A icle
H4K20me1and H3K27me3a e concu en ly loaded
on o he inac i e X ch omosome bu dispensable
o inducing gene silencing
Sjoe d J D Tjalsma
1,†
, Mayako Ho i
2,†
, Yuko Sa o
2,3,†
, Au elie Bousa d
1
, Aki o Ohi
2
,
Ana Cl
audia Raposo
4
, Julia Roensch
1
, Agnes Le Saux
1
, Jumpei Nogami
5
, Kazumi su Maeha a
5
,
Tomoya Kuji ai
6
, Te suya Handa
3
, Sand a Bag
es-A nal
7
, Yasuyuki Ohkawa
5
, Hi oshi Ku umizaka
6
,
Sim~
ao Teixei a da Rocha
4
, Jan J ˙
Zylicz
1,7,8,*
, Hi oshi Kimu a
2,3,**
& Edi h Hea d
9,10,***
Abs ac
Du ing X ch omosome inac i a ion (XCI), in emale placen al
mammals, gene silencing is ini ia ed by he Xis long non-coding
RNA. Xis accumula ion a he X leads o en ichmen o speci ic ch o-
ma in ma ks, including PRC2-dependen H3K27me3and SETD8-
dependen H4K20me1. Howe e , he dynamics o his p ocess in
ela ion o Xis RNA accumula ion emains unknown as is he
in ol emen o H4K20me1in ini ia ing gene silencing. To ollow XCI
dynamics in li ing cells, we de eloped a gene ically encoded,
H3K27me3-speci ic in acellula an ibody o H3K27me3-min body.
By combining li e-cell imaging o H3K27me3,H4K20me1, heXch o-
mosome and Xis RNA, wi h ChIP-seq analysis we unco e concu en
accumula ion o bo h ma ks du ing XCI, albei wi h dis inc genomic
dis ibu ions. Fu he mo e, using a Xis BandC epea mu an ,
which s ill shows gene silencing on he X bu no H3K27me3deposi-
ion, we also ind a comple e lack o H4K20me1en ichmen . This
demons a es ha H4K20me1is dispensable o he ini ia ion o
gene silencing, al hough i may ha e a ole in he ch oma in
compac ion ha cha ac e ises acul a i e he e och oma in.
Keywo ds emb yonic s em cells; H4K20me1; he e och oma in; polycomb;
X inac i a ion
Subjec Ca ego y Ch oma in, T ansc ip ion & Genomics
DOI 10.15252/emb .202051989 | Recei ed 31 Oc obe 2020 | Re ised 22
Decembe 2020 | Accep ed 7Janua y 2021
EMBO Repo s (2021)e51989
In oduc ion
Dynamic changes o he ch oma in landscape allow o imely
execu ion o de elopmen al and di e en ia ion p og ammes.
Indeed, ch oma in modi ie s o en ein o ce signalling cues o ini i-
a e and/o main ain an exac ansc ip ional ou come (S ase ich
e al, 2014; Zylicz e al, 2015; Jambheka e al, 2019). One powe ul
model, whe e such egula ion akes place, is X ch omosome inac i-
a ion (XCI) in emale mammals. Al hough much is known abou
he in ol emen o ce ain ch oma in ma ks in ini ia ing XCI ( e-
iewed in Zylicz & Hea d, 2020), i was hus a impossible o ack
ch oma in ea angemen s and he inac i e X ch omosome (Xi) in
li ing cells. Wha is mo e, he ole and dynamics o some his one
modi ica ions du ing XCI emains enigma ic.
In eu he ian emale mammals, XCI is ini ia ed a a ound he ime
o implan a ion, when each cell andomly inac i a es one o he wo
X ch omosomes (Lyon, 1962). This p ocess can be modelled in i o
by di e en ia ing emale mouse emb yonic s em cells (ESC) (Ras an
& Robe son, 1985). Bo h in i o and in i o, XCI depends on he
coa ing o he Xi by a long non-coding RNA called Xis (X-inac i e-
speci ic ansc ip ) (Penny e al, 1996). Xis RNA accumula ion
along he Xi induces no only he silencing o o e a 1,000 genes bu
also a cascade o ch oma in al e a ions ( e iewed in Zylicz & Hea d,
1Mammalian De elopmen al Epigene ics G oup, Ins i u Cu ie, CNRS UMR3215, INSERM U934, PSL Uni e si y, Pa is, F ance
2G adua e School o Bioscience and Bio echnology, Tokyo Ins i u e o Technology, Yokohama, Japan
3Cell Biology Cen e , Ins i u e o Inno a i e Resea ch, Tokyo Ins i u e o Technology, Yokohama, Japan
4Faculdade de Medicina, Ins i u o de Medicina Molecula , Jo~
ao Lobo An unes, Uni e sidade de Lisboa, Lisboa, Po ugal
5Di ision o T ansc ip omics, Medical Ins i u e o Bio egula ion, Kyushu Uni e si y, Fukuoka, Japan
6Ins i u e o Quan i a i e Biosciences, The Uni e si y o Tokyo, Tokyo, Japan
7The No o No disk Founda ion Cen e o S em Cell Biology, Copenhagen, Denma k
8Depa men o Physiology, De elopmen and Neu oscience, Uni e si y o Camb idge, Camb idge, UK
9EMBL Heidelbe g, Heidelbe g, Ge many
10 Coll
ege de F ance, Pa is, F ance
*Co esponding au ho . Tel: +45 23839889; E-mail: jan.zy[email p o ec ed]
**Co esponding au ho . Tel: +81 45 924 5742; E-mail: [email p o ec ed].jp
***Co esponding au ho . Tel: +49 6221 3878 201; E-mail: [email p o ec ed]
†
These au ho s con ibu ed equally o his wo k
ª2021 The Au ho s. Published unde he e ms o he CC BY 4.0license EMBO epo s e51989 |2021 1o 17
2020). Xis is a modula non-coding RNA wi h speci ic egions play-
ing dis inc oles. I s 50A- epea egion is i al o he induc ion o
gene silencing (Wu z e al, 2002). I does so by ec ui ing SPEN, a
key RNA-binding p o ein, which in eg a es many ep essi e
complexes including NCOR/SMRT and HDAC3 (McHugh e al,
2015; Zylicz e al, 2019; Dossin e al, 2020). On he o he hand, he
B and o a lesse ex en C epea egions a e i al o he ec ui men
o Polycomb-g oup ep essi e complexes (PRC) (Bousa d e al,
2019; Cologno i e al, 2019; Nes e o a e al, 2019). Indeed, upon
Xis RNA coa ing Xi becomes apidly en iched wi h he PRC1-depen-
den H2AK119Ub and only subsequen ly wi h PRC2-dependen
H3K27me3 (Pla h e al, 2003; Sil a e al, 2003; de Napoles e al,
2004; Zylicz e al, 2019). Recen da a indica e ha he B and C
epea egion o Xis RNA di ec ly ec ui s hnRNPK, which in u n
binds non-canonical PRC1 allowing o apid H2AK119Ub deposi-
ion (Almeida e al, 2017; Pin acuda e al, 2017; Bousa d e al, 2019;
Cologno i e al, 2019). H2AK119Ub could po en ially be ecognised
by PRC2 co ac o s allowing o de no o deposi ion o H3K27me3
(Blackledge e al, 2014). Impo an ly, Polycomb accumula ion is
dispensable o he ini ia ion o gene silencing, as Xis mu an s lack-
ing he B and C epea egion can s ill induce XCI albei wi h sligh ly
lowe e iciency (Bousa d e al, 2019; Nes e o a e al, 2019).
Ins ead, PRC2 enables s able main enance o he ep essed s a e,
pa icula ly in he con ex o ex a-emb yonic lineages (Kalan y
e al, 2006). The dynamics o accumula ion o PcG ma ks ha e been
s udied using ChIP-seq, al hough i emains unclea how p ecisely
his co ela es wi h Xis RNA coa ing o he Xi. Indeed, he ime ha
lapses be ween Xis RNA up egula ion and H3K27me3 deposi ion is
no known. No how his ela es o o he ep essi e his one ma ks.
Ano he modi ica ion ha apidly accumula es on he Xi is
H4K20me1 (Kohlmaie e al, 2004), al hough i s ole in XCI is no
well unde s ood. H4K20me1 is deposi ed by SETD8 (also called P -
Se 7) (Nishioka e al, 2002) and is hough o egula e a a ie y o
p ocesses ela ing o ansc ip ion, ch omosome condensa ion, DNA
eplica ion and he DNA damage esponse ( e iewed in an Nuland
& Gozani, 2016). Consis en wi h he in ol emen o H4K20me1 in
cell di ision, i ansien ly accumula es du ing G2 and mi osis,
whe eas H4K20me2 and -me3 do no show simila luc ua ions
du ing he cell cycle (Wu e al, 2010). This dynamics is achie ed by
s able exp ession o SETD8 du ing G2/M when i s phospho yla ion
p e en s he in e ac ion wi h he anaphase-p omo ing complex
(APC). Du ing la e mi osis, CDC14 dephospho yla es SETD8 hus
s imula ing i s deg ada ion by APC. On he o he hand, PHF8,
H4K20me1 deme hylase, is deg aded by APC du ing ea ly G2 (Lim
e al, 2013) oge he allowing o e icien deposi ion o H4K20me1
and success ul p og ession h ough mi osis. Fu he mo e,
H4K20me1 accumula es a cen ome es whe e i p omo es kine o-
cho e assembly and hus co ec ch omosome seg ega ion (Ho i
e al, 2014). The egula ion o H4K20me1 le els could also be
achie ed by modula ing he e iciency o u he me hyla ion by
SUV420H1/2 o deme hyla ing H4K20me2/3 by speci ic enzymes.
DPY-21 media es he la e a eac ion in Caeno habdi is elegans, and
i s wo mouse o hologs (RSBN1/RSBN1L) e ain speci ic deme hy-
lase ac i i y in i o (B ejc e al, 2017). Howe e , he biological unc-
ion o hese enzymes emains unknown. The s udy o H4K20me1
du ing XCI has been hampe ed by he ac ha SETD8 is absolu ely
equi ed o he p og ession h ough mi osis and emb yos lacking i
die be o e he onse o XCI (Oda e al, 2009; Shika a e al, 2020).
While he genomic dis ibu ion o H4K20me1 a a s ably inac i a ed
X was p e iously desc ibed (Calab ese e al, 2012), he dynamics o
H4K20me1 accumula ion du ing XCI ini ia ion in ela ion o Xis
RNA and H3K27me3 deposi ion emains unclea . Fu he mo e, he
p ecise dis ibu ion o i s H4K20me1 en ichmen along he X and
he molecula mechanisms unde lying i s dynamic accumula ion on
he Xi ha e no p e iously been in es iga ed.
To add ess hese ques ions, i is impo an o ollow he spa io-
empo al dynamics o his one modi ica ions as hey accumula e on
he X ch omosome in li ing cells. To his end, we ha e p e iously
de eloped gene ically encoded, modi ica ion-speci ic in acellula
an ibodies o min bodies, by using a single-chain a iable agmen
(scF ) o he speci ic an ibody wi h a luo escen p o ein (Sa o e al,
2013; Sa o e al, 2016). The exp ession o min bodies enables acking
o changes in endogenous his one modi ica ion le els wi hou
a ec ing cell cycle p og ession and de elopmen al p ocesses in such
model o ganisms as ission yeas , nema ode, d osophila and zeb a ish
(Sa o e al,2013;A aie al, 2017). He e, o ollow XCI dynamics in
li ing cells, we de eloped a min body speci ic o he PRC2-dependen
H3K27me3. In addi ion, we implemen ed sgRNA-dCas9 sys em o
isualise bo h X ch omosomes in li ing cells (Chen e al, 2013; An on
e al, 2014). We used hese no el ools in combina ion wi h an
H4K20me1-speci ic min body (Sa o e al, 2016; Sa o e al, 2018) and a
me hod o isualise endogenous Xis RNA in li e cells (Masui e al,
2018; Dossin e al, 2020). This allowed us o show ha he e a e s ik-
ing simila i ies in he accumula ion dynamics o bo h H3K27me3 and
H4K20me1 ma ks on he Xi. Fu he ChIP-seq analysis con i med
concu en accumula ion o H4K20me1 and H3K27me3 du ing XCI
albei wi h a he dis inc genomic dis ibu ions and co ela ions wi h
gene silencing dynamics. By using cells exp essing mu an Xis RNA,
we demons a e ha H4K20me1 accumula ion, jus like ha o PcG-
dependen H3K27me3, elies on he Xis RNA B and C epea s. This
also e eals ha bo h ma ks a e dispensable o he ini ia ion o gene
silencing. Taken oge he , ou analysis unco e s ha he H4K20me1
and H3K27me3 his one ma ks accumula e a he Xi wi h compa able
dynamics bu wi h a he di e en dis ibu ions. Reliance o bo h
ma ks on he Xis -BC egion o hei en ichmen sugges s a mechanis-
ic link be ween Polycomb and H4K20me1 accumula ion du ing acul-
a i e he e och oma in o ma ion on he X ch omosome. These
obse a ions also imply ha he gene al unc ion o H4K20me1 may
be in ch oma in compac ion ha cha ac e ises acul a i e he e och o-
ma in a he han in he ini ia ion o gene silencing.
Resul s
H3K27me3min body (2E12LI) speci ically acks H3K27me3in
li ing cells
Me hods o acking polycomb-dependen his one ma ks in li ing
cells ha e been lacking up un il now. Such ools would be hugely
bene icial o he s udy o dynamic epigene ic p ocesses such as XCI.
To add ess his, we decided o gene a e an H3K27me3-speci ic min -
body. Upon sc eening o mouse hyb idomas, we selec ed a 2E12
clone, which exp esses an H3K27me3-speci ic an ibody. We de e -
mined he cDNA sequence o he IgG hea y and ligh chains in 2E12
by deep sequencing (Kuniyoshi e al, 2016) and cloned he a iable
agmen s by PCR o cons uc a min body exp ession ec o (Figs 1A
2o 17 EMBO epo s e51989 |2021 ª2021 The Au ho s
EMBO epo s Sjoe d JD Tjalsma e al
and EV1A). To alida e ha he min body p ope ly localises o
H3K27me3-en iched ch oma in in li ing cells, we used mouse e a o-
ca cinoma cell line MC12 (Abe e al, 1988). MC12 is a mix u e o
diploid and e aploid emale cells ha bou ing one o wo Xi, on
which he min body is expec ed o be concen a ed. When exp essed
in MC12, he supe olde (s ) GFP e sion o 2E12 min body (2E12-
s GFP) localised mos ly in cy oplasm and ended o o m agg ega es
(Fig EV1B, le ), sugges ing ha 2E12-s GFP was no able o unc ion
in cells, p obably due o a ypical p oblem in olding and/o s uc-
u al s abili y associa ed wi h he in acellula exp ession o scF
(Ca aneo & Biocca, 1999; Ewe e al, 2004). To p e en agg ega e
o ma ion, we pe o med PCR-based andom mu agenesis and an-
sien ans ec ion in o MC12 cells. One mu an ha localised p e e en-
ially in he nucleus wi h ocal en ichmen (p esumably Xi) had Me
86 o Leu subs i u ion in he hea y chain (M86L) (Fig EV1B, middle).
Changes in localisa ion sugges ed ha M86L mu a ion imp o ed he
min body unc ion o bind o H3K27me3. To gain insigh in o he
con ibu ion o M86L mu a ion in scF s uc u e, we looked in o he
a omic models buil based on he X- ay s uc u e o he mos simila
scF (Fig EV1C, le ). The modelling indica ed ha Me 86 is loca ed
wi hin a hyd ophobic co e o he hea y chain, and i s subs i u ion o
Leu appea s o ill he space in he co e be e han he o iginal Me
o s eng hen hyd ophobic in e ac ions. We ha e p e iously obse ed
ha he hyd ophobici y o an amino acid in a hyd ophobic co e is
impo an o he unc ionali y o a H4K20me1-min body (Sa o e al,
2016). Howe e , he 2E12M86L-s GFP was s ill no exclusi ely
loca ed in he nucleus and we we e unable o ob ain cells s ably
exp essing i . Thus, we decided o u he s eng hen scF olding
wi h an addi ional mu a ion a Me 158 in he ligh chain’s hyd opho-
bic co e (Fig EV1C, igh ). We cons uc ed h ee mu an s in which
Me 158 was subs i u ed o Ile, Leu and Val in 2E12M86L. Among he
h ee mu a ions, M185I mu an was he mos en iched in he nucleus
wi h e iden oci (Fig EV1B, igh ). To con i m ha nuclea oci o
2E12M86L, M185I (2E12LI)-s GFP ep esen H3K27me3-en iched Xi,
we employed immuno luo escence (IF) on ansien ly exp essing
cells using p e iously cha ac e ised speci ic an ibodies di ec ed
agains H3K27me3 and H3K9me3 (CMA327 and CMA318, espec-
i ely (Chand a e al, 2012)). The nuclea oci we e colocalised wi h
a H3K27me3-speci ic an ibody, bu no wi h H3K9me3-speci ic an i-
body ha is concen a ed in Hoechs -dense pe icen ome ic he e-
och oma in (Fig 1B). These da a a e consis en wi h he min body
oci ep esen ing Xi in li ing cells. The u ili y o 2E12LI min body
was u he demons a ed by ime-lapse imaging o s ably exp essing
MC12 cells, which allowed o he acking o Xi du ing in e phase
and mi osis (Fig 1C and Mo ie EV1).
To u he alida e he speci ici y o 2E12LI o H3K27me3, we
pe o med bo h biochemical and cell-based assays. We i s e alua ed
he binding speci ici y o bac e ially exp essed and pu i ied 2E12LI-
s GFP in i o using a modi ied his one pep ide a ay (Fig EV2A).
H3K27me3-con aining pep ides we e highligh ed o e o he pep ides,
ega dless o he neighbou ing R26 modi ica ions (Fig EV2B). The
2E12LI-s GFP binding was, howe e , occluded by S28 phospho yla-
ion (S28ph), as commonly obse ed o me hyl-speci ic an ibodies
(Kimu a e al, 2008; Hayashi-Takanaka e al,2011;Kimu a,2013).
Nex , we assessed whe he he pu i ied 2E12LI-s GFP can selec i ely
bind o H3K27me3 in cells by manipula ing he le el o speci ic
me hyla ion. HeLa cells we e ans ec ed wi h HaloTag- agged lysine
deme hylases KDM6B and KDM4D and ixed o s aining wi h he
speci ic an ibodies o 2E12LI-s GFP. Consis en ly wi h he subs a e
speci ici y o hese KDMs, IF indica ed ha HeLa cells o e exp essing
Halo-KDM6B and Halo-KDM4D exhibi ed d as ic dec ease o
H3K27me3 and H3K9me3, espec i ely (Fig EV2C and D). The esul
o s aining wi h pu i ied 2E12LI-scF was simila o H3K27me3-speci-
ic an ibody, showing dec eased le els o H3K27me3 by Halo-KDM6B
o e exp ession and no changes in H3K27me3 by Halo-KDM4D (Fig
EV2E and F). Taken oge he wi h he immuno luo escence pa e n
and biochemical analysis, we concluded ha 2E12LI-s GFP selec i ely
binds o H3K27me3 o e he o he modi ica ions including H3K9me3.
Fo con enience, we now call 2E12LI used wi h a luo escen p o ein
as H3K27me3-min body.
Simul aneous acking o X ch omosome loci, H3K27me3and
H4K20me1du ing XCI
In o de o isualise he dynamics o his one modi ica ions du ing
XCI, we decided o de elop a me hod o iden i y he X ch omosome
min body
VHVLFluo escen p o ein
scF
A B
H3K27me3-min body (2E12LI)
H3K9me3 (IF)H3K27me3 (IF)
Hoechs 33342
25:0024:45 25:15
25:4525:30 26:00
VH
VL
FP
H3K27me3
min body
H3K27me3 (IF) H3K9me3 (IF)
min body
C
min body
Figu e 1. Es ablishing H3K27me3min body o isualise H3K27me3in li ing cells.
A Schema ic ep esen a ion o he min body. H3K27me3-speci ic single-chain a iable agmen (scF ) is gene ically used wi h a luo escen p o ein (FP).
B Immuno luo escence (IF) alida ion o min body speci ici y. Mouse MC12 cells, which s ably exp ess H3K27me3-min body (s GFP), a e labelled wi h an ibodies speci ic
o H3K27me3(Cy5) and H3K9me3(Cy3). DNA is s ained wi h Hoechs 33342. Single con ocal sec ions a e shown. A ows ma k Xi. Scale ba =10 μm.
C Time-lapse imaging o a di iding MC12 cell s ably exp essing H3K27me3-min body (s GFP). P ojec ion images o 7con ocal sec ions wi h 2μm in e als a e shown
wi h elapsed ime (hh:mm). Scale ba =10 μm.
ª2021 The Au ho s EMBO epo s e51989 |2021 3o 17
Sjoe d JD Tjalsma e al EMBO epo s
in li e cells. We used an sgRNA-dCas9 (single guide RNA-nuclease
dead Cas9) sys em o label X-linked loci in li ing cells (An on e al,
2014; Ma e al, 2015). We sc eened epe i i e sequences loca ed
speci ically on he mouse X ch omosome and chose 32 mic osa el-
li e sequences as a ge candida es. The exp ession ec o s o
sgRNAs ha bind o he sequences we e cons uc ed and co ans-
ec ed wi h an exp ession ec o o EGFP- agged dCas9 in o MC12
cells. Among hem, 4 sgRNAs (mX2, mX8, mX18 and mX26)
enabled isualising X-linked loci in li ing cells (Fig EV3). We
selec ed mX8 (nea Xis locus) and mX26 (Dxz4 locus) o label X
AB
XIC
Xis
Cen
Dxz4
mouse Ch . X
mX26
mX8
76Mb
103Mb
C
dCas9/mX8+26
dCas9/mX8+26
Min Body (H3K27me3)
-LIF, 3 days +LIF
Min Body
(H3K27me3)
dCas9/mX8+26
Min Body
(H3K27me3)
Min Body
(H4K20me1)
dCas9/mX8+26
Min Body
(H3K27me3)
dCas9/mX8+26
Min Body
(H4K20me1)
0:00 h s 1:00 h s 2:00 h s 3:00 h s 4:00 h s 5:00 h s 6:00 h s
D
Min Body-mChe y
H3K27me3
-LIF
Xi
Xa
H3K27me3
dCas9
3 s GFP
sgRNA
mX26
mX8
Female ESC
Figu e 2. Simul aneous isualisa ion o his one ma ks and X ch omosome loci in li ing cells.
A Schema ic diag am o CRISPR/dCas9-3×s GFP a ge ing loci on mouse X ch omosome. sgRNA mX26 and mX8 a ge mic osa elli e epea s on Dxz4and Xis loci,
espec i ely.
B Expe imen al design: emale ESCs a e enginee ed o s ably exp ess wo sgRNAs, dCas9-3xs GFP and a nuclea localisa ion signal (NLS)- used H3K27me3-min body
(mChe y). Upon ESC di e en ia ion by LIF wi hd awal H3K27me3accumula ion occu s a he inac i a ing X (Xi) bu no a he ac i e X (Xa). G een oci allow ma king
o wo X-linked loci (Dxz4and Xis ).
C Li e imaging o emale mESCs as in (B) in undi e en ia ed condi ions (+LIF) o upon 3days o LIF wi hd awal. Double a owheads ma k Xa and single a owhead
ma ks Xi. Shown a e maximum in ensi y p ojec ions o 11 z-plane con ocal sec ions. Scale ba =10 μm.
D ESCs as in (B) we e enginee ed o exp ess H3K27me3-min body (SNAP/JF646) and H4K20me1-min body (mChe y). Cells we e cul u ed in he absence o LIF o
2days. Time-lapse images o 12 z-plane con ocal s ack we e acqui ed e e y 1h. Maximum in ensi y p ojec ion images a e shown wi h elapse ime (hh:mm).
A owheads ma k Xi. Scale ba =10 μm.
4o 17 EMBO epo s e51989 |2021 ª2021 The Au ho s
EMBO epo s Sjoe d JD Tjalsma e al
ch omosomes and used 3×s GFP- agged dCas9 (3×s GFP-dCas9), o
ampli y he signal o u he analyses using ESCs.
To ack H3K27me3 du ing X inac i a ion, we es ablished emale
mouse ESCs (PGK12.1) exp essing wo sgRNA (mX8 and mX26),
3×s GFP-dCas9 and H3K27me3-min body (mChe y e sion) (Fig 2
A and B). In undi e en ia ed ESCs, X ch omosome loci we e iden i-
ied as 2 pai s o 3×s GFP-dCas9 spo s and H3K27me3-min body
was no en iched in ei he allele (Fig 2C, uppe panels). In con as ,
3 days a e wi hd awal o leukaemia inhibi o y ac o (LIF),
H3K27me3-min body accumula ed a ound one pai o 3×s GFP-
dCas9 spo s (Fig 2C, lowe panels). This esul demons a ed ha
H3K27me3-min body oge he wi h he sgRNA-dCas9 sys em allows
o acking he dynamics o his one modi ica ion in li ing and di -
e en ia ing cells. Since H4K20me1, unlike H4K20me2/3, also
becomes en iched on Xi (Fig EV4A; Kohlmaie e al, 2004), we
sough o e eal he ela i e accumula ion kine ics o he H3K27me3
and H4K20me1 his one modi ica ions du ing XCI. We es ablished
ESCs exp essing he wo sgRNAs, 3×s GFP-dCas9, H3K27me3-min -
body (SNAP-Tag e sion, isualised wi h JF646) and H4K20me1-
min body (mChe y e sion, (Sa o e al, 2016)) and induced di -
e en ia ion (3d LIF wi hd awal). Time-lapse imaging e ealed ha
whene e H3K27me3 accumula ion was isible a he Xi, some le el
o H4K20me1 en ichmen was also disce nible (Fig 2D), sugges ing
ha bo h ep essi e ma ks accumula e concu en ly du ing XCI.
This ini ial analysis in di e en ia ing emale ESCs con i med he
success ul use o bo h H3K27me3- and H4K20me1-min bodies o
ollow en ichmen o hese wo ma ks on he X ch omosome.
Howe e , gi en he asynch onous na u e o andom XCI in di e en-
ia ing ESCs and he a he weak signal om he agged X-linked
loci, i was challenging o dis inguish he Xi om he ac i e X (Xa)
p io o a signi ican accumula ion o bo h ep essi e ch oma in
ma ks on he Xi. To amelio a e ou analysis, we he e o e decided
o ollow Xis RNA i sel in li ing cells alongside H3K27me3 o
H4K20me1.
H3K27me3accumula es concu en ly wi h H4K20me1a he Xi
In o de o ack he ela i e dynamics o H3K27me3, H4K20me1
and Xis RNA du ing XCI, we combined he use o min bodies wi h
he inducible Xis -Bgl sys em (Masui e al, 2018; Dossin e al, 2020;
Fig 3A). The la e model is based on he hyb id (Mus musculus
cas aneus x C57BL/6) TX1072 emale ESCs line allowing o doxycy-
cline (DOX) induc ion o he endogenous Xis gene om C57BL/6
(B6) allele (Schulz e al, 2014). By adding DOX, we can induce Xis
A
DOX inducible
p omo e
Xis BglSL
Xis
BglG-FP
Xis RNA
min body-FP
H4K20me1/H3K27me3
+DOX
Xis
Xi
Xa
H3K27me3/
H4K20me1
01 23 45
1
1.2
ime since Xis accumula ion [h s]
1.4
1.6
No malised en ichmen
Me geBglG (Xis )
min body
(H3K27me3) Me geBglG (Xis )
min body
(H4K20me1)
0:00h s1:00h s2:00h s3:00h s4:00h s5:00h s6:00h s
0:00h s1:00h s2:00h s3:00h s4:00h s5:00h s6:00h s
B
C
H3K27me3
H4K20me1
emale ESC 2i+LIF
*
*
*
Figu e 3. Simul aneous isualisa ion o his one ma ks and Xis RNA in li ing cells.
A Schema ic ep esen a ion o he expe imen al design. Female mouse ESC line was used, in which Xis can be induced om one allele (TX1072) and his allele also
ha bou s an a ay o 18 Bgl s em loops (BglSL) inse ed in o he 7
h
exon o Xis gene. BglG used o a luo escen p o ein (GFP o mChe y) de ec s Xis RNA as i binds
o BglSL. Cells also s ably exp ess a min body allowing he de ec ion o H3K27me3(GFP) o H4K20me1(mChe y).
B Cells we e ea ed wi h DOX o induce Xis exp ession and ime-lapse imaging was pe o med wi h images acqui ed e e y 15 min. Maximum in ensi y p ojec ion
images a e shown wi h elapse ime (hh:mm). A owheads ma k Xi accumula ing Xis RNA. Scale ba =5μm.
C Li e imaging analysis o a e age H3K27me3( ed) and H4K20me1(blue) accumula ion a he Xi. A e age no malised min body en ichmen is shown wi h shading
ep esen ing 25 and 75 qua iles. Signal was calcula ed s a ing om he i s accumula ion o Xis RNA. A leas 30 cells we e analysed. *unpai ed - es
P- alue <0.05.
ª2021 The Au ho s EMBO epo s e51989 |2021 5o 17
Sjoe d JD Tjalsma e al EMBO epo s

exp ession and hus educe he le el o he e ogenei y obse ed
du ing XCI in di e en ia ing ESCs. In addi ion, 18 Bgl RNA s em
loops (BglSL) we e knocked in o he 7
h
exon o DOX-inducible Xis
(Fig EV4B) (Dossin e al, 2020). To ack Xis RNA accumula ion,
BglSL a e isualised by a BglG p o ein used o ei he EGFP o
mChe y and exp essed om he Rosa26 o TIGRE locus (Fig EV4C
and D). The cell lines enabling Xis RNA isualisa ion a e named
TX-Xis -EGFP and TX-Xis -mChe y. These cell lines we e used o
gene a e s ably exp essing PiggyBac ansgenes o he H3K27me3-
o H4K20me1-min bodies (Fig EV4C and D). Two ESC lines, TX-
Xis -EGFP; H4K20me1-mChe y and TX-Xis -mChe y; H3K27me3-
s GFP we e used o subsequen analyses. Impo an ly, he use o
Xis -BglSL o he min bodies did no a ec he e iciency o gene
silencing du ing XCI (Fig EV4E).
In o de o isualise he ea lies e en s ollowing Xis RNA accu-
mula ion, we s a ed li e imaging o ESCs 5 min a e DOX induc-
ion (Fig 3B, Mo ies EV2–EV3). Imaging was pe o med o a leas
6 h in 15-min in e als allowing o single-cell acking and de ailed
quan i a i e analysis. Xis RNA domains appea ed a e 2–6h o
DOX induc ion (Fig EV4F) and we e segmen ed using he Xis -
EGFP/mChe y signal. The min body en ichmen wi hin such
domains was measu ed and ollowed in indi idual cells h oughou
he leng h o he expe imen (Fig 3C). O no e, unlike H3K27me3-
min body, H4K20me1-min body shows inc eased nuclea signal
du ing G2/M phase o he cell cycle, hus acking he oscilla ions in
H4K20me1 le els (Sa o e al, 2016). We hus excluded mi o ic cells
om he analysis due o e y high o e all le els o H4K20me1.
Nex , we ollowed he en ichmen o H3K27me3/H4K20me1 signal
in a leas 30 nuclei indi idually synch onised o he ime poin
when Xis RNA en ichmen was i s obse ed (i.e., ypically 2–6h
a e DOX addi ion). A signi ican en ichmen o bo h his one ma ks
was obse ed wi hin he Xis RNA domain abou 45 min a e Xis
RNA appea ance (adj. P- alue <0.05; Fig 3B and C). Ini ial accu-
mula ion o H3K27me3 and H4K20me1 ollowed e y simila
dynamics. Howe e , a e 2.5 h he wo ma ks signi ican ly di e ged
(Fig 3C). Indeed, H3K27me3 con inued o apidly accumula e, while
he inc ease in H4K20me1 en ichmen signi ican ly slowed down
(Fig 3C). Thus, we ha e success ully pe o med simul aneous ack-
ing o Xis RNA and ep essi e his one ma ks. In summa y, he
en ichmen o bo h H3K27me3 and H4K20me1 is delayed compa ed
o Xis RNA accumula ion by abou 45 min. H3K27me3 shows a
con inued accumula ion o e ime while ha o H4K20me1 e en u-
ally slows down. This e eals ha he wo ma ks sha e simila
dynamics o en ichmen bu only a he ea ly s ages o XCI.
Allele-speci ic na i e ChIP-seq e eals simila i ies be ween
H3K27me3and H4K20me1accumula ion dynamics
Gi en he abo e esul s indica ing ha al hough H3K27me3 and
H4K20me1 a e bo h ini ially en iched soon a e Xis RNA accumula-
ion, hei pa e ns di e ge a e a ew hou s, we decided o in es iga e
he molecula dis ibu ions o bo h hese ma ks along he Xi. To his
end, we pe o med allele-speci ic na i e ChIP-seq (nChIP-seq) o
H4K20me1 in he hyb id emale cell line TX1072 (Schulz e al, 2014).
Thanks o an inducible Xis gene on he B6 ch omosome, DOX addi ion
leads o apid gene silencing and ch oma in ma k al e a ions ha can
be measu ed allelically hanks o he high a e o polymo phism
be ween he B6 and Cas X ch omosomes (Zylicz e al, 2019). Following
DOX ea men , we acked H4K20me1 accumula ion on he Xis RNA-
coa ed B6 allele ac oss i e ime poin s a up o 4-h esolu ion on biolog-
ical duplica es (Fig 4A). We compa ed all ou esul s wi h a ma ched
published da ase o H3K27me3 (Zylicz e al, 2019). Impo an ly, we
alida ed he speci ici y o H4K20me1 an ibody using a pep ide a ay
(Fig EV5A) and con olled o Xis RNA induc ion e iciency (Fig
EV5B). Upon sequencing, eads we e spli acco ding o con en o
allele-speci ic single nucleo ide polymo phisms (SNPs: B6 mapping o
Xi; Cas mapping o Xa). Such allelic in o ma ion was analysed, e eal-
ing p og essi e en ichmen o B6-speci ic eads (o igina ing om Xi)
upon Xis induc ion o H4K20me1 as is he case o H3K27me3 (Fig
EV5C). We analysed ela i e B6- ead en ichmen wi hin 10-kb windows
ac oss he whole X ch omosome no malised o =0 h (Fig 4B). This
e ealed H4K20me1 accumula ion a e 8 h o DOX induc ion, a ime
poin when H3K27me3 also s a s o acc ue bu signi ican ly la e han
ini ial H2AK119Ub en ichmen (Zylicz e al, 2019). The le els o en ich-
men o H4K20me1 a e lowe han o H3K27me3 and seem o each
hei pla eau ea lie . To be e isualise he iming o accumula ion, we
no malised bo h ma ks o hei e ec i e dynamic ange, i.e. o a e age
accumula ion a e 24 h o DOX induc ion (Fig 4C). This con i med ha
a e 8 h o DOX ea men , he e is concu en ini ia ion o H4K20me1
and H3K27me3 accumula ion. To quan i y his u he , we plo ed he
B6- eaden ichmen ela i e o =0 h o each 10-kb window as a unc-
ion o ime and i ed a sigmoidal cu e (see Ma e ials and Me hods).
To ex ac he in o ma ion abou ela i e iming o his one ma k accu-
mula ion, we ob ained he ime when each cu e eaches i s maximum
slope (e ec i e dose 50%, ED50). ED50 analysis e ealed ha
H4K20me1 eaches i s mos e icien accumula ion p io o H3K27me3
(Fig 4D) bu la e han H2AK119Ub (Zylicz e al, 2019). This is in line
wi h H4K20me1 achie ing i s pla eau signi ican ly be o e H3K27me3.
Thus, ou analysis con i med li e imaging obse a ions ha H4K20me1
accumula es concu en ly wi h H3K27me3 bu quickly eaches maxi-
mum en ichmen (Fig 3C).
We nex examined he deg ee o which he dis ibu ions o he
wo ma ks o e lapped ac oss he Xi. We ound ha H4K20me1
becomes p e e en ially en iched a loci in p oximi y o he Xis gene
(g een ba ) as well as a egions ha Xis RNA i s in e ac s wi h (so
called “en y si es”, black ba s (Pin e e al, 2012); Fig 4E). The
ini ial en ichmen o H4K20me1 on he Xis -coa ed X ollows he
same pa e n p e iously de ec ed o no only H3K27me3 bu also o
PRC1-dependen H2AK119Ub (Fig 4E; Zylicz e al, 2019). To con i m
his obse a ion, we in es iga ed ch omosome-wide co ela ion
be ween H3K27me3 and H4K20me1 accumula ion wi hin di e en
genomic windows (Fig 4F). Consis en ly wi h ou ini ial obse a ion,
we ound s iking co ela ion in he accumula ion o bo h ma ks
ac oss in e genic windows as well as bodies o silen genes. In
con as , he bodies o genes ha we e ac i e p io o DOX ea men
(i.e. ini ially ac i e genes) showed a much lowe Pea son’s co ela-
ion wi h ρ=0.2, implying ha bo h ma ks di e in hei co ela ion
wi h ansc ip ion (see below). Wi h he excep ion o ini ially ac i e
genes, nChIP-seq o H4K20me1 e ealed ha i s accumula ion
ollows a s ikingly simila pa e n and dynamics o H3K27me3.
H4K20me1accumula es in e genically and is dispensable o
XCI ini ia ion
The abo e nChIP-seq analysis e ealed simila i ies and di e ences
be ween H4K20me1 and H3K27me3 accumula ion on he X du ing
6o 17 EMBO epo s e51989 |2021 ª2021 The Au ho s
EMBO epo s Sjoe d JD Tjalsma e al
XCI. While in es iga ing genes ha a e ini ially ac i e and hen
become silenced ollowing Xis induc ion, we obse ed ha
H4K20me1 is s ongly biallelically en iched (p ema ked) a an-
sc ibed gene bodies, p io o silencing (Fig 5A). This con as s
s ikingly wi h he dis ibu ion o H3K27me3, which ne e
p ema ks ini ially ac i e genes (Zylicz e al, 2019). We con i med
ha his H4K20me1 en ichmen a ansc ibed genes is a gene al
ea u e bo h on he X ch omosome (Fig 5B) and au osomes (Fig
EV5D). This is in line wi h p e ious epo s indica ing ha
H4K20me1 co ela es wi h ansc ip ional elonga ion and is
en iched a ac i e gene bodies (Beck e al, 2012; Kapoo -Vazi ani &
Ve ino, 2014; Veloso e al, 2014). Upon Xis induc ion (+DOX),
we ound a signi ican accumula ion o H4K20me1 on he Xi a
ini ially ac i e p omo e s and in e genic egions (Fig 5A and B).
This esembles he pa e n obse ed o H3K27me3, wi h he
in iguing ca ea ha H4K20me1 does no seem o u he accumu-
la e a bodies o ini ially ac i e genes, a he , H4K20me1 le els
emain cons an . Thus, H4K20me1 and H3K27me3 accumula ion
pa e ns a e s ikingly simila wi hin in e genic egions bu di e
in he bodies o ini ially ac i e genes (Fig 4F). Nex , o e alua e he
ela ionship be ween H4K20me1 accumula ion and he p ocess o
gene silencing, we sepa a ely analysed genes inac i a ed ea ly and
la e upon DOX ea men (Fig 5C). This e ealed ha while
H4K20me1 de no o accumula ion ollowing Xis induc ion is
es ic ed o in e genic and p omo e egions, i occu s mo e e i-
cien ly in he p oximi y o apidly silenced genes e.g. o Rn 12
(Fig 5A). All in all, hese indings sugges ha p io o XCI,
H4K20me1 co ela es wi h ac i e ansc ip ion (in gene bodies),
bu ollowing Xis induc ion H4K20me1 becomes en iched de no o
a egions su ounding genes ha a e apidly silenced.
To u he explo e he po en ial ela ionship o H4K20me1
en ichmen and H3K27me3 deposi ion du ing XCI, we decided o
in es iga e ch oma in s a es in Xis mu an cell lines. Cu en
models sugges ha H3K27me3 is deposi ed hanks o a complex
H3K27me3
H4K20me1
0.0
0.2
0.4
0 4 8 12162024
ime [h s]
B6 (Xi) ead accumula ion el. o =0h
B
0.0
0.5
1.0
1.5
04812162024
ime [h s]
no malised B6 (Xi) ead accumula ion
C
H3K27me3
H4K20me1
n=9069 n=4971
0
8
16
24
H3K27me3 H4K20me1
D
ED50 [h s]
DOX inducible
p omo e
Xis
Xis
B6
Cas
ESC 2i+LIF
(TX1072)
B6 Xi
Cas Xa
H4K20me1
allele-speci ic
nChIP-seq
DOX ea men
0, 4, 8, 12, 24h
A
2
1
0
B6 (Xi) ead accumula ion el. o =0h
1
0
0 50 100 150
H3K27me3 - 12h DOX
H4K20me1- 12h DOX
E
X ch omosome posi ion [Mb]
*
0.0
-0.1
0.1
0.0 0.1 0.2
Ini ially ac i e gene bodies
ρ = 0.2
0.0 0.1 0.2 0.3
0.0
0.1
Inac i e gene bodies
ρ = 0.656
0.1 0.20.0
0.1
In e genic egions
ρ = 0.828
0.0
H3K27me3 accumula ion
24h s [Lo
g
2 scale]
H4K20me1 accumula ion
24h s [Log2 scale]
H3K27me3 accumula ion
24h s [Lo
g
2 scale]
H3K27me3 accumula ion
24h s [Lo
g
2 scale]
F
Figu e 4. Na i e ChIP-seq e eals H4K20me1and H3K27me3co-accumula ion du ing XCI.
A Schema ic ep esen a ion o he expe imen al design. The hyb id TX1072 mouse emale ESC line was used, in which Xis can be induced om he endogenous B6
allele. Time cou se nChiP-seq o H4K20me1was pe o med and compa ed o a published H3K27me3da ase (GSE116480) (Zylicz e al,2019).
B Quan i ica ion o a e age H4K20me1(blue) and H3K27me3( ed) en ichmen a he Xi (B6allele) compa ed wi h =0hin10-kb windows spanning he whole
ch omosome. Shading is he in e qua ile ange.
C As in (B) bu shown is no malised B6en ichmen o he a e age accumula ion a 24 h.
D Pai wise compa ison o H4K20me1and H3K27me3accumula ion dynamics (ED50) a he X ch omosome. All 10-kb windows wi h ED50 <24 h a e plo ed. * ai ed
Wilcoxon ank-sum es P- alue <0.05.
EH4K20me1(blue) and H3K27me3( ed) accumula ion ac oss he Xi a e 12 h o DOX ea men . The black line is a locally es ima ed sca e plo smoo hing (LOESS)
eg ession on all 10-kb windows (do s). Below each plo shown is he Xis locus (g een ba ) and Xis en y si es (black ba s).
F Co ela ion be ween H3K27me3and H4K20me1accumula ion a e 24 h o DOX ea men a ini ially ac i e gene bodies (le ), inac i e gene bodies (middle) and 10-
kb in e genic windows spanning he X ch omosome ( igh ). All scales a e loga i hmic. All co ela ions (ρ) a e wi h P<0.01 om Pea son’s co ela ion es .
ª2021 The Au ho s EMBO epo s e51989 |2021 7o 17
Sjoe d JD Tjalsma e al EMBO epo s
machine y, in ol ing ini ial PRC1 ec ui men by he B and C
epea s o Xis and subsequen PRC2 ec ui men hanks o
H2AK119Ub en ichmen (Almeida e al, 2017; Pin acuda e al, 2017;
Bousa d e al, 2019; Cologno i e al, 2019; Nes e o a e al, 2019).
We he e o e es ed whe he H4K20me1 en ichmen a he Xi is also
dependen on he B and C egions o Xis RNA. To his end, we
employed p e iously es ablished male ESC lines ha bou ing a DOX-
inducible, endogenous Xis in ei he i s ull leng h o m (Xis FL)o
lacking he B and C epea egions (Xis ΔBC) (Fig 6A), in which
H3K27me3 changes on he Xi had al eady been mapped (Bousa d
e al, 2019). This sys em ecapi ula es hallma ks o XCI, namely
ch omosome-wide Xis coa ing, X-linked gene silencing and he e-
och oma in o ma ion (Bousa d e al, 2019). A e DOX ea men ,
we pe o med IF/RNA FISH expe imen s o quan i y he e iciency
o H4K20me1 en ichmen a he Xis RNA domains (Fig 6B). While
he global le el o H4K20me1 was compa able in bo h cell lines, i
showed, as p edic ed, s iking a iabili y be ween cells due o di -
e en cell cycle s ages. Ne e heless, we obse ed comple e loss o
H4K20me1 en ichmen om he Xis RNA domains when Xis ΔBC
was exp essed; a esul eminiscen o wha was p e iously
obse ed o H3K27me3 (Fig 6C) (Bousa d e al, 2019). These da a
indica e ha H4K20me1 en ichmen depends on he same egions o
Xis RNA ha a e in ol ed in ec ui ing PRC1 and lead o subse-
quen PRC2 ec ui men . Which ac o s di ec ly allow o
H4K20me1 en ichmen a he Xi s ill emains unclea , howe e .
Ano he impo an conclusion om his esul pe ains o gene
silencing. Indeed, he Xis ΔBC RNA is able o ini ia e gene silencing,
albei a sligh ly lowe e iciency (Bousa d e al, 2019). This implies
ha he de no o accumula ion o H4K20me1 is la gely dispensable
o he ini ia ion o gene silencing, simila ly o bo h Polycomb-asso-
cia ed H3K27me3 and H2AK119Ub.
Discussion
We epo he spa io- empo al dynamics o wo ea ly ch oma in
changes, H4K20me1 and H3K27me3 du ing he o ma ion o acul a-
i e he e och oma in in XCI. Using a unique combina ion o li e-cell
imaging and ch oma in p o iling in he same cell sys ems, we e eal
he ela i e iming and dis ibu ions o ch oma in en ichmen o
hese ma ks du ing X inac i a ion. Ou s udy p o ides insigh s in o
he p ocess o XCI and new ools o he s udy o epigene ic
p ocesses in gene al. Indeed, we use he powe ul echnology o
min bodies, gene ically encoded luo escen p obes ha can de ec
speci ic p o eins and hei modi ica ions. In his way, we isualise
he dis ibu ion and le els o speci ic his one modi ica ions, in longi-
udinal single-cell analyses o he epigene ic p ocess o XCI. We
p e iously epo ed on he de elopmen o min bodies speci ic o
H4K20me1 and H3K9ac (Sa o e al, 2013; Sa o e al, 2016). The
H3K27me3 min body de eloped he e will enable he s udy o PRC2-
dependen epigene ic mechanisms beyond XCI.
Ou esul s e eal ha i s de ec able en ichmen o bo h
H3K27me3 and H4K20me1 on he Xis -coa ed X ch omosome occu s
abou 45 min ollowing Xis RNA coa ing, hus ~3–7 h a e DOX
ea men . This sugges s ha bo h ma ks become en iched on he Xi
in a Xis -dependen bu likely indi ec mechanism. I should be
no ed ha p o eins such as SPEN which a e ec ui ed di ec ly by
Xis RNA (Chu e al, 2015; McHugh e al, 2015; Minajigi e al, 2015)
ARn 12
10kb
0h
4h
8h
12h
24h
0h
4h
8h
12h
24h
3-
3-
3-
3-
3-
7-
7-
7-
7-
7-
H3K27me3
H4K20me1
Pgk1
10kb
0h
4h
8h
12h
24h
0h
4h
8h
12h
24h
3-
3-
3-
3-
3-
7-
7-
7-
7-
7-
H3K27me3
H4K20me1
B6 (Xi) Cas (Xa) O e lap (Xi&Xa)
-30kb 30kbTSS TES -30kb 30kbTSS TES
0.01
0.02
0.01
0.02
0.01
0.02
H4K20me1
ini ially ac i e genes inac i e genes
mean B6 (Xi) en ichmen
-30kb 30kbTSS TES -30kb 30kbTSS TES
0.01
0.02
H3K27me3
ini ially ac i e genes inac i e genes
0h 4h 8h 12h 24h
0h 4h 8h 12h 24h
B
C
-30kb 30kbTSS TES -30kb 30kbTSS TES
0.1
0.3
0.5
0.1
0.3
0.5
0h 4h 8h 12h 24h
H4K20me1
ea ly silenced genes la e silenced genes
mean B6 (Xi)
en ichmen
Figu e 5.H4K20me1accumula es in e genically.
A Genome b owse acks showing H3K27me3( op) and H4K20me1(bo om) accumula ion a a gene silenced apidly (Rn 12) o mo e slowly (Pgk1). Allele-speci ic acks
we e o e laid (B6in ed and Cas in blue). No e s ong H4K20me1bi-allelic p ema king a gene bodies.
B A e age H3K27me3( op) and H4K20me1(bo om) en ichmen a he B6allele o e ini ially ac i e o inac i e genes 30 kb a he X ch omosome. Shown a e da a o
all ime poin s.
C A e age H4K20me1en ichmen a he B6allele o e ea ly and la e silenced genes 30 kb a he X ch omosome. Shown is da a o all ime poin s.
Da a in o ma ion: TSS: ansc ip ion s a si e; TES: ansc ip end si e.
8o 17 EMBO epo s e51989 |2021 ª2021 The Au ho s
EMBO epo s Sjoe d JD Tjalsma e al
show immedia e colocalisa ion wi h Xis du ing ini ia ion o XCI
based on li e-cell imaging using he same ESC sys em (Dossin e al,
2020). While he ini ial en ichmen o bo h H4K20me1 and
H3K27me3 ollows simila dynamics, hey soon di e ge wi h
H3K27me3 accumula ing mo e e icien ly. Wha is mo e, H4K20me1
shows a unique cell cycle dynamics wi h ansien ly inc eased le els
du ing G2/M (Rice e al, 2002). In line wi h hese di e ences, ou
nChIP-seq analysis e ealed ha bo h ma ks also show dis inc
genomic dis ibu ion especially in he con ex o ini ially ac i e
genes. Thus, while accumula ion o bo h ma ks depends on a
common Xis -media ed mechanism hei sp eading and long- e m
en ichmen is egula ed independen ly.
He e, we e eal ha bo h H3K27me3 and H4K20me1 accumula-
ion depends on he Xis B+C egion (Bousa d e al, 2019).
Howe e , nei he SETD8 no PRC2 co e componen s ha e been
iden i ied as di ec Xis RNA-binding p o eins (Chu e al, 2015;
McHugh e al, 2015; Minajigi e al, 2015). Ins ead, PRC1 is hough
o be ec ui ed by his egion ia he hnRNPK ac o . Fu he mo e,
SETD8 does no in e ac wi h any o he i e p o eins, including
hnRNPK, ecen ly iden i ied o bind wi h he Xis B+C epea s
(Bousa d e al, 2019). I hus emains unclea how SETD8, i a all,
is ec ui ed o he Xi and whe he i depends on he non-canonical
PRC1 complexes, as p oposed o PRC2 (Almeida e al, 2017; Pin a-
cuda e al, 2017). Indeed, PRC2 ec ui men depends on he non-
canonical PRC1, which e y apidly deposi s H2AK119Ub. This
modi ica ion in u n is p oposed o ec ui PRC2 ia i s co ac o s,
e.g. JARID2 (Coope e al, 2016). Whe he a simila mechanism
media es H4K20me1 en ichmen emains unclea . A ac o po en-
ially in ol ed in indi ec SETD8 ec ui men is a Polycomb-g oup
p o ein L3MBTL1, which is no howe e a co ac o o ei he PRC1
o PRC2. L3MBTL1 is a binde o H4K20me1 (Kalakonda e al,
2008) bu i also in e ac s wi h SETD8 and H3K27me1/2 (Kalakonda
e al, 2008). Thus, L3MBTL1 may ec ui SETD8, when H3K27 s a s
o become me hyla ed by PRC2 bu he H3K27me1/2 en ichmen
du ing XCI has no ye been s udied. In line wi h his model, loss o
EED, a co e PRC2 componen , has p e iously been epo ed o esul
in pa ially educed H4K20me1 en ichmen a he Xi (Schoe ne
e al, 2006). Thus, PRC2 con ibu es o H4K20me1 accumula ion bu
o he pa hways mus also ope a e. Al e na i ely, i is possible ha
SETD8 is no ec ui ed o he Xi; ins ead, i could be he enzymes
ca alysing H4K20me2 deme hyla ion o H4K20me1, which become
en iched du ing XCI. Indeed, in C. elegans dosage compensa ion,
which consis s o a educ ion o X-linked gene ac i i y in XX indi id-
uals, is pa ly dependen on an H4K20me2-speci ic deme hylase
(B ejc e al, 2017). In his sys em, e icien con e sion o H4K20me2
o H4K20me1 a he X ch omosome by DPY-21 p omo es gene
silencing du ing he main enance phase. RSBN1 and RSBN1L a e
mouse o hologs o DPY-21 and seem o e ain speci ic his one
deme hylase ac i i y (B ejc e al, 2017). Whe he RSBN1/RSBN1L
ha e a ole in H4K20 me hyla ion dynamics in mammalian cells and
whe he hese ac o s play a ole in XCI me i s u he in es iga ion.
Finally, i is also possible ha H4K20me1 en ichmen a he Xi is
due o hinde ed con e sion o highe me hyla ion s a es by SUV4-
20H1/2.
Iden i ying he B and C epea s o Xis RNA as he key egion o
H4K20me1 en ichmen du ing XCI sugges s ha his ma k may only
play a mino ole in ini ia ing gene silencing. Indeed, cells exp ess-
ing Xis :ΔBC can induce X ch omosome gene silencing du ing XCI
albei wi h lowe e iciency (Bousa d e al, 2019). The ac ha
H4K20me1 is highly en iched a ansc ibed gene bodies bo h on he
X ch omosome (be o e XCI) and au osomes would also sugges ha
his ma k is no in ol ed in ansc ip ional ep ession. Indeed,
p e ious s udies epo ed a posi i e co ela ion be ween H4K20me1
en ichmen and he a e o ansc ip ional elonga ion (Veloso e al,
2014). SETD8 was also p oposed o acili a e RNA Polyme ase II
(RNAPII) elease om p omo e s (Kapoo -Vazi ani & Ve ino, 2014;
Nikolaou e al, 2017). The ini ial dis ibu ion o H4K20me1 is in line
wi h such indings. Howe e , his ma k also pe sis s a genes which
become apidly silenced du ing XCI, e.g. Rn 12 (Fig 5A). Thus, an-
sc ip ional silencing does no esul in apid H4K20me1 deple ion a
AB
Me geXis RNA H4K20me1
Xis FLXis ΔBC
C
D
C
BE
F
A
A
A
A
A
A
DOX inducible
p omo e
Xis
IF+RNA FISH
+DOX -LIF
48h s
Xis
Xi
H4K20me1
Xa
Male TXY ESC
DE
F
A
A
A
A
A
A
Xis FL
Xis ΔBC
0
25
50
75
100
0
25
50
75
100
% o Xis domains wi h en ichmen
H4K20me1 H3K27me3
(Bousa d e al., 2019)
*
*
Xis FL Xis ΔBC Xis FL Xis ΔBC
Figu e 6.H4K20me1accumula ion elies on Xis B +C epea egion.
A Schema ic ep esen a ion o he expe imen al design. By DOX addi ion, male TXY mouse ESC lines allow o exp ession o ull leng h Xis FL o Xis ΔBC. Cells induced
o 48 h in he absence o LIF we e used o combined IF and RNA FISH.
B Rep esen a i e image o IF/RNA FISH o Xis RNA and H4K20me1in cells exp essing Xis FL o Xis ΔBC. A owheads poin o he Xis RNA domains. Scale ba =5μm.
C G aph ep esen s he mean % S De o Xis RNA domains en iched o H4K20me1(le ) o H3K27me3( igh ) in cells exp essing Xis WT o Xis ΔBC. Shown a e
a e ages om a leas 2independen expe imen s; minimum o 50 Xis RNA domains we e coun ed pe expe imen ; only P- alues co esponding o signi ican
di e ences om unpai ed S uden ’s - es compa ing mu an s o Xis FL a e indica ed as * (P- alue <0.05). Da a o H3K27me3was ex ac ed om published da a
(Bousa d e al,2019).
ª2021 The Au ho s EMBO epo s e51989 |2021 9o 17
Sjoe d JD Tjalsma e al EMBO epo s
Liao Y, Smy h GK, Shi W (2014) ea u eCoun s: an e icien gene al pu pose
p og am o assigning sequence eads o genomic ea u es. Bioin o ma ics
30:923 –930
Lim HJ, Dimo a NV, Tan MK, Sigoillo FD, King RW, Shi Y (2013) The G2/M
egula o his one deme hylase PHF8is a ge ed o deg ada ion by he
anaphase-p omo ing complex con aining CDC20.Mol Cell Biol 33:4166 –
4180
Lyon MF (1962) Sex ch oma in and gene ac ion in he mammalian X-
ch omosome. Am J Hum Gene 14:135 –148
Ma H, Nase i A, Reyes-Gu ie ez P, Wol e SA, Zhang S, Pede son T (2015)
Mul icolo CRISPR labeling o ch omosomal loci in human cells. P oc Na l
Acad Sci USA 112:3002 –3007
Ma in M (2011) Cu adap emo es adap e sequences om high- h oughpu
sequencing eads. EMBne jou nal 17:10
Masui O, Hea d E, Koseki H (2018) Li e imaging o Xis RNA. Me hods Mol Biol
1861:67 –72
McHugh CA, Chen CK, Chow A, Su ka CF, T an C, McDonel P, Pandya-Jones A,
Blanco M, Bu gha d C, Mo adian A e al (2015) The Xis lncRNA in e ac s
di ec ly wi h SHARP o silence ansc ip ion h ough HDAC3.Na u e 521:
232 –236
Minajigi A, F obe g JE, Wei C, Sunwoo H, Kesne B, Cologno i D, Lessing D,
Paye B, Boukhali M, Haas W e al (2015) A comp ehensi e Xis
in e ac ome e eals cohesin epulsion and an RNA-di ec ed ch omosome
con o ma ion. Science 349: aab2276
de Napoles M, Me moud JE, Wakao R, Tang YA, Endoh M, Appanah R,
Nes e o a TB, Sil a J, O e AP, Vidal M e al (2004) Polycomb g oup
p o eins Ring1A/B link ubiqui yla ion o his one H2A o he i able gene
silencing and X inac i a ion. De Cell 7:663 –676
Nes e o a TB, Wei G, Coke H, Pin acuda G, Bowness JS, Zhang T, Almeida M,
Bloechl B, Moind o B, Ca e EJ e al (2019) Sys ema ic allelic analysis
de ines he in e play o key pa hways in X ch omosome inac i a ion. Na
Commun 10:3129
Nikolaou KC, Moulos P, Ha okopos V, Chalepakis G, Talianidis I (2017) Km 5a
con ols hepa ic me abolic pa hways by acili a ing RNA Pol II elease
om p omo e -p oximal egions. Cell Rep 20:909 –922
Nishioka K, Rice JC, Sa ma K, E djumen -B omage H, We ne J, Wang Y,
Chuiko S, Valenzuela P, Temps P, S ewa d R e al (2002) PR-Se 7is a
nucleosome-speci ic me hyl ans e ase ha modi ies lysine 20 o his one
H4and is associa ed wi h silen ch oma in. Mol Cell 9:1201 –1213
an Nuland R, Gozani O (2016) His one H4lysine 20 (H4K20) me hyla ion,
expanding he signaling po en ial o he p o eome one me hyl moie y a
a ime. Mol Cell P o eomics 15:755 –764
Oda H, Okamo o I, Mu phy N, Chu J, P ice SM, Shen MM, To es-Padilla ME,
Hea d E, Reinbe g D (2009) Monome hyla ion o his one H4-lysine 20 is
in ol ed in ch omosome s uc u e and s abili y and is essen ial o mouse
de elopmen . Mol Cell Biol 29:2278 –2295
Penny GD, Kay GF, Shea down SA, Ras an S, B ockdo N (1996) Requi emen
o Xis in X ch omosome inac i a ion. Na u e 379:131 –137
Pin acuda G, Wei G, Rous an C, Ki mizi as BA, Solcan N, Ce ase A, Cas ello A,
Mohammed S, Moind o B, Nes e o a TB e al (2017) hnRNPK ec ui s
PCGF3/5-PRC1 o he Xis RNA B- epea o es ablish polycomb-media ed
ch omosomal silencing. Mol Cell 68:955 –969
Pin e SF, Sad eye RI, Yildi im E, Jeon Y, Ohsumi TK, Bo owsky M, Lee JT
(2012) Sp eading o X ch omosome inac i a ion ia a hie a chy o de ined
Polycomb s a ions. Genome Res 22:1864 –1876
Pla h K, Fang J, Mlyna czyk-E ans SK, Cao R, Wo inge KA, Wang H, de la
C uz CC, O e AP, Panning B, Zhang Y (2003) Role o his one H3lysine 27
me hyla ion in X inac i a ion. Science 300:131 –135
Quinlan AR, Hall IM (2010) BEDTools: a lexible sui e o u ili ies o
compa ing genomic ea u es. Bioin o ma ics 26:841 –842
Rami ez F, Dunda F, Diehl S, G uning BA, Manke T (2014) deepTools: a
lexible pla o m o explo ing deep-sequencing da a. Nucleic Acids Res 42:
W187 –191
Ras an S, Robe son EJ (1985) X-ch omosome dele ions in emb yo-de i ed
(EK) cell lines associa ed wi h lack o X-ch omosome inac i a ion. J
Emb yol Exp Mo phol 90:379 –388
Rice JC, Nishioka K, Sa ma K, S ewa d R, Reinbe g D, Allis CD (2002) Mi o ic-
speci ic me hyla ion o his one H4Lys 20 ollows inc eased PR-Se 7
exp ession and i s localiza ion o mi o ic ch omosomes. Genes De 16:
2225 –2230
Rizk G, La enie D, Chikhi R (2013) DSK: k-me coun ing wi h e y low
memo y usage. Bioin o ma ics 29:652 –653
Robinson MD, McCa hy DJ, Smy h GK (2010) edgeR: a Bioconduc o package
o di e en ial exp ession analysis o digi al gene exp ession da a.
Bioin o ma ics 26:139 –140
Sa o Y, Mukai M, Ueda J, Mu aki M, S ase ich TJ, Ho ikoshi N, Kuji ai T, Ki a
H, Kimu a T, Hi a S e al (2013) Gene ically encoded sys em o ack
his one modi ica ion in i o.Sci Rep 3:2436
Sa o Y, Kuji ai T, A ai R, Asakawa H, Oh suki C, Ho ikoshi N, Yamaga a K,
Ueda J, Nagase T, Ha aguchi T e al (2016) A gene ically encoded p obe
o li e-cell imaging o H4K20 monome hyla ion. J Mol Biol 428:3885 –
3902
Sa o Y, S ase ich TJ, Kimu a H (2018) Visualizing he dynamics o inac i e X
ch omosomes in li ing cells using an ibody-based luo escen p obes.
Me hods Mol Biol 1861:91 –102
Schoe ne S, Sengup a AK, Kubicek S, Mech le K, Spahn L, Koseki H,
Jenuwein T, Wu z A (2006) Rec ui men o PRC1 unc ion a he ini ia ion
o X inac i a ion independen o PRC2and silencing. EMBO J 25:3110 –
3122
Scho a G, Sengup a R, Kubicek S, Malin S, Kaue M, Callen E, Celes e A,
Pagani M, Op a il S, De La Rosa-Velazquez IA e al (2008) A ch oma in-
wide ansi ion o H4K20 monome hyla ion impai s genome in eg i y and
p og ammed DNA ea angemen s in he mouse. Genes De 22:2048 –
2061
Schulz EG, Meisig J, Nakamu a T, Okamo o I, Siebe A, Pica d C, Bo ensz ein
M, Sai ou M, Blu hgen N, Hea d E (2014) The Two ac i e X ch omosomes
in emale ESCs block exi om he plu ipo en s a e by modula ing he
ESC signaling ne wo k. Cell S em Cell 14:203 –216
Shika a D, Yamamo o T, Honda S, Ikeda S, Minami N (2020)H4K20
monome hyla ion inhibi ion causes loss o genomic in eg i y in mouse
p eimplan a ion emb yos. J Rep od De 66:411 –419
Sil a J, Mak W, Z e ko a I, Appanah R, Nes e o a TB, Webs e Z, Pe e s AH,
Jenuwein T, O e AP, B ockdo N (2003) Es ablishmen o his one h3
me hyla ion on he inac i e X ch omosome equi es ansien ec ui men
o Eed-Enx1polycomb g oup complexes. De Cell 4:481 –495
S ase ich TJ, Hayashi-Takanaka Y, Sa o Y, Maeha a K, Ohkawa Y, Saka a-
Sogawa K, Tokunaga M, Nagase T, Nozaki N, McNally JG e al (2014)
Regula ion o RNA polyme ase II ac i a ion by his one ace yla ion in
single li ing cells. Na u e 516:272 –275
T oje P, Li G, Sims 3 d RJ, Vaque o A, Kalakonda N, Boccuni P, Lee D,
E djumen -B omage H, Temps P, Nime SD e al (2007)L3MBTL1,a
his one-me hyla ion-dependen ch oma in lock. Cell 129:915 –928
Veloso A, Ki kconnell KS, Magnuson B, Biewen B, Paulsen MT, Wilson TE,
Ljungman M (2014) Ra e o elonga ion by RNA polyme ase II is associa ed
wi h speci ic gene ea u es and epigene ic modi ica ions. Genome Res 24:
896 –905
16 o 17 EMBO epo s e51989 |2021 ª2021 The Au ho s
EMBO epo s Sjoe d JD Tjalsma e al

Wilson DS, Kee e AD (2000) Random mu agenesis by PCR. Cu P o oc Mol
Biol 51:8.3.1–8.3.9
Wu S, Wang W, Kong X, Congdon LM, Yokomo i K, Ki schne MW, Rice
JC (2010) Dynamic egula ion o he PR-Se 7his one
me hyl ans e ase is equi ed o no mal cell cycle p og ession. Genes
De 24:2531 –2542
Wu z A, Rasmussen TP, Jaenisch R (2002) Ch omosomal silencing and
localiza ion a e media ed by di e en domains o Xis RNA. Na Gene 30:
167 –174
Zylicz JJ, Die mann S, Gunesdogan U, Hacke JA, Cougo D, Lee C, Su ani MA
(2015) Ch oma in dynamics and he ole o G9a in gene egula ion and
enhance silencing du ing ea ly mouse de elopmen . Eli e 4:e09571
Zylicz JJ, Bousa d A, Zume K, Dossin F, Mohammad E, da Rocha ST, Schwalb
B, Syx L, Dingli F, Loew D e al (2019) The implica ion o ea ly ch oma in
changes in X ch omosome inac i a ion. Cell 176:182 –197
Zylicz JJ, Hea d E (2020) Molecula mechanisms o acul a i e
he e och oma in o ma ion: an X-ch omosome pe spec i e. Annu Re
Biochem 89:255 –282
License: This is an open access a icle unde he
e ms o he C ea i e Commons A ibu ion License,
which pe mi s use, dis ibu ion and ep oduc ion in
any medium, p o ided he o iginal wo k is p ope ly
ci ed.
ª2021 The Au ho s EMBO epo s e51989 |2021 17 o 17
Sjoe d JD Tjalsma e al EMBO epo s