An integrated in silico and SPR-based workflow identifies FDA-approved drugs as previously unrecognized SIRT3 binders

Ataxia–Telangiectasia (AT) is a rare disorder characterized by ATM deficiency and mitochondrial dysfunction. Because SIRT3 contributes to mitochondrial homeostasis in ATM-deficient settings, it represents a relevant molecular context for repurposing-oriented target-engagement studies. Here, we present an integrated computational and biophysical workflow designed to identify experimentally tractable SIRT3 binders within a clinically approved chemical space. A curated dataset of 2342 FDA-approved drugs was screened using hierarchical GNINA-based virtual screening, SAveRUNNER network analysis, triplicate 300 ns molecular dynamics simulations, post-MD redocking, MM/PBSA calculations, and experimental SPR validation. Four compounds were prioritized for further investigation: binimetinib, olaparib, mizolastine, and teniposide. Surface plasmon resonance (SPR) demonstrated direct binding of all four compounds to recombinant SIRT3, with equilibrium dissociation constants in the nano- to low-micromolar range. Binimetinib showed the highest apparent affinity (KD = 21.28 ± 3.5 nM), followed by olaparib (84 ± 5.7 nM), mizolastine (138 ± 3.5 nM), and teniposide (1470± 58 nM). Computational analyses supported the compatibility of these ligands with dynamically relaxed SIRT3 binding-site conformations and enabled the efficient prioritization of experimentally validated binders. These results support the value of an integrated in silico/SPR pipeline for identifying SIRT3-binding scaffolds. However, the functional consequences of binding, isoform selectivity, and biological relevance in ATM-deficient disease models remain to be established.