Validation of the repurposing of the methotrexate scaffold for the design of Janus kinase modulators with potential inhibitory activity

Abstract

Polycythaemia vera, a rare blood cancer, has been associated with a mutation in the JAK2V617F gene, in turn leading to the activation of the JAK/STAT signalling pathway. This receptor is therefore a target in the identification of molecules that target myeloproliferative neoplasms. Antagonism can reverse underlying marrow pathologies, disrupting erythrocytosis and other inflammatory responses. The identification of lowdose methotrexate as a selective and potent JAK2 antagonist indicates, that there is potential for its repurposing, since current treatments are associated with the increased risk of transformation of polycythaemia vera to acute leukemia. This study uses the methotrexate scaffold as the lead molecule in order to recognize high affinity ligands using two different approaches: virtual screening and de novo techniques. Conformational analysis was performed on the methotrexate scaffold. The methotrexate molecule was modelled and docked into the apo JAK2 receptor. This process generated 20 different conformers. The optimal conformer was identified by calculating the ligand binding energy in kcalmol-1 and the ligand binding affinity in pKd for each conformer, and plotting them on a line graph. The molecule that was the most energetically feasible and that had the highest affinity was chosen as the best optimal conformer. In the Virtual Screening approach, the methotrexate molecule was used as the template to generate a cohort of Lipinski Rule compliant hit molecules. These hits were docked into a generated protomol and the resultant 991 compliant molecules were ranked according to their affinity for the JAK2 ligand binding pocket. In the de novo approach 3 seed structures were modelled, based on the optimal methotrexate molecule, and growth was allowed, through a fragmentation exercise, within the pharmacophoric space, in order to generate novel molecules. A number of filters were applied to ensure that the best molecules are Lipinski Rule compliant, and hence able to penetrate the blood brain barrier. Lipinski Rule compliant de novo molecules were generated and separated into different families based on their pharmacophoric similarities, and then ranked according to affinity. The optimal molecules obtained from these two approaches will be retained for further validation and optimization.