乳腺癌瘤内微环境表观遗传异质性的系统分析及新型分子特征挖掘

Different microenvironments (sub-regions) in the same tumor share the same genetic code, but show epigenetic heterogeneity. Intratumor epigenetic heterogeneity is closely related to tumor heterogeneity and leads to drug-resistant, metastasis and poor prognosis of tumors. The system analysis of intratumor epigenetic heterogeneity would be helpful for understanding of its impact on biological behavior and clinical phenotypes of cancer. In this project, we use the HER2-positive breast cancer as a model, and apply high-throughput sequencing technologies to map whole genome wide DNA methylome, exome and transcriptome of different sub-regions of breast tumor. Bioinformatic tools are developed based on Shannon entropy and statistic model to identify intratumor heterogeneity related multidimensional omics changes across different sub-regions and map the profile of molecular features. Using the Logistics regress, we identify the genetic and epigenetic risk factors associated with intratumor hypoxia microenvironments. The maximum likelihood model is trained to optimize the epigenetic regulatory markers which drive intratumor heterogeneity. These markers would be further evaluated by molecular biology experiments. At last, all informatic algorithms and software are integrated to construct an integrated platform for identification of epigenetic heterogeneity markers across intratumor microenvironments and prognosis. Results of this project will not only help to further understanding of tumor heterogeneity, but also provides an important reference for molecular targeted therapy of cancer.

同一肿瘤内部共享相同遗传密码的不同微环境（部位）间存在表观遗传异质性，与肿瘤异质性密切相关，会导致肿瘤的抗药性、转移及预后不良。肿瘤内微环境间表观遗传异质性的系统分析有利于理解瘤内异质性对癌症生物学行为及临床表型的影响。本项目以预后不良的HER2阳性乳腺癌为模型，应用高通量测序技术测定瘤内不同部位的全基因组DNA甲基化组、外显子组及全RNA转录组，开发信息熵和统计学算法从多维组学角度筛选瘤内异质性相关的差异并绘制分子特征图谱；利用Logistics回归筛选肿瘤乏氧微环境相关的遗传/表观遗传风险因子，训练最大似然模型优选驱动瘤内异质性的表观遗传调控标记，并利用分子生物学实验开展表观遗传标记的验证。最后，通过集成本课题开发的信息学算法和软件构建整合的肿瘤微环境表观遗传异质性标记识别及预后评估技术平台。本项目的研究成果不仅有利于对肿瘤异质性的进一步理解，也为肿瘤的分子靶向治疗提供重要参考。

肿瘤内的异质性反映了癌细胞和肿瘤微环境（如乏氧）之间相互作用的潜在细胞和分子机制，癌细胞亚群的存在使肿瘤内或与转移部位之间在遗传、表型及行为特征上出现了差别，通过深入挖掘瘤内异质性的特征及受肿瘤微环境影响的分子水平改变将有利于精确判断肿瘤性质并找到有效及持久的治疗方法。作为一种异质性很强的肿瘤，不同的乳腺癌只有根据其遗传和表观遗传特征进行个体化治疗，才能解决目前乳腺癌治疗的困境。本项目通过新一代测序技术绘制 HER2 阳性乳腺癌乏氧微环境相关的瘤内表观遗传异质性分子景观图谱。系统分析肿瘤微环境的差别与遗传/表观遗传的结构区域特征间的关系, 识别乳腺癌瘤内异质性的表观遗传特异性标记。我们利用TCGA数据库中的669例乳腺癌样本，探索了DNA甲基化特征性的特异性预后亚型。通过使用对生存有显著影响的3869个CpG位点进行一致性聚类获得了九个DNA甲基化亚组，揭示出DNA甲基化分类可以解释乳腺癌分子亚群的异质性及预后的差别并有助于开发新的特定亚型个体化治疗的靶标。项目通过整合DNA甲基化和拷贝数变异分析，探讨了基于拷贝数变异的CIMP与抗癌药物反应的关系，并通过评估肿瘤样本的细胞系组成推测癌症患者对某些抗肿瘤药物的反应。结果发现由HCC1419细胞系组成的拷贝数扩增的CIMP-H乳腺癌肿瘤样本对药物AKT抑制剂GSK690693是敏感的。同时，通过本项目开发的数据库SEA和DiseaseMethversion2.0可以查询到乳腺癌相关的调控元件，为乳腺癌异质性分析提供帮助。本项目中开发的信息学算法和软件提供了整合的肿瘤微环境表观遗传异质性标记识别及预后评估技术平台。