Genome-wide analysis reveals conserved and
divergent features of Notch1/RBPJ binding in
human and murine T-lymphoblastic leukemia cells
Hongfang Wanga,1, James Zoub,1,2, Bo Zhaoc, Eric Johannsenc, Todd Ashwortha, Hoifung Wonga, Warren S. Peard,
Jonathan Schuge, Stephen C. Blacklowf, Kelly L. Arnettf, Bradley E. Bernsteing, Elliott Kieffc,2, and Jon C. Astera,2
aDepartment of Pathology andcDepartments of Medicine and Microbiology and Molecular Genetics, Brigham and Women’s Hospital, Boston, MA 02115;
bSchool of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138;dDepartment of Pathology andeDepartment of Genetics,
University of Pennsylvania Medical School, Philadelphia, PA 19104;fDepartment of Cancer Biology, The Dana Farber Cancer Institute, Boston, MA 02115;
andgThe Howard Hughes Medical Institute, Department of Pathology, Massachusetts General Hospital, Boston, MA 02114
Contributed by Elliott Kieff, June 6, 2011 (sent for review March 31, 2011)
Notch1 regulates gene expression by associating with the DNA-
binding factor RBPJ and is oncogenic in murine and human T-cell
progenitors. Using ChIP-Seq, we find that in human and murine
T-lymphoblastic leukemia (TLL) genomes Notch1 binds preferentially
to promoters, to RBPJ binding sites, and near imputed ZNF143, ETS,
and RUNX sites. ChIP-Seq confirmed that ZNF143 binds to ∼40% of
Notch1 sites. Notch1/ZNF143 sites are characterized by high Notch1
and ZNF143 signals, frequent cobinding of RBPJ (generally through
sites embedded within ZNF143 motifs), strong promoter bias, and
relatively low mean levels of activating chromatin marks. RBPJ and
ZNF143 binding to DNA is mutually exclusive in vitro, suggesting
RBPJ/Notch1 and ZNF143 complexes exchange on these sites in
cells. K-means clustering of Notch1 binding sites and associated
motifs identified conserved Notch1-RUNX, Notch1-ETS, Notch1-
RBPJ, Notch1-ZNF143, and Notch1-ZNF143-ETS clusters with differ-
ent genomic distributions and levels of chromatin marks. Although
Notch1 binds mainly to gene promoters, ∼75% of direct target
genes lack promoter binding and are presumably regulated by
enhancers, which were identified near MYC, DTX1, IGF1R, IL7R,
and the GIMAP cluster. Human and murine TLL genomes also have
many sites that bind only RBPJ. Murine RBPJ-only sites are highly
enriched for imputed REST (a DNA-binding transcriptional repres-
sor) sites, whereas human RPBJ-only sites lack REST motifs and are
more highly enriched for imputed CREB sites. Thus, there is a con-
served network of cis-regulatory factors that interacts with Notch1
to regulate gene expression in TLL cells, as well as unique classes of
divergent RBPJ-only sites that also likely regulate transcription.
Signaling is mediated through a series of proteolytic cleavages,
the last carried out by γ-secretase, that permit the Notch in-
tracellular domain (NICD) to translocate to the nucleus and
form a transcriptional activation complex with the DNA-binding
factor RBPJ [also known as CSL in vertebrates and Su(H) in
flies; for recent review, see ref. 1]. This complex in turn recruits
a Mastermind (MAML) family member, other coactivators, and
mediator complex components, leading to regulated target gene
expression. NICD-containing transcription activation complexes
turn over rapidly because of transcription-coupled degradation.
In the absence of NICD, RBPJ binds several transcriptional
corepressors, supporting models in which RBPJ acts like a tran-
scriptional switch. However, genome-wide studies testing this
model have yet to be done in higher organisms.
Notch effects vary with dosage and the chromatin state of the
signal-receiving cell. Dysregulated Notch signaling underlies
certain developmental disorders and cancers, most notably
T-lymphoblastic leukemia/lymphoma (TLL), in which most tumors
have somatic Notch1 gain-of-function mutations that increase
otch receptors participate in a highly conserved signaling
pathway that regulates development and tissue homeostasis.
NICD1 levels (for review, see ref. 2). Dysregulated RBPJ-
dependent gene expression is also seen in other cancers, par-
ticularly B-cell tumors associated with Epstein-Barr virus (EBV)
(3–6). However, activating Notch mutations are virtually unique
to TLL, indicating an unusual susceptibility of T-cell progenitors
to transformation by Notch.
Recent data indicate that transcription factors bind widely
throughout genomes and that individual transcription-factor
binding sites show substantial evolutionary divergence (7, 8). To
identify conserved relationships of likely fundamental impor-
tance, as well as potentially interesting points of evolutionary
divergence, we performed ChIP deep-sequencing (ChIP-Seq) for
Notch1 and RBPJ in human and murine TLL cell lines.
Notch1 and RBPJ Binding Sites in Human TLL Cell Genomes. In initial
studies, duplicate ChIP-Seq libraries were prepared from the
Notch1-dependent human TLL cell line CUTLL1 (9) with RBPJ
and Notch1 antisera. Analysis of pooled data identified 3,846
Notch1 binding sites and 2,112 RBPJ binding sites [false-
discovery rate (FDR) <0.01] (Fig. 1A). Only 36% of Notch1
peaks overlapped with RBPJ peaks, and 66% of RBPJ peaks
overlapped with Notch1 peaks. Most Notch1 and RBPJ binding
sites were in promoters [regions within 2 kb of annotated tran-
scriptional start sites (TSS)]. In line with studies in Drosophila
suggesting RBPJ is stabilized on genomic DNA by NICD (10),
RBPJ and Notch1 ChIP-Seq signals were higher at sites where
both bound (Fig. 1B) (P < 10−6for both comparisons).
Transcription Factor Motifs Enriched at Sites of Notch1 and RBPJ
Binding in Human TLL Cells. To determine factors that influence
genomic Notch1 binding, recombinant RBPJ protein-binding
microarray (PBM) data (11) were used to identify the highest
affinity RBPJ binding site within 100 bp of the center of each
Notch1 binding site. Both Notch1/RBPJ and Notch1-only sites
were equally enriched for high-affinity RBPJ binding sequences
Author contributions: H. Wang, B.Z., E.K., and J.C.A. designed research; H. Wang, B.Z.,
T.A., H. Wong, and K.L.A. performed research; E.J., W.S.P., J.S., S.C.B., and B.E.B. contrib-
uted new reagents/analytic tools; H. Wang, J.Z., B.Z., B.E.B., E.K., and J.C.A. analyzed data;
H. Wang, J.Z., E.K., and J.C.A. wrote the paper.
The authors declare no conflict of interest.
Data deposition: The sequences reported in this paper have been deposited in the Gene
Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession nos.
GSE29544 and GSE29600).
See Commentary on page 14715.
1H. Wang and J.Z. contributed equally to this work.
2To whom correspondence may be addressed. E-mail: firstname.lastname@example.org, ekieff@rics.
bwh.harvard.edu, or email@example.com.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| September 6, 2011
| vol. 108
| no. 36www.pnas.org/cgi/doi/10.1073/pnas.1109023108
compared with random genomic sequences (P < 10−10), suggesting
that many Notch1-only sites also bind RBPJ. The promoter
localizations of Notch1/RBPJ and Notch1-only sites were similar
(63% and 59%, respectively). Failure to detect RBPJ at Notch1
sites with RBPJ consensus sequences may stem from shielding of
RBPJ epitopes. Other Notch1-only sites may result from Notch1
binding to other chromatin-associated proteins or be an artifact of
Notch1 cross-linking via long-range chromatin loops.
Further clues came from a search for transcription factor
motifs enriched within 250 bp of RBPJ and Notch1 binding sites.
For Notch1 sites, the most enriched motif (compared with
overall genomic frequency) was that of ZNF143, followed by
those for ETS and RUNX factors (P < 10−50for each) (Fig. 1C).
ZNF143 was also the most enriched motif associated with RBPJ
binding sites, followed by the motifs for CREB, ETS factors, and
RUNX factors (P < 10−50) (Fig. 1C). Of note, the ZNF143
consensus motif contains an embedded high-affinity RBPJ site.
Overall, CREB consensus motifs were found in association with
46% of the RBPJ-only sites, compared with 25% of Notch1/
RBPJ cosites (P < 10−6) (Fig. S1A). Compared with RBPJ sites
without CREB motifs, the 588 RBPJ/CREB sites had lower-
affinity RBPJ binding sites (P < 10−10) (Fig. S1B), but higher RBPJ
signals (Fig. S1C) (P < 10−10), suggesting determinants other
than DNA binding affinity (e.g., protein–protein interactions)
contribute to RBPJ association with imputed CREB sites.
Confirmation of ZNF143 Association with RBPJ and Notch1 Binding
Sites. Western blotting detected NICD1, RBPJ, ZNF143, the
ETS factors GABPA and ETS1, RUNX1, and CREB in three
human and two murine Notch1-dependent TLL cell lines (Fig.
S2A). In local ChIP assays of randomly selected Notch1 binding
elements with imputed ZNF143, GABPA, ETS1, or RUNX1
sites, 17 of 17 tested sites showed binding by each factor (Fig.
S2B), suggesting most imputed sites are occupied by the imputed
ETS factors and RUNX factors bind widely to Jurkat TLL cell
genomes (12, 13), are required during early stages of T-cell de-
velopment (14), and interact with Notch1 in early hematopoiesis
(15), but little is known about ZNF143. ZNF143 ChIP-Seq was
thus carried out in CUTLL1 cells, resulting in identification of
3,684 ZNF143 binding sites (FDR < 0.01) (Fig. 2A). ZNF143
also showed a bias for promoters, particularly bidirectional
promoters, 285 of which bound ZNF143 in CUTLL1 cells. The
most highly enriched motif near ZNF143 peaks was that of
ZNF143 (Fig. 2B), followed by NFY1, ETS, and YY1, all of
which also often bind bidirectional promoters (16, 17). Re-
markably, 40% (n = 1,551) of Notch1 peaks lay within 100 bp
of ZNF143 peaks, and 14% (n = 544) overlay RBPJ/Notch
“copeaks” (Fig. 2A). Of ZNF143 sites overlapping with Notch1
peaks, 67% (n = 1,044) contained the ZNF143 consensus motif,
and of these 57% (n = 591) had an embedded high-affinity RBPJ
binding site. ZNF143 signals were higher at sites where Notch1
bound, as were Notch1 signals at sites of ZNF143 binding (Fig.
2C) (P < 10−100for each). Consistent with these associations,
Notch1 and ZNF143 signals correlated at cosites (R2= 0.66),
with cosites with significant RBPJ signals having the highest
Notch1/ZNF143 signals (Fig. 2D).
On average the positions of ZNF143 and Notch1 copeaks co-
incided exactly (within 2 bp), closely approximating the overlap of
RBPJ and Notch1 copeaks (Fig. 2E). This finding suggests that
Notch1 associates with ZNF143 cosites via RBPJ, which could
either co-occupy sites with ZNF143 or exchange with ZNF143 on
chromatin. Consistent with the latter possibility, binding of RBPJ
was mutually exclusive (Fig. 2F), suggesting that in cells RBPJ/
Notch1 and ZNF143 complexes bind sequentially rather than si-
multaneously. It follows that the correlation of ZNF143 and
Notch1 signal strength (Fig. 2D) is a function of binding site ac-
cessibility (e.g., chromatin “openness”) rather than cobinding of
ZNF143 and Notch1 to these sites.
Identification of Distinct Classes of Notch1 and RBPJ Binding Sites. K-
means clustering of all 3,846 Notch binding sites revealed five
major clusters named RUNX, ETS, ZNF-ETS, ZNF, and RBPJ
for the predominant cofactors (Fig. 3A). The RUNX cluster
showed a nonpromoter bias, whereas the ZNF143 and ZNF143-
ETS clusters had marked promoter biases and high Notch1 sig-
nals (Fig. 3A). To further characterize these clusters, Chip-Seq
for activating H3K4me1 (enhancer) and H3K4me3 (promoter)
histone marks and repressive H3K27me3 marks was done using
CUTLL1 cells. The clusters showed subtle but statistically sig-
nificant differences in H3K4me3 signals at promoters, with the
ETS and RUNX clusters having highest marks (Fig. 3B) (P <
0.01) and the greatest nucleosome displacement (Fig. 3C), and
the ZNF143 cluster having the lowest. Consistent with these
associations, mean-expression levels of genes with Notch1 pro-
moter binding were highest for the ETS cluster and lowest for
the ZNF cluster (Fig. 3B) (P < 0.01). More striking differences
were observed at nonpromoter Notch1 binding sites, with the
ETS and RUNX clusters having the highest H3K4me1 signals
and greatest nucleosome displacement and the RBPJ, ZNF143-
ETS, and ZNF143 clusters the lowest (Fig. 3C) (P < 0.001).
The chromatin marks of ZNF143-only and RBPJ-only (no
Notch1) binding sites were also assessed. Promoter and inter-
genic ZNF143 sites without Notch1 binding had lower active
chromatin marks (Fig. S3 A and B) (P < 0.0001 for both com-
parisons) and higher repressive marks (H3K27me3) (Fig. S3C)
(P < 10−100), suggesting ZNF143 associates with repressive
complexes. Similarly, RBPJ-only sites had lower intergenic H3-
K4me1 and promoter H3K4me3 signals (Fig. S3 D and E)
(P < 0.05), and of these RBPJ-only sites, those with CREB
motifs had lower intergenic H3K4me1 and promoter H3K4me3
signals than those without (Fig. S3 D and E) (P < 0.05). These
and overlap of RBPJ and Notch1 binding peaks in CUTLL1 cells. Promoter
regions are defined as sequences within 2 kb of the TSS of annotated genes.
(B) Notch1 and RBPJ ChIP-Seq signals (expressed as reads per kilobase per
million aligned reads, RPKM) at Notch1-only sites, RBPJ-only sites, and RBPJ/
Notch1 cosites. (C) Enriched transcription factor motifs lying ±250 bp of
Notch1 and RBPJ binding sites. A RBPJ consensus motif embedded within the
extended ZNF143 motif is boxed.
Notch1 and RBPJ binding sites in human TLL cells. (A) Distribution
Wang et al.PNAS
| September 6, 2011
| vol. 108
| no. 36
findings are consistent with a repressive role for RBPJ in the
absence of NICD1.
Genomic Notch1 Binding Sites and Target Gene Regulation. To
identify robust direct Notch1 target genes in CUTLL1 cells, we
used a γ-secretase-inhibitor (GSI) washout method (18) that
permits Notch1 reactivation in the presence of cycloheximide.
High-confidence direct canonical Notch1 target genes were de-
fined by a twofold or greater increase in expression within 4 h of
GSI washout that was insensitive to cycloheximide and sensitive
to dominant-negative MAML1, a specific inhibitor of canonical
Notch1 signaling. Two-hundred forty-five genes met these cri-
teria (Dataset S1), including previously identified target genes,
such as HES1, HES4, HES5, DTX1, and MYC (18, 19). Notch1
bound the promoters of 61 (25%) of these genes (Fig. S4), an
enrichment (P < 10−4, binomial test) over the total fraction of
genes with Notch1 binding to their promoters (2,325 of 15,340
TLL cells. (A) Distribution of ZNF143 binding peaks and overlap with RBPJ
and Notch1 sites in CUTLL1 cells. (B) Enriched transcription factor motifs ly-
ing ±250 bp of ZNF143 binding sites. (C) ZNF143 and Notch1 ChIP-Seq signals
(expressed as RPKM) at ZNF143 sites with and without Notch1, and Notch1
promoter (Pro), and nonpromoter sites with and without ZNF143. (D) Cor-
relation of Notch1 and ZNF143 signals on individual cosites sites with or
without RBPJ signals. (E) Overlap of Notch1 sites with ZNF143 and RBPJ
binding sites in CUTLL1 cells, expressed as the distance in base pairs between
the centers of Notch1 and ZNF143 copeaks (Left) and Notch1 and RBPJ
copeaks (Right). (F) EMSA done with an oligonucleotide containing the SKP2
promoter ZNF143/Notch1 cosite. ZNF143 and RBPJ were added at concen-
trations of 280 and 407 nM, respectively. In one lane, an unlabeled probe
was added in 200-fold excess over the labeled probe (4 μM:20 nM).
Frequent association of ZNF143 and Notch1 binding sites in human
distinct classes of putative response elements in TLL cells. (A) K-means clus-
tering of Notch1 binding sites in CUTLL1 cells. In the clusters shown at the
left, each red line represents a genomic Notch1 binding site where cobinding
(RBPJ, Znf143) or a motif (ETS, RUNX, CREB) for the indicated factor is seen.
The numbers at the right are the fraction of binding sites within each cluster
that are found within promoters of annotated genes and the associated
Notch1 and RBPJ ChIP-Seq signal strengths (RPKM). (B) Mean histone
H3K4me3 signals (Left) and relative gene expression (Right) for genes with
Notch1 promoter binding by cluster designation. Normalized gene expres-
sion was determined by expression profiling of CUTLL1 cells. (C) Distribution
of H3K4me3 signals in Notch1-binding promoters (Left) and H3K4me1 sig-
nals in Notch1-binding intergenic sites (Right) by cluster designation. “0”
marks the center of the associated Notch1-binding peaks.
Transcription factors associated with Notch1 binding sites define
| www.pnas.org/cgi/doi/10.1073/pnas.1109023108 Wang et al.
genes screened, 15.1%). The remaining target genes are pre-
sumably regulated through enhancers, a possibility consistent
with the presence of at least one Notch1 binding site within
100 kb of the promoters of 127 of 179 target genes (69%) lacking
Notch1 promoter binding.
For each target gene, the Notch binding site closest to the
transcriptional start site was designated the most likely regula-
tory element (summarized in Dataset S1). RBPJ binding was
detected by ChIP-Seq at 52% of these sites, an enrichment over
the 36% overlap between Notch1 and RBPJ binding genome-
wide (P < 10−6, binomial test). The 179 target genes without
promoter binding showed larger expression changes in response
to Notch1 than the 66 target genes that did (P < 0.05), indicating
that Notch1-responsive enhancers are relatively potent inducers
of target gene expression.
RBPJ/Notch1 binding sites near some target genes merit com-
ment (Fig. S5). HES1, a gene involved in TLL and normal T-cell
development (20), has two high-confidence and one low-confi-
dence binding peaks. HES1 is unusual among target genes in
CUTLL1 cells in having high levels of bivalent H3K4me3 and
H3K27me3 chromatin marks, a combination that in embryonic
stem cells defines genes that are poised to respond to differenti-
ation cues (21); the significance of this chromatin structure
near HES1 in TLL cells remains to be determined. Candidate
enhancers (defined by high levels of H3K4me1 marks) were
identified near other direct Notch1 target genes, including: the
clustered GIMAP genes, which have been implicated in TLL cell
survival (22); DTX1; and IL7R, IGF1R, and MYC, all of which
contribute to Notch1-dependent TLL cell growth (18, 19, 23, 24).
Local ChIP confirmed Notch1/RBPJ binding to the candidate
MYC enhancer, which also bound p300, CREB binding protein,
and RNA polymerase II (Fig. S6A). Although a promoter binding
site for RBPJ/Notch1 has been reported in IL7R in CUTLL1 cells
(23), we observed no binding to the IL7R promoter in CUTLL1
cells by ChIP-Seq, whereas local ChIP confirmed binding of
human TLL lines (Fig. S6B). Similarly, Notch1 binding to a candi-
Conservation of Factors Associated with Notch1 Binding Sites and
Divergence of Factors Associated with RBPJ-Only Binding Elements.
To determine conserved aspects of RBPJ/Notch1 interactions
with TLL genomes, ChIP-Seq analyses were done on two Notch1-
dependent murine T-acute lymphoblastic leukemia cell lines,
T6E (25) and G4A2 (26). T6E cell contained 1,587 Notch1 sites
and 2,776 RBPJ sites (FDR < 0.01) (Fig. 4A). The Notch1
binding-site distribution was similar to that in CUTLL1 cells,
with a bias for promoters. Furthermore, the transcription factor
motifs associated with T6E cell Notch1 binding sites were the
same as in CUTLL1 cells and in the same rank order: ZNF143,
ETS, and RUNX. The fraction of Notch sites overlapping with
each cofactor was almost identical in mouse and human (Fig.
4B), and K-means clustering identified the same five classes of
Notch1 binding clusters (Fig. 4C). For unclear reasons, fewer
high-confidence Notch1 and RBPJ binding sites (548 and 668,
respectively) were identified in G4A2 cells (Fig. S7), but the
same associations were found.
In contrast to the conservation of motifs and cofactor com-
binations, Notch1 binding to orthologous elements was quite
divergent. By strict criteria (see Materials and Methods), of the
245 direct Notch1 target genes identified in human CUTLL1
cells, only 50 genes (20.4%) showed Notch1 binding to orthol-
ogous elements in murine T6E cells (Dataset S1). Well-charac-
terized target genes with conserved Notch1 binding sites include
HES1 and HES5, and divergent Notch1 binding sites were ob-
served in other conserved target genes, such as DTX1. Conserved
RBPJ binding to orthologous elements was highly associated
with conserved Notch1 binding (Fisher’s test, P < 10−16), and
conserved imputed ZNF143, ETS, and CREB sites showed
smaller significant associations (Table S1).
Unexpectedly, the distribution of RBPJ binding sites was highly
divergent in human and murine TLL cells. Only 36% of RBPJ
sites were in promoters and only 35% of RBPJ sites overlapped
with Notch1 sites in murine T6E cells (Fig. 4A); similar dis-
tributions were observed in murine G4A2 cells (Fig. S7). The
motif most highly enriched near RBPJ binding sites in T6E cells
(Fig. 4D) and G4A2 cells was that of REST, a DNA-binding
transcriptional repressor (27), followed by the motifs for CREB
and ETS. Overall, 895 of 2,776 (32%) RBPJ binding sites in T6E
sites were outside of promoters and none bound Notch1; thus,
these sites explain the divergence in RBPJ binding distributions
between human and murine TLL genomes. Analyses using PBM
data showed that RBPJ/REST sites generally lack high-affinity
RBPJ binding motifs (Fig. S8A), suggesting that RBPJ binding to
these sites involves protein-protein interactions. In contrast, only
6 of 2,112 (0.3%) of the RBPJ sites in human CUTLL1 cells were
associated with REST motifs. To confirm REST binds CUTLL1
cell genomes, local ChIP for REST was done on four conserved
RBPJ/REST sites identified in T6E cells. All four sites showed
of 2,319 REST and 10,529 RBPJ binding sites identified by ChIP-
seq [FDR <0.01 (28)] in EBV-transformed lymphoblastoid cell
lines, only 12 (0.1%) overlapped. Thus, although REST binding
sites may be conserved, RBPJ/REST colocalization in genomes
may be restricted to murine cells.
in murine and human TLL cells. (A) Distribution and overlap of RBPJ and
Notch1 binding sites in the genome of the murine TLL cell line T6E. Promoter
regions are defined as sequences within 2 kb of the TSS of annotated genes.
(B) Conservation of the proportions of Notch1 binding sites associated with
ZNF143, ETS, RUNX, and CREB motifs in the genomes of human CUTLL1 cells
and murine T6E cells. The fraction of Notch1 binding sites that contain the
indicate motif ±250 bp of Notch1 peaks is shown. (C) Conservation of RUNX,
ETS, RBPJ, ZNF143-Ets, and ZNF143 clusters in murine T6E cells. (D) Enriched
transcription factor motifs lying ±250 bp of RBPJ peaks in murine T6E cells.
(E) Divergence of REST and conservation of CREB motifs near RBPJ peaks in
murine T6E cells and human CUTLL1 cells. The fraction of RBPJ binding sites
that contain the indicated motif ±250 bp of RBPJ peaks is shown.
Conserved and divergent features of Notch1 and RBPJ binding sites
Wang et al.PNAS
| September 6, 2011
| vol. 108
| no. 36
A second interspecies difference in RBPJ-only sites was the
presence of a higher fraction of RBPJ/CREB sites in human TLL
elements (Table S2) (Fisher’s test P < 10−16). Thus, unlike RBPJ/
REST sites, there appears to be selective pressure favoring conser-
vation of RBPJ-only sites associated with CREB motifs.
Analysis of RBPJ/Notch1 interactions with TLL cell genomes
identified conserved features with implications for transcrip-
tional regulation as well as unexpected points of evolutionary
divergence. Although Notch1 binds predominantly to promoters,
promoter binding is a poor predictor of Notch1 responsiveness;
all told, 97% of genes with Notch1 promoter binding do not
respond robustly when Notch1 is activated. This finding is in line
with studies in yeast, which found only 3% overlap between
promoter occupancy and response to transcription factor per-
turbations (29). Similar findings have been noted in Drosophila
(30). In nonresponsive genes with promoters bound by Notch1,
loss of Notch1 may be compensated by other transcription fac-
tors, or Notch1 might only affect gene expression when other
signaling pathways are also perturbed. The only positive pre-
dictor of Notch1 responsiveness in TLL cells was RBPJ cobin-
ding, but only a small subset of genes with RBPJ/Notch1 pro-
moter binding are Notch1 responsive, and other determinants
must exist. The ChIP-Seq studies described here also revealed
a large number of candidate Notch1-binding enhancers, which
presumably regulate target genes that lack Notch1 promoter
binding. Additional studies using methods such as chromosome
conformation capture will be necessary to link these elements to
target gene regulation.
Another variable likely to influence the Notch1 responsiveness
of genes are cofactors that bind nearby. We identified a signifi-
cant enrichment for overlapping ZNF143 binding sites and ETS
and RUNX factor motifs near Notch1 binding sites in human
and murine TLL cells. These cofactors occur in nonrandom
combinations, pointing to a conserved cis-regulatory transcrip-
tion factor network that forms the basis for a new understanding
of Notch1 function in this cellular context. ETS factors are in-
volved in T-cell specification and development (31, 32). Simi-
larly, RUNX factors interact with Notch1 during hematopoietic
stem cell development (15, 33), act upstream of Notch1 in early
T-cell development (14), and (when deficient) predispose mice
to TLL (34). An open question in understanding Notch1 func-
tion in T-cell progenitors is the identity of factors that “condi-
tion” the epigenome, allowing NICD1 access to regulatory
elements that control T-cell development or, which when over-
active, contribute to TLL. ETS and RUNX family members, by
virtue of their association with Notch1 binding sites in TLL
genomes, are candidates for such roles.
Unlike ETS and RUNX factors, ZNF143 has no known role in
T-cell development, TLL, or Notch signaling. ZNF143 is a ubiq-
uitously expressed zinc finger protein of unknown function origi-
nally identified as a regulator of selenomethionine tRNA gene
expression (35). Our ChIP-Seq analyses confirm prior studies
suggesting that ZNF143 localizes to promoters (36), particularly
those driving transcription bidirectionally (17). Gene pairs regu-
lated by bidirectional promoters are highly conserved in verte-
brates (37);whether this is a consequence of convergent functions
sequence adjacent toa promoter with bidirectional activity during
early vertebrate evolution is uncertain. ZNF143 motifs also ap-
pear to be enriched near hepatocyte SREBP-1 binding sites (38),
suggesting ZNF143 is a general factor with broad functions. An
issue emerging from our work is whether sites where RBPJ/
Notch1/ZNF143 colocalize load ternary complexes or are instead
occupied sequentially by RBPJ/Notch1 and ZNF143 complexes.
In vitro studies suggest that binding of these factors to ZNF143
motifs is mutually exclusive, supporting a complex exchange
model, but further work is needed to confirm this relationship in
cells and to understand its basis.
Finally, we have identified major classes of genomic RBPJ sites
that fail to bind Notch1 or do so very inefficiently. Both murine
and human TLL genomes contain numerous RBPJ-only binding
sites enriched for CREB motif that have relatively high levels of
repressive chromatin marks, tend to flank genes with low expres-
sion levels, and often lack high-affinity RBPJ binding motifs. A
second RBPJ-only site is associated with motifs for REST, a co-
repressor best known for its role in regulating neurogenesis (27)
simplex virus latency (39). RBPJ/REST sites are common in mu-
rine TLL cells, but essentially undetectable in human TLL cells or
EBV-transformed human B cells, presumably because of di-
vergence in factors that mediate recruitment of RBPJ to REST
sites. One important Notch-independent RBPJ-dependent func-
tion in mammals involves the association of RBPJ with the tran-
independent RBPJ functions exist and that the RBPJ “transcrip-
in higher vertebrates.
Materials and Methods
Cell Culture. Cell lines were cultured in RPMI medium 1640 (Invitrogen)
supplemented with 10% FBS, 2 mM L-glutamine, penicillin, and streptomycin;
T6E cell medium was additionally supplemented with 1 mM sodium pyruvate
and 5 μM 2-mercaptoethanol.
ChIP-Seq. ChIP was done using the ChIP assay kit protocol (Millipore). ChIP-seq
libraries were constructed per the Illumnia ChIP DNA library preparation kit.
High-throughput sequencing was performed using the Illumina Genome
Analyzer II. Reads were mapped to human genome hg18 or mouse genome
mm9. Mapped reads ranged from 1 × 107to 3.4 × 107per library. ChIP-seq
data (accession# GSE29600) are available through the Gene Expression
Local ChIP. Antibodies used for local ChIP assays are listed below. After DNA
purification, real-time PCR was performed in triplicate on the CFX96 Real-
Time PCR System (Bio-Rad), using input DNA as a standard and preimmune or
nonimmune Ig as a control. Enrichment was estimated using the ΔΔCT
method. Primer sequences used in local ChIP assays are available on request.
RNA Preparation and Gene-Expression Profiling. Induction of Notch1 signaling
through GSI washout was performed as previously described (18); additional
details are available in SI Materials and Methods. Total RNA prepared with
RNeasy Mini kit (Qiagen) was subjected to expression profiling in triplicate
on Affymetrix human genome U133 plus 2.0 arrays. Affymetrix expression
data were normalized using the robust multichip average package in R.
Gene-expression profiling data (accession number GSE29544) are available
through the Gene Expression Omnibus.
Analysis of ChIP-Seq Data ChIP-Seq data were processed using QUEST (41).
High-confidence peaks were defined by ChIP enrichment of ≥ threefold and
FDR < 0.01. The FDR was determined by running the same peak calling
analysis on a randomly selected subset of the input ChIP-seq data. Peaks for
two transcription factors were called overlapping when the centers of the
peaks were within 100 bp of each other. All binding signals were normalized
to reads per kilobase per million mapped. Binding sites in the mouse ge-
nome (version mm9) were mapped to orthologous sequences in the human
genome (version hg18) using the University of California Santa Cruz Liftover
tool. A site was called conserved if Notch1 binding was present in both
species within 100 bp of the center of the binding peak in CUTLL1 cells.
Motif Analysis and Clustering. Enriched motifs within 250 bp of the center of
each binding peak were determined using HOMER (42) in the default setting.
Controls were genomic sequences selected at random with DNA percent GC
content similar to the binding peaks. The hypergeometric test was used to
quantify enrichment of motifs in DNA regions adjacent to binding peaks. For
cluster analysis, each binding site was represented as a vector defined by the
| www.pnas.org/cgi/doi/10.1073/pnas.1109023108Wang et al.
presence (i) or absence (ii) of a given cofactor (based on ChIP-Seq) or binding
PBM Analysis. PBM data were obtained from Del Bianco et al. (11). Each 8-mer
sequence was associated with a z-score based on RBPJ affinity. The 100-bp
region around each Notch1 and RBPJ binding site were scanned for the
motif with the highest z-score.
Antibodies. Antibodies used in these studies are given in SI Material
Statistical Methods. Unless otherwise specified, statistical comparisons were
performed and P values were calculated using the Mann-Whitney U test.
Electrophoretic Mobility Shift Assay. Recombinant RBPJ was prepared as pre-
viouslydescribed (43). HumanZNF143was expressed inEscherichiacoliRosetta
(de3)pLys cells and purified by sequential chromatography on Ni-NTA affinity
matrix followed by a MonoQ ion exchange column. An oligonucleotide con-
taining the SKP2 promoter ZNF143/Notch1 cobinding site (GGAAAACTA-
CAATTCCCAGCTTGGGCCTAGC) (ZNF143 site underlined, embedded RBPJ site
in bold) was end-labeled with32P and mixed at a final concentration of 20 nM
with 200 ng of ZNF143 (280 nM) or 200 ng RBPJ (407 nM) in 20 mM Hepes, pH
7.9, containing 60 mM KCl, 10 mM DTT, 5 mM MgCl2, 10 μg/mL poly-dIdC, and
0.2 mg/mL albumin. After 30 min at 30 °C, the mixtures were loaded onto
a 10% native Tris-glycine gel, separated by electrophoresis, and analyzed
ACKNOWLEDGMENTS. We thank Jeremiah Huang for technical assistance
and the University of Pennsylvania Diabetes and Endocrinology Center for
the use of the Functional Genomics Core (P30-DK19525). This work was
supported in part by National Institutes of Health Grant P01CA119070 and
a Specialized Center of Research grant from the Leukemia and Lymphoma
Society (to J.C.A., S.C.B., and W.S.P.); The Howard Hughes Medical Institute,
the National Human Genome Research Institute, and the National Institutes
of Health Epigenome Mapping Centers Consortium, Encyclopedia Of DNA
Elements Grant U54 HG004570, and Roadmap Epigenomics Grant U01
ES017155 (to B.E.B.); National Institutes of Health Grants R01CA047006,
R01CA131354, and R01 CA085180 (to E.K.); and a National Science Founda-
tion Graduate Student Fellowship (to J.Z.).
1. Kopan R, Ilagan MX (2009) The canonical Notch signaling pathway: Unfolding the
activation mechanism. Cell 137:216–233.
2. Aster JC, Blacklow SC, Pear WS (2011) Notch signalling in T-cell lymphoblastic
leukaemia/lymphoma and other haematological malignancies. J Pathol 223:262–273.
3. Henkel T, Ling PD, Hayward SD, Peterson MG (1994) Mediation of Epstein-Barr virus
EBNA2 transactivation by recombination signal-binding protein J kappa. Science 265:
4. Hsieh JJ, Hayward SD (1995) Masking of the CBF1/RBPJ kappa transcriptional
repression domain by Epstein-Barr virus EBNA2. Science 268:560–563.
5. Robertson ES, Lin J, Kieff E (1996) The amino-terminal domains of Epstein-Barr virus
nuclear proteins 3A, 3B, and 3C interact with RBPJ(kappa). J Virol 70:3068–3074.
6. Lee S, et al. (2009) Epstein-Barr virus nuclear protein 3C domains necessary for
lymphoblastoid cell growth: Interaction with RBP-Jkappa regulates TCL1. J Virol 83:
7. Dowell RD (2010) Transcription factor binding variation in the evolution of gene
regulation. Trends Genet 26:468–475.
8. Schmidt D, et al. (2010) Five-vertebrate ChIP-seq reveals the evolutionary dynamics of
transcription factor binding. Science 328:1036–1040.
9. Palomero T, et al. (2006) CUTLL1, a novel human T-cell lymphoma cell line with t(7;9)
rearrangement, aberrant NOTCH1 activation and high sensitivity to gamma-secretase
inhibitors. Leukemia 20:1279–1287.
10. Krejcí A, Bray S (2007) Notch activation stimulates transient and selective binding of
Su(H)/CSL to target enhancers. Genes Dev 21:1322–1327.
11. Del Bianco C, et al. (2010) Notch and MAML-1 complexation do not detectably alter
the DNA binding specificity of the transcription factor CSL. PLoS ONE 5:e15034.
12. Hollenhorst PC, Shah AA, Hopkins C, Graves BJ (2007) Genome-wide analyses reveal
properties of redundant and specific promoter occupancy within the ETS gene family.
Genes Dev 21:1882–1894.
13. Hollenhorst PC, et al. (2009) DNA specificity determinants associate with distinct
transcription factor functions. PLoS Genet 5:e1000778.
14. Rothenberg EV (2007) Regulatory factors for initial T lymphocyte lineage specification.
Curr Opin Hematol 14:322–329.
15. Burns CE, Traver D, Mayhall E, Shepard JL, Zon LI (2005) Hematopoietic stem cell fate
is established by the Notch-Runx pathway. Genes Dev 19:2331–2342.
16. Lin JM, et al. (2007) Transcription factor binding and modified histones in human
bidirectional promoters. Genome Res 17:818–827.
17. Anno YN, et al. (2010) Genome-wide evidence for an essential role of the human Staf/
ZNF143 transcription factor in bidirectional transcription. Nucleic Acids Res 39:
18. Weng AP, et al. (2006) c-Myc is an important direct target of Notch1 in T-cell acute
lymphoblastic leukemia/lymphoma. Genes Dev 20:2096–2109.
19. Palomero T, et al. (2006) NOTCH1 directly regulates c-MYC and activates a feed-
forward-loop transcriptional network promoting leukemic cell growth. Proc Natl
Acad Sci USA 103:18261–18266.
20. Wendorff AA, et al. (2010) Hes1 is a critical but context-dependent mediator of
canonical Notch signaling in lymphocyte development and transformation. Immunity
21. Bernstein BE, et al. (2006) A bivalent chromatin structure marks key developmental
genes in embryonic stem cells. Cell 125:315–326.
22. Chadwick N, et al. (2009) Identification of novel Notch target genes in T cell
leukaemia. Mol Cancer 8:35.
23. González-García S, et al. (2009) CSL-MAML-dependent Notch1 signaling controls
T lineage-specific IL-7Ralpha gene expression in early human thymopoiesis and
leukemia. J Exp Med 206:779–791.
24. Haluska P, et al. (2006) In vitro and in vivo antitumor effects of the dual insulin-like
growth factor-I/insulin receptor inhibitor, BMS-554417. Cancer Res 66:362–371.
25. Pear WS, et al. (1996) Exclusive development of T cell neoplasms in mice transplanted
with bone marrow expressing activated Notch alleles. J Exp Med 183:2283–2291.
26. Ashworth T, et al. (2010) Deletion-based mechanisms of Notch1 activation in T-ALL:
key roles for RAG recombinase and a conserved internal translational start site in
Notch1. Blood 116:5455–5464.
27. Ballas N, Grunseich C, Lu DD, Speh JC, Mandel G (2005) REST and its corepressors
mediate plasticity of neuronal gene chromatin throughout neurogenesis. Cell 121:
28. Zhao B, et al. (2011) Epstein-Barr virus exploits intrinsic B-lymphocyte transcription
programs to achieve immortal cell growth. Proc Natl Acad Sci USA 108:14902–14907.
29. Hu Z, Killion PJ, Iyer VR (2007) Genetic reconstruction of a functional transcriptional
regulatory network. Nat Genet 39:683–687.
30. Li XY, et al. (2008) Transcription factors bind thousands of active and inactive regions
in the Drosophila blastoderm. PLoS Biol 6:e27.
31. Anderson MK, Hernandez-Hoyos G, Diamond RA, Rothenberg EV (1999) Precise
developmental regulation of Ets family transcription factors during specification and
commitment to the T cell lineage. Development 126:3131–3148.
32. Yu S, Zhao DM, Jothi R, Xue HH (2010) Critical requirement of GABPalpha for normal
T cell development. J Biol Chem 285:10179–10188.
33. Burns CE, et al. (2009) A genetic screen in zebrafish defines a hierarchical network of
pathways required for hematopoietic stem cell emergence. Blood 113:5776–5782.
34. Kundu M, et al. (2005) Runx1 deficiency predisposes mice to T-lymphoblastic
lymphoma. Blood 106:3621–3624.
35. Myslinski E, Krol A, Carbon P (1998) ZNF76 and ZNF143 are two human homologs of
the transcriptional activator Staf. J Biol Chem 273:21998–22006.
36. Myslinski E, Gérard MA, Krol A, Carbon P (2006) A genome scale location analysis of
human Staf/ZNF143-binding sites suggests a widespread role for human Staf/ZNF143
in mammalian promoters. J Biol Chem 281:39953–39962.
37. Yang MQ, Taylor J, Elnitski L (2008) Comparative analyses of bidirectional promoters
in vertebrates. BMC Bioinformatics 9(Suppl 6):S9.
38. Seo Y-K, et al. (2009) Genome-wide analysis of SREBP-1 binding in mouse liver
chromatin reveals a preference for promoter proximal binding to a new motif. Proc
Natl Acad Sci USA 106:13765–13769.
39. Gu H, Liang Y, Mandel G, Roizman B (2005) Components of the REST/CoREST/histone
deacetylase repressor complex are disrupted, modified, and translocated in HSV-1-
infected cells. Proc Natl Acad Sci USA 102:7571–7576.
40. Hori K, et al. (2008) A nonclassical bHLH Rbpj transcription factor complex is required
for specification of GABAergic neurons independent of Notch signaling. Genes Dev
41. Valouev A, et al. (2008) Genome-wide analysis of transcription factor binding sites
based on ChIP-Seq data. Nat Methods 5:829–834.
42. Heinz S, et al. (2010) Simple combinations of lineage-determining transcription
factors prime cis-regulatory elements required for macrophage and B cell identities.
Mol Cell 38:576–589.
43. Nam Y, Sliz P, Song L, Aster JC, Blacklow SC (2006) Structural basis for cooperativity in
recruitment of MAML coactivators to Notch transcription complexes. Cell 124:
Wang et al. PNAS
| September 6, 2011
| vol. 108
| no. 36