Supplementary MaterialsSupplementary Information 41467_2020_14777_MOESM1_ESM. environment, and emphasizes the power Dihydroberberine of using defined genetic model systems. gain-of-function (-catGOF) and loss-of-function (Bmpr1aLOF) mutations. We showed that Icam2 these mice developed very specific salivary gland SCCs within Dihydroberberine 100 days after birth which contained highly self-renewing Wnt-dependent CD24+CD29+ CSCs which, upon isolation and injection into NOD/SCID mice, produced fast-growing tumors16,17. These CSCs showed high activity of the stem cell-associated SSEA1 marker as well as nuclear -catenin and Wnt-specific target genes such as which were not found in additional subpopulations within the tumor16. To gain a more fundamental understanding of tumorigenesis, we here used single-cell transcriptomics together with our Wnt-dependent double-mutant salivary gland SCC mouse model16, 17 to systematically study CSCs inside a controlled establishing in vivo. Our setup (Fig.?1a) enabled us to build a high-resolution salivary gland cell atlas, to dissect tumor heterogeneity in Dihydroberberine a whole tissue environment and to identify CSC-like cells de novo directly from stable tumor samples. We display that tumor-specific epithelial cells consist of luminal- and basal-like cells as well as a small, but unique CSC-like human population. Further molecular characterization together with pathway and lineage analyses allowed us to Dihydroberberine infer and reconstruct a powerful trajectory of the tumor progression. We found that upon activation of gain- and loss-of-function mutations in basal cells, tumorigenesis is initiated by manifestation of an EMT signature and proceeds through heterogeneous populations of CSC-like cells driven by differential Wnt signaling, before differentiating into luminal-like cells. Our work reveals several genes and manifestation patterns that may be fundamental in the rules of tumorigenesis, and provides a novel and unbiased approach to study CSCs from a developmental perspective. Open in a separate windowpane Fig. 1 A comprehensive salivary gland cell atlas.a Experimental strategy to systematically dissect the cellular diversity in stable tumors. Submandibular salivary glands were separately dissected, dissociated and single, live cells isolated by FACS. Cells were immediately fixed in methanol and further processed to profile their transcriptomes by a high-throughput droplet-based single-cell approach. Each biological replicate corresponds to the cells of one submandibular gland from a control or tumor-bearing, woman or male mouse at a defined stage as indicated. b tSNE representation of single-cell data from control salivary glands demonstrates cells cluster into 14 organizations based on their transcriptome similarity. Clusters are coloured and shaded according to the manifestation of both novel and known marker genes for epithelial and non-epithelial cell types. Turquoisebluegreen: luminalacinarductal,?pink: basal,?purple: myoepithelial,?shades of brown, yellow and orange: non-epithelial (immune, endothelial, fibroblasts, T/NK). c Anatomical sketch of the female submandibular gland based on single-cell transcriptome data, available literature (observe text for referrals) and validations in cells sections by immunofluorescence. Results Single-cell RNA sequencing of salivary gland tumors To identify and characterize the cellular heterogeneity that is specific to the solid tumor context, we first founded controlled ways to dissociate tumor-bearing (double-mutant: -catGOF; Bmpr1aLOF) and control salivary glands into high-quality single-cell suspensions (Fig.?1a). After dissociation, deceased cells and enucleated cellular debris were excluded and live intact cells acquired by fluorescence-activated cell sorting (FACS) (Supplementary Fig.?1a). Cells were directly sorted into methanol for fixation18, and further processed to profile their transcriptomes by a high-throughput droplet-based approach (Drop-Seq)19. In total, 26 single-cell RNA libraries were generated Dihydroberberine from 12 control and 14 double-mutant (tumor-bearing) salivary glands of either woman or male mice from an early and a late tumor stage at postnatal days 40 (P40) and 90 (P90), respectively (Fig.?1a). To validate our experimental approach, we compared all single-cell samples, computationally pooled by disease status (control or double-mutant), to bulk mRNA-seq data that were generated from equivalent, freshly dissected but unprocessed, salivary glands (Supplementary Fig?1b). Although gene manifestation levels correlated better within experimental methods and samples grouped by genotype, correlations between all samples were generally high (and were indicated in cell populations with strongly sex-dependent representation (Supplementary Fig.?3) and identified several other marker genes with related patterns (Supplementary Fig.?4a). Interestingly, we also mentioned that Dcpp1+ cells were more abundant.