Vaccine reverse engineering Burton focuses

“Vaccine reverse engineering” (Burton, 2002) focuses on inferring the structure of improved immunogens based on interactions with neutralizing antibodies. While “epitope focused antigen design” (Correia et al., 2014) and other structural approaches (Kulp and Schief, 2013) show promise in identifying the three dimensional structure of antigens that could elicit mAb-like neutralizing responses, it also is recognized that the link between physical structure and landscapes recognized by mAbs can be analyzed using antigenic approaches (Dormitzer et al., 2008) that examine the binding specificity of mAbs. For the many neutralizing mAbs with unknown three-dimensional structures of antigen-antibody complexes, identification of antigens that can optimally present the critical epitopes to the immune system remains a challenge. We reasoned that a given mAb could be antigenically characterized by identifying a series of tightly binding antigens that, collectively and uniquely define the binding specificity of the antibody, and thus explored whether probing neutralizing antibodies with a high-complexity random peptide library could allow identification of optimized antigens that capture key information about a mAb of interest, holding the potential to generate immune responses that have properties similar to that of the mAb. To increase the chance of successful antigen selection, we used messenger ribonucleic recommended reading (mRNA) display, which generates random peptide libraries with complexity as high as >1013 unique sequences (Takahashi et al., 2003). Each peptide is covalently bound to its encoding mRNA through a puromycin linkage, enabling in vitro selection of peptide libraries (ranging from 1 to over 100 amino acids) that bind to targets of interest (including antibodies) and the use of reverse transcriptase polymerase chain reaction (RT-PCR) and sequencing to determine the nucleic acid sequence corresponding to the peptides that bind the target. The high library complexity of mRNA display allows selection of extremely rare and high affinity binding partners, and its use in epitope mapping has been described (Baggio et al., 2002; Ja et al., 2005). In the present study, we integrated mRNA display and high throughput sequencing (mRNA display-HTS) to identify peptide mimotopes that capture important information from HCV neutralizing mAb41 (Duan et al., 2012). This approach enabled high-resolution mapping of this mAb\'s binding specificity. MAb41 was previously obtained from mice immunized with a HCV genotype 1a (GT1a) wild type peptide fragment named peptide A (pA) (Duan et al., 2012), comprising amino acid residues 412 to 447 of the E2 protein. MAb41 neutralizes cell culture HCV (HCVcc) in a GT1a-specific manner (Duan et al., 2012). Using mRNA display-HTS to select for peptides that bind to mAb41, we hoped to identify peptide mimotopes that capture key information unique to this neutralizing antibody\'s antigen binding sites, potentially identifying peptide antigens that can present mAb-specific epitopes in an optimal context. While these peptide mimotopes likely will have limitations as stand-alone vaccine candidates, the identified antigens could potentially serve as optimized core epitope components in the context of scaffolds or other antigen presentation strategies. We evaluated the binding affinity of the selected peptides to the original mAb, and further investigated their ability to stimulate immune responses.
Materials and Methods
Results
Discussion We did not observe clear correlations between relative mAb41 binding affinities of peptides as assessed by different methods (the mRNA display selection, or in Octet or ELISA assays) and virus neutralization of induced polyclonal serum, suggesting that there may be differences among the binding interactions detected by these methods. However, all methods indicated strong mAb41 binding of p41_1, p41_3, p41_4, and p41_5. p41_2 showed both relatively lower mAb41 affinity by ELISA and weaker neutralizing immune responses. Improved neutralizing responses induced by mAb41-selected peptides than pA could also be mediated by the duplicated W(L/I)XX(L/I) motif and/or the acidic residues that allow a more favorable presentation of the epitope. It is possible that pA presents the critical residues in a less favorable manner either due to the loss of optimal conformation when removed from its native context (Fibriansah et al., 2014), or the lack of mAb41 preferred features (duplicated W(L/I)XX(L/I) motif and acidic residues). This is evidenced by the absence of detectable mAb41-like antibodies in anti-pA polyclonal sera, despite the fact that this peptide was used for the initial immunizations that resulted in the isolation of mAb41 (Duan et al., 2012). These results suggest that mAb41 was a rare species in the antibody pool originally induced by pA. Although some anti-pA antisera showed GT1a neutralizing activity, this activity could have been mediated by non-mAb41-like antibodies, because pA also contains other neutralizing epitopes (Duan et al., 2012). In contrast, p41_1 and p41_3 induced a higher frequency of mAb41-like antibodies, targeting specific residues within the critical motif, suggesting a more favorable presentation of the WLXXL epitope than in pA. These results indicate that mimotopes identified by mRNA display-HTS have potential as improved vaccine antigens that shift the immune response towards the production of epitope-specific antibodies with desired activity, achieving a result similar to that of other epitope-focused vaccine design approaches (Correia et al., 2014). The mRNA display-HTS approach could also be applied to identification of linear mimetics for mAbs against conformational epitopes. Due to the structural complexity of conformational epitopes, linear peptides may not be able to recapitulate the full three dimensional structure of the original antigens. Using more complex libraries (for example, of 80mer (Keefe and Szostak, 2001; Cho et al., 2000)) could help to address this limitation.