How evolution endowed membrane enzymes with specific abilities, and then tuned them to the needs of different cells, is poorly understood. We examined whether statistical coupling analysis (SCA) can be applied to rhomboid proteases, the most widely distributed membrane proteins, to identify amino acid “sectors” that evolved independently to acquire a specific function. SCA revealed three coevolving residue networks that form two sectors. Sector 1 determines substrate specificity, but is paradoxically scattered across the protein, consistent with dynamics driving rhomboid-substrate interactions. Sector 2 is hierarchically composed of a subgroup that maintains the catalytic site, and another that maintains the overall fold, forecasting evolution of rhomboid pseudoproteases. Changing only sector 1 residues of a “recipient” rhomboid converted its substrate specificity and catalytic efficiency to that of the “donor.” While used only twice over a decade ago, SCA should be generally applicable to membrane proteins, and our sector grafting approach provides an efficient strategy for designing enzymes.

进化如何赋予膜酶特定的能力，然后将它们调整到不同细胞的需要，我们知之甚少。我们检查了统计耦合分析 (SCA) 是否可以应用于菱形蛋白酶（分布最广泛的膜蛋白），以识别独立进化以获得特定功能的氨基酸“部门”。SCA 揭示了三个共同进化的残留物网络，它们形成了两个部门。扇区 1 确定底物特异性，但自相矛盾地分散在整个蛋白质中，与驱动菱形-底物相互作用的动力学一致。扇区 2 由一个维持催化位点的亚组和另一个维持整体折叠的亚组分层组成，预测菱形假蛋白酶的进化。仅改变“受体”菱形的扇区 1 残基，将其底物特异性和催化效率转化为“供体”的底物特异性和催化效率。虽然十多年前只使用过两次，但 SCA 应该普遍适用于膜蛋白，我们的部门接枝方法为设计酶提供了有效的策略。