PhD thesis

Drugs work only when they adopt the right 3D shape at the right moment. In my PhD thesis, we ask a simple but powerful question: Which shapes do promising molecules prefer in realistic conditions, and how do those shapes relate to their biological effects? To answer it, the work studies two families of compounds with both rigid and flexible parts: cannabinoid-inspired molecules (related to Δ8-THC and CBD) and fullerene derivatives.

For the cannabinoid series, computer simulations were used to map out the shapes the molecules naturally adopt in solution and inside their target receptors (CB1 and CB2, key proteins of the endocannabinoid system). Data-driven 3D maps (3D-QSAR) then revealed which regions around the molecule benefit most from bulk (steric fit) and charge (electrostatics). The results show that good “shape fit” is the dominant driver of activity, while electrical interactions fine-tune binding. Importantly, building the models in the context of the receptor improved predictive power. These insights guided the design of new cannabinoid analogues predicted to bind more strongly.

For the fullerene series, the work focused on HIV-1 protease, an enzyme the virus needs to mature. Simulations captured how the enzyme’s flexible “flaps” move: they close over an inhibitor when bound and sit semi-open without one. Understanding this motion helped explain how fullerene derivatives lodge in the active site. Again, 3D-QSAR models converted these structural insights into concrete design rules, proposing next-generation inhibitor candidates.

Why it matters: The thesis demonstrates a practical roadmap—from conformation analysis to target-aware modeling to rational design—that turns molecular shape into actionable drug-discovery decisions. The approach is general, offering a template for designing better ligands across diverse targets, and pointing to testable candidates for cannabinoid receptors and HIV-1 protease. For details:

https://refubium.fu-berlin.de/handle/fub188/7753