通过生物模拟纳米颗粒介导的原位合成深度学习设计的结合物-死亡标志物嵌合体，降解GPX4蛋白用于前列腺癌治疗

Abstract

Glutathione peroxidase 4 (GPX4) is a central guardian against ferroptosis and a critical therapeutic target for castration‐resistant prostate cancer (CRPC). However, traditional GPX4 inhibitors are often limited by irreversible covalent binding and systemic toxicity. Here, we utilized the deep learning algorithm Bindcraft to design high‐affinity de novo binders targeting GPX4. To achieve GPX4 degradation, we developed a novel modality termed binder‐degron chimera (bdC), which integrates a de novo‐designed binder with a C‐terminal combinational degron. Mechanistically, bdC recruits the CUL2‐RING E3 ubiquitin ligase complex, leading to proteasome‐mediated GPX4 degradation. To circumvent the delivery barriers inherent to macromolecular therapeutics, the bdC was encoded into a plasmid vector and encapsulated within a biomimetic nanocarrier featuring a PSMA‐targeted and CD47‐camouflaged surface (pc‐CMNP). This platform enables the in situ synthesis of bdCs specifically within prostate cancer cells, driving robust GPX4 degradation and triggering potent ferroptosis. Our work establishes a versatile framework combining deep‐learning generative binders, combinational degrons, and biomimetic nanotechnology to degrade intracellular targets and provides a novel therapeutic modality for CRPC. A deep‐learning‐designed binder is fused to a combinatorial degron to create a binder‐degron chimera (bdC) that recruits the CUL2‐RING ubiquitin ligase complex. Encapsulation in a prostate‐cancer‐targeted, CD47‐camouflaged nanocarrier enables tumor‐specific in situ bdC expression, triggering GPX4 degradation and ferroptosis, thereby establishing a versatile platform for targeted protein degradation in castration‐resistant prostate cancer.