An ideal HIV/AIDS vaccine should elicit high levels of protective antibody and cytotoxic T lymphocyte to neutralize viral infection and to eliminate virus infected cells, respectively. Because of the extensive genetic diversity of HIV and its devious immune evasion strategies, such a vaccine has yet to be discovered. To change this situation, in-depth analysis of vaccine-elicited immune responses is particularly important. To this end, the first evidence of efficacy for AIDS vaccine candidate in the RV144 trial has demonstrated a modest level of 31.2% against HIV risk,1 which is related to the binding IgG antibodies targeting V1V2 region of gp120, rather than to bNAbs.2,3 Therefore, it is of great interest to investigate the antigen–antibody interaction to reveal the major antigenic determinants (MAD) of vaccine-elicited antibody responses in protection.
Significant progresses have been made in understanding the role of antibody responses against simian immunodeficiency virus (SIV) infection in rhesus macaques. Similar to HIV infection, the protective role of antibodies has been demonstrated by passive immunization experiments using plasma4–6 or purified immunoglobulin derived from SIV-infected long-term nonprogressor macaques.7 Recently, a mimic study of RV144 in rhesus macaques indicated that the level of V1V2-binding antibodies was inversely correlated with SIV risk.8 It, therefore, becomes necessary to investigate the role of V1V2-directed antibodies in other vaccine-elicited protection studies.
We recently reported that mucosal priming with a replicating modified vaccinia Tiantan virus (MVTTgpe) together with a nonreplicating Ad5gpe boost regimen (referred as the MVTTgpe-based regimen here) elicits durable protection against pathogenic SIVmac239 infection in rhesus monkeys.9 Although robust, cellular immune responses and higher titers of neutralizing antibodies against SIVmac1A11 were discovered, the role of vaccine-elicited antibody responses in protection remains poorly understood. In this study, we develop a yeast surface display (YSD)-based antigen library of SIVmac239 full envelope glycoprotein for both quantitative and qualitative measurement of vaccine-elicited antibody responses. Using this newly established technique, we sought to define the MAD of predominant SIV antigenic domains and their relevance to vaccine-induced immune protection in Rhesus macaques.
Plasmid, Yeast Strain, and Monoclonal Antibodies
The plasmid pCTCON2-T for yeast surface display and yeast clone Saccharomyces cerevisiae EBY100 have been described previously.10,11 Three SIV-specific monoclonal antibodies (mAbs) KK8, KK65, and VM18S were obtained from NIH AIDS Research and Reference Reagent Program.
Animals and Study Design
We tested 12 colony-bred Chinese rhesus macaques (Macaca mulatta), which were selected, housed, and maintained in accordance with the standards of the Institutional Animal Care and Use Committee (IACUC). The experimental design of our monkey study has been described previously.9 Briefly, as shown in Figure 2A, 12 monkeys were divided into 3 groups: (1) 4 monkeys received MVTTgpe mucosal prime and Ad5gpe intramuscular boost immunization (MVTTgpe-based vaccine regimen), (2) 4 monkeys received a homologous, intramuscular prime and boost with the Ad5gpe vaccine (Ad5gpe-based vaccine regimen), and (3) 4 monkeys received PBS as a sham control group. At week 24 after the final vaccination, each animal was challenged intrarectally with a high dose of TCID50 pathogenic SIVmac239. Immunized sera G1 and G2 were collected at week 6 after the final immunization from the 2 different vaccination groups. Recalled immune sera G3 and G4 were collected at week 7 after infection from the same vaccination groups. Early infection sera G5 and simian AIDS sera G6 were collected at week 7 and 56/66 after infection from the sham control group, respectively. All serum samples were stored at −80°C until use.
Construction of YSD-Based Antigen Library
The full-length SIVmac239-env gene (2640 bp) was amplified by polymerase chain reaction (PCR). The PCR products were then purified and randomly digested into small fragments (50–100 bp) with DNase I. The digested fragments were then reassembled to longer 100–900 bp fragments by PCR amplification, A tailed, and then ligated to the pCTCON2-T vector as described previously.11,12 The competent EBY100 cells were transformed with purified ligation products through electroporation, then transferred to a flask with 200 mL synthetic dextrose casamino acids (SDCAA) media, and incubated for about 24 hours. Serial dilutions of the cell suspension were placed on the SDCAA plates to estimate the library size (The size of SIV-Env yeast libraries was about 106). Conditions for yeast growing and induction of surface antigen expression in solution have been described previously.13
Fluorescence-Activated Cell Sorting and Sequence Analysis
Induced yeast cells were incubated with 50 μL of monkey sera or mouse anti-SIV mAbs (0.2 μg/test) at 4°C for 1 hour. After washing, yeasts were incubated with Phycoerythrin (PE) conjugated antimonkey IgG (clone SB108a, SouthernBiotech) or PE-conjugated anti-mouse IgG (BioLegend) at 4°C for another 1 hour. Finally, the stained yeasts were analyzed by fluorescence-activated cell sorting (FACS) using Aria III Flow Cytometer (BD Biosciences, San Jose, CA). After confirmed by FACS and fluorescence microscope (Nikon, Tokyo, Japan), the positive yeast clones were cultured for plasmid isolation using OmegaSpin Plasmid kit (Omega Bio-Tek Inc., Norcross, GA). The isolated plasmids were then transformed into Escherichia coli competent cells for plasmid extraction and DNA sequencing. The sequences were analyzed using Sequencher 4.9 (GeneCodes Corp., Ann Arbor, MI).
Algorithm for Antigenic Domain and MAD Mapping
The potential antigenic domains based on the selected YSD-Env fragments were analyzed using a designed computer algorithm for sequence scanning and clustering as described previously.11 Accordingly, we also calculated the area under the curve (AUC), which reflects the probability of reactive antigenic regions or domains for a quantitative analysis. The core stretch of 11–29 residues at the peak of each antigenic domain is defined as MAD.
Structure Modeling of Antigenic Domains
The structures of SIVmac239-gp120 were built through MODELLER9v7 program. The initial geometry of SIVmac239-gp120 was homology-modeled based on the structure of SIVmac32H (Protein Data Bank code2BF1) as templates. Accordingly, it was accomplished by energy minimization in the force field of MMFF94 seconds14 in the MOE software package (Chemical Computing Group, Montreal, Quebec, Canada).
YSD-Based FACS Titration Assays
The binding titers of antibodies in monkey sera against various antigenic domains were determined by FACS. Single yeast clone expressing V1 (aa106-158) or V2 (aa160-182) fragment, or 6–8 yeast clones expressing SIVmac239-Env fragments covering V1V2 (aa42-258), or V3/CD4-BS (aa307-473), or ecto-gp41 (aa531-668) were chosen. After induction of expression, these yeast clones were equally mixed and added into 96-well plates (about 105 per well). The CD20-expressing yeast clones were used as negative control. Serial diluted monkey sera were then added and incubated with yeasts at 4°C for 1 hour. After washing, yeast cells were then incubated with PE-conjugated antimonkey IgG, and the fluorescence intensity was detected by Calibur Flow Cytometer (BD Biosciences). The antibody binding level was determined according to geometric mean fluorescence intensity, and the binding titer was defined as the serum dilution at which the mean fluorescence intensity of the test serum was twice that of the negative control.
Flow cytometry software analysis was performed using FlowJo 7.6 (Tree Star Inc., Ashland, OR). Correlates were determined by Spearman rank correlation tests with GraphPad Prism 5.01 (GraphPad Software Inc., La Jolla, CA). Two-tailed P values were calculated for all analyses. SPSS programs (Version 16.0) were used for multivariable statistical analysis (SPSS Inc., Chicago, IL). The differences were considered statistically significant when P values were <0.05.
A YSD-Based Antigen Library of SIVmac239-Env for Mapping of Antigenic Determinants
A combinatorial antigen library was constructed to present overlapping and random SIVmac239 envelope fragments on the surface of yeast cells (see Figure S1, Supplemental Digital Content, https://links.lww.com/QAI/A625), which allows the comprehensive mapping of antigenic determinants in the entire SIVmac239-gp160. We first validated this library by testing SIV-specific mAbs KK65 (recognize a linear epitope at V1 region) and KK8 (recognize a conformational epitope spanning V1V2 region).15,16 We found that the MAD of KK65, as defined by the most frequently recognized antigenic sequences, is located at aa143-162 of V1 region (Fig. 1A), almost identical to the previously described position (aa140-160). Interestingly, we found that the conformational mAb KK8 primarily directed to the MAD at aa145-163 in C-terminal of V1 region (Fig. 1B), overlapping with the KK65 epitope. Subsequently, we investigated another conformational mAb VM18S (SIVmac1A11-gp130), whose MAD remains unknown.17 As shown in Figure 1C, the sorted sequences span a wider range in gp120 region (aa34 to 366), but the MAD of VM18S was identified at residue aa166-187 of V2 region. These results, therefore, demonstrated that our new method is robust and highly specific in identifying both linear and some conformation-dependent MAD in V1V2 region.
The MVTTgpe-Based Vaccine Regimen Induces Distinct Profiles of Antibody Responses in Rhesus Macaques
We then sought to analyze quantitatively the antigenic determinants of SIVmac239-specific polyclonal antibody responses in rhesus monkeys of our recent vaccine study.9 These sera were collected 6 weeks after the last vaccination. The schedule of vaccination and sampling is depicted in Figure 2A. To determine the distribution and frequency of antigenic determinants recognized by immunized sera, a total of 169 and 136 positive YSD clones to G1 and G2 were randomly obtained and sequenced, respectively (Fig. 2B). We found that the reactive SIV fragments for antibody recognition in both G1 and G2 covered the entire SIVmac239-Env while they were concentrated in V1V2, V3, and the main CD4 binding site (V3/CD4-BS), ecto-domain of gp41 (ecto-gp41), and cytoplasmic tail (CT) of gp41 regions. Using YSD-based FACS titration assays, we found that the titer of V1V2-specific antibodies was significantly higher in G1 than that in G2 (P < 0.05, Fig. 2C), whereas no significant differences were found in titer of anti-V3/CD4-BS (P = 0.3216) or anti-gp41 (P = 0.1597) antibodies between G1 and G2 (Fig. 2D, E). Consistently, the enzyme-linked immunosorbent assay (ELISA) titer of V1V2-specific antibody response in G1 was also significantly higher than that in G2 (P = 0.0022, see Figure S2C, Supplemental Digital Content, https://links.lww.com/QAI/A625). Critically, the anti-V1V2 antibody titer was inversely correlated with peak viral load (P = 0.0102, Fig. 2F). In addition, similar levels of anti-V1V2 antibody responses were observed with another 4 rhesus monkeys vaccinated with the MVTTgpe-based vaccine regimen (see Figure S3, Supplemental Digital Content, https://links.lww.com/QAI/A625), the study II animals as we previously described.9 Because these sera were collected at a different time point (week 2 instead of week 6 after the last vaccination), we were not able to conduct a correlation analysis using all 8 monkeys together. Nevertheless, our results demonstrated that the MVTTgpe-based vaccine regimen induced consistently higher levels of anti-V1V2 antibodies in rhesus macaques.
The MVTTgpe-Based Vaccine Regimen Uniquely Induces Antibody Responses Against the MAD in V2 Region
Based on the mapping of MAD of SIV-specific mAbs (Fig. 1), we sought to characterize antigenic domains and corresponding MADs recognized by polyclonal antibodies using the same computer algorithm as previously described.11 Six discrete antigenic domains were selected by G1 and G2 immunize sera, including domain 1 at N-terminus of gp120, domain 2 at V2 and C2 regions, domain 3 at V3 to V4 regions, domain 5 at the ecto-gp41 region, domains 7 and 8 at the CT region (Fig. 3A). We found that G1 and G2 sera share MAD (aa329-346) of domain 3 in V3 region, the MAD (aa602-622) of domain 5, and the MAD (aa729-764) of domain 7 in ecto-gp41 region. Although the fraction of anti-V3/CD4-BS antibodies was higher in G1 than that in G2, the titer of V3/CD4-BS-specific antibodies had no significant difference between G1 and G2 (Fig. 2D). This is because only 1 of the 4 monkey sera in G1 developed high anti-V3/CD4-BS antibody titer.
Critically, G1 sera preferentially recognized the MAD (aa163-181) in domain 2 of V2 region (13% AUC) overlapping with the MAD of mouse mAb VM18S (aa166-187), whereas G2 sera predominantly bound MAD (aa122-139) in domain 1 of V1 region (17% AUC) (Fig. 3A; see Table S1, Supplemental Digital Content, https://links.lww.com/QAI/A625). Because the major difference of predominantly recognized MADs between G1 and G2 was found in V1V2 region, we further measured the antibody binding titer against specific YSD-based MADs in V1 and V2, respectively. As a result, a significant higher titer of anti-V2 antibodies was found in G1 than that in G2 (P < 0.001, Fig. 4A), whereas the binding titers of anti-V1 antibodies were equivalent between G1 and G2 (P = 0.4367, Fig. 4B). The higher titer of V2-specific antibodies likely validated the dominant recognition of V2 region in 4 G1 macaques. Significantly, the titer of V2-specific antibodies was inversely correlated to both the peak and the set-point viral loads (P = 0.0022 and P = 0.0458, Fig. 4C, D). Our findings, therefore, provided evidence that the MVTTgpe-based vaccine regimen preferentially induced more robust antibody responses against V2 region as compared with the Ad5gpe-based regimen, likely reflecting the preference of B-cell recognition in response to the different types of vectored vaccines. It should be noted that a multivariate bootstrap linear regression analysis did not reveal a significant role of all types of immune responses in either set-point or peak viral loads (see Table S2, Supplemental Digital Content, https://links.lww.com/QAI/A625). More animals should be studied in future experiments to address this observation.
The MAD of V2 is Not Preferentially Recognized by Infection-Elicited Polyclonal Antibody Responses
To understand if vaccine-elicited B-cell memory response would respond quickly in control of experimentally inoculated pathogenic SIVmac239, G3 and G4 sera after viral challenge were collected for analysis (Fig. 2A). We found that except for the fully protected animal in G3, the overall titers of V1V2-specific antibody of G3 and G4 were higher than vaccine-elicited antibody responses (Fig. 4E vs Fig. 2C), suggesting that SIVmac239 infection indeed boosted B-cell memory responses to V1V2 region. There was, however, no significant difference in terms of V1V2-specific antibody titers between G3 and G4 (P = 0.608, Fig. 4E). Besides the 6 antigenic domains as described for G1 and G2 (Fig. 3A), we found additional domains 4 and 6 were reactive to G3 and G4 sera (Fig. 3B) after SIVmac239 infection. Unexpectedly, the identified MADs in each domain were quite similar between G3 and G4 (Fig. 3A; see Table S1, Supplemental Digital Content, https://links.lww.com/QAI/A625). Critically, the MAD of domain 2 (aa140-158) of G3 and G4 was located in V1 region (aa113-167), which is likely infection-specific and different from that of G1 sera. Therefore, different from the preferential selection of V2 MAD-specific antibodies by the MVTTgpe-based vaccination, SIVmac239 infection unlikely presents this V2 MAD preferentially for effective B-cell recognition in recalled antibody responses.
To further confirm whether or not SIVmac239 infection leads to V2 MAD-specific antibody responses, we also analyzed G5 and G6 sera from sham control group (Fig. 2A). We consistently detected similar titer of anti-V1V2 antibodies in G5 as compared with G3 and G4 sera, but less in G6 (Fig. 4E). Meantime, the anti–ecto-gp41 antibody titer was significantly lower in G5 than in G6 (P = 0.0169, Fig. 4F). Critically, the predominant MAD of domain 2 in G5 and G6 was not located in V2 but in V1 regions (aa144-163, Fig. 3C). This finding is consistent with previous findings that the V1 epitope is highly immunogenic by measuring antisera from SIV-infected macaques.16,18,19 These results provide further evidence that the MAD of V2, which was uniquely recognized by G1 sera, was not a predominant target of B-cell stimulation during the different stages of SIV infection in macaques.
Structural Analysis of the Mapped Antigenic Domains in V1V2 Region
Because SIV gp120 core crystal structure has been previously solved,20 we performed the analysis of mapped antigenic MADs. The homology-modeled structures indicate that the V2 loop (green) and part of the V1 loop (red) are exposed on the surface of gp120 as color-coded (Fig. 5). The footprint of each MAD in V1V2 region for antibody binding is color-coded in blue. We found that the immunodominant MAD2 (aa163-181) of G1 in V2 region was actually exposed on the surface (Fig. 5A). In contrast, the immunodominant MAD1 (aa122-139) of G2 in V1 region was concealed inside the structure shielded by V2 loop (Fig. 5B). Similarly, the immunodominant MAD2 (aa151-161) of G5 and G6 in V1 region was also hidden under V2 (Fig. 5C). Because of the predominant immunogenicity observed, these hidden MADs are apparently opened up for B-cell recognition during Ad5gpe-based vaccination and SIVmac239 infection, which are different from the MVTTgpe-based vaccination. Our results provide evidence that the MVTTgpe-based vaccination is able to shift B-cell recognition by targeting the well-exposed MAD in V2 region for better antibody induction and probable protection.
In this study, we report a new YSD-based antigen library of SIVmac239-gp160 to investigate quantitatively not only the linear epitope but also some conformational antigenic determinants recognized by SIV-specific mAbs and polyclonal antibodies. Using this new technique, we were able to confirm the MAD of mAb KK65 at aa143-162 of V1 region as previously described.15 For the first time, we were also able to define the MAD of mAb KK8 aa145-163 at C-terminal V1 region and the MAD of mAb VM18S at residue aa166-187 in V2 region. These 2 mAbs have previously been considered to be specific to conformational epitopes without well-defined sequences. Notably, the mAb KK8 was able to neutralize a glycan mutant of SIVmac239 after the removal of glycosylation sites at residues 146 and 156 within our mapped MAD but not the original SIVmac239 strain, indicating the significance of this MAD for KK8 recognition.17 As for the mapping of polyclonal antibody responses, we characterized sera in rhesus monkeys elicited by vaccination (eg, G1 and G2), vaccination plus SIVmac239 challenge (eg, G3 and G4), and experimental SIVmac239 infection alone (eg, G5 and G6) (Fig. 2A). We were able to define 8 antigenic domains and related MADs across entire SIVmac239-Env. Our findings of MAD in the immunodominant ecto-gp41 region are also consistent to previous studies (Fig. 3). The MAD of domain 5 in G1–G6 is located between 2 heptad repeats (HR1 and HR2) in ecto-gp41 region, which matches the immunodominant Gnann epitope found in both HIV and SIV infection.19,21 The MAD of domain 7 corresponds to another previously defined immunodominant Kennedy epitope.16,22,23 Moreover, other 6 antigenic domains correspond to linear epitopes identified in infected monkey sera by peptide-based ELISA.19,24 Our results, therefore, have provided the quantitative analysis of antigenic determinants in terms of conformation, AUC, and precise MAD mapping, which are new and useful for better understanding of both monoclonal and polyclonal antibody responses. We, however, acknowledge the limitation of our YSD-system that it is unable to map gp160 trimer-dependent conformational structures (eg, CD4-induced antibody, data not shown).25,26It is possible that some viral epitopes displayed by yeast are not properly glycosylated.
Different types of virus-vectored vaccines may elicit distinct polyclonal antibody responses against SIV infection. We recently reported that MVTTgpe-based vaccine regimen elicits protective immunity against pathogenic SIVmac239 intrarectal challenge in rhesus monkeys.9 Because this regimen (G1 in Fig. 2A) resulted in better viremia control and slower disease progression than that of the Ad5gpe-based vaccine (G2), we aimed to determine the role of SIV-specific polyclonal antibody responses in protection. We first confirmed previous findings that although G1 and G2 showed similar ELISA-binding titer to SIVmac239, G1 had a higher titer of neutralizing antibodies than G2 against SIVmac1A11 (see Figures S2A and S2B, Supplemental Digital Content, https://links.lww.com/QAI/A625).9 Using the newly established YSD library, we then found that the recognition frequency of V2 region is significantly higher in G1 sera than in G2. In particular, G1 sera preferentially recognize the predominant MAD (aa163-181) in V2 region, whereas G2 sera primarily react with MAD (aa122-139) in V1 region (Fig. 3A). Antibodies to the MAD in V2 have resulted in the higher antibody binding titer of G1 to V1V2 than that of G2 animals (Figs. 2C, 4A, B). Because the major vaccination difference between G1 and G2 is the mucosal prime using the MVTTgpe-based live vaccine, our findings provided evidence that there are likely distinct antigenic determinants of B-cell recognition in response to the different types of virus-vectored vaccine. In analog to the RV144 AIDS vaccine clinical trial and the mimic study in monkeys,8 our data provided additional supporting evidence that the vaccinia-based prime may play a critical role in eliciting V2-directed antibody responses in a heterologous prime and boost vaccine regimen.
An inverse association between anti-V1V2–binding antibodies and the infection risk has been discovered in the RV144 trial.2,3,27 Studies in rhesus macaques also indicated that the level and avidity of anti-V2–binding antibodies were likely associated with vaccine-mediated protection against SIVmac251 acquisition,8,28 with an Env-V2 sieve effect seen in breakthrough infections that lend support to the importance of immune response against Env-V2 in reducing HIV-1 and SIV risk.29,30 Our preliminary data, however, have yet established correlative relationships between MVTTgpe-elicited anti-V2 antibodies and virion capture, ADCC, or reducing viral infectivity by the interaction with integrin α4β7-expressing cells as previously described by others (data not shown).8,31,32 Nevertheless, our results provide supporting evidence that vaccinia MVTTgpe-elicited anti-V2 antibodies probably play a role in protection by several lines of evidence: (1) the significantly higher levels of anti-V2 antibodies (eg, titer and AUC) in G1 as compared with G2 (Figs. 3A, 4A), (2) an inverse association between anti-V2–binding antibodies and the peak/set-point viral load (Fig. 4C, D), (3) the loss of anti-V2 antibody responses in G6 of macaques with progression to simian AIDS (Figs. 3C, 4E), and (4) significantly delayed disease progression of G1 but not G2 macaques.9 Future investigation, however, remains necessary to understand the possible protective mechanism mediated by MVTTgpe-elicited anti-V2 antibodies. Notably, the sterile protection was achieved in only 1 macaque of MVTTgpe-elicited protection.9 It will be important to determine if the levels of MVTTgpe-elicited anti-V2 antibodies would confer better protection against SIV acquisition in a repeat, low-dose challenge model system.8,28 Because pathogenic SIVmac239 infection evades MVTTgpe vaccine-elicited B-cell responses to V2 region by shifting immunodominance to V1 and gp41, it probably represents a previously unrecognized mechanism of immune evasion by AIDS viruses, which needs to be further investigated. Because of the lack of recalled anti-V2 antibody responses by SIVmac239 infection, future vaccine strategies must be able to maintain a high titer of V2-directed antibodies for protection, which would require repeat vaccinations, a daunting challenge for virus-vectored vaccines because of the issues of pre-existing antivector immunity.
1. Rerks-Ngarm S, Pitisuttithum P, Nitayaphan S, et al.. Vaccination with ALVAC and AIDSVAX to prevent HIV-1 infection in Thailand. New Engl J Med. 2009;361:2209–2220.
2. Haynes BF, Gilbert PB, McElrath MJ, et al.. Immune-correlates analysis of an HIV-1 vaccine efficacy trial. New Engl J Med. 2012;366:1275–1286.
3. Zolla-Pazner S, Decamp A, Gilbert PB, et al.. Vaccine-induced IgG antibodies to V1V2 regions of multiple HIV-1 subtypes correlate with decreased risk of HIV-1 infection. PLoS One. 2014;9:e87572.
4. Putkonen P, Thorstensson R, Ghavamzadeh L, et al.. Prevention of HIV-2 and SIVsm infection by passive immunization in cynomolgus monkeys. Nature. 1991;352:436–438.
5. Van Rompay KK, Berardi CJ, Dillard-Telm S, et al.. Passive immunization of newborn rhesus macaques prevents oral simian immunodeficiency virus infection. J Infect Dis. 1998;177:1247–1259.
6. Lewis MG, Elkins WR, McCutchan FE, et al.. Passively transferred antibodies directed against conserved regions of SIV envelope protect macaques from SIV infection. Vaccine. 1993;11:1347–1355.
7. Haigwood NL, Watson A, Sutton WF, et al.. Passive immune globulin therapy in the SIV/macaque model: early intervention can alter disease profile. Immunol Lett. 1996;51:107–114.
8. Pegu P, Vaccari M, Gordon S, et al.. Antibodies with high avidity to the gp120 envelope protein in protection from simian immunodeficiency virus SIVmac251 acquisition in an immunization regimen that mimics the RV-144 Thai trial. J Virol. 2013;87:1708–1719.
9. Sun C, Chen Z, Tang X, et al.. Mucosal priming with a replicating-vaccinia virus-based vaccine elicits protective immunity to simian immunodeficiency virus challenge in rhesus monkeys. J Virol. 2013;87:5669–5677.
10. Boder ET, Wittrup KD. Yeast surface display for screening combinatorial polypeptide libraries. Nat Biotechnol. 1997;15:553–557.
11. Zuo T, Shi X, Liu Z, et al.. Comprehensive analysis of pathogen-specific antibody response in vivo based on an antigen library displayed on surface of yeast. J Biol Chem. 2011;286:33511–33519.
12. Lin Z, Li S, Chen Y. Identification of viral peptide fragments for vaccine development. Methods Mol Biol. 2009;515:261–274.
13. Chao G, Lau WL, Hackel BJ, et al.. Isolating and engineering human antibodies using yeast surface display. Nat Protoc. 2006;1:755–768.
14. Halgren TA. MMFF VI. MMFF94s option for energy minimization studies. J Comput Chem. 1999;20:720–729.
15. Kent KA, Gritz L, Stallard G, et al.. Production and of monoclonal antibodies to simian immunodeficiency virus envelope glycoproteins. AIDS. 1991;5:829–836.
16. Kent KA, Rud E, Corcoran T, et al.. Identification of two neutralizing and 8 non-neutralizing epitopes on simian immunodeficiency virus envelope using monoclonal antibodies. AIDS Res Hum Retroviruses. 1992;8:1147–1151.
17. Johnson WE, Sanford H, Schwall L, et al.. Assorted mutations in the envelope gene of simian immunodeficiency virus lead to loss of neutralization resistance against antibodies representing a broad spectrum of specificities. J Virol. 2003;77:9993–10003.
18. Benichou S, Venet A, Beyer C, et al.. Characterization of B-cell epitopes in the envelope glycoproteins of simian immunodeficiency virus. Virology. 1993;194:870–874.
19. Silvera P, Flanagan B, Kent K, et al.. Fine analysis of humoral antibody response to envelope glycoprotein of SIV in infected and vaccinated macaques. AIDS Res Hum Retroviruses. 1994;10:1295–1304.
20. Chen B, Vogan EM, Gong H, et al.. Structure of an unliganded simian immunodeficiency virus gp120 core. Nature. 2005;433:834–841.
21. Gnann JW Jr, Nelson JA, Oldstone MB. Fine mapping of an immunodominant domain in the transmembrane glycoprotein of human immunodeficiency virus. J Virol. 1987;61:2639–2641.
22. Kennedy RC, Henkel RD, Pauletti D, et al.. Antiserum to a synthetic peptide recognizes the HTLV-III envelope glycoprotein. Science. 1986;231:1556–1559.
23. Eberle J, Loussert-Ajaka I, Brust S, et al.. Diversity of the immunodominant epitope of gp41 of HIV-1 subtype O and its validity for antibody detection. J Virol Methods. 1997;67:85–91.
24. McBride BW, Corthals G, Rud E, et al.. Comparison of serum antibody reactivities to a conformational and to linear antigenic sites in the external envelope glycoprotein of simian immunodeficiency virus (SIVmac) induced by infection and vaccination. J Gen Virol. 1993;74:1033–1041.
25. Edinger AL, Ahuja M, Sung T, et al.. Characterization and epitope mapping of neutralizing monoclonal antibodies produced by immunization with oligomeric simian immunodeficiency virus envelope protein. J Virol. 2000;74:7922–7935.
26. White TA, Bartesaghi A, Borgnia MJ, et al.. Three-dimensional structures of soluble CD4-bound states of trimeric simian immunodeficiency virus envelope glycoproteins determined by using cryo-electron tomography. J Virol. 2011;85:12114–12123.
27. Yates NL, Liao HX, Fong Y, et al.. Vaccine-induced env V1-V2 IgG3 correlates with lower HIV-1 infection risk and declines soon after vaccination. Sci Transl Med. 2014;6:228ra239.
28. Barouch DH, Liu J, Li H, et al.. Vaccine protection against acquisition of neutralization-resistant SIV challenges in rhesus monkeys. Nature. 2012;482:89–93.
29. Sina S, Tovanabutra S, Sanders-Buell E, et al.. Evidence for env-V2 sieve effect in breakthrough SIVMAC251 infections in rhesus monkeys vaccinated with Ad26/MVA and MVA/Ad26 constructs. Retrovirology. 20129(Suppl 2):O32.
30. Rolland M, Edlefsen PT, Larsen BB, et al.. Increased HIV-1 vaccine efficacy against viruses with genetic signatures in Env V2. Nature. 2012;490:417–420.
31. Liao HX, Bonsignori M, Alam SM, et al.. Vaccine induction of antibodies against a structurally heterogeneous site of immune pressure within HIV-1 envelope protein variable regions 1 and 2. Immunity. 2013;38:176–186.
32. Arthos J, Cicala C, Martinelli E, et al.. HIV-1 envelope protein binds to and signals through integrin alpha4beta7, the gut mucosal homing receptor for peripheral T cells. Nat Immunol. 2008;9:301–309.