Background Melanoma is one of the deadliest forms of skin cancer, but it is curable if detected early. Some melanomas closely mimic benign moles (melanocytic nevi), making diagnosis challenging and increasing the risk of diagnostic errors, morbidity, and mortality. Early melanoma originates in the epidermis, the outermost epithelial layer of the skin; however, the role of epidermal keratinocytes in melanoma initiation and progression is not fully understood, leaving their potential as diagnostic or therapeutic targets largely unexplored. To identify key gene signatures of the keratinocyte microenvironment in melanoma progression, we utilized single cell spatial transcriptomics, which enables in-situ gene expression analysis of individual cells while simultaneously accounting for their spatial context within the tumor microenvironment.

Methods We used spatial transcriptomics to visualize in situ gene expression of ~6000 genes at a single-cell-level within seven samples: three invasive melanomas, one melanoma in situ, one dysplastic nevus, one common nevus, and one basal cell carcinoma. Multiple fields of view (FOVs) consisting of the epidermal components of both lesional tissue and adjacent non-lesional skin were included from each sample. To visualize cell types and expression patterns across all samples simultaneously, we performed dimensionality reduction and cell clustering using the R package Seurat. Subsequently, the smiDE package was used to perform differential gene expression of keratinocytes, and clusterProfiler was used for Gene Ontology enrichment analysis.

Results Our analysis spanned 144 of FOVs, including a total of 147,541 cells across all samples. Cell clustering revealed multiple keratinocyte populations, with some populations more prevalent in malignant samples. In these keratinocytes, the top enriched pathways included keratinocyte differentiation, epidermal proliferation, viral immune response, and wound healing. Initial results discovered 33 upregulated genes within the epidermal keratinocyte microenvironment of melanomas but not nevi, including but not limited to damage associated molecular patterns (DAMPs) S100A7, S100A8, and S100A9, and wound-associated keratins KRT6B and KRT6C. Wound-associated keratins and S100A8/9 were also upregulated in the basal cell carcinoma microenvironment, suggesting that they are associated with both melanocytic and keratinocytic cutaneous malignancies. Additionally, upregulated genes unique to melanomas included DSC2, DYNLL1, LCN2, and RAB11A.

Conclusion Our findings identify a keratinocyte population in the tumor microenvironment of melanoma, which exhibits a wound-like proliferative state. Our results build a foundation for identifying biomarkers and therapeutic targets in the epithelial microenvironment of melanoma.