Diversity of sugar-diphospholipid-utilizing glycosyltransferase families

Peptidoglycan polymerases, enterobacterial common antigen polymerases, O-antigen ligases, and other bacterial polysaccharide polymerases (BP-Pols) are glycosyltransferases (GTs) that build bacterial surface polysaccharides. These integral membrane enzymes share the particularity of using diphospholipid-activated sugars and were previously missing in the carbohydrate-active enzymes database (CAZy; www.cazy.org). While the first three classes formed well-defined families of similar proteins, the sequences of BP-Pols were so diverse that a single family could not be built. To address this, we developed a new clustering method using a combination of a sequence similarity network and hidden Markov model comparisons. Overall, we have defined 17 new GT families including 14 of BP-Pols. We find that the reaction stereochemistry appears to be conserved in each of the defined BP-Pol families, and that the BP-Pols within the families transfer similar sugars even across Gram-negative and Gram-positive bacteria. Comparison of the new GT families reveals three clans of distantly related families, which also conserve the reaction stereochemistry.