133 of the adaptive immune system.55 Furthermore, the detection limits of both techniques may show discrepancies: some microbes detected by fecal metagenomics may not necessarily elicit strong antibody responses, whereas highly immunogenic microbes may not all be detected by metagenomics. Although we restricted the microbiome concordance analysis to patients who were profiled within one year of sampling (in both directions), there could still be a certain amount of temporal heterogeneity that overshadows potential associations between the two data entities. The fact that we did not observe strong concordance with the gut microbiome corroborates the findings from a previous study that also found few associations, mostly those involving peptides occurring in < 5% of individuals.12 Furthermore, it should be noted that fecal sampling was employed for generation of metagenomics data, whereas locally present (mucosal) microbial communities are different and may have stronger immune reactivity. Strengths of the present study included the extensive characterization of the study cohort with multiple layers of information, such as the availability of detailed patient and fecal metagenomics data, which enable detailed assessment of specific antibody signatures. Another highlight pertains to the successful application of the PhIP-Seq technology in the context of IBD, providing a high-resolution, high-throughput platform to study human antibody responses and enabling the characterization of both current antibody responses and responses provoked upon past antigen exposure. This temporal stability allows us to capture antibody responses that are saved in our immunological memory, adding a unique resource of biological information to existing technologies such as fluorescence-activated cell sorting (FACS), IgA-Seq or BugFACS methods.22,56,57 Previously, the unique longitudinal stability of the PhIP-Seq technology was shown to surpass that of metagenomics shotgun sequencing.12 Furthermore, PhIP-Seq accurately informs us about the exact nature of the bound antigens, which remains largely unknown using conventional IgA-Seq or BugFACS techniques as they usually only provide information about the presence of certain antibody-coated microbes, while capturing only a fraction of the full antibody repertoire. Some limitations of this study also warrant recognition. For example, the majority of patients with IBD in our cohort were in disease remission, which limited our ability to study differences in antibody responses with varying degrees of disease activity or in new-onset patients. In addition, our assessment of disease activity was limited to clinical and serological parameters, as endoscopic data or fecal calprotectin levels were not sufficiently available at time of sampling. Finally, some technical aspects of PhIP-Seq limit further characterization of antibody responses, including the length restriction of included peptides (up to 64 amino acids for the present library) and the absence of peptides from conformational epitopes (while all linear epitopes were represented). Second, only protein antigens are incorporated in the library, whereas other immunogenic molecules like lipids and glycans are not represented. However, non-protein antigens usually generate T-cell-independent antibody responses, which are characterized by production of antibodies with lower affinity but higher avidity.58 The current characterization of antibody-bound peptides may therefore represent only strong (high-affinity) and specific (low avidity) immunological responses, which may be expected to be more relevant biomarkers than low-affinity, high-avidity antibody responses. In the same line of thought, detected associations with single peptides should be interpreted cautiously, whereas associations covering multiple peptides for the same protein may be more reliable. In addition, cautious interpretation is The antibody epitope repertoire in IBD
RkJQdWJsaXNoZXIy MjY0ODMw