Identification workflow of endotoxins by pyrolysis-gas chromatography-mass spectrometry based on a database and chemometrics

Single Shot Pyrolysis-Gas Chromatography-Mass Spectrometry (SS Py-GC-MS) has been previously demonstrated to directly evaluate the endotoxins of Gram-negative bacteria with minimal sample preparation. The analytical method was quick and sensitive, but data processing was time-consuming, especially when comparing two sets of data that were quite similar. In this research article, the data handling workflow was optimized, mainly using two approaches for comparing critical compound quantities to distinguish endotoxins from different sources. Firstly, a library of endotoxin biomarkers for bacteria profiling was constructed and the identification workflow was automated. Bacterial endotoxins could be fingerprinted down to the subspecies level. The approach was then tested on environmental, food, and water samples subjected to minimal sample preparation, and the results were highly promising. In the second approach, Chemometrics was utilized to create a similar but more encompassing database with predictive capability. The results supported the library technique, indicating that distinct bacteria species and subspecies produce endotoxins with varied amounts of each critical compound that could be identified using Py-GC-MS. A Hotelling T-2 range plot was successful in differentiating Gram-negative bacterial endotoxins from Fungal mycotoxins and Cyanobacterial endotoxins. This illustrates the potential for the second approach for endotoxins which might not have been discovered. Additionally, the correlation between the molecular structure of lipopolysaccharides (LPS) and the endotoxin's virulence was observed for the first time, using a quantitative model of supervised Multivariate Analysis (MVA).