Intracellular processing of the gp160 HIV-1 envelope precursor. Endoproteolytic cleavage occurs in a cis or medial compartment of the Golgi complex.

The intracellular processing of the gp160 HIV-1 envelope precursor was characterized in acutely infected CD4+ T cells. Our data show that gp160 undergoes endoproteolytic cleavage by a nonacid dependent protease(s) in the rough endoplasmic reticulum-Golgi complex, within cis or medial cisternae, and is not transported to the cell surface. Two-dimensional electrophoretic pulse-chase analysis indicates that it takes greater than 2 h for gp160 to be transported from the rough endoplasmic reticulum to the site of action of sialyltransferases in the trans Golgi. Evidence is presented that gp160 is subject to mannose trimming in the Golgi complex, which is inhibited by 1-deoxymannojirimycin (a specific Golgi alpha-mannosidase I inhibitor). Preliminary data also suggest that gp120 is post-translationally modified by sialylated O-linked oligosaccharides.

The intracellular processing of the gp160 HIV-l envelope precursor was characterized in acutely infected CD4+ T cells. Our data show that gp160 undergoes endoproteolytic cleavage by a nonacid dependent protease in the rough endoplasmic reticulum-Golgi complex, within cis or medial cisternae, and is not transported to the cell surface. Two-dimensional electrophoretic pulse-chase analysis indicates that it takes greater than 2 h for gp160 to be transported from the rough endoplasmic reticulum to the site of action of sialyltransferases in the truns Golgi. Evidence is presented that gp160 is subject to mannose trimming in the Golgi complex, which is inhibited by l-deoxymannojirimycin (a specific Golgi cr-mannosidase I inhibitor). Preliminary data also suggest that gp120 is posttranslationally modified by sialylated O-linked oligosaccharides.
Human immunodeficiency virus (HIV-l) is the etiologic agent of the acquired immunodeficiency syndrome (AIDS) and its related disorders. CD4 has been definitively established to be the cell surface receptor responsible for HIV-l target cell tropism (1,2). HIV-l is productively infective in a wide spectrum of CD4-bearing cell types including T cells (3), B cells (4), monocytes/macrophages (5,6), Langerhans cells (i'), follicular dendritic cells (8,9), glial cells (lo-13), and cells of the colonic epithelium (14) and GI tract (15). The HIV-l gpl20 envelope moiety is responsible for virion-specific binding to cell surface CD4 (1,2,16), and the gp41 transmembrane component putatively mediates virus-to-cell membrane fusion, an event requisite for viral nucleocapsid penetration into the cytosol (17). gp120 and gp41 are biosynthetically derived from a single 160-kDa precursor polyprotein (18)(19)(20), and are believed to be associated only by noncovalent molecular interactions (21). The gp160 precursor molecule is translated from a 4.3-kilobase spliced mRNA encoding the enu message (22). This message produces a 90-kDa polypeptide that is cotranslationally modified in the lumen of the rough endoplasmic reticulum by the addition of approximately 31 N-asparagine-linked oligomannosyl carbohydrate residues (GlcsMangGlcNAcp-) (18,21,23).
The formation of gp120 and gp41 is consequent to endoproteolytic cleavage of the gp160 polypeptide backbone. There are two potential trypsin-sensitive cleavage sites within the * This work was supported in part by National Institutes of Health Grants HL33811 and AI25922 and bv the Medical Research Service of the Veterans Administration.
TGe costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
envelope coding region which, if hydrolyzed, could function to release the N-terminal hydrophobic domain of gp41. In particular, the highly conserved Arg-Glu-Lys-Arg sequence appears to be the preferred site of gp160 endoproteolytic cleavage as determined by site-directed mutagenesis (24). The nature and intracellular localization of protease enzyme(s) involved in gp160 endoproteolysis are to date speculative. However, cleavage of the Rous sarcoma virus (RSV)' envelope precursor (Pr95) has been associated with a putative novel Golgi endopeptidase that is highly lysine-specific (25). Functionally, endoproteolytic cleavage of the gp160 HIV-l envelope precursor has been found to be requisite for the production of infectious virions (24). However, the precise intracellular localization of this cleavage has not been identified.
Other retroviruses, such as murine leukemia virus (26,27) and RSV (28) encode envelope precursors (gPr90 and Pr95, respectively) which are endoproteolytically cleaved in the RER-Golgi complex proximal to the site where N-asparaginelinked mannose residues are extended to form complex oligosaccharide side chains. Alternatively, the influenza envelope precursor, HAo, is cleaved by a cellular protease (29) approximately at the time of its insertion into the plasma membrane.
In this study we have characterized and compared Nasparagine-linked oligosaccharide side chains of gp160 and gp120 envelope determinants derived from radiolabeled CD4+ T cells acutely infected with HIV-l to identify the intracellular site of endoproteolytic cleavage of the gp160 envelope precursor. The data indicate that the cleavage reaction takes place in the cis or medial cisternae of the RER-Golgi complex close to the site of action of Golgi a-mannosidase I. Nuclear) (except for 'Z51-labeled proteins) and exposed to Kodak X-Omat AR film at -70 "C.

RESULTS
Physical Characterization of HIV-l Envelope Determinants by Two-dimensional Gel Analysis-HIV-l envelope moieties were immunoprecipitated from acutely infected VB cells radiolabeled with [35S]cysteine and [35S]methionine using purified IgG from an HIV-l seropositive patient (HW). The respective p1 values of gp160, gp120, and gp41 were then resolved by two-dimensional gel electrophoresis (Fig. lA). The gp160 envelope precursor focuses principally as a contiguous neutral to basic species over a pH range of 7.0 to greater than 7.5. Alternatively, cell-associated gp120 resolves into multiple discrete species with a wide acidic pH range between 5.0 and 7.5, and gp41 focuses broadly across a pH gradient of 6.0 to greater than 7.5. It is of note that gp41, unlike gp160 and gp120, is not consistently immunoprecipitated with HW, as evidenced by its frequent absence upon SDS-PAGE analysis. The predicted p1 values of the naked polypeptide backbone chains of all HIV-l envelope determinants are basic (approximately 8.0), as calculated from the known amino acid sequence of multiple clonal isolates (21). More specifically, in the LAV-la clone (33) the number of basic amino acid residues (arginine, histidine, lysine) are in excess compared to the number of acidic amino acid residues (aspartate, glutamate) by a factor of 1.35, 1.40, and 1.29, in gp160, gp120, and gp41, respectively. Thus, the acidic p1 of both gp120 and gp41 relative to the basic p1 of gp160 is attributed to differential biosynthetic processing. In particular, both gpl20 and gp41 must acquire abundant negatively charged moieties posttranslationally subsequent to endoproteolytic cleavage of gp160. These data are inconsistent with the hypothesis that gp160 is cleaved at the site of viral budding or after exiting the Golgi complex as is the case for HA,,, the influenza virus envelope precursor (29), since most posttranslational charge modifications of membrane spanning N-linked glycoproteins occur in the Golgi network.
Additional two-dimensional analysis was performed on HIV-l envelope proteins immunoprecipitated with monospecific anti-HIV-l envelope antisera (env 2-3 and PB46) ( Effect of Glycosidase Enzyme Digestion on the Physical Properties of gp160 and gplZO-Cell-associated gp120 was digested with neuraminidase in order to explore the molecular basis for its broad resolution into multiple distinct acidic species upon isoelectric focusing. The two-dimensional gel displayed in Fig. 2B demonstrates that desialylation of gpl20 with neuraminidase results in its disappearance from the p1 range of 5.0 to 7.5 ( Fig. 2A). This loss was confirmed not to be secondary to proteolysis. Upon nonequilibrium pH gradient electrophoresis (32), the desialylated species resolve at a basic pH greater than 7.5 (data not shown). One-dimensional SDS-PAGE analysis shows that after neuraminidase digestion, the apparent molecular mass of gp120 is reduced by approxi-  focusing represents extensive progressive sialylation during biosynthesis. The gp160 HIV-l envelope precursor, in contrast to gp120, is marginally sialylated as evidenced by its relative resistance to neuraminidase. Although there is a slight reduction in the apparent molecular weight of gp160 noted by one-dimensional SDS-PAGE analysis consequent to neuraminidase digestion ( . The extensive sialylation of gp120 relative to gp160 signifies that endoproteolytic cleavage of gp160 occurs before gp120 attains sufficient sialic acid residues to confer a shift in p1 to the acidic range. It has previously been determined that virion-associated gp120 is comprised of greater than 20 potential N-glycosylation sites (21) and that the N-linked carbohydrates present are equally represented by both high mannose and complex sialic acid-containing carbohydrate glycans (34). Since those sialyltransferase enzymes responsible for the addition of sialic acid to Golgi (Ymannosidase trimmed N-asparagine-linked oligosaccharide side chains are known to reside in the tram Golgi complex (35,36), we conclude that gp160 is endoproteolytically cleaved in or proximal to this compartment. It is well established that both gp160 and gp120 are exquisitely sensitive to digestion with endo H (18,23,34). Endo H selectively cleaves oligomannosidic N-asparagine-linked oligosaccharides between the two N-acetylglucosamine residues which bind the core sugars to the peptide backbone (37, 38). The biosynthetic conversion of endo H-sensitive high mannose N-linked glycans into complex-type glycans occurs in the medial and truns Golgi (35, 36) and confers endo H resistance. Cell-associated gp160 and gp120 immunoprecipitated with either HW or PB12 (rabbit anti-gp41) were subjected to endo H digestion and subsequently analyzed by onedimensional SDS-PAGE. PB12 specifically recognizes the polypeptide backbone of gp160, but is completely nonreactive with gp120 (Fig. 3A from cell lysates using env 2-3. Selected immunoprecipitates were subjected to neuraminidase, PNGase F, or neuraminidase and PNGase F digestion. One-dimensional SDS-PAGE analvsis was then carried out under reducing conditions in 8% polyacrylamide. Lane 1, nonimmune rabbit serum; lane 2, env 2-3, control; lane 3, env 2-3, neuraminidase-treated; lane 4, env 2-3, PNGase F-treated; lane 5, env 2-3, neuraminidasetreated followed by PNGase F digestion. apparent molecular masses of 70, 90, and 100 kDa (Fig. 3B, lane 4). Only the 90-kDa protein was detected with PB12 (data not shown), demonstrating that this species represents the endo H digestion product of gp160. Moreover, the apparent molecular mass of the polypeptide backbone of gp160 is also approximately 90 kDa, consistent with the notion that the native envelope precursor polypeptide is modified primarily with endo H-sensitive N-linked oligomannosyl glycans. However, the observation that gp160 is marginally sialylated suggests that gp160 is also modified with rare endo H-resistant complex N-linked and/or O-linked oligosaccharide side chains.
When the bulk of N-linked oligosaccharides side chains are removed from gp120 with endoglycosidase F, the apparent molecular mass of the residual polypeptide backbone is estimated to be approximately 60 kDa (39). Therefore, the obser-vation that the 70-and lOO-kDa proteins displayed in Fig. 3B (lane 4) are not detected with PB12 (data not shown), is consistent with the conclusion that these proteins represent the endo H digestion products of cell-associated gp120. Since virion-associated gp120 is substituted by roughly equal amounts of oligomannosidic and complex-type glycans (34), the 70-kDa endo H digestion product seen in Fig. 3B (lane 4) likely reflects immature g-p120 whose N-linked oligosaccharide side chains have been only partially processed into c ?mplex forms. The increased endo H resistance of the lOO-kDa species relative to the 70-kDa species supports this hypothesis. Thus, these data are consistent with the proposal that endoproteolytic cleavage of gp160 occurs in or proximal to those compartments of the Golgi apparatus involved in the synthesis of complex-type iv-linked side chains (i.e. the melial and truns cisternae). The formation of complex carbonydrates is ultimately responsible for conferring endo H resistance to gp120 (34,40).
Endo H specifically hydrolyzes N-linked high mannose oligosaccharides which range in size from a trimannosyl core [ManLul+6(Mancul+3)Man~l+4GlcNAc/3l+4GlcNAc-] up to derivatives with 50 mannose moieties (41). Endo D, however, has restricted specificity for only those iv-linked glycans that have an unsubstituted al-t3 mannosyl linkage at the C-2 position (38, 42). Although endo D-sensitive oligosaccharides are naturally occurring carbohydrate substituents (38), cell-associated gp160 and gp120 both appear to be completely endo D-resistant (Fig. 3C, lane 3). The finding that virtually all gp160 N-linked oligosaccharides are sensitive to endo H, yet remain completely resistant to endo D, raises the possibility that putative core N-linked mannose high chains (Mans9GlcNAc2-) associated with gp160 are not trimmmed by Golgi a-mannosidase I (35, 36), so that unsubstituted al+3 mannosyl residues requisite for endo D digestion cannot be generated. However, evidence is presented subsequently in this report demonstrating that at least some if not all of the gpl60-associated N-linked oligomannosyl side chains do in fact undergo Golgi a-mannosidase I trimming (see Fig. 8). Therefore, once generated, free 011+3 mannosyl residues must be short-lived, being rapidly modified by the addition of N-acetylglucosamine in a @1+2 linkage via Golgi N-acetylglucosaminyltransferase (35, 36). The same reasoning can also be applied to explain the observed resistance of gp160 HIV-l Envelope Precursor Cleaved in Go& gpl20-associated high mannose N-linked glycans to endo D digestion.
PNGase F is an enzyme with broad specificity that hydrolyzes high-mannose, hybrid, and bi-, tri-, and tetraantennary N-linked oligosaccharides directly at the glycosyl asparagine linkage, thus liberating polypeptides free of all N-linked oligosaccharide side chains (43). Cell-associated gp160 and gp120 immunoprecipitated with either HW or env 2-3 antibody were subjected to PNGase F digestion and subsequently analyzed by one-and two-dimensional gel electrophoresis. With env 2-3, two faster migrating protein species appeared with apparent molecular masses of approximately 90 and 60 kDa by onedimensional SDS-PAGE (Fig. 4, lane 4). Identical findings were also noted using HW (data not shown). Only the 90-kDa protein was detected with PB12 antibody (data not shown), verifying that this species represents deglycosylated gp160. On the other hand, the 60-kDa protein species is the PNGase F digestion product of gp120. Therefore, removal of all Nlinked oligosaccharides from gp160 and gp120 generates protein products with apparent molecular weights identical to those estimated for the respective envelope polypeptide backbones. Upon two-dimensional gel electrophoresis, the 60-kDa protein focuses as multiple discrete species in a pH range between 5.3 and 6.5 and the 90-kDa protein focuses as a contiguous species over a pH range of 5.8 to 7.5 (Fig. 5, B and C). Therefore, gp160 and gp120 both undergo an acidic shift in p1 subsequent to PNGase F hydrolysis of associated Nlinked glycans. Inasmuch as the naked polypeptide backbones of gp160 and gp120 are both predicted to have a basic p1 (approximately 8.0) according to their amino acid sequence (33), these findings indicate that significant negative charge modifications are associated with moieties other than complex N-linked oligosaccharides. During PNGase F-mediated hydrolysis of asparagine-linked oligosaccharides, the deglycosylated asparagine residues are converted to aspartic acid. In the LAV-la isolate, there are approximately 24 potential Nglycosylation sites on the polypeptide backbone of gp120 and 7 on gp41 (21, 33). If PNGase F-mediated aspartic acid conversion on gp160 and gp120 is taken into account, the predicted p1 values of the deglycosylated products are then calculated to be 7.4 and 7.1, respectively. Therefore the observed acidic shift in p1 noted for gp160 and gp120 after VB cells acutelv infected with HIV-1 were metabolically labeled with [35S]cysteine and [35S]methionine for 5 h. HIV-1 envelope determinants were immunoprecipitated from cell lysates using either HW or env 2-3. Selected immunoprecipitates were subjected to PNGase F digestion or to neuraminidase treatment followed by PNGase F digestion. Immunoprecipitates were analyzed by two-dimensional gel electrophoresis as per Fig. 1. A, HW, contrdl; B, HW, PNGase F-treated: C. env 2-3. PNGasetreated, D, env '2-3, neuraminidasetreated followed by PNGase F digestion.
PNGase F digestion (Fig. 5, B and C) can only be partially attributed to the generation of additional carboxylic acid residues on the peptide backbone.
Alternatively, sialylated O-linked glycans may also contribute net negative charge. O-Linked glycosylation has been observed to be a conserved feature of retroviral envelope products (44), where carbohydrate substituents are covalently bound to the hydroxy groups of serine and threonine residues (45,46). In a preliminary effort to search for sialylated Olinked glycans, cell-associated gp160 and gp120 immunoprecipitated with env 2-3 were first treated with neuraminidase and followed by digestion with PNGase F. PNGase F principally removes all N-linked oligosaccharides (43), suggesting that any resistant glycans represent O-linked substituents. One-dimensional SDS-PAGE analysis of the digestion products reveals two protein species of 90 and 60 kDa (Fig. 4, lane 5), indistinguishable from those generated by PNGase F alone (Fig. 4, lane 4). However, upon two-dimensional gel electrophoresis, the 90-and 60-kDa species generated by neuraminidase/PNGase F digestion both demonstrate a shift in p1 toward a more basic pH (Fig. 5D), compared with the 90-and 60-kDa species generated by PNGase F digestion alone (Fig.  5C). More specifically, the p1 of the PNGase F-derived 60-kDa protein is shifted from a pH range between 5.3 and 6.5 to a pH range between 5.6 and 6.8 consequent to neuraminidase digestion. Similarly, neuraminidase treatment causes a shift in the p1 of the PNGase F-derived 90-kDa protein from a pH range between 5.7 and 7.5 to a pH range between 6.3 and 7.5. The effect of neuraminidase on the p1 of the 90-and 60-kDa PNGase F digestion products suggests that both gp160 and gp120 contain sialylated O-linked oligosaccharides, inasmuch as PNGase F can be assumed to hydrolyze all N-linked glycans (43). Although unlikely, the possibility cannot be excluded that PNGase F digestion is not complete, and that the observed neuraminidase sensitivity simply reflects cleavage of sialic acid from sialylated complex N-linked oligosaccharides which are resistant to PNGase F. VB cells acutely infected with HIV-l were pulse-labeled with [?S]cysteine and [35S]methionine for 2 h. washed in PBS. and then chased for 1, 3, and 5 h in nonradioactive medium containing 10 @g/ml of cycloheximide (Calbiochem). Lysates were immunoprecipitated by HW and analyzed by two-dimensional gel electrophoresis as per Fig. 1. A, 2-h pulse (no chase); B, l-h chase; C, 3-h chase; and D, 5-h chase. The super&ant remaining after the 5h chase was clarified and Triton X-100 was added to a final concentration of 2%. HIV-1 proteins were then immunoprecipitated with HW and analyzed by two-dimensional electrophoresis (E). In E, 2% ampholytes (1% pH 2.5-5 and 1% pH 5-8) were employed for isoelectric focusing, and 10% polyacrylamide was employed for separation in the second dimension. the de novo synthesis of gp160 and the initial appearance of acidic sialylated forms of gp120 should approximate the total time required for gp160 to traverse the RER, undergo endoproteolytic cleavage, and enter the truns compartment of the Golgi complex. In an effort to estimate the duration of this interval, VB cells acutely infected with HIV-l were initially pulsed with [35S]cysteine and [35S]methionine for 2 h, followed by a l-, 3-, and 5-h chase in complete nonradioactive medium. Cycloheximide was also added during the chase to block de novo protein synthesis. HIV-l envelope moieties were immunoprecipitated from cell samples taken at each time point using HW and analyzed by two-dimensional gel electrophoresis. Fig. 6A reveals that after 2 h of radiolabeling, gp160 is clearly detectable with a p1 ranging from pH 7.0 to 7.5, whereas there is absolutely no evidence of synthesis of any gpl20 that focuses within a pH range between 5.0 and 7.5. However, after a l-h chase there is the sudden appearance of multiple species of gpl20 which resolve across an acidic pH gradient of 5.0-7.3 (Fig. 6B). The p1 of gp160 remains essentially unchanged. These findings indicate that it takes longer than two hours for the cleavage products of newly synthesized gp160 to be subject to extensive sialylation in the trans Golgi. Neuraminidase completely abolishes the acidic p1 of gp120 immunoprecipitated from infected cells radiolabeled for 5 h (Fig. 2), which substantiates the conclusion that the acidic gp120 species appearing in Fig. 6B are consequent to sialylation. One-dimensional SDS-PAGE analysis of the sample used in Fig. 6A reveals that endoproteolytic cleavage of gp160 occurs within 2 h of radiolabeling, before gp120 acquires sufficient sialic acid residues to be rendered acidic (data not shown, see Fig. 8, lane 7). However, half-life studies are required to more precisely determine the exact time required for endoproteolysis. In general, the rate limiting step in the transport of any newly synthesized glycoprotein to the cell surface appears to be at the level of exit from the RER (25,47,48). Although other explanations are plausible, the data presented in Fig. 6, A and B, are most consistent with the idea that the delay in the appearance of extensively sialylated kD -120 -@ -24 410 5!0 610 710 p,H gp120 (i.e. greater than 2 h) is secondary to prolonged retention of gp160 in the RER. Fig. 6, C and D, reveals that as the duration of the chase is extended to 3 and 5 h, gp120 continues to resolve as multiple species across a pH gradient of 5.ti to 7.3. Fig. 6E is a two-dimensional gel analysis of HIV-lassociated proteins immunoprecipitated with HW from the supernatant collected after the 5-h pulse-chase. Only p24 gag and gp120 appear as prominent proteins. The p1 range of the supernatant-associated gp120 is more acidic (pH 5.0-6.5) than that of cell-associated gp120 (pH 5.0-7.5), suggesting that mature g-p120 constitutes those species which are more heavily sialylated. The resolution of p24 gag into eight predominant forms is also notable. As shown in Fig. 6, gp160 appears to disappear as the chase progresses toward 3 h. This observation is consistent with the known precursor product relationship between gp160 and gp120. Cleavage of gp160 Occurs Proximal to the Site of Action of Fucosyltransferase Enzymes in the Go& Complex-The addition of crl+6-linked fucose to the most proximal core Nacetylglucosamine residue of iv-linked oligosaccharides confers endo H resistance by completely blocking the action of this enzyme (40). However, there are also endo H-resistant N-linked glycans which do not contain fucose. Nonetheless, the observation that gp160 does not display any endo H resistance, whereas gp120 does show significant partial resistance (see Fig. 3), is consistent with the notion that gp160 and gpl20 are differentially fucosylated. In an attempt to confirm this hypothesis, HIV-l envelope determinants were immunoprecipitated from acutely infected cells metabolically labeled with [3H]fucose and subjected to one-dimensional SDS-PAGE analysis. The results shown in Fig. 7, lane 4, reveal that while [3H]fucose is readily incorporated by gp120, no incorporation into gp160 is noted. These data, therefore, indicate that fucosylation of gp120 occurs subsequent to the endoproteolytic cleavage of gp160. The fact that fucosylation of N-linked oligosaccharides has been shown to localize the late medial and truns Golgi (35,36) further substantiates the conclusion that cleavage of gp160 occurs in or proximal to these compartments. VB cells acutely infected with HIV-1 were pulse-labeled with [35S]cysteine and ['"Slmethionine for 1 h in the presence (lanes 2, 4, 6, 8, and 10) or absence (lanes 1,3,5, 7, and 9) of 1 mM dMM. Those cells pulsed in the presence of dMM were also pretreated with the drug for 1 h before the addition of radiolabel. After the pulse, cells were washed, and chased for 0.5, 1, and 2 h in nonradioactive medium containing 10 rg/ml of cycloheximide. Those cells which were pulsed in the presence of 1 mM dMM were also chased with drug at the same concentration (lanes 2 4, 6, 8, and 10). HIV-l envelope determinants were immunoprecipitated from all lysates using HW. One-dimensional SDS-PAGE analysis was subsequently performed under reducing conditions in 8% polyacrylamide. The Effect of Inhibition of Golgi a-Mannosidase Z on Oligosaccharide Processing of gp160 and gpl20-VB cells acutely infected with HIV-l were first pulsed with 35S-labeled amino acids for 1 h, followed by a 05, l-, and 2-h chase in complete nonradioactive medium supplemented with cycloheximide. In a parallel experiment cells were treated with I-deoxymannojirimycin (dMM) (Genzyme), a specific inhibitor of Golgi (Ymannosidase I (49), for 1 h preceding radiolabeling and throughout the duration of the entire pulse-chase. HIV-l envelope substituents were then immunoprecipitated from cell samples obtained at each respective time point and analyzed by one-dimensional SDS-PAGE. The data in Fig. 8 (lanes 7 and 9) reveal that de nouo synthesized gp160 is converted to a slightly lower molecular weight form approximately 2-4 h subsequent to the addition of radiolabel, as evidenced by its faster migration in polyacrylamide. Most importantly, this observed loss in apparent molecular weight is completely abrogated in cells treated with dMM (lanes 8 and 10). Thus, it appears that gp160 is subject to significant Golgi Lu-mannosidase I-mediated trimming of N-linked oligomannosyl side chains. Available biochemical data indicate that the site of action of Golgi cu-mannosidase I localizes to a cis or medial Golgi compartment (35,36). Therefore, the observation that the susceptibility of gp160 to the enzymatic effects of Golgi Lu-mannosidase I does not predominantly manifest until 2-4 h subsequent to its initial de nouo synthesis provides supporting evidence that gp160 has a long residency in the RER. This is consistent with the findings described in Fig. 6, which indicate that it takes between 2 and 3 h for newly synthesized gp160 to traverse the RER, undergo endoproteolytic cleavage, and enter the tram Golgi network.
Although cleavage of gp160 is definitively evident within 2 h of its initial synthesis (Fig. 8, lane 7), the resultant species of gp120 are not extensively sialylated (see Fig. 6A). However, extensive sialylation does proceed subsequently (Fig. 6B). As shown in Fig. 8 (lane 9), gp120 undergoes a significant increase in apparent molecular weight between 2 and 4 h after the initial synthesis of gp160, which is represented at least in part by the addition of sialic acid residues. Cleavage of gp160 appears unaffected in cells treated with dMM. However, the inhibitory effects of dMM on Golgi a-mannosidase I do result in the generation of species of gpl20 with aberrant apparent molecular weight, as evidenced by faster migration in polyacrylamide (Fig. 8, lanes 8 and 10). The appearance of these aberrant low molecular weight species can be explained on the basis that dMM interferes with the addition of fucose and sialic acid required for the formation of complex N-linked glycans by inhibiting Golgi cu-mannosidase I-mediated trimming of high mannose oligosaccharide side chains.
Based on the data presented in Fig. 8, we propose that cleavage of gp160 occurs close to the site of action of Golgi (Ymannosidase I in the cis or medial Golgi cisternae (35, 36), since mannose trimming of gp160 shows a temporal correlation with the progressive appearance of gp120. Small quantities of apparent immature forms of gp120 are also encountered after 1 h of radiolabeling, which remain unaffected by dMM as assessed by molecular weight criteria (Fig. 8, lanes  3-6). This presumptive finding suggests that endoproteolytic cleavage of gp160 may occur proximal to the site of action of Golgi a-mannosidase I.

The gp160 HIV-1 Envelope Precursor Is Not Transported
to the Cell Surface in Acutely Infected CD4+ T Cells-Onedimensional SDS-PAGE analysis was carried out on proteins immunoprecipitated with either HW or anti-CD4 murine monoclonal antibody (OKT4) from acutely infected VB cells surface-labeled with lz51. Fig. 9 shows that although cell surface gp120 is readily detected with both HW (lane 4) and OKT4 (lane 2), there is a complete absence of gp160. The gp120 encountered with OKT4 can be explained on the basis of coprecipitation inasmuch as VB cells express high levels of surface CD4 (30). These data suggest that gp160 is not typically transported to the plasma membrane in CD4+ T cells acutely infected with HIV-l.
Evidence That gpl60 Binds to CD4-CD4 was immunoprecipitated from acutely infected VB cells radiolabeled with [""Slcysteine and [35S]methionine using murine monoclonal antibody (OKT4). In addition, envelope substituents were immunoprecipitated with HW which served as a molecular weight reference. Subsequent one-dimensional SDS-PAGE analysis reveals that both gp120 and gp160 are coprecipitated with anti-CD4 antibody (Fig. 10, lane 4) and do not appear in uninfected cells (data not shown). It is noteworthy that gp160 is coprecipitated by OKT4 far less efficiently than gp120. The protein migrating below gp120 in Fig. 10, lane 4, is believed to be cellular in origin in that it is typically found in uninfected cells as well. CD4 itself is noted at 55 kDa. With HW, gp160, gp120, gp41, and p24 are clearly detectable (Fig.  10, lane 2), as well as a 55kDa protein, which likely reflects coprecipitation of CD4 by anti-envelope antibodies. These data are consistent with previous reports which indicate that there is a specific binding interaction between soluble forms of gp160 and CD4 (50, 51). However, glycosylation patterns of soluble gp160 may not be representative of those species of gp160 synthesized in acutely infected cells. This is clearly evidenced in one particular case in which a soluble form of gp160 is demonstrated to lose approximately 30 kDa of its apparent molecular mass upon neuraminidase digestion (51). The fact that gp160 synthesized in acutely infected VB cells is nonfucosylated and marginally sialylated, yet still demonstrates putative binding to CD4, suggests that sialic acid moieties constitutive to complex carbohydrates may not be paramount to this molecular interaction.
gp160 Is Cleaved by a Non-acid-dependent Protease-The data in this report provide evidence that gp160 undergoes cleavage in neutral compartments of the Golgi complex. Therefore, the cellular protease responsible for cleavage of the gp160 polypeptide backbone would not be expected to be low pH-dependent. In order to confirm this hypothesis, neutralization of intracellular acidic vesicular compartments was accomplished using monensin, a carboxylic acid ionophore. We have previously established that maximal neutralization of these compartments (pH 6.4) is readily achieved in VB cells treated with monensin at a final concentration of 20 pM, and is sustained for greater than 4 h (30). Radioimmunoprecipitation of envelope substituents was therefore performed using acutely infected cells treated with 20 pM monensin. The resultant immunoprecipitates were then analyzed by onedimensional SDS-PAGE. The data in Fig. 11 (lane 4) reveal that 20 pM monensin only marginally inhibits the efficiency of proteolytic cleavage of gp160 in VB cells which were radiolabeled for 4 h. However, the putative gp120 cleavage product migrates with an apparent molecular mass of 100 kDa which is attributed to aberrant glycosylation. This observation is in agreement with the work of Dewar et al. (52) who report similar findings. The lOO-kDa protein is not reactive with PB12 antibody, indicating that its peptide backbone is that of gp120 (data not shown). In addition it is only slightly sensitive to neuraminidase (Fig. 11, lane 5) compared with g-p120 derived from untreated cells (Fig. 11, lane 3). The lOO-kDa protein resolves at a basic pH of 8.0 as determined by nonequilibrium pH gradient electrophoresis (data not shown). These findings demonstrate that although 20 pM monensin does not significantly affect the cleavage of gp160 in VB cells, it does disrupt tram Golgi-mediated sialylation of gp120. Other data not shown here reveal that the lOO-kDa protein is also incompletely fucosylated. It is well known that monensin can impede the transport of certain viral membrane proteins from the medial to trans Golgi cisternae (53). Therefore, the inhibition of sialylation and fucosylation of gp120 observed in monensin-treated VB cells suggests that HIV-l envelope translocation through the Golgi may be similarly blocked and is consistent with the notion that endoproteolytic cleavage of gp120 occurs in cis or medial cisternae.
The production of gp120 is clearly detectable in cells treated with effective neutralizing doses of chloroquine (500 PM) and NH&l (20 mM) (data not shown). The apparent molecular weight of the resultant gp120 species generated in the presence of these other lysosomotropic agents is indistinguishable from gp120 immunoprecipitated from untreated cells. Definitive species of gpl20 were detected as early as 2 h after labeling gp160 HIV-l Envelope Precursor Cleaved in Golgi of infected cells treated with 20 pM monensin, 500 pM chloroquine, or 20 mM NH&l (data not shown). We therefore conclude that endoproteolytic cleavage of gp160 is mediated by a non-acid-dependent protease. However, it has been reported that 40 mM NH&l inhibits the production of gp120 after a 2-h pulse-chase in a CD4+ T cell line (20). Based on this finding it was postulated that cleavage of gp160 requires an intracellular acidic compartment, or alternatively that NH&l interferes with the transport of gp160 to the intracellular site at which cleavage takes place. The data presented in this study are most consistent with the latter interpretation.

DISCUSSION
In this study we provide definitive evidence, based upon the physical characterization of N-asparagine-linked glycans, that endoproteolytic cleavage of the gp160 HIV-l envelope precursor occurs in the RER-Golgi complex. The end products of this cleavage reaction give rise to the formation of species of gpl20 and gp41 which subsequently undergo extensive Golgi-mediated carbohydrate modifications in order to generate mature envelope determinants. N-Asparagine-linked oligosaccharides contribute significantly to the respective molecular mass of both gp160 and gp120 (18,39).
Our results clearly show that gp160 is cleaved proximal to those compartments of the Golgi apparatus which are responsible for converting N-linked mannose type glycans into complex oligosaccharides. More specifically, upon steady-state [3H]fucose labeling, we found that gp120 was readily radiolabeled, while gp160 remained devoid of any detectable 3H incorporation (Fig. 7). These data indicate that endoproteolytic cleavage of gp160 must occur before the resultant cleavage products arrive in those compartments of the Golgi complex responsible for the fucosylation of N-linked glycans: i.e. the late medial and tram cisternae. In addition, gp120, unlike gp160, is also subject to extensive truns Golgi-mediated sialylation, which likely occurs subsequent to fucosylation, but definitively requires antecedent endoproteolytic cleavage of gp160 (Fig. 6). The observation that the HIV-l gp160 envelope precursor is cleaved in a prefucosyltransferase-containing compartment of the Golgi complex is not a unique finding, inasmuch as the envelope precursors of other retroviruses such as murine leukemia virus (gPr90) and RSV (Pr95) also require antecedent endoproteolytic cleavage in order to acquire susceptibility to Golgi-mediated fucosylation (26-28). HIV-l, therefore, differs significantly from the influenza virus, in that the influenza envelope precursor (HAJ is cleaved subsequent to exiting the Golgi, approximately at the same time of its insertion in the plasma membrane (29). Thus, the cleavage of gp160 itself does not play a direct role in the assembly of nascent HIV-l particles at the cell surface of productively infected cells.
The putative envelope precursor cleavage recognition sequences of RSV and HIV-l are both composed of basic tetrapeptides that share Lys-Arg residues as terminal amino acids (25). A novel Golgi endopeptidase, highly specific for lysine residues has been implicated in the cleavage of the Pr95 RSV envelope precursor (25). The observation that the envelope precursors of RSV and HIV-l both undergo endoproteolytic cleavage proximal to the site of action of fucosyltransferase enzymes in the Golgi complex suggests that a similar lysine-specific Golgi endopeptidase may be involved in the cleavage of gp160. Studies employing monensin indicate that the actual protease enzyme(s) responsible for the endoproteolytic cleavage of gp160 do(es) not depend upon low pH for activity (Fig. 11). This finding is consistent with the notion that gp160 is not cleaved in intracellular acidic vesicular compartments, but rather in neutral cisternae of the Golgi complex.
Pulse-chase experiments demonstrate that the initiation of Golgi a-mannosidase I mediated trimming of gp160 correlates temporally with the progressive generation of gp120 (Fig. 8).
Although the topographic organization of N-glycosylation within the Golgi complex exhibits cell type-specific variability (36), we conclude that cleavage of gp160 occurs close to the site of action of Golgi cY-mannosidase I, thus putatively localizing endoproteolysis to cis or medial Golgi compartments. However, the possibility cannot be excluded that there may be several sites within the RER-Golgi complex where cleavage can be executed.
The gp160 envelope precursor appears to have a long residence within the RER as judged by pulse-chase analysis, in that oligosaccharide trimming by Golgi a-mannosidase I does not become clearly evident until approximately 2 h following translation (Fig. 8). It is assumed that this long residence in the RER is a function of the time required for gp160 to undergo those modifications (i.e. correct folding, oligomerization, disulfide bond formation, and oligosaccharide trimming) prerequisite for its transport into the Golgi complex (28,29). This is supported by the work of Fennie and Lasky (54), who report that a truncated form of gp120 expressed in Chinese hamster ovary cells requires a relatively lengthy folding time in the RER to attain appropriate tertiary StNCture.
Our data indicate that gp160 is not typically transported and expressed on the surface of CD4+ T cells acutely infected with HIV-l (Fig. 9). However, there have been reports in which cell surface gp160 has been described in a variant clonal T cell line (50) as well as with genetic envelope recombinants expressed in cells of either non-T cell lineage (24) or nonhuman origin (51). We therefore propose that there are specific stretches within the HIV-l envelope coding region which, if mutated, result in aberrant trafficking of gp160 secondary to disruption of the integrity of important retention and sorting signals intrinsic to its molecular structure. This hypothesis is supported by studies with RSV, which demonstrate that a single amino acid change in the cleavage region of the envelope precursor (Pr95) results in significant changes in its glycosylation and intracellular routing, such that it acquires the ability to be expressed at the cell surface (25). Moreover, the trafficking of gp160 may also vary considerably as a function of the cell type in which it is being expressed, in that retention and sorting signals may be differentially processed. Therefore, although the transport of gp160 to the cell surface has been documented in certain systems, it would appear that such expression in acutely infected CD4+ T cells may not represent a general phenomenon and may be both viral isolate-as well as host cell-dependent. Lastly, the efficiency of endoproteolytic cleavage may also be a determining factor as to whether or not gp160 will be routed to the cell surface.
The functional importance of the observed hypersialylation of gp120 remains elusive. The removal of terminal sialic residues from mature secretory gp120 with neuraminidase has been reported to have no effect on the ability of gp120 to bind to CD4 (54). Moreover, our observation that gp160 is coprecipitated by anti-CD4 antibodies in acutely infected CD4+ T cells (Fig. lo), despite the fact that it is only marginally sialylated, supports the idea that abundant sialic residues are not essential to the binding interaction between gp120 and CD4. Alternatively, it is reasonable to postulate that the negative charge contributed by sialic acid residues may play a paramount role in maintaining a stable molecular associa-tion between gp120 and gp41 in the absence of any putative covalent disulfide bonds. It is of particular interest to note that the envelope of caprine-arthritis encephalitis virus (a  goat lentivirus) is also hypersialylated, and it has been shown that the envelope-associated sialic residues confer the virus with resistance to degradation by proteolytic enzymes as well as to neutralization by antibodies in postinfection serum (55). Therefore, it is conceivable that the hypersialylation of gp120 likewise suffices to protect HIV-l from host immune surveillance.
Presumptive data are provided suggesting that gp120 is modified with sialylated O-linked oligosaccharides (Fig. 5). The presence of O-linked glycans has been firmly established to be constituent to the envelope gene products of other retroviruses including murine leukemia virus and feline leukemia virus (44). Nonetheless, the exact functional role of these O-linked oligosaccharide side chains remains speculative at this time. 14.

18.
In summary, based on our findings, the following sequence of events can be envisioned during HIV-l envelope biosynthesis in VB cells. Initially, in the RER, the nascent envelope precursor polypeptide is modified by the transfer of core Glc3Man9GlcNAcs-residues to selected asparagine sites. The resultant gp160 product is then retarded within the RER for an indeterminate period of time which we estimate to be 2 h, before transport into the Golgi complex proper. Once in the cis and medial cisternae gp160 is subject to mannose trimming and endoproteolytic cleavage at approximately the same time. The cleavage products (gp12O/gp41) are soon after translocated to the late medial and tram Golgi where extensive complex carbohydrate modification of gp120 takes place. Mature envelope is then transported to the cell surface and incorporated into budding virions.