Gut microbial and metabolomics profiles reveal the potential mechanism of fecal microbiota transplantation in modulating the progression of colitis-associated colorectal cancer in mice

Abstract

Purpose

Intestinal flora promotes the pathogenesis of colorectal cancer (CRC) through microorganisms and their metabolites. This study aimed to investigate the composition of intestinal flora in different stages of CRC progression and the effect of fecal microbiota transplantation (FMT) on CRC mice.

Methods

The fecal microbiome from healthy volunteers (HC), colorectal adenoma (CRA), inflammatory bowel disease (IBD), and CRC patients were analyzed by 16s rRNA gene sequencing. In an azoxymethane (AOM)/dextran-sulfate-sodium (DSS)-induced CRC mouse, the effect of FMT from HC, CRA, CRC, and IBD patients on CRC mice was assessed by histological analysis. Expression of inflammation- EMT-associated proteins and Wnt/β-catenin pathway were assessed using qRT-PCR and western blot. The ratio of the fecal microorganisms and metabolomics alteration after FMT were also assessed.

Result

Prevotella, Faecalibacterium, Phascolarctobacterium, Veillonella, Alistipes, Fusobacterium, Oscillibacter, Blautia, and Ruminococcus abundance was different among HC, IBD, CRC, and CRA patients. HC-FMT alleviated disease progression and inflammatory response in CRC mice, inhibited splenic T help (Th)1 and Th17 cell numbers, and suppressed the EMT and Wnt/β-catenin pathways in tumor tissues of CRC mice. IBD-FMT, CRA-FMT, and CRC-FMT played deleterious roles; the CRC-FMT mice exhibited the most malignant phenotype. Compared with the non-FMT CRC mice, Muribaculaceae abundance was lower after FMT, especially lowest in the IBD-FMT group; while Lactobacillus abundance was higher after FMT and especially high in HC-FMT. Akkermansia and Ileibacterium abundance increased after FMT-HC compared to other groups. Metabolite correlation analysis revealed that Muribaculaceae abundance was significantly correlated with metabolites such as Betaine, LysoPC, and Soyasaponin III. Lactobacillus abundance was positively correlated with Taurocholic acid 3-sulfate, and Ileibacterium abundance was positively correlated with Linoleoyl ethanolamide.

Conclusion

The different intestinal microbiota communities of HC, IBD, CRA, and CRC patients may be attributed to the different modulation effects of FMT on CRC mice. CRC-FMT promoted, while HC-FMT inhibited the progress of CRC. Increased linoleoyl ethanolamide levels and abundance of Muribaculaceae, Akkermansia, and Ileibacterium and reduced Fusobacterium might participate in inhibiting CRC initiation and development. This study demonstrated that FMT intervention could restore the intestinal microbiota and metabolomics of CRC mice, suggesting FMT as a potential strategy for CRC therapy.

Introduction

Colorectal cancer (CRC) is one of the most prevalent malignancies; it also accounts for a major cause of cancer-related fatalities globally [1]. Epidemiologic data show that, in 2020, CRC accounted for 9.4% of cancer-related fatalities and almost 10% of all cancer incidence globally, just slightly less than lung cancer [2]. This presents a significant burden to the global economy and public health. Therefore, it is particularly important to explore the pathogenesis of CRC.

CRC is predominantly disseminated, with approximately 85% of colorectal cancers originating from colorectal adenomas (CRA). The evolution of colorectal adenoma cancer usually takes 5–15 years, creating an optimal window for clinical prevention and treatment of CRA and CRC [2]. Presently, the study of the correlation between inflammation and tumor development and progression has emerged as an important focus within the field of tumor immunology. Numerous clinical and basic studies have confirmed that inflammatory bowel disease (IBD) is tightly associated with CRC development [3]. The presence of CRC in inflammatory bowel disease (IBD) is frequently regarded as a prototype of tumorigenesis induced by inflammatory response. Chronic inflammation causes DNA damage induced by oxidative stress, thereby activating oncogenes and inactivating tumor-suppressive genes. The hallmarks of oxidative damage and DNA double-strand breaks are gradually increasing within the inflammation-atypical hyperplasia-cancer sequence. The inflammatory and oncogenic processes are caused by the host’s immune response and the intestinal microorganism and its products [4, 5]. Therefore, an in-depth study of the role of the intestinal microorganism and its interrelationship with CRC is crucial to the exploration of the pathogenesis of CRC and the development of therapeutic options.

As a key component of the intestinal tract, the intestinal micro-ecosystem is important for the regulation of gut homeostasis and host health as it is capable of protecting the mucosal barrier, nutrient metabolism, and immunity [6]. The gut microbiota was categorized into 3 groups according to their effect on the intestinal tract, such as physiological bacteria, conditioned pathogens, and pathogens [7].

Alterations in the intestinal and exterior environments could lead to a decreased ratio of the dominant intestinal microbiota and confer an advantage for survival to pathogens or conditioned pathogens. Consequently, several studies have delved into fecal microbiota transplantation (FMT) to treat cancer by restoring microbiota-host homeostasis [8, 9]. However, the current understanding of the mechanism of disruption or restoration of the intestinal microbiota affecting CRC development remains limited.

In this study, the relationship between the intestinal microbiome and CRC progression was assessed by analyzing the microbiome of fresh fecal specimens from HC, IBD, CRA, and CRC patients, as well as analyzing the intestinal microbiome and metabolomic from the FMT-treated CRC mouse model. The results show that FMT was able to reverse the imbalance of the dominant intestinal microorganism flora and metabolic disturbances and reduce the severity of CRC.

Materials and methods

Clinical samples

A total of 118 preoperative fecal specimens from patients with IBD (n = 31), CRA (n = 36), CRC (n = 32), and Healthy Control (HC, n = 19) were collected for 16 S rRNA gene sequencing in this study. This study was conducted with the approval of the Ethics Committee of Xiangya Hospital of Central South University (CSU-2022-0114). All patients were made aware of the study protocol and signed a written informed consent.

The clinicopathological characteristics of the patients were also collected and recorded in this study, including information on gender, age, date of surgery, preoperative tumor stage, and chemotherapy status. Standard exclusion criteria included permanent ostomy, distant metastases, chronic kidney disease or cirrhosis, ischemic heart disease with unstable angina, patients with class III or IV chronic heart failure or acute myocardial infarction during the last six months, patients with a history of chronic diarrhea, history of diabetes, or history of autoimmune disease, individuals who had used antibiotics or probiotics during the last three months prior to the sample collection, individuals with a history of abdominal surgery for other reasons were excluded. Clinical information is provided in Table S1.

Each patient provided a fresh fecal sample at the hospital, which was immediately divided into two aliquots. One portion was immediately placed in liquid nitrogen and sent to Wuhan BGI Technology Co. Ltd. for 16s rRNA gene sequencing, and the other portion was used for FMT fluid preparation.

16 S rRNA gene sequencing and analysis

The DNA isolation from feces was performed by Wuhan BGI Technology Co. Ltd. About 100–200 mg of fecal specimen was transferred into the centrifuge tube containing grinding beads and supplemented with 1 mL of Buffer ATL (tissue lysis buffer containing SDS)/ polyvinylpyrrolidone (PVP)-10. The specimen was ground using a grinding machine, followed by a 20-minute incubation at 65℃. After centrifugation at 14,000×g for 5 min, the supernatant was transferred to a new tube, and 0.6 mL Buffer phenol chloroform isoamylalcohol (PCI) was added, the mixture was mixed thoroughly by vortexing for 15 s. Following a second centrifugation at 18,213×g for 10 min, transfer the supernatant to a deep-well plate containing a magnetic bead binding solution. The deep well plate was transferred to the proper place of the machine and the corresponding program was started in Kingfisher. The DNA was then transferred to a 1.5 mL centrifuge tube. Universal primers were designed with V3-V4 16 S-rRNA gene fragments. PCR amplification was carried out with a forward primer 341 F: CCTACGGGGNGGCWGCAG and a reverse primer 805R: GACTACHVGGGGTATCTAATCC. The V3-V4 region amplification products were in equal molar amounts, and the paired ends were sequenced on an Illumina NovaseQ6000 PE250 system (Illumina, CA, USA). DADA2 (version 1.2.0) was used to denoise raw data. The chimeric sequences were removed via the UCHIME algorithm. Operational Taxonomic Units (OTU) clustered with a 97% identify cutoff using the Ribosomal Database Project (RDP) classifier algorithm and the SILVA database (SSU132). The Quantitative Insights Into Microbial Ecology (QIIME) tool (version 1.9) was used for α-diversity (Chao, ACE, Shannon, and Simpson) and β-diversity analysis. Based on the Unweighted UniFrac distance and analysis of similarity (ANOSIM) were used to assess the overall differences in the β-diversity of gut microbiota among different groups using the vegan package of R language (version 4.0.2). PICRUSt 2.0 (version 2.4.1) was used to predict functional profiles of gut microbiota. Predicted functional genes were categorized in the Kyoto Encyclopedia of Genes and Genome (KEGG).

CRC mouse model and FMT

Male C57BL/6 mice (six weeks gold, weighed 18–20 g) were procured from SLAC laboratory animal company (Hunan, China). All mice were kept in plastic cages at a temperature of 24–25 °C, relative humidity of 60–65%, and a light-dark cycle of 12 h, and were allowed to take food and water ad libitum. All experiments involving animals were conducted with the approval of the Medical Ethics Committee of Xiangya Hospital, Central South University.

C57BL/6 mice were acclimatized to feeding for one week. Next, pseudo-germ-free CRC mice were modeled in this study using the following method: 0.2 g/L of ampicillin (A8180, Solarbio, Beijing, China), neomycin (N8090, Solarbio), metronidazole (M8060, Solarbio) and 0.1 g/L of vancomycin (S17059, Yuanye Bio, Shanghai, China) were added to the drinking water daily for a total of two weeks. Following the last dose of antibiotics, mice were subjected to an intraperitoneal injection of one dose of 10 mg/kg of AOM (HY111375, MedChemExpress, NJ, USA). A week after the AOM injection, the mice were allocated at random into 5 groups (n = 6): the NC group, the HC-FMT group, the CRC-FMT group, the CRA-FMT group, and the IBD + FMT group. The FMT procedures were performed as previously described with small modifications [10, 11]. 3 people from HC (F/M 2/1, 31.66 ± 13.31 years), 3 patients with IBD (F/M 1/2, 27.33 ± 4.51 years), 3 patients with CRA (F/M 1/2, 50.33 ± 29.74 years), and 3 patients with CRC (F/M 1/2, 64.67 ± 10.12 years) were randomly selected as fecal donors for FMT. The fecal supernatant was prepared as previously described [10, 12]. Briefly, 1 g of fecal was then suspended in 5 mL PBS. The fecal slurry was subsequently filtered by 0.25 mm stainless steel mesh and centrifuged at 1000 rpm for 5 min at 4 °C, and the suspension was collected. Equal amounts of fecal supernatants from 3 HC, 3 IBD, 3 CRA, and 3 CRC donors were mixed and used as a single source for mice in the HC-FMT group, IBD-FMT group, CRA-FMT group, and CRC-FMT group respectively. The 4 groups of FMT mice were enemas (200 µL/each) using the fecal supernatants from CRC patients, CRA patients, IBD patients, and the HC group twice weekly. The NC group mice were enemas with PBS as a negative control. On the same day of the first enemas, 2% DSS (9011-18-1, MP Biomedicals, CA, USA) was supplemented into the drinking water of the mice for 7 days. Normal drinking water was used during the recovery period (14 days). 3 cycles of 2% DSS treatment were performed. Fresh feces from all groups were harvested at the end of experiments, preserved in liquid nitrogen, and sent to Shanghai WeiHuan Biotechnology Co. Ltd for mouse fecal 16s RNA gene sequencing and untargeted metabolomics studies. The 16s rRNA gene sequencing procedure is the same as 2.2.

The mice were weighed and euthanized. The spleens of the mice were collected and weighed to calculate the ratio of spleen weight to body weight. Meanwhile, the colon was resected from the ileocecal region to the anal verge by cutting it longitudinally along the main axis and flushing it with phosphate-buffered saline (pH 7.4). Finally, the mouse colon was photographed. Tumor volume was computed by the formula: volume = 1/2 × length × width².

Hematoxylin and eosin (HE) staining

After being fixed with 4% paraformaldehyde and paraffin-embedded as previously described [13], the mouse colon tissue samples were sectioned into 4-µm slices, followed by staining as per the protocols of the HE staining kit (G1120, Solarbio, Beijing, China) [14]. Finally, the sections were sealed with neutral resin. An optical microscope (Olympus, Japan) was used to observe the section and capture images.

Immunohistochemistry (IHC) staining

After antigen repair, blocking endogenous peroxidase and non-specific binding sites as previously described [15], the sections were incubated with primary antibody: Ki-67 (AF0198, 1:100, Affinity Bioscience, Changzhou, China) overnight at 4 ℃ in a wet box. Next, the streptavidin-biotin complex (SABC)- horseradish peroxidase (HRP) Kit (P0615, Beyotime, Shanghai, China) was used for further staining. Briefly, the sections were incubated with biotin-labeled goat-anti rabbit IgG for 1 h at room temperature and SABC working solution for 30 min at room temperature. Finally, the sections were further incubated with DAB chromogenic solution (P0202, Beyotime) for 10 min in light-deprived conditions and observed under a microscope.

Quantitative real-time reverse-transcription PCR (qRT-PCR)

TRIzol (15596018, Invitrogen) was used to extract total RNA from mouse tumor tissue samples. The Reverse Transcription Kit (RR037B, TaKaRa, Tokyo, Japan) was used as per the kit’s protocols to perform reverse transcription. Gene expression was detected using A Jena qTower 3.0G Fluorescent Quantitative PCR Instrument (Germany), and the reaction conditions were conducted as per the protocols of the BeyoFast™ SYBR Green qPCR Mix (D7260, Beyotime). Briefly, initial pre-denaturation was conducted at 95℃ for 2 min, followed by 40 cycles of denaturation at 95℃ for 5 s, annealing at 60℃ for 30 s, and elongation at 72℃ for 30 s. qRT-PCR was set up with 3 replicates per reaction. GAPDH was used for the normalization of the relative expression of target genes, which was calculated using the 2-ΔΔCt method, ΔΔCt = experimental group (Ct target gene-Ct internal reference) -control group (Ct target gene-Ct internal reference). The amplification primer sequences of PTGS2, MMP9, CTNNB1, TNF-α, GAPDH gene are detailed in Table 1.

Table 1 qRT-PCR primer sequences

Western blot

RIPA lysis solution (P0013, Beyotime) was used to extract total protein from mouse colon tumor tissues, thereby yielding protein samples. A bicinchoninic acid (BCA) protein assay kit (P0011, Beyotime) was used to measure the protein content, and then the corresponding protein volume was taken and supplemented into the sampling buffer (P0015, Beyotime) and mixed. A 5-minute heating in a boiling water bath was conducted to denature the protein. The proteins were subjected to 10% Sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) electrophoresis (80 V, 30 min) until the bromophenol blue was migrated into the gel; next, a higher voltage (120 V) was used for 1–2 h. The membrane transfer onto a polyvinylidene difluoride (PVDF) membrane (ISEQ15150, Millipore) was performed in an ice bath, with a current of 220 mA for 120 min. Subsequently, after 1–2 min washing within the washing solution, the membrane was kept in the blocking solution for 60 min at room temperature (RT). Primary antibodies (GAPDH (5174 S, 1:1000, Cell Signaling Technology [CST], Boston, USA), TNF-α (60291-1-Ig, 1:1000, Proteintech, Wuhan, China), COX2 (12375-1-AP, 1. 1000, Proteintech), MMP9 (10375-2-AP, 1:1000, Proteintech), CTNNB (8480, 1:1000, CST), E-cadherin (3195, 1:1000, CST), N-cadherin (#13116, 1. 1000, CST), Vimentin (5741, 1:1000, CST), Snail (3879, 1:1000, CST) and CD133 (ab271092, 1:1000, Abcam) were incubated at 4 °C for one night in a shaker. On the following day, the membrane was rinsed thrice in the membrane washing solution (10 min each). The membrane was further incubated with secondary antibodies (horseradish peroxidase-conjugated goat anti-rabbit/mouse IgG, 1:1000, A0208/A0216, Beyotime) for 2 h at RT. The membrane was rinsed thrice in TBST (10 min for each) and was supplemented dropwise with the ECL solution, and then the chemiluminescence imaging system (Bio-rad) was applied to detect the membrane. ImageJ software was applied to perform quantitative analysis.

Flow cytometry for Th1 and Th17 cells

The spleen single-cell suspensions were collected by digesting mouse spleens within Hank’s balanced salt solution containing 0.1 mg/mL collagenase D (11088858001, Roche, Switzerland) and 50 U/mL DNase I (09873562001, Roche) for 30 min at 37 °C. The ammonium chloride-potassium lysis buffer was used to remove erythrocytes. Finally, the cell suspensions were filtered through a 70-µm cell filter, washed with PBS, and resuspended for flow cytometry analysis.

Following the staining of surface markers (CD4-APC, 100412, BioLegend, CA, USA), cells were rinsed in a staining buffer and then fixed and permeabilized using the Fix&Perm Cell Permeabilization Kit (GAS004, Invitrogen, CA, USA). Next, antibodies against INF-γ-PE (163504, BioLegend) or IL-17 A-PE (506904, BioLegend) were used to stain cells. Lastly, after two washings, cells were re-suspended in a staining buffer before flow cytometry analysis (Novocyte, CA, USA).

CRC mouse fecal metabolomics and data analysis

Mouse fecal specimens were maintained within a 1.5-mL Eppendorf tube. Next, 20 µL of each of the internal standards Lyso PC17:0 (0.01 mg/mL) and L-2-chlorophenylalanine solution were supplemented. The supernatants from each tube were isolated using ultrasonography and centrifugation before being extracted using crystal syringes. 0.22 μm microfilters were used to filter the supernatants, which were then transferred into a glass vial for liquid chromatography/mass spectrometry (LC/MS). Subsequently, the supernatants were analyzed as revealed previously using an AB SCIEX Triple TOF 6600 System and an ACQUITY UHPLC system. Agilent Masshunter Qualitative Analysis B.08.00 software (Agilent Technologies, USA) was applied to pre-process raw LC-MS data. In the R software platform, the XCMS program was used in peak identification, retention time correction, and automatic integration pretreatment. The data were subsequently subjected to internal standard normalization. The qualitative method of metabolites is: searching in the Human Metabolome Database (HMDB) [16] and Metlin [17] for accurate molecular weight comparison. Adduct manner: [M + H] + and [M + Na] + was selected in positive mode, [M-H]- in negative mode. Mass error value: 30 PPM.

LC-MS data were quality assessed using principal component analysis (PCA). The between-group differentially expressed metabolites were identified using orthogonal partial least squares discriminant analysis (OPLS-DA) and t-test. Thresholds were set: projected important variable (VIP) > = 1 and p < 0.05.

Statistical analysis

All experimental data obtained were presented in terms of mean ± standard deviation (SD). The t-test, Wilcoxon rank sum test, or one-way ANOVA was carried out to examine statistical differences, and the Pearson correlation coefficient was performed to assess correlations. GraphPad Prism (8.0.1), and SPSS (25.0) were applied for data analysis. A P-value of less than 0.05 was regarded as statistically significant.

Results

Characterization of intestinal microorganisms in patients with IBD, CRA and CRC

In this study, intestinal microorganism specimens were first harvested for 16 S rRNA gene sequencing from healthy individuals (HC) (n = 19), IBD (n = 31), CRA (n = 36) and CRC (n = 32) patients. A total of 8,301,013 raw reads and an average of 70347.57 ± 905.51 reads per sample were obtained. A total of 8,064,595 clean reads, with an average of 68344.03 ± 479.26 per sample, remained for downstream analysis. The number of species and differences between groups was assessed using α- and β-diversity analyses. In Fig. 1A, flora richness was assessed using Sobs, Chao1 and ace indices, while the microbial diversity of the samples was assessed using Shannon and Simpson indices. The richness and diversity indices between the intestinal flora were slightly higher in patients in the CRC and CRA groups compared to the HC group without a statistically significant difference (P > 0.05). In contrast, the richness of the Sobs, Chao, and Ace indices of the flora within the IBD group showed a dramatic decrease compared to the other three groups. In terms of bacterial diversity, Simpon and Coverage indices were significantly increased compared to the other three groups and statistically different (P < 0.05). UniFrac analysis revealed no remarkable difference between microbiota from the CRA group and CRC group, but substantial differences between the other two groups (P < 0.001) (Fig. 1B).

Fig. 1

Characterization of intestinal microorganisms in patients with IBD, CRA, CRC

A: Alpha diversity consists of the following metrics: sobs, Chao1, and ace indices for assessing colony richness, Shannon and Simpson indices for assessing microbial diversity of samples, and coverage for assessing sequencing depth. B: Beta diversity was calculated using Unweighted UniFrac. C: Heatmap of the relative top20 abundance of the intestinal flora in each study group at the genus level

* p < 0.05, ** p < 0.01, *** p < 0.001

Moreover, Fig. 1C shows the heat map of the relative abundance of intestinal flora. A total of 136 flora were annotated at the genus level, which were sorted according to relative abundance from largest to smallest. The top 20 abundance bacteria were taken as the dominant flora (Table 2). Prevotella, Faecalibacterium, Phascolarctobacterium, Veillonella, Alistipes, Fusobacterium, Oscillibacter, Blautia and Ruminococcus abundance were different among groups (Kruskal-Wallis test, p < 0.05). Trends in the distribution of abundance for each group of samples are shown by box plots (Fig. 2, t-test). Blautia abundance was lowest in IBD compared to normal, CRA and CRC. The abundance of Faecalibacterium [18], which has anti-inflammatory properties, and Ruminococcus [19], a beneficial bacterium, was decreased in the feces of patients with IBD, CRA, and CRC than the HC without statistically significant differences (t-test, Table S2).

Table 2 The abundance top 20 intestinal bacteria from HC, IBD, CRA and CRC patients at genus level

Fig. 2

Differences in top 20 intestinal bacteria genus abundance between groups *p < 0.05, ** p < 0.01, t-test

Metabolic prediction of intestinal microbiota in HC and patients with IBD, CRA, CRC

A total of 1,953 metabolites were predicted by PICRUSt analysis. Of these, 7 metabolite levels in the Top100 abundance metabolite were significantly different among the four sample groups (Fig. 3A; Table 3). Specifically, the levels of Acetolactate synthase, Iron-chelate-transporting ATPase, Peroxiredoxin, Cystathionine beta-lyase, RNA helicase, Ribonucleoside- diphosphate reductase were significantly elevated, while relatively high levels of Protein-N(pi)-phosphohistidine–sugar phosphotransferase were also observed in the IBD group.

Fig. 3

Metabolic prediction of intestinal microbiota in HC and patients with IBD, CRA, CRC

A: Box plots demonstrating the levels of the predicted seven intestinal differential metabolites in HC and patients with IBD, CRA, and CRC. B: The correlations between intestinal bacterial genera and predicted differential metabolites were analyzed by Pearson correlation coefficients. * P < 0.05, ** P < 0.01, *** P < 0.001

Table 3 Metabolites predictions based on Picrust

The correlation between the top 20 intestinal bacterial genera and metabolites was explored using the Pearson correlation analysis (Fig. 3B). Particular attention was paid to the relationship between differential bacterial genera among groups (Prevotella, Faecalibacterium, Phascolarctobacterium, Veillonella, Alistipes, Fusobacterium, Oscillibacter, Blautia, and Ruminococcus) and the 7 metabolites. The results show that Prevotella abundance was negatively correlated with acetolactate synthease, iron-chelate-transporting ATPase, peroxiredoxin, cystathionine beta-lyase, RNA helicase, and ribonucleoside-diphosphate reductase; Faecalibacterium abundance was negatively correlated with Iron-chelate-transporting ATPase level, protein-N(pi)-phosphohistidine–sugar phosphotransferase, peroxiredoxin, RNA helicase and ribonucleoside-diphosphate reductase; Veillonella abundance was positively correlated with acetolactate synthase, iron-chelate-transporting ATPase, peroxiredoxin and ribonucleoside-diphosphate reductase; Fusobacterium abundance was positively correlated with ribonucleoside-diphosphate reductase; Blautia abundance was positively correlated with acetolactate synthase (p < 0.01).

Intestinal flora can influence tumor growth in mice with AOM/DSS-induced colitis-associated carcinomas

An AOM/DSS-induced colitis-associated carcinoma mouse model with pseudo-germ-free was further established for FMT to explore the role of intestinal flora in CRC. The schematic diagram of the animal experiment is detailed in Fig. 4A. The top20 abundance bacteria results among the donors’ samples showed that the abundance of Facelibacterium, Lachnospiracea_incertea_sedis, Fusobacterium, Enterococcus and Ruminococcus abundance was different among HC, IBD, CRA, CRC groups. In addition, the abundance of Facelibacterium and Ruminococcus was higher in HC donors (Fig.S1). First, the number of tumors, histopathological changes, and cell proliferation in mouse colon tissues were examined. AOM/DSS-induced precancerous lesions in mice were observed by histopathological analysis (Fig. 4B), marked by crypt destruction, infiltration of inflammatory cells, and tumor formation. Colon lesions could be ameliorated using healthy crowd-fecal microbiota transplantation (HC-FMT). Conversely, colon injury was exacerbated by the use of CRC-FMT and IBD-FMT. Ki67 acts as a cell proliferation marker, with abundant expression within malignancy tissues. Therefore, Ki67 expression within the colon was analyzed (Fig. 4C). The IHC result shows that the expression level of tumor Ki67 was ameliorated and decreased by HC-FMT relative to the NC group, while Ki67 expression showed to be elevated by CRC-FMT and IBD-FMT. Meanwhile, the number and volume of tumors of mice in each group were also counted (Fig. 4D-E). Compared to the NC group, the number, and volume of tumors were reduced in the HC-FMT group but with no statistical difference (P > 0.05), while the number and volume of tumors showed to be dramatically elevated within the CRC-FMT group and the IBD-FMT group compared to HC-FMT group (P < 0.05).

Fig. 4

Intestinal microbiota can influence tumor growth in mice with AOM/DSS-induced colitis-associated carcinomas

A: Animal experimental protocol. B: Representative gross images of mice colonic tissue samples from all treatment groups (top), histopathological alterations determined by H&E staining (middle & bottom, scale bar = 200 & 50 μm). C: IHC staining to detect Ki67 expression in the colonic tissues of mice from all groups (scale bar = 20 μm). D-E: Number & volume of tumors in colon tissues (n = 6)

* P < 0.05, vs. the NC mice. ## P < 0.01, vs. HC-FMT mice

Taken together, intestinal flora exerts a critical regulatory effect on the AOM/DSS-induced colon cancer model, and treatment with fecal transplantation of microbiota can ameliorate or worsen colitis-induced carcinogenesis.

Effect of intestinal microbiota on inflammatory factors and immune cell infiltration

Due to the changes in inflammatory cell infiltration observed in the intestinal tumor tissues from all groups, the expression levels of inflammatory factors TNF-α and COX-2 in tumor tissues were further examined. qRT-PCR and western blot results show that HC-FMT considerably decreased TNF-α and COX-2 gene and protein levels in tumor tissues, compared to the NC mice, while the expression levels of inflammatory factors were markedly up-regulated by CRC-FMT, CRA-FMT, and IBD-FMT. Moreover, CRC-FMT, CRA-FMT, and IBD-FMT significantly promoted the upregulation of TNF-α and COX-2 gene and protein expression compared with HC-FMT (Fig. 5A-B, E).

Fig. 5

Effect of intestinal flora on inflammatory factors and immune cell infiltration

A-E: TNF-α, COX-2, MMP9 and CTNNB expression levels in tumor tissues of mice in each group (n = 3) were detected using qRT-PCR (A-D) and western blot (E). F: E-cadherin, N-cadherin, Vimentin, Snail and CD133 levels in intestinal tissues of mice in each group (n = 3) were detected using Western blot. G: Spleen/body weight ratio of mice in each group (n = 6). H-I: Flow cytometry to detect changes in Th1 (CD4 + IFN-γ) and Th17 (CD4 + IL-17 A) cell numbers in the spleens of mice in each group (n = 3)

* P < 0.05, ** P < 0.01, vs. the NC mice. ## P < 0.01, vs. HC-FMT mice

Matrix metalloproteinases (MMP) [20] and epithelial-mesenchymal transition (EMT) [21] are critical for tumor invasion and metastasis. Reportedly, the Wnt/β-catenin pathway plays an important regulatory role in inflammatory signaling, favoring tumorigenesis [22]. Therefore, the expression of related proteins was further examined. HC-FMT dramatically reduced MMP9, CTNNB1, N-cadherin, Vimentin, Snail, and CD133 levels, and elevated E-cadherin levels compared with the NC group. Meanwhile, CRC-FMT, CRA-FMT and IBD-FMT exerted opposite effects on the expression of the above proteins. Compared with HC-FMT, the CRC-FMT, CRA-FMT and IBD-FMT groups were able to reverse the mRNA levels and protein expression of the above factors (Fig. 5C-F).

Moreover, the immune cell types in the spleens of each group of mice were detected by flow cytometry. No remarkable difference between groups was observed within mice spleen /body weight ratios (Fig. 5G). Compared to the NC group, Th1 and Th17 cell numbers were remarkably decreased by HC-FMT. Conversely, Th1 and Th17 cell numbers were shown to be dramatically elevated by CRC-FMT, CRA-FMT and IBD-FMT. Compared with HC-FMT, Th1 and Th17 cell numbers showed to be considerably increased within the CRC-FMT, CRA-FMT and IBD-FMT groups (Fig. 5H-I). The above results indicate that HC-FMT can inhibit Th1 and Th17 cell overactivation and alleviate the inflammation in colon cancer.

FMT alters the intestinal microbiota composition in CRC mice

The mouse feces were collected from each group for 16 S rRNA gene sequencing. A total of 3,550,423 raw reads and an average of 118347.43 ± 15609.84 reads per sample were obtained. A total of 3,454,984 clean reads, with an average of 115166.10 ± 15298.97 per sample, remained for downstream analysis. The species number within the community and differences between groups were analyzed using Alpha and Beta diversity. No remarkable intergroup differences were observed within the abundance and diversity of mouse intestinal flora among the groups (Fig. 6A-F). β-diversity was calculated using ANOSIM, which reflects intergroup differences in microbial communities. As shown by Fig. 6G, intergroup differences within microbiota communities were observed (P < 0.05).

Fig. 6

FMT alters the intestinal microbiota composition in CRC mice after FMT

A-F: Pielou, Shannon, and Simpson to estimate microbial diversity in the samples, Observed features, Chao1, and ACE to assess the abundance of flora in the community. G: ANOSIM to calculate β-diversity. H: Top 20 dominant flora species at genus level among groups

Next, further tests were performed to detect whether FMT treatment exerted an effect on the dominant intestinal flora of mice. The top 20 dominant flora in the intestines of CRC mice in each group at the genus level are shown in Fig. 6H. Figure 7; Table 4 and Table S3 show the abundance of each dominant flora in each group, of which Muribaculaceae, Bacteroides, Akkermansia, Parasutterella, [Ruminococcus]_torques_group, Fusobacterium, Rikenellaceae_RC9_gut_group, Bifidobacterium, Odoribacter, Alloprevotella, Ileibacterium, and Erysipelatoclostridium had differences in abundance among groups (Kruskal-Wallis test, P < 0.05). A pairwise comparison revealed that the abundance of Muribaculaceae after FMT was lower compared to the NC group (the non-FMT group) and was lowest in the IBD-FMT group, while Lactobacillus was all higher than that of the NC group after FMT and was highest in HC-FMT. Akkermansia and Ileibacterium abundance was increased after FMT-HC treatment compared to other groups. The abundance of Fusobacterium was not different in the HC-FMT and NC groups and showed to be elevated within the other IBD-, CRA- and CRC-FMT treatment mice compared to the HC-FMT mice.

Fig. 7

Box plots showed the relative abundance of top20 abundance intestinal microbiota after FMT. *P < 0.05, ** P < 0.01, *** P < 0.001

Table 4 The abundance top 20 intestinal bacteria from CRC mice with HC-FMT, IBD-FMT, CRA-FMT and CRC-FMT

Metabolomics analysis of intestinal contents in FMT-treated mice

The metabolic alterations within the intestinal microbiota after AOM/DSS modeling and FMT treatment were characterized using LC-MS-based untargeted metabolomics analysis of fecal specimens. Data quality control results show that QC samples aggregated overlap in the PCA analysis plots in NEG and POS modes (Fig S2A-B), indicating that the extracted sample peak area data were stable, and the method was reproducible enough for the next statistical analysis. Firstly, a clear group-based clustering pattern of the samples in each group was shown in the PCA plots, indicating that FMT treatment was able to alter the composition of the intestinal metabolome of CRC mice (Fig S2A-B). Partial least squares discriminant analysis (PLS-DA) further confirmed that the fecal metabolic profiles of mice can be altered by FMT treatment (Fig S2C-D). The permutation test results are shown in Fig S2E-F. The current model has good fitting accuracy and can effectively differentiate the intestinal microorganism samples from each group of mice.

The dominant metabolites and functions in each group of mice were further analyzed. Figure 8A-B shows the differential metabolites in each group of mice in NEG and POS modes. In NEG mode, 1648 metabolites were screened based on VIP (Variable Importance in the Projection) of the PLS-DA model (VIP ≥ 1) for 675 metabolites. The top 20 most abundant metabolites with statistical differences among groups were further screened by the Kruskal-Wallis test (P < 0.001) (Fig S3, Table 5). Within the POS model, 656 metabolites were screened from 1585 metabolites based on VIP (VIP ≥ 1) of the PLS-DA model, and then the top 20 most abundant metabolites with the statistical differences were screened (P < 0.001, Fig S4, Table 6). These differential metabolites were performed with KEGG pathway enrichment analysis, indicating that these metabolites were significantly linked to pyruvate metabolism, glycolysis/gluconeogenesis, serine, and threonine, tryptophan metabolism, aminoacyl-tRNA biosynthesis, steroid hormone biosynthesis, among other functions (Fig. 8C).

Fig. 8

Metabolomics of intestinal microorganisms in FMT-treated mice

A-B: Heatmaps show intergroup differences in metabolites of intestinal content in NEG and POS modes in each group. C: KEGG enrichment to analyze the metabolic processes associated with differential metabolites in NEG and POS models

Table 5 The top20 most abundant metabolites with the statistical differences in NEG mode

Table 6 The top20 most abundant metabolites with the statistical differences in POS mode

Correlation analysis of intestinal microbiota and metabolites in FMT-treated mice

The correlation between the differential metabolites and microbiota after FMT transplantation (vs. NC group t-test, p < 0.05.) is shown using Pearson correlation analysis. With NEG differential metabolites, the abundance of Muribaculaceae was found to be highly significantly positively correlated with garcinone A, (R)-Pantothenic acid 4’-O-b-D-glucoside and significantly negatively related to Janthitrem C. The abundance of Lactobacillus was shown to be highly significantly positively related to Taurocholic acid 3-sulfate (Fig. 9A). With POS differential metabolites, the abundance of Muribaculaceae was significantly negatively correlated with Betaine, LysoPC (20:3(5Z,8Z,11Z)/0:0) and Soyasaponin III, while Ileibacterium abundance was significantly positively correlated with Linoleoyl ethanolamide. (Fig. 9B).

Fig. 9

Correlation analysis of intestinal microbiota and metabolites in FMT-treated mice

A: Correlation analysis of top20 differential metabolites in NEG mode and differential microbiota. B: Correlation analysis of top20 differential metabolites in POS mode and differential microbiota. *P < 0.05, **P < 0.01

Discussion

Colorectal cancer (CRC) is one of the most frequent malignancies globally and the second leading cause of cancer-related deaths, posing a significant health burden globally. Data from 16s rRNA gene sequencing and experimental models have confirmed the underlying impact of the intestinal microbiota on regulating CRC progression [23]. Herein, fecal samples from patients with different types of IBD, CRA and CRC were clinically collected to perform 16 S gene rRNA gene sequencing and fecal metabolomics analysis. The relative abundance of Faecalibacterium and Parabacteroides was found to be decreased in the IBD, CRA, and CRC groups compared to the HC group. Faecalibacterium, as one of the main butyrate-producing bacteria in the intestinal tract, was expanding its clinical relevance and applications and was found to have anti-inflammatory and intestinal microbiota-modulating characteristics in treating IBD, Crohn’s disease and CRC [18, 24]. F. prausnitzii, a probiotic in the genus Faecalibacterium, has been reported to significantly reduce the frequency and formation of abnormal crypt foci in AOM-induced colon cancer in rats [25]. F. prausnitzii inhibits colorectal cancer cell proliferation in vitro, whose numbers in the intestine are negatively correlated with IBD and CRC activities [25, 26]. In this study, Faecalibacterium was significantly down-regulated in the intestinal flora of patients in the IBD, CRA, and CRC groups, which is in line with previous studies [27,28,29]. The decreased abundance of Faecalibacterium may also be associated with colorectal inflammatory lesions as well as tumorigenesis. Combined with the results of metabolic prediction, Faecalibacterium was found to be negatively correlated with the level of Iron-chelate-transporting ATPase protein-N(pi)-phosphohistidine–sugar phosphotransferase, peroxiredoxin, RNA helicase, and ribonucleoside-diphosphate reductase. Iron chelation may play a therapeutic role in CRC by promoting CRC cell apoptosis [30]. Peroxiredoxins are frequently dysregulated in cancer and are positively associated with colitis and colon cancer [31]. Targeting ribonucleotide reductases is an effective strategy in anti-cancers, including colon cancer [32, 33]. Therefore, Faecalibacterium may positively impact CRC therapy by regulating those metabolic pathways.

Previous experiments have demonstrated that FMT improves drug response in patients treated with radiotherapy and other antitumor agents and overcomes melanoma patients’ resistance to anti-PD-L1 treatment [34]. Previous experiments have demonstrated that FMT in CRC patients promotes intestinal adenoma formation in mice in a pseudo-germ-free mouse model, FMT in healthy controls was able to normalize intestinal flora by reducing tumor growth in a CRC mouse model [35]. In this study, the colorectal cancer mouse model was established by AOM/DSS. FMT experiments were performed on CRC mice using clinically collected fecal samples. An obvious increase was observed in the colorectal tumor number in the CRC-FMT group compared to the NC group (non-FMT mice) as well as the HC-FMT group, and Ki-67 expression within the tumors also showed to be remarkably elevated. The results demonstrate that the gut flora of CRC patients indeed promoted the development and proliferation of colorectal cancer in mice. Similar results were obtained within the CRA-FMT and the IBD-FMT groups, but the difference was most significant in the CRC-FMT group. FMT from HC effectively alleviated symptoms in CRC mice.

The beneficial effects of HC-FMT might be attributed to the higher abundance of potentially beneficial bacteria, such as Faecalibacterium and Ruminococcus, in healthy individuals, which are known for their anti-inflammatory properties and maintenance of gut barrier integrity [18, 36]. In contrast, the exacerbation of colon injury observed in the CRC-FMT and IBD-FMT groups might be attributed to an imbalance in microbial composition, despite the absence of a significant increase in harmful bacteria in our findings. This phenomenon is consistent with previous studies where sterile mice receiving fecal transplants from CRC patients exhibited increased precancerous lesions and mild inflammatory responses. These changes were associated with a rise in pathogenic bacteria such as Bacteroides and a reduction in beneficial bacteria like Faecalibacterium, which likely contributed to the activation of oncogenic pathways such as Wnt and Notch, thereby promoting disease progression [37]. Additionally, it has been shown that gut microbiota from CRC patients can disrupt the intestinal barrier, induce chronic low-grade inflammation, and activate pathways that accelerate the development of adenomas [38]. These mechanisms offer a plausible explanation for the exacerbation of colon injury in our study following FMT from CRC and IBD patients, as well as the protective effects observed with HC-FMT.

Chronic inflammation exerts a crucial effect on CRC development. As previously reported, soluble mediators exert a pivotal effect on maintaining the chronic inflammation microenvironment, such as IL-6, TNF-α and COX-2. These mediators are beneficial in promoting angiogenesis, growth, and migration/invasion of CRC-arising cells [39]. TNF-α promotes tumor metastasis by activating oncogenes leading to DNA damage, and induces the aggregation and adhesion of tumor-associated inflammatory cells, accelerates tumor infiltration and dissemination, and exerts a critical effect on promoting CRC development. TNF-α gene and protein are significantly highly expressed and thus can be used as an indicator to predict the prognosis of CRC within the tumor tissues of advanced CRC patients [40]. Elevated COX-2 expression has been found in tumor tissues of approximately 50% of CRA and 85% of CRC patients, and high COX-2 expression predicts poorer survival in CRC patients [41]. It has been well documented that inhibition of the COX-2 validation pathway by regular utilization of nonsteroidal anti-inflammatory drugs, such as aspirin and celecoxib, in the period of 10–15 years can reduce the relative risk of the organism developing CRC by 40–50% [42]. Therefore, blocking these cytokines can help prevent CRC progression. In this study, HC-FMT significantly inhibited TNF-α and COX-2 levels within tumor tissue samples of CRC mice. These results further confirm that reducing these inflammatory mediators inhibits CRC progression.

MMPs are a class of zinc-dependent neutral endopeptidases with hydrolytic activity against extracellular matrix (ECM) proteins. Among the 24 identified MMPs, MMP9 is undetectable within healthy tissues but exhibits high expression within inflammation and several malignancies [43]. The function of MMP9 in CRC remains elusive. Previous research reported that inhibition of MMP9 expression suppressed liver metastasis of CRC [44]. In this study, HC-FMT was found to significantly inhibit MMP9 expression, which supports the potential of MMP9 as a therapeutic target for malignancies.

EMT [21] and Wnt/β-catenin pathway [22] are important for tumor invasion and metastasis and have also been shown to be key regulators of inflammatory signaling that promotes tumorigenesis. The results show that HC-FMT could participate in regulating the expression levels of proteins associated with EMT and Wnt/β-catenin pathways. This finding enriches the novel mechanism of FMT in the treatment of malignant tumors, especially CRC.

Intestinal bacteria play an important role in regulating intestinal homeostasis by influencing immunity. Th cells (including Th1, Th2 and Th17 cells) have been proven to promote IBD progression by inducing multiple inflammatory pathways [45]. In this study, HC-FMT was found to effectively reduce the Th1 and Th17 cell numbers in the spleen. In contrast, FMT from patients with IBD, CRA and CRC significantly upregulated the amount of Th1 and Th17 cells. This result indicates that targeting Th1 and Th17 cells might be a novel strategy for treating CRC.

The changes in intestinal microbiota in CRC mice treated with FMT were examined. The results show that the abundance of Akkermansia and Ileibacterium was significantly upregulated in HC-FMT-treated mice, while Fusobacterium was significantly downregulated. Akkermansia was shown to be reduced in abundance in the intestinal tract of CRC and CRA patients [46], while supplementation with Akkermansia significantly inhibited the occurrence and development of CRC [47]. Meanwhile, Ileibacterium is a reported beneficial intestinal bacterium [48], Fusobacterium can promote CRC progression through multiple mechanisms, including WNT/β-catenin signaling activation and immunosuppression promotion [47]. Moreover, Muribaculaceae abundance was found to be significantly reduced within the intestines of mice treated with CRC-, CRA- and IBD-FMT. As previously reported, decreased Muribaculaceae might be linked to CRC progression [48]. These results indicate that elevating the abundance of Muribaculaceae, Akkermansia, and Ileibacterium and reducing the abundance of Fusobacterium might participate in inhibiting CRC initiation and development. Further investigation of the mechanism of these florae in CRC will provide important guidance for developing new therapeutic strategies.

Within the analysis of the correlation between intestinal flora and metabolites, Muribaculaceae was found to be negatively correlated with LysoPC(20:3(5Z,8Z,11Z)/0:0) Previous studies reported that glycerophospholipids and sphingolipids could be biomarkers for monitoring patients with CRC [49]. The upregulation of LysoPC(20:3(5Z,8Z,11Z)), LysoPE(0:0/22:4) and LysoPE(0:0/20:2) were considered positively related to the anti-colon cancer effect of Gegen Qinlian decoction + PD-1 therapy in mice [50]. In this study, LysoPC(20:3(5Z,8Z,11Z)) levels were reduced by CRC-FMT. Ileibacterium abundance was positively correlated with linoleoyl ethanolamide. The anti-inflammatory effects of linoleoyl ethanolamide have been reported in obesity and dermatitis [51, 52]. In the present study, the level of linoleoyl ethanolamide was significantly increased in the HC-FMT group and decreased in the CRC-FMT group, suggesting linoleoyl ethanolamide might be involved in the anti-inflammatory effects of HC-FMT.

In the present study, differential metabolites were found to be significantly associated with the functions of pyruvate metabolism, glycolysis/gluconeogenesis, metabolism of glycine, serine and threonine, tryptophan metabolism, aminoacyl-tRNA biosynthesis, and steroid hormone biosynthesis. Several studies have reported abnormalities in intestinal microbial pyruvate metabolism [53] and glycine, serine, and threonine metabolism [54] in patients with CRC, but larger cohort studies may be needed to establish associations between CRC and these metabolic pathways. Furthermore, glycolysis is a typical feature of glucose metabolism in tumor cells [55]. The decrease in tryptophan content has been reported to be proportional to disease progression and decreased quality of life in CRC patients [56]. Some reports indicate that CRC is a steroid hormone-sensitive tumor [57]. Meanwhile, aminoacyl-tRNA biosynthesis requires aminoacyl-tRNA synthetases, an essential and universally distributed family of enzymes that play a critical role in protein synthesis, several of which are positively linked to colorectal cancer progression [58]. Based on these studies, targeted therapies against glycolysis/gluconeogenesis, tryptophan metabolism, aminoacyl-tRNA biosynthesis, and steroid hormone biosynthesis have been reported to have potential in the treatment of CRC. Further studies could explore the mechanism of action of these metabolic pathways in CRC development and treatment and provide guidance for the development of new personalized therapeutic strategies.

This study has several limitations that should be considered when interpreting the results. First, while the α-diversity indices did not significantly differ between the CRC, CRA, and HC groups, this result may be due to the similarity in overall richness and diversity, even though the specific composition and abundance of microbial species varied significantly. These differences likely underlie the distinct pathways and functions observed between the groups. The relatively small sample sizes, particularly the smaller number of healthy controls, might have reduced the statistical power, potentially masking subtle differences in diversity and introducing bias in comparisons between healthy individuals and diseased groups. Future studies with larger and more balanced sample sizes are necessary to better capture these differences and strengthen the generalizability of the findings. Additionally, our study’s nine-week FMT treatment revealed significant changes in CRC progression in AOM/DSS-induced mice, but the long-term effects of FMT remain unexplored. While notable microbial and metabolic alterations were observed within this timeframe, the potential for late-emerging effects remains unexamined. Extending the duration of FMT treatment and follow-up in future research is crucial for understanding the long-term implications of these interventions on CRC development and recurrence.

Our study demonstrates that FMT from HC could effectively inhibit CRC progression in mice, while FMT from CRC patients exacerbates the disease. The findings suggest that restoring a healthy gut microbiota might help control CRC progression. Furthermore, our results highlight specific bacterial taxa, such as Akkermansia and Ileibacterium, that might contribute to the anti-tumor effects of HC-FMT, highlighting their beneficial roles in gut health and cancer prevention. Nonetheless, as FMT transplants the complete living gut microbiota from the donor, it could bring the danger of introducing multidrug-resistant bacteria and transferring unidentified pathogenic objects. Numerous cases and studies have documented postoperative norovirus enteritis, E. coli bacteremia, cytomegalovirus (CMV) infection, fungal and parasite contamination, along with additional challenges following FMT [59]. In the future, prospective clinical trials are necessary to evaluate the efficacy and safety of FMT as a therapeutic for colon cancer.