The role of human immunodeficiency virus type 1 envelope glycoproteins in virus infection.

Enveloped viruses enter cells by a two-step process. The first step involves the binding of a viral surface protein to receptors on the plasma membrane of the host cell. After receptor binding, a membrane fusion reaction takes place between the lipid bilayer of the viral envelope and host cell membranes. This fusion reaction can occur either at the plasma membrane or in acidic endosomes following receptor-mediated endocytosis. In either case, the membrane fusion reaction delivers the viral nucleocapsid into the host cytoplasm, allowing the infection to proceed. Viral proteins embedded in the lipid bilayer of the viral envelope (known variously as surface, spike, or envelope proteins) catalyze receptor binding and membrane fusion reactions. The critical involvement of these viral proteins in receptor binding and membrane fusion has stimulated intensive investigation aimed at understanding the mechanisms by which these proteins function. In this article, we provide a brief overview of the roles envelope (Env) glycoproteins play in the human immunodeficiency virus type 1 (HIV-1) life cycle.

tion sites in gp120 from the HTLV-III B HIV-1 isloate and at least three of the five sites in the ectodomain of gp41 appear to be utilized ( Fig. 1; 4 , 14). It has been suggested that HIV-1 Env also contains O-linked carbohydrates (15).
Proteolytic cleavage of gp160 in the Golgi is inefficient (12) and is catalyzed by a host protease at a Lys/Arg-X-Lys/Arg-Arg motif (where X is any amino acid) that is highly conserved among viral Env glycoprotein precursors (16 -19). Several studies have suggested that the host enzyme responsible for cleaving gp160 (and other viral Env precursors) is furin or a furin-like protease (20,21). Other enzymes may also be capable of mediating gp160 precursor processing, since cleavage can occur in a furin-deficient cell line (22), and a basic pair of amino acids at the cleavage site is not absolutely required for gp160 processing (18). Following gp160 cleavage, the oligomeric, noncovalently associated gp120-gp41 complexes are transported to the cell surface, where they are incorporated into budding virions.

Env Incorporation into Virus Particles
Because of the role HIV-1 Env plays in receptor binding and membrane fusion (see below), the virion incorporation of Env is essential for the formation of infectious virus particles. In certain virus systems (e.g. the alphaviruses), an interaction between the Env protein intracytoplasmic tail and the viral capsid has been demonstrated directly, and this interaction is required for virus release (23)(24)(25). In the case of retroviruses, which do not require Env expression for virus assembly and release (for review, see Ref. 26), the picture is less clear. Evidence derived from Env and Gag mutagenesis and pseudotyping studies has accumulated over the past decade both for and against the existence of a specific interaction between the TM cytoplasmic tail and the matrix protein (MA), which forms the membrane-proximal component of the retroviral core (27)(28)(29)(30)(31)(32)(33)(34)(35). In a recent study, it was demonstrated that mutations in HIV-1 MA that blocked the virion incorporation of full-length HIV-1 Env did not affect the incorporation of heterologous retroviral Env glycoproteins with short cytoplasmic tails or HIV-1 Env mutants containing large truncations in the gp41 cytoplasmic tail (36). This latter finding implies that the incorporation of Env glycoproteins with long cytoplasmic tails (i.e. lentiviral Env glycoproteins) depends upon a specific interaction between sequences in the cytoplasmic tail of the TM glycoprotein and the HIV-1 MA, whereas the incorporation of Env glycoproteins with short cytoplasmic tails into HIV-1 virions does not (36).

CD4 Binding
The initial step in HIV-1 infection involves the binding of virionassociated gp120 to the cell surface molecule CD4, which serves as the major receptor for HIV-1 and the related HIV-2 and simian immunodeficiency viruses (SIVs) (37)(38)(39). The Env determinants of CD4 binding map to gp120, in particular C3 and C4 (40 -42). CD4 binding to gp120 induces conformational changes in both gp120 and gp41 that result in the exposure of Env domains (see below) that are thought to be involved directly in the membrane fusion reaction (43)(44)(45)(46).
Following the identification of CD4 as the primary receptor for HIV, it was determined that soluble CD4 (sCD4) could neutralize virus infectivity (47)(48)(49)(50)(51). This neutralization was demonstrated to be primarily a result of an enhanced shedding of gp120 from virions following treatment with sCD4 (52)(53)(54). Initially, it was suggested that the ability of sCD4 to neutralize HIV-1 might be exploited therapeutically. Unfortunately, however, primary, non-laboratoryadapted HIV-1 isolates are neutralized poorly by sCD4 (55,56), in part as a result of the relative resistance of primary isolates to sCD4-induced gp120 shedding (57)(58)(59)(60), thereby diminishing the utility of sCD4 as a therapeutic agent. In fact, in related HIV-2 and SIV systems, sCD4 has been reported to actually enhance virus infectivity (61). It is currently unclear what role, if any, CD4induced gp120 shedding plays in HIV-1 Env function (62,63).
In addition to binding CD4 on the cell surface during the early phase of virus infection, HIV-1 Env associates with CD4 intracellularly soon after gp160 synthesis in the ER. Pulse-chase and transport inhibition studies suggested that within approximately 30 min of synthesis, gp160 adopts a conformation suitable for CD4 binding (64). The association of Env and CD4 early in the transport pathway leads to the down-regulation of CD4 expression from the surface of Env-expressing cells (65)(66)(67)(68). This decrease in the level of cell surface CD4 may reduce the ability of Env-expressing cells to become infected with additional virions (66), a phenomenon, described for other retroviruses (69), known as superinfection interference. It has also been proposed that the association between Env and CD4 early in the transport pathway may be cytotoxic (70,71).

Membrane Fusion
The ability to induce fusion between the lipid bilayer of the viral envelope and host cell membranes is a central feature of Env glycoprotein function. Env expression in an infected cell can also lead to cell-to-cell fusion, or syncytium formation, with neighboring CD4 ϩ cells, a process that contributes to HIV cytopathicity in culture and possibly in vivo (72)(73)(74). In addition to domains required for gp160 proteolytic cleavage and CD4 binding (discussed above), a number of determinants in both gp120 and gp41 have been postulated to play a role in membrane fusion.
The first domain recognized as being directly involved in HIV-1 Env-induced membrane fusion was the highly hydrophobic sequence at the amino terminus of gp41. A number of studies involving the hemagglutinin protein of orthomyxoviruses (i.e. influenza A) and the F protein of paramyxoviruses had previously suggested that an analogous domain, referred to as the fusion peptide, plays a role in the membrane fusion function of these proteins (for reviews see Refs. 19 and 75-79). Analysis of lentiviral Env glycoproteins indicated that single amino acid changes in the highly hydrophobic amino termini of the HIV-1, HIV-2, and SIV TM glycoproteins blocked Env-induced syncytium formation (80 -84). An extensive literature, reviewed elsewhere (76,78,79), details the irreversible conformational changes in the influenza virus hemagglutinin and paramyxovirus F proteins that lead to the exposure, or "activation" of the fusion peptide. As mentioned above, the HIV-1 Env glycoprotein also undergoes a series of conformational changes following CD4 binding, one outcome of which is the exposure of the gp41 fusion peptide (43). Several fusion peptides may act in concert to destabilize the lipid bilayer of the target membrane by forming a "fusion pore" between the two bilayers (77). The hypothesis that Env glycoproteins behave cooperatively to promote membrane fusion derives support from the finding that substitutions of polar amino acids in the fusion peptide of HIV-1 gp41 elicit a transdominant negative effect on syncytium formation and virus infectivity (85,86) and from studies demonstrating a synergistic effect of high levels of sCD4 on HIV-1 neutralization (87).
In addition to the amino-terminal fusion peptide, other domains within gp41 have been reported to play a role in the fusion process. Mutations in a putative leucine zipper motif in the gp41 ectodomain (88) blocked syncytium formation and virus infectivity without affecting Env oligomerization, transport, processing, or CD4 binding (89,90); a peptide based on the sequence of this region also inhibited syncytium formation and virus infection (91). Substitution of two charged residues in the membrane-spanning domain of gp41 also perturbed Env-induced membrane fusion (92).
In a number of studies, deletion, frameshift, or premature translation termination mutations were introduced into the cytoplasmic tails of the HIV-1, HIV-2, or SIV TM glycoproteins (40, 93-102), which, as noted above, are unusually long compared with those of other retroviruses. In some cases, these deletions enhanced Envinduced membrane fusion, suggesting that sequences in the gp41 cytoplasmic tail may modulate Env fusogenicity (94,96,97,99,101,102).
In gp120, primarily two regions are involved in membrane fusion. A number of studies determined that antibodies to V3 were capable of neutralizing virus infectivity (103-109) without affecting virus binding to CD4 (108,109). Mutational analyses demonstrated that single amino acid substitutions within the HIV-1 V3 loop, and the analogous domain of HIV-2, blocked Env-induced syncytium formation (83,110,111) and virus infectivity (83,112). More recent studies have also implicated the V1/V2 region in membrane fusion. Mutations within V1/V2 were reported to block syncytium formation without affecting the gp120-gp41 interaction or CD4 binding (113), and the transfer of V2 sequences from syncytium-inducing Env glycoproteins conferred the ability to induce fusion on non-syncytium-inducing Env glycoproteins (114,115). Consistent with a role for V1/V2 in membrane fusion, antibodies to this region are capable of neutralizing virus infectivity (116,117).
It has been postulated for a number of years that molecules other than CD4 may be necessary for membrane fusion induced by HIV-1 Env. The following observations suggest that factor(s) provided by human cells are required for HIV-1 Env-induced membrane fusion: (i) expression of human CD4 in murine cells does not confer upon them the ability to support HIV-1 infection (39), (ii) in a cell-fusion reaction, the target cell must be of human origin, whereas the Env-expressing cell can be of non-human origin (118), and (iii) the formation of some somatic cell hybrids between human cells and CD4-expressing non-human cells can overcome the fusion defect observed in human CD4-expressing non-human cells, suggesting that the inability of CD4 ϩ murine cells to support HIV infection is due to the absence of factor(s) on murine cells, rather than the presence of a mouse cell-specific interfering function (119 -121). Although non-CD4 molecules have been reported to serve as alternative HIV-1 receptors on CD4 Ϫ cells (122), no widely accepted CD4 co-receptor has been identified. It was suggested by Callebaut et al. (123) that CD26 (dipeptidyl-peptidase IV) conferred susceptibility to HIV-1 infection upon CD4-expressing murine (NIH 3T3) cells. A number of groups, however, failed to confirm a role for CD26 in HIV-1 infection or syncytium formation (124 -128). Recent protease digestion data suggest that the factor(s) provided by human cells may be nonproteinaceous (129).

Tissue Tropism
An additional function of the HIV-1 Env glycoprotein is to determine the cell-type specificity, or tissue tropism, of virus infection. In culture, HIV-1 typically infects either cells of the monocyte/ macrophage lineage or immortalized T-cell lines, but rarely both. Primary virus isolates obtained from infected individuals during the early, asymptomatic phase of infection are frequently nonsyncytium-inducing and macrophage-tropic, and cells of the monocyte/macrophage lineage are thought to be important targets for virus infection in vivo (for review, see Ref. 130). HIV-1 isolates which are syncytium-inducing and capable of productively infecting T-cell lines tend to arise late in infection after the onset of AIDS-defining symptoms (131). In fact, it has been argued that the evolution in vivo of syncytium-inducing, T-cell line-tropic variants may play a causal role in disease development (74). As would be predicted for a property determined by Env, the block to infection in nonpermissive cells appears to be primarily at the level of entry, presumably resulting from a defect in membrane fusion (132,133). Interestingly, both macrophage-tropic and T-cell line-tropic isolates are capable of efficiently infecting primary human CD4 ϩ T-lymphocytes.
Studies conducted in a number of laboratories have concluded that sequences within gp120 are responsible for determining the tissue tropism of HIV-1. The V3 loop, discussed above in the context of membrane fusion, plays a central role in tropism. The introduction of sequences encompassing the V3 loop from macrophage-tropic clones to T-cell line-tropic clones is able to confer macrophage tropism upon certain T-cell line-tropic clones (134 -138). It is clear, however, that a combination of sequences both within, and outside, V3 is required for optimal macrophage infec- tion. The exchange of additional sequences adjacent to V3, particularly amino-terminal to V3, greatly enhances the ability of chimeric viruses to infect macrophages (139,140). It appears that V3 loop conformation differs between macrophage-tropic and T-cell line-tropic Env glycoproteins (140,141), and that residues both within and outside V3 influence V3 conformation (142).
The mechanism by which the V3 loop affects cell-type tropism has not been elucidated. It has been suggested by some that a sequence near the tip of the V3 loop serves as the target for proteolytic cleavage, and that this V3 cleavage event activates the membrane fusion potential of HIV-1 Env (143,144). It was proposed that differences in tropism correlate with altered sensitivity of the V3 loop to proteolytic cleavage (141,145), although other investigators failed to find a correlation between these properties (146), and it has not been established that V3 cleavage plays any role in HIV-1 infection. Another model for V3 loop function invokes the existence of a cell-type specific, non-CD4 receptor molecule with which the V3 loop interacts. To date, however, no such receptor has been identified.

Env Interactions
In the discussion above, we have focused largely on the functions of discrete domains within the HIV-1 Env glycoprotein. It is becoming increasingly clear, however, that most Env functions require the interaction between nonadjacent sequences within gp120 or between gp120 and gp41. Although attempts to obtain a crystal structure of HIV-1 Env have thus far been unsuccessful, a variety of biological, biochemical, and immunological data have provided information about Env interactions. In an early demonstration of the importance of Env interactions, a debilitating mutation in C2 affecting infectivity could be reversed by changes in C1 and V3, suggesting a functional interaction between these domains of gp120 (147). More recently, the analysis of a revertant obtained from a V1/V2 envelope chimera demonstrated the existence of a functional interaction between V1/V2 and C4 (148). The analysis of chimeras between syncytium-inducing and non-syncytium-inducing Env glycoproteins also support an interaction between V1/V2 and sequences near the carboxyl terminus of gp120 (114).
Several groups have used antibody binding analyses to identify interacting regions within gp120. This approach has provided evidence for interactions between V1/V2 and C4; V3 and C1, C2, and C4; and C1, C2, and C5 (117, 149 -153). An interaction between V3 and C4 is further supported by the observation that treatment of gp120 with soluble CD4 (which binds C4) enhances the binding of anti-V3 monoclonal antibodies (152), and that single amino acid mutations in C4 increase binding by a V3-specific antibody (153). Although some of these findings may be ascribed to indirect effects on protein folding rather than direct, functional interactions, these studies are consistent with the concept that while distinct domains within HIV-1 Env are involved in specific functions, complex interactions between, and within, these domains are essential for the full range of biological activities required for productive infection.