#2263 Computational drug repurposing screen targeting PLA2R antibody binding

Abstract

Listen

Translate article

Abstract Background and Aims Approximately 60% of patients with membranous nephropathy have antibodies against the phospholipase A2 receptor (PLA2R). The cysteine rich (CysR) epitope of PLA2R is a likely site of protein-antibody interaction. Current treatment regimens for membranous nephropathy involve maximal conservative management or systemic immunosuppression in severe cases. We have developed protocols and programs for an in silico high-throughput repurposing screen using structural modelling, docking, and molecular dynamics (MD). Uniquely, our approach uses a bespoke interface on Google Colab for graphic processing units to run the MD simulations. We describe their application to a repurposing screen aiming to identify larger licensed drug molecules capable of impeding antibody-protein interactions at the site of the CysR epitope. Method A complete model of PLA2R was obtained from I-TASSER using the canonical sequence on the UNIPROT database. This structure was equilibrated using a five nanosecond molecular dynamics (MD) simulation. The simulation was prepared using CHARMM-GUI and run on the GROMACS-ON-COLAB platform. The centroid of the MD system between three and five nanoseconds was used for docking experiments using PLANTS PLP with a study area that included the CysR epitope. A library of compounds listed in the British National Formulary was used for the docking studies. The results would be analysed by PLANTS energy and by key residue overlap score. Results Viable docking results were obtained for 891 of the 1510 compounds tested. 60 of these compounds were in contact with at least 10% of the residues in the CysR region. By restricting these results to compounds in the lowest 20% of the total sample for calculated binding energy, 22 compounds remained. 16 of these compounds have been assessed for their safety for oral or intravenous use. These sixteen compounds were then identified for further investigation by MD to assess the stability of the identified interactions. The sixteen compounds include Lymecycline, Trazodone, Mirabegron, Sulfasalazine, Methotrexate, Rosuvastatin and Montelukast. Conclusion Structural bioinformatics methods provide a relatively fast and economical option for identifying protein-drug interactions that could result in new treatment strategies. The compounds identified for investigation are used for an array of different health conditions. However, they all share the chemical properties of large molecular weight and have the potential to form multiple hydrogen bonds. If validated experimentally, these compounds could provide new treatment options for PLA2R antibody positive membranous nephropathy.