Inflammation-mediated fibroblast activation and immune dysregulation in collagen VII-deficient skin

Abstract

Inflammation is known to play a critical role in all stages of tumorigenesis; however, less is known about how it predisposes the tissue microenvironment preceding tumor formation. Recessive dystrophic epidermolysis bullosa (RDEB), a skin-blistering disease secondary to COL7A1 mutations and associated with chronic wounding, inflammation, fibrosis, and cutaneous squamous cell carcinoma (cSCC), models this dynamic. Here, we used single-cell RNA sequencing (scRNAseq) to analyze gene expression patterns in skin cells from a mouse model of RDEB. We uncovered a complex landscape within the RDEB dermal microenvironment that exhibited altered metabolism, enhanced angiogenesis, hyperproliferative keratinocytes, infiltration and activation of immune cell populations, and inflammatory fibroblast priming. We demonstrated the presence of activated neutrophil and Langerhans cell subpopulations and elevated expression of PD-1 and PD-L1 in T cells and antigen-presenting cells, respectively. Unsupervised clustering within the fibroblast population further revealed two differentiation pathways in RDEB fibroblasts, one toward myofibroblasts and the other toward a phenotype that shares the characteristics of inflammatory fibroblast subsets in other inflammatory diseases as well as the IL-1-induced inflammatory cancer-associated fibroblasts (iCAFs) reported in various cancer types. Quantitation of inflammatory cytokines indicated dynamic waves of IL-1α, TGF-β1, TNF, IL-6, and IFN-γ concentrations, along with dermal NF-κB activation preceding JAK/STAT signaling. We further demonstrated the divergent and overlapping roles of these cytokines in inducing inflammatory phenotypes in RDEB patients as well as RDEB mouse-derived fibroblasts together with their healthy controls. In summary, our data have suggested a potential role of inflammation, driven by the chronic release of inflammatory cytokines such as IL-1, in creating an immune-suppressed dermal microenvironment that underlies RDEB disease progression.

Keywords: dermal microenvironment; epidermolysis bullosa; fibroblast activation; immune suppression; inflammation; interleukin-1; single-cell RNA sequencing; squamous cell carcinoma.

Figure 1

Single-cell RNA-Seq identifies distinct cell populations and transcriptional changes in Col7a1^−/− mouse skin. (A) UMAP plots showing the cell-type composition of the wild-type (WT) and knockout (KO) mouse samples. 0: keratinocytes (KRT), 1: fibroblasts (FB), 2: chondrogenic fibroblasts (CHFB), 3: vascular endothelial cells (VEC), 4: lymphatic endothelial cells (LEC), 5: perivascular cells (PVC), 6: Schwann cells (SCH), 7: mast cells (MC), 8: lymphocytes (LYM), 9: neutrophils (NEU), and 10: antigen-presenting cells (APC). (B) Dot plots showing the expression of gene markers by each cell type in the dataset. The color represents the mean expression in each cell type, and the size of the dots represents the percentage of cells within the population that express each marker. (C) Barplots showing the number of cells of each cell type in WT and KO mice. ^*p-value < 0.05; ^**p-value < 0.01; ^****p-value < 0.0001; χ² contingency test. (D) Heat map highlighting upregulated genes in KO samples among all cell types. Color reflects the log-fold difference of gene expression in KO compared to WT. Highlighted genes are grouped together by function: carbohydrate metabolism (Carbohydrate met.), hypoxia, lipid metabolism (Lipid met.), amino acid metabolism (AA met.), angiogenesis, and cytokine and chemokine production (Cytokine).

Figure 2

Distinct inflammatory and myofibroblast subsets in RDEB mouse dermal fibroblasts. (A) UMAP plots showing the fibroblast subtypes found in wild-type (WT) and knockout (KO) mice. The four subpopulations of interest are highlighted in color, and the rest of the cells are shown in gray. All the subpopulations are described in Supplementary Figure S3. (B) UMAP plots showing the gene ontology (GO) term scores of hypoxia, glycolysis, stress, and cytokine expression. (C) UMAP plot of fibroblast clusters 0, 1, 2, 3, and 4 with partition-based graph abstraction (PAGA) nodes and edges. Two connected nodes are likely to be related in the differentiation trajectory. (D) Heatmap of key fibroblast trajectory genes with cells ordered based on trajectory distance. (E–J) UMAP plots of genes involved in pattern recognition receptors (PRRs) (E), encoding receptors for key inflammatory cytokines (F), upregulated in cluster 0/Pdpn⁺Il1rl1⁺ (G), involved in myofibroblast differentiation (H), involved in angiogenesis (I), and defining fibroblast lineages (J). (K) Immunohistochemistry staining of vimentin (VIM) and IL-1R1 in the WT and KO paw skin.

Figure 3

Pervasive recruitment and activation of immune cells in RDEB skin. (A) UMAP plots showing the immune cell types in wild-type (WT) and knockout (KO) mice. 0: ɑβ T cells (ɑβT), 1: γδ T cells (γδT), 2 and 3: neutrophils (Neu), 4 and 5: Langerhans cells (LC), 6, 7, and 8: macrophage/dendritic cells (Mφ/DC), 9: mast cells (MC), 10: basophils (Bas), 11: plasma cells (PC), and 12: B cells (BC). (B) Heat map highlighting ɑβ T-cell gene expression at a single-cell level. Cells were grouped by experimental condition (KO vs. WT) and ordered along the x-axis by expression of Pdcd1 and Tcf7. (C, D) Accumulation of CD8⁺PD-1⁺ T cells and CD68⁺PD-L1⁺ macrophages in immunohistochemically stained KO paw skin over a 3-week time course. Images were representative of three biological samples per group. Time points for staining are 0 (D0), 7 (D7), 11 (D11), and 21 (D21) days after birth. The staining in the 11-day-old wild-type (D11 WT) paw skin was representative of all WT time points. CD8/PD-1 and CD68/PD-L1 staining in C7^hypo mouse skin is included in Supplementary Figure S4. (E) Heat map on select genes, grouped together by function: immune suppressive (Suppressive), immune stimulatory (Stim.), proinflammatory (Proinf.), IL-1 processing (IL1 proc.), chemoattractants (Chemoatt.), and chemokine receptors (Chemorec.). Color reflects the log-fold difference of gene expression in KO compared to WT. (F, G) Heatmaps highlighting Langerhans cell and neutrophil gene expression at a single cell level. Cells were grouped together by WT and KO and activated (Act. LC/Act. Neu) and nonactivated (LC/Neu) Langerhans cells/neutrophils.

Figure 4

Dysfunctional epidermal barrier formation and keratinocyte-immune cell crosstalk in RDEB. (A) UMAP plots showing keratinocyte subtypes in wild-type (WT) and knockout (KO) mouse samples. (B) Significantly downregulated and (C) upregulated genes in KO keratinocytes shown by UMAP plots of their expression in WT and KO keratinocytes. Images were representative of three biological samples per group. (D) Representative KRT16 immunohistochemical staining in the epidermis of KO and WT mice. (E) UMAPs highlighting inflammatory genes upregulated in keratinocytes of KO mice.

Figure 5

RDEB mice exhibit distinct waves of inflammation after birth. (A, B) Quantitation of cytokine concentrations in paw skin lysates of RDEB and wild-type (WT) mice. Results show box and scatter plots of RDEB (n = 46–73) and WT (n = 36–48). ^*p-value < 0.05; ^**p-value < 0.01; ^***p-value < 0.001; ^****p-value < 0.0001. Data between age groups of RDEB samples were analyzed by ANOVA with Tukey’s correction, and data between RDEB and WT samples within each age group were analyzed by unpaired Student’s t-test. (A) Comparison of cytokine log concentrations in RDEB and WT paw skin. (B) Individual cytokine expression differential between RDEB and WT mouse paws grouped by age (d0 = 0 days old, d1–d5 = 1 to 5 days old … d21–d30 = 21 to 30 days old). Representative cytokine levels spread by age are presented in Figure S7. (C, D) Quantitation of IL-1α (C) and sST2 (D) concentrations in the plasma of RDEB and WT mice. Results show box and scatter plots of RDEB (n = 10–13) and WT (n = 11–13) grouped by age. (E, F) Immunohistochemistry analysis of NF-κB (phospho-IκBɑ (Ser32), p105/p50, and C-Rel) and JAK/STAT (pSTAT3) signaling in RDEB and WT paw skin sections. A magnified inset image in (F) is shown in Supplementary Figure S8C. Images were representative of three biological samples per group. Scale bar: 75 µm.

Figure 6

PDPN expression in wild-type (WT) and RDEB mice and in vitro inflammatory responses of WT and RDEB mouse-derived fibroblasts. (A–C) Immunohistochemistry (IHC) staining highlighting the colocalization of PDPN with vascular cells (lymphatic vessel endothelial hyaluronan receptor 1 (LYVE1)), macrophage/dendritic cells (CD68), and fibroblasts (vimentin (VIM)) in the skin of RDEB and WT mouse paws. Images were representative of three biological samples per group. White arrows in (A) identify PDPN⁺LYVE1⁻ cells, and white arrowheads in (B) identify PDPN⁺CD68⁺ cells in RDEB paw skin. In the WT paw skin, PDPN⁺ cells were mainly colocalized with LYVE⁺ vascular cells. (D) IHC analysis of PDPN, αSMA, and VIM staining in D21 RDEB mouse skin. (E, F) In vitro responses of RDEB and WT mouse fibroblasts (Fb) under basal conditions and following the treatment with TNF, IFN-γ, IL-1α, and TGF-β1, respectively. *p-value < 0.05; **p-value < 0.01; ***p-value < 0.001. (E) Median fluorescence intensity (MFI) of PDPN staining measured by flow cytometry and (F) ST2 in the fibroblast-conditioned medium (CM) quantitated by ELISA analysis. Data between treatments of fibroblasts were analyzed by ANOVA with Tukey’s correction, and data between RDEB fibroblasts and normal controls within each treatment were analyzed by paired Student’s t-test. (G) Immunocytochemistry staining of αSMA in WT and RDEB fibroblasts under basal condition and following treatment with TGF-β1, IL-1α, and IL-1α/IFN-γ.

Figure 7

In vitro inflammatory responses of RDEB patient-derived and normal control fibroblasts. RDEB patient-derived (EB Fb) and normal control (NC Fb) fibroblasts were cultured for 24 h with cytokines (TNF: 100 ng/mL, IFN-γ: 100 ng/mL, IL-1α: 4 ng/mL, IL-6: 2 ng/mL, and TGF-β1: 5 ng/mL) in basal media. (A) Flow cytometry analysis of PDPN expression measured by median fluorescence intensity (MFI). Results show box and scatter plots of RDEB patients (n = 6) and normal controls (n = 3) with two replicates each, grouped by cytokine treatment. (B) ST2 concentrations measured in conditioned media (CM) by ELISA. Results show box and scatter plots of RDEB patients (n = 6) and normal controls (n = 3) with one to two replicates each, grouped by cytokine treatment. ^*p-value < 0.05. Data between treatments of fibroblasts were analyzed by ANOVA with Tukey’s correction, and data between RDEB fibroblasts and normal controls within each treatment were analyzed by paired Student’s t-test. See Supplementary Figure S9 for overlaid flow cytometry plots of PDPN and membrane-bound ST2 (ST2L) expression of a representative RDEB patient-derived fibroblast at a basal level and following the treatment of cytokines. (C) Immunocytochemistry staining of PDPN and IL-1R1 in EB Fb and NC Fb under basal conditions and following treatment with IL-1α, IFN-γ, and TGF-β1, respectively. Scale bar: 150 µm.