(590d) Microsphere-Assisted Peptide Screening (MAPs): High-Throughput Identification of Promiscuous MHCII-Binding Peptides for T-Cell Epitope Vaccine Designs | AIChE

(590d) Microsphere-Assisted Peptide Screening (MAPs): High-Throughput Identification of Promiscuous MHCII-Binding Peptides for T-Cell Epitope Vaccine Designs


Smith, M. - Presenter, University of Michigan
Wen, F., University of Michigan
T cells recognize antigenic peptides as part of a peptide-major histocompatibility complex (pMHC) via T-cell receptors (TCRs). The cognate peptide is termed a T-cell epitope and can be used to develop a T-cell epitope vaccine. However, before T-cell epitopes can be identified and used as vaccines, candidate peptides must first be presented on MHC molecules. Therefore, peptide-MHC binding is the most discriminating step in the identification of T-cell epitopes and significant effort has been dedicated to identifying peptides that bind a specific MHC protein. This has proven to be a challenge in the field as MHC proteins are extremely polymorphic which results in different T-cell epitope repertoires in different persons. MHC allelic diversity is particularly challenging in the context of designing T-cell epitope vaccines, where in order to provide protection to a diverse population, the selected peptides should bind promiscuously to many different MHC alleles. Identifying peptide-MHC binders is even further complicated for class II MHC (MHCII), as MHCII have open-ended binding grooves that can accommodate long, variable-length peptides. Several computational and experimental systems have been developed to determine if a specific peptide will bind to an MHC allele. Current computational systems using complex matrix based systems and machine learning algorithms are incredibly high-throughput and can accurately predict peptide binding to many MHCI alleles. Despite the successful application of computational approaches to peptide-MHCI binding, even the most sophisticated in silico systems struggle to accurately predict peptides binding to MHCII. In contrast, most experimental systems are capable of providing incredibly accurate quantitative peptide-binding data for both MHCI and MHCII alleles. However, these systems tend to be labor-intensive, low-throughput, and are often applied to limited sets of MHC alleles and thus can miss promiscuously-binding peptides.

To address some of the limitations of current experimental systems used to screen peptide-MHC binding, we developed a novel high-throughput screening method to rapidly identify promiscuously binding peptides. Microsphere-assisted peptide screening (MAPs) coupled with our modular baculovirus-MHCII expression system allows one to screen large peptide libraries for binding to many MHCII alleles. The MAPs approach consists of performing peptide-exchange reactions in which an epitope-tagged peptide is loaded on to one of four common MHCII alleles. These exchanged-pMHC complexes are loaded to the surface of microspheres and analyzed using flow cytometry to determine the relative peptide binding. We validated the MAPs system by screening a 20mer peptide library with known binding data and analyzed the results using receiver operating characteristic (ROC) curves. This ROC analysis demonstrated the MAPs system is rapid and accurate with an area under curve greater than 0.90. We then applied the MAPs approach to successfully identify four promiscuously-binding peptides within a 26-member peptide-library of the rotavirus capsid protein VP7. Interestingly, structural analysis of the rotavirus VP7 protein revealed that these four promiscuously-binding peptides share a number of residues with known antibody neutralization sites on the native VP7 trimer. Finally, the complete peptide-binding dataset for the VP7 library screen was compared to peptide binding predictions by computational methods within the IEDB Analysis Resource including SMM-align, NN-align, and NetMHCIIpan. The predictive-algorithms captured many of the peptide-binders observed using the MAPs system; however, the computational predictions had a false-negative rate of roughly 20% including the four promiscuously binding peptides discovered using MAPs. Therefore, in addition to serving as a robust experimental platform for identifying promiscuously binding peptides for potential T cell epitope vaccines, MAPs should also contribute to the improvement of computational systems that predict peptide-MHCII binding in silico.