A. Muciniphila Suppresses Colorectal Tumorigenesis by Inducing TLR2/NLRP3-Mediated M1-Like TAMs

Colorectal cancer is one of the most frequently diagnosed cancers and a major health care burden worldwide (1). Accumulating evidence suggests that epigenetic alterations, environmental factors, and gut microbiota contribute to the initiation and progression of colorectal cancer (2–4). Patients with colorectal adenomas, especially those with the inherited cancer-predisposition syndrome familial adenomatous polyposis (FAP), have a high risk of developing colorectal cancer (5). Adenomatous polyposis coli (APC) is the tumor-suppressor gene that is mutated in individuals who have FAP; it also is frequently mutated in patients who have sporadic colorectal cancer (6, 7).

A role for gut microbiota in the initiation and development of colorectal cancer is emerging. For example, FAP patients harboring bacterial biofilms of Escherichia coli and Bacteroides fragilis have increased susceptibility to colorectal cancer (8). Fusobacterium nucleatum is a key oncogenic bacterium that promotes colorectal cancer cell proliferation (9, 10), mediates chemotherapy resistance via induction of autophagy (11), and facilitates colorectal cancer metastasis by modulating KRT7-AS/KRT7 (12). In addition, Bacteroides fragilis promotes tumor growth by mediating tumor immune escape (13). Most of these studies on colorectal cancer focused on oncogenic bacteria, however, and the potential role of beneficial bacteria remains largely unknown.

Akkermansia muciniphila (A. muciniphila) is a gram-negative anaerobic bacterium belonging to the Verrucomicrobia phylum. It resides in the human gut mucosa where it degrades mucin and exhibits strong adhesion to gut epithelial cells (14). A. muciniphila is a promising probiotic, because it can improve metabolic health (15) and glucose homeostasis (16), and it can prolong lifespan in progeroid mice (17). A. muciniphila also alleviates acute and chronic colitis induced by dextran sulfate sodium (DSS; refs. 18, 19). Moreover, A. muciniphila improves the efficacy of cancer immunotherapy by increasing the recruitment of CCR9⁺CXCR3⁺CD4⁺ T lymphocytes (20) and inducing antigen-specific T-cell responses (21). These studies imply that A. muciniphila is a highly promising probiotic. However, whether A. muciniphila can protect against colorectal cancer by modulating the tumor immune microenvironment remains unclear.

In this study, we aimed to identify the effect of A. muciniphila on colorectal cancer pathogenesis and the underlying molecular mechanisms of these effects. We found that the abundance of A. muciniphila was significantly reduced in patients with colonic adenoma or colorectal cancer. Supplementation with A. muciniphila suppressed colonic tumorigenesis in Apc^Min/+ mice. Mechanistically, we showed that A. muciniphila induced M1-like macrophage activation in vivo and in vitro, an effect mediated via TLR2/NLRP3-dependent signaling. These results highlight the protective effects of A. muciniphila in colorectal cancer, which occurs through modulation of M1-like macrophage response, and provide a therapeutic approach for probiotic-based modulation of the tumor immune microenvironment.

Fresh stool samples were obtained from 50 patients with colorectal cancer, 36 patients with colonic adenoma, and 38 healthy subjects undergoing physical examination at the Sir Run Run Shaw Hospital of Zhejiang University School of Medicine from 2018 to 2019 (cohort 1). Fresh tumor and paired normal tissues were obtained from 64 patients with colorectal cancer who underwent surgical resection at Sir Run Run Shaw Hospital from 2017 to 2018 (cohort 2). Eighty-one fresh endoscopic colorectal cancer tissues were collected in the validation cohort from Affiliated Lishui Hospital of Zhejiang University School of Medicine (2017–2018; cohort 3). All samples were refrigerated in liquid nitrogen until use. All samples were anonymously coded in accordance with local ethical guidelines, written informed consent was obtained, and the protocol was approved by Clinical Research Ethics Committee of the Sir Run Shaw Hospital and Affiliated Lishui Hospital, Zhejiang University School of Medicine. All aspects of the study were conducted in accordance with the principles of the Declaration of Helsinki.

Bacterial DNA from human fecal samples was extracted using QIAGEN stool kits (Cat. #51604) and bacterial genomic DNA (gDNA) was extracted from human tissues using QIAGEN DNA mini kits (Cat. #56304) according to the manufacturers' protocol. qRT-PCR was performed to assess the A. muciniphila genes Universal Eubacteria 16S and PGT using a ROCHE LightCycler480 System (Rotor gene 6000 Software). Each reaction was performed in triplicate in 10 μL reaction system containing SYBR Premix Ex Taq (Cat. #RR820A, Takara), primers and 100 ng template gDNA. Relative abundance was calculated by -ΔCt method. Universal Eubacteria 16S was used as internal reference gene for stool samples. The PGT gene was used as an internal control for tissue samples. Primers used are listed in the Supplementary Table S1.

GMrepo is an annotated human gut metagenomic data repository for microbiota. The relative abundance of bacteria at the species level was calculated for each stool sample, with the total abundance values of 100%. Using GMrepo RESTful APIs for R (version 3.6.1, https://www.r-project.org) and the RStudio (version 1.1.442, https://www.rstudio.com) software, the relative abundance of A. muciniphila in the stool samples of the healthy individuals and patients with colorectal cancer was acquired from the GMrepo database. The data quality was assessed by consulting the description of the samples and Supplementary Data of related publications. Then, the relative abundance of A. muciniphila for the healthy individuals and patients with colorectal cancer was analyzed.

A. muciniphila was purchased from the ATCC (A. muciniphila BAA-835). Bacterial gDNA was extracted and amplified, and then 16S ribosomal sequencing (27F, 1492R) was performed to confirm bacterial strain by Tsingke Biotech. The bacterium was cultured in Brain Heart Infusion (BHI; Cat. #237500, BD Difco) supplemented with 0.05% mucin Type II (Cat. #84082–64–4, Sigma-Aldrich) under an atmosphere of 10% H₂, 10% CO₂, and 80% N₂ in an AW500SG anaerobic workstation (ELECTROTEK) for 72 hours at 37°C. The nonpathogenic commensal intestinal bacteria, E. coli strain DH5a (Code No.9057, Takara), which was used as a negative control, was cultured in Luria-Bertani medium (Cat. #A507002 Sangon Biotech) at 37°C. When the optical density (OD) at 600 nm of A. muciniphila reached 0.5, bacteria were centrifuged at 5,000 rpm for 10 minutes and suspended by 1 × 10⁹ CFU/300 μL in anaerobic PBS (Cat. #E607008, Sangon Biotech) for further application. For analysis of the viability of A. muciniphila, the bacterial suspension was inoculated on BHI-containing 1% agarose (Cat. # A505255, Sangon Biotech) and then incubated for at least 3 days at 37°C in the anaerobic incubator.

HCT116 human colon cancer cells were obtained from the ATCC at the beginning of this project. Cells were maintained at 37°C under 5% CO₂ in McCoy's 5A (Cat. #GNM16600, Genom) with 10% (vol/vol) FBS (Cat. #10270, Gibco) supplemented with 1% (vol/vol) penicillin and streptomycin (Cat. #P1400, Solarbio) and maintained in culture for a maximum of 2 months or 10 passages. Murine cell lines were also obtained from the ATCC at the beginning of our study and maintained in culture for a maximum of 2 months or 10 passages. CT26 murine colon carcinoma cancer cells and Raw264.7 murine macrophages were cultured at 37°C under 5% CO₂ in RPMI-1640 with 10% (vol/vol) FBS (Cat. #10270, Gibco) supplemented with 1% (vol/vol) penicillin and streptomycin (Cat. #P1400, Solarbio). All cells tested negative for Mycoplasma contamination and were authenticated on the basis of short tandem repeats fingerprinting before use.

Mice were sacrificed and soaked with 75% alcohol for 10 minutes for disinfection. The bone marrow cavities of the leg bones were opened, flushed with DMEM (GNM12800, Genom). The supernatants were passed through 200-mesh filter and the cell suspension was collected after centrifugation for 5 minutes at 500 × g 4°C. Then cells were incubated in ACK lysis buffer (Cat. #R1010, Solarbio) for 5 minutes to remove the red blood cells. Cells were cultured in DMEM (GNM12800, Genom) with 10% (vol/vol) FBS (Cat. #10270, Gibco) supplemented with 2% (vol/vol) penicillin and streptomycin (Cat. #P1400, Solarbio) and 20 ng/mL mouse macrophage colony stimulating factor (M-CSF; Cat. #CB34, Novoprotein). DMEM containing M-CSF was renewed every 2 days until macrophages were induced. Six days later, bone marrow–derived macrophages (BMDM) were confirmed by flow cytometry for F4/80 (Cat. #565411, Clone T45–2342, 1 μg/mL, BD Biosciences).

All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) of Zhejiang University (ZJU). All animal experiments strictly adhered to protocols, policies, and ethical guidelines formulated by our IACUC. Apc^Min/+ mice were purchased from Nanjing Biomedical Research Institute of Nanjing University (NBRI), China. BALB/C nude mice and C57L/B6 mice were purchased from Shanghai SLAC Laboratory Animal, China. C57BL/6J NLRP3^tm1Bhk (NLRP3^−/−) mice were purchased from The Jackson Laboratory. All mice were maintained in ventilated cages with 12-hour light/dark cycles, constant temperature and humidity, enriched water and ad libitum feeding under specific pathogen–free (SPF) conditions.

Before intragastric bacteria administration, male C57BL/6J Apc^Min/+ mice (6–8 weeks of age, 20 g) were fed with 2 mg/mL streptomycin (Cat. #MB1275, Meilunbio) in the drinking water for 7 days to ensure the consistency of regular microbiota and facilitate A. muciniphila colonization. Apc^Min/+ mice were randomly assigned to three groups. Mice in the A. muciniphila and E. coli groups were administrated with 1 × 10⁹ CFU A. muciniphila or DH5a suspended in 300 μL sterile anaerobic PBS, respectively, every 2 days for 3 months. The control group was administered with PBS. Two cycles of 10-day 1% DSS (Cat. #160110, MP Biomedicals) was given in drinking water to accelerate tumorigenesis. To suppress NLRP3 in Apc^Min/+ mice, the NLRP3 inhibitor MCC950 (10 mg/kg body weight, Cat. #CP-456773, Selleck) was intraperitoneally injected during the final 6 weeks of the experiment. The body weight of mice was measured every week, and mouse anal prolapse was assessed by observation during the final month of the experiment.

At the indicated time intervals, colon and spleen tissues were harvested after fasting. Colon tissues were photographed and the number and size of tumors were measured. Tumor sizes (diameter) were quantified as <1 mm, 1 to 2 mm, 2 to 3 mm, or >3 mm. Tumor load was calculated as the sum of all tumor diameters in a single mouse.

Female BALB/C nude mice (3–4 weeks of age, 15 g) were kept in SPF conditions. HCT116 or CT26 cells were washed twice with PBS and harvested using trypsin-EDTA solution (Cat. # GNM25200, Genom). 3 × 10⁶ HCT116 or CT26 cells were mixed with A. muciniphila, DH5α (MOI = 10:1) or PBS with 25 μL Matrigel matrix (Cat. #354234, Corning Biocoat), and then injected (100 μL per mouse) subcutaneously into the right flank of nude mice.

For inhibitor treatments, 3 × 10⁶ HCT116 or CT26 cells mixed with A. muciniphila (MOI = 10:1) or PBS were injected (100 μL per mouse) subcutaneously into the right flank of BALB/C nude mice. One week later, IL-1RA (3 mg/kg body weight; Cat. #HY-108841, MCE) or S-Methylisothiourea sulfate (SMT; 3 mg/kg body weight; Cat. #HY-79457, MCE) or MCC950 (10 mg/kg body weight) were intraperitoneally injected into the mice every two days.

For NLRP3^−/−mice (3–4 weeks of age, 15 g), CT26 cells were washed and harvested using trypsin-EDTA solution (Cat. # GNM25200, Genom). 5 × 10⁶ CT26 cells were mixed with A. muciniphila (MOI = 10:1) or PBS with 25 μL Matrigel matrix (Cat. #354234, Corning Biocoat), and then injected (100 μL per mouse) subcutaneously into the right flank of C57BL/6J wild-type or NLRP3^−/−mice.

After 7 days implantation, tumor volume was monitored every two days and calculated as follows: Volume = 0.54 × L × W², where L is the longest diameter and W is the shortest diameter. At the terminal time, the tumor weights were recorded.

Male C57L/B6 mice (6–8 weeks of age, 20 g) were given 2 mg/mL streptomycin (Cat. #MB1275, Meilunbio) in drinking water for 7 days to deplete most bacteria. The mice were then randomly assigned to vehicle, A. muciniphila (1 × 10⁹ CFU/d) or E. coli (1 × 10⁹ CFU/d) groups, respectively. Oral gavage lasted 14 days, followed by 7 days of 3% DSS (Cat. #160110, MP Biomedicals) in drinking water. Three weeks later, mice were sacrificed for further analysis.

Colorectal tumors were fixed overnight with 10% formalin at room temperature and then embedded in paraffin. Sections of 5 μm were stained with hematoxylin and eosin (H&E) for pathologic analysis. For IHC, sections of paraffin-embedded tissues were stained by PCNA-specific antibody (Cat. #GB11010, diluted 1:1000, Servicebio) and iNOS-specific antibody (Cat. #GB11119, diluted 1:1000, Servicebio), and visualized by DAB staining (Cat. #G1212, Servicebio) according to the manufacturer's instructions.

For immunofluorescence staining, after deparaffinization, paraffin-embedded tissues were incubated with NLRP3-specific antibody (Cat. #GB11300, diluted 1:1000, Servicebio) CD68-specific antibody (Cat. #GB14043, diluted 1:500, Servicebio). Sections were visualized by FITC-TSA (G1222, diluted 1:1000, Servicebio) and CY3-TSA (G1223, diluted 1:2000, Servicebio), respectively. Finally, sections were counterstained with 40,6-diamidino-2-phenylindole (DAPI; G1012, Servicebio). Sections were imaged with NIKON digital sight DS-FI2 (NIKON Eclipse ci). For each section, the number of double-positive (CD68⁺NLRP3⁺) cells was analyzed in at least 5 randomly selected high-power fields.

Colon tissues were collected from cecum to anus without lymph and adipose tissue. Colon tissues were then cut into small pieces and put into D-Hanks buffer (Cat. #MA0039, Meilunbio). The tissues were incubated in D-Hanks buffer supplemented with 1 mmol/L DTT (Cat. #MB3047–1, Meilunbio) and 5 mmol/L EDTA (Cat. #MB2514, Meilunbio) on a shaker (150 g) for 30 minutes at 37°C. Any remaining colon tissues were cut into 1-mm pieces and further digested in Hanks buffer (Cat. #MA0041, Meilunbio) supplemented with 1 mg/mL Type IV collagenase (Cat. #A005318, Sangon Biotech) for 1 hour at 37°C with 150 g shaking. After complete digestion, the cell suspension was passed through a 200-mesh filter and then centrifuged at 300 × g for 5 minutes. The isolated colon lamina propria cells were further analyzed.

Colon lamina propria cells were counted (5 × 10⁶ cells/per sample), and the surface of the cells was stained for 30 minutes at room temperature using Fixable viability stain 510 (Cat. #564406, 2 μg/mL, BD Biosciences) for live cell staining, then cells were divided into two wells. One sample was stained with a panel of antibodies designed to analyze the surface of myeloid cells: Alexa Fluor 700-CD45 (Cat. #560510, Clone 30-F11, 1 μg/mL, BD Biosciences), BV421-F4/80 (Cat. #565411, Clone T45–2342, 1 μg/mL, BD Biosciences), AF488-CD11b (Cat. #557672, Clone M1/70, 1 μg/mL, BD Biosciences), PE-MHC-II (Cat. #557000, Clone M5/114, 1 μg/mL, BD Biosciences), PECP-CY5.5-Gr-1 (Cat. #108427, Clone RB6–8C5, 1 μg/mL, BioLegend), PE-CY7-CD11c (Cat. #117317, Clone N418, 1 μg/mL, BioLegend), and APC-CY7-Ly6C (Cat. #128025, Clone HK1.4, 1 μg/mL, BioLegend). Another sample was stained with a panel of antibodies designed to analyze the surface of T cells: Alexa Fluor 700-CD45, PECP-CY5.5-CD3 (Cat. #551163, Clone 145–2C11, 1 μg/mL, BD Biosciences), BV605-CD4 (Cat. #563151, Clone RM4–5, 1 μg/mL, BD Biosciences) and APC-CY7-CD8 (Cat. #561967, Clone 53–6.7, 1 μg/mL, BD Biosciences). Next, cells were permeabilized with Foxp3/Transcription Factor Staining Buffer Set (Cat. #00–5523–00, eBioscience) and stained for 60 minutes at room temperature with APC-CD206 (Cat. #17–2061–82, Clone MR6F3, 2 μg/mL, eBioscience) for intracellular staining. Samples were analyzed using Flow Cytometer (BD Biosciences). Subsequent analysis was performed with FlowJo software (Tree Star Inc.). Gating strategy is shown in the Supplementary Fig. S1.

BMDMs or Raw264.7 cells were seeded at a density of 5 × 10⁵ cells per well in 6-well plate and cultured in DMEM or RPMI-1640 medium with 10% (vol/vol) FBS (Cat. #10270, Gibco) overnight. BMDMs and Raw264.7 cells were incubated with A. muciniphila at an MOI of 100:1 for 24 hours. Finally, BMDMs and Raw264.7 cells were digested by trypsin-EDTA solution (Cat. #GNM25200, Genom) for further analysis. For the NLRP3-deficient assay, BMDMs from wild-type or NLRP3^−/− mice were cultured with A. muciniphila at an MOI of 100:1 for 24 hours for further analysis. Raw264.7 cells were treated with the NLRP3 inhibitor MCC950 (10 nmol/L) for 2 hours followed by incubating with PBS or A. muciniphila at an MOI of 100:1 for 24 hours. For the TLR2-deficient assay, BMDMs and Raw264.7 cells were treated with the TLR2 inhibitor CPT22 (10 μmol/L; Cat. #HY-108471, MCE) for 2 hours followed by incubation with PBS or A. muciniphila for 24 hours.

For the in vitro tumor-suppressive effect assay, BMDMs were incubated with A. muciniphila for 24 hours, then HCT116 or CT26 cells were seeded at a density of 5 × 10⁴ cells per well into the lower chamber of a 24-well Transwell plate (Cat. #3413, Corning) and A. muciniphila–treated BMDMs (1 × 10⁵ per well) were placed into the upper chamber. Cell viability assays were performed at 24, 48, and 72 hours to assess the proliferation of HCT116 or CT26 cells. For cytokine inhibitor assays, CT26 cells cocultured with A. muciniphila–treated BMDMs were treated with vehicle, IL1RA (2 μmol/L) or SMT (20 μmol/L), then the cell viability was evaluated at 24, 48, and 72 hours. For the NLRP3-deficient assay, wild-type or NLRP3^−/− BMDMs were incubated with A. muciniphila for 24 hours, and then HCT116 or CT26 cells were cocultured with A. muciniphila–treated wild-type or NLRP3^−/− BMDMs and cell viability was evaluated at 24, 48, and 72 hours. Each test was repeated in triplicates.

For cell viability assays, HCT116 or CT26 cells with different treatments were seeded at 2 × 10³ cells per well in 96-well plates. Cell viability was then analyzed using a Cell Counting Kit (CCK-8; Cat. #CK04, Dojindo) according to the manufacturer's instructions. Briefly, after removing the medium, cells were incubated with CCK8 for 2 hours and the absorbance was determined at 450 nm by a spectrophotometer at 24, 48, and 72 hours, respectively. Each test was repeated five times.

Colorectal tumors were cut into small parts and mechanically homogenized with a tissue disruptor (Tiss-24, Jinxin Biotechnology) in PBS. The homogenate was centrifuged at 12,000 × g for 10 minutes and the supernatant was collected. Then protein concentration was quantified with BCA protein assay kits (Cat. #PC0020, Solarbio) according to the manufacturers' protocol. The levels of the M1 macrophage–related cytokines TNFα, IL23, and IL27 and the M2 macrophage–related cytokines IL4, IL5, and IL10 were measured by cytokine profiling assay using Cytokine 17-Plex Mouse ProcartaPlex Panel (Cat. #EPX170–26087–901, Thermo Fisher Scientific) according to the manufacturer's instructions. The levels of cytokines were recorded by Luminex technology using Bio-Plex 200 machine (Bio-Rad Laboratories).

RNA was extracted from BMDMs using 1 mL RNAiso Plus reagent (Cat. #9108, Takara). Total RNA was reverse transcribed using PrimeScript RT reagent Kit (Cat. #RR047A, Takara). RT-qPCR was performed using SYBR Premix Ex Taq (RR820A, Takara) in the Light Cycler480 Real-Time PCR System (Roche) using cDNA. Each reaction was performed in triplicate in 10 μL reactions containing SYBR Premix Ex Taq (Cat. #RR820A, Takara), primers and 200 ng template cDNA. Three replicates were performed. cDNA was amplified by PCR under the following condition: 95°C for 2 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for 30 seconds. The mouse primers of Tnfα, Il6, Il12, Inos, Pcna, Pclaf, Ki67, Rfc1, Tlr2, Tlr4, Tlr5, and Tlr9 and human primers of NLRP3 and TLR2 used are listed in Supplementary Table S1. Relative mRNA expression was calculated using comparative cycle method (2^−ΔΔCt). Gapdh was used as the internal reference gene for mice tissues and Β-actin was used as an internal control for human tissues.

Colorectal tumors or BMDMs were homogenized in RIPA buffer (Cat. #R0010, Solarbio). The homogenate was centrifuged at 4°C for 15 minutes at 15,000 × g and then the supernatant was collected. Protein concentration was quantified with BCA protein assay kits (Cat. #PC0020, Solarbio). Proteins were separated using 10% SDS polyacrylamide gels, and then transferred onto PVDF membranes. The membranes were blocked with 5% skimmed milk for 1 hour and then immunoblotted with NLRP3-specific antibody (Cat. #ab214185, diluted 1:1,000, Abcam), IL1β-specific antibody (Cat. #12242, diluted 1:1,000, CST), iNOS-specific antibody (Cat. #13120, diluted 1:1,000, CST), TLR2-specific antibody (Cat. #13744, diluted 1:1,000, CST), p-p65–specific antibody (Cat. #3033, diluted 1:1,000, CST), iκBα-specific antibody (Cat. #4814, diluted 1:1,000, CST) at 4°C overnight. Membranes were then incubated with Goat anti-Rabbit IgG-HRP (Cat. #HA1001, diluted 1:10,000, HUABIO) or Goat anti-Mouse IgG-HRP (Cat. #HA1006, diluted 1:10,000, HUABIO) at room temperature for 1 hour and bands were visualized using an ECL kit (Cat. #FD8000, Fdbio science). β-actin (Cat. # M1210–5, diluted 1:3,000, HUABIO) was used as a loading control.

RNA sequencing (RNA-seq) data of 308 normal tissues from the GTEx dataset and 380 colon carcinomas (with 51 normal tissues) from The Cancer Genome Atlas (TCGA) were accessed from the UCSC Xena public data hub (University of California, https://xena.ucsc.edu). The RNA-seq gene expression was normalized as RSEM norm_count. Inclusion criteria from TCGA TARGET GTE cohort was set as sample type (normal tissue, solid tissue normal and primary tumor) and primary site (colon and rectum). Data were analyzed with the GraphPad Prism 7.0 (GraphPad Software), R (version 3.6.1), and R Bioconductor packages. The Mann-Whitney test was conducted to compare normal tissues and colon carcinomas groups. Spearman correlation analysis was conducted for NLRP3 and TLR2 correlation analysis.

Data were expressed as mean ± standard deviation (SD) or standard error of mean (SEM). Differences between groups were analyzed by Student t test, one-way ANOVA test, Kruskal–Wallis test, Mann–Whitney test or Wilcoxon-matched pairs signed-rank test. Spearman correlation analysis was used for correlation analysis. A P value <0.05 was considered statistically significant. All statistical analyses were performed using GraphPad Prism 7.04 software (GraphPad Software, Inc.) and SPSS 19.0 for Windows (SPSS Inc.).

We examined fecal A. muciniphila abundance in healthy subjects and patients with colonic adenoma or colorectal cancer (cohort 1). qPCR revealed that A. muciniphila abundance was significantly decreased in patients with colorectal cancer or colonic adenoma compared with healthy controls (Fig. 1A). Consistent with this observation, the GMrepo database indicated that fecal A. muciniphila abundance in patients with colorectal cancer was significantly lower as compared with healthy individuals (Fig. 1B). In addition, we showed that A. muciniphila abundance was decreased in tumor tissue compared with their corresponding adjacent normal mucosa (cohort 2; Fig. 1C). These results suggested that A. muciniphila was depleted in colorectal cancer.

Apc^Min/+ mice is a model of spontaneous intestinal adenomas (22). We validated that A. muciniphila could colonize Apc^Min/+ mice by analyzing the abundance of A. muciniphila in stool and colon tissues by qRT-PCR (Supplementary Fig. S2). We then administrated A. muciniphila, E. coli or PBS to Apc^Min/+ mice that were pretreated with antibiotic for 7 days (Fig. 1D). After 12 weeks of treatment, we observed that A. muciniphila suppressed colon tumorigenesis in Apc^Min/+ mice, as assessed by tumor number (5.333 ± 0.7638 vs. 9.714 ± 1.04, P <0.01), tumor size (10% vs. 20% for tumors with size >3 mm) and tumor load (9.056 ± 1.621 vs.15.11 ± 1.654, P <0.05), as compared with PBS control (Fig. 1E and F). In contrast, E. coli had no significant effect on tumor number and load as compared with the PBS control group (Fig. 1E and F). Colon tumors from A. muciniphila–treated mice had decreased expression of the cell proliferation marker PCNA (Fig. 1G). In addition, A. muciniphila–treated Apc^Min/+ mice had decreased spleen weight (Fig. 1H). To validate the tumor-suppressive effect of A. muciniphila, we injected HCT116 or CT26 cells treated with PBS, A. muciniphila or E. coli (MOI = 10:1) into BALB/C nude mice, and found that A. muciniphila treatment suppressed tumor growth compared with E. coli or PBS control (Supplementary Fig. S3). Collectively, these results indicated that A. muciniphila could suppress colon tumorigenesis in vivo.

Tumor-infiltrating immune cells play an important function in promoting or inhibiting tumor initiation and progression (23). To determine whether A. muciniphila modulates immune responses in Apc^Min/+ mice, we analyzed CD8⁺ T cells, CD4⁺ T cells, dendritic cells (DC), myeloid-derived suppressor cells (MDSC), and M1- and M2-like tumor-associated macrophages (TAM) from colon lamina propria by flow cytometry assays. We found that the frequency of M1-like TAMs was increased in A. muciniphila–treated Apc^Min/+ mice compared with mice gavaged with PBS (28.1% vs. 11.9%, P < 0.05), whereas the frequency of other immune subtype cells showed no significant difference between these two groups (Fig. 2A and B).

We further used the DSS model to investigate the function of bacteria on the gut–bone marrow axis in damaged mucosa. We observed that A. muciniphila but not E. coli increased the levels of M1-like monocytes (CD45⁺Ly6C⁺MHCII⁺) in colon, blood, and bone marrow (Supplementary Fig. S4), which indicated that a gut–bone marrow axis is ignited after oral gavage with A. muciniphila (but not E. coli). Consistent with enrichment of M1-like TAMs in the presence of A. muciniphila, M1-like TAM–related cytokines, including IL23, TNFα, and IL27 were significantly induced in tumor homogenate in mice treated with A. muciniphila (Fig. 2C). However, there was no significant difference in M2-like TAM–associated cytokines such as IL10, IL5, and IL4 (Fig. 2C). Tumors from A. muciniphila–treated mice also showed higher expression of the M1-like TAM–related marker iNOS compared with control mice, suggesting that A. muciniphila could stimulate an M1-like TAM response (Supplementary Fig. S5).

To confirm whether A. muciniphila induced M1-like macrophage polarization in vitro, we treated BMDMs from C57BL/6J mice with A. muciniphila or E. coli (Fig. 2D). We observed that A. muciniphila induced more M1-like macrophages compared with E. coli or control (Fig. 2E). In addition, the mRNA expression levels of M1-like macrophage–related markers, including Inos, Tnfa, and Il6, were significantly increased after A. muciniphila treatment compared with control (Fig. 2F). We also observed similar findings in the macrophage Raw264.7 cell line (Supplementary Fig. S6).

To determine whether induction of M1-like macrophages mediates the tumor-suppressive effect of A. muciniphila, we cocultured A. muciniphila–treated BMDMs with HCT116 or CT26 cells in Transwell plates. A. muciniphila–treated BMDMs significantly inhibited colorectal cancer cell proliferation (Fig. 2G). In addition, the cell proliferation markers Pcna, Pclaf, Ki67, and Rfc1 were significantly decreased in CT26 cells cocultured with A. muciniphila–treated BMDMs (Fig. 2H). To determine which TAM-associated factor might contribute to tumor suppression, we cocultured A. muciniphila–treated BMDMs with CT26 cells in the presence of IL1RA (IL1 receptor antagonist) or SMT (iNOS inhibitor). IL1RA, but not SMT, abolished the antiproliferative effect of A. muciniphila–treated BMDMs (Fig. 2I). Furthermore, we observed that IL1RA treatment partly reversed the growth inhibitory effect of A. muciniphi on CT26 and HCT116 tumors in nude mice (Fig. 2J; Supplementary Fig. S7). Taken together, these data demonstrated that A. muciniphila specifically promoted an M1-like TAM response in the colorectal cancer microenvironment and that this could suppress tumorigenesis in vivo and in vitro.

Accumulating evidence indicates that macrophage polarization involves multiple pathways (24–26), including the NLRP3 pathway (27). NLRP3 plays a critical and well-defined role in the host innate immune response to various microbes (28). To determine whether A. muciniphila induces an M1-like TAM response via the NLRP3 pathway, we examined NLRP3 expression in Apc^Min/+ mice. NLRP3 was significantly induced in A. muciniphila–treated Apc^Min/+ mice, as evidenced by immunofluorescence staining and Western blot assay (Fig. 3A). Moreover, protein expressions of NLRP3, pro-IL1β, and IL1β (p17) in both BMDMs and Raw264.7 cells were significantly increased by A. muciniphila treatment as compared with E. coli or PBS treatment (Fig. 3B).

To confirm whether A. muciniphila stimulates M1-like TAM response in an NLRP3-dependent manner, BMDMs were isolated from NLRP3^−/− mice, and we found that expression of iNOS was reduced in A. muciniphila-NLRP3^−/− BMDMs (Fig. 3C). Similarly, treatment with MCC950, an NLRP3 inhibitor, decreased iNOS expression in Raw264.7 cells (Fig. 3C). In addition, the A. muciniphila–mediated M1-like macrophage response was impaired in NLRP3^−/− BMDMs compared with wild-type BMDMs (Fig. 3D). Furthermore, NLRP3^−/− BMDMs showed significantly attenuated A. muciniphila–induced expression of M1-like macrophage–related molecules, including Inos, Tnfa, Il6, and Il12 (Fig. 3E). These data indicated that the M1-like TAM response induced by A. muciniphila was NLRP3 dependent.

To confirm our hypothesis that A. muciniphila suppresses colorectal cancer through NLRP3-dependent mechanism, we cocultured A. muciniphila-treated wild-type or NLRP3^−/− BMDMs with HCT116 and CT26 cells. The growth inhibitory effect of A. muciniphila was attenuated in NLRP3^−/− BMDMs compared with wild-type BMDMs (Fig. 4A). To validate this in vivo, we administrated A. muciniphila with or without NLRP3 inhibitor MCC950 to Apc^Min/+ mice. Consistent with an essential role for NLRP3 on the tumor-suppressive effect of A. muciniphila, cotreatment of MCC950 with A. muciniphila promoted tumor growth, as shown by increased colon tumor number, colon tumor load and small intestine tumor number (Fig. 4B and C). Importantly, MCC950 also abolished A. muciniphila–induced M1-like TAM response (Fig. 4D). Concordantly, the tumor-suppressing effect of A. muciniphila in a HCT116 xenograft model was partially abolished by cotreatment with MCC950 (Fig. 4E). Furthermore, we found that the tumor-suppressive effect of A. muciniphila on syngeneic CT26 subcutaneous tumors was abrogated by cotreatment with MCC950 or by genetic loss of NLRP3 (Fig. 4F and G). Collectively, the in vitro and in vivo data supported that NLRP3 deficiency eliminated the tumor-protective effect of A. muciniphila on colorectal cancer.

Previous results suggest that A. muciniphila specifically activates human HEK-Blue cells expressing TLR2, but not cells expressing TLR5 or TLR9 (29). To elucidate how A. muciniphila interacts with macrophages to promote NLRP3 activation, we explored the expression of TLR2, TLR4, TLR5, and TLR9 in macrophages incubated with A. muciniphila. We observed that A. muciniphila stimulated Tlr2 mRNA and protein expression in BMDMs and Raw264.7 cells as compared with PBS control, whereas the expressions of Tlr4, Tlr5, and Tlr9 were unchanged (Fig. 5A and B). Meanwhile, we found that phospho-p65 was upregulated and the NF-κB inhibitor IκB-α was downregulated after treatment of BMDMs with A. muciniphila (Fig. 5B). Increased protein expression of TLR2 by A. muciniphila was further confirmed in Apc^Min/+ mice (Fig. 5B).

To determine whether macrophage TLR2 recognizes A. muciniphila and whether it has a role in NLRP3 activation during A. muciniphila-induced M1-like macrophage response, we performed TLR2 loss-of-function assays with its inhibitor CPT22 in BMDMs or Raw264.7 cells. We found that the A. muciniphila–mediated M1-like macrophage response was impaired by CPT22 treatment of BMDMs or Raw264.7 cells (Fig. 5C; Supplementary Fig. S8). CPT22 also attenuated the A. muciniphila–mediated expression of phospho-p65, NLRP3, pro-IL1β, and iNOS in BMDMs or Raw264.7 cells (Fig. 5D), as well as mRNA levels of M1-like macrophage–related molecules, including Inos, Tnfa, Il6, and Il12 (Fig. 5E). These results suggested that TLR2 might be involved in the activation NF-κB/NLRP3 and macrophage phenotype polarization induced by A. muciniphila.

Given that A. muciniphila promotes M1 macrophage polarization via upregulation of NLRP3 and TLR2 in vitro and in mice, we evaluated the clinical association between A. muciniphila abundance and macrophages, NLRP3 and TLR2 levels. We first measured A. muciniphila at DNA level to define low (−ΔC_t <−10) or high (−ΔC_t >−10) abundance in colorectal cancer tissues. A high abundance of A. muciniphila was associated with increased expression levels of the M1-like macrophage marker iNOS in colorectal cancer tissues (cohort 2; Fig. 6A). Consistent with this observation, we found that A. muciniphila was associated with higher expression of macrophage NLRP3 (CD68⁺NLRP3⁺ dual positive) in colorectal cancer tissues, as determined by immunofluorescence staining (cohort 2; Fig. 6B). Moreover, we checked NLRP3 and TLR2 transcript expressions in 64 matched colorectal cancer and normal tissues. NLRP3 and TLR2 expressions were reduced in colorectal cancer tissue, and TLR2 expression was correlated with NLRP3 activation (R = 0.3656, P < 0.0001; Fig. 6C). In addition, we confirmed that A. muciniphila was associated with the increased NLRP3 and TLR2 expressions, and the positive relationship between NLRP3 and TLR2 was confirmed in an independent clinical cohort (cohort 3; R = 0.7902, P < 0.0001; Fig. 6D). We also validated the above results using TCGA GTEx datasets and found that NLRP3 and TLR2 were reduced in colorectal cancer tissues and there was a positive correlation between the two (R = 0.7111, P < 0.0001; Fig. 6E).

Gut dysbiosis is involved in pathologic processes, including intestinal inflammation and carcinogenesis (30, 31). However, few studies have reported on the role of beneficial bacteria in colorectal cancer (32, 33). Herein, we first identified that A. muciniphila suppressed colorectal tumorigenesis in Apc^Min/+ mice by targeting TAMs in the tumor immune microenvironment. Mechanistically, we revealed that A. muciniphila activates TLR2/NF-κB and NLRP3, leading to increased M1-like TAMs and the suppression of colonic tumorigenesis.

Our study found that A. muciniphila is depleted in patients with colorectal cancer (Fig. 1). A. muciniphila is an intestinal symbiotic bacterium whose abundance is decreased in many pathologic conditions such as obesity, type 2 diabetes (15), hypertension (34), and colitis-associated colorectal cancer (35). Dietary factors such as ethanol (36), Western diet (37), and a diet low in fermentable oligosaccharides, disaccharides, monosaccharides, and polyols (38) also decrease the abundance of A. muciniphila. However, the underlying cause of decreased A. muciniphila remains poorly understood and needs further exploration.

Previous studies have shown that gut microbiota can regulate macrophages in the tumor microenvironment (33, 39). As an essential innate immune population for maintaining homeostasis, macrophages displayed high plasticity (40). Typically, macrophages in the tumor microenvironment can be switched between M1 and M2 phenotypes, and the “phenotype switch” of macrophages is considered a potential therapeutic target for regulating tumorigenesis (41). Here, we demonstrated that tumor burden was alleviated by A. muciniphila through M1-like TAM response. Furthermore, we validated a direct effect of A. muciniphila on TAMs in vitro (Fig. 2). M1-like TAMs are characterized as proinflammatory and displaying antitumor activity (42). M1-like TAMs highly expresses iNOS, and produce large amounts of proinflammatory cytokines, including TNFα, IL12, and IL23 (43). One study reported that A. muciniphila enhances the proportion of CD8⁺ CTLs in mesenteric lymph nodes and the colon during AOM-DSS induced carcinogenesis (35). Our study showed that A. muciniphila had no influence on the proportion of CD8⁺ CTLs in the colon from Apc^Min/+ mice.

TAM-based immunotherapy includes blocking TAM infiltration or promoting M1-like TAM responses (44, 45). One approach to TAM-based immunotherapy is to directly eliminate all macrophages in the microenvironment, such as by blocking the CCL2 signal to inhibit macrophage chemotaxis (46), or to use clodronate liposome to eliminate macrophages in vivo (47). Our study found that A. muciniphila inhibited the proliferation of colorectal cancer cells in the presence of macrophages (Fig. 2), suggesting that A. muciniphila–trained M1-like TAMs might provide a novel therapeutic target in the tumor microenvironment.

Multiple signaling pathways have been implicated in M1-like macrophage response, such as the PI3K/Akt pathway (24), PPARs (25) and the mTOR pathway (26). We discovered that knockout of NLRP3 significantly impaired the A. muciniphila–mediated M1-like macrophage response and abrogated the tumor suppressive effect of A. muciniphila in vivo and in vitro. We thus established that NLRP3 is essential for the crosstalk between A. muciniphila and macrophage in colorectal tumorigenesis (Figs. 3 and 4). NLRP3 is an important part of the NALP3 inflammasome. Activated NLRP3 can recruit caspase-1, which cleaves pro-IL1β into the mature IL1β (p17), thereby participating in pathogen identification and immune response (48). Previous studies show that high NLRP3 expression in macrophages promotes metastasis of colon tumors to the liver (49). NLRP3 is activated in Kupffer cells, which recruits NK cells to kill tumor cells (27). Collectively, the effect of NLRP3 on colorectal cancer could be cell type– and context-dependent.

TLRs are an important component of host defense mediated by the innate immune system (50). TLR4-LPS interactions are thought to be a common pathway for sensing gram-negative bacteria, and can activate the NF-κB pathway (51, 52). However, some studies suggest that the LPS of A. muciniphila only has weak stimulatory effect on TLR4 (14). A. muciniphila is reported to regulate cell glucose and lipid metabolism by TLR2 (29). We thus hypothesized that TLR2 might act as the receptor through which A. muciniphila induced macrophage M1 polarization. Indeed, we observed that TLR2 was increased after A. muciniphila treatment and that TLR2 inhibition significantly impaired the A. muciniphila–mediated M1-like macrophage response in vitro (Fig. 5). Consistent with our findings, TLR2 plays an important role in the innate immune response by recognizing microbial lipoproteins and lipopeptides (53).

In conclusion, our findings provide what we believe to be novel insights into the molecular mechanisms for the regulation of macrophage by A. muciniphila in the tumor microenvironment (Fig. 7). Our work provides evidence for a probiotic role of A. muciniphila in colorectal cancer, which could be harnessed to regulate intestinal immune homeostasis for colorectal cancer prevention and treatment.

No disclosures were reported.

L. Fan: Investigation, writing–original draft. C. Xu: Resources, validation. Q. Ge: Conceptualization. Y. Lin: Investigation. C.C. Wong: Writing–review and editing. Y. Qi: Software. B. Ye: Resources. Q. Lian: Resources. W. Zhuo: Conceptualization. J. Si: Project administration. S. Chen: Funding acquisition, investigation. L. Wang: Supervision, writing–review and editing.

The authors thank Xin Wang and Dr. Wei Liu (Zhejiang Academy of Agricultural Sciences, China) for providing A. muciniphila and culture method instruction. They thank Di Wang (Institute of Immunology, Zhejiang University School of Medicine) for providing NLRP3^tm1Bhk (NLRP3^−/−) mice. The authors also appreciate Dr. Pin Wu (Cancer Institute, Second Affiliated Hospital, Zhejiang University School of Medicine) for helpful discussion and revisiting the immune phenotype. This work was supported by the National Natural Science Foundation of China (grant number 82072623, to L. Wang) and Natural Science Foundation of Zhejiang (grant number LGD21H160002, to S. Chen), and was partly funded by National Key R&D Program of China (grant number 2016YFC1303200).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.