Reciprocal interactions between malignant cells and macrophages enhance cancer stemness and M2 polarization in HBV-associated hepatocellular carcinoma

Abstract

Background: The tumor microenvironment of cancers has emerged as a crucial component in regulating cancer stemness and plays a pivotal role in cell-cell communication. However, the specific mechanisms underlying these phenomena remain poorly understood. Methods: We performed the single-cell RNA sequencing (scRNA-seq) on nine HBV-associated hepatocellular carcinoma (HCC) patients. The heterogeneity of the malignant cells in pathway functions, transcription factors (TFs) regulation, overall survival, stemness, as well as ligand-receptor-based intercellular communication with macrophages were characterized. The aggressive and stemness feature for the target tumor subclone was validated by the conduction of in vitro assays including sphere formation, proliferation, Annexin V apoptosis, flow cytometry, siRNA library screening assays, and multiple in vivo preclinical mouse models including mouse hepatoma cell and human HCC cell xenograft models with subcutaneous or orthotopic injection. Results: Our analysis yielded a comprehensive atlas of 31,664 cells, revealing a diverse array of malignant cell subpopulations. Notably, we identified a stemness-related subclone of HCC cells with concurrent upregulation of CD24, CD47, and ICAM1 expression that correlated with poorer overall survival. Functional characterization both in vitro and in vivo validated S100A11 as one of the top downstream mediators for tumor initiation and stemness maintenance of this subclone. Further investigation of cell-cell communication within the tumor microenvironment revealed a propensity for bi-directional crosstalk between this stemness-related subclone and tumor-associated macrophages (TAMs). Co-culture study showed that this interaction resulted in the maintenance of the expression of cancer stem cell markers and driving M2-like TAM polarization towards a pro-tumorigenic niche. We also consolidated an inverse relationship between the proportions of TAMs and tumor-infiltrating T cells. Conclusions: Our study highlighted the critical role of stemness-related cancer cell populations in driving an immunosuppressive tumor microenvironment and identified the S100A11 gene as a key mediator for stemness maintenance in HCC. Moreover, our study provides support that the maintenance of cancer stemness is more attributed to M2 polarization than the recruitment of the TAMs.

Keywords: Cancer stem cell; Hepatocellular carcinoma; Macrophage polarization; Single-cell RNA sequencing; Tumor heterogeneity.

Figure 1

scRNA-seq of hepatocellular carcinoma revealed major cell types. (A) Schematic overview of the generation of scRNA-seq data. Nine HCC samples were collected. (B) t-SNE plot of 31,664 high-quality cells. Each point is a single cell colored by cluster assignment. (C) Violin plots of canonical marker genes expression by cell type with highest log-normalized expression value labeled. (D) t-SNE plot, color-coded for expression (gray to red) of marker genes for the major cell types as indicated. (E) Representative donut plots depicting distribution of overall immune cell composition. (F) The tumor cell lineage compositions inferred by scRNA-seq data of nine HCC sample. The low panel shows the HCC-defined cell types (rows) by patient (columns). The size of the circle represents, for each specific cell type, the fraction of corresponding cell (among the total quality-control passed cells) in each individual. The circles are color coded by defined cell types. The histogram on the top shows, for each individual sample, the accumulation of the raw number of each cell type. (G) Linear regression showing the inverse correlation between T cells and macrophages proportion. Pearson's correlation R values are shown and the significance of the difference between two data sets was measured by two-tailed Student's t-test. Only cases with at least 500 non-malignant cells were included.

Figure 2

Identification of cancer cell subpopulations. (A) t-SNE plot of cancer cells, color-coded by the subsets. (B) t-SNE plot, color-coded for expression (gray to red) of marker genes for the subsets as indicated. (C) Differential activity pathways in cancer cell subsets (scored by GSVA for each cell). Red represents upregulated pathways; blue represents down-regulated pathways. (D) Heatmap showing the regulon activity of cancer cell subsets estimated by SCENIC, depicting the top three transcription factors estimates from one cluster versus all the other clusters. (E) Kaplan-Meier survival plots of LIHC patients (n = 371) from TCGA, the expressions of top ten genes were used to stratify patients into binary subgroups (high and low).

Figure 3

CSC-associated heterogeneity in HCC. (A) t-SNE plot, color-coded for expression (gray to red) of typically reported cancer stem cell markers in cancer cells. (B) DotPlot visualization of the relative expression of cancer stem cell markers in each subset, where the size represents the percentage of cells expressing the gene of interest and the color indicates the scaled average expression of the gene of interest across the various. (C) Intra-heterogeneity at the gene expression level for a panel of CSC markers. (D) Violin plot showing stemness signature in subpopulations of HCC. The middle box shows the median and interquartile range (IQR 25th-75th percentiles). (E) Bar plot showing the fraction of cells with triple expression of CD24/CD47/ICAM1 in each patient.

Figure 4

S100A11 enhanced cancer cells stemness both in vivo and vitro. (A) Bar plot showing the expression log2FC of up/down-regulated DEGs. Significance was determined by Wilcoxon test at p-value < 0.05. (B) and (C) Spheroid formation assay showed that S100A11 knockdown by two independent shRNA sequences dramatically reduced the ability of PLC/PRF/5 and CLC7 cells to form spheres compared with the non-treated control (NTC) after 10-day incubation. Scale bar, 250 µm. The data are reported as mean ± SEM (**p < 0.01). (D) and (E) S100A11 knockdown and non-target control (NTC) PLC/PRF/5 cells treated with and without cisplatin were analyzed by Annexin V assays using flow cytometry detection. (F) and (G) In vivo limiting dilution assays with varying numbers of NTC and S100A11-knockdown cells for both PLC/PRF/5 and CLC7 cells subcutaneously injected into NOD-SCID mice (n = 4/group), respectively, were performed. Tumor incidence was examined at day 70 post-inoculation and frequency of tumor initiating cells was calculated.

Figure 5

Dissection and subclusters of lymphocytes/monocytes in HCC. (A) t-SNE plot of lymphocytes (exclude B cells), color-coded by the subsets. (B) t-SNE plot, color-coded for expression (gray to red) of marker genes for the subsets as indicated. (C) Differential activity pathways in the five lymphocyte subsets (scored by GSVA for each cell). Red represents upregulated pathways; blue represents down-regulated pathways. (D) Heatmap showing the regulon activity of five lymphocyte subsets estimated by SCENIC, depicting the top five transcription factors from one cluster versus all other clusters. (E) t-SNE plot of monocytes, color-coded by the subsets. (F) t-SNE plot, color-coded for expression (gray to red) of marker genes for the subsets as indicated. (G) Heatmap showing the regulon activity of ten monocyte subsets estimated by SCENIC, depicting the top five transcription factors from one cluster versus all the other clusters. (H) Differential activity pathways in the ten monocyte subsets (scored by GSVA for each cell). Red color represents upregulated pathways; blue color represents down-regulated pathways.

Figure 6

Interrogation of cell-cell communication between TAMs and cancer cells in HCC. (A) Heatmap showing the number of potential ligands and receptors among the cell components. (B) Heatmap showing the number of potential ligand-receptor pairs between TAM subsets and cancer cell subsets (TAMs as ligands source and cancer cells as receptor source). (C) Module scores of M2 signatures for each TAMs subpopulation (Genes list in from Azizi et al.). (D) Ligand-receptor pair expression according to cell type. Ligands are indicated in the left panel, and receptors are indicated in the right panel. (E) Module scores of M2 signatures for CD24/CD47/ICAM1high HCC tumors compared to the rest tumor samples in the TCGA-LIHC dataset. Significance was determined by Wilcoxon test at p-value < 0.05.

Figure 7

Reciprocal interactions between TAM polarization and cancer cell stemness by co-culture assays. (A) Barplot representing the mean expression value of the CSC markers using qPCR, normalized to three reference genes. Error bars indicate the standard error. *p < 0.05, **p < 0.01, ***p < 0.001. (B) Barplot representing the mean expression value of M1/M2 macrophage markers using qPCR, normalized to three reference genes. Error bars indicate the standard error. *p < 0.05, **p < 0.01, ***p < 0.001.