Comprehensive Analysis of Type 2 Innate Lymphoid Cell ( <scp>ILC2</scp> ) Heterogeneity and Identification of <scp>CD81</scp> as a Marker for Pathogenic <scp>IL</scp> ‐13 ⁺ <scp>ILC2s</scp> in Atopic Dermatitis Skin

Atopic dermatitis (AD) is a chronic inflammatory skin disease characterized by type 2 immune responses, in which group 2 innate lymphoid cells (ILC2s) contribute via IL-13 and IL-5 production [1]. While ILC2s are known contributors to epithelial barrier dysfunction and type 2 inflammation, their pathogenic subsets remain poorly defined [2]. To address this, we performed single-cell RNA sequencing (scRNA-seq) on Cd45+ lineage-negative (Lin−) skin-infiltrating cells from two AD murine models: IL-18 transgenic (IL18Tg) [3] and caspase-1 transgenic (KCASP1Tg) [4] mice (Supporting Information S1). The overall t-SNE plot of all skin cells obtained from sorted Lin−CD45+ cells is shown in Figure S1A. Although cells were FACS-sorted to match the ILC definition, substantial contamination remained. We therefore reapplied the ILC definition in silico, yielding the purified ILC population shown in Figure S1B, in which ILC2s formed a distinct cluster for further analysis. As shown in Figure 1A–C, ILC2 clusters segregated into six transcriptionally distinct clusters (Clusters 0–5). These subsets reflect functional heterogeneity, including resident-like, inflammatory-responsive, type 2 cytokine-producing, IL-18–responsive, immature-like, and ILC1-like states (Supporting Information S2). Among them, Cluster 2 was characterized by high IL-13 production and is likely to represent a pathogenic ILC2 population. Cluster 5 exhibited ILC1-like features (e.g., Ifng+Prf1+Gzmc+), raising the possibility of contamination by non-ILC2 subsets. Next, we focused on two key cytokine pathways that regulate ILC2 activation. IL-33–ST2 (Il1rl1, interleukin 1 receptor–like 1) and IL-18–IL18R (Il18r1, interleukin 18 receptor 1) pathways, which are involved in barrier disruption and chronic inflammation, respectively. Il13 expression was highest in Il18r1+Il1rl1+ cells, and Il18r1−Il1rl1+ cells also showed elevated Il13 levels (Figure 1D). Il18r1 was enriched in Clusters 3–5, while Il1rl1 was higher in Clusters 0–3 (Figure 1E,F). Although Cluster 2 showed the highest Il18r1+Il1rl1+ frequency, Clusters 0, 1, and 3 also exhibited co-expression, indicating overlap with the pathogenic subset (Figure 1G). While we did not perform trajectory analysis due to limited cell numbers per cluster, the observed co-expression patterns imply potential developmental relationships. To identify a more selective marker for the pathogenic subset, we found Cd81 as a defining feature. Cd81, a member of the tetraspanin family known for T and NK cell activation [5], was significantly upregulated in Cluster 2 and in Il13+ ILC2s compared to Il13− ILC2s (Figure 2A,B). Further analysis showed that Cd81 expression was highest in Il5+/Il13+ ILC2s, and reduced in Ifng+ ILC1s but partially retained in Ifng+ or Il17a+ ILC2s (Figure 2C). To validate this at the protein level, we analyzed CD81 expression in skin and draining lymph nodes (dLNs) from IL18Tg mice (n = 18). CD81 was predominantly expressed in IL-13+ ILCs, with significantly higher levels than in IFN-γ+ ILCs in dLNs (Figure 2D). CD81 marks type 2 cytokine–producing ILC2s, likely pathogenic in chronic skin inflammation. To assess clinical relevance, we analyzed public scRNA-seq data from AD patients before and after Dupilumab treatment (n = 1 untreated, n = 5 treated for 16 weeks, n = 4 treated for 1 year) [6]. Both ILCs and ILC2s gradually declined after treatment, and ILC2s were no longer detected after 1 year of Dupilumab therapy (Figure 2E,F). IL13 expression in CD81+ ILC2s was reduced at 16 weeks. Although not statistically significant because of the small sample size, a trend toward higher IL13 expression was observed in CD81+ cells (Figure 2G). In summary, CD81 is a novel surface marker of pathogenic ILC2s, enriched in IL-13+/IL-5+ subsets and conserved across murine and human skin. These findings suggest that CD81 may serve as a disease activity marker and a potential therapeutic target in AD. We acknowledge Bangert C, Rindler K, Krausgruber T et al. for their study “Persistence of mature dendritic cells, T(H)2A, and Tc2 cells characterize clinically resolved atopic dermatitis under IL-4Rα blockade” (Sci Immunol 2021; 6), which provided the scRNA-seq data used in our analysis of CD81 expression and cytokine production in human skin ILC2s. This work was supported in part by Japan Society for the Promotion of Science (JSPS) 24K11464forYI and 25K11582forKY. The authors declare no conflicts of interest. The processed scRNA-seq count matrices generated in this study have been deposited in the NCBI Gene Expression Omnibus (GEO) under BioSample accession numbers SAMN50536478 and SAMN50536479. Figure S1: Single-cell RNA-seq analysis of skin ILCs from KCASP1Tg and IL18Tg mice. (A) t-SNE plot of all analyzed skin cells (10,217 cells) obtained from FACS-sorted Lin−CD45+ cells, including 1714 cells from KCASP1Tg mice (n = 45) and 8503 cells from IL18Tg mice (n = 75). Although sorting was based on the ILC definition, substantial contamination remained. (B) t-SNE plot showing the purified ILC population obtained by reapplying the ILC definition in silico. ILCs were defined as CD45+ cells lacking lineage-specific markers for T cells (Cd3e, Cd3d, Cd3g, Cd4, Cd8a, Trgc1), B cells (Cd19, Ms4a1, Pax5), macrophages (Adgre1, Cd68), dendritic cells (Itgax, Zbtb46), neutrophils/plasmacytoid dendritic cells (pDCs; Ly6g, Siglech), and monocytes (Itgam). Among these, ILC2s were further defined as Gata3+ ILCs. The red box highlights the ILC2 population (672 cells), consisting of 37 cells from KCASP1Tg mice and 635 from IL18Tg mice, which was the focus of subsequent analyses. Data S1: Supporting Information 1. Data S2: Supporting Information 2. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.