μMap Photoproximity Labeling Enables Small Molecule Binding Site Mapping

Abstract

The characterization of ligand binding modes is a crucial step in the drug discovery process and is especially important in campaigns arising from phenotypic screening, where the protein target and binding mode are unknown at the outset. Elucidation of target binding regions is typically achieved by X-ray crystallography or photoaffinity labeling (PAL) approaches; yet, these methods present significant challenges. X-ray crystallography is a mainstay technique that has revolutionized drug discovery, but in many cases structural characterization is challenging or impossible. PAL has also enabled binding site mapping with peptide- and amino-acid-level resolution; however, the stoichiometric activation mode can lead to poor signal and coverage of the resident binding pocket. Additionally, each PAL probe can have its own fragmentation pattern, complicating the analysis by mass spectrometry. Here, we establish a robust and general photocatalytic approach toward the mapping of protein binding sites, which we define as identification of residues proximal to the ligand binding pocket. By utilizing a catalytic mode of activation, we obtain sets of labeled amino acids in the proximity of the target protein binding site. We use this methodology to map, in vitro, the binding sites of six protein targets, including several kinases and molecular glue targets, and furthermore to investigate the binding site of the STAT3 inhibitor MM-206, a ligand with no known crystal structure. Finally, we demonstrate the successful mapping of drug binding sites in live cells. These results establish μMap as a powerful method for the generation of amino-acid- and peptide-level target engagement data.

Figure 1.

(a) Identification of drug candidate binding sites is a significant challenge in drug development. (b) This work: photocatalytic proximity labeling (μMap) as a platform for the mapping of drug-target binding sites.

Figure 2:

(a) in vitro targeted labeling of bovine carbonic anhydrase (CA) using benzenesulfonamide iridium (1). (b) Western blot analysis of in vitro labeling demonstrates successful targeted labeling of CA over bovine serum albumin (BSA). (c) Workflow for label-free mass spec binding site identification. (d) Illustrative MS2 spectra for peptide YGDFGTAAQQPDGLAVVGVFLK showing modification of glutamine 135. (e) Residues detected for the binding site mapping of bovine CA with sulfonamide photocatalyst (1) using a modified peptide cutoff of >2 Log2FC intensity vs. free photocatalyst control. PDB: 6SKS

Figure 3.

(a) Residues detected for the binding site mapping of recombinant BRD4 using (+)-JQ-1 photocatalyst (3). Applied modified peptide cutoff at >4 Log2FC intensity vs. inactive enantiomer control. PDB: 3MFX. (b) Binding site mapping of recombinant BTK using dasatinib photocatalyst (4). Applied modified peptide cutoff at >3.85 Log2FC intensity vs. off compete control. PDB: 6AUB. Ligand shown is CGI2815. Distances from ligand binding site were calculated from PDB: 3K54 (c) Binding site mapping of recombinant CDK2 using AT7519 photocatalyst (5). Applied modified peptide cutoff at >2.5Log2FC intensity vs. off-compete control. PDB: 2VU3 (d) Residues detected for the binding site mapping of recombinant CRBN using lenalidomide photocatalyst (6) using a modified peptide cutoff of >3 Log2FC intensity vs. off compete control. PDB: 6BN7. (e) Binding site mapping of the FKBP12-rapamycin-mTOR ternary complex with rapamycin photocatalyst (7) using a modified peptide cutoff of >6 Log2FC intensity vs. off compete control for mTor and >3.5 Log2FC intensity vs. off compete control for FKBP12. PDB:8ERA

Figure 4.

(a) Recent work suggests that MM-206 (9) may be an allosteric, instead of orthosteric, STAT3 inhibitor. We investigated this via the MM-206-catalyst conjugate (10) (b) Residues detected for the binding site mapping of recombinant STAT3 (130-688) with MM-206 photocatalyst (10) using a modified peptide cutoff across two replicates with >1.5 Log2FC intensity vs. off compete control. PDB: 6NUQ. (c) Workflow for live cell binding site mapping in Jurkat cells. (d) Volcano plot showing enriched proteins resulting from μMap, showing exclusive selectivity of BRD proteins, including BRD4. (e) Modified residues detected in live cell binding studies of BRD4; all of which are within 10 angstroms of the small molecule.