Fragment-Hopping-Based Discovery of a Novel Chemical Series of Proto-Oncogene PIM-1 Kinase Inhibitors

Abstract

A new chemical series, triazolo[4,5-b]pyridines, has been identified as an inhibitor of PIM-1 by a chemotype hopping strategy based on a chemically feasible fragment database. In this case, structure-based virtual screening and in silico chemogenomics provide added value to the previously reported strategy of prioritizing among proposed novel scaffolds. Pairwise comparison between compound 3, recently discontinued from Phase I clinical trials, and molecule 8, bearing the selected novel scaffold, shows that the primary activities are similar (IC50 in the 20 to 150 nM range). At the same time, some ADME properties (for example, an increase of more than 45% in metabolic stability in human liver microsomes) and the off-target selectivity (for example, an increase of more than 2 log units in IC50 vs. FLT3) are improved, and the intellectual property (IP) position is enhanced. The discovery of a reliable starting point that fulfills critical criteria for a plausible medicinal chemistry project is demonstrated in this prospective study.

Deep Scaffold Hopping with Multi-modal Transformer Neural Networks

“…Note that the improvements are stable after fine-tuning 5% pairs, suggesting that the bioactivity feature is easy to capture compared to structural ones. and significantly improved off-target selectivity as well as ADME property 16 .…”

Deep Scaffold Hopping with Multi-modal Transformer Neural Networks

“…Next, we chose PIM-1 kinase (CHEMBL2147), a well-studied target for antitumor drugs, as a representative to mimic a real-world scaffold hopping process. To search for novel inhibitors of the PIM-1 kinase, Saluste and co-workers once reported a typical fragment hopping by replacing imidazopyridazine scaffold with triazolopyridine, which maintained the primary activity and significantly improved off-target selectivity as well as ADME property 16 .…”

“…4,[8][9] Many of them utilized matching algorithms based on two-dimensional (2D) or threedimensional (3D) molecular representations [10][11][12][13][14] . Others developed fragment replacement techniques, including fingerprints and combining experimental information [15][16][17][18][19] . However, almost all methods published to date relied exclusively on a predefined database to select a molecule or a fragment, with the differences between approaches arising solely from the searching algorithms of the database, the ways to define similarity of compound pairs or the contents of the scanned database.…”

Computational Scaffold Hopping: Cornerstone for the Future of Drug Design?

“…Recent trends in computational scaffold hopping include an increasing focus on natural products as starting points [10] and on methods for fragment replacements in active compounds [11][12][13]. Although scaffold hopping calculations were 'traditionally' carried out at the level of complete compounds, new search tools and descriptors have been introduced for comparing scaffolds directly and identifying promising replacements [13][14][15].…”