High Resolution Free Energy Landscape Analysis of Protein Folding Dynamics

尽管人们已经对蛋白质折叠成天然状态的机制进行了多年的研究，但它仍然是生物学中一个重要的问题。此外，既然错误折叠已被证明是一系列疾病的根源，那么了解决定多肽链是否会折叠成其原始状态或聚合状态的因素就变得更加重要。虽然蛋白质折叠领域普遍被认为是成熟的，但在许多基本问题上仍存在许多争议，包括变性状态下残余结构的重要性，折叠步骤和动力学屏障的性质以及途径多样性的程度。这些问题可以用一种严格的方法来分析，将折叠动力学描述为自由能版图上的扩散。然而，尽管它们具有基本的重要性，但定量准确的蛋白质自由能格局尚未确定。最先进的实验技术缺乏必要的空间和时间分辨率。景观的性质只能间接地被探测。原则上，模拟可以提供（无限）高的时空分辨率，这是构建定量准确的自由能景观所必需的。最近，由于硬件和仿真方法的进步，小的快速折叠蛋白质的折叠的真实模拟在计算上变得负担得起。目前的记录保持者是Shaw等人在具有里程碑意义的论文中报道的模拟。本文对含有15个折叠展开事件的FIP35蛋白在显式水中的全原子平衡折叠进行了“蛮力”模拟，其折叠速率和固有结构与实验一致。更多的模拟正在进行中。这开启了一个新的激动人心的时代，当详细了解围绕蛋白质折叠动力学的有争议的问题成为可能。在这种模拟中获得的轨迹包含折叠动力学的详细信息，其大小为许多tb，通常对自动化分析提出了很大的挑战。此外，根据自由能景观对蛋白质折叠动力学进行精确的定量分析是出了名的困难。传统的方法，即使是基于坚实的物理直觉，往往导致次优结果与简单的自由能景观，隐藏固有的折叠动力学的复杂性。这样的景观，在本质上，通过模拟带走了最大的优势——丰富的详细信息。我们将应用新开发的严格方法来确定蛋白质折叠的高分辨率定量准确的自由能景观。稍后，该方法将扩展到基于单分子实验中记录的轨迹来确定这种景观。

Understanding the mechanism by which proteins fold to their native state remains a problem of fundamental interest in biology, in spite of the fact that it has been studied for many years. Moreover, now that misfolding has been shown to be the source of a range of diseases, a knowledge of the factors that determine whether a polypeptide chain will fold to its native state or aggregate has become all the more important. While the field of protein folding generally is considered to be mature, much controversy still remains on many fundamental topics including the importance of residual structure in denatured states, the nature of folding steps and kinetic barriers as well the extent of pathway diversity. These questions can be analysed in a rigorous way by describing the folding dynamics as diffusion on a free energy landscape. However, in spite of their fundamental importance the quantitatively accurate free energy landscapes of proteins are yet to be determined. The state of the art experimental techniques lack the necessary spatial and temporal resolution. Properties of the landscapes can be probed only indirectly. Simulation, in principle, can provide (infinitely) high spatial and temporal resolution, necessary for the construction of the quantitatively accurate free energy landscapes. Recently, due to advances in the hardware and simulation methodology realistic simulation of folding of small fast-folding proteins became computationally affordable. The current record holder is the simulation reported in the landmark paper by Shaw et al. The paper presents a ``brute-force'' 200 $\mu$s detailed all-atom equilibrium folding simulation of FIP35 protein in explicit water that contains 15 folding-unfolding events with the folding rate and the native structure in agreement with experiment. Many more simulations are now in progress. Which opens a new exciting era, when a detailed understanding of the controversial issues surrounding protein folding dynamics becomes possible. The trajectories obtained in such simulations, which contain the detailed information about the folding dynamics, are many terabytes in size and generally present a big challenge for an automated analysis. Moreover, accurate quantitative analysis of protein folding dynamics in terms of free energy landscapes is notoriously difficult. Conventional approaches, even though being based on solid physical intuition, often lead to suboptimal results with simple free energy landscapes which hide the inherent complexity of folding dynamics. Such landscapes, in essence, through away the biggest advantage of the simulation - the wealth of detailed information. We will apply the newly developed rigorous approach to determine high-resolution quantitatively accurate free energy landscape for protein folding. Later the approach will be extended to determine such landscape based on trajectories recorded in single molecule experiments.