Targeting PPIs with Peptides & Small Molecules

Pure-DEL technologies features the solid phase-based synthesis of ultra-large libraries of highly-pure and chemically diverse DNA-encoded small macrocyclic peptides with drug-like properties. Pure-DELs can be screened at once in affinity-based selections and I will present the results of Pure-DEL selections for a variety of "undruggable" targets.

Transcription factors and other intracellular PPIs remain difficult to drug because they are dynamic, poorly pocketed, and inaccessible to most biologics. This presentation describes live-cell peptide library platforms in which genetically encoded peptides are chemically constrained during selection, allowing direct enrichment of cell-tolerated, biostable antagonists that disrupt transcription factor–DNA and protein–protein interactions. The approach integrates intracellular cyclisation, functional selection, and downstream biophysical and cellular validation.

Cyclins A and B regulate cell-cycle processes through recruitment of substrates to the RxL hydrophobic patch. We describe the discovery and clinical translation of an orally bioavailable macrocyclic inhibitor targeting this intracellular protein–protein interaction site. Structure-guided medicinal chemistry optimised passive permeability, metabolic stability, and oral bioavailability within beyond-Rule-of-5 chemical space. Integrated cross-species ADME, PK, allometric scaling, and IVIVE approaches prospectively predicted human pharmacokinetic parameters and systemic exposure with good agreement to Phase 1 clinical data (NCT06577987). These findings demonstrate that translational ADME and PK approaches used for small molecules can successfully predict human pharmacokinetics for orally bioavailable beyond-Rule-of-5 macrocycles.

DNA replication stress (DRS), a cancer hallmark driving genomic instability, stems from deregulated replication and cell-cycle progression, oncogene activation, or chemotherapy, causing fork stalling and collapse. RAD52 repairs collapsed forks and helps cancer cells tolerate DRS, making it an attractive target, especially in BRCA1/2-deficient or RAD52-amplified tumours. Despite limited ligandable pockets, fragment screening identified carboxylic acids binding RAD52 single-stranded DNA groove. Fragment growing, macrocyclisation, and Cys64-directed optimisation produced covalent RAD52 inhibitors with nanomolar binding affinity, sub-micromolar cellular activity, and promising mouse pharmacokinetics. These first covalent RAD52 inhibitors provide validated tools to probe RAD52 biology and support anticancer drug discovery efforts.

Werner syndrome helicase (WRN) is a promising target in MMRd/MSI-H cancers. This presentation highlights fragment-based screening to identify multiple novel allosteric WRN pockets and the subsequent optimization of fragment hits , emphasizing the challenges of targeting this dynamic helicase. We also outline workflows that leveraged synergies between DEL and FBLD hits and guided DEL hit selection using fragment-derived structural and biophysical data.

14-3-3s are regulatory proteins for LRRK2. Upon phosphorylation-dependent binding of 14-3-3 to LRRK2, kinase activity and propensity for oligomerization and aggregation of LRRK2 are reduced. We are developing molecular glues that bind to the interface of LRRK2 and 14-3-3, stabilize the inhibitory binding of 14-3-3 to LRRK2 and counteract the above-mentioned PD-related LRRK2 activities. We present the protein crystallography-guided optimization of these  cooperative 14-3-3 molecular glues toward cell active compounds.

Targeting intrinsically disordered proteins remains challenging in drug discovery. We describe Fx168, a low-micromolar affinity peptide that binds the disordered p53 transactivation domain (TAD) and disrupts the p53–FOXO4 interaction. The NMR solution structure of the resulting stable complex reveals that Fx168 induces secondary structure in p53 TAD. Together, these findings demonstrate that p53 TAD is a tractable target and establish Fx168 as a promising starting point for further optimisation.

Oncogenic transcription factors are prized therapeutic targets historically considered undruggable due to high levels of intrinsic disorder. Nuage addresses this challenge through its pioneering drug discovery platform, which enables the identification and optimisation of selective covalent small molecules that target transient, higher-order druggable states adopted by intrinsically disordered proteins. We present the scientific basis of our innovative approach and its application to ASCL1 for the treatment of small-cell lung cancer.

INTRAMETICS is a proprietary platform designed to deliver synthetic peptidomimetics that exploits intramolecular interactions in intrinsically disordered proteins (IDPs) to block their function. Based on this technology, IDP Pharma has achieved a leading position in the field by building a strong pipeline of first-in-class drug candidates exclusively directed at IDP targets, with two products currently under clinical validation.

Macrocycles are nature’s preferred choices to generate large cell-permeable bioactive molecules. They are also increasingly considered as modalities for difficult-to-bind proteins but strategies to best exploit their potential are only beginning to emerge. We used derivatisation of linkers as an underexplored approach for macrocycle optimisation. This revealed high conformational plasticity of the macrocycles that could be controlled by minor linker modifications, resulting in several improved ligands. Our results highlight linkers as an opportunity for macrocyclic drug development, show how linker derivatisation can improve the performance of macrocycles, and emphasises the need to track macrocyclic scaffold evolution at a three-dimensional level.

Cyclic peptides can engage targets beyond the reach of small molecules, including protein-protein interactions, while retaining the potential for oral bioavailability. Yet their discovery remains challenging due to the vast chemical space. We present Boltz-MCMC, a deep learning–based method that designs cyclic peptides containing non-canonical amino acids directly from target surfaces. Screening 1,472 designs, we identified low-µM thrombin inhibitors and MDM2 binders, establishing Boltz-MCMC as a programmable route to cyclic peptide hit discovery.

Driven by advances in structural AI and deep generative models, we can now treat peptide design as a fully programmable software layer, allowing us to navigate and exploit peptide sequence space far beyond the constraints of nature. We will examine the de novo design of complex molecular interfaces, showing how AI natively navigates unconventional structural geometries, engineering functional bias, and conquering elusive protein complexes.