Activation of nuclear receptor CAR amelioratesdiabetes and fatty liver diseaseBingning Donga, Pradip K. Sahab, Wendong Huangc, Wenling Chend, Lutfi A. Abu-Elheigae, Salih J. Wakile,1,Robert D. Stevensf, Olga Ilkayevaf, Christopher B. Newgardf, Lawrence Chanb, and David D. Moorea,b,1

aDepartment of Molecular and Cellular Biology and bDivision of Diabetes, Endocrinology, and Metabolism, Department of Medicine, and eVerna and MarrsMcLean Department of Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, TX 77030; cDepartment of Gene Regulation and DrugDiscovery, Beckman Research Institute, City of Hope National Medical Center, Duarte, CA 91010; dDepartment of Molecular Genetics, University of TexasSouthwestern Medical Center, Dallas, TX 75390-9046; and fDepartment of Pharmacology and Cancer Biology, Sarah W. Stedman Nutrition and MetabolismCenter, Duke University Medical Center, Durham, NC 27710

Contributed by Salih J. Wakil, September 9, 2009 (sent for review July 7, 2009)

Constitutive androstane receptor CAR (NR1I3) has been identifiedas a central mediator of coordinate responses to xenobiotic andendobiotic stress. Here we use leptin-deficient mice (ob/ob) andob/ob, CAR�/� double mutant mice to identify a metabolic role ofCAR in type 2 diabetes. Activation of CAR significantly reducesserum glucose levels and improves glucose tolerance and insulinsensitivity. Gene expression analyses and hyperinsulinemic eugly-cemic clamp results suggest that CAR activation ameliorates hy-perglycemia by suppressing glucose production and stimulatingglucose uptake and usage in the liver. In addition, CAR activationdramatically improves fatty liver by both inhibition of hepaticlipogenesis and induction of �-oxidation. We conclude that CARactivation improves type 2 diabetes, and that these actions of CARsuggest therapeutic approaches to the disease.

insulin resistance � nuclear receptor � xenobiotics

Type 2 diabetes mellitus (T2D) is primarily characterized byinsulin resistance and leads to uncontrolled glucose and lipid

metabolism (1, 2). In diabetic subjects, the major insulin-resistant organs are liver, muscle, and adipose tissue. Liver is themajor organ for endogenous glucose production and lipogenesis,both of which are tightly controlled by various metabolic andnutritional factors (3). Diabetic liver is unresponsive to insulinsuppression of glucose output, but continues to produce largeamounts of lipids. Hepatic overproduction of glucose and fattyacids further impairs insulin signaling thereby establishing avicious cycle (4, 5). Liver is also a principal organ of drugmetabolism, which depends on a variety of phase I oxidizingenzymes, primarily cytochrome P450’s (CYPs) with broad sub-strate specificities, as well as phase II conjugating enzymes andphase III transporters (6). CAR (constitutive androstane recep-tor) and PXR (pregame X receptor) constitute two essentialmembers of the nuclear receptor family. They function as sensorsof toxic derivatives from xenobiotic and endobiotic metabolism.In general, CAR appears to be more essential in response toendogenous stimuli, while PXR appears to act primarily inxenobiotic induction of drug metabolism. Highly expressed inthe liver, CAR is activated by phenobarbital (PB) and a groupof structurally diverse agents referred to as ‘‘phenobarbital like,’’such as TCPOBOP (TC) (7).

A link between drug metabolism and diabetes was suggestedby reports that PB treatment decreases plasma glucose andimproves insulin sensitivity, not only in a diabetic rodent model(8), but also in human T2D patients (9). Moreover, treatmentswith CAR activators PB and TC repress the expression of thehepatic gluconeogenic enzymes phosphoenolpyruvate car-boxykinase (PEPCK) and glucose-6-phosphatase (G6P) inmouse liver (10) and rat hepatocytes (11). Recently, this effecthas been attributed to the inhibitory effect of CAR on insulindependent activity of FOXO1 (12).

We used ob/ob and ob/ob CAR�/� double mutant mice toinvestigate the role of CAR in T2D. Our results show that the

antidiabetic effects of the more specific CAR agonist TCPOBOP(TC) are lost in the double mutant mice, and that activation ofCAR significantly affects both glucose and lipid homeostasis. Inparticular, CAR activation represses gluconeogenesis, whileincreasing hepatic glucose uptake and utilization. It also re-presses lipogenic gene expression via a mechanism dependent atleast in part on inactivation of oxysterol ligands for the liver Xreceptors LXR� and �, and increases �-oxidation via an unex-pected pathway associated with decreased expression of acetyl-CoA carboxylase 2 (ACC2).

ResultsCAR Activation Improves Glucose Metabolism in ob/ob Mice. To studythe role of CAR in T2D, we crossed the CAR�/� allele into theob/ob background to generate double mutant mice (ob/ob,CAR�/�). Loss of CAR function in the ob/ob background hadno effect on a number of parameters, including body weight, foodintake, fat-to-body weight ratio, and serum levels of triglycerides,cholesterol, and free fatty acids (Fig. S1 A–F). The lack of effecton serum triglycerides contrasts with a recent report using ob/ob,CAR�/� mice generated using a distinct CAR�/� allele (13). Inthat study, the effect on serum triglycerides was attributed to anactivation of PPAR�, which we did not observe in our ob/ob,CAR�/� mice (Fig. S1 G and H). We conclude that our ablationof CAR in the ob/ob background does not affect the obesephenotype.

To test whether diabetes and obesity affect CAR activity,10-week-old ob/ob and ob/ob CAR�/� mice were treated withthe CAR specific agonist TCPOBOP (TC) for a week. Asexpected, expression of CAR targets Cyp2B10 and Cyp3A11were robustly elevated in the TC-treated ob/ob mice, but not inthe ob/ob, CAR�/� mice or vehicle treated ob/ob mice. Thus, theob/ob background has no apparent effect on either the basalactivity or the agonist response of CAR (Fig. S2).

To initially investigate how CAR influences glucose metab-olism, we monitored glucose levels in ob/ob and ob/ob, CAR�/�

mice in response to an acute 1-week TC treatment. Serumglucose levels were significantly decreased in TC-treated ob/ob,but not ob/ob, CAR�/� mice (Fig. 1A). In i.p. glucose tolerancetests (GTT), TC-treated ob/ob mice displayed significantly im-proved glucose tolerance compared with control treated ob/obmice, as shown by reduced blood glucose levels at each timepoint, and this effect was also completely absent in the double

Author contributions: B.D., P.K.S., W.H., L.A.A.-E., S.J.W., C.B.N., L.C., and D.D.M. designedresearch; B.D., P.K.S., R.D.S., and O.I. performed research; W.C. contributed new reagents/analytic tools; B.D., P.K.S., W.H., L.A.A.-E., S.J.W., R.D.S., O.I., C.B.N., L.C., and D.D.M.analyzed data; and B.D. and D.D.M. wrote the paper.

The authors declare no conflict of interest.

1To whom correspondence may be addressed. E-mail: swakil@bcm.imc.edu ormoore@bcm.tmc.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0909731106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0909731106 PNAS � November 3, 2009 � vol. 106 � no. 44 � 18831–18836

PHYS

IOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

mutant ob/ob CAR�/� mice (Fig. 1B). The decreased serumglucose levels were associated with appropriately decreasedserum insulin levels, indicating that CAR activation alleviateshyperinsulinemia in the ob/ob mice and improves glucose tol-erance (Fig. 1C).

Since T2D is a chronic disease, we examined the effects oflong-term CAR activation. This longer term treatment alsominimizes the impact of the CAR-dependent hepatomegaly thatoccurs over the first week of treatment. Starting at 6 weeks ofage, ob/ob and ob/ob, CAR�/� mice were treated with one doseof TC per week for 1 month. ob/ob mice gradually develop T2Dover this time course, as revealed by markedly elevated serumglucose in the vehicle treated ob/ob group (Fig. S3). TC treat-ment significantly attenuated this diabetes progression in ob/obmice, but not in ob/ob, CAR�/� mice (Fig. S3 A and B). TheTC-treated ob/ob mice showed a dramatic improvement in theGTT, which was not observed in the ob/ob, CAR�/� mice (Fig. 2A).

To critically determine whether CAR activation improves

insulin sensitivity, 1-month TC-treated or control ob/ob micewere studied using the hyperinsulinemic euglycemic clamp. TCtreatment resulted in a dramatic 3-fold increase in glucoseinfusion rate (GIR), demonstrating markedly improved insulinsensitivity (Fig. 2CII). TC treatment also strongly repressed boththe basal glucose production and hepatic glucose production(Fig. 2 CI and CIII). The difference in GIR can be mostlyattributed to the insulin-induced reduction of hepatic glucoseproduction because peripheral glucose uptake remains the samein both groups (Fig. 2CIV). These results suggest that activationof CAR improves liver insulin sensitivity mainly through sup-pression of de novo glucose production.

To rule out concerns related to the potential impact of theabsence of leptin signaling in the ob/ob mice, we studied theeffect of CAR activation in the diet induced obesity model ofinsulin resistance. Wild-type and CAR�/� mice were fed with ahigh-fat diet (45% of calories) for 2 months and treated withvehicle or TC for another month while maintaining the same

Fig. 1. Activation of CAR represses serum glucose level and improves glucose tolerance after 1-week treatment in ob/ob mice. (A) Fed and Fasted blood glucoselevels were measured in each group. (n � 5, *P � 0.01) (B and C) After 16 h of fasting, mice were performed glucose tolerance test. Blood glucose and plasmainsulin values were assessed. (n � 6, *P � 0.01)

18832 � www.pnas.org�cgi�doi�10.1073�pnas.0909731106 Dong et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

diet. Consistent with the ob/ob results, TC treatment signifi-cantly improved glucose tolerance in wild-type mice, but not inthe CAR�/� mice (Fig. 2D).

CAR Regulates Glucose Metabolic Genes. Both the reduced fastingserum glucose level and decreased hepatic glucose productionduring the clamp in the TC-treated ob/ob mice indicate thatgluconeogenesis is strongly suppressed by CAR activation. Con-sistent with previous studies in lean mice (12, 14), we found thatexpression of the gluconeogenic genes PEPCK and G-6-Pase(G6P) was repressed by TC treatment in the ob/ob mice, but notin the ob/ob, CAR�/� double mice (Fig. 3). Previous studies haveshown that phenobarbital treatment increases glucose uptakeand utilization in the liver of ob/ob mice, as indicated byincreased activity of key enzymes such as hexokinase (HK) andphosphogluconate dehydrogenase (PGD) (9, 15). Hexokinasescatalyze glucose phosphorylation, the essentially irreversible

first step of the glycolytic pathway (16). In accord with theprevious studies of enzymatic activities, CAR activation induceshexokinase mRNA (Fig. 3), but does not affect glucokinase geneexpression. PGD, the rate-limiting enzyme in the pentose phos-phate pathway, is also induced by approximately 2-fold in theTC-treated ob/ob mice but not in the ob/ob, CAR�/� doublemutant mice (Fig. 3). As expected, the responses of these genesto CAR activation in high-fat-diet-fed mice were similar to thoseobserved in the ob/ob mice (Fig. S4). In transient transfectionswith proximal promoter constructs, CAR was able to transac-tivate the hexokinase promoter and, to a lesser extent, the PGDpromoter, indicating that both genes are primary CAR targets(Fig. S5). Overall, we conclude that CAR regulates glucosehomeostasis by suppressing gluconeogenesis and inducing glu-cose uptake and consumption in the liver.

CAR Activation Improves Fatty Liver in ob/ob Mice. Fatty liver diseaseis highly associated with T2D. To determine the effect of CARactivation on hepatic fat accumulation, liver samples from1-month control and TC-treated ob/ob mice were examined forhistology and lipid content. TC treatment significantly improvedfatty liver histology and decreased hepatic triglycerides in ob/obmice, and these responses were absent in the ob/ob, CAR�/�

mice (Fig. 4 A and B and Fig. S6).Reduced hepatic lipid content can be due to decreased de

novo hepatic lipogenesis or increased fatty acid �-oxidation. Wefirst examined expression of lipogenic genes in response to1-month TC treatment. Compared to vehicle treated ob/ob mice,TC-treated ob/ob mice showed significant repression of not onlySREBP-1c mRNA, but also its downstream genes, such as FASand SCD-1 (Fig. 4C). Again, these responses were absent in thedouble mutant ob/ob CAR�/� mice. These results are consistentwith the previous demonstration that TC treatment decreasesSREBP-1c protein levels (17), and strongly suggest that CARactivation suppresses lipogenesis.

Liver SREBP-1c is regulated by insulin through liver Xreceptor (LXR) signaling (18, 19). Cholesterol sulfotransferase2B1b (SULT2B1b) attenuates LXR signaling by sulfating andinactivating oxysterol agonist ligands (20). Sulfotransferases, likeother phase II drug metabolism enzymes, are potential CARtargets (7, 21), and CAR activation does induce SULT2B1bexpression (Fig. 5A). We used SULT2B1b null mice to test therelevance of this induction for the effect of TC on SREBP-1c

A

D

Control TCS

erum

glu

cose

(mg/

dl)

Minutes after glucose injection

ob/ob CAR-/-

0100200300400500600700

0 15 30 45 60 75 90 105 120*

ob/ob

0100200300400500600700

0 15 30 45 60 75 90 105 120

* ***

Control TreatedB C

TC TreatedG

luco

seU

ptak

e (m

g/g/

min

)

0102030405060

BG

P (

mg/

kg/h

r)

051015202530

HG

P (

mg/

kg/h

r)

05

10152025

GIR

(m

g/kg

/hr)

*

(I) *(II)

*

(III)

0510152025

Soleus WAT

(IV)

Ser

um in

sulin

(ng/

ml)

05

101520253035

ob/ob,CAR-/-ob/ob

**

Minutes after glucose injection

Ser

um g

luco

se(m

g/dl

)

WT 45 kcal% fat

CAR-/- 45 kcal% fat

ControlTC

0100200300400500600700

0 15 30 45 60 75 90 105 1200

100200300400500600700

0 15 30 45 60 75 90 105 120

** *

**

Fig. 2. Activation of CAR improves glucose tolerance and insulin sensitivityafter1-monthtreatment inob/obmiceandDIOmice. (A)ob/obandob/obCAR�/�

mice were treated with TC or corn oil control for 1 month. After 16 h of fasting,glucose tolerance tests were performed. Blood glucose values were assessed. (n�6, *P � 0.01) (B) Serum insulin level was measured after 6 h fasting in eachtreatmentgroups. (n�4,*P�0.01) (C)ob/obmicefrom1-monthtreatmentwereperformed hyperinsulinemic-euglycemic clamp studies. Basal glucose production(CI), glucose infusion rate (CII), hepatic glucose production (CIII) and glucoseuptakefromperipheral tissue (CIV)wereassessedduringclamp. (n�4,*P�0.05)(D) Wild-type and CAR�/� mice (n � 6) were fed on 45 Kcal% diet or control for2 months and treated with TC or control for 1 month. After 16 h of fasting, micewere performed glucose tolerance test. Glucose levels at each time point weremeasured. (n � 6, *P � 0.01)

β-Actin

PEPCK

G-6-Pase

TC: - - + + - - + +ob/ob CAR-/-ob/ob

ob/ob,CAR-/-ob/ob,CAR-/-mR

NA

rel

ativ

e va

lue

0.5

1.5

0

1

2 HK

0

PGD

ob/ob ob/ob

**

123456

Control Treated TC Treated

Fig. 3. CAR regulates glucose metabolizing genes expression. Ob/ob andob/ob CAR�/� mice were treated with TC or corn oil control for 1 month. Livertotal RNA was isolated from mice of different treatments and equal amountsof RNA were pooled from two individual mice and loaded on one lane.Northern blot was performed with Pepck and G6p probes. Hexokinase andPgd were analyzed by quantitative RT-PCR. (n � 4, *P � 0.01)

Dong et al. PNAS � November 3, 2009 � vol. 106 � no. 44 � 18833

PHYS

IOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

expression. As expected, CAR induction of Cyp2B10 expressionwas unaltered in the SULT2B1b null mice, and repression ofPEPCK expression was also unaffected (Fig. S7A). TC repres-sion of SREBP-1c expression was lost in the Sult2B1b knockoutmice (Fig. 5B), while the repression of SCD-1 expression wasattenuated. Expression of the LXR target genes ABCG5 andABCG8 was also suppressed by CAR activation (Fig. S7B),confirming the suppression of LXR signaling. These results

indicate that this indirect pathway of LXR ligand inactivationplays an important role in CAR suppression of lipogenesis,although other mechanisms are likely to contribute.

We used a metabolomic approach to further characterize thephysiological role of CAR in lipid metabolism. Comparing aseries of metabolites in TC vs. control treated liver samplesrevealed a marked induction of acylcarnitines, essential inter-mediates in mitochondrial fatty acid import and �-oxidation, inob/ob mice but not in ob/ob, CAR�/� mice (Fig. 6A). Threeimportant organic acids in the TCA cycle, fumarate, malate, and�-ketoglutarate, were also induced by CAR activation (Fig. 6A).The absence of a similar response of pyruvate indicates that theinduced TCA products are derived from increased �-oxidation,rather than from glycolysis. Consistent with this, serum ketonebodies, an indicator of increased �-oxidation, were increased byTC treatment in ob/ob mice, but not in ob/ob, CAR�/� mice (Fig.6B). �-oxidation was examined directly in primary hepatocytesfrom wild-type and CAR�/� mice treated with TC or vehicle for3 days. As expected, [14C]-palmitate oxidation was significantlyincreased by TC in the wild-type hepatocytes, whereas therewere no differences in TC vs. control groups from CAR�/� mice(Fig. 6B).

We examined expression of a number of genes related to keysteps in �-oxidation, including PPAR�, CPT-1, CPT-2, anddifferent acyl-CoA dehydrogenases, but did not observe anobvious basis for this increase. However, �-oxidation is potentlysuppressed by malonyl-CoA, the initial committed product inlipogenesis and an allosteric inhibitor of the carnitine-dependentpathway of mitochondrial fatty acid import (22, 29). Consistentwith a report that ACC2 is a direct SREBP-1c target (23), wefound that ACC2 and also ACC1 mRNA expression was signif-icantly suppressed by CAR activation (Fig. 6C). Thus, CARactivation could promote �-oxidation indirectly by suppressingmalonyl-CoA production.

DiscussionPrevious studies have shown beneficial effects of PB on insulinsensitivity in both mouse models (8) and human T2D patients (8,9), but the basis for these antidiabetic effects has been unclear.We used two complementary approaches to critically test thehypothesis that CAR mediates these effects. The first is thepharmacologic substitution of TC, a much more specific mouseCAR agonist that does not activate AMP kinase, for PB. Thesecond is the genetic introduction of the CAR�/� allele into theob/ob mouse model of obesity and T2D. TC induces antidiabeticeffects very similar to those described for PB in the ob/ob model,and those effects are lost in the ob/ob, CAR�/� double mutants,definitively establishing CAR as the mediator of such antidia-betic effects. While this manuscript was in review, similarantidiabetic effects of TC were reported (24). TC treatmentdecreased body weight in those studies, possibly due to toxiceffects that do not occur in our mice. This decreased body weightcould amplify the antidiabetic effect, but is clearly not essentialfor the TC activity that we observed or the previous results withPB, which also did not decrease body weight (8).

Increased hepatic glucose production is a major contributor tohyperglycemia in T2D. The hyperinsulinemic euglycemic clampstudies show that CAR activation markedly improves insulinsensitivity and decreases hepatic glucose production in the ob/obmice. In accord with this, and also with both earlier studies of PBeffects on enzyme activities (8) and more recent studies (12, 14),TC treatment suppresses expression of the key gluconeogenicgenes PEPCK and G6P. CAR also induces important genes inglucose uptake (hexokinase), and also utilization via the pentosephosphate pathway (PGD), which supports drug metabolism bygenerating NADPH, an obligate cofactor for cytochrome P450reductase. Outside of the context of the xenobiotic response,activation of the pentose phosphate shunt has also been sug-

ob/o

bob

/ob,

CA

R-/

-

TC

TCCO

CO

A

Hepatic Lipids

(µg/mg protein)

ob/ob ob/ob,CAR-/- ob/ob ob/ob,CAR-/-01020304050

*

048

1216

*

TG NEFA

mR

NA

Rel

ativ

eva

lue

SREBP-1c FAS

0123456

**

00.51

1.52

2.5SCD-1

**0

5

10

15

20

*

ob/ob,CAR-

/-

ob/ob,CAR-/-

ob/ob,CAR-/-

ob/ob ob/ob ob/ob

B

C

control TC

Fig. 4. CAR activation reduces liver lipids deposition and lipogenic geneexpression in ob/ob mice. (A) Liver samples from ob/ob and ob/ob CAR�/� miceof 1-month TC or control treatment were assessed for oil red O staining. (B)Hepatic triglyceride and NFFA were measured for quantitative liver lipids. (n �4, *P � 0.05) (C) Liver RNA was extracted and gene expression analyzed byquantitative RT-PCR. Gene names were shown on top of each figure. (n � 4,

**P � 0.01, *P � 0.05).

control TC

mR

NA

Rel

ativ

e va

lue

A B

ob/ob,

CAR-/-

ob/ob01530456075

SULT2B1b **

WT Sult2b1-/-0

1

2

3

4

WT Sult2b1-/-

SREBP-1c SCD-1

0246810

***

Fig. 5. SULT2B1 plays an important role in CAR mediated suppression oflipogenic gene expression. (A) Ob/ob and ob/ob CAR�/� mice were treatedwith TC or corn oil control for 1 month. Liver RNA was extracted and geneexpression analyzed by quantitative RT-PCR. Gene names were shown on topof each figure. (n � 4, **P � 0.01, *P � 0.05) (B) Wild-type and SULT2B1knockout mice were treated with TC or control for 3 days. Liver RNA wasextracted and gene expression was assessed by quantitative RT-PCR. Lipogenicgene expressions were compared between WT and SULT2B1 KO groups. (n �4, **P � 0.01, *P � 0.05)

18834 � www.pnas.org�cgi�doi�10.1073�pnas.0909731106 Dong et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

gested to contribute to the suppression of hyperglycemia byPPAR� activation (25).

Elevated liver triglycerides are tightly associated with insulinresistance and T2D, and PB treatment alleviates fatty liver inrodents (8, 26), and possibly human T2D patients (27). Ourresults with TC are also consistent with another recent studyusing the methionine and choline deficient diet to induce fattyliver (28). We identified two mechanisms for the impact of CARactivation on hepatic steatosis: suppression of lipogenesis andinduction of �-oxidation. TC treatment decreased expression ofthe lipogenic transcription factor SREBP-1c and its downstreamgenes. These responses are consistent with a recent report thatTC treatment decreased nuclear levels of the mature SREBP-1cprotein (17).

Exploring the molecular basis for decreased SREBP-1c ex-pression, we found that CAR induces the phase II enzymeSULT2B1b, which sulfates and inactivates oxysterol agonists forLXR, a well known activator of SREBP-1c expression. Thesuppressive effect of CAR activation on lipogenic gene expres-sion is substantially impaired in SULT2B1b knockout mice.Thus, it is likely that this indirect pathway of LXR ligandinactivation contributes significantly to the decreased SREBP-1cexpression, and that there are additional mechanisms, includingthe previously described induction of INSIG-1 as primary CARtarget (17).

While we observed a clear increase in hepatic �-oxidation, wedid not find substantial effects of CAR activation on geneexpression of key components of this pathway. However, ourresults are consistent with recent studies showing that PBtreatment induces serum ketone bodies, indicating increased�-oxidation (30), and that CAR activation induces enzymesinvolved in fatty acid microsomal omega-oxidation and peroxi-somal �-oxidation (28). Consistent with the decreased triglyc-eride accumulation in the TC-treated ob/ob livers, metabolomic

profiling revealed markedly elevated levels of acylcarnitines andTCA cycle organic acids, suggesting increased �-oxidation. Thiswas confirmed by observations that CAR activation results inelevated levels of serum ketone bodies and, more directly,increases palmitate oxidation in TC-treated hepatocytes.

Decreased levels of malonyl-CoA provide a plausible mech-anism for this increase. Production of malonyl-CoA by ACC1and ACC2 initiates lipogenesis, and strongly suppresses theopposing �-oxidation pathway by allosteric inhibition of CPT-1enzymatic activity, which decreases fatty acid transport intomitochondria. Decreasing activity of ACC1 and particularlyACC2 has beneficial effects on both hepatic steatosis and insulinresistance (31), and very similar effects are achieved by overex-pression of malonyl CoA decarboxylase (32). CAR activationsignificantly represses both ACC1 and ACC2 expression,strongly suggesting that CAR induces �-oxidation by decreasingmalonyl CoA levels. Recent studies have shown that SCD-1ablation protects against hepatic steatosis by the combinedeffects of decreased lipogenesis and increased �-oxidation (33).Hence, reduced SCD-1 expression by CAR activation may alsocontribute to this induced �-oxidation.

In summary, our results document the beneficial impact ofspecific CAR activation in T2D and fatty liver disease, andidentify mechanisms that can account for such effects. PB is notused for treating T2D patients because of its undesirable sideeffects, such as promoting hepatocyte proliferation and increas-ing drug-to-drug interaction. As with other nuclear receptors,however, it may be possible to limit or even eliminate theseproblems using selective CAR modulators that retain beneficialeffects on glucose and lipid metabolism, but not undesirableeffects on drug metabolism and proliferation.

Experimental ProceduresMetabolic Studies. Glucose tolerance test was performed by i.p. injectingglucose (0.625g/kg for obese mice and 1.5g/kg for lean mice). Blood glucose

Control TC

Fum

arat

e

Acy

lcar

nitin

es (

uM)

0

0.05

0.1

0.15

0.2

0.25

0.0000.0020.0040.0060.0080.0100.0120.014

ob/o

b,C

AR

-/-

0.0000.0050.0100.0150.0200.0250.030

0.1

0.2

0.3

0.4

0.5

ob/o

b

0C

4/C

i4

C5'

s

C4-

OH C6

* *

*

*

C10

:3

C16

:2

C18

:2

C18

:1

C18

C18

:2-O

H

C20

* *

*

*

**

*

Org

anic

aci

d (p

mol

/ul)

050

100150200250

Mal

ate

*

050

100150200250

�-K

eto-

glut

arat

e

Pyr

uvat

e

**

ob/o

b,C

AR

-/-

ob/o

b

A

050

100150200250300

ob/ob ob/ob,CAR -/-

*

0

1

2

3

4

mR

NA

R

elat

ive

valu

e

012345

ob/ob ob/ob, CAR-/-

Acc1 Acc2 Acc1 Acc2

**

**

[14 C

]pal

mita

te

CP

M/1

05ce

lls

03006009001200150018002100

ob/ob ob/ob,CAR -/-

**

Ket

one

Bod

ies

(µm

ol/L

)

B C

Fig. 6. CAR activation induces �-oxidation in the liver. (A) Ob/ob and ob/ob CAR�/� mice were treated with TC or corn oil control for 1 month. Liver sampleswere used for measurement of a series of acylcarnitines with fatty acyl side chains of the indicated lengths, and also the indicated organic acids by stable isotopedilution mass spectrometry. (n � 5, *P � 0.01) (B) Ob/ob and ob/ob CAR�/� mice were treated with TC or corn oil control for 1 month. Mice were fasted for 6 hand serum ketone bodies were measured. (n � 6, *P � 0.05) WT and CAR�/� mice were treated with TC or control for 3 days. Primary hepatocytes were measuredand performed �-oxidation assay. (n � 3, *P � 0.01) (C) Liver total RNA from 1 month treated ob/ob and ob/ob CAR�/� mice were examined for ACC1 and ACC2expression. (n � 4, *P � 0.01)

Dong et al. PNAS � November 3, 2009 � vol. 106 � no. 44 � 18835

PHYS

IOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

was measured by tail bleeds at 0, 15, 30, 60, and 120 min post glucose dose.Hyperinsulinemic clamp (insulin dose is 15 mU/kg/min) was performed andcalculated as described in our previous publication (34). Acylcarnitines andorganic acids were measured using stable isotope dilution techniques aspreviously described (35, 36). Acylcarnitine species were measured using flowinjection tandem mass spectrometry (MS/MS) and sample preparation meth-ods described previously (37, 38). Organic acids were quantified using apreviously described method that utilizes Trace GC Ultra coupled to a TraceDSQ MS operating under Excalibur 1.4 (Thermo Fisher Scientific) (35).

Gene Expression Analysis. Total RNA was extracted from mouse liver usingTRIzol reagent (Invitrogen). Equivalent amounts of RNA from each treatmentgroup were pooled, and 20 �g was used for Northern blot analysis. The

primers used for generation of each cDNA probe were described previously(39). All of the blots were stripped and hybridized subsequently with the�-Actin probe as the internal control. For Q-PCR experiment, RNA was reverse-transcribed by using SuperScript™ III RT (Invitrogen). Samples were run byusing SYBR green (Applied Biosystems) and compared with levels of GAPDH asa control. Thermal cycling was carried out with an ABI prism 7500 sequencedetection system (Applied Biosystems). Primers are purchased from Qiagen.

ACKNOWLEDGMENTS. We thank Drs. Yoon Kwang Lee, Ke Ma, and Rui Xiaofor helpful comments and discussion. This work was supported by NationalInstitutes of Health Grants R01 DK46546 (to D.D.M.) R01-DK068037 (to L.C.)and Diabetes and Endocrinology Research Center at Baylor College of Medi-cine Grant P30-DK079638.

1. Bouche C, Serdy S, Kahn CR, Goldfine AB (2004) The cellular fate of glucose and itsrelevance in type 2 diabetes. Endocr Rev 25:807–830.

2. McGarry JD (2002) Banting lecture 2001: Dysregulation of fatty acid metabolism in theetiology of type 2 diabetes. Diabetes 51:7–18.

3. Accili D (2004) Lilly lecture 2003: The struggle for mastery in insulin action: Fromtriumvirate to republic. Diabetes 53:1633–1642.

4. Biddinger SB, Kahn CR (2006) From mice to men: Insights into the insulin resistancesyndromes. Annu Rev Physiol 68:123–158.

5. Postic C, Dentin R, Girard J (2004) Role of the liver in the control of carbohydrate andlipid homeostasis. Diabetes Metab 30:398–408.

6. Remmer H, Merker HJ (1963) Drug-Induced Changes in the Liver Endoplasmic Reticu-lum: Association with Drug-Metabolizing Enzymes. Science 142:1657–1658.

7. Qatanani M, Moore DD (2005) CAR, the continuously advancing receptor, in drugmetabolism and disease. Curr Drug Metab 6:329–339.

8. Karvonen I, Stengard JH, Huupponen R, Stenback FG, Sotaniemi EA (1989) Effects ofenzyme induction therapy on glucose and drug metabolism in obese mice model ofnon-insulin dependent diabetes mellitus. Diabetes Res 10:85–92.

9. Lahtela JT, Arranto AJ, Sotaniemi EA (1985) Enzyme inducers improve insulin sensitivityin non-insulin-dependent diabetic subjects. Diabetes 34:911–916.

10. Manenti G, Dragani TA, Della Porta G (1987) Effects of phenobarbital and 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene on differentiated functions in mouse liver. ChemBiol Interact 64:83–92.

11. Argaud D, Halimi S, Catelloni F, Leverve XM (1991) Inhibition of gluconeogenesis inisolated rat hepatocytes after chronic treatment with phenobarbital. Biochem J280:663–669.

12. Kodama S, Koike C, Negishi M, Yamamoto Y (2004) Nuclear receptors CAR and PXRcross talk with FOXO1 to regulate genes that encode drug-metabolizing and gluconeo-genic enzymes. Mol Cell Biol 24:7931–7940.

13. Maglich JM, Lobe DC, Moore JT (2009) The nuclear receptor CAR (NR1I3) regulatesserum triglyceride levels under conditions of metabolic stress. J Lipid Res 50:439 –445.

14. Miao J, Fang S, Bae Y, Kemper JK (2006) Functional inhibitory cross-talk betweenconstitutive androstane receptor and hepatic nuclear factor-4 in hepatic lipid/glucosemetabolism is mediated by competition for binding to the DR1 motif and to thecommon coactivators, GRIP-1 and PGC-1alpha. J Biol Chem 281:14537–14546.

15. Stengard JH, et al. (1987) Hepatic drug metabolism and the activities of NADPHgenerating enzymes and glucose-6-phosphatase in phenobarbital treated geneticallyobese (ob/ob) mice. Biomed Pharmacother 41:389–396.

16. Wilson JE (2003) Isozymes of mammalian hexokinase: Structure, subcellular localiza-tion and metabolic function. J Exp Biol 206:2049–2057.

17. Roth A, et al. (2008) Regulatory cross-talk between drug metabolism and lipid ho-meostasis: Constitutive androstane receptor and pregnane X receptor increase Insig-1expression. Mol Pharmacol 73:1282–1289.

18. Horton JD, Bashmakov Y, Shimomura I, Shimano H (1998) Regulation of sterol regu-latory element binding proteins in livers of fasted and refed mice. Proc Natl Acad SciUSA 95:5987–5992.

19. Chen G, Liang G, Ou J, Goldstein JL, Brown MS (2004) Central role for liver X receptorin insulin-mediated activation of Srebp-1c transcription and stimulation of fatty acidsynthesis in liver. Proc Natl Acad Sci USA 101:11245–11250.

20. Chen W, Chen G, Head DL, Mangelsdorf DJ, Russell DW (2007) Enzymatic reduction ofoxysterols impairs LXR signaling in cultured cells and the livers of mice. Cell Metab 5:73–79.

21. Saini SP, et al. (2004) A novel constitutive androstane receptor-mediated and CYP3A-independent pathway of bile acid detoxification. Mol Pharmacol 65:292–300.

22. Saggerson D (2008) Malonyl-CoA, a key signaling molecule in mammalian cells. AnnuRev Nutr 28:253–272.

23. Abu-Elheiga L, Oh W, Kordari P, Wakil SJ (2003) Acetyl-CoA carboxylase 2 mutant miceare protected against obesity and diabetes induced by high-fat/high-carbohydratediets. Proc Natl Acad Sci USA 100:10207–10212.

24. Gao J, He J, Zhai Y, Wada T, Xie W (2009) The constitutive androstance receptor is ananti-obesity nuclear receptor that improves insulin sensitivity. J Biol Chem 284:25984–25992.

25. Lee CH, et al. (2006) PPARdelta regulates glucose metabolism and insulin sensitivity.Proc Natl Acad Sci USA 103:3444–3449.

26. Omura K, et al. (2007) Gene expression profiling of rat liver treated with serumtriglyceride-decreasing compounds. J Toxicol Sci 32:387–399.

27. Sotaniemi EA, Arranto AJ, Sutinen S, Stengard JH (1983) Treatment of noninsulin-dependent diabetes mellitus with enzyme inducers. Clin Pharmacol Ther 33:826–835.

28. Baskin-Bey ES, Anan A, Isomoto H, Bronk SF, Gores GJ (2007) Constitutive androstanereceptor agonist, TCPOBOP, attenuates steatohepatitis in the methionine choline-deficient diet-fed mouse. World J Gastroenterol 13:5635–5641.

29. Koves TR, et al. (2008) Mitochondrial overload and incomplete fatty acid oxidationcontribute to skeletal muscle insulin resistance. Cell Metab 7:45–56.

30. Kiyosawa N, et al. (2004) Molecular mechanism investigation of phenobarbital-induced serum cholesterol elevation in rat livers by microarray analysis. Arch Toxicol78:435–442.

31. Savage DB, et al. (2006) Reversal of diet-induced hepatic steatosis and hepatic insulinresistance by antisense oligonucleotide inhibitors of acetyl-CoA carboxylases 1 and 2.J Clin Invest 116:817–824.

32. An J, et al. (2004) Hepatic expression of malonyl-CoA decarboxylase reverses muscle,liver and whole-animal insulin resistance. Nat Med 10:268–274.

33. Dobrzyn P, et al. (2004) Stearoyl-CoA desaturase 1 deficiency increases fatty acidoxidation by activating AMP-activated protein kinase in liver. Proc Natl Acad Sci USA101:6409–6414.

34. Ma K, Saha PK, Chan L, Moore DD (2006) Farnesoid X receptor is essential for normalglucose homeostasis. J Clin Invest 116:1102–1109.

35. Jensen MV, et al. (2006) Compensatory responses to pyruvate carboxylase suppressionin islet beta-cells. Preservation of glucose-stimulated insulin secretion. J Biol Chem281:22342–22351.

36. Ferrara CT, et al. (2008) Genetic networks of liver metabolism revealed by integrationof metabolic and transcriptional profiling. PLoS Genet 4:e1000034.

37. Chen J, et al. (2004) Determination of SCH 211803 by nanoelectrospray infusion massspectrometry: Evaluation of matrix effect and comparison with liquid chromatogra-phy-tandem mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci809:205–210.

38. Wu JY, et al. (2004) ENU mutagenesis identifies mice with mitochondrial branched-chain aminotransferase deficiency resembling human maple syrup urine disease. J ClinInvest 113:434–440.

39. Zhang J, Huang W, Chua SS, Wei P, Moore DD (2002) Modulation of acetaminophen-induced hepatotoxicity by the xenobiotic receptor CAR. Science 298:422–424.

18836 � www.pnas.org�cgi�doi�10.1073�pnas.0909731106 Dong et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0