Structuur-gebaseerd ontwerp van kleine moleculen en eiwit liganden tegen het LEDGF PWWP domain

Lens epithelium-derived growth factor p75 (LEDGF/p75) is a chromatin reader that binds to nucleosomes carrying di-and trimethylation marks on Lys36 of histone H3 (H3K36me2/3) through its PWWP domain. As I explain in the Introduction (Chapter 1), this interaction plays critical roles in transcriptional regulation, with relevance for disease including HIV аs well as for oncogenesis, particularly in MLL-rearranged leukemia. Despite the therapeutic relevance of this domain, previously no active molecules were known that could interrogate its function or modulate its activity. Addressing this unmet need, the thesis sets out to discover and develop ligands for the LEDGF PWWP domain to inhibit its interaction with H3K36me2/3-modified nucleosomes, using a structure-based, multi-platform strategy that integrates small-molecule discovery and protein-based modalities (Chapter 2: Research objectives). In Chapter 3, the search for a suitable crystallographic system for fragment-based screening (FBS-X) is described. Previous attempts to crystallize the LEDGF PWWP domain directly yielded no structures amenable to soaking-based screening. As a surrogate, the wild-type PWWP domain of hepatoma-derived growth factor-related protein 2 (HRP-2), a close paralog of LEDGF, was explored. However, these crystals only formed in high ionic strength conditions incompatible with fragment soaking and displayed rapid degradation. The instability was traced to oxidation of a surface-exposed cysteine (Cys64) located near a major crystal lattice contact. To overcome this limitation, the Cys64Ser mutation was introduced, markedly enhancing crystal stability and enabling growth in low-salt, PEG-based conditions compatible with fragment soaking. The resulting mutant crystals were well-diffracting and long-term stable, providing the essential crystallographic platform for FBS-X. Chapter 4 presents the FBS-X campaign carried out using the crystal of the engineered HRP-2 PWWP mutant. This screen identified 68 fragments binding within the H3K36me2/3 recognition pocket. Selected fragment series were elaborated through iterative cycles of structure-based design, medicinal chemistry, and in vitro testing. This process yielded compounds with double-digit micromolar affinities and high ligand efficiencies. Nevertheless, further improvement of the binding affinity was hampered by the small size of the binding pocket. Extending interactions beyond the canonical pocket was identified as a prerequisite for achieving higher affinity, yet no fragments from this campaign engaged such peripheral surfaces. Identifying binders to such peripheral sites to guide small-molecule optimization was therefore the aim of the alternative strategies explored in the subsequent chapters of the thesis. In Chapter 5, an AI-guided de novo protein design approach was applied to explore additional binding surfaces on the LEDGF PWWP domain. Using the RFdiffusion platform and extensive in silico filtering, four synthetic binders (JUBO1-JUBO4) were designed to contact both the H3K36me2/3 pocket and adjacent underexplored regions of the LEDGF PWWP domain. These proteins were expressed, purified, and characterized using a range of biophysical assays. JUBO2 displayed low-micromolar affinity to the PWWP domain in orthogonal measurements, validating the computational design strategy. The co-crystal structure of JUBO4 bound to the LEDGF PWWP domain revealed a previously inaccessible binding interface extending towards peripheral residues. This structural insight enabled expansion of the pharmacophore model to incorporate interactions within both the H3K36me2/3 pocket and peripheral regions. An initial attempt to exploit one such new contact using a small-molecule scaffold confirmed binding at the predicted site, although poor solubility hindered further optimization. Chapter 6 describes a further approach to map the LEDGF PWWP binding landscape using immune-derived nanobodies (Nbs). A llama and an alpaca were immunized with the target domain, and phage display yielded nine sequence-distinct Nbs. Seven of those Nbs interacted with the LEDGF PWWP domain with nanomolar affinity, and most also bound to the HRP-2 PWWP domain. While no co-crystal structures of Nb-PWWP domain complexes were obtained, likely due to conformational heterogeneity or Nb oligomerization, apo Nb structures confirmed canonical immunoglobulin folds. In preliminary in vitro assays, NbH10 appeared to reduce the LEDGF/p75-H3K36me3 nucleosome interaction, suggesting a possible inhibitory effect. However, additional studies are required to validate and fully characterize its potential as an inhibitor and as a tool for probing LEDGF/p75 function. Chapter 7 (Discussion and future horizons) integrates the findings from all ligand discovery strategies employed. The successful engineering of a robust crystallographic platform enabled the first systematic fragment-based exploration of the LEDGF/HRP-2 PWWP domain. The structural constraints revealed by this work—particularly the limited depth and small volume of the H3K36me2/3 pocket—explain the difficulty in achieving high-affinity small molecules. New routes using AI-designed protein binders (JUBOs) and immune-derived Nbs provided complementary approaches to map peripheral binding surfaces and expand the pharmacophore model. Together, these advances provide a foundation for future LEDGF PWWP inhibitor development, including proteolysis-targeting chimera (PROTAC)-based degradation strategies and improved Nb tools for mechanistic studies. To conclude, this thesis advances the structural understanding of the LEDGF PWWP domain, explores its ligandability, and delivers molecular tools and binding models that open new opportunities for targeted inhibition and mechanistic exploration.