ABSTRACT
We have previously shown that
in vivo
estradiol-dependent dephosphorylation of MDBP-2-H1 (a member of the histone H1 family) correlates with the loss of
in vitro
preferential binding to methylated DNA. To study the effects
of the phosphorylation/dephosphorylation of MDBP-2-H1 on the expression of the avian vitellogenin II gene, we optimised
an
in vitro
transcription system using HeLa nuclear extracts. We show that in the absence
of the phosphorylated form of MDBP-2-H1 from rooster, methylation of the
vitellogenin II
promoter does not affect the transcription. Addition of purified MDBP-2-H1 from rooster to the
in vitro
transcription system inhibits transcription more efficiently from a methylated than an unmethylated DNA template. Dephosphorylation of rooster MDBP-2-H1 by phosphatase treatment or estradiol treatment of rooster lead
to the loss of inhibitory activity of the protein when added to the
in vitro
transcription assays. These findings indicate that the phosphorylation of MDBP-2-H1 is essential for the repression of the transcription. Taken
together these results establish the relationship between the dephosphorylation
of MDBP-2-H1 caused by estradiol, the down regulation of its binding activity
to methylated DNA and the derepression of
vitellogenin II
transcription.
DNA methylation which occurs in almost every higher organism, appears to be an important component in the regulation of vertebrate gene
expression (
1
). Several laboratories have shown that the methylation of cytosines situated
within CpG dinucleotides resulted in the inhibition of transcription from viral, plant and animal tissue-specific genes. Methylated and non-methylated cloned DNA has been introduced into cells via
transfection (
2
,
3
) or microinjection (
4
,
5
) or has been tested in
in vitro
transcription systems (
6
-
9
). Although it is assumed that DNA methylation affects protein-DNA interaction and induces some conformational changes of the gene (
1
), the mechanism of transcriptional inhibition is poorly understood. Two models
have been proposed to explain how promoter methylation causes transcriptional
repression. 5 methyl-cytosine may inhibit directly the binding of transcription factors to DNA
(
10
,
11
) or it may favour the binding of specific repressor proteins that render the
DNA inaccessible to transcription factors (
12
). However, the detailed mechanism of how these proteins interact with
methylated DNA and how precisely this interaction affects gene expression
remains unknown. For example methyl-CpG binding proteins (MeCP1 and MeCP2), that form complexes with a variety
of unrelated sequences that are methylated at CpG, have been described to be
associated with centromeric heterochromatin (
13
). One of the methylated-DNA binding proteins, MDBP-2-H1, belongs to the histone H1 family and binds
in vivo
and
in vitro
with high affinity (
K
d
10
-9
M) to the methylated promoter region of the avian vitellogenin II gene (
14
,
15
). The DNA binding activity of this protein is down regulated in egg laying hens
and estradiol treated roosters (
16
). MDBP-2-H1 is a phosphoprotein and its state of phosphorylation decreases in
response to estradiol treatment (
17
). Importantly, phosphorylation of this protein is essential for its
in vitro
binding to methylated DNA.
Using an
in vitro
transcription assay, the present study provides evidence that the inhibitory
effect of DNA methylation on the transcription of the avian vitellogenin II
gene is an indirect mechanism and that the phosphorylated form of MDBP-2-H1 selectively inhibits transcription from methylated templates. In
addition, we report that the phosphorylation of MDBP-2-H1 is essential for the repression of the transcription. The
implication of these findings are discussed in the general context of DNA
methylation and gene inactivation.
Heparin-Sepharose was purchased from Pharmacia. [[gamma]-
32
P]ATP (3000 Ci/mmol; 1 Ci = 37 GBq) was obtained from Amersham. 17-[beta]-estradiol was purchased from Serva (Heidelberg). CpG methylase (
Sss
I methylase) and
S
-adenosylmethionine were obtained from New England Biolabs whereas polynucleotide kinase and
Hpa
II endonuclease were from Biofinex Praroman (CH 1724; Switzerland). Ribonuclease-free DNase I and S1 nuclease were purchased from Boehringer Mannheim and calf intestinal phosphatase attached to beaded agarose was from Sigma. HeLa nuclear extracts and ribonucleoside triphosphates were
obtained from Promega.
Where indicated, mature or immature white Leghorn roosters received a single
intramuscular injection of 17[beta]-estradiol (40 mg/ Kg) dissolved in propylene glycol (40
mg/ml).
Liver nuclei from roosters were prepared as described by Sierra (
18
) with the following modifications: for estradiol-treated rooster livers, the buffer contained 1.9 M sucrose and for the non-treated animals the buffer contained 2 M sucrose. Nuclei were then lysed with 1/10 vol 4 M (NH
4
)
2
SO
4
for 30 min on ice. After centrifugation for 60 min at 300 000
g
to remove chromatin, the supernatant was dialyzed two times for 2 h periods
against 100 vol dialysis buffer (25 mM HEPES pH 7.6, 2 mM benzamidine, 100 mM KCl, 1 mM
dithiothreitol and 10% glycerol). The nuclear proteins were then fractionated at room temperature by FPLC on 10 ml Heparin-Sepharose columns by step elution gradients (0.1-1 M KCl) (
17
). Individual fractions were concentrated by means of Centriprep-10 concentrators from Amicon. Aliquots of the 0.5 M KCl fraction, containing
MDBP-2-H1, were stored frozen at -80oC. The protein concentration was determined according to
Bradford (
19
) using bovine serum albumin as a standard.
Experiments were carried out essentially as described by Pawlac
et al
. (
14
). Sonicated
E.coli
DNA (1 [mu]g) was used as non-specific competing DNA. Double-stranded oligonucleotides used for these experiments were end
labelled with the polynucleotide kinase reaction or with the Klenow fragment of
DNA polymerase I (Boehringer Mannheim).
The fraction containing MDBP-2-H1 was mixed in a 25 [mu]l reaction with either calf intestinal phosphatase attached to
beaded agarose (CIP-agarose) (0.1 U/[mu]g protein) or agarose beads only and stirred gently at 37oC for 30 min. The protein solution was separated from the CIP-agarose or agarose by centrifugation at 10 000
g
for 10 min at 4oC. The protein concentration was determined according to Bradford (
19
) using bovine serum albumin as a standard.
For the construction of the plasmid template (pVT 13) used for
in vitro
transcription, a 126 bp
Eco
RI-
Hin
dIII fragment of the avian vitellogenin II gene, containing 73 bp of the 5' flanking region and the first 52 bp of the first exon was fused to a 144
bp
Hin
dIII-
Cla
I fragment of the cauliflower mosaic virus (CaMV) ORFII protein gene and cloned
into pBR322 (Fig.
1
). Template DNA (MetpVT13) was methylated with CpG methylase (M.
Sss
I) according to the manufacturer's instructions. After 2 h incubation at 37oC, the DNA was extracted twice with phenol, once with chloroform, precipitated with 1/10 vol 3 M sodium acetate (pH 7.5) and 3 vol ethanol and the pellet was resuspended in 1* TE (10 mM Tris-HCl pH 7.5, 1 mM EDTA). The extent of methylation was tested
with
Hpa
II endonuclease and analyzed on 1% agarose gel.
Oligonucleotides were synthesized with an Applied Biosystems oligonucleotide
synthesizer and gel purified. Equal moles of complementary strands were heated to 90oC for 1 min and annealed by slow cooling to room temperature. The sequence
of the methylated oligonucleotide from the avian vitellogenin II gene is 5'-TTCACCTTmCGCTATGAGGGGGATCATACTGGCA-3' (nucleotide positions +2 to +34) (
16
).
A reaction mixture (final volume 25 [mu]l) contained 19 [mu]g HeLa nuclear extracts, 400 ng supercoiled DNA template in 9 mM HEPES,
pH 7.9, 12 mM KCl, 3 mM MgCl
2
, 0.09 mM EDTA, 0.22 mM dithiothreitol, 9% (vol/vol) glycerol and 0.03 mM phenylmethylsulfonyl fluoride. The negative control contained 2 [mu]g/ml [alpha]-amanitin. The reaction mixture was preincubated at 4oC for 10 min, after which transcription was initiated by
the addition of 400 [mu]M of each ribonucleoside triphosphates and incubation was continued at 30oC for 60 min. The reaction mixture was then treated with proteinase K
(300 [mu]g/ml). The volume was brought up to 300 [mu]l with 1* TE (pH 7.5) and samples were extracted twice with phenol, once
with chloroform and ethanol precipitated. A 15 min DNase I (70 U/ml) treatment followed according to the manufacturer's instructions (Boehringer Mannheim), and the RNA was purified by phenol/chloroform extractions and
ethanol precipitated. For detection of the CaMV transcript from
vitellogenin II
promoter, RNA was heated to 75oC for 10 min in 20 [mu]l of hybridization solution (0.4 M NaCl, 10 mM Tris-HCl pH 7.5 and 1 mM EDTA) and hybridized overnight at 40oC with 1 ng
32
P end-labeled oligonucleotide (2.5 * 10
8
c.p.m./[mu]g) 5'-TCTAAAAGGGATTTTACTT-CCTTTAGTTGGCTCGAAATCCG-3'. Detection of specific transcript was
carried out by S1 nuclease protection assay (
20
). Hybridized sample was digested for 60 min at 37oC in 300 [mu]l S1 nuclease buffer (0.3 M NaCl, 60 mM sodium acetate pH 4.5, 4 mM
ZnSO
4
and 7.5 [mu]g salmon sperm DNA) with 660 U/ml S1 nuclease. Following ethanol precipitation
the transcription product was separated on a 12% (w/v) polyacrylamide-4 M urea denaturating gel. Radioactive bands were
visualized by using a PhosphorImager and IMAGEQUANT software (Molecular Dynamics). Each
in vitro
transcription experiment was repeated three times to confirm the
reproducibility of the results.
The presence of the phosphorylated form of MDBP-2-H1 in the rooster liver, binding selectively to methylated DNA, has
been previously documented by
in vitro
DNA binding studies (
14
-
16
). To study the mode of action of this liver protein more closely, a
heterologous
in vitro
transcription system consisting in HeLa nuclear extracts and a DNA template
with vitellogenin II promoter region was optimised. Figure
1
A shows the plasmid construct (pVT13) used for the
in vitro
transcription. The details of the construction are given in the methods
section. Figure
1
B, lane 1 shows that in the absence of the repressor MDBP-2-H1, newly synthesized transcripts (TP) were detected. The plasmid
without the
vitellogenin
II
promoter region gave no transcription (data not shown) whereas the template
with
vitellogenin II
promoter region gave a specific transcription product TP (Fig.
1
B, lane 1). The transcripts were generated by RNA polymerase II as demonstrated by their sensitivity to a low concentration of [alpha]-amanitin (Fig.
1
B, lane 2). To determine the effect of methylation on the transcription from the
vitellogenin
II
promoter, two CpG sites of the promoter were methylated with the CpG methylase
Sss
I (Fig.
1
A). Transcription was carried out
in vitro
with the HeLa transcription system which does not contain any MDBP-2-H1. The absence of MDBP-2-H1 in HeLa nuclear extracts was demonstrated by gel
shift assays and by immunoblot analysis with calf thymus histone H1 antibodies
(data not shown). Template DNA and nuclear extracts were added at zero time and upon 10 min
incubation at 4oC, rNTPs were added to the incubation mixture and the incubation was
continued for 60 min at 30oC. Figure
2
shows the effect of increasing concentrations of methylated and non-methylated templates on the transcription. Under our experimental
conditions, both the methylated and unmethylated DNA templates gave a
comparable level of transcription. These results indicate that in the absence
of the phosphorylated form of MDBP-2-H1 there is, in the
in vitro
transcription system, no difference in the level of transcription from methylated or unmethylated DNA templates, therefore, in this case there is no direct
inhibitory effect of methylated DNA on the binding of transcription factors.
In recent years evidence has accumulated that histone H1, the internucleosomal
histone of higher eukaryotes, acts
in vitro
and
in vivo
as a transcriptional repressor of specific genes (
21
-
24
). Histone H1 has been shown to be enriched in inactive chromatin (
25
) and associated with nucleosomes containing 5 methyl-cytosine (
26
). Conversely, it is depleted in active genes (
27
) and is absent in CpG-rich islands from active housekeeping genes (
28
). Several recent reports have established a correlation between DNA
methylation, binding of histone H1 and repression of gene transcription (
8
,
9
). In one case, Johnson
et al.
(
9
) showed that the histone variant H1c had a preferential inhibition of
transcription from methylated DNA. Our results show that MDBP-2-H1, also a member of the histone H1 family, selectively inhibits
in vitro
transcription from the methylated
vitellogenin II
promoter. In the absence of this H1 variant, methylation of the promoter region
does not affect the transcription. A gel mobility shift assay carried out with
methylated and unmethylated labeled oligonucleotides could not demonstrate the
presence of a specific methylated DNA binding protein in total nuclear extracts
prepared from HeLa cells and in the 0.5 M KCl fraction eluted from Heparin-Sepharose column (data not shown). These finding demonstrate that DNA methylation
confers its inhibitory effect on
vitellogenin II
transcription indirectly by a mechanism that involves MDBP-2-H1. However, we do not know how the interaction of MDBP-2-H1 with methylated DNA affects gene transcription. We
propose that the binding of MDBP-2-H1 to several sites in the methylated promoter of
vitellogenin II
(for example nucleotide positions +2 to +34) could lead to a steric occlusion
of the basic transcriptional machinery and/or of transcription factors from their binding sites on the DNA.
Since histone H1 plays a crucial role in the chromatin structure (
29
), it is reasonable to assume that DNA methylation and subsequent binding of
MDBP-2-H1 to
vitellogenin II
promoter favor the formation of heterochromatin. The relation between DNA methylation,
chromatin structure and the gene activity could be examined by
in vitro
transcription of chromatin reconstituted in the absence or presence of MDBP-2-H1 (
30
). Furthermore, previous work from our laboratory has shown that MDBP-2-H1 mediates the binding of another nuclear protein that does not
itself bind to methylated DNA (
15
,
31
). The identity and the role of this protein in the transcriptional regulation
of
vitellogenin II
,
however,
remain to be determined.
The histone H1 family consists of several different somatic variants which are
present in different species, cell types and developmental stages (
32
). This genetic microheterogeneity is increased by several reversible post-translational modifications (
33
) that can influence H1-DNA or H1-H1 interactions which modulate the chromatin structure. We do not
know to which variants MDBP-2-H1 belongs but we do know that this protein is phosphorylated at
several residues in the nuclei of rooster liver and that the phosphorylation of
MDBP-2-H1 is essential
in vitro
for its binding to methylated DNA (
17
). Phosphorylation of histone H1 is a complex modification and the literature is
undecided about its role in the chromatin condensation/decondensation (
34
-
36
). Our present results show that the phosphorylation of MDBP-2-H1 is essential for the repression of
vitellogenin
II
transcription. We suggest that in our case, DNA methylation and phosphorylated
MDBP-2-H1 binding to methylated DNA are two major factors involved in the
chromatin condensation and the subsequent repression of
vitellogenin
II
transcription.
Treatment of adult rooster with estradiol triggers a partial dephosphorylation
of MDBP-2-H1 (
17
), and a sharp decrease of its binding activity to a methylated sequence
(nucleotide positions +2 to 34) of the
vitellogenin
II
promoter (
16
). The data of the present study demonstrate that these changes result in the
loss of the inhibitory activity of MDBP-2-H1 on the
vitellogenin
II
transcription. In addition, we have previously observed dramatic estradiol-induced changes in the chromatin structure of this region (
37
,
38
). These parameters combined with a gradual demethylation of the CpG pair of the
MDBP-2-H1 binding site (
16
) contribute to the maximum stimulation of
vitellogenin II
transcription. Taken together these results establish clearly the relationship
between the dephosphorylation of MDBP-2-H1 caused by estradiol, the down regulation of its binding to
methylated DNA, the modification of the chromatin structure and the derepression of
vitellogenin II
transcription. However, the mechanism by which estradiol induces a partial
dephosphorylation of MDBP-2-H1 is still unknown. We consider two models that are not mutually
exclusive and that could explain the dephosphorylation process. (i) Estradiol could induce directly
or indirectly a specific histone H1 phosphatase. (ii) The binding of the
estradiol-receptor complex to the
vitellogenin II
promoter could disrupt the chromatin structure and expose MDBP-2-H1 to a specific histone H1 phosphatase already present in the
nuclei. However, the nature and the sequence of these specific events involved
in the dephosphorylation of MDBP-2-H1 under estadiol control, remain to be established. Furthermore the
entire sequence of MDBP-2-H1 needs to be determined since recently it has been conclusively
shown that the total population of histone H1 does not have any preferential
binding to methylated DNA (
39
; J.P.J., unpublished data) and that the methylation of CpG sequences does not
influence the binding of total histone H1 to nucleosomes (
40
).
We are grateful to Michel Siegmann for the technical help in preparation of MDBP-2-H1 from rooster liver nuclear extracts and to Drs Jerzy Paszkowski,
Edward Oakeley and Jean Louis Couderc for critically reading the manuscript.
+
Present address: Unité de Nutrition Cellulaire et Moléculaire, INRA de Theix, 63122 Saint Genès Champanelle, France
REFERENCES
Return