Pan-Cancer Analyses Refine the Single-Cell Portrait of Tumor-Infiltrating Dendritic Cells

Abstract Dendritic cells (DC) are pivotal orchestrators of antitumor immunity. DC-based antitumor treatments are being actively developed, but effective clinical responses have not yet been achieved. Further exploration of DC heterogeneity in the tumor microenvironment and across cancer types could provide insights for developing DC-based immunotherapies. In this study, we integrated single-cell RNA sequencing data of DCs from more than 2,500 samples across 33 cancer types and established a comprehensive blueprint of human DCs. Several rare subsets of DCs infiltrated the tumors, including AXL+SIGLEC6+ DCs and Langerhans cell–like DCs, and displayed functional potentials marked with distinct transcriptomic characteristics. Computational analyses demonstrated that the Langerhans cell–like subset could be an additional cellular origin of tumor-enriched LAMP3+ DCs and that distinct cellular origins are associated with the pleiotropic functional potentials of LAMP3+ DCs. Furthermore, this DC atlas enabled the development of a machine learning model to guide DC annotation for subsequent single-cell analysis and prioritization of a valuable target for enhancing antitumor DC vaccination. This integrative resource provides a panoramic view to unravel the complexity of tumor-infiltrating DCs and offers valuable insights for developing therapies targeting DCs. Significance: The comprehensive dissection of tumor-infiltrating dendritic cells redefined cell subsets with different regulations, tissue preferences, and functional potentials and provided an atlas as a rich resource with promising applications. This article is part of a special series: Driving Cancer Discoveries with Computational Research, Data Science, and Machine Learning/AI.