BET-targeting PROteolysis TArgeting Chimeras (PROTACs) outperform their parent inhibitors in preclinical prostate cancer (PCa) models, yet their enhanced potency is expected to amplify dose-limiting on-target toxicities. To widen the therapeutic window, we designed a two-tier selective prodrug strategy enabling glutathione (GSH)-responsive and PCa-targeted delivery of the BET PROTAC MZ1 caged via a carbonate moiety. Our design integrates a prostate-specific membrane antigen (PSMA) ligand with a GSH-cleavable disulfide linker to achieve tumor-associated activation. Systematic optimization of the commonly used but hydrolytically labile carbonate–disulfide motif led to secondary carbonate variant 2a with markedly improved stability. In cellular systems, this analogue demonstrated strong disulfide dependence, confirming tight GSH control over prodrug activation. Importantly, while resistance to premature hydrolysis was significantly enhanced compared to the primary carbonate, GSH-mediated cleavage and the subsequent MZ1-releasing cyclization step proceeded with minimal kinetic penalty. In contrast, selective PSMA-mediated uptake could not be demonstrated. Molecular dynamics simulations revealed unexpected intramolecular folding that generates a compact prodrug conformation in which acidic functionalities are effectively masked. This structural feature may promote passive membrane permeability, which underscores the complexity of dual-targeting strategies. Collectively, our study establishes design principles for stable yet GSH-responsive carbonate–disulfide prodrugs and provides a framework for the rational development of PROTAC prodrugs, emphasizing the importance of coordinated optimization of their individual components to achieve predictable biological behavior.