Lectins with a β-propeller fold bind glycans on the cell surface through multivalent binding sites and appropriate directionality. These proteins are formed by repeats of short domains, raising questions about evolutionary duplication. However, these repeats are difficult to detect in translated genomes and seldom correctly annotated in sequence databases. To address these issues, a database has been developped and is now available on internet at https://www.unilectin.eu/propeller.
We defined the blade signature of the five types of β-propellers using 3D-structural data. With these templates, we predicted 3887 β-propeller lectins in 1889 species and organised this new information in a searchable online database. The data reveals a widespread distribution of β-propeller lectins across species. Prediction also emphasises multiple architectures and led to uncover a novel β-propeller assembly scenario. This was confirmed by producing and characterizing a predicted protein coded in the genome of Kordia zhangzhouensis. The crystal structure shows a new intermediate in the evolution of β-propeller assembly and demonstrates the power of our tools.