Cell, Vol.

81,687-693, June 2, 1995,Copyright© 1995by Cell Press

Identification of a Nuclear Receptor

That Is Activated by Farnesol Metabolites

Barry M. Forman, 1 Elizabeth Goode, 2 Jasmine Chen, 1 Introduction

Anthony E. Oro, ~ David J. Bradley, 3.
Thomas Perlmann, ~ Daniel J. Noonan, 4 Leo T. Burka, 5 Molecular cloning studies have demonstrated that recep-
Trevor McMorris, e William W. Lamph, 7 tors for steroid, retinoid, vitamin D, and thyroid hormones
Ronald M. Evans, ~,8 and Cary Weinberger ~ comprise a superfamily of regulatory proteins that are
8 Howard Hughes Medical Institute structurally and functionally related (Evans, 1988; Leid et
1 Gene Expression Laboratory al., 1992a). Known as nuclear hormone receptors, these
The Salk Institute for Biological Studies proteins bind to cis-acting elements in the promoters of
10010 North Torrey Pines Road their target genes and modulate gene expression in re-
La Jolla, California 92037 sponse to hormone. Nuclear receptors can be classified
2 Receptor Biology Group based on their DNA-binding properties (Evans, 1988;
5 Organic Chemistry Group Glass, 1994). The glucocorticoid, estrogen, androgen,
Laboratory of Reproductive and Developmental progestin, and mineralocorticoid receptors bind as homo-
Toxicology dimers to hormone response elements organized as in-
National Institute of Environmental Health Sciences verted repeats (Glass, 1994). A second class of receptors,
Research Triangle Park, North Carolina 27709 including the retinoic acid receptor (RAR), the thyroid hor-
3 Laboratory of Cell Biology mone receptor (T3R), the vitamin D3 receptor, the fatty acid/
National Institutes of Health peroxisome proliferator-activated receptor (PPAR), and the
Bethesda, Maryland 20892 ecdysone receptor (EcR), bind as heterodimers with a com-
4 Department of Biochemistry mon partner, retinoid X receptor (RXR) (Yu et al., 1991 ; Hal-
University of Kentucky lenbeck et al., 1992; Kliewer et al., 1992; Leid et al., 1992b;
Lexington, Kentucky 40506 Marks et al., 1992; Zhang et al., 1992), the 9-cis retinoic
6 Department of Chemistry acid receptor (Heyman et al., 1992; Levin et al., 1992).
University of California, San Diego An important advance in the characterization of this su-
La Jolla, California 92093 perfamily of regulatory proteins has been the delineation
7 Ligand Pharmaceuticals, Incorporated of a growing number of gene products (orphan receptors)
9393 Towne Centre Drive that possess the structural features of hormone receptors,
San Diego, California 92121 but that lack known ligands. The search for hormonal acti-
vators for these newly discovered orphan receptors has
created exciting areas of research on previously unknown
Summary
signaling pathways (Issemann and Green, 1990; Heyman
et al., 1992; Levin et al., 1992). The ability to identify novel
Nuclear hormone receptors comprise a superfamily of
hormonal systems has important implications in physiol-
ligand-modulated transcription factors that mediate
ogy as well as human disease and its treatment.
the transcriptional activities of steroids, retinoids, and
As receptors have been identified for all known nuclear-
thyroid hormones. A growing number of related pro-
acting hormones, a question arises as to the types of mole-
teins have been identified that possess the structural
cules that may activate orphan receptors. Products of in-
features of hormone receptors, but that lack known
termediary metabolism act as transcriptional regulators in
ligands. Known as orphan receptors, these proteins
bacteria and yeast, suggesting that they may serve similar
represent targets for novel signaling molecules. We
functions in higher organisms (Tomkins, 1975; O'Malley,
have isolated a mammalian orphan receptor that forms
1989). A crucial biosynthetic pathway in higher eukaryotes
a heterodimeric complex with the retinoid X receptor.
is the mevatonate pathway, which leads to the synthesis
A screen of candidate ligands identified farnesol and
of cholesterol, bile acids, porphyrin, dolichol, ubiquinone,
related metabolites as effective activators of this com-
carotenoids, retinoids, vitamin D, steroid hormones, and
plex. Farnesol metabolites are generated intracellu-
farnesylated proteins. The farnesol derivative, farnesyl py-
larly and are required for the synthesis of cholesterol,
rophosphate, is a key metabolic intermediate, since it is
bile acids, steroids, retinoids, and farnesylated pro-
the last precursor common to all branches of the mevalo-
teins. Intermediary metabolites have been recognized
nate pathway (Goldstein and Brown, 1990). We describe
as transcriptional regulators in bacteria and yeast. Our
an orphan nuclear receptor named farnesoid X-activated
results now suggest that metabolite-controlled intra-
receptor (FXR) that is activated by farnesol and related
cellular signaling systems are utilized by higher or-
molecules. FXR provides an example of a vertebrate tran-
ganisms.
scription factor that is regulated by an intracellular metabo-
*PresentAddress:Departmentof Medicine,TheJohnsHopkinsHospi- lite and suggests the existence of a potentially novel verte-
tal, Baltimore,Maryland21287. brate-signaling pathway.
Cell
688

A A

4 I~eG~YPr~rHis5euG~nA~ThrAspG~uPheA~aL~uS~rG~uAsn5euPheG~yVa~uThrG~uH~sA~aA~aG~yPr~Leu

iF
34G1yG1nAsnLeuAapLeuG1uSerTYrSerPr~TYrAs~AsnVa~G1nPhe~r~G1nVa1G1nPr~G~nI1eSer~erSerSerTyrTyr VP-RXR
64Se~Asn~euG~yPheTyr~r~nG~r~G1~spTrp~rSe~pr~G~yLeu~rG1u5e~rgArgMetPr~ThrG~uSerVa~Tyr
94G~n~G~uThrG~u~a~er~et~r~Va~Thr~ysLysPr~Ar~MetA~aA~aSer~erA~aG~yArgI~e~s~yAspG~uL~u
I VP-RXR
184 • -- =5e~laGluCysLeuLeu~rGluIlsGln~sLysSerLysArgLe~rgLysAsnValLysCl~sAla VP-RXR
214AspG~n~rVa~AsnG~uAspSerG~uG~yAr~A~p~e~gG~Va~S~rThrThrLys~eu~sAr~G1ULysT~rG~u5eu~hr vP-PPAR
244ValAspGlnGln~rLeeuAsp~rll~tAspSe~rSerLysGl~rg~tProGln~lulle~rAsnLysIleS~uLysGlu~
274G~PheSerA1aG~uG~uA~n~heLeuI~eL~uThrG~e~A~aThrSerHi~Va1G~nI~e~euVa~G~uPheThr~y~Ar~eu~r~ VP-VDR
304G~yPheG~n~rLeuAspHi~G~spG~n~eA~a~eu~uLy~G~y~erA~aVa~G1~et~heL~uArgSe~A~aG~uZ~ePhe VP-T3R
534AsnL~sLYsLeu~r~A~aG~yHisA~s~LeuLeuG~uG~uArgI~eArgLysSerG~yI~eSerAspG~u~rI~eThr~r~MetPhe
364SerPh~rLysSerVa~G~yG~Leu~y~MetThrG~nG~G~uTyrA~LeuLeuThrA~aI~eVa~I~e~uSerPrcAspAr~G~n
394~yr~Ly~AapArgG~A~a~u5y~euG~nG~Pr~LeuLeuAspVa~Le~G~LysLe~sL~I~e~rG1n~r~G~n 10 20 30 40 50 60 70 80 90 100

424~euLeuGlyArgLe~Th~GluLeuArgThrPheAsnNisHisHlaAlaGluMetLeuMetSerTr~ArgVal Reporter Activity

454[A~H~Lys~hr~roLeuLeuCysGI~IIeTrpABpValGI~
EcRE EcRE m IR1 IR1 m •

1 124 i * 189 250 , D 469 --H~-- x--HEP-~( ~

FXR 0C IF r r II ¢r II rrl
X

+ + + +

ii . . . . "---- 1~7~ EcR

264 32~ 427 653 - LL CE LL -- LL rr LL -- LL ¢r LL -- LL 17" LL

Figure 1. FXR Is a Member of the Nuclear Receptor Superfamily

(A) The nucleotide and amino acid sequence of rat FXR. The 2.1 kb
cDNA encodes a 469 amino acid open reading frame (ORF). The ORF
begins with an initiation codon that appears eight codons downstream
of an in-frame stop codon and ends with a TGA stop codon (asterisk).
In vitro translation of FXR-derived RNA results in a protein with a
relative molecular mass (M,) of 54,000, close to the predicted Mr of
53,934 (data not shown). Amino acids comprising the conserved DBD
(Cys-124-Met-189) are in white letters on black background. The LBD
(Leu-250-GIn-469) is boxed. The FXRal cDNA contains a short inter-
spersed repetitive DNA element (Sutcliffe et al., 1984) in the 3' untrans- Figure 2. FXR and RXR Interact in Cells and In Vitro
lated region followed by a polyadenylation signal. (A) The LBDs of FXR and RXR(I interact specifically in cells. CV-1
(B) Amino acid sequence comparison between rat FXR and Drosophila cells were transiently transfected, as indicated, with CMV promoter-
EcR. Similarity between the DBDs and LBDs is schematically repre- driven expression vectors (100 ng per 10~ cells) containing the yeast
sented as percent amino acid identity. Amino acid regions comprising GAL4 DBD alone (GAL411-147]), GAL4 linked to the FXR LBD (GAL4-
each domain are numbered accordingly. FXR[181-469]), the 78 amino acid herpesvirus VP16 transactivation
domain alone (data not shown), or the VP16 domain linked to the
amino-terminal end of the LBDs for human RXR(~(VP-RXR[203-462]),
Results and Discussion mouse PPARa (VP-PPAR[155-468]), human vitamin D receptor
(VDR) (VP-VDR[92-427]), human T3R~(VP-T3R[173-456]), or human
RAR~ (VP-RAR[156-462]). All cells were cotransfected with a thymi-
Receptor Gene Isolation dine kinase (TK)-Iuciferase reporter construct (300 ng per 105 cells)
A degenerate oligonucleotide probe corresponding to the containing four copies of the yeast GAL4 upstream activating se-
highly conserved P b o x / D N A recogn!tion helix (TCEGCK quence and a CMV-driven 13-galactosidaseexpression vector (500 ng/
(GN)FF) of the nuclear receptor superfamily DNA-binding 105 cells) as internal control. The only combination resulting in signifi-
cant activation was GAL4-FXR plus VP-RXR. As expected from previ-
d o m a i n (DBD) was used to isolate a cDNA for FXR from ous in vitro studies (Yu et al., 1991; Hallenbeck et al., 1992; Kliewer
rat liver (Figure 1A). A similar cDNA has recently been et al., 1992; Leid et al., 1992b; Marks et al., 1992; Zhang et al., 1992),
cloned from m o u s e liver (Seol et al., 1995). Examination VP-PPAR, VP-VDR, VP-T3R, and VP-RAR interacted productively
of the amino acid sequence of FXR reveals that it is a with GAL4-RXR (K. Umesono, personal communication; R. M. E, un-
published data), indicating that these VP16 chimeras are functionally
m e m b e r of the nuclear receptor superfamily (Evans, 1988)
expressed.
and is most closely related to the insect EcR (Koelle et (B) FXR and RXR bind cooperativelyto DNA. Mobility shift experiments
al., 1991). FXR and EcR share 8 1 % a m i n o acid identity were performed using the indicated 32P-labeledDNA probes. Each
within their DBDs (Figure 1B). The carboxy-terminal li- probe is flanked by Hindlll overhangs. Their sequences are as follows
(Yao et al., 1992): hsp27-EcRE, GGTTCA A TGCACT; EcRE= (11 N-
gand-binding domain (LBD) is a c o m p l e x region encoding
hsp27-EcRE), C_GI-FCA A TGCACA__; IR-1, AGGTCA A TGACCT;
subdomains for ligand binding, dimerization, and tran- IR-lm, AGAACA A TGTTCT. The position of mutations introduced into
scriptional activation (Forman and Samuels, 1990; Glass, the hsp27-EcRE and the idealized IR-1 are underlined above and
1994). FXR and EcR LBDs exhibit 3 4 % sequence identity indicated in the figure (X). The gel was autoradiographed for 1.5 hr
(59% similarity) throughout the LBD (Figure 1B). Closer with an intensifying screen.
inspection reveals that similarity is highest within the hep-
tad repeats required for receptor dimerization (Forman
and Samuels, 1990). dimer (usp is a Drosophila homolog of RXR) (Thomas et
al., 1993; Y a o et al., 1993). This prompted us to e x a m i n e
FXR-RXR Complexes Bind DNA whether FXR also interacts with RXR. To do so, we m a d e
Previous studies have shown that the physiologically ac- use of a two-hybrid system modified for use in m a m m a l i a n
tive form of EcR is an EcR-ultraspiracle (EcR-usp) hetero- cells (Nagpal et al., 1993). The LBD/dimerization d o m a i n
A SignalingPathwayfor FarnesolMetabolites
689

= q,,rmone
J
of FXR was fused to the yeast GAL4 DBD (GAL4-FXR). 12-

As seen in Figure 2A, the GAL4 DBD or the GAL4-FXR A 10

chimera is incapable of stimulating transcription from a
reporter construct containing the GAL4 upstream activat- 8¸
ing sequence. Similarly, a fusion protein containing the >=
herpesvirus VP16 transactivation domain linked to the
n-
RXRa LBD (VP-RXR) is inactive when expressed alone
or with the GAL4 DBD (Figure 2A). However, when GAL4- 2.
FXR and VP-RXR are coexpressed, the reporter is acti-
vated 500-fold, indicating that FXR and RXRa interact ef- FXR RXR FXR+RXR
ficiently in cells (Figure 2A). No interaction could be
detected between FXR and other nuclear hormone recep-
100' ~'-~
tors, indicating that FXR and RXR~ associate in a highly NO Hormone
JH III ~.
specific fashion. 80' Spec tic L gand
The amino acids comprising the DNA recognition helix
(P box, Cys-141-Lys-145) of FXR and EcR are identical,
suggesting that the FXR-RXR~ complex could recognize
40.
an ecdysone response element (EcRE): Electrophoretic
mobility shift analysis was performed using ~P-labeled
DNA and in vitro translated FXR and RXRm As seen in
20 | i
Figure 2B, neither FXR nor RXR~ was capable of high
EcR+USP RXR T3R RAR
affinity binding to the hsp27-EcRE. However, when
mixed, the two proteins bind cooperatively. Binding is spe-

,i
cific, as indicated by the inability of the FXR-RXR complex C 300 -

I
to recognize the mutated EcRE (EcREm). The hsp27- 250 - i NO Hormone
JH Ill I
EcRE consists of two imperfect core-binding sites ar- LG69
200 JH III + LG69
ranged as inverted repeats separated by 1 nt (IR-1). Ac-
cordingly, the binding of FXR-RXR was further examined 150
on an idealized IR-1 containing two consensus half-sites.
loo
As seen in Figure 2B, the FXR-RXR complex bound coop-
eratively to the idealized IR-1, but not to a mutant IR-1 50'
containing substitutions within the half-sites (IR-lm).
FXR RXR FXR+RXR
Activation by Farnesoids
Figure3. HormonallyControlledActivity of FXR-RXR
We sought to determine whether FXR possessed tran-
(A)JH III activatesFXR-RXRheterodimers.CV-1cellsweretransiently
scriptional activity that could be hormonally controlled.
transfected with hsp27-EcREx 5 ,'~MTV-luciferase(1000 ng per 105
Based on the identification of an EcRE as a DNA target, cells)alone(minussign)or with expressionvectorsfor rat FXR,human
a reporter plasmid was constructed containing five copies RXRa, or both (50 ng per 10s cells). Reporteractivity was assayed
of the hsp27 response element linked to a truncated aftertreatingcellswith or without50 p.MJH Ill (cis- 10,11-epoxy-3,7,11-
mouse mammary tumor virus promoter (Yao et al., 1993). trimethyl-trans-trans-2,6-dodecadienoic acidmethylester).JH III failed
to activateFXR-RXR complexesusing the parental AMTV reporter
This reporter was cotransfected into CV-1 cells alone or construct that lackedthe EcREs (data not shown).
with expression vectors for FXR, RXR~, or both. Cotrans- (B) JH III fails to activateother nuclear receptors.The activityof the
fected cells were treated with a variety of potential ligands followingreceptor/luciferasereporterpairs was assayedin the pres-
and monitored for changes in luciferase activity. Candi- enceof 50 pM JH III or the indicatedreceptor-specificligand:Drosoph-
date ligands or activators (or both) included a variety of ila G-EcR-usp/hsp27-EcREx 5 ~MTV (1000ng per 105cells), human
RXR~dCRBPII-TK, human T3RI~/TREpx2-TK,and human RARe./
low molecular weight molecules that have been reported DR5x 2-TK. TK-reporterconstructs (300 ng per 105 cells) and CMV-
to have specific biological activity. Unexpectedly, juvenile driven receptorexpressionvectors (100 ng per 10s cells)were added
hormone III (JH III) (Moore, 1990) (50 I~M) elicited a strong as indicated.
induction (10-fold) of luciferase activity in cells expressing (C) FXR-RXR is synergisticallyactivatedby JH III and LG69. CV-1
cells weretransientlytransfectedas in (A), but treatedwithoutor with
both FXR and RXR (Figure 3A). Other potential ligands, 50 I~MJH III, 100 nM LG69, and JH III plus LG69. Datapointswere
including steroids, retroretinoids, eicosanoids, and bile normalizedfor differences in transfection efficiency and plotted as
acids, had no effect. JH III appears to be specific for the relative activity where the untreatedreporteris defined to have an
FXR-RXR~ complex, since it failed to activate EcR-usp, activity of 1 U.
RXR, T3R, or RAR (Figure 3B).
Although JH III activates FXR-RXR, it fails to activate
either FXR or RXR alone (Figures 3A and 3B). This is subsequent transcriptional activation require coexpres-
reminiscent of the Drosophila EcR, which requires RXR/ sion of EcR with RXR/usp. Thus, while the EcR-RXR/
usp for transcriptional activity (Yao et al., 1992, 1993; usp heterodimer is the physiologically active complex, the
Thomas et al., 1993). EcR itself binds ecdysteroids with ability to respond to ecdysone is determined by the EcR
low affinity (Yao et al., 1993); high affinity binding and component of the complex. Since the EcR-RXR hetero-
Cell
690

A ~.~ ~ OH phate, a key precursor in the mevalonate biosynthetic

pathway (Goldstein and Brown, 1990). Accordingly, we
decided to test whether metabolites derived from the mev-
alonate pathway in mammalian cells could also serve as
activators of the FXR-RXRa complex. Indeed, farnesol
(trans-trans or mixed isomers, 50 ~.M) served as a strong
activator of FXR-RXRa (Figure 4A), whereas other fame-
juvenile hormone III
soids such as farnesal, farnesyl acetate, and geranylgera-
geranylgeraniol niol possessed weaker activity. In contrast, little or no acti-
squalene vation was seen with 50 p.M concentrations of geraniol,
cholesterol
farnesoic acid, squalene (Figure 4A), methoprene, meva-
Ionate, squalene epoxide, squalene dioxide, lanosterol,
1 3 5 7 11
24,25-epoxycholesterol, pregnenalone, dehydroepian-
Fold Activation
drosterone, bile acids, or 10 I~M 25-hydroxycholesterol
(data not shown). Mevalonate (200 ~M) displayed weak
activity, provided cells were cotransfected with a mevalo-
20'
nate transporter protein (Kim et al., 1992; data not shown).
• JH III Since FXR-RXR is not activated by potent JH analogs
15' I--*- F. . . . . . I I f (methoprene; data not shown) and since the JH receptor
has not been identified, it remains unclear what relation-
ship, if any, exists between FXR and the putative insect
receptor for JH.
5"
Activation of classical nuclear receptors occurs at physi-
ological concentrations of circulating hormones. To be rel-
evant, activation of FXR should occur at physiologic fame-
; ;o&~;2;a?a;4'o&~; soid concentrations. The levels of endogenous farnesoids
concentration(gM) have been difficult to determine, owing to their rapid me-
Figure 4. Profilesof FXR-RXR Activity Using Various Isoprenoids tabolism and potential sequestration by intracellular and
(A) Activationof FXR-RXRby variousfarnesolmetabolites.CV-1 cells extracellular binding proteins. However, intracellular con-
were transientlytransfected as in Figure 3A with expressionvectors centrations may be inferred from the Michaelis constant of
for rat FXRand human RXRmCellsweretreatedwith 50 I~Mconcentra- enzymes that utilize isoprenoid substrates. These values
tions of each compound. Similar results were obtained with all-trans have been reported to be in the range of 0.5-10 I~M for
and mixed isomers of farnesol and farnesoic acid. Data are plotted protein farnesyltransferase and isopentenyl pyrophos-
as fold activation relativeto untreated cells. Transfectionswere per-
formed as described in Figure 3. Farnesoicacid was synthesizedas phate isomerase (Rilling and Chayet, 1985; Reiss et al.,
described above (Kulkarniet al., 1987). 1990; Gomez et al., 1993). Furthermore, down-regulation
(B) Dose response profile. The experimentwas performed as in (A) of 3-hydroxy-3-methyl-glutaryl coenzyme A reductase ac-
with the indicatedconcentrationof JH III and farnesol(mixedisomers). tivity by a mevalonate-derived nonsterol occurs when mev-
alonate is added to cells at concentrations in excess of 100
I~M (Brown and Goldstein, 1980; Nakanishi et al., 1988).
dimer is composed of two functional receptors, the com- Together, these reports suggest that physiologic concen-
plex can be activated independently by ecdysteroids or trations of farnesoids are in the micromolar range. Accord-
9-cis RA and can be activated synergistically by both li- ingly, dose response studies were performed using in-
gands (Thomas et al., 1993). The structural and functional creasing concentrations of farnesol and JH III, the two
similarities between EcR and FXR prompted us to exam- most effective FXR-RXR activators (Figure 4B). Activation
ine whether the FXR-RXR complex could also be syner- of FXR-RXR required concentrations of 5-50 t~M, sug-
gistically activated by JH Ill and the RXR-specific ligand gesting that FXR-RXR activity is regulated at appropriate
LG69 (Kurokawa et al., 1994). Using the hsp27-EcRE re- concentrations. Similarly, regulation of PPAR and sterol-
porter (Figure 3C), we activated the FXR-RXR~ complex regulatory element-binding protein 1 (SREBP-1) is known
17-fold by 50 I~M JH Ill, 76-fold by 100 nM LG69, and to require 5-100 I~M doses of fatty acids (Green and Wahli,
212-fold bythe combination of JH III and LG69. This syner- 1994) or cholesterol (Wang et al., 1994), levels that are
gistic activity required coexpression of FXR with RXR~ consistent with the presumed intracellular concentrations
(Figure 3C), RXRI], or RXR~, (data not shown). The ability of these compounds.
of J H III to synergize with saturating doses of LG69 or 9-cis
RA (data not shown) suggests that these two compounds Expression of FXR mRNA in
have distinct targets within the FXR-RXR complex. As Isoprenoidogenic Tissues
LG69 is thought to be an RXR-specific ligand (Kurokawa One expectation of an intracellular metabolic activator is
et al., 1994), these results imply that JH III responsiveness that it would be synthesized in the same tissues in which
is determined by the FXR component of the FXR-RXR its receptor is expressed, Northern blot analysis revealed
complex. restricted expression of a 2,3 kb FXR-specific transcript
JH III is a metabolic derivative of farnesyl pyrophos- in liver and kidney (Figure 5). In situ hybridization and
A Signaling Pathwayfor Farnesol Metabolites
691

FXR Conkol RXR~

A

--28 s

rat
E19.5
--18S

KJ ~

Figure 5. Expressionof FXR mRNA

Nissl FXR Nissl FXR
Northern blot analysisof rat tissues. The source of the poly(A)÷ RNA
is shown above each lane. Muscle refers to skeletal muscle.
rat
E19.5
histochemistry (Figure 6) indicated that FXR expression
is limited to the liver, kidney, and gut of rat embryonic
day 19.5 (E19.5) embryo sections (Figure 6A). Background FXR FXR
levels were seen in other tissues and in experiments using
a control probe (Figure 6B). As expected (Mangelsdorf et adult adult
al., 1992), mRNA for the heterodimerizing partner RXRI3 rat mouse
is also found in the liver, kidney, and gut, as well as other
embryonic tissues (Figure 6C). FXR expression in the gut
is prominent in the intestinal villi (Figures 6D-6G). In the
Figure 6. In Situ Hybridizationand Histochemistry
E19.5 kidney, highest levels of expression occur in the
Antisense cRNA probesfrom the truncated mouse FXR cDNA or full-
renal tubules (Figures 6D-6G). In the adult rat kidney, FXR length mouseRXR~cDNAwere usedas indicated.Control represents
is expressed in areas rich in renal tubules: the medullary a truncated rat glucocorticoid receptor sense cRNA probe. (A)-(C)
rays and medullary stripe (Figure 6H). FXR expression is representserial sagittal sectionsfrom an E19.5 rat embryo.Areas of
also detected in the adrenal cortex of the adult mouse positivehybridizationappearwhite.(D)and (E),respectively,are bright
(Nissl) and dark-field photomicrographsof the boxed region in (C),
(Figure 61). Thus, FXR expression is restricted to the liver,
hybridizedwith a FXR antisenseprobe. Similarly,(F) and (G) are de-
gut, adrenal gland, and kidney: tissues known to have rived from the boxed region in (D). (H) is a section from an adult rat
significant flux through the mevalonate pathway (Edmond (9-weekold) kidney. The control probe revealed near-backgroundhy-
et al., 1976; Righetti et al., 1976; Wiley et al., 1977). bridization (data not shown). (I) contains adult mouse (8-weekold)
adrenal gland and adjacent renal cortex and liver. Abbreviations:L,
liver; G, gut; K, kidney; P, pancreas.Scale bars: (A-C) 4 mm; (D-E)
Metabolic Intermediates as Transcriptional 1 ram; (F-G) 0.3 ram; (H-I) 2 mm. Sections were apposed to Kodak
Activators X-OMAT film for 10 days (A-C and H-J) and then coated with nuclear
Our results identify an orphan nuclear receptor that is acti- emulsion (D-G) and exposedfor 16 weeks.
vated by farnesol derivatives. A question that remains
unanswered is the mechanism by which farnesoids acti-
vate FXR. One possibility is that farnesoid treatment re- lipoprotein receptor gene regulator SREBP-1 is main-
sults in the prenylation of FXR. This seems unlikely, since tained in an inactive form by cholesterol (Wang et al.,
FXR lacks a known prenylation sequence (Moores et al., 1994). As with PPAR and SREBP-1, the FXR ligand, as
1991; Khosravi-Far et al., 1992) and is not prenylated by well as FXR target genes, remains to be identified. None-
an in vitro prenyltransferase system (Khosravi-Far et al., theless, these regulatory systems define a novel paradigm
1992; data not shown). Alternatively, JH III or farnesol (or of metabolite-controlled intracellular signaling in verte-
both) could potentially serve as ligands for FXR. To date, brates (O'Malley, 1989). Such signaling provides a means
we have not been able to demonstrate specific binding of to regulate responses to intracellular metabolites in a cell-
JH III or farnesol (or both) to the FXR-RXR complex. Thus, autonomous fashion. By transducing metabolic cues into
a more likely possibility is that a farnesoid, farnesoid me- genomic responses, FXR, PPAR, and SREBP-1 may pro-
tabolite, or farnesoid-induced metabolite may serve as an vide examples for the metabolic code proposed by Tom-
authentic ligand for FXR. This is conceptually similar to kins (1975).
the activation of RXR by all-trans RA and ultimately
leads to the identification of retinoid X as 9-cis RA (Man- Experimental Procedures
gelsdorf et al., 1990; Heyman et al., 1992; Levin et al.,
cDNA Isolation
1992). A degenerate29-mer consensus oligonucleotide(5'-ACC TGT GAG
The demonstration that FXR-RXR is regulated by fame- GGC TGC AAR GKY FrC TTC AA 3'-) correspondingto the highly
sold derivatives provides an example of gene regulation conserved P box/DNA recognitionhelix (TCEGCK(G/V)FF)of the nu-
by intracellular metabolites. Other examples include the clear receptor superfamily DBD was used to probe a ~.gt11 mouse
hepatomaHepa-lclc7 cDNA library under low stringencyconditions
PPAR, a fatty acid-activated orphan receptor (Gottlicher
(Issemannand Green, 1990).An incomplete850 bp mouseFXR cDNA
et al., 1992) that regulates genes involved in fatty acid clone was obtained. This clone was subsequentlyused to screen a
metabolism (Green and Wahli, 1994), and adipocyte differ- regenerating rat liver cDNA library. The FXRal clone was obtained
entiation (Tontonoz et al., 1994). Similarly, the low density from this screen and sequencedby the dideoxysequencingmethod.
Cell
692

Cell Culture and Tranefection evidence for TR isoform-specific transcriptional regulation in vivo.
CV-1 cells were grown in DMEM supplemented with 10% AGI-X8 Proc. Natl. Acad. Sci. USA 91,439-443.
resin-charcoal-stripped calf bovine serum, 50 U/ml penicillin G, and Brown, M. S., and Goldstein, J. L. (1980). Multivalent feedback regula-
50 lig/ml streptomycin sulfate (DMEM-CBS) at 37°C in 50/oCO2. Cells tion of HMG CoA reductase, a control mechanism coordinating iso-
were plated to 50%-80% confluence 1 day prior to transfection using prenoid synthesis and cell growth. J. Lipid Res. 21,505-517.
phenol red-free DMEM with 10%oresin-charcoal-stripped fetal bovine
Edmond, J., Fogelman, A. M., and Popjak, G. (1976). Mevalonate
serum (DMEM-FBS). Cells were transfected by lipofection using N-[1-
metabolism: role of kidneys. Science 193, 154-156.
(2,3-dioleoyloxy)propyl]-N,N,N-ammonium methyl sulfate (DOTAP)
according to the instructions of the manufacturer (Boehringer Mann- Evans, R. M. (1988). The steroid and thyroid hormone receptor super-
helm). Reporter constructs (300-1000 ng per 10s cells), cytomegalovi- family. Science 240, 889-895.
rus (CMV)-driven receptor (50-100 ng per 105 cells), and IB-galacto- Forman, B. M., and Samuels, H. H. (1990). Interactions among a sub-
sidase expression vectors (500 ng per 105 cells) were added as family of nuclear hormone receptors: the regulatory zipper model. Mol.
indicated. After 2 hr, the liposomes were removed, and cells were Endocrinol. 4, 1293-1301.
treated for 40 hr with phenol red-free DMEM-FBS containing the indi- Glass, C. K. (1994). Differential recognition of target genes by nuclear
cated compounds: 50 p.M JH III, 100 nM muristerone A, 100 nM 9-cis receptor monomers, dimers, and heterodimers. Endocrinol. Rev. 15,
RA, 100 nM "1"3(L-triiodothyronine), 1 llM trans RA (all-trans RA), or 391-407.
100 nM LG69 (4-II-(3,5,5,8,8-pentamethyl-5,6,7,8-tetrahydro-2-na p- Goldstein, J. L., and Brown, M. S. (1990). Regulation of the mevalonate
thyl)-l-ethenyllbenzoic acid). Cells were harvested and assayed for
pathway. Nature 343, 425-430.
luciferase and ~-galactosidase activity. All points were performed in
triplicate and varied by less than 10%. Experiments were repeated Gomez, R., Goodman, L. E., Tripathy, S. K., O'Rourke, E., Manne,
three or more times with similar results. V., and Tamanoi, F. (1993). Purified yeast protein farnesyltransferase
is structurally and functionally similar to its mammalian counterpart.
Biochem. J. 289, 25-31.
Electrophoretic Mobility Shift Assays
Electrophoretic mobility shift assays were performed using proteins Gottlicher, M., Widrnark, E., Li, Q., and Gustafsson, J. A. (1992). Fatty
translated in a rabbit reticulocyte lysate system (TNT; Promega). Pro- acids activate a chimera of the clofibric acid-activated receptor and
teins (1 p.I)were incubated for 20 rain at room temperature with 100,000 the glucocorticoid receptor. Proc. Natl. Acad. Sci. USA 89, 4653-4657.
cpm of Klenow-labeled probes in 10 mM Tris (pH 6.0), 100 rnM KCI, Green, S., and Wahli, W. (1994). Peroxisome proliferator-activated
60/0 glycerol, 0.05% NP-40, 1 rnM DTT, 100 ng/id poly(dl, dC) and then receptors: finding the orphan a home. Mol. Cell Endocrinol. 100, 149-
electrophoresed through a 50/o polyacrylamide gel in 0.5x TBE (45 153.
mM Tris base, 45 mM boric acid, 1 mM EDTA). Hallenbeck, P. L., Marks, M. S., Lippoldt, R. E., Ozato, K., and Niko-
dem, V. M. (1992). Heterodimerization of thyroid hormone (-FH) recep-
RNA Expression Studies tor with H-2RIIBP (RXRI~) enhances DNA binding and TH-dependent
For Northern blot analysis, poly(A)÷ RNA (10 p.g) from the indicated transcriptional activation. Proc. Natl. Acad. Sci. USA 89, 5572-5576.
rat tissues was electrophoresed through a 1%oagarose gel under dena-
Heyman, R. A., Mangelsdorf, D. J., Dyck, J. A., Stein, R. B., Eichele,
turing conditions and transferred to a filter. The filter was hybridized
G., Evans, R. M., and Thaller, C. (1992). 9-cis retinoic acid is a high
to the mouse FXR-truncated cDNA that was ~P-labeled by the random
affinity ligand for the retinoid X receptor. Cell 68, 397-406.
primer method (Mangelsdorf et al., 1992) (5 x 10s cpm/lig). This probe
corresponds to rat FXR sequences spanning amino acids 1-297, Issemann, I., and Green, S. (1990). Activation of a member of the
which encode the amino terminus, the DBD, and a portion of the LBD steroid hormone receptor superfamily by peroxisome proliferators. Na-
of FXR. Hybridization was performed overnight at 65°C in 500 mM ture 347, 645-650.
sodium phosphate (dibasic:monobasic, 7:3), 1 mM ethylenedi- Khosravi-Far, R., Clark, G. J., Abe, K., Cox, A. D., McLain, T., Lutz,
aminetetracetic acid, 1% bovine serum albumin, and 7% SDS. The R. J., Sinensky, M., and Der, C. J. (1992). Ras (CXXX) and Rab (CC/
filter was washed twice in 2 x SSC (1 x SSC is 0.15 M NaCI, 0.015 CXC) prenylation signal sequences are unique and functionally dis-
M sodium citrate) at room temperature and twice in 1 x SSC at 55°C tinct. J. Biol. Chem. 267, 24363-24368.
and then autoradiographed with an intensifying screen at -70°C for Kim, C. M., Goldstein, J. L., and Brown, M. S. (1992). cDNA cloning
5 days. In situ hybridizations were performed as described elsewhere of MEV, a mutant protein that facilitates cellular uptake of mevalonate,
(Bradley et al., 1994). and identification of the point mutation responsible for its gain of func-
tion. J. Biol. Chem. 267, 23113-23121.
Kliewer, S. A., Urnesono, K., Mangelsdorf, D. J., and Evans, R. M.
Acknowledgments (1992). Retinoid X receptor interacts with nuclear receptors in retinoic
acid, thyroid hormone and vitamin D3 signalling. Nature 355, 446-
The authors would like to thank T. Cooke, H. Kobayashi, C. McFadden, 449.
and T. Mears for excellent technical assistance. We thank Dr. K. Umes- Koelle, M. R., Talbot, W. S., Segraves, W. A., Bender, M. T., Cherbas,
ono for providing expression vectors for VP16 fusion proteins. We P., and Hogness, D. S. (1991). The Drosophila EcR gene encodes an
thank Drs. M. Brown, M. Brownstein, D. Dixon, L. Gilbert, J. Goldstein, ecdysone receptor, a new member of the steroid receptor superfamily.
M. McKeown, H. Towle, and S. Young for valuable discussions. We Cell 67, 59-77.
also thank Dr. R. Chavez for providing farnesoic acid; Dr. T. Spencer Kulkarni, Y. S., Niwa, M., Ron, E., and Snider, B. B. (1987). Synthesis
for squalene oxide, squalene dioxide, and epoxycholesterol; Dr. R. of terpenes containing the bicyclo[3.1.1]heptane ring system by the
Heyman for LG69; Dr. L. Gilbert for reagents; M. Kricker for assistance intramolecular [2+2] cycloaddition reaction of vinylketenes with al-
with nucleotide sequence formatting; and M. Herkenharn for photo- kenes. J. Org. Chem. 52, 1568.
graphic advice. R. M. E is an Investigator of the Howard Hughes Medi-
cal Institute (HHMI) at the Salk Institute for Biological Studies. This Kurokawa, R., Direnzo, J., Boehm, M., Sugarman, J., Gloss, B., Rosen-
work was supported by the HHMI, the National Institutes of Health feld, M. G., Heyrnan, R. A., and Glass, C. K. (1994). Regulation of
retinoid signalling by receptor polarity and allosteric control of ligand
(R. M. E., D. J. N., and C. W.), and the Tobacco-Related Disease
Research Program (B. M. F.) binding. Nature 371, 528-531.
Leid, M., Kastner, P., and Chambon, P. (1992a). Multiplicity generates
Received February 27, 1995; revised March 23, 1995. diversity in the retinoic acid signalling pathways. Trends Biochem. Sci.
17, 427-433.
References Leid, M., Kastner, P., Lyons, R., Nakshatri, H., Saunders, M., Zach-
arewski, T., Chert, J.-Y., Staub, A., Gamier, J.-M., Mader, S., and
Bradley, D. J., Towle, H. C., and Young, W. S. (1994). ~ and 13thyroid Chambon, P. (1992b). Purification, cloning, and RXR identity of the
hormone receptor (TR) gene expression during auditory neurogenesis: HeLa cell factor with which RAR or TR heterodimerizes to bind target
A Signaling Pathway for Farnesol Metabolites
693

sequences efficiently. Cell 68, 377-395. Erratum: Cell 71(5). Yu, V. C., Delsert, C., Andersen, B., Holloway, J. M., Devary, O. V.,
Levin, A. A., Sturzenbecker, L. J., Kazmer, S., Bosakoweki, T., Husel- Naar, A. M., Kim, S. Y., Boutin, J. M., Glass, C. K., and Rosenfeld,
ton, C., Allenby, G., Speck, J., Kratzeisen, C., Rosenberger, M., and M. G. (1991). RXRI~: a coregulator that enhances binding of retinoic
Lovey, A. (1992). 9-cis retinoic acid stereoisomer binds and activates acid, thyroid hormone, and vitamin D receptors to their cognate re-
the nuclear receptor RXRa. Nature 355, 359-361. sponse elements. Cell 67, 1251-1266.
Mangelsdorf, D. J., Ong, E. S., Dyck, J. A., and Evans, R. M. (1990). Zhang, X. K., Hoffmann, B., Tran, P. B., Graupner, G., and Pfahl, M.
Nuclear receptor that identifies a novel retinoic acid response pathway. (1992). Retinoid X receptor is an auxiliary protein for thyroid hormone
Nature 345, 224-229. and retinoic acid receptors. Nature 355, 441-446.
Mangelsdorf, D. J., Borgmeyer, U., Heyman, R. A., Zhou, J. Y., Ong,
GenBank Accession Number
E. S., Oro, A. E., Kakizuka, A., and Evans, R. M. (1992). Characteriza-
tion of three RXR genes that mediate the action of 9-cis retinoic acid.
The accession number for the sequence reported in this paper is
Genes Dev. 6, 329-344.
U 18374.
Marks, M. S., Hallenbeck, P. L., Nagata, T., Segars, J. H., Appella,
E., Nikodem, V. M., and Ozato, K. (1992). H-2RIIBP (RXR~) heterodim-
erization provides a mechanism for combinatorial diversity in the regu-
lation of retinoic acid and thyroid hormone responsive genes. EMBO
J. 11, 1419-1435.
Moore, D. D. (1990). Diversity and unity in the nuclear hormone recep-
tors: a terpenoid receptor superfamily. New Biol. 2, 100-105.
Moores, S. L., Schaber, M. D., Mosser, S. D., Rands, E., O'Hara, M. B.,
Garsky, V. M., Marshall, M. S., Pompliano, D. L., and Gibbs, J. B.
(1991 ). Sequence dependence of protein isoprenylation. J. Biol. Chem.
265, 14603-14610.
Nagpal, S., Friant, S., Nakshatri, H., and Chambon, P. (1993). RARs
and RXRs: evidence for two autonomous transactivation functions
(AF-1 and AF-2) and heterodimerization in vivo. EMBO J. 12, 2349-
2360.
Nakanishi, M., Goldstein, J. L., and Brown, M. S. (1988). Multivalent
control of 3-hydroxy-3-methylglutaryl coenzyme A reductase: mevalo-
nate-derived product inhibits translation of mRNA and accelerates
degradation of enzyme. J. Biol. Chem. 263, 8929-8937.
O'Malley, B. W. (1989). Did eucaryotic steroid receptors evolve from
intracrine gene regulators? Endocrinology 125, 1119-1120.
Reiss, Y., Goldstein, J. L., Seabra, M. C., Casey, P. J., and Brown,
M. S. (1990). Inhibition of purified p21 'as farnesyl:protein transferase
by Cys-AAX tetrapeptides. Cell 62, 81-88.
Righetti, M., Wiley, M H., Murrill, P. A., and Siperstein, M. D. (1976).
The in vitro metabolism of mevalonate by sterol and non-sterol path-
ways. J. Biol. Chem. 251, 2716-2721.
Rilling, H. C., and Chayet, L. T. (1985). Biosynthesis of cholesterol.
In Sterols and Bile Acids, H. Danielsson and J. Sjovall, eds. (New
York: Elsevier Science), pp. 1-39.
Seol, W., Choi, H. S., and Moore, D. D. (1995). Isolation of proteins
that interact specifically with the retinoid X receptor: two novel orphan
receptors. Mol. Endocrinol. 9, 72-85.
Sutcliffe, J. G., Milner, R. J., Gottesfeld, J. M., and Reynolds, W.
(1984). Control of neuronal gene expression. Science 225,1308-1315.
Thomas, H. E., Stunnenberg, H. G., and Stewart, A. F. (1993). Hetero-
dimerization of the Drosophila ecdysone receptor with retinoid X recep-
tor and ultraspiracle. Nature 362, 471-475.
Tomkins, G. M. (1975). The metabolic code. Science 189, 760-763.
Tontonoz, P., Hu, E., and Spiegelman, B. M. (1994). Stimulation of
adipogenesis in fibroblasts by PPARy2, a lipid-activated transcription
factor. Cell 79, 1147-1156.
Wang, X., Sato, R., Brown, M. S., Hua, X., and Goldstein, J. L. (1994).
SREBP-1, a membrane-bound transcription factor released by sterol-
regulated proteolysis. Cell 77, 53-62.
Wiley, M. H., Howton, M. M., and Siperstein, M. D. (1977). The quantita-
tive role of the kidneys in the in vivo metabolism of mevalonate. J.
Biol. Chem. 252, 548-554.
Yao, T. P., Segraves, W. A., Oro, A. E., McKeown, M., and Evans,
R.M. (1992). Drosophila ultraspiracle modulates ecdysone receptor
function via heterodimer formation. Cell 71, 63-72.
Yao, T. P., Forman, B. M., Jiang, Z., Cherbas, L., Chen, J. D.,
McKeown, M., Cherbas, P., and Evans, R. M. (1993). Functional ecdy-
sone receptor is the product of EcR and ultraspiracle genes. Nature
366, 476-479.