The single-cell genomic atlas of cervical cancer.

A. The schematic design of sample collection, single-cell RNA sequencing, data processing and clinical validating through 10X genomics platform. The sketch of scRNA-sq machine, as well as the plot of dimension-reduction visualization, was originally cited from the official website of 10X genonics: https://www.10xgenomics.com. B. UMAP plots of all single cells by original clustering and cell type re-clustering, according to the published lineage-specific marker genes, respectively. C. Violin plots demonstrating the expression of marker genes that correspond to each of the major cell types. D. UMAP plots demonstrating the most typical marker gene specific to each type of cell cluster. E. UMAP plot and histogram plot showing the occupation ratios of the major cell types in each individual sample, with HPV status and histological type summarized. F. UMAP (up) and histogram (down) plots to show the differences of distribution and proportion of each cell type between different histological types (ADC vs. SCC).

The scRNA-seq data reveals the malignant features of tumor epithelial cells.

A. UMAP plots showing the distribution of epithelial cells in 15 samples and the sub-clustering into 12 clusters according to ectopic gene expressions. Each cluster of epithelial cells was named using the most highly enriched gene. B. Heatmap plot showing the annotation of each epithelial sub-cluster with top 5 DEGs. C. UMAP plots demonstrating the most specific gene as the marker for each sub-cluster. D. UMAP (left) and histogram (right) plots to compare the differences of distribution and proportion of each epithelial cell sub-cluster between different histological types (ADC vs. SCC). E. UMAP plot of CytoTRACE showing the thermal imaging projection of predicted developmental order (left) and histogram plot showing the ranking of CytoTRACE scores (right). F. UMAP plot of malignancy analysis demonstrating the cell malignancy features (up) and ranking of scores (down). G. GOBP analyses showing the top 15 enriched signaling pathways of Epi_10_CYSTM1 that are more active in ADC than SCC (up), particularly pathways related to cell-to-cell interactions (down). The histogram data are transformed from -Log10(P-value) for visualization. Abbreviations: Ig: immunoglubolin; MHC: major histocompatibility complex.

SLC26A3 is identified as a potential prognostic and diagnostic indicator for lymph node metastasis of CC patients.

A. Heatmap showing the top 50 DEGs of Epi_10_CYSTM1 cluster in comparison to all the other sub-clusters (Red: Epi_10_CYSTM1 cluster; Green: other epithelial cell sub-clusters). B. Violin plots showing the expression difference between two groups (top), with UMAP plots (bottom) demonstrating the specificity of each candidate gene, filtering SLC26A3 as the most identical marker for this sub-cluster. C. Kaplan-Meier curve showing the overall survival rates of CC patients stratified by the top 50 genes-scaled signature of Epi_CYSTM1. D. IHC staining showing the protein expression of SLC26A3 in surgically-resected ADC samples which are classified as early stages (FIGO stage I-IIA) and late stages (FIGO stage IIIC1∼2p). Images from 6 individual cases were shown as representatives for each group. E. IHC staining showing the protein expression of SLC26A3 in biopsy ADC samples which are classified as early stages (FIGO stage I-IIA) and late stages (FIGO stage IIIC1∼2p). Images from 4 individual cases were shown as representatives for each group. The method of H-score is used and the scoring system is as follows: negative (0), weak (1), intermediate (2), and strong (3). Expression is quantified by the H-score method.

The association between post-surgical upstaging and clinical characteristics in patient cohort 1 (from Xiangya hospital)

The association between post-surgical upstaging and clinical characteristics in patient cohort 2 (from Hunan cancer hospital)

The association between clinical characteristics and SLC26A3 protein expression via IHC tested on post-surgical samples

The association between clinical characteristics and SLC26A3 protein expression via IHC tested on biopsy small specimens

Cellular and molecular heterogeneity of T cells in ADC.

A. UMAP plots showing the sample distribution (left), original cell clustering of T cells (middle, 8 clusters in total) and re-clustering of T cell subtypes (right, 4 major sub-clusters: exhausted T cell, cytotoxic T cell, regulatory T cell and activated T cell) according to acknowledged marker genes. B. Violin plot showing the expression of specific marker genes that annotate each sub-type of T cells. C. UMAP plots showing the widely recognized classification markers that denote each type of T cell sub-cluster. D. Dot heatmap plot of row-scaled expression of overexpressed genes for each sub-cluster of T cells. E. Dot heatmap plots that demonstrate the level of marker genes representing the signaling pathways of immune checkpoint activation and inhibition for grouped sub-clusters. F. Differences of distribution and proportion of each T cell sub-cluster between different histological types (ADC vs. SCC). G. GOBP analysis of Treg-enriched immunity-related pathways that are more active in ADC than SCC. The histogram data are transformed from -Log10 (P-value). Abbreviation: Treg: regulatory T cell.

The heterogeneity of tumor associated neutrophils in ADC.

A. UMAP plots showing the distribution of neutrophils in 15 samples (left), original clustering (middle) the re-clustering (right) into 3 sub-clusters according to gene markers of tumor-associated neutrophils (TANs). B. Dot heatmap showing the top 5 DEGs in each sub-cluster of TANs. C. UMAP plots for annotation of each sub-type of TANs with published marker genes presented: pro-tumor TANs, anti-tumor TANs and TANs with isg. D. UMAP and histogram plots to compare the differences of distribution and proportion of each sub-types of TANs between different histological types. E. GOBP analyses of signaling pathways that are more active in ADC than SCC, in terms of the pro-tumor TANs cluster. The histogram data are transformed from -Log10 (P-value). F. Kaplan-Meier curve showing the overall survival rate of CC patients stratified by the top 90 genes-scaled signature of pro-TANs.

Phenotype diversity of plasma/B cells in ADC.

A. UMAP plots showing the distribution of plasma/B cells in 15 samples (left) and the re-clustering into 6 clusters (right) according to ectopic expressions of genes. B. Plot heatmap showing the annotation of each sub-cluster with top 5 DEGs. C. UMAP plots demonstrating the most specific genes as the marker for each sub-cluster. D. UMAP and histogram plots to compare the differences of distribution and proportion of each plasma/B cell sub-cluster between different histological types. E. GOBP analysis of immunity-related signaling pathways that are more active in ADC, in terms of Plasma/B_01_IGHA2 sub-cluster. The histogram data are transformed from -Log10 (P-value). F. Kaplan-Meier curve showing the overall survival rate of CC patients stratified by the top 50 genes-scaled signature of Plasma/B_01_IGHA2.

The cellular interaction modules of sub-clusters of T cells, neutrophils and tumor epithelial cells.

A. Circle plots showing the interacting networks between epithelial cell sub-clusters and T cell sub-clusters via the pathway of ALCAM, by comparing ADC and SCC. B. Circle plots simplified from A to show the interactions among target cell clusters via ALCAM pathway. C. Circle plots showing the interacting networks between epithelial cell sub-clusters and T cell sub-clusters via the pathway of MHC-II, by comparing ADC and SCC. D. Circle plots simplified from C to show the interactions among target cell clusters via MHC-II pathway. From A to D, The direction of each arrow shows the regulation from outputting cells to incoming cells. The width of the each line shows the predicted weight and strength of regulation. E. Bubble plot showing the probability of ligand-to-receptor combination of each pathway between two different target sub-clusters of cells, by comparing ADC with SCC. F. Circle plots showing Tregs regulate epithelial cells via the TGF-β pathway, which is solely activated in ADC. G. Bubble plot showing the probability of ligand-to-receptor combination of TGF-β pathway between Tregs and epithelial cells, by comparing ADC with SCC. The pathways of ADGRE5, CD46, GZMA and NAMPT are used as negative controls. H. Dual IF staining confirming that in SLC26A3high regions of CC tissues, more FOXP3+ cells are recruited than in SLC26A3low regions (left). The numbers of recruited FOXP3+ cell are quantified using histogram plot (right). Three individual samples with ROI were calculated and P<0.01 were marked with **, showing significant difference. I. Multiplexed IF staining confirming the interaction between CD6 (on FOXP3+ cells) and ALCAM (on SLC26A3+ epithelial cells) in the ALCAM pathway. J. Multiplexed IF staining showing that the recruitment of FOXP3+ cells towards SLC26A3+ cells might induce EMT (marked with E-cadherin) and increase the stemness (marked with ALDH1A1) of tumor cells, via TGF-β pathway.