论文部分内容阅读
目的以毒理基因组学方法建立预测遗传毒性致癌物与非遗传毒性致癌物的分类器,探索暴露时间对其预测能力的影响并验证其性能。方法原代小鼠肝细胞模型经2个遗传毒性致癌物黄曲霉素B1和苯并芘,2个非遗传毒性致癌物硫代乙酰胺和匹立尼酸处理24和48 h后,对差异表达基因运用基因芯片预测分析筛选出分类器。通过基因集富集分析研究分类器中基因的功能,并运用STRING数据库预测分类器中基因编码蛋白之间的相互关系。进一步运用2个额外的致癌物验证分类器的预测性能。最后还通过Quanti Gene Multiplex实验验证了基因芯片数据。结果经基因芯片预测分析筛选的48 h分类器优于24 h分类器,分类器中的基因涉及p53通路、肿瘤坏死因子-α信号通路、脂肪酸代谢相关基因集、过氧化物酶体增殖物激活受体通路等。分类器中的基因形成致癌蛋白-蛋白相互作用关系网络图和代谢相关蛋白-蛋白相互作用网络图。经验证48h分类器对2个额外的致癌物预测可能率接近100%,Quanti Gene Multiplex实验结果与芯片数据有较高的一致性。结论成功建立了预测分类器并验证其性能。该分类器可用于分辨潜在的遗传毒性致癌物和非遗传毒性致癌物,并对未知化合物可能的作用机制进行预测,有望成为药物非临床安全性评价致癌性试验体外替代方法之一。
Objective To establish a classifier to predict genotoxic carcinogens and non-genotoxic carcinogens by toxicological genomics and to explore the effect of exposure time on its predictive ability and to verify its performance. Methods The primary mouse hepatocyte model was treated with 2 genotoxic carcinogens aflatoxin B1 and benzopyrene, 2 non-genotoxic carcinogens thioacetamide and pirnidin for 24 and 48 h, Gene expression using gene chip prediction analysis of screening out the classifier. The gene function in the classifier was studied by enrichment of gene sets, and the relationship between the gene encoding proteins in the classifier was predicted by STRING database. Further use of 2 additional carcinogens to validate the predictive performance of the classifier. Finally, gene chip data were verified by Quanti Gene Multiplex experiments. Results The 48 h classifier screened by gene chip predictive analysis was better than the 24 h classifier. The genes involved in the classifier were p53 pathway, tumor necrosis factor-α signaling pathway, fatty acid metabolism related gene set, peroxisome proliferator activated Receptor pathway and so on. Genes in classifiers form oncogenic protein-protein interaction network maps and metabolic-related protein-protein interaction networks. The validated probability of 48h classifier predicting 2 extra carcinogens is nearly 100%, and the results of Quanti Gene Multiplex experiment are in good agreement with the data of the chip. Conclusion The predictive classifier has been successfully established and its performance verified. The classifier can be used to distinguish potential genotoxic carcinogens and non-genotoxic carcinogens and to predict the possible mechanism of action of unknown compounds, which is expected to become an in vitro alternative to nonclinical safety assessment of carcinogenicity of drugs.