Redefining bacterial Wzy polymerase families via three-dimensional structure-based clustering

In many bacterial species, Wzy polymerases are used in the synthesis of surface polysaccharides (e.g. capsules and O-antigens) to form the glycosidic linkage between oligosaccharide units. Currently, there are major gaps in our understanding of structural diversity and specificity-determining features of Wzy proteins needed to guide polysaccharide-targeting control methods. Here, we used Alphafold3 to model 889 Wzy structures from 11 genera (Gram-negative=10; Gram-positive=1) and show via clustering analyses that structures fall into three families, designated WzyA, WzyB and WzyC. Each were strongly associated with an anomeric linkage configuration. Novel Hidden Markov Models (HMMs) were generated with sensitivity superior to 18 pre-existing HMMs when tested against >20,000 polysaccharide genes from multiple bacteria, including 19 genera in the order Enterobacteriales. We further show that our HMMs can identify previously undetected wzy outside polysaccharide loci, highlighting important implications for polysaccharide characterisation and in silico typing schemes for all species that utilise Wzy.