Department of Biological Sciences, University of Bergen, N-5020 Bergen, Norway.
Computational Biology Unit, Department of Informatics, University of Bergen, N-5020 Bergen, Norway.
J Chem Theory Comput. 2020 Apr 14;16(4):2550-2560. doi: 10.1021/acs.jctc.9b01194. Epub 2020 Mar 9.
Cation-π interactions play an important role in biomolecular recognition, including interactions between membrane phosphatidylcholine lipids and aromatic amino acids of peripheral proteins. While molecular mechanics coarse grain (CG) force fields are particularly well suited to simulate membrane proteins in general, they are not parameterized to explicitly reproduce cation-π interactions. We here propose a modification of the polarizable MARTINI coarse grain (CG) model enabling it to model membrane binding events of peripheral proteins whose aromatic amino acid interactions with choline headgroups are crucial for their membrane binding. For this purpose, we first collected and curated a dataset of eight peripheral proteins from different families. We find that the MARTINI CG model expectedly underestimates aromatics-choline interactions and is unable to reproduce membrane binding of the peripheral proteins in our dataset. Adjustments of the relevant interactions in the polarizable MARTINI force field yield significant improvements in the observed binding events. The orientation of each membrane-bound protein is comparable to reference data from all-atom simulations and experimental binding data. We also use negative controls to ensure that choline-aromatics interactions are not overestimated. We finally check that membrane properties, transmembrane proteins, and membrane translocation potential of mean force (PMF) of aromatic amino acid side-chain analogues are not affected by the new parameter set. This new version "MARTINI 2.3P" is a significant improvement over its predecessors and is suitable for modeling membrane proteins including peripheral membrane binding of peptides and proteins.
阳离子-π 相互作用在生物分子识别中起着重要作用，包括膜磷脂脂质与外周蛋白芳香族氨基酸之间的相互作用。虽然分子力学粗粒 (CG) 力场特别适合模拟一般的膜蛋白，但它们没有被参数化来明确再现阳离子-π 相互作用。在这里，我们提出了对可极化 MARTINI CG 模型的修改，使其能够模拟外周蛋白的膜结合事件，这些蛋白的芳香族氨基酸与胆碱头基的相互作用对于它们的膜结合至关重要。为此，我们首先收集并整理了来自不同家族的 8 种外周蛋白的数据集。我们发现，MARTINI CG 模型预计会低估芳香族-胆碱相互作用，并且无法重现我们数据集中外周蛋白的膜结合。可极化 MARTINI 力场中相关相互作用的调整可显著改善观察到的结合事件。每个结合在膜上的蛋白质的取向与全原子模拟和实验结合数据的参考数据相当。我们还使用阴性对照来确保不高估胆碱-芳香族相互作用。最后，我们检查新参数集是否会影响膜性质、跨膜蛋白和芳香族氨基酸侧链类似物的膜穿入平均力势 (PMF)。新版本"MARTINI 2.3P" 相较于其前身有了显著改进，适合于模拟膜蛋白，包括肽和蛋白质的外周膜结合。
