Characterization of HLA-DR immunopeptidome of bronchoalveolar lavage cells in patients with newly diagnosed rheumatoid arthritis and healthy current-smoker controls

Abstract

Evidence suggests that self-tolerance is breached in the lung prior to the clinical onset of rheumatoid arthritis (RA) in the joints. The human leukocyte antigen DR (HLA-DR) shared epitope (SE) represents the strongest genetic risk factor for sero-positive RA. However, to our knowledge, the HLA-DR immunopeptidome of the RA lung and its link to HLA-DR genotype has not been investigated to date. The objective of this study was to optimize the methods for characterizing the HLA-DR immunopeptidome of lung immune cells and apply it to newly diagnosed RA patients versus current-smoker healthy controls, as well as to investigate the connection with the HLA-DR genotype. The HLA-DR immunopeptidome method was improved to facilitate characterization from as few as 6 million bronchoalveolar lavage (BAL) cells per subject, consisting primarily of alveolar macrophages. This method was applied to newly diagnosed RA patients naive to treatment (n=9, LURA cohort), as well as healthy current-smoker controls (n=10, COSMIC cohort). For five of the RA patients, a 6-month follow-up after initiation of the standard-of-care treatment regime was also included. After isolation and purification, peptide samples were separated by nano-flow liquid chromatography coupled to an Orbitrap mass spectrometer equipped with ion mobility device (FAIMS). Mass spectra acquired in data dependent acquisition mode were then searched against a human proteome database. Subsequently, the identified peptides were deconvoluted to their predicted binding HLA-DR allele using MHCMotifDecon based on the sequenced genotype of the individual. An optimized sample preparation and analytic method enabled the detection of over 23,000 peptides from over 3,000 source proteins with between 1,000 and 5,000 peptides identified per sample. Notably, the application of FAIMS with three compensation voltages allowed for efficient transfer of 2+, 3+, and 4+ peptide ions while removing singly charged background ions. Hierarchical clustering revealed that the immunopeptidome was more driven by the HLA-DR genotype than by RA disease or sex. However, since the HLA-DR genotype is a strong risk factor for RA, these results are convoluted. When deconvoluting the peptides to their predicted binding allele, the HLA-DRB1 alleles *01:01, *04:01, *04:04, *04:05, *04:07, and *10:01 were consistently assigned more peptides than other alleles. Except for *04:07 these alleles belong to the SE risk factor alleles, providing a potential explanation between HLA-SE and RA pathogenesis. Native peptides from known citrullinated and non-modified RA autoantigens (such as -enolase and calreticulin) were detected and validated as binders in prediction algorithms. No significant differences were found between base line and follow-up (post-treatment) samples from RA patients. Taken together, this data characterizes the HLA-DR immunopeptidome in the lung of early RA in an unprecedented manner, which together with future immunogenicity studies will help our understanding of the connection between the lung and the pathogenesis of RA. Finally, more peptides predicted to bind to SE alleles and *04:07 compared to other alleles demands further study on the relative expression of HLA-DR alleles and presentation mechanisms to understand the implications for RA.