The similarities in morphology and efficacy of cultured epithelial sheets under SFF and SFD conditions
The scheme of the study was depicted in Fig. 1. Three patients were included in this study (Fig. 2a). Based on previous literature and our own experience with this treatment, factors that can affect the outcome included the age of the patient, the ratio of keratinocyte and melanocyte in the skin sheet, the site of the transplantation, the viability of the donor cells, the type of vitiligo, the stable period of vitiligo (no isomorphic reaction and depigmentation for more than 6 months), and previous treatment. In order to minimize the interference of the above factors, we enrolled young patients with stable vulgaris vitiligo who had failed to repigmentation after traditional drug treatment for more than 1 year, and have had a stable period for more than 1 year. For each enrolled patient, the skin obtained from the donor site was divided into two parts, one culturing the epithelial sheets under serum- and feeder-free conditions, and the other culturing the epithelial sheets under serum- and feeder-dependent conditions (Fig. 1). After the culture, the appearance and thickness of the SFF and SFD epithelial sheets were evaluated by HE staining (Fig. 2b). It was found that both methods cultured stratified cell skin grafts with epithelioid structure, and there was no significant difference in thickness between the two methods (Fig. 2c). The epithelial sheets obtained by the two methods were transplanted to the patient's skin at the same time. Both methods achieved excellent (full repigmentation) results at the 1-year follow-up after transplantation (Fig. 2d).
Specific cell types of keratinocytes and fibroblasts in the epithelial sheets compared with normal skin revealed by scRNA-seq
Healthy skin of 3 patients with stable vitiligo were obtained by surgical excision. Its derived epithelial sheets were cultured under SFD and SFF conditions, respectively. Cells dissociated from these 9 samples were subjected to scRNA-seq. After stringent quality controls (Supplementary Fig. S1a, 1b), transcriptomes of 79200 cells were obtained. A total of 16 distinct cell clusters were identified, according to unsupervised hierarchical clustering (Fig. 3a; Supplementary Fig. S1c, 1d). We formed cluster-specific marker genes using differential gene expression analysis to describe the identity of each cell cluster (Fig. 3b; Supplementary Fig. S1e-k). The unbiased cluster identifier was frequently a widely recognized specific cell marker, such as TYRP1, PMEL, and MLANA for melanocytes, KRT14, KRT10, and KRT5 for keratinocytes, COL1A1, COL1A2, and FN1 for fibroblasts. It is noteworthy that we redefined a cluster of fibroblasts using an additional marker, called ZNF90, on the basis of established makers for fibroblasts. Therefore, the 16 clusters included 7 clusters of keratinocytes (granular, spinous, LAMB3 + basal, KRT15 + basal, suprabasal-1, suprabasal-2, proliferation), 2 clusters of fibroblasts (ZNF90 + and DCN+), as well as melanocytes, endothelial cells, T cells, Langerhans cells, mast cells, and neurons (Fig. 3a). Immunofluorescence staining has verified the presence of these clusters in epithelial sheets and normal skin. (Fig. 3c; Supplementary Fig. S2)
Changes in cell composition could shed light on how melanocytes maintain their own stability and carry out melanin synthesis. The cell proportions in each cluster were analyzed, and that exhibited heterogeneity within and among samples (Fig. 3d). Compared with healthy skin, the proportions of melanocytes, granular keratinocytes, LAMB3 + basal keratinocytes, ZNF90 + fibroblasts increased in epithelial sheets cultured under both SFD and SFF conditions (Fig. 3d, e, f, g). When comparing SFF epithelial sheets to SFD epithelial sheets, the proportions of these 4 cell clusters showed an increase in ZNF90 + fibroblasts and LAMB3 + basal keratinocytes, with a decrease in granular keratinocytes and melanocytes (Fig. 3f, g).
Putative Lamb3 + basal Keratinocyte And Znf90 + fibroblast Differentiation Trajectories
While UMAP analysis revealed keratinocyte and fibroblast heterogeneity, we also questioned whether they adhere to particular differentiation trajectories. By ordering keratinocytes in pseudotime, a triangular trajectory was arranged with 2 probable differentiation paths from the vertex to the bottom right and the left corners, respectively (Fig. 4a). Differential gene expression analysis showed that LAMB3 representing basal keratinocytes and ITGB1 related to epithelial stem cell characteristics were highly expressed at the early stage, while LCE3D/E representing granular keratinocytes and S100A8/A9 mediating keratinocyte growth inhibition were upregulated at a later stage (Fig. 4b). We investigated the modules of sets of co-regulated genes among all 7 keratinocytes clusters and observed module 4 and 15 were relatively upregulated in LAMB3 + keratinocytes (Fig. 4c). When these 2 modules were projected to the pseudotime map, they could be found out at the right corner at the early stage (Fig. 4d). Next, we conducted pathway enrichment analysis of module 4 and 15 using Reactome (Fig. 4e) and GO database (Fig. 4f), and found that laminin interactions, ECM proteoglycans, integrin cell surface interactions, kinase activity, transmembrane signaling receptor activity, and so on were enriched in module 4 and 15, which suggested that LAMB3 + basal keratinocytes participated in extracellular matrix homeostasis and exhibited vigorous cellular metabolism.
Ordering of fibroblasts in pseudotime arranged them into a hybrid of an initial cyclical trajectory and a subsequent linear trajectory (Fig. 4g). DCN and COL6A1-expressing cells were mainly dispersed towards the beginning of the trajectory, whereas ZNF90 and HIST3H2A expression levels increased near the end of the trajectory (Fig. 4h). In other words, cell trajectory analysis represented a differentiation pattern from DCN + to ZNF90 + fibroblasts. Module 10, 7, 16, 8, and 9 were upregulated in ZNF90 + fibroblasts (Fig. 4i), which were consistent with dynamic changes in feature plots (Fig. 4j). Moreover, pathway enrichment analysis suggested that anaphase transition cell signaling in mitosis, positive regulation of apoptotic process, keratinization cornified envelope formation, epidermal and epithelial cell differentiation, vasculature development, and so on were enriched in ZNF90 + fibroblasts (Fig. 4k, l). These results demonstrated that ZNF90 + fibroblasts were more mature differentiated than DCN + fibroblasts, more importantly, were associated with more developed cellular activity and stronger interactions with epidermal cells and vascular endothelial cells.
Specific Cellular Functions Of Lamb3 + basal Keratinocytes And Znf90 + fibroblasts
To elucidate the cellular functions of LAMB3 + basal keratinocytes and ZNF90 + fibroblasts in melanocyte development, we identified DEGs using GSVA via REACTOME and GO database in 2 groups, LAMB3 + basal keratinocytes and other keratinocytes, and ZNF90 + fibroblasts and other fibroblasts.
DEGs mainly correlated with the activation of extracellular matrix (Fig. 5a, b; Supplementary File. S1), including “extracellular matrix organization”, “fibronectin matrix formation”, “collagen formation”, “collagen fibril organization” and so on, when comparing LAMB3 + basal keratinocytes to other keratinocytes. Moreover, upregulated genes were associated with melanocyte proliferation (Fig. 5b). WNT signaling, activating the differentiation of melanocyte precursors15, was enriched, including “beta-catenin independent WNT signaling”, “positive regulation of WNT signaling pathway planar cell polarity pathway”, “WNT signaling pathway calcium modulating pathway” and so on (Fig. 5a, b). FGF, an essential growth factor for melanocytes16, was upregulated via “signaling by FGFR2”, “regulation of cell chemotaxis to fibroblast growth factor”, and so on (Fig. 5a, b).
As to fibroblasts, GSVA analyses suggested that pathways involved in collagen, including “assembly of collagen fibrils and other multimeric structures”, “Intrinsic pathway of fibrin clot formation”, “collagen chain trimerization”, “collagen formation”, and so on, were enriched in ZNF90 + fibroblasts compared to DCN + fibroblasts (Fig. 5c, d; Supplementary File. S1). It may help fibroblasts to adhere to surrounding cells, such as “positive regulation of cell adhesion mediated by integrin” and “cell junction maintenan0thesis”, “keratinization”, “hair follicle maturation”, “epithelial mesenchymal cell signaling”, and so on (Fig. 5c, d).
Potential ligand-receptor interactions between LAMB3 + basal keratinocyte and ZNF90 + fibroblast clusters and melanocyte cluster in epithelial sheets
To characterize the cell–cell communication paradigm between LAMB3 + basal keratinocyte, and ZNF90 + fibroblast clusters and melanocyte clusters in epithelial sheets, we performed an analysis using CellPhoneDB, which contains a repository of ligand–receptor interactions and offers a framework for deducing cell-cell communication networks between two cell types from single-cell transcriptomics dataset. In the heatmap showing the numbers of inter-populations communications with each other in SFF sheets and SFD sheets (Fig. 6a), we observed more abundant interactions among cells in SFF epithelial sheets, in particular, of which LAMB3 + basal keratinocytes displayed the richest interactions with other cell types, including melanocytes.
Next, we established ligand–receptor pairs between LAMB3 + basal keratinocytes and melanocytes in SFF versus SFD sheets (Supplementary File. S2). Significantly altered signals from LAMB3 + basal keratinocytes associated with melanocyte homeostasis including JAG, NOTCH and EGF signaling17 were observed in both SFF and SFD sheets, whereas more interactions were involved in SFF sheets (Fig. 6b). For example, EGFR and its receptors including MIF, GRN, and COPA were upregulated in the interactions between LAMB3 + basal and melanocytes from both sheets, while JAG1-NOTCH2/3/4 and NOTCH1-NOV ligand–receptor pairs were significantly altered only in SFF sheets. Additionally, we investigated variations in ligand signals transmitted by melanocytes. In both sheets, CD44 that facilitates cell adhesion by hyaluronic acid18, and CD46 that protects melanocytes against complement-mediated damage19 was highly expressed in melanocytes, thereby activating LAMB3 + keratinocytes via several patterns such as CD44-FGFR2, CD44-HBEGF, and CD46-JAG1 pairs (Fig. 6b). Notably, KIT-KITLG pairs that play an important role in melanocyte migration and survival20, and NRP2-VEGFA/SEMA3C pairs that influence cell migration and angiogenesis21 were significantly altered only in SFF sheets. Representative ligand–receptor circle figures also indicated that JAG, NOTCH, and CD44 signaling interactions increased between 2 cells in SFF sheets compared to SFD sheets (Fig. 6c).
Moreover, ligand–receptor pairs between ZNF90 + fibroblasts and melanocytes were investigated (Supplementary File. S2). JAG1-NOTCH, NRP2-VEGFA/SEMA3C, and MIF-TNFRSF14 pairs were significantly altered in SFF sheets where melanocytes express receptors and respond to ligand signals from ZNF90 + fibroblasts (Fig. 6b, left), while only MIF-TNFRSF14 pairs that might induce hypopigmentation22 to maintain melanocytes homeostasis was dramatically changed in SFD sheets (Fig. 6b, right). Moreover, ligand signals that were significantly overexpressed in melanocytes including NRP2, KIT, CD44, and CD46 were accepted by ZNF90 + fibroblasts from SFF sheets (Fig. 6b, left), while only KIT signaling was enhanced in melanocytes from SFD sheets (Fig. 6b, right). These results were also supported in the circle figures (Fig. 6c). Taken together, more interactions between LAMB3 + basal keratinocyte and ZNF90 + fibroblast clusters and melanocyte cluster were observed in SFF sheets compared to SFD sheets.