Design, modular synthesis and screening of 58 shape-diverse 3-D fragments
Fragment screening is a tried and tested method of finding quality hit matter for the target of interest, fragment-based drug discovery (FBDD) is thus widely adopted in both academic and industrial settings where ligand-target co-crystal or Cryo-EM structures can be obtained. For example, we carry out FBDD as part of the ASAP AViDD centre (https://asapdiscovery.org/), working from fragment datasets obtained by the Diamond Light Source against viral targets such as the SARS-CoV-2 main protease (MPro), the DENV RNA-dependent RNA polymerase (RdRp) and more.
There are numerous approved drugs that were optimised from fragment hits, and even more clinical and preclinical candidates (such as ASAP-0017445!: https://dndi.org/press-releases/2025/open-science-approach-delivers-preclinical-candidate-broad-spectrum-coronavirus-antiviral/). Check out the Practical Fragments blog (https://practicalfragments.blogspot.com/2024/02/fragments-in-clinic-2024-edition.html), authored by Dan Erlanson of Frontier Medicines for news, stories and discussions relating to FBDD. While revolutionary in its own right, FBDD has been made increasingly more efficient in recent years, thanks to the adoption of new technologies and computational tools.
Typically, fragment screening campaigns involve small (hundreds- to thousands-strong) fragment libraries, these libraries must be designed carefully so as to cover the largest amount of chemical space as possible. The authors note that as FBDD has matured, there has been a desire to move towards more sp3-rich fragment libraries. Moving away from small (hetero)aromatic-focused libraries that could contain promiscuous binders that are less readily deployed in hit-to-lead efforts (“unsociable” fragments). The 3D aspect of their shape will of course always add overall diversity to the fragment set.
While the term “sp3-rich” is used above, this does not, of course, directly imply that a particular sp3-rich fragment will show significant 3D shape, compared to 2D benzene. To predict 3D shape, we use methods such as the Principal Moment of Inertia (PMI, https://pubs.acs.org/doi/abs/10.1021/ci025599w) and plane-of-best-fit (PBF, https://pubs.acs.org/doi/full/10.1021/ci300293f), which involve the generation of 3D conformations of the test molecule. These are powerful and scalable methods to quickly assess the 3D shape diversity of compound libraries.
This paper outlines a workflow to obtain “sociable” fragments, i.e. those obtained via robust and modular synthetic methods that can be rapidly elaborated once the hits are identified. An sp3 “scaffold”, such as a cyclopentane or a piperidine is proposed, followed by the attachment of (hetero)aromatic rings by established chemistry, while staying within reasonable product size and lipophilicity bounds, such as: MW<300 Da, clogP<3.
Suzuki-coupling of enol triflates followed by hydrogenation over palladium yielded cis-1,2-disubstituted sp3 scaffolds with (hetero)aromatic substituents. Interestingly, an in-silico investigation (see figure S2) into the shape diversity of the corresponding trans- isomers concluded that they were the flatter (less 3D shape) stereoisomer, perhaps due to the adoption of favourable bis-equatorial conformations. I wonder if this would still be true for bulkier substituents?
Additional chemistry included enolate arylation with (hetero)aryl halides and also enolate alkylation with (hetero)benzyl bromides, both giving branched products where the sp3 core formally becomes a spirocyclic ring. Using such readily available building blocks (boronic acids, aryl halides, benzyl halides) guarantees that once a decent binder is identified, the first analogues can be quickly prepared to investigate the Structure-Activity Relationships (SAR).
Upon field-testing their fragment libraries, the team found crystallographic hits against 2 SARS-CoV-2 targets: the MPro and also the macrodomain (Non-Structural protein 3, NSP3 “mac1”). It is always interesting to see the similarities and differences in the structures of what one team finds compared to another working on the same target, ASAP did some work on mac1, and also the Fraser and Schoichet Labs out of UCSF in 2023 (https://doi.org/10.1073/pnas.2212931120). The cis-cycloalkyl carboxylic acid placing the acidic group in the oxy-anion hole is present across series that all teams found. The library was also deployed against an oncology target: MGATV, providing hits that could be further optimised into potent inhibitors.