DNA Methylation: Basic Mechanisms - Part 7 docx

Establishment and Maintenance of DNA Meth
y
lation Pa tterns in Mammals 19
3
et a
l
. 2001; C
h
en et a
l
. 2004). Suc
h
a
l
oca
l
ization
p
attern seems to
b
e
d
e
-
p
endent on H3-K9 meth
y
lation, as Dnmt3b and HP1 fail to concentrate a
t
heter ochromatic foci in Suv39h1 and Suv39h2 double knockout cells (Lehn
-

ertz et a
l
. 2003). Co-IP ex
p
eriments s
h
ow t
h
at Dnmt3a an
d
Dnmt3
bf
orm
complexes with HP1, apparentl
y
in a Suv39h-independent manner (Fuks e
t
a
l
. 2003; Le
h
nertz et a
l
. 2003
)
. Dnmt3a an
d
Dnmt3
bh
ave a

l
so
b
een s
h
own to
associate with H3-K9 meth
y
ltransferase activit
y
(Fuks et al. 2003; Lehnertz e
t
a
l
. 2003). One stu
d
ys
h
ows t
h
at Dnmt3a, via its ATRX-
h
omo
l
ogy
d
omain,
d
i-
rectl

y
interacts with Suv39h1 (Fi
g
. 1; Fuks et al. 2003). A separate stud
y
show
s
t
hat the H3-K9 meth
y
ltransferase activities associated with Dnmt3b in wild
-
t
ype an
d
Suv39
hd
ou
bl
e
k
noc
k
out ce
ll
s are equa
ll
yro
b
ust, suggesting t

h
a
t
Dnmt3b forms one or more histone-DNA meth
y
lation complexes containin
g
Suv39
h
-unre
l
ate
d
H3-K9 met
h
y
l
trans
f
erases (Le
h
nertz et a
l
. 2003)
.
3.
2
.
1
0

S
UMO-1
,
Ubc9
,
PIAS1
,
and PIASx
α
The small ubiquitin-related protein SU MO-1 posttranslationall
y
modifies
m
any proteins wit
h
ro
l
es in
d
iverse processes inc
l
u
d
ing regu
l
ation o
f
tran-
scription, c
h

romatin structure, an
d
DNA repair. SUMO-1 is
l
i
g
ate
d
to
ly
sine
residues in substrate proteins via a three-step enz
y
matic process involvin
g
a heterodimeric E1 activatin
g
enz
y
me (SAE1/SAE2), an E2 con
j
u
g
atin
g
en
-
z
y
me (Ubc9), and a number of E3 li

g
atin
g
enz
y
mes (PIAS proteins, RanBP2
,
an
d
Pc2). In contrast to u
b
iquitination, sumoy
l
ation
d
oes not promote pr otei
n
d
e
g
radation but instead modulates several other as pects of protein function,
inc
l
u
d
ing su
b
ce
ll
u

l
ar
l
oca
l
ization,protein–proteininteractions,protein–DN
A
interactions, and enz
y
matic activit
y
(Gill 2004).
Using yeast two-
h
y
b
ri
d
screens, two groups
h
ave i
d
enti
fi
e
d
severa
l
com
-

p
onents of the sumo
y
lation machiner
y
as Dnmt3a- and Dnmt3b-interactin
g
p
artners. These include Ubc9, PIAS1, and PIAS
x
α
.
Th
e
in
te
r
act
i
o
n
sa
r
e
f
u
r-
th
er con
fi

rme
db
yco-
l
oca
l
ization, co-IP, an
d
GST pu
ll
-
d
own experiments
.
Muta
g
enesis anal
y
ses map the interaction domain to the N-terminal re
g
ion
s
of
Dnmt3a an
d
Dnmt3
b
(Fig. 1). Dnmt3a an
d
Dnmt3

b
can
b
e sumoy
l
ate
d
wh
en co-trans
f
ecte
d
wit
hSU
M
O
-1 in ce
ll
sorw
h
en incu
b
ate
d
wit
h
recom
b
i
-

nant E1 (SAE1/SAE2), U
b
c9, an
d
SUMO-1 in t
h
e presence o
f
ATP (Kang et
al. 2001; Lin
g
et al. 2004). In co-transfection experiments, overexpression of
SUMO-1 inhibits Dnmt3a-HDAC interaction and relieves Dnmt3a-mediate
d
t
ranscriptiona
l
repression o
f
areporter
g
ene (Lin
g
et a
l
. 2004). T
h
ese re-
sults su
gg

est that sumo
y
lation ma
y
re
g
ulate the functions of Dnmt3a an
d
Dnmt3
b
.
19
4
T
.
C
hen · E. Li
3.
2
.
11
Dnmt3L
As discussed above, Dnmt3L belon
g
s to the Dnmt3 famil
y
, but does not
h
ave enzymatic activity. Dnmt3L contains an ATRX-
h

omo
l
ogy
d
omain t
h
a
t
is closel
y
related to that of Dnmt3a and Dnmt3b. Its C-terminal re
g
ion shows
s
equence
h
omo
l
ogy to t
h
ecata
l
ytic
d
omain o
f
Dnmt3a an
d
Dnmt3
b

,
b
ut
l
ac
k
s
s
ome residues known to be critical for enz
y
matic activit
y
, includin
g
the P
C
d
ipepti
d
eatt
h
e active site (Fig. 1; Aapo
l
aeta
l
. 2001; Hata e t a
l
. 2002). T
h
e

expression pattern of
Dnmt3L
i
s strikin
g
l
y
similar to that o
f
Dnmt3a
a
n
d
D
nmt3b
durin
g
mouse development (Hata et al. 2002). Genetic studies have
d
emonstrate
d
t
h
at
D
nmt3
L
,l
i
k

e
D
nmt3
a
,
is essentia
lf
or t
h
e esta
bl
is
h
men
t
of
g
enomic imprintin
g
.Althou
g
h disruption of D
nmt3L
i
n the z
yg
ote does
not a
ff
ect em

b
ryonic
d
eve
l
opment,
D
nmt3
L
−
/
−
/
/
f
ema
l
es
f
ai
l
to esta
bl
is
h
ma
-
ternal meth
y
lation imprints in the ooc

y
tes, which leads to loss of monoalleli
c
expression o
f
mat erna
ll
yimprinte
d
genes an
dd
eve
l
opmen ta
ld
e
f
ects in t
h
e
offs prin
g
,an
d
D
nmt3
L
−
/
−

/
/
ma
l
es s
h
ow
d
e
f
ects in spermato
g
enesis (Bourc’
h
is
and Bestor 2004; Bourc ’his et al. 2001; Hata et al. 2002). Dnmt3L has been
sh
own to
d
irect
ly
interact wit
h
Dnmt3a an
d
Dnmt3
b
via t
h
eir C-termina

l
re
g
ions, resultin
g
in stimulation of the catal
y
tic activit
y
of these de novo
met
h
y
l
trans
f
erases (Fig. 1; C
h
e
d
in et a
l
. 2002; Gow
h
er et a
l
. 2005; Hata et a
l.
2002; Sueta
k

eeta
l
. 2004). In vitro assa
y
ss
h
ow t
h
at comp
l
ex
f
ormation
b
e
-
tween Dnmt3a an
d
Dnmt3L acce
l
erates DNA an
d
A
d
oMet
b
in
d
ing to Dnmt3a
(

Gow
h
er et a
l
. 2005
)
. Moreover, Dnmt3L
h
as
b
een s
h
own to associate wit
h
H
DAC1 via its ATRX-homolo
gy
domain and function as a transcriptional re-
pressorinreporter s
y
stems(Fi
g
.1;Aapo
l
aeta
l
. 2002; Dep
l
us et a
l

. 2002). Ta
k
e
n
to
g
ether, Dnmt3L ma
y
re
g
ulate
g
enomic imprintin
g
b
y
enhancin
g
the activit
y
o
f
Dnmt3a or
b
y increasing t
h
e accessi
b
i
l

ity o
f
Dnmt3a to imprinte
dl
oci.
4
Conclud
i
n
g
Remarks
Over the past several
y
ears, our understandin
g
of the molecular mecha
-
nisms
b
yw
h
ic
h
DNA met
h
y
l
ation patterns are esta
bl
is

h
e
d
an
d
maintaine
dh
a
s
been
g
rowin
g
steadil
y
. The identification of a
g
rowin
g
number of chromatin-
associate
d
proteins t
h
atinteract wit
h
oneor more Dnmts supports t
h
e
h

ypot
h-
esis t
h
at c
h
romatin structure an
d
c
h
romatin proteins p
l
a
y
important ro
l
es i
n
t
h
eregu
l
ation o
f
t
h
e activities an
d
speci
fi

cities o
f
DNA met
h
y
l
trans
f
erases. I
t
sh
ou
ld b
enote
d
,
h
o wever, t
h
at man
y
o
f
t
h
e Dnmt-interactin
g
partners wer
e
identified b

y
candidate approaches or
y
east two-h
y
brid screens. Much need
s
to
b
e
d
one to veri
f
yt
h
ese interactions. Moreover, wit
h
t
h
e exception o
f
a
f
e
w
Establishment and Maintenance of DNA Meth
y
lation Pa tterns in Mammals 19
5
cases suc

h
as t
h
e Dnmt3a–Dnmt3L interaction, t
h
e
f
unctiona
l
im
pl
ications o
f
t
hese interactions remain lar
g
el
y
unknown due to the lackof
g
enetic evidence
.
A
nother challen
g
e we are facin
g
is how to assemble the individual interactin
g
p

roteins into regu
l
atory comp
l
exes an
d
pat
h
ways. In t
h
e
f
uture, we expect t
o
see
m
o
r
e stud
i
es t
h
at add
r
ess t
h
ese
i
ssues.
Re

f
eren
c
e
s
Aapo
l
aU,Ly
l
eR,Kro
h
n K, Antonara
k
is SE, Peterson P (2001) Iso
l
ation an
d
initia
l
characterization of the mouse Dnmt3 l
g
ene. C
y
to
g
enet Cell Genet 92:122–126
Aapola U, Liiv I, Peterson P (2002) Im printin
g
re
g

ulator DNMT3L is a transcriptiona
l
repressor associated with histone deacet
y
lase activit
y
. Nucleic Acids Res 30:3602
–
3608
Ao
k
i A, Sueta
k
e I, Miyagawa J, Fujio T, C
h
ijiwa T, Sasa
k
i H, Tajima S (2001) Enzymati
c
properties o
fd
e novo-type mouse DNA (cytosine-5) met
h
y
l
trans
f
erases. Nuc
l
ei

c
Aci
d
s Res 29:3506–351
2
Aoki K, Men
g
G, Suzuki K, Takashi T, Kameoka Y, Nakahara K, Ishida R, Kasai M
(1998) RP58 associates with condensed chromatin and mediates a se
q
uence
-
sp
ecific transcri
p
tional re
p
ression. J Biol Chem 273:26698–2670
4
AraujoFD, Croteau S,S
l
ac
k
AD,Mi
l
utinovic S, Bigey P, Price GB, Zannis-Hajopou
l
os M
,
Sz

y
f M (2001) The DNMT1 tar
g
et reco
g
nition domain resides in the N terminus.
J
Biol Chem 276:6930–6936
Bachman KE, Rountree MR, Ba
y
lin SB (2001) Dnmt3a and Dnmt3b are transcriptional
re
p
ressors that exhibit uni
q
ue localization
p
ro
p
erties to heterochromatin. J Bio
l
Ch
em 276:32282–3228
7
Bannister AJ, Zegerman P, Partri
d
ge JF, Mis
k
aEA,T
h

omas JO, A
ll
s
h
ire RC, Kouzari-
d
es T (2001) Se
l
ective recognition o
f
met
h
y
l
ate
dl
ysine 9 on
h
istone H3
b
yt
h
e
HP1 c
h
romo
d
omain. Nature 410:120–124
Beard C, Li E, Jaenisch R (1995) Loss of meth
y

lation activates Xist in somatic but not
in embr
y
onic cells. Genes Dev 9:2325–2334
Bestor T, LaudanoA,Mattaliano R, In
g
ram V(1988) Clonin
g
and sequencin
g
ofacDN
A
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
omai
n
o
f
t
h
e mamma
l
ian enzymes is re
l
ate
d
to
b
acteria
l
restriction met
h
y
l
trans
f
erases.
JMo
l
Bio
l
203:971–983

Bird A (2002) DNA meth
y
lation patterns and epi
g
enetic memor
y
. Genes Dev 16:6–21
Bird AP, Wolffe AP (1999) Meth
y
lation-induced repression—belts, braces, and chro-
matin.
C
ell
99
:451–454
Bourc’
h
is D, Bestor TH (2004) Meiotic catastrop
h
ean
d
retrotransposon reactivation
in ma
l
e germ ce
ll
s
l
ac
k

ing Dnmt3L. N ature 431:96–99
Bourc’
h
is D, XuGL, LinCS, Bo
ll
man B, BestorTH (2001)Dnmt3L an
d
t
h
e esta
bl
is
h
ment
of maternal
g
enomic imprints. Science 294:2536–253
9
Brehm A, Miska EA, McCance D
J
, Reid
J
L, Bannister A
J
, Kouzarides T
(
1998
)
Retinoblastoma protein recruits histone deacet
y

lase to repress transcription
.
N ature 3
9
1:5
9
7–60
1
19
6
T
.
C
hen · E. Li
B
renner C, De
p
lus R, Didelot C, Loriot A, Vire E, De Smet C, Gutierrez A, Danovi D
,
Bernar
d
D, Boon T, Pe
l
icci PG, Amati B, Kouzari
d
es T,
d
eLaunoitY,DiCroceL
,
Fu

k
s F (2005) Myc represses transcription t
h
roug
h
recruitment o
f
DNA met
h
y
l-
trans
f
erase corepressor. EMBO J 24:336–346
C
ameron EE, Bac
h
man KE, Myo
h
anen S, Herman JG, Bay
l
in SB (1999) Synergy o
f
demeth
y
lation and histone deacet
y
lase inhibition in the re-expression of
g
enes

silenced in cancer. Nat
G
enet 21:103–10
7
C
hedin F, Lieber MR, Hsieh CL (2002) The DNA meth
y
ltransferase-like protei
n
DNMT3L stimulates de novo meth
y
lation b
y
Dnmt3a. Proc Na tl Acad Sci U S
A
99:1
6
91
6
–1
6
92
1
Ch
en T, Li E (2004) Structure an
df
unction o
f
eu
k

aryotic DNA met
h
y
l
trans
f
erases
.
Curr Top Dev Bio
l
60:55–89
C
hen T, UedaY, Xie S,LiE(2002) AnovelDnmt3a isoform
p
roduced from an alternativ
e
p
romoter localizes to euchr omatin and its ex
p
ression correlates with active d
e
novo meth
y
lation. J Biol Chem 277:38746–38754
C
hen T, Ueda Y, Dod
g
e JE, Wan
g
Z, Li E (2003) Establishmen t and maintenance of

genomic met
h
y
l
ation patterns in mouse em
b
ryonic stem ce
ll
s
b
y Dnmt3a an
d
Dnmt3
b
.Mo
l
Ce
ll
Bio
l
23:5594–560
5
Ch
en T, Tsujimoto N, Li E (2004) T
h
e PWWP
d
omain o
f
Dnmt3a an

d
Dnmt3
b
is
require
df
or
d
irecting DNAmet
h
y
l
ation to t
h
emajorsate
ll
ite repeats atpericentric
heter ochromatin. Mol
C
ell Biol 24:
9
048–
9
05
8
C
hen
g
X, Roberts RJ (2001) AdoMet-dependent meth
y

lation, DNA meth
y
ltransferases
and base flippin
g
. Nucleic Acids Res 29:3784–379
5
Ch
uang LS, Ian HI, Ko
h
TW, Ng HH, Xu G, Li BF (1997) Human DNA-(cytosine-5
)
met
h
y
l
trans
f
erase-PCNAcomp
l
ex asa target
f
or p21WAF1.Science 277:1996–200
0
D
ep
l
us R, Brenner C, Burgers WA, Putmans P, Kouzari
d
es T,

d
e Launoit Y, Fu
k
sF
(2002) Dnmt3L is a transcriptional repressor that recruits histone deacet
y
lase.
Nuc
l
eic Aci
d
s Res 30:3831–3838
D
iCroceL
,
Raker VA
,
Corsaro M
,
Fazi F
,
Fanelli M
,
Faretta M
,
Fuks F
,
Lo Coco F
,
Kouzarides T, Nervi C, Minucci S, Pelicci PG (2002) Meth

y
ltransferase recruitmen
t
and DNA h
y
permeth
y
lation of tar
g
et pr omoters b
y
an onco
g
enic transcriptio
n
factor.
S
cience 2
9
5:107
9
–1082
D
ing F, C
h
ai
ll
et JR (2002) In vivo sta
b
i

l
ization o
f
t
h
e Dnmt1 (cytosine-5)
-
meth
y
ltransferase protein. Proc Natl Acad Sci U S A 99:14861–14866
D
od
g
e JE, Ramsaho
y
e BH, Wo ZG, Okano M, Li E (2002) De novo meth
y
lation o
f
MMLV provirus in embr
y
onic stem cells: CpG versus non-CpG meth
y
lation. Gene
28
9
:41–4
8
D
o

d
ge JE, O
k
ano M, Dic
k
F, Tsujimoto N, C
h
en T, Wang S, Ue
d
a Y, Dyson N, Li
E
(2005) Inactivation o
f
Dnmt3
b
in mouse em
b
ryonic
fib
ro
bl
asts resu
l
ts in DN
A
h
ypomet
h
y
l

ation, c
h
romosoma
l
insta
b
i
l
ity, an
d
spontaneous immorta
l
ization
.
JBio
l
C
h
em 280:17986–17991
D
on
g
A, Yoder JA, Zhan
g
X, Zhou L, Bestor TH, Chen
g
X (2001) Structure of human
DNMT2, an eni
g
matic DNA meth

y
ltransferase homolo
g
that displa
y
s denaturant
-
resistant bindin
g
to DNA. Nucleic Acids Res 29:439–448
D
unaief JL, Strober BE, Guha S, Khavari PA, Alin K, Luban J, Be
g
emann M, Crab
-
tree GR, Goff SP (1994) The retinoblastoma
p
rotein and BRG1 form a com
p
le
x
an
d
cooperate to in
d
uce ce
ll
cyc
l
e arrest. Ce

ll
79:119–13
0
Establishment and Maintenance of DNA Meth
y
lation Pa tterns in Mammals 19
7
Esteller M, Fra
g
a MF, Paz MF, Campo E, Colomer D, Novo FJ, Calasanz MJ, Galm O,
Guo M, Benitez J, Herman JG (2002) Cancer epigenetics an
d
met
h
y
l
ation. Scienc
e
29
7:
180
7–
1808
Fatemi M, Hermann A, Pra
dh
an S, Je
l
tsc
h
A (2001) T

h
eactivityo
f
t
h
e murine DN
A
met
h
y
l
trans
f
erase Dnmt1 is contro
ll
e
db
y interaction o
f
t
h
ecata
l
ytic
d
omain
with the N-terminal part of the enz
y
me leadin
g

to an allosteric activation of th
e
enz
y
me after bindin
g
to meth
y
lated DNA. J Mol Biol 309:1189–1199
Fen
g
Q, Zhan
g
Y (2001) The MeCP1 complex represses transcription throu
g
h pref-
erential bindin
g
, remodelin
g
, and deacet
y
latin
g
meth
y
lated nucleosomes. Gene
s
D
e

v1
5:8
2
7
–
83
2
Freitag M, Hic
k
ey PC, K
hl
a
f
a
ll
a
h
TK, Rea
d
ND, Se
lk
er EU (2004) HP1 is essentia
lf
or
DNA met
h
y
l
ation in neurospora. Mo
l

Ce
ll
13:427–43
4
Fu
k
s F, Burgers WA, Bre
h
mA,Hug
h
es-Davies L, Kouzari
d
es T (2000) DNA met
h
y
l
-
trans
f
erase Dnmt1 associates wit
hh
istone
d
eacety
l
ase activity. Nat Genet 24:88
–
91
Fuks F, Bur
g

ers WA, Godin N,Kasai M,Kouzarides T (2001) Dnmt3a bindsdeacet
y
lases
and is recruited b
y
a sequence-specific repressor to silence transcription. EMBO
J
20:2536–254
4
Fuks F, Hurd PJ, Deplus R, Kouzarides T (2003) The DNA meth
y
ltransferases associate
with HP1and the SUV39H1histone meth
y
ltransferase. Nucleic Acids Res 31:2305–
2
3
12
Ge Y-Z, Pu M-T, Gowher H, Wu H-P, Din
g
J-P, Jeltsch A, Xu G-L (2004) Chromati
n
targeting o
fd
e novo DNA met
h
y
l
trans
f

erases
b
yt
h
e PWWP
d
omain. J Bio
l
C
h
em
2
7
9
:
2
5447–
2
5454
Gi
ll
G (2004) SUMO an
d
u
b
iquitin in t
h
enuc
l
eus:

d
i
ff
erent
f
unctions, simi
l
ar mec
h
a
-
n
i
sms? Genes Dev 18:2046–2059
Gowher H, Liebert K , Hermann A, Xu G,
J
eltsch A
(
2005
)
Mechanism of stimulation o
f
catal
y
ticactivit
y
of Dnmt3Aand Dnmt3B DNA-(c
y
tosine-C5)-meth
y

ltransferase
s
b
y
Dnmt3L. J Biol Chem 280:13341–13348
GrignaniF,DeMatteisS,NerviC,TomassoniL,Ge
l
mettiV,CioceM,Fane
ll
iM
,
Rut
h
ar
d
t M, Ferrara FF, Zamir I, Seiser C, Lazar MA, Minucci S, Pe
l
icci PG (1998
)
Fusion proteins o
f
t
h
e retinoic aci
d
receptor-a
l
p
h
a recruit

h
istone
d
eacety
l
ase i
n
promye
l
ocy tic
l
eu
k
aemia. Natur e 391:815–81
8
Hata K, O
k
ano M, Lei H, Li E (2002) Dnmt3L cooperates wit
h
t
h
e Dnmt3
f
ami
l
yo
fde
novo DNA met
h
y

l
trans
f
erases to esta
bl
is
h
materna
l
imprints in mice. Deve
l
op
-
ment 12
9
:1
9
83–1
993
Hendrich B,Bird A(1998) Identification andcharacterizationof afamil
y
ofmammalia
n
meth
y
l-CpG bindin
g
proteins. Mol Cell Biol 18:6538–654
7
Hermann A, Sc

h
mitt S, Je
l
tsc
h
A (2003) T
h
e
h
uman Dnmt2
h
as r esi
d
ua
l
DNA-
(cytosine-C5) met
h
y
l
trans
f
erase activity. J Bio
l
C
h
em 278:31717–3172
1
Hsie
h

CL (1999) In vivo activity o
f
murine
d
e novo met
h
y
l
trans
f
erases, Dnmt3a an
d
Dnmt3
b
.Mo
l
Ce
ll
Bio
l
19:8211–8218
Hun
g
MS, Karthike
y
an N, Huan
g
B, Koo HC, Ki
g
er J, Shen CJ (1999) Drosophila

proteins related to vertebrate DNA (5-c
y
tosine) meth
y
ltransferases. Proc Nat
l
Acad
S
ci
US
A
9
6:11
9
40–11
9
4
5
19
8
T
.
C
hen · E. Li
I
ida T, Suetake I, Ta
j
ima S, Morioka H, Ohta S, Obuse C, Tsurimoto T (2002) PCNA
c
l

amp
f
aci
l
itates action o
f
DNA cytosine met
h
y
l
trans
f
erase 1 on
h
emimet
h
y
l
ate
d
DNA. Genes Ce
ll
s 7:997–100
7
J
ac
k
son JP, Lin
d
rot

h
AM, Cao X, Jaco
b
sen SE (2002) Contro
l
o
f
CpNpG DNA met
h
y
l
a-
tion
b
yt
h
e KRYPTONITE
h
istone H3 met
h
y
l
trans
f
erase. N ature 416:556–560
J
aenisch R (1997) DNA meth
y
lation and imprintin
g

:wh
y
bother? Trends Genet 13:323–
3
2
9
J
aenisch R, Bird A (2003) Epi
g
enetic re
g
ulation of
g
ene expression: how the
g
enom
e
integrates intrinsic an
d
environmenta
l
signa
l
s. Nat Genet 33 Supp
l
:245–254
J
ones PL, Veenstra GJ, Wa
d
e PA, Vermaa

k
D, Kass SU, Lan
d
s
b
erger N, Strou
b
ou
l
is 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
K
aneda M, Okano M, Hata K, Sado T, Tsu
j
imoto N, Li E, Sasaki H (2004) Essential role
forde novoDNAmeth
y
ltransferasesDnmt3a in paternal andmaternal imprintin
g.
N ature 42
9
:
9
00–
9
03
K
an
g
ES, Park CW, Chun
g
JH (2001) Dnmt3b, de novo DNA meth
y
ltransferase, in-
teracts wit
h
SUMO-1 an

d
U
b
c9 t
h
roug
h
its N-termina
l
region an
d
is su
b
ject to
mo
d
i
fi
cation
b
y SUMO-1. Bioc
h
em Biop
h
ys R es Commun 289:862–868
K
imura H, S
h
iota K (2003) Met
h

y
l
-CpG-
b
in
d
ing protein, MeCP2, is a target mo
l
ecu
le
f
or maintenance DNA met
h
y
l
trans
f
erase, Dnmt1. J Bio
l
C
h
em 278:4806–4812
K
unert 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
iates
DNA meth
y
lation in Drosophila. Development 130:5083–509
0
L
achner M, O’Carroll D, Rea S, Mechtler K, Jenuwein T (2001) Meth
y
lation of histone
H3 l
y
sine 9 creates a bindin
g
site for HP1 proteins. Nature 410:116–12
0
L
ehnertz B, Ueda Y, Deri
j
ck AA, Braunschwei
g
U, Perez-Bur

g
os L, Kubicek S, Chen T,
Li E, Jenuwein T, Pet ers AH (2003) Suv39h-mediated histone H3 l
y
sine 9 meth
y
la-
tion
d
irects DNA met
h
y
l
ation to major sate
ll
ite repeats at pericentric
h
eteroc
h
ro-
matin. Curr Bio
l
13:1192–120
0
L
ei H, O
h
SP, O
k
ano M, Juttermann R, Goss KA, Jaenisc

h
R, Li E (1996) De novo DNA
cytosine met
h
y
l
trans
f
erase activities in mouse em
b
ryonic stem ce
ll
s. Deve
l
op
-
ment 122:31
9
5–3205
L
eonhardt H, Pa
g
e AW, Weier H-U, Bestor TH (1992) A tar
g
etin
g
sequence direct
s
DNA meth
y

ltransferase to sites of DNA r eplication in mammalian nuclei. Cel
l
7
1
:865
–
873
L
ewis 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
L
i E (2002) Chroma tin modification and epi
g
enetic repro
g
rammin
g
in mammalian
develo

p
ment. Nat Rev Genet 3:662–673
L
i E, Bestor TH, Jaenisc
h
R (1992) Targete
d
mutation o
f
t
h
e DNA met
h
y
l
trans
f
eras
e
gene resu
l
ts in em
b
ryonic
l
et
h
a
l
ity. Ce

ll
69:915–92
6
L
i E, Bear
d
C, Jaenisc
h
R (1993) Ro
l
e
f
or DNA met
h
y
l
ation in genomic imprinting
.
N
ature 366:362–365
L
ian
g
G, Chan MF, Tomi
g
ahara Y, Tsai YC, Gonzales FA, Li E, Laird PW, Jones PA (2002)
Cooperativit
y
between DNA meth
y

ltransferases in the maintenance meth
y
lation
of re
p
etitive elements. Mol Cell Biol 22:480–49
1
Establishment and Maintenance of DNA Meth
y
lation Pa tterns in Mammals 19
9
Lin RJ, Na
gy
L, Inoue S, Shao W, Miller WH Jr, Evans RM (1998) Role of the histone
d
eacety
l
ase comp
l
ex in acute p romye
l
ocy tic
l
eu
k
aemia. Nature 391:811–81
4
Ling Y, San
k
pa

l
UT, Ro
b
ertson AK, McNa
ll
yJG,KarpovaT,Ro
b
ertson KD (2004)
Mo
d
i
fi
cation o
fd
e novo DNA met
h
y
l
trans
f
erase 3a (Dnmt3a)
b
y S UMO-1 mo
d
u-
l
at es its interaction wit
hh
istone
d

eacety
l
ases (HDACs) an
d
its capacity to repress
transcription. Nuc
l
eic Aci
d
s Res 32:598–610
Liu K, Wan
g
YF, Cantemir C, Muller MT (2003) Endo
g
enous assa
y
sofDNAmeth
y
l
-
transferases: evidence for differential activities of DNMT1
,
DNMT2
,
and DNMT
3
in mammalian cells in vivo. Mol
C
ell Biol 23:270
9

–271
9
Liu Y, Oa
k
e
l
ey EJ, Sun L, Jost JP (1998) Mu
l
tip
l
e
d
omains are invo
l
ve
d
in t
h
e targeting
o
f
t
h
e mouse DNA met
h
y
l
trans
f
erase to t

h
eDNArep
l
ication
f
oci. Nuc
l
eic Aci
ds
R
es 26:1038–104
5
Liu Z, Fisher RA
(
2004
)
RGS6 interacts with DMAP1 and DNMT1 and inhibits DMAP
1
transcriptional repr essor activit
y
. J Biol Chem 279:14120–14128
Luo RX, Posti
g
o AA, Dean DC (1998) Rb interacts with histone deacet
y
lase to repress
transcription. Ce
ll
92:463–473
Ly

k
o F, Ramsa
h
oye BH, Kas
h
evs
k
yH,Tu
d
or M, Mastrange
l
o MA, Orr-Weaver TL,
Jaenisc
h
R (1999) Mamma
l
ian (cytosine-5) met
h
y
l
trans
f
erases cause genomi
c
DNA met
h
y
l
ation an
dl

et
h
a
l
ity in Drosop
h
i
l
a. Nat Genet 23:363–36
6
L
y
ko F, Whittaker AJ, Orr-Weaver TL, Jaenisch R (2000) The putative Drosophila
meth
y
ltransferase
g
ene dDnmt2 is contained in a transposon-like element and is
expressed specificall
y
in ovaries. Mech Dev 95:215–21
7
Ma
g
na
g
hi-Jaulin L, Groisman R, Na
g
uibneva I, Robin P, Lorain S, Le Villain JP
,

Troalen F, Trouche D, Harel-Bellan A (1998) Retinoblastoma
p
rotein re
p
resses
transcription
b
y recruiting a
h
istone
d
eacety
l
ase. Nature 391:601–605
Ma
l
agnac F, Bartee L, Ben
d
er J (2002) An Ara
b
i
d
opsis SET
d
omain protein require
df
o
r
maintenance
b

ut not esta
bl
is
h
ment o
f
DNA met
h
y
l
ation. EMBO J 21:6842–6852
Mee
h
an RR, Lewis JD, McK ay S, K
l
einer EL, Bir
d
AP (1989) I
d
enti
fi
cation o
f
a mam
-
ma
l
ian protein t
h
at

b
in
d
s speci
fi
ca
ll
y to DNA containing met
h
y
l
ate
d
CpGs. Ce
ll
5
8:4
99
–507
MuromotoR,Su
g
i
y
ama K, Takachi A, Imot o S, Sat o N, Yamamoto T, Oritani K, Shimo
-
da K, Matsuda T (2004) Ph
y
sical and functional interactions between Daxx an
d
DNA meth

y
ltransferase 1-associated protein, DMAP1. J Immunol 172:2985–299
3
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
ve
s
a
h

istone
d
eacety
l
ase comp
l
ex. N ature 393:386–389
Nie
l
sen SJ, Sc
h
nei
d
er R, Bauer UM, Bannister AJ, Morrison A, O’Carro
ll
D, Firestein R
,
C
l
eary M, Jenuwein T, Herrera RE, Kouzari
d
es T (2001) R
b
targets
h
istone H
3
meth
y
lation and HP1 to promoters. Nature 412:561–56

5
Okano M, Xie S, Li E (1998a) Clonin
g
and characterization of a famil
y
of novel
mammalian DNA (c
y
tosine-5) meth
y
ltransferases. Nat Genet 19:219–220
Okano M, Xie S, Li E (1998b) Dnmt2 is not re
q
uired for de novo and maintenance
meth
y
lation of viral DNA in embr
y
onic stem cells. Nucleic Acids Res 26:2536–254
0
O
k
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
57

200
T
.
C
hen · E. Li
P
annin
g
B, Jaenisch R (1996) DNA h
y
pometh
y
lation can activate Xist expression an
d
si
l
ence X-
l
in
k
e
d
genes. Genes Dev 10:1991–2002
P
ate
l
JH, Lo
b
o
d

aAP,S
h
owe MK, S
h
owe LC, McMa
h
on SB (2004) Ana
l
ysis o
f
genomic
targets revea
l
scomp
l
ex
f
unctions o
f
MYC. Nat Rev Cancer 4:562–568
P
e
l
engaris S, K
h
an M, Evan G (2002) c-MYC: mo re t
h
an just a matter o
fl
i

f
ean
dd
eat
h
.
Nat Rev
C
ancer 2:764–77
6
P
eterson EJ, Bo
g
ler O, Ta
y
lor SM (2003) p53-mediated repression of DNA meth
y
ltrans-
ferase 1 expression b
y
specific DNA bindin
g
. Cancer Res 63:6579–658
2
P
inarbasi E, Elliott J, Hornb
y
DP (1996) Activation of a
y
east pseudo DNA meth

y
l-
transferase b
y
deletion of a sin
g
le amino acid. J Mol Biol 257:804–813
P
ra
dh
an S, Kim GD (2002) T
h
e retino
bl
astoma gene pro
d
uct interacts wit
h
main
-
tenance
h
uman DNA (cytosine-5) met
h
y
l
trans
f
erase an
d

mo
d
u
l
ates its activity
.
EMB
O
J 21:779–788
P
radhan S,Bacolla A,Wells RD,Roberts RJ(1999) RecombinanthumanDNA(c
y
tosine-
5) meth
y
ltransferase. I. Exp ression, purification, and com parison of de novo an
d
maintenance meth
y
lation. J Biol Chem 274:33002–3301
0
Q
iu C, Sawa
d
aK,Z
h
ang X, C
h
eng X (2002) T
h

e PWWP
d
omain o
f
mamma
l
ian DNA
met
h
y
l
trans
f
erase Dnmt3
bd
e
fi
nes a new
f
ami
l
yo
f
DNA-
b
in
d
ing
f
o

ld
s. Nat Struct
Bio
l
9:217–22
4
R
ei
k
W, Dean W, Wa
l
ter J (2001) Epigenetic reprogramming in mamma
l
ian
d
eve
l
op
-
ment.
S
cience 2
9
3:108
9
–10
93
R
heeI,JairKW,YenRW,Len
g

auer C, Herman JG, Kinzler KW, Vo
g
elstein B, Ba
y
lin SB
,
Schuebel KE (2000) CpG meth
y
lation is maintained in human cancer cells lackin
g
DNM
T
1. N ature 404:1003–100
7
Rh
ee I, Bac
h
man KE, Par
k
BH, Jair KW, Yen RW, Sc
h
ue
b
e
l
KE, Cui H, Fein
b
erg AP
,
Lengauer C, Kinz

l
er KW, Bay
l
in SB, Voge
l
stein B (2002) DNMT1 an
d
DNMT3
b
cooperate to si
l
ence genes in
h
uman cancer ce
ll
s. Natur e 416:552–55
6
R
obert MF, Morin S, Beaulieu N, Gauthier F, Chute IC, Barsalou A, MacLeod AR
(
2003
)
DNMT1 is required to maintain CpG meth
y
lation and aberrant
g
ene silencin
g
in
human cancer cells. Nat

G
enet 33:61–6
5
R
o
b
ertson KD, Ait-Si-A
l
iS,Yo
k
oc
h
iT,Wa
d
e PA, Jones PL, Wo
lff
e AP (2000) DNMT
1
f
orms a comp
l
ex wit
h
R
b
, E2F1 an
d
HDAC1 an
d
represses transcription

f
ro
m
E2F-respons
i
ve promoters. Nat Genet 25:338–34
2
R
ountree MR, Bachman KE, Ba
y
lin SB (2000) DNMT1 binds HDAC2 and a new co-
re
p
ressor, DMAP1, to form a com
p
lex at re
p
lication foci. Nat Genet 25:269–27
7
Sh
iro
h
zu H, Ku
b
otaT,KumazawaA,Sa
d
oT,C
h
ijiwa T, Inaga
k

i K, Sueta
k
eI,TajimaS
,
Wa
k
ui K, Mi
k
i Y, Hayas
h
iM,Fu
k
us
h
ima Y, Sasa
k
i H (2002) T
h
ree nov e
l
DNMT3
B
mutations in Japanese patients wit
h
ICF syn
d
rome. Am J Me
d
Genet 112:31–3
7

S
ta
ll
er P, Peu
k
ert K, Kiermaier A, Seoane J, Lu
k
as J, Karsun
k
yH,MoroyT,Barte
k
J,
Massague J, Hane
l
F, Ei
l
ers M (2001) Repression o
f
p15INK4
b
expression
b
yMy
c
t
h
roug
h
association wit
h

Miz-1. Nat Ce
ll
Bio
l
3:392–399
S
tec I, Na
g
l SB, van Ommen GJ, den Dunnen JT (2000) The PWWP domain: a potential
protein-protein interaction domain in nuclear proteins influencin
g
differentia
-
tion? FEB
S
Lett 473:1–
5
S
ueta
k
eI,S
h
inoza
k
i F, Miyagawa J, Ta
k
es
h
ima H, Tajima S (2004) DNMT3L stimu
l

ates
t
h
e DNAmet
h
y
l
ationactivity o
f
Dnmt3a an
d
Dnmt3
b
t
h
roug
h
a
d
irect interaction.
JBio
l
C
h
em 279:27816–27823
Establishment and Maintenance of DNA Meth
y
lation Pa tterns in Mammals 20
1
Tamaru H, Selker EU (2001) A histone H3 meth

y
ltransferase contro lsDNA meth
y
lation
i
n Neurospora crassa. Nature 414:277–28
3
Tan g LY, Re
dd
y MN, Ras
h
eva V, Lee TL, Lin MJ, Hung MS, S
h
en CK (2003) T
h
e eu-
k
aryotic DNMT2 genes enco
d
e a new c
l
ass o
f
cytosine-5 DNA met
h
y
l
trans
f
erases.

JBio
l
C
h
em 278:33613–3361
6
Tatematsu KI, Yamazaki T, Ishika wa F (2000) MBD2-MBD3 co m
p
lex binds to hemi
-
meth
y
lated DNA and forms a complex containin
g
DNMT1 at the replication foc
i
in late S
p
hase. Genes Cells 5:677–688
Trouc
h
eD,LeC
h
a
l
ony C, Muc
h
ar
d
t C, Yaniv M, Kouzari

d
es T (1997) RB an
dhb
rm
cooperate to repress t
h
eactivation
f
unctions o
f
E2F1. Pr oc Nat
l
Aca
d
SciUSA
94:112
6
8–11273
Tweedie S, N
g
HH, Barlow AL, Turner BM, Hendrich B, Bird A (1999) Vesti
g
esofaDNA
meth
y
lation s
y
stem in Drosophila melano
g
aster? Nat Genet 23:389–390

Van d en W
y
n
g
aert I, Spren
g
el J, Kass SU, Lu
y
ten WH (1998) Clonin
g
and anal
y
sis of
a novel human putative DNA meth
y
ltransferase. FEBS Lett 426:283–289
Wade PA, Ge
g
onne A, Jones PL, Ballestar E, Aubr
y
F, Wolffe AP (1999) Mi-2 comple
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
Wei n
b
erg RA (1995) T
h
e retino
bl
astoma prot ein an
d
ce
ll
cyc
l
econtro
l
.Ce

ll
81:323–330
Wilkinson CR, Bartlett R, Nurse P, Bird AP (1995) The fission
y
east
g
ene pmt1
+
encodes a DNA meth
y
ltransferase homolo
g
ue. Nucleic Acids Res 23:203–21
0
Wu S, Cetinka
y
aC,Munoz-AlonsoMJ,vonderLehrN,BahramF,Beu
g
er V, Eilers M,
Leon J, Larsson LG (2003) M
y
c represses differentiation-induced p21CIP1 ex
-
pression via Miz-1-dependent interaction with the p21 core promoter. Onco
g
ene
22:351–3
6
0
Xie 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
:87–95
Yoder JA, Bestor TH (1998) A candidate mammalian DNA meth
y
ltransferase relate
d

to pmt1p of fission
y
east. Hum Mol Genet 7:279–28
4
Yoder JA, Soman NS, Verdine GL, Bestor TH (1997) DNA (c
y
tosine-5)-meth
y
ltrans-
f
erases in mouse ce
ll
san
d
tissues. Stu
d
ies wit
h
amec
h
anism-
b
ase
d
pro
b
e. J Mo
l
Bio
l

270:385–39
5
00 Prelims ToHell 28_4_05.qxd 28/04/05 14:55 Page ii
This page intentionally left blank
CTMI
(
2006
)
301:203–22
5

c
Springer-Ver
l
ag Ber
l
in Hei
d
e
lb
erg 200
6
Molecular Enzymology of Mammalian DN
A
Methyltrans
f
erase
s
A
.Je

l
tsc
h(
✉
)
Sc
h
oo
l
o
f
Engineering an
d
Science, Internationa
l
University Bremen, Campus Ring 1
,
28759 Bremen, German
y
a.
j

1In
t
r
oduct
i
o
n

2
04
2Cata
l
ytic Mec
h
anism o
f
DNA-(Cytosine-C5)-MTases
.
20
6
2.1 Reaction Mec
h
anism o
f
DNA-(Cytosine-C5)-MTases 206
2.2 Base Flippin
g
208
3Tar
g
et Sequence Specificit
y
of Mammalian DNA MTase
s

21
0
3

.1 Specificit
y
of Dnmt1 for Hemimeth
y
latedDNA 210
3
.2 CG and Non-CG Meth
y
lation b
y
Dnmt3AandDnmt3B 211
3
.3 Flankin
g
SequencePreferenceofMammalianDNAMTases 211
3
.4 Speci
fi
city o
f
Dnmt2 21
2
4 Processi vity o
f
DN A Met
h
y
l
ation
b

y Mamma
l
ian DNA MTase
s
.

213
4.1 Processi vity o
f
Dnmt1 21
3
4.2 Processi vity o
f
Dnmt3A an
d
Dnmt3B 214
5
Control of DNA MTase Activit
y
in Mammalian S
y
stem
s

21
5
5.
1All
oste
ri

c
A
ct
iv
at
i
o
n
o
fDnm
t
1

21
6
5.2 Stimula tion of Dnmt3A and Dnmt3B b
y
Dnmt3L 21
7
6FuturePers
p
ectives

21
8
Re
f
erence
s

219
A
b
stract
D
NA met
h
y
l
ation is an essentia
l
mo
d
i
fi
cation o
f
DNA in mamma
l
st
h
at is
invo
l
ve
d
in gene regu
l
ation,
d

eve
l
opment, genome
d
e
f
ence an
dd
isease. In mamma
ls
3f
ami
l
ies o
f
DNA met
h
y
l
trans
f
erases (MTases) comprising (so
f
ar) 4 mem
b
ers
h
av
e
b

een
f
oun
d
: Dnmt1, Dnmt2, Dnmt3A an
d
Dnmt3B. In a
dd
ition, Dnmt3L
h
as
b
ee
n
i
d
enti
fi
e
d
as a stimu
l
ator o
f
t
h
e Dnmt3A an
d
Dnmt3B enzymes. In t
h

is review t
he
enz
y
molo
gy
of the mammalian DNA MTases is described, startin
g
with a depiction
of the catal
y
tic mechanism that involves co valent catal
y
sis and base flippin
g
.Sub-
sequentl
y
, important mechanistic features of the mammalian enz
y
me are discussed
includin
g
the specificit
y
of Dnmt1 for hemimeth
y
lated tar
g
et sites, the tar

g
et sequenc
e
specificit
y
of Dnmt3A, Dnmt3B and Dnmt2 and the flankin
g
sequence preferences o
f
Dnmt3A and Dnmt3B. In addition, the processivit
y
of the meth
y
lation reaction b
y
20
4
A
. Jeltsc
h
D
nmt1, Dnmt3A an
d
Dnmt3B is reviewe
d
. Fina
ll
y, t
h
econtro

l
o
f
t
h
ecata
l
ytic activity
of
mamma
l
ian MTases is
d
escri
b
e
d
t
h
atinc
l
u
d
es t
h
eregu
l
ation o
f
t

h
eactivityo
f
Dnmt1
b
y its N-termina
ld
omain an
d
t
h
e interaction o
f
Dnmt3A an
d
Dnmt3B wit
h
Dnmt3L.
Th
ea
ll
osteric activation o
f
Dnmt1
f
or met
h
y
l
ation at unmo

d
i
fi
e
d
sites is
d
escri
b
e
d.
Wh
erever possi
bl
e, corre
l
ations
b
etween t
h
e
b
ioc
h
emica
l
properties o
f
t
h

e enzyme
s
a
n
d
t
h
eir p
h
ysio
l
ogica
lf
unctions in t
h
ece
ll
are in
d
icate
d
.
1
I
ntr
odu
ct
ion
The first mammalian DNA meth
y

ltransferase (MTase) activit
y
was discov-
ere
db
yRazin’s group in t
h
e ear
l
y 1980s (Gruen
b
aum et a
l
. 1982). T
h
e enzyme
responsi
bl
e
f
or t
h
is activit
y
is ca
ll
e
d
Dnmt1 to
d

a
y
[t
h
e name
d
erives
f
ro
m
DN
A
m
et
h
y
l
t
r
ans
f
erase; t
h
e systematic nomenc
l
ature o
f
DNA MTases is
d
e-

s
cri
b
e
d
in Ro
b
erts et a
l
. (2003)]. T
h
e murine Dnmt1 enz
y
me was t
h
e
fi
rst
mammalian DNA MTase to be cloned and expressed recombinantl
y
(Bestor
et a
l
. 1988; Pra
dh
an et a
l
. 1997). Durin
g
t

h
e
l
ast
d
eca
d
e, t
h
ree more mem
b
er
s
of the mammalian Dnmt enz
y
me famil
y
have been discovered and clone
d
(
Fig. 1; reviews: C
h
en an
d
Li 2004; Hermann et a
l
. 2004a). A
ll
t
h

ese enzyme
s
contain a
d
omain o
f
approxima te
ly
400–500 amino aci
d
resi
d
ues, w
h
ic
h
i
s
c
h
aracterise
db
yt
h
e presence o
f
10 conserve
d
amino aci
d

moti
f
s, s
h
are
d
b
etween pro
k
ar
y
otic an
d
eu
k
ar
y
otic DNA-(c
y
tosine-C5)-MTases (reviews
:
Chen
g
1995; Jeltsch 2002). The catal
y
tic centre and coenz
y
me bindin
g
site o

f
M
Tases resi
d
ewit
h
in t
h
is
d
omain. In a
dd
ition, t
h
e Dnmt1 an
d
t
h
e Dnmt3 en-
Fi
g. 1
D
omain organisation o
f
t
h
e mamma
l
ian Dnmts. T
h

e mamma
l
ian met
h
y
l
trans-
f
erases are divided into an N-terminal
p
art and a C-terminal
p
art. The C-termina
l
p
art shows stron
g
amino acid sequence homolo
gy
to prokar
y
otic DNA-(c
y
tosine-C5)-
M
Tase and contains 10 conserved catal
y
tic amino acid motifs (indicated b
y
R

oman
numerals
) characteristic for this enz
y
me famil
y
Molecular Enz
y
molo
gy
of M ammalian DNA Meth
y
ltransferase
s
205
zymes
h
ar
b
our
l
arge N-termina
l
regu
l
atory parts (reviews: C
h
en an
d
Li 2004

;
Hermann et al. 2004a). The N-terminal re
g
ulator
y
domain of Dnmt1 contains
d
ifferent motifs and subdomains which interact with man
y
other protein
s
(C
h
uang et a
l
. 1997; Fu
k
seta
l
. 2003; Liu an
d
Fis
h
er 2004; Margot et a
l
. 2003
;
P
radhan and Kim 2002; Robertson et al. 2000; Rountree et al. 2000). One
examp

l
eo
f
t
h
ese int eracting proteins is t
h
epro
l
i
f
erating ce
ll
nuc
l
ear antigen
(PCNA) known as processivit
y
factor for the DNA pol
y
merase
s
ε
/
δ
(
C
h
uan
g

et a
l
. 1997; Maga an
d
Hu
b
sc
h
er 2003). It seems t
h
at t
h
e N-terminus is
f
ormin
g
a platform for bindin
g
of proteins involved in chromatin condensation,
g
ene
re
g
ulation and DNA replication. In addition, Dnmt1 has a role in mismatch
repair o
f
mamma
l
ian ce
ll

s(Kimeta
l
. 2004; Wang an
d
James S
h
en 2004)
.
D
nmt1 has a stron
g
preference for meth
y
lation of hemimeth
y
lated C
G
sites
(
Fatemi et a
l
. 2001; Gruen
b
aum et a
l
. 1982; Hermann et a
l
. 2004
b)
,w

h
ic
h
implicates it as havin
g
a function in maintenance of the meth
y
lation pattern o
f
th
eDNAa
f
ter rep
l
ication. Dnmt1
k
noc
k
-out mice
d
ie
d
uring em
b
ryogenesis
;
embr
y
os show almost complete loss of DNA meth
y

lation (Li et al. 1992).
Interestin
g
l
y
, the catal
y
tic domain of Dnmt1 is inactive in the absence of th
e
N-terminal part (Fatemi et al. 2001), which implies an important re
g
ulator
y
f
unction of the N-terminal domain on the enz
y
me.
D
nmt2 is t
h
e sma
ll
est enzyme among t
h
eeu
k
aryotic MTases an
d
it com-
p

rises onl
y
the catal
y
tic domain (Fi
g
. 1). It has a ver
y
slow turnover rat
e
(Hermann et a
l
. 2003; Kunert et a
l
. 2003; Liu et a
l
. 2003; Tang et a
l
. 2003). T
he
p
rotein is conserved in man
y
eukar
y
otic species (also some that onl
y
have low
o
r even undetectable levels of DNA methylation like Drosophila melanogaste

r
or
S
chizosaccharom
y
ces pombe
)
. The biolo
g
ical function of Dnmt2 is not
k
nown, althou
g
h it has been associated to lon
g
evit
y
in
D
. melano
g
aster
(
Li
n
e
ta
l
. 2004
)

.
The mammalian Dnmt3 enz
y
me famil
y
consists of three different protein s,
Dnmt3A, Dnmt3B an
d
Dnmt3L (Fig . 1). T
h
eregu
l
atory N-t ermina
ld
omain o
f
Dnmt3A and Dnmt3B is not essential for catal
y
sis (Gowher and Jel tsch 2002;
R
either et al. 2003). Bothenzymes contain anATRX-like Cys-rich domain (als
o
c
alled PHD domain
)
and a PWWP domain, which are involved in interactions
w
ith other proteins and targeting to heterochromatin (Aapola et al. 2002
;
B

ac
h
man et a
l
. 2001; C
h
en an
d
Li 2004; Fu
k
seta
l
. 2003; Ge et a
l
. 2004). Despit
e
significant amino acid sequence and biochemical similarities, Dnmt3A and
Dnmt3B
h
ave
d
istinct
b
io
l
ogica
l
ro
l
es. Dnmt3B is responsi

bl
e
f
or met
h
y
l
ation
o
f pericentromeric satellite re
g
ions(Hansenetal.1999;Okano et al. 1999; Xu et
a
l
. 1999
)
. Dnmt3
B
−/
−
k
noc
k
-out mice
d
ie
d
uring t
h
e

l
ate em
b
ryonic stage an
d
th
eem
b
r
y
os
l
ac
k
met
hyl
ation in pericentromeric repeat re
g
ions (O
k
ano e
t
al. 1999). Loss of Dnmt3B activit
y
in human leads to ICF (immunodeficienc
y,
centromere insta
b
i
l

ity,
f
acia
l
anoma
l
ies) syn
d
rome, a genetic
d
isor
d
er t
h
at
206
A
. Jeltsc
h
is accompanie
db
y
l
ow met
h
y
l
ation in t
h
e pericentromeric sate

ll
ite regions
of chromosomes 1, 9 and 16
(
Ehrlich 2003
)
. Dnmt3A knock-out mice sho
w
develo
p
mental abnormalities and die a few weeks after birth (Okano et al.
1
999). T
h
is enzyme
h
as
b
een associate
d
wit
h
t
h
e met
h
y
l
ation o
f

sing
l
ecopy
g
enes and retrotransposons (Bourc’his and Bestor 2004; Bourc’his et al. 2001;
H
ata et a
l
. 2002) an
d
it is require
df
or t
h
e esta
bl
is
h
ment o
f
t
h
e genomic imprin
t
durin
gg
erm cell development (Kaneda et al. 2004). The N-terminal part o
f
D
nmt3L is s

h
orter t
h
an t
h
ose o
f
Dnmt3A an
d
Dnmt3B an
d
on
l
y con tains
the PHD domain. The C-terminal part of this protein is truncated and all it
s
“
catal
y
tic” motifs are crippled, indicatin
g
it cannot be an active DNA MTase
.
D
nmt3L acts as a stimu
l
ator o
f
t
h

ecata
l
ytic activity o
f
Dnmt3A an
d
Dnmt3B
activit
y
(Chedin et al. 2002; Gowher et al. 2005; Suetake et al. 2004).
In t
h
e
f
o
ll
owing sections, t
h
e enzymo
l
ogy o
f
t
h
e mamma
l
ian DNA MTase
s
will be reviewed. Startin
g

with a description of the catal
y
tic mechanism, som
e
important mec
h
anistic
f
eatures
l
i
k
et
h
e
d
egree o
f
speci
fi
city
f
or t
h
e target
b
as
e
and preference fo r flankin
g

sequences, the processivit
y
of DNA meth
y
lation
and the mechanism ofcontrol ofenz
y
me activit
y
will be discussed. It is writte
n
under the presumption that a detailed knowled
g
e of the enz
y
mes’ pr operties
is an essential prerequisite for the understandin
g
of their cellular roles
.
2
Catal
y
t
i
c Mechan
i
sm of DNA-(C
y
tos

i
ne-C5)-MTase
s
A
ll
DNA MTases use t
h
e coenz
y
m
e
S-
a
d
enos
yl-
l
-
met
h
ionine
(
A
d
oMet
)
as t
he
s
ource

f
or t
h
e met
h
y
l
group
b
eing trans
f
erre
d
to t
h
eDNA
b
ases. T
h
e met
h
y
l
g
roup of A doMet is bound to a sulphonium centre, which activates it towards
nucleophilic attack. The AdoM et bindin
g
site is remarkabl
y
conservedinal

l
D
NA (an
d
a
l
so non-DNA) MTases. It is create
db
y resi
d
ues
f
rom t
h
emoti
fs
I–III and X, which form conserved contacts to almost ever
y
h
y
dro
g
en bon
d
d
onor an
d
acceptor o
f
t

h
eA
d
oMet an
d
,ina
dd
ition, severa
lh
y
d
rop
h
o
b
ic
interactions to the cofactor. The roles of man
y
of these residues have bee
n
con
fi
rme
db
y mutagenesisexperiments in pro
k
aryotic MTases (review: Je
l
tsc
h

2002
)
.
2
.1
Reaction Mechanism of DNA-(Cytosine-C5)-MTase
s
The reaction mechanism of c
y
tosine-C5 meth
y
lation was uncovered for th
e
prokar
y
otic DNA-(c
y
tosine-C5)-MTase M.HhaI (Fi
g
. 2; Wu and Santi 1985;
Wu an
d
Santi 1987). A
k
ey
f
eature o
f
t
h

ecata
l
ytic process is a nuc
l
eop
h
i
l
i
c
Molecular Enz
y
molo
gy
of M ammalian DNA Meth
y
ltransferase
s
207
F
i
g.
2
S
tructure o
f
t
h
epro
k

aryotic M.H
h
aI DNA MTase. T
he
l
e
f
tpar
t
sh
ows t
h
e protein
in sc
h
ematic view, in t
he
rig
h
tpar
t
on
l
yt
h
eDNAiss
h
own to i
ll
ustrate t

h
erotationo
f
t
h
e target
b
ase out o
f
t
h
eDNA
h
e
l
i
x
attack of the enz
y
me on the carbon-6 of the tar
g
et c
y
tosine. This attack
is per
f
orme
db
yt
h

et
h
io
l
group o
f
t
h
e cysteine resi
d
ue t
h
at is part o
f
t
h
e
conserve
d
PCQ moti
f
in t
h
e active site o
f
c
y
tosine-C5-MTases (moti
f
IV)

.
T
h
is reaction is cata
l
yse
db
yt
h
eprotonationo
f
t
h
e cytosine N3 positio
n
carrie
d
out
by
t
h
e
gl
utamic aci
d
o
f
t
h
e amino aci

d
moti
f
ENV (moti
f
VI).
Theformationofthecovalentbondactivatesthec
y
tosine C5 atom towards
nuc
l
eop
h
i
l
ic attac
k
on t
h
e met
hyl g
roup
l
ea
d
in
g
to t
h
ea

dd
ition o
f
t
h
e met
hyl
g
roup to carbon-5. The reaction c
y
cle is closed b
y
the elimination of the 5-
p
osition proton an
d
t
h
et
h
io
l
moiety, w
h
ic
h
reso
l
ves t
h

ecova
l
ent interme
d
iat
e
and re-establishes aromaticit
y
(review: Jeltsch 2002)
.
T
h
is
d
escription o
f
t
h
ecata
l
ytic mec
h
anism o
f
DNA-(cytosine C5)-MTases
by
acom
b
ination o
f

cova
l
ent cata
ly
sis an
d
aci
db
ase cata
ly
sis is supporte
d
b
y
a lar
g
ebod
y
of experimental evidence: The covalent reaction intermediat
e
b
etween met
hyl
ate
d
DNA an
d
t
h
e active site c

y
steine
h
as
b
een o
b
serve
d
i
n
all structures of DNA-(c
y
tosine-C5)-MTase in complex with DNA known s
o
f
ar
(
K
l
imasaus
k
as et a
l
. 1994; Reinisc
h
et a
l
. 1995
)

.Ina
dd
ition, t
h
ecova
l
en
t
intermediate has been detected biochemicall
y
with several DNA MTases in
-
c
l
u
d
ing Dnmt1 an
d
Dnmt3A (C
h
en et a
l
. 1991; Hanc
k
et a
l
. 1993; Osterman e
t
al. 1988; Reither et al. 2003; Santi et al. 1984; W
y

sz
y
nski et al. 1993; Yoder et al
.
1997) an
d
cova
l
ent comp
l
ex
f
ormation
h
as
b
een s
h
own to invo
l
ve t
h
e cystein
e
residue in the PCQ mo tif
(
Chen et al. 1991; Everett et al. 1990; Hanck et al.
1993; Reither et al. 2003). In addition, the importance of the c
y
steine residue

in moti
f
IV
f
or cata
l
ysis
b
ypro
k
aryotic MTases
h
as
b
een
d
emonstrate
db
y
208
A
. Jeltsc
h
s
ite-
d
irecte
d
mutagenesis (Hur
d

et a
l
. 1999; Wyszyns
k
ieta
l
. 1992, 1993). T
he
formation of a stable covalent intermediate comprisin
g
the enz
y
me and th
e
tar
g
et base is the basis of the efficient inhibition of DNA MTases b
y
c
y
tidine
ana
l
ogues incorporate
d
into DNA, w
h
ic
h
current

l
yis
b
eing investigate
d
wit
h
res
p
ect to its thera
p
eutic
p
otential (review: Gowher and Jeltsch 2004).
Surprising
l
y, in t
h
e case o
f
t
h
e Dnmt3A cata
l
ytic
d
omain, t
h
eg
l

utamic aci
d
residue in motif VI has been shown to be ver
y
important for activit
y
, but the
remova
l
o
f
t
h
e active site cysteine resi
d
ue
d
i
d
not resu
l
tinacomp
l
ete
l
oss o
f
catal
y
tic activit

y
(Reither et al. 2003). This findin
g
su
gg
ests that, in additio
n
to covalent catal
y
sis, other mechanisms of enz
y
me catal
y
sis are operativ
e
in DNA MTases (at
l
east in t
h
e case o
f
Dnmt3A) suc
h
as positioning o
f
t
h
e
tar
g

et base and the cofactor with respect to each other and stabilisation of
t
h
e transition state o
f
met
h
y
l
group trans
f
er. I n t
h
is context, it is interestin
g
to note that Dnmt3A purified from
Escherichia coli
but a
l
so
fr
o
min
sect ce
ll
s
sh
ows on
l
yre

l
ative
l
y
l
ow turnover rates (Ao
k
ieta
l
. 2001; Gow
h
er an
d
Je
l
tsc
h
2001; Okano et al. 1998
)
. This indicates that the active site of Dnmt3A is not
in an ideal conformation and the c
y
steine residue is not ideall
y
positioned t
o
perform a nucleophilic attack on the C6 position. It mi
g
ht be possible that
a

covalent modification of the enz
y
me or an interaction with another protei
n
cou
ld
in
d
uceacon
f
ormationa
l
c
h
ange o
f
t
h
ecata
l
ytic site t
h
at activates t
he
enz
y
me and switches the catal
y
tic mechanismto the covalent catal
y

sis schem
e
(
Reit
h
er et a
l
. 2003). T
h
e mamma
l
ian Dnmt1 enzyme mig
h
t
b
eaprece
d
ent
for this kind of activation, because althou
g
h the full-len
g
th enz
y
me is hi
g
hl
y
a
ctiv e, its ca tal

y
ticdomainisnotactiveinanisolatedform,whichimpliestha
t
a
n interaction of the catal
y
tic domain with the rest of the protein is essentia
l
for the catal
y
tic domain to adopt a catal
y
ticall
y
competent conformation
.
2.
2
Base Fl
i
pp
i
n
g
The first X-ra
y
structure of a DNA-(c
y
tosine-C5)-MTase incomplex with DN
A

was determined with M.HhaI (Klimasauskas et al. 1994; Fi
g
. 3). It demon-
s
trated that DNA MTases completel
y
rotate their tar
g
et base out of the DNA
h
e
l
ix prior to its met
h
y
l
ation, a process ca
ll
e
db
ase
fl
ipping. A
f
ter
b
ase
fl
ip-
pin

g
the tar
g
et c
y
tosine is no lon
g
er buried in the double helix of the DNA
b
ut is turne
d
a
b
out its
fl
an
k
ing sugar-p
h
osp
h
ate
b
on
d
ssuc
h
t
h
at it project

s
out into the catal
y
tic pocket of the enz
y
me. The base pairin
g
h
y
dro
g
en bonds
a
re
b
ro
k
en an
d
t
h
e stac
k
ing interactions wit
h
t
h
ea
d
jacent

b
ase pairs ar
e
l
ost durin
g
this process. Base flippin
g
has been observed in all MTase-DNA
com
p
lex structures known so far (Goedecke et al. 2001; Klimasauskas et al.
1
994; Reinisc
h
et a
l
. 1995) an
d
a
l
so in many ot
h
er enzymes interacting wit
h
Molecular Enz
y
molo
gy
of M ammalian DNA Meth

y
ltransferase
s
2
0
9
Fi
g
.
3
Chemistr
y
of the DNA meth
y
lation reactio
n
DN A,
f
or examp
l
e many DNA repair enzymes (reviews: C
h
eng an
d
Ro
b
ert
s
2
001; Ro

b
erts an
d
C
h
en
g
1998). It
b
rin
g
st
h
e tar
g
et
b
ase into c
l
ose contact
t
o the enz
y
me, allowin
g
for the intricate chemical reactions to occur and for
accurate reco
g
nition o
f

t
h
e
fl
ippe
db
ase, an important requirement
f
or t
h
e
f
unction of DNA repair
g
l
y
cos
y
lases. Also , it is a prerequisite for the catal
y
ti
c
m
ec
h
anism as
d
escri
b
e

d
a
b
ove,
b
ecause it ma
k
es t
h
e C5, C6 an
d
N3
p
ositions
o
f the c
y
tosine accessible to the enz
y
me.
T
h
e structure o
f
M.H
h
aI is typica
lf
or a
ll

enzymes o
f
t
h
e DNA-(cytosine
-
C
5)-MTase
f
ami
ly
(reviews: C
h
en
g
1995; C
h
en
g
an
d
Ro
b
erts 2001). It com-
p
rises two
d
omains. T
h
e

l
arger, cata
l
ytic
d
omain is conserve
d
among a
ll
en-
z
y
mes o
f
t
h
is t
y
pe. It consists o
f
acentra
l
, para
ll
e
l
, 6-stran
d
e
d

β
-s
h
eet
fl
an
k
e
d
b
y
α
-helices. The domain can be divided int o two subdomains, one formin
g
th
e
b
in
d
ing poc
k
et
f
or t
h
e
fl
ippe
d
target

b
ase, t
h
eot
h
er
f
or t
h
eA
d
oM et co
f
ac
-
2
1
0
A
. Jeltsc
h
tor. T
h
e structures o
fb
ot
h
su
bd
omains are simi

l
ar, an
d
t
h
ecata
l
ytic
d
omain
most likel
y
arose b
yg
ene duplication (Malo ne et al. 1995). The smal ler domai
n
is involved in the reco
g
nition of the tar
g
et sequence and structurall
y
divers
e
(
review: Je
l
tsc
h
2002). T

h
eon
l
y structure o
f
a mamma
l
ian MTase cata
l
ytic
domain solved so far is that of Dnmt2 (Don
g
et al. 2001). The protein is folded
very simi
l
ar
l
y to M.H
h
aI; un
f
ortunate
l
y, t
h
e reason(s) w
h
yt
h
e Dnmt2 enzyme

h
as onl
y
aver
y
low catal
y
tic activit
y
cannot be deduced from its structure
.
3
Target Sequence Spec
i
f
i
c
i
t
y
of Mammal
i
an DNA MTase
s
All mammalian DNA MTases modif
y
DNA at CG sites. However, the de
g
ree o
f

s
peci
fi
city
f
or t
h
e target sequence an
d
t
h
epre
f
erence
f
or
d
i
ff
erent met
h
y
l
ation
s
tates o
f
t
h
e tar

g
et site varies consi
d
era
bly
amon
g
t
h
e
d
i
ff
erent enz
y
mes.
3
.
1
S
pecificit
y
of Dnmt1 for Hemimeth
y
lated DNA
In 1982, Razin’s group iso
l
ate
d
DNA MTase activity

f
rom mamma
l
ian ce
ll
st
h
at
d
isp
l
a
y
e
d
aver
yh
i
gh
pre
f
erence
f
or
h
emimet
hyl
ate
d
CG sites (Gruen

b
au
m
et al. 1982). Later this enz
y
me was identified as Dnmt1, but, usin
g
oli
g
onu-
c
l
eoti
d
esu
b
strates,
d
i
ff
erent
f
actors
f
or t
h
epre
f
erence o
fh

emimet
h
y
l
ate
d
D
NA over unmeth
y
lated were found which ran
g
e from 2- to 50-fold (Fatem
i
et a
l
. 2001; F
l
ynneta
l
. 1996; Pra
dh
an et a
l
. 1999; To
ll
e
f
s
b
o

l
an
d
Hutc
h
ison 1995
,
1
997). T
h
ese
d
i
ff
erences cou
ld b
e
d
ue to
d
eviations in t
h
e experimenta
l
setu
p
(
a
ll
osteric activation, see Sect. 5.1),

d
i
ff
erent su
b
strates,
d
i
ff
erent sources o
f
t
h
e p roteins an
dd
i
ff
erent
d
e
g
rees o
f
purit
y
. For examp
l
e, Bestor reporte
d
i

n
the earl
y
1990s that treatment of Dnmt1 with proteases leads to the loss of pref-
erence
f
or
h
emimet
hyl
ate
d
tar
g
et sites (Bestor 1992). Simi
l
ar
ly
,weo
b
serve
d
that the preference for hemimeth
y
lated DNA decreased from its ori
g
ina
l
l
eve

l
o
f
a
b
out 50-
f
o
ld d
uring pro
l
onge
d
storage o
f
t
h
e enzyme (Fatemi et a
l.
2001). In t
h
e context o
fl
on
g
er
h
emimet
hyl
ate

d
DNA, a 24-
f
o
ld
pre
f
erence
f
or
a
h
emimet
h
y
l
ate
d
target site
h
as
b
een
d
etecte
d
(Hermann et a
l
. 2004
b

). Invivo
,
t
h
is propert
y
is ver
y
important, as it ena
bl
es t
h
e enz
y
me to cop
y
t
h
e existin
g
meth
y
lation pattern of the DNA after DNA replication and, therefore, to wor
k
as a maintenance MTase. T
h
e
h
i
gh

speci
fi
cit
y
o
f
Dnmt1
f
or
h
emimet
hyl
ate
d
tar
g
et sitesisa fascinatin
g
exampleofmolecular r ec o
g
nition, because thep res-
ence o
f
a sing
l
e met
h
y
l
group switc

h
es on t
h
e enzyme’s activity at
h
emimet
h
y
-
l
ated CG sites. The detaile d mechanism of this process is not
y
et known
.
Molecular Enz
y
molo
gy
of M ammalian DNA Meth
y
ltransferase
s
2
1
1
3.
2
CG and Non-CG Methylat
i
on by Dnmt3A and Dnmt3

B
B
oth Dnmt3A and Dnmt3B do not differentiate between unmeth
y
lated and
hemimeth
y
lated substrates, and both are involved in de novo DNA meth
y
la-
t
ion in vivo (Gow
h
er an
d
Je
l
tsc
h
2001; O
k
ano et a
l
. 1998, 1999). Interestin
gly
,
Dnmt3A and Dnmt3B also meth
y
late c
y

tosine residues in a non-CG contex
t
in vitro (Ao
k
ieta
l
. 2001; Gow
h
er an
d
Je
l
tsc
h
2001; Hsie
h
1999; Ramsa
h
oye e
t
al. 2000). Dependin
g
on the substrat e and assa
y
s
y
stem, the activit
y
at non-CG
sites varies

b
etween 0.5% an
d
10% o
f
t
h
eactivityo
b
serve
d
at CG sites. In gen
-
eral, CA sites wer e found the second-best substrate for Dnmt3A and Dnmt3B.
Met
h
y
l
ation o
f
non-CG sites
b
y Dnmt3A
h
as
b
een
d
etecte
d

a
l
so in mouse
DN A (Do
dg
eeta
l
. 2002). However, since Dnmt1 cannot maintain t
h
is as
y
m
-
m
etric meth
y
lation, the biolo
g
ical function of this activit
y
is not known. On
e
cou
ld
specu
l
ate t
h
at non-CG met
h

y
l
ation is important to ensure a rapi
d
onse
t
o
fastron
g
repression of
g
ene expression durin
g
earl
y
embr
y
o
g
enesis. After
some time, w
h
en a
dd
itiona
l
epigenetic mec
h
anisms
l

i
k
e
h
istone mo
d
i
fi
catio
n
and chromatin condensation have become effective, the non-CG meth
y
latio
n
m
ig
h
tno
l
onger
b
e require
d
.
3.3
Flankin
g
Sequence Preference of Mammalian DNA MTases
A
n

ot
h
e
rf
acet
in
t
h
e
DN A in
te
r
act
i
o
n
o
fm
a
mm
a
li
a
nDNAMT
ases
i
st
h
e
i

r
fl
ankin
g
sequence preferences. Since it contains onl
y
two bases, the reco
g
ni
-
t
ion sequence of these enz
y
mes is much shorter than t
y
pical DNA interaction
sites of prot eins of that size, which are in the ran
g
e of 8 to 14 base pairs
.
T
h
ere
f
ore, it is
l
i
k
e
l

yt
h
at interactions
b
etween t
h
e protein an
d
t
h
eDNA
also occur outside of the central CG site, which could lead to
p
references o
f
m
et
h
y
l
ation o
f
CG sites wit
h
in a certain sequence context. Suc
hd
i
ff
erences
are usuall

y
called “flankin
g
sequence preference” and the
y
are conceptuall
y
d
istinct from the “sequence specificit
y
”, because a chan
g
e in the flanks wil
l
o
nl
y
modif
y
the rate of meth
y
lation, while a chan
g
e in the central tar
g
et site
w
ill abolish meth
y
lation. The flankin

g
sequence preferences of Dnmt3A an
d
Dnmt3B
h
ave
b
een stu
d
ie
d
in
d
etai
l
. Dnmt3A ex
h
i
b
its strong stran
d
pre
f
er-
ence for CG sites flanked by pyrimidines and a loose con sensus sequence o
f
Y
NCGY (Lin et a
l
. 2002). Later, t

h
e consensuses sequence cou
ld b
ere
fi
ne
d
and extended also to Dnmt3B, showin
g
that both enz
y
mes p refer meth
y
lation
of
CG sites in a RCGY context an
dd
is
f
avour YCGR sites
(
Han
d
aan
dJ
e
l
tsc
h
2

005). Interestin
g
l
y
, the rates of meth
y
lation of substrates differin
g
in 4 base
p
airs on each site of the central CG site varied by more than more than 500-
f
o
ld
. Comparing t
h
ese num
b
ers wit
h
t
h
e actua
l
pre
f
erence
f
or CG over C
A

2
12
A
. Jeltsc
h
in a given sequence context, w
h
ic
h
is approximate
l
y 10- to 100-
f
o
ld
,one
h
a
s
to conclude that the concept of flankin
g
sequence and c en tral site is not full
y
applicable to Dnmt3A and Dnmt3B, because chan
g
es in the flankin
g
sequenc
e
in

fl
uence t
h
e reaction rat e to a simi
l
ar
d
egree as a c
h
ange o
f
t
h
ecentra
l
CG
to CA. The flankin
g
sequence preferences of Dnmt1 for the meth
y
lation at
unmet
h
y
l
ate
d
CG sites
h
ave

b
een stu
d
ie
d
as we
ll
,
d
emons trating t
h
e enzym
e
s
hows a clear preference for meth
y
lation within a CCGG context (R. Go
y
al
an
d
A. Je
l
tsc
h
,in
p
re
p
aration).

Interestin
g
l
y
, a statistical anal
y
sis of human DNA meth
y
lation patterns
revealed that there is a clear correlation between the avera
g
e meth
y
lation
l
eve
l
o
f
CG sites an
d
t
h
eir
fl
an
k
ing sequence t
h
at c

l
ose
l
y
fi
ts to t
h
e
fl
an
k
ing
s
e
q
uence
p
references of Dnmt3A and Dnmt3B (Handa and Jeltsch 2005). Thi
s
fi
n
d
ing
d
emonstrates t
h
at t
h
e intrinsic pre
f

erences o
f
Dnmt3A an
d
Dnmt3B
for certain tar
g
et sites shaped the human epi
g
enome. However, the biolo
g
ical
im
pl
ications o
f
t
h
ese
q
uence
p
re
f
erences o
f
t
h
e Dnmt3A an
d

Dnmt3B
d
enov
o
M
Tases mi
g
ht extend even to immunolo
gy
. DNA containin
g
unmeth
y
lated C
G
dinucleotide sequences is imm uno
g
enic in mammals (Krie
g
2002; Rui et al.
2003). Inseveralreports it hasbeenshown thatDNAwith CG flankedb
y
purin
e
at t
h
e5

end and p
y

rimidine at the
3

end has a hi
g
her immuno
g
enic respons
e
w
h
en compare
d
to ot
h
er sequences (K
l
inman et a
l
. 1996; Krieg 2002). T
h
is
consensus sequence is i
d
entica
l
to t
h
e
h

i
gh
pre
f
erence consensus sequence
f
or Dnmt3A an
d
Dnmt3B. T
h
ere
f
ore, t
h
ose
fl
an
k
ing sequences t
h
at ren
d
e
r
h
i
gh
immuno
g
enicit

y
to unmet
hyl
ate
d
CG
d
inuc
l
eoti
d
e sites
b
e
l
on
g
to t
he
most
p
referred consensus se
q
uence for de novo DNA MTases and hence have
t
h
e
l
owest pro
b

a
b
i
l
it
y
to
b
e unmet
hyl
ate
d
in t
h
e
h
uman DNA. T
h
ere
by
,t
he
risk of an auto immune response
g
enerated from self-DNA is minimised. Thi
s
o
b
serva tion in
d

icates co-evo
l
ution o
fd
e novo DNA MTases an
d
t
h
e immune
sy
stem in context with CG dinucleotides and the flankin
g
sequences (Handa
an
d
Je
l
tsc
h
2005).
3
.4
S
pec
i
f
i
c
i
t

y
of Dnmt
2
T
h
esu
b
strate speci
fi
cit
y
o
f
t
h
e Dnmt2 enz
y
me is sti
ll
not
f
u
lly
un
d
erstoo
d.
The human enz
y
me has a preference for CG sites (Hermann et al. 2003

)
w
h
ereas D. me
l
anogaster Dnmt2 was
f
oun
d
to
p
re
f
er CT an
d
CA sites (Kunert
et al. 2003
)
. It is not clear whether or not these differences are due to the
amino aci
dd
i
ff
erences
b
etween
b
ot
h
enzymes, w

h
ic
h
are on
l
ymo
d
erate
.
H
owever, a
ll
t
h
ese stu
d
ies are
h
ampere
dby
t
h
e
l
ow met
hyl
ation activit
y
o
f

t
h
e
enz
y
mes, leadin
g
to an insufficient statistical samplin
g
. Therefore, additiona
l
experiments will be required to resolve this issue
.
Molecular Enz
y
molo
gy
of M ammalian DNA Meth
y
ltransferase
s
2
1
3
4
Process
i
v
i
ty of DNA Methylat

i
on by Mammal
i
an DNA MTase
s
Since DNA MTases are enz
y
mes t
h
at wor
k
on a
l
on
g
po
ly
meric su
b
strat
e
containin
g
several potential tar
g
et sites, the processivit
y
of the meth
y
lation

reaction is an important issue
f
or t
h
is c
l
ass o
f
enz
y
mes. Here, processivit
y
is defined as the preference of the enz
y
me to transfer more than one meth
yl
g
roup to one DNA mo
l
ecu
l
ewit
h
out re
l
ease o
f
t
h
eDNA.

4
.1
Processivity of Dnmt
1
Evidence for a
p
rocessi ve reaction mechanism of Dnmt1 dates back to 198
3
wh
en Bestor an
d
Ingram
d
emonstrate
d
t
h
at Dnmt1 met
h
y
l
ates
l
onger su
b-
strates faster than shorter ones (Bestor and In
g
ram 1983). Recentl
y
,lon

g
hemimeth
y
lated substrates were used to stud
y
the processivit
y
of Dnmt1 i
n
m
ore detail usin
g
aph
y
siolo
g
ical substrate. This stud
y
demonstrated tha
t
Dnmt1 modifies DNA in a hi
g
hl
y
processive reaction, and durin
g
the pro
-
cessive movement on t
h

e DNA it accurate
l
ycopiest
h
e exiting met
h
y
l
atio
n
p
attern (Hermann et al. 2004b). Such processive meth
y
lation of DNA implies
th
at Dnmt1 moves a
l
ong t
h
eDNAa
f
ter eac
h
turnover. T
h
e mec
h
anism o
f
t

his movement is not
y
et clear; it mi
g
ht invo lv e a slidin
g
and a hoppin
g
pro
-
cess. It a
l
so is not
k
nown i
f
Dnmt1 moves on t
h
eDNAwit
h
a
d
irectiona
l
p
reference
.
I t is temptin
g
to speculate that the abilit

y
of Dnmt1 to meth
y
late DN A i
n
a
p
rocessive reaction an
d
to interact wit
h
PCNA are co-a
d
a
p
tations t
h
atena
ble
t
he enz
y
me to bind to the replication fork in vivo and meth
y
late nascent DNA
imme
d
iate
l
ya

f
ter DNA rep
l
ication. However, its cata
l
ytic activity mig
h
tno
t
sufficetocopewiththehi
g
hdensit
y
of CG sites in heterochromatin. Therefore,
Dnmt1 mig
h
timpe
d
et
h
e progression o
f
t
h
erep
l
ication
f
or
k

i
f
it remaine
d
t
i
g
htl
y
attached to the replication fork durin
g
replication of heterochromatic
DNA. To avoid this
p
otential com
p
lication, one could su
pp
ose that Dnmt
1
is released from the replication fork durin
g
the heterochromatin replication
p
hase, and that the methylation of heterochromatic DNA is restored after
rep
l
ication
h
as ta

k
en p
l
ace. T
h
is mo
d
e
l
is supporte
db
yt
h
e
fi
n
d
ing t
h
at t
he
t
ime
g
ap between replication and meth
y
lation is lar
g
er for the heterochro-
m

atic t
h
an
f
or t
h
eeuc
h
romatic DNA
(
Gruen
b
aum et a
l
. 1983; Leon
h
ar
dt
et al. 1992; Lian
g
et al. 2002). Furthermore, it has been demonstrated tha
t
Dnmt3A and Dnmt3B also play a role in the preservation of meth ylation lev-
els at heterochr omatic DNA (Chen et al. 2003; Lian
g
et al. 2002; Rhee et al
.
2
002)
.

2
14
A
. Jeltsc
h
4
.2
Process
i
v
i
ty of Dnmt3A and Dnmt3
B
S
imilar experiments with Dnmt3A and Dnmt3B
y
ielded the interestin
g
result
that Dnmt3A modified DNA in a distributive reaction, but Dnmt3B was
p
ro-
cessive (Gow
h
er an
d
Je
l
tsc
h

2001, 2002). T
h
is was an unex
p
ecte
d
o
b
servation
because the catal
y
tic domains of Dnmt3A and Dnmt3B are about 84% identi
-
ca
l
in amino aci
d
sequence. However, among t
h
e 44 amino aci
d
resi
d
ues t
h
a
t
are not identical between human and murine Dnmt3A and Dnmt3B catal
y
tic

domains, 15 include char
g
ed residues. The exchan
g
es o bserved amon
g
thes
e
residues are hi
g
hl
y
biased such that, in the end, Dnmt3B carries 6 more pos-
itive char
g
es than Dnmt3A. Therefore, Dnmt3B has a much more positivel
y
c
h
arge
d
DN A
b
in
d
ing c
l
e
f
tt

h
an Dnmt3A, w
h
ic
h
cou
ld
exp
l
ain w
h
y Dnmt3B
meth
y
lates DNA in a processive reaction whereas Dnmt3A is distributive
(
Fig. 4; Gow
h
er an
d
Je
l
tsc
h
2002)
.
The difference in the kinetic mechanisms of the catal
y
tic domains of
D

nmt3A an
d
Dnmt3B cou
ld b
ere
l
ate
d
to t
h
e
d
istinct
b
io
l
ogica
lf
unctions
of these enz
y
mes in the cell, because satellite 2 repeats (one of the ma
j
or tar
-
g
ets of Dnmt3B) are exceptionall
y
rich in CG sites when compared with th
e

rest o
f
t
h
e genome (Gow
h
er an
d
Je
l
tsc
h
2001). Dnmt3B iswe
ll
suite
d
to mo
d
i
f
y
F
i
g
.4
M
odels of the catal
y
tic domains of Dnmt3A and Dnmt3B. The models wer
e

p
repared usin
g
M.H
ha
Iastem
p
late as described in (Gowher and Jeltsch 2002). Th
e
s
ur
f
ace o
f
t
h
e proteins was co
l
oure
d
accor
d
ing to t
h
ee
l
ectrostatic potentia
l
ca
l

cu
l
ate
d
u
sing Swiss PDB viewer version 3.7.
b
2. To i
ll
ustrate t
h
e
l
ocation o
f
t
h
eDNA
b
in
d
ing
cl
e
f
tint
h
e enzymes, t
h
e DNA as seen in t

h
eM
.
H
ha
I-DNA comp
l
ex is s
h
own i
n
orange
,
th
eA
d
oMet is s
h
own in
green
Molecular Enz
y
molo
gy
of M ammalian DNA Meth
y
ltransferase
s
2
1

5
th
ese regions,
b
ecause a
f
ter targeting to t
h
e DNA it can met
h
y
l
ate severa
l
cy-
t
osine residues in a processive reaction. The distributive reaction mechanis
m
o
f Dnmt3A mi
g
ht explain wh
y
it cannot replace Dnmt3B at satellite repeats i
n
v
ivo, a
l
t
h

oug
h
t
h
e Dnmt3A enzyme can met
h
y
l
ate t
h
ese regions
.
S
o if the processive mechanism has such obvious advanta
g
es, wh
y
di
d
Nature invent
d
istri
b
utive enzymes
l
i
k
e Dnmt3A? One a
d
vantage o

f
a
d
is
-
t
ributive enz
y
me could be that its activit
y
is under better control, because it
h
as to
b
e
d
irecte
d
to t
h
eDNA
f
or eac
h
sing
l
e met
h
y
l

ation event. T
h
ere
f
ore,
a distributive enz
y
me depends on a mechanism tar
g
etin
g
it to the sites o
f
action much more so than a processive enz
y
me, where one tar
g
etin
g
event
w
i
ll l
ea
d
to t
h
e trans
f
er o

f
severa
l
met
h
y
l
groups to t
h
eDNA.In
l
ine wit
h
t
hese considerations, Dnmt3A has been associated with the meth
y
lation of
sing
l
e-copy genes an
d
retrotransposons (Bourc’
h
is an
d
Bestor 2001, 2004;
Hata et al. 2002) and it is critical to the establishment of the
g
enomic imprint
d

uring germ ce
ll d
eve
l
opmen t (Kane
d
aeta
l
. 2004). T
h
ere
f
ore, Dnmt3A i
s
involved in the meth
y
lation of defined tar
g
et sites, whereas Dnmt3B (at least
as far as the meth
y
lation of heterochromatic repeats is concerned) catal
y
se
s
t
he complete meth
y
lation of lar
g

e DNA domains. One c ould envisa
g
e tha
t
Dnmt3A contacts a tar
g
etin
g
factor and thereb
y
keeps indirect contact (via
th
e targeting protein) to t
h
eDNA.T
h
is mec
h
anism wou
ld
a
ll
ow
f
or e
ffi
cient
m
eth
y

lation of the DNA at sites that are determined b
y
the specificit
y
of th
e
t
argeting c omp
l
ex
.
5
Contr ol of DNA MTase Act
i
v
i
ty
i
n Mammal
i
an System
s
T
h
emec
h
anism
by
w
h

ic
h
mamma
l
ian DNA MTases create a speci
fi
cDNA
m
eth
y
lation pattern that carries additional information is one of the most
f
ascinating questions regar
d
ing t
h
e
f
unction o
f
t
h
ese enzymes. A
l
t
h
oug
h
t
h

e
exact mechanism of pattern
g
eneration is no t certain, it clearl
y
depends on
th
econtro
l
o
f
t
h
e enzyme’s activity
b
y
d
i
ff
erent instances t
h
at inc
l
u
d
econtro
l
of g
ene transcription, cova
l

ent mo
d
i
fi
cation an
d
interaction wit
h
re
g
u
l
ator
y
p
roteins. T
h
e transcri
p
tiona
l
contro
l
o
f
mamma
l
ian Dnmts
h
as

b
een reviewe
d
recent
ly
(Pra
dh
an an
d
Esteve 2003
b
)an
d
is
b
e
y
on
d
t
h
escopeo
f
t
h
is review
,
w
hich focuses on enz
y

molo
gy
. Dnmt1 isolated from mammalian cell lines
h
as
b
een s
h
own to carr
y
some p
h
osp
h
or
yl g
roups (G
l
ic
k
man et a
l
. 1997)
.
However, the functional relevance of this modification is not
y
et known, an
d
it is not c
l

ear i
fp
ost-trans
l
ationa
l
mo
d
i
fi
cations occur wit
h
Dnmt3A, Dnmt3
B
o
r Dnmt2 as well. In the followin
g
para
g
raphs the interactions of MTases with
regu
l
atory proteins wi
ll b
e
d
iscusse
d
.
2

1
6
A
. Jeltsc
h
5.1
A
lloster
i
cAct
i
vat
i
on of Dnmt1
S
urprising
l
y, t
h
eiso
l
ate
d
cata
l
ytic
d
omain o
f
Dnmt1 is not cata

l
ytica
ll
yactive,
althou
g
h it con tains all the amino acid motifs characteristic for c
y
tosine-C
5
M
Tases (Fatemi et a
l
. 2001; Margot et a
l
. 2000; Zimmermann et a
l
. 1997). T
h
es
e
results demonstrate that the N-terminal part of Dnmt1 has an importan
t
ro
l
eincontro
ll
ing t
h
eactivityo

f
t
h
e protein, suc
h
t
h
at Dnmt1’s N-termina
l
part could be considered a “re
g
ulator
y
protein”. A similar observation wa
s
alread
y
made b
y
Bestor (1992) b
y
demonstratin
g
that a pro teol
y
tic cleava
ge
o
f
Dnmt1 just

b
etween t
h
ecata
l
ytic
d
omain an
d
t
h
e N-termina
ld
omain
l
ea
ds
to a stron
g
l
y
increased activit
y
of Dnmt1 to wards unmeth
y
lated tar
g
et site
s
(

Bestor 1992). In t
h
is stu
d
y, t
h
eC-an
d
N-termina
l
parts o
f
Dnmt1 most
l
i
k
e
l
y
remained in contact, but the proteol
y
tic cleava
g
e induced a conformational
c
h
ange t
h
at activate
d

t
h
e enzyme.
Interestin
g
l
y
, Dnmt1 bears at least two separate DNA bindin
g
sites, at leas
t
one in the N-terminal part and one in the C-terminal part (Arau
j
o et al. 2001;
F
atemi et al. 2001; Fl
y
nn and Reich 1998). The enz
y
me can interact with its
tar
g
et DNA and, in addition, with a second DNA molecule that function
s
as an a
ll
osteric regu
l
ator. Bin
d

ing to met
h
y
l
ate
d
DNA activates Dnmt1
f
or
meth
y
lation of unmodified tar
g
et sites (Bacolla et al. 1999; Fatemi et al. 2002
;
F
atemi et a
l
. 2001). Stea
d
y-state
k
inetic experiments
d
emonstrate t
h
at t
h
e
N-terminal part o f Dnmt1 has a repressive function on the catal

y
tic domain,
which is relieved after binding of methylated DNA to the N-terminus (Bacolla
et al. 2001). Experimental evidence su
gg
ests that bindin
g
of meth
y
lated DN
A
occurs within the Zinc-domain, which forms a direct
p
rotein/
p
rotein co ntact
to t
h
ecata
l
ytic
d
omain o
f
t
h
e enzyme (Fatemi et a
l
. 2001) or to a s
h

ort moti
f
in
between the PCNA interaction site and the nuclear localisation signal (NLS
)
(
Pra
dh
an an
d
Esteve 2003a
)
.Givent
h
ese resu
l
ts, at
l
east t
h
ree
d
i
ff
erent state
s
of Dnmt1 can be distin
g
uished: The isolated catal
y

tic domain is inactiv
e
towards hemimethylated and unmethylated DNA. With unmethylated DNA
the full-len
g
th enz
y
me shows low activit
y
. In t he presence of meth
y
lated DNA
,
the activity of Dnmt1 is much higher, suggesting that the N-terminal part has
two e
ff
ects: (1) It stimu
l
ates t
h
e C-termina
l
part
f
or genera
l
activity an
d
(2
)

either unmethylated DNA binding to the N-terminal part inhibits the enzym
e
or
b
in
d
ing o
f
met
h
y
l
ate
d
DNA stimu
l
ates t
h
e enzyme,
l
ea
d
ing to an increase
d
meth
y
lation of unmodified sites
.
T
h

is a
ll
osteric activation is a surprising e
ff
ect, as it means t
h
at,int
he
presence of meth
y
lated DNA, Dnmt1 loses specificit
y
for hemimeth
y
lated
D
NA and also starts working as a de novo MTase. Therefore, activated Dnmt
1
is
l
ess accurate in copying an existing met
h
y
l
ation pattern, w
h
ic
h
at
fi

rst
Molecular Enz
y
molo
gy
of M ammalian DNA Meth
y
ltransferase
s
2
1
7
sig
h
t appears as a mis-a
d
aptation
f
or a maintenance MTase. A
f
ter a
ll
osteric
stimulation, Dnmt1 has a similar activit
y
on unmeth
y
lated and hemimeth
y
-

latedDNA,su
gg
estin
g
that this enz
y
me could also have a role in de novo
m
et
h
y
l
ation o
f
DNA. Activate
d
Dnmt1 cou
ld
support Dnmt3A an
d
Dnmt3
B
in de novo meth
y
lation, a conclusion that is in a
g
reement with in vivo dat
a
d
emonstrating Dnmt1 is require

df
or
d
enovomet
h
y
l
ation (Liang et a
l
. 2002)
and overexpression of Dnmt1 can cause de novo meth
y
lation of DNA (Bin-
isz
k
iewicz et a
l
. 2002). T
h
is assumption is a
l
so supporte
db
yt
h
e
fi
n
d
ing t

h
at
Dnmt1 and Dnmt3A interact with each other
(
Datta et al. 2003; Kim et al.
2
002)
.
T
h
ea
ll
osteric activation mec
h
anism o
f
Dnmt1 ma
k
es DNA met
h
y
l
ation
behave in an all-or-none fashion, beca use some meth
y
lation will alwa
y
sat
-
t

ract more met
h
y
l
ation. In a
dd
ition, epigenetic signa
ll
ing comprises severa
l
p
ositive feedback loops: Initial DNA meth
y
lation could induce histone 3 l
y
-
sine 9 met
h
y
l
ation or
h
istone
d
eacety
l
ation (Cameron et a
l
. 1999; Fa
h

rner
et al. 2002; Sarraf and Stancheva 2004; Tariq et al. 2003). These responses in
t
urn could tri
gg
er additional DNA meth
y
lation (Bachman et al. 2003; Jack-
son et al. 2002; Lehnertz et al. 2003; Tamaru and Selker 2001
)
. Furthermore,
m
eth
y
lation of DNA could attract MeCP2 that itself would tar
g
et Dnmt1 to
th
e DNA (Kimura an
d
S
h
iota 2003). T
h
ere
f
ore, in a stea
d
y-state situation on
ly

completel
y
unmeth
y
lated and full
y
meth
y
lated re
g
ions of the DNA coexist,
wh
ic
h
are separate
db
yc
h
romatin
b
oun
d
ary e
l
ements. T
h
is a
ll
-or-none
b

e-
haviour mi
g
ht increase the efficienc
y
of epi
g
enetic circuits in switchin
g
o
n
and off
g
ene expression. These mechanisms also explain the observation tha
t
m
eth
y
lation tends to spread from heavil
y
meth
y
lated re
g
ions of the DNA into
nei
g
hbourin
g
unmeth

y
lated re
g
ions, which is often observed in cancer cells.
5.2
S
t
i
mulat
i
on of Dnmt3A and Dnmt3B b
y
Dnmt3
L
De novo meth
y
lation b
y
Dnmt3A and Dnmt3B is re
g
ulated b
y
at least one
additional protein, namel
y
Dnmt3L, which shows clear homolo
gy
to the
Dnmt3A and 3B enz
y

mes (Aapola et al. 2000). However, Dnmt3L carrie
s
m
utations wit
h
in a
ll
conserve
d
DNA-(cytosine-C5)-MTase moti
f
s. T
h
is o
b-
servation su
gg
ests that Dnmt3L adopts the t
y
pical MTase fold, but it does
not
h
ave cata
l
ytic activity. In co-trans
f
ection experiments, Dnmt3L
h
as
b

een
shown to stimulate DNA meth
y
lation b
y
Dnmt3A in human cell lines (Chedi
n
et a
l
. 2002). In vitro stu
d
ies
d
emonstrate
d
an appro ximate
l
y 15-
f
o
ld
activa
-
t
ion of Dnmt3A and Dnmt3B b
y
Dnmt3L (Gowher et al. 2005). Biochemica
l
studies demonstrate Dnmt3L directl
y

interacts with Dnmt3A and Dnmt3B
v
ia its C-termina
ld
omain
(
Gow
h
er et a
l
. 2005; Hata et a
l
. 2002; Sueta
k
eeta
l.