Berberine ameliorates chronic kidney disease through inhibiting the production of gut-derived uremic toxins in the gut microbiota

Abstract

At present, clinical interventions for chronic kidney disease are very limited, and most patients rely on dialysis to sustain their lives for a long time. However, studies on the gut–kidney axis have shown that the gut microbiota is a potentially effective target for correcting or controlling chronic kidney disease. This study showed that berberine, a natural drug with low oral availability, significantly ameliorated chronic kidney disease by altering the composition of the gut microbiota and inhibiting the production of gut-derived uremic toxins, including p-cresol. Furthermore, berberine reduced the content of p-cresol sulfate in plasma mainly by lowering the abundance of g_Clostridium_sensu_stricto_1 and inhibiting the tyrosine–p-cresol pathway of the intestinal flora. Meanwhile, berberine increased the butyric acid producing bacteria and the butyric acid content in feces, while decreased the renal toxic trimethylamine N-oxide. These findings suggest that berberine may be a therapeutic drug with significant potential to ameliorate chronic kidney disease through the gut–kidney axis.

Graphical abstract

Berberine ameliorates chronic kidney disease through the gut–kidney axis by regulating the metabolism of nephrotoxins such as p-cresol by the gut microbiota.

1. Introduction

Kidney disease is a global public health problem, affecting more than 750 million people worldwide¹. An estimated 1.7 million people die each year from acute kidney injury². According to the 2015 Global Burden of Disease Study, approximately 1.2 million people died of chronic kidney disease (CKD)¹, and more than 2 million people died in 2010 due to the inability to receive dialysis.

The association between the gut microbiota and human physiological systems depends on changes in the gut environment due to external factors (dietary intake and drug treatment)³. The “gut–kidney axis” hypothesis was proposed by Mohan K. Raizada in 2018⁴, which innovatively associates the crosstalk between the gut microbiota and the kidney. CKD induces changes in the composition and metabolic activity of the gut microbiota, and renal failure changes the gut microbiota along with the nutrient and endogenous metabolomes into a clinical phenotype characterized by the production of uremic toxins⁵. The aforementioned alterations are associated with changes in the gut wall structure, which in turn impair the barrier function of the gut wall and lead to leakage of bacterial endogenous metabolites, bacterial wall products, and viable bacteria into the blood circulation⁶. Therefore, the gut microbiota will become a new therapeutic target for improving the prognosis of CKD, including relieving uremic symptoms, rescuing the metabolic changes of the gut microbiota, reducing cardiovascular complications, and alleviating abnormal immunity⁷.

The development of CKD is associated with increased blood concentrations of various endogenous metabolites excreted by healthy kidneys, including the blood urea nitrogen (BUN) and creatinine⁸. Studies both in postcolectomy CKD patients⁹ and in CKD germ-free mice have shown that microbiota-derived metabolites contribute significantly to the uremic serum metabolome¹⁰.

The best-known gut-derived uremic toxins are p-cresol sulfate and indoxyl sulfate. p-Cresol sulfate is a protein-related uremic toxin. The gut microbiota metabolizes the gut dietary aromatic amino acids such as tyrosine and phenylalanine to produce p-cresol, and the latter will be absorbed and metabolized into p-cresol sulfate in the liver¹¹. p-Cresol sulfate is usually eliminated in the urine through tubular secretion. Nevertheless, in CKD, p-cresol sulfate accumulates in plasma and increases the risk to development of cardiovascular and kidney diseases^12,13. p-Cresol sulfate causes oxidative stress in white blood cells¹⁴, and stimulates the release of endothelial granules, which in turn causes endothelial injury, while stimulating the renin–angiotensin–aldosterone system/transforming growth factor-β pathway, inducing epithelial–mesenchymal-like transition, leading to renal injury and fibrosis¹⁵. Increased p-cresol sulfate levels in CKD patients are associated with worse prognosis¹⁶.

Indoxyl sulfate is another product of tryptophan metabolism by intestinal flora. Dietary tryptophan reaching the colon is converted into indole by the gut flora and absorbed into the systemic circulation, where it is further metabolized by the liver to form indoxyl sulfate⁵. It was reported that subjects with normal renal function who consumed a high-protein diet for 2 weeks had higher serum levels and urinary excretions of indoxyl sulfate than subjects on a low-protein diet¹⁷. Indoxyl sulfate binds to plasma protein, and the binding rate is close to 90%, which will affect the dialysis behavior of indoxyl sulfate and limit its clearance rate¹⁸. In addition, indoxyl sulfate has a prooxidant effect with reduction of nitric oxide bioavailability, and indoxyl sulfate alters endothelial cell and endothelial progenitor cell migration, regeneration and control vascular smooth muscle cells proliferation. And thus, high concentration of indoxyl sulfate induces endothelial dysfunction implicated in cardiovascular morbidity and mortality during CKD¹⁸.

Trimethylamine N-oxide (TMAO) is another uremic toxin which is derived from gut microbiota-produced trimethylamine. The plasma TMAO concentrations in CKD patients is elevated¹⁹. Increased plasma concentration of TMAO is associated with tubulointerstitial fibrosis, collagen deposition, and increased biomarkers of renal injury²⁰. Thus, removal of uremic toxins is the main difficulty and challenge in the treatment of CKD at all stages. And elucidating the production mechanism of uremic toxins and how to reduce the production of uremic toxins based on the “gut–kidney axis” is a hotspots and difficulties in CKD research.

Strategies for gut microbiota intervention in CKD include fecal microbiota transplantation, specific gut microbiota-targeted interventions, and the use of prebiotics, probiotics, or dietary interventions to alleviate or control disease progression^21,22. Supplementation with certain probiotics and prebiotics reduces the production of certain gut-derived uremic toxins. In a randomized, placebo-controlled, crossover trial, 6 weeks of synbiotic therapy (prebiotic powder and probiotic capsules) in adult patients with predialysis CKD significantly reduced plasma levels of uremic toxin (p-cresol sulfate)²³. Engineered bacteria with specific phenotypic functions can be a potential approach to control CKD-associated metabolite levels in host. It was reported that genetic deletion the tryptophanase of bacteria abolishes indole production in vitro, which can be applied for the CKD-treatment in the future²⁴.

Berberine (BBR) is one of the main active components of the traditional Chinese medicine Coptidis Rhizoma, which is currently used in the clinic to treat intestinal infections. In recent years, it was reported that BBR has therapeutic effects on metabolic diseases, such as hyperlipidemia, hyperglycemia, etc.25, 26, 27, 28, 29. The structural properties of BBR result in low solubility and poor permeability (in ionic form). And as a substrate for P-glycoprotein, significant efflux occurs after BBR is absorbed, resulting in low bioavailability of BBR³⁰. And thus, BBR can accumulate in the intestine, which provides a basis for BBR to act on the gut microbiota. BBR can alter the intestinal flora structure and the concentrations of endogenous intestinal flora metabolites in various disease models31, 32, 33, 34, 35, 36 to exert its medicinal effect, which is a new hot topic in research on the effective substance and therapeutic mechanism of BBR.

2. Materials and methods

2.1. Chemicals and reagents

Berberine hydrochloride was purchased from J&K Scientific Co., Ltd. (Beijing, China). Trimethylamine oxide, sodium butyrate, adenine, creatinine, tryptophan, tyrosine, 4-hydroxyphenylacetic acid, 4-hydroxyphenylpropionic acid, indole and p-cresol were purchased from Beijing Solarbio Biotechnology Co., Ltd. (Beijing, China). Indoxyl sulfate and p-cresol sulfate were purchased from Sigma–Aldrich (St. Louis, USA). The purity of all standards was above 98%. Oxytetracycline, erythromycin, and cefadroxil were purchased from Beijing Solabio Biotechnology Co., Ltd. (Beijing, China). Chromatography-grade acetonitrile and methanol were obtained from Fisher Scientific (Fairlawn, USA). Deionized water was purchased from Hangzhou Wahaha Group Co., Ltd. (Hangzhou, China). Other chemical reagents were purchased from Sinopharm Chemical Reagent Co., Ltd. (Beijing, China). Reinforced clostridium medium (RCM) was purchased from Beijing Solarbio Biotechnology Co., Ltd. (Beijing, China). An H&E staining kit was purchased from Beijing Solarbio Biotechnology Co., Ltd. (Beijing, China). Creatinine detection kits, urea nitrogen detection kits, and inflammatory factors detection kits (TNF-α, IL-6, IL-1β) were purchased from Nanjing Jiancheng Bioengineering Institute (Nanjing, China).

Clostridium sporogenes (ATCC 19404) and Clostridium perfringens (ATCC 13124) were purchased from Guangdong Huankai Microorganism Technology Co., Ltd. (Guangzhou, China). The Lactobacillus rhamnosus (ATCC7469) was purchased from the American Type Culture Collection (ATCC).

2.2. Animals

Male Sprague–Dawley (SD) rats (180–200 g) were provided by Beijing Vitalriver Laboratory Animal Technology Co., Ltd. (Beijing, China). All animals had free access to food and water, and were housed in a ventilated room with a circulation of 12 h of light and 12 h of darkness. The temperature was maintained at 20–24 °C and humidity of 40%–60%. Animals were fasted for 12 h before the experiment and had free access to water. This study was approved by the Experimental Animal Ethics Committee of the Chinese Academy of Medical Sciences and Peking Union Medical College (No. 00003403, date of approval, 14 Jun 2021), and strictly followed the instruction of Organizational Guidelines and Ethics Guidelines of the Experimental Animal Ethics Committee.

2.3. Animal studies

The animals (200 g male SD rats) were randomly divided into 6 groups, 10 in each group, including the control group (Control), CKD model group (Model), BBR treatment group (BBR), antibiotic intervention group (Antibiotic), probiotic treatment group (L. rhamnosus) and sodium butyrate treatment group (Sodium Butyrate). The CKD model was initially established from SD rats with 2 months of a CKD-inducing formula feed containing 0.25% (w/w) adenine. During modeling, serum creatinine and BUN levels were measured. After the model was successfully established, drug intervention was performed for 1 month. At the end of the experiment, the animals were sacrificed after anesthesia, and the blood, feces and the kidney were collected for further study.

The details of drug interventions were as follows: 1) Control group (Control): fed with maintained feed; 2) Model group (Model): fed with a CKD-inducing formula feed containing 0.25% (w/w) adenine; 3) BBR treatment group (BBR): fed with CKD-inducing formula feed + oral administration of BBR (100 mg/kg/day); 4) antibiotic intervention group (Antibiotic): fed with a CKD-inducing formula feed + gavage with mixed antibiotics (oxytetracycline 300 mg/kg/day; erythromycin 300 mg/kg/day; cefadroxil 100 mg/kg/day); 5) probiotic treatment group (L. rhamnosus): fed with a CKD-inducing formula feed + gavage of probiotics (1 × 10¹⁰ CFU/day); 6) sodium butyrate treatment group (Sodium Butyrate): fed with CKD-inducing formula feed + gavage of sodium butyrate (600 mg/kg/day).

The L. rhamnosus (ATCC7469) was purchased from the ATCC, and cultured anaerobically in MRS broth medium to an appropriate concentration, and diluted to 1 × 10¹⁰ CFU/mL for drug delivery.

2.4. Kidney H&E staining

H&E staining was performed as follows. The paraffin sections were first dewaxed with xylene (I) and xylene (II) for 5 min and then treated with an ethanol gradient (anhydrous ethanol for 5 min, 95% ethanol for 2 min, 80% ethanol for 2 min, and 70% ethanol for 2 min, followed by distilled water for 2 min). Deparaffinized tissue sections were stained with a hematoxylin staining solution for 20 min and rinsed with tap water. After application of a differentiation solution for 30 s, tissue samples were immersed in water for 15 min. The samples were stained with an eosin staining solution for 30 s and rinsed under running water. After soaking in water for 5 min, the samples were dehydrated with an ethanol gradient, cleared in xylene, and sealed with neutral glue. Finally, it was observed and photographed with an optical microscope.

2.5. Detection of plasma creatinine, blood urea nitrogen and inflammatory factors

At two months after modeling and the end of intervention, the mice were fasted for 12 h before serum samples were collected. The detection of creatinine, BUN, and inflammatory factors (TNF-α, IL-6, IL-1β) content was carried out according to the kit instructions.

2.6. Fecal microbiome 16S rRNA detection

Feces of rats were collected after 1 month of treatment. Microbial DNA in fecal samples was extracted using the E.Z.N.A. Soil DNA Kit (Omega Biotek, USA) according to the instructions. The V3–V4 region of the microbial 16S rRNA gene was amplified with primer pairs 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′). PCR products were then extracted from 2% agarose gels and purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, USA). Purified amplicons were sequenced using Illumina MiSeq to analyze the diversity of bacterial groups in animal feces from each group.

2.7. Establishment of an LC–MS/MS method for plasma renal function-related metabolites

The LC–MS/MS 8060 mass spectrometer (Shimadzu) equipped with an electrospray ion (ESI) source was used for quantitative analysis of the plasma renal function-related metabolites. The detection method was established based on multiple reaction mode (MRM). The mass spectrometry parameters were set as follows: 1) nebulizing gas flow: 2.7 L/min; 2) heating gas flow: 9.8 L/min; 3) drying gas flow: 9.8 L/min; 4) interface temperature: 300 °C; 5) DL temperature: 250 °C; 6) heating block temperature: 400 °C. Quantitative transitions for the metabolites were as follows: indoxyl sulfate: MRM− 212.20 → 79.90 (CE: 22.0); and p-cresol sulfate: MRM− 187.20 → 107.00 (CE: 25.0); TMAO: MRM+ 76.15 → 58.05 (CE: −23.0); creatinine: MRM+ 114.35 → 44.15 (CE: −18.0).

An Alltima C18 (150 mm × 4.6 mm, 5 μm, Grace, USA) was used for separation of the metabolites. The mobile phase was 0.05% ammonia in water (phase A)–acetonitrile (phase B). The elution mode was set as gradient elution, and the flow rate was 0.4 mL/min. And the column temperature was maintained at 40 °C. The specific phase B gradient was as follows: 0.01 min: 20%; 2 min: 20%; 6 min: 80%; 8 min: 95%; 8.01 min: 20%; and 12 min: stop.

Corresponding methodological validation was carried out by determination of specificity, linearity, intra- and inter-day accuracy and precisions, recovery and stability as described in our previous study³¹.

2.8. Establishment of an LC–MS/MS method for fecal renal function-related metabolites

The LC–MS/MS 8060 mass spectrometer (Shimadzu) equipped with an ESI source was used for quantitative analysis of the fecal renal function-related metabolites. The mass spectrometry parameters were set as follows: nebulizing gas flow: 2.7 L/min; 2) heating gas flow: 9.8 L/min; 3) drying gas flow: 9.8 L/min; 4) interface temperature: 300 °C; 5) DL temperature: 250 °C; and 6) heating block temperature: 400 °C. The quantification transitions of the metabolites were as follows: tryptophan: MRM− 203.35 → 115.90 (CE: 21.0); p-cresol: SIM– 107.15; indole: SIM– 116.05; tyrosine: MRM− 180.30 → 163.20 (CE: 15.0); and butyric acid: SIM– 87.10.

An xSelect HSS PFP (100 mm × 2.0 mm, 1.8 μm, Waters, USA) was used for separation of the metabolites. The mobile phase was 0.05% ammonia in water (phase A)–acetonitrile (phase B). The elution mode was gradient elution, and the flow rate was 0.4 mL/min. And the column temperature was maintained at 40 °C. The specific phase B gradient was as follows: 0.01 min: 10%; 2 min: 10%; 6 min: 99%; 7.30 min: 99%; 7.31 min: 10%; and 13 min: stop.

Corresponding methodological validation was carried out by determination of specificity, linearity, intra- and inter-day accuracy and precisions, recovery and stability as described in our previous study³¹.

2.9. Plasma and stool sample preparation

50 μL of the collected plasma samples were added to 3 times volume of the methanol containing 100 ng/mL glipizide (IS) for extraction the metabolites and the precipitation of the protein. After centrifugation at 13,400 × g for 5 min, 10 μL of the supernatant was collected for analysis. The collected fresh feces were added to 3 times volume of double-distilled water, mixed to obtain homogenate, and centrifuged at 13,400 × g for 5 min. Then, 100 μL of supernatant was added to 3 times volume of methanol containing 100 ng/mL glipizide. After vortexing, the samples were centrifuged at 13,400 × g for 5 min, and 10 μL of the supernatant was injected for analysis.

2.10. In vitro incubation of BBR and SD rat intestinal flora

After 6 SD rats were sacrificed, the colonic contents were collected, and sterilized anaerobic medium was added at a ratio of 1.0 g:20 mL, with gentle agitation. After filtration, the medium containing the intestinal flora (mixed medium) was placed under N₂ and preincubated at 37 °C for 30 min before use. The incubation system consisted of 10 μL of BBR in methanol (final concentrations: 50, 100, and 200 μg/mL, respectively, n = 4) or pure methanol as control and 990 μL of mixed medium. Incubation was performed in a shaking incubator (Longyue Instrument Co., Ltd., Shanghai, China) at 37 °C and 200 rpm. During the experiment, the incubation system was maintained in an anaerobic environment. The reaction was terminated by adding 3-fold volume of 100 ng/mL glipizide in methanol (IS) at 0, 6, 12 and 24 h, respectively. And then each sample was centrifuged at 13,400 × g for 10 min in a refrigerated centrifuge at 4 °C to precipitate proteins. 5 μL of the supernatant was analyzed by an LCMS-8060 to detect the contents of p-cresol, indole, tyrosine, and tryptophan in the incubation system.

2.11. Study of metabolites in tyrosine–p-cresol pathway in vitro

An analytical method for detection of a series of intermediates in the tyrosine–p-cresol pathway was established by the LC–MS/MS 8060 mass spectrometer (Shimadzu) equipped with an ESI source. The mass spectrometry parameters were set as follows: 1) nebulizing gas flow: 2.7 L/min; 2) heating gas flow: 9.8 L/min; 3) drying gas flow: 9.8 L/min; 4) interface temperature: 300 °C; 5) DL temperature: 250 °C; and 6) heating block temperature: 400 °C. Quantitative transitions for the compounds were as follows: 4-hydroxyphenylacetic acid: MRM− 151.00 → 93.20 (CE: 15.0); p-cresol: SIM– 107.15; 4-hydroxyphenylpropionic acid: MRM− 165.25.00 → 121.00 (CE: 13.0); and tyrosine: MRM− 180.30 → 163.20 (CE: 15.0).

An xSelect HSS PFP (100 mm × 2.0 mm, 1.8 μm, Waters, USA) was used for substance separation. And the mobile phase was 0.05% ammonia in water (phase A)–acetonitrile (phase B). The elution method was gradient elution with a flow rate of 0.4 mL/min. The column temperature was maintained at 40 °C. The specific phase B gradient was as follows: 0.01 min: 10%; 2 min: 10%; 6 min: 99%; 7.30 min: 99%; 7.31 min: 10%; and 13 min: stop.

Corresponding methodological validation was carried out by determination of specificity, linearity, intra- and inter-day accuracy and precisions, recovery and stability as described in our previous study³¹.

2.12. BBR molecular virtual docking analysis

Docking analysis between BBR and TyrB were conducted using Discovery Studio Client software (v16.1.0.15350). The crystal structure of TyrB 4WD2 was obtained from the Protein Data Bank (PDB) database and the docking between BBR and TyrB was performed using the CDOCKER approach as the docking algorithm for our docking study. During the docking process, the protein was kept rigid while the ligands were treated as fully flexible and a final minimization step was used to refine the docked positions. In addition, the parameters were set to the default values.

2.13. Inhibitory effect of BBR on two Clostridium strains

Two strains, Clostridium sporogenes (ATCC 19404) and Clostridium perfringens (ATCC 13124), were purchased from Guangdong Huankai Microbial Technology Co., Ltd. Bacteria were activated using reinforced Clostridium medium (RCM). And then two Clostridium strains were expanded with the initial concentrations of 1 × 10⁶ CFU/mL, respectively. For the inhibitory experiment, different concentrations of BBR (0, 8, 40 and 100 μg/mL) were incubated with these two Clostridium strains in an anaerobic box for 8 h. The OD₆₀₀ was measured using a UV spectrophotometer from Shimadzu Corporation, Japan.

2.14. Statistical analysis methods

Data were analyzed using Prism version 5.0 (GraphPad Software, La Jolla, CA, USA). Clustering correlation heatmap with signs was performed using the OmicStudio tools at https://www.omicstudio.cn. The data in the figure are expressed as the mean ± standard deviation (S.D.), and the differences between groups were analyzed by two-sided t test. P values less than 0.05 were considered statistically significant (∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001).

3. Results

3.1. BBR ameliorated adenine-induced CKD in rats

The experimental scheme was shown in Fig. 1A. Rat-maintained feed with supplement of 0.25% adenine was used for the CKD modeling. After 2 months of modeling, the drug interventions were conducted for 1 month. The Antibiotic group was introduced to evaluate the role of gut microbiota in the efficacy of BBR on CKD. And pseudo-germ-free states were established using an antibiotic formula which was commonly applied in our laboratory³¹. Furthermore, our previous work has shown that BBR can stimulate the production of short-chain fatty acid of gut microbiota³⁷, especially the butyric acid, and the latter has been reported to have protective effect on renal function³⁸. Since the butyric acid is too irritating to the gastric mucosa, the Sodium Butyrate group (the sodium butyrate, the butyric acid sodium salt, was administered to the rats) was introduced to study the effect of BBR-related short-chain fatty acid on CKD. Since the short-chain fatty acids are the secondary metabolites of the gut microbiota, it is possible to choose a strain of bacteria as short-chain fatty acids producer for the probiotic intervention on CKD. Previous studies have shown the renal protective effects of a short-chain fatty acid producer, L. rhamnosus, on a 5/6 nephrectomy mouse model³⁹ and a cisplatin-induced chronic nephrotoxicity model⁴⁰. Meanwhile, pilot experiments showed that L. rhamnosus could not produce typical uremic toxins such as indole and p-cresol. Thus, L. rhamnosus was chosen as a probiotic for treatment of CKD in L. rhamnosus group.

Figure 1.

BBR ameliorated 0.25% adenine-induced CKD in rats. (A) Experimental grouping and scheme. (B, C) BUN (B) and serum creatinine levels (C) after 2 months of consumption of a 0.25% adenine diet in CKD model rats. (D, E) The levels of serum creatinine (D) and BUN (E) after 4 weeks of treatment in 0.25% adenine-induced CKD model rats. The data are shown as mean ± S.D., ∗∗P < 0.01, ∗∗∗P < 0.001, n = 10. (F) H&E staining results of kidneys in 0.25% adenine-induced CKD model rats after 4 weeks of treatment. Vacuoles with unclear boundaries which indicated the kidney damage were marked with rad arrows, and the glomerular was marked with green arrows (scale bar, 100 μm).

During the experiment, only 2 animals in the model group and 2 animals in the Antibiotic group died due to the severe kidney failure (swell bodies and kidneys surrounded by the gelatinous substance can be observed), and none of the animals in the other groups died. The modeling results are shown in Fig. 1B and C. Two months after feeding with 0.25% adenine supplementary food, the BUN and creatinine levels in the Model group increased significantly (∗∗∗P < 0.001). Compared with the Control group, the BUN and the creatinine levels in Model group increased by 82.5% and 669.4%, respectively, indicating that the CKD model was successfully established.

After 2 and 4 weeks of drug or probiotic interventions, the creatinine level in serum samples was measured. The results are shown in Supporting Information Fig. S1A (2 weeks) and Fig. 1D (4 weeks). After 2 weeks of drug or probiotic interventions (Fig. S1A), the serum creatinine level was significantly lower in the BBR group (reduced by 15.6%, ∗P < 0.05) and the L. rhamnosus group (reduced by 14.9%, ∗∗∗P < 0.001), compared with the Model group, while the serum creatinine level in Sodium Butyrate group was declined slightly but not significantly (reduced by 6.2%). The CKD progression was more severe in the Antibiotic group. The serum creatinine level was higher by 14.0% in Antibiotic group. After 4 weeks of drug or probiotic intervention (Fig. 1D), the serum creatinine level in the BBR group (reduced by 29.5%, ∗∗P < 0.01), the L. rhamnosus group (reduced by 28.6%, ∗∗P < 0.01) and Sodium Butyrate group (reduced by 29.8%, ∗∗∗P < 0.001) was declined significantly, compared to Model group. And the therapeutic effects of above three interventions were similar, while the serum creatinine level in Antibiotic group also demonstrated a slight decrease with no significance.

Next, the BUN level in serum samples was analyzed after 2 and 4 weeks of continuous drug or probiotic interventions, and the results are shown in Fig. S1B (2 weeks) and Fig. 1E (4 weeks). After 2 weeks of drug or probiotic interventions, the BBR group (reduced by 11.5%, ∗P < 0.05) and Sodium Butyrate group (reduced by 10.7%, ∗P < 0.05) showed a significant decrease in BUN content compared with that in the Model group, while the BUN level in the L. rhamnosus group was slightly lower by 3.6% with no significant difference. Meanwhile, there was no significant difference between the Antibiotic group and the Model group. After 4 weeks of drug or probiotic interventions, the BUN content of the BBR group, the L. rhamnosus group and the Sodium Butyrate group was declined significantly by 35.2% (∗∗∗P < 0.001), 29.1% (∗∗∗P < 0.001) and 26.9% (∗∗∗P < 0.001), respectively, compared with that of the Model group. And there was no significant difference between the Antibiotic group and the Model group.

Then, the kidneys of CKD model rats were sectioned, and H&E staining was performed after sectioning to evaluate the kidney protective effect of the BBR. The results are shown in Fig. 1F. Compared with the Control group, a large number of vacuoles with unclear boundaries (red arrow) can be observed in the kidney of the Model group, and the glomerular structure was completely destroyed and atrophied. After the interventions of BBR and the sodium butyrate, the number of vacuoles was greatly reduced, and the glomerular structure was maintained (green arrow), which indicated the protective effect on CKD of BBR and sodium butyrate. While in the L. rhamnosus group, vacuoles with unclear boundaries were still existed but the becomes smaller, and the glomerular damage was still can be found but less severe. Thus, the L. rhamnosus intervention still showed a certain renal protection effect in which the overall damage was between the Model group and the BBR group. However, the renal function was more severely damaged in the antibiotic group, since more vacuoles with larger area can be observed, and the glomerular structure was completely destroyed.

The inflammatory factors including TNF-α, IL-6 and IL-1β in serum were detected to evaluate the CKD-related inflammation progression. As shown in Supporting Information Fig. S9A–S9C, the TNF-α (∗∗∗P < 0.001), IL-6 (∗∗∗P < 0.001) and IL-1β (∗∗P < 0.01) levels in the Model group were significantly higher compared with those in Control group. Meanwhile, the TNF-α, IL-6 and IL-1β content of the BBR group declined significantly by 35.8% (∗∗P < 0.01), 40.2% (∗∗P < 0.01) and 32.1% (∗P < 0.05), respectively, compared with that of the Model group. The TNF-α, IL-6 and IL-1β content of the Antibiotic group was found to decrease due to its anti-inflammatory properties. The inflammatory factors in Sodium Butyrate group and L. rhamnosus group were also lower but not significantly than those in Model group (except IL-6, ∗P < 0.05).

3.2. BBR alters the composition of the intestinal flora in rats with CKD

To further study the compositional changes of gut microbiota in CKD states and the effect of BBR on gut microbiota in CKD model, and to explore the relationship between gut microbiota imbalance and host disease, the animal feces were firstly collected, and 16S rRNA analysis was conducted. The results are shown in Fig. 2, and Supporting Information Figs. S2–S5. α-Diversity indicated the number of microbial species in one single sample and the proportion of each species. Chao1 index, Shannon index, PD whole tree, and Observed species are the most used indices to measure α-diversity. As shown in Figs. 2A, B and S2A and S2B, the Chao1 index, Shannon index, PD whole tree, and observed species of the Model group were decreased compared with those of the Control group, indicating that the species diversity and uniformity of the intestinal flora in CKD model with were significantly perturbated and the macroscopic changes in intestinal flora occurred with the progress of CKD. With the application of antibiotics, the gut microbiota in the animals were significant inhibited, and the Chao1 index, Shannon index, PD whole tree, and observed species of the Antibiotic group decreased significantly, indicating the diversity of the gut flora was severely reduced under antibiotic treatment. The similar observations were found in the BBR group. α-Diversity in BBR group has shown a downtrend since the BBR is a natural antibiotic. Of note, the degree of decline of α-diversity in BBR was slighter than that in the Antibiotic group. So, further comparisons such as β-diversity are needed to analyze the similarities and differences between BBR and antibiotics in the regulation of the intestinal flora under the CKD state. Finally, the L. rhamnosus group and Sodium Butyrate group which showed unchanged or slightly improved α-diversity, compared with the Model group, meanwhile, the α-diversities in the L. rhamnosus group and Sodium Butyrate group were approaching the Control group, indicating the interventions of L. rhamnosus and sodium butyrate restored the disturbance of intestinal flora under the CKD state to a certain extent.

Figure 2.

BBR modulated gut microbiota diversity in CKD model rats. (A, B) BBR affected the α-diversity of gut bacteria in CKD model: (A) Chao1 index; (B) Shannon index. (C) β-Diversity of gut bacteria after BBR treatment in CKD model (PCA). (D) Volcano plot analysis of the intestinal flora of the Model group and the Control group at the genus level. (E) Heatmap of the gut microbiota in the CKD model at the genus level. The data are shown as mean ± S.D., ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, n = 6.

β-Diversity is an index used to measure the similarity of the composition of bacterial groups between different samples (especially between different groups), that is, to reveal the differences in the composition of bacterial groups between samples. Thus, the β-diversity in each group was analyzed in Fig. 2C. According to the PCA (principal component analysis) results, the Model group and the Control group can be effectively distinguished in Fig. 2C, indicating that the intestinal flora constituents in the Model group and the Control group were quite different. At the same time, it was observed that the Antibiotics group was clearly separated from the Model group and the Control group, indicating that the intestinal flora was greatly changed under the action of antibiotic treatment. Additionally, the BBR group was also located from the Control group and the Model group. Of note, the BBR group was far distant from the Antibiotic group, indicating the unique and complex ability of BBR to regulate the intestinal flora, which may be related to the mechanism of therapeutic effect of BBR on CKD. Finally, the introduction of sodium butyrate brought the Sodium Butyrate group and the Control group closer in the PCA, while the L. rhamnosus group was located between the Control group and the Model group, verifying the state that the interventions of L. rhamnosus and sodium butyrate restored the disturbance of intestinal flora under the CKD state to a certain extent.

To further analyze the specific composition of the intestinal flora under the CKD state, we analyzed the intestinal flora at the phylum and genus levels in each group. Supporting Information Fig. S3 shows the difference of the gut microbiota at phylum level. The intestinal flora is mainly composed of bacteria from the phyla Firmicutes and Bacteroidota. Except for the antibiotic group, the sum of these two groups accounted for more than 90%, which were the absolute dominant strains. And it was worth noting that the abundance of Firmicutes and Bacteroidota was significantly decreased in the Antibiotic group, replaced by a significant enrichment of Proteobacteria strains. Next, we analyzed the well-known Firmicutes/Bacteroidota ratio, which was first studied in obese patients, and the lower Firmicutes/Bacteroidota ratio correlated with a healthier state⁴¹. In Fig. S4A and S4B. The ratio of Firmicutes/Bacteroidota was significantly increased in the Model group, in which there was a significant decrease in the abundance of Bacteroidota bacteria, while BBR significantly reduced this ratio. This function of BBR was mainly achieved by increasing the abundance of bacteria in the phylum Bacteroidota.

To further screen the genus with significant differences in abundance, we analyzed all 172 detected bacteria genera. The genera of which fold change > 2 and P value < 0.05 were selected and was shown in the form of volcano map. As demonstrated in Fig. 2D, the blue dots represented the bacterial genera with significant differences in abundance between the Model group and Control group. Next, the diversity of abovementioned bacterial genera with significant differences in all groups was analyzed and demonstrated in a heatmap. The results are shown in Fig. 2E. In the Model group and the Control group, it can be seen that g_Turicibacter, g_Clostridium_sensu_stricto_1, g_Romboutsia, g_Corynebacterium, g_Adlercreutzia, g_Jeotgalicoccus, g_Bifidobacterium, g_Candidatus_Saccharimonas, and g_Candidatus_Soleaferrea were significantly enriched in the Model group, whereas g_UCG-009, g_Eubacterium_xylanophilum_group, g_NK4A214_group, g_Lachnospiraceae_NK4A136_group, g_Rothia, g_Lachnospiraceae_UCG-001, g_Vagococcus, g_Streptococcus, g_Lachnospiraceae_UCG-006, g_Colidextribacter and g_Alistipes were significantly depleted in the Model group. Among the abovementioned genera, we found that various genera were closely related to the progression of CKD. g_Clostridium_sensu_stricto_1 was highly enriched in CKD models, and it was reported to express a 4-hydroxyphenylacetic acid decarboxylase gene, which encodes the 4-hydroxyphenylacetic acid decarboxylase, and the latter can mediate the production of p-cresol⁴². The uremic toxin sorbent AST-120 can reduce the abundance of g_Clostridium_sensu_stricto_1 and fecal p-cresol content⁴². g_Romboutsia has been reported to be positively correlated with circulating TMAO, which is a typical uremic toxin⁴³. g_Adlercreutzia has been reported to be positively correlated with plasma phenyl sulfate, and there is a significant correlation between phenyl sulfate and proteinuria, and inhibition of phenyl sulfate production can reduce proteinuria in diabetic mice⁴⁴. g_Jeotgalicoccus has been reported to be associated with renal calcium oxalate crystals. The intervention of BBR significantly reduced the abundance of abovementioned g_Clostridium_sensu_stricto_1 (∗∗P < 0.01, Fig. 5B), g__Adlercreutzia (∗∗P < 0.01, Fig. 5C), g__Romboutsia (∗P < 0.05, Fig. S5A), and g__Jeotgalicoccus (∗P < 0.05, Fig. S5B). We can infer that BBR could regulate the abundance of the renal toxin producer and thus regulate the renal toxin in the body, which maybe promote the efficacy of the renal protection of BBR.

Figure 5.

(A) Heatmap of the correlation analysis between uremic toxins and gut bacteria at the genus level. (B) BBR reduced the abundance of g_Clostridium_sensu_stricto_1 in SD rats with CKD (n = 6). (C) BBR reduced the abundance of g_Adlercreutzia in SD rats with CKD (n = 6). (D) Fold changes in the bacteria genera which positive correlated with renal toxins between the BBR group and the Model group (expressed by log₂ fold change) (n = 6). (E) The content of butyric acid in the feces of SD rats with CKD after 4 weeks of BBR treatment (n = 10). (F) Fold changes in the butyric acid producing bacteria between the BBR group and the Model group (expressed by –log₂ fold change) (n = 6). The data are shown as mean ± S.D., ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

3.3. BBR reduced gut-derived uremic toxin production in vivo

Since the BBR regulated the potential renal toxin producer in the gut microbiota, it was necessary to evaluate the renal toxin regulatory ability of BBR. A highly sensitive analysis method for creatinine, TMAO, indoxyl sulfate and p-cresol sulfate in blood samples and a method for tryptophan, p-cresol, indole, and tyrosine in fecal samples were established using LC–MS/MS. All analysis methods have been methodologically validated (Supporting Information Tables S1–S3, S5, S6, Figs. S10 and S11). Fig. 3A shows the mass spectra of indole and p-cresol in the fecal samples of the Control group and the Model group, and the content of p-cresol in the fecal samples of the Control group was extremely low.

Figure 3.

BBR modulated gut microbiota metabolite levels in CKD model SD rats. (A) Mass spectrometry chromatograms of p-cresol and indole in feces samples (black curve, Control group; red curve, Model group; the content of p-cresol in the feces samples of the Control group was extremely low). (B) Schematic diagram of the tyrosine–p-cresol–p-cresol sulfate and tryptophan–indole–indoxyl sulfate production pathway. (C) The content of tryptophan in the feces of rats with CKD after 4 weeks of BBR treatment. (D) Indole content in feces of rats with CKD after 4 weeks of BBR treatment. (E) The content of indoxyl sulfate in the plasma of rats with CKD after 4 weeks of BBR treatment. (F) The content of tyrosine in the feces of rats with CKD after 4 weeks of BBR treatment. (G) The content of p-cresol in the feces of rats with CKD after 4 weeks of BBR treatment. (H) Plasma levels of p-cresol sulfate in rats with CKD after 4 weeks of BBR treatment. The data are shown as mean ± S.D., ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, n = 10.

The tryptophan–indole–indoxyl sulfate pathway and the tyrosine–p-cresol–p-cresol sulfate pathway are illustrated in Fig. 3B. In brief, the indole is metabolized from the tryptophan by the gut microbiota, and then indole is absorbed into the body and metabolized into indoxyl sulfate by the liver. And the tyrosine can be metabolized into p-cresol, and the latter will continue transform into p-cresol sulfate under the metabolic effect of the liver. After 2 months of modeling with 0.25% adenine supplementary feed, the metabolites in the abovementioned pathways were detected (Supporting Information Fig. S6A–S6G). The concentration of tryptophan in the feces of the Model group was significantly lower (reduced by 59.1%, ∗∗∗P < 0.001, Fig. S6A) than the Control group. Meanwhile, a significant increase was also observed in the concentration of indole in feces (increased by 53.4%, ∗∗P < 0.01, Fig. S6B), which indicated the changes in the intestinal bacteria of the CKD model resulted in the increased utilization of tryptophan to produce more indole. The higher concentration of indoxyl sulfate in plasma (increased by 941.2%, ∗∗∗P < 0.001, Fig. S6C) was also confirmed the increased production of indole by the gut microbiota under the CKD state. At the same time, the fecal tyrosine concentration in the Model group also declined significantly (reduced by 27.2%, ∗P < 0.05, Fig. S6D), while the abundance of the p-cresol in feces, the most downstream metabolite of tyrosine, significantly increased by 331.9% (∗∗∗P < 0.001, Fig. S6E). And the concentration of p-cresol sulfate in the plasma was also significantly higher by 2552.2% (∗∗∗P < 0.001, Fig. S6F), which indicated the changes in the intestinal bacteria of the CKD model resulted in the increased activity of tyrosine–p-cresol–p-cresol sulfate pathway. Furthermore, the renal toxin TMAO in the plasma was also higher after 2 months of modeling than Control group (increased by 542.3%, ∗∗∗P < 0.001, Fig. S6G). The findings above were consistent with the previous report in which the intestinal flora of CKD shifts from glycolytic fermentation to proteolytic fermentation⁴⁵, increasing the utilization of amino acids and resulting in a significant impact on the body.

After drug and probiotic interventions, the tryptophan–indole–indoxyl sulfate pathway and the tyrosine–p-cresol–p-cresol sulfate pathway were evaluated. As shown in Fig. 3C, after 4 weeks of interventions, the fecal content of tryptophan was significantly higher in the BBR group (∗P < 0.05), Antibiotic group (∗∗∗P < 0.001), L. rhamnosus group (∗∗∗P < 0.001) and Sodium Butyrate group (∗∗∗P < 0.001), compared with that in the Model group, respectively. At the same time, As shown in Fig. 3F, after 4 weeks of interventions, the fecal content of tyrosine was significantly increased in the BBR group (∗∗∗P < 0.001), Antibiotic group (∗∗∗P < 0.001), L. rhamnosus group (∗∗∗P < 0.001) and Sodium Butyrate group (∗∗P < 0.01), compared to the Model group. Thus, we can infer that BBR, L. rhamnosus and sodium butyrate could significantly reduce the utilization of tyrosine and tryptophan by the gut microbiota. Next, the contents of indole and p-cresol, the downstream metabolites of tryptophan and tyrosine were determined. As shown in Fig. 3D, compared with the Model group, the concentration of indole in feces was slightly lower in BBR group, Antibiotic group, and L. rhamnosus group, but with no significance. However, as shown in Fig. 3G, after 4 weeks of treatment, fecal content of p-cresol in the BBR group (∗∗P < 0.01), the Antibiotic group (∗∗∗P < 0.001), L. rhamnosus group (∗P < 0.05), and the Sodium Butyrate group (∗∗P < 0.01) was significantly reduced, compared with Model group. Among them, BBR (reduced by 54.0%) and antibiotics (reduced by 96.3%) showed the best efficacy on inhibition the production of p-cresol by the gut microbiota. Then, the correlation study showed that the content of p-cresol in the feces was positively correlated with inflammatory factors (TNF-α, R² = 0.3602, P = 0.0005, IL-6, R² = 0.1892, P = 0.0163, IL-1β, R² = 0.1950, P = 0.0146, Fig. S9D), which was consisted with the therapeutic effect on CKD of BBR since BBR downregulated the content of p-cresol in the feces. The indoxyl sulfate and the p-cresol sulfate contents in the plasma were also detected after 4 weeks of treatment. As shown in Fig. 3E, the use of BBR (∗∗P < 0.01), L. rhamnosus (∗∗∗P < 0.001) and the sodium butyrate (∗∗∗P < 0.001) significantly reduced indoxyl sulfate content in plasma, while the Antibiotic group showed no obvious effect. And the p-cresol sulfate content in plasma samples was shown in Fig. 3H. Compared with the Model group, p-cresol sulfate content in the BBR group and the Antibiotics group was significantly lower by 72.0% (∗∗P < 0.01) and 87.1% (∗∗∗P < 0.001), respectively. However, the L. rhamnosus and the sodium butyrate showed no significant effect on reduction of p-cresol sulfate. Furthermore, the TMAO levels in plasma, which is closely related to renal vascular sclerosis and renal fibrosis, showed a significant reduction in the BBR group (reduced by 29.8%, ∗P < 0.05) and the Antibiotic group (reduced by 42.9%, ∗∗∗P < 0.001), compared with the Model group (Supporting Information Fig. S7A).

Thus, BBR showed a strong effect of reducing uremic toxins, especially uremic toxins in the regulation of tyrosine–p-cresol–p-cresol sulfate pathway, thereby reducing the concentration of p-cresol sulfate in plasma.

3.4. BBR regulated the tyrosine–p-cresol pathway in the gut microbiota in vitro

Since BBR showed a strong effect of inhibition of renal toxins, especially the tyrosine–p-cresol–p-cresol sulfate pathway in vivo, we next focused on the direct interaction between the BBR and gut microbiota, and the regulatory ability of BBR on tyrosine–p-cresol pathway in the gut microbiota in vitro. A sensitive determination method for the tyrosine–p-cresol pathway was established, and the analysis method has been methodologically validated (Tables S1, S4–S6, Figs. S10 and S11). Six SD rats were anesthetized and sacrificed. After collecting the mixed colonic contents, BBR and rat intestinal bacteria were incubated together for 24 h, and the regulation of uremic toxin synthesis pathway in the incubation system was detected at 0, 6, 12 and 24 h, respectively.

As shown in Supporting Information Figs. S8A and 4B, it was observed that the tryptophan and the tyrosine in the incubation system were consumed by the gut microbiota with prolonged incubation time and were almost completely run out at 24 h (Figs. S8A and 4B). After BBR was added at a final concentration of 100 μg/mL, metabolism of tyrosine by intestinal bacteria was significantly inhibited (∗P < 0.05, Fig. 4B), while BBR had little effect on tryptophan metabolism by intestinal bacteria (Fig. S8A). Moreover, as shown in Figs. S8B and 4D, the concentration of indole (metabolite of tryptophan) and p-cresol (metabolite of tyrosine) was gradually increased. And BBR significantly inhibited the production of p-cresol in the incubation system (∗P < 0.05, Fig. 4C), but the production of indole was not obviously affected (Fig. S8B), which was consistent with the in vivo results.

Figure 4.

BBR modulated the production of gut microbiota-associated uremic toxins in vitro. (A) Schematic diagram of the tyrosine-p-cresol pathway. (B) The effect of BBR on tyrosine metabolism in SD rat colon contents (n = 4). (C) The effect of BBR on p-cresol production in SD rat colon contents (n = 4). (D) BBR at different concentrations inhibited tyrosine metabolism in SD rat colon contents in a dose-dependent manner (n = 3). (E) Different concentrations of BBR inhibited the production of 4-hydroxyphenylacetic acid in SD rat colon contents in a dose-dependent manner (n = 3). (F) Different concentrations of BBR inhibited the production of p-cresol in SD rat colon contents in a dose-dependent manner (n = 3). The data are shown as mean ± S.D., ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

Next, we mainly focused on regulatory effect of BBR on the tyrosine–p-cresol pathway in vitro since BBR significantly inhibited the production of p-cresol and had little effect on indole in the in vitro incubation system. Fig. 4A illustrates the p-cresol production pathway by the gut microbiota. In brief, tyrosine first can be catabolized into 4-hydroxyphenylpyruvate by aromatic amino acid aminotransferase (Aat). And then the latter is metabolized to 4-hydroxyphenylacetic acid under the action of a series of enzymes including ferredoxin oxidoreductase A (PorA). 4-Hydroxyphenylpyruvate can also be metabolized into hydroxyphenylpropionic acid under the action of a series enzymes including phenyllactate dehydrogenase (FldH). Finally, the 4-hydroxyphenylacetic acid is metabolized to p-cresol under the action of hydroxyphenylacetate decarboxylase (Hpd). Another pathway to produce p-cresol is the direct cleavage of p-cresol from tyrosine⁴⁶. Thus, a quantitative method for tyrosine, 4-hydroxyphenylacetic acid, 4-hydroxyphenylpropionic acid and p-cresol was established by LC–MS/MS. And then the concentration of tyrosine, 4-hydroxyphenylacetic acid, 4-hydroxyphenylpropionic acid and p-cresol was analyzed in the in vitro intestinal bacterial incubation system with various BBR concentrations (final concentration: 0, 50, 100, 200 μg/mL) at 0, 1, 2, 4 and 12 h, respectively. As shown in Fig. 4D–F, except that 4-hydroxyphenylpropionic acid was not detected in the in vitro intestinal bacterial incubation system, and we observed that BBR significantly reduced the utilization of the tyrosine by the gut microbiota at 12 h (∗P < 0.05, Fig. 4D) and showed a dose-dependent manner (reduced by 119.3% at 50 μg/mL, 446.3% at 100 μg/mL, and 579.2% at 200 μg/mL). At the same time, the production of 4-hydroxyphenylacetic acid was also significantly inhibited by BBR with a dose-dependent manner (reduced by 14.5% at 50 μg/mL, 36.4% at 100 μg/mL, ∗P < 0.05, and 62.3% at 200 μg/mL, ∗∗∗P < 0.001). And finally, the content of p-cresol was significantly decreased in a dose-dependent manner (reduced by 20.6% at 50 μg/mL, 37.7% at 100 μg/mL, ∗P < 0.05, and 65.2% at 200 μg/mL, ∗∗P < 0.01).

The above results showed that BBR dose-dependently reduced the production of p-cresol and 4-hydroxyphenylacetic acid, but increased the content of tyrosine, which once again proved that BBR regulated tyrosine–p-cresol synthesis pathway in vitro.

3.5. Correlation analysis between gut-derived uremic toxins and the intestinal flora

Next, since the BBR showed a unique microbial spectrum and regulated the production renal toxins, it is necessary to elucidated the relationship between the gut microbiota and the uremic toxins and the role of BBR plays in the treatment of CKD. We performed a correlation analysis of the detected renal function indices and several uremic toxins with the gut microbiota. The results are shown in Fig. 5A. It was observed that g_Butyricicoccus, g_Proteus, g_Lactococcus, g_Eubacterium_oxidoreducens_group, g_Lachnospiraceae_NK4A136_group, g_Family_XIII_UCG-001, g_Shuttleworthia, g_Anaerovorax, g_Streptococcus, g_Lachnospiraceae_UCG-006, g_Lachnospiraceae_UCG-008, g_Eubacterium_siraeum_group, g_Colidextribacter, and g_Roseburia were negatively correlated with plasma creatinine, plasma TMAO, plasma p-cresol sulfate and plasma indoxyl sulfate. Of note, a large number of butyric acid-producing bacteria were found among these strains, including g_Butyricicoccus⁴⁷, g_Lachnospiraceae_NK4A136_group⁴⁸, g_Lachnospiraceae_UCG-006⁴⁹, g_Lachnospiraceae_UCG-008⁴⁹ and g_Roseburia⁵⁰. And compared with the Model group, BBR increased the abundance of abovementioned genera (Fig. 5F).

Since butyric acid was reported to have renal protective effect on CKD³⁸, and butyric acid-producing bacteria, which were increased by BBR, were negatively correlated with CKD uremic toxins, the butyric acid in the feces samples was detected (Fig. 5E). The fecal butyric acid content in Model group was significantly decreased compared by that in the Control group (∗∗∗P < 0.001). After 4 weeks of interventions, there was a significant uptrend in the fecal content of butyric acid in the BBR group (∗∗∗P < 0.001), L. rhamnosus group (∗∗∗P < 0.001) and Sodium Butyrate group (∗∗∗P < 0.001). However, the fecal content of butyric acid in the Antibiotic group was at a very low level (only 9.6% of the Control group) and showed no obvious changes compared with that in the Model group. Combined with the results in Section 3.3, we can infer that although antibiotics showed a strong inhibitory effect on uremic toxins, antibiotics also inhibited the production of beneficial metabolites such as butyric acid, which may be the reason for the more severe damaged kidney in the Antibiotics group. However, BBR up-regulated butyric acid production in feces as well as down-regulate the renal toxins, which may be related to the renal protective effect of BBR on CKD.

Meanwhile, as shown in Fig. 5A, we also found that g_Clostridium_sensu_stricto_1, g_Adlercreutzia, g_Faecalibaculum, g_Corynebacterium, g_Bifidobacterium, and g_Candidatus_Soleaferrea were positively correlated with plasma creatinine, plasma TMAO, plasma p-cresol sulfate and plasma indoxyl sulfate. And compared with the Model group, BBR decreased the abundance of abovementioned genera (Fig. 5D).

Combined with the results in Section 2.2, g_Clostridium_sensu_stricto_1 was reported to produce p-cresol⁴³, and g_Adlercreutzia was reported to produce phenol⁴⁴, which was corresponded to the effect of BBR in regulating uremic toxins in vivo and in vitro, especially on the tyrosine–p-cresol–p-cresol sulfate pathway. Meanwhile, it was observed that in vivo experiment, g_Clostridium_sensu_stricto_1 and g_Adlercreutzia were significantly enriched in the model group, and BBR could significantly deplete these two genera (Fig. 5B and C). thus, g_Clostridium_sensu_stricto_1 and g_Adlercreutzia could be used for the therapeutic targets for CKD. Since the BBR significant regulated the renal toxins in the tyrosine–p-cresol–p-cresol sulfate pathway, we can infer that g_Clostridium_sensu_stricto_1 may be the main efficacy target for BBR on treatment of CKD.

3.6. BBR reduced g_Clostridium_sensu_stricto_1 abundance in the gut microbiota

In the abovementioned experiments, we have found that BBR could down-regulate the tyrosine–p-cresol–p-cresol sulfate pathway, and BBR could significantly deplete the abundance of g_Clostridium_sensu_stricto_1 in CKD animal feces. Thus, there was a possibility that g_Clostridium_sensu_stricto_1 may be the main efficacy target for BBR on treatment of CKD. In this section, we mainly focused on the role of g_Clostridium_sensu_stricto_1 in the therapeutic effect of BBR on CKD.

First, gene functional abundance concerning the renal toxins pathways in the gut microbiota was evaluated using PICRUSt. PICRUSt⁵¹ is an algorithm for predicting gene functional abundance based on marker gene sequences. PICRUSt1 performs functional prediction only on 16S high-throughput sequencing data. According to the information in the KEGG (Kyoto Encyclopedia of Genes and Genomes) database, the KEGG Orthology (KO) Pathway and EC information can be obtained, and the gene abundance of each functional category can be calculated. The results are shown in Fig. 6A and B. The gene functional abundance of tryptophan catabolism pathway was significantly enriched in the Model group (∗∗P < 0.01), while tryptophan metabolism pathway in the BBR group (∗∗P < 0.01), the L. rhamnosus group (∗∗P < 0.01) and the Sodium Butyrate group (∗∗P < 0.01) was significantly depleted. Meanwhile, the gene functional abundance of tyrosine metabolic pathway was also significantly enriched in the Model group (∗P < 0.05), and the gene functional abundance of tyrosine metabolic pathway in the BBR group (∗P < 0.05), Antibiotic group (∗P < 0.05), L. rhamnosus group (∗P < 0.05) and Sodium Butyrate group (∗P < 0.05) was significantly depleted. Next, we analyzed the KO data and found the key enzymes in tryptophan metabolism and tyrosine metabolism. TnaA (tryptophanase, EC: 4.1.99.1) mainly mediates the production of indole from tryptophan. TyrB (aromatic amino acid aminotransferase, EC: 2.6.1.57) mainly mediates tyrosine to generate 4-hydroxyphenylpyruvate, and the latter generates p-cresol through a series of reactions. Meanwhile, the abundance predictions of TnaA and TyrB in each group were examined (Figs. S7B, 6C). The predicted abundance of TyrB was higher in the fecal samples of the Model group compared with that of the Control group (∗∗P < 0.01), while the BBR group showed significantly reduced TyrB abundance (Fig. 6C), indicating that BBR can affect the tyrosine–p-cresol pathway by the regulation of the TyrB enzyme, which was also consistent with the results in Section 2.3. We next obtained the crystal structure of TyrB (PDB 4WD2) from the PDB database, and then performed virtual molecular docking analysis. Fig. 6D shows the virtual molecular docking between BBR and TyrB. When BBR was contacted with TyrB, the two molecules exhibited strong docking ability with a binding free energy of −18.65 kcal/mol. The two-dimensional schematic diagram of BBR binding to TyrB. There were a large number of hydrophilic bonds (hydrogen bonds and attractive charge bonds) in the active site of TyrB, which may be the main binding force between BBR and TyrB. Meanwhile, the predicted abundance of TnaA was higher in the fecal samples of the Model group compared with that of the Control group (∗P < 0.05, Fig. S7B). However, the BBR group showed no significant effect on TnaA abundance (Fig. S7B), which corresponded with the fact that BBR showed little effect on tryptophan–indole pathway.

Fig. 6.

(A, B) KEGG prediction of the tryptophan breakdown pathway (A) and tyrosine breakdown pathway (B) in the CKD model under BBR intervention (n = 6). (C) KEGG prediction of the abundance of TyrB, a key enzyme in the tyrosine metabolism pathway, in the CKD model under BBR intervention (n = 6). (D) Molecule docking principle between BBR and TyrB and 2D schematic diagram of the binding of BBR to TyrB. (E) Correlation of g_Clostridium_sensu_stricto_1 with p-cresol. (F) Correlation of g_Clostridium_sensu_stricto_1 with p-cresol sulfate. (G) Inhibitory effect of BBR on Clostridium sporogenes (n = 3). (H) Inhibitory effect of BBR on Clostridium perfringens (n = 3). The data are shown as mean ± S.D., ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

Next, we analyzed the correlation of g_Clostridium_sensu_stricto_1 with p-cresol and p-cresol sulfate in CKD model. The results are shown in Fig. 6E and F. g_Clostridium_sensu_stricto_1 had a significant positive correlation with p-cresol in fecal samples (P < 0.0001, r² = 0.44) and p-cresol sulfate in plasma samples (P = 0.0067, r² = 0.19). Then, we investigated whether BBR has an inhibitory effect on the growth of common strains of Clostridium spp. Since few strains of Clostridium spp. that can directly produce p-cresol have been reported, we selected two common Clostridium species (Clostridium sporogenes and Clostridium perfringens) for further study. Clostridium sporogenes is a recognized model strain of Clostridium which has been reported to have the ability to metabolize tyrosine to produce 4-hydroxyphenylpyruvate, 4-hydroxyphenylacetic acid and 4-hydroxyphenylpropionic acid⁵², among which 4-hydroxyphenylacetic acid is the precursor to p-cresol. The abovementioned 2 strains of Clostridium species were incubated with different concentrations of BBR (0, 8, 40, 100 μg/mL), the results are shown in Fig. 6G and H. BBR significantly inhibited the growth of Clostridium sporogenes (Fig. 6G) and C. perfringens (Fig. 6H) and showed a dose-independent manner, which was also consistent with the in vivo results.

4. Discussion

CKD is a serious threat to human health. With the progression of CKD, renal toxins increasingly accumulate in the human body and cannot be excreted by kidneys. Most of these renal toxins are derived from the metabolism of the main nutrients in the diet by the intestinal flora⁵³. Meanwhile, the concept of the gut–kidney axis has been gradually elucidated in recent years, and more correlations between intestinal flora metabolites and changes in renal function have been revealed. Therefore, it is possible to improve renal function in patients with CKD based on the gut microbiota. Strategies for CKD treatment with gut microbiota intervention include fecal microbiota transplantation, prebiotics, probiotics, dietary interventions, or pharmacological interventions to alleviate disease progression²¹. In particular, drug interventions based on the gut microbiota for the treatment of CKD have become a new focus of current research. For example, AST-120 (an activated carbon adsorbent that adsorbs indole, a renal toxin from gut microbiota) has been reported to alleviate CKD by reducing the level of indoxyl sulfate in the blood¹⁷. In this study, BBR has shown a significant therapeutic effect on the progression of CKD. BBR can decrease the BUN and serum creatinine of adenosine-induced CKD rat models and ameliorate the CKD progression. And this mechanism of BBR-ameliorating-CKD involves in the regulation the renal-toxin produce gut bacteria.

Most of the protein-derived gut-produced uremic toxins are the intestinal bacterial degradation products from aromatic amino acids, including p-cresol (mainly generated from tyrosine) and indole (mainly derived from tryptophan), which further enter the blood to generate p-cresol sulfate and indoxyl sulfate phenol under the metabolic effect of the liver. Among the uremic toxins-producer in gut bacteria, indole-producing strains are distributed widely in various phyla, such as Bacteroides (Gram-negative)²⁴. However, p-cresol-producing strains are distributed relatively concentrated, which mainly belong to Enterococcaceae, Clostridiaceae, Staphylococcaceae and Enterobacteriaceae. Notably, these strains are known to coding tyrosine phenol-lyase which mediates the production of tyrosine-derived phenols⁵⁴. In addition, most of the p-cresol-producing strains belongs to Gram-positive bacteria (apart from the strains in Enterobacteriaceae, which is Gram-negative). In this study, we mainly focused on the protective effect of BBR on CKD based on the regulation of tyrosine–p-cresol–p-cresol sulfate pathway and the tryptophan–indole–indoxyl sulfate pathway of gut microbiota. From our data, it was shown that BBR had a more significant effect on the tyrosine–p-cresol-sulfate–p-cresol pathway but a weaker effect on the tryptophan–indole–indoxyl sulfate pathway. This phenomenon can be explained by the antibacterial spectrum of BBR, in which BBR is more sensitive to Gram-positive bacteria and has a weaker effect on the Gram-negative bacteria. In this study, BBR significantly decreased the abundance of Clostridium spp. in CKD models which belongs to Gram-positive bacteria. In addition, BBR can also directly inhibit the growth of Clostridium spp. in vitro, which is consistent with the antibacterial spectrum. However, BBR did not show significant inhibition ability on the growth of indole-producing strains, most of which belongs to Gram-negative bacteria.

Since the antibacterial spectrum of BBR is unique, it was clear that BBR significantly regulated the intestinal flora from the 16S rRNA analysis of the gut microbiota in the animal feces. In the PCA analysis, the clustering of the gut flora in BBR group was in the middle of the Control group, the Model group, and the Antibiotic group, which demonstrated the complex regulation of gut microbiota by BBR, and may be related to the mechanism of treatment of CKD. Meanwhile, it was observed that the application of BBR resulted in a slight decrease in α diversity, which was also a significant manifestation of the bacteriostatic effect of BBR as an antibiotic. However, the bacteriostatic effect of BBR was different from that of most antibiotics. Meanwhile, BBR was found to increase the abundance of butyric acid-producing bacteria, which was consisted with our previous work³⁷.

Butyric acid has been reported to play a role in renal dysfunction in acute kidney injury and CKD by reducing inflammation³⁸. At the same time, butyric acid can directly inhibit histone deacetylase, which regulates gene expression, and the activation and expression of the latter are related to slowing the progression of renal fibrosis, renal injury in lupus nephritis, and podocyte injury in diabetic nephropathy⁵⁵. Butyric acid also acts as an agonist for G protein-coupled receptors, thereby exerting effects on improving kidney function³⁸. There have been studies to explore the direct effect of transplantation of fecal microbiota rich in short chain fatty acids and butyric acid on the treatment of diseases⁵⁶. In our study, direct introduction of BBR-upregulated butyric acid also achieved a therapeutic effect on CKD. And we also found the depletion of the short-chain fatty acid worsened the kidney damage in CKD. In the Antibiotic group, the intestinal flora structure was suffered a completely disturbance. Although the triple antibiotic intervention led to the reduction in p-cresol production by gut microbiota, the antibiotics also significantly inhibited the production of butyric acid at the same time, which may explain the significant deterioration of renal function and the death of animals in the antibiotic group. Therefore, CKD interventions based on the gut microbiota need to address the protection of short-chain fatty acid production while depleting gut-derived uremic toxins.

In addition, antibiotics themselves have been shown to have a variety of side effects, including neurologic toxicity, coagulopathy, nephrotoxicity, hypoglycemia, hematologic toxicity, etc.⁵⁷ It was reported that renal failure patients are often found the reduced gastrointestinal motility or uremic gastroparesis, reflecting the decrease in either intestinal first-pass metabolism or extrusion of drugs (mediated by P-glycoprotein), which increases the bio-availability of several drugs^58,59. Meanwhile, patients with renal failure often have reduced muscle mass and subcutaneous fat, both of which may reduce distribution volume of antibiotics⁶⁰. And Nephrotoxins such as p-indoxyl sulfate occupy serum albumin, further affecting the renal clearance of antibiotics. And renal failure affects the metabolism of antibiotics by inhibiting key enzymatic systems in the liver, intestine and kidney⁶⁰. The above mentioned may further result in the accumulation of antibiotics in the body, which worsen the CKD pathophysiology. Thus, the use of antibiotics in the course of CKD requires special attention.

The use of probiotics in CKD intervention based on the gut flora often produces unexpected effects. In this study, L. rhamnosus was mainly used, which is a well-reported probiotic strain that can regulate host immunity and improve cardiovascular health61, 62, 63. The main mechanism was mostly related to the production of short-chain fatty acids. The use of L. rhamnosus restored the reduction in butyric acid content in feces of the CKD model. Furthermore, L. rhamnosus was reported to regulate TMAO level and CD4⁺ T cell induced-type I inflammation, which may be contributed to mitigate the development of CKD⁶³. Also, L. rhamnosus was reported to reduce the pathogenicity of Candida albicans, during which the energy metabolism of C. albicans was forced to change, resulting in abnormal regulation of virulence-related genes⁶⁴. And the Intestinal wall integrity can be improved after introduction of L. rhamnosus⁶⁵, which may further decrease the leakage of renal toxins. Furthermore, combination of BBR and butyric acid or L. rhamnosus maybe a potential intervention for the treatment of CKD in the future.

The amino acid-derived uremic toxin needs to be metabolized through a variety of enzymes in the intestinal bacteria. Clostridium is the main genus of phenolic producing bacteria⁵². Some of Clostridium can use amino acids as their sole carbon source to generate phenolic. The major tyrosine end products of Clostridium species include phenol, p-cresol and 4-hydroxyphenylacetic acid (precursor of p-cresol)¹⁶. In our study, in addition to showing the direct killing and inhibitory effects of BBR on Clostridium spp., BBR also inhibited the production of p-cresol and p-cresol precursors, such as 4-hydroxyphenylacetic acid, in the incubation system of intestinal bacteria, which suggested that BBR may also affect the metabolic enzymes of intestinal bacteria that mediate the production of amino acid-derived uremic toxin. There are two main ways to produce p-cresol. The first pathway refers to the direct metabolism from tyrosine to p-cresol through tyrosine lyase (ThiH). The second one is via a serious of enzymes, including tyrosine aminotransferase B (TyrB), phenyllactate dehydrogenase (FldH), phenyllactate dehydratase (FldBC) and etc., by which the tyrosine is successively metabolized into 4-hydroxyphenylacetic acid, which is finally metabolized into p-cresol⁵². The p-cresol metabolism of Clostridium is mainly involved in the latter pathway, which corresponded to the fact that BBR inhibited the production of p-cresol and 4-hydroxyphenylacetic acid. With the help of PICRUSt, we found that BBR reduced the relative abundance of the enzymes involved in tryptophan metabolism pathway and tyrosine metabolism pathway in gut microbiota. And the relative abundance of TyrB (mainly mediates tyrosine to generate 4-hydroxyphenylpyruvate) was found to be decreased under the action of BBR. Furthermore, the virtual molecular docking between BBR and TyrB also proved the potential effect of BBR on the metabolic enzymes of intestinal bacteria that mediate the production of amino acid-derived uremic toxin.

In this study, we also found that the plasma TMAO, a renal toxin, was decreased after the intervention of BBR. Recently, it was found that TMAO is also derived from the gut microbiota^28,31. Oral BBR in animals lowered TMAO biosynthesis in intestine through interacting with the enzyme/co-enzyme of choline–trimethylamine lyase and flavin-containing monooxygenase in the gut microbiota. And this action was performed by dihydroberberine (a reductive metabolite of BBR by nitroreductase in the gut microbiota), via a vitamin-like effect down-regulating choline–TMA (trimethylamine)–TMAO production pathway. Moreover, TMA and TMAO in patients with atherosclerosis were decreased by 38% and 29% in feces, and 37 and 35% in plasma, after 4 months on BBR³¹.

Furthermore, there was a downward trend of indole in feces after 4-weeks of BBR treatment in vivo. And the content of indoxyl sulfate in the plasma was also downregulated by BBR. Thus, we can infer that BBR could influence the production or excretion of indoxyl sulfate and may have the influence on the tryptophan–indole–indoxyl sulfate pathway in the gut microbiota since BBR reduced the abundance of the enzymes of tryptophan metabolism in gut microbiota. Further study needed to be done to elucidated the mechanism of BBR on regulating indoxyl sulfate.

Rhizoma Coptidis has been used for thousands of years, and its safety as well as BBR has been proven. Its unique mechanism of action on the intestinal flora provides great possibilities for the new use of old drugs, especially the development of interventions based on the intestinal flora. In summary, based on the strong regulatory effect of BBR on intestinal bacteria, we proposed the possibility of BBR intervention in CKD. BBR group, Sodium Butyrate group for which many studies have reported that BBR can promote the production of short-chain fatty acids³⁷, and probiotic fecal transplant (a potential intervention for CKD, and L. rhamnosus was reported to produce butyric acid) was established to evaluate the renal protective effect on CKD of BBR. BBR can significantly inhibit the production of amino acid-derived uremic toxins, especially the tyrosine–p-cresol-sulfate–p-cresol pathway. And the unique spectrum of BBR decreased the abundance of phenol-producing Clostridium and increased the abundance of short-chain fatty acid-producing bacteria. g_Clostridium_sensu_stricto_1 may be the main efficacy target for BBR on treatment of CKD. The above results suggested that BBR may be a potential therapeutic drug for ameliorating CKD (Fig. 7). The newly discovered mechanism by which BBR acts on the gut–kidney axis can provide new inspiration for the study of difficult-to-absorb natural medicines.

Figure 7.

BBR can ameliorate CKD through the gut–kidney axis by regulating the metabolism of nephrotoxins such as p-cresol by the gut microbiota.

5. Conclusion

In this study, the therapeutic effect of BBR on CKD was investigated, and BBR was revealed to deplete gut-derived uremic toxins and improve the structure of the intestinal flora in CKD. Moreover, BBR can inhibit intestinal bacterial metabolism of tyrosine to produce p-cresol, and the mechanism may be related to the inhibition of Clostridium proliferation by BBR. Based on the above findings, this study first indicated that BBR can ameliorate CKD through the gut–kidney axis, suggesting that BBR has significant potential for the treatment of CKD.

Acknowledgments

This project was supported by the National Key R&D Program of China (No. 2022YFA0806400), the CAMS Innovation Fund for Medical Sciences (CIFMS; Nos. 2022-I2M-JB-011, 2022-I2M-2-002, and 2021-I2M-1-007, China), National Natural Science Foundation of China (Nos. 82173888 and 81973290), and Beijing Key Laboratory of Non-Clinical Drug Metabolism and PK/PD study (Z141102004414062, China). We would like to thank Shimadzu (China) Co., Ltd. for technological support.

Author contributions

Yan Wang, Jiandong Jiang: conceptualization and methodology. Libin Pan, Hang Yu, Yan Wang: original draft preparation, writing, review, and editing. Jie Fu, Mengmeng Bu: investigation. Libin Pan, Hang Yu, Hui Xu, Zhengwei Zhang, Jiachun Hu, Xinyu Yang, Haojian Zhang, Jinyue Lu: validation. Linbin Pan, Hang Yu, Yan Wang: formal analysis. Yan Wang, Jiandong Jiang: supervision.

Conflicts of interest

The authors declare no conflicts of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: