Martí-Renom M A, Stote R H, Querol E, Avilés F X, Karplus M
Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain.
J Mol Biol. 1998 Nov 20;284(1):145-72. doi: 10.1006/jmbi.1998.2071.
The folding of the potato carboxypeptidase inhibitor (PCI) from partially unfolded conformations by the introduction of native disulfide bond constraints was studied by molecular dynamics simulations in explicit solvent. PCI consists of a globular core (Cys8 to Cys34), two flexible terminal regions (Glu1 to Ile7 and Glu35 to Gly39) and three loop regions characteristic of the family of proteins known as knottins. To generate unfolded conformations, two high temperature (600 K) simulations were performed; one with the native disulfide bonds intact (N600), and one with the disulfide bonds broken (ND600). For comparison purposes, two simulations at 300 K were done; one with the native disulfide bonds (N300), and one with the disulfide bonds broken (ND300). The N300 simulation reached an energetic equilibrium within a few picoseconds and maintained a stable structure during the 500 ps simulation. The three other simulations led to partial unfolding. The largest changes were observed in ND600 simulation with an rms deviation of over 5 A and radius of gyration 12.5% larger than the crystal structure value. Six structures from the ND600 simulation and one from the N600 simulation were used as starting structures for nine refolding simulations with somewhat different protocols for reforming the native disulfide bonds; in all cases the disulfides were reformed at 600 K and the temperature was decreased to 300 K for equilibration of the folded structures. Except for one structure that was significantly misfolded (final rms of 6.64 A with respect to N300), the other folding simulations recovered the native simulation structure (N300) to within rms differences ranging from 1.8 to 3.2 A for the main-chain of the core, relative to the N300, the X-ray and the NMR structures. Of particular interest is the internal and overall refolding behavior of the three loop regions. The more unfolded starting structures led to smaller rms values for the folded structures. Several energetic and solvation models were used to evaluate the X-ray, NMR, N300 and refolded structures. Although most models can distinguish the X-ray, NMR and N300 from the refolded structures, there is no correlation between the rms values of the latter and their estimated stability. Implications of the present results for protein folding by simulations and database search methods are discussed.
通过在显式溶剂中的分子动力学模拟，研究了引入天然二硫键约束后，马铃薯羧肽酶抑制剂（PCI）从部分未折叠构象的折叠过程。PCI由一个球状核心（Cys8至Cys34）、两个柔性末端区域（Glu1至Ile7和Glu35至Gly39）以及三种被称为“扭结蛋白”家族蛋白质所特有的环区域组成。为了生成未折叠构象，进行了两次高温（600K）模拟；一次是天然二硫键保持完整（N600），另一次是二硫键断裂（ND600）。为作比较，还进行了两次300K的模拟；一次是天然二硫键存在（N300），另一次是二硫键断裂（ND300）。N300模拟在几皮秒内达到能量平衡，并在500皮秒的模拟过程中保持稳定结构。其他三次模拟导致了部分展开。在ND600模拟中观察到的变化最大，均方根偏差超过5埃，回转半径比晶体结构值大12.5%。来自ND600模拟的六个结构和来自N600模拟的一个结构被用作九次重折叠模拟的起始结构，这些模拟在重新形成天然二硫键的方案上略有不同；在所有情况下，二硫键在600K时重新形成，然后温度降至300K以使折叠结构达到平衡。除了一个结构严重错误折叠（相对于N300的最终均方根为6.64埃）外，其他折叠模拟将天然模拟结构（N300）恢复到相对于N300、X射线和NMR结构，核心主链的均方根差异在1.8至3.2埃范围内。特别令人感兴趣的是三个环区域的内部和整体重折叠行为。起始结构展开程度越高，折叠结构的均方根值越小。使用了几种能量和溶剂化模型来评估X射线、NMR、N300和重折叠结构。尽管大多数模型能够将X射线、NMR和N300与重折叠结构区分开来，但后者的均方根值与其估计的稳定性之间没有相关性。讨论了本研究结果对通过模拟和数据库搜索方法进行蛋白质折叠的意义。
