DNA Methylation: Basic Mechanisms - Part 2 ppsx

28
A.
Br
e
r
oeta
l
.
o
f
Dnmt1. Since Dnmt1 is a ca ta
l
ytica
ll
ys
l
ow enzyme (Pra
dh
an et a
l
. 1997)
,
its prolon
g
ed association in G2 and M-phases with chromatin could allow
s
ufficient time for full meth
y
lation of all hemimeth
y
lated sites, in particu

-
l
ar at
h
eavi
l
y met
h
y
l
ate
dh
eteroc
h
romatic sequences (Easwaran et a
l
. 2004)
.
In addition, Dnmt1 has been reported to interact with histone deacet
y
lases
(
HDACs
)(
Fu
k
seta
l
. 2000; Ro
b

ertson et a
l
. 2000; Rountree et a
l
. 2000
)
an
d
mi
g
ht serve as a loadin
g
platform for these chromatin modifiers. Concomi-
tant
l
y, met
h
y
l
-CpG-
b
in
d
ing
d
omain (MBD) proteins, recognizing t
h
e new
l
y

g
enerated modified CpGs, have been also shown to recruit HDACs (Jones et al.
1
998; Nan et al. 1998; N
g
et al. 1999) and can thereb
y
further contribute to the
rep
l
ication o
f
t
h
e
h
istone mo
d
i
fi
cations upon DNA rep
l
ication. In t
h
is regar
d
,
there is increasin
g
evidence of crosstalk between histone modifications and

D
NA met
h
y
l
ation.In para
ll
e
l
to t
h
ese mec
h
anisms
f
or rep
l
icationo
f
epigenetic
info rmation, the random distribution of “old” histones between the two repli
-
cate
d
DNA stran
d
simp
l
ies t
h

at mo
d
i
fi
cation s suc
h
as
h
istone met
h
y
l
ation ar
e
passed onto the nucleosomes assembled at the newl
y
replicated strands. Fac
-
tors such as HP1, which reco
g
nizes specific meth
y
lation forms of histone H3
(
Lachner et al. 2001), can then bind the replicated chromatin, recruit histone
meth
y
ltransferases (HMTs) (Lehnertz et al. 2003) and “spread” the histon
e
met

h
y
l
ation mar
k
sontot
h
ea
d
jacent, p revious
l
y
d
eacety
l
ate
dh
istones
.
A
lthou
g
h man
y
enz
y
mes have been described that can actuall
y
add meth
yl

g
roups to t
h
eDNA,muc
hl
ess is
k
nown a
b
out DN A
d
emet
h
y
l
ases. T
h
e exis
-
tence of such enz
y
mes, however, is almost certain, since active demeth
y
latio
n
of the paternal
g
enome durin
g
preimplantation development has been ev-

idenced (Ma
y
er et al. 2000). Similarl
y
, there must be demeth
y
lases, whic
h
can remove imprints in the course of
g
erm cell development, in order to se
t
t
h
enove
l
parenta
l
i
d
entity. Can
d
i
d
ate enzymes
f
or DNA
d
emet
h

y
l
ation in
-
clude, on the one hand,
g
l
y
cos
y
lases, which in effect resemble a “base excision
D
NA repair activity” w
h
ere t
h
e met
h
y
l
ate
d
cytosines are r emove
d
,resu
l
t
-
in
g

in an abasic site and sin
g
le strand breaks that have to be consecutivel
y
re
p
aired (Jost et al. 2001; Vaira
p
andi 2004). Another
p
ro
p
osed mechanism
includes direct demeth
y
lation of 5mC, via the meth
y
lated CpG bindin
g
pro
-
tein MBD2 (Bhattacharya et al. 1999). Since MBD2 has also been reporte
d
to
b
einvo
l
ve
d
in 5mC-

d
epen
d
ent transcriptiona
l
repression (Hen
d
ric
h
an
d
Tweedie 2003) (see following section), it was proposed that it might exer
t
a
d
ua
l
, promoter-speci
fi
cro
l
easarepressort
h
roug
hb
in
d
ing o
f
5mC an

d
as
an activator throu
g
h active DNA demeth
y
lation (Detich et al. 2002). However,
t
h
e
d
emet
h
y
l
ating activity o
f
MBD2 cou
ld
not yet
b
erepro
d
uce
d
an
d
is
h
ence

disputed (Vairapandi 2004).
Re plication and Translation of Epi
g
enetic Information
29
Fig. 2a,
b
R
eplication of epi
g
enetic info rmation
.
a
Are
p
lication fork is shown wher
e
Dnmt1 associated with the replication machiner
y
(
g
reen box)iscop
y
in
g
the meth
y-
l
ation mar
k(

m
)at
h
emimet
h
y
l
ate
d
CpG sites, w
h
ic
h
are t
h
en recognize
d
an
db
oun
d
b
y methyl-CpG-binding domain (MBD) proteins. Both MBD proteins and Dnmt1 re-
c
ruit histone deacet
y
lases (HDACs), thereb
y
maintainin
g

the deacet
y
lated chromatin
state.
b
Th
e same rep
l
ication
f
or
k
is s
h
own
f
rom a nuc
l
eosoma
l
view. Nuc
l
eosomes
a
r
esho
wn
as
blue circles
,

wit
h
met
h
y
l
ate
dh
istone H3 tai
l
sa
s
fi
lled yellow square
s
a
n
d
5
mC a
s
red dot
s
. Histones bearing repressive methylated lysine residues are distribute
d
r
an
d
om
l

yontorep
l
icate
dd
aug
h
ter stran
d
s. Bin
d
ing o
f
HP1 to met
h
y
l
ate
dh
istones
c
an recruit
h
istone met
h
y
l
trans
f
erase (HMT) t
h

at mo
d
i
f
y
l
ysine resi
d
ues o
f
t
h
e new
ly
i
ncorporated histones (light blue circles
)
30
A.
Br
e
r
oeta
l
.
4
Translat
i
on of DNA Methylat
i

on
T
h
e precise mo
d
eo
f
action o
fh
ow DNA met
h
y
l
ation mo
d
u
l
ates transcriptio
n
is far fr om bein
g
understood. In fact, different mechanisms could accoun
t
f
or contro
ll
ing gene expression at
d
i
ff

erent
l
oci. T
h
oug
h
DNA met
h
y
l
ation
in
g
eneral is associated with transcriptional silencin
g
, in some cases meth
y-
l
ation
h
as
b
een s
h
own to in
d
uce ex
p
ression. T
h

is
h
as
b
een
d
emonstrate
d
for the imprinte
d
I
gf2
locus, where meth
y
lation of a differentiall
y
meth
y
-
l
ated re
g
ion (DMR) o n the ma ternal chromosome prev ents b indin
g
of CTCF
(
CCCTC-
b
in
d

ing
f
actor), w
h
ic
h
resu
l
ts in a positive en
h
ancer
f
unction (Be
ll
and Felsenfeld 2000; Hark et al. 2000; Kanduri et al. 2000; Szabo et al. 2000).
Transcriptiona
l
si
l
encing me
d
iate
db
y met
h
y
l
ation o
f
CpGs near promoter

re
g
ions is thou
g
ht to occur b
y
at least two different mechanisms. One pos
-
s
i
b
i
l
ity is t
h
at met
h
y
l
ation o
f
speci
fi
c target sites simp
l
ya
b
o
l
is

h
es
b
in
d
in
g
of transcription factors or transcriptional activ ators b
y
sterical hindrance
.
Another increasin
g
l
y
important mechanism involves the specific reco
g
ni
-
tion and bindin
g
of factors to meth
y
latedDNA,tri
gg
erin
g
different kinds of
downstream responses, entailin
g

(or not) further chromatin modifications.
In mamma
l
s, t
h
ere are severa
lk
nown met
h
y
l
-CpG-
b
in
d
ing proteins. T
h
e
M
BD protein famil
y
members share a conserve
d
m
eth
y
l-CpG
-
b
indin

g
d
o
m
ain
(
MBD
)(
Hen
d
ric
h
an
d
Bir
d
1998
)
.W
h
i
l
e MeCP2, MBD1, an
d
MBD2
h
ave
b
een
s

hown to act as transcriptional repressors, MBD4 appears to be involv ed i
n
r
educing the mutational risk from potential C
→
T
transitions
,
which resul
t
from deamination of 5mC. A fifth member of the MBD famil
y
, MBD3 does not
b
ind to methylated DNA (Hendrich and Tweedie 2003), but is a constituent o
f
t
h
e NuRD (nuc
l
eosome remo
d
e
l
ing an
dh
istone
d
eacety
l

ation) corepresso
r
complex. A further, recently detected 5mC-binding protein is Kaiso, whic
h
sh
ows no sequence conservation wit
h
MBD proteins
b
ut a
l
so
f
unctions a
s
a
transcriptional repressor (Prokhortchouk et al. 2001). In contrast to MBDs,
Kaiso appears to bind via a zinc-finger motif in a sequence-specific manner at
s
equences containin
g
two s
y
mmetricall
y
meth
y
lated CpGs. A recen t stud
y
in

X
enopus rev ealed an essential role of Kaiso as a methylation-dependen t globa
l
transcriptiona
l
rep ressor
d
uring ear
l
y
d
eve
l
opment (Ruzov et a
l
. 2004)
.
In mammals, the MBD family comp rises five members: MBD1–4 and
M
eCP2. A
ll
o
f
t
h
em except MBD3 s
h
are a
f
unctiona

l
MBD t
h
at is responsi
ble
for tar
g
etin
g
the proteins to 5mC sites. In mouse cells this can be readil
y
see
n
b
yt
h
e increase
d
concentration o
f
MBD proteins at pericentric
h
eter oc
h
ro
-
matin, which is hi
g
hl
y

enriched in 5mC (Lewis et al. 1992; Hendrich and Bir
d
1
998). A summary of the mouse MBD protein family and their domains i
s
sh
owninFig.3.
Re plication and Translation of Epi
g
enetic Information
3
1
F
i
g.
3
O
rganization o
f
t
h
e mouse MBD protein
f
ami
l
y
.
Num
b
er

s
r
epresent amino aci
d
pos
i
t
i
ons
.
co
RID
,
cor epressor interacting
d
omain;
C
XX
C
,Cys-ric
hd
omain
;
(
E
)
1
2
,
G

l
u
repea
t;
(
GR
)
11
,
G
l
y-Arg repeat;
MBD
,
met
h
y
l
-CpG-
b
in
d
ing
d
omain
;
Hh
H-GPD,DN
A
N-Gl

y
cos
y
lase domain
;
T
R
D
,
transcri
p
tional re
p
ressor domain
M
BD2 an
d
3s
h
ow a
h
ig
h
conservation, s
h
aring t
h
e same genomic struc
-
t

ure except for their intron len
g
th (Hendrich et al. 1999a). Since homolo
g
ou
s
expresse
d
sequence tags (ESTs)
f
or MBD2/3 were a
l
so
f
oun
d
in inverte
b
rates,
it is thou
g
ht to represent the ancestral protein from which all other famil
y
m
embers have been derived (Hendrich and Tweedie 2003). The increase i
n
number of 5mC bindin
g
pr oteins from invertebrates to vertebrates is believe
d

t
o have paralleled the increase in DNA meth
y
lation (see Sect. 2, “DNA Meth
y-
l
ation”), as t
h
is wou
ld h
ave ena
bl
e
d
a
fi
ne-tuning o
f
met
h
y
l
ation-
d
epen
d
en
t
silencin
g

on the one hand, as well as lowered the mutational risks emer
g
in
g
f
rom s
p
ontaneous
d
eamination on t
h
eot
h
er (Hen
d
ric
h
an
d
Twee
d
ie 2003)
.
In mammals, MBD3 does not bind to meth
y
lated CpGs due to two amin
o
aci
d
su

b
stitutions wit
h
in t
h
e MBD
(
Saito an
d
Is
h
i
k
awa 2002
)
.Ot
h
er verte-
brates,however,suchasfro
g
s, have two MBD3 forms, one of which retains a
5mC-binding ability (Wade et al. 1999). Seq uence homology predicts a similar
situation
f
or t
h
epu
ff
er
fi

s
h
an
d
t
h
eze
b
ra
fi
s
h
(Hen
d
ric
h
an
d
Twee
d
ie 2003).
32
A.
Br
e
r
oeta
l
.
M

BD3 in mamma
l
s is a constituent o
f
t
h
eNuRDcore
p
ressor com
pl
ex. NuRD
is found in man
y
or
g
anisms incl udin
g
plants and pla
y
s an important role in
transcriptional silencin
g
via histone deacet
y
lation. Thou
g
h MBD3 has bee
n
sh
own to

b
e essentia
lf
or em
b
ryonic
d
eve
l
opment (Hen
d
ric
h
et a
l
. 2001), it
s
function within the NuRD multi
p
rotein com
p
lex hasstill to be clarified. MBD
2
interacts wit
h
t
h
e NuRD comp
l
ex ma

k
ing up t
h
e MeCP1 comp
l
ex (met
h
y
l-
CpG-bindin
g
protein), which was actuall
y
the first meth
y
l-CpG-bindin
g
ac
-
tivity iso
l
ate
d
in mamma
l
s(Mee
h
an et a
l
. 1989). In spite o

f
t
h
e many potentia
l
bindin
g
sites of MBD2, it does not appear to act as a
g
lobal transcriptional
repressor. In fact, onl
y
one tar
g
et
g
ene of MBD2 has been described until now
,
an
d
t
h
at i
s
I
l4
d
uring mouse T ce
ll d
i

ff
erentiation (Hutc
h
ins et a
l
. 2002). Her
e
l
oss of MBD2 has been shown to correlate with a leak
y
instead of a complet
e
repression. Consequent
l
y, it
h
as
b
een
h
ypot
h
esize
d
t
h
at MBD2 mig
h
trat
h

e
r
act in “fine-tunin
g
” transcriptional contr ol b
y
reducin
g
transcriptional noise
at genes, w
h
ic
h
are a
l
rea
d
ys
h
ut o
ff
(Hen
d
ric
h
an
d
Twee
d
ie 2003). A

l
terna
-
tivel
y
, the lack of a
g
lobal de-repression o f meth
y
lated
g
enes u pon MBD
2
l
oss could be explained b
y
redundanc
y
amon
g
MBD famil
y
members. Studies
abro
g
atin
g
several MBD proteins at the same time will help to answer this
question. An interestin
g

phenot
y
pe of MBD2
−
/
−
mi
ce
i
st
h
at a
ff
ected
f
e
m
a
l
e
anima
l
sneg
l
ect t
h
eir o
ff
spring
d

ue to an un
k
nown neuro
l
ogica
l
e
ff
ect (Hen
-
d
ric
h
et a
l
. 2001). MBD2
b
is an iso
f
orm t
h
at is
g
enerate
dby
usin
g
an a
l
ternativ

e
trans
l
ation start co
d
on generating a protein t
h
at
l
ac
k
s 140 N-termina
l
amino
aci
d
s(Hen
d
ric
h
an
d
Bir
d
1998). Surprisin
gly
,it
h
as
b

een reporte
d
to possess
a demeth
y
lase activit
y
(see previous section and Bhattachar
y
a et al. 1999). I
n
g
ene reporter assa
y
s, it was even s
h
own to act as a transcriptiona
l
activator
(
Detich et al. 2002). Thus, it has been
p
ro
p
osed that MBD2 could act as bot
h
a transcriptiona
l
repressor an
d

stimu
l
ator. It s
h
ou
ld b
ea
dd
e
d
,t
h
oug
h
,t
h
at
other
g
roups hav e not been able to reprod uce the demeth
y
lase activit
y
o
f
M
BD2
b
,sot
h

e existence o
f
t
h
is activity is sti
ll
controversia
l
(
d
iscusse
d
in
Wa
d
e 2001
)
.
MBD1 is exceptional amon
g
the transcriptionall
y
rep ressive MBDs, since i
t
can suppress transcription
f
rom
b
ot
h

met
hyl
ate
d
an
d
unmet
hyl
ate
d
promot
-
ers in transient transfection assa
y
s(Fu
j
ita et al. 1999). Four splicin
g
isoforms
h
ave
b
een
d
escri
b
e
d
in
h

umans (Fujita et a
l
. 1999) an
d
t
h
ree in mouse (Jor
-
g
ensen et al. 2004), with the ma
j
or difference bein
g
the presence of thre
e
versus two CXXC cysteine-ric
h
regions (see Fig. 3). T
h
e presence o
f
t
h
e mos
t
C-terminal CXXC motifs in mouse was shown to be responsible for its bindin
g
to unmet
h
y

l
ate
d
sites (Jorgensen et a
l
. 2004) an
df
or its capacity to si
l
ence un-
meth
y
lated reporter constructs (Fu
j
ita et al. 1999). The repression potential of
M
BD1 seems to rel
y
on the recruitment of HDACs, althou
g
h, most probabl
y,
d
i
ff
erent ones
f
rom t
h
ose engage

d
in MBD2 (an
d
MeCP2) si
l
encing (Ng et
Re plication and Translation of Epi
g
enetic Information
33
a
l
. 2000). Simi
l
ar to MBD2, MBD1
−/−
mice ex
h
i
b
it neuro
l
ogica
ld
e
fi
ciencies,
as the
y
show reduced neuronal differentiation and have defects in spatia

l
learnin
g
as well as in hippocampus lon
g
-term potentiation (Zhao et al. 2003)
.
M
BD4 is t
h
eon
l
y mem
b
er wit
h
in t
h
e MBD
f
ami
l
yt
h
at is not invo
l
ve
d
in transcriptional re
g

ulation. Instead, it appears to be implicated in reduc
-
ing t
h
e muta tiona
l
ris
k
t
h
at is imminent in genomes wit
hh
ig
h
met
h
y
l
atio
n
levels, b
y
transitions of 5m
C
→
T
via deamination. This transitio n poses a bi
g-
g
er pro

bl
em
f
or t
h
e DNA repair mac
h
inery t
h
an C
→
U
transitions
,
w
h
ic
h
result from the deamination of unmeth
y
lated c
y
tosines, since the former re
-
sults in G–T mismatches, in which the mismatched base (G or T) cannot
rea
d
i
l
y

b
ei
d
enti
fi
e
d
. In contrast, uraci
l
in G–U mismatc
h
es can easi
l
y
b
e
p
inpointed as the “wron
g
” base, since it is not a constituent of DNA. Accord-
ing
l
y, MBD4 possesses a C-termina
l
g
l
ycosy
l
ase moiety t
h

at can speci
fi
ca
ll
y
remove Ts from G–T mismatches (Hendrich et al. 1999b; see Fi
g
. 3). In fact,
its pre
f
erre
db
in
d
ing su
b
strate is 5mCpG/TpG, i.e., t
h
e
d
eamination pro
d
uct
o
f the 5mCpG/5mCpG dinucleotide. Indeed mutation frequenc
y
anal
y
sis in
MBD4

−
/
−
mice revealed an approximatel
y
threefold increase in
C
→
T
t
r
a
n
-
sitions at CpGs compare
d
to wi
ld
-t
y
pe ce
ll
s (Mi
ll
ar et a
l
. 2002; Won
g
et a
l

.
2
002), which supports the idea of MBD4 bein
g
a mutation attenuator
.
S
ince MeCP2 was t
h
e
fi
rst met
h
y
l
-CpG-
b
in
d
ing protein to
b
ec
l
one
d
an
d
t
he second meth
y

lated DNA bindin
g
activit
y
to be isolated after MeCP1, i
t
is o
f
ten re
f
erre
d
to as t
h
e
f
oun
d
ing mem
b
er o
f
t
h
e MBD
f
ami
l
y. A sing
l

e
m
eth
y
lated CpG dinucleotide has been shown to be sufficient for bindin
g
(Lewis et al. 1992). In transient transfection assa
y
s with meth
y
lated
g
ene re-
p
orter i
n
X
eno
p
u
s
and in mice it was demonstrated that Me
C
P2 functions
as a transcri
p
tional re
p
ressor, at least in
p

art via interaction with the Sin3
c
orepressor comp
l
ex, w
h
ic
h
contains
h
istone
d
eacety
l
ases 1 an
d
2 (J ones et
al. 1998; Nan et al. 1998). An approximatel
y
100-amino-acid-containin
g
tran-
scriptiona
l
repression
d
omain (TRD) in t
h
emi
ddl

eo
f
t
h
e protein
h
as
b
een
shown to be critical for transcriptio nal silencin
g
(Nan et al. 1997). Apart from
t
he recruitment of HDACs
,
MeCP2 has been shown to associate with a histon
e
m
eth
y
ltransferase activit
y
specificall
y
modif
y
in
g
histone H3 at l
y

sine 9, whic
h
is known to repre sent a transcriptionally repressive chromatin label (Fuks e
t
a
l
. 2003). In a
dd
ition, MeCP2
h
as recent
l
y
b
een
f
oun
d
to interact wit
h
com-
p
onents of the SWI/SNF-related chromatin-remodeling complex, suggesting
anove
l
potentia
l
MeCP2-
d
epen

d
ent si
l
encing mec
h
anism (Hari
k
ris
h
nan et
al. 2005). Moreover, MeCP2 can induce compaction of oli
g
onucleosomes in
v
itro, w
h
ic
h
cou
ld
a
dd
itiona
ll
y suppress transcription in vivo t
h
roug
h
a
d

ens
e
c
hromatin conformation that is incompatible with the bindin
g
of factors rel-
e
vant for transcriptional activation (Georgel et al. 2003). In summary, MeCP
2
c
ou
ld
trans
l
ate t
h
e DNA met
h
y
l
ation mar
kd
irect
l
y
b
y preventing t
h
eacces
s

34
A.
Br
e
r
oeta
l
.
o
f
transcriptiona
l
acti vators to promo ters/en
h
ancers or in
d
irect
l
y
b
yeit
h
e
r
recruitin
g
modifiers o f histones such as histone deacet
y
lases (see also Fi
g

.2)
and meth
y
ltransferases or b
y
compactin
g
chromatin.
W
it
h
t
h
ei
d
ea in min
d
t
h
at MeCP2 mig
h
tactasag
l
o
b
a
l
transcription
rep ressor, it was ver
y

surprisin
g
that an exp ression profilin
g
anal
y
sis c om
-
paring MeCP2 nu
ll
mice wit
h
norma
l
anima
l
s revea
l
e
d
on
l
ysu
b
t
l
ec
h
anges in
the mRNA profiles of brain tissues (Tudor et al. 2002). This apparent lack o

f
gl
o
b
a
ld
e-rep ression in t
h
ea
b
sence o
f
MeCP2 resem
bl
es a simi
l
ar situatio
n
as desc
ri
bed
f
o
r MBD
2
−
/
−
m
ice

(
as
d
iscusse
d
ear
l
ier in t
h
is section
)
. Possi
ble
reasons for this observation could be either that other MBD
p
roteins ca
n
compensate
f
or t
h
e
l
oss o
f
MeCP2, or t
h
at t
h
ec

h
anges in transcription
l
eve
ls
induced b
y
MeCP2 deficienc
y
are so small that the
y
are undetectable with
current microarray tec
h
no
l
ogy. T
h
is supports t
h
e rationa
l
et
h
at MBDs mi g
ht
act as re
d
ucers o
f

transcriptiona
l
noise rat
h
er t
h
an to s
h
ut
d
own active
g
ene
s
(
Hen
d
ric
h
an
d
Twee
d
ie 2003). On t
h
eot
h
er
h
an

d
,itcou
ld
we
ll b
et
h
at MeCP
2
rep resses
g
enes in a tissue- an
d
/or time-speci
fi
c
f
as
h
ion. Matarazzo an
d
Ron-
nett, for example, usin
g
a proteomic approach, found substantial difference
s
in protein
l
eve
l

s
b
etween MeCP2-
d
e
fi
cient an
d
wi
ld
-t
y
pe mice (Matarazzo
and Ronnett 2004). Importantl
y
, the
y
showed that the de
g
ree of difference
s
varie
dd
epen
d
ing on t
h
e ana
l
yze

d
tissue (o
lf
actory epit
h
e
l
ium vs o
lf
actor
y
b
u
lb
)an
d
t
h
ea
g
eo
f
t
h
e anima
l
s(2vs4wee
k
sa
f

ter
b
irt
h
). Apart
f
romapo-
tentia
l
g
l
o
b
a
l
e
ff
ect, MeCP2
h
as recent
l
y
b
een
l
in
k
e
d
to t

h
eregu
l
ation o
f
two
s
pecific tar
g
et
g
enes. The
g
enes of Hair
y
2a in
X
eno
p
u
s
(
Stancheva et al. 2003
)
and brain-derived neurotro
p
ic factor (BDNF) in rat (Chen et al. 2003) and
m
ice (Martinowic
h

et a
l
. 2003)—
b
ot
h
are proteins invo
l
ve
d
in neurona
ld
eve
l-
opment and differentiation—have meth
y
lated promoters with bound M eCP2,
w
h
ic
h
is re
l
ease
d
upon transcriptiona
l
activation. Recent
l
y MeCP2 was s

h
ow
n
t
o be involved in the transcriptional silencin
g
of the imprinted
g
en
e
Dlx5
v
ia
th
e
f
ormation o
f
ac
h
romatin
l
oo
p
structure (Hori
k
eeta
l
. 2005)
.

MeCP2 is expre ssed ubiquitousl
y
in man
y
tissues of humans, rats, and
m
ice, althou
g
h at variable levels. Several lines of evidence ar
g
ue that MeCP2
expression increases durin
g
neuronal maturation and differ entiation (Shah
-
bazian et al. 2002b; Jun
g
et al. 2003; Balmer et al. 2003; Cohen et al. 2003;
M
u
ll
aney et a
l
. 2004). In a recent stu
d
y, it was s
h
own t
h
at MeCP2 an

d
MBD
2
protein levels increase also durin
g
mouse m
y
o
g
enesis alon
g
with an increase
in DNA met
h
y
l
ation at pericentric
h
eteroc
h
romatin (Brero et a
l
. 2005). More
-
over, it was demonstrated that MeCP2 and MBD2 are responsible f or a ma
j
o
r
reorganization o
f

pericentric
h
eter oc
h
romatin
d
uring termina
ld
i
ff
erentia-
t
ion that leads to the formation of lar
g
e heterochro matic clusters (Brero et
al. 2005). This findin
g
provides the link between a protein(s) (MeCP2/MBD2
)
an
d
c
h
romatin organization an
d
assigns it a
d
irect ro
l
einc

h
anges o
f
t
h
e
Re plication and Translation of Epi
g
enetic Information
35
3D c
h
romatin topo
l
ogy
d
uring
d
i
ff
erentiation. T
h
e
l
atter represents yet an-
o
ther level of epi
g
enetic information be
y

ond the molecular composition of
c
hr
o
m
at
in
.
In agreement wit
h
its su
b
strate speci
fi
city, MeCP2
l
oca
l
izes main
l
yat
h
eav
-
il
y
meth
y
lated DNA re
g

ions. I n mouse nuclei, for example, MeCP2 intensel
y
d
ecorates
p
ericentric
h
eter oc
h
romatin (Lewis et a
l
. 1992). In
h
uman ce
ll
s
,
ho wever, the intranuclear distribution of MeCP2 was found to deviate fro
m
th
e pattern in mouse, in t
h
at it
d
i
d
not strict
l
yco
l

oca
l
ize wit
h
met
h
y
l
ate
d
DN A, pericen tric sa tellite sequences, or heterochromatic re
g
ions [visual
-
ized b
y
intense
4

-
6

-
diamidino-2-phen
y
lindole (DAPI) stainin
g
; Koch and
Stra t
l

ing 2004]. Intriguing
l
y, t
h
eaut
h
ors
f
oun
d
an a
dd
itiona
lb
in
d
ing a
ffi
nity
o
f MeCP2 for TpG dinucleotides andproposed asequence-specific bindin
g
de-
fi
ne
db
ya
d
jacent sequences. By using animmunoprecipitation approac
h

,t
h
e
y
revea
l
e
d
an association o
f
MeCP2 wit
h
retrotransposa
bl
ee
l
ements, especia
lly
w
it
h
A
l
u sequences, an
d
wit
h
putative matrix attac
h
ment regions (MARs). I

n
th
is respect, it s
h
ou
ld b
ea
dd
e
d
t
h
at t
h
e MeCP2
h
omo
l
o
g
in c
h
ic
k
en (name
d
A
RBP) was ori
g
inall

y
isolated as a MAR bindin
g
acti vit
y
(von Kries et al.
1991), even
b
e
f
ore rat MeCP2 was actua
lly d
escri
b
e
df
or t
h
e
fi
rst time (Lewi
s
et al. 1992),
y
et its homolo
gy
to the rat protein was noticed onl
y
later (Weitzel
et a

l
. 1997). Interesting
l
y, ARBP/MeCP2
b
in
d
ing in c
h
ic
k
en appears not t
o
b
e
d
epen
d
ent on CpG met
hyl
ation (Weitze
l
et a
l
. 1997). Since t
h
eresu
l
ts in
h

uman ce
ll
swereo
b
taine
d
using a
b
reast cancer ce
ll l
ine (MCF7), it wi
ll be
interestin
g
to investi
g
ate
f
urt
h
er
h
uman ce
ll
t
y
pes, inc
l
u
d

in
g
primar
y
ce
ll
s
,
t
o further clarif
y
MeCP2 bindin
g
specificit
y
in human cells
.
Two stu
d
ies
h
ave
l
ate
ly
reporte
d
a secon
d
MeCP2 sp

l
icin
g
iso
f
orm, w
h
ic
h
y
ields a protein with a sli
g
htl
y
different N-terminal end, due to the utiliza-
t
ion o
f
an a
l
ternative trans
l
ation start co
d
on
(
Kriaucionis an
d
Bir
d

2004;
Mnatzakanian et al. 2004; Fi
g
. 3). Surprisin
g
l
y
this new MeCP2 mRNA ap-
p
ears to
b
emuc
h
more a
b
un
d
ant in
d
i
ff
erent mouse an
dh
uman tissues t
h
a
n
t
he ori
g

inall
y
described isof orm. Fluorescentl
y
ta
gg
ed fusions of both pro-
t
eins, thou
g
h, show the same subnuclear distribution in cultured mouse cell
s
(Kriaucionis and Bird 2004). An antibod
y
raised a
g
ainst the “old” isoform
w
as shown to reco
g
nize also the novel variant (Kriaucionis and Bird 2004).
C
onsequent
l
y, in previous immunocytoc
h
emica
l
stu
d

ies most pro
b
a
bl
y
b
ot
h
isoforms have been detected. The differences between both isoforms are onl
y
su
b
t
l
e, wit
h
t
h
e new protein
h
aving a 12 (
h
uman) an
d
17 (mouse) amino aci
d
lon
g
er N-terminus followed b
y

adiver
g
ent stretch o f 9 amino acids. Since
neit
h
er t
h
e MBD nor t
h
e TRD are a
ff
ecte
db
yt
h
ec
h
anges,
b
ot
h
proteins are
anticipated to be functionall
y
equivalent
.
As already noted, MeCP2 expression appears to be correlated with dif-
f
erentiation an
dd

eve
l
o
p
ment. Its im
pl
ication in neurona
ld
i
ff
erentiation is
36
A.
Br
e
r
oeta
l
.
f
urt
h
er supporte
db
y its invo
l
vement in a
h
uman neuro
d

eve
l
opmenta
ld
is
-
order called Rett s
y
ndrome (RTT). The s
y
ndrome was ori
g
inall
y
described
in 1966 b
y
the Austrian pediatrician Andreas Rett, but its
g
enetic basis was
revea
l
e
d
on
l
y recent
l
y (Amir et a
l

. 1999). At
l
east 80% o
f
RTT cases are cause
d
b
y
spontaneo us mutations in theMeCP2
g
ene (see Kriaucionis andBird 2003),
w
h
ic
h
is
l
oca
l
ize
d
on X
q
28 (Amir et a
l
. 1999). RTT is t
h
e secon
d
most

f
re
q
uen
t
form of female mental retardation after Down s
y
ndrome, and its incidenc
e
is approximate
l
y two
f
o
ld h
ig
h
er t
h
an p
h
eny
lk
etonuria (Je
ll
inger 2003). RT
T
is dia
g
nosed in 1:10,000–1:22,000 female births, with affected

g
irls bein
g
het
-
eroz
yg
ous for th
e
Me
C
P
2
m
utation (Kriaucionis and Bird 2003); con sequentl
y,
t
h
ep
h
enotype is cause
db
yt
h
ece
ll
st
h
at
d

o not express
f
unctiona
l
pr otein
due to random inactivation of the X chromosome containin
g
the wild-t
y
pe
copy o
f
MeCP2
.
Most mutations
f
oun
d
in RTT
p
atients are
l
ocate
d
wit
h
in t
h
e
functional domains, i.e., within the MBD and the TRD of MECP2, but several

mutations
h
ave a
l
so
b
een
f
oun
d
in t
h
e C-termina
l
region, w
h
ere no concrete
function has
y
et been assi
g
ned
.
R
ecentl
y
, however, it was shown that the C-terminal domain of MeCP2 is
crucial at compactin
g
oli

g
onucleosomes into dense hi
g
her order conforma-
tions in vitro (Geor
g
el et al. 2003). Interestin
g
l
y
, this activit
y
was found to
b
ein
d
epen
d
ent o
f
CpG met
h
y
l
ation o
f
t
h
eo
l

igonuc
l
eosoma
l
arrays, w
h
ic
h
parallels the findin
g
s in human and chicken where MeCP2 bindin
g
was als
o
f
oun
d
at non-met
h
y
l
ate
d
sites (see a
b
ove) (Weitze
l
et a
l
. 1997; Koc

h
an
d
S
tratlin
g
2004). Moreover, the C-terminal domain of MeCP2 was found to
s
pecificall
y
bind to the
g
roup II WW domain found in the splicin
g
factor
s
formin-bindin
g
pr otein (FBP) and HYPC (Buschdorf and Stratlin
g
2004). Al
-
thou
g
h the functional role of this association has
y
et to be unraveled, various
mutations wit
h
in t

h
is C-termina
l
region were s
h
own to corre
l
ate wit
h
aRTT
phenot
y
pe. In mouse models for RTT, animals carr
y
in
g
mutations in the
C-terminus genera
ll
yex
h
i
b
it a
l
ess-severe p
h
enotype t
h
an t

h
ose wit
h
anu
ll
mutation (Shahbazian et al. 2002a). Mice wher e MeCP2 was conditionall
y
knockedoutonlyinbraintissueyieldedthesamephenotypeasthatwhere
the whole animal was affected, su
gg
estin
g
that the observable phenot
y
pe is
l
argely due to a failure of proper brain develo pment (Chen et al. 2001; Guy
et a
l
. 2001
)
. Mutations in MeCP2, moreover,
h
ave
b
een s
h
own to corre
l
ate

with phenotypes containing clinical features of X-link ed mental retardatio
n
(
Couvert et a
l
. 2001), Ange
l
man syn
d
rome (Watson et a
l
. 2001), an
d
autism
(
Carne
y
et al. 2003; Zappella et al. 2003). In conclusion, RTT is a
g
ood exampl
e
ill
ustrating t
h
at not on
l
yaret
h
e esta
bl

is
h
ment an
d
rep
l
ication o
f
met
h
y
l
ation
marks pivotal for a normal development—as is shown b
y
the severe pheno-
types caused by loss of Dnmt functions—but the correct translation of D NA
met
h
y
l
ation mar
k
sisacritica
l
prerequisite
f
or norma
l
ontogeny.

Re plication and Translation of Epi
g
enetic Information
37
5
Ou
t
look
T
h
e esta
bl
is
h
ment an
d
sta
bl
e maintenance o
f
epi
g
enetic mar
k
sont
h
e
g
enom
e

at each cell divisio n as well as the translation of this epi
g
enetic information
into
g
enome expression an
d
sta
b
i
l
it
y
is crucia
lf
or
d
eve
l
opment an
dd
i
ff
er-
entiation. This role of epi
g
enetic re
g
ulator
y

mechanisms in the realization o
f
th
e genome
h
as
b
een c
l
ear
l
y esta
bl
is
h
e
db
yt
h
e
fi
n
d
ing o
f
mutations a
ff
ect
-
in

g
epi
g
enetic re
g
u
l
ators in
h
uman
d
iseases (RTT an
d
ICF s
y
n
d
rome) an
d
th
e severity o
f
p
h
enotypes in anima
l
mo
d
e
l

s carrying mutations in t
h
e
d
i
f-
f
erent components o
f
t
h
ese pat
h
wa
y
s. In a
dd
ition,
gl
o
b
a
l
an
dl
oca
l
c
h
an

g
es
in meth
y
lation patterns of the
g
enome are found in most tumors and have
,
th
ere
f
ore, tri
gg
ere
d
intense researc
h
into t
h
eir usa
g
e as new tumor
d
ia
g
nostic
t
ools and therapeutic tar
g
ets.

Anot
h
er recent
l
y emerging an
d
exciting area o
f
resear c
h
w
h
ere manipu
l
at-
in
g
epi
g
enetic information is of fundamental importance is stem cell therap
y
an
d
anima
l
c
l
oning. In a reverse
d
way to

d
i
ff
erentiation, resetting or repro-
g
rammin
g
o
f
t
h
eepi
g
enetic state o
f
a
d
i
ff
erentiate
dd
onor ce
ll
appears to
be
o
ne of the ma
j
or difficulties in animal clonin
g

b
y
nuclear transfer (reviewed
,
e.
g
., in S
h
ieta
l
. 2003). Besi
d
es
h
avin
g
a
f
un
d
amenta
l
impact
f
or
b
asic re-
search, understandin
g
the nature of epi

g
enetic information and its plasticit
y
in (a
d
u
l
t/em
b
ryonic) stem ce
ll
sisa
k
ey prerequisite
f
or success
f
u
l
c
l
inica
l
applications of cell replacement therapies in re
g
enerative medicine.
Acknowled
g
ements Work in the author’s laboratories is funded b
y

the Volkswa
g
ens-
ti
f
tung an
d
t
h
e Deutsc
h
eForsc
h
ungsgemeinsc
h
a
f
t.
References
Aapo
l
aU,Ly
l
eR,Kro
h
n K, Antonara
k
is SE, Peterson P (2001) Iso
l
ation an

d
initia
l
c
h
aracterization o
f
t
h
e mouse Dnmt3
l
gene. Cytogenet Ce
ll
Genet 92:122–126
Aguirre-Arteta AM, Grunewa
ld
I, Car
d
oso MC, Leon
h
ar
d
t H (2000) Expression o
f
an a
l
ternative Dnmt1 iso
f
orm
d

uring musc
l
e
d
i
ff
erentiation. Ce
ll
Growt
h
Di
ff
e
r
11:551–55
9
Amir RE, Van den Ve
y
ver IB, Wan M, Tran CQ, Francke U, Zo
g
hbi HY (1999) Rett
sy
ndrome is caused b
y
mutations in X-linked MECP2, encodin
g
meth
y
l-CpG-
bindin

g
protein 2. Nat Genet 23:185–188
Ar
b
er W, Linn S (1969) DNA mo
d
i
fi
cation an
d
restriction. Annu Rev Bioc
h
em 38:467
–
500
Bac
h
man KE, Rountree MR, Bay
l
in SB (2001) Dnmt3a an
d
Dnmt3
b
are transcriptiona
l
re
p
ressors that exhibit uni
q
ue localization

p
ro
p
erties to heterochromatin. J Bio
l
C
hem 276:32282–3228
7
38
A.
Br
e
r
oeta
l
.
B
almer D, Goldstine J, Rao YM, LaSalle JM (2003) Elevated meth
y
l-CpG-bindin
g
protein 2 expression is acquire
dd
uring postnata
lh
uman
b
rain
d
eve

l
opment an
d
is corre
l
ate
d
wit
h
a
l
ternative po
l
ya
d
eny
l
ation. J Mo
l
Me
d
81:61–6
8
B
e
ll
AC, Fe
l
sen
f

e
ld
G (2000) Met
h
y
l
ation o
f
aCTCF-
d
epen
d
ent
b
oun
d
ary contro
ls
imprinte
d
expression o
f
t
h
eIg
f
2 gene. Nature 405:482–48
5
B
estorT,Lau

d
ano A, Matta
l
iano R, Ingram V (1988)C
l
oning an
d
sequencing o
f
acDNA
enco
d
ing DNA met
h
y
l
trans
f
erase o
f
mouse ce
ll
s. T
h
e car
b
oxy
l
-termina
ld

omain
of the mammalian enz
y
mes is related to bacterial restriction meth
y
ltransferases
.
J
Mol Biol 203:971–983
B
hattachar
y
a SK, Ramchandani S, Cervoni N, Sz
y
f M (1999) A mammalian protein
wit
h
speci
fi
c
d
emet
h
y
l
ase activity
f
or mCpG DNA. Nature 397:579–583
B
ourc’

h
is D, Xu GL, Lin CS, Bo
ll
man B, Bestor TH (2001) Dnmt3L an
d
t
h
e esta
bl
is
h
men
t
o
f
materna
l
genomic imprints. Science 294:2536–253
9
B
rero A, Easwaran HP, Nowa
k
D, Grunewa
ld
I, Cremer T, Leon
h
ar
d
t H, Car
d

oso M
C
(2005) Met
h
y
l
CpG
b
in
d
ing proteins in
d
uce
l
arge sca
l
ec
h
romatin reorganizatio
n
d
uring termina
ld
i
ff
erentiation. J Ce
ll
Bio
l
169:733–743

B
uschdorf JP, Stratlin
g
WH (2004) A WW domain bindin
g
re
g
ion in meth
y
l-CpG-
bindin
g
protein MeCP2: impact on Rett s
y
ndrome. J Mol Med 82:135–14
3
C
ardoso MC, Leonhardt H (1999) DNA meth
y
ltransferase is activel
y
retained in the
c
y
toplasm durin
g
earl
y
development. J Cell Biol 147:25–3
2

C
arlson LL, Pa
g
e AW, Bestor TH (1992) Properties and localization of DNA meth
y
l
-
trans
f
erase in preimp
l
antation mouse em
b
ryos: imp
l
ications
f
or genomic im
-
pr
i
nt
i
ng. Genes Dev 6:2536–2541
C
arney RM, Wo
l
pert CM, Ravan SA, S
h
a

hb
azian M, As
hl
ey-Koc
h
A, Cuccaro ML
,
Vance JM, Perica
k
-Vance MA (2003) I
d
enti
fi
cation o
f
MeCP2 mutations in a series
o
ff
ema
l
es wit
h
au tistic
d
isor
d
er. Pe
d
iatr Neuro
l

28:205–211
C
hen RZ, Akbarian S, Tudor M, Jaenisch R (2001) Deficienc
y
of meth
y
l-CpG bindin
g
protein-2 in CNS neurons results in a Rett-like phenot
y
pe in mice. Na t Gene
t
2
7:3
2
7
–
331
Ch
en T, Ue
d
a Y, Xie S, Li E (2002) A nove
l
Dnmt3a iso
f
orm pro
d
uce
df
rom ana

l
ternativ
e
promoter
l
oca
l
izes to euc
h
romatin an
d
its expression corre
l
ates wit
h
active
de
novo met
h
y
l
ation. J Bio
l
C
h
em 22:2
2
Ch
en WG, C
h

ang Q, Lin Y, Meissner A, West AE, Gri
ffi
t
h
EC, Jaenisc
h
R, Green
-
ber
g
ME (2003) Dere pression of BDNF transcription involves calcium-de penden
t
phosphor
y
lation of MeCP2. Science 302:885–889
C
huan
g
LS, Ian HI, Koh TW, N
g
HH, Xu G, Li BF (1997) Human DNA-(c
y
tosine-5
)
meth
y
ltransferase-PCNAcomplex as a tar
g
et forp21WAF1.Science277:1996–200
0

C
ohen DR, Matarazzo V, Palmer AM, Tu Y,
J
eon OH, Pevsner
J
, Ronnett GV
(
2003
)
Expression of MeCP2 in olfactor
y
receptor neurons is developmentall
y
re
g
ulated
an
d
occurs
b
e
f
ore synaptogenesis. Mo
l
Ce
ll
Neurosci 22:417–42
9
C
ouvert P, Bienvenu T, Aquaviva C, Poirier K, Moraine C, Gen

d
rot C, Ver
l
oes A
,
Andres C, Le Fevre AC, Souville I, Steffann J, des Portes V, Ro
p
ers HH, Yntema HG
,
Fr
y
ns JP, Briault S, Chell
y
J, Cherif B (2001) MECP2 is hi
g
hl
y
mutated in X-linked
mental retardation. Hum Mol
G
enet 10:
9
41–
9
4
6
C
remer T, Cremer C (2001) Chromosome territories, nuclear architecture and
g
en

e
regu
l
ation in mamma
l
ian ce
ll
s. Nat Rev Genet 2:292–30
1
Re plication and Translation of Epi
g
enetic Information 3
9
Cummin
g
s DJ, Tait A, Goddard JM (1974) Meth
y
lated bases in DNA from Parameciu
m
aure
l
ia. Bioc
h
im Biop
h
ys Acta 374:1–11
Detic
h
N, T
h

e
b
erge J, Szy
f
M (2002) Promoter-speci
fi
cactivationan
dd
emet
h
y
l
atio
n
b
y MBD2/
d
emet
h
y
l
ase. J Bio
l
C
h
em 277:35791–35794
Doerfler W (1983) DNA meth
y
lation and
g

ene activit
y
. Annu Rev Biochem 52:93–124
Easwaran HP, Schermelleh L, Leonhardt H, Cardoso MC (2004) Re
p
lication-
independen t chromatin loadin
g
of Dnmt1 durin
g
G2 and M phases. EMBO Rep
5
:1181–118
6
Easwaran HP, Leon
h
ar
d
tH,Car
d
oso MC (2005) Ce
ll
cyc
l
e mar
k
ers
f
or
l

ive c e
ll
ana
l
yses.
Ce
ll
Cyc
l
e 4:453–45
5
Felsenfeld G, Groudine M (2003) Controllin
g
the double helix. Nature 421:448–453
Fisher AG, Merkenschla
g
er M (2002) Gene silencin
g
, cell fate and nuclear or
g
anisation.
Curr O
p
in Genet Dev 12:193–197
Fujita N, Ta
k
e
b
ayas
h

iS,O
k
umura K, Ku
d
oS,C
h
i
b
aT,SayaH,Na
k
ao M (1999
)
Met
h
y
l
ation-me
d
iate
d
transcriptiona
l
si
l
encing in euc
h
romatin
b
y met
h

y
l
-CpG
b
in
d
ing protein MBD1 iso
f
orms. Mo
l
Ce
ll
Bio
l
19:6415–6426
Fu
k
s F, Burgers WA, Bre
h
mA,Hug
h
es-Davies L, Kouzari
d
es T (2000) DNA met
h
y
l
-
transferase Dnmt1 associates with histone deacet
y

lase activit
y
. Nat Genet 24:88
–
91
Fuks F, Hurd PJ, Wolf D, Nan X, Bird AP, Kouzarides T (2003) The meth
y
l-CpG-
b
in
d
ing protein MeCP2
l
in
k
sDNAmet
h
y
l
ation to
h
istone met
h
y
l
ation. J Bio
l
C
h
em 278:4035–404

0
Gau
d
et F, Ri
d
eout WM 3r
d
, Meissner A, Dausman J, Leon
h
ar
d
t H, Jaenisc
h
R (2004
)
Dnmt1 expression in pre- an
d
postimp
l
antation em
b
ryogenesis an
d
t
h
e mainte
-
nance of IAP silencin
g
. Mol Cell Biol 24:1640–164

8
Geor
g
el PT, Horowitz-Scherer RA, Adkins N, Woodc ock CL, Wade PA, Hansen J
C
(2003) Chromatin compaction b
y
human MeCP2. Assembl
y
of novel secondar
y
chromatin structures in the absence of DNA meth
y
lation. J Biol Chem 278:32181
–
3
21
88
Gorovs
k
y MA, Hattman S, P
l
eger GL (1973) (6N)Met
h
y
l
a
d
enine in t
h

enuc
l
ear DNA
o
f
a eucaryote, Tetra
h
ymena pyri
f
ormis. J Ce
ll
Bio
l
56:697–70
1
Guy J, Hen
d
ric
h
B, Ho
l
mes M, Martin JE, Bir
d
A (2001) A m ouse Mecp2-nu
ll
mutatio
n
causes neuro
l
ogica

l
symptoms t
h
at mimic Rett syn
d
rome. Nat Genet 27:322–326
Hans en RS, Wi
j
men
g
a C, Luo P, Stanek AM, Canfield TK, Weemaes CM, Gartler SM
(1999) The DNMT3B DNA meth
y
ltransferase
g
ene is mutated in the ICF immun-
odeficienc
y
s
y
ndrome. Proc Natl Acad Sci U S A 96:14412–1441
7
Harikrishnan KN, Chow MZ, Baker EK, Pal S,Bassal S, Brasacchio D, Wan
g
L, Crai
g
JM,
Jones PL, Sif S, El-OstaA(2005)BrahmalinkstheSWI/SNFchromatin-remodelin
g
comp

l
ex wit
h
MeCP2-
d
epen
d
ent transcriptiona
l
si
l
encing. Nat Genet 37:254–264
Har
k
AT, Sc
h
oen
h
err CJ, Katz DJ, Ingram RS, Levorse JM, Ti
l
g
h
man SM (2000) CTCF
me
d
iates met
h
y
l
ation-sensitive en

h
ancer-
bl
oc
k
ing activity at t
h
e H19/Ig
f
2
l
ocus.
N ature 405:486–48
9
Hattman S, Kenn
y
C, Ber
g
er L, Pratt K (1978) Comparative stud
y
of DNA meth
y
latio
n
in three unicellular eucar
y
otes. J Bacteriol 135:1156–1157
Hen
d
ric

h
B, Bir
d
A (1998) I
d
enti
fi
cationan
d
c
h
aracterization o
f
a
f
ami
l
yo
f
mamma
l
ia
n
meth
y
l-CpG bindin
g
proteins. Mol Cell Biol 18:6538–654
7
40

A.
Br
e
r
oeta
l
.
H
endrich B, Tweedie S (2003) The meth
y
l-CpG bindin
g
domain and the evolvin
g
role
o
f
DNA met
h
y
l
ation in anima
l
s. Tren
d
s Genet 19:269–277
H
en
d
ric

h
B, A
bb
ott C, McQueen H, C
h
am
b
ers D, Cross S, Bir
d
A (1999a) Genomi
c
structure an
d
c
h
romosoma
l
mapping o
f
t
h
e murine an
dh
uman M
bd
1, M
bd
2
,
M

bd
3, an
d
M
bd
4 genes. Mamm Genome 10:906–91
2
H
en
d
ric
h
B, Har
d
e
l
an
d
U, Ng HH, Jiricny J, Bir
d
A (1999
b
)T
h
et
h
ymine g
l
ycosy
l

as
e
MBD4 can
b
in
d
to t
h
epro
d
uct o
fd
eamination at met
h
y
l
ate
d
CpG sites. Nature
4
0
1
:30
1–
304
H
endrich B, Gu
y
J, Ramsaho
y

e B, Wilson VA, Bird A (2001) Closel
y
related protein
s
MBD2 and MBD3 pla
y
distinctive but interactin
g
roles in mouse development.
G
enes Dev 15:710–723
H
o
ll
i
d
ay R, Pug
h
JE (1975) DNA mo
d
i
fi
cation mec
h
anisms an
d
gene activity
d
urin
g

d
eve
l
opment. Science 187:226–232
H
orike S, Cai S, Mi
y
ano M, Chen
g
JF, Kohwi-Shi
g
ematsu T (2005) Loss of silent-
chromatin loopin
g
and impaired imprintin
g
of DLX5 in Rett s
y
ndrome. Na
t
G
enet 37:31–40
H
otc
hk
iss RD (1948) T
h
e quantitative separation o
f
purines, pyrimi

d
ines, an
d
nuc
l
e
-
osi
d
es
b
y paper c
h
romatograp
h
y. J Bio
l
C
h
em 175:315–332
H
owe
ll
CY, Bestor TH, Ding F, Lat
h
am KE, Mertineit C, Tras
l
er JM, C
h
ai

ll
et JR (2001
)
Genomic imprinting
d
isrupte
db
yamaterna
l
e
ff
ect mutation in t
h
e Dnmt1 gene.
Ce
ll
104:829–838
H
utchins AS, Mullen AC, Lee HW, S
y
kes KJ, Hi
g
h FA, Hendrich BD, Bird AP, Reiner SL
(2002) Gene silencin
g
quantitativel
y
controls the function of a developmental
trans-activator. Mo l
C

ell 10:81–
9
1
J
e
ll
inger KA (2003) Rett syn
d
rome—an up
d
ate. J Neura
l
Transm 110:681–70
1
J
ones PL, Veenstra GJ, Wade PA, Vermaak D, Kass SU, Landsber
g
er N, Strouboulis J
,
Wo
lff
e AP (1998) Met
h
y
l
ate
d
DNA an
d
MeCP2 recruit

h
istone
d
eacety
l
ase to
repress transcr
i
pt
i
on. Nat Genet 19:187–19
1
J
or
g
ensen HF, Ben-Porath I, Bird AP (2004) Mbd1 is Recruited to both meth
y
lated an
d
nonmeth
y
lated CpGs via distinct DNA bindin
g
domains. Mol Cell Biol 24:3387–
3
3
95
J
ost JP, Oa
k

e
l
ey EJ, Z
h
u B, Benjamin D, T
h
iry S, Siegmann M, Jost YC (2001) 5-
Meth
y
lc
y
tosine DNA
g
l
y
cos
y
lase participates in the
g
enome-wide loss of DN
A
meth
y
lation occurrin
g
durin
g
mouse m
y
oblast differentiation. Nucleic Acids Re

s
2
9
:4452–446
1
J
un
g
BP, Ju
g
loff DG, Zhan
g
G, Lo
g
an R, Brown S, Eubanks JH (2003) The expression o
f
met
h
y
l
CpG
b
in
d
ing
f
actor MeCP2 corre
l
ates wit
h

ce
ll
u
l
ar
d
i
ff
erentiation in t
h
e
d
eve
l
oping rat
b
rain an
d
in cu
l
ture
d
ce
ll
s. J Neuro
b
io
l
55:86–9
6

K
an
d
uriC,PantV,Lou
k
inov D, Pugac
h
eva E, Qi CF, Wo
lff
eA,O
hl
sson R, Lo
b
a
-
nen
k
ov VV (2000) Functiona
l
association o
f
CTCF wit
h
t
h
einsu
l
ator upstream
o
f

t
h
eH19geneisparento
f
origin-speci
fi
can
d
met
h
y
l
ation-sensitive. Curr Bio
l
10:853–85
6
K
oc
h
C, Strat
l
ing WH (2004) DNA
b
in
d
ing o
f
met
h
y

l
-CpG-
b
in
d
ing protein MeCP2 in
h
uman MCF7 ce
ll
s. Bioc
h
emistr
y
43:5011–502
1
K
riaucionis S, Bir
d
A (2003) DNA met
hyl
ation an
d
Rett s
y
n
d
rome. Hum Mo
l
Genet 12
Supp

l
2:R221–R22
7
Re plication and Translation of Epi
g
enetic Information
4
1
Kriaucionis S, Bird A (2004) The ma
j
or form of MeCP2 has a novel N-terminu
s
generate
db
ya
l
ternative s p
l
icing. Nuc
l
eic Aci
d
s Res 32:1818–1823
Kumar S, C
h
eng X, K
l
imasaus
k
as S, Mi S, Pos

f
ai J, Ro
b
erts RJ, Wi
l
son GG (1994)
T
h
e DNA (cytosine-5) met
h
y
l
trans
f
erases. Nuc
l
eic Aci
d
s Res 22:1–1
0
Kunert N, Mar
h
o
ld
J, Stan
k
e J, Stac
h
D, Ly
k

o F (2003) A Dnmt2-
l
i
k
e protein me
d
iate
s
DNA met
h
y
l
ation in Drosop
h
i
l
a. Deve
l
opment 130:5083–509
0
Lachner M, O’Carroll D, Rea S, Mechtler K, Jenuwein T (2001) Meth
y
lation of histon
e
H3 l
y
sine 9 creates a bindin
g
site for HP1 proteins. Nature 410:116–120
Lehnertz B, Ueda Y, Deri

j
ck AA, Braunschwei
g
U, Perez-Bur
g
os L, Kubicek S, Chen T,
Li E, Jenuwein T, Peters AH (2003) Suv39h-mediated histone H3 l
y
sine 9 meth
y
la-
tion directs DNA meth
y
lation to ma
j
or satellite repeats at pericentric heterochro
-
matin.
C
urr Biol 13:11
9
2–120
0
Leon
h
ar
d
t H, Page AW, Weier HU, Bestor TH (1992) A targeting sequence
d
irect

s
DNA met
h
y
l
trans
f
erase to sites o
f
DNA rep
l
ication in mamma
l
ian nuc
l
ei. Ce
ll
71:8
6
5–873
Lewis JD, Meehan RR, Henzel WJ, Maurer-Fo
gy
I, Jeppesen P, Klein F, Bird A (1992)
Purification, se
q
uence, and cellular localization of a novel chromosomal
p
rotein
that binds to meth
y

lated DNA. Cell 69:905–91
4
Li E, Bestor TH, Jaenisch R (1992) Tar
g
eted m utation of the DNA meth
y
ltransferase
gene resu
l
ts in em
b
ryonic
l
et
h
a
l
ity. Ce
ll
69:915–92
6
Margot JB, Aguirre-Arteta AM, Di Giacco BV, Pra
dh
an S, Ro
b
erts RJ, Car
d
oso MC
,
Leon

h
ar
d
t H (2000) Structure an
df
unction o
f
t
h
e mouse DNA met
h
y
l
trans
f
erase
gene: Dnmt1 s
h
ows a tripartite structure. J Mo
l
Bio
l
297:293–30
0
Martinowic
h
K, Hattori D, Wu H, Fouse S, He F, Hu Y, Fan G, Sun YE (2003) DNA
meth
y
lation-related chromatin remodelin

g
in activit
y
-dependent BDNF
g
ene re
g
-
ulation.
S
cience 302:8
9
0–8
9
3
Matarazzo V, Ronnett GV (2004) Temporal and re
g
ional differences in the olfac
-
tor
y
proteome as a consequence of MeCP2 deficienc
y
.ProcNatlAcadSciUS
A
101:77
6
3–77
68
Ma

y
er W, Niveleau A, Walter J, Fundele R, Haaf T (2000) Demeth
y
lation of the z
yg
otic
paternal
g
enome. Nature 403:501–502
Meehan RR, Lewis JD, McKa
y
S, Kleiner EL, Bird AP (1989) Identification of a mam
-
malian protein that binds specificall
y
to DNA containin
g
meth
y
lated CpGs. Cell
5
8:4
99
–507
Mertineit C, Yo
d
er JA, Ta
k
eto T, Lair
d

DW, Tras
l
er JM, Bestor TH (1998) Sex-s
p
eci
fi
c
exons contro
l
DNA met
h
y
l
trans
f
erase in mamma
l
ian germ ce
ll
s. Deve
l
opment
12
5:
889
–
89
7
Millar CB, Gu
y

J , Sansom OJ, Selfrid
g
e J, MacDou
g
all E, Hendrich B, K ei
g
htle
y
PD,
Bishop SM, Clarke AR, Bird A (2002)Enhanced CpG muta bilit
y
and tumori
g
enesis
in MBD4-deficient mice.
S
cience 2
9
7:403–405
Mnatza
k
anian GN, Lo
h
i H, Munteanu I, A
lf
re
d
SE, Yama
d
a T, MacLeo

d
PJ, J ones JR
,
Sc
h
erer SW, Sc
h
anen NC, Friez MJ, Vincent JB, Minassian BA (2004) A previous
ly
uni
d
enti
fi
e
d
MECP2 open rea
d
ing
f
rame
d
e
fi
nes a new protein iso
f
orm re
l
evant
to Rett syn
d

rome. Nat Genet 36:339–341
Mu
ll
aney BC, Jo
h
nston MV, B
l
ue ME (2004) Deve
l
opmenta
l
expression o
f
met
h
y
l
-CpG
bindin
g
protein 2 is d
y
namicall
y
re
g
ulated in the rodent brain. Neuroscience
123:
9
3

9
–
9
4
9
42
A.
Br
e
r
oeta
l
.
N
an X, Campo
y
FJ, Bird A (1997) MeCP2 is a transcriptional repr essor with abundan
t
b
in
d
ing sites in genomic c
h
romatin. Ce
ll
88:471–481
N
an X, Ng HH, Jo
h
nson CA, La

h
erty CD, Turner BM, Eisenman RN, Bir
d
A (1998
)
Transcriptiona
l
repression
b
yt
h
e met
h
y
l
-CpG-
b
in
d
ing protein MeCP2 invo
l
ves
a
h
istone
d
eacety
l
ase comp
l

ex. Nature 393:386–38
9
N
g HH, Z
h
ang Y, Hen
d
ric
h
B, Jo
h
nson CA, Turner BM, Er
d
jument-Bromage H
,
Tempst P, Rein
b
ergD,Bir
d
A (1999) MBD2 is a transcriptiona
l
repressor
b
e-
lon
g
in
g
to the MeCP1 histone deacet
y

lase complex. Nat Genet 23:58–6
1
Ng
HH, Jeppesen P, Bird A (2000) Active repression of meth
y
lated
g
enes b
y
the chro-
mosomal
p
rotein MBD1. Mol Cell Biol 20:1394–1406
Ok
ano M, Xie S, Li E (1998) C
l
oning an
d
c
h
aracterization o
f
a
f
ami
l
yo
f
nove
l

mam
-
ma
l
ian DNA (cytosine-5) met
h
y
l
trans
f
erases. Nat Genet 19:219–22
0
Ok
ano M, Be
ll
DW, Ha
b
er DA, Li E (1999) DNA met
h
y
l
trans
f
erases Dnmt3a an
d
Dnmt3
b
are essentia
lf
or

d
e novo met
h
y
l
ation an
d
mamma
l
ian
d
eve
l
opment. Ce
ll
99
:
2
47–
2
5
7
P
radhan S, Talbot D, Sha M, Benner
J
,HornstraL,LiE,
J
aenisch R, Roberts R
J(
1997

)
Baculovirus-mediated expression and characterization of the full-len
g
th murin
e
DNA meth
y
ltransferase. Nucleic Acids Res 25:4666–4673
P
rokhortchouk A, Hendrich B, Jor
g
ensen H, Ruzov A, Wilm M, Geor
g
iev G, Bird A
,
Pro
kh
ortc
h
ou
k
E (2001) T
h
e p120 catenin partner Kaiso is a DNA met
h
y
l
ation
-
d

epen
d
ent transcriptiona
l
repressor. Genes Dev 15:1613–1618
R
iggs AD (1975) X inactivation,
d
i
ff
erentiation, an
d
DNA met
h
y
l
ation. Cytogenet Ce
ll
G
enet 14:9–2
5
R
obertson KD (2002) DNA meth
y
lation and chromatin—unravelin
g
the tan
g
led web.
Onco

g
ene 21:5361–537
9
R
obertson KD, Uzvol
gy
iE,Lian
g
G, Talmad
g
eC,Sume
g
iJ,GonzalesFA,JonesP
A
(1999) T
h
e
h
uman DNA met
h
y
l
trans
f
erases (DNMTs) 1, 3a an
d
3
b
:coor
d

inate
mRNA expression in norma
l
tissues an
d
overexpression in tumors. Nuc
l
eic Aci
d
s
R
es 27:2291–2298
R
obertson KD, Ait-Si-Ali S, Yokochi T, Wade PA,
J
ones PL, Wolffe AP
(
2000
)
DNMT
1
forms a com
p
lex with Rb, E2F1 and HDAC1 and re
p
resses transcri
p
tion fro
m
E2F-res

p
onsive
p
romoters. Nat Genet 25:338–34
2
R
ountree MR, Bac
h
man KE, Ba
yl
in SB (2000) DNMT1
b
in
d
s HDAC2 an
d
anewco-
re
p
ressor, DMAP1, to form a com
p
lex at re
p
lication foci. Nat Genet 25:269–27
7
R
uzov A, Dunican DS, Pro
kh
ortc
h

ou
k
A, Pennings S, Stanc
h
eva I, Pro
kh
ortc
h
ou
k
E,
Meehan RR (2004) Kaiso is a
g
enome-wide repressor of transcription that i
s
essential for am
p
hibian develo
p
ment. Develo
p
ment 131:6185–619
4
S
aito M, Ishikawa F (2002) The mCpG-bindin
g
domain of human MBD3 does not bind
to mC
p
G but interacts with NuRD/Mi2 com

p
onents HDAC 1 and MTA2. J Bio
l
C
h
em 277:35434–35439
S
e
lk
er EU (1997) Epigenetic p
h
enomena in
fil
amentous
f
ungi: use
f
u
l
para
d
igms or
repeat-in
d
uce
d
con
f
usion? Tren
d

s Genet 13:296–301
S
hahbazian M, Youn
g
J, Yuva-Pa
y
lor L, Spencer C, Antalff
y
B, Noebels J, Armstron
g
D
,
Pa
y
lor R, Zo
g
hbi H (2002a) Mice with truncated MeCP2 recapitulate man
y
Rett
s
y
ndrome features and dis pla
y
h
y
peracet
y
lation o f histone H3. Neuron 35:243–
2
54

Re plication and Translation of Epi
g
enetic Information
43
Shahbazian MD, Antalff
y
B, Armstron
g
DL, Zo
g
hbi HY (2002b) Insi
g
ht into Rett
s
yn
d
rome: MeCP2
l
eve
l
s
d
isp
l
ay tissue- an
d
ce
ll
-speci
fi

c
d
i
ff
erences an
d
corre
l
ate
wit
h
neurona
l
maturation. Hum Mo
l
Genet 11:115–12
4
S
h
iW,Za
kh
artc
h
en
k
oV,Wo
lf
E (2003) Epigenetic reprogramming in mamma
l
ia

n
nuc
l
ear trans
f
er. Di
ff
erentiation 71:91–11
3
Spector DL (2003) The d
y
namics of chromosome or
g
anization and
g
ene re
g
ulation.
Ann
u
R
e
vBi
oc
h
e
m
7
2
:573

–
608
Srinivasan PR, Bore
k
E (1964) Enzymatic a
l
teration o
f
nuc
l
eic aci
d
structure. Scienc
e
1
45:54
8
–55
3
Stanc
h
eva I, Co
ll
ins AL, Van
d
en Veyver IB, Zog
hb
iH,Mee
h
an RR (2003) A mutant

form of MeCP2 proteinassociated with human Retts
y
ndromecannot be displace
d
from meth
y
lated DNA b
y
notch in Xenopus embr
y
os. Mol Cell 12:425–435
Szabo P, Tan
g
SH, Rentsendor
j
A, Pfeifer GP, Mann JR (2000) Maternal-specific foot
-
prints at putative CTCF sites in the H19 imprintin
g
control re
g
ion
g
ive evidence
for insula tor function.
C
urr Biol 10:607–610
Tariq M, Pasz
k
ows

k
i J (2004) DNA an
dh
istone met
h
y
l
ation in p
l
ants. Tren
d
s Gene
t
20
:
2
44–
2
5
1
Tucker KL, Talbot D, Lee MA, Leonhardt H, Jaenisch R (1996) Com
p
lementation
of meth
y
lation deficienc
y
in embr
y
onic stem cells b

y
a DNA meth
y
ltransferas
e
mini
g
ene. Proc Natl Acad Sci USA 93:12920–12925
Tudor M, Akbarian S, Chen RZ, Jaenisch R (2002) Transcriptional profilin
g
ofamous
e
model for Rett s
y
ndrome reveals subtle transcriptional chan
g
es in the brain. Proc
N atl Acad
S
ci
US
A
99
:15536–1554
1
Twee
d
ie S, C
h
ar

l
ton J, C
l
ar
k
V, B ir
d
A (1997) Met
h
y
l
ation o
f
genomes an
d
genes at t
he
invertebrate-vertebrate boundar
y
. Mol Cell Biol 17:1469–147
5
Vairapandi M (2004)Characterization of DNA demeth
y
lation in normal and cancerou
s
cell lines and the re
g
ulator
y
role of cell c

y
cle proteins in human DNA demeth
y
las
e
activit
y
. J Cell Biochem 91:572–58
3
von Kries JP, Buhrmester H, Stratlin
g
WH (1991) A matrix/scaffold attachment re
g
io
n
bindin
g
protein: identification,purification, and mode of bindin
g
. Cell 64:123–135
Wade PA (2001) Meth
y
l CpG bindin
g
proteins: couplin
g
chromatin architecture to
g
ene re
g

ulation. Onco
g
ene 20:3166–317
3
Wa
d
e PA, Gegonne A, Jones PL, Ba
ll
estar E, Au
b
ry F, Wo
lff
e AP (1999) Mi-2 comp
l
e
x
coup
l
es DNA met
h
y
l
ation to c
h
romatin remo
d
e
ll
ing an
dh

istone
d
eacety
l
ation
.
Nat
G
enet 23:62–66
Watson P, Black G, Ramsden S, Barrow M, Super M, Kerr B, Cla
y
ton-Smith J (2001
)
An
g
elman s
y
ndrome phenot
y
pe associated with mutations in MECP2, a
g
en
e
encodin
g
a meth
y
l CpG bindin
g
protein. J Med Genet 38:224–22

8
Weitze
l
JM, Bu
h
rmester H, Strat
l
ing WH (1997) C
h
ic
k
en MAR-
b
in
d
ing protein ARBP
is
h
omo
l
ogous to rat m et
h
y
l
-CpG-
b
in
d
ing protein MeCP2. Mo
l

Ce
ll
Bio
l
17:5656
–
5666
Won
g
E, Yan
g
K, Kura
g
uchi M, Werlin
g
U, Avdievich E, Fan K, Fazzari M, Jin B
,
Br own AM, L ipkin M, Edelmann W (2002) Mbd4 inactivation increases Cri
g
ht
-
arrowT transition mutations and promotes
g
astrointestinal tumor formation.
Proc Natl Acad
S
ci
US
A
99

:14
9
37–14
9
4
2
Wyatt GR (1951) Recognition an
d
estimation o
f
5-met
h
y
l
cytosine in nuc
l
eic aci
d
s.
JBioc
h
em (To
k
yo) 48:581–584
44
A.
Br
e
r
oeta

l
.
X
ie S, Wan
g
Z, Okano M, No
g
ami M, Li Y, He WW, Okumura K, Li E (1999) Clonin
g
,
expression an
d
c
h
romosome
l
ocations o
f
t
h
e
h
uman DNMT3 gene
f
ami
l
y. Gen
e
23
6

:87–9
5
X
u GL, Bestor TH, Bourc’
h
is D, Hsie
h
CL,TommerupN,BuggeM,Hu
l
tenM,QuX
,
Russo JJ, Viegas-Pequignot E (1999) C
h
romosome insta
b
i
l
ity an
d
immuno
d
e
fi-
cienc
y
s
y
ndrome caused b
y
mutations in a DNA meth

y
ltransferase
g
ene. Natur
e
402:187–1
91
Y
oder JA, Bestor TH (1998) A candidate mammalian DNA meth
y
ltransferase related
to pmt1p of fission
y
east. Hum Mol Genet 7:279–284
Y
o
d
er JA, Yen RWC, Vertino PM, Bestor TH, Bay
l
in SB (1996) New 5’ regions o
f
t
he
murine an
dh
uman genes
f
or DNA (cytosine-5)-met
h
y

l
trans
f
erase. J Bio
l
C
h
em
2
7
1
:
31092
–
31097
Z
appella M, Meloni I, Lon
g
o I, Canitano R, Ha
y
ek G, Rosaia L, Mari F, Renieri A (2003
)
Stud
y
of MECP2
g
ene in Rett s
y
ndrome variants and autistic
g

irls. Am J Me
d
Genet B Neurops
y
chiatr Genet 119:102–10
7
Z
hao X, Ueba T, Christie BR,Barkho B, McConnell M
J
, Nakashima K, Lein ES, EadieBD
,
Willhoite AR, Muotri AR, Summers RG, Chun J, Lee KF, Ga
g
e FH (2003) Mic
e
lackin
g
meth
y
l-CpG bindin
g
protein 1 have deficits in adult neuro
g
enesis an
d
h
ippocampa
lf
unction. Proc Nat
l

Aca
d
Sci U S A 100:6777–678
2
CTMI
(
2006
)
301:45–66

c
Springer-Ver
l
ag Ber
l
in Hei
d
e
lb
erg 200
6
DNA Methyltransferases: Facts, Clues, Mysterie
s
C
. Brenner · F. Fuks (
✉
)
La
b
oratory o

f
Mo
l
ecu
l
ar Viro
l
og y, Facu
l
ty o
f
Me
d
icine, Free University o
f
Brusse
l
s
,
808 route
d
e Lenni
k
, 1070 Brusse
l
s, Be
l
gium
ff
u

k
s@u
lb
.ac.
b
e
1Intro
d
uction 4
6
2 DNMTs: M ug S
h
ots an
d
Knoc
k
out 4
6
2.1 DNM
T
Structure 4
8
2
.
2
D
nmt
K
noc
k

outinMice 4
9
2.3 DNMT Met
h
y
l
trans
f
erase Activity: A Comp
l
exIssue 5
0
3 How Do DNMTs Interfere with Transcri
p
tion
?
51
3
.1 Cross-talk and Transcriptional Silencin
g
51
3
.2 Histone and DNA Meth
y
lation: Mutual Boostin
g
and Feedback Loops . . . 5
3
4 How Are DNMTs Tar
g

eted to Precise DNA Sequences
?
55
4.1 C
h
romatin-Base
d
Targeting 5
6
4.2 Targeting o
f
DNMTS
b
yDNA-Boun
d
TranscriptionFactors 5
9
4.3 T
h
eRNATrigger 6
0
5C
onclusion
s
62
R
e
f
e
r

e
n
ces
63
A
b
str
a
c
t
D
NA meth
y
lation pla
y
s a pivotal role durin
g
development in mammals and i
s
central to transcriptional silencin
g
. The DNA meth
y
ltransferases (DNMTs) are respon
-
sible for the
g
eneration of
g
enomic meth

y
lation patterns leadin
g
to
g
ene silencin
g
,bu
t
t
h
eun
d
er
l
ying mo
l
ecu
l
ar
b
asis remains
l
arge
l
ys
h
rou
d
e

d
in mystery. Here we revie
w
our current understandin
g
of the mechanisms b
y
which DNMTs repress transcription
and how the
y
are tar
g
eted to preferred DNA sequences. Emer
g
in
g
evidence points
to an essentia
l
an
d
intricate we
b
o
f
interactions
b
etween DNMTs an
d
t

h
ec
h
romati
n
environment in w
h
ic
h
t
h
ey
f
unction. T
h
e recent i
d
enti
fi
cation o
f
nove
l
transcription
f
actors recruiting t
h
e DNMTs may open new avenues o
f
researc

h
into t
h
eorigino
f
DNA meth
y
lation patterns. Thanks to these emer
g
in
g
clues, researchers have be
g
un
to lift the veil on the multi-faceted DNMTs, but there remains fascinatin
g
work ahea
d
for whoever wants to full
y
understand DNMTs and their role in the mammalian cell
.
46
C
. Brenner · F. Fuks
1
I
ntr
odu
ct

ion
D
NA meth
y
lationis a ma
j
or ep i
g
enetic event. Itis a post-replicative, reversi ble
,
an
dh
erita
bl
ec
h
emica
l
mo
d
i
fi
cation o
f
DN A invo
l
ve
d
in regu
l

ating a
d
iverse
ran
g
e of biolo
g
ical processes in vertebrates, plants, and fun
g
i. The presen
t
c
h
apter
d
ea
l
s main
l
ywit
h
DNA met
h
y
l
ation in mamma
l
s, an
d
particu

l
ar
ly
h
umans an
d
mice
.
In mamma
l
s, DNA met
h
y
l
ation occurs pre
d
ominant
l
y at cytosine resi
d
ues
l
ocated within CpG din ucleotides and is associated with
g
ene silencin
g
.The
distribution of CpG dinucleotides in the mammalian
g
enome is uneven

an
d
non-ran
d
om. Met
h
y
l
ate
d
DNA is most a
b
un
d
ant in
h
eter oc
h
romatin
-
containin
g
bulk DNA such as parasitic sequences, retrotransposons, and var
-
ious repeat e
l
ements. Most unmet
h
y
l

ate
d
CpG
d
inuc
l
eoti
d
es are
f
oun
d
i
n
“
CpG islands,” i.e., small stretches of CpG-rich DNA found in the 5

re
g
u
l
a
-
tory regions o
f
a
l
most
h
a

lf
o
f
t
h
e genes o
f
t
h
e genome (Bir
d
2002)
.
DNA meth
y
lation has a crucial role in normal mammalian developmen
t
and pla
y
sama
j
or role in
g
ene expression, X-chromosome inactivation in
f
ema
l
es, an
dg
enomic imprintin

g
.Ita
l
so contri
b
utes to t
h
e sta
b
i
l
it
y
an
d
in
-
te
g
rit
y
of the
g
enome b
y
inacti vatin
g
bulk DNA. Altered meth
y
lation p atterns

,
wit
h
genome-wi
d
e
h
ypomet
h
y
l
ation an
d
region-speci
fi
c
h
ypermet
h
y
l
ation
,
are frequentl
y
fo und in cancers (Jones and Ba
y
lin 2002)
.
How

d
oes DNA met
h
y
l
ation
l
ea
d
to gene si
l
encing? How are DNA met
h
y
l
a
-
tion patterns established and maintained? These are amon
g
the most pressin
g
and intri
g
uin
g
questions in the DNA meth
y
lation field. Mechanistic insi
g
hts

into these questions have come from the identification and characterization
of several dedicated enz
y
mes called DNA meth
y
ltransferases (DNMTs). Thes
e
k
ey regu
l
ators o
f
DNA met
h
y
l
ation are t
h
e
f
ocus o
f
t
h
is c
h
apter, w
h
ic
hl

ea
ds
the reader on a trail that starts with the structure of these
p
roteins and
p
ro-
g
resses t
h
roug
h
t
h
e mec
h
anisms
b
yw
h
ic
h
t
h
ey re press transcription an
d
what we know about their tar
g
etin
g

to preferred DNA sequences. Emphasis
is laid on emer
g
in
g
evidence of an intimate connection between DNMTs and
c
hr
o
m
at
in
st
r
uctu
r
e.
2
D
NMTs: Mu
g
Shots and Knockou
t
D
NMTs cata
l
yze met
h
y
l

ation at position 5 o
f
t
h
e cytosine ring, usin
g
S
-
a
d
enos
yl
-met
h
ionine as t
h
e met
hyl g
roup
d
onor. On t
h
e
b
asis o
f
sequence
h
omolo
gy

, DNMTs are divided into three families: DNMT1, DNMT2, and
D
NMT3. T
h
is t
h
ir
df
ami
l
y
h
as t
h
ree mem
b
ers: DNMT3A, DNMT3B, an
d
DNA Meth
y
ltransferases: Facts, Clues, M
y
sterie
s
4
7
DNMT3L (Fig. 1). T
h
e structures an
d

enzymatic acti vities o
f
t
h
ese protein
s
and the correspondin
g
knocko ut phenot
y
pes are reviewed in the followin
g
sectio
n
s.
F
i
g.
1
T
h
e mamma
l
ian DNA met
h
y
l
trans
f
erases (DNMTs). T

h
ree c
l
asses o
f
DNMT
s
are
k
nown. Most o
f
t
h
ese proteins possess an N-termina
l
regu
l
atory
d
omain an
d
aC
-
termina
l
cata
l
ytic
d
omain,

b
ut DNMT2
l
ac
k
st
h
eregu
l
atory
d
omain an
d
DNMT3L is
cata
l
ytica
ll
y inactive. Speci
fi
c conserve
d
moti
f
sare
d
epicte
d[
C
y

s
,
cysteine-ric
hd
o
-
ma
i
n;
PHD
,p
l
ant
h
omeo
d
omain (ATRX-
l
i
k
e);
PWWP
,
pro
l
ine- an
d
tryptop
h
ane-ric

h
d
omain]. T
h
e
l
engt
h
o
f
eac
h
protein is in
d
icate
d
in amino aci
d
s. T
h
e
th
ir
d
co
l
umn
roug
hl
yout

l
ines t
h
ep
h
enotypes resu
l
ting
f
rom
D
nm
t
k
noc
k
out in mice. T
h
e met
h
y
l-
transferase activit
y
of each DNMT (present of not; de novo and/or maintenance) i
s
desc
ri
bed
in

t
h
e
f
ar ri
g
ht colum
n
48
C
. Brenner · F. Fuks
2
.1
DNMT Structur
e
A DNMT genera
ll
y comprises two
d
omains:a we
ll
-conserve
d
cata
l
ytic
d
omain
in the carbox
y

-terminal part of the protein and a more variable r e
g
ulator
y
d
omain in t
h
e amino-termina
l
region. Dnmt1 was t
h
e
fi
rst enzyme to
b
e
isolated as a mammalian DNMT and the onl
y
one identified via a biochemica
l
assay (Bestor et a
l
. 1988; Yen et a
l
. 1992). It
h
as t
h
e
l

argest amino-termina
l
domain of all known DNMTs. Responsible for import into the nucleus and for
zinc bindin
g
, this domain also mediates pro tein–protein interactions.
Expression o
f
t
h
e gene
D
nmt
1
is
h
ig
h
in pro
l
i
f
erating ce
ll
san
d
u
b
iquitou
s

i
n somatic cells. Durin
gg
ameto
g
enesis, expression of the
g
ene from sex
-
sp
eci
fi
c
p
romoters an
d
5

exons resu
l
ts in sex-s
p
eci
fi
c Dnmt1 iso
f
orms w
h
os
e

b
io
l
o
g
ica
lf
unctions are sti
ll
quite o
b
scure (Mertineit et a
l
. 1998; Do
h
ert
y
et a
l
. 2002). In t
h
e mouse, a Dnmt1 iso
f
orm ca
ll
e
d
Dnmt1o,
f
or “oocyte

-
s
peci
fi
c,” is expresse
d
in t
h
e ooc
y
te an
d
pre-imp
l
antation em
b
r
y
o. It seems
to be required onl
y
durin
g
asin
g
le S-phase in the 8-cell mouse embr
y
ot
o
maintain met

hyl
ation patterns at imprinte
dl
oci (Howe
ll
et a
l
. 2001).
The observation that meth
y
lation persists in mouse embr
y
onic stem cell
s
l
ac
k
ing t
he
D
nmt
1
g
ene
l
e
d
researc
h
ers to postu

l
ate t
h
at ot
h
er DNMTs must
exist. Screenin
g
of expressed sequence ta
g
(EST) databases for sequences
containing moti
f
so
f
t
h
e conserve
d
cata
l
ytic
d
omain
l
e
d
to t
h
ei

d
enti
fi
cation
of three candidates: Dnmt2, Dnmt3a, and Dnmt3b
(
Okano et al. 1998a; Yode
r
and Bestor 1998).
Dnmt2 contains onl
y
the DNMT motifs; its
g
ene is expressed, albeit to lo
w
l
evels, in man
y
human and mouse tissues (Yoder and Bestor 1998). The role
o
f
t
h
is protein remains enigmatic (see Sect. 2.3).
The
g
enes D
nmt3a
a
n

d
Dnmt3b
show ver
y
hi
g
h expre ssion durin
g
em-
b
ryogenesis an
d
gametogenesis
b
ut muc
hl
ower expression in
d
i
ff
erentiate
d
so
m
at
i
ct
i
ssues.
Tw

o
Dnm
t3a a
n
dse
v
e
n Dnm
t3b
i
so
f
o
rm
s
h
a
v
ebee
n
de-
s
cribed, featuring specific expression patterns during development and in
adult tissues. Ver
y
little is known about the biolo
g
ical importance of individ-
ual isoforms (Okano et al. 1998a; Chen et al. 2002). To elucidate the s
p

ecific
f
unction o
f
eac
h
Dnmt3a an
d
Dnmt3
b
iso
f
orm, it wi
ll b
e necessary to carry
out
g
enetic anal
y
ses based on isoform-specific
g
ene disruption
.
Structura
ll
y, Dnmt3a an
d
Dnmt3
b
s

h
are, in a
dd
ition to t
h
ecata
l
ytic site in
the C-terminal re
g
ion, two conserved domains in the amino-terminal re
g
ion
:
t
h
epro
l
ine- an
d
tryptop
h
an-ric
h
PWWP
d
omain an
d
t
h

e cysteine-ric
h
PH
D
domain (for plant homeodomain)
.
The PWWP domain has been found in more than 60 eukaryotic proteins
imp
l
icate
d
in transcriptiona
l
regu
l
ation an
d
c
h
romatin organization (Stec e
t
DNA Meth
y
ltransferases: Facts, Clues, M
y
sterie
s
4
9
a

l
. 2000). T
h
estructureo
f
t
h
e mouse Dnmt3
b
PWWP
d
omain is
k
nown (Qi
u
et al. 2002). This domain probabl
y
allows tar
g
etin
g
of Dnmt3a and Dnmt3
b
t
o pericentric heterochromatin, as it is sufficient for bindin
g
to metaphase
c
h
romosomes an

d
promotes met
h
y
l
ation o
f
nuc
l
eosoma
l
DNA (C
h
en et a
l.
2
004; Ge et al. 2004). The PWWP domain of Dnmt3b binds nonspecificall
y
to
DNA(Qiueta
l
. 2002); t
h
at o
f
Dnmt3a s
h
ows
l
itt

l
eDNA-
b
in
d
ing a
b
i
l
ity (C
h
en
et al. 2004
)
.
T
h
e secon
d
conserve
dd
omain o
f
t
h
e N-termina
l
region, t
h
e PHD

d
omain
,
is conserved also in the third member of the DNMT3 famil
y
, Dnmt3L. The
P
HD domains of these proteins most closel
y
resemble the imperfect PHD mo
-
t
i
ff
oun
d
in ATRX, a putative mem
b
er o
f
t
h
e SNF2
f
ami
l
yo
f
ATP-
d

epen
d
ent
chromatin remodelin
g
proteins. A muta ted ATRX
g
ene has been found i
n
severa
l
X-
l
in
k
e
d
menta
l
retar
d
ation
d
isor
d
ers
(
Gi
bb
ons et a

l
. 2000
)
.T
h
e PHD
d
omain mediates protein–protein interactions and functions as a transcrip-
t
iona
l
rep ressor
d
omain (Burgers et a
l
. 2002).
2.2
Dnm
t
Kn
o
ck
ou
t
i
nM
i
ce
DNA met
h

y
l
ation c
h
anges in a
h
ig
hl
yorc
h
estrate
d
way in t
h
ecourseo
f
mouse
d
evelopment. This involves both
g
enome-wide and
g
ene-specific demeth
y
la
-
t
ion an
dd
enovomet

h
y
l
ation (Li 2002). As mentione
d
a
b
ove, DNA met
h
y
l
a
-
t
ion is essential to mammalian development. This is vividl
y
illustrated b
y
tar-
g
eteddisruption of DNMT
g
enes in mice, whichca usesembr
y
onic(D
nmt
1
a
n
d

Dnmt3b
) or post-natal
(
Dnmt3a
)
mortalit
y
(Li et al. 1992; Okano et al. 1999).
D
nmt
1
−
/
−
/
/
mice die around embr
y
onic da
y
(E)8.5, at the onset of
g
astrula
-
t
ion. Ana
l
yses o
fd
ea

d
em
b
ryos
h
ave revea
l
e
d
genome-wi
d
e
d
emet
h
y
l
ation,
biallelic expression of several (but not all) imprinted
g
enes, and aberrant
expression o
f
Xist, a
l
ong, non-co
d
ing RNA invo
l
ve

d
in X-c
h
romosome inac-
t
ivation in
f
ema
l
es
(
Li et a
l
. 1992
).
D
nmt3a
−/
−
mice die 4 weeks after birth; the
y
displa
y
severe intestinal
d
e
f
ects an
d
impaire

d
spermato
g
enesis. As
f
or Dnmt3
b
−
/
−
mice, t
h
e
y
s
h
ow
d
emeth
y
lation of minor satellite DNA, mild neural tube defects, and embr
yo
m
orta
l
ity at E14.5–E18.5 (O
k
ano et a
l
. 1999). W

h
en
b
ot
h
D
nmt3
a
a
n
d
D
nmt3
b
are disrupted in mice, doubl
y
homo z
yg
ous [
Dnmt3a
−/
−
,
D
nmt3
b
−/
−
]embr
y

o
s
h
aveap
h
enotype simi
l
ar to t
h
at o
f
D
nmt
1
−
/
−
em
b
ryos, s
h
owing
d
eve
l
opmen-
t
a
l
arrest at t

h
e presomite sta
g
ean
d
a
d
istorte
d
neura
l
tu
b
earoun
d
E8.
5
(O
k
ano et a
l
. 1999)
.
M
ice wit
h
a
d
isrupte
d

D
nmt
2
g
ene are via
bl
ean
df
erti
l
e, wit
h
minor
d
e
f
ect
s
(Okano et al. 1998b). This is in a
g
reement with results obtained o
n
Dnmt2
−/−
em
b
ryonic stem (ES) ce
ll
s. T
h

ese ce
ll
sarevia
bl
ean
d
s
h
ow no o
b
vious a
l
ter
-
50
C
. Brenner · F. Fuks
ation o
f
t
h
eir DNA met
h
y
l
ation pattern (O
k
ano et a
l
. 1998

b
). As men tione
d
in the next section, this mild phenot
y
pe of
D
nmt2
−
/
−
i
spro
b
a
bly l
in
k
e
d
t
o
the ver
y
low enz
y
matic activit
y
of the DNMT2 protein (Hermann et al. 2003).
D

nmt3
L
−
/−
m
ice are via
bl
e,
b
ut ma
l
es are steri
l
ean
d
t
h
e
h
eterozygou
s
pro
g
en
y
of homoz
yg
ous females die in utero and show complete loss of ma-
terna
l

genomic imprinting (Hata et a
l
. 2002). T
h
is p
h
enotype is in
d
istinguis
h-
able from that of conditional knockout mice havin
g
a disru pted
Dnmt3a
g
en
e
in germ ce
ll
son
l
y. T
h
is
h
ig
hl
ig
h
ts t

h
e crucia
l
ro
l
eo
f
Dnmt3L an
d
Dnmt3a i
n
m
aternal imprintin
g
(Kaneda et al. 2004). A stud
y
also su
gg
ests that Dnmt3L
is an important cofactor for Dnmt3a (Chedin et al. 2002). Dnmt3L ma
y
addi
-
t
iona
ll
y
b
einvo
l

ve
d
in retrotransposon si
l
encing
d
uring premeiotic genome
s
cannin
g
in male
g
erm cells (Bourc’his and Bestor 2004), since deletion of
D
nmt3
L
i
n ear
l
yma
l
e germ ce
ll
sprevents
d
e novo met
h
y
l
ation o

fd
isperse
d
retrotransposons and causes meiotic failure in spermatoc
y
tes
.
2
.
3
DNMT Meth
y
ltransferase Activit
y
: A Complex Issu
e
D
NMTs have commonl
y
been classified as either “maintenance” (DNMT1
)
or “
d
e novo” (DNMT3) met
h
y
l
trans
f
erases. T

h
is c
l
assi
fi
cation is
b
ase
d
o
n
t
he observation that Dnmt1 interacts with proliferatin
g
-cell nuclear anti
g
en
(
PCNA) (C
h
uang et a
l
. 1997), an auxi
l
iary component o
f
t
h
eDNArep
l

ication
complex, and localizes to replication foci (Leonhar dt et al. 1992). Yet it i
s
emerging with increasing clarity that this classification is far too simplistic
.
I
n human colorectal cancer cells, for example, there is evidence that DNA
m
ethylation patterns are maintained not by DNMT1 alone but by cooperatio
n
b
etween DNMT1 an
d
DNMT3B (R
h
ee et a
l
. 2000, 2002; Ting et a
l
. 2004). T
h
e
e
ff
ects of
Dnmt3a
a
n
d
D

nmt3b
disru
p
tion in ES cells likewise indicate tha
t
b
ot
h
Dnmt3a an
d
Dnmt3
b
are invo
l
ve
d
in maintaining DN A met
h
y
l
ation
patterns (Chen et al. 2003). Dnmt1, on the other hand, shows little or no d
e
novo methylation activity in vivo. Li and coworkers have recently propose
d
a model for the action of these three DNMTs
(
Chen et al. 2003
)
: DNMT1

would be the main maintenance enzyme, acting with high efficiency bu t not
f
u
ll
accuracy. DNMT3A an
d
DNMT3B, via t
h
eir
d
enovoactivity,wou
ld
act a
s
“
proofreaders,” restoring CpG methylation at sites left untouched by DNMT1.
DNMT3L s
h
ows no met
h
y
l
trans
f
erase activity,
b
ut it is nevert
h
e
l

ess in
-
volved in the re
g
ulation of DNA meth
y
lation. As mentioned above, it con
-
t
ri
b
utes particu
l
ar
l
y to esta
bl
is
h
ing genomic imprinting
d
uring gametogene
-
s
is. It would appear to act as a cofactor for Dnmt3a, enhancin
g
the latter’s d
e
novo activity (Bourc’his et al. 2001; Bourc’his and Bestor 2004; K a neda et al.
2004

)
.
DNA Meth
y
ltransferases: Facts, Clues, M
y
sterie
s
5
1
A
l
t
h
oug
h
DNMT2, as mentione
d
a
b
ove,
h
as retaine
d
on
l
yoneo
f
t
h

e
d
o-
m
ains characteristic of DNMTs, the meth
y
ltransferase domain, it was not
shown until recentl
y
to be catal
y
ticall
y
active (Hermann et al. 2003). It was
a
l
so s
h
own to
d
isp
l
ay a certain sequence speci
fi
city
f
or centromeric struc
-
t
ures. This recent observation will likel

y
revive interest in this still-m
y
steriou
s
m
em
b
er o
f
t
h
e DNMT
f
ami
l
y.
3
How Do DNMTs Interfere w
i
th T ranscr
i
pt
i
on?
DNMTs participate in
g
ene silencin
g
, but how? It has been known for man

y
y
ears that DNA meth
y
lation and chromatin structure are connected. In mam
-
m
a
l
ian genomes,
f
or examp
l
e,
h
ig
hl
eve
l
so
f
DNA met
h
y
l
ation coinci
d
ewit
h
heterochromatic re

g
ions (Razin and Cedar 1977). Also, meth
y
lated CpG is
-
l
an
d
s(suc
h
as t
h
ose o
f
t
h
e
f
ema
l
e-inactivate
d
Xc
h
romosome) a
pp
ear in
closed, transcriptionall
y
silent chromatin with deacet

y
lated histones, whereas
unmeth
y
lated islands in
g
ene promoters are transcriptionall
y
favorable an
d
have an open chromatin structure with hi
g
hl
y
acet
y
lated histones (Bird and
W
olffe 1999)
.
T
h
e mec
h
anistic
b
asis o
f
t
h

e
l
in
kb
etween DNA met
h
y
l
ation an
d
c
h
romatin
structure has lon
g
remained obscure, but the recent explosion in knowled
g
e
o
n
h
ow c
h
romatin organization mo
d
u
l
ates gene transcription
h
as pave

d
t
he
w
a
y
towards elucidatin
g
this link.As described below andSect. 3.2 with special
emphasis on DNMTs, it is now increasin
g
l
y
clear that DNA meth
y
lation an
d
chromatin or
g
anization work hand in hand to repress
g
ene expression
.
3.1
Cross-talk and Transcr
i
pt
i
onal S
i

lenc
i
n
g
Initial
p
a
p
ers from the laboratories of A. Bird and A. Wolffe were the firs
t
t
ounvei
l
amec
h
anistic co nnection
b
etween DNA met
hyl
ation an
dh
istone
m
odification. The
y
showed that meth
y
l-CpG bindin
g
domain (MBD) pro

-
t
eins, w
h
ic
h
se
l
ective
l
y recognize met
h
y
l
ate
d
CpG
d
inuc
l
eoti
d
es, are compo-
nents of—or establish contacts with—histone deacet
y
lase (HDAC) complexes
(Jones et a
l
. 1998; Nan et a
l

. 1998). HDACs remove acety
l
groups
f
rom
h
iston
e
t
ai
l
san
dh
e
l
p to maintain nuc
l
eosomes in a compact, transcriptiona
lly
si
l
en
t
s
t
a
t
e
.
Next, a much more direct connection between CpG meth

y
lation an
d
d
eacet
y
lation was identified: DNMTs appear to repress transcription throu
gh
recruitment o
fh
istone
d
eacety
l
ases (Burgers et a
l
. 2002). T
h
e
f
act t
h
at eac
h
5
2
C
. Brenner · F. Fuks
D
NMT associates wit

h
HDAC prompts t
h
e question: W
h
yist
h
is contact nec-
essar
y
? One clue mi
g
ht lie in the abilit
y
of DNMTs to act as maintenanc
e
and/or de novo meth
y
ltransferases. A challen
g
e for the cell is to restor e in
new
l
yrep
l
icate
d
DN A t
h
ec

h
romatin structure nee
d
e
d
to maintain t
h
e tran-
s
criptional activit
y
states dictated b
y
chromatin modifications. In the cas
e
o
f
at
l
east one maintenance enzyme, DNMT1, its association wit
h
HDAC is
particularl
y
attractive: It occurs predominantl
y
at replication foci durin
g
the
l

ate S-
ph
ase, w
h
en most o
f
t
h
e
h
eteroc
h
romatin is
d
u
pl
icate
d
(Rountree et a
l
.
2000). DNMT1 ma
y
thus be necessar
y
to ensure that the histones formin
g
th
e
nucleosomes assembled at newl

y
replicated sites are deacet
y
lated.
A
n unexpecte
dfi
n
d
ing
h
as emerge
df
rom t
h
e stu
d
yo
f
t
h
e DNMT–HDA
C
interaction, mediated b
y
the non-catal
y
tic N-terminal portion of the DNMT.
Intriguing
l

y,transcriptiona
l
si
l
encing
d
oes not require preservation o
f
DNMT
enz
y
matic activit
y
. In addition, Dnmt3L can still recruit the HDAC repressiv
e
mac
h
inery
d
espite its
l
ac
k
o
f
DNMT activity (Dep
l
us et a
l
. 2002)

.
It thus seems that DNMTs can carr
y
out some HDAC-associated function
s
independentl
y
of their abilit
y
to meth
y
late CpG sites, at least in certain cir
-
cumstances. Althou
g
h these observations remain to be confirmed in vivo, it
is temptin
g
to speculate that DNMTs are more versatile than initiall
y
antic-
ipate
d
.Inot
h
er wor
d
s, t
h
ey may

b
emu
l
ti
f
acete
d
proteins per
f
orming ot
h
er
function s in addition to meth
y
lation of CpG dinucleotides
.
More recent
l
y, DNMTs
h
ave
b
een imp
l
icate
d
in anot
h
er c
h

romatin-re
l
ate
d
transcriptional repression process, invo lvin
g
meth
y
lation of histone H3 at
ly
sine 9. This connection was first evidenced in the ascom
y
cete fun
g
us
Neu
-
ros
p
ora crassa andintheplant
A
rabido
p
sis thalian
a
by
E. Selker’s and S.
Jacobsen’s
g
roups, respectivel

y
(Jackson et al. 2002; Selker et al. 2003). It wa
s
sh
own t
h
atmutationsint
h
egenes
d
im-
5
of
Neurospora an
d
k
r
y
ptonite o
f
Ara
-
b
idopsis result in loss of DNA meth
y
lation in these or
g
anisms. Excitement
a
rose

f
rom t
h
e
fi
n
d
ing t
h
at t
h
ese genes enco
d
e H3-K9
h
istone met
h
y
l
trans-
f
e
r
ases.
The mechanisms linking DNA methylation to histone methylation remain
unclear and there are likel
y
several wa
y
s that connect these two epi

g
eneti
c
events. In Arabidopsis, the adaptor pro tein LHP1 (the homolog of the mam
-
ma
l
ian
h
eteroc
h
romatin protein 1, HP1) is not nee
d
e
d
to maintain DN
A
methylation, and at least deacetylase HDA6 is instead required (Bender 2004).
In Neurospora,
h
owever, t
h
e HP1 protein cou
ld b
e a possi
bl
e
l
in
kb

etwee
n
D
NA and histone meth
y
lation, since it has been shown that HP1 is require
d
f
or DNA met
h
y
l
ation (Se
lk
er et a
l
. 2003). Accor
d
ing to t
h
e current wor
k
ing
model in Neurospora, meth
y
lation at H3-K9 b
y
DIM5 would create a bindin
g
p

latform for HP1. This ada
p
tor
p
rotein would then recruit the DIM2 DNMT.
In t
h
is way,
h
istone met
h
y
l
ation wou
ld
in
fl
uence DNA met
h
y
l
ation
.