Male Balb/c mice aged 8 weeks were purchased from Beijing Vital River Lab (Beijing, China) and acclimated for 1 week in SPF laboratory animal center (Suzuki et al., 2006). In week 1 of the experiment, the mice were intraperitoneally injected with the carcinogen azoxymethane (AOM, 10 mg/kg, Sigma). In week 2, the mice were free to drink water containing the proinflammatory agent dextran sodium sulfate (DSS, 2.5%, MP Biomedicals). In week 3 and week 4, the mice were given drinking water without DSS. Such treatment from week 2 to week 4 was repeated for two additional cycles (Figure 1A) to establish the CRC mouse model as described previously (Neufert et al., 2007). The mice were randomly allocated into three experimental groups, 4 mice per cage, 5 cages in each group. The study was approved by the Ethics Committee of the Harbin Medical University (HMUIRB20180013).

Fecal samples for FMT were freshly collected from healthy, untreated and age-matched Balb/c donor mice. The collected fecal samples were thoroughly mixed in PBS (200 mg/ml) and centrifuged for 1 min at 1,500 rpm. The supernatant was then separated for transplantation. Each recipient mouse received 1 ml of the microbial solution by enema once every 3 days for 9 weeks. PBS enema of equal volume was performed as a negative control for FMT.

Paraffin sections of colorectal tissues were stained with hematoxylin for 2 min, and then with eosin for 1 min. After dehydrating and hyalinizing, respectively, with ethanol and xylene, the tissues were sealed with neutral resin, and then imaged.

The colorectal tissues freshly dissected in RPMI 1640 medium (HyClone) were stained in dark with 30 μM dihydroethidium (DHE, Invitrogen) for 5 min at room temperature. The tissues were then washed and immediately imaged by laser scanning confocal microscope at 420/590 nm. DHE exhibited blue-fluorescence in the cytosol. Once it was oxidized and intercalated with the DNA, fluorescence in the cell nucleus would be turned into a bright fluorescent red color. The relative red fluorescence intensity was analyzed by fluorescence microscope analysis system (Olympus).

The mice were kept for 1 week without any treatment after the completion of the whole FMT plan in order to exclude the negative impact of residual transplant fluid on the analysis of the reconstructed intestinal flora. Then, the mice were transferred to a clean cage for the collection of fresh fecal samples. Two tubes of feces were collected per cage, and 5–6 tubes of feces from different cages in same experimental group were used for high-throughput sequencing of intestinal microbiota (GENEWIZ, Suzhou, China). The PacBio sequencing platform was employed to obtain the full-length of 16S rDNA for its advantages of longer length of gene reads and higher accuracy. Accordingly, the data for analysis included five or six replicates in each experimental group and were submitted to NCBI (PRJNA606977). Microbial diversity was analyzed based on the representative OTU sequence with 97% similarity.

Paraffin sections were dewaxed in xylene and rehydrated in ethanol, respectively. Antigen retrieval was performed in a microwave oven for 2 min at high power and then for 10 min at low power, followed by cooling down to room temperature. The activity of endogenous peroxidase was quenched with 3% H₂O₂ for 10 min. Tissue blocking was performed with goat serum for 1 h at room temperature. Tissue sections were then incubated at 4°C over-night with primary antibodies for T lymphocytes (CD3, ARG52744, Arigo), helper T lymphocytes (CD4, ab183685, Abcam), cytotoxic T lymphocytes (CD8, ab217344, Abcam), Treg lymphocytes (FOXP3, GB11093, Servicebio), natural killer cell (CD49b, AGR57601), or macrophages (F4/80, ARG22476, Arigo). The sample were then incubated with horseradish peroxidase labeled secondary antibody (1: 500) for 1 h at 37°C. Color development was performed using the DAB peroxidase substrate kit (ZSGB-BIO) according to the manufacturer’s instruction. The sections were finally dehydrated in ethanol and xylene, and sealed with neutral resin. The image signal recorded by microscope was converted into digital signal by Image J to analyze the distribution and staining intensity of positive markers.

In order to explore the effect of intestinal microbiota on the intestinal inflammatory responses in CRC mice, we detected the expressions of genes coding for the cellular mediators of inflammation and immunity by the Mouse Cancer Inflammation & Immunity Crosstalk RT² Profiler PCR Array (QIAGEN). The genes are as follows: Ackr3, Aicda, Bcl2, Bcl2l1, Ccl2, Ccl20, Ccl22, Ccl28, Ccl4, Ccl5, Ccr1, Ccr10, Ccr2, Ccr4, Ccr5, Ccr7, Ccr9, Cd274, Csf1, Csf2, Csf3, Ctla4, Cxcl1, Cxcl10, Cxcl11, Cxcl12, Cxcl2, Cxcl5, Cxcl9, Cxcr1, Cxcr2, Cxcr3, Cxcr4, Cxcr5, Egf, Egfr, Fasl, Foxp3, Gbp2b Gzma, Gzmb H2-D1, H2-K1, Hif1a, Ido1, Ifng, Igf1, Il10, Il12a, Il12b, Il13, Il15, Il17a, Il1a, Il1b, Il1r1, Il2, Il22, Il23a, Il4, Il5, Il6, Irf1, Kitl, Mif, Myc, Myd88, Nfkb1, Nos2, Pdcd1, Ptgs2, Spp1, Stat1, Stat3, Tgfb1, Tlr2, Tlr3, Tlr4, Tlr7, Tlr9, Tnf, Tnfsf10, Trp53, and Vegfa.

Tissue proteins from each sample were denatured in 5× SDS-PAGE loading buffer and electrophoresis was performed on 12% tris-glysine agarose gels. The separated proteins were transferred to nitrocellulose membrane followed by blocking with 5% non-fat milk for 2 h at room temperature. The membrane was then incubated at 4°C over-night using primary antibodies for IFNg (ARG21502, Arigo), IL1a (ARG56667, Arigo), IL6 (ARG56625, Arigo), IL12a (bs-0767R, Bioss), IL12b (bs-10641R, Bioss), IL17a (ARG55256, Arigo), TNFa (ARG10158, Arigo), IL10 (ARG21475, Arigo), TGFb (ARG10002, Arigo), CXCR4 (ARG54674, Arigo), STAT3 (ARG54150, Arigo) and Tubulin (AC015, Abclonal). This was followed by incubation with horseradish peroxidase labeled secondary antibody (AS028, AS014, Abclonal; bs-0296G-HRP, Bioss) for 2 h at room temperature. Band intensity was measured by Tanon-5200 and Tanon MP.

Statistical analysis was performed by the Student–Newman–Keuls-q test and Log-rank test, and the statistical comparisons were performed by Graphpad Prism v5. All data were expressed as mean ± SD. A two-tailed value of p < 0.05 was taken to indicate statistical significance.

The schedule of making a colorectal cancer mouse model and performing intestinal microbiota transplantation are shown in Figure 1A. A large number of cancerous foci were observed in the intestine of the CRC mice, with the average number of 17.6 and the average diameter of 4.4 mm, whereas in CRC mice that accepted FMT, the average number of the foci was reduced to 9.8 with the average diameter of 2.3 mm (Figures 1B–D). The CRC mice showed the typical characteristic of intestinal tissue shortening induce by inflammatory adhesion and epithelial hyperplasia, while FMT significantly inhibited the intestinal tissue shortening (Figure 1E). Disordered intestinal crypts with cell dysplasia and hyperplasia could be clearly observed in the CRC mice with or without FMT (Figure 1F). The loss of body weight is a significant feature of CRC. The results showed that the body weight of CRC mice was significantly less than that of mice in the normal control group at the same feeding time point, with a difference of as much as 4.5 g at the end of the experiment. However, the difference was reduced by the treatment of FMT (Figure 1G). The survival of CRC mice treated with FMT prolonged significantly from the fifth week. At the end of the experiment, this value was elevated from 55% (11/20) to 80% (16/20) by the treatment of FMT (Figure 1H). A large amount of superoxide is produced during the rapid division of cancer cells, which in turn greatly facilitates carcinogenesis. In CRC mice, the release of superoxide was significantly increased in the intestine in comparison with the mice in the control group, as indicated by the bright DHE fluorescence around the crypts, whereas the release was significantly reduced in CRC mice accepted FMT, as shown by the dim red fluorescence (Figure 1I).

A total of 291,444 gene reads were obtained, with an average of 12,301 reads and 3,589 OTUs per sample after being processed by the quality filters. The intestinal microbiota was categorized to 15 phyla, 23 classes, 31 orders, 60 families, 148 genera and 203 species. In comparison with normal control mice, the abundance of intestinal microbiota was significantly altered in CRC mice at the level of phyla, family, genus and species, respectively (Figures 2A–D). Anosim analysis showed that the microbial diversity was also statistically different between mice of the CRC and control groups (Figures 2E,F). By the treatment with FMT, the microbial abundance and diversity of CRC mice were both significantly improved as shown by the shannon index analysis (Figure 2G). The phylogenetic structure of the microbiota in CRC mice was also significantly different from that in normal mice, but FMT altered the microbial structure in CRC mice toward that of the normal control mice (Figures 2H,I). Compared with CRC mice, PBS treated mice showed no significant improvement in the abundance, diversity and phylogenetic structure of intestinal microbiota.

Based on 16 s rDNA sequencing data, the classification of intestinal bacteria is generally accurate to the genus level. At the genus level, the dominant bacteria in mice of the normal control group included those in Lactobacillus (9.19%), Alistipes (3.69%), Alloprevotella (4.99%), Ruminiclostridium (1.50%), Saccharimonas (1.11%), Oscillibacter (0.55%), Roseburia (0.47%), Muribaculum (0.45%), Anaeroplasma (0.44%), Arthromitus (0.27%), and genera of Muribaculaceae (25.53%), Lachnospiraceae (20.52%), Prevotellaceae (7.51%), Ruminococcaceae (4.55%) and Erysipelotrichaceae (0.18%), which were all significantly decreased in the CRC mice. After FMT, most of these taxa increased and approached the normal levels. In contrast, some bacteria accounted for large proportions of the gut microbiota in CRC mice, including Akkermansia (40.72%), Bacteroides (13.83%), the E. coli complex (9.79%), Clostridium (0.60%), Turicibacter (0.54%), Azospirillum (0.29%), Paeniclostridium (0.20%), Mucispirillum (0.16%), Parabacteroides (0.15%), Erysipelatoclostridium (0.14%), and genera of Rikenellaceae (0.41%), some of which could be regarded as the marker of CRC mice through the analysis of receiver operating characteristic curve, that were Akkermansia, Escherichia-Shigella, Rikenellaceae RC9-gut-group, Clostridium_sensu_stricto_1, Parabacteroides and Turicibacter (Figures 2C,J). The proportions of these dominant taxa were significantly reduced after the treatment of FMT. Taking the advantage of detecting longer gene sequences by PacBio platform for more accurately classification of bacteria at species level, we obtained 90 species in mice of the control group, including Lactobacillus murinus (0.24%), Bacteroides acidifaciens (0.12%), Lactobacillus reuter (0.77%), Azospirillum sp.-47–25 (0.02%), and Clostridium sp.-Clone-27 (0.42%), with 113 additional taxa (OTUs) remaining to be classified for a formal taxonomic status at the species level since strains containing these gene sequences have not been included in the current database. Most notably, bacteria that are not colonized in normal mice and FMT treated mice, were detected in CRC mice such as Clostridium_perfringens (0.20%), Escherichia_coli_DEC14A (0.06%), Odoribacter_sp._N54.MGS-14 (0.02%), and Escherichia_coli_TOP291 (0.02%; Figure 2D).

Although the microbial disorder of CRC mice has been largely reversed, some intestinal bacteria are still difficult to colonize in the intestine through FMT, that were Alloprevotella, Ruminococcaceae_UCG-014, Lachnospiraceae_UCG-001, Marvinbryantia, and Eubacterium-xylanophilum-group as revealed by LefSe analysis (Figure 2K). It is necessary to isolate and culture these bacteria for multiple transplantation to increase the probability of their colonization.

Intestinal immune cells are located at intestinal epithelium that separates the body from outside environment as an impermeable barrier. Due to this specific location, these cells directly contact enterocytes and are in proximity to antigens in the gut lumen. The dysbiosis of intestinal microbiota may trigger immune disorder via abnormal recruitment of immune cells in the intestine. In CRC mice with intestinal microbial dysbiosis, we found that the infiltration of nearly all T cells were significantly increased as shown by CD3 labeled positive cells, indicating robust immune responses. These increased tumor-infiltrating lymphocytes include CD4⁺ T cells, Foxp3⁺ Treg cells. In addition to adaptive immunity, innate immunity was also involved, manifested as elevated accumulation of F4/80⁺ macrophages and CD49b⁺ NK cells. Interestingly, the enrichment of CD8⁺ T cells and much more aggregation of CD4⁺ T cells, CD49b⁺ NK cells were observed in CRC mice accepted with FMT, as well as reduced population of Foxp3⁺ Treg cells, F4/80⁺ macrophages, indicating the augmentation of anti-cancer efficacy (Figure 3).

On intestinal tumorigenesis, intestinal microbe translocate across altered epithelium and stimulate the recruitment of immune cells to release cytokines. These molecules within CRC microenvironment are sensitive indicators of microbiota-host homeostasis. In this study, dozens of genes enriched in the intestine of CRC mice were identified, including pro-inflammatory factors Ifng, Il1a, Il6, Il12a, Il12b, Il17a, and Tnf; immunosuppressive factors Il13 and Tgfb1; enzymatic modulators Gzma; chemokine and interleukin receptors Ccr5, Cxcr3, and Cxcr4; other cytokines Spp1 and Tnfsf10; toll-like receptor Tlr3; transcription factors Foxp3, and Stat3. By the treatment with FMT, the gene expressions of some molecules were altered significantly, such as suppressed expressions of Il1a, Il6, Il12a, Il12b, Il17a, Tgfb1, Cxcr4, Spp1, Foxp3, Stat3, and increased expressions of Tnf, Il10, and Ccr5 (Figure 4A).

The differential expressions of these genes were further confirmed at protein level. In comparison with the normal control mice, the protein expressions of IL1a, IL6, IL12a, IL12b, IL17a, TGFb, STAT3, TNFa, IFNg and CXCR4 were significantly elevated in CRC mice, resulting in severe inflammatory responses manifested as hematochezia and lethargy, as well as activation of oncogene transcription. Interestingly, in mice received FMT, the expressions of these proteins were significantly altered, including down-regulation of pro-inflammatory factors IL1a, IL6, IL12a, IL12b, IL17a and up-regulation of anti-inflammatory factor IL10, contributing to obvious suppression of intestinal inflammation. Furthermore, the repressed expressions of growth factor TGFb, transcription factor STAT3 and elevated expressions of TNFa, IFNg, CXCR4 together promoted the anti-cancer efficacy (Figure 4B).

The expressions of pro-inflammatory factors IL1a, IL6, IL12a, IL12b, IL17a, growth factor TGFb and transcription factor STAT3 were positively correlated with Azospirillum sp.-47–25, Clostridium sensu stricto-1, Escherichia-Shigella, Akkermansia, Turicibacter, Bacteroides, Mucispirillum, Rikenellaceae RC9_gut_group and Erysipelatoclostridium. It was indicated that the expressions of TGFb and STAT3 were affected by intestinal bacteria related to inflammation and possibly the interaction between inflammatory environment and these microbe may promote the carcinogenic process. Therefore, these microbe were of great potential to serve as anti-inflammation and anti-cancer targets. In contrast, their expressions were negatively correlated with Ruminococcaceae UCG-014, Ruminiclostridium_9, Muribaculum, Anaeroplasma, Candidatus_Arthromitus and Candidatus_Saccharimonas, which show the potential to be developed as microecological agents for the treatment of colitis and CRC. In addition, the abundance of Muribaculum and Candidatus_Arthromitus in PBS treated mice were relatively higher than that in CRC mice, which are easy to colonize in the intestine, and PBS washing provide them more space for colonization. Their increased colonization may be the reason for the decreased protein expressions of IL12b and TGFb in PBS treated mice. Moreover, the expressions of cytokines which were able to directly kill cancer cells such as TNFa, IFNg and CXCR4 were positively correlated with Odoribacter, Lachnospiraceae UCG-006, Desulfovibrio, and negatively correlated with Alloprevotella, Ruminiclostridium, Prevotellaceae UCG-001, Oscillibacter and Lachnospiraceae NK4A136. These intestinal bacteria also show the potential to be developed as microecological agents and anti-cancer targets (Figures 4C,D).