Active Form of Notch Imposes T Cell Fate in Human Progenitor Cells¹

The development of mature T cells from CD34⁺ progenitor cells involves a series of cell fate choices and differentiation steps that direct cells along one of several distinct developmental pathways. In this respect, Notch1, a mammalian analog of Drosophila Notch and a member of the Notch multigene family that controls binary cell fate decisions in several developmental systems, might be critical in T cell development (1, 2, 3, 4). The gene expression requirements for T cell differentiation have been well defined in the murine system using the powerful approach of gain-of-function and loss-of-function studies with transgenic and gene-disrupted mice. An essential role for Notch1 during development of T cells in thymus as well as in secondary lymphoid organs such as lymph nodes and spleen has been reported recently (5). Nevertheless, it was the identification of Notch1 as the gene involved in chromosomal translocations with the TCR-β gene in a subset of human T cell leukemias that gave a first clue as to the important role of Notch in T cell regulation (6). These Notch translocations result in the expression of truncated Notch1 polypeptides that lack most of the extracellular domain and constitutively activate the Notch pathway. In addition, induced overexpression of constitutively active Notch1 (ICN)³ in mouse bone marrow (BM) stem cells causes T cell leukemia, suggesting a causative role for Notch1 in T cell oncogenesis (7). A role for Notch signaling in promoting the commitment of lymphoid progenitor cells to the T cell lineage has been recently proposed and reviewed (8, 9, 10). Critical findings in this respect were a severe block in the further differentiation of the most immature CD4⁻CD8⁻ thymocytes after inducible deletion of Notch1 in murine BM stem cells, whereas the development of other hematopoietic populations was apparently undisturbed (5).

To elucidate the importance of Notch1 activation on the differentiation of human hematopoietic cells, we transduced cord blood (CB) CD34⁺ progenitor cells with ICN and evaluated their differentiation toward different hematopoietic lineages by the following approaches. First, in coculture experiments with the murine BM MS-5 cell line (11), differentiation toward B lymphocyte, monocyte, and NK lymphocyte lineages was evaluated by the determination of the expression of CD19, CD14, and CD56, respectively. The determination of the coexpression of CD7 and cyCD3 allowed the evaluation of differentiation toward T/NK precursor cells. Second, the introduction of transduced human CD34⁺ progenitor cells in SCID-nonobese diabetic (NOD) mouse FD14 thymi followed by fetal thymus organ culture (FTOC) allowed us to estimate the impact of Notch1 on human T lymphocyte differentiation in vitro. Finally, the transduced human CD34⁺ progenitor cells were i.v. injected in SCID-NOD mice to estimate the impact of Notch1 on human lymphopoiesis in vivo.

cDNA encoding a constitutively active form of Notch consisting of the intracellular domain (ICN) (base pairs 5308–7665; amino acids 1770–2555) (12) was subcloned into the multicloning site of the retroviral vector MSCV-EGFP (kindly provided by N. Carlesso, Massachusetts General Hospital, Charlestown, MA) (13). The puromycin amino transferase gene was cloned in the pBlue(II)KS vector (Stratagene, La Jolla, CA) into the KpnI/ClaI restriction sites. Propagated plasmids were purified with resin columns (Qiagen, Hilden, Germany). The Phoenix-A cell line, which was derived from 293T cells (kindly provided by Dr. P. Achacoso and Dr. G. P. Nolan, Stanford University School of Medicine, Stanford, CA) (14), was cotransfected with the pBlue(II)KS-puro plasmid and either the MSCV-ICN-EGFP or the MSCV-EGFP plasmid (control) using calcium phosphate precipitation (Life Technologies, Paisley, U.K.). After three rounds of selection with IMDM containing puromycine (2 μg/ml), cells were cultured for 48 h in IMDM. Cell culture supernatant from confluent cultures in 175-cm² tissue culture flasks (Falcon; BD Labware, Franklin Lakes, NJ) was collected 24 h after refreshment of the medium. Pooled supernatants were spun (350 × g for 10 min at room temperature) and aliquots were stored at −70°C until use.

Jurkat cells (American Type Culture Collection, Manassas, VA) were cultured in complete IMDM as described previously (15). Child thymi and CB samples were obtained and used following the guidelines of the Medical Ethical Commission of the University Hospital of Ghent.

CD34⁺ CB cells were purified by positive selection with CD34 MACS beads (Miltenyi Biotec, Bergisch Gladbach, Germany) and stained with CD34-allophycocyanin, CD8α-FITC, CD4-FITC, CD3-FITC, CD19-FITC, and CD7-FITC (all mAb from BD Immunocytometry Systems, Mountain View, CA). Subsequently, CD34⁺Lin⁻ cells were sorted for by flow cytometry (FACSVantage; BD Immunocytometry Systems). CD34⁺ thymocytes were purified by positive selection with MACS beads (Miltenyi Biotec) and sorted for CD1⁻CD4⁻CD8⁻CD3⁻CD34⁺ progenitor cells by flow cytometry. Purity of the cells was always at least 98%.

Sorted CD34⁺Lin⁻ CB cells were resuspended in complete IMDM supplemented with 100 ng/ml recombinant human c-kit ligand (stem cell factor (SCF)), 100 ng/ml flt3/flk-2 ligand (FL), and 20 ng/ml thrombopoietin (TPO) (all cytokines from R&D Systems, Abingdon, U.K.) and cultured in 96-well round-bottom plates (Falcon) for 48 h. Sorted CD34⁺ thymocytes were cultured in complete IMDM supplemented with 10 ng/ml SCF and 10 ng/ml IL-7 for 24 h and then cells were put in RetroNectin (Takara Biomedicals, Otsu Shiga, Japan)-coated 96-well flat-bottom plates (Falcon) and cultured with the same volume (100 μl) of retroviral supernatants, supplemented with cytokines (to keep the final cytokine concentrations unchanged). After 24 h, cells were harvested to determine transduction efficiency by flow cytometric analysis of enhanced green fluorescent protein (EGFP) expression and were used in subsequent assays. ICN overexpression was also determined by immunoblotting and flow cytometric analysis of intracellular ICN labeled with a mouse mAb against the cdc10-NCR region of mNotch1 (a kind gift from L. A. Milner, Fred Hutchinson Cancer Research Institute, Seattle, WA) (16, 17), revealed with second-step PE-labeled anti-mouse Ig Ab.

Cell lysates were run on 4–12% Bis-Tris polyacrylamide gels (Nupage; Invitrogen, Carlsbad, CA) in 2-(N-morpolino) ethane sulfonic acid buffer in reducing conditions and proteins were blotted on polyvinylidene fluoride membranes (Invitrogen). Blots were stained with a mAb against ICN (L. A. Milner) and anti-mouse Ig alkaline phosphatase conjugate (Santa Cruz Biotechnology, Santa Cruz, CA).

Isolation of murine embryonic thymic lobes, incubation with human cells using the hanging drop procedure, and organotypic cultures were performed as described previously (18). NOD-LtSz-scid/scid (NOD-SCID) mice, originally purchased from The Jackson Laboratory (Bar Harbor, ME), were obtained from our own specific pathogen-free breeding facility. NOD-SCID mice were treated according the guidelines of the Laboratory Animal Ethical Commission of the University Hospital of Ghent.

After 3–4 wk of FTOC, cells were harvested. After blocking with anti-mouse FcRγII/III (clone 2.4.G2; a kind gift of Dr. J. Unkeless, Mount Sinai School of Medicine, New York, NY) to avoid aspecific staining of the murine cells, cells were stained with anti-mouse CD45-CyChrome (BD PharMingen, San Diego, CA) in combination with one or more of the following anti-human Abs: CD4-allophycocyanin, TCR-γδ-PE, CD3-allophycocyanin, CD34-allophycocyanin (all from BD Immunocytometry Systems), CD8β-PE, and TCR-αβ-PE (Coulter, Miami, FL). Human viable cells were gated by exclusion of propidium iodide and anti-mouse CD45-positive cells, and were examined for the expression of the Ags by fluorescence analysis performed on a FACSCalibur using CellQuest Pro software (BD Immunocytometry Systems).

After transduction of CD34⁺Lin⁻ purified CB cells, 1000 cells were incubated in 24-well plates precoated with confluent murine marrow-derived MS-5 cells (11) (kindly provided by L. Coulombel, Institut Gustave Roussy, Villejuif, France) in IMDM supplemented with 5% human serum and 5% FCS. For assessment of CD34 maintenance and myeloid differentiation, the following mix of six human recombinant cytokines was used: 50 ng/ml SCF, 50 ng/ml FL, 20 ng/ml IL-7, 10 ng/ml IL-15, 5 ng/ml IL-2, and 10 ng/ml TPO. For assessment of B lymphoid differentiation, 50 ng/ml SCF and 20 ng IL-7 was used. For the assessment of NK cells, 50 ng/ml SCF, 5 ng/ml IL-2, and 10 ng/ml IL-15 were used (all reagents from R&D Systems).

After 2 wk of culture cells were collected and counted under the microscope, and their phenotype was assessed by flow cytometry after labeling with the following mAbs: CD34-allophycocyanin, CD19-PE, CD14-PE, CD56-allophycocyanin, HLA-DR-allophycocyanin, CD7-PE, CD4-PE or allophycocyanin, CD3-PE or allophycocyanin. For intracellular staining, cells were fixed and permeabilized using Fix and Perm (Caltag Laboratories, San Francisco, CA) according to the guidelines of the manufacturer.

NOD-SCID mice aged 8–10 wk were given a sublethal dose of whole-body irradiation (350 cGy, 12–15 cGy/min) with a cobalt radiation source and injected i.p. with 200 μg of TM-β1, a rat mAb functionally blocking the mouse IL-2Rβ chain (19) (kindly provided by Dr. T. Tanaka, Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan).

Within 24–48 h after irradiation, the mice were injected i.v. with 75,000–100,000 human transduced CD34⁺ CB cells. Eight to 10 wk after injection, mice were killed and peripheral blood, thymus, spleen, and femora were used for analysis. Cell suspensions from these organs were put on a 70-μm cell strainer (Falcon). RBCs were lysed with hypotonic lysing buffer. Cells were counted, cell viability was checked with eosin (>85%), and, after blocking the FcR of mouse cells, cells were labeled with mAbs and analyzed by flow cytometry.

A paired Student t test was used and differences were considered to be significant when p < 0.05.

Fig. 1,A shows a diagram of the complete Notch protein, of the active intracellular part of the Notch protein that is formed after protein processing upon ligation of the receptor, and of the constructs that were used in the study. The efficiency of MSCV-ICN-EGFP or control MSCV-EGFP retroviral transduction was assessed by Western blot (Fig. 1,B) and flow cytometric analysis (Fig. 1 C). Independent experiments showed a mean infection efficiency of 20 and 27% with MSCV-ICN-EGFP or control MSCV-EGFP retroviruses, respectively, for Jurkat cells. Infection was more variable in CD34⁺ cells, ranging from 0.7 to 12.7% of cells (average 5.3 ± 3.2, n = 8) expressing EGFP for MSCV-ICN-EGFP and ranging from 2.7 to 24.1% (average 14.8 ± 6.6) for control MSCV-EGFP retrovirus, after transduction during 24 h. After another 24 h, the proportion of cells stably expressing EGFP increased to 10.6 ± 6.2 and 23.2 ± 9.9% for the cells transduced with MSCV-ICN-EGFP and control MSCV-EGFP retrovirus, respectively.

Western blot analysis showed that ICN of the appropriate length (98 kDa) was found in both the transduced Jurkat and CB progenitor cells after culture in FTOC (Fig. 1,B), whereas control Jurkat cells and control vector-transduced CB progenitor cells were negative. In a representative experiment, wherein the proportion of ICN-transduced cells increased to 30% in FTOC after 30 days of culture, cells were stained intracellularly with anti-Notch mAb. It is clear that Notch could be detected in ICN-transduced CB cells (Fig. 1,C, upper panel), whereas control vector-transduced cells were negative (Fig. 1,C, lower panel). Moreover, there was a good correlation between the intensity of ICN staining and EGFP fluorescence in the transduced CB cells, showing that EGFP intensity is a parameter for intensity of Notch production (Fig. 1 C, upper panel). Isotype control Abs did not stain transduced or untransduced cells aspecifically (data not shown).

It has been shown previously that murine BM-derived MS-5 cells support B cell differentiation from human CD34⁺ CB cells (20). Depending on the mix of cytokines added to the culture, either the growth of stem cells with a CD34 phenotype is supported or the differentiation toward myelocytes or lymphocytes is promoted. We used a combination of following cytokines: SCF plus FL plus IL-2 plus IL-7 plus IL-15 plus TPO. These conditions allow the growth of stem cells and the differentiation toward monocytes. In a series of four to six separate experiments these conditions triggered active proliferation, and after 2 wk of culture the number of the human cells increased ∼1000-fold for the nontransduced or control-transduced CD34⁺ CB cells, whereas the number of ICN-transduced cells increased 2650-fold (Table I). However, ICN-transduced CD34⁺ CB cells had a significant lower frequency of CD34⁺ cells (∼10-fold) as compared with nontransduced cells in the same culture, or as compared with control vector-transduced progenitor cells (Table II).

Similarly, in the same culture condition, ICN-transduced cells had a significantly lower frequency of CD14⁺ cells (∼5-fold) as compared with nontransduced cells or compared with progenitor cells, whether they were transduced with the control vector or not (Table II and Fig. 2 A). Taken together, it is clear that not only the frequencies but also the absolute numbers of CD34⁺ and CD14⁺ cells were decreased in the ICN-transduced cells.

Interestingly, when CD34⁺ cells were cocultured on MS-5 stromal cells with SCF and IL-7, known to promote B lymphocyte differentiation, ICN-transduced cells almost did not generate any CD19⁺ B cells. The nontransduced cells in the same culture had a significant number of CD19⁺ cells (Table II and Fig. 2 B). This number was comparable to the number of CD19⁺ B lymphocytes in both fractions, nontransduced or transduced, observed in cultures where progenitor cells were transduced with the control vector.

Both control vector nontransduced and transduced CD34⁺ cells yielded a very low number of cells (<2%) with a CD7⁺cyCD3⁺ phenotype. In contrast, the frequency of CD7⁺cyCD3⁺ cells was significantly more abundant in ICN-transduced cells (>80%) (Table II and Fig. 2,B). This phenotype is analogous to the phenotype of early T/NK precursor cells that arise when CD34⁺ progenitor cells are cultured in FTOC (21). These culture conditions resulted in a much higher increase (∼1390-fold) in the number of ICN-transduced (EGFP⁺) cells as compared with the control-transduced cells and the nontransduced cells from both control or ICN-transduced cells, which increased from 176- to 202-fold (Table I). Taken together, it is clear that the 30-fold lower frequency of CD19⁺ B cells and the 80-fold higher frequency of early T/NK precursors also reflect changes in the absolute numbers of those cells, and that the increase in cell number is due to the expansion of T/NK precursors in the ICN-transduced cell population.

To further demonstrate that the CD7⁺ cells that were obtained from the ICN-transduced CB cells after culture on MS-5 cells with IL-7 and SCF were a bipotent T/NK population we did the following experiment. First, CD7⁺ cells were generated by culture of ICN-transduced CB cells on MS-5 cells with SCF and IL-7 for 2 wk, which we will refer to as primary culture. Then the EGFP⁺CD7⁺CD56⁻ cells obtained in the primary culture were sorted by flow cytometry. As shown in Fig. 3 we obtained a population that was >99% pure. Intracellular staining for cyCD3 on those sorted cells showed that >90% were positive (data not shown). This EGFP⁺CD7⁺CD56⁻ population was put in a secondary coculture on MS-5 cells with SCF, IL-2, and IL-15. After 1 wk we were able to obtain human cells that were 85% CD56⁺, indicating the NK potential of the EGFP⁺CD7⁺CD56⁻ population (Fig. 3). In addition, EGFP⁺CD7⁺CD56⁻ cells from the primary culture were seeded in D15 fetal thymus by hanging drop, and after 16 days of FTOC we obtained human cells that were >75% CD3⁺; almost all the CD3⁺ cells were TCR-γδ⁺ cells (Fig. 3). This shows the T cell potential of the EGFP⁺CD7⁺CD56⁻ population.

Conversely, in the presence of SCF/IL-2/IL-15, CD56⁺ NK cells were significantly more abundant in ICN-transduced cells as compared with the moderate amounts in nontransduced cells. A comparable lower number of CD56⁺ NK cells was obtained in both populations, transduced or nontransduced, when the control vector was used (Table II and Fig. 2,C). The increase in absolute number of cells in these culture conditions was ∼120-fold for all cell fractions, control or ICN-transduced cells and nontransduced cells (Table I). Taken together, it is clear that not only the frequencies but also the absolute numbers of CD56⁺ cells were increased in the ICN-transduced cells.

To assess T cell differentiation in vitro, CB CD34⁺ progenitor cells were transferred to NOD-SCID FD14 thymic lobes and cultured in FTOC. In this microenvironment, human T cells develop (15, 18). We found a significantly higher amount of TCR-γδ cells in ICN-transduced thymocytes than in the nontransduced cells, which were comparable to control vector-transduced or nontransduced cells (Table III and Fig. 4, A and B). Accordingly, the number of CD3⁺ cells was also significantly increased as compared with the cells that were nontransduced or to the cells of the FTOC seeded with CB cells transduced with control vector, irrespective of whether those cells were transduced or not (Table III and Fig. 4, A and B). The other populations, including TCR-αβ⁺ and CD4⁺CD8⁺ double positive (DP) cells, were not significantly different, although the frequency of CD34⁺ cells was slightly diminished in the ICN-transduced cell population (data not shown). However, the large CB-to-CB variation in the number of TCR-αβ and CD4⁺CD8⁺ DP cells obtained in the 11 FTOC performed did not allow a good estimation of the influence of ICN on those subsets. Interestingly, in cultures where a high number of TCR-αβ and DP cells were generated, ICN-transduced CD34⁺ cells generated a lower number of TCR-αβ DP cells than the nontransduced cells and the control vector-transduced CD34 cells (data not shown).

CD34⁺ cells from the thymus are already at a further step of T cell commitment and therefore produce a consistently higher number of TCR-αβ DP cells in FTOC. For this reason, we looked at the influence of ICN overexpression on the fate of those cells in FTOC. This approach confirmed the higher frequency of TCR-γδ cells in the ICN-transduced cells as compared with the nontransduced cells, or to the control vector-transduced or nontransduced cells. In addition, the frequency of TCR-αβ cells was significantly lower in the ICN-transduced cells as compared with the nontransduced cells or to the control vector-transduced or nontransduced cells (Table III and Fig. 4, C and D). The same goes for the frequency of DP cells, as these cells were significantly lower in the ICN-transduced cells as compared with the nontransduced cells, or to the control vector-transduced or nontransduced cells (Table III). The frequency of CD3⁺ cells did not differ significantly.

To assess T cell differentiation in vivo, ICN-transduced CD34⁺ progenitor cells from CB were injected i.v. in 8- to 10-wk-old NOD-SCID mice that were sublethally irradiated and treated with TMβ-1 mAb 1 or 2 days before injection. We have shown previously that CD34⁺ CB cells give rise to myeloid cells and to B and T lymphoid cells in these mice (22, 23). When examined 2 mo after injection, in the BM of four of five injected mice we found cells that strongly expressed EGFP, and thus ICN. Those cells coexpressed CD4 and CD8, a phenotype normally found only in the thymus. All these cells were CD1⁺; the majority were CD3⁺TCR-αβ⁺. We noticed a small population of CD4⁺CD8⁻ cells (3%) that were partly CD1⁺ (and therefore most likely represent immature CD4 precursor cells) and partly CD1⁻, a phenotype compatible with mature CD4 single positive (SP) cells. Because we could not find CD14⁺ cells, we conclude that monocytes were not present in this CD4⁺ cell fraction. This is in contrast with the untransduced cells, because CD4⁺ cells in the EGFP⁻ cell fraction were CD1⁻CD3⁻CD8⁻ and partly CD14⁺ and can therefore be considered monocytes (data not shown).

CD19⁺ B lymphocytes were nearly absent within the EGFP⁺ population. In addition, in the nontransduced EGFP⁻ cell fraction we found no DP cells and a high number of CD19⁺ B lymphocytes (Fig. 5).

Previously, it has been shown that Notch-1 mRNA is present in human precursors (CD34⁺lin⁻), indicating a role for Notch interactions in hematopoiesis (24). In mouse, protein expression of Notch-1 and Notch-2 has been detected in BM and in Lin⁻Sca-1⁺c-kit⁺ precursor cells, whereas the Notch ligand Jagged-1 has been detected in BM stromal cells. When Lin⁻Sca-1⁺c-kit⁺ precursor cells were cultured with the appropriate cytokines, addition of beads coated with Jagged-1 increased the number of CFU-C colonies, suggesting that Notch interactions may promote self renewal of early precursor cells (25). Carlesso et al. (13) have also been able to show that primary human CD34⁺ cells retain their ability to form immature colonies and that coculture of CD34⁺ CB progenitor cells with 3T3 monolayers expressing Jagged-2 results in a delay of CD34 down-modulation. Varnum-Finney et al. (26) have shown that activated Notch1 can result in the in vitro outgrowth of cell lines with multipotential, indicating that Notch can immortalize stem cells and promote their self renewal. In our culture system, we did not observe a delay in CD34 down-regulation upon ICN transduction in CD34⁺ CB progenitor cells. The different outcome in coculture experiments might be due to the differences among feeder cells or to the composition and the amounts of cytokines used.

Because an activated form of Notch1 is able to inhibit G-CSF-induced granulocytic differentiation of 32D myeloid progenitors (17), it is indicated that Notch activity plays a negative role in mediating cell fate decisions in the myeloid lineage. This is compatible with the lower number (absolute and frequency) of CD14 cells that we obtained after overexpression of ICN in human CD34⁺ progenitor cells. The inhibition of monocytic and granulocytic differentiation reported by Carlesso et al. (13) is in line with our observations that the number of generated CD14⁺ cells is diminished upon ICN transduction. Preferential transduction of a cell population with lower differentiation capacity toward monocytes is unlikely, because both EGFP⁺ and EGFP⁻ cell fractions after transduction with a control vector produced CD14⁺ cells in equal frequencies and absolute numbers. It is possible that, upon ICN transduction, not only was the differentiation of monocytes impaired, but their survival was impaired as well. Recently it has been shown that, in cell suspension cultures wherein Notch was activated in isolated human monocytes by immobilized Delta, apoptosis of monocytes was induced and differentiation into macrophages was impaired (27).

The ability of ICN to preferentially inhibit the B cell fate in lymphoid differentiation conditions was evident by comparing the proportion of cells in the EGFP⁺CD19⁺ population. It is strikingly evident that almost no CD19⁺ cells were EGFP⁺ in the ICN-transduced cultures. The percentage of CD19⁺ cells was 12% in the EGFP⁻ and <1% in the EGFP⁺ fraction. These results clearly indicate that the B cell fate is negatively influenced by ICN. The negative role of Notch on B cell development is well documented in the murine system. In mice, retrovirally transduced BM cells that overexpress ICN do not develop into B cells in lethally irradiated recipients (28). Moreover, immature T cells with a CD4⁺CD8⁺ phenotype accumulate in the BM of these chimeric mice. Our observations provide the first direct evidence that expression of an active form of Notch1 inhibits human B lymphopoiesis. Here we show that, in culture conditions that allow B cells to develop, retrovirally transduced CD34⁺ progenitor cells that overexpress ICN scarcely develop into B lymphocytes. Also, when injected i.v. in immunodeficient mice, ICN-transduced cells do not develop into B lymphocytes. In contrast, those transduced cells develop ectopically in human CD4⁺CD8⁺CD3⁺TCR-αβ⁺ T cells in the BM. Therefore, we can conclude that also in humans these gain-of-function experiments clearly show that Notch inhibits the B cell fate and promotes the T cell fate.

There is ample evidence from our experiments that the T cell fate is favored in ICN-transduced cells. After 2 wk of MS-5 coculture assays, ICN-transduced cells developed in CD7⁺cyCD3⁺ cells, a phenotype that is considered to identify pre-T/NK cells (21). After 3 wk CD7⁺cyCD3⁺CD4⁺ cells were found, a phenotype that is considered to identify T precursor cells (21). In addition, when those CD7⁺ cells obtained after 2 wk of MS-5 coculture assay were purified by flow cytometry, those cells generated NK cells in secondary MS-5 coculture assays with IL-2, IL-15, and SCF after 1 wk, whereas T cells were generated in FTOC after 16 days. We can conclude that these experiments confirm that ICN favors the generation of progenitor cells with T/NK potential.

When ICN-transduced CD34⁺ precursor cells were cultured in FTOC, differentiation toward CD7⁺cyCD3⁺CD4⁺ cells (data not shown) and toward TCR-αβ and TCR-γδ cells was seen. Whereas generation of TCR-αβ Τ cells was unaffected, a significant increase in the number of TCR-γδ T cells was observed. This is in contrast with studies in mice wherein reduced, rather than increased, activity of Notch favors the γδ T cell fate over the αβ T cell fate (29). In FTOC, starting with CD34⁺ cells from human thymus, one obtains a faster kinetic of T cell development and a higher yield of TCR-αβ, DP, CD4 SP, and CD8 SP cells, because thymic CD34⁺ cells are already a further step of differentiation toward the T cell lineage than CD34⁺ cells from CB. Using ICN-transduced thymic CD34⁺ cells we observed a significant increase in the number of TCR-γδ T cells, whereas no changes in CD4 SP and CD8 SP cells (data not shown) and even a decrease in TCR-αβ⁺ T cells and DP cells were observed. This is in contrast with the observation that expression of an activated form of Notch-1 in murine thymocytes under the lck promoter leads to an increase in CD8⁺ SP TCR-αβ⁺ cells and a parallel decrease in CD4⁺ T cells (30). These discrepancies can be related to the dose effect of Notch or to species differences. It is clear that complete deletion of the Notch1 gene in mice by cre/lox recombination leads to severe thymic atropy due to a block of the expansion and differentiation of immature CD4⁻CD8⁻ T cell precursors (5). In studies where a reduction in Notch gene expression was examined, it was shown that T cells from mice that are heterozygous for Notch1 (N^+/−) adopt preferentially the TCR-γδ and CD4 TCR-αβ fate as compared with the T cell fate of wild-type mice that are homozygous for Notch1 (N^+/+) (29). The same conclusion could be drawn from studies wherein the use of γ-secretase inhibitors leads to Notch deficiency by inhibition of the formation of the intracellular domain of Notch, which is essential for signal transduction in vivo (16). In these studies, a high dose of γ-secretase inhibitors administered to murine FTOC leads to a complete block of T cell development at an early stage of differentiation, resulting in an increase in CD4⁻CD8⁻ double negative thymocytes and a decrease in CD4⁺CD8⁺ DP thymocytes. A lower dose of γ-secretase inhibitors leads to changes at later steps of differentiation, resulting in a higher number of TCR-γδ and CD4 SP thymocytes (31). Thus, complete deficiency of Notch leads to an early block of T cell development, whereas partial deficiency of Notch allows further T cell development, but at the expense of TCR-αβ and CD8 SP development. Based on these findings obtained in murine models, one would expect that overexpression of ICN in human CD34⁺ progenitor cells should also favor these cells to adopt the TCR-αβ and CD8 SP cell fate. However, we consistently found in the mixed human-mouse FTOC that human cells adopted more frequently the TCR-γδ fate and that the frequency of CD8 SP cells was not increased. This contradiction could be ascribed to the fact that overexpression ICN may be detrimental, that the mixed human-murine FTOC is not optimal to allow the effect on TCR-αβ development to be seen, or that species differences are important.

Our data are remarkably in line with data recently obtained by Jaleco et al. (32), using human CD34⁺ progenitor cells in a different approach. Instead of introducing the active form of Notch in hematopoietic stem cells, these authors transduced a ligand for Notch, Jagged-1, or Delta-1 in the S17 murine stromal cell line. They were able to show that Delta-1-transduced S17 stromal cells induced CD34⁺ CB cells to adopt the T/NK developmental pathway and even to allow the differentiation of a few DP CD4⁺CD8⁺ cells, whereas the B cell pathway was inhibited. In contrast, Jagged-1-transduced S17 stromal cells did not have an effect on the differentiation of CD34⁺ CB cells. It is clear that our results not only confirm but also extend those experimental data in the following ways. First, the specificity of the ligands for the Notch receptors is largely unknown (25). Theoretically each of the four known Notch receptors may be triggered in the experiments of Jaleco et al. (32). Our approach is a direct proof that activation of Notch1 alone suffices to impose the CD34⁺ progenitor cells to adopt a T/NK pathway and to inhibit their differentiation potential toward B cells and myeloid cells. Furthermore, it is obvious that the presence of the few DP cells with a CD4/CD8αα DP phenotype that were seen in coculture with the S17 Delta-transduced stromal cells is suggestive for a transition toward a further step of differentiation of the T cell lineage; nevertheless, these CD4⁺CD8⁺ DP cell numbers are small (32). In our hands, in vivo experiments have shown a high number of CD4/CD8αβ DP TCR-αβ cells in the BM of reconstituted mice. These cells are phenotypically similar to CD4⁺CD8⁺ DP cells normally found in the thymus. Importantly, we found that there was a dose-dependent effect: when Notch1 ICN was expressed at lower levels as estimated by the lower expression of the reporter gene EGFP, the CD4/CD8αβ DP TCR-αβ cells were not present (data not shown). Therefore, we assume that it is likely that the differential effect seen between the two ligands of Notch, Delta-1 and Jagged-1, may be related to the efficiency of those ligands to activate the Notch receptor. In this respect, the authors were able to show that Hes-1 up-regulation was clearly seen when the cells received signals from Delta-1-transduced S17 stromal cells and not from Jagged-1-transduced S17 stromal cells (32). It remains an open question whether this reflects a physiological difference in the triggering capacity of those two ligands or whether the expression levels of the ligands were different.

Finally, ICN-transduced cells developed ectopically into CD4⁺CD8⁺CD3⁺TCR-αβ⁺ T cells in BM when injected i.v. in immunodeficient mice. These findings are in line with previous murine studies, wherein ICN promotes the differentiation of cells with similar CD4⁺CD8⁺CD3⁺TCR-αβ⁺ cells in BM (33). These cells may be considered pretumoral cells, because mice harboring ICN transgenes for Notch1 (7, 33, 34, 35) or Notch3 transgenes (36) develop acute leukemia-like tumors. We were not able to find transduced cells in the spleen, thymus, or blood. It is likely that the transduction procedure, wherein CD34⁺ cells are precultured over 24 h, inhibited the homing capacity of those cells. Our study indicates that mice injected with human progenitor cells transduced with Notch1-ICN accumulate human T cells in the BM with an unusual phenotype. Additional experiments are now performed to address whether these cells eventually develop into tumoral cells in function of length of time.

We were not able to show non-cell-autonomous modifications. We found that ICN nontransduced cells (EGFP⁻) had a remarkably similar distribution of cell subpopulations (B lymphocytes, monocytes, NK cells, and immature progenitor cells) as compared with control-transduced and nontransduced cells in the MS-5 coculture experiments (see Table II). Similarly, the distribution of the different cell subpopulations (CD3, TCR-γδ, TCR-αβ, CD4⁺CD8⁺) of the ICN nontransduced cells and the control-transduced and nontransduced cells were comparable in the thymuses in FTOC experiments (see Table III). This was also the case for the in vivo experiments, where, as expected, a significant amount of B cells were produced in the ICN nontransduced cell population (Fig. 4). This is in contrast with a recent study reported by Kawamata et al. (37) wherein mice reconstituted with Notch1ICN stem cells also showed a profound suppression of the B lineage in the nontransduced compartment. It is possible that a close cell contact within the BM microenvironment, together with a high transduction efficiency, is required to observe an influence on the surrounding cells.

In conclusion, our results show that, in human Notch, signaling determines the adoption of the T cell fate at the expense of the B cell fate, and that continuous stimulation of Notch drives the cells to develop ectopically into CD4⁺CD8⁺ DP cells in the BM. Therefore, a thorough knowledge of the Notch signaling pathway is of high potential relevance to modulate the T/B fate of human lymphocytes. These strategies are needed to shorten the lag time in T cell recovery after stem cell transplantation or after highly active antiretroviral therapy in HIV patients, and to address Notch-induced leukemogenesis.

We thank L. A. Milner (Fred Hutchinson Cancer Research Institute) for the gift of mouse anti-human Notch, Dr. J. Unkeless (Mount Sinai School of Medicine) for the gift of clone 2.4.G2, Dr. G. P. Nolan (Stanford University School of Medicine) for the gift of the Phoenix-NA packaging cell line, Dr. L. Coulombel for the murine marrow-derived MS-5 cells, Dr. T. Tanaka (Tokyo Metropolitan Institute of Medical Science) for the mAb anti-mouse IL-2Rβ chain, Dr. D. Scadden and A. Carlesso (Massachusetts General Hospital) for the MSCV-ICN-IRES-EGFP construct and MSCV-IRES-EGFP control vector, Prof. Dr. G. Van Nooten (Department of Heart Surgery, Ghent University Hospital) for the supply of human thymic tissue, Caroline Collier and Greet De Smet for animal care, and Christian De Boever for artwork.