Insights in the regulation of cholesterol 7α-hydroxylase gene reveal a target for modulating bile acid synthesis

Authors

  • Nico Mitro,

    1. Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Facoltà di Farmacia, Università degli Studi di Milano, Milan, Italy
    Current affiliation:
    1. Department of Cell Biology, The Scripps Research Institute, Stein Research Building, Room 212, 10666 North Torrey Pines Road, La Jolla, CA 92037
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Cristina Godio,

    1. Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Facoltà di Farmacia, Università degli Studi di Milano, Milan, Italy
    Current affiliation:
    1. Department of Cell Biology, The Scripps Research Institute, Stein Research Building, Room 212, 10666 North Torrey Pines Road, La Jolla, CA 92037
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Emma De Fabiani,

    1. Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Facoltà di Farmacia, Università degli Studi di Milano, Milan, Italy
    Search for more papers by this author
  • Elena Scotti,

    1. Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Facoltà di Farmacia, Università degli Studi di Milano, Milan, Italy
    Search for more papers by this author
  • Andrea Galmozzi,

    1. Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Facoltà di Farmacia, Università degli Studi di Milano, Milan, Italy
    Search for more papers by this author
  • Federica Gilardi,

    1. Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Facoltà di Farmacia, Università degli Studi di Milano, Milan, Italy
    Search for more papers by this author
  • Donatella Caruso,

    1. Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Facoltà di Farmacia, Università degli Studi di Milano, Milan, Italy
    Search for more papers by this author
  • Ana Belen Vigil Chacon,

    1. Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Facoltà di Farmacia, Università degli Studi di Milano, Milan, Italy
    Search for more papers by this author
  • Maurizio Crestani

    Corresponding author
    1. Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Facoltà di Farmacia, Università degli Studi di Milano, Milan, Italy
    • Laboratorio “Giovanni Galli” di Biochimica e Biologia Molecolare dei Lipidi e di Spettrometria di Massa, Dipartimento di Scienze Farmacologiche, Università degli Studi di Milano, via Balzaretti, 9, 20133 Milan, Italy
    Search for more papers by this author
    • fax: (39) 02-50318391


  • Potential conflict of interest: Nothing to report.

Abstract

The transcription of the gene (CYP7A1) encoding cholesterol 7α-hydroxylase, a key enzyme in cholesterol homeostasis, is repressed by bile acids via multiple mechanisms involving members of the nuclear receptor superfamily. Here, we describe a regulatory mechanism that can be exploited for modulating bile acid synthesis. By dissecting the mechanisms of CYP7A1 transcription, we found that bile acids stimulate the sequential recruitment of the histone deacetylases (HDACs) 7, 3, and 1, and of the corepressor SMRTα (silencing mediator of retinoid and thyroid receptors-α) and the nuclear corepressor. Bile acids, but not the farnesoid X receptor–selective agonist GW4064, increase the nuclear concentration of HDAC7, which promotes the assembly of a repressive complex that ultimately represses CYP7A1 transcription. Interestingly, despite its high basal expression level, small heterodimer partner (SHP) is associated with the CYP7A1 promoter only at a later stage of bile acid repression. Gene silencing with small interfering RNA confirms that HDAC7 is the key factor required for the repression of CYP7A1 transcription, whereas knockdown of SHP does not prevent the down-regulation of CYP7A1. Administration of the HDAC inhibitors valproic acid or trichostatin A to genetically hypercholesterolemic mice increases Cyp7a1 messenger RNA and bile acid synthesis and consequently markedly reduces total plasma and low-density lipoprotein cholesterol. Conclusion: By using a combination of molecular, cellular, and animal models, our study highlights the importance of HDACs in the feedback regulation of CYP7A1 transcription and identifies these enzymes as potential targets to modulate bile acid synthesis and for the treatment of hypercholesterolemia. (HEPATOLOGY 2007.)

Hepatic bile acid biosynthesis is quantitatively the most important metabolic pathway to maintain cholesterol balance in mammals.1 Cholesterol 7α-hydroxylase (CYP7A1) is a liver-specific enzyme that catalyzes the rate-limiting reaction of the so-called “classic” pathway of bile acid synthesis and represents a major checkpoint of cholesterol homeostasis.2–4 CYP7A1 is transcriptionally regulated by bile acids returning to the liver via the portal vein5–7 through multiple mechanisms. In particular, bile acids trigger the farnesoid X receptor (FXR, NR1H4)/small heterodimer partner (SHP, NR0B2) regulatory cascade8, 9 that leads to CYP7A1 repression.10, 11 Additionally, bile acids initiate the c-Jun N-terminal kinase–dependent cascade that ultimately suppresses CYP7A1 expression via FXR-mediated induction of fibroblast growth factor-19 (FGF-19)12 or protein kinase C.13, 14 We also demonstrated an FXR/SHP- independent mechanism of feedback regulation of CYP7A115, 16 whereby bile acids cause the dissociation of coactivators from hepatocyte nuclear factor-4α (NR2A1). Finally, studies in Shp null mice show that bile acids repress gene transcription in a SHP-independent fashion.17, 18 However, the contribution of the FXR/SHP-independent pathway in the regulation of CYP7A1 transcription in response to bile acids is still unclear. The full understanding of these mechanisms may be exploited to develop new therapies that decrease plasma cholesterol and possibly treat patients affected by gallstone disease.

Hypercholesterolemia is a well-known risk factor for atherosclerosis and cardiovascular disease. Several therapeutic approaches have been shown to reduce the rate of morbidity and mortality associated with hypercholesterolemia and atherosclerosis. For example, inhibitors of cholesterol synthesis (statins) and absorption (ezetimibe) have shown several beneficial effects related to their cholesterol-lowering activity.19, 20 However, a more significant decrease of plasma cholesterol levels would be desirable, particularly for patients at high risk of coronary artery disease, such as those with genetic defects in lipoprotein and lipid metabolism (for example, familial hypercholesterolemia).21

It has been shown that adenovirus-mediated overexpression of CYP7A1 in genetically hypercholesterolemic (Ldl-r−/−) mice lowers plasma low-density lipoprotein cholesterol levels,22 indicating that the CYP7A1 gene represents an ideal target for treatment of hypercholesterolemia. However, to date, ion exchange resins are the only agents that can increase bile acid synthesis and, consequently, decrease plasma cholesterol levels.

In this study, we performed a detailed dissection of the events underlying the bile acid–mediated feedback regulation of CYP7A1, which indicates that bile acids repress CYP7A1 transcription rapidly via an FXR-independent pathway, whereas the FXR-dependent regulation occurs only after several hours. In particular, we found that histone deacetylase 7 (HDAC7) is the critical factor for the bile acid–mediated repression of CYP7A1 gene transcription. Finally, we show that valproic acid (VPA) or trichostatin A (TSA), two known and structurally unrelated inhibitors of HDAC activity,23–26 prevent the feedback regulation by bile acids in vitro and de-repress Cyp7a1 in vivo, leading to a dramatic reduction of blood cholesterol in Ldl-r−/− mice.27 Our analysis highlights a complementary mechanism for the suppression of bile acid biosynthesis and identifies HDACs as possible targets for the treatment of hypercholesterolemia.

Abbreviations

CA, cholic acid; CDCA, chenodeoxycholic acid; CYP7A1, cholesterol 7α-hydroxylase; ChIP, chromatin immunoprecipitation; FXR, farnesoid X receptor; FGF-19, fibroblast growth factor-19; HDAC, histone deacetylase; mRNA, messenger RNA; PBS, phosphate-buffered saline; SHP, small heterodimer partner; SMRT-α, silencing mediator of retinoid and thyroid receptors-α; siRNA, short interfering RNA; TSA, trichostatin A; VPA, valproic acid.

Materials and Methods

Reagents and Plasmids.

Chenodeoxycholic acid (CDCA), cholic acid (CA), calyculin A, TSA, and VPA were obtained from Sigma. The GW4064 agonist was kindly donated by Dr. K. Bamberg (Astra-Zeneca, Mölndal, Sweden). The plasmid pcDNA3HA–peroxisome proliferator-activated receptor-γ coactivator 1α was provided by Dr. Anastasia Kralli (The Scripps Research Institute, La Jolla, CA).

Cell Cultures, RNA Extraction, and Quantitative Analysis.

HepG2 cells (American Type Culture Collection, Manassas, VA; no. CRL-10741) were cultured as described.16 HepG2 cells were treated with either 25 μM CDCA, 25 μM CA, 5 μM GW4064, 1.5 mM VPA, 200 nM TSA, or 10 nM calyculin A at the indicated combinations for different times. Total RNA was extracted, and specific messenger RNA (mRNA) was quantitated via real-time PCR, using the Sybr Green kit (Bio-Rad Laboratories, Milan, Italy) as described.16 Primer sequences are available in Table 1. We monitored the specificity of the amplified products by melting curve analysis. Experiments were performed in triplicate and repeated twice with different cell preparations.

Table 1. List of Primers Used for Real-Time Q-PCR Analyses
Sequence nameGene Bank accession numberPrimer sequence
  1. Abbreviations: FOR, forward; REV, reverse.

human CYP7A1 mRNAM93133FOR 5′-AAACGGGTGAACCACCTCTAGA-3′
  REV 5′-AACTCAAGAGGATTGGCACCA-3′
human SHP mRNANM_021969FOR 5′-CTGAGGAAGGCCACTGTCTTG-3′
  REV 5′-GCTATGTGCACCTCATCGCA-3′
human Apo C-III mRNANM_000040FOR 5′-TTCTCTGAGTTCTGGGATTTGGA-3′
  REV 5′-GGAGCTCGCAGGATGGATAG-3′
human NTCP mRNAL21893FOR 5′-CGTCTGCCCCATTCAACTTC-3′
  REV 5′-GAACAACATGAACACCAGGATGA-3′
human Apo A-I mRNANM_000039FOR 5′-GCATTTCTGGCAGCAAGATGA-3′
  REV 5′-ACAGTGGCCAGGTCCTTCACT-3′
human HDAC7 mRNAAF239243FOR 5′-TAGCTCACGAGAAGCCACTTG-3′
  REV 5′-ATGAGCTTCATTCCTCCAATGC-3′
human HDAC1 mRNANM_004964FOR 5′-ACGGGATTGATGACGAGTCCT-3′
  REV 5′-AGTCTGAGCCACACTGTAAGACCA-3′
human HDAC2 mRNANM_001527FOR 5′-TTGGAGGAGGTGGCTACACAA-3′
  REV 5′-ACAATCAAGGGCAACTGCAGT-3′
human HDAC3 mRNANM_003883FOR 5′-TTGAGTTCTGCTCGCGTTACA-3′
  REV 5′-GCCCAGTTAATGGCAATATCACA-3′
human FGF19NM_005117FOR 5′-TCGGAGGAAGACTGTGCTTTC-3′
  REV 5′-CTTCTCGGATCGGTACACATTG-3′
human FXR mRNANM_005123FOR 5′-AACATAGCTTCAACCGCAGACC-3′
  REV 5′-GAAATGGCAACCAATCATGTACA-3′
mouse Cyp7a1 mRNANM_007824FOR 5′-ATCAAAGAGCGCTGTCTGGGT-3′
  REV 5′-GCGTTAGATATCCGGCTTCAAAC-3′
mouse Cyp8b1 mRNAAF090319FOR 5′-TTATTCGGCTACACCAAGGACA-3′
  REV 5′-CAAATCGACGGAACTTCCTGA-3′
mouse Shp mRNANM_011850FOR 5′-AAGGGCACGATCCTCTTCAA-3′
  REV 5′-CCAGGGCTCCAAGACTTCAC-3′
mouse Fxr mRNANM_009108FOR 5′-CACGAAGATCAGATTGCTTTGC-3′
  REV 5′-CCGCCGAACGAAGAAACAT-3′
mouse Hnf-4 mRNAD29015FOR 5′-CAAGAGGTCCATGGTGTTTAAGG-3′
  REV 5′-CGGCTCATCTCCGCTAGCT-3′
mouse Apo A-I mRNANM_009692FOR 5′-CTGGGACACTCTGGGTTCAAC-3′
  REV 5′-CACCCAATCTGTTTCTTTCTCCA-3′
mouse Hmg-CoA red mRNANM_008255, XM_127496FOR 5′-GCACCATGCCATCGATAGAGA-3′
  REV 5′-CACATACAATTCGGGCAAGCT-3′
mouse Apo B mRNAXM_137955FOR 5′-GACTGTGTCTGATTTCCCCTCAAT-3′
  REV 5′-TCCTCATCCTCCAGTCCTGAA-3′
mouse L-CptI mRNANM_013495FOR 5′-CCTTCAGCGAGTAGCGCATAG-3′
  REV 5′-GCATGATTGCAAAGATCAATCG-3′
mouse Lxr α mRNAAJ132601FOR 5′-GCCTCTCTACTTGGAGCTGGTC-3′
  REV 5′-GAGCCATGAATGAGCTGCAA-3′
mouse Lxr β mRNANM_009473, XM_358686FOR 5′-CGAGCTAGCCATCATCTCGG-3′
  REV 5′-AACTGCAAGAACCCTGGCAC-3′
mouse Ftf mRNANM_030676FOR 5′-CCTCCTGAGTCTCGCACAGG-3′
  REV 5′-CAAACTCCCGCTGATCGAAC-3′
mouse Hdac7 mRNANM_019572FOR 5′-TGTTGACTGGGATGTTCACCA-3′
  REV 5′-GGGAAATGTAGAGCACACTGGG-3′
human CYP7A1 BAREL13460FOR 5′-CTGTTGTCCCCAGGTCCG-3′
  REV 5′-AAGTGGTAGTAACTGGCCTTGAACT-3′
human CYP7A1 TATA boxL13460FOR 5′-CTCTGATTAGAAAGGGAAGGATGC-3′
  REV 5′-AATGGCTAATTGTTTGCTTTGTCA-3′
human CYP7A1 upstreamL13460FOR 5′-CATAATTCAGTCACCTCCTACCAGG-3′
  REV 5′-AGGCATGGTAGTGTGACATGGTT-3′
human CYP7A1 3′-untranslated regionM93133FOR 5′-TTAGCTACTCGAGAGGCCAAAGAA-3′
  REV 5′-CCTTAGGAAAAAACAATCTGCCAAT-3′
mouse/human 18S rRNAX03205, X00686FOR 5′-CGGCTACCACATCCAAGGAA-3′
  REV 5′-CCTGTATTGTTATTTTTCGTCACTACCT-3′

Chromatin Immunoprecipitations.

Chromatin immunoprecipitation (ChIP) assays were performed as described.16 Immunoprecipitations were performed with antibodies against hemagglutin (HA)-tag (Roche) to immunoprecipitate the transfected HA-tagged peroxisome proliferator-activated receptor-γ coactivator 1α (PGC-1), whereas the following antibodies were used to immunoprecipitate the endogenously expressed proteins: cyclic adenosine monophosphate response element binding–protein binding protein (CBP), acetyl-H3, acetyl-H4 (Upstate Biotechnology, Charlottesville, VA), HDAC1 to HDAC8, RNA polymerase II, SHP, silencing mediator of retinoid and thyroid receptors-α (SMRTα), and nuclear corepressor N-CoR (Santa Cruz Biotechnology, Santa Cruz, CA). Specificity of antibodies used for ChIP assays was assessed by western blotting. Specific genomic DNA regions were quantitated via real-time PCR.16 Primer sequences are available in Table 1. Each time point was replicated in three separate dishes. To check the specificity of immunoprecipitated DNA regions, we also amplified distal regions, which confirmed lack of interaction with the analyzed nuclear factors. The specificity of amplified regions was assessed by performing a melting curve at the end of each real-time PCR reaction. Data are expressed as fold above background and were calculated according to the following formula: (Ct ChIP − Ct Input) at tn − (Ct no antibody − Ct Input) at tn = ΔΔCt Recruitment (fold above background) = 2−ΔΔCt, where “Ct ChIP” is the threshold cycle obtained with samples immunoprecipitated with a given antibody at the indicated time (tn), “Ct Input” is the value obtained with total genomic DNA, and “Ct no antibody” refers to the controls with no antibody added for each immunoprecipitation as a background value. The background levels from a control antibody (anti-mouse IgG) was comparable to the values obtained with no antibody.

Immunocytochemistry.

HepG2 cells were cultured on coverslips with or without 25 μM CDCA. Cells were fixed, washed, permeabilized, and blocked with phosphate-buffered saline (PBS) containing 10% serum, 1% bovine serum albumin, and 0.5% Tween 20 for 30 minutes at room temperature. Cells were incubated with a polyclonal antibody against HDAC7 (diluted 1:20 in PBS containing 1% serum) overnight at 4°C. The Alexa Fluor 488 goat anti-rabbit IgG (diluted 1:500 in PBS containing 1% serum) (Invitrogen) was added as a secondary antibody for 1 hour at room temperature. Alexa Fluor 546 phalloidin (diluted 1:200 in PBS) and TOPRO-3 iodide (diluted 1:500) (Invitrogen) were added for 15 minutes at room temperature to stain the cytoskeleton and nucleus, respectively. Coverslips were mounted and analyzed with a confocal microscope (Nikon Eclipse TE2000-S, Radiance 2100, Bio-Rad Laboratories). Image acquisition was performed with Laser Sharp software (Bio-Rad Laboratories). For the quantitation of HDAC7-containing nuclei, we counted at least 3 different fields in each coverslip in 2 distinct coverslips.

Inhibition of HDACs with Short Interfering RNA.

Predesigned short interfering RNA (siRNA) oligonucleotides to knock down HDAC1, HDAC3, HDAC7, SHP, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (siGenome siRNA SMART pools) (Dharmacon, Lafayette, CO) were transfected into HepG2 cells according to the manufacturer's instructions. Dharmacon's siControl nontargeting siRNA pool, which is a pool of four nontargeting siRNAs, and siControl GAPDH (Supplementary Fig. 4), which is a pool of siRNA oligonucleotides targeting GAPDH, were used as negative controls. In cells transfected with siControl siRNA, the response of CYP7A1 to bile acids did not differ to those of mock-transfected cells (data not shown). Thirty hours after transfection, HepG2 cells were treated with 25 μM CDCA, and at the end of the incubation total RNA was extracted and analyzed via real-time quantitative PCR (Q-PCR). To rule out unspecific silencing (that is, off-target effects), we tested the effect of knockdown with siRNA by monitoring the mRNA levels of different HDACs and the response to bile acids of other bile acid–regulated genes (Supplementary Figs. 1-3).

Animal Studies.

Male 10-week-old Ldl-r/ mice (genetic background C57BL/6J), 7-9 animals/group, were injected intraperitoneally for 7 days with 205 mg valproic acid/kg body weight twice a day, 1 mg trichostatin A/kg body weight once a day, or vehicle (0.9% NaCl). Plasma alanine aminotransferase and aspartate aminotransferase levels were determined with an enzymatic kit (Sentinel, Milan, Italy) at the end of the treatment, and no statistically significant difference between control and treated groups was observed. Animals had free access to food and water and the light cycle was from 7:00 A.M. to 7:00 P.M. Total RNA was extracted from livers, reverse-transcribed, and analyzed via real-time Q-PCR. Primer sequences are available in Table 1. All animal experiments were approved by the University of Milan and the Italian Ministry of Health and were conducted while strictly following European Community (EEC Directive no. 609/86) and local regulations for animal care (Italian Legislative Decree no. 116, January 27, 1992).

Fecal Bile Acid Excretion.

Mice were placed in metabolic cages for the collection of stools over a 72-hour period. Dried feces (0.2 g) were extracted in 4 mL of 75% ethanol at 50°C for 2 hours. The extracts were centrifuged and supernatants were diluted 20-fold with 65 mM phosphate buffer (pH 7.0). Bile acid concentration was measured with an enzymatic kit (Sentinel). The daily feces output (g/day per 100 g body weight) and fecal bile acid content (μmol/g) were used to calculate the rate of bile acid excretion (μmol/day/100 g body weight).28

Liver Bile Acid and Cholesterol Content.

Two hundred milligrams of livers were homogenized in 2 mL PBS, and 2 × 105 disintegrations per minute of [3H]cholesterol were used to normalize for recovery. Total lipids were extracted by adding 10 mL chloroform/methanol (2:1). The organic phase was recovered and dried, and radioactivity was counted via liquid scintillation. For the determination of bile acid and cholesterol content in liver extracts, samples were diluted 20-fold with 65 mM phosphate buffer (pH 7.0) and bile acid and cholesterol concentration was measured with an enzymatic kit (Sentinel). After correcting for the percentage of recovery, values were expressed as the mean ± standard deviation (SD) of nanograms total bile acids per gram of liver.

Determination of Plasma Cholesterol.

Total plasma cholesterol was determined with an enzymatic kit (Sentinel). Cholesterol distribution in plasma lipoprotein fractions was determined via fast protein liquid chromatography (FPLC) using a Superose 6 column (Amersham Biosciences, Milan, Italy). One-milliliter fractions were collected and assayed for cholesterol with an enzymatic kit (Sentinel).

Statistical Analyses.

Statistical analyses were performed via 1-way analysis of variance, using GraphPad Prism version 4.0 (GraphPad Software, San Diego, CA).

Results

Bile Acids Affect CYP7A1 Transcription via Chronologically Distinct Mechanisms.

To distinguish between FXR/SHP-independent and FXR/SHP-dependent pathways of CYP7A1 regulation, we determined the mRNA levels of CYP7A1 in the presence of CDCA or the FXR-selective agonist GW4064. CDCA reduces CYP7A1 mRNA levels at 1, 16, and 40 hours of incubation (Fig. 1A), whereas GW4064 affects the mRNA levels only at 16 and 40 hours (Fig. 1B). We also observed that SHP mRNA rises only after 16 hours but not after 1 hour of treatment with CDCA (Fig. 1A) or GW4064 (Fig. 1B), consistent with the inhibition mediated by the FXR pathway. The mRNA levels of another bile acid–repressed gene, apolipoprotein C-III,29 also decrease only 16 hours after the addition of CDCA (Fig. 1C) or GW4064 (Fig. 1D), suggesting that the rapid reduction of mRNA levels in response to bile acids may be limited to certain genes, including CYP7A1. On the other hand, the mRNA levels of FGF-19 as well as SHP rise at 16 hours but not at 1 hour after incubation with CDCA (Fig. 1C) or with GW4064 (Fig. 1D), indicating that the early response of CYP7A1 to bile acids and the induction of FGF-19 and SHP are dissociated. As reported,30, 31 CDCA increases the expression of FXR (Fig. 1E), whereas CA does not affect the mRNA levels of FXR, CYP7A1, and SHP (Fig. 1E,F). None of the compounds used in these experiments affect the viability of HepG2 cells (data not shown). Because GW4064 affects CYP7A1 and SHP mRNA levels only after 16 hours, we infer that the initial inhibition elicited by CDCA does not involve FXR/SHP inhibitory pathway.10, 11

Figure 1.

Analysis of gene expression in HepG2 cells in response to CDCA, GW4064 and CA. (A) CYP7A1 and SHP mRNA levels in HepG2 cells treated with 25 μM CDCA for the indicated times. (B) CYP7A1 and SHP mRNA levels in HepG2 cells treated with 5 μM GW4064 for the indicated times. (C) Apolipoprotein C-III and FGF-19 mRNA levels in HepG2 cells treated with CDCA. (D) Apolipoprotein C-III and FGF-19 mRNA levels in HepG2 cells treated with GW4064. (E) FXR mRNA levels in HepG2 cells treated either with CDCA or CA for the indicated times. (F) Effect of CA treatment on CYP7A1 and SHP mRNA levels. (*) and (**) indicate statistical significant difference with P < 0.05 and P < 0.01, respectively.

Bile Acids Induce the Ordered Recruitment of a Repressive Complex on the CYP7A1 Promoter.

To unravel the molecular mechanisms whereby bile acids mediate the early inhibition of CYP7A1 transcription, we analyzed the factors recruited in vivo to this promoter and the local modifications of histones by ChIP assay. As previously shown,16 CDCA causes the dissociation of CBP and PGC-1 from CYP7A1 promoter (Fig. 2A), which is accompanied by the recruitment of the corepressors SMRTα and N-CoR (Fig. 2B) and of HDAC7 and HDAC3 (Fig. 2D). HDAC1 appears on the promoter only at 16 hours (Fig. 2D). CDCA does not affect the association of HDAC2 to the CYP7A1 promoter (Fig. 2D), suggesting that it is not involved in the regulatory pathway induced by bile acids. The other HDACs tested (that is, HDAC4, HDAC5, HDAC6, and HDAC8) are not recruited to the CYP7A1 promoter (data not shown). Interestingly, SHP is recruited to the CYP7A1 promoter only 16 hours after the addition of CDCA, which indicates that SHP is poorly associated to this promoter in the absence of bile acids and that it contributes to the bile acid–mediated repression of CYP7A1 transcription only at a later stage (Fig. 2B). This result correlates with SHP mRNA levels, which are fully induced only at 16 hours (Fig. 1A,B). We checked the modifications of histone tails induced by bile acids and observed that histone H4 deacetylation occurs only at 16 hours (Fig. 2C), whereas the acetylation state of histone H3 does not change (Fig. 2C).

Figure 2.

Analysis of factors recruited to CYP7A1 promoter in response to bile acids. HepG2 cells were treated with 25 μM CDCA for the indicated times and ChIP analyses were performed with antibodies against PGC-1 and CBP (A), with antibodies against SHP, SMRTα and nuclear corepressor (N-CoR) (B), with the antibodies against acetyl-histone H3 and acetyl-histone H4 (C), with antibodies against HDAC1, HDAC2, HDAC3, HDAC7 (D), with the antibody against RNA polymerase II (E). The amplified sequences of immunoprecipitated genomic fragments were from nucleotide (nt) −252 to nt −113 (including the bile acid-responsive element) of the human CYP7A1 promoter for the ChIP assays in (A,B,D) and from nt −90 to nt +26 (including the TATA box) of the human CYP7A1 promoter for the ChIP assays in (C) and (E). In (E), the 3′ untranslated region from nt 2009 to nt 2141 of the CYP7A1 cDNA was amplified. The CYP7A1 upstream region from nt −1935 to nt −1833 was amplified as a negative control. Data are mean ± SD of the fold above the background value (no antibody). (*) and (**) indicate statistical significant difference with P < 0.05 and P < 0.01, respectively.

Finally, we monitored the kinetics of RNA polymerase II on the core promoter and at the 3′ untranslated region of CYP7A1 and found that the total amount of RNA polymerase II had already decreased at 1 hour on both the TATA box and the 3′ untranslated region of the CYP7A1 promoter (Fig. 2E), which probably reflects the early transcriptional repression elicited by bile acids.

Comparative studies on apolipoprotein C-III promoter show that CDCA inhibits only after 16 hours (data not shown). These results suggest that bile acids promote, at an early stage, the recruitment of a repressive complex containing corepressors and HDACs on CYP7A1 promoter.

Inhibition of HDAC7 Expression Prevents the Negative Effects of Bile Acids on CYP7A1 Transcription.

We next sought to confirm the role of specific HDACs in the bile acid–induced repression of CYP7A1 transcription. When HDAC7 is silenced in HepG2 cells, bile acids do not decrease CYP7A1 mRNA (Fig. 3A,B). The basal mRNA levels of SHP are not affected in HDAC7-depleted cells and are normally induced upon treatment with CDCA (Supplementary Fig. 1), indicating that despite the normal expression and induction of SHP, bile acids cannot inhibit CYP7A1 transcription when HDAC7 is silenced. The mRNA levels of HDAC1, HDAC2, and HDAC3 and of other genes relevant to lipid metabolism and transport are not affected in HDAC7-silenced cells (Supplementary Fig. 1). In contrast, siRNA against SHP (Fig. 3C,D), HDAC1 (Fig. 3E,F), HDAC3 (Fig. 3G,H), and GAPDH (Supplementary Fig. 4) do not affect the response of HepG2 cells to CDCA. Notably, knockdown of HDCA7 does not block CYP7A1 repression by GW4064 (Supplementary Fig. 5). In summary, these results underscore the key role of HDAC7 in the feedback regulation of CYP7A1 gene transcription by bile acids.

Figure 3.

HDAC7 is the key factor for the bile acid-mediated repression of CYP7A1 gene transcription. (A,B) Specific siRNA oligonucleotides to silence HDAC7 were transfected into HepG2 cells, which were treated with 25 μM CDCA or with 0.1% ethanol as vehicle (ctrl) for the indicated times. Cells were harvested for RNA extraction and quantitation of CYP7A1 (A) and HDAC7 (B) mRNA by real time Q-PCR. (C,D) Specific siRNA oligonucleotides to silence SHP were transfected into HepG2 cells, which were treated with 25 μM CDCA or ethanol for the indicated times and analyzed for specific mRNA content as indicated. (E,F) Specific siRNA oligonucleotides to silence HDAC1 were transfected into HepG2 cells, which were treated with 25 μM CDCA or ethanol for the indicated times and analyzed for specific mRNA content as indicated. (G,H) Specific siRNA oligonucleotides to silence HDAC3 were transfected into HepG2 cells, which were treated with 25 μM CDCA or ethanol for the indicated times and analyzed for specific mRNA content as indicated. Scrambled oligonucleotides were used as negative controls in each RNA interference experiment (CTRLsiRNA). Data are expressed as mean ± SD. (*) and (**) indicate statistical significance at P < 0.05 and P < 0.01, respectively.

Bile Acids Induce the Nuclear Localization of HDAC7.

To assess if CDCA affects the intracellular localization of HDAC7, we used immunocytochemistry and found that HDAC7 is mostly localized in the cytoplasm in control HepG2 cells32 (Fig. 4A), whereas HDAC7 is mainly localized in the nuclear compartment in HepG2 cells exposed to CDCA for 1 hour and for 16 hours (Fig. 4A,B). In contrast, GW4064 does not affect the intracellular localization of HDAC7 (GW4064) (Fig. 4A). This experiment shows that bile acids induce the rapid nuclear localization of HDAC7 in an FXR-independent fashion, an event that most likely facilitates its recruitment on the CYP7A1 promoter.

Figure 4.

Bile acids induce the nuclear localization of HDAC7 via an FXR-independent pathway. (A) HepG2 cells were cultured on coverslips and treated with 25 μM CDCA or 5 μM GW4064 for the indicated times. Coverslips were incubated with a polyclonal antibody against HDAC7 and then with Alexa Fluor 488 goat anti-rabbit IgG secondary antibody. Alexa Fluor 546 phalloidin and TOPRO-3 iodide were added to stain the cytoskeleton and the nucleus, respectively. Images were finally merged into a single layer and analyzed by confocal microscopy (magnification was 40×). (B) For the quantitation of HDAC7-containing nuclei, we counted at least 3 different fields in each coverslip in 2 distinct coverslips. Data are expressed as mean ± SD of the percentage of nuclei positive to HDAC7 staining. (**) indicates statistical significance at P < 0.01.

To further support the role of HDAC7 in bile acid–mediated down-regulation of CYP7A1 transcription, we asked whether the inhibition of the nuclear import of HDAC7 could prevent the repressive effect of bile acids. For this purpose, we used calyculin A, a phosphatase inhibitor previously shown to inhibit the nuclear import of HDAC7 due to the interaction with cytosolic adaptor proteins 14-3-3.33 HepG2 cells were incubated for 1 hour in the absence or presence of combinations of calyculin A and CDCA. In the presence of calyculin A, CDCA cannot promote the nuclear import of HDAC7 (Fig. 5A). Calyculin A also prevents the CDCA-mediated decrease of CYP7A1 expression (Fig. 5B).

Figure 5.

Phosphatase inhibitors prevent the translocation of HDAC7 to the nucleus and the transcriptional repression of CYP7A1 by bile acids. HepG2 cells were treated with combinations of 10 nM calyculin A and 25 μM CDCA as indicated, for 1 hour and analyzed for the subcellular localization of HDAC7 by confocal microscopy (magnification was 20×) (A) and CYP7A1 mRNA by real-time Q-PCR (B). Data are expressed as mean ± SD. (**) indicates statistical significance at P < 0.01.

HDAC Inhibitors Prevent the Feedback Regulation of CYP7A1 Transcription.

The sequential recruitment of HDAC7, HDAC3, and HDAC1 on the CYP7A1 promoter led us to hypothesize that HDAC inhibitors may prevent this negative regulation. Therefore, we measured CYP7A1 mRNA levels in HepG2 cells incubated with CDCA, the HDAC inhibitors VPA and TSA, and a combination of CDCA and VPA or TSA. As expected, CDCA decreased CYP7A1 transcript levels but VPA and TSA completely prevented the negative effect of bile acids (Fig. 6A). Possible toxic effects of VPA and TSA were ruled out by a cytotoxicity test (data not shown). The same experiment was also performed with GW4064 but in this case the HDAC inhibitors do not prevent the repression elicited by GW4064 (Fig. 6B).

Figure 6.

HDAC inhibitors prevent the feedback regulation of CYP7A1 by bile acids. (A) HepG2 cells were cultured in the presence of different combinations of 25 μM CDCA and 1.5 mM VPA or 200 nM TSA for 16 hours as indicated (C+T= CDCA+TSA; C+V= CDCA+VPA). CYP7A1 mRNA levels are expressed relative to the control wells treated with vehicle (ethanol, EtOH) and are expressed as mean ± SD. (B) HepG2 cells were cultured in the presence of different combinations of 5 μM GW4064 (GW) and 1.5 mM VPA or 200 nM TSA for 16 hours as indicated (GW+T= GW4064+TSA; GW+V= GW4064+VPA). CYP7A1 mRNA levels are expressed relative to the control wells treated with vehicle (ethanol, EtOH) and are expressed as mean ± SD. (*) and (**) indicate statistical significance versus control (EtOH) at P < 0.05 and P < 0.01, respectively.

HDAC inhibitors reduce low-density lipoprotein cholesterol in vivo by increasing the expression of Cyp7a1 and bile acid synthesis.

The detailed analysis of the molecular mechanisms underlying the regulation of CYP7A1 transcription suggests that targeting HDACs with an appropriate inhibitor may release Cyp7a1 from the physiological repression caused by bile acids in vivo and, consequently, may also decrease blood cholesterol as a result of its enhanced conversion to bile acids. To test this hypothesis and to verify the effects of HDAC inhibitors in vivo, we performed experiments with the Ldl-r/ mouse, an animal model of hypercholesterolemia caused by the genetic deficiency of the low-density lipoprotein receptor. We chose VPA and TSA because they are well tolerated in animals even at high doses.24, 26 VPA and TSA elevate Cyp7a1 mRNA levels dramatically in Ldl-r/ mice (Fig. 7A). Interestingly, we found that the mRNA levels of Shp are also elevated in mice treated with both HDAC inhibitors (Fig. 7A), as a consequence of the increased total bile acid concentration in the liver and fecal bile acid excretion (Fig. 7B), the latter considered an index of bile acid synthesis.28 Notably, both VPA and TSA reduce total plasma cholesterol levels very effectively, whereas hepatic cholesterol does not decrease (Fig. 7C). FPLC analysis of lipoprotein fractions shows that the administration of VPA or TSA decreases low-density lipoprotein cholesterol in Ldl-r/ mice whereas the high-density lipoprotein peak is not altered (Fig. 7D). Finally, to make sure that the decrease of plasma cholesterol is due primarily to the enhanced expression of the Cyp7a1 gene, we also measured the expression of a subset of genes relevant to lipid metabolism and found no changes (Fig. 7E) that could contribute to the hypocholesterolemic effect of VPA and TSA. Altogether, these results reveal for the first time that the inhibition of HDACs, by enhancing the conversion of cholesterol to bile acids, may be a therapeutic strategy for the treatment of hypercholesterolemia.

Figure 7.

The HDAC inhibitors, valproic acid and trichostatin A, increase Cyp7a1 expression and reduce low-density lipoprotein cholesterol in vivo. (A) Ldl-r/ mice were treated with saline (white bars) or VPA (dashed bars) or TSA (black bars) and at the end liver samples were extracted for the determination of Cyp7a1 and Shp mRNA levels. (B) Equal aliquots of the livers from the different treatment groups were extracted and total bile acid content was measured. Stools were collected in the last 3 days of the treatment for determination of fecal bile acid excretion. (C) Total plasma and hepatic cholesterol from Ldl-r/ mice treated with intraperitoneal injections of saline (white bars), VPA (dashed bars) or TSA (black bars). (D) Plasma samples from Ldl-r/ mice treated with saline, VPA or TSA were pooled and aliquots of 0.5 mL were analyzed by FPLC to determine cholesterol distribution in lipoprotein fractions. (E) Ldl-r/ mice were treated with saline (white bars), VPA (dashed bars), or TSA (black bars). Liver samples were analyzed by real-time Q-PCR to determine mRNA levels of genes involved in lipid metabolism. Data are expressed as mean ± SD. (**) indicates statistical significance with P < 0.01.

Discussion

The kinetic analysis of mRNA levels and of cofactors recruited on the CYP7A1 promoter allowed us to show that bile acids repress CYP7A1 transcription in two waves, the first one involving a repressive complex containing HDAC3 and HDAC7, the latter the FXR/SHP cascade. The combination of results obtained by ChIP, RNA interference, and immunocytochemistry analysis suggests that the early stage of CYP7A1 transcriptional repression is due to the recruitment of a repressor complex on this promoter containing HDAC7, HDAC3, the corepressors SMRTα, and N-CoR that promotes the dissociation of RNA polymerase II.

The FXR/SHP-dependent repression occurs only after several hours, as demonstrated by the delayed increase of SHP mRNA. One may argue that the basal intracellular concentrations of SHP may be already sufficient to sustain the repression of CYP7A1 gene transcription elicited by bile acids. However, ChIP assay demonstrates that SHP does not associate to the CYP7A1 gene promoter during the initial exposure of liver cells to bile acids. The indispensable role of HDAC7 at early time points, demonstrated in the knockdown experiments (Fig. 3A) and in the experiments with calyculin A (Fig. 5), suggests that the initial events triggered on CYP7A1 gene promoter by HDAC7 in response to bile acids are strictly required. We also showed that the silencing of SHP in cell culture does not prevent the repression of CYP7A1 by bile acids. Moreover, even if SHP mRNA is normally induced by bile acids in HDAC7-silenced HepG2 cells (Supplementary Fig. 1) as well as in mice with elevated bile acid concentration in the liver after treatment with HDAC inhibitors (Fig. 7A), this does not seem to be sufficient to achieve the repression of CYP7A1 transcription. Thus, consistent with previous observations showing that Shp null mice still remain susceptible to repression by bile acids17, 18 our results support the concept that SHP is only partially responsible for the transcriptional repression of the CYP7A1 gene. On this basis, we propose that SHP intervenes at a later stage, after other factors have triggered the repression of CYP7A1 gene transcription. It has been demonstrated, in fact, that SHP interacts with HDAC134 and both these factors are recruited to the CYP7A1 promoter only at 16 hours after bile acid treatment32 (Fig. 2B,D). Hence, HDAC1 and SHP might cooperate to repress the CYP7A1 transcriptional machinery at the later time point. Our results fit and expand the recent observations of the groups of Talianidis35 and Kemper,34, 36 in that HDAC7 and the associated SMRTα and HDAC3 prepare the CYP7A1 promoter by altering the array of transcription factors and coregulators associated to it and by marking local histones with deacetylation, a prerequisite necessary for the subsequent methylation of lysine 9 of histone H3 by G9a.36 The crucial role played by HDACs in the negative regulation of CYP7A1 could explain the lack of repression of Cyp7a1 in mice treated with VPA and TSA, despite the increased hepatic bile acid concentration and SHP mRNA levels.

VPA, an anticonvulsant drug,37 is also an inhibitor of HDAC activity. Here, we demonstrate that it also lowers plasma cholesterol very effectively. This result correlates with and provides a mechanism to explain the previously reported hypocholesterolemic effect of VPA in epileptic patients.38–41 Furthermore, we show that another structurally unrelated HDAC inhibitor, such as TSA, reduces plasma cholesterol, confirming the hypocholesterolemic properties of these compounds. In principle, HDAC inhibitors may also determine the sharp reduction of plasma cholesterol by affecting the expression of other genes that regulate lipid metabolism and transport. However, our data on mRNA levels of apolipoprotein B-100, HMG-CoA reductase, and of other genes relevant to lipid metabolism suggest that the effect is mostly mediated by increased bile acid biosynthesis. To the best of our knowledge, besides ion exchange resins, no other pharmacological treatment stimulates the expression of Cyp7a1 to this extent, while at the same time decreasing plasma cholesterol dramatically. It should be stressed, however, that in this study VPA and TSA were used mainly as tools to demonstrate that pharmacological inhibition of HDAC activity may be a promising approach to strongly up-regulate CYP7A1 and consequently lower plasma cholesterol. Future investigations should be aimed at identifying more selective HDAC inhibitors that decrease plasma cholesterol by stimulating its conversion to bile acids.

In conclusion, we define here the molecular events and the factors that determine the rapid bile acid–mediated repression of CYP7A1 transcription. The relevant consequence of this study is the identification of HDACs, and in particular HDAC7, as possible targets for the design of hypocholesterolemic agents that may help in the prevention of the associated risk of atherosclerosis and coronary artery disease, especially in those patients that do not respond very well to other hypocholesterolemic drugs. Moreover, modulation of bile acid synthesis by targeting these mechanisms may also be exploited for the treatment of liver diseases.

Acknowledgements

We thank Drs. Enrique Saez (The Scripps Research Institute, La Jolla, CA), Iannis Talianidis (Biomedical Sciences Research Center “Alexander Fleming”, Vari, Greece), Walter Wahli (Center for Integrative Genomics, Université de Lausanne, Lausanne, Switzerland), Ronald Evans (The Salk Institute for Biological Studies, San Diego, CA) and Laura Calabresi (Dipartimento di Scienze Farmacologiche, Università degli Studi di Milano, Italia) for critically reading the manuscript and for helpful discussion. We are grateful to Drs. E. Verdin (The Gladstone Institute of Virology and Immunology, San Francisco, CA), E. Seto (Moffitt Cancer Center, Tampa, FL), R. M. Evans (The Salk Institute, San Diego, CA), I. Talianidis (Biomedical Sciences Research Center “Alexander Fleming”, Vari, Greece), and K. Bamberg (Astra-Zeneca, Mölndal, Sweden) for kindly providing plasmids, antibodies, and other reagents. We also thank Drs. Laura Calabresi, Giulio Simonutti, and Danilo Norata (Dipartimento di Scienze Farmacologiche, Università di Milano) for help with FPLC analyses of plasma lipoproteins and immunocytochemistry and Miss Elda Desiderio Pinto for administrative assistance.

Ancillary