CAREER: Constructing Multi-scale Dynamical Ensembles of Ribonucleic Acids (RNAs)

核糖核酸(RNA)对不同结构进行采样的能力是它们许多生物学作用的基础。许多调节细菌、病毒和人类基因表达的RNA是通过改变其结构来对刺激做出反应的。例如，在细菌中，RNA元素在特定代谢物存在的情况下改变其分子结构，如维生素B12。这些代谢物的结合和随后的结构变化使表达这些RNA的有机体能够对环境做出反应，并根据需要开启或关闭基因表达。然而，可视化RNA采用的结构仍然是分子生物物理学中的一个基本挑战。该项目寻求开发和应用新的方法来整合实验和计算数据来描述RNA的结构。该项目还将紧密结合研究和教育，以培养高中、本科生和研究生应用计算分子生物物理学的基础知识。在这个项目中，学生们将利用现代机器学习技术，类似于面部识别软件中使用的技术，从生物物理测量中提取RNA结构信息。外展活动的目的将是让高中生积极参与计算密集型研究，特别是那些来自历史上招募不足的群体的学生。具体地说，该项目旨在开发确定动态系综的方法，即给定RNA可以填充的构象状态的集合。为了了解RNA是如何折叠和发挥功能的，我们必须确定它们的系谱，以及它们在识别蛋白质或配体期间或在引入突变或化学修饰期间如何变化。然而，表征RNA的动态系综仍然是一项艰巨的任务，因为未知变量的数量超过了可以通过实验测量的数量。此外，许多构象以低丰度存在，寿命短，因此很难通过实验检测到。更具挑战性的是确定系综中不同的构象如何相互转换，因为这需要描述异常低密度的过渡态和复杂的路径，这些路径可能很难梳理出来。在这个项目中，将使用新的综合方法来确定RNA的动态系综，这些方法是使用现成的实验数据进行初始化的。在这个项目中，核磁共振化学位移将被用来指导RNA的模拟并确定它们的动力学系综。因此，第一个目标是发展多尺度的方法来从RNA的结构中估计化学位移。第二个目标是利用多尺度化学位移预测方法来构建RNA的多尺度动态集成。最后一个目标是开发一种方法，使用多尺度动态集成来构建RNA的过渡路径集成。最终，该项目将导致稳健、简化和集成的方法，超越静态结构确定，以构建动态集成，捕获功能RNA可访问的构象状态范围，同时描述这些状态之间的转换。这种结构信息将把对RNA驱动的生化过程的理解从描述性提高到预测性。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The ability of ribonucleic acids (RNAs) to sample distinct structures underlie many of their biological roles. Many RNAs that regulate gene expression in bacteria, viruses, and humans do so by changing their structure in response to stimuli. In bacteria, for instance, RNA elements switch their molecular structure in the presence of specific metabolites, such as vitamin B12. The binding of such metabolites and the subsequent structural changes enable organisms that express these RNAs to respond to their environment and turn ON or OFF gene expression as needed. However, visualizing the structures that RNAs adopt remains a fundamental challenge in molecular biophysics. This project seeks to develop and apply new methods to integrate experimental and computational data to describe the structure of RNAs. This project will also tightly integrate research and education to train high school, undergraduate, and graduate students in the fundamentals of applied computational molecular biophysics. During this project, students will leverage modern machine learning techniques, similar to techniques used in facial recognition software, to extract RNA structural information from biophysical measurements. Outreach activities will be aimed at actively engaging high school students in computationally-intensive research, especially those from historically under-recruited groups. Specifically, this project aims to develop methods to determine the dynamical ensemble, i.e., the collection of conformational states that a given RNA can populate. To understand how RNAs fold and function, one must determine their ensembles and how they change during the recognition of proteins or ligands or during the introduction of mutations or chemical modifications. Characterizing dynamical ensembles of RNA, however, remains a formidable task because the number of unknown variables exceeds that which can be measured experimentally. Additionally, many conformers exist in low-abundance and have short lifetimes and are therefore difficult to detect experimentally. Even more challenging is determining how different conformers in an ensemble interconvert as this requires characterization of exceptionally lowly-populated transition states and complex pathways that can be difficult to tease out. In this project, the dynamical ensembles of RNAs will be determined using novel integrative approaches that are initialized using readily available experimental data. In this project, NMR chemical shifts will be used to guide simulations of RNA and to determine their dynamical ensembles. Therefore, the first objective is to develop multi-scale methods to estimate chemical shifts from structures of RNA. The second objective is to leverage multi-scale chemical shift prediction methods to construct multi-scale dynamical ensembles of RNA. The last objective is to develop a methodology that uses multi-scale dynamical ensembles to construct transition-path ensembles of RNA. Ultimately, this project will lead to robust, streamlined, and integrated methodologies that go beyond static structure determination to construct dynamical ensembles that capture the range of conformational states accessible to functional RNAs, while simultaneously describing the transitions between these states. This structural information will advance the understanding of RNA-driven biochemical processes from descriptive to predictive.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.