Targeting farnesyl pyrophosphate synthase of Trypanosoma cruzi by fragment-based lead discovery

Trypanosoma cruzi (T. cruzi) is the causative agent of Chagas disease (CD), which mostly affects underprivileged populations in South and Central America. The current standard of care for this disease are the two empirically discovered drugs Benznidazole and Nifurtimox. They show low efficacy, difficulties in administration and severe side effects. Moreover, there are T. cruzi strains that have formed resistances. Thus, the development of a safe and efficient drug is urgently needed. T. cruzi is dependent on isoprenoid biosynthesis as ergosterol and other 24 alkylsterols are essential metabolites that cannot be acquired by other mechanisms. Therefore, it was hypothesised that enzymes along this pathway are promising drug targets. A number of compounds targeting these enzymes were tested and have been shown to inhibit parasite growth. Among those enzymes is farnesyl pyrophosphate synthase (FPPS), a key branch-point enzyme in the isoprenoid pathway, which is in the focus of this work. It catalyses the synthesis of farnesyl pyrophosphate (FPP), a C15 building block in sterol biosynthesis and in protein prenylation of signalling proteins. Bisphosphonates (BPs) are known active site directed FPPS inhibitors, which exhibit ideal pharmacokinetics to target bone mineral and are used to treat bone diseases. BPs can also combat T. cruzi flagellates but are not ideal to treat CD due to their pharmacokinetics. In the search for new chemotypes, several non-BP inhibitors that bind to another pocket were found for human FPPS (hFPPS) by fragment based screening (FBS). Recently, it was shown that the product of FPPS, farnesyl pyrophosphate (FPP), can bind to this pocket and locks the enzyme in an open and inactive state, thus showing the allosteric character of this pocket. The current work aims at the discovery of non-BP inhibitors of T. cruzi FPPS (TcFPPS), which could be starting points for the development of a treatment against CD. Towards this goal, recombinant expression in E. coli cells and purification by means of IMAC and SEC yielded pure und homogenous TcFPPS (chapter 5.1). This includes unlabelled, 13C15N labelled and in vivo biotinylated avi-tagged TcFPPS. Furthermore, a novel, reliable, highly reproducible, and well-diffracting crystallization system was established. The system exhibits excellent properties for FBS as it was compatible with different types of 96-well plates, including 2 drop and 3 drop SwissCi/MRC, as well as CrystalDirect™ plates. Apo crystals were stable for up to 24 h in 15% DMSO and allowed collection of data sets with a diffraction limit of around 1.6 Å. The highest achieved diffraction limit was 1.28 Å for a soaked TcFPPS crystal (PDB ID 6R09). The allosteric region in TcFPPS was investigated by means of sequence analysis and structural superimposition of various orthologous FPPSs (chapter 5.2). This revealed that the allosteric region is less conserved than the active site. Differences among residues in equivalent positions that form the allosteric site were observed, which is surprising if it is assumed that all FPPSs can be product inhibited as hFPPS. A remarkable finding is that residue Phe50 in TcFPPS is an exception in an otherwise highly conserved position. It causes steric hindrance of the pocket in TcFPPS. An attempt to reposition established allosteric inhibitors of hFPPS showed binding affinity to TcFPPS but the two obtained crystal structures demonstrated their binding to sites on the protein surface (sites S1 and S2, PDB IDs 6R08 and 6R07, respectively). The Novartis core and fluorine library (1336 and 482 compounds) were screened on TcFPPS, which resulted in 63 and 45 validated fragment hits, respectively (chapter 5.3). Performing the same screen with T. brucei FPPS (TbFPPS), the causative agent of African sleeping sickness and counter screening on hFPPS led to unique, pairwise and triple binders demonstrating selectivity at the early stage of FBS. Strikingly, TcFPPS has generally more binders than TbFPPS, and TcFPPS has many unique hits when compared to TbFPPS. Subsequent crystallization experiments with the core library hits resulted in 3D structures of two TcFPPS complexes. One ligand binds to the homodimer interface (site S12) and the other one in the active site. The latter was identified by using the statistical analysis tool Pan-Dataset Density Analysis (PanDDA). FBS by X-ray crystallography at the XChem facility in Harwell, UK, and the HTXlab in Grenoble, France, were conducted (chapter 5.4). The XChem screen identified 35 fragment binders (PDB IDs 5QPD – Z, 5QQ0 – 9, 5QQA – C) in binding sites that were distributed over the entire protein. This includes the active site, the allosteric site, the homodimer interface, sites on the surface and a new site in close proximity to the active site. Strikingly, the first two fragments binding to the allosteric site of TcFPPS in its open state were identified. Rotation of the phenyl side chain of Phe50 led to opening of the former closed pocket. The HTXlab screen identified additional binders for the active and allosteric site. In total 1244 data sets were collected and analysed. This process was accelerating using PanDDA. The first fragment-to-lead optimization by means of virtual screening using the web-based platform ANCHOR.QUERY was based on fragment hit LUY (chapter 5.5). Compounds were synthesised using one-pot one-step multi-component reactions. Synthesis of 11 compounds (MCR 1 – 11) was successful, but poor solubility was detrimental in subsequent testing on TcFPPS and crystallization experiments did not lead to a structure model of a complex. A second fragment to lead optimization using a fragment merging approach for chemical optimization was based on the active site directed binders AWM, LVV, LUY, LDV and AWV (chapter 5.6). A library of 12 compounds (MCN 1 – 12) was synthesised by reductive amination. X-ray structures revealed unexpected binding modes for compounds MCN-1, -4 and -8. Instead of retaining the binding site of the fragment, the merged compounds bind to the surface directed binding site S1 (PDB IDs 6R09, 6R0A, 6R0B). Nevertheless, the 42 crystal structures of TcFPPS-fragment-complexes provided in this work will pave the way for future drug discovery campaigns for CD. The large diversity of the fragments’ scaffolds and different binding sites are potential starting points for inhibitors with different physicochemical properties and a novel mode of action that might help to overcome the limitations related to the BP scaffold