Rational Design of SIRT2 Inhibitors Using Combined 3D-QSAR, Docking, and Machine Learning Methods

Sirtuin 2 (SIRT2) is a key regulator of apoptosis, DNA repair, and the cell cycle, making it a promising target for anticancer therapy. Achieving high selectivity remains challenging due to the conserved NAD⁺ binding site among sirtuin family members. Derivatives of 5-((3-amidobenzyl)oxy)nicotinamide are among the most potent and selective SIRT2 inhibitors, providing a foundation for further exploration. [1] In this study, a 3D-QSAR model was constructed using 86 nicotinamide-based SIRT2 inhibitors, complemented by GRIND-derived pharmacophore modelling. [2] The model, developed using partial least squares regression, achieved R² = 0.89 and Q² = 0.69, with no outliers in the applicability domain. Model interpretation highlighted the positive influence of steric features, secondary amide groups, and nitrogen atoms in the pyridine ring on SIRT2 activity. Fine-tuning the propynyl and central phenyl groups produced derivatives with predicted enhanced potency. [3] Molecular docking on SIRT1–3 isoforms enabled the development of two machine learning classification models (Naive Bayes and k-nearest neighbours) to predict SIRT1/2 and SIRT2/3 selectivity profiles. Several selective SIRT2 inhibitors were identified in silico, and these will be synthesised and tested in vitro, supporting the development of novel SIRT2-targeted anticancer therapies. [3]

By integrating 3D-QSAR, selectivity modelling, and ADMET predictions, several promising selective SIRT2 inhibitors were identified. [3] These candidates will be synthesised and tested in vitro to validate their efficacy and selectivity, advancing the development of novel SIRT2-targeted anticancer therapies.

References

1. Ai T et al., 2023, Molecules 28:7655, https://doi.org/10.3390/molecules28227655
2. Chen L et al., 2016, U.S. Patent US2016/0376238A1
3. Ilić A et al., 2024, Comput. Biol. Chem. 113:108242, https:/doi.org/10.1016/j.compbiochem.2024.108242