Machine learning-guided composite ionic liquid-based system for dual-drug delivery targeting redox homeostasis and STAT3-PI3K axis in psoriasis therapy.

Abstract

Excessive accumulation of reactive oxygen and nitrogen species (RONS) exacerbates inflammatory responses and contributes to the progression of psoriasis. In particular, ROS activate the STAT3 pathway, inducing abnormal proliferation of keratinocytes and aggravating local inflammation. Moreover, interactions between macrophages and keratinocytes can further exacerbate disease progression. However, current therapeutic strategies have limited efficacy due to poor transdermal permeability and insufficient target specificity. To address these limitations, we have developed a machine learning (ML)-guided framework that integrates virtual screening, experimental validation, and mechanistic analysis into the design of transdermal ionic liquids (ILs). Using this approach, we successfully identified highly efficient transdermal ILs and developed a composite ionic liquids (CIL) delivery system capable of releasing H₂S. This CIL platform enables the co-delivery of the APTSTAT3-9R peptide and catalase (CAT) directly to psoriatic lesions, implementing a dual therapeutic strategy: (1) inhibition of STAT3 phosphorylation to suppress keratinocyte hyperproliferation, and (2) regulation of redox homeostasis and macrophage polarization via local release of H₂S and CAT. In vivo studies have shown that CIL@CA can effectively alleviate IMQ-induced psoriasis symptoms in mice. In this study, a novel ML-driven ILs-based drug delivery system was developed, offering a promising strategy for the treatment of inflammatory skin diseases.