Non-targeted LC-MS and CE-MS for biomarker discovery in bioreactors: Influence of separation, mass spectrometry and data processing tools

Liquid separation coupled to mass spectrometry is often used for non-targeted analyses in various fields, such as metabolomics. However, the combination of non-standardized methods, various mass spectrometers (MS) and processing tools for data evaluation affect biomarker discovery potentially. Here, we present a comprehensive study of these factors based on non-targeted liquid chromatography coupled to time-of-flight (TOF) and Orbitrap MS and capillary zone electrophoresis to Orbitrap analyses of the same bioreactor samples, describing the correlation of its gas yield with changing feature signal intensity. The three datasets were processed with MZmine 2 and XCMS online and subsequential Partial Least Square Regression (PLSR) with Variable Importance in Projection (VIP) ranking for feature prioritization. The six feature tables were compared to evaluate their overlap of shared features and the influence of the processing software and MS instrument on the VIP values and fold changes. The overlaps, defined as a fraction of one feature table found in the comparative table, were from 27% to 57% for the comparison of MZmine and XCMS and from 15% to 50% between Orbitrap and TOF data sets, respectively. Considering the most relevant features only (VIP >1.5), the overlaps were increased significantly in all cases from 26% to 95%. For the same data set, both VIP values and fold changes were well correlated, however, varied significantly between TOF and Orbitrap. CE-MS showed higher total feature numbers compared to LC-MS, most likely due to its more appropriate selectivity, different sample preparation, and/or the sensitive nano-ESI interface. Since only less than 10% of MS/MS data overlapped, CE-MS provided complementary information to LC-MS. Overall, our systematic study proves the benefits of using different separation techniques and processing tools but also indicates a significant influence of mass spectrometry on comprehensive biomarker discovery. [Display omitted] •Choice of separation, MS instrument and data processing tool can strongly affect the outcome.•Many important features are most likely missed in most non-targeted studies.•PLSR enabled differentiation between reactor conditions and biomarker discovery.•CE and LC are complementary techniques for a more comprehensive metabolome coverage.