Melatonin relieves hepatic lipid dysmetabolism caused by aging via modifying the secondary bile acid pattern of gut microbes

Abstract

It has been reported that aging-generated gut microecosystem may promote host hepatic lipid dysmetabolism through shaping the pattern of secondary bile acids (BAs). Then as an oral drug, melatonin (Mel)-mediated beneficial efforts on the communication between gut microbiota and aging host are still not clearly. Here, we show that aging significantly shapes the pattern of gut microbiota and BAs, whereas Mel treatment reverses these phenotypes (P < 0.05), which is identified to depend on the existence of gut microbiota. Mechanistically, aging-triggered high-level expression of ileac farnesoid X receptor (FXR) is significantly decreased through Mel-mediated inhibition on Campylobacter jejuni (C. jejuni)-induced deconjugation of tauroursodeoxycholic acid (TUDCA) and glycoursodeoxycholic acid (GUDCA) (P < 0.05). The aging-induced high-level of serum taurine chenodeoxycholic acid (TCDCA) activate trimethylamine-N-oxide (TMAO)-triggered activating transcriptional factor 4 (ATF4) signaling via hepatic FXR, which further regulates hepatic BAs metabolism, whereas TUDCA inhibits aging-triggered high-level of hepatic ATF4. Overall, Mel reduces C. jejuni-mediated deconjugation of TUDCA to inhibit aging-triggered high-level expression of hepatic FXR, which further decreases hepatic TMAO production, to relieve hepatic lipid dysmetabolism.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-022-04412-0.

Introduction

Although it has been more than 60 years since the identification of Mel, the studies regarding the effects of Mel on the anti-aging have been brought into focus in the last two decades [1]. Initial studies have observed that when young pineal glands are grafted to the old animals or exogenous melatonin is supplemented, both significantly increase the life span of experimental animals [2]. Similar hypothesis claims that aging is secondary physiology of aged pineal gland, which is positively correlated with low-level expression of Mel [2, 3]. Since the background of these studies is originated from the strong link between circadian rhythm disruption and aging, most of the physiological sites are selected in the nervous system. Recently, increasing investigations contribute to widely revealing the beneficial effects of Mel on peripheral organs/tissues [4]. For instance, numerous studies have demonstrated that oral melatonin alleviates high-fat diet-induced obesity by improving gut microbiota [5–7]. Further, genome sequencing in clinical microbiology and animal experiments have indicated the pattern of gut microbiome in aged individuals to be differ from the young [8, 9], and the aging mouse microbiome has obesogenic characteristics [10]. However, Mel has rarely been reported that this oral drug improves aging-induced lipid dysmetabolism via shaping the structure of gut microbiome.

Aging-induced hepatic steatosis may be related to decreased transport of insulin across the sinusoidal endothelium [11], reduction in autophagic flux [12], or chronic low-level inflammation [13], leading to the buildup of toxic free fatty acids in the liver. Among gut microbiota-mediated regulatory pathways, microbial metabolites are the most watched target for the relative research, such as short chain fatty acids, lipopolysaccharide, and trimethylamine (TMA) [14, 15]. In recent years, Clinical studies have indicated high abundance of TMA-producing strains are positively associated with type 2 diabetes and other chronic metabolic diseases [16–20]. TMAO, a gut microbiota-derived secondary metabolite, is synthesized by TMA under hepatic flavin-containing monooxygenase 3 (FMO3) catalyzing [21]. Compared to the young, aged individuals have shown a significantly higher TMAO concentration in circulation [22, 23]. Considering the high incidence of cardiovascular diseases and metabolic syndrome in the aging population, the hypothesis has been advanced that TMAO may be a key linkage between gut microbiota and aging-induced dysmetabolism. However, relative research has been rarely reported at present.

BAs are synthesized by cholesterol (CHOL) in hepatocytes and then secreted into small intestines, which play a vital role in lipid metabolism. Based on enterohepatic circulation, most BAs are reabsorbed into the liver and recycled, the other are excreted by stool [24, 25]. Besides maintaining the homeostasis of lipid uptake and CHOL metabolism, BAs also bind to corresponding BAs receptors and participate in the regulation of energy metabolism and immunity [26, 27]. In the synthesis of BAs, CHOL is catalyzed by the liver-related enzyme systems, cholesterol 7α-hydroxylase (Cyp7a1) and oxysterol 7-αhydroxylase (Cyp7b1), to produce primary BAs that are free or conjugated with taurine and glycine, such as chenodeoxycholic acid (CDCA) and glycine chenodeoxycholic acid (GCDCA). Secondary BAs are defined as gut microbial metabolites, which are deconjugated or converted from primary BAs by bacterial enzymes bile salt hydrolase (BSH) and 7α-dehydroxylase [28, 29]. Since enterohepatic circulation leads to the reabsorption of intestinal bile acids, secondary BAs may exert potentially modulation in the crosstalk from gut microbiota to host distal organs/tissues. Moreover, as a common observation in human and rodent models, the pattern of BAs in aged individuals is significantly different from the young [30, 31]. Aging-generated gut microecosystem may promote host hepatic dysmetabolism through shaping the pattern of secondary BAs. In this investigation, we discuss the mechanisms of oral Mel inhibiting the production of hepatic TMAO and alleviating liver lipid accumulation by the structure of gut microbiome and the pattern of BAs.

Materials and methods

Animal experiment

Female C57BL/6 J wild-type mice (Eight-week-old) were purchased from the Laboratory Animal Center of the Fourth Military Medical University (Xi’an, China). In an environmentally controlled room (temperature: 25 ± 1 °C; humidity: 55 ± 5%; 12-h light/dark cycle; 07:00–19:00 for light), all animals had ad libitum access to water and food.

Mice were randomly divided into various groups (n = 5) and housed in a signal cage (30 × 20 × 13 cm) and placed a layer of bedding (corn cob for animal bedding, Beijing, BEIJING KEAO XIELI FEED Co., Ltd), which were fed a standard diet (fat provided 30–31% of total energy, Beijing, Boaigang Biotechnology Co. Ltd). The groups include young mice (Control, reared for 2 weeks), aging mice (AP, reared for 12 months), aging mice with Mel supplementation (AP + Mel), and the other treatments. Melatonin treatment was performed by supplying it in drinking fluid (0.4 mg/mL, Sigma-Aldrich, St. Louis, MO, USA) in accordance with previous study by Yin et al. [32]. Melatonin was dissolved in 1% ethanol and diluted in drinking water. The bottled melatonin water was covered by a layer of aluminum foil to prevent light [32, 33]. Moreover, to avoid melatonin degradation by bacterial activities in water, the water by high temperature setrilization was used to prepare melatonin solution and the bottles were daily sterilized by 121 °C. The drinking fluid of all groups was changed daily. According to our previous study, on average, each mouse drank 8.10 ± 0.71 mL/day of melatonin water [34], so the melatonin intake in this study was approximately 108 mg/kg BW/day. To further confirm whether the changes in the bile acid profile are related to melatonin-mediated, we further orally took various BAs (sodium salt form) in the melatonin-treated aged mice for 4 weeks [35]. The supplemental concentration of BAs is 50 mg/kg/day (Sigma-Aldrich, St. Louis, MO, USA) [35], mainly including CDCA, ursodeoxycholic acid (UDCA), GCDCA, TCDCA, GUDCA, TUDCA. Caffeic acid phenethyl ester (CAPE, a BSH inhibitor) administration was performed by intragastric administration as previously described (CAPE: 75 mg/kg/day, Sigma-Aldrich, St. Louis, MO, USA) [35].

In various gene interference studies, aged mice fed for more than 12 months were intraperitoneally injected with the corresponding AAV-shRNA at 10:00 a.m., once a week for 3 weeks. Each injection dose was 1 × 10⁷PFU/mouse. Then each mouse was given corresponding Mel (0.4 mg/mL) or BAs (GUDCA, GCDCA and TCDCA) at 50 mg/kg/day. All adeno-associated virus products were constructed by Gene Pharma (Shanghai, China), and these vectors mainly include AAV-shRNA of AMPKα, Bmal1, Cyp7a1 and Cyp7b1. The target sequences of mice AMPKα, Bmal1, Cyp7a1, Cyp7b1 Lenti-virus were: 5'-CCCATCTTATAGTTCAACCAT-3', 5'-CGCGGAGGAAATCATGGAAAT-3', 5'-CACTTGTTCAAGACCGCACAT-3', 5'-CATACACAATGACCCGGAAAT-3'. Their interference efficiency analysis is shown in Supplementary Figure S1A.

The feed intake and body weight of the mice were monitored daily. Per mice were sacrificed at 8:00 a.m. after the last treatment. Blood samples were collected by orbital bleeding. The ileum, pineal and liver were immediately collected, weighed, and stored in liquid nitrogen until testing. Portion of these tissues were fixed in 4% paraformaldehyde in phosphate buffer and prepared for frozen sectioning which was stained using Oil Red O. The size of cells and stained area were analyzed using a cell profiler software (cellSens, OLYMPUS, Tokyo, Japan).

Oil red O staining

Frozen sections (4–8 um) of liver were dried at room temperature for 15–20 min. 100% isopropanol was used to incubate it for 5 min (avoid bringing water into oil red O), and dyed with 5% oil red O working solution (Beyotime, Shanghai, China). It was washed with 85% isopropanol solution for 3 min, and hematoxylin stained for 1–1.5 min. Finally, it was washed and sealed with glycerin gelatin. The sections were observed under a microscope (Olympus, Tokyo, Japan).

Gut microbiota analysis

Microbial sequencing was provided by novogene (Beijing, China). Total DNA of microorganisms was extracted from all 15 samples sing CTAB/SDS method. The procedure was performed according to the instructions. DNA concentration and purity was monitored on 1% agarose gels. According to the concentration, DNA was diluted to 1 ng/µL using sterile water. 16S rRNA/18SrRNA/ITS genes of distinct regions (16S V4/16S V3/16S V3-V4/16S V4-V5, 18S V4/18S V9, ITS1/ITS2, Arc V4) were amplified used specific primer (e.g. 16S V4: 515F-806R, 18S V4: 528F-706R, 18S V9: 1380F-1510R, et al.) with the barcode. PCR was performed using the following conditions: initial denaturation at 98 °C for 1 min, followed by 30 cycles of denaturation at 98 °C for 10 s, annealing at 50 °C for 30 s, and elongation at 72 °C for 30 s. Finally 72 °C for 5 min. Then, mixture PCR products was purified with Qiagen Gel Extraction Kit (Qiagen, Germany). Sequencing libraries were generated usingTruSeq^® DNA PCR-Free Sample Preparation Kit (Illumina, USA) following manufacturer's recommendations and index codes were added. The library quality was assessed on the Qubit@ 2.0 Fluorometer (Thermo Scientific) and Agilent Bioanalyzer 2100 system. At last, the library was sequenced on an Illumina NovaSeq platform and 250 bp paired-end reads were generated.

Sequences analysis were performed by Uparse software (Uparse v7.0.1001, http://drive5.com/uparse/). Sequences with ≥ 97% similarity were assigned to the same Operational Taxonomic Unit (OUT). Representative sequence for each OTU was screened for further annotation. OTU table was obtained using QIIME (Version 1.7.0) software and further combined with clustering analysis to form a heatmap. Alpha diversity is applied in analyzing complexity of species diversity for a sample through 6 indices, including Observed-species, Chao1, Shannon, Simpson, ACE, Good-coverage. All indices in our samples were calculated with QIIME and displayed with R software (Version 2.15.3). Principal Coordinate Analysis (PCoA) was performed to get principal coordinates and visualize from complex, multidimensional data. PCoA analysis was displayed by WGCNA package, stat packages and ggplot2 package in R software.

BAs analysis

Stool and serum samples were collected from each group and prepared by precipitation. In the supernatants, BAs analysis were as described in previous report by Sun et al. [35]. Briefly, chlorpropamide (Sigma-Aldrich, St. Louis, MO, USA) was added into the samples as an internal standard. The concentrations of BAs were detected by a UPLC/Synapt G2-Si QTOF MS system with an ESI source (Waters Corp., Milford, MA). Chromatographic separation was operated on an Acquity BEH C18 column (100 mm × 2.1 mm i.d., 1.7 μm, 45℃, 0.4 ml/min). The mobile phase included a mixture of 0.1% formic acid in water and 0.1% formic acid in acetonitrile. The gradient elution was applied and MS detection proceeded in negative mode. A mass range from m/z 50 to 850 was acquired. Standards for all BAs were used to identify the different BAs metabolites detected by LC–MS.

Fecal and serum biochemical parameters

Serum samples were tested to detect multiple biochemical parameters using Cobas c-311 coulter chemistry analyzer, including cholesterol (CHOL, Roche, Shanghai, China), triglyceride (TG, Roche, Shanghai, China), high-density lipoprotein (HDL, Roche, Shanghai, China), low-density lipoprotein (LDL Roche, Shanghai, China). The liver tissue was homogenized, dispersed and centrifuged by adding a chloroform/methanol solution 20 times the volume of the liver. According to the manufacturer's instructions, an Elisa kit (BluGene Biotech, Shanghai, China) was used to measure liver total CHOL (TC) and TG levels. Fecal lipid was determined in accordance with previous studies [36]. In brief, fecal samples were collected from all groups when the mice were sacrificed. After being dried at 60 °C to a constant weight, samples were ground to a fine powder and then extracted three times with ethanol 95% at 60 °C. After filtration and evacuation, fecal TG and CHOL were detected using commercial kits (Abcam, Cambridge, UK).

The analysis of serum Mel and FGF15 levels was performed according to the instructions of the mouse melatonin Elisa kit (AMEKO, Shanghai, China) or the sandwich ELISA Kit (LifeSpanBioSciences, Inc., Seattle, WA). Briefly, 40ul of serum samples and 50 ul of standards were added to wells which had been pre-coated with the target specific capture antibody of Mel or FGF15, and then anti-Mel/anti-FGF15 antibody and 50 ul streptavidin-HRP were added respectively, and the plate was covered with a sealing membrane and incubated at 37 °C for 60 min. The plate was washed five times, and the chromogenic solution and stop solution were added in sequence, and the optical density (OD value) of each well was measured at a wavelength of 450 nm using a SpextraMax i3 Multi-mode Microplate Reader (Molecular Devices, USA). Finally, the concentration of each sample is calculated according to the standard curve. The standard curve is shown in Supplementary Figure S1B.

Serum TMA and TMAO were measured referred on previous report from Chen et al. [21]. In brief, d9-TMAO and d9-TMA were used as the internal control, and the serum TMA and TMAO were analyzed by mass spectrometry analysis using a Shimadzu HPLC and AB Sciex 4000-Qtrap hybrid linear ion trap triple quadrupole mass spectrometer. Chromatographic separation was operated on a Primesep 100 column (100 mm × 2.1 mm i.d., 3 μm, 30 ℃, 0.5 ml/min). The mobile phase included a mixture of 0.1% formic acid in water and 0.1% formic acid in acetonitrile. The gradient elution was applied and MS detection proceeded in positive mode.

C. jejuni isolation, identification, culture and in vitro treatment test

Stool samples were collected from mice in each group and placed in 1 ml of phosphate buffered saline (PBS). The isolation of C. jejuni was performed referred previous report from Kim et al. [37]. Briefly, the samples were centrifugated and equal volumes of supernatant were homogenized with Bolton broth (Solarbio, Beijing, China) before streaking onto plain modified charcoalcefoperazone-deoxycholate agar (mCCDA, Oxoid Ltd., Hampshire, UK). Following aerobic incubation at 42 °C for 48 h, 10 random colonies, morphologically resembling C. jejuni, were selected and inoculated onto plain mCCDA to be further selected. Colonies suspected as Campylobacter were transferred to Müller-Hinton agar (MHA, Oxoid Ltd.) and microaerophilically incubated at 42 °C for 48 h. Next, C. jejuni was confirmed by PCR for each colony (Supplementary Table S1). HipO was selected as a target gene for distinguishing C. jejuni from other microbes. Finally, the abundance of C. jejuni was determined through using OD value at 600 nm.

To study the effect of melatonin on C. jejuni, we treated C. jejuni with melatonin concentrations of 1, 2, and 4 mM, respectively. The selection of melatonin concentration referred to the treatment concentration of Yin [5] and the previous aged mice (0.4 mg/mL).

Fecal microbiota transplantation (FMT) and mono-colonization

To deplete original gut microbiota, recipient mice were pre-treated with drinking water dissolved in multiple antibiotics (1 g/L streptomycin, 0.5 g/L ampicillin, 1 g/L gentamicin, and 0.5 g/L vancomycin) for 10 days according to Yin et al. [5, 32]. The antibiotic-containing water was then replaced with regular water, and the microbiota-depleted mice received transplants of the donor microbiota. Recipient mice were colonized by a single oral gavage (100 μl/mouse) with either fecal supernatant (final concentration 5%) from AP group mice or a single colonization of C. jejuni (1 × 10¹⁰ c.f.u./ml).

BSH analysis

The BSH enzyme activity analysis referred to the description of Wang et al. [38]. The cultured bacterial cells or stool sample (50 mg) in 250 μl PBS (pH 7.4) by vortexing for 1 min, then were placed in an ice bath and ultrasonically lysed for 90 s at 30 s intervals. The lysate was centrifuged at 15,000 rpm for 30 min at 4 °C. The protein concentration was measured with BCA protein analysis kit (Pierce, Rockford, IL, USA). The protein concentration of the sample was diluted to 2 mg/ml using PBS working solution. The diluted protein solution was co-cultured with 0.1 mM d4-TCDCA in sodium acetate buffer (pH 5.2) for 20 min at 37 °C. Finally, methanol was added to the mixed solution and centrifuged at 4 °C and 15,000 rpm for 20 min. The supernatant was tested for BSH activity by UPLC-TQMS (Waters, Milford, MA, USA).

Immunoblotting analyses

According to previous studies [34], proteins were extracted from ileum and liver using lysing buffer (Solarbio, Beijing, China). Roughly 30 μg protein were separated by electrophoresis (12% and 5% SDS-PAGE gels) and then transferred onto PVDF nitrocellulose membranes (Millipore, MA, USA) which were further blocked by 5% non-fat milk in Tris-Tween buffered saline buffer for 2 h. After blocking, these membranes were incubated with various primary antibodies, including anti-FXR, anti-FMO3, anti-extracellular regulated protein kinases (ERK), anti-phosphorylated PERK (p-PERK), anti-sterol-regulatory element binding proteins (SREBP1), anti-fatty acid synthase (FASN), anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and HRP-conjugated secondary antibody (Abcam, Cambridge, UK). Finally, proteins were imaged using chemiluminescent peroxidase substrate (Millipore, Massachusetts, USA), and quantified using ChemiDoc XRS system (Bio-Rad, Richmond, CA, USA).

Quantitative real-time PCR

The extraction and reverse transcription of total RNA from ileum and liver were performed (TRIpure Reagent kit and M-MLV reverse transcriptase kit, Takara, Dalian, China) according to previous studies, respectively [34]. Primers were designed in accordance with mouse sequence and passed to Invitrogen (Shanghai, China) to synthesize. All primer names and sequences are summarized in Supplementary Table S2. Quantitative PCR was performed in 25 μl reaction system containing specific primers and SYBR Premix (Vazyme Biotech, Nanjing, China). Amplification was performed in the ABI StepOne plusTM RT-PCR System (Carlsbad, CA). The reaction program is set to 95 °C, 10 s (pre-change period), 95 °C, 5 s, 60 °C, 30 s (PCR reaction period), a total of 40 cycles. The levels of mRNA were normalized in relevance to Gapdh. The relative RNA expressions were analyzed using the method of 2^−ΔΔCt [34].

Metabolomic analyses

25 mg of liver samples from AP and Control groups was weighted to an EP tube respectively, and 500 μl extract solution (acetonitrile: methanol: water = 2: 2: 1, with isotopically labeled internal standard mixture) was added. After 30 s vortex, the samples were homogenized at 35 Hz for 4 min and sonicated for 5 min in ice-water bath. The homogenization and sonication cycle were repeated for three times. Then the samples were incubated for 1 h at − 40 °C and centrifuged at 12,000 rpm for 15 min at 4 °C. The resulting supernatant was transferred to a fresh glass vial for analysis. The quality control sample was prepared by mixing an equal aliquot of the supernatants from all of the samples. LC–MS/MS were performed using an UHPLC system (Thermo Fisher Scientific, Waltham, MA, USA) with a UPLC BEH Amide column (2.1 mm × 100 mm, 1.7 μm) coupled to Q Exactive HFX mass spectrometer (Orbitrap MS, Thermo). The mobile phase consisted of 25 mmol/L ammonium acetate and 25 ammonia hydroxide in water(pH = 9.75)(A) and acetonitrile (B). The analysis was carried with elution gradient as follows: 0–0.5 min, 95% B; 0.5–7.0 min, 95–65% B; 7.0–8.0 min, 65–40% B; 8.0–9.0 min, 40% B; 9.0–9.1 min, 40–95% B; 9.1–12.0 min, 95% B. The column temperature was 30 °C. The auto-sampler temperature was 4 °C, and the injection volume was 3 μl. The QE HFX mass spectrometer was used for its ability to acquire MS/MS spectra on information-dependent acquisition mode in the control of the acquisition software (Xcalibur, Thermo). MetaboAnalyst 4.0 (https://www.metaboanalyst.ca) was used for statistical analysis of metabolomics data. Principal component analysis and partial least squares–discriminant analysis were performed to distinguish differentially enriched metabolic profiles in each group. T tests were used to examine the differences in metabolite concentrations between the Control and AP groups and calculate P values. We defined significance using a P-value threshold of ≤ 0.05. Multiple metabolites were represented in a heatmap. Z-scores were clustered using correlation distance measure from Heatmap function in XLStat 19.5 software (Addinsoft, Paris, France).

Plasmid transfection and dual-luciferase reporter assay

A 2000 bp fragment of mouse SREBP1 promoter was amplified by PCR and fused to the pGL-3 basic vector (Promega, Madison, WI, USA). AML12 cells were transfected after one day culturing in 96-well plate. The built reporter was referred to SREBP1 2000-Luc. Deuterogenic SREBP1 500-Luc and SREBP1 300-Luc reporters were generated from SREBP1 2000-Luc via deletion and relegation, which contained 500 and 300 bp of SREBP1 promoter, respectively. Using the SREBP1 2000-Luc plasmid as a template, mutant SREBP1 reporter plasmids were generated. By co-transfection with SREBP1 promoter vector, AML12 cells were treated together with empty vector (pcDNA3.1) or pc-ATF4-encoding vector (over-expression vector) using X-tremeGENETM transfection reagent (Roche, Basel, Switzerland). Overexpression plasmid vector of ATF4 (pc-ATF4) were constructed in our lab (Forward: 5′-CCCAAGCTTACATGACCGAGATGAGCTTCC-3′, reverse: 5′-CCGCTCGAGTTACGGAACTCTCTTCTTCCC-3′; Hind III: AAGCTT, Xho I: CTCGAG, His Tag: CATCATCACCATCACCAT). The ATF4 gene sequence was cloned into the pcDNA3.1 vector (Invitrogen, Carlsbad, California, USA) by standardized procedures. All plasmid transfection procedures were performed in accordance with the manufacturer’s instructions. Luciferase activity was determined as previous described after transfection for 48 h [39].

Primary cell culture and vectors infection

Primary hepatocytes were performed as described from Zhou et al. through Hepatocyte Isolation System (Worthington Biochemical Corporation, Lakewood, NJ, USA) [40]. Briefly, a drug perfusion system was built on mice liver and multiple veins around it, including vena cava and portal vein. By perfusion pump, plain CMF-HBSS was injected in liver via tube inserted into portal vein. After 7–10 min, enzyme buffer solution was switched to perfuse until full digestion of liver. After stopping the pump, liver was gently placed in culture dish and removed the undigested tissue. Then, hepatocytes were collected via gradient centrifugation.

The control vector was pGLVU6-GFP. For virus vectors study, by co-transfection with ATF4 promoter vector, primary hepatocytes were transfected with control vector or recombinant FXR and PERK adenovirus interference vector (sh-FXR and sh-PERK) for 48 h at the titer of 1 × 10⁹ IFU/ml, and the control vectors were pAd-GFP. All the vectors were constructed by Gene Pharma (Shanghai, China). The target sequences of FXR and PERK Lenti-virus were 5′-CCGTTTAATGAAGTGCGATTA-3′ and 5′-GCCACTTTGAACTTCGGTATA-3′. Their interference efficiency analysis is shown in Supplementary Figure S1C. 24 h after transfection, cells were treated with 50 μM TMAO or 50 μM TCDCA as indicated. TMAO (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in water to prepare a stock solution (5–10 mM) and stored at -20 °C, and freshly diluted in the medium. Cells were harvested 24 h after treatment. The dual luciferase reporter gene detection kit (Promega, Madison, Wisconsin, USA) was used to detect firefly and renilla luciferase activity according to the manufacturer's instructions.

Statistical analysis

Experimental data were analyzed by Student’s t-test and one-way analysis of variance (ANOVA) in SAS v8.0 (SAS Institute, Cary, NC, USA). Statistical significance between two groups was analyzed using the Student’s t-test. Differences among three or more groups were evaluated using one-way ANOVA. Individual means were compared using Fisher’s least significant difference. Mean ± SEM represented each data set and significance in statistical difference was defined as P < 0.05, with P < 0.01 indicating a highly significant difference.

Results

Aging promotes lipid absorption and impairs the melatonin production in mice

According to the method of Camell et al. [41], we constructed mice aging models, and detected the serum P16, a biomarker of aging, by ELISA (Fig. 1A). The results showed that compared to the control (young mice), the level of serum P16 in aging mice were remarkedly elevated (P < 0.05, Fig. 1B). Furthermore, the levels of TG and CHOL in the serum of aging mice were significantly increased compared to the control group; whereas their contents were obviously reduced in stool (P < 0.05), suggesting that aging promotes lipid absorption (Fig. 1C, D). Interestingly, we further found that serum or stool melatonin levels and pineal key rate-limiting enzyme in melatonin biosynthesis (AANAT and ASMT) mRNA levels were significantly lower in aging mice (P < 0.05, Fig. 1E–G), but intestinal AANAT and ASMT levels were not altered (Figure S2A). In addition, melatonin receptors (MT1 and MT2) were also significantly reduced (P < 0.05, Fig. S2E, F). These data indicate that aging promotes lipid absorption and impairs the melatonin production of the pineal gland.

Fig. 1.

Lipid absorption increased and the serum Mel level decreased in aging mice. A The program of animal experiment (n = 5). B Serum P16 protein levels between different groups (n = 5). C, D TG and CHOL levels in serum and stool of aged and young mice (n = 5). E, F Serum and stool Mel levels (n = 5). G The expression of AANAT and ASMT in the pineal (n = 5). Values are means ± SEM. P < 0.05 represents a significant difference between two groups

Aging shapes the structure of gut microbiome and the pattern of BAs

Further, we investigated the composition of fecal microbiota via 16S rRNA sequencing. The plot from the partial least squares-discriminant analysis documented that the gut microbial community in aging mice was substantially isolated from the control (Fig. 2A). Consistently, compared with the control, the Shannon index and Simpson index (α-diversity) of aging mice all changed significantly (P < 0.05, Fig. 2C, D). However, the Observed species have not shown a significantly difference between the AP group and the control (P > 0.05, Fig. 2B). As shown in heat map (Fig. 2E), we compared the abundance of different intestinal flora and found that relative abundances of multiple strains changed significantly with the increase of age.

Fig. 2.

Aging shapes the pattern of gut microbiome and bile acids. A PCoA plot analysis from each sample (n = 5). B–D Shannon, Simpson, and Obeserved Species index (n = 5). E Heatmap of the abundance of representative gut microbiota from each group at the level of genus. The data of each group is displayed as an average value. F Various BAs levels in stool (nmol/g, n = 5). The BAs determined were cholic acid (CA), chenodeoxycholic acid (CDCA), α-muricholic acid (αMCA), β-muricholic acid (βMCA), ursodeoxycholic acid (UDCA), glycoursodeoxycholic acid (GUDCA), tauroursodeoxycholic acid (TUDCA), deoxycholic acid (DCA), Glycine deoxycholic acid (GDCA), taurine deoxycholic acid (TDCA), glycine chenodeoxycholic acid (GCDCA), taurine chenodeoxycholic acid (TCDCA), lithocholic acid (LCA), glycine lithocholic acid (GLCA), hyodeoxycholic acid (HDCA), taurine hyodeoxycholic acid (THDCA), taurocholate acid (TCA), glycocholic acid (GCA). G The levels of BAs in serum of young and aged mice (nM) (n = 5). H ASBT and MRP2 expression in the ileum (n = 5). I Total fecal bile acid level (nmol/g, n = 5). J The ratio of conjugated bile acid to non-conjugated bile acid in stool(n = 5). K The mRNA expression levels of key enzymes in bile acid synthesis include classical pathways (Cyp7a1 and Cyp8a1), alternative pathways (Cyp27a1 and Cyp7b1) and conjugated binding (Baat) (n = 4). Values are means ± SEM. P < 0.05 represents a significant difference between two groups. NS represents no significant difference between groups (p > 0.05)

Moreover, LC–MS/MS analysis indicated aging significantly shaped the pattern of BAs in comparison with the control. Compare to the young, the concentrations of CA, CDCA and its conjugate BAs (GCDCA and TCDCA) in stool were substantially up-regulated (P < 0.05), whereas αMCA, βMCA, UDCA and its conjugate BAs (GUDCA and TUDCA) were obviously reduced (P < 0.05, Fig. 2F). To evaluate the potential for alterations in bile acid uptake for enterohepatic circulation, we further examined the changes in serum BAs and ileal bile acid transport and reabsorption receptors (ASBT and MRP2). Among various BAs, TCDCA in aging mice were dramatically increased compared to the control (P < 0.05), whereas the increased levels of CA and CDCA were not significant (P = 0.05); serum αMCA, βMCA, UDCA, and TUDCA were obviously decreased (P < 0.05, Fig. 2G). Further, the mRNA levels of ASBT and MRP2 in the ileum of aging mice were not significantly different from those of the control (P > 0.05, Fig. 2H). This indicates that aging has no effect on the uptake of BAs. In addition, total BAs in stool from aged mice was significantly increased in comparison with the control, meanwhile the proportion of conjugated BA to unconjugated BA has shown to be obviously decreased in AP group (P < 0.05, Fig. 2I, J). We analyzed the expression of key enzymes for the synthesis of BAs. The result identified that the transcription of hepatic Cyp7a1 and prostacyclin synthase (Cyp8a1) were obviously increased in aging mice, whereas the expressions of sterol 27-hydroxylase (Cyp27a1) and Cyp7b1 mRNA were significantly reduced in AP group (P < 0.05, Fig. 2K). Compared to the control, the transcriptional level of Baat was markedly up-regulated (P < 0.05, Fig. 2K). These findings indicate aging to significantly shape the structure of gut microbiome and the pattern of BAs.

Oral melatonin inhibits ileal FXR signaling through gut microbiota in aged mouse

Considering the close relationship between BAs and lipid uptake, we further detected the effects of oral melatonin for 4 weeks on lipid absorption in aging mice (Fig. 3A). The results showed that oral melatonin had no effect on the feed intake of aging mice, but after 3 weeks of oral administration, the body weight and the ratio of liver weight/body weight of AP + Mel group decreased significantly (P < 0.05, Fig. 3B and Fig. S2B, C). Furthermore, oral melatonin obviously reduced TG and CHOL uptake in aging mice (AP + Mel group) in comparison with the AP group (P < 0.05), meanwhile the analysis of serum apolipoprotein (HDL and LDL) also supported above observations (Fig. 3C–E). Supplementing melatonin (AP + Mel group) significantly increased the serum and stool melatonin level (P < 0.05, Fig. 3F and Fig. S2D) and the expression of MT1/2 in aging mice (Fig. S2E, F).

Fig. 3.

Oral Mel inhibits ileac FXR signaling via gut microbiota in aged mice. A Schematic diagram of Mel treatment in aging mice. B Liver weight/body weight ratio in each group (n = 5). AP: Aging mice. AP + Mel: Aging mice treated with Mel (0.4 mg/ml). Control: Young mice. **Represents P < 0.01 for AP and Control. ^#Represents a significant difference between AP and AP + Mel (P < 0.05), and ^##represents P < 0.01 for AP and AP + Mel. C, D Fecal and serum levels of TG and CHOL in mice with various treatments (n = 5). E Serum levels of HDL and LDL in mice with various treatments (n = 5). F Serum Mel levels in each group (n = 5). G The mRNA expression of FXR in the ileum of aging mice under different concentrations of Mel (n = 4). *P < 0.05 compared with the other groups. H The mRNA expression of Shp and Fgf15 in the ileum of mice with different treatments (n = 5). I Serum FGF15 levels in each group (n = 5). J The effect of antibiotics and Mel treatment on the expression of FXR, Shp and Fgf15 (n = 4). AP: Aged mice. AP + Antibiotics: Aged mice treated with multiple antibiotics (1 g/L streptomycin, 0.5 g/L ampicillin, 1 g/L gentamicin, and 0.5 g/L vancomycin). AP + Antibiotics + Mel: Aged mice treated with multiple antibiotics (1 g/L streptomycin, 0.5 g/L ampicillin, 1 g/L gentamicin, and 0.5 g/L vancomycin) and Mel(0.4 mg/ml). K The effect of FMT and Mel treatment on the expression of FXR, Shp and Fgf15 (n = 4). Control: Young mice. Control + Antibiotics: Young mice treated with treated with multiple antibiotics. Control + Antibiotics + APT: young mice transplanted with intestinal flora of aged mice. L The effect of Mel treatment after interference with AMPKα or Bmal1 on the expression of FXR, Shp and Fgf15 in aged mice (n = 4). Values are means ± SEM. Same markup represents no significant difference between two groups (P > 0.05), different markup represents a significant difference (P < 0.05) compared with the other groups. P < 0.05 represents a significant difference between two groups

Previous studies document that that melatonin alleviates neuroinflammation in depressed rats via gut FXR signaling [42]. Further other studies showed that the activation of intestinal FXR signaling and the high-level expression of downstream Shp and Fgf15 contribute to triggering hepatic lipid dysmetabolism [43, 44]. Thus, we evaluated the transcription level of FXR in the ileum of aging mice and found that oral Mel inhibited the ileal FXR mRNA level in a dose-dependent manner (Fig. 3G). Moreover, Mel supplementation markedly decreased the transcriptional level of Shp and Fgf15, and serum levels of Fgf15 also validated the results of mRNA (P < 0.05, Fig. 3H, I). To investigate whether oral Mel-mediated inhibition of ileal FXR signaling dependent on gut microbiota, we sought to observe these phenotypes in the same administrations under the depletion of gut microbiota by antibiotic mixtures. As shown in Fig. 3J, antibiotics-mediated depletion of gut microbiota significantly down-regulated ileal FXR signaling in comparison with AP group (P < 0.05), and Mel supplementation did not exhibit further inhibition of ileal FXR signal (P > 0.05). In turn, Mel showed further inhibition of FXR signaling after microbial supplementation in aging mice (P < 0.05, Fig. 4C). Ileal FXR signaling was obviously activated in young mice under fecal microbiota transplantation from aged mice, even higher than the control (young mice) (P < 0.05, Fig. 3K), indicating that Mel-mediated ileum Inhibition of FXR signaling is dependent on the gut microbiota.

Fig. 4.

Mel inhibits ileac FXR signaling through decreasing C. jejuni-mediated deconjugation of GUDCA and TUDCA in aging mice. A The effects of Mel and BAs treatment on the expression of FXR, Shp and Fgf15 in aged mice (n = 4). B The effect of Mel and BAs treatment on the expression of FXR, Shp and Fgf15 in aged mice after interference with Cyp7a1 and Cyp7b1 (n = 4). C The effect of Mel and BAs treatment on the expression of FXR, Shp and Fgf15 in aged mice colonized with C. jejuni (n = 4). D Relative mRNA levels in young mice with various treatments (n = 4). Control + APT: young mice transplanted with intestinal flora of aged mice. Control + C. jejuni: young mice colonized with C. jejuni. Control + TCDCA: young mice treated with TCDCA. E The growth curves of C. jejuni under melatonin treatment with various dosages in vitro (1 mM, 2 mM, 4 mM) (OD600nm, n = 6), right panel: partial enlargement. F BSH activity of C. jejuni treated with Mel (4 mM, n = 5). G, H The C. jejuni-mediated unconjugation of GUDCA and TUDCA under various treatments (n = 5). I Relative mRNA levels (n = 4). CAPE: A BSH inhibitor with a treatment concentration of 75 mg/kg/day. Values are means ± SEM. Same markup represents no significant difference between two groups (P > 0.05), different markup represents a significant difference (P < 0.05) compared with the other groups. P < 0.05 represents a significant difference between two groups

Other studies reported that AMPKα signaling is involved in lipid uptake of enterocytes and Mel-mediated Bmal1 targeted activation in aging mice [45, 46]. In view of this, we further investigated whether AMPKα and Bmal1 participate in oral Mel-mediated ileal FXR inhibition via the gut microbiota in aging mice. The results showed that the mRNA expressions of FXR, Shp and Fgf15 in the ileum of aging mice were not affected by sh-AMPKα or sh-Bmal1 treatments (P > 0.05), whereas Mel supplementation contributed to inhibiting ileal FXR signaling and the transcriptions of downstream Shp and Fgf15 (P < 0.05), indicating that oral Mel-mediated inhibition of ileal FXR signaling is independent of AMPKα or Bmal1 signaling (Fig. 3L).

Mel inhibits ileal FXR signaling through decreasing C. jejuni-mediated deconjugation of GUDCA and TUDCA in aged mice

To identify whether the alternat pattern of BAs is responsible for oral melatonin-mediated inhibition of ileal FXR, we continued the treatment of different BAs on aging mice that were orally administered melatonin. Compared to the AP group, CDCA and its conjugate BAs (GCDCA and TCDCA) supplementations exhibited a statistically increase in the transcription of FXR and downstream Shp and Fgf15 (P < 0.05, Fig. 4A). Moreover, mice with co-treatment of sh-Cyp7a1 and sh-Cyp7b1 have shown significant inhibitions on the transcriptional levels of FXR, Shp, and Fgf15 in comparison with AP group (P < 0.05, Fig. 4B). Notably, Mel supplementation did not further decrease the transcriptional levels of FXR, Shp, and Fgf15 on the mice with co-treatment of sh-Cyp7a1 and sh-Cyp7b1 (P > 0.05, Fig. 4B). Meanwhile, GUDCA treatment contributed to further inhibiting ileal FXR and downstream Shp and Fgf15, whereas GCDCA supplementation obviously elevated their mRNA expressions (P < 0.05, Fig. 4B).

16S rRNA sequencing showed that aging induced a significant increase in the number of C. Jejuni (Fig. 2E), but its mechanism of action is not clear. Other studies have reported that C. jejuni is one of the prevalent causes of bacterial-derived diarrheal illness, and that infection induces complications such as metabolic disease, intestinal inflammation, and colorectal cancer (Figure S3) [47, 48]. Its pathogenicity is closely related to host BAs (Figure S3A). To investigate the effects of C. jejuni on dysmetabolism of aging mice, we preformed FMT and mono-colonization on aged and young mice, respectively. The results showed that C. jejuni colonization statistically up-regulated the transcriptional levels of FXR, Shp, and Fgf15 on aging mice, whereas Mel supplementation reversed this risk (P < 0.05, Fig. 4C). Furthermore, GCDCA and TCDCA supplementations further increased the mRNA expression of FXR, Shp, and Fgf15 in aging mice with co-treatment of C. jejuni and Mel (P < 0.05, Fig. 4C). Compared to the control (young mice), recipient young mice with gut microbiota from aged mice have also shown significant up-regulations of FXR, Shp, and Fgf15 transcription (P < 0.05), meanwhile C. jejuni colonization and TCDCA treatment also markedly induced these phenotypes (P < 0.05, Fig. 4D).

To further explore the effect of Mel on C. jejuni, we measured its growth curve with no or different concentrations of Mel in vitro. The results showed that Mel supplementation inhibited the proliferation of C. jejuni in a dose-dependent manner after 4 h of cultivation (Fig. 4E). Furthermore, compared with no addition of Mel (Control), Mel treatment (4 mM) of C. jejuni in vitro significantly reduced the its BSH activity (P < 0.05), suggesting that Mel enable C. jejuni to reduce the ability of digesting BAs (Fig. 4F). I further used Mel or CAPE (a BSH inhibitor) treatment to analyze the deconjugation of GUDCA and TUDCA in aged mice colonized by C. jejuni. As shown in Fig. 4G and H, we found that Mel treatment obviously inhibited C. jejuni-mediated deconjugation of GUDCA and TUDCA (P < 0.05). Notably, co-treatment of CAPE and Mel rarely reduced C. jejuni-mediated deconjugation of GUDCA and TUDCA in comparison with Mel treatment (P > 0.05, Fig. 4G, H). Finally, we identified CAPE to inhibit ileal FXR signaling and the mRNA expression of downstream Shp and Fgf15 in vivo (Fig. 4I). Overall, these data indicate that Mel inhibited the activation of ileal FXR signaling by decreasing C. jejuni-mediated deconjugation of GUDCA and TUDCA in aging mice.

Aging-induced high-level of TCDCA promotes hepatic TMAO production

For further study, we need to identify the downstream targets of TCDCA and TUDCA-mediated FXR signaling in regulating liver lipid metabolism in aging mice. We compared the hepatic metabolomes from aging mice to the control. Cluster analysis showed that there was a significant difference in the composition of metabolites between the two groups (Fig. 5A). From all alternat metabolites, we found hepatic TMAO to be significantly increased in aging mice (Fig. 5B). To verify the difference of hepatic TMAO between the two groups, we tested the concentration of TMAO in liver and serum, as well as TMA, the precursor of TMAO. Notably, serum and hepatic TMAO from aging mice was significantly higher than the young (P < 0.05, Fig. 5C, E), but the level of serum TMA has not exhibited a statistical difference (P > 0.05, Fig. 5D). We further detected the relative protein of hepatic FMO3 due to high-level of TMAO in serum and liver from aging mice. Compared with the control (young mice), the liver FMO3 protein level of aging mice was significantly increased (P < 0.05, Fig. 5F). Meanwhile, we also observed that the relative protein of hepatic FXR were up-regulated in the AP group (Fig. 5F).

Fig. 5.

Aging-mediated high-level of TCDCA promotes hepatic TMAO production in mice. A PCoA plot analysis from each sample (n = 3). B Heatmap of the abundance of representative hepatic metabolites from each group (n = 3). C TMAO levels in the liver of young and aged mice (n = 5). D, E Serum TMA and TMAO levels in young and aged mice (n = 5). F Relative protein levels (n = 5). G The effect of interference with FXR or TCDCA treatment on serum TMAO in aging mice (μmol/L, n = 5). H The effect of interference with FXR or TCDCA treatment on the relative protein levels of aging mice (n = 5). I Serum TMAO levels of young mice (n = 5). J Relative protein levels of young mice (n = 5). K Serum TMA and TMAO levels in young mice after TCDCA or TCDCA and TUDCA supplemented synergistically (μmol/L, n = 5). Values are means ± SEM. P < 0.05 represents a significant difference between two groups

Previous studies reported that FMO3 expression was induced by dietary BAs through a mechanism involving FXR [49]. To further investigate whether aging-induced high-level of serum TCDCA and hepatic FXR are involved in modulating TMAO production, we tested the level of hepatic TMAO in the mice with co-treatment of sh-FXR vector and TCDCA. The results demonstrated that TCDCA supplementation further promote the high-level expression of hepatic TMAO in aging mice (P < 0.05), whereas supplementing TCDCA after interfering with the expression of FXR will not have a significant effect (P > 0.05, Fig. 5G). Meanwhile, the protein expression of FMO3 in liver was also consistent with the changes of TMAO (Fig. 5H). Similarly, we also observed these same phenotypes and trends in young mice (Fig. 5I, J). Based on young mice, the co-treatment of TCDCA and TUDCA did not decrease the level of serum TMA but markedly reduced the TMAO level in the blood circulation compared with TCDCA treatment (P < 0.05, Fig. 5K), indicating that TUDCA relieved the production of TMAO. Taken together, these findings demonstrated that aging-induced high-level of TCDCA promotes hepatic TMAO production through activating hepatic FXR/FMO3 signaling.

TCDCA-activated ATF4 promotes SREBP1 expression at the transcriptional level

Previous study has identified TMAO to specifically bind to PERK, which activates ATF4 to induce endoplasmic reticulum (ER) stress in hepatocytes [21]. Here, we further isolated and cultured primary hepatocytes in vitro to study their relationship. TMAO treatment indeed activated PERK/ATF4 signaling by luciferase reporter assay (Fig. 6A, B). It was worth noting that TCDCA supplementation also significantly activated PERK/ATF4 signaling, although it was lower than TMAO treatment (P < 0.05, Fig. 6A). Moreover, the co-treatment of TCDCA and TMAO did not show a higher level of activation of PERK/ATF4 signaling than TMAO treatment (P > 0.05), supporting that TCDCA regulates hepatocytes via TMAO pathway (Fig. 6A). To further identify whether FXR and PERK are responsible for TCDCA-mediated regulation via TMAO pathway in hepatocytes, we performed same administrations under sh-FXR or sh-PERK treatment. The results showed that TCDCA supplementation did not obviously activate PERK/ATF4 signaling under the interference of FXR (P > 0.05), whereas TMAO treatment was not affected by sh-FXR administration (Fig. 6C). Furthermore, both TCDCA and TMAO supplementations could not trigger PERK/ATF4 signaling in hepatocytes under sh-PERK treatment (Fig. 6D). These data suggest that TCDCA activates PERK/ATF4 via FXR signaling which increases TMAO production in hepatocytes.

Fig. 6.

TCDCA-activated ATF4 promotes the expression of SREBP1 at the transcriptional level. A Luciferase reporter assay under various treatments (n = 6). The concentration of primary hepatocytes treated by TMAO was 50uM, and TCDCA was also 50 uM. Vehicle: Control vector. sh-FXR: Interfering vector with the FXR. sh-PERK: Interfering vector with the PERK. B–D Relative protein levels under the same treatment as (A) (n = 5). E The binding sites of ATF4 in the promoter region of SREBP1. F Luciferase activity was corrected for Renilla luciferase activity and normalized to the control activity (n = 6). Values are means ± SEM. Same markup represents no significant difference between two groups (P > 0.05), different markup represents a significant difference (P < 0.05) compared with the other groups. *P < 0.05 compared with the other groups

Since SREBP1 is a core transcription factor that promotes hepatic lipid synthesis, we further investigated whether ATF4 affect the expression of SREBP1 at the transcriptional level. On the SREBP1 promoter of mouse, ATF4 has been predicted to own two potential binding domains (Fig. 6E). Using the luciferase assay, two binding sites were identified as the promotor positions of SREBP1 (Fig. 6F). Taken together, these data suggest that TCDCA activates PERK/ATF4 via FXR signaling which further increases the expression of SREBP1 at the transcriptional level.

TUDCA supplementation inhibits aging-triggered hepatic lipid dysmetabolism

As shown in result 4, Mel reduced the BSH activity of C. jejuni and inhibited its growth. We further assessed the microbial profile of aging mice after oral administration of melatonin. As expected, 16S rRNA sequencing indicated that oral Mel inhibited the abundance of C. jejuni in aging mice (Fig. 7A). Consistently, oral melatonin significantly inhibited fecal BSH activity in aging mice (P < 0.01, Fig. 7B). Compared to AP group, the content of CA, CDCA, GCDCA, and TCDCA were significantly reduced in stool from aging mice with oral Mel treatment, whereas the concentration of αMCA, βMCA, UDCA, GUDCA and TUDCA were obviously increased (P < 0.05, Figure S4A). Moreover, the analysis results of BAs in serum are consistent with those in stool (Figure S4B). Since TUDCA has been identified as an inhibitor of FXR by previous studies [35, 50], we further explored the effect of TUDCA on ATF4 expression in young mice based on the TCDCA-induced aging model. Using TUDCA or Mel treatment, we confirmed that TUDCA supplementation significantly decreased TCDCA-induced high-level expression of ATF4 (P < 0.05), whereas the Mel treatment did not exhibit an obvious inhibition on the basis of supplementing TCDCA (P > 0.05, Fig. 7C). Furthermore, Supplementing TUDCA alone also reduced luciferase activity of ATF4 (P < 0.05, Fig. 7C). Compared with the untreated group, TCDCA treatment significantly increased the liver weight/body weight ratio, Hepatic TG and total CHOL levels of young mice, while TUDCA treatment significantly reversed this trend (P < 0.05, Fig. 7D–F). However, with TCDCA supplementation, oral melatonin cannot improve the hepatic lipid dysmetabolism of TCDCA-induced aging. We further analyzed the protein levels of key regulators of fatty acid and triglyceride biosynthesis (SREBP1) and its downstream targets (FASN), and the results showed that supplementation with TCDCA significantly increased SREBP1 and FASN protein levels, but the results of TUDCA treatment were opposite (P < 0.05, Fig. 7G). Hepatic slide with Oil Red O staining reflected same tendency among each group (Fig. 7H), supporting that TUDCA inhibits aging-triggered liver lipid accumulation, and oral Mel does not directly affect the expression of ATF4 to regulate liver lipid metabolism.

Fig. 7.

TUDCA supplementation inhibits aging-triggered hepatic lipid dysmetabolism. A Heatmap of the abundance of representative gut microbiota from each group at the level of genus (n = 5). The data of each group is displayed as an average value. B Analysis of BSH activity in stool of each group (n = 5). C Luciferase reporter assay under various treatments (n = 5). D, E The liver weight/body weight ratio (D), liver TG (E) and TC (F) levels of each treatment group (n = 5). G Relative protein levels under the same treatment as (C) (n = 5). H Representative Oil Red O staining (Red dot) of hepatic slides from mice with various treatments, bar: μm (× 200, n = 5). Values are means ± SEM. Same markup represents no significant difference between two groups (P > 0.05), different markup represents a significant difference (P < 0.05) compared with the other groups. P < 0.05 represents a significant difference between two groups

Discussion

Apart from CA and CDCA, primary BAs pool in rodent also includes α/βMCA which are different from human [51]. Although previous studies documented multiple BAs to be ligands of FXR, their binding to FXR exerts a physiological difference [44, 52, 53]. In mammals, FXR is a ligand-regulated nuclear receptor that modulates BAs biosynthesis, transport, secretion and numerous lipometabolism, which is widely distributed in the intestines and liver [54]. Among various BAs, CA, CDCA and its conjugate BAs (GCDCA and TCDCA) are well-established FXR agonists, whereas α/βMCA are considered as inhibitors of FXR [52, 53]. In a recent study, Sun et al. identified GUDCA and TUDCA to be inhibitors of FXR [35]. Also, FXR-FGF15 axis has been affirmed to involve in gut microbiota-mediated de novo synthesis of hepatic bile acids, and the inhibition of FXR signaling contributes to improving obesity-induced metabolic dysfunction in mice [24, 43]. Meanwhile, oral UDCA has been observed to alleviate hepatic disorder via inhibiting intestinal FXR signaling and decreasing the serum level of FGF15 [55]. Overall, we hypothesized that the alternation of gut microbiome-triggered shaped pattern of BAs may be essential for dysmetabolism in aged individual, and intestinal FXR signaling and downstream Shp and Fgf15 can be regarded as bio-markers representing the level of systematical/hepatic dysmetabolism. Herein, we found aging significantly shaped the pattern of microbial community and BAs in comparison with the young. It is noteworthy that the concentrations of CDCA, TCDCA and GCDCA, which were considered as FXR agonists, have been detected to be significantly increased in ileum of aged mice, whereas inhibitory BAs of FXR (UDCA, GUDCA and TUDCA) were decreased. This phenotype is consistent with the subsequent findings on the activation of intestinal FXR signal and the high transcriptional expression of downstream Shp and Fgf15, meanwhile the high-level of serum TG, CHOL and TMAO have also been tested, which was consistent with previous research [35, 56, 57]. Notably, TUDCA supplementation substantially reversed aging/TCDCA treatment-induced these risks, indicating that FXR signaling is definitely play a vital role in regulating dysmetabolism in aged mice.

In addition to promoting the decomposition and absorption of fat from diet, BAs also exerts on antibacterial activity in intestines [58–60]. However, BAs-mediated broad-spectrum antimicrobial effect is not limited to pathogenic bacteria [58]. Instead, some strains evolve the ability to resist this bacteriostasis during long-term symbiotic condition [58]. C. jejuni is identified as a prevalent Gram-negative pathogen causing gastroenteritis [61]. As a food-borne bacterial pathogen, C. jejuni widely colonizes a variety of domestic animals with asymptomatic infection, so it is difficult to be prevented by screening healthy livestock and poultry [62]. Previous research has documented that C. jejuni developed resistance in the environment of low BAs concentrations via secreting outer membrane vesicles, resulting in enhancing self-colonization [63]. Besides, C. jejuni is able to keep proliferation under deoxycholate treatment, its differentiated system of adaptation to bile acid stresses enables bacterial cell to maintain activity in the culture of lowest inhibitory concentration of DCA and CDCA [64]. Moreover, it has shown to sensitive to GUDCA-mediated bacteriostatic action [64]. Thus, we hypothesized that aging-triggered alteration in the pattern of BAs may be associated with high abundance of C. jejuni in aged mice. In this study, we confirmed that Campylobacter jejun contributed to reversely activating ileal FXR signaling via BSH-mediated deconjugation of TUDCA and GUDCA, which was supported by previous research.

In liver, FMO3 has been identified to be positively correlated with insulin resistance, cardiovascular diseases, and other metabolic disorders [65]. Moreover, it has been reported to be regulated by upstream hepatic FXR signaling [49]. Both of clinical research and rodent model of insulin resistance have reported hepatic FMO3 to be significantly elevated, whereas FMO3 knockout contributed to preventing mice from hypertension, hyperlipidemia and metabolic syndrome [21, 65]. Further investigations have demonstrated that FMO3 is responsible for processing circulating TMA into TMAO and then trigger insulin resistance [66–68]. Based on these studies, we hypothesized that aging-induced pattern of BAs may increase TMAO production through triggering hepatic FXR/FMO3 regulatory pathway. Thus, we performed a comparison of the effects of various BAs on TMAO production via hepatic FXR/FMO3 signaling. Herein, the results documented aging-induced elevated TCDCA to contribute to activating hepatic FXR/FMO3 signaling and then promoting TMAO secretion. In turn, TUDCA supplementation reversed this risk. Oral Mel also inhibited hepatic TMAO production via decreasing C. jejun-mediated deconjugation of TUDCA.

In hepatocytes, since ER stress has a close association with lipid dysmetabolism and TMAO specifically binds to PERK, which activates ATF4 to induce ER stress in hepatocytes [21], we thought that high expression of TMAO may activate PERK/ATF4 to regulate BAs metabolism. In view of the key role of SREBP1 in modulating the metabolic homeostasis of BAs and lipid [69, 70], we detected whether the activation of PERK/ATF4 contributed to regulating the expression of SREBP1. Our results determined that aging-induced high-level of TCDCA contributed to activating PERK/ATF4 pathway through FXR signaling, which further promoted SREBP1 expression at the transcriptional level. Notably, Mel did not directly inhibit SREBP1 expression via hepatic FXR signal.

However, some limitations also remain in this study. First, aged pattern of gut microbiota-mediated shaping in the structure of BAs has been only focused, without considering the causal relationship between aged pattern of gut microbiome and aging. Therefore, following studies need a long-term of experimental cycle to deeply discuss the relationship between gut microbiota and host aging. Second, in this study we only regarded TMAO as a representative indicator, without assessing other parallelly dysmetabolic phenotypes, such insulin resistance and glucose intolerance. Finally, all the results of this study occurred on rodents, and the translational aspects of the findings needs to be greatly enhanced by comparison to human fecal samples. But clinically, oral Mel-mediated alterations in the gut microbiome provide data support for a potential anti-aging therapy—alleviating hepatic lipid metabolism disturbances by shaping BAs patterns.

In summary, the present results suggest that (i) secondary BAs is a key pathway for bio-information in the crosstalk between gut microbiota and distal organ; (ii) TUDCA and TCDCA are important regulator for hepatic TMAO production; (iii) aging-induced the high-level of TCDCA activates TMAO-triggered PERK/ATF4 signaling via hepatic FXR, which further promotes SREBP1 transcription and regulates hepatic BA metabolism; (iv) oral melatonin reduces C. jejuni-mediated deconjugation of TUDCA to inhibit aging-triggered high-level expression of hepatic FXR/FMO3, which further decreases hepatic TMAO production and lipid dysmetabolism (Fig. 8).

Fig. 8.

Oral melatonin inhibited hepatic TMAO production through inhibiting C. jejuni-mediated degradation of TUDCA, resulting the inhibition of FXR/FMO3 signaling and dysmetabolism of bile acid production in aged mice

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Every author contributed to this work. DW: conceptualization, supervision, methodology, writing—review and editing. YL: investigation, resources, writing—original draft preparation. MC: visualization, data curation. CL: validation, format modification. QW: formal analysis. CS: writing—review and editing, funding acquisition.

Funding

This work was supported by the grants from the National Key Research and Development Program of China (2021YF1000602) and the Key Research and Development Projects in Shaanxi province (2021NY-020) and Natural Science Foundation of China (U1804106) and Qinghai Fundamental Scientific and Technological Research Plan (2018-ZJ-721) and the Scientific Research Guiding Plan Topic of Qinghai Hygiene Department (2018-wjzdx-131).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Conflict of interest

Authors declared that there is no conflict of interest.

Ethics approval

All animal experiments were performed in accordance with the guidelines and regulations and approved by the Animal ethics committee (Approval Number: DK2019052145, Northwest A&F University, Yangling, Shaanxi).

Consent to participate

Not applicable.

Consent for publication

Not applicable.