Cohesin and CTCF differentially affect chromatinarchitecture and gene expression in human cellsJessica Zuina,1, Jesse R. Dixonb,c,d,1, Michael I. J. A. van der Reijdena, Zhen Yeb,e, Petros Kolovosa,Rutger W. W. Brouwerf,g, Mariëtte P. C. van de Corputa, Harmen J. G. van de Werkena, Tobias A. Knochh,i,Wilfred F. J. van IJckenf, Frank G. Grosvelda,j, Bing Renb,e,2, and Kerstin S. Wendta,2

aDepartment of Cell Biology, iBiophysical Genomics, Department of Cell Biology, fCenter for Biomics, jCancer Genomics Center, Erasmus Medical Center, 3015GE, Rotterdam, The Netherlands; bLaboratory of Gene Regulation, Ludwig Institute for Cancer Research, La Jolla, CA 92093; cMedical Scientist TrainingProgram, dBiomedical Sciences Graduate Program, eDepartment of Cellular and Molecular Medicine, Institute of Genomic Medicine, Moores Cancer Center,San Diego School of Medicine, University of California, San Diego, La Jolla, CA 92093; gNetherlands Bioinformatics Centre, 6500 HB, Nijmegen, TheNetherlands; and hGenome Organization and Function, Bioquant Centre/German Cancer Research Center, 69120 Heidelberg, Germany

Edited* by Richard A. Young, Massachusetts Institute of Technology, Cambridge, MA, and approved October 31, 2013 (received for reviewSeptember 20, 2013)

Recent studies of genome-wide chromatin interactions haverevealed that the human genome is partitioned into many self-associating topological domains. The boundary sequences be-tween domains are enriched for binding sites of CTCC-bindingfactor (CTCF) and the cohesin complex, implicating these two fac-tors in the establishment or maintenance of topological domains.To determine the role of cohesin and CTCF in higher-order chro-matin architecture in human cells, we depleted the cohesin com-plex or CTCF and examined the consequences of loss of thesefactors on higher-order chromatin organization, as well as thetranscriptome. We observed a general loss of local chromatininteractions upon disruption of cohesin, but the topologicaldomains remain intact. However, we found that depletion of CTCFnot only reduced intradomain interactions but also increased inter-domain interactions. Furthermore, distinct groups of genes be-come misregulated upon depletion of cohesin and CTCF. Takentogether, these observations suggest that CTCF and cohesin contrib-ute differentially to chromatin organization and gene regulation.

Hi-C | transcriptional regulation | 4C | HOX cluster

Recent studies of the topological organization of the genomesuggest that CTCC-binding factor (CTCF) and cohesin

might be involved in establishment or maintenance of topologicaldomains in the mammalian genome, as their binding sites areenriched at the boundaries of these domains (1). It was proposedthat CTCF and cohesin might work together to facilitate long-range interactions in the genome (2). First, CTCF and thecohesin complex, consisting of the core subunits SMC3, SMC1,RAD21, and STAG1/SA1 or STAG2/SA2, were found to coloc-alize extensively throughout mammalian genomes (3–5). Second,both factors are involved in mediating long-range interactions(6–11). Finally, cohesin was shown to be important for CTCF’schromatin insulation function (3–5), whereas CTCF is necessaryto recruit cohesin to the shared binding sites but not to chro-matin (3). CTCF and cohesin have also been recently correlatedwith both interaction frequency and gene expression duringdifferentiation (12), indicating that they may play major roles inmediating the impacts of chromatin structure on gene regulation.However, the exact mechanisms these factors use to contributeto chromatin structure and gene regulation are unclear, as de-pletion of these factors has not yet been systematically tested ona genome-wide basis. Whether the two factors work in concertor independently, through mechanisms, such as long-range en-hancer looping (13) or chromatin insulation (2) to controlchromatin structure and gene expression, is unknown. To de-termine the role of cohesin and CTCF in higher-order chromatinarchitecture in human cells, we depleted the cohesin complex orCTCF and examined the consequences of loss of these factors ondomain structure and gene expression.

ResultsProteolytic Cleavage of RAD21 Leads to Loss of Long-Range ChromatinInteractions. To understand the contribution of cohesin to genomeorganization, we generated a HEK293T cell line containing anepisome-based vector allowing doxycycline-inducible expressionof siRNA targeting endogenous RAD21 and a RAD21-EGFPvariant containing a recognition site for Human rhinovirus 3C(HRV) protease (RAD21cv) (14) (Fig. 1 A and B). Three daysafter induction, RAD21cv replaced up to 90% the endoge-nous RAD21 (SI Appendix, Fig. S1A) and was incorporated inthe cohesin complex (SI Appendix, Fig. S1B). Subsequent trans-fection of these cells (transfection efficiency ∼80–90%) witha construct expressing HRV protease led to full cleavage ofRAD21cv within 24 h (Fig. 1C); TEV (tobacco etch mosaicvirus) protease was used as negative control.With this system we could rapidly remove the cohesin complex

from interphase chromosomes, similar to natural RAD21 cleavageby separase in mitosis (15). Similar systems have been used beforeto study cohesin in yeast and fly (14, 16, 17).Fractionation of HRV- or TEV-transfected RAD21cv cells

into soluble and chromatin-bound fraction showed that RAD21cv

Significance

For the 2m DNA to fit into the tiny cell nucleus, it is wrappedaround nucleosomes and folded into loops clustering togetherin domains. Genome function depends on this 3D-organization,especially on-going dynamic processes like transcription. Tech-niques studying the network of DNA contacts genome-widehave recently revealed this 3D architecture, but the proteinfactors behind this are not understood. We study two proteinsthat are known to help form DNA loops: cohesin and CTCC-binding factor (CTCF). Respective depletion and analysis ofDNA contacts genome-wide show that CTCF is required toseparate neighboring folding domains and keep cohesin inplace, whereas cohesin is important for shaping the domains.Consistently, we observe different changes of gene expression.

Author contributions: J.Z., J.R.D., F.G.G., B.R., and K.S.W. designed research; J.Z., J.R.D.,M.I.J.A.v.d.R., Z.Y., M.P.C.v.d.C., W.F.J.v.I., and K.S.W. performed research; J.Z., J.R.D., Z.Y.,P.K., R.W.W.B., H.J.G.v.d.W., T.A.K., and F.G.G. analyzed data; and J.Z., J.R.D., B.R.,and K.S.W. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.

Data deposition: The sequence reported in this paper has been deposited in the GenBankdatabase (accession no. GSE44267).

See Commentary on page 889.1J.Z. and J.R.D. contributed equally to this work.2To whom correspondence may be addressed. E-mail: biren@ucsd.edu or k.wendt@erasmusmc.nl.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1317788111/-/DCSupplemental.

996–1001 | PNAS | January 21, 2014 | vol. 111 | no. 3 www.pnas.org/cgi/doi/10.1073/pnas.1317788111

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

had fully replaced wild-type RAD21 and bound with a comparablelevel to chromatin. Transfection of RAD21cv cells with HRVprotease led to release of RAD21cv from chromatin, but notTEV protease. CTCF remained chromatin-bound (Fig. 1D).The chromatin-bound RAD21 fraction was reduced by 70–80%,estimated from ChIP/quantitative PCR (qPCR) experimentstargeting several cohesin binding sites (Fig. 1E). This rapid de-struction of cohesin allowed us to study the immediate effect ofcohesin loss on chromatin organization and transcription, with-out interfering with cohesin’s function in cell division, becauseneither a shift in cell cycle distribution nor cells arrested in mi-tosis were observed (SI Appendix, Fig. S2).To test whether the cleavage of cohesin affects long-range

chromatin interactions, we used the 4C technique (18, 19), whichidentifies all regions interacting with one “viewpoint.” To ex-amine the interior and the borders of one topological domain (1)(Fig. 1F) after RAD21 cleavage, we selected six viewpoints (Vp1–Vp6) overlapping cohesin/CTCF sites (Fig. 1 G and H) inchr11p15.5 comprising the noncoding RNA H19, the insuline-like growth factor 2 (IGF2) gene, and other imprinted genes

(here, the H19/IGF2 domain), which we previously used to es-tablish the role of cohesin in chromatin insulation (3).In control cells (RAD21cv/TEV) we observe, as reported be-

fore (8), that the IGF2 promoter region (Vp1) interacts stronglywith an intergenic region between H19 and IGF2 and oneviewpoint there confirms this (Vp2). Further contacts of Vp1 andVp2 persist to the region upstream ofH19 (Vp3) and over a 500-kbregion until the proximal keratin associated protein gene cluster(KRTAP) near the domain boundary. One viewpoint (Vp4) ina cohesin-depleted region shows only weak interactions. A view-point at the upstream boundary (Vp5) shows weak interactionswith both domains and another viewpoint within the neighboringdomain downstream (Vp6) consistently shows interactions until thedomain boundary. We observed similar interaction profiles in thebreast endothelial cell line 1-7HB2 (abbreviated HB2), indicatingtheir conservation between cell lines (SI Appendix, Fig. S3).Cleavage of RAD21 led to a global loss of interactions across

the entire domain at all viewpoints (Fig. 1H, red line). Correla-tion analysis of the interactions identified by the different view-points in the control 4C experiments revealed close correlationbetween viewpoints Vp1–Vp5, consistent within their localiza-tion in the same topological domain. This correlation was re-duced after RAD21 cleavage, reflecting the loss of interactions(SI Appendix, Fig. S4). A control experiment with a cell linelacking the HRV cleavage site in RAD21-EGFP (RAD21wt) didnot show altered cohesin binding and long-range interactionsafter transfection with HRV protease (SI Appendix, Fig. S5).These results strongly support that cohesin plays a role in higher-order chromatin structure within this domain.

RAD21 Depletion Predominantly Reduces Shorter-Range InteractionsWithin Topological Domains. To investigate whether cohesin playsa general role in topological domain organization, we performedHi-C experiments with control cells (RAD21cv/TEV) and afterRAD21-cleavage (RAD21cv/HRV). We obtained greater than370 million nonredundant uniquely mapping read pairs for bothcontrol and RAD21-cleaved cells, split between two biologicalreplicates for each condition and high correlation between thereplicates was observed (Pearson correlation 0.90/0.90) (SI Ap-pendix, Fig. S6). We normalized the Hi-C interaction frequenciesaccording to the iterative correction method (20). For eachcontrol and RAD21 cleavage replicate we located the topologi-cal domains using a previously described algorithm (1). Of note,the resolution of Hi-C data is dependent upon the depth of se-quencing for each sample. With our current sequencing depth,we can readily identify topological domains and analyze relation-ships between interaction frequencies and various genomic ele-ments or DNA binding factors. However, current high-throughputsequencing technologies are still insufficient to identify differencesin interaction frequency between individual binding sites on agenomewide level using Hi-C data. Therefore, our analysis focuseson aggregate genomewide trends in interaction frequency andtheir relationship to cohesin and CTCF binding patterns.To determine genomewide cohesin binding, we performed

ChIP-seq for the cohesin subunit SMC3 in control cells (RAD21cv/TEV). As observed previously (1), cohesin sites were enriched atthe boundaries of topological domains (SI Appendix, Fig. S7A),although this was only seen for SMC3 sites overlapping with CTCF(SI Appendix, Fig. S7 C and D).To correlate overall Hi-C interaction frequencies with SMC3

binding, we divided the genome into 40-kb bins and stratifiedthe interacting bin-pairs according to whether SMC3 binds toboth bins (“SMC3 2×”), only one bin (“SMC3 1×”), or none(“none”) (Fig. 2A). We observed a higher interaction frequencyin control Hi-C experiments between bin-pairs containingSMC3 sites on both ends than when only one or no SMC3 sitewas present (Fig. 2B), supporting the hypothesis that cohesinmediates long-range chromatin interactions genomewide. Uponcleavage of RAD21, we observed an overall loss in local chromatininteraction frequency, primarily occurring at distances up to 2 Mb,with a maximum in the range between 100 and 200 kb (Fig. 2C).

RAD21 EGFPsiRNA RAD21

HRV Cleavage site

P

DoxycyclineA

B

C

D

Time after HRV protease transfection:0 4 7 9 14 20 24h

RAD21-EGFPCleaved product

SMC3

EGFP

γ-tubulin

Blot:

Supernatant Chromatin

RAD21-EGFPEndogenous RAD21

RAD21-EGFPC-terminalcleavage product

*This antibody does not recognize the C-terminal RAD21 cleavage product

F

G

KRTAP5 CTSD H19 IGF2

DBDB

Scalechr11:

(IMR90)Domains

1 Mb hg181,500,000 2,000,000 2,500,000 3,000,000

RefseqGenes

1 23 4 5 6 7ChIP/qPCR Primer

H

129

10,000 Vp6

EChIP:Cells:IgG EGFPIgGEGFP

Cohesin binding sites Negative site

RAD21cv/TEVRAD21cv/TEVRAD21cv/HRVRAD21cv/HRV

0

0.1

0.2

0.3

0.4

0.5

6 5 1 2 3 4 7 8

%IP

ofinp

ut

RAD21*

γ-tubulin

CTCF

EGFP

SA1/2

Blot:

Topo II

4C with six viewpoints: RAD21cv/TEVRAD21cv/HRV

190

10,000 Vp1

RP

M Viewpoint

Transfection:dox induction:

_SMC3 ChIP-seq

_40

0

Primer

TEVHRV

- TEVHRV

-+ + + +- -

HEK293T stable cell line

Expression of RAD21cvDepletion of wt RAD21

TEV

+ dox

Transfection

72 hours

for 24 hours HRV

RAD21

SMC1 SMC3

RAD21-EGFP

43

Vp25,000

RP

MR

PM

183

Vp5

RP

M

10,000147

Vp4

RP

M

10,000296

Vp3

RP

M

10,000

Fig. 1. Cohesin cleavage reduces long-range interactions within the H19/IGF2 domain. (A) Scheme of the expression construct. (B) Outline of the ex-periment. (C ) Time course showing full cleavage of RAD21cv 24 h afterHRV transfection. (D) Fractionation of uninduced cells (−dox), and trans-fected (TEV or HRV) RAD21cv cells into soluble and chromatin-boundfraction. Blotting for RAD21 shows full replacement of endogenous RAD21by RAD21cv after induction (+dox). Detection of the RAD21cv C-terminalEGFP-tag shows full cleavage of RAD21cv after HRV transfection (+dox,HRV) and release of RAD21cv as well as the cohesin subunits STAG1 andSTAG2 (SA1/2) from chromatin. CTCF binding to chromatin is not affected.(E) ChIP-qPCR with anti-EGFP targeting the RAD21cv EGFP-tag shows a re-duced ChIP signal after HRV transfection at cohesin sites. (F–H) The effectof RAD21 cleavage on long-range interactions was tested by 4C at sixdifferent viewpoints in two topological domains of chromosome 11. (F)Domain identification in IMR90 cells (domain boundaries, blue boxes) (1). (G)Cohesin sites (SMC3) determined in control cells. Primer pairs used forqPCR in E are indicated. (H) The 4C interaction profiles for six differentviewpoints (highlighted in green) without (RAD21cv/TEV) and with RAD21cleavage (RAD21cv/HRV). Data are displayed as reads per million (RPM) andonly interactions above a cutoff based on P value < 0.05 are displayed.

Zuin et al. PNAS | January 21, 2014 | vol. 111 | no. 3 | 997

CELL

BIOLO

GY

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

The decrease in chromatin interaction frequencies after RAD21cleavage was the highest when both interacting regions werebound by SMC3 (Fig. 2C, Inset). The observed loss of interactions,tested for example on bins 240-kb apart, is statistically significant(P = 3 × 10−197, Wilcoxon test) (Fig. 2D).We next investigated the effects of cohesin complex de-

struction on topological domains. The positions of most topo-logical domains did not markedly change upon cleavage ofRAD21 and their pattern is still readily apparent in the in-teraction heat maps. Although we consistently called fewerdomains in the RAD21-depleted cells, there was a strongoverlap between domain boundaries identified in the controland RAD21-depleted cells (Fig. 2 E and F). This observation isnot surprising because cohesin and CTCF likely are not theonly factors responsible for long-range interactions (12, 13).However, consistent with the previously described general lossin interaction frequency, we also observed a clear reduction ininteraction frequency after RAD21-depletion, both within andbetween domains (SI Appendix, Fig. S8A). Interestingly, the de-gree of depletion in interaction frequency within domains wasmost pronounced when one or both interacting bins were asso-ciated with a boundary (SI Appendix, Fig. S8B). We performed3D-FISH (21) and measured distances between cosmid-basedFISH probes located at boundaries of one domain comprisingthe homeobox D (HOXD) genes (Fig. 2G). Consistent withHi-C results, we observed significantly increased distances (P =5,632 × 10−4, ANOVA) between the FISH probes after RAD21cleavage (Fig. 2 H and I). Consequently, these results suggestthat the cohesin complex contributes to the self-associationwithin topological domains, in part by promoting interactionsbetween regions near the boundaries. However, cohesin depletiondoes not appear to contribute to the positioning and segregationof neighboring domains.

Cohesin and CTCF Shape the Topological Domain Organization inNonredundant Ways. To determine the role of CTCF in mediat-ing chromatin interactions and to compare it to the effects ofRAD21 cleavage, we performed Hi-C experiments in duplicatefor CTCF and control siRNA knockdowns in HEK293T cells(knockdown efficiency 80%) (SI Appendix, Fig. S9 A and B). Weobtained between 95 and 288 million unique reads for each rep-licate, with high correlation between them (Pearson correlation0.93/0.94) (SI Appendix, Fig. S6). Similar to SMC3, CTCF bindingsites are enriched at the boundaries of topological domains incontrol cells (SI Appendix, Fig. S7B). Interestingly, CTCF sites atdomain boundaries are more closely aligned to the CTCF bindingmotifs than CTCF binding sites found within topological domains(SI Appendix, Fig. S7E). CTCF binding also correlates with thestrength of Hi-C interaction frequency, where interacting bin-pairsbound by CTCF on each side form stronger interactions comparedwith regions with only one or no CTCF sites (Fig. 3 A and B).Upon knockdown of CTCF, we observed a loss of interactions

within topological domains but, in contrast to RAD21 cleavage,we also observed a gain of interactions (Fig. 3 C and D). Detailedanalysis of the changes in interaction frequencies versus dis-tance between bins, localization of bins in the same (intra-domain) or different topological domains (interdomain), andalso whether bins have SMC3 or CTCF sites, revealed very in-teresting details (Fig. 3 E–J). RAD21 depletion appeared to mostmarkedly affect interacting loci separated by 100–200 kb (Figs. 2Cand 3E), but CTCF knockdown appeared to most prominentlyaffect interacting loci separated by less than 100 kb (Fig. 3H).This finding implies that depletion of CTCF or cohesin affectsinteractions within topological domains differently. However,in both cases the loss of interactions was more pronounced forintradomain interactions (Fig. 3 E and H, blue line), and in thecase of RAD21 depletion, even more when interacting bins have

177,000,000

TEV1 TEV2

TEV1

TEV2

HRV1 HRV1

HRV2

HRV2

1,560(1,747)

(1,747)

(1,830) (1,830)

(1,674) (1,674)

(1,383)

(1,383))

1,497(

1,283 1,227

SMC3 2x:

SMC3 1x:

None:

or

102 4 6 80Distance (Mbp)

1.00.5 2.01.5-0.4

-0.3

-0.2

-0.1

0.0

∆ N

orm

aliz

ed In

tera

ctio

n Fr

eque

ncy

(HR

V-T

EV

)

SMC3 2xSMC3 1xNone

0

-0.6

-0.8

-0.4

-0.2

0

Distance (Mbp)

E

0

1

2

3

4

5

6

7

Nor

mal

ized

Inte

ract

ion

Freq

uenc

y24

0 kb

p in

tera

ctio

ns

TEV HRV

F

p = 3x10-197

I

0

0.5

1.0

1.5

Dist

ance

betw

een

prob

es (μ

m)

n=97n=92

p=0.0005632

TEV HRVTransfected RAD21cv cells

H

0.5 μm2 μm

TEV HRVTransfected RAD21cv cells:

40kb

SMC3

D

TEVNormalizedInteractionFrequency 0

6

-2

+2∆ NormalizedInteractionFrequency

HRV-TEV

chr2:

cosmids

500 kb hg18

RefSeq Genes

0

3.0

Nor

mal

ized

Inte

ract

ion

Freq

uenc

y

Control (TEV)

RAD21-depleted (HRV)

0

3.0

Nor

mal

ized

Inte

ract

ion

Freq

uenc

y

Scalechr2:

5 Mb hg18155,000,000 160,000,000

RND3

RBM43

NMI

TNFAIP6MIR4773-2MIR4773-1

RIF1

NEBARL5A

CACNB4STAM2

FMNL2

PRPF40AARL6IP6

RPRM

GALNT13

LOC100144595

KCNJ3 NR4A2GPD2

GALNT5

ERMN

CYTIPACVR1C

ACVR1

UPP2CCDC148-AS1

CCDC148

PKP4

DAPL1TANC1

WDSUB1

BAZ2B

MARCH7

CD302

LY75PLA2R1

ITGB6

RBMS1

MIR4785

TANK

PSMD14

TBR1

SLC4A10DPP4GCG

FAPIFIH1GCA

KCNH7

10 _

1 _ _

-5 _0 -

_

-5 _0 -

DI

SMC3 ChIP-Seq

TEVDC

DI

DC

5

5HRV

A

C

GSMC3 2xSMC3 1xNone

Nor

mal

ized

Inte

ract

ion

Freq

uenc

y

2.0

1.8

1.6

1.4

1.2

1.0Fold

Cha

nge

over

“Non

e”

0

2

4

6

8

10

12

Distance (Mbp)0 0.5 1.0 1.5 2.0

Distance (Mbp)

B

0 0.5 1.0 1.5 2.0

Fig. 2. Cohesin cleavage reduces interactions within topological domains. (A) Stratification of the Hi-C interaction map based on SMC3 binding. Clustering ofinteracting 40-kb bins for presence of SMC3 (brown peak) at both interacting bins (2×), at one bin (1×) or no SMC3 (none). (B) The normalized interactionfrequency is plotted versus the distance of interacting bins for the different bin clusters (SMC3 2×, 1×, and none). (Inset) In the fold-change relative to the“none” category the SMC3 2× cluster has highest interactions frequencies. (C) Cohesin destruction reduces the interactions between bins. The change ofinteraction frequency after RAD21 cleavage (HRV-TEV, dark-green curve) is plotted relative to the distance between interacting bins and reveals a reductionof interactions at distances up to 4 Mb. (Inset) The loss of interactions for the SMC3 2×, 1×, and none categories. (D) Interaction frequencies are sig-nificantly reduced after RAD21 cleavage, shown here at the example of bins 240-kb apart. (E ) Normalized Hi-C interaction frequencies observed inRAD21cv cells transfected with either TEV or HRV protease are shown. SMC3 ChIP-sequencing, topological domains positions (DC, domain calls) anddirectionality index (DI) are shown. Arrows indicate regions with significant changes. (F ) Comparison of topological domain boundary calls between Hi-Creplicates (TEV1/TEV2; HRV1/HRV2) and control (TEV) and RAD21-depleted cells (HRV). Variations between the respective replicates and between controland RAD21 cleavage experiments are comparable, indicating that the number of domains does not change. (G) Position of cosmid-based DNA-FISH probesat the topological domain including the HOXD locus. The color of the cosmid probes (red, green) corresponds to the DNA-FISH images in (H). Arrows markthe interactions visualized by DNA-FISH. (H) DNA-FISH using the cosmid probes shown in G in control cells (TEV) and after RAD21 cleavage (HRV). Themarked DNA-FISH signals (white boxes) are shown enlarged at the right side of each panel. Consistent with the Hi-C experiments, we observe separationof the FISH signals after RAD21 cleavage. (I) Distances between the FISH-probes observed in the TEV and HRV experiments. The P value was calculatedusing an ANOVA test on the log distances.

998 | www.pnas.org/cgi/doi/10.1073/pnas.1317788111 Zuin et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

SMC3 sites (Fig. 3F). In the CTCF knockdown experiment wedid not observe this (Fig. 3I).Most remarkably, CTCF and cohesin depletion differentially

altered interactions between topological domains (Figs. 3 E andH, yellow lines). RAD21 depletion again led to reduced inter-actions (Fig. 3E, yellow line), but CTCF depletion led to a gainof interactions between neighboring domains (Fig. 3H, yellowline). This increase was seen over a distance scale up to 2 Mb,consistent with the range of domain sizes (SI Appendix, Fig.S11B), and was more pronounced when CTCF-binding sites wereinvolved (Fig. 3G and J). This suggests that CTCF is necessary tomaintain topological domain boundaries. Interactions gained byCTCF depletion could involve cohesin, which is now delocalized(3) but still chromatin-bound (SI Appendix, Fig. S9C). Consis-tently, the largest gains in interdomain interaction frequencyafter CTCF knockdown occurred between bins containingCTCF or cohesin sites (SI Appendix, Fig. S10). Taken together,our observations suggest that cohesin and CTCF shape geno-mic structure on the level of topological domains in a non-redundant manner.

Chromatin Substructures Within Topological Domains also Depend onCohesin and CTCF. Frequently, topological domains appear to har-bor substructures (SI Appendix, Fig. S11A), which have been re-cently described as “subdomains” or “sub-TADs” (12). Thissuggests that some topological domains have a nonuniform interior

structure, and certain factors may influence these intradomainstructures. Cohesin and CTCF have recently been shown to beenriched at the boundaries of subdomains (12). To determinehow cohesin or CTCF depletion affects these substructures,we modified our previously applied hidden Markov model (1)to identify subdomains. We identified ∼2,600 subdomains incontrol experiments (RAD21cv/TEV and control siRNA)with a median size of 520 kb, compared with a 1,080-kb me-dian size for topological domains (SI Appendix, Fig. S11B).Overall, after RAD21 cleavage similar alterations in in-teraction frequencies could be observed when consideringboth domains and subdomains for intra- and interdomaininteractions (SI Appendix, Fig. S11 C and D). However, afterCTCF depletion the loss of interactions within subdomainsseemed to be more pronounced compared with interactionswithin entire topological domains, and the gain in interactionswas stronger between topological domains compared with be-tween subdomains (SI Appendix, Fig. S11 C and E). Therefore,CTCF appears to function within domains less as an “insulatingfactor” and more that it appears to be contributing to non-uniform structures that exist within some topological domains.In contrast, cohesin appears to function, regardless of its ge-nomic context, as a factor that contributes to the associationbetween loci (SI Appendix, Fig. S11D). The observed differ-ences between these two factors suggest that cohesin mayfunction primarily by causing the interaction between loci

5 _

-5 _5 _

-5 _RefSeq Genes

5 _

-5 _5 _

-5 _RefSeq Genes

CTCF

Intra-Domain

SMC3 ≥1xSMC3 None

Distance (Mbp)

0.10

-0.2-0.3-0.4-0.5-0.6

-0.1

F

Inter-Domain

SMC3 ≥1xSMC3 None

0.10

-0.2-0.3-0.4-0.5-0.6

-0.1

Distance (Mbp)

G

Intra-Domain

CTCF ≥1xCTCF None

Distance (Mbp)

0.2

-0.2

-0.4

-0.6

-0.8

0

I

Inter-Domain

CTCF ≥1xCTCF None

Distance (Mbp)

0.2

-0.2

-0.4

-0.6

-0.8

0

J

0.10

-0.2-0.3-0.4-0.5-0.6

Distance (Mbp)

E-0.1

Intra-DomainInter−Domain

All

Distance (Mbp)

0.2

-0.2

-0.4

-0.6

-0.8

0

H

Intra-DomainInter−Domain

All

+1

-1

4

0

4

0

Nor

mal

ized

Inte

ract

ion

Freq

uenc

y

Nor

mal

ized

Inte

ract

ion

Freq

uenc

y

∆ N

orm

aliz

edIn

tera

ctio

nFr

eque

ncy

(HR

V -

TEV

)

TEV

C

Scalechr6:

2 Mbhg18 113,000,000 114,000,000 115,000,000 116,000,000 117,000,000

0 -

0 -

Control (TEV)

RAD21-cleavage (HRV)

DI

DIHRV

DC

DC

Δ (HRV-TEV)

+1

-1

4

0

4

0

Nor

mal

ized

Inte

ract

ion

Freq

uenc

y

Nor

mal

ized

Inte

ract

ion

Freq

uenc

y

∆ N

orm

aliz

edIn

tera

ctio

nFr

eque

ncy

(siC

TCF-

siC

ontro

l)

siControl

siCTCF

D

Scalechr6:

2 Mbhg18 113,000,000 114,000,000 115,000,000 116,000,000 117,000,000

0 -

0 -

siControl

siCTCF

DI

DI

DC

DC

Δ (siCTCF-siControl)

A15

10

5

0

BCTCF 2x:

CTCF 1x:

None:

or

CTCF 2xCTCF 1xNone

Distance (Mbp)0 0.5 1.0 1.5 2.0

0 0.5 1.0 1.5 2.0Distance (Mbp)

1.8

1.6

1.4

1.2

1.0Fo

ld C

hang

e

40kb

Nor

mal

ized

IF

0 0.5 1.0 1.5 2.0

∆ N

orm

aliz

ed IF

* (

HR

V-T

EV

)

0 0.5 1.0 1.5 2.0

0 0.5 1.0 1.5 2.0

0 0.5 1.0 1.5 2.0

0 0.5 1.0 1.5 2.0

0 0.5 1.0 1.5 2.0

∆ N

orm

aliz

ed IF

* (c

iCTC

F-si

Con

trol)

∆ N

orm

aliz

ed IF

* (

HR

V-T

EV

)∆

Nor

mal

ized

IF*

(H

RV

-TE

V)

∆ N

orm

aliz

ed IF

* (c

iCTC

F-si

Con

trol)

∆ N

orm

aliz

ed IF

* (c

iCTC

F-si

Con

trol)

Fig. 3. CTCF depletion reduces the function of domain boundaries. (A) As in Fig.2A, interacting 40-kb bins were analyzed for the presence of CTCF at one orboth interacting bins. (B) The normalized interaction frequency was plotted versus the distance between bins for each class of interactions (CTCF 2×, CTCF 1×,and none). (Inset) In the fold-change relative to the “none” category, the CTCF 2× class has a higher interaction frequency than CTCF 1× with a maximum forbins 100- to 200-kb apart. (C) The differential interaction map (HRV-TEV) displays changes in interaction frequencies after RAD21 cleavage (red, gain ofinteractions; blue, loss of interactions). RAD21 depletion leads predominantly to reduced interactions within domains. The domain identification (domaincalls, DC; directionality index, DI) is also shown. (D) Similar to C, but showing the changes in interaction frequencies at the same region after CTCF de-pletion by siRNA. A similar pattern of reduced intradomain interaction frequency as in C is observed, visible as blue outline of the domains in the dif-ferential plot (siCTCF-siControl). CTCF depletion yields increased interdomain interactions, visible as red signals between domains. (E ) Quantification ofthe average change of interaction frequencies after RAD21 depletion, analyzed separately for intradomain (blue) and interdomain (yellow) interactions.In both cases, RAD21 depletion leads to a reduced interaction frequency. (F and G) The frequency change of intra- (F ) and interdomain (G) interactionswas analyzed for the presence of SMC3 on the interacting bins. The loss of interactions is in both cases correlated with SMC3-binding (F, purple; G,orange). (H) Quantification of the average change of interaction frequencies after CTCF siRNA depletion separated for intradomain (blue) and inter-domain (yellow) interactions. CTCF depletion leads to a reduced interaction frequency within and to an increased interaction frequency betweendomains. (I and J) The change of interaction frequency of intra- (I) and interdomain (J) interactions was further analyzed for the presence of CTCF on theinteracting bins. The gain of interactions is more pronounced for interdomain interactions (J) and is stronger when CTCF-sites are present in the inter-acting bins (J, orange).

Zuin et al. PNAS | January 21, 2014 | vol. 111 | no. 3 | 999

CELL

BIOLO

GY

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

throughout the genome, and CTCF may play a greater role indetermining the specificity of interactions and the identity oftopological domains.

Cohesin and CTCF Depletion Affect Gene Expression in DifferentWays. Loss of cohesin or CTCF seems to affect chromatinstructure in different ways. Given the intimate relationships be-tween chromatin structure and gene regulation, we predict thatloss of these two factors would also affect gene expression dif-ferentially. To test this, we performed RNA-sequencing (RNA-seq) in the control (RAD21cv/TEV) and RAD21-depleted(RAD21cv/HRV) cells, as well as CTCF RNAi and mock-trea-ted cells. In both cases we observed only modest changes in geneexpression (SI Appendix, Tables S1–S4), consistent with earlierobservations (3). We observe 48 and 161 differentially expressedgenes (false-discovery rate, FDR < 5%) for RAD21 and CTCFdepletion, respectively, but very little overlap between these sets(Fig. 4A). Among the genes with reduced expression after RAD21depletion are several Hox genes (HOXA11AS, HOXA-AS3,HOXB-AS3, HOXB5, HOXC9) (Fig. 4B). We validated the re-duced expression of HOXB-AS3, HOXA-AS3 and H19 by RT-PCR and qPCR (Fig. 4C). Hox genes have been shown to beregulated by antisense transcription as well as the topologicalorganization of the locus and the contact to remote enhancers(22, 23), but so far only for the HOXA cluster has a rolecohesin and CTCF at the barrier element been shown (24).Among genes that are differentially expressed after CTCF de-

pletion, we observed a clear enrichment of CTCF-binding at theirpromoters (Fig. 4 D and E), with a median distance from thetranscription start site to the nearest CTCF binding site being only191 bp (SI Appendix, Fig. S12B). In contrast, genes that are dif-ferentially expressed after cohesin depletion are not directly boundat their promoter by SMC3 (Fig. 4F), although they are locatedcloser to SMC3 binding sites than would be expected at random(median distance ∼4 kb) (SI Appendix, Fig. S12A). This findingindicates that altered expression of genes after RAD21 cleavagemaybe a product of higher-order chromatin structural changes, for ex-ample at the HOXA and HOXB cluster (SI Appendix, Fig. S13). Tovalidate this finding, we analyzed interactions of cohesin-regulatedgenes with DNase hypersensitive sites as markers for potential distalgene regulatory regions at a restriction fragment-level resolution.We observed that cohesin-regulated genes lose more interactionswith distal DNaseI hypersensitive sites than with noncohesin-regulated Ref-seq genes (Fig. 4G). These results suggest thatcohesin may regulate gene expression by affecting the interactionfrequency of genes with distal regulatory elements, whereas CTCFmay directly regulate genes by binding at their promoters.

DiscussionTo reveal the role of CTCF and cohesin in the 3D organizationof the human genome, we acutely depleted either the cohesincomplex or CTCF from cells and examined the changes inchromatin organization using a combination of 4C, Hi-C, and3D-FISH. We show that cohesin and CTCF contribute differ-entially to the topological domain architecture. First, CTCF orcohesin-bound regions are more likely to interact with each otherthan other genomic sites, extending recent observations at sixgenetic loci in mouse ES cells (12) and supporting a critical rolefor these factors in the folding of the chromatin fiber. Second,disruption of the cohesin complex by proteolytic cleavage leadsto loss of shorter-range chromatin interactions. Surprisingly, thisloss of chromatin interaction is not accompanied by breakingdown of topological domain organization. Third, depletion ofCTCF also reduces the intradomain interactions but, in contrastto cohesin removal, it also leads to increased interactions between

Myc

CDKN1A

HoxA-AS3

HoxB-AS3+2 -2

Log (Treatment/Control)

A

BC

G

E

H19

00.20.40.60.8

11.2

TEV HRV Cont. CTCF

Rel

ativ

eex

pres

sion

RAD21cv siRNA

HOXA-AS3

00.20.40.60.8

11.2

Rel

ativ

eex

pre s

sio n

TEV HRV Cont. CTCFRAD21cv siRNA

HOXB-AS3

00.20.40.60.8

11.2

Rel

ativ

eex

p re s

sion

TEV HRV Cont. CTCFRAD21cv siRNA

I

Cohesin cleavage CTCF RNAi

J

0.4

0.3

0.2

0.1

0>50%

reduction>50%gain

Frac

tion

of In

tera

ctio

ns

p = 1.92x10e-5

p = 0.38

Cohesin-dependent genesAll Refseq genes

chr7: 50 kb27,150,000 27,200,000

HOXA1HOTAIRM1

HOXA2HOXA3

HOXA4HOXA-AS3

HOXA5

HOXA6

HOXA-AS3HOXA7

HOXA9

HOXA-AS4HOXA10

HOXA10

HOXA11

HOXA11-AS

HOXA13HOTTIP

_

__

_

2

2-2

-2

HR

VTE

VR

NA

-seq

-

+

+

-

TEVHRVDC

D

F

Scalechr6:

100 kb131,950,000 132,000,000 132,050,000 132,100,000

ARG1MED23

ENPP3

50 _

0 _0.5 _

0 _0 _

-0.5 _0.5 _

0 _0 _

-0.5 _

RN

A-s

eq

-

+

+

-

CTCF

H

CohesinCTCF

Normal

siC

onto

l si

CTC

F siCTCFvs siControl

RAD21cv/HRVvs RAD21cv/TEV

25

20

15

10

5

CTC

F/In

put

0

TSS-2.5kb +2.5kb

All genes

CTCF regulatedUp-regulatedDown-regulated

10

8

6

4

2

0

SM

C3/

Inpu

t

RAD21 regulatedAll genes

-2.5kb +2.5kbTSS

Fig. 4. Transcriptional changes after cohesin cleavage and CTCF depletion.(A) Changes in expression levels after RAD21 cleavage (Upper) and CTCFdepletion (Lower) (FDR < 0.05) are ranked from highest to lowest. Only veryfew genes behave similarly in both experiments. (B) Expression of HOXAgenes changes after RAD21 cleavage. Normalized RNA-seq read coverage isshown for RAD21cv/TEV and RAD21cv/HRV cells (+strand, purple; −strand,turquoise). HOX genes differentially expressed with FDR < 0.2 are marked inred. (C) qPCR confirmation of reduced HOXB-AS3 and HOXA-AS3 expressionafter RAD21 cleavage. CTCF depletion did not lead to a consistent reduction,as also seen in the analysis of the RNA-seq data (SI Appendix, Table S1).Transcription of the H19 noncoding RNA was reduced after CTCF depletionand also by RAD21 cleavage. (mean n = 3 ± SD). (D) Transcription of theENPP3 gene is increased after CTCF knockdown. Normalized RNA-seq readcoverage are shown for control siRNA and CTCF siRNA (+strand, purple; −strand,turquoise). CTCF binding sites are at the promoter and also intragenic. Theup-regulation was confirmed by RT-PCR to depend solely on CTCF knock-down (SI Appendix, Fig. S9F). (E) Position of CTCF sites analyzed relative totranscription start sites of all genes (black) and genes with altered expressionafter CTCF depletion (blue). Each line represents the average fold-enrich-ment of CTCF relative to input over a ±2.5-kb window surrounding thepromoters of genes bound by CTCF. CTCF is enriched at the transcriptionstart site of differentially expressed, in particular down-regulated genes(green), but it localizes more in the gene body at up-regulated genes. (F)Similar to E, except showing the fold-enrichment of SMC3 over input at thepromoter of genes altered after RAD21 cleavage. SMC3 does not appear tobe enriched at the promoter of the genes regulated by cohesin depletion.(G) Analysis of changes in interaction frequency between restriction frag-ments containing a promoter and restriction fragments containing a distalDNaseI hypersensitive site (DHS). Shown is the fraction of genes that displaya 50% reduction or 50% increase in interaction frequency after RAD21cleavage for either cohesin-regulated genes (orange) or all Ref-seq genes(black). Cohesin-regulated genes are enriched for a loss of interactions withrestriction fragments containing distal DHS sites relative to all Ref-seq genes(Fisher’s exact test). (H–J), Models describing the different changes of chro-mosomal interactions after cohesin cleavage (I) and CTCF depletion (J). (H)Cohesin and CTCF shape long-range interactions. (I) RAD21 cleavagedestroys cohesin and leads to reduced interactions within domains. CTCF

binding does not change and can still influence chromatin topology andmaintain domain identity. (J) CTCF depletion leads to more dynamic domainsand interactions across domain boundaries, normally prevented by CTCF’sinsulation function, potentially involving nonspecifically localizing cohesin.

1000 | www.pnas.org/cgi/doi/10.1073/pnas.1317788111 Zuin et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

neighboring domains. In light of the interaction between cohesinand CTCF and their colocalization, we propose that cohesin ismainly involved in chromatin interactions within topologicaldomains, whereas CTCF is important for their spatial segregation(Fig. 4 H–J). CTCF is bound stably to the chromatin (25) andlikely maintains boundaries by determining the localization ofcohesin. Without CTCF, cohesin would no longer localize prop-erly and, hence, form nonspecific interactions reaching beyondboundaries. This theory suggests that the removal of CTCF andcohesin would have different effects on gene expression, and in-deed there is very little overlap between genes affected by theremoval of CTCF or cohesin. Correlating these genes withcohesin and CTCF binding sites revealed that differentiallyexpressed genes after CTCF depletion tended to have CTCFsites close to the promoter. Genes differentially expressed aftercohesin cleavage do not correlate with cohesin; instead, a corre-lation with neighboring DNase hypersensitive sites marking po-tential regulatory elements suggests that these genes have lostcontact to neighboring enhancers. Several misregulated genesbelong to the developmentally important HOX clusters, whoseproperly timed and localized expression is regulated by topo-logical domain organization and remote enhancers (22, 23). In atleast two HOX clusters (HOXA, HOXB) we observed a loss ofinteractions after cohesin cleavage (Fig. 2G, and SI Appendix,Fig. S13). This finding shows that defects in cohesin or its reg-ulatory proteins could influence topological domain organizationof developmental genes, and hints how cohesin defects might belinked to the phenotype observed in patients diagnosed withCornelia de Lange syndrome, a “cohesinopathy” that is associ-ated with defects in multiple organ systems (26).The finding that the topological domain structure remains

largely intact after cohesin (and CTCF) depletion is surprising,considering the high density of cohesin occupancy at topologicaldomain boundaries, and the extensive colocalization betweenCTCF and cohesin. However, protein depletion is never complete,and the transfection efficiency is also never 100%. Therefore, the

observed effects might be weaker than under full depletion, whichcan only be addressed in the future by more sophisticated gene-ablation methods.However, our general finding is independent from the depletion

efficiency and another study by Seitan et al. (27, 28), using a dif-ferent experimental system where Rad21 is depleted rather slowlyover 10 d in mouse T-cells, also found that deletion of Rad21 failsto interrupt topological domain formation (27). Thus, although thecohesin complex is important for the chromosomal interactionswithin topological domains, it is not required for the formation ofthese domains. Taken together, our results provide an initial modelfor understanding the mechanisms of higher-order chromatin or-ganization and its relationship to gene expression.

Materials and MethodsHEK293T stable cell lines containing episomes coding for RAD21cv orRAD21wt and RAD21 siRNA were grown for 3 d in the presence of doxy-cycline until RAD21 was replaced by the engineered RAD21 versions, trans-fected with either control protease (TEV) or cleavage protease (HRV), andharvested after 24 h.

A detailed description of all methods can be found in the SI Appendix.

ACKNOWLEDGMENTS. We thank L. Schöckel and O. Stemmann for DNAconstructs; R. Stadhouders, E. Soler, and R.-J. Palstra for advice on 4C;R. van der Linden for FACS cell cycle analysis; L. Edsall and S. Kuan for assistancein next-generation DNA sequencing; A. de Klein and B. Eussen for providingcosmids; and N. Galjart and J.-M. Peters for antibodies. This work was sup-ported in part by the Netherlands Organization for Scientific Research viaALW2PJ/11029 (J.Z.) and the EpiGenSys project/ERASysBio+ initiative in theEU FP7 ERA-NET Plus program (F.G.G. and T.A.K.), the Erasmus Medical Cen-ter (K.S.W.), the Dutch Royal Academy (F.G.G.), the Netherlands Organiza-tion for Health Research and Development (E-RARE network TARGET-CdLS)(K.S.W.), the Netherlands Genomics Initiative (Zenith and Medical Epige-netics) (H.J.G.v.d.W. and F.G.G.) and the EU Systems Biology Consortium(SyBoSS) (F.G.G.). Work in the B.R. laboratory was supported by the LudwigInstitute for Cancer Research, the California Institute of Regenerative Med-icine (RN2-00905-1), and the National Institutes of Health.

1. Dixon JR, et al. (2012) Topological domains in mammalian genomes identified byanalysis of chromatin interactions. Nature 485(7398):376–380.

2. Phillips JE, Corces VG (2009) CTCF: Master weaver of the genome. Cell 137(7):1194–1211.

3. Wendt KS, et al. (2008) Cohesin mediates transcriptional insulation by CCCTC-bindingfactor. Nature 451(7180):796–801.

4. Parelho V, et al. (2008) Cohesins functionally associate with CTCF on mammalianchromosome arms. Cell 132(3):422–433.

5. Rubio ED, et al. (2008) CTCF physically links cohesin to chromatin. Proc Natl Acad SciUSA 105(24):8309–8314.

6. Splinter E, et al. (2006) CTCF mediates long-range chromatin looping and local histonemodification in the beta-globin locus. Genes Dev 20(17):2349–2354.

7. Kurukuti S, et al. (2006) CTCF binding at the H19 imprinting control region mediatesmaternally inherited higher-order chromatin conformation to restrict enhancer accessto Igf2. Proc Natl Acad Sci USA 103(28):10684–10689.

8. Nativio R, et al. (2009) Cohesin is required for higher-order chromatin conformationat the imprinted IGF2-H19 locus. PLoS Genet 5(11):e1000739.

9. Mishiro T, et al. (2009) Architectural roles of multiple chromatin insulators at thehuman apolipoprotein gene cluster. EMBO J 28(9):1234–1245.

10. Hadjur S, et al. (2009) Cohesins form chromosomal cis-interactions at the develop-mentally regulated IFNG locus. Nature 460(7253):410–413.

11. Hou C, Dale R, Dean A (2010) Cell type specificity of chromatin organization mediatedby CTCF and cohesin. Proc Natl Acad Sci USA 107(8):3651–3656.

12. Phillips-Cremins JE, et al. (2013) Architectural protein subclasses shape 3D organiza-tion of genomes during lineage commitment. Cell 153(6):1281–1295.

13. Kagey MH, et al. (2010) Mediator and cohesin connect gene expression and chro-matin architecture. Nature 467(7314):430–435.

14. Schöckel L, Möckel M, Mayer B, Boos D, Stemmann O (2011) Cleavage of cohesinrings coordinates the separation of centrioles and chromatids. Nat Cell Biol 13(8):966–972.

15. Uhlmann F, Lottspeich F, Nasmyth K (1999) Sister-chromatid separation at anaphaseonset is promoted by cleavage of the cohesin subunit Scc1. Nature 400(6739):37–42.

16. Gruber S, Haering CH, Nasmyth K (2003) Chromosomal cohesin forms a ring. Cell112(6):765–777.

17. Pauli A, et al. (2008) Cell-type-specific TEV protease cleavage reveals cohesin functionsin Drosophila neurons. Dev Cell 14(2):239–251.

18. Stadhouders R, et al. (2013) Multiplexed chromosome conformation capture se-quencing for rapid genome-scale high-resolution detection of long-range chromatininteractions. Nat Protoc 8(3):509–524.

19. Simonis M, Kooren J, de Laat W (2007) An evaluation of 3C-based methods to captureDNA interactions. Nat Methods 4(11):895–901.

20. Imakaev M, et al. (2012) Iterative correction of Hi-C data reveals hallmarks of chro-mosome organization. Nat Methods 9(10):999–1003.

21. van de Corput MP, et al. (2012) Super-resolution imaging reveals three-dimensionalfolding dynamics of the β-globin locus upon gene activation. J Cell Sci 125(Pt 19):4630–4639.

22. Alexander T, Nolte C, Krumlauf R (2009) Hox genes and segmentation of the hind-brain and axial skeleton. Annu Rev Cell Dev Biol 25:431–456.

23. Noordermeer D, et al. (2011) The dynamic architecture of Hox gene clusters. Science334(6053):222–225.

24. Kim YJ, Cecchini KR, Kim TH (2011) Conserved, developmentally regulated mechanismcouples chromosomal looping and heterochromatin barrier activity at the homeoboxgene A locus. Proc Natl Acad Sci USA 108(18):7391–7396.

25. Li Y, et al. (2013) Characterization of constitutive CTCF/cohesin loci: A possible role inestablishing topological domains in mammalian genomes. BMC Genomics 14(1):553.

26. Deardorff MA, Clark DM, Krantz ID (2005) Cornelia de Lange Syndrome. GeneReviewseds Pagon RA, et al. (Univ of Washington, Seattle). Available at http://www.ncbi.nlm.nih.gov/books/NBK1104/. Accessed August 30, 2013.

27. Seitan VC, et al. (2013) Cohesin-based chromatin interactions enable regulated geneexpression within preexisting architectural compartments. Genome Res, 10.1101/gr.161620.113.

28. Seitan VC, et al. (2011) A role for cohesin in T-cell-receptor rearrangement andthymocyte differentiation. Nature 476(7361):467–471.

Zuin et al. PNAS | January 21, 2014 | vol. 111 | no. 3 | 1001

CELL

BIOLO

GY

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020