Control of RNA function by conformational design

在过去的二十年里，RNA已经成为一个主要的研究焦点。核酶、核糖开关和小的非编码RNA参与了大量的细胞过程，这使得人们对RNA的结构和构象如何与其功能联系在一起以及如何通过构象设计来控制RNA功能产生了浓厚的兴趣。因此，很明显，更复杂的设计目标不适用于人工序列设计，而需要使用基于计算机的优化方法。另一方面，由于能量模型中即使是很小的误差也可以显著改变不同构象之间的平衡，计算RNA结构预测往往不够准确，不足以立即产生功能RNA分子。然而，计算模型非常有效地能够将搜索空间减少到可以通过实验探测的少数候选对象。通过将计算机辅助预测与湿实验室实验相结合，我们的目标是开发用于设计功能RNA分子的协议和软件工具。特别是，将对三个系统进行评估：(I)自剪接和(Ii)来自发夹核酶的自复制RNA，以及(Iii)自诱导RNA开关。各个系统的设计将建立在以前的实验工作的基础上，并将通过对RNA折叠动力学、RNA复合体的形成和构象分布的计算预测来强有力地指导和/或优化，作为成功解决更复杂设计问题的基本工具。反过来，对理论预测的实验测试将使我们能够确定计算模型的缺陷。将使用反复的建模和测试来反复改进模型设置和参数。

Over the past two decades RNA has become a major focus of research. The discovery of ribozymes, riboswitches and small noncoding RNAs being involved in a large number of cellular processes has led to much interest in the understanding of how RNA structure and conformation are linked to its function and how RNA function can be controlled by conformational design. Accordingly, it has become evident that more complex design targets are not amenable to manual sequence design, but require the use of computer based optimization methods. On the other hand, since even small errors in the energy model can significantly shift the balance between alternative conformations, computational RNA structure prediction is often not accurate enough to yield immediately functional RNA molecules. The computational models are, however, very efficiently able to reduce the search space to a small number of candidates that can be probed experimentally. By combining computer-aided prediction with wet lab experiments, we aim at developing protocols and software tools for the design of functional RNA molecules. In particular, three systems will be evaluated: (i) self-splicing and (ii) self-replicating RNAs derived from the hairpin ribozyme, and (iii) self-induced RNA switches. Design of the respective systems will build up on previous experimental work, and will be strongly guided and/or optimized by computational prediction of RNA folding kinetics, RNA complex formation and conformational distribution as an essential tool to successfully solve more complex design problems. In turn, experimental testing of the theoretical predictions will allow us to identify shortcomings of the computational models. Repeated rounds of modelling and testing will be used to iteratively refine model setup and parameters.