80 Figure 2 | Co-occurrence network. Weighted gene network analysis identified 22 different antibody-bound peptide co-occurrent modules with at least 10 members. A. A minimum spanning tree was used to create the network of peptides belonging to one of the 22 modules. Nodes represent peptides, and node size is proportional to the peptide prevalence. Edges bind nodes with at least 0.3 Pearson correlation (between binary profiles). Colors represent different taxonomic sources of the peptide. Shades group modules and are labeled “M + module number”. B. Pie charts showing the taxonomic relative composition of each module. Pie charts are grouped in three categories. At right, category (1) indicates modules composed of different peptides from the same species. At left, category (2) indicates modules composed of structurally related peptides. At bottom is category (3) in which a mix of unrelated peptides from different organisms are seen. Category (3) may overlap with modules where the majority of peptides belong to category (1) or (2). Peptide enrichment is associated to HLA, FUT2 and IGHV genetic regions Our observation that both common environments and genetic relations within families affect the antibody-bound peptide repertoire (Figure 1E) made us wonder about the specific drivers of repertoire variability. Genetics are known to influence antibody repertoires,67-70 but the exact contribution of genetic and environmental factors to bacterial and, especially, commensal gut microbiota immune-reactivity is incompletely characterized. We estimated the proportion of antibody-bound peptide presence/absence variability accounted for by common genetic variation, i.e. its heritability (H2), using common genetic variants in 1,255 unrelated individuals. We saw an overall moderate genetic contribution to the variability of antibody-bound peptides enrichment (mean H2 = 0.1, median = 0.06, min = 0, max = 0.96) (Supplementary Table 1.1). A total of 35/2,814 antibody-bound peptides showed very high heritability (H2 ≥ 0.5), while a substantial number (597/2,814) had a relatively high heritability (H2 ≥ 0.2). Using the highly heritable antibody-bound peptides (H2 ≥ 0.5), we then computed genetic correlations in order to determine similar genetic signals across antibody-bound peptide presence. We found a correlation of 0.47 between the matrices of presence/absence and genetic correlations (Mantel test, P < 1x10-04, 9,999 permutations) (Supplementary Figure 1C). We also observed hubs of highly genetically correlated groups of peptides inwhich the genetic signatures aremore correlated than antibody-bound peptide presence itself (Supplementary Figure 1C). This indicates the existence of a common genetic architecture explaining the presence of antibody-bound peptides. Next, we set out to uncover specific loci contributing to the observed heritability. We ran a genome-wide association study (GWAS) on 1,3640,125 genotyped and imputed SNPs in 2,815 peptides. To reduce the false discovery rate (FDR) and increase the power of the analysis, we meta-analyzed the results of our LLD GWAS with those of a dataset that used the same PhIP-Seq libraries in the context of inflammatory bowel disease (IBD) (490 participants), bringing us up to a total of 1,745 individuals29 (Supplementary Table 2.2). At study-wide significance threshold (P < 5.67x10-11), we identified three genetic loci associated to 149 antibody-bound peptides. These were located in chromosome 6 (Human leukocyte antigen (HLA) locus), chromosome 14 (Immunoglobulin heavy chain variable (IGHV) region) and chromosome 19 (fucosyltransferase 2 (FUT2) gene) (Figure 3A). Chapter 3
RkJQdWJsaXNoZXIy MjY0ODMw