ABSTRACT
cAMP-responsive-element (CRE)-binding factors interaction with nucleosomal DNA has been investigated
in vitro
on the human c-
fos
promoter. Analysis of nucleosome reconstitution of this promoter shows a
preferential nucleosome positioning on the proximal promoter sequences,
including the CRE centered at -60 relative to the start site of transcription. CRE-binding protein (CREB) and modulator protein (CREM) are unable to interact with their recognition site incorporated in a nucleosome. However, competition between transcription factor
binding and nucleosome assembly allows CREM binding and induces important modifications in the nucleosomal structure suggesting the displacement of nucleosomes. These
findings imply that binding of transcription factors to the CRE prior to cAMP
induction might be required to prevent the incorporation of this element in a
nucleosome.
In eucaryotes, gene transcription is the result of a complex interplay between transcription factors and chromatin. The interaction of
trans
-acting factors with specific DNA sequences, which is a prerequisite for
the transcription process, must be achieved in a chromatin environment (
1
-
3
). The packaging of DNA into nucleosomes severely restricts the access of
factors to their binding sites. The DNA wrapped in approximately two turns
around the histone core (one tetramer of H3/H4 and two dimers of H2A/H2B) is
highly bent and the access on one side of the double helix is occluded because
it faces the histone octamer (
4
). Thus, factor binding to nucleosomal DNA may depend on the way the sequence is
incorporated into the nucleosomes. Both the position of the boundaries of the
nucleosome on the DNA (translational positioning) and the orientation of the
double helix around the octamer surface (rotational positioning) are important
determinants for nucleosome factor binding (
5
,
6
). Specific nucleosome organization has been shown to be a common feature among
promoter regions and has been found to be critical for proper gene
transcription by providing a defined architectural conformation for
transcription to take place (
7
,
8
).
Binding of transcription factors to chromatin can be achieved in different ways.
Binding sites may be left free of histone-DNA interactions by being located in
a linker region between nucleosomes (
9
-
11
). Some transcription factors, including the glucocorticoid receptor (
12
-
14
), GAL4 (
15
,
16
) and Max (
17
), have been found to be able to bind DNA within the nucleosome, forming a
ternary complex. In some cases, histone modifications such as acetylation
appear to facilitate protein interaction with nucleosomal DNA (
18
).
Trans
-acting factor binding can also disrupt nucleosomes in order to permit
stable binding for itself or for other factors (
16
,
19
). There are however, a number of examples where nucleosomes have been found to
inhibit
trans
-acting factors binding and/or repress transcription (
20
-
22
). In some instances, access to DNA can only be gained after remodeling of the
chromatin structure which may be induced by other
trans
-acting factors or by chromatin-remodeling factors such as the SWI/SNF complex (
23
). Other studies have demonstrated that the inhibition of transcription factor
binding by chromatin can be prevented if transcription factors interaction with DNA precedes nucleosome assembly. Thus, DNA replication might provide an opportunity for a number of factors to access their sequences before it is packaged into chromatin (
24
-
26
). Recent evidence suggests that promoters can be preset for rapid induction of
transcription as a result of a combined interaction of transcription factors and nucleosomes following DNA replication (
27
).
The c-
fos
proto-oncogene belongs to the class of `immediate early genes' which are
involved in converting extracellular and intracellular signals into changes in
gene expression. c-
fos
transcription is induced rapidly and transiently by a variety of stimuli such
as phorbol esters, growth factors, neurotransmitters, cAMP and others (
28
). A number of
cis
-acting elements have been defined in the promoter region, which allow
activation or repression of transcription by distinct signal transduction pathways. The SRE (Serum Response Element) centered at position -300 with respect to the transcription initiation site is the molecular
target of the protein kinase C (PKC) signal transduction pathway through
binding of the SRF (Serum Response Factor) and associated factors (
29
). c-
fos
transcription is induced by the cAMP/PKA pathway through a major cAMP element
(CRE) centered at -60 (
30
,
31
). This element is recognized by the protein CREB (CRE-binding protein) and a family of structurally related factors referred to
as the CREB/ATF family (
32
). These factors belong to the basic-leucine zipper (b-Zip) class of transcription factors and bind CRE as homo- or heterodimers. A recently cloned member of this family,
CREM (CRE modulator) generates both activators and repressors of transcription through alternative splicing (
33
).
Distinct signal transduction pathways activate c-
fos
transcription independently of protein synthesis within minutes of exposure to activators (
28
). Interestingly, the c-
fos
regulatory elements appear to be constitutively occupied
in vivo
, suggesting that the c-
fos
promoter might be organized in a preset chromatin structure to allow a rapid
response to stimuli (
34
,
35
). To understand the molecular basis of c-
fos
regulation, it is essential to determine how nucleosomes are organized on this
promoter and how this structure affects the binding of regulatory factors. We have addressed this question by analyzing the nucleosome organization on the human c-
fos
promoter in an
in vitro
reconstitution system.
We present evidence that the proximal region of the c-
fos
promoter contains DNA sequences capable of positioning a population of rotationally phased nucleosomes
in vitro
. Pre-bound nucleosomes inhibit CREB and CREM binding to the CRE. However,
concomitant incubation of transcription factors and histones during the
nucleosome reconstitution process allows
trans
-acting factor binding and subsequent remodeling of the nucleosomal
structure. These findings suggest that the formation of a pre-bound transcription factor complex during chromatin assembly could be a
necessary step to mediate a rapid response to cAMP induction.
The 465 bp fragment used in reconstitutions contains the human c-
fos
promoter sequences -404/+42 and was obtained by a
Sst
II-
Bgl
II digestion of plasmid FC-3 (
36
). A -222/+42 fragment obtained by PCR was subcloned into the
Bam
HI-
Eco
RI sites of plasmid pUC19 to give pF2. A 264 bp
Eco
RI-
Bam
HI fragment (-222/+42) and a 234 bp
Eco
RI-
Pst
I fragment (-222/+12) derived from pF2 were used in reconstitution experiments.
Fragments were end-labeled using the T4 DNA polynucleotide kinase (New England Biolabs).
Purified core histones were prepared from chicken erythrocyte nuclei using hydroxylapatite chromatography (
18
,
37
). Nucleosomes were reconstituted by salt/urea dialysis as described (
18
) in 200 [mu]l total volume. The final DNA concentration was 20 [mu]g/ml and the histone:DNA (mass to mass) ratio was 0.8:1 (a 1:1 ratio
corresponds approximately to one nucleosome core particle per 160 bp of DNA).
The assembly process was 90% efficient as detected by mobility shift assay. In
the case of co-reconstitution with CREB and CREM proteins (or control extract), 5 [mu]g of proteins were simply added to the initial histone-DNA mix and the reconstitution conditions were kept identical.
pETCREB and pETCREM constructs have already been described (
38
). Bacterial extracts were prepared exactly as described (
38
). Protein concentration was determined by Biorad assay and the proteins were visualized on polyacrylamide-SDS gel. As reported, the recombinant
proteins consisted of >80% of the total protein lysate (
38
). Binding assays were done in 25 mM Tris, pH 7.5, 6 mM MgCl
2
, 0.5 mM EDTA, 0.5 mM dithiothreitol 10% glycerol in 20 [mu]l final volume for 15 min at 20oC.
Hydroxyl radical footprinting was performed as described (
18
). For DNase I analysis, 1 mM MgCl
2
(final concentration) was added to 20 [mu]l samples and digestion was carried out for 2-3 min at 20oC with 0.01 mg/ml DNase I (Sigma) for naked DNA and with 0.1
mg/ml DNase I for reconstitutes. Exonuclease III digestions were performed in
20 [mu]l reactions in 10 mM Tris, pH 7.5, 25 mM KCl, 5% glycerol, 1 mM
dithiothreitol for 15 min at 20oC. For naked DNA samples, 2-20 U of
Exo
III (Promega) were added as 20-200 U were used to digest reconstitution samples. Footprinting
experiments were analyzed on 6% polyacrylamide gels containing 7.5 M urea.
A radiolabeled fragment from the human c-
fos
promoter spanning from -404 to +42 (relative to the start of transcription) was used to examine
the organization of nucleosomes after
in vitro
reconstitution with purified histones (Fig.
1
). This DNA fragment contains all the known c-
fos
regulatory elements (
28
). Nucleosomes were assembled by high-salt/urea dialysis at moderate (0.8:1 histones:DNA by mass) densities of
nucleosomes to avoid non physiological close-packing of nucleosomes. Under these conditions two nucleosomes are expected to form on the DNA fragment. Assemblies were
analyzed by mobility shift assays to verify the presence of retarded complexes
(data not shown).
Several members of the CREB/CREM family of factors were bacterially expressed in
pET vectors as previously described (
38
) and tested for their ability to bind free or reconstituted c-
fos
promoter fragments. As seen by DNase I footprinting, bacterially expressed
CREB, CREM[beta] and CREM[tau] bind DNA efficiently at the CRE element as seen by the clear
protection obtained on this element (Fig.
5
A, lanes 3-5). However, incorporation of the CRE in a nucleosome prevents
interaction of the factors with their recognition site even when high amounts
of proteins (20-fold excess) were added (Fig.
5
A, lanes 8-13). Exclusion from the nucleosomes was complete since mobility shifts
also failed to detect any binding of CRE proteins on the reconstituted
nucleosomes (data not shown). We obtained similar results with other CREM
proteins [CREM[alpha] and ICER (
40
), data not shown]. We conclude that the presence of positioned nucleosomes on
the CRE prevents interaction of
trans
-acting factors.
Our results show that CRE-binding proteins can successfully compete with histones for binding on the
c-
fos
promoter. Preformed rotationally phased nucleosomes positioned on the c-
fos
proximal promoter prevent the interaction of CREB and CREM with their binding
site. However, co-incubation of the
trans
-acting factors with the histones during the reconstitution process allows
transcription factor binding to the CRE on the nucleosome-reconstituted promoter. We found that the nucleosomal organization was disturbed upon transcription factor binding, probably
reflecting the displacement of the nucleosomes on sequences immediately
adjacent to the CRE.
Nucleosomes can be specifically positioned by DNA sequences. Nucleosome positioning depends on DNA structural features such as flexibility
or intrinsic curvature rather than precise sequence requirement (
41
). Nucleosome reconstitution on a DNA fragment comprising the entire c-
fos
promoter
in vitro
revealed a specific nucleosome positioning on the proximal promoter sequences
but not on the distal part of the promoter. Under the conditions of
reconstitutions used which allow the formation of two nucleosomes per DNA
template, we conclude that the proximal part of the promoter contains sequence-specific nucleosome positioning signals which direct preferential
nucleosome assembly whereas on the distal part, the nucleosomes are randomly
arranged. Further analysis using DNase I, OH-radical and
Exo
III cleavage on a smaller DNA fragment comprising the proximal promoter
sequences suggested the presence of four nucleosomes with identical rotational
settings but translationaly staggered by one helical repeat. Such multiple
nucleosome positioning has previously been reported for other DNA sequences
in vitro
and
in vivo
(
42
-
45
). In the absence of translational signals such as specific DNA structural
features or linker histones, nucleosomes have been shown to adopt different
translational positions possibly by being mobile (
43
-
45
).
The major consequence of this multiple positioning on the c-
fos
promoter is that the CRE is incorporated in all four translational positions.
The single rotational phasing directs the CRE major groove towards the histone
core, preventing the interaction of CREB and CREM. Thus, the minimum
requirement for these factors to bind may not accommodate to the constraints
imposed on the DNA in a nucleosome. The CRE 8 bp palindromic sequence is
recognized by the basic region adjacent to the leucine-zipper such that the two positively-charged [alpha]-helices are in contact with the two halves of the
palindrome in the major groove of the DNA helix (
46
). As a result, the CRE is slightly bent towards the leucine-zipper. In addition, flanking bases on each side of the core 8 bp sequence
appear to be important for CREB binding (
47
). Thus, it seems unlikely that CREB or CREM can interact with their site in a
nucleosome in any rotational position, although this remains to be tested.
Phosphorylation of CREB and CREM by PKA has been reported to have a positive
effect on DNA binding although this remains controversial (
32
). We found that the phosphorylation of CREB and CREM by PKA had no effect on
the interaction of these factors with nucleosomal DNA (data not shown). Thus,
an alternative for CREB/CREM binding to the CRE would be to interact before
nucleosome assembly.
The addition of CREM in the reconstitution assays resulted in the concomitant
binding of the transcription factor and the nucleosomes. This was observed in
three different assays. First, in DNAse I footprinting experiments, we observed
both the protection on the CRE and the nucleosomal cleavage pattern indicating
the binding of CREM in presence of nucleosomes. Second, in mobility shift
assays, co-incubation of CREM[beta] and histones resulted in the appearance of larger complexes than
reconstitutions done in presence of either CREM[beta] or histones alone, confirming the interaction of CREM on the nucleosome-reconstituted c-
fos
promoter. Finally, as seen by the presence of new
Exo
III-resistant boundaries upstream of the CRE, it appeared that an important
remodeling of the nucleosomal structure had occurred upon CREM binding. This
could be explained either by a destabilization of the nucleosomes by the CRE-bound factors allowing the
Exo
III to penetrate in the nucleosomal structure, or by a displacement of the
nucleosomes on new translational positions immediately upstream from this
element. However, the size of the CREM dimers (CREM[beta], 66 kDa; CREM[tau], 84 kDa) and the tight contacts they make with DNA argues against
their incorporation in a nucleosome. Therefore, this result more likely
reflects the displacement of nucleosomes onto new translational positions,
leaving the rotational phasing unchanged. Earlier studies have documented
nucleosome positioning by specific DNA binding proteins. Specific interaction
of proteins with DNA was shown to generate arrays of positioned nucleosomes
whereas a random organization was observed in absence of factor binding (
48
-
50
). However, in the case of the c-
fos
promoter described here, nucleosomes were already positioned in the absence of
CRE-binding proteins. Thus, it appears that
trans
-acting factors interaction with the CRE could provide a new boundary on pre-existing sequence-dependent nucleosome positioning signals.
Nucleosome assembly is a stepwise reaction involving a first interaction of the
H3/H4 tetramer and the subsequent addition of two H2A/H2B dimers (
51
). The association of H2A/H2B has been shown to be inhibitory to some
trans
-acting factors binding (reviewed in
8
). The structure of the nucleosome is known to change dramatically with ionic
environment consistent with an increasingly relaxed secondary structure as the
salt concentration is raised (
52
,
53
). We did not detect any interaction of CREB/CREM after reconstitution of the c-
fos
promoter with H3/H4 tetramers (data not shown). Thus, the interaction of CRE-binding factors during the reconstitution process is likely to precede the
association of histones with DNA or could occur at intermediate ionic strength
(0.2-0.7 M NaCl) when the histone-DNA interactions can be easily disrupted.
As stable nucleosomes restrict the access to the CRE, interaction of CREB/CREM
in vivo
with their recognition site could be mediated by a disruption of the
nucleosomal structure. A number of factors have been shown to destabilize and
remodel nucleosomes, promoting trans-acting factors binding. One is the SWI/SNF complex, originally found in
yeast (
54
) but for which a homolog has also been identified in human cells (
55
). This complex appears to interact with nucleosomal DNA and alter histone-DNA contacts, assisting transcription factor binding (
23
). NURF, another factor recently purified from
Drosophila
is able to alter nucleosomal arrays and facilitate the interaction of the GAGA
transcription factor (
56
). Also, a yet unidentified factor from
Drosophila
embryos has been reported to promote chromatin reorganization (
57
). It would be interesting to determine if these factors are able to disrupt the positioned nucleosomes on the c-
fos
promoter to allow CREB/CREM binding after nucleosome assembly.
Alternatively it is possible, as it has been suggested earlier (
24
,
25
,
58
), that CRE-binding proteins interact with the CRE following DNA replication by
competing with the histones for binding to DNA. Although there is, at the moment, no information about the organization of the nucleosomes and transcription factors on the c-
fos
proximal promoter sequences
in vivo
, there is evidence that CREB-related proteins might be bound to the CRE before cAMP induction. Earlier
studies have suggested that the c-
fos
CRE might be constitutively occupied
in vivo
in various mouse tissues and in a human cell line (
34
). Also, on the phosphoenolpyruvate carboxykinase (PEPCK) gene, the CRE site is
occupied by CREB before induction by cAMP suggesting that phosphorylation of
CREB by PKA occurs on dimers already bound to their site (
59
). Similarly, two AP-1 binding sites on the c-
jun
promoter appear constantly occupied
in vivo
, independently of phorbol ester stimulation or UV irradiation (
60
). The SRE on the c-
fos
promoter has also been found occupied by a complex of transcription factors
before induction by epidermal growth factor and these protein-DNA contacts remain unchanged during gene activation and subsequent repression (
35
). Thus, the pre-establishment of protein-DNA complexes before induction might be a common mechanism for
immediate early genes for rapid and transient response to extracellular
signals. Our findings suggest that CREB/CREM binding prior to nucleosome
assembly might be a necessary step to prevent a negative regulation by
chromatin and to provide a defined nucleosomal organization for transcription to take place.
A recent investigation of the regulation of c-
fos
transcription in transgenic mice have suggested that the regulatory elements in
c-
fos
function interdependently, as point mutations in any one of the SIE, SRE, FAP
or CRE sequences abolished transcription from the c-
fos
promoter (
61
). Again, this observation implies that the regulatory elements are
constitutively occupied and suggests that induction
in vivo
requires a sophisticated arrangement of several transcription factors. Interestingly, a nucleosome positioned between the CRE and the FAP site (see Fig.
1
) could bring the CRE and the upstream regulatory elements fairly close
together.
We thank Robert Haché and Alan Wolffe for critical reading of the manuscript. C.S.-P. was supported by fellowships from the Swiss National Science
Foundation, the Association pour la Recherche sur le Cancer and the Centre
National de la Recherche Scientifique.
*To whom correspondence should be addressed at present address: Loeb Medical
Research Institute, 725 Parkdale Avenue, Ottawa, Ontario K1Y 4E9, Canada. Tel:
+1 613 798 5555; Fax: +1 613 761 5354; Email: cschild@civich.ottawa.on.ca
REFERENCES
Return