Hepatology International (2022) 99-7ORIGINAL ARTICLEAbnormal bile acid‑microbiota crosstalk promotes the developmentof hepatocellular carcinomaRui Shen1 · Lixin Ke1 · Qiao Li2 · Xi Dang1 · Shunli Shen1 · Jianming Shen1 · Shaoqiang Li1 · Lijian Liang1 ·Baogang Peng1 · Ming Kuang1 · Yi Ma3 · Zhonghan Yang4 · Yunpeng Hua1Received: 11 October 2021 / Accepted: 3 January 2022 / Published online: 24 February 2022 The Author(s) 2022AbstractBackground Gut microbiota and microbe-derived metabolites are involved in the development of HCC. Bile acids (BAs)are the most important gut microbiota-modulated endogenous signaling molecules.Methods We tested serum bile acid levels and gut microbiome compositions in patients with HCC, chemical-inducedHCC mouse models (DEN-HCC mice) and mouse orthotopic implanted liver tumor models with vancomycin treatment(vancomycin-treated mice). Then, we screened an important kind of HCC-related BAs, and verified its effect on the growthof HCC in vivo and in vitro.Results We found that the remarkably decreasing percentages of serum secondary BAs in the total bile acids of patients andDEN-HCC mice, especially, conjugated deoxycholic acids (DCA). The relative abundance of the bile salt hydrolase (BSH)rich bacteria (Bifidobacteriales, Lactobacillales, Bacteroidales, and Clostridiales) was decreased in the feces of patients andDEN-HCC mice. Then, in vancomycin-treated mice, vancomycin treatment induced a reduction in the BSH-rich bacteriaand promoted the growth of liver tumors. Similarly, the percentage of conjugated DCA after vancomycin treatment wassignificantly declined. We used a kind of conjugated DCA, Glyco-deoxycholic acid (GDCA), and found that GDCA remarkably inhibited the growth of HCC in vivo and in vitro.Conclusions We conclude that the remarkably decreasing percentages of serum conjugated DCA may be closely associatedwith HCC, which may be induced by the reducing gut BSH-rich bacteria. The mechanisms may be correlated with conjugatedDCA directly inhibiting the growth and migration of HCC cells.Keywords Primary bile acids · Secondary bile acids · Gut microbiota · Bile salt hydrolase · Glyco-deoxycholic acid ·Bifidobacteriales · Lactobacillales · Bacteroidales · Clostridiales · TumorigenesisRui Shen, Lixin Ke, Qiao Li, Xi Dang, and Shunli Shencontributed equally to the article and should be considered co-firstauthors.* Baogang [email protected]* Ming [email protected]* Yi [email protected]* Zhonghan [email protected]* Yunpeng [email protected] author information available on the last page of the article13Vol:.(1234567890)AbbreviationsHCC Hepatocellular carcinomaBAs Bile acidsUHPLC-MS/MS Ultra-high performance liquidchromatography tandem massspectrometryBSH Bile salt hydrolaseDEN-HCCmice Chemical-induced HCCmouse modelsVancomycin-treated mice Mouse orthotopic implantedliver tumor modelsDEN DiethylnitrosamineLCA Lithocholic acidUDCA Ursodeoxycholic acidHDCA Hyodeoxycholic acidCDCA Chenodeoxycholic acidDCA Deoxycholic acid

Hepatology International (2022) 16:396–411α-MCA α-Muricholic acidβ-MCA β-Muricholic acidCA Cholic acidGUDCA Glycoursodeoxycholic acidGCDCA Glycochenodeoxycholic acidGDCA Glycodeoxycholic acidGCA Glycocholic acidTUDCA Tauroursodeoxycholic acidTHDCA Taurohyodeoxycholic acidTCDCA Taurochenodeoxycholic acidTDCA Taurodeoxycholic acidT-α-MCA Tauro α-Muricholic acidT-β-MCA Tauro β-Muricholic acidTCA Taurocholic acidHSDH 7-Alpha-hydroxysteroiddehydrogenaseFXR Farnesoid X receptorTGR5 G Protein coupled bile acidreceptor 1IntroductionGlobally, hepatocellular carcinoma (HCC) is still the fourthmost common cause of cancer-related deaths, with nearly800,000 new cases annually, despite many recent advancesin the diagnosis and treatment of HCC [1–4]. Therefore, itis urgent to understand the mechanisms of HCC occurrenceand progression and to find novel approaches to predict ortreat HCC. It is well known that hepatocarcinogenesis isclosely related to chronic liver injury resulting from hepaticinflammation, which is mainly attributed to hepatitis B virusinfection. Furthermore, increasing evidence also suggeststhat other intrahepatic and systemic factors likely play significant roles in the process of carcinogenesis and progression,such as the gut microbiota, microbe-derived metabolites andbile acids (BAs) [5–8].The gut microbiota has been thought to play relevant rolesin physiological conditions of human health, such as digestion,vitamin B synthesis, immunomodulation, and the promotion ofangiogenesis and nerve function [9]. Recently, increased interest has also focused on the specific role of the intestinal microbiota in various metabolic diseases, including alcoholic liverdisease, nonalcoholic fatty liver disease, liver cirrhosis, andeven HCC, because the liver is the first organ to be exposed togut-derived toxic factors through the portal vein [10]. A studyshowed that fecal microbial diversity was increased from cirrhosis to early HCC with cirrhosis [11]. The gut microbiotacan promote the development of HCC through the gut-liveraxis in animal models [12, 13], and probiotics can inhibit thegrowth and tumor angiogenesis of HCC by regulating the gutbacteria of mice [14]. Bifidobacterium enhances antitumorimmunity by enhancing the function of dendritic cells and397efficacy of anti-PD-L1 therapy [15]. However, until now, themechanisms by which the gut microbiota promotes HCC havenot yet been clarified.It is well known that the gut microbiota can convert theprimary BAs chenodeoxycholic acid (CDCA) and cholic acid(CA) into the secondary BAs lithocholic acid (LCA), deoxycholic acid (DCA) and ursodeoxycholic acid (UDCA) throughdeconjugation by bile salt hydrolase (BSH) and downstreammodifications by 7-alpha-dehydroxylase or 7-alpha-hydroxysteroid dehydrogenase (HSDH) [16, 17]. As microbe-derivedmetabolites, BAs are involved in the induction of hepatocellular injury, in addition to facilitating lipid absorption [5]. Ingeneral, the hydrophobic bile acids LCA, DCA, and CDCA arecytotoxic, and the hydrophilic bile acid UDCA and its derivative taurourso-deoxycholic acid (TUDCA) are cytoprotective[18]. In addition, BAs also have direct or indirect antimicrobialeffects to modulate the constitution of the microbiota, whichin turn influences the size and composition of the BA pool[19]. Recently, accumulating evidence has demonstrated thatbile acid–microbiota crosstalk plays a crucial role in gastrointestinal carcinogenesis [20]. Yoshimoto et al. also showedthat the gut bacterial metabolite DCA promoted the development of obesity-associated HCC, which was induced with7,12 dimethylbenzanthracene (DMBA) in a mouse model [6].In contrast, the hydrophilic bile acid TUDCA diminished liverovergrowth and tumorigenesis in mice [21]. Xie et al. showedthat the levels of hydrophobic BAs in plasma and liver weresubstantially increased in a nonalcoholic steatohepatitis-hepatocellular carcinoma (NASH-HCC) mouse model, includingDCA, tauro-cholate acid (TCA), tauro-chenodeoxycholate acid(TCDCA), and tauro-lithocholate acid (TLCA). Furthermore,2% cholestyramine feeding significantly prevented HCC development by enhancing the intestinal excretion of hydrophobicBAs [22]. However, the characteristics of bile acids in HCCpatients have not yet been reported, and it remains unclear howthe gut microbiota influences the levels and species of bileacids in patients with HCC. Therefore, we hypothesized thatthe gut microbiota-bile acid axis was closely associated withthe development of HCC and that it was very valuable to probethe mechanisms of and find novel targets for the diagnosis andtreatment of HCC.In this study, we revealed the unique gut microbial spectrum and bile acid spectrum of patients and mice with HCC,explored the correlation between host microbes and bile acids,and further confirmed the protective role of hydrophilic conjugated secondary bile acids on HCC.13

398Hepatology International (2022) 16:396–411Materials and methodsHuman subjectsThe study was approved by the Ethics Review Committee ofthe First Affiliated Hospital of Sun Yat-sen University. Allparticipants were recruited from the Department of HepaticSurgery at Sun Yat-sen University First Affiliated Hospital,which included 20 individuals with hepatitis B virus (HBV)related HCC and 15 healthy controls. Written informed consent was obtained from all participants. None of the individuals were positive for hepatitis C virus (HCV), consumedexcessive alcohol, or received chemotherapy before sampling. The main clinical characteristics of human subjectswere summarized in Table 1. In addition, 11 patients hada single tumor with a median diameter of 6.2 cm (range:3.2–12.5 cm), and 9 patients had multiple tumors with amedian diameter of 5.1 cm (range: 2.8–20.3 cm). Accordingto the BCLC staging system, the number of stage A, B, andC patients was 12 (60%), 5 (25%), and 3 (15%), respectively.All blood samples were set at room temperature for30 min and were then centrifuged at 3000g for 20 minto obtain the serum. Fecal samples were collected on thesame day, snap frozen in dry ice, and stored at 80 untilanalysis.Cell cultureThe H22 mouse HCC cell line, SUN-449 and HepG2 humanHCC cell lines, and LO2 human hepatocyte cell line werepurchased from the Shanghai Cell Collection (Shanghai,Table 1  Patient characteristics for serum bile acid analysisGroupAge(year)Gender(F/M)BMI(kg/m2)AFP (μg/L) 20 20ALTASTAlbuminChild-PaughABHealthy control(N 15)HCC(N 20)p value27.73 10.75F12/M319.8 2.8351.30 10.18F16/M421.2 4.75p 0.001p 0.05p 0.0515 (100%)0 (0%)18.73 5.9619.53 4.1040.21 3.565 (25%)15 (75%)86.85 125.8576.60 58.9237.09 4.05p 0.001p 0.026p 0.001p 0.02415 (100%)0 (0%)19 (95%)1 (5%)––HCC hepatocellular carcinoma; ALT alanine aminotransferase; ASTaspartate aminotransferase; AFP alpha fetoprotein; F female; M male13China) and maintained in Dulbecco’s modified Eagle’smedium (DMEM) with 10% fetal bovine serum (FBS)(Gibco by Life Technologies, Bleiswijk, the Netherlands).Cells were cultured in a cell incubator with 5% CO2 at 37 C.Diethylnitrosamine (DEN) and carbon tetrachloride(CCl4)‑induced HCC C57BL/6J mouse modelMale C57BL/6 mice were purchased from Vital River Laboratories (Beijing, China). C57BL/6J mice (6 weeks old) weredivided into the following two groups (n 10 in each group):(1) HCC group, HCC was induced by the intraperitoneal(i.p.) injection of diethylnitrosamine (DEN) (100 mg/kg)at 6 weeks of age followed by 6–12 biweekly injections ofcarbon tetrachloride (0.5 ml/kg i.p. dissolved in corn oil)unless stated otherwise; (2) control group, i.p. with corn oilas vehicle, double distilled (Ddwater) (25 mg/kg i.p.) wasgiven at day 15 postpartum, and 6–12 weekly injections ofcorn oil (0.5 ml/kg i.p.) [12]. The mice were sacrificed atweek 32, and the size of the liver tumors was measured.Blood serum samples were collected for BA assessment.Stool samples were collected for 16S RNA analysis. Allsamples were stored at 80 until analysis.Orthotopic C57BL/6 mouse hepatic tumor modelwith gut microflora dysbiosisMale C57BL/6 mice (6–8 weeks old) were randomly dividedinto two groups (n 5 each group). Gut bacterial dysbiosis in the vancomycin group was induced using vancomycin (mainly sterilized gram-positive bacteria, 500 mg/l)in drinking water for 4 weeks. Mice in the control groupdrank sterile water directly. H22 mouse HCC cells (1   106in 200 µl DMEM) were injected subcutaneously into theflanks of C57BL/6 mice to generate implanted tumors. After2 weeks, the subcutaneous tumors were resected and dicedinto 1 mm3 cubes, which were then implanted in the left lobeof the liver to make orthotopic transplantation tumors inC57BL/6 mice in the vancomycin group and control group.All of the mice were killed after 2 weeks, and the size of theliver tumors was measured. Blood serum samples were collected for BA assessment. Stool samples were collected for16S RNA analysis. All samples were stored at 80 untilanalysis. All studies were conducted with the approval ofthe Institutional Animal Care and Use Committee (IACUC)of the First Affiliated Hospital of Sun Yat-sen University.Subcutaneous tumor transplant model with GDCAtreatmentEight-week-old, athymic BALB/c nu/nu female mice werepurchased from Gempharmatech Co., Ltd (Nanjing, China).Mice were randomly divided into two groups: GDCA group

Hepatology International (2022) 16:396–411(n 5) and control group (n 5). Nude mice were subcutaneously transplanted with SUN-449 cells (5   106 cellsin 150 μl PBS). On day 10 after cell injection, mice weretreated with GDCA (200 mg/kg every day) or PBS (equalamount) by gastric gavage. Treatments were maintained forone month. Then, mice were sacrificed, and tumor tissueswere removed, weighed, and photographed. Tumor volumeswere determined by measuring length (l) and width (w) andcalculating volume (V 0.5 l w2) at the indicated timepoints.399samples with regards to species complexity. Beta diversityon both weighted and unweighted UniFrac were calculatedby QIIME software (Version 1.7.0). Principal coordinateanalysis (PCoA) was performed with the WGCNA package,stat packages and ggplot2 package in R software (Version2.15.3).Plate cloning formation experimentSerum samples were prepared by precipitation. In addition,50 μl of sample was transferred to an EP tube. After theaddition of 200 μl of extraction solvent (acetonitrile-methanol, 1:1, containing 0.1% formic acid and 312.5 nmol/l internal standard), the samples were vortexed for 30 s, sonicatedfor 10 min in an ice-water bath, incubated at 40 C for 1 hand centrifuged at 12,000 rpm and 4 C for 15 min. Theclear supernatants were transferred to an autosampler vialfor ultra-high performance liquid chromatography tandemmass spectrometry (UHPLC-MS/MS) analysis.SUN-449 and HepG2 human HCC cells and LO2 humanhepatocyte cells were harvested and diluted with DMEM orRPMI 1640 medium. Then, SUN-449 and HepG2 cells wereseeded at 1000 cells/well in a 6-well plate, while LO2 cellswere seeded at 2000 cells/well in a 6-well plate. The treatment groups were treated with GDCA (0.5 mM), and thecontrol groups were treated with PBS. The culture mediumwas changed every 3 days, and the cells were cultured for14 days. Then, the cells were fixed with 4% formaldehydeand stained with crystal violet staining solution for 15 min.Next, the cells were air-dried at room temperature, and theplates were imaged. The number of colonies in each wellwas manually counted. Clone formation rate clone number/number of inoculated cells 100% [23].16S rRNA sequencing and analysisCell counting kit‑8 (CCK‑8)Human and mouse stool sample collection was describedpreviously. Briefly, the sample was divided into five aliquotsof 200 mg and immediately stored at 80 C. Total DNAin feces was isolated using a DNA extraction kit (Tiangen,China). The V3-V4 hypervariable regions of the bacterial16S rRNA gene were amplified with primers 338F (5′-ACT CCT ACG GGA GGC AGC AG-3′) and 806R (5′-GGA CTA CHVGGG T WT CTAAT-3′). PCR products were purifiedusing the GeneJET Gel Extraction Kit (Thermo Scientific).Sequencing libraries were generated using an IlluminaTruSeq DNA PCR-Free Library Preparation Kit (Illumina,USA) following the manufacturer’s recommendations, andindex codes were added. The sequencing was performedby an Illumina HiSeq platform (Novogene BioinformaticsTechnology Co., Ltd.). Sequence analysis was performedby Uparse software (Uparse v7.0.1001, http:// drive5. com/ uparse/). Sequences with 97% similarity were assigned tothe same OTUs. The obtained OTU sequences were groupedat the phylum and order levels.SUN-449 cells, HepG2 HCC cells and LO2 human hepatocytes were treated with GDCA (0.5 mM) or PBS for the cellgrowth test, which was detected by a Cell Counting Kit-8(CCK-8) (Dojindo, Japan) according to the manufacturer’sprotocol. The cells (2000 cells/well) were cultivated in96-well plates for 24, 48, 72, 96, and 120 h, incubated with10 μl of CCK-8 plus 100 μl of DMEM for 2 h, and finallyplaced in a microplate reader (BioTek Synergy2, Winooski,VT, USA) to measure the absorbance at 450 nm [24].Bile acid analysisBacterial diversity and taxonomic analysisAlpha diversity was applied to analyze the complexity ofspecies diversity for a sample. All of these indices reflecting our samples were calculated with QIIME (Version1.7.0) and displayed with R software (Version 2.15.3).Beta diversity analysis was used to evaluate differences inWound haling and transwell assaysThe cell migration of SUN-449 and HepG2 HCC cells wasevaluated using wound healing assays and Transwell assays.The treatment groups were treated with GDCA (0.5 mM),and the control groups were treated with PBS. Wound healing assays were performed as previously described by Lianget al. [25]. Briefly, the cells were seeded at 1   104 cells/well in 6 cm dishes. After the cells formed a monolayer, ascratch wound was made with the tip of a 200-μl pipette tip.Photographs were taken at 0, 8, 16 and 24 h after wounding.Migration distances were measured using ImageJ software(National Institutes of Health, Bethesda, MD, USA).Transwell assays were performed using C orning Transwell polycarbonate membrane cell culture inserts(pore size, 8.0 μm). A 200 μl aliquot of a 20,000 cells/ml suspension (SUN-449 and HepG2 HCC cells) was13

400resuspended in serum-free medium supplemented withGDCA and seeded into the upper chamber of the polycarbonate membrane. Subsequently, 600 μl of mediumcontaining 15% fetal bovine serum (FBS) was added tothe lower well of the migration plate. After incubating for24 h at 37 C, cells in the upper layer of the membranewere scraped, and cells in the lower layer were stainedwith crystal violet staining solution. Then, the cells werephotographed and counted under a phase contrast microscope [26].Cell apoptosisHepatology International (2022) 16:396–411Additional statistical analysisStudent’s t-tests or Mann–Whitney U tests with two-taileddistribution were performed to examine significant differences between the two experimental groups. A p value 0.05was considered to be significant. All analyses were performed with GraphPad Prism 7.0 and SPSS 25.0 software.ResultsSerological bile acids in HCC patients and miceSUN-449 and HepG2 HCC cells were treated with GDCA(0.5 mM) or PBS for cell apoptosis, which was detected24 h after treatment using an Annexin V FITC apoptosis detection kit according to the manufacturer’s protocol(Dojindo, Japan). The cell apoptosis rate was calculatedas follows: cell apoptosis rate (%) (early apoptoticcells advanced apoptotic cells)/total cell number 100%[27].The levels of serum total BAs in the HCC group(5251 1460 nM) were higher than those in the healthycontrols (4626 1015 nM), but this difference didnot reach statistical significance (p 0.736) (Fig. 1a).We found there was a significantly lower ratio of secondary BAs to primary BAs in the serum of the HCCgroup (0.18 0.02) than in the healthy control group(0.52 0.11) (p 0.008) (Fig. 1b). Specifically, the percentages of conjugated (HCC group 11.0% 1.3%, healthyFig. 1  Serum bile acids in patients with HCC and healthy controls. aPlotted in the bar graph are total serum bile acids (MEAN SEM) inthe serum of patients with HCC and healthy controls. b Ratio of secondary bile acids (DCA, GDCA, GUDCA, and UDCA) to Primarybile acids (CA, GCA, GCDCA, and CDCA). c Percent of conjugatedsecondary bile acids (GDCA and GUDCA) in the serum of patientswith HCC and healthy controls. d Percent of unconjugated secondarybile acids (DCA and UDCA) in the serum of patients with HCC andhealthy controls. HCC hepatocellular carcinoma; N healthy controls.*p 0.05, **p 0.01, ***p 0.00113

Hepatology International (2022) 16:396–411control group 17.1% 2.4%) (p 0.028) and unconjugatedsecondary BAs (HCC group 3.8% 0.7%, healthy controlgroup 12.4% 2.8%) (p 0.009) in the total BAs wereboth significantly reduced in the HCC group (Fig. 1c, d).To investigate the percentage of secondary BAs in thetotal BAs was decreased in HCC patients, we generatedmouse liver cancer models through the combinative induction of diethylnitrosamine (DEN) and hepatotoxin carbontetrachloride (CCl4) (Fig. 2a). We found that the level ofserum total BAs was significantly increased in the DENHCC mouse group (11,244 3690 nM) compared to thatin the control mouse group (1556 407 nM) (p 0.001)(Fig. 2b). The ratio of secondary BAs to primary BAs wasalso remarkably reduced in the DEN-HCC mouse group(DEN-HCC mouse group 0.24 0.04, control mouse group0.60 0.14) (p 0.029) (Fig. 2c). Compared with thoseof the normal control mouse group, the percentages ofconjugated (DEN-HCC mouse group 3.6% 0.7%, controlmouse group 10.3% 1.1%) (p 0.001) and unconjugatedsecondary BAs (DEN-HCC mouse group 14.9% 2.5%,Fig. 2  Serum bile acids in patients with chemical-induced mice andnormal control mice. a Liver images from chemical-induced mice andnormal control mice. b Plotted in the bar graph are Mice total serumbile acids (MEAN SEM). c Ratio of secondary bile acids (DCA,TDCA, TUDCA, UDCA, HDCA and THDCA) to primary bile acids(CA, TCA, TCDCA, CDCA, α-MCA, β-MCA, Tα-MCA, Tβ-MCA).401control mouse group 24.1% 3.5%) (p 0.048) of theDEN-HCC mouse group were also decreased (Fig. 2d, e).Characterization of gut microbiome compositionalprofiles in HCC patients and miceIt is well known that the gut microbiota can affect the metabolism of bile acids and change the composition of bile acids[8]. To display microbiome β-diversity, we used principalcoordinate analysis (PCoA) coupled with unweighted UniFrac distances and found a clear separation between HCCpatients and healthy controls (Fig. 3a). Moreover, to displaythe overlaps between two groups, we used a Venn diagramand observed that 1262 of the 2109 OTUs were sharedbetween the 2 groups (Fig. 3b). We found 699 of 1961 OTUswere unique to HCC patients, while only 148 of 1410 OTUswere unique to healthy persons (Fig. 3b).Among the bacterial compositions, the bacterial phylaBacteroidetes, Firmicutes, Proteobacteria, and Actinobacteria were the most abundant bacteria. Compared with healthycontrols, Actinobacteria was significantly decreased in HCCd Percent of conjugated secondary bile acids (TDCA, TUDCA, andTHDCA) in the serum of chemical-induced mice with HCC and normal control mice. e Percent of unconjugated secondary bile acids(DCA, UDCA, and HDCA) in the serum of chemical-induced micewith HCC and normal control mice. T Chemical-induced mice; Nnormal control mice; *p 0.05, **p 0.01, ***p 0.00113

402Hepatology International (2022) 16:396–411Fig. 3  System composition spectrum of gut microbiome in HCC andhealthy controls. Bile acid biosynthesis, transport and metabolism.a Principal Co-ordinates Analysis (PCoA) of bacterial beta diversitybased on the unweighted UniFrac distances. Each node representseach sample; HCC and N subjects are colored in red and blue, respectively. b A Venn diagram displaying the overlaps between groups. cRelative abundance of the top 10 microbiota at the Phylum level inHCC and N. d Relative abundance of the top ten microbiota at theorder level in HCC and N. e, f BSH include species in order. HCChepatocellular carcinoma; N healthy controls; *p 0.05; **p 0.01,***p 0.001patients (p 0.02). In addition, Bacteroidales, Lactobacillales, Selenomonadales, Verrucomicrobiales, and Enterobacteriales were increased in HCC, while Clostridiales, Fusobacteriales, Pasteurellales, and Burkholderiales were decreasedin HCC, but these differences between them were not statistically significant (Fig. 3c). At the order level, probioticBifidobacteriales, belonging to the phylum Actinobacteria,was significantly decreased in HCC patients (p 0.026)13

Hepatology International (2022) 16:396–411403(Fig. 3d). Since the production of secondary bile acidsrequires the participation of BSH enzymes from Bifidobacteriales, Lactobacillales, Bacteroidales, and Clostridiales[18], we checked the abundance of BSH-rich bacteria inHCC patients (76.6% 4.0%) and observed that it was lowerthan that in healthy controls (80.3% 2.5%); however, theirdifference did not reach statistical significance (p 0.462,Fig. 3e).To investigate the HCC-related changes in the gutmicrobiome, we collected fecal samples from mice withDEN-induced HCC and controls. After using PCoA todisplay microbiome β-diversity, we found two distinctenterotypes between the two groups (Fig. 4a). Furthermore, a Venn diagram showed that 437 of the 609 OTUswere shared between the 2 groups. Notably, 109 of 546Fig. 4  System composition spectrum of gut microbiome in DENinduced HCC mice. a Principal Co-ordinates Analysis (PCoA) ofbacterial beta diversity based on the unweighted UniFrac distances.Each node represents each sample. T and N subjects are colored inred and blue, respectively. b Venn diagram of OTUs in two groups.c Relative abundance of the top ten microbiota at the phylum levelin two groups. d Relative abundance of the top microbiota at theorder level in two groups. e, f BSH include species in order. T Chemical-induced mice; N normal control mice; *p 0.05; **p 0.01,***p 0.00113

404Hepatology International (2022) 16:396–411To further confirm our hypothesis that the decrease inBSH-rich bacteria is involved in the development of HCCthrough downregulating the levels of secondary BAs, weused vancomycin to treat C57BL/6 mice and then generated orthotopic implanted liver tumor models. We found thatthe tumor weight in the vancomycin treatment group washigher than that in the control group (p 0.075, Fig. 5a).Furthermore, we used 16S rDNA to analyze the gut microbiota between the two groups and found that the abundanceof BSH-rich bacteria in the vancomycin treatment group(20.0% 3.4%) was significantly lower than that in the control group (93.0% 2.2%) (p 0.009) (Fig. 5b). To furtherobserve the effect of vancomycin treatment on serum bileacids, we found that the concentration of serum total BAsin the vancomycin treatment group (3895 1495 nM) washigher than that in the control group (3026 1079 nM), butthe difference was not statistically significant (p 0.644,Fig. 5c). Interestingly, the ratio of secondary BAs to primary BAs of vancomycin treatment group (0.05 0.01)was significantly lower than that of the control group(0.35 0.11) (p 0.032) (Fig. 5d). However, the percentage of conjugated secondary BAs in vancomycin treatmentgroup (5.1% 1.0%) was significantly lower than that incontrol group (6.3 1.5%) (p 0.522) (Fig. 5e), though thepercentage of unconjugated secondary BAs was significantlyreduced in vancomycin treatment group (0.1% 0.1%) compared with control group (17.8% 6.5%) (p 0.02) andFig. 5f).Fig. 5  The decrease of BSH include species after vancomycin treatment leads to a decrease in secondary BAs, which promotes thedevelopment of HCC. a Images of tumors from each group andtumor weight at the time of sacrifice. b, c BSH include speciesin order. d Plotted in the bar graph are Mice total serum bile acids(mean SEM). e Ratio of secondary bile acids (DCA, TDCA,TUDCA, UDCA, HDCA and THDCA) to Primary bile acids (CA,TCA, TCDCA, CDCA,α-MCA,β-MCA, Tα-MCA and Tβ-MCA). fConjugated secondary bile acids (TDCA, TUDCA, and THDCA) inthe serum of vancomycin treatment group mice and control groupmice. g Unconjugated secondary bile acids (DCA, UDCA, andHDCA) in the serum of vancomycin treatment group mice and control group mice. VAN vancomycin treatment group mice; CON control group mice; *p 0.05; **p 0.01, ***p 0.001OTUs were unique to mice with HCC, while only 63 of500 OTUs were unique to control mice (Fig. 4b).In addition, the bacterial phyla Bacteroidetes, Firmicutes, Proteobacteria, and Actinobacteria were still the mostabundant bacteria in the two groups (Fig. 4c). Comparedwith control mice, phylum unidentified Bacteria were significantly increased in HCC (p 0.004), and Proteobacteriawere significantly decreased in HCC (p 0.001) (Fig. 4c). Atthe order level, Erysipelotrichales (p 0.002), unidentifiedBacteria (p 0.006), and Coriobacteriales (p 0.02), wereremarkably increased in mice with HCC, while Clostridiales (p 0.005), Desulfovibrionales (p 0.001), and Enterobacteriales (p 0.007) were significantly decreased inHCC (Fig. 4d). We also found that the abundance of BSHrich bacteria in DEN-induced HCC mice (70.7% 6.5%)was markedly lower than that in normal control mice(91.0% 0.6%) (p 0.007) (Fig. 4e).Antibiotic vancomycin decreased the abundanceof BSH‑rich bacteria, lowered the levelsof secondary BAs, and induced tumor growth13

Hepatology International (2022) 16:396–411We found that the percentages of GDCA and DCA inHCC patients were decreased significantly (Fig. 6a–d).Similarly, in the DEN-induced HCC mice, the percentagesof TUDCA, TDCA, DCA, and THDCA were all significantly reduced. (Fig. 7a–f). Further analysis found that thepercentages of UDCA, TDCA, DCA, THDCA and HDCAin the vancomycin treatment group were significantlyreduced (Fig. 8a–f). Through the above data, we observeda remarkable reduction of serum conjugated DCA (a kind405of conjugated secondary BAs) in HCC patients, DENinduced HCC mice and vancomycin-treated mice.GDCA inhibits HCC growth in vivo and in vitroAs a kind of common conjugated DCA in humans, GDCAwas used to treat human HCC cell lines, including SUN449 and HepG2 cell lines. We found that GDCA markedlydecreased the clone formation rates of SUN-449 and HepG2cell lines compared with LO2 human hepatocyte cell linesFig. 6  Percent of secondary bileacids in the serum of patientswith HCC and healthy controls.a–d Percent of GUDCA,GDCA, UDCA, and DCA in theserum of patients with HCC andhealthy controls. HCC Hepatocellular Carcinom

The H22 mouse HCC cell line, SUN-449 and HepG2 human HCC cell lines, and LO2 human hepatocyte cell line were purchased from the Shanghai Cell Collection (Shanghai, China) and maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) (Gibco by Life Technologies, Bleiswijk, the Netherlands).