Regulation of cholesterol and bile acid homeostasis by the cholesterol 7α-hydroxylase/steroid response element-binding protein 2/microRNA-33a axis in mice


  • Potential conflict of interest: Nothing to report.

  • This work was supported by grants R37DK058379 and R01DK044442 (to J.Y.L.C.) from the National Institute of Diabetes Digestive and Kidney Diseases, National Institutes of Health. J.F. is an awardee of the National Research Service Award (F32 DK096784).

Address reprint requests to: John Chiang, Ph.D., Department of Integrative Medical Sciences, Northeast Ohio Medical University, 4209 State Route 44, Rootstown, OH 44272. E-mail:; fax: 330-325-5910.


Bile acid synthesis not only produces physiological detergents required for intestinal nutrient absorption, but also plays a critical role in regulating hepatic and whole-body metabolic homeostasis. We recently reported that overexpression of cholesterol 7α-hydroxylase (CYP7A1) in the liver resulted in improved metabolic homeostasis in Cyp7a1 transgenic (Cyp7a1-tg) mice. This study further investigated the molecular links between bile acid metabolism and lipid homeostasis. Microarray gene profiling revealed that CYP7A1 overexpression led to marked activation of the steroid response element-binding protein 2 (SREBP2)-regulated cholesterol metabolic network and absence of bile acid repression of lipogenic gene expression in livers of Cyp7a1-tg mice. Interestingly, Cyp7a1-tg mice showed significantly elevated hepatic cholesterol synthesis rates, but reduced hepatic fatty acid synthesis rates, which was accompanied by increased 14C-glucose-derived acetyl-coenzyme A incorporation into sterols for fecal excretion. Induction of SREBP2 also coinduces intronic microRNA-33a (miR-33a) in the SREBP2 gene in Cyp7a1-tg mice. Overexpression of miR-33a in the liver resulted in decreased bile acid pool, increased hepatic cholesterol content, and lowered serum cholesterol in mice. Conclusion: This study suggests that a CYP7A1/SREBP2/miR-33a axis plays a critical role in regulation of hepatic cholesterol, bile acid, and fatty acid synthesis. Antagonism of miR-33a may be a potential strategy to increase bile acid synthesis to maintain lipid homeostasis and prevent nonalcoholic fatty liver disease, diabetes, and obesity. (Hepatology 2013;53:1111–1121)


ATP-binding cassette


acetyl-CoA carboxylase


bile salt export pump


carbohydrate response element-binding protein


coenzyme A


cytochrome P450


cholesterol 7α-hydroxylase


Cyp7a1 transgenic mice


sterol 12α-hydroxylase


endoplasmic reticulum


fatty acid synthase


farnesoid X receptor


high-density lipoprotein


high fat diet


insulin-induced gene




Ingenuity Pathway Analysis


insulin resistance


low-density lipoprotein receptor


liver pyruvate kinase




messenger RNAs


multidrug-resistant protein 3


nonalcoholic fatty liver disease


National Center for Biotechnology Information


sodium taurocholate cotransporting polypeptide




polymerase chain reaction


peroxisome proliferator-activated receptor gamma


SREBP cleavage-activating protein


standard error


small heterodimer partner


steroid response element-binding protein


solute transporter 2a2


3'-untranslated region


Western diet


wild type

Bile acids are synthesized from cholesterol exclusively in the liver.[1] The rate of bile acid synthesis is mainly controlled by transcriptional regulation of cholesterol 7α-hydroxylase (CYP7A1),[1] which encodes the rate-limiting enzyme in the classic bile acid synthesis pathway. When bile acid levels increase, bile acids repress their own synthesis and stimulate biliary lipid secretion. These mechanisms allow the liver to efficiently maintain lipid homeostasis. Bile acids activate farnesoid X receptor (FXR) and the G-protein-coupled receptor, TGR5, and also several cell-signaling pathways to regulate bile acid synthesis and lipid metabolism.[1] Pharmacological activation of either FXR or TGR5 receptor has been shown to improve lipid, glucose, and energy homeostasis, glucose tolerance, and insulin sensitivity.[2, 3] Paradoxically, loss of FXR in obese and diabetic mice reduced body weight and improved peripheral insulin sensitivity,[4] and decreasing bile acid pool size with the specific FXR agonist, GW4064, caused increased susceptibility to diet-induced obesity, fatty liver, and hypertriglyceridemia.[5] It is likely that activation of different bile acid signaling in different mouse models might have different effects on hepatic metabolism, diabetes, and obesity. In Cyp7a1 transgenic (Cyp7a1-tg) mice, both CYP7A1 enzyme activity and bile acid pool size are doubled,[6] biliary cholesterol and bile acid secretion are stimulated, and serum cholesterol is decreased, whereas serum triglyceride levels remain the same.[7] These metabolic changes caused by increased CYP7A1 expression result in significantly improved lipid homeostasis and protection against hepatic steatosis, insulin resistance (IR), and obesity.[6] Therefore, further study is necessary to understand the participation of bile acid synthesis in the regulation of metabolic homeostasis, nonalcoholic fatty liver disease (NAFLD), and diabetes.

Bile acid metabolism is closely linked to whole-body cholesterol homeostasis; bile acid synthesis and bile-acid–facilitated biliary cholesterol secretion are the only significant pathways for cholesterol elimination from the body. Furthermore, the liver acquires cholesterol through dietary absorption, receptor-mediated uptake, and de novo synthesis. Intracellular cholesterol/oxysterols play an important role in the regulation of cholesterol synthesis through the transcriptional factor, sterol response element-binding protein 2 (SREBP2).[8] Upon increased intracellular cholesterol levels, SREBP2 precursor (125 kDa) forms a complex with insulin-induced gene (INSIG) and SREBP cleavage-activating protein (SCAP), which is retained in the endoplasmic reticulum (ER) membrane. When cholesterol levels decrease, SCAP escorts SREBP2 precursor to the Golgi, where two steroid-sensitive proteases (S1P and S2P) cleave an N-terminal fragment (68 kDa), subsequently translocating into the nuclei to activate its target genes, including low-density lipoprotein receptor (LDLR) and key genes involved in de novo cholesterol synthesis.[8]

microRNAs (miRs) are small noncoding RNAs that, after base pairing with complementary sequences of target messenger RNAs (mRNAs), promote mRNA degradation or inhibit protein synthesis. miR-33a, encoded by intron 16 of the SREBP2 gene, has recently been shown to regulate cellular cholesterol homeostasis,[9] biliary bile acid secretion,[10] and fatty acid oxidation.[11] Additionally, when cellular cholesterol levels decrease, miR-33a expression is coinduced with SREBP2 mRNA. miR-33a inhibits ATP-binding cassette (ABC)A1 and ABCG1 to reduce cellular cholesterol efflux. Studies in mice treated with anti-miR-33a or in genetic miR-33a-deficient mice showed miR-33a antagonism induced ABCA1 in macrophages and liver, increased serum high-density lipoprotein (HDL) levels, and promoted macrophage-to-feces reverse cholesterol transport.[12] Additionally, miR-33a antagonism promoted regression of atherosclerosis in mice and nonhuman primates.[13, 14] These studies suggest that miR-33a acts in a synergistic manner with SREBP2 to regulate cellular cholesterol homeostasis.

The aim of this study was to investigate the potential effect of stimulation of bile acid synthesis on hepatic lipid metabolism using Cyp7a1-tg mice as a model. Here, we report that bile acid synthesis plays an important role in integrating intracellular cholesterol sensing and homeostasis by modulating the liver SREBP2/miR-33a axis. Our study suggests the antagonism of miRNA-33a to induce CYP7A1 and bile acid synthesis may be a potential therapeutic approach to treat NAFLD and diabetes.

Materials and Methods

Cyp7a1-tg Mice

Cyp7a1-tg mice overexpressing rat Cyp7a1 complementary DNA under an ApoE3 hepatic control region have been described previously.[6] “Humanized” CYP7A1 mice expressing human CYP7A1 from a BAC clone on a mouse cyp7a1 knockout background were generated as described previously.[15] Mice were maintained under a 12-hour light (6 a.m. to 6 p.m.) and 12-hour dark (6 p.m. to 6 a.m.) cycle. Male wild-type (WT) and Cyp7a1-tg mice were fed chow or Western diet (WD; 42% fat calories, 0.2% cholesterol, Harlan-Teklad 88137; Harlan Teklad, Madison, WI) for 4 months. The local institutional animal care and use committee approved all animal protocols.

Microarray and Pathway Analysis

A MouseRef-8 v2.0 Expression BeadChip kit (BD-202-0202; Illumina, San Diego, CA) was used for microarray analysis. Raw microarray data were log2 transformed and processed with background correction and quintile normalization. Quality control analyses were applied to detect outlier samples. Expression signals with an Illumina detection threshold <0.05 across all samples were used. Linear models and the empirical Bayes method in Limma[16] were used to access differential expression between the control and transgenic groups. Those genes that satisfied the false discovery rate adjusted P value <0.05 or raw P value of <0.001, whichever was more stringent (Benjamini-Hochberg's method), and fold-change threshold of 1.5 were identified for inclusion in the functional pathway and network analysis. Functional profiling of differentially affected biological processes and pathways between transgenic and control mice were evaluated using publicly available tools (e.g., National Center for Biotechnology Information [NCBI], Ensemble, FatiGO, FatiScan, Mouse Genome Informatics, and Kyoto Encyclopedia of Genes and Genomes) and commercial pathway analysis databases, such as Metacore (Metacore; and Ingenuity Pathway Analysis (IPA; Ingenuity Systems; Consensus findings from these tools were used to interpret and understand the biological mechanisms behind the data. Microarray data have been deposited to the NCBI's Gene Expression Omnibus repository (accession no.: GSE38872).

Measurement of Cholesterol and Fatty Acid Synthesis Rate

Mice were fasted for 4 hours and were intraperitoneally (IP) injected with 10 µCi of [1-14C]-sodium acetate (PerkinElmer, Waltham, MA). Mice were sacrificed 30 minutes after injection, and ∼250 mg of liver were rinsed in ice-cold 1× phosphate-buffered saline. Liver tissue was saponified in a 2.2-mL mixture of 50% KOH/95% ethanol (1:10, v/v) at 70°C overnight. 3H-cholesterol (1 µCi) was added to the same tube as a recovery control. Sterols were extracted in 3 mL of hexane, dried, and redissolved in a 300-μL mixture of acetone/ethanol (1:1, v/v). Sterols were then precipitated with 1 mL of digitonin (0.5% in 95% ethanol) overnight at room temperature. Saponified fatty acids were acidified and extracted with petroleum ether.

Measurement of Fecal Sterol Excretion

WT and Cyp7a1-tg mice were IP injected with a single dose of glucose (8 g/kg) containing 5 µCi of [1-14C] glucose (PerkinElmer) as an isotope tracer. Mice were allowed free access to standard chow and water, and feces were collected for 3 consecutive days. Fecal samples were then homogenized in a 5-mL mixture of 50% KOH/95% ethanol (1:10, v/v) at 70°C overnight. 3H-cholesterol (1 µCi) was added to the same tube as a recovery control. Neutral sterols were separated from bile acids by extraction in 6 mL of hexane, dried, and redissolved in a 1-mL mixture of acetone/ethanol (1:1, v/v) and were then precipitated with 3 mL of digitonin (0.5% in 95% ethanol) overnight at room temperature.

Recombinant Adenovirus

Adenovirus-expressing miR-33a was purchased from Applied Biological Materials, Inc. (Richmond, British Columbia, Canada). Adeno-null, which does not express a gene product, was purchased from Vector Biolabs (Philadelphia, PA). Adenovirus was administered at 2 × 109 pfu/mouse through the tail vein. Experiments were carried out 7 days postinjection.

Statistical Analysis

Results were expressed as mean ± standard error (SE), unless noted. Statistical analysis was performed by the Student t test; P < 0.05 indicates statistical significance.


Microarray Gene Expression Profiling in Livers of Cyp7a1-tg Mice Revealed a Tight Link Between Bile Acid Synthesis and Cholesterol Metabolism

To obtain molecular insight into the role of bile acid synthesis in maintaining hepatic lipid homeostasis, we used microarray gene profiling to identify differentially expressed genes in livers of chow-fed and Western high-fat-diet (HFD)-fed Cyp7a1-tg and WT mice. Under chow-fed conditions, 77 genes were identified as differentially expressed with a 2-fold induction or a 50% inhibition, whereas under WD-fed conditions, 144 genes were differentially expressed (Supporting Fig 1A). There were 52 differentially expressed genes that were identified by comparison under both chow-fed and WD-fed conditions. Because the expressions of these 52 genes are likely genotype dependent, but independent of dietary condition, we speculate that some of these genes may be responsible for the improved metabolic homeostasis in Cyp7a1-tg mice. Remarkably, of all 35 up-regulated genes in Cyp7a1-tg mice, most genes are clustered in cholesterol metabolism, with 12 of the top 13 up-regulated genes directly involved in cholesterol biosynthesis, esterification, transport, and regulation (Table 1). IPA identified sterol biosynthesis as the top differentially regulated pathway in Cyp7a1-tg mice, followed by tryptophan metabolism, lipopolysaccharide/interleukin-1–mediated inhibition of retinoid X receptor function, bile acid synthesis, and metabolism of xenobiotics by cytochrome P450 (CYP) (Supporting Table 1). Some of the results were confirmed by quantitative real-time polymerase chain reaction (PCR) analysis. Table 2 shows real-time PCR analysis of expression of key regulatory genes in cholesterol metabolism, bile acid synthesis and detoxification, and fatty acid metabolism in chow-fed and WD-fed WT and Cyp7a1-tg mouse liver. HMG-CoA (coenzyme A) reductase and HMG-CoA synthase gene expression was induced more than 10-fold in chow-fed and HFD-fed Cyp7a1-tg mice, compared to WT mice. Both microarray and real-time PCR detected higher SREBP2 mRNA in Cyp7a1-tg mice (Tables 1 and 2), and mature SREBP2 protein was markedly increased in livers of Cyp7a1-tg mice (Supporting Fig. 2). Other SREBP2-induced genes, such as LDLR, CYP51, and PCSK9, were also induced. Taken together, these data support the activation of a SREBP2-regulated cholesterol metabolic network in Cyp7a1-tg mice. It is well known that SREBP2 maturation is repressed by cholesterol. Consistently, all SREBP2 target genes were down-regulated upon feeding WT mice a cholesterol-rich WD (Tables 1 and 2). Interestingly, WD feeding did not repress induction of cholesterologenic genes in Cyp7a1-tg mice (Tables 1 and 2), suggesting that increasing bile acid synthesis has a dominant positive effect on hepatic cholesterol synthesis.

Table 1. Microarray Analysis of Genes Related to Cholesterol and Fatty Acid Metabolism
  1. n = 4.

  2. a

    P < 0.05 versus WT+C group.

  3. Abbreviations: Tg, Cyp7a1-tg; C, chow.

Table 2. Gene Expression Analysis by Quantitative Real-Time PCR
  1. n = 4.

  2. a

    P < 0.05 versus WT+C group.

  3. Abbreviations: Tg, Cyp7a1-tg; C, chow.

Cholesterol metabolism
SREBP21.00 ± 0.183.10 ± 0.25a0.55 ± 0.04a3.23 ± 0.40a
LDLR1.00 ± 0.272.70 ± 0.27a0.55 ± 0.04a2.40 ± 0.10a
CYP511.00 ± 0.2614.70 ± 1.25a0.27 ± 0.03a19.30 ± 1.27a
HMGCoAR1.00 ± 0.3010.80 ± 2.20a0.36 ± 0.03a13.70 ± 0.50a
HMGCS11.00 ± 0.207.99 ± 2.18a0.77 ± 0.0619.55 ± 1.21a
ABCG81.00 ± 0.301.80 ± 0.20a2.60 ± 0.03a2.70 ± 0.30a
PCSK91.00 ± 0.3015.70 ± 1.60a0.30 ± 0.01a25.70 ± 2.40a
Insig11.00 ± 0.306.90 ± 1.10a0.60 ± 0.02a8.60 ± 1.00a
StarD41.00 ± 0.304.90 ± 0.48a0.30 ± 0.02a5.10 ± 0.57a
CES11.00 ± 0.100.23 ± 0.04a1.17 ± 0.100.40 ± 0.04a
Bile acid synthesis/detoxification
BSEP1.00 ± 0.141.10 ± 0.0502.40 ± 0.30a3.57 ± 0.450a
CYP3A111.00 ± 0.100.84 ± 0.1001.00 ± 0.100.90 ± 0.200
SULT2a21.00 ± 0.201.78 ± 0.270a1.38 ± 0.400.70 ± 0.400
MRP31.00 ± 0.070.70 ± 0.1000.90 ± 0.080.60 ± 0.010a
CYP7A11.00 ± 0.020.04 ± 0.001a0.67 ± 0.10a0.02 ± 0.001a
CYP8B11.00 ± 0.120.05 ± 0.001a0.28 ± 0.01a0.05 ± 0.001a
SHP1.00 ± 0.200.90 ± 0.1602.35 ± 0.73a3.80 ± 1.00a
Fatty acid metabolism
SREBP11.00 ± 0.031.66 ± 0.03a4.53 ± 0.90a1.60 ± 0.30a
FAS1.00 ± 0.063.80 ± 0.30a1.70 ± 0.13a5.60 ± 0.30a
ACC1.00 ± 0.281.90 ± 0.30a0.80 ± 0.141.90 ± 0.20a
ChREBP1.00 ± 0.201.56 ± 0.16a1.96 ± 0.13a0.98 ± 0.19
L-PK1.00 ± 0.101.86 ± 0.15a2.10 ± 0.20a0.90 ± 0.16
CD361.00 ± 0.050.18 ± 0.06a1.02 ± 0.100.24 ± 0.03a
PPARγ1.00 ± 0.150.17 ± 0.03a1.50 ± 0.08a0.20 ± 0.03a
PPARα1.00 ± 0.151.14 ± 0.251.73 ± 0.15a0.92 ± 0.30

Dissociation of Hepatic Bile Acid Metabolism and Lipogenic Gene Expression From Fatty Acid Synthesis in Cyp7a1-tg Mice

In Cyp7a1-tg mice, endogenous mouse CYP7A1 and sterol 12α-hydroxylase (CYP8B1) mRNA levels were decreased as the result of increased bile acid feedback (Table 2). However, FXR target genes small heterodimer partner (SHP), involved in the regulation of bile acid synthesis, and canalicular bile salt export pump (BSEP), involved in bile acid efflux, were not identified by microarray analysis and their mRNA levels were not induced (Table 2). Solute transporter 2a2 (SULT2a1), involved in the efflux of sulfoconjugated xenobiotics and bile acids, was increased in Cyp7a1-tg mice, indicating increased excretion of conjugated bile acids and xenobiotics. Multidrug resistant protein 3 (MRP3, ABCC3), the basolateral efflux transporter of conjugated bile acid expressed under cholestatic conditions, was reduced in hepatocytes of WD-fed Cyp7a1-tg mice (Table 2), consistent with no cholestatic injury in these mice. SREBP1c was induced 66%, much less than SREBP2 in Cyp7a1-tg mice versus WT mice. WD feeding strongly induced SREBP1c in WT mice by 5.5-fold, but only 1.6-fold in Cyp7a1-tg mice. In fatty acid synthesis pathway, a FXR target gene fatty acid synthase (FAS) was strongly induced, but the rate-limiting enzyme acetyl-CoA carboxylase (ACC) was induced only 90% in Cyp7a1-tg mice versus WT mice. However, microarray analysis did not indicate differential expression of any fatty acid synthesis genes, and IPA did not identify fatty acid metabolism as a top regulated pathway. Interestingly, mRNA levels of CD36, a major hepatic fatty acid transporter, were reduced in Cyp7a1-tg. Peroxisome proliferator-activated receptor gamma (PPARγ), involved in the induction of hepatic fatty acid synthesis, was markedly reduced in both chow- and WD-fed Cyp7a1-tg mice. Liver pyruvate kinase (L-PK) and carbohydrate response element-binding protein (ChREBP), involved in lipogenesis, were increased in chow-fed, but decreased in WD-fed, Cyp7a1-tg mice, compared to respective WT mice. These data suggest that reduced free fatty transport to hepatocytes and fatty acid synthesis in hepatocytes may prevent hepatic steatosis in Cyp7a1-tg mice.

Given that induction of hepatic bile acid synthesis in Cyp7a1-tg mice is associated with increased expression of cholesterologenic and lipogenic genes, we injected 14C-labeled sodium acetate to chow-fed WT and Cyp7a1-tg mice to study hepatic fatty acid and cholesterol synthesis rate. As estimated by pmole of 14C-acetate incorporated into fatty acids and sterols, Fig. 1A shows that acetyl-CoA was mainly used for fatty acid synthesis in WT liver. Interestingly, cholesterol synthesis rate was increased ∼12-fold, whereas fatty acid synthesis rate was decreased ∼60% in Cyp7a1-tg mice, resulting in approximately equal incorporation of 14C-acetate into cholesterol and fatty acids.

Figure 1.

Cyp7a1-tg mice have increased cholesterol synthesis and decreased fatty acid synthesis in the liver. (A) Hepatic cholesterol and fatty acid synthesis rates were measured as described in Materials and Methods. Activities of 3H and 14C were determined in a scintillation counter. Cholesterol and fatty acid synthesis rates were expressed as 14C radioactivity derived from 14C-acetate (adjusted by internal recovery standard 3H-cholesterol radioactivity) and incorporated into cholesterol or fatty acids per minute per g of liver tissue. (B) WT and Cyp7a1-tg mice were IP injected with a single dose of glucose (8 g/kg) containing 5 µCi [1-14C] of glucose, and fecal 14C sterol amount was estimated as described in Materials and Methods. Fecal acidic sterols (bile acids) and neutral sterols (cholesterol) derived from 14C-glucose were estimated by measuring the radioactivity of 14C incorporated, adjusted by internal standard 3H-cholesterol radioactivity, and expressed as percentage of total administrated 14C activity. Results are expressed as mean ± SE: *P < 0.05 versus WT; **P < 0.05 versus 14C incorporation into cholesterol in WT mice (n = 4-5).

During the postprandial state, acetyl-CoA derived from glycolysis is used for both lipogenesis and cholesterologenesis. Induction of cholesterol synthesis provides cholesterol substrate to stimulate CYP7A1 activity and bile acid synthesis and, subsequently, stimulates fecal excretion of cholesterol and bile acids. To test the potential contribution of this route to hepatic lipid metabolism, we administered 14C-glucose to mice and measured 14C radioactivity in fecal neutral and acidic sterols. Figure 1B shows that fecal 14C radioactivity in neutral, acidic, and total sterols was markedly and rapidly increased in day 1 in Cyp7a1-tg mice, compared to WT mice. Fecal samples from Cyp7a1-tg mice contained significantly higher 14C radioactivity, accounting for ∼15% of 14C-glucose administered, compared to WT mice feces, which contained only ∼2% of 14C-glucose administered. In addition, the majority of fecal 14C radioactivity was recovered as neutral sterols. Fecal acidic sterols (bile acids) were increased 2-fold in Cyp7a1-tg mice. In summary, these data suggest that in Cyp7a1-tg mice, stimulation of bile acid synthesis may divert acetyl-CoA from fatty acid synthesis to cholesterol synthesis and secretion to decrease lipogenesis.

miR-33a Is Induced in Cyp7a1-tg Mice and Inhibits CYP7A1 mRNA and Bile Acid Synthesis in WT Mice

Because SREBP2 and miR-33a are coexpressed by the SREBP2 gene, we hypothesized that miR-33a might be also induced in Cyp7a1-tg mice to participate in the regulation of cholesterol metabolism. Indeed, hepatic miR-33a expression was coinduced with SREBP2 in Cyp7a1-tg mice under both chow- and WD-feeding conditions (Fig. 2A,B).[9] These results suggest that increasing bile acid synthesis in Cyp7a1-tg mice may induce miR-33a expression by inducing cholesterol-regulated SREBP2 expression.

Figure 2.

Induction of CYP7A1 induces SREBP2 and miR-33a expression in Cyp7a1-tg mice. WT and Cyp7a1-tg mice were fed a standard chow diet or WD for 4 months. mRNA expression of (A) SREBP2 and (B) miR-33a was measured by real-time PCR. Results are expressed as mean ± SE: *P < 0.05 versus WT of same diet (n = 4).

To further test whether miR-33a regulates bile acid metabolism, we used adenovirus-mediated gene delivery to overexpress miR-33a specifically in WT mouse liver (Supporting Fig. 3). mRNA analysis by real-time PCR showed that overexpression of miR-33a reduced the mRNA expression of CYP7A1 and CYP8B1 and Na+-dependent taurocholate cotransport peptide (NTCP), the basolateral bile acid uptake transporter (Fig. 3A). mRNA levels of BSEP, ABCG5, and ABCG8 were also reduced by miR-33a (Fig. 3A). As a positive control, miR-33a inhibited ABCA1 and carnitine palmitoyl-CoA transferase 1 mRNA (Fig. 3A).[9-11] Consistent with down-regulation of CYP7A1 mRNA, miR-33a overexpression reduced microsomal CYP7A1 enzyme activity by ∼40% (Fig. 3B) and total bile acid pool size by ∼25% (Fig. 3C). In addition, miR-33a reduced total serum cholesterol levels by ∼50% (Fig 3D), but increased hepatic cholesterol content by ∼20% (Fig 3E). Such changes in serum and hepatic cholesterol levels are likely resultant from inhibition of both ABCA1 and CYP7A1.

Figure 3.

Effects of hepatic miR-33a overexpression on bile acid and cholesterol metabolism in mice. WT C57BL6J mice were administered adenovirus-expressing miR-33a (Ad-miR-33a) or control adenovirus (Ad-null) by tail vein injection and were sacrificed after 7 days. (A) Hepatic mRNA expression was measured by real-time PCR. (B) Hepatic CYP7A1 enzyme activity. (C) Total bile acid pool size. (D) Serum cholesterol. (E) Hepatic cholesterol. Results are expressed as mean ± SE: *P < 0.05 versus Ad-null controls (n = 4).

To investigate whether miR-33a regulation of CYP7A1 is conserved in the human CYP7A1 gene, we used adenovirus-mediated gene delivery to overexpress miR-33a in “humanized” mice, which express the human CYP7A1 gene. miR-33a overexpression resulted in ∼40% reduction of human CYP7A1 mRNA and ∼25% reduction in total bile acid pool size (Supporting Fig. 4A,C). As a positive control, miR-33a repressed ABCA1 mRNA in the “humanized” CYP7A1 mouse liver (Supporting Fig. 4B). We next transfected miR-33a mimic or miR-33a hairpin inhibitor into HepG2 cells to confirm our in vivo observations. miR-33a mimic dose dependently decreased mRNA levels of CYP7A1, CYP8B1, and ABCA1 (positive control) (Fig. 4A,B). In addition, antagonism of miR-33a by a miR-33a hairpin inhibitor dose dependently increased mRNA levels of CYP7A1, CYP8B1, and ABCA1 (Fig. 4D-F). In summary, these data suggest that miR-33a regulates CYP7A1 and bile acid synthesis and may coordinately regulate hepatic cholesterol and bile acid homeostasis.

Figure 4.

Regulation of CYP7A1 mRNA by miR-33a in HepG2 cells. HepG2 cells were transfected with miRIDIAN miR-33a mimic (A-C) or miRIDIAN miR-33a hairpin inhibitor (D-F) at 50 and 100 nM for 48 hours. Respective controls were used to equally adjust the total amount transfected among samples. mRNA expression was measured by real-time PCR 48 hours after transfection. Assays were performed in triplicate and expressed as mean ± standard deviation: *P < 0.05 versus controls.

Identification of miR-33a Target Sequence in the 3'-Untranslated Region of Human CYP7A1 mRNA

miR-mediated target gene repression commonly occurs through binding to the 3'-untranslated region (3'-UTR) in target mRNAs, which usually results in degradation of the mRNA and inhibition of protein translation. To identify the potential miR-33a target site in the human CYP7A1 3'-UTR, we cloned two human CYP7A1 3'-UTR fragments (nucleotides [nts] 1-200 and nts 203-982) downstream of a luciferase gene in the pMir REPORT luciferase vector and tested the effect of miR-33a mimic on reporter activity. Cotransfection of miR-33a mimic in transient transfection assay of pMir-hCYP7A1 (1-200) reporter in HepG2 cells resulted in ∼40% inhibition of reporter activity (Fig. 5A), but showed no effect on pMir-hCYP7A1 (203-982) reporter activity (Fig. 5B). These results suggest that the nt 1-200 region of the human CYP7A1 3'-UTR may contain a potential miR-33a target site. Analysis of the sequence in this region identified a putative seed-match sequence for miR-33a binding (Fig. 5C). Mutations of this putative seed-match sequence resulted in elevated basal reporter activity and abolished the inhibitory effect of miR-33a mimic on the mutant reporter (Fig. 5A). As a positive control, miR-33a mimic repressed ABCA1 3'-UTR reporter activity, as expected (Fig. 5D). These results suggest that a putative miR-33a-binding site, located in the 3'-UTR of human CYP7A1 mRNA, is functional in mediating the inhibitory effect of miR-33a. In vitro and in vivo studies in both WT and humanized CYP7A1-tg mice showed that this miR-33a-mediated regulatory mechanism is functionally conserved in humans and mice. However, we have not identified a functional miR-33a target site in the mouse cyp7a1 mRNA 3'-UTR.

Figure 5.

Identification of a putative miR-33a target site in human CYP7A1 3'-UTR region. (A) Effect of miR-33a mimic on WT (pMir-hCYP7A1 [1-200]) and mutant (pMir-hCYP7A1 [1-200]) reporter activity in HepG2 cells. (B) Effect of miR-33a mimic on human pMir-hCYP7A1 (203-982) reporter activity in HepG2 cells. (C) Putative miR-33a target site in the human CYP7A1 3'-UTR. (D) Effect of miR-33a mimic on human ABCA1 3'-UTR reporter activity in HepG2 cells. Assays were performed in triplicate and expressed as mean ± standard deviation: *P > 0.05 versus controls.


In this study, we used Cyp7a1-tg mice as an experimental model to demonstrate that stimulating bile acid synthesis significantly affects hepatic lipid metabolism and homeostasis, as well as to elucidate the underlying molecular mechanism for bile acid signaling in preventing diet-induced hepatic steatosis, IR, and obesity. We demonstrate that stimulating de novo bile acid synthesis results in decreased lipogenesis through mechanisms independent of hepatic FXR signaling. This study unveiled complex links between bile acid, cholesterol, and fatty acid metabolism. We also uncovered a novel role for miR-33a in the coordinated regulation of hepatic bile acid and cholesterol metabolism. We found that in response to increased conversion of cholesterol to bile acids, SREBP2 is induced to stimulate cholesterol synthesis to provide a substrate to CYP7A1, and that miR-33a is coinduced to reduce CYP7A1 mRNA translation. This feed-forward activation of CYP7A1 enzyme activity by cholesterol and feedback inhibition of CYP7A1 translation by miR-33a provide a rapid posttranscriptional mechanism for regulation of bile acid synthesis to maintain hepatic lipid homeostasis.

We first showed that a 2-fold to 3-fold stimulation of hepatic CYP7A1 enzyme activity resulted in marked induction of cholesterol synthetic genes and de novo cholesterol synthesis rate in Cyp7a1-tg mice.[6] Stimulation of cholesterol catabolism to bile acids resulted in activation of SREBP2 and all SREBP2-regulated genes in cholesterol metabolism.[17] The ER is a cholesterol-poor organelle,[18] and intracellular cholesterol/oxysterol levels are critical in the regulation of the SREBP2-mediated cholesterol metabolism network. Thus, colocalization of CYP7A1 and SREBP2 in the ER provides advantages for efficient sensing of intracellular cholesterol metabolism, which further suggests that CYP7A1 enzyme activity may play a key regulatory role in cholesterol metabolism. It is well known that cholesterol synthesis increases during the postprandial state to meet increasing demands for cholesterol.[19] We recently reported that feeding rapidly and markedly induced CYP7A1 mRNA expression and increased CYP7A1 enzyme activity by ∼2-fold in mice.[15] Furthermore, CYP7A1 mRNA expression peaked 3 hours after refeeding, whereas HMG-CoA reductase mRNA was minimally affected at 3 hours, but increased by ∼12-fold 6 hours after refeeding.[15] We hypothesize that rapid nutritional induction of CYP7A1 may play a role in stimulating postprandial cholesterol synthesis and lipid homeostasis. Upon food intake, bile acids released into the intestine induce fibroblast growth factor 15, which may be transported to hepatocytes to inhibit bile acid synthesis.[20] This mechanism may reduce CYP7A1 to basal levels after the postprandial period. Postprandial increase in bile acid synthesis is also supported by a recent report that serum bile acid concentrations increased after oral glucose challenge in patients with normal glucose tolerance, but this response was blunted in patients with impaired glucose tolerance.[21] Interestingly, Roux-en-Y gastric bypass rapidly improved IR and glucose tolerance and is associated with higher serum bile acid levels.[22, 23] Reduced bile acid circulation back to the liver in bypass patients may stimulate bile acid synthesis and signaling, which stimulates energy metabolism and glucagon-like peptide 1 to improve insulin sensitivity and reduce weight.

This study unexpectedly revealed that marked induction of CYP7A1 enzyme activity in mouse liver resulted in dissociation of SREBP1c-dependent lipogenic gene expression and hepatic fatty acid synthesis rate. Increased SREBP1c and its targets, FAS and ACC, in Cyp7a1-tg mice (despite a 2-fold to 3-fold enlarged bile acid pool) suggests that CYP7A1 enzyme activity, presumably by modulating cholesterol catabolism, may have a predominant role in SREBP1c maturation over the repressive effect of bile acids on SREBP1c-regulated lipogenesis. Furthermore, these results suggest that reduced hepatic fatty acid synthesis rate in Cyp7a1-tg mice is unlikely a direct result of transcriptional repression of hepatic lipogenic genes by the bile acid/FXR/SHP pathway, as previously reported.[24] Circulating bile acids modulate peripheral energy expenditure, which could indirectly affect hepatic lipogenesis in Cyp7a1-tg mice.[6, 25] Our results here support such a mechanism that stimulation of bile acid and cholesterol synthesis could have a negative effect on de novo lipogenesis by limiting cellular acetyl-CoA availability for fatty acid synthesis. Activation of bile acid and cholesterol synthesis, and decreasing lipogenesis, may result in a metabolic shift that directs acetyl-CoA toward sterol synthesis for biliary excretion and fecal elimination. It should be noted that our study, using Cyp7a1-tg mice as a model, does not necessarily contradict results from other bile-acid–treated experimental models because we have shown that increasing de novo bile acid synthesis did not result in bile acid accumulation in the liver, likely as a result of efficient bile acid secretion.

Finally, this study identified that a novel miR-33a-mediated repression of CYP7A1, as a result of SREBP2 induction, could be part of the feedback loop to reduce bile acid synthesis. The recent discovery of coexpression of SREBP2 and miR-33a, as well as down-regulation of ABCA1 by miR-33a, provided the first evidence that miR-33a down-regulates cellular cholesterol efflux to HDL in response to decreased cellular cholesterol levels to maintain hepatic lipid homeostasis.[9] Our study provides further evidence that miR-33a inhibition of CYP7A1 and bile acid synthesis may also contribute to maintaining cholesterol homeostasis. Cholesterol/oxysterols might also repress miR-33a levels to increase CYP7A1 expression as well as cholesterol efflux transporters.[9] Figure 6 shows a proposed mechanism for the regulation of cholesterol homeostasis by a CYP7A1/SREBP2/miR-33a axis, based on this study, and the well-recognized mechanism for maintaining cholesterol homeostasis and pool by intracellular cholesterol or oxysterol levels.[8] Increased CYP7A1 enzyme activity results in increased cholesterol catabolism and decreased intracellular cholesterol, which leads to proteolytic activation of SREBP2 and subsequent stimulation of de novo cholesterol synthesis and LDLR-mediated cholesterol uptake to reduce serum cholesterol. Simultaneously, SREBP2 activation of its own gene transcription coinduces miR-33a, which down-regulates cholesterol efflux transporters and bile acid synthesis. These changes result in increased intrahepatic cholesterol, which subsequently represses SREBP2 and miR-33a expression. This mechanism integrates bile acids and cholesterol metabolism to control lipid homeostasis at both transcriptional and posttranscriptional levels. Thus, CYP7A1 may play a central role in sensing intracellular cholesterol levels by converting excess hepatic cholesterol to bile acids, thus activating SREBP2 and miR-33a, which inhibits CYP7A1 translation as a rapid feedback mechanism.

Figure 6.

Regulation of cholesterol and bile acid metabolism by the CYP7A1/SREBP2/miR-33a axis. This figure illustrates the coordinated regulation of cholesterol and bile acid metabolism based on this study. Under conditions when cellular cholesterol decreases or cholesterol catabolism increases, SREBP2 is activated. To prevent cellular cholesterol from decreasing, SREBP2 induces LDLR-mediated LDL cholesterol (LDL-C) uptake into hepatocytes. In addition, SREBP2 induces a number of key genes to stimulate de novo cholesterol synthesis. On the other hand, SREBP2 binds to its own gene promoter to induce SREBP2 gene transcription, which also results in coinduction of miR-33a. Increased miR-33a down-regulates ABCA1 and ABCG1 to reduce cellular cholesterol efflux to HDL. Meanwhile, miR-33a inhibits CYP7A1 and bile acid synthesis to inhibit cholesterol catabolism. Under conditions when cellular cholesterol increases, cholesterol represses these SREBP2-miR-33a-regulated pathways, whereas cholesterol feed forward induces CYP7A1 (in mouse) to convert cholesterol into bile acids and to stimulate biliary cholesterol secretion by ABCG5/G8 heterodimers. This feed-forward activation of CYP7A enzyme activity by cholesterol and feedback inhibition of CYP7A1 translation by miR-33a provide a rapid posttranscriptional mechanism for regulation of bile acid synthesis to maintain hepatic lipid homeostasis.

Inducing CYP7A1 activity by targeting miR-33a may be a potential therapeutic approach to improve metabolic homeostasis. This study suggests that the cardioprotective effects of miR-33a antagonism can be attributed not only to stimulating HDL biogenesis, but also bile acid synthesis, the final step in macrophage-to-feces reverse cholesterol transport. In this study, we also showed that mRNA of CYP8B1, NTCP, and BSEP were repressed upon miR-33a overexpression in mice, indicating that miR-33a antagonism also stimulates enterohepatic bile acid circulation. Interestingly, a recent study reported that BSEP and ATP8B1 (a canalicular phospholipid flippase) are targets of miR-33a,[10] and that antagonism of miR-33a increased biliary bile acid and phospholipid secretion in mice. Therefore, miR-33a antagonism may be a potential strategy for increasing bile acid synthesis to treat NAFLD, diabetes, and obesity.


The authors acknowledge microarray analysis by Banu Gopalan and Jie Na (Cleveland Clinic Foundation Genomic Core).