BACKGROUND: Acinetobacter baumannii is a common nosocomial pathogen and strain-typing methods play an important role in hospital outbreak investigations and epidemiologic surveillance. We describe a method for identifying strain-specific peptide markers based on LC-MS/MS profiling of digested peptides. This method classified a test set of A. baumannii isolates collected from a hospital outbreak with discriminatory performance exceeding that of MALDI-TOF mass spectrometry. METHODS: Following the construction of a species "pan-peptidome" by in silico translation and digestion of whole genome sequences, a hypothetical set of genome-specific peptides for an isolate was constructed from the disjoint set of the pan-peptidome and the isolate's calculated peptidome. The genome-specific peptidome guided selection of highly expressed genome-specific peptides from LC-MS/MS experimental profiles as potential peptide markers. The species specificity of each experimentally identified genome-specific peptide was confirmed through a Unipept lowest common ancestor analysis. RESULTS: Fifteen A. baumannii isolates were analyzed to derive a set of genome- and species-specific peptides that could be used as peptide markers. Identified peptides were cross-checked with protein BLAST against a set of 22 A. baumannii whole genome sequences. A subset of these peptide markers was confirmed to be present in the actual peptide profiles generated by multiple reaction monitoring and targeted LC-MS/MS. The experimentally identified peptides separated these isolates into 6 strains that agreed with multilocus sequence typing analysis performed on the same isolates. CONCLUSIONS: This approach may be generalizable to other bacterial species, and the peptides may be useful for rapid MS strain tracking of isolates with broad application to infectious disease diagnosis.
