PD-L1 AND ICOSL DISCRIMINATE HUMAN SECRETORY AND HELPER DENDRITIC CELLS IN CANCER, ALLERGY AND AUTOIMMUNITY [CITE-seq]

In this study, we define a novel phenotypic and functional dichotomy of human dendritic cells (DC), with strong relevance to cancer, allergy, and autoimmunity. Using 16 different stimuli in vitro (130 observations), we describe two states of human activated DC. PD-L1highICOSLlow Secretory DC produced large amounts of inflammatory cytokines and chemokines but induced very low levels of T helper (Th) cytokines following DC-T co-culture; conversely PD-L1lowICOSLhigh Helper DC produced low levels of secreted factors but induced high levels and a broad range of Th cytokines. Secretory DC were aligned with mature migratory LAMP3+ DC in three different cancer types, atopic dermatitis, and lupus nephritis by single-cell transcriptomics. They were associated with T cell inflammation, overexpressed stimulatory and inhibitory checkpoints, acquired increased cell-cell communication, and were associated with good prognosis in head and neck squamous cell carcinoma, and with response to checkpoint blockade in Melanoma. This functional dichotomy of DC has broad implications in inflammation and immunotherapy. Fresh samples of HNSCC and breast tumor tissues of untreated patients were obtained from the pathology departments of the Institut Curie and the Pitié-Salpêtrière hospital. After tissue dissociation and single-cell suspension generation, cell sorting was done by FACS to enrich for immune cell populations. Single-cell sequencing of final samples was performed using Single Cell 3’ kit V1 (Breast cancer sample), v2 (HNSCC Pt1) and v3 (HNSCC Pt2) (10X Genomics), according to the manufacturer protocol (10X Genomics), including the CITE-seq libraries (HNSCC samples). Sample quality was checked with Bioanalyzer Agilent 2100 using a HighSensitivity DNA chip (Agilent Genomics). Samples were sequenced on an Illumina HiSeq (Breast cancer sample) and Novaseq (HNSCC samples) with a depth of sequencing of 50000 reads/cells (Breast cancer) and 100000 reads/cells (HNSCC samples).