AWD13299 Admin Supplement to Support Undergraduate Summer Research Experiences

1 R35 GM 146821 -01的项目摘要 尽管蛋白酶在疾病病理生理学中起主要作用，但在疾病诊断和治疗中的相应挑战仍然存在。 蛋白酶药物发现是设计或分离选择性抑制或 激活目标蛋白酶。改进现有的蛋白酶药物和开发新的药物 蛋白酶靶向已被证明是一个反复的、费力的且经常不成功的过程。认识 配体的性质最终决定了其调节功能和结合机制， 这项研究提出了两个假设。首先，如果分子被直接选择 根据大型库的调节功能，它们的性质将直接关系到它们的 功能，而不是其约束力。其次，如果调节剂的结合机制是 确定，配体性质和机制之间的函数关系可以 并可能将这些发现扩展到相关的蛋白酶。拟议研究 追求三个方向，总体目标是改变蛋白酶配体的发现， 蛋白酶生物化学从迭代努力到数据驱动，最终预测 流程.第一个研究方向将建立一个机器学习（ML）引导的高 通量筛选平台，其直接基于 它们如何改变蛋白酶功能。在这里，属性-功能关系将训练机器 用于函数预测和ML引导的库设计的学习算法将显著减少 蛋白酶调节剂的搜索空间，同时探索远端调节多样性 全面地在第二个研究方向中，该平台将被扩展以分离 β-分泌酶和胰岛素降解酶的基于纳米抗体的底物选择性调节剂， 这两种蛋白酶是阿尔茨海默病和2型糖尿病的关键治疗靶点， 分别精细地重编程蛋白酶的底物选择性的能力可以 彻底改变了如何研究和药物多特异性酶，并导致成功的靶向 以前不可用的蛋白酶。第三个研究方向将实现深度突变 扫描协议，以绘制蛋白酶的调节景观，并确定如何 调节剂改变蛋白酶活性和底物选择性。该方法将识别 调节剂的构象表位，绘制耐药性，表征新的远端位点，以及 发现远距离远端通讯总的来说，这些研究的长期回报 是建立基于三元关系的可推广的配体设计准则， 配体性质、结合机制/蛋白酶结构和调节功能，使人们能够 更好地了解蛋白酶如何工作以及如何控制它们。

Project Summary for 1 R35GM146821-01 Although proteases play major roles in disease pathophysiology, a consequential challenge in protease drug discovery is to design or isolate a specific ligand that selectively inhibits or activates a target protease. Improving current protease drugs and developing drugs for new protease targets has proven an iterative, arduous, and often unsuccessful process. Recognizing that the property of a ligand ultimately dictates its modulatory function and binding mechanism, the proposed research postulates two hypotheses. First, if molecules are selected directly based on their modulatory function from large libraries, their properties will directly relate to their function, rather than their binding capabilities. Second, if the binding mechanism a modulator is determined, functional relationships between ligand properties and mechanism can be developed and possibly extended these findings to related proteases. The proposed research pursues three directions, with an overall objective to transform protease ligand discovery and protease biochemistry from iterative endeavors to data-driven, and ultimately predictive processes. The first research direction will establish a machine learning (ML)-guided high- throughput screening platform that isolates protein-based protease modulators directly based on how they alter protease function. Here, property-function relationships will train machine learning algorithms for function prediction and ML-guided library design will significantly reduce the search space for protease modulators while exploring distal regulation diversity more comprehensively. In a second research direction, this platform will be extended to isolate nanobody-based substrate selective modulators of β-secretase and insulin-degrading enzyme, two proteases that are key therapeutic targets in Alzheimer's disease and Type-2-Diabetes, respectively. The ability to finely reprogram the substrate selectivity of proteases can revolutionize how to study and drug polyspecific enzymes and lead to successfully targeting previously undruggable proteases. The third research direction will implement deep mutational scanning protocols to map the modulatory landscape of proteases and determine how modulators alter protease activity and substrate selectivity. This approach will identify conformational epitopes of modulators, map drug resistance, characterize novel distal sites, and uncover long-range distal communication. Taken together, the long-term payoff of these studies is to establish generalizable ligand design guidelines based on ternary relationships between ligand property, binding mechanism/protease structure and modulatory function, enabling one to better understand how proteases work and how to control them.