Single-cell analyses reveal cannabidiol rewires tumor microenvironment via inhibiting alternative activation of macrophage and synergizes with anti-PD-1 in colon cancer

Abstract

Colorectal tumors often create an immunosuppressive microenvironment that prevents them from responding to immunotherapy. Cannabidiol (CBD) is a non-psychoactive natural active ingredient from the cannabis plant that has various pharmacological effects, including neuroprotective, antiemetic, anti-inflammatory, and antineoplastic activities. This study aimed to elucidate the specific anticancer mechanism of CBD by single-cell RNA sequencing (scRNA-seq) and single-cell ATAC sequencing (scATAC-seq) technologies. Here, we report that CBD inhibits colorectal cancer progression by modulating the suppressive tumor microenvironment (TME). Our single-cell transcriptome and ATAC sequencing results showed that CBD suppressed M2-like macrophages and promoted M1-like macrophages in tumors both in strength and quantity. Furthermore, CBD significantly enhanced the interaction between M1-like macrophages and tumor cells and restored the intrinsic anti-tumor properties of macrophages, thereby preventing tumor progression. Mechanistically, CBD altered the metabolic pattern of macrophages and related anti-tumor signaling pathways. We found that CBD inhibited the alternative activation of macrophages and shifted the metabolic process from oxidative phosphorylation and fatty acid oxidation to glycolysis by inhibiting the phosphatidylinositol 3-kinase-protein kinase B signaling pathway and related downstream target genes. Furthermore, CBD-mediated macrophage plasticity enhanced the response to anti-programmed cell death protein-1 (PD-1) immunotherapy in xenografted mice. Taken together, we provide new insights into the anti-tumor effects of CBD.

Keywords: Cannabidiol; Colorectal cancer; Macrophage; Tumor microenvironment; scATAC-seq; scRNA-seq.

Graphical abstract

Fig. 1

Cannabidiol (CBD) inhibits the growth of transplanted colon cancer in mice. MC38 cancer cells (1 × 10⁶) were inoculated subcutaneously into each mouse (n = 5). When the tumor grew to 100 mm³, the mice were randomly divided into a vehicle group (i.p., once a day), a 5-fluorouracil (5-Fu) group (25 mg/kg, i.p., once every other day), and a CBD different dose group (5 mg/kg and 10 mg/kg, i.p., once per day). (A) Representative images of the tumor. (B) Tumor growth curve. (C) Tumor weight. (D) Tumor volumes of individual mice. (E, F) Terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay for apoptotic cells in tumor tissue. (G, H) Immunohistochemistry (IHC) for detecting Ki67 in the tumor tissue. Data are represented as mean ± standard error of mean. ^∗P < 0.05 and ^∗∗P < 0.01.

Fig. 2

Single-cell RNA sequencing (scRNA-seq) analysis of cannabidiol (CBD) inhibiting the growth of MC38 transplanted tumor. (A) Schematic diagram of scRNA-seq and single-cell ATAC sequencing (scATAC-seq). (B) The uniform manifold approximation and projection (UMAP) visualization shows unsupervised scRNA-seq clustering, revealing 5 major clusters (left) and 15 minor clusters (middle) in two groups (right). (C) Expression of example marker genes. (D) Heatmap of row-wise Z-score-normalized mean expression of selected marker genes. (E) 3D pie chart of the major cluster proportion in vehicle and CBD. (F) The overall networks of cell-cell communication within myeloid sub-populations between vehicle and CBD. DCs: dendritic cells; NK: nature killer; Treg: regulatory T cells.

Fig. 3

Cannabidiol (CBD) promotes the polarization of M2-like macrophages to M1-like macrophages. (A) The uniform manifold approximation and projection (UMAP) of myeloid subclusters in vehicle and CBD. (B) Dot plots showing representative top marker genes across the myeloid subtypes. (C) Heatmap showing the expression of marker genes (top 50) in the indicated myeloid subtypes. Selected marker genes are highlighted. (D) Pseudo-time trajectory analysis of myeloid cells based on group, cell subtypes, and pseudo-time (upper panel) and faceted plots showing cell types in pseudo-time series (lower panel). (E) Trajectory tree showing the differentiation of monocytes and macrophages subtypes. (F) Ridge plots showing the differentiation of monocyte and macrophage subtypes along the pseudo-time. (G) Chordal graph showing the differential interaction numbers of macrophage sub-populations interacting with cancer cells between vehicle and CBD. (H) Gene set enrichment analysis (GSEA) showing significant enrichment of selected hallmark pathways in CBD M1-like macrophages compared with vehicle. (I) Ridge plots showing the gene set variation analysis (GSVA) of significant hallmark pathways in CBD M2-like macrophages compared with vehicle. (J, K) The cnet plot showing the relationship between the selected pathways with downstream genes. DCs: dendritic cells; PI3K: phosphatidylinositol 3-kinase; Akt: protein kinase B; mTOR: mechanistic target of rapamycin; IL: interleukin; JAK: Janus kinase.

Fig. 4

Identified specific regulators of the maintenance of macrophage subtypes identity. (A) Dot plot showing the transcription factor (TF) regulons that regulate macrophage subtypes predicted by pySCENIC. The top five TFs are highlighted. (B) Heatmap showing the enrichment of the TF in specific macrophage subtypes between vehicle and cannabidiol (CBD). (C) Heatmaps showing different expression patterns of the indicated signature genes among myeloid subtypes calculated by gene set variation analysis (GSVA). (D) Gene regulatory networks (GRNs) by the top 10 TF regulons and their targeted genes. TF regulons are highlighted by yellow rectangles, and target genes are shown by circles. Pie charts of nodes indicate the relative gene expression level between the two groups. (E) Dot plot showing the inflammation-related genes in M1-like macrophages between the two groups. (F) Raincloud plots showing the selected gene expression difference for M2-like macrophages from vehicle and CBD groups. Data are represented as mean ± standard error of mean. ^∗∗∗P < 0.001. DCs: dendritic cells.

Fig. 5

Integrative analysis of myeloid cells of single-cell RNA sequencing (scRNA-seq) and single-cell ATAC sequencing (scATAC-seq) data in the vehicle and cannabidiol (CBD) treatment group. (A, B) Uniform manifold approximation and projection (UMAP) plot showing the joint clustering of scRNA-seq (blue) and scATAC-seq (red) data in colon. Cells in the right UMAP are colored based on cell types annotated by scRNA-seq data. (C) Marker genes expression based on scATAC-seq data. (D) Bar plot showing the number of reproducible peaks identified from each cluster. The peaks are classified into four categories: distal, exonic, intronic, and promoter. (E) Dot plot showing the identification of positive transcription factor (TF) regulators through correlation of chromVAR TF deviation scores and gene expression in cell groups (Pearson correlation r > 0.5, adjusted P-value <0.01, and deviation difference in the top 25th percentile). (F) TF footprint for the Stat1, Stat3, Irf3, and Irf5 motif with the subtraction using the Tn5 bias normalization method. (G) Dot plot showing the M1-like macrophage-related genes between the two groups. (H) Profile of the Stat1, Stat3, Irf3, and Irf5 TF motif activity, gene chromatin accessibility, and gene expression (inferred from scRNA-seq). (I) Genome track visualization of the Stat1 locus (chr16:36,616,076−36,716,077). Inferred peak-to-gene links for distal regulatory elements are shown below. (J) Heatmap showing the positive TF regulators for which gene expression is positively correlated with TF deviation (inferred by chromVAR) across the M1-like macrophages cell trajectory. DCs: dendritic cells; TSS: transcriptional start site.

Fig. 6

Cannabidiol (CBD) suppresses alternative activation of macrophage in vitro. (A) Immunofluorescence staining of tumor tissue from vehicle or CBD treatment mice with CD206 and inducible NO synthase (iNOS). (B–D) Bone marrow-derived macrophages (BMDMs) were stimulated with media or interleukin-4 (IL-4) (20 ng/mL) then treated with dimethyl sulfoxide (DMSO) (vehicle control) or increasing concentrations (μM) of CBD for 24 h. (B) Flow cytometry data for the M2-specific surface marker CD206 (left) and M1 surface marker CD86 (right). (C) Quantitative real-time polymerase chain reaction (qRT-PCR) results showing mRNA expression of Arg1, Cd206, Chil3, and Retnla normalized to β-actin. (D) qRT-PCR results showing mRNA expression of Il12b and Nos2 normalized to β-actin. (E, F) BMDMs were stimulated with medium or IL-4 and treated with DMSO or CBD for 24 h. Immunofluorescence images showing the expression of CD206 (E) and iNOS (F). (G) BMDMs were stimulated with medium or IL-4 (20 ng/mL) and treated with DMSO or CBD for 24 h then co-cultured with MC38-mCheery cells for another 48 h. (H) Time course images of organoid on day 1 and day 4 (n = 6–8 for each time point). Data are represented as mean ± standard error of mean. ^∗P < 0.05 and ^∗∗P < 0.01. ns: no significance.

Fig. 7

Cannabidiol (CBD) regulates macrophage polarization by inhibiting phosphatidylinositol 3-kinase (PI3K)/protein kinase B (Akt) signaling. (A) Oxygen consumption rate (OCR) and (B) extracellular acidification rate (ECAR) of interleukin-4 (IL-4)-primed bone marrow-derived macrophages (BMDMs) treated with dimethyl sulfoxide (DMSO) or CBD for 24 h. (C–E) BMDMs were stimulated with media or IL-4 (20 ng/mL) then treated with DMSO (vehicle control) or increasing concentrations (μM) of CBD for 24 h. (C) Lactic acid content, (D) malondialdehyde (MDA) content, and (E) adenosine triphosphate (ATP) production in polarized macrophage. (F) Bubble plot showing the expressions of oxidative phosphorylation-related genes from vehicle and CBD groups. (G) BMDMs grown on coverslips treated with media or IL-4 with or without CBD. The cells were analyzed according to the section titled ‘BODIPY’ staining for microscopy. (H, I) BMDMs cells were pretreated with DMSO or increasing concentrations of CBD for 2 h and then stimulated with media or IL-4 for the indicated time. The protein levels of p-PI3K, PI3K, p-Akt, and Akt were analyzed by Western blot. β-Actin is shown as a loading control. (J) Raincloud plots show the expression of glycolysis and lipid metabolism gene in macrophages. Data are represented as mean ± standard error of mean. ^∗P < 0.05, ^∗∗P < 0.01, and ^∗∗∗P < 0.001. FCCP: fluoro-carbonyl cyanide phenylhydrazone; 2-DG: 2-deoxy-d-glucose; Gapdh: glyceraldehyde-3-phosphate dehydrogenase.

Fig. 8

The combination of cannabidiol (CBD) and programmed cell death protein 1 (PD-1) antibody shows stronger anti-tumor effect. (A–C) Effects of CBD and anti-PD-1 on tumor growth. (B) Tumor volume and (C) tumor weight of MC38-derived tumors in the subcutaneous C57BL/6 mouse model. (D) Hematoxylin & eosin (H&E) and (E) proliferating cell nuclear antigen (PCNA) staining of tumor sections. Representative immunofluorescent staining for M1 cells (inducible NO synthase (iNOS)) and (F) M2 cells (CD206) and (G) CD8⁺ T cells in tumor. (H) Percentage of CD206 and iNOS positive cells in panel. Data are represented as mean ± standard error of mean. ^∗P < 0.05 and ^∗∗P < 0.01. DAPI: 4’-6-diamidino-2-phenylindole; GZMB: granzyme B.

Fig. 9

Graphic illustration of the anti-tumor mechanism of cannabidiol (CBD) in colon cancer. The tumor microenvironment (TME) of all cell types in mouse MC38 xenograft tumors after CBD treatment was investigated by single-cell RNA sequencing (scRNA-seq) and single-cell ATAC sequencing (scATAC-seq). CBD inhibits the alternative activation of macrophages and shifts the metabolic process from oxidative phosphorylation and fatty acid oxidation to glycolysis by inhibiting the phosphatidylinositol 3-kinase (PI3K)-protein kinase B (Akt) signaling pathway and related downstream target genes, which relieves the inhibitory immune microenvironment and restores the intrinsic anti-tumor effects of macrophages. Furthermore, CBD-mediated macrophage plasticity enhances the response to anti-programmed cell death protein 1 (PD-1) immunotherapy in xenografted mice. OXPHOS: oxidative phosphorylation; DCs: dendritic cells; RTKs: receptor protein tyrosine kinases; IFN-γ: interferon-gamma.

figs1

Supplementary Figure 1. CBD suppresses colon tumor growth in mice. (A) Body weight of mice. (B) Representative hematoxylin-eosin-stained section of the tumor. Scale bar: 100 μm. (C) qRT - PCR for Ifnγ, Gzmb, Perforin, and FasL expression in the tumor. Data are represented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01. Scale bar: 100 μm. qRT - PCR (quantitative real-time polymerase chain reaction).

figs2

Supplementary Figure 2. Quality control (QC) of scRNA-seq and scATAC-seq data. (A–B) Violin plots for scRNA-seq data quality control (A) and after quality control (B) measures by sample. Plots show the number of genes (nFeature_RNA), the number of UMIs (nCount_RNA), the percent of mitochondrial content (percent.mt), and the percent of ribosome content (percent.rb) for samples of vehicle and CBD. (C) Scatter plot showing the TSS enrichment score vs. unique nuclear fragments per cell. The color of the dots represents the density of each point in the plot. (D) Fragment size distributions of vehicle and CBD (left). Aggregate TSS insertion profiles centered at all TSS regions. The cells shown passed the ArchR QC thresholds for each sample (right). CBD (cannabidiol); TSS (transcriptional start site), scRNA-seq (single-cell RNA sequencing); scATAC-seq (single-cell ATAC sequencing); UMAP (uniform manifold approximation and projection).

figs3

Supplementary Figure 3. Cell–cell communication analysis for major cell clusters. (A) Bar plots showing the proportion of cells (left) and the number of differentially expressed genes (DEGs) in minor cell types of different samples (right). (B) The number of inferred interactions and interaction strength for two groups. (C) The differential interaction numbers and strength of each cell type for the vehicle and CBD groups. (D) The relative information flow between the two groups. (E) The bubble plot shows significant up-regulated ligand–receptor pairs between sender and receiver cells, colored according to group types. CBD (cannabidiol); DCs (dendritic cells); NK nature killer; Treg (regulatory T cells).

figs4

Supplementary Figure 4. Pseudo-time trajectory analysis and scATAC-seq data analysis. (A) Heatmap showing the dynamic changes of top 100 genes expression along the pseudo-time trajectory. (B) UMAP plot showing the joint clustering of scRNA-seq (blue) and scATAC-seq (red) data in vehicle and CBD. Cells in the right UMAP are colored based on cell types annotated by scRNA-seq data. (C) Marker genes expression based on scATAC-seq data. (D) Hypergeometric enrichment of TF motifs in myeloid cells marker peaks. (E) Heatmap showing chromatin accessibility and gene expression of 56,287 significantly (R > 0.5 and FDR < 0.01) linked peak-gene pairs. (F) Heatmap showing row-normalized chromatin accessibility of 11,133 cell type-specific peaks (FDR < 0.1, log2FC > 2). (G) Heatmap showing the dynamic of motif deviation score, gene-score, chromatin accessibility, and RNA expression across the macrophages pseudo-time trajectory. scATAC-seq (single-cell ATAC sequencing); UMAP (uniform manifold approximation and projection); scRNA-seq (single-cell RNA sequencing); CBD (cannabidiol); TF (transcription factor); FDR (false discovery rate); DCs (dendritic cells); M1 (M1-like macrophages); M2 (M2-like macrophages).

figs5

Supplementary Figure 5. CBD inhibits tumor growth in a macrophage-dependent manner. (A) Schematic of experiment workflow. (B) Tumor growth curves of MC38 with intravenous injection of clodronate liposomes (Clo-Lip) and control liposomes (Ctrl-Lip) as indicated (B, Ctrl-Lip: n = 5, Clo-Lip: n = 7). (C–D) Tumor photos (C) and tumor weight (D). (E) Immunofluorescence for F4/80 staining of tumor tissues. Scale bar: 100 μm. Data are represented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01. ns = no significance. CBD (cannabidiol).

figs6

Supplementary Figure 6. Transcriptome dynamics of lymphoid cells. (A) UMAP plot of lymphoid cell in group and sub-clusters. (B) Proportion of lymphoid cell sub-populations between vehicle and CBD. (C) Chordal graph showing the differential interaction numbers of lymphoid cell sub-populations interacting with cancer cells between vehicle and CBD. (D) Stacked violin plots and UMAP plots of expression for selected marker genes. (E) Cell–cell communication analysis showing the overall signaling patterns for major cell clusters between vehicle and CBD. (F) Bubble plot showing genes expression difference for lymphoid cells from vehicle and CBD groups. UMAP (uniform manifold approximation and projection); CBD (cannabidiol); NK nature killer; Treg (regulatory T cells).

figs7

Supplementary Figure 7. CBD inhibits anti-inflammatory macrophage polarization and promotes pro-inflammatory macrophage polarization. (A) BMDMs were treated with different concentrations of CBD as indicated for 24 h. The cells were harvested, stained with Annexin V and PI, and detected by flow cytometry. (B) BMDMs were treated with different concentrations of CBD for 24 h, and cell survival was detected by cell counting kit-8 (CCK8). (C) BMDMs were treated with a different concentration of CBD for 24 h; cell lysates were probed by anti-p62, anti-LC3B, and anti-Gapdh. (D) Percentage of CD206+ BMDMs (left) and CD86+ BMDMs (right). (E) BMDMs were stimulated with media or IL-4 (20 ng/mL) then treated with DMSO (vehicle control) or CBD (10 μM) at different times as indicated. Gene expressions were determined by qRT - PCR. (F) Schematic diagram of coculture of macrophages and tumor cells. (G) BMDMs were stimulated with medium or IL-4 and treated with DMSO or CBD for 24 h, then co-cultured with MC38-mCheery cells for another 48 h. Scale bar: 100 μm. (H) THP-1 cells were stimulated with phorbol-12-myristate-13-acetate (PMA, 500 nM) for 3 h to induce M0 macrophages, then cells were treated with recombinant human IL-4 (20 ng/mL) in DMSO or CBD (10 μM) containing medium to differentiate into M2 macrophages. Gene expressions were determined by qRT - PCR. (I) Schematic image for organoid co-culture with macrophages. (J) Organoid volume. n = 6–8 for each time point. Data are represented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01. ns = no significance. CBD (cannabidiol); BMDM (bone marrow-derived macrophages); qRT – PCR (quantitative real-time polymerase chain reaction); IL-4 (interleukin-4); DMSO (dimethyl sulfoxide).

figs8

Supplementary Figure 8. CBD inhibits the activation of PI3K-AKT signaling in alternative activation of macrophages. (A–B) BMDMs were pre-treated with DMSO or increasing concentrations of CBD for 2 h, then stimulated with media or IL-4 for the indicated time. Protein levels of p-JAK1, JAK1, p-STAT6, and STAT6 were analyzed by western blot. β-Actin is shown as a loading control. (C) OCR of BMDMs treated with medium or IL-4 for 24 h. (D–E) The mix respiration (D) and basic respiration (E) of OCR were calculated. (F–G) The proton efflux rate (PER) was derived from glycolysis (F), and the basic glycolysis is shown (G). (H) BMDMs were stimulated with media or IL-4 (20 ng/mL) then treated with DMSO (vehicle control) or increasing concentrations (μM) of CBD for 24 h. Neutral lipid content (BODIPY 493) was compared between vehicle and CBD treatment groups. Scale bar: 100 μm. (I) BMDMs were pre-treated with DMSO or increasing concentrations of CBD for 2 h then stimulated with media or IL-4 for the indicated time. Protein levels of p-PI3K and p-AKT were analyzed by western blot. β-Actin shown as a loading control. (J) Bubble plots show the expression of genes related to macrophage function in two groups. Data are represented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01. CBD (cannabidiol); BMDM (bone marrow-derived macrophages); IL-4 (interleukin-4); DMSO (dimethyl sulfoxide); OCR (oxygen consumption rate).

figs9

Supplementary Figure 9. CBD enhances the immune checkpoint blockade-triggered immune responses. MC38 cells were inoculated subcutaneously into mice, and the mice were treated with vehicle or CBD, anti-PD-1, or a combination of CBD and anti-PD-1. (A) IHC score of PCNA. (B) Gating strategies to determine the proportion of infiltrating T cells in tumors as well as T cell function. (C) Flow cytometry analysis of CD8+ infiltrating T cells proportion in the tumor. (D) Flow cytometric analysis of CD8+ infiltrating T cell function in tumors. Data are represented as mean ± SEM. ∗P < 0.05, ∗∗P < 0.01. ns = no significance. CBD (cannabidiol); PD-1 (programmed cell death protein 1); IHC (immunohistochemistry); PCNA (proliferating cell nuclear antigen).