 |
INTRODUCTION |
Development is a multistep process involving interactions of a
large number of trans-acting factors with specific
cis-regulatory elements leading to the activation and
repression of many genes. Oct-3/4 is the earliest expressed
transcription factor that is known to be crucial in murine
pre-implantation development (1). The Oct-3/4
gene is a member of the POU family of transcription factors; it is
expressed in embryonal stem
(ES)1 and in embryonal
carcinoma (EC) cells (1, 2). Oct-3/4 expression is down-regulated in
these cells upon induction to differentiate with retinoic acid (RA)
(3-8). The Oct-3/4 gene is expressed throughout
the pre-implantation embryo (6, 9). Oct-3/4 protein is present in
unfertilized oocytes, and the zygotic expression is activated prior to
the 8-cell stage (9-11). In the embryo Oct-3/4 is abundantly and
uniformly expressed in all cells through the morula stage (11). Its
expression is down-regulated in trophectoderm cells and becomes
restricted to cells of the inner cell mass in the blastocyst (9).
Oct-3/4 expression is high and persists through day 7.5 in the
unsegmented presomitic mesoderm, decreasing anteriorly to posteriorly
as the somatic lineages form (7). From day 8.5, expression cannot be
detected in any somatic tissue but is restricted to the premigratory
progenitor germ cells (9, 12).
Recently, using conventional gene targeting technology, it was found
that the activity of Oct-3/4 is essential for the identity of the
pluripotential founder cell population in the inner cell mass (13).
Furthermore, it was shown that a critical amount of Oct-3/4 is required
to sustain stem cell self-renewal, and any up- or down-regulation
induces divergent developmental programs (14). A very modest modulation
of Oct-3/4 levels in ES cells results in a dramatic readout. Thus, it
is now recognized that the mere presence of Oct-3/4 protein does not
define pluripotency because it is the level of expression that is the
key to regulation.
One of the molecular mechanisms that regulate gene expression is
methylation (15). DNA methylation plays a crucial role in normal
embryogenesis by acting in cis to modulate protein-DNA interactions. Several studies have demonstrated that DNA methylation patterns go through dynamic changes during embryogenesis. A global loss
of DNA methylation occurs in the mammalian preimplantation embryo prior
to the 16-cell morula stage, and the DNA remains unmethylated
throughout blastulation (16, 17). Following implantation, an extensive
wave of de novo methylation occurs, and only genes containing CpG islands, such as housekeeping genes, escape this process
(16, 18, 19). At a later stage of development, tissue-specific genes
undergo selective demethylation (20, 21).
Because Oct-3/4 plays a crucial role in regulating initial
differentiation decisions in early development and has a unique expression pattern, we studied the dynamics of the generation of its
methylation pattern and the role of methylation in controlling Oct-3/4 expression. We have shown previously (22)
that treatment of EC cells with RA reduces
Oct-3/4 expression that is associated with
increased methylation and changes in chromatin structure in the
immediate upstream regulatory region that includes the promoter (P) and
proximal enhancer (PE) elements. In this study we show that in
vivo the Oct-3/4 gene is unmethylated from
the blastula stage and starts to undergo de novo methylation
at 6.5 dpc. The Oct-3/4 gene remains modified in
all adult somatic organs analyzed. We show that during embryogenesis
Oct-3/4 remains unmethylated at a time when
genome wide methylation occurs; thus it was interesting to find out
whether it carries a cis-demethylation element that protects
the gene from global modification mechanisms. By using a series of
stable transfection assays into EC cells, we have found
that the Oct-3/4 gene does harbor a
cis-specific demodification element that includes the
Oct-3/4 PE sequence. Furthermore, mutagenesis at
the protein-binding sites of this sequence clearly affected its
demethylation activity. By using a mini-transgene containing the PE
sequences only, we have shown that in the embryo the PE protects the
DNA from de novo methylation in 6.25 dpc epiblast cells, and
when it is mutated the sequence becomes methylated. In vitro
methylation of the Oct-3/4 upstream sequence
inhibits its ability to induce transcription. Moreover, we have shown
that DNA methylation inhibits the endogenous
Oct-3/4 basal expression in vivo.
| |
EXPERIMENTAL PROCEDURES |
Cells--
Murine P19 (23) EC cells and L8 cells
were maintained as described previously (8). p53
/ and
p53/ Dnmt1/ fibroblasts were grown from
9 dpc embryos obtained by mating p53/
Dnmt1/+ mice. These cells (grown in Dulbecco's modified
Eagle's medium containing 15% fetal calf serum) appear to be
immortal, still not having undergone senescence even after 100 passages. Nearest neighbor analysis (24) revealed that less than 5% of
CpG residues in these cells are methylated, as compared with 73% for
normal fibroblasts. These cells were treated with 50 ng/ml TSA for
24 h in order to study Oct-3/4 transcription.
Plasmids and Oligonucleotides--
The constructs 342 and
pOct-luc have been described previously (see Refs. 25 and 26,
respectively). pOctPE-luc was constructed by cloning the PCR
amplification product spanning positions 
1215 to +155 of the
Oct-3/4 upstream region containing the PE and P sequences into the PstI site of pOct-luc. The pre-existing P
was deleted by digestion with HindIII- and the plasmid was
religated. The primers used were Nsi 2 and Nsi 3.
The following plasmids were constructed by digesting the
different constructs with NsiI and subcloning them into the
PstI site in plasmid 342.
pPEwt was constructed by cloning the PCR amplification product spanning
positions 1215 to 900 of the Oct-3/4 PE
region. The primers used were Nsi 2 and Nsi 1.
pPEPwt was constructed by cloning the PCR amplification product
spanning the 1215 to +155 positions of the
Oct-3/4 PE and P region. The primers used were
Nsi 2 and Nsi 3.
pPE1A* construct is identical to the above described pPEwt plasmid
except for a 16-bp mutation inserted in the 1A site. Two distinct PCR
amplification reactions were done on pBluescript KS() containing
the 1215 to 541 BamHI-SauIIIA upstream
region of the Oct-3/4 gene. The primers used were
RARE1A and Primer 2 and RARE1Arev and Primer 1. Products were mixed,
amplified with Primers 1 and 2, digested with
BamHI/ScaI, and subcloned into BamHI/EcoRV sites in pBluescript KS() to create
pmRARE1A. pPE1A* was constructed by cloning the PCR amplification
product of pmRARE1A using the primers Nsi 1 and Nsi 2 into the
PstI site in 342.
pPE1B* was constructed by cloning the PCR amplification product
spanning positions 1215 to 942 of the Oct-3/4 PE region, harboring a deletion of the 1B site. The primers used were Nsi 2 and
Nsi 4.
pPE1A*1B* construct is identical to the above described pPE1B* plasmid
except for a 16-bp mutation inserted in the 1A site. pPE1A*1B* was
constructed by cloning the PCR amplification product of pmRARE1A using
the primers Nsi 2 and Nsi 4.
pPEP1A*1B* construct is identical to the above described pPEPwt plasmid
except for a 16-bp mutation inserted in the 1A site and a 28-bp
deletion of the 1B site. pmRARE1A was subjected to PCR amplification
with Primer 2 and Primer RARE1Brev; the product was digested with
BamHI/ScaI and ligated to
ScaI/BamHI fragment from pPEPwt. The ligation
fragment was subjected to PCR amplification using the primers Nsi 2 and
Nsi 3 and subcloned into the PstI site in 342 to create
pPEP1A*1B*. Primers and PCR amplification conditions are described in
Table I.
View this table:
[in this window]
[in a new window]
|
Table I
Primers for plasmid constructions
All PCR amplification reactions were carried out as follows: 2 min at
94 °C; 24 cycles of 1 min at 94 °C, 30 s at 60 °C, and
30 s at 72 °C, and finally 3 min at 72 °C. All amplified
fragments were checked by sequencing.
|
|
Methylation in Vitro and Transfection--
Plasmid DNAs were
methylated in vitro as described previously (27). P19 and L8
cells were transfected by the calcium phosphate precipitation method
(28). Cells (5 × 105) were plated 24 h before
transfection and cotransfected with 15 µg of methylated plasmid
together with 1.5 µg of a plasmid carrying the neo
selection gene. Stably transfected cells were selected by G418 (250 mg
per ml, Invitrogen), and clonally propagated.
Preparation of Preimplantation and Post-implantation
Embryos--
Pre-implantation embryos were obtained as described
previously (29). The post-implantation embryos were dissected from the uterus of the foster mothers at dpc 6.25, 6.5, 7.5, 8.5, 12.5, or
14.5.
PCR Methylation Analysis--
Genomic DNAs were extracted from
the following groups: 1) a pool of 10-20 preimplantation blastocysts;
2) a pool of epiblast cells dissected from 7 embryos of 6.25 dpc; 3)
pools of post-implantation embryos at stages 6.5, 7.5, 8.5, 12.5, and
14.5 dpc.
DNA samples were divided into 3 aliquots that were
subjected to digestion with PvuII alone and PvuII
together with HpaII or HhaI. The PCRs were
carried out with 5 ng of DNA using the following primers: for analyzing
HpaII sites 1-4 primers Hp1a and Hp1b, Hp2a and Hp2b, Hp3a
and Hp3b, Hp4a and Hp4b were used, respectively. In order to analyze
the HhaI site, primers Hp3a and Hp4b were used. To assay the
J
region that lacks HpaII or HhaI sites, we used 5GL-2 and J1110 primers. Primers and PCR amplification
conditions are described in Table II.
View this table:
[in this window]
[in a new window]
|
Table II
Primers for methylation analysis and transgenic mice identification
All PCR amplification reactions were carried out as follows: 2 min at
95 °C; 29 cycles of 45 s at 95 °C, 30 s at 56 °C,
and 45 s at 72 °C, and finally 5 min at 72 °C. The PCR
products were electrophoresed on 1-3% agarose gels, stained with
ethidium bromide, and photographed.
|
|
HpaII and HhaI sites in the first intron of the
apoAI gene were studied using "fragment I" primers as described
previously (16).
Whole Cell Extracts (WCEs)--
WCEs and electrophoretic
mobility shift assays (EMSAs) were done as described previously (30).
1Awt and mut1A oligonucleotides (described in Table
III) were used for EMSAs.
Luciferase Analysis--
The differentially methylated pOct-luc
or pOctPE-luc constructs were transiently cotransfected with the
pCMV-Renilla luciferase construct into P19 EC cells using
the calcium phosphate precipitation method (28). Luciferase activity
was determined 48 h post-transfection (Dual-luciferase Reporter
Assay system kit, Promega) and was corrected for transfection
efficiency by measuring the Renilla luciferase activity.
RT-PCR--
Total RNA was collected from the cells with Trizol
(Sigma) following the manufacturer's instructions. Reverse
transcription was carried out following the manufacturer's protocol
(Promega). RT-PCR analysis of Oct-3/4 transcription in
p53/, p53//Dnmt1/, and
blastocyst cells were carried out using the primers 251L and Oct4-a.
PCR was performed using 200 mM of the four dNTPs, 100 ng of
each primer, and 0.1 µl of [32P]dCTP. A second
half-nested PCR was carried out in the same reaction conditions using
the following primers: Oct4-5' and Oct4-a. One-third of the reaction
products was subjected to electrophoresis on a 7.2% polyacrylamide
gel. Primers and PCR amplification conditions are described in Table
IV.
View this table:
[in this window]
[in a new window]
|
Table IV
Primers for RT-PCR analysis
Amplification was performed as follows: 4 min at 94 °C, 40 cycles of
30 s at 94 °C, 30 s at 62 °C, and 1 min at 72 °C,
and, finally, 5 min at 72 °C.
|
|
Transgenic Mice--
The 1141- and 1099-bp DraI
fragments, harboring the wt or double mutated Oct-3/4 PE
sequences, were isolated from pPEwt and pPE1A*1B* plasmids,
respectively. Fragments were purified from low melting point agarose
gels by a gel extraction kit (Qiagen). DNA samples were microinjected
into the pronucleus of (C57BL/6 × BALB/c)F1 fertilized mouse eggs
and transferred into pseudo-pregnant CB6/F1 foster mother. Transgenic
mice were identified by PCR amplification using primers
DraI-3' and Hp1b (Table II).
Methylation-dependent PCR Analysis of Transgenic
Mice--
Epiblast cells were isolated from 6 wt PEwt and 13 double
mutated PE1A*1B* 6.25 dpc transgenic embryos as described (31). DNA
from the wt and mutated pools of embryos was extracted, and 10 ng were
subjected to digestion with PvuII in the presence or absence
of HpaII. Products were analyzed by PCR amplification using
primers DraI-3' and Hp1b (Table II) flanking
HpaII site 1 in the Oct-3/4 PE. As a
control, we assayed the J region that lacks HpaII sites
using 5GL-2 and J1110 primers. Primers and PCR amplification
conditions are described in Table II.
| |
RESULTS |
The Generation of Oct-3/4 Methylation Pattern during
Embryogenesis and in the Adult--
Four HpaII and one
HhaI sites were analyzed in the
Oct-3/4 1370-bp upstream region (Fig.
1A). To study the
methylation status of the Oct-3/4 gene in early
mouse embryos, genomic DNA was isolated from mouse embryos at various
stages of development. We adopted a PCR-based assay in which genomic
DNA was subjected to digestion with or without HpaII or
HhaI methyl-sensitive restriction enzymes. Digestion
conditions were carefully titrated. After digestion, oligonucleotide
primers flanking each site were used. A PCR product is observed only
when the site is methylated and refractory to digestion. It was shown
previously (16) that when properly calibrated, this assay is linear
over a wide range of DNA concentrations and can be used to measure
accurately the degree of DNA methylation at specific sites. By using
this PCR assay, we show that the HhaI and the four
HpaII sites tested are clearly unmethylated in blastocysts and in epiblast cells isolated from 6.25 dpc embryos. These sites become methylated in 6.5 dpc embryos almost to the full extent (Fig.
1B). Control experiments were included to validate our
results. Tail DNA and DNA from P19 cells served as controls for
methylated and unmethylated DNA, respectively. Moreover, as an internal
control, a primer pair that amplifies a J sequence that does not
encompass either HpaII or HhaI sites was included
in each reaction. As expected, the fragment was amplified from all DNA
samples to a similar level (Fig. 1B). We have quantified the
methylation status of each site by PhosphorImager analysis and found
that different sites undergo modification to various levels, and for
most of these sites the level of methylation is sustained throughout
embryogenesis (Fig. 1D). Similar results were obtained from
quantifications of Southern blots containing DNA isolated from 6.5, 8.5, 12.5, and 14.5 dpc embryos, P19 cells, and murine tails (data not
shown).
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 1.
In vivo methylation pattern
of the Oct-3/4 gene throughout embryogenesis and in
the adult. A, map of the 5' 1354-bp regulatory region
of the Oct-3/4 gene. The arrow
indicates the initiation of transcription. Selected restriction enzymes
sites are shown (X, XbaI; B,
BamHI; Hh, HhaI, and
HpaII/MspI (Hp/M)), and
Hp/M restriction sites are numbered 1-4 and indicated by
arrowheads. The promoter (P) and proximal
enhancer (PE) regions are boxed. DNA samples that
were analyzed also by Southern blots were digested with
BamHI and XbaI in the presence or absence of
HpaII or HhaI and hybridized with two radioactive
probes, the 805-bp BamHI/XbaI fragment and the
542-bp XbaI/BamHI region. B, PCR
analysis of the Oct-3/4 regulatory region.
Genomic DNA was isolated from blastocysts, epiblasts of 6.25 dpc embryos, 6.5, 8.5, and 14.5 dpc embryos, and P19 EC
cells, and tail. DNA was digested with PvuII in
the presence or absence of the indicated enzymes (Hp,
HpaII; Hh, HhaI). Oligonucleotide
primers flanking HpaII sites 1-4 (rows 1-4,
respectively) and HhaI site (row Hh) of the
Oct-3/4 upstream regulatory element were used for
PCR amplification. As a control each DNA was amplified with primers
5SGL-2 and J1110 which amplify the J gene that do not encompass
either HpaII or HhaI sites. PCR products were
electrophoresed on 1-3% agarose gels, stained with ethidium bromide,
and photographed. C, PCR analysis of the apoAI gene.
Genomic DNA was isolated from blastocysts and epiblast of 6.25 dpc
embryos. Oligonucleotide primers flanking HpaII and
HhaI sites of the apoAI gene were used for PCR
amplification. Each sample was amplified with control primers
(described in B), and products were analyzed on agarose
gels. D, the methylation level of HpaII
sites 1-4 and the HhaI (Hh) site throughout
embryogenesis and in the adult mice were quantified by PhosphorImager
after subtracting the lane background signal and graphed. Percent
methylation for each site, y axis, and embryonic stage
(dpc), tail and kidney DNA, x axis, are shown. Every
experiment was repeated 3 times and means and error bars are
given. E, kidney and tail DNA were digested with
BamHI/XbaI in the presence or absence of
HpaII and HhaI restriction enzymes,
electrophoresed on a 1% agarose gel, blotted, and probed either with
the 805-bp BamHI/XbaI fragment or with the 549-bp
XbaI/BamHI fragment. The methylation level of
HpaII sites 1-4 and HhaI site were quantified by
PhosphorImager and graphed. Methylation sites, DNA source, and percent
methylation are indicated.
|
|
In order to investigate whether this pattern of modification serves as
the prototype for the basic stable methylation pattern in somatic
cells, we analyzed the methylation status of the
Oct-3/4 upstream regulatory region in adult
kidney and tail DNA. As can be seen in Fig. 1D, the cleavage
patterns of all sites analyzed of both DNA samples were roughly similar
to those found in 8.5-day embryos. Results from PhosphorImager analysis
of the corresponding Southern blots were graphed (Fig. 1E),
and it is interesting to note that the HhaI site located in
the P region and the juxtaposed HpaII site 3 are almost
fully methylated, serving as a core methylation center, whereas sites
flanking this center (HpaII sites 1, 2, and 4) are partially
methylated. These experiments clearly show that the methylation pattern
established during development faithfully propagates itself to adulthood.
The PE Fragment Directs EC-specific
Demethylation--
The apoA1 gene, similar to other tissue-specific
genes analyzed, undergoes de novo methylation due to the
onset of global methylation that occurs at post-implantation embryos
(16, 17). As can be seen in Fig. 1C, the apoA1 sequences are
methylated in the 6.25 dpc epiblast cells, at the time when
Oct-3/4 is still unmethylated. Thus, the
Oct-3/4 gene remains unmethylated at a time when
the apoA1 gene undergoes de novo methylation. One possible mechanism to explain the ability of the Oct-3/4
gene to escape the first de novo methylation wave is the
involvement of a cis-element that either actively
demethylates the newly methylated Oct-3/4 or
protects it from de novo methylation. At this stage of our research we cannot differentiate between these two alternatives, and
for simplicity reasons this element will be designated a
protection/demethylation or demodification element.
In the first series of experiments we wished to determine whether a
long (1370 bp) fragment from the upstream regulatory element of the
Oct-3/4 gene could bring about demethylation of
an in vitro methylated substrate when integrated into the
genome of P19 EC cells. This element contains the P,
intervening, and PE sequences. We inserted this fragment
into a vector (25) that was engineered to harbor a non-CpG island
sequence as well as recognition sites for methylation-sensitive
restriction endonucleases (designated pPEPwt, Figs. 1A and
2A). We generated stably transfected cell lines containing
the above described in vitro methylated (by HpaII and HhaI methylases) construct and examined the methylation
status of the insert as well as the vector sequences. To minimize the effect of particular integration sites and copy number, we analyzed more than 10 independent clones for each of the constructs tested and
chose those with a similar copy number. The sites that were analyzed
were the same HpaII and HhaI sites that were
analyzed in the endogenous Oct-3/4 gene (Fig. 1)
and an extra HhaI site in the vector sequence (Fig.
2A). All sites analyzed were
undermethylated (Fig. 2B, lanes 2 and 3),
indicating that this fragment harbors a demodification element that is
able to demethylate sequences inside and outside the fragment.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 2.
The PE is an EC cis-specific
demethylation element. A, pPEwt and pPEPwt maps. The
pPEwt and pPEPwt constructs were obtained by sub-cloning the
Oct-3/4 PE region spanning positions 1215 to 900
and the Oct-3/4 PE and P regions spanning
positions 1215 to +155, respectively, into the PstI site
of plasmid 342 (25). Selected restriction sites, EcoRI,
DraI, and HindIII, present in the vector sequence
are shown. Black arrowheads indicate
HpaII/MspI restriction sites, and the gray
arrowheads indicate HhaI sites present in the vector
and in the PEP insert sequences. P and PE regions are boxed.
The 826-bp DraI fragment (without any insert) was
used as a radioactive probe. B, P19 EC cells were
separately transfected with methylated constructs, pPEwt and pPEPwt
emphasized by the diagram. DNA was digested with DraI in the
presence or absence of the indicated enzymes (Hp,
HpaII; Hh, HhaI), electrophoresed on a
1% agarose gel, blotted, and probed with the 826-bp DraI
fragment (containing no Oct-3/4
insert). C, L8 fibroblast cells were
transfected with methylated pPEwt construct. DNA was digested with
DraI in the presence or absence of the indicated enzymes
(Hp, HpaII; M, MspI;
Hh, HhaI), electrophoresed on a 1% agarose gel,
blotted, and probed with the 826-bp DraI fragment.
|
|
Next we analyzed the ability of the P alone, the proximal enhancer (PE)
alone, and the P and PE elements without the intervening sequences to
induce demethylation. It was clear that both the P and the intervening
sequences are dispensable for the demethylation process (data not
shown). In striking contrast, we found that the PE fragment (pPEwt)
plays a key role in directing the demethylation reaction (Fig.
2B, compare lanes 4 and 5). This PE
was previously found to be a critical element for P19 EC cell-specific
expression and for RA-mediated down-regulation of the Oct-3/4 protein
(1, 2).
We wished to determine whether the Oct-3/4
demodification element is specific for EC cells or alternatively
functions in other cells as well. We transfected in vitro
methylated pPEwt into L8 fibroblast cells and performed Southern blot
analysis. This analysis showed that the integrated construct was
resistant to HpaII and HhaI digestion and
therefore maintained methylation at these sites (Fig. 2C,
compare lane 1 to lanes 3 and 4).
Taken together, these experiments clearly indicate that the
Oct-3/4 PE is required for directing
demodification and for mediating demethylation of the Oct-3/4 gene in a cell-specific manner.
The 1A Site Is Crucial for the Demethylation Activity--
This
cell specificity encouraged us to delineate the demodification element
responsible for Oct-3/4 demethylation in EC
cells. It was shown previously (3, 5) that the PE contains two binding
sites, 1A and 1B, which bind distinct factors in vitro, but
only site 1A, which contains a conserved Sp1-like binding site, is
occupied in vivo in undifferentiated ES and EC cells. To
define whether the 1A and 1B sites are required for the demethylation process, the 1A site was mutated, and the 1B sequence was deleted in
the context of both, the PE only (designated pPE1A*1B*), and the PE
together with the P (designated pPEP1A*1B*). The pPEP1A*1B* was
generated in order to analyze the ability of the mutated PE to induce
demethylation not only upon vector sequences, which can be analyzed
using pPE1A*1B*, but also upon HpaII and HhaI sites located in the Oct-3/4 P region. As can
clearly be seen in Fig. 3A,
mutating the 1A site and deleting the 1B site abolished almost
completely the ability of the PE to induce demodification of the vector
and P sequences. These results strongly suggest that either 1A, 1B, or
both sites play a critical role in the demethylation process and are
most likely responsible for its cell type and developmental stage
specificity.
View larger version (39K):
[in this window]
[in a new window]
|
Fig. 3.
The 1A site in the PE is crucial for its
demethylation activity. A and B, P19 EC
cells were transfected with the following methylated plasmids
separately: pPE1A*1B* and pPEP1A*1B* (A), pPE1A* and pPE1B*
(B), as emphasized by the diagram. DNA was digested with
DraI in the presence or absence of the indicated enzymes
(Hp, HpaII; Hh, HhaI),
electrophoresed on a 1% agarose gel, blotted, and probed with the
826-bp DraI fragment. The faint bands observed are either
from a cross-reaction with endogenous sequences (A,
lanes 1-4), or due to less than 10% demethylation
(A, lanes 2, 3, 5, and
6).
|
|
To assess the roles of these two individual motifs in the process of
demethylation, we designed two additional mutants that contained either
mutation in the 1A sequence (designated pPE1A*) or a deletion of the 1B
(designated pPE1B*). These mutations were only subcloned in the context
of the PE because our previous results showed that all the
HpaII and HhaI sites present in the pPEPwt construct were similarly modified, suggesting that the demodification element works in a regional manner. In vitro methylated
pPE1A* and pPE1B* DNA were stably transfected into P19 cells, and DNA from more than 20 independent transfected clones were analyzed. This
analysis revealed that deletion of the 1B sequence (leaving the 1A site
intact) did not alter the ability of the demethylation element to
demodify the Oct-3/4 HpaII site 1, and
reduced its ability to demethylate the HhaI site located in
the vector backbone (Fig. 3B, lanes 4-6).
Strikingly, mutations in the 1A site were sufficient to abolish almost
completely the ability of the Oct-3/4 PE element
to induce demodification (Fig. 3B, lanes 1-3).
These data strongly suggest that the 1A site is a crucial part of the Oct-3/4 demodification element. Furthermore, it
is clear that the 1B site is not sufficient to bring about
demethylation. Taken together, the above results define a 270-bp
segment (that does not contain the 1B site) from the
Oct-3/4 PE as an embryonal specific demethylation element.
Because we have shown that demethylation of the
Oct-3/4 gene is mostly guided by the 1A site, we
attempted to find whether the above described mutations in this site
result in loss of the PE binding activity. We carried out EMSAs using
P19 cell extracts. Binding of this extract to the labeled 1Awt 32-bp
(harboring the protected 1Awt region (3, 5)) probe generated three
specific complexes (Fig. 4, A-C,
lane 1) as follows: two prominent complexes, numbers 1 and 3 (the
ratio between these two complexes changes in different experiments),
and one faint complex (number 2). All three complexes were specifically
competed by a 50-fold molar excess of a cold 1Awt
oligonucleotide (Fig. 4A, lane 2, and
B, lane 4) but not by an unrelated
oligonucleotide (Fig. 4A, lane 3). To test
whether the Sp1 transcription factor binds the PE1A GC-rich Sp1-like
binding site located at the PE1A, we performed a supershift experiment
with antibodies directed against Sp1, and we found that indeed these
antibodies supershifted a major part of complex 1 and not the other
complexes (Fig. 4A, lane 5). These results
indicate that one of the proteins in complex 1 is the Sp1 transcription
factor.
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 4.
Mutations in the 1Awt site abolish PE1A
binding activity. A, the 32P-labeled 1Awt
oligonucleotide was incubated with WCEs prepared from P19 EC cells
(lanes 1-5). Binding reactions were performed in the
absence of competitors (lane 1) or in the presence of a
50-fold molar excess of unlabeled 1Awt (lane 2), the
nonspecific octa oligonucleotides (lane 3), or in the
presence of 1 µg of preimmune serum (P.I., lane 4) and 1 µg of anti-Sp1 antibody (lane 5). Specific DNA-protein
complexes are numbered 1-3. The supershifted band is
indicated by an asterisk. The free DNA (F)
migrated to the bottom of the gel. B, the
32P-labeled 1Awt oligonucleotide was incubated with WCEs
prepared from P19 EC cells (lanes 1-10). Binding reactions
were performed in the absence of competitors (comp.)
(lanes 1 and 6) or in the presence of 5-100-fold
molar excess of unlabeled 1Awt (lanes 2-5) and mut1A
(lanes 7-10) oligonucleotides. C, the
32P-labeled 1Awt (lane 1) and mut1A (lanes
2 and 3) oligonucleotides were incubated with WCE
prepared from P19 EC cells. Binding reactions were performed in the
absence of antibodies (lanes 1 and 2), or in the
presence of 1 µg of anti-Sp1 antibody (lane 3).
|
|
To determine whether the mutations in the PE1A, which abolished the
demethylation activity, also abolish protein binding to the 1A
sequence, we conducted EMSA with the 1Awt as the labeled probe, and we
compared the ability of cold 1Awt and mut1A (that harbors the same
mutations as in pPE1A*) oligonucleotides to compete for specific
complex formation. All three complexes were almost completely inhibited
by a 20-fold molar excess of the cold 1Awt oligonucleotide (Fig.
4B, lane 3). In contrast, oligonucleotide mut1A
did not compete at all (lanes 7-10). Moreover, binding of P19 extracts to the labeled mut1A oligonucleotide resulted in either
faint or nonspecific complexes (Fig. 4C, lanes 2 and 3). Furthermore, mutations in PE1A completely abolished
Sp1 binding to the DNA as seen by the inability of antibodies directed
against Sp1 to supershift the complex (Fig. 4C, lane
3).
The PE Fragment Protects Itself from de Novo Methylation in
Vivo--
On the basis of our above described data showing that the
Oct-3/4 gene is unmethylated at 6.25 dpc embryo,
and identifying the Oct-3/4 PE as an
embryonal-specific demethylation element, we generated transgenic
mice in order to test whether the PE element is also a major
determinant in protecting the Oct-3/4 gene from de novo methylation in vivo. To this end, we
generated two kinds of transgenic embryos. One contains the wt PE
element and the other harbors the mutated PE1A*1B*, which is mutated in
the 1A site and deleted at the 1B site. Epiblast cells were isolated from six embryos of 6.25 dpc (31), and the methylation status of
HpaII site 1, located inside the PE, was determined
by methylation-dependent PCR analysis that was repeated 3 times. In agreement with the EC transfection experiments (Fig. 3), the
wt PE element stays unmethylated (Fig.
5), indicating that it is able to protect
itself from de novo methylation occurring in these cells at
this time in development. Interestingly, the mutated PE element
underwent substantial de novo methylation in the 13 analyzed
6.25 dpc embryos, indicating that an intact element is required to
prevent local de novo methylation in vivo. Thus,
the same mutations, located in the Sp1-like binding site, that destroy
the PE demethylation and binding activities in EC cells also eliminate
the ability of this sequence to protect itself from de novo
methylation in the post-implantation embryo.
View larger version (38K):
[in this window]
[in a new window]
|
Fig. 5.
PCR analysis of the Oct-3/4
PE in transgenic mice. Six PEwt and 13 double-mutated
PE1A*1B* 6.25 dpc transgenic embryos were collected. Epiblast cells
were isolated, and DNA was extracted and subjected to digestion with
PvuII in the presence (+) or absence () of
HpaII (Hp). Oligonucleotide primers flanking
HpaII site 1 were used for PCR amplification. As a control,
each DNA was amplified with primers 5SGL-2 and J1110 which amplify
the J gene that does not encompass HpaII sites. PCR
products were electrophoresed on 2% agarose gels, stained with
ethidium bromide, and photographed. This analysis was repeated 3 times,
and a representative experiment is shown.
|
|
Oct-3/4 Methylation and Transcription--
The
Oct-3/4 gene undergoes de novo
methylation both during embryogenesis (described above) and also in EC
cells as a consequence of RA treatment (22). To find out if methylation
plays a significant role in the regulation of
Oct-3/4 gene expression, two kinds of experiments
were performed. The pOct-luc plasmid described previously (26), containing the Oct-3/4 P region from
position 413 (XbaI site) to +48
(BanI), upstream from the luciferase reporter gene, as well as the newly generated pOctPE-luc harboring the P and the PE
elements (Fig. 6A) were
methylated in vitro, using either HpaII or
HhaI methylases. Subsequently, the unmethylated and
methylated plasmids were introduced transiently into P19 cells, and the
luciferase activity was determined. Both plasmids behaved similarly to
each other. As shown previously (8, 26, 30), the unmethylated plasmids
drive efficient transcription when transiently transfected into P19
cells. Transfectants containing the HhaI-methylated plasmids show a slightly reduced level of luciferase activity. Interestingly, HpaII methylation of the Oct-3/4
regulatory elements (either of the P alone or the P together with the
PE) reduced luciferase activity to about 30%, relative to the activity
of the unmethylated constructs (Fig. 6B). These results
demonstrate that the transcriptional activity of the
Oct-3/4 upstream region is down-regulated by DNA methylation under transient transfection conditions.
View larger version (10K):
[in this window]
[in a new window]
|
Fig. 6.
Methylation represses
Oct-3/4 upstream regulatory elements activity.
A, maps of the pOct-luc and pOctPE-luc reporter
plasmids. The XbaI/BanI fragment, spanning
positions 413 to +48 of the Oct-3/4 P region, was inserted
upstream to the luciferase reporter gene in pBluescript II KS to
generate pOct-luc (26). The BamHI/BamHI fragment,
spanning positions 1215 to +155 of the Oct-3/4 region
containing the P and PE elements, was inserted upstream to the
luciferase reporter gene in pBluescript II KS to generate pOctPE-luc.
B, the pOct-luc or the pOctPE-luc constructs were
transiently cotransfected into P19 EC cells either un-methylated or
methylated with the indicated enzymes (Hp, HpaII
methylase; Hh, HhaI methylase) together with the
pCMV-Renilla luciferase construct. Luciferase and
Renilla luciferase activities were determined 48 h
post-transfection, and each separate transfection was normalized to the
Renilla luciferase activity. Percent luciferase activity
values are relative to the un-methylated respective construct that was
set at 100%. Percent luciferase activity values, presented as means of
3 independent experiments ± S.D., are shown. C,
RT-PCR analysis of p53/ fibroblasts (lanes 1 and 2), p53/ Dnmt1/
fibroblasts (lanes 3 and 4), and blastocysts
(lane 5) for Oct-3/4 transcripts with (+) or
without () TSA treatment. The mouse adenine
phosphoribosyltransferase (APRT) gene was used as a
control.
|
|
These results prompted us to consider whether the same may be true for
basal transcription associated with the Oct-3/4
locus in vivo. In order to test this possibility, we took
advantage of a p53//Dnmt/ (Dnmt; DNA
methyltransferase) fibroblast cell line developed in Dr. Howard
Cedar's laboratory. The p53/ knock out allows the
propagation of a Dnmt/ cells for a longer time (32).
Unlike wild type (wt) or p53/ fibroblasts, almost all
of the CpG residues in these cells are unmethylated (data not shown).
As shown in Fig. 6C, Oct-3/4 RNA synthesis was readily
detected in p53//Dnmt/ fibroblasts,
whereas no Oct-3/4 could be detected in p53/ cells,
using sensitive RT-PCR analysis. These data suggest that DNA
methylation inhibits transcription of the endogenous
Oct-3/4 gene as well.
It was shown previously that treatment of cells with the histone
deacetylase inhibitor trichostatin A (TSA) (33, 34) partially overcomes
the effects of DNA methylation on gene expression (35, 36). In order to
find whether this is also the case for the Oct-3/4 locus, we treated the cells with TSA and
examined Oct-3/4 basal transcription. Treatment
of p53//Dnmt/ cells resulted in no
induction of the Oct-3/4 transcript level, probably due to the already acetylated nature of the
Oct-3/4 chromatin surrounding the unmethylated
substrate (Fig. 6C). Interestingly, TSA treatment of
p53/ cells also did not help to induce any detectable
level of Oct-3/4 transcription, most probably due
to the ability of DNA methylation to inhibit basal transcription that
cannot be overcome by a change in the acetylation status of the chromatin.
| |
DISCUSSION |
The Oct-3/4 gene provides a unique model
system to study the molecular parameters that underlie the generation
of the methylation status of genes that are expressed during early
embryogenesis and are repressed following implantation around
gastrulation. By analyzing the Oct-3/4
methylation status during embryogenesis, we have shown that the
Oct-3/4 upstream regulatory region is
unmethylated in blastocysts and in 6.25 dpc epiblast cells and becomes
methylated in 6.5 dpc embryos almost to the full extent, similar to
adult tissues (Fig. 1, B and D). Interestingly,
it seems that there is a core methylation center in the PE,
encompassing HhaI and HpaII site 3, which is
almost fully modified, whereas sequences that flank this center are
gradually less methylated as a function of their distance from the
center (Fig. 1E).
DNA methylation is known to have profound effects on both chromatin
structure and gene expression (37-39). Our current and previously
published data (22) indicate that Oct-3/4
methylation status is highly compatible with
Oct-3/4 expression. Our experiments clearly
demonstrate that methylation inhibits transcription of transfected
genes regulated by the Oct-3/4 upstream region
(Fig. 6B). Furthermore, methylation inhibits transcription
of the endogenous Oct-3/4 gene, because in cells
in which most of the CpG residues are unmethylated
Oct-3/4 transcripts are readily detected. TSA treatment that modifies the chromatin structure did not induce Oct-3/4 transcription from a methylated
substrate, indicating that in these cells histone acetylation cannot
override inhibition by DNA methylation (Fig. 6C).
Recently, it was shown that relative minor changes in Oct-3/4 protein
expression exert a dramatic effect. For instance a 50% decrease in the
Oct-3/4 protein levels causes the cell to differentiate into
trophectodermal cells (14). A consequence of this finding is the
necessity for fine-tuning of Oct-3/4 gene
regulation. Indeed, a number of studies (26, 40-42) have
shown that several orphan members of the nuclear receptor
gene superfamily are involved in repression of
Oct-3/4 expression. Our study indicates that repression of Oct-3/4 expression is brought about
not only by trans-acting transcriptional repressors but also
by epigenetic mechanisms, such as DNA methylation. It is possible that
silencing of Oct-3/4 by methylation is a secondary event to another
epigenetic mechanism such as chromatin modifications. This in fact
could explain our findings that in blastocysts no methylation is
observed (Fig. 1B), although the gene is repressed in
trophoblasts at this stage.
Oct-3/4 Demodification Element and
Embryogenesis--
Oct-3/4 expression is
dependent on at least three upstream cis-regulatory regions: the P, the
PE, and a distal enhancer (3, 5, 8, 12, 41, 43). Our results clearly
show that the PE element is not only a transcriptional enhancer but it
is also a cis-demodification element that
demethylates/protects the DNA in a cell- and stage-specific manner. The
observation that enhancer sequences are part of several identified
demodification elements (20, 27, 44-46) and the fact that
demethylation does not seem to occur as a consequence of
transcriptional activation (47-49) strongly suggest that enhancers may
have two separate and independent functions, first to target genes for
epigenetic modification and second to stimulate RNA synthesis.
The PE sequence in the Oct-3/4 gene harbors two
sites, 1A and 1B, that in vitro bind distinct and currently
unidentified factors. The 1A site, located between conserved regions of
homology 2 and 3 (50), contains the sequence GGGAGGG that is conserved
among mouse, human, and bovine Oct-3/4 upstream
region. This site is occupied in vivo in undifferentiated EC
and ES cells (3, 5). Repression of Oct-3/4
transcription through the PE by RA treatment is coupled with the
displacement of factors from site 1A, indicating that in cells
expressing the Oct-3/4 gene, transcriptional
activators bind this site. Interestingly, our data show that site 1A,
and not 1B, plays a critical role in the demethylation reaction (Fig. 3B). Thus, our results correlate well with the previously
published data showing that site 1A is the crucial site for
transcriptional activation in vivo.
A number of previously published studies (16, 17) indicate that most of
the CpG sites in the genome, located in expressed and non-expressed
sequences, are unmethylated in the blastula. Between implantation and
gastrulation a wave of de novo methylation re-establishes
the overall methylation pattern, which is then maintained throughout
life in the somatic cells of the animal, except for the demethylation
of some tissue-specific genes during differentiation (51). Our data
indicate that during embryogenesis Oct-3/4
remains unmethylated, whereas other genes undergo de novo methylation (similarly to genes on the inactive X chromosome (52)). Theoretically, it is possible that the overall wave of de
novo methylation does modify the Oct-3/4
gene along with other sequences, but it continuously undergoes
demethylation due to the presence of an active demodification element.
The other possibility is that Oct-3/4 harbors a
cis-element that is bound by proteins that protect the gene
from the methyltransferases that operate at this time of embryogenesis.
Our studies do not differentiate between the two possibilities;
however, because the DNA is initially unmethylated, we postulate that
in vivo the Oct-3/4 PE element
protects the DNA from de novo methylation in 6.25 dpc
post-implantation embryos (Fig. 5). Because the tissue composition of
the 6.5 dpc embryo is different from the composition of the 6.25 dpc
embryo, it is possible that in a small number of cells that do express
the Oct-3/4 protein in the 6.5 embryo the PE element still protects the
gene from de novo methylation. It is highly likely that this
element is bound by trans-acting factors that prevent access
to the DNA from the methylation machinery. It was shown previously that
DNA-binding proteins are able to protect sites from de novo
methylation as well as specify sites for demethylation (49, 53). These
factors may recruit additional cell-specific proteins (with histone
modification capabilities (54)) to the region, thereby protecting the
local region from de novo methylation under circumstances
where the sequence is initially unmethylated or demethylating the
region when the sequence is first methylated.
Once we have identified the PE as a demethylation/protection element in
the Oct-3/4 gene, it is pertinent to ask where
else (besides in epiblast cells) this element is active in
vivo. There are two obvious cell types where it may play a role.
The first candidates are cells in the post-implantation embryo. Because Oct-3/4 expression persists through 7.5 dpc
embryo in the unsegmented presomitic mesoderm, decreasing anteriorly to
posteriorly as the somites form until 8.5 dpc (1, 7, 9), it is logical
to suggest that the Oct-3/4
demethylation/protection element protects the gene from becoming
modified in these subpopulations of cells as well.
Additionally, the PE demethylation element could play a role during
differentiation and maturation of gametes of both sexes. Throughout
spermatogenesis de novo methylation is very well
orchestrated, occurring later than global methylation in somatic cells
and at least in three waves as follows: the first at 15.5-17.5 dpc,
the second in the vas deferens, and the third that methylates imprinted genes only (55, 56). The Oct-3/4 gene stays
unmethylated almost to the last developmental stage of spermatogenesis,
in the face of the global de novo methylation wave that
occurs in pro-spermatogonia (56). It is reasonable to suggest that the
novel Oct-3/4 demodification element protects the
Oct-3/4 gene from the first but not from the
second or third waves of de novo methylation.
In the female gonads, Oct-3/4 is expressed until
15.5 dpc (when de novo methylation occurs) and is
down-regulated as the oocyte enters prophase I (at 16.5 dpc) to be
again expressed in oocytes within primary follicles and at
the onset of oocyte growth and folliculogenesis (1-2 dpc) (9, 10). The
methylation pattern of the Oct-3/4 gene during
oogenesis has not been studied. However, there is a striking
correlation between the global de novo methylation timing
and Oct-3/4 down-regulation. Thus, because in
oocytes Oct-3/4 is re-expressed after birth, it
is logical to suggest that the PE demethylation element may also play a
role in protecting the Oct-3/4 gene from
undergoing de novo methylation at 15.5 dpc and therefore
allowing for its later re-expression during oogenesis. Thus, the PE
element protects the gene during early development and is a very good
candidate for a germ cell-specific demethylating element.
It was shown previously (18, 19) that the mouse has a
pre-programmed mechanism to ensure that CpG islands at the 5' end of
all housekeeping genes remain unmethylated to allow expression. Sp1-like binding sites are frequently associated with CpG islands, where they protect the islands from undergoing de novo
methylation (18, 19). We propose that the mouse has also a similar
mechanism to ensure undermethylation and expression of genes that are
specifically expressed through a particular window of time during
embryogenesis, when de novo methylation occurs. A
representative of this group of genes is Oct-3/4.
Indeed, our data clearly show that the Oct-3/4 PE1A site that is GC-rich and is related to the consensus Sp1-binding site binds several proteins. We have shown that in an in
vitro assay the Sp1 transcription factor is one of them. Most
interestingly, the 1A site protects the DNA from de novo
methylation in vivo (Fig. 5). This is not to say that every
Sp1-binding site harbors the ability to protect the surrounding
sequence from de novo methylation. In fact, it was shown
previously (25) that an Sp1-binding site on its own is not sufficient
for demethylation in vivo. However, it is logical to suggest
that during early stages of development specific Sp1-like binding sites
together with as yet unidentified cis sequences not only
protect/demethylate CpG islands in housekeeping genes but also modify
sequences located in cell-specific genes, which do not contain CpG
islands but contain other permissive sequences, such as the
Oct-3/4 PE element.
Methylation of the Oct-3/4 gene is most likely a
part of a regulatory process that takes place independently of the
major wave of de novo methylation. In the
Oct-3/4 gene case the methyl moieties are
actually added to the DNA later than in genes that are not expressed at
this particular time in development. This process may resemble the late
addition of methyl groups to CpG islands on the inactive X chromosome
(57). On the basis of our above described data, we propose the
following model for the generation of Oct-3/4
methylation pattern. During the post-implantation period and most
probably during germ cell formation, specific trans-acting factors bind to the PE element and protect the surrounding sequences from the global de novo methylation which lead in turn
to expression of the gene. During later stages of embryogenesis, these
factors are probably absent, and the Oct-3/4 gene
undergoes de novo methylation and silencing. This could
happen by a default mechanism due to the lack of proteins that bind to
the PE sequences or through binding of different kinds of transcription
factors (Fig. 7).
View larger version (9K):
[in this window]
[in a new window]
|
Fig. 7.
A model for generation of Oct-3/4
methylation pattern. During the early stages of
embryogenesis (up to 8.0 dpc) and most probably during spermatogenesis
and oogenesis, specific trans-acting factors
(red) bind the 1A site in the PE element and protect the
surrounding sequences from global de novo methylation which
leads in turn to Oct-3/4 undermethylation and
expression. At later stages of embryogenesis (post-8.5 dpc) and in the
adult, these factors are either absent (left part of Fig. 7)
or different trans-acting factors (purple
hexamer) bind (right part of the figure), and the
Oct-3/4 undergoes methylation de novo
and silencing.
|
|
TOP
ABSTRACT
INTRODUCTION
To whom correspondence should be addressed: Dept. of Experimental
Medicine and Cancer Research, The Hebrew University-Hadassah Medical
School, Jerusalem 91120, Israel. Tel.: 972-2-6758362; Fax:
972-2-6414583; E-mail: yberg@md2.huji.ac.il.
