Scholarlink

<p>The human apoptosis regulator Bcl-2 (Bcl-2) protein plays a pivotal role in many blood cancers such as chronic lymphocytic leukemia (CLL), acute myeloid leukemia (AML), and multiple myeloma (MM). By inhibiting apoptosis, Bcl-2 contributes to tumor growth and resistance to chemotherapy. Therefore, inhibiting the anti-apoptotic function of Bcl-2 has emerged as a promising approach to enhance therapeutic efficacy in hematologic malignancies. In this study, we used neural network machine learning models, molecular docking, and molecular dynamics simulations to screen a flavonoid library for potent Bcl-2 inhibitors. We computed four types of molecular fingerprints- Morgan, RDKit, AtomPair, and TopologicalTorsion- of compounds used in Bcl-2 bioassays and generated predictive neural network models based on these descriptors. The quantitative structure-activity relationship (QSAR) model developed with Morgan fingerprints demonstrated superior performance, with an R2 value of 0.804. Among 4857 flavonoids, 83 compounds were predicted to be ‘Active’. Subsequently, triplicate molecular docking simulations identified several flavonoids with favorable docking scores relative to Venetoclax, the FDA-approved Bcl-2 inhibitor used to treat CLL, AML, and MM. The top ten docking hits were validated using replicated 200-ns molecular dynamics simulations, where the flavonoid Comp-7 exhibited the most stable root-mean-square deviation (RMSD), which converged within the first 20 ns and remained below 0.2&#xa0;nm throughout the simulation. Analysis of root-mean-square fluctuation (RMSF), radius of gyration, and hydrogen bonding also supported the stability of the interactions between Comp-7 and Bcl-2. Overall, the flavonoid Comp-7 was predicted as a Bcl-2 inhibitor by the neural network QSAR model, showed a favorable docking score, and exhibited stable interactions in molecular dynamics simulations. Our results can be used to develop effective Bcl-2 inhibitors for the treatment of CLL, AML, and MM.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning and molecular simulations reveal a prenyl plant flavonoid as a potent Bcl-2 inhibitor in cancer therapy

  • 期刊论文

  • 发表日期：

    2026年6月2日

查看全文
  • Abtin Tondar,
  • Asim Kumar Bepari,
  • Hasan Mahmud Reza,
  • Laura Calvet Liñán,
  • David Hervás-Marín

摘要

The human apoptosis regulator Bcl-2 (Bcl-2) protein plays a pivotal role in many blood cancers such as chronic lymphocytic leukemia (CLL), acute myeloid leukemia (AML), and multiple myeloma (MM). By inhibiting apoptosis, Bcl-2 contributes to tumor growth and resistance to chemotherapy. Therefore, inhibiting the anti-apoptotic function of Bcl-2 has emerged as a promising approach to enhance therapeutic efficacy in hematologic malignancies. In this study, we used neural network machine learning models, molecular docking, and molecular dynamics simulations to screen a flavonoid library for potent Bcl-2 inhibitors. We computed four types of molecular fingerprints- Morgan, RDKit, AtomPair, and TopologicalTorsion- of compounds used in Bcl-2 bioassays and generated predictive neural network models based on these descriptors. The quantitative structure-activity relationship (QSAR) model developed with Morgan fingerprints demonstrated superior performance, with an R2 value of 0.804. Among 4857 flavonoids, 83 compounds were predicted to be ‘Active’. Subsequently, triplicate molecular docking simulations identified several flavonoids with favorable docking scores relative to Venetoclax, the FDA-approved Bcl-2 inhibitor used to treat CLL, AML, and MM. The top ten docking hits were validated using replicated 200-ns molecular dynamics simulations, where the flavonoid Comp-7 exhibited the most stable root-mean-square deviation (RMSD), which converged within the first 20 ns and remained below 0.2 nm throughout the simulation. Analysis of root-mean-square fluctuation (RMSF), radius of gyration, and hydrogen bonding also supported the stability of the interactions between Comp-7 and Bcl-2. Overall, the flavonoid Comp-7 was predicted as a Bcl-2 inhibitor by the neural network QSAR model, showed a favorable docking score, and exhibited stable interactions in molecular dynamics simulations. Our results can be used to develop effective Bcl-2 inhibitors for the treatment of CLL, AML, and MM.