Mapping protein binding sites by photoreactive fragment pharmacophores

Fragment-based drug discovery (FBDD) has revolutionized drug development over the past three decades, leading to the approval of several successful drugs. The strategy leverages the advantages of screening smaller, less complex fragment libraries (≤16 heavy atoms) which offer better chemical space coverage and higher hit rates compared to larger compound libraries (≤36 heavy atoms). However, the weak affinities of fragments necessitate highly sensitive biophysical assays for detection. Traditional methods such as thermal shift assays, NMR, and surface plasmon resonance are often used, but these can be limited in sensitivity and/or require specialized equipment. Photoaffinity labeling, which involves a pharmacophore and a photoreactive group, has emerged as a powerful approach to overcome these limitations. Fully functionalized fragments (FFFs) incorporating photoaffinity tags have been successfully employed, but often suffer from uneven pharmacophore coverage within available libraries. This research aims to address these challenges by developing a new fragment screening platform that combines the advantages of pharmacophore optimization and photoaffinity labeling.

Fragment-based drug discovery (FBDD) has significantly impacted drug discovery in recent years. The approach emphasizes the screening of small, less complex molecules (fragments) that provide better chemical space coverage than larger molecules, leading to higher hit rates. However, the weak affinity of fragments requires sensitive detection methods. Biochemical assays, while simpler, lack sensitivity and structural information. Biophysical methods like thermal shift assays, NMR, and surface plasmon resonance offer higher sensitivity but require more resources. Photoaffinity labeling, combining a pharmacophore with a photoreactive group, has become increasingly popular for detecting fragment hits. While fully functionalized fragments (FFFs) have shown success, existing FFF libraries often exhibit uneven distribution of pharmacophores. The SpotXplorer methodology has previously been developed for fragment library design to maximize pharmacophore coverage. This study combines SpotXplorer's design principles with photoaffinity labeling to enhance the efficiency of fragment screening.

This study developed and validated a pharmacophore-optimized photoaffinity library (PhP) containing 100 diazirine-tagged fragments. The library was designed using the SpotXplorer technology to ensure optimal coverage of experimentally validated fragment-protein binding pharmacophores. The Enamine primary amine collection was utilized as the starting point for the fragment library, with primary amines serving as the attachment points for the diazirine photoaffinity tag. A parallel synthesis approach was used to couple the amines with the photoaffinity tag. The resulting library was screened against three benchmark proteins (carbonic anhydrase II, myoglobin, and lysozyme) and three therapeutically relevant oncology targets (BRD4-BD1, KRasG12D, and STAT5B-NTD). A plate-based format was employed, with each fragment incubated with the target protein, followed by irradiation to enable crosslinking. The labeled protein-fragment complexes were then detected by mass spectrometry (MS). The binding sites of confirmed hits were further identified through techniques like enzymatic digestion, X-ray crystallography, HSQC NMR, and molecular modeling. For STAT5B-NTD, the screening sensitivity was enhanced by conjugating an iridium-based photocatalyst to the protein, boosting the photoaffinity labeling efficiency. Microscale thermophoresis (MST) was used to validate fragment binding to STAT5B-NTD. The anti-proliferative activity of selected fragment hits was assessed using MTT and CellTiter-Blue assays against relevant cancer cell lines.

The PhP library effectively mapped binding sites for all six protein targets, showcasing improved performance over traditional methods. The library yielded multiple fragment hits against even challenging targets, such as KRasG12D and STAT5B-NTD. For BRD4-BD1, the hits successfully bound to both the primary and secondary binding sites. For KRasG12D, three confirmed hits identified five previously unexplored allosteric pockets or protein-protein interaction surfaces. One of these hits, PhP072, demonstrated antiproliferative activity against KRas-dependent cancer cells. For the unliganded STAT5B-NTD, 24 hits were identified, with two validated hits exhibiting micromolar affinity and antiproliferative activity against leukemia cell lines. The conjugation of an iridium-based photocatalyst significantly enhanced the screening sensitivity, especially on a standard UHPLC-MS system, doubling the number of hits against STAT5B-NTD. The overall hit rates for the PhP library were significantly higher compared to previous photoaffinity-based fragment screening platforms, particularly for challenging targets.

The results demonstrate the efficacy of the PhP library in identifying fragment hits and mapping binding sites, even for notoriously challenging targets. The higher hit rates and improved binding site exploration compared to existing methods highlight the advantages of combining pharmacophore optimization with photoaffinity labeling. The identification of novel binding sites for KRasG12D and the first reported fragment hits for STAT5B-NTD offer promising starting points for drug development. The successful adaptation of the methodology to standard LC-MS platforms expands the accessibility of this powerful screening technique. The enhanced sensitivity achieved by photocatalyst conjugation opens new avenues for screening low-affinity fragments. While some fragment hits showed promiscuity, the demonstrated on-target affinities and cellular activities suggest their potential as viable starting points for hit-to-lead optimization.

This study successfully introduced a novel fragment screening platform using pharmacophore-optimized photoaffinity fragments (PhPs). The PhP approach demonstrated higher hit rates and improved exploration of protein binding sites compared to existing methods, particularly for challenging or unliganded targets. Methodological improvements, including the use of readily available LC-MS systems and a bioconjugated photocatalyst to enhance sensitivity, broaden the accessibility of this technology. The identified fragment hits against BRD4-BD1, KRasG12D, and STAT5B-NTD offer valuable starting points for drug discovery efforts. Future research could focus on expanding the PhP library, exploring different photoreactive groups, and investigating the potential of this technology for other challenging targets.

The study primarily focused on in vitro assays and further in vivo studies are necessary to confirm the efficacy and safety of the identified fragment hits. Some fragment hits showed promiscuity, binding to multiple proteins. While the on-target effects were demonstrated, the contribution of off-target effects cannot be fully excluded at this stage. The photocatalyst conjugation approach improved sensitivity but also increased the occurrence of some side reactions. Further optimization of this strategy might be needed for specific applications. The generalizability of the enhanced screening sensitivity achieved using the photocatalyst remains to be tested across a broader range of target proteins.