From Deep-Sea Natural Product to Optimized Therapeutics: Computational Design of Marizomib Analogs

The proteasome β5 subunit plays a central role in protein degradation and is an established therapeutic target in glioblastoma. Marizomib (MZB), a natural β5 inhibitor, has shown promising anticancer activity, yet suboptimal pharmacological properties limit its clinical translation. Using a comprehensive computational approach, this study aimed to identify and characterize novel MZB analogs with improved binding affinity, stability, and drug-like profiles. An integrative in silico study was performed, including molecular docking, frontier molecular orbital (FMO) analysis, pharmacophore modeling, molecular dynamics (MD) simulations over 200 ns, MM/PBSA binding free energy calculations, and per-residue energy decomposition. ADMET profiling evaluated the pharmacokinetic and safety properties of MZB and top-performing analogs. Docking and pharmacophore modeling revealed strong complementarity between MZB analogs and the β5 catalytic pocket. MD simulations showed that MZBMOD-77 and MZBMOD-79 exhibited exceptional structural stability with low RMSD values (0.40-0.42 nm), persistent binding within the active site cavity, and significant disruption of hydrogen-bond networks in the active loop regions Ala19-Lys33 and Val87-Gly98. MM/PBSA analysis confirmed their superior binding free energies (-19.99 and -18.79 kcal/mol, respectively), surpassing native MZB (-6.26 kcal/mol). Per-residue decomposition highlighted strong contributions from Arg19, Ala20, Lys33, and Ala50. ADMET predictions indicated improved oral absorption, reduced toxicity, and favorable pharmacokinetics compared to native MZB. This integrative computational study identifies MZBMOD-77 and MZBMOD-79 as promising next-generation proteasome β5 inhibitors. These analogs mimic and enhance the inhibitory mechanism of native MZB, offering potential candidates for further optimization and preclinical development in glioblastoma therapy.