A icle
Accu a e Recycling o Pa en al His ones
Rep oduces he His one Modi ica ion Landscape
du ing DNA Replica ion
G aphical Abs ac
Highligh s
dChOR-seq de e mines his one PTM occupancy on newly
eplica ed DNA
dHis one PTM posi ional in o ma ion is p ese ed h ough
pa en al his one ecycling
dPa en al H3K27me3 domains a e s able and inhe i ed o
daugh e cells
dRes o a ion o his one PTM le els ollows ma k- and locus-
speci ic kine ics
Au ho s
Naza e Re e o
´n-Go
´mez,
C is ina Gonza
´lez-Aguile a,
Ka hleen R. S ewa -Mo gan, ...,
Janus Schou Jakobsen,
Cons ance Alabe , Anja G o h
Co espondence
anja.g o h@b ic.ku.dk
In B ie
His one modi ica ions a e a co e
componen o he epigenome. Re e o
´n-
Go
´mez e al. de elop ChOR-seq o p o ile
his one modi ica ions a e DNA
eplica ion and ind ha he genomic
localiza ion o modi ied pa en al his ones
is p ese ed on daugh e s ands while
new his one modi ica ion o es o e p e-
eplica ion le els ollows ma k- and
locus-speci ic kine ics.
Re e o
´n-Go
´mez e al., 2018, Molecula Cell 72, 239–249
Oc obe 18, 2018 ª2018 The Au ho s. Published by Else ie Inc.
h ps://doi.o g/10.1016/j.molcel.2018.08.010
Molecula Cell
A icle
Accu a e Recycling o Pa en al His ones
Rep oduces he His one Modi ica ion Landscape
du ing DNA Replica ion
Naza e Re e o
´n-Go
´mez,
1,2,6
C is ina Gonza
´lez-Aguile a,
1,3,6
Ka hleen R. S ewa -Mo gan,
1,2,6
Na aliya Pe yk,
1,2
Valen in Flu y,
1,2
Simona G aziano,
1,2
Jens Vils up Johansen,
1
Janus Schou Jakobsen,
1,4
Cons ance Alabe ,
1,5
and Anja G o h
1,2,7,
*
1
Bio ech Resea ch and Inno a ion Cen e (BRIC), Uni e si y o Copenhagen, Facul y o Heal h and Medical Sciences,
Uni e si y o Copenhagen, 2200 Copenhagen, Denma k
2
The No o No disk Cen e o P o ein Resea ch (CPR), Uni e si y o Copenhagen, Facul y o Heal h and Medical Sciences,
Uni e si y o Copenhagen, 2200 Copenhagen, Denma k
3
P esen add ess: Andalusian Cen e o Molecula Biology and Regene a i e Medicine (CABIMER), CSIC/Uni e si y o
Se ille, Se ille 41092, Spain
4
P esen add ess: Symphogen A/S, 2750 Balle up, Denma k
5
P esen add ess: Cen e o Gene Regula ion and Exp ession, School o Li e Sciences, Uni e si y o Dundee, Dundee DD1 5EH, UK
6
These au ho s con ibu ed equally
7
Lead Con ac
*Co espondence: [email p o ec ed]k
h ps://doi.o g/10.1016/j.molcel.2018.08.010
SUMMARY
Ch oma in o ganiza ion is dis up ed genome-wide
du ing DNA eplica ion. On newly syn hesized DNA,
nucleosomes a e assembled om new nai e his-
ones and old modi ied his ones. I emains unknown
whe he he landscape o his one pos - ansla ional
modi ica ions (PTMs) is ai h ully copied du ing
DNA eplica ion o he epigenome is pe u bed.
He e we de elop ch oma in occupancy a e eplica-
ion (ChOR-seq) o de e mine his one PTM occu-
pancy immedia ely a e DNA eplica ion and ac oss
he cell cycle. We show ha H3K4me3, H3K36me3,
H3K79me3, and H3K27me3 posi ional in o ma ion
is ep oduced wi h high accu acy on newly syn he-
sized DNA h ough his one ecycling. Quan i a i e
ChOR-seq e eals ha de no o me hyla ion o
es o e H3K4me3 and H3K27me3 le els occu s
ac oss he cell cycle wi h ma k- and locus-speci ic
kine ics. Collec i ely, his demons a es ha accu-
a e pa en al his one ecycling p ese es posi ional
in o ma ion and allows PTM ansmission o
daugh e cells while modi ica ion o new his ones
gi es ise o complex epigenome luc ua ions ac oss
he cell cycle ha could unde lie cell- o-cell he e o-
genei y.
INTRODUCTION
The o ganiza ion o euka yo ic genomes in o ch oma in in lu-
ences all DNA-based p ocesses, including gene exp ession
and DNA epai . Ch oma in o ganiza ion is pa icula ly impo an
o es ablishing and main aining cell- ype-speci ic ansc ip-
ional p og ams and hus unde lies epigene ic cell memo y (Allis
and Jenuwein, 2016; Halley-S o and Gu don, 2013). Howe e ,
he basic mechanisms ha ensu e p opaga ion o ch oma in
s a es du ing DNA eplica ion and ac oss cell di ision emain un-
clea (Alabe and G o h, 2012; Allis and Jenuwein, 2016; Al-
mouzni and Ceda , 2016).
The nucleosome is he basic uni o ch oma in, in which 146
base pai s o DNA a e w apped a ound a his one co e composed
o a cen al his one H3-H4 e ame lanked by wo his one
H2A-H2B dime s. His ones a e deco a ed wi h a la ge a ie y
o pos - ansla ional modi ica ions (PTMs) ha con ibu e o
he es ablishmen and main enance o ac i e and ep essed
ch oma in s a es (Pa el and Wang, 2013). Many o hese egula-
o y modi ica ions a e ound on his one H3. His one H3 lysine
4 i-me hyla ion (H3K4me3), his one H3 lysine 36 i-me hyla ion
(H3K36me3), and his one H3 lysine 79 i-me hyla ion
(H3K79me3) ma k ac i e ch oma in, wi h H3K4me3 en iched in
p omo e egions and H3K36me3 and H3K79me3 en iched in
gene bodies (Rando, 2007). Con e sely, i-me hyla ion o his-
one H3 lysine 27 (H3K27me3) dema ca es la ge , ansc ip ion-
ally silen domains (Schue eng ube e al., 2017).
His one modi ica ions associa ed wi h bo h ansc ip ional
silencing and ac i a ion a e p oposed o play a cen al ole in
epigene ic cell memo y (Allis and Jenuwein, 2016; Campos
e al., 2014; Halley-S o and Gu don, 2013), implying ha
his one-based in o ma ion mus be ans e ed o daugh e cells
du ing mi o ic cell di ision. Howe e , he p ocess o DNA eplica-
ion is dis up i e and leads o he disassembly o nucleosomes
in o H3-H4 e ame s and H2A-H2B dime s (Jackson, 1987,
1988, 1990; Xu e al., 2010). Elec on mic oscopy and in i o
eplica ion o simian i us 40 (SV40) mini-ch omosomes ha e
shown ha 1–2 pa en al nucleosomes a e des abilized ahead
o he eplica ion o k (Gasse e al., 1996; McKnigh and Mille ,
1977) and ha his ones a e eleased om DNA, bu kep in close
Molecula Cell 72, 239–249, Oc obe 18, 2018 ª2018 The Au ho s. Published by Else ie Inc. 239
This is an open access a icle unde he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/).
p oximi y, du ing o k passage (G uss e al., 1993; Madamba
e al., 2017). Sho ly a e o k passage, nucleosome densi y is
es o ed on he wo daugh e DNA s ands h ough a combina-
ion o e-deposi ion ( ecycling) o old his ones and inco po a ion
o newly syn hesized his ones (Alabe and G o h, 2012;
Almouzni and Ceda , 2016; Annunzia o, 2015). Recen nucleo-
some mapping expe imen s in D osophila S2 cells and yeas
ha e e ealed ha nucleosome occupancy is inc eased a ound
ac i e p omo e and enhance egions sho ly a e DNA eplica-
ion (Fennessy and Owen-Hughes, 2016; Ramachand an and
Heniko , 2016; Vasseu e al., 2016), bu i is unclea whe he
his e lec s new his one inco po a ion o dispe sal o posi ioned
pa en al nucleosomes.
A cen al ques ion in epigene ics is he e o e how genome-
wide ch oma in dis up ion du ing DNA eplica ion migh be
compa ible wi h inhe i ance o genomic his one modi ica ion pa -
e ns o daugh e cells. Quan i a i e p o eomic analyses o new
and old his ones in human cells ha e shown ha old and new his-
ones (H3, H4, H2A, and H2B) a e mixed in a 1:1 a io on newly
eplica ed DNA (Alabe e al., 2015) and ha old his one H3-H4
a e ecycled wi h hei modi ica ions (Alabe e al., 2015; Pesa-
en o e al., 2008; Scha e al., 2009; Xu e al., 2011; Zee e al.,
2012). Old his one H3-H4 dime s do no mix wi h new ones (Jack-
son, 1987, 1990; Xu e al., 2010), a guing ha in ac , old H3-H4
e ame s wi h hei PTM in o ma ion a e ans e ed on o newly
syn hesized DNA. Howe e , i is no known how p ecisely old his-
ones a e e-inco po a ed on he new daugh e DNA s ands ela-
i e o hei o me genomic posi ion o whe he his ones and hei
associa ed ma ks a e dispe sed du ing DNA eplica ion. This is
pa icula ly impo an because modi ied pa en al his ones may
di ec modi ying enzymes owa d new his ones in hei icini y
(Aude gon e al., 2015; B own e al., 2017; Pa el and Wang,
2013; Raguna han e al., 2015) and allos e ic egula ion o he
PRC2 complex by H3K27me3 acili a es a posi i e eed o wa d
loop (Jiao and Liu, 2015; Ma gue on e al., 2009). Ma hema ical
modeling has es ima ed ha old his ones a e einco po a ed
wi hin 400 bp o hei o iginal genomic loca ion in yeas (Rad-
man-Li aja e al., 2011). Howe e , he e-deposi ion o pa en al
his ones has no been acked di ec ly.
To unde s and how he PTM landscape is duplica ed du ing
DNA eplica ion, i is necessa y o elucida e whe e and when
modi ied his ones a e deposi ed wi hin a gi en genomic locus
pos eplica ion. We ha e de eloped a echnology o analyze
ch oma in occupancy a e DNA eplica ion by nex -gene a ion
sequencing, e med ChOR-seq. ChOR-seq can ack he occu-
pancy o p o eins and his one PTMs a e eplica ion o k pas-
sage genome-wide. Gi en ha newly syn hesized his ones a e
de oid o i-me hyla ion ma ks a he ime o deposi ion (Alabe
e al., 2015; Ba -Zi e al., 2016; Jasencako a e al., 2010; Loyola
e al., 2006), ChOR-seq p o ides a means o ack ecycling o
old modi ied his ones and o di ec ly measu e po en ial eplica-
ion-dependen displacemen o p e-exis ing his one PTMs.
Using ChOR-seq o ack H3K4me3, H3K36me3, H3K79me3,
and H3K27me3, we ind ha PTM occupancy pa e ns a e ep o-
duced on newly eplica ed DNA wi h high accu acy in bo h
ep essed and ac i e genomic egions, demons a ing ha he
posi ional in o ma ion o his one ma ks is ai h ully inhe i ed o
daugh e s ands du ing DNA eplica ion. We hen ack es o a-
ion o H3K4me3 and H3K27me3 le els by quan i a i e ChOR-
seq ime cou se analysis and ind ha de no o his one me hyl-
a ion a e DNA eplica ion inc eases he le el o he ma ks wi hin
egions al eady dema ca ed by modi ied pa en al his ones.
No ably, we ind ha es o a ion o his one PTM le els ollows
ma k- and locus-speci ic kine ics, a guing ha he epigenome
is unde going complex changes ac oss he cell cycle ha could
unde lie cell- o-cell he e ogenei y.
RESULTS
ChOR-Seq T acks P o ein and PTM Occupancy on
Replica ed DNA
ChOR-seq is based on sho (10–20 min) pulse labeling o epli-
ca ed DNA wi h a nucleo ide analog (EdU) ollowed by sequen ial
ch oma in immunop ecipi a ion (ChIP) o a speci ic his one PTM.
Labeled DNA is hen bio inyla ed ia Click-IT and isola ed by
bio in-s ep a idin pull-down p io o analysis by nex -gene a-
ion sequencing (Figu e 1A). To in es iga e his one modi ica ion
pa e ns a e DNA eplica ion, we i s pe o med ChOR-seq
expe imen s o H3K27me3 and H3K4me3. Since H3K27me3
and H3K4me3 a e ma ke s o ep essed and ac i e ch oma in,
espec i ely, his app oach allowed us o assess he ChOR-
seq me hod in dis inc egions o he genome. We also
pe o med ChOR-seq o pan-his one H3 o ack o e all nucleo-
some occupancy. To in o m on p e- eplica ion his one PTM po-
si ion, we used S phase synch onized HeLa S3 and ca ied ou
s anda d ChIP-seq o H3K4me3 and H3K27me3 in o al ch o-
ma in p io o DNA labeling (pa en al ChIP) (Figu e 1B; Fig-
u e S1A). These H3K4me3 and H3K27me3 en ichmen p o iles
om S-phase-synch onized cells we e la gely iden ical o
genome-wide maps o H3K4me3 and H3K27me3 in asynch o-
nous HeLa S3 cells a ailable om ENCODE (Be ns ein e al.,
2005)(Figu e S1B), con i ming ha pa en al ChIP-seq is a sui -
able baseline o assessing ou ChOR-seq da a. Ou synch oni-
za ion se up also allowed us o e i y he speci ic isola ion o
eplica ed DNA by compa ison o eplica ion iming da a a ail-
able o HeLa S3 cells (Figu es S1C and S1D) (ENCODE P ojec
Conso ium, 2012).To op imize co e age o ansc ip ionally
ac i e and ep essed loci, we labeled eplica ing DNA in ea ly
S phase and mid S phase, espec i ely, co esponding o he
eplica ion iming o hese egions (Comoglio and Pa o, 2014;
Julienne e al., 2013; Pope e al., 2014). We ha es ed samples
o ChOR-seq immedia ely a e EdU pulse labeling (nascen
ch oma in) and a selec ed la e ime poin s o ack ch oma in
ma u a ion (ma u e ch oma in) (Figu e 1B; Figu e S1A). Finally,
o allow la e quan i a i e analyses o his one PTM le els
du ing ma u a ion, we spiked in EdU-labeled ch oma in om
D osophila S2 cells (Figu e 1A) (Bonhou e e al., 2014).
ChOR-seq p o iles o pan-H3 showed a high co ela ion
( = 0.86) wi h eplica ed DNA p o iles (EdU pull-downs) (Fig-
u e 1C; Figu e S1E). This was expec ed due o he apid es o a-
ion o nucleosome occupancy on newly eplica ed DNA
(Annunzia o, 2015; McKnigh and Mille , 1977) and p o ided
con i ma ion ha ChOR-seq de e mines occupancy speci ically
on eplica ed DNA. Fu he con i ma ions o ChOR-seq speci-
ici y we e ins ances in which egions o pa en al ChIP-seq
en ichmen lacked ChOR-seq signal owing o insu icien
240 Molecula Cell 72, 239–249, Oc obe 18, 2018
eplica ion o he locus a he ime o EdU labeling (Figu es 1C
and 1D; Figu e S1F). Ou synch oniza ion app oach o ChOR-
seq cap u ed app oxima ely 70% and 77% o he pa en al
H3K4me3- and H3K27me3-en iched egions, espec i ely (Fig-
u e 1E). Impo an ly, 94% o H3K27me3 and 92% o H3K4me3
loci in eplica ed egions we e also iden i ied by ChOR-seq (Fig-
u e 1E). This was also ue when newly eplica ed DNA was
labeled di ec ly by bio in-dUTP ins ead o EdU coupled wi h
Click-IT chemis y, and he spiked-in D osophila ch oma in
was omi ed (Figu es S1G and S1H). We we e he e o e con i-
den ha ChOR-seq was a obus and sensi i e me hod ha
could di ec ly assess his one PTM occupancy on eplica ed
DNA genome-wide.
The His one Modi ica ion Landscape Is Accu a ely
Rep oduced on Newly Syn hesized DNA
To add ess how accu a ely his one PTM p o iles a e copied
du ing eplica ion, we compa ed occupancy pa e ns o
ou modi ica ions—H3K27me3, H3K4me3, H3K36me3, and
A
ChOR-seq
EdU labelling
Ch oma in
C oss-linking
D. melanogas e
EdU labelled
ch oma in
(spike-in)
ChIP
Click-i
S ep a idin
pull-down
High- h oughpu
sequencing
B
C
D
DNA
labelling
(EdU)
Nascen
ChOR-seq
Pa en al
ChIP
Release
in o S phase
HeLa S3
H3
H3K27me3
ChIP-seq
Replica ed
DNA(Mid S)
H3
H3K27me3
ChOR-seq
2Mb
Ma u e Ch oma in
ch 1:151,495,060-165,790,665
200kb
H3K4me3
(ChIP-seq)
H3K4me3
(ChOR-seq)
ch 19-7,047,900-8,713,217
E
100
0
0
0
0
0
100
100
100
100
0
0
0
100
100
100
0
20
40
60
80
100
H3K27me3H3K4me3
Synch oniza ion
co e age
% o Pa en al peaks in
eplica ed egions
Pe cen age o Max.
0
20
40
60
80
100
H3K27me3H3K4me3
ChOR-seq
co e age
% o ChOR-seq peaks in
pa en al eplica ed egions
Replica ed
DNA(Ea ly S)
Pe cen age o Max.
Figu e 1. T acking His one PTM Occupancy
a e DNA Replica ion wi h ChOR-Seq
(A) O e iew o he ChOR-seq p o ocol.
(B) Expe imen al se up. HeLa S3 cells we e
eleased in o S phase om a hymidine block.
Pa en al and nascen ch oma in we e collec ed 1 h
be o e o immedia ely a e EdU labeling, espec-
i ely. The EdU label was hen chased and ma u e
ch oma in ha es ed a selec ed ime poin s along
he cell cycle.
(C and D) Pa en al ChIP-seq and nascen ChOR-
seq p o iles o pan-H3 and H3K27me3 (C) and
H3K4me3 (D). Replica ed DNA p o iles a e shown in
blue. Signal is scaled as pe cen age o maximum a
he locus depic ed.
(E) Ba plo s showing he synch oniza ion
co e age (le ) and ChOR-seq co e age ( igh ) in
he H3K4me3 and H3K27me3 da ase s. Pe cen -
age is calcula ed om peaks subse ed in o 500 bp
non-o e lapping windows.
See also Figu e S1.
H3K79me3—in p e- eplica i e and
nascen ch oma in by pa en al ChIP-seq
and ChOR-seq, espec i ely. Locally, we
obse ed ha his one modi ica ion pa -
e ns we e p ese ed du ing eplica ion
(Figu e 2A; Figu e S2A). Plo ing a e aged
signal o e si es o expec ed en ichmen
o each ma k con i med ha his posi-
ion p ese a ion occu ed genome-wide
(Figu e 2B; Figu e S2B). Hea maps o
signal o e expec ed si es o en ichmen
e ealed ha his held ue o all le els
o PTM en ichmen (Figu e 2C). Pa sing
H3K4me3 egions by exp ession le el
also showed ha he accu acy o pa en al
his one deposi ion was una ec ed by
pa en al PTM le els (Figu e S2C). Blu ing
o PTM occupancy a si es o expec ed
en ichmen would ha e indica ed dispe sal o pa en al his ones
du ing DNA eplica ion. The a e age p o iles o pa en al and
nascen PTM signals did no show any indica ion o blu ing o
eplica ion-dependen dispe sal o his one PTMs. We u he
de e mined he mean di e ence in localiza ion be ween nascen
and pa en al H3K4me3 peaks a indi idual loci o be app oxi-
ma ely 170 bp (Figu e S2D). This is below he esolu ion o ou
ChOR-seq analysis gi en by an a e age DNA agmen size o
250 bp (Figu e S2E). We hus conclude ha pa en al his ones
deco a ed wi h PTMs a e e-inco po a ed in o eplica ed DNA
wi hin 250 bp o hei p e- eplica ion posi ion.
H3K4me3 Is Res o ed wi hin 6 h Pos Replica ion
ChOR-seq analysis o nascen ch oma in showed ha his one
H3K4me3 occupancy pa e ns we e accu a ely ep oduced
on newly eplica ed DNA, bu i emained unclea whe he
he H3K4me3 landscape was, in ac , ully es o ed o
ch oma in ma u a ion would be equi ed o modi ica ion o
new his ones. This was pa icula ly impo an o add ess as
Molecula Cell 72, 239–249, Oc obe 18, 2018 241
H3K4me3 cell-cycle dynamics ha e no been esol ed by
mass spec ome y due o echnical limi a ions. Resol ing
H3K4me3 es o a ion kine ics is, howe e , amenable o quan i-
a i e ChOR-seq (qChOR-seq), which akes ad an age o a
D osophila ch oma in spike-in o no malize ead coun s,
e ealing quan i a i e di e ences in signal be ween samples
ha a e los wi h con en ional da a p ocessing me hods. We
he e o e ca ied ou H3K4me3 qChOR-seq on nascen ch o-
ma in and on ma u e ch oma in ha es ed 1 h la e , wi h cells
s ill in S phase (T1); 6 h la e , when cells eached G2/M bu
had no passed h ough mi osis (T6); and 12 h la e , when cells
had passed h ough mi osis and we e in G1 o he nex cell
cycle (T12) (Figu e 3A; Figu e S3A). While he aw ChOR-seq
signals (RPM) we e highly simila (Figu e S3B), no maliza ion
using spiked-in EdU-labeled D osophila ch oma in (RRPM)
e ealed a subs an ial accumula ion o H3K4me3 du ing he
i s 6 h o ch oma in ma u a ion (Figu e 3B; Figu e S3B).
Genome-wide, we obse ed ma ked gains in H3K4me3 signal
be ween T0 and T1 and T1 and T6 bu no u he inc ease
AC
Z-sco e
Domain
bo de
2kb
Pa en al
Nascen
H3K27me3H3K27me3
ch 1:35,000,000-40,000,000
100
0
100
0
100
0
1Mb
Pa en al
Nascen
Pe cen age o Max.
Replica ed DNA
Pa en al Nascen
−2kb 2kb
H3K27me3
Pa en al en ichmen
Pe cen age o Max.
25
50
75
100
0
Domain bo de
−2kb 2kb
H3K4me3
Z-sco e
TSS−2kb 2kb
Pa en al
Nascen
−1
0
1
2
H3K4me3
Replica ed DNA
Pa en al
Nascen
100
0
100
0
100
0
ch 7:97,327,705-99,176,888
400nb
Pe cen age o Max.
Pe cen age o Max.
Pa en al en ichmen
Pa en al Nascen
H3K4me3
25
50
75
100
0
−2kb 2kb −2kb 2kb
Dis ance o TSS
H3K36me3
TSS TTS−2kb 2kb
Pa en al
Nascen
−1
0
1
Z−sco e
H3K36me3
Replica ed DNA 400nb
Pa en al
Nascen
100
0
100
0
100
0
ch 2:181,456,702-184,646,339
Pe cen age o Max.
H3K36me3
Pa en al Nascen
Pe cen age o Max.
TSS TTS
Dis ance o TSS and TTS
25
50
75
100
0
Pa en al en ichmen
−2kb 2kb −2kb 2kb
TSS TTS
H3K79me3
−1
0
1
2
3
Pa en al
Nascen
Z−sco e
TSS TTS−2kb 2kb
H3K79me3
Replica ed DNA
Pa en al
Nascen
ch 9:11,939,329-17,291,440
1Mb
100
0
100
0
100
0
Pe cen age o Max.
Pe cen age o Max.
Pa en al Nascen
H3K79me3
25
50
75
100
0
Pa en al en ichmen
TSS TTS
Dis ance o TSS and TTS
−2kb 2kb −2kb 2kb
TSS TTS
B
−1
−2kb
0
0.5
1
−0.5
Figu e 2. The His one H3 PTM Landscape Is
Accu a ely Rep oduced upon Replica ion o
Ac i e and Rep essed Genomic Loci
(A) His one PTM p o iles om ChIP-seq
(pa en al) and ChOR-seq (nascen ) o H3K27me3,
H3K4me3, H3K36me3, and H3K79me3. Repli-
ca ed DNA p o iles a e shown in blue. Signal is
scaled as pe cen age o maximum a he locus
depic ed.
(B) A e age p o iles o pa en al and nascen
H3K27me3, H3K4me3, H3K36me3, and
H3K79me3. H3K27me3 signal is plo ed ac oss
4 kb cen e ed on bo de s o eplica ed H3K27me3
domains. H3K4me3 signal is plo ed ac oss 4 kb
cen e ed on eplica ed TSSs. H3K36me3 and
H3K79me3 signal is plo ed om 2 kb ups eam o
2 kb downs eam o eplica ed open eading
ames. All da a shown is Zsco e no malized.
(C) Hea maps o pa en al and nascen H3K27me3,
H3K4me3, H3K36me3, and H3K79me3 signal
ac oss he egions desc ibed in (B). Colo in ensi y
ep esen s pe cen age o maximum le els se
sepa a ely o pa en al and nascen samples.
See also Figu e S2.
be ween T6 and T12 (Figu es 3C and
3D). No ably, his gain o H3K4me3
occu ed wi hin he H3K4me3 egions
al eady p esen in nascen ch oma in
and did no lead o expansion o
H3K4me3 peaks o e ime (Figu e 3D;
Figu e S3C). Thus, while he genomic
loca ion o H3K4me3 is es ablished a
he ime o DNA eplica ion, es o a ion
o H3K4me3 le els is uncoupled om
DNA eplica ion simila o o he his one
di- and i-me hyla ion ma ks (Alabe
e al., 2015). Howe e , es ablishmen o
H3K4me3 on new his ones is la gely
comple e by G2, p io o mi osis, in
con as o ep essi e ma ks like H3K27me3 and H3K9me3
ha a e es o ed p ima ily in G1 o he nex cell cycle (Alabe
e al., 2015).
Exp ession and CpG Con en P edic H3K4me3
Res o a ion Kine ics
The ad an age o qChOR-seq is ha i p o ides bo h quan i a-
i e and posi ional in o ma ion abou PTM signal. We he e o e
nex asked whe he H3K4me3 es o a ion occu s wi h di e en
kine ics in di e en pa s o he genome. To add ess his, we
compa ed qChOR-seq signal ac oss ou ime cou se and
conside ed a locus es o ed when i eached H3K4me3 le els
close o ha obse ed a he 12-h ime poin (Figu e 3E; STAR
Me hods). This e ealed ha app oxima ely 50% o H3K4me3
egions a e es o ed wi hin 1 h o eplica ion (R1), and he
emaining 50% a e es o ed wi hin 6 h o eplica ion (R6)
(Figu e 3E). H3K4me3 da a om ENCODE, p oduced in
asynch onous HeLa cells, showed highe signal o e R1 egions
(Figu e 3F). Since signal a H3K4me3 peaks posi i ely co ela e
242 Molecula Cell 72, 239–249, Oc obe 18, 2018

wi h bo h he exp ession le el o associa ed genes and he CpG
densi y o he unde lying DNA, we nex compa ed R1 and R6
egions wi h espec o hese wo cha ac e is ics. Consis en
wi h ou p edic ions based on he H3K4me3 signal om
ENCODE, ansc ip ion s a si es (TSSs) wi hin R1 egions
we e mo e highly exp essed (Figu e 3G), and R1 egions we e
mo e CpG dense (Figu e 3H) han hose in he R6 es o a ion
ca ego y. This implica es bo h ansc ip ion and DNA sequence
con en as impo an de e minan s o H3K4me3 es o a ion
kine ics. Taken oge he , ou qChOR-seq ime cou se has bo h
de ined H3K4me3 es o a ion kine ics globally and e ealed
impo an , si e-speci ic di e ences in es o a ion a es, com-
G1G2
G1G2
G1G2
H3K4me3
le els
R0
R1
R6
R12
T0 T1 T6 T12
Res o a ion
0 200 400 600 800
T0 T1 T6 T12
RRPM
H3K4me3
0 100 200 300 400
TSS
T12
T6
T1
T0
RRPM
0R1R6R21R
% o egions
0
20
40
60
80
100
k
ch 1:42,772,780-43,438,021
0
300
0
300
0
300
0
300
H3K4me3 (RRPM)
T0
T1
T6
T12
A
M
G2 G1
S
T0
T6
B
E
CD
H3K4me3
(ENCODE)
010203040
RPKM
R6R1
F
0 5 10 15
FPKM
R6R1
T ansc ip ion
le els
G
T1
T12
G1G2
H
Log CpG Densi y
CpG con en
R6R1
2468
Figu e 3. H3K4me3 Res o a ion Is Comple e
wi hin 6 h wi h Fas es Kine ics in Highly
Exp essed P omo e s
(A) Ou line o H3K4me3 qChOR-seq ime cou se
analysis. Cell cycle p og ession was moni o ed by
FACS analysis o DNA con en .
(B) Compa ison o H3K4me3 nascen and ma u e
qChOR-seq p o iles.
(C) Boxplo o H3K4me3 qChOR-seq signal in
eplica ed pa en al peaks subse ed in o 25 bp
non-o e lapping windows.
(D) A e age p o iles o H3K4me3 qChOR-seq
signal ac oss 4 kb cen e ed on eplica ed TSSs.
In (B)–(D), signal is quan i a ed using e e ence-
adjus ed eads pe million (RRPM).
(E) Le : scheme o s a egy used o pa se
H3K4me3-en iched egions by es o a ion kine ics.
Regions we e de ined as R0, R1, R6, o R12 based
on he ime poin a which R12 H3K4me3 le els
we e eached. Righ : ba cha o he p opo ion
o H3K4me3-en iched egions in each es o a ion
ca ego y. Regions a e de ined as 500 bp non-
o e lapping windows in eplica ed pa en al peaks.
(F) Boxplo o ENCODE H3K4me3 signal in R1 and
R6 egions. Signal is quan i a ed using eads pe
kilobase pe million (RPKM).
(G) Boxplo o RNA-seq signal o e genes associ-
a ed wi h R1 and R6 p omo e s. RNA-seq da a a e
om Mo aza i e al. (2008). Signal is quan i a ed
using agmen s pe kilobase pe million (FPKM).
(H) Boxplo showing he CpG densi ies o CpG
islands o e lapping R1 and R6 egions. CpG con-
en da a a e om Illingwo h e al. (2010).
See also Figu e S3.
plemen ing exis ing knowledge abou
H3K4me3 biology while adding c ucial
and no el insigh s in o he p opaga ion o
his ma k.
High PRC2 Occupancy Si es Show
Fas e H3K27me3 Res o a ion
We ha e p e iously de ined he es o a ion
kine ics o H3K27me3 a he global le el
using quan i a i e mass spec ome y
(Alabe e al., 2015). Bulk mass spec om-
e y me hods, howe e , a e unable o
de ec si e-speci ic di e ences in PTM
es o a ion. The e o e, o add ess how he H3K27me3 land-
scape de elops ac oss he cell cycle and e eal whe he
pa icula genomic loci es o e wi h as e kine ics han bulk
H3K27me3, we ca ied ou H3K27me3 qChOR-seq ime cou se
analyses. We included es o a ion imes o 4 h , when cells we e
in G2 phase; 10 h , co esponding o ea ly G1 phase; and 24 h ,
whe e we a es ed cells a he G1/S ansi ion o a oid e- eplica-
ion o he domains (Figu e 4A; Figu e S4A). Applying spike-in
no maliza ion o allow quan i a i e compa ison (Figu e S4B)
e ealed ha H3K27me3 qChOR-seq signal g adually accumu-
la ed ac oss all ime poin s, wi h he majo inc ease aking place
a e mi osis in daugh e cells (Figu e 4B), co obo a ing he
Molecula Cell 72, 239–249, Oc obe 18, 2018 243
p e ious esul s om quan i a i e mass spec ome y (Alabe
e al., 2015). Impo an ly, his g adual inc ease in H3K27me3
le els was e iden bo h a he le el o indi idual domains (Fig-
u e 4C) and a he le el o whole ch omosomes (Figu e 4D),
wi h he gain in H3K27me3 being es ic ed o egions al eady
dema ca ed by H3K27me3 in nascen ch oma in. Looking wi h
highe esolu ion, we obse ed ha H3K27me3 domain bo de s
we e ai h ully dema ca ed a all ime poin s (Figu e 4E; Fig-
u e S4C). Ch oma in es o a ion hus inc eases H3K27me3
le els wi hin domains wi hou changing hei wid h, compa able
o how H3K4me3-en iched egions we e es o ed.
To iden i y genomic loci wi h di e en ial es o a ion a es, we
pa sed all eplica ed H3K27me3 egions acco ding o when he
maximal H3K27me3 le el was eached. We conside ed a egion
es o ed when eaching a le el close o ha obse ed a he
24-h ime poin (Figu e 4F; STAR Me hods). Wi h his de ini ion,
we ound ha abou 80% o he analyzed H3K27me3 egions
we e es o ed wi h e y slow kine ics, aking up o 24 h o each
hei inal le el (R24; Figu e 4F). We also iden i ied a subs an ial
numbe o si es es o ed wi hin 10 h (R10; Figu e 4F), bu almos
A
ch 1:116,085,090-117,034,685
123456
−2kb 2kb
T0
T4
T10
T24
H3K27me3 (RRPM)
H3K
3
27m
e3
e3
(RR
(
(
PM)
)
)
T0
T4
T10
T24
0
20
0
20
0
20
0
20
200kb
B
D
T0 T4 T10 T24
H3K27me3
T0 T4 T10 T24
0 1020304050
RRPM
F
H3K27me3
le els
R0
R4
R10
R24
T0 T4 T10 T24
0 20406080100
R0 R4 R10 R24
% o egions
G
40 45 50
−5kb 0 5kb
R10
R24
RPKM
40 45 50 55 60
−5kb 0 5kb
RPKM
R10
R24
C
Res o a ion
R10 R24
04080120
RPKM
EZH2
(ENCODE)
R10 R24
0 20406080100
RPKM
H3K27me3
(ENCODE)
M
G2 G1
S
T0
T4
T10
T24
G1G2
G1G2
G1G2
G1 G2
E
H
RRPM
H3K27me3
Domain bo de
Ch 20
H3K27me3
RRPM
EZH2
(ENCODE)
H3K27me3
(ENCODE)
Figu e 4. High PRC2 Occupancy Si es Show
Fas e H3K27me3 Res o a ion
(A) Ou line o H3K27me3 qChOR-seq ime cou se
analysis. Cell cycle p og ession was moni o ed by
FACS analysis o DNA con en .
(B) Boxplo s o H3K27me3 qChOR-seq signal in
eplica ed pa en al peaks subse ed in o 2 kb
non-o e lapping windows.
(C) Compa ison o H3K27me3 nascen and ma u e
qChOR-seq p o iles.
(D) Hilbe cu es o H3K27me3 qChOR-seq signal
o e ch omosome 20 a he indica ed ime
poin s. Colo ed a eas e lec he size and signal o
H3K27me3-en iched domains.
(E) A e age p o iles o H3K27me3 qChOR-seq
signal ac oss 4 kb cen e ed on he bo de o epli-
ca ed H3K27me3 domains.
In (B)–(E), signal is quan i a ed using e e ence-
adjus ed eads pe million (RRPM).
(F) Top: scheme o s a egy used o pa se
H3K27me3-en iched egions by es o a ion
kine ics. Regions we e de ined as R0, R4, R10, o
R24 based on he ime poin a which R24
H3K27me3 le els we e eached. Bo om: ba cha
o he p opo ion o H3K27me3-en iched egions
in each es o a ion ca ego y. Regions a e de ined
as 2 kb non-o e lapping windows in eplica ed
pa en al peaks.
(G) Boxplo s o ENCODE H3K27me3 signal (le )
and ENCODE EZH2 signal ( igh ) in R10 and R24
egions. Signal is quan i a ed using eads pe kilo-
base pe million (RPKM).
(H) A e age p o iles o ENCODE H3K27me3 signal
(le ) and ENCODE EZH2 signal ( igh ) ac oss 10 kb
cen e ed on R10 and R24 egions.
See also Figu e S4.
none showing es o a ion p io o mi osis
(R0 and R4; Figu e 4F). These esul s sup-
po a model in which old, ecycled
H3K27me3-ma ked his ones con ibu e
signi ican ly o he ch oma in landscape ansmi ed o daugh e
cells, while modi ica ion o new his ones eplenishes H3K27me3
le els mainly a e cell di ision. Compa ison wi h H3K27me3 and
EZH2 ENCODE da a e ealed ha he as e es o ing si es, on
a e age, had highe H3K27me3 le els and EZH2 occupancy
(Figu e 4G). Fu he , hese egions co esponded o peaks wi hin
ENCODE H3K27me3 and EZH2 domains, in con as o he
slowes es o ing egions (Figu e 4H). Ou qChOR-seq analyses
he e o e demons a e ha H3K27me3 es o a ion kine ics a e
locus speci ic, wi h high PRC2 occupancy p omo ing he mos
e icien H3K27me3 es o a ion and bo de egions es o ing
mo e slowly.
Pa en al H3K27me3 Domains A e S able ac oss he
Cell Cycle
Ou qChOR-seq analyses in unpe u bed sys ems accu a ely
e ealed PTM es o a ion dynamics genome-wide bu le open
he ques ion o PTM domain s abili y in he absence o a es o a-
ion mechanism. New his ones a e deposi ed la gely wi hou
me hyla ion ma ks, including H3K27 me hyla ion (Alabe e al.,
244 Molecula Cell 72, 239–249, Oc obe 18, 2018
2015; Jasencako a e al., 2010; Loyola e al., 2006). H3K27me3,
which spans la ge domains and is ca alyzed by a single me hyl-
ans e ase, EZH2, was he e o e an ideal ma k o in es iga e
his ques ion. To add ess he ela i e con ibu ions o ecycled
pa en al his one H3K27me3 and de no o his one H3 K27 i-
me hyla ion o he inhe i ance o H3K27me3 o daugh e cells,
we he e o e pe o med H3K27me3 qChOR-seq analysis in he
p esence o an inhibi o o EZH2 (Come e al., 2016; Høj eld
e al., 2018; Knu son e al., 2013) o block new i-me hyla ion
o H3K27 on new and old his ones.
We added he EZH2 inhibi o o cells sho ly be o e EdU label-
ing and pe o med qChOR-seq o in o m on he inhe i ance
o pa en al his ones ca ying H3K27me3 bo h immedia ely
ollowing DNA eplica ion (nascen , T0) and ac oss mi osis o
daugh e cells (24 h pos EdU labeling, T24). H3K27me3
ch 1:116,085,090-117,034,685
H3K27me3 (RRPM)
0
10
0
10
0
10
T0 T24
+ EZH2i
H3K27me3 (RRPM)
A
H3K27me3 ChOR-seq
(scaled RRPM)
T0 T0+i
B
C
E
H
020406080
RPKM
H3K27me3
020406080
RPKM
EZH2
ch 1:35,000,000-40,000,000
7
0
100
0
7
0
Nascen + EZH2i
1Mb
Pa en al
Nascen
Replica ed DNA Pa en al
−1 −0.5 0 0.5 1
Nascen
Nascen
+ EZH2i
−2kb Domain
bo de
2kb
Z-sco e
H3K27me3H3K27me3
D
F
H3K27me3 ChOR-seq (RRPM)
T0+i T24+i
Ch 20
Ch 20
200kb
T0
T24+i
T0+i
100
0
120
100
% o Regions
80
60
40
20
0
H3K27me3 loss
High Mode a e Low
010203040
T0 T24
G
Mode a e Low Mode a e Low
scaled RRPM
RRPM
RRPM Pe cen age
o Max.
Figu e 5. Pa en al H3K27me3 Domains A e
S able ac oss he Cell Cycle
(A) P o iles o pa en al H3K27me3 ChIP-seq
(g ay) and nascen H3K27me3 qChOR-seq in he
absence ( ed) o p esence (pu ple) o EZH2 inhib-
i o . Replica ed DNA is shown in blue. Replica ed
DNA and pa en al ChIP-seq signal is scaled as a
pe cen age o maximum a he locus depic ed;
nascen qChOR-seq signal is quan i a ed using
e e ence-adjus ed eads pe million (RRPM).
(B) A e age p o iles o pa en al H3K27me3 ChIP-
seq (g ay) nascen H3K27me3 qChOR-seq signal
in he absence ( ed) o p esence (pu ple dashes)
o EZH2 inhibi o . Signal is shown ac oss 4 kb
cen e ed on he bo de o eplica ed H3K27me3
domains and Zsco e no malized.
(C) Hilbe cu es o nascen (T0) H3K27me3
qChOR-seq signal o e ch omosome 20 in he
absence o p esence o EZH2 inhibi o . Scaled
RRPM alues a e shown o compa e he occu-
pancy landscape (no absolu e in ensi ies).
(D) Boxplo s o H3K27me3 qChOR-seq signal a
T0 and T24 and in he absence o p esence o
EZH2 inhibi o . Signal is calcula ed om 2 kb non-
o e lapping windows in eplica ed pa en al peaks.
(E) P o iles o H3K27me3 qChOR-seq a T0 ( ed)
and a T0 and T24 in he p esence o EZH2 inhibi o
(ligh and da k pu ple, espec i ely).
(F) Hilbe cu es o T0 and T24 H3K27me3
qChOR-seq signal o e ch omosome 20 in he
p esence o EZH2 inhibi o .
In (D)–(F), signal is quan i a ed using RRPM.
(G) Ba cha showing he p opo ion o H3K27me3
egions ha exhibi high, mode a e, and low
qChOR-seq signal loss in he p esence o EZH2
inhibi o . Regions we e de ined as high, mode a e,
o low loss by compa ing T0 and T24 qChOR-seq
signal in he p esence o EZH2 inhibi o .
(H) Boxplo s o ENCODE H3K27me3 signal (le )
and ENCODE EZH2 signal ( igh ) in mode a e and
low loss egions. Signal is quan i a ed using eads
pe kilobase pe million (RPKM).
See also Figu e S5.
nascen qChOR-seq e ealed ha he
H3K27me3 occupancy pa e ns in
nascen ch oma in we e la gely una -
ec ed by lack o EZH2 ac i i y (Figu e 5A). The de ini ion o
H3K27me3 domain bo de s and pa e n o H3K27me3 domains
we e unchanged in he p esence o inhibi o , indica ing ha
H3K27me3 posi ional in o ma ion was main ained pos eplica-
ion (Figu es 5B and 5C). These esul s demons a e ha he
nascen H3K27me3 landscape is he esul o pa en al his one
ecycling, wi h li le, i any, con ibu ion om de no o me hyla ion
e en s.
We nex compa ed ou T0 and T24 da a, bo h wi h and wi hou
inhibi o . Signal compa ison be ween all samples e ealed ha
he accumula ion o H3K27me3 du ing ch oma in ma u a ion
was en i ely dependen on de no o H3K27 me hyla ion (Fig-
u e 5D), as p edic ed. Unexpec edly, howe e , he H3K27me3
landscape gene a ed om pa en al his one ecycling a he
ime o eplica ion pe sis ed in daugh e cells 24 h pos DNA
Molecula Cell 72, 239–249, Oc obe 18, 2018 245
eplica ion (Figu es 5E and 5F), a guing ha nucleosome
u no e and/o deme hyla ion do no gene ally e ode he
H3K27me3 landscape.
To in es iga e whe he sus ained EZH2 inhibi ion causes
H3K27me3 loss in ce ain local genomic egions, we de ined
egions o low, mode a e, and high H3K27me3 loss by
compa ing qChOR-seq signal a T0 and T24 om EZH2 inhibi-
o - ea ed cells (see STAR Me hods). While he la ge majo i y
o H3K27me3 loci did no change subs an ially, 3.5% o he
egions showed a mode a e educ ion in H3K27me3 (Figu e 5G).
These egions o mode a e H3K27me3 loss we e cha ac e ized
by somewha lowe H3K27me3 and EZH2 occupancy compa ed
o he mo e s able egions (Figu e 5H). In D osophila,
H3K27me3-ma ked egions, such as Polycomb g oup esponse
elemen s, ha e been iden i ied as agile high- u no e si es
(Deal e al., 2010). Howe e , in ou inhibi o - ea ed HeLa cells,
egions o mode a e H3K27me3 loss did no co ela e wi h
highe occupancy o he eplacemen his one a ian H3.3 (Fig-
u e S5A), sugges ing ha nucleosome u no e is no su icien
o explain he dec ease in signal obse ed. Taken oge he ,
H3K27me3 qChOR-seq in he absence o de no o H3K27
me hyla ion e ealed ha domains laid down a he ime o
eplica ion a e ema kably s able, unde sco ing his one ecy-
cling du ing DNA eplica ion as a subs an ial con ibu o o he
inhe i ance o H3K27me3 in daugh e cells wi hou challenges
by deme hyla ion o his one exchange.
DISCUSSION
He e we de elop ChOR-seq o de e mine he occupancy o
modi ied his ones on eplica ed DNA. We ack me hyla ed
his ones associa ed wi h ac i e and ep essed ch oma in and
ind ha hei posi ion on newly eplica ed DNA mi o s hei
posi ion p io o eplica ion. This demons a es ha he his one
modi ica ion landscape can wi hs and he dis up i e p ocess
o DNA eplica ion. Se e al lines o e idence poin owa d accu-
a e ecycling o modi ied pa en al his ones as he unde lying
mechanism. Fi s , new his ones a e de oid o i-me hyla ion
(Alabe e al., 2015; Ba -Zi e al., 2016; Scha e al., 2009; Xu
e al., 2011), a guing ha we mainly de ec p e-exis ing ma ks
on old ecycled his ones. Second, H3K27me3 occupancy pa -
e ns a e accu a ely copied in he absence o EZH2 ac i i y.
Thi d, quan i a i e ChOR-seq analysis showed a s ong inc ease
in H3K4me3 and H3K27me3 signal in ensi y du ing ch oma in
ma u a ion, a guing ha de no o i-me hyla ion occu s in a
manne uncoupled om DNA eplica ion. We did no de ec
dispe sal o ei he H3K4me3, H3K36me3, H3K79me3, o
H3K27me3 ma ks as a esul o DNA eplica ion, which would
be he p edic ed ou come i pa en al his ones we e ei he ully
eleased and e-inco po a ed a a di e en eplica ion o k o
main ained loosely a he o k and e-deposi ed haphaza dly.
O no e, canonical his ones H3.1/2 and he eplacemen a ian
H3.3 a e no di e en ia ed in ou analysis because hey a e
ecycled wi h equal e iciency du ing DNA eplica ion and depos-
i ed unmodi ied de no o (Alabe e al., 2015; Jasencako a e al.,
2010; Loyola e al., 2006). Thus, all his one H3 a ian s can
con ibu e o p e-ma ked pa en al his ones in nascen ch oma in
and be subjec o de no o me hyla ion du ing ch oma in es o a-
ion. A ecen in i o s udy ound ha his one posi ioning in o -
ma ion is los in Xenopus ex ac s when la ge T an igen ac s as
he eplica i e helicase (Madamba e al., 2017) and sugges ed
ha dispe sal o his ones migh be mo e limi ed when eplica ion
is media ed by he MCM2-7 helicase, as i is in human cells. Ou
da a show ha pa en al his ones a e e-inco po a ed wi hin
250 bp o hei o iginal posi ion in human cells, which a gues
ha his one e ic ion and e-deposi ion a he o k mus be igh ly
coo dina ed. In human cells, his one-based in o ma ion is hus
e ained wi h highe p ecision han p edic ed by ma hema ic
modeling in yeas (Radman-Li aja e al., 2011) and obse ed in
Xenopus in i o eplica ion sys ems (Madamba e al., 2017),
wo sys ems in which eplica ion-independen his one exchange
is high. Collec i ely, his a gues ha ecycling o pa en al
his ones a he eplica ion o k is highly accu a e, ep oducing
he landscape o his one modi ica ions on newly syn hesized
DNA, albei wi h a lowe ampli ude due o dilu ion by new nai e
his ones.
Combining a spike-in app oach wi h ChOR-seq, we we e able
o measu e wi h base-pai esolu ion how his one ma k le els
eco e ed a e DNA eplica ion. The s eng h o spike-in
ChOR-seq is ha i allows quan i a i e compa ison o es o a ion
kine ics ac oss he genome. Using ChOR-seq o ob ain spa ial
in o ma ion, we ound ha he si e o occupancy is ixed a
he ime o eplica ion and modi ica ions hen accumula e wi h
kine ics inhe en o he ma k and genomic ea u es o he locus
as cells p og ess h ough he cell cycle. Mass spec ome y anal-
ysis had p e iously iden i ied H3K27me3 and H3K9me3 as
pa icula ly slow- eco e ing ma ks pos eplica ion (Alabe
e al., 2015; Scha e al., 2009; Xu e al., 2011). ChOR-seq
co obo a ed hese esul s and u he e ealed ha es o a ion
kine ics a e no uni o m h oughou he genome. While es o a-
ion o mos H3K27me3 domains is slow, con inuing a e mi osis
in daugh e cells, si es o high H3K27me3 and PRC2 occupancy
eco e as e han o he egions. Likewise, ou esul s indica e
ha si es wi h he highes H3K4me3 le els, co esponding o
CpG-dense, highly exp essed p omo e s (Chen e al., 2014;
Illingwo h e al., 2010; Mikkelsen e al., 2007), also a e i s o
gain H3K4me3 a e eplica ion. Ou esul s show ha he as
majo i y o H3K4me3 peaks a e es o ed p io o mi osis. This
means ha he H3K4me3 landscape ansmi ed o daugh e
cells is no a ec ed by di e en ial es o a ion a es, in con as
o he H3K27me3 landscape whe e he si es o he highes
H3K27me3 and PRC2 occupancy ha e gained ela i ely mo e
signal han o he egions. This may no be de e mined solely
by me hyl ans e ase kine ics bu could also be in luenced by
eplica ion iming, since ea ly- eplica ing ch oma in is gene ich
and ansc ip ionally ac i e while la e- eplica ing ch oma in
ends o be he e och oma ic (Ri e a-Mulia and Gilbe , 2016).
Impo an ly, i a gues ha he epigenome is no ixed o a gi en
cell bu should be conside ed as a dynamic landscape changing
h oughou he cell cycle wi h ega ds o he o al le el o all
ma ks, he ela i e en ichmen o indi idual ma ks ac oss
di e en si es, and he abundance o di e en ma ks ela i e o
each o he .
E en hough he co ec posi ion and ela i e abundance o
his one ma ks a e main ained du ing DNA eplica ion, new his-
one deposi ion ep esen s a majo challenge o he epigenome,
246 Molecula Cell 72, 239–249, Oc obe 18, 2018
Da a Sequencing and P ocessing
ChOR-seq, ChIP-seq, eplica ed DNA and inpu samples om wo independen ime cou se expe imen s o each his one ma k we e
sequenced a he Danish High- h oughpu DNA Sequencing Cen e (h ps://seqcen e .ku.dk) and a he Bio ech Resea ch and Inno-
a ion Cen e (BRIC) (h ps://www.b ic.ku.dk) using Illumina HighSeq 4000 and Nex Seq 500 machines o ob ain 50 bp and 75 bp
single-end eads, espec i ely. Reads we e aligned o he Feb ua y 2009 human genome assembly (GRCh37/hg19) and he Ap il
2006 D. melanogas e genome assembly (BDGP R5/dm3) by Bow ie (Langmead e al., 2009) using pa ame e s -m1–bes . PCR
duplica es we e emo ed and uniquely mapped eads we e ex ended o 250 bp (H3K4me3, H3K36me3, H3K79me3) o 500bp
(H3K27me3) o accoun o he a e age lib a y size. Replica ed DNA was ex ended o 250 bp in ea ly S ime cou ses and o
500 bp in mid S ime cou ses. Reads we e hen summed in 25 bp (H3K4me3, H3K36me3, H3K79me3, ea ly S eplica ed DNA) o
500 bp (H3K27me3, H3, mid S eplica ed DNA) non-o e lapping bins, unless o he wise speci ied, and eads we e no malized o
eads pe million (RPM). Mapping and subsequen analysis o he da a we e done using a local Galaxy se e and cus om R sc ip s.
In ChOR-seq expe imen s wi h exogenous DNA, human H3K4me3, H3K36me3, H3K79me3, H3K27me3 and his one H3 eads we e
di ided by o al D osophila unique mapped eads o ge quan i a i e in o ma ion in he o m o e e ence-adjus ed RPM (RRPM) as
desc ibed in O lando e al. (2014). When compa ing occupancy pa e ns be ween pa en al and nascen samples RPM o RRPM
alues we e no malized o pe cen age o maximum. This no maliza ion mos app op ia ely ep esen s he da a, since in he pa en al
sample we pu i y ma ks om he whole genome, while in nascen we only ob ain signal om EdU-labeled domains, c ea ing di e -
ences in signal:noise a ios ha ende di ec compa isons inapp op ia e.
PTM Dis ibu ion Analyses
All da a in igu es co espond o eplica e 1 o H3K4me3, H3K36me3, H3K79me3 and H3K27me3, espec i ely, unless o he wise
speci ied. Since he ChOR-seq me hod pu i ies newly eplica ed ch oma in, only bins o e lapping wi h eplica ed egions in each
expe imen we e conside ed o analysis. Fo H3K4me3, H3K36me3, and H3K79me3, peak calling was pe o med wi h MACS
(Zhang e al., 2008) s anda d pa ame e s and INPUT as con ol ile. Fo H3K27me3, peak calling was pe o med wi h MACS
‘‘b oad domain’’ pa ame e s using eplica e-ma ched pa en al ChIP-seq o his one H3 as a con ol. H3K4me3, H3K36me3,
H3K79me3 and H3K27me3 domains we e de ined using pa en al ChIP-seq signal. Replica ed egions we e de ined using
MACS de aul ‘‘b oad domain’’ pa ame e s and eplica e-ma ched pa en al ChIP-seq o his one H3 as a con ol and il e ed o
egions wi h a q- alue g ea e han o equal o 0.05. Pa en al peaks we e subse ed in o 25 bp windows o H3K4me3,
H3K36me3, and H3K79me3 and 500 bp windows o H3K27me3, and only windows ha o e lapped wi h eplica ed egions
we e included in downs eam analyses. To calcula e he mean di e ence in peak localiza ion a indi idual loci, we selec ed
H3K4me3 pa en al and nascen peaks o e lapping only once and compu ed he absolu e dis ance in bp be ween o e lapping
peaks a bo h ends.
Hilbe plo s we e c ea ed using Hilbe is so wa e (Ande s, 2009). A e age p o iles and hea maps we e gene a ed using Seqplo
(S empo and Ah inge , 2016). Fo H3K4me3 analyses, a e age p o iles we e cen e ed on TSSs. Fo H3K27me3 analyses, a e age
p o iles we e cen e ed a he bo de s o H3K27me3 domains. Since H3K27me3 domain bo de s o en o e lapped wi h he bo de s o
eplica ed DNA egions, we only included H3K27me3 bo de s ha we e a leas 5 kb om he bo de o a eplica ed egion in all
analyses. Fo H3K36me3 and H3K79me3, a e age p o iles we e cen e ed o e open eading ames. Signal inside he open eading
ame was no malized using he ‘‘ancho ed poin ’’ pa ame e in Seqplo o co ec o gene leng h di e ences. When indica ed, signal
om a e age p o iles we e no malized using z-sco e z=xm
s o ocus on dis ibu ion di e ences, whe e mis he mean o he
popula ion and s he s anda d de ia ion.
Res o a ion Ca ego ies and Analyses
To s udy H3K4me3 es o a ion eads we e summed in 500bp bins ha o e lapped wi h bo h pa en al H3K4me3 peaks and eplica ed
egions. Regions we e hen classi ied in o R0, R1, R6 and R12 ca ego ies acco ding he ime hey need o each T12 H3K4me3 le els.
Only bins wi h R12/R(X) a ios g ea e han 1.5- old and p esen in bo h eplica es we e conside ed o analysis. CpG islands o e -
lapping R1 and R6 egions we e espec i ely assigned o ha egion o analysis; CpG islands o e lapping bo h a R1 and a R6 egion
we e disca ded. To ensu e he CpG densi ies, calcula ed o coo dina es in he hg18 genome assembly, emained accu a e, R1 and
R6 egion coo dina es we e con e ed om hg19 o hg18 using UCSC Li O e be o e de ining o e laps.
To s udy he es o a ion dynamics o H3K27me3, eads we e summed in 2 kb bins ha o e lapped wi h bo h pa en al H3K27me3
peaks and eplica ed egions. Regions we e hen classi ied in o R0, R4, R10 and R24 ca ego ies acco ding he ime hey need o
each T24 H3K27me3 le els. Only bins wi h R24/R(X) a ios g ea e han 1.5- old and p esen in bo h eplica es we e conside ed
o analysis.
Fo analysis o he H3K27me3 loss a e in he p esence o EZH2 inhibi o , H3K27me3 loss ca ego ies we e de ined as he
2 kb windows p esen in eplica ed H3K27me3 pa en al peaks ha showed a T0/T24 old change smalle han 1.5 (low), be ween
1.5 and 3 (mode a e) and bigge han 3 (high). Only bins ha i he c i e ia in he wo independen eplica es we e conside ed o
he analysis.
Molecula Cell 72, 239–249.e1–e5, Oc obe 18, 2018 e4

Cell Cycle Analyses
Fo analysis o cell cycle p og ession, synch onized cells we e ixed wi h 70% e hanol and labeled wi h p opidium iodide (10 mg/mL)
o 30 min in he da k, be o e analysis on a FACSCalibu machine. FACS p o iles we e analyzed by FlowJo 10.0.8 so wa e.
QUANTIFICATION AND STATISTICAL ANALYSIS
The s a is ical es s applied in his s udy a e s a ed in he igu e legends and we e calcula ed using cus om R sc ip s. In Figu e S1E,
- alues co espond o Pea son co ela ion. In boxplo s, he bo om and op o boxes indica e he 25 h and 75 h pe cen iles, espec-
i ely, and middle lines indica e medians. Whiske s indica e he lowes and highes da a poin s wi hin 1.5 3in e qua ile ange om
he box.
DATA AND SOFTWARE AVAILABILITY
Replica ion iming was ob ained om Repli-seq da ase s GEO: GSM923449 (ENCODE P ojec Conso ium, 2012). ChIP-seq o asyn-
ch onous HeLa cells o H3K27me3 (GEO: GSM733696), H3K4me3 (GEO: GSM733682), and EZH2 (GEO: GSM1003520) we e aken
om Be ns ein e al. (2005). Asynch onous HeLa H3.3 ChIP-seq (GEO: GSM788633) was ob ained om Ray-Galle e al. (2011).
Exp ession le els o genes we e ob ained om RNA-seq (GEO: GSM958735) (Mo aza i e al., 2008). Posi ions o TSSs we e aken
om he able o TSSs iden i ied in Mo aza i e al. (2008). Exons, in ons, 50UTRs and 30UTRs we e de ined using Re Seq anno a-
ions. CpG densi ies o CpG islands associa ed wi h p omo e s we e aken om Illingwo h e al. (2010). All o iginal da a gene a ed in
his s udy we e deposi ed a GEO: GSE110354.
e5 Molecula Cell 72, 239–249.e1–e5, Oc obe 18, 2018