Online Poster Portal

Atopic dermatitis (AD) is a heterogeneous inflammatory skin disease in which pathway-selective therapies can produce divergent clinical responses, suggesting that distinct immune architectures may determine therapeutic vulnerability. Although type 2 cytokine blockade has transformed AD care, many patients have partial, inadequate, or non-durable responses, highlighting the need to define inflammatory circuits beyond the canonical type 2 axis. Here, we integrated clinically curated human single-cell RNA sequencing data with state-matched allergen-based murine dermatitis models to identify immune and metabolic programs that license therapeutic response.

Across 14 public human skin single-cell RNA sequencing datasets, we identified 32 clinically curated active lesional AD samples and 12 healthy controls suitable for inflammatory-state analysis. Within the helper-like T-cell compartment, lesional AD resolved along a reproducible AD2–AD3 inflammatory axis. AD2-like lesions were enriched for GATA3-positive T helper 2 (Th2)-like cells and type 2 immune programs, whereas AD3-like lesions were enriched for RORC-positive Th17-like cells and interleukin-17 (IL-17)-associated type 3 programs. Network-level analysis linked these states to distinct cytokine circuits, intercellular communication programs, epithelial response modules, and metabolic signatures. Benchmarking against psoriasis supported AD3-like disease as a type 3-polarized AD state rather than a psoriasis-equivalent profile.

To functionally test these inflammatory architectures, we established complementary house dust mite (HDM)-based murine dermatitis models. HDM/alum sensitization generated type 2-dominant dermatitis aligned with human AD2-like inflammation, whereas HDM/complete Freund’s adjuvant sensitization generated type 3-biased dermatitis aligned with human AD3-like inflammation. Disruption of interleukin-4 and interleukin-13 signaling attenuated AD2-like dermatitis but failed to suppress and modestly augmented AD3-like pathology. In contrast, AD3-like inflammation depended on the interleukin-23/JAK2/interleukin-17 axis, mitochondrial oxidative phosphorylation, and carnitine palmitoyltransferase 1-dependent fatty acid β-oxidation in effector CD4⁺ T cells. Accordingly, targeting interleukin-17/interleukin-23 signaling, Janus kinase 2 activity, or fatty acid metabolism ameliorated AD3-like inflammation.

Together, these findings establish AD2-like and AD3-like inflammation as therapeutically distinct immune-metabolic architectures rather than descriptive transcriptomic states. By linking human single-cell inflammatory programs to state-matched murine perturbation models, this work identifies cytokine-, JAK2-, and metabolism-based vulnerabilities that may guide molecular stratification and precision therapeutic development in AD.