Computational Studies of Protein-Protein Interactions

关于活细胞中蛋白质分子的最基本问题之一是它们相互作用的方式。蛋白质如何识别并与正确的伴侣结合是细胞内和细胞间信息交流的方式之一。这就是细胞在分子水平上处理信息和做出决定的方式。原子水平的信息对于解释蛋白质在结构和动力学方面的特定相互作用是必不可少的。该研究项目包括发展和改进基于计算机的“分子动力学”方法，以准确计算和预测蛋白质如何相互结合。分子动力学（MD）使用牛顿的经典方程，力=质量x加速度，模拟所有原子的运动作为时间的函数。MD模拟虽然近似于真实世界，但提供了原子运动时间过程的详细信息，而这些信息是实验无法获得的。基于原子模型和MD模拟的方法是有利的，因为它不依赖于任何关于蛋白质结合的特定假设。能够通过计算预测蛋白质可能的结合伙伴，将为更好地了解活细胞打开大门。除了发展对科学界有广泛用途的方法外，该项目还将向本科和研究生阶段的年轻科学家（包括代表性不足的少数民族成员）提供跨学科培训和指导，并向高中生提供服务。该研究项目包括改进理论和计算方法，以准确地计算和预测分子如何相互结合。根据统计热力学理论，控制分子结合的量是结合自由能。这个数学上定义良好的量可以用计算机模拟感兴趣的分子的原子模型来计算。分子动力学（MD）模拟可以帮助阐明在原子水平上控制生物分子结合的基本原理。MD包括构建大分子的详细原子模型，并使用牛顿的经典方程F=MA，从字面上模拟所有原子作为时间函数的动态运动。微观力（牛顿方程中的“F”）可以用由简单解析函数构造的势函数（也称为力场）来近似。验证力场的准确性也是基于MD仿真计算的重要用途。计算出的轨迹虽然近似于真实世界，但提供了关于原子运动时间过程的详细信息，这是不可能通过实验获得的。这一领域的努力可能会产生巨大的影响，因为生物学正在进入一个定量时代，这需要预测分子结合的能力。该研究的第一个具体目标是设计和开发新的计算方法，以解决蛋白质-蛋白质相互作用所带来的具体挑战。对大量可获得的溶剂化蛋白质构型进行采样是至关重要的。简单地运行更长时间的无偏MD不会导致成功，需要特殊的增强采样方法来实现这一点。第二个目标是验证和测试计算方法，并评估原子力场的准确性，以已知的蛋白质-蛋白质复合物为基础。重要的是要知道结合自由能、熵和焓分解以及缔合的动力学速率的结果是否正确。在最后阶段，目标集中在测试计算方法与日益具有挑战性的情况。该项目由生物科学理事会分子和细胞生物科学部的分子生物物理集群和数学与物理科学理事会化学部的化学理论、模型和计算方法项目共同资助。

One of the most basic questions about the protein molecules in the living cell is the manner by which they interact with one another. How proteins recognize and bind (stick) to the correct partner(s) is one of the ways information is communicated within and between the cells. This is how cells process information and make decisions at the molecular level. Atomic-level information is essential to explain the specific interactions between proteins in terms of structure and dynamics. The research project consists in developing and improving the computer-based approach of "molecular dynamics" to calculate and predict accurately how proteins bind to one another. Molecular dynamics (MD) consists of using Newton's classical equation, force = mass x acceleration, to simulate the motions of all the atoms as a function of time. The MD simulation, although an approximation to the real world, provides detailed information about the time course of the atomic motions that cannot be accessed experimentally. The approach based on atomic models and MD simulations is advantageous because it does not rely on any particular assumption about protein binding. Being able to predict the possible binding partners of a protein from computations will open the door to a much better understanding of living cells. In addition to developing methods that will be of broad utility to the scientific community, the project will provide inter-disciplinary training and mentoring to young scientists at the undergraduate and graduate level (including members of under-represented minorities), as well as outreach to high school students. The research project consists in improving the theoretical and computational methods to calculate and predict accurately how molecules bind to one another. According to the theory of statistical thermodynamics, the quantity that controls the association of molecules is the binding free energy. This mathematically well-defined quantity can be calculated using computer simulations of atomic models of the molecules of interest. Molecular dynamics (MD) simulation can help elucidate the fundamental principles governing the binding of biological molecules at the atomic level. MD consists of constructing detailed atomic models of the macromolecules and using Newton's classical equation, F=MA, to literally simulate the dynamical motions of all the atoms as a function of time. The microscopic forces (the "F" in Newton's equation) are approximated by using a potential function, also called a force field, constructed from simple analytical functions. Validating the accuracy of the force field is also an important use of computations based on MD simulations. The calculated trajectory, though an approximation to the real world, provides detailed information about the time course of the atomic motions, which is impossible to access experimentally. Efforts in this area are likely to have a large impact because biology is entering a quantitative era that requires an ability to predict the binding of molecules. The first specific objective of the research is the design and development of novel computational methodologies addressing the specific challenges presented by protein-protein interactions. Sampling the large number of accessible configurations of the solvated proteins is critical. Simply running longer unbiased MD is not going to lead to success, and special enhanced sampling methods are needed to achieve this. The second objective is to validate and test the computational methodology and assess the accuracy of the atomic force field on the basis of well-known protein-protein complexes. It is important to know if the results are correct for the binding free energy, the entropy and enthalpy decomposition, and the kinetic rate of association. In the last stage, the objectives are focused on testing the computational method with respect to increasingly challenging situations.This project is jointly funded by the Molecular Biophysics Cluster in the Division of Molecular and Cellular Biosciences in the Directorate for Biological Sciences and the Chemical Theory, Models, and Computational Methods Program in the Division of Chemistry in the Directorate of Mathematical and Physical Sciences.