A unique Oct4 interface is crucial for reprogramming to ... · Oct4 in reprogramming, we determined the structure of its POU domain bound to DNA. We show that the linker between the

L E T T ERS

A unique Oct4 interface is crucial for reprogrammingto pluripotencyDaniel Esch1,5, Juha Vahokoski2,5, Matthew R. Groves2, Vivian Pogenberg2, Vlad Cojocaru1, Hermann vom Bruch1,Dong Han1, Hannes C. A. Drexler1, Marcos J. Araúzo-Bravo1, Calista K. L. Ng3,4, Ralf Jauch3,Matthias Wilmanns2,6 and Hans R. Schöler1,6

Terminally differentiated cells can be reprogrammed topluripotency by the forced expression of Oct4, Sox2, Klf4 andc-Myc1,2. However, it remains unknown how this leads to themultitude of epigenetic changes observed during thereprogramming process. Interestingly, Oct4 is the only factorthat cannot be replaced by other members of the same familyto induce pluripotency3–5. To understand the unique role ofOct4 in reprogramming, we determined the structure of itsPOU domain bound to DNA. We show that the linker betweenthe two DNA-binding domains is structured as an α-helix andexposed to the protein’s surface, in contrast to the unstructuredlinker of Oct1. Point mutations in this α-helix alter or abolishthe reprogramming activity of Oct4, but do not affect its otherfundamental properties. On the basis of mass spectrometrystudies of the interactome of wild-type and mutant Oct4, wepropose that the linker functions as a protein–proteininteraction interface and plays a crucial role duringreprogramming by recruiting key epigenetic players to Oct4target genes. Thus, we provide molecular insights to explainhow Oct4 contributes to the reprogramming process.

Proteins of the POU (Pit1, Oct1/Oct2, UNC-86) family contain abipartite DNA-binding domain, comprising a POU-specific domain(POUS) with four α-helices and a POU homeodomain (POUHD) withthree α-helices. These subdomains are tethered by a linker region thatis not only hypervariable in both sequence and length but also exhibitsno homology to the different POU factors (Fig. 1a). Although theatomic structures of both POU subdomains of Oct1 complexed withDNA have been elucidated, the properties of the linker connecting thePOUS and POUHD remain elusive, as the linker structure has not beenvisible in previously solved structures6–9. In addition, the linker wasalso not revealed in the recently determined crystal structure of the

1Department for Cell and Developmental Biology, Max Planck Institute for Molecular Biomedicine, Röntgenstrasse 20, Münster D-48149, Germany. 2EMBL-Hamburg,Notkestrasse 85, Hamburg D-22603, Germany. 3Laboratory for Structural Biochemistry, Genome Institute of Singapore, 60 Biopolis Street, Singapore 138672,Singapore. 4School of Biological Sciences, Nanyang Technological University, Singapore 639798, Singapore. 5These authors contributed equally to this work.6Correspondence should be addressed to M.W. or H.R.S. (e-mail: [email protected] or [email protected])

Received 22 October 2012; accepted 17 December 2012; published online 3 February 2013; DOI: 10.1038/ncb2680

Oct6 complex10, although this linker is shorter than the Oct1 linkerand similar to the Oct4 linker (17 amino acids), suggesting that Oct6contains an unstructured linker, just like Oct1. Although POUS andPOUHD have been vigorously studied, the linker has received littleattention, but it has been shown to be important in the maintenanceof pluripotency11. POU proteins bind as monomers to the so-calledoctamer motif of DNA or form distinct homodimers based on thepseudo-palindromic PORE and the palindromic MORE sequence mo-tifs. They also heterodimerize with other POUmembers, as well as withmembers of the Sox family through the HMG (DNA-binding) domain(SOX2HMG). Oct4 is a POU family member specific to the mammaliangermline, encoding a transcription factor that is found in oocytes andis zygotically expressed in pluripotential cells of the pregastrulationembryo12. The protein is primarily expressed in pluripotential cells andin the germ cell lineage, and its function has been proved to be criticalfor both13. Most interestingly, Oct4 cannot be replaced by any otherPOU factor, as such a transcription factor cocktail does not lead to thesuccessful derivation of induced pluripotent stem (iPS) cells.Here we investigated the features that make Oct4 unique within

the POU family members and their role in the reprogrammingprocess. We have crystallized the POU domain of Oct4 (Oct4POU)on the PORE DNA-binding motif complex and determined thestructure of the complex at 2.8 Å resolution (Supplementary Table S1),similarly to the already known Oct1POU:PORE structure7. In contrastto previous structures6,7,10,14, most of the POUS–POUHD connectinglinker, except for residues 87–89, is visible in our structure (Fig. 1b,cand Supplementary Fig. S1a–c). Although the protein–DNA contactsare similar to those found in Oct1POU:DNA complexes (SupplementaryFig. S1d)6,7, the amino-terminal part of the linker is folded as an α-helix.This additional α-helix 5 (α5) interacts with helices α2 and α4 of POUS

(Fig. 1b,c) mostly by Van derWaals interactions, except for one specifichydrogen bond between Tyr 25 of the POUS and Gln 81 of the linker.

NATURE CELL BIOLOGY ADVANCE ONLINE PUBLICATION 1

© 2013 Macmillan Publishers Limited. All rights reserved.

mailto:[email protected]

mailto:[email protected]

http://www.nature.com/doifinder/10.1038/ncb2680

L E T T ERS

MEEED

1

Pit1a

b c

Oct1Oct2Oct6Oct4

Pit1

Oct1

Oct2

Oct6

Oct4

Pit1Oct1Oct2Oct6Oct4

SPPAK

PSSPA

EDDSL

ILLSQ

REEDK

EEEDE

EEEEE

QQQQQ

QQQQQ

TGGAA

NDDDD

AGGGG

AKKTV

VLLLL

HYYYF

GGGGG

40

SNNNK

EDDVV

FFFFF

SSSSS

SSSSS

QQQQQ

TTTTT

TTTTT

IIIII

CSSCC

RRRRR

FFFFF

50

FFFFL

KKKKK

KKKKK

WWWWW

70

LLLLV

ENNEE

EDDEE

AAATA

EEEDD

75

NNNNN

AMMMM

60

CCCCC

KKKKK

LLLLL

LLLLL

SEENE

ILLLL

KKKKR

APPPP

EEEEE

NAAAA

LLLLL

LLLLL

QNNQQ

VVVVV

30

VQQQQ

KKKKK

KKKKT

RRRRK

RRRRR

20

IIIII

FFFFF

AAAAA

NKRKK

ETTQL

FFFFL

LLLLL

LLLLL

ELLLL

AAAAT

LMMLL

TTTTT

GGGGG

GGGGG

YFFFY

DEEDM

Q

N

T

S

N

V

L

M

S

N

G

S

S

S

E

A

S

V

G

N

L

D

D

S

L

Y

S

S

P

Q

N

T

S

T

E

76

77

78

79

80

81

82

-

A

L

N

I

-

S

P

L

C

E

S

S

D

K

K

P

P

K

S

V

S

N

I

E

G

A

Q

A

T

A

L

L

A

L

N

N

S

Q

V

E

S

S

G

Q

-

P

P

-

-

-

G

S

-

-

-

L

L

-

-

-

G

G

-

-

-

A

F

-

-

-

E

D

-

-

-

G

G

-

-

-

L

L

-

-

-

N

P

-

-

-

-

G

-

-

15

26

27

17

17

RRRRA

93

RRRRK

RRRRR

TTTTT

TSSSS

IIIII

100

SEEEE

VTTVN

ANNGR

AIVVV

KRRKR

DVFGW

AAAAS

LLLLL

EEEEE

RKKST

110

HSSHM

FFFFF

GMLLL

EEAKK

HNNCC

SQQPP

KKKKK

PPPPP

STTSS

SSSAL

120

QEEHQ

EDEEQ

IIIII

MTLTT

RLLGH

MIILI

AAAAA

EEEDN

EQQSQ

LLLLL

130

NNHQG

LMMLL

EEEEE

KKKKK

EEEED

VVVVV

RRRRR

VVVVV

WWWWW

FFFFF

CCCCC

NNNNN

RRRRR

RRRRR

QQQQQ

RKKKK

KKKKK

RRRRR

VII

MS

KNNTS

152

EEEEG

VIIVV

KKKKR

RKKKK

KRRKR

α1

α1 α2 α3

α2

α5

α3 α4

α1 α1

α2 α1

α4

α5

α2

V71D75

N77

D73

W70

L37

T33

D29V36

L32 C84

K85V90

Q91A92

Q81 Y25L80

E82I83

S86

D72

K69

N76L23

E78

N79α3α4

α3

α2α5

α3

α2

α1

α5

α2

α3

α4

α1

Figure 1 The crystallographic structure of the Oct4POU:PORE complex.(a) Sequence alignment between the DNA-binding domains of differentPOU proteins. Strictly conserved residues are highlighted in yellow. Theα-helices defining the protein tertiary structure are indicated underneaththe alignment coloured in green for POUS, in red for the linker and inblue for POUHD. (b) Schematic representation of the Oct4POU homodimerbound to the PORE DNA. Oct4POU is coloured by domain (POUS in greenand POUHD in blue). The linker is highlighted in red. Residues that werenot visible in the electron density map are inferred with a dotted line.The DNA is shown in yellow. The α-helices are labelled according to the

corresponding number. (c) Magnified view of the linker and its neighbouringprotein residues. The residues with at least one atom within 7Å of anyatom in the linker are shown in a ball-and-stick representation colouredaccording to the type of residue (lysine in dark blue; aspartic acid andglutamic acid in red; asparagine in dark orange; glutamine in light orange;cysteine, serine and threonine in pink; valine, leucine and isoleucine inlight blue; tyrosine and tryptophan in purple). The hydrogen bonds areshown as black dotted lines. The residues that were not resolved in theelectron density map are shown as a dotted red line. This colouring ismaintained throughout this manuscript.

Val 36 of the POUS plays a prominent role by interfacing with thecarboxy-terminal part of the linker α5 (Supplementary Fig. S1b). Thetwo residues Asn 79 and Leu 80 are completely exposed to the surface inthe linker structure and are invariant among all known Oct4 sequencesbut are not conserved in other members of the Oct family (Fig. 1a),suggesting a distinct functional role in Oct4. The amino acid residuesof the helix point away from POUS, with the exception of Gln 81, whichis located in a microenvironment comprising residues in helix α2 ofPOUS (Supplementary Fig. S1b). Interestingly, the amino acids of helixα5 are also highly conserved in otherOct4 orthologues (Fig. 2a).The Oct4POU:PORE structure lacks electron density between residues

Cys 84 and Leu 89 (Fig. 1b,c and Supplementary Fig. S1c). To completeour model and to visualize the linker on other DNA motifs, we

built comparative models of Oct4POU bound as a homodimer orheterodimer with Sox2HMG to four distinct DNA elements and foundthat the residues of the linker, although situated in the proximity ofthe Oct4POU:Sox2HMG interaction interface, do not contribute to thisinteraction with the exception of Glu 82 and Lys 85, which may formtransient salt bridges with complementary residues in Sox2 (Supple-mentary Fig. S2). To investigate whetherα5 plays an important functionin reprogramming, we designed a set of three Oct4 variants (Fig. 2b).First, as Oct6 has already been shown to have no reprogrammingpotential4, we replaced the linker in Oct4 with the linker from Oct6.The other two mutants, V36K and Q81R, were designed to interferespecifically with the POUS-linker interface in α5, and their biologicalfunction was assessed in a reprogramming assay (Fig. 2c–e).

2 NATURE CELL BIOLOGY ADVANCE ONLINE PUBLICATION


L E T T ERS

b

e

f

c

d

POUS POUHD

C-terminalN-terminal

TAD TADLinker

1 15176 92

WT

Q81R

V36K

Oct6linker

Xenopus

Medaka

Zebrafish

SSSGSPTNLDKIAAQGR

NNENLQEICKSETLVQA

NNENLQEICKSETLVQA

NNENLQEIISRGQIIPQV

NSENPQDMYKIERVFVDT

TSENPQDMYKIERVFADT

NNENLREICKSETLVQA

...LGVLF...

...LGVLF...

...LGVLF...

...LGVLF...

...LGVLF...

...LGVLF...

...LGKLF...

MouseHuman

Xenopus

MedakaZebrafish

NN

N

TN

NN

N

SS

EE

E

EE

NN

N

NN

LL

L

PP

QQ

Q

QQ

EE

E

DD

II

I

MM

CC

I

YY

KK

S

KK

SA

R

II

EE

G

EE

TT

Q

RR

LL

I

VV

––

I

FF

VV

P

AV

QQ

Q

DD

AA

V

TT

a

WT

Q81R

V36K

7 days 14 days 20 daysMouse 34 191 218Xenopus 30 179 211

0 0 0Medaka

Zebrafish

Zebrafish

0 0 0

050

100150200250300350400450

Mouse Xenopus Medaka

20 days 14 days 7 days

7 days 14 days 20 daysWT 34 163 200

Oct6 linker 0 0 0V36K 6 94 145Q81R 0 0 0

050

100150200250

GFP

+ c

olon

ies

GFP

+ c

olon

ies

Oct

4 ex

pre

ssio

n

300350400

WT Oct6 linker V36K Q81R


0

200

400

600

800

1,000

1,200

1,400

1,600

WT Oct6 linker V36K Q81R

Oct6linker

Phase SSEA-1 APGFP

Figure 2 The effect of mutations in the Oct4 linker on reprogrammingactivity. (a) Sequence alignment of the Oct4 linker region from variousspecies. The conserved residues from the mammalian Oct4 that arealso conserved in at least one other species analysed are shown inyellow. (b) Representation of mouse Oct4 protein (WT) and differentgenetic modifications of the linker region. Amino acids that arereplaced are coloured in red. (c) The number of GFP-positive iPScell colonies formed by replacing the linker with the correspondingsequence from other species. The data from one of three independentreprogramming experiments are shown (for the complete data seeSupplementary Table S2). (d) In the left columns, iPS cell colonies

are shown under phase contrast and GFP fluorescence. In the rightcolumns, immunohistochemical staining of the iPS cell colonies forSSEA-1 and AP is shown. Scale bars, 75 µm. (e) The number ofGFP-positive iPS cell colonies formed at different time points duringreprogramming (7, 14 and 20 days) with proteins containing differentlinker mutations in comparison with mouse WT Oct4. The data fromone of two independent reprogramming experiments are shown (for thecomplete data, see Supplementary Table S2). (f) The relative transcriptlevels of the mutations observed with the qRT–PCR experiment. Themean values of three replicates started from the same reprogrammingexperiment are shown with the error bars representing the s.d.

The reprogramming efficiency of the Oct4 mutants was comparedwith that of the wild-type (WT) Oct4 protein in parallel fashion (seeSupplementary Methods). Using OG2 mouse embryonic fibroblast(MEF) cells that drive green fluorescent protein (GFP) expressionunder control of the Pou5f1 (Oct4) distal enhancer, reprogrammingefficiency was scored by counting the number of GFP-positivecolonies. Indeed, the Oct4 with the linker from Oct6 and theQ81R mutation led to a complete loss-of-function phenotype, asnot a single GFP-positive colony had formed within 20 days ofreprogramming (Fig. 2d,e). As viral titres play a vital role in thereprogramming process, we analysed the expression levels of the

viral Oct4 variants using quantitative PCR with reverse transcription(qRT–PCR; Fig. 2f). Viral transcript levels were found to be comparable,and thus they could not account for the loss-of-function phenotype.We characterized the mutants in terms of protein turnover, proteinlocalization, DNA binding and transactivation potential comparedwith WT Oct4 (Supplementary Fig. S3). The Oct4 with the linkerfrom Oct6 showed reduced DNA-binding activity to the W and POREmotifs, which is reflected by reduced transactivation activity, possiblyproviding an explanation for the observed loss-of-function phenotype.Interestingly, the Q81R mutation did not affect essential transcriptionfactor properties of the protein.



L E T T ERS

Oct4 DAPI

WT

76A

77A

78A

79A

80A

d

ba

c

e

fOct1

Oct6

Oct4

Oct

4 ex

pre

ssio

n

GFP

+ c

olon

ies

W sequence PORE sequence

OverlayPhase

Oct4

Tubulin

Oct4

Tubulin

Oct4

Tubulin

Oct4

Tubulin

Oct4

Tubulin

Oct4

Tubulin

WT

76A

77A

78A

79A

80A

0 8 16 24 30 96 (h)

Oct4

Oct6

Probe Probe

WT

76A

77A

78A

79A

80A

Neg Neg 80A

79A

78A

77A

76A

WT

050

100150200250300350400450

WT 76A 77A 78A 79A 80A 81A 82A


1

10

100

1,000

WT 76A 77A 78A 79A 80A 81A 82A

0

1

2

3

4

5

6

7

Rel

ativ

e lu

cife

rase

act

ivity

Oct4 mutant

WT 76A 77A 78A 79A 80A Neg

Figure 3 Characterization of alanine mutations on the linker segment. (a) Thenumber of GFP-positive iPS cell colonies formed for the correspondingalanine mutations on the linker segment was calculated. The data fromone of three independent reprogramming experiments are shown (for thecomplete data, see Supplementary Table S2). (b) The relative transcriptlevels for WT Oct4 and Oct4 with alanine mutations on the linker segmentwere assessed using qRT–PCR. The mean values from three replicatesstarted from the same reprogramming experiment are shown with the s.d. aserror bars. (c) Localization of the alanine mutations on the linker segmentsas shown by phase contrast and Oct4 immunostaining with counterstaining

(DAPI). Scale bars, 50 µm. (d) The in vivo stability of the Oct4 mutantsis compared with that of WT Oct4 by using cycloheximide inhibition oftranslation over 96h. (e) Oct4 alanine mutations were assessed for theability to bind to the W and PORE sequences. (f) The relative transcriptionalactivity of WT Oct4 and Oct4 with alanine mutations was measured usingan oligomerized octamer-containing oligonucleotide as an enhancer (6W;ref. 25). The mean values of three independent biological replicates startedfrom the same reprogramming experiment are shown with the error barsrepresenting the s.d. Uncropped versions of the electrophoretic mobilityimages are shown in Supplementary Fig. S5.

To confirm our findings and further map the amino acids importantto the reprogramming process, we replaced the mouse Oct4 linkerwith its Xenopus, zebrafish and medaka orthologues, and tested forreprogramming activity. Although the chimaeric protein with the

Xenopus linker was found to be functional, proteins with either themedaka and zebrafish linker sequences did not give rise to any iPScell colonies (Fig. 2c and Supplementary Fig. S4a and Table S2). Thisdata set confirmed the importance of the linker region in the biological



L E T T ERS

Chd4

Ctbp2

Hcfc1

Hdac1

Hells

Mta2

Mta3

Ogt

Smarca4

Smarcc1

Pardo

Ding

Berg

Ewsr1 Hdac1 Msh2 Prmt1

Rbm14 Ruvbl2 Set Smc1a

Actr3 Ahnak Ankrd17

Ap2m1 Atp1a1 Atxn10

Bzw2 Calu Cct2

Cct3 Cct4 Cct5

Cct7 Cnn3 Cnot1

Ctnnb1 Dbt Dnaja2

Eef2 Elavl1 Etf1

Flnb Fubp3 Gcn1l1

Hdgf Hk2 Hsp1

Hspg2 Ilf3 Lamb1

Mcm5 Nono Numa1

Paf1 Parp1 Ppp1cb

Ppp1cc Rfc4 Rpn1

Rpn2 Serpinh1 Snd1

Tcof1 Tcp1 Tpr1

Upf1

14

10

846

Acin1 Cul4b Ddb1 Hnrnpu Kpna2

Kpna3 Msh6 Nudc Pp2r1a Psmb6

Rpa1 Ssrp1 Supt16h Xrcc6

Common

Chd4

80A WT 80A WT

WT

80A WT 80A WT

60

70

80

90

100

110

120a c

b

d

80100120140160180200

Mta3

60

Q81L80 N77

N79E82

N76

E78

80

100

140

120

Msh6

80

100

120

140

160Msh2

50

150

100

250

200

300

Smarca4

80

100

120

140

Smarcc1

80A WT 80A

Pep

tide

coun

ts (%

)

Pep

tide

coun

ts (%

)

Pep

tide

coun

ts (%

)

Pep

tide

coun

ts (%

)

Pep

tide

coun

ts (%

)

Pep

tide

coun

ts (%

)

Figure 4 Surface view of the Oct4POU:PORE complex. (a) Thesolvent-accessible surfaces of the DNA-binding domains of Oct4 areshown in green for POUS, in blue for POUHD, and in red for the linker. TheDNA is shown in yellow ribbons. (b) A magnified view of the linker surfacerotated by 90◦ relative to that in a. The residues mutated to alanine arehighlighted. The colour scheme is the same as in a, except for the linkerresidues that are coloured according to the type of residue as in Fig. 1.(c) Boxplots of candidate proteins Chd4, Msh6 and Smarca4, which areunder-represented in the L80A pulldowns in comparison with the WT Oct4.

The whiskers correspond the minimum and maximum observed values of thepercentage of peptide counts normalized to the average peptide counts inthe case of the WT Oct4; the box spans from the first to the third quartile;the bold line in the box corresponds to the median. (d) Comparison ofour Oct4 interactome with the three published Oct4 network studies15–17.Overlapping proteins in all three published data sets are listed. Candidatesthat were under-represented in the L80A mutant and detected by all threestudies are highlighted in red, and candidates from only one published dataset are highlighted in orange.

NATURE CELL BIOLOGY ADVANCE ONLINE PUBLICATION 5© 2013 Macmillan Publishers Limited. All rights reserved.

L E T T ERS

Table 1 The intensity of the proteins that are commonly detected in all three replicates and the published interactome studies on Oct4.

Mt-154 Mt-406 Mt-432 Wt-154 Wt-406 Wt-432

Chd4 18,981,917 14,797,190 1,991,505 30,508,831 19,737,231 2,628,261Ctbp2 15,134,948 24,162,382 8,881,419 21,109,575 29,267,851 4,120,841Hcfc1 2,872,290 7,161,581 2,483,793 4,919,465 6,742,908 1,882,718Hdac1 23,159,511 35,204,275 33,317,271 21,012,796 26,978,942 15,619,726Hells 6,516,943 3,330,736 1,576,757 5,214,580 2,790,398 1,006,093Mta2 16,089,147 11,496,410 885,722 16,500,153 11,496,410 344,610Mta3 4,249,351 8,467,608 1,143,120 2,146,365 7,731,482 1,564,548Ogt 4,258,756 1,143,120 2,842,064 3.104,843 1,064,624 3,983,403Smarca4 6,033,749 11,952,742 48,944 6.631,674 11,084,689 183,747Smarcc1 20,729,606 42,283,859 2,508,807 29,998,777 33,317,271 1,740,677

function of Oct4 during reprogramming and provided a basis for adetailed mutagenesis screen. The sequence alignment of the linkerregions in the mouse, Xenopus, zebrafish and medaka revealed thatmouse and Xenopus linkers share only a few amino acids that are notconserved in zebrafish andmedaka (Fig. 2a).Next, we investigated whether the linker residues exposed to the

surface of the protein may be required for the biological activity ofOct4. We performed an alanine scan on the most exposed and highlyconserved linker segment, including residues of α5 (Asn 76–Glu 82,Figs 1a and 2a). The Oct4 mutations Q81A and E82A showed areprogramming efficiency comparable to that of the WTOct4 (Fig. 3a).In marked contrast, the point mutations N76A, N77A and N79Aled to the formation of significantly fewer iPS cell colonies (Fig. 3aand Supplementary Fig. S4a and Table S2). The strongest effect wasobtained with L80A, which abolished any iPS cell colony formationin the investigated time period. qRT–PCR analysis demonstratedthat viral overexpression of the mutant proteins was comparable tothat of the WT protein (Fig. 3b) and thus could not account for theloss-of-function phenotype. As expression levels of the mutants andthe WT protein were comparable, we examined the localization ofthe mutants by immunohistochemistry (Fig. 3c), protein turnover bycycloheximide-mediated inhibition of translation (Fig. 3d), sequence-specific DNA-binding activity (Fig. 3e), heterodimer formation withSox2 by electrophoretic mobility shift assay (EMSA) and modelling(Supplementary Fig. S2), and transactivation potential by the luciferaseassay (Fig. 3f). We could not find any significant differences incomparison to the WT protein. Examination of our models revealedthat these amino acids are entirely exposed to the surface of the protein(Fig. 4a,b). Thus, we deduced that mutating the L80A residue couldpotentially disturb an interaction surface with yet unknown additionalfactors. As the Oct4 L80A mutant led to complete loss-of-function inreprogramming, we used this variant for further investigations.To elucidate the molecular mechanism underlying the biological

function of this linker residue, we designed strep-tagged WT Oct4 andL80A Oct4 constructs. Nuclear extracts were prepared in triplicates,and protein complexes were analysed by mass spectrometry to enablelabel-free quantification of the interacting proteins.The role of the Oct4 interactome in embryonic stem cells has

been assessed by three independent studies15–17, which have describednumerous transcription factors and major remodelling complexesto be direct interaction partners of Oct4. Here we compared theinteractome of the L80A mutant with that of the WT protein bylabel-free mass spectrometry in a quantitative manner. The massspectrometry analysis revealed that the WT and L80A constructs

pulled down most of the interacting proteins with comparableaffinity (Supplementary Table S3). Within the intersection of thethree published data sets, only two proteins exhibited a significantlyreduced intensity in the mutant interactome (Fig. 4c,d, Table 1 andSupplementary Table S4). The first was Smarca4, a helicase of the BAFremodelling complex previously shown to improve reprogrammingefficiency using the iPS cell methodology18. Furthermore, Smarca4 isknown to interact with key epigenetic repressors and activators19, andwe propose that it is recruited by Oct4 to downstream target genesto enforce euchromatin/heterochromatin transitions of Oct4 targetgenes during reprogramming. It is worth noting that we detectedexclusively esBAF components and not pBAF components in ourinteractome study, highlighting the selectivity of this interaction andits importance for the induction of pluripotency. The second proteinwas Chd4, a helicase of the NuRD complex, which also is known tosafeguard pluripotency by maintaining the bivalent mark (H3K273m)in embryonic stem cells20. In the mutant pulldown, the amount ofChd4 was found to be significantly reduced (Fig. 4c,d). When onlyone published data set was used for filtering, Msh6 was found to beanother potential interaction candidate. Msh6 recognizes T/G DNAmismatches and could be involved in the DNA nucleotide excisionprocess after deamination of methylcytosines by AID and APOBEC(refs 21,22). However, Msh6−/− embryonic stem cells have beenshown to give rise to viable offspring. In contrast to the quadruplemutant (76A, 77A, 79A, 80A), the L80A mutant exhibited only apartial loss-of-function phenotype in the rescue experiments withTC4 cells23 (Supplementary Fig. S4b). This observation may explainwhy only a mild reduction in intensity was observed. The rescueexperiment also confirms that the integrity of the linker and theidentified amino acids are important not only for reprogramming,but also to maintain the pluripotent state.Here we show that a unique property of the linker distinguishes Oct4

from other members of the POU family and that the integrity of thelinker is essential for successful reprogramming. Our data also stronglyindicate that the linker plays an important role in reprogramming byrecruiting key epigenetic players to sites occupied by Oct4, suggestingthat Oct4 may serve as a recruiting platform during the epigenetictransition from a differentiated to a pluripotent cell state. A moredetailed analysis of the orchestration of the series of events mediatingthis transition and the interaction of the epigenetic players withthe linker interface will provide further insights into the molecularprocesses underlying reprogramming and may prompt futurestudies to direct induced transdifferentiation towards different celllineages24. �



L E T T ERS

METHODSMethods and any associated references are available in the onlineversion of the paper.

Note: Supplementary Information is available in the online version of the paper

ACKNOWLEDGEMENTSX-ray diffraction experiments were performed on the ID23-1 beamline at theEuropean Synchrotron Radiation Facility (ESRF), Grenoble, and X12 and X13at EMBL Hamburg, DESY. We are grateful to E. Fioravanti of the ESRF forproviding assistance in using beamline ID23-1. We thank G. Bourenkov forinvaluable assistance in crystallographic problem resolution and data collection.We thank A. Nolte for help with mass spectrometry measurements. We alsothank R. Grindberg for critically reading, and A. Malapetsas and J. Bruder forediting the manuscript. We acknowledge the financial support from DFG (SPP1356Pluripotency and Cellular Reprogramming priority program) AJ DFG SI 1695/1-2and AJ DFG CO 975/1.

AUTHOR CONTRIBUTIONSD.E. and J.V. performed the experiments and wrote the manuscript; J.V., M.R.G.and V.P. collected and analysed the crystallographic data; V.C. performed thecomparative modelling experiments and wrote the manuscript; H.v.B. performedthe cycloheximide experiments; H.C.A.D. performed the MS experiments; M.J.A-B.analysed the proteomic data sets; D.H. performed the rescue experiments; andC.N. assessed the heterodimer formation ability of Oct4 and Sox2. All authorscommented on the manuscript; and R.J., M.W. and H.R.S. as co-senior authorssupervised the project and assisted in writing the manuscript.

COMPETING FINANCIAL INTERESTSThe authors declare no competing financial interests.

Published online at www.nature.com/doifinder/10.1038/ncb2680Reprints and permissions information is available online at www.nature.com/reprints

1. Takahashi, K. & Yamanaka, S. Induction of pluripotent stem cells frommouse embryonic and adult fibroblast cultures by defined factors. Cell 126,663–676 (2006).

2. Yu, J. et al. Induced pluripotent stem cell lines derived from human somatic cells.Science 318, 1917–1920 (2007).

3. Shi, Y. et al. Induction of pluripotent stem cells from mouse embryonic fibroblasts byOct4 and Klf4 with small-molecule compounds. Cell Stem Cell 3, 568–574 (2008).

4. Nakagawa, M. et al. Generation of induced pluripotent stem cells without Myc frommouse and human fibroblasts. Nat. Biotechnol. 26, 101–106 (2008).

5. Feng, B. et al. Reprogramming of fibroblasts into induced pluripotent stem cells withorphan nuclear receptor Esrrb. Nat. Cell Biol. 11, 197–203 (2009).

6. Klemm, J. D., Rould, M. A., Aurora, R., Herr, W. & Pabo, C. O. Crystal structure ofthe Oct-1 POU domain bound to an octamer site: DNA recognition with tetheredDNA-binding modules. Cell 77, 21–32 (1994).

7. Reményi, A. et al. Differential dimer activities of the transcription factor Oct-1 byDNA-induced interface swapping. Mol. Cell 8, 569–580 (2001).

8. Reményi, A. et al. Crystal structure of a POU/HMG/DNA ternary complex suggestsdifferential assembly of Oct4 and Sox2 on two enhancers. Genes Dev. 17,2048–2059 (2003).

9. Williams, D. C. Jr, Cai, M. & Clore, G. M. Molecular basis for synergistictranscriptional activation by Oct1 and Sox2 revealed from the solution structureof the 42-kDa Oct1.Sox2.Hoxb1–DNA ternary transcription factor complex. J. Biol.Chem. 279, 1449–1457 (2004).

10. Jauch, R., Choo, S. H., Ng, C. K. & Kolatkar, P. R. Crystal structure of the dimericOct6 (POU3f1) POU domain bound to palindromic MORE DNA. Proteins 79,674–677 (2011).

11. Nishimoto, M. et al. Oct-3/4 maintains the proliferative embryonic stem cell statevia specific binding to a variant octamer sequence in the regulatory region of theUTF1 locus. Mol. Cell Biol. 25, 5084–5094 (2005).

12. Pesce, M., Gross, M. K. & Schöler, H. R. In line with our ancestors: Oct-4 and themammalian germ. Bioessays 20, 722–732 (1998).

13. Kehler, J. et al. Oct4 is required for primordial germ cell survival. EMBO Rep. 5,1078–1083 (2004).

14. Klemm, J. D. & Pabo, C. O. Oct-1 POU domain-DNA interactions: cooperativebinding of isolated subdomains and effects of covalent linkage. Genes Dev. 10,27–36 (1996).

15. Van den Berg, D. L. et al. An Oct4-centered protein interaction network in embryonicstem cells. Cell Stem. Cell 6, 369–381 (2010).

16. Pardo, M. et al. An expanded Oct4 interaction network: implications for stem cellbiology, development, and disease. Cell Stem Cell 6, 382–395 (2010).

17. Ding, J., Xu, H., Faiola, F., Ma’ayan, A. & Wang, J. Oct4 links multiple epigeneticpathways to the pluripotency network. Cell Res. 22, 155–167 (2012).

18. Singhal, N. et al. Chromatin-remodeling components of the BAF complex facilitatereprogramming. Cell 141, 943–955 (2010).

19. Trotter, K. W. & Archer, T. K. The BRG1 transcriptional coregulator. Nucl. Recept.Signal. 6, e004 (2008).

20. Hu, G. & Wade, P. A. NuRD and pluripotency: a complex balancing act. Cell StemCell 10, 497–503 (2012).

21. Rai, K. et al. DNA demethylation in zebrafish involves the coupling of a deaminase,a glycosylase, and gadd45. Cell 135, 1201–1212 (2008).

22. Bhutani, N. et al. Reprogramming towards pluripotency requires AID-dependent DNAdemethylation. Nature 463, 1042–1047 (2010).

23. Niwa, H., Miyazaki, J. & Smith, A. G. Quantitative expression of Oct-3/4 definesdifferentiation, dedifferentiation or self-renewal of ES cells. Nat. Genet. 24,372–376 (2000).

24. Adachi, K. & Schöler, H. R. Directing reprogramming to pluripotency by transcriptionfactors. Curr. Opin. Genet. Dev. 22, 416–422 (2012).

25. Schöler, H. R., Balling, R., Hatzopoulos, A. K., Suzuki, N. & Gruss, P. Octamerbinding proteins confer transcriptional activity in early mouse embryogenesis. EMBOJ. 8, 2551–2557 (1989).







http://www.nature.com/reprints

METHODS DOI: 10.1038/ncb2680

METHODSPurification and assembly of the Oct4POU:PORE complex. The POU domainof mouse Oct4 (Oct4POU) was cloned previously7 (mouse Pou5f1 residues131–282) and overexpressed in the bacterial strain BL21 (DE3) RIL (Novagen)in ZYM-5052 at 20 ◦C. The pellet was resuspended (50mM HEPES at pH 7.5,250mM NaCl and 0.01% thioglycerol), lysed by sonication and centrifuged(18,000g ). The remaining pellet was resuspended (50mM HEPES at pH 7.5,6M Gu–HCl and 20mM imidazole at pH 8.0), and the denatured fraction wascentrifuged (18,000g ). The Oct4POU protein was purified using immobilizedmetal affinity chromatogram (Ni–NTA, QIAGEN) and eluted (50mM HEPESat pH 7.5, 6M Gu–HCl and 300mM imidazole at pH 8.0). The Oct4POU:PORE(5′-TCACATTTGAAAGGCAAATGGA-3′; ref. 7) complex was formed and refoldedin a one-step process, dialysing all components (15mg Oct4POU, 1mg TEV proteaseand 200 nmol DNA in 5–8ml total volume) in buffer (50mM HEPES at pH 7.5,150mM NaCl and 5mM dithiothreitol) at 4 ◦C overnight and purifying by sizeexclusion chromatography (16/60 Superdex 75 column, GE Healthcare; 50mMHEPES at pH 7.5, 150mM NaCl and 1mM dithiothreitol). Fractions containingthe complex (verified by EMSA and SDS–PAGE) were pooled and concentratedby ultrafiltration after incubation for 15–30min at room temperature (relativemolecular mass cutoff 10,000; Millipore).

Crystallization and structure determination. The Oct4POU:PORE complex wascrystallized by vapour diffusion in the presence of 320mM Na/K phosphate (pH5.7), using the EMBL Hamburg HTP screening facility26. The crystal was collectedone day after setting up the drops and transferred into a solution with glycerol,which was added stepwise to final concentrations of 8, 20 and 38%. An X-raydata set covering 90◦ was collected at beamline ID23-1 at the ESRF Grenoble. Thedata were integrated and reduced with XDS and scaled using XSCALE (ref. 27).The P41 space group with a (h,−k,−l) twinning operator (twinning fraction 0.29)was determined after analysing the reflections using PHENIX.XTRIAGE (ref. 28).The structure was solved by molecular replacement using PHASER (ref. 29) withthe Oct1POU:PORE complex (1HF0) as the template7. One dimeric Oct4POU:DNAcomplex per asymmetric unit was found. The refinement was performed stepwisein REFMAC 5.6 (ref. 30) using restrained non-crystallography symmetry operatorsandmanualmodel building in COOT (ref. 31). Ramachandran plots were calculatedusing MOLPROBITY (ref. 32; 93.7% of residues were in the most favoured regions,and 5.2% in the allowed regions). Structure determination details are given inSupplementary Table S1.

EMSA. Proteins were overexpressed in BL21(DE3) bacteria and purified usingimmobilized metal affinity chromatography followed by ion-exchange chromatog-raphy and gel filtration. Whole-cell extracts were prepared as described previously33.Cells (2.2×106) were scraped in lysis buffer (20mMHEPES, 150mMNaCl, 0.2mMEDTA at pH 8.0 and 25% glycerol). DNA fragments were labelled with [γ -32P] ATPusing polynucleotide kinase. The protein sequences are listed in SupplementaryTable S5, DNA sequences in Supplementary Table S6 and primers in SupplementaryTable S7.

Luciferase assay. The transactivation potential was tested with the Dual-GloLuciferase Assay System (Promega) by measuring luciferase activity after 48 h.HEK293T cells (4× 105) were transfected with 100 ng of effector DNA, 100 ngof pTK-RL (Renilla luciferase) and 800 ng of reporter constructs (6W 37tk-Luc;refs 25,34).

qRT–PCR analysis. Total RNA was extracted using the RNeasy Mini Kit(QIAGEN). Complementary DNAs were synthesized with the High Capacity cDNAArchive Kit (Applied Biosystems). Transcript levels were determined using the ABIPRISM Sequence Detection System 7900. Gene expression was normalized to thehousekeeping gene Hprt1.

Localization of Oct4. NIH3T3 cells were infected with pMX virus and collectedafter 24 h. Cells were fixed in 4% paraformaldehyde (PFA) for 10min and quenchedwith 50mM glycine in PBS. The cell membrane was permeated with Triton X-100.Nuclear localization was probed with Oct4 antibody (Santa Cruz, sc-8628, 1:200).Nuclei were counterstained with Hoechst (Sigma-Aldrich H2261).

SSEA-1 and AP staining. Cells were fixed with 4% PFA for stage-specificembryonic antigen 1 (SSEA-1) staining with SSEA-1 antibody (DevelopmentalStudiesHybridomaBank,MC-480, 1:50).Naphtol and fast redwere used for alkalinephosphatase (AP) staining (15min in dark).

Cell culture. OG2 MEF, HEK293T, HEK293 and NIH3T3 cells were maintainedin low-glucose DMEM containing 10% FBS (Biowest), 1× PSG, 1× non-essential

amino acids (NEAA; PAA) and 1×β-mercaptoethanol (Invitrogen). OG2 MEFswere isolated as described previously1. Mouse embryonic stem cell and iPS cellcultures were maintained in DMEM containing 15% knockout serum replacement(Invitrogen), 5% FBS, 1× PSG, 1× NEAA, and 1 × β-mercaptoethanol, with1000 units of leukaemia inhibitory factor (LIF) on feeder layers of gamma-irradiated MEFs or 2,000 units of LIF under feeder-free conditions, as previouslydescribed35.

Virus production. HEK293T cells were seeded (density of 2.2× 106 cells per100-mm dish). The following day, cells were transfected with pMX-based retroviralvectors using Fugene 6 (Roche). Cells were incubated overnight at 37 ◦C with 5%CO2. The medium was replaced 24 h after transfection with 6ml fresh medium.The virus-containing supernatant was collected and filtered (0.45 µm, Millex-HV,Millipore) 48 h after infection. The supernatant was supplemented with 6 µgml−1

protamine sulphate (Sigma-Aldrich) before infection. Viral stocks were generatedsimultaneously to ensure equivalent virus production among different experiments.

Reprogrammingand rescue assay. MEFswere seeded in a gelatinized 6-well plate(density 5 ×104 cells per well). Simultaneously, viral supernatant was added on theOG2 MEF cells. Viral expression was analysed by qRT–PCR. The supernatant wasremoved and cells were washed twice with PBS 24 h after infection. The mediumwas replaced with embryonic stem cell medium (high-glucose DMEM with LIF)72 h after infection. Cells were maintained in culture until GFP-positive coloniesappeared. The number of colonies obtained with the WT and with each mutantOct4 were compared after 7, 14 and 20 days. For the rescue assay, the proteins wereoverexpressed in TC4 cells. The morphology of the colonies was assessed after 96 hof doxycycline treatment23.

Strep-tagged purification of proteins. SF lentivirus encoding strep-tagged WTOct4 and the mutants was produced in HEK293T cells by co-transfection withpCL-Eco. SF lentivirus tomato was used as a negative control. The virus-containingsupernatant was filtered. OG2 embryonic stem cells were infected. Cells (50×106)were collected, and extracts were applied to Strep-Tag/Strep-Tactin affinity columns.Pulldown quality was assessed by western blot analysis for Oct4 using SC-9081antibody (Santa Cruz).

Mass spectrometry. Three independent experiments were analysed by massspectrometry (MS) for label-free quantification. Proteins from the strep-taggedpulldown were processed for liquid chromatography (LC)–MS/MS analysis with thefilter-aided sample preparation method as described previously36. Proteins elutedfrom the Strep-Tactin beads with desthiobiotin were reduced by dithiothreitol(0.1M; 45min at 56 ◦C) followed by alkylation for 20min in the dark at roomtemperature using 55mM iodoacetamide. After exchanging the urea buffer to50mM NH4HCO3, proteins were digested with trypsin (2.5 µg per sample) byincubation overnight at 37 ◦C. Peptides were collected by centrifugation, desaltedusing STAGE-Tips37 and analysed by LC-MS/MSwith anEasy-nLCnanoflow system(Proxeon Biosystems) that was online-coupled to an LTQ Orbitrap Velos massspectrometer (Thermo Scientific) through a nanoelectrospray source (ProxeonBiosystems) and a fused-silica capillary emitter column filled with C18 reversed-phasematerial (Dr.Maisch, ReproSil-Pur 120, C18-AQ, 3 µm). Peptides were loadedat the maximum flow rate at 200 bar and separated at 250 nlmin−1 running a lineargradient from 7 to 35% Buffer B in 90min, 35–60% Buffer B in 20min and 60–98%Buffer B in 6min (Buffer A: 0.5% acetic acid; Buffer B: 80% acetonitrile and 0.5%acetic acid).

The mass spectrometer was operated in the positive-ion mode (spray voltage2.2 kV, heated capillary maintained at 225 ◦C) using Xcalibur software, automati-cally switching to a data-dependent mode between survey scans in the mass rangeof m/z 300–1650 and MS/MS acquisition. Collision-induced MS/MS spectra fromthe 15 most intense ion peaks in the MS were collected (target value of the Orbitrapsurvey scan was 1,000,000, resolution 60,000, and the lock mass 445.12). Precursorion charge state screening was enabled, excluding unassigned charge states as wellas singly charged species. Dynamic exclusion was activated, allowing maximum 500entries and a retention period of 180 s.

Raw data were processed with MaxQuant software (v. 1.0.13.13) and the Mascotdatabase. Data were searched against the International Protein Index sequencedatabase (mouse IPI, v. 3.60) concatenated with reversed-sequence versions of allentries. The parameters were: minimum length of 6 amino acids, maximum of 2missed cleavages, fixed carbamidomethylation of cysteines, variable oxidation ofmethionines and acetylation at the N termini. Maximum allowed mass deviationwas 7 ppm for MS and 0.5Da for MS/MS scans. Proteins were considered identifiedif there were at least two matching peptides, with one unique to the protein.The false discovery rate (FDR) was 1% for both the peptide and the proteinidentifications.

NATURE CELL BIOLOGY


DOI: 10.1038/ncb2680 METHODS

Modelling Oct4–DNA complexes. We built 100 comparative models in MOD-ELLER (ref. 38) for each of the four Oct4POU:Oct4POU:DNA or Oct4POU:Sox2HMG:DNA complexes formed on the following DNA elements: PORE(templates: Oct4POU:PORE (3L1P), Oct1POU:PORE (1HFO; ref. 7), Oct1POU:Sox2HMG:FGF4 (1GT0; ref. 8)); MORE (templates: Oct1POU:MORE (1E3O; ref. 7),Oct6POU:MORE (2XSD; ref. 10), 3L1P); FGF4 (templates: 3L1P, 1GT0 and Oct1POU:Sox2HMG:HOXB1 (1O4X; ref. 9)); and HOXB1 (templates: 3L1P, 1O4X, 1GT0).We refined the loop between residues Cys 84 and Leu 89 and generated 25 loopmodels38,39 for each model. The refinement procedure included energy minimiza-tion and molecular dynamics simulations. We selected the 25 best-scoring models(using the normalized DOPE score) and clustered them according to the root meansquare deviation of the linker.

Primary accession codes. Protein Data Bank (the Oct4POU:PORE structure):3L1P.

Proteome Commons Database (the WT and mutant Oct4 interactome data):T7mU/n8vauq1d4M3P8ZEnECJx7AV3CqyDygo7RXFfAmRAC/KV6Xf3JZYU04Fn2NlBngbErGvevwXkY6eNvv8tnVmEJkAAAAAAAAglw==

Reference accession codes. Uniprot (Swissprot): P20263 (mouse Pou5f1 se-quence).

26. Mueller-Dieckmann, J. The open-access high-throughput crystallization facility atEMBL Hamburg. Acta Crystallogr. D 62, 1446–1452 (2006).

27. Kabsch, W. Evaluation of single-crystal X-ray diffraction data from a position-sensitive detector. J. Appl. Crystallogr. 21, 916–924 (1988).

28. Adams, P. D. et al. PHENIX: a comprehensive Python-based system formacromolecular structure solution. Acta Crystallogr. D 66, 213–221 (2010).

29. McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Crystallogr. 40,658–674 (2007).

30. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refinement of macromolecularstructures by the maximum-likelihood method. Acta Crystallogr. D 53,240–255 (1997).

31. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development ofCoot. Acta Crystallogr. D 66, 486–501 (2010).

32. Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecularcrystallography. Acta Crystallogr. D 66, 12–21 (2010).

33. Sauter, P. & Matthias, P. Coactivator OBF-1 makes selective contacts with both thePOU-specific domain and the POU homeodomain and acts as a molecular clamp onDNA. Mol. Cell Biol. 18, 7397–7409 (1998).

34. Tomilin, A. et al. Synergism with the coactivator OBF-1 (OCA-B, BOB-1) is mediatedby a specific POU dimer configuration. Cell 103, 853–864 (2000).

35. Kim, J. B. et al. Pluripotent stem cells induced from adult neural stem cells byreprogramming with two factors. Nature 454, 646–650 (2008).

36. Wisniewski, J. R., Zougman, A., Nagaraj, N. & Mann, M. Universal samplepreparation method for proteome analysis. Nat. Methods 6, 359–362 (2009).

37. Rappsilber, J., Ishihama, Y. & Mann, M. Stop and go extraction tips formatrix-assisted laser desorption/ionization, nanoelectrospray, and LC/MS samplepretreatment in proteomics. Anal. Chem. 75, 663–670 (2003).

38. Sali, A. & Blundell, T. L. Comparative protein modelling by satisfaction of spatialrestraints. J. Mol. Biol. 234, 779–815 (1993).

39. Fiser, A., Do, R. K. & Sali, A. Modeling of loops in protein structures. Protein Sci. 9,1753–1773 (2000).

NATURE CELL BIOLOGY


http://www.rcsb.org/pdb/explore/explore.do?structureId=3L1P

https://proteomecommons.org/dataset.jsp?i=78271

S U P P L E M E N TA RY I N F O R M AT I O N

WWW.NATURE.COM/NATURECELLBIOLOGY 1

DOI: 10.1038/ncb2680

Esch et al. - A unique Oct4 interface is crucial for reprogramming to pluripotency

Supplementary Figure 1

4

A unique Oct4 interface is crucial for reprogramming to pluripotencyA unique Oct4 interface is crucial for reprogramming to pluripotency

Figure S1 Details of the Oct4POU:PORE complex structure (see complementary information in Figure 1). (a) Superposition of the Oct4POU:PORE and Oct1POU:PORE complex structures. Oct4POU and the DNA are coloured as in main Figure 1. Oct1POU is shown in grey. The missing residues in the linkers of Oct1POU and Oct4POU are inferred with dotted lines (grey for Oct1POU; red for Oct4POU). (b) Surface view highlighting the cavity of the POUS domain in which Q81 is docked. (c) Stereo images of the the electron density map of residues 70 to 86 in both

Oct4POU protomers in two different orientations. The type of map is 2Fo-Fc and the contour level has been set to 1.0 . (d) Interactions between Oct4POU and the PORE motif. In the figure, a NUCPLOT diagram of the crystal structure of Oct4POU:PORE is presented. After the interacting residue, the corresponding chain identifier is indicated (A or B). The amino and hydroxy groups forming a hydrogen bond are shown in blue and red upper case letters, respectively. Residues in the POUS domain surrounding this cavity are labelled.


http://www.nature.com/doifinder/10.1038/ncbxxxx


2 WWW.NATURE.COM/NATURECELLBIOLOGY



5

A unique Oct4 interface is crucial for reprogramming to pluripotencyA unique Oct4 interface is crucial for reprogramming to pluripotency

Figure S2 Models of Oct4POU bound to four DNA elements. Each panel shows a view of a complete model and a zoomed-in surface view of the linker region. The colouring is as in Figure 1. The Sox2HMG is shown in pink. (a), (b) Oct4POU:PORE complex (homodimer). (a) Model in which the residues E87, T88, and L89, missing in the crystal structure, are exposed to the surface. (b) Model in which K85 is oriented towards POUS, establishing a salt bridge with D29. (c), (d) Oct4POU:MORE complex (homodimer). (c) Model in which E87, T88, V90 are exposed to the surface. (d) Model in which E87 is oriented toward the interface between the POUS and the N-terminal tail of the POUHD domain. (e), (f)

Oct4POU:Sox2HMG:FGF4 ternary complex. (e) Model in which K85 and E87 of the Oct4 linker establish additional interaction with residues from the C-terminal tail of Sox2HMG. (f) Model in which K85 forms a hydrogen bond with D29 of the POUS domain, while E87 is exposed to the surface. (g), (h) Oct4POU:Sox2HMG:HOXB1 (UTF1-like) ternary complex. (g) Model with the highlighted residues oriented as in (e). (h) Model with the highlighted residues oriented as in (f). (i) Electrophoretic mobility shift assay (EMSA) using the purified Oct4POU and Sox2HMG on the Nanog probe. The L80A mutant is able to form a heterodimer with Sox2, comparable to the WT Oct4:Sox2 heterodimer.




Esch et al. - A unique Oct4 interface


6

A unique Oct4 interface is crucial for reprogramming to pluripotencyis crucial for reprogramming to pluripotency

Figure S3 Characterization of the initial mutations of the linker segment (see complementary information in Figure 2). (a) The cellular distribution of the mutant proteins is analyzed by phase contrast (Phase) and Oct4 immunostainings (Oct4) with DAPI counterstaining (DAPI). Scale bar is 50 μm (b) The in vivo stability of Oct4 mutants is compared with that of WT Oct4 by using cycloheximide inhibition of translation over 96 hours. (c) Oct4 alanine mutations were tested for the ability to bind to the

proximal octamer enhancer (6W) sequence, by using bacterially expressed purified protein and crude cellular extracts from 293T cells. (d) The relative transcriptional activity of WT Oct4 was compared with that of the mutants by the luciferase assay using 6W. The mean values of three replicates started from different cell extracts from the same reprogramming experiment are shown with the error bars representing the standard deviation.



4 WWW.NATURE.COM/NATURECELLBIOLOGY

7



Figure S4 Characterization of the Oct4 mutants from the alanine scan and the Oct4 orthologues (see complementary information in Figures 2 and 3) (a) Fluorescence stereolupe images of (green fluorescent protein-positive [GFP+] and stagespecific embryonic antigen 1–positive [SSEA-1+]) colonies during reprogramming for proteins with orthologue linker exchanges and from the alanine scan. Scale bar is 5 mm (b) Rescue experiments with TC4

cells were performed as described23. In contrast to the reprogramming experiments, here the L80A mutant had a rescue index of 0.2, while the quadrupole mutant showed a complete loss-of-function phenotype. Scale bar is 100 μm. The mean values of three independent experiments (biological replicates) are shown with the error bars representing the standard deviation.






8

A unique Oct4 interface is crucial for reprogramming to pluripotency

A unique Oct4 interface is crucial for reprogramming to pluripotency

Figure S5 Uncropped electrophoretic mobility images showed in Figures 2 and 3.


Documents

A unique Oct4 interface is crucial for reprogramming to ... · Oct4 in reprogramming, we determined the structure of its POU domain bound to DNA. We show that the linker between the