A Protein Interaction Network for Pluripotency Of

8/6/2019 A Protein Interaction Network for Pluripotency Of

http://slidepdf.com/reader/full/a-protein-interaction-network-for-pluripotency-of 1/5

LETTERS

A protein interaction network for pluripotency of

embryonic stem cellsJianlong Wang1, Sridhar Rao1, Jianlin Chu1, Xiaohua Shen1, Dana N. Levasseur1, Thorold W. Theunissen1

& Stuart H. Orkin1,2

Embryonic stem (ES) cells are pluripotent1,2 and of therapeuticpotential in regenerative medicine3,4. Understanding pluripotency at the molecular level shouldilluminatefundamental propertiesof stem cells and theprocess of cellular reprogramming. Through cellfusion the embryonic cell phenotype can be imposed on somaticcells, a process promoted by the homeodomain protein Nanog 5,

which is central to the maintenance of ES cell pluripotency 6,7.Nanog is thought to function in concert with other factors suchas Oct4 (ref. 8) and Sox2 (ref. 9) to establish ES cell identity. Here we explore the protein network in which Nanog operates in mouseES cells. Using affinity purification of Nanog under native condi-tions followed by mass spectrometry, we have identified physically associated proteins. In an iterative fashion we also identifiedpartners of several Nanog-associated proteins (including Oct4), validated the functional relevance of selected newly identifiedcomponents and constructed a protein interaction network. Thenetwork is highly enriched fornuclear factors that are individually critical for maintenance of the ES cell state and co-regulated ondifferentiation. The network is linked to multiple co-repressorpathways and is composed of numerous proteins whose encoding

genes are putative direct transcriptional targets of its members.This tight protein network seems to function as a cellular modulededicated to pluripotency.

To recover Nanog with its associated proteins, we used a biotiny-lation/proteomics approach10. Nanog complementary DNA bearingan amino-terminal Flag epitope and a short peptide tag that serves asa substrate for in vivo biotinylation (Fig. 1a) was expressed at about20% of the endogenous level in ES cells previously engineered toexpress Escherichia coli biotin ligase BirA (Fig. 1b, left). Low-levelexpression reduces the likelihood that non-physiological interactionswill be identified or that the stoichiometry of protein complexes willbe perturbed11. Biotinylated Nanog (bioNanog) is efficiently recov-ered from ES cell nuclear extracts with streptavidin (SA)-agarose(Fig. 1b right), and is biologically active, because its expression

attenuates the dependence of ES cells on leukaemia inhibitory factor(LIF; Supplementary Fig. S1). Cells coexpressing BirA and bioNanogare functionally equivalent to wild-type ES cells in stem-cell markergene expression (Supplementary Fig. S1). On size fractionation of EScell nuclear extracts, bioNanog and endogenous Nanog behave simi-larly, broadly distributed with an apparent molecular mass of morethan 160 kDa to 1 MDa, far greater than that of the predicted poly-peptide (34 kDa) (Fig. 1c).

To identify Nanog-associated proteins we used single-step captureon streptavidin beads10, as well as tandem purification with anti-Flagimmunoprecipitation followed by capture with streptavidin. Nuclearextracts of ES cells expressing either BirA alone or BirA plus bioNanogwere processed in parallel. Total peptides sequenced and proteins

identified by whole-lane liquid chromatography–tandem mass spec-trometry (LC–MS/MS) are summarized in Fig. 1d (left panel; see alsoSupplementary Data). With criteria described in the Supplementary Methods, putative Nanog-associated proteins were chosen (Fig. 1d,right panel). The candidates, which are predominantly transcriptionfactors or components of transcriptional complexes, include proteins

previously studied in an ES cell context (for example Oct4 or Dax1)andfall into three groups. Thefirstgroup includes proteins present inat least two of three independent one-step purifications and tandempurification. These proteins (Sall1, Sall4, Rif1, Tif1b, Mybbp, Dax1and Nac1) are prime interaction candidates. The second groupincludes proteins (Zfp281, Err2, Elys, Oct4, Zfp198, NF45 andHDAC2) that may be part of unstable or transient complexes thatdissociate during tandem purification. Proteins in the third group(REST, Sp1 and Wapl) seem ‘masked’ in one-step purificationsbut are recovered in the tandem procedure. Proteins identified by our approach may interact directly with Nanog, or more often indir-ectly through their association with other components of a proteincomplex.

We then validated selected Nanog-interacting candidates by per-

forming co-immunoprecipitation experiments. Interactions betweenNanog and Dax1, Nac1, Zfp281 and Oct4 were confirmed in hetero-logous 293T (Supplementary Fig. S2) and ES (Fig. 2a–d) cells tran-siently transfected with constructs as indicated (see the figure legendfor details). In addition, we generated ES cells expressing biotinylatedNac1, Zfp281, Dax1 and Oct4, at or below endogenous protein levels(Supplementary Fig. S3), for affinity capture with anti-streptavidin-agarose and western blotting with anti-streptavidin–horseradish per-oxidase (HRP) antibody. We confirmed that Nanog co-purifies withbioNac1, bioZfp281, bioDax1 and bioOct4 (Fig. 2e) and that Nac1 co-purifies with bioNanog (Fig. 2f). Furthermore, we confirmed theassociation of Rif1 with endogenous Nanog (Fig. 2g), as well as ecto-pically expressed Nanog (Fig. 2h, i).

To test the functional relevance of selected components, we

employed inhibition of endogenous RNAs by short hairpin RNA(shRNA; Supplementary Fig. S4). If Nanog-associated proteins influ-ence Nanog’s transcriptional activity, their loss of function would beexpected to affect ES cell pluripotency. We first examined shRNAsdirected to Dax1 and Sall4 in ES cells that express green fluorescentprotein (GFP) under thecontrol of Oct4 regulatory elements.In bothinstances a loss of pluripotency was observed, as revealed by dimin-ished Oct4 expression and concomitant derepression of selected lin-eage-specific markers (Supplementary Fig. S5). Recent conditionalgene targeting of Dax1 (ref. 12) and Sall4 (ref. 13) provides inde-pendent support for their crucial function in ES cells.

Further functional validation included shRNA inhibition of two Nanog-associated proteins, Nac1 and Zfp281, not previously

1Division of Hematology–Oncology, Children’s Hospital and Dana Farber Cancer Institute, Harvard Medical School, Harvard Stem Cell Institute, 2Howard Hughes Medical Institute,Boston, Massachusetts 02115, USA.

Vol 444 | 16 November 2006 | doi:10.1038/nature05284

364

NaturePublishingGroup©2006



examined in ES cells. Nac1 is a BTB-domain containing protein14,15

related to Drosophila bric-a-brac/tramtrack, whichprevents inappro-priate neural gene expression. Zfp281, the mouse homologue of human zinc-finger protein ZBP99, may regulate cell proliferation,differentiation and oncogenesis16. We infected wild-type ES cells withretrovirus containing an shRNA vector carrying a GFP cassette(Supplementary Fig. S4) and isolated GFPhigh clones harbouringNac1 and Zfp281 shRNAs. Two clones each with similar extents of RNA knockdown (about 80% for Nac1 and about 70% for Zfp281;

Fig. 3a) were studied in detail. Nac1 or Zfp281 knockdown cellsmaintained on a feeder layer formed small colonies, yet they retainedproper ES cell morphology (Fig. 3b, top panels). On passage to gela-tin in the presence of LIF, knockdown cells grew poorly and formeddisorganized clumps of small cells (Fig. 3b, bottom panels). Strikingderepression of primitive endoderm (Gata6/4 ), mesoderm/visceralendoderm (Bmp2 ) and neuroectoderm (Isl1) markers was observed,despite normal expression of the stem-cell markers (Nanog , Oct4 and Rex1; Fig. 3c). Validation of these findings was achieved throughthe use of an independent shRNA directed to Zfp281 and by exam-ination of gene-targeted Nac11/2 ES cells (Supplementary Figs S4

and S6). To assess the relevance of Nac1 and Zfp281 with respect totranscriptional control, we examined chromatin occupancy of theseproteins at the Gata6 promoter in wild-type ES cells. As shown inFig. 3d, Nac1 and Zfp281, as well as Nanog, are highly enriched at theGata6 promoter. These data strongly implicate Nac1 and Zpf281,along with Nanog, in the tight control of this crucial target geneand also show that derepression of Gata6 expression in knockdowncells is not due to off-target effects of shRNAs.

The broad pattern of Nanog on size fractionation, and the iden-

tification and validation, both physically and functionally, of asso-ciated proteins suggest that Nanog is a component of multipleprotein complexes. By performing similar proteomic analyses onseveral of the Nanog-associated proteins, as well as another stem-cell marker protein Rex1 (ref. 17), we sought to identify additionalpreviously unknown proteins and components shared betweencomplexes and to develop a protein network (Fig. 4a). The sizedistributions of complexes containing Oct4, Dax1, Nac1, Zfp281and Rex1 overlap that of Nanog (Supplementary Fig. S7). Weemployed ES cell lines expressing bioDax1, bioNac1, bioZfp281,bioOct4 and bioRex1 (Fig. 2e, and Supplementary Fig. S3) for the

Anti-

Nanog

a b

Anti-SA

MSGLNDIFEAQKIEWHEGAPSSR

Co-express in J1 ES CellsCo-express in J1 ES cells

+

Purify biotinylated protein

using SA

NE IP anti-SA-

agarose

1 2 Anti-SA–HRP

=

-bioNanog

endNanog

bioNanog

bioNanog

bioNanog

endbioP

bioNanog

bioNanog

endNanog

Total lysates

++

+–

++

+–

––

BirAbioNanog

BirA

c

1 2 3

Anti-SA–HRP

Anti-Nanog

FL

Bio

01+2+4Zfp281

07+2+1Err2

01+2+1Elys

02+0+1Oct4

22+0+2Nac1

10+1+2Dax1

00+2+5Zfp198

02+1+0HDAC2

04+0+1NF45

28+14+15Mybbp

1511+16+13Tif1β

312+5+5Rif1

10+0+1Wapl

1

12

8

1

4

Peptide nos

(Tandem)

0+1+0Sp1

0+0+1REST

4+2+14Sall4

3+2+1Sall1

4+7+8Nanog

Peptide nos

(MS1+MS2+MS3)Protein ID

T o t a l 2 , 5 3 0

p e p t i d e s ( 1 7 8

p r o t e i n s )

T o t a l 3 , 4 1 0

p e p t i d e s ( 2 6 6

p r o t e i n s )

132 6620

107 8810

168 9810

1 10 30

d

Tandem

One-step

(MS2)

One-step

(MS1)

One-step

(MS3)

Void 700 kDa 440 kDa 158 kDa232 kDa

bioNanogBirA

Figure 1 | Protein complexes containing Nanog protein in mouse ES cells.

a, Strategy for biotinylation of Nanog in ES cells. Flag (FL) and 23-amino-acid biotin (Bio) tags are indicated. SA, streptavidin. b, Western blotanalyses of expression of biotinylated Nanog (bioNanog) as well asendogenous Nanog (endNanog) (left), and binding of bioNanog to SA beads(right). IP, immunoprecipitation. c, Western blots of fractions using anti-Nanog (top)and anti-SA–HRP antibodies(bottom).d, Summary ofLC–MS/MS results from three independent one-step purifications and a tandempurification. Totalpeptidessequencedand proteins identifiedin BirA (black circles) and bioNanog (red circles) samples are indicated. Components of Nanog complexes selected with the criteria described in Supplementary

Methods are listed in the table.

D a x 1 V 5

B i r A V 5

I P

S A

( b i o N

a n o g )

N a c 1 V 5

B i r A V

5

BirAbioNanog

+–

++

Anti-SA

bioNanog

bio

Nanogbio

Nanog

bioNanog/lgG

endNanog

Anti-Nac1

Nac1

N a n

o g V 5

B i r A

V 5

I P

a n t i - V 5

WB anti-Rif1

WB anti-Rif1

i

Rif1

Rif1

Rif1 Anti-SA

Anti-Rif1

f

IP

Anti-

Nanog

I n p u t

*

BirA

–

BirAbioNac1

BirAbioDax1

BirAbioZfp281

BirAbioOct4

I P S A

I n p u t

Z f p 2 8

1 V 5

B i r A V

5

I n p u t

I P S A

( b i o N a n o g )

I n p u t

I n p u t

O c t 4 V

5

B i r A V

5

WB

BirAbioNanog

+–

++ WB

WB

h

*

I P a n t i - V 5

I n p u t

<BirA

<BirA

Anti-Nanog

Anti-V5

Anti-Nanog

Anti-V5

WBb dc

Nanog

Nanog

WB anti-Nanog

WB anti-Nanog

Rif1

Rif1 Anti-

Rif1

Anti-

Rif1

Anti-

SA

P I s

e r u m

A n t i - R

i f 1

A n t i - N

a n o g

e

g

a

Figure 2 | Confirmation of Nanog association by co-immunoprecipitation

in ES cells. a–d, Co-immunoprecipitation of Nanog and Dax1 (a), Nac1(b),Zfp281(c)andOct4(d) in ES cells expressing BirA without (a, b)orwith(c, d) bioNanog. The asterisk marks a faint band in a. IP,immunoprecipitation. e, f, Co-immunoprecipitation of Nanog and Nac1,Zfp281, Dax1 and Oct4 in ES cells expressing bioNac1, bioZfp281, bioDax1,bioOct4 (e) and bioNanog (f). g–i, Co-immunoprecipitation of Nanog andRif1 in ES cells. In g, nuclear extracts from ES cells expressing BirA andbioNanog were incubated with preimmune (PI) serum, anti-Rif1 antibody,and anti-Nanog antibody, respectively, followed by capture with protein-A/G agarose. The asterisk indicates endogenous Nanog immunoprecipitatedby anti-Rif1 antibody. Note that bioNanog migrates together with the IgGband. Antibodies for IP and western blot (WB) analyses are indicated. Inh and i, experiments similar to those in a and f were performed except thatdifferent constructs and antibodies were used.

NATURE | Vol 444 | 16 November 2006 LETTERS

365




affinity purification of complexes, as described for bioNanog.Proteins were identified with the criteria applied for the selectionof Nanog-associated proteins and are summarized in Fig. 4a.Multiple proteins are shared between the different complexes. Aprotein interaction map (or mini-interactome)18 summarizes theseobservations (Fig. 4b). Several features suggest that the network isspecialized to establish and/or maintain ES cell pluripotency.

First, the network is remarkably enriched for proteins that havebeen shown, or are likely, to be required individually for controllingthesurvival or differentiationof theinner cell mass, or aspects of early development. These include 15 proteins (Fig. 4b, green circles) withknown phenotypic alleles in published knockout studies or commu-

nicated to us (Supplementary Table S1), and 4 proteins (blue circles)from shRNA-mediated inhibition described previously 19 or in thisstudy (Fig. 3, and Supplementary Figs S5 and S6). Three other pro-teins (yellow circles) have known phenotypes that are not consistentwith an early requirement in development (Supplementary TableS1). Thus, more than 80% of proteins for which knockout or knock-down studies have been performed seem essential for early devel-opment and/or ES cell properties. To approximate the ‘expected’fraction of essential genes, we curated 300 random transcriptionfactor genes and found that less than 15% of those expressed in EScells had knockout phenotypes consistent with an early requirement(data not shown).

Second, most genes encoding the proteins within the network areco-regulated, and specifically downregulated, on ES cell differenti-ation. In Fig. 4c, a rank order depicts the expression of transcripts

for the network proteins relative to Rex1. For comparison, a ‘control’set of randomly chosen gene transcripts was similarly analysed(Supplementary Fig. S8). We chose Rex1 as a comparator becauseit is abruptly and extensively downregulated on the differentiation of ES cells.Co-regulation of genes in the network is consistent with theirparticipation in a common cellular function or pathway.

Third, in reinforcement of this hypothesis, we find that many (atleast 56%) of the genes encoding proteins of the mini-interactomehave been reported as putative Nanog and/or Oct4 targets inrecent ChIP-on-chip20 or ChIP-PET19 analyses of chromatinoccupancy in human or mouse ES cells (Supplementary Table S2).Thus, ‘downstream’ gene targets of Nanog and Oct4 also serve as

‘upstream’ effectors in the pluripotency network. Although this fea-ture may stabilize aspects of the ES cell state, it also renders thenetwork vulnerable to attack at multiple points rather than at very few 21.

Fourth, the network is linked to several cofactor pathways, whichare largely involved in mediating transcriptional repression. Theseinclude the histone deacetylase NuRD (P66b and HDAC2), poly-comb group (YY1, Rnf2 and Rybp) and SWI/SNF chromatin remod-elling (BAF155) complexes. Of potential interest, Rex1 and Oct4, ascontrasted with Nanog or its closest partners (for example Dax1,Sall4 and Nac1), are associated with polycomb components. Nanogis linked to HDAC/NuRD through its association with Sall1/4 (ref.22) and Nac1 (ref. 23). The involvement of multiple co-repressorcomplexes provides both a method of regulating different sets of target genes and a fail-safe mechanism to prevent differentiation

b

N a n o

g O c t 4 R e

x 1

G a t a 6

G a t a 4

H N F 4

a

L a m b 1

B m p 2

B m p 4 C d

x 2 F g

f 5 I s l 1

B r a c h y

u r y

a

Nac1

Zfp281

0 . 1 6

0 . 3 4

Nac1.2872 shRNA Zfp281.1691 shRNAMock

2 1 . 9

B i o t i n - t a g g e d / B i r A

s h R N A / m o c k

s h R N A / m o c k

1 0 1 . 6

0 . 8 1

. 6

1 . 1

1 . 0 2

. 2

1 . 3

6 9 . 6

bioNac1 bioNanog bioZfp281

1 . 8 1 . 9 1 . 8

1 1 . 9

9 . 0

1 . 7 2 . 5

6 . 6 8

. 8

0 . 9

4 . 2

1 . 0

1 5 . 0

1 . 7 1 . 1 2 . 1

4 6 . 6

2

5 . 9

2 . 4 4 . 0

0 . 8

8 . 6

1 . 6 0 . 9 0 . 9

9 . 9

0 . 1 9

C l o n e 1

C l o n e 2

C l o n e 1

C l o n e 2

0 . 3 1

c d

Figure 3 | Functional validation by RNA-mediated interference.

a, Quantitative reverse transcriptase–polymerase chain reaction (RT–PCR)analysis of Nac1 andZfp281 gene expression (two cloneseach) after shRNA-mediated knockdown in wild-type ES cells. b, Morphology of ES cellsinfected with virus containing empty shRNA vector (Mock; left), Nac1shRNA (middle) or Zfp281 shRNA (right). ES cells were grown either onfeeders (top) or on gelatin in the presence of LIF (bottom). c, Quantitative

RT–PCR analysis of stem-cell markers as well as lineage-specific markers asindicated in Nac1.2872 knockdown cells (top) and Zfp281.1691 knockdowncells (bottom). d, Quantitativereal-time chromatin immunoprecipitation of biotin-tagged proteinson Gata6 (blackbars) aswell asGapdh (greybars) andb2 microglobulin (white bars) promoter sequences (as negative controls).Results are shown as means and s.e.m. from two clones in c and threeindependent experiments in d.

LETTERS NATURE | Vol 444 | 16 November 2006

366




Rex1Nanog

RequiemRequiemRequiemErr2Err2

Sall4Sall4Sall4Sall4Sall4

Tif1βTif1βTif1β

(Rif1)(Rif1)(Rif1)Rif1

Rex1

complexes

Oct4

complexes

Zfp281

complexes

Nac1

complexes

Dax1

complexes

Nanog

complexes

BAF155Cdk1Cdk1BAF155Dax1Dax1

Sp1

P66β

HDAC2

NF45

Arid3b

Zfp219

EWS

RYBP

Oct4

Zfp281

EWS

(Mybbp)

Wdr18

P66β

Btbd14a

Arid3b

Ari3a

Zfp281

Nac1

Sall1

Oct4

Mybbp

Wdr18Sp1Sp1

RESTREST

PELOHDAC2

WaplWapl

Rnf2NF45

Prmt1Elys

Rai14Zfp198

Etl1Zfp609Zfp281

Nac1Nac1

YY1Sall1

Oct4Oct4

Oct4Dax1 Nac1 Zfp281

Nanog Rex-1a

Time (days)

Rex1Dax1MybbpEtl1Err2

Oct4Tif1βElysPrmt1Wdr18RESTRif1BAF155Zfp281EWSRai14Sall1Sall4NanogNac1NF45HDAC2WaplSp1Zfp219Btbd14aCdk1Zfp609P66β

RequiemYY1Rnf2PeloZfp198Rybp

Arid3b Arid3a

0 0 . 2

5

0 . 5

0 . 7

5

1 1 . 5

2 4 7 9 1 4

c

Nanog

Dax1

Zfp281

Nac1

Oct4

Rex1

Sall1

Zfp198

Elys

NF45

Wapl HDAC2

REST

Sp1

Err2

Sall4

Rif1

Mybbp

Zfp609Pelo

EWS

Cdk1Requiem

Arid3a

Arid3b

Btbd14a

P66β

Wdr18

BAF155

RybpZfp219

YY1

Etl1

Rai14

Prmt1

Rnf2

Tif1β

b

Figure 4 | A protein interaction network in ES cells. a, Schematic depictionof the strategy for mapping protein–protein interaction network in ES cells.Starting with the primary tagged baits Nanog and another stem-cell markerprotein Rex1, interaction partners of both Nanog and Rex1 were identified.The validated interactors such as Dax1, Nac1, Zfp281 and Oct4 served assecondary tagged baits for the identification of tertiary interacting partners.Selected high-confidence components of multiprotein complexes are listedin the table. Proteins shared by multiple tagged baits are shaded similarly.Proteins in parentheses are present predominantly in tagged samples withsome peptide presence in one of the controls during one-step purifications.b, Depiction of the features of the interactome. Proteins with red labels are

tagged baits for affinity purification. Green lines indicate interactionsconfirmed by co-immunoprecipitation and the red line indicates interactionconfirmedin the literature27. Greencircles indicate proteins whoseknockoutresults in defects in proliferation and/or survival of the inner cell mass orother aspects of early development; blue circles indicate proteins whosereduction by RNA-mediated interference results in defects in self-renewaland/ordifferentiation of ES cells; yellowcircles areproteinswhoseknockoutresults in laterdevelopmentaldefects;white circles denote proteins for whichno loss-of-function data are available. c, Co-regulation of proteins in thenetwork on differentiation. Microarray data were curated from theliterature28 and are presented in rank order relative to Rex1.

NATURE | Vol 444 | 16 November 2006 LETTERS

367


A Protein Interaction Network for Pluripotency Of

Documents

Transcript of A Protein Interaction Network for Pluripotency Of