Decoding the mammalian regulatory genome

尽管人类基因组序列在16年前就完成了，但我们目前还无法确定该序列中绝大多数的功能。我们知道如何阅读基因编码区内的序列密码，但这些基因组的98%是基因调控元件，在特定细胞中打开/关闭基因。这些调控元素是特定细胞类型的指令。例如，它们导致大脑、血液和心脏细胞中不同组基因的激活。我的研究是研究一种被称为增强剂的激活型调控元件。增强子是由大约500个碱基组成的短序列，当特定基因的蛋白质产物需要时，它们会启动特定的基因。尽管我们知道增强子对复杂生物体的正常发育很重要，而且往往是不同物种身体规划的具体差异的原因，但我们实际上对赋予它们功能的序列密码知之甚少。当我们知道增强子的位置时，我们无法扫描基因组序列来寻找增强子或解释序列不同部分的功能。这就像是用一种你看不懂的语言写的书。或者，因为我们已经对100多种哺乳动物的基因组进行了完整的测序，整个图书馆的书籍都是你不能阅读的。我的团队最近的工作确定了关于增强子监管代码的新信息。我们使用计算生物学来预测多能干细胞(PSC)增强子中的保守功能单位，方法是通过对5种哺乳动物的序列保守。我们发现，有更多的序列单位对增强子活性做出贡献，这比之前所理解的要多。需要10个以上的结合转录因子的位点才能产生强大的增强子活性。我们现在对PSCs的代码有了足够的了解，可以生成功能类似于天然增强剂的合成序列；我们可以编写自己的句子。在接下来的5年里，我的研究团队将扩展我们在一种细胞类型中开发的方法，以预测多种不同细胞类型的增强子调控代码。我们将解决三个关键问题：1)为什么增强子活性需要如此多的转录因子结合位点？2)不同细胞类型的调控密码有何不同？3)完全解码哺乳动物基因组的最佳计算方法是什么？从长远来看，完全解码基因组序列将使我们能够预测调控序列的变化将如何影响特定基因的表达。这将对我们理解种群的发育、进化和表型变异的根本原因产生广泛的影响。

Despite completion of the human genome sequence 16 years ago we are currently not able to determine the function of the vast majority of this sequence. We understand how to read the sequence code within gene coding regions but these are 98% of the genome are gene regulatory elements that turn genes on/off in specific cells. These regulatory elements are the instructions for specific cell types. They cause the activation of different groups of genes in brain, blood and heart cells, for example. My research studies an activating type of regulatory element called enhancers. Enhancers are short sequences of about 500 bases that turn on specific genes when their protein product is needed. Although we know that enhancers are important for the proper development of complex organisms and are often the cause of specific differences in the body plan of different species we actually know remarkably little about the sequence code that gives them their function. We are not able to scan the genome sequence to find enhancers or interpret the function of different parts of the sequence when we do know the location of an enhancer. This is like having a book written in a language you can't read. Or, as we have fully sequenced genomes for more than 100 mammals, an entire library of books that you can't read. Recent work by my team has determined new information about the enhancer regulatory code. We used computational biology to predict the conserved functional units in pluripotent stem cell (PSC) enhancers using sequence conservation across 5 mammals. We found there is a larger number of sequence units contributing to enhancer activity than was previously appreciated. More than 10 sites that bind transcription factors are required for robust enhancer activity. We now understand enough of the code for PSCs to make synthetic sequences with function similar to natural enhancers; we can write our own sentences. Over the next 5 years my research team will expand the approach we developed in one cell type to predict the enhancer regulatory code for multiple different cell types. We will address three key questions: 1) Why are so many transcription factor binding sites needed for enhancer activity? 2) How is the regulatory code different in different cell types? 3) What is the best computational approach to fully decode mammalian genomes? In the long term fully decoding genome sequences will allow us to predict how regulatory sequence changes will affect the expression of specific genes. This will have widespread impact on our understanding of development, evolution and the underlying cause of phenotype variation in populations.