Active Form of Notch Members Can Enforce T Lymphopoiesis on Lymphoid Progenitors in the Monolayer Culture Specific for B Cell Development¹

In the ontogenical process, stem cells and progenitor cells are exposed to numerous cell fate-related choices. These govern their development into certain cell lineages by certain gene expressions regulating cell commitment in a lineage-specific fashion. T and B lymphoid cells are recognized as being generated from hemopoietic stem cells (HSC)³ through a common lymphoid progenitor stage. Recent reports demonstrated several molecules that determine cell fate at the branch point for T and B lineages from lymphoid progenitors. For B cells, the products of E2A and Pax5 are specifically required at the early development stages in hemopoietic cells. Without these proteins, the initiation of B cell development is arrested (1, 2). For T lymphopoiesis, Notch1 protein is well accepted as such a key molecule specific for T lineage determination.

The Notch genes encode transmembrane receptors that are highly conserved from invertebrate to mammals. These receptors bind to their ligands, such as Delta-like (Dll) or Jagged family members, resulting in the generation of a proteolytic fragment of the intracellular region of Notch (ICN), which moves to the nucleus as a signal transducer (3, 4). Notch-mediated signaling regulates the progression of cell differentiation or specification of cell fate at the branch point of distinct cell lineages, contributing to a large number of developmental systems (5). Recently, it was reported that conditional targeting of the Notch1 gene results in the specific defect of T cell development and a significant increase of B cells in the thymus (6). A reciprocal experimental result also showed that the enforced expression of the active form of Notch, ICN, induced the ectopic appearance of CD4 and CD8 double-positive (DP) cells in bone marrow and inhibited B cell development (7). These results demonstrated that Notch1 has an indispensable role in the specification of T cell lineage at the branch point. However, the molecular mechanism for these phenomena remains to be clarified.

To resolve the detailed mechanism of how Notch is involved in T cell development, a simple culture system inducing T cell development is needed, similar to that of B cells. Since certain cell lines were established from bone marrow stromal cells, studies on B cell development have progressed well on the basis of hemopoietic cell cultivations on monolayers of cell lines (8, 9). In contrast, the in vitro development of T cells is not easily induced on the monolayer of stromal cells, but requires the stereoscopic structure of thymic stromal cells such as fetal thymus organ culture (FTOC) or reaggregation thymus organ culture (10). Moreover, it is still unresolved why such a particular structure is important for T cell development, even though that seems to mimic the thymic organ. It is also unclear which molecule(s) are involved in the thymus-specific ability for T cell development.

To address the issue of how Notch proteins control T cell commitment, we tried to establish a culture system for testing the induction effect of Notch signaling without a thymic environment. Using the in vitro system we established, we demonstrated for the first time that the introduction of the ICN1 can induce T cells expressing TCR on the cell surface in the absence of a thymic environment but in a B cell-specific culture condition. We also showed for the first time that other Notch family members, Notch2 and Notch3, have the potential to induce T cell development as well as Notch1 does.

The retroviral vectors MIGR1, MIG ICN1, MIG dANK (7), and pMY-IRES-GFP were kindly donated by Drs. W. Pear (University of Pennsylvania Medical Center, Philadelphia, PA) and T. Kitamura (University of Tokyo, Tokyo, Japan), and murine ICN1, ICN2, or ICN3 were constructed by cloning into EcoRI/NotI sites of pMY-IRES-GFP. These vectors were transfected transiently into BOSC23, and the culture supernatants were harvested as the source of the retrovirus. The purified lineage markers (Lin; TER119, Gr-1, Mac-1, and CD19)-negative, c-kit-positive cells (>98% purity) from embryonic day 15.5 (E15.5) murine fetal liver (FL) were infected with the above retrovirus through centrifugation (7500 rpm; 1.5 h) in the presence of polybrene (8 μg/ml) and stem cell factor (SCF; 100 ng/ml) and further cultured at 37°C for 3 h. After washing with the medium, cells were used as the gene-transfected hemopoietic precursors that could be traced by the expression of green fluorescence protein (GFP).

The gene-transfected hemopoietic precursors (10⁴) were seeded on the monolayer of OP-9 (11, 12) or PA-6 (9) (these cell lines were kindly donated by Drs. T. Nakano (Osaka University, Osaka, Japan) and S. Nishikawa (Kyoto University, Kyoto, Japan), respectively) in 6-well culture plates (Corning, Corning, NY) with IL-7 (10 ng/ml) or IL-7 and SCF (50 ng/ml) (PeproTech, London, U.K.) for 1 wk. To interrupt the direct contact between hemopoietic cells and the stromal layer, the cell culture insert with a 0.4-μm pore size membrane (BD Biosciences Labware, Franklin Lakes, NJ) was used in the culture. After the culture, growing cells were harvested and analyzed by flow cytometry. For further culture, gene-transfected cells were sorted out by FACStar^Plus (BD Biosciences, Mountain View, CA) and seeded again on newly prepared stromal layers.

FITC-conjugated TER119 and biotinylated c-kit (ACK4) mAbs were produced in our laboratory. FITC-conjugated Mac-1 (M1/70) and CD19 (1D3), PE-conjugated CD4 (GK1.5), allophycocyanin-conjugated CD8 (53-6.7) and CD44 (IM7), and biotinylated CD45.2 (104) mAbs were purchased from BD Biosciences. FITC-conjugated Gr-1 (RB6-8C5); PE-conjugated B220 (RA3-6B2), Thy1.2 (53-2.1), TCR γδ (UC7-13D5), and CD25 (PC61.5); allophycocyanin-conjugated CD19; and Cy-5-conjugated Gr-1 mAbs were purchased from eBioscience (San Diego, CA). Flow cytometric analysis was conducted as described previously (13). The cytoplasmic staining for CD3ε chain was performed as follows: 2 × 10⁵ cells were fixed in 1% paraformaldehyde for 10 min at room temperature, permeabilized, and pretreated with 0.5% saponin buffer (10 mM HEPES (pH 7.4), 5% FCS) containing 5 μg of hamster IgG1 (BD Biosciences) or CD3ε mAb (145-2C11; hamster IgG1; eBioscience) for 10 min. Allophycocyanin-conjugated hamster IgG1 (BD Biosciences) or CD3ε mAb (145-2C11; eBioscience) (0.1 μg each) was then added to the cell suspension. After 20 min, cells were washed with 0.1% saponin buffer and analyzed by FACSCalibur (BD Biosciences) using CellQuest software (BD Biosciences).

mRNA from sorted GFP-positive cells (>99% purity) from the culture on the monolayer with IL-7, which were transfected with MIGR1, ICN1, or dANK, was isolated by RNeasy mini kit (Qiagen, Hilden, Germany), and transcribed with oligo(dT) and Ominiscript reverse transcriptase (Qiagen). PCR was performed for 35 cycles for 45 s at 94°C, 1 min at 58°C, and 1 min at 72°C. PCR products were electrophoresed on 1.2% agarose gel and visualized by ethidium bromide staining. The oligonucleotides used to detect GATA3, pTα, and β₂-microglobulin (β₂m) were the following: GATA3 sense, 5′-TCTCACTCTCGAGGCAGCATGT-3′, GATA3 antisense, 5′-GGTACCATCTCGCCGCCACAG-3′; pTα sense, 5′-GGCACCCCCTTTCCGTCTCT-3′, pTα antisense, 5′-GTCCAAATTCTGTGGGTGGGA-3′; λ5 sense, 5′-CTTGAGGGTCAATGAAGCTCAGAGTA-3′, λ5 antisense, 5′-CTTGGGCTGACCTAGGATTG-3′; and β₂m sense, 5′-ATGGCTCGCTCGGTGACCCTA-3′, β₂m antisense, 5′-TCATGATGCTTGATCACATGTCTC-3′.

Fetal thymic lobes were obtained from E15.5 C57BL/6 embryos and placed on insert filters (Nucleopore, Whatman, NJ) in 24-well culture plates (Corning) at the air-medium interface. The lobes were cultured with 1.35 mM 2-deoxyguanosine (dGuo) for 5–6 days and then rinsed with medium without dGuo. Sorted GFP-positive cells from the cultures were aliquotted at 5000/well in Terasaki plates (Sumitomo Bakelite, Tokyo, Japan), and one dGuo-treated lobe per well was added. The cells and lobes were incubated for 48 h as hanging drop cultures. After 48 h, the lobes were removed, rinsed, and cultured on insert filters in 24-well plates for 5 days.

In this study, Lin-negative and c-kit-positive FL cells (c-kit⁺ FL cells) of day 15.5 (E15.5) were used as a source of hemopoietic progenitor cells. To examine the effect of Notch1-mediated signaling on T cell development in vitro, the isolated c-kit⁺ FL cells were transfected with retrovirus vector encoding ICN1 and GFP, as described previously (7). Two other vectors used as control carried GFP plus a nonfunctional mutant of ICN1 (dANK) or GFP gene alone (MIGR), respectively. Just after the infection, the c-kit⁺ FL cells were seeded on the monolayer of bone marrow-derived stromal cell line, OP-9, which was originally established for B cell development (11, 12). One week after the culture with IL-7, the cultured cells showed vigorous growth (>100 times) with a relatively large proportion of GFP⁺ cells (>25%) (Fig. 1). In the cultures of the c-kit⁺ FL cells transfected with MIGR or dANK, most of the harvested cells expressed B cell markers such as B220 and CD19, and a few cells were Gr-1⁺ myeloid cells. This expression pattern is ordinarily found in the culture with hemopoietic progenitor cells on OP-9. In contrast, almost all GFP⁺ cells transfected with ICN1 came to express a T cell marker, Thy-1, at the same high density as typical thymocytes, but not CD19 or Gr-1. The Thy1⁺ cells were CD4 and CD8 double-negative (DN), and the majority were CD25^highCD44^low, corresponding to the third stage of DN thymocytes (Fig. 1 and data not shown). These results indicated that the expression of ICN1 can skew the cell fate of c-kit⁺ FL cells into the T cell lineage even on the OP-9 monolayer culture specific for B and myeloid cells.

To confirm whether Thy-1⁺ cells appearing in OP-9 monolayer culture had already expressed T cell-specific characteristics, GFP⁺ cells harvested from the cultures were subjected to further analysis (Fig. 2). The harvested cells were fixed, permeabilized, and stained with anti-CD3ε Ab for the intracellular CD3ε chain as a T cell marker. As shown in Fig. 2,A, the ICN1-transfected cells had CD3ε chain in the cytoplasm, although such cytoplasmic CD3ε-positive cells were not found in either GFP⁺ or GFP⁻ cells transfected with MIGR or dANK. At the same time, GFP⁺ cells sorted from every infection were subjected to RT-PCR to determine the expressions of pTα and GATA3 genes (Fig. 2,B). As was expected, pTα expression, known to be enhanced by Notch signaling (14, 15), was clearly shown only in the cultured GFP⁺ cells with ICN1 (Fig. 2,B, middle panel), but not in the other two GFP⁺ cells transfected with control genes (Fig. 2,B, right and left panels) or in ICN1-transfected NIH 3T3 fibroblasts (data not shown). The GATA3 gene, which is specifically expressed in T cells but not in B cells, was also detectable only in ICN1-bearing cells (Fig. 2 B). These results indicated that the Thy-1⁺ cells induced in the culture under ICN1 expression are committed to becoming a genuine T cell lineage.

To determine whether these Thy-1⁺ cells possess potentials for progressing to more developed stages of T cell lineage, the purified GFP⁺ cells were further cultured in the two systems. First, Thy-1⁺ cells were cultured in dGuo-treated fetal thymic lobes by hang-drop organ culture that had been well established for the in vitro induction of T cell development from immature T cells as well as hemopoietic progenitors. In FTOC, the GFP⁺Thy-1⁺ cells with ICN1 developed to become CD4⁺CD8⁺ DP cells and TCR-bearing cells, indicating that Thy-1⁺ cells appearing on OP-9 monolayer culture can develop to a more advanced stage in the thymic environment (Fig. 3 A). However, TCR β-high-positive or CD4 single-positive (SP) mature T cells could not be detected in this culture. This might be consistent with the phenotype reported in the previous studies that the enforced expression of ICN1 blocked DP-to-SP cell differentiation (16, 17).

Next, we examined the further developmental potentials of the same Thy-1⁺ cells on the B cell-specific OP-9 monolayer culture. After 5-day culture with IL-7, Thy-1⁺ cells had proliferated remarkably (>50 times), and the relatively large proportion became CD4⁺CD8⁺ DP cells with intermediate density of TCR β- or TCR γ-bearing cells (Fig. 3, B and C). This is the first evidence that genuine T cells expressing TCR can be induced in vitro without any thymic environment by ICN. CD4 or CD8 SP mature T cells with the high expression of TCR β could not be found. This may be consistent with the finding that the DP-to-SP cell differentiation requires the thymus-specific environment, especially epithelial cells, or it may be due to the effect of ICN1 described above in this subsection (16, 17). In contrast to Thy-1⁺ cells with ICN1, GFP⁺ cells transfected with MIGR or dANK did not develop into T cell marker-bearing cells even after they were subjected to FTOC or longer OP-9 monolayer culture with IL-7 (Fig. 3 B and data not shown). These results clearly demonstrated that a thymic environment is not always required for the appearance of TCR β-bearing DP cells if Notch signaling is present.

It is notable that γδ T cells were found in a relatively large proportion in both thymic and monolayer cultures (Fig. 3). This result may not be consistent with the previous proposal (18, 19) that Notch signaling alters the cell fate of a T cell progenitor at the branch point of αβ/γδ T cell lineages. This will be discussed in Discussion.

In the culture condition for B lymphopoiesis from HSC, certain stromal cells such as OP-9 are known to be required (12, 13). As shown in the aforementioned results, ICN1 can enforce hemopoietic progenitors into T cell lineage even in the culture condition specific for B and myeloid cells. These results raise the question of whether or not in vitro T lymphopoiesis actually requires stromal cells and/or IL-7 other than ICN1. To address this issue, we first used another bone marrow-derived stromal cell line, PA-6 (9), instead of OP-9, because PA-6 had been reported to lack IL-7 production (20). In PA-6 monolayer culture, c-kit⁺ FL cells were induced to be Thy-1⁺ cells with ICN1 and to be CD19⁺ cells without ICN1, respectively, in the presence of IL-7. However, without IL-7, neither Thy-1⁺ nor CD19⁺ cells appeared on the PA-6 monolayer (Fig. 4 A), indicating the requirement of IL-7 for the cell commitment of both T and B lineages from hemopoietic progenitors, although the possibility was not excluded that IL-7 is critical for the survival or expansion of the lineage-committed cells. In the culture of c-kit⁺ FL cells without stromal cells, only a few Thy-1⁻CD19⁻ cells were detectable regardless of the infected retroviruses even if IL-7 or IL-7 plus SCF was present (data not shown). These results suggest that bone marrow-derived stromal cells support the in vitro lymphopoiesis for T and B lineages by their products such as surface or secreted molecules.

Thus, we examined whether the direct interaction between stromal cells and c-kit⁺ FL cells is necessary for Thy-1⁺ cell appearance. As shown in Fig. 4 B, Thy-1⁺ cells in the presence of ICN1 and CD19⁺ cells in the presence of MIGR or dANK appeared in the culture separated by 0.4-μm pore size filter between c-kit⁺ cells and OP-9 monolayer, if IL-7 plus SCF were present. However, no lymphocyte was obtained only with IL-7 (data not shown), suggesting that SCF was quite significant for such lymphopoiesis in this system and that the membrane-bound SCF might be effective to support the cell survival or growth in the monolayer culture. These results indicated that direct interaction with stromal cells is not indispensable for the in vitro lymphopoiesis from c-kit⁺ cells, although with reduced induction efficiency, but that stromal cell-derived soluble factor(s) are required in addition to IL-7 and SCF. It is also suggested that the differentiations of both T and B cell lineages are supported by this stromal cell-derived factor(s), besides Notch signaling.

Notch proteins are composed of several members in mammals, but it is not clear whether the intracellular domain of every Notch protein has a similar function. To resolve this issue, the intracellular domains of other Notch genes, Notch2 and Notch3 (ICN2 and ICN3), were prepared and introduced into c-kit⁺ FL cells, instead of Notch1, by retrovirus vectors. As depicted in Fig. 5 A, ICN2 and ICN3 showed effects on the appearance of Thy-1⁺ cells similar to that of ICN1, but neither contributed to inducing cells expressing CD19 or Gr-1. Thus, ICN2 and ICN3 also induced T lymphopoiesis but inhibited the differentiation to B and myeloid cells on the monolayer culture of bone marrow stromal cells. This indicates that, if Notch proteins are cleaved by interaction between their extracellular portions and the corresponding receptor, the resultant intracellular domains may have common signaling for T lymphopoiesis in vitro at least in Notch1, -2, and -3.

In the further recultures, ICN2 and ICN3 could induce the differentiation to the DP stage of the gene-transfected Thy1⁺ cells as shown, as well as ICN1 could (Fig. 5 B), but their efficiencies were lower than that of ICN1. These results suggested that ICN2 and ICN3 had a potential to induce the Notch signal that is required for the differentiation, but the signaling magnitude might be different within the members of Notch family.

This study provides the first evidence of the in vitro induction of hemopoietic progenitors into T cells without a thymic environment, if ICN1 was exogenously introduced. On the monolayer culture of bone marrow-derived stromal cells, which is substantially specific for B cell development, murine c-kit⁺ FL cells with ICN1 were all forced to be T cells expressing TCR, CD3, CD4, and CD8, other than Thy-1.

Recently, there were several reports suggesting that constitutive Notch signaling is involved in the efficient commitment of HSC into the T cell lineage in vitro; in a suspension culture of murine HSC transfected with ICN1 in the presence of various growth factors without stromal cells, Lin⁻c-kit⁺Sca-1⁺ cells became Gr-1⁻CD25⁺Thy-1⁺ cells, but they still expressed both c-kit and Sca-1, markers of the stem cells (21). In another cell suspension culture, human CD34⁺ HSC with ICN1 became CD7⁺ cells but did not express other T cell markers on the surface, suggesting that they are at least at the T/NK precursor-like stage (22, 23). These murine and human cells developed from HSC in culture did not show the surface expression of TCR even though cytoplasmic CD3ε was detectable. Another previous study demonstrated that CD4⁺CD8⁺CD3⁺ cells were detected in the bone marrow after transplantation of ICN1-transfected HSC (7). However, these reports did not clarify what specific or common elements actually induce the appearance of CD4⁺CD8⁺ cells in bone marrow except ICN1.

In our culture, c-kit⁺ FL cells transfected with ICN1 developed into T cells expressing CD4, CD8, and TCR by the monolayer culture of OP-9, a stromal cell line, but did not develop without OP-9, suggesting that ICN1 alone is not sufficient for inducing T cell development from HSC. Although OP-9 produces IL-7 and SCF (T. Nakano, unpublished observations), IL-7 alone or IL-7 plus SCF did not promote the differentiation of T and B cells from c-kit⁺ FL cells in suspension culture without OP-9 (data not shown). Flt-3/Flk-2 ligand is a potent candidate of soluble factors to support B lymphopoiesis in cooperation with IL-7 and SCF, but the combination of these three factors could not reconstitute the differentiation of T and B cells from c-kit⁺ FL cells if OP-9 cell was absent (data not shown). However, as depicted in separated culture (Fig. 4 B), direct interaction between c-kit⁺ FL cells and OP-9 is not necessarily required for lymphopoiesis of T and B cells. Collectively, it is suggested that, in addition to Notch signaling, the in vitro induction of T cell development requires OP-9-derived soluble factor(s) other than IL-7, SCF, and Flt-3 ligand, which are shared in B cell development in vitro.

Past studies using transgenic mice with the active form of Notch1, ICN1, suggested that Notch1-mediated signaling alters cell fate from γδ to αβ T cell lineage (18). In our experiments using ICN1-transfected cells, the TCR β-chain was highly expressed on the surface of immature thymocytes and, unexpectedly, γδ T cells also appeared in a DN cell fraction (Fig. 3). Since the pTα gene was recently reported to be a target of Notch signaling, the presence of ICN1 may induce a higher expression of pTα whose product can generate the heterodimer with TCR β-chain to be pre-TCR complex easily. Consequently, this was connected with the expression of TCR β-chain on ICN1-transfected DN cells (Fig. 3 C) and resulted in the efficient transition from DN to DP cells of TCR αβ T lineage with a marked proliferation. Our result is consistent with a very recent finding (24) that the development of γδ T cell lineage is not affected by the inactivation of Notch1 gene at the DN2 intrathymic stage. Thus, it is likely that Notch-mediated signaling may promote the appearance of numerous TCR β-bearing DP T cells but does not shut off the differentiation pathway to γδ T cell lineage as is not seen in B cells, which are completely inhibited from their development by ICN1 at the early branch point.

The Notch family is composed of four members, Notch1 to -4, in mammals, and among them, Notch1, -2, and -3 are reported to function as signaling mediators just after binding with their physiological ligands, Dll and Jagged family members (25, 26). The present study (Fig. 5) clearly demonstrated that, in addition to Notch1, the intracellular domain of Notch2 or Notch3 is also able to induce T cells in culture, although the efficiency of the transition from DN to DP was varied. At the same time, these signalings inhibit B cell development. This implies that the signaling mediated by Notch2 or Notch3 has the potential to induce T lymphopoiesis as well as that by Notch1. Since a conditional gene-targeting study of Notch1 showed the impairment of T cell development even if the other Notch family genes were intact (6), Notch1 has been highlighted as playing a critical role in the cell fate decision at the branch point. However, the question remains why Notch2 and/or Notch3 cannot compensate for the defect of Notch1. In fact, Notch1 and Notch2 transcripts are easily detectable in Lin⁻c-kit⁺Sca-1⁺ cells (our unpublished data). Taking together these findings with our present data, it is predictable that the active forms of Notch2 and -3 may not be equally generated as Notch1 is in the thymus, and that, consequently, both Notch2 and -3 proteins may not be involved in the thymic development pathway. This assumption may be supported by our recently established conditional targeting of mice for Notch2, in which normal T cell development was observed (35). To confirm this, the binding specificity to the ligands of each Notch or the distribution should be clarified.

In the physiological condition, the transcripts of Notch ligands, Dll and Jagged family members, were detected in the bone marrow as well as in the thymus, and partly in hemopoietic cells (27, 28, 29, 30). In a recent study, the enforced expression of Dll-1 in bone marrow-derived stromal cells showed induction of the in vitro differentiation of T/NK precursors, but such induction was not detectable in the enforced expression of another Notch ligand, Jagged-1 (23). These results suggest distinct functions between Dll-1 and Jagged-1 and/or different expression patterns of these ligands between thymic and bone marrow stromal cells. The in vitro induction of T cell development needs the stereoscopic structure of thymic stromal cells, which may facilitate more close cell-to-cell contact between thymic stromal cells and hemopoietic cells. Such close interaction of thymic stromal cells and lymphoid cells may also contribute to efficient binding between Notch proteins and Notch ligands if their expression is not sufficient. If that is the case, the enforced expression of Dll-1 on bone marrow stromal cells may replace the T cell induction ability by the stereoscopic structure of thymic stromal cells. In addition to the ligands, there is considerable information about regulators of Notch signaling (31). Among them, Fringe, a glycosyltransferase, was reported to have the interesting ability to antagonize Notch signaling and block T cell development in transgenic mice (32, 33). Collectively, how Notch signaling is induced and regulated in the bone marrow and in the thymus is critical for understanding the molecular mechanism of lymphoid commitment to T/B cells.

Very recently, Schmitt and Zúñiga-Pflücker (34) demonstrated the importance of Notch signaling for the T lymphopoiesis from the side of a Notch ligand in vitro. These results, together with our results, have clarified the pivotal roles of Notch signaling for T cell commitment from both sides, Notch ligand and active form of Notch which induces Notch signaling. However, it was not shown whether Jagged-1 can work effectively as well as Dll-1 on the surface of OP-9 for the induction of murine T cells in vitro. Further analysis with these molecules on OP-9 should make clear how Notch ligands affect their receptor in the thymus, which is a key to understanding T cell commitment in vivo.

We thank Drs. W. Pear and T. Kitamura for providing us with retrovirus vectors, and Y. Okada and S. Ueno for operating the cell sorter. We also thank I. Tsuchiya and M. Onoe for technical assistance.