Abstract
Proximity labeling approaches have been widely utilized to define protein interactomes. Due to the inherent promiscuity of proximity labeling using TurboID-based approaches, identification and adoption of appropriate labeling controls is a pivotal step to mitigate background interference and enhance interactome assignment accuracy. Here, we evaluate the effectiveness of both expression controls and data normalization strategies in generating high-confidence interactome maps. We demonstrate that the extent of control of TurboID protein expression is strongly correlated with overall signal intensity and the number of identified proteins from streptavidin-enrichments. Discordant expression levels between the bait and control samples result in high-frequency false-negative and false-positive identifications. Data normalization strategies help correct these expression differences but also introduce data distortion for proteins with high or low endogenous expression. Using the ubiquitin ligases RNF10 and HUWE1 as bait proteins, we demonstrate that matching TurboID expression between control and bait proteins allows for similar sampling of non-specific interactions. Using a matched expression strategy results in significantly reduced background interference and increases the accuracy of interactome assignments. These results document the need to alter proximity-labeling experimental workflows to include the generation of matched expression controls to enhance proximity labeling proteomics interactome mapping robustness and reproducibility.