γ分泌酶催化过程中底物识别及进入机制的多尺度计算模拟研究

Gamma-secretase is a promiscuous protease that cleaves transmembrane proteins in membrane. This enzyme plays a key role in the pathology of the Alzheimer's disease (AD) by releasing pathogenic amyloid-beta peptides from their precursor proteins. Although gamma-secretase is an appealing inhibition target for treatment of AD, design of effective gamma-secretase-based inhibitors is hurdled by the lack of an atomic picture of how the enzyme works. In particular, it remains unknown how the enzyme recognizes its substrates and how the bound substrate enters the catalytic site of the enzyme. In this research, we will employ molecular dynamics (MD) simulations to characterize structures and structural evolution of the enzyme and its substrates during the processes of the substrate recognition and substrate entry. MD simulations of these processes are difficult due to demanding computational cost for sampling. Thus, multiple complementary approaches will be applied to overcome the difficulty. We will combine a hybrid-resolution model with enhanced sampling techniques and kinetic network analysis to study the substrate-recognition mechanism and to identify substrate-specific patterns of recognition. We will also develop an efficient method to explore enzyme conformations that are ready for substrate entry. Then the Markov state model will be employed to study how the enzyme undergoes structural transitions for the substrate entry. In addition, we will combine the hybrid-resolution model and enhanced sampling techniques to construct a structural ensemble of a loop in the enzyme that is known to inhibit the enzymatic activity. The constructed structural ensemble will be employed further to examine if the loop blocks the access of substrate or water. The studies proposed here will not only generate new insights into the catalytic mechanism of gamma-secretase and into inhibitor design, but will also reveal novel mutations for experimental test in future.

γ分泌酶与阿尔茨海默症密切相关，它通过切割蛋白释放出可以致病的多肽。由于对酶工作机制不甚了解，至今尚未成功设计出有效的γ分泌酶抑制剂。目前还不清楚γ分泌酶识别底物蛋白的机制和底物到达催化位点的过程。在本研究中，我们将运用分子动力学模拟来表征这两个过程中酶和底物构象变化的原子细节。为了克服传统方法下模拟成本过分高昂这一困难，我们将会采用多个互补的办法来开展本项目。首先，我们将结合混合尺度模型、增强抽样和动力学网络分析方法，来研究酶识别特异底物的机制。我们还将发展方法用来寻找可供底物进入的酶的构象，并运用马尔科夫状态模型探索底物进入过程中γ分泌酶构象的动态转变。我们还将结合混合尺度模型和增强抽样方法，表征一段可抑制酶活力的柔性区域结构，并基于这些结构，评估该区域是否能阻止底物或者溶剂水的进入。上述研究不但可以对γ分泌酶催化机制的理解及抑制剂的设计提供新思路，而且还能发掘全新突变供实验验证。

本项目的主要目的是通过计算模拟阐明γ-分泌酶催化反应过程中底物与酶结合，进入活性口袋，并转变为反应前置状态构象等一系列过程的分子机制。通过粗粒化模拟，我们发现底物APP首先与γ-分泌酶亚结构域PS1的延伸表面结合，并存在多个结合状态，而底物跨膜螺旋的柔性程度对APP能否识别PS1的碳端结构域至关重要。另外，另一种底物NOTCH虽然具有刚性跨膜螺旋，但是由于其序列与APP不同，仍能够很好的与碳端结构域结合，因此这些发现加深了对γ-分泌酶既能够识别多种底物又具有底物选择性这一特性的机制理解。另外，为了研究底物进入过程，我们研发了一套能够处理柔性跨膜螺旋的通用粗粒化模拟方法。通过该方法，我们发现PS1中TMD2的开关运动对于底物进入尤为重要；此外，我们还发现磷脂膜中的脂质分子会和底物竞争结合口袋，这表明结合口袋的排空对于酶催化性能也十分重要。我们进一步发现当底物进入结合口袋后，其切割位点的酰胺键可以沿肽链方向旋转，与催化残基D257和D385中质子化的那一个残基的羧基侧链的氧原子形成氢键，而另一个催化残基则与切割位点形成水分子介导的相互作用。这表明两个催化残基均有可能帮助底物进入反应前置状态，以便酶进行后续水解反应。最后，我们还运用多尺度模拟研究TMD6和TMD7之间柔性区域构象分布，以期理解该区域对酶活性自抑制的机制。但是该区域构象十分复杂，我们为此研发了一套混合尺度的通用模型，可以准确采集多种水溶液中以及膜内蛋白质柔性区域的构象。该方法将会在后续研究中用来解答上述机制问题。总体而言，本项目较为深入地揭示了γ-分泌酶底物识别及进入这一过程不同方面的分子机制，并且开发了分别针对跨膜螺旋柔性区域和膜外柔性区域的通用分子模拟模型。这些工作不但为后续γ-分泌酶的机制研究和调节剂设计提供思路，也对其他跨膜蛋白机制的计算模拟提供方法学的支持。