Yu Lei, Li Da-Wei, Brüschweiler Rafael
Department of Chemistry and Biochemistry, The Ohio State University, Columbus, Ohio 43210, United States.
Campus Chemical Instrument Center, The Ohio State University, Columbus, Ohio 43210, United States.
J Phys Chem B. 2021 Jan 28;125(3):798-804. doi: 10.1021/acs.jpcb.0c10078. Epub 2021 Jan 14.
The prevalence of intrinsically disordered proteins (IDPs) and protein regions in structural biology has prompted the recent development of molecular dynamics (MD) force fields for the more realistic representations of such systems. Using experimental nuclear magnetic resonance backbone scalar -coupling constants of the intrinsically disordered proteins α-synuclein and amyloid-β in their native aqueous environment as a metric, we compare the performance of four recent MD force fields, namely, AMBER ff14SB, CHARMM C36m, AMBER ff99SB-, and AMBER ff99SBnmr2, by partitioning the polypeptides into an overlapping series of heptapeptides for which a cumulative total of 276 μs MD simulations were performed. The results show substantial differences between the different force fields at the individual residue level. Except for ff99SBnmr2, the force fields systematically underestimate the scalar (H,Hα)-couplings due to an underrepresentation of β-conformations and an overrepresentation of either α- or PP conformations. The study demonstrates that the incorporation of coil library information in modern MD force fields, as shown here for ff99SBnmr2, provides substantially improved performance and more realistic sampling of the local backbone dihedral angles of IDPs as reflected by the good accuracy of the computed scalar (H,Hα)-couplings with less than 0.5 Hz error. Such force fields will enable a better understanding of how structural dynamics and thermodynamics influence the IDP function. Although the methodology based on heptapeptides used here does not allow the assessment of potential intramolecular long-range interactions, its computational affordability permits well-converged simulations that can be easily parallelized. This should make the quantitative validation of intrinsic disorder observed in MD simulations of polypeptides with experimental scalar -couplings widely applicable.
内在无序蛋白质（IDP）及其区域在结构生物学中的普遍存在促使了分子动力学（MD）力场的发展，以便更真实地描述此类系统。我们以天然水环境中内在无序蛋白质α-突触核蛋白和淀粉样β蛋白的实验核磁共振主链标量耦合常数为指标，通过将多肽划分为一系列重叠的七肽来比较四个最新MD力场（即AMBER ff14SB、CHARMM C36m、AMBER ff99SB-和AMBER ff99SBnmr2）的性能，为此总共进行了276微秒的MD模拟。结果表明，在单个残基水平上，不同力场之间存在显著差异。除了ff99SBnmr2，其他力场由于β构象表示不足以及α或PP构象表示过度，系统地低估了标量（H，Hα）耦合。该研究表明，如ff99SBnmr2所示在现代MD力场中纳入卷曲库信息，可显著提高性能，并更真实地采样IDP的局部主链二面角，计算得到的标量（H，Hα）耦合具有良好的准确性，误差小于0.5赫兹即可体现。这样的力场将有助于更好地理解结构动力学和热力学如何影响IDP功能。尽管此处使用的基于七肽的方法无法评估潜在的分子内长程相互作用，但其计算成本较低，允许进行易于并行化的收敛良好的模拟。这应使基于实验标量耦合的多肽MD模拟中观察到的内在无序的定量验证得到广泛应用。
