Development of a Neoepitope Prioritization Platform to Identify Immunogenic Neoantigens from Oncogenic FGFR3-TACC3 Fusion in Glioblastoma

This study investigates the FGFR3-TACC3 fusion as a tumor-specific neoantigen source and establishes an end-to-end computational and experimental framework for the identification, prioritization, and validation of immunogenic neoantigens derived from chimeric RNAs. The FGFR3-TACC3 fusion, generated through an intrachromosomal rearrangement on chromosome 4, produces a unique junctional sequence that constitutively activates oncogenic signaling while remaining absent in normal tissues, thereby representing a highly specific therapeutic target. To evaluate its neoantigenic potential, 27 peptides (9-11 mers) spanning the fusion junction were assessed for HLA class I binding using NetMHCpan 4.1, MixMHCpred 3.0, and BigMHC 1.0, alongside an in vitro flow cytometry-based binding assay. Eleven peptides, including 10 fusion-derived neoantigens, were validated as strong binders to HLA-A*01:01. NetMHCpan 4.1 demonstrated superior predictive performance, exhibiting a significant inverse correlation with experimental binding and high classification accuracy (AUC = 0.9669). An experimentally derived threshold (EL Rank < 32.7) achieved maximal sensitivity while maintaining high specificity. This was further refined to a more stringent cutoff (EL Rank < 25.7) using an expanded multi-fusion dataset comprising FGFR3-TACC3, CD74-NRG1, and ESR1-PRKN across multiple HLA alleles, improving predictive robustness and specificity. Functional validation using ELISpot assays demonstrated that 9 of 10 fusion derived in vitro validated binders induced significant IFN-γ secretion from CD8 T cells, confirming their immunogenicity. Cross-reactivity assessment using the CrossDome pipeline identified two neoantigens with potential T cell off-target effects, enabling safety-informed prioritization of candidate peptides. Integration of single-cell transcriptomic and TCR repertoire analyses revealed that peptide stimulation induces coordinated activation of immune programs, including T cell activation, cytokine signaling, proliferation, and metabolic pathways. The best peptide pool elicited increased abundance weighted clonotype diversity and convergence without evidence of oligoclonal dominance. Clonal analysis further identified antigen-responsive CD8 T cell populations with conserved CDR3 features, supporting structural stability and antigen specificity. Collectively, this work establishes a scalable and reproducible workflow integrating fusion detection, fusion reconstruction, neoepitope prediction, experimental validation, and safety filtering. This platform enables both fusion-centric and peptide-centric strategies for multi-epitope vaccine design and TCR-based therapies, while introducing data-driven refinement of NetMHCpan 4.1 thresholds to enhance translational precision in neoantigen selection.

Name thesis:degree_name

Doctor of Philosophy

Discipline thesis:degree_discipline

Biochemistry

Grantor

University of Houston

Year dc:date.issued

2026

Author dc:creator

• Thevasagayampillai, Shiyanth 1994-

Advisor dc:contributor.advisor

• Gunaratne, Preethi

Committee members dc:contributor.committeemember

• Chin, Yo-Lin
• Antunes, Dinler A
• Sumazin, Pavel

• TCR
• Cancer
• Vaccine
• FGFR3-TACC3
• Fusion
• Neoantigen

Language dc:language.iso

English

Handle dc:identifier.uri

https://hdl.handle.net/10657/21509

OAI identifier oai:identifier

oai:uh-ir.tdl.org:10657/21509