Parabacteroides johnsonii inhibits the onset and progression of colorectal cancer by modulating the gut microbiota

Abstract

Background

Colorectal cancer (CRC) is the third most prevalent malignant tumor and the second leading cause of cancer-related deaths globally. The genus Parabacteroides is an important component of the gut microbiota. P. distasonis and P. goldsteinii are reported probiotics, and their roles in CRC have been investigated in related studies. However, the association between P. johnsonii and CRC remains unknown.

Methods

P. johnsonii (10–42) and Lactococcus formosensis (22–2) were isolated from healthy human feces. 29 mice that demonstrated normal feeding and activity were randomly assigned to four groups: normal control (NC group), CRC model (IC group), P. johnsonii (PJ group), and L. formosensis (LO group). Colonic tumor tissues from the IC, PJ, and LO groups and normal colon tissues from the NC group were then collected for HE staining and immunohistochemical staining. Fecal samples from mice during the hyperproliferative and adenoma phases were collected for Metagenomic sequencing and metabolite analysis.

Results

P. johnsonii intervention reduced the number and slowed the growth of colonic tumors, improved tumor histological scores, and decreased microenvironmental inflammation levels. P. johnsonii improved the composition of intestinal flora in mice with colon cancer, increased gut microbial species diversity, and maintained gut microbiota stability. Furthermore, P. johnsonii intervention increased the abundance of Bifidobacterium pseudolongum and Lactobacillus, which play a role in ameliorating AOM/DSS-induced gut microbiota dysbiosis. P. johnsonii intervention affected the metabolic pathways, including amino sugar degradation and galactose metabolism, sphingolipid synthesis, amino acid synthesis, and polyphenol synthesis pathways, with the tryptophan metabolism pathway as the primary pathway being affected.

Conclusion

Our study profiled the P. johnsonii administration reduces the number of tumors and lower tumor staging in AOM/DSS-induced colon cancer mice by modulating gut microbiota and its metabolites at early stages.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-025-06675-0.

Introduction

Colorectal cancer (CRC) is the third most prevalent malignant tumor and the second leading cause of cancer-related deaths globally. Studies have revealed a worrying trend, such as the annually increasing proportion of CRC cases diagnosed in individuals under 50 years old, with a tendency toward late-stage diagnosis, despite recent declines in the overall incidence and mortality rates of CRC [1]. In addition to genetic, lifestyle, and environmental factors, an increasing number of large-scale metagenomic studies have indicated that CRC development and progression are closely associated with gut microbiota dysbiosis [2]. The human colon harbors trillions of microorganisms that collectively form the gut microbiota. This microbiota is a crucial part of the human body, playing a crucial role in maintaining overall health. The gut microbiota exists in a delicate balance in a healthy individual. However, the composition and function of the gut microbiota change if this balance is disrupted, potentially causing diseases and adversely affecting human health [3]. Research has revealed that gut microbiota dysbiosis occurs when pathogenic bacteria displace harmless commensal ones. These pathogenic bacteria invade the intestinal wall, trigger inflammation, and produce carcinogenic signals and metabolites, damaging the host [1, 4]. Modulating the gut microbiota has become evident to potentially prevent and inhibit cancer development and progression as the understanding of how gut microbiota promotes cancer expands.

Probiotics are live microorganisms that confer health benefits to the host when administered in adequate amounts. Research has revealed that probiotics exert therapeutic effects, or improve the efficacy of other treatments in various diseases through multiple mechanisms [5].The genus Parabacteroides is an important component of the gut microbiota. Currently, 13 species of Parabacteroides have been isolated and identified from the feces of healthy humans, including P. distasonis, P. goldsteinii, and P. johnsonii. Members of this genus generally regulate the immune responses, suppress inflammatory responses, participate in carbohydrate metabolism, and produce short-chain fatty acids (SCFAs) [6].

Considering these characteristics, Parabacteroides appeared as a promising source of novel probiotics. Research has revealed that the abundance of Parabacteroides species is significantly reduced in the feces of patients with CRC compared to healthy individuals. P. distasonis and P. goldsteinii are reported probiotics, and their roles in CRC have been investigated in related studies [7, 8]. However, the association between P. johnsonii and CRC remains unknown. Lactic Acid Bacteria (LAB), particularly strains of Lactobacillus and Bifidobacterium, have been extensively recognized as promising probiotics due to their multifaceted health benefits, including potent antioxidative activity, broad-spectrum antimicrobial effects, enhancement of host immune responses, and chemopreventive potential against malignancies [9–11]. However, the genus Lactococcus, another critical member of LAB, remains underexplored in the context of CRC intervention. Emerging evidence suggests that lactic acid-producing cocci may exhibit antitumorigenic properties. For instance, Yu Jun et al. demonstrated that Streptococcus thermophilus attenuates colorectal tumorigenesis via β-galactosidase-mediated modulation of the tumor microenvironment [12], while Lactococcus lactis HkyuLL 10 suppresses CRC progression through alpha-mannosidase-dependent restoration of gut microbial homeostasis [13]. Despite these advances, the functional role of Lactococcus formosensis in CRC pathogenesis remains uncharted, we hypothesize that Lactococcus formosensis may similarly exert prophylactic or therapeutic effects against CRC by targeting microbiota-host interactions. The transition from the hyperproliferative phase to the adenoma phase is a critical period in CRC development. Effective early intervention during this stage may help prevent further progression to CRC. Therefore, in this study, we developed an AOM/DSS-induced CRC mouse model to investigate the effects of P. johnsonii and Lactococcus formosensis on CRC development. Additionally, we collected fecal samples from mice during the hyperproliferative and adenoma phases for metagenomic analysis and conducted metabolomic analysis on fecal samples from the adenoma phase to understand the differences in gut microbiota at various stages of CRC formation and the changes in metabolites during the adenoma phase, aiming to determine the underlying mechanisms.

Materials and methods

Experimental animals and strains

Beijing Huafukang Biotechnology Co., Ltd supplied 7-week-old SPF-grade male C57BL/6 mice, weighing 20 ± 2 g. The study was approved by the Animal Ethics Committee of the University of Science and Technology Beijing (Approval Number: HFK-AP-20230110). P. johnsonii (10–42) and Lactococcus formosensis (22–2) were isolated from healthy human feces, identified by 16 S rRNA gene sequencing, cultured in anaerobic GAM medium at 37℃, and preserved in 10% skim milk freeze-dried powder. The viability of bacteria in the freeze-dried powder was assessed using the dilution plate counting method.

Experimental design

All C57BL/6 mice were individually housed in cages (temperature: 22 ± 2℃, relative humidity: 55 ± 5%, 12-hour light/dark cycle), with free access to standard feed and water. Bedding, feed, and water were replaced daily at 8 PM. After a one-week acclimation period, 29 mice that demonstrated normal feeding and activity were randomly assigned to four groups: normal control (NC group, n = 7), CRC model (IC group, n = 8), P. johnsonii (PJ group, n = 7), and L. formosensis (LO group, n = 7). Mice in the IC, PJ, and LO groups were intraperitoneally injected with 12.5 mg/kg of AOM on day 7 of the experiment. After 5 days, they were given drinking water containing 2.5% DSS for 6 days, followed by normal water for 15 days, with a total of 3 cycles (6 days of 2.5% DSS water + 15 days of normal water) over 63 days. The NC group received an equivalent volume of sterile saline intraperitoneally. Concurrently, PJ group mice were gavaged daily with 2 × 10⁹ CFU of P. johnsonii; LO group mice were gavaged daily with 2 × 10⁹ CFU of L. formosensis; NC and IC groups were gavaged daily with an equivalent volume of sterile saline until the end of the DSS-induced inflammation cycle (Fig. 1). All mice were then euthanized, and their colons were collected to determine the number of colonic tumors. Tumor tissues were subjected to hematoxylin and eosin (HE) staining and immunohistochemistry for histological assessment. Fresh fecal samples were obtained from the IC and PJ groups and the IC, PJ, NC, and LO groups at the end of the DSS-induced inflammation phases 2 and 3, respectively. The samples were placed in sterile cryovials, rapidly frozen in liquid nitrogen, and stored at − 80℃. After collecting all samples, metagenomic sequencing and metabolite analysis of the mouse gut microbiota were conducted.

Fig. 1.

Experimental design of AOM/DSS-treated colon cancer models in the C57BL/6 mice. ICH: at the DSS-induced inflammation hyperproliferative phase in the colorectal cancer group; PJH: at the DSS-induced inflammation hyperproliferative phase in the PJ probiotic intervention group; IC: at the DSS-induced adenoma phase in the colorectal cancer group; PJ: at the DSS-induced adenoma phase in the PJ probiotic intervention group; LO: at the DSS-induced adenoma phase in the LO probiotic intervention group

HE staining and immunohistochemistry

Mice were euthanized at the cervical dislocation site. Colonic tumor tissues from the IC, PJ, and LO groups and normal colon tissues from the NC group were then collected. The tissues were fixed in formalin, dehydrated in ethanol, embedded in paraffin, and sectioned. The sections were subjected to HE staining and immunohistochemical staining for Ki67, CD11b, and β-catenin. The histologic scoring criteria for HE staining refer to the scoring criteria used in the study by Man SM et al. [14]. The staining was conducted using the SABC-AP Kit with Anti-Rabbit IgG (IHC&ICC) following the manufacturer’s instructions. Images were captured under a microscope, and the Ki67, CD11b, and β-catenin expressions were evaluated using ImageJ software to calculate the integrated optical density. This evaluation was conducted for the colonic tumor tissues in the IC, PJ, and LO groups and the normal colon tissues in the NC group.

Fecal metagenomic sequencing

DNA was collected from samples using the E.Z.N.A. Soil DNA Kit (Omega Bio-tek, USA). After extraction, DNA concentration and purity were evaluated, and DNA integrity was verified using 1% agarose gel electrophoresis. DNA was fragmented using a Covaris M220 (Gene Company, China), and fragments of approximately 400 bp were selected for establishing paired-end libraries. Library preparation was conducted using the NEXTFLEX Rapid DNA-Seq Kit (Bioo Scientific, USA), including adapter ligation, bead selection, polymerase chain reaction amplification, and bead recovery steps. Shanghai Meiji Biomedical Technology Co., Ltd conducted the metagenomic sequencing using the Illumina NovaSeq platform (Illumina, USA).

Metagenomic data were assembled using MEGAHIT (version 1.1.2), and contigs of ≥ 300 bp were selected as the final result. Open reading frames were predicted using Prodigal, and genes of ≥ 100 bp in length were translated into amino acid sequences. Gene sequences from all samples were clustered using CD-HIT (version 4.6.1) to establish a non-redundant gene set. High-quality reads were aligned to the non-redundant gene set using SOAPaligner (version 2.21) to estimate gene abundance. Amino acid sequences of the non-redundant gene set were compared against the NR database and Kyoto Encyclopedia of Genes and Genomes (KEGG) database using Diamond (version 0.8.35) for species and functional annotations, and abundance information was calculated.

Fecal metabolomics

Solid samples of 50 mg were added to a 2-mL centrifuge tube along with 6-mm grinding beads and 400-µL extraction solvent (methanol: water = 4:1, containing 0.02 mg/mL of internal standard L-2-chlorophenylalanine). The sample was ground using a cryogenic tissue grinder (− 10℃, 50 Hz, 6 min) and then subjected to low-temperature ultrasonic extraction (5℃, 40 kHz, 30 min). The mixture was left at − 20℃ for 30 min and then centrifuged (4℃, 13,000 g, 15 min), with the supernatant used for liquid chromatography–mass spectrometry (LC-MS/MS) analysis. Equal volumes of supernatant from all samples were mixed to prepare QC samples, with a QC sample analyzed after every 5–10 test samples. The analysis was conducted using a Thermo Fisher Scientific UHPLC-Q Exactive HF-X system. Chromatographic separation was conducted on an ACQUITY UPLC HSS T3 column (100 mm × 2.1 mm i.d., 1.8 μm; Waters, Milford, USA). Mobile phase A was 95% water + 5% acetonitrile (containing 0.1% formic acid) and mobile phase B was 47.5% acetonitrile + 47.5% isopropanol + 5% water (containing 0.1% formic acid). The injection volume was 3 µL, and the column temperature was 40℃. Mass spectrometry was conducted in both positive and negative ion modes, with ion spray voltage set to 3500 V for positive mode and − 3500 V for negative mode, an ion transfer tube temperature of 325℃, and normalized collision energies of 20–40–60 V. The first-order mass spectrometry resolution was set to 60,000, and the second-order resolution was set to 7,500. Data were collected in DDA mode. The mass spectrometry data were processed using ProgenesisQI software and matched with the HMDB and Metlin databases. The final data were preprocessed, normalized, and log10-transformed on the Meiji cloud platform to gather a standardized data matrix for further analysis.

Cell viability assay and colony formation experiment

Cell viability was determined using the CCK-8 assay. HCT116, HT29, and NCM460 cells in the logarithmic growth phase were trypsinized and resuspended to prepare single-cell suspensions. Cells were seeded into 96-well plates at a density of 6.0 × 10³ cells per well and incubated for 24 h to allow adherence. The original medium was then replaced with fresh medium containing varying concentrations (2.5%, 5%, 10%, or 20%) of P. johnsonii or L. formosensis culture supernatants (PJCS, LOCS), with normal saline serving as the control. Cells were cultured for 0, 1, 2, 3, or 4 days. After incubation, 10 µL of CCK-8 solution was added to each well, followed by 2 h of incubation. Absorbance at 450 nm (OD450) was measured using a microplate reader. For the colony formation assay, 1,000 cells from each cell line (HCT116, HT29, NCM460) were seeded into 6-well plates and treated with the same concentrations of bacterial supernatants or saline. Cells were maintained in a 37 °C, 5% CO₂ incubator, with regular medium replacement and daily observation. The experiment was terminated when macroscopic colonies (diameter > 50 μm or containing > 50 cells) became visible. Colonies were fixed with methanol, washed with PBS, stained with 0.1% crystal violet for 10 min, rinsed thoroughly, air-dried, and photographed for analysis.

Bioinformatics statistical analysis

Metagenomic analysis was conducted using the mothur software to calculate alpha diversity metrics, including Chao 1 and Shannon index. Differences in alpha diversity between groups were evaluated using the Kruskal–Wallis H test. Principal Coordinates Analysis (PCoA) based on Bray-Curtis distance was used to assess the similarity of microbial community structures between samples, combined with PERMANOVA non-parametric testing to identify the significance of the differences in microbial community structures between groups. LEfSe analysis (Linear Discriminant Analysis [LDA] ≥ 3, p < 0.05) was used to determine bacterial taxa with significant differences in abundance from phylum to genus levels between groups. Species with Spearman correlation|r| of > 0.6 and p-values of < 0.05 were selected for correlation network analysis.

The R package ropls (Version 1.6.2) was used to conduct principal component analysis and orthogonal partial least squares discriminant analysis (OPLS-DA) on the preprocessed data matrix for metabolomic data analysis. Metabolites with variable importance in projection (VIP) values of > 1 from the OPLS-DA model were selected. Statistical Package for the Social Sciences version 27.0 was utilized for parametric or non-parametric testing to determine metabolites with p-values of < 0.05 and|log2 (Fold Change)| of > 2. Differential metabolites were defined as those meeting the criteria of VIP of > 1, p-values of < 0.05, and|log2(Fold Change)| of > 2. Spearman’s rank correlation analysis was performed to assess associations between differentially abundant bacterial taxa, with statistically significant correlations defined as Benjamini-Hochberg false discovery rate (FDR)-adjusted p < 0.05 and an absolute correlation coefficient (|r|) > 0.6, and the KEGG database (https://www.kegg.jp/kegg/pathway.html) was used to annotate the metabolic pathways of differential metabolites. The Python package scipy.stats was used for pathway enrichment analysis, and Fisher’s exact test was utilized to identify the most relevant biological pathways associated with experimental treatments. Statistical significance was defined as p-values of < 0.05 (*p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). All data analyses were performed on the Meiji Bio cloud platform (https://cloud.majorbio.com).

Results

P. johnsonii significantly reduces tumor number and improves histopathological scores in mouse Colon cancer

All mice were euthanized by cervical dislocation on the final day of the experiment, and their colons were examined. We observed tumors in the colons of mice in the IC, PJ, and LO groups. However, the number of tumors in the colons of mice in the PJ intervention group was significantly reduced compared to the IC group (Student’s t-test, p < 0.01), whereas the number of tumors in the colons of the LO intervention group did not demonstrate a significant change compared to the IC group (Student’s t-test, p > 0.05). This indicates that oral P. johnsonii administration attenuated the tumor loadin mice (Fig. 2A). Additionally, HE staining and histopathological scoring of colon tumor tissues from all groups indicated that oral P. johnsonii administration significantly improved the histopathological scores of colon tumors (Student’s t-test, p < 0.05), whereas Lactococcus formosensis demonstrated no significant effect (Student’s t-test, p > 0.05) (Fig. 2B).

Fig. 2.

Colon cancer tumor statistics and HE Staining of tumor tissues (A) Representative images of colon cancer specimens from the IC, PJ, and LO groups and comparison of the number of colon tumors among these groups. (B) Representative HE-stained sections of colon tumor tissues from the NC, PJ, IC, and LO groups and comparison of histopathological scores of colon tumor tissues among the IC, PJ, and LO groups (statistical significance defined as p-values of < 0.05, * p < 0.05, ** p < 0.01, ns, no significance, by Student’s t-test)

P. johnsonii slows tumor proliferation and improves inflammation levels in the tumor microenvironment in mice

All mice were euthanized by cervical dislocation on the final day of the experiment, and their colon tissues were collected for immunohistochemical staining of Ki-67, β-catenin, and CD11b to evaluate tumor proliferation and inflammation levels in the tumor microenvironment.

The colon tumor tissues in the IC group demonstrated significantly increased positive staining for Ki-67 and β-catenin, indicating high tumor proliferation. Additionally, significant lymphocyte infiltration and severe tissue vacuolation were observed, indicating rapid tumor growth and high malignancy. Compared to the IC group, P. johnsonii intervention significantly reduced the expression of Ki-67 and β-catenin in the AOM/DSS-induced colon tumors (p < 0.05, p < 0.001), indicating slower tumor proliferation. The LO group demonstrated an increase in Ki-67 and β-catenin expression compared to the IC group, but the changes were not significant (p > 0.05) (Fig. 3A and B).

Fig. 3.

Comparison of Ki67, β-catenin, and CD11b immunohistochemical staining in colon tumor tissues of the IC, PJ, and LO groups and normal colon tissues of the NC group. (A) Ki67 Immunohistochemical Staining: Representative Ki67 immunohistochemical staining images and tissue scores of colon tumors from the IC, PJ, and LO groups and normal colon tissues from the NC group. (B) β-catenin Immunohistochemical Staining: Representative β-catenin immunohistochemical staining images and tissue scores of colon tumors from the IC, PJ, and LO groups and normal colon tissues from the NC group. (C) CD11b Immunohistochemical Staining: Representative CD11b immunohistochemical staining images and tissue scores of colon tumors from the IC, PJ, and LO groups and normal colon tissues from the NC group. (Statistical significance is indicated as follows: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001)

CD11b is a marker for tumor microenvironment inflammation. Our study revealed that CD11b expression was significantly higher in the colon tumors of the IC group compared to the normal colon tissue of the NC group (p < 0.001), indicating severe inflammation. P. johnsonii intervention significantly reduced CD11b expression in the tumor tissues compared to the IC group (p < 0.01), demonstrating a reduction in inflammation. Additionally, the LO group demonstrated an increase in CD11b expression compared to the IC group, but with no significance (p > 0.05), indicating no significant improvement in tumor microenvironment inflammation (Fig. 3C).

P. johnsonii improves the composition of intestinal flora in mice with colon cancer

Metagenomic analysis revealed significant differences in the α-diversity of gut microbiota among different groups, with the Ace index demonstrating significant variation (p < 0.05). The IC group exhibited a significant reduction in gut microbiota richness compared to the NC group (Nemenyi test, p < 0.05), indicating that the tumor model markedly reduced microbiota diversity (Fig. 4A). Although the Shannon index did not show significant differences (p = 0.149), a trend toward increased microbial species diversity was observed in the P. johnsonii-treated mice and approached the levels seen in the control group, indicating that P. johnsonii may be associated with changes in gut microbiota diversity that could influence the tumor microenvironment (Fig. 4B).

Fig. 4.

Analysis of α and β Diversity of Gut Microbiota in Mice at Six Sampling Points: NC, LO, IC, ICH, PJ, and PJH. (A) Comparison of Ace Index: The Ace index of gut microbiota at different sampling points (NC, LO, IC, ICH, PJ, and PJH). Significant differences in microbiota richness are indicated (*p < 0.05). (B) Comparison of Shannon Index: The Shannon index of gut microbiota at different sampling points. No significant differences are observed in this index, but overall microbial diversity is assessed. (C) PCoA and PERMANOVA Analysis at Six Sampling Points: PCoA analysis and PERMANOVA results demonstrate differences in the microbial composition among NC, LO, IC, ICH, PJ, and PJH groups. (D) PCoA and PERMANOVA Analysis at Three Time Points: PCoA and PERMANOVA analysis comparing the NC, IC, and ICH groups over three time points to assess changes in microbial composition

PCoA and PERMANOVA analyses at the genus level of the gut microbiota revealed significant differences in microbial composition between groups (Adonis, R² = 0.308, p < 0.05; Adonis, R² = 0.377, p < 0.05). The microbial composition was distinctly different from that in other groups in colon cancer model mice, particularly in comparison to the IC group. This indicates that significant differences in gut microbiota composition were observed between the overproliferation and adenoma stages (Fig. 4C, D).

In addition, the species-level intestinal flora principal components of mice at the PJH and PJ groups were significantly different from those at the ICH and IC groups (Adonis, R² = 0.264, p < 0.001; Adonis, R² = 0.197, p < 0.001), suggesting that oral administration of P. johnsonii altered the composition of the intestinal flora of mice in both the hyperproliferative and adenomatous phases (Figures S1A, B). There was no significant difference in the main composition of the intestinal flora of mice at the PJ and PJH groups compared with mice in the NC group (Adonis, R2 = 0.19, p = 0.059), suggesting that oral administration of P. johnsonii improved the composition of the flora in colon cancer mice, bringing them close to the intestinal flora composition of normal mice, and helping to maintain the stability of the flora (Figure S1C). However, the main components of the intestinal flora of mice at the LO groups were not significantly different from those at the IC group (Adonis, R² = 0.088, p = 0.249), suggesting that administration of Lactococcus formosensis did not have a significant effect on the composition of the intestinal flora of colon cancer mice (Figure S1D).

Analysis of the microbiota associated with the intervention of P. johnsonii in colon cancer

Figure 5A shows the dominance of the phyla Bacteroidetes and Firmicutes at the phylum level. Among these, Verrucomicrobia was more abundant in the IC group (22.4%) than in the ICH group (5.26%), suggesting a possible association with colon tumor development. The abundance of Actinobacteria is 9.75% in the PJ group, suggesting a potential role in colon cancer modulation, though further research is needed to confirm this. Figure 5B illustrates that Akkermansia muciniphila is a major contributor in the ICH and IC groups at the species level, accounting for 21.6% and 5%, respectively, and belongs to the phylum Verrucomicrobia. Bifidobacterium pseudolongum is the main contributor in the PJ group, accounting for 7.4%, and belongs to the phylum Actinobacteria. Figure 5C presents the LEfSe analysis, indicating that the genera Bifidobacterium and Lactobacillus have higher LDA scores compared to the CRC model group in the PJ group. This indicates that these probiotics play a more significant role in the PJ group and may have a positive effect on gut health. In contrast, the genus Erysipelatoclostridium has a higher LDA score in the IC group, consistent with previous studies, indicating its association with colon cancer development and progression. Under Bacteroides uniformis intervention, Bifidobacterium pseudolongum and Lactobacillus may play crucial roles in improving colon cancer outcomes.

Fig. 5.

Composition and differential analysis of gut microbiota in mice at six sampling points: NC, LO, IC, ICH, PJ, and PJH.(A) Classification and abundance of gut microbiota at the phylum level.(B) Classification and abundance of species B levels of gut microbiota.(C) LEfSe analysis (LDA score of ≥ 3)

Interaction between characteristic microbiota in the overproliferation and adenoma stages of the PJ and IC groups

We conducted a correlation analysis of Erysipelatoclostridium, Bifidobacterium pseudolongum, and Lactobacillus, as well as the associated microbiota, to better understand the interactions among microbiota. Erysipelatoclostridium in the IC group, demonstrated significant negative correlations with various harmful microbiota (such as Treponema, Ruminococcus, Prevotella, and Faecalibacterium), indicating a crucial role and complex interactions of these microbiota in the tumor environment. Bifidobacterium pseudolongum and Lactobacillus in the PJ group were strongly positively correlated with other beneficial bacteria (such as Olsenella and Turicibacter), indicating oral administration of P. johnsonii may improve the flora environment in colon cancer by promoting an increase in beneficial intestinal bacteria.Additionally, the presence of some opportunistic pathogens, such as Escherichia and Pseudomonas, was observed, further reflecting the complex interactions among the microbiota (see Fig. 6).

Fig. 6.

The mutual network diagram of bacteria associated with Bifidobacterium, Lactobacillus, and Erysipelatoclostridium. A circle represents a species, and the lines between the circles indicate a significant correlation between these two species. The red and green lines denote positive and negative correlations, respectively. The thicker the lines, the greater the absolute value of the correlation coefficient. Bacteria with P-values of < 0.05 and| r| of > 0.6 were selected for display

The metabolic pathways involved in the P. johnsonii intervention include the degradation of glycosaminoglycans, galactose metabolism, sphingolipid synthesis, amino acid synthesis, and polyphenol synthesis pathways

We performed iPath metabolic pathway analysis on the macrogenome of mouse intestinal flora and determined the predominant and distinctive metabolic pathways annotated in each group of mice (Fig. 7A). In the PJ group, key metabolic pathways identified included glycosaminoglycan degradation, galactose metabolism, sphingolipid synthesis, amino acid synthesis, and polyphenol synthesis (Fig. 7B). Further research into these metabolic pathways may help elucidate the potential mechanisms through which P. johnsonii is associated with changes in colon cancer development, potentially leading to new therapeutic targets.

Fig. 7.

Analysis of related metabolic Pathways in mouse gut microbiota across groups. (A) iPath Analysis of Metabolic Pathways for mouse gut microbiota at six sampling points: NC, LO, IC, ICH, PJ, and PJH. (B) Metabolic Pathways Affected by Parabacteroides johnsonii Intervention

Differential Metabolites between the P. johnsonii group and the IC group during the adenoma stage mainly include quinolines, amino acids, and fatty acids

We collected fecal samples from the healthy, IC, and PJ groups for metabolomics analysis at the end of the DSS-induced inflammation phase III (adenoma stage) to understand the differences in intestinal metabolites during the adenoma stage. The OPLS-DA analysis of metabolites revealed significant differences between the IC and PJ groups during the adenoma stage, with good intra-group reproducibility (Fig. 8A). Differential metabolites between the two groups were selected using p-values of < 0.05 and VIP of > 1 and displayed in a volcano plot. Compared to the PJ group, 53 metabolites were upregulated, and 45 were downregulated in the IC group (Fig. 8B). P. johnsonii oral administration significantly improved ovalicin, 4-amino-5-hydroxymethyl-2-methylpyrimidine, 3-keto sphingosine, 22-hydroxydocosaenoic acid, sterculic acid, doxercalciferol, and S-β-cholanic acid. It diminished xanthurenic acid compared to the AOM/DSS group (Fig. 8C and J). Additionally, K-W rank sum tests on metabolites from the NC, PJ, and IC groups during the adenoma stage revealed uptrend of PG, PC, and Lyso-PC in the IC group, whereas PC was decreased after P. johnsonii intervention. Moreover, 4-(2-Amino-3-hydroxyphenyl)-2, 4-dioxobutanoic acid, a butyrate derivative, was reduced in the IC group compared to the NC group (p < 0.05) but increased following P. johnsonii intervention (p < 0.05) (Fig. 8K).

Fig. 8.

Metabolomics Analysis of IC and PJ Groups (P < 0.05 indicates statistical significance; * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). (A) OPLS-DA plot of metabolites during the adenoma stage for IC and PJ groups. (B) Volcano plot of differential metabolites between the IC and PJ groups during the adenoma stage (larger scatter points indicate higher VIP values. Significantly upregulated, downregulated, and non-significant metabolites are indicated in red, blue, and gray, respectively). (C-J) Differences in metabolites between the IC and PJ groups during the adenoma stage. (K) Comparative analysis of metabolites during the adenoma stage among the NC, PJ, and IC groups

The tryptophan metabolism pathway is closely associated with P. johnsonii intervention

The tryptophan metabolism pathway was significantly enriched in both metagenomic and metabolomic analyses (Fig. 9A and B). Therefore, we integrated the results from these two omics approaches and annotated the relevant enzymes and metabolites on the KEGG tryptophan metabolism pathway. The results indicate that multiple branches of the tryptophan metabolism pathway in the PJ group, particularly the kynurenine and indole pathways, showed significant metabolic alterations (Fig. 9C). Key enzymes including K01667 (tnaA, p < 0.05, fold change 0.7), K00128 (ALDH, p < 0.05, fold change 1.5), K01501 (E3.5.5.1, p < 0.05, fold change 2.3), and K04103 (ipdC, p > 0.05, fold change 4.5) exhibited significant or non-significant alterations across groups, while metabolites like xanthurenic acid (p < 0.05, fold change 0.8) were significantly downregulated and indoxyl (p < 0.05, fold change 1.1) was significantly upregulated in the PJ group. These changes suggest that P. johnsonii is associated with alterations in key enzyme and metabolite levels, potentially influencing the activation of the tryptophan metabolism pathway, particularly the kynurenine and indole pathways. This activation may influence CRC-related gut microbiota changes, suggesting a potential role in modulating disease-associated metabolic pathways.

Fig. 9.

Integrated metagenomic and metabolomic analysis of the tryptophan metabolism pathway. (A) KEGG enrichment analysis of the tryptophan metabolism pathway in the adenoma stage of the IC and PJ groups (P < 0.05 indicates statistical significance, * p < 0.05). (B) Enrichment analysis of differential metabolites in the adenoma stage between the IC and PJ groups. The size of the points represents the number of differential metabolites, and the color indicates significance. (C) Integrated analysis of enzymes and metabolites associated with the tryptophan metabolism pathway. Squares represent enzymes, and circles represent metabolites. Light green and green indicate downregulation in the PJ group, whereas light red and dark red denote upregulation in the PJ group

P. johnsonii culture supernatant inhibits CRC cell viability and colony formation

To evaluate the tumor-suppressive effects of P. johnsonii and L. formosensis in vitro, functional analyses were conducted using two CRC cell lines (HCT116, HT29) and a normal colonic epithelial cell line (NCM460) as a control. Results demonstrated that P. johnsonii culture supernatant significantly reduced the viability of CRC cell lines in a concentration-dependent manner. Notably, 10% and 20% P. johnsonii supernatants exhibited pronounced inhibitory effects from day 3 to day 4 of treatment (p < 0.0001), while no significant impact was observed on the viability of normal colonic epithelial cells (p > 0.05). In contrast, L. formosensis culture supernatant showed no significant alteration in viability across all tested cell lines (p > 0.05) (Fig. 10A–C). Consistent with these findings, colony formation assays revealed a marked reduction in colony numbers for HCT116 and HT29 cells treated with P. johnsonii supernatant (p < 0.0001, p < 0.001) (Fig. 10D). These data collectively suggest that molecules secreted by P. johnsonii can suppress both viability and clonogenic capacity in CRC cells.

Fig. 10.

Effects of P. johnsonii, L. formosensis culture supernatants (2.5%, 5%, 10%, 20%), and saline control on viability and colony formation in HCT116, HT29, and NCM460 cell lines (p-values calculated using two-way ANOVA for viability assays and one-way ANOVA for colony formation; * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). (A) Viability of HCT116 cells treated with varying concentrations of P. johnsonii or L. formosensis supernatants versus saline control. (B) Viability of HT29 cells treated with varying concentrations of P. johnsonii or L. formosensis supernatants versus saline control. (C) Viability of NCM460 normal colonic epithelial cells treated with varying concentrations of P. johnsonii or L. formosensis supernatants versus saline control. (D) Colony formation capacity of HCT116, HT29, and NCM460 cells treated with P. johnsonii or L. formosensis supernatants at indicated concentrations versus saline control

Discussion

The gut microbiota significantly affects the overall health of the host, with dysbiosis having a direct or indirect association with the occurrence of various diseases, including cancer [1, 15]. Recent studies suggest that gut microbiota alterations contribute to cancer development, including CRC, though the precise extent varies across studies. Patients with CRC exhibit decreased gut microbiome biodiversity, with a significant increase in pathogenic bacteria and a notable decrease in beneficial symbiotic bacteria, compared to healthy individuals [16]. Regulating the gut microbiota to prevent or treat CRC has become a focus of research in recent years, considering the complex and close association between the gut microbiota and CRC.

The genus Parabacteroides is a core member of the human gut microbiota, with a detection rate of > 90% in gut microbiome databases. Moreover, it is detected across various stages, indicating its close association with human health at different phases [17]. Parabacteroides help digest high-fiber diets that we cannot otherwise process and their abundance increases with resistant starch diets. The membrane component of P. distasonis is responsible for suppressing pro-inflammatory cytokine production in CRC cell lines [18]. Other studies have revealed that P. distasonis possesses anti-inflammatory and anti-tumor properties, mediated through TLR4, MYD88, and Akt signaling reduction, and by promoting apoptosis [19]. These results are consistent with the observed reduction in P. distasonis levels in CRC mouse models. Parabacteroides has been identified and isolated into over 20 species, with 13 being common gut bacteria. However, most studies on Parabacteroides have focused on the following four species: P. distasonis (model species), P. goldsteinii, P. johnsonii, and P. merdae. Currently, both P. distasonis and P. goldsteinii have been associated with CRC development [20, 21]. However, P. johnsonii has not yet been reported. In this study, we developed an AOM/DSS-treated cancer model in C57BL/6 mice to investigate the effects of P. johnsonii on improving CRC in mice. The results indicate that oral P. johnsonii administration regulates the gut microbiota, alters related metabolites, and reduces tumor burden in AOM/DSS-treated C57BL/6 mice.

In this study, AOM/DSS-induced cancer model mice showed decreased intestinal flora abundance during the adenoma stage, whereas mice administered P. johnsonii showed enhanced intestinal flora diversity and flora composition that was significantly different from that of the cancer model group of mice and closer to that of the intestinal flora of normal mice, suggesting that oral P. johnsonii may affect tumorigenesis by modulating intestinal flora diversity and altering the composition of intestinal flora to influence tumorigenesis. This is the first report of P. johnsonii bacteria intervening in the gut microbiota of CRC mice and exerting a positive effect.

We analyzed the composition of gut microbial populations at both phylum and species levels to investigate their potential association with CRC. Significant changes in microbial composition occurred at the DSS-induced inflammation overgrowth period (ICH). The gut microbiota composition of mice receiving P. johnsonii and Lactococcus formosensis interventions was similar to that of the control group (NC group), indicating that P. johnsonii and Lactococcus formosensis mitigate the degree of microbial changes associated with CRC. Notably, Verrucomicrobia accounted for 22.4% and 5.26% in the IC and ICH groups, respectively, indicating that these bacteria may play a crucial role in colorectal tumor development. Nielson T Baxter has revealed that Verrucomicrobia increases the susceptibility to CRC and is closely associated with colorectal tumor occurrence [22].

Our results at the species level indicate that Akkermansia muciniphila was the main contributor in the ICH and IC groups, accounting for 21.6% and 5%, respectively, and belongs to the Verrucomicrobia phylum. Akkermansia muciniphila is considered a next-generation probiotic for certain disease prevention and treatment [23]. Some studies have revealed that the abundance of A. muciniphila is higher or increased in patients with CRC compared to healthy individuals. However, the reduction in A. muciniphila abundance is associated with severe symptoms of CRC, indicating that A. muciniphila does not play a role in CRC development [24]. Additionally, A. muciniphila administration improved the gene expression of proliferation-related molecules or cell proliferation markers such as S100A9, Dbf4, and Snrpd1 [25]. Other studies indicate that A. muciniphila may promote intestinal inflammation and tumorigenesis. Overall, the role of A. muciniphila in CRC development or inhibition is unclear and remains controversial [26].

The abundance of Bifidobacterium pseudolongum, Lactobacillus, and Actinobacteria at the phylum level was higher in the PJ group, indicating that these probiotics play a more important role in the PJ group and may have positive effects on gut health [27]. In contrast, the LDA values of Erysipelatoclostridium and Parasutterella in the IC group were higher, consistent with previous studies, indicating that these two bacteria are closely associated with CRC development and progression [28]. In 2012, Tjalsma et al. proposed the bacterial driver–passenger model of CRC, stating that pathogenic bacteria (“driver bacteria”) colonize the intestinal mucosa, driving CRC through persistent inflammation, increased cell proliferation, or genotoxic substance production. Subsequently, the tumor microenvironment changes, and the “driver bacteria” are gradually replaced by bacteria (“passenger bacteria”) that have a competitive advantage in the new environment, promoting further CRC development [29]. Although Akkermansia, Lachnospiraceae, Muribaculaceae, and Bacteroidales are often considered beneficial, their increased presence during the hyperproliferation phase and adenoma phase suggests a potential role as passenger bacteria in CRC progression. We studied the interactions among the microbial communities to further clarify the effect of the PJ group on the gut microbiota. Our results revealed that Erysipelatoclostridium was significantly negatively correlated with harmful bacteria associated with CRC in the IC group, such as Treponema, Ruminococcus, Prevotella, and Faecalibacterium, indicating its crucial role in the tumor environment [30]. In contrast, beneficial bacteria in the PJ group, such as Bifidobacterium pseudolongum and Lactobacillus, significantly increased and demonstrated strong positive correlations with other beneficial bacteria, including Olsenella and Pseudomonas, indicating their potential role in improving the gut environment. In summary, the IC group was dominated by harmful bacteria, whereas the PJ group was dominated by beneficial bacteria, indicating the positive effect of P. johnsonii intervention on the gut microbiota.

We performed iPath metabolic pathway analyses on the mouse intestinal flora macrogenome, in which the metabolic pathways annotated to the mouse intestinal flora macrogenome administered with P. johnsonii including amino polysaccharide degradation, galactose metabolism, sphingolipid synthesis, amino acid synthesis, and polyphenol synthesis. The galactose metabolism pathway may be involved in cancer cell metabolism and can inhibit breast cancer cell growth [31]. Abnormal lipid and amino acid metabolism may be important factors in cancer development. Changes in the composition, distribution, and content of lipids in cell membranes and cells, as well as lipid metabolism abnormalities, have increasingly been recognized as potential mechanisms in various tumors. Growing evidence indicates that sphingolipid and amino acid synthesis pathways are closely associated with tumor development [32–34]. Sphingomyelinase regulates ceramide generation to inhibit colon cancer cell proliferation. The hydrolysis of sphingomyelin to ceramide activates endoplasmic reticulum stress pathways (e.g., PERK/eIF2α/CHOP), inducing apoptosis. Sphingosine kinase (SphK1) inhibitors block sphingosine-1-phosphate (S1P) production, suppressing metastasis. SphK1 overexpression correlates with poor prognosis in colorectal cancer patients, as S1P promotes epithelial-mesenchymal transition (EMT) and angiogenesis through S1PR1/3 receptors. Sphingolipid metabolic reprogramming enhances antitumor immunity by modulating CD8⁺ T cell function, as elevated S1P levels in the tumor microenvironment impair CD8⁺ T cell effector activity [35–38]. Polyphenols exert multifaceted anticancer effects in colorectal carcinogenesis through antioxidant, anti-inflammatory, microbiota-modulating, antiproliferative, and pro-apoptotic mechanisms. They reduce oxidative DNA damage by chelating metal ions and activating the Nrf2 pathway, while suppressing inflammation-driven oncogenic signaling (e.g., NF-κB, STAT3). Polyphenols promote probiotic growth, inhibit procarcinogenic bacteria, and stimulate beneficial metabolite production (e.g., short-chain fatty acids, tryptophan derivatives). Their antiproliferative effects involve regulation of Wnt/β-catenin and PI3K/Akt pathways to arrest the cell cycle at G1/S phase, coupled with caspase cascade activation to induce apoptosis. Specific polyphenols (e.g., myricetin, luteolin) downregulate VEGF and MMPs, inhibiting tumor angiogenesis and EMT [39–43]. Further research into these metabolic pathways may reveal the molecular mechanisms by which oral P. johnsonii administration mitigates CRC development and identifies new therapeutic targets.

The fecal metabolomics analysis of mice in the IC and PJ groups during the adenoma stage revealed that the content of quinoline metabolites in the intestinal metabolites of mice in the IC group was significantly altered. Specifically, 1,5-Isoquinolinediol and Isoquinoline were elevated in the IC group, whereas 2,8-quinolinediol and 4,6-dihydroxy-2-quinolinecarboxylic acid (xanthurenic acid) were decreased. These metabolite alterations suggest a potential link between quinoline metabolism and CRC, though further studies are needed to confirm its role [44]. Studies have revealed that 8-hydroxyquinoline (HQ) and its derivatives, when complexed with metal ions (such as copper and iron), inhibit cell proliferation and tumor growth in vivo [45]. Additionally, xanthurenic acid, equivalent to 8-hydroxykynurenic acid and possessing the structure of 8-hydroxyquinoline, exhibits anticancer properties; thus, a reduction in xanthurenic acid contributes to CRC progression. Pathway enrichment analysis of the differential metabolites in the KEGG database indicated that the primary enriched pathways include cutin, suberine and wax biosynthesis, and thiamine metabolism. Research has revealed that Lactiplantibacillus plantarum-12 alleviates CRC symptoms by modulating the cutin, suberine, and wax biosynthesis pathways [46]. This indicates that P. johnsonii may mitigate CRC through its effects on cutin, suberine, and wax biosynthesis, and thiamine metabolism.

Further integrating the macro-genomic and metabolomic results, we found that the tryptophan metabolic pathway was significantly enriched in KEGG pathway enrichment analyses in both histologies, especially showing significant changes in the kynurenine(Kyn) and indole pathways. In vitro experiments demonstrated that P. johnsonii culture supernatant suppresses CRC cell viability and clonogenic capacity, confirming the tumor-growth inhibitory role of its metabolites.

The tryptophan metabolism pathway is considered a universal therapeutic target for various diseases, including tumors, cancer, neurodegenerative diseases, infectious diseases, autoimmune diseases, and metabolic diseases [47, 48]. Studies have revealed that microbiome-derived tryptophan catabolites mediate the chemopreventive effects of statins on CRC [49]. The Kyn pathway is the main pathway for tryptophan degradation [50], and molecules produced by this pathway are critical in immune response, inflammation, and oxidative stress [51]. Kynurenine reduces the activity of natural killer cells, dendritic cells, or proliferating T cells, while kynurenic acid promotes monocyte extravasation and controls cytokine release. High kynurenine levels also increase the proliferative and migratory capacity of cancer cells [52].The rate-limiting enzyme of the Kyn pathway, indoleamine 2,3-dioxygenase (IDO), has enhanced expression in tumors, and downstream metabolites (e.g., Kyn) activate β-catenin signaling and promote proliferation of mouse colon cancer cells [53]. The aryl hydrocarbon receptor (AhR) is a sensor of tryptophan metabolites and an effective immunomodulator. It has been shown that a variety of tryptophan-derived metabolites activate the AhR and transmit immunosuppressive signals in the tumor microenvironment [54], and that tumor cells can take advantage of this mechanism of tryptophan degradation in the host body to evade immune control and thus grow undisturbed [55]. A recent article on immunity revealed that AhR on tumor-associated macrophages is activated by indole, a tryptophan metabolite produced by the gut microbiota, which inhibits inflammatory T-cell infiltration and promotes tumor growth [56]. Small molecule inhibitors targeting IDO and tryptophan 2,3-dioxygenase (TDO) are undergoing preclinical and clinical evaluation for oncology treatment due to their ability to alleviate immunosuppression [57, 58]. In addition Parabacteroides johnsonii was found to convert tryptophan into indolepropionic acid (IPA) and indoleacetic acid (IAA).IPA inhibits tumor-associated inflammation by enhancing intestinal barrier function through the activation of AhR and decreasing the release of pro-inflammatory factors (e.g., IL-1β, TNF-α).IAA directly induces apoptosis of colon cancer cells and inhibits their IAA can directly induce apoptosis and inhibit proliferation of colon cancer cells [59]. Tryptophan metabolites (e.g., IPA) inhibit tumor-associated macrophage (TAMs) differentiation to the immunosuppressive M2 phenotype through the AhR pathway, thereby suppressing tumor immune escape [48]. These studies suggest that the tryptophan pathway has a bidirectional regulatory role for tumor development. Significant changes in some key enzymes (e.g., Tryptophanase A, Aldehyde Dehydrogenase) and metabolites (e.g., Xanthurenic Acid, Indoxyl) on the tryptophan pathway were found in our study, suggesting that P. johnsonii is associated with alterations in key enzyme and metabolite levels, potentially influencing the activation of the tryptophan metabolism pathway. This activation may influence CRC-related gut microbiota changes, suggesting a potential role in modulating disease-associated metabolic pathways. However, to confirm whether microbiota-driven metabolic changes are functionally relevant to CRC suppression and to identify whether tryptophan metabolites directly affect CRC progression, it is necessary to supplement indole derivatives or deprive xanthurenic acid in CRC cells, and assess whether these metabolites mimic the effects of P. johnsonii in reducing tumor growth.

Our study analyzed the changes in gut microbiota from the hyperproliferative phase to the adenoma phase of colon tumors and the alterations in metabolites during the adenoma phase, as well as the role of P. johnsonii and Lactococcus formosensis in these two stages. Our results reveal differences in gut microbiota composition between hyperproliferative and adenoma phases in colon cancer mice, confirming the early gut microbiota changes associated with colon cancer development. The discovery of new bacteria may help explain novel mechanisms of colon cancer occurrence and provide new potential drug targets.We have confirmed, for the first time, that oral P. johnsonii administration reduces the number of tumors and lower tumor staging in AOM/DSS-induced colon cancer mice by modulating gut microbiota and its metabolites, indicating that P. johnsonii suppresses colon cancer occurrence and development at early stages. Our experimental results provide new insights into colon cancer development and offer useful guidance for targeted early intervention in patients with colon cancer.

This study has some limitations. First, our study did not establish a control group of normal mice administered only P. johnsonii to assess whether there were potential side effects of P. johnsonii administration and the baseline effects of P. johnsonii administration on the gut microbiota. In the absence of such control, we cannot exclude that the observed benefits may be related to P. johnsonii and host interactions, rather than acting solely through modulation of the microbiota and metabolites. Second, the focus of this study on the hyperproliferative and adenomatous stages of tumors to determine a preventive effect, while not including the heteroproliferative and ultimately adenocarcinomatous stages, limits the understanding of how changes in the dynamics of the intestinal flora are associated with tumor progression throughout its entire course, and in particular does not allow for an assessment of whether the administration of P. johnsonii is still capable of exerting an inhibitory effect in the context of late-stage tumor growth or whether it may be rendered ineffective by a change in the tumor micro environment. Therefore, the current conclusions are mainly applicable to early intervention, and its long-term efficacy needs to be further validated by extending the observation period and incorporating more disease-stage studies. This study is performed in the AOM/DSS-induced CRC mouse model, which does not fully recapitulate human CRC pathophysiology. This model primarily induces inflammation-driven tumorigenesis rather than the genetic mutations observed in human colorectal cancer. Additionally, several studies have explored gut microbiota’s role in inhibiting colon cancer, but most are limited to animal models or cell experiments, which is a limitation of our study. The effectiveness, safety, and specific usage of probiotics in patients with colon cancer need further investigation and clarification, and our future research should focus on the clinical translation of probiotics.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Liu J and Zhang Y contributed equally to this study. Dong ZW and Gu GL designed the research; Liu J and Zhang Y conceived of the study, collected data and written manuscript; Xu LX sorted out the data. All authors have read and agreed to the published version of the manuscript.

Funding

Supported by Clinical Program of Air Force Medical University, (School Science [2022] No.7); Excellent Talent Program of Air Force Medical Center, (22BJQN004).

Data availability

The data presented in this study are available on request from the author.

Declarations

Ethics approval and consent to participate

Not applicable.

Institutional review board statement

The animal study was reviewed and approved by the Animal Ethics Committee of the University of Science and Technology Beijing (Approval Number: HFK-AP-20230110).

Consent for publication

Not applicable.

Conflict of interest

The authors declare that they have no conflicts of interest.