Emerging research has revealed regulation of colorectal cancer metabolism by bacteria. Fusobacterium nucleatum (Fn) plays a crucial role in the development of colorectal cancer, however, whether Fn infection modifies metabolism in patients with colorectal cancer remains unknown. Here, LC-MS/MS-based lipidomics identified the upregulation of cytochrome P450 monooxygenases, primarily CYP2J2, and their mediated product 12,13-EpOME in patients with colorectal cancer tumors and mouse models, which increased the invasive and migratory ability of colorectal cancer cells in vivo and in vitro by regulating the epithelial–mesenchymal transition (EMT). Metagenomic sequencing indicated a positive correlation between increased levels of fecal Fn and serum 12,13-EpOME in patients with colorectal cancer. High levels of CYP2J2 in tumor tissues also correlated with high Fn levels and worse overall survival in patients with stage III/IV colorectal cancer. Moreover, Fn was found to activate TLR4/AKT signaling, downregulating Keap1 and increasing NRF2 to promote transcription of CYP2J2. Collectively, these data identify that Fn promotes EMT and metastasis in colorectal cancer by activating a TLR4/Keap1/NRF2 axis to increase CYP2J2 and 12,13-EpOME, which could serve as clinical biomarkers and therapeutic targets for Fn-infected patients with colorectal cancer.

Significance:

This study uncovers a mechanism by which Fusobacterium nucleatum regulates colorectal cancer metabolism to drive metastasis, suggesting the potential biomarker and therapeutic utility of the CYP2J2/12,13–EpOME axis in Fn-infected patients.

Colorectal cancer is one of the most common cancers worldwide (1). There are accumulating evidence suggesting that gut microbiota contributes to the tumorigenesis and progression of colorectal cancer, potentially through proinflammatory responses, microbial metabolites, and interference with cancer cells in tumor microenvironment (2). For instance, recent studies have identified some specific colorectal cancer-related oncogenic bacteria such as Fusobacterium nucleatum (Fn), Peptostreptococcus anaerobius, certain strains of Escherichia coli and Bacteroides fragilis (3), which not only modulate tumor immunity but also promote cancer by mediating abnormal intracellular biosynthesis (4–6). However, the mechanisms underlying gut bacteria in colorectal cancer development have not been fully understood yet. Therefore, investigating the novel oncogenic roles and associated mechanisms of gut microbiota will undoubtedly benefit the discovery of novel personalized diagnostic or therapeutical strategies in colorectal cancer.

Recently, emerging studies have suggested specific bacteria promote or inhibit the occurrence and development of colorectal cancer through metabolism alteration. A recent study has shown that sulfur-metabolizing bacteria can convert sulfur from the diet into genotoxic hydrogen sulfide, which may be correlated with colorectal cancer progression (7). Holdemanella biformis in a human setting can inhibit the proliferation of colorectal cancer cells by producing short-chain fatty acids (8). Previously, using the miR-21-knockout mice model, we proved Fn promotes the proliferative and migratory ability of colorectal cancer cells through TLR4/MYD88/NF-κB/miR-21 pathway and serves as a prognostic biomarker for patients with colorectal cancer (4, 9). Despite our accumulating work regarding Fn in colorectal cancer, whether it exerts its oncogenic role through regulating metabolism-related mechanisms remains unknown.

The alteration of metabolism usually results in the changes in metabolome, which could be qualitatively and/or quantitatively monitored by metabolomics, a cutting-edge technique of omics approaches. Recently, metabolomics has been employed to investigate the metabolic changes in colorectal cancer (10, 11). Here, by orchestratedly using LC-MS/MS-based targeted metabolomics, metagenomic sequencing, cellular assays, and germ-free gene-knockout mice model, we reported that Fn infection upregulates cytochrome P450 2J2 (CYP2J2)/12,13-epoxyoctadecenoic acid (12,13-EpOME) axis in colorectal cancer via TLR4/Keap1/NRF2 signaling pathway, which finally promotes the development of colorectal cancer. This study not only extends our knowledge of Fn in the colorectal cancer field but also provides novel clinical biomarkers and potential therapeutic targets for Fn-infected patients with colorectal cancer.

Patients and specimen collection

This study was approved by the Institutional Review Board for Clinical Research of Shanghai Tenth People's Hospital affiliated to Tongji University. Written informed consent was obtained from all patients and healthy controls. The serum, feces, and colonic tissue samples were collected at Shanghai Tenth People's Hospital affiliated to Tongji University from 2017 to 2019. Thirty age-matched healthy controls and 30 pre-operative colorectal cancer serum/feces samples were used for metabolomics and metagenomics screening. Eighteen fresh-frozen colonic tissues from healthy controls (by endoscopy) and 141 fresh-frozen tumor samples from patients with colorectal cancer (by surgery) were used to analyze and validate the target gene expression correlation. The information of patients and controls was provided in Supplementary Table S1.

Culture conditions

The Fn strain was obtained from ATCC (F nucleatum subsp. nucleatum ATCC 25586) and was cultured following the protocol described previously (4). In brief, Fn was grown in Columbia blood agar supplemented with 5 μg/mL haemin, 5% defibrinated sheep blood (5%), and 1 μg/mL vitamin K1 (Sigma-Aldrich) in an anaerobic glove box with 85% N2, 10% H2, and 5% CO2 at 37°C. The colorectal cancer cell lines HCT116 (RRID:CVCL_0291), LoVo (RRID:CVCL_0399), SW480 (RRID:CVCL_0546), and Caco2 (RRID:CVCL_0025) and the human normal colonic fibroblast cell line CCD-18Co (RRID:CVCL_2379) were provided by the ATCC. The human normal colon epithelial cell line NCM460 (RRID:CVCL_0460) was obtained from the INCELL Corporation. All the cell lines were verified by short-tandem repeat analysis and cultured in appropriate conditions as described previously (4, 12). Cells were also tested for mycoplasma using the MycoAlert Mycoplasma Detection Kit (Lonza). All experiments were done with cell lines cultivated less than 20 passages since their procurement.

Plasmids, lentivirus short-hairpin RNA construction, and transfection

To stably knockdown CYP2J2 in the LoVo cell line, the synthesized DNA oligos containing short hairpin RNAs (shRNA) sequences were cloned into pLVX-shRNA2-Puro vectors and verified by DNA sequencing. Then, the lentivirus was synthesized and transfected into HEK-293T cells using Lipofectamine 3000 (Thermo Fisher Scientific). Cells transfected with vehicle vector were used as controls. After incubation for 48 hours, the cultured lentiviral supernatant was collected and purified. LoVo cells were transfected with the recombinant shRNA lentiviruses at the multiplicity of infection (MOI) of 50 for subsequent experiments. Puromycin (final concentration: 5 μg/mL) was used to select stable clones.

To overexpress CYP2J2 in the HCT116 cell line, the insert sequence was constructed and cloned into the GV492 vector (Genechem) and applied to transfected colorectal cancer cells using lentivirus. HCT116 cells were transfected with the recombinant lentiviruses at an MOI of 10 for subsequent experiments. Puromycin (final concentration: 2 μg/mL) was used to select stable clones. Lentivirus for manipulating the expression of NRF2, and Keap1 were synthesized using the pCDH vector. The empty vector pCDH was used as controls. The primer sequences used in plasmid construction are presented in Supplementary Table S9.

Experimental metastasis assay

Experimental protocols were approved by the Institutional Animal Research Committee of Shanghai Tenth People's Hospital and complied with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. HCT116 and LoVo cells were treated with 12,13-EpOME or DMSO vehicle in complete medium for 48 hours. Colorectal cancer cell lines with stable CYP2J2 knockdown were co-cultured with Fn at an MOI of 1000:1 or PBS for 6 hours. Next, the cells were injected into the spleens of 6 to 8 weeks old male Balb/C-nu mice as described (13). Mice were sacrificed at 14 days after tumor cell injection. Liver samples were collected and subjected to histologic examination.

Mouse models

Experimental protocols were approved by the Institutional Animal Research Committee of Shanghai Tenth People's Hospital and complied with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. Cyp2j5 knockout (Cyp2J5/−) mice and their wild-type (WT) littermates were purchased from Nanjing Biomedical Research Institute of Nanjing University. Cyp2j5/− and WT mice were derived into germ-free (GF) conditions by embryo transfers, and breeding colonies were established in the GF facility at Shanghai SLAC Laboratory Animal Co. Ltd. Other WT mice were housed and reared in specific pathogen-free (SPF) and barrier conditions with autoclaved food and water at Shanghai SLAC Laboratory Animal Co. Ltd. The bodyweight of each mouse was measured once a week.

Protocol 1

Fourteen 8- to 9-week-old C57BL/6J mice were divided into two groups at random: control (n = 7) and colorectal cancer (n = 7). Then, the animals in the colorectal cancer group were treated with a single dose of azoxymethane (AOM, 12 mg/kg body weight, i.p.) followed by 5 successive days of 2.5% dextran sodium sulfate (DSS) in the drinking water, and then were given regular drinking water. At week 9 after the AOM injection, the mice were sacrificed to harvest serum and colon/tumor tissues for analysis. The animals in the control group received an intraperitoneal injection of the same volume of saline.

Protocol 2

For the rAAV-CYP2J2 mouse model (14), 24 8- to 9-week-old C57BL/6J mice were randomized to four groups (six animals/group) as follows: saline control, rAAV-GFP control, rAAV-CYP2J2, and rAAV-CYP2J2 + Linoleic acid (LA). Animals received a single intraperitoneal injection of either saline or rAAV (5 × 1012 vector genomes/mice). Then, animals were treated with AOM/DSS as the above-mentioned protocol. Additional LA was provided in a formulated high LA diet. The composition of the high LA diet and control diet were detailed in Supplementary Table S10. The concentrations of LA and total fat in the diets were measured according to the protocols detailed in the China guidelines GB/T 21514–2008 and GB/T 6433–2006, respectively.

Protocol 3

Thirty-two 8-week-old germ-free C57BL/6J Cyp2j5−/− and WT mice were divided into six groups at random: WT control (n = 4), WT colorectal cancer (n = 6), WT Fn (n = 6), KO control (n = 4), KO colorectal cancer (n = 6), and KO Fn (n = 6). Each group included equal male and female mice. Then, animals were treated with AOM/DSS as the above-mentioned protocol. PBS-resuspended 109 colony-forming units of Fn or PBS were administrated into germ-free Cyp2j5−/− and WT mice by gavage every day. At the indicated time intervals, the mice were anesthetized with 40 mg/kg pentobarbital sodium and sacrificed.

Statistical analysis

Differences in the quantitative data between two groups were performed using the unpaired or paired two-tailed Student t test or Mann–Whitney U test, where appropriate. Comparisons of means among multiple groups were determined by one-way ANOVA followed by Tukey's or Games-Howell post-hoc comparison test. The relationships between the abundance of Fn and the expression level of CYP2J2/12,13-EpOME were analyzed by using linear regression. The associations between the patient categorical characteristics were analyzed using Pearson χ2 test or Fisher exact test, where appropriate. Overall survival was defined as the time interval from surgery to the date of death or the last follow-up. The survival curves were constructed using Kaplan–Meier model and log-rank tests were used to determine the statistical difference. P values < 0.05 were designated as significantly different (*, P < 0.05; **, P < 0.01; ***, P < 0.001). FDR adjusted P values were used in metagenomics analysis. All statistical analyses were done using GraphPad Prism 6 software (GraphPad lnc., RRID:SCR_002798) or IBM SPSS Statistics 20.0 software (IBM lnc., SPSS, RRID:SCR_002865).

See Supplementary Materials and Methods for details on Metagenomic sequencing, Assessment of bioactive lipids by LC/MS-MS, transwell migration and invasion assay, IHC staining and scoring, immunofluorescence, Western blot, qPCR, RNA sequencing, recombinant adeno-associated virus vector production, dual‐luciferase reporter assay, chromatin immunoprecipitation, and FISH.

Data and materials availability

All the RNA sequencing data were submitted to the National Center for Biotechnology Information Sequence Read Archive (accession number SRP318879). Additional data related to this paper may be requested from the authors. Email: qinhuanlong@tongji.edu.cn (H.Q.).

CYP monooxygenases and EpOMEs are increased in colorectal cancer patients and AOM/DSS-induced colorectal cancer mice

To discover the dominant changes in the metabolome of oxylipin profile in colorectal cancer, an LC-MS/MS-based targeted metabolomics was utilized to analyze about 50 metabolites of polyunsaturated fatty acids (PUFA) in the serum of 30 pre-operative patients with colorectal cancer and 30 age-matched healthy controls (Fig. 1A, cohort 1, Supplementary Table S1). As a result, 33 metabolites were detected and were summarized in Supplementary Table S2. Orthogonal partial least squares discriminant analysis (OPLS-DA) of these data resulted in a visualized separation of the colorectal cancer group from the controls (Fig. 1B). S-plot analysis showed that three metabolites, including CYP monooxygenases-derived 12,13- and 9,10-EpOME, and lipoxygenases-derived 12-hydroxyeicosatetraenoic acid (12-HETE), were the major compounds contributing to the difference between patients with colorectal cancer and healthy control subjects (Fig. 1C). The concentrations of both EpOMEs were significantly elevated in the serum of patients with colorectal cancer, whereas no significant difference was observed for 12-HETE (Fig. 1D; Supplementary Table S2). Then, qPCR was performed to detect the expression of the primary genes responsible for EpOME production, including CYP2C8, CYP2C9, CYP2C18, CYP2C19, and CYP2J2 in colorectal cancer tissues and normal colonic tissues (cohort 2, Supplementary Table S1). The mRNA expression of CYP2J2 and CYP2C8 was significantly upregulated in colorectal cancer tissues when compared with those of controls and CYP2J2 was the highest expressed one (Fig. 1E; Supplementary Fig. S1A). Also, the results of Western blot analysis were in accordance with the major results of qPCR (Fig. 1F; Supplementary Fig. S1B). In addition, qPCR was used to detect the mRNA expression of the five CYPs genes in human colorectal cancer cell lines (HCT116, LoVo, SW480, and Caco2), normal human colon epithelial cell line (NCM460), and the human normal colonic fibroblast cell line CCD-18Co. The result showed CYP2J2 was abundant in HCT116, LoVo, SW480 cell lines, but deficient in Caco-2, NCM460, and CCD-18Co cell lines (Fig. 1G). The results of the other four CYPs genes were presented in Supplementary Figs. S1C–S1F.

Figure 1.

CYP monooxygenases and EpOMEs are increased in patients with colorectal cancer and AOM/DSS-induced colorectal cancer mice. A, Scheme of the clinical study. B, OPLS-DA analysis score scatter plots for metabolic profiles of control subjects and patients with colorectal cancer (n = 30 per group). C, S-plot of the OPLS-DA model of the concentrations of eicosanoids in control subjects and patients with colorectal cancer (n = 30 per group). D, Concentrations of 9,10-EpOME and 12,13-EpOME in the serum of control subjects and patients with colorectal cancer (n = 30 per group; ***, P < 0.001, unpaired Student t test). E, Gene expression of CYP monooxygenases in the colon tissues of control subjects and tumor tissues of patients with colorectal cancer (n = 17–18 per group; **, P < 0.01, unpaired Student t test). F, Western blotting analysis of CYP monooxygenases expression in the colon tissues of control subjects and tumor tissues of patients with colorectal cancer (n = 5 per group). G, Gene expression of CYP2J2 in human colorectal cancer cells (HCT116, LoVo, SW480, CaCO2), normal human colon epithelial cell (NCM460), and the human normal colonic fibroblast cell line CCD-18Co. H, Scheme of animal experiment. I and J, Concentrations of 12,13-EpOME and 9,10-EpOME in the serum (I) and colon/tumor tissues (J) of control healthy mice and colorectal cancer mice (n = 6–7 per group; **, P < 0.01, unpaired Student t test). K and L, qPCR (K) and Western blotting analysis (L) of Cyp2j5 in the colon tissues of control healthy mice and tumor tissues of colorectal cancer mice (n = 5–7 per group; **, P < 0.01, unpaired Student t test). The results are expressed as mean ± SEM.

Figure 1.

CYP monooxygenases and EpOMEs are increased in patients with colorectal cancer and AOM/DSS-induced colorectal cancer mice. A, Scheme of the clinical study. B, OPLS-DA analysis score scatter plots for metabolic profiles of control subjects and patients with colorectal cancer (n = 30 per group). C, S-plot of the OPLS-DA model of the concentrations of eicosanoids in control subjects and patients with colorectal cancer (n = 30 per group). D, Concentrations of 9,10-EpOME and 12,13-EpOME in the serum of control subjects and patients with colorectal cancer (n = 30 per group; ***, P < 0.001, unpaired Student t test). E, Gene expression of CYP monooxygenases in the colon tissues of control subjects and tumor tissues of patients with colorectal cancer (n = 17–18 per group; **, P < 0.01, unpaired Student t test). F, Western blotting analysis of CYP monooxygenases expression in the colon tissues of control subjects and tumor tissues of patients with colorectal cancer (n = 5 per group). G, Gene expression of CYP2J2 in human colorectal cancer cells (HCT116, LoVo, SW480, CaCO2), normal human colon epithelial cell (NCM460), and the human normal colonic fibroblast cell line CCD-18Co. H, Scheme of animal experiment. I and J, Concentrations of 12,13-EpOME and 9,10-EpOME in the serum (I) and colon/tumor tissues (J) of control healthy mice and colorectal cancer mice (n = 6–7 per group; **, P < 0.01, unpaired Student t test). K and L, qPCR (K) and Western blotting analysis (L) of Cyp2j5 in the colon tissues of control healthy mice and tumor tissues of colorectal cancer mice (n = 5–7 per group; **, P < 0.01, unpaired Student t test). The results are expressed as mean ± SEM.

Close modal

To further confirm the findings from clinical subjects, AOM/DSS-induced colorectal cancer model was constructed using C57BL/6 mice. First, the targeted LC/MS-MS was used to compare the 9,10- and 12,13-EpOMEs in the serum and tumor/colon tissues of healthy mice and colorectal cancer mice (Fig. 1H). As a result, the concentrations of 9,10- and 12,13-EpOME were significantly increased in both the serum and tumor tissues of colorectal cancer mice as compared with controls (Fig. 1I and J). Then, the mRNA expression of the major genes responsible for EpOME production was detected in the tumor tissues of colorectal cancer mice and normal colon tissues of healthy mice. The expression of Cyp2j5 (corresponding to human CYP2J2) and other genes (Cyp2c39, Cyp2c65, Cyp2c40, Cyp2j13, and Cyp2j6) were significantly upregulated in the tumor tissues of colorectal cancer mice as compared with those of controls (Fig. 1K; Supplementary Fig. S1G). The following Western blot confirmed the upregulation of Cyp2j5 at the protein level in tumor tissues of colorectal cancer mice (Fig. 1L; Supplementary Fig. S1H). Collectively, our results highlight that EpOMEs and CYP2J2 (the major enzyme accounting for EpOMEs production), are increased in colorectal cancer serum/tissues from both clinical samples and mouse model.

CYP2J2 promotes the invasion and migration of colorectal cancer cells in vivo and in vitro through epithelial–mesenchymal transition

To clarify the oncogenic role of CYP2J2 on colorectal cancer cells, overexpression of CYP2J2 in colorectal cancer cell line HCT116 and knockdown of CYP2J2 in colorectal cancer cell line LoVo, were first performed respectively (Fig. 2A). In the invasion and migration assays, CYP2J2 overexpression increased the number of invasive and migratory HCT116 cells, whereas it was opposite for CYP2J2 knockdown in LoVo cells (Fig. 2B and C). Then, the expressions of epithelial–mesenchymal transition (EMT)–related proteins (vimentin, Snail, and E-cadherin) were detected. Immunofluorescence staining and Western blot analysis showed that CYP2J2 overexpression upregulated vimentin and Snail, but downregulated E-cadherin in HCT-116 cells, whereas it was opposite for CYP2J2 knockdown in LoVo cells (Fig. 2D and E). Furthermore, phalloidin staining for F-actin revealed that CYP2J2 overexpression contributed to spindle-like mesenchymal morphology in HCT-116 cells, whereas CYP2J2 knockdown LoVo cells looked mostly like epithelial cobblestones in appearance (Fig. 2F). Moreover, a murine model of liver metastasis was constructed through injecting colorectal cancer cells into the spleens of Balb/c-nu mice. The images of representative harvested livers and hematoxylin-eosin (H&E) staining for metastasis were shown in Fig. 2G and H. The result demonstrated CYP2J2 overexpression increased the metastasis area (Fig. 2I), whereas it is opposite for CYP2J2 knockdown. Collectively, our results suggest that CYP2J2 promotes invasion and migration of colorectal cancer cells through activating EMT.

Figure 2.

CYP2J2 promotes the invasion and migration of colorectal cancer cells in vivo and in vitro. A, The representative images of Western blot showing upregulated or downregulated CYP2J2 expression in HCT116 or LoVo cells. OE, overexpression of CYP2J2; KD, knockdown of CYP2J2. B and C, Migration and invasion assays (200×) for HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their matched control group (n = 5 per group; ***, P < 0.001, unpaired Student t test). D, IF assay of E-cadherin and vimentin expressions (400×) in HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their matched control group. E, EMT-related protein expression in HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their matched control group. F, Representative images of the cytoskeleton in HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their matched control group. G and H, Representative images of harvested livers (G) and H&E staining (H) for metastasis obtained from mice that received spleen injection with HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their control cells (n = 7–8 per group). I, The percentages of liver surfaces covered with macrometastatic nodules were quantified (n = 7–8 per group; *, P < 0.05; ***, P < 0.001, unpaired Student t test). The results are expressed as mean ± SEM.

Figure 2.

CYP2J2 promotes the invasion and migration of colorectal cancer cells in vivo and in vitro. A, The representative images of Western blot showing upregulated or downregulated CYP2J2 expression in HCT116 or LoVo cells. OE, overexpression of CYP2J2; KD, knockdown of CYP2J2. B and C, Migration and invasion assays (200×) for HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their matched control group (n = 5 per group; ***, P < 0.001, unpaired Student t test). D, IF assay of E-cadherin and vimentin expressions (400×) in HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their matched control group. E, EMT-related protein expression in HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their matched control group. F, Representative images of the cytoskeleton in HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their matched control group. G and H, Representative images of harvested livers (G) and H&E staining (H) for metastasis obtained from mice that received spleen injection with HCT116-CYP2J2OE, LoVo-CYP2J2KD, and their control cells (n = 7–8 per group). I, The percentages of liver surfaces covered with macrometastatic nodules were quantified (n = 7–8 per group; *, P < 0.05; ***, P < 0.001, unpaired Student t test). The results are expressed as mean ± SEM.

Close modal

12,13-EpOME promotes colorectal cancer invasion, migration, and tumorigenesis

To investigate whether EpOMEs are responsible for the oncogenic effect of CYP2J2, the concentration of 9,10- and 12,13-EpOME was first detected in the culture supernatant of colorectal cancer cells after CYP2J2 overexpression or knockdown. As a result, the concentration of EpOMEs, especially 12,13-EpOME, was significantly increased in CYP2J2-overexpressed HCT116 cells, while it was significantly decreased in CYP2J2-knockdown LoVo cells as compared with the controls (Supplementary Fig. S2A). Then, colorectal cancer cells were incubated with 12,13-EpOME to investigate whether it exerts similar oncogenic roles as CYP2J2 upregulation. As expected, supplementing 12,13-EpOME increased the invasive and migratory ability of both HCT-116 and LoVo cells (Supplementary Figs. S2B and S2C). Supplementing 12,13-EpOME also decreased E-cadherin expression, increased vimentin as well as Snail expression (Supplementary Figs. S2D and S2E), and contributed to spindle-like mesenchymal morphology (Supplementary Fig. S2F). Furthermore, using the liver metastasis model, we found 12,13-EpOME treated colorectal cancer cells were more likely to induce liver metastasis as compared with controls (Supplementary Figs. S2G and S2H). This observation was also supported by liver metastasis area comparison (Supplementary Fig. S2I).

Since LA serves as a substrate for CYP2J2 to product 12,13-EpOME, we further constructed a recombinant adeno-associated virus (rAAV) containing CYP2J2 gene to upregulate CYP2J2 expression in C57BL/6J mice. After dividing mice into four groups: Control, rAAV-GFP, rAAV-CYP2J2, rAAV-CYP2J2 + LA (supplementing mice with additional LA by a high LA diet), colorectal cancer was induced by AOM/DSS treatment. The experimental procedure was shown in Supplementary Fig. S3A. High LA diet treated rAAV-CYP2J2 group were more likely to have anorectal prolapse (33.3% vs. 16.7%, 0%, 0%) and gut dilatation (83.3% vs. 50%, 16.7%, 33.3%) as compared with rAAV-CYP2J2 group, rAAV-GFP group, and control group (Supplementary Fig. S3B). The representative images of harvested intestines, H&E staining, and Ki-67 staining for tumors were shown in Supplementary Figs. S3C and S3D, respectively. In addition, rAAV-CYP2J2 combined with the treatment of high LA diet have significantly reduced colon length (Supplementary Fig. S3E), more harvested tumors (Supplementary Fig. S3F) as well as tumor burden (Supplementary Fig. S3G). Most importantly, the concentration of 12,13-EpOME was significantly increased in the serum and tumors of LA treated mice as compared with controls (Supplementary Figs. S3H and S3I). Taken together, these results suggested 12,13-EpOME, as an oncogenic metabolite of CYP2J2, promoted the development of colorectal cancer.

CYP2J2/12,13-EpOME axis is clinically relevant with fusobacterium nucleatum infection in patients with colorectal cancer

Accumulating evidences demonstrate that intestinal bacteria interact with colorectal cancer host genes (15). For clarifying whether the oncogenic role of CYP2J2 and its metabolite 12,13-EpOME are relevant with gut microbiota, metagenomic sequencing was first performed on the stool samples from patients with colorectal cancer and healthy controls in Cohort 1. The details of the generated metagenomic reads were provided in Supplementary Table S3. The species accumulation curve indicated the adequacy of sampling efforts (Supplementary Fig. S4A) and the Venn diagram displayed the shared or unique operational taxonomic units of patients with colorectal cancer and healthy controls (Supplementary Fig. S4B). To display microbiome differences between groups, β diversity was calculated using Bray–Curtis distance, and principal coordinates analysis (PCoA) was performed. The results presented a significantly separated distribution between patients with colorectal cancer and healthy controls (Supplementary Fig. S4C). Then, the discrepant species were selected using the criteria of log20 (fold change) > 1 and P-value < 0.01. In the comparative analysis, Fusobacterium nucleatum (Fn), Clostridium_sp_CAG_567, Xanthophyllomyces dendrorhous, Prevotella_sp_CAG_386, Bacillus_sp_CPSM8, and Morganella_sp_EGD_HP17 were found to be enriched in patients with colorectal cancer, whereas Firmicutes_bacterium_CAG_41 was deficient, as compared with those of healthy controls (Fig. 3A). The details of the mentioned bacteria were provided in Fig. 3B. Especially, the abundance of Fn is approximately 24.67 times more than that of healthy controls (Supplementary Fig. S4D). More interestingly, the stool Fn level was significantly positively correlated with the serum 12,13-EpOME (r = 0.4442; P = 0.0139, Fig. 3C) whereas the correlation of serum 12,13-EpOME with other six bacteria were nonsignificant (Supplementary Table S4).

Figure 3.

CYP2J2/12,13-EpOME axis is clinically relevant to Fusobacterium nucleatum infection in patients with colorectal cancer. A and B, Volcano plot (A) and details of differentially abundant species calculated by P values and fold changes (B). Each dot represents a species. The red dots represent significantly increased species, and the green dots represent significantly decreased species in patients with colorectal cancer, compared with healthy controls (n = 30 per group). The Wilcoxon rank-sum test was used to test for differences. C, The correlation of fecal Fn relative abundance and serum 12,13-EpOME level in patients with colorectal cancer (n = 30; *, P < 0.05 by two-tailed nonparametric Spearman correlation). D, Representative images of FISH staining of Fn (red) and CYP2J2 (green) in the normal mucosa and colorectal cancer tissues. E, Expression of CYP2J2 mRNA in colorectal cancer samples with high (n = 23) or low (n = 22) Fn relative abundance, and their correlation was analyzed by Spearman rank correlation test. F and G, Representative IHC images of CYP2J2 expression in colorectal cancer tissues with positive (n = 44) or negative (n = 34) Fn (F), and their correlation in colorectal cancer samples was analyzed by Spearman rank correlation test in the validation cohort (G). H, The Fn positive rate and CYP2J2 expression of colorectal cancer tissues in different TNM stages (n = 78). I, Kaplan–Meier survival curves for 228 patients with colorectal cancer within III/IV stage from the TCGA database stratified by CYP2J2 expression [log-rank (Mantel–Cox) test]. CRC, colorectal cancer.

Figure 3.

CYP2J2/12,13-EpOME axis is clinically relevant to Fusobacterium nucleatum infection in patients with colorectal cancer. A and B, Volcano plot (A) and details of differentially abundant species calculated by P values and fold changes (B). Each dot represents a species. The red dots represent significantly increased species, and the green dots represent significantly decreased species in patients with colorectal cancer, compared with healthy controls (n = 30 per group). The Wilcoxon rank-sum test was used to test for differences. C, The correlation of fecal Fn relative abundance and serum 12,13-EpOME level in patients with colorectal cancer (n = 30; *, P < 0.05 by two-tailed nonparametric Spearman correlation). D, Representative images of FISH staining of Fn (red) and CYP2J2 (green) in the normal mucosa and colorectal cancer tissues. E, Expression of CYP2J2 mRNA in colorectal cancer samples with high (n = 23) or low (n = 22) Fn relative abundance, and their correlation was analyzed by Spearman rank correlation test. F and G, Representative IHC images of CYP2J2 expression in colorectal cancer tissues with positive (n = 44) or negative (n = 34) Fn (F), and their correlation in colorectal cancer samples was analyzed by Spearman rank correlation test in the validation cohort (G). H, The Fn positive rate and CYP2J2 expression of colorectal cancer tissues in different TNM stages (n = 78). I, Kaplan–Meier survival curves for 228 patients with colorectal cancer within III/IV stage from the TCGA database stratified by CYP2J2 expression [log-rank (Mantel–Cox) test]. CRC, colorectal cancer.

Close modal

Because CYP2J2 is responsible for 12,13-EpOME production, we next tested whether CYP2J2 expression is correlated with Fn. The FISH was first used to detect the Fn infection (red) and CYP2J2 expression (green) in colorectal cancer tissues and normal colonic tissues (cohort 2). As shown in Fig. 3D, compared with normal colonic tissues, Fn was found to invade colorectal cancer tissues and was frequently accompanied by CYP2J2 overexpression. To further clarify the correlation between Fn and CYP2J2, the Fn level and mRNA expression of CYP2J2 were detected in 45 fresh-frozen colorectal cancer tissues (cohort 3, Supplementary Table S1). The samples were divided into Fn high-level group (n = 23) and Fn low-level group (n = 22) using the median of Fn genomic DNA amount detected by qPCR as a cut-off. The correlation analysis revealed the Fn level and CYP2J2 expression were significantly correlated in colorectal cancer tissues (P = 0.0472, Fig. 3E). To validate this result, a validation cohort enrolling 78 patients with colorectal cancer was employed (cohort 4, Supplementary Table S1). The CYP2J2 expression was detected using IHC and the representative staining images were shown in Fig. 3F. Moreover, CYP2J2 expression was significantly correlated with Fn level in colorectal cancer tissues (r = 0.62, P = 0.0001, Fig. 3G). Further analysis suggested positive Fn combined with CYP2J2 overexpression is more frequently observed in colorectal cancer tissues of patients with III/IV stage (Fig. 3H).

To further validate our results, we evaluate the impact of CYP2J2 on the survival rate of patients enrolled in The Cancer Genome Atlas (TCGA) database. Of all 542 colorectal cancer cases in the TCGA, the transcriptomic and clinical data of 228 patients with colorectal cancer at stage III/IV were extracted for further analyses. The basic clinical characteristics of the enrolled patients were presented in Supplementary Table S5. The patients with high expression of CYP2J2 in tumors (based on the gene expression median value) had a significant shorter overall survival than those with low-expression of CYP2J2 (P = 0.042; Fig. 3I). Taken together, these data strongly demonstrated a close clinical correlation between Fn and CYP2J2/12,13-EpOME axis in patients with colorectal cancer.

Fn promotes colorectal cancer development partly by upregulating CYP2J2

To determine whether Fn affected CYP2J2/12,13-EpOME axis, colorectal cancer cells were treated with Fn for 1, 2, 3, and 6 hours for the measurement of CYP2J2 expression. As a result, Fn treatment increased CYP2J2 expression at both mRNA and protein levels in HCT-116 and LoVo cells in a time-dependent manner (Fig. 4A and B). Several other cytochrome P450 monooxygenases were also significantly upregulated by Fn treatment (Supplementary Figs. S4E and S4F). As expected, Fn treatment also increased the concentration of 9,10- and 12,13-EpOME in the culture supernatants of colorectal cancer cells (Fig. 4C, left). Meanwhile, the concentrations of 12,13-EpOME in the centrifuged pellet of colorectal cancer cells were significantly increased as compared with that in the blank culture medium, centrifuged pellet of Fn, and its medium (Fig. 4C, right).

Figure 4.

Fn upregulates CYP2J2 to promote colorectal cancer invasion and migration. A and B,Fn increased the mRNA (A) and protein (B) expression of CYP2J2 in HCT116 and LoVo cells time dependently (three biological replicates; **, P < 0.01; ***, P < 0.001, one-way ANOVA followed by Tukey or Games-Howell post-hoc comparison test). C, Concentrations of 12,13-EpOME and 9,10-EpOME in the supernatants and pellets of cultured colorectal cancer cells, Fn, and blank culture medium (n = 6 per group; *, P < 0.05; **, P < 0.01; ***, P < 0.001, one-way ANOVA followed by Tukey or Games–Howell post-hoc comparison test). D, Effect of Fn on CYP2J2 expression in LoVo-CYP2J2KD and control cells (three biological replicates; ***, P < 0.001, unpaired Student t test). E and F, The migration, invasion (200×) assay, and EMT marker expression detection (IF; 400×) of Fn-treated LoVo-CYP2J2KD cells and control cells. G, Representative images of harvested livers and H&E staining for metastasis obtained from mice that received spleen injection with Fn-treated LoVo-CYP2J2KD cells and control cells. H, The percentages of liver surfaces covered with macrometastatic nodules were quantified (n = 7 per group; ***, P < 0.001, unpaired Student t test). The results are expressed as mean ± SEM.

Figure 4.

Fn upregulates CYP2J2 to promote colorectal cancer invasion and migration. A and B,Fn increased the mRNA (A) and protein (B) expression of CYP2J2 in HCT116 and LoVo cells time dependently (three biological replicates; **, P < 0.01; ***, P < 0.001, one-way ANOVA followed by Tukey or Games-Howell post-hoc comparison test). C, Concentrations of 12,13-EpOME and 9,10-EpOME in the supernatants and pellets of cultured colorectal cancer cells, Fn, and blank culture medium (n = 6 per group; *, P < 0.05; **, P < 0.01; ***, P < 0.001, one-way ANOVA followed by Tukey or Games–Howell post-hoc comparison test). D, Effect of Fn on CYP2J2 expression in LoVo-CYP2J2KD and control cells (three biological replicates; ***, P < 0.001, unpaired Student t test). E and F, The migration, invasion (200×) assay, and EMT marker expression detection (IF; 400×) of Fn-treated LoVo-CYP2J2KD cells and control cells. G, Representative images of harvested livers and H&E staining for metastasis obtained from mice that received spleen injection with Fn-treated LoVo-CYP2J2KD cells and control cells. H, The percentages of liver surfaces covered with macrometastatic nodules were quantified (n = 7 per group; ***, P < 0.001, unpaired Student t test). The results are expressed as mean ± SEM.

Close modal

Because we have discovered a strong link between Fn and colorectal cancer, we speculated Fn may exert an oncogenic role partly through CYP2J2/12,13-EpOME axis. To test whether CYP2J2 is crucial for the oncogenic role of Fn, CYP2J2 knockdown LoVo cells were treated with Fn for 6 hours for detecting the expression of CYP2J2 (Fig. 4D). The results showed that knockdown of CYP2J2 expression significantly alleviated the promoting role of Fn in the invasion and migration of colorectal cancer cells in vitro (Fig. 4E). Knockdown of CYP2J2 expression was observed to alleviate Fn induced EMT related molecular phenotypes (Fig. 4E and F). Moreover, using the liver metastasis model (Fig. 4G), knockdown of CYP2J2 expression alleviated the Fn-promoted metastasis of LoVo cells in vivo, as supported by comparison of metastasis area (Fig. 4H).

To confirm whether CYP2J2 is crucial for Fn-mediated colorectal cancer development, a Cyp2j5 knockout (Cyp2j5−/−) germ-free mice model was engineered (Fig. 5A). The knockout efficiency was validated by sequencing for Cyp2j5 (Supplementary Table S6). The germ-free mice were treated with Fn during the AOM/DSS induced tumorigenesis, and the concentration of Cyp2j5 product-12,13-EpOME in the serum of mice was detected for further validation (Fig. 5B and C). Compared with WT and Fn-treated mice, Fn-treated Cyp2j5−/− mice were less likely to undergo anorectal prolapse (Fig. 5D). The representative images of harvested intestines, H&E staining for tumors were shown in Fig. 5E and F, respectively. The results also showed that Fn-treated Cyp2j5−/− mice had more colon length, fewer tumor numbers, and tumor loads than Fn-treated WT mice (Fig. 5G,I). Taken together, these findings suggest Fn promotes colorectal cancer development partly through upregulating CYP2J2/EpOME axis.

Figure 5.

Fn upregulates Cyp2j5 to promote the development of colorectal cancer in a germ-free murine model. A, Representative images of the macroscopic appearance of germ-free (GF)-Cyp2j5WT and GF-Cyp2j5−/− mice. B, Scheme of animal experiment. C, Concentrations of 12,13-EpOME in the serum of GF-Cyp2j5 WT and GF-Cyp2j5−/− mice treated with or without Fn during the AOM/DSS induced tumorigenesis (n = 4–6 per group). *, P < 0.05; **, P < 0.01, one-way ANOVA followed by Tukey or Games–Howell post-hoc comparison test. D and E, Representative images of the macroscopic appearance of the anus (D) and colon (E) in mice. F, Representative images of H&E staining in the tumor tissues. G–I, Quantification of colon length (one-way ANOVA followed by Tukey or Games–Howell post-hoc comparison test; G), tumor numbers (Mann–Whitney U test; H), and tumor burden (I) in mice (n = 4–6 per group). *, P < 0.05; **, P < 0.01. The results are expressed as mean ± SEM. CRC, colorectal cancer; ns, nonsignificant.

Figure 5.

Fn upregulates Cyp2j5 to promote the development of colorectal cancer in a germ-free murine model. A, Representative images of the macroscopic appearance of germ-free (GF)-Cyp2j5WT and GF-Cyp2j5−/− mice. B, Scheme of animal experiment. C, Concentrations of 12,13-EpOME in the serum of GF-Cyp2j5 WT and GF-Cyp2j5−/− mice treated with or without Fn during the AOM/DSS induced tumorigenesis (n = 4–6 per group). *, P < 0.05; **, P < 0.01, one-way ANOVA followed by Tukey or Games–Howell post-hoc comparison test. D and E, Representative images of the macroscopic appearance of the anus (D) and colon (E) in mice. F, Representative images of H&E staining in the tumor tissues. G–I, Quantification of colon length (one-way ANOVA followed by Tukey or Games–Howell post-hoc comparison test; G), tumor numbers (Mann–Whitney U test; H), and tumor burden (I) in mice (n = 4–6 per group). *, P < 0.05; **, P < 0.01. The results are expressed as mean ± SEM. CRC, colorectal cancer; ns, nonsignificant.

Close modal

Fn upregulates CYP2J2 through Keap1/NRF2 cascade via TLR4/AKT signaling pathway

To explore the regulatory mechanisms of Fn on CYP2J2 expression, RNA sequencing was performed in HCT116 and Fn-treated HCT116 cells. Significant differentially expressed genes were selected using the criteria as follows: 1, P value < 0.05; 2, log2 counts per million reads > 2. Gene ontology (GO) functional analysis showed these genes participated in several transmembrane transport and lipid flow-related functions such as transport vesicle membrane, membrane raft polarization, plasma membrane, and oxidation-reduction process (Supplementary Fig. S5A). The representative significantly expressed genes were shown in the heat map (Fig. 6A; Supplementary Table S7). On the basis of sequencing data, bioinformatics analysis from the Kyoto Encyclopedia of Genes and Genomes (KEGG) database, and literature review (4, 16), we hypothesized that Fn is likely to upregulate CYP2J2 expression through Keap1/NRF2 cascade via toll-like receptor (TLR)4/AKT signaling pathway (Fig. 6B). It is well known that Fn can influence TLR4 activation in colorectal cancer (4). To have a better understanding in our model, we directly quantified TLR4 expression and tested the above hypothesis by qPCR. The results revealed that Fn treatment increased the mRNA expression of TLR4 and NRF2, but decreased the expression of Keap1 in HCT-116 and LoVo cells (Fig. 6C). At protein levels, the expression of TLR4, p-AKT, NRF2 were upregulated, whereas that of Keap1 was decreased in Fn treated colorectal cancer cells (Fig. 6D). In addition, overexpression of Keap1 in colorectal cancer cells resulted in downregulation of NRF2 and CYP2J2 (Fig. 6E and F). Finally, knockdown of NRF2 in colorectal cancer cells suppressed the expression of CYP2J2 expression (Fig. 6G and H).

Figure 6.

Fn upregulates CYP2J2 through Keap1/NRF2 cascade via TLR4/AKT signaling pathway. A, Representative differentially expressed genes between Fn-treated and control HCT116 cells (n = 3 per group) in RNA sequencing. B, Bioinformatics analysis from the KEGG database based on sequencing data. C and D, Effect of Fn on the expression of TLR4, Keap1, NRF2, and p-AKT in HCT116 and LoVo cells (three biological replicates; *, P < 0.05; ***, P < 0.001, unpaired Student t test). E and F, mRNA (E) and protein (F) expression of Keap1, NRF2, and CYP2J2 in Keap1 overexpressed HCT116 and LoVo cells (three biological replicates, *, P < 0.05; ***, P < 0.001, unpaired Student t test). G and H, mRNA (G) and protein (H) expression of NRF2, and CYP2J2 in HCT116 and LoVo cells treated by NRF2 knockdown. The results are expressed as mean ± SEM. **, P < 0.01.

Figure 6.

Fn upregulates CYP2J2 through Keap1/NRF2 cascade via TLR4/AKT signaling pathway. A, Representative differentially expressed genes between Fn-treated and control HCT116 cells (n = 3 per group) in RNA sequencing. B, Bioinformatics analysis from the KEGG database based on sequencing data. C and D, Effect of Fn on the expression of TLR4, Keap1, NRF2, and p-AKT in HCT116 and LoVo cells (three biological replicates; *, P < 0.05; ***, P < 0.001, unpaired Student t test). E and F, mRNA (E) and protein (F) expression of Keap1, NRF2, and CYP2J2 in Keap1 overexpressed HCT116 and LoVo cells (three biological replicates, *, P < 0.05; ***, P < 0.001, unpaired Student t test). G and H, mRNA (G) and protein (H) expression of NRF2, and CYP2J2 in HCT116 and LoVo cells treated by NRF2 knockdown. The results are expressed as mean ± SEM. **, P < 0.01.

Close modal

Because NRF2 is a transcription factor, we assumed the TLR4/Keap1/NRF2 pathway can regulate CYP2J2 expression in a promoter-dependent manner. Bioinformatic analysis was exploited to decipher potential NRF2 (NFE2L2) binding sites in upstream of the transcription start site of CYP2J2. Interestingly, the seven-consensus binding sequences for NRF2 in the promoter region (2,000 bp upstream from the transcription start site) of CYP2J2 were identified (Supplementary Fig. S5B; Supplementary Table S8). To determine whether NRF2 can bind to these regions and regulate CYP2J2 expression, two predicted binding sites were randomly selected and luciferase reporter plasmids were generated containing either WT or MT NRF2 binding sites in the CYP2J2 promoter region to examine the direct interaction between NRF2 and CYP2J2 promoter (Fig. 7A and B). The results revealed that NRF2 overexpression at both binding sites increased the activity of the WT CYP2J2 promoter in colorectal cancer cells (Fig. 7C; Supplementary Fig. S5C). Moreover, chromatin immunoprecipitation assay also supported that the binding of NRF2 with the CYP2J2 promoter was inhibited after NRF2 knockdown in colorectal cancer cells (Fig. 7D; Supplementary Fig. S5D). These evidence suggest that Fn treatment activates Keap1/NRF2 cascade via TLR4/AKT signaling pathway, thus transcriptionally upregulates CYP2J2 expression.

Figure 7.

NRF2 regulates CYP2J2 expression in a promoter-dependent manner. A and B, Schematic diagram and details of two representative MT NRF2-binding sites in the CYP2J2 promoter. C, Relative luciferase activities of reporter genes construct driven by the WT or MT CYP2J2 promoter in HCT116 cells (chr1:59892357–59892456; three biological replicates; ***, P < 0.001, one-way ANOVA followed by Tukey or Games–Howell post-hoc comparison test). The results are expressed as mean ± SEM. D, Chromatin immunoprecipitation (ChIP) assay shows that NRF2 knockdown effectively inhibited the binding of NRF2 with the promoter of the CYP2J2 gene in HCT116 cells. E, The illustration of the hypothetical mechanism by which Fn promotes the tumorigenesis, invasion, and migration of colorectal cancer by regulating CYP2J2/12,13-EpOME axis via TLR4/AKT/Keap1/NRF2 signaling pathway.

Figure 7.

NRF2 regulates CYP2J2 expression in a promoter-dependent manner. A and B, Schematic diagram and details of two representative MT NRF2-binding sites in the CYP2J2 promoter. C, Relative luciferase activities of reporter genes construct driven by the WT or MT CYP2J2 promoter in HCT116 cells (chr1:59892357–59892456; three biological replicates; ***, P < 0.001, one-way ANOVA followed by Tukey or Games–Howell post-hoc comparison test). The results are expressed as mean ± SEM. D, Chromatin immunoprecipitation (ChIP) assay shows that NRF2 knockdown effectively inhibited the binding of NRF2 with the promoter of the CYP2J2 gene in HCT116 cells. E, The illustration of the hypothetical mechanism by which Fn promotes the tumorigenesis, invasion, and migration of colorectal cancer by regulating CYP2J2/12,13-EpOME axis via TLR4/AKT/Keap1/NRF2 signaling pathway.

Close modal

This study reported a novel mechanism of Fn as a specific oncogenic bacterium in promoting tumorigenesis and metastasis of colorectal cancer (Fig. 7E). By using metabolomics and microbiomics analyses of clinical samples, we found the levels of Fn, CYP2J2, and its oncogenic metabolite 12,13-EpOME are heavily enriched in patients with colorectal cancer. We then discovered that Fn may promote colorectal cancer development and EMT-mediated metastasis partly through the CYP2J2/12,13-EpOME axis. Finally, the mechanism investigation further revealed Fn infection activated TLR4/AKT/NRF2 signaling pathway to increase CYP2J2 expression at the transcription level, which contributes to the 12,13-EpOME production. These findings collectively suggested that the CYP2J2/12,13-EpOME axis could serve as a promising clinical biomarker and therapeutical target for Fn-infected patients with colorectal cancer.

The mechanisms of specific gut microbiota on tumorigenesis and metastasis remain largely undiscovered although the oncogenic roles of gut microbiota have received intensive attention during the past decade. Previous studies have proved that the altered metabolism of cancer cells results in the transformed tumor microenvironment that promotes a tolerogenic environment with hindered DNA repair, thereby supporting tumor growth (17, 18). Meanwhile, emerging evidence have pointed to the crucial oncogenic role of cellular intrinsic metabolites in bacteria-induced colorectal cancer development. For example, enteroinvasive Escherichia coli and Salmonella dublin can upregulate the expression of the iNOS gene in colorectal cancer cells to produce nitric oxide, which eventually causes nitrite accumulation and stimulates cell oxidative stress (19). Furthermore, the enterotoxigenic Bacteroides fragilis can also upregulate the expression of spermine oxidase (SMO) in colorectal cancer cells, which in turn leads to the generation of SMO-dependent reactive oxygen species and induces DNA damage (20). Fn, as a well-established oncogenic bacterium, has been found to promote the growth, metastasis, and chemotherapy resistance of colorectal cancer (4, 21). However, to our knowledge, whether it could affect the intrinsic metabolites of colorectal cancer cells is poorly investigated.

For addressing this issue, we first selected potential oncogenic metabolites using clinical samples and AOM/DSS-induced colorectal cancer mice. As a result, we found levels of 12,13 and 9,10-EpOMEs were significantly increased in the serum samples of both patients with colorectal cancer and colorectal cancer mice as compared with the healthy controls. In addition, we found the expression of CYP monooxygenases that were responsible for EpOME production was abnormally upregulated in colorectal cancer tissues. The expression of CYP monooxygenases is elevated by hypoxia status, implying that hypoxic tumor microenvironment may contribute to their overexpression in colorectal cancer tissues (22, 23). Furthermore, since the serum level of 12,13-EpOME and the tissue expression of CYP2J2 were the highest in patients with colorectal cancer, we therefore focused on the CYP2J2/12,13-EpOME axis in the following work.

Previous studies have found that EpOMEs were closely associated with various human diseases such as multiple organ failure and adult respiratory distress syndrome (24). Its regulatory gene-CYP2J2 was found to be involved in tumor progression and anticancer therapy resistance (25, 26). To clarify the role of CYP2J2/12,13-EpOME axis in colorectal cancer development, we first knockdown or overexpressed CYP2J2 expression in colorectal cancer cells and found that the concentration of 12,13-EpOME was accordingly changed in the culture supernatant of colorectal cancer cells. The functional assay demonstrated that both CYP2J2 overexpression and 12,13-EpOME treatment dramatically promoted the invasion and migration of colorectal cancer cells in vitro, more importantly, resulting in a mesenchymal phenotype. This observation inspired us to hypothesize that the CYP2J2/12,13-EpOME axis may contribute to the EMT program in colorectal cancer, a well-established molecular mechanism mediating colorectal cancer invasion and metastasis (27). The following EMT-related molecular detection demonstrated both of them decreased E-cadherin expression, but increased vimentin and Snail expression, successfully confirming our hypothesis. Furthermore, using the liver metastasis model, we proved their pro-metastatic role in vivo. LA serves as a crucial substrate for EpOMEs and a recent study have suggested it may participate in the malignant transformation of ulcerative colitis into colorectal cancer (28, 29). We found the LA supplement contributed to the CYP2J2-mediated production of 12,13-EpOME and colon tumorigenesis in the AOM/DSS mice model. Taken together, the above findings collectively suggested CYP2J2/12,13-EpOME axis functioned as a crucial driver in the development of colorectal cancer.

Using metagenomics techniques, we established a close clinical correlation between Fn and 12,13-EpOME in patients with colorectal cancer. In colorectal cancer tissues from the training and validation cohort, we also found Fn level/positivity was significantly correlated with CYP2J2 expression both at mRNA and protein level. More importantly, simultaneous Fn infection and CYP2J2 overexpression in colorectal cancer tissues were more significantly associated with advanced TNM stage than other phenotypes. Finally, we found high CYP2J2 expression was correlated with poor overall survival in TCGA patients with regional or/and distant metastasis, directly supporting our observations of functional assays in vivo and in vitro.

To further clarify how Fn activated the CYP2J2/12,13-EpOME axis in colorectal cancer cells, we performed RNA sequencing in colorectal cancer cells treated with or without Fn. The result and our literature review have pointed to its potential regulation in TLR4/AKT/Keap1/NRF2 signaling pathway. TLR4 is a well-known crucial receptor for recognizing Fn interaction with colorectal cancer cells to activate the intracellular downstream signals (4, 21). The Fn-derived molecule-Autoinducer-2 was found to promote colorectal cancer development by promoting macrophage M1 polarization via ATK-related signaling (30). In addition, previous studies have found that bacterial intervention can regulate the Keap1/NRF2 signaling pathway, and overexpression of NRF2 may participate in tumorigenesis (31–33). In colorectal cancer, Keap1/NRF2 signaling was suggested to be involved in tumor metastasis (34). Furthermore, mounting evidence have suggested TLR-4 signaling regulating inflammation response through NRF2 pathway (35). For validating the speculated regulatory signaling pathway, we downregulated NRF2 expression, and upregulated Keap1 expression respectively in colorectal cancer cells and then detected the expression change of upstream/downstream molecules. The results firmly supported that the CYP2J2 expression was regulated by TLR4/AKT/Keap1/NRF2 signaling. Previous studies have suggested that genes of the CYP family could be transcriptionally activated, we subsequently investigated whether the CYP2J2 gene is transcriptionally regulated by NRF2 gene (36). Our data illustrated that the NRF2 gene could bind with the promoter of the CYP2J2 gene to directly activate its transcription and upregulated its expression, finally contributing to 12,13-EpOME production. Therefore, Fn may exert its oncogenic role partly through TLR4/AKT/Keap1/NRF2 signaling mediated CYP2J2/12,13-EpOME axis. According to the classic “driver-passenger model” by Tjalsma and colleagues (37), Fn was considered to be a “bacterial passenger” that rarely colonized in the normal tissues but preferred the tumor microenvironment. However, our findings and accumulating studies revealed the oncogenic features of Fn in colorectal cancer, according with the concept of “bacterial drivers.” Therefore, the definite role of Fn as a bacterial “passenger” or “driver” in colorectal cancer carcinogenesis is still a matter of debate.

Taken together, our current findings demonstrate that Fn infection activates TLR4/AKT/Keap1/NRF2 signaling to upregulate CYP2J2 expression in colorectal cancer cells, which then increases the production of 12,13-EpOME and finally results in the development of colorectal cancer. Our study not only identified the CYP2J2/12,13-EpOME axis as a novel clinical indicator and therapeutic target for Fn-infected colorectal cancer but also highlighted the crucial role of tumor cell metabolism altered by gut microbiota in the colorectal cancer field.

No disclosures were reported.

C. Kong: Conceptualization, data curation, software, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. X. Yan: Validation, writing–original draft, writing–review and editing. Y. Zhu: Investigation, methodology. H. Zhu: Methodology, writing–review and editing. Y. Luo: Investigation. P. Liu: Writing–review and editing. S. Ferrandon: Methodology, writing–review and editing. M.F. Kalady: Supervision, writing–review and editing. R. Gao: Validation. J. He: Investigation. F. Yin: Validation. X. Qv: Validation. J. Zheng: Resources. Y. Gao: Resources. Q. Wei: Resources. Y. Ma: Conceptualization, supervision, methodology, writing–review and editing. J.-Y. Liu: Conceptualization, data curation, software, supervision, investigation, methodology. H. Qin: Conceptualization, supervision, project administration, writing–review and editing.

The present work was supported by the National Nature Science Foundation of China (nos. 81972221, 81730102, 81902422, 81972709, 81920108026, and 81871964), Emerging Cutting-Edge Technology Joint Research projects of Shanghai (SHDC12017112), Tongji University Subject Pilot Program (1501141201), Special Construction of Integrated Traditional Chinese Medicine and Western Medicine in Shanghai General Hospital (ZHYY-ZXYJHZX-1-201704), Social Development Project of Yangzhou science and technology bureau (YZ2020078), Program of Jiangsu Commission of Health (No. M2020024), Shanghai Sailing Program (18YF1419400), the National Ten Thousand Plan Young Top Talents (for Dr. Yanlei Ma), the Shanghai Young Top Talents (No. QNBJ1701), the Shanghai Science and Technology Development Fund (no. 19410713300), the CSCO-Roche Tumor Research Fund (No. Y-2019Roche-079), and the Fudan University Excellence 2025 Talent Cultivation Plan (for Dr. Yanlei Ma). The authors thank Chaoyang Xiu (analyzed and interpreted of data; Shanghai Tenth People's Hospital, Tongji University) and Hua Gao (analyzed and interpreted of data; Tongji University School of Medicine).

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.

1.
Bray
F
,
Ferlay
J
,
Soerjomataram
I
,
Siegel
RL
,
Torre
LA
,
Jemal
A
. 
Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
.
CA Cancer J Clin
2018
;
68
:
394
424
.
2.
Marchesi
JR
,
Adams
DH
,
Fava
F
,
Hermes
GD
,
Hirschfield
GM
,
Hold
G
, et al
The gut microbiota and host health: a new clinical frontier
.
Gut
2016
;
65
:
330
9
.
3.
Wong
SH
,
Yu
J
. 
Gut microbiota in colorectal cancer: mechanisms of action and clinical applications
.
Nat Rev Gastroenterol Hepatol
2019
;
16
:
690
704
.
4.
Yang
Y
,
Weng
W
,
Peng
J
,
Hong
L
,
Yang
L
,
Toiyama
Y
, et al
Fusobacterium nucleatum increases proliferation of colorectal cancer cells and tumor development in mice by activating TLR4 signaling to NFkappaB, upregulating expression of microRNA-21
.
Gastroenterology
2016
;
152
:
851
66
.
5.
Long
X
,
Wong
CC
,
Tong
L
,
Chu
ESH
,
Ho Szeto
C
,
Go
MYY
, et al
Peptostreptococcus anaerobius promotes colorectal carcinogenesis and modulates tumour immunity
.
Nat Microbiol
2019
;
4
:
2319
30
.
6.
Tsoi
H
,
Chu
ESH
,
Zhang
X
,
Sheng
J
,
Nakatsu
G
,
Ng
SC
, et al
Peptostreptococcus anaerobius induces intracellular cholesterol biosynthesis in colon cells to induce proliferation and causes dysplasia in mice
.
Gastroenterology
2017
;
152
:
1419
33
.
7.
Nguyen
LH
,
Ma
W
,
Wang
DD
,
Cao
Y
,
Mallick
H
,
Gerbaba
TK
, et al
Association between sulfur-metabolizing bacterial communities in stool and risk of distal colorectal cancer in men
.
Gastroenterology
2020
;
158
:
1313
25
.
8.
Zagato
E
,
Pozzi
C
,
Bertocchi
A
,
Schioppa
T
,
Saccheri
F
,
Guglietta
S
, et al
Endogenous murine microbiota member Faecalibaculum rodentium and its human homologue protect from intestinal tumour growth
.
Nat Microbiol
2020
;
5
:
511
24
.
9.
Shi
C
,
Yang
Y
,
Xia
Y
,
Okugawa
Y
,
Yang
J
,
Liang
Y
, et al
Novel evidence for an oncogenic role of microRNA-21 in colitis-associated colorectal cancer
.
Gut
2016
;
65
:
1470
81
.
10.
Yachida
S
,
Mizutani
S
,
Shiroma
H
,
Shiba
S
,
Nakajima
T
,
Sakamoto
T
, et al
Metagenomic and metabolomic analyses reveal distinct stage-specific phenotypes of the gut microbiota in colorectal cancer
.
Nat Med
2019
;
25
:
968
76
.
11.
Shen
C
,
Xuan
B
,
Yan
T
,
Ma
Y
,
Xu
P
,
Tian
X
, et al
m(6)A-dependent glycolysis enhances colorectal cancer progression
.
Mol Cancer
2020
;
19
:
72
.
12.
Lu
L
,
Xie
R
,
Wei
R
,
Cai
C
,
Bi
D
,
Yin
D
, et al
Integrin α5 subunit is required for the tumor supportive role of fibroblasts in colorectal adenocarcinoma and serves as a potential stroma prognostic marker
.
Mol Oncol
2019
;
13
:
2697
714
.
13.
Yu
HK
,
Kim
JS
,
Lee
HJ
,
Ahn
JH
,
Lee
SK
,
Hong
SW
, et al
Suppression of colorectal cancer liver metastasis and extension of survival by expression of apolipoprotein(a) kringles
.
Cancer Res
2004
;
64
:
7092
8
.
14.
Cai
Z
,
Zhao
G
,
Yan
J
,
Liu
W
,
Feng
W
,
Ma
B
, et al
CYP2J2 overexpression increases EETs and protects against angiotensin II-induced abdominal aortic aneurysm in mice
.
J Lipid Res
2013
;
54
:
1448
56
.
15.
Tomkovich
S
,
Gharaibeh
RZ
,
Dejea
CM
,
Pope
JL
,
Jiang
J
,
Winglee
K
, et al
Human colon mucosal biofilms and murine host communicate via altered mRNA and microRNA expression during cancer
.
mSystems
2020
;
5
:
e00451
19
.
16.
Wang
Y
,
Li
C
,
Cheng
K
,
Zhang
R
,
Narsinh
K
,
Li
S
, et al
Activation of liver X receptor improves viability of adipose-derived mesenchymal stem cells to attenuate myocardial ischemia injury through TLR4/NF-κB and Keap-1/Nrf-2 signaling pathways
.
Antioxid Redox Signal
2014
;
21
:
2543
57
.
17.
Chen
LL
,
Xiong
Y
. 
Tumour metabolites hinder DNA repair
.
Nature
2020
;
582
:
492
4
.
18.
Costa
A
,
Scholer-Dahirel
A
,
Mechta-Grigoriou
F
. 
The role of reactive oxygen species and metabolism on cancer cells and their microenvironment
.
Semin Cancer Biol
2014
;
25
:
23
32
.
19.
Witthoft
T
,
Eckmann
L
,
Kim
JM
,
Kagnoff
MF
. 
Enteroinvasive bacteria directly activate expression of iNOS and NO production in human colon epithelial cells
.
Am J Physiol
1998
;
275
:
G564
71
.
20.
Goodwin
AC
,
Destefano Shields
CE
,
Wu
S
,
Huso
DL
,
Wu
X
,
Murray-Stewart
TR
, et al
Polyamine catabolism contributes to enterotoxigenic Bacteroides fragilis-induced colon tumorigenesis
.
Proc Natl Acad Sci U S A
2011
;
108
:
15354
9
.
21.
Yu
T
,
Guo
F
,
Yu
Y
,
Sun
T
,
Ma
D
,
Han
J
, et al
Fusobacterium nucleatum promotes chemoresistance to colorectal cancer by modulating autophagy
.
Cell
2017
;
170
:
548
63
.
22.
du Souich
P
,
Fradette
C
. 
The effect and clinical consequences of hypoxia on cytochrome P450, membrane carrier proteins activity and expression
.
Expert Opin Drug Metab Toxicol
2011
;
7
:
1083
100
.
23.
Jing
X
,
Yang
F
,
Shao
C
,
Wei
K
,
Xie
M
,
Shen
H
, et al
Role of hypoxia in cancer therapy by regulating the tumor microenvironment
.
Mol Cancer
2019
;
18
:
157
.
24.
Wang
W
,
Yang
J
,
Edin
ML
,
Wang
Y
,
Luo
Y
,
Wan
D
, et al
Targeted metabolomics identifies the cytochrome P450 monooxygenase eicosanoid pathway as a novel therapeutic target of colon tumorigenesis
.
Cancer Res
2019
;
79
:
1822
30
.
25.
Karkhanis
A
,
Hong
Y
,
Chan
ECY
. 
Inhibition and inactivation of human CYP2J2: implications in cardiac pathophysiology and opportunities in cancer therapy
.
Biochem Pharmacol
2017
;
135
:
12
21
.
26.
Allison
SE
,
Chen
Y
,
Petrovic
N
,
Zhang
J
,
Bourget
K
,
Mackenzie
PI
, et al
Activation of ALDH1A1 in MDA-MB-468 breast cancer cells that over-express CYP2J2 protects against paclitaxel-dependent cell death mediated by reactive oxygen species
.
Biochem Pharmacol
2017
;
143
:
79
89
.
27.
Vu
T
,
Datta
PK
. 
Regulation of EMT in colorectal cancer: a culprit in metastasis
.
Cancers
2017
;
9
:
171
.
28.
Hildreth
K
,
Kodani
SD
,
Hammock
BD
,
Zhao
L
. 
Cytochrome P450-derived linoleic acid metabolites EpOMEs and DiHOMEs: a review of recent studies
.
J Nutr Biochem
2020
;
86
:
108484
.
29.
Tang
Q
,
Cang
S
,
Jiao
J
,
Rong
W
,
Xu
H
,
Bi
K
, et al
Integrated study of metabolomics and gut metabolic activity from ulcerative colitis to colorectal cancer: the combined action of disordered gut microbiota and linoleic acid metabolic pathway might fuel cancer
.
J Chromatogr A
2020
;
1629
:
461503
.
30.
Wu
J
,
Li
K
,
Peng
W
,
Li
H
,
Li
Q
,
Wang
X
, et al
Autoinducer-2 of Fusobacterium nucleatum promotes macrophage M1 polarization via TNFSF9/IL-1β signaling
.
Int Immunopharmacol
2019
;
74
:
105724
.
31.
Yu
Y
,
Wang
C
,
Wang
A
,
Yang
W
,
Lv
F
,
Liu
F
, et al
Effects of various feeding patterns of Bacillus coagulans on growth performance, antioxidant response and Nrf2-Keap1 signaling pathway in juvenile gibel carp (Carassius auratus gibelio)
.
Fish Shellfish Immunol
2018
;
73
:
75
83
.
32.
Hos
NJ
,
Ganesan
R
,
Gutierrez
S
,
Hos
D
,
Klimek
J
,
Abdullah
Z
, et al
Type I interferon enhances necroptosis of Salmonella Typhimurium-infected macrophages by impairing antioxidative stress responses
.
J Cell Biol
2017
;
216
:
4107
21
.
33.
Ganan-Gomez
I
,
Wei
Y
,
Yang
H
,
Boyano-Adanez
MC
,
Garcia-Manero
G
. 
Oncogenic functions of the transcription factor Nrf2
.
Free Radic Biol Med
2013
;
65
:
750
64
.
34.
Chang
LC
,
Fan
CW
,
Tseng
WK
,
Hua
CC
. 
Associations between the Nrf2/Keap1 pathway and mitochondrial functions in colorectal cancer are affected by metastasis
.
Cancer Biomark
2020
;
27
:
163
71
.
35.
Mohan
S
,
Gupta
D
. 
Crosstalk of toll-like receptors signaling and Nrf2 pathway for regulation of inflammation
.
Biomed Pharmacother
2018
;
108
:
1866
78
.
36.
Manikandan
P
,
Nagini
S
. 
Cytochrome P450 structure, function and clinical significance: a review
.
Curr Drug Targets
2018
;
19
:
38
54
.
37.
Tjalsma
H
,
Boleij
A
,
Marchesi
JR
,
Dutilh
BE
. 
A bacterial driver-passenger model for colorectal cancer: beyond the usual suspects
.
Nat Rev Microbiol
2012
;
10
:
575
82
.