Download Gene silencing in mammalian cells and the spread of DNA

ª
Oncogene (2002) 21, 5388 – 5393
2002 Nature Publishing Group All rights reserved 0950 – 9232/02 $25.00
www.nature.com/onc
Gene silencing in mammalian cells and the spread of DNA methylation
Mitchell S Turker*,1,2
1
Center for Research on Occupational and Environmental Toxicology, Oregon Health and Science University, Portland,
Oregon, OR 97201, USA; 2Department of Molecular and Medical Genetics, Oregon Health and Sciences University, Portland,
Oregon, OR 97201, USA
Aberrant gene silencing in mammalian cells is associated
with promoter region methylation, but the sequence of these
two events is not clear. This review will consider the
possibility that gene silencing is not a single event, but
instead a series of events that begins with a dramatic drop in
transcription potential and ends with its complete cessation.
This transition will be portrayed as a chaotic process that
ensues when transcription levels drop and DNA methylation
begins spreading haltingly towards the diminished promoter.
According to this view, silencing is stabilized when the
promoter region is ‘captured’ by the spread of DNA
methylation near or into its transcription factorbinding sites.
Oncogene (2002) 21, 5388 – 5393. doi:10.1038/sj.onc.
1205599
Keywords: DNA methylation; methylation spreading;
silencing
Introduction
This review is meant to be speculative. It will focus on
methylation-associated gene silencing in mammalian
cells and it will take the view that a decrease in
transcription is just the first step in a gradual process
that involves a feedback loop between diminishing
transcription and the spread of DNA methylation. In
my opinion normal methylation patterns represent a
balancing act between competing forces; open chromatin established by an active promoter and the spread of
DNA methylation from a pre-existing focus (Turker,
1999). If so, a significant diminishment in transcription
would lead to a loss of this balance allowing DNA
methylation to spread towards the promoter. The
presence of variegated methylation patterns near
silenced promoters will be used to build a model in
which transcription smolders before being extinguished.
Methylation patterns result from opposing forces that
promote and block DNA methylation
Many regions of the mammalian genome are methylated at one or more CpG sites (Bird, 2002) and exhibit
*Correspondence: MS Turker, CROET, L606, Oregon Health
Sciences University, 3181 SW Sam Jackson Park Road, Portland,
OR 97201, USA; E-mail: [email protected]
allele specific methylation patterns (Chandler et al.,
1987; Silva and White, 1988; Turker et al., 1989). These
patterns are formed during embryonic development
(Reik et al., 2001) beginning with genome wide
demethylation that commences shortly after fertilization (Monk et al., 1987; Howlett and Reik, 1991;
Santos et al., 2002). Remethylation of most of the
genome occurs after the blastocyst implants (Monk et
al., 1987; Okano et al., 1999; Santos et al., 2002) and
continues at a slower pace during the remainder of
development. It is believed that this demethylation/
remethylation process converts gamete specific methylation patterns to those specific to somatic cells.
Remethylation of the genome can be divided into
two steps involving different forms of de novo
methylation (Turker, 1999). The first is de novo
methylation that occurs at discrete CpG sites throughout the genome due to the presence of cis-acting
methylation signals (Figure 1a). Once CpG sites within
cis-acting signals are methylated they can act as foci
for a second form of de novo methylation, which is
spreading to distal CpG sites (Figure 1b,c). After a
CpG site becomes methylated via spreading this
modification is preserved via maintenance methylation,
which involves the targeting of DNMT1 to hemimethylated DNA shortly after replication (Leonhardt
et al., 1992). Most spreading of methylation ceases
about the time of birth, although low level spreading
may continue during an individual’s lifetime (Issa,
2000; Ahuja and Issa, 2000). While the phenomenon of
spreading has not been defined mechanistically, I
suggest that it represents a self-perpetuating interaction
between chromatin modifying proteins and DNA
methylation. According to this view, DNA methylation
closes chromatin via the attraction of repressive
complexes (Bird and Wolffe, 1999). The presence of
these complexes will, in turn, alter nearby chromatin
making it accessible to the spread of methylation. The
net effect of this interaction is the gradual spreading of
methylation along the DNA molecule until it run into
a counteracting force in the form of open and active
chromatin (Figure 1c).
Based on our work with the mouse Aprt gene region,
I have proposed that the initial de novo methylation
events are signaled by cis-acting elements that are
dispersed throughout the genome (Turker, 1999)
(Figure 1a). For Aprt, we have identified two upstream
B1 repetitive elements as providing a significant de novo
methylation signal (Yates et al., 1999). They reside
Silencing and DNA methylation spreading
MS Turker
5389
Figure 1 Dynamic processes shape methylation patterns. Methylation is established at CpG sites (closed circles) by cis acting sequences (striped rectangle) (a) and then begins spreading to
unmethylated CpG sites (open circles) in both directions (only
one direction is shown) (b) until it reaches an open chromatin domain (c) established by the expressed promoter (flag, ‘+’ represents relative level of expression). The junction between the
open chromatin domain and methylation spreading is considered
to be a boundary between the repressive complexes that bind
methylated CpG sites and open chromatin. Shifting of the boundary can move methylated CpG sites into the open chromatin domain, causing them to become demethylated (d)
within a larger DNA fragment that we have termed a
methylation center (Mummaneni et al., 1993). Other
retroposons that are also suspected of signaling de novo
DNA methylation are B2 (Hasse and Schultz, 1994),
Alu (human equivalent of B1) (Magewu and Jones,
1994; Graff et al., 1997), LINE-1 (Liang et al., 2002),
and mouse IAP (Walsh et al., 1998). De novo
methylation of retroposons may be a protective
function to block their transposition (Yoder et al.,
1997). Once the B1 elements upstream of Aprt become
methylated, methylation spreads towards the promoter
(Figure 1b,c), which is comprised of three consensus
Sp1 binding sites. It has also been shown in transgenic
mice that DNA methylation can spread from an
integrated retrovirus to flanking sequences in the
mouse genome (Jähner and Jaenisch, 1985). Similarly,
it has been shown in plants that short retroposons can
act as nucleation centers for de novo DNA methylation
and that DNA methylation then spreads from these
centers to distal sites (Arnaud et al., 2000).
The Aprt promoter provides a counteracting force to
spreading as demonstrated by two types of studies. One
used embryonic cells and developing embryos to show
that removal or site directed mutation of one or more
of the Sp1 binding sites allows methylation to spread
into the promoter (Mummaneni et al., 1995, 1998;
Brandeis et al., 1994; Macleod et al., 1994). Because the
site directed mutations were designed to eliminate
transcription factor binding it was assumed that protein
binding plays an important role in blocking the spread
of DNA methylation, presumably by maintaining an
open chromatin domain around the promoter region. A
second set of studies has shown that the addition of
Sp1 binding sites can prevent transgene (Siegfried et al.,
1999) and retroviral (Hejnar et al., 2001) silencing.
Additional studies provided evidence that DNA
methylation patterns are not static in somatic cells, as
often assumed, but instead can vary at individual CpG
sites (Silva et al., 1993; Turker et al., 1989). For mouse
Aprt, we showed that a CpG site located between the
Aprt promoter and the methylation center is methylated on approximately 25% of alleles, and that this
level persisted even when the cells were subcloned
(Turker et al., 1989). These observations can best be
explained by assuming that methylation for CpG sites
on a given allele can be either gained or lost.
By combining the results from the above studies, I
proposed a model depicting a ‘dynamic equilibrium’
between forces that support the spread of DNA
methylation and those that actively block this spread
(Turker, 1999). The frontline where these forces meet
(Figure 1c) can be considered a boundary that separates
repressive complexes from those that keep chromatin in
an open and expressed conformation (Gerasimova and
Corces, 2001). An A-rich sequence in the human
GSTP1 gene has also been shown to form a boundary
between methylated and unmethylated DNA (Millar et
al., 2000). For mouse Aprt, the upstream methylation
center and the promoter provide the opposing forces of
DNA methylation spreading and blocking. According
to the model, the boundary can shift back and forth
along the DNA molecule and methylated CpG sites
caught behind the boundary are at risk to switch from
being methylated to being unmethylated (Figure 1d). If
this model is correct, it follows that the protective force
of the Aprt promoter will extend at least several
hundred bases upstream from the end of its associated
nuclease sensitivity (Cooper et al., 1993) and nucleosome positioning (Macleod et al., 1994) regions.
Although a mechanism for direct removal of methyl
groups from DNA has been proposed (Bhattacharya et
al., 1999) and debated (Bird, 2002), the boundary
model is more consistent with a replication dependent
mechanism (Santos et al., 2002; Rougier et al., 1998) in
which maintenance methylation is blocked during S
phase (Bird, 2002). A final aspect of this model is that
methylation of demethylated sites will occur anew via
spreading when the front line shifts again towards the
promoter (not shown). Fortunately for Aprt, the
frontline of the battle is located several hundred base
pairs away from the promoter, and while some shifting
of this frontline can occur, it rarely moves close enough
to the promoter to jeopardize gene expression. The last
section of this review will deal with what happens when
silencing begins, and the blocking force is weakened
thereby allowing the spread of methylation towards the
promoter. First I will consider some things we know
and do not know about silencing.
Gene silencing in mammalian cells
A simple definition of gene silencing is the conversion
of an actively expressed allele to one that is not
Oncogene
Silencing and DNA methylation spreading
MS Turker
5390
expressed. This term usually denotes aberrant loss of
expression. In mammalian cells, gene silencing is most
often associated with hypermethylation of the promoter region and a closed chromatin conformation
(Robertson and Jones, 2000; Rountree et al., 2001;
Baylin et al., 2001; Jones and Wolffe, 1999).
Chromatin conformation is controlled to a large
extent by modification of lysine residues on histones
tails (Jenuwein and Allis, 2001). These modifications
include acetylation/deacetylation and methylation/
demethylation at lysine positions 4, 9 and 14. Histone
acetylation, deacetylation and methylation are enzymatically catalyzed, but histone demethylation may
require a more complex process. In general, histone
acetylation is associated with active chromatin and
deacetylation is associated with closed chromatin.
DNA methyl binding proteins attract histone deacetylases, which appears to explain the relationship
between DNA methylation and gene repression (Bird
and Wolffe, 1999; Wade, 2001). For histone methylation the picture is more complex because methylation
of lysine 4 is associated with active chromatin and
methylation of lysine 9 is correlated with inactive
chromatin (Jenuwein and Allis, 2001).
The term ‘histone code’ (Jenuwein and Allis, 2001)
has been used to describe the various potential
combinations of histone modifications. Because many
different combinations of histone tail modifications can
potentially exist, it is possible that the level of gene
expression for a given allele can vary from very high to
undetectable. If so, the specific histone modifications
that are present near the promoter on a given allele
could determine the probability that the promoter will
be able to bind to transcription factor, the strength of
this interaction, and perhaps for how long it will last.
This interaction, in turn, could dictate how far the
boundary is located away from the promoter. As
discussed in the previous section, DNA outside the
boundary would be methylated and histones outside
the boundary would have repressive modifications.
However, these modifications would pose little threat
to the promoter as long as the boundary is present. But
what would happen if transcription were disrupted
transiently for a given allele? During that time the
boundary would be weakened significantly or disappear, which would initiate a battle for the allele
between repressive forces that are trying to silence
expression permanently and a promoter that is
struggling to regain high level transcription.
The outcome of a battle for a given allele is not
predetermined as evidenced by observations such as
position-effect-variegation (PEV) for eye color in
Drosophila (Wakimoto, 1998; Weiler and Wakimoto,
1995). Eye color PEV is caused by the translocation of
an active gene region to a heterochromatic (i.e.
repressive) region, leading to cell-to-cell variation for
expression of an eye pigment gene. This gene is termed
white because the normally red eye cells appear white
when expression is lost. Since each cell inherits exactly
the same translocation event, the cell-to-cell difference
in color demonstrates the lack of a predetermined
Oncogene
outcome for each translocated allele. DNA methylation
occurs at extremely low levels in the fly (Gowher et al.,
2000; Lyko et al., 2000) and thus is unlikely to be
involved in gene repression. This suggests that allelic
battles for expression can occur in the absence of DNA
methylation, and by extension that chromatin modifications play the central role. Not surprisingly, it has
been shown that altered expression of a variety of
chromatin modifying proteins (Wakimoto, 1998; Weiler
and Wakimoto, 1995) can influence the percentage of
red versus white cells in a given eye. Cell to cell
variation has also been shown in the mouse, perhaps
best visualized as variegated coat color due to an
insertion of an IAP retroviral element into the agouti
locus that causes cell-to-cell variation in silencing of
this allele (Duhl et al., 1994; Michaud et al., 1994). In
these cases, DNA methylation of the IAP element has
been correlated with loss of agouti expression.
An important component for the silencing process is
the initiation event, which unfortunately is a part of
the process that has not been defined mechanistically
(Baylin and Herman, 2000). If DNA methylation
initiates silencing in mammalian cells by spreading
past the protective boundary and recruiting repressive
complexes, loss of transcription should be stable once
it occurred. Consistent with this notion are numerous
studies showing that the methylated promoters are
stably inactivated and not expressed at detectable
levels, presumably because methyl-binding proteins
attach to the methylated CpG sites and attract
repressive complexes (Bird and Wolffe, 1999; Rountree
et al., 2001; Wade, 2001).
The problem with methylation spreading being the
trigger for silencing in mammalian cells is that evolving
processes have been observed in those instances where
it has been possible to follow silencing. In a study of
p16 inactivation in human mammary epithelial cells
that escape senescence after stable transfection with the
E6 papillomavirus gene, it was shown that escape from
senescence was correlated initially with a decrease but
not complete loss of p16 expression (Wong et al.,
1999). Significantly, at this early time point some p16
alleles were found to be unmethylated in the promoter
region. Eventually, as the cells continued to be
passaged, promoter region methylation increased and
the levels of p16 expression became undetectable.
Additional evidence to support a gradual silencing
process came from a study of silencing of a retroviral
LTR (long terminal repeat) fused to a GFP reporter
gene in mouse leukemia cells (Lorincz et al., 2000). In
this study, silenced GFP constructs were reactivated by
treatment with the deacetylating agent trichostatin A
(TSA) soon after construct silencing occurred. The
reactivated alleles were then found to gradually silence
again, but for many cellular passages the silenced
alleles could be reactivated anew by treatment of the
cells with either TSA or the demethylating agent 5-azadC. However, with continued passage the silenced
alleles eventually reached a point where they became
refractory to either TSA or 5-aza-dC exposures by
themselves. Increased methylation levels were corre-
Silencing and DNA methylation spreading
MS Turker
5391
lated with loss of responsiveness to TSA or 5-aza-dC.
Similarly, in a recently completed study (submitted) we
have found that mouse Aprt constructs exhibited at
least two stages of silencing in a differentiated cell type.
The first stage was characterized by ‘leaky’ (i.e. readily
detectable) low-level transcription in most cells and
very high spontaneous reactivation frequencies. At this
stage many of the silenced alleles exhibited methylation
patterns virtually identical to those observed for
expressing alleles. The second stage, which apparently
required the first stage as an intermediate, was
characterized by loss of transcription and loss of
spontaneous reactivation. Increased levels of construct
methylation correlated with stabilization of the silencing process.
In all three studies mentioned above the observed
methylation patterns were variegated rather than
complete. Variegated patterns are defined by alternating regions of methylated and unmethylated CpG sites
(see Figure 2 for an unpublished example from our
work). Moreover, in all three studies the variegated
patterns were different from allele to allele, even
though clonally derived cell populations were examined. These observations can only be explained by
assuming that DNA methylation patterns are evolving
on silenced alleles, and that silencing is a process not a
single event. Similarly, a study that correlated transcription and DNA methylation levels for the human
p15 gene in acute leukemias also led to the suggestion
of an evolving process, at least for methylation changes
(Cameron et al., 1999). An evolving silencing process
would be in marked contrast to a wide spectrum of
mutational events, which require no more than two cell
divisions to become established (Turker, 1998). The
next section describes how an initial decrease in
transcription can start a process that causes the spread
of methylation and the formation of alleles with
variegated methylation.
Figure 2 Example of a variegated methylation patterns for unstably silenced alleles. Unpublished work from the author’s group
showing bisulphite sequencing analysis for 19 CpG sites (circles)
on 14 Aprt alleles isolated from a clonally derived cell population
expressing low levels of activity and high spontaneous reversion
frequencies. In these constructs methylation initiates mostly at
CpG sites 1 – 3 and methylation spreads from these CpG sites
to the promoter region, which contains 2 Sp1 binding sites.
CpG sites 13 and 16 are within these binding sites. Closed and
open circles represent methylated and unmethylated CpG sites, respectively. The distances separating CpG sites represents relative
distances that they are apart on the Aprt plasmid that was examined. Typical of silenced alleles obtained from a clonal population, only two of the 14 alleles exhibit identical methylation
patterns
Silencing and the spread of methylation
There are two ways to consider the relation between
silencing and the spread of DNA methylation. One is
that the spread of DNA methylation into the promoter
causes silencing via the attraction of chromatin
modifying factors. This, in turn, would lead to a loss
in transcription factor binding. According to this
scenario silencing, as defined as loss of transcription,
would be a rapid and relatively stable process.
However, the presence of silenced alleles with variegated methylation patterns in tumor material (Melki et
al., 2000; Song et al., 2001; Dodge et al., 1998; Nosaka
et al., 2000; Cameron et al., 1999) and cultured cells
(Wong et al., 1999; Lorincz et al., 2000; Domann et al.,
2000; Liang et al., 2000) suggests an alternative
scenario in which silencing is a gradual and evolving
process that involves promoter region methylation as a
very late event.
I speculate that silencing initiates with a dramatic,
albeit incomplete and still undefined, drop in transcription
that
could
result
from
spontaneous,
environmentally-induced, or malignancy-related perturbations in expression acting at the level of individual
alleles. Silencing then progresses via series of mutually
reinforcing steps that gradually reduce the probability
of transcription factor binding. This creates a situation
in which a given promoter is bound by transcription
factor for periods of time interspersed with increasing
periods of time when the promoter is unoccupied.
Because the boundary that acts to prevent methylation
from spreading towards the promoter is a product of
transcription factor binding, a chaotic scenario will
ensue in which the boundary to methylation spreading
will alternately disappear and reappear (Figure 3). As a
result methylation will spread towards the promoter in
a halting process, slowed or even reversed occasionally
when transcription factor binding occurs and the
boundary is formed anew. At the same time, because
of the interrelationship between histone modification
and DNA methylation, repressive modifications will
move towards the promoter, also in a halting and
reversible process. Moreover, because the location of a
boundary should be a function of the strength and
longevity of transcription factor biding, which in turn
could be affected by the proximity of DNA methylation, the boundary would develop closer and closer to
the promoter each time it is reformed.
This evolving situation will create interesting circumstances, such as when a boundary forms upstream of
methylated CpG sites (Figure 3). Such a transient event
would allow methylated CpG sites to become
unmethylated if transcription and hence the open
domain, is restored and maintained during replication.
However, because boundaries are forming and disappearing continuously, there is no certainty that an
unmethylated CpG site caught outside the protective
domain will become methylated, or alternatively that a
methylated site caught inside the protective domain will
become unmethylated. Instead, because the boundary
is being formed, lost, and formed again, there are
Oncogene
Silencing and DNA methylation spreading
MS Turker
5392
Figure 3 Silencing as a process. An expressed allele (a) with a
boundary between DNA methylation spreading and the open
chromatin region of the expressed promoter (flag with ‘+’) undergoes a dramatic loss in transcription (flag with ‘7’), which eliminates boundary allowing methylation to spread closer to
promoter. A series of restoration and loss of promoter expression
(c – g) ensues, with transcription levels decreasing as methylation
encroaches. Finally, a sufficient density of methylation and/or
methylation of critical CpG sites (filled in flag) occurs leading
to stabilization of silencing (h). A by-product of this process is
the formation of variegated methylation patterns
continue inexorably because the spread of DNA
methylation and closing of chromatin due to histone
modification will feedback upon each other. During
malignant progression this process will be hastened by
selection for alleles with progressively lower levels of
expression because loss of normal regulation of cell
growth and sometimes DNA repair pathways are
critical for the development of cancerous cells (Turker,
1998). While it may be possible for a rare allele to
reverse the process, particularly if selection for
expression is present, for most alleles DNA methylation will continue spreading towards the promoter as
the interaction between the promoter and transcription
factor weakens in strength and decreases in duration.
Moreover, by influencing chromatin conformation via
the attraction of methyl-binding proteins, the spread of
methylation will play a role in decreasing this
interaction, further facilitating its spread. The battle
finally ends when a sufficient number of CpG sites near
or within the promoter and/or critical CpG sites
become methylated to preclude further interaction
between the promoter and transcription factor, though
there may be remaining unmethylated CpG sites that
will eventually succumb to methylation.
Omissions and conclusion
increasing probabilities for each CpG site being
methylated as the boundary is pushed gradually
towards the promoter, but no guarantees. The net
result of this ongoing process is the gradual advance of
DNA methylation towards the promoter and the
formation of alleles with variegated methylation
patterns. The reader is reminded that Figure 3 shows
the CpG methylation history for only a single allele
and that each allele could potentially have its own
unique history. Therefore, countless variegated patterns
will be observed when ‘sampling’ the methylation
pattern for a silenced gene (Figure 2). Finally, by
combining the concept of shifting boundaries shown in
Figure 1 with the concept that CpG sites caught in
these shifting regions will sometimes, but not always,
change methylation status (Figure 3), it is also possible
to explain variegated DNA methylation patterns that
have been reported for non-silenced regions of the
genome (Liang et al., 2002; Millar et al., 2000; Hansen
et al., 2000; Bourc’his et al., 2001; Hanel and Wevrick,
2001).
Unlike position effect variegation, which is established and fixed during development, the process
described above for a dividing cell population will
To keep this review focused, I omitted or discussed
minimally a number of potential additional players.
These include maintenance methylation, de novo
methylation, the different DNA methyltransferases
and their functions, non-CpG methylation, demethylation mechanisms, the aberrant mechanisms that cause
the initial diminishment of transcription, genomic
damage, and selective processes. I also omitted
discussion of retroviral silencing in embryonal cells,
which represents a normal process where loss of
expression clearly precedes DNA methylation (Cherry
et al., 2000; Stewart et al., 1982; Pannell et al., 2000;
Gautsch and Wilson, 1983). Despite these omissions, I
have hopefully stimulated the reader to think about
gene silencing in mammals as a process instead of an
event, and not a very tidy process at that.
Acknowledgements
The author’s current work on DNA methylation and
silencing is supported by NIH grant R01 CA92114. I
would like to thank Drs Maura Pieretti and Phil Yates for
many helpful comments on this manuscript.
References
Ahuja N, Issa JP. (2000). Histol. Histopathol., 15, 835 – 842.
Arnaud P, Goubely C, Pélissier T, Deragon J-M. (2000).
Mol. Cell. Biol., 20, 3434 – 3441.
Baylin SB and Esteller MRountree MRBachman KESchuebel KHerman JG. (2001). Hum. Mol. Genet., 10, 687 – 692.
Baylin SB, Herman JG. (2000). Trends Genet., 16, 168 – 174.
Oncogene
Bhattacharya SK, Ramchandani S, Cervoni N and Szyf M.
(1999). Nature, 397, 579 – 583.
Bird A. (2002). Genes Dev., 16, 6 – 21.
Bird APWolffe AP. (1999). Cell, 99, 451 – 454.
Bourc’his DXu GLLin CSBollman BBestor TH. (2001).
Science, 294, 2536 – 2539.
Silencing and DNA methylation spreading
MS Turker
5393
Brandeis M, Frank D, Keshet I, Siegfried Z, Mendelsohn M,
Nemes A, Temper V, Raxin A and Cedar H. (1994).
Nature, 371, 435 – 438.
Cameron EE, Baylin SB and Herman JG. (1999). Blood, 94,
2445 – 2451.
Chandler LA, Ghazi H, Jones PA, Boukamp P and Fusenig
NE. (1987). Cell, 50, 711 – 717.
Cherry SR, Biniszkiewicz D, van Parijs L, Baltimore D and
Jaenisch R. (2000). Mol. Cell. Biol., 20, 7419 – 7426.
Cooper GE, Bishop PL and Turker MS. (1993). Somat. Cell.
Mol. Genet., 19, 221 – 229.
Dodge JE, List AF and Futscher BW. (1998). Int. J. Cancer,
78, 561 – 567.
Domann FE, Rice JC, Hendrix MJ and Futscher BW. (2000).
Int. J. Cancer, 85, 805 – 810.
Duhl DMJ, Vrieling H, Miller KA, Wolff GL and Barsh GS.
(1994). Nat. Genet., 8, 59 – 65.
Gautsch JW and Wilson MC. (1983). Nature, 301, 32 – 37.
Gerasimova TI and Corces VG. (2001). Annu. Rev. Genet.,
35, 193 – 208.
Gowher H, Leismann O and Jeltsch A. (2000). EMBO J., 19,
6918 – 6923.
Graff JR, Herman JG, Myöhänen S, Baylin SB and Vertino
PM. (1997). J. Biol. Chem., 272, 22322 – 22329.
Hanel ML and Wevrick R. (2001). Mol. Cell. Biol., 21,
2384 – 2392.
Hansen RS, Stoger R, Wijmenga C, Stanek AM, Canfield
TK, Luo P, Matarazzo MR, D’Esposito M, Feil R, Gimelli
G, Weemaes CM, Laird CD and Gartler SM. (2000). Hum.
Mol. Genet., 9, 2575 – 2587.
Hasse A and Schultz WA. (1994). J. Biol. Chem., 269, 1821 –
1826.
Hejnar J, Hajkova P, Plachy J, Elleder D, Stepanets V and
Svoboda J. (2001). Proc. Natl. Acad. Sci. USA, 98, 565 –
569.
Howlett SK and Reik W. (1991). Development, 113, 119 –
127.
Issa JP. (2000). Curr. Top. Microbiol. Immunol., 249, 101 –
118.
Jähner D and Jaenisch R. (1985). Nature, 315, 594 – 597.
Jenuwein T and Allis CD. (2001). Science, 293, 1074 – 1080.
Jones PL and Wolffe AP. (1999). Semin. Cancer Biol., 9,
339 – 347.
Leonhardt H, Page AW, Weirer HU and Bestor TH. (1992).
Cell, 71, 865 – 873.
Liang G, Chan MF, Tomigahara Y, Tsai YC, Gonzales FA,
Li E, Laird PW and Jones PA. (2002). Mol. Cell. Biol., 22,
480 – 491.
Liang G, Robertson KD, Talmadge C, Sumegi J and Jones
PA. (2000). Cancer Res., 60, 4907 – 4912.
Lorincz MC, Schubeler D, Goeke SC, Walters M, Groudine
M and Martin DI. (2000). Mol. Cell. Biol., 20, 842 – 850.
Lyko F, Ramsahoye BH and Jaenisch R. (2000). Nature, 408,
538 – 540.
Macleod D, Charlton J, Mullins J and Bird AP. (1994). Genes
Dev., 8, 2282 – 2292.
Magewu AN and Jones PA. (1994). Mol. Cell. Biol., 14,
4225 – 4232.
Melki JR, Vincent PC, Brown RD and Clark SJ. (2000).
Blood, 95, 3208 – 3213.
Michaud EM, van Vugt MJ, Bultman SJ, Sweet HO,
Davisson MT and Woychik RP. (1994). Genes Dev., 8,
1463 – 1472.
Millar DS, Paul CL, Molloy PL and Clark SJ. (2000). J. Biol.
Chem., 275, 24893 – 24899.
Monk M, Boubelik M and Lehnert S. (1987). Development,
99, 371 – 382.
Mummaneni P, Bishop PL and Turker MS. (1993). J. Biol.
Chem., 268, 552 – 558.
Mummaneni P, Walker KA, Bishop PL and Turker MS.
(1995). J. Biol. Chem., 270, 788 – 792.
Mummaneni P, Yates P, Simpson J, Rose J and Turker MS.
(1998). Nucl. Acids Res., 26, 5163 – 5169.
Nosaka K, Maeda M, Tamiya S, Sakai T, Mitsuya H and
Matsuoka M. (2000). Cancer Res., 60, 1043 – 1048.
Okano M, Bell DW, Haber DA and Li E. (1999). Cell, 99,
247 – 257.
Pannell D, Osborne CS, Yao S, Sukonnik T, Pasceri P,
Karaiskakis A, Okano M, Li E, Lipshitz HD and Ellis J.
(2000). EMBO J., 19, 5884 – 5894.
Reik W, Dean W and Walter J. (2001). Science, 293, 1089 –
1093.
Robertson KD and Jones PA. (2000). Carcinogenesis, 21,
461 – 467.
Rougier N, Bourc’his D, Gomes DM, Niveleau A, Plachot
M, Páldi A and Viegas-Péquignot E. (1998). Genes Dev.,
12, 2108 – 2113.
Rountree MR, Bachman KE, Herman JG and Baylin SB.
(2001). Oncogene, 20, 3156 – 3165.
Santos F, Hendrich B, Reik W and Dean W. (2002). Dev.
Biol., 241, 172 – 182.
Siegfried Z, Eden S, Mendelsohn M, Feng X, Tsuberi BZ and
Cedar H. (1999). Nat. Genet., 22, 203 – 206.
Silva AJ, Ward K and White R. (1993). Dev. Biol., 156, 391 –
398.
Silva AJ and White R. (1988). Cell, 54, 145 – 152.
Song SH, Jong HS, Choi HH, Inoue H, Tanabe T, Kim NK
and Bang YJ. (2001). Cancer Res., 61, 4628 – 4635.
Stewart CL, Stuhlmann H, Jähner D and Jaenisch R. (1982).
Proc. Natl. Acad. Sci. USA, 79, 4098 – 4102.
Turker MS. (1998). Semin. Cancer Biol., 8, 407 – 419.
Turker MS. (1999). Semin. Cancer Biol., 9, 329 – 337.
Turker MS, Swisshelm K, Smith AC and Martin GM. (1989).
J. Biol. Chem., 264, 11632 – 11636.
Wade PA. (2001). Bioessays, 23, 1131 – 1137.
Wakimoto BT. (1998). Cell, 93, 321 – 324.
Walsh CP, Chaillet JR and Bestor TH. (1998). Nat. Genet.,
20, 116 – 117.
Weiler KS and Wakimoto BT. (1995). Annu. Rev. Genet., 29,
577 – 605.
Wong DJ, Foster SA, Galloway DA and Reid BJ. (1999).
Mol. Cell. Biol., 19, 5642 – 5651.
Yates PA, Burman RW, Mummaneni P, Krussel S and
Turker MS. (1999). J. Biol. Chem., 274, 36357 – 36561.
Yoder YA, Walsh CP and Bestor TH. (1997). Trends Genet.,
13, 335 – 340.
Oncogene