Abstract

   This study was designed to explore the role of gut microbiota and its
   metabolites in the treatment of estrogen-induced cholestasis (EIC) in
   rats with a soybean pulp-rich diet and to clarify the effects of
   daidzein (DAI), a principal active ingredient of soybean pulp. The
   findings demonstrated that the soybean pulp-rich diet could relieve
   cholestasis by decreasing the levels of total bile acids (TBA) and
   alkaline phosphatase and enhancing the bile flow rate. Through gut
   microbiota and metabolomics analyses, it was revealed that this diet
   might alter the abundances of certain bacterial taxa including
   Akkermansia, Bacteroides, and Turicibacter, thus influencing lipid
   metabolism, tryptophan metabolism, and steroid metabolism, which led to
   disparities between the groups fed with and without the soybean
   pulp-rich diet. Moreover, the soybean pulp-rich diet could modulate the
   abundances of Prevotella spp. and Tyzzerella, reducing EIC by
   regulating lipid metabolism and short-chain fatty acids synthesis.
   Notably, DAI treatment significantly alleviated the abnormalities in
   serum TBA, alanine aminotransferase, and aspartate aminotransferase
   levels and mitigated the liver tissue damage in the EIC model. In
   summary, during cholestasis, variations in gut microbiota and
   metabolite profiles occurred. The intervention of soybean pulp affected
   the abundances of bacteria (such as Prevotella spp. and Tyzzerella) and
   regulated lipid metabolism-related pathways. Importantly, DAI was
   identified as a crucial component for the protective effects associated
   with the soybean pulp diet.

   Keywords: Soybean pulp, Daidzein, Cholestasis, Estrogen, Intestinal
   flora, Metabolomics

Graphical abstract

   [41]Image 1
   [42]Open in a new tab

Highlights

     * •
       The study reveals differences in microbial structure, function, and
       metabolites in soybean pulp-rich and pulp-free diets.
     * •
       The study first reveals soybean pulp-rich diets alleviate EIC,
       supported by gut microbiota and metabolomic analyses.
     * •
       The study provides the first evidence that daidzein is a key
       component in soybean pulp that alleviates EIC.

1. Introduction

   Estrogen-induced cholestasis (EIC) is one of the common liver diseases
   in clinics, it is characterized by the destruction of bile acid (BA)
   homeostasis, which leads to the accumulation of BA in the liver and
   causes cholestatic liver injury ([43]Dong et al., 2019; [44]Ming et
   al., 2021). EIC mainly occurs during oral contraceptives, pregnancy, or
   hormone replacement therapy, which can lead to premature delivery,
   amniotic fluid meconium pollution, intrauterine fetal death, and other
   adverse outcomes ([45]Wadie et al., 2021). At present, the main
   pathogenic mechanism of EIC is related to dysfunction of BA
   homeostasis, decreased fluidity of liver cell membrane, oxidative
   stress, inflammatory response, and changes in tight junctions among
   liver cells ([46]Zu et al., 2021). Due to the complexity and
   variability of the etiology and complications of EIC, the existing
   therapeutic drugs (e.g. ursodeoxycholic acid) do not elicit a sensitive
   response in all patients ([47]Hassan et al., 2024). Therefore, it is
   particularly important to develop new treatments for EIC. The
   intestinal microbiota is one of the major regulators of estrogen. The
   microbiota play an important role in the female reproductive endocrine
   system by interacting with estrogens, androgens and other hormones
   ([48]Baker et al., 2017). Imbalance in the composition of the gut
   microbiota can lead to a variety of diseases such as pregnancy
   complications, adverse pregnancy outcomes, and hepatobiliary disease
   ([49]Gao, 2024; [50]Qi et al., 2021). The gut microbiota and their
   metabolites, such as microbially derived short-chain fatty acids (SCFA)
   and biotransformed BA, have been shown to influence physiological
   functions and immune homeostasis in the gut and within the system
   ([51]Rizzetto et al., 2018).

   The increasing consumption of soybean is closely related to the
   potential health benefits. Soybean contains many physiological active
   components, such as soybean phytosterol, soybean isoflavones (like
   daidzin), and soybean polysaccharides ([52]Chatterjee et al., 2018).
   Soybean oil is a vegetable oil derived from soybeans and is high in fat
   but very low in protein. It has been noted that soy lipid products are
   considered to be the greatest contributor to intestinal
   failure-associated liver disease, and that reduction in the use of soy
   lipids has led to improvements in markers of cholestasis and liver
   function ([53]Lim et al., 2019). By reducing the dose of soybean lipid
   emulsion may play an important role in preventing parenteral
   nutrition-associated cholestasis ([54]Maselli et al., 2024). Soybean
   pulp is a high-protein feed left over from the extraction of oil from
   soybean and is low in fat, and high in protein and fiber, therefore
   soybean pulp-rich diet may have the potential to alleviate cholestasis.
   At the same time there are fewer studies on soybean pulp meal and its
   role in cholestasis has not been clarified. Daidzein (DAI) is the
   product of glucosylated daidzin, and it is also the most abundant
   isoflavone in soybeans ([55]Laddha and Kulkarni, 2023). Modern
   pharmacological studies have shown that daidzein can reduce
   inflammation, oxidative stress, osteoporosis, diabetes and
   cardiovascular disease, but there are few studies on hepatobiliary
   diseases.

   Metabolomics is the study of metabolites and metabolism, which enables
   researchers to investigate how genes and the environment interact. As a
   result, metabolomics has emerged as a novel diagnostic tool in clinical
   and biological research, as it directly reflects cellular phenotypes
   and contributes to understanding the downstream effects of the central
   dogma ([56]Li et al., 2023). To maintain metabolic homeostasis, the gut
   microbiota can influence the composition, diversity, and function of
   the microbial community ([57]Mao et al., 2023). Disruption in the
   microbial composition and host metabolism can lead to various diseases.
   Therefore, elucidating the alterations in gut microbiota composition
   and function, as well as their impact on metabolic homeostasis in
   diseases, is of significant importance for the prevention and treatment
   of these conditions.

   In this study, we firstly determined that soybean pulp-rich dietary
   feeding has a protective effect on EIC model rats; and secondly, we
   used 16S rRNA sequencing and non-targeted metabolomics analysis to
   study the changes in the structure of the intestinal flora and
   metabolites in the rats fed soybean pulp-rich diets, so as to elucidate
   the mechanism of the action of soybean pulp-rich diet in the treatment
   of EIC from the point of view of intestinal flora and their
   metabolites; Finally, the effectiveness of DAI, the main active
   ingredient of soybean pulp, was determined in the EIC model, which
   provides a reference basis for the material basis of consuming soybean
   pulp-rich diet to alleviate cholestatic liver.

2. Materials and methods

2.1. Experimental design of intervention effect of soybean pulp-rich diet on
EIC rats

   Due to the complexity of experimental manipulation in pregnant females
   and the high estrogenic effect in vivo, we chose male rats as
   experimental subjects. Healthy male Wistar rats (7–8 weeks old,
   220 ± 20g) were supplied by Beijing SPF Biotechnology Co., Ltd. All
   Wistar rats were raised in the Animal Experiment Center of Fifth
   Medical Center of the PLA General Hospital. Under standard conditions
   (55–60% humidity, 20–22 °C temperature, and 12-hour dark/bright cycle),
   all animals were allowed to drink and eat freely. All animal
   experiments were approved by the Animal Ethics Committee of the Fifth
   Medical Center of the PLA General Hospital (No. IACUC-2018-0024).
   Animal-related experiments were conducted in strict accordance with the
   standard animal research guidelines. All animals were free to eat
   soybean pulp-free chow and soybean pulp-rich chow (Beijing Keao Xieli
   Feed Co., Ltd.) for 7 days for adaption. A total of 32 Wistar rats were
   used in this experiment. They were randomly divided into four groups
   (n = 8): soybean pulp-free chow-fed blank control group (QC) and model
   group (QM), and soybean pulp-rich chow-fed blank control group (PC) and
   model group (PM). The doses of 17α-ethinylestradiol (EE) were selected
   based on the methods described in the previously published studies,
   which have established and validated this particular dosing regimen for
   inducing cholestasis in similar experimental models ([58]Wu et al.,
   2016; [59]Yu et al., 2016). Every morning, EE (5 mg/kg; Sigma-Aldrich,
   MO, USA) was subcutaneously injected in the model groups (injection
   volume, 0.1 ml/100g; powdered EE was fully dissolved in
   1,2-propanediol) for 5 consecutive days to induce cholestasis as
   previous studies. The blank control groups were fed with different
   diets and injected with different volumes of 1,2-propanediol according
   to body weight.

2.2. Experimental design of DAI intervention on EIC rats

   24 healthy Wistar rats were purchased and acclimatized for 7 d (soybean
   pulp-free diet) and randomly divided into a blank control group, a
   model group, a DAI (Solarbio, Beijing, China) 200 mg/kg group, and a
   DAI 600 mg/kg group, with 6 rats in each group. Except for the rats in
   the Control group injected subcutaneously with the corresponding volume
   of 1,2-propanediol, all other groups were injected subcutaneously with
   the corresponding volumes of EE solution (5 mg/kg, injection volume of
   0.1 mL/100 g, once a day) for 5 consecutive days. At the same time of
   subcutaneous injection of EE solution, DAI group was gavaged with
   corresponding concentrations of DAI solution (dissoleve in CMC-Na;
   gavage volume 1 ml/100g, once a day) for 5 consecutive days, and the
   control group and model group were gavaged with corresponding volumes
   of CMC-Na solution for 5 consecutive days.

2.3. Measurement of bile excretion rate

   Total bile flow was calculated using the differential weight method:
   total bile flow (μL) = [EP tube plus bile mass (g) - pre-weighed blank
   EP tube mass (g)] × 1000. Bile density was set at 1 g/mL. Bile flow was
   calculated based on total bile flow and time: bile flow = total bile
   flow (μL)/120 min/100 g body weight.

2.4. Histopathological evaluation of the liver and comparison of liver weight
coefficients

   The collected fresh liver tissues were fixed in a 4% paraformaldehyde
   solution for 48 h. The tissues were then dehydrated, embedded in
   paraffin, and cut into 5 μm thick sections. The sections were stained
   with hematoxylin and eosin (HE) reagents and morphological changes in
   the liver tissue were observed under a light microscope. Liver
   index = liver weight/rat body weight × 100%.

2.5. Determination of biochemical indicators of liver function

   To assess liver function, the levels of TBA and alkaline phosphatase
   (ALP) in the serum were detected by enzymatic kits, respectively
   (Nanjing Jiancheng Institute of Biological Engineering, Nanjing,
   China). Serum levels of alanine aminotransferase (ALT) and aspartate
   aminotransferase (AST) were measured by a serum biochemistry analyzer
   (Mindray, Shenzhen, China).

2.6. 16S rRNA sequencing analysis

   DNA quality and concentration from total faecal bacterial DNA were
   examined by using the PowerSoil DNA Isolation Kit (MoBio Laboratories,
   Carlsbad, CA). 16S rDNA V3-V4 (338–806) forward primers
   (ACTCCTACGGGAGGCAGCAG) and reverse primers (GGACTACHVGGGTWTCTAAT) were
   used to amplify the V3 to V4 regions of the bacterial 16S rRNA gene,
   and PCR amplification was performed using 2xTaq Plus Master Mix and the
   corresponding primers, followed by PCR product electrophoresis (1%
   agarose gel electrophoresis). The total volume of PCR amplification
   reaction was 25 μl. The reaction mixture consisted of 30 ng of DNA
   sample, 1 μL of forward and reverse primers (5 μM), 3 μL of BSA
   (2 ng/μL), and 12.5 μl of 2xTaq Plus Master Mix. The following thermal
   cycling conditions were used: initial denaturation at 95 °C for 5 min,
   denaturation at 95 °C for 45 s for 28 cycles, primer annealing at 55 °C
   for 50 s, extension at 72 °C for 45 s, and a final extension at 72 °C
   for 10 min. 16S rRNA gene amplicons were sequenced on the Illumina
   MiSeq platform. Operational Taxonomic Units (OTUs) were selected by a
   97% sequence similarity threshold using the Qiime (1.8.0) and Vsearch
   (2.7.1) platforms.

2.7. Faecal metabolomics analysis

2.7.1. Sample processing procedure

   50 mg of each faecal sample were weighed, mixed with pre-cooled
   methanol according to the mass to volume ratio of 1:20, and then
   homogenized (60Hz, 10s, 6 times) and shaked vigorously for 2 min. The
   mixture was centrifuged at 14000 rpm, 4 °C for 10 min. 500 μL of
   supernatant was collected for centrifuge again at 14000 rpm, 4 °C for
   10 min, and then 200 μL of supernatant was collected for metabolomic
   analysis.

2.7.2. Chromatographic conditions

   Samples were separated by an ACQUITY UPLC HSS T3 C18 column
   (2.1 mm × 100 mm, 1.8 μm) through a Waters Acquity™ UPLC liquid phase
   system. Mobile phase A was 0.1%-formic acid aqueous solution, and
   mobile phase B was 0.1%-formic acid acetonitrile solution. Gradient
   elution was set as followed: 0–6.0 min, 5%–45% B; 6.0–8.0 min, 45%–75%
   B; 8.0–12.0 min, 75%–85% B; 12.0–12.5 min, 85%–100% B; 12.5–14.0 min,
   100% B; 14.0–14.5 min, 100%–5% B; 14.5–16.0 min, 5% B. Column
   temperature was set at 40 °C, the flow rate was 0.3 mL/min, and the
   injection volume was 3 μL.

2.7.3. Mass spectrometry conditions

   Mass Spectrometrums were analyzed by Xevo G2-XS Q-Tof mass spectrometer
   with Lock-spray interface equipped with ESI ion source. Data
   acquisition was performed in MS^E Continuum mode through MassLynx v.4.1
   mass spectrometry workstation. Accurate mass determination was
   performed using leucine enkephalin (Leucine Enkephalin, ESI^− m/z
   554.2615, ESI ^+ m/z 556.2771) solution as a locking mass solution. The
   MS conditions were set as followed: Capillary voltage, ESI^− 2.5 kV,
   ESI^+ 3.0 kV; ion source temperature, 140 °C; desolvation gas
   temperature, 450 °C; cone hole voltage, 40 V; cone hole gas flow rate,
   50 L/h; desolvation gas flow rate, 800 L/h; collision energy, 10–45 V;
   interval scan time 0.2 s; Mass scan range, m/z 50–1200.

2.7.4. Data extraction and pattern recognition analysis

   MS data were extracted for analysis by using Progenesis QI v.2.4
   software. Data were normalized using MetaboAnalyst v.5.0 and population
   distributions of all samples were assessed using principal component
   analysis (PCA). Differences between the two groups were analyzed by
   orthogonal partial least squares discriminant analysis (OPLS-DA).
   Differential metabolite IDs were determined by statistical analysis
   using the R package “fdrtool” with the screening criteria of VIP >1,
   P < 0.05, and |log[2]FC| > 2. Finally, differential metabolite IDs were
   determined using the HMDB database ([60]https://hmdb.ca/), and
   MetaboAnalyst v.5.0 ([61]http://www.metaboanalyst.ca) were used to
   identify differential metabolite names, potential marker enrichment
   pathways, and biological roles. Finally, Spearman correlation
   coefficients between gut microbiota and metabolic pathways were
   analyzed using the online analysis website
   ([62]https://www.chiplot.online).

2.8. Statistical analysis

   The results were statistically analyzed using the statistical software
   SPSS v.22.0. Data are expressed as the mean ± SD. Significant
   differences were assessed by Student's t-test (comparison between two
   groups) or one-way ANOVA (comparison among multiple groups), and P
   values lower than 0.05 was considered statistically significant. Origin
   v.2022 software was used to perform data visualization.

3. Results

3.1. Protective effect of soybean pulp-rich diet against estrogen-induced
cholestasis

   None of rats died during the entire experiment. After quantitative
   analysis of liver weight, liver index, TBA level, ALP level, and bile
   excretion rate among groups, we found significant differences between
   the QC and QM groups, and between the PC and PM groups (P < 0.01)
   ([63]Fig. 1A–E). Notably, significant differences were found between
   the PM and QM groups in the analysis of serum TBA and ALP. The PM group
   had lower levels of TBA and ALP, which could mean that consuming a
   soybean pulp-rich diet could reduce the severity of EIC. As the most
   intuitive indicator of cholestasis, the rate of bile excretion in the
   QM group was lower than in the PM group ([64]Fig. 1F). The pathological
   results showed that there was no inflammatory infiltration of
   hepatocytes in the liver of rats in the QC group and the PC group, the
   arrangement was regular and orderly, no substantial lesions of
   hepatocytes were seen, and there was no abnormality in the central vein
   and confluence zone. QM rats exhibit extensive nucleus contraction and
   necrosis in liver tissue. The hepatocytes in the PM group were
   irregularly arranged, with obvious inflammatory cell infiltration, bile
   duct hyperplasia, nucleus contraction, and intercellular space
   enlargement, but there was no obvious large-scale hepatocytes necrosis
   ([65]Fig. 1G). Together, these results suggest that a soybean pulp-rich
   diet attenuates a range of pathologic responses to EIC in liver
   compared with a soybean pulp-free diet.

Fig. 1.

   [66]Fig. 1
   [67]Open in a new tab

   Pathophysiology of EIC rats after soybean pulp-rich diet consumption.
   (A) Changes in liver weight of rats in each group (n = 8). Liver weight
   measured as it reflects liver size changes related to diet impact on
   metabolism. (B) Changes in the liver index of rats in each group
   (n = 8). Liver index (liver-to-body weight ratio) shows overall liver
   adaptation to diet. (C) Changes of TBA in the serum of rats in each
   group (n = 8). TBA in serum chosen as biomarker for cholestasis due to
   diet. (D) Changes of ALP in the serum of rats in each group (n = 8).
   ALP measured for detecting liver or biliary diseases caused by diet.
   (E) Changes in bile excretion rate of rats in each group (n = 5). Bile
   excretion rate was determined to reflect biliary function in response
   to diet. (F) The secreted bile in 2h was collected from the bile duct.
   (G) Histopathological changes in the liver of rats in each group
   ( × 100, × 200). Histopathology shows microscopic liver tissue changes
   validating biochemical data.

3.2. Elevated alpha diversity and altered overall microbial composition of
the QM and PM groups of bacteria

   In this study, 16S rRNA sequencing was performed on 24 fecal samples,
   with 6 samples randomly selected from each group. Sequence alignment at
   a 97% similarity threshold identified a total of 4135 OTUs
   ([68]Supplementary Table S1), with the following distributions: 2789
   OTUs in the QC group, 2662 OTUs in the QM group, 2611 OTUs in the PC
   group, and 3195 OTUs in the PM group. Specifically, 182 OTUs were
   identified exclusively in the QC group, 179 OTUs in the QM group, 156
   OTUs in the PC group, and 360 OTUs in the PM group ([69]Fig. 2A). The
   Rank-abundance curve indicated differences in both the richness and
   evenness of microbial communities among the four groups ([70]Fig. 2B).
   Alpha diversity analysis revealed that estrogen intervention
   significantly increased the Simpson and Shannon indices ([71]Fig. 2C
   and D), suggesting that estrogen markedly enhanced microbial richness
   and diversity in the rat gut. Partial least squares discriminant
   analysis (PLS-DA) results ([72]Fig. 2E) demonstrated distinct
   separation among the four groups, indicating notable differences in
   microbial composition. Beta diversity, calculated using Bray-Curtis
   distances and visualized by principal coordinate analysis (PCoA),
   showed significant differences in fecal microbiota distribution between
   QC and QM, PC and PM, PC and QC, as well as PM and QM groups (all
   P < 0.05, PERMANOVA, [73]Fig. 2F–I). These findings indicate that QM
   and PM groups exhibit different diversity and microbial distance
   metrics compared to their respective control groups, while the
   similarities in diversity between PC and QC as well as PM and QM groups
   are driven by their unique microbial profiles.

Fig. 2.

   [74]Fig. 2
   [75]Open in a new tab

   Bacterial diversity of fecal microbiota associated with each group
   (n = 6). (A) Venn graph showing the OTU numbers of gut microbiota from
   four groups. (B) The Rank-abundance curve. (C–D) Alpha-diversity
   (Shannon and Simpson index) in each group. (E) PLS-DA analysis of gut
   microbiota based on four group's OTU data. (F–I) Beta diversity between
   groups calculated by PCoA and PERMANOVA with Bray-Curtis distance. ns,
   P > 0.05; ∗, P < 0.05; ∗∗, P < 0.01.

3.3. Taxonomic signatures of microbiota in PC and QC groups

   The analysis of microbial community composition revealed that the
   dominant bacterial phyla in both groups were Firmicutes, Bacteroidota,
   Verrucomicrobiota, and Actinobacteriota. Notably, the relative
   abundance of Verrucomicrobiota was significantly increased in the QC
   group (P < 0.05) ([76]Fig. 3A and B). LEfSe analysis identified
   differential biomarkers across the phylum to species level ([77]Fig.
   3C), with 2 phyla, 2 classes, 6 orders, 6 families, 12 genera, and 8
   species being potential candidate biomarkers. The QC group was
   characterized by a predominance of 28 taxa, including Verrucomicrobiae,
   gut_metagenome, Akkermansia, Verrucomicrobiales, and Verrucomicrobiota.
   In contrast, the PC group exhibited dominant taxa such as
   Lactobacillus_intestinalis_DSM_6629, Lactobacillus_reuteri,
   Clostridium_sp._ATCC_29733, Eubacterium_siraeum_group, and
   Lachnospiraceae_FCS020_group. Statistical analysis indicated that in
   normal healthy rats, a soybean pulp-rich diet significantly increased
   the relative abundance of UCG-008, Eubacterium_nodatum_group,
   Lachnospiraceae_FCS020_group, and Eubacterium_siraeum_group (P < 0.05),
   while it significantly decreased the relative abundance of Akkermansia,
   Romboutsia, Bacteroides, Turicibacter, Eisenbergiella, and UCG-007
   (P < 0.05) ([78]Fig. 3D).

Fig. 3.

   [79]Fig. 3
   [80]Open in a new tab

   Effects of soybean pulp-rich diet on the overall structure of
   microbiota in normal healthy rats. (A) Average percent of community
   abundance at the phylum-level. (B) Relative community abundances of
   Firmicutes, Bacteroidota, Verrucomicrobiota, and Actinobacteriota
   community abundance ratio. (C) Histogram of LDA represented significant
   difference in abundance of gut bacteria between the PC and QC groups. A
   high LDA score indicates great effect of species abundance on the
   difference between groups. (D) Average abundance percents of biomarker
   community at genus level. ns, P > 0.05; ∗, P < 0.05; ∗∗, P < 0.01.

3.4. Taxonomic signatures of microbiota in PM and QM groups

   From the perspective of overall bacterial community composition, both
   groups exhibited dominant phyla including Firmicutes, Bacteroidota,
   Verrucomicrobiota, and Actinobacteriota, with no significant
   differences in abundance observed ([81]Fig. 4A and B). LEfSe analysis
   was employed to identify differential biomarkers between the groups at
   the phylum to species level ([82]Fig. 4C). The results revealed three
   orders, five families, six genera, and one species as potential
   biomarkers. The QM group showed a notable prevalence of 11 taxa,
   including Oscillospiraceae, Christensenellaceae, Christensenellales,
   NK4A214_group, and Christensenellaceae_R-7_group, while the PM group
   was characterized by dominant taxa such as Prevotellaceae_NK3B31_group,
   Prevotella, Tyzzerella, and gut_metagenome. Statistical analysis
   indicated that, at the genus level, soybean pulp-rich diets
   significantly increased the relative abundance of
   Prevotellaceae_NK3B31_group, Prevotella, and Tyzzerella (P < 0.05),
   while significantly decreasing the relative abundance of NK4A214_group,
   Christensenellaceae_R-7_group, and Monoglobus (P < 0.05) ([83]Fig. 4D).

Fig. 4.

   [84]Fig. 4
   [85]Open in a new tab

   Effects of soybean pulp-rich diet on the overall structure of
   microbiota in EIC rats. (A) Average percent of community abundance at
   the phylum-level. (B) Relative community abundances of Firmicutes,
   Bacteroidota, Verrucomicrobiota, and Actinobacteriota community
   abundance ratio. (C) Histogram of LDA represented significant
   difference in abundance of gut bacteria between the PM and QM groups.
   (D) Average abundance percents of biomarker community at genus level.
   ns, P > 0.05; ∗, P < 0.05; ∗∗, P < 0.01.

3.5. Effects of soybean pulp-rich diet on intestinal metabolic functions

   Functional metabolic pathways of microbiota were predicted using
   PICRUSt2. The analysis revealed a total of 142 pathways in both PC and
   QC groups ([86]Supplementary Table S2), with 11 pathways showing
   significant differences between these two groups. These include Lysine
   degradation, Penicillin and cephalosporin biosynthesis, Amoebiasis,
   Toxoplasmosis, Valine, leucine, and isoleucine degradation,
   Selenocompound metabolism, Streptomycin biosynthesis, Steroid
   biosynthesis, Steroid hormone biosynthesis, Carotenoid biosynthesis,
   and Photosynthesis. Soybean pulp-rich diet significantly increased the
   abundance of Penicillin and cephalosporin biosynthesis, Amoebiasis,
   Toxoplasmosis, Streptomycin biosynthesis, and Photosynthesis pathways
   in normal healthy rats (P < 0.05), while significantly decreasing the
   abundance of Lysine degradation, Valine, leucine, and isoleucine
   degradation, Selenocompound metabolism, Steroid biosynthesis, Steroid
   hormone biosynthesis, and Carotenoid biosynthesis pathways (P < 0.05)
   ([87]Fig. 5A). In the PM and QM groups, 155 pathways were predicted
   ([88]Supplementary Table S3), with 5 pathways showing significant
   differences between these two groups. These pathways include D-Arginine
   and D-ornithine metabolism, Sphingolipid metabolism, Nicotinate and
   nicotinamide metabolism, Flavonoid biosynthesis, and Selenocompound
   metabolism. A soybean pulp-rich diet significantly increased the
   abundance of D-Arginine and D-ornithine metabolism in the intrahepatic
   cholestasis of pregnancy (ICP) model rats (P < 0.05), while
   significantly decreasing the abundance of Sphingolipid metabolism,
   Nicotinate and nicotinamide metabolism, Flavonoid biosynthesis, and
   Selenocompound metabolism pathways (P < 0.05) ([89]Fig. 5B).

Fig. 5.

   [90]Fig. 5
   [91]Open in a new tab

   Differential functional analysis of intestinal microbiota. (A)
   Differential function of intestinal microbiota between PC and QC
   groups. (B) Differential function of intestinal microbiota between PM
   and QM groups. ns, P > 0.05; ∗, P < 0.05; ∗∗, P < 0.01.

3.6. Untargeted metabolomics for faecal metabolic profiling of PC and QC
groups

   To elucidate the potential gut microbiome metabolic phenotypes
   associated with soybean pulp-rich diet feeding, we performed a
   metabolic analysis of the gut contents from the PC and QC groups of
   rats. Results from the principal component analysis (PCA) score plots
   indicated significant differences in the principal components of
   metabolites between these two groups ([92]Fig. 6A). Supervised
   orthogonal partial least squares discriminant analysis (OPLS-DA) models
   also revealed marked differences in metabolite profiles between these
   two groups ([93]Fig. 6B). A total of 50 metabolites, closely associated
   with the gut microbiome, showed significant alterations
   ([94]Supplementary Table S4). Of these, 31 metabolites exhibited a
   notable increase in relative abundance compared to the QC group
   ([95]Fig. 6C). Our metabolomics data revealed that the majority of
   these differential metabolites belong to lipid classes (including
   cholesteryl esters, phospholipids, sphingolipids, and sterol lipids),
   with fewer belonging to flavonoids, secondary bile acids, steroid
   hormones and their metabolites, tryptophan metabolism, vitamins, and
   branched-chain amino acids (BCAAs). Pathway enrichment analysis
   conducted using MetaboAnalyst v.5.0 ([96]https://www.metaboanalyst.ca)
   identified 11 enriched pathways, among which primary bile acid
   biosynthesis, steroid hormone biosynthesis, and steroid biosynthesis
   showed significant differences between the groups (P < 0.05) ([97]Fig.
   6D).

Fig. 6.

   [98]Fig. 6
   [99]Open in a new tab

   Faecal metabolomics analysis of PC group vs QC group. (A) PCA scores of
   untargeted metabolite analysis for both groups of metabolic samples
   under ESI^+ and ESI^−. (B) The plot of OPLS-DA scores for metabolite
   analysis between ESI^+ and ESI^− groups. (C) Stratified cluster
   analysis of faecal metabolites from the PC and QC groups. (D) The
   bubble plot of the KEGG enrichment analysis shows differential
   metabolites.

3.7. Untargeted metabolomics for faecal metabolic profiling of PM and QM
groups

   In order to reveal the relationship between the cholestatic mitigating
   effect of soybean pulp-rich diet and the metabolic phenotype of the gut
   microbiome, we analyzed the gut contents of ECI rats. The PCA score
   plot results showed a clear trend of separation of metabolite principal
   components between the two groups ([100]Fig. 7A). Supervised pattern
   recognition by OPLS-DA also obtained significant changes between these
   two groups in the metabolomics analysis ([101]Fig. 7B). The
   metabolomics analysis revealed significant changes in 48 metabolites
   closely associated with gut microbiota ([102]Supplementary Table S5),
   with 22 of these metabolites showing a notable increase in relative
   abundance compared to the QM group ([103]Fig. 7C). Our metabolomics
   results indicate that most of these differential metabolites belong to
   lipid classes (including cholesteryl esters, sterol lipids,
   phospholipids, prenol lipids, and sphingomyelins), while a smaller
   fraction are related to carbohydrates, flavonoids, SCFAs, steroid
   hormones and their metabolites, and tryptophan metabolism. Pathway
   enrichment analysis conducted using MetaboAnalyst v.5.0 identified 14
   significantly enriched pathways, with primary bile acid biosynthesis,
   steroid hormone biosynthesis, linoleic acid metabolism, steroid
   biosynthesis, and glycerophospholipid metabolism showing significant
   differences between these two groups (P < 0.05) ([104]Fig. 7D).

Fig. 7.

   [105]Fig. 7
   [106]Open in a new tab

   Faecal metabolomic analysis of PM group vs QM group. (A) PCA scores of
   metabolic samples from both groups under ESI^+ and ESI^−. (B) The plot
   of OPLS-DA scores for ESI^+ and ESI^− analyses. (C) Hierarchical
   clustering analysis of faecal metabolites from PM and QM groups. (D)
   Bubble plot of KEGG enrichment analysis showing differential
   metabolites.

3.8. Correlations between gut microbiota and metabolites

   We further explored the functional associations between the disrupted
   gut microbiota and altered circulating microbiome-related metabolites
   through Spearman correlation analysis. The results revealed strong
   correlations between several specific gut bacteria and characteristic
   metabolites.

   In the PC and QC groups ([107]Fig. 8A), Eisenbergiella and Veillonella
   exhibited significant positive correlations with DAI levels. The
   chenodeoxycholic acid glycine conjugate showed a significant positive
   correlation with Turicibacter levels and a significant negative
   correlation with Eubacterium_nodatum_group levels. Glycocholic acid was
   significantly positively correlated with Eubacterium_nodatum_group and
   Eubacterium_siraeum_group levels, and significantly negatively
   correlated with Turicibacter, Eisenbergiella, and Veillonella levels.
   Tryptophan metabolites (3-Methylindole, 1H-Indole-3-acetamide, and
   Indoleacetic acid) were significantly positively correlated with
   Turicibacter, Eisenbergiella, and Veillonella levels, and significantly
   negatively correlated with Eubacterium_nodatum_group levels. L-Leucine
   exhibited significant positive correlations with
   Lachnospiraceae_FCS020_group and Eubacterium_nodatum_group levels, and
   significant negative correlations with Akkermansia, Turicibacter,
   Eisenbergiella, and Veillonella levels. Eubacterium_nodatum_group
   levels were significantly positively correlated with vitamin D
   metabolites (Calcidiol and 24-Hydroxycalcitriol) and significantly
   negatively correlated with Turicibacter and Eisenbergiella levels. The
   androgen metabolites 19-Hydroxyandrost-4-ene-3,17-dione and
   19-Oxoandrost-4-ene-3,17-dione exhibit a significant positive
   correlation with the levels of Akkermansia and Veillonella, while
   showing a significant negative correlation with the levels of
   Eubacterium_nodatum_group.

Fig. 8.

   [108]Fig. 8
   [109]Open in a new tab

   Correlation between gut metabolites and microbial communities based on
   Spearman's rank correlation coefficient. (A) Correlation matrix between
   differential metabolites in the PC and QC groups and genus-level
   microbial biomarkers. (B) Correlation matrix between differential
   metabolites in the PM and QM groups and genus-level microbial
   biomarkers. Colours and size are distributed according to Spearman's
   correlation coefficient. Correlation coefficients with absolute values
   greater than 0.7 were considered to have strong correlations.

   In both the PM and QM groups ([110]Fig. 8B),
   Prevotellaceae_NK3B31_group levels show a significant negative
   correlation with carbohydrate metabolites (D-Maltose, D-Sedoheptulose
   7-phosphate, and L-Fucose). Daidzein correlates positively with
   NK4A214_group levels and negatively with Prevotellaceae_NK3B31_group
   levels. 6-Hydroxyhexanoic acid exhibits a significant positive
   correlation with Prevotellaceae_NK3B31_group levels. Tryptophan
   metabolites (Tryptamin, N-Methyltryptamine, 1H-Indole-3-acetamide, and
   Indoleacetic acid) are significantly negatively correlated with
   Prevotellaceae_NK3B31_group levels. NK4A214_group levels show a
   significant positive correlation with N-Methyltryptamine and
   1H-Indole-3-acetamide. Prevotellaceae_NK3B31_group and Prevotella are
   significantly negatively correlated with androgen metabolites
   (19-Hydroxyandrost-4-ene-3,17-dione, 7α-Hydroxytestosterone, and
   19-Oxoandrost-4-ene-3,17-dione). Major estrogen metabolites
   (16β-Hydroxyestrone and 2-Methoxyestrone) show significant positive
   correlations with NK4A214_group and Monoglobus levels and a significant
   negative correlation with Prevotellaceae_NK3B31_group levels.

3.9. DAI significantly ameliorated injury in EIC rats

   Given that DAI is one of the major components of soybean pulp diet, we
   included DAI as a key bioactive component of soybean pulp diet in this
   part of the study (all groups of rats consumed a soybean pulp-free
   diet). The results of liver weight and liver index showed that compared
   with the blank group, the liver weight and liver index in the model
   group rats were significantly increased (P < 0.01), and after treatment
   with daidzein, the liver index was significantly decreased (P < 0.01)
   ([111]Fig. 9A and B). Serum biochemical analyses revealed that,
   compared to the model group, administration of daidzein significantly
   reduced the levels of TBA, ALT, and AST in rat serum (P < 0.01)
   ([112]Fig. 9C–E). Furthermore, histopathological examination using HE
   staining demonstrated that daidzein treatment notably alleviated
   pathological manifestations such as hepatic tissue necrosis,
   inflammatory infiltration, and pyknosis of cell nuclei, as compared to
   the model group ([113]Fig. 9F).

Fig. 9.

   [114]Fig. 9
   [115]Open in a new tab

   The level of DAI in metabolites and pharmacodynamic results of DAI
   intervention in EIC modeling. (A) Changes in liver weight of rats in
   each group (n = 6). (B) Changes in the liver index of rats in each
   group (n = 6). (C) Changes of TBA in the serum of rats in each group
   (n = 6) (D) Changes of ALT in the serum of rats in each group (n = 6).
   (E) Changes of AST in the serum of rats in each group (n = 6). (F)
   Histopathological changes in the liver of rats in each group
   ( × 100, × 200). ∗∗, P < 0.01; ∗, P < 0.05 compared with control group.
   ^##, P < 0.01; ^#, P < 0.05 compared with the model group.

4. Discussion

   Estrogen is a steroid hormone that is closely related to the
   functioning of the cardiovascular, nervous and reproductive systems
   ([116]Yoh et al., 2023). Of note, excess estrogens and their
   metabolites can lead to cholestasis, and pregnant women, premenopausal
   women, women receiving oral contraceptives, or patients using hormone
   replacement therapy are all susceptible ([117]Bach et al., 2020).
   Estrogen severely impairs the synthesis, transport, and metabolic
   pathways of bile acids, disrupts downstream bile acid homeostasis,
   leads to the influx of bile acids into the bloodstream, and reduces the
   flow rate of bile ([118]Brouwers et al., 2015). Our research findings
   conclusively demonstrate that exposure to EE significantly induces
   intrahepatic cholestasis in rats, regardless of whether they were fed a
   soybean pulp-free diet or a soybean pulp-rich diet. Notably, rats fed
   with a soybean pulp-rich diet exhibited markedly milder cholestasis, as
   evidenced by the levels of TBA, ALP, and bile flow rate, when compared
   the QM groups. We investigated the potential causes of this phenomenon
   through 16S rRNA sequencing and metabolomics analyses and identified
   DAI as one of the key components responsible for the protective effects
   of soybean pulp.

4.1. Soybean pulp-rich diets alter the gut microbiota and its metabolites in
healthy rats

   Diet plays a pivotal role in shaping gut microbiota composition and
   diversity. Although pharmacological assessments revealed no significant
   disparities between the QC and PC groups, comparing the gut microbiota
   and metabolomics of these two groups is crucial to unravel how a
   soybean pulp-rich diet impacted the gut microbiota and its metabolites
   in healthy rats. Notably, relative to the QC group, the PC group
   exhibited remarkable alterations in β-diversity, taxonomic unit
   abundances, and potential metabolic functions. Our study focused on the
   genus-level biomarkers identified via LEfSe analysis. Akkermansia
   muciniphila, constituting 1–3% of the healthy human gut microbiota,
   plays a beneficial role in modulating both intestinal and systemic
   diseases, influencing lipid metabolism through mechanisms such as
   improved insulin sensitivity and reduced body fat accumulation
   ([119]Hasani et al., 2021; [120]Pellegrino et al., 2023). Bile acid
   metabolism in the gut is complex, involving bacteria with bile salt
   hydrolase (BSH) and 7α-dehydroxylase, with Bacteroides converting
   primary conjugated bile acids to deconjugated bile acids, forming
   secondary bile acids ([121]Chen et al., 2024; [122]Jia et al., 2018).
   Turicibacter, with relative abundance up to 20% in rodent small
   intestines and about 0.5% in human fecal microbiota ([123]Mo et al.,
   2021), also contains BSH and affects bile and lipid composition, though
   its relationship with dietary fat and obesity is debated ([124]Lynch et
   al., 2023). Veillonella, a strict anaerobic Gram-negative coccus found
   predominantly in mammalian oral and gut microbiota ([125]Rocha et al.,
   2023), ferments short-chain organic acids to produce SCFAs such as
   acetate and propionate ([126]Mashima et al., 2021). Eubacterium, a core
   genus in the human gut microbiota, is considered beneficial for human
   health similarly to Lactobacillus and Bifidobacterium strains, with
   SCFAs like propionate and butyrate being particularly advantageous
   ([127]Gibson et al., 2017; [128]Louis and Flint, 2017). Eubacterium
   spp. plays a significant role in regulating inflammation, immune
   responses, gut barrier integrity, blood glucose, and cholesterol
   homeostasis ([129]Mukherjee et al., 2020). Tryptophan (Trp) has
   anti-inflammatory effects in mammals, with its metabolites, including
   tryptamine, indole propionic acid (IPA), and other indole derivatives,
   playing crucial roles in immune, metabolic, and neuronal responses
   ([130]Gao et al., 2018; [131]Gupta et al., 2023). Bacteroides bacteria
   degrade Trp to produce indole compounds, and Eubacterium spp. are
   involved in Trp metabolism ([132]Lv et al., 2022; [133]Wang et al.,
   2024). Vitamin D-induced changes in the gut microbiota composition
   favor Bacteroides, which positively modulates cancer immunity
   ([134]Giampazolias et al., 2024). Cholesterol regulates itself and
   serves as the backbone for all steroid hormones and vitamin D analogs
   ([135]Schade et al., 2020). Differential biomarkers primarily regulate
   lipid metabolism, with some affecting Trp, vitamin, and steroid
   metabolism. PICRUSt2 results indicate significant differences in amino
   acid and steroid hormone metabolism. Enrichment pathways related to 50
   metabolites show noticeable differences in primary bile acid
   biosynthesis, steroid hormone biosynthesis, steroid biosynthesis,
   glycerophospholipid metabolism, and Trp metabolism. In conclusion, a
   soybean pulp-rich diet may induce differences between the PC and QC
   groups by modulating the abundance of Akkermansia, Bacteroides,
   Turicibacter, Veillonella, and Eubacterium spp., thereby affecting
   lipid, Trp, and steroid metabolism in the gut microbiota.

4.2. Soybean pulp-rich diets alleviated cholestasis by modulating the
composition and metabolic functions of the gut microbiota

   Soybean is a beneficial food in human nutrition and is widely utilized
   in traditional Eastern cuisines. Epidemiological studies have linked
   soybean consumption to potential benefits in reducing the risk of
   chronic diseases such as obesity, cardiovascular disease, insulin
   resistance/type II diabetes, certain types of cancer, and immune system
   disorders ([136]Marventano et al., 2017; [137]Richter et al., 2015).
   Some studies suggest that soybean consumption may improve
   cardiovascular health by lowering cholesterol levels and blood pressure
   ([138]Rizzo et al., 2022). However, researchers seem to be very
   unenthusiastic about the pharmacological effects of soybean pulp, a
   secondary processed product of soybeans. Our preliminary observations
   revealed differences between soybean pulp-rich and soybean pulp-free
   diets in the EIC model, with soybean pulp-rich diets showing a
   significantly milder cholestasis. Multiple lines of evidence indicate
   that dysbiosis of the gut microbiota is closely associated with the
   occurrence of cholestasis, particularly in patients with ICP ([139]Tang
   et al., 2023; [140]Yu et al., 2023). We hypothesize that this
   phenomenon may be related to changes in the gut microbiota. We also
   focused on several genus-level biomarker differences identified through
   LEfSe analysis. Prevotella spp., a common, abundant, and consistent
   feature of mammalian-associated microbiota, has been found in various
   human body sites including the skin, oral cavity, and gastrointestinal
   tract ([141]Tett et al., 2021). Studies have confirmed that a
   Prevotella spp.-rich gut microbiota can enhance weight loss, lower
   cholesterol levels, and limit the effects of Bifidobacteria
   ([142]Eriksen et al., 2020; [143]Hjorth et al., 2018). Prevotella spp.
   is also a proficient producer of SCFAs propionate derived from
   arabinoxylans and oligosaccharides ([144]Chen et al., 2017). SCFAs can
   directly affect the intestinal environment. As previous studies
   demonstrated, they play a regulatory role in maintaining the integrity
   of the intestinal barrier ([145]Seethaler et al., 2022). A strengthened
   intestinal barrier has the potential to limit the translocation of
   endotoxins, which are closely associated with the inflammatory cascade
   related to cholestasis ([146]Li et al., 2019; [147]Sun et al., 2023).
   We observed a significant positive correlation between Prevotella spp.
   and 6-Hydroxyhexanoic acid. Prevotella spp. may also play a role in
   preterm birth associated with ICP, as Prevotella spp. colonization
   during pregnancy has been linked to full-term delivery, despite being a
   well-known bacterium associated with preterm birth ([148]Park et al.,
   2022). Furthermore, Prevotella is closely related to the metabolism of
   dietary soybean, with studies showing that soybean consumption can
   improve obesity, insulin sensitivity, adipose tissue inflammation, and
   arterial stiffness through beneficial changes in the gut microbiota,
   with Prevotella being a key taxon ([149]Cross et al., 2017). We
   observed a significant increase in the relative abundance of Prevotella
   spp. in the PM group, with Spearman correlation analysis showing
   significant positive correlations between Prevotella spp. and most
   lipid metabolism-related products, and significant negative
   correlations with carbohydrate levels. Tyzzerella primarily functions
   in the fermentation of dietary fibers and other complex carbohydrates
   to produce SCFAs such as acetate, propionate, and butyrate ([150]Xu et
   al., 2022). Acetate, one of the SCFAs produced, has been demonstrated
   to possess anti-inflammatory properties ([151]Deleu et al., 2023). In
   the context of cholestasis, where inflammation is the primary driver of
   liver injury ([152]Ahmed et al., 2021), the increased production of
   acetate by Tyzzerella may potentially suppress the inflammatory
   response. The relative abundance of Tyzzerella was significantly higher
   in the PM group and was notably correlated with lipid metabolism
   products and 6-Hydroxyhexanoic acid levels.

   Through PICRUSt2 analysis, substantial disparities between the PM and
   QM groups emerged in key metabolic pathways, including amino acid
   metabolism, lipid metabolism, and flavonoid biosynthesis. Moreover, an
   in-depth exploration of the 48 differential metabolites associated with
   the gut microbiota disclosed pronounced differences in several enriched
   pathways, such as primary bile acid biosynthesis, steroid hormone
   biosynthesis, glycerophospholipid metabolism, and tryptophan
   metabolism. Collectively, these findings suggest that a diet rich in
   soybean pulp hold the potential to alleviate EIC. This beneficial
   effect may be attributed to its ability to modulate the abundances of
   specific microbiota taxa, notably Prevotella spp. and Tyzzerella. By
   doing so, it can exert a profound influence on lipid metabolism and the
   synthesis of short-chain fatty acids (SCFAs) within the gut microbiota,
   ultimately contributing to the mitigation of EIC.

4.3. DAI significantly alleviates EIC

   Research indicates that the principal bioactive components of soybeans
   include proteins or peptides, isoflavones, saponins, and protease
   inhibitors ([153]Belobrajdic et al., 2023; [154]Kang et al., 2010).
   Among these, DAI, a major isoflavone, has garnered considerable
   attention in recent years. DAI exhibits a broad range of
   pharmacological properties, including potential therapeutic effects in
   cancer, neurodegenerative diseases, cardiovascular diseases, diabetes
   and its complications, osteoporosis, and dermatological conditions
   ([155]Laddha and Kulkarni, 2023; [156]Li et al., 2022). As a primary
   metabolite of isoflavones in soybean pulp, DAI can influence various
   physiological processes through interactions with endogenous ketone
   enzymes ([157]Chadha et al., 2017). Our analysis also revealed
   differences in DAI levels among the fecal metabolites of four groups of
   rats. Administration of different doses of DAI (200 mg/kg and
   600 mg/kg) to the EIC rat model resulted in significant alleviation of
   abnormal serum levels of TBA, ALT, and AST, as well as a reduction in
   the severity of liver tissue damage. We have confirmed that DAI is a
   key component contributing to the protective effects of soybean pulp in
   the EIC model.

5. Conclusion

   In summary, through comprehensive analysis including 16S rRNA
   sequencing and untargeted metabolomics, along with correlations among
   cholestasis parameters, gut microbiota, and metabolites, it is shown
   that a soybean pulp-rich diet relieved estrogen-induced cholestasis by
   modulating gut microbiota composition and metabolic functions.
   Specifically, soybean pulp intervention impacted the abundance of
   bacteria such as Prevotella spp. and Tyzzerella, and regulates lipid
   metabolism-related pathways. Moreover, DAI is identified as a key
   component for the soybean pulp-related protective effects. Although the
   crucial role of gut microbiota and their metabolites is confirmed,
   further research such as fecal microbiota transplantation or
   specific-strain supplementation is required to enhance these findings.

CRediT authorship contribution statement

   Jiawei Wang: Conceptualization, Methodology, Software, Formal analysis,
   Writing – original draft. Qichao Hu: Data curation, Writing – original
   draft. Jianying Wang: Methodology, Formal analysis, Writing – original
   draft. Liwei Lang: Methodology, Visualization, Investigation. Shizhang
   Wei: Resources, Supervision. Haotian Li: Software, Validation. Xiao Ma:
   Visualization, Writing – review & editing, Resources. Yanling Zhao:
   Conceptualization, Funding acquisition, Resources, Supervision, Writing
   – review & editing. Xuelin Zhou: Conceptualization, Resources,
   Supervision, Writing – review & editing.

Declaration of competing interest

   The authors declare no conflicts of interest.

Acknowledgements