Updated project metadata. Proximity-dependent labelling methods are based on promiscuous labeling enzymes that produce reactive molecules that covalently bind neighbor proteins. Labeled proteins can be then purified and identified using affinity-purification coupled to mass spectrometry methods23. Proximity-dependent biotin identification (BioID)24 uses a promiscuously active Escherichia coli biotin ligase (BirA*) generated by a point mutation (R118G) to biotinylate lysines in nearby proteins within an estimated range of 10 nm25. By fusing BirA* to specific proteins, BioID efficiently identifies interactors at physiological levels in living cells26. It has been extensively used in the Ub field, for instance, to identify substrates of E3 ligases27,28. Recently, a more efficient version of BioID, termed TurboID, has been developed29, being this more suitable for transient protein-protein interaction (PPI) detection. Several studies have developed split-versions and applied “protein fragment complementation” to BioID and TurboID, where proximal biotinylation is dependent on the proximity of the fusion partners, opening new opportunities for spatial and temporal identification of complex-dependent interactomes30,31. To study how SUMOylation and SUMO-SIM interactions can lead to other roles and fates for particular substrates poses particular challenges. SUMOylation occurs transiently and often in a small percentage of a given substrate. Modified proteins can be readily deSUMOylated and SUMO can be recycled and passed to other substrates. SUMO-SIM interactions are also difficult to analyze due to their weak affinity (Kd 1-100 μM). To overcome those technical issues, we developed SUMO-ID, a new strategy based on Split-TurboID to identify interactors of specific substrates dependent on SUMO conjugation or interaction. Using PML as a model, we demonstrate that SUMO-ID can enrich for factors that depend on PML-SUMO interaction. Importantly, the identified proteins are represented among proximal interactors of PML identified using full-length TurboID. We also applied SUMO-ID to a less-characterized SUMO substrate, Spalt Like Transcription Factor 1 (SALL1), and identified both known and novel interactors that depend on intact SUMOylation sites in SALL1. SUMO-ID is thus a powerful tool to study transient and dynamic SUMO-dependent interaction events. The developed methodology is generic and therefore widely applicable in the Ub and UbL field to identify readers of these modifications for individual target proteins to improve our insight in non-covalent signal transduction by Ub and UbL.