Fig. 2: Identification and transcriptional characterization of a malignant epithelial cluster.
From: Single cell deciphering of progression trajectories of the tumor ecosystem in head and neck cancer
UMAP projection of 7054 epithelial cells from 26 samples colored by groups (A), diploid/aneuploid status (B), and clusters (C). D The Kaplan-Meier curves showed patients with higher infiltration of cluster1 epithelial cells are associated with worse overall survival (OS) in TCGA-HNSCC cohort (n = 494, 247 samples for each group). Dot plot of top 5 hallmarks (E) and 6 metabolic pathways (F) for differentially expressed genes (DEGs) in each epithelial cluster. G Potential trajectory of epithelial cells inferred by Monocle2. The trajectory was divided into three states indicated as S1, S2, and S3. H Heatmap shows normalized activity of top 5 transcription factors (TF) regulons predicted by the SCENIC algorithm (left) and the relative expression (z-score) of top 5 TF genes (right) in epithelial cell clusters. I Heatmap shows the dynamic changes in TF and regulons expression along the pseudotime. The expression of TFDP1 is shown in the right panel. J Representative images of multiplex immunohistochemistry (mIHC) staining of TFDP1+ epithelial cells in HNSCC tumor and nonmalignant samples. Scale bar, 100 μm and 50 μm. The quantitative results are shown on the right. Upper line: TFDP1+ epithelial cell ratio in different stages (n = 69). Lower line: The Kaplan-Meier OS curves of validation cohort patients stratified by TFDP1+ expression level (n = 52, 26 samples for each group). K The Kaplan-Meier curves of samples with high (n = 372) and low (n = 122) TFDP1 expression level in TCGA-HNSCC cohort (n = 494). L Images of Transwell assays for migration and invasion in different cell lines with TFDP1 overexpression or knockdown. Scale bar = 100 μm. The quantitative analysis is shown on the right. n = 3 biologically independent experiments. Data represent mean ± SD. P values were calculated by two-side Student’s t-test in J and L, by one-way ANOVA test in E and F, and by two-sided log-rank test in D, J and K. Source data are provided as a Source Data Fig. 2A–L.
