Abstract
The lipid A domain of the endotoxic lipopolysaccharide layer of Gram-negative bacteria is comprised of a diglucosamine backbone to which a variable number of variable length fatty acyl chains are anchored. Traditional characterization of these tails and their linkages by nuclear magnetic resonance (NMR) or mass spectrometry is time-consuming and necessitates databases of pre-existing structures for structural assignment. Here, we introduce an automated de novo approach for characterization of lipid A structures that is completely database-independent. A hierarchical decision-tree MS(n) method is used in conjunction with a hybrid activation technique, UVPDCID, to acquire characteristic fragmentation patterns of lipid A variants from a number of Gram-negative bacteria. Structural assignments are derived from integration of key features from three to five spectra and automated interpretation is achieved in minutes without the need for pre-existing information or candidate structures. The utility of this strategy is demonstrated for a mixture of lipid A structures from an enzymatically modified E. coli lipid A variant. A total of 27 lipid A structures were discovered, many of which were isomeric, showcasing the need for a rapid de novo approach to lipid A characterization.