Identification of STAT3 phosphorylation inhibitors using generative deep learning, virtual screening, molecular dynamics simulations, and biological evaluation for non-small cell lung cancer therapy

Abstract

The development of phosphorylation-suppressing inhibitors targeting Signal Transducer and Activator of Transcription 3 (STAT3) represents a promising therapeutic strategy for non-small cell lung cancer (NSCLC). In this study, a generative model was developed using transfer learning and virtual screening, leveraging a comprehensive dataset of STAT3 inhibitors to explore the chemical space for novel candidates. This approach yielded a chemically diverse library of compounds, which were prioritized through molecular docking and molecular dynamics (MD) simulations. Among the identified candidates, the HG110 molecule demonstrated potent suppression of STAT3 phosphorylation at Tyr705 and inhibited its nuclear translocation in IL6-stimulated H441 cells. Rigorous MD simulations further confirmed the stability and interaction profiles of top candidates within the STAT3 binding site. Notably, HG106 and HG110 exhibited superior binding affinities and stable conformations, with favorable interactions involving key residues in the STAT3 binding pocket, outperforming known inhibitors. These findings underscore the potential of generative deep learning to expedite the discovery of selective STAT3 inhibitors, providing a compelling pathway for advancing NSCLC therapies.

Keywords: Apoptosis; Drug design; Molecular dynamics; NSCLC; Signal transducer and activator of transcription 3.

Fig. 1

Chemical structures of STAT3 inhibitors with different scaffolds

Fig. 2

The flowchart for generative deep learning model, virtual screening, and biological assays to identify merging STAT3 inhibitors

Fig. 3

Generation of a virtual compound library against STAT3 using the cRNN-ased generative model. A schematic illustration of library using GDL methods. B chemical space of generated space (gray dot), source space (red dot), and target space (blue dot) by transfer learning. C the distributions of molecular weight, logP, QED, and logS of targeted data and the generated data by GDL generation methods

Fig. 4

HG106 and HG110 efficiently suppress cancer cell growth. A representative image (left panels) and quantification (right panels) of concentration-manner colony formation assays of cancer cells treating with HG106 and HG110, respectively. B bright-field images of soft agar colony formation of A549 cells treated with HG106 and HG110, respectively. All the data were presented as Mean±S.D. Statistical analysis was conducted by one-way ANOVA (***p < 0.001)

Fig. 5

A The percentage of apoptotic cells quantitatively measured by using flow cytometry for the H441 cells stained by Annexin V/PI. Cells were treated with HG106 and HG110 at concentration of 5 uM for 24 h, respectively. DMSO was selected as control. B Immunoblot bands and quantitative measurement revealed apoptosis-associated protein expression levels after dose-dependent treatment against H441 cells. C Immunoblot analysis reveals the effect of HG110 on the phosphorylation levels of STAT3 at Tyr705 compared with HG106 at 24 h in H411 and H1299 cells. GAPDH was selected as inner standard. D HG110 suppressed phosphorylation levels of STAT3 at 24 h. E Immunofluorescent staining revealed the impacts of HG110 on IL6-promoted nucleus translocation of STAT3 in H441 cells. Scale bar is XX um. All the data were presented as Mean±S.D. Statistical analysis was conducted by one-way ANOVA (*p < 0.05, **p < 0.01, ***p < 0.001)

Fig. 6

Molecular dynamic simulation to reveal the binding model of HG106 and HG110 with STAT3. A Molecular structure, binding model, and planar residue–ligand interaction of HG106. B Analysis of residue root mean square deviation (RMSD) in compound HG106 and STAT3 complex for protein (STAT3) stability. C Profile of hydrogen bonds between HG106 and STAT3 among molecular dynamic simulation. D Molecular structure, binding model, and planar residue–ligand interaction of HG110. E Analysis of residue root mean square deviation (RMSD) in compound HG110 and STAT3 complex for protein (STAT3) stability. F Profile of hydrogen bonds between HG110 and STAT3 among molecular dynamic simulation