Proximity labeling has emerged as a prominent, reliable tool for obtaining local proteomes from a wide range of cell-types. Two major classes of labeling reagents, peroxidase based (APEX family), or biotin-ligase based (BioID family) have been developed in parallel. These two approaches are often used interchangeably, or chosen based on availability of reagents, however each may produce a biased proteome which should be considered during experimental design. We compared proximity labeling with TurboID or APEX2 in HEK293 cells across cytosol, nucleus, and membrane compartments. Both enzymes enriched compartment-specific proteomes, validated by GO terms, but showed distinct protein profiles. TurboID identified more membrane proteins, favoring identification of proteins associated with RNA processing and protein localization, while APEX2 enriched for proteins involved in metabolic pathways. Trypsin digestion highlighted biases from TurboID's lysine biotinylation, which we show can be mitigated by an endoproteinase GluC digestion during sample prep, yet these differences persist to some degree. We find that TurboID suits broader proteomic studies whereas APEX2 targets specific signaling pathways. We therefore show that strategic enzyme and protease selection is critical for optimizing proximity labeling-based proteomic studies, advancing cellular proteome mapping.
Battison et al. (2025) studied this question.