The analysis of the Human Leukocyte Antigen (HLA)-associated peptide (HLAp) ligand repertoire by mass spectrometry (MS)-based immunopeptidomics has become indispensable for modern vaccine design and personalized immunotherapy.^1,2 Peptide antigens from endogenous proteins are typically presented via HLA class I complexes to immune cells. These are particularly relevant for defining the cancer-associated or -specific HLA ligandome which can be exploited as immunotherapy targets. However, human sample material is scarce and precious, and the identification of HLAps is challenging due to their individual specificity, high diversity, low abundance, and high dynamic range.^1,2,3 Therefore, there is a strong need to develop high-sensitivity immunopeptidomics workflows to profile the immunopeptidome and identify potential disease-relevant HLAps.

The two main methods for the enrichment of HLA class I immunopeptides are Immunoprecipitation (IP) and Mild Acid Elution (MAE). IP typically uses bead-coupled antibodies to enrich HLA-peptide complexes from lysed cells. MAE utilizes a mild acid to strip HLA-class I bound (HLAIps) ligands from the surface of whole cells.⁴ While the latter is often regarded as less specific, we show that HLAIp enrichment using an optimized MAE HLA class I sample preparation workflow enables immunopeptidome analysis from less than 1x10⁶ cells with high specificity. More specifically, we show that our protocol enriches a high proportion (>80%) of potential HLAIp peptides (8-13mers) comprising 75% predicted HLA ligands with the respective binding motifs. Moreover, further optimization increased the identification of 8-13mers by 50% while maintaining the HLAIp enrichment specificity. Using two cell lines JY and Raji harbouring distinct HLA-I Alleles and thus also ligand motifs, we further evaluated our optimized MAE protocol by comparing MAE and IP bulk preparations (40x10⁶ cells per sample). Irrespective of the cell line, IP and MAE enriched on average over 7,000 potential HLAIps with high specificity (>85% 8-13mers) comprising mainly predicted binders (>75%). Importantly, >80% of predicted binders were shared between both protocols. This suggests that no evident bias is introduced using our optimized MAE instead of IP for HLAIp enrichment. Utilizing an Evosep LC coupled to a timsTOF Ultra MS, we identified more than 2500 HLAIps per 1x10⁶ cells of JY and Raji in Thunder-DDA-PASEF⁵ and library-based HLA-tailored DIA analyses further leveraged HLAIp identification 1.5-fold.

In summary, we optimized MAE sample preparation and MS methods to enable immunopeptidomics profiling of samples using less than 1x10⁶ cells. Although the last steps for method optimization and validation are still ongoing, this workflow may enable the analysis of sorted cell populations to study HLAIp presentation in the heterogenous environment of the tumor.