MAFG Is a Transcriptional Repressor of Bile Acid Synthesis and ...

Article

MAFG Is a Transcriptional
Repressor of Bile AcidSynthesis and Metabolism
Graphical Abstract

Highlights

d FXR activation induces expression of many transcriptional

repressors including MafG

d MAFG represses bile acid synthetic genes and alters bile acid

composition

d Bile acid synthesis and metabolism genes have MAFG

response elements (MAREs)

d MAFG is an important regulator of bile acid negative feedback

regulation

de Aguiar Vallim et al., 2015, Cell Metabolism 21, 298–310February 3, 2015 ª2015 Elsevier Inc.http://dx.doi.org/10.1016/j.cmet.2015.01.007

Authors

Thomas Q. de Aguiar Vallim,

Elizabeth J. Tarling, ...,

Masayuki Yamamoto,

Peter A. Edwards

[email protected] (T.Q.d.A.V.),[email protected] (P.A.E.)

In Brief

de Aguiar Vallim et al. identify MAFG as an

FXR target gene that functions as a

transcriptional repressor of bile acid

synthetic genes, thus altering the

composition of the bile acid pool. These

studies identify a molecular mechanism

for the negative feedback regulation of

bile acid synthesis.

mailto:[email protected]


http://dx.doi.org/10.1016/j.cmet.2015.01.007

http://crossmark.crossref.org/dialog/?doi=10.1016/j.cmet.2014.01.007&domain=pdf

Cell Metabolism

Article

MAFG Is a Transcriptional Repressorof Bile Acid Synthesis and MetabolismThomas Q. de Aguiar Vallim,1,* Elizabeth J. Tarling,1 Hannah Ahn,1 Lee R. Hagey,2 Casey E. Romanoski,3 Richard G. Lee,4

Mark J. Graham,4 Hozumi Motohashi,5 Masayuki Yamamoto,6 and Peter A. Edwards1,7,*1Division of Cardiology, David Geffen School of Medicine, University of California, Los Angeles, CA 90095, USA2Department of Medicine, University of California, San Diego, La Jolla, CA 92093, USA3Department of Cellular and Molecular Medicine, University of California, San Diego, La Jolla, CA 92093, USA4ISIS Pharmaceuticals, Carlsbad, CA 92010, USA5Department of Gene Expression Regulation, Institute of Development, Aging and Cancer, Tohoku Medical Megabank Organization,

Sendai 980-8573, Japan6Department of Medical Biochemistry, Tohoku Medical Megabank Organization, Sendai 980-8573, Japan7Department of Biological Chemistry, University of California, Los Angeles, CA 90095, USA

*Correspondence: [email protected] (T.Q.d.A.V.), [email protected] (P.A.E.)


SUMMARY

Specific bile acids are potent signaling moleculesthat modulate metabolic pathways affecting lipid,glucose and bile acid homeostasis, and the micro-biota. Bile acids are synthesized from cholesterolin the liver, and the key enzymes involved in bileacid synthesis (Cyp7a1, Cyp8b1) are regulated tran-scriptionally by the nuclear receptor FXR. We haveidentified an FXR-regulated pathway upstream of atranscriptional repressor that controls multiple bileacid metabolism genes. We identify MafG as anFXR target gene and show that hepatic MAFG over-expression represses genes of the bile acid syntheticpathway and modifies the biliary bile acid composi-tion. In contrast, loss-of-function studies usingMafG+/� mice causes de-repression of the samegenes with concordant changes in biliary bile acidlevels. Finally, we identify functional MafG responseelements in bile acid metabolism genes using ChIP-seq analysis. Our studies identify a molecular mech-anism for the complex feedback regulation of bileacid synthesis controlled by FXR.

INTRODUCTION

Bile acids function both as detergents that facilitate lipid absorp-

tionandas endogenous ligands that regulatemetabolic pathways

through activation of several nuclear receptors, including the

farnesoid X receptor (FXR, Nr1h4) as well as TGR5, a G protein-

coupled receptor (de Aguiar Vallim et al., 2013a). Although FXR

plays a particularly important role in maintaining bile acid homeo-

stasis, numerous studies have shown that FXR directly regulates

many genes that affect multiple metabolic cascades (de Aguiar

Vallim et al., 2013a; Evans and Mangelsdorf, 2014). Consistent

with these findings, mice lacking FXR exhibit not only dysregu-

lated bile acid metabolism, but also abnormal lipoprotein (Sinal

et al., 2000) and glucose metabolism (Duran-Sandoval et al.,

2005; Zhang et al., 2006), increased hepatic susceptibility to

certain toxins (Lee et al., 2010), increased levels of ileal bacteria,

and impaired barrier function of intestinal epithelia (Inagaki et al.,

2006). Reduced FXR signaling is also associated with obesity,

possibly as a result of bile acid-dependent modulation of the mi-

crobiota (Li et al., 2013; Ridaura et al., 2013). Moreover, a recent

study demonstrated that in mice, the benefits of bariatric surgery

may be mediated by FXR signaling to modulate bile acid-depen-

dent effects on the microbiota (Ryan et al., 2014). Thus, the find-

ings that elevatedbile acid levels in humans and/ormice areasso-

ciatedwithgastrointestinal diseases, hepatoxicity, alteredplasma

lipoprotein levels, and aberrant glucose metabolism suggest that

abnormal control of the bile acid pool can have broad physiolog-

ical effects (de Aguiar Vallim et al., 2013a; Kuipers et al., 2014).

Although negative feedback of bile acid synthesis was first

described over 50 years ago (Beher et al., 1961), the precise

mechanisms by which bile acids mediate this repression are still

not fully understood. The enzymatic catabolism of cholesterol, or

hydroxysterols, to form primary bile acids occurs via either the

classic or alternative pathways (Figure 1A). These two pathways

generate approximately 75% and 25%, respectively, of the total

primary bile acids and involve at least 17 enzymes (Russell, 2003)

(Figure 1A). Within the classic pathway, CYP7A1 is the rate-

limiting enzyme, while CYP8B1 regulates the synthesis of cholic

acid (Li-Hawkins et al., 2002) and thus regulates the bile acid

pool composition (Figure 1A). The transcription of both Cyp7a1

andCyp8b1 are particularly responsive to end product feedback

control (Russell, 2003). In contrast, the alternative bile acid

pathway involves CYP7B1 and CYP27A1 (Figure 1A) (Russell,

2003). Little is known about the regulation of the genes that

encode enzymes of the alternative pathway, or downstream of

CYP7A1 and CYP8B1. Nonetheless, the finding that a number

of diseases result from mutations of CYP7A1, CYP7B1,

CYP27A1, HSD3B7, AMACR, or AKR1C4 (Akr1c14 in mice) (Fig-

ure 1A) emphasizes the importance of maintaining normal bile

acid synthesis and homeostasis.

In humans and mice, one of the most abundant bile acids is

cholic acid (CA). Humans also have high levels of chenodeox-

ycholic acid (CDCA). In contrast, mice almost quantitatively

298 Cell Metabolism 21, 298–310, February 3, 2015 ª2015 Elsevier Inc.




http://crossmark.crossref.org/dialog/?doi=10.1016/j.cmet.2015.01.007&domain=pdf

convert CDCA to muricholic acid (MCA) (Figure 1A) (de Aguiar

Vallim et al., 2013a; Russell, 2003). The negative feedback

regulation of bile acid synthesis is largely dependent on activa-

tion of hepatic and/or intestinal FXR (Kim et al., 2007). Such

activation results in the induction of small heterodimerizing

partner (Shp, Nr0b2) in the liver (Kerr et al., 2002; Wang

et al., 2002) and Fgf15 (mouse) or FGF19 (human) in the small

intestine (Inagaki et al., 2005, 2006). SHP does not bind DNA

directly, but rather binds to other transcription factors such

as HNF4a and LRH-1 to impair their function (Bavner et al.,

2005; Goodwin et al., 2000; Lu et al., 2000). In contrast, intes-

tinally derived FGF15/19 is secreted into the blood before

binding to the FGFR4/b-Klotho receptor on the surface of he-

patocytes to initiate an incompletely understood pathway that

leads to repression of Cyp7a1 (Inagaki et al., 2005). In contrast

to the detailed studies detailing the mechanisms that con-

trol Cyp7a1, the mechanisms involved in the repression of

Cyp8b1, and thus cholic acid synthesis, are less well under-

stood. Nonetheless, FXR activation is known to repress

Cyp8b1 expression by mechanisms that may also involve

SHP and FGF15/19 (Kerr et al., 2002; Kong et al., 2012;

Wang et al., 2002).

Here, we identify a previously unrecognized FXR-regulated

pathway involving MAFG (V-Maf Avian Musculoaponeurotic

Fibrosarcoma Oncogene Homolog G), a member of the MAF

family of transcription factors. We show that MafG is a direct

target gene of FXR, and hepatic overexpression of MAFG in

mice represses genes encoding enzymes of the classic and

alternative pathways. In addition, overexpression of MAFG in

mice resulted in a decrease in biliary cholic acid levels and an in-

crease in muricholic acid levels, a finding consistent with the

greater inhibition of Cyp8b1 as compared to Cyp7a1. Finally,

we utilize loss-of-function approaches (knockdown of MafG

with antisense oligonucleotides, MafG+/� mice) to show that a

50%–80% loss of hepatic MAFG protein results in both de-

repression of many of the same genes, including Cyp8b1, and

an increase in biliary cholic acid levels. In conclusion, our results

identify an FXR-MafG pathway that functions in the feedback

repression of bile acid metabolism by modulating the composi-

tion of the bile acid pool.

A B C

D

Figure 1. FXR Activation Represses Most Bile Acid Synthesis Genes

(A) Schematic diagram of the major hepatic enzymes involved in bile acid synthesis. Genes in red are repressed, while genes in yellow are unchanged, and the

gene in green is induced following FXR activation.

(B–D) Hepatic expression of genes encoding enzymes of the (B) classic or (C) alternative bile acid synthetic pathway or (D) remaining genes involved in primary

bile acid synthesis. mRNA levels were measured by qRT-PCR following treatment of wild-type or Fxr�/� mice (7–9 mice/group) for 3 days with GW4064 or

GSK2324 at 60 mpk/day. All data are shown as mean ± SEM. Asterisks indicate statistically significant differences comparing WT or KO vehicle-treated against

agonist-treated mice (*p < 0.05; **p < 0.01; ***p < 0.001).

Cell Metabolism 21, 298–310, February 3, 2015 ª2015 Elsevier Inc. 299

RESULTS

FXR Activation Represses Both the Classic andAlternative Bile Acid Synthetic PathwaysMultiple studies have shown that activated FXR leads to

repressed transcription of both Cyp7a1 and Cyp8b1 (Kerr

et al., 2002; Kong et al., 2012; Wang et al., 2002). While the

mechanisms regulating Cyp7a1 expression have been exten-

sively studied, much less is known about how FXR represses

Cyp8b1 or whether FXR regulates the expression of other genes

involved in the two bile acid synthetic pathways. In our initial

studies, we treated wild-type and Fxr�/� mice (KO) for 3 days

with either GW4064, a widely used FXR agonist (Maloney

et al., 2000), or GSK2324, a water-soluble derivative of

GW4064 that exhibits increased potency (Bass et al., 2011). As

expected, both agonists led to robust repression of both

Cyp7a1 andCyp8b1 in wild-type, but not Fxr�/�mice (Figure 1B).

Importantly, both agonists also resulted in an FXR-dependent

repression of Cyp7b1 and Cyp27a1 in the alternative pathway

(Figure 1C). In addition, we measured the hepatic mRNA levels

of a number of additional genes that encoded enzymes involved

in bile acid synthesis. We now show for the first time that FXR

activation in wild-type, but not Fxr�/� mice, results in repression

of numerous bile acid synthetic genes, including Acox2,

Akr1c14, Hsd3b7, Hsd17b4, Scp2, and Slc27a5 (Figure 1D). In

contrast, Amacr and Cyp39a1mRNA levels are unchanged (Fig-

ure 1D), while Akr1d1 is modestly induced after FXR activation,

consistent with ChIP-seq data that identify a putative intronic

FXRE in the Akr1d1 locus (Figure S1A). Overall, the repression

of most bile acid synthetic genes was not as pronounced as

that observed forCyp8b1 andCyp7a1 (Figure 1B). Nevertheless,

these studies demonstrate that treatment of mice with two

different synthetic FXR agonists results in repression of genes

involved in both the classic and alternative bile acid synthetic

pathways, consistent with a central role for FXR in regulating

all aspects of bile acid synthesis.

FXR Activation Induces the Expression of SeveralTranscriptional RepressorsThe nuclear receptor FXR binds to its cognate response element

(FXRE) as an FXR:RXR heterodimer and functions almost exclu-

sively as a transcriptional activator (de Aguiar Vallim et al.,

2013a). Nonetheless, although activation of FXR leads to induc-

tion of many hepatic genes, it also results in repression of

numerous genes involved in bile acid metabolism (Figure 1D).

A number of studies have demonstrated that the mechanisms

involved in the repression of Cyp7a1 and Cyp8b1 are indirect

and are the result of the FXR-dependent increased expression

of Shp and Fgf15/19 that encode proteins that function to inhibit

transcription of specific genes (de Aguiar Vallim et al., 2013a;

Kuipers et al., 2014). One additional mechanism by which FXR

causes a reduction in specific genes is through miRNAs. For

example, we recently identified miR-144 as an FXR-regulated

miRNA that subsequently targets ABCA1 (de Aguiar Vallim

et al., 2013b). Based on these earlier studies, we hypothesized

that FXR activation might increase the expression of additional

repressors that function to control metabolic pathways.

To identify such putative transcriptional repressors, we re-

analyzed the data from our prior ChIP-seq study that had been

used to identify global hepatic FXR response elements (FXREs)

(Chong et al., 2010). Gene ontology analysis identified a signifi-

cant enrichment in transcription factors containing FXREs

(Chong et al., 2010). Consequently, we searched this subset of

FXRE-containing genes, focusing specifically on genes anno-

tated to have transcriptional repressor activity. Our analysis

identified four putative transcriptional repressor genes, namely

Shp, a well-characterized FXR target gene, v-maf musculoapo-

neurotic fibrosarcoma oncogene homolog G (avian) (MafG),

cysteine-rich intestinal protein 2 (Crip2), and zinc finger protein

385a (Zfp385a). We also identified a fifth putative transcriptional

repressor, oligodendrocyte transcription factor 1 (Olig1), which

we show can be regulated by FXR agonists (Figure S1B) and

contains an FXRE in its genomic loci. However, since the hepatic

expression of Olig1 is very low (data not shown), we have not

studied this gene further.

To confirm the presence of FXREs at the loci of these putative

transcriptional repressors, we analyzed a second independent

FXR ChIP-seq dataset from mouse livers (Thomas et al., 2010).

This analysis verified that Crip2 (Figure 2A), MafG (Figure 2B),

Zfp385a (Figure 2C), and Shp (Figure S1C) contained one or

more FXREs at their genomic loci. To investigate whether these

genes were regulated in response to FXR and FXR agonists, we

measured the hepatic expression ofCrip2,MafG, and Zfp385a in

wild-type and Fxr�/� mice (KO) pre-treated for 3 days with either

GW4064 or GSK2324. We utilized a dose of 60 mpk/day to

directly compare the effects of the two agonists. In all experi-

ments, Shp and/or Bsep, both well-characterized FXR-target

genes (Ananthanarayanan et al., 2001; Goodwin et al., 2000),

served as positive controls. Hepatic MafG, Crip2, Zfp385a, and

Shp mRNA levels were all significantly induced following treat-

ment of wild-type mice, but not Fxr�/� mice (KO), with either

FXR agonist (Figure 2D), demonstrating that induction was FXR

dependent. Further, induction of each gene was greater after

treatment with GSK2324 as compared to GW4064 (Figure 2D),

consistent with increased potency of GSK2324. In the case of

MAFG, the levels of protein were induced 2- to 3-fold following

treatment of mice with either GW4064 or GSK2324, and this ef-

fect was specific, as it was not observed in Fxr�/� mice treated

with either agonist (Figure 2E).

In order to determine whether cholic acid, an endogenous

ligand that activates a number of receptors, including FXR

(Makishima et al., 1999; Parks et al., 1999), induces these

same genes, we fed wild-type and Fxr�/� mice (KO) a diet con-

taining non-toxic levels of cholic acid (0.2%) for 7 days. The

hepatic expression of MafG and Zfp385a (Figure 2F), as well as

the positive controls Bsep (Figure 2F) and Shp (Figure S1D),

were modestly induced when wild-type, but not Fxr�/� mice,

were fed the cholic acid-enriched diet. In contrast, the increase

in Crip2 mRNA levels did not reach statistical significance (Fig-

ure 2F). Taken together, these results demonstrate that MafG,

Crip2, and Zfp385a, which encode putative transcriptional re-

pressors, are induced by specific FXR agonists (GSK2324 and

GW4064), while MafG and Zfp385a are also induced by cholic

acid, an endogenous FXR ligand.

To determine the optimal dose of GSK2324, we treated

wild-type mice for 3 days with vehicle or 10, 30, or 100 mpk

GSK2324. Induction of the classic FXR target genes Shp

and Bsep was dependent upon the dose of GSK2324 with


A B C

D E

FG

H I

Figure 2. Identification of Transcriptional Repressors as Direct FXR Target Genes

(A–C) ChIP-seq analysis of hepatic FXR from Chong et al. (2010) (top) and Thomas et al. (2010) (bottom) at (A) Crip2, (B) MafG, and (C) Zfp385a genomic loci.

(D) Hepatic expression of Shp,MafG, Crip2, and Zfp385a in C57BL/6 wild-type (WT) and Fxr�/� (KO) mice treated with vehicle, GW4064, or GSK2324 for 3 days

(n = 7–9 mice/group).

(E) Western blotting analysis and quantification of MAFG protein in livers of C57BL/6 wild-type and Fxr�/�mice treated with vehicle or GW4064 (top), or GSK2324

(bottom) for 3 days.

(F) Hepatic expression of Bsep,Crip2,MafG, and Zfp385a in C57BL/6 wild-type and Fxr�/�mice fed either a control (Ctr) or 0.2% cholic acid (CA) diet for 7 days.

(G) Hepatic expression ofCrip2,MafG, and Zfp385a in C57BL/6 wild-typemice treated with vehicle (Ctr) or 10, 30, or 100mpk/day of GSK2324 for 3 days (n = 4–8

mice/group).

(H) Hepatic expression ofCrip2,MafG, and Zfp385a following treatment of wild-type mice with a single injection of GSK2324 (30mpk) 1, 2, or 4 hr before sacrifice

(n = 6 mice/group).

(I) Hepatic expression of Crip2, MafG, and Zfp385a in littermate C57BL/6 wild-type (Flox) or liver-specific Fxr�/� mice (L-KO) treated with GSK2324 for 3 days

(n = 7–9 mice/group). All data are shown as mean ± SEM. Asterisks indicate statistically significant differences comparing WT or KO vehicle-treated against

agonist-treated mice (**p < 0.01; ***p < 0.001).


near-maximal effects at 30 mpk (Figure S2A). Induction of Crip2,

MafG, and Zfp385a was also dependent upon the dose of

GSK2324 (Figure 2G). Further, a single dose of GSK2324 at 30

mpk resulted in a significant induction of Crip2, MafG, and

Zfp385a (Figure 2H) andShp andBsep (Figure S2B)mRNA levels

within 1 hr. To investigate whether induction of Crip2,MafG, and

Zfp385a in response to GSK2324 is dependent upon hepatic

FXR expression, we treated wild-type (Fxrflox/flox) mice or litter-

mates lacking hepatic Fxr (FxrL�/L�; L-KO) for 3 days with the

agonist at 30 mpk. While GSK2324 treatment of Fxrflox/flox mice

led to increased hepatic expression of Crip2, MafG, Zfp385a,

and Shp, induction of these genes was not observed following

GSK2324 treatment of FxrL�/L� mice (Figure 2I; Figure S2C).

Taken together, our results demonstrate that Crip2, MafG, and

Zfp385a are acutely and robustly induced following activation

of hepatic FXR.

MafG Overexpression In Vivo Decreases Cyp8b1 andBiliary Cholic Acid LevelsTo investigate the potential repressive effects of MafG, Crip2,

and Zfp385a on gene expression, we generated, and then in-

jected, adenoviral vectors to acutely overexpress these three

proteins in the livers of wild-type mice. We then measured the

hepatic mRNAs of the two major bile acid synthetic enzymes,

as these are robustly repressed following activation of FXR (Fig-

ure 1B). Ad-MafG, but not Ad-Crip2 or Ad-Zfp385a treatment, re-

sulted in a decrease in the hepatic levels of Cyp8b1 (Figure 3A).

In contrast, the expression of Cyp7a1 was unaffected by any of

these treatments (Figure 3A).

The MafG-dependent repression of Cyp8b1 suggests that

MAFG may regulate the synthesis of cholic acid and alter the

bile acid pool composition. To determine whether MAFG overex-

pression could indeed change the biliary pool composition,

we treated a new cohort of mice with either Ad-control or Ad-

MafG adenovirus. Changes in the composition of the bile acid

pool are relatively slow under normal conditions, since only 5%

of the bile acids are excreted each day during multiple enterohe-

patic cycles (de Aguiar Vallim et al., 2013a; Hofmann and Hagey,

2008). Consequently, we fed mice either control diet (chow) or

the same diet supplemented with a bile acid sequestrant

(0.25% Colesevelam/Welchol) for 7 days prior to treatment

with Ad-control or Ad-MafG. Bile acid sequestrants bind bile

acids in the intestine, preventing their re-absorption in the ileum

and promoting bile acid excretion in the feces (Hofmann and Ha-

gey, 2008). The result is impaired enterohepatic re-circulation of

bile acids, de-repression of genes involved in bile acid synthesis,

and changes in the bile acid pool size (Hofmann and Hagey,

2008; Kong et al., 2012). As expected, Ad-MafG increased the

hepatic expression of MAFG mRNA (Figure 3B) and protein (Fig-

ure 3C and full blot in Figure S2D), regardless of the presence

or absence of the bile acid sequestrant in the diet. The bile

acid sequestrant-containing diet increased basal expression of

Cyp8b1, as expected (Figure 3D). Importantly, Ad-MafG treat-

ment of both the control and Colesevelam-fed mice resulted in

a significant repression ofCyp8b1mRNA (Figure 3D). Consistent

with this decrease inCyp8b1mRNA,we observed a concomitant

decrease in cholic acid and an increase in muricholic acid levels

in the bile of Ad-MafG-treated mice (Figure 3E). Moreover, the

MafG-dependent changes in bile acids were qualitative, not

quantitative, as total bile acid levels in the liver, intestine, or

gall bladder (Figures S2E–S2G) were similar in mice treated

with Ad-control, Ad-MafG, Ad-Crip2, or Ad-Zfp385a. Thus, these

two studies demonstrate that the MafG-dependent repression

of Cyp8b1 is sufficient to decrease cholic acid and increase

muricholic acid levels, even after acute (7 days) and modest in-

creases (2- to 3-fold) in MAFG protein.

To determine whether the pathway described here is

conserved in human cells, we treated the human hepatoma

cell line HepG2 with increasing concentrations of CDCA, a

natural FXR agonist (Makishima et al., 1999; Parks et al., 1999).

Consistent with our observations in mouse liver, CDCA

treatment increased the expression of both MAFG (Figure 3F)

and SHP (Figure S2H). As expected, CDCA treatment also

decreased CYP7A1 and CYP8B1 expression in a dose-depen-

dent manner (Figure 3G). Finally, Ad-MafG, but not Ad-Crip2

or Ad-Zfp385a, treatment of HepG2 cells decreased CYP8B1

expression without affecting CYP7A1 (Figure 3H). Thus, we

have identified a pathway whereby MAFG represses CYP8B1

expression both in mice and in human cells.

To further characterize the FXR-dependent regulation of

MafG, we carried out more in-depth analysis. The MafG gene

is reported to contain two transcriptional start sites that corre-

spond to exon 1a or exon 1b (Katsuoka et al., 2005a) (Figure 3I).

However, our analysis of hepatic RNA-seq data fromMenet et al.

(2012) (Figure 3I, upper panel) together with the data of Katsuoka

et al. (2005a) suggest that in the liver, MafG is preferentially

transcribed from exon 1b (Figure 3I). Thus, the putative FXRE

identified by ChIP-seq analysis (Figure 2B) would reside in the

hepatic MafG proximal promoter that lies upstream of exon 1b

(Figure 3I). Consequently, we generated a luciferase reporter

gene controlled by the hepatic MafG promoter (upstream of

exon 1b). Treatment of cells with GSK2324, following co-trans-

fection of plasmids expressing FXR and the reporter gene, led

to a robust increase in luciferase activity (Figure 3J). In contrast,

the FXR- and GSK2324-dependent increase in luciferase activity

was abolishedwhen the FXRE, which corresponds to an inverted

repeat 1 (IR-1), was mutated in the MafG promoter construct

(Figure 3J). Taken together, these data demonstrate that MafG

is a bona fide FXR target gene that contains a functional FXRE

in its hepatic proximal promoter.

MAFG Overexpression in Mouse Liver RepressesNumerous Genes Involved in Bile Acid MetabolismMAFG is a member of the small MAF family of transcription

factors, composed of MAFG, MAFF, and MAFK, that lack an

activation domain and therefore are considered transcriptional

repressors (Motohashi et al., 2002). Small MAF proteins can

bind to DNA as either homo- or hetero-dimers and function as

transcriptional repressors. Alternatively, they can heterodimerize

with transcriptional activators to induce gene expression (Moto-

hashi et al., 2002). Consistent with their role as transcription

factors, we show that epitope-tagged MAFG localizes to the

nucleus (Figure S3A). MafG is expressed in several metabolic

tissues, including the liver (Figure S3B), the major site for bile

acid synthesis.

To determine if MAFG overexpression regulated additional

genes of bile acid metabolism, in addition to Cyp8b1, we carried

out gene expression profiling of livers of mice treated with either


F

A B C D E

G H

I J

Figure 3. MafG Overexpression Represses Cyp8b1 mRNA and Reduces Biliary Cholic Acid Levels

(A) Hepatic expression of Cyp7a1 and Cyp8b1 following treatment of C57BL/6 mice with Ad-control (Ad-Ctr), Ad-Crip2, Ad-MafG, or Ad-Zfp385a adenoviruses

for 5 days (n = 7–8 mice/group).

(B–E) Hepatic levels of (B)MafGmRNA, (C) MAFG protein, and (D) Cyp8b1mRNA and (E) taurine-conjugated biliary bile acid levels in C57BL/6 wild-type treated

with Ad-control or Ad-MafG adenovirus for 7 days and fed either a control or Colesevelam-containing diet (Colesev) for 7 days prior to and 7 days post-adenovirus

treatment (n = 8–9 mice/group).

(F and G) Expression levels of (F)MAFG or (G) CYP7A1 and CYP8B1 in human HepG2 cells treated with 100, 150, or 200 mM chenodeoxycholic acid (CDCA) for

24 hr (n = 4 wells/condition).

(H) Expression levels of CYP7A1 and CYP8B1 in HepG2 cells infected with Ad-Control, -Crip2, -MafG, or -Zfp385a for 36 hr (n = 3–4 wells/condition).

(I) RNA-seq (Menet et al., 2012) (top) and FxrChIP-seq (Chong et al., 2010) (bottom) analysis of theMafG genomic loci showing locations of MafG exons and FXRE

in the putative MafG proximal promoter.

(J) Wild-type and FXRE mutant (mutated bases are bolded and underlined) MafG promoter (MafG prom) constructs upstream of a luciferase reporter gene were

transfected into HepG2 cells with increasing amounts of a FXR expression plasmid and treated with vehicle or GSK2324 for 24 hr. Luciferase activity was

normalized to b-galactosidase and expressed as fold change. All data are shown as mean ± SEM. Asterisks indicate statistically significant differences versus

controls (**p < 0.01; ***p < 0.001). Different letters (a–d) indicate statistically significant differences (p < 0.05).


control or MafG adenovirus (Figure 4A). MAFG overexpression

repressed 554 genes and activated 833 genes. Induction of

genes in response to MAFG overexpression is not unexpected,

as MAFG can heterodimerize with transcriptional activators,

most notably NRF2 and NRF3, to activate specific genes (Kat-

suoka et al., 2005a, 2005b). Gene ontology analysis (Huang

et al., 2009) of the hepatic genes that are repressed following

MAFG overexpression revealed a significant enrichment in pri-

mary bile acid synthesis genes (Figure 4B) that included

Cyp8b1 and Cyp7b1, as well as the bile acid importer, sodium

taurocholate co-transporting polypeptide (Ntcp, gene symbol

Slc10a1). Interestingly, these same genes are also repressed in

mouse liver following treatment with FXR agonists (Figures 1B–

1D), suggesting that the repression of these genes by FXR ago-

nists is indirect and likely dependent upon induction of MAFG

mRNA and protein levels.

We also determined the expression of the remaining bile acid

synthesis genes in Ad-control and Ad-MafG-treated mice that

had been fed either normal chow or a diet supplemented with

0.25% Colesevelam. Ad-MafG treatment resulted in repression

of almost all bile acid synthesis genes, including Cyp7b1 and

Cyp27a1, independent of the diet (Figure 4C). Interestingly, com-

parison of the data shown in Figures 4C and 1D demonstrates

that the bile acid synthetic genes that are repressed following

FXR activation are also repressed following overexpression of

MAFG. The repression of bile acid synthesis genes is not univer-

sal, since Amacr and Cyp39a1 mRNA levels were unchanged

following treatment ofmicewith FXR agonists or following hepat-

ic overexpression of MAFG (Figures 1D and 4C). Together, these

data suggest that MAFG is an important transcriptional regulator

of bile acid synthesis andmay play an important role inmediating

the FXR-dependent repression of genes involved in bile acid

metabolism.

Loss of MAFG Causes De-Repression of Multiple BileAcid Synthetic Genes and Increases Biliary Cholic AcidLevelsTo further evaluate the role of MafG in regulating bile acid meta-

bolism, we investigated the effect of loss of MafG. Short-term

silencing of MafG in isolated mouse hepatocytes using three

distinct shRNA constructs (Figure 5A) resulted in de-repression

of Cyp8b1 (Figure 5B). Further, siRNA-mediated silencing of

A B

C

Figure 4. MAFG Regulates Several Genes Involved in Bile Acid Synthesis

(A) Microarray analysis of livers from mice treated with either Ad-control or Ad-MafG for 7 days. Lines delineate fold-change cut-off (1.5-fold). Red and blue dots

indicate genes that are repressed or induced genes, respectively (n = 3/condition).

(B) KEGG pathway analysis for categories that were significantly enriched from global analysis of repressed genes in (A).

(C) Hepatic expression of genes involved in bile acid synthesis or transport following treatment of C57BL/6 wild-type with Ad-control or Ad-MafG adenovirus for

7 days and fed either a control or Colesevelam-containing diet for 7 days prior to and 7 days post-adenovirus treatment (n = 8–9 mice/group). All data are shown

as mean ± SEM. Asterisks indicate statistically significant differences versus controls on the same diet (**p < 0.01; ***p < 0.001).


MAFG mRNA and protein levels in human HepG2 cells (Fig-

ure 5C) also led to de-repression of CYP8B1 (Figure 5D).

Together, these results suggest that MafG is a critical negative

regulator of Cyp8b1 expression in both mice and humans. We

then generated an antisense oligonucleotide (ASO) to silence

MafG in vivo. Acute treatment with the MafG ASO significantly

decreased the levels ofMafGmRNA (Figure 5E) and protein (Fig-

ure 5F, full blot in Figure S4A). Notably,Cyp8b1, but not Cyp7a1,

was de-repressed in mice after MafG silencing with the ASO

treatment (Figure 5G), thus recapitulating our in vitro findings in

an in vivo setting.

To determine whether complete loss ofMafG also affected bile

acidhomeostasis,weobtainedMafG+/�mice,whichare reported

to generate viableMafG�/� mice on a mixed background (Shavit

et al., 1998).We then backcrossedMafG+/�mice onto aC57BL/6

background for 10 generations to be consistent with the genetic

background of all other mice used in the current studies. Unex-

pectedly, we failed to recover any MafG�/� mice on a C57BL/6

background (data not shown). We conclude that complete loss

ofMafG on a C57BL/6 background is lethal, likely a result of pro-

nounced neurological disorders previously reported in MafG�/�

mice on amixed genetic background (Shavit et al., 1998). Conse-

quently, our studies are limited to heterozygous MafG+/� mice

and their wild-type littermates. As expected, hepatic MafG

mRNA and protein levels are decreased approximately 50%

in MafG+/� mice (Figure 5H). Importantly, Cyp8b1, but not

Cyp7a1, mRNA levelswere induced/de-repressed in theMafG+/�

mouse liver (Figure 5I). Other genes, such as Cyp7b1 and

Cyp27a1, thatwehaveshownare repressed followingMafGover-

expression (Figure 4C) were also de-repressed in the livers of

chow-fed MafG+/� mice (Figure 5I). Importantly, partial loss of

MafGmRNA and protein led to a significant increase in the biliary

levels of cholic acid and decreased muricholic acid levels in

MafG+/� mice (Figure 5J), consistent with de-repression of

Cyp8b1. MafG+/� mice do not have significantly altered total

bile acid levels in liver, intestine, or gall bladder (Figures S4B–

S4D), suggesting that MAFG regulates the bile acid pool compo-

sition, but not the pool size. We quantified the expression of

multiple genes encoding enzymes of the bile acid synthetic

pathway in the livers of wild-type and MafG+/� mice. Partial loss

of MAFG caused de-repression of several genes, including

Acox2, Akr1d1, Akr1c14, Hsd17b4, Ntcp, and Scp2 (Figure 5K).

Collectively, these results support the hypothesis that hepatic

MAFG functions asa repressor ofCyp8b1andcholic acid synthe-

sis as well as a regulator of bile acid metabolism in vivo.

Identification of MAFG Binding Sites at Multiple GenesInvolved in Bile Acid Synthesis and MetabolismTo investigate the molecular mechanism for the MafG-depen-

dent repression of Cyp8b1 as well as additional target genes,

A

H I J K

B C D E F G

Figure 5. Loss of MafG Results in De-Repression of Several Bile Acid Synthetic Genes(A and B) Expression of (A) MafG and (B) Cyp8b1 in primary mouse hepatocytes treated with control (Ad-sh-LacZ) or three differentMafG shRNA adenoviruses

(Ad-sh-MafG 1–3) (n = 4 wells/condition).

(C and D) Shown are (C)MAFGmRNA and protein (inset; bA, b-actin) and (D)CYP8B1 expression in HepG2 cells treated with control orMAFG siRNA (n = 3 wells/

condition).

(E) Hepatic MafG mRNA in C57BL/6 wild-type mice treated with control or MafG ASO (100 mpk) for 3 days (n = 9 mice/group).

(F) Hepatic MafG protein from mice treated as in (E).

(G) Hepatic Cyp7a1 and Cyp8b1 mRNA from mice treated as in (E).

(H and I) Shown are (H)MafGmRNA and protein (top) and (I)Cyp7a1,Cyp8b1,Cyp27a1, and Cyp7b1mRNA levels in littermate wild-type andMafG+/� (Het) mice

(n = 7–11 mice/group).

(J) Biliary bile acid levels were determined from individual littermate wild-type and MafG+/� mice (n = 7–11/group).

(K) Hepatic expression of bile acid synthesis genes in littermate wild-type andMafG+/�mice (n = 7–11mice/group). All data are shown as mean ± SEM. Asterisks

indicate statistically significant differences versus controls or wild-type (*p < 0.05; **p < 0.01; ***p < 0.001).


we generated an adenovirus construct to overexpress a biotin-

ligase recognition peptide (BLRP)-tagged MafG. Consistent

with our studies using untagged MAFG (Ad-MafG) (Figures 3A–

3D), treatment ofmicewith Ad-BLRP-MafG resulted in increased

hepatic MafG mRNA (Figure S5A) and protein (Figure 6A) and

decreased Cyp8b1, Cyp7b1, and Cyp27a1 expression (Fig-

ure S5B), suggesting the BLRP epitope does not interfere with

MAFG function. We then used ChIP analysis to identify MAFG

bound to MAFG response elements (MAREs). First, as controls

we show that in mouse liver, BLRP-tagged MAFG was enriched

at MAREs that had been previously identified in Nqo and G6pdx

in studies using cultured cells (Hirotsu et al., 2012) (Figure S5C).

We then carried out ChIP-seq analysis from livers of mice treated

with Ad-BLRP-MafG as well as Ad-BLRP (control), using the

same anti-BLRP antibody. Global analysis of all peaks for

MAFG revealed that 46% reside in intergenic regions, as

compared to 41% in introns, while there was a modest enrich-

ment in proximal promoters (Figure 6B). Motif enrichment anal-

ysis of sequences for the top 20,000 MAFG peaks (representing

the largest number of reads per site) identified the consensus

MARE (Figure 6C, top). This sequence is highly similar to the

MARE described previously for MAFG homodimers (Hirotsu

et al., 2012). Analysis of all MAFG ChIP-seq peaks (n = 68,754)

identified a MARE that contained a consensus sequence of

GTCAGC (Figure 6C, bottom) but was otherwise different from

that found in the top 20,000 peaks. Presumably, the latter

A

D

G H I

E F

B C

Figure 6. Identification of MAREs in Bile Acid Synthetic Genes

(A) Hepatic levels of MAFG protein levels (detected using anti-BLRP or anti-MAFG antibodies) in C57BL/6 mice treated with either control (Ad-BLRP) or BLRP-

tagged MAFG adenovirus (Ad-BLRP-MafG).

(B) Global frequency of hepatic MAFG binding sites (MAREs) across the genome relative to gene location (expressed as a percentage).

(C) Table showing the motif for the top 20,000 peaks in MAFG ChIP-seq peaks (top) and in all 68,754 peaks (bottom), with statistical significance and percent

occurrence in peaks (target) or background (bkgd).

(D) ChIP-seq analysis of MAREs in chromatin isolated from livers of mice treated with Ad-BLRP (control; top) or Ad-BLRP-MafG (bottom) at Cyp8b1 locus.

(E) ChIP analysis of MafG occupancy at the Cyp8b1 promoter region determined by qRT-PCR (primer locations to scale, y axis).

(F) Wild-type and MARE Cyp8b1 promoter luciferase constructs were transfected into HepG2 cells and co-transfected with increasing amounts of a MAFG

expression plasmid. Luciferase activity was normalized to b-galactosidase and expressed as fold change.

(G and H) ChIP-seq analysis of MAREs in chromatin isolated from livers of mice treated with Ad-BLRP (control; top) or Ad-BLRP-MafG (bottom) at loci for (G)

Cyp27a1 or (H) Cyp7b1.

(I) Cyp7b1 promoter-luciferase reporter transfected into HepG2 cells together with increasing amounts of a MAFG expression plasmid. Luciferase activity was

normalized to b-galactosidase and expressed as fold change. All data are shown as mean ± SEM. Asterisks indicate statistically significant differences from

control (**p < 0.01; ***p < 0.001).


MARE represents various binding sites for complexes containing

different MAFG-containing heterodimers. The MARE motif iden-

tified in the top MAFG 20,000 peaks was present in 59% of all

peaks, whereas the same motif was present in 4% of the Ad-

BLRP-control (Figure 6C). Similarly, analysis for all 68,754 peaks

identified the differentMAREmotif in 42%of peaks, compared to

4% in the background control (Figure 6C).

We next searched the Cyp8b1 locus of the MAFG ChIP-seq

data. This identified a number of MAREs within 10 kb of the

Cyp8b1 gene (Figure 6D, lower panel) and one peak present

near the transcriptional start site (TSS). We confirmed the enrich-

ment of MAFG in the Cyp8b1 proximal promoter by qRT-PCR

ChIP analysis (Figure 6E). Complementary analysis of the

Cyp8b1 promoter (0.5 kb) using luciferase reporter assays

demonstrated a dose-dependent repression following MAFG

overexpression (Figure 6F). In contrast, mutation of the MARE

within the Cyp8b1 promoter resulted in de-repression of the

luciferase reporter gene, which was no longer repressed by

MAFG overexpression (Figure 6F). These results demonstrate

direct binding of MAFG to multiple sites upstream of Cyp8b1,

and to one site in the proximal promoter, that identify the molec-

ular mechanism for the MAFG-dependent repression ofCyp8b1.

MAFG ChIP-seq analysis also identified MAREs in the pro-

moter and/or intronic regions of Cyp27a1 (Figure 6G, lower

panel) and Cyp7b1 (Figure 6H, lower panel). In contrast, peaks

for MAFG binding sites were not present in ChIP-seq data from

mice treated with the control Ad-BLRP (Figures 6D, 6G, and

6H, upper panels). Analysis of the proximalCyp7b1 promoter us-

ing a luciferase reporter gene showed that MAFG overexpres-

sion reduced luciferase activity in a dose-dependent manner

(Figure 6I). Further, we also identified MAREs in several other

bile acid synthetic genes, including Acox2, Akr1d1, Akr1c14,

Ntcp, Hsd17b4, and Scp2 (Figures S5D–S5I), but not in the

100 kb upstream of Cyp7a1 (Figure S5J). Together, these data

demonstrate that MAFG directly regulates several bile acid

metabolism genes.

Interestingly, liver-specific LRH-1-deficient mice also have

decreased Cyp8b1, unchanged Cyp7a1, and altered bile acid

composition (Lee et al., 2008; Mataki et al., 2007). We therefore

investigated whether MAFG binding sites were associated with

LRH-1 occupancy in the liver. UsingChIP-seq analysis for hepat-

ic LRH-1 sites (Chong et al., 2012), we only identified a small

number of genes that had both MAFG and LRH-1 binding sites

(282 of 10,351; Figure S6A), and these genes were enriched in

genes of negative regulation of metabolic processes, but not

bile acid synthesis genes (Figure S6B). Taken together, these

results suggest that MAFG is unlikely to repress transcription

of multiple bile acid synthetic genes by displacing LRH-1.

In conclusion, our extensive studies identify a pathway

involving the nuclear receptor FXR and the FXR-target gene

MafG that functions to repress transcription of Cyp8b1 as well

as multiple bile acid genes, including Acox2, Akr1d1, Akr1c14,

Cyp7b1, Cyp27a1, Hsd17c14, Ntcp, and Scp2, and thus modu-

late bile acid homeostasis (Figure 7).

DISCUSSION

The current studies identify an FXR-MAFG pathway that controls

the transcription of multiple genes involved in both the classic

and alternative pathways of bile acid synthesis, and bile acid

transport. We show that the MafG gene is a direct target

of FXR and that MAFG subsequently represses many genes

involved in bile acid synthesis and metabolism (Figures 1, 2, 3,

4, and 7). Further, we show that loss of R50% hepatic MAFG

leads to de-repression of many of these genes (Figure 5). Impor-

tantly, we demonstrate that MAFG binds to MAREs associated

with the same repressed genes (Figures 6 and 7). Cyp8b1 or

Cyp7b1 promoter-reporter assays provided additional insight

into the functional importance of selected MAREs (Figure 6).

The identification of a MAFG-dependent regulation of Cyp8b1,

Acox2, Akr1d1, Akr1c14, Cyp27a1, Cyp7b1, Hsd17c4, Ntcp,

and Scp2 suggests a concerted action of MAFG in regulating

various aspects of bile acid metabolism that has not been

previously appreciated (Figure 1A).

Consistent with the finding that hepatic overexpression of

MAFG inmice repressesCyp8b1, we show that under these con-

ditions there is a decrease in biliary cholic acid and an increase

in muricholic acid levels (Figure 3) without increasing bile acid

levels in liver, intestine, or gall bladder (Figures S2E–S2G). This

finding is consistent with Cyp8b1 encoding the regulatory

enzyme for cholic acid synthesis from 7-hydroxycholesterol

and the earlier observation that Cyp8b1�/� mice not only fail to

synthesize cholic acid, but exhibit increased muricholic acid

levels in bile, without a change in the bile acid pool size (Li-Haw-

kins et al., 2002). In contrast, loss of MAFG, as a result of partial

gene ablation or silencing, caused de-repression ofCyp8b1 (Fig-

ure 5) and an increased ratio of cholic acid:muricholic acid

without altering total bile acid levels in liver, intestine, and gall

bladder (Figures S4B–S4D). This change in the bile acid compo-

sition is expected to alter the hydrophobicity. However, the

Figure 7. Summary for the Role of MAFG in Regulating Bile Acid

Synthesis and Transport

The cartoon shows the activation of theMafG gene following FXR activation by

various ligands (yellow boxes) and the subsequent targeting of MAFG protein

(purple) to several MAREs. MAREs identified in the current study that lie within

the genomic loci of MAFG-repressed genes (shown in red) that encode pro-

teins involved in bile acid synthesis or metabolism are identified. Below is a

simplified version of the classic and alternative bile acid synthetic pathways

that generate cholic acid (CA), chenodeoxycholic acid (CDCA), or muricholic

acid (MCA).


physiologic consequences of such a change onmetabolism as a

whole are unknown and will require additional studies.

We did not observe changes in Cyp7a1 mRNA levels in the

MafG+/�mice or in ASO-treated wild-typemice (Figure 5), further

supporting the specificity of the regulation of specific bile acid

genes by MAFG. Nonetheless, after prolonged MAFG overex-

pression, we have observed some repression of Cyp7a1 (data

not shown). This effect was not consistent across all our studies.

Since MafG silencing and MafG+/� mice do not exhibit changes

in Cyp7a1 expression (Figure 5), and MAFG ChIP-seq analysis

at the Cyp7a1 locus did not identify MAFG binding sites (Fig-

ure S5J), we suggest that the repression of Cyp7a1 may be

indirect.

The current studies suggest that MAFG represents a compli-

mentary pathway that is critical for the regulation of bile acid ho-

meostasis. Previous studies reported that the FXR-dependent

regulation of Cyp8b1 involves Shp (Kerr et al., 2002; Wang

et al., 2002), although the authors suggested at that time that

additional unknown pathways were likely to play a role in the

repression of bile acid synthetic genes. Earlier in vitro studies

had shown that SHP can repress luciferase reporter gene activity

by binding to and inhibiting HNF4a and LRH-1 transcription

factors that normally activate Cyp7a1 and/or Cyp8b1 (Goodwin

et al., 2000; Lu et al., 2000). Nonetheless, the effects of hepatic

overexpression of SHP on the expression of Cyp7a1 or

Cyp8b1 in vivo are at best very modest (Kir et al., 2012; Kong

et al., 2012), while the effects of SHP overexpression on other

bile acid synthetic genes or bile acid composition have not

been reported. Further, Kong et al. (2012) reported that treat-

ment of Shp�/� mice with the FXR agonist GW4064 resulted in

near-normal repression of both Cyp8b1 and Cyp7a1. Taken

together, these results contrast with the broad and significant

repression of numerous bile acid synthetic genes and the

decreased levels of cholic acid we observe in mice following

MAFG overexpression (Figures 3 and 5).

A second pathway that leads to repression ofCyp7a1 involves

activation of FXR in enterocytes and the resulting increase in

Fgf15 (mouse) or FGF19 (humans) and subsequent secretion of

the protein (Inagaki et al., 2005). FGF15/19 binds to the cognate

receptor, FGFR4/b-klotho, resulting in repression of Cyp7a1

(Potthoff et al., 2012). Whether the FGF15/19 pathway plays a

role in the regulation of Cyp8b1 is unclear at the present time.

A recent study also demonstrated a co-requirement for SHP in

mediating the effects of FGF19 repression of Cyp7a1 (Kir et al.,

2012). In contrast, a separate study showed that injection of

FGF15 protein into Shp�/� mice resulted in near-normal repres-

sion of both Cyp7a1 and Cyp8b1 (Kong et al., 2012). Thus, it ap-

pears that the precise role of SHP in mediating the FGF15/19

and/or FXR-dependent repression of Cyp8b1 and/or Cyp7a1 re-

mains to be established. Nonetheless, the finding that Cyp7a1

and/or Cyp8b1 mRNA levels are induced/de-repressed in cells

or the livers of mice deficient for either Shp (Kerr et al., 2002;

Wang et al., 2002), Fgfr4/b-klotho (Kong et al., 2012), or MafG

(Figure 5) suggest that SHP, FGF15/19, and MAFG represent

three complimentary pathways that control bile acid synthesis

and composition. Indeed, the existence and complexity of the

complementary pathways highlight the fact that tight regulation

of bile acid homeostasis is required, and dysregulation can

lead to various metabolic diseases.

The role of MAFG in bile acid metabolism has not been

previously appreciated. The MAF family of proteins is divided

into small (MAFG, MAFF, and MAFK) and large (cMAF, MAFA,

MAFB) members (Kannan et al., 2012). The small members

contain a DNA-binding domain and a basic leucine zipper but

lack the transcriptional activation domain found in the large

family members (Kannan et al., 2012; Motohashi et al., 2002).

Small members of the family can form homodimers or hetero-

dimers that bind MAREs to repress transcription of target genes

(Kurokawa et al., 2009). However, the small MAF proteins can

also dimerize with transcriptional activators, such as NRF2, a

member of the Cap ‘n’ Collar family of transcription factors, to

induce genes involved in the stress response and detoxification

(Kannan et al., 2012; Motohashi et al., 2002). At the current time,

the factors that control the formation of homodimers versus

heterodimers of the small MAF proteins are poorly understood.

Feedback repression of bile acid synthesis in response to

accumulating bile acids is critical for the normal maintenance

of bile acid homeostasis and for the prevention of hepato-toxicity

that occurs with elevated levels of bile acids. The identification of

the pathway described here may have important implications in

disease since bile acid metabolism is linked to several metabolic

disorders, including cardiovascular disorders, diabetes, and

specific types of cancer.

EXPERIMENTAL PROCEDURES

GSK2324 was dissolved in water and administered to mice via intraperitoneal

(i.p.) injection at 30 mg/kg body weight (mpk) unless otherwise stated. In ex-

periments where GW4064 and GSK2324 were compared, agonists were dis-

solved in water containing 0.5% Tween 80, and mice were treated once daily

with either drug or vehicle alone at 60 mpk for 3 days via i.p. injection. Unless

otherwise stated, mice were fasted for 4–6 hr after the last treatment with FXR

agonists prior to removal of tissues. All animal experiments were carried out

according to NIH guidelines and were approved by the Office of Animal

Research Oversight (OARO) at UCLA. For MafG ASO studies, male 12-

week-old C57BL/6 mice (Jackson Laboratory) were dosed once with either

control or MafG ASO at 100 mpk, and 3 days later, mice were fasted overnight

and livers collected the following morning (9–11 a.m.). All adenoviruses were

prepared in BSL2 category facilities. Briefly, cDNAs for mouse MafG, Crip2,

and Zfp385a were cloned from whole-liver cDNA into pAdTrack CMV plasmid

and prepared as described in Bennett et al. (2013). For animal experiments,

13 109 plaque-forming units (PFU) were used, and tissueswere collected after

5–7 days, and for cell culture studies, a moi of 1–10 was used, and cells were

harvested for analysis after 24–48 hr. For gene expression analysis, RNA was

isolated using QIAZOL according to the manufacturer’s instructions (QIAGEN)

and rDNaseI treated before cDNA was synthesized (Life Technologies). qRT-

PCR analysis was carried out using primers described in Table S1, and gene

expression data were normalized to Tbp and/or 36B4/Rplp0. Western blotting

analysis was carried out from liver samples (approximately 100 mg of tissue)

homogenized in 1 ml of RIPA buffer supplemented with protease inhibitor

complex (Roche). Protein was quantified using the BCA assay (Thermo Fisher

Scientific), and 10–30 mg of protein was loaded on pre-cast gels (Bio-Rad).

Protein was transferred to PVDF membranes (Millipore), probed with anti-

bodies described in the Supplemental Experimental Procedures, and detected

with ECL reagent (Sigma) or ECL Prime (GE Healthcare) using a GE Image

Quant LAS 4000 detection system (GE Healthcare). Bile acid analysis was

carried out from biliary material by HPLC, and the major taurine-conjugated

species were detected by measuring absorbance measured at 205 nm and

compared to known bile acid standards. Total bile acids were measured in

liver, intestine, and gall bladder as described in the Supplemental Experi-

mental Procedures. For promoter-reporter studies, mouse MafG promoter

(2 kb), Cyp8b1 promoter (0.5 kb), and Cyp7b1 promoter (1 kb) were amplified

from mouse genomic DNA (C57BL/6) using KAPA HiFi polymerase (Kapa) and


cloned into pGL4.10[luc2] plasmid (Promega). Luciferase reporter constructs

were transfected using Fugene HD (Promega) (n = 6 wells per condition) ac-

cording to manufacturer’s instructions into human HepG2 or Hep3B cells

(ATCC) and plated onto 48-well dishes. For MafG ChIP analysis, mice were

treated with Ad-BLRP or Ad-BLRP-MafG for 5 days. Livers were fixed in

PBS containing 1% formaldehyde, nuclei were isolated, and chromatin was

sheared by sonication for 25–30 cycles using BioRuptor Twin (Daigenode)

and immunoprecipitated using a BLRP antibody (Avi-tag, GeneScript) as

described in the Supplemental Experimental Procedures. For ChIP-seq,

immunoprecipitated DNA was used for library preparation (Kapa Biosystems)

and sequenced by the UCLA UNG Core. Analysis of ChIP- and RNA-seq, as

well as microarray analysis, is described in the Supplemental Experimental

Procedures.

Statistics

All bars shown are mean ± SEM. The comparison of different groups was

carried out with Student’s t test, one- and two-way ANOVA, and differences

under p < 0.05 were considered statistically significant (*p < 0.05, **p < 0.01,

***p < 0.001, and different letters indicate at least p < 0.05).

SUPPLEMENTAL INFORMATION

Supplemental Information includes Supplemental Experimental Procedures,

six figures, and two tables and can be found with this article online at http://

dx.doi.org/10.1016/j.cmet.2015.01.007.

AUTHOR CONTRIBUTIONS

T.Q.d.A.V. and P.A.E. conceived the project and wrote the manuscript. L.R.H.

carried out bile acid analysis, H.M. and M.Y. providedMafG+/� mice, and R.L.

and M.G. provided the MafG ASO. C.R. carried out ChIP-seq bioinformatics

analysis. All authors provided feedback during preparation of the manuscript.

T.Q.d.A.V. executed all experiments and was aided by H.A. and E.J.T.

ACKNOWLEDGMENTS

We thank Drs. Timothy Willson and David Deaton for the kind gift of GSK2324

and Peter Tontonoz and his lab for helpful discussions. We also thank Tieyan

Han, Joan Cheng, Christina Cheung, and Elizabeth Nam for excellent technical

assistance. This work was supported in part by United States Public

Health Service, grants 1R01DK102559-01 and the Laubish fund at UCLA

(to P.A.E.); American Heart Association (AHA) Beginning Grant In Aid

(13BGIA17080038) and NIH K99HL118161 (to E.J.T); and AHA Postdoctoral

Fellowship (12POST11760017) and NIH NHLBI K99HL12348501 (to C.E.R.).

T.Q.d.A.V. was supported by an AHA Scientist Development Grant

(14SDG18440015), University of California Los Angeles (UCLA) Clinical and

Translational Science Institute (CTSI) grant (UL1TR000124), a UCLA Center

for Ulcer Research and Education (CURE):Digestive Diseases Research Cen-

ter (DDRC) grant (DK41301), and a UCLA Diabetes Research Center (DRC)

grant (DK063491). R.G. Lee andM.J. Grahamare employees and shareholders

of Isis Pharmaceuticals.

Received: September 18, 2014

Revised: November 20, 2014

Accepted: January 13, 2015

Published: February 3, 2015

REFERENCES

Ananthanarayanan, M., Balasubramanian, N., Makishima, M., Mangelsdorf,

D.J., and Suchy, F.J. (2001). Human bile salt export pump promoter is trans-

activated by the farnesoid X receptor/bile acid receptor. J. Biol. Chem. 276,

28857–28865.

Bass, J.Y., Caravella, J.A., Chen, L., Creech, K.L., Deaton, D.N., Madauss,

K.P., Marr, H.B., McFadyen, R.B., Miller, A.B., Mills, W.Y., et al. (2011).

Conformationally constrained farnesoid X receptor (FXR) agonists: heteroaryl

replacements of the naphthalene. Bioorg. Med. Chem. Lett. 21, 1206–1213.

Bavner, A., Sanyal, S., Gustafsson, J.A., and Treuter, E. (2005). Transcriptional

corepression by SHP: molecular mechanisms and physiological conse-

quences. Trends Endocrinol. Metab. 16, 478–488.

Beher, W.T., Baker, G.D., Anthony, W.L., and Beher, M.E. (1961). The feed-

back control of cholestrol biosynthesis. Henry Ford Hosp. Med. Bull. 9,

201–213.

Bennett, B.J., de Aguiar Vallim, T.Q., Wang, Z., Shih, D.M., Meng, Y., Gregory,

J., Allayee, H., Lee, R., Graham, M., Crooke, R., et al. (2013). Trimethylamine-

N-oxide, a metabolite associated with atherosclerosis, exhibits complex

genetic and dietary regulation. Cell Metab. 17, 49–60.

Chong, H.K., Infante, A.M., Seo, Y.K., Jeon, T.I., Zhang, Y., Edwards, P.A., Xie,

X., and Osborne, T.F. (2010). Genome-wide interrogation of hepatic FXR re-

veals an asymmetric IR-1 motif and synergy with LRH-1. Nucleic Acids Res.

38, 6007–6017.

Chong, H.K., Biesinger, J., Seo, Y.K., Xie, X., and Osborne, T.F. (2012).

Genome-wide analysis of hepatic LRH-1 reveals a promoter binding prefer-

ence and suggests a role in regulating genes of lipid metabolism in concert

with FXR. BMC Genomics 13, 51.

de Aguiar Vallim, T.Q., Tarling, E.J., and Edwards, P.A. (2013a). Pleiotropic

roles of bile acids in metabolism. Cell Metab. 17, 657–669.

de Aguiar Vallim, T.Q., Tarling, E.J., Kim, T., Civelek, M., Baldan, A., Esau, C.,

and Edwards, P.A. (2013b). MicroRNA-144 regulates hepatic ATP binding

cassette transporter A1 and plasma high-density lipoprotein after activation

of the nuclear receptor farnesoid X receptor. Circ. Res. 112, 1602–1612.

Duran-Sandoval, D., Cariou, B., Percevault, F., Hennuyer, N., Grefhorst, A.,

van Dijk, T.H., Gonzalez, F.J., Fruchart, J.C., Kuipers, F., and Staels, B.

(2005). The farnesoid X receptor modulates hepatic carbohydrate metabolism

during the fasting-refeeding transition. J. Biol. Chem. 280, 29971–29979.

Evans, R.M., and Mangelsdorf, D.J. (2014). Nuclear Receptors, RXR, and the

Big Bang. Cell 157, 255–266.

Goodwin, B., Jones, S.A., Price, R.R., Watson, M.A., McKee, D.D., Moore,

L.B., Galardi, C., Wilson, J.G., Lewis, M.C., Roth, M.E., et al. (2000). A regula-

tory cascade of the nuclear receptors FXR, SHP-1, and LRH-1 represses bile

acid biosynthesis. Mol. Cell 6, 517–526.

Hirotsu, Y., Katsuoka, F., Funayama, R., Nagashima, T., Nishida, Y.,

Nakayama, K., Engel, J.D., and Yamamoto, M. (2012). Nrf2-MafG hetero-

dimers contribute globally to antioxidant and metabolic networks. Nucleic

Acids Res. 40, 10228–10239.

Hofmann, A.F., and Hagey, L.R. (2008). Bile acids: chemistry, pathochemistry,

biology, pathobiology, and therapeutics. Cell. Mol. Life Sci. 65, 2461–2483.

Huang, W., Sherman, B.T., and Lempicki, R.A. (2009). Systematic and integra-

tive analysis of large gene lists using DAVID bioinformatics resources. Nat.

Protoc. 4, 44–57.

Inagaki, T., Choi, M., Moschetta, A., Peng, L., Cummins, C.L., McDonald, J.G.,

Luo, G., Jones, S.A., Goodwin, B., Richardson, J.A., et al. (2005). Fibroblast

growth factor 15 functions as an enterohepatic signal to regulate bile acid

homeostasis. Cell Metab. 2, 217–225.

Inagaki, T., Moschetta, A., Lee, Y.K., Peng, L., Zhao, G., Downes, M., Yu, R.T.,

Shelton, J.M., Richardson, J.A., Repa, J.J., et al. (2006). Regulation of antibac-

terial defense in the small intestine by the nuclear bile acid receptor. Proc. Natl.

Acad. Sci. USA 103, 3920–3925.

Kannan, M.B., Solovieva, V., and Blank, V. (2012). The small MAF transcription

factors MAFF, MAFG and MAFK: current knowledge and perspectives.

Biochim. Biophys. Acta 1823, 1841–1846.

Katsuoka, F., Motohashi, H., Engel, J.D., and Yamamoto, M. (2005a). Nrf2

transcriptionally activates the mafG gene through an antioxidant response

element. J. Biol. Chem. 280, 4483–4490.

Katsuoka, F., Motohashi, H., Ishii, T., Aburatani, H., Engel, J.D., and

Yamamoto, M. (2005b). Genetic evidence that small maf proteins are essential

for the activation of antioxidant response element-dependent genes. Mol. Cell.

Biol. 25, 8044–8051.

Kerr, T.A., Saeki, S., Schneider, M., Schaefer, K., Berdy, S., Redder, T., Shan,

B., Russell, D.W., and Schwarz, M. (2002). Loss of nuclear receptor SHP




impairs but does not eliminate negative feedback regulation of bile acid syn-

thesis. Dev. Cell 2, 713–720.

Kim, I., Ahn, S.H., Inagaki, T., Choi, M., Ito, S., Guo, G.L., Kliewer, S.A., and

Gonzalez, F.J. (2007). Differential regulation of bile acid homeostasis by the

farnesoid X receptor in liver and intestine. J. Lipid Res. 48, 2664–2672.

Kir, S., Zhang, Y., Gerard, R.D., Kliewer, S.A., and Mangelsdorf, D.J. (2012).

Nuclear receptors HNF4a and LRH-1 cooperate in regulating Cyp7a1 in vivo.

J. Biol. Chem. 287, 41334–41341.

Kong, B., Wang, L., Chiang, J.Y., Zhang, Y., Klaassen, C.D., and Guo, G.L.

(2012). Mechanism of tissue-specific farnesoid X receptor in suppressing the

expression of genes in bile-acid synthesis in mice. Hepatology 56, 1034–1043.

Kuipers, F., Bloks, V.W., and Groen, A.K. (2014). Beyond intestinal soap—bile

acids in metabolic control. Nat. Rev. Endocrinol. 10, 488–498.

Kurokawa, H., Motohashi, H., Sueno, S., Kimura, M., Takagawa, H., Kanno, Y.,

Yamamoto, M., and Tanaka, T. (2009). Structural basis of alternative DNA

recognition by Maf transcription factors. Mol. Cell. Biol. 29, 6232–6244.

Lee, Y.K., Schmidt, D.R., Cummins, C.L., Choi, M., Peng, L., Zhang, Y.,

Goodwin, B., Hammer, R.E., Mangelsdorf, D.J., and Kliewer, S.A. (2008).

Liver receptor homolog-1 regulates bile acid homeostasis but is not essential

for feedback regulation of bile acid synthesis. Mol. Endocrinol. 22, 1345–1356.

Lee, F.Y., de Aguiar Vallim, T.Q., Chong, H.K., Zhang, Y., Liu, Y., Jones, S.A.,

Osborne, T.F., and Edwards, P.A. (2010). Activation of the farnesoid X receptor

provides protection against acetaminophen-induced hepatic toxicity. Mol.

Endocrinol. 24, 1626–1636.

Li, F., Jiang, C., Krausz, K.W., Li, Y., Albert, I., Hao, H., Fabre, K.M., Mitchell,

J.B., Patterson, A.D., andGonzalez, F.J. (2013). Microbiome remodelling leads

to inhibition of intestinal farnesoid X receptor signalling and decreased obesity.

Nat. Commun. 4, 2384.

Li-Hawkins, J., Gafvels, M., Olin, M., Lund, E.G., Andersson, U., Schuster, G.,

Bjorkhem, I., Russell, D.W., and Eggertsen, G. (2002). Cholic acid mediates

negative feedback regulation of bile acid synthesis in mice. J. Clin. Invest.

110, 1191–1200.

Lu, T.T., Makishima, M., Repa, J.J., Schoonjans, K., Kerr, T.A., Auwerx, J., and

Mangelsdorf, D.J. (2000). Molecular basis for feedback regulation of bile acid

synthesis by nuclear receptors. Mol. Cell 6, 507–515.

Makishima, M., Okamoto, A.Y., Repa, J.J., Tu, H., Learned, R.M., Luk, A., Hull,

M.V., Lustig, K.D., Mangelsdorf, D.J., and Shan, B. (1999). Identification of a

nuclear receptor for bile acids. Science 284, 1362–1365.

Maloney, P.R., Parks, D.J., Haffner, C.D., Fivush, A.M., Chandra, G., Plunket,

K.D., Creech, K.L., Moore, L.B., Wilson, J.G., Lewis, M.C., et al. (2000).

Identification of a chemical tool for the orphan nuclear receptor FXR.

J. Med. Chem. 43, 2971–2974.

Mataki, C., Magnier, B.C., Houten, S.M., Annicotte, J.S., Argmann, C.,

Thomas, C., Overmars, H., Kulik, W., Metzger, D., Auwerx, J., and

Schoonjans, K. (2007). Compromised intestinal lipid absorption in mice with

a liver-specific deficiency of liver receptor homolog 1. Mol. Cell. Biol. 27,

8330–8339.

Menet, J.S., Rodriguez, J., Abruzzi, K.C., and Rosbash, M. (2012). Nascent-

Seq reveals novel features of mouse circadian transcriptional regulation.

eLife 1, e00011.

Motohashi, H., O’Connor, T., Katsuoka, F., Engel, J.D., and Yamamoto, M.

(2002). Integration and diversity of the regulatory network composed of Maf

and CNC families of transcription factors. Gene 294, 1–12.

Parks, D.J., Blanchard, S.G., Bledsoe, R.K., Chandra, G., Consler, T.G.,

Kliewer, S.A., Stimmel, J.B., Willson, T.M., Zavacki, A.M., Moore, D.D., and

Lehmann, J.M. (1999). Bile acids: natural ligands for an orphan nuclear recep-

tor. Science 284, 1365–1368.

Potthoff, M.J., Kliewer, S.A., and Mangelsdorf, D.J. (2012). Endocrine fibro-

blast growth factors 15/19 and 21: from feast to famine. Genes Dev. 26,

312–324.

Ridaura, V.K., Faith, J.J., Rey, F.E., Cheng, J., Duncan, A.E., Kau, A.L., Griffin,

N.W., Lombard, V., Henrissat, B., Bain, J.R., et al. (2013). Gut microbiota from

twins discordant for obesity modulate metabolism in mice. Science 341,

1241214.

Russell, D.W. (2003). The enzymes, regulation, and genetics of bile acid syn-

thesis. Annu. Rev. Biochem. 72, 137–174.

Ryan, K.K., Tremaroli, V., Clemmensen, C., Kovatcheva-Datchary, P.,

Myronovych, A., Karns, R., Wilson-Perez, H.E., Sandoval, D.A., Kohli, R.,

Backhed, F., and Seeley, R.J. (2014). FXR is a molecular target for the effects

of vertical sleeve gastrectomy. Nature 509, 183–188.

Shavit, J.A., Motohashi, H., Onodera, K., Akasaka, J., Yamamoto, M., and

Engel, J.D. (1998). Impaired megakaryopoiesis and behavioral defects in

mafG-null mutant mice. Genes Dev. 12, 2164–2174.

Sinal, C.J., Tohkin, M., Miyata, M.,Ward, J.M., Lambert, G., andGonzalez, F.J.

(2000). Targeted disruption of the nuclear receptor FXR/BAR impairs bile acid

and lipid homeostasis. Cell 102, 731–744.

Thomas, A.M., Hart, S.N., Kong, B., Fang, J., Zhong, X.B., and Guo, G.L.

(2010). Genome-wide tissue-specific farnesoid X receptor binding in mouse

liver and intestine. Hepatology 51, 1410–1419.

Wang, L., Lee, Y.K., Bundman, D., Han, Y., Thevananther, S., Kim, C.S., Chua,

S.S., Wei, P., Heyman, R.A., Karin, M., and Moore, D.D. (2002). Redundant

pathways for negative feedback regulation of bile acid production. Dev. Cell

2, 721–731.

Zhang, Y., Lee, F.Y., Barrera, G., Lee, H., Vales, C., Gonzalez, F.J., Willson,

T.M., and Edwards, P.A. (2006). Activation of the nuclear receptor FXR im-

proves hyperglycemia and hyperlipidemia in diabetic mice. Proc. Natl. Acad.

Sci. USA 103, 1006–1011.


Cell Metabolism, Volume 21

Supplemental Information

MAFG Is a Transcriptional Repressor of Bile Acid Synthesis and Metabolism

Thomas Q. de Aguiar Vallim, Elizabeth J. Tarling, Hannah Ahn, Lee R. Hagey, Casey E. Romanoski,

Richard G. Lee, Mark J. Graham, Hozumi Motohashi, Masayuki Yamamoto, and Peter A. Edwards

0

1

2

3

0

1

2

35

10

15

20Olig1

mR

NA

(Fol

d C

hang

e)

GSK2324 FXR

+ − −

WT

− + −

GW4064

***

*** B

+ − −

KO

− + −

Nr0b2/Shp

C Shp

mR

NA

(Fol

d C

hang

e)

FXR

Diet

***

WT

Ctr CA

D

KO

Ctr CA

** *

Supplementary Figure 1

Akr1d1

A

0

1

2

3

4

0

1

2

3

45678

mR

NA

(Fol

d C

hang

e)

Shp Bsep

Ctr 10 30 100 Ctr 10 30 100 GSK2324

*** **

*

***

***

***

A

0

2

4

6

FXR

+ − + −

Flox L-KO

Shp

***

0

1

2

3

468101214

mR

NA

(Fol

d C

hang

e)

GSK2324 Ctr 1h 2h 4h

Bsep Shp

Ctr 1h 2h 4h

*** **

* **

*

*** **

* **

*

***

C


mR

NA

(Fol

d C

hang

e)

CDCA (µM) 0 100 150 200

SHP

***

***

***

B

mR

NA

(Fol

d C

hang

e)

GSK2324

0

20

40

60

80

100

Ad-

Ctr

µmol

/BW

Ad-

Crip

2

Ad-

Maf

G

Ad-

Zfp3

85a

Hepatic BA

0

50

100

150

200

Ad-

Ctr

µmol

/BW

Ad-

Crip

2

Ad-

Maf

G

Ad-

Zfp3

85a

Intestinal BA

0

5000

10000

15000

Ad-

Ctr

µmol

/BW

Ad-

Crip

2

Ad-

Maf

G

Ad-

Zfp3

85a

Biliary BA H

D E IB: MafG

IB: βA

Control Diet Colesevelam Diet

Ad-Control Ad-Control Ad-MafG Ad-MafG

F G


A B N-FLAG

MafG

FLAG DAPI Merge

C-FLAG MafG

FLAG Control

*** ***

0

1

2

3

5

10

Adr

enal

Aor

ta

BAT

Bra

in

Hea

rt

Duo

denu

m

Jeju

num

Ileum

Col

on

Kid

ney

Live

r

Lung

Mus

cle

Sple

en

Skin

Thym

us W

AT

MafG mRNA

mR

NA

(Fol

d C

hang

e)

mR

NA

(Fol

d C

hang

e)

27

27 30

24

27

28

28

28

28

26

26

26

27

27

27

30

27


A IB: MafG

IB: β-Actin

Control ASO MafG ASO

0

5

10

15

20

25

0

50

100

150

200

250

0

5000

10000

Hepatic Bile Acid

Intestinal Bile Acid

Biliary Bile Acid

µmol

/BW

µmol

/BW

µmol

/BW

p=0.08

B C D

0.0

0.5

1.0

1.5

0

1

2

34567

Nqo1 MARE

G6pdx MARE #1

G6pdx MARE #3

G6pdx ARE

Chr 11 (-ve)

Ad-

BLR

P

Ad-

BLR

P-M

afG

Ad-

BLR

P

Ad-

BLR

P-M

afG

Ad-

BLR

P

Ad-

BLR

P-M

afG

Ad-

BLR

P

Ad-

BLR

P-M

afG

Ad-

BLR

P

Ad-

BLR

P-M

afG

% o

f Inp

ut


Scp2

Control ChIP-Seq

MafG ChIP-Seq

Akr1d1

Control ChIP-Seq

MafG ChIP-Seq

Akr1c14

Control ChIP-Seq

MafG ChIP-Seq

Cyp8b1

mR

NA

(Fol

d C

hang

e)

Ad-

BLR

P

Ad-

BLR

P-M

afG

**

*

C D

Acox2

Control ChIP-Seq

MafG ChIP-Seq

E F G

Hsd17b4

MafG ChIP-Seq

Control ChIP-Seq H I

B

0

5

10

MafG m

RN

A (F

old

Cha

nge)

Ad-

BLR

P

Ad-

BLR

P-M

afG

**

* A

** ***

Cyp7b1 Cyp27a1

Ad-

BLR

P

Ad-

BLR

P-M

afG

Ad-

BLR

P

Ad-

BLR

P-M

afG

Slc10a1/Ntcp

Control ChIP-Seq

MafG ChIP-Seq

Cyp7a1

100kb MafG ChIP-Seq

Control ChIP-Seq

* * * *

* * * * *

* *

J


Term ID! Term Name!Enrichment

p-value!Genes in Term!

Target Genes in

Term!

Total Target

Genes in Total!

Fraction of

Targets in Term!

GO:0009892!negative

regulation of metabolic process!

3.67E-11! 1715! 53! 243! 22%!

Top 20,000 !MafG peaks!

Top 10,633 !LRH-1 peaks!

282! 10,351!19,718!

A

B

Supplemental Figure Legends Supplemental Figure 1 (related to Figure 2)– Identification of

Transcriptional Repressors as Direct FXR Target Genes. (A) ChIP-Seq

analysis of hepatic FXR from (Chong et al., 2010) (top) and (Thomas et al., 2010)

(bottom) at Akr1d1 genomic locus. (B) Hepatic expression of Olig1 in C57BL/6

wild-type and Fxr−/− mice treated with vehicle, GW4064 or GSK2324 for 3 days

(n=7-9 mice/group). (C) ChIP-Seq analysis of hepatic FXR from (Chong et al.,

2010) (top) and (Thomas et al., 2010) (bottom) at Shp (gene symbol Nr0b2)

genomic locus. (D) Hepatic expression of Shp in C57BL/6 wild-type and Fxr−/−

mice fed either a control or 0.25% cholic acid (CA) diet for 7 days. All data shown

as mean ± SEM. Asterisks indicate statistically significant differences comparing

WT or KO vehicle treated against agonist treated mice (*p<0.05; **p<0.01; ***

p<0.001).

Supplemental Figure 2 (related to Figure 3) – Regulation of Crip2, MafG and

Zfp385a by FXR and Effect of MafG Overexpression on Total Bile Acid

Levels.

(A) Hepatic expression of Shp and Bsep in C57BL/6 wild-type mice treated daily

with vehicle, or 10, 30 or 100mpk of GSK2324 for 3 days (n=4-8 mice/group). (B)

Hepatic expression of Shp and Bsep in C57BL/6 wild-type treated with 30 mpk of

GSK2324 for 1, 2 or 4h before sacrifice (n=6 mice/group). (C) Hepatic expression

of Shp in littermate C57BL/6 wild-type (Flox) or liver-specific Fxr−/− mice (L-KO)

treated with GSK2324 (n=7-9 mice/group). (D) Western blotting analysis of MafG

and beta actin for Ad-control or Ad-MafG-treated livers (n=5 mice/group) fed

either control or colesevelam (Colesev) diet. Quantification of Western blots are

shown in Fig 3C. (E-G) Total bile acid levels measured following extraction from

liver (E), intestine (F) or directly from biliary fluid (G), normalized to body weight

(n=7/8 mice/group). (H) Expression levels of SHP in human HepG2 cells treated

with 100,150 or 200µM chenodeoxycholic acid (CDCA) for 4 hours (n=4

wells/condition). All data shown as mean ± SEM. Asterisks indicate statistically

significant differences comparing WT or KO vehicle treated against agonist

treated mice, or in H, comparing vehicle with CDCA treatments (*** p<0.001).

Supplemental Figure 3 (related to Results section: MAFG Overexpression

in Mouse Liver Represses Numerous Genes Involved in Bile Acid

Metabolism) – Nuclear Localization and Tissue Expression of MafG. (A)

Nuclear localization of MAFG protein following transfection of Hep3B cells with

pcDNA3.1 expressing MAFG containing a FLAG peptide at either the N- or C-

termini (identified with anti-FLAG antibody). (B) Expression of MafG mRNA in

various tissues isolated from 8-week old C57BL/6 male mice (n=3 mice per

tissue). Cp values, indicative of mRNA abundance, are listed for each tissue in

white. All data are shown as mean ± SEM.

Supplemental Figure 4 (related to Figure 5)– Loss of MafG Does not Affect

Total Bile Acid Levels. (A) Western blotting analysis of MAFG and β-ACTIN in

livers of mice treated with control ASO or MafG ASO (100mpk) for 3 days.

Quantification of the Western blots is shown in Fig. 5F. (B-D) Total bile acid

levels measured following extraction from liver (B), intestine (C) or directly from

biliary fluid (D), normalized to body weight (n=8-9 mice/group).

Supplemental Figure 5 (related to Figure 6) – Identification of MAFG-

Response Elements (MAREs) in Known and Newly Identified MAFG Target

Genes. (A) Hepatic levels of MafG mRNA or (B) Cyp8b1 mRNA levels in

C57BL/6 mice treated with either control (Ad-BLRP) or BLRP-tagged MafG

adenovirus (Ad-BLRP MafG) (n=6-7 mice/group). All data shown as mean ±

SEM. Asterisks indicate statistically significant differences comparing BLRP-

Control versus BLRP MafG (** p<0.01; *** p<0.001). (C) Chromatin

immunoprecipitation (ChIP) analysis of hepatic BLRP-MafG occupancy at the

Nqo1 and G6pdx loci determined by RT-qPCR using primers described in Hirotsu

et al. (Hirotsu et al., 2012). (D-I) ChIP-Seq analysis of MAREs in chromatin

isolated from livers of mice treated with Ad-BLRP (control; top) or Ad-BLRP MafG

(bottom) at loci for (D) Acox2 (E) Akr1d1, (F) Akr1c14, (G) Ntcp, (H) Hsd17b4, (I)

Scp2 and (J) Cyp7a1. Significant peaks are highlighted with *.

Supplemental Figure 6 (related to Results section: Identification of MAFG

Binding Sites at Multiple Genes Involved in Bile Acid Synthesis and

Metabolism) – Co-occupancy of Genes by MafG and LRH1. (A) Venn diagram

for ChIP-Seq analysis from liver LRH1 and MafG binding sites, showing 282

genes are co-bound by MafG and LRH1. (B) Gene ontology analysis of 282

genes identifies negative metabolic processes as a significantly enriched

category.

Supplemental Table 1 - RT-qPCR Primer Sequences

Gene Forward Reverse Ref

Mouse Crip2 TGCTGAGCATGATGGGAAG CGGTCTGAGGCTTCTCGTAG

Mouse MafG GACCCCCAATAAAGGAAACAA TCAACTCTCGCACCGACAT

Mouse Zfp385a GTCTGTCAGATCCGCTTCAAT GGCGATTACCCTTGTAGTGC

Mouse Bsep

(Abcb11)

AAGCTACATCTGCCTTAGACACAG

AA

CAATACAGGTCCGACCCTCTCT (Schmidt et

al., 2010)

Mouse Shp

(Nr0b2)

CGATCCTCTTCAACCCAGATG AGGGCTCCAAGACTTCACACA (Schmidt et

al., 2010)

Mouse Cyp7a1 AGCAACTAAACAACCTGCCAGTAC

TA

GTCCGGATATTCAAGGATGCA (Schmidt et

al., 2010)

Mouse Cyp8b1 GCCTTCAAGTATGATCGGTTCCT GATCTTCTTGCCCGACTTGTAGA (Schmidt et

al., 2010) Mouse Cyp7b1 TAGCCCTCTTTCCTCCACTCATA GAACCGATCGAACCTAAATTCCT (Schmidt et

al., 2010)

Mouse Cyp27a1 GCCTCACCTATGGGATCTTCA TCAAAGCCTGACGCAGATG (Schmidt et

al., 2010)

Mouse Cyp39a1 ACCTATGATGAGGGCTTTGAGTA CCATCTTTTGGATTTTGACCA

Mouse Acox2 AGATTGGGCCTATAGGGAAGA CACCGGGAGGTACCAAGAA

Mouse Akr1d1 GAAAAGATAGCAGAAGGGAAGGT GGGACATGCTCTGTATTCCATAA

Mouse Akr1c14 TTATTCATTTCCCAATGGCTTT TTTTCCATGTTCGTCTCGTG

Mouse Amacr GGCAGGTCATCGATTCAAG GCTGGGTTTTCCACAGGA

Mouse Hsd17b4 GGGAGCAGTACTTGGAGCTG TCAGCAATAACTGCTTCACATTTT

Mouse Hsd3b7 CGCTTTGGAGGTCGTCTATT CAGTATGTGCATCCAAGCAAC

Mouse Ntcp

(Slc10a1)

GAAGTCCAAAAGGCCACACTATGT ACAGCCACAGAGAGGGAGAAAG (Schmidt et

al., 2010)

Mouse Scp2 GATTGCTTCTCTGTCAATGAACTC ACCAGGGTTCCACCTTGTC

Mouse Slc25a7 AGGGTTTTTGCATTCCTGTG TTGGTTCTTTCGAACCTTGG

Mouse 36B4

(Rplp0)

CACTGGTCTAGGACCCGAGAAG GGTGCCTCTGGAGATTTTCG

Mouse Tbp CTCAGTTACAGGTGGCAGCA ACCAACAATCACCAACAGCA

Human MAFG GTGGACAGGAAGCAGCTCA TATTGGGGGTCGTCATAACC

Human

CYP8B1

CTGGGCAACATGCTTCAGT ACTTGTCCTGCATAGCTGAGG

Human 36B4

(RPLP0)

GGTGCCTCTGGAGATTTTAG CACTGGTCTCGGGCCCGAGAA

Human TBP CGGCTGTTTAACTTCGCTTC CACACGCCAAGAAACAGTGA

Supplemental Table 2 – BLRP-MafG ChIP qPCR Primer Sequences

Primer Pair Forward Reverse Ref

Cyp8b1 -2652bp AACTGCAAGCAGCATTTCTTC GAAGCCAGGGCAGGAATC

Cyp8b1 -2299bp GAGTTCTCCCCAGCAAGGA GCCCCAGAGATGGGATAGTC

Cyp8b1 -1874bp CTAATGGCAACTAAGGAGACACC TCTCCATCAGACATCTGCTAGTTT

Cyp8b1 -1211bp GCCAGACACTGTCCTAAGAGC GAGTACACCCATGTCAGTGATTT

T

Cyp8b1 -890bp CAGGGAGAGATCCTGACTCAA GGGTTCTCTGGAAAAGCAGA

Cyp8b1 -382bp TCATTGGCTAGCTCATTCACTAA ACAGACACGCACTTTAAAAACAA

T

Cyp8b1 -83bp TCCCTGTGCCAGCTAACTAGA GTTCCTGCCCTTGGACTTT

Cyp8b1 +39bp GAGCTGACAAGTGGAGCTCAG GCAGGGCTCCTAGCACTGTA

Cyp8b1 +485bp AACTCAACCAGGCCATGC GCACCCAGACTCGAACCTT

G6pdx ARE GTGGCAGGGGACTGGTCTGC TGTCAGTTGCAGGCTGAGCCAA (Hirotsu et

al., 2012)

G6pdx MARE #1 GGCCTTGCAGGAGTGAGGCA AGCCGACCCTCAGTCGCAGT (Hirotsu et

al., 2012)

G6pdx MARE #3 GGGAGCCTGAGCCCAATGGC AAGCCCAGCTGGCAGCAAGT (Hirotsu et

al., 2012)

Nqo1 GCACGAATTCATTTCACACGAGG GGAAGTCACCTTTGCACGCTAG (Hirotsu et

al., 2012)

Negative CTAGGCCACAGAATTGAAAGATC

T

GTAGGTGGAAATTCTAGCATCAT

CC

Extended Experimental Procedures Animals. All animals were bred and housed in a pathogen-free animal facility

and unless otherwise stated, were maintained on a C57BL/6 background

(Jackson Laboratories). The generation of liver- and intestine-specific Fxr−/−mice

and their respective (littermate) wild-type (flox) controls, and whole body Fxr−/−

mice was previously described (de Aguiar Vallim et al., 2013). Littermate wild-

type and MafG+/− mice on a mixed background, were generated after re-

derivation at UCLA from MafG+/− mating pairs.

Primary Hepatocyte Isolation: Primary hepatocytes were isolated as described

previously from wild-type C57BL/6 mice (de Aguiar Vallim et al., 2013).

Bile Acid Analysis. Gall bladders were removed from all experimental mice after

a 4-6 hour fast and bile removed and stored at -80ºC. Bile acid species were

measured by the HPLC System. Conjugated bile acids were analyzed by high-

pressure liquid chromatography using a Kinetex 5m C-18 100A 250 x 4.6 mm

column (Phenomenex, Torrance, CA) with isocratic elution at 0.75 ml/min. The

eluting solution was composed of a mixture of methanol and 0.01M KH2PO4

(67.4% v/v), adjusted to an apparent pH of 5.25 with H2PO4. Bile acids were

quantified by measuring their absorbance at 205 nm and were identified by

matching their relative retention times with those of known standards.

Total bile acids were extracted from liver and intestine by homogenizing a piece

(approx. 250mg) of liver or the entire small intestine with contents in 96%

ethanol. Samples were incubated at 55°C overnight, and centrifuged at 2000xg

the following day, where bile acids in the supernatant were collected. Pellets

were re-suspended in the same volumes described above of 96% ethanol and

incubated at 55°C for a second overnight period. Samples were centrifuged as

above, and supernatants combined, with the pellet being resuspended in half

volume of chloroform:methanol (2:1 ratio) at incubated at room temperature for a

final overnight period. Total bile acids were quantified with a calorimetric assay

(Diazyme) calculated using a standard curve from different concentrations of Na-

tauro-cholic acid. For biliary bile acids, samples were diluted 1:1000 and

assayed, for intestine, extracted samples were diluted 1:10 and for liver,

undiluted extracted samples were used for the assay. Bile acid levels were then

corrected for total body weight.

RNA Isolation, RT-qPCR Quantification. Liver samples (approximately 100mg)

were removed from mice and immediately flash-frozen in liquid nitrogen and

stored at -80ºC. Frozen tissue was then homogenized in Qiazol (Qiagen) and

extracted according to the manufacturer’s instructions. RNA was then DNAse-

treated (rDNAseI, Ambion/Life Technologies) for 1 hour at 37ºC RNA was re-

extracted by phenol:chloroform:isoamyl alcohol (25:24:1, pH 6.6, Life

Technologies). Complementary DNA (cDNA) was synthesized from 500ng of

RNA using the High Capacity cDNA Reverse Transcriptase kit (Applied

Biosciences/Life Technologies) according to manufacturers’ instructions. RT-

qPCR analysis and microarray analysis are described in Supplemental Materials.

ChIP, ChIP-Seq and RNA-Seq Analysis. The biotin-ligase recognition peptide

(BLRP) (Heinz et al., 2010) was cloned at the N-terminus of MafG, and used to

generate at BLRP-tagged MafG adenovirus. C57BL/6 mice were infused with

adenovirus particles as described above. Liver chromatin was isolated as

described in detail in Supplemental Materials. FXR ChIP-Seq data have been

previously described (Chong et al., 2010; Thomas et al., 2010). BED files were

converted to BigWig files and visualized using the IGV browser or UCSC genome

browser. Publically available RNA-Seq data (Menet et al., 2012) BedGraph files

were downloaded from GEO datasets and visualized using IGV browser. For

MAFG ChIP-Seq analysis, libraries from IP chromatin isolated as described

above were prepared using the Kapa Library Prep kit (Kapa Bioscience),

sequences using the Illumina HiSeq sequencer (Illumina) by the UNGC at UCLA.

Data analysis was carried out using Galaxy suite and visualized as described

above. Peak and motif analysis was carried out using HOMER as previously

described (Heinz et al., 2010). LRH1 ChIP-Seq peaks (identified using MACS2)

(Chong et al., 2012) were converted to coordinates in mm10 and compared with

MafG peaks identified as described above.

Western Blotting. Antibodies used for Western Blotting were MAFG (Genetex;

GTX114541; 1:1000), β-ACTIN (Sigma, 1:5000), BLRP (Avi-tag; Genescript

1:1000), GAPDH (Genetex, 1:2000), LAMIN A/C (Santa Cruz, 1:1000) and for

immuno-staining M2 FLAG (Sigma, 1:1000). For ASO-treated, or Ad-MafG

treated livers (Colesevelam), Western blots were quantified using HRP detection

with ECL prime reagent (GE Healthcare) according to manufacturer’s

instructions. HRP signal detection was determined electronically using GE Image

Quant LAS 4000 system and parameters set strictly below the saturation point.

Densitometric analysis was carried out using Quantity One software (BioRad)

and relative protein levels expressed as fold change from control after

normalization to β-actin protein levels, with vehicle, Ad-control or control ASO-

treated mice set to 1.

Promoter Reporter Analysis and Cell Culture. Mouse MafG promoter (2kb) or

Cyp8b1 promoter luciferase reporter constructs (3kb, 1kb and 0.5kb) were

generated by amplifying these regions from mouse genomic DNA using KAPA

HiFi polymerase (Kapa) and cloning them into pGL4.10[luc2] plasmid (Promega).

Mutations were made with a nested PCR approach, and primers were designed

with the Quick Change Site-Directed Mutagenesis website (Agilent). Luciferase

reporter constructs were transfected using Fugene HD (Promega) according to

manufacturer’s instructions into human HepG2 cells (ATCC) or Hep3B cells

(ATCC), plated onto 48-well dishes (n=6 wells/condition). After 24 hours, medium

was replaced and cells were treated with either vehicle (DMSO or water) or

GSK2324 (1µM in water) in medium containing 10% charcoal-stripped serum

(Omega Scientific) for a further 24 hours. For siRNA transfections, HepG2 cells

were seeded in 6 well plates transfected with 1µM siRNA (MafG siRNA; Cat

#SI00036701, CCGGGTATTTATTGCTGTACA, or Allstars negative control

(SI03650318) Qiagen) for 24 hours and then medium was replaced with 10%

charcoal-stripped serum for further 24 hours before cells were harvested in

Qiazol for RNA analysis or RIPA buffer for protein analysis.

Anti-sense oligonucleotides (ASO). A chimeric 16-mer phosphorothioate

oligonucleotide targeted to mouse MafG (5’- GGCCAATACGCCGTCA -3’) or

control (5’- TAATGTCTGATAACTC -3’) containing 2′-4’ constrained ethyl groups

at bases underlined was synthesized and purified as described in (Seth et al.,

2010).

ChIP Analysis: One lobe of the liver was frozen in liquid nitrogen, and the

remainder (~80%) of the liver was then macerated in PBS containing 1%

formaldehyde and protease inhibitor cocktail 1 (PIC1, comprised of 1µg/ml

leupeptin, 1.4µg/ml pepstatin, 0.2mg/ml PMSF (Sigma), 1mM EGTA, 1mM

EDTA) for 10 minutes with gentle rotation at room temperature. Crosslinking was

then stopped by the addition of glycine (0.125M) and samples were further

incubated for 5 minutes at room temperature. Fixed livers were then placed on

ice and then centrifuged (2,000 x g for 10 minutes) and the pellet resuspended in

8ml of PBS containing protease inhibitor cocktail. Resuspended pellets were

gently homogenized and again centrifuged as above. The resulting pellet was

resuspended in 5ml lysis buffer (5mM PIPES, pH 8.0, 85mM KCl, 0.5% NP-40

alternate) supplemented with protease inhibitor complex 2 (PIC2, 10µg/ml

leupeptin, 5µg/ml pepstatin, 0.2mg/ml PMSF, 50µg/ml ALLN, Sigma). Nuclei

were released after 30 strokes using a Dounce homogenizer and collected after

centrifugation as above. Pellets were resuspended in 6ml homogenization buffer

(10mM HEPES, pH 7.6, 25mM KCl, 1mM EDTA, 1mM EGTA, 1M Sucrose, 10%

glycerol, 0.15mM spermine, supplemented with PIC1) and layered onto 3ml of

the same buffer. Nuclei were then pelleted at 26,000rpm for 1 hour (Beckman

SW41 rotor) and stored at -80ºC. Nuclear pellets were re-suspended in 0.3ml

nuclear lysis buffer (50mM Tris pH 7.6, 10mM EDTA and 1% SDS), and diluted

with 0.6ml immunoprecipitation (IP) dilution buffer (0.01% SDS, 1.1% Triton

x100, 167mM NaCl, 16.7mM Tris pH 7.6, 1.2mM EDTA). For sonication, 0.3ml

(1/3) of nuclear lysate was sonicated for 25-30 cycles 30 seconds on 30 seconds

off at 4ºC with BioRuptor twin sonicator (Diagenode). Sonicated chromatin was

then further diluted to 1ml with IP dilution buffer, which is sufficient for three ChIP

reactions. The BLRP antibody (Genescript) was conjugated to Protein G

Dynabeads (Life Technologies) in PBS containing 0.5% BSA overnight at 4ºC

with gentle rotation. Sonicated DNA (0.33ml made up to 1ml in IP dilution buffer)

was then incubated overnight with Protein G-BLRP beads at 4ºC with gentle

rotation. For input samples, 10% was used. Samples were then washed twice

with wash buffer I (20mM Tris, pH 7.4, 150mM NaCl, 0.1% SDS, 1% Triton X100,

2mM EDTA), three times with alternate wash buffer III (10mM Tris, pH 7.4,

250mM LiCl, 1% NP-40 alternate, 0.7% Deoxycholate, 1mM EDTA), two washes

with 0.2% Triton X100 TE buffer and two final washes with 50mM NaCl

containing TE buffer. Chromatin was recovered and DNA from IP and input was

isolated after reverse crosslinking (incubating at 65ºC overnight in 0.3M NaCl and

Proteinase K digestion) and DNA purified using gel/PCR extraction kit (Clontech).

For qPCR analysis, samples were diluted 1:10 and run in triplicate as described

above. A standard curve for PCR was generated from 4 log serial dilutions of

input samples and data expressed as percentage of input. Primer sequences for

ChIP analysis are provided in Table S2.

RT-qPCR Analysis: RT-qPCR standards were prepared from an aliquot from

each cDNA reaction that was then pooled. Standards which were diluted over 3

log range (dilution range 1:5, 1:10, 1:50, 1:100, 1:500, 1:1000). Quantitative PCR

was carried out with primers designed to cross exon-exon boundaries using the

Roche UPL primer design website. RT-qPCR primer sequences are provided in

Table S1. Quantitative PCR was carried out in triplicate for each sample in 384-

well format using Kapa LC480 SYBR green mix (Kapa Biosciences). Quantitative

PCR was carried out using a Lightcycler 480 (Roche) and concentrations were

determined from the standard curve using the efficiency corrected method (2nd

derivative max, Roche). Relative quantification was determined by normalizing

the expression of each gene to a housekeeper. Two housekeepers were used to

normalize gene expression data (36B4 and Tbp).

Microarray Analysis: Hepatic RNA was isolated from 3 mice treated with Ad-

Control or Ad-MafG and global gene expression measured by the UCLA

Neuroscience Genomics Core (UNGC) using Illumina microarrays (Illumina).

Microarray analysis was carried out using Genome Suite (Illumina) and

differential gene expression determined from probes that that had expression of

p>0.05, (approximately 10,000 genes left over after filtering). Of those, genes

that were altered 1.5 fold were considered differentially expressed. Gene

enrichment analysis was carried out using DAVID and pathway analysis using

KEGG (Huang da et al., 2009).

Supplemental References Chong, H.K., Biesinger, J., Seo, Y.K., Xie, X., and Osborne, T.F. (2012).

Genome-wide analysis of hepatic LRH-1 reveals a promoter binding preference

and suggests a role in regulating genes of lipid metabolism in concert with FXR.

BMC Genomics 13, 51.


X., and Osborne, T.F. (2010). Genome-wide interrogation of hepatic FXR reveals

an asymmetric IR-1 motif and synergy with LRH-1. Nucleic Acids Res 38, 6007-

6017.


and Edwards, P.A. (2013). MicroRNA-144 regulates hepatic ATP binding

cassette transporter A1 and plasma high-density lipoprotein after activation of the

nuclear receptor farnesoid X receptor. Circ Res 112, 1602-1612.

Heinz, S., Benner, C., Spann, N., Bertolino, E., Lin, Y.C., Laslo, P., Cheng, J.X.,

Murre, C., Singh, H., and Glass, C.K. (2010). Simple combinations of lineage-

determining transcription factors prime cis-regulatory elements required for

macrophage and B cell identities. Mol Cell 38, 576-589.

Hirotsu, Y., Katsuoka, F., Funayama, R., Nagashima, T., Nishida, Y., Nakayama,

K., Engel, J.D., and Yamamoto, M. (2012). Nrf2-MafG heterodimers contribute

globally to antioxidant and metabolic networks. Nucleic Acids Res 40, 10228-

10239.

Huang da, W., Sherman, B.T., and Lempicki, R.A. (2009). Systematic and

integrative analysis of large gene lists using DAVID bioinformatics resources. Nat

Protoc 4, 44-57.

Menet, J.S., Rodriguez, J., Abruzzi, K.C., and Rosbash, M. (2012). Nascent-Seq

reveals novel features of mouse circadian transcriptional regulation. Elife 1,

e00011.

Schmidt, D.R., Holmstrom, S.R., Fon Tacer, K., Bookout, A.L., Kliewer, S.A., and

Mangelsdorf, D.J. (2010). Regulation of bile acid synthesis by fat-soluble

vitamins A and D. J Biol Chem 285, 14486-14494.

Seth, P.P., Vasquez, G., Allerson, C.A., Berdeja, A., Gaus, H., Kinberger, G.A.,

Prakash, T.P., Migawa, M.T., Bhat, B., and Swayze, E.E. (2010). Synthesis and

biophysical evaluation of 2',4'-constrained 2'O-methoxyethyl and 2',4'-constrained

2'O-ethyl nucleic acid analogues. J Org Chem 75, 1569-1581.

Thomas, A.M., Hart, S.N., Kong, B., Fang, J., Zhong, X.B., and Guo, G.L. (2010).

Genome-wide tissue-specific farnesoid X receptor binding in mouse liver and

intestine. Hepatology 51, 1410-1419.

Supplemental Figure Legends Supplemental Figure 1 (related to Figure 2)– Identification of

Transcriptional Repressors as Direct FXR Target Genes. (A) ChIP-Seq

analysis of hepatic FXR from (Chong et al., 2010) (top) and (Thomas et al., 2010)

(bottom) at Akr1d1 genomic locus. (B) Hepatic expression of Olig1 in C57BL/6

wild-type and Fxr−/− mice treated with vehicle, GW4064 or GSK2324 for 3 days

(n=7-9 mice/group). (C) ChIP-Seq analysis of hepatic FXR from (Chong et al.,

2010) (top) and (Thomas et al., 2010) (bottom) at Shp (gene symbol Nr0b2)

genomic locus. (D) Hepatic expression of Shp in C57BL/6 wild-type and Fxr−/−

mice fed either a control or 0.25% cholic acid (CA) diet for 7 days. All data shown

as mean ± SEM. Asterisks indicate statistically significant differences comparing

WT or KO vehicle treated against agonist treated mice (*p<0.05; **p<0.01; ***

p<0.001).

Supplemental Figure 2 (related to Figure 3) – Regulation of Crip2, MafG and

Zfp385a by FXR and Effect of MafG Overexpression on Total Bile Acid

Levels.

(A) Hepatic expression of Shp and Bsep in C57BL/6 wild-type mice treated daily

with vehicle, or 10, 30 or 100mpk of GSK2324 for 3 days (n=4-8 mice/group). (B)

Hepatic expression of Shp and Bsep in C57BL/6 wild-type treated with 30 mpk of

GSK2324 for 1, 2 or 4h before sacrifice (n=6 mice/group). (C) Hepatic expression

of Shp in littermate C57BL/6 wild-type (Flox) or liver-specific Fxr−/− mice (L-KO)

treated with GSK2324 (n=7-9 mice/group). (D) Western blotting analysis of MafG

and beta actin for Ad-control or Ad-MafG-treated livers (n=5 mice/group) fed

either control or colesevelam (Colesev) diet. Quantification of Western blots are

shown in Fig 3C. (E-G) Total bile acid levels measured following extraction from

liver (E), intestine (F) or directly from biliary fluid (G), normalized to body weight

(n=7/8 mice/group). (H) Expression levels of SHP in human HepG2 cells treated

with 100,150 or 200µM chenodeoxycholic acid (CDCA) for 4 hours (n=4

wells/condition). All data shown as mean ± SEM. Asterisks indicate statistically

significant differences comparing WT or KO vehicle treated against agonist

treated mice, or in H, comparing vehicle with CDCA treatments (*** p<0.001).

Supplemental Figure 3 (related to Results section: MAFG Overexpression

in Mouse Liver Represses Numerous Genes Involved in Bile Acid

Metabolism) – Nuclear Localization and Tissue Expression of MafG. (A)

Nuclear localization of MAFG protein following transfection of Hep3B cells with

pcDNA3.1 expressing MAFG containing a FLAG peptide at either the N- or C-

termini (identified with anti-FLAG antibody). (B) Expression of MafG mRNA in

various tissues isolated from 8-week old C57BL/6 male mice (n=3 mice per

tissue). Cp values, indicative of mRNA abundance, are listed for each tissue in

white. All data are shown as mean ± SEM.

Supplemental Figure 4 (related to Figure 5)– Loss of MafG Does not Affect

Total Bile Acid Levels. (A) Western blotting analysis of MAFG and β-ACTIN in

livers of mice treated with control ASO or MafG ASO (100mpk) for 3 days.

Quantification of the Western blots is shown in Fig. 5F. (B-D) Total bile acid

levels measured following extraction from liver (B), intestine (C) or directly from

biliary fluid (D), normalized to body weight (n=8-9 mice/group).

Supplemental Figure 5 (related to Figure 6) – Identification of MAFG-

Response Elements (MAREs) in Known and Newly Identified MAFG Target

Genes. (A) Hepatic levels of MafG mRNA or (B) Cyp8b1 mRNA levels in

C57BL/6 mice treated with either control (Ad-BLRP) or BLRP-tagged MafG

adenovirus (Ad-BLRP MafG) (n=6-7 mice/group). All data shown as mean ±

SEM. Asterisks indicate statistically significant differences comparing BLRP-

Control versus BLRP MafG (** p<0.01; *** p<0.001). (C) Chromatin

immunoprecipitation (ChIP) analysis of hepatic BLRP-MafG occupancy at the

Nqo1 and G6pdx loci determined by RT-qPCR using primers described in Hirotsu

et al. (Hirotsu et al., 2012). (D-I) ChIP-Seq analysis of MAREs in chromatin

isolated from livers of mice treated with Ad-BLRP (control; top) or Ad-BLRP MafG

(bottom) at loci for (D) Acox2 (E) Akr1d1, (F) Akr1c14, (G) Ntcp, (H) Hsd17b4, (I)

Scp2 and (J) Cyp7a1. Significant peaks are highlighted with *.

Supplemental Figure 6 (related to Results section: Identification of MAFG

Binding Sites at Multiple Genes Involved in Bile Acid Synthesis and

Metabolism) – Co-occupancy of Genes by MafG and LRH1. (A) Venn diagram

for ChIP-Seq analysis from liver LRH1 and MafG binding sites, showing 282

genes are co-bound by MafG and LRH1. (B) Gene ontology analysis of 282

genes identifies negative metabolic processes as a significantly enriched

category.

Supplemental Table 1 - RT-qPCR Primer Sequences

Gene Forward Reverse Ref

Mouse Crip2 TGCTGAGCATGATGGGAAG CGGTCTGAGGCTTCTCGTAG

Mouse MafG GACCCCCAATAAAGGAAACAA TCAACTCTCGCACCGACAT

Mouse Zfp385a GTCTGTCAGATCCGCTTCAAT GGCGATTACCCTTGTAGTGC

Mouse Bsep

(Abcb11)

AAGCTACATCTGCCTTAGACACAG

AA

CAATACAGGTCCGACCCTCTCT (Schmidt et

al., 2010)

Mouse Shp

(Nr0b2)

CGATCCTCTTCAACCCAGATG AGGGCTCCAAGACTTCACACA (Schmidt et

al., 2010)

Mouse Cyp7a1 AGCAACTAAACAACCTGCCAGTAC

TA

GTCCGGATATTCAAGGATGCA (Schmidt et

al., 2010)

Mouse Cyp8b1 GCCTTCAAGTATGATCGGTTCCT GATCTTCTTGCCCGACTTGTAGA (Schmidt et

al., 2010) Mouse Cyp7b1 TAGCCCTCTTTCCTCCACTCATA GAACCGATCGAACCTAAATTCCT (Schmidt et

al., 2010)

Mouse Cyp27a1 GCCTCACCTATGGGATCTTCA TCAAAGCCTGACGCAGATG (Schmidt et

al., 2010)

Mouse Cyp39a1 ACCTATGATGAGGGCTTTGAGTA CCATCTTTTGGATTTTGACCA

Mouse Acox2 AGATTGGGCCTATAGGGAAGA CACCGGGAGGTACCAAGAA

Mouse Akr1d1 GAAAAGATAGCAGAAGGGAAGGT GGGACATGCTCTGTATTCCATAA

Mouse Akr1c14 TTATTCATTTCCCAATGGCTTT TTTTCCATGTTCGTCTCGTG

Mouse Amacr GGCAGGTCATCGATTCAAG GCTGGGTTTTCCACAGGA

Mouse Hsd17b4 GGGAGCAGTACTTGGAGCTG TCAGCAATAACTGCTTCACATTTT

Mouse Hsd3b7 CGCTTTGGAGGTCGTCTATT CAGTATGTGCATCCAAGCAAC

Mouse Ntcp

(Slc10a1)

GAAGTCCAAAAGGCCACACTATGT ACAGCCACAGAGAGGGAGAAAG (Schmidt et

al., 2010)

Mouse Scp2 GATTGCTTCTCTGTCAATGAACTC ACCAGGGTTCCACCTTGTC

Mouse Slc25a7 AGGGTTTTTGCATTCCTGTG TTGGTTCTTTCGAACCTTGG

Mouse 36B4

(Rplp0)

CACTGGTCTAGGACCCGAGAAG GGTGCCTCTGGAGATTTTCG

Mouse Tbp CTCAGTTACAGGTGGCAGCA ACCAACAATCACCAACAGCA

Human MAFG GTGGACAGGAAGCAGCTCA TATTGGGGGTCGTCATAACC

Human

CYP8B1

CTGGGCAACATGCTTCAGT ACTTGTCCTGCATAGCTGAGG

Human 36B4

(RPLP0)

GGTGCCTCTGGAGATTTTAG CACTGGTCTCGGGCCCGAGAA

Human TBP CGGCTGTTTAACTTCGCTTC CACACGCCAAGAAACAGTGA

Supplemental Table 2 – BLRP-MafG ChIP qPCR Primer Sequences

Primer Pair Forward Reverse Ref

Cyp8b1 -2652bp AACTGCAAGCAGCATTTCTTC GAAGCCAGGGCAGGAATC

Cyp8b1 -2299bp GAGTTCTCCCCAGCAAGGA GCCCCAGAGATGGGATAGTC

Cyp8b1 -1874bp CTAATGGCAACTAAGGAGACACC TCTCCATCAGACATCTGCTAGTTT

Cyp8b1 -1211bp GCCAGACACTGTCCTAAGAGC GAGTACACCCATGTCAGTGATTT

T

Cyp8b1 -890bp CAGGGAGAGATCCTGACTCAA GGGTTCTCTGGAAAAGCAGA

Cyp8b1 -382bp TCATTGGCTAGCTCATTCACTAA ACAGACACGCACTTTAAAAACAA

T

Cyp8b1 -83bp TCCCTGTGCCAGCTAACTAGA GTTCCTGCCCTTGGACTTT

Cyp8b1 +39bp GAGCTGACAAGTGGAGCTCAG GCAGGGCTCCTAGCACTGTA

Cyp8b1 +485bp AACTCAACCAGGCCATGC GCACCCAGACTCGAACCTT

G6pdx ARE GTGGCAGGGGACTGGTCTGC TGTCAGTTGCAGGCTGAGCCAA (Hirotsu et

al., 2012)

G6pdx MARE #1 GGCCTTGCAGGAGTGAGGCA AGCCGACCCTCAGTCGCAGT (Hirotsu et

al., 2012)

G6pdx MARE #3 GGGAGCCTGAGCCCAATGGC AAGCCCAGCTGGCAGCAAGT (Hirotsu et

al., 2012)

Nqo1 GCACGAATTCATTTCACACGAGG GGAAGTCACCTTTGCACGCTAG (Hirotsu et

al., 2012)

Negative CTAGGCCACAGAATTGAAAGATC

T

GTAGGTGGAAATTCTAGCATCAT

CC

Extended Experimental Procedures Animals. All animals were bred and housed in a pathogen-free animal facility

and unless otherwise stated, were maintained on a C57BL/6 background

(Jackson Laboratories). The generation of liver- and intestine-specific Fxr−/−mice

and their respective (littermate) wild-type (flox) controls, and whole body Fxr−/−

mice was previously described (de Aguiar Vallim et al., 2013). Littermate wild-

type and MafG+/− mice on a mixed background, were generated after re-

derivation at UCLA from MafG+/− mating pairs.

Primary Hepatocyte Isolation: Primary hepatocytes were isolated as described

previously from wild-type C57BL/6 mice (de Aguiar Vallim et al., 2013).

Bile Acid Analysis. Gall bladders were removed from all experimental mice after

a 4-6 hour fast and bile removed and stored at -80ºC. Bile acid species were

measured by the HPLC System. Conjugated bile acids were analyzed by high-

pressure liquid chromatography using a Kinetex 5m C-18 100A 250 x 4.6 mm

column (Phenomenex, Torrance, CA) with isocratic elution at 0.75 ml/min. The

eluting solution was composed of a mixture of methanol and 0.01M KH2PO4

(67.4% v/v), adjusted to an apparent pH of 5.25 with H2PO4. Bile acids were

quantified by measuring their absorbance at 205 nm and were identified by

matching their relative retention times with those of known standards.

Total bile acids were extracted from liver and intestine by homogenizing a piece

(approx. 250mg) of liver or the entire small intestine with contents in 96%

ethanol. Samples were incubated at 55°C overnight, and centrifuged at 2000xg

the following day, where bile acids in the supernatant were collected. Pellets

were re-suspended in the same volumes described above of 96% ethanol and

incubated at 55°C for a second overnight period. Samples were centrifuged as

above, and supernatants combined, with the pellet being resuspended in half

volume of chloroform:methanol (2:1 ratio) at incubated at room temperature for a

final overnight period. Total bile acids were quantified with a calorimetric assay

(Diazyme) calculated using a standard curve from different concentrations of Na-

tauro-cholic acid. For biliary bile acids, samples were diluted 1:1000 and

assayed, for intestine, extracted samples were diluted 1:10 and for liver,

undiluted extracted samples were used for the assay. Bile acid levels were then

corrected for total body weight.

RNA Isolation, RT-qPCR Quantification. Liver samples (approximately 100mg)

were removed from mice and immediately flash-frozen in liquid nitrogen and

stored at -80ºC. Frozen tissue was then homogenized in Qiazol (Qiagen) and

extracted according to the manufacturer’s instructions. RNA was then DNAse-

treated (rDNAseI, Ambion/Life Technologies) for 1 hour at 37ºC RNA was re-

extracted by phenol:chloroform:isoamyl alcohol (25:24:1, pH 6.6, Life

Technologies). Complementary DNA (cDNA) was synthesized from 500ng of

RNA using the High Capacity cDNA Reverse Transcriptase kit (Applied

Biosciences/Life Technologies) according to manufacturers’ instructions. RT-

qPCR analysis and microarray analysis are described in Supplemental Materials.

ChIP, ChIP-Seq and RNA-Seq Analysis. The biotin-ligase recognition peptide

(BLRP) (Heinz et al., 2010) was cloned at the N-terminus of MafG, and used to

generate at BLRP-tagged MafG adenovirus. C57BL/6 mice were infused with

adenovirus particles as described above. Liver chromatin was isolated as

described in detail in Supplemental Materials. FXR ChIP-Seq data have been

previously described (Chong et al., 2010; Thomas et al., 2010). BED files were

converted to BigWig files and visualized using the IGV browser or UCSC genome

browser. Publically available RNA-Seq data (Menet et al., 2012) BedGraph files

were downloaded from GEO datasets and visualized using IGV browser. For

MAFG ChIP-Seq analysis, libraries from IP chromatin isolated as described

above were prepared using the Kapa Library Prep kit (Kapa Bioscience),

sequences using the Illumina HiSeq sequencer (Illumina) by the UNGC at UCLA.

Data analysis was carried out using Galaxy suite and visualized as described

above. Peak and motif analysis was carried out using HOMER as previously

described (Heinz et al., 2010). LRH1 ChIP-Seq peaks (identified using MACS2)

(Chong et al., 2012) were converted to coordinates in mm10 and compared with

MafG peaks identified as described above.

Western Blotting. Antibodies used for Western Blotting were MAFG (Genetex;

GTX114541; 1:1000), β-ACTIN (Sigma, 1:5000), BLRP (Avi-tag; Genescript

1:1000), GAPDH (Genetex, 1:2000), LAMIN A/C (Santa Cruz, 1:1000) and for

immuno-staining M2 FLAG (Sigma, 1:1000). For ASO-treated, or Ad-MafG

treated livers (Colesevelam), Western blots were quantified using HRP detection

with ECL prime reagent (GE Healthcare) according to manufacturer’s

instructions. HRP signal detection was determined electronically using GE Image

Quant LAS 4000 system and parameters set strictly below the saturation point.

Densitometric analysis was carried out using Quantity One software (BioRad)

and relative protein levels expressed as fold change from control after

normalization to β-actin protein levels, with vehicle, Ad-control or control ASO-

treated mice set to 1.

Cell Culture. For GSK2324 treatments, cells were transfected for 24 hours and

then treated with either vehicle (DMSO or water) or GSK2324 (1µM in water) in

medium containing 10% charcoal-stripped serum (Omega Scientific) for a further

24 hours. For siRNA transfections, HepG2 cells were seeded in 6 well plates

transfected with 1µM siRNA (MafG siRNA; Cat #SI00036701,

CCGGGTATTTATTGCTGTACA, or Allstars negative control (SI03650318)

Qiagen) for 24 hours and then medium was replaced with 10% charcoal-stripped

serum for further 24 hours before cells were harvested in Qiazol for RNA analysis

or RIPA buffer for protein analysis.

Anti-sense oligonucleotides (ASO). A chimeric 16-mer phosphorothioate

oligonucleotide targeted to mouse MafG (5’- GGCCAATACGCCGTCA -3’) or

control (5’- TAATGTCTGATAACTC -3’) containing 2′-4’ constrained ethyl groups

at bases underlined was synthesized and purified as described in (Seth et al.,

2010).

ChIP Analysis: One lobe of the liver was frozen in liquid nitrogen, and the

remainder (~80%) of the liver was then macerated in PBS containing 1%

formaldehyde and protease inhibitor cocktail 1 (PIC1, comprised of 1µg/ml

leupeptin, 1.4µg/ml pepstatin, 0.2mg/ml PMSF (Sigma), 1mM EGTA, 1mM

EDTA) for 10 minutes with gentle rotation at room temperature. Crosslinking was

then stopped by the addition of glycine (0.125M) and samples were further

incubated for 5 minutes at room temperature. Fixed livers were then placed on

ice and then centrifuged (2,000 x g for 10 minutes) and the pellet resuspended in

8ml of PBS containing protease inhibitor cocktail. Resuspended pellets were

gently homogenized and again centrifuged as above. The resulting pellet was

resuspended in 5ml lysis buffer (5mM PIPES, pH 8.0, 85mM KCl, 0.5% NP-40

alternate) supplemented with protease inhibitor complex 2 (PIC2, 10µg/ml

leupeptin, 5µg/ml pepstatin, 0.2mg/ml PMSF, 50µg/ml ALLN, Sigma). Nuclei

were released after 30 strokes using a Dounce homogenizer and collected after

centrifugation as above. Pellets were resuspended in 6ml homogenization buffer

(10mM HEPES, pH 7.6, 25mM KCl, 1mM EDTA, 1mM EGTA, 1M Sucrose, 10%

glycerol, 0.15mM spermine, supplemented with PIC1) and layered onto 3ml of

the same buffer. Nuclei were then pelleted at 26,000rpm for 1 hour (Beckman

SW41 rotor) and stored at -80ºC. Nuclear pellets were re-suspended in 0.3ml

nuclear lysis buffer (50mM Tris pH 7.6, 10mM EDTA and 1% SDS), and diluted

with 0.6ml immunoprecipitation (IP) dilution buffer (0.01% SDS, 1.1% Triton

x100, 167mM NaCl, 16.7mM Tris pH 7.6, 1.2mM EDTA). For sonication, 0.3ml

(1/3) of nuclear lysate was sonicated for 25-30 cycles 30 seconds on 30 seconds

off at 4ºC with BioRuptor twin sonicator (Diagenode). Sonicated chromatin was

then further diluted to 1ml with IP dilution buffer, which is sufficient for three ChIP

reactions. The BLRP antibody (Genescript) was conjugated to Protein G

Dynabeads (Life Technologies) in PBS containing 0.5% BSA overnight at 4ºC

with gentle rotation. Sonicated DNA (0.33ml made up to 1ml in IP dilution buffer)

was then incubated overnight with Protein G-BLRP beads at 4ºC with gentle

rotation. For input samples, 10% was used. Samples were then washed twice

with wash buffer I (20mM Tris, pH 7.4, 150mM NaCl, 0.1% SDS, 1% Triton X100,

2mM EDTA), three times with alternate wash buffer III (10mM Tris, pH 7.4,

250mM LiCl, 1% NP-40 alternate, 0.7% Deoxycholate, 1mM EDTA), two washes

with 0.2% Triton X100 TE buffer and two final washes with 50mM NaCl

containing TE buffer. Chromatin was recovered and DNA from IP and input was

isolated after reverse crosslinking (incubating at 65ºC overnight in 0.3M NaCl and

Proteinase K digestion) and DNA purified using gel/PCR extraction kit (Clontech).

For qPCR analysis, samples were diluted 1:10 and run in triplicate as described

above. A standard curve for PCR was generated from 4 log serial dilutions of

input samples and data expressed as percentage of input. Primer sequences for

ChIP analysis are provided in Table S2.

RT-qPCR Analysis: RT-qPCR standards were prepared from an aliquot from

each cDNA reaction that was then pooled. Standards which were diluted over 3

log range (dilution range 1:5, 1:10, 1:50, 1:100, 1:500, 1:1000). Quantitative PCR

was carried out with primers designed to cross exon-exon boundaries using the

Roche UPL primer design website. RT-qPCR primer sequences are provided in

Table S1. Quantitative PCR was carried out in triplicate for each sample in 384-

well format using Kapa LC480 SYBR green mix (Kapa Biosciences). Quantitative

PCR was carried out using a Lightcycler 480 (Roche) and concentrations were

determined from the standard curve using the efficiency corrected method (2nd

derivative max, Roche). Relative quantification was determined by normalizing

the expression of each gene to a housekeeper. Two housekeepers were used to

normalize gene expression data (36B4 and Tbp).

Microarray Analysis: Hepatic RNA was isolated from 3 mice treated with Ad-

Control or Ad-MafG and global gene expression measured by the UCLA

Neuroscience Genomics Core (UNGC) using Illumina microarrays (Illumina).

Microarray analysis was carried out using Genome Suite (Illumina) and

differential gene expression determined from probes that that had expression of

p>0.05, (approximately 10,000 genes left over after filtering). Of those, genes

that were altered 1.5 fold were considered differentially expressed. Gene

enrichment analysis was carried out using DAVID and pathway analysis using

KEGG (Huang da et al., 2009).

Supplemental References Chong, H.K., Biesinger, J., Seo, Y.K., Xie, X., and Osborne, T.F. (2012).

Genome-wide analysis of hepatic LRH-1 reveals a promoter binding preference

and suggests a role in regulating genes of lipid metabolism in concert with FXR.

BMC Genomics 13, 51.


X., and Osborne, T.F. (2010). Genome-wide interrogation of hepatic FXR reveals

an asymmetric IR-1 motif and synergy with LRH-1. Nucleic Acids Res 38, 6007-

6017.


and Edwards, P.A. (2013). MicroRNA-144 regulates hepatic ATP binding

cassette transporter A1 and plasma high-density lipoprotein after activation of the

nuclear receptor farnesoid X receptor. Circ Res 112, 1602-1612.

Heinz, S., Benner, C., Spann, N., Bertolino, E., Lin, Y.C., Laslo, P., Cheng, J.X.,

Murre, C., Singh, H., and Glass, C.K. (2010). Simple combinations of lineage-

determining transcription factors prime cis-regulatory elements required for

macrophage and B cell identities. Mol Cell 38, 576-589.

Hirotsu, Y., Katsuoka, F., Funayama, R., Nagashima, T., Nishida, Y., Nakayama,

K., Engel, J.D., and Yamamoto, M. (2012). Nrf2-MafG heterodimers contribute

globally to antioxidant and metabolic networks. Nucleic Acids Res 40, 10228-

10239.

Huang da, W., Sherman, B.T., and Lempicki, R.A. (2009). Systematic and

integrative analysis of large gene lists using DAVID bioinformatics resources. Nat

Protoc 4, 44-57.

Menet, J.S., Rodriguez, J., Abruzzi, K.C., and Rosbash, M. (2012). Nascent-Seq

reveals novel features of mouse circadian transcriptional regulation. Elife 1,

e00011.

Schmidt, D.R., Holmstrom, S.R., Fon Tacer, K., Bookout, A.L., Kliewer, S.A., and

Mangelsdorf, D.J. (2010). Regulation of bile acid synthesis by fat-soluble

vitamins A and D. J Biol Chem 285, 14486-14494.

Seth, P.P., Vasquez, G., Allerson, C.A., Berdeja, A., Gaus, H., Kinberger, G.A.,

Prakash, T.P., Migawa, M.T., Bhat, B., and Swayze, E.E. (2010). Synthesis and

biophysical evaluation of 2',4'-constrained 2'O-methoxyethyl and 2',4'-constrained

2'O-ethyl nucleic acid analogues. J Org Chem 75, 1569-1581.

Thomas, A.M., Hart, S.N., Kong, B., Fang, J., Zhong, X.B., and Guo, G.L. (2010).

Genome-wide tissue-specific farnesoid X receptor binding in mouse liver and

intestine. Hepatology 51, 1410-1419.

MAFG Is a Transcriptional Repressor of Bile Acid Synthesis and ...

Documents

Transcript of MAFG Is a Transcriptional Repressor of Bile Acid Synthesis and ...