DNA Methylation: Basic Mechanisms - Part 5 pot

Function of DNA Meth
y
latio
n
12
7
fi
e
ld
in a conventiona
l
sense. Since t
h
eaut
h
or is a
l
so one o
f
t
h
ee
d
itors o
f
t
h
is series
o
f
C

urr en t Topics in Immuno
l
ogy a n
d
Micro
b
io
l
ogy on DNA met
h
y
l
ation, to w
h
ic
h
contri
b
utions
b
ymanyo
f
our co
ll
eagues in t
h
is
fi
e
ld h

ave
b
een invite
d
,t
h
eaut
h
or’
s
conscience is a
ll
eviate
d
t
h
at
h
e
h
as not cite
d
many o
f
t
h
ere
l
evant an
d

exce
ll
ent report
s
b
yot
h
ers. T
h
ec
h
oice o
f
vira
l
mo
d
e
l
systems in mo
l
ecu
l
ar
b
io
l
ogy is we
ll f
oun

d
e
d
.
Over many
d
eca
d
es, viruses
h
ave prove
d
t
h
eir inva
l
ua
bl
ean
d
pioneering ro
l
eastoo
l
s
in molecular
g
enetics. When our interest turned to the demonstration of
g
enome-wide

patterns of DNA meth
y
lation, we focused mainl
y
on the human
g
enome. The followin
g
topics in DNA meth
y
lation will be treated in detail: (1) The de novo meth
y
lation of in
-
te
g
rated forei
g
n
g
enomes; (2) the lon
g
-term
g
ene silencin
g
effect of sequence-specific
promoter meth
y
lation and its reversal; (3) the properties and specificit

y
of patterns o
f
DNA meth
y
lation in the human
g
enome and their possible relations to patho
g
enesis
;
(4) t
h
e
l
ong-range g
l
o
b
a
l
e
ff
ects on ce
ll
u
l
ar DNA met
h
y

l
ation an
d
transcriptiona
l
pro-
fil
es as a consequence o
ff
oreign DNA insertion into an esta
bl
is
h
e
d
genome; (5) t
h
e
patterns o
f
DN A met
h
y
l
ation can
b
e consi
d
ere
d

part o
f
ace
ll
u
l
ar
d
e
f
ense mec
h
anism
against
f
oreign or repetitive DNA; w
h
ic
h
ro
l
e
h
as
f
oo
d
-ingeste
d
DNA p

l
aye
d
in t
he
e
l
a
b
oration o
f
t
h
is mec
h
anism? T
h
e interest in pro
bl
ems re
l
ate
d
to DNA met
h
y
l
atio
n
h

as sprea
d
—
l
i
k
et
h
e mec
h
anism itse
lf
—into many neig
hb
oring
fi
e
ld
s. T
h
enatureo
f
the transcriptional pro
g
rams orchestratin
g
embr
y
onal and fetal development, chro-
matin structure,

g
enetic imprintin
g
,
g
enetic disease, X chromosome inactivation, and
tumor biolo
gy
are but a few of the areas of research that hav e incorporated studies o
n
the importance of the hitherto somewhat ne
g
lected fifth nucleotide in man
yg
enomes.
Even the fl
y
researchers now have to cope with the presence of this nucleotide, i
n
however small quantities it exists in the
g
enome of their model or
g
anism, at least
d
uring em
b
ryona
ld
eve

l
opment. T
h
e
b
u
lk
o
f
t
h
e experimenta
l
wor
k
accomp
l
is
h
e
d
in
t
h
eaut
h
or’s
l
a
b

oratory
h
as
b
een s
h
ou
ld
ere
db
ymanyverymotivate
d
un
d
ergra
d
uat
e
an
d
gra
d
uate stu
d
ents an
db
yanum
b
er o
f

ta
l
ente
d
post
d
octora
l
researc
h
ers. T
h
ei
r
contri
b
utions are re
fl
ecte
d
in t
h
e
l
ist o
f
re
f
erences in t
h

is c
h
apter. We
h
ave a
l
so
h
a
d
t
h
e
goo
dl
uc
k
to receive
f
un
d
ing t
h
roug
h
anum
b
er or organizations as ac
k
now

l
e
d
ge
d.
1
Intr
odu
cti
on
T
h
eresu
l
ts o
f
researc
h
on t
h
e
b
ioc
h
emistr
y
an
db
io
l

o
gy
o
f
DNA met
hyl
atio
n
ha ve
g
rown into a sizable bod
y
of scientific information. This series within
Cu rrent Topics in Micro
b
io
l
ogy an
d
Immuno
l
ogy
w
i
ll
provi
d
e a summar
y
o

f experimental work and evolvin
g
concepts. A sin
g
le chapter like this on
e
cannot, o
f
course, even attempt to p resent an a
d
equate overview o
f
t
h
is rapi
dly
d
eve
l
opin
gfi
e
ld
.T
h
is c
h
apter
h
as t

h
ere
f
ore
b
een restricte
d
to a s
y
nopsis o
f
se
l
ecte
d
wor
k
per
f
orme
d
in t
h
eaut
h
or’s
l
a
b
oratory

b
etween 1975 an
d
2005.
F
or a
l
on
g
time, man
y
co
ll
ea
g
ues in mo
l
ecu
l
ar
b
io
l
o
gy
resiste
d
reco
g
niz-

in
g
the fact that the fifth nucleotide in DNA, 5-meth
y
l-deox
y
c
y
tidine (5-mC),
exerts
d
ecisive
f
unctions in c
h
romatin structure an
d
in genetic contro
l
mec
h
-
1
2
8
W.
D
oe
rfl
er

anisms. Wit
h
5-mC,
h
owever, t
h
e arguments
h
ave
fi
na
ll
y
b
ecome too strong to
bei
g
nored.Nevertheless, textbooks still preach the existenc eoffour, insteado
f
five, nucleotides in DNA. Of course, it is
g
ood and essential scientific practic
e
to cast most critica
l
scrutiny on new c
l
aims an
dd
eman

d
amp
l
ean
dd
e
fi
nitive
experimental proof. A lar
g
e number of researchers hav e now provided thi
s
proo
f
,an
d
many o
f
t
h
e
fi
n
d
ings wi
ll b
e summarize
d
in t
h

ese v o
l
umes. My
own
g
roup started contributin
g
to the honin
g
of problems related to DNA
met
h
y
l
ation in t
h
emi
d
-1970s, an
d
t
h
is artic
l
e presents a
d
etai
l
e
d

summary o
f
ou
rr
esu
l
ts t
h
at
h
a
v
ebee
n
adduced s
in
ce t
h
e
n
a
n
d stood t
h
e test o
f
t
im
e.
F

or
further information
,
the reader can consult the references cited herein an
d
p
revious reviews t
h
at
h
ave
b
een
p
u
bl
is
h
e
d
as our wor
kp
rocee
d
e
d
(Doer
fl
e
r

1
981, 1983, 1995, 1996, 2000; Doerfler et al. 1988, 2001)
.
T
h
e
d
iscovery o
f
5-mC (Hotc
hk
iss 1948) in eu
k
aryotic, particu
l
ar
l
yin
mammalian, DNA has provoked a challen
g
in
g
search for i ts functional si
g
nif-
icance. T
h
is searc
h
is

b
y no means comp
l
ete
d
,an
d
active investigations on nu-
merous unsolved questions are still con tinuin
g
. The modification of c
y
tidin
e
(
C) to 5-mC, apparentl
y
the onl
y
one amon
g
the nucleotides in mammalian
D
NA, is introduced post-replicationall
y
b
y
several DNA meth
y
ltransferases

(
DMTases) that are chosen de pendin
g
on the functional context o f their en-
zymatic activity: DNA can
b
e met
h
y
l
ate
dd
e novo, sti
ll
a most enigmati
c
s
eries of events, or a
g
iven pattern of DNA meth
y
lation in the
g
enome can be
maintaine
d
upon rep
l
ication. In t
h

is
l
atter mo
d
eo
f
maintenance met
h
y
l
ation,
the parental DNA strand with the 5-mC residue still in place can serve a
s
the template to direct the DMTases to modif
y
the newl
y
s
y
nthesized DN
A
complement. Altho u
g
h several DMTases have been well characterized (for
areview,seethechapterb
y
T. Chen and E. Li, this volume), it is not clea
r
w
h

et
h
er any one o
f
t
h
em
b
y itse
lf
su
ffi
ces to
f
aci
l
itate eit
h
er o
f
t
h
e two mo
d
e
s
of DNA meth
y
lation. In addition to the enz
y

matic activit
y
pr oper, the func
-
tion o
f
t
h
ese enzymes seems to
d
epen
d
critica
ll
yont
h
econ
f
ormation o
f
t
h
e
l
ocal chromatin se
g
ment in which the DNA is to be meth
y
lated. Since o ur un-
derstanding of chr omatin structure is incomplete, we cannot expect to obtai

n
a comprehensive description of the enz
y
matic activities of the DMTases. It ap-
pears more realistic to p ropose a c omplex interplay between DNA-chromatin
s
tructure an
d
speci
fi
cc
h
oices o
f
enzymatic
f
unctions in w
h
ic
h
a
dd
itiona
l
reg
-
ulatory proteins have to participate. In experimen tal terms, DNA methylatio
n
activities cannot
b

erea
l
istica
ll
y assesse
db
yre
l
ying on t
h
e measurement o
f
enz
y
matic function usin
g
a naked DNA templat e, since the actuall
y
opera-
tiona
l
temp
l
ate
f
or DMTases is a DNA-c
h
romatin comp
l
ex wit

h
site-speci
fi
c,
s
tochasticall
y
malleable functions that are tar
g
eted to individual loci in th
e
g
enome. It will be some time before these processes can be elucidated or eve
n
mimic
k
e
db
ycurrenttec
h
no
l
ogy
.
Function of DNA Meth
y
latio
n
12
9

H
ow can we approac
h
a
f
unctio na
l
ana
l
ysis o
f
DNA met
h
y
l
ation in eu
k
ary-
o
tic, particularl
y
in mammalian, s
y
stems? One important parameter in un-
d
erstandin
g
this functional D NA modification is to realize that 5-mC residue
s
are not intro

d
uce
d
ran
d
om
l
y
b
ya
f
ortuitous
l
y acting enzymatic mec
h
anism.
In contrast, hi
g
hl
y
specific patterns in the distribution of 5-mC residues exis
t
a
ll
over t
h
e genome. T
h
ese patterns appear to
b

e
d
i
ff
erent in eac
h
ce
ll
type
and in each re
g
ion of the
g
enome. It will require a ma
j
or effort to determine
th
ese patterns o
f
DNA met
h
y
l
ation in a
ll
parts o
f
t
h
e mamma

l
ian, speci
fi
ca
ll
y
in the human,
g
enome.
In reco
g
nizin
g
the ver
y
si
g
nificant accomplishment of determinin
g
the
nuc
l
eoti
d
e sequence o
f
t
h
e
h

uman genome, I su
b
mit t
h
at t
h
e tas
kh
as not
been com
p
leted without the inclusion of the fifth nucleotide. Of co urse, it i
s
t
ec
h
nica
ll
y impossi
bl
eto
d
i
ff
erentiate
b
etween a C- an
d
a 5-mC-resi
d

ue
by
t
he conventional sequencin
g
reaction. The application of the bisulfite pro-
t
oco
l
o
f
t
h
e genomic sequencing reaction (Frommer et a
l
. 1992; C
l
ar
k
et a
l
.
1994) is a demandin
g
pro
j
ect, particularl
y
when it has to be extended t
o

m
an
y
kilobases of DNA sequence. Nevertheless, this method is, at least fo
r
t
he time bein
g
,theonl
y
reliable procedure to ascertain levels and patterns
o
f DNA meth
y
lation. B
y
appl
y
in
g
the bisulfite reaction, one can detect all
5-mC resi
d
ues in a sequence. T
h
e
h
uman epigenome project
h
as just

b
ee
n
initiated on an explorator
y
basis and will ha ve to cope with the fact that
p
atterns o
f
DNA met
h
y
l
ation can
b
e
d
i
ff
erent
f
rom ce
ll
type to ce
ll
typ
e
and, of course, in each se
g
ment of the

g
enome (Beck and Olek 2003). I
n
my
laborator
y
, we have investi
g
ated meth
y
lation patterns in several areas of
t
he human
g
enome to obtain a first impression of the t
y
pes of patterns (see
Sect. 6 of this chapter). The structure of the
g
enome inside its chromatin
casing an
d
its regu
l
atory
f
unctions appear to
d
epen
d

on t
h
ese patterns o
f
DNA meth
y
lation. The function of the
g
enome w ill not be understood be-
f
ore t
h
ecomp
l
etion o
f
t
h
e ana
l
ysis o
f
t
h
ese patterns. Hence, t
h
e stu
d
yo
f

m
ore complex bio medical problems will undoubtedl
y
escape a tho rou
g
hl
y
informed experimental approach before this analysis has been finished. Cur-
rentl
y
available data impl
y
that in 90 human
g
enes in the ma
j
or histocom-
p
atibility co mplex (MHC) of multiple tissues and individuals, the majorit
y
of
regions were
h
ypo- or
h
ypermet
h
y
l
ate

d
.T
h
e patterns were tissue-speci
fi
c,
interindi vidually variable, and correlated with gene expression (Rakyan et al
.
2
004
).
The followin
g
G
edankenmodel
l
m
a
y
aid the conceptual visualization of
a more genera
lf
unction o
f
patterns o
f
DNA met
h
y
l

ation across t
h
eentir
e
g
enome. The model is based upo n the notion that 5-mC re sidues are modu
-
lators of DNA-
p
rotein interactions, as
p
ro
p
osed earlier (Doerfler 1983), and
th
ese mo
d
u
l
ators cou
ld f
aci
l
itate an
d
en
h
ance or a
b
rogate suc

h
interactions
.
130
W.
D
oe
rfl
er
T
h
e
d
irection in w
h
ic
h
t
h
ese mo
d
u
l
ations wor
k
-
d
epen
d
sont

h
etypeo
f
protein and DNA sequences in functionall
y
crucial interactions
.
Ima
g
ine a bare wall represented here b
y
the plain nucleotide sequence o
f
A
,
C
,
G
,
an
d
T resi
d
ues onto w
h
ic
h
e
l
a

b
orate
d
ecorations
h
ave to
b
e attac
h
e
d.
Chromatin proteins then are t he decorations that even tuall
y
contribute t
o
t
h
ec
h
romatin structures an
d
cou
ld b
e speci
fi
c
f
or
d
i

ff
erent segments o
f
t
he
g
enome. Now, we insert into the blank wall the 5-mC “pe
g
s” to which protein
s
b
in
d
or are pro
h
i
b
ite
df
rom
b
in
d
ing. Wit
h
t
h
is
fi
rst an

d
essentia
l
set o
f
DN A
-
protein interactions, a central
g
enome-associated scaffold will be
g
enerated
that then will be able to inau
g
urate further protein and/or RNA assemblies
unti
l
t
h
e
fi
na
l
, yet enigmatic, c
h
romatin structure
h
as
b
een esta

bl
is
h
e
d.
L
ocal specificities in this structure will, ofcourse,be determined b
y
the site-
s
peci
fi
cpatterno
f
DNA met
h
y
l
ation t
h
at t
h
us assumes a
f
unctiona
ll
y crucia
l
role in this assembl
y

pr ocess. There are several, but one particular, proble
m
wit
h
t
h
is mo
d
e
l
:Itisnotapparentw
h
et
h
er t
h
e generation o
f
a given pattern o
f
D
NA meth
y
lation arises before or after chromatin forma tion. Possibl
y
,both
events are interdependent and develop concomitantl
y
. Upon DNA replication
,

an established and inheritable pattern of DNA meth
y
lation is, of course
,
maintained b
y
the arra
y
of 5-mC residues that are still preserved after DN
A
re
pl
ication in t
h
e
p
arenta
l
stran
d
an
d
t
h
at can serve as a tem
pl
ate
f
or t
h

e
insertion of meth
y
l
g
roups in the newl
y
s
y
nthesized DNA complement. In
t
h
is way, patterns o
f
DNA met
h
y
l
ation are propagate
d
an
d
in
h
erite
d
.T
he
meth
y

lation patterns in turn promote the site-specific chromatin structures
.
A
further tantalizin
g
aspect arises f rom the fact that DNA meth
y
lation
patterns are erased earl
y
in embr
y
onic development and are thereupon re-
imposed b
y
an unknown mechanism of de novo DNA meth
y
lation that canno
t
avai
l
itse
lf
o
f
t
h
etemp
l
ate pattern on t

h
ecomp
l
ementary stran
d
o
f
DNA.
Conversel
y
, the fixation of de novo meth
y
lation patterns on inte
g
rated forei
gn
D
NA or in t
h
e course o
f
em
b
ryonic
d
eve
l
opment mig
h
t

b
e
d
irecte
db
y
l
oca
l
c
hr
o
m
at
in
st
r
uctu
r
es t
h
at t
h
e
nw
ou
l
d
h
a

v
etobe
“r
e
m
e
m
be
r
ed
”
e
v
e
nin
t
h
e
absence of the fifth nucleotide. It is this crucial interde
p
endence between
meth
y
lation pattern and chromatin structure that we cannot
y
et satisfacto ril
y
explain. RNA could conceivably serve as a mediator for this functional gap in
time an
d

structure. T
h
is mo
d
e
l
is
b
ase
d
on t
h
e
fi
n
d
ing t
h
at eac
h
in
d
ivi
d
ua
l
s
egment of the genome is tightly associated with a given pattern of DN
A
met

h
y
l
ation an
d
, consequent
l
y, o
f
c
h
romatin structure. T
h
e same or a ver
y
s
imilar site-specific pattern can also be conve
y
ed to forei
g
n DNA subsequent
to its insertion into a speci
fi
c segment o
f
t
h
e mamma
l
ian genome

.
In part, this model has been deduced from the observation that the site
-
s
pecific re-integration of an unmethylated mouse gene, the B lymphocyt
e
tyrosine
k
inase (BLK) gene, into t
h
e mouse genome
b
y
h
omo
l
ogous recom
b
i
-
Function of DNA Meth
y
latio
n
1
31
nation
l
ea
d

stot
h
e reesta
bl
is
h
ment o
f
t
h
e origina
l
an
d
aut
h
entic DNA met
h
y-
lation pattern in the inte
g
rate at its authentic site (Hertz et al. 1999; Sect. 2.3.3)
.
In contrast, when the BLK
g
ene randoml
y
hits host DNA sequences and re-
com
b

ines t
h
ere
b
y a non-
h
omo
l
ogous mec
h
anism, patterns o
f
DNA met
h
y
l
a-
t
ion are completel
y
different from the authentic pattern in the BLK
g
ene. Fo
r
awor
k
ing
h
ypot
h

esis, we assume t
h
at eac
h
genome segment is c
h
aracterize
d
b
y
a “meth
y
lation memor
y
.” Its biochemical correlate is not known but mus
t
some
h
ow
b
ere
l
ate
d
to to
p
ica
l
c
h

romatin structure as we
ll
as
l
oca
l
DMTase
ty
pe, concentrations, and activities, as well as auxiliar
y
functions.
The most intensel
y
studied function of DNA meth
y
lation in eukar
y
otic
g
enomes is t
h
at o
f
promoter activity an
dl
ong-term gene si
l
encing. Startin
g
in the late 1970s, our laborator

y
has re
g
ularl
y
contributed to the elaboration
of
t
h
is conce pt (Doer
fl
er 1981, 1983,
f
or reviews). In conjunction (an
d
again
in an interdependent mode), DNA (5-mC) and chroma tin (histone acet
y
la-
t
ion an
d
met
h
y
l
ation) mo
d
i
fi

cation s co
ll
a
b
orate in t
h
e
l
ong-term si
l
encing o
f
p
romoters, and thus assume an essential function in re
g
ulatin
g
the activit
y
o
f specific
g
enome se
g
men ts. In recent
y
ears, these mechanisms have been
reco
g
nized to be of importance also for the understandin

g
of more complex
biomedical problems, in particular those that are related to
g
enetic imprint
-
ing, em
b
ryonic an
df
eta
ld
eve
l
opment, genetic
d
isease, an
d
tumor
b
io
l
ogy
.
H ere, we have another fine example of how basic research on fundamental
m
ec
h
anisms in mo
l

ecu
l
ar genetics can eventua
ll
y
h
e
l
pusun
d
erstan
d
practi
-
cal problems in biomedical research. Without turnin
g
to the stud
y
of simpler
experimental s
y
stems—e.
g
., viral models—in the elucidation of promo ter
silencin
g
b
y
DNA meth
y

lation and related histone modifications, it woul
d
ha ve been im
p
ossible to a
pp
roach more com
p
lex
p
roblems in mammalia
n
o
rganisms or in p
l
ants
.
2
T
he De Novo Methylation of Integrated Foreign DN
A
2
.
1
Cho
i
ce of Exper
i
mental System
s

T
h
ere are many exce
ll
ent examp
l
es
d
ocumenting t
h
at t
h
e stu
d
yo
f
vira
l
sys-
t
ems
h
as
l
e
d
to t
h
e
d

iscover
y
o
ff
un
d
amenta
l
mec
h
anisms in pro
k
ar
y
otic an
d
eu
k
aryotic mo
l
ecu
l
ar genetics. Since, in most instances, virus rep
l
ication
h
a
s
t
ore

ly
on t
h
euti
l
ization o
f
ce
ll
u
l
ar mec
h
anisms, it cannot
b
e surprisin
g
t
h
at
v
iruses have been efficientl
y
exploited as Tro
j
an horses inside the cellular
m
i
l
ieus.

13
2
W.
D
oe
rfl
er
In t
h
e 1970s, our
l
a
b
oratory was invo
l
ve
d
in
d
etai
l
e
d
ana
l
yses o n t
he
mode of adenoviral DN A inte
g
ration into the host

g
enome in adenovirus
-
transformed cells and in adenovirus t
y
pe 12 (Ad12)-induced hamster tumo
r
ce
ll
s. In t
h
e c ourse o
f
t
h
ese stu
d
ies, it
b
ecame
f
easi
bl
etoprovet
h
at integrate
d
forei
g
n (adenoviral) DNA was de novo meth

y
lated (Sutter et al. 1978). Whe
n
we su
b
sequent
l
y were a
bl
eto
d
ocument t
h
e
fi
rst inverse re
l
ations
h
ip
b
etwee
n
g
enetic activit
y
of inte
g
rated viral
g

enes and the extent of their meth
y
lation
(
Sutter an
d
Doer
fl
er 1980), it was o
b
vious t
h
at t
h
is experimenta
l
system cou
ld
be applied to fundamental studies on the re
g
ulation of
g
enetic activit
y
and
on the biolo
g
ical function of DNA meth
y
lation (Doerfler 1983).

A
secon
d
semina
l
o
b
servation, w
h
ic
h
emp
h
asize
d
t
h
e
b
io
l
ogica
l
impor-
tance of DNA meth
y
lation, came from detailed investi
g
ation s on patterns of
D

NA met
h
y
l
ationint
h
e5

-upstream regions o
f
two ran
d
om
l
yc
h
osen
h
uma
n
g
enes, tumor necrosis
f
actor (TNF)
-
α
a
n
d
TNF

-
β
.
T
h
e
g
enomic sequencin
g
met
h
o
d
origina
ll
y
d
eve
l
ope
db
yC
h
urc
h
an
d
Gi
lb
ert (1984), t

h
oug
hd
i
ffi
cu
l
tt
o
use at t
h
e time,
h
e
l
pe
d
consi
d
era
bly
in t
h
ese stu
d
ies. In t
h
epromoteran
d
5


-
upstream re
g
ions o f the TNF
-
α
a
n
d
TNF-
β
g
enes, we found cell t
y
pe-specific
a
n
d
interin
d
ivi
d
ua
lly h
i
ghly
conserve
d
patterns in t

h
e
d
istri
b
ution o
f
5-m
C
residues that a
g
reed to the nucleotide site amon
g
individuals from differen
t
et
h
nic origins (Koc
h
ane
k
et a
l
. 1990, 1991). Simi
l
ar, t
h
oug
hl
ess precise, ev-

idence came from lar
g
e, rando ml
y
chosen se
g
ments of the human
g
enom
e
(
Be
h
n-Krappa et a
l
. 1991), many o
f
t
h
em repetitive DNA sequences. T
h
ese
results implied that hi
g
hl
y
specific patterns of DNA meth
y
lation existed an
d

most likel
y
had to have a fundamentall
y
importan t function. It was not eas
y
in those da
y
s to convince others that a s
y
stematic endeavor to determin
e
patterns of DNA meth
y
lation was not
j
ust a descriptive exercise but had to
b
e initiate
d
to
l
earn a
b
out t
h
ewi
d
er gamut o
f

possi
b
i
l
ities wit
hf
unctiona
l
implications. Hopefull
y
, the human epi
g
enome pro
j
ect will help provide more
evi
d
ence t
h
an t
h
e stu
d
yo
f
a sing
l
e, pioneering,
l
a

b
oratory cou
ld
possi
bl
y
h
av
e
a
dduced with limited means 15
y
ears a
g
o
.
A
t that time, we also soughtthe collabo ration of clinical researchers inorder
to extend the basic concepts derived from simpler experimental s
y
stems t
o
more com
p
lex biomedical
p
roblems. In the course of these studies, it becam
e
even more o
b

vious t
h
at t
h
emo
d
e
ld
eve
l
ope
d
wit
h
t
h
ea
d
enovirus syste
m
could reliably guide all our efforts. In collaboration with several groups
,
we
d
etermine
d
patterns o
f
DNA met
h

y
l
ation in t
h
ePra
d
er-Wi
ll
i/Ange
l
ma
n
re
g
ions of the human
g
enome (Zeschni
g
k et al. 1997a, b; Schumacher et al.
1
998), in t
h
e p romoter regions o
f
t
h
e RET protooncogene (Munnes et a
l
. 2000)
,

of the FMR1
g
ene (Schwemmle et al. 1997; Genç et al. 2000) and of several
g
enes of the erythrocyte membrane (Remus et al. 2001, 2005).
Function of DNA Meth
y
latio
n
1
33
2.
2
T
he State of Methylat
i
on
i
n DNA V
i
ral Genome
s
2.2.
1
M
any DNA V
i
r
i
on Genomes Are Unmethylated, Others Are Methylate

d
Mo
d
u
l
ation as a
b
i
d
irectiona
ll
y active parameter in DNA-protein interac-
t
ions can
b
eexemp
l
i
fi
e
dby
t
h
eactivit
y
o
f
t
h
e restriction en

d
onuc
l
eases Dpn
I
an
d
D
p
nII. D
p
nI c
l
eaves t
h
enuc
l
eoti
d
ese
q
uences
G
6m
A
TC on
l
yw
h
en t

h
e
A
resi
d
ue in t
h
ereco
g
nition sequence is met
hyl
ate
d
,w
h
ereas DpnII is in
h
i
b-
ited b
ya
6
m
A
residue in this se
q
uence (for review, see McClelland and Nelson
1988). Hence, a met
h
y

l
ate
d
nuc
l
eoti
d
e can o
b
struct or
f
aci
l
itate w
h
i
l
e
b
eing
required for the activit
y
of a restriction endonuclease, i.e., for the interaction
b
etween a given nuc
l
eoti
d
e sequence an
d

t
h
e protein t
h
at speci
fi
ca
ll
yrec-
og
nizes t
h
is sequence. A simi
l
ar
ly
instructive examp
l
eisnotavai
l
a
bl
e
f
or
a
5-mC-containing recognition sequence o
f
a restriction en
d

onuc
l
ease.
Amon
g
t
h
e DNA containin
g
vira
lg
enomes, examp
l
es o
f
comp
l
ete
ly
un
-
m
eth
y
lated as well as completel
y5

-
CG
-

3

meth
y
lated virion DNA molecules
exist. T
h
e encapsi
d
ate
d
virion DNA o
f
t
h
e
h
uman a
d
enoviruses (Günt
h
ert e
t
al. 1976) is unmeth
y
lated. In strikin
g
contrast, the double-stranded
g
enom

e
of f
rog virus 3 (FV3), an iri
d
ovirus, is comp
l
ete
l
y met
h
y
l
ate
d
in a
ll 5

-
CG
-3

d
inuc
l
eoti
d
es
(
Wi
ll

is an
d
Grano
ff
1980; Sc
h
etter et a
l
. 1993
)
.Aswi
ll b
e
d
is-
cusse
dl
ater, t
h
e intrace
ll
u
l
ar FV3 virion DNA
b
ecomes quic
kl
y remet
h
y

l
ate
d
after replication. Possibl
y
, due to the specific nucleotide sequence of the FV3
g
enome, the viral and/or cellular proteins, which have to interact with thi
s
v
ira
lg
enome in t
h
e course o
f
vira
l
transcription an
d
rep
l
ication, are not in
h
i
b-
ited b
y
FV3 DNA meth
y

lation. Some of them ma
y
even require a meth
y
lated
g
enome
f
or
f
u
ll
activity.
We have used several techniques—includin
g
total h
y
drol
y
sis of the virio
n
DN A
f
o
ll
owe
db
y
b
i

d
irectiona
l
c
h
romatograp
h
yan
d
e
l
ectrop
h
oresis (Gün
-
t
hert et al. 1976)—that allow the separation of C from 5-mC residues, as wel
l
as
g
enomic sequencin
g
methods (Wienhues and Doerfler 1985; Kämmer and
Doerfler 1995
)
, to demonstrate that the virion DNA as well as the free, i.e., no
t
host cell
g
enome inte

g
rated, intracellular adenovirus DNA in productivel
y
o
r
a
b
ortive
l
yin
f
ecte
d
ce
ll
s(Var
d
imon et a
l
. 1980) remains unmet
h
y
l
ate
d
.Int
h
e
latter stud
y

, restriction endonucleases were used to document the absence of
5-mC resi
d
ues at
l
east in t
h
e HpaII recognition sequences 5

-
CCGG
-
3

.
T
h
e intrace
ll
u
l
ar
g
enomes o
f
t
h
e episoma
lly
persistin

g
Epstein-Barr viru
s
(EBV)
h
ave
b
ecome met
h
y
l
ate
d
to a certain extent (Ern
b
erg et a
l
. 1989). Simi
-
l
ar
ly
,t
h
e
g
enome o
f
anot
h

er persistin
g
virus,
h
erpesvirus saimiri, in
ly
mp
h
oi
d
t
umor cell lines has been shown to be extensivel
y
meth
y
lated (Desrosiers et
a
l
. 1979).
13
4
W.
D
oe
rfl
er
T
h
e retrovira
l

progenomes can a
l
so
b
ecome met
h
y
l
ate
d
(ear
l
yre
f
erence
s
on this topic are e.
g
., Conklin et al. 1982; Jähner et al. 1982)
2
.
2
.
2
S
YREC, an Ad12 Recomb
i
nant Genome That Carr
i
es Unmethylated Cellular DNA

W
h
en A
d
12 was seria
lly
propa
g
ate
d
on
h
uman ce
ll
sincu
l
ture,avariantA
d
1
2
g
enome arose t
h
at constitute
d
a natura
ll
y generate
d
recom

b
inant
b
etwee
n
t
h
e
l
e
f
t t ermina
l
2,081 nuc
l
eoti
d
es o
f
A
d
12 DNA an
d
a
l
ar
g
epa
l
in

d
romi
c
fra
g
ment of cellular DNA. This viral recombinant could be separated fro
m
t
h
eaut
h
entic A
d
12 virions
d
ue to its
l
ower
b
uoyant
d
ensity
b
yequi
l
i
b
riu
m
s

edimentation in CsCl densit
yg
radients (Deurin
g
et al. 1981). The existenc
e
o
f
t
h
is symmetrica
l
recom
b
inant (SYREC) prove
d
t
h
at recom
b
ination cou
ld
occur
b
etween vira
l
an
d
ce
ll

u
l
ar DNA in
h
uman ce
ll
st
h
at
h
ave
b
een pro
d
uc-
tive
l
yin
f
ecte
d
wit
hh
uman A
d
12 (Deuring an
d
Doer
fl
er 1983). T

h
ece
ll
u
l
ar
D
NA in t
h
is
h
u
g
epa
l
in
d
romic
g
enome o
f
some 34
kb
wit
h
i
d
entica
ll
e

f
ten
d
s
of A d12 on either terminus of the recombinan t
g
enome comprised cellula
r
D
NA sequences o
fb
ot
h
t
h
e unique an
d
repetitive t
y
pes. Interestin
gly
,t
h
es
e
cellular DNA sequences were completel
y
unmeth
y
lated in the virion recom-

b
inant,
b
ut t
h
e same ce
ll
u
l
ar DNA sequences were
h
ig
hl
y met
h
y
l
ate
d
in t
h
e
h
uman ce
ll
u
l
ar
g
enome

f
rom w
h
ic
h
t
h
e
yh
a
db
een ori
g
ina
lly d
erive
d
(Deur
-
ing et a
l
. 1981). T
h
is
fi
n
d
ing
d
emonstrates t

h
at
f
ree a
d
enovirion DNA remain
s
devoid of 5-m
C
in the same human cell nucleus in which adenovirion DNA
replicates and in which the meth
y
lation of cellular DNA is maintained i
n
s
peci
fi
cpatterns.Apparent
ly
,t
h
ece
ll
u
l
ar DMTases
f
ai
l
to

g
ain access to t
h
e
free virion DNA, possibl
y
because adenovirus DNA can avail itself of its own
,
s
peci
fi
c, virio n genome-enco
d
e
d
mec
h
anism o
f
DNA rep
l
ication wit
h
t
h
ea
d
e-
novirus terminal protein (TP), its viral DNA pol
y

merase (pol), and the DNA
b
in
d
ing protein (DBP). A
l
ternative
l
y, it is conceiva
bl
et
h
at
f
ree intranuc
l
ear
adenovirus DNA becomes protected from de novo meth
y
lation b
y
bindin
g
to
s
pecific proteins. Adenoviral DNA replication is at least partl
y
independent
of the cellular replication machiner
y

, except for the requirement for nuclea
r
factors I, II, and III that mi
g
ht not be linked to an
y
of the cellular DMTases.
On t
h
eot
h
er
h
an
d
,t
h
e intrace
ll
u
l
ar
l
y
l
ocate
d
,episoma
l
DNA o

f
EBV must
be ti
g
htl
y
associated with the replication s
y
stem for cellular DN A with whic
h
it rep
l
icates in sync
h
rony. T
h
us, t
h
e EBV episomes mig
h
t
b
einc
l
ose contac
t
with cellular DMTases and become meth
y
lated
.

T
h
ea
d
enovirus SYREC mo
l
ecu
l
ean
d
its a
b
i
l
ity to rep
l
icate in
h
uman ce
lls
in the presence of a helper adenovirus with an intact authentic viral
g
enome
h
as been the model for the construction of the
g
utless adenovirus vectors of
t
h
et

h
ir
d
generation (Koc
h
ane
k
et a
l
. 1996
b
). T
h
ese researc
h
ers
h
ave
b
een
Function of DNA Meth
y
latio
n
1
35
a
bl
etoseparatet
h

erecom
b
inant virus
f
rom its wi
ld
-type precursor a
l
so
by
equilibri um sedimentation in CsCl densit
yg
radients
.
2.2.3
S
uppress
i
on of the Frequenc
y
of 5

-
C
G-
3

D
i
n

u
cle
o
t
id
es
i
n the Gen
o
me
s
o
f the Small Eukar
y
otic Viruses
T
h
e
d
inuc
l
eoti
d
e
5

-CG
-3

is sta tistica

ll
yun
d
errepresente
d
in a
ll b
ut
f
our o
f
t
he small viruses with a
g
enome size of less than 30 kb (Karlin al. 1994).
In t
h
e
l
arger vira
l
genomes, t
h
ea
b
un
d
ance o
f
t

h
is
d
inuc
l
eoti
d
e
f
o
ll
ows sta-
t
istical expectations. The retrotransposons in eukar
y
otic
g
enomes are als
o
characterized b
y
low values of
5

-
CG
-3

d
inucleotides. There are several

p
os
-
si
bl
e interpretations
f
or t
h
ese p
h
enomena: (1) met
hyl
ation e
ff
ects
d
urin
g
t
he
p
roviral states of some of these
g
enomes, which would lead to their silencin
g,
(2)
d
inuc
l

eoti
d
e stac
k
ing energies, (3) mutation mec
h
anisms, or (4) se
l
ection
d
urin
g
evolution.
2.3
De Novo Methylat
i
on of Fore
i
gn DNA That Was Integrate
d
i
nto the Mammalian Genom
e
2
.
3
.
1
S
tud

i
es on Inte
g
rated Ad12 Genomes
i
n T ransformed or Tumor Cells
Inthecourseofinvesti
g
ations on the mode of Ad12 DNA inte
g
ration in
Ad
12-trans
f
orme
dh
amster ce
ll
s
by
usin
g
restriction en
d
onuc
l
eases, t
h
e
d

e
novo meth
y
lation of inte
g
rated forei
g
n DNA was discovered. The
g
enerated
f
ragments o
f
ce
ll
u
l
ar DNA were separate
db
ye
l
ectrop
h
oresis on agarose ge
l
s
and further anal
y
zed b
y

Southern blottin
g
(Southern 1975) and h
y
bridizatio
n
t
o
32
P
-
l
a
b
e
l
e
d
A
d
12 DNA or, more speci
fi
ca
ll
y, to t
h
e
32
P
-

l
a
b
e
l
e
d
termina
l
f
ra
g
ments o
f
A
d
12 DNA. In t
h
is wa
y
,t
h
e termina
l
vira
l
DNA
f
ra
g

ments
l
in
k
e
d
t
o the immediatel
y
abuttin
g
cellular DNA se
g
ments could be identified
.
In an attempt to
g
enerate sma
ll j
unction
f
ra
g
ments t
h
at c ou
ld b
emor
e
e

asil
y
anal
y
zed, frequent-cuttin
g
restrictases like HpaII were emplo
y
ed. In
th
ese experiments, we
d
iscovere
d
t
h
at t
h
e integrate
df
orm o
f
A
d
12 DNA was
n
ot effectiv el
y
cleaved b
y

HpaII, whereas virion DNA, previousl
y
shown to b
e
u
nmet
h
y
l
ate
d
(Günt
h
erteta
l
. 1976), was rea
d
i
l
ycut.T
h
ese
d
ata imp
l
ie
d
t
h
a

t
th
einte
g
rate
d
A
d
12 DNA
h
a
db
ecome
d
e novo met
hyl
ate
d
upon inte
g
ration
into t
h
e esta
bl
is
h
e
dh
amster genome (see Sutter et a

l
. 1978; Doer
fl
er 1982;
Doer
fl
er et a
l
. 1983,
f
or reviews
).
This inter
p
retation could be
p
roved when the isoschizomeric restrictio
n
e
n
d
onuc
l
ease pair HpaII an
d
MspI
b
ecame avai
l
a

bl
e. Bot
h
enzymes rec ognize
136
W.
D
oe
rfl
er
t
h
ese
q
uence
5

-CCGG
-3

.
Ms
p
Ic
l
eaves irres
p
ective o
f
t

h
e
p
resence o
f
a 5-m
C
residue in the 3

-
l
ocate
d
C-position in t
h
ereco
g
nition sequence,
b
ut HpaI
I
is capable of cleavin
g
onl
y
the unmeth
y
lated sequence (Waalwi
j
k and Flavel

l
1
978). A
l
ong t
h
ese
l
ines, t
h
e integrate
d
A
d
12 DNA,
l
i
k
e virion A
d
12 DN
A
s
tudied as a control, was completel
y
cleaved b
y
MspI, whereas HpaII could
c
l

eave on
l
yt
h
e virion DNA to comp
l
etion. Integra te
d
A
d
12 DNA was cut in-
comp
l
ete
ly b y
HpaII an
d
was t
h
us reco
g
nize
d
to
b
e
5

-
CCGG

-
3

met
hyl
ate
d
i
n
d
istinct patterns. Here, we were a
bl
eto
d
ocument one o
f
t
h
e ear
l
y examp
l
es
f
o
r
the notion that forei
g
n DNA inserted into established mammalian
g

enome
s
became heavil
y
meth
y
lated (Sutter et al. 1978; Sutter and Doerfler 1980). This
now common
l
yrepro
d
uce
dfi
n
d
ing was
l
ater extrapo
l
ate
d
to numerous ot
h
er
eukar
y
otic
g
enomes, includin
g

those of plants (for review, see Me
y
er 1995).
T
h
e
h
uman papi
ll
omaviruses (HPVs) 16 an
d
18 integrate
d
into t
he
g
enomes o f cells from human cervical carcinomas ar e als o meth
y
lated i
n
f
unctio na
ll
y
d
istinct patterns (Ba
d
a
l
et a

l
. 2004).
2.3.2
Si
te of In
i
t
i
at
i
on of De Novo Meth
y
lat
i
on: S
i
te of Fore
i
gn DNA Integrat
i
on
in the Recipient Genom
e
In later studies, we demonstrated in numerous Ad12-transfo rmed hamster
ce
ll l
ines an
d
particu
l

ar
l
yinA
d
12-in
d
uce
dh
amster tumor ce
ll
st
h
at inte-
g
rated Ad12 DNA is an excellent substrate for the action of cellular DMTase
s
(
Kuhlmann et al. 1982a, b; Orend et al. 1991). The patterns
g
enerated in dif
-
ferent cell lines and tumors exhibited some similarities but did not appear
to be identical. Extent and pattern of meth
y
lation of inte
g
rated forei
g
nDNA
were

d
irecte
d
rat
h
er
b
yt
h
e site o
f
integration in t
h
e recipient genome t
h
an
b
y
the nucleotide sequence of the forei
g
n DNA (Orend et al. 1995a), althou
g
h
t
h
e
l
atter cou
ld h
avesomein

fl
uence as we
ll
.Int
h
is context
,
t
h
eo
b
servatio
n
was of interest that a cloned E1 se
g
ment of adenovirus DNA
g
enomicall
y
fixe
d
b
y
transfection into hamster cells and inte
g
rated at several different loci i
n
the host hamster
g
enome became meth

y
lated to different extents or could re-
main h
y
po- or unmeth
y
lated (Orend et al. 1995a). Hence, the site of in sertion
o
ff
oreign DNA
h
a
d
to
b
e a strong
d
eterminant in its su
b
sequent
d
enov
o
meth
y
lation
.
T
h
e sites o

f
initiation o
fd
enovomet
h
y
l
ation were
d
etermine
db
y using
the cloned HindIII DNA fra
g
ments of Ad12 DNA as h
y
bridization probes o
n
Hp
aII- or Ms
p
I-c
l
eave
d
DNA
f
rom
d
i

ff
erent A
d
12-in
d
uce
dh
amster tumors
.
These DNA fra
g
ments were separated b
y
electrophoresis on a
g
arose
g
els and
s
ubsequently analyzed by Southern blotting and hybridization to th
e
32
P
-
l
a
b
e
l
e

d
A
d
12 DNA
f
ragments. T
h
eresu
l
ts o
f
a
l
arge num
b
er o
f
experiments
Function of DNA Meth
y
latio
n
1
37
(Oren
d
et a
l
. 1995a)
d

emonstrate
d
t
h
at
d
e novo met
h
y
l
ation was initiate
d
in
-
side the inte
g
rated Ad12 DNA molecules in two paracentrall
y
located re
g
ions
o
f the Ad12
g
enomes. De novo meth
y
lation did not commence at or close to
th
e termini o
f

A
d
12 DNA t
h
at were
l
in
k
e
d
to ce
ll
u
l
ar DNA. T
h
e termini
,
i
n
f
act, remained h
y
pometh
y
lated, possibl
y
because c ontinued h
y
pometh

y
lation
an
d
expression o
f
t
h
e gene pro
d
ucts
f
rom t
h
eA
d
12 regions E1 (
l
e
f
t terminus)
and E4 (ri
g
ht terminus) were likel
y
selected for in Ad12-transformed cell
s
an
d
in A

d
12-in
d
uce
d
tumors
.
We hav e investi
g
ated the site of initiation of de novo meth
y
lation in in
-
t
e
g
rated Ad12
g
enomes also at the nucleotide level b
y
appl
y
in
g
the bisul-
fi
te genomic sequencing reaction. W
h
en CsC
l

-puri
fi
e
d
A
d
12, 10
6
–10
7
pl
a
q
u
e
f
ormin
g
units per animal, is in
j
ected intramuscularl
y
into newborn hamster
s
(
Me socricetus auratus), numerous tumors o
fd
i
ff
erent sizes

d
eve
l
o
p
in t
h
e
animals’ peritoneal cavities (see also Sect. 2.3.6). In these intraperitoneal tu-
m
ors o
fd
i
ff
erent sizes, t
h
e paracentra
ll
y
l
ocate
d
regions o
f
integrate
d
A
d
12
g

enomes were ana
ly
ze
dby
t
h
e
b
isu
lfi
te protoco
lf
or t
h
e state o
f
met
hyl
ation.
M
eth
y
lation levels did not exhibit an unequivocal relation to tumor size.
Initiation of DNA meth
y
lation, moreover, was not emanatin
g
from a specific
nucleotide or set of nucleotides in the re
g

ion previousl
y
shown to be a site o
f
initiation o
fd
e nov o met
h
y
l
ation. Initiation was rat
h
er regiona
l
an
d
appeare
d
t
o emer
g
e from several sites within this re
g
ion (Orend et al. 1995a; Hohlwe
g
et a
l
. 2003). Hence,
d
e novo met

h
y
l
ation seems to commence in a region an
d
not at a sin
g
le specific nucleotide. Of course, this experimental approach doe
s
not allow us to derive a definite correlation between the time of forei
g
nDN
A
inte
g
ration and that of the initiation of DNA meth
y
lation.
In a related t
y
pe of experiment, a plasmid construct, which contained the
E2A
l
ate promoter o
f
a
d
enovirus type 2 (A
d
2) an

d
t
h
epro
k
aryotic gene
f
o
r
t
he chloramphenicol acet
y
ltransferase (CAT) as a reporter, was transfected
into
h
amster ce
ll
s. T
h
e HpaII-premet
h
y
l
ate
d
or unmet
h
y
l
ate

d
pA
d
2E2AL
-
C
AT
g
ene construct was
g
enomicall
y
fixed in hamster cells b
y
co-transfection
w
ith the unmeth
y
lated pSV2-neo plasmid. In this plasmid, the earl
y
simia
n
v
irus (SV)40 promoter controlled the neom
y
cin phosphotransferase
g
ene that
f
acilitated the selection of transgenic cells. Stability of methylation status and

expression o
f
t
h
e CAT gene were assesse
d
in a num
b
er o
f
c
l
ona
l
transgenic ce
ll
lines (Müller and Doerfler 1987). The foreign DNA was integrated frequently
in mu
l
tip
l
e tan
d
ems o
f
t
h
etrans
f
ecte

d
p
l
asmi
d
. Among 19 c
l
ona
l
ce
ll l
ines
,
t
he unmeth
y
lated c onstruct remained in that state, and in 18 of these lines the
C
AT gene was continuous
l
y expresse
d
. Among 14 ce
ll l
ines transgenic
f
or t
h
e
p

remeth
y
lated construct, 7 lines failed to express the test trans
g
ene, and the
t
hr
ee 5

-
CCGG
-
3

s
ites in its late E2A promoter remained almost completel
y
m
et
h
y
l
ate
d
.In5ce
ll l
ines t
h
e promoter remaine
d

part
l
y met
h
y
l
ate
d
an
d
t
h
e
138
W.
D
oe
rfl
er
CAT gene was on
l
y wea
kl
y expresse
d
.In2ce
ll l
ines, t
h
e premet

h
y
l
ate
d
pro
-
moter lost the 5m-C modification alto
g
ether, and the CAT
g
ene was stron
g
l
y
ex
p
ressed (Müller and Doerfler 1987).
2
.3.3
Factors Determining De Novo Meth
y
lation: Reinsertion of a Mouse Gen
e
i
nt
o
Its A
u
thent

i
cP
o
s
i
t
ion
We further pursued the
g
eneral question of which factors would affect the d
e
novo DNA met
h
y
l
ation in mamma
l
ian genomes. T
h
e mouse B
l
ymp
h
ocy t
e
t
y
rosine kinase (BLK)
g
ene was re-inte

g
rated b
y
homolo
g
ous reco mbinatio
n
into the
g
enome of mouse embr
y
onal stem (ES) cells. Two different plasmid
constructs containin
g
that
g
ene were used for these experiments. One con
-
s
truct also carried the weak E2A late
p
romoter of Ad2 DNA in front of the
l
uci
f
erase gene. In t
h
e secon
d
,t

h
is gene was contro
ll
e
db
yt
h
e strong ear
ly
S
V40 promoter. Upon homin
g
throu
g
hhomolo
g
ous recombination to th
e
aut
h
entic c
h
romosome 14 or
b
y
h
etero
l
ogous recom
b

ination to many
d
i
ff
er-
ent loci in the mouse
g
enome, meth
y
lation patterns in the inte
g
rates were
assessed b
y
restriction with the meth
y
lation-sensitive endonuclease HpaII o
r
the insensitive MspI. The mouse BLK
g
ene reinserted into the
g
enome b
y
ho
-
molo
g
ous recombination had reestablished the identical meth
y

lation pattern
c
h
aracteristic
f
or t
h
eaut
h
entic, non-manipu
l
ate
d
mouse BLK gene (Hertz e
t
al. 1999). The extent of de novo meth
y
lation in the DNA se
g
ments ad
j
acent t
o
t
h
e BLK gene in t
h
e integrate
d
construct

d
epen
d
e
d
on t
h
e promoter presen
t
in the plasmid construct and on the location of the recombined construc
t
in t
h
e ES genome. In
h
omo
l
ogous
l
y inserte
d
DNA, w
h
ic
h
carrie
d
t
h
e wea

k
Ad2 promoter, de novo meth
y
lation was exten sive. Presence of the stron
g
S
V40 promoter led to h
y
pometh
y
lation or no meth
y
lation at all. When the
en
h
ancer sequence was remove
df
rom t
h
e SV40 promoter, it a
l
so
b
ecame
h
y
-
permeth
y
lated. All randoml

y
inte
g
rated constructs, independent of the t
y
pe
o
f
promoter or en
h
ancer inc
l
u
d
e
d
, were
h
ypermet
h
y
l
ate
d
in patterns
d
i
ff
erent
from the ori

g
inal meth
y
lation pattern of the mouse BLK
g
ene.
W
econc
l
u
d
e
d(
Hertz et a
l
. 1999
)
:
1
. That an authentic mouse
g
ene reinserted into its ori
g
inal
g
enomic site wa
s
r
emet
h

y
l
ate
d
to t
h
ei
d
entica
l
pattern as previous
l
y present on t
h
e target
a
n
do
n
t
h
ea
ll
e
li
cs
i
te
2. That heterolo
g

ous recombination to randoml
y
tar
g
eted loci did not confe
r
t
h
e mouse BLK gene-speci
fi
c met
h
y
l
ation pattern
3
. That promoter stren
g
th in a construct was able to influence the pattern o
f
methylation imposed de novo on the inserted construct aft er homologous
r
ecom
b
ination in t
h
e mouse genome
Function of DNA Meth
y
latio

n
13
9
2.3.4
De Novo DNA Methylat
i
on: An Anc
i
ent Cellular Defense Mechan
i
sm
?
The de novo meth
y
lation of inte
g
rated forei
g
n DNA is a phenomenon widel
y
d
ocumente
d
t
h
roug
h
out t
h
ep

h
y
l
ao
f
eu
k
aryotic organisms, in mamma
l
ian
s
as well as in plants (f or review, see Me
y
er 1995). The currentl
y
establishe
d
g
enomes
h
ave evo
l
ve
d
over many mi
ll
ennia. We ten
d
to assume t
h

at t
he
evolution of
g
enomes is a continuous process that has started ri
g
ht after th
e
be
g
innin
g
of or
g
anismic life and probabl
y
before or
g
anisms arose and will
continue as
l
ongast
h
is
b
io
l
ogica
l
system can

b
e maintaine
d
.Evenun
d
er
experimental conditions, man
y
or
g
anisms have been shown t o be capable of
accepting an
d
accommo
d
ating
f
oreign DNA, a
l
t
h
oug
h
wit
h
unp re
d
icta
bl
e

—
advanta
g
eous or catastrophic—consequences for the acc eptor cell. Since all
ce
ll
sincu
l
ture or organisms in environment-c
h
a
ll
enge
dl
i
f
earesu
b
ject t
o
strin
g
entconditionsofselection,eventhecellthathasbeenforcedb
y
its innate
recombination mechanisms to tolerate the
g
enomic in sertion o f forei
g
nDNA

can avail itself of an ancient defense mechanism a
g
ainst the
g
enetic activit
y
o
fforei
g
n DNA that could carr
y
active
g
enes. Since promoter meth
y
latio
n
h
as
b
een i
d
enti
fi
e
d
as part o
f
amec
h

anism
f
or t
h
e
l
ong-term si
l
encing o
f
g
enes and DNA se
g
ments, the de novo meth
y
lation of inte
g
rated forei
g
nDN
A
can
b
econtemp
l
ate
d
as suc
h
a

d
e
f
ense mec
h
anismorat
l
eastasanintegra
l
p
art of it (Doerfler 1991; Yoder et al. 1997). A lar
g
e part of 5-mC residues is
f
ound in the parasitic sequence elements of retrotransposons and endo
g
enou
s
retroviruses that constitute more than 35% of the human
g
enome. Perhap
s
intra
g
enomic parasites are reco
g
nized b
y
their hi
g

hcop
y
number. Lon
g
-ter
m
inactivation
b
y DNA met
h
y
l
ation a
l
so entai
l
st
h
e possi
b
i
l
ity o
f
5-mC resi
d
ue
s
bein
g

deaminated to Ts followed b
y
permanent inactivation (Bestor 1998)
.
2.3.5
A
re Integrated Fore
i
gn DNA Sequences Stab
i
l
i
zed b
y
H
y
permeth
y
lat
i
on
?
The Ad12-transformed cell line T637 was obtained b
y
the in vitro transforma-
t
ion of BHK21 hamster cells with Ad12 (Strohl et al. 1967) and carries abou
t
15 copies o
f

A
d
12 DNA integrate
d
at a sing
l
ec
h
ro mosoma
ll
ocus (Sta
b
e
l
et a
l
1980; Knoblauch et al. 1996; Schröer et al. 1997). Uninte
g
rated, free Ad12 DN
A
cou
ld
never
b
e
d
etecte
d
in any o
f

t
h
ea
d
enovirus-trans
f
orme
d
ce
ll l
ines or i
n
A
d12-induced tumor cells. The inte
g
rated Ad12 DNA is meth
y
lated in func-
t
iona
ll
y signi
fi
cant patterns (Sutter an
d
Doer
fl
er 1980; Ho
hl
weg et a

l
. 2003)
.
U
pon continuous propa
g
ation in cell culture, a small number of morpholo
g-
ical revertants arose from the cell line T637 (Groneber
g
et al. 1978). Thes
e
revertants ex
h
i
b
ite
d
a
fib
ro
bl
astic p
h
enotype in contrast to t
h
emoreroun
d
e
d

,
1
4
0
W.
D
oe
rfl
er
e
p
it
h
e
l
ioi
d
a
pp
earance o
f
t
h
e T637 ce
ll
s. T
h
e revertants can
b
ese

l
ecte
dd
u
e
to their resilience to spent, acidic medium in which T637 cells detach
.
W
e have studied several of these revertants with res
p
ect to their conten
t
o
f
resi
d
ua
l
vira
l
DNA
(
Eic
k
et a
l
. 1980
)
.Inoneo
f

t
h
ese revertants, TR12
,
onl
y
one complete cop
y
of Ad12 DN A and an approximatel
y
3.9-kb-lon
g
fra
g
-
ment
f
rom t
h
erig
h
t terminus o
f
A
d
12 DNA persist in t
h
e integrate
d
state (N

.
H
ochstein, I. Muiznieks, and W. Doerfler, manuscript in preparation). Studie
s
using met
h
y
l
ation-sensitive restriction en
d
onuc
l
eases rev ea
l
e
d
t
h
at t
h
einte
-
g
rated Ad12 DNA in the revertant cell line TR12 was even more extensivel
y
meth
y
lated than in the cell line T637 from which TR12 ori
g
inated (Orend et

a
l
. 1995
b
). Pre
l
iminary resu
l
ts a
dd
uce
df
rom experiments using t
h
e
b
isu
lfi
te
s
equencin
g
method confirm these data (N. Hochstein and W. Doerfler, un-
pu
bl
is
h
e
d
). We propose t

h
at
h
yper- or near
l
ycomp
l
ete
l
y
5

-
CG
-3

met
h
y
l
ate
d
f
orei
g
n DNA sequences wou
ld b
e more sta
bly
inte

g
rate
d
t
h
an
l
ess comp
l
ete
ly
met
h
y
l
ate
df
oreign genomes. Repetitive sequences,
l
i
k
een
d
ogenous retrovi
-
ra
l
retrotransposons, appear to
b
esi

g
ni
fi
cant
ly b
ut not comp
l
ete
ly
met
hyl
ate
d
(
Heller et al. 1995).
2
.3.6
De Novo Methylat
i
on of Ad12 and of Cellular DNA
i
n Hamster Tumors
T
h
e insertion o
ff
oreign DNA into esta
bl
is
h

e
d
mamma
l
ian genomes
h
as con
-
s
equences
f
or t
h
e inserte
df
orei
g
nDNAan
df
or t
h
e recipient
h
ost
g
enome.
We have chosen to stud
y
these events in Ad12-induced hamster tumor cells
,

amo
d
e
l
in
b
asic researc
h
as we
ll
as
f
or its si
g
ni
fi
cance in vira
l
onco
l
o
gy
.
When Ad12 is in
j
ected subcutaneousl
y
into newborn hamsters within 24
h
a

f
ter
b
irt
h
,un
d
i
ff
erentiate
d
tumors
d
eve
l
o
p
at t
h
e site o
f
to
p
ica
l
a
ppl
icatio
n
of the virus within a few weeks in 70

%
–90
%
of the animals that survive in-
jection. As mentione
d
a
b
ove, upon t
h
e intramuscu
l
ar injection o
f
A
d
12 int
o
t
h
e
gl
utea
l
re
g
ion, numerous tumors are
f
oun
d

intraperitonea
lly
. Histo
l
o
g
i
-
call
y
, these tumors exhibi t Homer-Wri
g
ht rosette-like structures indicative of
primiti ve neuroecto
d
erma
l
tumors (PNET)
.
The tumor cells ex
p
ress
p
roteins that are characteristic both for neu-
roepit
h
e
l
ia
l

as we
ll
as
f
or mesenc
h
yma
l
ce
ll
s(Ho
hl
weg et a
l
. 2003, 2004)
.
Each tumor cell carries multiple copies of inte
g
rated Ad12 DNA. Free vi
-
ra
l
DN A
h
as never
b
een
f
oun
d

in any o
f
t
h
ese tumor ce
ll
s. T
h
ece
ll
s
f
ro
m
an individual tumor carr
y
the viral inte
g
rates, with f ew exceptions, all a
t
a sing
l
ec
h
romosoma
ll
ocation t
h
at is i
d

entica
l
in a
ll
ce
ll
so
f
a given tu-
mor. Amon
g
60 different tumors investi
g
ated, onl
y
one showed two differen
t
chromosomal loci to be occupied b
y
Ad12 DNA when investi
g
ated b
y
the
fl
uorescent in situ
h
y
b
ri

d
ization (FISH) tec
h
nique. W
h
en we compare
d
t
he
Function of DNA Meth
y
latio
n
14
1
sites o
f
A
d
12 DNA integration in more t
h
an 100
d
i
ff
erent
h
amster tumors
,
t

he sites of viral DNA inte
g
ration in the host cell
g
enome were differen
t
f
rom tumor to tumor (Doerfler 1982; Kuhlmann and Doerfler 1982; Hil
g
er-
Evers
h
eim an
d
Doer
fl
er 1997). We, t
h
ere
f
ore,
f
avor a mo
d
e
l
o
f
c
l

ona
l
origi
n
o
f these Ad12-induced tumors. Aside fr om the de novo meth
y
lation of th
e
v
ira
l
integrates in a
ll
o
f
t
h
eA
d
12-in
d
uce
d
tumors (see a
l
so Sect. 2.3), t
h
ere
can be chan

g
es in the extent of DNA meth
y
lation also in the host
g
enom
e
(He
ll
er et a
l
. 1995). Mor eover, t
h
e transcription patterns o
f
ce
ll
u
l
ar genes
in different tumors can be ver
y
similar but there are also differences. Th
e
p
atterns of transcription of the inte
g
rated Ad12
g
enes are st rikin

g
l
y
simi-
l
ar in a
ll
A
d
12-in
d
uce
dh
amster tumors ana
l
yze
d
an
d
resem
bl
et
h
ose pat-
t
erns in the Ad12-transformed hamster cell line T637 (Hohlwe
g
et al. 2003,
2
004

).
2.3.
7
Loss of Ad12 Genomes Is Compat
i
ble w
i
th Ma
i
ntenance of the Oncogen
i
c Phenot
y
pe
U
pon continuous cell culture, some of the Ad12-induced hamster tumor cells,
w
hich carried multiple copies of inte
g
rated and de novo meth
y
lated Ad1
2
g
enomes,
l
ost t
h
e vira
l

DNA sequences comp
l
ete
l
yora
l
most comp
l
ete
l
y
.
Surprisin
g
l
y
, these revertants devoid of Ad12 DNA and notabl
y
lackin
g
its
l
e
f
t terminus wit
h
t
h
e trans
f

orming E1 r egion o
f
t
h
e vira
l
genome retaine
d
t
heir onco
g
enic phenot
y
pes when rein
j
ected into hamsters (Kuhlmann et al.
1982; P
f
e
ff
er et a
l
. 1999). We consi
d
er t
h
is resu
l
t
h

ig
hl
y signi
fi
cant in t
h
a
t
it demonstrates that Ad12 is capable of inducin
g
tumors, which keep thei
r
o
ncogenic properties in animals, even when the E1 region is completely los
t
f
rom t
h
ese tumor ce
ll
s. Many researc
h
ers in t
h
e
fi
e
ld
o
f

a
d
enovirus tumo
r
biology consider the E1 region of the adenoviral genome akin to an oncogen
e
an
d
its main tenance paramount in preserving t
h
e oncogenic potentia
l
o
f
t
he
adenovirus-induced tumor cells. However, this notion has been experimen
-
t
a
ll
y
d
erive
d
main
l
y
b
yusingratem

b
ryo
fib
ro
bl
asts t
h
at were trans
f
orme
d
i
n
culture throu
g
h the transfection of adenovirus t
y
pe 5 (Ad5) DNA fra
g
ments
(for review, see Zantema and van der Eb 1995). Obviously, tumor induction
b
y
Ad12 in animals can be a quite different process and can pro babl
y
not
appropriately be mimicked by transfection experiments in cell culture. I
n
Sect. 7 o
f

t
h
is c
h
apter, I pro pose an a
l
ternate way o
fl
oo
k
ing at t
h
e mec
h
anis
m
o
f viral onco
g
enesis. In this view,
g
lobal chan
g
es in the cellular
g
enome as
a consequence o
f
vira
l

DNA integration are invo
k
e
d
as an importan t part o
f
t
he mechanism of viral onco
g
enesis in addition to the expression of a sin
g
le
v
ira
l
“oncogene” (Doer
fl
er 2000; Doer
fl
er et a
l
. 2001).
1
42
W.
D
oe
rfl
er
2

.3.8
DNA Methylat
i
on
i
n non-CpG D
i
nucleot
i
des; Hem
i
methylated DNA
T
h
e esta
bl
is
h
ment o
fd
enovopatternso
f
DNA met
h
y
l
ation in mamma
l
ia
n

g
enomes is characterized b
y
the
g
radual spreadin
g
of meth
y
lation, which has
b
een
d
ocumente
d
to occur across mu
l
tip
l
e copies o
f
in tegrate
d
a
d
enoviru
s
g
enomes as we
ll

as, at t
h
enuc
l
eoti
d
e
l
eve
l
,int
h
einte
g
rate
d
E2A promoter
of Ad2 DNA (Müller and Doerfler 1987; Toth et al. 1989). A few
5

-CG
-3

s
equences can remain
h
emimet
hyl
ate
df

or sev era
l
ce
ll g
enerations
b
e
f
ore t
h
e
y
become totall
y
meth
y
lated. Hemimeth
y
lation ma
y
be a transient phenomeno
n
b
ut cou
ld
a
l
so persist
f
or a certain perio

d
in speci
fi
c segments o
f
a transgene
or in the
g
enome in
g
eneral. In the Ad2-transformed cell line HE2, the E2A
promoter in t
h
e integrate
d
A
d
2 genome is
h
eavi
l
y met
h
y
l
ate
d
not on
l
yina

ll
5

-
CG
-3

d
inuc
l
eoti
d
es
b
ut a
l
so in some5

-
C
A-
3

an
d5

-
C
T-3


d
inuc
l
eoti
d
es
(
Tot
h
et al. 1990).Evidence for the occurrenceof5-mC innon-
5

-CG
-3

d
in
uc
l
eot
i
des
h
as a
l
so
b
een presente
dby
Wo o

d
coc
k
et a
l
. (1987). T
h
ere is at present n
o
plausible explanation concernin
g
whether and how this t
y
pe of meth
y
latio
n
can
b
e maintaine
df
o
ll
owing DNA rep
l
ication.
2
.
3
.

9
In
i
t
i
at
i
on and Spread
i
ng of De Novo Meth
y
lat
i
on
The mechanism of de novo meth
y
lation is not well understood. From th
e
resu
l
ts o
f
experiments per
f
orme
d
wit
h
t
h

ea
d
enovirus s
y
stem, t
h
e
f
o
ll
owin
g
conclusions a
pp
ear well founded
:
1
. In a trans
g
ene t
h
e size o
f
A
d
12 DNA, 34,125-
b
p(Spren
g
e

l
et a
l
. 1994)
,
de novo meth
y
lation starts in internal re
g
ions of the
g
enome (Orend et
al
. 1991, 1995a) an
d
sprea
d
s
f
rom t
h
ere across t
h
e transgene (Tot
h
et a
l
.
1
989, 1990). In transformed or tumor cells, the earl

y
re
g
ions, particularl
y
E
1, can
b
epart
l
yspare
df
rom
d
e novo met
h
y
l
ation,
b
ecause t
h
eir gen
e
p
roducts are required durin
g
the selection for the onco
g
enic phenot

y
pe.
2. Initiation o f de novo meth
y
lation is re
g
ional and not confined to one o
r
af
ew contiguous
5

-CG
-3

d
inuc
l
eoti
d
es (Ho
hl
weg et a
l
. 2003). In A
d
12-
i
nduced tumors the extent of Ad12 trans
g

ene meth
y
lation is not onl
y
d
e
p
en
d
ent on tumor size
.
3
. Transcribed re
g
ions of the trans
g
ene are h
y
pometh
y
lated, and inactive
s
e
g
ments are h
y
permeth
y
lated (Sutter and Doerfler 1980; Muiznieks an
d

Doer
fl
er 1994a; Munnes an
d
Doer
fl
er 1997
).
4
. Spreadin
g
of DNA meth
y
lation is not entirel
y
conti
g
uous. For unknow
n
r
easons
,
certa
i
n5

-
CG
-3


dinucleotides can remain unmeth
y
lated (N.
Hoc
h
stein, I. Muiznie
k
s, an
d
W. Doer
fl
er, manuscri
p
tin
p
re
p
aration).
Function of DNA Meth
y
latio
n
14
3
5. T
h
e extent o
fd
e novo met
h

y
l
ation an
d
t
h
e spee
d
o
f
its sprea
d
ing seem to
be determined b
y
the chromosomal site of trans
g
ene localization, and t
o
a lesser extent b
y
the nucleotide sequence of the trans
g
ene (Orend et al.
1
995; Hertz et a
l
. 1999
)
.

6. The reestablishment of the authentic pattern of DNA meth
y
lation in th
e
correctl
y
reinserted BLK
g
ene in mouse embr
y
onic stem cells (Hertz e
t
a
l
. 1999) imp
l
ies t
h
at eac
h
segment o
f
t
h
e mamma
l
ian genome is capa
-
ble of exertin
g

a memor
y
function that could be directl
y
related to th
e
mec
h
anism o
f
maintenance met
h
y
l
ation. Speci
fi
cc
h
romatin structures a
t
different sites of the
g
enome ma
y
be decisive in directin
g
the meth
y
lation
reaction. In sequences a

d
joine
d
to t
h
e BLK gene, t
h
e
d
e novo met
h
y
l
atio
n
patterns depended on the stren
g
th of the viral promoter in the construct
.
The weaker E2A late
p
romoter o f Ad2 DNA led to a mo re extensi ve de
novo meth
y
lation than the presence of the stron
g
er ear l
y
promoter o
f

S
V40 DN
A.
7
. In the same set of ex
p
eriments (Hertz et al. 1999), we noted that the sam
e
re-inte
g
rated BLK
g
ene, when it was inserted at randoml
y
selected sites in
the
g
enome, was h
y
permeth
y
lated in patterns completel
y
different fro
m
t
h
e origina
l
BLK gene pattern an

d
in
d
epen
d
ent
l
yo
f
promoter strengt
h
.
8.
Wh
e
nA
d
12-
t
r
a
n
s
f
o
rm
ed ce
ll
so
rA

d
12-in
duced
h
a
m
ste
r
tu
m
o
r
ce
ll
s
w
e
r
e
continuouslypassagedinculture,revertantsarosethathadlostallorapart
o
f
t
h
emu
l
tip
l
e copies o
f

integrate
d
A
d
12 DNA. T
h
et
h
en-persisting A
d
1
2
D
NA genomes in the revertants seemed to be more completely methylated
t
h
an t
h
e
l
ost copies. Hence, t
h
ei
d
ea was put
f
orwar
d
t
h

at t
h
e
l
eve
l
so
f
DNA
meth
y
lation of trans
g
enes mi
g
ht be related to their stabilit
y
of fixation
in the host genome (Orend et al. 1995b). Moreover, continuous cultur
e
o
f
d
iff
e
r
e
n
t
A

d
12-in
duced tu
m
o
r
ce
ll
s
l
ed so
m
et
im
es to t
h
ese
l
ect
i
o
n
of
cell lines with very similar integration patterns (Orend et al. 1994). It i
s
conceiva
bl
et
h
at t

h
e integration sites in t
h
ece
ll
sse
l
ecte
d
t
h
is way were
most com
p
atible with the survival of these cells in culture
.
3
Inverse Corr elat
i
ons Between Promoter Act
i
v
i
ty and Methylat
i
on
T
h
e
fi

e
ld
o
f
DNA met
h
y
l
ationin mamma
l
ian ce
ll
s was pioneere
db
yt
h
e
d
iscov
-
er
y
o
f
inverse corre
l
ations
b
etween t
h

e extent o
f
se
g
menta
l
DNA met
hyl
ation
and the
g
enetic activit
y
of these se
g
ments. In inte
g
rated Ad12 DNA in A d12-
t
rans
f
orme
d
ce
ll
s, t
h
e ear
l
yvira

l
genes are transcri
b
e
d
,an
d
t
h
e
l
ate A
d
12
1
44
W.
D
oe
rfl
er
g
enes responsi
bl
e
f
or t
h
esynt
h

esis o
f
virion capsi
d
proteins are permanent
l
y
s
ilenced. Hence, in Ad12-transformed cells or in Ad12-induced tumor cells
,
l
ate viral
g
ene products and mature virions are not s
y
nthesized (Ortin et al.
1
976). T
h
us, A
d
12-trans
f
orme
d
ce
ll
sprovi
d
e

d
a suita
bl
etoo
l
to stu
d
yt
he
l
evels of DNA meth
y
lation in distinct sections of the viral
g
enome and t
o
d
ocument inverse corre
l
ations to gene transcription
f
or t
h
e
fi
rst time (Sutter
and Do erfler 1979, 1980; Vardimon et al. 1980, 1981). A particularl
y
clea
r

examp
l
e
h
as
b
een o
ff
ere
db
yA
d
2-trans
f
orme
d
ce
ll l
ines HE1, HE2, an
d
HE3
(
Cook and Lewis 1979). The E2A re
g
ion of Ad2 DNA is not expressed in cel
l
l
ines HE2 and HE3, whereas cell line HE1 does express the E2A re
g
ion of

t
h
e integrate
d
A
d
2DNA(Jo
h
ansson et a
l
. 1978). Accor
d
ing
l
y, t
h
e5

-
CCGG
-3

dinucleotides (HpaII sites) in the promoter re
g
ion of this
g
ene are unmeth
y-
l
ate

d
in ce
ll l
ine HE1
b
ut met
h
y
l
ate
d
in ce
ll l
ines HE2 an
d
HE3 (Var
d
imon e
t
a
l
. 1980). At t
h
e time, one
h
a
d
to re
ly
on t

h
e ana
ly
ses o
f
a
l
imite
d
num
b
er o
f
5

CG
-3

d
inuc
l
eoti
d
es w
h
ose state o
f
met
h
y

l
ation cou
ld b
e assesse
d
on
l
y
b
y
t
h
e use o
f
met
hyl
ation-sensitive restriction en
d
onuc
l
eases, in t
h
is examp
l
e
of H
p
aII and Ms
p
I(5


-
CCGG
-
3

). These seminal observ ations started a burs
t
o
f
simi
l
ar stu
d
ies wit
h
numerous vira
l
an
d
ce
ll
u
l
ar
g
enes an
d
con
fi

rme
d
a
l
-
most without exception the initial observations that su
gg
ested that specifi
c
promoter met
h
y
l
ation patterns are instrumenta
l
in t
h
e
l
ong-term si
l
encing o
f
eukar
y
otic
g
enes (for review, see Doerfler 1983; Munnes and Doerfler 1997)
.
To t

h
is
d
ay ,
h
owever, we
d
onotun
d
erstan
d
w
h
et
h
er t
h
ere
h
ave to
b
e a speci
fic
number or pattern of 5-mC residues in a promoter to assure its lon
g
-ter
m
s
ilencin
g.

3.1
Pr
odu
ct
i
ve Vers
u
sA
bo
rt
i
ve Infect
io
n
o
f Cells w
i
th A
d
1
2
A
d
12 interacts wit
hh
uman ce
ll
sinapro
d
uctive in

f
ection
l
ea
d
ing to t
he
sy
nthesis o f a lar
g
enumberofpro
g
en
y
virions. In contrast, the replication o
f
Ad12 in hamster cells is completely blocked (Doerfler 1969, 1991), possibl
y
because onl
y
few copies of Ad12 DNA are capable of reachin
g
the nucleus of
the hamster c ells (D. Webb, M. Hösel, B. Schmitz, et al., submitted). However,
t
h
e expression o
f
t
h

eA
d
12 genome in
h
amster ce
ll
s appears to
b
e suppresse
d
in several steps of the normal replication cycle. Nevertheless, viral DNA ca
n
b
e trace
d
in t
h
enuc
l
eus o
f
t
h
ea
b
ortive
l
yin
f
ecte

dh
amster ce
ll
san
dl
imite
d
transcription of the earl
yg
enes of non-inte
g
rated Ad12 has been documented
(
Ortin et a
l
. 1976; D. We
bb
, M. Höse
l
,B.Sc
h
mitz, et a
l
., su
b
mitte
d)
.T
h
ere is

,
h
owever, no meth
y
lation of the free intranuclear Ad12 DNA in productivel
y
or abortively infected cells (Vardimon et al. 1980). Obvious ly, methylatio
n
o
ff
ree vira
l
DNA, even in t
h
ea
b
ortive system,
d
oesnotservetoregu
l
ate o
r
Function of DNA Meth
y
latio
n
14
5
t
o inactivate t

h
e
l
ate A
d
12 vira
l
genes t
h
at are not transcri
b
e
d
in a
b
ortive
ly
infected hamster cells.
3
.2
T
he Act
i
vel
y
Transcr
i
bed Genome of Frog V
i
rus 3 Is Completel

y
5

-
C
G-
3

-Met
hyl
ate
d
T
h
e
hy
permet
hyl
ate
d
state o
f
t
h
e virion-encapsi
d
ate
d
or o
f

t
h
e intrace
ll
u
l
ar
FV3 genomes (Wi
ll
is an
d
Grano
ff
1980) in
fi
s
h
or mamma
l
ian ce
ll
s
h
as taug
h
t
us t
h
at t
h

e
b
io
l
o
g
ica
l
si
g
ni
fi
cance o
f
DNA met
hyl
ationcannot
b
esc
h
ematica
lly
interpreted and depends entirel
y
on the biolo
g
ical s
y
stem studied. While the
inverse corre

l
ations
d
escri
b
e
d
a
b
ove
h
o
ld
true
f
or most systems investigate
d
so far, there are notable exce
p
tions to this “rule” and, of course, no such
stringent
d
ogmata in
b
io
l
ogy. It
h
as
b

een
d
ocumente
d
t
h
at t
h
evira
l
L1140 gen
e
is active
ly
transcri
b
e
dl
ate a
f
ter in
f
ection o
ffi
s
h
ce
ll
swit
h

FV3, a
l
t
h
ou
gh
it is
m
et
h
y
l
ate
d
in a
ll 5

-
CG
-
3

d
inuc
l
eoti
d
es (Munnes et a
l
. 1995). In FV3-in

f
ecte
d
fi
s
h
or
h
amster ce
ll
s, a trans
f
ecte
d
L1140 promoter-in
d
icator
g
ene construc
t
is active in the unmeth
y
lated or full
y
5

-CG
-3

-meth

y
lated form. When th
e
same construct is met
hyl
ate
d
on
ly
in t
h
e
5

-
CCGG
-3

(
HpaII
)
sequences, it
s
activit
y
is reduced. Compatibilit
y
of the meth
y
lation of an immediate-earl

y
FV3
p
romoter wit
h
itsactivetranscri
p
tion
h
as a
l
so
b
een re
p
orte
d
(T
h
om
p
so
n
et al. 1988). These data confirm the special meth
y
lation requirements of this
p
romoter inFV3 DNA. S
p
ecia

lp
ro
p
erties o
f
t
h
e FV3DNA-
p
rotein interaction
s
m
a
y
account fo r these unexpected activit
y
patterns. It would be interestin
g
t
o stud
y
in
g
reater biochemical detail the transcription of FV3
g
enes and the
p
roteins invo
l
ve

d
in t
h
eir re
g
u
l
ation.
4
Site-Specific Promoter Methylation and Gene Silencing
T
h
e
fi
n
d
in
g
o
f
inverse corre
l
ations
b
etween promoter activit
y
an
d
extent o
f

DNA meth
y
lation led to the concept that sequence-specific promoter meth
y
-
l
ations exert a regu
l
atory
f
unctionongeneactivity.Inor
d
er to provi
d
emore
d
irect evidence for this interpretation, we devised experiments in which
anum
b
er o
f
promoter-in
d
icator gene constructs were teste
df
or t
h
eir geneti
c
activities in the unmeth

y
lated or in the meth
y
lated state at 48 h after transfec-
t
ion into mammalian cells in culture. In
g
eneral, these data corroborated the
earlier interpretation of promot er inactivation b
y
pr omoter meth
y
lation, al
-
t
hou
g
h this experimental approach could, of course, not help decide whether
in an intact mamma
l
ian genome promo ter met
h
y
l
ation was t
h
e cause or con
-
sequence of promoter inactivation. The former possibilit
y

, however, remain
s
th
emore
l
i
k
e
l
y exp
l
anation
.
1
4
6
W.
D
oe
rfl
er
4
.1
The E2A Promoter of Ad2 DNA
In a first set of experiments, the ooc
y
te s
y
stem fro
m

Xeno
p
us laevi
s
was
adapted to test unmeth
y
lated or meth
y
lated promoter-
g
ene constructs for
g
enetic activities. T
h
ec
l
one
d
E2A re
g
ion o
f
A
d
2 DNA was t
h
en 5

-

CCGG
-3

-
meth
y
lated with the HpaII DMTases or was left unmeth
y
lated. Subsequentl
y
,
eit
h
er construct was microinjecte
d
into t
h
enuc
l
ei o
f
X
.
l
aevis oocytes. T
h
e
meth
y
lation status of these constructs was maintained in the ooc

y
te nuclei. At
4
8
h
a
f
ter microinjection, t
h
e unmet
h
y
l
ate
d
construct was transcri
b
e
d
in t
h
e
ooc
y
te nuc
l
ei, w
h
i
l

et
h
e met
hyl
ate
d
construct was si
l
ence
d
(Var
d
imon et a
l
.
1
982a). Transcri
p
tion was initiate
d
at t
h
eaut
h
entic E2A
l
ate
p
romoter o
f

A
d2
D
NA. Contro
l
constructs carr
y
in
g
t
h
e unmet
hyl
ate
dh
istone
h
22
g
ene were
activel
y
transcribed when co-in
j
ected with the meth
y
lated and silenced E2
A
construct. Hence, t
h

ere was no evi
d
ence
f
or possi
bl
e unspeci
fi
cin
h
i
b
itory e
f-
fects exerted b
y
the in vitro premeth
y
lated construct. Modification of the E2
A
construct
b
yt
h
eBsuRI(
5

-GG*CC
-3


) DNMT
d
i
d
not inactivate transcri
p
tio
n
(
Var
d
imon et a
l
. 1982
b
). T
h
ese
d
ata provi
d
e
dd
irect evi
d
ence
f
or t
h
enotion

t
h
at
5

-CG
-3

s
equence-speci
fi
cpromotermet
h
y
l
ation was invo
l
ve
d
in t
h
e
s
i
l
encin
g
o
f
eu

k
ar
y
otic
g
enes.
The s
y
stem was further refined b
y
separatin
g
the promoter of the E2A
g
ene
f
rom its
b
o
dy
an
dby
preparin
gb
ot
h
DNA
f
ra
g

ments in quantitativ
e
a
m
ou
n
ts.
W
et
h
e
n
5

-
CCGG
-3

-meth
y
lated either the promoter or the
g
ene
b
o
d
yparto
f
t
h

e constructs. Su
b
sequent
l
y, t
h
e met
h
y
l
ate
d
promoter was re
-
l
i
g
ate
d
to t
h
e unmet
hyl
ate
db
o
dy
o
f
t

h
e E2A
g
ene an
d
, converse
ly
,t
h
e un-
met
h
y
l
ate
d
promoter was reattac
h
e
d
to t
h
e met
h
y
l
ate
d
E2A gene sequence.
Upon microin

j
ection into t
h
enuc
l
ei o
f
X.
l
aevi
s
ooc
y
tes, on
ly
t
h
e construct,
in which the promoter had been meth
y
lated, was inactivated. The construc
t
wit
h
an unmet
hyl
ate
d
promoter
b

ut a met
hyl
ate
dg
ene
b
o
dy
was active
ly
tran
-
s
cribed (Lan
g
ner et al. 1984). We interpreted these data to demonstrate tha
t
s
equence-speci
fi
c promoter met
h
y
l
ation
l
e
d
to gene inactivation, w
h

ereas t
he
meth
y
lation of the bod
y
of this
g
ene did not affect its activit
y.
4.
2
The E1A Pr
o
m
o
ter
of
A
d
12 DN
A
A simi
l
ar set o
f
ex
p
eriments was
p

er
f
orme
d
wit
h
constructs t
h
at carrie
d
t
he
c
hl
oramp
h
enico
l
acet
yl
trans
f
erase (CAT)
g
ene as an in
d
icator
f
or
g

ene activ-
ity un
d
er t
h
econtro
l
o
f
t
h
e E1A regu
l
atory region o
f
A
d
12 DNA. Met
h
y
l
atio
n
o
f
t
h
e two HpaII (
5


-
CCGG
-3

)
or o
f
t
h
e seven H
h
aI
(
5

-GCGC
-3

)
sequence
s
in this promoter inactivated the CAT
g
ene or severel
y
decreased its activ-
ity at 48
h
a
f

ter t
h
etrans
f
ection o
f
t
h
eseconstructsintomouseLt
k
−
ce
lls
Function of DNA Meth
y
latio
n
14
7
(Krucze
k
an
d
Doer
fl
er 1982, 1983). Severa
l
a
dd
itiona

l
sites in t
h
e
p
romoter
o
f the E1A
g
ene of Ad12 were meth
y
lated, and the activit
y
of the modified
p
romoter was assessed with the CAT indicator
g
ene. The C-residue meth
y
la
-
t
ion o
f
two A
l
uI sites
(5

-AGC

T
-
3

)
d
ownstream
f
rom t
h
eTATA
b
ox
h
a
d
no
effect on promoter activit
y
. However, when one EcoRI (5

-GAATTC-
3

)se-
q
uence, 281
bp
u
p

stream, or o ne Ta
q
I(5

-T
CGA-3

)site
d
ownstream
f
rom t
h
e
TATA si
g
na
l
in t
h
e promoter was
d
eox
y
a
d
enosine met
hyl
ate
d

,t
h
epromote
r
b
ecame si
l
ent (Kne
b
e
l
an
d
Doer
fl
er 1986). Deoxya
d
enosine met
h
y
l
ation o
f
an
M
b
oI
(
5


-G
AT
C
-
3

) sequence
d
ownstream o
f
t
h
eTATAsi
g
na
lh
a
d
no e
ff
ect
.
A
pparentl
y
, meth
y
lated nucleotides introduced at hi
g
hl

y
specific promote
r
l
ocations can p
l
ay an important ro
l
eint
h
e
d
ownregu
l
ation o
f
t
h
eA
d
12 E1A
p
romoter at least in transfection ex
p
eriments. Since N
6
-
mA i
s
n

ot
kn
o
wn
to
o
ccur in mamma
l
ian DN A
,
t
h
ee
ff
ect o
f
N
6
-mAonpromoteractivity
h
as
b
een
unexpecte
d
.
In an extension o
f
t
h

is ex
p
erimenta
l
a
pp
roac
h
,a
dd
itiona
l
vira
l
an
d
non-
v
ira
l
eu
k
ar
y
otic promoters were teste
df
or t
h
eir sensiti vit
y

towar
d
s
5

-
CG
-3

o
r
5

-
CCGG
-
3

meth
y
lation. The CAT or luciferase
g
ene was used as activit
y
in
d
icator 24
h
a
f

ter t
h
e trans
f
ection into
d
i
ff
erent
h
uman ce
ll l
ines
(
HeLa,
P
A-1, 293). The meth
y
lation of all 5

-
CG
-
3

sequences b
y
the SssI DNMT
inactivate
d

t
h
e E2A
l
ate promoter o
f
A
d
2DNA,t
h
e
h
uman cytomega
l
oviru
s
p
romoter, t
h
e TNF
-
α
p
romoter, t
h
e
h
erpes simp
l
ex virus t

hy
mi
d
ine
k
inase
p
ro moter, an
dd
ecrease
d
t
h
eactivityo
f
t
h
e SV40 ear
l
y promoter (Muiznie
k
s
and Doerfler 1994a). In some experiments, HpaII meth
y
lation
j
ust led t
o
adecreaseinthe
g

enetic activit
y
of some of these constructs
.
4.3
T
he L1140 Promoter of Fro
g
Virus FV3 DN
A
The resistance of the late L1140 or an earl
y
FV3 promoter to complete
5

-
CG
-3

m
et
h
y
l
ation an
d
its
f
u
ll

activity in
fi
s
h
or mamma
l
ian ce
ll
sint
h
ecomp
l
ete
l
y
m
et
hyl
ate
d
state
h
as
b
een
d
escri
b
e
d

(T
h
ompson et a
l
. 1988; Munnes et a
l.
1995)
.
4
.
4
T
he p10 Promoter of the AcNPV Insect V
i
ru
s
A
construct, w
h
ic
h
containe
d
t
h
epromotero
f
t
h
e p10

g
ene o
f
t
h
e insect virus
Autograp
h
aca
l
i
f
ornic
a
nuc
l
ear po
l
y
h
e
d
rosis virus (AcNPV) an
d
t
h
eCAT
in
d
icator

g
ene, was active in AcNPV-in
f
ecte
d
S
p
o
d
o
p
tera
f
ru
g
i
p
er
da
i
nsec
t
ce
ll
sat
1
8
h
a
f

te
r
t
r
a
n
s
f
ect
i
o
n
o
f
t
h
eco
n
st
r
uct.
Wh
e
n
t
h
et
hr
ee 5


-
CCGG
-
3

(HpaII) sites in t
h
epromoteran
d
its
d
ownstream region were met
h
y
l
ate
d
,t
h
e
1
4
8
W.
D
oe
rfl
er
p10 gene promoter was si
l

ence
d
(Kne
b
e
l
et a
l
. 1985). A
l
t
h
oug
h
insect ce
ll
sma
y
contain onl
y
minor amounts of 5-mC, the activit
y
of an AcNPV insect viru
s
promoter could be shown to be sensitive to sequence-specific meth
y
lation
.
4
.

5
Human Alu Sequences Transcr
i
bed b
y
RNA Pol
y
merase III
We
h
ave a
l
so
d
emonstrate
d
t
h
at t
h
epo
ly
merase III transcription o
f
A
l
use
-
quences associated with the human an
g

io
g
enin, the tissue plasmino
g
en acti-
vator (tPA), or t
he
α
1
-g
l
o
b
in gene is in
h
i
b
ite
db
y
5

-CG
-3

m
et
h
y
l

ation o
f
t
h
es
e
s
equences (Koc
h
ane
k
et a
l
. 1993). T
h
eir met
hyl
ation a
l
so inter
f
eres wit
h
t
h
e
b
in
d
ing o

f
proteins to t
h
eBcontro
l
region o
f
t
h
ese A
l
u sequences (Koc
h
ane
k
et a
l
. 1995
)
.
4
.6
Bending of Promoter DNA Sequences Due to Methylation
?
The site-specific meth
y
lation in
5

-

CCGG
-3

(H
p
aII),
5

-CGCG
-3

(
FnuDII)
,
or in 5

-CG
-3

(
SssI) sequences o
f
t
h
e E2A promoter, t
h
epo
l
ymerase III
-

transcri
b
e
d
virus-associate
d
RNA I (VAI)
g
ene o
f
A
d
2DNAoro
f
t
h
e
h
uma
n
angiogenin gene-associate
d
A
l
u sequence can a
l
ter t
h
ee
l

ectrop
h
oretic mo-
b
i
l
it
y
o
f
t
h
ese DNA sequences in non-
d
enaturin
g
po
ly
acr
yl
ami
d
e
g
e
l
s. T
h
i
s

findin
g
indicates that the bendin
g
of the tested sequences mi
g
ht be altered b
y
D
NA met
h
y
l
ation (Muiznie
k
san
d
Doer
fl
er 1994
b
)
.
5
A
n Adenov
i
rus E1A Gene Product or the Stron
g
E

nhancer of Human Cytomegalovirus Can Overcome
the Transcr
i
pt
i
on-Inact
i
vat
i
ng Effect of Promoter Meth
y
lat
i
on
The removal of the meth
y
l
g
roup from 5-mC in a meth
y
lated promoter in the
a
b
sence o
f
DNA re
pl
ication seems to
b
e a rare event. Hence, ot

h
er mec
h
anisms
for transient reactivation of a permanentl
y
meth
y
lated promoter appear t
o
b
e require
d
.O
f
course, experimenta
ll
y, t
h
e met
h
y
l
ate
d
E2A
l
ate promoter in
t
h

eA
d
2-trans
f
orme
d
ce
ll l
ine HE3 can
b
e
d
emet
hyl
ate
d
an
d
reactivate
dby
g
rowing t
h
ece
ll
sincu
l
ture in t
h
e presence o

f
50 µM 5-azacyti
d
ine (5-aza-C)
,
an in
h
i
b
itor o
f
maintenance met
hyl
ation (Knust et a
l
. 1989). T
h
is approac
h
provides sup port of principle but does not adequatel
y
mimic the situation i
n
a
b
io
l
ogica
l
system.

Function of DNA Meth
y
latio
n
14
9
In
h
uman 293 ce
ll
s, w
h
ic
h
carry t
h
e
l
e
f
t terminus o
f
A
d
5DNAc
h
romoso
-
m
all

y
inte
g
rated and expre ss the E1 re
g
ion of Ad5 constitutivel
y
, the inacti
-
v
atin
g
effect of 5

-
CCGG
-
3

meth
y
lation of an E2A promoter construct of Ad2
DN A is re
l
ease
d
or mar
k
e
dl

y
d
ecrease
d
(Langner et a
l
. 1986). We
h
ave a
l
so
shown that the E1A
g
ene encodin
g
the 13S RNA and the 289-amino acid (aa
)
p
rotein o
f
A
d
2, a we
ll
-
k
no wn transactivator o
f
genes (F
l

int an
d
S
h
en
k
1989;
Nevins et a
l
. 1995), is responsi
bl
e
f
or t
h
e reversa
l
o
f
t
h
e inactivatin
g
e
ff
ect o
f
E2A promoter met
h
y

l
ation (Weiss
h
aar et a
l
. 1988). It is un
k
nown
b
yw
h
ic
h
m
ec
h
anism t
h
e 289-aa E1A
f
unction is capa
bl
eo
f
e
ff
ectin
g
t
h

is reactivation.
The meth
y
lated E2A promoter did not lose its
5

-CCGG
-3

m
eth
y
l
g
roups i
n
th
e reactivation
p
rocess at 48
h
a
f
ter trans
f
ection. Moreover, t
h
eaut
h
entic

c
ap site of this promoter was used in the transcription followin
g
reactivatio
n
(Weiss
h
aar et a
l
. 1988; Knust et a
l
. 1989).
S
imi
l
ar
ly
,t
h
e5

-
CCGG
-
3

m
et
hyl
ate

d
E2A promoter o
f
A
d
2DNAwasactive
wh
en t
h
e strong imme
d
iate ear
l
yen
h
ancer o
f
HCMV DNA was inserte
d
into t
h
e promoter-in
d
icator
g
ene construct in a position eit
h
er imme
d
iate

ly
antecedent to the
p
romoter o r several thousand n ucleotides remote from i
t
(Kne
b
e
l
-Mörs
d
or
f
et a
l
. 1988). Transcription was initiate
d
correct
ly
at t
he
authentic cap site of the E2A
g
ene, and
5

-
CCGG
-3


meth
y
lation remained
u
na
l
tere
d
at
l
east
d
uring t
h
e
d
uration o
f
t
h
e transient expression experiment.
5.
1
Promoter Methylation and Protein Bindin
g
This topic has been extensivel
y
investi
g
ated in several laboratories (addressed

in man
y
of the chapters of this volume). In the E2A promoter s
y
stem of A d
2
DN A, t
h
e in vitro met
h
y
l
ation o
f5

-
CCGG
-
3

s
e
q
uences at nuc
l
eoti
d
es +24, +6,
and −215 relative to nucleotide +1, the site of transcri
p

tional initiation, was
d
emonstrate
d
to
l
ea
d
to transcri
p
tiona
l
inactivation in transient ex
p
ressio
n
stud
i
es
in
X. laevis
ooc
y
tes (Lan
g
ner et al. 1986), in mammalian cells (Lan
g
ne
r
et al. 1986), after the

g
enomic fixation of the promoter in mammalian cell
s
(Müller and Doerfler 1987), and in a cell-free transcription s
y
stem usin
g
nuclear extracts from human HeLa cells (Do brzanski et al. 1988). DNA fra
g-
m
ents 50 or 73
b
pin
l
engt
h
—w
h
ic
h
comprise t
h
e+24an
d
+6
5

-
CCGG
-

3

sequences of the E2A promoter of Ad2 DNA in the unmeth
y
lated, meth
y
-
l
ate
d
,or
h
emimet
h
y
l
ate
d
state—were incu
b
ate
d
wit
h
part
l
y puri
fi
e
d

nuc
l
ear
extracts
f
rom
h
uman HeLa ce
ll
s. Protein
b
in
d
in
g
to t
h
ese DNA preparations
w
as assesse
db
ye
l
ectrop
h
oretic mo
b
i
l
ity s

h
i
f
t assays (EMSAs). T
h
e
f
ormation
of
one o
f
t
h
eo
b
serv e
d
DN A-protein com p
l
exes in t
h
is s
y
stem was compro-
m
ised when the construct was meth
y
lated or hemimeth
y
lated (Hermann e

t
a
l
. 1989). T
h
eresu
l
ts o
f
t
h
e necessary competition experiments con
fi
rme
d
t
h
e
150
W.
D
oe
rfl
er
interpretation t
h
at speci
fi
c promoter met
h

y
l
ation inter
f
ere
d
wit
h
t
h
e
b
in
d
in
g
of nuclear prot eins from human cells. There was evidence that the AP2 tran
-
s
criptio n factor was amon
g
the proteins sensitive to promoter meth
y
lation in
t
h
is system (Hermann an
d
Doer
fl

er 1991)
.
6
P
atterns of DNA Methylation in the Human Genome
Amorepro
f
oun
d
un
d
erstan
d
in
g
o
f
t
h
emu
l
ti
f
acete
db
io
l
o
g
ica

lf
unction
s
of DNA meth
y
lation in mammalian and other
g
enomes will remain elusiv
e
un
l
ess we
h
ave at
h
an
d
t
h
ecomp
l
ete nuc
l
eoti
d
e sequences o
f
t
h
ese

g
enomes
includin
g
the fifth nucleotide. Researchers interested in the function of 5-
mC
h
ave, t
h
ere
f
ore,
b
een
d
isappointe
db
yt
h
eot
h
erwise a
d
mira
bl
eresu
l
t
s
o

f
t
h
e Human Genome Pro
j
ect. T
h
e
h
uman epi
g
enome pro
j
ect
h
as
b
ee
n
initiate
d
,an
d
its resu
l
ts, once at
l
east part
l
ycomp

l
ete
d
,wi
ll
un
d
ou
b
te
dl
y
fill
a serious
g
ap in t
h
e anatom
y
o
f
t
h
e
h
uman
g
enome. In t
h
e ear

ly
1990s, m
y
l
aborator
y
be
g
an, as a pilot pro
j
ect as it were, to stud
y
DNA meth
y
lation
patterns in v arious parts o
f
t
h
e
h
uman
g
enome. A part o
f
t
h
ese resu
l
ts

h
as
been summarized (Doerfler 2000). A more complete surve
y
will be presented
h
ere. Our stu
d
ies a
l
so
h
a
d
t
h
e aim o
f
contri
b
uting to t
h
eun
d
erstan
d
ing o
f
epi
g

enetic mechanisms and of human disease
.
6.1
Inter
i
nd
i
v
i
dual Concordance
i
n Human DNA Methylat
i
on Pattern
s
We
h
ave as
k
e
d
t
h
e question o
fh
ow tig
h
t
l
y preserve

d
patterns o
f
DNA met
h
y-
l
ation actuall
y
are in the promoter and
5

-upstream re
g
ions of a human
g
ene
among severa
l
in
d
ivi
d
ua
l
so
fd
i
ff
erent et

h
nic origins
.
The human
g
enome, like man
y
other eukar
y
otic
g
enomes, is character-
ized b
y
the existence of complex patterns of DNA meth
y
lation that reflect,
in an unidentified wa
y
, sta tes of
g
ene activit
y
and inactivit
y
and, equall
y
important and related, the chromosomal structure of the (human)
g
enome

.
T
h
e5

-
upstream an
d
promoter regions o
f
t
h
e
h
uman genes
f
or TNF
-
α
an
d
TNF
-
β
w
ere investi
g
ated with the bisulfite sequencin
g
technique (Frommer

et a
l
. 1992; C
l
ar
k
et a
l
. 1994)
f
or t
h
e
p
resence o
f
5-mC r esi
d
ues (Koc
h
ane
k
et a
l
. 1990, 1991). Human DNA was
d
erive
df
rom perip
h

era
lbl
oo
dg
ran
-
u
l
ocytes,
l
ymp
h
ocytes, or
f
rom sperm. T
h
eresu
l
ts in
d
icate
d
t
h
at patterns
o
f
DNA met
hyl
ation, at

l
east in t
h
ese
g
enome se
g
ments, were interin
d
ivi
d
-
uall
y
hi
g
hl
y
conserved. Thus, in the TNF
-
α
D
NA from
g
ranuloc
y
tesof15
i
n
d

ivi
d
ua
l
so
f
A
f
rican, Caucasian, or C
h
inese origin, t
h
e 5-mC resi
d
ues wer
e
Function of DNA Meth
y
latio
n
1
51
consistent
l
y
f
oun
d
in 5


-CG
-3

d
inuc
l
eoti
d
e
p
ositions minus IX, minus X, mi-
nus XI (upstream of the cap site), and in position plus XVI (downstream o
f
it). Ver
y
different distributions of 5-mC residues were observed in human cell
l
ines HL60,
J
ur
k
at, an
d
RPMI 1788. T
h
e TNF-
α
g
ene is transcri
b

e
d
in
h
uma
n
g
ranuloc
y
tes
.
Avery
d
i
ff
erent resu
l
t emerge
df
or t
h
epromoteran
d
upstream region
s
o
f the human
g
ene for TNF-
β

th
at is not transcri
b
e
d
in
h
uman
g
ranu
l
o
-
cy tes. A
ll
13
5

-
CG
-
3

d
inuc
l
eoti
d
es in t
h

is segment were met
h
y
l
ate
d
, two on
l
y
h
emimet
hyl
ate
d
.A
g
ain, t
h
is pattern
h
e
ld
true in t
h
e
g
ranu
l
oc
y

tes
f
rom nin
e
d
ifferent individuals. The same sequence was completel
y
unmeth
y
lated i
n
h
uman
l
ymp
h
ocy tes
f
rom t
h
e same in
d
ivi
d
ua
l
s, in sperm an
d
in t
h

e
h
uman
cell lines RPMI 1788 and HL60, but almost completel
y
meth
y
lated in cell lin
e
J
ur
k
at (Koc
h
ane
k
et a
l
. 1990)
.
These data document that meth
y
lation patterns in human DNA can be ver
y
d
i
ff
erent in
d
i

ff
erent ce
ll l
ines,
b
ut can
b
einterin
d
ivi
d
ua
ll
y
h
ig
hl
yconcor
d
ant.
Moreover, patterns of DNA meth
y
lation in specific
g
enome se
g
ments can var
y
a
g

reat deal
.
The patterns of DNA meth
y
lation in the human TNF
-
α
an
d
TNF
-
β
g
ene
s
in
g
ranuloc
y
tes, monoc
y
tes and in several cases of acute (AML) or chronic
m
ye
l
oi
dl
eu
k
emia (CML) were

f
oun
d
to
b
e very simi
l
ar, except
f
or one AML
,
in which the re
g
ion in the TNF
-
α
g
ene was comp
l
ete
ly
unmet
hyl
ate
d
,an
d
severa
ll
eu

k
emia cases in w
h
ic
h
many sites in t
h
e TNF
-
β
g
ene were on
ly
h
emimet
hyl
ate
d
(Koc
h
ane
k
et a
l
. 1991). I n T an
d
B
ly
mp
h

oc
y
tes o
f
man
y
individuals and in a number of Hod
g
kin and non-Hod
g
kin l
y
mphomas, bot
h
th
e TNF-
α
an
d
TNF
-
β
g
enes were un- or
hy
pomet
hyl
ate
d
.T

h
eDNAinHeL
a
cells in culture was completel
y
meth
y
lated in the upstream and promoter re-
g
ions o
fb
ot
h
genes (Koc
h
ane
k
et a
l
. 1991). I
fl
eu
k
emia- or
l
ymp
h
oma-speci
fi
c

p
atterns should exist, the
y
are ver
y
complex and not readil
y
reco
g
nizable b
y
th
is type o
f
ana
l
ysis.
We
h
ave a
l
so compare
d
met
hyl
ation patterns
by
HpaII (
5


-
CCGG
-3

)
an
d
HhaI (
5

-
GCGC
-3

) cleava
g
es of human DNA from European and Japanes
e
in
d
ivi
d
ua
l
s across a
b
out 500
kb
o
f

ran
d
om
ly
se
l
ecte
d
DNA sequences in t
h
e
human
g
enome and found complete interindividual con
g
ruence of patterns
b
yt
h
is met
h
o
d
o
f
a
d
mitte
dl
y interme

d
iate sensitivity (Be
h
n-Krappa et a
l.
1991)
.
6.
2
M
eth
y
lat
i
on Patterns
i
n Genet
i
call
y
Impr
i
nted Reg
i
ons of the Human Genome
The Prader-Willi/An
g
elman re
g
ion on chromosome 15q11-q13 of the huma

n
g
enome is genetica
ll
yimprinte
d
, i.e., on t
h
e materna
ll
yan
d
on t
h
e pat erna
ll
y