The experimental results described here indicate that a significant fraction of the anti-V3 antibody-mediated neutralization resistance of the conC sequence maps directly to the antibody-binding domain of the V3 crown. Furthermore, the epitope-independent structural feature by which the subtype C V3 crown resists neutralization by a variety of anti-V3 antibodies appears to be a rigid N-terminal non-β-strand conformation at positions 12 to 14 of the V3 loop. This effect is exclusive of the more commonly observed mechanism of antibody escape, that is, mutations of key neutralization epitope side chains, such as R18Q for 447-52D which we have shown results in a distinguishable, antibody-specific resistance. The combination of the loss of key neutralization epitope amino acid side chains with rigidity or non-β-strand structure results in total resistance of the psV bearing these V3 loop properties to neutralization by the antibody in question. Since these are intrinsic features of the V3 loop sequences, this phenomenon would apply to circulating viruses bearing these properties in their V3 loops as well, specifically subtype C viruses. Our dissection of the effects of single V3 loop point mutations shows that the effects of each point mutation is complex, and the backbone effect is combinatorial to all of the V3 loop positions simultaneously. Thus, no single amino acid position is solely responsible for the conC structural phenomenon. The convergence of three completely independent sets of data—(1) known crystallographic structures of V3 peptides bound to antibody, (2) patterns of psV neutralization, and (3) validated ab initio folding simulations—strongly supports these conclusions.
Our observation suggests that a flexible, β-strand structure at positions 12 to 14 is required for anti-V3 antibody-mediated neutralization, and indeed this region is bound by many anti-V3 antibodies. Nevertheless, it cannot be concluded that antibody binding alone underlies this structure-activity relationship. Neutralization is a multistep process with antibody binding being only one step. One can imagine a rigid V3 loop crown shape that is perfectly complementary to an antibody-combining site and therefore binds the antibody in vitro, but the virus may nevertheless be neutralization resistant due to the effects of this selfsame rigidity at other steps in the neutralization process. For example, neutralization-relevant V3 loop interactions with several other surfaces of gp120 may be affected by the rigidity in the V3 loop crown. For this reason, it is possible that structural rigidity in the V3 loop crown may also influence neutralization by non-V3-targeted antibodies by inhibiting intermediate conformations involving the V3 loop in the series of conformational changes that likely comprise the overall neutralization process. Indeed, the conC psV exhibits mildly increased resistance to the non-V3 Ab b12 ().
The unique resistance of conC to a wide variety of subtype A and subtype B derived anti-V3 antibodies may be informed by the observation of low variability of the consensus C sequence in circulating subtype C strains [26
]. If neutralization via the V3 loop is a strong selection pressure on circulating subtype C viruses, then infective V3 loop sequences harboring resistance to anti-V3-loop-antibody mediated neutralization might be observed at a higher rate and exhibit fewer escape mutations (vary less in sequence). As a corollary, subtype A and B derived anti-V3 antibodies may not be very effective as vaccine tools to combat subtype C, an observation that has previously been suggested for subtype B [27
]. Different strategies for interrogation of subtype C infected HIV+ sera may be required to uncover novel, effective neutralizing antibody responses to this subtype. On the other hand, a diversity of V3 loop sequences are present in subtype C in addition to the dominant conC and conC-like sequences. It is unlikely that the effect we have observed is universal in subtype C, and some subtype C strains may exhibit V3 loop flexibility or encode compensatory changes in other parts of gp120 to afford efficient neutralization by anti-V3 antibodies.
The psV neutralization data correlates strongly with the results of the folding simulations. This correlation extends previous observations suggesting that the 12 β
-hairpin residues (positions 10 to 22) of the V3 crown—including the currently known anti-V3 antibody combining sites—are sufficiently flexible in situ in the V3 loop to behave essentially as free peptides or as an autonomously folded subdomain [18
]. Ab initio
folding simulations of the V3 loop crown may therefore visualize at low resolution the dynamic structural ensemble of some V3 loop crowns in silico
, a potentially important high-throughput capability for mapping structure-(neutralization) activity relationships in the V3 loop crown. It should be noted that the identification of two properties that are more easily assessed in a dynamic ensemble of many conformations—secondary structure and rigidity—facilitated the comparative interpretation of our library of folding data. Observing the structural tendency at a very specific location—positions 12 to 14 in the V3 loop—also facilitated the study. More subtle and global structural patterns, including overall fold assignment and absolute energetic stability for the whole domain, may be more difficult to discern and assess across different foldings for the purpose of correlations with experiments.
The strong correlation of the psV neutralization measurements with observed structural features in the folding simulations also establishes the SF162 chimeric psV system as one that provides, even to fine resolution, a consistent virologic background across multiple experiments and V3 loop sequence variations. The combination of chimeric psV neutralization measurements with ab initio folding simulations allows detailed quantitative dynamic structure-neutralization activity relationships to be mapped out for the V3 loop. Such studies would be difficult with crystallography, which is low throughput and does not evaluate dynamic structure.
Antibody epitopes in the V3 loop may occur broadly in HIV-1 viruses, but antibodies appear to be limited in accessing these epitopes presumably due to the effects of “masking” glycans and nearby variable domains [17
]. In this work, the comparison of neutralization tests with folding simulations may have revealed an obscure structural explanation for epitope-independent variations in antibody-mediated neutralization. The fact that some of the resistance to antibody-mediated neutralization maps directly to the antibody binding area of the V3 loop influences the view of masking: some of the observed masking of V3 loop epitopes may be intrinsic to the V3 loop itself and not due to outside factors such as glycans and the other variable loops of gp120, although both factors are likely operative to different degrees in any given strain. The approach described here could potentially be modified to “localize” the masking of the V3 loop epitope in primary HIV-1 isolates, at least to V3 loop or “outside-V3 loop” locations, by identifying anti-V3 loop antibody-resistant primary isolate sequences in which this “local masking” is present.
HIV-1 strains that evolve to partially mask receptor interacting surfaces in order to hide those vulnerable surfaces from the immune system trade a loss of infective efficiency for a gain in camouflage protection. The coincidence of determinants for infection (chemokine receptor binding surfaces) and immune detection (broadly neutralizing epitopes) in the V3 loop likely requires such a tradeoff for best viral fitness. Since a wide variety of anti-V3 antibodies appear to adhere to the antibody resistance mechanism described here, the work suggests that this tradeoff is accomplished in subtype C partly by the adoption of V3 loop structural shapes that are inefficient for both antibody-mediated neutralization and coreceptor binding, that is, rigid non-β
-strand conformations. As the same viral mechanisms that produce this feature of the V3 loop beget the structural features of the other four variable loops, it is likely that the same tradeoff is exploited by viral evolution for functional regions of the V1/V2, V4, and V5-loops. If anti-variable loop antibodies play a significant enough role in the human protective response to circulating HIV-1 strains, the phenomenon we have described here may explain the concurrent diverging observations of decreased fitness (poorer receptor usage) and increased natural spread (successful immune evasion) in subtype C. Indeed, the recent identification of anti-V2 loop Abs as the only known inverse correlate of risk for HIV infection suggests that antibodies to variable loops indeed play a significant role in the human protective response from circulating HIV-1 strains [28