Tyagi Manoj, Gowri Venkataraman S, Srinivasan Narayanaswamy, de Brevern Alexandre G, Offmann Bernard
Laboratoire de Biochimie et Génétique Moléculaire, Université de La Réunion, BP 7151, 15 avenue René Cassin, 97715 Saint Denis Messag Cedex 09, La Réunion, France.
Proteins. 2006 Oct 1;65(1):32-9. doi: 10.1002/prot.21087.
Analysis of protein structures based on backbone structural patterns known as structural alphabets have been shown to be very useful. Among them, a set of 16 pentapeptide structural motifs known as protein blocks (PBs) has been identified and upon which backbone model of most protein structures can be built. PBs allows simplification of 3D space onto 1D space in the form of sequence of PBs. Here, for the first time, substitution probabilities of PBs in a large number of aligned homologous protein structures have been studied and are expressed as a simplified 16 x 16 substitution matrix. The matrix was validated by benchmarking how well it can align sequences of PBs rather like amino acid alignment to identify structurally equivalent regions in closely or distantly related proteins using dynamic programming approach. The alignment results obtained are very comparable to well established structure comparison methods like DALI and STAMP. Other interesting applications of the matrix have been investigated. We first show that, in variable regions between two superimposed homologous proteins, one can distinguish between local conformational differences and rigid-body displacement of a conserved motif by comparing the PBs and their substitution scores. Second, we demonstrate, with the example of aspartic proteinases, that PBs can be efficiently used to detect the lobe/domain flexibility in the multidomain proteins. Lastly, using protein kinase as an example, we identify regions of conformational variations and rigid body movements in the enzyme as it is changed to the active state from an inactive state.
基于被称为结构字母表的主链结构模式对蛋白质结构进行分析已被证明非常有用。其中，一组由16个五肽结构基序组成的、被称为蛋白质模块（PBs）的结构已被识别出来，并且大多数蛋白质结构的主链模型都可以基于此构建。PBs能够将三维空间简化为一维空间，形式为PBs序列。在此，首次对大量比对后的同源蛋白质结构中PBs的替换概率进行了研究，并将其表示为一个简化的16×16替换矩阵。通过基准测试该矩阵在比对PBs序列方面的效果（类似于氨基酸比对），使用动态规划方法来识别密切相关或远缘相关蛋白质中的结构等效区域，从而对该矩阵进行了验证。所获得的比对结果与诸如DALI和STAMP等成熟的结构比较方法非常可比。还研究了该矩阵的其他有趣应用。我们首先表明，在两个叠加的同源蛋白质之间的可变区域中，通过比较PBs及其替换分数，可以区分保守基序的局部构象差异和刚体位移。其次，以天冬氨酸蛋白酶为例，我们证明PBs可以有效地用于检测多结构域蛋白质中的结构域/结构域灵活性。最后，以蛋白激酶为例，我们识别了该酶从无活性状态转变为活性状态时的构象变化区域和刚体运动区域。
