Berberine might block colorectal carcinogenesis by inhibiting the regulation of B-cell function by Veillonella parvula

Abstract

Background:

Colorectal carcinogenesis and progression are related to the gut microbiota and the tumor immune microenvironment. Our previous clinical trial demonstrated that berberine (BBR) hydrochloride might reduce the recurrence and canceration of colorectal adenoma (CRA). The present study aimed to further explore the mechanism of BBR in preventing colorectal cancer (CRC).

Methods:

We performed metagenomics sequencing on fecal specimens obtained from the BBR intervention trial, and the differential bacteria before and after medication were validated using quantitative polymerase chain reaction. We further performed Apc^Min/+ animal intervention tests, RNA sequencing, flow cytometry, immunohistochemistry, and enzyme-linked immunosorbent assays.

Results:

The abundance of fecal Veillonella parvula (V. parvula) decreased significantly after BBR administration (P = 0.0016) and increased through the development from CRA to CRC. Patients with CRC with a higher V. parvula abundance had worse tumor staging and a higher lymph node metastasis rate. The intestinal immune pathway of Immunoglobulin A production was activated, and the expression of TNFSF13B (Tumor necrosis factor superfamily 13b, encoding B lymphocyte stimulator [BLyS]), the representative gene of this pathway, and the genes encoding its receptors (interleukin-10 and transforming growth factor beta) were significantly upregulated. Animal experiments revealed that V. parvula promoted colorectal carcinogenesis and increased BLyS levels, while BBR reversed this effect.

Conclusion:

BBR might inhibit V. parvula and further weaken the immunomodulatory effect of B cells induced by V. parvula, thereby blocking the development of colorectal tumors.

Trial Registraion:

ClinicalTrials.gov, No. NCT02226185.

Introduction

Colorectal cancer (CRC) is one of the most common malignancies in the world. Currently, CRC accounts for up to 10% of all cancer cases, ranking fifth among new tumor cases, and fifth in tumor-related deaths wordwide.^[1,2] CRC is a multifactorial and multistep disease, the etiology of which has not yet been fully elucidated. Its risk factors are mainly genetic and environmental factors. Approximately 95% of CRC cases develop from colorectal adenoma (CRA). Colorectal tumors have become a serious threat to human health; hence, it is of great value to explore their pathogenesis in depth.

Humans have long coevolved in symbiosis with microorganisms that colonize the gastrointestinal tract and other organs of the body. The gut microbiota forms a special system with a complex composition and structure in the human body, forging deep interrelationships with host mucosal epithelial cells and immune cells, and having a vital role in the human immune system.^[3] Previous studies have revealed significant differences in the species and their abundance of the intestinal microbiota among normal subjects, and patients with CRA and CRC.^[4,5] Regardless of whether it is adenoma or adenocarcinoma, there is still plenty of evidence that disturbance of the intestinal flora is closely related to the evolution of colorectal tumors.^[6,7]

Berberine (BBR) is extracted from Coptidis Rhizoma, a traditional Chinese herb that has been used in China for more than 3000 years. A multicenter, double-blind, randomized placebo-controlled trial (RCT) demonstrated that oral administration of BBR significantly reduced the recurrence of CRA after endoscopic treatment, providing direct evidence and a potential drug for the prevention of CRA–CRC progression.^[8] However, the mechanism of action of BBR is still unclear. Clinical trials have found that BBR regulates metabolism by altering the human gut microbiota.^[9,10] Our previous animal experiments confirmed that BBR inhibited colorectal tumorigenesis in mice by modulating the intestinal microecological structure.^[11] Thus, the action of BBR most likely involves in the regulation of intestinal microecology and the maintenance of intestinal homeostasis. However, how BBR prevents CRC development by altering the gut microbiota, its critical effector species, and potential mechanism remains unknown.

Based on the previous clinical intervention trial and an animal study, we further applied clinical specimen analysis, in vivo and in vitro intervention experiments to determine the mechanism by which BBR prevents CRC by altering the gut microbiota.

Methods

Ethical approval

Approval for the study was obtained from the ethics committee of Renji Hospital, Shanghai JiaoTong University School of Medicine (No. RA-2019-349). All study participants provided written informed consent, and the study was conducted following the guidelines of the Declaration of Helsinki.

CRC patient sample collection

CRC patients were recruited from Renji Hospital, Shanghai Jiaotong University School of Medicine. All subjects were Han Chinese ethnic origin. Tumor tissue specimens (immersed in RNA later solution), formalin-fixed paraffin tissue samples, and preoperative stool samples were collected from the subjects and preserved at -80°C.

Preliminary clinical intervention trial and sample collection

Participants aged 18–75 years who had one to six CRAs completely resected under endoscopy within 6 months were randomly assigned (1:1) to receive BBR or placebo. Trial drugs were taken orally at 0.3 g twice per day for more than one year. Subjects were reexamined annually by colonoscopy. The primary outcome was recurrence of adenoma at follow-up colonoscopy. The Trial was available at ClinicalTrials.gov (No. NCT02226185). The detailed trial design and results were described in a previously published study.^[8] We collected stool and serum samples from both groups of subjects before taking the drug within one week after colonoscopy in the first and/or second year, respectively.

Stool sample processing and metagenomic sequencing

All stool samples were preserved at -80°C immediately after collection. Stool specimens were subjected to microbial genomic DNA extraction using a QIAamp PowerFecal DNA Kit (Qiagen, Hilden, Germany) according to the manufacturer's guidance. DNA concentrations were measured using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), and samples with concentrations less than 10 ng/μL or an absorbance value (A_{260 nm}/A_{280 nm}) below 1.8 or above 2.2 were rejected. The extracted DNA samples were stored at -80°C.

Agarose gel electrophoresis was applied to examine the integrity of the DNA samples, and 1 μg of DNA was taken and disrupted using ultrasound. We conducted polymerase chain reaction (PCR) and cyclization to construct the sequencing library, followed by high-throughput sequencing using the Illumina HiSeq sequencing platform (Illumina Inc., San Diego, CA, USA) in dual-end sequencing mode on multiple samples. According to the distribution characteristics of the low mass fraction of the Illumina sequencing data, quality control analysis of the pretreatment data was carried out using FastQC software (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/), Trimmomatic (https://www.usadellab.org/cms/?page=trimmomatic) was applied to remove low-quality reads, and the information statistics were complied using Fastq-stat function from the ea-utils package (https://expressionanalysis.github.io/ea-utils/).

Validation by real-time quantitative PCR (qPCR)

Ten microliters of SYBR Green II (Takara, Japan) were applied for the qPCR system, and 80 ng of fecal DNA sample was used as the amplification template. qPCR amplification was performed using a StepOnePlus^TM (ABI, Foster City, CA, USA) instrument. The amplification program consisted of 40 cycles in total, comprising denaturation at 95°C for 30 s, followed by annealing at 95°C for 5 s, and extension at 60°C for 30 s. DNA samples were assessed in three replicates and average cycle threshold (CT) values were calculated. The relative abundance of the target bacterial strain was defined as the ΔCT value obtained by subtracting the CT value of the target bacterial strain from the CT value of the total bacterium (16S ribosomal ribonucleic acid [rRNA]). The primer sequences were as follows: Veillonella parvula (V. parvula) Forward GGTGAATACG TTCCCGG; V. parvula Reverse TACGGCTACCTTG TTACGACTT; 16S rRNA Forward GGTGAATACGT TCCCGG; 16S rRNA Reverse TACGGCTACCTTGT TACGACTT. All primers were produced by Sangon Biotech (Shanghai, China).

RNA extraction and RNA sequencing

An RNAsimple Total RNA kit (Tiangen, Beijing, China) was used to extract RNA from the collected surgical tissue samples from our single-center patients with CRC. Then, a NanoDrop 2000 Spectrophotometer (Thermo Scientific, USA) was used to measure the RNA concentration. Absorbance (A_{260 nm/280 nm}) values larger than 2.0 or lower than 1.8 were removed. The extracted RNA was further treated with deoxyribonuclease to avoid DNA contamination and monitored for quality on an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA). Samples were sequenced on the Illumina HiSeq 4000 platform (Illumina Inc., San Diego, CA, USA) for 2 × 150 bp double-end sequencing. Sequencing reads were aligned to the human genome GRCh38 using HISAT2 (http://ccb.jhu.edu/software/hisat2/index.shtml).^[12] FeatureCounts was used to quantify the transcriptome using genome annotation from GENCODE v22.^[13] Differential expression analysis of count files was performed using the DESeq2 package, following standard normalization procedures.^[14]

Immunohistochemistry

For immunohistochemistry, paraffin-embedded CRC tumor tissue was cut into 4-μm sections and subjected to acid hyperbaric induction for antigen repair, and then peroxidase blocking was performed to remove endogenous hydrogen peroxide. After blocking with 10% sheep serum (diluted in phosphate-buffered saline [PBS]), sections on slides were incubated with rabbit anti-CD19 monoclonal antibody (A19013, Abclonal, Wuhan, China) overnight at 4°C in a humidified chamber. The slides were subsequently incubated with a horseradish peroxidase (HRP)-labeled secondary antibodies for 30 min at room temperature and then stained with a substrate solution containing 3,3΄-diaminobenzidine (DAB).

Animal experiment

Apc^Min/+ mice are common models for spontaneous CRC. To explore the effect of V. parvula on CRC, 40 male Apc^Min/+ mice aged 5–6 weeks were divided into two groups: (1) PBS treatment and (2) V. parvula treatment. Mice were given a mixture of antibiotics (0.2 g/L ampicillin, neomycin, and metronidazole and 0.1 g/L vancomycin) supplemented in their drinking water two weeks before gavage. V. parvula (purchased from the American Type Culture Collection [ATCC], Manassas, VA, USA) was cultured in 1252 ATCC medium–Clostridium perfringens medium (CM149; Oxoid, Basingstoke, UK) with 1.5% sodium lactate (60% solution). After 1 day, the bacteria were collected, resuspended in PBS, and then administered to mice via gavage (1 × 10⁸ colony forming unit [CFU] in 100 μL of PBS per mouse) once daily for 12 weeks. The control mice received the same regimen except that the 100 μL of PBS contained no V. parvula. Mice were sacrificed after 14 weeks, and colon tissue, tumor tissue, spleen, fecal and serum samples were collected. To investigate whether the effect of V. parvula on CRC was dependent on encoding B lymphocyte Stimulator (BLyS), 40 male Apc^Min/+ C57B/6 mice aged 5–6 weeks old were divided into four groups: (1) PBS + isotype control, (2) V. parvula + isotype control, (3) PBS + anti-BLyS antibodies, and (4) V. parvula +anti-BLyS antibodies. For mice treated with anti-BLyS or isotype, 5 mg/kg B-cell Activating Factor (BAFF)-neutralizing antibody (clone 10F4; GSK, Brentford, UK) or isotype was administered intraperitoneally once a week for 12 weeks. To investigate whether BBR could block the promotion effect of V. parvula on CRC, 5- to 6-week-old male Apc^Min/+ C57B/6 mice were divided into four groups: (1) PBS; (2) V. parvula; (3) BBR; and (4) V. parvula + BBR. For mice treated with BBR by gavage, 100 mg/kg of a BBR–PBS suspension was gavaged once per day for 12 weeks and the interval between BBR and V. parvula gavage in mice was at least four hours.^[15] All colonic tissues, tumor tissues, feces, serum and spleen samples were collected on day 80.

Flow cytometry

The complete mouse colon was dissected and divided into four segments. To isolate lamina propria immune cells, the colon segments were shaken in 0.5 mol/L ethylenediaminetetraacetic acid (EDTA)/1 mmol/L dithiothreitol (DTT) to strip the intestinal epithelial cells and the intraepithelial lymphocyte layer. The remaining tissue was digested with deoxyribonuclease (0.5 mg/mL) and collagenase (1 mg/mL) to obtain a single cell suspension. For flow cytometry, cells were stained with the following combinations of fluorescently coupled monoclonal antibodies: anti-CD3 (clone: 145-2C11), anti-CD4 (clone: RM4-5), anti-CD45 (clone: 30-F11), anti-CD19 (clone: 1D3), and anti-CD21/CD35 (clone: 7E9). Samples were collected on BD LSRFortessa™ X-20 flow cytometer (BD Biosciences, San Jose, CA, USA) and analyzed using FlowJo software (BD Biosciences).

Enzyme-linked immunosorbent assay (ELISA)

To detect BLyS levels in serum, we used the Mouse BAFF/BLyS/TNFSF13B Quantikine ELISA Kit (MBLYS0; R&D SYSTEMS, Minneapolis, MN, USA). We performed the experiment according to the manufacturer's instruction manual. Briefly, the serum was diluted 50–100 times and coincubated at room temperature with anti-BLyS antibodies for two hours, followed by incubation with the reaction substrate for 30 minutes and detection of the optical density at 450 nm.

Bacteria culture and constructing a growth curve

V. parvula (ATCC^® 17745™) was purchased from the ATCC and was cultured in an anaerobic incubator at 37°C in Veillonella medium. Absorbance (A_{600 nm}) was measured every 4 h. We used the serial dilution plate counting method to determine the relationship between the concentration of the bacteria solution and the A_{600 nm} value. Different concentrations of BBR (20 μmol/L or 50 μmol/L) were added into the bacterial culture medium, and the A_{600 nm} was measured every 4 h to determine its effect on bacterial growth.

Statistical analysis

All statistical analyses were performed by SPSS (version 25.0, IBM, USA) or SAS (version 9.4, SAS Institute Inc, USA) or the program R (www.r-project.org) or GraphPad Prism (version 9, Dotmatics, Boston, MA, USA). Statistical values were expressed as the mean ± standard error of mean [SEM] in the in vivo and in vitro experiments. Two-side Student's t-test was used to compare two groups. The Analysis of Variance (ANOVA) was used to compare the results among different multiple groups and then further analyzed by Tukey's multiple comparison test as a post-hoc analysis. Statistical significance was shown by a P value <0.05.

Results

Decrease of V. parvula in feces after administration of BBR

Gut microbiota diversity was compared between the groups before and after the administration of BBR, and no significant difference was obtained for the Shannon–Wienier index and Simpson's index between the two groups (P >0.05). Unlike the changes in the gut microbiota caused by antibiotics, no obvious effect on the overall α-diversity was observed after taking BBR. Therefore, we concluded that for patients taking BBR for a long time, their intestinal microflora might remain in a self-stable state relative to the individual, and their overall intestinal microecological environment would be stable and diverse [Figure 1A].

Figure 1.

Decrease of V. parvula in feces after administration of BBR. (A) There was no significant difference in Simpson's index showing the overall alpha diversity among the four groups. (B) PcoA of unweighted UniFrac analysis was performed to assess the alterations of fecal bacterial composition structure before and after the use of BBR. (C) Heat map showing bacteria with significant changes in abundance before and after the use of BBR. (D) The fecal bacterial abundance differed between the adenoma recurrence group and non-recurrence group. A cutoff value of ≥2.0 was used for the LDA as shown. (E) qPCR detection of the abundance of V. parvula in the feces samples of participants before and after the use of BBR (n = 128). (F) qPCR detection of the abundance of V. parvula in the feces samples of participants before and after the use of placebo (n = 144).The data were displayed as the mean ± SEM. The statistical analysis methods were Kruskal–Wallis H test (A) and a paired t-test (E). BBR: Berberine; CT: Cycle threshold; LDA: Linear discriminant analysis; ns: No significance; PCoA: Principal coordinate analysis; qPCR: Quantitative polymerase chain reaction; V. parvula: Veillonella parvula; SEM: Standard error of measurement.

The abundance of each gene in each sample and the microbial species richness were calculated based on the read comparison in the metagenomics sequencing results. The relative abundance of the different bacteria in each sample was calculated with the aim of screening out bacteria whose relative abundance changed significantly before and after drug administration. The difference in β-diversity at the operational taxonomic unit (OTU) level was evaluated using principal coordinate analysis (PCoA). Consistent with the α-diversity differences, the β-diversity for the stool microbiome was not differentially separated before and after BBR intervention [Figure 1B]. The heat map showed that the bacteria with the most significant changes (P <0.05) in relative abundance before and after taking BBR included V. parvula, Akkermansia muciniphila, Clostridium cellulovorans, and Eubacterium limosum at the species level and Anaerococcus, Clostridium, Solitalea, Pedobacter, and Roseburia at the genus level. In the placebo group, there were no such changes before and after administration [Figure 1C]. At the same time, we discovered that the relative abundance of V. parvula was also significantly increased in the recurrence group [Figure 1D]. Thus, we focused on V. parvula for further study.

qPCR validation showed that the amount of V. parvula was significantly decreased after taking BBR in CRA patients (P = 0.0016) [Figure 1E], and no obvious difference in the abundance of fecal V. parvula was detected before and after treatment in the placebo group [Figure 1F].

V. parvula abundance was related to the development of CRC

qPCR confirmed that V. parvula abundance was higher in the relapsed population than in the non-relapsed population (P = 0.0334) [Figure 2A].

Figure 2.

V. parvula abundance is related to the development of CRC. (A) qPCR detection of the abundance of V. parvula in the feces of recurrence (n = 55) and non-recurrence (n = 73) participants after BBR treatment. (B) qPCR detection of the abundance of V. parvula in the feces of patients with CRC (n = 52) and HC (n = 51). (C) qPCR detection of the abundance of V. parvula in the feces of patients with CRC at different stages, stage I (n = 6), stage II (n = 18), stage III (n = 23), and stage IV (n = 5). (D) qPCR detection of the abundance of V. parvula in the feces of patients with CRC with metastasis (n = 28) and without metastasis (n = 24). The statistical analysis methods were unpaired t-test (A, B, D) and one-way ANOVA followed by Turkey's test (C). ANOVA: Analysis of variance; BBR: Berberine; CRC: Colorectal cancer; CT: Cycle threshold; qPCR: Quantitative polymerase chain reaction; HC: Healthy control; V. parvula: Veillonella parvula; SEM: Standard error of measurement.

To further explore the correlation between V. parvula and the progression of colorectal tumors, we collected stool samples from patients with CRC in our center. We found that the abundance of V. parvula was higher in patients with CRC than in the normal population (P = 0.0010) [Figure 2B]. Moreover, CRC patients with a high abundance of V. parvula had a worse tumor stage [Figure 2C] and a higher lymph node metastasis rate [Figure 2D], supporting the cancer-inducing effect of V. parvula.

In addition, we performed multivariate linear regression analysis to show the correlation between V. parvula abundance and CRC disease phenotype indicators, including patients' age, sex, tumor size, pathological differentiation, penetration, and American Joint Committee on Cancer (AJCC) stage. The results showed that V. parvula abundance correlated positively with AJCC stage in CRC (β = 2.74, P = 0.0003) [Table 1].

Table 1.

The correlation between the abundance of Veillonella parvula and CRC disease phenotype indicators.

Variables	S	β	P value
Sex	1.13	-1.07	0.3523
Age	0.07	-0.06	0.4135
Maximum diameter	0.32	-0.11	0.7438
Pathological differentiation	1.32	0.55	0.6826
Penetration	0.71	-1.24	0.0867
AJCC staging	0.69	2.74	0.0003

AJCC: American Joint Commission on Cancer; CRC: Colorectal cancer.

V. parvula promoted CRC progression

After antibiotic cleansing, Apc^Min/+ mice in the group gavaged with V. parvula grew more (P = 0.0089) and larger tumors (P = 0.0481) compared with those in the control group [Figure 3A–D], indicating that V. parvula could promote CRC progession. The colonization abundance increased siginificantly with V. parvula gavage [Figure 3E].

Figure 3.

V. parvula promotes CRC progression. (A) Schematic diagram showing the experimental design and timeline of the Apc^Min/+ mice model. (B) Typical colon images in each group of Apc^Min/+ mice (tumors are marked with red arrows). (C) The number of colon tumors in each group of Apc^Min/+ mice. (D) Colon tumor load in each group of Apc^Min/+ mice, PBS (n = 15), and V. parvula (n = 15). (E) qPCR detection of the abundance of V. parvula in the feces of each group of Apc^Min/+ mice, PBS (n = 15), and V. parvula (n = 15). The data are displayed as the mean ± SEM. The statistical analysis methods were unpaired t-test (C) and Mann–Whitney test (D,E). ABX: Antibiotic cocktail; CFU: Colony forming units; CRC: Colorectal cancer; CT: Cycle threshold; PBS: Phosphate-buffered saline; qPCR: Quantitative polymerase chain reaction; SEM: Standard error of measurement; V. parvula: Veillonella parvula.

V. parvula was associated with the Immunoglobulin A (IgA)-producing intestinal immune pathway

RNA sequencing was performed on surgical tissue samples, and the results showed that 29 related pathways were significantly regulated in the V. parvula high-abundance group. Neuroactive ligand–receptor interaction is involved in nerve conduction, and autoimmune thyroid disease may be associated with other autoimmune diseases. The gut microbiota maintains normal immune tolerance in the intestine, and intestinal immune pathway plays a critical role in the development of colorectal tumors. Therefore, we selected the intestinal immune network for IgA production for follow-up analysis. The expression of TNFSF13B (encoding BLyS) and the genes encoding its receptors (interleukin-10 and TGF-β), which are representatives of the pathway, were significantly upregulated in CRC tissues from the high V. parvula group [Figure 4A–C]. Immunohistochemistry of clinical samples showed a higher proportion of B cells in the CRC tissue of patients with high abundance of V. parvula (P = 0.0075) [Figure 4D]. Moreover, correlation scatter graphs showed that V. parvula correlated positively with the B-cell positive rate and BLyS expression in CRC tissues [Figure 4E,F]. In the mouse model, flow cytometry of immune cells in the intestinal lamina propria was conducted, which showed that the intestinal B-cell infiltration level increased significantly with V. parvula gavage (P = 0.0085) [Figure 4G]. Moreover, the serum BLyS level in Apc^Min/+ mice also increased after V. parvula intervention (P <0.0001) [Figure 4H].

Figure 4.

V. parvula is associated with the IgA-producing intestinal immune pathway. (A) A bubble plot showing up-regulated signaling pathways analyzed by GSEA in the tumor tissue of patients with CRC in the V. parvula-high group compared with those in V. parvula-low group. (B) The GSEA enrichment plot shows that the intestinal immune network for IgA production was significantly enriched in the tumor tissue of patients with CRC in the V. parvula-high group. (C) Heat map showing the expression levels of each gene contained in the intestinal immune network for IgA production. (D) Representative images of immunohistochemical staining of B cells (marked by CD19) in CRC tissues with high and low abundance of V. parvula (left, original magnification at 200 ×), the positive rate of B cells in CRC tissues with high (n = 19) and low (n = 19) V. parvula abundance (right). (E) The correlation between the abundance of V. parvula and the positive rate of B cells in CRC tissues (n = 40). (F) The correlation between the abundance of V. parvula and the relative mRNA expression of BLyS in CRC tissues (n = 25). (G) Flow cytometry analysis of the proportion of B cells of CD45⁺ cells in the lamina propria of Apc^Min/+ colon tissues, PBS (n = 10), and V. parvula (n = 11). (H) ELISA analysis of the BLyS concentration in serum of Apc^Min/+ mice, PBS (n = 15). The statistical analysis methods were Mann–Whitney test (D right), Spearman correlation test (E,F), and unpaired t-test (G,H). BLyS: B lymphocyte stimulator; CRC: Colorectal cancer; CT: Cycle threshold; ELISA: Enzyme-linked immunosorbent assay; GS: Gene sets; GSEA: Gene set enrichment analysis; IgA: Immunoglobulin A; KEGG: Kyoto encyclopedia of genes and genomes; MSigDB: Molecular signatures database; PBS: Phosphate-buffered saline; SEM: Standard error of measurement; V. p/V. parvula: Veillonella parvula.

B lymphocytes might be involved in the pro-carcinogenic effect of V. parvula

After using an anti-BLyS antibody in the mouse model, the serum BLyS level and CD19⁺B-cell subsets in the spleen and colon were significantly decreased [Figure 5A–C], and the number and size of colon tumors induced by V. parvula were significantly reduced [Figure 5D–F]. In addition, isotype and anti-BLyS antibody had no effect on V. parvula colonization [Figure 5G], indicating that intestinal B lymphocytes were induced by V. parvula and might be involved in colorectal carcinogenesis. The pro-carcinogenic effect of V. parvula could be weakened after blocking the production of B lymphocytes using anti-BLyS antibodies.

Figure 5.

B lymphocytes might be involved in the pro-carcinogenic effect of V. parvula. (A) Schematic diagram showing the experimental design and timeline of the Apc^Min/+ mice model. (B) ELISA analysis of the BLyS concentration in serum of Apc^Min/+ mice, isotype (n = 5), and anti-BLyS (n = 5). (C) Flow cytometry analysis of the proportion of B cells of CD45⁺ cells in the spleen (up) and lamina propria (down) of Apc^Min/+ colon tissues, isotype (n = 5), and anti-BLyS (n = 5). (D) Typical colon images in each group of Apc^Min/+ mice (tumors are marked with red arrows). (E) The number of colon tumors in each group of Apc^Min/+ mice, isotype-PBS (n = 12), isotype-V. parvula (n = 13), anti-BLyS-PBS (n = 5), and anti-BLyS-V. parvula (n = 7). (F) Colon tumor load in each group of Apc^Min/+ mice, isotype-PBS (n = 12), isotype-V. parvula (n = 13), anti-BLyS-PBS (n = 5), and anti-BLyS-V. parvula (n = 7). (G) qPCR detection of the abundance of V. parvula in the feces of Apc^Min/+ mice, isotype-PBS (n = 12), isotype-V. parvula (n = 13), anti-BLyS-PBS (n = 5), and anti-BLyS-V. parvula (n = 7). The data are displayed as the mean ± SEM. The statistical analysis methods were an unpaired t-test (B,C) and a two-way ANOVA test (E–G). ABX: Antibiotic cocktail; ANOVA: Analysis of variance; BLyS: B lymphocyte stimulator; CT: Cycle threshold; CFU: Colony forming units; ELISA: Enzyme-linked immunosorbent assay; ns: No significance; PBS: Phosphate-buffered saline; qPCR: quantitative polymerase chain reaction; SEM: Standard error of measurement; V. parvula: Veillonella parvula.

BBR inhibited the pro-carcinogenic effect of V. parvula

The growth curve showed that V. parvula growth in vitro was inhibited by BBR in a dose-dependent manner [Figure 6A]. In animal experiments, the number and size of colon tumors in Apc^Min/+ mice were reduced after BBR intervention, suggesting that BBR could inhibit the tumor-promoting effect induced by V. parvula [Figure 6B–E]. In addition, BBR decreased the abundance of V. parvula in mice feces [Figure 6F], representing evidence that BBR directly inhibited V. parvula. To explore the effect of BBR intervention on the pro-carcinogenic effect of V. parvula, the serum BLyS levels and lamina propria B-cell infiltration levels were detected in four groups of mice, treated by PBS, V. parvula, BBR, and V. parvula + BBR, respectively. The results showed that V. parvula up-regulated the BLyS expression and the proportion of B cell, while BBR intervention down-regulated the V. parvula-induced serum BLyS expression and B-cell infiltration in colon tissue [Figure 6G,H]. Finally, we demonstrated the relationship between BBR intervention and BLyS expression in the clinical trial patients. The serum BLyS concentration decreased significantly after taking BBR and also differed significantly between the BBR intervention group and the placebo group [Figure 6I–K].

Figure 6.

BBR inhibited the pro-carcinogenic effect of V. parvula. (A) The growth curve of V. parvula treated with BBR (n = 3). (B) Schematic diagram showing the experimental design and timeline of the Apc^Min/+ mice model. (C) Typical colon images in each group of Apc^Min/+ mice (tumors are marked with red arrows). (D) The number of colon tumors in each group of Apc^Min/+ mice, PBS-PBS (n = 8), PBS–V. parvula (n = 12), BBR–PBS (n = 8), and BBR–V. parvula (n = 11). (E) Colon tumor load in each group of Apc^Min/+ mice, PBS–PBS (n = 8), PBS-V. parvula (n = 12), BBR-PBS (n = 8), and BBR-V. parvula (n = 11). (F) qPCR detection of the abundance of V. parvula in the feces of Apc^Min/+ mice, PBS–PBS (n = 8), PBS–V. parvula (n = 12), BBR–PBS (n = 8), and BBR–V. parvula (n = 11). (G) ELISA analysis of the BLyS concentration in serum of Apc^Min/+ mice, PBS–PBS (n = 6), PBS–V. parvula (n = 6), BBR–PBS (n = 5), and BBR–V. parvula (n = 5). (H) Flow cytometry analysis of the proportion of B cells of CD45⁺ cells in the lamina propria of Apc^Min/+ colon tissues, PBS–PBS (n = 7), PBS–V. parvula (n = 8), BBR–PBS (n = 7), and BBR–V. parvula (n = 8). (I) ELISA detection of the BLyS concentration in serum of clinical trial subjects after placebo (n = 31) or BBR (n = 30) intervention. (J) ELISA detection of the BLyS concentration in serum of clinical trial subjects before (n = 30) and after (n = 30) BBR intervention. (K) ELISA detection of the BLyS concentration in serum of clinical trial subjects before (n = 31) and after (n = 31) placebo intervention. The statistical analysis methods were two-way ANOVA (A,D–H), an unpaired t-test (I), and a paired t-test (J,K). ABX: Antibiotic cocktails; ANOVA: Analysis of Variance; BBR: Berberine; BLyS: B lymphocyte Stimulator; CFU: Colony-forming units; CT: cycle threshold; CRC: Colorectal cancer; DMSO: Dimethyl sulfoxide; ELISA: Enzyme-linked immunosorbent assay; ns: No significance; PBS: Phosphate-buffered saline; qPCR: Quantitative polymerase chain reaction; SEM: Standard error of mean.

Discussion

In the study, we identified V. parvula, a new potential "driver" bacteria of CRC. V. parvula might promote the progression of adenoma–carcinoma by activating intestinal immune pathway with increased BLyS expression and B-cell infiltration. BBR could directly inhibit V. parvula and its pro-carcinogenic effect [Figure 7].

Figure 7.

Schematic diagram of the relationships among BBR, V. parvula, and BLyS in the development of CRC. BBR: Berberine; BLyS: Encoding B lymphocyte Stimulator; CRC: Colorectal cancer; V. parvula: Veillonella parvula.

The composition and function of the gut microbiome in patients with CRC are significantly different from those in healthy subjects, and the abundances of some bacterial species, such as Fusobacterium nucleatum, Bacteroides fragilis, Streptococcus gallic acidophilus, Streptococcus pepticus, and Enterococcus faecalis, are significantly increased in CRC tissues. Increasing evidence shows that microbial dysbiosis can be observed not only after the formation of colorectal tumors, but also before the malignant transformation of cells or in the early stages of the tumor, which may directly contribute to tumorigenesis.^[7,16] Microbial dysbiosis leads to immune dysfunction of the intestinal mucosa, cell proliferation, and increased susceptibility to carcinogens.^[17]

BBR, a common isoquinoline alkaloid, was originally extracted from the rhizome of Coptis and has been used to treat bacterial diarrhea in China for thousands of years. A previous RCT identified the protective effect of BBR against colorectal tumors; however, its mechanism is not clear. A pharmacokinetic study showed that BBR needs to be converted into its active form under the action of intestinal bacteria,^[18] suggesting that the mechanism of BBR might be associated with the gut microbiota. Metagenomic sequencing of fecal samples from the intervention trial and real-time qPCR verified that the relative abundance of V. parvula decreased after one year of oral BBR. Furthermore, in vivo and in vitro, BBR was found to show a direct inhibitory effect on V. parvula.

V. parvula is a gram-negative anaerobic micrococci belonging to the genus Veillonella, which is part of the normal flora parasitizing the human oral, intestinal, and respiratory tracts.^[19] Previous studies have shown that V. parvula is associated with poor prognosis in ventilator-associated pneumonia, bone and joint or soft tissue abscesses, and cirrhosis.^[20] A recent study indicated that V. parvula harnessed nitrate respiration in the inflammatory reponse to colonize in the intestine.^[21] Hence, V. parvula could induce certain digestive diseases, for example, inflammatory bowel disease (IBD) and CRC. V. parvula is the only anaerobic coccus capable of fermenting lactate to produce acetate and propionate, which is associated with Fusobacterium nucleatum colonization.^[22] In the current study, the relative abundance of V. parvula increased with the CRA–adenocarcinoma sequence. Patients with CRC with high fecal V. parvula abundance had a worse tumor stage, higher rates of lymph node metastasis, and poorer prognosis. Combined with the above results, we believe that V. parvula might be the key effector of BBR in preventing CRA recurrence or even CRC.

As one of the important antigens in the intestine, the gut microbiota maintains normal immune tolerance in the intestine,^[23] stimulates the maturation of intestinal B cells, participates in constructing intestinal immune defense, and maintains intestinal homeostasis.^[24] In the present study, mRNA sequencing of surgical CRC tissue revealed that V. parvula might exert protumor effects through the intestinal immune network for IgA production. The representative gene of this pathway, TNFSF13B, the encoding BLyS and the genes encoding its receptors, is mainly expressed by B cells, and BLyS levels are associated with the function of B cells. Some intestinal B-cell subtypes are linked to poor prognosis in patients with malignancies.^[25] A recent study demonstrated that colon tumors enriched with B cells, particularly certain B-cell subtypes, showed an immunosuppressive property, and were characterized by antigen presentation, microsatellite instability-high (MSI-H), and an increased tumor mutation burden, which could be a predictive biomarker for immunotherapy.^[26] Intestinal bacteria might act on the interaction between tumor-infiltrating immune cells and host cells, thus affecting tumor development and prognosis.^[27]

In the Apc^Min/+ mice model, we verified that V. parvula could regulate the differentiation and function of B cells and increase the level of B cells in tumor tissues by increasing the level of BLyS in the serum, thus exerting an immunosuppressive effect and promoting transition through the CRA–CRC sequence. The pro-carcinogenic effect of V. parvula could be inhibited using BBR or using an anti-BLyS antibody. Serum BLyS level and B-cell infiltration in CRC tissue could also be reduced by BBR or the anti-BLyS antibody. The specific mechanism of how V. parvula affects B-cell infiltration and intestinal immune pathway remains to be further investigated.

In conclusion, our study suggested that BBR might inhibit V. parvula and further weaken the immunomodulatory effect of B cells against CRC induced by V. parvula, thereby blocking CRA–carcinoma progression. Of course, in-depth mechanistic research is necessary for the future.

Acknowledgements

We thank all patients and individuals for their participation in our study.

Funding

This project was supported by the grants from the National Key R&D Program of China (No. 2020YFA0509200), the National Natural Science Foundation of China (Nos. 82002622, 81830081, 31970718, and 81972203), the Shanghai Municipal Health Commission, Collaborative Innovation Cluster Project (No. 2019CXJQ02), and the Youth Project of Shanghai Municipal Health Commission (No. 20194Y0096).

Conflicts of interest

None.