Hahn Matthew W, Conant Gavin C, Wagner Andreas
Department of Biology, Box 90338, Duke University, Durham, NC 27708, USA. mwhahn@ ucdavis.edu
J Mol Evol. 2004 Feb;58(2):203-11. doi: 10.1007/s00239-003-2544-0.
Genetic networks show a broad-tailed distribution of the number of interaction partners per protein, which is consistent with a power-law. It has been proposed that such broad-tailed distributions are observed because they confer robustness against mutations to the network. We evaluate this hypothesis for two genetic networks, that of the E. coli core intermediary metabolism and that of the yeast protein-interaction network. Specifically, we test the hypothesis through one of its key predictions: highly connected proteins should be more important to the cell and, thus, subject to more severe selective and evolutionary constraints. We find, however, that no correlation between highly connected proteins and evolutionary rate exists in the E. coli metabolic network and that there is only a weak correlation in the yeast protein-interaction network. Furthermore, we show that the observed correlation is function-specific within the protein-interaction network: only genes involved in the cell cycle and transcription show significant correlations. Our work sheds light on conflicting results by previous researchers by comparing data from multiple types of protein-interaction datasets and by using a closely related species as a reference taxon. The finding that highly connected proteins can tolerate just as many amino acid substitutions as other proteins leads us to conclude that power-laws in cellular networks do not reflect selection for mutational robustness.
基因网络显示出每个蛋白质相互作用伙伴数量的宽尾分布，这与幂律一致。有人提出观察到这种宽尾分布是因为它们赋予网络对突变的鲁棒性。我们针对两个基因网络评估了这一假设，即大肠杆菌核心中间代谢网络和酵母蛋白质相互作用网络。具体而言，我们通过其一个关键预测来检验该假设：高度连接的蛋白质对细胞应该更重要，因此受到更严格的选择和进化约束。然而，我们发现，在大肠杆菌代谢网络中，高度连接的蛋白质与进化速率之间不存在相关性，而在酵母蛋白质相互作用网络中只有微弱的相关性。此外，我们表明在蛋白质相互作用网络中观察到的相关性是功能特异性的：只有参与细胞周期和转录的基因显示出显著相关性。我们的工作通过比较多种类型蛋白质相互作用数据集的数据并使用密切相关的物种作为参考分类单元，揭示了先前研究人员相互矛盾的结果。高度连接的蛋白质能够容忍与其他蛋白质一样多的氨基酸替换这一发现使我们得出结论，细胞网络中的幂律并不反映对突变鲁棒性的选择。
