Fragment-based drug discovery (FBDD) has been established as a key bottom-up concept in the field of medicinal chemistry [1]. By screening small, low-complexity compounds (fragments) to discover potential drug leads, FBDD has already resulted in seven approved drugs and 52 candidates under clinical investigations. The success of FBDD lies in its ability to efficiently explore chemical space using smaller libraries of fragments, compared to traditional high-throughput screening, ultimately leading to higher hit rates. However, FBDD has its own challenges, particularly in the detection and characterization of weak fragment binders. Here, typically quite resource-intensive techniques, such as X-ray screening, NMR or surface plasmon resonance (SPR) are used for primary hit detection.

To address the challenges in fragment hit detection, photoaffinity-tagged fragment libraries have emerged as a powerful tool [2]. These libraries consist of fragments equipped with a photoreactive group that, upon irradiation with light, forms a covalent bond with the target protein. Covalent labelling stabilizes the fragment-protein interaction, enabling the use of mass spectrometry (MS) for hit detection. Besides providing direct evidence of binding, this approach also allows for the identification of the binding site on the protein, offering invaluable structural information that can guide further optimization of the fragment into a drug lead.

While photoaffinity tagging offers clear advantages, it is not without limitations. One major issue is the limited photosensitivity of the commonly used diazirine tag, which can reduce the efficiency of the covalent labelling process, and thus directly impact the hit rate by missing some opportunities for hit identification. Another significant limitation of fragment libraries in general is the redundancy of the core fragments in the library. Some fragments are promiscuous binders, interacting with multiple targets, while other fragments (up to even 63% as reported by a Novartis study [3]) never appear as hits.

In response to these challenges, we have developed a pharmacophore-optimized photoaffinity (PhotoXplorer) fragment library that utilizes two innovative concepts: the SpotXplorer method for the optimization of the core fragments, and the application of an iridium-based photocatalyst to enhance labelling efficiency.

The SpotXplorer method, detailed in our 2021 study [4] (and introduced in a previous Nature Chemistry Community blog post), aims to maximize the pharmacophore coverage of fragment libraries. Our key finding was that because the number of small-molecule endogenous signalling agents (hormones, neurotransmitters, etc.) is finite, the number of protein binding sites (fine-tuned by evolution to recognize these small molecules) must be finite as well. As a result, the unique 3D binding motifs between small molecule and protein can be represented by a finite number of pharmacophores—key structural motifs that are responsible for the biological activity of a molecule. By ensuring that the fragment library covers as many unique pharmacophores as possible, the chances of identifying a true binding event are significantly increased (Figure 1).

Figure 1. SpotXplorer fragment pharmacophores are tagged by the photoactivable diazirine moiety. The resulting PhotoXplorer fragments are screened against relevant drug targets.

Here, SpotXplorer was applied to design a set of fragments that not only cover a broad range of pharmacophores but are also equipped with a diazirine tag (PhotoXplorer library). This innovative library has been successfully tested against several protein targets, demonstrating higher hit rates and better exploration of potential binding sites compared to traditional fragment libraries. These include a “conventional” target (BRD4 bromodomain) with 6000+ known ligands where we have identified fragment binders against the well-known acetyl-lysine (AcK) binding site, as well as two other, allosteric sites. Turning to more difficult targets, our screening resulted in several fragment hits against multiple orthogonal protein-protein interaction (PPI) sites of the challenging oncotarget, KRAS^G12D.

Additionally, to overcome the bottleneck of the diazirine tag’s limited photosensitivity, we have introduced an iridium-based photocatalyst [5]. This catalyst enhances the efficiency of the photoreactive process by facilitating Dexter-energy transfer, which activates the diazirine tag even at lower energy levels. This will ultimately result in increased labelling efficiency, higher overall hit rates and an improved detection sensitivity of the screening platform, making it easier to identify weak binders that might otherwise go undetected. We have successfully showcased this approach by identifying multiple fragment hits against a challenging and yet unliganded oncotarget, the N-terminal domain of the STAT5B transcription factor, where traditional methods had previously failed to identify viable hits (Figure 2).

Figure 2. Identification of STAT5B NTD hits in the presence and absence of a photocatalyst.

Overall, we have advanced the field of photoreactive fragment libraries by integrating two key concepts: the SpotXplorer approach for the optimization of pharmacophore coverage, and the use of a photocatalyst for enhanced labelling efficiency. By addressing the limitations of redundancy and photosensitivity, these innovations increase the likelihood of identifying meaningful fragment hits, even against challenging or previously “undruggable” targets.

You can find the the paper at https://doi.org/10.1038/s42004-024-01252-w

References

1. Murray, C. W. and Rees, D. C. “The rise of fragment-based drug discovery.“ Nature Chemistry. 1 (2009), 187–192. doi: 10.1038/nchem.217
2. Parker, Christopher G. et al. “Ligand and target discovery by fragment-based screening in human cells.” Cell. 168 (2017): 527. doi: 10.1016/j.cell.2016.12.029
3. Kutchukian, Peter S. et al. „Large scale meta-analysis of fragment-based screening campaigns: privileged fragments and complementary technologies.” SLAS Discovery. 20 (2015): 588–596. doi: 10.1177/1087057114565080
4. Bajusz, Dávid et al. “Exploring protein hotspots by optimized fragment pharmacophores.” Nature Communications, 12 (2021): 3201. doi: 10.1038/s41467-021-23443-y

   5.  Trowbridge, A. D. et al. “Small molecule photocatalysis enables drug target identification via energy transfer.” Proc. Natl Acad. Sci. 119 (2022): e2208077119. doi: 10.1073/pnas.2208077119