Sali A, Shakhnovich E, Karplus M
Department of Chemistry, Harvard University, Cambridge, MA 02138.
J Mol Biol. 1994 Feb 4;235(5):1614-36. doi: 10.1006/jmbi.1994.1110.
A three-dimensional lattice model of a protein is used to investigate the properties required for its folding to the native state. The polypeptide chain is represented as a 27 bead heteropolymer whose lowest energy (native) state can be determined by an exhaustive enumeration of all fully compact conformations. A total of 200 sequences with random interactions are generated and subjected to Monte Carlo simulations to determine which chains find the ground state in a short time; i.e. which sequences overcome the folding problem referred to as the Levinthal paradox. Comparison of the folding and non-folding sequences is used to identify the features that are required for fast folding to the global energy minimum. It is shown that successful folding does not require certain attributes that have been previously proposed as necessary for folding; these include a high number of short versus long-range contacts in the native state, a high content of the secondary structure in the native state, a strong correlation between the native contact map and the interaction parameters, and the existence of a high number of low energy states with near-native conformation. Instead, the essential difference between the folding and the non-folding sequences is the nature of the energy spectrum. The necessary and sufficient condition for a sequence to fold rapidly in the present model is that the native state is a pronounced energy minimum. As a consequence, the thermodynamic stability of the native state of a folding sequence has a sigmoidal dependence on temperature. This permits such a sequence to satisfy both the thermodynamic and the kinetic requirements for folding; i.e. the native state predominates thermodynamically at temperatures that are high enough for folding to be kinetically possible. The applicability of the present results to real proteins is discussed.
使用蛋白质的三维晶格模型来研究其折叠成天然状态所需的特性。多肽链被表示为一个由27个珠子组成的杂聚物,其最低能量(天然)状态可通过对所有完全紧凑构象的穷举枚举来确定。生成了总共200个具有随机相互作用的序列,并对其进行蒙特卡罗模拟,以确定哪些链能在短时间内找到基态;即哪些序列克服了被称为 Levinthal 悖论的折叠问题。通过比较折叠和非折叠序列来识别快速折叠到全局能量最小值所需的特征。结果表明,成功折叠并不需要某些先前被认为是折叠所必需的属性;这些属性包括天然状态下大量的短程与长程接触、天然状态下二级结构的高含量、天然接触图与相互作用参数之间的强相关性以及存在大量具有近天然构象的低能态。相反,折叠和非折叠序列之间的本质区别在于能谱的性质。在当前模型中,序列快速折叠的充要条件是天然状态是一个明显的能量最小值。因此,折叠序列天然状态的热力学稳定性对温度具有 S 形依赖性。这使得这样的序列能够同时满足折叠的热力学和动力学要求;即在温度足够高以使折叠在动力学上成为可能时,天然状态在热力学上占主导地位。讨论了当前结果对真实蛋白质的适用性。