Deleterious KOs in the HLA Class I Antigen Processing and Presentation Machinery Induce Distinct Changes in the Immunopeptidome

Abstract

The human leukocyte antigen (HLA) processing and presentation machinery (APPM) is altered in various diseases and in response to drug treatments. Defects in the machinery may change presentation levels or alter the repertoire of presented peptides, globally or in an HLA allele-restricted manner, with direct implications for adaptive immunity. In this study, we investigated the immunopeptidome landscape across a panel of isogenic HAP1 cell line clones, each with a KO of a single gene encoding a key protein in the APPM, including B2M, TAP1, TAP2, TAPBP, IRF2, PDIA3, ERAP1, GANAB, SPPL3, CANX, and CALR. We applied immunopeptidomics and proteomics to assess the successful gene KOs on the protein level, understand how these proteins participate in antigen presentation, and contextualize protein expression and antigen presentation. We validated the absence of the KO proteins in the respective samples and found that knocking-out an APPM component leads to the loss of peptide subsets that are normally presented on the control wildtype cells. We assessed the immunopeptidomes qualitatively and quantitatively, considering factors like peptide diversity, peptide length distribution, and binding affinity to the endogenously expressed HLA alleles in HAP1 cells. We demonstrated prominent HLA allele-restricted alterations in several KO conditions. The absence of CALR, CANX, and TAP1 led to significant changes in HLA allele-specific presentation levels. Overall, this work represents the first systematic analysis of how the absence of individual APPM components, knocked out in a single cell line under controlled conditions, affects the immunopeptidome. This approach could facilitate the creation of predictive tools capable of prioritizing HLA-bound peptides likely to be presented when presentation defects occur, such as in cancer and viral infections.

Keywords: antigen processing and presentation machinery; human leukocyte antigen class I; immunopeptidomics.

Graphical abstract

Fig. 1

Proteomics and immunopeptidomics investigation of the antigen processing and presentation machinery.A, investigated mediators of HLA modification (GANAB), folding (CANX, CALR), assembly (CANX, CALR, PDIA3), loading (CALR, PDIA3, TAPBP, TAP1, TAP2), editing (TAPBP), and translocation (GANAB). B, sample processing for proteomics and immunopeptidomics. C, tools used for data processing and analysis. HLA, human leukocyte antigen.

Fig. 2

Library generation for high-sensitivity immunopeptidomics.A, HLA allele–wise length distribution of peptides in the spectral library (n = 23,069). B, predicted HLA allele proportions of bound peptides in the spectral library (n = 23,069). C, motifs of 9-mer peptides in the library with their associated HLA allele (n = 7027). D, number of peptide identifications in B2M KO samples. E, length distribution of peptides identified in B2M KO samples (n = 497). F, binding affinity %-rank of peptides detected in B2M KO samples. HLA, human leukocyte antigen.

Fig. 3

Immunopeptidome diversity and HLA surface expression.A, immunopeptidome diversity, measured by the number of different peptides identified per replicate. B, the median fluorescence intensity (MFI) distribution of negative control (blue) and cells (orange). C, linear regression calculating the correlation between HLA-A, -B, -C surface expression and immunopeptidome diversity (R² = 0.69). D, length distribution of immunopeptidome per sample. Asterisks indicate KO conditions with significantly different fraction of 9-mers than observed in wildtype samples. HLA, human leukocyte antigen.

Fig. 4

Immunopeptidome subsets in KOs and upregulated immunopeptidome presentation.A, overlap coefficients of individual replicates with all three wildtype replicates. Asterisks indicate KO conditions with significantly different overlap coefficients than observed within wildtype samples. B, example of intensity, that is, presentation level, comparison between the immunopeptidome of a TAP2 and a wildtype biological replicate. The dots cast to either the right side or the bottom were not identified in one of the samples. C, number (spelled out next to bar) and fraction (x-axis) of upregulated immunopeptides as derived with paired t tests (Bonferroni-corrected p value ≤0.05, fold change >2). D, mapping of upregulated (blue) or nonupregulated (red) immunopeptides to protein blocks 1 to 100 (x-axis) of respective source proteins. E, proportion of peptides that bind to a given HAP1 HLA allele (allele distribution) of upregulated (blue) or nonupregulated (red) immunopeptides. HLA, human leukocyte antigen.

Fig. 5

Immunopeptidome sampling from proteome.A, for each sample, the histograms show in how many other conditions the identified immunopeptides were found. The side plot above shows the median intensity for all peptides in a histogram bin, per sample. Gray dots indicate a peptide population below 100. B, number of proteins per sample that are presented in the immunopeptidome. C, overlap coefficients of proteins that are presented in the immunopeptidome. Overlaps are calculated between each sample and the three wildtype replicates. D, regression of relative protein intensity (MaxLFQ rank score) and the relative protein sampling density (sample density rank score). Only proteins that were identified as expressed in the proteomics data and as presented in the immunopeptidomics data were included in the analysis. E, fraction of all immunopeptides detected (y-axis) mapping to protein blocks 1 to 100 (x-axis), per replicate. Asterisks indicate Bonferroni-corrected p values <0.05 when comparing a respective KO condition to the wildtype.

Fig. 6

APPM KO consequences are HLA restricted.A, proportion of peptides that bind to a given HAP1 HLA allele (allele distribution), per condition. Asterisks indicate HLA allele fractions that deviate significantly from the allele fraction observed in the wildtype. Nonbinder fractions were not evaluated. B, median of intensity rank score of all peptides bound by a given HLA allele, per condition. Asterisks indicate HLA allele–wise peptide intensity rank scores that deviate significantly from the intensity rank scores observed in the wildtype. Nonbinder intensities were not evaluated. C, %-rank distribution of peptides detected in TAP1, TAP2, TAPBP KO clones, as well as in wildtype HAP1 cells. Side plot indicates the absolute number of peptides depicted in distribution. Vertical, dark gray line indicates the typical cutoff for immunopeptidomics experiments to assign peptides as binders and nonbinders. D, length distribution of peptides identified in TAP1, TAP2, TAPBP KO clones, as well as in wildtype HAP1 cells. Peptide populations are divided by the HLA allele peptides bind. Asterisks indicate statistically significant (p ≤ 0.05, Student's t test, Bonferroni-corrected) differences between condition and wildtype for each peptide length, per HLA allele. E, motif of 9-mers in TAP1, TAP2, TAPBP KO clones, as well as in wildtype HAP1 cells, divided by which HLA allele peptides bind. Asterisks indicate Bonferroni-corrected p values <0.05 when comparing a respective KO condition to the wildtype. F, average peptide length as well as median %-rank of peptides found in individual samples (each dot a sample), peptide populations divided by which HLA allele peptides bind. Lines connecting samples indicate groups of replicates. APPM, antigen processing and presentation machinery; HLA, human leukocyte antigen.

Fig. 7

Noncanonical (NC) immunopeptides populate the PAKC immunopeptidome.A, tile plot where each column represents a unique NC immunopeptide and tiles indicate that said immunopeptide was detected in the sample indicated on the y-axis. Dark tiles indicate a %-rank <2 for at least one of the HAP1 HLA I alleles (binder). Top-side plot indicates to what classical HLA I alleles the peptide is predicted to bind to best, the right-side plot indicates the total number of NC immunopeptides identified per sample. B, fraction of NC immunopeptides across the entire immunopeptidome per sample. Asterisks indicate Bonferroni-corrected p values <0.05 when comparing a respective KO condition to the wildtype. C, differences in allele distribution between immunopeptides derived from canonical (PC) and NC peptides. “|--|” indicates Bonferroni-corrected p values <0.05 when comparing fractions of PC and NC immunopeptides for a given allele within a condition. HLA, human leukocyte antigen; PAKC, panel of APPM KO cell.

Sup fig 5