Ripoll D R, Piela L, Vásquez M, Scheraga H A
Baker Laboratory of Chemistry, Cornell University, Ithaca, New York 14853-1301.
Proteins. 1991;10(3):188-98. doi: 10.1002/prot.340100304.
In connection with the accompanying paper to test various models for the hydration of polypeptides, we have explored a limited portion of the conformational energy hyperspace of the small protein bovine pancreatic trypsin inhibitor (BPTI) with the aid of two search methods developed in this laboratory. A series of low-energy conformations was obtained as a result of this study. These conformations constitute a set of local minima in the conformational energy space of the molecule as described by the ECEPP/2 (Empirical Conformational Energy Program for Peptides) potential energy function, without the inclusion of hydration. Five different initial conformations were used in this exploration: the first corresponds to an energy-refined structure based on the crystallographic coordinates (4PTI) provided by Deisenhofer and Steigemann and reported previously by Meirovitch and Scheraga. The remaining four initial conformations were obtained by using a Variable-Target-Function procedure, applied to the experimental Cartesian coordinates (5PTI) reported by Wlodawer et al. The self-consistent electrostatic field (SCEF) and the electrostatically driven Monte Carlo (EDMC) methods were used to search the conformational space. The SCEF and EDMC methodologies assume that a polypeptide or protein molecule is driven toward the native structure mainly by the action of the electrostatic interactions. Application of these methodologies led to a set of conformations (up to 50 kcal/mol lower than the starting ones) with ECEPP/2 energies lower than any of those that we had previously found. Application of both methods to the initial conformation generated from 4PTI led to a series of low-energy conformations exhibiting similar rms deviations with respect to the experimental data (4PTI) as did the starting conformation. However, statistical analysis of the runs that had started from the conformations generated by using the variable-target-function procedure (and applying the EDMC method) indicated that the rms deviations of the atomic positions of the new low-energy conformations tended to increase as the energy improved, when compared with the X-ray data from which the starting conformations had been generated. The structures with the lowest energies also had radii of gyration smaller than the experimentally observed one. These results indicated a need to include hydration in the potential function, and provided the conformations used in the accompanying paper to test various hydration models.
为了检验多肽水合作用的各种模型，结合随附的论文，我们借助本实验室开发的两种搜索方法，对小蛋白牛胰蛋白酶抑制剂（BPTI）构象能量超空间的有限部分进行了探索。这项研究得到了一系列低能量构象。这些构象在不考虑水合作用的情况下，由ECEPP/2（肽的经验构象能量程序）势能函数描述，构成了分子构象能量空间中的一组局部最小值。在这次探索中使用了五种不同的初始构象：第一种对应于基于Deisenhofer和Steigemann提供的晶体学坐标（4PTI）并由Meirovitch和Scheraga先前报道的能量优化结构。其余四种初始构象是通过将可变目标函数程序应用于Wlodawer等人报道的实验笛卡尔坐标（5PTI）获得的。使用自洽静电场（SCEF）和静电驱动蒙特卡罗（EDMC）方法搜索构象空间。SCEF和EDMC方法假设多肽或蛋白质分子主要通过静电相互作用的作用趋向于天然结构。应用这些方法导致了一组构象（比起始构象低达50千卡/摩尔），其ECEPP/2能量低于我们之前发现的任何构象。将这两种方法应用于由4PTI生成的初始构象，导致了一系列低能量构象，相对于实验数据（4PTI），其均方根偏差与起始构象相似。然而，对从使用可变目标函数程序生成的构象开始的运行（并应用EDMC方法）进行的统计分析表明，与生成起始构象的X射线数据相比，新的低能量构象的原子位置的均方根偏差倾向于随着能量的提高而增加。能量最低的结构的回转半径也小于实验观察到的回转半径。这些结果表明需要在势能函数中包含水合作用，并提供了随附论文中用于测试各种水合模型的构象。
