(Received for publication, November 5, 1996, and in revised form, February 4, 1997)
From the Department of Biochemistry and Molecular Pathology, Northeastern Ohio Universities College of Medicine, Rootstown, Ohio 44272-0095
The cholesterol 7-hydroxylase gene
(CYP7A) is transcriptionally regulated by a number of
factors, including hormones, bile acids, and diurnal rhythm. Previous
studies have identified a region from nucleotides (nt)
74 to
55 of
the rat CYP7A promoter that enhanced bile acid repression of the SV40
early promoter, as assayed with a luciferase reporter gene in
transiently transfected HepG2 cells. The rat CYP7A promoter/reporter
activity was strongly stimulated by cotransfection with an expression
plasmid encoding the nuclear hormone receptor chicken ovalbumin
upstream promoter transcription factor II (COUP-TFII) in a
dose-dependent manner. Site-directed mutagenesis in the
region of nt
74 to
55 altered this stimulation. Recombinant
COUP-TFII expressed in HepG2 or COS-1 cells were found to bind to nt
74
55 and nt
149
128 probes by electrophoretic mobility shift
assay (EMSA) and by supershifting the corresponding band with
COUP-TFII-specific antibodies. The region of nt
176
117 was
previously mapped as a retinoic acid response region and was found to
bind retinoid X receptor (RXR). EMSA supershift assays of wild-type and
mutant oligomers using antibody against RXR revealed that the sequences
between nt
145 and
134 were important for RXR binding. We conclude
that COUP-TFII stimulates the transcriptional activity of the rat CYP7A
promoter by binding to the sequences between nt
74 to
54 and nt
149 to
128. RXR may stimulate CYP7A gene transcription
by binding to a direct repeat of the hormone response element separated
by one nucleotide located at nt
146
134.
Cholesterol 7-hydroxylase (EC 1.14.13.17) catalyzes the first
and rate-limiting step in a pathway that converts cholesterol to bile
acids. The catabolism of cholesterol to bile acids in the liver is the
main mechanism for elimination of cholesterol from the body and thus
plays an important role in maintaining cholesterol homeostasis (1). The
gene CYP7A is transcriptionally regulated by a number of
factors, including bile acids, cholesterol, hormones, and circadian
rhythm (2-6).
The CYP7A promoter structure is typical of a DNA-dependent
RNA polymerase II promoter in that, upstream of the TATA box, there are
several cis-acting elements that regulate the
transcriptional activity of this promoter (7-9). Transient
transfection assay of chimeric CYP7A promoter/luciferase constructs in
HepG2 cells revealed that the region from nt 416 to +32 of the rat
CYP7A gene contained the promoter and regulatory domains
conferring the activation of transcription by dexamethasone and
retinoic acid and suppression by bile acids, phorbol esters, and
insulin (2, 3, 9, 10). We have mapped two footprints that are protected
from DNase I digestion using rat liver nuclear extracts (11). Footprint
(Fp)1 A is located between nt
81 and
35, and FpB is located between nt
148 and
129. Nucleotide
sequences in these two regions are highly conserved among homologous
CYP7A genes of different species. These footprints contain
liver-enriched transcription factors binding sites and hormone response
elements. A putative bile acid response element (BARE) was mapped to
the FpA region, which lead us to propose that a bile acid responsive
nuclear receptor may be mediating bile acid response (11).
The molecular mechanism of bile acid repression is still unknown.
Determining what factors interact with these BARE sequences is a first
step in elucidating the nature of the bile acid response. A nuclear
protein factor, which was found to bind a repeated sequence -65TCAAGTTCAAG-54 was named direct repeat binding protein (DRBP). Binding of DRBP to BARE was diminished when liver nuclear extract prepared from rats fed deoxycholate was used in EMSA (11). The sequence
from nt 73 to
57 contains an imperfect direct repeat of the hormone
response element (HRE, AGGTCA) separated by four nucleotides (DR4).
According to the 3-4-5 rule for the binding specificity of the
steroid/thyroid hormone family of transcription factors (12), a DR4
would be predicted to act as a thyroid hormone response element, hence
"7
TRE" as the tentative name for this DR4 sequence (11).
However, CYP7A promoter constructs containing these sequences did not
respond to thyroid hormones in HepG2 cells (2). A number of nuclear
factors have affinity for DR4 sequences, including TR, XONR, and LXR
that form heterodimers with RXR (13, 14). COUP-TFII (or ARP-1) of
orphan nuclear receptor family with promiscuous binding specificities
also bind a DR4 sequence (15, 16). COUP-TFs are believed to influence
developmental changes, which is supported by the report that the
COUP-TFII promoter responds to differentiation signal, retinoic acid
(17). Recently, we have mapped a complex retinoic acid response region
located between
176 and
117 of the rat CYP7A promoter (10).
Using transient transfection assay of CYP7A promoter/reporter activity
in HepG2 cells, we have screened a number of known steroid/thyroid
hormone nuclear receptors for their effects on gene transcription.
COUP-TFII produced one of the strongest effects on transcription. Here,
we provide evidence that orphan receptor COUP-TFII activates
transcription of rat CYP7A by binding to nt 74 to
54 and to nt
149 to
128 and that an RXR homodimer may bind to a DR1 motif
between
146 to
134.
DNA oligomers were synthesized by National
Biosciences (Plymouth, MN) or Life Technologies, Inc. DNA restriction
and modifying enzymes, reporter lysis buffer, luciferase assay system,
and the reporter vectors pGL2-Basic and pGL2-Promoter were purchased
from Promega (Madison, WI). The radioactive isotopes
[-32P]dCTP (3000 Ci/mol) and
[
-35S]dATP (1200 Ci/mmol) Sequenase sequencing grade
were obtained from ICN (Costa Mesa, CA) and DuPont NEN, respectively.
The DNA purification systems used were the GeneClean kit, from BIO 101, Inc. (La Jolla, CA), and the Qiagen plasmid kit, acquired from Qiagen
Inc. (Chatsworth, CA). Expression plasmids for the transcription factors were the generous gifts of Dr. W. Chen for pLen4S (HNF4) and
pL-H3
(HNF3
), Dr. P. Johnson for pMEXC/EBP (C/EBP
) and pMEXCRP2.seq (LAP), and Dr. M.-J. Tsai for pTF3A (COUP-TFII). Antiserum
against COUP-TFII was a kind gift of Dr. Tsai. Affinity-purified rabbit
polyclonal antibodies against RAR (sc-773; cross-reacts with RAR
1,
2,
1,
2,
1, and
2 isoforms), RXR (sc-774; cross-reacts with RXR
,
, and
isoforms), C/EBP
(sc-61), and
c-erbA
1 (sc-76) were obtained from Santa Cruz
Biotechnology, Inc. (Santa Cruz, CA). Other reagents were from
Sigma.
The human hepatoma cell line (HepG2, ATCC HB8065) and African green monkey kidney cells (COS-1, ATCC CRL1650) were obtained from the American Type Culture Collection, Rockville, MD. The cells were grown in a 1:1 mixture of Dulbecco's modified Eagles medium and F12, (Life Technologies, Inc.) supplemented with 100 units/ml penicillin G-streptomycin sulfate (Celox Corp., Hopkins, MN) and 10% (v/v) heat-inactivated fetal calf serum (JRH Biosciences, Lenexa, KS).
Plasmid ConstructionThe sequences of the CYP7A promoter
from nt 200 to +32 were amplified by polymerase chain reaction (PCR)
using p-416/+32 (3) as the template. Primers were designed to introduce
a 5
KpnI and a 3
XhoI site for cloning into the
luciferase reporter vector, pGL2-Basic (Promega) resulting in the
plasmid p-200/+32. The mutations in the sequences from nt
74 to
54
(see Fig. 1A) were introduced by linking two fragments
synthesized by PCR with mutagenic primers containing restriction sites
at the 5
- or 3
-end and using p-416/+32 as a template. The region from
nt
74 to
54 was replaced with a SpeI site
() by linking the nt
416/
74 (SpeI) and
nt
53(SpeI)/+32 fragments to generate p-416
74/
54 SpeI. To delete the SpeI site, p-416
74/54
SpeI was digested with SpeI, treated with Mung
bean nuclease, and then religated to generate deletion mutant
p-416
74/54. p200
74/54 was obtained by PCR using
p-416
74/54 as a template. p-416LSM
74/
71
(linker-scanning mutation) was
constructed by ligating a nt
70 to +32
(SpeI/XhoI) fragment to
416
74/
54Spe
digested with SpeI and XhoI. p-416
74/
65 was isolated as a spontaneously arising deletion mutant.
p-416LSM-59/
54, p-416LSM
70/
66, and p-416IM
64/
63
(insertion mutation) were constructed similarly
to p-416LSM
74/
71. All promoter/reporter constructs were cloned into
pGL2-Basic vector. All constructs were confirmed by sequencing with
Sequenase 2.0 (U. S. Biochemicals, Corp., Cleveland, OH).
Transfection Assays
Confluent cultures of HepG2 cells grown
in 12-well tissue culture plates were transfected with DNA by calcium
phosphate method as described previously (3, 11). Luciferase activities
were determined with the luciferase assay kit (Promega) according to manufacturer instructions using a Lumat LB9501 luminometer (Berthold Systems, Inc., Pittsburgh, PA). Luciferase activities were normalized for transfection efficiencies by dividing relative light units by
-galactosidase activity expressed from cotransfected pCMV
plasmid
(Clontech).
-Galactosidase activities were determined using
O-nitrophenyl-
-D-galactopyranoside as a
substrate (18). The average of the corrected luciferase activity from
cell extracts is given, and error bars indicate the standard deviation
of activity from triplicate samples. All transfections were repeated at
least two times. Statistical significances were analyzed by Student's t test using Sigma Plot software (Jandel
Scientific, San Rafael, CA).
Nuclei from HepG2 and COS-1
cells were isolated essentially as described (19, 20). Recombinant
COUP-TFII was prepared by transfecting confluent HepG2 or sub-confluent
COS-1 cells in 100-mm tissue culture plates with 20 µg of pTF3A and 1 µg of pCMV. The monolayers were overlaid with DMEM/F12 containing
5 mM butyric acid and 10% fetal calf serum after glycerol
shock, and nuclei were harvested 40 h later. Transfection
efficiency was monitored by
-galactosidase activity.
Double-stranded
synthetic probes for EMSA were prepared by heating equal molar amounts
of complementary synthetic oligomers to 95 °C in 2 × SSC (0.3 M NaCl, 0.03 M Na3 citrate, pH 7.0)
and allowing them to cool to room temperature. The resulting
double-stranded fragments were designed with single-stranded 5
overhangs for end-labeling by incorporating [
-32P]dCTP
(3000 Ci/mol) with the Klenow fragment of DNA polymerase I. Oligonucleotides blunted with non-labeled dNTPs were used as cold
competitors in EMSA. Labeled fragments were isolated from a 15%
polyacrylamide gel or purified through two G-50 spin columns. Binding
reactions were initiated with the addition of 2 µg of nuclear
extracts to 100,000 cpm of oligomer probe dissolved in 20 µl of
buffer containing 12 mM HEPES, pH 7.9, 50 mM
KCl, 1 mM EDTA, 1 mM dithiothreitol, and 15%
glycerol, and 2 µg of poly(dI-dC)·poly(dI-dC). After incubation for
the time indicated in the legends to Figs. 2, 3, 4, 5, samples were run on
4% polyacrylamide gels, dried, and autoradiographed. EMSAs were
quantitated with IP Lab Gel software (Signal Analytics, Corp., Vienna,
VA) in conjunction with a PhosphorImager 445Si (Molecular Dynamics,
Sunnyvale, CA).
Cotransfection of the rat
CYP7A promoter fragment from nt 416 to +32 fused to the luciferase
gene (p416/+32) (Fig. 1A) with a plasmid
encoding recombinant COUP-TFII (pTF3A) was found to strongly stimulate
reporter activity (Fig. 1B). The dose-response curve
resulting from cotransfecting with increasing amounts of pTF3A with
p-200/+32 was indistinguishable from the stimulation curve of p-416/+32
(Fig. 1B) despite lower overall luciferase reporter activity
(Fig. 1C), indicating the COUP-TF activation elements were
downstream of position nt
200. Deletion of nt
74 to
54 (Fig.
1A) reduced the stimulatory effect of COUP-TFII (Fig. 1B).
To define the sequences important for COUP-TFII activation, a number of
deletion () and linker scanning mutations (LSM) were introduced into
the nt
74 to
54 region. The dose-dependent stimulation of the promoter activity of the reporter plasmids by COUP-TFII fell
into roughly three groups. In the first group, the wild-type promoter
fragments (p-416/+32 and p-200/+32) showed the greatest stimulation of
transcriptional activity, and the effect was positive at all points of
the curve (Fig. 1B). The second group, which includes
p-416LSM
74/
71, p-416LSM
70/
66, p-416
74/
65,
p-416LSM
59/
54, p-416
74/
54, and p200
74/54 displayed
considerably flatter dose-response curves (Fig. 1B). The
mutations in this group destroyed one or both of the putative HRE
half-sites, AGGTCA, either by completely deleting the sequence
(p-416
74/
65, p-416
74/
54, and p200
74/54) or by
replacing the sequence with an SpeI restriction site while preserving the relative spacing of sequences on either side of the
mutation (p-416LSM
74/
71, p-416LSM
70/
66, and p-416LSM
59/
54) (Fig. 1A). Reporter plasmids with both HRE half-sites
deleted (p-416
74/
54 and p200
74/54) retained significant
stimulation by COUP-TFII (approximately 3-4 fold), suggesting there
may be additional COUP-TFII sites elsewhere in the promoter. Consistent with the hypothesis that the putative COUP-TFII activation elements were downstream of position nt
200, p-200
74/54 was stimulated to
approximately the same extent as p-416
74/
54, which has the same
deletion but was 216 base pairs longer (Fig. 1B).
The third category was represented by an insertion mutant (IM),
p-416IM64/
63 that displayed an activation intermediate between the
first and second groups. This plasmid, in which the HRE half-sites were
intact but the spacing was increased from 4 to 5 base pairs (Fig.
1A), was strongly activated but to a somewhat lesser extent than the wild type, reaching a maximum of 1740 ± 99% verses
2960 ± 209%, respectively, at a transcription factor plasmid
mass to reporter plasmid mass ratio of 0.2 (Fig. 1B).
The various mutations affected the basal activity. Mutant reporter
plasmids p-41674/54, p-416LSM
74/
71, p-416LSM
59/
54, p-416
74/65, and p200
74/54 had increased basal promoter activity relative to their parental plasmids (p-416/+32 and p-200/+32), as would
be observed if a repressor binding site was mutated. With one
exception, the basal promoter activity also fell into three groups,
corresponding with the COUP-TFII responsiveness of the reporter plasmid
(Fig. 1C). The promoter/reporter constructs with the lowest
response to COUP-TFII had the highest basal activities (p-416
74/54
and p-416LSM
74/
71). p-416IM
64/
63 basal transcriptional activity
was not significantly higher than the wild type. The exception was
p-416LSM
70/
66, which was activated by COUP-TFII to the same extent
as p-416LSM
59/
54 and p-416
74/65 but did not display the same
transcriptional activation over p-416/+32. The activation of
p-416
74/
54 over p-416/+32 was not observed in the kidney cell
line, COS-1.
To determine if COUP-TF activates transcriptional activity
by directly binding to the sequence between nt 74 and
54, the ability of COUP-TF to interact with the CYP7A sequences in
vitro, was measured with the EMSA. Also, other promoter fragments
from nt
200 to
74 were screened to determine if there were
additional COUP-TFII binding sites.
Nuclear extracts were prepared from HepG2 cells transfected with pTF3A,
an expression plasmid specifically encoding COUP-TFII. The gel shift
patterns of the nt 72 to
59 and nt
65 to 54 oligomers are similar
in HepG2 cells, regardless of COUP-TFII overexpression (Fig.
2B). Two extra bands were shifted with nt
74 to
53 probe that were enhanced by COUP-TFII overexpression (Fig.
2). The nt
74 to
53 included the putative DR4, which is overlined
in Fig. 2A; the nt
72 to
59 or nt
65 to 54 oligomers
do not. The diffuse nature of these bands was attributed to the low
abundance of the nuclear factors.
Recombinant COUP-TFII expressed in the kidney cell line COS-1, bound
the nt 74 to
53 (Fig. 2C), indicating the binding of COUP-TF was not dependent on liver-specific factors. The increased band
shift with COS-1 extracts was attributed to the higher activity of the
MT2 promoter of pTF3A, which drives transcription of the recombinant
gene encoding COUP-TFII, in COS-1 cells relative to HepG2 cells. This
presumably results in higher levels of expression of recombinant
COUP-TFII in this cell line. Overexpression of COUP-TFII was associated
with a reduction in DRBP binding, even in probes unable to be shifted
by COUP-TFII (nt
72/
59 and
65/
54). The reduction in DRBP
binding may be attributed to the reduced proportion of DRBP in the
given mass of protein in COS-1 nuclear extracts overexpressed with
recombinant COUP-TFII. Alternatively, if COUP-TFII interacts with DRBP
in solution, the overexpressed COUP-TFII may be sequestering the DRBP
and forms larger complexes with the probe nt
74/
53. COUP-TFII did
not shift the probes lacking the COUP-TFII binding site (nt
72/
59
and
65/
54) (Fig. 2C). It is apparent that COUP-TFII
interacts with the DR4 of nt
74 to
53 and DRBP interacts with
AGTTCAAG sequence (Fig. 2A).
To further confirm that COUP-TFII binds to this sequence, antiserum
raised against COUP-TF was added to an EMSA using the nt 74 to
55
oligomer as a probe of nuclear extract from HepG2 cells. The band that
increased in intensity when shifted by nuclear extracts from HepG2
overexpressing COUP-TFII was supershifted by the anti-COUP-TF antibody
(Fig. 3), indicating that COUP-TFII was interacting
directly with the region. Extracts prepared from cells overexpressing
recombinant C/EBP
, LAP, HNF4, or HNF3
did not produce any changes
in the band-shift pattern (data not shown), and antiserum directed
against C/EBP
(potential binding site at nt
52 to
41) and c-erbA
(a homologue of the thyroid receptor that is regulated by
phosphorylation events) did not change the gel shift pattern (Fig. 3).
Antibody against RAR and RXR did not supershift the nt
74 to
53
probe. However, when performed with extracts prepared from HepG2 cells
treated with all trans-retinoic acid, the ligand for RAR,
the nt
74 to
54 probe was weakly supershifted with anti-RXR (data
not shown). DRBP, the major nuclear factor that binds to the sequence
was not affected by the antibody, indicating that the proteins
responsible for that band are distinct from COUP-TFII and RXR. These
results confirm that the COUP-TFII binds to DR4 motif, which does not
bind thyroid hormone receptor. C/EBP
, LAP, HNF4, and HNF3
do not
bind to this region of CYP7A promoter.
The probe nt 149 to
128 shifted one band in
HepG2 nuclear extracts. Antibody against COUP-TFII supershifted the
oligomer and changed the pattern of bands as if a complex was disrupted (Fig. 4). The amount of probe supershifted was not
increased with increasing amounts of antibody, indicating that antibody
was in excess. Antibody against RXR supershifted a band in HepG2
nuclear extract treated with retinoic acid. The antibody directed
against RAR did not supershift the probe even with extract prepared
from HepG2 cells treated with all trans-retinoic acid.
Retinoic acid treatment may have increased the expression of endogenous
RXR and showed the supershift with anti-RXR.
Once it had been determined that the antibody specific for RXR
supershifted the nt 149 to
128 probe, nucleotides important for the
binding of RXR were determined by EMSA with mutant oligonucleotides and
the antibody against RXR. Double-stranded oligonucleotides were
synthesized in which three bases were changed by transversions. Trinucleotide mutations in nt
145 to
134 completely abolished or
reduced the supershift by antibody against RXR (Fig. 5).
Mutations in nt
133/
131 did not have any effect on supershift by
antibody against RXR. These results indicate that RXR was interacting
with the DR1 motif located from nt
146 to
134.
Cotransfection assays of COUP-TFII expression plasmid with
wild-type and mutant CYP7A proximal promoter constructs and EMSA results indicate that COUP-TFII interacts with nt 74 to
53 and nt
149 to
128 of the rat CYP7A promoter. The region of nt
146 to
134 was shown by EMSA to bind RXR and contains HRE half-sites separated by one nucleotide (DR1), the preferred binding site for RXR
homodimer (14). The pattern of response of the mutants in nt
74 to
54 suggested that the nt
74 to
54 was the element with the
strongest potential for COUP-TF activation of transcription. The
deletion of the sequences from nt
74 to
54 results in the increase
of transcriptional activity of CYP7A promoter/reporter plasmids. This
sequence was previously found to mediate the bile acid response and to
interact with DRBP, whose binding was sensitive to deoxycholate feeding
in rats. Fig. 6 is a model which shows the binding sites
for these transcription factors on the rat CYP7A proximal promoter.
Interactions of transactivators COUP-TFII and RXR with DRBP and their
competition for binding to these bile acid response elements may
regulate the CYP7A gene transcription by bile acids and hormones. The
CCAAT enhancer binding proteins (C/EBP
and C/EBP
or LAP) (21)
consensus binding site flank the COUP-TFII binding site at nt
74 to
54. The C/EBP family of transcription factors are temporally
regulated and are thus important developmental regulators of gene
expression. COUP-TFII may recruit C/EBP on to the apolipoprotein A1
promoter (22). The DBP B-site, which also is the LAP binding site, was
located in the nt
149 to
118 region (23). However, antibody
directed against C/EBP
and LAP did not supershift the nt
149/
118
probe. This is consistent with the lack of effect of these
transcription factors on CYP7A promoter activity we observed. The
juxtaposition of the COUP-TF and C/EBP binding sites at nt
74/
54
and nt
149/
118 may prevent the weak binding of these factors.
In light of the well-characterized negative effect of COUP-TFII on gene
promoters, removing a COUP-TFII binding site in CYP7A promoter may
explain the much higher activity of the deletion mutants (Fig. 1).
However, several recent reports support our finding that COUP-TFII also
can function as a transactivator (24-27). COUP-TFII binds as a stable
homodimer to many variations of GGTCA sequences in direct repeats and
interferes with vitamin D3, thyroid hormone, and retinoic
acid receptors by competing for the AGGTCA sites, the cognate sequences
for the steroid/thyroid hormone receptors (15). However, if the
negative effect of COUP-TFII is due to interfering with a positive
factor, over-expression of COUP-TFII should repress the transcriptional
activity of the CYP7A promoter. COUP-TFII activated transcription of
the CYP7A promoter at all COUP-TFII expression plasmid to reporter
plasmid ratios tried. The deletion of the strong proximal COUP-TFII
site at nt 74 to
55 may allow enhancers in the promoter to interact
more readily with the transcriptional machinery and resulting in higher
transcriptional activity. COUP-TFII has been reported to interact with
TFIIB (28) and to affect the transcriptional activity by interacting
with other transcription factors. For example, COUP-TFII is required for activation of the apolipoprotein A1 promoter by RXR (22). As the
COUP-TFII and RXR both recognized the nt
149 to
128 probe, it is
possible that COUP-TFII may interact with RXR. The heterodimer of
COUP-TFII and RXR is known to bind a DR1 motif and transrepress gene
transcription (13).
Despite the finding that COUP-TF binds to a region responsive to bile
acid repression, the role of COUP-TFII in the bile acid regulation of
CYP7A is not clear and requires further study. The region from nt 65
to
54 can confer the bile acid response to SV40 promoter, but the
probe does not bind COUP-TFII in vitro (Fig. 2, B
and C). However, the DRBP and COUP-TFII apparently share
binding sequence, the 3
COUP-TFII half-site overlaps with the DRBP
binding site (Fig. 2A). COUP-TF clearly influences
cholesterol metabolism, in that this transcription factor has
previously been shown to affect the transcription of the promoters for
apolipoprotein AI (16), apolipoprotein AII, apolipoprotein B (29),
apolipoprotein CIII (29, 30) and cholesteryl ester transfer protein
gene (27).
The physiological role of the sequences from nt 74 to
54 may be to
restrain CYP7A expression. This down-regulation may be necessary
because of the cytotoxicity of bile acids and the need to preserve
cholesterol for synthesis of steroid hormones and membrane components.
The exact interactions of the factors involved has yet to be defined;
however, the data are consistent with a model in which COUP-TFII, RXR,
and DRBP interact with and bind to a negative element in the region
between nt
74 and
54 and a positive element from nt
149 to
118
(10) and lead to the regulation of the CYP7A gene transcription in
response to signals from bile acids and hormones. This work contributes
to the increasing body of evidence indicating COUP-TF has a prominent
role in transcriptional regulation of lipid metabolism and that factor
binding to the nt
74 to
54 and nt
149 to
128 regions are
important determinants of CYP7A transcription.
The transcription factor expression plasmids
were the generous gifts of Drs. W. Chen (HNF4 and HNF3), P. Johnson
(C/EBP and LAP), and M.-J. Tsai (COUP-TFII). Antibody against
COUP-TFII was obtained from Dr. Tsai.