摘要
目的本研究旨在利用人工智能(artificial intelligence,AI)技术,对La蛋白抑制剂候选化合物4424-1120的成药性、代谢途径及药动学特性进行全面预测和评估,为其进一步开发提供理论依据。方法首先确定La蛋白的结构及其关键结合位点;然后,基于La蛋白的X射线晶体结构,采用Schr9dinger软件的Maestro模块进行虚拟筛选,筛选出具有潜在抑制活性的化合物;最后,采用AI成药性预测平台(optADMET)对筛选得到的候选化合物4424-1120进行了成药性和代谢预测,包括理化性质计算、代谢途径模拟、毒性评估以及药动力学参数和转运体相互作用预测,并进行了综合成药性风险评估。结果化合物4424-1120在理化性质预测中显示出随pH变化的显著特征,其溶解度在pH 6~9范围内较高(最大值58.32 mg·mL^(-1)),分配系数在中性及碱性条件下趋于稳定。酸解离常数(acid dissociation constant,pKa)预测结果显示化合物具有3个主要pKa值(2.17、3.81和9.38),在胃肠道pH范围内主要以pKa_Mol_3和pKa_Mol_2形式存在,其中在pH值6~8范围内pKa_Mol_2占比高达96.55%。代谢预测表明,CYP2C9和CYP3A4共同介导其主要代谢产物的生成,CYP_Mol_1和CYP_Mol_4为主要代谢产物,各占代谢总量的28%。此外,UGT1A1和UGT1A8参与其葡糖醛酸化代谢。毒性预测显示化合物在Ames试验和染色体畸变预测中均为阴性,无人类乙型延迟整流钾通道(human ether-à-go-go-related gene,hERG)抑制、致敏性、光毒性风险,急性毒性[半数致死量(lethal dose 50%,LD50)值为486.21 mg·kg^(-1)]和亚慢性毒性[半数毒性剂量(toxic dose 50%,TD50)值为每天44.26 mg·kg^(-1)]较低,可能存在肝毒性风险。本研究还通过图像展示了候选化合物4424-1120的潜在毒性位点,直观反映了其结构与毒性风险之间的关系,为后续优化提供了参考依据。药动学预测显示化合物口服吸收迅速(吸收率99.99%,生物利用度94.64%),血浆浓度在2.48 h达到峰浓度1230.86 ng·mL^(-1),半衰期为12.05 h。综合成药性分析表明,化合物成药性风险低,具有进一步开发潜力。结论AI技术在La蛋白特异性抑制剂的成药性预测中展现出高效性和准确性,为候选化合物4424-1120的评估提供了重要支持。4424-1120具备良好的成药性特征和较低的毒性风险,药动学特性优良,为后续实验验证和新药研发奠定了基础。
OBJECTIVE To utilize artificial intelligence(AI)technology to comprehensively predict and evaluate the druggability,metabolic pathways,and pharmacokinetic properties of the La protein inhibitor candidate compound 4424-1120,providing a theoretical basis for its further development.METHODS The structure of La protein and its key binding sites were first determined.Based on the X-ray crystal structure of La protein,the Maestro module of Schrödinger software was used for virtual screening to identify compounds with potential inhibitory activity.Finally,the optADMET platform developed by Tianzhi Yaocheng was employed to predict the druggability and metabolism of the candidate compound 4424-1120,including physicochemical property calculations,metabolic pathway simulations,toxicity assessments,pharmacokinetic parameters,and transporter interaction predictions.A comprehensive druggability risk assessment was also conducted.RESULTS The physicochemical property predictions showed that compound 4424-1120 exhibited significant solubility characteristics that varied with pH,with high solubility in the pH range of 6-9(maximum of 58.32 mg·mL^(-1)).The distribution coefficient remained stable under neutral and alkaline conditions.Acid dissociation constant(pKa)predictions revealed three main pKa values(2.17,3.81,and 9.38),with the compound existing primarily as pKa_Mol_3 and pKa_Mol_2 in the gastrointestinal pH range,and pKa_Mol_2 accounting for up to 96.55%at pH 6-8.Metabolism predictions showed that CYP2C9 and CYP3A4 jointly mediated the formation of its main metabolites,CYP_Mol_1 and CYP_Mol_4,each accounting for 28%of total metabolism.Additionally,UGT1A1 and UGT1A8 were involved in glucuronidation metabolism.Toxicity predictions indicated that the compound was negative in Ames tests and chromosomal aberration predictions,with no human ether-à-go-go-related gene(hERG)channel inhibition,sensitization,or phototoxicity risks.Acute toxicity(lethal dose 50%,LD50:486.21 mg·kg-1)and subchronic toxicity(toxic dose 50%,TD50:44.26 mg·kg·d^(-1))were low,but there is a potential risk of hepatotoxicity.The study also visualized the potential toxicity-related regions of compound 4424-1120,directly reflecting the relationship between its structure and toxicity risks,providing reference points for further optimization.Pharmacokinetic predictions showed that the compound is rapidly absorbed orally(absorption rate 99.99%,bioavailability 94.64%),with peak plasma concentration(ρmax)of 1230.86 ng·mL^(-1) reached at 2.48 h and a half-life(t1/2)of 12.05 h.Comprehensive druggability analysis indicated low druggability risk for the compound,suggesting its potential for further development.CONCLUSION AI technology demonstrates efficiency and accuracy in predicting the druggability of La protein-specific inhibitors,providing strong support for evaluating compound 4424-1120.Compound 4424-1120 exhibits favorable druggability characteristics,low toxicity risk,and excellent pharmacokinetic properties,laying a solid foundation for subsequent experimental validation and drug development.
作者
李俊慜
李月妍
刘才平
汤静
LI Junmin;LI Yueyan;LIU Caiping;TANG Jing(Obstetrics&Gynecology Hospital of Fudan University,Shanghai 200090,China;Children′s Hospital of Chongqing Medical University,National Clinical Research Center for Child Health and Disorders,Key Laboratory of Child Development and Disorders of the Ministry of Education,Chongqing 401122,China;Tianzhi Yaocheng Technology(Chongqing),Co.,Ltd.,Chongqing 401329,China)
出处
《中国药学杂志》
北大核心
2025年第15期1632-1640,共9页
Chinese Pharmaceutical Journal
基金
中华医学会临床药学分会2023年度临床药学科研基金项目资助(Z-2021-46-2101-2023)
上海市青浦区卫生健康系统第五轮学科带头人培养计划项目资助(XD2023-10)。
关键词
人工智能
LA蛋白
特异性抑制剂
成药性
预测
artificial intelligence
La protein
specific inhibitor
druggability
prediction
