EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Virtual screening of Nox4 Inhibitors for knee osteoarthritis: A study combining AI-based virtual screening with computer-aided strategies

Ziteng Ma, Wei Wang, Yixin Chen, Zehong Lin, Fengbiao Wu, Yonghao Gao, Zhenning Li, Honghai Zhou

Letters in Drug Design & Discovery · 2026-09 · 23 (9) · 100598

DOI: 10.1016/j.lddd.2026.100598 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Letters in Drug Design & Discovery, 2026-09. Verfasst von Ziteng Ma und 7 weiteren Autor:innen. Die Arbeit nennt 51 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Knee osteoarthritis (KOA) is a common degenerative joint disease whose pathogenesis is closely associated with reactive oxygen species (ROS)-induced synovial cell death. As a key pro-oxidant enzyme, Nox4 has long been considered a promising therapeutic target; however, the development of specific inhibitors has been hindered by the lack of high-resolution experimental structures.

Methods

This study combined artificial intelligence (AI) with a multiscale computational framework to perform virtual screening for novel natural Nox4 inhibitors. Based on a high-confidence Nox4 structure predicted by AlphaFold, a hierarchical virtual screening was conducted on the COCONUT database (approximately 695,000 compounds). Following screening against Lipinski and QED pharmacophore criteria, deep-learning-driven DiffDock and CB-Dock were sequentially applied to narrow down the pool of candidate compounds. To exclude computationally unstable candidates and evaluate binding affinity, 100 ns full-atom molecular dynamics (MD) simulations and MM-PBSA calculations were subsequently performed.

Results

The tiered screening narrowed the candidate compounds down to two lead compounds: CNP0500966.0 and CNP0277172.2. The 100 ns molecular dynamics simulations showed that the root-mean-square deviation (RMSD) for both complexes converged, indicating excellent conformational stability within the Nox4 active pocket. MM-PBSA calculations revealed that CNP0500966.0 exhibits exceptional binding affinity (total binding free energy of −31.18 kcal/mol, significantly superior to the −20.31 kcal/mol of CNP0277172.2), which is primarily attributed to van der Waals forces and strong hydrophobic interactions.

Conclusion

This study not only identified computationally predicted antioxidant candidates for the treatment of knee osteoarthritis but also provided a reliable computational framework for hard-to-target therapeutic targets.

Zitieren

APA

Ma, Z., Wang, W., Chen, Y., Lin, Z., Wu, F., Gao, Y., Li, Z., & Zhou, H. (2026). Virtual screening of Nox4 Inhibitors for knee osteoarthritis: A study combining AI-based virtual screening with computer-aided strategies. Letters in Drug Design & Discovery, 23(9), 100598. https://doi.org/10.1016/j.lddd.2026.100598

Vancouver

Ma Z, Wang W, Chen Y, Lin Z, Wu F, Gao Y, et al. Virtual screening of Nox4 Inhibitors for knee osteoarthritis: A study combining AI-based virtual screening with computer-aided strategies. Letters in Drug Design & Discovery. 2026;23(9):100598. doi:10.1016/j.lddd.2026.100598

BibTeX
@article{Ma2026,
  author = {Ma, Ziteng and Wang, Wei and Chen, Yixin and Lin, Zehong and Wu, Fengbiao and Gao, Yonghao and Li, Zhenning and Zhou, Honghai},
  title = {Virtual screening of Nox4 Inhibitors for knee osteoarthritis: A study combining AI-based virtual screening with computer-aided strategies},
  journal = {Letters in Drug Design & Discovery},
  year = {2026},
  volume = {23},
  number = {9},
  pages = {100598},
  doi = {10.1016/j.lddd.2026.100598}
}
Kostenlos starten →