Fragment Based Design of GLS1 Allosteric Inhibitors for Cancer Therapy

MOTIVATION Glutamine is the second most utilized energy source for cancer cells after glucose, playing a crucial role in their growth and survival. Glutaminase 1 (GLS1) is the rate-limiting enzyme in the glutaminolysis pathway and thus represents a promising therapeutic target for developing innovative antitumor agents, particularly effective on triple negative breast cancer. Two main classes of GLS1 inhibitors, based on their different binding mode, are reported in literature: the substrate active site inhibitors and the allosteric site inhibitors. Despite the intense efforts made, to date, only two GLS1 inhibitors (CB-839 and IPN60090) have entered clinical trials. Therefore, this research field remains to be explored to improve the effectiveness of anticancer therapy. Hence, is here described an in silico fragment-based approach targeted to the rational design of new glutaminase allosteric site inhibitors. METHODS 47 crystallographic structures of the enzyme in apo-form or in complex with its natural substrate or various inhibitors available for GLS1 are deposited in the Protein Data Bank. 5HL1 was chosen as the most appropriate three-dimensional structure of the target protein. Initial ligand structures for virtual screening were taken from an in-house dataset of approved drugs from AIFA and Drug Bank. Docking studies were carried out using Autodock-Vina with the Vinardo scoring function, which proved to be the best performing in reproducing the interactions between ligands and GLS1 as known from experiments. Then, a fragment-based approach was applied on the best virtual screening results. New molecules were generated through fragments combination. The selection of the best candidates during all three steps (parent drugs, fragments, newly assembled molecules) was based on their energy ranking and their conformational ability to superimpose with X-ray binding pose of CB-839 (selected as template). RESULTS Among the known allosteric sites, we focused on the one most widely explored and characterized by structural biology, where CB-839 and its analogues bind. This allosteric site is defined by the solvent-exposed region at the dimer interface of the enzyme. CB-839 binds in this allosteric pocket, bridging the two GLS1 monomers. This specific ligand binding mode requires a long molecule, made up of a “head”, a “linker” and a “tail” and capable of adopting a U-shaped conformation, that allows the enzyme inactivation. Based on these considerations, a sequential docking methodology was applied to computationally predict starting points for fragment combination. The best initial parent drugs from the virtual screening were fragmented into “head”, “linker” and “tail” and a fragment library was created. The library was docked into the allosteric binding site to identify the best fragments, according to energy ranking and CB-839 pose overlapping. The top solutions from each fragments’ group, head, linker and tail, were re-assembled to generate new molecules. The final ligands will be docked into the allosteric site. The most promising molecules emerging from this strategy will be further validated by means of molecular dynamics studies prior to chemical synthesis and biological evaluation.