Synthesis of the Env Polyprotein

Env polyproteins are synthesized from a spliced version of the genomic RNA from which the gag-, pro-, and pol-coding sequences have been removed. In most cases, the initiation codon used for translation is encoded by the env gene and brought closer to the 5′end of the mRNA as a result of splicing. However, in the case of the ASLVs, the AUG used for Env synthesis is the same one that is used for the synthesis of the Gag and Gag-Pro-Pol proteins (Hunter et al. 1983; Fitch et al. 1984). In this case, the splice donor used to make the env mRNA falls just inside the gag gene (Hackett et al. 1982), and the initial Env product contains the first six amino acid residues of Gag.

Env is synthesized by the same machinery used by the cell for the synthesis of other surface and secreted proteins. Translation is initiated on free ribosomes, and very soon thereafter, the leader peptide begins to emerge. This amino-terminal segment of the env gene product varies in length from virus to virus (up to 20 kD in the case of FIV; Verschoor et al. 1993), but it invariably contains a short hydrophobic signal peptide (Fig. 9), which is recognized by signal recognition particle (SRP; for reviews, see Walter and Johnson 1994; Wolin 1994; Lutcke 1995). The SRP then docks on the surface of the RER, and the hydrophobic signal peptide is inserted into the lipid bilayer. The nascent Env protein is cotranslationally translocated across the membrane and into the lumen of the RER.

Figure 9

Organization of retroviral Env glycoproteins. Targeting of Env proteins to the ER is mediated by a small hydrophobic signal peptide (small black rectangle) contained within a larger leader sequence at the amino terminus of the nascent protein. The leader (more...)

Concurrent with translation, two modifications of the nascent Env protein occur, both of which are identical to those that take place on typical host-cell glycoproteins. First, the entire leader peptide sequence is proteolytically removed by the action of a cellular protease (signal peptidase) within the ER. Second, the nascent protein becomes glycosylated as identical blocks of preassembled oligosaccharides (glucose₃-mannose₉-N-acetylglucosamine₂) are transferred from a lipid intermediate (dolichol) to the nascent polypeptide (Hirschberg and Snider 1987; Abeijon and Hirschberg 1990). The sites of covalent attachment are asparagine (N) residues within the canonical sequence for N-linked sugars: Asn-X-Ser/Thr, where X is any amino acid except proline. The number and position of glycosylation sites vary widely among retroviruses, as would be expected from the high degree of variation in Env amino acid sequences. For example, the Env protein of HIV-1 has approximately 30 glycosylation sites, whereas the Env protein of MLV has only 6.

Modification of the N-linked sugars begins almost as soon as they are added to the protein. Specifically, all three of the glucose residues and one of the nine mannose residues on each block of sugars are removed in the ER by the sequential action of three cellular enzymes: glucosidase I, glucosidase II, and ER mannosidase. This trimming can result in a minor but noticeable shift of Env proteins to a lower apparent molecular weight during short pulse-chase experiments (see, e.g., Wills et al. 1984). At least some of the oligosaccharides must have important roles in the proper folding of Env because treatment of cells with tunicamycin, an inhibitor of the dolichol-sugar intermediate, typically results in unglycosylated Env molecules that are trapped in the ER (see, e.g., Stohrer and Hunter 1979; Pinter et al. 1984; Sarkar 1986; Delwart and Panganiban 1990; Pique et al. 1992). However, in some cases (e.g., MLV), individual sites of N-linked carbohydrate can be eliminated by mutagenesis without affecting the transport of the Env protein (Felkner and Roth 1992), whereas in other cases (HIV), mutagenesis results in a block only after transport to the Golgi (Fenouillet and Jones 1995). Similar results have been obtained with influenza virus hemagglutinin (see, e.g., Gallagher et al. 1992).

As the ribosomes approach the end of the env mRNA, a long (27–40 residues) and very hydrophobic amino acid sequence followed by a basic amino acid emerges to prevent the Env protein from being fully released into the lumen of the ER. If this transmembrane anchor is eliminated by introducing a stop codon immediately before its coding sequence, then the resulting env gene product is found free within the lumen and is secreted from the cell (Stephens and Compans 1986; Perez et al. 1987; Berman et al. 1988). The amino acids following the membrane anchor (from <30 to >100 residues, depending on the virus) extend into the cytosolic space (and later, the interior of the virion) and are referred to as the cytoplasmic tail. The residues on the luminal side of the membrane eventually reside on the outside of the cell (and virion) and constitute the external domains of the protein.

Folding and Oligomerization of Env

Before Env polyproteins can transit to the cell surface, they must be correctly folded and assembled into oligomeric structures (for reviews, see Einfeld and Hunter 1991; Saraste and Kuismanen 1992; Doms et al. 1993). The ASLV glycoproteins form trimers (Einfeld and Hunter 1988). While there has been much discussion in the literature on whether the HIV-1 glycoprotein forms dimers, trimers, or tetramers (Pinter et al. 1989; Schawaller et al. 1989; Doms et al. 1990; Earl et al. 1990; Owens and Compans 1990; Weiss et al. 1990; Thomas et al. 1991; Shugars et al. 1996), the most recent structural information suggests that this glycoprotein also forms a trimer (Chan et al. 1997). The oligomerization process is very slow in the ER and appears to be the overall rate-limiting step in transport to the cell surface (Einfeld and Hunter 1988). In the typical cases of ASLV and HIV-1, Env proteins have a half-time of transport of about 2 hours (Wills et al. 1984; Earl et al. 1991; Salzwedel et al. 1993). In contrast, the half-time for the influenza virus hemagglutinin protein, which also must oligomerize before leaving the ER, is only about 20 minutes (Gething et al. 1986; Gething and Sambrook 1990).

The lumen of the ER has two characteristics that make proper folding difficult. First, it is a very active site of synthesis in which proteins are constantly being made and exported. Because of this, there is always a high concentration of incompletely folded proteins of all types in close proximity to one another, and this increases the likelihood of unwanted aggregates forming on the transiently exposed hydrophobic faces. The second problem is that the lumen of the ER, unlike the cytosol, is not a reducing environment. Because of this, each cysteine that emerges from the ribosome can potentially form an undesirable disulfide bond with any other exposed cysteine, whether it is located within the same polypeptide chain or within an unrelated molecule.

To prevent undesirable hydrophobic interactions and assist in folding, there are high concentrations of chaperone proteins resident in the ER (for reviews, see Doms et al. 1993; Hammond and Helenius 1995). Among the best characterized of these are the immunoglobulin heavy-chain-binding protein (BiP; Haas 1994) and calnexin and calreticulin (Bergeron et al. 1994; Williams 1995), all of which have been demonstrated to interact with Env polyproteins (Earl et al. 1991; Otteken and Moss 1996). BiP has a peptide-binding site that recognizes exposed aliphatic residues on proteins as they emerge from the ribosome (Flynn et al. 1991; Blond-Elguindi et al. 1993). These interactions stabilize the folding- and assembly-competent states of the Env polyprotein until the individual subunits can find one another and properly associate. Calnexin and calreticulin are unrelated to BiP and “capture” nascent proteins in the ER by functioning as lectins with specificity for an early intermediate in the processing of N-linked oligosaccharides (glucose₁-mannose₉-N-acetylglucosamine₂); subsequent chaperone function is mediated through polypeptide interactions. Interactions of Env proteins with BiP and calnexin/calreticulin prevent incompletely assembled oligomers from being prematurely exported out of the ER.

To correct errors of incorrectly linked cysteines, the ER also contains protein disulfide isomerase. This enzyme cleaves all accessible disulfide bonds in nascent proteins to allow many new combinations of cysteine pairs to be tried (for review, see Noiva 1994). As the protein reaches its properly folded state and a corresponding low overall free energy, it becomes quite stable in the lumen of the ER and resistant to interactions with newly synthesized proteins that are still unfolded. As would be expected from the differences in the number and positions of cysteines found in the Env proteins of different viruses, their patterns of disulfide bonds vary greatly. In some cases (e.g., M-PMV and HIV), these bonds are normally found only within individual SU and TM domains (Bradac and Hunter 1986; Owens and Compans 1990), and the Env cleavage products are held together by noncovalent interactions (Helseth et al. 1991; Ivey-Hoyle et al. 1991; Brody and Hunter 1992; Brody et al. 1994a) which can be easily broken to allow shedding of SU (see, e.g., Bird et al. 1992). In other cases, disulfide bonds always (ASLV; Leamnson and Halpern 1976) or occasionally (MLV; Leamnson et al. 1977; Witte et al. 1977) form bridges between SU and TM.

The region of the Env polyprotein that participates in the assembly of oligomers seems to be contained entirely within the TM sequence (Doms et al. 1990; Earl et al. 1990; Thomas et al. 1991; Einfeld and Hunter 1994). TM mutants that abolish oligomerization remain trapped in the ER. In contrast, when the SU sequence is expressed in the absence of TM, it is secreted from the cell in a monomeric form (Einfeld and Hunter 1994), and mutants that prevent the stable interaction of SU and TM have been found which result in the release of biologically active SU from the cell surface (Brody et al. 1994a). In some cases, the TM protein can be transported to the cell surface in the absence of SU, but only if the proper oligomerization takes place (Einfeld and Hunter 1994). Interactions between TM molecules do not require the sequences of the hydrophobic transmembrane anchor or the cytoplasmic tail because these can be completely removed (Earl and Moss 1993; Hallenberger et al. 1993) or replaced with the glycosylphosphatidylinositol (GPI) membrane attachment signal without affecting the transport of the Env polyprotein to the cell surface or oligomerization (Gilbert et al. 1993; Salzwedel et al. 1993). The nature of the TM interaction is not completely understood. A leucine zipper-like motif is found within the HIV-1 gp41 ectodomain (Gallaher et al. 1989). Its oligomerization has been proposed to be involved both in gp160 oligomerization and in a conformational change in Env required for fusion (Dubay et al. 1992a; Wild et al. 1994; Bernstein et al. 1995; Poumbourios et al. 1995).

Premature Receptor Interactions

By the time the Env polyprotein is properly oligomerized in the ER, it is already competent for interaction with its receptor, even though several modifications remain to take place along the secretory pathway. This has been demonstrated in several ways. For example, cells that are normally susceptible to reticuloendotheliosis virus (REV) become resistant to infection when they express an Env derivative that is trapped in the ER (Delwart and Panganiban 1989). Similar results have been obtained with murine retroviruses (Matano et al. 1993). Hence, interaction of the receptor with the mutant form of Env prevents transport of the receptor to the cell surface.

In the case of HIV-1, a special mechanism involving the Vpu protein serves to prevent premature Env-receptor interaction. Transport of the fully folded Env polyprotein is blocked when coexpressed with forms of the CD4 receptor that have been modified to contain an ER retention signal (Buonocore and Rose 1990; Raja et al. 1993). Coexpression of the wild-type receptor and Env also results in inefficient transport of both molecules, which end up in complexes as demonstrated by direct coimmunoprecipitation experiments (Crise et al. 1990; Jabbar and Nayak 1990; Bour et al. 1991; Crise and Rose 1992). The Vpu protein of HIV-1 has the ability to bind to the cytoplasmic domain of CD4 and induce its degradation (for review, see Jabbar 1995; see also Willey et al. 1992a,b, 1994; Chen et al. 1993; Vincent et al. 1993; Raja et al. 1994; Schubert and Strebel 1994; Bour et al. 1995; see also below Assessory Proteins and Assembly), even in the absence of the HIV-1 Env protein (i.e., when CD4 is retained in the ER by treating the cells with the transport inhibitor brefeldin A; Willey et al. 1992a). Down-regulation of CD4 by Vpu allows the fully oligomerized Env protein to be transported to the cell surface and packaged into the progeny virions.

Mechanisms to prevent intracellular interactions between Env and its receptor have not yet been described in other retroviral systems. In those cases where the steady-state level of Env exceeds that of its receptor (e.g., ASLV; Bates et al. 1993; Young et al. 1993; Gilbert et al. 1994), interactions between the two provide an effective mechanism to block superinfection by other viruses that utilize the same receptor (i.e., interference). Moreover, expression of Env in the absence of other viral proteins confers resistance (see, e.g., Federspiel et al. 1989). In such cases, there is no need to prevent Env-receptor interactions along the secretory pathway because the excess Env protein is free for transport to the cell surface and packaging into virions.

Transport, Cleavage, and Further Modifications of Env

Although Env proteins become competent for receptor interactions soon after they oligomerize, they do not become fusogenic until much later, after cleavage occurs to separate the SU sequence from the TM (Perez and Hunter 1987; Freed et al. 1989; Dong et al. 1992b). This exposes the fusion peptide, which is very hydrophobic and located at the amino terminus of TM (Fig. 9) (Gallaher 1987; Bosch et al. 1989). Cleavage occurs in the Golgi apparatus (Wills et al. 1984; Stein and Engleman 1990; Bedgood and Stallcup 1992), and the cellular protease thought to be responsible is furin, or a closely related enzymatic activity (Hallenberger et al. 1992; E.D. Anderson et al. 1993; Morikawa et al. 1993; Decroly et al. 1994; Ohnishi et al. 1994; Gu et al. 1995). These endoproteases are related to bacterial subtilisin and are encoded by the host cell (for reviews, Barr 1991; Van de Ven et al. 1993). They reside in the lumen of the Golgi where they cleave immediately after the basic sequence Arg-X-(Lys/Arg)-Arg in proteins that possess this motif, which includes all of the Env proteins. The transport of oligomerized Env proteins to and through the cis, medial, and trans compartments of the Golgi apparatus is vesicular in nature and driven entirely by cell-encoded mechanisms.

As mentioned earlier, modification of N-linked oligosaccharides actually begins in the ER. The subsequent changes in sugar composition that take place as the Env proteins travel through the Golgi apparatus are similar to those experienced by typical cellular glycoproteins. These changes are extensive and often heterogeneous from one carbohydrate chain to the next. What occurs typically is a sequential removal of three of the eight remaining mannose residues (by Golgi mannosidase I), the addition of a single residue of N-acetylglucosamine (by N-acetylglucosamine transferase I), and the removal of another mannose residue (by Golgi mannosidase II). At this point, the Env protein resides in the medial compartment of the Golgi apparatus, and the modified oligosaccharides are resistant to removal by Endo H, a highly specific endoglycosidase. The conversion of a glycoprotein to the Endo-H-resistant state is readily monitored by shifts in electrophoretic mobility; however, in the case of Env glycoproteins, this type of analysis is sometimes difficult to follow due to the large number of oligosaccharides and the failure of all of them to be completely processed to an Endo-H-resistant form by the cellular modification enzymes.

As the Env glycoprotein is transported further, yet another mannose is removed (by Golgi mannosidase II), leaving just three of the original nine. Subsequently, many additional sugars (N-acetylglucosamine, galactose, sialic acid, and fucose) are added while the protein is in the trans Golgi compartment. These additions are quite variable in number, further contributing to the heterogeneity of the final Env products. Thus, whereas the initial Env polyprotein (in the ER) is a relatively homogeneous species, the final products (SU and TM) invariably migrate as broad fuzzy bands when analyzed by denaturing polyacrylamide gel electrophoresis.

Both O-linked glycosylation and sulfation of Env glycoproteins have been shown to occur, but their significance is unknown. O-linked carbohydrates are typically attached to the hydroxyl group of serine and threonine residues after transport of proteins to the Golgi, and this modification has been identified within certain Env proteins including that of HIV-1 (Pinter and Honnen 1988; Bernstein et al. 1994). Sulfation has been demonstrated for the Env proteins of HIV-1, HIV-2, and simian immunodeficiency virus (SIV) (Bernstein and Compans 1992). Although sulfate is known to be added to tyrosines in some proteins, in the case of Env, it appears to be added to N-linked oligosaccharides. Removal of carbohydrate with N-glycosylase F also removes all of the sulfate from the Env protein. However, sulfate is not completely removed by treatment with Endo H, which suggests that modification takes place after the protein is transported to the Golgi apparatus, an idea supported by the finding that sulfation is inhibited by brefeldin A (Bernstein and Compans 1992), an inhibitor that blocks transport from the ER to the Golgi (Lippincott-Schwartz et al. 1989; Pelham 1991). Although sulfation increases the negative charge of the Env protein, there is no evidence at the moment to suggest that this modification is important in the replication of retroviruses.