Fast Computations for Structural Transitions in Proteins

描述（由申请人提供）：缺乏对蛋白质构象转变的可靠描述代表了结构生物学知识体系中的一个关键空白。不仅构象转变是许多蛋白质功能的核心-从酶到马达蛋白-而且结构中间体也是过渡态抑制剂药物的公认目标。因此，我们提出了一种新的计算方法来研究戏剧性的构象转变（a）钙调蛋白（CAM），它介导的基因表达，肌肉收缩到有丝分裂的基本过程，和（B）肌球蛋白，引起肌肉收缩的蛋白马达。新的方案将建立在非常有希望的初步结果的基础上，这些结果展示了前所未有的生理时间尺度。结合残差水平的建模和精细网格离散化，该方法已经被证明能够无偏动态模拟的几十个构象转换的两个72残基域的钙调蛋白。事实上，该协议每天在廉价的单处理器台式计算机上生成若干个转换事件。这种高效率将使人们能够利用有限的计算机资源来研究肌球蛋白。一个多层次的建模方法是这个建议的核心。从初始“粗粒度”模型生成的结果（例如，残留水平）将使用一系列逐步更精确的力场进行改进。与离散化所实现的高质量采样相结合，多级建模将允许对计算数据的高度置信度。事实上，模拟结果-包括结构中间体的模型和动力学关键残基的身份-将用于设计实验（由合作者进行），旨在提高我们对结构事件的详细理解。由于其速度和简单性，新协议形成了一个理想的基础上，一个软件工具的价值，专家和非专家一样。提出了分发方便用户的软件包的详细计划，源代码将提供给专家用户修改。

DESCRIPTION (provided by applicant): The lack of reliable descriptions of conformational transitions in proteins represents a critical gap in the body of structural biology knowledge. Not only are conformational transitions at the heart of many proteins' functions --- from enzymes to motor proteins --- but structural intermediates also serve as a well-established class of targets for transition-state inhibitor drugs. We therefore propose a novel computational approach to study the dramatic conformational transitions in (a) calmodulin (CAM), which mediates essential processes from gene expression to muscle contraction to mitosis, and (b) myosin, the protein motor that causes muscle contraction. The new protocol will build on extremely promising preliminary results, which demonstrate unprecedented access to physiological timescales. Combining residue-level modeling and fine-grid discretization, the approach has already proven capable of unbiased dynamic simulation of dozens of conformational transitions in each of the two 72-residue domains of calmodulin. In fact, the protocol generates several transition events per day on an inexpensive, single-processor desktop computer. This high efficiency will enable the study of myosin with modest computer resources. A multi-level approach to modeling is central to this proposal. Results generated from initial "coarse grained" models (e.g., residue-level) will be refined using a series of progressively more accurate force fields. Combined with the high-quality sampling enabled by discretization, multi-level modeling will permit a high degree of confidence in the computational data. Indeed, the simulation results --- including models of structural intermediates and the identities of kinetically critical residues --- will be used to design experiments (to be performed by collaborators) aimed toward improving our detailed understanding of structural events. Because of its speed and simplicity, the new protocol forms an ideal basis for a software tool of value to expert and non-expert alike. A detailed plan for distributing user-friendly software packages is presented, and the source code will be available for modification by expert users.