Theoretical Studies of Aqueous Solvation of Proteins

一野敏子MCB 98-08116水在蛋白质的结构和功能中起着重要的作用，如折叠、酶活性和与其他分子的相互作用。然而，水是一种复杂的溶剂，其性质仍未被完全了解。水与蛋白质的相互作用特别复杂，因为它们是大分子，具有各种各样的官能团，没有规则的结构。因此，蛋白质水溶液溶剂化理论的发展对于理解蛋白质的结构和功能至关重要。本研究的长期目标是1)在生物大分子的计算机模拟中开发快速、准确的溶剂效应处理方法；2)了解蛋白质溶剂化的本质，包括疏水效应和离子溶剂化等现象。具体目标是：1)将PI的新水模型纳入蛋白质的计算机模拟；2)将PI的大分子积分方程理论应用于蛋白质；3)通过PI的非线性连续统理论和模拟研究蛋白质的水溶液溶剂化。主要方法有统计力学理论和分子动力学计算机模拟。在第一个目标中，PI小组先前开发的新的软粘偶极子（SSD）水模型将用于蛋白质的计算机模拟。SSD模型相对于通常用于生物模拟的三位点模型是一个相当大的进步，因为它具有更好的结构、介电性和动力学特性，并且在蒙特卡罗模拟中快了7倍，在分子动力学模拟中快了3到4倍。固态水分子由具有点偶极子的伦纳德-琼斯球和模拟氢键相互作用的四面体“粘”势组成。能量参数是简单地根据水分子的位置和方向给出的，而三位点模型是根据氧和两个氢的位置给出的。SSD模型将在CHARMM（一个分子力学计算机程序）中实现，并在蛋白质的分子动力学模拟中进行测试。此外，还提出了进一步的溶剂化性能测试，包括显式电子极化。在第二个目标中，先前由PI小组开发的球形分子溶剂化理论将用于蛋白质。该模型允许用隐式模型取代计算机模拟中的显式溶剂，该模型准确地描述了水溶液溶剂化的重要特征，减少了计算机时间。该理论使用统计力学积分方程理论来模拟溶剂化的分子性质，该理论在预测简单液体的结构方面非常成功。具体地说，一个近似的Ornstein-Zernike方程理论与一个改进的超网状链闭包被用来预测水分子的位置和方向在溶质周围的分布。这种方法的重要特点是保留了完整的三维分布，并将分布的密度部分和定向部分分离。该理论预测了分子性质，如氢键，这是无法通过简单的连续介质模型得到的，但仍然保持正确的宏观极限静电。该理论将用于预测蛋白质的溶剂化能。最后，先前在PI的小组开发的离子溶剂化理论将用于分析蛋白质和简单模型系统的分子动力学模拟。这将增加对蛋白质-水界面水结构的理解；特别是取向结构，它决定了溶剂的极化能。该理论是基于Onsager和Kirkwood的工作的准连续体方法，它预测了水分子的方向，但不包括特定的分子相互作用。它超越了波恩的简单的溶剂化理论，包括非线性效应，如介电饱和和最近修改的电伸缩。此外，还将研究该理论的进一步扩展，如启发式地包括分子效应和增加对复杂溶质（如蛋白质）的预测能力。2. 水在蛋白质的结构和功能中起着重要的作用，如折叠、酶活性和与其他分子的相互作用。然而，水是一种复杂的溶剂，其性质仍未被完全了解。水与蛋白质的相互作用特别复杂，因为它们是大分子，具有各种各样的官能团，没有规则的结构。因此，蛋白质水溶液溶剂化理论的发展对于理解蛋白质的结构和功能至关重要。本研究的长期目标是1)在生物大分子的计算机模拟中开发快速、准确的溶剂效应处理方法；2)了解蛋白质溶剂化的本质，包括疏水效应和离子溶剂化等现象。主要方法有统计力学理论和分子动力学计算机模拟。具体目标是实现和扩展PI小组先前开发的模型/理论，包括1)水的软粘性偶极子（SSD）模型2)球形分子的溶剂化理论3)离子溶剂化理论。SSD模型相对于通常用于生物模拟的三位点模型是一个相当大的进步，因为它具有更好的结构、介电性和动力学特性，并且在蒙特卡罗模拟中快了7倍，在分子动力学模拟中快了3到4倍。它将被用于蛋白质的计算机模拟测试。溶剂化模型允许用隐式模型代替计算机模拟中的显式溶剂，在更少的计算机时间内，描述分子性质，如氢键，但保持正确的宏观极限静电。离子理论是分析蛋白质和简单模型系统的分子动力学模拟。这将增加我们对蛋白质-水界面水结构的理解；特别是取向结构，它决定了溶剂的极化能。

Toshiko Ichiye MCB 98-08116 1. TECHNICAL ABSTRACT Water plays an important role in the structure and function of proteins such as in folding, enzymatic activity, and interactions with other molecules. However, water is a complex solvent whose properties are still not completely understood. The interaction of water with proteins is particularly complex because they are large molecules with a variety of functional groups and no regular structure. Thus, the development of theories of aqueous solvation for proteins is crucial to understanding their structure and function. The long-term goals of this study are 1) to develop fast, accurate treatments of solvent effects in computer simulations of biological macromolecules and 2) to understand the nature of protein solvation, including phenomena such as the hydrophobic effect and ionic solvation. Specific aims are: 1) to incorporate PI's new model of water into computer simulations of proteins, 2) to apply PI's macromolecular integral equation theory to proteins, and 3) to investigate aqueous solvation of proteins via PI's nonlinear continuum theory and simulations. The main methods are statistical mechanical theory and molecular dynamics computer simulations. In the first aim, the new soft, sticky dipole (SSD) model of water previously developed by PI's group will be implemented for computer simulations of proteins. The SSD model is a considerable advance over the three-site models commonly used for biological simulations because it has better structural, dielectrical and dynamical properties and yet is seven times faster in Monte Carlo and three to four times faster in molecular dynamics simulations. An SSD water molecule is composed of a Lennard-Jones sphere with a point dipole and a tetrahedral "sticky" potential mimicking hydrogen bonding interactions. The energy parameters are simply given in terms of the position and orientation of a water molecule, whereas the three-site models are given in terms of the positions of the oxygen and the two hydrogens. The SSD model will be implemented in CHARMM, a molecular mechanics computer program, and tested in molecular dynamics simulations of proteins. In addition, further testing of the solvation properties and including explicit electronic polarization are proposed. In the second aim, a theory for solvation of globular molecules previously developed by PI's group will be implemented for proteins. This model allows the replacement of explicit solvent in computer simulations with an implicit model that accurately describes the important features of aqueous solvation with less computer time. The theory models the molecular nature of solvation using statistical mechanical integral equation theories, which have been highly successful in predicting the structure of simple liquids. Specifically, an approximate Ornstein-Zernike equation theory with a modified hypernetted chain closure is used to predict the distribution of water molecule positions and orientations around a solute. The important features of this approach are the retention of the full three-dimensional distribution and the separation of the density and orientational parts of the distribution. The theory predicts molecular properties such as hydrogen bonding, which cannot be obtained by simple continuum models, and yet maintains the correct macroscopic limit electrostatics. The theory will be used to predict solvation energies of proteins. Finally, a theory of ionic solvation previously developed in PI's group will be used to analyze molecular dynamics simulations of proteins and simple model systems. This will increase the understanding of water structure at the protein-water interface; in particular, the orientational structure, which determines the polarization energy of the solvent. The theory is a quasi-continuum approach based on the works of Onsager and Kirkwood that predicts the orientation of water molecules but does not include specific molecular interactions. It goes beyond simple Born th eories of solvation by including non-linear effects such as dielectric saturation and, with a recent modification, electrostriction. Also, further extensions of the theory will be investigated, such as including molecular effects heuristically and increasing the predictive capabilities for complex solutes such as proteins. 2. Non-technical Water plays an important role in the structure and function of proteins such as in folding, enzymatic activity, and interactions with other molecules. However, water is a complex solvent whose properties are still not completely understood. The interaction of water with proteins is particularly complex because they are large molecules with a variety of functional groups and no regular structure. Thus, the development of theories of aqueous solvation for proteins is crucial to understanding their structure and function. The long-term goals of this study are 1) to develop fast, accurate treatments of solvent effects in computer simulations of biological macromolecules and 2) to understand the nature of protein solvation, including phenomena such as the hydrophobic effect and ionic solvation. The main methods are statistical mechanical theory and molecular dynamics computer simulations. The specific aims are to implement and extend models/theories previously developed by PI's group which include 1) a soft, sticky dipole (SSD) model of water 2) a theory for solvation of globular molecules, and 3) a theory of ionic solvation. The SSD model is a considerable advance over the three-site models commonly used for biological simulations because it has better structural, dielectrical and dynamical properties and yet is seven times faster in Monte Carlo and three to four times faster in molecular dynamics simulations. It will be tested for computer simulations of proteins. The solvation model allows the replacement of explicit solvent in computer simulations with an implicit model that, in less computer time, describes molecular properties such as hydroge n bonding and yet maintains the correct macroscopic limit electrostatics. The ionic theory is to analyze molecular dynamics simulations of proteins and simple model systems. This will increase our understanding of water structure at the protein-water interface; in particular, the orientational structure, which determines the polarization energy of the solvent.