Computational workflow for discovering small molecular binders for shallow binding sites by integrating molecular...

Area of Science:

• Biochemistry
• Molecular Biology
• Pharmacology

Background:

• Signal transducer and activator of transcription 3 (STAT3) is crucial for cellular processes and often dysregulated in cancer.
• STAT3 inhibition is a promising anti-cancer strategy, but challenges exist due to the protein's SH2 domain and water molecule interactions.
• Discovering potent STAT3 inhibitors is complex, necessitating advanced computational approaches.

Purpose of the Study:

• To extract pharmacophores from molecular dynamics simulations of STAT3-ligand complexes.
• To develop a predictive model using genetic function algorithm and machine learning (GFA-ML) for STAT3 inhibitor bioactivity.
• To identify novel STAT3 inhibitors by screening chemical databases.

Main Methods:

• Molecular dynamics (MD) simulations were performed on STAT3 SH2 domain-ligand complexes.
• Pharmacophores were extracted from MD frames.
• A GFA-ML model was employed to correlate pharmacophores with bioactivity, using augmented ligand conformer datasets.
• The NCI database was screened using the developed pharmacophore model.

Main Results:

• A single, significant pharmacophore representing STAT3-ligand binding was identified after 188 ns of MD simulation.
• The GFA-ML model effectively accounted for bioactivity variations among inhibitors.
• Database screening identified a low micromolar inhibitor with predicted binding to the STAT3 SH2 domain.

Conclusions:

• The study successfully identified a key pharmacophore for STAT3 inhibition, crucial for understanding ligand binding.
• The developed GFA-ML approach provides a robust method for discovering potent STAT3 inhibitors.
• This research offers a promising avenue for developing novel anti-cancer therapeutics targeting the STAT3 pathway.