How many post-translationally proteasomal spliced HLA-I peptides can be confidently identified in large scale MS/MS data?

Abstract

Mass spectrometry-based immunopeptidomics is the method of choice for characterizing human leukocyte antigen binding peptides (HLAp) which play a key role in adaptive immunity. Recently two studies [1,2] reported that a large portion (20-40%) of HLAp’s originate from proteasomal splicing events. Both studies used large-scale mass spectrometry measurements of HLAp’s and bioinformatics analysis: the first study created a large sequence database containing all cis-spliced peptides to search the HLAp MS/MS spectra and the second study was based on de novo sequencing and subsequent filtration of cis-and trans-spliced peptides. We performed a reanalysis of the data of 1 and could not confirm most of the identifications reported in these studies 3,4 . Our results indicate that the proportion of HLAp that could be explained by proteasomal splicing was maximally a few percent. More recently, we reanalyzed part of the data from 2 . Comparison of predicted and observed hydrophobicity for the reported spliced and nonspliced peptides, and reanalysis of the data using state of the art MS/MS search engines indicate that the fraction of spliced peptides reported was vastly overestimated. Here we discuss the issues that arise when searching spliced peptides in large scale MS/MS data. We propose practical MS/MS search strategies and methods to properly control the error rate. We would like to emphasize that spliced peptides identifications from large-scale immunopeptidomics measurements should be carefully validated and confirmed with further biological experiments.