|Home | About | Journals | Submit | Contact Us | Français|
Despite the frequent observation of masking of HIV-1 neutralization epitopes, its extent has not previously been systematically assessed either for multiple epitopes presented by individual viruses or for individual epitopes across multiple viral strains. Using a recently developed method to identify amino acid sequence motifs required for recognition by HIV-1-neutralizing monoclonal antibodies (mAbs), we visualized the patterns of masking of specific epitopes targeted by mAbs in a diverse panel of HIV-1 isolates. We also calculated a specific masking intensity score for each virus based on the observed neutralization activity of mAbs against the epitopes in the virus. Finally, we combined these data with estimates of the conservation of each mAb-targeted epitope in circulating HIV-1 strains to estimate the effective neutralization potential (EN) for each mAb. Focusing on the V3 loop of gp120 as a prototype neutralization domain, we found that the V3 loop epitope targeted by mAb 2219 is one of the least masked mAbs and it has the highest EN. Interestingly, although the V3 loop epitope targeted by mAb 3074 is present in over 87% of all viruses, it is 82.2% masked, so its EN is lower than that for mAb 2219. Notably, 50% of the viruses that mAb 3074 is able to neutralize are classified as subtype C viruses, while 70% or more of the viruses neutralized by mAbs 2219, 2557 or 447-52D are classified as subtype B. Thus, neutralization epitopes (in this case, in the V3 loop) have differential patterns of masking and also display distinct patterns of distribution among circulating HIV-1 viruses. Both factors combine to contribute to the practical vaccine value of any single epitope/mAb. Here we have developed a quantitative score for this value. These results have important implications for rational design of vaccines designed to induce neutralizing Abs by revealing epitopes that are minimally masked and maximally reactive with neutralizing Abs.
The envelope glycoprotein of HIV-1 (gp120) is one of the primary targets for antibodies that neutralize HIV and therefore contains many neutralization epitopes that vaccines need to target [1, 2]. Passive transfer of gp120-specific neutralizing antibodies confers protection from HIV and SHIV challenge in several animal models [3–5]. So antibody-mediated neutralization of HIV-1 strains could potentially protect against or limit infection. However, the tremendous antigenic variation of HIV is an obstacle to vaccine development and presents at least two immunologic and disease-relevant conundrums. Firstly, the bewildering variety of different HIV-1 isolates present antigenically (conformationally) distinct B-cell epitopes (although some cross-reactive epitopes do exist among viral isolates). Secondly, the presence of a specific set of amino acid residues comprising a neutralization epitope in a particular HIV-1 strain is a necessary, but not a sufficient, condition for neutralization by an antibody specific for that epitope. Those epitopes, when present, may be obscured or protected by other parts of the virus to varying degrees, a poorly understood phenomenon known as masking [6, 7]. Alterations in either epitope conformation or accessibility can be the basis for viral escape from neutralizing antibodies, and epitope accessibility can theoretically be divided into masking due to conformational protein changes or to glycan occlusion [8, 9]. Thus, the study of the phenomenon of masking includes evaluation of the multiple mechanisms by which Abs effectively, or ineffectively, neutralize HIV. For example, there is no way to ascribe an experimentally observed loss of Ab-mediated neutralizing activity to viral amino acid mutations in the epitope or to extra-epitope masking without a strict assessment of the presence and integrity of the epitope in the virus; both mechanisms result indistinguishably as loss of neutralization activity.
We recently developed a novel technology to identify individual anti-V3 loop mAb neutralization epitopes solely from the viral sequence . This capability permits a quantitative assessment of masking by revealing the neutralization variability between different HIV-1 viral isolates that are known to contain the same epitope. We utilized this capability to develop a masking signature for four different anti-HIV V3 loop mAbs, a V3 epitope masking score for each of the viruses, and a measure of the effective neutralization capability (EN) of each mAb against the worldwide population of circulating viruses.
The viral panel and neutralization data used in this study was published in Hioe et al. (2010) . Descriptions of the panel of 98 viruses used may be found in that study, but briefly, all subtypes were represented and all levels of neutralization susceptibility (i.e., the “tier” of the virus) were represented. Epitopes were mapped in the viral sequence by the method of Cardozo et al. (2009) , including new neutralization epitope sequence motifs from Swetnam et al. (2009) . Briefly, the “signature motif” for a V3 neutralization epitope consists of those V3 loop amino acid side chains in a crystal structure of the complex of a V3 loop peptide with the mAb that are significantly buried in a pocket on the molecular surface of the mAb. The previous study demonstrated that the signature motif derived in this manner is highly predictive of sensitive and specific virus neutralization by the mAb when controlling for exposure of the epitope in the virus.
Some of the viral sequences included in this study exhibited unidentified amino acids at certain positions, shown as X in the sequence in Figure 1. Also, the complete panel of viruses was not tested for neutralization with 447-52D therefore the IC50 for the untested viruses was classified in Figure 1 as N/A. Only four mAbs (2219, 3074, 2557, 447-52D) were evaluated, as they were the only ones with both neutralization data and known signature motifs for the epitopes they target.
The epitope targeted by mAbs 2219, 3074, 2557 or 447-52D (Table 1) was classified as “masked” if its signature motif was determined to be present in a virus, but the result of the in vitro neutralization assay showed no detectable neutralization by the mAb of that same virus at a maximum mAb concentration of 50 ug/ml. The percentage of masked viruses was calculated as the total number of viruses bearing a particular mAb-targeted epitope minus the number of viruses with the mAb epitope that were detectably neutralized by that mAb, this number was then divided by the total number of viruses with the epitope to obtain the final “masked” score shown in Table 2.
The effective neutralization score was calculated as:
where “global conservation percentage” is the percentage of circulating viruses containing the mAb-targeted epitope, as taken from Swetnam et al. (2009) ; and “masked percentage” is again the percentage of viruses containing a mAb-targeted epitope that were not detectably neutralized, as shown in Table 2.
The signature motifs for the epitopes targeted by the four mAbs evaluated in this study are shown in Table 1. Since it is a prototype neutralization domain, we studied only the V3 loop of HIV-1 gp120, which is a highly sequence-variable and frequently masked, but also highly immunogenic, region of the virus. Each of the four neutralizing anti-V3 loop mAbs previously isolated from HIV-infected human subjects targets epitopes near or overlapping each other in the crown of the V3 loop[12, 13]. Among the neutralization results for each mAb against a panel of 98 pseudoviruses from multiple subtypes derived from both acute and chronic HIV infections in standard in vitro neutralization assays [14, 15], none of the viruses, except two, lacking mAb epitopes were neutralized by the corresponding mAb (Figure 2, red bars), indicating that the signature motifs for the epitopes targeted by these mAbs are highly specific. The exceptions are sequences DJ263.8 and MW965.26. These latter pseudoviruses were neutralized by 447-52D (Figure 2D), but do not contain its sequence motif (P16+R18); instead they both contain a glutamine, Q in position 18 at the turn region on the V3 loop tip.
If the mAb-targeted epitopes in these viruses were completely unmasked, every green bar in Figure 2 would be near the X-axis (as with pseudovirus Bal.26 in Figure 2A). If the mAb-targeted epitopes in these viruses were completely masked, each green bar would be maximum height. Instead, each mAb neutralized a different spectrum of viruses bearing their targeted epitope, and the potency of the neutralization activity varied widely, with IC50 values ranging from <1.0 to 48.5 ug/ml. Thus, each neutralization epitope targeted by each specific mAb has a unique masking signature across diverse strains of viruses. This masking signature is depicted in “bar code” fashion in Figure 2. Of the four mAbs, 2219 neutralizes the largest number of pseudoviruses containing its epitope. The 2219 neutralization epitope is unmasked in 25/68 (36.8%) of the viruses containing the 2219 epitope. The epitope targeted by mAb 447-52D is the least masked, unmasked in 16/36 (44.4%) viruses containing the 447-52D epitope (Table 2). The epitopes targeted by the other two mAbs, 3074 and 2557 are highly masked (>70% unresponsive viruses containing the epitopes), but no epitope is completely masked across all the viruses in which it occurs (Table 2).
Three of the 98 pseudoviruses tested do not bear any of the four signature sequence motifs defined by the four mAbs tested (red boxes in Figure 1); the rest contain one or more. Among those containing the investigated epitopes, SF162.LS, Bx08.16, BaL.26, SS1196.1, 271-11, 25710-2.43, MW965.26, DJ263.8, CAP210.2.00.G3 and MGRM-D-034 were the only viruses that were sensitive to neutralization by every mAb whose epitope they contained. The remaining viruses bore at least one mAb-targeted epitope that was not neutralized by the mAb in vitro. Several viruses were insensitive to neutralization across the board, despite exhibiting all of the four mAb target neutralization epitopes: MGRM-Acute-B-018, MGRM-Acute-B-004, MGRM-Chronic-B-034, AC10.029, PVO.4, SC422661.8, TRJO4551.58 and H029.12. Four of the 16, and 7 of the 25 of pseudoviruses neutralized by mAbs 3074 and 2219, respectively contain multiple antibody epitopes that were not neutralized by other mAbs. Thus, a diversity of masking patterns is observed in this panel of viruses: some viruses mask all their epitopes, while others mask none, and several differentially mask only some of their epitopes, but are susceptible to neutralization through their unmasked epitopes. Interestingly, mAb 2219 targets an epitope that is maximally unmasked in subtype B viruses, while mAb 3074 targets an epitope that is maximally unmasked in subtype C viruses (Table 3).
Neutralizing mAbs are used in rational vaccine development to guide the design of immunogens that mimic their targeted epitopes. The most valuable mAbs for this purpose may be those that are estimated to neutralize, without modification, the maximum number of actual circulating HIV-1 viruses causing AIDS globally. This estimate (which we define as the mAb effective neutralization score, EN) has two components: 1) the frequency that the epitope occurs in circulating viruses, and 2) whether and how completely it is masked. Using the results of this study in conjunction with a previous study , we estimated the EN score for each mAb (Table 2). MAb 2219 has the highest EN. While mAb 3074 has nearly the same EN as mAb 2557, despite the fact that the former targets an epitope that occurs in nearly all circulating HIV-1 viruses while the latter targets an epitope present in only 52% of viruses.
Understanding the phenomenon of epitope masking is critical for elucidating viral escape mechanisms, since either protein conformational or glycan pattern changes in envelope structure can mediate viral escape without alteration of epitopes [6, 7, 9]. Understanding masking is also important for designing vaccines that will induce neutralizing antibodies in vivo that recapitulate the specificity and activity of known broadly neutralizing HIV antibodies. This “reverse vaccinology” approach utilizes the 3D structure of epitopes bound to these antibodies for immunogen design, but the efficacy of antibodies elicited by those immunogens is strongly influenced by the accessibility of the epitopes on diverse viruses. Furthermore, the variability of masking among different HIV strains may be related to the inherent resistance or sensitivity of an epitope to mutational escape by the mechanism of extra-epitope conformational masking: if the epitope is relatively unmasked across diverse viruses, there may be a significant cost to viral fitness to mask it. Hence, the epitope may be particularly valuable from a vaccine design point of view as it may be resistant to viral escape mutations that mask it.
In this study, we described a method to quantify both the level of masking of the array of epitopes in single viruses and the pattern of masking across viral isolates for individual epitopes within a naturally immunogenic region of gp120, the V3 loop. The results show that masking does indeed strongly influence antibody-mediated neutralization to the point that its contribution is as high or greater than the structural conservation of the epitope across circulating strains (Table 2). The results also show that the epitopes targeted by mAbs 447-52D and 2219 are the least masked of the V3 epitopes, suggesting that there is a viral fitness cost to masking/escaping these epitopes, which are conserved in the sequences of 11% and 56% of circulating HIV-1 viruses, respectively.
The V3 loop was previously observed to be highly masked, and the neutralizing activities of anti-V3 antibodies against HIV-1 isolates are affected both by sequence variation within V3 and by epitope masking by the V1/V2 domain  and carbohydrate moieties . We found that the masking pattern of V3 epitopes across diverse viruses is highly variable, but surprisingly none of the V3 loop epitopes were completely masked across all the viruses tested. Different V3 loop epitopes, such as those targeted by mAbs 2219 and 3074, have different masking patterns despite the fact that they occur in the exact same location in gp120. The 2219-targeted epitope is much less masked than the 3074-targeted epitope in subtype B strains while the 3074-targeted epitope is unmasked more frequently in subtype C strains. This differential pattern also suggests that different extra-epitope masking mechanisms may be operative in these two subtypes.
Based on this study, a set of viruses that completely masks V3 loop epitopes and a mutually exclusive set of viruses that expose all the studied V3 loop epitopes have been identified. Mapping glycan site changes in these viruses is relatively straightforward, so these two sets may be useful for identifying the positions of carbohydrates on gp120 that are responsible for at least one type of extra-epitope V3 loop epitope masking using a compare-and-contrast statistical approach such as that used to predict co-receptor usage by HIV-1 viruses . Non-glycan conformational extra-epitope masking may also be studied by subtraction of the glycan-dependent set using the two sets of viruses. For this purpose, additional data from larger samples of viral isolates will be needed for each of the two sets, and these can be added simply by testing for neutralization of more pseudoviruses with known sequences using the same (and perhaps additional) mAbs.
Identification of the signature motifs for neutralization epitopes was the key to conducting this study [10, 12]. A crucial test of the significance and specificity of these signature motifs was whether we would observe many viruses that were neutralized by a mAb despite a demonstrated absence of the signature motif for the epitope targeted by the mAb. In fact this was not the case, and only two of the total 98 viruses tested were neutralized by any mAb (both by the 447-52D mAb) despite the lack of the corresponding mAb-targeted neutralization epitope. The motifs are derived from crystal structures. A newer crystal structure demonstrates that 447-52D is an unusual mAb that can adopt a different mode of binding for viruses lacking the epitope signature motif . There is thus a secondary 447-52D-targeted motif that has not yet been clarified and was not included in this study. This motif is likely responsible for the two outliers. In practice then, our results support the view that the motifs may be 100% specific, which makes them useful in a variety of antibody-based investigations on populations of HIV-1 viruses.
This is the first report of EN for individual mAbs. This score, although prototyped here only for the V3 loop, can theoretically be applied to any HIV-1 epitope, or indeed to any neutralization epitope in any pathogen. The score may be very useful in efforts to reverse-engineer immunogens that are designed to mimic HIV-1-neutralization epitopes and elicit protective antibodies targeting those epitopes in mammalian serum. Previous results suggested that mAb 3074 may be the most useful mAb for such purposes due to the almost universal occurrence of its epitope in circulating HIV strains , but this new score reveals that mAb 2219 may be equally or perhaps more valuable due to the maximally unmasked nature of its epitope. The results suggest that although viral escape mutations that disturb the integrity of the 3074-targeted epitope may be uncommon, the activity of neutralizing antibodies targeting the same epitope may be more susceptible to viral escape mutations that introduce conformational or glycan-mediated masking of the 3074 epitope in some subtypes. On the other hand, the 2219-targeted epitope is less susceptible to viral escape mutations that introduce masking in subtype B.
An important vaccine design strategy might be the use of polyvalent “cocktail” vaccines in which epitopes that complement each other are combined to induce the broadest response. Interestingly, the EN range for mAbs 2219 and 3074 minimally overlap, with only six viruses neutralized by both mAbs. A combined antibody response to both these epitopes may thus be more effective than either one alone, potentially neutralizing 36% of globally circulating viruses in combination. Notably, both 2219-like and 3074-like neutralizing antibodies have previously been elicited in rabbits using rationally designed immunogens (S. Zolla-Pazner, personal communication), so such combinations are vaccine-ready. The EN thus guides the choice of rational immunogen combinations for use in polyvalent vaccines.
The authors would like to thank Miroslaw K. Gorny and Phillipe Nyambi helpful discussions in forming and editing the manuscript. The work was supported by grants from the Bill and Melinda Gates Foundation (#38631 to SZP), the NIH, including OD004631 (TC), AI084119 (TC), AI36085 (SZP), and AI27742 (NYU Center for AIDS Research), the NSF, including 0333389 training grant supporting (AA), and research funds from the Department of Veterans Affairs.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.