Hierarchy of Notch-Delta interactions promoting T cell lineage commitment and maturation.

Abstract

Notch1 (N1) receptor signaling is essential and sufficient for T cell development, and recently developed in vitro culture systems point to members of the Delta family as being the physiological N1 ligands. We explored the ability of Delta1 (DL1) and DL4 to induce T cell lineage commitment and/or maturation in vitro and in vivo from bone marrow (BM) precursors conditionally gene targeted for N1 and/or N2. In vitro DL1 can trigger T cell lineage commitment via either N1 or N2. N1- or N2-mediated T cell lineage commitment can also occur in the spleen after short-term BM transplantation. However, N2-DL1-mediated signaling does not allow further T cell maturation beyond the CD25(+) stage due to a lack of T cell receptor beta expression. In contrast to DL1, DL4 induces and supports T cell commitment and maturation in vitro and in vivo exclusively via specific interaction with N1. Moreover, comparative binding studies show preferential interaction of DL4 with N1, whereas binding of DL1 to N1 is weak. Interestingly, preferential N1-DL4 binding reflects reduced dependence of this interaction on Lunatic fringe, a glycosyl transferase that generally enhances the avidity of Notch receptors for Delta ligands. Collectively, our results establish a hierarchy of Notch-Delta interactions in which N1-DL4 exhibits the greatest capacity to induce and support T cell development.

Full text

Page 1

The Journal of Experimental Medicine
ARTICLE
JEM © The Rockefeller University Press $15.00
Vol. 204, No. 2, February 19, 2007 331–343 www.jem.org/cgi/doi/10.1084/jem.20061442
331
T cells, like other cells of the blood system, are
derived from pluripotent hematopoietic stem
cells (HSCs). The major site of T cell develop-
ment is the thymus. Thus, descendants of HSCs
migrate to the thymus, where they undergo a
program of maturation, proliferation, and dif-
ferentiation. They pass through a CD4−CD8−
double-negative (DN) developmental stage,
followed by a CD4+CD8+ double-positive (DP)
stage, before undergoing positive or negative
selection to generate single-positive (SP) CD4+
and CD8+ T cells that migrate to the peri-
phery. The CD4−CD8− DN cells represent
the most immature thymic subset that can be
further subdivided into four developmental
stages (DN1–4), based on their diff erential ex-
pression of CD44 and CD25, maturing from
the CD44+CD25− (DN1) to the CD44+CD25+
(DN2) to the CD44−CD25+ (DN3) to the
CD44−CD25− (DN4) stages (1–3). Many dif-
ferent signaling pathways have been shown to
be involved in T lymphocyte development.
One of these pathways is the Notch cascade,
which has received a lot of attention in recent
years because of its involvement in T lineage
commitment, T cell maturation, and peripheral
T cell function (4, 5). Notch proteins compose
a family of four transmembrane receptors that
infl uence cell fate decisions and diff erentiation
processes in many diff erent organisms (6).
Notch signaling is triggered upon binding of
ligands of the Jagged and Delta family. This
leads to a cascade of proteolytic cleavages that
release the intracellular cytoplasmic domain of
Notch receptors, which subsequently trans-
locates to the nucleus, where it binds to the
Hierarchy of Notch–Delta interactions
promoting T cell lineage commitment
and maturation
Valerie Besseyrias,1 Emma Fiorini,1 Lothar J. Strobl,2 Ursula Zimber-Strobl,2
Alexis Dumortier,3 Ute Koch,3 Marie-Laure Arcangeli,4 Sophie Ezine,4
H. Robson MacDonald,1 and Freddy Radtke1,3
1Ludwig Institute for Cancer Research, Lausanne Branch, University of Lausanne, 1066 Epalinges, Switzerland
2Institute of Clinical Molecular Biology and Tumor Genetics, GSF-National Research Center for Environment and Health,
81377 Munich, Germany
3Swiss Institute for Experimental Cancer Research, 1066 Epalinges, Switzerland
4Institut National de la Santé et de la Recherche Médicale, U591, Université Paris V, 75730 Paris Cedex 15, France
Notch1 (N1) receptor signaling is essential and suffi cient for T cell development, and
recently developed in vitro culture systems point to members of the Delta family as being
the physiological N1 ligands. We explored the ability of Delta1 (DL1) and DL4 to induce
T cell lineage commitment and/or maturation in vitro and in vivo from bone marrow (BM)
precursors conditionally gene targeted for N1 and/or N2. In vitro DL1 can trigger T cell
lineage commitment via either N1 or N2. N1- or N2-mediated T cell lineage commitment
can also occur in the spleen after short-term BM transplantation. However, N2–DL1–
mediated signaling does not allow further T cell maturation beyond the CD25+ stage due
to a lack of T cell receptor 훃 expression. In contrast to DL1, DL4 induces and supports T cell
commitment and maturation in vitro and in vivo exclusively via specifi c interaction with N1.
Moreover, comparative binding studies show preferential interaction of DL4 with N1,
whereas binding of DL1 to N1 is weak. Interestingly, preferential N1–DL4 binding refl ects
reduced dependence of this interaction on Lunatic fringe, a glycosyl transferase that gen-
erally enhances the avidity of Notch receptors for Delta ligands. Collectively, our results
establish a hierarchy of Notch–Delta interactions in which N1–DL4 exhibits the greatest
capacity to induce and support T cell development.
CORRESPONDENCE
Freddy Radtke:
Freddy.Radtke@isrec.unil.ch
Abbreviations used: DL1,
Delta1; DN, double negative;
DP, double positive; EGFP,
enhanced GFP; ES, embryonic
stem; HPRT, hypoxanthine
guanine phosphoribosyl trans-
ferase; HSC, hematopoietic stem
cell; ISP, immature SP; KLS,
CD117+lin−Sca1+; Lfng, Lunatic
fringe; MZB cell, marginal zone
B cell; N1, Notch1; SP, single
positive.
V. Besseyrias, E. Fiorini, L.J. Strobl, and U. Zimber-Strobl
contributed equally to this work.
The online version of this article contains supplemental material.

Page 2

332 NOTCH–DELTA INTERACTIONS DURING T CELL DEVELOPMENT | Besseyrias et al.
RBP-J transcription factor and thereby activates transcription.
Notch signaling itself can be regulated by several modulators,
such as the family of Fringe proteins, which are glycosyl
transferases that add N-acetylglucosamine to certain epidermal
growth factor–like repeats of Notch receptors, promoting
Notch signaling in response to Delta ligands and inhibiting
Jagged-mediated Notch signaling (7).
The best-established role for Notch signaling in the
hematopoietic system is the essential function of Notch1
(N1) in T cell fate specifi cation. Conditional inactivation of
the N1 (8, 9) or RBP-J (10) genes in adult BM progenitors
results in B cell development within the thymus at the ex-
pense of T cell lineage commitment, suggesting that N1/
RBP-J–mediated signaling is important to induce T cell de-
velopment and to simultaneously block B cell development.
Although multiple Notch receptors such as N1, N2, and
N3—as well as the ligands Jagged1, Jagged2, Delta1 (DL1),
and DL4—are expressed on thymocytes and/or thymic epi-
thelium (11–18), T lineage commitment appears to be medi-
ated via the N1 receptor in a nonredundant manner. This
is consistent with the fi nding that conditional inactivation
of the N2 gene does not aff ect T cell development but is
instead necessary for marginal zone B cell (MZB cell) speci-
fi cation (19). Moreover, N3 gene–targeted mice do not ex-
hibit any hematopoietic phenotype (20). Further support
for the essential role of Notch signaling in T cell lineage
commitment is derived from gain-of-function studies, as
overexpression of a constitutively active form of N1 (21, 22)
or DL4 (13, 23, 24) induces ectopic T cell development
in the BM and simultaneously blocks B cell development.
Thus, these reciprocal loss- and gain-of-function studies
indicate that N1 signaling is necessary and suffi cient for T cell
lineage commitment.
An additional nonredundant function of N1 during thy-
mocyte maturation was revealed by conditional inactivation
of the N1 gene in immature thymocytes. N1 defi ciency in
DN thymocytes leads to a partial block of αβ T cell develop-
ment at the pre-TCR checkpoint because of defective V to
DJβ rearrangement (25). Although N1 seems to be a key
player during T lineage commitment and T cell maturation,
several issues are still controversial or unknown. For example,
the expression of multiple Notch ligands on thymic epithelial
cells leads to the question of which ligand triggers the physio-
logical N1 signal for T lineage commitment and/or maturation.
Figure 1. N2 signaling is suffi cient to induce T lineage commit-
ment in vitro but not in vivo. (A) Mixed BM chimeric mice were
analyzed 8 wk after reconstitution with a 1:2 mixture of WT (CD45.1+)
and Ctrl (N1lox/lox), N1−/−, N2−/−, or N1N2−/− (CD45.2+) BM-derived
populations. Representative FACS analysis of thymocytes stained with
anti-CD117, -CD44, and -CD25 antibodies after gating on donor
(CD45.2+)-derived lineage-negative cells (top). Representative FACS
analysis of thymocytes stained with anti-B220 and -CD19 antibodies
after gating on donor (CD45.2+)-derived lineage-negative cells (bottom).
Representative FACS profi les are derived from experiments in which fi ve
mice of each genotype were analyzed. (B) BM KLS cells were sorted from
Ctrl, N1−/−, N2−/−, and N1N2−/− mice and cultured on OP9-DL1 cells for
10 d (top) and 18 d (bottom). Cells from these cultures were analyzed
for the expression of CD44 and CD25 (top) and for the presence of
B220+CD19+ B cells (bottom). Representative FACS profi les are derived
from four individual experiments. (C) Deletion PCR analysis for the N1
gene was performed on genomic DNA from sorted CD25− (corresponding
to DN1) and CD25+ (corresponding to DN2/DN3) cells derived from N1−/−
and Ctrl animals cultured for 10 d on OP9-DL1. (D) Semiquantitative
RT-PCR for the expression of N1, N2, and tubulin was performed on
sorted BM KLS cells. Three serial dilutions (threefold) of template RNA
are shown for the indicated genes.

Page 3

JEM VOL. 204, February 19, 2007
333
ARTICLE
Ligands of the Jagged family (Jagged1 and Jagged2) can be
excluded as being essential during these processes, because
conditional inactivation of Jagged1 (26) does not perturb he-
matopoiesis and Jagged2-defi cient mice show only a minor
decrease in γδ T cells, whereas αβ T cell development ap-
pears normal (27). Thus, members of the Delta-like family
seem to be the crucial ligands, because expression of DL1 or
DL4 on stromal cells can induce T cell development of
human or mouse hematopoietic progenitors (28–30). Inter-
estingly, conditional inactivation of the DL1 gene in hemato-
poietic cells leads to the loss of MZB cells, indicating that DL1
signals via N2 to specify this subclass of splenic B cells (30).
Surprisingly, loss of DL1, even in thymic epithelium, does
not perturb T cell development, indicating that DL1 is also
dispensable for T cell lineage commitment or T cell matura-
tion in vivo (19, 30). Because DL4-expressing stromal cells
can also induce T cell development in vitro (18, 30), it is
conceivable that loss of DL1 function in vivo is compensated
by DL4.
In this paper, we explore the ability of DL1 and DL4 to
induce T cell lineage commitment and/or to infl uence T cell
maturation in vitro and in vivo via interactions with N1 and/
or N2. Our results show that DL1 and DL4 exhibit diff erent
Notch receptor specifi cities and that T cell fate specifi cation
is mediated by specifi c Notch receptor–ligand interactions.
RESULTS
N2 compensates for the loss of N1 during T cell
commitment in vitro but not in vivo
To further characterize the Notch-dependent events of T cell
development, we examined in vivo versus in vitro T lineage
commitment. For the in vivo analysis, we set up mixed BM
chimeras using BM cells of conditional gene-targeted mice in
which either the N1 or N2 genes alone or both N1 and N2
can be inactivated simultaneously in HSCs (using the IFN-
α–responsive Mx-Cre system). Therefore, CD45.1+ WT
lethally irradiated mice were injected with either CD45.2+
control (N1lox/lox; Ctrl), N1-defi cient (N1−/−), N2-defi cient
(N2−/−), or N1N2–double-defi cient (N1N2−/−) BM pro-
genitors. Thymic T cell development was analyzed 8 wk after
transplantation (Fig. 1 A). As previously described, inactiva-
tion of the N1 gene in BM progenitors results in a block at or
before the earliest intrathymic precursor stage (8). Immature
B cells develop in the thymus from incoming N1−/− BM
progenitors, demonstrating that N1 signaling is essential for T
lineage commitment in vivo (Fig. 1 A) (9). N2−/− BM pro-
genitors reconstituted the T cell lineage as effi ciently as BM
cells derived from control animals. No ectopic B cell devel-
opment was observed in the thymus of N2−/− BM chimeras.
The only detectable hematopoietic phenotype caused by the
inactivation of N2 was the loss of MZB cells in the spleen
(unpublished data). In contrast, chimeric mice reconstituted
with BM progenitors double defi cient for both N1 and N2
recapitulated the phenotype of the N1−/− BM chimeras, sug-
gesting that T lineage commitment in vivo is exclusively
dependent on N1 signaling.
To further characterize the N1-dependent events of T
lineage commitment in vitro, we have made use of the OP9-
DL1 culture system (29) in which purifi ed (CD117+lin−Sca1+;
KLS) HSCs from either Ctrl, N1−/−, N2−/−, or N1N2−/− BM
progenitors were sorted and cultured on DL1-expressing
OP9 stromal cells. Surprisingly, after 10 d of culture, N1−/−
HSCs principally gave rise to DN2 (CD44+CD25+) and
DN3 (CD44−CD25+) T cell progenitors (Fig. 1 B). No dif-
ference was observed in the ability of Ctrl or N1−/− HSCs to
diff erentiate into immature thymocytes. To rule out the pos-
sibility that these OP9-DL1–generated DN2 and DN3 pro-
genitor cells were derived from precursor cells that had
escaped N1 deletion, both subsets were sorted and analyzed
by PCR for the successful inactivation of the N1 gene. The
PCR analysis shows the expected bands characterizing the
fl oxed and the inactivated N1 alleles of the sorted cells, re-
spectively, confi rming that the OP9-DL1–generated DN2
and DN3 cells were indeed the progeny of N1−/− HSCs
(Fig. 1 C).
Because N2 is expressed together with N1 on HSCs
(Fig. 1 D), it is conceivable that N2 is able to compensate for
the loss of N1 function during T lineage commitment when
N1−/− HSCs are cultured on DL1-expressing OP9 cells. To
test this hypothesis, we sorted and cultured N2−/− and
N1N2−/− HSCs on DL1-expressing OP9 cells. After 10 d of
OP9-DL1 culture, N2−/− HSCs gave rise to immature
DN1–3 T cell progenitors similar to Ctrl and N1−/− HSCs
(Fig. 1 B). In contrast, N1N2−/− HSCs do not develop into
T cell progenitors; instead, they exhibit a developmental
block at the putative DN1 stage characterized by the accu-
mulation of CD44+CD25− cells. This phenotype is very
reminiscent of the defect observed in vivo in inducible N1−/−
mice, where CD44+CD25− cells accumulated in the thymus
and were identifi ed as B220+CD19+ B cells (Fig. 1 A) (8).
Therefore, Ctrl, N1−/−, N2−/−, and N1N2−/− HSCs were
assessed for their ability to develop into B cells on OP9-
DL1–expressing stromal cells. Only HSCs derived from
N1N2−/− HSCs developed into B cells after 18 d of culture,
whereas Ctrl, N1−/−, or N2−/− cells did not. These data con-
fi rm the hypothesis that DL1-mediated N2 signaling can
compensate for the loss of N1 function during T lineage com-
mitment in vitro and that the presence of either N1 or N2
alone is suffi cient to block B cell development.
N2 cannot compensate for the loss of N1 function
during T cell maturation
Notch signaling is not only essential for T lineage commit-
ment but is also continuously required for the successful dif-
ferentiation of all DN thymocyte subsets into CD4+CD8+
DP cells (14). Because N2 can instruct N1−/− HSCs to adopt
a T cell fate on OP9-DL1–expressing stromal cells, it is
conceivable that N2 signaling would be suffi cient to allow
the subsequent DN to DP transition. To this end, Ctrl and
N1−/− HSCs were cultured for 28 d on OP9-DL1 stromal
cells and subsequently analyzed for the development of DP
T cells. Although Ctrl HSCs diff erentiated very effi ciently

Page 4

334 NOTCH–DELTA INTERACTIONS DURING T CELL DEVELOPMENT | Besseyrias et al.
into DP T cells, >90% of the N1−/− cells appeared to be
blocked in the DN compartment (Fig. 2 A).
These data show that N2 signaling, although suffi cient for
T cell commitment of BM HSCs, is not suffi cient for T cell
maturation. Impaired DN to DP transition has also been ob-
served in mice in which the N1 gene was inactivated in im-
mature thymocytes using the lck-Cre transgene (25). In these
mice, N1 defi ciency leads to a partial block of αβ T cell de-
velopment at the pre-TCR checkpoint because of inhibition
of VDJβ rearrangement. This partial block is characterized by
a substantial decrease in the proportion of N1−/− DN3 and
DN4 cells expressing intracellular TCRβ protein (25). To
investigate whether the block observed in the in vitro culture
system at the DN to DP transition might also be caused by
ineffi cient expression of a TCRβ chain, intracellular TCRβ
staining was performed on WT, DN3, or DN4 thymocytes
or on in vitro–generated DN T cell progenitors derived from
Ctrl or N1−/− HSCs. Approximately 60% of the in vitro–
generated DN3 and DN4 T cell progenitors derived from
Ctrl HSCs have an in-frame TCRβ rearrangement and, thus,
stained positive for intracellular TCRβ, compared with 29%
of DN3 and 94% of DN4 thymocytes in vivo (Fig. 2 B).
However, only 5 and 8% of the DN3 and DN4 cells, re-
spectively, derived from N1−/− HSCs were icTCRβ+. These
results show that DL1-mediated N2 signaling is not suffi cient
to allow diff erentiation of DN immature cells into DP T cell
progenitors in the absence of N1 because of impaired TCRβ
rearrangement and/or expression.
N2 mediates T lineage commitment in vivo
in the absence of N1 signaling at extrathymic sites
after BM transplantation
Although N1 is the key receptor for T lineage commitment
in the thymus, our in vitro data raise the possibility that, un-
der certain conditions (when encountering the DL1 ligand),
the N2 receptor might be able to induce T lineage commit-
ment at extrathymic sites in vivo.
Recent studies by Maillard et al. (31), Lancrin et al. (32),
and Arcangeli et al. (33) showed that early T cell develop-
ment occurs in the spleen and LNs after BM transplantation.
This pool of splenic T cell progenitors can effi ciently con-
tribute to donor-derived thymopoiesis by migrating from
the spleen to the thymus (31), where they complete T cell
maturation. The generation of these splenic T cell progen-
itors after BM transplantation is Notch signaling dependent
(31), because it is blocked by the expression of a dominant-
negative form of the Notch coactivator mastermind-like 1
(31). However, it is not clear whether these extrathymi cally
derived T cell progenitors are generated in a N1- and/or
N2-dependent manner. To address this question, we trans-
planted CD45.2+ Ctrl, N1−/−, N2−/−, and N1N2−/− BM
cells into lethally irradiated CD45.1+ C57BL/6 recipients.
Donor-derived lin− cells in the spleen were analyzed 12 d
after BM transplantation by staining for Thy1.2, CD44,
and CD25. As previously reported, Thy1.2 and CD44
staining identifi ed two populations within the lin− donor-
derived cells in mice receiving Ctrl BM cells (32, 33). The
CD44+Thy1.2− population has previously been shown to
have multilineage potential, whereas the CD44lo/−Thy1.2+
population is T lineage restricted (Fig. 3) (33). The Thy1.2+
cells appear to be heterogeneous for the expression of CD25,
as ?50% express this marker, which normally defi nes the
DN2 and DN3 subsets of immature thymocytes. A similar
population of splenic Thy1.2+ T cell progenitors was iden-
tifi ed in hosts receiving Ctrl, N1−/−, and N2−/− BM. In
contrast, no Thy1.2+ cells were observed in the lin− donor-
derived population of hosts receiving N1N2−/− BM (Fig. 3).
These results demonstrate that splenic T cell progenitors can
be generated in the absence of N1 after BM transplantation
in a N2-dependent manner. Thus, either N2 or N1 signal-
ing is suffi cient for T lineage commitment in the spleen after
BM transplantation.
DL1 and DL4 ligands exhibit different Notch
receptor specifi cities
The simplest hypothesis to explain the discrepancy in the
ability of N2 to compensate for the loss of N1 during T lin-
eage commitment in vitro but not in vivo is that the N2
Figure 2. N2 cannot compensate for the loss of N1 function
during T cell maturation in vitro. (A) KLS cells from Ctrl and induced N1−/−
mice were sorted and cultured on OP-DL1 cells for 20 d. A representative
fl ow cytometric analysis of CD4 versus CD8 of WT thymocytes, and Ctrl
and N1−/− KLS cells cultured on OP9-DL1 are shown. (B) Indicated cells
were electronically gated on lineage-negative DN thymocytes and ana-
lyzed for the expression of CD44 and CD25. Representative histograms
for intracellular TCRβ (iTCRβ) expression on DN3 and DN4 thymocytes
derived from Ctrl and N1−/− KLS cells 20 d after culture on OP9-DL1 are
shown. The numbers above the bars indicate the percentage ± SD of
iTCRβ+ cells (n = 4 for WT thymocytes and in vitro culture experiments).

Page 5

JEM VOL. 204, February 19, 2007
335
ARTICLE
gene, although being expressed on HSCs, is not expressed in
the earliest thymocyte progenitors. To test this hypothesis,
semiquantitative RT-PCR was performed on all DN subsets
derived from either BM HSCs that have been cultured
on OP9-DL1 cells or from WT thymocytes. As shown in
Fig. 4, both N1 and N2 were expressed in all DN subsets
(even in DN1, CD117+CD44+CD25− subsets), irrespective
of whether they were generated in vitro or were derived from
thymocytes in vivo. Furthermore, no noticeable diff erences
were observed in the expression levels of both genes in the
diff erent DN subsets (Fig. 4). Thus, the hypothesis that the
absence of N2 expression in early thymocyte progenitors ac-
counts for the diff erential outcome of T lineage commitment
in vivo versus in vitro is unlikely. However, an alternative
explanation for this discrepancy is that the outcome of Notch
signaling might be dependent on the specifi city and/or avid-
ity of certain ligand–receptor interactions. Although DL1 is
able to interact with N1 and N2 to trigger T lineage commit-
ment in vitro, it is conceivable that other Notch ligands may
also trigger T lineage commitment, but only upon selective
interaction with one specifi c Notch receptor. An attractive
alternative ligand for N1 is DL4, a close homologue of DL1,
which has also been shown to trigger T lineage commitment
of WT precursors in vitro when expressed on OP9 cells (30).
To test whether DL1 and DL4 can mediate T cell develop-
ment equally effi ciently in vitro, sorted WT BM HSCs were
cultured on DL1- or DL4-expressing OP9 cells and analyzed
side by side. Although the expression level of DL4 was slightly
lower than DL1 on OP9 cells (as judged by GFP expression;
Fig. 5 A, left), WT HSCs developed very effi ciently into DP
thymocytes in both cases within 30 d of culture (Fig. 5 A,
right). To investigate whether DL1 and DL4 exhibit diff erent
Notch receptor specifi cities, Ctrl, N2−/−, and N1−/− HSCs
were cultured side by side on either DL1- or DL4-expressing
OP9 cells. After 11 d of culture, Ctrl HSCs cultured on either
DL1 or DL4 progressed to the DN2/DN3 stage, although
the progression appeared to be slightly more rapid on OP9-
DL4–expressing cells. Whereas all three genotypes developed
into DN1, DN2, and DN3 T cell progenitors on DL1-
expressing OP9 cells, only Ctrl and N2−/− HSCs, but not
N1−/− HSCs, developed into DN1, DN2, and DN3 T cell
progenitors when cultured on OP9-DL4 cells (Fig. 5 B).
Within the same time frame, N1−/− HSCs could only diff er-
entiate into a putative DN1 population (CD44+CD25−) on
OP9-DL4 cells. These data indicate that N1−/− HSCs either
exhibit a developmental block at the DN1 to DN2 transition
or that they may have adopted a B cell fate similar to the
phenotype observed after inducible inactivation of N1 in BM
progenitors in vivo (Fig. 1 A) (8, 9). Therefore, the diff erent
OP9-DL4 cultures were stained with antibodies to the pan–B
cell markers B220 and CD19. As shown in Fig. 5 B (right),
B220+CD19+ B cells were only observed when N1−/− but
not Ctrl or N2−/− BM HSCs were cultured on OP9-DL4
cells. These results demonstrate that DL4 and DL1 are not
equivalent in their ability to trigger T lineage commitment,
as DL4 can only induce T lineage commitment of BM HSCs
Figure 3. N2 is suffi cient to specify T lineage progenitors in the
spleen after BM transplantation. CD45.2+ Ctrl, N1−/−, N2−/−, or
N1N2−/− BM cells were injected into lethally irradiated CD45.1+ hosts. The
spleens of host mice were analyzed 12 d after BM transplantation. Repre-
sentative fl ow cytometric analyses of donor-derived lineage-negative cells
for the expression of CD44 and Thy1.2, and Thy1.2 and CD25, respectively,
are shown. Data are representative of four independent experiments.
Figure 4. Expression of N1 and N2 in immature DN thymocytes.
cDNA was prepared from sorted cells cultured on OP9-DL1 cells for 16 d
(corresponding to the DN1–4 subsets), and WT DN1–4 thymocyte subsets.
Transcripts of N1 and N2 were analyzed by semiquantitative RT-PCR of
threefold dilutions of the cDNA. The cDNA input was normalized accord-
ing to the expression of the control tubulin gene.

Page 6

336 NOTCH–DELTA INTERACTIONS DURING T CELL DEVELOPMENT | Besseyrias et al.
that express N1. Moreover, DL4-mediated N2 signaling is
not suffi cient for T cell commitment nor is it suffi cient for
the inhibition of B cell development in vitro, suggesting that
DL4 signals specifi cally through the N1 receptor, whereas
DL1 can signal through both N1 and N2. To investigate
whether DL1 and DL4 exhibit diff erent binding avidities for
N1 and/or N2, DL1- and DL4-IgG fusion proteins were
generated (Fig. 6 A) and assessed for their ability to bind to
Notch receptors expressed on thymocytes. As shown in Fig.
6 B, DL4-IgG fusion proteins bind immature DN thymo-
cytes very effi ciently. Binding of DL4-IgG is already observed
in the DN1 subset, peaks at the DN2 subset, gradually de-
clines through the DN3, DN4, and immature SP (ISP) sub-
sets, and is virtually absent in more mature DP and CD4
and CD8 SP thymocytes (Fig. 6 C). Surprisingly, DL1-IgG
fusion proteins did not stain thymocytes above background
levels of the IgG isotype control (unpublished data), with
the exception of the DN1 and DN2 subsets, which stained
weakly above background. These results demonstrate that
the DL4 fusion protein has a considerably higher binding
avidity to Notch receptors present on immature thymocytes
compared with DL1.
To investigate whether diff erences in the binding avidity
of DL1- and DL4-IgG fusion proteins translate into diff eren-
tial Notch target gene induction, immature DN thymocytes
were cultured on either DL1- or DL4-IgG–coated plastic
dishes or on DL1- or DL4-expressing OP9 stromal cells. Ex-
pression of the Notch target genes Deltex1 and Hes1 was sub-
sequently analyzed by semiquantitative RT-PCR. Similar to
the hierarchy observed in binding assays, induction of Deltex1
gene expression was stronger when Notch signaling was trig-
gered by DL4 compared with DL1. However, diff erences in
Hes1 gene expression were less pronounced or, as in the OP9
culture system, not observed (Fig. 6 D). These results indicate
that diff erential Notch–Delta binding avidity correlates in
some cases with diff erences at the level of gene expression.
However, this correlation appears to be target gene dependent,
suggesting that diff erent target genes respond to diff erent
threshold levels of Notch signaling.
Because the binding assays were performed with WT
thymocytes, we were unable to distinguish binding of DL4-
IgG to the N1 and/or N2 receptor. Our results presented in
Fig. 5 show that DL4, in contrast to DL1, cannot induce T
lineage commitment via N2. This result is compatible with
two hypotheses: either DL4 cannot bind effi ciently to N2 or,
alternatively, DL4 can bind N2 but cannot transmit a signal
via the N2 receptor. To distinguish between these two possi-
bilities, we examined the binding effi ciency of DL1- and
DL4-IgG fusion proteins to 293T cells transiently expressing
either N1- or N2–enhanced GFP (EGFP) fusion proteins. As
shown in Fig. 7 (B and C, top), DL4-IgG fusion proteins
bind N1 very effi ciently, whereas binding to the N2 receptor
is not detectable above background. Interestingly, DL1-IgG
fusion proteins bind N1 weakly but do not bind N2 above
levels of the IgG isotype control.
Modulation of Notch–Delta binding avidity
by Lunatic fringe (Lfng)
A recent report suggests that the sensitivity of Notch recep-
tors to DL ligands in thymocytes is regulated by the glycosyl-
transferase Lfng (34). Although only N1–DL4 binding can be
detected in transiently transfected 293T cells (Fig. 7 B, top),
it is possible that this situation refl ects the fact that 293T cells
express very low levels of Lfng as measured by semiquantita-
tive RT-PCR (Fig. 7 A). To investigate directly whether the
avidity of Notch–Delta binding might be dependent on Lfng
expression, 293T cells transiently expressing either N1 or N2
were cotransfected with an expression plasmid encoding Lfng
and subsequently analyzed in binding assays using DL1- and
DL4-IgG fusion proteins. Interestingly, overexpression of
Lfng in 293T cells allows DL1 to bind effi ciently to N1 and
N2. Moreover, DL4-Fc can now also bind N2 (Fig. 7 B
and C, bottom). These results directly demonstrate that the
binding avidity of DL ligands to Notch receptors is Lfng
dependent, with the possible exception of the DL4–N1 inter-
action. They furthermore highlight the possibility that Notch
signaling might be regulated in an important way at the level
of Lfng expression.
Figure 5. Comparison of DL1- and DL4-mediated T cell develop-
ment in vitro. (A) Histograms show fl ow cytometric analyses of GFP
expression of uninfected OP9-cells (dashed line) and DL1 (shaded histo-
gram) and DL4 (continuous line) retrovirally transduced OP9 cells. Ctrl BM
HSCs were sorted and cultured side by side on OP9-DL1 and -DL4 cells
and anaylzed by fl ow cytometry for the expression of CD4 and CD8 30 d
after culture. (B) Sorted BM HSCs derived from either Ctrl, N1−/−, or
N2−/− mice were cultured for the indicated times on either OP9-DL1 or
-DL4 cells and subsequently analyzed by fl ow cytometry for the expres-
sion of CD44 and CD25 (electronically gated on lineage-negative cells)
and the presence of B220+CD19+ B cells. Data are representative of four
independent experiments.

Page 7

JEM VOL. 204, February 19, 2007
337
ARTICLE
Enforced expression of DL4 but not DL1 induces effi cient
T cell development in vivo
In vitro, DL4 is unable to induce T lineage commitment in
the absence of a functional N1 receptor, suggesting that DL4
must specifi cally interact with N1 to specify the T lineage.
To exclude that this observation is caused by a peculiarity of
our in vitro culture system, we investigated the ability of DL1
and DL4 to induce T cell development in vivo in the pres-
ence and absence of N1. Previous studies demonstrated that
retroviral overexpression of DL4 in hematopoietic cells is
suffi cient to promote thymus-independent T cell development
to the DP stage in vivo (13, 23, 24). We therefore transduced
CD45.2+ Ctrl and N1−/− BM cells with a retrovirus express-
ing either GFP alone (MIG), or DL1 or DL4 together with
GFP, and subsequently transplanted these cells into lethally
irradiated CD45.1+ C57BL/6 mice. The BM transduction
effi ciency of MIG and MIG expressing either DL1 or DL4
virus (based on GFP expression) was between 55 and 60% for
both Ctrl and N1−/− BM cells (Fig. 8 A). Reconstituted hosts
were analyzed 9 wk after transplantation for the presence of
GFP+ donor-derived cells in PBLs. 72% of Ctrl PBLs and
64% of N1−/− PBLs in host mice that were transplanted with
MIG-transduced BM cells were GFP+. Comparable percent-
ages of PBLs were GFP+ in hosts receiving either DL1- or
DL4-expressing Ctrl and N1−/− BM cells, indicating that the
relative number of virus-expressing Ctrl and N1−/− donor
cells was comparable even 9 wk after transplantation (Fig.
8 B, right). Only forced expression of DL4 but not DL1
resulted in the effi cient development of DP T cells (Fig. 8,
B–D). DL4-induced DP T cells were exclusively found in
the PBLs (13%), BM (86%), and spleen (48%) of Ctrl but not
of N1−/− chimeras, suggesting that enforced DL4 expression
can only induce T cell development of N1-expressing pro-
genitors in vivo. These results confi rm our in vitro results us-
ing the DL4-expressing OP9 cells.
D I S C U S S I O N
The data presented in this study provide compelling evidence
for a hierarchy of Notch–Delta interactions promoting T cell
lineage commitment and maturation. Using the well-char-
acterized OP9 stromal cell culture system, we unexpect-
edly found that DL1 can trigger T cell lineage commitment
Figure 6. Binding of purifi ed DL1- and DL4-IgG fusion proteins to
thymocytes. (A) DL1- (lanes 1 and 3) and DL4-IgG fusion proteins (lanes
2 and 4) before (lanes 1 and 2) and after (lanes 3 and 4) purifi cation over
a protein A column were stained with Coommassie blue (left). A Western
blot analysis of DL1- (lane 5) and DL4-IgG (lane 6) fusion proteins using
an anti–human IgG–horseradish peroxidase–conjugated antibody is also
shown (right). (B) The purifi ed DL1- and DL4-IgG fusion proteins were
used to stain immature WT thymocytes. Thymocytes were stained with
lineage cocktail and anti-CD117, -CD44, and -CD25 antibodies together
with DL1- or DL4-IgG fusion proteins. Representative histograms
show the staining of DL1- (gray line) and DL4-IgG (bold line) fusion
proteins or IgG isotype control (continuous line) gated on the DN1
(CD117+CD44+CD25−), DN2 (CD117+CD44+CD25+), DN3 (CD44−CD25+),
and DN4 (CD44−CD25−) thymocyte subpopulations. (C) Representative
histograms showing staining of DL1- (gray line) and DL4-IgG (bold line)
fusion proteins gated on ISP (CD8+TCRβ−), DP (CD4+CD8+), CD4SP
(CD4+TCRβ+), and CD8SP (CD8+TCRβ+) thymocytes. In the more mature
thymocyte subsets (ISP, DP, and SP), DL1-IgG staining was indistinguish-
able from the IgG isotype control, which is therefore not shown in C. Data
are representative of four independent experiments, and numbers within
the histograms indicate means ± SD of the mean fl uorescence intensity
for DL4. (D) Semiquantitative RT-PCR for the Notch target genes Deltex1
and Hes1. cDNA was prepared from DN thymocytes, either cultured for
20 h on IgG-, (DL1-IgG) DL1-Fc–, and DL4-Fc–coated plastic dishes or
on OP9-DL1 or -DL4 cells. Transcripts of Deltex1 and Hes1 were analyzed
by semiquantitative RT-PCR of threefold dilutions of the cDNA. The
cDNA input was normalized according to the expression of the control
HPRT gene.

Page 8

338 NOTCH–DELTA INTERACTIONS DURING T CELL DEVELOPMENT | Besseyrias et al.
(to the DN CD25+ stage) via either N1 or N2 receptors
expressed on BM precursors. In contrast, DL4 could only in-
duce T cell lineage commitment via N1. Concomitant with
T cell lineage commitment, inhibition of B cell development
was observed for N1–DL1, N1–DL4, and N2–DL1 receptor–
ligand pairs but not for N2–DL4. Collectively, our data dem-
onstrate novel aspects of both the specifi city and redundancy
of Notch receptor–ligand interactions in the induction of
T cell lineage commitment on OP9 stromal cells. Further-
more, they highlight the fact that N2 is unable to promote
thymic T cell lineage commitment in vivo despite its ability
to induce T cell commitment in vitro via DL1. This unex-
pected ability of N2 to substitute for N1 in the promotion
of T cell lineage commitment was also observed in vivo in a
short-term BM transplantation model, originally derived to
quantitate pluripotential stem cells (35). In this system, spleen
colonies in lethally irradiated recipients derived from single
BM stem cells contain multiple hematopoietic lineages, in-
cluding committed T cell precursors expressing Thy-1 and
CD25 (32, 33). Development of these extrathymic T cell
precursors was recently shown to depend upon Notch sig-
naling, because it is inhibited by a dominant-negative form
of the Notch coactivator mastermind-like 1 (31). Our data
using conditional Notch knockout BM precursors clearly
demonstrate that expression of either N1 or N2 is suffi cient
to promote T cell lineage commitment in the spleen after
transplantation. The relevant Notch ligands responsible for
T cell commitment in this system have not been identifi ed.
Nevertheless, it is clear that DL1 is expressed on both B cells
and DCs in the spleen (30), and DL1 is the nonredundant
ligand responsible for N2-mediated MZB cell fate specifi ca-
tion in that organ (30). By analogy with the OP9 system,
it is thus tempting to speculate that N2–DL1 interactions
account for the ability of N2 to substitute for N1 in promot-
ing extrathymic T cell lineage commitment after short-term
BM transplantation.
Notch–Delta interactions are not only required for T cell
lineage commitment but also for further maturation of DN
CD25+ immature thymocytes to the DP stage both in vivo
(25) and in vitro (14). A critical aspect of this maturation
process is productive VDJ rearrangement of the TCRβ lo-
cus, leading to expression of a TCRβ protein and functional
pre-TCR. In the OP9 system, it is known that both DL1 and
DL4 can promote TCRβ rearrangement and progression to the
DP stage when WT BM precursors are plated (18). However,
DL1 was unable to promote the further maturation of DN
CD25+ precursors that had developed via N2 (i.e., in the
absence of N1). Furthermore, this defect in the generation
of DP thymocytes was accompanied by a failure to express
TCRβ protein in DN CD25+ cells. Collectively, these data
indicate that N2–DL1 interactions, though suffi cient to pro-
mote T cell lineage commitment in vitro, are unable to in-
duce subsequent T cell maturation. Thus, it is possible that
N2–DL1 interactions in the OP9 system are of lower avid-
ity than N1–DL1 and, especially, N1–DL4 interactions, as
further suggested by direct binding assays of DL1 and DL4
fusion proteins on N1- or N2-transfected 293T cells. Never-
theless, N2–DL1 interactions are readily detected in the pres-
ence of high levels of Lfng in transfected 293T cells, raising
the possibility that the failure of N2–DL1 to induce T cell
maturation in vitro could be explained by limiting concen-
trations of Lfng in thymic progenitors. Alternatively, N2 may
signal less effi ciently than N1 because of its weaker transacti-
vation domain (4).
An even more stringent requirement for Notch–Delta
interactions in T cell maturation was observed in an in vivo
model where DL1 and DL4 were retrovirally transduced in
WT and N1−/− BM precursors that were subsequently used
to reconstitute lethally irradiated hosts. In agreement with
several other reports (13, 23, 24), large numbers of DP cells
developed after 9 wk in the BM, spleen, and PBLs of
mice reconstituted with DL4-expressing WT BM precursors.
In contrast, DL1-expressing WT BM precursors did not
Figure 7. Binding of DL1- and DL4-IgG fusion proteins to N1 and N2.
(A) Semiquantitative RT-PCR for the Lfng gene. cDNAs were prepared from
DN thymocytes and 293T cells. Transcripts were analyzed by semi-
quantitative RT-PCR of fi vefold dilutions of the cDNA. The cDNA input
was normalized according to the expression of the control HPRT gene.
293T cells were transiently transfected with N1- (B) or N2-EGFP (C)
together with or without Lfng expression vectors and stained 48 h after
transfection with either human IgG1 isotype control or DL1- or DL4-IgG
fusion proteins. Data are representative FACS profi les of four independent
experiments. Extremely high EGFP-expressing cells were gated out as the
fusion proteins were trapped inside the cells.

Page 9

JEM VOL. 204, February 19, 2007
339
ARTICLE
generate detectable numbers of DP cells. Importantly, DL4-
induced generation of DP cells was totally dependent on
N1, because it did not occur when reconstitution was
performed with DL4-transduced N1−/− BM precursors.
These results point to a highly specifi c N1–DL4 interaction
as being essential for extrathymic T cell maturation to the
DP stage in vivo when Delta expression is restricted to
hematopoietic cells. The unique N1–DL4 specifi city in
this system could be related to the fact that the N1–DL4
interaction is less dependent on Lfng than other Notch–
Delta interactions. According to this scenario, putative low
levels of Lfng in HSCs could restrict their ability to undergo
T cell maturation in the BM unless they encounter DL4
on hematopoietic cells. This stringent requirement may be
overcome in the OP9-DL1 system, where high levels of
expression of DL1 and/or other co-stimulatory properties of
OP9 stromal cells may compensate for the putatively weaker
N1–DL1 interaction.
The hierarchal nature of Notch–Delta interactions in im-
mature thymocytes raises the important issue of whether high
avidity Notch–Delta binding correlates with Notch signal-
ing. In this respect, analysis of Notch target genes in DN thy-
mocytes stimulated by DL1 or DL4 (expressed on OP9 cells
or immobilized on plastic) revealed a considerably better in-
duction of Deltex1 by DL4, consistent with the hierarchy of
ligand binding. However, induction of Hes1 in DN thymo-
cytes by DL1 and DL4 was comparable. Collectively, these
data favor a scenario in which diff erences in Notch–Delta
binding avidity translate into diff erences in some Notch sig-
naling outcomes but not in others, presumably because acti-
vation of downstream Notch signaling pathways is hierarchal
in nature. According to this model, the ability of N1–DL1
interactions to drive T cell lineage commitment and matura-
tion on OP9 cells in vitro would refl ect a low-threshold
Notch signaling requirement.
Finally, it is worth noting that the hierarchy of Notch–
Delta interactions described in this study has potential impli-
cations for T cell lineage commitment and maturation under
physiological conditions in the thymus. In this context,
it is of particular interest that N2 is capable of promoting
T cell commitment via interaction with DL1 (but not DL4)
both on OP9 stromal cells and during extrathymic T cell
development in the spleen. This raises the obvious question
of why N2 cannot compensate for N1 during T cell lin-
eage commitment in the thymus. Because N2 is expressed on
HSCs and early intrathymic T cell precursors, one possible
Figure 8. Comparative in vivo analysis of DL1 and DL4 for their
ability to promote ectopic T cell development. (A) Bar diagrams show
percentages of GFP+ Ctrl and N1−/− BM cells after transduction with the
control virus (MIG) and virus expressing DL1 and DL4. (B) Histograms
show the percentage of GFP+ cells within total PBLs in host mice 9 wk
after BM transplantation. Flow cytometric analysis for the presence of
CD4- and CD8-expressing T cells was performed on PBLs (B) and BM and
spleen (C) of host animals that were transplanted with WT and N1−/−
BM cells expressing the indicated ligands.

Page 10

340 NOTCH–DELTA INTERACTIONS DURING T CELL DEVELOPMENT | Besseyrias et al.
explanation for the inability of N2 to support T cell com-
mitment would be that DL1 is not present or available in
the thymus. Because N2 cannot promote T cell commitment
via DL4 even in the sensitive OP9 stromal system, it is
unlikely that N2–DL4 interactions would induce T cell
development in vivo. This scenario would therefore imply
that DL4 is in fact the physiological ligand for N1 during
intrathymic T cell development and, more generally, that
tissue-specifi c compartmentalization of Delta family members
is a mechanism to assure Notch receptor–ligand specifi city in
cell fate determination.
A second aspect of hierarchal Notch–Delta interac-
tions that is relevant to the identifi cation of the physiologi-
cal ligand of N1 during thymus development is the unique
ability of DL4 to specifi cally bind to immature thymocytes
with apparent high avidity. Conditional inactivation of N1
in BM precursors (8, 9) and at early stages of intrathymic
development (25) has clearly demonstrated that signaling
via N1 is required in immature DN thymocytes until they
have completed VDJβ rearrangement at the CD25+ DN3
stage, whereas later N1 inactivation (from the DN4 stage
onwards) has no impact on subsequent thymus development
(36). Interestingly, binding of DL4 by N1 in thymus subsets
closely parallels this functional requirement, because DL4
binding is high from the DN1 to DN3 stages and declines
in DN4 to become undetectable in subsequent DP and SP
stages. This result is again consistent with the hypothesis
that DL4 is the physiological N1 ligand responsible for both
T cell lineage commitment and subsequent thymic matu-
ration. Moreover, the strict correlation between N1–DL4
binding and N1 function during this process further suggests
that N1 signaling on developing thymocytes is regulated
at the level of ligand binding. Because N1–DL4 binding
activity appears to be relatively independent of Lfng (at least
in transfected 293T cells), it could be speculated that N1
function during thymic maturation is largely controlled at
the level of N1 expression.
Expression studies of DL1 and DL4 in the thymus, though
not defi nitive, also favor the hypothesis that DL4 may be the
physiological N1 ligand for T-lineage commitment and mat-
uration. Thus, semiquantative PCR analysis indicates that
DL4 is more strongly expressed than DL1 in the embryonic
(18) as well as adult (15, 30) thymus epithelium. More con-
vincingly, in situ hybridization studies, as well as lacZ gene–
targeted (knock-in) reporter mice demonstrated clearly that
DL4 is expressed at relatively high levels in situ in both em-
bryonic and adult thymus (13, 16, 17), whereas DL1 expres-
sion is barely or not detectable (unpublished data) (16, 37). At
the protein level, one group has reported broad expression of
DL1 in the adult thymus (14); however, the specifi city of the
polyclonal anti-DL1 antibody used in that study has been
challenged (15). Collectively, these data are consistent with
the possibility that DL4 rather than DL1 is the physiological
thymic ligand for N1. Indeed, conditional inactivation of
DL1 in the thymic epithelium does not impair T cell devel-
opment (30). Nevertheless, it remains possible that DL1 and
DL4 function redundantly in N1-mediated T cell lineage
commitment. Conditional gene targeting of DL4 will be
required to defi nitively resolve this important issue.
MATERIALS AND METHODS
Mice and induction of the Cre-mediated inativation of the fl oxed
N1 and N2 genes. Experiments were performed according to Swiss guide-
lines and authorized by the veterinary authorities of the Canton de Vaud
(authorization nos. 1099.2 and 1099.3). CD45.1+ C57BL/6 mice were pur-
chased from the Jackson Laboratory. N1lox/lox&Mx-Cre mice were generated as
previously described (8). For N2lox/lox mice, the generation of the N2 target-
ing vector was based on a genomic DNA fragment including the exons b to h
and the 5′ part of exon i (according to the nomenclature previously described
[38]) from the mouse N2 locus. Exons d and e (coding for the C-terminal
part of the RAM23 domain and nuclear localization sequence) were fl anked
by loxP sites. Cre recombinase–mediated N2 inactivation results in the
expression of a truncated N2 protein, lacking the intracellular part. The gen-
erated mice carrying the loxP-targeted exons d and e are referred to here as
the fl oxed N2 allele (N2lox).
Generation of the N2 targeting vector and targeting of embryonic
stem (ES) cells. Construction of the targeting vector based on the geno-
mic N2 phage DNA clone 1NT2-2 (provided by Y. Hamada, National
Institute of Basic Biology, Okazaki, Japan) containing the distal exons a to i
(exon c turned out to consist of two small neighboring exons, designated
by us as c1 and c2). A 9.5-kb SphI fragment from 1NT2-2 encompassing
N2 exons b to h and the 5′ part from exon i was subcloned into the
backbone vector pHEBOpl-mod, which was previously created from the
plasmid pHEBO (39) by replacement of a 630-bp ClaI/SalI fragment with a
polylinker sequence and subsequent deletion of a 2.9-kb MluI fragment.
Within the genomic N2 SphI fragment, insertions were made at three posi-
tions: (a) at the BsrBI site between exons c2 and d, a loxP-fl anked neomycin
resistance gene cassette (NeoR) from pEasyFlox (provided by Marat Alim-
zhanov, Harvard Medical School, Boston, MA; unpublished data; reference 40)
containing an additional SacI/SstI recognition site between the 5′ loxP
site and NeoR was inserted; (b) at the XhoI site between exons e and f, a
loxP sequence with a SacI/SstI recognition site at its 3′ end was inserted;
and (c) a thymidine-kinase expression cassette (XhoI/PvuI fragment from
pEasyFlox) was cloned into the SphI site at the 3′ end of the genomic N2
sequence (within exon i).
The fi nal targeting vector pU1496-21 was sequenced, linearized by
NotI, and electroporated into BALB/c-derived ES cells. G418-resistant and
ganciclovir-sensitive colonies were screened for homologous recombination
by Southern blot analysis. The cellular DNA was digested with SstI (a SacI
isoschizomer) and hybridized with a specifi c N2 probe (a 490-bp SacI/SphI
fragment located just upstream of the 5′ end of the genomic N2 sequence in
the targeting vector). The WT N2 and the targeted N2 alleles were identi-
fi ed as 7.2- and 2.9-kb fragments, respectively. Clones with homologous
recombination were further confi rmed by hybridization with an internal probe
(NeoR) and a 3′ external probe (271-bp SphI/SacI fragment from exon i).
Of 768 analyzed ES cell clones, four exhibited correct recombination with
the targeting vector.
In one correctly targeted ES cell clone (1G10), the loxP-fl anked NeoR
was deleted in vitro by transfection with the cre expression vector pIC-cre
(41). The resulting single-cell clones were screened for correct deletion of
the NeoR cassette by Southern blotting, as decribed in the previous para-
graph. The fl oxed N2 allele could be identifi ed as a 5.7-kb fragment. Two
of the ES cell clones with NeoR deletions were reconfi rmed by sequencing
and injected into C57BL/6 blastocysts, which were then transferred into
foster mothers to obtain chimeric mice. Floxed N2 gene –targeted mice were
provided by K. Rajewsky (Harvard Medical School, Boston, MA).
Activation of the Cre recombinase was performed as previously de-
scribed (8, 9). In brief, Ctrl, (N1lox/lox), N1lox/lox&Mc-Cre, N2lox/lox&Mx-Cre, and
N1/N2lox/lox&Mx-Cre mice received fi ve i.p. injections of 250 μg polyI-polyC

Page 11

JEM VOL. 204, February 19, 2007
341
ARTICLE
(pIpC; Sigma-Aldrich) at 2-d intervals. Competitive mixed BM chimeras
were set up as previously described (9). In brief, lethally irradiated mice (950
rads 24 h before transfer) that had been treated i.p. 48 h before BM trans-
plantation with 100 μg anti-NK1.1 monoclonal antibodies were reconsti-
tuted with a 1:2 mixture (5 × 106:10 × 106) of CD45.1+ WT and Ctrl,
N1lox/lox&Mc-Cre, N2lox/lox&Mx-Cre, or N1/N2lox/lox&Mx-Cre BM for mixed chimeras.
Mice were maintained on antibiotics (Bactrim) containing water, and re-
constitution of BM and lymphoid organs by donor-derived cells was analyzed
8 wk later.
Flow cytometry and cell sorting. The following monoclonal anti-
body conjugates were purchased from eBioscience: CD117 (2B8)-PE and
-PE-Cy5.5; Sca-1 (D7)-PE and -APC; CD19 (MB-19.1)-PE and (6D5)-
PE-Cy5.5; B220 (RA3.6B2)-PE-Cy5.5 and –Alexa Fluor 647; CD44
(IM781)-PE-Cy5.5; CD25 (PC61)-APC; CD4 (L3T4)-PE-Cy5.5; CD45.2
(104)-PE-Cy5.5; TCRβ (H57)-PE and -APC; CD161 (PK136)-FITC;
CD90.1 (HIS15)-PE; and CD90.2 (30H12)-PE. TCRγδ (GL3)-PE and
TCRβ (H57)-biotin were purchased from BD Biosciences. CD25 (PC61)-
PE was purchased from Caltag. CD19 (ID3)–Alexa Fluor 647; B220
(RA3.6B2)–Alexa Fluor 647; CD4 (GK1.5)-FITC and -PE; CD8α (53.6.7)-
FITC and –Alexa Fluor 647; CD45.1 (A20)–Alexa Fluor 647; CD45.2
(104)–Alexa Fluor 647; TCRβ (H57)-FITC; TCRγδ (GL3)-FITC; Gr1
(RB6.8C5)-FITC; Ter119-FITC; CD11b (M1/70)-FITC, -PE, and –Alexa
Fluor 647; and CD3 (17A2)-FITC were purifi ed from hybridoma super-
natants and conjugated in our laboratory according to standard protocols.
Alexa Fluor 647 conjugates were prepared using the appropriate Alexa Fluor
protein labeling kits (Invitrogen). APC and PE conjugates were prepared
using kits purchased from Prozyme. Intracellular staining for TCRβ was
performed as previously described (42). Single-cell suspensions were stained
with the respective antibodies and analyzed using a FACSCalibur or FAC-
Scanto fl ow cytometer (Becton Dickinson). The cells were sorted with a
FACSVantage or a FACSAria fl ow cytometer (Becton Dickinson). Dead
cells and debris were eliminated by appropriate gating on forward and side
scatter. The data were analyzed using either CellQuest Pro (BD Biosciences)
or FlowJo (TreeStar, Inc.) software.
OP9 cell co-cultures. OP9 stromal cells engineered to express GFP and
the mouse DL1 gene (OP9-DL1 cells, provided by J.C. Zuniga-Pfl ücker,
University of Toronto, Toronto, Canada) or GFP and the mouse DL4 gene
(OP9-DL4 cells provided by A. Cumano, Institut Pasteur, Paris, France)
were cultured in αMEM supplemented with 20% FBS (Sigma-Aldrich).
HSCs were isolated from adult mouse BM and sorted as lin−CD117hiSca-1hi.
HSCs were seeded at 4 × 103 cells/well onto 80% confl uent monolayers of
OP9 cells (24-well plates) in DMEM with 10% FBS. Every third day, 1 ml
of the culture supernatant was exchanged with fresh medium. Co-cultures
were harvested by pipetting at the time points indicated in the fi gures, and
contaminating OP9 cells were eliminated by fi ltering the lymphocytes
through a 70-μm cell strainer (BD Biosciences) before replating or fl ow cy-
tometric analysis. All co-cultures were performed in the presence of 5 ng/ml
rmIL-7 and rhFlt3L (PeproTech).
Semiquantitative RT-PCR. Total RNA was isolated using TRIzol reagent
(Invitrogen), and semiquantitative PCR was performed using the Onestep
RT-PCR kit (QIAGEN). All PCR reactions were performed using the same
serially diluted RNA samples normalized to an α tubulin– or hypoxanthine
guanine phosphoribosyl transferase (HPRT)–specifi c signal. Gene–specifi c
primer sequences were as follows: N1, (forward) 5′-T G T G A C A G C C A G T G-
C A A C T C -3′ and (reverse) 5′-G C A G T G C T T C C A G A G T G C C A -3′; N2,
(forward) 5′-A C A T C A T C A C A G A C T T G G T C -3′ and (reverse) 5′-G G C-
A G C T G C T G T C A A T A A T G -3′; tubulin, (forward) 5′-T C A C T G T G C C T-
G A A C T T A C C -3′ and (reverse) 5′-G G A A C A T A G C C G T A A A C T G C -3′;
mouse and human HPRT, (forward) 5′-A A G G A G A T G G G A G G C C A T C-
A C -3′ and (reverse) 5′-C T T G T C T G G A A T T T C A A A T C C A A C -3′; Del-
tex1, (forward) 5′-C A C T G G C C C T G T C C A C C C A G C C T T G G C A G G -3′
and (reverse) 5′-G G G A A G G C G G G C A A C T C A G G C C T C A G G -3′; Hes1,
(forward) 5′-A T C A T G G A G A A G A G G C G A A G G G -3′ and (reverse) 5′-T G A-
T C T G G G T C A T G C A G T T G G -3′; and mouse and human Lfng, (forward)
5′-C G C G C C A C A A G G A G A T G A C G T T C -3′ and (reverse) 5′-T G G G C-
A C C T G C T G C A G G T T C T -3′. PCR products were resolved by agarose
gel electrophoresis and visualized by ethidium bromide staining. All PCR
products shown correspond to the expected molecular size (Figs. 1, 4,
6, and 7).
Purifi ed DL1- and DL4-IgG were also used to stimulate in vitro DN
thymocytes. In brief, 24-well plates were coated with 10 μg/ml protein A
overnight at 4°C. 10 μg/ml of purifi ed DL1- and DL4-IgG was added to the
plates for 1 h at 4°C. Lineage-depleted DN thymocytes were subsequently
plated on DL fusion protein–coated plates at 2 × 106 cells/well. After 20 h
of culture, DN thymocytes were harvested to analyze gene expression by
semiquantitative RT-PCR.
Spleen CFU (CFU-S) assay. CD45.1+ C57BL/6 mice were exposed to
lethal whole-body irradiation (950 rads) from a 137Cs source and maintained
on water containing antibiotics (Bactrim). The next day, total BM cells
(12 × 104) from either Ctrl, N1−/−, N2−/−, or N1N2−/− mice were injected
i.v. via the retroorbital sinus. After 12 d, spleens of the recipient mice were
removed, and single-cell suspensions were prepared for FACS analysis.
In vitro stimulation of 5-FU–treated BM cells. 1 wk after the last polyI-
polyC injection, N1lox/lox and N1lox/lox&Mx-Cre mice were treated with 3 mg/20 g
body weight of 5-fl uorouracil (Sigma-Aldrich). 5 d later, BM cells were
harvested and, after red blood cell lysis, single-cell suspensions were prepared
in stem cell–activating (SA) medium containing IMDM supplemented with
10% FBS, 100 ng/ml rmSCF (R&D Systems), 50 ng/ml rmTPO (R&D
Systems), and 50 ng/ml rmFlt-3L (R&D Systems). The cells were incubated
overnight at 37°C in 5% CO2 before the retroviral infection.
Retrovirus production and infection procedure. Empty (MIG) or re-
combinant (DL1 and DL4) retroviruses (provided by A. Freitas, Institut Pas-
teur, Paris, France) were obtained after transfection of Bosc23 packaging
cells using lipofectamine 2000 (Invitrogen). Retrovirus-containing super-
natants were collected 48 h after transfection. 3-cm petri dishes were coated
with 1 ml of RetroNectin (r-fi bronectin fragment CH-296; TaKara) solu-
tion (12.5 μg/ml in PBS) for 2 h at room temperature. After removal of the
RetroNectin, 2 ml PBS + 2% BSA was added to each dish for 30 min. After
washing the plates with PBS, the retroviral supernatant was added to the
coated plates for 1 h at 37°C. The prestimulated BM cells were spun down,
resuspended in 1 ml of fresh SA medium, and added to the retroviral super-
natant. The next day, 1–3 × 106 cells were injected i.v. into lethally irradi-
ated CD45.1+ C57BL/6 mice.
To ensure similar functionality and/or expression of DL1 and DL4, the
retroviruses were assessed before BM infection by Western blot analysis,
as well as by infecting OP9 cells that were subsequently tested for their
ability to induce T cell development in vitro. Based on these criteria, DL1
and DL4 retroviruses were equivalent (Fig. S1, available at http://www.jem
.org/cgi/content/full/jem.20061442/DC1).
Expression plasmids and transfections. The mouse cDNA coding for
the extracellular domain of either DL1 or DL4 was cloned via HindIII/
BamHI and EcoRI/SalI, respectively, into the PS 521 expression vector (43)
to generate DL1- and DL4-IgG fusion proteins. The corresponding expres-
sion vectors were transfected into 293T cells using the calcium-phosphate
method, and IgG fusion proteins were subsequently purifi ed over protein A
columns according to the manufacturer’s instructions (HiTrap rProtein A
FF; GE Healthcare). Purity of the fusion proteins was verifi ed by Coomassie
blue staining and Western blot analysis. Full-length N1 and N2 cDNAs were
cloned in frame into the pEGFP-N1 expression vector (CLONTECH Lab-
oratories, Inc.). The mouse cDNA coding for the Lfng was cloned via
BamHI into pcDNA3.1− (Invitrogen). 293T cells were transiently trans-
fected with 5 μg N1- and N2-EGFP expression vectors, respectively, with
or without the Lfng expression vector. 48 h later, the cells were indirectly

Page 12

342 NOTCH–DELTA INTERACTIONS DURING T CELL DEVELOPMENT | Besseyrias et al.
stained with DL1- or DL4-IgG fusion proteins (0.5 μg/106 cells). Binding
of the fusion proteins to the Notch receptors was detected using biotinylated
anti–human IgG antibodies.
Online supplemental material. Fig. S1 shows a functional test for DL1-
and DL4-expressing retroviruses. Online supplemental material is available at
http://www.jem.org/cgi/content/full/jem.20061442/DC1.
We thank F. Grosjean and E. Devevre for cell sorting, P. Schneider for the PS521
expression vector and helpful suggestions, and A. Wilson for discussions and critical
reading of the manuscript.
This work was supported in part by the Swiss National Science Foundation
(Förderungsprofessur to F. Radtke), a Marie Heim-Vögtlein Fellowship (PMP DB-
110307/1 to E. Fiorini), the Swiss Cancer League, and grants from the Deutsche
Forschungsgemeinschaft (SFP 243, STR-461/3-2, und SFB 684).
The authors have no confl icting fi nancial interests.
Submitted: 7 July 2006
Accepted: 30 December 2006
R E F E R E N C E S
1. Rothenberg, E.V. 2000. Stepwise specifi cation of lymphocyte develop-
mental lineages. Curr. Opin. Genet. Dev. 10:370–379.
2. Godfrey, D.I., and A. Zlotnik. 1993. Control points in early T-cell
development. Immunol. Today. 14:547–553.
3. Rodewald, H.R., and H.J. Fehling. 1998. Molecular and cellular events
in early thymocyte development. Adv. Immunol. 69:1–112.
4. Radtke, F., A. Wilson, S.J. Mancini, and H.R. MacDonald. 2004.
Notch regulation of lymphocyte development and function. Nat.
Immunol. 5:247–253.
5. Maillard, I., T. Fang, and W.S. Pear. 2005. Regulation of lymphoid de-
velopment, diff erentiation, and function by the Notch pathway. Annu.
Rev. Immunol. 23:945–974.
6. Artavanis-Tsakonas, S., M.D. Rand, and R.J. Lake. 1999. Notch sig-
naling: cell fate control and signal integration in development. Science.
284:770–776.
7. Haines, N., and K. Irvine. 2003. Glycosylation regulates Notch signal-
ling. Nat. Rev. Mol. Cell Biol. 4:786–797.
8. Radtke, F., A. Wilson, G. Stark, M. Bauer, J. van Meerwijk, H.R.
MacDonald, and M. Aguet. 1999. Defi cient T cell fate specifi ca-
tion in mice with an induced inactivation of Notch1. Immunity. 10:
547–558.
9. Wilson, A., H.R. MacDonald, and F. Radtke. 2001. Notch1-defi cient
common lymphoid precursors adopt a B cell fate in the thymus. J. Exp.
Med. 194:1003–1012.
10. Han, H., K. Tanigaki, N. Yamamoto, K. Kuroda, M. Yoshimoto, T.
Nakahata, K. Ikuta, and T. Honjo. 2002. Inducible gene knockout of
transcription factor recombination signal binding protein-J reveals its
essential role in T versus B lineage decision. Int. Immunol. 14:637–645.
11. Hasserjian, R.P., J.C. Aster, F. Davi, D.S. Weinberg, and J. Sklar. 1996.
Modulated expression of notch1 during thymocyte development. Blood.
88:970–976.
12. Felli, M.P., M. Maroder, T.A. Mitsiadis, A.F. Campese, D. Bellavia,
A. Vacca, R.S. Mann, L. Frati, U. Lendahl, A. Gulino, and I.
Screpanti. 1999. Expression pattern of notch1, 2 and 3 and Jagged1
and 2 in lymphoid and stromal thymus components: distinct ligand-
receptor interactions in intrathymic T cell development. Int. Immunol.
11:1017–1025.
13. Yan, X.Q., U. Sarmiento, Y. Sun, G. Huang, J. Guo, T. Juan, G. Van,
M.Y. Qi, S. Scully, G. Senaldi, and F.A. Fletcher. 2001. A novel Notch
ligand, Dll4, induces T-cell leukemia/lymphoma when overexpressed
in mice by retroviral-mediated gene transfer. Blood. 98:3793–3799.
14. Schmitt, T.M., M. Ciofani, H.T. Petrie, and J.C. Zuniga-Pfl ucker.
2004. Maintenance of T cell specifi cation and diff erentiation requires re-
current Notch receptor–ligand interactions. J. Exp. Med. 200:469–479.
15. Lehar, S.M., J. Dooley, A.G. Farr, and M.J. Bevan. 2005. Notch ligands
Delta 1 and Jagged1 transmit distinct signals to T-cell precursors. Blood.
105:1440–1447.
16. Tsukamoto, N., M. Itoi, M. Nishikawa, and T. Amagai. 2005. Lack of
Delta like 1 and 4 expressions in nude thymus anlages. Cell. Immunol.
234:77–80.
17. Benedito, R., and A. Duarte. 2005. Expression of Dll4 during mouse
embryogenesis suggests multiple developmental roles. Gene Expr.
Patterns. 5:750–755.
18. Mohtashami, M., and J.C. Zuniga-Pfl ucker. 2006. Three-dimensional
architecture of the thymus is required to maintain delta-like expression
necessary for inducing T cell development. J. Immunol. 176:730–734.
19. Saito, T., S. Chiba, M. Ichikawa, A. Kunisato, T. Asai, K. Shimizu, T.
Yamaguchi, G. Yamamoto, S. Seo, K. Kumano, et al. 2003. Notch2 is
preferentially expressed in mature B cells and indispensable for marginal
zone B lineage development. Immunity. 18:675–685.
20. Krebs, L.T., Y. Xue, C.R. Norton, J.P. Sundberg, P. Beatus, U.
Lendahl, A. Joutel, and T. Gridley. 2003. Characterization of Notch3-
defi cient mice: normal embryonic development and absence of genetic
interactions with a Notch1 mutation. Genesis. 37:139–143.
21. Pui, J.C., D. Allman, L. Xu, S. DeRocco, F.G. Karnell, S. Bakkour,
J.Y. Lee, T. Kadesch, R.R. Hardy, J.C. Aster, and W.S. Pear. 1999.
Notch1 expression in early lymphopoiesis infl uences B versus T lineage
determination. Immunity. 11:299–308.
22. De Smedt, M., K. Reynvoet, T. Kerre, T. Taghon, B. Verhasselt,
B. Vandekerckhove, G. Leclercq, and J. Plum. 2002. Active form
of Notch imposes T cell fate in human progenitor cells. J. Immunol.
169:3021–3029.
23. Dorsch, M., G. Zheng, D. Yowe, P. Rao, Y. Wang, Q. Shen, C.
Murphy, X. Xiong, Q. Shi, J.C. Gutierrez-Ramos, et al. 2002. Ectopic
expression of Delta4 impairs hematopoietic development and leads to
lymphoproliferative disease. Blood. 100:2046–2055.
24. de La Coste, A., E. Six, N. Fazilleau, L. Mascarell, N. Legrand, M.P.
Mailhe, A. Cumano, Y. Laabi, and A.A. Freitas. 2005. In vivo and in
absence of a thymus, the enforced expression of the notch ligands delta-1
or delta-4 promotes T cell development with specifi c unique eff ects.
J. Immunol. 174:2730–2737.
25. Wolfer, A., A. Wilson, M. Nemir, H.R. MacDonald, and F. Radtke.
2002. Inactivation of Notch1 impairs VDJbeta rearrangement and allows
pre-TCR-independent survival of early alpha beta lineage thymocytes.
Immunity. 16:869–879.
26. Mancini, S.J., N. Mantei, A. Dumortier, U. Suter, H.R. Macdonald,
and F. Radtke. 2005. Jagged1-dependent Notch signaling is dispens-
able for hematopoietic stem cell self-renewal and diff erentiation. Blood.
105:2340–2342.
27. Jiang, R., Y. Lan, H.D. Chapman, C. Shawber, C.R. Norton, D.V.
Serreze, G. Weinmaster, and T. Gridley. 1998. Defects in limb, cra-
niofacial, and thymic development in Jagged2 mutant mice. Genes Dev.
12:1046–1057.
28. Jaleco, A.C., H. Neves, E. Hooijberg, P. Gameiro, N. Clode, M.
Haury, D. Henrique, and L. Parreira. 2001. Diff erential eff ects of Notch
ligands Delta-1 and Jagged-1 in human lymphoid diff erentiation. J. Exp.
Med. 194:991–1002.
29. Schmitt, T.M., and J.C. Zuniga-Pfl ucker. 2002. Induction of T cell de-
velopment from hematopoietic progenitor cells by delta-like-1 in vitro.
Immunity. 17:749–756.
30. Hozumi, K., N. Negishi, D. Suzuki, N. Abe, Y. Sotomaru, N. Tamaoki,
C. Mailhos, D. Ish-Horowicz, S. Habu, and M.J. Owen. 2004. Delta-
like 1 is necessary for the generation of marginal zone B cells but not
T cells in vivo. Nat. Immunol. 5:638–644.
31. Maillard, I., B.A. Schwarz, A. Sambandam, T. Fang, O. Shestova, L.
Xu, A. Bhandoola, and W.S. Pear. 2006. Notch-dependent T-lineage
commitment occurs at extrathymic sites following bone marrow trans-
plantation. Blood. 107:3511–3519.
32. Lancrin, C., E. Schneider, F. Lambolez, M.L. Arcangeli, C. Garcia-
Cordier, B. Rocha, and S. Ezine. 2002. Major T cell progenitor
activity in bone marrow–derived spleen colonies. J. Exp. Med. 195:
919–929.
33. Arcangeli, M.L., C. Lancrin, F. Lambolez, C. Cordier, E. Schneider,
B. Rocha, and S. Ezine. 2005. Extrathymic hemopoietic progenitors
committed to T cell diff erentiation in the adult mouse. J. Immunol.
174:1980–1988.

Page 13

JEM VOL. 204, February 19, 2007
343
ARTICLE
34. Visan, I., J.B. Tan, J.S. Yuan, J.A. Harper, U. Koch, and C.J. Guidos.
2006. Regulation of T lymphopoiesis by Notch1 and Lunatic fringe-
mediated competition for intrathymic niches. Nat. Immunol. 7:634–643.
35. Till, J.E., and C.E. McCulloch. 1961. A direct measurement of the
radiation sensitivity of normal mouse bone marrow cells. Radiat. Res.
14:213–222.
36. Wolfer, A., T. Bakker, A. Wilson, M. Nicolas, V. Ioannidis, D.R.
Littman, P.P. Lee, C.B. Wilson, W. Held, H.R. MacDonald, and F.
Radtke. 2001. Inactivation of Notch 1 in immature thymocytes does not
perturb CD4 or CD8 T cell development. Nat. Immunol. 2:235–241.
37. Visan, I., J.S. Yuan, J.B. Tan, K. Cretegny, and C.J. Guidos. 2006.
Regulation of intrathymic T-cell development by Lunatic Fringe-
Notch1 interactions. Immunol. Rev. 209:76–94.
38. Hamada, Y., Y. Kadokawa, M. Okabe, M. Ikawa, J.R. Coleman, and
Y. Tsujimoto. 1999. Mutation in ankyrin repeats of the mouse Notch2
gene induces early embryonic lethality. Development. 126:3415–3424.
39. Sugden, B., K. Marsh, and J. Yates. 1985. A vector that replicates as a
plasmid and can be effi ciently selected in B-lymphoblasts transformed by
Epstein-Barr virus. Mol. Cell. Biol. 5:410–413.
40. Schenten, D., V.L. Gerlach, C. Guo, S. Velasco-Miguel, C.L. Hladik,
C.L. White, E.C. Friedberg, K. Rajewsky, and G. Esposito. 2002. DNA
polymerase kappa defi ciency does not aff ect somatic hypermutation in
mice. Eur. J. Immunol. 32:3152–3160.
41. Gu, H., Y.R. Zou, and K. Rajewsky. 1993. Independent control
of immunoglobulin switch recombination at individual switch re-
gions evidenced through Cre-loxP-mediated gene targeting. Cell.
73:1155–1164.
42. Wilson, A., M. Capone, and H.R. MacDonald. 1999. Unexpectedly
late expression of intracellular CD3epsilon and TCR gammadelta pro-
teins during adult thymus development. Int. Immunol. 11:1641–1650.
43. Schneider, P. 2000. Production of recombinant TRAIL and TRAIL
receptor: Fc chimeric proteins. Methods Enzymol. 322:325–345.