Metabolic proteomics has been widely used to characterize dynamic protein networks in many areas of biomedicine, including in the arena of tissue aging and rejuvenation. Bioorthogonal noncanonical amino acid tagging (BONCAT) is based on mutant methionine-tRNA synthases (MetRS) that incorporates metabolic tags, for example, azidonorleucine [ANL], into newly synthesized proteins. BONCAT revolutionizes metabolic proteomics, because mutant MetRS transgene allows one to identify cell type-specific proteomes in mixed biological environments. This is not possible with other methods, such as stable isotope labeling with amino acids in cell culture, isobaric tags for relative and absolute quantitation and tandem mass tags. At the same time, an inherent weakness of BONCAT is that after click chemistry-based enrichment, all identified proteins are assumed to have been metabolically tagged, but there is no confirmation in mass spectrometry data that only tagged proteins are detected. As we show here, such assumption is incorrect and accurate negative controls uncover a surprisingly high degree of false positives in BONCAT proteomics. We show not only how to reveal the false discovery and thus improve the accuracy of the analyses and conclusions but also approaches for avoiding it through minimizing nonspecific detection of biotin, biotin-independent direct detection of metabolic tags, and improvement of signal to noise ratio through machine learning algorithms.
Keywords: antibody array; bioorthogonal; biotinylated proteins; false positive; machine learning; mass spectrometry; metabolic; proteomics.