Dysbiosis in the Gut‐Liver Axis Is Associated With Low Bone Mass During Murine Cholestasis

ABSTRACT

The gut‐liver axis is the bidirectional relationship between the gut microbiota and the liver. Dysbiosis in the gut‐liver axis and disrupted bile acid homeostasis contribute to cholestatic liver disease pathogenesis. Patients afflicted with cholestasis have accelerated bone loss, a higher incidence of fractures, and are at risk of developing osteoporosis. However, the mechanisms underlying bone loss are largely unknown. The study purpose was to investigate the role of the gut‐liver axis and bile acid signaling on skeletal homeostasis during cholestasis. Male C57BL/6J specific‐pathogen‐free mice were administered 3,5‐diethoxycarbonyl‐1,4‐dihydrocollidine from age 11 to 15 weeks to induce cholestasis. 16s rDNA sequencing was performed on colonic contents. Livers were processed for qRT‐pCR. Trabecular and cortical bone were analyzed by micro‐CT. Osteoclast/osteoblast outcomes were determined by histomorphometry. Bile acid concentrations in serum and bone marrow were assessed by mass spectrometry. MC3T3‐E1 and RAW 264.7 cells were stimulated with bile acids at concentrations found in the bone marrow to determine their effects on osteoblasts and osteoclasts. Cholestatic mice had less bone mass than controls, attributed to increases in osteoclasts and decreases in osteoblasts. Following cholestatic injury, mice show dysbiotic shifts in their colonic bacteriome, increased expression of hepatic bile acid efflux transporters, and elevated bone marrow bile acids. In vitro, bile acids from the bone marrow of cholestatic mice suppressed osteoblastogenesis and promoted osteoclastogenesis, which was rescued by stimulating cells with a farnesoid X receptor agonist. This study introduces the gut‐liver axis as a novel regulator of skeletal homeostasis during cholestatic liver disease through dysregulated bile acid signaling.

Dysbiosis in the gut‐liver axis drives a dysregulation in bile acid metabolism that increases bile acid efflux into systemic circulation and the bone marrow. Bile acid signaling in the marrow promotes osteoclastogenesis and suppresses osteoblastogenesis, which drives bone loss during cholestatic liver disease in mice.

Abbreviations

ALP

Alkaline phosphatase

ALT

alanine aminotransferase

ASBT

apical sodium‐dependent bile acid transporter

AST

aspartate aminotransferase

BSH

bile salt hydrolase

CA

cholic acid

CLR

centered log‐ratio

DCA

deoxycholic acid

DDC

3,5‐diethoxycarbonyl‐1,4‐dihydrocollidine

GPBAR1/TGR5

g protein‐coupled bile acid receptor 1

IBAT

ileal bile acid transporter

LDA

linear discriminant analysis

LEfSe

linear discriminant analysis effect size

OCN

osteocalcin

PXR

pregnane X receptor

RANKL

receptor activator of NF‐κB ligand

SPF

specific‐pathogen‐free

TCA

taurocholic acid

TCDCA

taurochenodeoxycholic acid

TRAP

tartrate‐resistant acid phosphatase

TαMCA

tauro‐α‐muricholic acid

TβMCA

tauro‐β‐muricholic acid

TωMCA

tauro‐ω‐muricholic acid

UPLC

ultra‐performance liquid chromatography

VDR

vitamin D receptor

1. Introduction

Within the gastrointestinal tract resides the gut microbiota, a diverse community of bacteria, fungi, viruses, and other microorganisms that interact with the host to regulate host immunity, nutrient absorption, and metabolism [1, 2, 3]. The gut and liver are linked through the portal vein, forming what is known as the gut‐liver axis. This bidirectional relationship between the gut microbiota and the liver has implications in the development and homeostasis of host tissues. Dysbiosis in the gut‐liver axis has been associated with the onset of chronic liver conditions [4, 5, 6, 7], including cholestatic liver disease, which is characterized by impaired bile flow and the accumulation of toxic bile in the liver [8, 9].

Bile acids are metabolites synthesized in the liver from cholesterol and are conjugated with glycine or taurine before being excreted into the intestines [10, 11, 12]. In the gut, microbes deconjugate and metabolize bile acids [10, 11, 12]. In health, bile acids support nutrient absorption and act as antimicrobial agents that shape the gut microbiota and protect the liver [10, 11, 12]. From the intestines, bile acids can also recirculate back to the liver or can enter systemic circulation and act as signaling molecules at distant tissues [10, 11, 12]. During cholestasis, the abnormal accumulation of bile acids and gut microbiota dysbiosis can increase gut permeability and enhance liver inflammation and fibrosis, thereby driving the progression of cholestatic liver disease [13, 14, 15, 16, 17].

Hepatic osteodystrophy is a complication of cholestatic liver disease characterized by dysregulated bone turnover, reduced bone mineral density, and increased fracture risk. Patients with cholestatic liver disease lose bone at double the rate compared to age and sex‐matched populations [18]. Roughly 30% of patients with cholestatic liver disease have an increased prevalence of fragility fractures [19]. Moreover, between one‐third and one‐half of patients with cholestatic liver disease who are being evaluated for liver transplantation have severe bone disease or osteoporosis [20, 21]. Currently, there are no evidence‐based guidelines for treating metabolic bone disease in these patients.

Although hepatic osteodystrophy is well documented in patients with cholestatic liver disease, the biological mechanisms driving bone loss remain largely undefined. In patients with Alagille syndrome, bone mineral density and bone mineral content are negatively correlated with total serum bile acids [22]. Further, patients afflicted with primary sclerosing cholangitis and osteoporosis have elevated serum conjugated bile acids compared to patients with high bone mass [23]. Based on these clinical findings, it appears that bile acids may contribute to low bone mass in patients with cholestatic liver diseases.

Prior research has suggested that the gut microbiota signaling actions influence bone cell functions and skeletal homeostasis through a direct gut‐bone axis [24, 25]. Emerging evidence has demonstrated that the liver mediates microbiota actions on the skeleton, revealing a gut‐liver‐bone axis [26, 27, 28]. Notably, we previously demonstrated that gut microbiota dysbiosis causes increases in serum conjugated bile acids that suppress osteogenesis in mice [28]. While this report demonstrated bile acids as novel mediators of gut microbiota signaling actions on bone, to our knowledge, no known studies have quantified bile acid concentrations within the bone marrow and assessed their effects on bone cell differentiation and function.

The purpose of this study was to investigate the role of the gut‐liver axis and bile acids on bone loss during murine cholestasis. To induce cholestasis, mice were administered 3,5‐diethoxycarbonyl‐1,4‐dihydrocollidine (DDC). Reduced bone mass in cholestatic mice was attributed to dysbiotic shifts in the gut‐liver axis and bile acid dyshomeostasis. Elevated levels of conjugated bile acids in the bone marrow of cholestatic mice were linked to suppressed osteoblast activity and enhanced osteoclastogenesis. Agonizing the farnesoid X receptor (FXR) rescued the bile acid‐mediated decreases in osteoblast function and increases in osteoclastogenesis. Findings from this preclinical investigation demonstrate that bile acids may function as direct regulators of skeletal outcomes in the context of cholestatic liver disease.

2. Materials and Methods

2.1. Animals

Male 11‐week‐old specific‐pathogen‐free (SPF) C57BL/6J mice (RRID:IMSR_JAX:000664) were housed in ventilated cages in an SPF vivarium at the University of Pittsburgh. To induce cholestasis, mice were fed a diet containing 0.1% DDC (Bioserve, Frenchtown, NJ) or normal mouse chow from age 11 to 15 weeks. Animals were sacrificed at age 15 weeks. Blood was collected from the inferior vena cava and serum was isolated at time of sacrifice. Serum was analyzed by the University of Pittsburgh Medical Center Clinical Chemistry Laboratory to assess alkaline phosphatase (ALP), total bilirubin, alanine aminotransaminase (ALT), and aspartate aminotransferase (AST).

Animals were maintained on a 12 h:12 h light:dark schedule. Room temperature and humidity were maintained within the advised ranges per the NIH Guide for Care and Use of Laboratory Animals, as reported by Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines [29]. All animal studies were performed in accordance with the guidelines of the Institutional Animal Use and Care Committee at the University of Pittsburgh School of Medicine.

2.2. Micro‐Ct Analysis

Femurs were fixed in 10% neutral buffered formalin for 48 h at room temperature and then stored in 70% ethanol. Femurs were scanned using a Scanco μCT50 scanner (Scanco Medical, Brüttisellen, Switzerland) at the University of Pittsburgh Center for Craniofacial Regeneration Microcomputed Tomography at a 6 μm³ voxel size. Using the Scanco software, three‐dimensional images were reconstructed, and trabecular and cortical bone morphometry and bone mass were analyzed. Distal femur trabecular bone was assessed starting 300 μm proximal to the distal growth plate and extending 1800 μm (300 slices) proximally. A fixed threshold of 298 was used to identify mineralized tissue. Cortical bone was assessed in a 166‐slice segment of the femur mid‐diaphysis using a threshold of 368. Analysis was performed, and data are reported based on the Guidelines for Assessment of Bone Microstructure in Rodents Using Micro‐Computed Tomography [30].

2.3. Histology/Histochemistry

Livers were fixed in 10% neutral buffered formalin for 48 h at room temperature. Histological changes were evaluated by hematoxylin and eosin staining. 200X images were acquired using the Zeiss Axiovert 40 CFL inverted microscope.

Tibiae and femur were fixed in 10% neutral buffered formalin for 48 h at room temperature and then stored in 70% ethanol. Tibiae were decalcified using an ion‐exchange decalcification unit (Biocare Medical, Pacheco, CA, USA). Femur was decalcified in 14% EDTA for 21 days at room temperature. Tibia and femur were embedded in paraffin. Five μm serial frontal sections were cut through the proximal tibia. Five micrometers serial sagittal sections were cut through the distal femur.

Femurs were deparaffinized and rehydrated in graded ethanols. Femurs were then stained with tartrate‐resistant acid phosphatase (TRAP) and counterstained with hematoxylin for histomorphometric analysis of osteoclasts. TRAP+ cells lining the trabecular bone were scored as osteoclasts. The region of interest was the secondary spongiosa, initiated 300 μm proximal to the growth plate and extending 1800 μm proximally and 50 μm from endocortical surfaces. Images were acquired at 200X using the Nikon Eclipse NI‐E scope.

Tibia were deparaffinized, rehydrated with graded ethanols, and briefly washed in 1X PBS. Antigen retrieval was performed by incubating the samples with Proteinase K for 20 min at 37°C. Samples were cooled to room temperature and washed in deionized water. Specimens were blocked for 10 min at room temperature with SuperBlock blocking buffer (Thermo Scientific, Waltham, MA, USA). Specimens were then incubated with anti‐osteocalcin (OCN) primary antibody (AB10911, Millipore Sigma, Burlington, MA, USA; RRID:AB_1587337) diluted 1:100 in 0.1% BSA in PBS overnight at 4°C. Samples were washed with 0.2% Tween 20 in PBS and then incubated in secondary antibody diluted 1:500 in 0.1% BSA in PBS for 30 min at room temperature. Immunohistochemistry was performed using Vectastain ABC Elite Kit (Vector Laboratories, Burlingame, CA) and developed using DAB (Vector Laboratories). Samples were counterstained with hematoxylin. OCN+ cells lining trabecular bone were scored as osteoblasts. The region of interest was the secondary spongiosa, initiated 300 μm proximal to the growth plate and extending 1800 μm proximally and 50 μm from endocortical surfaces. 200X images were acquired using the Nikon Eclipse NI‐E scope.

2.4. 16S rDNA Sequencing Analysis

Colon contents were isolated and submitted to the University of Pittsburgh Center for Medicine and the Microbiome. DNA was extracted using the Qiagen DNeasy Powersoil Kit (Qiagen, Germantown, MD, USA) and processed per the manufacturer's protocols. V4 variable regions of bacterial 16S rDNA were amplified by PCR. Sequencing was performed on the Illumina Miseq v2 platform (Illumina, San Diego, CA, USA). Fastq files were filtered and processed using the DADA2 pipeline (RRID:SCR_023519) in the R statistical programming software (version 4.4.1) [31]. Paired reads with pathological errors and low‐quality scores were truncated. Sequencing errors were removed from data sequences using the DADA2 sample interference algorithm [32]. Then, the paired sequences were merged, and the chimera sequences were removed. The SILVA database (version 138.2; RRID:SCR_006423) was used to assign taxonomy to the amplicon sequencing variants. Sample data and the resulting microbial abundance table with corresponding taxonomic information were transferred to the phyloseq R statistical package (version 1.50.0) for analysis. A minimum abundance threshold of 0.01% was applied to exclude low‐abundance amplicon sequencing variants. α‐diversity was measured by the Chao1, Shannon, and Simpson indices. Relative abundance and centered‐log ratio abundance were measured across taxa. Abundances were measured using unpaired t‐tests with false discovery rate post hoc tests to correct for multiple comparisons [33]. Abundances across taxa are found in File S1. β‐diversity was measured using the Bray–Curtis and Jaccard dissimilarity matrices. The matrices were generated in the R statistical programming software, and the principal coordinate analysis plots were generated. The adonis function in the vegan R statistical package (version 2.6‐10) was used to carry out permutational multivariate analysis of variance to determine the significance in the Bray–Curtis and Jaccard matrices [34]. Linear discriminant analysis (LDA) effect size (LEfSe) analysis was carried out using the microbiomeMarker R package (version 1.13.2) to identify bacterial taxa that enriched or depleted during cholestasis [35, 36]. A p value cutoff of 0.05 and an LDA score of 4 were used to identify differentially abundant taxa. Plots were generated using ggplot2 (version 4.0.1). The 16S sequencing reads are available from the Sequence Read Archive Database under the BioProject accession number PRJNA1325370.

2.5. Liver Quantitative Real‐Time PCR mRNA Analysis

Liver was flash‐frozen, pulverized, and homogenized in TriZol Reagent (Invitrogen, Carlsbad, CA, USA) to isolate total RNA. The Nanodrop Lite spectrophotometer (Thermo Fisher Scientific, Pittsburgh, PA, USA) was used to quantify total RNA. cDNA was synthesized utilizing the SuperScript IV Synthesis Kit (Invitrogen, Carlsbad, CA, USA). The CFX96 Touch Real‐Time PCR Detection System (Bio‐Rad Laboratories, Hercules, CA, USA), using SYBR Green (Thermo Fisher Scientific, Pittsburgh, PA, USA), was used for RT‐PCR studies. Relative quantification of mRNA was determined via the 2^−ΔΔCT method [37, 38] using glyceraldehyde‐3‐phosphate dehydrogenase (Gapdh) expression as an endogenous housekeeping gene. Assays were performed in technical duplicate reactions and averaged for each biological replicate. Primer sequences are provided in Table 1.

TABLE 1.

List of primers used for quantitative RT‐PCR.

Primer	Sequence	Reference number
Gapdh (Forward)	5′‐GGAGAAACCTGCCAAGTATGA‐3′	275321430
Gapdh (Reverse)	5′‐TCCTCAGTGTAGCCCAAGA‐3′	275321431
Cyp7a1 (Forward)	5′‐TGGGCATCTCAAGCAAACAC‐3′	148516448
Cyp7a1 (Reverse)	5′‐TCATTGCTTCAGGGCTCCTG‐3′	148516449
Cyp27 (Forward)	5′‐TGCCTGGGTCGGAGGAT‐3′	151029518
Cyp27 (Reverse)	5′GAGCCAGGGCAATCTCATACTT‐3′	151029519
Cyp8b1 (Forward)	5′‐GCCCTTACTCCAAATCCTACCA‐3′	In‐house
Cyp8b1 (Reverse)	5′‐TCGCACACATGGCTCGAT‐3′	In‐house
Ntcp (Forward)	5′‐CACCATGGAGTTCAGCAAGA‐3′	151029516
Ntcp (Reverse)	5′‐AGCACTGAGGGGCATGATAC‐3′	151029517
Oatp4 (Forward)	5′‐GATCCTTCACTTACCTGTTCAA‐3′	In‐house
Oatp4 (Reverse)	5′‐CCTAAAAACATTCCACTTGCCATA‐3′	In‐house
Bsep (Forward)	5′‐ACACCATTGTATGGATCAACAGC‐3′	147406727
Bsep (Reverse)	5′‐CACCAACTCCTGCGTAGATGC‐3′	147406728
Mrp2 (Forward)	5′‐GCTTCCCATGGTGATCTCTT‐3′	151373038
Mrp2 (Reverse)	5′‐ATCATCGCTTCCCAGGTACT‐3′	151373039
Mrp3 (Forward)	5′‐TGAGATCGTCATTGATGGGC‐3′	In‐house
Mrp3 (Reverse)	5′‐AGCTGCGAGCGCAGGTCG‐3′	In‐house
Mrp4 (Forward)	5′‐TTAGATGGGCCTCTGGTTCT‐3′	In‐house
Mrp4 (Reverse)	5′‐GCCCACAATTCCAACCTTT‐3′	In‐house
Cyp2b10 (Forward)	5′‐CAATGGGGAACGTTGGAAGA‐3′	In‐house
Cyp2b10 (Reverse)	5′‐TGATGCACTGGAAGAGGAAC‐3′	In‐house
Cyp3a11 (Forward)	5′‐CCACCAGTAGCACACTTTCC‐3′	In‐house
Cyp3a11 (Reverse)	5′‐TTCCATCTCCATCACAGTATCA‐3′	In‐house

2.6. Bile Acid Quantification

Bile acid profiling was performed as previously described [39]. Briefly, 40 μL of serum or bone marrow homogenate was mixed with 120 μL methanol/acetonitrile (1:1, v/v) and centrifuged at 15000 rpm for 10 min. The supernatants were transferred to a separate Eppendorf tube for a second centrifugation at 15000 rpm for 10 min. Two microliters of the supernatants were injected into the UPLC‐QTOFMS (Waters, Milford, MA) system for analysis. The hydrophobicity index of the bile acid pool in liver and serum was calculated based on the Heyman index values of each bile acid species.

2.7. In Vitro Studies

2.7.1. Osteoblast Mineralization Studies

MC3T3‐E1 subclone 4 cells were purchased from ATCC (Manassas, VA, USA; RRID:CVCL_5440). Cells were grown in growth media (ascorbic acid‐free, α‐minimum essential medium supplemented with 10% FBS and 1% PS [100 U/mL penicillin–streptomycin]). For mineralization studies, cells were plated in growth media at a density of 1 × 10⁵ cells per well in 12‐well plates. The following day, culture media was replaced with osteogenic media (growth media, 50 mg/mL ascorbic acid, 10 mM β‐glycerophosphate, 10% FBS and 1% PS). To evaluate the effects of bone marrow bile acids on osteogenesis, cultures were stimulated in osteogenic media alone (negative), osteogenic media supplemented with bile acids at concentrations found in the bone marrow of control mice (control; 3.32 ng/mL cholic acid, 79.93 ng/mL DCA, 15.13 ng/mL TCA, 14.87 ng/mL TβMCA), or osteogenic media supplemented with bile acids at concentrations found in cholestatic mice (DDC; 110.66 ng/mL CA, 186.69 ng/mL DCA, 199.30 ng/mL TCA, 22.61 ng/mL TCDCA, 21.87 ng/mL TαMCA, 309.83 ng/mL TβMCA, 77.74 ng/mL TωMCA). Bile acids were purchased from Cayman Chemical (Ann Arbor, MI, USA). Cultures were supplemented with 1 μM GW4064 to agonize FXR (MedChemExpress, Monmouth Junction, NJ, USA). Media was changed every other day for 20 days. Mineralization was quantified by the alizarin red method. Alizarin red stained cultures were de‐stained with 10% cetylpyridinium chloride. Optical density was measured at 405 nm to quantify alizarin concentrations. Data is presented as the average concentration in three independent experiments performed in duplicate.

2.7.2. Osteoclastogenesis Studies

RAW 264.7 murine macrophage cells were purchased from ATCC (Manassas, VA, USA; RRID:CVCL_4093). Cells were grown in growth media (Dulbecco's modified eagle medium [DMEM]‐high glucose supplemented with 10% FBS and 1% PS). RAW 264.7 cells were passed and plated at 5 × 10³ cells per well in a 96‐well plate. Twenty‐four hours after plating, culture media were replaced with osteoclastogenic media (DMEM‐high glucose, supplemented with 50 ng/mL RANKL). To evaluate the effects of bone marrow bile acids on osteoclastogenesis, cultures were stimulated in osteoclastogenic media alone (negative), osteoclastogenic media supplemented with bile acids at concentrations found in the bone marrow of control mice (control; 3.32 ng/mL cholic acid (CA), 79.93 ng/mL deoxycholic acid (DCA), 15.13 ng/mL taurocholic acid (TCA), 14.87 ng/mL tauro‐β‐muricholic acid (TβMCA)), or osteoclastogenic media supplemented with bile acids at concentrations found in cholestatic mice (DDC; 110.66 ng/mL CA, 186.69 ng/mL DCA, 199.30 ng/mL TCA, 22.61 ng/mL taurochenodeoxycholic acid (TCDCA), 21.87 ng/mL TαMCA, 309.83 ng/mL TβMCA, 77.74 ng/mL TωMCA). Culture media were changed every other day. Cultures were supplemented with 1 μM GW4064 to agonize FXR (MedChemExpress, Monmouth Junction, NJ, USA). After 4 days, cultures were fixed and stained by the TRAP method. Four images per well were imaged at 200X. TRAP+ cells with ≥ 3 nuclei were scored as osteoclasts. Image analysis was carried out using ImageJ Fiji version 1.53s (NIH; https://imagej.net/ij/index.html; RRID:SCR_003070). Data is presented as the average concentration in three independent experiments performed in triplicate.

2.8. Statistical Analysis

Unpaired 2‐tailed t‐tests were performed to compare outcomes in control and DDC‐treated mice. For 16S rDNA sequencing analysis, relative and center log ratio abundances were analyzed using unpaired t‐tests with false discovery rate (5%) two‐stage step‐up method to correct for multiple comparisons [40]. Permutational multivariate analysis of variance was performed using the adonis function in the vegan R statistical package (version 2.6‐10) to determine the significance in the Bray–Curtis and Jaccard dissimilarity matrices. Two‐way ANOVA with Tukey's post hoc test was performed for in vitro osteoblast and osteoclast assays. Analyses were carried out utilizing GraphPad Prism 10 (GraphPad Software Inc., La Jolla, CA; RRID:SCR_002798) or the R statistical analysis software (version 4.4.2). Data are reported as mean ± standard error of the mean. Significance presented as *p < 0.05, **p < 0.01, ***p < 0.001. N values for in vivo and sequencing studies represent biological replicates. N values reported for in vitro experiments represent independent experiments.

3. Results

3.1. DDC Feeding Induces Cholestatic Liver Injury and Reduces Bone Mass

C57BL/6J mice were administered 0.1% DDC diet from ages 11 to 15 weeks. As expected, DDC caused liver injury characterized by ductular reaction, inflammation, and onion‐skin fibrosis (Figure 1a) [41, 42, 43]. Serum biochemistry revealed that markers for biliary injury, alkaline phosphatase (ALP, Figure 1b) and total bilirubin (Figure 1c), and markers for hepatocyte injury, alanine aminotransferase (ALT, Figure 1d) and aspartate aminotransferase (AST, Figure 1e), were increased following DDC feeding. Further, the liver weight to body weight ratio was increased in DDC‐treated versus control‐treated mice (Figure 1f). Body and femur lengths were similar in DDC‐treated versus control‐treated mice, implying normal skeletal growth was not disrupted (Figure 1g,h).

FIGURE 1.

Cholestatic mice have reduced bone mass. Male C57BL/6J specific pathogen‐free (SPF) mice were administered DDC from ages 11 to 15 weeks; euthanized at age 15 weeks. (a) Representative H&E‐stained liver sections (200X). (b–e) Serum biochemistry; (b) alkaline phosphatase (ALP); (c) total bilirubin; (d) alanine aminotransferase (ALT); (e) aspartate aminotransferase (AST). (f) Liver weight per body weight (%). (g) Body length (cm). (h) Femur length (cm). (i–l) Micro‐CT analysis of distal femur trabecular bone; (i) representative images; (j) trabecular bone mineral density (Tb. BMD); (k) bone volume per tissue volume (BV/TV). (l) Trabecular spacing (Tb. Sp). (m–p) Micro‐CT analysis of femur mid‐diaphysis cortical bone; (m) representative images; (n) cortical bone mineral density (Ct. BMD); (o) cortical area per tissue area (Ct. Ar/T. Ar); (p) cortical thickness (Ct. Th). n = 4/group; unpaired t‐test; reported as mean ± SEM; *p < 0.05 vs. control, **p < 0.01 vs. control, ***p < 0.001 vs. control.

Distal femur micro‐CT analysis (Figure 1i–l, Figure S1a–c) showed similar trabecular bone mineral densities in control versus cholestatic mice (Figure 1i,j). Mice afflicted with cholestasis had a roughly 25% decrease in bone volume fraction (Figure 1i,k), which was attributed to an increase in trabecular separation (Figure 1i,l), since trabecular thickness (Figure S1a), number (Figure S1b), and connectivity density (Figure S1c) were similar in control and cholestatic mice. Micro‐CT analysis of cortical bone parameters in the mid‐diaphysis of the femur (Figure 1m–p) showed a blunted cortical bone mineral density (Figure 1m,n). Cortical bone area fraction (Figure 1l,m,o) and cortical thickness (Figure 1m,p) were similar in control mice and mice with cholestasis.

3.2. Chronic Cholestasis Enhances Osteoclastogenesis and Suppresses Osteoblastogenesis

Histomorphometric analysis of TRAP‐stained distal femur sections was performed to determine if osteoclastogenesis is enhanced following cholestatic liver injury (Figure 2a–c). The number and interface of TRAP+ osteoclasts lining trabecular bone in the distal femur were increased in cholestatic mice versus control mice (Figure 2a–c). Immunohistochemical staining of osteocalcin (OCN) and histomorphometric analysis of proximal tibial sections were carried out to evaluate osteoblastogenesis (Figure 2d–f). The frequency of osteoblasts lining trabecular bone was similar in cholestatic versus control mice (Figure 2d,e). The osteoblast interface with trabecular bone was decreased in cholestatic mice versus controls (Figure 2d,f). Together, these findings suggest that the low bone mass phenotype in mice afflicted with cholestasis is attributed to decreases in osteoblasts and increases in osteoclasts.

FIGURE 2.

Cholestasis enhances osteoclastogenesis, suppresses osteoblastogenesis, and increases bone marrow adiposity. Male C57BL/6J specific pathogen‐free (SPF) mice were administered DDC from ages 11 to 15 weeks; euthanized at age 15 weeks. (a–c) Histomorphometric analysis of tartrate‐resistant acid phosphatase–positive (TRAP) osteoclasts in the distal femur: (a) representative images; (b) osteoclast number per bone perimeter (N. Oc/B. Pm); (c) osteoclast perimeter per bone perimeter (Oc. Pm/B. Pm). (d–f) Histomorphometric analysis of osteocalcin (OCN+) cells in the proximal tibia: (d) representative images; (e) osteoblast number per bone perimeter (N. Ob/B. Pm); (f) osteoblast perimeter per bone perimeter (Ob. Pm/B. Pm). n = 4/group; unpaired t‐test; reported as mean ± SEM; *p < 0.05 vs. control, **p < 0.01 vs. control.

3.3. Chronic Cholestasis Induces Shifts in the Gut Bacteriome

Dysbiotic shifts are associated with the pathogenesis of cholestatic liver diseases. Further, gut microbiota dysbiosis has been shown to suppress osteogenesis and bone strength, leading to low bone mass [26, 27, 28, 44, 45, 46, 47, 48]. Therefore, we performed 16S rDNA sequencing on colonic contents from control and cholestatic mice to evaluate changes in the gut microbiota.

Alpha diversity measures within‐sample diversity, which describes the number of different taxa (richness) and taxa distribution (evenness) [49, 50, 51]. The Chao1 index, which measures richness [49, 50, 51], was increased in cholestatic mice (Figure 3a). The Shannon index measures evenness, and the Simpson index measures the alpha diversity by accounting for richness and evenness [49, 50, 51]. Both the Shannon and Simpson indices were similar in control and cholestatic mice (Figure S2a,b). Beta diversity measures the similarity of microbiota between two or more communities [52, 53]. The Bray–Curtis dissimilarity measurement accounts for the abundance and presence/absence of taxa in each data set [52, 53]. The Jaccard distance is based on the presence/absence of taxa without accounting for abundance information [52, 53]. Permutational multivariate analysis of variance of the beta diversity showed significant differences in the Bray–Curtis (p = 0.033, Figure 3b) and Jaccard (p = 0.029, Figure S2c) matrices when comparing the colonic bacteriomes of cholestatic mice versus control mice.

FIGURE 3.

Cholestasis causes dysbiotic changes in the colonic bacteriome. Male C57BL/6J specific pathogen‐free (SPF) mice were administered DDC from ages 11 to 15 weeks; euthanized at age 15 weeks. Genomic DNA was isolated from colonic contents, and 16S rDNA sequencing was performed. (a) Bacterial α diversity (Chao1 index). Unpaired t‐test; reported as mean ± SEM; ***p < 0.001 vs. control. (b) Principal coordinate analysis of Bray–Curtis distances indicates visual and statistical separation of microbiota communities in DDC versus Control mice. Permutational multivariate analysis of variance, p = 0.033. Genera level abundances normalized by (c) centered log ratio and (d) total sum scaling (relative abundance, %). (e, f) Linear discriminant analysis effect size (LEfSe) of 16S sequencing studies. Bacteria were considered significant with a linear discriminant analysis score ≥ 4 and a p value cutoff of 0.05. Visualized as a (e) dot plot and (f) cladogram to show phylogenetic distribution of microbiota associated with control and cholestatic mice. n = 4/group; unpaired t‐test with false discovery rate post hoc test, reported as mean ± SEM; *p < 0.05 vs. control, **p < 0.01 vs. control, ***p < 0.001 vs. control.

To evaluate differences in the abundances of taxa, we normalized abundance data by centered log‐ratio (CLR), which uses the geometric mean of the read counts within a sample, or by total sum scaling (relative abundance), which measures abundances based on total read counts (File S1). At the phylum‐level, the CLR and relative abundances of Actinomycetota were increased, and Pseudomonadota were decreased (Figure S2d,e; File S1). CLR abundance of Verrucomicrobiota was increased in cholestatic mice versus controls (Figure S2d). At the genus‐level, most detected taxa exhibited significant changes in cholestatic mice based on CLR normalization (Figure 3c). The relative abundance of Parasutterella was decreased, and Clostridium sensu stricto 1 was increased following DDC administration (Figure 3d).

Since compositional changes in the gut microbiota were observed in cholestatic mice, we performed LEfSe analysis to identify differentially abundant taxa that may serve as biomarkers of cholestasis and low bone mass (Figure 3e,f). Eleven genera were identified with LDA scores ≥ 4, and a p value ≤ 0.05 (Figure 3e,f). In CTRL mice, the genera Ligilactobacillus, Parasutterella, and Lachnospiraceae UCG‐006 were enriched (Figure 3e,f). In cholestatic mice, the genera Bifidobacterium, Turicibacter, HT002, Clostridium sensu stricto 1, Limosilactobacillus, Rumiinococcus, Akkermansia, and Coriobacteriaceae UCG‐002 were significantly enriched (Figure 3e,f). 16S rDNA sequencing outcomes demonstrate that chronic cholestasis is associated with shifts in the colonic bacteriome.

3.4. Chronic Cholestasis Disrupts Hepatic Bile Acid Metabolism and Transport

Since gut dysbiosis and cholestatic liver disease disrupt hepatic bile acid homeostasis [13, 54, 55, 56], we assessed hepatic bile acid metabolism and transport genes. Bile acid synthesis cytochrome (Cyp) enzymes Cyp27 and Cyp8b1 were downregulated in cholestatic mice (Figure 4a). The expression of the basolateral bile acid uptake transporters, sodium‐taurocholate cotransporting polypeptide (Ntcp/Slc10a1) and organic anion transporting polypeptide‐4 (Oatp4/Slco1b2), was downregulated in cholestatic mice (Figure 4b). The expression of bile salt export pump (Bsep/Abcc11), the bile acid apical transporter, was also downregulated in mice afflicted with cholestasis (Figure 4b). Mrp2, another apical bile acid transporter, was upregulated following DDC administration (Figure 4b). The expression of bile acid efflux transporters (Mrp3, Mrp4, Figure 4b) and detoxification enzymes (Cyp2b10, Cyp3a11, Figure 4c) was also upregulated in cholestatic mice.

FIGURE 4.

Cholestasis disrupts hepatic bile acid metabolism and transport and alters serum bile acid concentrations. Male C57BL/6J specific pathogen‐free (SPF) mice were administered DDC from ages 11 to 15 weeks; euthanized at age 15 weeks. Liver qRT‐PCR analysis of (a) bile acid synthesis genes (Cyp7a1, Cyp27, Cyp8b1), (b) bile acid transporters (Ntcp, Oatp4, Bsep, Mrp2, Mpr3, Mrp4), and (c) detoxification enzymes (Cyp2b10, Cyp3a11). qRT‐PCR outcomes analyzed by 2^−ΔΔCT method, normalized to Gapdh. (d–h) Mass spectrometry was used to measure the abundance of bile acid species in serum; (d) hydrophobicity index of serum bile acid pool was calculated based on Heuman index for each bile acid species; (e) average distribution of serum bile acids; (f) serum primary bile acid concentration; (g) serum secondary bile acid concentration; (h) serum conjugated bile acid concentration. n = 4/group; unpaired t‐test; reported as mean ± SEM; *p < 0.05 vs. control, **p < 0.01 vs. control, ***p < 0.001 vs. control.

We collected serum and performed ultra‐performance liquid chromatography (UPLC) coupled with quadrupole time‐of‐flight mass spectrometry to evaluate changes in serum bile acids in mice afflicted with cholestatic liver injury (Figure 4d–h, Table 2). The hydrophobicity of serum bile acids in cholestatic mice was significantly lower than in control mice (Figure 4d). These findings indicate that there is a shift toward a more soluble bile acid profile in serum in cholestatic mice, which reduces bile acid toxicity and facilitates their secretion into the blood. Following DDC administration, there were significant shifts in the proportions of conjugated primary and secondary bile acids, including a roughly 20% increase in the prevalence of TβMCA (Figure 4e). Cholestatic mice exhibited increases in the concentrations of most bile acid species in serum (Table 2). Overall, the concentrations of serum primary, secondary, and conjugated bile acids were elevated in cholestatic mice versus control mice (Table 2, Figure 4f–h).

TABLE 2.

Serum bile acid concentrations (ng/mL) of control and DDC‐treated mice.

Bile acid (ng/mL)	Control (average ± SEM)	DDC (average ± SEM)	p
Total bile acids	1222.145 ± 219.774	371 226.350 ± 16 009.609	4.304E‐07***
Cholic acid (CA)	40.767 ± 9.183	10 156.610 ± 4352.417	5.911E‐02
Deoxycholic acid (DCA)	14.535 ± 1.327	129.716 ± 43.819	3.920E‐02*
Taurocholic acid (TCA)	400.878 ± 60.134	120 465.628 ± 16 007.494	2.891E‐04***
Taurodeoxycholic acid (TDCA)	16.671 ± 1.343	357.521 ± 89.044	8.865E‐03**
Taurochenodeoxycholic acid (TCDCA)	30.962 ± 2.685	3400.377 ± 210.520	3.780E‐06***
Taurohyodeoxycholic acid (THDCA)	22.542 ± 2.099	1165.179 ± 153.103	2.987E‐04***
Tauroursodeoxycholic acid (TUDCA)	24.180 ± 3.326	298.750 ± 20.747	1.238E‐05***
α‐muricholic acid (αMCA)	0.485 ± 0.291	86.664 ± 27.086	1.904E‐02*
β‐muricholic acid (βMCA)	27.444 ± 6.220	5721.593 ± 1772.288	1.830E‐02*
ω‐muricholic acid (ωMCA)	19.642 ± 3.868	3429.046 ± 1035.616	1.657E‐02*
Tauro‐α‐muricholic acid (TαMCA)	64.894 ± 9.165	5267.705 ± 367.462	7.768E‐06***
Tauro‐β‐muricholic acid (TβMCA)	346.067 ± 73.734	175 162.075 ± 2025.774	1.637E‐10***
Tauro‐ω‐muricholic acid (TωMCA)	312.532 ± 0.164	45 421.040 ± 2984.737	5.299E‐06***
Glycocholic acid (GCA)	0.544 ± 0.164	164.428 ± 38.250	5.180E‐03**

Note: Unpaired 2‐tailed t‐test, reported as mean ± SEM.

^* p < 0.05 vs. control.

^** p < 0.01 vs. control.

^*** p < 0.001 vs. control.

Gut microbes encode enzymes to deconjugate bile acids and transform primary bile acids into secondary bile acids [10, 11, 12]. Compositional shifts in the gut microbiota can alter the microbial metabolism of bile acids [10, 11, 12]. In addition, during cholestasis, basolateral bile acid efflux is induced to protect against hepatobiliary damage [57, 58, 59]. Based on our 16S sequencing findings and liver qRT‐PCR studies, it appears that dysbiosis in the gut‐liver axis causes dysregulation in bile acid homeostasis and drives bile acid efflux into systemic circulation.

3.5. Agonizing FXR Rescues Bile Acid‐Induced Decreases in Osteoblast Function and Increases in Osteoclastogenesis In Vitro

Bile acids that escape the enterohepatic loop can enter systemic circulation and act as signaling molecules at distant tissue sites. Bile acids and their receptors have been shown to influence bone cell actions [60, 61]. Prior work has demonstrated that gut microbiota dysbiosis increases the concentration of serum conjugated bile acids that suppress osteogenesis [28]. While this work introduced that bile acids mediate gut microbiota signaling actions on bone [28], no known studies have evaluated bile acid concentrations in the bone marrow and determined their effects on bone cell differentiation and function. Therefore, we performed UPLC mass spectrometry to detect changes in the bile acid profile in the bone marrow of cholestatic mice versus control mice and evaluated the role of these bile acids on osteoblastogenesis and osteoclastogenesis (Figure 5).

FIGURE 5.

Agonizing FXR rescues bile acid‐induced decreases in osteoblast function and increases in osteoclastogenesis in vitro. (a) Mass spectrometry was used to measure the abundance of bile acid species in bone marrow (ng/mL). Cholic acid (CA), deoxycholic acid (DCA), taurocholic acid (TCA), taurochenodeoxycholic acid (TCDCA), tauro‐α‐muricholic acid (TαMCA), tauro‐β‐muricholic acid (TβMCA), tauro‐ω‐muricholic acid (TωMCA). n = 4/group; unpaired t‐test; reported as mean ± SEM; **p < 0.01 vs. control, ***p < 0.001 vs. control. (b) MC3T3‐E1 subclone four cells were stimulated in osteogenic media alone (negative) or osteogenic media with the bile acid concentrations found in the bone marrow of control mice (control) or cholestatic mice (DDC). Cultures were also co‐stimulated with GW4064 to agonize FXR. Alizarin red staining was performed to assess osteoblast function. Data presented as alizarin red concentrations (μM). Two‐way ANOVA with Tukey's post hoc test; reported as mean ± SEM; *p < 0.05, ***p < 0.001. n = 3/group; each data point represents an independent experiment performed in duplicate and averaged. (c–f) RAW 264.7 cells were treated with 50 ng/mL RANKL to initiate the fusion of RAW 264.7 cells into small multinucleated cells. Cultures were stimulated with osteoclastogenic media alone (negative), or osteoclastogenic media with the bile acid concentrations found in the bone marrow of control mice (control) or cholestatic mice (DDC). Cultures were also co‐stimulated with GW4064 to agonize FXR. Tartrate‐resistant acid phosphatase (TRAP) method was performed to assess osteoclastogenesis; (c) representative images; (d) number of osteoclasts per well (N. Oc/Well); (e) average osteoclast area (Oc. Ar/Oc); (f) nuclei number per osteoclast (N. Nc/Oc). Two‐way ANOVA with Tukey's post hoc tests; reported as mean ± SEM; *p < 0.05. n = 3/group; each data point represents the average of three independent experiments performed in triplicate.

In the bone marrow of control mice, cholic acid (CA), deoxycholic acid (DCA), taurocholic acid (TCA), and tauro‐β‐muricholic acid (TβMCA) were detected (Figure 5). In cholestatic mice, there were increases in total bone marrow bile acids, including a 32‐fold increase in CA, a 1.34‐fold increase in DCA, a 12‐fold increase in TCA, and a 20‐fold increase in TβMCA (Figure 5a). The bile acids taurochenodeoxycholic acid (TCDCA), TαMCA, and TωMCA were also found in the bone marrow of cholestatic mice, which were not detected in the bone marrow of control mice (Figure 5a).

To discern the impact of these bile acids on bone cells, we carried out in vitro studies stimulating murine MC3T3‐E1 cells (osteoblasts) [62] and murine RAW 264.7 macrophages (osteoclastic cells) [63, 64] with the bile acids at concentrations found in the bone marrow of control or DDC mice. Mineralization of MC3T3‐E1 cells was assessed by alizarin red staining following stimulation with osteogenic media (αMEM, 10% FBS, 1% Penicillin–Streptomycin [PS] supplemented with 50 mg/mL ascorbic acid and 10 mM β‐glycerophosphate) for 20 days (Figure 5b). Cultures were stimulated with osteogenic media alone (negative), or osteogenic media supplemented with the bile acid profile found in the bone marrow of control mice (control) or cholestatic mice (DDC) (Figure 5b). Mineralization potential was similar in cultures treated with osteogenic media alone and control cultures (Figure 5b). Stimulating MC3T3‐E1 cells with osteogenic media containing the bile acid profile found in cholestatic mice suppressed mineralization compared to cultures stimulated with osteogenic media alone or with the bile acid profile from control mice (Figure 5b).

RAW 264.7 cells were treated in osteoclastogenic media (DMEM high glucose, 10% FBS, 1% PS supplemented with 50 ng/mL receptor activator of NF‐κB ligand [RANKL]) to induce osteoclastogenesis. RAW 264.7 cultures were stimulated in osteoclastogenic media alone (negative) or osteoclastogenic media with the bile acid concentrations found in the bone marrow of control mice (control) or cholestatic mice (DDC) (Figure 5c–f). Osteoclast number, size, and nuclei per osteoclast were similar in cultures stimulated with osteoclastogenic media alone versus cultures stimulated with bile acids at the concentrations found in the bone marrow of control mice (Figure 5c–f). Together with the osteoblast mineralization findings, these results suggest that bile acids in the bone marrow do not impact bone cell functions and bone turnover under healthy conditions. The number of nuclei per osteoclast was increased following stimulation with osteoclastogenic media supplemented with bile acids at concentrations found in the bone marrow of cholestatic mice, compared with negative and control cultures (Figure 5c,f). In vitro findings showing that bile acids found at the concentration in the bone marrow of cholestatic mice impair osteoblast function and promote osteoclastogenesis parallel histomorphometric observations made in vivo. Therefore, disruptions in bile acid homeostasis appear to contribute to the low bone mass phenotype observed during chronic cholestasis.

The FXR is a prominent bile acid receptor expressed on osteoblasts and osteoclasts [65, 66]. Activation of FXR promotes osteogenesis and suppresses osteoclastogenesis [65, 66, 67, 68, 69, 70]. Prior work has linked changes in serum conjugated bile acids to suppressed osteogenesis in mice due to attenuated FXR signaling [28]. In the bone marrow of cholestatic mice, there were increases in bile acids that weakly agonize or antagonize FXR. For example, TCA and TCDCA are weaker agonists at FXR than their unconjugated forms, and TαMCA, TβMCA, and TωMCA are potent FXR antagonists [71, 72, 73, 74, 75].

To determine the role of FXR signaling in bile acid‐mediated effects on bone cells during cholestasis, MC3T3‐E1 and RAW 264.7 cells were stimulated with the base media supplemented with bile acids found in the bone marrow of control and cholestatic mice in the presence of GW4064, an FXR agonist. In cultures treated with bile acids at concentrations found in the bone marrow of cholestatic mice, GW4064 increased osteoblast mineralization (Figure 5b) and significantly decreased the number of nuclei per osteoclast to levels similar to those in control cultures (Figure 5c,f). These findings suggest that during cholestasis, bile acids repress osteoblast function and enhance osteoclastogenesis, likely due to reduced FXR activation. Further, agonizing FXR likely protects against bile acid‐induced bone loss.

4. Discussion

This report demonstrates that dysbiosis in the gut‐liver axis during cholestatic liver disease induces bone loss through alterations in bile acid metabolism and signaling. Prior research has demonstrated that gut microbiota affects the skeleton through a gut‐liver‐bone axis [26, 27]. Notably, our prior work identified bile acids as mediators of gut‐liver signaling actions on bone cell function and osteogenesis [28]. In mice with cholestasis, gut dysbiosis was associated with disruption of bile acid metabolism and increased bone marrow bile acid concentrations. Changes in bone marrow bile acids during cholestasis suppressed osteoblast function and enhanced osteoclastogenesis in vitro, which were rescued when cultures were stimulated with an FXR agonist. This study provides further evidence that bile acids contribute to gut microbiota effects on bone and introduces the gut‐liver axis as a potential regulator of low bone mass in patients afflicted with cholestatic liver disease.

The results reported herein show that bile acids have a direct role in inducing bone loss during cholestatic liver disease. Agonizing FXR in vitro rescued the impaired osteoblast mineralization potential and enhanced osteoclastogenesis caused by treating cells with bile acids at concentrations found in the bone marrow of cholestatic mice. Agonizing FXR in vivo has hepatoprotective effects against cholestatic liver injury [76, 77, 78, 79, 80]. Moreover, prior work has shown that FXR agonists promote bone formation and suppress bone resorption in ovariectomized and osteoarthritic mice [67, 68]. This suggests that FXR agonist therapy in vivo may also protect against cholestasis‐induced bone loss.

In addition to FXR regulating bone cell functions, other bile acid receptors, such as G protein‐coupled bile acid receptor 1 (GPBAR1/TGR5), can influence bone cell differentiation and function. Loss of TGR5 alters the bone marrow microenvironment by decreasing the bone volume fraction and marrow adiposity in long bones [81, 82]. The vitamin D receptor (VDR) and the pregnane X receptor (PXR) are nuclear bile acid receptors that also support bone formation and suppress bone resorption [60, 83, 84]. Lithocholic acid (LCA) is a potent activator of TGR5 and PXR [85]. DCA is also a strong TGR5 agonist [85]. VDR is activated by LCA and its derivatives, whereas CDCA, CA, DCA, and MCAs do not activate VDR [86]. LCA was not detected in the bone marrow of control mice or mice afflicted with cholestasis. Furthermore, there were no significant differences in DCA levels in the marrow between control and cholestatic mice. Together, these findings imply that changes in the bone marrow bile acid profile induced by DDC‐feeding suppress osteoblast function and promote osteoclastogenesis by inhibiting FXR signaling, not TGR5, PXR, or VDR‐mediated signaling.

16S sequencing findings reflect the potential presence/absence of gut microbes in cholestatic liver disease. We identified eight enriched genera and three depleted genera in cholestatic mice, which may contribute to the pathogenesis of cholestasis and low bone mass. For example, Ruminococcus was enriched following DDC feeding. Increases in this genus have been reported in patients with primary sclerosing cholangitis [87] and have been positively correlated with bone resorption and the risk of osteoporosis [88]. The genus Ligilactobacillus was depleted in mice fed DDC. Ligilactobacilli are commonly used as probiotics due to their role in regulating the host immune response and metabolism [89, 90]. L. murinus , previously Lactobacillus murinus [91], administration protects against cholestatic liver injury in DDC‐fed mice [92]. L. salivarus has protective roles against bone resorption in rats [93], and L. ruminius presence is reduced in patients with osteoporosis versus patients with normal bone mass [94]. Future studies are necessary to elucidate the role of specific bacteria in driving low bone mass during cholestasis and whether this phenotype is driven by the enrichment of pathogenic microbes and/or the depletion of health‐promoting microbes.

Bile salt hydrolases (BSH) are expressed by bacteria in mice and humans to deconjugate taurine and glycine from bile acids that enter the intestines. In humans, cholestasis is associated with shifts in the gut microbiota and reduced BSH expression and enzymatic activity [95]. Decreasing BSH activity is linked to suppressed FXR activity, attributed to increases in TβMCA and conjugated bile acids, which are FXR antagonists and weak agonists, respectively [96, 97, 98]. In the serum and bone marrow of mice fed DDC, we observed increases in total conjugated bile acids, which suggests that BSH activity is likely suppressed in mice with cholestasis. These findings parallel a clinical study demonstrating that patients with primary sclerosing cholangitis and low bone mineral density have increased serum conjugated bile acids compared with patients with normal bone mass [23]. We suspect that BSH expression and activity are suppressed during cholestasis, which increases serum and bone marrow levels of weak FXR agonists and FXR antagonists. In turn, FXR signaling in bone cells is impaired, suppressing bone formation, increasing osteoclastogenesis, and leading to low bone mass.

The observed accumulation of conjugated bile acids suggests that BSH activity is suppressed during cholestasis. However, our 16S rDNA sequencing findings are limited to evaluating the taxonomic composition of the gut microbiota at the genus level and are unable to assess microbial function. Future metatranscriptomic and targeted metabolomic studies are needed to elucidate the functional activity of the gut microbiota to determine whether BSH activity is suppressed and contributes to low bone mass in cholestatic liver disease. Another limitation of this study is that the gut microbiome composition differs between mice and humans, which affects bile acid metabolism [99, 100, 101, 102, 103]. Mice produce MCAs, which are typically present at low concentrations in humans [99, 100, 101, 102, 103]. In addition, murine bile acids are predominantly conjugated with taurine, whereas in humans, they are predominantly conjugated with glycine [99, 100, 101, 102, 103]. Despite compositional differences, the murine and human gut microbiota have 85%–95% functional overlap [104]. This further supports the need for metatranscriptomic and targeted metabolomic studies to define translatable, functional changes in the gut microbiota during cholestasis and how these changes contribute to attenuated FXR signaling and low bone mass.

Findings from this study demonstrate that bile acids are potential mediators of gut‐liver signaling actions on bone during cholestatic liver disease. Therefore, preventing bile acid efflux into systemic circulation and the bone marrow may be a potential therapeutic strategy to prevent cholestasis‐induced bone loss. Prior work has shown that the bile acid sequestrant cholestyramine ameliorates bone loss in mice with experimental irritable bowel syndrome, which was attributed to reduced bile acid content [105]. Ileal bile acid transporter (IBAT) inhibitors are a new class of drugs used to treat Alagille syndrome and progressive familial intrahepatic cholestasis [106, 107]. IBAT inhibitors disrupt the ileal reuptake of bile acids, preventing their recirculation and reducing the total bile acid pool size [106, 107]. Currently, it is unclear whether IBAT inhibitors protect against bile acid‐induced bone loss during cholestasis. Based on our study, suppressing bile acid efflux would likely attenuate bone loss by decreasing bile acid concentrations in the bone marrow. Future studies are needed to determine the efficacy of bile acid sequestrants and IBAT inhibitors as viable therapeutic strategies for treating cholestasis‐induced bone loss.

Our preclinical findings demonstrate that dysbiosis in the gut‐liver axis is associated with low bone mass during murine cholestasis, which is attributed to bile acid‐induced increases in osteoclastogenesis and suppressed osteoblast function. Currently, there are limited studies evaluating the mechanisms underlying low bone mass and increased prevalence of fractures in patients afflicted with cholestatic liver disease. In patients with Alagille syndrome, bone mineral density and bone mineral content are negatively associated with total serum bile acids [22]. Further, patients with primary sclerosing cholangitis and osteoporosis show elevated levels of serum conjugated bile acids compared to those with high bone mass [23]. Together with this preclinical work, it is likely that disruptions in the gut‐liver axis and bile acid metabolism/signaling contribute to skeletal dysfunction in patients afflicted with cholestatic liver disease. However, clinical studies are required to discern the relationship between the gut‐liver axis, bile acids, and skeletal outcomes in patients with cholestatic liver disease.

Author Contributions

M.D.C. conceived the study; K.N.‐B. and M.D.C. designed the study; B.H., J.F., R.C., and M.D.C. acquired and analyzed data; B.H., R.C., J.F., P.C., K.N.‐B., M.D.C. interpreted the data; B.H. and M.D.C. wrote the manuscript; all authors revised and approved the final version of the manuscript; B.H. and M.D.C. take responsibility for the integrity of the data analysis.

Funding

This work was supported by the American Liver Foundation (ALF), Pitt | Pittsburgh Liver Research Center, University of Pittsburgh (PLRC), and HHS | NIH | National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (R01DK103775).

Conflicts of Interest

Kari Nejak‐Bowen is a consultant for Surrozen Inc.

Supporting information

Figure S1: Cholestatic mice have reduced bone mass. (a–c) Micro‐CT analysis of distal femur trabecular bone; (a) trabecular thickness (Tb. Th); (b) trabecular number (Tb. N); (c) connectivity density (Conn. D). n = 4/group; unpaired t‐test; reported as mean ± SEM.

Figure S2: Cholestasis causes dysbiotic changes in the colonic bacteriome. Male C57BL/6J specific pathogen‐free (SPF) mice were administered DDC from ages 11 to 15 weeks; euthanized at age 15 weeks. Genomic DNA was isolated from colonic contents, and 16S rDNA sequencing was performed. (a) Bacterial α‐diversity (Shannon index). (b) Bacterial α‐diversity (Simpson index). (c) Principal coordinate analysis of Jaccard distances indicates visual and statistical separation of microbiota communities in DDC versus Control mice. Permutational multivariate analysis of variance, p = 0.029. Phylum‐level abundances normalized by (d) centered log ratio and (e) total sum scaling (relative abundance, %). n = 4/group; unpaired t‐test with false discovery rate post hoc test, reported as mean ± SEM; **p < 0.01 vs. control.

Table S1. Cholestasis casues dysbiotic changes in the colonic bacteriome. Male C57BL/6J specific pathogen‐free (SPF) mice were administered DDC fro ages 11 to 15 weeks; euthanized at age 15 weeks. Genomic DNA was isolated from colonic contents, and 16S rDNA sequencing was performed. Abundances across taxa were measured using centered log ratio and total sum scaling (relative abundance, %) normalization. n = 4/group.

Data Availability Statement

The data supporting these findings of this study are available in the Materials and Methods, Results, and/or Supporting Information of this article. Further information can be obtained upon request from the corresponding author.