82 and DQ2.5. For example, the hydrogen bonds occurring between the Tyrosine 60 (Tyr60) and Tryptophan 61 (Trp61) present in the beta chain of both DR3 and DQ2.5 interact with Glutamic acid (Glu) and Threonine (Thr) in the peptide core. By contrast, although we could model the peptide binding into the negative control DR14, the majority of the peptide’s amino acids are located outside of the binding site and in the opposite direction compared to DR3 and DQ2 (Figure 3B). In addition, we selected two other highly associated HLA–peptide complexes to explore in detail: (1) the combination of the peptide Lactococcus phage (YP_009222335.1 hypothetical protein LfeInf_097) with the DR15 haplotype (DRB1*0301), which showed the strongest studywide association (OR = 13.3, p = 1.44x10-47) (Supplementary Figure 2A), and (2) a combination of a peptide from the Human mastadenovirus minor core protein with the associated DR4-DQ8 haplotype (encoded by the DRB1*0401 and DQA1*0301-DQB1*0302 genes) (DRB1*0401, OR = 5.69, p = 4.45x10-15; DQA1*0301, OR = 2.55, p = 2.12x10-18; DQB1*0302, OR = 3.14, p = 4.17x10-20) (Supplementary Figure 2B). We observed a positive identification of the peptide core matching known deconvolution motifs, as well as a favorable binding prediction for the Lactococcus phage peptide to DR15 and for the Human mastadenovirus peptide to DR4-DQ8 haplotypes. Similarly, the bindingmodemodeling of the peptide cores to the HLA-II complexes resulted in energetically favorable binding energy calculations and Kd in the nanomolar range (Lactococcus phage–DR15, 1.6x10-7 M; Humanmastadenovirus–DR4/DQ8, 1.2x10-7M and 1.3x10-7M, respectively). These results suggest that the identified HLA–peptide associations point to biologically relevant processes in which a specific HLA complex can preferentially bind and display the specific peptide sequence. A second study-wide significant signal in our GWAS pointed to the IGHV region in chromosome 14 that encodes the immunoglobulin heavy chain variable domain. Here, we found 16 associated peptides in 11 leading loci within the region. The majority of SNPs (11/16) were located in non-coding regions around the IGHV gene, whereas Ovis aries casein protein (representing the primary sheep’s milk allergy food allergen) was associated with a missense variant that changes Glycine, a non-polar amino acid, for Arginine, a positively charged amino acid. Next to the Ovis aries casein peptide, the top peptides associated to this region are bacteriarelated (Bacteroides uniformis, Blautia producta and Lactobacillus plantarum) or viral (Influenza A, Lactobacillus phage and Norwalk virus). The strongest association was observed in Lactobacillus plantarum (aggregation promoting factor) and Lactobacillus phage (endolysin). We found a third study-wide significant signal in the FUT2 gene in chromosome 19. This gene status controls the secretion or non-secretion (homozygous for loss of function) of the H-antigen, an oligosaccharide. Thus, we subsequently ran the analysis in a dominant/recessive model to increase power and detected three study-wide significant peptides, all of which originally belonged to Norwalk virus polyproteins and were negatively associated with the same leading variant, rs2251034 (A>G,3’ UTR). This variant is in high linkage with an early-stop variant in FUT2 that is known to stop the secretion of the H-antigen, rs601338 (A>G, R2 = 0.85, 1000G, CEU population). FUT2 secretor status has been previously associated with multiple phenotypes, including infection susceptibility,72 gut microbiome,73,74 human milk oligosaccharides75 and cardiovascular traits.36 Our finding supports the previously reported association between Norwalk virus susceptibility and FUT2 secretor status,76 since this virus requires the H type-1 oligosaccharide ligand for successful attachment in the cell surface. Chapter 3
RkJQdWJsaXNoZXIy MjY0ODMw