Thesis title: DEVELOPMENT AND APPLICATION OF BIOINFORMATICS METHODS IN THE ANALYSIS OF THE INTERACTIONS BETWEEN AURORA-A KINASE AND ORTHOSTERIC INHIBITORS, AND IN THE DEVELOPMENT OF NEW TANKYRASE INHIBITORS
Aurora kinase proteins are a family of serine-threonine kinases implicated in several processes of cell division. Kinases play crucial roles in regulating a wide range of cellular processes, by catalysing phosphate transfer to serine, threonine, and tyrosine residues . Most protein kinases are excellent drug targets, especially in oncology, because they are the basis of dysregulations that can lead to the onset of cancer (e.g., Imatinib/Gleevec targeting BCR-Abl tyrosine kinase, c-Kit receptor tyrosine kinase and PDGF receptor) . However, kinase proteins share a high similarity of their conserved ATP binding sites. This observation presents a significant challenge in the discovery of non-promiscuous inhibitors. Because the nucleotide-binding sites of kinases are the most promising, yet the most highly conserved, the two main challenges for the discovery of new inhibitors are 1) to develop compounds that selectively bind to a specific nucleotide-binding site of a specific kinase and 2) to obtain new, highly potent compounds that can compete with millimolar-level ATP concentrations. The combination of selectivity and potency presents a serious challenge. We tackled this goal using a computational approach. In particular, the aim of the first part of this thesis is to improve the docking prediction and scoring performance of Virtual Screening (VS) for small molecules with potential inhibitory activity on Aurora-A protein kinase, with the main of obtaining in the future a kinase-focused VS server.
Therefore, this project can be divided into four parts:
• Dataset creation and analysis of crystallographic structures, belonging to the Aurora-A kinase;
• Clustering and analysis of ligands, and development of class-based pharmacophores for each set of scaffolds;
• Performing docking calculations in basic mode, with pharmacophoric maps, with flexible residues and with both;
• Analysis of the results.
The second part of this thesis focuses on the development of new compounds able to inhibit tankyrase enzymes. Tankyrases belong to the Diphtheria toxin-like ADP-ribosyltransferase (ARTD) enzyme superfamily, also known as poly (ADP-ribose) polymerases (PARPs). They catalyse a covalent post-translational modification reaction, where they transfer ADP-ribose units from NAD+ to target proteins. Tankyrases are involved in many cellular processes and their roles in telomere homeostasis, Wnt signalling and in several diseases including cancers have made them interesting drug targets.
All members of the PARP family catalyse the addition of a portion of ADP-ribose from NAD+ to acceptor proteins, such as various amino acid side chains, and the subsequent elongation of the polymer. Among these, Tankyrase 1 (TNKS1, PARP-5a) and Tankyrase 2 (TNKS2, PARP5b) are particular in that, in addition to the conserved C-terminal poly ADP-ribosylating catalytic domain, they include two distinct protein-protein interaction domains: a sterile α-motif and a cluster of five ankyrin repeats, which mediate self-multimerization and recognition of target proteins, respectively.
The previously identified MC2050 inhibitor  was co-crystallised in the PARP-1 protein. TNKS1 and TNKS2 are very similar to each other in their domain construction, they share 82% sequence identity and due to their unique domain structure, they differ from the other four poly (ADP-ribosyl) transferases not only in their overall domain structure but also in their functions.
Based on this evidence, and using a structure-based computational approach, the aim of this work was to develop a series of MC2050 analogues with selectivity on TNKS. Starting from the structure of 2-mercapto-quinazolone, a series of compounds were synthesised with the characteristic of protruding from the nicotinamide binding site to create additional and specific interactions with the adenosine subunit of TNKS. We used molecular docking simulations in order to predict the binding mode of a previously reported 2-mercapto-quinazolinone-based PARP inhibitor (MC2050) in PARP-1 catalytic domain and validated the accuracy of this model by X-ray co-crystal structure. Superimposition of MC2050 from PARP-1 co-crystal structure within the TNKSs catalytic domains uncovered that quinazolinone H-bonding could be preserved and paved the way for rational designing derivatives. By this approach, we generated a focused library of compounds in which the 2-mercapto-quinazolinone scaffold was functionalized with different aromatic systems installed on a piperazine moiety used as a linker.
In conclusion, herein we reported a computationally driven drug design, structure-based approach, that with a limited, but extremely focused, a synthetic effort led to the discovery of a bis-quinazolinone-based nicotinamide-adenosine dual-binder as a picomolar and selective TNKS-2 inhibitor, that has been selected as a lead compound for further optimization studies.
1. Mitchison TJ, Salmon ED. “Mitosis: a history of division” Nat Cell Biol (2001):5,E17-E21
2. P. Cohen, DR. Alessi. “Kinase drug discovery--what's next in the field?” ACS Chem Biol. (2013):18,96-104.
3. Kim, M. K. Novel insight into the function of tankyrase. Oncol. Lett. 2018, 16, 6895−6902