Dual-Targeting MDM2 and CDK4: An Integrated Machine Learning and Molecular Simulation Pipeline Identifies Novel Fungal Metabolites for Cancers with Co-Dysregulated p53 and Rb Pathways

• First computational report identifying fungal metabolites as predicted high‑affinity dual MDM2/CDK4 inhibitors via an integrated machine‑learning and simulation pipeline. • Novel leads demonstrate superior dual-target affinity and thermodynamics compared to clinical reference inhibitors. • Atomic-resolution mechanism reveals key interactions: a salt bridge with CDK4-Asp158 and complete hydrophobic occupation of the MDM2 cleft. • In silico ADMET profiling predicts high oral bioavailability, low CNS penetration, and reduced cardiotoxicity risk. • Proposed therapeutic rationale restricted to TP53‑wild‑type, MDM2‑amplified tumors. Resistance to single-target therapies necessitates innovative strategies in oncology, particularly for aggressive cancers characterized by co-dysregulated p53 and Rb pathways. Here, we present an integrated computational pipeline that identifies novel fungal metabolites as predicted dual inhibitors of human E3 ubiquitin-protein ligase (MDM2) and cyclin-dependent kinase 4/cyclin D1 (CDK4). We developed robust machine learning QSAR models (test R² = 0.83 for CDK4, 0.81 for MDM2) to virtually screen a curated fungal metabolome library. This approach identified three lead compounds, CNP0341466.2, CNP0341466.5, and CNP0356395.1 that demonstrated exceptional dual-target affinity, with molecular docking scores ranging from -11.6 to -10.8 kcal/mol for CDK4 and -11.20 to -9.2 kcal/mol for MDM2, significantly outperforming the selective inhibitors abemaciclib (-10.90 kcal/mol) and nutlin-3 (-8.7 kcal/mol). Molecular dynamics simulations (2 × 100 ns) confirmed the stability of these complexes, with mean backbone RMSD values as low as 0.16 nm for MDM2 and consistently below 0.30 nm for CDK4 across replicate trajectories. Thermodynamic profiling via MM/GBSA revealed superior binding free energies for the leads (ΔGbind: -36.29 ± 0.17 to -25.81 ± 0.14 kcal/mol for CDK4; -29.71 ± 0.16 to -22.55 ± 0.10 kcal/mol for MDM2). Per-residue decomposition quantified pivotal interactions, including a dominant salt bridge with CDK4-Asp158 (-6.40 ± 0.05 kcal/mol) and complete hydrophobic encapsulation of the MDM2 cleft by Leu54, Val93, and Ile99 (contributions up to -3.27 ± 0.02 kcal/mol). Density functional theory (DFT) calculations confirmed the electronic stability and favorable frontier orbital properties of the lead scaffolds. In silico ADMET profiling further supported their therapeutic potential, predicting high oral absorption, minimal CNS penetration, and a markedly reduced cardiotoxicity risk relative to clinical benchmarks. This work not only establishes a validated computational pipeline for dual-target drug discovery but also identifies specific fungal-derived chemotypes as predicted first-in-class leads for overcoming resistance in cancers with concurrent pathway dysregulation. Importantly, this study demonstrates the untapped potential of the fungal metabolome as a rich source of complex, evolutionarily optimized scaffolds for challenging multi-target applications.