Berchanski Alexander, Eisenstein Miriam
Department of Biological Chemistry, Weizmann Institute of Science, Rehovot 76100, Israel.
Proteins. 2003 Dec 1;53(4):817-29. doi: 10.1002/prot.10480.
Comparative modeling methods are commonly used to construct models of homologous proteins or oligomers. However, comparative modeling may be inapplicable when the number of subunits in a modeled oligomer is different than in the modeling template. Thus, a dimer cannot be a template for a tetramer because a new monomer-monomer interface must be predicted. We present in this study a new prediction approach, which combines protein-protein docking with either of two tetramer-forming algorithms designed to predict the structures of tetramers with D2 symmetry. Both algorithms impose symmetry constraints. However, one of them requires identification of two of the C2 dimers within the tetramer in the docking step, whereas the other, less demanding algorithm, requires identification of only one such dimer. Starting from the structure of one subunit, the procedures successfully reconstructed 16 known D2 tetramers, which crystallize with either a monomer, a dimer or a tetramer in the asymmetric unit. In some cases we obtained clusters of native-like tetramers that differ in the relative rotation of the two identical dimers within the tetramer. The predicted structural pliability for concanavalin-A, phosphofructokinase, and fructose-1,6-bisphosphatase agrees semiquantitatively with the observed differences between the several experimental structures of these tetramers. Hence, our procedure identifies a structural soft-mode that allows regulation via relative rigid-body movements of the dimers within these tetramers. The algorithm also predicted three nearly correct tetramers from model structures of single subunits, which were constructed by comparative modeling from subunits of homologous tetrameric, dimeric, or hexameric systems.
比较建模方法通常用于构建同源蛋白质或寡聚体的模型。然而，当建模的寡聚体中亚基数量与建模模板中的不同时，比较建模可能不适用。因此，二聚体不能作为四聚体的模板，因为必须预测新的单体 - 单体界面。我们在本研究中提出了一种新的预测方法，该方法将蛋白质 - 蛋白质对接与两种用于预测具有D2对称性的四聚体结构的四聚体形成算法中的任何一种相结合。两种算法都施加了对称性约束。然而，其中一种算法在对接步骤中需要识别四聚体内的两个C2二聚体，而另一种要求较低的算法只需要识别一个这样的二聚体。从一个亚基的结构开始，这些程序成功重建了16种已知的D2四聚体，它们在不对称单元中与单体、二聚体或四聚体一起结晶。在某些情况下，我们获得了类天然四聚体的簇，这些簇在四聚体内两个相同二聚体的相对旋转方面有所不同。伴刀豆球蛋白A、磷酸果糖激酶和果糖 - 1,6 - 二磷酸酶的预测结构柔韧性与这些四聚体的几个实验结构之间观察到的差异半定量地一致。因此，我们的程序识别出一种结构软模式，该模式允许通过这些四聚体内二聚体的相对刚体运动进行调节。该算法还从单亚基的模型结构中预测了三个几乎正确的四聚体，这些单亚基模型结构是通过对同源四聚体、二聚体或六聚体系统的亚基进行比较建模构建的。
