Radical diffusion, not lifetime, determines the range of peroxidase-based proximity labelling.

Proximity labelling offers a powerful strategy for mapping the molecular composition in the vicinity of a protein of interest. Here, we employed APEX2- and HRP-mediated biotinylation in the Drosophila follicular epithelium to analyse the apical, lateral and basal membrane proteomes, using Cadherin99c (Cad99c), Fasciclin 3 (Fas3) and Nidogen (Ndg) as baits. We unexpectedly found that standard peroxidase-based labelling conditions produced a strong basal biotinylation signal, even when the tagged cargo localized apically or laterally. This arises from the long-range diffusion of phenoxy radicals, far exceeding the presumed ∼20 nm labelling radius. The basement membrane acts as a high-capacity sink for these radicals, owing to its abundance of electron-rich amino acids. Titrating the concentration of biotin-phenol or biotin-SS-tyramide or shortening reaction times restored spatially faithful labelling at both the plasma membrane and in intracellular compartments. Our results reveal that the distance over which proteins are labelled by peroxidase-based proximity labelling is not limited by the lifespan of the biotin radicals and can extend over many micrometres when radical production exceeds the number of reactive targets. This approach therefore requires careful optimization to avoid misleading spatial signatures.