Guided by a new crystal structure that includes the complete gp41-interactive region of the CD4-bound HIV-1 gp120 (Pancera et al., 2010
), we identified a network of interactions involving the gp120 inner domain that modulate gp120-gp41 association, CD4 binding affinity, and susceptibility to inactivation by sCD4.
Prior to receptor engagement, gp120 must maintain its non-covalent association with gp41 and prevent gp41 from prematurely undergoing transitions to lower-energy conformations. In both unliganded and CD4-bound states, two HIV-1 gp120 elements, a 7-stranded β-sandwich and the N- and C-terminal extensions from this sandwich, play a major role in mediating gp120-gp41 association (Helseth et al., 1991
; Binley et al., 2000
; Yang et al., 2003
; Sen et al., 2008
; Wang et al., 2008
). Two other gp120 regions apparently contribute to the interaction with gp41 in the unliganded Env gp trimer. The first region involves Layer 1, which is predicted to reside near the trimeric axis of the Env gp complex (Kwong et al., 2000
; Pancera et al., 2010
) and thus would be well-positioned to interact with gp41. Indeed, changes in two Layer 1 residues (H72 and A73) did decrease subunit association. Unexpectedly, however, most changes in Layer 1 residues (including those interacting with Layer 2) exerted modest or no effects on gp120-gp41 association. Thus, Layer 1 may play a modulatory rather than a primary role in the gp120-gp41 interaction. Such a role has been suggested for valine 65 in the Layer 1 α0 helix, based on the co-selection of gp120 and gp41 changes during virus adaptation to T-cell lines (Leavitt et al., 2003
A second region important for gp120-gp41 association is Layer 2, particularly the distal end of the α1 helix. Most of these Layer 2 mutants mediate cell-cell fusion efficiently and are recognized by conformation-dependent antibodies, and thus are not globally misfolded. Many of the Layer 2 residues implicated in gp120-gp41 association are not surface-exposed in the CD4-bound conformation of gp120. Thus, any direct interaction between these Layer 2 residues and gp41 would require the existence of a different gp120 conformation in the unliganded state.
Other studies have suggested the possibility that the distal elements of Layer 2 and the adjacent β2-β3 strands, which compose part of the bridging sheet, undergo conformational changes upon CD4 binding (Kwong et al., 1998
; Chen et al., 2005
; Pan et al., 2005
; Rits-Voloch et al., 2006
). Indeed, in some structures of gp120 without CD4 (Chen et al., 2005
; Zhou et al., 2007
; Chen et al., unpublished data), the α1 helix is partially or completely unwound, potentially allowing direct contacts between some of the residues implicated in our study and gp41. In applying these observations to the unliganded Env gp complex, however, two caveats should be kept in mind. First, in all prior studies, gp120 inner domain components were deleted or modified to promote crystallization. Second, in light of the possibility that the inner domain interacts with gp41, structures of this region that are relevant to the unliganded Env gp trimer may require the inclusion of gp41 components.
Changes in the Layer 1-Layer 2 interface apparently decrease the spontaneous sampling of the CD4-bound conformation by HIV-1 gp120, lowering the on-rate, binding affinity and neutralization potency of CD4i antibodies (, Figure S1
and Kassa et al., 2009
). CCR5 binding is also sensitive to these Layer 1-Layer 2 changes. By contrast, the initial contact of CD4 with gp120, reflected in its on-rate, is minimally affected by inner domain changes, and thus appears not to depend upon the sampling of the CD4-bound conformation by gp120 (Zhou et al., 2007
; Rits-Voloch et al., 2006
). Rather, changes in the integrity of the Layer 1-Layer 2 interface influence the off-rate of the CD4-gp120 complex; this effect cannot be explained by direct contacts of Layer 1 or Layer 2 residues with CD4 (Pancera et al., 2010
). The S375W change, which fills the Phe 43 cavity, completely compensates for the inability of inner domain mutants to decrease the off-rate of bound CD4; however, in the absence of CD4, the S375W change is not sufficient to allow spontaneous sampling of the conformation required for efficient recognition by CCR5 and CD4i antibodies. Thus, Layers 1 and 2 in the inner domain contribute to the ability of gp120 to make the transition from the unliganded to the CD4-bound state; subtle differences exist in the requirements for binding CD4 on the one hand and CD4i antibodies/CCR5 on the other.
Alteration of gp120 residues in the Layer 1-Layer 2 interface (H66A, W69L, L111A and P212A) resulted in increased resistance to neutralization by sCD4. Differences in the affinity of monomeric gp120 for CD4 likely contribute to the sCD4 resistance of these mutants (Figure S3
), as has been observed for other HIV-1 Env gp variants with decreases in CD4 binding (Thali et al., 1991
). Consistent with this, the S375W change restored the CD4-binding affinity of the H66A, W69L and L111A mutants, and reverted the phenotypes of resistance to both sCD4 neutralization and sCD4 induction of gp120 shedding. Of interest, changes in the α0 helix of Layer 1 have been reported to segregate with sCD4 resistance during the passage of a primary HIV-1 isolate in a T-cell line (Orloff et al., 1995
Soluble CD4 induces an activated state in the HIV-1 Env gps that rapidly decays into functionally inactive forms (Haim et al., 2009
). Alterations in Layer 1 (e.g., H66A and W69L) apparently affect this decay process, even when sCD4 binding to the Env gps is restored by the S375W change in the Phe 43 cavity. During infection of CD4-negative, CCR5-positive target cells, viruses with the H66A/S375W and W69L/S375W mutant Env gps were activated at low sCD4 concentrations. During infection of CD4-expressing target cells, these mutant viruses were neutralized as efficiently by sCD4 as viruses with wild-type Env gps. Soluble CD4-induced shedding of gp120 was comparable for the wild-type, H66A/S375W and W69L/S375W Env gps. Therefore, the H66A/S375W and W69L/S375W Env gps apparently bind sCD4 efficiently; nonetheless, these mutants demonstrate activation at high sCD4 concentrations that, in parallel, do not activate wild-type HIV-1. In this respect, these HIV-1 mutants resemble SIV (Allan et al., 1992; Schenten et al., 1999
). Thus, the H66A and W69L changes stabilize the sCD4-activated HIV-1 Env gp intermediate, slowing its decay into functionally inactive forms. The observed correlation between the longevity of the sCD4-activated state and the duration of sCD4-induced exposure of the gp41 HR1 region (Haim et al., 2009
) is consistent with a role of Layer 1 in modulating conformational transitions in gp41. That H66A/S375W and W69L/S375W Env gps differ from wild-type Env gps in sCD4-induced activation-decay but not in gp120 shedding underscores previous suggestions that the functionally inactive products of these two processes are distinct (Si et al., 2004
; Haim et al., 2009
The involvement of Layer 1 and Layer 2 residues in gp41 association and in CD4 binding suggests a model for the unliganded gp120 in the assembled Env gp trimer, and for the conformational switch that occurs upon CD4 binding (). According to this model, in the unliganded state, Layer 2 interacts with the gp41 ectodomain. By virtue of its own interaction with gp41, Layer 1 is shifted from its CD4-bound conformation, thus enhancing the opportunity for Layer 2 to contact gp41. The binding of CD4 to the wild-type HIV-1 Env gps results in the apposition of Layer 1 and Layer 2, as seen in the recent Pancera et al. crystal structure (). The Layer 1-Layer 2 interaction reduces the off-rate of the bound CD4. The Layer 1-Layer 2 interaction also allows the gp41 ectodomain segment to undergo additional conformational changes involved in virus entry; this process may involve both the liberation of gp41 ectodomain segments through the loss of previous interactions with Layer 1 and Layer 2 residues, and the creation of new bonds between gp120 and gp41 that favor the formation of the pre-hairpin intermediate. This model helps to explain the relative lability of the sCD4-bound state (Haim et al., 2009
; Kassa et al., 2009
). Altered gp41 interactions with Layers 1 and 2 resulting from sCD4 binding might destabilize the gp41 ectodomain and either prematurely promote conformational changes related to those involved in virus entry or predispose to off-pathway inactivating events. Future studies will test the specific predictions of this model.
A model of the activation of the HIV-1 Env gps triggered by binding to the CD4 receptor
The inner domain residues implicated in the CD4-induced conformational transition are well-conserved within the HIV-1 and SIVcpz lineage (Figure S4
). Of interest, the corresponding residues are also conserved within the HIV-2/SIVsm lineage, although differences between the HIV-1/SIVcpz and HIV-2/SIVsm lineages are evident. For example, residue 375 is naturally a tryptophan in the HIV-2/SIVsm viruses (Kuiken et al., 2008
); in HIV-1, substitution of tryptophan for serine 375 fills the Phe 43 cavity and diminishes the impact of alterations in Layers 1 and 2 on CD4 binding. Alanine substitutions in the distal part of Layer 2 (residues 206-214) in SIV gp120 have been previously reported to decrease gp41 association (Rits-Volloch et al., 2006
); however, two equivalent HIV-1 gp120 mutants (F210A and P212A) studied herein exhibited wild-type levels of subunit association. Thus, while both immunodeficiency virus lineages preserve the potential to form a network of inner domain interactions, lineage-specific differences exist, probably driven by distinct biological necessities.