88 Discussion In this study, we aimed to characterize the antibody epitope repertoire in the blood of a Dutch population and reveal which factors contribute to its variation. In particular, the factors that contribute to the generation of antibodies against microbiota and different allergens remain elusive. Here, we combined phenotypic and genetic information together with the immuneinterrogation of 2,815 common peptides from microbes, viruses, allergens and self-peptides to study this variability. Using population, family and longitudinal samples, we identified the antibody profile in the general population, assessed the stability of antibodies after 4 years and investigated the effect of genetic and environmental factors on individual immune profiles. The relation between genetics and antibody repertoire has been extensively described67-70 but has been limited to a relatively small number of antibodies until now. PhIP-Seq has recently enabled the investigation of the genetic contribution to antibody variability on a much broader scale, although it has mainly been investigated for viruses, toxins and virulence factors20,70 and not for other antigens such as allergens and gut microbiota–derived proteins. Here, we identified three genomic regions highly associated with the variability of antibody-bound peptide repertoires. As expected, we replicated the relation between HLA loci and antibody- bound peptide prevalence.20,70,90,91 Through imputation of HLA alleles, amino acids and structural variants, we also set out to uncover the specific HLA variations that allow the peptide to be displayed. Our structural simulations of the HLA alleles agree with the observed association patterns, supporting the hypothesis that the strong associations are due to HLA-display capabilities. For the first time, we report massive and specific HLA associations to more than 500 peptides at a high confidence level. This association data will be used in the future to further understand HLA–peptide interactions by modeling possible residue interactions. Our findings also support previous observations, such as the association of FUT2 and Norwalk virus peptides76 that is explained by the attachment of the viral particle to the epithelia of FUT2-secretor cells.92 We also observed association in the IGHV locus that was not previously reported in relation to antibody profiles. This association is in a complex genetic region as several genes with multiple isoforms coexist in the genome that are hard to address with microarrays.93 In addition, we lack information about the rearrangements that this gene undergoes during B-cell maturation. Nevertheless, although we cannot directly interpret the relation between variation and peptide recognition, this is a genetic region that is expected to contribute to antibody-bound peptide variability. Interestingly, our study did not identify the previously reported association of the nucleoredoxin gene (NXN) with S. pyogenes’ M3 Streptolysin O (SLO) protein,20 although we do find a weak positive association between rs4968063 and the prevalence of this antibody-bound peptide in the combined LLD and IBD cohort (p = 0.01). In the present study, we observe a lack of concordance between meta-analyzed fecal microbial composition and PhIP-Seq-based epitope repertoires, which is in line with findings from studies using the exact same library of antigens in a healthy population-based Israeli cohort and in a disease cohort consisting of patients with IBD.21,29 The top associations do not present clear relationships between specific microbial taxa and antibody-bound peptides, which could be explained in various ways. First, this apparent lack of association might point to past events, such as microbial translocation, that may have triggered long-lasting immunity that was captured Chapter 3
RkJQdWJsaXNoZXIy MjY0ODMw