Atopic Dermatitis Expands Stem-Like Keratinocytes with Divergent Immune, Stress and Maturation Programs

Supplemental Data of the manuscript in preparation with the title "Atopic Dermatitis Expands Stem-Like Keratinocytes with Divergent Immune, Stress and Maturation Programs". Abstract: The epidermis maintains homeostasis through a tightly regulated balance of self-renewal, proliferation, and differentiation, orchestrated by distinct keratinocyte subpopulations with specialized functions. Understanding how these processes are disrupted in atopic dermatitis is essential for elucidating disease mechanisms. Using a supervised machine learning model trained on single-cell transcriptomes from previously defined basal subpopulations, we annotated keratinocyte subtypes in an independent single-cell RNA-sequencing dataset of healthy adult and AD epidermis. We identified basal III as the interfollicular stem cell-containing compartment, preferentially localized at the base of rete ridges and atop dermal papillae. Basal I and II were classified as transit-amplifying cells, while basal IV corresponded to early differentiation stages. In AD, basal III cells split into two transcriptionally distinct subpopulations and partially lost their characteristic marker expression. One subpopulation retained stem-like gene expression but showed reduced expression of basal membrane anchoring genes, while the other maintained anchoring features but prematurely activated differentiation programs. Basal III cells occupied a more advanced position in pseudotime and upregulated the production of pro-inflammatory cytokines, along with markers of cellular stress, apoptosis, premature proliferation, and stress-associated differentiation. While this population was transcriptomically expanded in AD, transient amplifying basal I/II were not. These findings suggest a potential paradigm shift in AD pathogenesis, indicating basal III cells as central contributor to disrupted epidermal homeostasis, aberrant differentiation and potentially pro-inflammatory signaling. We used public single-cell RNA-sequencing data from GSE153760 for analysis of healthy and AD epidermis and single-cell RNA-sequencing data from GSE147482 for the annotation of basal cell types. All code and the corresponding environments are available on GitHub https://github.com/FPardow/KCinAD. An interactive interface of the key datasets used in this publication can be found here: https://kcinad.shinyapps.io/shinyapp/. Pictures of immunohistochemistry stainings are available from the corresponding author upon request.