Development of new LC-MS tools to investigate new biological entities immunogenicity

Protein therapeutics are expanding rapidly, and with them the need to anticipate and mitigate unwanted immune responses that can lead to anti-drug antibodies, loss of efficacy, and adverse events. This dissertation addresses preclinical T cell–dependent immunogenicity risk by developing and validating liquid chromatography–mass spectrometry (LC-MS) workflows, centered on MHC-Associated Peptide Proteomics (MAPPs), and by integrating structural MS with computational modeling to characterize antigen presentation and antibody–antigen interfaces. The work is framed within the broader context of immunogenicity assessment, where in vitro, in silico, and analytical evidence must be combined to inform risk management for biotherapeutics. The thesis first establishes and optimizes a robust MAPPs assay workflow using monocyte-derived dendritic cells, evaluating technical reproducibility, culture conditions, bead chemistries, and donor selection strategies. It characterizes peptide recovery and abundance, examines the impact of donor HLA genotype on peptide repertoires, and benchmarks positive controls to ensure assay performance and interpretability. Using these optimized conditions, the dissertation demonstrates concordance between experimentally eluted MHC II ligands and in silico predictions (e.g., NetMHCpan), across representative test systems including a therapeutic antibody (Infliximab), an allergen (Bet v1a), and a model protein (KLH). These results support the use of MAPPs to complement sequence-based prediction, refine neo-epitope discovery, and strengthen preclinical immunogenicity assessment. Building on peptide-level insights, the work extends structural characterization with cross-linking mass spectrometry (XL-MS) to map epitopes and paratopes in clinically relevant antibody–antigen complexes (Infliximab–TNFα, Cetuximab–EGFR, Matuzumab–EGFR). Cross-link constraints are integrated with relative surface accessibility and epitope/paratope probability to guide docking and integrative modeling, improving interface resolution and model plausibility. The dissertation further benchmarks multiple computational pipelines, including AlphaFold Multimer and a cross-link-aware approach (AlphaLink2), showing that XL-MS restraints and interface-focused predictors enhance complex modeling fidelity and correlate with accuracy metrics (e.g., iRMSD, DockQ). Finally, it evaluates peptide–HLA II complex modeling across loci and motif orientations, analyzing how cross-link data, peptide length, and model confidence scores affect structural accuracy; dedicated case studies (e.g., CLIP–HLA DRB1*01:01) illustrate the practical gains of integrative strategies. Collectively, the dissertation provides two principal contributions: it evidences the correlation between MAPPs-derived MHC II ligands and in silico predictions in preclinical settings, and it establishes mass spectrometry—particularly structural MS—as a critical validation and discovery enabler that complements computational methods for epitope identification and complex modeling. These advances suggest actionable paths to de-risk new biological entities by combining immunopeptidomics with integrative modeling, thereby strengthening immunogenicity risk assessment strategies in early development