a truncated version of gp70, generated from the 78-kDa protein as described above. In support of this inference is an apparent precursor-product relationship between the 78-and 66-kDa proteins. However, since the 66-kDa protein was only detected in BFA-treated cells, its relevance is uncertain. It may be that in control cells the small amount of the 78-kDa protein generated from gPrgO""' does not leave the ER (i.e. it is locally degraded). But, in the presence of BFA, the protease(s) involved in the production of gp70 and p15 from gPr9O""' relocates from the Golgi complex to the ER, where it can act on the 78-kDa protein to yield the 66 kDa truncated form of gp70.
Localization of Processing Events-The cleavage of gPr9O""' to gp70 and p15 appears to occur in the cis Golgi, based on the following lines of evidence: (i) gPr90""' accumulated in the presence of BFA, with no evidence of cleavage to gp70; (ii) removal of BFA resulted in the appearance of gp70, concomitantly with the disappearance of gPr90'"'; (iii) gPrSO'"', gp70, and the two related proteins (78 and 66 kDa) were endo H sensitive, indicating that proteolytic processing of gPr9O""" to gp70 takes place before they reach the middle Golgi compartment, where endo H resistance is acquired; and (iv) gPr90'"" and the 78-kDa protein were endo D resistant, whereas gp70 and the 66-kDa protein were sensitive to the glycosidase, suggesting that, in our experimental conditions, the cleavage event cannot be resolved in time from the acquisition of endo D sensitivity in all cases. It is believed that endo D sensitivity is conferred upon arrival in the cis Golgi compartment, when oligosaccharides are trimmed to the Man5GlcNAc2 structure (for review, see Kornfeld and Kornfeld, 1985). Therefore, the cleavage of gPr9O'"' to gp70 must occur distal to the BFA block in MEL cells (i.e. past the ER) but not past the cis Golgi. The proteolytic processing of gPr9O""' to gp70 and p15 in an early Golgi compartment in MEL cells is in contrast to similar processing of the envelope proteins in other retroviruses, which takes place in a late Golgi compartment (for review, see Krausslich and Wimmer, 1988).
During BFA treatment, the 66-kDa protein appeared at the time when SFFV gp55 and the 78-kDa protein became partially resistant to endo H (>2 h). This time frame coincides with the delayed acquisition of 0-linked oligosaccharides by MEL cell glycophorins under similar conditions (Ulmer and Palade, 1989a). Therefore, the protease(s) responsible for the production of the 66-kDa protein (perhaps the same one(s) that converts gPr9O'"' to gp70) probably relocates to the ER from the Golgi complex along with the enzymes involved in oligosaccharide processing. It is interesting to note that not all molecules of gp70 found in shed virus were completely endo H resistant, indicating that terminal oligosaccharide processing is not a prerequisite for transport of gp70 to the cell surface or assembly into virions. In fact, in some strains of MuLV, proteolytic processing of gPr9O'"' to gp70 is not required for its export (Famulari and English, 1981;.

SUMMARY
BFA has enabled us to study in detail several aspects of the processing of viral envelope proteins in MEL cells. First, a small fraction of the MuLV envelope precursor, gPrSO'"', undergoes an apparent proteolytic cleavage in its N terminus while in the ER to yield a 78-kDa protein. Second, the normal proteolytic processing of gPr9O'"' that gives rise to gp70 and p15 apparently occurs in a cis Golgi compartment, unlike the cleavage of other retroviral envelope proteins which is effected in a trans Golgi compartment. Finally, SFFV gp55, which is for the most part not transported out of the ER to the Golgi complex, undergoes turnover a t a rate that is substantially reduced by BFA. The site of gp55 degradation remains to be established.