(Received for publication, September 4, 1996, and in revised form, January 9, 1997)
From the Department of Pharmacology, University of Minnesota Medical School, Minneapolis, Minnesota 55455
A negative, regulatory DNA element from the mouse
cellular retinoic acid-binding protein I gene promoter was identified.
This DNA element, located approximately 1 kilobase upstream from the transcription initiation site of this gene, contained a pair of direct repeats (DRs) separated by 4 base pairs (DR4,
TTGG). By examining a series of
reporters deleted or mutated within this DR4 region, it was concluded
that the core sequence of this DR4, including both repeats and the
spacer, was required for suppressive activity in the mouse embryonal
carcinoma cell line P19. From gel retardation experiments, it was
concluded that both repeated sequences were essential for specific
protein binding, but the spacer sequence was not as critical. Specific
residues required for protein binding to this DR4 were identified. In
P19 cells, retinoic acid induced the binding of nuclear factors to DR4
and suppressed the activities of the reporters containing this DR4. Co-expression of retinoic acid receptor or thyroid hormone receptor
1 (T3R
1) significantly inhibited
the expression of this reporter in P19 cells. Gel retardation with
in vitro-synthesized nuclear receptors demonstrated
specific binding of this DR4 by T3R
1
monomers, homodimers, or heterodimers of
T3R
1/retinoid receptor X
. A biological
function of DR4 in crabp-I gene regulation in P19 cells was
suggested.
Retinoic acid (RA)1 exerts pleiotropic effects in animals, and the effects are mediated through various cellular components (for review, see Refs. 1-4). The nuclear receptors for RA, including retinoic acid receptors (RARs) and retinoid receptors X (RXRs), regulate gene expression by enhancing or suppressing the transcriptional machinery via binding to DNA sequences called RA response elements (RAREs). In the cytosol, two cellular retinoic acid-binding proteins (CRABPs) exist for specific binding to RA, designated as CRABP-I and CRABP-II. It is suggested that these cytosolic receptors for RA are involved in RA metabolism, thereby controlling the amount of RA molecules available to RARs/RXRs (for review, see Refs. 3 and 4). In developing embryos, both CRABPs are highly expressed. In adult animals, CRABP-I is ubiquitously expressed at low levels but is highly expressed in RA-sensitive tissues such as the eye and the testis (5-8). In contrast, CRABP-II is specific to the skin (9, 10). Studies in cultured cells showed that CRABP-II expression was directly induced by RA via an RARE located in its promoter region, whereas the regulation of CRABP-I expression involved various signaling pathways such as protein kinases (11), DNA methylation (12), and RA (13). Based upon the promoter sequence, the mouse crabp-I gene was characterized as a housekeeping gene (14). However, the upstream region of this gene contained numerous inverted repeat sequences and putative binding sites for transcription factors, indicating a complex regulatory mechanism involved in its cell type- and developmental stage-specific expression (15). The bovine crabp-I gene was also characterized (16) and seemed to have the same genomic organization and a similar promoter sequence as the mouse gene, suggesting a highly conserved regulatory mechanism for the expression of this gene among animal species.
Because the crabp-I knockout mice displayed no apparent
phenotypes (17-19), the function of CRABP-I in animals remained
unclear. However, it was demonstrated in transgenic mice (20) and
embryonal carcinoma cells (21) that elevated CRABP-I expression was
associated with abnormal cellular differentiation and RA-responsive
gene expression. It was also shown in embryonic palate cells that the expression of RAR, transforming growth factor
3, and
tenascin was altered as a result of introducing anti-crabp-I
oligonucleotides into the cultures (22). In addition, recent
biochemical studies provided more evidence for a role of
crabp-I in RA catabolism (23, 24). It is possible that an
abnormally high level of crabp-I expression may disturb RA
concentration, thereby affecting gene expression in specific cells at a
critical time. Therefore, it is suggested that the level of
crabp-I expression must be tightly controlled for certain
cell types.
The study of the mouse crabp-I genomic structure revealed
several interesting features within 3 kb in the upstream region of the
promoter, such as a GC content of greater than 70%, 9 pairs of
inverted repeats, 5 copies of GC boxes (Sp-1 sites, GGGCGG), and
several potential binding sites for transcription factors, including
putative hormone response elements (HREs) and consensus sequences for
AP-1 and AP-2 binding (14, 15). The biological activity of this 3-kb
upstream sequence was demonstrated in transgenic mouse embryos and cell
cultures using an Escherichia coli -galactosidase (lacZ)
reporter (15). By examining a series of lacZ reporters deleted in
various portions of the upstream sequence, the minimal promoter and
cell type-specific regulatory regions were determined. From studies of
transgenic mouse embryos, it was concluded that the information for
developmental stage-specific regulation was encoded within this 3-kb
upstream region where both positive and negative DNA elements were
located. Most notably, a DNA fragment conferring a strongly suppressive
activity for this promoter was located approximately 1000-1200 bp
upstream from the transcription initiation site. This 200-bp DNA
segment also functioned in heterologous promoters and was able to
interact specifically with proteins of nuclear extracts isolated from
P19 mouse embryonal carcinoma cells in which this gene expression was
regulated by RA. Within this region, a putative HRE was identified,
which contained two pairs of direct repeats (DRs), each separated by 5 bp (DR5) and 4 bp (DR4), respectively. To determine if this putative
response element is a functional negative response element and to
better understand the regulatory mechanism for crabp-I gene
expression, we conducted experiments aimed at characterizing this
negative HRE in terms of its biological activities in cultured cells
and its protein-binding properties. We now report the studies of
detailed analysis of this negative HRE sequence demonstrated in both
gel retardation and transfection experiments.
P19 cells were maintained in
-minimum Eagle's medium supplemented with 2.5% fetal calf serum
and 7.5% calf serum as described (15).
The reporter deletion constructs were made by mutating the original CRABP-lacZ reporter (15) with either restriction enzyme-generated fragments or polymerase chain reaction-generated DNA fragments. All the mutations were confirmed by DNA sequencing.
Cell Transformation and the Quantitation of LacZ Reporter ActivitiesTo determine the biological activities of mutants
truncated at various portions of the DNA segment containing this HRE,
P19 cells were plated in 24-well plates (5 × 104
cells/well) and transfected with each truncated reporter DNA using the
calcium phosphate precipitation method. Protein concentrations were
determined using a Bio-Rad protein assay kit. The specific lacZ
activity (SLA) was determined between 24 and 40 h, using orthonitrophenyl--D-galactopyranoside (Sigma) as
substrate, as described previously (15). For oligonucleotide-generated
point mutations, the reporter activity was shown as relative lacZ
activity (RLA) by comparing its SLA to the SLA of the parental
construct. To determine the effects of nuclear receptor expression on
the putative HRE activity in P19, cells were co-transfected with the lacZ reporter and one of the nuclear receptor expression vectors. The
expression vectors for the nuclear receptors were made by inserting
each corresponding cDNA into a cytomegalovirus expression vector
(25). The RLA of each co-transfection experiment was determined by
comparing its SLA to the SLA of the control, where the RLA was
arbitrarily set at 100%. For all of the assays, triplicate cultures
were used in each experiment, and three to five independent experiments
were conducted to obtain the means and standard errors of the mean
(S.E.).
Gel retardation experiments using
nuclear extracts of cultured cells were as described previously (11,
15). Nuclear extracts from P19 cells were prepared using the method of
Standke et al. (26). Gel retardation experiments using
in vitro-translated proteins were conducted according to an
established protocol (27). Each RAR and RXR cDNA was inserted into
pGem2 (Promega, Madison, WI) for in vitro transcription and
translation by using the TNT T7-coupled reticulocyte lysate system
(Promega). A yeast expression vector and an antibody for
T3R1 were kindly provided by Dr. H. Towle
(Department of Biochemistry, University of Minnesota) (28). The protein
extract was incubated with 0.4 ng of probe (5 × 104
cpm) in 50 µl of reaction buffer (20 mM HEPES, pH 7.6, 0.1% Nonidet P-40, 50 mM KCl, 1 mM
-mercaptoethanol, 2 µg of poly(dI
dC), and 20% glycerol) at
room temperature for 30 min. The reaction mixture was then subjected to
polyacrylamide gel electrophoresis as described (15). For supershift
experiments, the T3R
1-specific antibody was
added to protein-DNA complex and incubated for another 30 min at room
temperature. For competition experiments, DR4 and a palindromic thyroid
hormone response element (TRE) (Ref. 29; 5
-TCGAGATCTCAGGTCATGACCTGAGATC-3
) was used.
Previous studies showed that an approximately 200-bp DNA
fragment (nucleotide position 1180 to
993, according to our
previous numbering system described in Ref. 15) of the mouse
crabp-I gene promoter exerted a suppressive activity when it
was fused to either the crabp-I minimal promoter or other
heterologous promoters in P19 cells (15). To determine the exact DNA
sequence required for the negative activity, serial deletion mutants
were made, designated as
1180,
1110,
1103,
1046, and
993
(Fig. 1A). The SLAs of these five truncated
mutants and the parental reporter were determined in P19 cells as shown
in Fig. 1B. Deletions
1180,
1110,
1103, and
1046
retained the negative activity, but a further deletion to
993
position lost the suppressive activity. Therefore, the sequence between
1046 and
993 encoded the suppressive activity, which contained two
overlapping DRs, one of the DR5 type
(AGGAA) and the other of the DR4
type (TTGG).
To determine which of the two DRs was responsible for this negative
regulatory activity, deletions truncated at various portions of this
54-bp sequence were made, designated as 1027,
1020, and
1012 as
shown in Fig. 2A. The
1027 retained the
complete DR5 and DR4, the
1020 was deleted in the left-side repeat of DR5, and the
1012 retained only the intact DR4. Fig. 2B
shows the RLA of each deletion mutant as compared with the activity of
the reference reporter,
993. It seemed that the requirement for the
suppressive activity was the addition of the DR4 sequence to the
reference reporter
993, as shown by the fully suppressed activity of
the
1012 construct. Thus, the DNA sequence containing the suppressive
activity was located between
1012 and
993, which contained the DR4
sequence.
To examine if this DNA fragment could interact with specific nuclear
factors from P19, gel retardation was conducted by first using the
54-bp fragment as the probes, as shown in Fig.
3A. As expected, this fragment was
specifically retarded by P19 nuclear extracts (lane 1),
which could be competed out specifically by either the 54-bp fragment
(lanes 2-4) or a larger fragment covering the 54-bp
sequence (lanes 5-7), but not by an adjacent DNA fragment (lanes 8-10). To further define the sequence responsible
for protein binding to this 54-bp region, various sequences from this
54-bp region were used as the competitors as shown in Fig.
3B. The retarded band could be competed out by the unlabeled
54-bp fragment (lanes 2 and 3), a fragment
containing the DR4 sequence (lanes 8 and 9), and
a RARE derived from the RAR promoter (lanes
10 and 11; sequence reported in Ref. 25), but not by
the fragments containing a half site of the DR5 (lanes 4 and
5) or the entire DR5 sequence (lanes 6 and
7). Thus, in consistence with the transfection results (Fig.
2), protein factors binding to this 54-bp fragment were specific to the
DR4 site but not the DR5 site. In addition, these protein factors were
able to interact with the RARE derived from the RAR
promoter. It was concluded that the DR4 of the crabp-I
promoter, between
1012 and
993, functioned as a negative DNA
element in P19 cells and shared some common protein factors with a
typical RARE of the DR5 type.
Characterization of the DR4 Negative DNA Element
To further
characterize this new negative DNA element, mutations at specific
residues of this DR4 site were made by polymerase chain reaction
mutagenesis, and their biological activities and protein-binding
properties were determined in transfection and gel retardation
experiments, respectively. The first series of mutants were mutated,
three bases at a time, from the wild-type fragment (labeled with HRE)
and designated as m1, m2, m3 and m4, respectively (Fig.
4A). These mutant DNA fragments were first tested for competition with the wild-type DNA fragment in gel retardation experiments as shown in Fig. 4B. Like the
wild-type DNA fragments (lanes 2 and 3), the m1
(lanes 4 and 5), m4 (lanes 10 and
11), and RARE (lane 12) fragments were able to
efficiently compete with the wild-type fragments. In contrast, the m2
(lanes 6 and 7) and m3 (lanes 8 and
9) mutants failed to compete in these protein-DNA
interactions. The same results were obtained by using probes prepared
from the DR4 fragments (data not shown). Thus, the m2 and m3 sequences
were important for specific protein binding to this DR4 element,
whereas mutations in the m1 and m4 regions had no significant effects
on protein binding.
To determine the effects of these mutations on the biological activity
of this DR4, trinucleotide mutant reporters were made by replacing the
wild-type DR4 sequence in the 1020 construct with each mutant
sequence and were tested in transfection experiments as shown in Fig.
4C. In consistence with the gel retardation results, the m1
mutation did not affect the suppressive activity. Interestingly, the
suppressive activity was partially affected in the m2 and m4 mutants
but was completely abolished in the m3 mutant. Thus, it was concluded
that the 5
-flanking sequence of DR4 was not required for either the
biological activity or protein binding of this DR4, whereas the DR4
core sequence, covering the regions of m2, m3, and m4, was essential
for its biological activity, although mutation in the m4 sequence had
no significant effects on protein binding to this DNA element.
To further define the specific residues required for DNA binding, point
mutations (Fig. 5A) were made within the m2
and m3 regions by polymerase chain reaction, and the mutant DNA
fragments were examined to determine if they could compete with the
probes prepared from the wild-type DR4 fragments in gel retardation
experiments. As shown in Fig. 5B, mutations in any single
residue within the m2 region (lanes 4-6), a combination of
any two residues (lanes 7-9), or the entire m2 region
(lane 3) rendered this DNA fragment unable to compete with
the wild-type DNA fragment. As expected, the wild-type DNA fragment
competed efficiently (lane 2). Thus, it was concluded that
all three residues of the m2 region were critical for specific protein
binding to this DNA element. Using the same strategy, point and double
mutations within the m3 region were made and tested to see if they
could compete with the wild-type DR4 fragment as shown in Fig.
5C. Like the m3 mutant (lane 3), the m3-1
mutation (lane 4) failed to compete with the wild-type fragment, but both the m3-2 (lane 5) and m3-3 (lane
6) mutations were able to complete with the wild-type DNA
sequence. Consistently, the two double mutants containing the m3-1
mutation, including m3-1.2 (lane 7) and m3-1.3 (lane
9), failed to compete, but the m3-2.3 mutant (lane 8)
successfully competed with the wild-type DNA fragment. As a control,
the wild-type DR4 fragment (lane 2) successfully competed in
the reaction. Thus, it was concluded that the m3-1 position was the
most critical residue within the m3 region for specific protein
binding.
RA Induction of Nuclear Factors Binding to the Negative HRE of crabp-I Promoter
It was shown previously that RA affected
crabp-I gene expression in P19 cells. The early effect of RA
on this gene seemed to be a suppression, but the late effect was a slow
accumulation of the mRNA, possibly due to the stabilization of
mRNA (13). To locate the DNA sequence responsible for the effect of
RA, the wild-type CRABP-lacZ reporter, the DR4 deleted reporter
(993), the DR4 wild-type reporter (
1046), and DR4 mutated reporters (m2, m3, and m4) were transfected into P19, followed by the addition of
vehicle or RA (10
7 M) for 24 h (Fig.
6A). RA exerted an inhibitory effect on the wild-type reporter (CRABP-lacZ), but not the DR4 deleted (
993) or DR4
mutated reporters (m2, m3, and m4). The inhibition on DR4 wild-type
reporter (
1046) was not clear because the reporter activity was too
low. We then tested if protein factors binding to this negative HRE
were induced by RA in P19 cells. As shown in Fig. 6B, the
retarded bands were specifically competed by DR4 fragment (lanes
2 and 7) or RARE (lanes 5 and 10)
but not by adjacent sequences (L, lanes 3 and 8;
DR5, lanes 4 and 9). In addition, protein factors
binding to this DR4 were strongly induced by RA (lane 6) as
compared with control (lane 1). To examine if RA
specifically induced nuclear factors binding to the DR4, a probe
derived from the minimal promoter sequence (
153 to
104,
5
-GCCTTAGGGCGGGGAGTAGTCGGGCTCACCCCTCGTGGGCCACCCCCCGCCC-3
) containing Sp-1 sites was tested in parallel experiments
(lanes 14-16). Proteins binding to the minimal promoter
sequence were not induced by RA (lane 16) as compared with
the control (lane 15), whereas proteins binding to the DR4
sequence were induced by RA significantly (lane 13) as
compared with its corresponding control (lane 12). Thus, it
was concluded that protein factors binding to this negative HRE were
specifically induced by RA, supporting a biological function of this
negative HRE in RA-regulated crabp-I expression.
Biological Activities and DR4 Binding of T3R
Because many genes could be induced by RA in
embryonic stem cells (30) and P19 cells,2
such as several RARs/RXRs, thyroid hormone receptors (31), and some
homeobox genes, we then tested the effects of three RARs, RXR, and T3R
1 on the
wild-type reporter and DR4 mutants in co-transfection experiments as
shown in Fig. 7A. It seemed that the
expression of RAR
or T3R
1
suppressed the wild-type CRABP-lacZ reporter activity to approximately
25% and 40%, respectively. RAR
suppressed this
reporter approximately 40%, whereas RAR
had no
significant effects on this reporter. T3R
1
suppressed the m3 mutant approximately 35%, and neither the RARs nor
the RXR
had any effects on this mutant. Similar results
were observed for the m2 and m4 mutants (data not shown). Thus, it was
concluded that the expression of RAR
or
T3R
1 inhibited the expression of CRABP-lacZ
reporter, possibly through the negative HRE containing a DR4
element.
To determine if T3R1 or RAR
could specifically bind to this DR4, gel retardation experiments were
conducted using proteins expressed in vitro or in yeast as
shown in Fig. 7B. T3R
1 was able
to bind to this DR4 as monomers (labeled a, lane
1), homodimers (labeled b, lane 1), or
heterodimers with RXR
(labeled c, lane
7). Surprisingly, RAR
could not bind to this DR4 as
either homodimers or heterodimers with RXR
(data not
shown). RXR
alone could not bind to this sequence either (lanes 5 and 6). The specificity of retarded
bands was demonstrated in competition experiments (lanes 2, 3, 8, and 9). To provide direct evidence for
T3R
1 or
T3R
1/RXR
binding to this DR4
sequence, an antibody specific to T3R
1 was
used in supershift experiments as shown in lane 4 (for
T3R
1 monomers and dimers) and lane
10 (for T3R
1/RXR
heterodimers). It was concluded that this DR4 sequence could
specifically interact with T3R
1 as monomers,
homodimers, or heterodimers of
T3R
1/RXR
.
We have identified and characterized a negative HRE located
between nucleotide positions 1012 and
993 of the mouse
crabp-I gene upstream region, approximately 1 kb 5
to the
transcription initiation site. A DR4 sequence
(TTGG) was located in this HRE,
which exerted a strongly suppressive effect on the reporter activity.
In addition, RA inhibited the expression of this reporter and induced
specific nuclear factors binding to this HRE. Co-expression of
RAR
or T3R
1 significantly inhibited the expression of this reporter, and
T3R
1 was able to bind to this DR4 sequence
as monomers, homodimers, or heterodimers formed with
RXR
. The entire DR4 sequence was required for the
negative biological activity, but only the repeated sequences were
critical for specific protein binding.
The expression of many RA-regulated genes is mediated through the
nuclear receptor superfamily in either positive or negative fashion,
depending upon the combination of nuclear receptors available to the
cells at a specific time (for reviews, see Refs. 1, 32, and 33). We
have obtained evidence that the depletion of retinoids dramatically
stimulated the expression of both the endogenous crabp-I and
the CRABP-lacZ reporter in transgenic mouse embryos, whereas RA
suppressed the expression of this reporter in transgenic mouse
embryos.3 In consistence with this
observation, the current study demonstrated that RA induced nuclear
factors binding to this HRE and inhibited the activities of reporters
carrying this HRE in P19. In other studies, the
T3R gene was shown to be induced by RA (31). In this study, it was clearly demonstrated that
T3R
1 was able to bind to this HRE and that
its expression suppressed the CRABP-lacZ reporter activities. Thus, it
was highly possible that RA-mediated suppression of crabp-I
expression mainly involved the T3R family. However, because
many genes can be induced by RA, it remains to be determined if this
DR4 can be bound by other nuclear receptor complexes.
It was demonstrated in P19 cells that the expression of the crabp-I gene was under the control of a complicated regulatory machinery involving various signaling pathways such as protein kinases (11), DNA methylases (12), and RA (13). In response to RA treatment, the endogenous crabp-I gene was first suppressed and then induced in P19 cells, which required protein synthesis (13). However, it was also reported that crabp-I gene expression was not responsive to RA in cell types such as F9 cells (9), skin cells, and fibroblasts (10). A recent report showed that crabp-I was suppressed by RA at a high concentration but was induced by RA at a low concentration in embryonic stem cells (30). The expression of this gene might respond to various cellular signals in different cell types. This was supported by the fact that the upstream DNA segments of this gene contain both strongly positive (11, 15) and negative (15) activities. Thus, the expression of crabp-I was differentially regulated in various cell types under different conditions. Nevertheless, the RARs, RXRs, and T3Rs seemed to play important roles in the regulation of crabp-I gene expression.
Previously, a model was proposed for the regulatory elements
controlling crabp-I gene expression (15). It was
hypothesized that both positive and negative regulatory mechanisms were
needed for the control of crabp-I gene expression. An AP-1
element located further upstream of this negative HRE was shown to be
involved in the elevation of CRABP-I expression in P19 cells (11). The minimal promoter activity was under the control of four Sp-1 binding sites (15). In addition, demethylation of the upstream region of this
gene was closely associated with a high level of crabp-I expression in both mouse embryos and P19 cells (12). The study shown
here demonstrated one additional mechanism to silence the crabp-I gene, mediated through
T3R1/RXR
binding to this negative HRE located between the two major positive regulatory regions.
Thus, the expression of this gene in a specific cell is dependent upon
the DNA status and the kind of nuclear factors available to the cell.
The model has been updated by the addition of new information generated
from this study, as shown in Fig. 8, in which the
activities of the negative HRE have been confirmed and the factors
involved in this negative regulation are shown in bold. It
will be interesting to determine what are the factors binding to the
sequences flanking this HRE (shown with questions marks in
Fig. 8) and how these positive and negative elements interact with each
other inside specific cell types where trans-acting factors are
constantly affected by the signals presented to the cells.
Although mice deficient in this gene seemed normal phenotypically, several studies showed that overexpression of this gene altered cellular differentiation (20) or the expression of certain developmental genes (21). Thus, the absence of this gene was not devastating to animal survival, but overexpression of this protein affected cell differentiation in both transgenic mice and cell cultures. A tightly controlled regulatory mechanism for crabp-I gene expression, especially a mechanism to keep the gene silenced under a specific condition, could be important in certain developmental processes such as cellular differentiation. It would be interesting to examine if mutation in this promoter would cause any pathological consequences in the animals.
We thank Drs. R. Evans and V. Giguere
for providing the RAR and RXR cDNAs. We thank Dr. H. Towle for
providing the cDNA as well as the antibody and yeast expression
vector for T3R1. We also thank J. R. Wilkinson for help in supershift experiments.