Genomes & Developmental Control
The nucleosome remodeling factor (NURF) regulates genes involved in
Drosophila innate immunity
So Yeon Kwona, Hua Xiaob, Bradley P. Gloverc, Robert Tjianc, Carl Wub, Paul Badenhorsta,⁎
aInstitute of Biomedical Research, University of Birmingham, Edgbaston, B15 2TT, UK
bLaboratory of Biochemistry and Molecular Biology, National Cancer Institute, Bldg. 37, Rm. 6060, NIH, Bethesda, MD 20892, USA
cDepartment of Molecular and Cell Biology, University of California at Berkeley, Berkeley, CA 94720, USA
Received for publication 28 September 2007; revised 25 January 2008; accepted 27 January 2008
Available online 12 February 2008
The Drosophila nucleosome remodeling factor (NURF) is an ISWI-containing chromatin remodeling complex that catalyzes ATP-dependent
nucleosome sliding. By sliding nucleosomes, NURF has the ability to alter chromatin structure and regulate transcription. Previous studies have
shown that mutation of Drosophila NURF induces melanotic tumors, implicating NURF in innate immune function. Here, we show that
NURF mutants exhibit identical innate immune responses to gain-of-function mutants in the Drosophila JAK/STAT pathway. Using
microarrays, we identify a common set of target genes that are activated in both mutants. In silico analysis of promoter sequences of these
defines a consensus regulatory element comprising a STAT-binding sequence overlapped by a binding-site for the transcriptional repressor Ken.
NURF interacts physically and genetically with Ken. Chromatin immunoprecipitation (ChIP) localizes NURF to Ken-binding sites in
hemocytes, suggesting that Ken recruits NURF to repress STAT responders. Loss of NURF leads to precocious activation of STAT target
© 2008 Elsevier Inc. All rights reserved.
Keywords: NURF; Chromatin remodeling; JAK/STAT pathway; ISWI; Drosophila; Innate immunity
The condensation of DNA in chromatin is a key determinant
that controls whether genetic information is available to be
expressed. By altering nucleosome dynamics, genes can be
rendered inaccessible or made available to the transcription
machinery. There are at least two mechanisms by which altered
nucleosome states can be induced. Firstly, post-translational
modification of the histone tails can change associations
between histones and DNA, and between neighboring nucleo-
somes, altering chromatin flexibility and conformation (re-
viewed in Kouzarides, 2007; Li et al., 2007). Another approach
is through the deployment of ATP-dependent chromatin
remodeling factors (reviewed in Saha et al., 2006; Choudhary
and Varga-Weisz, 2007). These multi-subunit protein com-
plexes utilize the energy of ATP hydrolysis to catalyze either
histone variant exchange, nucleosome sliding or nucleosome
The principal activity of the ISWI family of ATP-
dependent chromatin remodeling factors is energy-dependent
nucleosome sliding (Xiao et al., 2001). The nucleosome re-
modeling factor (NURF) is the founding member of this
family. By sliding nucleosomes, NURF has the potential to
expose transcription factor binding sites, thereby allowing
transcription activation. Consistent with this, studies of NURF
mutants have shown that NURF is required for transcription
activation in vivo (Deuring et al., 2000; Badenhorst et al.,
2002, 2005; Barak et al., 2003). However, nucleosome sliding
may also mask transcription factor binding sites, depending
on histone location, and repress transcription. Support for this
notion has been provided by studies of the orthologous yeast
Isw2 remodeling complex. Isw2 is needed to repress early
meiotic genes and MATa-specific genes (Goldmark et al.,
2000; Ruiz et al., 2003). Preliminary analysis of Drosophila
mutants lacking the large NURF-specific subunit (Nurf301)
Available online at www.sciencedirect.com
Developmental Biology 316 (2008) 538–547
⁎Corresponding author. Fax: +44 121 414 3599.
E-mail address: email@example.com (P. Badenhorst).
0012-1606/$ - see front matter © 2008 Elsevier Inc. All rights reserved.
revealed the presence of inflammatory or melanotic tumors
(Badenhorst et al., 2002). Melanotic tumors previously have
been shown to form after dysregulated activation of the JAK/
STAT pathway (Luo et al., 1995), suggesting that NURF may
be a negative regulator of JAK/STAT target genes.
Most known negative regulators of the JAK/STAT pathway –
for example SOCS proteins (Starr et al., 1997), the PIAS family
of inhibitors (Liu et al., 1998) and protein tyrosine phosphatases
(Baeg et al., 2005; Muller et al., 2005) – operate post-
transcriptionally to decrease activation of pathway components,
causing reduced transcription of target genes. However, it is
possible that mechanisms also exist to repress STAT target
promoters at the level of transcription, either in the absence of
signal, or to shut-off promoters after an initial signaling event.
Some evidence for this is provided by studies of the Drosophila
Zn-finger protein Ken and Barbie (Ken) and its mammalian
ortholog Bcl6. Binding sites for these transcription factors
partially overlap the STAT consensus element and over-
expression of either can down-regulate STAT-responsive
genes (Dent et al., 1997; Harris et al., 2005; Arbouzova et al.,
2006). To date, however, no well-defined involvement of
chromatin remodeling complexes in repression of the JAK/
STAT pathway has been demonstrated. Here, using whole
genome expression profiling, we have identified NURF as co-
repressor of a large set of JAK/STAT target genes. We also
provide evidence that NURF is recruited by Ken to repress a
subset of STAT target genes.
Materials and methods
Genetics and Drosophila strains
Flies were raised at 25 °C. The following genotypes were used for
microarray analysis, RT-PCR and immunofluorescence: w1118; Nurf3012/
Nurf3018(as described in Badenhorst et al., 2005) and hopTum(Luo et al.,
1995). Nurf301 mutations were maintained over the balancer chromosome
TM6B, Tb, Hu. Homozygous Nurf301 mutant larvae were selected as Tb+ as
described in Badenhorst et al. (2002, 2005).
Whole genome expression analysis
Whole genome expression profiling was performed using Affymetrix
Drosophila Genome Arrays. RNA for microarrays was collected from samples
of Nurf301 mutant, wild type and hopTumthird instar larvae that had reached
the wandering stage. Each sample comprised 30–50 animals that had been
cultured independently, and at low densities to avoid over-crowding. RNA
was isolated from larval samples using Trizol reagent (Invitrogen).
Precipitated RNA was resuspended in RNase-free water and treated for
15 min with 1U amplification grade DNaseI (Gibco BRL). Samples were
further purified using RNAeasy spin-columns (Qiagen), after which RNA was
labeled and hybridized to Affymetrix Drosophila Genome Arrays according to
the manufacturer's instructions. Microarrays were washed using an Affyme-
trix GeneChip Fluidics Station 400. Image data was captured using an
Affymetrix GeneChip array scanner and converted to numerical output using
Microarray Analysis Suite version 5.0. Further analysis and comparison was
using Genespring software. Normalization treatments used in Genespring
were: Per Chip (Normalize to 50th percentile); Per Gene (Normalize to
specific samples — in this case the mean of the signals of all wild-type
samples); Data Transformation (Set measurements less than 0.0 to 0.0).
Microarray datasets are available from ArrayExpress (http://www.ebi.ac.uk/
arrayexpress/), normalized expression values for common up-regulated genes
are listed in Supplementary Table 1.
Semi-quantitative RT-PCR analysis
High purity mRNA was obtained from 10 larvae of each genotype by
magnetic selection using μMacs according to manufacturer's instructions
(Miltenyi Biotech, Auburn, CA). Hemocytes were isolated from third instar
larvae by puncturing larvae in a small volume of HyQCCM-3 medium
(Hyclone) containing protease inhibitors (Complete, Boehringer). Contamina-
ting fat body was removed by centrifugation at 260×g for 10 min at 4 °C, a step
that also pellets blood cells. The less dense fat body remains in suspension and
can be removed. High purity mRNAwas obtained from blood cells of 400 larvae
by using μMacs columns as above. mRNA was reverse transcribed using
Superscript II (Invitrogen) at 42 °C. PCR was performed for 25–32 cycles using
specific primers as described (Badenhorst et al., 2005). Primer sequences are
available on request.
Drosophila hemocytes (from 10 larvae for wild-type and 2 larvae for
Nurf301 and hopTummutants) were suspended in 200 μl HyQCCM-3 medium
with 20% FCS and spun onto slides at 400 rpm for 5 min using a Shandon
Cytospin II slide centrifuge. Slides were fixed with 4% paraformaldehyde for
10 min, washed in PBS, and blocked overnight in PBS containing 0.1% saponin
(Sigma) and 10% FCS. Primary antibody incubation was performed in PBS
containing 0.1% saponin and 10% FCS for 1 h at room temperature. Mouse
MAb L1b (Sinenko et al., 2004) was used at 1:60 dilution; mouse MAb
CF.6G11 (anti-Mys) was used at 1:50 dilution; rabbit anti-βνintegrin (Yee and
Hynes,1993) was used at 1:500 dilution.FITC-conjugated secondaryantibodies
(Jackson ImmunoResearch) were used at 1:1000 in PBS containing 0.1%
saponin for 30 min at room temperature. Samples were mounted in Vectashield
containing DAPI (Vector Laboratories) and observed using a confocal
microscope (Zeiss Axiovert 100M).
Protein interaction assay
Purified recombinant FLAG-tagged NURF complex bound to anti-Flag
agarose beads (Sigma) was incubated with in vitro transcribed and translated
35S-labeled Ken (TnT, Promega). Bound proteins were analyzed by SDS-PAGE
followed by autoradiography as described (Badenhorst et al., 2005). Full length
Stat92E and Ken cDNAs were cloned into pET-30a (Novagen) and were
overexpressed in Escherichia coli strain BL21. E. coli was lyzed by sonication
in column buffer (50 mM NaH2PO4, 300 mM NaCl, 1% Triton-X100, 10 mM
Imidazole, 10 mM β-mercaptoethanol) and centrifuged at 25,000×g for 10 min
at 4 °C. The resultant supernatants were bound to Ni-NTA agarose resin
(Qiagen). Stat- or Ken-bound Ni-NTA agarose resin was incubated with NURF
complex and bound proteins were analyzed by western blot, using Rabbit anti-
Nurf301 antibodies (see below) at a dilution 1:1000.
Chromatin immunoprecipitation (ChIP)
ChIP was performed using Rabbit polyclonal anti-Nurf301 antibodies that
had been raised against an MBP fusion protein to Nurf301 amino acids 1036–
1628. This antibody has been validated by Western analysis and detects bands of
the predicted size. This antibody also exhibits nuclear staining when used for
immunofluorescence on larval imaginal discs. Western and immunofluores-
cence signals are not observed in extracts of null Nurf301 mutants or in Nurf301
mutant tissue. Hemocytes were isolated as described above in batches from 50
3rd instar larvae at a time. Hemocytes were fixed using 1% formaldehyde in
PBS containing protease inhibitors (Complete, Boehringer) for 15 min at 25 °C.
Hemocytes were washed three times using ice cold PBS containing protease
inhibitors and pelleted after each wash by centrifugation at 260×g for 5 min at
4°.Fixed hemocytepellets were stored at −80 °C until use.Hemocytesfrom 300
w1118or Nurf3012homozygous mutant 3rd instar larvae were used to prepare
soluble chromatin using the protocol of the Chromatin Immunoprecipitation
Assay Kit (Upstate Biotechnology). Wild-type and Nurf301 mutant input
chromatin amounts were normalized by protein assay (Bio-Rad Protein Assay).
Samples were pre-cleared using Protein A-conjugated magnetic beads (Dynal)
for 30 min at room temperature, followed by incubation with anti-Nurf301
539 S.Y. Kwon et al. / Developmental Biology 316 (2008) 538–547
coated Protein A-conjugated magnetic beads (Dynal) for 2 h at room
temperature. Immune complexes were recovered by magnetic selection, washed
and eluted using the protocol of the Chromatin Immunoprecipitation Assay Kit.
Target DNA abundance in ChIP eluates was assayed by quantitative PCR with
addition of 0.2 μCi [α-32P]-deoxycytosine 5′-triphosphate (Perkin Elmer,
specific activity 6000 Ci/mMol) as a tracer before the amplification step. Primer
pairs for PCR were CG5791: 5′-CCATTCATGTTTACGTCTGG-3′ and 5′-
CGCAATTCGATGGGGTTTTC-3′; and dei: 5′-AGCAACCGGCCTCACG-
CACA-3′ and 5′-CTTCGGTACGGTTTTCGCGC-3′.
Identical hemocyte phenotypes in Nurf301 and hopTummutants
In Drosophila, defense against pathogens is mediated by an
effective innate immune system comprising the fat body and
circulating leukocyte-like cells (hemocytes) (Crozatier and
Meister, 2007). Three hemocyte types occur: plasmatocytes
(which resemble human macrophages and monocytes), crystal
cells and lamellocytes. Lamellocytes encapsulate pathogens that
are too large to be engulfed by plasmatocytes. In wild-type
uninfected third instar larvae they are rarely detected in
circulation (Rizki, 1957; see Fig. 1). However, ectopic activ-
ation of the JAK/STAT signaling pathway triggers lamellocyte
differentiation. In mutants that express a constitutively-active
variant of the Drosophila JAK (hopTum), lamellocyte number is
increased to greater than 50% of circulating hemocytes (Luo et
al., 1995). One consequence of over-production of lamellocytes
is the formation of inflammatory tumors, or melanotic tumors.
We have shown previously that mutants lacking NURF subunits
exhibit melanotic tumors. Moreover, loss-of-function Nurf301
mutants enhance hopTumtumor phenotypes (Badenhorst et al.,
These results suggest that NURF may act as a negative
regulator of the JAK/STAT pathway. As a first step to test this
hypothesis, we examined if excess lamellocytes were also
observed in Nurf301 mutants as is the case for hopTummutants.
As shown in Figs. 1A and D, lamellocytes make up less than 1%
of the hemocytes in wild-type preparations. In contrast,
numerous lamellocytes were detected in hopTumand Nurf301
mutant (Figs. 1B, C) blood preparations. Lamellocyte frequency
was increased from 0.7% in wild-type animals to 71% and 57%
in hopTumand Nurf301 mutants respectively (Fig. 1D).
MAb L1-positive cells in hopTumand Nurf301 mutants were
more heterogeneous in size than wild type, ranging from 15 μM
in diameter (marginally larger than plasmatocytes) to 50 μM in
diameter (the size of wild-type lamellocytes). Clusters or
aggregates of lamellocytes were typically observed in both
mutant backgrounds (Figs. 1B, C), suggesting that the adherent
properties of hemocytes in hopTumand Nurf301 lamellocytes
were altered. Consistent with this, expression of the two known
Drosophila β-integrin subunits – Myospheroid (Mys) and βν-
integrin (βInt-ν) – was increased in both hopTumand Nurf301
lamellocytes (Fig. 2). This was accompanied by up-regulation
of α-integrin subunits (αPS4, see microarray data below),
indicating that functional adhesion complexes are formed,
thereby accounting for the presence of inflammatory tumors in
Whole genome expression profiling of Nurf301 and hopTum
The similarity of the hemocyte phenotypes of loss of
function NURF mutants and gain-of-function JAK mutants
supports the proposal that a normal function of NURF is to
repress gene targets of the Drosophila JAK/STAT pathway. If
indeed NURF represses JAK/STAT targets, whole genome
expression profiling of Nurf301 and hopTummutants should
reveal an overlapping set of genes with increased expression. As
shown in Fig. 3A, 148 genes were regulated in common
(Pb0.05 and fold change relative to wild type N or b2.5). When
only genes with increased expression in both mutants were
considered, a marked overlap was detected. 119 genes showed
elevated expression in both hopTumand Nurf301 mutants. This
corresponded to 37% of the genes that are increased in Nurf301
mutants and 40% of the genes that are increased in hopTum
mutants. In contrast, there is little overlap in genes that are
down-regulated in both mutants.
These comparisons were made using a 2.5-fold change in
gene expression as cut-off. By imposing a stringent cut-off, it
Fig. 1. Increased lamellocyte number in Nurf301 and hopTummutants.
Hemocytes from (A) wild-type, (B) hopTum, and (C) Nurf301 mutant third
instar larvae were immunostained with the lamellocyte-specific antibody MAb
L1b. MAb L1b staining is revealed in green, DAPI-stained nuclei are shown in
blue. Lamellocyte number is increased upon (B) activation of the JAK/STAT
pathway, or (C)lossof NURF.(D) Lamellocytefrequency wasdeterminedas the
ratio of MAb L1-positive cells relative to total hemocyte number (marked by
DAPI staining) in a field. Values are the mean±S.D. of 10 determinations for
540S.Y. Kwon et al. / Developmental Biology 316 (2008) 538–547
is possible that we excluded genes that, although co-regulated,
showed smaller increases in expression. As such, the overlap
in gene sets (Fig. 3A) represents a robust test of co-regulation.
To provide an additional measure of similarity we used linear
regression to compare the Nurf301 and hopTumtranscriptomes.
This revealed that fold change of genes with a statistically-
significant increase in expression in hopTummutants (Pb0.05
and fold change N1.0) correlated well with fold change in
Nurf301 mutants (Fig. 3B, r=0.84). In contrast, regression
analysis of genes with a statistically-significant decrease in
expression in hopTummutants (fold change b1.0) indicated
little correlation with Nurf301 profiles (r=0.04, data not
Gene ontology (GO http://www.geneontology.org/) classifi-
cation of the 119 genes that are up-regulated by at least 2.5 fold
in both mutant backgrounds confirmed that 25% of these
correspond to known defense response genes (Fig. 3C).
However, 33% of the genes identified have no previously
ascribed function. We speculate that some of these may be novel
innate immune targets. Support for this was provided by
comparison with published microarray datasets. 51% of these
genes had been detected in previous microarray analyses of
immune-challenged Drosophila (De Gregorio et al., 2001;
Boutros et al., 2002; Asha et al., 2003).
As shown in Fig. 3D, genes that are up-regulated included
components of the antibacterial humoral response — for
example Thiolester containing protein I, II and IV (TepI,
TepII, TepIV), Cecropin B (CecB), and six members of the
Immune induced molecule gene family including, IM1, IM2,
IM3, IM4, IM10 and IM23. Other defence-response genes
that showed increased expression include the hemocyte
expressed Scavenger receptor class C type I (Sr-CI), the
related Scavenger receptor class C type IV (Sr-CIV), and the
hemolymph phenol oxidase Diphenol oxidase A3 (Dox-A3).
Numerous serine proteases were also elevated. Serine protease
cascades play a vital role in coagulation, melanization and
activation of the Toll signaling pathway (Kambris et al.,
2006). Other potential targets revealed by microarrays include
cell adhesion molecules for example α- and β-integrin
subunits (αPS4 and mys), Fermitin 1 (Fit1 — the Droso-
phila homologue of mig-2), Vinculin (Vinc) and the matrix
glycoprotein Papilin (Ppn). Components of signal transduc-
tion pathways including the GTPase Rac2, GTPase activating
proteins (CG32560, Rapgap1 and RhoGAP18B) and G-protein
coupled receptors (mthl2 and mthl4) were also up-regulated.
Finally the transcription factors pointed (pnt), delilah (dei),
U-shaped (ush) and the Drosophila homologues of C/EBPζ/
CHOP10 (CG7839) and SNAP190 (CG2702) were also
revealed as potential targets of NURF and the Drosophila
Validation of microarray data
To verify the results of our microarray analysis we analyzed
transcript levels of selected targets using semi-quantitative RT-
PCR. Good correspondence between fold change determined by
microarrays and RT-PCR was observed for the majority of
transcripts. As shown in Fig. 4, marked increases in transcript
levels of TepI, CecB, IM1, IM23, Dox-A3 and εTry were
observed. Up-regulation of transcript levels of the transcription
factor dei was also confirmed.
By examining transcript levels in purified fat bodies and
hemocytes we were able to show that changes in expression
seen in whole animals were due to altered transcription in
immune-competent tissues (Figs. 5B, C). By examining gene
expression in isolated hemocytes we could also normalize
expression to hemocyte number, providing an important
control. These experiments indicated that increases in the levels
of components of the antibacterial humoral response were due
predominantly to increased transcription in the fat body (Fig.
5C). Conversely, the increased levels of the cell adhesion and
signaling molecules, and transcription factors observed in
whole animals were largely accounted for by increased
transcription in hemocytes (Fig. 5B). Additionally, this data
confirmed that for most known hemocyte-expressed genes
higher transcript abundance was due to transcriptional up-
regulation in hemocytes and not simply a consequence of
increased hemocyte number in mutants.
Fig. 2. Increased β-integrin expression in Nurf301 and hopTumhemocytes.
Hemocytes from wild-type, hopTum, and Nurf301 mutant third instar larvae were
immunostained with (A–C) antibodies against the Drosophila β-integrin
subunit Myospheroid (Mys), and (D–F) antibodies against the variant β-
integrin, βν-integrin (βInt-ν). Antibody staining is shown in green, nuclei
stained with DAPI are shown in blue.
541 S.Y. Kwon et al. / Developmental Biology 316 (2008) 538–547
In silico promoter analysis of NURF and STAT targets
Our microarray experiments showed that a common set of
genes were up-regulated in loss of function NURF and gain-of-
function JAK mutants, demonstrating that NURF normally
represses gene targets of the Drosophila JAK/STAT pathway.
The key question, however, is how this repression is achieved
and how is NURF recruited to these promoters. Previous studies
have shown that NURF can be recruited to genes by interaction
with DNA-binding transcription factors (Xiao et al., 2001;
Badenhorst et al., 2005). If NURF is recruited to repress JAK/
STAT target genes by a dedicated DNA-binding transcription
factor, conserved sequence elements corresponding to the
binding sites of this factor should be identifiable within the
promoters of our common set of target genes. Accordingly, we
performed an unbiased in silico search for common cis-
Fig. 3. Whole genome expression profiling identifies co-regulated genes in Nurf301 and hopTummutants. (A) Venn diagrams show significant overlap in the genes
with increased expression in Nurf301 and hopTummutants, but little overlap in genes with reduced expression. Nurf301-specific genes are indicated in yellow, hopTum-
specific genes in blue. Genes regulated in common are shown in grey. (B) Linear regression analysis of the 1162 genes that exhibit a statistically significant (Pb0.05
and fold change N1.0) increase in expression in hopTummutants, reveals that fold change in hopTumand Nurf301 mutants correlates well. (C) Gene ontology (GO)
classification of the 119 genes that are up-regulated by at least 2.5 fold in both mutant backgrounds indicates that 25% correspond to known defence related genes. (D)
Graph shows expression relative to wild-type for Nurf301 and hopTummutants. Colorbar denotes the magnitude of increase. Genes with STAT-binding consensi within
1 kb of the transcription start-sites are indicated with a blue circle.
542 S.Y. Kwon et al. / Developmental Biology 316 (2008) 538–547
elements that are overrepresented within the proximal promoter
regions of the identified NURF and JAK/STAT target genes.
start site of the 119 genes regulated in common in Nurf301 and
hopTummutants were collated and analyzed using the MEME
motif-finding algorithm (Bailey and Elkan, 1994). MEME
searches identified a putative regulatory element TTC[AGT]
[ACT][GT]GAA[AT] within this promoter set. This motif
includes the consensus STAT binding-site (TTCNNNGAA)
(Fig. 6A). However, we noticed that the STAT binding-site in
adenine residue. Recently it has been shown that the Drosophila
Bcl6 homolog Ken and Barbie (Ken) binds to a core element
GAAA that can overlap the STATconsensus (Arbouzova et al.,
2006). Both Ken and Bcl6 can repress JAK/STAT target genes
that contain their binding sites (Dent et al., 1997; Harris et al.,
2005; Arbouzova et al., 2006). As such, the regulatory element
we identified potentially corresponded to a Stat92E-binding site
with an overlapping binding site for the repressor Ken (Fig. 6A).
We examined how frequently this element occurred within
the NURF and STAT target promoters and observed that 49
out of 119 (41%) promoters contain consensus STAT-binding
Fig. 5. Tissue-specific changes in gene expression in Nurf301 and hopTum
mutants. (A) Schematic of third instar larvae indicating the principal innate
immune tissues, fat body (fb) and circulating hemocytes (he). (B) Expression of
selected targets in circulating hemocytes isolated from wild type, Nurf301 and
hopTummutant third instar larvae. (C) Expression of selected targets in fat body
tissue isolated from the corresponding animals. In panels B and C transcript
abundance is normalized to rp49 as a loading control. Fat body protein 1 (Fbp1)
and Hemese (He) transcripts provide an indicator of the purity of the tissue
preparations. High-level expression of Fbp1 and He should only be detected in
fat body samples and hemocyte samples respectively. Reduction of Fbp1
expression in the Nurf301 fat body is due to the established requirement of
NURF for ecdysone-triggered induction Fbp1 transcription (Badenhorst et al.,
Fig. 4. Validation of microarray data by semi-quantitative RT-PCR. Abundance
of selected transcripts was analyzed in RNA isolated from whole third instar
wild type, Nurf301 and hopTummutant larvae. Transcript abundance is
normalized to rp49 and Hemese as loading controls.
543 S.Y. Kwon et al. / Developmental Biology 316 (2008) 538–547
sequences within 1 kb of the transcription start-site,
suggesting that they are likely to be direct targets of the
JAK/STAT pathway. 70% of these STAT sites are flanked by
a downstream A (Supplementary Fig. 1A) indicating that
they are also potential Ken-binding sites. A comparable size
dataset of genes that show least change in expression in
Fig. 6. NURF interacts with the transcriptional repressor Ken. (A) In silico pattern search analysis of NURF and STAT target promoters identifies a conserved
consensus element TTC[AGT][ACT][GT]GAA[AT]. This corresponds to an over-lapping Stat92E- and Ken-binding site. (B) FLAG-tagged recombinant NURF
complex pulls-down in vitro translated Ken (lane 4) but does not bind the control protein Luciferase (lane 2). Neither recombinant Stat92E-coated (lane 12) nor
uncoatedNi-NTAagaroseresin(His-beads,lanes6, 10)pulls-downrecombinantNURFcomplex. In contrast,Ken-coated Ni-NTAagaroseresinwasableto pull-down
recombinant NURF complex(lane 8). NURF was revealed by western blot using anti-Nurf301 antibodies.(C) Structure of the CG5791 and dei loci showingSTATand
Ken bindingsites. STAT-binding sitesare indicatedby ovals.SequencesflankingSTAT-bindingsites are shownin brackets.STAT-bindingsequences(TTCNNNGAA)
are underlined. Ken-binding sequences (GAAA) are indicated by arrows. Regions amplified during ChIP PCR are shown by shaded bars. (D) Nurf301 ChIP signals
can be detected in wild-type hemocytes overlapping the Ken-binding sites in both the CG5791 and dei promoters (lane 3). These ChIP signals are not detected in
Nurf301 mutant hemocytes (lane 4). PCR signals from wild-type and Nurf301 ChIP inputs are shown in lanes 1 and 2 respectively. (E) Loss of one copy or either
Nurf301 or Ken enhances tumor phenotypes seen in hopTum/+ adult female flies. Enhancement was greatest after simultaneous reduction of Ken and Nurf301
levels. More than 100 flies of each genotype were scored.
544S.Y. Kwon et al. / Developmental Biology 316 (2008) 538–547
Nurf301 mutants relative to wild-type samples reveals that
23% possess consensus STAT-binding sequences within 1 kb
of the transcription start-site, and 15% possess STAT-binding
sites with overlapping Ken-consensi (Supplementary Fig.
1B). This suggests an enrichment of STAT- and Ken-binding
sequences within the putative NURF target promoters.
NURF interacts with Ken
To test if Ken does indeed recruit NURF to target
promoters we examined whether NURF and Ken interact
physically and genetically. First, we tested physical interac-
tions between NURF and Ken. Pull-down assay revealed that
in vitro translated Ken could interact with FLAG-tagged
recombinant NURF complex. As shown in Fig. 6B (lane 4)
FLAG-tagged recombinant NURF was able to pull-down
Ken, but did not show any interaction with a control protein
(Luciferase) that should not interact with NURF (Fig. 6B,
lane 2). We confirmed NURF interactions with Ken by
performing the reciprocal pull-down experiment. As shown in
Fig. 6B (lane 8) Ken-coated beads were able to precipitate
recombinant NURF complex. In contrast, no interaction with
NURF complex was observed using uncoated beads (Fig. 6B,
lanes 6 and 10, His-beads) or Stat92E-coated beads (Fig. 6B,
Next we determined whether NURF could be detected at any
of the putative STAT/NURF target promoters that contain Ken
binding sites. CG5791 and dei are both upregulated in Nurf301
and hopTummutant hemocytes (Fig. 5B). As shown in Fig. 6C,
both promoters contain STAT-binding consensi that are over-
lapped by binding sites for Ken. We isolated hemocytes from
wild-type and Nurf301 mutant larvae and performed chromatin
immunoprecipitation (ChIP) using polyclonal antibodies
against Nurf301, the only NURF-specific subunit of NURF.
Nurf301 ChIP signals could be detected in wild-type hemocytes
using primer-sets that overlap the Ken-binding sites (Fig. 6D,
lane 3). In contrast no ChIP signal was detected in null Nurf301
mutant hemocytes (Fig. 6D, lane 4). These results demonstrate
that NURF can be localized to promoters that contain Ken
binding-sites and that CG5791 and dei are direct targets of
Lastly, we tested genetic interactions between NURF and
Ken. We exploited the known melanotic tumor phenotype
observed in hopTummutant larvae. As noted previously
(Badenhorst et al., 2002), tumor incidence in animals carrying
one copy of the gain-of-function hopTummutant was increased
by reduction in Nurf301 levels (Fig. 6E). Reduction of Ken
levels also increased tumor frequency. However simultaneous
reduction in Nurf301 and Ken levels shows a synergistic effect
on tumor production (Fig. 6E). Tumor incidence was raised
from 22% to 66% when a mutant allele of both Ken and
Nurf301 was present. Taken together, these data indicate that
NURF interacts with Ken. These results extend the repertoire
of transcription factors shown to interact with NURF, with Ken
an addition to previously demonstrated binding partners of
NURF — the GAGA factor, GAL4-VP16, HSF (Xiao et al.,
2001) and EcR (Badenhorst et al., 2005).
Given the potential catastrophic effects of inappropriate
activation of signaling cascades, it is essential that the gene
targets of signaling pathways are maintained in a repressed state
in the absence of activating ligand (reviewed in Barolo and
Posakony, 2002). The ability of chromatin to repress gene
expression has long been postulated (Stedman and Stedman,
1950; Huangand Bonner,1962). It is assumed that packaging of
DNA into nucleosomes, and positioning of nucleosomes over
gene regulatory elements can block transcription. Members of
the ISWI family of ATP-dependent chromatin remodeling
enzymes are key regulators of nucleosome positioning and, in
this report, we have shown that NURF activity is required to
maintain repression of JAK/STAT target genes. Repression by
NURF is consistent with other studies of ISWI chromatin
remodeling enzymes. For example, the yeast Isw2 remodeling
complex is required for transcriptional repression (Goldmark et
al., 2000; Ruiz et al., 2003). In humans, the Snf2h-containing
chromatin remodeling complex NoRC slides nucleosomes to
silence rRNA genes (Li et al., 2006). More recently, ISWI in
African trypanosomes has been demonstrated to silence variant
surface glycoprotein gene expression sites (Hughes et al.,
Although a chromatin remodeling enzyme (SWI/SNF) is
required for activation of STAT-inducible genes (Ni and
Bremner, 2007), this is the first report to implicate a
chromatin remodeling enzyme in repression of JAK/STAT
target genes. There is, however, evidence that covalent histone
modification is involved in repression of JAK/STAT target
genes. The co-repressor SMRT suppresses induction of
STAT5 target genes. This suppression is blocked by the
addition of the histone deacetylase inhibitor TSA (Nakajima
et al., 2001), implying a chromatin component in repression.
In Drosophila, mutations in the heterochromatin component
HP1 have been shown to enhance tumor formation in hopTum
gain-of-function JAK mutants (Shi et al., 2006), further
implying a connection between chromatin, JAK/STAT and
We cannot exclude that some of the genes that show
increases in expression in the Nurf301 and hopTummutants may
be indirect targets of NURF. Changes in the proportion of
lamellocytes in these mutant backgrounds may affect transcrip-
tion of some genes, for example the Drosophila β-integrin
subunit mys. Nevertheless by ChIP we have been able to show
that NURF is located at the promoters of two potential targets,
CG5791 and dei. Importantly, NURF-biding coincides with
recognition sequences for STATare overlapped by binding sites
for the transcriptional repressor Ken. In addition, we have
shown that NURF physically interacts with Ken, providing a
means by which NURF can be recruited to JAK/STAT target
Ken is an ortholog of the mammalian proto-oncogene Bcl6
and, like Bcl6, can down-regulate JAK/STAT target genes
(Dent et al., 1997; Harris et al., 2005; Arbouzova et al., 2006).
Our data suggests a mechanism by which Ken represses
transcription. We propose that in unstimulated conditions (see
545 S.Y. Kwon et al. / Developmental Biology 316 (2008) 538–547
model, Fig. 7) Ken binds to JAK/STAT target promoters and
recruits NURF. NURF-mediated nucleosome-sliding then
establishes a repressed chromatin configuration that blocks
transcription, perhaps by positioning a nucleosome over the
transcription start site. Upon stimulation Stat92E enters the
nucleus, binds target promoters and, in addition to recruiting
co-activators, displaces Ken and thus NURF. The promoter is
switched from a repressive to active chromatin state, and
transcription can occur. In NURF mutants, we suggest that
repressive nucleosome positions are either not established or
maintained and, consequently, JAK/STAT targets are not
silenced. As a result, transcription can occur in the absence
of JAK/STAT activation.
More than two decades ago, Travers and colleagues
proposed that underlying DNA sequence can influence
nucleosome positioning (Satchwell et al., 1986), with some
sequences favoring, and others destabilizing nucleosomes.
Recent computational analysis has revealed that sequences at
yeast transcription start-sites encode nucleosomes that are
intrinsically unstable (Segal et al., 2006). The NURF-related
yeast Isw2 chromatin remodeling complex is able to override
these refractory sequences, positioning nucleosomes over them,
to block promoters (Whitehouse and Tsukiyama, 2006).
Interestingly, in ISW2 mutants, nucleosomes revert to thermo-
dynamically favorable positions exposing the promoter. We
speculate that Drosophila transcription start-sites may similarly
be refractory to nucleosomes. Normally, at JAK/STAT targets,
NURF overrides these sequences but, in NURF mutants, these
transcription start sites may similarly be exposed.
In the case of the innate immune system, prompt activation
of signaling cascades such as the JAK/STAT pathway in
response to pathogens are essential for survival. However, it is
also paramount that in the absence of challenge the innate
immune system be held in check or regulated, to prevent
inappropriate damage. In humans chronic immune-mediated
inflammatory conditions are characterized by the abnormal or
continued episodic activation of these pathways leading to
disease. We have shown that Drosophila NURF has a vital
function in preventing ectopic activation of the JAK/STAT
pathway. In the absence of NURF, Drosophila develop an
immune-mediated inflammatory syndrome — melanotic
tumors. Given the conservation of NURF between Drosophila
and humans, it is tempting to speculate that human NURF may
function to hold inflammatory pathways in check.
We thank I. Ando and R. Hynes for generously providing
flies and reagents. We also thank C. Buckley, and M. Mortin
for critical reading of the manuscript. This work was partly
supported by the intramural research program of the
National Cancer Institute (C.W.) and by the BBSRC (S.K.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.ydbio.2008.01.033.
Fig. 7. Model for NURF repression of JAK/STAT target genes. In unstimulated conditions (wt unstimulated) Ken binds to recognition sequences in JAK/STAT target
promoters and recruits NURF. NURF-induced nucleosome sliding establishes a repressed chromatin configuration on the promoter blocking transcription. After
stimulation (wt normal activation), Stat92E enters the nucleus, binds its recognition sequence and displaces Ken and NURF. The repressed chromatin conformation is
no longer maintained. In NURF mutants, a repressed chromatin structure is not established and JAK/STAT targets are not silenced. Transcription activation in the
absence of Stat92E nuclear entry can occur.
546S.Y. Kwon et al. / Developmental Biology 316 (2008) 538–547
References Download full-text
Arbouzova, N.I., Bach, E.A., Zeidler, M.P., 2006. Ken & barbie selectively
regulates the expression of a subset of Jak/STAT pathway target genes. Curr.
Biol. 16, 80–88.
Asha, H., Nagy, I., Kovacs, G., Stetson, D., Ando, I., Dearolf, C.R., 2003.
Analysis of Ras-induced overproliferation in Drosophila hemocytes.
Genetics 163, 203–215.
Badenhorst, P., Voas, M., Rebay, I., Wu, C., 2002. Biological functions of the
ISWI chromatin remodeling complex NURF. Genes Dev. 16, 3186–3198.
Badenhorst, P., Xiao, H., Cherbas, L., Kwon, S.Y., Voas, M., Rebay, I., Cherbas,
P., Wu, C., 2005. The Drosophila nucleosome remodeling factor NURF is
required for Ecdysteroid signaling and metamorphosis. Genes Dev. 19,
Baeg, G.H., Zhou, R., Perrimon, N., 2005. Genome-wide RNAi analysis of
JAK/STAT signaling components in Drosophila. Genes Dev. 19,
Bailey, T.L., Elkan, C., 1994. Fitting a mixture model by expectation
maximization to discover motifs in biopolymers. Proceedings of the Second
International Conference on Intelligent Systems for Molecular Biology.
AAAI Press, Menlo Park, CA, pp. 28–36.
Barak, O., Lazzaro, M.A., Lane, W.S., Speicher, D.W., Picketts, D.J.,
Shiekhattar, R., 2003. Isolation of human NURF: a regulator of Engrailed
gene expression. EMBO J. 22, 6089–6100.
Barolo, S., Posakony, J.W., 2002. Three habits of highly effective signaling
pathways: principles of transcriptional control by developmental cell
signaling. Genes Dev. 16, 1167–1181.
Boutros, M., Agaisse, H., Perrimon, N., 2002. Sequential activation of signaling
pathways during innate immune responses in Drosophila. Dev. Cell 3,
Choudhary, P., Varga-Weisz, P., 2007. ATP-dependent chromatin remodelling:
action and reaction. Subcell. Biochem. 41, 29–43.
Crozatier,M., Meister, M.,2007. Drosophilahaematopoiesis. Cell. Microbiol.9,
De Gregorio, E., Spellman, P.T., Rubin, G.M., Lemaitre, B., 2001. Genome-
wide analysis of the Drosophila immune response by using oligonucleotide
microarrays. Proc. Natl. Acad. Sci. U. S. A. 98, 12590–12595.
Dent, A.L., Shaffer, A.L., Yu, X., Allman, D., Staudt, L.M., 1997. Control of
inflammation, cytokine expression, and germinal center formation by
BCL-6. Science 276, 589–592.
Deuring, R., Fanti, L., Armstrong, J.A., Sarte, M., Papoulas, O., Prestel, M.,
Daubresse, G., Verardo, M., Moseley, S.L., Berloco, M., Tsukiyama, T., Wu,
C., Pimpinelli, S., Tamkun, J.W., 2000. The ISWI chromatin-remodeling
protein is required for gene expression and the maintenance of higher order
chromatin structure in vivo. Mol. Cell 5, 355–365.
Goldmark, J.P., Fazzio, T.G., Estep, P.W., Church, G.M., Tsukiyama, T., 2000.
The Isw2 chromatinremodeling complex represses early meiotic genes upon
recruitment by Ume6p. Cell 103, 423–433.
Harris, M.B., Mostecki, J., Rothman, P.B., 2005. Repression of an interleukin-4-
responsive promoter requires cooperative BCL-6 function. J. Biol. Chem.
Huang, R.C., Bonner, J., 1962. Histone, a suppressor of chromosomal RNA
synthesis. Proc. Natl. Acad. Sci. U. S. A. 48, 1216–1222.
Hughes, K., Wand, M., Foulston, L., Young, R., Harley, K., Terry, S., Ersfeld,
K., Rudenko, G., 2007. A novel ISWI is involved in VSG expression site
downregulation in African trypanosomes. EMBO J. 26, 2400–2410.
Kambris, Z., Brun, S., Jang, I.H., Nam, H.J., Romeo, Y., Takahashi, K., Lee,
W.J., Ueda, R., Lemaitre, B., 2006. Drosophila immunity: a large-scale in
vivo RNAi screen identifies five serine proteases required for Toll
activation. Curr. Biol. 16, 808–813.
Kouzarides, T., 2007. Chromatin modifications and their function. Cell 128,
Li, J., Langst, G., Grummt, I., 2006. NoRC-dependent nucleosome positioning
silences rRNA genes. EMBO J. 25, 5735–5741.
Li, B., Carey, M., Workman, J.L., 2007. The role of chromatin during
transcription. Cell 128, 707–719.
Liu, B., Liao, J., Rao, X., Kushner, S.A., Chung, C.D., Chang, D.D., Shuai, K.,
1998. Inhibition of Stat1-mediated gene activation by PIAS1. Proc. Natl.
Acad. Sci. U. S. A. 95, 10626–10631.
Luo, H., Hanratty, W.P., Dearolf, C.R., 1995. An amino acid substitution in the
Drosophila hopTum-l Jak kinase causes leukemia-like hematopoietic
defects. EMBO J. 14, 1412–1420.
Muller, P., Kuttenkeuler, D., Gesellchen, V., Zeidler, M.P., Boutros, M., 2005.
Identification of JAK/STAT signalling components by genome-wide RNA
interference. Nature 436, 871–875.
Nakajima, H., Brindle, P.K., Handa, M., Ihle, J.N., 2001. Functional interaction
of STAT5 and nuclear receptor co-repressor SMRT: implications in negative
regulation of STAT5-dependent transcription. EMBO J. 20, 6836–6844.
Ni, Z., Bremner, R., 2007. Brahma-related gene 1-dependentSTAT3 recruitment
at IL-6-inducible genes. J. Immunol. 178, 345–351.
Rizki, M.T.M., 1957. Alterations in the hemocyte population of Drosophila
melanogaster. J. Morphol. 100, 437–458.
Ruiz,C.,Escribano,V., Morgado,E.,Molina,M., Mazon,M.J.,2003.Cell-type-
dependent repression of yeast a-specific genes requires Itc1p, a subunit of
the Isw2p–Itc1p chromatin remodelling complex. Microbiology 149,
Saha, A., Wittmeyer, J., Cairns, B.R., 2006. Chromatin remodelling: the
industrial revolution of DNA around histones. Nat. Rev., Mol. Cell Biol. 7,
Satchwell, S.C., Drew, H.R., Travers, A.A., 1986. Sequence periodicities in
chicken nucleosome core DNA. J. Mol. Biol. 191, 659–675.
Segal, E., Fondufe-Mittendorf, Y., Chen, L., Thåström, A., Field, Y., Moore,
I.K., Wang, J.P., Widom, J., 2006. A genomic code for nucleosome
positioning. Nature 442, 772–778.
Shi, S., Calhoun, H.C., Xia, F., Li, J., Le, L., Li, W.X., 2006. JAK signaling
globally counteracts heterochromatic gene silencing. Nat. Genet. 38,
Sinenko, S.A., Kim, E.K., Wynn, R., Manfruelli, P., Ando, I., Wharton, K.A.,
Perrimon, N., Mathey-Prevot, B., 2004. Yantar, a conserved arginine-rich
protein is involved in Drosophila hemocyte development. Dev. Biol. 273,
Starr, R., Willson, T.A., Viney, E.M., Murray, L.J., Rayner, J.R., Jenkins, B.J.,
Gonda, T.J., Alexander, W.S., Metcalf, D., Nicola, N.A., Hilton, D.J., 1997.
A family of cytokine-inducible inhibitors of signalling. Nature 387,
Stedman, E., Stedman, E., 1950. Cell specificity of histones. Nature 166,
Whitehouse, I., Tsukiyama, T., 2006. Antagonistic forces that position
nucleosomes in vivo. Nat. Struct. Mol. Biol. 13, 633–640.
Xiao, H., Sandaltzopoulos, R., Wang, H.M., Hamiche, A., Ranallo, R., Lee,
K.M., Fu, D.G., Wu, C., 2001. Dual functions of largest NURF subunit
NURF301 in nucleosome sliding and transcription factor interactions.
Mol. Cell 8, 531–543.
Yee, G.H., Hynes, R.O., 1993. A novel tissue-specific integrin subunit β-v
expressed in the midgut of Drosophila melanogaster. Development 118,
547S.Y. Kwon et al. / Developmental Biology 316 (2008) 538–547