MOLECULAR AND CELLULAR BIOLOGY, May 1993, p. 3113-3121
Copyright © 1993, American Society for Microbiology
Vol. 13, No. 5
The Orphan Receptor Rev-ErbAot Activates Transcription via
a Novel Response Element
HEATHER P. HARDING AND MITCHELL A. LAZAR*
Endocrinology Division, Department ofMedicine, and Department of Genetics, University ofPennsylvania
SchoolofMedicine, Philadelphia, Pennsylvania19104
Received 18 December 1992/Returned for modification 1 February 1993/Accepted 11 February 1993
Rev-ErbAca (Rev-Erb) is a nuclear hormone receptor-related protein encoded on the opposite strand of the
at-thyroid hormone receptor (TR) gene. This unusual genomic arrangement may have a regulatory role, but the
conservation of human and rodent Rev-Erb amino acid sequences suggests that the protein itself has an
important function, potentially as a sequence-specific transcriptional regulator. However, despite its relation-
ship to the TR, Rev-Erb bound poorly to TR binding sites. To determine its DNA-binding specificity in an
unbiased manner, Rev-Erb was synthesized in Escherichia coli, purified, and used to select specific binding sites
from libraries of random double-stranded DNA sequences. We found that Rev-Erb binds to a unique site
consisting of a specific 5-bp A/T-rich sequence adjacent to a TR half-site. Rev-Erb contacts this entire
asymmetric 11-bp sequence, which is the longest nonrepetitive element specifically recognized by a member of
the thyroid/steroid hormone receptor superfamily, and mutations in either the A/T-rich orTR half-site regions
abolished specific binding. The binding specificity of wild-type Rev-Erb was nearly identical to that of C- and
N-terminally truncated forms. This binding was not enhanced by retinoid X receptor, TR, or other nuclear
proteins, none ofwhich formed heterodimers with Rev-Erb. Rev-Erb also appeared to bind to the selected site
as a monomer. Furthermore, Rev-Erb activates transcription through this binding site even in the absence of
exogenous ligand. Thus, Rev-Erb is a transcriptional activator whose properties differ dramatically from those
of classical nuclear hormone receptors, including the TR encoded on the opposite strand of the same genomic
Rev-Erb (also known as earl) is encoded on the noncoding
strand of the a-thyroid hormone receptor (TR) gene, such
that its mRNA is complementary to that of the TR splice
variant, TRac2 (also called c-ErbAa2) (34, 41). This unusual
relationship is likely to serve an important role in the
regulation of gene expression from this locus (31, 42).
Rev-Erb mRNA is expressed in many cell types but is most
abundant in muscle, brown fat, and brain (34). Furthermore,
Rev-Erb protein is highly conserved between rats and hu-
mans (35) and is translated in vivo (18), suggesting an
important function. Indeed, Rev-Erb contains a putative
DNA binding domain (DBD) containing two zinc fingers of
the CyS4 type which are characteristic of steroid and thyroid
hormone receptors (12, 15).
Rev-Erb is an orphan receptor (45) most similar to the
TR/retinoic acid receptor (RAR) subgroup of nuclear recep-
tors (13, 34). These receptors all contain an identical se-
quence in the first zinc finger of the DBD, referred to as the
P box, which specifies binding to response elements contain-
ing the sequence AGGTCA (4, 10, 54), often arranged in
tandem as direct or inverted repeats with variable spacing
(43, 55). The unique actions of peroxisomal proliferators,
vitamin D, thyroid hormone (T3), and retinoic acid are due at
least in part to the ability of their receptors to differentially
recognize direct repeats spaced by 1, 3, 4, and 5 bp,
respectively (26, 55). Furthermore, these receptors bind to
DNA with highest affinity as heterodimers with retinoid X
receptors (RXRs) (8, 24, 25, 36, 39, 60, 61) and COUP-TF (5,
52), two members of the TR/RAR subgroup. Rev-Erb shares
considerable sequence identity in the domains implicated in
receptor homo- and heterodimerization (13, 23, 39, 44, 50).
We considered whether Rev-Erb functions as a ligand-
dependent transcriptional activator, as a constitutive tran-
scriptional activator, or as a heterodimerization partner for
other members of the TR/RAR subgroup. To address these
issues, an iterative, polymerase chain reaction (PCR)-based
DNA binding site selection technique (6, 59) was used to
determine the ideal binding site for Rev-Erb. Remarkably,
the selected Rev-Erb element (RevRE) is composed of a
unique 5-bp A/T-rich sequence adjacent to a single AGG
TCA half-site. Competition experiments confirmed the im-
portance of the entire RevRE for DNA binding, and methy-
lation and acylation interference assays demonstrated that
bases in both the half-site and the A/T flank of the RevRE are
directly contacted by Rev-Erb. Rev-Erb did not form func-
tional heterodimers with TR, RXR, or other nuclear proteins
and appeared to bind to the RevRE as a monomer. However,
Rev-Erb had the ability to activate transcription via the
RevRE site in the absence of added ligand. Therefore
Rev-Erb acts via a mechanism distinct from that of better-
studied nuclear receptors, including the TR which is en-
coded on the opposite strand of the same genomic locus. In
addition, this is to our knowledge the first report demonstrat-
ing that two transcriptional activators are encoded on over-
lapping genes transcribed from opposite strands of the same
MATERUILS AND METHODS
Synthesis of Rev-Erb proteins in bacteria. Glutathione
S-transferase (GST) fusion proteins were produced in Esch-
erichia coli from pGEX-2TK (22). GST-RevT, containing
amino acids 21 to 292 (numbered as in the human cDNA) of
Rev-Erb, was generated by introducing via PCR a stop
codon and StuI and EcoRI restriction sites following nucle-
HARDING AND LAZAR
otide 1058 in the rat Rev-Erb cDNA (34) and subcloning the
BamHI-EcoRI fragment of the PCR product into pGEX-
2TK. GST-Rev-Erb, containing all of the Rev-Erb sequences
except the N-terminal 21 amino acids, was produced by
subcloning the HindIII-NaeI fragment of the rat Rev-ErbAa
cDNA into the HindIII-StuI sites of GST-RevT, which
removed the stop codon at amino acid 292 and substituted
the entire additional coding region of the rat Rev-ErbAax
cDNA, followed by its native stop codon after amino acid
614. Proteins were expressed in E. coli DH5at cells as
previously described for production of TRa (23) and purified
to 50 to 75% homogeneity by glutathione-Sepharose chro-
Synthesis of Rev-Erb in reticulocyte lysate. Full-length
Rev-Erb was synthesized in reticulocyte lysates from a
BamHI-to-NaeI fragment of the human Rev-Erb cDNA
subcloned into pBluescript (pBS), using the TNT-T7 kit
(Promega). Truncation products were made by runoff tran-
scription/translation on cDNA templates restricted with
XhoI (to synthesize Rev-199), EcoRV (to synthesize Rev-
288), and EcoRI (to synthesize Rev-554).
Preparation of DNA fragments. For DNA binding site
selection studies, the bottom strands of N4AGGTCAN12 and
N20 were 5'-AGACGGATCCATTGCAACCTCCCCCN12
GATCTGTAGGAATTCGGA-3', respectively. These oligo-
nucleotides were made double stranded by using Klenow
enzyme and amplified by PCR with primers A and B, which
TGACTCCGAATTCCTACAG-3' and 5'-TCGTAAGCTGA
PCR mixtures contained 10 ng of template, 200 ng (1.75
nM) each of primers A and B, 0.2 mM deoxynucleoside
triphosphates, 2 U of Taq polymerase, and 1x PCR buffer (5
mM KCl, 10 mM Tris [pH 8.4], 5 mM MgCl2, 0.1 mg of
gelatin per ml) in 100 ,ul. Cycling conditions used to generate
the initial double-stranded DNA were 4 cycles of 94°C for 1
min, 52°C for 1 min, and 72°C for 3 min followed by 25 cycles
of 94°C for 15 s, 62°C for 15 s, and 72°C for 15 s. In
subsequent rounds of PCR, a single cycle of 94°C for 1 min,
62°C for 1 min, and 72°C for 3 min was followed by 25 cycles
of 94°C for 15 s, 62°C for 15 s, and 72°C for 15 s. Labeled
probes were obtained in a 50-,ul reaction by using the same
conditions but with a final concentration of 0.5 mM dATP,
dGTP, and d1TP and 30 ,uCi of [32P]dCTP. The reaction
mixture was incubated at 95°C for 1 min, 37°C for 2 min, and
72°C for 3 min for a single cycle, and the probe was
separated from free nucleotides on a Nick column (Pharma-
cia). PCR products were purified on 7.5% nondenaturing
EMSA. For the
(EMSA), the standard 30-p 1 binding reaction mixture con-
tained labeled (30,000 cpm) or unlabeled (30 to 90 ng) DNA
and Rev-Erb (from bacteria or reticulocyte lysate as de-
scribed above) in lx binding buffer plus 200 ,ug of poly(dI-
dC) per ml. After incubation at room temperature for 20 min,
reaction mixtures were loaded on a 7.5% polyacrylamide gel
and the complexes were separated in 0.5 x Tris-borate-
EDTA at room temperature. Gels were dried prior to auto-
radiography. Bacterially expressed TRal and RXR (also
known as H2RIIBP ), which have been described previ-
ously (5), and reticulocyte lysate-synthesized RXRa (38)
were used in the heterodimerization studies.
Selection of binding sites. The selection method of Black-
well et al. (6, 7) was modified as follows. The double-
stranded randomers (30 to 90 ng) were prepared as described
above, incubated with 500 ng of purified GST-RevT, and
subjected to EMSA as previously described (30). Bound
DNA was excised from the portion of the gel between the
free probe and the wells, eluted into buffer (0.5 M ammo-
nium acetate, 10 mM MgCl2, 0.1% sodium dodecyl sulfate, 1
mM EDTA) at 37°C for 8 to 16 h, PCR amplified and gel
purified as described above, and subjected to EMSA with
GST-RevT. This process was repeated six times. Fragments
were further selected by affinity chromatography with GST-
RevT prebound to glutathione-Sepharose (Pharmacia) at 4°C
for 2 h in 10 mM N-2-hydroxyethylpiperazine-N'-2-ethane-
sulfonic acid (HEPES; pH 7.9)-80 mM KCl-1 mM dithio-
threitol-5% glycerol-25 ,ug of denatured herring sperm DNA
per ml and then washed extensively with the same buffer.
Bound protein and DNA were eluted with 50 mM reduced
glutathione. DNA bound to GST-RevT was purified by
phenol-chloroform extraction, PCR amplified, gel purified,
and subjected to a second round of affinity chromatography.
The pool selected from N4AGGTCAN12 was subjected to
one additional round of EMSA. Fragments were subcloned
into XbaI-SphI sites of pBS, PCR amplified with primers A
and B, and subjected to EMSA with GST-RevT. Inserts in
positive transformants were amplified by PCR, gel purified,
labeled by a single round of PCR as described above, and
then studied in the EMSA. Strongly binding inserts from 19
transformants selected from N4AGGTCAN12 and 23 se-
lected from N20 were sequenced by the dideoxy-chain ter-
Methylation and acylation interference assays. The RevRE-
containing insert from a selected clone (ACTCCCAGAAK
TGGGTCAT) was labeled on either strand by using Klenow
enzyme to fill in one 3' overhang (XbaI or HindIlI digested)
prior to digestion with the second enzyme. The gel-purified,
labeled DNA was modified by treatment with dimethyl
sulfate (DMS) or diethylpyrocarbonate (DEPC) as described
by Sturm et al. (51). The modified probes (300,000 cpm) were
incubated with 5 ,ug of GST-RevT or GST-Rev-Erb in a 40-,ul
reaction mixture and subjected to EMSA. The protein-DNA
complexes and free probe were eluted from the wet gel, and
the DNA was cleaved with piperidine (3) and electrophore-
sed on a sequencing gel.
Cell culture, transfection, and CAT assays. The RevRE-
TKCAT reporter plasmid was made by ligating four copies
of the sequence GATCCAGAATGTAGGTCAGGATC (the
Rev-Erb binding site is underlined) into the BamHI site of
pUTKAT-3 (46) in the following orientation:
additional nucleotides at both ends of the RevRE resulted in
separation between AGGTCA half-sites of 13 bp in the
arrangement, 20bp in the +---+arrangement, and 6bpin the
--+- arrangement. JEG-3 choriocarcinoma cells were main-
tained in Dulbecco's modified Eagle's medium (DMEM)
supplemented with 10% bovine calf serum and switched to
DMEM plus 10% anion-exchange resin and charcoal-treated
serum (47) 12 h prior to calcium phosphate transfection (32)
in pairs or triplicate with S ,ug of RevRE-TKCAT or pUT-
KAT-3, along with 5 ,ug of plasmid pCDM (2) unmodified or
expressing rat TRol (33) or human Rev-Erb (35) cDNA. The
fold activation of RevRE-TKCAT varied somewhat for
different preparations of the Rev-Erb expression plasmid; 5
,ug ofP-galactosidaseexpression plasmid pCH110 (Pharma-
cia) was also included to control for transfection efficiency.
After 16 h, cells were treated for 2 min with 20% dimethyl
sulfoxide in serum-free medium and then washed prior to
addition of fresh DMEM plus 10% stripped serum. Cell
MOL. CELL. BIOL.
NOVEL BINDING SITE OF ORPHAN RECEPTOR Rev-ErbAa
extracts were prepared 24 h after dimethyl sulfoxide shock,
normalized for protein concentration, and assayed for chlor-
amphenical acetyltransferase (CAT) activity. Expression of
,-galactosidase varied insignificantly, and Rev-Erb had no
significant effect on transcriptional activity of the enhancer-
less reporter plasmid.
Rev-Erb binds to an asymmetric site composed of an
AlT-rich region 5' to a single AGGTCA half-site. Since the
DBD of Rev-Erb is as similar to those ofTR and RAR as the
latter are to each other (amino acid identities of 49% [33 of
67] in all three, 57% between TR and Rev-Erb, 61% between
RAR and Rev-Erb, and 60% between TR and RAR ), we
initially studied the binding of Rev-Erb to an inverted repeat
of the AGGTCA motif (TREp), which is bound by both TR
and RAR. A GST-RevT fusion protein containing amino
acids 21 to 292 was produced in bacteria for use in these
studies (the highly conserved DBD is from amino acids 132
to 199). However, although TR and RAR bound TREp well
(5), binding of GST-RevT to TREp was barely detectable
(data not shown; see Fig. 5).
To better understand the DNA-binding properties of Rev-
Erb, we devised a PCR-based strategy based on the se-
quence amplification and binding (SAAB) method of Black-
well et al. (6), in which the binding site is selected from a
pool of random oligonucleotides by iterative binding, sepa-
ration of bound from free, and PCR amplification. Hypoth-
esizing that Rev-Erb, like TR, RAR, vitamin D3 receptor,
PPAR (peroxisomal proliferator-activated receptor), and re-
lated receptors, would recognize two AGGTCA half-sites
with a specific orientation and spacing (26, 43, 55), we
initially selected from a biased pool of oligomers containing
four randomly generated base pairs 5' to one AGGTCA
motif followed by 12 random base pairs (N4AGGTCAN12).
Six rounds of EMSA selection using GST-RevT were fol-
lowed by two rounds of selection on GST-RevT immobilized
on beads and one additional round of EMSA selection (Fig.
1A). After the ninth round of selection, there was an
-70-fold enhancement of RevT binding to the selected pool
(Fig. 1B). The selected material was subcloned into pBS,
and inserts from 100 individual clones were assayed for
DNA binding to GST-RevT in the EMSA (data not shown).
Of these, approximately 30% contained inserts which bound
tightly to Rev-Erb, and 19 of these were sequenced. Surpris-
ingly, a second AGGTCA motif did not emerge. Instead, as
shown in Fig. 2A, the consensus sequence (A-[A/T]-N-T)
was nearly invariant 5' to the AGGTCA in strongly binding
fragments, while no preference for specific 3' sequences,
including a second TR half-site, was ascertained.
To confirm the requirement for the AGGTCA motif in an
unbiased manner and to determine whether Rev-Erb had
additional sequence preference 5' to the A/T-rich region
(only four random bases were present 5' to the AGGTCA in
the initial pool of oligonucleotides), the selection was re-
peated with a starting pool of oligomers containing 20
random base pairs (N20). Selection of binding sites was
confirmed in the EMSA (Fig. 1C), and the selected pool of
fragments was subcloned into pBS as described above.
Sequences of 20 fragments which were determined to bind
tightly to Rev-Erb are shown in Fig. 2B. This analysis
proved in an unbiased manner that a single TR half-site was
necessary for Rev-Erb binding, although it was noted that a
G residue can substitute for the first A in the AGGTCA.
Furthermore, these studies confirmed the requirement for
EMSA with GST-RexvT
Elute Bound Oligo
Elute Bound Oligo
Subclone into Plasmids
PCR Amplifv Insert from
EMSA vith GST-RexT
Sequence Strongly-Binding Fragments
GST-RevT(.d): 0.5 1
FIG. 1. Selection of a consensus Rev-Erb binding site. (A)
Strategy. N4AGGTCAN12 and N20 are the two starting pools of
oligonucleotides (N indicates A, C, G, or T at random) which were
amplified by PCR using primers A and B (see Materials and
Methods). (B and C) Enhancement of Rev-Erb DNA binding in
selected pools. After the indicated number of rounds, selected
oligonucleotides were 32p labeled and studied in the EMSA with
-190 and 380 ng of GST-RevT.
the 5' A-(A/T)-N-T sequence and revealed an additional
preference for an A or T as the fifth base upstream from the
AGGTCA. No evidence for a second AGGTCA half-site or
any further upstream or downstream components of the
RevRE emerged from the starting material (N20).
Mutational analysis of the AlT-rich region and AGGTCA
half-site. The relative importance of specific bases within the
RevRE was evaluated with a series of mutant RevREs,
whose ability to compete with a labeled consensus RevRE
for binding to Rev-Erb was studied. The data from these
experiments are summarized in Fig. 3A and shown in Fig.
3B. The middle panel of Fig. 3B shows that mutants 9-llC,
8-1OC, and 7-9C failed to compete for binding to GST-RevT
even at a 625-fold molar excess, confirming the importance
of the AlT-rich sequence. Indeed, point mutations in this
region were as effective in preventing binding as were single
VOL. 13, 1993
3116 HARDING AND LAZAR
(N4) AGGTCA (N12) N20
10 9 8
AA N T AG G T CA
of the Rev-Erb binding
7 6 5 4 3 21
FIG. 2. Consensus
Strongly binding fragments selected from the N4AGGTCAN12 and
N20 oligonucleotides were sequenced, and the frequency of each
nucleotide at the indicated positions is shown. The numbering
system is relative to the AGGTCA. *, the AGGTCA sequence
present in positions 1 to 6 of N4AGGTCAN12.
base substitutions in the AGGTCA motif (compare mutants
A1OC and T9C with mutant AlC or C2G). Within the
A/T-rich region, bases close to the TR half-site were more
important than the extreme 5' nucleotides. Variation of the
nucleotide at position 8 (G8A in Fig. 3A) competed as well as
did the consensus oligonucleotide, as predicted from the
SAAB selection (data not shown).
The binding site selection and mutational analysis de-
scribed above were performed with GST-RevT, which is
truncated at amino acid 292 (of 614). To determine whether
the C terminus of Rev-Erb influenced DNA-binding speci-
ficity, we synthesized GST-Rev-Erb, which contains the
intact C terminus and lacks only the N-terminal 21 amino
acids. Figure 3B (top) shows that the mutant oligonucle-
otides competed for binding to GST-Rev-Erb in a pattern
which was virtually identical to that observed for GST-
RevT. The competition analysis was also repeated with
full-length, wild-type Rev-Erb synthesized in reticulocyte
lysate (RL-Rev-Erb) in order to determine whether the GST
moiety and absence of the N-terminal 21 amino acids influ-
enced the DNA-binding specificity of the GST fusion pro-
teins (Fig. 3B, bottom). Here, too, mutations in the A/T-rich
region and AGGTCA motif profoundly reduced binding
affinity, and RL-Rev-Erb displayed nearly the same binding
preferences as did GST-RevT and GST-Rev-Erb. Thus, the
DNA-binding specificity of Rev-Erb is determined entirely
by amino acids 21 to 292. Note that RL-Rev-Erb formed two
complexes with the consensus RevRE, consistent with the
two major translation products of 70 and 55 kDa which were
seen when the cDNA is translated in vitro (data not shown)
(35). The smaller species could result from initiation at the
second in-frame methionine residue (amino acid 107), as is
used by one form of rat Rev-Erb (34), or could represent a
C-terminal degradation product. In any event, no differences
were observed in the DNA-binding specificities of these two
Rev-Erb contacts both the half-site and the A/T-rich flank of
the selected binding site. The ability of Rev-Erb to contact
1. I - :.3
_ CON 11-13C`C'e
13 12 11
FIG. 3. DNA-binding specificity of Rev-Erb. (A) Consensus
RevRE sequence and mutants used for competition. The consensus
sequence determined from the binding site selection is underlined
within the CON oligonucleotide. Mutants are named by the base
substitution which is underlined and bold for each oligonucleotide.
Competition results shown in panel B are summarized as follows:
++ +, the level of competition of the CON oligonucleotide with
itself; + +, 50% competition between 25- and 125-fold molar excess;
+, 50% competition between 125- and 625-fold molar excess; -, no
competition detected at 625-fold molar excess. (B) Competition for
binding to Rev-Erb. Unlabeled consensus and mutant RevREs were
added in the molar excesses shown to binding reaction mixtures
containing the labeled consensus RevRE and either GST-RevT
(middle), GST-Rev-Erb (top), or RL-Rev-Erb (bottom). Unlabeled
competitors were in 25-, 125-, and 625-fold molar excess, as indi-
cated by the bars above the lanes.
bases within the A/T-rich and (A/G)GGTCA regions of the
RevRE was determined using methylation (DMS) and acy-
lation (DEPC) interference assays, in which protein-DNA
interaction was prevented by modification of purines. DMS
interference detected contact with G residues, while DEPC
interference identified important A and G (A>G) residues.
Analysis ofboth strands of the RevRE by these two methods
revealed that modification of bases within the A/T-rich and
GGGTCA sites completely interfered with binding, with the
exception that acylation of the downstream A in the
GGGTCA motif did not interfere with binding despite the
clear preference during selection (Fig. 4). Furthermore, the
interference patterns for GST-RevT and GST-Rev-Erb were
indistinguishable with both methods. Thus, Rev-Erb con-
tacts a span of 10 to 11 bp, and the DNA binding of a
C-terminal truncation was again nearly identical to that of
Rev-Erb binds to DNA in the absence of other proteins,
most likely as a monomer. We studied the ability of Rev-Erb
to form DNA-binding heterodimers with other nuclear re-
ceptors, including RXR, which heterodimerizes with TR and
RAR to stabilize their interaction with DNA. Figure SA
shows that RXR and TR did not bind to the RevRE.
MOL. CELL. BIOL.
..+, + -f
NOVEL BINDING SITE OF ORPHAN RECEPTOR Rev-ErbAa
FIG. 4. DNA contact by Rev-Erb determined by methylation (DMS) and acylation (DEPC) interference analysis. Results are shown with
both lower and upper strands of a selected fragment, using both GST-RevT and GST-Rev-Erb. DMS treatment modifies G residues; DEPC
cleaves atA and G residues (A>G [18, 19]). Arrows point to bases whose modification interferes withbinding. Lightlyshaded arrow at bottom
indicates weak contact of the G at position 8. Cleavage patterns of free (F) and bound (B) DNAs are shown.
Furthermore, although Rev-Erb is homologous to TR and
RAR in C-terminal regions involved in heterodimerization
(23, 39, 44, 50), Rev-Erb did not bind to the RevRE as a
heterodimer with RXR, with TR, or with proteins in liver
nuclear extract which have been shown to heterodimerize
with TR and RAR. In contrast, TR bound to TREp as
monomer, homodimer, and heterodimer with RXRa and
as previously demonstrated (8, 24, 25, 36, 39, 60, 61).
However, Rev-Erb did not bind to TREp as a heterodimer
with TR or with RXR and bound poorly to direct repeats of
the AGGTCA motif spaced 3, 4, and 5 bp apart (the
preferred response elements of vitamin D receptor, TR, and
RAR, respectively) unless the spacer sequences provided
the A/T-rich region of the RevRE (18). Thus, despite the
sequence conservation between Rev-Erb, TR, and RAR,
Rev-Erb has unique DNA-binding characteristics, which
suggests a distinct mode of action and potential target genes.
To better understand which regions of the Rev-Erb protein
were required for DNA binding, full-length and truncated
forms of the protein were synthesized in reticulocyte lysate.
As discussed earlier, in vitro translation of the wild-type
Rev-Erb cDNA resulted in two major protein species, and
two complexes were seen in the EMSA (see also Fig. 3B and
Fig. SA and B), the more greatly retarded of which was
presumably full-length Rev-Erb (Fig. 5C). Figure SC shows
that RevT-288 (truncated after amino acid 288) bound to the
RevRE, which was not surprising since this protein is nearly
identical to the Rev-Erb portion of GST-RevT. However,
DNA binding was eliminated by truncation immediately
after the second zinc finger (RevT-199), which removes
amino acids distal to the P and D boxes, a region recently
shown to be important for the DNA-binding specificity of
two nuclear receptor superfamily members, NGFI-B (also
known as nur77 [19, 40]) and RXRI
between 21 and 288, but not the zinc fingers alone, is
necessary and sufficient for DNA binding. Removal of the
last 60 amino acids from the C terminus of Rev-Erb (Rev-
554) also eliminated DNA binding, suggesting that the C
terminus of Rev-Erb is required to expose this domain in the
The truncated and full-length Rev-Erb formed complexes
with very different mobilities, as expected from their molec-
ular weights. RevT-288 and full-length Rev-Erb were mixed
prior to EMSA in order to determine whether these com-
plexes contained Rev-Erb monomers or homodimers; obser-
vation of an intermediately migrating complex would be
strong evidence for dimer formation (27, 53). In contrast,
even though both RevT-288 and full-length Rev-Erb contain
the D box which is involved in homodimerization of other
nuclear receptors (17, 20, 37), Fig. SC shows that no inter-
mediate complexes were observed when the translation
products were mixed, suggesting that each species bound as
a monomer. This experiment does not rule out the possibility
that sequences deleted from RevT-288 are important for
(58). Thus, a region
VOL. 13, 1993
HARDING AND LAZAR
Rev-ErbA: - + - +- +- + - +
TR IRXRCx RXRB Liver
+ + + +
nuclear extract was incubated with the RevRE in the presence or absence of RL-Rev-Erb. The first two lanes contained the RevRE with or
without RL-Rev-Erb. (B) Rev-Erb and TR were incubated with TREp separately, together, or with RXRa or -a. The TR monomer (TR-M),
homodimer (TR-D), and heterodimers (TR/RXR) are indicated. (C) C-terminal deletions of Rev-Erb. Human Rev-Erb (RL-Rev-Erb) contains
614 amino acids. C-terminal deletions (RevT-554, RevT-288, and RevT-199) contain the indicated number of amino acids due to digestion of
the cDNA with EcoRI, EcoRV, and XhoI, respectively, prior to transcription/translation. The apparent confluence of the signal from the free
probe is due to greater exposure time of this autoradiograph.
5. DNA-binding properties of Rev-Erb. (A and B) DNA binding in the presence of other proteins. (A) TRal, RXRa, RXR3, or liver
homodimerization. However, the bindings of RevT-288 and
wild-type Rev-Erb were approximately equal, and lack of
Rev-Erb homodimer formation would explain why the larger
and smaller Rev-Erb translation products formed two dis-
crete complexes and not a third, intermediately migrating
complex containing a heterodimer of the two translation
Rev-Erb activates transcription via the RevRE. To examine
the ability of Rev-Erb to regulate gene transcription, four
copies of the RevRE were inserted upstream to the thymi-
dine kinase promoter driving expression of CAT. Separation
between direct repeats of the AGGTCA half-sites was 13 bp,
and separation between inverted repeats was 6 bp, as a result
of cloning sites in the RevRE-containing insert used to
generate this reporter (see Materials and Methods). TRal
had no effect on expression from the RevRE-containing
reporter (Fig. 6), although it did mediate strong, T3-depen-
dent activation with a T3 response element substituted for
PNd szRi xC)
FIG. 6. Evidence that Rev-Erb activates transcription from the
Rev-RE. JEG-3 cells were cotransfected with a CAT expression
plasmid driven by the minimal thymidine kinase promoter, contain-
ing four copies of the RevRE upstream, along with a TRax1 or
Rev-Erb expression plasmid. After transfection, cells were incu-
bated for 24 h with DMEM plus 10% stripped calf serum, supple-
mented with 10 nM T3 and 10% calf serum where indicated.
the RevRE in the reporter plasmid (32) (data not shown).
Remarkably, however, a Rev-Erb-expressing plasmid stim-
ulated transcription of the RevRE-containing reporter gene
12- to 15-fold (Fig. 6). T3 (which has been suggested to be a
ligand for Rev-Erb , although we have not observed
specific binding ) and serum had no effect on transcrip-
tional activation by Rev-Erb. These data indicate that Rev-
Erb is a constitutive transcriptional activator or is activated
by a ligand present in stripped serum or in the JEG-3 cells
We have shown that Rev-Erb is a sequence-specific DNA-
binding protein and transcriptional activator. Rev-Erb and
TRot are thus the first examples of sequence-specific DNA-
binding proteins encoded on opposite strands of the same
eukaryotic genomic locus. Rev-Erb binds specifically to an
asymmetric 11-bp sequence, the longest nonrepetitive ele-
ment specifically recognized by a member of the thyroid/
steroid hormone receptor superfamily, and binds poorly to
tandemly arranged half-sites such as TREp. Rev-Erb also
binds poorly to direct repeats, unless the 5' or spacer
sequences provide the specific A/T-rich portion of the
RevRE (18) (suggesting a mechanism by which a subset of
hormone-responsive target genes could also respond to
Rev-Erb). The DNA-binding properties of Rev-Erb have
diverged from those of the TR and RAR despite the genomic
relationship between Rev-Erb and the TR which previously
led us to speculate that they arose by gene duplication and
inversion (34). These observations are consistent with recent
analyses of the primary amino acid sequences of all steroid/
thyroid hormone receptor superfamily members, which sug-
gest that Rev-Erb diverged from TRotl, TR,B, and RAR
before these diverged from each other (1, 29). The findings
that Rev-Erb seems to bind DNA as a monomer and does not
bind DNA as a heterodimer with RXR, TR, or TRIRAR
heterodimerization partners present in liver nuclear extract
are also interesting in this context, given the homology of
Rev-Erb in portions of the C terminus felt to be importantfor
MOL. CELL. BIOL.
NOVEL BINDING SITE OF ORPHAN RECEPTOR Rev-ErbAa
homo- and heterodimerization of TR and RAR (13, 45). Our
experiments have not ruled out the possibility that Rev-Erb
can form heterodimers with as yet undiscovered factors or
that Rev-Erb homo- or heterodimers bind to a site greater
than 20 bp in length, which would not have been detected by
our analyses. It is also formally possible that the bacterial
Rev-Erb and reticulocyte lysate-derived Rev-Erb do not
heterodimerize because of abnormal posttranslational mod-
ification of the proteins.
Rev-Erb's ability to activate transcription in the absence
of exogenous ligand
hormone receptors. It is important to note that the transac-
tivation by Rev-Erb was less than 10% of that which we
routinely observe for the T3-stimulated TR (using TREp or a
direct AGGTCA repeat with a 4-bp gap as the TRE), and it
is possible that the activity of Rev-Erb can be enhanced by
an as yet unknown ligand. Indeed, we do not know whether
transcriptional regulation by Rev-Erb requires an intracellu-
lar ligand that is present in JEG-3 cells or, alternatively,
reflects a truly autonomous function. Nonetheless, ligand-
independent transactivation has recently been demonstrated
for other orphan receptors (all with the identical P box as in
Rev-Erb and TR), including COUP-TF (21), HNF-4 (49), and
NGFI-B (11, 57). However, COUP-TF (28, 56) and HNF-4
(49) bind DNA as homodimers, with the COUP-TF ho-
modimer binding tightly to TREp (5) and to variably spaced
direct repeats of the AGGTCA half-site
COUP-TF forms DNA-binding heterodimers with TR and
RAR (5, 52) as well as RXR (24).
PCR-based, unbiased screening methods, referred to as
SAAB (6, 7) or cyclic amplification and selection of targets
(CASTing) (59), have been previously used to identify the
DNA-binding sites of transcription factors such as c-Myc (6,
7), myogenin (59), and p53 (14). The present report describes
the first successful application of a similar strategy to a
member of the nuclear receptor family of transcription
genetic selection in Saccharomyces cerevisiae which was
used to uncover a rat genomic DNA sequence specifically
bound by the orphan receptor NGFI-B (57). Both methods
are unbiased, but the yeast selection method has the advan-
tage of identifying functional binding sites in endogenous
genes, although it does not prove that the selected genomic
sequences actually function as natural response elements.
Also, if heterodimeric interactions can modulate DNA bind-
ing, the yeast system might yield different information than a
biochemical strategy using purified proteins. However, there
are also several advantages to the selection method which
we have used. While it does not address the possibility of
physiologically relevant heterodimerization partners, screen-
ing with purified protein in vitro ensures that the selected
binding sites are recognized by Rev-Erb in the absence of
other nuclear proteins, such as would be present in yeast
cells. Furthermore, the PCR-based method provides a large
number of independent binding sites which allow generation
of a consensus sequence. In contrast, the yeast genetic
selection method revealed only two NGFI-B binding sites,
both contained within a single genomic fragment.
Although the NGFI-B binding site (two A residues 5' to an
AGGTCA half-site ) competed only weakly for binding
to Rev-Erb (18), there are remarkable similarities between
the DNA-binding properties of Rev-Erb and NGFI-B. Both
bind to asymmetric elements containing a single TR half-site,
in contrast to the direct repeat motifs recognized by other
nuclear receptor homodimers and heterodimers. The DNA-
binding specificity of both Rev-Erb and NGFI-B is depen-
is unusual for steroid and thyroid
It is interesting to compare this method with
dent upon amino acid residues distal to the P and D boxes;
the fact that there is little similarity in the actual amino acid
residues in this region of the two proteins probably accounts
for their different DNA-binding specificities. Additionally, in
an assay similar to that employed here, NGFI-B was deter-
mined to bind DNA as a monomer (57), although its het-
erodimerization properties have not been studied. Further-
more, both Rev-Erb and NGFI-B can activate transcription
in the absence of exogenous ligand. These similarities sug-
gest that Rev-Erb and NGFI-B are prototypes of a subfamily
of orphan receptors capable of activating transcription from
complex elements composed of distinct sequences 5' to a
single TR half-site in the absence of endocrine ligand.
Another likely member of this subfamily is the ecdysone-
inducible E75 protein, which is involved in the metamorpho-
sis of Drosophila melanogaster (48), since it is 79% identical
to Rev-Erb in the DBD (with considerable similarity in the
adjacent region implicated in DNA-binding specificity). Fur-
ther study of the expression and properties of these and
additional members of this intriguing subfamily of receptor-
related transcription factors will be needed to shed light on
their biological functions.
We thank Cecilia Larson for help with sequencing, Debbie Katz
for constructing the human Rev-Erb expression plasmid, William
Kaelin for providing pGEX-2TK, Keith Blackwell for helpful dis-
cussions, and Andy Darrow for suggesting DEPC analysis. We also
thank Tom Kadesch and Myles Brown for helpful comments on the
This work was supported by NIH grants DK43806 and DK45586
and by an AFCR Foundation-Merck Early Career Development
Award to M.A.L.
1. Amero, S. A., R. H. Kretsinger, N. D. Montcrief, K. R. Ya-
mamoto, and W. R. Pearson. 1992. The origins of nuclear
receptor proteins: a single precursor distinct from other tran-
scription factors. Mol. Endocrinol. 6:3-7.
2. Aruffo, A., and B. Seed. 1987. Molecular cloning of a CD28
cDNA by a high-efficiency COS cell expression system. Proc.
Natl. Acad. Sci. USA 84:8573-8577.
3. Ausubel, F. M., R. Brent, R. Kingston, D. D. Moore, J. A.
Smith, J. G. Seidman, and K. Struhl (ed.). 1987. Current
protocols in molecular biology. Greene Publishing-Wiley Inter-
science, New York.
4. Beato, M. 1989. Gene regulation by steroid hormones. Cell
5. Berrodin, T. J., M. S. Marks, K. Ozato, E. Linney, and M. A.
Lazar. 1992. Heterodimerization among thyroid hormone recep-
tor, retinoic acid receptor, retinoid X receptor, chicken ovalbu-
min upstream promoter transcription factor, and an endogenous
nuclear protein. Mol. Endocrinol. 6:1468-1478.
6. Blackwell, T. K., L. Kretzner, E. M. Blackwood, R. N. Eisen-
man, and H. Weintraub. 1990. Sequence-specific DNA binding
by the c-Myc protein. Science 250:1149-1151.
7. Blackwell, T. K., and H. Weintraub. 1990. Differences and
similarities in DNA-binding preferences of MyoD and E2A
protein complexes revealed by binding site selection. Science
8. Bugge, T. H., J. Pohl, 0. Lonnoy, and H. G. Stunnenberg. 1992.
RXRa, a promiscuous partner of retinoic acid and thyroid
hormone receptors. EMBO J. 11:1409-1418.
9. Cooney, A. J., S. Y. Tsai, B. W. O'Malley, and M.-J. Tsai. 1992.
Chicken ovalbumin upstream promoter transcription factor
(COUP-TF) dimers bind to different GGTCA response ele-
ments, allowing COUP-TF to repress hormonal induction of the
vitamin D3, thyroid hormone, and retinoic acid receptors. Mol.
Cell. Biol. 12:4153-4163.
10. Danielsen, M., L. Hinck, and G. Ringold. 1989. Two amino acids
VOL. 13, 1993
HARDING AND LAZAR
within the knuckle of the first zinc finger specify DNA response
element activation by the glucocorticoid receptor. Cell 57:1131-
11. Davis, I. J., T. G. Hazel, and L. F. Lau. 1991. Transcriptional
activation by Nur77, a growth factor-inducible member of the
steroid hormone receptor superfamily. Mol. Endocrinol. 5:854-
12. Evans, R. M. 1988. The steroid and thyroid hormone receptor
superfamily. Science 240:889-895.
13. Forman, B. M., and H. H. Samuels. 1990. Interactions among a
subfamily of nuclear hormone receptors: the regulatory zipper
model. Mol. Endocrinol. 4:1293-1301.
14. Funk, W. D., D. T. Pak, R. H. Karas, W. E. Wright, and J. W.
Shay. 1992. A transcriptionally active DNA-binding site for
human p53 protein complexes. Mol. Cell. Biol. 12:2866-2871.
15. Green, S., and P. Chambon. 1988. Nuclear receptors enhance
our understanding of transcription regulation. Trends Genet.
16. Hamada, K., S. L. Gleason, B.-Z. Levi,
Appella, and K. Ozato. 1989. H-2RIIBP, a member of the
nuclear hormone receptor superfamily that binds to both the
regulatory element of major histocompatibility class I genes and
the estrogen response element. Proc. Natl. Acad. Sci. USA
17. Hard, T., E. Kellenbach, R. Boelens, B. A. Maler, K. Dahlman,
L. P. Freedman, J. Carlstedt-Duke, K. R. Yamamoto, J.-A.
Gustafsson, and R. Kaptein. 1990. Solution structure of the
glucocorticoid receptor DNA-binding domain. Science 249:157-
18. Harding, H. P., and M. A. Lazar. Unpublished observations.
19. Hazel, T. G., D. F. Nathans, and L. F. Lau. 1988. A gene
inducible by serum growth factors encodes a member of the
steroid and thyroid hormone receptor superfamily. Proc. Natl.
Acad. Sci. USA 85:8444-448.
20. Hirst, M. A., L. Hinck, M. Danielsen, and G. M. Ringold. 1992.
Discrimination of DNA response elements for thyroid hormone
and estrogen is dependent on dimerization of receptor DNA
binding domains. Proc. Natl. Acad. Sci. USA 89:5527-5531.
21. Hwung, Y.-P., D. T. Crowe, L.-H. Wang, S. Y. Tsai, and M.-J.
Tsai. 1988. The COUP transcription factor binds to an upstream
promoter element of the rat insulin II gene. Mol. Cell. Biol.
22. Kaelin, W. G., W. Krek, W. R. Sellers, J. A. DeCaprio, F.
Ajchenbaum, C. S. Fuchs, T. Chittenden, Y. Li, P. J. Farnham,
M. A. Blanar, D. M. Livingston, and E. K. Flemington. 1992.
Expression cloning of a cDNA encoding a retinoblastoma-
binding protein with E2F-like properties. Cell 70:351-364.
23. Katz, D., T. J. Berrodin, and M. A. Lazar. 1992. The unique
C-termini of the thyroid hormone receptor variant, c-erbAa2,
and the thyroid hormone receptor al mediate different DNA-
binding and heterodimerization properties. Mol. Endocrinol.
24. Kliewer, S. A., K. Umesono, R. A. Heyman, D. J. Mangelsdorf,
J. A. Dyck, and R. M. Evans. 1992. Retinoid X receptor-
COUP-TF interactions modulate retinoic acid signaling. Proc.
Natl. Acad. Sci. USA 89:1448-1452.
25. Kliewer, S. A., K. Umesono, D. J. Mangelsdorf, and R. M.
Evans. 1992. Retinoid X receptor interacts with nuclear recep-
tors in retinoic acid, thyroid hormone, and vitamin D3 signal-
ling. Nature (London) 355:446X449.
26. Kliewer, S. A., K. Umesono, D. J. Noonan, R. A. Heyman, and
R. M. Evans. 1992. Convergence of 9-cis retinoic acid and
erodimer formation of their receptors. Nature (London) 358:
27. Kumar, V., and P. Chambon. 1988. The estrogen receptor binds
tightly to its responsive element as a ligand-induced homodimer.
28. Ladias, J. A. A., and S. K. Karathanasis. 1991. Regulation of the
apolipoprotein AI gene by ARP-1, a novel member of the steroid
receptor superfamily. Science 251:561-565.
29. Laudet, V., C. Hanni, J. Coll, F. Catzeflis, and D. Stehelin. 1992.
S. Hirschfeld, E.
Evolution of the nuclear receptor gene superfamily. EMBO J.
30. Lazar, M. A., and T. J. Berrodin. 1990. Thyroid hormone
receptors form distinct nuclear protein-dependent and indepen-
dent complexes with a thyroid hormone response element. Mol.
31. Lazar, M. A., R. A. Hodin, G. R. Cardona, and W. W. Chin.
1990. Gene expression from the c-erbAa/Rev-ErbAa genomic
locus: potential regulation of alternative splicing by complemen-
tary transcripts from opposite DNA strands. J. Biol. Chem.
32. Lazar, M. A., R. A. Hodin, and W. W. Chin. 1989. Human
carboxyl-terminal variant of a-type c-erbA inhibits trans-activa-
tion by thyroid hormone receptors without binding thyroid
hormone. Proc. Natl. Acad. Sci. USA 86:7771-7774.
33. Lazar, M. A., R. A. Hodin, D. S. Darling, andW. W. Chin. 1988.
Identification of a rat c-erbAa-related protein which binds
deoxyribonucleic acid but does not bind thyroid hormone. Mol.
34. Lazar, M. A., R. A. Hodin, D. S. Darling, and W. W. Chin. 1989.
A novel member of the thyroid/steroid hormone receptor family
is encoded by the opposite strand of the rat c-erbAa transcrip-
tional unit. Mol. Cell. Biol. 9:1128-1136.
35. Lazar, M. A., K. E. Jones, and W. W. Chin. 1990. Isolation of
a cDNA encoding human Rev-ErbAa: transcription from the
non-coding DNA stand of a thyroid hormone receptor gene
results in a related protein which does not bind thyroid hor-
mone. DNA Cell. Biol. 9:77-83.
36. Leid, M., P. Kastner, R. Lyons, H. Nakshatri, M. Saunders, T.
Zacharewski, J. Chen, A. Staub, J.-M. Garnier, S. Mader, and
P. Chambon. 1992. Purification, cloning, and RXR identity of
the HeLa cell factor with which RAR or TR heterodimerizes to
bind target sequences efficiently. Cell 68:377-395.
37. Luisi, B. F., W. X. Xu, Z. Otwinowski, L. P. Freedman, K. R.
Yamamoto, and P. B. Sigler. 1991. Crystallographic analysis of
the interaction of the glucocorticoid receptor with DNA. Nature
38. Mangelsdorf, D. J., E. S. Ong, J. A. Dyck, and R. M. Evans.
1990. Nuclear receptor that identifies a novel retinoic acid
response pathway. Nature (London) 345:224-229.
39. Marks, M. S., P. L. Hallenback, T. Nagata, J. H. Segars, E.
Appella, V. M. Nikodem, and K. Ozato. 1992. H-2RIIBP (RXR3)
dimerization provides a mechanism for combinatorial diversity
in the regulation of retinoic acid and thyroid hormone respon-
sive genes. EMBO J. 11:1419-1435.
40. Milbrandt, J. 1988. Nerve growth factor induces a gene homol-
ogous to the glucocorticoid receptor gene. Neuron 1:183-188.
41. Miyajima, N., R. Horiuchi, Y. Shibuya, S. Fukushige, K. Mat-
subara, K. Toyoshima, and T. Yamamoto. 1989. Two erbA
homologs encoding proteins with different T3 binding capacities
are transcribed from opposite DNA strands of the same genetic
locus. Cell 57:31-39.
42. Munroe, S. H., and M. A. Lazar. 1991. Inhibition of c-erbA
mRNA splicing by a naturally occurring antisense RNA. J. Biol.
43. Naar, A. M., J.-M. Boutin, S. M. Lipkin, V. C. Yu, J. M.
Holloway, C. K. Glass, and M. G. Rosenfeld. 1991. The orien-
tation and spacing of core DNA-binding motif dictate selective
transcriptional responses to three nuclear receptors. Cell 65:
44. O'Donnell, A. L., E. D. Rosen, D. S. Darling, and R. Koenig.
1991. Thyroid hormone receptor mutations that interfere with
transcriptional activation also interfere with receptor interaction
with a nuclear protein. Mol. Endocrinol. 5:94-99.
45. O'Malley, B. W., and 0. M. Conneely. 1992. Orphan receptors:
in search of a unifying hypothesis for activation. Mol. Endo-
46. Prost, E., and D. D. Moore. 1986. CAT vectors for analysis of
eukaryotic promoters and enhancers. Gene 45:107-111.
47. Samuels, H. H., L. Stanley, and J. Casanova. 1979. Depletion of
L-3,5,3'-triiodothyronine and L-thyroxine in euthyroid calf se-
rum for use in cell culture studies of the action of thyroid
hormone. Endocrinology 105:80-85.
MOL. CELL. BIOL.
NOVEL BINDING SITE OF ORPHAN RECEPTOR Rev-ErbAat
48. Seagraves, W. A., and D. S. Hogness. 1990. The E75 ecdysone-
inducible gene responsible for the 75B early puff in Drosophila
encodes two new members of the steroid receptor superfamily.
Genes Dev. 4:204-219.
49. Sladek, F. M., W. M. Zhong, E. Lai, and J. E. Darnell. 1990.
Liver-enriched transcription factor HNF-4 is a novel member of
the steroid hormone receptor superfamily. Genes Dev. 4:2353-
50. Spanjaard, R. A., D. S. Darling, and W. W. Chin. 1991.
Ligand-binding and heterodimerization activities of a conserved
region in the ligand-binding domain of the thyroid hormone
receptor. Proc. Natl. Acad. Sci. USA 88:8587-8591.
51. Sturm, R., T. Baumruker, J. B. R. Franza, and W. Herr. 1987.
A 100-kD HeLa cell octamer binding protein (OBP100) interacts
differently with two separate octamer-related sequences within
the SV40 enhancer. Genes Dev. 1:1147-1160.
52. Tran, P., X.-K. Zhang, G. Salbert, T. Hermann, J. M. Lehmann,
and M. Pfahl. 1992. COUP orphan receptors are negative
regulators of retinoic acid response pathways. Mol. Cell. Biol.
53. Tsai, S. Y., J.-A. Carlstedt-Duke, N. L. Weigel, K. Dahlman,
J.-A. Gustafsson, M.-J. Tsai, and B. W. O'Malley. 1988. Molec-
ular interactions of steroid hormone receptor with its enhancer
element: evidence for receptor dimer formation. Cell 55:361-
54. Umesono, K., and R. M. Evans. 1989. Determinants of target
gene specificity for steroid/thyroid hormone receptors. Cell
55. Umesono, K., K. K. Murakami, C. C. Thompson, and R. M.
Evans. 1991. Direct repeats as selective response elements for
the thyroid hormone, retinoic acid, and vitamin D3 receptors.
56. Wang, L. H., N. H. Ing, S. Y. Tsai, B. W. O'Malley, and M.-J.
Tsai. 1991. The COUP-TFs compose a family of functionally
related transcription factors. Gene Expression 1:207-216.
57. Wilson, T. E., T. J. Fahrner, M. Johnston, and J. Milbrandt.
1991. Identification of the DNA binding site for NGF-IB by
genetic selection in yeast. Science 252:1296-1299.
58. Wilson, T. E., R. E. Paulsen, K. A. Padgett, and J. Milbrandt.
1992. Participation of non-zinc finger residues in DNA binding
by two nuclear orphan receptors. Science 256:107-110.
59. Wright, W. E., M. Binder, and W. Funk. 1991. Cyclic amplifi-
cation and selection of targets (CASTing) for the myogenin
consensus binding site. Mol. Cell. Biol. 11:4104-4110.
60. Yu, V. C., C. Delsert, B. Anderson, J. M. Holloway, 0. V.
Devary, A. M. Naar, S. Y. Kim, J.-M. Boutin, C. K. Glass, and
M. G. Rosenfeld. 1991. RXR,: a coregulator that enhances
binding of retinoic acid, thyroid hormone, and vitamin D
receptors to their cognate response elements. Cell 67:1251-
61. Zhang, X.-K., B. Hoffmann, P. B.-V. Tran, G. Graupner, and
M. Pfahl. 1992. Retinoid X receptor is an auxiliary protein for
thyroid and retinoic acid receptors. Nature (London) 355:441-
VOL. 13, 1993