Phage display of environmental protein toxins and virulence factors reveals the prevalence, persistence, and genetics of antibody responses

The ToxScan library tiles 14,430 unique protein sequences spanning 2,936 species across 1,312 genera (436 archaeal, 7,797 bacterial, 5,743 eukaryotic), as 95,601 overlapping 56-amino-acid peptides synthesized into a programmable T7 bacteriophage display library. PhIP-Seq applied to roughly 1,000 individuals turned a previously inaccessible portion of the human antibody reactome — environmental microbial toxins and virulence factors — into a single pooled measurement.

The study mapped pervasive anti-microbial antibody responses, showed that the diversity of those responses reaches a relatively stable set point in adulthood, and demonstrated that the MHC-II locus modulates which bacterial epitopes a person targets. In a cross-sectional cohort of 598 healthy adults, 848 peptides were reactive in more than 5% of participants — encompassing at least 364 non-overlapping epitopes, of which 549 had no significant homology to sequences in the Immune Epitope Database. 99.5% of healthy adults were seropositive for at least one of the S. aureus peptidoglycan hydrolases Sle1 or Atl, and anti-flagellin antibodies were linked to both Crohn's disease and juvenile dermatomyositis.

For IBD researchers using anti-flagellin as a Crohn's biomarker, the platform reproduces the established elevated anti-flagellin response at single-epitope resolution. For autoimmune-microbiome and rare-disease researchers, the novel anti-flagellin association in juvenile dermatomyositis — a pediatric autoimmune disease whose etiology has been opaque to standard microbiome surveys — is a template for hypothesis-generating environmental-trigger discovery in any rare or understudied autoimmune cohort. This opens a layer of population immunology that nucleic-acid-based microbiome surveys cannot see, capturing low-abundance, transient, or already-cleared encounters that DNA-based surveys routinely miss.