Human immunodeficiency virus (HIV) particles enter targeted cells through fusion of the viral lipid envelope with the cellular plasma membrane. Envelope glycoprotein spikes on the HIV surface are needed for this targeted fusion. Targeting is specified by interactions between gp120 and cellular receptor molecules, CD4 and chemokine receptors, which are co-opted from normal immunological functions. Fusion is catalyzed by the gp41 component in structural rearrangements triggered by the binding of gp120 to the receptors. Although the HIV envelope is antigenic, immune responses elicited by HIV infection are unable to control the virus. We expect structural studies to provide insight into mechanisms of cell invasion and immune evasion by HIV and guide the design of therapeutics and vaccines.
We are engaged in structural analyses of HIV gp120 and its interactions with the cellular receptors and the immune system. Much of the work involves x-ray crystallography, but we also carry out solution studies of the interactions. Progress has been made in the crystallographic analysis of gp120 molecules in complexes with receptors and antibody ligands (with Peter Kwong, Columbia; Joseph Sodroski and Richard Wyatt, Dana-Farber Cancer Institute; Raymond Sweet, SmithKline Beecham; and James Robinson, Tulane University). Progress is also being made in characterizing the energetics of interactions between gp120 and its ligands in solution (with Michael Doyle, SmithKline, and David Myszka, University of Utah). Other studies are directed at intervention in the HIV-CD4 interaction (with Irwin Chaiken, University of Pennsylvania, and James Samanen, SmithKline). (These studies are supported in part by grants from the National Institutes of Health.)
Structure of Core gp120
To prepare crystals of gp120, we removed certain loops and extensions of the molecule and nearly all of the heavy glycosylation of the protein. The resulting core gp120 molecules retain affinity for CD4 and many antibodies, and crystallization has also required complex formation with such ligands. Our first usable crystals were grown from core gp120 produced from the HXBc2 strain in a complex with the D1D2 two-domain fragment of CD4 and the antigen-binding fragment (Fab) of the 17b antibody. This structure has now been refined at 2.2-Å resolution. Analysis is also in progress for a core gp120 from the primary isolate Yu2, also in complex with CD4 (D1D2) and 17b (Fab).
Core gp120 comprises two major domains, inner and outer, and a number of excursions from the body of these domains. Two major excursions, one from each domain, form a bridging β-sheet structure that appears to require the CD4 interaction for its integrity. The glycosylation sites on gp120 are localized to surfaces outside the ligand-binding sites. From the orientation of CD4 as bound to gp120 and the location of the gp120 carboxyl terminus, the complexed structure can be placed plausibly relative to the viral and cell surfaces.
Although CD4 and gp120 associate with high affinity, there are comparatively few direct contacts in the structure. The relatively limited surface area buried into the interface includes two large interfacial cavities and involves elements from both inner and outer domains of gp120 as well as from the bridging-sheet minidomain. Key residues involved in the interaction are located together at the nexus of CD4 with the three gp120 domains. A remarkable 60 percent of the gp120 contacts with CD4 are from main-chain atoms.
Antibody and Chemokine-Receptor Interface
The 17b antibody binds to a CD4-induced epitope and, in keeping with this property, the gp120 surface to which the 17b Fab binds seems to depend on CD4 for its conformation. Chemokine-receptor binding to gp120 is likewise potentiated by prior binding to CD4, and CD4-induced antibodies such as 17b compete with the chemokine receptor. A mutational study, carried out on regions of Yu2 gp120 that are contiguous with the 17b-binding surface, has defined the binding site for the CCR5 chemokine receptor. This surface is highly conserved among HIV strains, and much of it is positively charged. The V3 loop of gp120, which was removed to form core gp120, is also critical for the interaction with chemokine receptors, and the acquisition of additional positively charged residues in the V3 loop is associated with the evolution of HIV strains from CCR5 to CXCR4 utilization.
Model of Trimeric Structure
Recombinant gp120 is monomeric in solution and in our crystals. We know, however, that gp120 is an oligomer in association with gp41 on the viral surface. Recent data indicate that HIV envelope spikes are trimeric. We have constructed a rough but instructive model of gp120 trimers. Our modeling constraints restricted carbohydrate from the interface, maximized conservative residues at the interface, and assured accessibility to ligands that bind to gp120 on virion spikes. This procedure was validated by application to influenza virus hemagglutinin, where the trimeric structure is known. Electrostatic features of this model correlate with chemokine-receptor–binding characteristics of various HIV strains.
Energetics of gp120-CD4 Binding
We have not been able to obtain crystals of gp120 molecules in the absence of ligands, which indicates that free gp120 might not be fully ordered. This suggestion is validated by thermodynamic results obtained by microcalorimety and evidence of conformational change from circular dichroism. The binding of CD4 to gp120 involves a dramatic 50 kcal/mol entropic penalty and high temperature dependence in heat capacity. These characteristics are of an unprecedented magnitude for protein-protein interactions but are similar to those that accompany protein folding, suggesting that a conformational ordering accompanies the complexation. Since the structure of CD4 is essentially the same in the complex as when free, the structural ordering that occurs must be in gp120.
Evasion of Immune Responses
Although HIV infection elicits a robust antibody response, few of these antibodies are effective in viral neutralization. Many bind to variable loops or to conserved surfaces that are only seen in monomeric gp120 shed from virions. Moreover, the more broadly neutralizing antibodies, such as 17b, tend to be raised only late in the course of disease. The structural properties of gp120 coupled with an analysis of epitopes for various antibodies suggest a variety of mechanisms by which HIV is able to evade humoral immune responses. It seems likely that the entropy of free gp120 makes it a moving target for the immune system. Surface features that are conserved for receptor binding are thereby hidden from the immune system until formed at the moment of binding to the target cell. Variable loops such as V2 may also mask receptor sites. Nearly all of the exposed, nonreceptor surface in the gp120 trimer is clad in a carbohydrate cloak and is essentially devoid of antibody epitopes. These carbohydrate moieties also occur on host proteins and are presumably therefore recognized as self. The CD4-binding surface is dominated by gp120 main-chain interactions, and the major interfacial cavity covers a highly variable surface. Mutations here could readily limit the activity of antibodies elicited to this surface to a narrow range of strains.
Mechanism of Cell Invasion
Structures of the envelope glycoproteins give clues about the mechanism of cell entry by HIV, but much about the suggested mechanisms remains speculative. We especially need information about the ectodomain of HIV envelope trimers. Since CD4 and many antibodies bind both to monomeric gp120 and to the virus, we can now surmise much about this aspect of the trimeric spike. The gp41 component, however, is likely to be structured very differently in the trimer than in isolation. The more thoroughly studied case of influenza virus is a helpful model. The low-pH form of influenza virus HA2 has characteristics like those of recombinant HIV gp41, whereas HA2 is radically different in the hemagglutinin ectodomain trimer. Despite gaps in our knowledge, structural evidence provides an outline of a mechanism for receptor-specified cell invasion by HIV. Trimeric gp120 on the virion probably has substantial interdomain flexibility and poorly structured loops. Binding to CD4 fixes the gp120 structure to form the conservative, positively charged chemokine-receptor–binding surface; this also properly positions gp120 for binding to the chemokine receptor. This binding somehow mechanically triggers a conformational change in gp41 at the place and time where the gp41 fusion peptide can enter the cellular membrane to initiate fusion to the viral envelope. Thereby, the viral contents become one with the cell.