Rational design of vaccines to elicit broadly neutralizing antibodies to HIV-1

Abstract

The development of a highly effective AIDS vaccine will likely depend on success in designing immunogens that elicit broadly neutralizing antibodies to naturally circulating strains of HIV-1. Although the antibodies induced after natural infection with HIV-1 are often directed to strain-specific or nonneutralizing determinants, it is now evident that 10%-25% of HIV-infected individuals generate neutralizing antibody responses of considerable breadth. In the past, only four broadly neutralizing monoclonal antibodies had been defined, but more than a dozen monoclonal antibodies of substantial breadth have more recently been isolated. An understanding of their recognition sites, the structural basis of their interaction with the HIV Env, and their development pathways provides new opportunities to design vaccine candidates that will elicit broadly protective antibodies against this virus.

Figure 1.

Mechanistic and atomic-level details of the HIV-1 viral spike, a fusion machine that also evades humoral detection. (Top row) Entry schematic. The viral spike is composed of three gp120 envelope glycoproteins (cyan) and three gp41 trans-membrane (red). In the prefusion conformation, the outer surface of the spike is covered with N-linked glycan, which is seen as “self” by the humoral immune system therefore virtually invisible to potentially neutralizing antibody. At the cell surface, binding to CD4 (yellow) induces large structural rearrangements, which include the formation of a binding site for a second requisite coreceptor (purple). Coreceptor binding induces a transient intermediate, with the amino-terminal fusion peptide of gp41 thrown into the target cell membrane while the carboxy-terminal gp41-trans-membrane region is buried in the viral membrane. Subsequent spike rearrangements resolve into a stable postfusion conformation, with the fusion peptide and trans-membrane regions of gp41 in close proximity. (Bottom row) Atomic-level structures, with polypeptide backbones shown in ribbon representation and N-linked glycans in stick representation. The unliganded conformation of HIV-1 gp120 in its viral spike conformation is unknown, but an atomic model has been solved for an SIV core (Chen et al. 2005), in which the N-linked glycans (cyan) cluster onto one face of gp120. The CD4-bound conformation of gp120 (Kwong et al. 1998; Huang et al. 2005) contains a four-stranded bridging sheet (blue) and a protruding V3 region (orange), both of which interact with CCR5 (Rizzuto et al. 1998), although only the amino-terminal region of the bound CCR5 structure has been determined (Huang et al. 2007). The gp41 conformations of HIV-1 in prefusion and intermediate stages is unknown, but the postfusion conformation (Chan et al. 1997; Weissenhorn et al. 1998) reveals a stable six-helix bundle, with structural similarity to other type 1 viral fusion machines.

Figure 2.

Sites of HIV-1 vulnerability to neutralizing antibody. (A) Electron tomogram of the HIV-1 viral spike (Liu et al. 2008), with docked atomic-level structure of gp120 (Pancera et al. 2010a) and with sites of vulnerability to antibody-mediated neutralization identified. (B) Initial site of CD4 attachment (Zhou et al. 2007; Chen et al. 2009). The site of vulnerability to antibody is shown as a cross-hatched yellow surface, with regions that induce conformational change and are binding sites for antibody F105 (blue) and antibody b13 (purple) or regions that extend outside of the site such as that recognized by antibody b12 (red). Neighboring N-linked glycan is shown in cyan. (C) MPER and membrane context. The MPER contains a number of highly conserved tryptophans, which are important for its entry function. (Left) Model of MPER in membrane from NMR/EPR measurements. The structure of MPER residues 662–683 is shown in the context of a DPC micelle (Sun et al. 2008), with residues required for recognition by neutralizing antibodies 2F5, z13e1, and 4E10 colored in red, green, and cyan, respectively. (Right) Model of MPER bound by broadly neutralizing antibody 2F5 as inferred from the crystal structure of the 2F5-epitope in complex with its gp41-MPER epitope (Ofek et al. 2004). The 2F5 antibody (partially shown with heavy chain in blue and light chain in gray) extracts its epitope from a helical conformation and induces an extended loop (Song et al. 2009). This has been modeled with the epitope as defined in the crystal structure (red) connected through a schematic dashed yellow line into the carboxy-terminal portion of the MPER (the structure is not known of the complete MPER when bound by the antibody nor the relative orientations of the amino- and carboxy-terminal portions of the MPER in this context). Virus neutralization by antibodies 2F5, z13e1, and 4E10, furthermore, also appears to require interactions with the surrounding membrane, likely mediated by extended CDR H3 loops, all of which contain hydrophobic motifs capable of interacting with membrane (Alam et al. 2009; Julien et al 2010; Ofek et al. 2010a,b; Scherer et al. 2010).

Figure 3.

Alternative forms of HIV Env serve as prototype immunogens for neutralizing antibody vaccines. Different forms of the HIV Env can be used to elicit neutralizing antibody responses. They range from the most complex form, the HIV trimer that most closely resembles the form found on the viral spike (upper row), monomeric forms, which include the gp120 core or resurfaced stabilized cores (second row), a region of the core that is composed primarily of the outer domain (OD) which includes the CD4-binding site (third row), or selected subdomains, such as the CD4-binding β-15 loop or MPER attached to a heterologous stable scaffold (bottom row). Modification of these prototypes by deletion of variable regions, removal, or addition of glycans, stabilization with disulfide bonds or addition of space filling mutations can serve to alter immunogenicity and elicit antibodies of the desired specificity.