【作者】 张文；

【导师】 瞿昆；

【作者基本信息】 中国科学技术大学 ， 生物信息学， 2022， 博士

【摘要】 视网膜是中枢神经系统的一部分,由多种类型的神经元(感光细胞、水平细胞、双极细胞、无长突细胞和视网膜神经节细胞)和胶质细胞组成,负责处理视觉信息。解析人类胚胎视网膜发育过程的分子机制对于视网膜疾病的预防和治疗是极其重要的。目前视网膜发育的研究大部分是基于模式动物,而有关人类胚胎视网膜发生的分子机制研究仍很少。近年来,高通量测序技术的发展为研究者们全面揭示器官的发育过程提供了可能。现有的组学研究已经描绘了人视网膜发育过程的基因表达和组蛋白修饰图谱,然而领域内仍缺乏对于人类胚胎视网膜发育的转录调控机制研究。随着体外分化技术的成熟,视网膜类器官为研究人类胚胎视网膜发育过程提供了新的模型,然而这两种分化发育过程在转录和染色质层面的异同仍未知。此外,基因表达具有时空特异性,构建视网膜细胞的空间表达图谱对于理解人类胚胎视网膜发育过程至关重要。因此本研究利用多组学技术,包括染色质可及性测序技术(ATAC-seq,scATAC-seq),RNA测序技术(RNA-seq,scRNA-seq),空间转录组技术(Spatial Transcriptomics)等,并通过整合表观基因组和转录组学数据揭示人类胚胎视网膜发育的时空转录调控机制。本论文的内容主要分为以下两个部分。第一部分工作是利用ATAC-seq研究人类胚胎视网膜发育过程的转录调控机制。本研究收集了不同发育时间点的人类胚胎视网膜和视网膜类器官样品,进行ATAC-seq和RNA-seq建库测序,系统、全面地揭示了人类胚胎视网膜与视网膜类器官在分化发育过程中染色质开放和基因表达的动态变化。研究结果显示视网膜类器官分化过程的染色质开放特征和转录特征与人类胚胎视网膜发育过程极为相似,表明视网膜类器官是研究人类胚胎视网膜分子机制的良好模型。此外,研究发现人类胚胎视网膜发育中期与神经分化相关的染色质区域处于双价修饰状态(H3K27me3和H3K4me3),而该区域在视网膜类器官中存在延后开放的现象。通过ATAC-seq数据与RNA-seq数据的整合分析,本研究鉴定了两个参与调控人类胚胎视网膜发育的转录因子NFIB和THRA,并利用视网膜类器官进行了功能验证。最后,我们重建了视网膜发育不同时期的转录调控网络。进一步,本研究下载了后续已发表的人类胚胎视网膜发育过程不同时间点的单细胞ATAC-seq数据,描绘了不同视网膜细胞亚群的染色质开放模式,并鉴定了细胞类型特异性的转录调控因子。此外,视网膜组织具有精细的空间结构,不同亚型的细胞在空间上有序排列,形成三个核层和两个突触层。为了研究视网膜细胞的空间发育图谱,在第二部分工作中,我们利用空间转录组技术研究人类胚胎视网膜发育过程中细胞和基因的空间分布模式。然而目前的空间转录组技术存在基因通量小(只能探测到100-1000个RNA分子)或空间分辨率低(每个空间位置点含有多个细胞)的缺陷,使得研究者难以获得单个细胞的完整转录组信息。为了克服空间转录组技术的局限性,研究者开发了许多整合空间转录组和单细胞转录组数据的算法工具,来预测基因或者细胞亚型的空间分布模式。为了选择合适的整合算法对视网膜空间组学数据进行分析,本研究设计了一套标准流程,并收集了 45对真实的空间转录组和单细胞转录组数据集,以及32对模拟数据集,用于比较16种整合算法在预测基因或者细胞空间分布方面的性能。之后,本研究收集了第11周和14周的人类胚胎视网膜样本进行空间转录组测序,并使用预测性能最优的Tangram和Cell2location这两个算法工具对视网膜空间转录组数据进行整合分析。本研究揭示了视网膜发育相关基因及细胞亚型的空间分布模式,并鉴定了多个视网膜空间结构域。最后,本研究通过拟时序分析构建了视网膜细胞的空间分化轨迹。综上所述,该项研究利用生物信息学方法全面揭示了人类胚胎视网膜发育和视网膜类器官分化过程中染色质可及性的动态变化过程;绘制了人类胚胎视网膜发育过程中调节元件和转录因子结合的模式图谱;鉴定了调控人类胚胎视网膜发育的转录因子NFIB和THRA,并利用视网膜类器官进行了功能验证;构建了人类胚胎视网膜发育的转录调控网络;鉴定了细胞类型特异性的染色质开放区域和转录因子结合模式;开发了一套无偏的,系统性的评估流程可用于评估空间转录组和单细胞转录组数据整合算法的性能;揭示了人类胚胎视网膜组织中关键基因和细胞类型的空间分布模式。该研究填补了领域内对人类胚胎视网膜转录调控研究的空缺,为今后研究人类胚胎视网膜发育以及改进视网膜类器官的分化系统提供了重要的研究基础和宝贵的数据资源。更多还原

【Abstract】 The retina is part of the central nervous system and consists of various types of neurons(photoreceptor cells,horizontal cells,bipolar cells,amacrine cells,and retinal ganglion cells)and glial cells,responsible for processing visual information.Deciphering the molecular mechanisms of human embryonic retinal development is important for retinal disease prevention and treatment.Most of the current research on retinal development is based on model animals,yet little is known about the molecular mechanisms of human embryonic retinogenesis.The development of high-throughput sequencing technology in recent years has made it possible for us to systematically and comprehensively study the process of organ differentiation and development.Existing omics studies have revealed gene expression and histone epigenetic maps in human retinal development,while the transcriptional regulatory mechanisms during human embryonic retinal development remain poorly understood.In addition,retinal organoids formed by human induced pluripotent stem cells in vitro provide a new model for studying human embryonic retinas.However,the similarities and differences between these two differentiation processes at the transcriptional and chromatin levels remain unclear.In addition,gene expression is spatiotemporally specific,and the construction of spatiotemporal expression profiles for human retina is crucial for us to understand the human embryonic retinal development.Therefore,we used ATAC-seq,scATACseq,Spatial Transcriptomics,and other multi-omics technologies to integrate epigenome and transcriptomics data to decipher the transcriptional regulation network in human embryonic retinal development.The content of this thesis is mainly divided into two parts.The first part of the work is to use ATAC-seq technology to study the transcriptional regulation mechanism of human embryonic retinal development.To systematically reveal the landscape of chromatin accessibility and gene expression in developing human retina and retinal organoids,we collected ATAC-seq and RNA-seq data of human embryonic retina and retinal organoids samples at different developmental time.Our results showed that the chromatin accessibility and transcriptional state of the retinal organoid differentiation process were similar to those of human embryonic retinal development,suggesting that retinal organoids would be a good model for studying the molecular mechanisms of human embryonic retina.Moreover,we found that the region related to neural differentiation in the middle stage of human embryonic retinal development was in a bivalently modified state(H3K27me3 and H3K4me3),and this region had a phenomenon of delayed opening in retinal organoids differentiation.Through integrated analysis of RNA-seq data and ATAC-seq data,we identified transcription factors NFIB and THRA as necessary regulators for human retinal development,and validated by retinal organoids.Finally,we reconstructed the transcriptional regulatory network at different stages of retinal development.Further,we collected published single-cell ATAC-seq data during human embryonic retinal development to identified the cell-type-specific chromatin open regions and transcriptional regulators of human embryonic retina at different developmental stages.In addition,human retina has a complex spatial structure,with cells spatially arranged to form three nuclear layers and two synaptic layers.To study the spatial development map of retinal cells,in the second part of the work,we used spatial transcriptomic technology to study the spatiotemporal expression profiles during retinal development.Yet,the current mainstream spatial transcriptome technologies have shortcomings such as low gene throughput(only a small number of RNA molecules can be detected)or low resolution(each spot contains multiple cells),respectively,making it difficult for researchers to obtain the complete transcriptome information of a single cell.To overcome the limitations of these spatial transcriptome technologies,researchers have developed many intergration methods that combing spatial transcriptome and single-cell transcriptome data to predict the spatial distribution patterns of genes or cellular subtypes.To choose appropriate integration methods for spatial data analysis,we designed a standard pipeline to compare the performance of 16 integrated algorithms in predicting the spatial distribution of genes or cells with 45 pairs of real spatial transcriptome and single-cell transcriptome datasets,and 32 pairs of simulated datasets.Afterwards,we obtained a spatiotemporal overview of human embryonic retina at gestation weeks 11 and 14,and performed an integrative analysis of spatial transcriptome data using Tangram and Cell2location.Our research revealed the spatial distribution of genes and retinal cell types,and systematically studied the functions of different spatial domains in the human retina by identifying functional spatial regions.Finally,we constructed the spatial differentiation trajectories of retinal cells by pseudotime analysis.In summary,this study used bioinformatics methods to comprehensively demonstrate the dynamic changes of chromatin accessibility during human embryonic retinal development and retinal organoids differentiation;draw the map of regulatory element and transcription factor binding during human embryonic retinal development;identified transcription factors NFIB and THRA as necessary regulators for human retinal development,and validated by retinal organoids;constructe the transcriptional regulatory network for human embryonic retinal development;identificatr the celltype-specific chromatin open regions and transcription factors;developed an unbiased,systematic evaluation pipeline for evaluating the performance of intergration methods for spatial transcriptome and single-cell transcriptome data;and reveal spatial distribution patterns of key genes and cell types for human embryonic retinal development.It fills the gap in the research on epigenetic regulation of human retinal development in the field,and provides an important research basis and valuable data resources for future research on human embryonic retinal development and improvement of retinal organoids differentiation systems.更多还原