E-pharmacophore and deep learning based high throughput virtual screening for identification of CDPK1 inhibitors of Cryptosporidium parvum

Abstract

Cryptosporidiosis, a prevalent gastrointestinal illness worldwide, is caused by the protozoan parasite Cryptosporidium parvum. Calcium-dependent protein kinase 1 (CpCDPK1), crucial for the parasite's life cycle, serves as a promising drug target due to its role in regulating invasion and egress from host cells. While potent Pyrazolopyrimidine analogs have been identified as candidate hit molecules, they exhibit limitations in inhibiting Cryptosporidium growth in cell culture, prompting exploration of alternative scaffolds. Leveraging the most potent compound, RM-1–95, co-crystallized with CpCDPK1, an E-pharmacophore model was generated and validated alongside a deep learning model trained on known CpCDPK1 compounds. These models facilitated screening Enamine's 2 million HTS compound library for novel CpCDPK1 inhibitors. Subsequent hierarchical docking prioritized hits, with final selections subjected to Quantum polarized docking for accurate ranking. Results from docking studies and MD simulations highlighted similarities in interactions between the cocrystallized ligand RM-1–95 and identified hit molecules, indicating comparable inhibitory potential against CpCDPK1. Furthermore, assessing metabolic stability through Cytochrome 450 site of metabolism prediction offered crucial insights for drug design, optimization, and regulatory approval processes.