Proximity Effects Leveraged in Ligand-Directed Chemical Labeling of Natural Proteins under Live Conditions

Covalent chemical labeling of proteins is central to chemical biology, offering functional modifications beyond imaging probes. While genetically engineered systems using self-labeling tags (e.g., HaloTag, SNAP-Tag) or genetic code expansion have enabled selective labeling in live cells, endogenous (naturally occurring) proteins remain difficult targets because genetic manipulation conferring selectivity cannot be conducted. To address this, ligand-directed chemistry (LDchem) was developed in which labeling reagents combine a ligand, a probe, and a cleavable electrophile. Ligand binding forms a transient complex that guides selective covalent modification, while preserving native protein function. A key feature of LDchem is the proximity effect, where reactivity is spatially restricted to residues adjacent to the ligand-binding site. Surprisingly, the proximity in LDchem often induces accelerated reactions and unexpected modifications such as ether or ester bond formation under mild physiological (aqueous) conditions. Such proximity effects include both local concentration (ligand affinity for concentrating reagents) and orientation (linker length and rigidity for impacting residue selectivity). Studies comparing electrophiles demonstrated reaction rates spanning 4 orders of magnitude, in some cases rivaling those of self-labeling enzyme tags or even fast click reactions. Notably, weaker intrinsic electrophiles can still yield efficient labeling when favorable orientation effects prolong the residential time of reactive groups near target residues. Proximity has also been exploited for protein functionalization. For example, tethering ¹⁹F-NMR probes to carbonic anhydrase converted in-cell to ¹⁹F-NMR biosensors that reported their ligand binding under native conditions. Similarly, attaching fluorescent dyes to receptors allowed evaluation of ligand binding kinetics by live-cell imaging. Recent advances extend LDchem into in vivo systems, notably, the live mouse brain. Injection of LDchem reagents into cerebrospinal fluid enabled selective receptor labeling across whole-brain scales visualized with tissue clearing technology. Furthermore, LDchem modified endogenous receptors with a photosensitizer to develop PhoxID, a photoproximity labeling method. In live brains, this enabled proteomic mapping of receptor interactomes with 1-10 min temporal resolution, unveiling developmental shifts in α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)-associated proximal proteins. LDchem has also been adapted for sensing the vicinity of a target receptor. For instance, AMPARs were converted into in situ fluorescent biosensors to detect matrix metalloproteinase MMP-9 activity in the receptor proximity within ∼10 nm, revealing region-specific and synapse-localized enzymatic activity in the brain. In conclusion, LDchem highlights how proximity effects can be harnessed to achieve selective, accelerated, and functional protein labeling under native biological conditions. Although precise prediction of proximity effects remains challenging due to complex local environments of a target protein, advances in artificial-intelligence-guided computational modeling promise rational design of labeling reagents. Such chemical tools are expected to transform in vivo chemical biology, enabling precise interrogation of protein microenvironments and the networks in the whole living systems.