The pathogenesis of inflammatory conditions like sepsis and autoimmune disorders is closely tied to overactive immune responses, with gasdermin D (GSDMD)-mediated pyroptosis being pivotal. Pyroptosis, a pro-inflammatory programmed cell death, is triggered when activated GSDMD's N-terminal fragment (GSDMD-NT) forms pores on mitochondrial and cellular membranes, causing uncontrolled cytokine release.

Recent studies found extracellular vesicles transfer mature GSDMD-NT pores to nearby cells, amplifying pyroptosis—especially in sepsis patients. However, current GSDMD inhibitors only target pore formation, not existing pores. This is clinically critical, as sepsis treatment often starts after pore development.

Urgent need exists for inhibitors blocking formed GSDMD-NT pores, but design is tough. The pores' large size (~21 nm), smooth surface, and dynamic structure hinder targeting. Also, full GSDMD inhibition disrupts its physiological roles, requiring precise regulation instead of total blockade.

To address these challenges, a research team from the Chengdu Institute of Biology of the Chinese Academy of Sciences employed TransForPep—a deep learning-based atomic generative model—to virtually screen peptide molecules capable of inhibiting mature GSDMD-NT pore function. The study was recently published in Nature Immunology.  

The team validated top candidate molecules across cell models, human alveolar lung organoids, and preclinical disease models, ultimately identifying a lead peptide: SK56. A key advantage of SK56 is its selectivity: it targets GSDMD-NT pores without interfering with interleukin-1β processing or GSDMD activation—two processes critical to normal immune function.

Experimental data confirmed SK56's therapeutic potential. In cell models, the peptide effectively suppressed macrophage pyroptosis and cytokine release, and entered cells after pore formation to reduce mitochondrial damage. Its mechanism of action relies on the cellular ESCRT membrane repair system, while also preventing dendritic cell overactivation and their uptake of GSDMD-NT membrane fragments from pyroptotic cells.

In human alveolar organoid models, SK56 stopped the spread of pyroptosis to neighboring cells. Most notably, in delayed-treatment sepsis models, SK56 improved mouse survival rates—even when administered after the disease model was established. These findings demonstrate that SK56 can block destructive inflammatory responses while preserving immune function.

SK56 represents a new candidate drug for the precise treatment of inflammation-driven diseases, including sepsis, chronic inflammation, and autoimmune disorders, the researchers noted. This research highlights the potential of AI-guided peptide design to target "undruggable" biological structures, opening new avenues for biopharmaceutical development.

This work received support from the National Natural Science Foundation of China's Youth Science Program, the National Defense Outstanding Youth Fund of the Bureau of Science, and other funding sources.