Charting the regulatory topography of the cell differentiation landscape with single-cell RNA-Seq.

 描述（由申请人提供）：单个干细胞产生一系列高度特化的成体细胞类型，以响应来自周围组织环境的精心调节的分子线索。然而，最近的证据对将成体细胞分类为离散类型提出了挑战。例如，免疫细胞可能更好地被描述为一个巨大的，连续的功能“景观”的居民。为了解决这些模型是否正确，我们必须能够测量单个细胞的完整基因表达谱，并观察它们在分化过程中在不同功能状态之间的移动。该提案旨在回答生物学中的一个基本问题：在细胞分化期间，基因表达景观有多连续？我假设单细胞转录组测序（RNA-Seq）可以用于直接可视化分化细胞所穿过的景观，并且跟踪细胞穿过它的路径将揭示控制细胞分化的基因调控网络。许多研究小组试图从微阵列或RNA-Seq获得的全局转录组测量结果中通过算法推断基因调控网络。不幸的是，从大量细胞表达数据推断调控网络的计算方法可能受到辛普森悖论的阻碍，这破坏了算法从表达数据准确重建网络所需的关键变异来源。辛普森悖论描述了当两个或两个以上的群体混合在一起时，一种趋势是如何改变或完全消失的。在细胞分化的时间序列表达分析中，辛普森悖论经常完全掩盖从一个状态到下一个状态的过渡期间发生的表达变化，因为每个批量测量包含两种状态的混合物。我们最近开发了一种名为Monocle的新算法，它构成了基因表达数据分析的重大突破，因为它克服了使用单细胞RNA-Seq的辛普森悖论。我将利用Monocle提供的新的调控信息来源，开发一种算法，重建转录景观和控制细胞分化的主动调控途径。然后，我将分析单核细胞来源的细胞谱系作为一个模型系统，用于辨别细胞是否落入离散状态，并解剖调节它们之间转换的途径。最近对单核细胞分化的分析表明，这一谱系比以前认为的更具可塑性，并且不那么明确，并且这种可塑性被怀疑有助于许多常见疾病，包括动脉粥样硬化，克罗恩病和多发性硬化。为了将我的方法扩展到非细胞自主调节，我将研究间充质干细胞如何通过细胞-细胞信号传导阻断单核细胞产生树突状细胞。如果成功的话，这里提出的研究将不仅提供一个控制单核细胞分化的途径图，而且提供一个通用的策略，用于阐明基因调控网络，这些网络控制着几乎任何组织细胞中的一系列动态过程。

 DESCRIPTION (provided by applicant): A single stem cell generates a staggering array of highly specialized adult cell types in response to carefully regulated molecular cues from the surrounding tissue environment. However, recent evidence has challenged the classification of adult cells into discrete types. Immune cells, for example might be better described as inhabitants of a vast, continuous functional "landscape". In order to resolve whether either of these models are correct, we must be able to measure the complete gene expression profile of individual cells and observe them moving between different functional states during differentiation. This proposal aims to answer a fundamental question in biology: how continuous is the gene expression landscape during cell differentiation? I hypothesize that single-cell transcriptome sequencing (RNA-Seq) can be used to directly visualize the landscape traversed by differentiating cells, and that tracking the paths cells take across it will reveal the gene regulatory networks that govern cell differentiation. Many groups have tried to algorithmically infer gene regulatory networks from global transcriptome measurements obtained with microarrays or RNA-Seq. Unfortunately, computational methods for inferring regulatory networks from bulk cell expression data have likely been hamstrung by Simpson's paradox, which destroys the crucial source of variation that an algorithm needs to accurately reconstruct networks from expression data. Simpson's paradox describes how a trend present in two or more groups of individuals changes or disappears entirely when those groups are mixed together. In time series expression analyses of cell differentiation, Simpson's paradox often completely obscures changes in expression that occur during a transition from one state to the next, because each bulk measurement contains a mixture of both states. We recently developed a new algorithm called Monocle that constitutes a major breakthrough in the analysis of gene expression data because it overcomes Simpson's paradox using single- cell RNA-Seq. I will exploit the new sources of regulatory information that Monocle makes accessible to develop an algorithm that reconstructs both the transcriptional landscape and the active regulatory pathways governing cell differentiation. I will then analyze the monocyte-derived cell lineage as a model system for discerning whether cells fall into discrete states and dissecting the pathways regulating transitions between them. Recent analyses of monocyte differentiation have suggested that this lineage is far more plastic and less sharply defined than previously believed, and this plasticity is suspected to contribute to many common diseases, including atherosclerosis, Crohn's disease, and multiple sclerosis. To extend my approach to non-cell- autonomous regulation, I will investigate how mesenchymal stem cells block dendritic cell generation from monocytes through cell-cell signaling. If successful, the research proposed here will provide not just a map of pathways governing monocyte differentiation, but a general strategy useful for illuminating the gene regulatory networks that govern a wide array of dynamic processes in cells of nearly any tissue.