Biomolecular recognition between proteins follows complex mechanisms, the understanding of which can substantially advance drug discovery efforts. Here, we track each step of the binding process in atomistic detail with molecular dynamics simulations using trypsin and its inhibitor bovine pancreatic trypsin inhibitor (BPTI) as a model system. We use umbrella sampling to cover a range of unbinding pathways. Starting from these simulations, we subsequently seed classical simulations at different stages of the process and combine them to a Markov state model. We clearly identify three kinetically separated states (an unbound state, an encounter state, and the final complex) and describe the mechanisms that dominate the binding process. From our model, we propose the following sequence of events. The initial formation of the encounter complex is driven by long-range interactions because opposite charges in trypsin and BPTI draw them together. The encounter complex features the prealigned binding partners with binding sites still partially surrounded by solvation shells. Further approaching leads to desolvation and increases the importance of van der Waals interactions. The native binding pose is adopted by maximizing short-range interactions. Thereby side-chain rearrangements ensure optimal shape complementarity. In particular, BPTI’s P1 residue adapts to the S1 pocket and prime site residues reorient to optimize interactions. After the paradigm of conformation selection, binding-competent conformations of BPTI and trypsin are already present in the apo ensembles and their probabilities increase during this proposed two-step association process. This detailed characterization of the molecular forces driving the binding process includes numerous aspects that have been discussed as central to the binding of trypsin and BPTI and protein complex formation in general. In this study, we combine all these aspects into one comprehensive model of protein recognition. We thereby contribute to enhance our general understanding of this fundamental mechanism, which is particularly critical as the development of biopharmaceuticals continuously gains significance.

中文翻译：

---

蛋白质-蛋白质结合作为两步机制：在 BPTI 和胰蛋白酶结合过程中预选相遇姿势

蛋白质之间的生物分子识别遵循复杂的机制，对其的理解可以大大推进药物发现工作。在这里，我们使用胰蛋白酶及其抑制剂牛胰蛋白酶抑制剂 (BPTI) 作为模型系统，通过分子动力学模拟以原子细节跟踪结合过程的每个步骤。我们使用伞式抽样来涵盖一系列解除绑定的途径。从这些模拟开始，我们随后在过程的不同阶段播种经典模拟，并将它们组合成马尔可夫状态模型。我们清楚地确定了三种动力学分离的状态（未结合状态、相遇状态和最终复合物）并描述了支配结合过程的机制。根据我们的模型，我们提出以下事件序列。相遇复合物的最初形成是由远程相互作用驱动的，因为胰蛋白酶和 BPTI 中的相反电荷将它们拉在一起。遭遇复合物的特点是预先对齐的结合伙伴，其结合位点仍部分被溶剂化壳包围。进一步接近会导致去溶剂化并增加范德华相互作用的重要性。通过最大化短程交互来采用原生绑定姿势。因此侧链重排确保了最佳的形状互补性。特别是，BPTI 的 P1 残基适应 S1 口袋和主要位点残基重新定向以优化相互作用。在构象选择范式之后，BPTI 和胰蛋白酶的结合能力构象已经存在于 apo 集合中，并且在这个提议的两步关联过程中它们的概率增加。驱动结合过程的分子力的详细表征包括许多方面，这些方面已被讨论为胰蛋白酶和 BPTI 结合以及一般蛋白质复合物形成的核心。在这项研究中，我们将所有这些方面结合到一个全面的蛋白质识别模型中。因此，我们有助于增强我们对这一基本机制的一般理解，随着生物制药的发展不断获得重要意义，这一点尤为重要。驱动结合过程的分子力的详细表征包括许多方面，这些方面已被讨论为胰蛋白酶和 BPTI 结合以及一般蛋白质复合物形成的核心。在这项研究中，我们将所有这些方面结合到一个全面的蛋白质识别模型中。因此，我们有助于增强我们对这一基本机制的一般理解，随着生物制药的发展不断获得重要意义，这一点尤为重要。驱动结合过程的分子力的详细表征包括许多方面，这些方面已被讨论为胰蛋白酶和 BPTI 结合以及一般蛋白质复合物形成的核心。在这项研究中，我们将所有这些方面结合到一个全面的蛋白质识别模型中。因此，我们有助于增强我们对这一基本机制的一般理解，随着生物制药的发展不断获得重要意义，这一点尤为重要。