Cannabidiol suppresses emergency MDSCs generation by disturbing EEF1B2-mediated C/EBPβ protein synthesis in colorectal adenomas

Abstract

Background

Colorectal cancer often develops from adenomas over years, necessitating early intervention. Myeloid-derived suppressor cells (MDSCs) are major immune suppressive cell types in colon cancer development from adenomas through early inflammation-induced emergency myelopoiesis. Cannabidiol (CBD) is reported to function in psychosis, coronavirus infection and some cancers through immune regulation. However, its target and underlying mechanisms in colorectal adenomas are unknown.

Methods

The antitumor effect of CBD was validated in two classical colorectal adenomas models including azoxymethane (AOM)/dextran sulfate sodium salt (DSS) induced mice model and high-fat fed Apc^min/+ mice model. Single-cell RNA sequencing was used to identified the immune environment change after CBD treatment in mice colorectal adenomas. Target responsive accessibility profiling was used to find the target of CBD in MDSCs. Subsequently, multiple immunology assays and molecular biology experiment were employed to explore the adenomas prevention mechanisms of CBD.

Results

Here, we found that CBD prevented the incidence of colorectal adenomas in AOM/DSS model and high-fat diet fed Apc^min/+ mice model. Our single-cell RNA sequencing data and the results of immunofluorescence revealed that CBD treatment significantly decreased the number of MDSCs in both two colon adenomas models. Mechanistically, CBD bound to the guanine nucleotide exchange factor domain of EEF1B2, inhibiting its function in translational elongation and subsequent C/EBPβ synthesis. This disruption suppressed the differentiation and generation of MDSCs, leading to enhanced T-cell activation and prevention of colorectal adenoma progression.

Conclusion

Our findings reveal EEF1B2-mediated C/EBPβ protein synthesis as a crucial pathway in MDSC generation and highlight the potential of CBD as an early intervention strategy for colorectal adenomas.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• It has been reported that cannabidiol (CBD) has immune regulation effect in some inflammatory diseases and tumors but its specific binding protein and underlying mechanism in myeloid-derived suppressor cells (MDSCs) are not well known.

WHAT THIS STUDY ADDS

• In this study, we found that CBD prevented the progression of colorectal adenomas via targeting inhibition the function of EEF1B2 to suppress the generation of MDSC from bone marrow in the condition of adenomas induced systemic inflammation. The underlying mechanism was that EEF1B2 inhibition prevented MDSC differentiation and generation through disturbing the protein synthesis of the key transcription factor C/EBPβ.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study implies that CBD may be a potential compound for clinical translation for colorectal adenomas in clinical use, which makes significant therapeutic implications in the early medical intervention for colorectal adenomas and is an effective strategy to inhibit MDSCs generation and relieve immune suppressive environment in MDSCs involved diseases.

Introduction

The incidence of colon cancer in the young population is rising at a pace of 1–2% per year.^{1 2} High intake of fat, red or processed meat is a risk factor. Immune checkpoint inhibitors (ICIs) are the first-line treatment for microsatellite instability-high or mismatch repair deficient stage 4 colorectal cancers because of the resistance. Recently, the study about the administration timing of programmed death-ligand 1 (PD-L1), an approach known as chronopharmacology,³ indicated a big difference of myeloid-derived suppressor cells (MDSCs) to PD-L1 treatment at different time point in colorectal cancer.⁴ Therefore, immune modulation at the right time is important. About 80–90% sporadic colorectal cancer arises from colorectal adenomas.^{5 6} Actually, it is about 5–10 years for the development of colorectal cancers from benign adenomatous polyps. Therefore, we have enough time to stop the progress of disease through early pharmacological intervention. However, there are few drugs for treatment of colorectal adenomas in clinic. We wondered if immune regulation at the stage of colorectal adenomas could effectively prevent the incidence of colorectal cancer. The immune-related adverse events caused by ICIs in gastrointestinal and liver is commonly seen in clinical trial, it is required to look for a new immunoregulation strategy with higher efficacy and less side effects for the early immune regulation of colorectal adenomas.

The efficacy of immunotherapy is limited because of the immune suppressive cells in tumor. MDSCs are a heterogeneous mixture of myeloid cells at different stages of differentiation but have some common features: ability to suppress adaptive immune responses, expression of both Gr-1 and CD11b in mice and activation of arginase I.^{7 8} MDSCs are further divided into two subtypes: polymorphonuclear MDSCs (PMN-MDSCs) and monocytic MDSCs (M-MDSCs). Up to date, there are no surface biomarkers that distinguish classical neutrophils from PMN-MDSCs or monocytes from M-MDSCs. Fortunately, with the help of single-cell RNA sequencing, we can separate them by different expressed gene signature.^{9 10} In tumor-associated chronic inflammation, MDSCs generated from bone narrow was induced in the stimulation of granulocyte-macrophage colony stimulating factor (GM-CSF), macrophage-stimulating factor (M-CSF) and interleukin (IL)-6 known as “emergency” granulopoiesis. MDSCs contributed to tumor progression in many cancer types. Strategies targeting MDSC include depleting MDSC, inhibition their immune suppressive function, blocking their recruitment to tumor site, promoting M-MDSC further differentiation and inhibiting their metabolism.¹¹ However, some therapies lack specificity that cannot distinguish MDSCs from their normal myeloid counterparts. Moreover, most studies focused on blocking MDSC accumulation into inflammatory site but had few reports about directly inhibiting MDSC generation or differentiation.¹² Therefore, identification of potential molecules that therapeutically inhibiting the generation and differentiation of MDSC is necessary.

The use of cannabis can be traced back to more than 4000 years.¹³ Cannabis was widely used in China, India, Persia, Middle East, Africa, Europe, and the USA. Cannabidiol (CBD) attracts more attention recently because its non-psychoactive properties. CBD binds to cannabinoid receptor 1/2 with very low affinity. CBD has been licensed by the Food and Drug Administration (FDA) and European Medicines Agency (EMA) for the treatment of childhood epilepsies. The clinical trials of CBD for the treatment of psychosis were list as the top clinical trial to watch for 2025.¹⁴ In many preclinical studies, CBD also have therapeutic activities in neurological disorders,^{15 16} psychological disorders,^{17 18} inflammatory diseases,19,22 mitochondrial diseases²³ and infection of SARS-CoV-2.²⁴ It is reported that CBD functions by binding to many membrane receptors. But its intracellular binding protein and the biological function of CBD in regulating cellular events in diseases is not well found.

Here, our data demonstrate that early CBD treatment effectively inhibits the progression of colorectal adenomas through preventing MDSCs generation and regulating its further differentiation. CBD binds to EEF1B2 and inhibits its function in protein synthesis of C/EBPβ, which interrupts the generation of MDSCs in tumor-related inflammatory condition. All these results revel the pharmacological action of CBD in remodeling immunosuppressive microenvironment and medical treatment of colorectal adenomas.

Materials and methods

Animal model, tumor challenge and treatment

C57BL/6 mice ((RRID:IMSR_JAX:000664)) and Apc^min/+mice (C57BL/6J background) were purchased from GemPharmatech (Nanjing, China).

The Apc^min/+ induced colorectal adenomas model: Apc^min/+ (C57BL/6J background) mice were fed with high-fat diet for 3 months to induced adenomas or polyps in intestinal.

Azoxymethane (AOM)/dextran sulfate sodium salt (DSS) induced inflammation-related colorectal adenomas model: C57BL/6J mice were first injected with 7.5 mg/mL AOM (Sigma, Cat#A5486) intraperitoneally. Then fed the mice with water containing 2% DSS for 5 days and replaced with normal water for 5 days. Repeated for three cycles.

MC38 syngeneic tumor model: MC38 cells (RRID: CVCL_B288) were suspended in phosphate-buffered saline (PBS) at the concentration of 1×10⁷ cells/mL and 100 µL of MC38 cells were inoculated subcutaneously into the right flank. If the tumor volume reached to 1,000 m³ and then sacrificed the mice as humane endpoints.

To evaluate the therapeutic effect of CBD (purchased from Xinping Hongshan Biotechnology, China), mice were treated with CBD (10 mg/kg) or vehicle every day from the first day of model construction. Normal C57BL/6J mice were as healthy control. To investigate the effect of MDSCs on CD4⁺ T-cell populations, mice were treated with anti-Gr-1 antibody (10 mg/kg, BioXcell) or isotype (10 mg/kg) two times a week.

Bone marrow transplantation: 6-week old female wild-type mice were first lethally irradiated (9 Gy) at two times with 4 hours interval and then intravenously injected with 5×10⁶ bone marrow cells from wild-type or heterozygous Eef1b2^+/- mice 5 hours later. After bone marrow reconstitution, the recipient mice were injected with 1×10⁶ cells into the right flank.

All animals were housed at 25°C with a humidity of 50±5% in a 12-hour light/dark cycle. All the animal studies were approved by the Science and Technology Ethics Committee of Nanjing University with IACUC-2407001.

Cells

MC38 cell line was purchased from American Type Culture Collection and cultured in Dulbecco's Modified Eagle's Medium supplemented with 10% fetal bovine serum (FBS), 100 U/mL of penicillin and 100 µg/mL of streptomycin. All cells were maintained at 37°C in 5% CO₂-humidified incubator.

Sample collection for single-cell RNA sequencing

In this study, single-cell RNA sequencing (scRNA-seq) was conducted on colon tissue samples from three groups of mice: wild-type C57BL/6 mice (Normal), high-fat diet (HFD) fed Apc^min/+ mice (vehicle), and CBD-treated HFD fed Apc^min/+ mice (CBD), with three biological replicates in each group (n=3).

Mice colon collection and processing for single-cell RNA sequencing

Freshly dissected colon tissues were initially rinsed with PBS, then transferred into GEXSCOPE Tissue Preservation Solution (Singleron Biotechnologies) to maintain cell viability, and stored on ice. Following this, samples were washed three times with Hanks’ Balanced Salt Solution and minced into small fragments. The tissue fragments were digested in Tissue Dissociation Solution (Singleron Biotechnologies) at 37°C for 15 min with gentle agitation to achieve a single-cell suspension. The cell suspension was filtered through a 40 µm cell strainer to remove debris and centrifuged at 300×g for 5 min. After discarding the supernatant, the cell pellet was resuspended in 1 mL PBS, followed by red blood cell removal using RBC lysis buffer. The cells were then centrifuged at 500×g for 5 min and resuspended in PBS. Cell viability and count were assessed using trypan blue staining, ensuring over 80% viability for each sample. For library preparation, the single-cell suspension was adjusted to a concentration of 1×10⁵ cells/mL and loaded onto GEXSCOPE microfluidics chips (Singleron Biotechnologies). Libraries were prepared according to the manufacturer’s protocol and sequenced on an Illumina NovaSeq platform with 150 bp paired-end reads.

scRNA-seq data processing and cell annotation

The scRNA-seq data from mouse colon tissue were processed using the Singleron Celescope (RRID:SCR_023553) tool for alignment, barcode assignment, and unique molecular identifier (UMI) counting, aligned to the mouse reference genome GRCm38. Filtered count matrices were then converted to sparse matrices using the Seurat package V.4.3.0 (RRID:SCR_016341), and quality control was applied by excluding cells expressing fewer than 200 genes or more than 6,000 genes, as well as cells with over 5% mitochondrial reads. The filtered data were subsequently log-normalized and scaled, adjusting for variation due to UMI counts and mitochondrial read percentages. To avoid batch effects among samples and experiments, data integration was performed using Harmony V.1.2.0 (RRID:SCR_023543), which allowed for the alignment of datasets across different conditions. Cell clustering was executed using the “FindClusters” function in Seurat with a resolution parameter of 0.4, identifying clusters based on shared gene expression profiles. Dimensionality reduction was performed with 43 principal components and visualized through the Uniform Manifold Approximation and Projection method. Finally, the “FindAllMarkers” function in Seurat was used to identify unique marker genes for each cell cluster, enabling further analysis of cell type-specific gene expression profiles across experimental conditions.

Calculation of immune-inhibitory scores for all cell clusters

Immune-inhibitory scores were assigned to all 16 cell clusters by calculating the average expression of an immune-inhibitory gene set (online supplemental file 2), based on the previous study.²⁵ Using normalized data, the mean expression of these genes within each cell population was computed to determine the immune-inhibitory score.

Identification of MDSC by scRNA-Seq analysis

MDSC populations were identified based on Kessenbrock et al’s study.¹⁰ Neutrophil and monocyte subpopulations were scored using an MDSC marker gene set (online supplemental file 3), assigning MDSC status to cells with the highest scores.

GO annotation analysis for neutrophils and monocytes

Gene Ontology (GO) annotation analysis for neutrophils and monocytes was performed using the R package clusterProfiler (V.4.13.3), applying a p value cut-off of 0.05, the Benjamini-Hochberg method for p value adjustment, and a Q value cut-off of 0.2. The results were visualized as heat maps generated with the R package pheatmap (V.1.0.12).

Single-cell preparation

For tumor tissue digestion, the tissues were cut into pieces and add 5–10 mL digestion solution (2 mg/mL collagenase IV, 40 µg/mL DNase, 0.5 mg/mL dispase II) in Dulbecco's Phosphate-Buffered Saline (DPBS) for 1 hour at 37°C in a rotatory shaker (at 80 rpm). After digestion, the enzymes were neutralized with DMEM plus 20% FBS. Then the cell suspensions were filtered using a 40 µm cell strainer to get the single-cell suspension.

For bone marrow preparation, bone marrow was harvested from tibia and femur. Red cells were removed by Tris-NH₄Cl lysis buffer. Then the cell suspensions were filtered using a 40 µm cell strainer to get the single-cell suspension.

For mouse spleen cell preparation, the spleen was digestion by the gentleMACS Dissociator (Miltenyi Biotec). Red cells were removed by Tris-NH₄Cl lysis buffer. Then cell suspensions were filtered using a 40 µm cell strainer to get the single-cell suspension.

Generation of bone marrow-derived myeloid-derived suppressive cells

Bone marrow was harvested from tibia and femur. Red cells were removed by Tris-NH₄Cl lysis buffer. Cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium with 10% FBS, 1X penicillin/streptomycin and 55 µM 2-mercaptoethanol at concentration of 6×10⁵ cells/mL. To induce differentiation of cells to MDSCs, GM-CSF (50 ng/mL, PeproTech, Cat# 96-315-03-20) and IL-6 (50 ng/mL, PeproTech, Cat# 96-216-16-10) were added into the medium and cultured for 4 days. Increasing concentrations of CBD or matching dimethylsulfoxide (DMSO) volume were added at the time of seeding.

T-cell suppression assay

MDSCs with different treatment were collected and resuspend in 1640 medium at the concentration of 1×10⁶ cells/mL. Spleen pan T cells were separated using pan T Cell Isolation Kit (Miltenyi Biotec Cat#130-095-130), labeled with Carboxyfluorescein Diacetate Succinimidyl Ester (CFSE, 2 µM, BioLegend, Cat#423801) and resuspend at the concentration of 2×10⁶ cells/mL. After that, pan T cells were cultured alone or co-cultured with MDSCs with different treatment at the ratio of 1:1, 2:1, 4:1 (T cells:MDSCs) in the presence of pre-coated anti-CD3 (5 µg/mL, Invitrogen, RRID:AB_468847) and soluble anti-CD28 (5 µg/mL, Invitrogen, RRID:AB_468921) in a 96-well plate for 4 days. After incubation, the proliferation of T cells was detected by flow cytometry.

RNA isolation and quantitative real-time PCR

Total RNA was extracted by TRIzol (TaKaRa). 1,000 ng of RNA was conducted reverse transcription to get complementary DNA using the HiScript III RT SuperMix for qPCR kit (Vazyme, Cat#R323). Then, performed real-time PCR detection with ChamQ Universal SYBR qPCR Master Mix (Vazyme, Cat#Q711) on ABI PRISM 7000 Sequence Detection System with specific primers listed in online supplemental table S2. The quantification of messenger RNA expression was determined with 2^−ΔΔCt method. All the target gene expressions were normalized to GAPDH or ACTIN.

Immunofluorescence staining

The tissues were fixed in 4% paraformaldehyde and embedded in paraffin. For immunohistochemistry staining, after washing three times with Phosphate Buffered Saline with Tween (PBST), the tissue sections were incubated with horseradish peroxidase -conjugated secondary antibodies for 1 hour at room temperature. After washing three times with PBST, diaminobenzidine staining was performed to show the indicated protein. Finally, the nuclei were stained with hematoxylin. The images were acquired using Olympus vs200.

Western blot

Cells were collected and lysed in cell lysis buffer (Beyotime, Cat#P0013). The cell lysates were quantified by BCA (bicinchoninic acid) assay. Then equal amounts of protein were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene difluoride membrane. The membranes were blocked with 5% skimmed milk (w/v) for 1 hour at room temperature and incubated with primary antibodies overnight at 4°C. After three time washing with PBST, the bands were incubated with the according HRP-conjugated secondary antibodies for 1 hour at room temperature. After three times of washing, the protein bands were immunoblotted using an enhanced luminescence kit.

Flow cytometry and cell sorting

Prepared single-cell suspensions were blocked with Fc receptor blocker and first stained with fixable viability dye and followed by the appropriate antibody cocktail at 4°C for 30 min. Antibodies used were listed in online supplemental table S1. Cells were washed with fluorescence ctivated cell sorter (FACS) buffer (PBS with 2% bovine serum albumin) and resuspended in FACS buffer. Cells were acquired using an Attune NxT (Invitrogen) for analysis and using SORP FACSARIA II (BD, RRID:SCR_009751) for FACS sorting.

Plasmid construction and transfection

The coding sequences of human EEF1B2 were constructed including wildtype, guanine nucleotide exchange factor (GEF) domain Ser6 point mutation (S6A), GEF domain Aps62 point mutation (D62V) and were amplified and cloned into pcdna3.1-EGFP vectors. The corresponding plasmids were transinfected into HEK293T cells (RRID:CVCL_0063) for 48 hours using the universal transfection reagent (Yeasen, Cat#40 808ES03).

Measurement of translation

To measure the incorporation of puromycin into the actively translated proteins, cells were treated with CBD for 24 hours or cycloheximide (CHX 100 ng/mL, Sigma, Cat# C7698) for indicated time and then puromycin (25 µM, Beyotime, Cat#ST551) was added into cells and cultured for 1 hour. Finally, harvest cells to get cell lysates for western blot analyses.

Target responsive accessibility profiling

The screening of CBD binding proteins was performed as described previously.²⁶ Briefly, three dishes of bone marrow cells were cultured in RPMI 1640 medium with GM-CSF and IL-6. At day 2 after seeding, we treated the attached myeloid cell with DMSO or CBD (5 or 10 µM) for 1 hour. Then, collected cells and washed for two times. We then lysed the cells by M-PER buffer (Thermo Scientific, Cat#78505) and used deuterium-coded formaldehyde and a borane-pyridine complex to label the proteinaceous lysine residues at room temperature. Subsequently, the target responsive accessibility profiling (TRAP)-labeled proteome was digested and further labeled by tandem mass tag reagents, following the Liquid Chromatograph Mass Spectrometer -based quantitative proteomics analysis. We assigned proteins carrying lysine residues with significant accessibility changes as target candidates of CBD. The ratio of the abundance of each TRAP-labeled peptide indicates the extent of accessibility change, and is intimately associated with ligand-binding affinity. Two-tailed Student’s t-test was carried out to assess whether the detected accessibility changes of labeled peptides are statistically significant. An intergroup p value (p<0.001) and TRAP ratio >2 or <0.5 was set as the cut-off to screen the target responsive peptides belonging to the CBD binding proteins from the whole quantified proteome.

Cellular thermal shift assays

To detect the binding of CBD to EEF1B2, we conducted cellular thermal shift assays as reported with modifications. HEK293T cells were transfected with pcDNA3.1-EEF1B2-EGFP plasmids for 48 hours then, the cells were collected and suspended in 800 µL of PBS with phenylmethanesulfonyl fluoride for ultrasonication and centrifuged at 12,000 g for 10 min to get the cell lysates. Subsequently, the cell lysates were divided equally into seven PCR tubes. Then, the cell lysates were treated with CBD for 30 min on ice. Then heated at various temperatures (60, 66, 72, 78, 84, 90, 96°C) for 5 min. Next, the lysates were centrifuged at 20,000 g for 20 min to get the soluble protein and boiled for 10 min in loading buffer and finally subjected to western blot analysis using EGFP antibody (RRID:AB_10709851).

Microscale thermophoresis assay

Microscale thermophoresis (MST) experiments were performed using Monolith NT.115 (Nano Temper) to detect the interaction between CBD and EEF1B2. Briefly, HEK293T cells were transfected with vector-EGFP plasmid or EEF1B2-WT-EGFP (S6A-EGFP, D62V-EGFP) plasmids for 48 hours. Then, collected cells for ultrasonication and centrifuged at 12,000 g for 10 min to get the cell lysis. We adjusted the fluorescence of vector plasmid and WT/ S6A /D62V plasmids to an equal intensity at 400–1,000. And then mixed the cell lysates with CBD at different concentration at a volume ratio of 1:1. Finally, using Monolith NT.115 for binding detection.

EEF1B2 protein expression and purification

Full length of homo EEF1B2 with His tag at its N termination were overexpressed by using Escherichia coli BL21 cells with induction of 0.5 mM isopropyl β‐D‐1‐thiogalactopyranoside at 16°C overnight. Cell pellets were suspended at 20°C in lysis buffer (50 mM Tris‐HCl, pH 7.4, containing 0.5 M NaCl, 40 mM imidazole, and 1 mM PMSF), frozen at −80°C overnight, thawed at 37°C, and broken up cells at 800 Pa for 5 min with high pressure cell cracker. Then the whole cell lysis was centrifugated (4°C, 10,000 g for 30 min). Then the supernatant was applied to an immobilized metal ion affinity chromatography column (5 mL) pre-equilibrated with Buffer‐1 and incubated for 1 hour at 4°C. Bound proteins were eluted after washing the columns with buffer‐1 with imidazole gradient (10 mM, 20 mM, 40 mM and 60 mM). The purity was confirmed by using SDS-PAGE.

Isothermal titration calorimetry assay

The binding affinity between CBD and EEF1B2 was examined using MicroCal isothermal titration calorimetry (ITC) 200. 50 µM of EEF1B2 protein was prepared in the buffer (50 mM Tris-HCl, 0.5 M NaCl, 1 mM PMSF) supplementary with the same DMSO as CBD. CBD was diluted with the same buffer with EEF1B2 protein to 500 µM. Titration were performed at 25°C using an initial injection of 0.4 µL followed by 20 injections of 2 µL with a 120 s interval. Finally, the data were analyzed by the Origin software to determine binding parameters.

Molecular docking of cannabidiol with EEF1B2

The structure model of the guanine nucleotide exchange factor domain of human EEF1B2 was obtained from the Protein Data Bank (PDB ID: 1B64, RRID:SCR_012820) and the structure of CBD was obtained from PubChem (compound CID: 644019, RRID:SCR_004284). The docking process was performed in AutoDock 2 software (RRID:SCR_012746), where the default settings were used.

Intrabone marrow injection of AAV-EEF1B2

Recombinant adeno-associated serotype eight virus (pcAAV2-CMV-3×flag-P2A-mCherry-WPRE) was generated by OBIO Technology. Intrabone marrow injection of AAV-EEF1B2 or AAV-mCherry were performed following general anesthesia. Viral vectors for EEF1B2 overexpression or control viral vectors were prepared at a dose of total 5×10¹¹ viral genomes/mice and pushed into the tibia and femur bone marrow cavity. The efficacy of EEF1B2 overexpression through AAV-EEF1B2 injection was measured by western blot and FACS.

Statistical analysis

Statistical analysis was performed using the GraphPad Prism V.8 or R statistical software V.4.3.3 (RRID:SCR_001905). Data from multiple experiments are presented as the means±SEM. Independent sample t-test was used for comparisons between two groups and one-way analysis of variance with Tukey’s multiple comparisons test was used for comparisons among three or more groups. Statistical tests were two-tailed and p<0.05 was considered statistically significant.

Data availability

The data that support the findings of this study have been deposited into Gene Expression Omnibus with accession number GSE277268. And other data included in this study are available on reasonable request by contact with from the corresponding author on request.

Results

CBD inhibits colorectal adenomas generation in both AOM/DSS and Apc^min/+ mouse model

We first established the intestinal benign adenomas using Apc^min/+ mice fed with HFD. The mice were simultaneously treated with CBD (figure 1A). We observed that CBD treatment significantly reduced adenoma formation (figure 1B and C). The body weights of mice were higher in the CBD-treated group compared with the vehicle group (figure 1D). Spleens from mice of vehicle group exhibited enlarged size and increased weight because of adenomas induced the systemic inflammation, whereas CBD mitigated the inflammation (figure 1E and F). The adiposity in mouse liver (figure 1G), serum T-CHO (total cholesterol) and triglyceride in CBD-treated mice decreased (figure 1H). Moreover, H&E staining indicated the smaller colorectal adenomas in mice treated with CBD (figure 1I). β-catenin and Ki67 also decreased after CBD treatment, indicating decreased cell malignant transformation and proliferation (online supplemental figure S1A and B). All the results demonstrated that CBD mitigated the phenotypes of systemic inflammation, disorder of lipid metabolism and intestinal adenomas in HFD fed Apc^min/+ mouse colorectal adenomas. We next constructed AOM/DSS induced inflammation- associated colorectal adenomas model (figure 1J). Consistent with the results in the HFD fed Apc^min/+ mice model, no obvious colon adenoma was observed in CBD-treated AOM/DSS mice (figure 1K–M). Because of the low disease burden, the body weight of CBD-treated mice increased at the late stage (figure 1N). Consistently, both the shortened colon length and increased intestinal permeability were rescued by CBD (figure 1O and P). H&E staining showed that CBD-treated mice developed reduced proportions of adenoma and low-grade dysplasia (figure 1Q), concomitant with significantly reduced Ki67 positive cells (online supplemental figure S1C). Colon inflammation was also alleviated as shown by decreased ROS level (online supplemental figure S1D and E). The consistent findings in the two colorectal adenoma mice model suggested that CBD was an effective compound in inhibiting the development of colorectal adenomas.

Figure 1. CBD ameliorates colorectal adenomas in mice models. (A) Experimental timeline and design (n=6/group). Apc^min/+ mice were fed with high-fat diet to induce colorectal adenomas with vehicle or CBD treatment for 12 weeks. (B) Representative pictures of mice colon form indicated groups and the red arrow indicated the adenomas. (C) The number of adenomas in mice colon. (D) Body weight of mice. (E) Representative pictures of spleen size from indicated groups. (F) Spleen weight of mice. (G) Representative pictures of mice livers. (H) T-CHO and TG levels in mice serum detected by ELISA. (I) Representative images of H&E staining of mice colon. Scale bar, 100 µm. (J) Experimental timeline and design. WT C57BL/6 mice were first injected with AOM and then give 2.5% DSS water for 5 days, replaced with fresh water for another 5 days and repeated three cycles to induce inflammation associated colorectal adenomas. The mice with colorectal adenomas treated with vehicle or CBD every day. WT C57BL/6 mice with no treatment were the normal control. Normal group (n=5), vehicle and CBD group (n=7/group). (K to M) The representative pictures of mice colon (K) and adenomas number (L) and size (M) in mice colon from different groups. (N) Body weight change of vehicle and CBD group. (O–P) Mice colon length (O) and intestinal permeability (P) of three groups. (Q) Representative images of H&E staining of colon in AOM/DSS model. Scale bar, 200 µm. *p<0.05, **p<0.01, ***p<0.001. P values were calculated by Student’s t-test or one-way ANOVA with Tukey’s multiple comparisons test. ANOVA, analysis of variance; AOM, azoxymethane; CBD, cannabidiol; DSS, dextran sulfate sodium salt; i.p., intraperitoneal injections; T-CHO, total cholesterol; TG, triglyceride.

scRNA-seq reveals CBD reduces the MDSCs in Apc^min/+ mice colon

APC gene mutation and the HFD lifestyle are essential risk factors of colorectal cancers, we thus collected the colon from Apc^min/+ mice fed with HFD to perform scRNA-seq (figure 2A). We first clustered all cells into 4 main types and 16 subtypes (figure 2B, online supplemental figure S2A-D). The ratio of myeloid cells increased about 10-folds to 23.55% in the colon adenomas mice (vehicle group) compared with 2.78% in normal healthy mice, but it decreased to 11.84% after CBD treatment (figure 2C). In particular, among myeloid cells, the percentage of neutrophils decreased after CBD treatment (figure 2D). Different expressed gene in neutrophils were showed after CBD treatment. The results were that CBD treatment decreased many feature genes of MDSCs (figure 2E). Consistently, neutrophils and monocytes exhibited the highest immune-inhibitory score (figure 2F), which led us to believe that neutrophils and monocytes in colorectal adenomas were MDSCs. The significantly enriched GO terms included myeloid leukocyte activation, myeloid cell differentiation, regulation of T-cell activation, regulation of inflammatory response and regulation of innate immune response (figure 2G). Then, we identified that N_C0 subcluster of neutrophils had the highest score of PMN-MDSC and M_C2 subcluster of monocytes had the highest score of M-MDSC (figure 2H–J, online supplemental figure S2E and F). CBD treatment decreased the number of the both two types of MDSCs (figure 2J). Then, because the MDSCs were immunosuppressive cells, we wondered the interaction between MDSCs and T cells. The cell–cell interactions between MDSCs and different types of T cells showed that CBD reduced the total interaction between MDSCs and T cells (online supplemental figure S2G), especially Th2 T cells and CD8 T cells (online supplemental figure S2H and I). Together, the results of scRNA-seq indicated that CBD decreased MDSCs in intestinal adenomas and it may be the reason for its function on alleviating colon adenomas progression.

Figure 2. CBD inhibits MDSCs in colon adenomas identified by scRNA-seq. (A) Apc^min/+ mice were fed with high fat diet to induce colon adenomas with vehicle or CBD treatment for 12 weeks. WT C57BL/6 mice were normal control. Then collected the mice colon from each group to perform scRNA-Seq (n=3/group). (B) UMAP plot of the major four types of cell cluster by scRNA-seq analysis. (C) The cell ratio of epithelial cells and myeloid cells with the largest changes among four major cells. (D) Composition of cell types in myeloid cells. (E) The different expressed genes (DEGs) in neutrophils regulated by CBD. Red dot represented the significantly upregulated genes in both vehicle group versus normal group and vehicle group versus CBD group. Blue dot represented the downregulated genes in both vehicle group versus normal group and vehicle group versus CBD group. (F) The immune inhibitory score was calculated in all 16 cell types by average expression of an immune-inhibitory gene set. (G) GO signaling pathway of different expressed genes in neutrophils and monocytes between three groups. (H) UMAP plot of the subclusters of neutrophils and monocytes. (I) Violin plots showing relative MDSC score ordered by the expression of MDSC gene signature in neutrophils and monocytes. (J) The proportion of PMN-MDSC and M-MDSC in three groups. CBD, cannabidiol; GO, Gene Ontology; MDSC, myeloid-derived suppressor cell; M-MDSC, monocytic MDSC; PMN-MDSC, polymorphonuclear MDSC; scRNA-seq, single-cell RNA sequencing; UMAP, Uniform Manifold Approximation and Projection.

CBD inhibits the generation and further differentiation of MDSCs from bone marrow in vitro

Next, we confirmed that CBD actually decreased the infiltration of total MDSCs (CD11b⁺Gr-1⁺) to colon site in both colorectal adenoma models (figure 3A, online supplemental figure S3A). Correspondingly, CD3⁺ T cells and interferon (IFN)-γ expression increased in mice colon after CBD treatment (figure 3B–D, online supplemental figure S3B and C). Interestingly, we found that CBD treatment significantly promoted the infiltration of T cells across stromal and tumor nest regions to the tumor central area of colorectal adenomas (online supplemental figure S3D). The mechanisms of MDSCs recruited to the tumor or inflammation site are widely reported,27,29 but how the differentiation of MDSCs is affected and how to regulate the generation of MDSCs from their original bone narrow are not clear. Hence, we next investigated the impact of CBD on the MDSCs generation from bone marrow. We excluded the cytotoxicity of CBD on colon epithelial cells and bone marrow cells (online supplemental figure S4). Then, we induced differentiation of bone marrow cells into MDSCs in the presence of GM-CSF and IL-6 for 4 days.³⁰ The immunosuppressive activity of MDSCs were identified by the T-cell suppression assay (online supplemental figure S5A and B). Functional genes of MDSCs increased with culture time (online supplemental figure S5C). CBD significantly reduced the percentage of M-MDSCs in CD11b⁺ myeloid cells along with increased percentage of Ly-6C^low/− Ly-6G⁺ cells in CD11b⁺ cells (figure 3E,F). The expression of key functional genes of MDSCs showed a dose-dependent decrease after a gradient of CBD treatment (figure 3G). By co-culture of T cells and MDSCs with or without CBD treatment, we observed that CBD significantly reduced MDSC suppressive activity indicated by the increased T-cell proliferation (figure 3H). But CBD at the dose of 5–10 µM did not directly affect T-cell proliferation (online supplemental figure S5D). Because M-MDSCs can further differentiate into macrophages and dendritic cells (DCs) when recruited into tissue, there is a strategy to eliminate MDSCs is to induce MDSCs further differentiation to M1-like macrophages or DCs.^{31 32} In our study, we noticed that when we extended culture time to 7 days, most adherent cells were non MDSCs (Ly6C⁻Ly6G⁻) (figure 3I, online supplemental figure S6A). So, we next detected the exact cell types in non-MDSCs. The results showed that CBD treatment significantly reduced the number of macrophages, and promoted the transformation of MDSCs to M1 macrophage and DCs from day 4 to day 7. But in non-CBD treated group, MDSCs differentiated into more macrophages and most of which were M2-like macrophages and less DCs (figure 3H, online supplemental figure S6B-D). Consistent with the results in vitro, CBD treatment decreased the total number of macrophages in mice colon adenomas (online supplemental figure S6E, F). Thus, these data suggested that CBD limited the generation and immunosuppressive activity of MDSCs, and promoted MDSCs further differentiation to M1-like macrophages and DCs.

Figure 3. CBD prevents MDSC generation from bone marrow cells in vitro. (A) The total Gr-1⁺ MDSCs (CD11b⁺Gr-1⁺, white arrow) in mice colon in different groups (n=5/group). Scale bar, 50 µm. (B) The CD3⁺ T cells in mice colon in different groups (n=5/group). Scale bar, 50 µm. (C) Expression of IFN-γ in mice colon in different groups (n=5/group). Scale bar, 100 µm. (D) The quantification of representative fluorescent images of Gr-1⁺ MDSCs, CD3⁺ T cells and IFN-γ from (A), (B) and (C), respectively. (E) Bone marrow cells were treated with DMSO or different dose of CBD (2.5–10 µM) supplemented with IL-6 (50 ng/mL) and GM-CSF (50 ng/mL) for 4 days to induce MDSCs. The representative FACS dot plot showed the effect of CBD on surface phenotype of MDSCs (n=3). (F) The statistical analysis of the proportion of Ly-6C^hiLy-6G⁻ (M-MDSCs) and Ly-6C^low/−Ly-6G⁺ (PMN-MDSCs) in figure 3E. (G) MDSC feature gene expression after CBD treatment detected by qPCR (n=3). (H) Suppressive activity of MDSCs against T-cell proliferation. Representative histograms of CFSE dilution (top) and the percentage of T-cell proliferation (bottom) were shown (n=3). (I) Bone marrow cells were treated with DMSO or CBD (10 µM) supplemented with IL-6 (50 ng/mL) and GM-CSF (50 ng/mL) and collected adherent cells for FACS analysis to determine the cell types at indicated day (n=3). *p<0.05, **p<0.01, ***p<0.001. P values were calculated by one-way ANOVA with Tukey’s multiple comparisons test. ANOVA, analysis of variance; CBD, cannabidiol; DMSO,dimethyl sulfoxide; GM-CSF, granulocyte-macrophage colony-stimulating factor ; FACS, fluorescence-activated cell sorting; IFN, interferon; IL, interleukin; MDSC, myeloid-derived suppressor cell; M-MDSC, monocytic MDSC; PMN-MDSC, polymorphonuclear MDSC; qPCR, quantitative PCR.

CBD inhibits the chronic inflammation induced MDSC generation and function in mice

To detect the impact of CBD on MDSCs in tumor induced systemic inflammation in vivo, we therefore used the syngeneic tumor model to induce MDSCs in mice (figure 4A). CBD inhibited tumor growth without affect the body weight of mice and reduced splenomegaly caused by systemic inflammation (figure 4B–C, online supplemental figure S7A-D). In the early stage (day 6) after tumor cell injection, both of M-MDSCs and PMN-MDSCs increased in peripheral blood (PB). Then, in the middle stage, M-MDSCs in PB decreased significantly and the percentage of M-MDSCs continuously reduced at day 18 after CBD treatment (figure 4D, upper, online supplemental figure S7E). But the percentage of Ly-6C^low/− Ly-6G⁺ cells in PB was not changed by CBD (figure 4D, lower). CBD inhibited the generation of M-MDSC in bone marrow and reduced the M-MDSCs in spleen and tumor (figure 4E–G). The percentage of Ly-6C^low/− Ly-6G⁺ cells from bone marrow and its recruitment into spleen and tumor were slightly decreased by CBD (figure 4E–G). The separated MDSCs from spleen and tumor in mice with CBD treatment had little immunosuppressive activity than that of vehicle group (figure 4H–J). T-cell activation in the tumor environment was also enhanced as indicated by increased CD8⁺ T-cell number and capability of secreting IFN-γ, perforin and granzyme B in CBD treated mice (figure 4K, online supplemental figure S8). We also observed that CD4⁺ T cells decreased after CBD treatment, so we evaluated which CD4⁺ T-cell populations were impacted. The results showed that CBD treatment decreased the percentage of Regulatory T cells and Th2 T cells in total CD4⁺ T cells and increased Th1, Tcm and Tem in total CD4⁺ T cells. Moreover, anti-Gr-1 antibody had the similar effect with CBD treatment on different CD4⁺ T-cell populations, so these results indicated the change of CD4⁺ T cells can be a direct effect of MDSCs change (online supplemental figure S9). Overall, all these results confirmed the similar effect of CBD in vivo with that in vitro on inhibiting generation of MDSCs and immunosuppressive function in a tumor induced chronic systemic inflammation.

Figure 4. CBD inhibits MDSCs generation and reshapes immune environment in mice. (A) The experimental design and timeline. MC38 cells were inoculated subcutaneously into WT C57BL/6 mice (n=6/group). After 6 days, the mice were randomly divided into the vehicle group and CBD group. Normal no treated mice were used as normal control. Then collected PB at day 6, day 13 and day18 for FACS analysis. Tumor, spleen and bone marrow were collected at the termination. (B) MC38 tumor volume in two groups. (C) MC38 tumor weight in two groups. (D–G) Proportion of Ly-6C^hiLy-6G⁻ (M-MDSCs) and Ly-6C^low/−Ly-6G⁺ (PMN-MDSCs) in PB (D), bone marrow (E), spleen (F) and tumor (G). (H) MDSCs separation from mice spleen and tumor by FACS. (I–J) Then MDSCs from spleen (I) or tumor (J) were co-culture with spleen T cells to determine the immune suppressive activity of MDSCs against T-cell proliferation. Representative histograms of CFSE dilution in different groups (left) and the percentage of T-cell proliferation (right) were shown (n=3). (K) The percentage of CD45⁺ cells, total CD3⁺ T cells, CD8⁺ T cells and CD4⁺ T cells in tumor from different groups. And the T-cell function in different groups were shown by the expression of cytotoxic cytokines detected by FACS. *p<0.05, **p<0.01, ***p<0.001. P values were calculated by Student’s t-test or one-way ANOVA with Tukey’s multiple comparisons test. ANOVA, analysis of variance; BM, bone marrow; CBD, cannabidiol; CFSE, Carboxyfluorescein succinimidyl ester; FACS, fluorescence-activated cell sorting; IFN, interferon; i.p., intraperitoneal injections; MDSC, myeloid-derived suppressor cell; M-MDSC, monocytic MDSC; PB, peripheral blood; PBS, phosphate-buffered saline; PMN-MDSC, polymorphonuclear MDSC.

Above we have confirmed the direct effect of CBD on MDSCs of colorectal adenomas, to explore the indirect effect of CBD on MDSCs from influencing colon tumor cells, we then used conditional medium (CM) from three types of colon epithelial cells to induce MDSCs generation (online supplemental figure S10A). The adherent bone marrow cell increased by culturing with CM from MC38 and CT26 and FACS results confirmed MDSCs induced only by incubated with CM from two colon cancer cells (online supplemental figure S10B and C). Functional genes of MDSCs also increased significantly after culturing with CM from two colon cancer cell lines (online supplemental figure S10D). Then, we treated MC28 or CT26 cells with CBD for 24 hours, then refreshed the culture medium for another 24 hours and collected CM to culture BM (online supplemental figure S10E). The FACS results showed that CBD pretreatment inhibited M-MDSCs generation in a dose-dependent manner but have no effect on PMN-MDSCs at the dose of 5 µM and 10 µM (online supplemental figure S10F). The results show that CM from CBD treated epithelial colon cancer cells caused less MDSCs generation and led to decreased key function gene expression of MDSCs such as Arg1 and Nos2 and Ptgs2. Besides, the gene expression of Cd300ld, involved in MDSCs recruitment was also decreased in MDSCs (online supplemental figure S10G). Therefore, all the results indicated that CBD also directly diminished the role of cancer cells in promoting MDSC generation and recruitment.

EEF1B2 is the direct intracellular target of CBD in MDSCs detected by TRAP

Here we used the TRAP approach to identify the target of CBD in MDSCs.³³ Bone marrow cells were cultured in the presence of GM-CSF and IL-6 for 2 days to let cells differentiation towards MDSCs and then CBD or DMSO was added into MDSCs and cultured for 1 hour (figure 5A). Since the protein binding with CBD had low labeling efficiency of lysine in MS spectrum, we identified EEF1B2 with downregulated lysine-labeled ratio after CBD treatment in a dose dependent manner as the binding protein of CBD (figure 5B). Furthermore, the cellular thermal shift assay supported the binding between CBD and EEF1B2 because the EEF1B2 protein becoming more stability after treated with CBD (figure 5C). Besides, ITC and MST assay also confirmed the binding between CBD and EEF1B2 and the affinity was 52.1 µM and 4.79 µM, respectively (figure 5D and E). Subsequently, we noticed that the binding peptide was part of the GEF domain of EEF1B2 protein (figure 5F). To further study the binding model of CBD to EEF1B2, a protein structure-based docking calculation was used with the crystal structure of GEF domain in PDB library (PDB ID: 1B64). CBD formed critical hydrogen bonding interactions with ASP62 and SER6 in GEF domain of EEF1B2, forming hydrophobic interactions with surrounding residues (figure 5G). To verify the two sites, we constructed point mutant plasmids of GEF domain of EEF1B2 (D62V and S6A). The results of MST assay showed that D62V mutation slightly decreased binding affinity and the S6A mutation completely disrupted the binding (figure 5H and I). Consistently with this, the MST results of D62V and S6A showed that the protein stability had no change in CBD treated cells compared with DMSO (online supplemental figure S11). The previous reported CBD binding protein such as CB1, CB2, GPR55, TRPV1, TRPA1, KCNQ2 and Cav3.2 had very low gene expression in MDSCs (online supplemental figure S12), so these proteins were not found in our TRAP results. Collectively, these results demonstrated the directly binding sites between CBD and the GEF domain of EEF1B2 in MDSCs.

Figure 5. Identification of EEF1B2 as the target of CBD by TRAP. (A) The workflow of TRAP experiment. Bone marrow cells were cultured in the presence of GM-CSF and IL-6 for 2 days to let cells differentiation towards MDSCs and then two doses of CBD or DMSO were added into MDSCs and cultured for 1 hour. Then collected cell for protein extraction and digestion for MS analysis. (B) Volcano plot showing the different lysine labeled peptide protein between CBD group (5 µM, 10 µM) and DMSO control, FC, fold change (ratio of CBD to DMSO). (C) CBD promoted the resistance of EEF1B2 to different temperature gradients by CETSA in HEK293T cells. (D) ITC assay to confirm the interaction between EEF1B2 and CBD. (E) MST assay to confirm the interaction between EEF1B2 and CBD. (F–G) Docking prediction of the binding sites of the GEF domain (PDB ID: 1B64) of EEF1B2 to CBD. (H–I) The two predicted binding sites of EEF1B2 to CBD were mutant and then the MST assay indicated the disrupted binding between EEF1B2-D62V/S6A and CBD. CAR, the central acidic region; CBD, cannabidiol; DMSO, dimethyl sulfoxide; FC, fold change; GEF, guanine nucleotide exchange factor; GM-CSF, granulocyte-macrophage colony-stimulating factor; IL, interleukin; ITC, isothermal titration calorimetry; MDSC, myeloid-derived suppressor cell; MST, micro scale thermophoresis; TRAP, target responsive accessibility profiling.

CBD inhibit EEF1B2 mediated C/EBPβ protein translation and synthesis for the differentiation and function of MDSCs

Pseudotime trajectory analysis showed the distribution of cell state was similar in normal and CBD treated groups (online supplemental figure S13A and B). Eef1b2 expression was high in the immature MDSCs states and decreased when in the progress to mature state (online supplemental figure S13C and D). The expression of Eef1b2 was higher in the immature cells with higher MDSCs score (online supplemental figure S13E-G). The gene and protein expression of EEF1B2 gradually increased when inducing MDSCs differentiated from bone marrow (figure 6A and B). Moreover, the fold change of EEF1B2 during MDSCs differentiation is higher than other eukaryotic elongation factors (online supplemental figure S14A). It is reported that EEF1B2 interacts with EEF1A1 with its GEF domain to function in peptide elongation and protein synthesis (figure 6C). CBD bound to GEF domain of EEF1B2 (figure 5), which disrupted its binding to EEF1A1 (figure 6D) and inhibited its function in protein synthesis but not influencing EEF1B2 protein expression (figure 6E and F and online supplemental figure S14B). The FACS results show that CBD inhibited MDSCs generation as the protein transcription inhibitor CHX (figure 6G). Moreover, EEF1B2 knockdown decreased the generation of both type of MDSCs from bone marrow cells and also downregulated the expression of functional gene of MDSCs (figure 6H–J, online supplemental figure S14C and D). Total EEF1B2 expression increased in mice colorectal adenomas and human colorectal cancers (online supplemental figure S14E and F) and the data from database showed that it had positive correlation with MDSCs infiltration level, had negative correlation with both CD4⁺ T cells and CD8⁺ T cells infiltration in colon cancer (online supplemental figure S14G-I).

Figure 6. CBD inhibits CEBP/β protein synthesis mediated by EEF1B2 in MDSCs generation. (A) Eef1b2 gene expression increased in bone marrow-derived MDSCs at different culture time detected by qPCR (n=3). (B) EEF1B2 protein expression increased in bone marrow-derived MDSCs at different culture time. (C) The work model of EEF1B2 interaction with EEF1A1 to modulated the GDP to GTP transition of EEF1A1 in translation process. (D) CBD disrupted the interaction between EEF1B2 and EEF1A1 detected by IP. (E) CBD inhibited translation in MDSCs generation. CHX was used as positive control to culture MDSCs for 4 or 8 hours to inhibit translation. (F) EEF1B2 overexpression promoted translation in MDSCs. (G) FACS results showing that CBD inhibited MDSCs generation through disrupting intracellular protein translation (n=3). CHX was used as a positive control to inhibit translation. (H) Representative FACS results showing that EEF1B2 knockdown prevented the generation of MDSCs from bone marrow cells (n=5). (I) The statistical analysis of the percentage of two MDSC subtypes in (L). (J) Functional gene expression decreased in MDSCs after EEF1B2 knockdown. (K) C/EBPβ protein expression increased in bone marrow-derived MDSCs at different culture time. (L) CBD inhibited C/EBPβ protein expression during MDSCs generation. (M) EEFB12 knockdown inhibited C/EBPβ protein expression. (N) EEF1B2 overexpression increased C/EBPβ protein expression. (O) EEF1B2 overexpression relieved the inhibitory effect of CBD on C/EBPβ protein expression during MDSCs generation. *p<0.01,**p<0.01, ***p<0.001. P values were calculated by Student’s t-test or one-way ANOVA with Tukey’s multiple comparisons test. ANOVA, analysis of variance; CBD, cannabidiol; CHX, cycloheximide; FACS, fluorescence-activated cell sorting; IP, immunoprecipitation; MDSC, myeloid-derived suppressor cell; ns, not significant; qPCR, quantitative PCR.

It is reported that C/EBP family are key transcription factors in MDSC generation.³⁴ Especially, C/EBPβ controls emergency granulopoiesis induced by inflammatory cytokines³⁵ and promotes the generation and function of MDSCs.³⁰ Our results showed that the CBD inhibited Cebpb gene in HFD fed Apc^min/+ mice (figure 2E). The expression of C/EBPβ was higher in MDSCs than its family member C/EBPα (online supplemental figure S15A). We found that the key transcription factor, C/EBPβ, increased during MDSCs differentiation from bone marrow cells (figure 6K). Interestingly, we observed that CBD inhibited C/EBPβ protein expression in MDSCs in a dose-dependent manner (figure 6L). We wondered whether the C/EBPβ protein expression was affected by CBD through EEF1B2. As our expected, EEF1B2 overexpression or knockdown really regulated protein expression of C/EBPβ (figure 6M and N). Importantly, EEF1B2 overexpression could rescue the inhibitory effect of CBD on C/EBPβ expression (figure 6O). Taken together, all these data suggested that CBD inhibited C/EBPβ protein transcription through targeting EEF1B2 and thus inhibited MDSCs generation. C/EBPβ expression was higher in mice colorectal adenomas tissue and human colorectal cancer tissues compared with the normal colon tissue (online supplemental figure S15B and C) and positively associated with high colon cancer stage (online supplemental figure S15D and E). Moreover, human C/EBPβ gene expression increased in two databases (online supplemental figure S15F and G). High C/EBPβ expression indicated a low survival time (online supplemental figure S15H). These results suggested targeting EEF1B2 may be a promising strategy to prevent the occurrence of colon cancer through inhibiting C/EBPβ-mediated MDSCs generation.

The effect of EEF1B2 overexpression or knockdown in MDSCs mediated tumor progression

To verify the direct role of EEF1B2 in the therapeutic effect of CBD in colorectal adenomas, we used Apc^min/+ mice to construct HFD diet induced colorectal adenomas model. As CBD targeted EEF1B2 to inhibit the generation of MDSCs and thus inhibited colorectal adenomas, we next overexpressed EEF1B2 through intrabone marrow injection of AAV-EEF1B2 in CBD-treated mice (figure 7A, online supplemental figure S16A and B). The results showed that EEF1B2 overexpression in bone marrow reversed the protective effect of CBD on the incidence of colorectal adenomas (figure 7B–D, online supplemental figure 16C and D). The spleen size from EEF1B2 overexpressed mice was larger than that of CBD treated mice (figure 7E,F). EEF1B2 overexpression caused the increased total MDSCs in colon adenomas and bone marrow, especially the percentage of M-MDSC in MDSCs (figure 7G, online supplemental figure S17A). Besides, EEF1B2 overexpression led to M-MDSCs increasing both in PB and spleen, while the percentages of PMN-MDSCs seemed to have no changes. Additionally, the proportions of CD8⁺ T cells in the spleen of EEF1B2 overexpressed mice were significantly reduced compared with the CBD treatment group (online supplemental figure S17B). We finally constructed the genetic Eef1b2 KO mice and performed bone marrow transplantation to reconstitute the immune system of recipient mice (online supplemental figure S17C,D). The results showed wildt-ype recipient mice reconstituted with heterozygous Eef1b2^+/- donor bone marrow cells had slower tumor growth than mice reconstituted with bone marrow cells from wild-type littermates donor mice (figure 7H–K). The M-MDSCs and PMN-MDSCs infiltrating into tumor both significantly decreased in recipient mice with reconstituted with heterozygous Eef1b2^+/- donor bone marrow cells (figure 7L). All the above results suggested that the effect of CBD on inhibiting MDSCs generation and colorectal adenomas depended on it targeting inhibition of EEF1B2 (figure 7M).

Figure 7. The effect of EEF1B2 overexpression or knockdown in MDSCs mediated tumor progression. (A) The experimental design and timeline. Apc^min/+ mice were all fed with high fat diet to induce colon adenomas. The mice were treated with vehicle or CBD. The mice in EEF1B2 overexpression (EEF1B2 OE) group were pretreated with AAV-EEF1B2-mCherry via intra-bone marrow injection for 3 weeks to let EEF1B2 overexpression, and then treated with CBD. (B) EEF1B2 OE promoted the number and size of adenomas in mice colon at experiment termination by endoscope. (C) The adenomas number in mice colon from different groups. (D) The H&E staining of colon indicated that the increased adenomas in EEF1B2 OE mice. Scale bar, 20 µm. (E) The spleen size increased in EEF1B2 OE mice. (F) The spleen weight increased in EEF1B2 OE mice. (G) The FACS plot showing the change of MDSCs in colon in different groups. (H) Experiment flowchart of bone marrow transplantation with wild-type littermates donor mice and heterozygous Eef1b2^+/- donor mice. (I) Tumor weight in indicated two groups (n=6). (J) MC38 tumor photos at experiment termination. (K) MC38 tumor growth curve in wild-type recipient mice reconstituted with bone marrow cells from wild-type or heterozygous Eef1b2^+/- donor mice (n = 6). (L) Representative FACS plots (left) and statistic analysis (right) of tumor-infiltrating MDSCs in indicated two groups (n=5 in the group with wild-type donor mice and n=6 in the group with heterozygous Eef1b2^+/- donor mice). (M) The graphic abstract of the effect of CBD on its targeted molecule EEF1B2 and C/EBPβ in MDSCs generation of myeloepoiesis and immune suppressive environment in colon adenomas. *p<0.05, **p<0.01, ***p<0.001. P values were calculated by one-way ANOVA with Tukey’s multiple comparisons test. ANOVA, analysis of variance; CBD, cannabidiol; CMP, common myeloid progenitor; CRC, colorectal cancer; FACS, fluorescence-activated cell sorting; MDSC, myeloid-derived suppressor cell; M-MDSC, monocytic MDSC; ns, not significant; ns, not significant; PMN-MDSC, polymorphonuclear MDSC.

Discussion

The early intervention is very essential at the onset of colorectal adenomas formation.³⁶ In our study, we found that CBD effectively reduced colorectal adenomas generation in both AOM/DSS and HFD-Apc^min/+ model. Although some data indicated that cancer sufferers were using cannabis medicinally,³⁷ the mechanisms of CBD in cancer remains vague³⁸ and the use of CBD for cancer prevention at its precancerous early stage even had no report. Early detection and early intervention at the initial stage of carcinogenesis is vital for cancer prevention. While the strategies for the early prevention of many precancer lesions is not well developed. Hence, our results implied that CBD may be a promising compound to regulate the early onset of inflammatory precancerous colorectal adenoma and may be useful in the prevention of many other precancer types.

Malignant tumor developed from a dynamic process and commonly divided into four stages: initiation, promotion, progression, and metastasis. Immune escape is the critical step of tumorigenesis and the immunosuppressive tumor environment is an important reason for immune escape. Our scRNA-seq data showed that after CBD treatment, the infiltrated MDSCs decreased and the immunosuppressive function of MDSC was inhibited by CBD (figure 2). The increased MDSCs in human colon cancer or mice CRC model was confirmed by many of other papers.2839,42 MDSCs are immunosuppressive cell and are considered as the “queen bee” of the tumor microenvironment. MDSCs promoted the formation of Tregs, inhibit T cells activation and the function of adaptive immune system, often mature into tumor-associated macrophages or inflammatory dendritic cells.⁴³ Eliminating MDSCs for colon cancer immunotherapy is promising therapeutic strategy. Our results showed that CBD regulated immune response by directly reducing the generation of MDSCs in cell model, Apc^min/+ mice fed with HFD and syngeneic model (figures24). However, contrary to our findings, in some reports, such as experimental autoimmune encephalomyelitis, experimental autoimmune hepatitis, and corneal pathological angiogenesis and inflammation, CBD attenuated inflammation through induction of MDSCs.^{21 44} However, in colon cancers the scientists reduced MDSCs through using anti-CD300ld antibody or anti-CXCR2 to disrupt MDSC recruitment to tumor site and finally overcome immune therapy resistance.^{27 42 45} The opposite results may be attributed to the different system immune environment, different target and signaling pathway in different diseases. But from another perspective, the anti-inflammation effects of CBD in our results are consistent with the previous studies because in our data CBD can reduce the systemic inflammation as indicated by the decreased spleen weight. Because the chronic inflammation and inflammatory cytokine are the leading factors for inducing MDSCs generation in diseases, CBD may also inhibit MDSCs through its anti-inflammation effect in inflammation induced colorectal adenomas model.

How to regulate the generation and differentiation of MDSCs is not well studied. We found that CBD inhibited the differentiation of MDSCs from bone marrow cells (figure 3). Besides, CBD inhibited the immunosuppressive function of MDSCs and promoted part of immature MDSCs differentiating to M1 macrophage and DC, which overrode the immunosuppressive tumor environment (figure 3). The abnormal myelopoiesis in cancer induced many immature cell states and the M-MDSCs were reported previously to further differentiate to PMN-MDSC but not macrophages and DCs governed by epigenetic silencing of the retinoblastoma gene.³¹ Another study found that monocyte-like precursors of granulocytes can differentiate to PMN-MDSC in tumor-bearing mice.⁴⁶ Therefore, regulating the transition and differentiation of immature myeloid cells may be a promising strategy to eliminate MDSCs.

The first cannabis derived CBD oral solution, Epidiolex, have used for the treatment of Lennox-Gastaut and Dravet syndromes.⁴⁷ A paper lists CBD as one of the eleven clinical trials that will shape medicine in 2025¹⁴ for its clinical trials for the treatment of psychosis. Many previous reports find that CBD had therapeutic effect in neurological disorders such as epilepsy and peripheral neuropathic pain. The differential expression of target cells, targets, and blood flow in tissues and organs lead to inconsistent doses of CBD in tumors and neurological diseases. The two may act alone or interact with each other, exerting a synergistic effect. Nowadays, it is increasingly understood as the regulation of various activities in the nervous system to the development of tissues, organ formation, homeostasis, plasticity, regeneration, and immune function. Due to the disruption and repositioning of developmental and regenerative mechanisms caused by the formation, growth, and progression of cancer, the nervous system may be involved in many aspects of cancer pathophysiology. On the contrary, cancer and cancer therapies can affect and reshape the nervous system, leading not only to neurological dysfunction but also potentially driving pathological feedback loops in malignant tumors.⁴⁸ Therefore, we can reasonably speculate that, in some tumors, CBD may also affect the function of the nervous system and thus affect tumor progression,⁴⁸ such as through the brain gut axis.⁴⁹ Because different dose of drug may have different target protein expressed on different kinds of tissue cells, the antitumor effects can be achieved through regulating neuron cells, directly function on cancer cells or regulating tumor immune microenvironment through different dose of CBD. In many preclinical studies, it is also used to manage various symptoms because of its complex pharmacological profile. CBD functions differently in different diseases by targeting its interacting proteins.^{50 51} So, target identification of CBD is critical for the exact mechanism of action. In our study, we identified EEF1B2 as the direct target of CBD in inhibiting MDSCs differentiation by binding to the GEF domain of EEF1B2 (figure 5). In our study, CBD functioned as a new compound to inhibit EEF1B2 activity and the key function of EEF1B2 in regulating MDSCs generation was first found. Moreover, EEF1B2 was an important protein to be involved in some cancers. Importantly, CBD was the first found compound that have inhibitory effect on EEF1B2, which may be used in many other diseases that EEF1B2 involved.

Previous study reported that CBD interacted with the CB1 and CB2 but with low affinity and interestingly CBD had been shown to be an antagonist at CB1, CB2.⁵¹ CBD is also an inverse agonist for 3 G-protein receptors, which has been reported for mitigating neuropathic pain.⁵² Besides, CBD directly inhibits the activity of fatty acid amide hydrolase^{53 54} and hepatic cytochrome P450 enzymes.⁵⁵ In contrast, CBD acts as a protein activator in many conditions. As a ligand, CBD activates 5-hydroxytryptamine 1A serotonin receptors to attenuating anxiety and depression.⁵⁶ CBD is an agonist of ionotropic cannabinoid receptors including the TRP channel superfamily and TRPA.^{53 57} The anti-inflammatory effect and regulation of mitochondrial function of CBD is partially attributed to activate PPARγ nuclear receptors.^{23 58} All these studies indicate that CBD is a multi-target compound and may have therapeutic potential in many different diseases.

The effect of EEF1B2 in MDSCs generation is also unknown. Moreover, there is no compound reported to be an inhibitor of EEF1B2. So, we first found that CBD is a compound to exhibit inhibitory effect to EEF1B2. We found that expression of C/EBPβ, a key transcription factor in MDSCs differentiation, was positively regulated by EEF1B2. CBD decreased C/EBPβ expression through inhibited the activity of EEF1B2 in protein transcription of C/EBPβ, which was confirmed by our results (figure 6). Consistent with this, a previous study found that C/EBPβ antagonist peptide lucicebtide (ST101) enhanced antitumor immune responses of anti-programmed cell death protein-1 (PD-1) in 4T1 tumor bearing mice through inducing macrophage polarization toward a pro-inflammatory phenotype which indicated that high C/EBPβ expression can inhibit antitumor immune response.⁵⁹ Moreover, another study found that C/EBPβ expression can be inhibited by Zbtb46 expression which skewed bone marrow precursors toward immunostimulatory myeloid lineage output, leading to an immune-hot tumor microenvironment. And then, Zbtb46 mRNA treatment synergized with anti-PD-1 immunotherapy to improve tumor management in preclinical models.⁶⁰ In these studies C/EBPβ was downregulated by the Zbtb46 overexpression to enhance the efficacy of anti-PD-1 immunotherapy to improve tumor management. Therefore, based on these preclinical studies, we have reason to speculate high C/EBPβ expression be associated to poor survival in immunotherapy-treated patients in clinical. Therefore, our data first confirmed the important role of EEF1B2 in MDSCs generation and EEF1B2 may be targeted in the case of tumor-associated MDSCs regulation.

CBD are used medicinally by people around the world for many years. However, because of the lack of a clear pharmacological mechanism by which these compounds exert their actions, especially from an immunological perspective, the process of CBD moving to clinic is long and hard. Here, we demonstrated the beneficial effect of CBD and the underlying mechanisms on the treatment of colorectal adenomas. It provides the evidence for the function of CBD on immune regulation. Nevertheless, CBD is already FDA/EMA-approved drug for several childhood epilepsies and seizures associated with tuberous sclerosis complex which may streamline its clinical translation for colorectal adenomas in clinical use, which makes significant therapeutic implications in the early medical intervention for colorectal adenomas and its postoperative recurrence.

There are some limitations in our study. First, the regulation of CBD on MDSCs generation and MDSCs immunosuppressive function through targeting EEF1B2 cannot be further validated at this time by mimicking the inhibition effect of CBD through knockdown of EEF1B2. Except its effect on MDSCs, whether and how CBD has some effect on intestinal epithelial cells remains to be further investigated. Besides, because EEF1B2 expression is high in brain and CBD is reported to have some effect in brain, whether CBD targeting EEF1B2 is universal in brain-related diseases such as neurodegenerative diseases and depression is not verified and deserved for further study.

In conclusion, our results found the important role of CBD in preventing the malignant progression of colorectal adenomas, which provides a potential therapeutic strategy for the early immune regulation of colorectal adenomas. Our findings indicate that CBD may work as an EEF1B2 inhibitor to complete with EEF1A1 to bind to the GEF domain of EEF1B2, thereby inhibiting EEF1B2 mediated translational elongation and protein synthesis of C/EBPβ in MDSC generation and differentiation. Our findings not only demonstrate the therapeutic effect of CBD on colorectal adenomas but also highlight its potential to eliminate MDSCs in many other diseases by targeting the EEF1B2-EEF1A1 interaction.

Data availability statement

Data are available upon reasonable request.