Programmable peptide-guided protein degradation

摘要 最近，超过600种人类蛋白质被列为关键的癌症靶点，其中近一半是 由于靶标的不稳定和活性，被标准小分子抑制方法认为是“难治的” 站点可访问性限制。通过重定向靶蛋白的泛素-蛋白酶体途径(UPS) 蛋白水解靶向嵌合体(PROTAC)技术提供了一种潜在的解决方案，使 快速、持续的靶向消耗以及比小分子更强的药理作用 抑制力。尽管如此，PROTAC与小分子一样面临着相似的发育障碍，而且不可能 轻松设计用于Motif或翻译后修饰特定的目标。为了解决这些障碍， 通过引入蛋白质介导的概念，研究努力转向了基因治疗方法。 蛋白质降解。在这里，E3泛素连接酶通过取代它们的天然底物结合来重定向 具有“现成”结合域的结构域，包括纳米抗体、抗体和DARPins，以产生 靶标特异性泛体。为了加强这个平台，我们最近利用了天然的蛋白质-蛋白质相互作用 用于开发算法流水线的信息，该算法流水线对靶标选择性多肽进行优先排序，这些多肽可以与 E3泛素连接酶共轭结构域诱导靶蛋白降解。在这个项目中，我们将增强 我们目前的方法能够为任何蛋白质开发这些泛体(UAbs)，包括那些 以小分子为基础的手段被认为是“难治的”。要做到这一点，我们将自动执行两部分算法 利用蛋白质语言建模的最新进展以及现有的共同复合体的管道 设计不同蛋白质靶标的多肽结合剂的数据库，包括那些具有已溶解共晶体的多肽结合剂 作为那些结构信息最少的生物。具体地说，我们的流水线将以用户指定的目标蛋白为 输入，并生成候选多肽结合子的优先列表作为输出，从而实现后续生成 用于靶标降解的UABs。通过基于库对库荧光的人类细胞和 随后将UAB mRNA包裹在脂质纳米粒(LNPs)中，我们将开发一种可扩展的方法来 测试和翻译我们的降解物以进行下游的活体验证。总而言之，这项工作将产生一个稳健的 多肽设计工具，将加强有针对性的蛋白质降解努力，并为 可编程蛋白质组编辑。

Summary Over 600 human proteins have been recently prioritized as key cancer targets, with nearly half being considered ‘intractable’ by standard small-molecule inhibition approaches, due to target instability and active site accessibility constraints. By redirecting the ubiquitin-proteasomal pathway (UPS) for targeted protein degradation, the proteolysis-targeting chimera (PROTAC) technology provides a potential solution, enabling rapid and continuous target consumption as well as the stronger pharmacological effects than small molecule inhibition. Nonetheless, PROTACs suffer from similar developmental hurdles as small molecules and cannot be easily designed for motif or post-translational modification-specific targeting. To address these hurdles, research efforts have shifted toward gene therapy approaches by introducing the concept of protein-mediated protein degradation. Here, E3 ubiquitin ligases are redirected by replacing their natural substrate binding domains with “off-the-shelf” binding domains, including nanobodies, antibodies, and DARPins, to generate target-specific ubiquibodies. To augment this platform, we recently exploited natural protein-protein interaction information to develop algorithmic pipelines that prioritize target-selective peptides which can be fused to the E3 ubiquitin ligase conjugation domains to induce target protein degradation. In this project, we will augment our current methods to enable the development of these ubiquibodies (uAbs) for any protein, including those deemed ‘intractable’ by small molecule-based means. To do this, we will automate a bipartite algorithmic pipeline that leverages recent advancements in protein language modeling as well as existing co-complex databases to design peptide binders to diverse protein targets, including those with solved co-crystals as well as those with minimal structural information. Specifically, our pipeline will take user-specified target proteins as inputs, and generate prioritized lists of candidate peptide binders as outputs, enabling subsequent generation of uAbs for target degradation. Through library-on-library fluorescence-based assays in human cells and subsequent encapsulation of uAb mRNA in lipid nanoparticles (LNPs), we will develop a scalable method to test and translate our degraders for downstream in vivo validation. In total, this work will generate a robust peptide design tool that will enhance targeted protein degradation efforts and lay the foundation for programmable proteome editing.