Tran Thanh Thuy, Nguyen Phuong H, Derreumaux Philippe
Laboratoire de Biochimie Théorique, UPR 9080, CNRS, Université Denis Diderot, Sorbonne Paris Cité IBPC, 13 rue Pierre et Marie Curie, 75005 Paris, France.
J Chem Phys. 2016 May 28;144(20):205103. doi: 10.1063/1.4951739.
Coarse-grained protein lattice models approximate atomistic details and keep the essential interactions. They are, therefore, suitable for capturing generic features of protein folding and amyloid formation at low computational cost. As our aim is to study the critical nucleus sizes of two experimentally well-characterized peptide fragments Aβ16-22 and Aβ37-42 of the full length Aβ1-42 Alzheimer's peptide, it is important that simulations with the lattice model reproduce all-atom simulations. In this study, we present a comprehensive force field parameterization based on the OPEP (Optimized Potential for Efficient protein structure Prediction) force field for an on-lattice protein model, which incorporates explicitly the formation of hydrogen bonds and directions of side-chains. Our bottom-up approach starts with the determination of the best lattice force parameters for the Aβ16-22 dimer by fitting its equilibrium parallel and anti-parallel β-sheet populations to all-atom simulation results. Surprisingly, the calibrated force field is transferable to the trimer of Aβ16-22 and the dimer and trimer of Aβ37-42. Encouraged by this finding, we characterized the free energy landscapes of the two decamers. The dominant structure of the Aβ16-22 decamer matches the microcrystal structure. Pushing the simulations for aggregates between 4-mer and 12-mer suggests a nucleus size for fibril formation of 10 chains. In contrast, the Aβ37-42 decamer is largely disordered with mixed by parallel and antiparallel chains, suggesting that the nucleus size is >10 peptides. Our refined force field coupled to this on-lattice model should provide useful insights into the critical nucleation number associated with neurodegenerative diseases.
粗粒度蛋白质晶格模型近似原子细节并保留基本相互作用。因此，它们适合以低计算成本捕捉蛋白质折叠和淀粉样蛋白形成的一般特征。由于我们的目标是研究全长Aβ1-42阿尔茨海默病肽的两个实验特征良好的肽片段Aβ16-22和Aβ37-42的临界核尺寸，晶格模型模拟能重现全原子模拟这一点很重要。在本研究中，我们基于OPEP（高效蛋白质结构预测优化势）力场为晶格蛋白质模型提出了一种全面的力场参数化方法，该方法明确纳入了氢键的形成和侧链方向。我们的自下而上方法始于通过将Aβ16-22二聚体的平衡平行和反平行β-折叠群体拟合到全原子模拟结果来确定其最佳晶格力参数。令人惊讶的是，校准后的力场可转移到Aβ16-22三聚体以及Aβ37-42二聚体和三聚体上。受这一发现鼓舞，我们对这两个十聚体的自由能景观进行了表征。Aβ16-22十聚体的主要结构与微晶结构匹配。对4聚体到12聚体聚集体的模拟表明，纤维形成的核尺寸为10条链。相比之下，Aβ37-42十聚体在很大程度上是无序的，平行链和反平行链混合，这表明核尺寸大于10个肽段。我们改进的力场与这个晶格模型相结合，应该能为与神经退行性疾病相关的临界成核数提供有用的见解。
