The formation of membraneless organelles in cells commonly occurs via liquid-liquid phase separation (LLPS), and is in many cases driven by multivalent interactions between intrinsically disordered proteins (IDPs). Molecular simulations can reveal the specific amino acid interactions driving LLPS, which is hard to obtain from experiment. Coarse-grained simulations have been used to directly observe the sequence determinants of phase separation but have limited spatial resolution, while all-atom simulations have yet to be applied to LLPS due to the challenges of large system sizes and long time scales relevant to phase separation. We present a novel multiscale computational framework by obtaining initial molecular configurations of a condensed protein-rich phase from equilibrium coarse-grained simulations, and back mapping to an all-atom representation. Using the specialized Anton 2 supercomputer, we resolve microscopic structural and dynamical details of protein condensates through microsecond-scale all-atom explicit-solvent simulations. We have studied two IDPs which phase separate in vitro: the low complexity domain of FUS and the N-terminal disordered domain of LAF-1. Using this approach, we explain the partitioning of ions between phases with low and high protein density, demonstrate that the proteins are remarkably dynamic within the condensed phase, identify the key residue-residue interaction modes stabilizing the dense phase, all while showing good agreement with experimental observations. Our approach is generally applicable to all-atom studies of other single and multi-component systems of proteins and nucleic acids involved in the formation of membraneless organelles.

中文翻译：

---

微秒长的原子模拟探测的蛋白质冷凝物的分子细节

细胞中无膜细胞器的形成通常通过液-液相分离（LLPS）发生，并且在许多情况下是由内在无序蛋白（IDP）之间的多价相互作用驱动的。分子模拟可以揭示驱动LLPS的特定氨基酸相互作用，这很难从实验中获得。粗粒度模拟已被用于直接观察相分离的序列决定因素，但空间分辨率有限，而全原子模拟由于系统规模大和与相分离相关的时间尺度长的挑战而尚未应用于LLPS。 。我们通过从平衡粗粒度模拟获得浓缩的富含蛋白质的相的初始分子构型，提出了一种新颖的多尺度计算框架，并映射回全原子表示。使用专门的Anton 2超级计算机，我们通过微秒级全原子显式溶剂模拟来解析蛋白质冷凝物的微观结构和动力学细节。我们研究了在体外相分离的两个IDP：FUS的低复杂性域和LAF-1的N端无序域。使用这种方法，我们解释了蛋白质密度高低相之间的离子分配，证明了蛋白质在浓缩相中具有显着的动态，确定了稳定稠密相的关键残基-残基相互作用模式，同时与实验观察。