Single-cell RNA sequencing reveals Shen-Bai-Jie-Du decoction retards colorectal tumorigenesis by regulating the TMEM131-TNF signaling pathway-mediated differentiation of immunosuppressive dendritic cells

Abstract

Colorectal tumorigenesis generally progresses from adenoma to adenocarcinoma, accompanied by dynamic changes in the tumor microenvironment (TME). A randomized controlled trial has confirmed the efficacy and safety of Shen-Bai-Jie-Du decoction (SBJDD) in preventing colorectal tumorigenesis. However, the mechanism remains unclear. In this study, we employed single-cell RNA sequencing (scRNA-seq) to investigate the dynamic evolution of the TME and validated cell infiltration with multiplex immunohistochemistry and flow cytometry. Bulk RNA sequencing was utilized to assess the underlying mechanisms. Our results constructed the mutually verifiable single-cell transcriptomic atlases in Apc ^Min/+ mice and clinical patients. There was a marked accumulation of CCL22⁺ dendritic cells (DCs) and an enhanced immunosuppressive action, which SBJDD and berberine reversed. Combined treatment with cholesterol and lipopolysaccharide induced characteristic gene expression of CCL22⁺ DCs, which may represent "exhausted DCs". Intraperitoneal injection of these DCs after SBJDD treatment eliminated its therapeutic effects. TMEM131 derived CCL22⁺ DCs generation by TNF signaling pathway and may be a potential target of berberine in retarding colorectal tumorigenesis. These findings emphasize the role of exhausted DCs and the regulatory mechanisms of SBJDD and berberine in colorectal cancer (CRC), suggesting that the multi-component properties of SBJDD may help restore TME homeostasis and offer novel cancer therapy.

Keywords: CCL22+ dendritic cells; CD4+ regulatory T cells; Colorectal tumorigenesis; Single-cell RNA sequencing; TMEM131; TNF signaling pathway; Traditional Chinese medicine; Tumor microenvironment.

Graphical abstract

Figure 1

The effect of Shen-Bai-Jie-Du decoction (SBJDD) on tumorigenesis and copy number variation (CNV) patterns in Apc^Min/+ mice. (A) Schematic diagram of the experimental workflow. (B–D) Comparison of tumor number and volume among groups. Red arrowheads indicate colorectal tumors. (E) Hematoxylin and eosin staining of the colorectum among groups. (F, G) Immunohistochemistry (IHC) analysis of Ki67-positive cells in colorectal tumor tissues. (H) UMAP plot displaying the clustering of 48,992 single cells. (I) Forty distinct cell subtypes were identified. Cell subtypes were annotated using known markers, and each subtype is color-coded. (J) UMAP plot of malignancy levels in epithelial cell subgroups. (K) Violin plot shows CNV scores of epithelial cells. Data are presented as mean ± SD (n = 7). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 compared to the normal group; ^##P < 0.01 compared to the Apc^Min/+ group.

Figure 2

Dynamic alterations and the development trajectory of Ccl22⁺ dendritic cells (DCs) in tumor microenvironment. (A) Heatmap of the observed to expected cell numbers (Ro/e) ratio for each cell cluster. ±, 0≤Ro/e < 1; +, 1≤Ro/e < 1.5; ++, 1.5≤Ro/e < 2; +++, 2≤Ro/e < 2.5; ++++, Ro/e ≥ 2.5; (B) AUCell scores of each cell subgroup in normal and colorectal cancer (CRC) groups based on the cancer genome atlas (TCGA) database. (C, D) Multiplex immunohistochemistry (mIHC) shows the infiltration of CCL22⁺ DCs in different groups. (E) UMAP plot displaying 9 myeloid cell subtypes. (F) Pseudo-time analysis mapping the dynamic evolution of myeloid cell subtypes during colorectal tumorigenesis and SBJDD treatment. (G) Pseudo-time trajectory analysis shows the developmental origin of Ccl22⁺ DCs from Wdfy4⁺ DCs. (H) Scatter plot of Pseudo time-dependent genes in Ccl22⁺ DCs. (I) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of pseudo-time-dependent upregulated genes in Ccl22⁺ DCs. Data are presented as mean ± SD (n = 3). ns, no significance; ∗∗P < 0.01, ∗∗∗P < 0.001.

Figure 3

The potential immunosuppressive action of CCL22⁺ DCs. (A) The interaction intensity between DCs and various T cell types in different groups. (B) Communication intensity and associated molecules between different DC subsets and Cd4⁺ regulatory T cells (Tregs). (C) mIHC shows the abundance and spatial distance changes of CCL22⁺ DCs and Tregs in different groups. (D) Deconvolution analysis based on the TCGA database shows the abundance of CCL22⁺ DCs in CRC and normal tissues. (E) Correlation analysis shows a significant association between the expression of characteristic genes of CCL22⁺ DCs and Tregs. (F) The positive rates of CD4⁺ FOXP3⁺ Tregs or CCL22⁺ DCs in a tissue microarray (TMA). The locations of column 1 (rows 1 to 9), column 2 (rows 1 to 6), and column (row 5) in the TMA are colorectal adenoma tissues, and the others are cancer tissues. Detailed characteristics of the TMA patients are provided in Table S1. (G) Correlation between the positive rates of CD4⁺ FOXP3⁺ Tregs and CCL22⁺ DCs in TMA. Data are presented as mean ± SD (n = 3). ∗∗P < 0.01, ∗∗∗P < 0.001.

Figure 4

Single-cell transcriptomic analysis in clinical paired samples. (A) Schematic diagram of single-cell RNA sequencing experiment on paired samples from patients with multiple colorectal adenomas. (B) Colonoscopy shows morphological changes of the same adenoma before and after SBJDD treatment. (C) UMAP plot displaying the clustering of 39,500 single cells in colorectal adenoma samples before and after SBJDD treatment. (D) CNV scores of epithelial cells before and after SBJDD treatment (n = 3). (E) Changes in cell ratio of each subtype before and after SBJDD treatment. (F) DC subtypes and development trajectory. (G) Similarity of DC subtypes in the two scRNA-seq results. ∗∗∗P < 0.001.

Figure 5

SBJDD inhibited colorectal tumorigenesis via CCL22⁺ DCs. (A) Measurement of total cholesterol levels in mouse serum (B) Differential expression of CD83 in M05 Ccl22⁺ DCs and M03 Wdfy4⁺ DCs. (C) Gene expression analysis of imDCs treated with lipopolysaccharide, cholesterol, and their combination. (D) Flow cytometry analysis of the mean fluorescence intensity (MFI) value of CTLA4 in CD4⁺CD25⁺ Tregs after co-culture with different DCs. (E, F) Comparison of tumor number and volume among groups of Apc^Min/+ mice. Red arrowheads indicate colorectal tumors. (G) Changes in body weight of Apc^Min/+ mice in different groups. (H) Hematoxylin and eosin staining of colorectal tumor tissues. (I) IHC analysis of Ki67-positive cells in colorectal tumor tissues. (J, K) mIHC analysis shows the infiltration of CCL22⁺ DCs and Tregs in the colorectal tumor tissues. (L) Measurement of total cholesterol levels in mouse serum. (M) qPCR analysis of Ccl22 and Ccr7 in colorectal tumor tissues. Data are presented as mean ± SD (n = 7). ns, no significance; ∗P < 0.05, ∗∗P < 0.01.

Figure 6

TMEM131 regulates CCL22⁺ DCs generation by the TNF signaling pathway. (A) The heatmap shows the differentially expressed genes (DEGs) in Ccl22⁺ DCs from different groups. (B) Venn diagram identifies intersecting genes between DEGs in different groups and pseudo-time-dependent genes. (C) Dynamic changes in Tmem131 expression during the differentiation of Ccl22⁺ DCs from Wdfy4⁺ DCs. (D, E) qPCR analysis of Tmem131 mRNA expression in colorectal tissue samples and Ccl22⁺ DCs. (F) qPCR analysis of Tmem131 and Ccl22 mRNA expression after TMEM131 knockdown. (G) DEGs after TMEM131 knockdown. (H) KEGG pathway analysis of upregulated/downregulated genes after TMEM131 knockdown. (I) The enrichment of Tmem131 and TNF signaling pathway in DC subtypes. (J) The correlation between TMEM131 and TNF pathway in TCGA datasets (n = 499). (K) qPCR analysis of Traf2 and Tnfa mRNA expression after TMEM131 knockdown. (L) Western blot analysis of TRAF2 protein levels after TMEM131 knockdown. (M) ELISA analysis of TNFA protein levels after TMEM131 knockdown. Data are presented as mean ± SD (n = 3). ns, no significance; ∗∗P < 0.01, ∗∗∗P < 0.001.

Figure 7

SBJDD and its active compound berberine modulated the function of CCL22⁺ DCs via binding with TMEM131. The chemical base peak ion (BPI) chromatogram of SBJDD plasma in the positive (A) and negative (B) ion mode using ultrahigh-performance liquid chromatography with quadrupole time-of-flight mass spectrometry. (C) Structural classification of prototype compounds in plasma. (D) The number of prototype compounds in plasma for each herb. (E) Tissue distribution of typical compounds of SBJDD. (F) The chemical structure of berberine. (G) Molecular docking results for TMEM131 (AlphaFold ID: AF-O70472-F1) with berberine. (H) Gene expression analysis of CCL22⁺ DCs treated with berberine (20 μmol/L). (I) The MFI value of CTLA4 in CD4⁺CD25⁺ Tregs after co-culture with Ccl22⁺ DCs treated with berberine. (J, K) The protein levels of TRAF2 and TNFA in Ccl22⁺ DCs after berberine treatment. (L) Surface plasmon resonance assay obtained separately on TMEM131-coated chip in the presence of different concentrations of berberine. (M) Diagram of the mechanism of SBJDD retards colorectal tumorigenesis. Data are presented as mean ± SD (n = 3). ns, no significance; ∗∗P < 0.01, ∗∗∗P < 0.001.