Pitfalls in HLA Ligandomics-How to Catch a Li(e)gand

Abstract

Knowledge about the peptide repertoire presented by human leukocyte antigens (HLA) holds the key to unlock target-specific cancer immunotherapies such as adoptive cell therapies or bispecific T cell engaging receptors. Therefore, comprehensive and accurate characterization of HLA peptidomes by mass spectrometry (immunopeptidomics) across tissues and disease states is essential. With growing numbers of immunopeptidomics datasets and the scope of peptide identification strategies reaching beyond the canonical proteome, the likelihood for erroneous peptide identification as well as false annotation of HLA-independent peptides as HLA ligands is increasing. Such "fake ligands" can lead to selection of nonexistent targets for immunotherapeutic development and need to be recognized as such as early as possible in the preclinical pipeline. Here we present computational and experimental methods that enable the identification of "fake ligands" that might be introduced at different steps of the immunopeptidomics workflow. The statistics presented herein allow discrimination of true HLA ligands from coisolated HLA-independent proteolytic fragments. In addition, we describe necessary steps to ensure system suitability of the chromatographic system. Furthermore, we illustrate an algorithm for detection of source fragmentation events that are introduced by electrospray ionization during mass spectrometry. For confirmation of peptide sequences, we present an experimental pipeline that enables high-throughput sequence verification through similarity of fragmentation pattern and coelution of synthetic isotope-labeled internal standards. Based on these methods, we show the overall high quality of existing datasets but point out limitations and pitfalls critical for individual peptides and how they can be uncovered in order to identify true ligands.

Keywords: HLA ligands; TCR targets; cancer immunotherapy; immunopeptidomics; peptide quality control.

Graphical abstract

Fig. 1

Protein coverage by class I HLA ligands.A, comparison of coverage for a representative gene (chromobox homolog 6, CBX6) showing presentation hotspot (anchor amino acids in bold) and potential contaminant (hemoglobin alpha 2, HBA2) with characteristic peptide ladders due to proteolytic degradation. Contaminations are colored in red while true HLA ligands are colored blue. B, distribution of protein coverage ratios in ZH2018. Estimated Gaussian mixture model of contamination distribution shown in red and ligand distribution in blue. The cutoff for assessment of contaminations at 1% false discovery rate is visualized as dashed line. C, distribution for peptide coverage ratio and D, HLA ligand propensity with analogical modeling as before.

Fig. 2

Contamination analyses-based proteolytic contamination count PCC ≥ 2.A, proteolytic fragments observed in monoallelic cell lines of AB2017 for different HLA alleles with low frequency of contamination on average but substantial increase in HLA-A∗68:02. B, proteolytic fragments as a function of sample input (tumor tissue weight) for lung cancer adenocarcinoma tissue showing low constant baseline levels of contaminations. C, number of proteolytic fragments in population-scale immunopeptidome dataset ZH2018 aggregated according to tissue highlighting three groups of degree of contamination: low (cell lines and uncultured liquid cancers), medium (solid tissue), and high (digestive organs and healthy liquid tissue).

Fig. 3

Peptide carryover observed after a single nanoLC-MS run of a representative pan-HLA class I (W6/32) peptidomics sample.A, carryover peptide IDs identified in database search of blank runs performed subsequently to the sample analysis. B, for 16 carryover peptides retained until Blank #3 different characteristics are shown including the percentile of MS1 peak area, retention time, prevalence in ZH2018 samples, and HLA allotype. C, MS1 peak areas and their percentile for baseline sample (blue) and the 16 carryover peptides (red). D, reduction of MS1 signal intensity for carryover peptides across repeated blank runs shown for all 16 peptides and fitted (red line) using an exponential decay model resulting in log₁₀Area(t) = 5.21 + 4.60e^−0.72t. E, sequence logo of HLA-B∗44:03 for peptides from AB2017. F, sequence logo for simulated data assuming carryover from HLA-A∗24:02 into B∗44:03. The 6% most abundant peptides from the A∗24:02 data were computationally admixed to B∗44:03, simulating consecutive LC-MS analysis without intermittent LC cleanup.

Fig. 4

In-source fragmentation.A, example of a source fragmentation of the TP53-derived A∗02:01 ligand TP53_187-197 (GLAPPQHLIRV) resulting in concomitant detection of the truncated C-terminal source fragment APPQHLIRV on an HLA-A∗02 specific immunoprecipitation using the antibody BB7.2. As the fragment fulfills the consensus motif for B∗07 and B∗35, it may be misinterpreted as a veritable HLA ligand. B, example of a peptide that can occur both as a source fragment and as a veritable HLA ligand. Extracted MS1 ion chromatograms are shown for a pan-HLA class I immunoprecipitation using the antibody W6/32 on a sample expressing two restricting allotypes (A∗03:01 & C∗04:01). The short variant LFDHAVSKF was detected as a source fragment of the A∗03:01 ligand at experimental RT 31.2 and as a veritable C∗04:01 ligand at experimental RT 35.6. A and B, the main panel summarizes precursor ion intensities of the two peptide species over retention time. The cutouts provide the underlying MS1 extracted ion chromatograms including precursor m/z. C, frequency distribution of N-terminal amino acid loss by in-source fragmentation. D, sequence logo of peptides showing loss of one N-terminal amino acid by source fragmentation. E, sequence logo of peptides with loss of two amino acids.

Fig. 5

Peptide sequence verification. Experimental disambiguation of conflicting peptide sequence annotations by spike-in of stable isotope-labeled internal standard (SIL-) peptides. A, differentially labeled SIL counterparts for two possible spectral annotations were spiked into an HLA peptidome sample previously found positive for the spectrum in DDA-MS and analyzed by targeted MS (PRM). For the putative proteasomally spliced sequence for which the IS elutes at RT 30.5 min, no endogenous unlabeled signal was detected. For the alternative sequence annotation corresponding to an S-acetylated canonical peptide, coelution of SIL internal standard and endogenous signal was observed, serving as peptide identity confirmation. The lower panel summarizes MS2 fragment ion intensities of the different peptide species over retention time. The cutouts provide the underlying MS2 extracted ion chromatograms. B and C, quality control of spiked SIL-peptides for trace isotopic impurity in direct infusion MS1 QC runs. Labeled peptide was detected at signal intensities >1e7 (arbitrary units) with full isotopic envelopes displayed in blue (M), purple (M + 1), brown (M + 2), whereas no discernible signal was detected for the unlabeled isotopologues.

Fig. 6

Synthetic artifacts.A–C, isotopic impurities of synthetic stable isotope-labeled (SIL-) peptides. A, MS2 extracted ion chromatograms of two SIL peptides analyzed by targeted MS. While the SIL peptide with C-terminal labeled valine (V∗, ¹³C₅¹⁵N, >99%) does not show any trace signal in the unlabeled channel (left panel), the peptide with C-terminal labeled phenylalanine (F∗ ¹³C₉¹⁵N, >99%) shows unlabeled signal at about 1000-fold lower intensity than in the labeled channel. B, comparison of two batches of SIL-peptide synthesis performed with different lots of labeled phenylalanine. The first synthesis batch (upper panel) shows prevalent detection of isotopic impurities for SIL-F∗ peptides. Synthesis of a new batch of SIL peptides with a new lot of SIL-F∗ showed no isotopic impurities (lower panel). C, tracing the origin of isotopic impurity for a SIL-lysine labeled peptide. Two isotopomers of the same sequence were synthesized using either C-terminal labeling (left panel, SIL-K∗ ¹³C₆¹⁵N₂, >99%, loaded to resin at service provider) or internal labeling with the same SIL-K∗ lot directly obtained from the vendor (right panel). D, truncated by-products of KRASG12V_2-35 peptide synthesis. Shown are MS1 extracted ion chromatograms of quality control direct infusion MS for the three most abundant synthetic products. The two most prevalent by-products show incomplete coupling in an internal -VVV- sequence previously described as the site of proteasomal splicing.