Bioinformatic process for the identification and characterization of bacterial repeat-in-toxin adhesins

Bacteria bind to host surfaces with ligand-binding domains (LBDs) at the distal tips of fibrillar adhesins. Blocking these initial binding interactions could potentially stop colonization that would otherwise lead to biofilm formation and infections. To develop such preventative strategies, it will be essential to know the full range of adhesins and their LBDs that are present in the various strains of a particular bacterial pathogen. The bioinformatic process we have developed to find RTX adhesins in Gram-negative bacteria starts with the collation of all predicted proteins from long-read genome sequences for the target species. Proteins are clustered by similarity of their C-terminal regions then scanned through InterPro to detect domains characteristic of RTX adhesins. Candidates are then modeled using AlphaFold3 and classified as RTX adhesins if they contain both an N-terminal retention domain and a C-terminal Type I Secretion Signal. Many species have several distinct RTX adhesins, most of which have LBDs, including putative carbohydrate-binding modules and von Willebrand Factor A-like domains, projecting from split domains. Similar adhesin structures can be shared by several species, possibly by descent or by acquisition though DNA uptake. Some species have an RTX adhesin of uncertain function that has no obvious LBDs.