Computational Discovery of Dual-Target LDHA/BRD4 Inhibitors Targeting the Lactate-Kla-B7-H3 Immunosuppressive Axis Through AI-Driven Virtual Screening.

Background/

Objectives

Immune evasion remains a critical barrier to effective hepatocellular carcinoma (HCC) therapy. Lactate dehydrogenase A (LDHA) drives lactate accumulation and histone lysine lactylation (Kla), reshaping the immunosuppressive microenvironment, while bromodomain-containing protein 4 (BRD4) sustains B7-H3 transcription via super-enhancer occupancy. Despite their synergistic roles in the lactate-Kla-B7-H3 immunosuppressive axis, no dual-target inhibitor simultaneously engaging both proteins has been reported. This study aimed to discover dual LDHA/BRD4 inhibitors from natural product libraries using an integrated AI-driven computational pipeline.

Methods

We established a multi-tier virtual screening cascade comprising Lipinski/QED drug-likeness filtration, DiffDock-based AI docking, QuickVina binding energy validation, PLIP interaction profiling, 200 ns all-atom molecular dynamics simulations, MM-GBSA binding free energy calculations, and density functional theory analysis. Natural product libraries from COCONUT and CMNPD databases (84,730 compounds post-filtration) were screened against both targets.

Results

High-throughput DiffDock screening identified 11 dual-target hits, from which CNP0038114.1 and CMNPD16582 emerged as prioritized lead candidates. All four protein-ligand complexes maintained structural stability throughout MD simulations, with MM-GBSA binding free energies ranging from -27.24 to -32.45 kcal/mol, predominantly driven by van der Waals interactions. DFT calculations revealed distinct electronic profiles: CNP0038114.1 exhibited a narrow HOMO-LUMO gap (2.718 eV) favoring charge-transfer reactivity, whereas CMNPD16582 displayed a larger gap (4.822 eV), suggesting superior chemical stability.

Conclusions

This computational study furnishes two novel natural product leads for targeting the lactate-Kla-B7-H3 immunosuppressive axis in HCC, establishing a generalizable AI-driven workflow for dual-target inhibitor discovery.