In this work, we present a number of structures of trimeric HIV-1 Env and antibody complexes obtained by cryo-electron microscopy, including a 9 Å map of a novel, previously unknown activated intermediate that precedes formation of the pre-hairpin intermediate. The discovery () that binding of the co-receptor site on gp120 can induce the same quaternary conformational changes in trimeric Env as those that occur with CD4 binding is surprising. It is conceivable that the ability of 17b to induce a conformational change similar to that observed with CD4 is strain-specific, and that strains other than HIV-1 BaL may have a lesser propensity to undergo this 17b-induced change in the absence of added sCD4. In this context, we reported previously that, while soluble KNH1144 Env trimers were capable of displaying an open, 17b-bound conformation, similar trimers from JR-FL displayed the 17b-bound conformation only in the presence of bound sCD4 
The differences in the quaternary structures of trimeric Env bound to either VRC01 or b12 were also unexpected; these findings provide a molecular basis for understanding the differences in neutralization efficiency of these two antibodies. A CD4 mimic may be expected to induce similar structural changes as sCD4 itself. However, we show now that this is not the case, based on determination of the quaternary structures of native trimeric Env bound to VRC01, VRC02 and VRC03 antibodies. Thus, despite the remarkable similarities in the interactions of sCD4 and VRC01 with gp120 (), VRC01 (and also VRC02, VRC03) binding retains the unliganded (closed) trimer conformation, while sCD4 binding results in transition of Env to the open trimer conformation. Initial studies of VRC01 binding 
suggested that VRC01 binding enhances binding of 17b to gp120 monomers. Our results now establish that this does not hold true for trimeric gp120, highlighting the importance of analyzing antibody binding to gp120 trimers in the context of the native virus. Furthermore, the evidence that 17b binding in intact viruses is actually blocked by VRC01 binding suggests a more general structural mechanism for neutralization by antibodies such as VRC01. Our results show that this “CD4 mimic” not only blocks CD4 binding by binding a similar region on Env, but that it actually prevents the opening of the spike necessary for exposure of the pre-hairpin intermediate, and by extension, infection.
Thus, a central finding from our comparative structural analyses is that binding of ligands, including antibodies with very similar footprints on gp120, as judged by crystallographic and mutagenesis studies, can have profoundly different outcomes for the conformation of trimeric gp120. Strains that are less susceptible to the type of conformational rearrangements in gp120 and gp41 that are required to accommodate b12 binding may therefore be less likely to be neutralized by antibodies such as b12, offering a possible explanation for the lower neutralization breadth of b12. The reason for poorer neutralization breadth of VRC03 relative to VRC01 may lie in lower affinity 
for gp120, since there appear to be no significant differences either in the X-ray structures of their respective complexes with gp120 monomers 
(PDB ID: 3ES8) or in the complexes formed with native trimeric Env (). Whether the formation of the open quaternary state of Env is always sufficient to expose gp41 and lead to fusion between viral and target cell membranes remains an unanswered question. It also is unclear whether there are subtle variations in the quaternary structures, such as differential exposure of the gp41 fusion peptide in the open states populated by distinct ligand binding combinations.
Our description of the structure at ~9 Å resolution suggests a new structural template for designing immunogens that can elicit antibodies targeting HIV at a vulnerable, pre-entry stage. Previous models for the mechanism of HIV entry have postulated that activation of Env by CD4 and co-receptor binding should lead to formation of a transient “pre-hairpin” intermediate, in which the N-terminal helices of gp41 become exposed and accessible to binding by entry inhibitors. No structure is available for this intermediate, but many biochemical experiments have been used to deduce its likely structural properties. In our work, we show that activation of Env by CD4/co-receptor binding leads to the formation of an activated intermediate where three N-terminal helices are nestled at the center of the complex, surrounded by trimeric gp120 that has moved outwards. It is likely that the presence of the additional disulfide bond and possible the Ile to Pro mutation in the SOSIP gp140 constructs stabilizes the activated intermediate. We further show that the N-terminal helices in this activated intermediate are not in the same compact structure that is observed for the post-fusion state. The differences are significant: the transition from the open quaternary conformation to the post-fusion state requires a change in angle by ~15° of each of the three N-terminal helices. In addition, the helices become more compactly packed as a result of their interaction with each other and the C-terminal portion of gp41. Our discovery, therefore, suggests a new template for immunogens; antibodies elicited using this model could potentially have greater potency than those currently elicited using peptide mimics based on the compact post-fusion structure.
Conversion of the activated three-helix structure to the post-fusion six-helix bundle is a key event common to the entry mechanisms of many viruses, such as HIV, influenza, Ebola and Moloney murine leukemia virus 
. Determination of the spatial arrangement of the gp41 N-terminal helices thus provides a structural template for vaccine design that is based on an experimentally observed intermediate state. This intermediate state could prove to be more immunogenic than templates derived from the structure of the post-fusion six-helix bundle state which occur at a late stage in the entry process 
. The N-terminal helix of gp41 is one of the most conserved regions in HIV-1 Env. Because the open Env conformation occurs prior to membrane fusion and the N-terminal helices are more exposed than in the compact organization of the post-fusion state, immunogens based on this conformation are likely to elicit antibodies that could be effective in blocking entry across a broad spectrum of HIV-strains. Our observations are also consistent with the finding that, in HIV-1, dissociation of gp120 from gp41 exposes an epitope located in the loop between N- and C-terminal segments of gp41 
, which is recognized by the murine monoclonal antibody KK20.
Overall, our findings are consistent with a model in which native trimeric HIV-1 is in a closed, but metastable conformation. Binding of ligands, such as sCD4, and co-receptor mimics, such as 17b, result in the formation of a dramatically different, activated intermediate conformation. In this conformation, gp120 and gp41 undergo coordinated structural changes, resulting in exposure of the gp41 fusion peptide through an opening at the apex of the spike. The structural analyses presented here demonstrate that different neutralizing antibodies block viral entry by distinct structural mechanisms. Antibodies such as VRC01 appear to neutralize HIV by binding the CD4 site and blocking activation of Env, thereby preventing the opening of the spike necessary for exposure of the fusogenic components of gp41 and subsequent viral entry. Antibodies such as b12 also appear to hold trimeric Env in the closed state, but with subtle rearrangements in the quaternary packing of gp120 in the trimer (, ). Therefore, the neutralization ability of b12 may be limited to strains in which packing constraints in trimeric Env are loose enough to allow these gp120 rotations, potentially explaining its lower neutralization efficacy compared to VRC01. In contrast to VRC01 and b12, 17b binding blocks entry by capturing a conformation in which the fusion peptide is physically prevented from contact with the target cell membrane by the bound antibody. In a recent study, Scheid et al 
found that broadly neutralizing activity in serum from HIV-infected patients arose not from one highly potent neutralizing antibody, but from the aggregate activity of several antibody specificities. It is clear, therefore, that multiple antibodies can bind the same Env trimer simultaneously, and that the same antibody can be accommodated on the gp120 surface in very different quaternary states of trimeric Env. Effective strategies for HIV neutralization may thus require elicitation of a variety of antibodies, including those like VRC01 that lock gp120 and gp41 in the unliganded conformation, antibodies like 17b that block access of the fusion peptide to the target cell by binding at the apex of Env, and antibodies such as 8066 
, D5 
or HK20 
that bind Env following its activation, but before fusion between viral and target cell membranes.