Introduction

Colorectal cancer (CRC) ranks as the second most prevalent c. cause of mortality among cancer-related deaths1. Colitis-associated colorectal cancer (CAC), a sub-type of CRC, accounts for approximately 2% of CRC cases and 10–15% of patients with inflammatory bowel disease. Colitis is a chronic inflammation condition that has an increasing prevalence, impacting over five million worldwide2. The persistent inflammation from colitis increases the risk of progression to CRC by promoting cancer-favorable conditions3. 5-Fluorouracil (5-FU) is a chemotherapeutic agent commonly used for CRC, impeding DNA replication and repair processes4,5. However, 5-FU administration is associated with toxicities and adverse side effects, including fever, fatigue, mucositis, stomatitis, nausea, vomiting, and diarrhea4,6,7. 5-Fu exacerbates the colitis symptoms and disrupts the gut microbiome, potentially further complicating the outcome and quality of life for patients with CAC8. The imbalance of gut microbiota would lead to chronic inflammation, disruption of host immunity and accumulation of virulence metabolites or toxins, fostering a diseases environment which promotes both the initiation and progression of CRC9,10.

Theabrownin (TB) and zearalenone (ZEA) have previously demonstrated anti-CRC properties11,12. TB is a fungal fermentation product derived from tea polyphenols extracted from Pu-erh tea13,14. During fermentation process, fungi species Aspergillus tubingensis, Aspergillus marvanovae, and Aspergillus fumigatus convert tea polyphenols, theaflavin and thearubigin, into TB. Previous studies demonstrated that TB attenuated CRC in the AOM/DSS model by suppressing the PI3K/Akt/mTOR pathway and increasing the abundance of SCFA-producing Prevotella and Alloprevotella11. Similarly, ZEA is a secondary metabolite from fungal species Fusarium and Gibberella15,16. They have previously been shown to suppress CRC in both in vivo and in vitro studies by regulating the Ras/Raf/ERK1/2 pathway and enriching SCFA-producing bacteria12,17,18,19.

Given their distinct mechanisms of action, it was hypothesized that TB and ZEA as a mixture could exert a synergistic effect against CAC through targeting oncogenic pathways, promoting gut microbiota composition balance. Furthermore, the anti-inflammatory effect and gut microbiota modulatory effect of TB and ZEA might counteract the adverse effects of 5-Fu. Subsequently, the study aimed to determine if the alone, mixture or in combination with 5-FU could inhibit AOM/DSS-induced colorectal carcinogenesis more effectively. Additionally, the study aimed to elucidate how TB and ZEA mixture affects CRC progression and the gut microbiome of AOM/DSS mice.

Methods

Chemicals and antibodies

Azoxymethane (AOM) was obtained from Sigma-Aldrich (St. Louis, MO, USA). Dextran sodium sulphate (DSS) was acquired from TdB Labs (Uppsala, Sweden). TB (> 99.3% purity) was bought from PURENZHIZAO (Yunnan, China). ZEA was attained from Sigma-Aldrich (St. Louis, MO, USA). The primary antibodies were purchased from Abcam (Cambridge, UK) and CST (Massachusetts, USA). The secondary antibodies were bought from Biorad (Hercules, CA, USA).

Animals and experimental setup

Male BALB/c mice at 6 weeks old were obtained from Centre for Comparative Medicine Research (CCMR, HKU). Upon receipt, the animals were housed for one week of acclimation in a 12 h light/dark cycle with normal chow diet and drinking water given ad libitum. The mice were randomly allocated into (1) Healthy control (H), (2) AOM/DSS control (AOM/DSS), (3) 5-FU (F), (4) TZ and (5) TZ + 5-FU. Afterwards, the mice underwent one week of voluntary gel administration training, in which the chow diet was withdrawn overnight, and 0.20 g of MediGel® Sucralose gel (Maine, USA) was provided the next day in individual cages. Once the whole gel was consumed, mice were returned to their house cages with a restored chow diet. The process repeated for two more days without chow diet restriction. The AOM/DSS model was then performed as previously described20. At the beginning of the experiment, mice were injected intraperitoneally with AOM (10 mg/kg) except for the healthy control. After one week, drinking water was replaced with DSS (2.5%) for one week, followed by a week of regular drinking water for three weeks. Regular drinking water was given as control. At week 5, treatments of 5-FU, TZ and TZ + 5-FU began. For 5-Fu and TZ + 5-FU groups, 5-FU (35 mg/kg) was injected intraperitoneally every week21. PBS was injected to other groups as a control. For TZ and TZ + 5-FU, TB (225 mg/kg bw) and ZEA (10 mg/kg bw) were infused into MediGel® Sucralose gel (Maine, USA) and provided to each mouse at 0.20 g every other day until the end of the experiment. Dosage of TB and ZEA were determined from our previous studies11,12. MediGel® Sucralose gel without modification was given to other groups as control (Fig. 1). At week 12, all animals were euthanized via intraperitoneal injection of sodium pentobarbital (150 mg/kg). All animals were confirmed to be deeply anaesthetized and unresponsive, with the absence of reflexes, prior to tissue collection. Animals that reached humane endpoints requiring early euthanasia, as determined by animal facility veterinarians, were also humanely euthanized using the same method and dosage. Colon was harvested and cut open longitudinally to determine the tumor count and colon length. Colon samples were collected. One part of the colon was fixed in 10% formalin for histology analysis and the other was frozen at −80 °C for biochemical analysis. Fecal samples were obtained and frozen at –80 °C on the day of experimental endpoint. Liver samples were weighed as well. All animal experiments were conducted in accordance with institutional guidelines and approved by the Committee on the Use of Live Animals in Teaching and Research of the University of Hong Kong (CULATR No. 5584-20), and this study is reported in accordance with the ARRIVE guidelines.

Fig. 1
figure 1

Schematic diagram of the supplementation of TZ on AOM/DSS mice model. AOM, azoxymethane; DSS, dextran sulfate sodium; and 5-FU, 5-fluorouracil.

Full size image

Colon histological analysis

One part of the colon samples was fixed in formalin, embedded as paraffin blocks and sectioned on slides. After deparaffinization, Haematoxylin and eosin (H&E) staining was performed according to manufacturer’s manual (BASO, Wuhan, China). Similarly, according to manufacturer’s manual (Abcam, Cambridge, UK), Sirius red staining was carried out and the percentage the positively red-stained area over the entire area of specimens was calculated to quantify the degree of colonic collagen fibrosis.

Immunohistochemistry (IHC) staining

IHC staining was performed utilizing the heat-induced antigen retrieval method with sodium citrate (pH6) or Tris-EDTA (pH9). The inhibition of endogenous peroxidase activity was performed with 3% H2O2. CAS-block reagent (Invitrogen, Waltham, MA, USA) blocking was carried out for 1 h before the incubation with primary antibodies. Specimens were incubated with primary antibodies (1:100) at 4 °C overnight and with secondary antibodies (1:250) at room temperature for 1 h, followed by DAB (Abcam, Cambridge, UK) chromogen reaction and haematoxylin counterstaining. The percentage of positively stained areas was determined with ImageJ software (NIH, USA). By multiplying the percentage and the intensity of staining graded from non-stained, 0; weakly stained, 1; moderately stained, 2; and, strongly stained, 3, the histoscore was obtained22.

Cytokines ELISA analysis

Colonic protein extraction was performed with homogenization in RIPA buffer with protease and phosphatase inhibitor (Sigma-Aldrich, St. Louis, MO, USA). extracted protein was collected from centrifugation. The total protein content was measured using DC protein assay (Biorad, CA, USA). ELISA analysis of IL-6, TNF-α and IL-17 A/F were performed using Mouse ELISA MAX™ Set (BioLegend, CA, USA) according to the instructions of manufacturer. The absorbance was measured using SpectraMax iD3 microplate readers (Molecular devices, CA, USA).

Western blot analysis

Colonic protein was diluted and electrophorized in 10% SDS-PAGE gel. After being transferred to polyvinylidene fluoride (PVDF) membrane, it was blocked with either 5% non-fat milk or BSA. For primary antibodies incubation, the membrane was exposed under anti-PI3K (1:1000), anti-p-PI3K (1:1000), anti-Akt1/2/3 (1:1000), anti-p-AKT1/2/3 (1:1000), anti-cyclin D1 (1:10,000) (Abcam, Cambridge, UK) at 4 °C overnight. On the next day, goat anti-rabbit IgG (H + L) HRP conjugate or goat anti-mouse IgG (H + L) HRP conjugate (1:4000, Biorad, CA, USA) were used for secondary antibody exposure for 1 h at room temperature. Protein bands visualization was completed with enhanced chemiluminescence reagents (Biorad, CA, USA) under the ChemiDox XRS + imaging system (Biorad, CA, USA).

Metagenomics

Fecal DNA were collected at the end of the experiment. Using the QIAamp® PowerFecal® Pro DNA kit (Qiagen, Hilden, Germany), fecal microbial DNA was extracted according to manufacturer’s instructions. Extracted DNA was sent to BGI Genomics (Shenzhen, China) for whole genome shotgun sequencing. Upon receipt of clean data, low-quality and artificial adapter reads were filtered out by SOAPnuke23. Bowtie2 was utilized to exclude the mouse contamination (GRCm39) and phix genomes by mapping to their genomes. Taxonomic profiles were obtained using Kaiju with parameter “−e 5” and the latest release (2023-05-10) of NCBI nr database including archaea, bacteria, viruses, fungi, and microbial eukaryotes24. Taxa either presented in less than 10% samples or with a relative abundance of less than 0.01% were removed from the downstream analysis. After aggregating the raw abundance table into species-level counts per million (CPM) table by R package phyloseq alpha and beta diversity were calculated through R package vegan25,26. To test the overall composition difference between groups, dunn test and PERMANOVA (via adonis2 from the R package vegan) were carried out for alpha diversity and beta diversity respectively. ANCOM-BC was used to compare the abundance of microorganisms between AOM/DSS with and without treatment27. Linear discriminant analysis (LDA) was also used to lessen the bias present in the various differential methods28,29. Only the microorganisms with fold change ≥ 2 or ≤ 1/2, LDA score (1og10 transformed) ≥ 2 and FDR ≤ 0.05 were considered differentially abundant. HUMAnN3(v3.7) was applied to concatenated paired-end read files for gene family and metabolic Metacyc pathway abundance estimate30. By using the HUMAnN3 regroup table function, Kyoto Encyclopedia of Genes and Genomes (KEGG)31,32 ontology terms (KO), KEGG module and Metacyc pathway profiles were obtained. To test differential abundances (DA) of KO and metabolic pathway abundance, ANCOM-BC was applied to the raw profiles27. DA KO and DA pathway were estimated at FDR value ≤ 0.1.

Short chain fatty acids (SCFAs) analysis on fecal content

According to previous studies, fecal SCFAs content was measured with gas chromatography-mass spectrometry (GC–MS)33,34. Fecal samples homogenization was conducted in 0.005 M sodium hydroxide with internal standard (10 µg/mL acetic acid-d4). By centrifugation at 13,200 × g at 4 °C for 20 min, the supernatant was collected and mixed with 0.5 mL of 1-propanol/pyridine (3:2, v/v) and 0.1 mL of propyl chloroformate for 1 min under 60 °C incubation in order to derivatize the SCFAs. Followed by the addition of 0.5mL hexane, vortexing and centrifuging at 2000 × g for 5 min. The top layer was collected for GC–MS analysis (Agilent 6890 N-5973 GC–MS, USA) based on the previous studies33. The concentration of SCFAs was attained from the calibration curves constructed using the response ratios of acetic acid, propionic acid and butyric acids against acetic acid-d4.

Statistical analysis

GraphPad Prism 10.2.2 (GraphPad software, San Diego, CA, USA) was used to perform statistical analysis, and the data in this study were presented as mean ± standard deviation (SD). One-way analysis of variance (ANOVA) followed by Tukey’s multiple comparisons test or the Kruskal–Wallis test followed by Dunn’s multiple comparisons test was employed. The correlation between two variables was calculated from the Spearman correlation coefficient. For multiple-testing corrections, a false discovery rate (FDR) (Benjamini–Hochberg) was used to adjust the p value. A p value < 0.05 was recognized as having statistical significance.

Results

Supplementation of TZ attenuated AOM/DSS-derived colorectal tumorigenesis and improved survival rate over 5-FU

TZ and TZ + 5FU supplementation significantly improved survival outcomes compared to the AOM/DSS group (p < 0.05) (Fig. 2A). All treatment groups, including 5-FU, TZ and TZ + 5-FU, significantly reduced total tumor count (p < 0.05) (Fig. 2B, D). In addition, TZ supplementation significantly reduced the amount of the total adenoma count compared to AOM/DSS group (Fig. 3E). Colon shortening, a hallmark of colitis, was recovered in TZ group (p < 0.05) (Fig. 2C). Such a finding is consistent with the disease activity index, in which we found that the TZ group significantly decreased when compared to the AOM/DSS model (Fig. S1). Furthermore, AOM/DSS-induced enlarged caecum weight was reduced in both TZ and TZ + 5-FU groups (p < 0.05) (Fig. 2E). Compared to 5-Fu, which significantly reduced liver weight when compared to the healthy group, supplementation of TZ with 5-FU has preserved liver weight (Fig. 2F).

Histopathological analysis demonstrated significant reductions in colonic crypt depth and hyperplasia score (p < 0.05) in the TZ and TZ + 5-FU groups and inflammation score in the TZ group (p < 0.05) relative to the AOM/DSS group (Fig. 3A–E). The suppression in inflammation was supported by colonic inflammatory cytokine levels. TZ and TZ + 5-FU suppressed colonic IL-6 and IL-17 A/F concentration (p < 0.05) when compared to the AOM/DSS group. All treatment groups significantly reduced the concentration of TNF-α (p < 0.05) (Fig. 2G). In terms of collagen fibrosis, TZ supplementation significantly decreased the percentage of Sirius Red-positive staining, indicating significantly lower levels of fibrotic tissue in the colon compared to the AOM/DSS group (p < 0.05) (Fig. 3F–G).

Fig. 2
figure 2

TZ attenuated colorectal tumorigenesis, suppressed colonic inflammatory cytokines, and improved the survival rate of AOM/DSS mice over 5-FU. (A) Survival rate percentage. (B) Representative macroscopic pictures of dissected colon. (C) Total tumor counts. (D) Colon length. (E) Caecum weight. (F) Liver weight. (G) Colonic concentration of IL-6, IL-17 A/F and TNF-α. n = 6–10. Data are presented as mean ± SD (n = 6–10 per group). Comparisons were performed using one-way ANOVA followed by Tukey’s multiple comparisons test. No “*” indicates p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Full size image
Fig. 3
figure 3

TZ ameliorated colorectal AOM/DSS-induced histopathology. (A) Representative microscopic pictures of colonic sections under H&E staining. (B) Colonic crypt depth. (C) Hyperplasia score. (D) Inflammation score. (E) Adenoma counts categorized by microadenoma, low-grade macroadenoma and high-grade macroadenoma. (F) Representative microscopic pictures of colonic sections under Sirius red staining. (G) Sirius red positive percentage. Data are presented as mean ± SD (n = 6–10 per group). Comparisons were performed using one-way ANOVA followed by Tukey’s multiple comparisons test. No “*” indicates p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Full size image

TZ promoted apoptosis through suppressing proliferative PI3K/Akt pathway

IHC staining was performed to assess the proliferative and apoptotic status. All treatment groups, TZ, 5-FU and TZ + 5-FU, all promoted the H-score of the apoptotic marker, c-casp3 (p < 0.05), while only TZ significantly reduced the proliferative marker, Ki-67 (p < 0.05) when compared to AOM/DSS group (Fig. 4A–C). Western blotting analysis showed that TZ suppressed the phosphorylation of PI3K and Akt (p < 0.05), as well as their downstream transcription factor cyclin D1 (p < 0.05) (Fig. 4D–G). Meanwhile, TZ + 5-FU only reduced the phosphorylation of Akt and the level of cyclin D1 (p < 0.05) (Fig. 4D–G). These results indicated the inhibitory effects of TZ on the PI3k/Akt pathway.

Fig. 4
figure 4

TZ promoted apoptosis by regulating the PI3K/Akt signaling pathway. (A) Representative microscopic pictures of colonic sections under IHC staining in 400x magnification. Immunohistochemistry analysis H-score of (B) Ki67, and (C) c-casp3. (D) Representative immunoblots of p-PI3K/PI3K; p-Akt1/2/3/Akt1/2/3 and cyclin D1. Quantitative analysis of the relative expression of (E) p-PI3K/PI3K, (F) p-Akt1/2/3/Akt1/2/3, and (G) cyclin D1. Data are presented as mean ± SD (n = 6–10 per group). No “*” indicates p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Full size image

TZ-induced changes in microbiota composition and SCFA levels

Short-chain fatty acids (SCFAs), gut microbial metabolites, have been shown to exert a protective effect against carcinogenesis. In this study, supplementation of TZ and TZ + 5-FU significantly elevated fecal acetate and propionate levels (p < 0.05) when compared to AOM/DSS group (Fig. 5A), suggesting a partial restoration of SCFA levels that were otherwise depleted under tumorigenic and inflammatory conditions. Acetate (r = −0.63, p = 0.00006) and propionate (r = −0.37, p = 0.027) levels were negatively correlated with tumor count, indicating that higher faecal acetate and propionate concentrations were associated with reduced tumor burden (Fig. S3A, B).

Given that SCFAs are primarily produced by gut microbiota, we then further investigated the effect of TZ on gut microbiota composition. Notably, the TZ + 5FU group showed a significantly higher observed species number when compared to the AOM/DSS group. However, no significant changes were observed in community diversity (Shannon index) (Fig. 5B). Moreover, Principal coordinates analysis (PCoA) based on PERMANOVA indicated that the microbiota structure differed among groups (R2 = 0.1567, P = 0.032) (Fig. 5C). We next examined the relative abundances of bacterial species across groups. TZ supplementation significantly enriched Helicobacter ganmani, Desulfovibrionaceae bacterium LT0009, Helicobacter sp. MIT 03-1616, and Candidatus Borkfalkia ceftriaxoniphila compared to the AOM/DSS group (all p < 0.05). TZ also suppressed Roseburia sp. CAG:303, Clostridium sp. MD294, Clostridium sp. CAG:510, and Mucispirillum schaedleri (p < 0.05) (Fig. 5D). These results suggest that TZ selectively targeted specific taxa rather than inducing significant shifts in overall community diversity. Notably, in the TZ + 5-Fu group, Lactobacillus johnsonii and Desulfovibrionaceae bacterium LT0009 were further enriched compared to the AOM/DSS group.

We further assessed the association of these bacterial changes with clinical markers, such as tumor count, cytokines, and SCFAs levels. Desulfovibrionaceae bacterium LT0009 negatively correlated with total tumor count and IL-6 levels (p < 0.05). Another TZ-enriched taxon, Helicobacter sp. MIT 03-1616 showed inverse correlations with total tumor count, IL-6 and TNF-α (p < 0.05), while positively correlating with all three major SCFAs, namely acetate, propionate and butyrate, and the colon length (p < 0.05). Similarly, Candidatus Borkfalkia ceftriaxoniphila positively correlated with propionate and butyrate (p < 0.05) (Fig. 5D). Overall, our findings the role of TZ in selectively modulating gut microbial profiles and restoring SCFA levels, which may contribute to reduced CRC tumorigenesis.

Furthermore, Metacyc pathway analysis was performed to predict the biosynthesis and metabolic functions by TZ. TZ supplementation upregulated PWY-7388: octanoyl-[acyl-carrier protein] biosynthesis (mitochondria, yeast) and PWY-7013: (S)-propane-1,2-diol degradation pathways, which are associated with the biosynthesis of anti-tumor lipoic acid and propionate, respectively (p < 0.05) (Fig. S2A). PWY-7013 was also positively correlated to the relative abundance of Desulfovibrionaceae bacterium LT0009 (p < 0.05), a bacterium significantly enriched by TZ (Fig. S2B).

Fig. 5
figure 5

TZ supplementation modulates short-chain fatty acid levels and gut microbiota composition. (A) Fecal short-chain fatty acid (SCFA) concentrations (acetate, propionate, butyrate). (B) Alpha diversity metrics. Observed species richness (left) and Shannon diversity index (right). (C) Principal coordinates analysis (PCoA) based on a Bray–Curtis dissimilarity matrix. (D) Heatmap of differentially abundant bacterial taxa (left) and their correlations (right) with endpoint parameters. Differential abundance based on ANCOM-BC score (FDR < 0.1, p < 0.05, Fold change > 2 or < 1/2, LDA score (log10) > 2) were labelled with “*”. Color intensity is proportional to log2 transformed fold change (Treatment/Control), while size is proportional to the value of -log10 transformed adjusted p (FDR-corrected). Data are presented as mean ± SD (n = 6–10 per group). No “*” indicates p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Full size image

TZ induced gut microbiota and mycobiota modulation promoted SCFA concentrations and anti-cancer metabolic functions

Fungi have been shown to influence tumour development in colon cancer through modulating the immune system and producing fungi-derived factors35. The result of the alpha diversity showed that there is an increase in observed species in TZ + 5-Fu group compared to AOM/DSS group, indicating an increase in microbial richness. However, no significant changes in the Shannon index were observed across groups (Fig. 6B). The PCoA combined with PERMANOVA (R2 = 0.1598, p = 0.0021) showed significant differences in overall fungal composition among the treatment groups (Fig. 6C).

In terms of fungal species, 24 species showed at least a 2-fold decrease in abundance while 11 species were increased. Notably, TZ administration significantly reduced Paramyrothecium foliicola, Hyaloscypha hepaticicola, Grosmannia clavigera, Hypoxylon sp. FL1150, [Candida] arabinofermentans, Actinomortierella wolfii, Monilinia vaccinii-corymbosi, Dentiscutata erythropus, Candolleomyces eurysporus, Roridomyces roridus, Batrachochytrium salamandrivorans, Geranomyces variabilis, Dimargaris verticillate, Penicillium egyptiacum, Batrachochytrium dendrobatidis, Piromyces sp. E2, Neofusicoccum parvum, Rozella allomycis, Morchella snyderi, Claviceps sorghi, Geosiphon pyriformis, Stachybotrys elegans, Coelomomyces lativittatus, and Mortierella sp. GBA39 (p < 0.05). Besides, TZ enriched Lachancea thermotolerans, Actinomortierella ambigua, Gamsiella multidivaricata, Podila humilis, Glomus cerebriforme, Xylariaceae sp. FL0594, Xylariomycetidae sp. FL0641, Pecoramyces sp. F1, Lactifluus volemus, Chytridium Lagenaria, and Aplosporella prunicola (p < 0.05) (Fig. 6D).

Spearman’s correlation with clinical markers, such as tumor count and cytokines, suggested that TZ-depleted Batrachochytrium salamandrivorans, Batrachochytrium dendrobatidis, Neofusicoccum parvum, Claviceps sorghi, and Mortierella sp. GBA39 are positively correlated to total tumor counts (p < 0.05). Coelomomyces lativittatus, and Mortierella sp. GBA39 were positively correlated to the colon shortening (p < 0.05).

Notably, Batrachochytrium dendrobatidis were positively correlated to all three detected pro-inflammatory cytokines IL-6, IL-17 A/F, and TNF-α (p < 0.05). Noticeably, Batrachochytrium salamandrivorans was correlated to total tumor count, and all three pro-inflammatory cytokines being measured (p < 0.05), while inversely correlated to colon length (p < 0.05). On the contrary, TZ enriched Xylariaceae sp. FL0594 were inversely correlated with total tumor count, and IL-6 level (p < 0.05) (Fig. 6D).

To clarify the interkingdom interactions and co-occurrence pattern of TZ-modulated gut microbiota and mycobiota, we constructed a co-abundance network of TZ-modulated bacterial and fungal taxa. TZ-enriched bacterial species, Desulfovibrionaceae bacterium LT0009, and Helicobacter sp. MIT 03-1616, were each negatively correlated to the tumor counts, while positively correlated (p < 0.05) with Xylariaceae sp. FL0594, a fungus inversely correlated to tumor counts. In contrast, TZ-depleted bacteria also appeared to favor the growth of fungal species that showed a positive correlation with CRC tumor count. Clostridium sp. MD294 was positively correlated to Claviceps sorghi, which positively correlated tumor counts (p < 0.05), while Roseburia sp. CAG:303 was negatively correlated to the beneficial Xylariaceae sp. FL0594 (p < 0.01). Such findings highlight the role of bacterial-fungal interactions in mitigating tumor burden (Fig. 6A).

Fig. 6
figure 6

TZ modulated gut mycobiota composition, induced bacterial–fungal Community Shifts. (A) Co-abundance network of microbial species in the TZ-treated group relative to the AOM/DSS group. Red nodes indicate taxa enriched by TZ, whereas blue nodes represent taxa depleted under TZ treatment. Edges correspond to significant Spearman correlations (p < 0.05): red edges denote positive correlations, and green edges signify negative correlations. (B) Alpha diversity metrics. Observed species richness (left) and Shannon diversity index (right). (C) Principal coordinates analysis (PCoA) based on a Bray–Curtis dissimilarity matrix. (D) Heatmap of differentially abundant bacterial taxa (left) and their correlations (right) with endpoint parameters. Differential abundance based on ANCOM-BC score (FDR < 0.1, p < 0.05, Fold change > 2 or < 1/2, LDA score (log10) > 2) were labelled with “*”. Color intensity is proportional to log2 transformed fold change (Treatment/Control), while size is proportional to the value of -log10 transformed adjusted p (FDR-corrected). Data are presented as mean ± SD (n = 6–10 per group). Comparisons were performed using one-way ANOVA followed by Tukey’s multiple comparisons test. No “*” indicates p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Full size image

Discussion

Our study demonstrated that TB, in combination with ZEA, effectively attenuates colitis-associated colon cancer by significantly reducing tumour burden and improving survival outcomes. Although the combination of TZ and 5-FU resulted in a significant reduction in the total tumour count, TZ alone emerged as the most effective treatment across CRC-related parameters. Notably, all animals from both TZ and TZ + 5-FU groups survived throughout the experimental period without exhibiting severe pain or distress necessitating premature euthanasia, suggesting that TZ may also improve the survival outcome of 5-FU-treated mice. The AOM/DSS model typically induces an enlarged caecum and shortening of colon length, serving as an indication of inflammation and carcinogenicity36,37. Both TZ and TZ + 5-FU treatments alleviated caecal enlargement, however, only TZ prevented colon shortening. In contrast, 5-FU alone failed to improve either measure, but 5-FU reduced liver weight compared to the healthy group, consistent with its known hepatotoxicity effects, including inflammation, collagen fibrosis and apoptotic cell death, which collectively reduce liver functions and cause a decline in liver weight38,39,40. By contrast, TZ not only did not cause liver damage, but it also suppressed the liver damage from 5-FU when combined with 5-FU.

Chronic inflammation is a critical element in the progression of CRC, especially in the progression of AOM/DSS model41. Hyperplastic or serrated colonic epithelium and thicken collagen fibers were reported to be chronic inflammatory responses that stimulate tumor growth and promote angiogenesis in CRC42,43,44. Both TZ and TZ + 5-FU relieved the increased colonic crypt depth and hyperplasia, yet only TZ reduced the inflammation score and collagen fibrosis. Furthermore, TZ and TZ + 5-FU suppressed the concentration of pro-inflammatory cytokines, IL-6, TNF-α and IL-17 A/F, whereas 5-FU alone only suppressed TNF-α. Such findings are consistent with a previous study, in which 5-FU can be pro-inflammatory, thereby undermining its anti-tumor efficiency in AOM/DSS mouse study and humans45,46. TZ not only ameliorated the inflammatory conditions in AOM/DSS mouse models but also mitigated the inflammatory response associated with 5-FU when administered in combination with it.

Oncogenic pathways are pivotal in cancer development, driving tumor initiation and progression via genetic mutations that lead to uncontrolled cell proliferation47,48. In this study, TZ, TZ + 5-FU and 5 -FU all increased apoptosis, as evidenced by the elevation of the c-casp3 marker. The PI3K/Akt signaling pathway plays a critical role in CRC growth, proliferation, and apoptosis49,50,51. TZ supplementation significantly suppressed the phosphorylation of PI3K and Akt, as well as downstream proliferative markers, cyclin D1 and Ki67. Inhibiting the PI3K and Akt kinases, along with the cyclin D1, plays a role in limiting CRC progression52,53,54,55,56,57. By contrast, TZ + 5-FU only reduced the phosphorylation of Akt and cyclin D1 protein expression, whereas 5-FU alone did not affect the PI3K/Akt pathway or other proliferative markers in this study. Since the PI3K/Akt pathway contributes to 5-FU resistance through metabolic reprogramming, enhancing drug efflux and interacting with the tumor microenvironment to enhance cancer cell survival58,59, targeting the pathway could restore 5-FU sensitivity60, but our data indicate no such effect, with TZ only providing better efficacy.

Metagenomics analysis revealed that TZ modulated both gut microbiota and mycobiota. Gut microbiota, Desulfovibrionaceae bacterium LT0009 and Helicobacter sp. MIT 03-1616 were enriched by TZ, which correlates with reducing tumor number and lower pro-inflammatory IL-6. Desulfovibrionaceae bacterium LT0009, also known as Taurinivorans muris, respires taurine and produces hydrogen sulfide (H2S), potentially prevent pathogen colonization61. H2S has a biphasic dose-response curve in cancer cells, inducing cell death at both low and high doses62,63. Further investigations could investigate if increasing endogenous H2S production through supplementing Desulfovibrionaceae bacterium LT0009, serve as a potential therapeutic strategy for CRC. Whereas there was not much information about Helicobacter sp. MIT03-1616, except that the bacteria was previously identified from rodents64. In the current study, Helicobacter sp. MIT03-1616 were found to be positively correlated to short chain fatty acids. Consistent with these findings, fecal acetate and propionate were elevated in the TZ group. Previous studies demonstrated that elevated levels of acetate and propionate, whether administered individually or in combination, inhibit the proliferation of CRC cells65,66,67,68. Acetate was reported to exert anti-CRC properties, which was previously demonstrated to reduce CRC tumor size in a xenograft mouse model69. Similar to butyrate, propionate also functions as a histone deacetylase (HDAC) inhibitor, which may influence PI3K/AKT signaling pathways70,71. The downstream effects of HDAC inhibition and modulation of PI3K/AKT signalling are critical for inducing anti-tumour responses and may help overcome PI3K/AKT inhibitors72,73. In addition, propionate has been shown to inhibit CRC by promoting the proteasomal degradation of the euchromatic histone-lysine N-methyltransferase 2 (EHMT2) and inhibiting PRMT1-associated mTOR signaling68,74. It also enhances the surface expression of NKG2D ligands on CRC cells, thereby promoting improved immune recognition and surveillance75. In addition, Candidatus Borkfalkia ceftriaxoniphila was also positively correlated with propionate and butyrate levels. Candidatus Borkfalkia ceftriaxoniphilla, also known as Borkfalkia ceftriaxensis, is a newly identified, low-abundance commensal species implicated in restoring the gut microbiota community following antibiotic treatment76.

As to the fungal species, TZ enriched Xylariaceae sp. FL0594 which correlated to reduced tumor counts. Noteworthily, Xylariaceae sp. FL0594, a recently sequenced genome under the Xylariaceae family, was reported to produce secondary metabolites with antimicrobial, antioxidant, and anti-cancer effects77,78. These compounds including a group of sesquiterpenoids and cytochalasans, exert their effects by inducing cytotoxicity and inhibiting cell proliferation79,80. Furthermore, TZ depleted multiple species which were correlated to elevated tumor counts and pro-inflammatory cytokines, including Batrachochytrium salamandrivorans, Batrachochytrium dendrobatidis, Claviceps sorghi, and Mortierella sp. GBA39. Whereas these species were successfully identified, data regarding these fungi remain dispersed, particularly concerning their impact on CRC81. Claviceps sorghi produces ergot alkaloids which can cause poisoning effects in both human and livestock82,83,84,85,86. It was primarily understood that Batrachochytrium salamandrivorans, Batrachochytrium dendrobatidis, Neofusicoccum parvum and Coelomomyces lativittatus exhibit pathogenic characteristics in non-mammalian species but their interaction with mice and humans are not well-documented87,88,89,90,91. Despite data on bacterial-fungal interactions in cancer context remain limited, our result shows that bacterial Desulfovibrionaceae bacterium LT0009, Helicobacter sp. MIT 03-1616 and fungal Xylariaceae sp. FL0594 were correlated. Their negative association with tumor counts and pro-inflammatory cytokines IL-6 levels suggest their potential as the biomarkers of TZ anti-CRC effects.

Apart from identifying the microbiota and mycobiota changes that TZ modulated, MetaCyc analysis was used to predict the functional impact of these community shifts. The TZ-enriched pathways PWY-7388 and PWY-7013 are both associated with CRC suppression. PWY-7388 corresponds to the mitochondrial octanoyl-[acyl-carrier protein] biosynthesis in yeast, which produces octanoyl-acyl-carrier protein (acyl-ACP), a precursor for biosynthesis of lipoic acid (LA) in Saccharomyces cerevisiae92,93. LA has been shown to selectively induce cell death selectively to CRC cells, while sparing normal colonocytes94. Moreover, LA mitigated inflammatory responses in ulcerative colitis by inhibiting NF-κB signaling, reducing levels of TNF-α and IL-6, and demonstrating antioxidant scavenging properties95,96. Additionally, PWY-7013 corresponds to the (S)-propane-1,2-diol degradation pathway which generates propionate97, a metabolite known to suppress colorectal cancer progression as discussed above.

Nonetheless, this study has several limitations. First, the in vivo CRC model used may not fully represent all forms of human colorectal cancer. Although the AOM/DSS model is widely used to mimic human CRC phenotypes, the findings derived from it may not be directly translatable to human diseases. Therefore, further validation using human CRC cell lines and clinical biopsies is warranted. Moreover, the experimental setup did not include separate TB-only and ZEA-only groups. Including these groups would help determine whether the observed effects of the TZ combination are synergistic or simply additive. In addition, the causal relationship between microbial changes and the tumour suppressive effects of TZ was not established in this study. The bacterial and fungal species identified as being correlated to TZ’s anti-CRC effects have only recently been sequenced and classified. As research on these species is still in its early stages, future studies could involve culturing and administering them individually in the AOM/DSS model to assess their direct contribution to CRC suppression. Our results also indicate that TZ + 5-FU did not produce additive or synergistic effects. TZ and TZ alone appeared to outperform the combination in several parameters. This may be due to the gut dysbiosis induced by 5-Fu, which could impair the microbiota modulatory efficiency of TZ and counteract its beneficial effects. Future research is warranted to explore this potential interaction. Lastly, while ZEA is often referred to as a mycotoxin, the dose used in this study (10 mg/kg) is close to the no-observed-effect level (NOEL) of 9.2 mg/kg body weight in mice72. However, the current study did not assess whether TB mitigates the potential adverse effects of ZEA, nor did it assess the toxicity profiles of the combined formulation. Comprehensive assessments of acute, sub-chronic, and chronic safety are essential before considering therapeutic application. Importantly, future translational studies must establish the long-term safety of TZ, particularly given the known estrogenic effects of ZEA.

Conclusion

The study demonstrated that the combination of theabrownin and zearalenone (TZ) significantly attenuated AOM/DSS-induced colorectal tumorigenesis. TZ demonstrated greater efficacy in alleviating CRC tumorigenesis than 5-FU, while the combination of TZ and 5-Fu did not yield additional or synergic benefits. TZ treatment improved survival rate, tumor count, colon length, colonic crypt depth, collagen fibrosis and colonic inflammation in AOM/DSS mice through the suppression of proliferative PI3K/AKT pathways and lowering pro-inflammatory cytokines. Moreover, TZ improved the gut microbiota and mycobiota, enhanced anti-CRC metabolic profile and increased fecal acetate and propionate concentration. The enrichment of the bacterial Desulfovibrionaceae bacterium LT0009, Helicobacter sp. MIT 03-1616 and fungal Xylariaceae sp. FL0594 were identified as potential microbial biomarkers for the suppression of CRC induced by TZ. Future translational studies should be performed to ensure the safety of TZ’s application in human, especially with the ZEA component. Overall, TZ displayed promising potential to be developed as a novel CRC therapeutic.