Due to the pressing need for an effective HIV-1 vaccine, and due to the limits of current Env-based immunogens to elicit neutralization breadth, we pursued HIV-1 Env structure-guided immunogen design to determine if this line of investigation will better elicit virus neutralizing antibodies. Here, we demonstrate that the structure-based redesigning of the HIV-1 envelope core glycoprotein increased folding and expression of a series of related, mutagenically stabilized molecules. Structure-guided protein design led to stabilization of both the CD4-binding site and the co-receptor-binding region of gp120. The data clearly demonstrated that thermodynamic stabilization of the co-receptor-binding site was associated with a marked increase in the on-rate of binding of the co-receptor mimetic, 17b antibody. Furthermore, when immunized into small animals, stabilization of the gp120 core resulted in a dramatic enhancement of the functional antibody response against the CD4-enhanced co-receptor binding region shared by HIV-1 and HIV-2. These results suggest that, in general, specific regions of an immunogen might be rendered more immunodominant by direct conformational stabilization (in this case, cysteine-pair mutagenesis) of that region resulting in reduced entropy. These results are appealing from a thermodynamic perspective, as the ligand affinity (ΔG
) can be more favorable with a reduction in entropy (−TΔS
). This thermodynamic relationship would predict that if a given epitope (or circumscribed region) is pre-fixed into a desired conformation, a ligand (i.e.,17b or perhaps a “naïve” B cell receptor) specific for that site will not be required to initiate “induced-fit” 
and will therefore bind to the site with an increased on-rate 
. Potentially, the overall affinity may increase (assuming no negative impact on the off-rate), and in regards to the B cell receptor, a faster on-rate may enhance epitope recognition, resulting in more efficient activation of that B cell. The finding described in this study, demonstrate that conformational stabilization of a discrete protein region can alter the quantity and quality of the antibody response to that protein region. They suggest that ligand stabilization or improved ligand affinity, especially ligand on-rate, might be used as parameters to focus functional antibody responses on specific regions on the surface of a complex, conformationally sensitive and multi-epitope protein.
Recently we reported that the elicitation of co-receptor-directed antibodies is dependent upon interaction of gp140 glycoprotein immunogens with endogenous primate CD4 molecules 
. A mechanism for this important observation is provided in this current study as we clearly demonstrate that mutagenic stabilization of g120, in a similar manner to that achieved by CD4 binding, locks the co-receptor-binding-site into a single conformation that is well recognized by the naïve B cell repertoire in rabbits. This principle might be applicable for viruses that undergo receptor-induced conformational changes to accomplish entry, and for which a vaccine is lacking (e.g., Ebola). Deletion of immunodominant variable regions of Env-based anti-viral subunit vaccines may also have broader applicability.
The protein redesign described here to improve expression revealed some interesting observations relative to recognition by the 17b antibody and implications on the bridging sheet in the coreV3S context. Our earlier data indicated that the previously crystallized core protein could not bind 17b unless induced by CD4 
. In this study, the previously described core protein was modified at the base of the V3 loop and at the base of the V1V2 loop. Somewhat unexpectedly, we observed that the newly designed coreV3S protein was recognized by 17b with very high affinity even in the absence of CD4, and that CD4 binding affinity of this protein was markedly improved. A plausible explanation of this modified antigenicity is that restoration of the β12 and β13 strands on the outer domain indirectly aids in formation of bridging sheet elements that are critical for 17b recognition. Restoration of these strands may also impart stability at the base of the Phe 43 cavity, located above (CD4 binding site, see ). These implications are somewhat in conflict with the unliganded SIV core structure, which shows the bridging sheet β-strands in a non-CD4-bound orientation, but perhaps represent differences in the structural elements present between core and coreV3S proteins and/or differences between HIV and SIV 
. However, these data are consistent with the initial analysis of 17b recognition, which revealed that 17b binds well to full-length gp120 possessing the V3 loop, but not at all to a V3-loop deleted protein 
, confirmed by our recent studies 
. Complete V3 loop deletion was performed in the original HIV-1 and SIV crystallized cores proteins. However, because interaction of CD4 with V3 loop-deleted gp120 completely restores 17b binding, this suggests that CD4 can compensate for the (artificial) instability of the bridging sheet region imparted by full truncation of the V3 stem. Yet unresolved, then, is the structure of the receptor-binding sites in the context of the static functional spike (i.e., pre-receptor bound state). The data described above are consistent with the model that the co-receptor-binding site can exist in the context of the static viral spike, but accessibility to antibody is limited unless steric constraints are reduced by receptor engagement.
In previous studies, we have shown that the Phe 43 cavity-filling mutations partially lock gp120 into the receptor/co-receptor-bound conformation 
. To achieve greater stabilization, we added two pairs of cysteines and additional cavity-filling mutations to core gp120 
and analyzed the thermodynamic effects of these mutations on core gp120. Interestingly, we showed clearly that 17b itself can stabilize gp120 into the conformation recognized by CD4 (B1
reactions) and that the Ds mutations used in this study have no further effect on this conformation (B2
reactions). However, it is noteworthy that, a relatively constant amount of entropy change (15.1 to 17.3 kcal/mol) was always detected upon addition of CD4 to gp120-17b complexes irrespective of the presence of the stabilizing mutations. We assume that this entropy is either accounted for stabilization of elements distal from the cysteine pairs themselves or, alternatively, results from some unanticipated solvent displacement effects.
Increases in 17b affinity or increased stabilization of the 17b epitope alone were not always associated with all differences in the immunogenicity described here. For example, between 2CC and 3CC, increased 17b affinity correlated very significantly with increased elicitation of CD4i antibodies, although the degree of stabilization achieved was similar in these immunogens. In contrast, the increase in 17b affinity was minimal from 3CC to 4CC, although there was a substantial difference in epitope stabilization. This difference correlated with the enhanced elicitation of CD4i antibodies by 4CC. Therefore, ligand affinity and epitope stabilization both contributed to the overall altered immune responses elicited by the current set of immunogens.
The same set of mutations that stabilized the co-receptor-binding site also stabilized the CD4bs, generally considered a more desired target of the study design because of the neutralization capacity of both CD4 itself and of the CD4bs antibody, b12. In fact, we clearly demonstrate the elicitation of CD4bs antibodies both by ligand cross-competition and by selective adsorption. However, the unmodified cores appear to elicit this type of activity more efficiently than the stabilized cores (see , Figure S6
and Figure S7
). It might be that the stabilization process itself subtly altered the CD4-binding surface on gp120 and actually reduced cross-reactivity with natural sequences found on the virus. Alternatively, the stabilized CD4i site perhaps became immuno-dominant and out-competed CD4bs-directed responses. Furthermore, although stabilization of the CD4bs, relative to its starting entropy, approached that of the co-receptor-binding site, the absolute values of CD4-related entropy and affinity did not. Analysis of the sera from one representative animal immunized with coreV3S group compared to one representative animal immunized with the 4CC protein demonstrated a shift of antibody response towards the 17b epitope, correlating with the increased 17b affinity. Consistent with these results, immunization studies using synthetic peptide immunogens indicate that the kinetics of antigen recognition influence epitope-driven repertoire selection and antibody maturation 
. Achieving slower ligand off-rate may have the potential to improve immune response, although that property may not always be approachable by structure-based design and might be dependent upon the context. This might be in part due to the uncertainty of the factors that define which elements on the surface of a complex protein are most immunogenic. It was suggested that all accessible domains on the surface of a multi-determinant antigen can potentially induce primary B cell responses 
. However, only those that interact with naive B cells with high affinity will generate an avid antibody response 
Here, we demonstrate in this complex model system that it is conformational fixation, associated with increased 17b on-rate and overall affinity, which drives the elicitation of functionally cross-neutralizing antibodies directed toward the gp120 co-receptor binding region. This class of antibodies was extensively studied for their unique properties of posttranslational tyrosine sulfation and preferential VH gene usage 
. Although to date there are no identified co-receptor binding-site-directed monoclonal antibodies that potently neutralize diverse primary isolates, our recent study implicates antibodies with specificities to this site contribute to neutralization in broadly reactive HIV-1 patient sera 
. CD4-induced antibodies were also associated with partial control of SHIV challenge in macaques 
. In the current study, the stabilized immunogens elicited moderate neutralization responses against a few Tier 1 HIV-1 isolates that are typically sensitive to antibodies directed to the gp120 variable loop 3, a component absent in the immunogens tested here. Additional analysis may be warranted to determine if the neutralization activity observed in selected sera is indeed mediated by CD4-induced antibodies as was determined here for the Tier 1 isolates, MW965 and SS1196. Thus, the co-receptor-binding region, usually occluded on most primary isolates, remains an intriguing target due to its conservation, especially if there exists an as yet-to-be-defined subset of antibodies that can access elements of this region on circulating isolates. In addition to CD4-induced responses, the principle established in this study may have important implications for proper stabilization of the CD4bs to generate more broadly cross-reactive and neutralizing antibodies to this heavily shielded, receptor-binding region towards the development of a broadly protective HIV-1 vaccine. Beyond HIV-specific vaccine development, the viral envelope glycoprotein and its ligands under study here provide a model system to establish “proof-of-principle” regarding targeted immunogenicity. Such principles may extend to the design of vaccines against other pathogens capable of humoral immune evasion.