Dendritic cells (DCs) are rare cells that are essential regulators of anti-tumour, anti-viral and vaccine responses by the immune system. DC populations are functionally diverse and present in most adult tissues, however human models of DC biology rely on differentiation from peripheral blood monocytes or cord blood progenitors. A critical gap for development of DC-based immunotherapies is understanding how faithfully these capture the repertoire of DC subsets or behaviors.
We built a computational platform to compare DC subtypes including freshly isolated and cultured subsets, by combining 342 samples from 15 expression profiling studies derived from 10 laboratories. The resulting combined transcriptional atlas revealed that the most common in vitro model, CD34+ HSC-derived DCs, are missing key immunoregulatory factors when compared to primary DCs, and further, that these can be partially rescued by DCs differentiated in vivo in humanized mouse models.
The atlas is implemented in the Stemformatics.org platform. Users can visually compare gene expression of DC samples categorized by tissue, sample source and disease or activation status. Also, users are allowed to project external single cell datasets for annotation of DC subsets, or other transcriptome platforms to benchmark new models of DC biology.