Chandrasekaran Jaikanth, Gopal Dhanushya, Sureshkumar Lokesh Vishwa, Santhiyagu Infant Xavier, Senthil Kumar Varsha, Munuswamy Bhuvaneshwari, Gani Beevi Fathima Harshatha Mohamed Yousuf, Agrawal Mohit
Department of Pharmacology, Sri Ramachandra Faculty of Pharmacy, Sri Ramachandra Institute of Higher Education and Research (Deemed to be University), Chennai, Tamil Nadu, 600116, India.
Department of Pharmacology, School of Medical and Allied Sciences, K.R. Mangalam University, Gurugram, Haryana, India.
Mol Divers. 2025 Mar 19. doi: 10.1007/s11030-025-11157-y.
The dysregulation of the cell cycle in cancer underscores the therapeutic potential of targeting WEE1 kinase, a key regulator of the G2/M checkpoint. This study harnessed artificial intelligence (AI)-driven methodologies, particularly the MORLD platform, to identify novel WEE1 inhibitors. Starting with clinically validated WEE1 inhibitors as references, we generated 20,000 structurally diverse compounds optimized for binding affinity, synthetic accessibility, and drug-likeness. A rigorous cheminformatics pipeline-comprising PAINS filtering, physicochemical property assessments, and molecular fingerprinting-refined this library to 242 promising candidates. Dimensionality reduction using UMAP and clustering via K-means enabled the prioritization of structurally unique leads. Molecular docking studies highlighted two compounds, MORLD5036 and MORLD6305, with exceptional binding affinities and interactions with key WEE1 active site residues. Molecular dynamics simulations and MM-GBSA binding free energy calculations further validated MORLD5036 as the most stable and potent inhibitor. Scaffold analysis revealed novel chemotypes distinct from existing inhibitors, enhancing potential for intellectual property. Comprehensive ADME profiling confirmed favorable pharmacokinetics, while synthetic accessibility evaluations indicated practicality for experimental validation. The identified lead compound, MORLD5036, exhibits favorable pharmacokinetics and novel chemotypes, positioning it as a potential therapeutic candidate for cancers reliant on WEE1-mediated cell cycle control. This integrated, AI-driven pipeline expedites the identification of next-generation WEE1 inhibitors, paving the way for advancements in precision oncology. Unlike traditional methods reliant on pre-existing datasets, this study leverages MORLD's reinforcement learning framework to autonomously generate inhibitors, enabling exploration of uncharted chemical space. These findings establish MORLD5036 as a computationally promising WEE1 inhibitor candidate warranting further experimental validation.
癌症中细胞周期的失调突出了靶向WEE1激酶的治疗潜力，WEE1激酶是G2/M检查点的关键调节因子。本研究利用人工智能（AI）驱动的方法，特别是MORLD平台，来识别新型WEE1抑制剂。以临床验证的WEE1抑制剂为参考，我们生成了20000种结构多样的化合物，这些化合物在结合亲和力、合成可及性和类药性方面进行了优化。一个严格的化学信息学流程——包括PAINS筛选、物理化学性质评估和分子指纹识别——将这个文库精炼为242个有前景的候选化合物。使用UMAP进行降维和通过K均值聚类能够对结构独特的先导化合物进行优先级排序。分子对接研究突出了两种化合物，MORLD5036和MORLD6305，它们具有出色的结合亲和力以及与WEE1关键活性位点残基的相互作用。分子动力学模拟和MM-GBSA结合自由能计算进一步验证了MORLD5036是最稳定和最有效的抑制剂。支架分析揭示了与现有抑制剂不同的新型化学类型，增强了知识产权潜力。全面的ADME分析证实了良好的药代动力学，而合成可及性评估表明了实验验证的实用性。所鉴定的先导化合物MORLD5036具有良好的药代动力学和新型化学类型，使其成为依赖WEE1介导的细胞周期控制的癌症的潜在治疗候选物。这种集成的、由AI驱动的流程加快了下一代WEE1抑制剂的识别，为精准肿瘤学的进展铺平了道路。与依赖现有数据集的传统方法不同，本研究利用MORLD的强化学习框架自主生成抑制剂，从而能够探索未知的化学空间。这些发现确立了MORLD5036作为一种在计算上有前景的WEE1抑制剂候选物，值得进一步的实验验证。
