Genetic ablation of Cyp8b1 preserves host metabolic function by repressing steatohepatitis and altering gut microbiota composition

Abstract

Both type 2 diabetes (T2D) and nonalcoholic steatohepatitis (NASH) are associated with reduced hepatic mitochondrial respiratory capacity. Cholic acid (CA) is the predominant 12α-hydroxylated bile acid that regulates hepatic lipid metabolism, and its circulating levels are negatively correlated with insulin resistance. Abolishing CA synthesis via the genetic disruption of the enzyme sterol 12α-hydroxylase (Cyp8b1^−/−) leads in resistance to diabetes and hepatic steatosis. Here, we show that long-term stimulation of hepatic lipogenesis leads to a severe impairment in overall metabolic and respiratory function in control mice (Cyp8b1^+/+) but strikingly not in Cyp8b1^−/− mice. Cyp8b1^−/− mice are protected from such metabolic impairments associated with T2D and NASH by inhibiting hepatic de novo lipogenic gene and protein expression and altering gut microbiota composition. The protective phenotype is compromised when NASH induction is independent of impairment in de novo lipogenesis (DNL). Consequently, Cyp8b1^−/− mice also show a reduction in hepatic inflammation and fibrosis along with a shift in antimicrobial dynamics in the small intestine. Our data show that the altered bile acid composition of Cyp8b1^−/− mice preserves metabolic and respiratory function by repressing hepatic DNL and driving favorable changes in gut antimicrobial responses.

INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD), is a diabetes-associated comorbid condition characterized by excessive hepatic fat accumulation and is diagnosed as liver fat accumulation exceeding 5% of the liver mass (11). An estimated 70% of type 2 diabetes (T2D) patients also suffer from NAFLD. A persistent imbalance in fatty acid uptake, synthesis, storage, secretion, and oxidation governs hepatic fat accumulation and is implicated in the pathogenesis of NAFLD (1). This imbalance is perpetuated in insulin-resistant states that promote further accumulation of hepatic fat by driving hepatic de novo lipogenesis (DNL) and cholesterol biosynthesis. DNL is a significant source of hepatic fat accumulation, accounting for 30% of the triglycerides (TG) in the liver of NAFLD patients (6, 26).

In 10–30% of cases, benign NAFLD progresses to nonalcoholic steatohepatitis (NASH), a deleterious condition adversely affecting liver function due to inflammation and fibrosis. Of the different lipid classes, toxic levels of cholesterol have been observed in preclinical models as well as in human NASH (8, 13). In fact, NASH can be distinguished from NAFLD by measuring the levels of hepatic cholesterol accumulation, its crystallization, and crown structure formation (14). Excessive intrahepatocytic cholesterol induces hepatic DNL via the activation of the LXR-SREBP1c pathway (9, 31). Recently, direct measurements in NASH patients revealed a 40% reduction in hepatic mitochondrial respiration and a higher incidence of insulin resistance (19). Previous studies have shown that dietary cholesterol induces JNK1-mediated mitochondrial injury, leading to a suppression of State 3 respiration in steatohepatitis (10, 22, 34).

The only two pathways of eliminating hepatic cholesterol are 1) its secretion into bile and 2) its conversion into bile acids (BA). Genetic ablation of the cholic acid (CA) synthesizing enzyme sterol 12α-hydroxylase (Cyp8b1) in mice leads to a lack of the feedback regulation of BA synthesis, leading to an upregulation of the rate-limiting enzyme of BA biosynthesis, cholesterol 7α-hydroxylase (CYP7A1) (20, 24). Increased levels of CYP7A1 and the absence of CA drive the excessive production of α- and β-muricholic acids (MCA), which are specific end products of the C6-hydroxylation of chenodeoxycholic acid (CDCA) that occurs in rodent livers. Until recently, the enzyme responsible for this C6-hydroxylation reaction was not identified. Recent evidence from the work of Takahashi et al. (39) has shown that Cyp2c70 is the rodent C6-hydroxylase that leads to the production of α- and β-MCAs in rodent liver.

We have previously shown that the genetic ablation of Cyp8b1 leads to an improvement in glucose homeostasis consequent to reduced intestinal lipid absorption (18). Recent observations from other groups are indicative of a role of Cyp8b1 in hepatic lipid accumulation (2, 3). However, the pathways involved in Cyp8b1-mediated regulation of host metabolism, especially in pathophysiological conditions, are not well understood. Here, we show that the inhibition of Cyp8b1 affects hepatic lipid accumulation and overall metabolic function by repressing DNL and gut microbiota homeostasis.

MATERIALS AND METHODS

Animals and diets.

We have previously reported the generation and maintenance of Cyp8b1^−/− mice on a C57BL/6NTac genetic background (18). In brief, Cyp8b1^+/− mice were purchased from the University of California, Davis, Knockout Mouse Project (KOMP) repository (Cyp8b1^{tm1(KOMP)Vlcg}). Cyp8b1^+/+ (control) and Cyp8b1^−/− (knockout) mice were obtained by breeding Cyp8b1^+/− mice (18). Our study used male and female mice aged 3–6 mo. The mice had ad libitum access to water and were fed commercially purchased irradiated diets for the indicated durations including Chow diet (TD.10401; Harlan Diets, Madison, WI), the calorically matched chow supplemented with 0.5% wt/wt cholesterol [high-cholesterol diet (HCD), TD10401 + 0.5% cholesterol]; commercially purchased from Harlan Diets, and the methionine choline-deficient diet (MCD, A02082002B; Research Diets, New Brunswick, NJ). Blood for all experiments was drawn from the medial saphenous vein in EDTA-coated capillary tubes. All experiments were approved by the University of British Columbia Animal Care Committee.

Analysis of body composition and energy metabolism.

Lean and fat masses were quantified by EchoMRI-100 (Echo Medical Systems, Houston, TX) with the quantitative magnetic resonance-based method. For simultaneous measurement of multiple metabolic parameters, mice were housed in PhenoMaster metabolic cages (TSE Systems, Bad Homberg, Germany) for 4 consecutive days. All values for V̇o₂, V̇co₂, respiratory exchange ratio (RER), locomotor activity, food and water intake are based on measurements obtained over 60 h after an initial 24-h acclimatization period. RER was calculated as RER = V̇co₂/V̇o₂and energy expenditure (EE) as EE = 3.185 × V̇o₂ + 1.232 × V̇co₂. Locomotor activity was tracked by the total counts of infrared beam breaks. Food and water intakes were monitored through weight sensors.

Biochemical parameters.

Levels of plasma TG, cholesterol, and cholesterol ester (CE) were measured using enzymatic kits according to manufacturer’s protocols (Infinity, Thermo Scientific, Middleton, CA). Levels of lactate dehydrogenase (LDH), were measured from the media after incubation of primary hepatocytes with LPS or TNFα at indicated concentration and time, using the LDH cytotoxicity assay kit according to the manufacturer’s instructions (Thermo Scientific,). The levels of serum 2-aspartate aminotransferase (AST) were determined using commercially available kits from Sigma-Aldrich (St. Louis, MO).

Hepatic and cellular TG, cholesterol, and CE estimation.

TG, cholesterol, and CE concentrations were determined from lipid extracts of liver lysates or cultured cells by using enzymatic kits as mentioned above. Briefly, liver samples from Cyp8b1^+/+ and Cyp8b1^−/− mice were weighed and homogenized in PBS, and lipids were extracted in Folch solvent mixture, chloroform-methanol (2:1 vol/vol), in a volume 20 times the weight of the sample, air-dried, and resuspended in 200 µl of sterile phosphate-buffered saline followed by enzymatic quantification. Cholesterol levels in the extracts were measured using the Amplex Red Cholesterol kit (Invitrogen-Molecular Probes), and TG was measured as above.

Assay for hepatic 4-hydroxyproline levels.

Hepatic levels of 4-hydroxyproline were measured as previously described (30). Results were plotted as milligrams of collagen per gram of liver by use of the following formula: mg collagen = (mg 4-hydroxyproline/ml) × dilution factor × conversion factor (7.5).

Fecal excretion of dietary cholesterol.

Measurement of fecal excretion of dietary cholesterol from fecal pellets collected over 72 h was performed as described previously (43).

Analysis of gut microbiota.

The pyrosequencing and bioinformatics analysis of bacterial 16S rRNA was performed commercially by Microbiome Insights (http://microbiomeinsights.com/; Vancouver, BC, Canada), using samples from the jejunum tissues of Cyp8b1^+/+ and Cyp8b1^−/− mice after 12 wk of HCD diet. A more detailed method is available upon request.

Isolation, culture, and treatment of primary hepatocytes and Kupffer cells.

Hepatocytes were isolated after portal cannulation and collagenase perfusion of the liver as previously published, with slight modifications (21, 25). Briefly, mice were anesthetized using a single terminal dose of Avertin (250 mg/kg), the portal vein was exposed and cannulated, and the inferior vena cava was incised to allow outflow. The liver was perfused serially with Hanks' balanced salt solution followed by Collagenase NB 4G (Serva electrophoresis, Heidelberg, Germany) in low-glucose DMEM and antibiotics (100 U/ml penicillin + 100 µg/ml streptomycin). Followed by low-speed centrifugation, the cell pellet was resuspended in DMEM-F12 supplemented with 10% FBS and filtered through a 70-µm cell strainer, washed, counted, and plated on collagen-coated (5 µg rat tail collagen/cm²) plates at a density of 3.75 × 10³ cells/cm². For the isolation of Kupffer cells (KCs), the nonparenchymal fraction was used. Media, antibiotics, and FBS were purchased from GIBCO, ThermoFisher Scientific (Burlington, ON, Canada).

Real-time PCR.

Total RNA was isolated from tissues using an RNA extraction kit (QIAGEN, Toronto, ON, Canada). One microgram of RNA was used to synthesize cDNA using the Superscript III First-Strand Synthesis kit (Life Technologies). Primer sequences used for measuring gene expression are found in Table 1. Quantitative real-time PCR was performed in an ABI Prism 7700 Sequence Detection System, using SYBR Green PCR Master Mix (Applied Biosystems, Warington, UK).

Table 1.

Primer sequences used for quantitative real-time PCR analysis

Gene Name	Primer Sequences (5′ to 3′)
Acetyl Co-A carboxylase	F- GAT GAA CCA TCT CCG TTG GC R- GAC CCA ATT ATG AAT CGG GAG TG
Fatty acid synthase	F- GGA GGT GGT GAT AGC CGG TAT R- TGG GTA ATC CAT AGA GCC CAG
Sterol Co-A desaturase-1	F- TTC TTG CGA TAC ACT CTG GTG C R- CGG GAT TGA ATG TTC TTG TCG T
Sterol response element binding protein-1	F- GAT GTG CGA ACT GGA CAC AG R- CAT AGG GGG CGT CAA ACA G
HMG- Co-A reductase	F- GAT GAA CCA TCT CCG TTG GC R- GAC CCA ATT ATG AAT CGG GAG TG
24-Dehydrocholesterol reductase	F- GAG GGC TTG GGA TAC TGC AC R- CCT TCA CAG GCC AAA TGG ATG
Sterol O-acyltransferase-2 (Soat2)	F- ACA AGA CAG ACC TCT TCC CTC R- ATG GTT CGG AAA TGT TGC ACC
Sterol response element binding protein-2	F- GCA GCA ACG GGA CCA TTC T R- CCC CAT GAC TAA GTC CTT CAA CT
Tumor necrosis factor-α	F- GAC GTG GAA CTG GCA GAA GAG R- TTG GTG GTT TGT GAG TGT GAG
Interleukin-1β	F- GCA ACT GTT CCT GAA CTC AAC T R- ATC TTT TGG GGT CCG TCA ACT
Interleukin-10	F- GCT CTT ACT GAC TGG CAT GAG R- CGC AGC TCT AGG AGC ATG TG
Collagen type Iα1	F- GCT CCT CTT AGG GGC CAC T R- CCA CGT CTC ACC ATT GGG G
Collagen type IIIα1	F- CTG TAA CAT GGA AAC TGG GGA AA R- CCA TAG CTG AAC TGA AAA CCA CC
TIMP metallopeptidase inhibitor 1	F- GCA ACT CGG ACC TGG TCA TAA R- CGG CCC GTG ATG AGA AAC T
Smooth muscle actin-α2	F- GTC CCA GAC ATC AGG GAG TAA R- TCG GAT ACT TCA GCG TCA GGA
Small heterodimer partner (Shp/Nr0b2)	F- TGG GTC CCA AGG AGT ATG C R- GCT CCA AGA CTT CAC ACA GTG
Bile salt export pump (Bsep/Abcb11)	F- TCT GAC TCA GTG ATT CTT CGC A R- CCC ATA AAC ATC AGC CAG TTG T
Cholesterol 7α-hydroxylase	F- AAA CTC CCT GTC ATA CCA CAA AG R- TTT CCA TCA CTT GGG TCT ATG C
Fibroblast growth factor 15	F- GAA GAC GAT TGC CAT CAA GGA R- CGA ATC AGC CCG TAT ATC TTG C
Regenerating islet-derived 3β	F- ACT CCC TGA AGA ATA TAC CCT CC R- CGC TAT TGA GCA CAG ATA CGA G
Regenerating islet-derived 3γ	F- ATG CTT CCC CGT ATA ACC ATC A R- GGC CAT ATC TGC ATC ATA CCA G

Western blotting.

Proteins from liver homogenates were resolved on self-made acrylamide gels and blotted onto polyvinylidene difluoride membranes. The immunoreactive proteins were detected using antibodies in Table 2 followed by chemiluminescent detection (GE Healthcare, Piscataway, NJ and Thermo Scientific, Rockford IL). Relative densities of protein of interest vs. housekeeping protein calnexin were calculated from the percent area under the curve generated using ImageJ.

Table 2.

Antibodies, stains, and respective usage information

Name and Product Code	Supplier	Dilution
Anti-smooth muscle actin-α2 (αSMA) [ab5694]	Abcam	1:500; PFA fixed no antigen recovery
Anti-fatty acid synthase (FAS) (C20G5) [3180]	Cell Signaling Technology	1:1,000
Anti-sterol 12α-hydroxylase (CYP8B1) [ab191910]	Abcam	1:1,000
Anti-sterol regulatory element binding protein (SREBP)-1 (2A4) [ab3259]	Abcam	1:1,000
Anti-sterol regulatory element binding protein (SREBP)-2 [ab30682]	Abcam	1:1,000
Anti-calnexin, Santa Cruz Biotechnology, (H-70) [sc-11397]	Santa Cruz Biotechnology	1:2,500
Anti-F4/80 antigen [MCA497GA]	AbD Serotec	1:200; PFA fixed no antigen recovery
LipidTOX Green	Molecular Probes	Per manufacturer’s instructions

Histochemistry, immunohistochemistry, and cytochemistry.

Standard histochemistry and immunohistochemistry procedures were applied to 5-µm sections of paraffin-embedded liver tissue sections, using the antibodies listed in Table 2.

Statistical analysis.

Grouped data were compared using two-parameter analyses of variance (2-way ANOVA) followed by Bonferroni's post hoc-test. Comparisons between data sets of unequal “n” were performed using the nonparametric Mann-Whitney test. For comparisons of equal “n”, Student’s t-test was used. All calculations were performed in GraphPad Prism or Excel.

RESULTS

Challenging mice with HCD induces hepatic steatosis within 12 wk.

Typical features of hepatic steatosis developed after a 12-wk challenge with a chow diet supplemented with 0.5% cholesterol (HCD) (Fig. 1, A and B). We observed a significant increase in micro- and macrovesicular steatosis and fibrosis as well as balloon cell formation (Fig. 1, A and B). This dietary regimen also significantly increased the levels of cholesterol in plasma and liver without causing weight gain compared with Chow diet (Fig. 1, C–E). These effects are known to be driven by the activation of the liver X receptor (LXR)-sterol regulatory element binding protein-1c (SREBP1c) pathway in the liver (31). It is also known that dietary cholesterol suppresses mitochondrial State 3 respiration in steatohepatitis via JNK1 (10, 22, 34). Therefore, we challenged Cyp8b1^+/+ and Cyp8b1^−/− mice with the 12-wk HCD regimen and measured various metabolic parameters.

Fig. 1.

Diet supplemented with 0.5% cholesterol (HCD) induces hepatic steatosis in 12 wk. A: photomicrographs of liver cross-sections stained with Mason's Trichrome stain at ×40 and ×200 magnification from 12-wk chow- and HCD-fed Cyp8b1^+/+ mice. B: photomicrograph of a liver cross section stained with Mason's Trichrome stain at ×400 from Cyp8b1^+/+ mice fed HCD for 12 wk and enlarged inset showing characteristic NASH features: arrowhead, microvesicular steatosis; arrow, macrovesicular steatosis; B, balloon cell. C: evolution of body weights over 12 wk of chow or HCD feeding and liver appearances at the end of the 12-wk feeding study in Cyp8b1^+/+ mice. D: plasma cholesterol levels over 12 wk of chow or HCD feeding study in Cyp8b1^+/+ mice. E: hepatic cholesterol levels from liver homogenates of Cyp8b1^+/+ mice fed chow or HCD at the end of the 12-wk feeding study. Values are means ± SE *P < 0.05, **P < 0.01, ***P < 0.001.

Cyp8b1^−/− mice are resistant to metabolic impairment induced by long-term HCD.

At the start of the feeding study, Cyp8b1^−/− mice already had lower body weights compared with Cyp8b1^+/+ mice (Fig. 2A). The rate of increase in body weight on the HCD did not differ between the groups (Fig. 2B), indicating that neither Cyp8b1 deletion nor the HCD (Fig. 2C) alters the rate of increase in body weight over time. The lower body weights of Cyp8b1^−/− mice were accounted for by a 67% (0.4 vs. 1.2g, P = 0.003) and 52% (0.15 vs. 0.30g, P = 0.02) reduction in gonadal and inguinal white adipose mass, respectively, measured after necropsy at the end of the 12-wk study period (Fig. 2C).

Fig. 2.

Absence of cholic acid reduces fat mass and increases energy expenditure. Body weight (A) and percent body weight (B) over time on HCD, organ weights after 12-wk HCD feeding (C), body mass composition (D), metabolic measurements over 60 h during 11th week of HCD food intake (E), oxygen consumption (F), carbon dioxide production (G), respiratory exchange ratio (RER) and histogram showing mean RER (H), and locomotor activity and mean locomotor activity (I) in Cyp8b1^+/+ and Cyp8b1^−/− mice. Values represent means ± SE; n = 3/group; *P < 0.05, **P < 0.01, ***P < 0.001.

In a separate cohort of mice, body composition of individual mice was examined in the 11th week of the HCD regimen, using EchoMRI, after which the animals were housed in metabolic cages for three nocturnal cycles. Body composition analysis revealed a 65% reduction (13 vs. 4.5 g, P = 0.01) in fat mass but a normal lean mass in Cyp8b1^−/− mice (Fig. 2D), confirming the observation of reduced adipose depot weights. Despite the reduced fat mass, food consumption was severalfold higher in Cyp8b1^−/− mice (Fig. 2E). The volume of O₂ inhaled tended to be higher; CO₂ exhaled was significantly higher in the Cyp8b1^−/− mice, especially during the dark period of the circadian cycle, and accounted for a significant elevation in their RER (night:0.9 vs. 0.7, P < 0.01; Fig. 2, F–H). The RER of Cyp8b1^+/+ mice did not show a typical diurnal rhythmicity; that remained low across all circadian phases, suggesting that Cyp8b1^+/+ mice depend typically on fat as a preferred energy substrate in contrast to Cyp8b1^−/− mice (Fig. 2H). Total locomotor activity during the dark period of the circadian cycle was significantly higher (4.7 × 10⁴ vs. 1.7 × 10⁴ counts, P < 0.001) in Cyp8b1^−/− mice (Fig. 2I). These data indicate that Cyp8b1^−/− mice are resistant to deleterious effects such as the blunting in mitochondrial respiration and reduction in locomotor activity seen in Cyp8b1^+/+ mice, indicating that Cyp8b1 deletion prevents the onset of metabolic derangements associated with steatohepatitis.

Cyp8b1^−/− mice are protected from hepatic steatosis in response to HCD.

The 12-wk HCD regimen induced hepatic steatosis and led to a progressive increase in the levels of circulating cholesterol. Interestingly, cholesterol levels remained significantly lower in Cyp8b1^−/− mice starting from week 4 of HCD feeding onward until the end of the 12-wk study (Fig. 3A). Liver weights at the end of the feeding study were 24% lower (0.9 vs. 1.2g, P = 0.006), as shown in Fig. 2C, and plasma AST levels were significantly lower (13.5 vs. 24.0 mU/ml, P = 0.04; Fig. 3B) in Cyp8b1^−/− mice. Hepatic steatosis was prominent in Cyp8b1^+/+ mice with the percentage of cross-sectional liver area covered by steatosis averaging 10.9% compared with only 4% (P = 0.0002) in Cyp8b1^−/− mice (Fig. 3, C and D). The levels of intrahepatic TG, CE, free cholesterol (FC), and total cholesterol (C) measured from hepatic lipid extracts were 56% (53 vs. 23 mg/g, P = 0.04), 42% (2.9 vs 1.7 mg/g, P = 0.009), 40% (2.0 vs. 1.2 mg/g, P = 0.009), and 42% (4.9 vs. 2.9 mg/g, P = 0.007) lower in Cyp8b1^−/− than in Cyp8b1^+/+ mice (Fig. 3, E–H). The area stained by LipidTOX Green was twofold (14.8 vs. 7.3 AUC, P = 0.0005) lower in Cyp8b1^−/− compared with Cyp8b1^+/+ primary hepatocytes, showing reduced intrahepatocyte lipid droplet accumulation (Fig. 3, I and J).

Fig. 3.

Cyp8b1^−/− mice are resistant to HCD-induced steatosis. Biweekly plasma cholesterol over time on HCD (A); plasma 2-aspartate aminotransferase (AST) levels after 12-wk HCD feeding (B) (n = 4–6/group). Hepatic H&E staining and steatosis area (C and D; n = 8–9/group), hepatic triglycerides (E), cholesterol ester (F), free cholesterol (G), total cholesterol (H), and lipid droplets in primary hepatocytes and area (n = 3/group; I and J), Values represent means ± SE; n = 3/group; *P < 0.05, **P < 0.01, ***P < 0.001.

Cyp8b1^−/− mice have suppressed hepatic DNL and reduced intestinal lipid absorption.

Since the upregulation of DNL is the major mechanism by which dietary cholesterol induces steatosis (31), we investigated hepatic lipogenic gene and protein expression in Cyp8b1^+/+ and Cyp8b1^−/− mice. The activation of the master transcriptional regulator of hepatic DNL, SREBP1c and the expression of its downstream lipogenic target genes were examined in Cyp8b1^+/+ and Cyp8b1^−/− mice at the end of the feeding study. Analysis of the maturation and nuclear translocation of SREBP1c revealed a 60% reduction (1.4 vs. 0.5% AUC, P < 0.01) in mature SREBP1c in Cyp8b1^−/− (Fig. 4, A and B). As expected, the protein bands for Cyp8b1 were absent in the liver homogenates of Cyp8b1^−/− mice (Fig. 4A). The reduction in SREBP1c protein levels were accompanied with a 1.8-fold (1.2 vs. 0.7% AUC, P = 0.001) and 1.5-fold (0.9 vs. 0.6% AUC, P = 0.008) reduction in the protein abundance of fatty acid synthase (FAS) and SCD1 in Cyp8b1^−/− liver homogenates compared with those of Cyp8b1^+/+ (Fig. 4, C–F). Additionally, the transcript levels of acetyl-CoA carboxylase-α (Acaca), Fas, and sterol-CoA desaturase 1 (Scd1), three key enzymes involved in DNL, were downregulated (fold change 1 vs. 0.49, 0.43, and 0.61, P = 0.04, 0.04, and 0.02, respectively) in the liver of Cyp8b1^−/− mice (Table 3). The relative transcript levels of Srebp1c were repressed (fold change 1 vs. 0.58, P = 0.04) in Cyp8b1^−/− compared with Cyp8b1^+/+ livers (Table 3). These data indicate that hepatic DNL is downregulated in Cyp8b1^−/− mice. Similar to our previous study as well as in reports from other groups, we also found that the fecal excretion of cholesterol and free fatty acids, measured over 72 h was significantly higher in Cyp8b1^−/− mice (2.8 vs. 7.1 mg/day, P = 0.04 and 1.8 vs. 3.0 mg/day, P = 0.002, respectively; Fig. 4, G and H). Total weight of the dried feces and the weight of the extracted lipids were also higher in Cyp8b1^−/− mice (Fig. 4, I and J). The levels of mature SREBP2 were increased, along with the transcript levels of its target gene Hmgcr, indicating that cholesterol biosynthesis in the liver of Cyp8b1^−/− mice is higher owing to reduced intestinal absorption (Fig. 4, K–L, and Table 3).

Fig. 4.

Repressed HCD-induced lipogenesis in livers of Cyp8b1^−/− mice. Hepatic protein abundance of: sterol regulatory element binding protein-1 (SREBP1; precursor = p, mature = m) and Cyp8b1 and quantification vs. calnexin (A and B), fatty acid synthase (FAS) and quantification against calnexin (C and D), SCD1 and quantification against calnexin (E and F), fecal excretion of cholesterol (G), fatty acids (H), fecal dry weights (I), and lipid extract mass (J) during 11th week of HCD feeding (n = 4/group). K and L: quantification of mature SREBP2 in liver homogenates of Cyp8b1^+/+ and Cyp8b1^−/− mice after 12-wk HCD feeding (n = 5–6/group). Values represent means ± SE. *P < 0.05, **P < 0.01, ***P < 0.001.

Table 3.

Relative transcript levels of indicated genes from livers, unless indicated, of Cyp8b1^−/− compared with Cyp8b1^+/+ mice after 12 wk of HCD feeding

Gene	Fold Change	P Value
Lipogenesis
Acetyl-CoA carboxylase-α	0.49*	0.04
Fatty acid synthase	0.43*	0.04
Sterol CoA desaturase-1	0.61*	0.02
Sterol response element binding protein-1	0.58*	0.04
Cholesterol biosynthesis
HMG- CoA reductase	2.09*	0.01
24-Dehydrocholesterol reductase	0.83	ns
Sterol O-acyltransferase-2	0.56**	0.004
Sterol response element binding protein-2	2.15*	0.02
Inflammation
Tumor necrosis factor-α	0.93	ns
Interleukin-1β	0.49**	0.002
Interleukin-6	1.20	ns
Interleukin-10	1.10	ns
Fibrosis
Collagen type Iα1	0.57*	0.02
Collagen type IIIα1	0.43*	0.01
TIMP metallopeptidase inhibitor 1	0.62*	0.03
Smooth muscle actin-α2	0.66*	0.03
Farsenoid X receptor targets
Small heterodimer partner (Shp/Nr0b2)	0.70*	0.02
Bile salt export pump (Bsep/Abcb11)	0.61*	0.03
Cholesterol 7α-hydroxylase	1.57**	0.007
Sterol 12α-hydroxylase (Cyp8b1)	0.0002***	0.0002
Small heterodimer partner (Shp/Nr0b2) (ileum)	0.20**	0.003
Fibroblast growth factor 15 (ileum)	0.41*	0.012

Values are means ± SE (n = 4/group).

^*P < 0.05,

^**P < 0.01,

^***P < 0.001.

Cyp8b1^−/− mice are not protected from hepatic steatosis and fibrosis in response to MCD diet.

To investigate whether mechanisms other than the repression of DNL could be involved in the protection of Cyp8b1^−/− mice against steatosis, we challenged Cyp8b1^+/+ and Cyp8b1^−/− mice with a MCD diet. The MCD diet induces hepatic steatosis via the suppression of fatty acid β-oxidation and fatty acid uptake and blocking the release of apoB-100 containing very-low-density lipoproteins (VLDL) (32, 33). The 3-wk MCD dietary regimen led to a significant induction in steatohepatitis in both Cyp8b1^+/+ and Cyp8b1^−/− mice. The degree of steatosis was similar in both the groups (Fig. 5, A and B). This was also evident from the levels of hepatic TG that did not differ significantly between the genotypes (Fig. 5C). Body weights between day 10 and day 14 of the 3-wk MCD diet feeding study remained significantly higher (Fig. 5D) in Cyp8b1^−/− compared with Cyp8b1^+/+ mice. However, cumulative weight loss at the end of the study did not significantly differ between the groups (Fig. 5E). Food intake on the MCD diet was also comparable between the genotypes (Fig. 5F). The levels of plasma alanine aminotransferase (ALT) were also similar between the groups (Fig. 5G). However, VLDL release was surprisingly higher (P = 0.005) in Cyp8b1^−/− than in Cyp8b1^+/+ mice (Fig. 5H). Therefore, Cyp8b1 deletion can counter the suppressive effect of MCD on hepatic VLDL release, but this is still insufficient to protect against the build-up of steatosis.

Fig. 5.

Cyp8b1 deletion preserves VLDL release but does not protect from methionine-choline-deficient diet (MCD)-induced hepatic steatosis. Hepatic H&E staining and quantification of steatotic area (A and B), hepatic triglyceride levels (C) in mice fed 3-wk MCD diet (n = 4/group), evolution of body weight during the course of 3-wk feeding study (D), cumulative weight loss at the end of the 3-wk MCD feeding study (E), food consumption on MCD diet measured over 72 h (F), plasma alanine aminotransferase (ALT) levels at the end of the 3-wk MCD diet feeding study (G), and VLDL release on penultimate day of 3-wk MCD diet feeding study (H) in Cyp8b1^+/+ and Cyp8b1^−/− mice. Values are means ± SE *P < 0.05, **P < 0.01.

Hepatic markers for fibrosis and inflammation are reduced in Cyp8b1^−/− mice.

Dietary cholesterol exacerbates inflammation and fibrosis in the liver and is a major dietary factor that can lead to the progression of NAFLD to NASH (44). At the end of the 12-wk HCD feeding study, the extent of liver fibrosis was quantified as a collagen-positive surface area using the Trichrome staining method. Compared with Cyp8b1^+/+ mice, the collagen-positive area in Cyp8b1^−/− mice was significantly lower (68%, P = 0.007) (Fig. 6, A and B). This was accompanied by a reduction in transcript levels of the genes encoding- collagen type Iα1 (Col1a1; 1 vs. 0.57-fold, P = 0.02) and collagen type Iα3 (Col1a3; 1 vs. 0.43-fold, P = 0.01) (Table 3). The mRNA expression of tissue inhibitor of metalloproteinase (Timp)1 was also reduced (1 vs. 0.62-fold, P = 0.03; Table 3). The levels of 4-hydroxyproline in tissues revealed a 32% reduction (2.6 vs. 1.75 mg/g, P = 0.015) in Cyp8b1^−/− liver homogenates (Fig. 6C), confirming the Trichrome staining findings.

Fig. 6.

Cyp8b1^−/− mice have lower hepatic fibrosis. Hepatic Trichrome staining and collagen-positive area after 12 wk HCD feeding (A and B; n = 7–9/group), levels of 4-hydroxyproline expressed as collagen (C; n = 4–5/group), immunohistochemical detection of hepatic α-SMA- positive area (D and E; n = 5/group), levels of various cytokines in liver homogenates of 12-wk HCD-fed Cyp8b1^+/+ and Cyp8b1^−/− mice (F; n = 5/group). Values represent means ± SE. *P < 0.05, **P < 0.01.

An important change that occurs during liver fibrosis is the increase in the expression of α-smooth muscle actin (α-Sma), an activation marker of the fibrogenic stellate cells. Therefore, we measured the mRNA and protein expression of α-Sma at the end of the feeding study. The transcript levels of α-Sma were 34% lower (1 vs. 0.66-fold, P = 0.03) in the livers of Cyp8b1^−/− mice (Table 3). Immunofluorescence staining and quantification measuring the α-SMA-positive area revealed a lower (2.0 vs. 0.8 P = 0.007) α-SMA positivity in the liver cross-sections of Cyp8b1^−/− mice compared with Cyp8b1^+/+ mice (Fig. 6, D and E).

The transcript levels of the proinflammatory cytokine interleukin-1β (Il1β) was significantly lower (1 vs. 0.49-fold, P = 0.002) in Cyp8b1^−/− livers (Table 3). We also measured the protein levels of inflammatory cytokines from liver homogenates and found that IL-1β levels were 15% lower (P < 0.05), whereas IL-10 levels were 21% higher (P < 0.05) in the Cyp8b1^−/− liver homogenates (Fig. 6F).

KCs from HCD-fed Cyp8b1^−/− mice accumulate fewer lipid droplets.

Increased intracellular lipid accumulation in KCs triggers activation of their proinflammatory activity. KCs acquire lipids via the uptake of lipoproteins and their phagocytosis of hepatocyte debris. Since we found that hepatocyte death and circulating cholesterol levels were lower in Cyp8b1^−/− mice, we hypothesized that KCs from Cyp8b1^−/− mice would have reduced lipid accumulation. To measure this, we isolated KCs from the livers of Cyp8b1^+/+ and Cyp8b1^−/− mice. Isolated cells were indeed F4/80-positive phagocytes (Fig. 7A). KCs from Cyp8b1^−/− mice showed reduced lipid droplet accumulation and a 2.2-fold reduction of total cholesterol measured from lipid extracts of isolated KCs (28.9 vs. 12.8, P = 0.03) compared with Cyp8b1^+/+ KCs at the end of the HCD feeding study (Fig. 6, B and C).

Fig. 7.

Reduced lipid accumulation in Cyp8b1^−/− Kupffer cells (KCs). KCs isolated form Cyp8b1^+/+ and Cyp8b1^−/− mice at the end of 12 wk of HCD feeding were F4/80-positive phagocytes (A) stained for lipid droplets (green; B) and their cholesterol levels from extracts (C) (n = 3/group). Values represent means ± SE. *P < 0.05.

Cyp8b1^−/−mice have qualitative and quantitative changes in small intestinal microbiota.

Almost 98% of the bile acids that are secreted in the duodenum are reabsorbed in the terminal ileum; hence, the small intestine is subjected to the highest concentrations of BA. In mice, obstruction of bile flow causes small intestinal bacterial overgrowth, which can be reversed by oral administration of bile acids (12). Supplementing the diet with 2 g/kg CA also leads to phylum-level alterations causing a 41% increase in Firmicutes representing CA as one of the bile acids that alters the composition of gut microbiota (15). Excessive dietary cholesterol has been recently linked to increased gut permeability and inflammasome-dependent intestinal inflammation and submucosal myeloid cell accumulation (27). Therefore, we hypothesized that the absence of CA in Cyp8b1^−/− mice and the reduction in intestinal cholesterol absorption might alter the composition of gut microbiota of Cyp8b1^−/− mice.

Microscopically, we observed submucosal accumulation of infiltrating lymphocytes in the jejunum of Cyp8b1^+/+ mice but not in those of Cyp8b1^−/− mice (Fig. 8A). Analysis of microbiota composition revealed that Cyp8b1^−/− mice have a characteristic gut microbiome which is distinct from that found in Cyp8b1^+/+ mice (Fig. 8, B and C). The Firmicutes-to-Bacteroidetes ratio was higher in Cyp8b1^−/− jejuna as a result of a 28% increase in the levels of Firmicutes and a 55% reduction in the levels of Bacteriodetes (Fig. 8C). The proportions of cholesterol-degrading phyla, Acidobacteria and Actinobacteria, were two and three times higher in Cyp8b1^−/− compared with those found in Cyp8b1^+/+ mice respectively (Fig. 8C). The levels of Deferribacteres in the jejunum of Cyp8b1^−/− mice were 50% lower than those in Cyp8b1^+/+ mice (Fig. 8C). The levels of Proteobacteria, which consist of several endotoxin-producing bacterial genera, were similar in both the groups (Fig. 8C).

Fig. 8.

Microbiota composition and expression of gut anti-microbial peptides are altered in Cyp8b1^−/− mice. Photomicrographs of Cyp8b1^+/+ and Cyp8b1^−/− mice jejunal cross sections stained with Masons’ Trichrome at the end of the 12-wk HCD feeding study. Enlarged insets shows jejunal villi. Yellow asterisks and dotted lines indicate submucosal accumulation of cells at the villus base with densely packed nuclei in the lamina propria (A); dendrogram showing relatedness in microbial ecologies between individual samples of Cyp8b1^+/+ and Cyp8b1^−/− mouse jejuna (B), scale represents arbitrary distance units; pie charts showing percent distribution of major microbial phyla (C) and percent abundance of various microbial genera (horizontal histogram = mean abundance) (D) from the jejunum of Cyp8b1^+/+ and Cyp8b1^−/− mice (n = 3–4/group) after the 12 wk HCD feeding; relative transcript levels of the indicated anti-microbial peptide genes in indicated tissues of Cyp8b1^+/+ and Cyp8b1^−/− mice after the 12 wk HCD feeding (E,F)(n = 3/group); quantification of RegIIIβ protein in ileal homogenates of Cyp8b1^+/+ and Cyp8b1^−/− mice after the 12-wk HCD feeding (G) (n = 3/group). Values represent mean ± SE; *P < 0.05, ** P < 0.01.

At the genus level, we identified an impressive increase in the proportion of Allobaculum (18-fold increase), which accounted for most of the increase seen in the number of Firmicutes in the Cyp8b1^−/− mice (Fig. 8D). The levels of Bifidobacterium were also increased in Cyp8b1^−/− mice sixfold (Fig. 8D). The levels of Lactobacillus, Turicibacter, and Desulfovibrio were lower, whereas those of Sutterella were higher in the jejuna of Cyp8b1^−/− mice. These changes in gut microbiota were accompanied by increased levels of two of the predominant antimicrobial peptide genes, regenerating islet-derived protein-3 (RegIII)β and -γ, that are expressed in the small intestine. The transcript levels of RegIIIβ were significantly upregulated (average 3.8-fold upregulation) in both the jejunum and ileum of Cyp8b1^−/− mice, and RegIIIβ protein levels were significantly (1.6-fold) elevated in the ileum of Cyp8b1^−/− mice (Fig. 8, E, F, and G). The relative mRNA expression of RegIIIγ was also upregulated an average of 3.75-fold in the jejunum and ileum of Cyp8b1^−/− mice (Fig. 8, E and F). These data indicate that the alterations in the microbiome elicit an altered host-microbial dynamic in the gut of Cyp8b1^−/− mice.

DISCUSSION

We show here that the genetic ablation of Cyp8b1 is protective against NAFLD and NASH. The complete absence of CA and the overall increase of MCA in Cyp8b1^−/− mice drives the phenotype characterized by a lower accumulation of hepatic lipids via the inhibition of intestinal cholesterol absorption and by repression of hepatic DNL.

Intracellular cholesterol overload has emerged as a critical determinant of inflammation in different cell types. The observation of cholesterol crystallization in NASH livers and the formation of crown-structures comprising of crystalline cholesterol, gives credence to the crucial pathological role of cholesterol accumulation in NASH (13). One of the most striking observations of our study was the protection from metabolic impairments that was conferred by the genetic deletion of Cyp8b1^−/−. Indeed, excessive levels of free cholesterol accumulate in hepatocyte mitochondrial membranes leading to a reduced mitochondrial respiratory output in the livers of NASH patients (22, 34).

One of the primary effects that excessive intracellular hepatic cholesterol accumulation has is the induction of DNL via the LXR-SREBP1c pathway (31). The induction of DNL provides fatty acid substrates for the formation of cholesterol esters, which reduces the amount of free cholesterol. The same mechanism that leads to a reduction in free cholesterol toxicity leads to a gradual build-up of stored fat in the hepatocytes leading to steatosis (36). The most profound reduction in cholesterol absorption is achieved when rodents are fed with MCA (42). MCA are derivatives of CDCA, produced in the liver of rodents through the recently identified C6-hydroxylase activity of the Cyp2c70 enzyme (39). The complete absence of CA from the bile acid pool of Cyp8b1^−/− mice drives the production of MCAs. Therefore, more hepatic cholesterol is converted into bile acids and less dietary cholesterol is absorbed in Cyp8b1^−/− mice, reducing intracellular cholesterol accumulation in the liver.

Increasing dietary cholesterol can singularly upregulate hepatic DNL enhancing cholesterol ester storage in hepatocytes. Consequently, Cyp8b1^−/− mice have increased SREBP2 maturation, likely reflecting that in Cyp8b1^+/+ mice, the HCD feeding represses endogenous cholesterol biosynthesis. The relative expression levels of the rate-limiting cholesterol biosynthetic enzyme HMGCR, an SREBP2 target gene were also upregulated in Cyp8b1^−/− livers. Interestingly, the co-regulation of both SREBP factors by cholesterol is uncoupled in insulin resistant states (41, 45). We have previously shown that Cyp8b1^−/− mice are insulin sensitive owing to elevated GLP-1 levels (18).

A reduction in intracellular cholesterol levels was not limited to Cyp8b1^−/− hepatocytes but was also observable in Cyp8b1^−/− Kupffer cells (KCs). Cholesterol accumulation in KCs is known to polarize them to the M1 macrophage-like phenotype in which the expression and secretion of IL10 is reduced, whereas that of IL1β is increased (38).

The highly hydrophilic nature of MCAs makes them poor mediators of intestinal cholesterol absorption and they were recently identified as physiological antagonists of the nuclear receptor FXR (35). Importantly, FXR directly regulates the expression of the lipogenic enzyme FAS via an upstream DR1 binding site (23). This indicates that when activated, FXR may directly regulate the synthesis of new fatty acid substrates for esterifying cholesterol via FAS, a possible explanation for the observed increase in the levels of LDL cholesterol (predominantly contains cholesterol esters) upon the administration of FXR agonist, obeticholic acid. However, other groups have also reported that conjugation state of the MCAs dictates binding to FXR in a tissue-specific manner and that the antagonistic activity may also result from FXR-independent mechanisms (17). It is therefore clear that DNL is repressed in the liver of Cyp8b1 knockout mice, but whether MCAs can directly mediate such repression remains a subject of further investigation.

Challenging mice with a MCD diet represses the release of VLDL-lipoproteins leading to steatosis (32). The release of VLDL lipoproteins measured on MCD feeding was significantly higher in Cyp8b1^−/− mice. However, the higher VLDL release was clearly insufficient to exert any consequential reduction in hepatic steatosis and plasma ALT levels in Cyp8b1^−/− mice. These data indicate that inhibition of hepatic lipogenic gene and protein expression and the reduction in intestinal cholesterol absorption are the main mechanisms that may account for the protective phenotype of Cyp8b1^−/− mice.

Depletion of gut microbiota using antibiotics causes an increase in the levels of MCA due to the lack of microbial MCA degradation. Such an increase in MCA levels exerts beneficial effects on liver steatosis and inflammation via the FXR-mediated regulation of intestinal ceramide synthesis (16). The intestine-specific pharmacological activation of FXR through a poorly absorbable FXR agonist, fexaramine, leads to similar beneficial effects on liver steatosis and inflammation (7). A commonality between these two seemingly opposing conditions of FXR modulation that yield similar benefits is the reduction in conjugated and unconjugated levels of CA, which is also common to the genetic ablation of Cyp8b1.

Bile acids are strong modifiers of gut microbial communities. Bile flow into the intestine regulates the number and type of bacteria that colonize the gut and as a result, ligation of the bile duct (BDL) in mice causes small intestinal bacterial overgrowth (12). The decrease in bile acid levels in the intestine of BDL mice is correlated with a decline in gram-positive species and enrichment in LPS-producing, gram-negative species. This overgrowth can be reversed by addition of bile acids to the diet of BDL mice in an FXR and gut barrier-dependant manner (12). Diet–induced obese mice have elevated levels of phylum Deferribacteres that include some pathobionts including mucus-degrading bacteria that are associated with inflammation (29). The levels of Deferribacteres in the intestine of Cyp8b1^−/− mice were reduced suggestive of reduced levels of inflammation-associated microbiota.

We detected characteristic alterations in the gut microbial communities of Cyp8b1^−/− mice. The absence of CA and the relative increase in MCA drive alterations in the microbiota that are similar to long-term calorie restriction (46). The most impacted genus was that of Allobaculum, which accounted for most of the differences seen in phylum Firmicutes. The levels of Allobaculum are reduced upon high-fat diet feeding and have been reported to negatively correlate with body weight and plasma leptin levels and positively correlate with lifespan (28, 37, 46). Interestingly, in rats the levels of MCAs also reduce with age (40). Cyp8b1^−/− mice also carried higher numbers of Actinobacteria and Acidobacteria, the two phyla that contain cholesterol-degrading species and also higher levels of Bifidobacterium which are known to contain cholesterol-lowering probiotic members (4, 5). These alterations may be a consequence of the increased availability of cholesterol to these symbionts within the jejunal lumen of Cyp8b1^−/− mice due to its reduced absorption by enterocytes.

Our data support the view that Cyp8b1 deletion exerts its beneficial effects on metabolic syndrome-associated NAFLD and NASH by inhibiting elevated DNL, reducing intestinal cholesterol absorption and altering intestinal microbiota composition.

GRANTS

This work was supported by a Canadian Institutes of Health Research (CIHR) grant (MOP 106684) to M. R. Hayden and a Natural Sciences and Engineering Research Council (NSERC) grant to B. Vallance. M. R. Hayden is a Killam Professor of Medical Genetics and a Canada Research Chair in Human Genetics at the University of British Columbia. B. Vallance is the Children with Intestinal and Liver Disorders (CHILD) Foundation Research Chair in Pediatric Gastroenterology. J. V. Patankar was supported by a Pfizer-Ripples of Hope postdoctoral fellowship.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.V.P., W.T.G., B.V., and G.N.I. conceived and designed research; J.V.P., C.K.W., and V.M. performed experiments; J.V.P., C.K.W., and V.M. analyzed data; J.V.P., C.K.W., V.M., W.T.G., B.V., and M.H. interpreted results of experiments; J.V.P. and V.M. prepared figures; J.V.P. drafted manuscript; J.V.P., W.T.G., B.V., G.N.I., and M.H. edited and revised manuscript; J.V.P. and M.H. approved final version of manuscript.