Binding Analysis and Structure-Based Design of Tricyclic Coumarin-Derived MTHFD2 Inhibitors as Anticancer Agents: Insights from Computational Modeling

Selectivity analysis of diaminopyrimidine-based inhibitors of MTHFD1, MTHFD2 and MTHFD2L

The mitochondrial enzyme methylenetetrahydrofolate dehydrogenase (MTHFD2) is involved in purine and thymidine synthesis via 1C metabolism. MTHFD2 is exclusively overexpressed in cancer cells but absent in most healthy adult human tissues. However, the two close homologs of MTHFD2 known as MTHFD1 and MTHFD2L are expressed in healthy adult human tissues and share a great structural resemblance to MTHFD2 with 54% and 89% sequence similarity, respectively. It is therefore notably challenging to find selective inhibitors of MTHFD2 due to the structural similarity, in particular protein binding site similarity with MTHFD1 and MTHFD2L. Tricyclic coumarin-based compounds (substrate site binders) and xanthine derivatives (allosteric site binders) are the only selective inhibitors of MTHFD2 reported till date. Nanomolar potent diaminopyrimidine-based inhibitors of MTHFD2 have been reported recently, however, they also demonstrate significant inhibitory activities against MTHFD1 and MTHFD2L. In this study, we have employed extensive computational modeling involving molecular docking and molecular dynamics simulations in order to investigate the binding modes and key interactions of diaminopyrimidine-based inhibitors at the substrate binding sites of MTHFD1, MTHFD2 and MTHFD2L, and compare with the tricyclic coumarin-based selective MTHFD2 inhibitor. The outcomes of our study provide significant insights into desirable and undesirable structural elements for rational structure-based design of new and selective inhibitors of MTHFD2 against cancer.

Integrated Virtual Screening and Validation toward Potential HPPD Inhibition Herbicide

4-Hydroxyphenylpyruvate dioxygenase (EC 1.13.11.27, HPPD) is one of the most widely studied herbicide targets and has gained significant attention. To identify potential effective HPPD inhibitors, a rational multistep virtual screening workflow was built, which included CBP models (based on the receptor-ligand interactions in the crystal complex), Hypogen models with activity prediction ability (according to the derivation of structure-activity relationships from a set of molecules with reported activity values), and a consensus docking procedure (consisting of LibDock, Glide, and CDOCKER). About 1 million molecules containing diketone or β-keto-enol substructures were filtered by Lipinski's rules, CBP model, and Hypogen model. A total of 12 compounds with similar docking postures were generated by consensus docking. Eventually, four molecules were screened based on the specific binding pattern and affinity of the HPPD inhibitor. The biological evaluation in vivo displayed that compounds III-1 and III-2 exhibited comparable herbicidal activity to isoxaflutole and possessed superior safety on various crops (wheat, rice, sorghum, and maize). The ADMET prediction (absorption, distribution, metabolism, excretion, and toxicity) showed that compound III possessed relatively good toxicological results. This work provides a theoretical basis and valuable reference for the virtual screening and molecular design of novel HPPD inhibition herbicides.