Oncogene (2002) 21, 5361 – 5379
ª 2002 Nature Publishing Group All rights reserved 0950 – 9232/02 $25.00
www.nature.com/onc
DNA methylation and chromatin – unraveling the tangled web
Keith D Robertson*,1
1
Epigenetic Gene Regulation and Cancer Section, National Cancer Institute, National Institutes of Health, Bethesda, Maryland,
MD 20892, USA
Methylation of cytosines within the CpG dinucleotide by
DNA methyltransferases is involved in regulating transcription and chromatin structure, controlling the spread
of parasitic elements, maintaining genome stability in the
face of vast amounts of repetitive DNA, and X
chromosome inactivation. Cellular DNA methylation is
highly compartmentalized over the mammalian genome
and this compartmentalization is essential for embryonic
development. When the complicated mechanisms that
control which DNA sequences become methylated go
awry, a number of inherited genetic diseases and cancer
may result. Much new information has recently come to
light regarding how cellular DNA methylation patterns
may be established during development and maintained in
somatic cells. Emerging evidence indicates that various
chromatin states such as histone modifications (acetylation and methylation) and nucleosome positioning
(modulated by ATP-dependent chromatin remodeling
machines) determine DNA methylation patterning. Additionally, various regulatory factors interacting with the
DNA methyltransferases may direct them to specific
DNA sequences, regulate their enzymatic activity, and
allow their use as transcriptional repressors. Continued
studies of the connections between DNA methylation and
chromatin structure and the DNA methyltransferaseassociated proteins, will likely reveal that many, if not all,
epigenetic modifications of the genome are directly
connected. Such studies should also yield new insights
into treating diseases involving aberrant DNA methylation.
Oncogene (2002) 21, 5361 – 5379. doi:10.1038/sj.onc.
1205609
Keywords: DNA methyltransferase; chromatin; DNA
methylation; histone deacetylase; histone methylase;
transcriptional repression
Introduction
DNA methylation is a complex process whereby one of
three DNA methyltransferases (DNMTs) catalyzes the
addition of a methyl group from the universal methyl
donor S-adenosyl-L-methionine, to the 5-carbon position of cytosine. This modification, occurring
predominantly within the CpG dinucleotide, is the
*Correspondence: KD Robertson; E-mail: [email protected]
most prevalent epigenetic modification of DNA in
mammalian genomes. There are currently three known,
catalytically active DNMTs, DNMT1, 3a, and 3b and
each one appears to play a distinct and critical role in
the cell (Bestor, 2000). CpG methylation profoundly
influences many processes including transcriptional
regulation, genomic stability, chromatin structure
modulation, X chromosome inactivation, and the
silencing of parasitic DNA elements (Baylin et al.,
2001; Jones and Laird, 1999; Robertson, 2001). These
diverse processes nevertheless appear to share a
common characteristic, that is, they all exert a
stabilizing effect which promotes genomic integrity
and ensures proper temporal and spatial gene expression during development.
Genomic DNA methylation patterns are not
randomly distributed. Rather, discrete regions, including most repetitive and parasitic DNA, are
hypermethylated, while other regions, such as CpGrich regions often associated with the regulatory
regions of genes (CpG islands), are hypomethylated
(Yoder et al., 1997). Furthermore, DNA methylation
patterns change dramatically during embryonic development. Genome wide demethylation after fertilization
is followed by waves of de novo methylation upon
embryo implantation. Not all sequences in the genome,
however, are demethylated upon fertilization and not
all sequences become de novo methylated after
implantation. These exceptions further emphasize the
regional specificity of genomic DNA methylation
patterning (Reik et al., 2001). Evidence of the great
importance of these methylation patterns can be
gleaned by examining the effects of disrupting them
in vivo. Engineered disruption of factors governing
DNA methylation patterns in mice has revealed their
vital role in embryonic development (Li et al., 1992;
Okano et al., 1999). Naturally occurring mutations in
genes involved in controlling DNA methylation
patterns, including one of the DNA methyltransferases,
result in ICF, Rett, ATRX and fragile X syndromes
(Robertson and Wolffe, 2000). Disruption of normal
DNA methylation patterns is one of the most common
features of transformed cells and a number of studies
have revealed that methylation changes are early events
in the tumorigenesis process and contribute directly to
transformation. In tumor cells the normal regulation of
the DNA methylation machinery is severely disrupted,
such that the regional specificity of methylation
patterns begins to be reversed, resulting in de novo
methylation of CpG islands and hypomethylation of
DNA methylation and cancer
KD Robertson
5362
repetitive DNA (Baylin et al., 2001; Jones and Laird,
1999; Robertson, 2001).
The role of DNA methylation in cancer has been
reviewed extensively and will not be discussed at length
here (Baylin et al., 2001; Jones and Laird, 1999;
Robertson, 2001). Rather, I will focus on emerging
data that may help to answer one of the most pressing
and intriguing questions in the DNA methylation field,
namely how cellular DNA methylation patterns are
established during mammalian development and then
properly maintained in somatic cells. Clues have been
contributed by numerous studies in the last few years,
which indicate that DNA methylation and chromatin
structure, or the ‘tightness’ of packaging of the DNA
in nucleosomes and the higher order structures they
form, are physically and functionally linked (Bird,
2002). For example, all known catalytically active
DNA methyltransferases interact with histone deacetylases and the use of inhibitors of each of these
processes has revealed that they work together to
repress transcription (Cameron et al., 1999). Studies in
plants, mice, and humans, using naturally occurring or
engineered mutations in chromatin remodeling
machines, have indicated that chromatin structure itself
may dictate cellular DNA methylation patterns (Bird,
2001). I will first discuss what is currently known about
the components of the cellular DNA methylation
machinery, namely the DNA methyltransferases. I will
then summarize current knowledge of the interacting
protein partners of the DNMTs and why the study of
such factors may yield clues to the means by which
methylation patterns are established. Lastly I will
discuss the connections between DNA methylation
and chromatin remodeling and describe a model of
how chromatin states may dictate genome-wide DNA
methylation patterning.
The mammalian DNA methylation machinery –
properties of the DNA methyltransferases (DNMTs)
In this section, I will provide an overview of the
enzymes responsible for establishing and maintaining
cellular DNA methylation patterns in mammalian cells,
namely the (cytosine-5) DNA methyltransferases
(Figure 1). Particular emphasis will be placed on the
domain structure of these proteins, their tissue-specific
patterns of expression, subnuclear localization, alternatively spliced isoforms, and their catalytic activity.
DNMT1
The (cytosine-5) DNA methyltransferase 1, now
properly referred to as DNMT1 (Figure 1), was the
first enzyme to be isolated as a mammalian DNA
methyltransferase, and more importantly, the only one
that was identified via biochemical fractionation
methodology (Bestor, 2000). All other genes described
below were identified by database search. Following
the isolation of murine DNMT1 (Bestor et al., 1988),
its human homolog was identified in 1992 (Yen et al.,
Oncogene
1992), which was mapped to chromosome 19p13.2.
Generally, DNMTs are believed to be composed of
two parts, a diversified amino terminal region and a
relatively conserved carboxy terminal region. DNMT1
has the largest amino terminal region of all the
mammalian DNMTs, which has roles in regulating
the activity of the carboxy terminal catalytic domain,
nuclear localization, zinc binding, and in mediating
protein – protein interactions (Figures 1 and 4, Table 1;
Bestor, 2000; Robertson, 2001). The carboxy terminal
region comprises the catalytic domain that is a
common feature of all cytosine DNA methyltransferases.
The biochemical and enzymatic properties of
DNMT1 have been studied in considerable detail.
DNMT1 has a significant preference for hemimethylated double-stranded DNA relative to unmethylated
double-stranded DNA. This unique property is why
DNMT1 is commonly referred to as the maintenance
methyltransferase (Bacolla et al., 1999; Flynn et al.,
1996; Glickman et al., 1997; Pradhan et al., 1997, 1999;
Yokochi and Robertson, 2002). Analysis of the
subnuclear localization of DNMT1 supports this
assignment. During the G1 and G2 phases of the cell
cycle, DNMT1 shows a diffuse nucleoplasmic distribution pattern, but associates with sites of DNA synthesis
(replication foci) throughout S phase (Leonhardt et al.,
1992; Liu et al., 1998). These results led researchers to
a classic and still attractive hypothesis. DNMT1, as the
primary maintenance methyltransferase, is required to
maintain the epigenetic information encoded by
genome-wide DNA methylation patterns due to the
semiconservative nature of DNA replication. This
process results in two hemimethylated daughter
chromosomes that must be fully methylated in order
for DNA methylation patterns to be properly maintained. In contrast to this simple model, the expression
patterns of DNMT1 in certain cell types has been
found to be somewhat paradoxical. If DNMT1
expression were strictly linked to DNA replication,
then expression of DNMT1 would correlate with
proliferative state of the cell. However, Northern blot
analysis showed that DNMT1 is highly expressed not
only in the placenta and lung, but also in low
proliferative tissues such as the heart and brain
(Robertson et al., 1999; Yen et al., 1992). This
unexpected result may suggest an additional function
for DNMT1 in addition to maintenance of methylation
at replication foci. Emerging lines of evidence suggest
that tissue-specific alternative splicing produces several
forms of DNMT1, which result in enzymes designed to
carry out distinct roles in DNA methylation metabolism.
DNMT1b
An alternative splice variant of human DNMT1 in
somatic tissues was reported by two groups and was
named DNMT1b (Bonfils et al., 2000; Hsu et al.,
1999). This isoform of DNMT1 contains an extra 48
base pairs between exons 4 and 5, generating a protein
DNA methylation and cancer
KD Robertson
5363
Figure 1 Schematic structure of all known mammalian DNA methyltransferases and DNMT-like proteins. Specific motifs involved
in localization, catalysis (PC active site), or of unknown function (HRX-like, ATRX-like, and PWWP) are indicated with boxes
with an additional 16 amino acids derived from this
region (Figure 2; Hsu et al., 1999). Dnmt1b mRNA is
ubiquitously expressed in all cell lines tested and the
protein possessed enzymatic properties comparable to
DNMT1 in vitro (Bonfils et al., 2000). One group
described the mRNA expression level of DNMT1b as
40 – 70% of the level of DNMT1 (Hsu et al., 1999),
while another group reported that DNMT1b protein
was present at only 2 – 5% of the level of DNMT1
protein (Bonfils et al., 2000). Thus, the relative
abundance of DNMT1b and DNMT1 in somatic
tissues remains unclear. The murine DNMT1 gene
also undergoes alternative splicing and potentially
generates an isoform (murine DNMT1b) similar to
human DNMT1b, but would differ by two amino acids
(Lin et al., 2000). A biological role for human and
murine DNMT1b in somatic tissues and the functional
differences between them in vivo remains unknown.
DNMT1o
Alternative splicing of sex-specific 5’ exons produces
at least two DNMT1 variants at the mRNA level,
which have been termed DNMT1o and DNMT1p
(Figure 2). A shorter isoform of DNMT1
(DNMT1o), which is specific to oocytes, yields a
large amount of enzymatically active protein which
accumulates in oocytes during the growth phase
(Carlson et al., 1992; Mertineit et al., 1998). In fact,
DNMT1o is the sole isoform of DNMT1 not only
in the oocyte but also in the preimplantation
embryo. It is noteworthy that DNMT1o exhibits
an unusual trafficking movement. The majority of
DNMT1o is localized in the cytoplasm, and
transiently relocates to the nucleus during the
eight-cell stage. This observation led to the suggestion that the cytoplasmic-nuclear translocation of
DNMT1o at a particular stage of embryogenesis is
essential for establishment of normal methylation
patterns in imprinted regions (Cardoso and Leonhardt, 1999; Doherty et al., 2002).
DNMT1p
A larger form of DNMT1 mRNA was detected
exclusively in pachytene spermatocytes (Mertineit et
al., 1998), and thus, the transcript was named
DNMT1p (Figure 2; Doherty et al., 2002). It was
originally concluded that DNMT1p was not translated,
possibly due to the inhibitory effect of multiple short
upstream open reading frames (Mertineit et al., 1998).
However, the patterns of transcription and translation
of this isoform remain controversial. In 2000, it was
reported that an alternative DNMT1 transcript in
skeletal muscle, specifically differentiated myotubes,
was identical to DNMT1p (Aguirre-Arteta et al.,
2000). It was also shown that this muscle-specific
isoform was translatable in both transfected cells and
in differentiated muscle cells. As with DNMT1o,
DNMT1p may play an important role during oogenesis, gametogenesis, or myogenesis.
DNMT2
In 1998, attempts to identify novel putative de novo
DNA methyltransferases in mammalian cells resulted
Oncogene
DNA methylation and cancer
KD Robertson
5364
Table 1
DNA methyltransferase-associated proteins involved in transcriptional repression and chromatin modification
DNA methyltransferase
Interacting protein
Function of interacting protein
How do they work together?
Dnmt1
HDAC1/2
Histone deacetylase
pRb
Tumor suppressor
Cell-cycle regulation
DMAP1
Co-repressor
PML-RAR
Oncogenic transcription factor
MBD2/3
Methyl-CpG binding proteins
Modification of chromatin by histone
deacetylation, targeting methylation?
Sequester DNMT1 in non-dividing cell,
target or modulate DNMT activity at
replication foci?
Recruiting other repressors, transcriptional
repression
DNA-binding and interaction with other
transcriptional co-regulators, targeting
methylation
Transcriptional repression in methylated
regions, possible targeting of DNMT1 to
hemimethylated DNA at replication foci?
HDAC1
Histone deacetylase
RP58
Transcription factor
PML-RAR
Oncogenic transcription factor
HDAC1
Histone deacetylase
SUMO-1/Ubc9
Sumo ligase
Dnmt3a
Dnmt3b
Modification of chromatin by histone
deacetylation, targeting methylation?
Sequence-specific DNA binding, targeting
repression, maybe methylation as well?
DNA-binding and interaction with other
transcriptional co-regulators, targeting
methylation
Modification of chromatin by histone
deacetylation, targeting methylation?
Modification of protein by sumoylation,
altered localization or enzymatic activity?
Figure 2 Alternatively spliced isoforms of DNMT1. The mRNA splicing structure of the relevant region of DNMT1 (the 5’ end of
the gene only) is indicated on the left and the resulting proteins are shown at the right. The presence or absence of important motifs
and protein – protein interaction sites is also shown for each of the DNMT1 variants
in the identification of the DNMT2 gene (Figure 1),
which shares homology with the pmt1 gene of fission
yeast (Okano et al., 1998b; Yoder and Bestor, 1998).
Although the human DNMT2 gene was mapped to
chromosome 10p12-10p14 (Yoder and Bestor, 1998),
the human genome project revealed that its actual
location is 10p15.1. DNMT2 mRNA is ubiquitously
expressed at very low levels (Okano et al., 1998b;
Oncogene
Yoder and Bestor, 1998). The characteristic motifs
found in all other active DNA methyltransferases,
including the conserved proline-cysteine dipeptide at
the active site, are well conserved in both the human
and murine DNMT2 genes. Interestingly, DNMT2 may
be the most highly conserved of all DNMTs across
species since, unlike DNMT1 and DNMT3, DNMT2like proteins have been found in yeast, plants, and the
DNA methylation and cancer
KD Robertson
5365
fruit fly Drosophila (Hung et al., 1999; Lyko et al.,
2000; The Arabidopsis Genome Initiative, 2000;
Wilkinson et al., 1995). At this time, however, no
methyl-transfer activity of the DNMT2 gene product
has been reported using in vitro assays (Dong et al.,
2001; Okano et al., 1998b; Yoder and Bestor, 1998),
although the DNMT2 protein does form stable DNAprotein complexes in vitro (Dong et al., 2001).
Elucidation of the exact function of this protein in
vivo will be of great importance. DNMT2 could, for
example, be the catalytic subunit of a DNA methylase
complex that is inactive when expressed without one or
more of its associated factors. Alternatively it may
possess a more complex DNA recognition sequence
beyond the CpG dinucleotide.
DNMT3a
Low level, stable de novo methylation activity could be
observed in embryonic stem (ES) cells lacking
DNMT1, supporting the notion that enzymes other
than DNMT1 contribute to de novo methylation in vivo
(Lei et al., 1996). DNMT3a and DNMT3b (Figure 1),
two genes sharing significant homology to DNMT1,
were identified by EST database searches in 1998 using
the conserved methyltransferase motifs as bait (Okano
et al., 1998a). The human DNMT3a gene was mapped
to chromosome 2p23 (Robertson et al., 1999) and
shows 96% amino acid identity to its murine counterpart (Xie et al., 1999). This is significantly higher than
the identity between human and murine DNMT1
(78%), and suggests a critical function for this enzyme
preserved throughout evolution. The carboxy terminal
portion of DNMT3a includes the catalytic motifs
highly conserved in all cytosine DNA methyltransferases. DNMT3a has been shown to be enzymatically
active both in vitro and in vivo by several groups,
although these reports differ somewhat in the exact
substrate preference of DNMT3a (Aoki et al., 2001;
Gowher and Jeltsch, 2001; Hsieh, 1999; Lyko et al.,
1999; Okano et al., 1998a; Yokochi and Robertson,
2002).
Proteins of the DNMT3 family commonly have a
cysteine-rich domain in the amino terminal region,
which is referred to as the PHD (plant homeodomain) region or ATRX-like domain (Figure 1)
because of its homology with the PHD region of the
ATRX gene. ATRX is a member of the SNF2/SWI2
family of ATP-dependent chromatin remodeling
enzymes. This similarity suggests that DNMT3
proteins may be associated with structural changes
in chromatin via protein – protein interactions at the
amino terminal region. DNMT3a transcripts were
ubiquitously expressed in adult tissues, most tumor
cell lines, early embryos, and embryonic stem (ES)
cells (Okano et al., 1998a; Robertson et al., 1999; Xie
et al., 1999). Northern blot analysis showed that
DNMT3a activity is particularly high at embryonic
day 10.5 (Okano et al., 1998a). Unlike DNMT3b, the
level of DNMT3a mRNA was less affected when cells
were arrested in the G0/G1 phase of the cell cycle,
suggesting that DNMT3a expression is not regulated
in a cell cycle-dependent manner like DNMT1
(Robertson et al., 2000b). In contrast to DNMT1,
DNMT3a was found at discrete nuclear foci throughout the cell cycle that were not associated with DNA
synthesis. During late S phase, however, when
heterochromatic regions of the genome are known
to be replicated, some of the DNMT3a-enriched foci
appeared to overlap with replication foci (Bachman
et al., 2001).
DNMT3b and its multiple isoforms
The human DNMT3b gene was mapped to chromosome 20q11.2 (Robertson et al., 1999; Xie et al., 1999)
and is 85% identical to murine DNMT3b (Figure 1).
The catalytic domain, located at the carboxy terminus,
is well conserved between DNMT3a and DNMT3b
(more than 80% identity), whereas their amino
terminal regions are poorly conserved (less than
30%). This further underscores the notion that the
variety in function of each of the DNMTs is most
likely due to their diverse N-terminal regions, while the
highly conserved C-terminal regions, which are
common among all mammalian DNMTs, have similar
functions. DNMT3b, like DNMT3a, was shown to be
an active DNA methylstransferase in vivo and in vitro
(Aoki et al., 2001; Hsieh, 1999; Okano et al., 1998a;
Qiu et al., 2002).
Compared to DNMT3a, the expression levels of
DNMT3b are very low in most tissues. The testis,
however, expressed high levels of DNMT3b, suggesting
a crucial function for DNMT3b in spermatogenesis
(Okano et al., 1998a; Robertson et al., 1999; Xie et al.,
1999). DNMT3b shows a diffuse distribution pattern
throughout the nucleus in NIH3T3 cells, while
targeting to pericentromeric heterochromatin was
observed in undifferentiated ES cells (Bachman et al.,
2001). As will be discussed later in this review in the
context of ICF syndrome, DNMT3b appears to be
specialized for the establishment and/or maintenance of
DNA methylation of the minor satellite repeats
(satellites 2 and 3 on human chromosomes 1, 9, and
16) and its co-localization with pericentromeric heterochromatin in consistent with this function.
In contrast to DNMT3a, there are several isoforms
(five for human and eight for mouse) of DNMT3b that
result from alternative splicing. Three major isoforms,
namely DNMT3b1, 3b2, and 3b3, have been identified
(Figure 3) (Okano et al., 1998a), and were shown to be
expressed in a tissue-specific manner (Robertson et al.,
1999). DNMT3b1 is the longest form and is usually
regarded as the typical gene product of DNMT3b. All
other splice variants encode smaller proteins.
DNMT3b2 also demonstrated methyltransferase activity in vitro, however DNMT3b3, an even shorter
isoform lacking 63 amino acids within the central
region of the catalytic domain, did not (Aoki et al.,
2001; Okano et al., 1998a; Qiu et al., 2002). Several
other splice variants, DNMT3b4 and DNMT3b5
(Figure 3) were identified and are expressed predomiOncogene
DNA methylation and cancer
KD Robertson
5366
Figure 3 Alternatively spliced isoforms of the human DNMT3b protein. Important motifs are denoted with boxes. Motifs I, IV,
VI, IX, and X correspond to the highly conserved methyltransferase motifs involved in catalysis. The shading in the amino-terminal
regions of DNMT3b4 and DNMT3b5 indicate that the exact splicing structure in region-1 has not been determined. The hatch
marks at the extreme C-terminus of DNMT3b5 denote a change in reading frame after splicing (Robertson et al., 1999)
nantly in the testis (Robertson et al., 1999). Interestingly, these forms of DNMT3b lack several of the
conserved catalytic motifs and possess additional novel
sequences resulting from frameshifts after alternative
splicing and are therefore unlikely to be catalytically
active. These splice variants may possess the critical
functions encoded by the N-terminal region of
DNMT3b, including protein – protein interaction sites,
but lack catalytic activity. They may therefore act to
inhibit some function of DNMT3b1-2, like de novo
methylation of certain sequences, by interacting with
the same set of targeting proteins.
DNMT3L
The human DNMT3L (DNMT3-like, Figure 1) gene
located on human chromosome 21q22.3 was originally
isolated by database analysis of the genome sequence
in 2000 (Aapola et al., 2000). The identification of the
murine DNMT3L gene (Aapola et al., 2001), which is
61% identical to its human counterpart, soon followed.
Similarity between DNMT3L and DNMT3a and
DNMT3b is restricted mainly to the cysteine-rich
region encompassing the PHD/ATRX-like region in
the amino terminus of DNMT3L. DNMT3L lacks the
critical catalytic motifs commonly seen in all other
DNA methyltransferases, including the ‘FGG’
sequence in conserved motif I, the ‘PC’ catalytic active
site in motif IV, and the ‘ENV’ sequence in motif VI
believed to be involved in cofactor binding. Thus, the
gene product is almost certainly catalytically inactive,
although this has yet to be tested directly. DNMT3L is
highly expressed in testis and mouse embryos (Aapola
et al., 2001), and is likely involved in gametogenesis for
the process of establishing genomic imprints (Bourc’his
et al., 2001).
Oncogene
Insights into functions of the DNA methylation
machinery derived from knockout models
The recent history of DNA methylation research has
been punctuated by several elegant studies using
murine knockout models to elucidate the role of
DNA methylation in general, and of the functions of
the individual DNMTs in development, transcriptional
regulation, chromatin structure maintenance and
genomic imprinting. Comparison of the transgenic
knockout mice with the corresponding knockout ES
cells provides additional data on the role of each of the
DNMTs in a pluripotent cell versus a more differentiated cell type. This section of the review will
summarize the effects of inactivating each of the
DNMTs at the individual cell and the whole organism
levels.
DNMT1
Murine ES cells lines homozygous for a DNMT1
knockout were obtained by targeted disruption (Li et
al., 1992). The mutant ES cells were viable and showed
normal morphology. Only residual DNA methyltransferase activity was observed in lysates derived from
these homozygous mutant cells. DNA from homozygous mutant cells was found to have a 5methylcytosine content of roughly 30% of the DNA
from wild type cells, in which about 60% of all CCGG
sites were methylated (that is, a 70% decrease in 5methylcytosine content in the CCGG sequence) (Li et
al., 1992). Interestingly, targeted disruption of the
DNMT1 gene in a human colorectal carcinoma cell line
yielded somewhat different results (Rhee et al., 2000).
DNA from the DNMT1 deficient HCT116 cell line had
a 5-methylcytosine content of about 80% of the DNA
DNA methylation and cancer
KD Robertson
5367
from wild type cells, in which approximately 4% of all
cytosines were methylated (that is, a 20% decrease in
total genomic 5-methylcytosine content). A direct
comparison of the results between these two experimental systems is somewhat difficult due to the
different methods used for measuring the amount of
DNA methylation in the original wild type cells.
Regardless of whether this discrepancy arises from
different experimental methods or from the radically
different cell types used in each study, these results
suggest that methyltransferases other than DNMT1
contribute significantly to the homeostasis of DNA
methylation patterns. Transgenic mice heterozygous for
the DNMT1 knockout were indistinguishable from
wild type littermates. Homozygous mutation of
DNMT1 was embryonic lethal and mutant embryos
failed to develop beyond mid-gestation, strongly
suggesting that DNA methylation is essential for
mammalian development. DNMT1 mutants demonstrated significant hypomethylation of several genes,
which affected a variety of epigenetic events including
genomic imprinting (Li et al., 1993), X chromosome
inactivation (Beard et al., 1995), and the suppression of
transcription from parasitic elements (Walsh et al.,
1998).
More recently, conditional knockouts of DNMT1
have been generated to examine its role in the
development and maintenance of specific tissues,
particularly the central nervous system. In one study,
the cre/lox system was used to delete the DNMT1 locus
from primary fibroblasts in culture (Jackson-Grusby et
al., 2001). The resulting DNMT1-null cells displayed
extensive genomic hypomethylation and uniform cell
death by apoptosis within 6 days of DNMT1-deletion.
The connection between reduced DNA methylation
and apoptosis was further underscored by showing that
interfering with p53 function, a key mediator of
apoptosis, prolonged the life span of the DNMT1-null
cells and reduced the level of apoptosis. Microarray
expression studies revealed that 4 – 10% of all genes
were upregulated and 1 – 2% were down-regulated
when DNMT1 was deleted and that these genes were
involved in a multitude of cellular functions (JacksonGrusby et al., 2001). Conditional deletion of DNMT1
in neural precursor cells early in development resulted
in genome-wide demethylation and embryonic lethality,
although brain structure appeared normal. Mice
delivered by caesarian section exhibited aberrant
breathing and their lungs failed to inflate, indicating
potential defects in respiratory rhythmogenesis or
neurotransmission. Deletion of DNMT1 at this early
developmental stage in a small fraction of neural
precursor cells resulted in a marked selection against
the demethylated cells such that the mice developed
normally and were devoid of demethylated, DNMT1deficient cells (Fan et al., 2001). Deletion of DNMT1
from post-mitotic neurons later in development was
compatible with both normal development and normal
DNA methylation patterns. Thus, these studies
strongly suggest that DNA methylation is critical for
normal mammalian development during embryogenesis
and may play a particularly important role in brain
development and function as has been suggested
previously (Fan et al., 2001; Robertson, 2001; Tucker,
2001).
DNMT1o
Since DNMT1o demonstrated a unique nuclearcytoplasmic trafficking movement during oogenesis
and preimplantation development, a functional role
for DNMTo in genomic imprinting in the oocyte was
proposed (Carlson et al., 1992; Mertineit et al., 1998).
Although males homozygous for a deletion of the
maternal-specific exon were normal and fertile, homozygous females were infertile. That is, most
heterozygous fetuses of homozygous females died
during the last third of gestation, indicating that
deletion of DNMT1o causes a pure-maternal phenotype during oogenesis (Howell et al., 2001). DNA
methylation at certain imprinted loci, but not the
whole genome, was lost in the heterozygous embryos.
Therefore it appears that DNMT1o is required to
maintain methylation patterns at specific imprinted loci
during the fourth embryonic S phase.
DNMT2
Murine ES cells with a homozygous disruption of the
DNMT2 gene appeared to be normal, suggesting that
DNMT2 is not essential for cell proliferation, growth,
and differentiation. No significant change in the
methylation status of both genomic DNA and newly
integrated provirus DNA was observed in the
DNMT27/7 ES cells. These results indicated that
DNMT2 is required for neither maintenance nor de
novo methylation in vivo (Okano et al., 1998b).
DNMT3
ES cell lines homozygous for a DNMT3a knockout
(DNMT3a7/7) retained their undifferentiated
morphology and showed normal de novo methylation
activity on newly integrated proviral DNA following
retroviral infection (Okano et al., 1999). DNMT3a
heterozygous mice were normal and fertile. Although
DNMT3a7/7 homozygous mice appeared to be
normal at birth, undergrowth of such mutant mice by
18 days was obvious and all of the animals died by 4
weeks of age.
A homozygous knockout of the DNMT3b gene in
ES cells yielded a phenotype similar to that of
DNMT3a disruption. DNMT3b7/7 ES cell lines
and embryos exhibited comparable degrees of demethylation to those of DNMT3a7/7 homozygous ES cells
and embryos and a comparable ability to de novo
methylate retroviral sequences (Okano et al., 1999).
However, an analysis of early embryos revealed clear
evidence that DNMT3a and DNMT3b have independent functions. Unlike DNMT3a7/7 mice, no viable
DNMT3b7/7 mice were recovered at birth. Interestingly, the minor satellite repeats in the pericentromeric
Oncogene
DNA methylation and cancer
KD Robertson
5368
region
were
significantly
hypomethylated
in
DNMT3b7/7 but not in DNMT3a7/7 cells,
indicating that the minor satellite repeats are specific
targets of DNMT3b. As will be discussed later,
naturally occurring mutations in the human DNMT3b
gene give rise to ICF syndrome, which is also
characterized by hypomethylation of pericentromeric
repeats (Hansen et al., 1999; Okano et al., 1999; Xu et
al., 1999).
DNMT3a7/7, DNMT3b7/7 double mutant ES
cells were also generated and were viable. However, de
novo methylation of newly integrated retroviral
sequences was no longer observed, suggesting that the
gene products of DNMT3a and DNMT3b have
overlapping functions in ES cells with regard to de
novo methylation of parasitic elements. Embryonic
defects in the double mutant mice were more severe
than those of DNMT3a7/7 single mutant mice.
Global methylation levels in ES cell lines and E9.5
day embryos were dramatically reduced in
DNMT3a7/7, DNMT3b7/7 knockout mutants
compared to the single mutants (Okano et al., 1999),
supporting the notion that DNMT3a and DNMT3b
have at least partially overlapping functions in the
establishment of cellular DNA methylation patterns
during development.
DNMT3L
Targeted disruption of DNMT3L gene was recently
reported (Bourc’his et al., 2001). Homozygous animals
of both sexes were viable but sterile. Bisulfite genomic
sequencing revealed that a defect in DNMT3L resulted
in the disruption of maternal methylation imprints in
homozygous oocytes, while genome-wide methylation
patterns appeared to be normal, indicating that the
DNMT3L protein contributes to establishment of
genomic imprints during oogenesis. As was discussed
in the previous section, DNMT3L is almost certainly
not a functional DNA methyltransferase therefore the
observation that its disruption results in any loss of
DNA methylation indicates that DNMT3L may target
other functional DNA methyltransferases, such as
DNMT3a and DNMT3b, to specific loci in the genome
via direct or indirect protein – protein interaction.
Chromatin-associated proteins that interact with
DNMTs
In vitro studies of the DNMTs have shown that they
exhibit little sequence preference beyond the CpG
dinucleotide, while the knockout studies emphasize the
non-random nature of DNA methylation and the
distinctive roles for the individual DNMTs. A prime
mediator of these distinctive roles is most likely the
complement of proteins that interact with the DNMTs
at specific stages of development and differentiation,
and within the environment of chromatin. A number of
proteins have now been identified which interact with
one or more of the DNMTs (Figure 4, Table 1;
Oncogene
Robertson, 2001). This section of the review will focus
specifically on chromatin-related proteins known to
associate with the DNMTs and discuss the functional
consequences of these interactions.
The retinoblastoma protein, Rb
The retinoblastoma protein, Rb, is a protein with
intimate links to transcriptional regulation in the
context of chromatin. Robertson et al. (2000a) initially
identified an interaction between Rb, E2F1, and
DNMT1 via biochemical fractionation, and this result
was recently confirmed by others (Figure 4; Pradhan
and Kim, 2002). The amino terminal region of
DNMT1 could interact with the A/B pocket region
of Rb (Robertson et al., 2000a), and also appears to be
capable of interacting with the B/C pocket (Pradhan
and Kim, 2002). Rb and DNMT1 cooperate to repress
transcription from E2F-responsive promoters in vivo
although this repression did not depend on the
catalytic activity of DNMT1 (Robertson et al.,
2000a). A recent report demonstrated that Rb binding
to DNMT1 inhibited the ability of DNMT1 to bind to
DNA in vitro and exerted a strong negative effect on
DNMT1 catalytic activity. Overexpression of Rb in
cells resulted in a significant reduction in total genomic
5-methylcytosine levels (Pradhan and Kim, 2002). Rb,
like DNMT1, also co-localizes with DNA replication
foci, specifically early S phase perinucleolar replication
foci (Kennedy et al., 2000). The role of Rb at these
sites remains unclear.
Hypophosphorylated Rb interacts with E2F family
members and inhibits their transactivation function. As
cells prepare to divide, Rb is phosphorylated and
dissociates from E2F, allowing it to recruit coactivators and stimulate transcription of genes involved
in cell cycle progression (Dyson, 1998). Interestingly,
Rb has been shown to repress transcription at E2Fresponsive promoters by recruitment of both HDACs
and histone methyltransferases (HMTs) and subsequent binding of the methylated lysine binding proteins
of the heterochromatin protein (HP) 1 family (Brehm
et al., 1998; Luo et al., 1998; Nielsen et al., 2001). As
will be discussed in the next section, DNMT1 can also
interact directly with HDAC1 and HDAC2 (Robertson
et al., 2000a; Rountree et al., 2000) and methylated
cytosine itself serves as a recognition site for another
class of repressors of the methyl-CpG binding protein
(MBD) family. The MBDs have also been shown to
recruit HDACs to methylated DNA and repress
transcription (Bird and Wolffe, 1999).
What are the functional consequences of the
interaction between Rb and DNMT1? We have
previously proposed a model wherein DNMT1 binding
to Rb at E2F-containing promoters may be a
mechanism to sequester DNMT1 from the genome
and prevent promiscuous DNA methylation events
(Robertson, 2001). This is supported by recent findings
that the catalytic activity of DNMT1 is inhibited by
binding to Rb (Pradhan and Kim, 2002). This study
also speculated that DNMT1 bound to Rb would not
DNA methylation and cancer
KD Robertson
5369
Figure 4 Summary of known DNMT1, DNMT3a, and DNMT3b interacting proteins. The interacting region on each DNMT protein is indicated with a horizontal bar. An arrow not pointing to a specific region indicates that the interaction domain has not been
mapped
be capable of binding to PCNA, another DNMT1associated protein thought to recruit DNMT1 to
replication foci. Therefore this model may require
revision to take into account these new observations.
The exact nature of the complexes formed between
DNMT1, Rb, and PCNA in silent chromatin versus
replication foci will require extensive additional study
however we propose a model (Figure 5) for how the
nature of these interactions may change during the cell
cycle and how they may effect the catalytic activity of
DNMT1. It is likely that temporal changes in the
complement of DNMT1-associated proteins, particularly Rb and HDAC2, which associate with early and
late replication foci, respectively, will be critical
modulators of DNMT1 activity at this site.
HDAC1 and HDAC2
Several studies have now shown that DNMT1,
DNMT3a, and DNMT3b associate with HDAC1 and
HDAC2 in vitro and in vivo (Figure 4 and Table 1;
Bachman et al., 2001; Fuks et al., 2000, 2001;
Robertson et al., 2000a). Histone deacetylases can
remove acetyl groups from the amino terminal core
histone tails, which are critical modulators of chromatin structure, leading to the assembly of tightly packed
chromatin and rendering the sequence inaccessible to
the transcription machinery (Jenuwein and Allis, 2001).
DNMT1-mediated transcriptional repression has been
shown to be comprised of both HDAC-dependent and
HDAC-independent components. The HDAC-dependency can be demonstrated using the HDAC inhibitor
trichostatin A (TSA), which relieves a substantial
amount of the DNMT1-mediated repression (Fuks et
al., 2000; Robertson et al., 2000a). DNMT1 was shown
to bind HDAC1 via a transcriptional repression region
other than the HRX-homology domain (Figure 4; Fuks
et al., 2000). A direct interaction between the amino
terminal half of DNMT1 and HDAC2 has also been
demonstrated (Figure 4; Rountree et al., 2000). At
present, there is no evidence to believe that there are
major functional differences between HDAC1 and
HDAC2 since they are highly similar proteins (85%
identical at the amino acid level in humans). Thus, the
interaction of DNMT1 with HDAC1 or HDAC2 likely
has the same functional consequence. The TSAinsensitive component of DNMT1 repression may be
mediated by its interaction with DNMT1-associated
protein (DMAP) 1 (Rountree et al., 2000). DMAP1
binds to the extreme amino terminus of DNMT1, a
region missing in certain germ-cell specific DNMT1
splice variants (Figures 2 and 4). DMAP1 interacts
with another potent transcriptional repressor TSG101,
although the existence of a DNMT1, DMAP1, and
TSG101 ternary complex has not been reported.
DMAP1 co-localizes with DNMT1 at replication foci
throughout S phase (Figure 5) and may affect catalytic
activity or targeting of DNMT1.
The DNMT3s also interact with HDAC1 through
their ATRX-homology/PHD regions (Bachman et al.,
Oncogene
DNA methylation and cancer
KD Robertson
5370
Figure 5 Model for how DNMT1 and Rb may function to regulate cellular DNA methylation patterns. In resting cells (top),
DNMT1 binding to Rb may inhibit the catalytic activity of DNMT1 and prevent unscheduled methylation of the genome. DNMT1
may also potentiate Rb-mediated repression of E2F-responsive genes. During early S phase (middle), DNMT1, PCNA, Rb, and
DMAP1 co-localize with replication foci. The catalytic activity of DNMT1 at this time, when transcriptionally active hypomethylated sequences are replicated, may be low due to the presence of Rb (denoted by thin arrow), to prevent aberrant methylation.
Alternatively, DNMT1 may be completely inactive in early S phase and maintenance methylation is accomplished by another
DNMT. In late S phase (bottom), Rb is no longer present at replication foci and HDAC2 begins to co-localize with DNMT1.
The departure of Rb may stimulate DNMT1 catalytic activity at a time when transcriptionally inactive hypermethylated regions
are replicated and a highly active enzyme would be necessary (denoted by heavy arrow). An ATP-dependent chromatin remodeling
enzyme or complex (‘ATPase’) appears to be crucial for proper DNA methylation throughout the replication process. Filled lollipops represent methylated CpG sites and open lollipops represent unmethylated CpG sites. The assembly of potential proteins during replication is shown for only one strand. The remainder of the macromolecular DNA replication machinery is not shown
2001; Fuks et al., 2001), which are not present in
DNMT1 (Figure 4 and Table 1). DNMT3a and
DNMT3b also exhibited a TSA-insensitive transcriptional repression component, however the protein –
protein interactions mediating this repression are
unknown (Bachman et al., 2001). Therefore, it is likely
that the repression capability of DNMT1 and
DNMT3s are the result of distinct protein – protein
interactions, although histone deacetylase activity
appears to be involved in both.
What is the functional consequence of the HDACDNMT interactions? This extremely important question remains unanswered. The potential importance is
further underscored by the now universal nature of the
association. All known catalytically active DNMTs
have been shown to interact with HDACs (Table 1).
One scenario that has been proposed previously relates
Oncogene
to the temporal association of DNMT1 and HDAC2
at replication foci (Figure 5). Heterochromatic,
hypermethylated sequences enriched in hypoacetylated
histones are replicated in late S phase. The recruitment
of HDAC2 to replication foci by DNMT1 may help to
coordinate remethylation of the newly synthesized
DNA and deacetylation of newly deposited histones
in late S phase (Rountree et al., 2000). It is possible
that DNMT recruitment to, and methylation of, a
specific genomic region then brings in HDACs to
deacetylate core histones in the newly methylated
region and allow for chromatin compaction and
transcriptional silencing. This implies that DNA
methylation is the primary event in transcriptional
silencing and chromatin modifications follow. Evidence
gathered from experimental systems in which gene
silencing occurs in a regulated manner, such as X
DNA methylation and cancer
KD Robertson
5371
chromosome inactivation in females and host genome
defense-mediated silencing and de novo methylation of
retroviral DNA, suggests that other mechanisms may
also exist. In X chromosome inactivation, transcriptional silencing and histone modifications (both
deacetylation and methylation) occur well before de
novo DNA methylation (Csankovszki et al., 2001;
Heard et al., 2001). Furthermore, transcriptional
silencing of newly introduced retroviral sequences
occurs before de novo methylation and can occur in
the complete absence of the de novo DNA methyltransferases (Pannell et al., 2000), which have been
shown to be responsible for methylation of this class of
sequence (Okano et al., 1999). Therefore an alternative
scenario is one in which HDACs are recruited to a
region destined to undergo long-term transcriptional
silencing via interaction with other proteins. Once the
region is deacetylated and possibly with the help of
other chromatin remodeling factors, the DNMTs gain
access to the DNA or are attracted to a particular
chromatin structural feature, the region is methylated,
and stable, heritable, long-term gene silencing and
chromatin compaction is achieved (Figures 5 and 6).
Alternatively, both models may operate and depend on
the transcriptional and replicative state of the cell.
PML – RAR
As has been mentioned previously, there is a great need
to better understand how cellular DNA methylation
patterns are targeted to, or restricted from, certain
regions of the genome. A fascinating recent study,
directly relevant to the targeting issue, showed that
DNMT1 and DNMT3a can interact with the leukemiapromoting PML – RAR fusion protein (Figure 4 and
Table 1; Di Croce et al., 2002). Acute promyelocytic
leukemia (APL) is caused by a reciprocal translocation
of the retinoic acid receptor a (RARa) gene on
chromosome 17 to one of several other chromosomes.
To generate the PML – RAR fusion, the most common
in APL, the RARa gene is fused to the promyelocytic
leukemia gene (PML) on chromosome 15. PML is a
critical component of discrete nuclear structures
referred to as PML-oncogenic domains (PODs, ND10
and nuclear bodies). PODs, which are composed of a
number of proteins, including Sp100, Sp140, SUMO-1
and Daxx, may be involved in transcriptional activation, DNA replication, and apoptosis (Li et al., 2000;
Muller et al., 2001; Zhong et al., 2000a,b). The PML –
RAR fusion protein, which retains the DNA and
ligand binding domains of RARa, disrupts the PODs,
but they can be restored by treatment of cells with
retinoic acid (RA). RA treatment and POD reorganization correlates with differentiation of the APL cells,
indicating that PODs may have a critical role in
differentiation of promyelocytes (Di Croce et al., 2002;
Zhong et al., 2000a).
Di Croce et al. (2002) showed that PML – RAR
could repress a model RA target gene, RARb2, and
that the repression coincided with de novo methylation
of the 5’ end of the endogenous RARb2 promoter CpG
island. Co-immunoprecipitation studies revealed an
interaction between PML – RAR and DNMT1 and
DNMT3a, and immunofluorescence studies showed colocalization of PML – RAR and DNMT1 and Dnmt3a
when over expressed. It appeared that DNMT3a could
interact with regions on both the PML and RAR
portions of the fusion protein. Treatments of cells with
the DNA methylation inhibitor 5-aza-2’-deoxycytidine
(5-aza-dC) or the HDAC inhibitor TSA, revealed that
the PML – RAR-mediated transcriptional repression
and differentiation blockage was due to the temporally
distinct recruitment of both histone deacetylases (at
early time points) and DNA methyltransferases (at
later time points) (Di Croce et al., 2002). The
functional significance of the interaction between
DNMT1 and DNMT3a and PML in normal cells is
unclear, although it is possible that the DNMTs may
carry out some critical function in the PODs.
Interestingly, another PML-associated POD protein
and transcriptional repressor, Daxx, has been shown to
associate with DNMT1 and it remains possible that
Daxx bridges the interaction between PML – RAR and
DNMT1 (Li et al., 2000; Michaelson et al., 1999;
Zhong et al., 2000b). Perhaps most significantly are the
implications of this work in cancer since there have
been no prior studies showing how regional hypermethylation events, common to so many tumor cells,
might occur.
RP58
DNMT3a and DNMT3b were recently shown to
interact directly with a protein called RP58 via the
ATRX-like domain (Figure 4 and Table 1; Fuks et al.,
2001). This is also the region of DNMT3a and
DNMT3b that interacts with HDAC1 (and most likely
HDAC2). RP58 is a sequence-specific zinc finger DNA
binding protein and transcriptional repressor associated with heterochromatin (Aoki et al., 1998; Meng
et al., 2000). It contains a POZ domain and several
Krüppel-type zinc fingers commonly seen in other
transcriptional repressors (Ryan et al., 1999). The
repression activity of RP58 was enhanced by coexpression of DNMT3a and the cooperative effect did
not require the catalytic activity of DNMT3a (Fuks et
al., 2001). This suggests that DNMT3a acts as a
structural component, rather than an active enzyme, in
this repression pathway. Nuclear localization studies
using a DNMT3a fragment lacking the catalytic
domain also support this notion. The isolated amino
terminal regulatory domain of DNMT3a co-localized
with heterochromatin-associated proteins like HP1a as
well as methyl-CpG binding proteins like MeCP2
(Bachman et al., 2001). Thus, the co-localization of
DNMT3a with these types of proteins suggests that it
may be an important component of densely methylated, pericentromeric heterochromatin. Although the
ability of DNMT3a and RP58 to cooperate in
transcriptional repression was not dependent on the
catalytic activity of DNMT3a, it still remains possible
that one function of RP58 could be to target DNMT3a
Oncogene
DNA methylation and cancer
KD Robertson
5372
Figure 6 Model for how HDACs, ATP-dependent chromatin remodeling enzymes, and DNA methyltransferases may cooperate to
set up region-specific DNA methylation patterns. The histones within a transcribed or transcriptionally competent region destined
for silencing (top) may first be deacetylated (and potentially methylated) and this likely initiates transcriptional silencing. It remains
to be determined exactly how the HDAC would first be targeted to this region. The chromatin remodeler can now recognize the
chromatin and mobilize nucleosomes in an ATP-dependent manner to allow the DNMT access to its target DNA sites or create
a particular chromatin signature or ‘epitope’ that is recognized by the DNMT or DNMT-containing complex. Ample evidence exists
for a DNMT-HDAC containing complex but it remains to be determined if the ATPase is directly or indirectly associated with the
DNMT complex. Once the region is methylated, the methylated cytosines will recruit methyl-CpG binding proteins (MBD) and
their associated co-repressors to further reinforce transcriptional silencing and chromatin compaction
to specific DNA sequences that are destined to become
de novo methylated. Post-translational modifications,
other protein cofactors expressed in a tissue-specific
manner (RP58 is highly expressed in the brain for
example (Meng et al., 2000)), missing from the cell
culture system used to first characterize this interaction,
could influence the function of the RP58 – DNMT3a
complex.
which contains a DNA modification module (DNMT),
a methylcytosine reocognition subunit (methyl-CpG
binding protein), and a histone modifying subunit
(HDAC). This complex could have roles in directing
DNMT1 to hemimethylated sequences following DNA
replication, silencing of genes during S phase, or
deacetylation of newly deposited histones in a manner
akin to the previously described DNMT-DMAP1HDAC2 interactions.
MBD2 and MBD3
A physical interaction between DNMT1 and methylCpG binding proteins was also reported (Figure 4 and
Table 1). DNMT1 was co-immunoprecipitated with
MBD2 and MBD3 as one potential complex (Tatematsu et al., 2000). MBD2 and MBD3 have recently
been reported to be components of the large macromolecular MeCP1 repressor complex, which is capable
of preferentially binding, remodeling, and deacetylating
methylated DNA-containing nucleosomes in vitro
(Feng and Zhang, 2001). MBD2 and MBD3 colocalized with DNMT1 at replication foci in late S
phase. Furthermore, the MBD2 – MBD3 complex
demonstrated binding affinity for both hemimethylated
and fully methylated DNA and repressed transcription
in a TSA-sensitive manner (Tatematsu et al., 2000).
Thus, these results suggest the possibility of an ultimate
‘all-in-one’ type transcriptional repression complex,
Oncogene
ICF syndrome – a disease caused by aberrant DNA
methylation and chromatin structure
The identification of numerous interactions between
DNA methyltransferases and chromatin associated
proteins like histone deacetylases, Rb, and RP58
provides a clear link between DNA methyltransferases
and transcriptional regulation and chromatin structure
modulation. Is there additional in vivo evidence,
particularly in human cells, that these processes are
connected? While knockout studies in mouse models
have been very revealing, the severity of the phenotype,
namely embryonic lethality for DNMT1 and
DNMT3b, limits the information that can be gained
to embryonic development. Are there models or
diseases involving less severe mutations in the DNA
methylation machinery that could provide clues? The
DNA methylation and cancer
KD Robertson
5373
recent finding of a human genetic disease caused by
mutations in a DNA methyltransferase gene, the only
such disease known, has provided a wealth of
information about the more subtle roles a particular
DNMT may play in determining genomic DNA
methylation patterns.
ICF syndrome (immunodeficiency, centromere
instability, facial anomalies) is a very rare recessive
disorder caused by mutations in the DNMT3b gene
(Hansen et al., 1999; Okano et al., 1999; Xu et al.,
1999). Most ICF patients are compound heterozygotes
for their DNMT3b mutations and, with one exception,
all of these mutations occur within the carboxy
terminal catalytic domain of DNMT3b and likely fully
or partially impair catalytic activity (Robertson, 2001;
Xu et al., 1999). Phenotypically, ICF syndrome is
characterized by a profound immunodeficiency with an
absence or severe reduction in at least two immunoglobulin isotypes, variable impairment in cellular
immunity, unusual facial features, neurologic and
intestinal dysfunction, and delayed developmental
milestones (Franceschini et al., 1995; Smeets et al.,
1994).
At the cytogenetic level, primary and cultured cells
from ICF patients exhibit marked elongation of
juxtacentromeric heterochromatin. This elongation
occurs most consistently on chromosomes 1 and 16,
and to a lesser extent chromosome 9. Abnormalities
which have been observed include multiradial chromosomes involving multiple arms (3 – 12) of one or more
of the decondensed chromosomes, whole-arm chromosome deletions or duplications, translocations,
centromeric breakage, and in rare cases telomeric
associations (Franceschini et al., 1995; Smeets et al.,
1994; Tuck-Muller et al., 2000). These observations
strongly suggest that defective forms of DNMT3b lead
to chromosome instability and large-scale changes in
chromatin structure.
Molecular aspects of ICF syndrome
One of the most consistent features of ICF syndrome is
juxtacentromeric repeat sequence hypomethylation on
chromosomes 1, 9, and 16 (Jeanpierre et al., 1993).
Interestingly, these chromosomes contain the largest
blocks of classical satellite long tandem repeat arrays
(satellite 2 for chromosomes 1 and 16, and satellite 3
for chromosome 9) adjacent to their centromeres
(Jeanpierre et al., 1993; Tagarro et al., 1994). These
regions are normally hypermethylated in somatic cells
and such methylation is likely essential for proper
centromere structure and stability (Jeanpierre et al.,
1993; Tuck-Muller et al., 2000). A recent study, using
bisulfite genomic sequencing, revealed that satellite 2
repeat methylation was reduced from roughly 70% in
normal lymphoblasts, to 20% in ICF cells (Hassan et
al., 2001). Although there is a drastic loss of
methylation from satellite sequences in ICF patients,
the overall reduction in cellular 5-methylcytosine levels
is relatively small (about 7% in primary ICF brain
tissue) further underscoring the idea that mutations in
DNMT3b lead to highly selective losses of methylation
from the genome (Kondo et al., 2000; Tuck-Muller et
al., 2000). Several other regions have been shown to
become hypomethylated in ICF patients including, asatellite and Alu repeats (Miniou et al., 1997;
Schuffenhauer et al., 1995), the non-satellite repeats
D4Z4 and NBL2 (Kondo et al., 2000), the H19 gene
(Schuffenhauer et al., 1995), and several genes (G6PD,
SYBL1, AR and PGK1) on the inactive X chromosome
of female ICF patients (Hansen et al., 2000). Biallelic
expression of several genes and advanced replication
timing of the normally late S phase replicating inactive
X chromosome have also been noted (Hansen et al.,
2000). The marked lack of autosomal genes in the list
of affected DNA sequences in ICF syndrome suggests a
specialized role for DNMT3b in gene-poor heterochromatin. The transcribed genes (as opposed to
repeats of some kind) most consistently affected in
ICF syndrome all appear to reside on the inactive X
chromosome, whose methylation is likely regulated in a
manner distinct from autosomal sequences.
DNA methylation and aberrant chromatin structure in
ICF syndrome
In ICF syndrome it appears that loss of methylation
from pericentromeric heterochromatin and the inactive
X chromosome results in aberrant chromatin structure.
Pericentromeric heterochromatin is massively decondensed and promoter regions of genes on the inactive
X chromosome showing reactivation demonstrate
increased susceptibility to nucleases, indicative of a
more open chromatin configuration (Hansen et al.,
2000). ICF syndrome also demonstrates that hypomethylation results in aberrant transcription, which
may also be directly related to chromatin structure
changes since a number of genes whose expression is
altered in ICF syndrome do not display DNA
methylation changes (Ehrlich et al., 2001). This
suggests that DNA methylation is critical for the
long-term maintenance of repressive condensed chromatin. Since the regions of the genome which
DNMT3b is responsible for methylating are never
properly methylated in ICF cells, the subsequent
recruitment of other proteins involved in maintaining
or reinforcing chromatin compaction, such as the
MBDs and their associated HDACs, or the MeCP1
repressor complex, never occurs (Bird, 2002; Feng and
Zhang, 2001). Therefore ICF syndrome can be
regarded not only as a disease of aberrant DNA
methylation patterns, but also of aberrant chromatin
structure. The chromatin structural changes may be
directly or indirectly related to the ability of DNMT3b
to act as a transcriptional repressor via both HDACdependent and independent mechanisms. Alternatively,
reduced methylation may result in global imbalances in
transcription factor binding by allowing transcription
factors access to sites that would normally be blocked
by closed chromatin configuration, or by excesses of
transcriptional repressors, such as the methyl-CpG
binding protein-containing complexes, that may gain
Oncogene
DNA methylation and cancer
KD Robertson
5374
promiscuous access to important transcriptional
control regions. The exact nature of defects in DNA
methylation and chromatin structure in ICF cells will
require considerable study, however, this disease
further emphasizes the connection between DNA
methylation, DNA methyltransferases, and chromatin
structure.
DNA methylation and chromatin remodeling
We have previously discussed connections between
DNA methyltransferases and chromatin in the context
of histone tail modifying proteins, namely histone
deacetylases. HDACs are associated with all active
DNA methyltransferases and likely play an important
role in determining which regions of the genome are
to be methylated. A significant body of data has
accumulated showing that hypermethylated regions
are transcriptionally inactive, enriched in hypoacetylated histones, and the chromatin is tightly packed,
reducing the access of transcription factors to these
regions (Eden et al., 1998). Work over the last few
years has provided tantalizing evidence for yet another
connection between DNA methylation and chromatin
– the possibility that DNA methylation is directly
connected to, or targeted by, chromatin remodeling
machines. In this last section I will discuss the
evidence for this linkage and propose models for
how these processes may be coordinated in mammalian cells.
ATP-dependent chromatin remodeling enzymes
Helicases are a large group of proteins, involved in a
host of RNA and DNA metabolic processes, which
possess a set of seven conserved motifs involved in
ATP binding and hydrolysis. The helicase group can be
broken down into several families. SNF2 family
members are involved in chromatin remodeling (ISWI),
transcription (SNF2), DNA repair (ERCC6), and
recombination (RAD54) (Table 2). SNF2 (sucrose
non-fermenter) was first identified in yeast as a gene
essential for transcription of genes involved in sucrose
fermentation and mating type switching (Eisen et al.,
1995). Since then, a large number of SNF2-like genes
have been discovered, most of which have homologs in
organisms ranging from yeast to humans. The SNF2
family can be further divided into three subfamilies, the
SNF2-like subfamily, the ISWI-like subfamily, and the
CHD subfamily, based on the presence of other
conserved domains (Table 2; Havas et al., 2001;
Varga-Weisz, 2001). None of the SNF2 family
members appear to act as helicases, instead they utilize
the energy derived from ATP hydrolysis to disrupt
histone-DNA interactions and allow for the physical
movement, or sliding, of nucleosomes on the DNA. In
this way SNF2 proteins can reorganize the chromatin
structure of a region, to one more permissive for
transcription factor binding and transcriptional activation, or to one with more regularly spaced, tightly
Oncogene
packed nucleosomes characteristic of transcriptionally
inactive heterochromatin (Vignali et al., 2000; Wolffe
and Pruss, 1996). Inherited mutations in SNF2-like
genes give rise to a number of human diseases,
including Werner, Bloom and Cockayne syndromes,
and Xeroderma pigmentosum (Ellis, 1997). In the next
section we will review the connections between DNA
methylation and chromatin remodeling enzymes of the
SNF2 family and finally propose a model for how
DNA methylation, histone modification, and chromatin remodeling may act in concert to establish and
maintain genomic DNA methylation patterns.
DDM1
The phenotypic consequences of mutations in the
Arabidopsis DDM1 (decrease in DNA methylation 1)
gene provided the first evidence that chromatin
remodeling may be essential for proper DNA methylation
patterning.
DDM1
is
not
a
DNA
methyltransferase, but rather a member of the SNF2
family (Jeddeloh et al., 1999). DDM1 does not readily
fit into one of the three SNF2 subfamilies listed in
Table 2 due to its lack of motifs commonly associated
with members of each family, but may be distantly
related to the ISWI-subfamily. Mutations in DDM1
result in loss of about 70% of the total genomic 5methylcytosine content, primarily at repetitive elements
like satellites and ribosomal DNA (Jeddeloh et al.,
1999; Martienssen and Henikoff, 1999). With increasing generations of DDM1 inbreeding, DNA
methylation at single copy loci, including imprinted
regions, is also lost, suggesting that DDM1 may be
involved in maintenance methylation following DNA
replication (Jeddeloh et al., 1999; Vielle-Calzada et al.,
1999). DDM1 mutant plants exhibited defects in
flowering time, floral and leaf morphology, and fertility
(Jeddeloh et al., 1999). Similar effects were observed in
plants expressing antisense to the Arabidopsis DNMT1like gene MET1, except that the defects in the MET1
plants tended to occur after fewer generations
(Finnegan et al., 1996). Demethylation and activation
of transposition from transposable elements has also
been observed in DDM1 mutant plants (Miura et al.,
2001).
ATRX
The ATRX gene is mutated in a human genetic disease
called X-linked, a-thalassemia mental retardation
(ATR-X) syndrome (Gibbons et al., 1997). Features
of this disease include developmental abnormalities,
severe mental retardation, facial dysmorphism, and athalassemia. Many of the mutations in the ATRX gene
occur within the PHD region, a region highly
homologous to the PHD regions of DNMT3a and
DNMT3b (Gibbons et al., 1997). Structurally, ATRX
belongs to the CHD subfamily (Table 2) of SNF2-like
proteins, although it has yet to be purified from cells
and shown to possess chromatin remodeling activity
(Havas et al., 2001). ATRX localizes to pericentromeric
DNA methylation and cancer
KD Robertson
5375
Table 2
Listing of SNF2-like proteins, and their proposed functions, in each of the three SNF2 subfamilies
SNF2 subfamily
Protein name
Proposed function(s)
Reference
SNF2-like
Swi2/Snf2 (S. cerevisiae)
DNA-stimulated ATPase, disrupt nucleosome
spacing and stimulate factor binding
Nucleosome spacing, transcription co-activator
DNA-dependent ATPase, remodeling
nucleosomal arrays, octamer transfer
Activities similar to BRG-1
Whitehouse et al., 1999
Nucleosome-dependent ATPase, generate
regularly-spaced nucleosomal array
Component of RSF, hACF, hCHRAC complexes,
DNA/histone/nucleosome stimulated ATPase,
nucleosome spacing and remodeling
Less well characterized, likely similar to
hSNF2H
Deuring et al., 2000;
Ito et al., 1999
LeRoy et al., 2000;
Poot et al., 2000
Nucleosome disruption activity
Mi-2 complex, nucleosome-stimulated ATPase,
interacts with HDAC
Similar to dCHD4, in a complex with MBD3,
DNA methylation-mediated gene silencing?
Unknown, DNA methylation remodeling?
Tran et al., 2000
Brehm et al., 2000
Brahma (Drosophila)
hBRG1 (Human)
hBRM (Human)
ISWI-like
ISWI (Drosophila)
hSNF2H
hSNF2L
CHD
CHD1 (S. cerevisiae)
CHD4 (Drosophila)
CHD4 (Xenopus)
ATRX
Kal et al., 2000
Wang et al., 1996
Havas et al., 2001
Wade et al., 1999
Gibbons et al., 2000
The list is by no means comprehensive and more complete descriptions can be found in Havas et al., 2001; Varga-Weisz, 2001; Vignali et al.,
2000
heterochromatin and may act as a transcriptional
regulator within a chromatin environment (Berube et
al., 2000). Interestingly, it was found that ATR-X
patients demonstrate DNA methylation defects in
select regions of the genome. The ribosomal DNA
repeats (where ATRX has also been shown to localize)
were significantly hypomethylated. Conversely, a Y
chromosome-specific repeat (DYZ2) was found to be
hypermethylated in ATR-X patients (Gibbons et al.,
2000). Therefore, unlike the effects of DDM1 mutation, ATRX mutations result in both aberrant losses
and gains in DNA methylation in the genome. This
result implies that aberrant chromatin structures, which
may be the result of an improperly functioning or
improperly targeted chromatin remodeling protein,
may be able to target DNA methylation to regions
that would not normally be methylated. This also
supports the notion that chromatin structure changes
may precede, and potentially dictate, patterns of DNA
methylation.
Lsh
Lsh (lymphoid-specific helicase, Hells, PASG), the
murine homolog of DDM1, was initially identified as
a protein highly expressed in lymphoid tissue and
thought to be involved in recombination (Jarvis et al.,
1996). Since then, its expression has been found to be
less restricted, however Lsh expression appears to be
tightly correlated with cell proliferation (Geiman and
Muegge, 2000; Raabe et al., 2001). Lsh knockout mice
were generated and developed relatively normally.
Lsh7/7 mice were born live but died shortly thereafter, possibly due to renal failure. The Lsh knockout
mice show a reduced birth weight, renal lesions, and
reduced numbers of lymphoid cells. T cells from
Lsh7/7 mice demonstrated defects in cell proliferation and high levels of apoptosis (Geiman and
Muegge, 2000). Given the homology between Lsh
and DDM1 (50% identity in the helicase region),
genomic DNA methylation patterns were examined in
the Lsh7/7 mice. Remarkably, the knockout mice
exhibited profound methylation defects (Dennis et al.,
2001). An examination of repetitive elements, including
the major and minor satellite repeats, intracisternal Aparticle retroviral sequences, and Sine B1 repeats, all of
which are normally heavily methylated in cells,
revealed significant hypomethylation. Several single
copy loci were also examined, including the b-globin,
Pgk-2 and Pgk-1 genes, and the imprinting control
region upstream of the H19 gene, and all revealed
losses of methylation relative to normal mice. Total
genomic 5-methylcytosine levels were reduced by 50 –
60% and all tissues appeared to be similarly affected
(Dennis et al., 2001). Thus it appeared that most, but
not all DNA methylation was affected by the Lsh
mutation and unlike ATRX mutation, resulted only in
losses of DNA methylation. No aberrant hypermethylation was reported. The Lsh work provides the most
compelling evidence that chromatin structure and
chromatin remodeling are critical determinants of
cellular DNA methylation patterns in mammalian
cells.
DNA methylation and chromatin structure – how does it
all fit together?
The consistent theme arising from studies on DDM1,
ATRX, and Lsh, is that these chromatin remodeling
enzymes, most likely acting within large macromolecular remodeling complexes, may remodel chromatin
Oncogene
DNA methylation and cancer
KD Robertson
5376
to allow the DNA methyltransferases access to their
target sites, or set up a particular chromatin configuration
which
is
recognized
by
DNA
methyltransferases and/or their associated proteins.
Thus one could imagine that, (1) the chromatin
remodeling enzyme is directly associated with a DNA
methylase complex which can both remodel chromatin
and methylate DNA or (2), the chromatin remodeling
complex may remodel a particular region destined to
undergo DNA methylation, depart, then the DNA
methylation machinery would follow. The work with
DDM1, ATRX, and Lsh firmly establishes an indirect
connection between DNA methylation and chromatin
remodeling but has not provided any evidence for a
direct connection, such as a direct interaction, between
a chromatin remodeling enzyme and a DNA methyltransferase (or DNA methyltransferase-associated
protein). Given the universal association of HDACs
with DNMTs it does not seem unreasonable to
speculate that a similar direct association between
one or more of the DNMTs and SNF2 family
members may soon come to light.
A model which has been proposed previously is that
proteins like DDM1 and Lsh may facilitate access of
the DNMT to newly replicated DNA following DNA
replication (Martienssen and Henikoff, 1999). DDM1
and Lsh may therefore remodel newly deposited
histones in such a way as to allow DNMT1 to carry
out its maintenance methylation function. The hemimethylated DNA would be converted to fully
methylated DNA, and the associated HDAC2 may
then deacetylate newly deposited histones and allow for
maintenance of a heterochromatic state once the DNA
replication machinery has passed (Rountree et al.,
2000). The strict correlation between Lsh expression
and cell proliferation lends support for this idea
(Raabe et al., 2001). This model is highly feasible for
methylation in the context of DNA replication
(maintenance methylation), however evidence from
other cellular processes, particularly related to de novo
methylation and methylation remodeling events occurring during embryogenesis, indicates that there may be
additional mechanisms. As was mentioned previously,
silencing of newly introduced retroviral sequences, and
of the X chromosome copy destined for inactivation,
occurs well before de novo DNA methylation of those
sequences. A similar situation may also be occurring
during the rapid genome-wide methylation remodeling
events (both demethylation and de novo methylation)
that occur during embryogenesis (Reik et al., 2001).
Therefore chromatin remodeling and histone modification would seem to set the stage for DNA methylation
in some cases. Additional support for this notion
comes from recent studies in Neurospora and Arabidopsis, which showed that loss of histone methylation
(on histone H3, lysine 9) resulted in a complete or
partial loss of genomic DNA methylation, respectively
(Tamaru and Selker, 2001; Jackson et al., 2002).
Homologous HMTs exist in mammalian cells (Rea et
al., 2000), but it remains unclear whether a similar
control system exists in mammals, which, unlike
Oncogene
Neurospora, absolutely require DNA methylation for
proper development.
How might DNA methylation and chromatin
remodeling be coupled mechanistically? How are the
HDACs also involved in this process? Unfortunately,
the in vitro biochemical properties of the DDM1,
ATRX and Lsh proteins have not been investigated.
The biochemical properties of other SNF2-like chromatin remodeling enzymes, including ones in the ISWI
subfamily, have been investigated in considerable detail
and we may be able to gain insights into potential
mechanisms of Lsh or DDM1 using the properties of
these related proteins as a model. In the case of
recombinant Drosophila ISWI, or the ISWI-containing
complex NURF, it has been demonstrated that the
histone H4 tail, which protrudes from the nucleosomal
core particle, is essential for ISWI chromatin remodeling activity (Clapier et al., 2001; Corona et al., 2002;
Hamiche et al., 2001). The H3 and H4 tails are subject
to numerous post-translation modifications including
acetylation, methylation and phosphorylation (Jenuwein and Allis, 2001). In particular, it appears that the
base of the H4 tail, combined with nucleosome-bound
DNA, is the ‘epitope’ recognized by ISWI in
chromatin. Interestingly, it was recently shown that
acetylation of lysines (K12 and K16) near the base of
the H4 tail inhibited the chromatin remodeling ability
of ISWI (Clapier et al., 2002). Indirect genetic evidence
presented in the previous section indicates that
chromatin remodeling is essential for proper DNA
methylation patterns. Thus we propose a model
(Figure 6) where these three activities, histone
deacetylase, ATPase, and DNA methyltransferase, rely
on each other. In this model, histone deacetylation
would occur first. This may be the point at which
transcriptional shutdown of the region would occur.
Access of the HDACs themselves to chromatin may
require other chromatin remodeling activities (Tong et
al., 1998). Following histone deacetylation (and
possibly methylation), the chromatin remodeling
complex binds and alters the chromatin structure or
nucleosome positioning in such a way as to directly
facilitate access of the DNMT to the nucleosomal
DNA or create a particular ‘epitope’ or signature
recognized by a DNMT or DNMT-associated
complex. The region is then methylated which locks
the chromatin in a silent mode, which would then be
further enhanced by the recruitment of methyl-CpG
binding proteins and their associated repressive activities (Bird, 2002). There is ample evidence for DNMTs
and HDACs as components of one complex. The ATPdependent chromatin remodeling enzyme may also be
part of the complex or act separately. Although
speculation at this time, this model is testable using
chromatin reconstituted in vitro and recombinant
remodeling proteins and will likely be the subject of
future studies. Previously proposed models, where
DNA methylation occurs first and is followed by
histone deacetylation and gene silencing (Baylin et al.,
2001), may also operate and are by no means excluded
by the model proposed here.
DNA methylation and cancer
KD Robertson
5377
Concluding remarks
The past decade has seen amazing advances in our
understanding of the ways in which DNA methylation
contributes to transcriptional regulation and tumorigenesis. Findings from the last few years in particular
have revealed the first glimpses of how the myriad
epigenetic control mechanisms that exist in mammalian
cells, including DNA methylation, histone acetylation,
histone methylation, and ATP-dependent nucleosome
positioning, may be directly connected. Histone
acetylation and methylation patterns may recruit
certain chromatin remodeling activities, which in turn
may create docking sites for DNA methyltransferase
complexes. The CpG methylated region may recruit
further chromatin modulatory activities, such as the
methyl-CpG binding protein complexes and their
associated repressive activities, and finally lock a given
region of the genome in a silent state. Emerging
evidence strongly suggests that histone modifications
set the stage for DNA methylation. Issues that still
need to be resolved include the following: (1) which
ATP-dependent chromatin remodeling machines are
involved in establishing and maintaining DNA methylation patterns; (2) is the association between the
ATPases and the DNMTs direct or indirect; (3) does
the nature of the interaction change between undifferentiated, differentiated, and transformed cellular states;
(4) do each of the three catalytically active DNMTs
recognize the same chromatin ‘epitope’ for methylation
or do the de novo and maintenance methyltransferases
utilize fundamentally different signals, and finally; (5)
have we indeed accounted for all of the DNA
methylating activities in mammalian cells? Such
important questions will likely be answered in the next
few years as intensive research identifies and characterizes DNMT-associated proteins, and in vitro DNA
methylation and chromatin remodeling systems are
established. Such findings will likely be highly relevant
to diseases involving aberrant DNA methylation
patterns, including cancer, as well as ICF, Rett, and
ATRX syndromes, and may provide the understanding
to devise completely novel strategies for reversing the
defects. Thus, our blurred image of the ‘tangled web’
of DNA methylation, histone modifications, and
chromatin remodeling is gradually resolving into that
of a precisely patterned, logically woven fabric.
Acknowledgments
KD Robertson is a Cancer Scholar supported by the
National Cancer Institute (NIH grant CA84535-01). I
thank Andrea Kahler Robertson for critical reading of
the manuscript. This article is dedicated to my former
mentor and friend, Alan Wolffe, who died tragically last
year.
References
Aapola U, Lyle R, Krohn K, Antonarakis SE and Peterson
P. (2001). Cytogenet. Cell Genet., 92, 122 – 126.
Aapola U, Shibuya K, Scott HS, Ollila J, Vihinen M, Heino
M, Shintani A, Kawasaki K, Minoshima S, Krohn K,
Antonarakis SE, Shimizu N, Kudoh J and Peterson P.
(2000). Genomics, 65, 293 – 298.
Aguirre-Arteta AM, Grunewald I, Cardoso MC and
Leonhardt H. (2000). Cell Growth Differ., 11, 551 – 559.
Aoki A, Suetake I, Miyagawa J, Fujio T, Chijiwa T, Sasaki H
and Tajima S. (2001). Nucleic Acids Res., 29, 3506 – 3512.
Aoki K, Meng G, Suzuki K, Takashi T, Kameoka Y,
Nakahara K, Ishida R and Kasai M. (1998). J. Biol.
Chem., 273, 26698 – 26704.
Bachman KE, Rountree MR and Baylin SB. (2001). J. Biol.
Chem., 276, 32282 – 32287.
Bacolla A, Pradhan S, Roberts RJ and Wells RD. (1999). J.
Biol. Chem., 274, 33011 – 33019.
Baylin SB, Esteller M, Rountree MR, Bachman KE,
Schuebel K and Herman JG. (2001). Hum. Mol. Genet.,
10, 687 – 692.
Beard C, Li E and Jaenisch R. (1995). Genes. Dev., 9, 2325 –
2334.
Berube NG, Smeenk CA and Picketts DJ. (2000). Hum. Mol.
Genet., 9, 539 – 547.
Bestor T, Laudano A, Mattaliano R and Ingram V. (1988). J.
Mol. Biol., 203, 971 – 983.
Bestor TH. (2000). Hum. Mol. Genet., 9, 2395 – 2402.
Bird A. (2001). Science, 294, 2113 – 2115.
Bird A. (2002). Genes. Dev., 16, 6 – 21.
Bird AP and Wolffe AP. (1999). Cell, 99, 451 – 454.
Bonfils C, Beaulieu N, Chan E, Cotton-Montpetit J and
MacLeod AR. (2000). J. Biol. Chem., 275, 10754 – 10760.
Bourc’his D, Xu G-L, Lin C-S, Bollman B and Bestor TH.
(2001). Science, 294, 2536 – 2539.
Brehm A, Langst G, Kehle J, Clapier CR, Imhof A,
Eberharter A, Muller J and Becker PB. (2000). EMBO
J., 19, 4332 – 4341.
Brehm AB, Miska EA, McCance DJ, Reid JL, Bannister AJ
and Kouzarides T. (1998). Nature, 391, 597 – 601.
Cameron EE, Bachman KE, Myohanen S, Herman JG and
Baylin SB. (1999). Nature Genet., 21, 103 – 107.
Cardoso MC and Leonhardt H. (1999). J. Cell Biol., 147,
25 – 32.
Carlson LL, Page AW and Bestor TH. (1992). Genes. Dev., 6,
2536 – 2541.
Clapier CR, Langst G, Corona DFV, Becker PB and
Nightingale KP. (2001). Mol. Cell. Biol., 21, 875 – 883.
Clapier CR, Nightingale KP and Becker PB. (2002). Nucleic
Acids Res., 30, 649 – 655.
Corona DFV, Clapier CR, Becker PB and Tamkun JW.
(2002). EMBO Rep., 3, 242 – 247.
Csankovszki G, Nagy A and Jaenisch R. (2001). J. Cell Biol.,
153, 773 – 783.
Dennis K, Fan T, Geiman T, Yan Q and Muegge K. (2001).
Genes. Dev., 15, 2940 – 2944.
Deuring R, Fanti L, Armstrong JA, Sarte M, Papoulas O,
Prestel M, Daubresse G, Verardo M, Moseley SL, Berloco
M, Tsukiyama T, Wu C, Pimpinelli S and Tamkun JW.
(2000). Cell, 5, 355 – 365.
Di Croce L, Raker AA, Corsaro M, Fazi F, Fanelli M,
Faretta M, Fuks F, Coco FL, Kouzarides T, Nervi C,
Minucci S and Pelicci PG. (2002). Science, 295, 1079 –
1082.
Oncogene
DNA methylation and cancer
KD Robertson
5378
Doherty AS, Bartolomei MS and Schultz RM. (2002). Dev.
Biol., 242, 255 – 266.
Dong A, Yoder JA, Zhang X, Zhou L, Bestor TH and Cheng
X. (2001). Nucleic Acids Res., 29, 439 – 448.
Dyson N. (1998). Genes. Dev., 12, 2245 – 2262.
Eden S, Hashimshony T, Keshet I, Cedar H and Thorne AW.
(1998). Nature, 394, 842.
Ehrlich M, Buchanan KL, Tsein F, Jiang G, Sun B, Uicker
W, Weemaes CMR, Smeets D, Sperling K, Belohradsky
BH, Tommerup N, Misek DE, Rouillard J-M, Kuick R
and Hanash SM. (2001). Hum. Mol. Genet., 10, 2917 –
2931.
Eisen JA, Sweder KS and Hanawalt PC. (1995). Nucleic
Acids Res., 23, 2715 – 2723.
Ellis NA. (1997). Curr. Opin. Genet. Dev., 7, 354 – 363.
Fan G, Beard C, Chen RZ, Csankovszki G, Sun Y, Siniaia
M, Biniszkiewicz D, Bates B, Lee PP, Kuhn R, Trumpp A,
Poon C-S, Wilson CB and Jaenisch R. (2001). J. Neurosci.,
21, 788 – 797.
Feng Q and Zhang Y. (2001). Genes. Dev., 15, 827 – 832.
Finnegan EJ, Peacock WJ and Dennis ES. (1996). Proc. Natl.
Acad. Sci. USA, 93, 8449 – 8454.
Flynn J, Glickman JF and Reich NO. (1996). Biochemistry,
35, 7308 – 7315.
Franceschini P, Martino S, Ciocchini M, Ciuti E, Vardeu
MP, Guala A, Signorile F, Camerano P, Franceschini D
and Tovo PA. (1995). Eur. J. Pediatr., 154, 840 – 846.
Fuks F, Bergers WA, Brehm A, Hughes-Davies L and
Kouzarides T. (2000). Nature Genet., 24, 88 – 91.
Fuks F, Burgers WA, Godin N, Kasai M and Kouzarides T.
(2001). EMBO J., 20, 2536 – 2544.
Geiman TM and Muegge K. (2000). Proc. Natl. Acad. Sci.
USA, 97, 4772 – 4777.
Gibbons RJ, Bachoo S, Picketts DJ, Aftimos S, Asenbauer
B, Bergoffen J, Berry SA, Dahl N, Fryer A, Keppler K,
Kurosawa K, Levin ML, Masuno M, Neri G, Pierpont
ME, Slaney SF and Higgs DR. (1997). Nature Genet,, 17,
146 – 148.
Gibbons RJ, McDowell TL, Raman S, O’Rourke DM,
Garrick D, Ayyub H and Higgs DR. (2000). Nature Genet.,
24, 368 – 371.
Glickman JF, Pavlovich JG and Reich NO. (1997). J. Biol.
Chem., 272, 17851 – 17857.
Gowher H and Jeltsch A. (2001). J. Mol. Biol., 309, 1201 –
1208.
Hamiche A, Kang J-G, Dennis C, Xiao H and Wu C. (2001).
Proc. Natl. Acad. Sci. USA, 98, 14316 – 14321.
Hansen RS, Stoger R, Wijmenga C, Stanek AM, Canfield
TK, Luo P, Matarazzo MR, D’Esposito M, Feil R, Gimelli
G, Weemaes CMR, Laird CD and Gartler SM. (2000).
Hum. Mol. Genet., 9, 2575 – 2587.
Hansen RS, Wijmenga C, Luo P, Stanek AM, Canfield TK,
Weemaes CMR and Gartler SM. (1999). Proc. Natl. Acad.
Sci. USA, 96, 14412 – 14417.
Hassan KMA, Norwood T, Gimelli G, Gartler SM and
Hansen RS. (2001). Hum. Mol. Genet., 109, 452 – 462.
Havas K, Whitehouse I and Owen-Hughes T. (2001). Cell.
Mol. Life Sci., 58, 673 – 682.
Heard E, Rougeulle C, Arnaud D, Avner P, Allis CD and
Spector DL. (2001). Cell, 107, 727 – 738.
Howell CY, Bestor TH, Ding F, Latham KE, Mertineit C,
Trasler JM, Trasler JM and Chaillet JR. (2001). Cell, 104,
829 – 838.
Hsieh C-L. (1999). Mol. Cell. Biol., 19, 8211 – 8218.
Hsu D-W, Lin M-J, Lee T-L, Wen S-C, Chen X and Shen CKJ. (1999). Proc. Natl. Acad. Sci. USA, 96, 9751 – 9756.
Oncogene
Hung M-S, Karthikeyan N, Huang B, Koo H-C, Kiger J and
Shen C-KJ. (1999). Proc. Natl. Acad. Sci. USA, 96,
11940 – 11945.
Ito T, Levenstein ME, Fyodorov DV, Kutach AK,
Kobayashi R and Kadonaga JT. (1999). Genes. Dev., 13,
1529 – 1539.
Jackson JP, Lindroth AM, Cao X and Jacobsen SE. (2002).
Nature, advance online publication, 17 March..
Jackson-Grusby L, Beard C, Possemato R, Tudor M,
Fambrough D, Csankovzki G, Dausman J, Lee P, Wilson
C, Lander E and Jaenisch R. (2001). Nature Genet., 27,
31 – 39.
Jarvis CD, Geiman T, Vila-Storm MP, Osipovich O, Akella
U, Candeias S, Nathan I, Durum SK and Muegge K.
(1996). Gene, 169, 203 – 207.
Jeanpierre M, Turleau C, Aurias A, Prieur M, Ledeist F,
Fischer A and Viegas-Pequignot E. (1993). Hum. Mol.
Genet., 2, 731 – 735.
Jeddeloh JA, Stokes TL and Richards EJ. (1999). Nature
Genet., 22, 94 – 97.
Jenuwein T and Allis CD. (2001). Science, 293, 1074 – 1080.
Jones PA and Laird PW. (1999). Nature Genet., 21, 163 – 166.
Kal AJ, Mahmoudi T, Zak NB and Verrijzer CP. (2000).
Genes Dev., 14, 1058 – 1071.
Kennedy BK, Barbie DA, Classon M, Dyson N and Harlow
E. (2000). Genes Dev., 14, 2855 – 2868.
Kondo T, Bobek MP, Kuick R, Lamb B, Zhu X, Narayan A,
Bourc’his D, Viegas-Pequignot E, Ehrlich M and Hanash
SM. (2000). Hum. Mol. Genet., 9, 597 – 604.
Lei H, Oh SP, Okano M, Juttermann R, Goss KA, Jaenisch
R and Li E. (1996). Development, 122, 3195 – 3205.
Leonhardt H, Page AW, Weier H and Bestor TH. (1992).
Cell, 71, 865 – 873.
LeRoy G, Loyola A, Lane WS and Reinberg D. (2000). J.
Biol. Chem., 275, 14787 – 14790.
Li E, Beard C and Jaenisch R. (1993). Nature, 366, 362 – 365.
Li E, Bestor TH and Jaenisch R. (1992). Cell, 69, 915 – 926.
Li H, Leo C, Zhu J, Wu X, O’Neil J, Park E-J and Chen JD.
(2000). Mol. Cell. Biol., 20, 1784 – 1796.
Lin M-J, Lee T-L, Hsu D-W and Shen C-KJ. (2000). FEBS
Lett., 469, 101 – 104.
Liu Y, Oakeley EJ, Sun L and Jost J-P. (1998). Nucleic Acids
Res., 26, 1038 – 1045.
Luo RX, Postigo AA and Dean DC. (1998). Cell, 92, 463 –
473.
Lyko F, Ramsahoye BH, Kashevsky H, Tudor M,
Mastrangelo M-A, Orr-Weaver TL and Jaenisch R.
(1999). Nature Genet., 23, 363 – 366.
Lyko F, Whittaker AJ, Orr-Weaver TL and Jaenisch R.
(2000). Mech. Dev., 95, 215 – 217.
Martienssen R and Henikoff S. (1999). Nature Genet., 22, 6 –
7.
Meng G, Inazawa J, Ishida R, Tokura K, Nakahara K, Aoki
K and Kasai M. (2000). Gene, 242, 59 – 64.
Mertineit C, Yoder JA, Taketo T, Laird DW, Trasler JM and
Bestor TH. (1998). Development, 125, 889 – 897.
Michaelson JS, Bader D, Kuo F, Kozak C and Leder P.
(1999). Genes. Dev., 13, 1918 – 1923.
Miniou P, Jeanpierre M, Bourc’his D, Barbosa ACC,
Blanquet V and Viegas-Pequignot E. (1997). Hum. Genet.,
99, 738 – 745.
Miura A, Yonebayashi S, Watanabe K, Toyama T, Shimada
H and Kakutani T. (2001). Nature, 411, 212 – 214.
Muller S, Hoege C, Pyrowolakis G and Jentsch S. (2001).
Nature Rev. Mol. Cell Biol., 2, 202 – 210.
DNA methylation and cancer
KD Robertson
5379
Nielsen SJ, Schneider R, Bauer U-M, Bannister AJ,
Morrison A, O’Carroll D, Firestein R, Cleary M, Jenuwein
T, Herrera RE and Kouzarides T. (2001). Nature, 412,
561 – 565.
Okano M, Bell DW, Haber DA and Li E. (1999). Cell, 99,
247 – 257.
Okano M, Xie S and Li E. (1998a). Nature Genet., 19, 219 –
220.
Okano M, Xie S and Li E. (1998b). Nucleic Acids Res., 26,
2536 – 2540.
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.
Poot RA, Dellaire G, Hulsmann BB, Grimaldi MA, Corona
DFV, Becker PB, Bickmore WA and Varga-Weisz PD.
(2000). EMBO J., 19, 3377 – 3387.
Pradhan S, Bacolla A, Wells RD and Roberts RJ. (1999). J.
Biol. Chem., 274, 33002 – 33010.
Pradhan S and Kim G-D. (2002). EMBO J., 21, 779 – 788.
Pradhan S, Talbot D, Sha M, Benner J, Hornstra L, Li E,
Jaenisch R and Roberts RJ. (1997). Nucleic Acids Res., 25,
4666 – 4673.
Qiu C, Sawada K, Zhang X and Cheng X. (2002). Nature
Struct. Biol., 9, 217 – 224.
Raabe EH, Abdurrahman L, Behbehani G and Areci RJ.
(2001). Dev. Dyn., 221, 92 – 105.
Rea S, Eisenhaber F, O’Caroll D, Strahl BD, Sun Z-W,
Schmid M, Opravil S, Mechtler K, Ponting CP, Allis CD
and Jenuwein T. (2000). Nature, 406, 593 – 599.
Reik W, Dean W and Walter J. (2001). Science, 293, 1089 –
1093.
Rhee I, Jair K-W, Yen R-WC, Lengauer C, Herman JG,
Kinzler KW, Vogelstein B, Baylin SB and Schuebel KE.
(2000). Nature, 404, 1003 – 1007.
Robertson KD. (2001). Oncogene, 20, 3139 – 3155.
Robertson KD, Ait-Si-Ali S, Yokochi T, Wade PA, Jones PL
and Wolffe AP. (2000a). Nature Genet., 25, 338 – 342.
Robertson KD, Keyomarsi K, Gonzales FA, Velicescu M
and Jones PA. (2000b). Nucleic Acids Res., 28, 2108 – 2113.
Robertson KD, Uzvolgyi E, Liang G, Talmadge C, Sumegi J,
Gonzales FA and Jones PA. (1999). Nucleic Acids Res., 27,
2291 – 2298.
Robertson KD and Wolffe AP. (2000). Nature Rev. Genet., 1,
11 – 19.
Rountree MR, Bachman KE and Baylin SB. (2000). Nature
Genet., 25, 269 – 277.
Ryan RF, Schultz DC, Ayyanathan K, Singh PB, Friedman
JR, Fredericks WJ and Rasucher FR. (1999). Mol. Cell.
Biol., 19, 4366 – 4378.
Schuffenhauer S, Bartsch O, Stumm M, Buchholz T,
Petropoulou T, Kraft S, Belohradsky B, Hinkel GK,
Meitinger T and Wegner R-D. (1995). Hum. Genet., 96,
562 – 571.
Smeets DFCM, Moog U, Weemaes CMR, Vaes-Peeters G,
Merkx GFM, Niehof JP and Hamers G. (1994). Hum.
Genet., 94, 240 – 246.
Tagarro I, Fernandez-Peralta AM and Gonzales-Aguilera
JJ. (1994). Hum. Genet., 93, 383 – 388.
Tamaru H and Selker EU. (2001). Nature, 414, 277 – 283.
Tatematsu K-i, Yamazaki T and Ishikawa F. (2000). Genes
Cells, 5, 677 – 688.
The Arabidopsis Genome Initiative. (2000). Nature, 408,
796 – 815.
Tong JK, Hassig CA, Schnitzler GR, Kingston RE and
Schreiber SL. (1998). Nature, 395, 917 – 921.
Tran HG, Steger DJ, Iyer VR and Johnson AD. (2000).
EMBO J., 19, 2323 – 2331.
Tucker KL. (2001). Neuron, 30, 649 – 652.
Tuck-Muller CM, Narayan A, Tsien F, Smeets DFCM,
Sawyer J, Fiala ES, Sohn OS and Ehrlich M. (2000).
Cytogenet. Cell Genet., 89, 121 – 128.
Varga-Weisz P. (2001). Oncogene, 20, 3076 – 3085.
Vielle-Calzada J-P, Thomas J, Spillane C, Coluccio A,
Hoeppner MA and Grossniklaus U. (1999). Genes Dev.,
13, 2971 – 2982.
Vignali M, Hassan AH, Neely KE and Workman JL. (2000).
Mol. Cell. Biol., 20, 1899 – 1910.
Wade PA, Gegonne A, Jones PL, Ballestar E, Aubry F and
Wolffe AP. (1999). Nature Genet., 23, 62 – 66.
Walsh CP, Chaillet JR and Bestor TH. (1998). Nature Genet.,
20, 116 – 117.
Wang W, Cote J, Xue Y, Zhou S, Khavari PA, Biggar SR,
Muchardt C, Kalpana GV, Goff SP, Yaniv M, Workman
JL and Crabtree GR. (1996). EMBO J., 15, 5370 – 5382.
Whitehouse I, Flaus A, Cairns BR, White MF, Workman JL
and Owen-Hughes T. (1999). Nature, 400, 784 – 787.
Wilkinson CRM, Bartlett R, Nurse P and Bird AP. (1995).
Nucleic Acids Res., 23, 1995.
Wolffe AP and Pruss D. (1996). Cell, 84, 817 – 819.
Xie S, Wang Z, Okano M, Nogami M, Li Y, He W-W,
Okumura K and Li E. (1999). Gene, 236, 87 – 95.
Xu G-L, Bestor TH, Bourc’his D, Hsieh C-L, Tommerup N,
Bugge M, Hulten M, Qu Russo JJ and Viegas-Pequignot E.
(1999). Nature, 402, 187 – 191.
Yen R-WC, Vertino PM, Nelkin BD, Yu JJ, El-Deiry W,
Cumaraswamy A, Lennon GG, Trask BJ, Celano P and
Baylin SB. (1992). Nucleic Acids Res., 20, 2287 – 2291.
Yoder JA and Bestor TH. (1998). Hum. Mol. Genet., 7, 279 –
284.
Yoder JA, Walsh CP and Bestor TH. (1997). Trends Genet.,
13, 335 – 340.
Yokochi T and Robertson KD. (2002). J. Biol. Chem., 277,
11735 – 11745.
Zhong S, Salomoni P and Pandolfi PP. (2000a). Nature Cell
Biol., 2, E85 – E90.
Zhong S, Salomoni P, Ronchetti S, Guo A, Ruggero D and
Pandolfi PP. (2000b). J. Exp. Med., 191, 631 – 639.
Oncogene