Azoxymethane and Dextran Sodium Sulfate were purchased from MP Biomedicals (Santa Ana, CA, USA). AB23A was purchased from Must Bio-Technology Co., Ltd (Chengdu, Sichuan, China.) ELISA kits were purchased from ABclonal Biotechnology (Wuhan, Hubei, USA). Protein lysis buffer was purchased from Beyotime (Shanghai, China). PVDF membrane was purchased from Millipore (Boston, MA, USA). ECL reagent kit was purchased from Thermo Fisher Scientific (USA). The information of all antibodies used in this study were shown in Supplementary Table 1 .

The Animal Ethics Committee of Fujian Medical University inspected and approved the animal experiments. (No: FJMU IACUC 2019-0066) and the operations on mice were carried out in compliance with Fujian Medical University’s ethical methods and programs for the use of laboratory animals. Shanghai Slake Laboratory Animal Co., Ltd. provided the mice (C57BL/6J male mice, 18-20 g) that needed in this study and raised in accordance with the ethical methods and programs of Fujian Medical University on the use of laboratory animals. Great attempts were made to keep animal cruelty to a minimum.

All experimental mice were placed in an SPF-level animal barrier. During animal experiments, the temperature in the animal laboratory was 22 ± 2°C, and the light and dark conversion cycle was 12 hours. After one week of free breeding and adaptation to the animal barrier environment, 30 mice were divided into three classes at random: a control group (n =10), an AOM/DSS-induced model (DSS) group (n =10), and an AB23A treatment group (n =10).

At the beginning of the experiment, mice in the DSS and AB23A groups received an intraperitoneal injection of AOM (10 mg/kg); mice in the normal group were intraperitoneally injected with physiological saline (10 mL/kg). On the seventh day, mice in the DSS and AB23A groups were given water containing 3% (w/v) DSS. After 7 days, normal sterile drinking water was provided for 14 days. The control group was given sterile drinking water throughout this procedure. This cycle was repeated thrice until mature tumors developed. The entire experiment period was 70 days. The AB23A group began intragastric administration of AB23A (50 mg/kg) and solvent (10% Captisol) daily on the first day of the experiment and the other two groups were given a corresponding dose of solvent (10% Captisol) daily (Meng et al., 2015; Meng et al., 2017). An overview of the animal experiment design is provided in the Figure 1A . Over the course of the experiment, mouse weights, stool shapes, and disease status were noted every day. At the end of the experiment, the mice were euthanized for sample collection.

For biochemical analysis, blood samples were centrifuged at 4000 g for 10 minutes at 4°C, transferred to EP tubes, and processed at 80°C. for biochemical analysis. The colon was carefully cut longitudinally to collect its contents. The colon was quickly washed with pre-chilled saline, and the number and location of tumors were recorded. One part of the pre-chilled colon tissue was fixed in 10% anhydrous formalin for HE, and another part was stored at −80°C for subsequent biochemical determination. During the experiment, all experimental tissue samples are performed on ice for as long as possible during operation.

The disease activity index (DAI), which is commonly used to evaluate the progress of CAC, was calculated according to the weight change of the mouse (such as whether it has increased or decreased), the external characteristics of the stool, and the blood in the stool. The range of DAI is 0 ~ 4. The detailed evaluation criteria are provided in Supplementary Table 2 .

The colon is placed on a blue background plate and the length will be measured on ice in time. The tissue was then fixed in tissue fixation fluid (10%) for one day for histological examination via staining with hematoxylin–eosin (HE). The histological scores ranged from 0 to 10 according to the extent of colon injury. The total scoring system was based on four parameters, namely, mucosal epithelium, cell infiltrate and edema, integrity of crypts, and goblet cell depletion. The detailed evaluation criteria are provided in Supplementary Table 3 .

Colon tissues that have been fixed in formaldehyde solution (10%) for over 24 h were washed thrice with PBS and then placed in ethanol for gradient dehydration for 1 h each time. Subsequently, the colon tissue was obtained, placed in xylene I and II solutions for transparent treatment, and then allowed to stand for 1 h. Afterward, paraffin embedding was carried out at 56-58 °C. The samples were sliced into sections after cooling at room temperature, placed in warm water for unfolding, mounted on glass slides, and then dried. The sections were soaked once more in xylene I and II solutions for dewaxing treatment, and immersed in 100%, 95%, 85%, and 75% ethanol for 1 min each time for rehydration. The parts were stained for 5 minutes with hematoxylin, then distinguished for 5-10 seconds with a 1 percent HCl solution, washed twice with water, and stained for 2 minutes with eosin. Ethanol dehydration treatment was carried out once more, followed by xylene I and II soaking for 3 and 5 min, respectively. Finally, the film was sealed with neutral gum and observed under a microscope.

The relative expression of inflammatory cytokines, including IL-6, TNF-α, IFN-γ, and IL-10, in serum and colon tissue was measured with the ELISA kit (ABclonal Biotechnology Co., Ltd., USA). Ice-cold normal saline and colon tissue were mixed at a ratio of 1:10 (w/v) and homogenized, then centrifuged at 5000 g for 15 minutes, and the supernatant was taken to measure inflammatory cytokines.

The purposed protein from the colon tissue were extracted via using tissue lysis buffer and denatured with sample loading buffer (5×) at 100°C. Total proteins were denatured and electrophoresed in 10% SDS-PAGE before being moved to the PVDF membrane. The membrane was blocked with blocking buffer for 2 hours at normal room temperature. and then incubated overnight on ice with the primary antibody (dilution, 1:1000). Thereafter, wash the membrane with TBST, 3 times for 10 minutes each time. Then incubation with the corresponding secondary antibody (dilution: 0.5:5000) was carried out for 2 hours at normal room temperature. The ECL reagent kit was used to detect the expression of the protein in the bands, and Image J software was used to show striped images (NIH, USA) and detected the luminous intensity of the bands. β-Actin was used to present the standard of same loading of protein.

Fecal samples of mice were collected on the 70th day and used for gut microbiota sequencing. Anal stimulation was used to stimulate mice to defecate. When the mouse’s defecation is about to end, the stool is immediately clamped with sterile forceps and placed in a sterile tube, and then immediately stored in a liquid nitrogen tube. The collected feces are finally stored in a refrigerator at -80°C. The statement in detail of our sequencing of feces of CAC mice can be find in Supplementary Material .

Bioinformatics analysis was performed through NovoMagic, a data analysis platform independently developed by Novogene Co., Ltd. Analysis tools, algorithms, codes, and data are visible in the Supplementary Material .

All data were expressed as mean ± SD. Multiple comparisons were performed using one-way ANOVA. When P < 0.05 or P < 0.01, the data were considered statistically significant. The experimental data calculation and analysis were operated via GraphPad Prism software (version 8.3.0).

Figure 1 shows that AB23A administration significantly ameliorates AOM/DSS-induced CAC, as evidenced by marked reductions in weight loss ( Figure 1C ), decreased tumor burden ( Figure 1D ), significantly lower DAI scores ( Figure 1E ), and relief of colonic shortening ( Figure 1F ). The incidence of dysplasia in each group was: 0% in the normal group; 100% in the DSS group and 100% in the AB23A group. Although the dysplasia rate of the AB23A intervention group was not improved compared with the DSS group, the size and number of colon tumors in mice were reduced. The tumor burden of the DSS group was significantly more severe than that of the AB23A group. Compared with the control group (0.00 ± 0.00), the DSS group showed a significant increase in colon tumor number (28.67 ± 4.23) whereas the AB23A group demonstrated a significant reduction in tumor number (13 ± 4.73; Figure 1G ). These results indicate that the CAC model had been successfully constructed and illustrate the primary macroscopic protection effect of AB23A. Compared with those of the control group, the colon tissue sections of the DSS model group showed very obvious pathological changes, including the destruction of colonic epithelial cells, distortion of crypts, the loss of goblet cells, colon epithelial damage, and varying inflammatory infiltration ( Figure 1H ). By contrast, the AB23A group revealed a marked reduction of these symptoms. The histological scores can directly reflect the severity of colon pathological damage ( Figure 1I ). The histological scores of the DSS group were clearly much higher than those of the control and treatment groups. This result indicates that intervention with AB23A alleviates the clinical signs and tumor development of AOM/DSS-induced CAC.

A wide variety of research shows that chronic inflammation is one of the primary risk factors for initiation, proliferation, or metastasis of cancer. Considering the inter-triggering mechanism of cancer progression and inflammation, we assessed the production of pro-inflammatory cytokines, including IL-6, TNF-α, and IFN-γ, and the anti-inflammatory cytokine IL-10 in CAC mice in colon tissue and serum. In the serum, compared with those in the DSS group, the pro-inflammatory cytokines in the AB23A group manifested significantly decreasing trends, especially IL-6 (P < 0.01), TNF-α (P < 0.05), and IFN-γ (P < 0.01). AB23A intervention significantly increased the expression of IL-10 (P < 0.05; Figure 2A ). In colon tissue ( Figure 2B ), pro-inflammatory cytokines in AB23A group, such as IL-6 (P < 0.05), TNF-α (P < 0.01), and IFN-γ (P < 0.05) significantly decreased compared with the DSS group, while the expression of IL-10 (P < 0.01) was suppressed. Thus, AB23A markedly alters the cytokine profile of AOM/DSS-induced CAC mice by suppressing the expression of proinflammatory-related cytokines and upregulating the expression of anti-inflammatory cytokines to relieve gut inflammation, which is considered an inducing factor of colon cancer.

Alterations in gut microbiota were found by the sequencing of the 16s rRNA genes extracted from the control, DSS, and AB23A groups. According to the sequencing results, 89,936 tags were detected in each sample on average. Through the evaluation of the quality of tags, the effective data for each sample is 86,363 tags. Finally, the effective ratio of quality control reached 66.68%. The obtained sequence was clustered into an OTU unit with 97% identity, and finally divided into 1113 OTU units. The OTU sequences were annotated to species. We found that the ends of the dilution curves of all samples tended to be flat ( Figure 3A ), thus indicating that the amount of our sequencing data is reasonable, that the distribution of species richness and uniformity in the sample is relatively concentrated, and that no particularly discrete sample is present.

α-Diversity indices, including Chao1, Shannon, Observed_species, Simpson, and ACE, are commonly used to for evaluating the richness and diversity of microbiota. The Simpson index has three display forms, namely Simpson’s Index (D), Simpson’s Index of Diversity (1-D), and Simpson’s Reciprocal Index (1/D), they have similar effects on reflecting community diversity but the calculated results are in different forms. We used the Simpson’s Index of Diversity (1-D) display format which reflected the trend that is consistent with the Shannon index. As shown in Figures 3B, C , the diversity indices, including the Shannon and Simpson indices, manifested similar tendencies. The DSS group had the lowest other α-diversity indices of the three groups, and the other two groups had similar values. The other α-diversity indices were presented in the supplementary material ( Supplementary Table 4 ). This result indicates significant microbiota destruction in the DSS group. In summary, AB23A intervention can obviously improve the abundance and diversity of the gut microbiota.

β-Diversity analysis was used to assess the variance of diversity among different groups. We performed β-diversity analysis to generate principal component analysis (PCA) and PCoA plots. As shown in Figures 3D, E , PCA and PCoA yielded clearer clusters in the control and AB23A groups than in the DSS group. β-Diversity analysis revealed Species distribution in the AB23A group tended to be closer to the control group.

To determine which bacteria are altered by AB23A and, in turn, affect disease progression in CAC, we analyzed the top 10 species with the highest abundance at different taxonomic levels. Compared with that in the control group, Proteobacteria increased whereas Firmicutes decreased at the phylum level in the DSS group ( Figures 4A, B ). At the family level, the proportions of Enterobacteriaceae, Akkermansiaceae, and Muribaculaceae increased whereas the proportions of Prevotellaceae, Rikenellaceae, Ruminococcaceae, and Bacteroidaceae decreased in the DSS group compared with those in the control group ( Figures 4C, D ). The proportions of these families in the AB23A group were roughly similar to those in the control group. We selected the top 35 genera with the highest abundance ( Figure 4E ) for in-depth analysis of differences among groups. The abundance and diversity of these genera in the DSS group clearly decreased. Compared with the control and AB23A groups, the DSS group showed higher proportions of Klebsiella, Citrobacter, and Akkermansia and lower proportions of Bacteroides, Alloprevotella, and Lactobacillus. The DSS administration increased the ratio of Proteobacteria to Firmicutes sharply ( Figure 4F ). AB23A intervention reversed this increase and restored the ratio of Proteobacteria to Firmicutes to levels similar to that in the control group.

In order to find the differential flora regulated by AB23A, LEfSe was performed and investigate the influences of AB23A on the gut microbiota of CAC mice. LEfSe analysis indicated that Proteobacteria (class Gammaproteobacteria), Enterobacteriales (order and family levels), Enterobacteriaceae, Muribaculaceae, and Klebsiella (genus and species levels) are the main kinds of bacteria that contribute to gut microbiota dysbiosis in the CAC mice of DSS group compared with the control group. ( Figure 5A ). Between DSS and AB23A, the microbiota in the AB23A group displayed the enrichment of Bacteroidetes at the genus and family levels, Clostridia (class and order levels), Ruminococcaceae, Bacteroides_acidifaciens and unidentified_Ruminococcaceae ( Figure 5B ). The relative abundances of significantly altered taxa are visualized in Figures 5C, D . In comparison to the two other categories, DSS administration induced a large rise in certain pathogens in Proteobacteria (e.g., Klebsiella, Citrobacter, and Enterobacteriaceae). At the same time, some SCFA-related bacteria (e.g., Rikenellaceae and Prevotellaceae) and potential beneficial bacteria, such as Bacteroides acidifaciens, Bacteroides vulgatus, and Lactococcus, decreased.

We studied TLR4, TLR5, NF-κB, and MAPK, the key efficient molecules that mediate pro-inflammatory signaling, in order to provide more insights into the modulation of inflammatory responses by AB23A. Activation of TLR4 (P < 0.05), TLR5 (P < 0.05) and MyD88 (P < 0.05) was inhibited by AB23A in CAC mouse colon tissue ( Figure 6A ). AB23A treatment led to consistent and remarkable reductions in the activation of IKKα (P < 0.05), IκBα (P < 0.05), and p65 (P < 0.01), the significant molecules needed to activate the NF-κB ( Figure 6B ). Moreover, AB23A greatly reduced the phosphorylation of p38-MAPK (P < 0.05), ERK (P < 0.05), and JNK (P < 0.05), the key MAPKs implicated production of pro-inflammatory mediators ( Figure 6C ). The relative protein expression results are shown in Figures 6D–F . Taken together, our data indicate that AB23A confers profound protective effects against CAC in mice by reducing inflammatory responses through inhibition of the activation of the NF-κB/MAPK/TLR/MyD88 signaling pathways.

As shown in Figure 7A , MUC-2 expression in CAC mice of DSS group was substantially decreased in comparison to the mice in the control group. (P < 0.01). However, pretreatment with AB23A significantly increased the expression of this protein (P < 0.05). AB23A intervention also markedly up-regulated the expression of tight junction (TJ) proteins, including junctional adhesion molecule A (JAM-A; P < 0.01), claudin-1 (P < 0.05), ZO-1 (P< 0.05), and occludin (P < 0.01), as compared with those in the DSS group ( Figures 7B–F ). Our data show that AB23A regulates the expression of TJ proteins and reduces intestinal permeability to maintain intestinal integrity and functionality.