Single cell transcriptional zonation of human psoriasis skin identifies an alternative immunoregulatory axis conducted by skin resident cells

Abstract

Psoriasis is the most common skin disease in adults. Current experimental and clinical evidences suggested the infiltrating immune cells could target local skin cells and thus induce psoriatic phenotype. However, recent studies indicated the existence of a potential feedback signaling loop from local resident skin cells to infiltrating immune cells. Here, we deconstructed the full-thickness human skins of both healthy donors and patients with psoriasis vulgaris at single cell transcriptional level, and further built a neural-network classifier to evaluate the evolutional conservation of skin cell types between mouse and human. Last, we systematically evaluated the intrinsic and intercellular molecular alterations of each cell type between healthy and psoriatic skin. Cross-checking with psoriasis susceptibility gene loci, cell-type based differential expression, and ligand-receptor communication revealed that the resident psoriatic skin cells including mesenchymal and epidermis cell types, which specifically harbored the target genes of psoriasis susceptibility loci, intensively evoked the expression of major histocompatibility complex (MHC) genes, upregulated interferon (INF), tumor necrosis factor (TNF) signalling and increased cytokine gene expression for primarily aiming the neighboring dendritic cells in psoriasis. The comprehensive exploration and pathological observation of psoriasis patient biopsies proposed an uncovered immunoregulatory axis from skin local resident cells to immune cells, thus provided a novel insight for psoriasis treatment. In addition, we published a user-friendly website to exhibit the transcriptional change of each cell type between healthy and psoriatic human skin.

Fig. 1. Clustering, marker genes and cell type assignment of human full-thickness skin scRNAseq from healthy and psoriatic donors.

A Schematic view of the workflow. B Left, human full-thickness skin cell types. Visualization using UMAP of skin cells, each point represents one cell. Color coding based on unique cell clusters, as shown at right beside. Right, UMAP plot was folked from left side and colored by samples. Ctrl 1–3, samples from healthy donors; Psor 1–3, samples from patients with psoriasis vulgaris. C Expression of selected top significant marker genes for five main cell type clusters; relative expression of each gene in each cluster was visualized as violin plot, and the raw counts of each gene were visualized as sparkline plot in that violin plot. D Radar plot visualization of cell-type probabilistic scores of human full-thickness skin cells in relation to our defined cell types. Each dot represented one cell. Color coding based on unique cell clusters. The position of each dot indicated the cell-type score between that cell and the training (defined) cell types which were indicated outside of each bend in the radar wheel. Almost all cells were correctly assigned to their defined cell types.

Fig. 2. Prediction of the anatomical location and evolutional conversation of human skin cell type.

A Radar plot visualization of cell-type probabilistic scores of published human trunk epidermis skin cells in relation to our defined cell types. B Radar plot visualization of cell-type probabilistic scores of published human full-thickness skin cells in relation to our defined cell types. C Violin plot visualization of cell-type probabilistic scores of published human trunk epidermal skin cells in relation to our defined cell types. X-axis represented our defined cell types, y-axis represented defined cell types from original study. Red frame indicated low similar (absent) cell types of our defined cell types in epidermal skin cell types. D Radar plot visualization of cell-type probabilistic scores of published mouse skin atlas cells in relation to our defined human skin cell types. E Radar plot visualization of cell-type probabilistic scores of published mice healthy or psoriatic skin cells in relation to our defined human skin cell types. For all radar plots, each dot represented one cell. Color coding based on their original defined cell clusters or disease condition. The position of each dot indicated the cell-type score between that cell and the training (defined) cell types which were indicated outside of each bend in the radar wheel.

Fig. 3. The difference of cellular composition and cell-type-based transcription in psoriasis.

A Percentage bar plot of cell type composition in healthy and psoriatic skin, y-axis represented the percent of total cell number of that cell type. Red color represented psoriatic skin and blue color represented healthy control skin. Significant changes (FDR < 0.01) were labeled as ** on the top. B Dot plot of cell number distribution of each cell type in each healthy and psoriasis donor. Dot size represented the cell number percent of that donor (x-axis) in that cell type (y-axis). Color indicated the cell types. C Dot plot of the significant existence of each susceptibility gene loci (x-axis) in each cell type (y-axis). Color indicated each cell type. D Bar plot of the raw counted of four selected HLA genes in cell type EpD_Foli. Red color represented psoriatic skin and blue color represents healthy control skin. Y-axis represents the mean value of the expression raw counts. ***, FDR p < 0.001; *, p < 0.05. E, F GO-bioprocess enrichment plot of all psoriasis upregulated genes in cell type EpD_Granular/spinous (E) and Mes_Fibro (F). Red frame highlighted the significant enriched GO terms and the terminally enriched GO-bioprocess were enlarged at the bottom of each plot.

Fig. 4. Molecular interactions among cell types and the regulatory potentiality from epidermal/mesenchymal cells to dendritic cells during psoriasis.

A Chord plot visualization of the ligand-receptor interactions among cell types. The defined cell types were indicated outside of each bend. Arc line between each cell types represented the ligand to receptor interactions, and the color indicated the ligand expressed cell types. B Chord plot visualization of the ligand-receptor interactions between top scored resident cell types and type 1 DC. The defined cell types were indicated outside of each bend. Arc line between each cell types represented the ligand to receptor interactions, and the colors indicated the ligand expressed cell types. C Sankey plot visualization of the interactions from the ligand of mesenchymal type cells to the receptors of type 1 DCs. The IL-17B/IL-17RB pair was highlighted as red. D Immunofluorescence staining of the epidermis/dermis junction of human skins, PDGFRB protein was stained as red, IL-17B as green, and DAPI as blue, the cellular co-expressions of PDGFRB and IL-17B were observed at the dermis layer close to the junction (pointed by white arrow) and at the layer junction. Scale bar, 100 µM. E, F Immunohistochemistry staining and quantification of PDGFRB (E) and IL-17B (F) in healthy and psoriatic skins. Host IgG of primary antibodies were used as negative control. In dermis layer, the average PDGFRB protein amount didn’t significantly change (E), but IL-17B protein significantly increased in psoriasis (F). Scale bar, 100 µM. Ctrl N = 6; Psor N = 10. **, p < 0.01. G, H The secreted protein levels of CXCL8, CCL20, IL-6, IL-1B, and IL-17B in skin fibroblasts (G) or skin basal cells (H) under TNF-α treatment were determined by ELISA. The protein concentrations were normalized to control group. N ≥ 3; *, p < 0.05; NS not significant. i Flow cytometry histograms of CD80 and CD86 expressions in DCs induced by IL-17B. LPS treatment was used as positive control.

Fig. 5. Schematic view of the immune regulation from local resident epidermal cells and mesenchymal cells to DCs.

Resident cells including basal cells, hair follicle-like cells, mesenchymal cells potentialize local inflammation by evoking MHC complex genes in psoriatic condition, and these cells can trigger dendritic cells for the inflammatory cascade through remodeling ECM and secreting cytokines such as LIF, IL-6, IL-17B, IL-36, and CD58.