Intestinal farnesoid X receptor signaling controls hepatic fatty acid oxidation

Abstract

In addition to maintaining bile acid, cholesterol and glucose homeostasis, farnesoid X receptor (FXR) also regulates fatty acid β-oxidation (FAO). To explore the different role of hepatic and intestinal FXR in liver FAO, FAO-associated metabolites, including acylcarnitines and fatty acids, and FXR target gene mRNAs were profiled using an integrated metabolomic and transcriptomic analysis in control (Fxr^fl/fl), liver-specific Fxr-null (Fxr^ΔHep) and intestine-specific Fxr-null (Fxr^ΔIE) mice, treated either with the FXR agonist obeticholic acid (OCA) or vehicle (VEH). Activation of FXR by OCA treatment significantly increased fatty acyl-CoA hydrolysis (Acot1) and decreased FAO-associated mRNAs in Fxr^fl/fl mice, resulting in reduced levels of total acylcarnitines and relative accumulation of long/medium chain acylcarnitines and fatty acids in liver. Fxr^ΔHep mice responded to OCA treatment in a manner similar to Fxr^fl/fl mice while Fxr^ΔIE mice responded differently, thus illustrating that intestinal FXR plays a critical role in the regulation of hepatic FAO. A significant negative-correlation between intestinal FXR-FGF15 and hepatic CREB-PGC1A pathways was observed after both VEH and OCA treatment, suggesting that OCA-induced activation of the intestinal FXR-FGF15 axis downregulates hepatic PGC1α signaling via inactivation of hepatic CREB, thus repressing FAO. This mechanism was confirmed in experiments based on human recombinant FGF19 treatment and intestinal Fgf15 depleted mice. This study revealed an important role for the intestinal FXR-FGF15 pathway in hepatic FAO repression.

1. Introduction

Mitochondrial fatty acid β-oxidation (FAO) is an essential pathway for maintaining energy homeostasis, especially during prolonged fasting as FAO is an alternative source for ATP production when glucose and glycogen storage is low [1]. Long-chain fatty acids are the most abundant fatty acids stored in the body as triglycerides, and more than 15 enzymes are required to oxidize long-chain fatty acids when the blood glucose levels drop. Fatty acids enter the cell via fatty acid transport systems (CD36, FABPs and FATPs) and are conjugated by acyl-CoA synthetases with CoA to form acyl-CoAs in the cytosol for further degradation or elongation. Carnitine participates in the transfer of long-chain acyl-CoAs to mitochondria to initiate FAO by the carnitine shuttle [2]. In this shuttle, membrane bound carnitine palmitoyltransferase 1 (CPT1) exchanges CoA with a carnitine molecule to produce acylcarnitines, that pass the inner mitochondrial membrane into the matrix facilitated by carnitine acylcarnitine translocase (CACT) and then CPT2 reconverts the inner-mitochondrial acylcarnitines to acyl-CoAs for subsequent FAO via degradation by a series of enzymes with different chain-length specificities, including very long-chain acyl-CoA dehydrogenase (VLCAD, coding by ACADVL), long-chain acyl-CoA dehydrogenase (VLCAD, coding by ACADL), medium-chain acyl-CoA dehydrogenase (MCAD, coding by ACADM) and short-chain acyl-CoA dehydrogenase (SCAD, coding by ACADS) [3]. Following the initial dehydrogenation, the resulting enoyl-CoAs undergo hydration, a second dehydrogenation step, and finally thiolytic cleavage, leading to the formation of a molecule of acetyl-CoA and a shortened acyl-CoA with two carbons less. The acetyl-CoA enters the tricarboxylic acid cycle for production of reducing equivalents for oxidative phosphorylation resulting in ATP production [4].

Hepatic peroxisome proliferator-activated receptor α (PPARα) and farnesoid X receptor (FXR) are activated under the fasted and fed state, respectively [5, 6]. PPARα directly regulating genes involved in fatty acid transport and FAO, regardless of whether fatty acids are in the diet [5–7]. This role for PPARα in FAO is supported by a large number of studies with synthetic agonists and animal models such as Ppara-null mice [5–7]. The most convincing study is that Wy-14643, a potent PPARα agonist, failed to induce the expression of hepatic FAO-related genes including Cyp4a1, Cyp4a3, Acox1, Ehhadh, and Acaa1a in Ppara-null mice [8]. However, during feeding, PPARα signaling returns to normal levels and shifts its activity to coordinate different pathways of de novo lipid synthesis, consequently supplying fatty acids for storage as hepatic TG in lipid droplets [5, 6, 9].

In contrast to PPARα, FXR is activated in the fed state by bile acids that return to the liver along with nutrients via enterohepatic circulation. In addition to maintaining bile acid homoeostasis, FXR exerts direct effects on metabolic pathways, including suppression of both gluconeogenesis and lipogeneses [10–12]. An earlier study reported that mice fed a cholic acid (CA)-containing diet exhibited reduced constitutive levels of the PPARα target genes mRNAs, Cyp4a1, Cyp4a3, Acox1, Ehhadh, and Acaa1a encoding FAO-metabolizing enzymes, when compared with chow-fed mice [8]. Additionally, reduced levels of these mRNAs were also observed due to the treatment of Ppara-null mice with CA or CA/Wy-14643. However, the magnitude of this effect was not as great as that observed after CA treatment of wild-type mice, suggesting the existence of PPARα-independent mechanism(s) by which FXR can affect the expression of PPARα target genes [5, 8]. The suppression of Pgc1a and Hnf4a, via activation of FXR by dietary CA, were also confirmed in a later study [13]. Furthermore, the expression of FAO associated mRNAs, including Acox1 and Acadl, was higher in Fxr-null mice than in control wild-type mice, regardless of whether the observation was conducted in the fed or fasted state [13].

The role of FXR signaling in FAO and the underlying mechanism are far less understood. Among the FXR target genes, intestinal Fgf15 (FGF19 in humans) and hepatic Shp were reported to be associated with FAO. Quantitative proteomic and genetic expression analyses showed that recombinant FGF19 treatment decreased the expression of proteins involved in fatty acid synthesis, i.e., FABP5, SCD1, and ACSL3, and increased the expression of ACOX1, involved in FAO [14]. Overexpression of FGF19 in the intestine resulted in inactivation of CREB and subsequent decrease in PGC1α [10, 15], which is associated with gluconeogenesis [10] and FAO [15, 16]. HNF4α, as part of the HNF4α/HES6 complex, maintains a repressed state of PPARα target genes involved in fatty acid transport and metabolism in fed animals [17]. SHP suppresses PPARγ through HNF4α, and CD36 is regulated by PPARγ, while FXR-SHP was predicted to inhibit the expression of CD36 [11], which subsequently reduces fatty acid uptake for FAO.

Acylcarnitines are biomarkers of mitochondrial FAO function. Incomplete FAO slows down the removal of excess lipids in the cell, resulting in an increase of long- and medium chain but a reduction of short chain acylcarnitines and fatty acids [4, 18, 19]. The current study aimed to explore the potential role of FXR in the regulation of FAO based on integrated analysis of the transcriptome and metabolome. The profiles of hepatic acylcarnitines and fatty acids as well as the expression of corresponding FAO-associated genes were determined in fed wide-type (Fxr^fl/fl), liver-specific Fxr-null (Fxr^ΔHep) and intestine-specific Fxr-null (Fxr^ΔIE) mice, treated with the clinically-used FXR agonist obeticholic acid (OCA), to elucidate the mechanism of FXR-mediated FAO repression. Activation of FXR by OCA treatment was observed to significantly suppress the expression of FAO-associated genes in Fxr^fl/fl mice, resulting in a decrease of acylcarnitine levels and accumulation of long/medium chain acylcarnitines and fatty acids in liver. Fxr^ΔHep mice, and not Fxr^ΔIE mice, were metabolically similar to Fxr^fl/fl mice, suggesting an intestinal FXR-dependent mechanism that may potentially play an important role in FAO repression in fed mice.

2. Materials and methods

2.1. Materials and reagents

OCA (>98.0%) was purchased from MedChemExpress (China). Bile Acid-Carnitine-Sterol Metabolite Library of Standards (IROA Technologies, Boston, USA) and 17 acylcarnitines standards (Sigma-Aldrich, St. Louis, MO, USA) were used for the confirmation of acylcarnitine identification. Fatty Acid (FA) Metabolite Library (IROA Technologies, Boston, USA) and 45 fatty acids reference standards (NU-CHEK PREP, MN, USA) were used for FAs identification. 11 BAs standards and 1 internal standard (Sigma-Aldrich, St. Louis, MO; Toronto Research Chemicals, North York, ON, Canada) were used for qualitative and quantitative analysis of BAs, detailed in the previous study[20]. Recombinant human FGF19 protein was purchased from ABclonal (China).

2.2. Animal studies

Six- to eight-week-old male littermate wild-type (Fxr^fl/fl and Fgf15^fl/fl), hepatocyte-specific Fxr-null (Fxr^ΔHep intestine-specific Fxr-null (Fxr^ΔIE) and intestine-specific Fgf15-null (Fgf15^ΔIE) mice on the C57BL/6N background were randomly assigned to 10 experimental groups including 5 vehicle (VEH) groups and 5 obeticholic acid (OCA) treatment groups, each with at least 9 mice/group. The mice were fed ad libitum a standard chow diet and water, kept in a 12 h light-dark cycle and presented no differences in body weight before treatment. The VEH-treated groups were treated with corn oil and the experimental groups were treated with OCA (20 mg/kg/day, dissolved in corn oil) 2, 24 and 48 h before killing at ZT 6. To determine the contribution of FGF19 to OCA-induced FAO, 30 male wild-type mice (6- to 8-week-old) were equally divided into three groups (including VEH, low FGF 19 dose (80 μg/kg) and high dose (160 μg/Kg) treated groups). Based on the assigned groups, the mice were intraperitoneally injected with FGF19 and VEH 2, 24 and 48 h before killing. After the organ samples were harvested, the intestine was scraped to harvest epithelial cells, and liver, which were then snap-frozen in liquid nitrogen. All samples were stored at −80°C until analysis. All the related animal studies were conducted in accordance with the Institute of Laboratory Animal Resources Guidelines and approved by the National Cancer Institute (NCI) Animal Care and Use Committee.

2.3. Metabolomic analysis

Sample preparation was conducted based on a previously study [21, 22] with some modification. For serum, a 50 μl aliquot was added to a 200 μl mixture of chloroform and methanol (2:1, v/v), vortexed for 5 min and centrifuged at 2000 rpm for 1 min. For tissues, to 50 mg of tissue, a 500 ul mixture of methanol and H₂O (4:3, v/v) was added followed by homogenization. Homogenization was repeated twice (each lasting 20 s) using a bead beater operated at a 60 Hz vibration frequency. After homogenization, chloroform was added with a volume equivalent to 70 μl of the mixture/80 μl of chloroform. The extracts were vortexed for 5 min and centrifuged at 2000 rpm for 1 min. For serum and tissue extracts, after centrifugation, the upper aqueous and lower organic phases were separately transferred into another two tubes as the hydrophilic and hydrophobic extracts and concentrated to dryness using a vacuum concentrator (Thermo Fisher Scientific, USA). The dried hydrophilic and hydrophobic extracts were reconstituted using ACN-H₂O (9:1, v/v) and H₂O-ACN-IPA (1:1:2, v/v/v), filtered using Millex-GV (0.22 μm), and finally stored at −20°C until analysis.

Data acquisition and analysis, and structural identification were carried out according to a previous study [23]. UltiMate 3000 Hyperbaric LC-Q Orbitrap MS system (Thermo Fisher Scientific, USA) equipped with an Acquity CSH C18 column (2.1 × 100 mm i.d., 1.7 μm particle size, Waters, Milford, MA, USA) was used to acquire the MS data using full scan-independent data acquisition (FS-DIA), with positive and negative electro-spray ionizations (ESI) applied for acylcarnitines and FAs, respectively. For acylcarnitines, their identification was based on the match of predicted retention time (RT) and pseudo-characteristic fragmentation ions (CFIs) as descripted in a previous study [23]. Similarly, FAs were identified following the same procedure and the predicted RT are listed in Table S1. Metabolites were confirmed using authentic reference standards. The extracted ion chromatograph (EIC) peak areas of the identified species from the hydrophilic and hydrophobic extracts were used for statistical analysis.

2.3. Bile acid analysis

Analysis of bile acids (BAs) was performed based on a previous study [20–22]. In short, BAs were extracted by adding 400 μl acetonitrile (containing 1 μm d5-TCA) and homogenizing twice (each lasting 20 s) using a bead beater operated at 60Hz of vibration frequency. After extraction, the sample was centrifuged at 2000 rpm for 1 min and the supernatant diluted using acetonitrile (dilution factor: 10) for analysis. ESI (−) full-scan data acquisition was conducted for all samples and calibration standard points (0, 0.1, 0.3, 1, 3 and 10 μmol) using UPLC-QToF (Waters Xevo G2) equipped a Waters Acquity BEH C18 column (2.1 × 100 mm).

2.4. Real-time PCR analysis

RNA was extracted from frozen intestine and liver using TRIzol reagent (Invitrogen, Carlsbad, CA) and then 2 μg used to synthesize cDNA with Superscript II reverse transcriptase (Invitrogen, Carlsbad, CA). Real-Time PCR (Quantstudio 6 Flex, ThermoFisher, US) was conducted following the default standard program suggested by the instrument vendor. Real-time PCR primer sequences are listed in Table S2. Messenger RNA levels were normalized to 18S ribosomal, Gapdh or Actb mRNA and expressed as fold change relative to the control group.

2.5. Data analysis

After Pareto scaling, performed principal component analysis (PCA) was carried out using SIMCA 16.0 (Sartorius Stedim Data Analytics AB, Umeå, Sweden). Student’s unpaired two-tailed t-test was used for two-group comparisons. Spearman correlation was used for correlation analysis. GraphPad Prism (GraphPad Software, Inc., La Jolla, CA, USA) was used to export graphs. All data values were expressed as the mean ± SEM. For fold-change (FC)-based discrimination analysis, the perturbation with FC within ±20% relative standard deviations (RSD) was considered as homeostasis regulation. The data was inputted into MetaboAnalyst (https://www.metaboanalyst.ca/MetaboAnalyst/home.xhtml) for multivariate exploratory ROC analysis and prediction.

3. Results

3.1. FXR activation by OCA inhibited hepatic FAO in WT-mice

In response to OCA treatment, the Fxr^fl/fl mice presented a hepatic acylcarnitine species profile in the PCA scores plot (Fig. 1A) with total acylcarnitines reduced by 31% (Fig. 1B) as compared to VEH-treated Fxr^fl/fl mice. Among the detected 99 acylcarnitine species, 14 species were upregulated with a fold change (FC) ≥1.2, while more than half (50 species, accounting for 51%) were downregulated with a FC ≤ 0.8 (Fig. 1C), indicating that FXR activation by OCA treatment altered hepatic FAO in fed mice. To further investigate the acylcarnitine subclasses contributing to the above difference, FC was estimated based on relative levels. Polyunsaturated, long and medium-chain, unsubstituted and hydroxyl-substituted species were elevated with an FC higher than 1.2, while the dicarboxyl (DC)-substituted species were reduced with an FC lower than 0.8 (Suppl. Fig. S1A) after OCA treatment. The decreases in total acylcarnitines and the relative accumulation of long/medium chain acylcarnitines suggested reduced FAO energy metabolism induced by FXR activation in fed Fxr^fl/fl mice.

Fig. 1. Activation of FXR by OCA treatment results in metabolic perturbation of FAO associated metabolites in Fxr^fl/fl mice.

(A) PCA demonstrating the discriminated hepatic acylcarnitines profile due to OCA treatment. (B) Histogram of reduced levels of total hepatic acylcarnitines in response to OCA treatment. Volcano map visualizing accumulation of long/medium chain acylcarnitines (C) and fatty acids (D) of liver in response to OCA treatment. S/UnS/MunS/PunS_AcCar, Acylcarnitine with saturated/unsaturated/mono-unsaturated/poly-unsaturated acyl chain; LC/MC/SC_AcCar, Acylcarnitine with long (n ≥ 12)/medium(6 ≤ n < 12)/short (n < 6) acyl chain; A/OH/(OH/DC)_AcCar, Acylcarnitine with unsubstituted (namely aliphatic)/hydroxyl-substituted/hydroxyl+dicarboxyl-substituted acyl chain; L/M/SCFA, Long (n ≥ 12) /medium(6 ≤ n < 12)/short (n < 6) carbon chain fatty acid; TFA, Total fatty acid; SFA, Saturated fatty acid; USFA, Unsaturated fatty acid.

3.2. FXR repressed expression of FAO associated genes

To determine whether the OCA-induced metabolic-perturbation in acylcarnitine levels was due to regulation of FAO-related genes, hepatic mRNAs associated with FA transport, β-oxidation and the electron transport chain (ETC) were investigated in fed Fxr^fl/fl mice. FABP and CD36 can facilitate long-chain fatty acid (LCFA) transport across the hepatocyte sinusoidal membrane for subsequent FAO [11, 24], and their downregulation suggested a decrease of LCFA and depressed FAO by FXR activation. After the trafficking, ACSL1 catalyzes fatty acid esterification to form fatty acyl-CoAs, and ACOT1 which in-turn catalyze acyl-CoA conversion to free fatty acid and CoA. Thus, the expression of these two enzymes determines the balance between cytoplasmic concentrations of acyl-CoAs, free CoA and fatty acids [25]. Consistent with the accumulation of the long/medium chain acylcarnitines induced by OCA treatment (Suppl. Fig. S1A), Acsl1 mRNA was decreased while Acot1 mRNA was increased (Fig. 2).

Fig. 2. Activation of FXR by OCA treatment influences the expression of FAO associated mRNAs in Fxr^fl/fl, Fxr^ΔHep, and Fxr^ΔIE mice.

Similarities/differences are visualized using mRNAs expression heatmaps, hierarchical clustering (Distance measure: Euclidean, Clustering algorithm) and t-test analysis (between VEH and OCA groups and between WT and intestinal/hepatic specific Fxr-null groups). For Fxr^fl/fl-VEH mice, their mRNA expression was calibrated as unit concentration of 1.0, and for the rest of mouse groups, the values are expressed as z-scores. The z-scores in Fxr^ΔHep-VEH and Fxr^ΔIE-VEH mice, and OCA treated mice are calculated based on mRNA expression of Fxr^fl/fl-VEH mice and VEH treated mice, respectively. * p < 0.05, ** p < 0.01.

PPARα directly regulates the transcription of genes involved in mitochondrial fatty acid uptake and β-oxidation [5–7]. PGC1α is a coactivator that cooperates with PPARα in the transcriptional control of long-chain FAO by upregulating the expression of several genes in the mitochondrial FAO pathway [26]. HNF4α is also a important transcriptional factor in the regulation of lipid mobilization and FAO [27]. The decreased levels of Ppara, Hnf4a and Pgc1a mRNAs (Fig. 2) were consistent with results obtained using CA-treated Fxr^fl/fl mice [13]. Cpt1 and Cpt2, Cact and Crat mRNAs encode four members of the carnitine acyltransferase family that are important in FAO. Their downregulation results in decreased import of long-chain fatty acids across the mitochondrial membranes for FAO and export of the FAO-produced acetyl-CoA as acetyl-carnitine [28]. The decreased Cd36, Fabp1, Cpt1, Cpt2, Cact and Crat mRNAs were correlated with reduced concentrations of acylcarnitines and accumulation of the long/medium chain acylcarnitines. CPT1 can be inhibited by elevated levels of malonyl-CoA which is formed by the ATP-dependent carboxylation of acetyl-CoA with catalysis of ACACA [28, 29]. Unexpectedly, Acaca mRNA was decreased instead of increased in response to OCA treatment (Fig. 2). This suggests that repression of Cpt1 mRNA was potentially not through the pathway of ACACA-mediated inhibition of malonyl-CoA. As expected, UCP2, the transporter of small molecules across the mitochondrial inner membrane for fatty acid metabolism, was decreased at the mRNA level in this study.

Acadvl, Acadl, Acadm, and Acads mRNAs encode enzymes that catalyze the initial dehydrogenation step of mitochondrial β-oxidation of very-long, long, medium, and short chain fatty acids, respectively. Among these enzyme, long-chain acyl-CoA dehydrogenase (encoded by Acadl) is the main enzyme in FAO and determines the distribution pattern of acylcarnitines and fatty acids [6]. The expression of Acadl and Acadm mRNAs were decreased by OCA treatment in line with the accumulation of long/medium chain acylcarnitines (Fig. 2). Following dehydrogenation, the three subsequent steps in β-oxidation leading to chain shortening and release of acetyl-CoA, are catalyzed by the mitochondrial trifunctional enzyme HADHA/HADHB [6]. In addition to these decreased mRNAs, other mRNAs encoding fatty acid β-oxidation enzymes, including Ehhadh and Acaa1a, were also decreased (Fig. 2). Similarly, MCT1, associated with import of ketone body and ketolysis was decreased at the mRNA (Slc16a) level in response to OCA treatment (Fig. 2). Due to FAO repression, less NADH and FADH₂ are potentially released in mitochondria, which results in less electrons deposited into the electron transport chain (ETC) at complexes I and II, consequently resulting in upregulation of the gene encoding NDUFA9, the largest of the five complexes of ETC (Fig. 2) that maintains ATP hemostasis.

Carnitine accumulates inside the cell and is maintained at a steady level by the high-affinity OCTN2 carnitine transporter in the liver. Increased Octn2 mRNA (Fig. 2) suggests the more OCTN2 is potentially needed for cellular homeostasis. VLDLR is responsible for production of TG-rich lipoproteins [29, 30]. Consistent with a previous study performed in an ob/ob-NASH mice [31], Vldlr mRNA was decreased after OCA treatment in Fxr^fl/fl mice (Fig. 2). The significantly elevated fatty acyl-CoA hydrolysis mRNA (Acot1) and decreased FAO-associated mRNAs (including Cpt1a, Cpt2, Pgc1a, Acaa1a, Acadm, Acsl1, Cact, Hnf4a, Cd36, and Fabp1) in respond to OCA treatment are generally consistent with the reduced total abundance and specific distribution patterns of hepatic acylcarnitine species.

3.3. Specific fatty acid profile

To further investigate the molecular details of the downregulation of FAO induced by OCA treatment, hepatic fatty acids were profiled in fed Fxr^fl/fl mice. Compared with the VEH group, OCA treatment led to an increase in long-chain fatty acids (e.g. FA(22:0), FA(28:8), FA(18:0) and LCFA) and total FAs (FC≥1.2, p<0.05) and a decrease in short and medium chain fatty acids (e.g. FA(8:0), FA(6:0) and FA(7:0) ) (FC≤0.8, p<0.05) (Fig. 1D). The accumulation of long-chain fatty acids was consistent with elevated relative-concentrations of long/medium chain acylcarnitines (Suppl. Fig. S1A), decreased expression of FAO associated mRNAs including Fabp1, Cd36, Cpt1 Cpt2, Acadvl, Acadl, Acadm, Acads and Acsl1, and high expression of Acot1 mRNA (Fig. 2). The accumulation of short- and medium-chain fatty acids could be due to the low-expression of Crat mRNA, encoding CRAT catalyzing the addition or removal of carnitine from medium-and short-chain acyl CoA in the mitochondrial matrix, facilitating the efflux of mitochondrial acetyl-CoA to the cytosol and buffering the intracellular pools of acetyl-CoA and carnitine [28].

3.4. OCA-induced hepatic FAO repression depends on intestinal FXR

In response to OCA treatment, Fxr^ΔHep and Fxr^ΔIE mice all presented reduced hepatic acylcarnitines similar to Fxr^fl/fl mice, suggesting that OCA treatment can potentially induce the perturbation of acylcarnitine-associated metabolism, irrespective of whether the intestinal and hepatic FXR are knocked out. However, the degree of reduction in acylcarnitines levels was different. For the Fxr^ΔHep mice, the percent reduction (up to 31%) in total acylcarnitines level was similar to Fxr^fl/fl mice, but higher than that in the Fxr^ΔIE mice (18%) (Fig. 3B).

Fig. 3. OCA-induced hepatic FAO repression depends on intestinal FXR.

(A) PCA mapping discriminated hepatic acylcarnitine profiles in Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice treated either with VEH or OCA. (B) Histogram of reduced levels of total hepatic acylcarnitines in response to OCA treatment in Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice. Volcano map visualizes metabolic perturbations of FAO associated metabolites- hepatic acylcarnitines (D and E) and fatty acids (F and G) in response to OCA treatment in Fxr^ΔHep (D and F) and Fxr^ΔIE (E and G). Correlation heatmap illustrates similarity/difference in acylcarnitines among Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice (C), as well as the association of fatty acids with the expression of Pgc1a mRNA (H). S/UnS/MunS/PunS_AcCar, Acylcarnitine with saturated/unsaturated/mono-unsaturated/poly-unsaturated acyl chain; LC/MC/SC_AcCar, Acylcarnitine with long (n ≥ 12)/medium(6 ≤ n < 12)/short (n < 6) acyl chain; A/OH/(OH/DC)_AcCar, Acylcarnitine with unsubstituted (namely aliphatic)/hydroxyl-substituted/hydroxyl+dicarboxyl-substituted acyl chain; L/M/SCFA, Long (n ≥ 12)/medium(6 ≤ n < 12)/short (n < 6) carbon chain fatty acid; TFA, Total fatty acid; SFA, Saturated fatty acid; USFA, Unsaturated fatty acid.

The differences of acylcarnitine profiles were visualized in the PCA plot (Fig. 3A). Fxr^ΔHep mice was not discriminated from Fxr^fl/fl mice with some observed overlap with Fxr^fl/fl mice, while Fxr^ΔIE mice were completely separated from Fxr^fl/fl mice (Fig. 3A). Additionally, in Fxr^ΔHep mice, as observed in the Fxr^fl/fl mice, the OCA treatment group was completely separated from the corresponding VEH group in the plot, while in the Fxr^ΔIE mice, the discrimination was inferior with a few observations not completely separated (Fig. 3A). This suggests that the effects of OCA on acylcarnitine metabolism were identical in the Fxr^fl/fl and Fxr^ΔHep mice, but different from that in the Fxr^ΔIE mice. These results were also confirmed by a spearman correlation-based heat map in which more similarity in the acylcarnitines profile was observed between the Fxr^fl/fl and Fxr^ΔHep mice than that between the Fxr^fl/fl and Fxr^ΔIE mice (Fig. 3C).

To further characterize the difference of the individual acylcarnitine species, a volcano map was used revealing that compared with Fxr^ΔHep mice, Fxr^ΔIE mice were markedly different from the Fxr^fl/fl mice with a larger number of acylcarnitine species with FC, either ≤ 0.8 or ≥ 1.2 (Suppl. Fig. S1B&S1C). In response to OCA treatment, Fxr^ΔHep and Fxr^ΔIE mice, as expected, presented the same decreased acylcarnitines as Fxr^fl/fl mice, with most of the detected acylcarnitine species having FC ≤ 0.8 (Fig. 1C, 3D and 3E). However, as noted above, more acylcarnitines were similarity distributed in Fxr^ΔHep mice than in Fxr^ΔIE mice when compared with Fxr^fl/fl mice, suggesting that hepatic FXR has minimal impact on FAO, while intriguingly intestinal FXR has a more important role in the regulation of hepatic FAO.

Fxr^ΔHep mice shared more acylcarnitine profile similarity and concentration with Fxr^fl/fl mice than did Fxr^ΔIE, regardless of OCA or VEH treatment, suggesting that the OCA-suppressed hepatic FAO was largely mediated by intestinal FXR signaling and the corresponding biomarkers included the upregulated AcCar(22:2) and AcCar(22:1-OH), and the downregulated SC_AcCar, saturated_AcCar, AcCar(16-DC+OH), AcCar(16-DC) and AcCar(9:1-DC) (Fig. 1C&3D).

3.5. FAO associated gene expression in Fxr^ΔHep and Fxr^ΔIE mice

The hepatic expression of genes associated with FFA transport and FAO were compared in Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice. For the Fxr^ΔHep mice. Fourteen mRNAs (58%) were increased, notably Cact, Acadm, Acadl, Acaa1a, Slc16a and Hnf4a with statistical significance (p<0.05), and 9 mRNAs (37%) were decreased, including Acot1, Octn2, Vldlr and Cd36 (Fig. 2). In the Fxr^ΔIE mice, the results were reversed; 8 mRNAs (33%) were increased, including Acot1, Slc16a, Ucp2 and Acaca with statistical significance (p<0.05), and 13 (54%) decreased, including Acadvl, Octn2, Acsl1, Crat, Hnf4a and Ppara with statistical significance (p<0.05) (Fig. 2). The mRNAs with the statistically significant difference between Fxr^ΔHep and Fxr^ΔIE were Cd36, Fabp1, Cact, Crat, Acaca, Acaa1a, Acadm, Acadl, Acadvl, Hadha, Acsl1, Ucp2, Acot1, Pgc1a, Ppara and Hnf4a (Fig. 2). For the Acsl1 mRNA, Fxr^ΔIE mice presented statistically decreased expression compared with Fxr^fl/fl and slightly decrease levels in response to OCA (Fig. 2). While for the Fxr^ΔHep mice, the expression of Acsl1 mRNA was similar to that of Fxr^fl/fl mice, either OCA treated or not (Fig. 2). This indicates that the regulation of hepatic Acsl1 is mediated more through intestinal FXR than through hepatic FXR.

In response to OCA treatment, as observed in Fxr^fl/fl mice, Fxr^ΔHep mice presented the same expression of all mRNAs except for increased Vldlr and Cd36 and decreased Octn2 mRNA in Fxr^ΔHep mice (Fig. 2). Results with Fxr^ΔIE, differ from the Fxr^fl/fl mice, which was characterized with different expression for 12 mRNAs (55%) including Cd36, Fabp1, Acot1, Slc16a, Crat, Acaca, Acadvl, Ehhadh, Pgc1a, Hnf4a, Scd1, and Vldlr (Fig. 2). Thus, Fxr^ΔHep mice, instead of Fxr^ΔIE mice, were more similar to control mice in acylcarnitine profiles and corresponding mRNA expression patterns, suggesting that intestinal FXR plays an important role in regulation of hepatic FAO.

3.6. Different fatty acid profiles in Fxr^ΔHep and Fxr^ΔIE mice

Except for the difference in individual fatty acid species, Fxr^ΔHep and Fxr^ΔIE mice presented a similar fatty acid distribution pattern with elevated concentrations for a large number of fatty acid species when compared with Fxr^fl/fl mice (Suppl. Fig. S2A&S2B). When OCA was administered, Fxr^ΔHep presented almost the same fatty acid profile as the Fxr^fl/fl mice, with increased (FC≥1.2, p<0.05) of long-chain fatty acids (e.g. FA(22:0), FA(28:8) and decreased of short and medium chain fatty acids (e.g. FA(8:0), FA(6:0) and FA(7:0)) (Fig. 1D&3F). However, in OCA-treated Fxr^ΔIE mice, the results were reversed as short chain fatty acids were increased and long chain fatty acids were decreased (Fig. 3G). The discriminated fatty acid profiles, especially the accumulation of long-chain FAs, was generally in line with the above-mentioned results - the upregulation of fatty acyl-CoAs hydrolysis to form FFA and CoA associated mRNAs (e.g., Acot1) as well as the decreased levels of fatty acyl-CoAs synthesis (e.g., Acsl1) and FAO (e.g., Cpt1a, Pgc1a and Cact) associated mRNAs. Along with this result, the acylcarnitines presented a reduced level pattern in response to OCA treatment. Additionally, the increase of the Acadvl mRNA in OCA-treated Fxr^ΔIE mice (Fig. 2) resulted in elevated levels of short and medium chain fatty acids and reduced levels of long chain fatty acids (Fig. 3F), contributing to the discrimination of Fxr^ΔIE mice from the Fxr^fl/fl and Fxr^ΔHep mice. More importantly, compared with the Fxr^fl/fl mice, a much better correlation (the negative: TFA, LCFA and SFA, the positive: S/MCFA) between the expression of Pgc1a mRNA and FAs were observed for the Fxr^ΔHep mice, but almost disappeared or reversed in the Fxr^ΔIE mice (Fig. 3H).

3.7. Different roles of hepatic and intestinal FXR signaling on FAO

Compared to Fxr^fl/fl mice, Fxr^ΔHep and Fxr^ΔIE mice have reduced or similar Shp mRNA expression in liver, respectively, whereas Fxr^ΔHep and Fxr^ΔIE mice have similar or reduced Shp and Fgf15 mRNA levels in intestine, respectively, (Fig. 4A&4B). OCA treatment did not alter Shp mRNA levels in liver, but increased Shp and Fgf15 mRNA levels in intestine of Fxr^ΔHep mice when compared with VEH-treated Fxr^ΔHep mice. However, OCA treatment increased Shp expression in liver of Fxr^ΔIE mice but intestinal Shp and Fgf15 mRNA levels were not affected when compared with VEH-treated Fxr^ΔIE mice (Fig. 4A&4B). The intestinal FXR-FGF15 axis and hepatic FXR-SHP axis are responsible for the regulation of hepatic Cyp7a1 expression. Activation of hepatic and intestinal FXR led to decreased Cyp7a1 mRNA levels while disruption of hepatic and intestinal FXR increased Cyp7a1 levels (Fig. 4A). Compared with the action of hepatic FXR-SHP signaling on Cyp7a1 mRNA levels in Fxr^ΔIE mice, intestinal FXR-FGF15 signaling in Fxr^ΔHep mice was more functional in maintaining CYP7A1 homeostasis (Fig. 4A), as OCA treatment repressed the expression of Cyp7a1 mRNA in livers of Fxr^ΔHep mice to similar levels found in Fxr^fl/fl mice, while OCA did not affect Cyp7a1 mRNA levels in livers of Fxr^ΔIE mice which was almost three and six times higher than that of VEH-treated Fxr^fl/fl mice and OCA-treated Fxr^ΔHep mice, respectively (Fig. 4A). Cyp8b1 mRNA presented a similar expression pattern as Cyp7a1 mRNA (Fig. 4A). These findings are generally in line with the hepatic BA pool analysis (Fig. 4D). Fxr^ΔHep and Fxr^ΔIE mice weren’t observed with obviously different BA profiles as Fxr^fl/fl mice due to BAs homeostasis regulation. However, in response to OCA treatment, as the regulation of Cyp7a1 and Cyp8b1 mRNAs, Fxr^fl/fl and Fxr^ΔHep mice presented the almost same BAs profile with the statistically significantly decreased concentrations in different BA classes, and Fxr^ΔIE mice presented the completely different BA profile.

Fig. 4. Hepatic and intestinal FXR signaling differently regulate FAO.

(A) mRNAs relative expression of hepatic FXR targets, Creb and Pgc1a. (B) mRNAs relative expression of intestinal FXR targets-Shp and Pgc1a. (C) Significantly negative correlations between intestinal Fgf15 mRNA and hepatic Creb, Pgc1a and Ppara mRNAs using Spearman analysis in Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice treated either with VEH or OCA. (D) Total concentration of different BA classes. Σ-BAs, total BAs. T-BAs, taurine-conjugated BAs. U-BAs, unconjugated BAs. P-BAs, primary BAs. S-BAs, secondary BAs. * p < 0.05, ** p <0.01, *** p < 0.001.

How FXR in different tissues influences hepatic FAO is less clear. The FXR targets SHP in liver and FGF15/19 in intestine regulate FAO by inhibiting the hepatic CREB-PGC1α pathway [10, 32, 33]. OCA treatment suppressed the hepatic CREB-PGC1α pathway in both Fxr^fl/fl and Fxr^ΔHep mice as revealed by decreased Creb and Pgc1a mRNA levels while suppression was not noted in Fxr^ΔIE mice (Fig. 4A). Compared with Fxr^fl/fl mice, OCA treatment failed to increase Shp mRNA levels in livers of Fxr^ΔHep mice, but still activated intestinal FXR-FGF15 signaling. However, OCA treatment increased hepatic Shp mRNA levels but did not change intestinal Fgf15 mRNA which was maintained at very low levels due to the lack of functional FXR in intestine of Fxr^ΔIE mice compared with Fxr^fl/fl mice (Fig. 4B). These data further implied that hepatic FAO might be repressed via activation of the intestinal FXR/FGF15 pathway. Indeed, negative correlations were observed between the intestinal Shp and Fgf15 mRNAs, and hepatic Pgc1a and Ppara mRNAs in Fxr^fl/fl, Fxr^ΔIE and Fxr^ΔHep mice treated either with VEH or OCA (Fig. 4C).

To further confirm the repression of hepatic FAO is mediated by the activation of intestinal FXR-FGF15 axis by OCA, experiments with human recombinant FGF19 treatment and intestine-specific Fgf15 disruption (Fgf15^ΔIE) mice were conducted. Consistent with the results in Fxr^fl/fl mice, OCA treatment led to a significant accumulation of long and medium-chain acylcarnitines and a reduction of short-chain acylcarnitines in the livers of Fgf15^fl/fl mice, while these alterations were disappeared in Fgf15^ΔIE mice (Fig. 5A). Moreover, the injection of recombinant FGF19 protein resulted in an increase of hepatic long and medium-chain acylcarnitines and a decrease of short-chain acylcarnitines in mice which was similar to OCA treatment. Alternatively, recombinant FGF19 protein was also found to directly upregulate the mRNA expression of Acot1 and a trend to downregulate the expression of Creb and Pgc1a in primary mouse hepatocytes as what was seen in the OCA-treated livers (Fig. 2 and Fig. 4A). Collectively, these data demonstrate the causality between the intestinal FXR/FGF15 pathway and hepatic FAO.

Fig. 5. Activation of intestinal FGF15/19 repressed hepatic FAO.

Heatmap of hepatic acylcarnitines in Fgf15^fl/fl and Fgf15^ΔIE mice treated either with VEH or OCA (A), and in WT mice treated with VEH or FGF19 (80 and 160 μg/Kg/day) (B). Expression of key mRNAs associated with FAO in response to FGF19 treatment in primary mouse hepatocytes(C). * p < 0.05.

3.8. Differences in lipid absorption among Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice

To investigate the differences in uptake of FFAs in intestine of Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice treated either with VEH or OCA, the FFAs in intestine and cecal contents were measured and their relative concentration profiles were visualized using heatmaps (Fig. 6A&6B). OCA treatment reduced FFA absorption in intestine and elevated deposit in cecal contents as revealed by the decreased FFA levels in intestine (Fig. 6A) and increased FFA levels in cecal contents (Fig. 6B). Intestinal FXR disruption diminished the OCA-induced changes in FFA absorption, while hepatic FXR disruption even reversed the trend (Fig. 6A&6B), suggesting that both liver and intestine FXR are involved in the regulation of lipid absorption.

Fig. 6. Differences in uptake of FFAs in intestine in Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice treated either with VEH or OCA.

FFAs in intestine (A) and cecal contents (B) profiled using heatmap, BAs concentration in cecal contents (C), gall bladder (D) and ileum (E) and expression of genes (F) facilitating FFAs uptake in intestine. * p < 0.05, ** p <0.01, *** p < 0.001.

FFAs uptake in intestine was reported to be potentially inhibited by the decreased BA pool size which is mediated by FXR activation [34], and facilitated by proteins such as CD36 and fatty acid binding proteins (FABPs) [35]. First, the BA profiles in liver (Fig. 4D), cecal contents (Fig. 6C), gall bladder (GB, Fig. 6D) and ileum (Fig. 6E) were determined and demonstrated that total BAs were generally decreased in liver and GB in response to OCA treatment in WT and Fxr^ΔHep mice, but not in Fxr^ΔIE mice (Fig. 4D and Fig. 6C–6E). As reported in a previous study [34], Fxr^fl/fl and Fxr^ΔIE mice treated with OCA presented a slightly positive association between FFAs and BAs, respectively. However, OCA-treated Fxr^ΔHep mice and VEH-treated Fxr^ΔHep and Fxr^ΔIE mice did not show this association.

Intestinal expression of Cd36, Fabp1, and Fabp2 mRNAs were elevated in Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice in response to OCA treatment, while the expression of Fabp6 mRNA which was a known intestinal FXR target, was only upregulated by OCA in Fxr^fl/fl and Fxr^ΔHep, but not Fxr^ΔIE mice (Fig. 6F). The expression profile of these genes was generally opposite to that of hepatic Cyp7a1 and partly opposite to that of BAs in liver, cecal contents, GB, and ileum. These results suggest the influence of FXR activation on FFA uptake in intestine is complex and might be mediated by multiple pathways, rather than based on proteins/genes related to FFA transport [35]. It should be noted that the suppression of hepatic BA synthesis by both liver and intestinal FXR signaling, which led to the reduction of BA pool size, is also an important factor in controlling lipid absorption. The alterations of BA profiles could at least partially explain the changes of lipid absorption in response to OCA treatment.

3.9. Multivariate exploratory ROC analysis

All above-mentioned results suggest that the intestinal FXR plays a much more pivotal role in the discrimination of OCA-induced hepatic acylcarnitines than hepatic FXR. A multivariate exploratory ROC analysis was performed using the hepatic acylcarnitine data obtained from Fxr^ΔHep mice. The multivariate ROC (Fig. 7A) curve showed that the OCA treatment group was completely discriminated from the VEH group with 0% false positive rate and 100% true positive rate achieved by support vector machines (SVM) analysis. The top 15 important features (Fig. 7B) obtained in the discrimination analysis, were used as the best biomarkers to build a classifier for the Fxr^fl/fl and Fxr^ΔIE mouse prediction. OCA-treated and VEH-treated Fxr^fl/fl mice were successfully recognized in the prediction, while for the Fxr^ΔIE mice, the VEH treated group was recognized and the OCA failed to be recognized (Fig. 7C). This result confirms that the mechanism of OCA-induced FXR meditation of FAO repression is intestinal-FXR dependent.

Fig. 7. Intestinal FXR has a pivotal role in the discrimination of OCA-induced hepatic acylcarnitines revealed by multivariate exploratory ROC analysis and prediction.

Multivariate ROC curve (A) based support vector machines (SVM) analysis was performed for hepatic acylcarnitines in Fxr^ΔHep mice. The top 15 important features (B) are obtained based on the evaluation of feature importance in Monte-Carlo cross validation (MCCV) and used as the best markers to build a classifier which is used to predict the new samples (C).

4. Discussion

In the current study, most hepatic FAO-related mRNAs were dysregulated in liver of OCA-treated Fxr^fl/fl mice, suggesting that FXR activation could induce the repression of hepatic FAO in fed mice. Along with these genes, the FAO-associated metabolites, including acylcarnitines and fatty acids, were perturbed as revealed by the accumulation of long/medium chain acylcarnitines and long-chain fatty acids, and the reduction of short and medium chain fatty acids in liver. These results, especially the reduced total acylcarnitines in liver, suggested a decreased efficiency of energy metabolism induced by OCA. An earlier study found that PPARα activation by Wy-14643 treatment or prolonged fasting led to FXR suppression in the liver of control mice, but not hepatocyte-specific Ppara-null mice, thus revealing crosstalk between PPARα and FXR through RXRα competition [20]. However, another previous study reported that CA, an FXR agonist, still downregulated FAO associated genes in Ppara-null mice [8], suggesting that FXR-induced FAO repression is potentially independent of PPARα. In the present study, the OCA-induced alterations of acylcarnitine and fatty acid profiles were even more significant in the hepatocyte-specific Fxr-null (Fxr^ΔHep) mice, but was not observed in intestine-specific Fxr-null (Fxr^ΔIE) mice. Additionally, the FFA profiles in intestine and cecal contents observed in Fxr^fl/fl, Fxr^ΔHep and Fxr^ΔIE mice treated with or without OCA suggested that both liver and intestine FXR are potentially involved in the regulation of lipid absorption. These observations indicate that OCA-induced hepatic FAO repression is more likely dependent on the direct action of intestine FXR rather than possible crosstalk between PPARα and FXR in the liver. In contrast to the PPARα-mediated FAO regulation in the fasted state, the current study indicated that the repression of hepatic FAO was mediated by a complicated inter-organ FXR-dependent mechanism in the fed state.

To dissociate the regulatory mechanism of FXR in different tissues on hepatic FAO function, Fxr^ΔIE and Fxr^ΔHep mice were used. The results showed that Fxr^ΔHep mice were more similar to the control Fxr^fl/fl mice than the Fxr^ΔIE mice in the profiles of acylcarnitines, fatty acids, BAs and the expression pattern of FAO-associated genes in the VEH-treated groups. OCA treatment inhibited hepatic FXR in both Fxr^fl/fl and Fxr^ΔHep mice, which was well characterized with the upregulated genes involved in fatty acyl-CoA hydrolysis and the downregulated genes involved in acyl-CoA synthesis and subsequent FAO, and with the reasonable downstream metabolic perturbations noted in profiles of hepatic acylcarnitines and fatty acids. In Fxr^ΔIE mice, the results are opposite, although OCA also repressed FAO to a much lesser extent. These results strongly suggest that the repression of hepatic FAO function by OCA is potentially intestine FXR dependent, which was also confirmed by the hepatic acylcarnitines-based multivariate ROC analysis and prediction. Both hepatic FXR-SHP and intestinal FXR-FGF15/19 pathways were previously reported to mediate FAO via PGC1α [10, 32, 33, 36, 37]. However, the repression of PGC1α expression was maintained in both wild-type and Shp-null mice in response to FGF19 treatment [10]. Thus, FGF15/19, other than SHP is potentially involved the mediating PGC1α repression. FGF15/19 is a postprandial hormone that regulates metabolism in response to nutritional status and its secretion is induced in the small intestine by bile acids acting through FXR in the fed state [38]. The negative correlation between Fgf15 and Pgc1a, as observed in this study, is consistent with previous results that over-release of FGF15/19 in the small intestine could suppress the expression of PGC1α and FAO associated genes including Scd1, Cpt1 and Acc2 in mouse liver [10, 36, 37]. This correlation is also consistent with a previous study where incubating primary hepatocytes with recombinant FGF19 suppressed expression of PGC1β, the homolog of PGC1α and the coactivator that regulates expression of an array of mitochondrial FAO-associated genes [39]. The potential mechanism of PGC1α repression was via CREB inactivation by FGF15/19 [10, 15, 40, 41]. The relationship between intestinal FGF15 and the hepatic CREB-PGC1α pathway was validated based on their mRNA analysis in this study.

PGC1α is a critical activator of several oxidative processes including adaptive thermogenesis and fatty acid oxidation [42, 43]. PGC1α can transduce extracellular stimuli to the transcriptional control of genes involved in cellular energy metabolism and function as a regulator of mitochondrial FAO by serving as a coactivator for PPARα in the transcriptional control of mitochondrial FAO [16, 44]. PGC1α was reported to be involved in the regulation of Cpt1 expression by cAMP in combination with HNF4α and CREB [44, 45]. A recent study showed that PGC1α can induce peroxisomal activity accompanying elevations in the expression of Acadvl, Acadl, Acox1, Crot, and Crat encoding the enzymes responsible for long- and very-long-chain-fatty acid oxidation by a peroxisomal-mitochondrial functional cooperation, as observed in human skeletal muscle cells [46]. Except for oxidative phosphorylation (OXPHOS) and the TCA cycle genes, FAO-related genes were downregulated in mice with adipose-restricted PGC1α deficiency [47]. The Pgc1a-null mice have increased fatty acid accumulation [48]. The positive impact of PGC1α on FA transport (including FAT/CD36, FABPs and FATPs) across plasma and mitochondrial membranes in insulin sensitive tissues was reported [24]. The OCA-induced FAO repression is potentially mediated by activation of the intestinal FXR-FGF15 pathway and the resulting decrease of hepatic CREB-PGC1α (Fig. 8). This conclusion was confirmed by evidence showing that FGF19 supplementation induced the accumulation of long- and medium-chain acylcarnitines in WT mice and OCA-induced long- and medium-chain acylcarnitine accumulation were not found in Fgf15^ΔIE mice.

Fig. 8. Potential mechanism of hepatic FAO repression induced by intestinal FXR activation.

Intestinally secreted FGF15/19 controlled by intestinal FXR signaling represses FAO by repressing hepatic PGC1α via inactivation of hepatic CREB.

Unexpectedly, the FAO repression induced by OCA treatment was still partially present in intestinal Fxr-null mice. The reduced concentrations of acylcarnitines still remained in response to OCA treatment in Fxr^ΔIE mice. More importantly, the Acot1 gene encoding fatty acyl-CoA thioesterase 1 (ACOT1) was elevated and the key FAO-associated gene mRNAs, including Ppara, Cpt1a, Cpt2, Ucp2, Acaa1a, Acadm and Cact, were reduced in response to OCA treatment. In contrast to Fxr^ΔHep and Fxr^fl/fl mice, the Fxr^ΔIE mice had elevated expression of Pgc1a in response to OCA treatment. These results suggested that besides intestinal FXR-FGF15-hepatic CREB-PGC1α axis, additional signaling pathways seemed to exist in the control of FAO repression. The depletion of intestinal FXR in Fxr^ΔIE mice led to a failure in negative feedback in hepatic BA synthesis. As expected, hepatic Cyp7a1 and Cyp8b1 mRNAs in Fxr^ΔIE mice were increased, and more BAs were synthesized in liver and excreted in GB in response to OCA when compared to Fxr^fl/fl mice. The high concentration of hepatic BAs and high expression levels of its corresponding mRNAs, and FAO repression were in line with previous studies showing that BAs including dietary CA can inhibit the expression of PPARα and its target genes in Fxr^fl/fl mice, independent of the presence of FXR [5, 8, 13]. Thus, in addition to the intestinal FXR-FGF15-hepatic PGC1α pathway, the synthesized BAs that directly regulated PPARα signaling also coordinate the FAO in the liver (Fig. 8).

Integrated analysis of the metabolome and transcriptome shows that the potent FXR agonist OCA can repress hepatic FAO in fed mice and the mechanism is potentially involved in two potential signaling pathways. One is the intestinal FXR-FGF15-hepatic CREB-PGC1α signaling cascade, and another is to repress PPARα and its targets via BA activation of nuclear receptors. In addition to the above-discussed mechanism involving FAs accumulation, another potential mechanism involves FXR activation leading to the reduction of CD36 that reduces FA uptake [29]. There are also reports indicating that FXR stimulates lipid metabolism based on a pathway in which FXR activates PPARα signaling [29, 49]. Reduced uptake of dietary lipids in intestine via FXR activation is one of the main contributors to an improvement in NASH patients taking OCA [34]. However, because of the potential for FXR-induced PPARα signaling, FXR-induced FAO activation could also potentially contribute to the improvement in NASH. Although enhanced FAO was reported to contribute to relieve steatosis, FAO repression was also found to alleviate excessive lipid accumulation in steatotic livers [50]. Additionally, a higher hepatic FAO was observed in NASH resulting in increased oxidative injury and NASH progression [29, 51]. Thus, understanding the role of FAO in NASH patients requires additional studies. Additionally, the complementary and overlapping FAO associated pathways controlled by FXR and PPARα need to be further explored. Therefore, the FXR-based molecular mechanisms underlying downregulation of FAO remain elusive and more studies are needed on the mechanism of OCA in NASH treatment.

Highlights:

• Activation of FXR by OCA treatment results in FAO suppression in fed state in mice.

• Fxr^ΔHep mice, instead of Fxr^ΔIE mice responds FAO-associated perturbation in a manner similar to Fxr^fl/fl mice in OCA treatment.

• Intestinal FXR-FGF15 pathway plays critical important role on hepatic FAO suppression.

• Intestinal FXR-FGF15 axis mediates downregulation of hepatic PGC1α by inactivation of hepatic CREB, resulting in FAO suppression.

• Elevated BAs inhibits expression of PPARα and its target genes in Fxr^ΔIE mice.

Abbreviations:

ACAA1A

acetyl-coenzyme A acyltransferase 1A

ACACA

acetyl-CoA carboxylase α

ACADM/L/VL

Acyl-CoA dehydrogenase medium/long/very long chain

ACOT1

acyl-CoA thioesterase 1

ACSL1

acyl-CoA synthetase long-chain family member 1

BAS

bile acids

CACT

carnitine-acylcarnitine translocase

CD36

cluster of differentiation 36

CPT1A

carnitine palmitoyltransferase 1A

CPT2

carnitine palmitoyltransferase 2

CrAT

carnitine acetyltransferase

CREB

cAMP regulatory element-binding protein

CYP7A1

cholesterol 7α-hydroxylase

CYP8B1

sterol 12α-hydroxylase

EHHADH

enoyl-coenzyme A, hydratase/3-hydroxyacyl coenzyme A dehydrogenase

FABP1

fatty acid binding protein 1

FAO

fatty acid β-oxidation

FGF15

mouse fibroblast growth factor 15

FXR

farnesoid X receptor

HADHA

hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit α

HNF4α

hepatic nuclear factor-4α

NDUFA9

NADH dehydrogenase (ubiquinone) 1 α subcomplex 9

OCTN2

Na(+)-dependent carnitine transport by organic cation transporter

PGC1α

peroxisome proliferator-activated receptor gamma, coactivator 1α

PPARα

nuclear receptors-peroxisome proliferator-activated receptor α

OCA

obeticholic acid

SHP

short heterodimer partner

SLC16A

14 members of the monocarboxylate transporter (MCT) family

UCP2

uncoupling protein 2

VLDLR

very-low-density lipoprotein receptor