Developing computational methods to identify of endogenous substrates of E3 ubiquitin ligases and molecular glue degraders

项目摘要/摘要： E1-E2-E3连接酶级联负责用泛素标记底物蛋白。添加泛素 然后引导标记的蛋白质沿着几条路径中的一条，包括将其标记为蛋白酶体介导的 退化。E3连接酶负责识别底物蛋白，从而编码特异性 泛素转运蛋白。人类蛋白质组由大约600个已知的E3连接酶组成，每个连接酶都有一个不同的 底物专一性，使其能够结合蛋白质组的指定子集。鉴于其中一个 泛素化的典型后果是标记要销毁的蛋白质，这并不令人惊讶 E3连接酶的失调或突变可导致细胞动态平衡的破坏：因此， E3连接酶与多种疾病有关，包括自身免疫性疾病和癌症。 有趣的是，已发现某些与疾病相关的突变会改变E3的底物特异性。 连接酶-这些突变不仅影响E3‘S正常互作组内的细胞蛋白质水平，而且 相反，它们改变了E3连接酶的相互作用体。类似的影响也被观察到从某些小的 分子，称为分子胶，也可以改变E3连接酶的底物专一性：这些化合物 通常将E3连接酶重定向到泛素化一些“新底物”，最终导致这种降解。 蛋白。因此，分子胶提供了靶向致病蛋白质的可能性，这些蛋白质以前是 被认为是无法下药的。然而，到目前为止，E3连接酶与其自身相互作用的暂时性 底物(和新底物)已成为鉴定内源和“可粘性”的瓶颈。 E3连接酶底物。为了解决这个问题，我在这里建议开发基于结构的尖端机器 学习方法：(1)计算确定E3连接酶的内源底物；(2)合理设计 分子胶可以降解传统上不能下药的目标蛋白质。在仔细地对 对于每项任务的底层方法，我将应用前者来全面编目三种底物 与疾病相关的特定E3连接酶。同时，我将应用我对后者的方法来设计分子 旨在降解ADAR1的胶水，ADAR1是一种关键蛋白质，可提高对免疫检查点封锁的抵抗力 因此，它是许多不同癌症干预的潜在靶点。超出了直接的范围 这项建议，我期望通过这些研究制定的方法将有助于阐明潜在的 许多其他E3连接酶的生物学特性，并将促进针对关键基因的分子胶降解物的开发 许多其他疾病的司机。

Project Summary/ Abstract: The E1-E2-E3 ligase cascade is responsible for tagging substrate proteins with ubiquitin. Addition of ubiquitin then directs the tagged protein along one of several paths, including marking it for proteasome-mediated degradation. The E3 ligase is responsible for recognition of substrate proteins, and thus encodes the specificity of ubiquitin transfer. The human proteome comprises about 600 known E3 ligases, each with a distinct substrate specificity that allows it to engage a prescribed subset of the proteome. Given that one of the prototypical consequences of ubiquitination is to mark a protein for destruction, it is unsurprising that dysregulation or mutation of E3 ligases can lead to a disruption of cellular homeostatic balance: accordingly, E3 ligases have been implicated in a wide variety of diseases including autoimmune disease and cancer. Intriguingly, certain disease-associated mutations have been found to alter the substrate specificity of an E3 ligase – these mutations not only impact the cellular levels of proteins within the E3’s normal interactome, but rather they change the E3 ligase’s interactome. Similar effects have also been observed from certain small molecules, termed molecular glues, that also modify the substrate specificity of an E3 ligase: these compounds typically redirect an E3 ligase to ubiquitinate some “neo-substrate”, ultimately leading to degradation of this protein. Thus, molecular glues afford the possibility of targeting disease-causing proteins that were previously thought to be undruggable. To date, however, the transient nature of E3 ligase’s interactions with their substrates (and neo-substrates) has served as a bottleneck for identifying both endogenous and “glue-able” substrates of E3 ligases. To address this, here I propose to develop cutting-edge structure-based machine learning methods to (1) computationally identify endogenous substrates of E3 ligases, and (2) rationally design molecular glues that degrade a traditionally undruggable target protein. After carefully benchmarking the underlying methods for each task, I will apply the former to comprehensively catalog substrates of three specific disease-relevant E3 ligases. In parallel, I will apply my approach for the latter to design molecular glues intended to degrade ADAR1, a key protein that promotes resistance to immune checkpoint blockade therapy and is thus a potential target for intervention in many different cancers. Beyond the immediate scope of this proposal, I anticipate that the methods developed through these studies will help illuminate the underlying biology of many other E3 ligases, and will facilitate development of molecular glue degraders targeting key drivers in many other diseases.