Dupilumab Effects on Innate Lymphoid Cell and Helper T Cell Populations in Patients with Atopic Dermatitis

Group 2 innate lymphoid cells (ILCs) are thought to contribute to the pathogenesis of atopic dermatitis (AD). IL-4 stimulates T helper type 2 (Th2) cells and ILC2s to proliferate and produce cytokines. Dupilumab, an antibody against the IL-4 receptor, is used in AD therapy. We speculated that its efficacy might involve blocking the activation of Th2 cells and ILC2s via IL-4. Here, we examined circulating Th2 cells and ILC2s in 27 Japanese patients with AD before and after the administration of dupilumab. Between 0 and 4 months after dupilumab administration, the percentages of Th2 cells and ILC2s were decreased. Notably, ILC2/3 ratio was decreased after dupilumab treatment. Interestingly, ILC2/3 ratio before dupilumab treatment were significantly higher in high responders than in low responders to dupilumab. To resolve the molecular signatures of the Th2 and ILC2s in AD, we sorted CD4+ T cells and ILCs from peripheral blood and analyzed their transcriptomes using the BD Rhapsody Single-cell RNA sequencing system. Between 0 and 4 months after dupilumab administration, the Th2 and ILC2 cluster gene signatures were downregulated. Thus, dupilumab might improve dermatitis by suppressing the Th2 cell and ILC2 populations and altering the Th2 and ILC2 repertoire in patients with AD.


INTRODUCTION
Group 2 innate lymphoid cells (ILCs) are thought to contribute to the pathogenesis of atopic dermatitis (AD) (Imai, 2019). ILC1s (NK cells), ILC2s, and ILC3s mirror the corresponding T helper type (Th) 1, Th2, and Th17 cells, respectively. Previously, we and others reported that IL-4 stimulates Th2 cells and ILC2s to proliferate and produce type 2 cytokines Motomura et al., 2014). Dupilumab, an antibody against IL-4 receptors, is widely used in AD therapy (Beck et al., 2014;Guttman-Yassky et al., 2019), although we speculated that its efficacy in AD treatment might involve blocking the proliferation and activation of Th2 cells and ILC2s via IL-4. A recent report demonstrated a gradual decrease in the percentage of Th2 cells after dupilumab treatment (Trichot et al., 2021). However, the effects of dupilumab on ILC2s are not fully understood.

RESULT
In this study, we examined the proportion and number of circulating Th2 cells and ILC2s in 27 Japanese patients with AD before and after dupilumab administration (details of patient information are provided in Table 1). Treatment began with a 600-mg loading dose of dupilumab, followed by 300 mg dupilumab every other week, combined with topical corticosteroids and/or tacrolimus. Eczema Area and Severity Index was measured to assess disease activity, and blood was collected at the baseline visit (before therapy) and week 16, with a window of assessment of approximately 7 days. Dupilumab treatment significantly improved dermatitis and decreased total serum IgE and serum CCL17 (also known as TARC) levels in patients with AD, although some patients were low responders ( Figure 1a). We used flow cytometry to assess the populations of Th2 cells (CD4 þ , CCR6 À , CXCR3 À , and CCR4 þ ) and ILC2s (Lin À , CD127 þ , and CRTH2 þ ) (Figure 1b and c). Between 0 and 4 months after dupilumab administration, the percentage of Th2 cells (among total CD4 þ T cells) and ILC2s (among all ILCs) were decreased from 16.4 AE 7.6% to 14.5 AE 5.8% (mean AE SD) and 28.1 AE 16.1% to 21.2 AE 12.1%, respectively (Figure 1d and e). The absolute numbers of Th2 cells and ILC2s were significantly decreased. Both Th2 cells and ILC2s were more depleted in high responders than in low responders to dupilumab, suggesting that these cells are important in IL-4 receptor signaling pathways in AD pathogenesis. The percentage and absolute number of Th17 cells tended to increase, although this was not statistically significant ( Figure 1f). The percentage and absolute number of ILC3s increased significantly after dupilumab treatment (Figure 1g). The absolute numbers of Th1 cells slightly tended to decrease (Figure 1h), and the percentage and absolute number of ILC1s were unchanged after dupilumab treatment ( Figure 1i).
Next, we examined Th2/17 and ILC2/3 balance. Th2/ Th17 ratio tended to decrease, although it was not statistically significant (Figure 2a). ILC2/3 ratio was very  Th2 Th1 Th17   Th2/17  ratio Th2 Th1 Th17 Th2 Th1 Th17 ILC2 ILC1 ILC3   ILC2/3  ratio ILC2 ILC1 ILC3 ILC2 ILC1 ILC3 TARC (pg/ml) significantly decreased after dupilumab treatment ( Figure 2b). ILC2/3 ratio and the absolute numbers of ILC2s before dupilumab treatment were significantly higher in high responders than in low responders to dupilumab (Figure 2c and d), suggesting that dupilumab might be more effective in patients whose immune balance is originally skewed toward ILC2. Transcriptomic analyses of pretreatment and posttreatment bulk tissue skin biopsy specimens from patients with AD treated with dupilumab have been reported ; however, these analyses did not distinguish the cell types. By contrast, single-cell RNA sequencing (scRNA-seq) is able to uncover cell-specific changes in gene expression. To resolve the molecular signatures of Th2 cells and ILC2s from patients with AD before and after dupilumab treatment, we sorted CD4 þ T cells and ILCs from peripheral blood and analyzed their transcriptomes using the BD Rhapsody scRNA-seq analysis system (BD Biosciences, San Jose, CA) ( Figure 3aeh) (Hasegawa et al., 2019). By using hierarchical clustering as described in Materials and Methods, T cells were split into four clusters, namely, naive T, Th1/17, Th2, and regulatory T cells ( Figure 3aec). Between 0 and 16 weeks after dupilumab administration, Th2 cluster gene signatures were remarkably changed ( Figure 3d). ILCs cells were grouped into three clusters: NK/ILC1, ILC2, and ILC3 (Figure 3eeg). Between 0 and 16 weeks after dupilumab administration, ILC2 cluster gene signatures were downregulated (Figure 3h). These findings suggest that dupilumab administration affects both Th2 and ILC2 gene signatures.

DISCUSSION
There are a few reports of scRNA-seq analysis of AD skin (He et al., 2020;Rojahn et al., 2020), but none has identified cell clusters corresponding to Th2 cells and ILC2s because those studies analyzed all cell types, including keratinocytes, fibroblasts, and lymphocytes. Therefore, we solved this problem by initially sorting CD4 þ T cells and ILCs and subsequently investigated isolated cells by using scRNA-seq.
The study has some limitations. We were not able to separate Th1 and Th17 populations by scRNA-seq analysis. Therefore, we used both flow cytometry and scRNA-seq analysis. The number of patients recruited is small, thus implying the need to validate our findings in a larger cohort of patients. In addition, future studies should investigate how ILC2 changes when AD is treated with antieIL-13 antibody or Jak inhibitor.
Our results demonstrating dupilumab's tendency to decrease Th2 cells and increase Th17 cells in the peripheral blood of patients with AD are consistent with those of a previous report (Trichot et al., 2021). We further showed by scRNA-seq analysis that dupilumab administration significantly altered Th2 cluster gene signatures. Increased frequency of circulating ILC2s in AD was previously reported (Mashiko et al., 2017); in vitro or animal experiments demonstrated that IL-4 is involved in the proliferation and activation of ILC2s (Imai, 2019). However, the actual mechanism has not been clarified in human AD. By comparing the number of ILCs in peripheral blood before and after administration of dupilumab in the same patient, we found that inhibition of IL-4 receptor signal reduces percentages and absolute numbers of ILC2s in peripheral blood of patients with AD. Furthermore, ILC2s were more depleted in high responders than in low responders to dupilumab (Figure 1e), suggesting that activation of ILC2s by IL-4 might be involved in human AD. ILC2s might be an indicator of the effect of dupilumab. By measuring ILC2/ILC3 ratio before administration of dupilumab, it may be possible to predict the effect of dupilumab (Figure 2c). Thus, dupilumab might improve dermatitis by suppressing circulating Th2 cell and ILC2 populations and altering the Th2/Th17 or ILC2/ILC3 repertoire in patients with AD.

Patients
This study was conducted at Hyogo College of Medicine Hospital, Nishinomiya, Japan, to assess the effects of dupilumab in patients with moderate-to-severe AD who visited the hospital from April 2018 to March 2020. This study was approved by the Ethics Review Board of Hyogo College of Medicine (No. 0032 and 0212) and conformed to the Declaration of Helsinki. Blood was collected after obtaining written, informed patient consent. The diagnosis of AD was made according to the Hanifin and Rajka criteria.

Definition of high and low responder to dupilumab
To analyze the factors dictating high or low clinical responsiveness to dupilumab, patients were subgrouped into two groups according to the rank order of percentage change of Eczema Area and Severity Index from baseline at week 16. Based on the classification used in previous reports (Nettis et al., 2020;Tavecchio et al., 2020), 75% of patients in this study were tentatively defined as high responders and the rest (25%) were tentatively defined as low responders ( Figure 1a).
scRNA-seq analysis CD45 þ CD4 þ T cells or Lin À CD45 þ CD127 þ cells were isolated from blood samples using a FACSAria III cell sorter (BD Biosciences). Targeted scRNA-seq analysis was performed using the BD Rhapsody Single-Cell Analysis System (BD Biosciences), according to the manufacturer's instructions. For the library construction, we used the BD Human Single-Cell Multiplexing Kit (#633781, BD Biosciences) and the BD Rhapsody Immune Response Targeted Panel for Human (#633750, BD Biosciences), which consisted of primer sets for 399 genes. Sequencing was performed using an Illumina HiSeq X (Illumina, San Diego, CA), and the fastq files were converted using BD Rhapsody Analysis Pipeline (BD Biosciences) and analyzed using the BD DataView software v.1.2.2 (BD Biosciences).

Hierarchical clustering
Hierarchical clustering is a method of cluster analysis that attempts to build a hierarchy of clusters. All analyses were performed in BD DataView software using MATLAB (R2014a, MathWorks, Natick, MA) (Joshi et al., 2017) built-in cluster function with the maxclust option, according to the manufacturer's instructions. The specific formula is described elsewhere (Joshi et al., 2017;Zhang et al., 2018).

Statistical analysis
Data were analyzed using GraphPad Prism version 8 (GraphPad Software, San Diego, CA). A two-tailed unpaired t-test was used for single comparisons. The Wilcoxon matched-pairs signed rank test was used for intragroup comparison. A P-value < 0.05 was considered statistically significant.

Data availability statement
No datasets were generated during this study.