Systems immunology integrates the complex endotypes of recessive dystrophic epidermolysis bullosa

Abstract

Endotypes are characterized by the immunological, inflammatory, metabolic, and remodelling pathways that explain the mechanisms underlying the clinical presentation (phenotype) of a disease. Recessive dystrophic epidermolysis bullosa (RDEB) is a severe blistering disease caused by COL7A1 pathogenic variants. Although underscored by animal studies, the endotypes of human RDEB are poorly understood. To fill this gap, we apply systems immunology approaches using single-cell high-dimensional techniques to capture the signature of peripheral immune cells and the diversity of metabolic profiles in RDEB adults, sampled outside of any opportunistic infection and active cancer. Our study, demonstrates the particular inflammation and immunity characteristics of RDEB adults, with activated / effector T and dysfunctional natural killer cell signatures, concomitant with an overall pro-inflammatory lipid signature. Artificial intelligence prediction models and principal component analysis stress that RDEB is not solely confined to cutaneous issues but has complex systemic endotypes marked by immune dysregulation and hyperinflammation. By characterising the phenotype-endotype association in RDEB adults, our study lays the groundwork for translational interventions that could by lessening inflammation, alleviate the everlasting suffering of RDEB patients, while awaiting curative genetic therapies.

Fig. 1. Immune landscape of RDEB adults.

a UMAP visualizing PhenoGraph-obtained clusters of major immune cell populations from a total of 21 whole blood samples (105.10³ cells/sample). UMAP comparing the distribution of major immune cell populations in 9 healthy controls and 12 RDEB adults. Colours vary according to cell abundance density. b Violin plots comparing the frequency (%) and the absolute counts (cells/μl) of indicated immune cell populations of 9 healthy controls (black triangles) and 12 RDEB adults (red circles) with median values presented by solid black line. Source data are provided as a Source Data file. Statistical analyses were performed with two-sided unpaired t-test. Asterisks represent the significant differences (*p < 0.05, **p < 0.01, ***p < 0.001). c Variability in single-cell distributions across Healthy and RDEB adult Formalin-fixed paraffin-embedded skin slices. Representative (three independent experiments) images of antibody staining and corresponding single-cell segmented images across histological subgroups. Displayed labelling patterns for each reconstructed image. Merge: panKeratin(green)/SMA (grey)/immune cell subsets (mixed colours). Enlarged images depicting the immune cell subsets present within the dotted square. Immune cell subsets depicted include CD4 T cells: CD4 (green), CD8 T cells: CD8 (magenta), Monocytes/Macrophages: CD68 (cyan), B cells: CD20 (bleu), Dendritic cells: DC-Lamp (yellow), neutrophils: MPO (red). d Opt-SNE visualizing PhenoGraph-obtained clusters of CD45⁺ immune cells from healthy (HC, n = 2) and RDEB adult (n = 2) skin biopsies. Each dot represents a single cell.

Fig. 2. Unsupervised PBMC profiling of RDEB adults.

a UMAP visualizing PhenoGraph-obtained clusters of major innate and adaptive immune cell populations from a total of 21 PBMC samples (25.10³ cells/sample). b UMAP comparing the distribution of adaptive immune cell populations in 9 healthy controls and 12 RDEB adults. Colours vary according to cell abundance density. The frequency (%) of each cell population is presented as bar graphs on the left side of the corresponding UMAP. c Violin plots comparing the absolute counts (cells/μl) of indicated immune cell populations in 9 healthy controls (black triangles) and 12 RDEB adults (red circles) with median values presented by solid black line. Source data are provided as a Source Data file. Statistical analyses were performed with two-sided unpaired t-test. Asterisks represent the significant differences (*p < 0.05, **p < 0.01, ***p < 0.001). d PCA representing mass cytometry data from 9 healthy controls (black circles) compared to that from 12 RDEB adults (red circles), PC1: 14.8%, PC2: 21.0%. Clustering significance was determined using Permutational multivariate analysis of variance (PERMANOVA).

Fig. 3. RDEB T cells nurse distinct signature and activity.

a, b opt-SNE from healthy controls (n = 9) and RDEB adults (n = 12) clustered and coloured by main CD4⁺ and CD8⁺ subsets with their respective abundances (%) by PhenoGraph clustering. Clustering performed with 8.10³ CD4⁺ cells/sample and 4.10³ CD8⁺ cells/sample. c, d Violin plots comparing the absolute counts (cells/μl) of 9 healthy controls (black triangles) and 12 RDEB adults (red circles) CD4⁺ and CD8⁺ T cell subsets with median values presented by solid black line. Source data are provided as a Source Data file. e, f CD2/CD3/CD28-PBMC-induced proliferation gated on CD4⁺ and CD8⁺ cells. Representative histograms of proliferation measured as loss (red line) of initial CFSE labelling (black line) are presented. Replication and Proliferation index (RI, PI respectively) are presented as scatter plots. Data are presented as mean values ± SEM from 6 healthy controls and 9 RDEB adults. Healthy controls (black triangles) and RDEB adults (red circles). Statistical analysis is performed with two-sided unpaired t-test. Asterisks represent the significant differences between healthy controls and RDEB adults (*p < 0.05, **p < 0.01). Source data are provided as a Source Data file.

Fig. 4. NK cells signature and activity in RDEB adults.

a opt-SNE from healthy controls (n = 9) and RDEB adults (n = 12) clustered and coloured by NK cell subsets with their respective abundance (%) determined by FlowSOM-based clustering. b Violin plots comparing the absolute counts (cells/μl) of NK cell subsets. c Degranulation of NK cells assessed by the expression of CD107a marker (left panel) and specific lysis (%) of K562 cells (right panel). Results are presented in scattered plots as mean values ± SEM from 6 healthy controls and 9 RDEB adults. d PMA/Ionomycin-PBMC-induced IFNγ and TNF production by NK cells presented as % positive cells with their respective geometric mean fluorescence. Results are presented in scattered plots as mean values ± SEM from 7 healthy controls and 7 RDEB adults. e The expression of activating and inhibitory NK cells receptors, and the exhaustion- and senescence-related markers. Results are presented in scattered plots as mean values ± SEM from 7 healthy controls and 7 RDEB adults (NKG2D, KIR2D, NKG2C, NKG2A), mean values ± SEM from 6 healthy controls and 6 RDEB adults (PD-L1 and LAG-3). Healthy controls (black triangles) and RDEB adults (red circles). Statistical analyses were performed with two-sided unpaired t-test. Asterisks represent the significant differences between healthy controls and RDEB adults (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Source data are provided as a Source Data file.

Fig. 5. Single-cell metabolic profiling of RDEB T and NK cells.

a Schematic presentation of single-cell energetic metabolism by profiling translation inhibition (SCENITH). Visualization of protein synthesis (PS) after puromycin incorporation by flow cytometry. Histograms display the level of PS in control (Co) and in the presence of specific metabolic inhibitors (DG, O or DGO). b–d Single-cell metabolic profile presented as scatter plots displaying the translation level, Fatty acid Amino acid oxidation (FAAO), glycolytic capacity, glucose dependence and mitochondrial dependence in CD4⁺, CD8⁺ cells and NK cells. Healthy controls (n = 6, black triangles) and RDEB adults (n = 6, red circles). All results are presented in scattered plots as mean values from 6 independent donors ± SEM. Statistical analyses were performed with two-sided unpaired t-test. Asterisks represent significant differences between healthy controls and RDEB adults (*p < 0.05, **p < 0.01). Source data are provided as a Source Data file.

Fig. 6. RDEB adults nurse a particular eicosanoids signature.

a Schematic representation of lipid mediators downstream polyunsaturated fatty acid (PUFA). b Heatmap representing mean values of serum eicosanoid lipid mediators’ relative abundance from healthy controls (n = 8) and RDEB adults (n = 12). Colours vary according to lipid abundance density. c–e Relative abundance of altered eicosanoids derived from acid arachidonic (AA) downstream cyclooxygenase (COX), lipoxygenase (LOX/FLAP), and cytochrome P450 (CYP450) pathways, linoleic acid (LA), and docosahexaenoic acid (DHA). Lipid mediators are presented in scattered plots as mean values of relative abundance/ml of sera ± SEM from 8 healthy controls and 12 RDEB adults. Healthy controls (black triangles) and RDEB adults (red circles). Statistical analysis is performed with two-sided unpaired t-test. Asterisks represent the significant differences between healthy controls and RDEB adults (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Source data are provided as a Source Data file.

Fig. 7. Mapping lipid rewiring in RDEB.

a Ingenuity pathway analysis (IPA) predicted canonical pathways upregulated and potentially involved in RDEB pathophysiology. The Y-axis shows the -log (p-value). The orange line indicates a threshold at -log (p-value) = 1.3, which represent p-value = 0.05. Statistical analysis was performed with two-sided fisher exact test. b Cytokines level (pg/ml) in serum samples presented in scattered plots as mean values of relative abundance/ml of sera ± SEM from 9 healthy controls and 12 RDEB adults. Statistical analyses were performed with two-sided unpaired t-test. Asterisks represent the significant differences (*p < 0.05, **p < 0.01, ****p < 0.0001). Source data are provided as a Source Data file. c IPA predicted diseases and bio-functions possibly related to RDEB. The x-axis shows the -log(p-value). The threshold at -log (p-value) = 1.3 is indicated. Statistical analysis was performed with two-sided fisher exact test.

Fig. 8. RDEB beyond skin and mucosa.

a Receiver operating characteristic (ROC) curves prediction of IscorEB, % wound area, SCC incidence, TXB2 and PGE2 level in RDEB patients. Prediction is based on Inflammation Immunity Score (IIS) data. The Y-axis shows the true prediction, while the X-axis shows false prediction. Curves are mean ± SD of curves coordinates obtained following 50 prediction iterations with independent training/validation cohort sampling. Mean area under the curve ± SD are indicated on graph. b PCA of RDEB adults IIS. Patients are distributed in 2 significant clusters. Clustering significance was determined using Permutational multivariate analysis of variance (PERMANOVA), and p-value is indicated on the graph. c Cytokines levels (pg/ml) in RDEB adults sera based on PCA clustering (upper panel). Relative abundance of eicosanoids (lower panel), presented as relative abundance/ml of sera. Results are presented in scattered plots as mean values ± SEM from 8 RDEB adults (cluster 1) and 4 RDEB adults (cluster 2). Statistical analyses were performed with two-sided unpaired t-test. Asterisks represent the significant differences (**p < 0.01, ***p < 0.001, ****p < 0.0001). Source data are provided as a Source Data file.