The tumour suppressor L(3)mbt inhibits neuroepithelial

ART I C L E S

The tumour suppressor L(3)mbt inhibitsneuroepithelial proliferation and acts oninsulator elementsConstance Richter1, Katarzyna Oktaba2, Jonas Steinmann1, Jürg Müller2 and Juergen A. Knoblich1,3

In Drosophila, defects in asymmetric cell division often result in the formation of stem-cell-derived tumours. Here, we show thatvery similar terminal brain tumour phenotypes arise through a fundamentally different mechanism. We demonstrate that braintumours in l(3)mbt mutants originate from overproliferation of neuroepithelial cells in the optic lobes caused by derepression oftarget genes in the Salvador–Warts–Hippo (SWH) pathway. We use ChIP-sequencing to identify L(3)mbt binding sites and showthat L(3)mbt binds to chromatin insulator elements. Mutating l(3)mbt or inhibiting expression of the insulator protein genemod(mdg4) results in upregulation of SWH pathway reporters. As l(3)mbt tumours are rescued by mutations in bantam or yorkieor by overexpression of Expanded, the deregulation of SWH pathway target genes is an essential step in brain tumour formation.Therefore, very different primary defects result in the formation of brain tumours, which behave quite similarly in theiradvanced stages.

Development of the Drosophila nervous system recapitulates manysteps in mammalian neurogenesis. Neurons in the adult fly brain arisefrom stem cells called neuroblasts that undergo repeated rounds ofasymmetric cell division during larval stages1,2. After division, onedaughter cell remains a neuroblast whereas the other is called theganglionmother cell and divides just oncemore into two differentiatingneurons. Most larval neuroblasts are inherited from the embryo3,but the so-called optic lobe neuroblasts located laterally on eachbrain lobe pass through a neuroepithelial stage and are therefore aparticularly suitable model for mammalian neurogenesis4,5. Duringearly larval stages, the neuroepithelial cells of the optic lobes proliferateand separate into the inner and outer optic anlagen6,7. During latelarval stages, neuroepithelial cells switch to a neurogenic mode. Onthe medial side, they generate optic lobe neuroblasts, which generatethe neurons of the medulla, the second optic ganglion4,8. Opticlobe neurogenesis is controlled by a wave of lethal of scute (l(1)sc)expression passing through the neuroepithelium from medial tolateral5. The activity of the JAK–STAT pathway inhibits neural waveprogression and thereby controls neuroblast number. Differentiationof neuroepithelial cells also involves theNotch, epidermal growth factor(EGF) and SWH pathways9–13.Characterization of Drosophila genes identified in brain tumour

suppressor screens has demonstrated that defects in neuroblast

1Institute of Molecular Biotechnology of the Austrian Academy of Science (IMBA), Dr. Bohr-Gasse 3, 1030 Vienna, Austria. 2European Molecular Biology Laboratory(EMBL), Meyerhofstraße 1, 69117 Heidelberg, Germany.3Correspondence should be addressed to J.A.K. (e-mail: [email protected])

Received 20 December 2010; accepted 24 June 2011; published online 21 August 2011; DOI: 10.1038/ncb2306

asymmetric cell division result in the formation of stem-cell-derivedtumours thatmetastasize and become aneuploid on transplantation14,15.These screens also identified lethal (3) malignant brain tumour16,17

(l(3)mbt ), a conserved transcriptional regulator18 that is also requiredfor germ-cell formation inDrosophila19. L(3)mbt binds to the cell-cycleregulators E2F (ref. 20) and Rb (ref. 21), but the relevance ofthese interactions is unclear. We show that in Drosophila, L(3)mbtregulates target genes of the SWH pathway that are important inproliferation and organ size control22–24. The SWHpathway is regulatedby the membrane proteins Expanded (Ex) and Fat, which activatea protein complex containing the kinases Hippo (Hpo) and Wartsto phosphorylate the transcriptional co-activator Yorkie25–27 (Yki).Yki acts together with the transcription factors Scalloped28–31 andHomothorax32 (Hth) to activate proliferative genes such as Cyclin Eand the microRNA (miRNA) bantam (ban)33 and Drosophila inhibitorof apoptosis 1 (diap1; also known as thread). On phosphorylation, Ykiis retained in the cytoplasm and its target genes are not activated. InDrosophila, the main role of the SWH pathway is to limit proliferationin imaginal discs and its absence leads to tumorous overgrowth34. Invertebrates, many homologues of key pathway members are tumoursuppressors, indicating that this function is conserved34.L(3)mbt contains three MBT domains that bind mono- or di-

methylated histone tails21,35. The results of biochemical experiments

NATURE CELL BIOLOGY VOLUME 13 | NUMBER 9 | SEPTEMBER 2011 1029

© 2011 Macmillan Publishers Limited. All rights reserved.

mailto:[email protected]

ART I C L E S

in vertebrates have indicated a role in chromatin compaction21, butwhether this role is conserved is not known. Results published whilethis paper was under review have shown that germline genes areupregulated in l(3)mbt mutant brains and are necessary for tumourformation36. Our data indicate that L(3)mbt is bound to insulatorsequences, which affect promoter–enhancer interactions and influencetranscription37,38. In Drosophila, the proteins CTCF, CP190, BEAF-32,Su(Hw), Mod(mdg4) and GAF are found at insulator sequences, buthow these factors act is unknown38.Our data show that tumour formation in l(3)mbt mutants is initiated

by the uncontrolled overproliferation of neuroepithelial cells in theoptic lobes due to the upregulation of proliferation control genesnormally repressed by the SWH pathway. L(3)mbt is located at DNAsequences bound by chromatin insulators and we propose that thefunction of L(3)mbt as a chromatin insulator is essential for repressingSWH target genes and preventing brain tumour formation.

RESULTSl(3)mbt tumours originate in the optic lobesl(3)mbt mutants—similarly to brat or lgl—showed a strong increasein the number of neural stem cells positive for the neuroblastmarker Deadpan (Fig. 1a and Supplementary Fig. S1a), resulting in anabnormal enlargement of the brain (Fig. 1a). However, in contrast tobrat and lgl mutants39–43, the asymmetric segregation of determinantswas unaffected (Fig. 1b and Supplementary Fig. S1b and data notshown). Instead, we observed abnormal enlargement of the optic lobes.In wild-type brains (Fig. 1c, control), the epithelial monolayer of theouter optic anlagen forms optic lobe neuroblasts at its medial edge(Fig. 1d,e). In l(3)mbt mutants, the neuroepithelia of both inner andouter optic anlagen were massively expanded (Fig. 1c, middle andright panels). Initially, this led to a delay in optic lobe neuroblastformation (Fig. 1c, top view) but later, optic lobe neuroblasts wereformed and their number was significantly increased (Fig. 1a,c). Unlikein wild-type brains, neuroblast formation was also seen in the centre ofthe optic lobe epithelium (Fig. 1c, close-up). To quantify the epithelialphenotype, we reconstructed optic lobe neuroepithelia (of inner andouter optic anlagen) in three dimensions (see Methods). Whereas opticlobe epithelia were on average 1.2×105 µm3 in wild-type third instarlarvae, their average size was significantly increased to 3×105 µm3 inl(3)mbt76 mutants and 8.25×105 µm3 in l(3)mbt76/Df(3R)D605 larvae(Fig. 1f). During first larval instar (L1), cell number andmitotic patternwithin the optic lobe of l(3)mbt mutants were normal and during earlysecond instar (L2 early), the inner and outer optic anlagen separatednormally (Fig. 1g). However, starting in late second instar (L2 late),the mutant epithelia expanded and started folding. As tumours didnot form in l(3)mbt clones and available RNA-mediated interference(RNAi) lines did not cause optic lobe phenotypes, we generated ashort hairpin/miRNA (shmiR; ref. 44) line to investigate cell autonomyof the overproliferation defect. Expression of l(3)mbtshmiR in centralbrain neuroblasts resulted in depletion of L(3)mbt protein; however,this did not cause overproliferation (Supplementary Fig. S1c,d).In contrast, when expressed in neuroepithelia of the optic lobes,l(3)mbtshmiR caused a strong overproliferation in the inner and outeroptic anlagen (Fig. 1h). Epithelial expansion was also seen in l(3)mbtmutant wing imaginal discs16 (data not shown). In most adult mutantwings, this resulted in a significant size increase (Supplementary

Fig. S1e,f) but occasionally, very small deformed wings were alsoobserved. Taken together, these data indicate that overproliferation ofneuroepithelial cells during late L2 stages initiates tumour formationin l(3)mbt mutants.

Signalling pathways controlling optic lobe proliferationAs staining of l(3)mbt mutants for aPKC and actin did not revealany change in apical basal polarity (Fig. 2a,b), we used the GAL4C855a

driver line to express dominant-active, dominant-negative and RNAiconstructs for the main signalling pathways, epigenetic complexesand epithelial polarity genes (Supplementary Fig. S2e). The GAL4C855a

driver is expressed in the inner and outer optic anlagen from firstto third larval instar and in imaginal disc epithelia (SupplementaryFig. S2a,b). We analysed epithelial tissue size in the inner opticanlagen, the outer optic anlagen and the imaginal discs and estimatedthe number of optic lobe neuroblasts (Supplementary Fig. S2c–eand see Methods).Activation of the EGF pathway promoted epithelial growthmainly in

the inner optic anlagen (Fig. 2c and Supplementary Fig. S2e). Activationof the JAK–STAT pathway increased epithelial size5 (Fig. 2d and Sup-plementary Fig. S2e), whereas deregulation of Dpp or overactivation ofFGF pathways did not cause any visible phenotypes (Fig. 2c,d).In contrast, inhibition of the SWH pathway resulted in a phenotype

similar to l(3)mbt mutants11 (Fig. 2f–i, Supplementary Fig. S2e).expanded (ex) RNAi in the optic lobes (Supplementary Fig. S2c) causedepithelial overproliferation similar to what has been described forex mutants11. Overexpression of Hpo together with the apoptoticinhibitor P35 strongly decreased the size of optic lobe epithelia (Fig. 2e).On expression of non-phosphorylatable Yki (ref. 26,27), the size ofneuroepithelia in the inner and outer optic anlagen was increasedfive- to tenfold (Fig. 2f,g), a phenotype also observed previously11. Asimilar albeit milder phenotype was observed on expression of the Ykitarget ban (Fig. 2h and Supplementary Fig. S2e). Thus, inhibition of theSWH pathway or overexpression of its target genes can recapitulate theincreased optic lobe proliferation seen in l(3)mbt mutants.

l(3)mbt tumour formation involves the SWH pathwayThe SWH pathway inhibits expression of diap1, ban, and—as partof a negative feedback loop—ex (Fig. 2i; ref. 45). To investigate SWHpathway activity, we used a diap1–GFP4.3 reporter, in which the secondtranscriptional start site (TSS2) of diap1 controls green fluorescentprotein (GFP) expression (see below for a map of the diap1 locusand ref. 30). In wild-type brains, diap1–GFP4.3 is expressed in theneuroepithelium of the optic lobes during second and early third larvalinstars but becomes restricted to three small stripes during mid thirdinstar (Fig. 3a). However, in l(3)mbt mutants, diap1–GFP4.3 remainedexpressed throughout the overgrowing neuroepithelium (Fig. 3b, notethat the high level of GFP expression in the control is in lamina cells).Moreover, in wing imaginal discs, diap1–GFP4.3 was upregulated inthe entire wing pouch in l(3)mbt mutants (Fig. 3b). The ex–LacZreporter was only moderately expressed in wild-type epithelia of theouter optic anlagen, but was upregulated in l(3)mbt mutants (Fig. 3c).A similar upregulation was seen in the wing imaginal disc (Fig. 3c). Toinvestigate ban-miRNA activity, we used a ban sensor–GFP carryingmultiple ban binding sites46. When ban is active, this sensor–GFP isdownregulated. In control optic lobes, the ban sensor–GFP was not

1030 NATURE CELL BIOLOGY VOLUME 13 | NUMBER 9 | SEPTEMBER 2011


ART I C L E S

OOA

Anterior Posterior

Medial

NENB

VNC

CB

NELamina

Medullaneurons

NB

Medial Lateral

BA

l(3)mbt ts1 L3 early l(3)mbt ts1 L3 late

CD

8–G

FP/D

ead

pan

/E-c

adhe

rin

OOA

IOA

OL NBs

OOA

IOA

Bra

in a

nter

ior

Control

Bra

in a

nter

ior

GAL4C855a> l(3)mbtshmiR

Mira

nda/

CD

8–G

FP

Bra

in c

ross

-sec

tion

OOA

IOA

IOA

h

Mira

nda/

PH

3

l(3)mbt76

Control

l(3)mbt76/Df(3R)D605

32 °C

32 °C

Dea

dp

an

g

Con

trol

l(3)m

bt7

6

L1

Act

in/P

H3

L2 early

Act

in

L2 late

OOA

IOA

OOA

IOA

OOAIOA

e

l(3)m

bt76 /T

M3

l(3)m

bt76

Df(3R)D

605/

l(3)m

bt76

Neu

roep

ithel

ial v

olum

e (×

105

µm

3)

P = 0.0451

NENE

Dorsal

Ventral

0

2

4

6

8

10

P = 0.0017

Cro

ss-s

ectio

nTo

p v

iew

Clo

se-u

p

db

Control

f

a

c

Lateral

Figure 1 l(3)mbt is necessary to prevent tumorous overproliferation of thelarval central nervous system. (a) Brains of control and l(3)mbt76/Df(3R)D605

larvae raised at 32 ◦C stained for Deadpan (anterior). Scale bars, 50 µm.(b) Central brain (CB) neuroblasts of control and l(3)mbt76 mutant larvaestained for Miranda and PH3 (for overview, see Supplementary Fig. S1b).Scale bars, 20 µm. (c) Brains of control and l(3)mbtts1 mutant larvaeexpressing GAL41407 >CD8–GFP and stained for Deadpan and E-cadherin.Top, surface views (anterior side) with optic lobe neuroepithelia (NE)outlined. Middle, cross-sections through brain; optic lobe tissues areoutlined (neuroepithelia of inner optic anlagen (IOA), neuroepithelia ofouter optic anlagen (OOA) and optic lobe neuroblasts (OL NBs)). Thewhite rectangles outline the areas shown in the close-up views (bottom).The white arrow points to folding of an inner optic anlage. Scale bars,50 µm (top and middle) and 20 µm (bottom). (d) Schematic illustration ofmid–late third instar larval brain (lateral view). Only outer optic anlagenare shown, which consist of neuroblasts (NB, dark blue), neuroepithelial

cells (red) and lamina cells (La, grey). Central brain and ventral nervecord (VNC) neuroblasts are depicted as larger circles (light blue) andcentral brain neurons are shown in green. (e) Schematic illustration ofneurogenesis in outer optic anlagen. Neuroepithelial cells (red) give rise tomedulla neuroblasts (blue) and lamina cells (grey). Neuroblasts give riseto medulla neurons (green). (f) Quantification of optic lobe neuroepithelialvolume for genotypes: l(3)mbt76/TM3 (N = 16), l(3)mbt76 (N = 5) andDf(3R)D605/l(3)mbt76 (N =4). N , number of brain hemispheres quantified.Error bars, s.e.m. (g) First and second instar stage optic lobes of control andl(3)mbt76 mutant brains stained for PH3 (left panel, L1) and actin. LateL2 inner and outer optic anlagen are clearly detectable in control brains,whereas they are indistinguishable in l(3)mbt76 mutants. Scale bars, 20 µm.(h) Outer optic anlagen (top and middle) and inner optic anlagen (bottom) ofthird instar brains expressing GAL4C855a > l(3)mbtshmiR >CD8–GFP at 29 ◦C(top and bottom left) and 32 ◦C (middle and bottom right; see Methods)stained for Miranda. Scale bars, 50 µm.



ART I C L E S

Overproliferation IOANo phenotype

Pos

terio

r b

rain

Act

in/M

irand

a/P

rosp

ero

GAL4C855a>Bsk-DN

IOAIOA

GAL4C855a>banControl

Act

in/M

irand

a

h

fControl GAL4C855a >YkiS168A

Act

in/M

irand

a/Y

kiS

168A

–RFP NB NE NE

NB

Contro

l 29

°C

GAL4C85

5a >YkiS16

8A –RFP

P = 0.0123

g

eGAL4C855a GAL4C855a>Hpo, P35

Act

in/P

rosp

ero/

Mira

nda

Yki

Yki

Ex

Mer

Mats

Hpo

Sav

WtsP

Kibra

Cytoplasm

i

NENE

NENE

0

5

10

15

Neu

roep

ithel

ial v

olum

e (×

10

5 μm

3 )

No phenotype Overproliferation OOA

GAL4C855a>HopTum-DA

Act

in/M

irand

a/P

rosp

ero

Ant

erio

r b

rain

OOAOOA

GAL4C855a>λ-HtlH3-DA (FGF)

d

GAL4C855a>λ-Top-DA

?

Fat

Crb

c

Control l(3)mbt 76

Act

in

Bra

inTo

p v

iew

, op

tic lo

be

Control l(3)mbt ts1

CD

8–G

FP/a

PK

C, C

nn

aC

ross

- se

ctio

n

b

Yki

ban diap1 ex

....Nucleus

HthSd

Figure 2 Candidate screen reveals a function for the SWH pathwayin optic lobe proliferation. (a) Optic lobes of control and l(3)mbtts1

mutant brains expressing GAL41407 >CD8–GFP and stained for aPKC andCnn (same channel, surface and cross-section views). Note the mitoticGAL41407 > CD8–GFP-positive cell within the mutant epithelium (whitearrow). Scale bars, 20 µm. (b) Optic lobe epithelium in cross-section view ofcontrol and l(3)mbt76 mutant brains stained for actin. Scale bars, 20 µm. (c)Larval brains expressing a dominant-active (DA) form of Egfr (λ-Top) showstrong overproliferation of inner optic anlagen (IOA) neuroepithelia, whereasa dominant-negative (DN) form of Bsk has no phenotype. Brains expressGAL4C855a and are stained for actin, Miranda and Prospero. Scale bars,50 µm. (d) Larval brains expressing a dominant-active form of JAK (HopTum)show overproliferation of outer optic anlagen (OOA, outline), whereas adominant-active form of FGF receptor λ-Htl-H3 has no phenotype. All

brains express GAL4C855a and are stained for actin, Miranda and Prospero.Scale bars, 50 µm. (e) Larval brains expressing Hpo and anti-apoptotic P35permanently activate the SWH pathway and show underproliferation of opticlobe neuroepithelia (NE, outlined). All brains express GAL4C855a and arestained for actin, Miranda and Prospero. Scale bars, 50 µm. (f) Larval brainsexpressing dominant-active (DA) YkiS168A–RFP show strong overproliferationof optic lobe neuroepithelia (outlined) but no overproliferation of neuroblasts(NB). All brains express GAL4C855a and are stained for actin and Miranda.Scale bars, 50 µm. (g) Quantification of optic lobe neuroepithelial volume ofcontrol (N =11) and YkiS168A–RFP-expressing (N =4) brains. N , number ofbrain hemispheres quantified. Error bars, s.e.m. (h) Larval brains expressingban miRNA show overproliferation of optic lobe neuroepithelia (outlined,arrows). All brains express GAL4C855a and are stained for actin and Miranda.Scale bars, 50 µm. (i) Schematic illustration of the SWH pathway.

detectable, indicating strong ban expression. The limited dynamicrange of the ban sensor–GFP did not allow us to detect further banupregulation in the optic lobes (data not shown). In l(3)mbt mutantwing discs, however, GFP was almost completely lost, indicating astrong increase in ban activity (Fig. 3d). Thus, L(3)mbt inhibits theexpression of SWH target genes. These effects were cell autonomousand not a consequence of tumour formation, because expressionof l(3)mbtshmiR in the posterior compartment of the wing disc usingGAL4en increased expression of diap1–GFP4.3 (Fig. 3e), ban (Fig. 3f)and ex–LacZ (Fig. 3g) only in this compartment without significantlyincreasing compartment size.

The SWH pathway regulates gene expression by excluding Yki fromthe nucleus26,27. Surprisingly, neither the subcellular localization of Ykinor the expression of the associated transcription factor Scalloped28–31

(Supplementary Fig. S3a,b and data not shown) was changed in opticlobe epithelia or wing discs of l(3)mbt mutants, indicating that L(3)mbtdoes not influence SWH signalling activity.To determine whether the SWH pathway is important for tumour

formation in l(3)mbt mutants, we investigated genetic interactions.In l(3)mbt and ban double mutants, neuroepithelial size wassignificantly decreased and tumours did not form (Fig. 4a andcompare with Supplementary Fig. S4a). Neuroepithelial size was



ART I C L E S

Bra

in

Control l(3)mbt76

ban

sen

sor–

GFP

Win

g d

isc

Win

g d

isc,

pro

ject

ion

b

d

Ant

erio

r

Mira

nda

/dia

p1–

GFP

4.3

/Act

ina

Mira

nda/

dia

p1–

GFP

4.3

DA

PI/

dia

p1–

GFP

4.3

Win

g d

isc,

pro

ject

ion

cControl l(3)mbt

Bra

in

ex–L

acZ

L(3)

mb

t/b

an s

enso

r–G

FP

Win

g d

isc,

pro

ject

ion

GAL4en>l(3)mbtshmiRf

Win

g d

isc,

pro

ject

ion

eW

ing

dis

c, p

roje

ctio

n

g

GAL4en>l(3)mbtshmiR

L(3)

mb

t/d

iap

1–G

FP4.

3L(

3)m

bt/

ex–L

acZ

GAL4en>l(3)mbtshmiR

La

NBNE

E2/Df

E2/76

Wild type L2/L3 Wild type L3 early Wild type L3 mid

Control l(3)mbt76

Figure 3 SWH target genes are misregulated in l(3)mbt mutants. (a) Simu-lated time course of optic lobe development from second to third larval instarstage in wild-type brains expressing diap1–GFP4.3 (outlined) and stained forMiranda and actin. Note the changing expression pattern of diap1–GFP4.3in optic lobes. Optic lobes consist of neuroblasts (NB), neuroepithelial cells(NE) and lamina cells (La). Scale bars, 50 µm. (b) Top, diap1–GFP4.3expressed in control and l(3)mbt76 mutant brains stained for Miranda(optic lobe neuroepithelia are outlined). Bottom, diap1–GFP4.3 expressedin control and l(3)mbt76 mutant wing discs stained for DAPI (maximumintensity projections). Each genotype contains one copy of diap1–GFP4.3.Scale bars, 50 µm. (c) Top, ex–LacZ expressed in control and l(3)mbtE2/Dfmutant brains stained for LacZ (optic lobe neuroepithelia are outlined).

Bottom, ex–LacZ expressed in control and l(3)mbtE2/76 mutant wing discsstained for LacZ (maximum intensity projections). Each genotype containsone copy of ex–LacZ. Scale bars, 50 µm. (d) ban sensor–GFP expressedin control and l(3)mbt76 mutant wing discs. Note that the level of bansensor–GFP is decreased in l(3)mbt76 mutants, reflecting an increase in banactivity. Each genotype contains one copy of ban sensor–GFP. Scale bars,50 µm. (e) Wing disc expressing GAL4en

> l(3)mbtshmiR and diap1–GFP4.3and stained for L(3)mbt (maximum intensity projection). Scale bars, 50 µm.(f) Wing disc expressing GAL4en

> l(3)mbtshmiR and ban sensor–GFP andstained for L(3)mbt (maximum intensity projection). Scale bars, 50 µm.(g) Wing disc expressing GAL4en

> l(3)mbtshmiR and ex–LacZ and stainedfor L(3)mbt and LacZ (maximum intensity projection). Scale bars, 50 µm.

decreased from 4.5 × 105 µm3 in ban1/+;l(3)mbt76 to less than1× 105 µm3 in ban1; l(3)mbt76 larvae at the same developmentalstage (Fig. 4b). Overexpression of Ex rescued the overproliferationin l(3)mbt76 mutants (Fig. 4c,d; optic lobe neuroepithelial size3.3 × 105 µm3 in l(3)mbt76, 1.8 × 105 µm3 in GAL4C855a > Ex,l(3)mbt76/+ control and in GAL4C855a > Ex, l(3)mbt76 rescuedbrains). Finally, tumour size in l(3)mbt mutants was decreased byremoving one copy of yki (Fig. 4e and Supplementary Fig. S4b–d;optic lobe neuroepithelial size 3 × 105 µm3 in l(3)mbt76/E2 and2.2 × 105 µm3 in ykiB5/+; l(3)mbt76/E2). Thus, the SWH pathway

and its target genes are important for tumour formation inl(3)mbt mutants.

L(3)mbt is a nuclear proteinTo analyse the subcellular localization of L(3)mbt, we generateda specific antibody (Fig. 5a). In Drosophila embryos and in larvae,L(3)mbt was nuclear in interphase and dispersed in the cytoplasmduring mitosis (Fig. 5a and data not shown). This staining wasspecific, because it was lost from l(3)mbt mutant larvae (data notshown and western blot analysis in Supplementary Fig. S5) and



ART I C L E S

P = 0.0045 P = 0.0014

Act

in/M

irand

a/P

rosp

ero

ban1, l(3)mbt76ban1/+, l(3)mbt76

l(3)mbtE2

Mira

nda/

E-c

adhe

rin/P

rosp

ero

Mira

nda/

E-c

adhe

rin/P

rosp

ero

c GAL4C855a>Ex, l(3)mbt76/+

Ex,l(3)mbt76

GAL4C855a >Ex, l(3)mbt76

Act

in/D

ead

pan

/Pro

sper

o

d

GAL4C85

5a >Ex, l(3

)mbt7

6 /+

Ex, l(3

)mbt7

6

GAL4C85

5a >Ex, l(3

)mbt7

60

1

2

3

4

a b

0

2

4

6

8

ban1 /+

,

l(3)m

bt76

ban1 ,

l(3)m

bt76ban

1 ,

l(3)m

bt76 /T

M6

ykiB5/+; l(3)mbtE2

Neu

roep

ithel

ial v

olum

e (×

105

µm3 )

P = 0.0003 P = 0.0043

Neu

roep

ithel

ial v

olum

e (×

105

µm

3 )

e

Figure 4 The SWH pathway shows genetic interaction with l(3)mbt.(a) Genetic interaction of ban1 and l(3)mbt76: ban1/+, l(3)mbt76 mutantbrain and ban1, l(3)mbt76 double-mutant brain stained for actin, Mirandaand Prospero. Double mutants show the ban1 phenotype (see alsoSupplementary Fig. S4a). Scale bars, 50 µm. (b) Quantification of opticlobe neuroepithelial (NE) volume of ban1, l(3)mbt76/TM6 (control) (N =5),ban1/+, l(3)mbt76 (N = 5) and ban1, l(3)mbt76 (N = 6) brains. N ,number of brain hemispheres quantified. Error bars, s.e.m. (c) Geneticinteraction of Ex and l(3)mbt76: GAL4C855a > Ex, l(3)mbt76/+ controlbrains, Ex, l(3)mbt76 mutant brains and GAL4C855a >Ex, l(3)mbt76 rescued

brains stained for actin, Deadpan and Prospero. Scale bars, 50 µm.(d) Quantification of optic lobe neuroepithelial volume of GAL4C855a

>Ex,l(3)mbt76/+ control brains (N = 5), Ex, l(3)mbt76 mutant brains (N = 6)and GAL4C855a > Ex, l(3)mbt76 rescued brains (N = 5). N , number ofbrain hemispheres quantified. Error bars, s.e.m. (e) Genetic interaction ofl(3)mbtE2 and ykiB5 : l(3)mbtE2 single-mutant brain and ykiB5/+; l(3)mbtE2

double-mutant brain stained for Miranda, E-cadherin and Prospero. HalvingYki levels in double mutants suppressed the l(3)mbtE2 phenotype (see alsoSupplementary Fig. S4b–d for further double-mutant brains as well asquantification). Scale bars, 50 µm.

on l(3)mbtshmiR expression (Supplementary Fig. S1d and Fig. 3e–g).The localization was confirmed in flies expressing functional redfluorescent protein (RFP)–L(3)mbt or GFP–L(3)mbt fusions (see alsoSupplementary Fig. S5a,b). Live-cell imaging of larval neuroblastsexpressing RFP–L(3)mbt revealed that the protein accumulatesin nuclear dots in interphase cells (Fig. 5b). When expressed insalivary glands, L(3)mbt localized to multiple bands on polytenechromosomes that were often characterized by a decreased levelof DAPI (4,6-diamidino-2-phenylindole) staining (Fig. 5c). Thus,L(3)mbt is a nuclear protein that most likely exerts its function byassociating with chromatin.

L(3)mbt binds to SWH pathway target lociTo examine chromatin binding, we carried out chromatin immuno-precipitation (ChIP) followed by quantitative PCR analysis fromthird instar larval brains and imaginal discs (see Methods). diap1can be transcribed from three promoters, TSS1, TSS2 and TSS3(refs 28,30). L(3)mbt bound moderately to TSS2 and strongly to TSS1(Fig. 6a). Consistent with this, diap1–GFP5.1 (which contains TSS1)was expressed in optic lobe neuroblasts and neurons and stronglyupregulated in l(3)mbt mutants (Fig. 6b). This is surprising becauseprevious reports concluded that TSS1 is not expressed in wing or eyeimaginal discs28,30. For TSS3, we detected weak binding of L(3)mbt;

however, the relevance of this interaction is unclear (Fig. 6a). L(3)mbtdid not bind to the bxd Polycomb response element in the Ubxlocus35 (Fig. 6a), but bound to other SWH target genes such as CyclinE (Fig. 6a; ref. 47).To determine binding sites of L(3)mbt on a genome-wide level, we

used Solexa sequencing. Results from two highly correlated indepen-dent ChIP experiments (Pearson correlation 0.8835, SupplementaryFig. S6a,b) showed that L(3)mbt bound close to TSSs (Fig. 6c and seeMethods). We assigned each bound region to the closest gene (seeMethods) and conducted a biological pathway (KEGG; Fig. 6d) andgene ontology (GO; Supplementary Fig. S6c) analyses of the predictedtarget genes48. As the SWH pathway is not yet annotated in thesedatabases, it was manually added to our analysis. SWH target genes areamong the top 10 pathways over-represented among L(3)mbt-boundgenes (data not shown). We identified seven out of eleven knownand predicted SWH target genes (ban, CycA, CycB, CycE, E2f, diap1,fj, ex, Mer, wg, Ser) and found a significant enrichment (P value0.025) of these targets among genes bound by L(3)mbt (Fig. 6f andSupplementary Fig. S6e; 3% false discovery rate, FDR). Importantly,the functional binding site at TSS1 of the diap1 locus (Fig. 6a) wasconfirmed (Fig. 6f). In addition, we found a low-occupancy peak at aconserved site 23 kilobases (kb) upstream of the ban transcription unit(Fig. 6f). Thus, L(3)mbt binds tomultiple SWH target genes.



ART I C L E S

Wild

typ

e la

te L

3

Overview brain

L(3)

mb

t/M

irand

a

Close-up CB NBs

Cen

tral

bra

in

neur

obla

stC

entr

al b

rain

ne

urob

last

Interphase 1 Prophase

Anaphase Interphase 2

GA

L414

07>

RFP

–L(3

)mb

t,C

D8–

GFP

GA

L414

07>

RFP

–L(3

)mb

t,C

D8–

GFP

a c

DA

PI/

GFP

–L(3

)mb

t

Polytene chromosomesGAL41407>GFP–L(3)mbt

b

NE

NE

NB

NB

Figure 5 L(3)mbt localizes to the nucleus. (a) Larval brain (anterior) stainedfor L(3)mbt and Miranda. Overview and close-up of central brain neuroblasts(CB NBs, outlined) in inter- and anaphase are shown. Note dispersion ofL(3)mbt in mitotic neuroblasts. Scale bars, 50 µm in overview and 10 µm inclose-up. NE, neuroepithelium, outlined in overview. (b) Stills of time-lapseimaging of mitotic central brain neuroblasts expressing GAL41407 >CD8–GFP

and RFP–L(3)mbt. Note that the level of RFP–L(3)mbt is decreased anddispersed in the cytoplasm during mitosis. Arrow points to RFP–L(3)mbtdots in interphase nuclei. Scale bars, 10 µm. (c) Close-up view of polytenechromosome expressing GAL41407 >GFP–L(3)mbt and stained for DAPI.Arrowheads point to DAPI-rich bands at which GFP–L(3)mbt is absent. Scalebars, 1 µm. (See also Supplementary Fig. S5.)

In our KEGG analysis, we also found enrichment for JAK–STATsignalling genes (Fig. 6d and Supplementary Fig. S6e). JAK–STATsignalling has previously been shown to regulate optic lobeneuroepithelial growth5. Indeed, the 10xSTAT92E–GFP sensor wassignificantly upregulated in both optic lobe neuroepithelia andwing discs of l(3)mbt mutants49 (Fig. 6e). Finally, we found ‘oocytematuration’ among the top 20 processes enriched in the KEGG andGO analyses (data not shown and Supplementary Fig. S6c). L(3)mbtbound close to the TSS of vasa (vas), bag of marbles (bam) and benigngonial cell neoplasm (bgcn) (Fig. 6g,h), which may explain the defects ingerm-cell formation in l(3)mbt mutants19,50.Microarray analysis of various Drosophila brain tumours has

identified a set of genes that are specifically deregulated in l(3)mbtmutants and are termed L(3)mbt signature (MBTS) genes36. L(3)mbtbound within the next±2 kb at 63% of these MBTS genes and at 85%of the MBTS genes with a described germline function (Fig. 6g andSupplementary Fig. S6d), indicating that the identified L(3)mbt-boundregions have in vivo relevance.

L(3)mbt binds to insulator elementsTo analyse L(3)mbt binding specificity, we searched for DNA motifsenriched among L(3)mbt binding sites. Among seven DNA consensusmotifs, fourmatched the consensus for the chromatin insulators CP190,BEAF-32, CTCF and Su(Hw) (Fig. 7a; refs 51–54), whereas three didnot match any known consensus motif (Supplementary Fig. S7a).Using published ChIP-on-chip (ChIP–chip) data51, we determinedthe overlap in binding sites (Fig. 7b). We found a strong overlapof L(3)mbt binding sites with class I chromatin insulators CP190,BEAF-32 and CTCF (Fig. 7b), whereas the overlap with the class IIinsulator protein Su(Hw) is smaller.The best-studied locus for insulator proteins is the HOX gene

cluster of the Bithorax complex37,55. Remarkably, L(3)mbt bindingsites within this locus correlated strongly with CP190 and CTCF,but less with Su(Hw) (Fig. 7c). To determine whether L(3)mbtbinding sites in the HOX cluster are functional, we analysed the

expression of the homeotic gene Abdominal-B (Abd-B). CTCF mutantsshow a characteristic downregulation of Abd-B at the posterior endof the larval ventral nerve cord55 (VNC). We found a significantreduction of Abd-B in l(3)mbt mutant larval central nervous systems(Fig. 7d) that closely resembled the change seen in CTCF mutants.Notably, a strong correlation between L(3)mbt and chromatin insulatorbinding was also observed at the SWH target genes ban and diap1(Supplementary Fig. S7b,c).As we did not observe any deregulation of the diap1–GFP4.3

reporter on RNAi-mediated knockdown of CTCF, CP190, BEAF-32and Su(Hw), we investigated the insulator protein Mod(mdg4) thatbinds to CP190 and Su(Hw) but not directly to DNA (refs 56,57).On RNAi-mediated knockdown ofmod(mdg4), ban sensor–GFP wasstrongly downregulated (Fig. 7e, indicating an increase in ban activity),whereas the diap1–GFP4.3 reporter was only mildly upregulated (datanot shown). Thus, insulator protein function is necessary to controlSWH target gene expression.

DISCUSSIONThe genes brat, lgl and dlg were identified as Drosophila braintumour suppressors. In all cases, defects in asymmetric cell divisioncause a huge expansion of the neuroblast pool39–42,58,59. This study,however, demonstrates a new mechanism for tumour formation.In l(3)mbt mutants, the neuroblast pool is expanded because anupregulation of SWH target genes results in a massive expansion ofneuroepithelial tissue. Why those neuroblasts proliferate indefinitelyon transplantation15 is not understood for any of thosemutants.Although the SWH pathway is essential for tumorigenesis in l(3)mbt

mutants, its overactivation cannot recapitulate the neuroblast tumourphenotype seen in l(3)mbt mutants (this study and ref. 11). Consistentwith the multifactorial origin of mammalian tumours, therefore, thecombined deregulation of several signalling pathways may be required.The Notch pathway may be involved as it regulates the formation ofoptic lobe neuroblasts from neuroepithelia10,12,13 and Notch pathwaygenes are bound by L(3)mbt (Supplementary Table S1 and data not



ART I C L E S

a

b

f

Control

dia

p1–

GFP

5.1/

Mira

nda

Bra

in

Sin

gle

pla

neP

roje

ctio

n

l(3)mbt76 early L3 l(3)mbt76 late L3

Per

cent

age

of in

put

diap1

–0.5

Control regions

1 2

–30 +8

Ubx

+4.7 +7.1+11.6

CycE

–0.15

L(3)mbt

dSfmbt

Win

g d

isc

pro

ject

ion

10xS

tat9

2E–G

FP

e Control l(3)mbt76

29 °C

Bra

in

10xS

tat9

2E–G

FP

Mira

nda

32 °C

g

h

Percentage of target genes in KEGG pathway

Annotated genesL(3)mbt target genes

1.0*

0.36

RibosomeNatural-killer-cell-mediated

cytotoxicityHeparan sulphate biosynthesis

mTOR signalling pathway

JAK–STAT signalling pathway

Dorso-ventral axis formationECM–receptor interaction Posterior probability ≥ 50%

Peak distance from TSS (bp)Fr

eque

ncy

1,000–1,000 0

L(3)mbt

Ph

Pho

dSfmbt

diap1

3% FDR

CG12030

ban

Ph

Pho

dSfmbt

L(3)mbt

–14 kb Hth Yki

–23 kb L(3)mbt

0.5% FDR

CG5895

200100

15105

15105

302010

200100

15105

15105

302010 MBTS genes with germline function

L(3)mbt-bound region within ± 2 kb 22/26 85%

L(3)mbt-bound region within ± 5 kb 25/26 96%

MBTS genes

MBTS genes:

64/102 63% L(3)mbt-bound region within ± 5 kb 77/102 75%

0.5% FDR

Nup358CG11875

RpS27bam

21,065 kb

200100

3% FDR

L(3)mbt-bound regions

0.5% FDR3% FDR

19,745 kb

bgcn

gbb

ytr

elF6 wibg TBPH

200100

vas

ck

15,075 kb

TfIIS

vig

CG33678

15,060 kb

200100

L(3)mbt-bound region within ± 2 kb

2.52.0

0

0.4

0.8

6543210

80

60

40

20

0

0 2 4 6 8 10

0.85*

0.70*0.52*0.45

0.45

0.5% FDR 3% FDR

16,030 kb 16,040 kb 610 kb 620 kb 630 kb 640 kb

0.5% FDR3% FDR

dc

Figure 6 L(3)mbt binds at the TSS and regulates SWH target genesand JAK–STAT pathway activity. (a) ChIP analysis at the diap1, CycE,Ubx and control loci in wild-type larval tissues for dSfmbt and L(3)mbt.ChIP signals at Polycomb response elements (green blocks) and otherregions are presented as the percentage of input chromatin (distancesfrom TSS are indicated in kilobases). Error bars, s.d. of three ChIPexperiments. (b) diap1–GFP5.1 (containing TSS1 of diap1) expressedin control and l(3)mbt76 mutant brains stained for Miranda (top, singleplane; bottom, maximum intensity projections). Scale bars, 50 µm.(c) Histogram of L(3)mbt binding frequency relative to the closestTSS. (d) KEGG pathway analysis of L(3)mbt target genes (3% FDR).Purple bars, percentage of L(3)mbt target genes in a particular KEGGpathway. Grey bars, percentage expected on the basis of all annotatedgenes. (e) 10xStat92E –GFP expressed in wing discs (top, maximumintensity projections) and brains (bottom, single planes) of l(3)mbt76/+

control and l(3)mbt76 mutant larvae stained for Miranda (bottom alone).Scale bars, 50 µm. (f) ChIP-seq tracks for L(3)mbt (purple) andChIP–chip tracks for the PcG proteins Ph, Pho and dSfmbt (grey) at thediap1 and ban loci. Purple and grey blocks represent bound regions(L(3)mbt-bound regions at 0.5% and 3% FDR). Green lines indicateL(3)mbt low-occupancy peak at −23 kb, and Hth and Yki binding siteat −14 kb (ref. 32) relative to the ban miRNA. (BDGP Release 5 (UCSCdm3) assembly.) (g) Overlap of genes specifically deregulated in l(3)mbtmutants36 (also called MBTS genes) with genes bound within ±2 kband ±5 kb by L(3)mbt. The number of L(3)mbt-bound genes versusthe total number of genes in each category as well as correspondingpercentages are shown. (h) ChIP-seq tracks for L(3)mbt (purple) atgermline genes vas, bam and bgcn. Rectangles outline L(3)mbt bindingsites at the TSS. Purple blocks indicate L(3)mbt-bound regions at 0.5%and 3% FDR.



ART I C L E S

b dC

P19

0in

form

atio

n co

nten

t∗∗∗

Mot

if en

richm

ent

L(3)

mb

tb

ackg

roun

d

info

rmat

ion

cont

ent

Position

Mot

if en

richm

ent

L(3)

mb

tb

ackg

roun

d

∗∗∗

Mot

if en

richm

ent

∗∗∗

BE

AF-

32C

TCF

Cla

ss I

Cla

ss II

4282,886 2,830

Su(Hw)L(3)mbt

2,2801,034 2,426

BEAF-32L(3)mbt

329 2,985 3,663

CP190L(3)mbt

1,4931,821 994

CTCFL(3)mbt

5 CTCF

5 BEAF-32

5 CP190

5 Su(Hw)

L(3)mbt200

3% FDR0.5% FDR

Ubx bxdabd-A

iab-4

Abd-B

IrcFas1

a

c

l(3)m

bt7

6C

ontr

ol

Dea

dp

an/A

bd

-B

Abd-B intensity VNC (normalized to Deadpan)

Abd-B mean fluorescent intensity/ Deadpan mean fluorescent intensity

l(3)mbt 76

Wild type

∗

Win

g d

isc

Mod

(md

g4)

ban

sen

sor-

GFP

GAL4en>mod (mdg4)-RNAie

00.

51.

01.

52.

00

0.5

1.0

1.5

2.0

info

rmat

ion

cont

ent

00.

51.

01.

52.

0

Su(

Hw

)in

form

atio

n co

nten

t0

0.5

1.0

1.5

2.0

4321 5 6 7 8 9 10 11

Position4321 5 6 7 8 109 12 1311 14 15

Position4321 5 6 7 8 109 12 1311 14 15 16 17 18 19

12,800 kb12,700 kb12,600 kb12,500 kb

00.

51.

01.

52.

0

∗∗∗

01.

02.

03.

0

L(3)

mb

tb

ackg

roun

d00.

51.

01.

52.

0

0 0.2 0.4 0.6 0.8

L(3)

mb

tb

ackg

roun

dMot

if en

richm

ent

00.

51.

01.

52.

0

Position4321 5 6 7 8 9 10 11

Figure 7 L(3)mbt binds at insulator-bound regulatory domains andinfluences Abd-B expression (a) De-novo-identified sequence motifssignificantly enriched in L(3)mbt-bound regions (0.5% FDR) correspondto motifs for the insulator-associated proteins CP190, BEAF-32, CTCFand Su(Hw) (refs 51,52). Discovered motifs are depicted as sequencelogos generated from a PWM. Histograms show motif enrichment inL(3)mbt-bound regions, computed as the ratio of the PWMmatch frequencyin the data set (purple) and the background (grey) frequency (*** P <0.001,Fisher’s exact test). (b) Venn diagrams showing the overlap of regions boundby L(3)mbt (0.5% FDR, purple) and regions bound by insulator-associatedproteins (grey) CP190, BEAF-32 and CTCF (class I insulator proteins),and Su(Hw) (class II insulator protein)51. (c) ChIP-seq tracks for L(3)mbt

(purple) and ChIP–chip tracks for insulator-associated proteins (grey)CP190, BEAF-32, CTCF and Su(Hw) (ref. 51) at the Bithorax complexlocus. Purple and grey blocks represent bound regions. For L(3)mbt, boundregions at 0.5% and 3% FDR are indicated. (BDGP Release 5 (UCSC dm3)assembly; see also Supplementary Fig. S7.) (d) Micrographs, ventral nervecord (VNC) of control (Canton S) and l(3)mbt76 mutant brains stainedfor Deadpan and Abd-B. Scale bars, 50 µm. Graph, quantification of theAbd-B mean fluorescence intensity normalized to Deadpan in control(N = 4) and l(3)mbt76 (N = 6) animals. Error bars, s.e.m., P = 0.041(one-tailed) (N , number of ventral nerve cords quantified). (e) Wing discexpressing GAL4en

>mod (mdg4)-RNAi and ban sensor–GFP and stainedfor Mod(mdg4). Scale bars, 50 µm.



ART I C L E S

shown). We also observe increased activity of the JAK–STAT pathway,a main regulator of optic lobe development5. Finally, the deregulationof germline genes in l(3)mbt mutants that has been described while thismanuscript was under review36 may provide another explanation.Our results indicate that L(3)mbt acts on insulator elements, which

isolate promoters from the activity of nearby enhancers acting onother genes37,38. Our analysis showed that L(3)mbt binding sitesoverlap with CP190, CTCF and BEAF-32, making the protein a class Ichromatin insulator51.The identification of a DNA consensus motif for a histone-binding

protein such as L(3)mbt is highly unexpected, as insulators are typicallynucleosome free51. The activity of these important transcriptionalregulators may be explained in several ways37,38. They may formphysical barriers blocking the interaction between enhancers andpromoters. Alternatively, they may mimic promoters and competewith endogenous promoters for enhancer interaction. Finally, theymay interact with each other or nuclear structures to form loopdomains that regulate transcriptional activity. Our data supportanother model in which insulators interact with histones on nearbynucleosomes and influence the structure of higher-order chromatin.Importantly, in the regions flanking CTCF binding sites, nucleosomesare enriched for histones that are mono- and di-methylated on H3K4or mono-methylated on H3K9 or H4K20 (ref. 60), the variantsto which MBT domains can bind in vitro18,21,35,61. As the humanL(3)mbt homologue L3MBTL1 was shown to compact nucleosomearrays in vitro21, a model becomes feasible in which simultaneousbinding to insulators and the surrounding nucleosomes decreasesflexibility and thereby restricts the ability of nearby enhancersto interact with promoters on the other side of the insulator.However, our data may equally well be worked into the otherprevalent models for insulator activity. As L(3)mbt is the onlyknown chromatin insulator, besides CTCF, that is conserved invertebrates, analysis of its homologues will be useful to distinguishbetween those possibilities.Optic lobe development resembles vertebrate neurogenesis4. Both

processes consist of an initial epithelial expansion phase followed byneurogenesis through a series of asymmetric divisions62. Together withprevious findings11, our data demonstrate that L(3)mbt and the SWHpathway are crucial regulators of the initial neuroepithelial proliferationphase. Interestingly, the SWH pathway has been implicated inregulating neural progenitors in the chicken embryo63 and it will beinteresting to determine the role of mammalian L(3)mbt in this process.It is remarkable that YAP is upregulated64 and L3MBTL3 is deleted in asubset of human medulloblastomas65. Medulloblastoma is the leadingcause of childhood cancer death and investigating the role of the SWHpathway may contribute to the progress in fighting this disease. �

METHODSMethods and any associated references are available in the onlineversion of the paper at http://www.nature.com/naturecellbiology

Note: Supplementary Information is available on the Nature Cell Biology website

ACKNOWLEDGEMENTSWe wish to thank I. Reichardt, N. Corsini, A. Fischer and C. Jueschke for commentson the manuscript; R. Neumueller and all former and present members of theKnoblich laboratory, as well as W. Wei, C. Girardot and J. Gagneur for discussions;R. Lehmann, D. Pan, L. Zhang, N. Tapon, B. Thompson, G. Halder, G. H. Baeg,

M. Labrador, the Flytrap Yale, the BloomingtonDrosophila Stock Center, the ViennaDrosophila RNAi Center and the Developmental Studies Hybridoma Bank for flystocks, antibodies and constructs; P. Serrano Drozdowskyj and M. Novatchkovafor bio-informatic support; M. Madalinski for affinity purification; E. Kleiner fortechnical assistance; K. Aumayr, P. Pasierbek and G. Schmauss for bio-opticssupport; and S. F. Lopez, S. Wculek and C. Valenta for fly work support. Work inJ.A.K.’s laboratory is supported by the Austrian Academy of Sciences, the AustrianScience Fund (FWF) and the EU FP7 network EuroSystem.

AUTHOR CONTRIBUTIONSC.R. conceived and conducted experiments, coordinated the project and wrotethe manuscript. K.O. conducted the bio-informatics analysis and conductedexperiments. J.S. wrote the peakfinder software and mapped the Solexa reads.J.M. supervised the project. J.A.K. initiated, designed and supervised the project,conceived experiments and wrote the paper.

COMPETING FINANCIAL INTERESTSThe authors declare no competing financial interests.

Published online at http://www.nature.com/naturecellbiologyReprints and permissions information is available online at http://www.nature.com/reprints

1. Doe, C. Q. Neural stem cells: balancing self-renewal with differentiation.Development 135, 1575–1587 (2008).

2. Neumuller, R. A. & Knoblich, J. A. Dividing cellular asymmetry: asymmetriccell division and its implications for stem cells and cancer. Genes Dev. 23,2675–2699 (2009).

3. Prokop, A. & Technau, G. M. The origin of postembryonic neuroblasts in the ventralnerve cord of Drosophila melanogaster. Development 111, 79–88 (1991).

4. Egger, B., Boone, J. Q., Stevens, N. R., Brand, A. H. & Doe, C. Q. Regulation ofspindle orientation and neural stem cell fate in the Drosophila optic lobe. NeuralDev. 2, 1 (2007).

5. Yasugi, T., Umetsu, D., Murakami, S., Sato, M. & Tabata, T. Drosophila optic lobeneuroblasts triggered by a wave of proneural gene expression that is negativelyregulated by JAK/STAT. Development 135, 1471–1480 (2008).

6. Hofbauer, A. & Campos-Ortega, J. A. Proliferation pattern and early differentiationof the optic lobes in Drosophila melanogaster. Roux Arch. Dev. Biol. 198,264–274 (1990).

7. Green, P., Hartenstein, A. Y. & Hartenstein, V. The embryonic development of theDrosophila visual system. Cell Tissue Res. 273, 583–598 (1993).

8. Toriya, M., Tokunaga, A., Sawamoto, K., Nakao, K. & Okano, H. Distinct functionsof human numb isoforms revealed by misexpression in the neural stem cell lineagein the Drosophila larval brain. Dev. Neurosci. 28, 142–155 (2006).

9. Wang, W. et al. Notch signaling regulates neuroepithelial stem cell maintenanceand neuroblast formation in Drosophila optic lobe development. Dev. Biol. 350,414–428 (2010).

10. Ngo, K. T. et al. Concomitant requirement for Notch and Jak/Stat signalingduring neuro-epithelial differentiation in the Drosophila optic lobe. Dev. Biol. 346,284–295 (2010).

11. Reddy, B. V., Rauskolb, C. & Irvine, K. D. Influence of fat-hippo and notchsignaling on the proliferation and differentiation of Drosophila optic neuroepithelia.Development 137, 2397–2408 (2010).

12. Egger, B., Gold, K. S. & Brand, A. H. Notch regulates the switch from symmetricto asymmetric neural stem cell division in the Drosophila optic lobe. Development137, 2981–2987 (2010).

13. Yasugi, T., Sugie, A., Umetsu, D. & Tabata, T. Coordinated sequential action of EGFRand Notch signaling pathways regulates proneural wave progression in the Drosophilaoptic lobe. Development 137, 3193–3203 (2010).

14. Gateff, E. Malignant neoplasms of genetic origin in Drosophila melanogaster.Science 200, 1448–1459 (1978).

15. Caussinus, E. & Gonzalez, C. Induction of tumor growth by altered stem-cell asym-metric division in Drosophila melanogaster. Nat. Genet. 37, 1125–1129 (2005).

16. Gateff, E., Loffler, T. & Wismar, J. A temperature-sensitive brain tumor suppressormutation of Drosophila melanogaster : developmental studies and molecularlocalization of the gene. Mech. Dev. 41, 15–31 (1993).

17. Wismar, J. et al. The Drosophila melanogaster tumor suppressor genelethal(3)malignant brain tumor encodes a proline-rich protein with a novelzinc finger. Mech. Dev. 53, 141–154 (1995).

18. Bonasio, R., Lecona, E. & Reinberg, D. MBT domain proteins in development anddisease. Semin. Cell Dev. Biol. 21, 221–230 (2010).

19. Yohn, C. B., Pusateri, L., Barbosa, V. & Lehmann, R. l(3)malignant brain tumor andthree novel genes are required for Drosophila germ-cell formation. Genetics 165,1889–1900 (2003).

20. Lewis, P. W. et al. Identification of a Drosophila Myb-E2F2/RBF transcriptionalrepressor complex. Genes Dev. 18, 2929–2940 (2004).

21. Trojer, P. et al. L3MBTL1, a histone-methylation-dependent chromatin lock. Cell129, 915–928 (2007).



http://www.nature.com/naturecellbiology/

http://www.nature.com/naturecellbiology

http://www.nature.com/reprints

http://www.nature.com/reprints

ART I C L E S

22. Reddy, B. V. & Irvine, K. D. The Fat and Warts signaling pathways: newinsights into their regulation, mechanism and conservation. Development 135,2827–2838 (2008).

23. Zeng, Q. & Hong, W. The emerging role of the Hippo pathway in cell contactinhibition, organ size control, and cancer development in mammals. Cancer Cell13, 188–192 (2008).

24. Saucedo, L. J. & Edgar, B. A. Filling out the Hippo pathway. Nat. Rev. Mol. Cell Biol.8, 613–621 (2007).

25. Huang, J., Wu, S., Barrera, J., Matthews, K. & Pan, D. The Hippo signaling pathwaycoordinately regulates cell proliferation and apoptosis by inactivating Yorkie, theDrosophila homolog of YAP. Cell 122, 421–434 (2005).

26. Dong, J. et al. Elucidation of a universal size-control mechanism in Drosophila andmammals. Cell 130, 1120–1133 (2007).

27. Oh, H. & Irvine, K. D. In vivo regulation of Yorkie phosphorylation and localization.Development 135, 1081–1088 (2008).

28. Wu, S., Liu, Y., Zheng, Y., Dong, J. & Pan, D. The TEAD/TEF family protein Scallopedmediates transcriptional output of the Hippo growth-regulatory pathway. Dev. Cell 14,388–398 (2008).

29. Zhao, B. et al. TEAD mediates YAP-dependent gene induction and growth control.Genes Dev. 22, 1962–1971 (2008).

30. Zhang, L. et al. The TEAD/TEF family of transcription factor Scalloped mediatesHippo signaling in organ size control. Dev. Cell 14, 377–387 (2008).

31. Goulev, Y. et al. Scalloped interacts with Yorkie, the nuclear effector of the Hippotumor-suppressor pathway in Drosophila. Curr. Biol. 18, 435–441 (2008).

32. Peng, H. W., Slattery, M. & Mann, R. S. Transcription factor choice in the Hipposignaling pathway: homothorax and yorkie regulation of the microRNA bantamin the progenitor domain of the Drosophila eye imaginal disc. Genes Dev. 23,2307–2319 (2009).

33. Thompson, B. J. & Cohen, S. M. The Hippo pathway regulates the bantam microRNAto control cell proliferation and apoptosis in Drosophila. Cell 126, 767–774 (2006).

34. Harvey, K. & Tapon, N. The Salvador–Warts–Hippo pathway—an emerging tumour-suppressor network. Nat. Rev. Cancer 7, 182–191 (2007).

35. Grimm, C. et al. Molecular recognition of histone lysine methylation by the Polycombgroup repressor dSfmbt. EMBO J. 28, 1965–1977 (2009).

36. Janic, A., Mendizabal, L., Llamazares, S., Rossell, D. & Gonzalez, C. Ectopicexpression of germline genes drives malignant brain tumor growth in Drosophila.Science 330, 1824–1827 (2010).

37. Valenzuela, L. & Kamakaka, R. T. Chromatin insulators. Annu. Rev. Genet. 40,107–138 (2006).

38. Bushey, A. M., Dorman, E. R. & Corces, V. G. Chromatin insulators: regulatorymechanisms and epigenetic inheritance. Mol. Cell 32, 1–9 (2008).

39. Lee, C. Y., Robinson, K. J. & Doe, C. Q. Lgl, Pins and aPKC regulate neuroblastself-renewal versus differentiation. Nature 439, 594–598 (2006).

40. Lee, C. Y., Wilkinson, B. D., Siegrist, S. E., Wharton, R. P. & Doe, C. Q. Brat is aMiranda cargo protein that promotes neuronal differentiation and inhibits neuroblastself-renewal. Dev. Cell 10, 441–449 (2006).

41. Bello, B., Reichert, H. & Hirth, F. The brain tumor gene negatively regulates neuralprogenitor cell proliferation in the larval central brain of Drosophila. Development133, 2639–2648 (2006).

42. Betschinger, J., Mechtler, K. & Knoblich, J. A. Asymmetric segregation of the tumorsuppressor brat regulates self-renewal in Drosophila neural stem cells. Cell 124,1241–1253 (2006).

43. Bowman, S. K. et al. The tumor suppressors Brat and Numb regulate transit-amplifying neuroblast lineages in Drosophila. Dev. Cell 14, 535–546 (2008).

44. Haley, B., Hendrix, D., Trang, V. & Levine, M. A simplified miRNA-basedgene silencing method for Drosophila melanogaster. Dev. Biol. 321,482–490 (2008).

45. Pan, D. Hippo signaling in organ size control. Genes Dev. 21, 886–897 (2007).46. Brennecke, J., Hipfner, D. R., Stark, A., Russell, R. B. & Cohen, S. M. Bantam

encodes a developmentally regulated microRNA that controls cell proliferation andregulates the proapoptotic gene hid in Drosophila. Cell 113, 25–36 (2003).

47. Harvey, K. F., Pfleger, C. M. & Hariharan, I. K. The Drosophila Mst ortholog,Hippo, restricts growth and cell proliferation and promotes apoptosis. Cell 114,457–467 (2003).

48. Bauer, S., Gagneur, J. & Robinson, P. N. GOing Bayesian: model-based gene setanalysis of genome-scale data. Nucleic Acids Res. 38, 3523–3532 (2010).

49. Bach, E. A. et al. GFP reporters detect the activation of the Drosophila JAK/STATpathway in vivo. Gene Exp. Patterns 7, 323–331 (2007).

50. Gateff, E. & Miyamoto, T. The tumor suppressor gene lethal (3) malignant brain tumor(l (3) mbt) of Drosophila melanogaster : multiple functions during development.Oncog. Cancer diagnosis 39, 135–141 (1990).

51. Negre, N. et al. A comprehensive map of insulator elements for the Drosophilagenome. PLoS Genet. 6, e1000814 (2010).

52. Zhao, K., Hart, C. M. & Laemmli, U. K. Visualization of chromosomal domains withboundary element-associated factor BEAF-32. Cell 81, 879–889 (1995).

53. Holohan, E. E. et al. CTCF genomic binding sites in Drosophila and the organisationof the bithorax complex. PLoS Genet. 3, e112 (2007).

54. Adryan, B. et al. Genomic mapping of Suppressor of Hairy-wing binding sites inDrosophila. Genome Biol. 8, R167 (2007).

55. Mohan, M. et al. The Drosophila insulator proteins CTCF and CP190 link enhancerblocking to body patterning. EMBO J. 26, 4203–4214 (2007).

56. Pai, C. Y., Lei, E. P., Ghosh, D. & Corces, V. G. The centrosomal protein CP190 is acomponent of the gypsy chromatin insulator. Mol. Cell 16, 737–748 (2004).

57. Gause, M., Morcillo, P. & Dorsett, D. Insulation of enhancer-promoter communicationby a gypsy transposon insert in the Drosophila cut gene: cooperation betweensuppressor of hairy-wing and modifier of mdg4 proteins. Mol. Cell Biol. 21,4807–4817 (2001).

58. Ohshiro, T., Yagami, T., Zhang, C. & Matsuzaki, F. Role of cortical tumour-suppressor proteins in asymmetric division of Drosophila neuroblast. Nature 408,593–596 (2000).

59. Peng, C. Y., Manning, L., Albertson, R. & Doe, C. Q. The tumour-suppressor geneslgl and dlg regulate basal protein targeting in Drosophila neuroblasts. Nature 408,596–600 (2000).

60. Fu, Y., Sinha, M., Peterson, C. L. & Weng, Z. The insulator binding protein CTCFpositions 20 nucleosomes around its binding sites across the human genome.PLoS Genet. 4, e1000138 (2008).

61. Klymenko, T. et al. A Polycomb group protein complex with sequence-specificDNA-binding and selective methyl-lysine-binding activities. Genes Dev. 20,1110–1122 (2006).

62. Gotz, M. & Huttner, W. B. The cell biology of neurogenesis. Nat. Rev. Mol. Cell Biol.6, 777–788 (2005).

63. Cao, X., Pfaff, S. L. & Gage, F. H. YAP regulates neural progenitor cell number viathe TEA domain transcription factor. Genes Dev. 22, 3320–3334 (2008).

64. Fernandez-L, A. et al. YAP1 is amplified and up-regulated in hedgehog-associated medulloblastomas and mediates Sonic hedgehog-driven neural precursorproliferation. Genes Dev. 23, 2729–2741 (2009).

65. Northcott, P. A. et al. Multiple recurrent genetic events converge on control of histonelysine methylation in medulloblastoma. Nat. Genet. 41, 465–472 (2009).



METHODS DOI: 10.1038/ncb2306

METHODSDrosophila strains and constructs. The l(3)mbt alleles used in this study were:l(3)mbtts1, l(3)mbtE2 (refs 16,17) and l(3)mbt76 (ref. 19; the original stock containeda suppressive second-site mutation, and the chromosome was cleaned duringrecombination to new FRT82B). All experiments and controls with l(3)mbt mutantswere conducted at 29 ◦C except if noted otherwise. Other mutant fly strains were:ex e1 (ref. 66); ykiB5 (ref. 25); ban1 (ref. 67); and Df(3R)D605 (Bloomington).The SWH and JAK–STAT pathway reporter lines were: ex697 (ex–LacZ; gift fromN. Tapon, London Research Institute, UK)66; diap1–GFP4.3 and diap1–GFP5.1(ref. 30); ban sensor–GFP (ref. 46); Sd–GFPtrapCA07575 (ref. 68); and 10xStat92E–GFP (ref. 49). See Supplementary Table S2 for a complete list of all fly strains. Inpathway–activity assays all genotypes contained only one copy of the respectivereporter.

UAS constructs for l(3)mbt and yki were cloned using the Gateway System(Invitrogen). l(3)mbt was PCR-amplified from DGRC complementary DNA cloneLD05287 and yki was cloned from mixed-stage cDNA. l(3)mbtshmiR was cloned asdescribed previously44. All experiments with l(3)mbtshmiR were conducted at 32 ◦Cexcept if noted otherwise (overproliferation phenotype in optic lobe neuroepitheliais observed only at 32 ◦C). The following driver lines were used: GAL41407 (ref. 69),GAL4C855a (ref. 70), GAL4c253 and GAL4en (all Bloomington Stock Center). All otherfly lines, genotypes and corresponding references as well as primer sequences arelisted in Supplementary Table S2.

For the candidate screen, UAS constructs were crossed to GAL4C855a at 29 ◦C.Phenotypes were determined from z stacks of stained brains from wanderingthird instar larvae and scored manually on a scale from −10 to +10 (0 =wild-type optic anlagen size; +10= l(3)mbt76 mutant optic anlagen; −10= strongunderproliferation).

Image acquisition and live-cell imaging. Images were acquired on a ZeissLSM510 Meta confocal microscope using ×25 (×0.7 zoom for imaginal discs and×1 zoom for brains) or ×60 oil immersion objectives (×3 zoom for polytenechromosomes). For live-cell imaging, larval brains were dissected in PBS, mountedin PBS on a round coverslip and covered with bioFOLIE25 (In Vitro Systems andServices, 96077317). Four-dimensional z stacks of 15–20 µm at 0.8–1 µm intervalswere acquired at 30 s intervals (×40 oil immersion objective). A wet chamberprevented evaporation.

Neuroepithelial tumour volume was quantified using Amira three-dimensionalreconstruction software. Z stacks of 40–80 µm were recorded at 2 µm intervals(×25 oil immersion objective at ×1 zoom). Optic lobe neuroepithelia wereoutlined manually (actin positive; Deadpan negative) in each z slice and thethree-dimensional surface and volume were determined automatically. Significancewas calculated by using an unpaired t -test withWelch’s correction and the data werevisualized as an aligned dot plot showing the mean± s.e.m.

Antibodies and immunohistochemistry. L(3)mbt antibodies were raised inguinea pigs against amino acids 1,211 to 1,477 of the L(3)mbt protein fusedto maltose-binding protein (affinity purified, immunofluorescence microscopy:1:200, ChIP: ∼5 µg; 10 µl for 250 µl chromatin). Other antibodies were rabbitanti-Miranda (1:200; ref. 42), rat anti-E-cadherin (1:100, DCAD2, DevelopmentalStudies Hybridoma Bank (DSHB), University of Iowa), guinea pig anti-Deadpan(1:1,000, gift from J. Skeath, Washington University in St Louis, USA), goatanti-aPKCzeta (1:200, Santa Cruz Biotechnology), mouse and rabbit anti-phosphohistone H3 (1:2,000, Cell Signaling), mouse anti-Prospero (1:50, MR1A, DSHB),mouse anti-Lamin (1:50, ADL67.10, DSHB), chicken anti-β-galactosidase (1:500,Abcam), rabbit anti-Yki (1:200, D. Pan, Johns Hopkins School of Medicine inBaltimore, USA), rabbit anti-dSfmbt531–980aa (ChIP: 20 µl, ref. 61), rabbit anti-Mod(mdg4) (ref. 71), Alexa Fluor 488 and 568 phalloidin (Invitrogen). Allsecondary antibodies were generated in donkey and were from Invitrogen.

Immunofluorescence microscopy experiments in larval brains and discs werecarried out as described previously42. Briefly, brain–disc complexes were dissectedin PBS and fixed for 15–20min in 4% paraformaldehyde in PBS with 0.1% TritonX-100 (1 change of fixative); antibodieswere diluted in 10%normal donkey serum inPBS with 0.1% Triton X-100; brains were mounted in Vectashield containing DAPI(Vector Laboratories).

Chromatin preparation and ChIP. Chromatin immunoprecipitation (ChIP)experiments were carried out essentially as described previously72. A detailedprotocol can be found at the Nature Protocol exchange (http://www.nature.com/protocolexchange/; http://dx.doi.org/10.1038/protex.2011.229). Briefly, 200 wild-type (Canton S orw1118) third instar larvae were dissected in ice-cold PBS. Carcasseswere stored at 4 ◦C in wash B (maximal overnight). Brain–disc tissues werehand-dissected in wash B andmaterial of 600–800 brain–disc complexes was pooledfor one chromatin preparation. Sonication of dissected material was carried out in0.5ml of radioimmunoprecipitation assay (RIPA) buffer in two steps: first, twicefor 2min with a tip sonicator (Omni-Ruptor 250, Omni-International; microtip,power output: 20, pulser: 60); second, three times for 10min in a Covaris machine(Covaris S, maximum all). This resulted in ∼500-base-pair (bp) fragments for

∼80% of the chromatin. After centrifugation (4 ◦C, 20min, 16,000g ), the solublechromatin in the supernatant was diluted, aliquoted and quick-frozen in liquidnitrogen before storage at−80 ◦C.

For ChIP assays, 110 or 250 µl of chromatin was incubated with antibodies(anti-L(3)mbt1211–1477aa and anti-dSfmbt531–980aa) at 4 ◦C overnight. DNA pelletswere re-suspended in 500 µlH2O before storage at −20 ◦C (10 µl per quantitativePCR reaction, triplicates). For ChIP-sequencing (ChIP-seq), 12 ChIPs were pooledinto 30 µlH2O and used for library preparation according to the Illumina ChIP-seqprotocol. Primer sequences are listed in Supplementary Table S2.

ChIP-seq mapping and data analysis. All uniquely mappable reads from twobiological replicates were mapped to the genome (FlyBase 5.27) using Bowtie 0.12.5(ref. 73). As our data set contained an unusually high number of reads (73 millionreads), we developed new peakfinder software called PyPeak that was optimized forrich data sets. For low-ranking peaks, PyPeak was comparable to the commonlyused MACS software74, whereas it was superior in identifying high ranking peaks(Supplementary Fig. S7d). PyPeak used the high number of sequences to refine theidentified peaks and often identified two independent binding sites where MACSpredicted one wide binding region (data not shown). The FDR was estimated asthe ratio of the number of peaks called in the control to the number of peakscalled for the ChIP data for a given threshold. To define a cutoff for ‘true boundregions’, we plotted the calculated FDR against the peak score (SupplementaryFig. S6b). We identified a highly stringent set of 3,314 L(3)mbt-bound regions ata 0.5% FDR, as well as a larger set of 4,572 bound regions at a 3% FDR, includingmore low-occupancy peaks. Analysis of the binding site distribution showed thatL(3)mbt bound preferentially between 700 bp upstream and 200 bp downstream ofTSSs (Fig. 6c). Within this promoter proximal section, the most frequent L(3)mbtbinding site was at 250 bp upstream of the TSS. Remarkably, this ‘peak’ containedtwo maxima with a distance of approximately 150 bp.

All ChIP-seq data sets have been deposited in the GEO repository (GSE29206).The peak-finding software is available under the terms of the GNU GeneralPublic License and can be downloaded at http://github.com/steinmann/pypeak. Thefrequency of peak distances with respect to the closest TSS was computed.

Data analyses were done using the Drosophila melanogaster BDGP Release5 (UCSC dm3) genome assembly and the Flybase 5.26 genome annotationrelease. ChIP-seq tracks were visualized with the Integrated Genome Browser75. Asbackground for all L(3)mbt target gene analyses, all annotatedD.melanogaster geneswere used.

We assigned to each of the 3,314 and 4,572 L(3)mbt-bound regions (0.5% FDRand 3% FDR) the gene with the closest TSS from the peak summit of the region.Bound regions farther than ±5 kb from a TSS were left without an assigned gene(311 of 3,314 or 9.3% and 415 of 4,572 or 9%). Totals of 2,730 (0.5% FDR)and 3,543 (3% FDR) unique genes were assigned as target genes (SupplementaryTable S1). The 2,730 (0.5% FDR) target genes were tested for enriched GO slim termannotations (http://www.geneontology.org) with a hypergeometric distribution test.As the number of tested terms is small, P values were not adjusted for multipletesting.

The Ontologizer application76 was adapted to examine the 3,543 (3% FDR)target genes for enrichment in KEGG pathways (http://www.genome.jp/kegg/,Release 53.0) and for enriched Gene Ontology (GO) term annotations (http://www.geneontology.org; biological process subontology). The significance (posteriorprobability) of this enrichment is based on a Bayesian model-based gene setanalysis48.

Motif analysis. De novo motif discovery was carried out on 100-bp-long regionscentred on the peak summit of the 3,314 L(3)mbt-bound regions (0.5% FDR)using RSAT (Oligo-analysis tool, Pattern-assembly tool (occ_sig> 2 and maximumsubstitution 1) and the Convert-matrix tool for position weight matrix (PWM)conversions)77 and MEME (parameters: distribution of zero or one occurrence persequence model, allowing sites on + and − strands, 6 as minimum and 25 asmaximummotif width, and as backgroundmodel a second-order Markov model)78

tools. The background in these and all following analyses consisted of repeat-maskedsequences around the TSS (−700 to+200 bp) of all annotatedD.melanogaster genes.The GGTT motif was found only using MEME. PWMs are listed in SupplementaryTable S2.

Binding-site predictions were generated with the Patser tool79 using the PWMs.The Patser score cutoffs varied for each motif, from ls2 to ls0. Motif enrichment wascomputed as the ratio of the observed motif frequency in the 3,314 L(3)mbt-boundregions (0.5% FDR) and in the background. Significance was assessed using aFisher’s exact test. The fraction of regions with a motif was assessed in the ranked4,572 L(3)mbt-bound regions (3%FDR) and the number ofmotifs identified in eachof the 3,314 L(3)mbt-bound regions (0.5% FDR) was counted.

For Venn diagram counts, two or more regions that overlap with at least onebase were merged and defined as a ‘common’ region. Overlaps were generatedusing the 3,314 L(3)mbt-bound regions (0.5% FDR) and regions bound byinsulator-associated proteins CP190 (6,648), BEAF-32 (4,706), CTCF (2,487,

NATURE CELL BIOLOGY


http://www.nature.com/protocolexchange/

http://www.nature.com/protocolexchange/

http://dx.doi.org/10.1038/protex.2011.229

http://github.com/steinmann/pypeak

http://www.geneontology.org

http://www.genome.jp/kegg/



DOI: 10.1038/ncb2306 METHODS

overlap CTCF-N and CTCF-C) and Su(Hw) (3,258, overlap of Su(Hw)-1 andSu(Hw)-2; ref. 51).

66. Boedigheimer, M. & Laughon, A. Expanded: a gene involved in the control of cellproliferation in imaginal discs. Development 118, 1291–1301 (1993).

67. Hipfner, D. R., Weigmann, K. & Cohen, S. M. The bantam gene regulates Drosophilagrowth. Genetics 161, 1527–1537 (2002).

68. Buszczak, M. et al. The Carnegie protein trap library: a versatile tool for Drosophiladevelopmental studies. Genetics 175, 1505–1531 (2007).

69. Luo, L., Liao, Y. J., Jan, L. Y. & Jan, Y. N. Distinct morphogenetic functions of similarsmall GTPases: Drosophila Drac1 is involved in axonal outgrowth and myoblastfusion. Genes Dev. 8, 1787–1802 (1994).

70. Hrdlicka, L. et al. Analysis of twenty-four Gal4 lines in Drosophila melanogaster.Genesis 34, 51–57 (2002).

71. Wallace, H. A., Plata, M. P., Kang, H. J., Ross, M. & Labrador, M. Chromatininsulators specifically associate with different levels of higher-order chromatinorganization in Drosophila. Chromosoma 119, 177–194 (2010).

72. Oktaba, K. et al. Dynamic regulation by polycomb group protein complexes controlspattern formation and the cell cycle in Drosophila. Dev. Cell 15, 877–889 (2008).

73. Langmead, B., Trapnell, C., Pop, M. & Salzberg, S. L. Ultrafast and memory-efficientalignment of short DNA sequences to the human genome. Genome Biol. 10,R25 (2009).

74. Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9,R137 (2008).

75. Nicol, J. W., Helt, G. A., Blanchard, S. G. J., Raja, A. & Loraine, A. E. The IntegratedGenome Browser: free software for distribution and exploration of genome-scaledatasets. Bioinformatics 25, 2730–2731 (2009).

76. Bauer, S., Grossmann, S., Vingron, M. & Robinson, P. N. Ontologizer 2.0–amultifunctional tool for GO term enrichment analysis and data exploration.Bioinformatics 24, 1650–1651 (2008).

77. van Helden, J. Regulatory sequence analysis tools. Nucleic. Acids Res. 31,3593–3596 (2003).

78. Bailey, T. L. & Elkan, C. Fitting a mixture model by expectation maximizationto discover motifs in biopolymers. Proc. Int. Conf. Intell. Syst. Mol. Biol. 2,28–36 (1994).

79. Hertz, G. Z. & Stormo, G. D. Identifying DNA and protein patterns withstatistically significant alignments of multiple sequences. Bioinformatics 15,563–577 (1999).

NATURE CELL BIOLOGY


Documents

The tumour suppressor L(3)mbt inhibits neuroepithelial