*Fo co espondence: me ja.
heinaniemi@ue . i (MH); olli.lohi@
s a .u a. i (OL)
†
These au ho s con ibu ed
equally o his wo k
Compe ing in e es s: The
au ho s decla e ha no
compe ing in e es s exis .
Funding: See page 21
Recei ed: 16 No embe 2015
Accep ed: 09 June 2016
Published: 19 July 2016
Re iewing edi o : Sco A
A ms ong, Memo ial Sloan
Ke e ing Cance Cen e , Uni ed
S a es
Copy igh Heina
¨niemi e al.
This a icle is dis ibu ed unde
he e ms o he C ea i e
Commons A ibu ion License,
which pe mi s un es ic ed use
and edis ibu ion p o ided ha
he o iginal au ho and sou ce a e
c edi ed.
T ansc ip ion-coupled gene ic ins abili y
ma ks acu e lymphoblas ic leukemia
s uc u al a ia ion ho spo s
Me ja Heina
¨niemi
1
*, Tapio Vuo enmaa
1,2†
, Susanna Teppo
3†
,
Minna U Kaikkonen
2†
, Ma ia Bou y-Lii and
1
, Juha Meh onen
1
, Hen i Niskanen
2
,
Vasilios Zacha iadis
4
, Saa a Laukkanen
3
, Thomas Liuksiala
3
, Kaisa Tei inen
3
,
Olli Lohi
3,5
*
1
School o Medicine, Uni e si y o Eas e n Finland, Kuopio, Finland;
2
A. I. Vi anen
Ins i u e o Molecula Sciences, Uni e si y o Eas e n Finland, Kuopio, Finland;
3
School o Medicine, Uni e si y o Tampe e, Tampe e, Finland;
4
Depa men o
Molecula Medicine and Su ge y, Ka olinska Ins i u e , S ockholm, Sweden;
5
Tampe e Uni e si y Hospi al, Tampe e, Finland
Abs ac P og ession o malignancy o o e disease equi es mul iple gene ic hi s. Ac i a ion-
induced deaminase (AID) can d i e lymphomagenesis by gene a ing o - a ge DNA b eaks a loci
ha ha bo highly ac i e enhance s and display con e gen ansc ip ion. The i s ac i e
ansc ip ional p o iles om acu e lymphoblas ic leukemia (ALL) pa ien s acqui ed he e e eal
s iking simila i y a s uc u al a ia ion (SV) si es. Speci ic ansc ip ional ea u es, namely
con e gen ansc ip ion and Pol2 s alling, we e de ec ed a b eakpoin s. The o e lap was mos
p ominen a SV wi h ecogni ion mo i s o he ecombina ion ac i a ing genes (RAG). We p esen
signal ea u e analysis o de ec ulne able egions and quan i ied om human cells how
con e gen ansc ip ion con ibu es o R-loop gene a ion and RNA polyme ase s alling. Wide
s alling egions we e cha ac e ized by high DNAse hype sensi i i y and unusually b oad H3K4me3
signal. Based on 1382 p e-B-ALL pa ien s, he ETV6-RUNX1 usion posi i e pa ien s had o e en-
old ele a ion in RAG1 while high exp ession o AID ma ked p e-B-ALL lacking common
cy ogene ic changes.
DOI: 10.7554/eLi e.13087.001
In oduc ion
In p ecu so lymphoblas ic leukemia, p ima y gene ic lesions o en a ise in u e o (Wiemels e al.,
1999;Mo i e al., 2002;Maia e al., 2003,Ba eman e al., 2015), while he onse o o e disease
equi es addi ional gene ic al e a ions. Whole-genome sequencing (WGS) o ETV6-RUNX1 (also
known as TEL-AML1) posi i e acu e leukemias sugges ed ha he seconda y lesions a e p edomi-
nan ly caused by o - a ge ac i i y o he RAG complex (Papaemmanuil e al., 2014). In a simila
ashion, he exp ession o he AID complex in mo e ma u e B cells is implica ed in genomic ins abil-
i y and de elopmen o lymphomas (Meng e al., 2014;Qian e al., 2014;Robbiani e al. 2015). To
da e, WGS in leukemia ha e been epo ed om se e al p e-B-ALL sub ypes (Ande sson e al.,
2015;Holm eld e al., 2013;Paulsson e al., 2015;Zhang e al., 2012), esul ing in a comp ehen-
si e cha ac e iza ion o he unde lying gene ic al e a ions. The e o e, he esea ch ocus on leukemia
gene ics is mo ing in o cha ac e iza ion o he mechanisms by which hese lesions occu and he con-
sequences o he esul ing clonal he e ogenei y.
An igen ecep o genes a e assembled om disc e e gene segmen s by RAG-media ed V(D)J
ecombina ion a si es o ecombina ion signal sequences (RSS) du ing ea ly lymphocy e
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 1 o 26
RESEARCH ARTICLE
de elopmen (Gelle 2002;Scha z and Swanson, 2011). Cells inco po a e mul iple s a egies o
con ol he ac ion o he RAG complex o app op ia e genomic loci: he exp ession o RAG1 and
RAG2 is limi ed o p ecu so s ages o lymphocy es, he ac i i y o he complex is a enua ed du ing
S-phase o cell cycle, and RAG clea age is di ec ed owa ds RSS pai con aining sequences
(Scha z and Swanson, 2011). The engagemen o RAG2 is u he limi ed by he his one modi ica-
ion H3K4me3, which is ypically ound a ansc ip ion s a si es (TSS) (Ma hews e al., 2007;
Teng e al., 2015). Howe e , RSS and RSS-like mo i s a e ound only a a ound 7–40% o b eak-
poin s a SV (genomic imbalance, ansloca ion o in e sion) si es (Ande sson e al., 2015;
Papaemmanuil e al., 2014). Fu he mo e, he RSS mo i s and H3K4me3 occu equen ly in he
genome sugges ing ha addi ional ea u es, possibly e en addi ional complexes including AID
(Swamina han e al., 2015), a e ele an o he gene ic ins abili y unde lying leukemia SV.
In lymphomas, AID o - a ge e ec s localize o in agenic supe -enhance (SE) and p omo e
a eas cha ac e ized by ansc ip ion om bo h s ands, i.e. con e gen ansc ip ion (con T)
(Meng e al., 2014). No ably, VH gene segmen ecombina ion by RAG a he IgH locus coincides
wi h sense- and an isense ansc ip ion (Bolland e al., 2004), which could be ele an also a o - a -
ge si es. Secondly, s alled polyme ases, which a e ound a exons, R-loops and ac i ely paused a
TSS egions (Jonke s and Lis, 2015), expose single s anded DNA, ec ui ing AID ia Sp 5 binding
(Pa i e al., 2010). Fu he mo e, he polyme ase complex displaces nucleosomes comple ely o
pa ially ( he H2A/H2B moie y), which in i o p omo es clea age by RAGs (Be ing on and Boyes,
2013). Despi e hese in iguing indings, he ele ance o ansc ip ion-coupled p ocesses has no
been sys ema ically cha ac e ized, and he clinical ele ance o RAG and AID exp ession in he di e -
en leukemia sub ypes emains unclea . RNA polyme ases engaged in o p ima y ansc ip ion ac oss
he genome can be measu ed using Global-Run-On sequencing (GRO-seq) (Kaikkonen e al.,
2013). The e o e, his me hod is ideally sui ed o dis inguish ea u es o ansc ip ion a SV si es,
including con T and RNA polyme ase s alling. To his end, we acqui ed he i s pa ien p o iles o
nascen ansc ip ional ac i i y in leukemic blas s ep esen ing se en cy ogene ic subg oups and pe -
o med in eg a i e analysis o a ious genome-wide p o iles and pa ien ansc ip omes.
eLi e diges Some o he mos common cance s ound in child en a e called p ecu so
leukemias, which may s a o de elop be o e bi h. Cance ous cells o en con ain al e a ions o he
gene ic in o ma ion in hei DNA. In p ecu so leukemias, he mos common gene ic changes in ol e
dele ing, adding o ea anging segmen s o he DNA sequence.
Se e al esea che s ha e sequenced he en i e DNA o childhood leukemia cells, wi h he esul
ha almos all o he gene ic al e a ions linked o hese condi ions ha e been ca alogued. These
e o s ha e shown ha ce ain DNA egions a e pa icula ly a ec ed by mu a ions, bu no one
knows why e o s occu so equen ly in hese egions.
Recen e idence also sugges s ha ansc ip ion – he p ocess o p oducing use ul molecules
om a s e ch o DNA – can play a ole in gene a ing gene ic al e a ions. Heina
¨niemi e al. ha e now
used a echnique called global un-on sequencing o measu e he ex en o ansc ip ion in many
di e en ypes o leukemia cells. This e ealed ha in he e o -p one DNA egions, wo p ocesses –
called con e gen ansc ip ion and ansc ip ional s alling – in e e e wi h ansc ip ion. Bo h
p ocesses empo a ily lea e he no mally double-s anded DNA unzipped as wo single s ands and
ee o nucleosomes, which makes DNA mo e ulne able o b eaking. This would explain how pieces
o DNA migh be los , added, o mo ed o cause he gene ic e o s ha lead o leukemia.
Fu he in es iga ion e ealed ha wo p o ein complexes called RAG and AID, which ea ange
segmen s o DNA in immune cells, a e likely o cause he e o s in he ulne able DNA egions.
Di e en amoun s o RAG and AID we e p esen in di e en sub ypes o leukemia cells, and hese
amoun s also a ied wi h he isk classi ica ion o he disease. Fu he s udies a e now needed o
in es iga e he exac oles o hese p o ein complexes. This could e en ually help scien is s de ise
s a egies o p o ec he DNA o people wi h leukemia om hese e o s, which could educe he
isk o he cance eoccu ing.
DOI: 10.7554/eLi e.13087.002
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 2 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
Figu e 1. In eg a i e analysis o ansc ip ion and high- ecu ence SV si es highligh s no el ansc ibed egions. (A) WGS da a om he ETV6-RUNX1 (51
cases; Papaemmanuil e al., 2014), high hype diploid (16 cases; Paulsson e al., 2015), hypodiploid (20 cases; Holm eld e al., 2013) and MLL-
Figu e 1 con inued on nex page
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 3 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
Resul s
In eg a i e analysis o ansc ip ion and genomic ins abili y in leukemic
cells
T ansc ip ional ac i i y om ALL cells ep esen ing se en di e en p e-B-ALL cy ogene ic sub ypes
was assayed using GRO-seq (bo h p ima y pa ien and cell line samples, see Supplemen a y ile 1
and Ma e ials and me hods), and join ly analyzed wi h WGS da a om he ETV6-RUNX1 (51 cases;
Papaemmanuil e al., 2014), high hype diploid (HeH, 16 cases; Paulsson e al., 2015), hypodiploid
(20 cases; Holm eld e al., 2013) and MLL- ea anged (22 cases a diagnosis and 2 elapses;
Ande sson e al., 2015) sub ypes o p ecu so B-ALL. GRO-seq signals and b eakpoin da a a e
shown in Figu e 1— igu e supplemen 1 a he CDKN2A locus, a signi ican SV si e in childhood
ALL (Sulong e al., 2009).
To sys ema ically iden i y egions wi h high equency o SV ac oss he genome, opologically-
associa ed domains (TADs) we e e ie ed based on HiC da a om B-lymphoid lineage cells
(Rao e al., 2014). TADs e lec he h ee-dimensional s uc u e o ch oma in. These na u al bound-
a ies o ansc ip ional ac i i y we e used o di ide he ch omosomes in o sub egions o analysis
(see Figu e 1—sou ce da a 1 and Ma e ials and me hods). To link ypical ansc ip ional ac i i y pa -
e ns and ho spo s o genomic ins abili y, we ela ed he b eakpoin equency wi h ch oma in
domains, as illus a ed in Figu e 1A (see also Figu e 1— igu e supplemen 2).
The mos equen SV egions encompass no el ansc ibed egions
An inc easing end o ansc ip ional ac i i y was obse ed when TADs we e compa ed based on
b eakpoin equency qua iles (see Ma e ials and me hods, Figu e 1— igu e supplemen 3). TADs
wi h highes SV coun a e shown in Figu e 1 (see also Figu e 1—sou ce da a 1 and Figu e 1—
Figu e 1 con inued
ea anged (22 cases; Ande sson e al., 2015) sub ypes o p ecu so B-ALL was in eg a ed wi h p o iles o ansc ip ional ac i i y assayed using GRO-
seq om ALL pa ien and cell line samples (see also Figu e 1— igu e supplemen 1 and Supplemen a y ile 1). HiC da a om B-lymphoid cells
(Rao e al., 2014) was used o de ine TADs based on he HiC in e ac ion equency, shown as g ey scale hea map, in o de o dis inguish TADs wi h
highes equency o SV. (B) The PAX5 and ZCCHC7 loci a e loca ed in he TAD shown ha has high SV equency in hype diploid, ETV6-RUNX1- and
MLL- usion posi i e pa ien s (4, 20 and 6 b eakpoin s, espec i ely, Figu e 1—sou ce da a 1). The GRO-seq signal p o iles om h ee p e-B-ALL
cy ogene ic sub ypes and no mal B-lymphoblas oid cells a e displayed as indica ed in he igu e (see also Figu e 1— igu e supplemen 4 and
Figu e 2— igu e supplemen 2). The y-axis shows he no malized ead densi y (plus s and in ed, minus s and in blue). con T egions egions a e
indica ed in pu ple and leukemia b eakpoin s in ed. The TSS egion o PAX5 o e laps con T ha co-localized wi h an in agenic SE (B-lymphoblas oid
H3K27ac ack is shown a he bo om). (C) A TAD wi h he same numbe o b eakpoin s (20) in ETV6-RUNX1 pa ien s is shown wi h signal om REH
cells (see also Figu e 1— igu e supplemen 4). Genomic anno a ions include he loca ion o GENCODE ansc ip s (in g een). A s ong ansc ip ion
signal is isible ha spans app oxima ely 500 kb nea he TAD bounda y, lacking anno a ed ansc ip s. A zoom-in panel shows he mos ecu en SV
si e. (D) The TAD isualized ep esen s a genomic egion ha ha bo s mos SV in HeH (see Figu e 1— igu e supplemen 5 o he hypodiploid SV
ho spo ). The GRO-seq signal ( ack om pa ien 1) indica es a no el locus wi h abundan ansc ip ion in leukemic samples ( e e o Figu e 1— igu e
supplemen 4 o all GRO-seq p o iles). The highes ecu ence o SV occu s a he con T o e lapping mid- egion (zoom-in panel), which has also wo
ETV6-RUNX1 b eakpoin s.
DOI: 10.7554/eLi e.13087.003
The ollowing sou ce da a and igu e supplemen s a e a ailable o igu e 1:
Sou ce da a 1. Iden i ied opologically associa ed domains.
DOI: 10.7554/eLi e.13087.004
Figu e supplemen 1. T ansc ip ional ac i i y in leukemic cells om pa ien s, cell lines and p ima y heal hy B-lineage cells is cap u ed in GRO-seq
signals.
DOI: 10.7554/eLi e.13087.005
Figu e supplemen 2. Summa y o da a used in he in eg a i e analysis.
DOI: 10.7554/eLi e.13087.006
Figu e supplemen 3. T ansc ip ional ac i i y in TADs binned by b eakpoin equency.
DOI: 10.7554/eLi e.13087.007
Figu e supplemen 4. Da a om all signal acks o egions displayed in Figu e 1.
DOI: 10.7554/eLi e.13087.008
Figu e supplemen 5. TAD wi h equen SV in hypodiploid pa ien s.
DOI: 10.7554/eLi e.13087.009
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 4 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
igu e supplemen 4). The PAX5 and ZCCHC7 genes a e loca ed wi hin a TAD egion wi h 20 b eak-
poin s in he ETV6-RUNX1, 4 in HeH and 6 in MLL sub ype (excluding he MLL- usion i sel )
(Figu e 1B). F equen SV we e also ound in TADs wi h no anno a ed coding genes (Figu e 1C, 20
b eakp in ETV6-RUNX1; Figu e 1D, 4 b eakp in HeH), ye GRO-seq exhibi ed ansc ip ion signal
spanning se e al hund ed housand base pai s in bo h egions, ypical o long non-coding ansc ip s
(Sun e al., 2015). The e was e idence o non-coding ansc ip s, based on Re seq and GENCODE,
bu none ma ched he same loca ion ( e e o Supplemen a y ile 2 o all genomic coo dina es
shown; a TAD wi h equen SV in hypodiploid sub ype is shown in Figu e 1— igu e supplemen 5).
The nascen ALL ansc ip omes hus e eal no el ansc ibed egions as ecu en SV-associa ed
ho spo s in he wo mos common ALL sub ypes.
Con e gen ansc ip ion and RNA polyme ase s alling a e p e alen a
genomic egions wi h equen b eakpoin e en s
The p e ailing no ion is ha ac i e ansc ip ion s a si es (TSS) in p e-B cells a e suscep ible o RAG
o - a ge ing due o he H3K4me3 ch oma in ma k (Ma hews e al., 2007;Teng e al., 2015). How-
e e , we no iced ha he ecu en b eakpoin s o en lied se e al kb downs eam o TSS, as
highligh ed in Figu e 1B and D (see inse s), and coincided wi h simul aneous ansc ip ion on bo h
s ands, ie. con T spanning a minimum o 100 bp. In close examina ion o he signal da a om leu-
kemia SV ho spo s, many o hese egions likely co espond o ansc ip ion om in agenic
enhance s ha gene a e enhance RNAs (eRNA) ha a e ypically a ew kb in size (Kaikkonen e al.
2013). In ag eemen , a signi ican en ichmen o b eakpoin s in enhance s o e lapping wi h con T
was obse ed (hype geome ic es P=0.00012 o in e genic and P=4.6e-08 o all enhance s iden i-
ied based on eRNA signal, see Ma e ials and me hods and Figu e 2—sou ce da a 1). An o e lap-
ping eRNA ansc ip a he TSS egion o PAX5, con i med by he ac i e enhance ch oma in
ma ke H3K27ac, led o con T ex ending nea ly 20 kb, wi h SV si es loca ed be ween 3.7–9.7 kb
downs eam o he TSS (Figu e 1B, see inse ).
Secondly, con T in he icini y o in agenic b eakpoin s was o en associa ed wi h localized ele a-
ion in he GRO-seq signal, as exempli ied a he ZCCHC7 and RAG loci (Figu e 2A, see also Fig-
u e 2— igu e supplemen 1). The obse ed signal ea u es we e highly ep oducible be ween
biological eplica es and sha ed among a subse o cy ogene ic g oups (Figu e 2— igu e supple-
men 2). We hypo hesized ha hey ep esen RNA polyme ase II (Pol2) s alling e en s. P e ious
analyses o Pol2 s alling ha e ocused on p omo e p oximal egions (Adelman and Lis, 2012). To
examine such e en s genome-wide and ac oss gene bodies, we de eloped a gene al analysis
app oach ha iden i ies change poin s wi hin gene egions and epo s hose wi h high ele a ion in
he signal le el (see Ma e ials and me hods and Figu e 2—sou ce da a 1 o he iden i ied egions)
(Killick e al., 2012). As addi ional con i ma ion, we analyzed s alling om Pol2 ChIP-seq in he REH
and Nalm6 cell lines (Figu e 2A). To dis inguish be ween di e en Pol2 complexes (Zhou e al.,
2012), an ibodies agains he se ine 2 o se ine 5 phospho yla ed Pol2 we e used (see
Ma e ials and me hods).
Genome-wide analysis o con T and Pol2 s alling (see Ma e ials and me hods and Figu e 2—
sou ce da a 1) subs an ia ed he ele ance o hese obse a ions: conside ing he b eakpoin e-
quency pe TAD size, he op anked TADs in each ALL sub ype ep esen ed genomic egions wi h
abundan con T and Pol2 s alling (Figu e 2B). Signi ican en ichmen was con i med o he uppe
qua iles (hype geome ic es P=0.00038 in ETV6-RUNX1, P=0.00018 in hype diploid, P=0.028 in
hypodiploid and P=0.00004 in MLL- ea anged). The inc eased o e lap was ound o b eakpoin s
wi h and wi hou RSS mo i s (deno ed as R-b eakp and NR-b eakp, see Figu e 2— igu e supple-
men 3 and Ma e ials and me hods) and i was p ese ed when o al ansc ip ional ac i i y was con-
side ed (Figu e 2— igu e supplemen 4). Fu he mo e, he dis inc ansc ip ional p o ile o
emb yonic s em cells (ES) had lowe o e lap (Figu e 2— igu e supplemen 5).
Fo compa ison, ch oma in segmen a ion o B-lymphoid cells was simila ly analyzed (see Fig-
u e 1—sou ce da a 1 and Figu e 2—sou ce da a 1). TADs wi h high numbe o b eakpoin s consis-
en ly had signi ican o e lap wi h ch oma in segmen s ep esen ing ac i e ansc ip ion ( e e o
Figu e 2—sou ce da a 1), suppo ing a ansc ip ion-coupled mechanism o he obse ed gene ic
ins abili y. We hen dis inguished egions wi h o e lap o he ansc ip ional ea u es de ined he e
wi hin ac i e p omo e s and enhance s. Compa ing hese agains he TAD SV equency qua iles
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 5 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine

Figu e 2. Con e gen ansc ip ion and Pol2 s alling cha ac e ize genomic egions wi h high numbe o b eakpoin e en s. (A) The GRO-seq signal in
he ETV6-RUNX1 posi i e REH cell line is shown o exempli y he co-occu ence o con T (in pu ple) and local ele a ion in GRO-seq signal (Pol2 s alling,
Figu e 2 con inued on nex page
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 6 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
(Figu e 2— igu e supplemen 6), as be o e, e ealed he mos p onounced en ichmen in con T/
Pol2 s all o e lapping egions.
Nex , we se ou o de ine wha may link con T and Pol2 s alling egions wi h AID and RAG
ec ui men . The signal ea u e de ec ion o con T (as in Meng e al., 2014) and Pol2 s alling (as
de ined he e) enables his on a genome-wide le el.
R-loop o ma ion and con e gen ansc ip ion co-occu wi h Pol2
s alling
RNA polyme ases a e expec ed o s all a egions ha bo ing R-loop o ming sequences (RLFS)
(Skou i-S a haki e al., 2014a;Jenja oenpun e al., 2015). The sensi i i y o DNA sequence o
o m R-loops can be compu a ionally p edic ed (Jenja oenpun e al., 2015) (see
Ma e ials and me hods). These RLFS mo i con aining egions exhibi ed a signi ican ly highe o e lap
wi h Pol2 s alling si es when compa ed o andom in agenic egions (Figu e 3B, empi ical P<0.001
in B-lineage and ES cells). A highly conco dan local RLFS mo i densi y and GRO-seq signal p o ile
was obse ed ac oss gene egions (Figu e 3— igu e supplemen 1A and B). The p o iles peaked
nea TSS, whe e he p esence o RLFS mo i s led o a signi ican ele a ion in he median GRO-seq
signal le el (Figu e 3— igu e supplemen 1, 2.1- old inc ease in B-lineage cells, Wilcoxon ank sum
es P<2.2e-16, 95% CI 2.1–2.3). As a second mechanism, collisions due o con T may hal ansc ip-
ion (P esco and P oud oo , 2002) in a dynamic and cell-speci ic manne . Acco dingly, highe an i-
sense signal a con T egions (see Ma e ials and me hods) inc eased he o e lap wi h Pol2 s alling
si es on he sense s and (Figu e 3B), in iguingly exceeding ha obse ed o RLFS mo i s
(Figu e 3A).
As an addi ional expe imen al alida ion o R-loops, we used DNA-RNA-immunop ecipi a ion
sequencing (DRIP-seq) esul s om ES cells (see Ma e ials and me hods) ha co espond o de ec-
ion o DNA-RNA hyb ids (Ginno e al., 2013). The 2.1- old ele a ion in median DRIP-seq signal con-
i med ha RLFS mo i s a o DNA-RNA hyb id o ma ion (Figu e 3C, Wilcoxon ank sum es
P<2.2e-16, 95% CI 2.0–2.1, see Figu e 3—sou ce da a 2 o each eplica e). Mo eo e , DRIP-seq
quan i ica ion showed 1.7- old highe median signal a con T-posi i e TSS egions (Figu e 3D, Wil-
coxon ank sum es P<2.2e-16, 95% CI 1.6–1.7). These esul s demons a e ha ansc ip ion s all-
ing occu s a RLFS and con T egions in mammalian cells ha associa es wi h R-loop o ma ion
based on e idence om ES cells.
Figu e 2 con inued
in ligh blue) a bo h R- and NR-b eakp (in ed and b own, espec i ely) ha eside wi hin in onic (ZCCHC7), TSS (RAG2) o pu a i e enhance egions
(RAG2). The ele a ed signal is also isible in Pol2 ChIP-seq signal (Pol2 S2P in g een, Pol2 S5P in o ange, inpu in g ey). See also Figu e 2— igu e
supplemen 1. The pe cen age o TAD spanned by con T (in B) o Pol2 s alling (in C) in p e-B/B-lymphoid cells is summa ized as boxplo s om TADs
di ided in o qua iles based on numbe o b eakpoin s pe bp (see also Figu e 1— igu e supplemen 3,Figu e 2— igu e supplemen 3–6). The
qua ile anges a e o exclusi e lowe and inclusi e uppe alue in he ange, as indica ed. Re e o Figu e 2—sou ce da a 1 o s a is ical analysis.
DOI: 10.7554/eLi e.13087.010
The ollowing sou ce da a and igu e supplemen s a e a ailable o igu e 2:
Sou ce da a 1. Iden i ied con T and Pol2 s alling egions.
DOI: 10.7554/eLi e.13087.011
Figu e supplemen 1. Da a om all signal acks o egions displayed in Figu e 2.
DOI: 10.7554/eLi e.13087.012
Figu e supplemen 2. The GRO-seq signal om eplica e samples gene a ed om ALL cells displayed a he PAX5/ZCCHC7 locus.
DOI: 10.7554/eLi e.13087.013
Figu e supplemen 3. Signal ea u e span o TADs o de ed sepa a ely by R-b eakp o NR-b eakp equency.
DOI: 10.7554/eLi e.13087.014
Figu e supplemen 4. Signal ea u e span no malized by o al ansc ibed a ea o TADs so ed by b eakpoin equency.
DOI: 10.7554/eLi e.13087.015
Figu e supplemen 5. O e lap o TADs wi h con T in ES cells.
DOI: 10.7554/eLi e.13087.016
Figu e supplemen 6. TAD analysis using p omo e and enhance ch oma in segmen s s a i ied by con T and Pol2 s alling.
DOI: 10.7554/eLi e.13087.017
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 7 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
Figu e 3. Indica ion o ansc ip ion-coupled gene ic ins abili y a leukemia SV ho spo s lacking RSS mo i s. (A) O e lap be ween RLFS mo i ha bo ing
in agenic egions and de ec ed Pol2 s alling si es in B-lineage and ES cells. The high o e lap o RLFS-posi i e egions is s a is ically signi ican
compa ed o andom egions (empi ical P is indica ed o 30% and 28% o e laps, espec i ely). (B) O e lap o de ec ed Pol2 s alling si es also inc eases
based on he s eng h o an isense signal le el o B-lineage and ES cell con T egions di ided in o qua iles. (C) The in luence o RLFS a TSS on ES
cell DRIP-seq signal le el is shown (Wilcoxon ank sum es P is indica ed). Inpu signal le els a e shown as con ol. (D) ES cell DRIP-seq signal is plo ed
simila ly as in C, om con T-posi i e and -nega i e TSS egions. The DRIP-signal is highe in con T-posi i e TSS (Wilcoxon ank sum es P is indica ed,
TSS wi h con T N = 11774, TSS wi hou con T N = 12092, e e o Figu e 3—sou ce da a 2 o s a is ical analysis based on sepa a e DRIP-seq
eplica es). (E) The pe cen ages o b eakpoin egions wi h no RSS mo i s o e lapping in agenic Pol2 s alling si es ound in B-lineage cells a e shown as
ba plo s. The mean o e lap obse ed in andom sampling is indica ed in g ey ba s ( u he s a is ical analysis is p esen ed in Supplemen a y ile 3).
Ca ego ies wi h inc easing cu -o o ecu ence (1: non- ecu en in dim colo , >1 and abo e: ecu en in da ke colo ) we e es ed. (F) O e lap wi h
RLFS, con T and anno a ed TSS is shown, as in E, o ETV6-RUNX1 NR-b eakp (see also Supplemen a y ile 3). (G) A schema ic model illus a ing how
ansc ip ion om bo h s ands (con T) o RLFS can locally a es he Pol2 complex leading o ec ui men o DNA damage-sensing complexes o
R-loops, such as AID o BRCA (Al e al., 2013,Ha chi e al., 2015), in an RSS-independen manne . (H) NR-b eakp ho spo wi h he highes
ecu ence (TPI1 locus) is shown. DRIP-seq signal (shown in ones o ed o e laid wi h inpu con ol signal in blue), and RLFS mo i s indica ed as a
magen a ba ack ep esen wo le els o independen da a ha we e in eg a ed wi h GRO-seq da a (signal om REH and ES cells is shown) o
cha ac e ize p ope ies o con T and Pol2 s alling egions. The b eakpoin da a (NR-b eakp in b own) and de ec ed con T (in pu ple) and Pol2 s alling
in B-lineage cells (in blue) a e shown. A he he ecu en b eakpoin si es an isense ansc ip ion o neighbo ing gene (SPSB2 p ima y ansc ip ) leads
o a b oad con T egion, as indica ed in he igu e. Ele a ed DRIP-signal indica es o ma ion o DNA-RNA hyb ids (see also Figu e 3— igu e
supplemen 3).
DOI: 10.7554/eLi e.13087.018
The ollowing sou ce da a and igu e supplemen s a e a ailable o igu e 3:
Sou ce da a 1. B eakpoin clus e ing o egions.
DOI: 10.7554/eLi e.13087.019
Sou ce da a 2. S a is ical analysis o sepa a e DRIP-seq and DNAse-seq eplica es.
DOI: 10.7554/eLi e.13087.020
Figu e supplemen 1. GRO-seq, RLFS and DRIP-seq signal p o iles ac oss genes.
DOI: 10.7554/eLi e.13087.021
Figu e supplemen 2. Venn diag ams compa ing SV wi hin Pol2 s alling egions based on GRO- and ChIP-seq p o iles.
DOI: 10.7554/eLi e.13087.022
Figu e supplemen 3. Da a om all signal acks o egions displayed in Figu e 3.
DOI: 10.7554/eLi e.13087.023
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 8 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
T ansc ip ional-coupled ins abili y a RSS-independen SV ho spo s
A mechanis ic link be ween R-loops and AID o - a ge ing has been es ablished in lymphomas
(Al e al., 2013). Wi h his in mind, we in es iga ed egions whe e o - a ge ing could occu ia R-
-loops by ocusing on b eakpoin s wi hou RSS-mo i s (da a shown in igu es ep esen s he 416
ETV6-RUNX1 NR-b eakp, e e o Figu e 3—sou ce da a 1 and Supplemen a y ile 3 o all s a is-
ical esul s). We obse ed signi ican genome-wide en ichmen o b eakpoin s wi h he in es i-
ga ed ansc ip ional ea u es (Figu e 3E and F, 29% o e lap wi h Pol2 s alling wi hin gene
egions, binomial es P=4.088e-07; 9% genome-wide o e lap wi h con T, P=5.16e-07). This en ich-
men o b eakpoin s o con T and Pol2 s alling egions was signi ican ac oss a wide ange o an-
sc ip ional ac i i y ( e e o Supplemen a y ile 3). Co-occu ence o b eakpoin s wi hin a 1-kb
window was used o dis inguish non- ecu en (one b eakpoin ) and ecu en (mo e han one
b eakpoin ) e en s (Figu e 3—sou ce da a 1). B eakpoin ecu ence was ound o inc ease he
o e lap wi h bo h Pol2 s alling (Figu e 3E)and con T (Figu e 3F). The mean o e lap obse ed in
1000- old andom sampling (g ey ba s) con i med he speci ici y o he o e lap (no e ha Pol2
s alling is analyzed om in agenic egions only). The b eakpoin s in Pol2 s alling si es we e con-
co dan wi h analysis using Pol2 ChIP-seq (by 78%) and hey co-localized wi h bo h Se 2 and Se 5
phospho yla ed o ms o Pol2 complex (Figu e 3— igu e supplemen 2). A schema ic model sum-
ma izing he possible unde lying mechanisms based on hese esul s is shown in Figu e 3G. The
dis inc in eg a ed genomic p o iles a e collec i ely depic ed a he TPI1 loci, ep esen ing an SV
ho spo wi h he highes numbe o NR-b eakp in ETV6-RUNX1 cases (Figu e 3H, see also Fig-
u e 3— igu e supplemen 3 and Figu e 2A). A he b eakpoin egion, bo h RLFS and con T a e
isible and o e lap he ele a ed DRIP-seq signal measu ed om ES cells.
Access o RAG clea age si es inc eases a Pol2 s alling egions
Nex , we ocused on deciphe ing whe he he ansc ip ional ea u es associa e wi h RAG o - a ge -
ing. We hypo hesized ha locally deple ed nucleosomes a ound he Pol2 complex (Be ing on and
Boyes, 2013) may enhance access o RSS/RSS-like sequences. To his end, we e ie ed DNAse
hype sensi i i y da a om ENCODE (The ENCODE P ojec Conso ium, 2012; see
Ma e ials and me hods). DNAse-seq signal peaks we e signi ican ly wide when o e lapping wi h Pol2
s alling si es (Figu e 4A). A 876 bp (95% CI, 855–896) inc ease was obse ed in B-lymphoblas oid
cells and 412 bp (95% CI, 395–429) in ES cells (Wilcoxon ank sum es P<2.2e-16 in bo h cell ypes,
see also Figu e 3—sou ce da a 2). This was ep oducibly obse ed using peaks loca ed wi hin gene
TSS, body o end egions (Figu e 4A). We selec ed TSS egions wi h RSS mo i s o close examina-
ion and ound ha Pol2 s alling si es a hese TSS we e signi ican ly wide han a o he TSS
(Figu e 4B), wi h a di e ence o 259 bp (95% CI, 79–475 bp, Wilcoxon ank sum es P=0.0024). Thus,
wide Pol2 s alling inc eases he likelihood o RSS mo i occu ence in accessible ch oma in. The wid h
o s alling did no co ela e posi i ely (Pea son’s co ela ion 0.11; 95% CI, 0.09 o 0.13) wi h he
ansc ip ion le el o he co esponding gene, indica ing ha s alling e en s, and no jus ac i e an-
sc ip ion, a e impo an . We u he analyzed he op 5% o wides Pol2 s alling egions by compa ing
hem o wides peaks om DNAse hype sensi i i y and ChIP o his one ma ks (see
Ma e ials and me hods). The odds a ios o he o e lap a e isualized as a hea map (see Figu e 4C,
OR>10 is shown in da kes colo one, e e o Figu e 4—sou ce da a 1 o mo e s a is ics). In addi-
ion o DNAse-seq and Pol2 ChIP peaks, he H3K4me3 was ound among he op ca ego y, con i med
also by ChIP-seq da a acqui ed om REH and Nalm6 cells (Figu e 4—sou ce da a 1).
Nex , he ETV6-RUNX1 R-b eakp (335; 156 in agenic) we e analysed o he genome-wide o e -
lap wi h he ansc ip ional ea u es. A 66% o e lap was ound wi h Pol2 s alling a in agenic egions
(binomial es P<2.2e-16) and a 44% genome-wide o e lap wi h con T (binomial es P<2.2e-16, see
also Figu e 3—sou ce da a 1 o join analysis ac oss p e-B-ALL sub ypes). The o e lap wi h Pol2
s alling had high ag eemen be ween GRO-seq and ChIP-seq (Figu e 3— igu e supplemen 2) and
i inc eased a ecu en R-b eakp (Figu e 4D). In addi ion, o e lap wi h con T (Figu e 4E) was con-
side able (91%) a egions wi h 4 o mo e b eakpoin s. In compa ison, egions wi h RLFS mo i s o
anno a ed TSSs showed less ma ked en ichmen (up o 36%) (Figu e 4E). Simila , as o NR-b eakp,
he signi ican o e lap wi h ansc ip ional ea u es was p ese ed a a wide ange o exp ession le -
els (Supplemen a y ile 3). A schema ic model ha links he ob ained esul s wi h ulne abili y o
RAG clea age is shown in Figu e 4F. As in Figu e 3I, he di e en p o iles a e depic ed a he SV
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 9 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
amoun s o MNase (0.5–20 U; #88216, The mo ishe , Ca lsbad, CA, USA) was added o he nuclei in
10 ml olume and incuba ed a 37˚C o 10 mins. To s op he eac ion, 100 ml o 2x Lysis bu e was
added o he eac ion (1% SDS, 40 mM EDTA, 100 mM T is-HCl pH 8.1) and samples we e sonica ed
using Bio up o (Diagenode) o 5 cycles (30 s - 30 s) o b eak he nuclei. The lysa e was clea ed by
cen i uga ion and supe na an was dilu ed wi h RIPA bu e ( o Pol2 an ibodies, 1X PBS, 1% NP-40,
0.5% Sodium deoxychola e, 0.1% SDS, PIC) o dilu ion bu e ( o H3K4me3; 20 mM T ix-HCl pH 7.4,
100 mM NaCl, 2 mM EDTA, 0.5% T i onX, PIC). The dilu ed lysa e was p e-clea ed by o a ing o
2 h a 4˚C wi h 60 ml 80% CL-4B sepha ose slu y (GE Heal hca e, UK). Be o e use, sepha ose was
washed wice wi h TE bu e , blocked o 1 h min a oom empe a u e wi h 0.5% BSA and 20 mg/ml
glycogen in 1 ml TE bu e , washed wice wi h TE and b ough up o he o iginal olume wi h TE.
The beads we e disca ded, and 1% o he supe na an we e kep as ChIP inpu . The p o ein o in e -
es was immunop ecipi a ed by o a ing he supe na an wi h 3–5 mg an ibody o e nigh a 4˚C.
An ibodies agains Se 2P (ca # ab5095, RRID:AB_304749) and Se 5P (ca # ab5131, RRID:
AB_449369) we e pu chased om Abcam (Camb idge, MA, USA). The Ab was cap u ed using 25 ml
blocked P o ein G Sepha ose 4 Fas Flow (GE Heal hca e, UK) and o a ing he sample o 2 h a
4˚C. Sepha ose was blocked as CL-4B abo e, excep ha i was o a ed o e nigh a 4˚C. The beads
we e pelle ed (1 min, 1000g, 4˚C) and he supe na an disca ded. The beads used o bind Se 2P/
5P Ab we e washed i e imes wi h 5X LiCl IP wash bu e (100 mM T is pH 7.5, 500 mM LiCl, 1%
NP-40, 1% Sodium deoxychola e) and wice wi h TE in 0.45 mm il e ca idges (Ul a ee MC,
Millipo e, Bed o d, MA, USA). The beads used o pull down H3K4me3 Ab we e washed h ee imes
wi h wash bu e I (20 mM T is/HCl pH 7.4, 150 mM NaCl, 0.1% SDS, 1% T i on X-100, 2 mM EDTA),
wice wi h bu e II (20 mM T is/HCl pH 7.4, 500 mM NaCl, 1% T i on X-100, 2 mM EDTA) and bu e
III (10 mM T is/HCl pH 7.4, 250 mM LiCl, 1% IGEPAL CA-630, 1% Na-deoxychola e, 1 mM EDTA),
once wi h TE + 0.2% T i onX and wice wi h TE. Immunop ecipi a ed ch oma in was elu ed wice
wi h 100 ml elu ion bu e (TE, 1% SDS). The NaCl concen a ion was adjus ed o 300 mM wi h 5 M
NaCl and c osslinks we e e e sed o e nigh a 65˚C. The samples we e sequen ially incuba ed a
37˚C o 2 h each wi h 0.33 mg/ml RNase A and 0.5 mg/ml p o einase K (bo h om The mo ishe ,
Ca lsbad, CA, USA). The DNA was isola ed using he ChIP DNA Clean & Concen a o (Zymo
Resea ch, I ine, CA, USA) acco ding o he manu ac u e ’s ins uc ions. Sequencing lib a ies we e
p epa ed om collec ed DNA by blun ing, A- ailing, adap o liga ion as p e iously desc ibed
(Heinz e al., 2010) using ba coded adap e s (Nex Flex, Bioo Scien i ic, Aus in, TX, USA). Be ween
he eac ions, he DNA was pu i ied using Se a-Mag SpeedBeads (The mo ishe , Ca lsbad, CA,
USA). Lib a ies we e PCR-ampli ied o 15–16 cycles, size selec ed o 230–350bp agmen s by gel
ex ac ion and single-end sequenced on a Hi-Seq 2000 (Illumina) o 50 cycles.
P ocessing o GRO-seq, ChIP-seq, DRIP-seq and HiC sequencing eads
The GRO-seq da a om lymphoblas oid cells (GSE39878, Wang e al., 2014b; GSE60456,
Co e e al., 2014), ES cells (GSE41009, Sigo a e al., 2013), DRIP-seq da a om ES cells
(GSE45530, Ginno e al., 2013) and HiC da a om human lymphoblas oid GM12878 cells
(GSM1551571, GSM1551572, GSM1551574, GSM1551575; Rao e al., 2014) we e downloaded
om SRA ( aw eads) and p ocessed simila ly as he new samples: eads we e quali y con olled and
subsequen ly aligned o he human hg19 e e ence genome e sion. Speci ically, he quali y o aw
sequencing eads was con i med using he Fas QC ool (h p://www.bioin o ma ics.bab aham.ac.uk/
p ojec s/ as qc/) and subsequen ly bases wi h poo quali y sco es we e immed ( equi ing a mini-
mum 97% o all bases in one ead o ha e a min ph ed quali y sco e o 10) using he Fas X oolki
(h p://hannonlab.cshl.edu/ as x_ oolki /). Samples sequenced on mul iple lanes we e pooled a e
quali y con ol. Read s acks we e collapsed om ChIP-seq iles using as x (collapse). The Bow ie
so wa e (bow ie-0.12.9 0.1.x) (Langmead e al., 2009) was used o aligning he GRO-seq, ChIP-
seq and DRIP-seq eads o he human genome ( e sion hg19). Up o wo misma ches and up o
h ee loca ions we e accep ed and he bes alignmen was epo ed o each ead. Fo he GRO-seq
eads his s ep was p eceded by emo ing eads mapping o RNA egions (Abundan Sequences as
anno a ed by iGenomes) and disca ding eads o e lapping wi h so-called blacklis ed egions ha
ep esen unusual low o high mappabili y as de ined by ENCODE, ibosomal and small nucleola
RNA (snoRNA) loci om ENCODE and u he manually cu a ed o he human genome (bed ile
wi h sequences is p o ided in Supplemen a y ile 6).
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 16 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine

HiC
Reads om pai ed-end sequencing we e sepa a ely il e ed and aligned o he genome using bow-
ie. The eads we e checked o MboI es ic ion si es be o e doing he alignmen s and he sequen-
ces a e GATC si es we e immed ou o imp o e mappabili y. The HOMER 4.3 (h p://home .
salk.edu/home ) so wa e was used in u he p ocessing o HiC-da a. Pai ed-end eads we e con-
nec ed and ead pai s wi h exac same ends we e only conside ed once and ead pai s we e
emo ed i hey we e sepa a ed by less han 1.5 he es ima ed sequencing inse leng h o emo e
likely con inuous genomic agmen s o e-liga ion e en s. Pai ed-end eads o igina ing om egions
o unusually high ag densi y we e le ou by emo ing eads om 10 kb egions ha con ain mo e
han i e imes he a e age numbe o eads. Backg ound model o no maliza ion o HiC-da a was
gene a ed wi h 50 kb esolu ion. The opological domains we e iden i ied using he HOMER com-
mand’ indHiCDomains.pl’ using a esolu ion o 50 kb. This analysis is based on a s a is ic e e ed o
as he ‘di ec ionali y index’, which desc ibes he endency o a gi en posi ion o in e ac wi h ei he
he ch oma in ups eam o downs eam om i s cu en posi ion.
GRO-seq
Combined agDi ec o ies om GRO-seq samples we e made by pooling he sequencing da a o
each cell and sample ype wi h agmen leng h se o 75. The indPeaks.pl p og am in he The
HOMER 4.3 so wa e (h p://home .salk.edu/home ) was used o iden i y de no o ansc ip s om
GRO-seq da a using pooled sequencing eads pe sample ype. Deeply sequenced REH, Nalm6 and
lymphoblas oid cells we e used o de ine signal ea u es in B-cell lineage and sepa a e analysis was
ca ied ou o ES cells (see Supplemen a y ile 1). Gaps we e allowed a non-mappable egions (-
s yle g oseq -uniqmap).
ChIP-seq
Peaks we e iden i ied using indPeaks (-s yle his one –size 1000).
Signal acks
BedG aph and bigWig iles we e gene a ed wi h eads in each sequencing expe imen no malized
o a o al o 10
7
mapped eads. The bigWig iles we e u he con e ed o ack hubs and isualized
as s and-speci ic, o e laid Mul iT acks as a cus om T ack Hub in he UCSC Genome b owse .
Genomic egions used in analyses
The hg19 genome e sion om UCSC (a ailable om iGenomes) was used o speci y ch omosome
leng hs in he analysis. The gene anno a ions om Re seq and UCSC known gene ables we e
e ie ed using he UCSC Table B owse (hg19, GRCh37 Genome Re e ence Conso ium Human Re -
e ence 37 (GCA_000001405.1)). Unique ansc ip coo dina es we e used in analysis, ha is, any
ansc ip s sha ing he same s a and end coo dina e we e conside ed oge he . The TSS egions
we e de ined as +/- 1 kb egions a ound he anno a ed s a coo dina e. Only ansc ip s mapping
o canonical ch omosomes we e kep , also hose on ch M we e emo ed.
Enhance s
Supe -enhance coo dina es om CD19+, CD20+ and HSC cells we e ob ained (Hnisz e al., 2013)
and me ged o isualiza ion o acks. De no o enhance de ec ion was pe o med om he deeply
sequenced REH, Nalm6 and lymphoblas oid cells based on he ansc ip iden i ica ion esul . T an-
sc ip s wi h leng h <15 kb and he cha ac e is ic bidi ec ionali y o co-localiza ion wi h enhance
loca ions de ined using DNAse and ch oma in ma ke da a we e used o dis inguish eRNAs (see Fig-
u e 2—sou ce da a 1 o da a).
Analysis o SV in con ex o ch omosome sub egions
TADs e lec he h ee dimensional s uc u e o ch oma in, o ming na u al bounda ies ha di ide
he ch omosomes in o sub- egions. To iden i y TADs wi h highes equency o b eakpoin s, HiC-
da a analysis was pe o med using HOMER 4.3. As ou goal is o gene a e a na u al di ision o he
genome in o sub- egions ha a e ele an in con ex o ansc ip ional egula ion, his app oach is
supe io o a bi a ily assigning sub- egions based on ixed windows. The p e-B-ALL b eakpoin s and
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 17 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
anno a ion da a (Ande sson e al., 2015;Holm eld e al., 2013;Papaemmanuil e al., 2014;
Paulsson e al., 2015) we e analyzed in con ex o TADs. Speci ically, TADs we e o e lapped wi h
sub ype-speci ic b eakpoin s (Figu e 1—sou ce da a 1 p esen s TADs so ed based on he coun o
b eakpoin s). Subsequen ly, TADs wi h b eakpoin s we e di ided in o qua iles based on b eakpoin
equency pe bp o analyze en ichmen o ea u e o e lap ha exceeds he genomic backg ound
le el. To ob ain he o al ansc ibed a ea wid h wi hin each TAD, he TAD coo dina es we e o e -
lapped wi h he de ec ed GRO-seq ansc ip s (bed ools in e sec -wao). The combined SV da a ep-
esen s in o al 1680 b eakpoin s and is he mos comp ehensi e collec ion o p e-B-ALL SV ha we
a e awa e o .
Ch oma in segmen a ion da a
B oadCh omHMM ch oma in segmen a ions we e ob ained om GM12878 and H1 ES cells includ-
ing he ollowing segmen ypes: 1_Ac i e_P omo e , 2_Weak_P omo e , 3_Poised_P omo e ,
4_S ong_Enhance , 5_S ong_Enhance , 6_Weak_Enhance , 7_Weak_Enhance , 8_Insula o ,
9_Txn_T ansi ion, 10_Txn_Elonga ion, 11_Weak_Txn, 12_Rep essed", 13_He e och om/lo,
14_Repe i i e/CNV, 15_Repe i i e/CNV. The sizes o he segmen s o each ype we e used o calcu-
la e he o al span om he genome. Each segmen ype was hen o e lapped wi h a combined bed
ile speci ying con T and Pol2 s alling egions. O e lapping and non-o e lapping pieces we e
e u ned and analyzed sepa a ely (bed ools in e sec , ollowed by bed ools sub ac wi h he o e -
lapping pieces gi en as pa ame e b).
Dis inguishing b eakpoin s based on RSS-like mo i s o ecu ence
Two ypes o b eakpoin s we e dis inguished based on RSS mo i anno a ion o esul in he ollow-
ing egion assignmen : egions con aining a consensus RSS/hep ame sequence mo i we e used o
ca ego ize co-localized b eakpoin s as pu a i e RSS-dependen b eakpoin s (R-b eakp: 447 in o al,
335 in he ETV6-RUNX1 sub ype), while egions de oid o ecogni ion sequence we e used o clas-
si y RSS-independen lesions (NR-b eakp: 938 in o al, 416 in he ETV6-RUNX1 sub ype). Regions
ha bo ing un esol ed b eakpoin s we e le ou om majo i y o analysis pe o med (285 egions ha -
bo ing 295 b eakpoin s in he ETV6-RUNX1 sub ype ha we e mainly isola ed and non- ecu en ).
The RSS assignmen o o he b eakpoin s was ob ained in he ollowing way: he esol ed b eak-
poin s we e ex ended o bo h sides by 10 bp, esul ing in a 21 bp egion. The MEME analysis in
Papaemmanuil e al. 2014 o 708 esol ed b eakpoin s om ETV6-RUNX1 pa ien s was eplica ed
and compa able sequence logos o ha epo ed p e iously we e ob ained and used o anno a e
RSS s a us. A p- alue cu -o o 0.003 was chosen o he MEME mo i scanning based on FIMO anal-
ysis o he ETV6-RUNX1 da a.
To e alua e ecu ence, he b eakpoin ends a 1 kb dis ance we e s i ched oge he o o m
egions (each wi h a leas one b eakpoin , see Figu e 3—sou ce da a 1), anno a ing he numbe o
b eakpoin s inside (BEDTools me geBed –d 1000 –n). O e lap o b eakpoin egions wi h RLFS, TSS,
con T and Pol2 s alling egions we e ob ained using BEDTools wi h 1 kb window. The o e lap e-
quencies we e compa ed o andom sampling o simila ly sized genomic egions. Fu he compa i-
sons we e pe o med sepa a ing ecu en (>1 b eakpoin pe s i ched egion) and non- ecu en
egions, and wi h inc easing he cu -o o he numbe o b eakpoin e en s pe s i ched egion.
The same was epea ed o b eakpoin s wi hin genes binned in o ou ca ego ies based on hei
ansc ip ion le el. The ansc ip egions we e quan i ied using da a om REH, Nalm6, and lympho-
blas oid cells, and no malized by RPKM. The maximum exp ession alue was o di ide ansc ip s
in o qua iles based on he exp ession le el.
Signal ea u e analysis
The isual examina ion o SV si es se ed as he i s s ep o de ine ansc ip ional ea u es o po en-
ial ele ance. This mo i a ed he analysis o egions wi h o e lapping ansc ip ion om bo h
s ands (con T) and local ele a ions in he signal (Pol2 s alling), wi h de ailed de ini ions gi en below.
Fo genome-wide analysis o signal ea u e o e lap wi h SV, ea u e acks om se e al samples
we e combined. This app oach was deemed mos app op ia e o add ess he dynamic na u e o
ansc ip ional ac i i y and o a oid missing egions ha due o high ecu ence o SV may be
dele ed in subse o leukemic cells s udied.
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 18 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
Con T
Con T egions we e iden i ied as ansc ip s ha o e lap on opposi e s ands by a leas 100 bp (as
in Meng e al., 2014). Subsequen ly, a combined bed ack was c ea ed o he leukemic and lym-
phoblas oid samples using bedTools me geBed command (-d 0). The da a om ES cells (GSE41009)
se ed as an independen con ol. The le el o con T was quan i ied using he HOMER p og am ana-
lyzeRNA.pl om bo h s ands sepa a ely and no malized by egion size. The minimum alue
ob ained pe egion (compa ing + and - s ands) was assigned as con T le el.
Pol2 s alling
Change-poin de ec ion is he ma hema ical p oblem o inding ab up changes in a signal, ypically
applied in con ex o ime se ies (Killick e al., 2012). Bo h app oxima e and exac me hods exis o
es ima ing he poin a which he s a is ical p ope ies o a sequence o obse a ions change. The
analysis o changepoin s in he signal mean was ca ied ou using unc ions implemen ed in he R
package ‘changepoin ’ (Killick and Eckley, 2014). An exac me hod wi h a o able compu a ional
cos was ecen ly in oduced in con ex o ime-se ies da a (Killick e al., 2012). This me hod called
PELT was selec ed o de ec he changepoin s om scaled (ze o mean, equal a iance) signal p o iles
calcula ed a 50-bp esolu ion (gene a ed bedG aph iles a e a ailable unde he GEO accession
GSE67540). The analysis was pe o med sepa a ely o each gene, using Bayesian In o ma ion C i e-
ion as a penal y e m wi h he changepoin coun ed as a pa ame e ( unc ion cp .mean wi h pa am-
e e s penal y = ’BIC1’, me hod =’PELT’). The inpu da ase ep esen ing p ima y ansc ip ion
ac i i y a gene loci was gene a ed by o e lapping he GRO-seq signal ile s and-speci ically wi h
ansc ip coo dina es om UCSC and Re seq (see genomic egions used). The analysis only consid-
e ed egions wi h anno a ion ma ch (in minimum 5% o iden i ied ansc ip co e ed by anno a ion; a
minimum o 50% o e lap wi h he iden i ied ansc ip ; anno a ed and de ec ed s a s do no di e
mo e han 10 kb). In o de o de ine Pol2 s alling si es, he signal le el be ween changepoin s we e
compa ed o he median ac oss he whole gene, and in e als abo e 90% quan ile we e epo ed as
s alled. Fo ChIP-seq, his cu -o was elaxed o 80% due o highe backg ound signal. No ice also
ha he e is no s and in o ma ion based on ChIP-seq. The analysis was ca ied ou sepa a ely o
each o he deeply sequenced (REH, Nalm6 and lymphoblas oid) GRO-seq da ase s and ChIP-seq
eplica es and subsequen ly me ged o one bed ile speci ying s alled egion coo dina es (bedTools
me ge –d 100). The ES GRO-seq da ase GSE41009 was p ocessed simila ly and used as an indepen-
den con ol. To s udy whe he he e was a ela ionship be ween he s alled egion size o e lapping
TSS egions and R-b eakp equency, he ollowing in e sec s we e calcula ed using BEDTools (in e -
sec Bed -wa | uniq): (i) o e lap o Pol2 s alling si es and TSS egions ha bo ing R-b eakp and (ii) o e -
lap o Pol2 s alling si es and TSS egions no ha bo ing R- o NR-b eakp. Subsequen ly, he sizes o
Pol2 s alling si es in each coo dina e ile we e calcula ed and he Wilcoxon ank sum es used o
e alua ing s a is ical signi icance o he di e ence in Pol2 s alling wid h. Secondly, op 5% wides
Pol2 s alling si es we e iden i ied and compa ed o op 5% wides peaks om ChIP-seq and DNAse-
seq p o iles (see below).
Signal compa ison a gene egions
The HOMER command anno a ePeaks.pl was used o c ea e a ansc ip ional p o ile o ac i e genes
(RPKM > 0.5) in ES, lymphoblas oid and REH cells by scaling he his og am o each egion (i.e 0–
100%) using a bin size o 100. RLFS mo i densi y was calcula ed ac oss genes wi h he BED ools co -
e age ool. A densi y plo ep esen ing RLFS equency ac oss gene egions was hen ob ained as
abo e.
DRIP-seq and RLFS mo i da a o R-loop de ec ion
Da a om eplica e DRIP-seq expe imen s wi h wo di e en es ic ion enzyme diges ions
(GSE45530) we e used in he analysis. Log2 signal le els we e quan i ied using HOMER a TSS
egions. S a is ical signi icance was es ima ed sepa a ely o he wo di e en es ic ion enzyme
diges ions. RLFS mo i sea ch was pe o med using he so wa e QmRLFS- inde ha p edic s R-loop
o ming sequences based on s uc u al models o known sequences (Jenja oenpun e al., 2015).
The as a inpu ile was gene a ed by ex ac ing DNA sequences based on he hg19 genome
e sion.
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 19 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
DNAse-seq da a and addi ional ChIP-seq da a o cha ac e ize wide Pol2
s alling e en s
The DNAse-seq peaks we e i s o e lapped wi h he Pol2 s alling egions de ec ed based on he
GRO-seq signal. Only peaks wi h a sco e abo e 15 we e conside ed. The wid h o he o e lapping
peaks was hen compa ed o he wid h o peaks wi h no o e lap using he Wilcoxon ank sum es .
Nex , op 5% wides peaks we e ob ained om he DNAse-seq and ChIP-seq da a ( e e o Fig-
u e 4—sou ce da a 1). The o e lap wi h 5% wides Pol2 s alling egions was subsequen ly e alua ed
using he BEDTools ishe ool.
T ansc ip ome da a
Gene exp ession da a om p e-B-ALL s udies was combined om mic oa ay da ase s e ie ed
om he NCBI GEO da abase as pa o a da a collec ion ep esen ing bo h heal hy and malignan
samples hyb idized o hgu133Plus2 genome-wide mic oa ays (in p epa a ion o submission). In
o al, 1382 p e-B-ALL samples we e included. P obe-le el quali y con ol was pe o med o exclude
samples wi h e y high di e ence in da a loca ion o dis ibu ion as measu ed by median and in e -
qua ile ange o aw p obe in ensi ies. Samples ha passed his il e ing we e p ocessed using he
RMA p obe summa iza ion algo i hm wi h p obe mapping o En ez Gene IDs ( om B ainA ay e -
sion 18.0.0, ENTREZG), ollowed by bias co ec ion using he R package ‘bias’. The Ba nes-Hu -SNE
algo i hm (compu a ionally as e app oxima ion o -SNE) implemen a ion om he R package
‘R sne’ (K ij he, 2015) was used o disco e nea -op imal ep esen a ion o sample dis ances in wo
dimensions (using pa ame e alues pe plexi y 30 and he a 0.5) using 15% genes wi h highes a i-
ance. The -SNE me hod belongs o dimensionali y educ ion me hods ha include also adi ional
me hods such as P incipal Componen Analysis. The main objec i e o he me hod is o accu a ely
place highly simila samples (he e based on he high-dimensional gene exp ession p o ile) o close
p oximi ies in lowe dimensions. The esul can be isualized in wo-dimensions as a sca e plo ha
allows obse ing sample g oups based on he molecula p o iles. Acco ding o ou expe ience, his
me hod p o ides be e sepa a ion be ween sample g oups compa ed o mo e adi ional me hods
o la ge he e ogeneous sample collec ions. To iden i y whe he a gi en gene was exp essed o
unexp essed in a sample, a Gaussian ini e mix u e model ( es ing equal and a iable a iance mod-
els, bes i chosen by BIC) was i ed by expec a ion-maximiza ion algo i hm o he p obe signals (R
package ‘mclus ’, e sion 4.3, F aley and Ra e y, 2002).
S a is ical es s
S a is ical signi icance was es ima ed using se e al es s o ensu e eliabili y, including es s ha ely
on assump ions abou da a dis ibu ions and empi ical es s ha ely on andomiza ion o da a
poin s. The s a is ical es s used, exac alues o N, de ini ions o cen e and dispe sion and p ecision
measu es a e indica ed in Resul s, in he espec i e supplemen a y ables o igu e legends.
Binomial es
Tes o independen andom ials wi h bina y (success/ ailu e) ou come, wi h eplacemen . This es
was used o assess he s a is ical signi icance o obse ing b eakpoin e en s o e lapping a ansc ip-
ional ea u e (Pol2 s alling o con T). Success in popula ion was de ined using 1 kb windows ac oss
he genome. The windows o e lapping he s udied ea u e was di ided by he o al numbe o 1 kb
windows analyzed. E.g. in he Pol2 s alling analysis, he o al numbe o windows o e lapping Pol2
s alling egions di ided by his numbe o 1 kb windows wi hin gene coo dina es (included o he
inpu o he change poin analysis), de ine p obabili y o success.
Hype geome ic es
Tes o independen andom ials wi h bina y (success/ ailu e) ou come, wi hou eplacemen . This
es was used o assess he s a is ical signi icance o obse ing g ea e han o equal o e lap e-
quency be ween b eakpoin s and an anno a ed se o genomic egions. E.g. o es o en ichmen
o b eakpoin s inside con T-posi i e enhance s, con T-posi i e enhance s wi h b eakpoin s de ine
sample success; all enhance s wi h b eakpoin s popula ion success; and sample aken is all con T-
posi i e enhance s ( om he popula ion o all enhance s). The ela ed Fishe ’s es (implemen ed in
BEDTools ishe ) was used o ob ain simila s a is ics wi h odds a ios.
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 20 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
Wilcoxon ank sum es (Mann-Whi ney es )
A nonpa ame ic wo-sided Wilcox es was pe o med o es ima e whe he wo samples (con inuous
alues, unknown dis ibu ion) come om he same popula ion (R unc ion wilcox. es ). This es was
applied o quan i ied signal le els compa ed be ween ca ego ies.
Random sampling
This es can be used o ob ain an empi ical es ima e o andom o e lap equencies. The sampling
was pe o med 1000- old wi hin he same genomic con ex as used in he analysis. To es ima e he
signi icance o o e lap be ween s i ched b eakpoin egions wi h e.g. con T egions, he s i ched
egions we e alloca ed andom genomic coo dina es, hus p ese ing he size dis ibu ion and
b eakpoin e en equencies wi hin s i ched egions. The obse ed andom egion o e lap was
used as he empi ical p- alue es ima e. Fu he , he z- es was used o e alua e whe he he e was
e idence o ejec he null hypo hesis ha he obse ed ea u e o e lap alue would belong o he
empi ical dis ibu ion ob ained.
Acknowledgemen s
We would like o hank Ville Hau ama
¨ki o commen s on signal analysis me hods and he EMBL
Gene Co e sequencing eam o he sequencing se ice p o ided. The wo k was suppo ed by
g an s om he Emil Aal onen Founda ion, Jane and Aa os E kko Founda ion, Finnish Cance Foun-
da ion, Academy o Finland, Sig id Juselius Founda ion,Finnish Cul u al Founda ion, Paulo Founda-
ion, Founda ion o Pedia ic Resea ch, he Compe i i e S a e Resea ch Financing o he Expe
Responsibili y a ea o Tampe e Uni e si y Hospi al, Uni e si y o Tampe e and Uni e si y o Eas e n
Finland.
Addi ional in o ma ion
Funding
Funde G an e e ence numbe Au ho
Suomen Kul uu i ahas o 00150214 Me ja Heina
¨niemi
Olli Lohi
I a
¨-Suomen Yliopis o Me ja Heina
¨niemi
The Finnish Cance Founda ion Me ja Heina
¨niemi
Emil Aal osen Sa
¨a
¨ io
¨Me ja Heina
¨niemi
Suomen Aka emia 276634 Me ja Heina
¨niemi
Tampe een Yliopis o Susanna Teppo
Saa a Laukkanen
Thomas Liuksiala
Olli Lohi
Sig id Juselius Founda ion Minna U Kaikkonen
Suomen Aka emia 277816 Olli Lohi
Jane ja Aa os E kon Sa
¨a
¨ io
¨Olli Lohi
Paulo Founda ion Olli Lohi
Las en au ien Tu kimussa
¨a
¨ io
¨Olli Lohi
Compe i i e S a e Resea ch
Financing o he Expe
Responsibili y a ea o Tampe e
Uni e si y Hospi al
Olli Lohi
The unde s had no ole in s udy design, da a collec ion and in e p e a ion, o he decision o
submi he wo k o publica ion.
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 21 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine

Au ho con ibu ions
MH, ST, MUK, Concep ion and design, Acquisi ion o da a, Analysis and in e p e a ion o da a,
D a ing o e ising he a icle; TV, OL, Concep ion and design, Analysis and in e p e a ion o da a,
D a ing o e ising he a icle; MB-L, JM, HN, TL, Acquisi ion o da a, Analysis and in e p e a ion o
da a, D a ing o e ising he a icle; VZ, Analysis and in e p e a ion o da a, D a ing o e ising he
a icle; SL, KT, Acquisi ion o da a, D a ing o e ising he a icle
Au ho ORCIDs
Me ja Heina
¨niemi, h p://o cid.o g/0000-0001-6190-3439
Susanna Teppo, h p://o cid.o g/0000-0003-2569-8030
Olli Lohi, h p://o cid.o g/0000-0001-9195-0797
E hics
Human subjec s: The s udy was app o ed by he Regional E hics Commi ee in Pi kanmaa, Tampe e,
Finland (#R13109). The s udy was conduc ed acco ding o he guidelines o he Decla a ion o Hel-
sinki, and a w i en in o med consen was ecei ed by he pa ien and/o gua dians.
Addi ional iles
Supplemen a y iles
.Supplemen a y ile 1. GRO-seq sample summa y. Desc ip ion o he pa ien and cell line GRO-seq
samples used in he analysis, including he cell cul u e condi ions, eplica e in o ma ion and he o al
numbe o pooled sequencing eads ob ained a e quali y il e ing and alignmen . A mo e de ailed
able o cul u ed samples wi h eplica e in o ma ion and accession codes is p o ided a he bo om.
Sample accession codes o al eady published and e-analyzed GRO-seq da a, and addi ional GRO-
seq da a displayed in Figu e 1— igu e supplemen 1 a e lis ed in wo kshee 2.
DOI: 10.7554/eLi e.13087.030
.Supplemen a y ile 2. Genomic coo dina es o egions displayed. The coo dina es o example
gene egions displayed in he main and supplemen a y igu es a e lis ed (hg19 human genome
e sion).
DOI: 10.7554/eLi e.13087.031
.Supplemen a y ile 3. B eakpoin ho spo analysis o genes binned by he ansc ip ion le el.
Hype geome ic es s a is ics o genes s a i ied by exp ession le el. B eakpoin o e lap wi h an-
sc ip ional ea u es was es ed wi hin he binned in agenic egions. Da a o ETV6-RUNX1 sub ype
and all p e-B-ALL sub ypes a e shown as sepa a e wo kshee s. Rela ed o Figu es 3 and 4.
DOI: 10.7554/eLi e.13087.032
.Supplemen a y ile 4. In agenic ecu en SV in ETV6-RUNX1 pa ien s wi h o e lap o ulne able
egions. The pa ien and egion iden i ie s o ecu en in agenic SV in ETV6-RUNX1 pa ien s a e
lis ed, epo ing sepa a ely hose co-localized wi h Pol2 s alling o con T egions.
DOI: 10.7554/eLi e.13087.033
.Supplemen a y ile 5. Clinical da a o pa ien s wi h high AICDA exp ession. S udy desc ip ion,
sample iden i ie , cy ogene ic g oup, age and da ase iden i ie a e lis ed o he pa ien s wi hin
high AICDA exp ession le el. S a is ical analysis es ing en ichmen o de ec ed AICDA exp ession in
high isk s udies is summa ized in wo kshee 2.
DOI: 10.7554/eLi e.13087.034
.Supplemen a y ile 6. Cus om blacklis ed genomic egions. Blacklis ed egions disca ded om he
analysis ha we e deemed o ep esen low-mappabili y, RNA and snoRNA loci based on GRO-seq
signal. Coo dina es e e o he hg19 human genome e sion.
DOI: 10.7554/eLi e.13087.035
Majo da ase s
The ollowing da ase s we e gene a ed:
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 22 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
Au ho (s) Yea Da ase i le Da ase URL
Da abase, license,
and accessibili y
in o ma ion
Heina
¨niemi M,
Teppo S, Kaikko-
nen MU, Bou y-
Lii and M, Lohi O
2015 ALL cells h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE67540
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE67540)
Heina
¨niemi M,
Teppo S, Lohi O
2015 Genome-wide mapping o TEL-
AML1 a ge s in acu e leukemia
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE67519
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE67519)
The ollowing p e iously published da ase s we e used:
Au ho (s) Yea Da ase i le Da ase URL
Da abase, license,
and accessibili y
in o ma ion
Wang IX, Co e LJ,
Kwak H, B ady L,
B uzel A, McDaniel
L, Richa ds AL, Wu
M, G unseich C, Lis
JT, Cheung VG
2014 RNA-DNA DIFFERENCES IN
NASCENT RNA
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE39878
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE39878)
Co e LJ, Ma ins
AL, Danko CG,
Wa e s CT, Siepel
A, Lis JT
2014 Analysis o ansc ip ion s a si es
om nascen RNA iden i ies a
uni ied a chi ec u e o ini ia ion a
mammalian p omo e s and
enhance s
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE60456
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE60456)
Sigo a AA, Mullen
AC, Molinie B,
Gup a S, O lando
DA, Guen he MG,
Almada AE, Lin C,
Sha p PA, Giallou -
akis CC, Young RA
2013 Di e gen ansc ip ion o lncRNA/
mRNA gene pai s in emb yonic
s em cells
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE41009
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE41009)
Ginno PA, Lim YW,
Lo PL, Ko I,
Che
´din F
2013 DNA-RNA Immunop ecipi a ion
sequencing (DRIP-seq) o human
NT2 cells
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE45530
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE45530)
Sanbo n AL, Rao
SS, Huang SC,
Du and NC, Hun -
ley MH, Jewe AI,
Bochko ID, Chin-
nappan D, Cu kos-
ky A, Li J, Gee ing
KP, Gni ke A, Mel-
niko A, McKenna
D, S ameno a EK,
Lande ES, Aiden
EL
2014 A h ee-dimensional map o he
human genome a kilobase
esolu ion e eals p inicples o
ch oma in looping
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE63525
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE63525)
Sands om R 2011 DNaseI Hype sensi i i y by Digi al
DNaseI om ENCODE/Uni e si y
o Washing on
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE29692
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE29692)
Sho esh N 2011 His one Modi ica ions by ChIP-seq
om ENCODE/B oad Ins i u e
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE29611
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE29611)
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 23 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
Sands om R 2011 CTCF Binding Si es by ChIP-seq
om ENCODE/Uni e si y o
Washing on
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE30263
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE30263)
Mye s R, Pauli F 2011 T ansc ip ion Fac o Binding Si es
by ChIP-seq om ENCODE/HAIB
h p://www.ncbi.nlm.nih.
go /geo/que y/acc.cgi?
acc=GSE32465
Publicly a ailable a
NCBI Gene
Exp ession Omnibus
(accession no:
GSE32465)
Re e ences
Adelman K, Lis JT. 2012. P omo e -p oximal pausing o RNA polyme ase II: eme ging oles in me azoans. Na u e
Re iews. Gene ics 13:720–731. doi: 10.1038/n g3293
Al FW, Zhang Y, Meng FL, Guo C, Schwe B. 2013. Mechanisms o p og ammed DNA lesions and genomic
ins abili y in he immune sys em. Cell 152:417–429. doi: 10.1016/j.cell.2013.01.007
Ande sson AK, Ma J, Wang J, Chen X, Gedman AL, Dang J, Naki andwe J, Holm eld L, Pa ke M, Eas on J,
Hue he R, K iwacki R, Rusch M, Wu G, Li Y, Mulde H, Raimondi S, Pounds S, Kang G, Shi L, e al. 2015. The
landscape o soma ic mu a ions in in an MLL- ea anged acu e lymphoblas ic leukemias. Na u e Gene ics 47:
330–337. doi: 10.1038/ng.3230
Ba eman CM, Alpa D, Fo d AM, Colman SM, W en D, Mo gan M, Kea ney L, G ea es M. 2015. E olu iona y
ajec o ies o hype diploid ALL in monozygo ic wins. Leukemia 29:58–65. doi: 10.1038/leu.2014.177
Benayoun BA, Pollina EA, Uca D, Mahmoudi S, Ka a K, Wong ED, De a ajan K, Daughe y AC, Kundaje AB,
Mancini E, Hi z BC, Gup a R, Rando TA, Bake JC, Snyde MP, Che y JM, B une A. 2014. H3K4me3 b ead h is
linked o cell iden i y and ansc ip ional consis ency. Cell 158:673–688. doi: 10.1016/j.cell.2014.06.027
Be ing on S, Boyes J. 2013. T ansc ip ion-coupled e ic ion o his ones H2A/H2B go e ns V(D)J ecombina ion.
The Eu opean Molecula Biology O ganiza ion Jou nal 32:1381–1392. doi: 10.1038/emboj.2013.42
Bolland DJ, Wood AL, Johns on CM, Bun ing SF, Mo gan G, Chakalo a L, F ase PJ, Co co an AE. 2004.
An isense in e genic ansc ip ion in V(D)J ecombina ion.. Na u e Immunology 5:630–637. doi: 10.1038/ni1068
Co e LJ, Ma ins AL, Danko CG, Wa e s CT, Siepel A, Lis JT. 2014. Analysis o nascen RNA iden i ies a uni ied
a chi ec u e o ini ia ion egions a mammalian p omo e s and enhance s. Na u e Gene ics 46:1311–1320. doi:
10.1038/ng.3142
F aley C, Ra e y AE. 2002. Model-based clus e ing, disc iminan analysis, and densi y es ima ion. Jou nal o he
Ame ican S a is ical Associa ion 97:611–631. doi: 10.1198/016214502760047131
Gelle M. 2002. V(D)J ecombina ion: RAG p o eins, epai ac o s, and egula ion. Annual Re iew o
Biochemis y 71:101–132. doi: 10.1146/annu e .biochem.71.090501.150203
Ginno PA, Lim YW, Lo PL, Ko I, Che´din F. 2013. GC skew a he 5’ and 3’ ends o human genes links R-loop
o ma ion o epigene ic egula ion and ansc ip ion e mina ion. Genome Resea ch 23:1590–1600. doi: 10.
1101/g .158436.113
Hao B, Naik AK, Wa anabe A, Tanaka H, Chen L, Richa ds HW, Kondo M, Taniuchi I, Kohwi Y, Kohwi-Shigema su
T, K angel MS. 2015. An an i-silence - and SATB1-dependen ch oma in hub egula es Rag1 and Rag2 gene
exp ession du ing hymocy e de elopmen . The Jou nal o Expe imen al Medicine 212:809–824. doi: 10.1084/
jem.20142207
Ha chi E, Skou i-S a haki K, Ven z S, Pinello L, Yen A, Kamienia z-Gdula K, Dimi o S, Pa hania S, McKinney KM,
Ea on ML, Kellis M, Hill SJ, Pa migiani G, P oud oo NJ, Li ings on DM. 2015. BRCA1 ec ui men o
ansc ip ional pause si es is equi ed o R-loop-d i en DNA damage epai . Molecula Cell 57:636–647. doi:
10.1016/j.molcel.2015.01.011
Heinz S, Benne C, Spann N, Be olino E, Lin YC, Laslo P, Cheng JX, Mu e C, Singh H, Glass CK. 2010. Simple
combina ions o lineage-de e mining ansc ip ion ac o s p ime cis- egula o y elemen s equi ed o
mac ophage and B cell iden i ies. Molecula Cell 38:576–589. doi: 10.1016/j.molcel.2010.05.004
Hnisz D, Ab aham BJ, Lee TI, Lau A, Sain -And e´ V, Sigo a AA, Hoke HA, Young RA. 2013. Supe -enhance s in
he con ol o cell iden i y and disease. Cell 155:934–947. doi: 10.1016/j.cell.2013.09.053
Holm eld L, Wei L, Diaz-Flo es E, Walsh M, Zhang J, Ding L, Payne-Tu ne D, Chu chman M, Ande sson A, Chen
SC, McCas lain K, Becks o J, Ma J, Wu G, Pa el SN, Hea ley SL, Phillips LA, Song G, Eas on J, Pa ke M, e al.
2013. The genomic landscape o hypodiploid acu e lymphoblas ic leukemia. Na u e Gene ics 45:242–252. doi:
10.1038/ng.2532
Huang FT, Yu K, Bal e BB, Selsing E, O uc Z, Khamlichi AA, Hsieh CL, Liebe MR. 2007. Sequence dependence
o ch omosomal R-loops a he immunoglobulin hea y-chain Smu class swi ch egion. Molecula and Cellula
Biology 27:5921–5932. doi: 10.1128/MCB.00702-07
Jenja oenpun P, Wongsu awa T, Yenamand a SP, Kuzne so VA. 2015. QmRLFS- inde : a model, web se e
and s and-alone ool o p edic ion and analysis o R-loop o ming sequences. Nucleic Acids Resea ch 43:
W527–W534. doi: 10.1093/na /gk 344
Jonke s I, Lis JT. 2015. Ge ing up o speed wi h ansc ip ion elonga ion by RNA polyme ase II. Na u e Re iews.
Molecula Cell Biology 16:167–177. doi: 10.1038/n m3953
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 24 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine
Kaikkonen MU, Spann NJ, Heinz S, Romanoski CE, Allison KA, S ende JD, Chun HB, Tough DF, P injha RK,
Benne C, Glass CK. 2013. Remodeling o he enhance landscape du ing mac ophage ac i a ion is coupled o
enhance ansc ip ion. Molecula Cell 51:310–325. doi: 10.1016/j.molcel.2013.07.010
Killick R, Eckley IA. 2014. changepoin : An R Package o Changepoin Analysis. Jou nal o S a is ical So wa e
58:1–19 . doi: 10.18637/jss. 058.i03
Killick R, Fea nhead P, Eckley IA. 2012. Op imal De ec ion o Changepoin s Wi h a Linea Compu a ional Cos .
Jou nal o he Ame ican S a is ical Associa ion 107:1590–1598. doi: 10.1080/01621459.2012.737745
K ij he J. 2015. R sne: T-Dis ibu ed S ochas ic Neighbo Embedding using Ba nes-Hu Implemen a ion. R
package e sion 0.10. h ps://CRAN.R-p ojec .o g/package=R sne.
Langmead B, T apnell C, Pop M, Salzbe g SL. 2009. Ul a as and memo y-e icien alignmen o sho DNA
sequences o he human genome. Genome Biology 10:R25. doi: 10.1186/gb-2009-10-3- 25
Maia AT, an de Velden VH, Ha ison CJ, Szczepanski T, Williams MD, G i i hs MJ, an Dongen JJ, G ea es MF.
2003. P ena al o igin o hype diploid acu e lymphoblas ic leukemia in iden ical wins. Leukemia 17:2202–2206.
doi: 10.1038/sj.leu.2403101
Ma hews AG, Kuo AJ, Ramo´ n-Maiques S, Han S, Champagne KS, I ano D, Galla do M, Ca ney D, Cheung P,
Ciccone DN, Wal e KL, U z PJ, Shi Y, Ku a eladze TG, Yang W, Gozani O, Oe inge MA. 2007. RAG2 PHD
inge couples his one H3 lysine 4 ime hyla ion wi h V(D)J ecombina ion. Na u e 450:1106–1110. doi: 10.
1038/na u e06431
Meng FL, Du Z, Fede a ion A, Hu J, Wang Q, Kie e -Kwon KR, Meye s RM, Amo C, Wasse man CR, Neube g
D, Casellas R, Nussenzweig MC, B adne JE, Liu XS, Al FW. 2014. Con e gen ansc ip ion a in agenic
supe -enhance s a ge s AID-ini ia ed genomic ins abili y. Cell 159:1538–1548. doi: 10.1016/j.cell.2014.11.014
Mo i H, Colman SM, Xiao Z, Fo d AM, Healy LE, Donaldson C, Hows JM, Na a e e C, G ea es M. 2002.
Ch omosome ansloca ions and co e leukemic clones a e gene a ed du ing no mal e al de elopmen .
P oceedings o he Na ional Academy o Sciences o he Uni ed S a es o Ame ica 99:8242–8247. doi: 10.1073/
pnas.112218799
Papaemmanuil E, Rapado I, Li Y, Po e NE, Wedge DC, Tubio J, Alexand o LB, Van Loo P, Cooke SL, Ma shall
J, Ma inco ena I, Hin on J, Gundem G, an Del FW, Nik-Zainal S, Jones DR, Ramak ishna M, Ti ley I,
S ebbings L, Le oy C, e al. 2014. RAG-media ed ecombina ion is he p edominan d i e o oncogenic
ea angemen in ETV6-RUNX1 acu e lymphoblas ic leukemia. Na u e Gene ics 46:116–125. doi: 10.1038/ng.
2874
Paulsson K, Lilljebjo
¨ n H, Bilogla A, Olsson L, Rissle M, Cas o A, Ba bany G, Fogels and L, No dg en A,
Sjo
¨g en H, Fio e os T, Johansson B. 2015. The genomic landscape o high hype diploid childhood acu e
lymphoblas ic leukemia. Na u e Gene ics 47:672–676. doi: 10.1038/ng.3301
Pa i R, Gazumyan A, Janko ic M, Di Vi gilio M, Klein I, Ansa ah-Sob inho C, Resch W, Yamane A, Reina San-
Ma in B, Ba e o V, Nieland TJ, Roo DE, Casellas R, Nussenzweig MC. 2010. Ac i a ion-induced cy idine
deaminase a ge s DNA a si es o RNA polyme ase II s alling by in e ac ion wi h Sp 5. Cell 143:122–133. doi:
10.1016/j.cell.2010.09.017
Pe anis E, Wang J, Ro hschild G, Lim J, Chao J, Rabadan R, Economides AN, Basu U. 2014. Noncoding RNA
ansc ip ion a ge s AID o di e gen ly ansc ibed loci in B cells. Na u e 514:389–393. doi: 10.1038/
na u e13580
P esco EM, P oud oo NJ. 2002. T ansc ip ional collision be ween con e gen genes in budding yeas .
P oceedings o he Na ional Academy o Sciences o he Uni ed S a es o Ame ica 99:8796–8801. doi: 10.1073/
pnas.132270899
Qian J, Wang Q, Dose M, P ue N, Kie e -Kwon KR, Resch W, Liang G, Tang Z, Ma he´ E, Benne C, Dubois W,
Nelson S, Vian L, Oli ei a TY, Janko ic M, Hakim O, Gazumyan A, Pa i R, Awas hi P, Song B, e al. 2014. B cell
supe -enhance s and egula o y clus e s ec ui AID umo igenic ac i i y. Cell 159:1524–1537. doi: 10.1016/j.
cell.2014.11.013
Rao SS, Hun ley MH, Du and NC, S ameno a EK, Bochko ID, Robinson JT, Sanbo n AL, Machol I, Ome AD,
Lande ES, Aiden EL. 2014. A 3D map o he human genome a kilobase esolu ion e eals p inciples o
ch oma in looping. Cell 159:1665–1680. doi: 10.1016/j.cell.2014.11.021
Robbiani DF, De oubaix S, Feldhahn N, Oli ei a TY, Callen E, Wang Q, Janko ic M, Sil a IT, Rommel PC, Bosque
D, Eisen eich T, Nussenzweig A, Nussenzweig MC. 2015. Plasmodium in ec ion p omo es genomic ins abili y
and AID-dependen B cell lymphoma. Cell 162:727–737. doi: 10.1016/j.cell.2015.07.019
Robe s KG, Mullighan CG. 2015. Genomics in acu e lymphoblas ic leukaemia: insigh s and ea men
implica ions. Na u e Re iews. Clinical Oncology 12:344–357. doi: 10.1038/n clinonc.2015.38
Scha z DG, Swanson PC. 2011. V(D)J ecombina ion: mechanisms o ini ia ion. Annual Re iew o Gene ics 45:
167–202. doi: 10.1146/annu e -gene -110410-132552
Scheidegge A, Nechae S. 2016. RNA polyme ase II pausing as a con ex -dependen eade o he genome.
Biochemis y and Cell Biology = Biochimie E Biologie Cellulai e 94:82–92. doi: 10.1139/bcb-2015-0045
Sigo a AA, Mullen AC, Molinie B, Gup a S, O lando DA, Guen he MG, Almada AE, Lin C, Sha p PA,
Giallou akis CC, Young RA. 2013. Di e gen ansc ip ion o long noncoding RNA/mRNA gene pai s in
emb yonic s em cells. P oceedings o he Na ional Academy o Sciences o he Uni ed S a es o Ame ica 110:
2876–2881. doi: 10.1073/pnas.1221904110
Skou i-S a haki K, P oud oo NJ. 2014a. A double-edged swo d: R loops as h ea s o genome in eg i y and
powe ul egula o s o gene exp ession. Genes & De elopmen 28:1384–1396. doi: 10.1101/gad.242990.114
Skou i-S a haki K, Kamienia z-Gdula K, P oud oo NJ. 2014b. R-loops induce ep essi e ch oma in ma ks o e
mammalian gene e mina o s. Na u e 516:436–439. doi: 10.1038/na u e13787
Heina
¨niemi e al. eLi e 2016;5:e13087. DOI: 10.7554/eLi e.13087 25 o 26
Resea ch a icle Genes and Ch omosomes Human Biology and Medicine