Activating Mutations of
NOTCH1 in Human T Cell Acute
Andrew P. Weng,1*. Adolfo A. Ferrando,2* Woojoong Lee,1
John P. Morris IV,2Lewis B. Silverman,2Cheryll Sanchez-Irizarry,1
Stephen C. Blacklow,1A. Thomas Look,2Jon C. Aster1-
Very rare cases of human T cell acute lymphoblastic leukemia (T-ALL) harbor
chromosomal translocations that involve NOTCH1, a gene encoding a trans-
membrane receptor that regulates normal T cell development. Here, we report
that more than 50% of human T-ALLs, including tumors from all major mo-
lecular oncogenic subtypes, have activating mutations that involve the ex-
tracellular heterodimerization domain and/or the C-terminal PEST domain of
NOTCH1. These findings greatly expand the role of activated NOTCH1 in the
molecular pathogenesis of human T-ALL and provide a strong rationale for
targeted therapies that interfere with NOTCH signaling.
T-ALL is an aggressive cancer that prefer-
entially affects children and adolescents. It is
commonly associated with acquired chromo-
sosomal translocations and other genetic or
epigenetic abnormalities, which lead to aber-
rant expression of a select group of tran-
scription factors (1). NOTCH1 was discovered
as a partner gene in a (7;9) chromosomal
translocation found in G1% of T-ALLs (2). It
encodes a transmembrane receptor that is
required for the commitment of pluripotent
progenitors to T cell fate (3) and the sub-
sequent assembly of pre–T cell receptor com-
plexes in immature thymocytes (4).
Cleavage of pro-NOTCH1 by a furinlike
protease during transit to the cell surface (5)
produces a NOTCH1 heterodimer comprised
of noncovalently associated extracellular
(NEC) and transmembrane (NTM) subunits
(6). The heterodimerization domain (HD) re-
sponsible for stable subunit association consists
of a 103 amino acid region of NEC (HD-N)
and a 65 amino acid region in NTM (HD-C)
(7). Physiologic activation of NOTCH recep-
tors occurs when ligands of the Delta-Serrate-
Lag2 (DSL) family bind to the NEC subunit
and initiate a cascade of proteolytic cleavages
in the NTM subunit. The final cleavage, cat-
alyzed by ,-secretase (8, 9), generates intra-
cellular NOTCH (ICN), which translocates to
the nucleus and forms a large transcriptional
activation complex that includes proteins of
the Mastermind family (10–12).
Prior work has shown that enforced
NOTCH1 signaling is a potent inducer of T-
ALL in the mouse (13–15) and is required to
sustain the growth of a human t(7;9)-positive
T-ALL cell line (16). To investigate the
possibility of a more general role for NOTCH
signaling in human T-ALL, we tested T-ALL
cell lines lacking the t(7;9) for NOTCH
dependency by treating these cells with a ,-
secretase inhibitor (17). Of 30 human T-ALL
cell lines tested, 5 showed a G0/G1cell-cycle
arrest that equaled or exceeded that of T6E, a
reference NOTCH1-dependent murine T-
ALL cell line (Fig. 1A). This drug-induced
growth suppression was abrogated by retro-
viral expression of ICN1 (Fig. 1B) and
reproduced (fig. S1) by retroviral expression
of dominant negative Mastermindlike-1 (16).
These results indicated that the growth of
these five cell lines depends on NOTCH-
Because physical dissociation of the
NOTCH extracellular domain has been linked
to receptor activation (6, 18), we reasoned
that the HD domain of NOTCH1 (7) could
be the site of gain-of-function mutations. A
second logical candidate region for oncogen-
ic mutations is the negative regulatory PEST
sequence lying at the C terminus of the
NOTCH1 NTM (19), as retroviral inser-
tions that cause deletion of this region have
been reported in murine T-ALL (14, 15).
Remarkably, sequencing revealed mutations
that involve both the HD-N domain and the
PEST domain in four of the five NOTCH-
dependent cell lines (summarized in Fig. 2).
Missense mutations affecting HD-N caused
nonconservative changes at amino acid po-
sitions that are invariant in vertebrate
NOTCH1 receptors (fig. S2). One cell line,
DND-41, had two different HD-N mutations
within the same NOTCH1 allele. The PEST
mutations were short insertions or deletions
causing shifts in reading frame that are
predicted to result in partial or complete
deletion of the PEST domain (fig. S4).
Sequencing of cDNAs revealed that the
1Department of Pathology, Brigham and Women’s
Hospital, Harvard Medical School, Boston, MA 02115,
USA.2Department of Pediatric Oncology, Dana Farber
Cancer Institute, Harvard Medical School, Boston, MA
*These authors contributed equally to this work.
.Present address: Department of Pathology, British
Columbia Cancer Agency, Vancouver, BC V5Z 4E6,
-To whom correspondence should be addressed.
Fig. 1. Identification of T-ALL cell lines that require NOTCH signals for
growth. (A) Effects of compound E, a ,-secretase inhibitor (GSI), on cy-
cling cell fractions (S and G2/M). After treatment for 4 to 8 days with
compound E (1 6M) or dimethyl sulfoxide (DMSO) carrier (mock), the
DNA content of propridium iodide-stained cell populations was deter-
mined by flow cytometry. (B) Abrogation of ,-secretase inhibitor-induced
cell-cycle arrest by MSCV-GFP-ICN1 retrovirus. Starting 2 days after
transduction with empty MSCV-GFP or MSCV-GFP-ICN1, cells were treated with either compound E (GSI, 1 6M) or DMSO carrier (mock) for 7 to 10
days. After staining with DRAQ5, the proportion of growth-arrested cells (G0þ G1fractions) in the GFP–and GFPþsubpopulations was determined by
R E P O R T S
www.sciencemag.orgSCIENCE VOL 3068 OCTOBER 2004
HD-N and PEST domain mutations lie in
cis in the same NOTCH1 allele in each of
the four cell lines tested (fig. S5). Normal
NOTCH1 cDNA clones were also identified
in each cell line, indicating that both alleles
are expressed. This is consistent with West-
ern blot analysis (Fig. 3), which revealed
that cell lines with HD-N and PEST do-
main mutations contained a polypeptide of
the expected size of NTM and additional
smaller polypeptides. We also sequenced a
subset of T-ALL cell lines that were insensi-
tive to the ,-secretase inhibitor. This revealed
NOTCH1 mutations in 9 out of 19 non-
responsive cell lines, including three lines
with dual HD and PEST domain mutations
(table S1). The failure of all cell lines with
mutations to respond to ,-secretase inhibitors
may result from these cell lines having been
maintained in tissue culture for many years.
We also identified frequent NOTCH1 HD
and PEST domain mutations in primary T-
ALL samples obtained from the bone mar-
row of 96 children and adolescents at the
time of diagnosis (summarized in Fig. 2). At
least one mutation was identified in 54 tumors
(56.2%). Mutations were seen in tumors as-
sociated with misexpression of HOX11 (2 of
3 cases), HOX11L2 (10 of 13, or 77%), TAL1
(12 of 31, or 39%), LYL1 (9 of 14, or 64%),
MLL-ENL (1 of 3), or CALM-AF10 (1 of 2)
(table S2), which together define the major
molecular subtypes of T-ALL (1). The HD
domain mutations in primary tumors were
clustered in a Bhot spot[ spanning residues
three L to P missense mutations originally
identified in the NOTCH-dependent T-ALL
cell lines, as well as deletions of 1 to 2 resi-
dues and short Bin-frame[ insertions (fig. S2).
In addition, a smaller number of missense
involved highly conserved residues (fig. S3).
PEST domain mutations included insertions or
deletions that induced a shift in reading frame
codons (fig. S4). In contrast with T-ALLs, B
cell ALLs (B-ALLs) (n 0 89) showed no
mutations in these regions of NOTCH1 (20).
Mutations were also absent from four remis-
sion bone marrow samples obtained from
patients whose T-ALLs harbored NOTCH1
mutations (20), indicating that these mutations
are acquired within the malignant clones.
To prove that HD domain mutations found
in T-ALL patients have effects on function,
NOTCH-sensitive reporter-gene assays were
conducted in human U2OS cells (Fig. 4).
Single L to P mutations within the HD-N do-
main at residues 1575, 1594, or 1601 caused
a 3- to 9-fold increase in luciferase activity,
whereas a T-ALL–associated PEST deletion
at position 2471 resulted in È1.5- to 2-fold
increase. More strikingly, each HD mutation
and the same PEST domain truncation in cis
resulted in 20- to 40-fold increases; in con-
trast, the same mutations in trans produced
lesser effects close to the average of each
mutation acting alone. The synergistic in-
teraction of HD and PEST domain mutations
in cis is consistent with a model (fig. S6)
in which (i) HD domain mutations enhance
,-secretase cleavage and increase the rate
of production of ICN1 and (ii) truncations
that remove the PEST domain increase
ICN1 half-life (19). The intermediate levels
of activation produced by these mutations
in trans presumably reflect competition be-
tween relatively weak and strong gain-of-
function NOTCH1 polypeptides for factors
required for processing and signaling. The
stimulatory effects of mutated transmembrane
NOTCH1 polypeptides were completely ab-
rogated by a ,-secretase inhibitor (Fig. 4),
which indicates a requirement for proteoly-
sis at the juxtamembrane site of NTM for sig-
nal transduction. In contrast, the stimulation
produced by ICN1, which is constitutively nu-
clear, was unaffected by ,-secretase inhibi-
tion (Fig. 4).
Several factors may explain the high
frequency of NOTCH1 mutations in T-ALL.
The requirement for NOTCH1 signals dur-
ing several stages of normal early T cell
development provides a functional basis for
its frequent involvement. Unlike the t(7;9),
which is created during attempted V-D-J$ re-
arrangement in committed T cell progenitors,
the common point mutations and insertions
described here could occur in multipotent
hematopoietic progenitors, which normally
express NOTCH1 (21). This would be pre-
dicted to induce daughter cells to adopt a
T cell fate (22) and thereby increase the pool
of cells at risk for additional leukemogenic
events, such as synergistic mutations af-
fecting NOTCH1 and misexpression of other
critical transcription factors. The NOTCH1
mutations we describe here are currently spe-
cific to human T-ALL among vertebrate can-
cers, but a mutation involving the putative
HD domain of the NOTCH homolog GLP-
1 causes massive germ-cell proliferation in
Caenorhabditis elegans (23), which suggests
that such mutations have a highly conserved
capacity to cause abnormal growth in spe-
cific cellular contexts.
Of potential clinical relevance, our findings
identify the NOTCH pathway as a rational
molecular therapeutic target in T-ALL.
Although up to 75% of T-ALL patients are
currently cured with very intensive cytotoxic
chemotherapy regimens (24), new therapies
are needed for patients with refractory dis-
ease, and less toxic, more efficacious drug
combinations would be generally beneficial.
Potent, specific inhibitors have already been
developed (25) as a result of the involvement
of ,-secretase in the production of amyloid-
ogenic peptides in patients with Alzheimer_s
disease. On the basis of the results reported
Fig. 2. NOTCH1 HD and PEST domain mutations in human T-ALL. A schematic representation
of human NOTCH1 shows the distribution and frequency of HD-N, HD-C, and PEST domain
mutations in NOTCH-dependent T-ALL cell lines (black arrowheads) and primary T-ALL samples
(white arrowheads). NEC, NOTCH1 extracellular domain; LNR, Lin/NOTCH repeats; HD-N and
HD-C, N-terminal and C-terminal halves, respectively, of the heterodimerization domain; NTM,
NOTCH transmembrane subunit; TM, transmembrane domain; RAM, RAM domain; ANK, ankyrin
repeat domain; TAD, transcriptional activation domain. The arrow denotes the site of furin
Fig. 3. Western blot analysis of lysates from
NOTCH-dependent T-ALL cell lines. Whole-cell
extracts were analyzed with a rabbit polyclonal
antibody raised against the transcriptional acti-
vationdomainofNOTCH1(13). The lower panel
is a longer exposure of the same blot that ac-
centuates the presence of additional NOTCH1
polypeptides of smaller size than the normal
NTM. Each lane contains 25 6g total protein.
R E P O R T S
8 OCTOBER 2004 VOL 306SCIENCE www.sciencemag.org
here, clinical trials are warranted to test
the efficacy and potential side effects (26)
of this class of NOTCH-pathway inhibitor in
patients with T-ALL.
References and Notes
1. A. A. Ferrando et al., Cancer Cell 1, 75 (2002).
2. L. W. Ellisen et al., Cell 66, 649 (1991).
3. F. Radtke et al., Immunity 10, 547 (1999).
4. A. Wolfer, A. Wilson, M. Nemir, H. R. MacDonald,
F. Radtke, Immunity 16, 869 (2002).
5. F. Logeat et al., Proc. Natl. Acad. Sci. U.S.A. 95, 8108
6. M. D. Rand et al., Mol. Cell. Biol. 20, 1825 (2000).
7. C. Sanchez-Irizarry et al., Mol. Cell. Biol., in press.
8. R. Francis et al., Dev. Cell 3, 85 (2002).
9. W. T. Kimberly et al., Proc. Natl. Acad. Sci. U.S.A.
100, 6382 (2003).
10. A. E. Wallberg, K. Pedersen, U. Lendahl, R. G. Roeder,
Mol. Cell. Biol. 22, 7812 (2002).
11. C. J. Fryer, E. Lamar, I. Turbachova, C. Kintner, K. A. Jones,
Genes Dev. 16, 1397 (2002).
12. Y. Nam, A. P. Weng, J. C. Aster, S. C. Blacklow, J. Biol.
Chem. 278, 21232 (2003).
13. J. C. Aster et al., Mol. Cell. Biol. 20, 7505 (2000).
14. C. D. Hoemann, N. Beaulieu, L. Girard, N. Rebai,
P. Jolicoeur, Mol. Cell. Biol. 20, 3831 (2000).
15. B. J. Feldman, T. Hampton, M. L. Cleary, Blood 96,
16. A. P. Weng et al., Mol. Cell. Biol. 23, 655 (2003).
17. Materials and methods are available as supporting
material on Science Online.
18. H. Kramer, Sci. STKE 2000, pe1 (2000).
19. N. Gupta-Rossi et al., J. Biol. Chem. 276, 34371
20. J. C. Aster, A. A. Ferrando, A. P. Weng, unpublished data.
21. L. M. Calvi et al., Nature 425, 841 (2003).
22. D. Allman et al., J. Exp. Med. 194, 99 (2001).
23. L. W. Berry, B. Westlund, T. Schedl, Development
124, 925 (1997).
24. C. H. Pui, M. V. Relling, J. R. Downing, N. Engl. J. Med.
350, 1535 (2004).
25. M. S. Wolfe, Nature Rev. Drug Discov. 1, 859 (2002).
26. G. T. Wong et al., J. Biol. Chem. 279, 12876 (2004).
27. We thank M. Wolfe for compound E and J.-P. Hezel
for expert technical assistance. Supported by NIH
grants CA82308 (J.C.A.), CA68484 and CA109901
(A.T.L.), CA94233 (S.C.B.), CA98093 (A.P.W.), and
CA21765 (St. Jude Cancer Center). A.A.F. is a fellow
of the Leukemia and Lymphoma Society.
Supporting Online Material
Materials and Methods
Figs. S1 to S6
Tables S1 and S2
30 June 2004; accepted 24 August 2004
of the E3 Ligase Itch
Min Gao,1Tord Labuda,1,2Ying Xia,1* Ewen Gallagher,1
Deyu Fang,3. Yun-Cai Liu,3Michael Karin1-
The turnover of Jun proteins, like that of other transcription factors, is
regulated through ubiquitin-dependent proteolysis. Usually, such processes are
regulated by extracellular stimuli through phosphorylation of the target
protein, which allows recognition by F box–containing E3 ubiquitin ligases. In
the case of c-Jun and JunB, we found that extracellular stimuli also modulate
protein turnover by regulating the activity of an E3 ligase by means of its
phosphorylation. Activation of the Jun amino-terminal kinase (JNK) mitogen-
activated protein kinase cascade after T cell stimulation accelerated degrada-
tion of c-Jun and JunB through phosphorylation-dependent activation of the E3
ligase Itch. This pathway modulates cytokine production by effector T cells.
Ubiquitin-dependent proteolysis controls turn-
over and abundance of transcription factors
and other regulatory proteins (1). Protein
ubiquitination requires the concerted action
of ubiquitin-activating enzyme (E1), ubiquitin-
conjugating enzymes (E2), and ubiquitin
ligases (E3) (2, 3). Extracellular stimuli can
regulate protein turnover through inducible
substrate phosphorylation which confers rec-
ognition by F box–containing E3 ligases (4).
Such E3 ligases, which are devoid of catalytic
activity, recognize only the phosphorylated
forms of their substrates (5). Transcription
factors regulated through ubiquitin-dependent
turnover include the Jun proteins, components
of the AP-1 transcription factor. The activity
of c-Jun and JunB is enhanced by phospho-
rylation of their transcriptional activation
domain by JNKs (6, 7). JNK-dependent
phosphorylation can also stabilize c-Jun
(8, 9). Recently, however, JNK-mediated
phosphorylation was shown to accelerate c-
Jun degradation by allowing its recognition
by the E3 ligase Fbw7-containing Skp/
Cullin/F-box protein complex (SCFFbw7)
(10). Here, we provide physiological and
biochemical evidence for another pathway
through which extracellular stimuli control
c-Jun and JunB abundance. This process is
based on inducible phosphorylation of an E3
ligase of the homology to the E6-associated
protein C terminus (HECT) family, which
increases its catalytic activity.
1Laboratory of Gene Regulation and Signal Trans-
duction, Department of Pharmacology, School of
Medicine, University of California, San Diego, 9500
Gilman Drive, La Jolla, CA 92093–0723, USA.
2Department of Medical Microbiology and Immunol-
ogy and Institute of Molecular Biology, University of
Copenhagen, 2200 Copenhagen N, Denmark.
sion of Cell Biology, La Jolla Institute for Allergy and
Immunology, San Diego, CA 92121, USA.
*Present address: Department of Environmental
Health, University of Cincinnati Medical Center,
Cincinnati, OH 45267–0056, USA.
.Present address: Department of Biological Chem-
istry, University of Michigan Medical School, Ann
Arbor, MI 48109–0606, USA.
-To whom correspondence should be addressed:
Fig. 4. HD and PEST do-
main mutations activate
NOTCH1 signaling syner-
gistically. Human U2OS
cells were transiently co-
transfected in 24-well
format with the indicated
pcDNA3 plasmids, a
erase reporter gene, and
an internal Renilla lucif-
erase internal control
plasmid, as described
previously (13). Twenty-
five ng of pcDNA3 plas-
mid was used per well,
except for experiments
with pcDNA3-ICN1, in
which 5 ng of plasmid per
well was used. ‘‘DeltaP’’
denotes the presence of a
deletion removing NOTCH1
residues 2473 to 2556.
Normalized luciferase ac-
tivities in whole-cell lysates were determined in triplicate and expressed relative to the activity in
lysates prepared from vector-transfected cells. Error bars represent standard deviations.
R E P O R T S
www.sciencemag.org SCIENCEVOL 306 8 OCTOBER 2004