Loss of dendritic cell potential is one of the major events in intrathymic T cell development, during which the progenitors become determined to the T cell lineage. However, it remains unclear whether this event occurs in synchrony with another important event, TCRβ chain gene rearrangement, which has been considered the definitive sign of irreversible T cell lineage commitment. To address this issue, we used transgenic mice in which GFP expression is controlled by the lck proximal promoter. We found that the double-negative (DN) 2 stage can be subdivided into GFP and GFP+ populations, representing functionally different developmental stages in that the GFPDN2, but not GFP+DN2, cells retain dendritic cell potential. The GFP+DN2 cells were found to undergo several rounds of proliferation before the initiation of TCRβ rearrangement as evidenced by the diversity of D-Jβ rearrangements seen in T cells derived from a single GFP+DN2 progenitor. These results indicated that the determination step of progenitors to the T cell lineage is a separable event from TCRβ rearrangement.

T and B lymphocytes use a common gene-recombination machinery to form their diverse AgRs, the TCR in T cells and BCR in B cells. The precise timing of segregation of these two cell lineages and the occurrence of gene rearrangement of TCR and BCR genes have long been a matter of dispute. Because V to DJ rearrangement of IgH chain genes in B cells and of TCRβ chain genes in T cells is strictly regulated in a lineage-specific manner (1, 2), completion of this rearrangement step has been regarded as a sign of irreversible commitment to the B and T cell lineages, respectively. However, several studies indicate that T and B cell lineages are separated much earlier than the initiation of these gene rearrangements. In the case of B cell progenitors, Allman et al. (3) demonstrated that early B cell lineage committed progenitors contain IgH genes in germline configuration, and proposed that B cell lineage commitment takes place before D-J rearrangement at the H chain locus; however, the extent of proliferation of the B cell lineage committed progenitors before the IgH rearrangement remains unknown. In the case of T cell progenitors, we have shown that cells that have lost B cell and myeloid potential can expand in the thymus by >1000-fold before they initiate TCRβ rearrangement (4, 5). Thus, commitment to the T cell lineage, as defined by the shutoff of B cell potential, should occur much earlier than the initiation of TCRβ gene rearrangement.

Progenitors for T cells that have shutoff B cell and myeloid potential are not necessarily fully T cell lineage committed, because a large proportion of early thymocytes retain dendritic cell (DC)3 potential (6, 7, 8). We have recently provided clonal evidence that early intrathymic T cell progenitors as well as prethymic T cell progenitors retain the potential to generate DCs and NK cells (9, 10, 11, 12). DCs are primarily considered as members of the myeloid family, because their major functions, such as phagocytosis and Ag-presenting activity, are similar to those performed by macrophages. Indeed, some previous studies have also pointed out that early intrathymic progenitors can produce macrophages (13, 14). Because the frequency of DC-generating progenitors is much higher than that of macrophage-generating progenitors, it is probable that the DC potential is more stably retained by T cell progenitors. Thus, we reasoned that the shutoff of DC potential in T cell progenitors can be considered as one of the most important events for them to acquire their T cell lineage identity.

In developmental biology, the commitment process is divided into specified and determined phases (15). According to this characterization, T cell lineage “specified” progenitors are nearly committed to the T cell lineage, but still retain other potentials such as NK and DC potentials and possibly residual macrophage potential, whereas T cell lineage “determined” progenitors are irreversibly committed to T cell lineage. Thus, hereafter, we refer to the point where the progenitors lose their DC potential as “T cell lineage determination.”

Some previous studies have focused on the DC shutoff point during intrathymic T cell development. Immature CD4CD8 (double-negative (DN)) thymocytes can be subdivided into subpopulations based on expression of CD44 and CD25; CD44+CD25 (DN1) cells represent the earliest population and they differentiate into CD44CD25+ (DN3) stage through CD44+CD25+ (DN2) stage. It has been shown that DN1 and DN2 cells but not DN3 cells are able to give rise to DC (6, 7, 8). Hence, the transition from DN2 to DN3 stage was assumed to be the point to shutoff the DC potential. It is also known that thymic progenitors undergo extensive proliferation at the DN2 stage (5, 16) and then enter a resting phase at the DN3 stage, where TCRβ rearrangement takes place. In general, differentiation of cells occurs when they are at the “resting” stage, as is the case here for TCRβ rearrangement. It is probable that the shutoff of DC potential additionally require a drastic change in the differentiation program by completely closing a broad spectrum of loci related to the myeloid lineage. The question addressed in the present study is whether the shutoff of DC potential also takes place at the DN3 stage in synchrony with the initiation of TCRβ gene rearrangement. We determined the developmental timing of these two events and found that DC potential is shutoff within the DN2 stage, while TCRβ rearrangement occurs at the DN3 stage, and that these two events are separated by several cell divisions.

C57BL/6 (B6) mice were purchased from SLC. B6Ly5.1 mice were maintained in our animal facility. Transgenic mice of B6 background carrying enhanced GFP that is expressed under the control of the proximal lck promoter (plck-GFP mice) (17) were maintained in our animal facility. Embryos at 15 days postcoitum (dpc) were obtained from timed pregnancies of B6, B6Ly5.1, and plck-GFP mice. The day of finding the vaginal plug was designated as 0 dpc.

The following Abs were purchased from BD Pharmingen: anti-Ly5.1 (A20), anti-Ly5.2 (104), anti-c-kit (2B8), anti-erythroid lineage cells (TER119), anti-Gr-1 (RB6-8C5), anti-B220 (RA3-6B2), anti-CD8 (53-6.7), anti-CD4 (H129.19), anti-NK1.1 (PK136), anti-CD3ε (145-2C11), anti-CD25 (PC61), anti-CD44 (IM7). CD4, CD8, CD3ε, TER, B220, Gr-1, and NK1.1 were used as lineage markers (Lin).

Recombinant murine (rm) stem cell factor (SCF), rmIL-2, rmIL-3, rmIL-7, rmGM-CSF, rmFlt3 ligand (Flt3L), rmIL-1α, and rmTNF-α were purchased from Genzyme Techne.

RPMI 1640 medium (Invitrogen Life Technologies) supplemented with 10% FCS (M.A. Bioproducts), l-glutamine (2 mM), sodium pyruvate (1 mM), sodium bicarbonate (2 mg/ml), nonessential amino acid solution (0.1 mM; Invitrogen Life Technologies), 2-ME (5 × 10−5 M), streptomycin (100 μg/ml), and penicillin (100 U/ml) were used. A mixture of the following cytokines—SCF (10 ng/ml), IL-3 (10 ng/ml), IL-7 (10 ng/ml), GM-CSF (10 ng/ml), Flt3L (10 ng/ml), IL-1α (10 ng/ml), and TNF-α (10 ng/ml)—was used as a cytokine mixture (7, 9). In the experiments shown in Fig. 1,C, 100 cells were cultured in wells of a 96-well plate (Costar). In the experiments shown in Fig. 1 D, single fetal thymus (FT) cells were individually cultured in the wells of a Terasaki plate (Nunc).

The procedures for the coculture with a deoxyguanosine (dGuo)-treated FT lobe under high oxygen submersion conditions have been described in detail previously (4, 5). Basically, single dGuo-treated FT lobes (B6Ly5.2) were placed into wells of a 96-well V-bottom plate, to which cells from B6Ly5.1 mice to be examined were added. Culture medium was supplemented with SCF (1 ng/ml) and IL-7 (3 ng/ml). The plates were centrifuged at 150 × g for 5 min at room temperature, placed into a plastic bag (Ohmi Odor Air Service), the air inside was replaced by a gas mixture (70% O2, 25% N2, and 5% CO2), and incubated at 37°C. After 10 days of cultivation, cells were harvested from each well. For the detection of T and NK cell potential, the culture system was the same as above except that IL-2 (1 ng/ml) was added to the culture medium (18).

Murine stromal cell line TSt-4 cells were retrovirally transduced with the murine DLL-1 gene (TSt-4/DLL-1) (12). Cells of FT subpopulations (200 or 500 cells/well) were cultured in a well of 24-well plate monolaid with TSt-4/DLL1. After cultivation, cells were harvested by trypsinization, stained in two colors with anti-CD44 and anti-CD25, and analyzed by a flow cytometer. Cells falling on lymphoid gate in scatter characteristics, gating out dead cells by propidium iodide (PI) staining, were flow cytometrically counted.

RT-PCR was performed as described previously (4, 12). Primers used were: PU.1 sense: 5′-AGATGCACGTCCTCGATACT-3′, PU.1 antisense: 5′-TTGTGCTTGGACGAGAACTG-3′; Gata-3 sense: 5′-TCGGCCATTCGTACATGGAA-3′, Gata-3 antisense: 5′-GAGAGCCGTGGTGGATGGAC-3′; CD3ε sense: 5′-ATCACTCTGGGCTTGCTGAT-3′, CD3ε antisense: 5′-TAGTCTGGGTTGGGAACAGG-3′; pTα sense: 5′-AACAGGTAGCTCCTGGCTGCA-3′, pTα antisense: 5′-CAGGAAGAACAAAGCCC-3′; Tcf-1 sense: 5′-CCAGCTTTCTCCACTCTACG-3′, Tcf-1 antisense: 5′-TCAAGGATGGGTGGGTGAAC-3′; ikaros sense: 5′-GAGGCATGGCCAGTAATGTT-3′, ikaros antisense: 5′-AGGCCGTTCACCAGTA TGAC-3′; Rag2 sense: 5′-CCCAGAGAACCACAGAAAAAT-3′, Rag2 antisense: 5′-TAACCACCCACAATAACAAAT-3′; β-actin sense: 5′-TCCTGTGGCATCCATGAAACT-3′; β-actin antisense: 5′-GAAGCACTTG CGGTGCACGAT-3′.

Cycling times and temperatures were as follows: denaturation at 94°C for 30 s, annealing at 60°C for 30 s, and elongation at 72°C for 30 s. Amplification was performed for 25 cycles for β-actin and 35 cycles for all other genes. The PCR products were electrophoresed through a 1.2% agarose gel and stained with ethidium bromide.

The amount of nuclear DNA was determined by PI staining as follows. Cells were fixed in 50% ethanol, washed, and incubated in PBS containing 1 mg/ml RNase at 37°C for 20 min. The cells were washed in PBS, resuspended in PBS containing 100 μg/ml PI, and analyzed by a flow cytometer.

In determining the pre-β proliferation of progenitors from normal mice, single cells were cultured with a dGuo-treated lobe for 12 days. Genomic DNA extracted from cells generated in each well (1000 cells equivalent) was PCR amplified using primers: Dβ1, 5′-TTATCTGGTGGTTTCTTCCAGC-3′; Jβ1.5, 5′-CAGAGTTCCATTTCAGAACCTAGC-3′; Dβ2, 5′-GCACCTGTGGGGAAGAAACT-3′; Jβ2.6, 5′-TGAGAGCTGTCTCCTACTATCGATT-3′. The reaction volume was 20 μl containing 5 μl of the cell extract (equivalent to 1000 cells), 1.5 μl of 10× PCR buffer, 0.16 μl of 25 mM dNTPs, 0.4 μl of each primer (10 mM), and 0.6 U of Taq polymerase. Thermocycling conditions were as follows: 5 min at 94°C followed by 35 cycles of 1 min at 94°C, 1 min at 60°C, 2 min at 72°C, and finally 10 min at 72°C. Resulting products were electrophoresed through an agarose gel and stained with ethidium bromide.

The extent of pre-β proliferation was calculated as described previously (5). The detected number of PCR bands does not necessarily represent the real number of rearranged TCR gene constructs actually present in T cells generated in a clonal culture, because overlap between bands is possible. Therefore, as the first step, the prediction of the actual number of D-J rearrangement constructs per clonal culture was statistically calculated. The next step is to estimate how many cells have initiated TCRβ rearrangement to generate the predicted number of D-J constructs. This step was done by simulation based on assumptions that 1) D-J rearrangement occurs in both alleles, 2) V-DJ rearrangement subsequently occurs in one allele and, if it is successful, V-DJ rearrangement does not occur in the other allele due to allelic exclusion, but if it fails, V-DJ rearrangement occurs in the second allele, and 3) all segments were used equally. It is important to understand that a D-J construct becomes undetectable by PCR when V-DJ rearrangement occurs in the locus. Because the probability for D1-J1, D2-J2, or D1-J2 constructs to remain detectable by PCR differs from each other, the equation for the mathematical calculation is separately given in each experiment based on D1-J1, D2-J2, and D1-J2 PCR data. In this study, PCR was performed for D1-J1.5 and D2-J2.6. When T cells derived from a single progenitor exhibit an average of m bands in the PCR analysis, the different equations predicting pre-β proliferation (N) are as follows: 1) from D1-J1.5 PCR data: n = 25.8 × (−ln(5 − m)/5); 2) from D2-J2.6 PCR data: n = 15.6 × (−ln(6 − m)/6). Theoretically, if PCR analysis and calculation is done on the same sample, these two calculations for pre-β proliferation should result in a similar value.

Some PCR bands are very faint and therefore judgment for positive bands sometimes depends upon the individual. In our analysis, band counting was performed by four researchers and it was found that the numbers fluctuate about ±5% on an average among the four persons. Thus, we showed one representative data set for band number and value of pre-β expansion, together with the estimated range for each value of pre-β expansion, assuming that the band number count allows a ±5% margin.

Transgenic mice in which the expression of GFP is driven by the proximal lck promoter (plck) (17) were used to characterize the early stages of T cell development in the thymus. Flow cytometric profiles for expression of CD44, CD25, and GFP in Lin cells of 15 dpc FT from plck GFP mice are shown in Fig. 1,A. GFP expression starts at DN2 stage, and DN2 cells can be subdivided into GFP and GFP+ populations. Similar DN2 subpopulations are also seen in the adult thymus (AT) of the plck-GFP mice (data not shown). Because the DN2 stage is also characterized by c-kit expression, we examined whether GFP expression is initiated in c-kit+ cells. Expression profiles of c-kit vs CD25 in GFP and GFP+ fractions in 15 dpc FT are shown in comparison with that of CD44 vs CD25 in Fig. 1 B. The results show that GFP expression starts at the c-kit+CD25+ stage.

DN1, GFPDN2, GFP+DN2, and DN3 cells from 15 dpc FT were sorted and cultured at 100 cells/well in the presence of cytokine mixture that can induce DC generation. Several thousands of DC were generated in wells where DN1 cells or GFPDN2 cells were cultured, while neither GFP+DN2 cells nor DN3 cells gave rise to DC (Fig. 1,C). We then determined the frequency of DC-generating progenitors in FT subpopulations by culturing single cells in wells of a Terasaki plate. GFPDN2 cells contained progenitors with DC potential at a level comparable to that of DN1 cells (Fig. 1 D). These data indicate that the DC potential is retained until the GFPDN2 stage while it is completely shut off during the transition from the GFPDN2 to GFP+DN2 stages. Similar results were obtained with AT progenitors (data not shown).

We also compared the NK potential of cells in GFPDN2 and GFP+DN2 subpopulations. One hundred cells were cultured with a dGuo lobe in the presence of IL-2. The GFPDN2 cells generated a distinct population of NK cells (NK1.1+CD3 cells), whereas a very small number of NK cells were seen in the culture of GFP+DN2 cells (Fig. 1 E). Thus, the shutoff of NK potential may also occur at around the GFP to GFP+ transition step.

To confirm the progenitor-progeny relationship of GFPDN2 and GFP+DN2 cells, we cultured GFPDN2 cells on a monolayer of the stromal cells TSt-4/DLL1 (12). Results are shown in Fig. 2. On day 1, half of the cultured cells expressed GFP, and on day 2, almost all cells were found to express GFP. During these 2 days of culture, an ∼4-fold cell increase was seen, suggesting that the transition from the GFPDN2 stage to the GFP+DN2 stage is accompanied by cell proliferation. In the CD44 vs CD25 flow cytometric profile of cells on day 1, cells that have already down-regulated CD44 are seen. Therefore, the possibility that some GFPDN2 cells initiate down-regulation of CD44 without passing through the GFP+DN2 stage cannot be ruled out.

We next examined the expression of several developmentally regulated genes by RT-PCR in the FT subpopulations (Fig. 3). The myeloid transcription factor PU.1 is expressed in cells at the GFPDN2 stage, although it is significantly down-regulated compared with the DN1 stage. The expression of PU.1 becomes undetectable at the GFP+DN2 stage, consistent with our finding that DC potential was completely lost at this stage. Reciprocally, expression of GATA3, a zinc finger transcription factor crucial for T cell development, is up-regulated at the GFP+DN2 stage. Expression of Rag2 at the GFP+DN2 stage was found to be as high as at the DN3 stage. Of note, molecules that are known to be required at the DN3 stage to proceed further to the DN4 stage, e.g., CD3ε and pTα, have already begun to be highly expressed at the GFP+DN2 stage. Therefore, the shutoff of DC potential in progenitors seems to take place concurrently with up-regulation of T cell lineage-specific genes.

The above findings indicated that the DC potential of thymocytes is shutoff within DN2 cells, a stage during which there is extensive proliferation. However, it seemed unclear whether the shutoff of DC potential takes place during or after the proliferative phase. Thus, we investigated whether GFP+DN2 cells undergo cell division before the TCRβ rearrangement.

We first analyzed the configuration of D-J loci of the TCRβ gene, which is the first gene where rearrangement of the TCR gene occurs, by genomic PCR in cells in the FT subpopulations. Previous reports suggested that D-J rearrangement of the TCRβ chain gene is initiated at the DN1 or DN2 stage in a small proportion of cells (19, 20, 21). Indeed, Dβ1-Jβ1 (D1-J1) rearrangements were detected in both GFP and GFP+DN2 cells, but the D1-J1 locus remained mostly unrearranged at this stage (Fig. 4,A). Rearrangements of the Dβ2-Jβ2 (D2-J2) locus were already detected in DN1 cells, but again the frequency of rearranged loci remained low up to the GFP+DN2 stage (Fig. 4 B). In both D1-J1 and D2-J2 loci, rearranged constructs dominate the unrearranged germline configuration at the DN3 stage. Thus, GFP+DN2 cells are closer to GFPDN2 cells than to DN3 cells in terms of their TCRβ gene rearrangement status.

Because some D-J rearrangements were detected in DN1 cells and GFPDN2 cells, we examined whether the DC generated from these cells harbor rearranged D-J constructs. DC generation was induced from DN1 and GFPDN2 cells and from Linc-kit+Sca-1+ 15 dpc fetal liver cells by culturing these cells in the same manner as in Fig. 1,C, and it was found that the DC derived from these cells carry the unrearranged germline configuration in both D1-J1 and D2-J2 loci (Fig. 4, C and D).

We then investigated whether GFP+DN2 cells undergo proliferation before they become DN3 cells. Analysis of the forward and side scatter characteristics of cells in these FT subpopulations indicated that GFP+DN2 cells are somehow smaller than GFPDN2 cells, but still much larger than DN3 cells (Fig. 4,E). Cell cycle analysis demonstrated that the GFP+DN2 fraction contains even more cycling cells than DN1 or GFPDN2 fractions (p < 0.05) (Fig. 4, F and G). These findings suggested that the GFP+DN2 cells undergo proliferation to some extent before entering the DN3 stage.

We then cultured GFP+DN2 cells on TSt-4/DLL1. Virtually all cells differentiated into CD44CD25+ cells within 1 day, and a 5-fold expansion in cell number was observed during 3 days of culture (Fig. 4, H and I). These data strongly suggested that GFP+DN2 cells undergo cell divisions before the TCRβ rearrangement.

To obtain more direct evidence that a proliferative phase exists between the DC shutoff and TCRβ rearrangement-initiation points, we cultured single progenitors with dGuo lobes and investigated whether diverse D-J rearrangements of the TCRβ locus were formed in the progeny T cells. The formation of Dβ-Jβ rearrangement diversity in this clonal culture indicates that the seeded cell has proliferated before the initiation of TCRβ gene rearrangement (5). Representative flow cytometric profiles of T cells generated from single progenitors in various FT subpopulations are shown in Fig. 5 A. Virtually all clones contained CD4+CD8+ (double-positive (DP)) cells, CD4+CD8 (CD4 single-positive (SP)) cells, and CD4CD8+ (CD8SP) cells. As expected, clones generated from more differentiated progenitors tended to contain more SP cells and fewer DN cells (DN3>DN2>DN1), although no significant difference was seen between clones originating from GFPDN2 and GFP+DN2 progenitors.

PCR analysis of Dβ2-Jβ2 rearrangements in cells of each clone detected the rearranged D2-J2 constructs (maximum six) formed in each clonal culture. It was seen that most of the clones from the GFP+DN2 population exhibited multiple bands (Fig. 5,B). The number of bands differs among GFP+DN2 clones, reflecting the heterogeneity in proliferation potential of these progenitors. The average number of bands in 16 GFP+DN2 clones was 3.75 (Table I). According to a mathematical prediction based on simulation of TCRβ chain gene rearrangement (5), it is estimated that these clones have produced an average of 15.3 cells before the TCRβ gene rearrangement. This result indicates that about four cell divisions on average have occurred between the shutoff of DC potential and the initiation of TCRβ rearrangement.

Previous studies have suggested that the D1-J1 rearrangement precedes the D2-J2 rearrangement (14, 21). This leaves the possibility remaining that D1-J1 rearrangement occurs before the time point to shutoff the DC potential. We therefore decided to analyze the rearrangement status of the D1-J1 locus together with the D2-J2 locus in cells generated in clonal cultures. A total of 9 GFPDN2 clones and 19 GFP+DN2 clones were analyzed (Fig. 5, C and D). Multiple D1-J1 bands were seen in most of the GFP+DN2 clones. The average number of D1-J1 bands among the five visible rearrangements in 19 GFP+DN2 clones was 2.47, while that of D2-J2 bands in six constructs was 3.89 (Table II). According to the mathematical prediction (as described in the Materials and Methods), seeded GFP+DN2 cells were estimated to have produced 17.6 cells on average before the initiation of D1-J1 rearrangement. Using the same sample, it was also predicted that 16.3 cells were formed before D2-J2 rearrangement. This value (16.3), which was quite similar to the value (15.3) obtained from D2-J2 PCR data in the previous experiment (Table I), was very close to the value (17.6) obtained from D1-J1 data. These results thus clearly indicated that these progenitors shutoff their DC potentials just before the initiation of rearrangement in either the D1-J1 or D2-J2 locus. The results also indicated that the timing to start the rearrangement of D1-J1 and D2-J2 loci is almost the same between the two.

The presence of a proliferation phase between lineage restriction and TCRβ rearrangement suggests that the shutoff of DC potential does not directly interact with the mechanisms that drive the TCRβ gene rearrangement and vice versa.

It has been unclear at which stage the intrathymic progenitors become completely determined to the T cell lineage. By using plck-GFP mice as the progenitor source, the present study clarified that the T cell lineage determination occurs at late DN2 stage, which is before the initiation of TCRβ gene rearrangement, and that a proliferation phase exists between the stage of T cell lineage determination and that of TCRβ gene rearrangement.

Fig. 6 schematically illustrates the timetable of early intrathymic T cell development before the TCRβ rearrangement. During DN1 and DN2 stages in FT, T cell progenitors undergo extensive (>10) cell divisions (4). Such an extensive proliferation before TCRβ rearrangement may be important in generating a diversity of TCRβ chains. This pre-β proliferation phase can be divided into two phases, i.e., the phase of T cell lineage specified progenitors and that of T cell lineage determined progenitors that have lost DC potential. In this context, it can be said that the ∼16 fold-expansion (i.e., four round cell divisions) revealed in the present study is the first task of the T cell lineage-determined progenitors for producing an extensively diversified T cell repertoire. Although this timetable of T cell lineage determination and proliferation is based on the findings on FT progenitors, our experimental results indicated that the timetable was almost the same in AT progenitors (data not shown), except that the earliest AT progenitors may undergo at least three times more cell divisions than the FT progenitors at DN1 stage (22).

We have previously reported that the DN1 population of the AT contains a small fraction of GFP+ cells, and these cells were considered to belong to a more advanced stage toward T cells, because they have lost the potential to generate B cells, NK cells, and DC (17). Such GFP+ DN1 cells were rarely seen in FT (Fig. 1, A and B). We speculate that GFP expression at the DN1 stage represents a noncanonical pathway occurring only in AT, and that in both the fetus and adult, the GFP expression at the DN2 stage represents the major developmental pathway toward αβ T cells.

Previous studies showed that rearranged D-J constructs of the TCRβ gene are detectable in DN1 and/or DN2 cells (19, 20, 21). Hence, it has been proposed that D-J rearrangement of the TCRβ gene proceeds gradually, starting at the DN1 or DN2 stage. We also detected rearranged Dβ-Jβ constructs in DN1 or DN2 cells (Fig. 4,A). However, detection of Dβ-Jβ constructs in cells at these early stages does not necessarily prove that some of the progenitors initiated their rearrangements of Dβ-Jβ at the DN1 or DN2 stages. As shown in the present study, even in GFP+DN2 clones, at least 3 bands among 11 possible bands (by D1-J1.5 and D2-J2.6 PCR) were detected in all 19 clones analyzed (Fig. 5, C and D). The failure to detect clones that contain less D-J constructs than three is not because we picked up only the cultures derived from very proliferative progenitors, because the plating efficiency of clonal culture is always >30% for GFP+DN2 cells, and all clones in which DP cell generation was observed were served for PCR. It might be still argued that D-J rearrangement at the DN1 or DN2 stages takes place solely on either the D1-J1 locus or the D2-J2 locus, thus allowing the possibility of generating diversified D-J rearrangements in progeny cells. However, our data do not support this possibility, because most GFP+DN2 clones already contain at least two bands for both loci (Fig. 5, C and D). Thus, it is quite unlikely that these T cell clones were derived from cells that had already initiated partial D-J rearrangement. Therefore, the initiation of Dβ-Jβ rearrangement at the DN1 or DN2 stage, even if it occurs, represents a very rare event.

It has been shown that in some experimental settings, B cells or macrophages carrying rearranged Dβ-Jβ constructs are generated from thymic progenitors (14, 23). These results were interpreted to indicate that T cell progenitors that have undergone Dβ-Jβ rearrangement still retain the potential to give rise to other lineage cells than the T cell lineage. However, in the case of B cells, it could alternatively be argued that the Dβ-Jβ rearrangements occurred after progenitors were segregated from the main pathway for T cells, because Rag and various genes involved in gene rearrangement are also expressed during B cell development. In the case of macrophages, Balciunaite et al. (14) showed that the frequency of macrophage colonies generated from DN2 cells that harbor rearranged Dβ-Jβ construct is only 3%. Their data therefore are consistent with our data in which rearranged Dβ-Jβ constructs were not detected in DC generated from GFPDN2 cells (Fig. 4, C and D). Another interpretation, based on the finding that the Dβ-Jβ constructs in B cells or myeloid cells derived from thymic progenitors are always that of the D1-J1 locus (14, 21, 23), proposed that the rearrangement of the D1-J1 locus precedes that of the D2-J2 locus. However, our present results did not support such a scenario, because GFP+DN2 cells were estimated to proliferate at a similar rate before the rearrangement of the D1-J1 and D2-J2 loci (Table II). The selective rearrangement of the D1-J1 locus in certain conditions may indicate that the D1-J1 and D2-J2 loci are differently regulated. Therefore, the possible explanation for selective D1-J1 rearrangement in thymus-derived nonthymic lineage cells is that the D2-J2 locus is more strictly regulated than the D1-J1 locus to prevent rearrangement in non-T lineage cells.

Another important characteristic of thymic T cell progenitors is whether the DC potential is retained in all T cell progenitors until they become GFP+ or whether some T cell progenitors shutoff the DC potential as early as the DN1 stage. Our previous data indicated that the frequency of progenitors exhibiting DC potential among progenitors that showed T cell potential in the DN1 and DN2 populations was ∼50–60% in the DN1 population and 10–20% in the DN2 population (9, 11, 22). However, considering that any clonal analysis has some limitation in sensitivity, the finding that over 50% of the T cell progenitors in the DN1 population have DC potential may imply that most DN1 T cell progenitors could have retained DC potential. Although our clonal analysis using a fetal thymic organ culture system showed that the DC-generating progenitors within the DN2 population are lower than that in the DN1 population, a more sensitive analysis by culturing cells in the presence of a cytokine mixture in Terasaki plates revealed that DN2 cells contain DC-generating precursors at a comparable frequency as DN1 cells (Fig. 1 D). Therefore, it is conceivable that the potential of thymic progenitors to generate DC might be retained until they enter the GFP+DN2 stage.

Collectively, we propose that the time point of T cell lineage determination precedes that of the initiation of TCRβ gene rearrangement by several cell divisions and the timings of these events are strictly regulated. In contrast, initiation of D-J rearrangement in the IgH gene in B lineage cells may not be so strictly regulated. Although it has been shown that the early B cell progenitors bear an unrearranged D-J locus in their IgH gene (3) as mentioned in the Introduction, some recent studies demonstrated that the early lymphoid progenitors in bone marrow express Rag genes and harbor the rearranged D-J construct in the IgH gene (24, 25). It was recently shown that non-B lineage cells, i.e., a subset of plasmacytoid DCs, bear a rearranged D-J IgH gene (26, 27). It might be interesting to study these differences in initiating the gene rearrangement between T and B cell lineages.

The DC potential retained in T cell progenitors may provide a clue as to the relationship among various hemopoietic cell lineages. We have previously shown that even after multipotent progenitors are segregated into branches for erythroid, T and B cell lineages, progenitors of each branch still retain myeloid potential (4, 28, 29, 30). This may indicate that the developmental program for erythroid, B and T cell lineages proceeds on a basis of an underlying prototypical myeloid program. From this perspective, the DC potential in T cell progenitors could represent a modified form of myeloid potential in a T cell branch. Thus, the shutoff of DC potential can be regarded as the final event for the progenitor to determine its identity as a T cell progenitor. Down-regulation of PU.1, a key transcription factor for the myeloid lineage (31, 32), was observed coincident with the shutoff of DC potential (Fig. 3). Therefore, it seems that the down-regulation of PU.1 at this transition point may account for the shutoff of myeloid potential in T cell progenitors. This transition step is also accompanied by the up-regulation of T cell lineage-specific molecules, e.g., pTα, CD3ε, which are essential to pass the DN3 stage checkpoint and to enter DN4 and DP stages.

In general, differentiation of cells occurs during their nonproliferating phase. In the case of T cell development, TCRβ and TCRα chain genes are rearranged at the DN3 and DP stages, respectively, when cells are in a quiescent status. Positive selection also proceeds at the DP stage. The shutoff of DC potential is also one of the dramatic events in thymic T cell development, but this event appears to occur in a midproliferation phase. Whether molecular mechanisms that actually shutoff DC potential take place in the G1 phase of cycling cells or at some resting G0 period within the DN2 stage remains to be clarified. In this context, the fact that GFP+DN2 cells contain more cycling cells than GFPDN2 cells (Fig. 4, D and E) suggests that the GFPDN2 stage contains some resting cells in which the determination process may take place, whereas GFP+DN2 cells contain only cells that have passed this critical step and are proliferating before they enter DN3 stage.

Although DC potential was found to be completely shutoff at the transition between the GFPDN2 and GFP+DN2 stages, NK potential does not seem to be so completely lost at this step. It is probable that the NK program has to be more or less maintained beyond this step or even beyond the DN3 stage, because thymic precursors have to produce NKT cells, which share various characteristics with NK cells, after TCR gene rearrangements (33). Further studies are needed to clarify the molecular mechanisms by which NK-generating potential is shutoff in relation to the shutoff of DC potential during T cell development.

We thank Drs. Peter D. Burrows and Wilfred T. V. Germeraad for critical reading of the manuscript.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by the Special Coordination Funds for Promoting Science and Technology from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

3

Abbreviations used in this paper: DC, dendritic cell; DN, double negative; dpc, days post coitum; rm, recombinant murine; SCF, stem cell factor; Flt3L, Flt3 ligand; FT, fetal thymus; dGuo, deoxyguanosine; PI, propidium iodide; AT, adult thymus; DP, double positive; SP, single positive.

1
Schlissel, M. S., P. Stanhope-Baker.
1997
. Accessibility and the developmental regulation of V(D)J recombination.
Semin. Immunol.
9
:
161
-170.
2
Sleckman, B. P., C. H. Bassing, C. G. Bardon, A. Okada, B. Khor, J. C. Bories, R. Monroe, F. W. Alt.
1998
. Accessibility control of variable region gene assembly during T-cell development.
Immunol. Rev.
165
:
121
-130.
3
Allman, D., J. Li, R. R. Hardy.
1999
. Commitment to the B lymphoid lineage occurs before DH-JH recombination.
J. Exp. Med.
189
:
735
-740.
4
Kawamoto, H., T. Ikawa, K. Ohmura, S. Fujimoto, Y. Katsura.
2000
. T cell progenitors emerge earlier than B cell progenitors in the murine fetal liver.
Immunity
12
:
441
-450.
5
Kawamoto, H., K. Ohmura, S. Fujimoto, M. Lu, T. Ikawa, Y. Katsura.
2003
. Extensive proliferation of T cell lineage-restricted progenitors in the thymus: an essential process for clonal expression of diverse T cell receptor β chains.
Eur. J. Immunol.
33
:
606
-615.
6
Wu, L., C. L. Li, K. Shortman.
1996
. Thymic dendritic cell precursors: relationship to the T lymphocyte lineage and phenotype of the dendritic cell progeny.
J. Exp. Med.
184
:
903
-911.
7
Saunders, D., K. Lucas, J. Ismaili, L. Wu, E. Maraskovsky, A. Dunn, K. Shortman.
1996
. Dendritic cell development in culture from thymic precursor cells in the absence of granulocyte/macrophage colony-stimulating factor.
J. Exp. Med.
184
:
2185
-2196.
8
Lucas, K., D. Vremec, L. Wu, K. Shortman.
1998
. A linkage between dendritic cell and T-cell development in the mouse thymus: the capacity of sequential T-cell precursors to form dendritic cells in culture.
Dev. Comp. Immunol.
22
:
339
-349.
9
Shen, H. Q., M. Lu, T. Ikawa, K. Masuda, K. Ohmura, N. Minato, Y. Katsura, H. Kawamoto.
2003
. T/NK bipotent progenitors in the thymus retain the potential to generate dendritic cells.
J. Immunol.
171
:
3401
-3406.
10
Ikawa, T., K. Masuda, M. Lu, N. Minato, Y. Katsura, H. Kawamoto.
2004
. Identification of the earliest prethymic T cell progenitors in murine fetal blood.
Blood
103
:
530
-537.
11
Masuda, K., M. Itoi, T. Amagai, N. Minato, Y. Katsura, H. Kawamoto.
2005
. Thymic anlage is colonized by progenitors restricted to T, NK and dendritic cell lineages.
J. Immunol.
174
:
2525
-2532.
12
Masuda, K., H. Kubagawa, T. Ikawa, C. C. Chen, K. Kakugawa, M. Hattori, R. Kageyama, M. D. Cooper, N. Minato, Y. Katsura, H. Kawamoto.
2005
. Prethymic T-cell development defined by the expression of paired immunoglobulin-like receptors.
EMBO J.
24
:
4052
-4060.
13
Lee, C. K., J. K. Kim, Y. Kim, M. K. Lee, K. Kim, J. K. Kang, R. Hofmeister, S. K. Durum, S. S. Han.
2001
. Generation of macrophages from early T progenitors in vitro.
J. Immunol.
166
:
5964
-5969.
14
Balciunaite, G., R. Ceredig, A. G. Rolink.
2005
. The earliest subpopulation of mouse thymocytes contains potent T, significant macrophage, and natural killer cell but no B-lymphocyte potential.
Blood
105
:
1930
-1936.
15
Slack, J. M. W..
1991
.
From Egg to Embryo: Regional Specification in Early Development
Cambridge University Press, New York.
16
Penit, C., B. Lucas, F. Vasseur.
1995
. Cell expansion and growth arrest phases during the transition from precursor (CD48) to immature (CD4+8+) thymocytes in normal and genetically modified mice.
J. Immunol.
154
:
5103
-5113.
17
Shimizu, C., H. Kawamoto, M. Yamashita, M. Kimura, E. Kondou, Y. Kaneko, S. Okada, T. Tokuhisa, M. Yokoyama, M. Taniguchi, et al
2001
. Progression of T cell lineage restriction in the earliest subpopulation of murine adult thymus visualized by the expression of lck proximal promoter activity.
Int. Immunol.
13
:
105
-117.
18
Ikawa, T., H. Kawamoto, S. Fujimoto, Y. Katsura.
1999
. Commitment of common T/natural killer (NK) progenitors to unipotent T and NK progenitors in the murine fetal thymus revealed by a single progenitor assay.
J. Exp. Med.
190
:
1617
-1626.
19
Capone, M., R. D. Hockett, Jr, A. Zlotnik.
1998
. Kinetics of T cell receptor β, γ, and δ rearrangements during adult thymic development: T cell receptor rearrangements are present in CD44+CD25+ pro-T thymocytes.
Proc. Natl. Acad. Sci. USA
95
:
12522
-12527.
20
Livak, F., M. Tourigny, D. G. Schatz, H. T. Petrie.
1999
. Characterization of TCR gene rearrangements during adult murine T cell development.
J. Immunol.
162
:
2575
-2580.
21
King, A. G., M. Kondo, D. C. Scherer, I. L. Weissman.
2002
. Lineage infidelity in myeloid cells with TCR gene rearrangement: a latent developmental potential of pro T cells revealed by ectopic cytokine receptor signaling.
Proc. Natl. Acad. Sci. USA
99
:
4508
-4513.
22
Lu, M., R. Tayu, T. Ikawa, K. Masuda, I. Matsumoto, H. Mugishima, H. Kawamoto, Y. Katsura.
2005
. The earliest thymic progenitors in adults are restricted to T, NK, and dendritic cell lineage and have a potential to form more diverse TCRβ chains than fetal progenitors.
J. Immunol.
175
:
5848
-5856.
23
Allman, D., A. Sambandam, S. Kim, J. P. Miller, A. Pagan, D. Well, A. Meraz, A. Bhandoola.
2003
. Thymopoiesis independent of common lymphoid progenitors.
Nat. Immunol.
4
:
168
-174.
24
Igarashi, H., S. C. Gregory, T. Yokota, N. Sakaguchi, P. W. Kincade.
2002
. Transcription from the RAG1 locus marks the earliest lymphocyte progenitors in bone marrow.
Immunity
17
:
117
-130.
25
Borghesi, L., L. Y. Hsu, J. P. Miller, M. Anderson, L. Herzenberg, M. S. Schlissel, D. Allman, R. M. Gerstein.
2004
. B lineage-specific regulation of V(D)J recombinase activity is established in common lymphoid progenitors.
J. Exp. Med.
199
:
491
-502.
26
Corcoran, L., I. Ferrero, D. Vremec, K. Lucas, J. Waithman, M. O’Keeffe, L. Wu, A. Wilson, K. Shortman.
2003
. The lymphoid past of mouse plasmacytoid cells and thymic dendritic cells.
J. Immunol.
170
:
4926
-4932.
27
Pelayo, R., J. Hirose, J. Huang, K. P. Garrett, A. Delogu, M. Busslinger, P. W. Kincade.
2005
. Derivation of 2 categories of plasmacytoid dendritic cells in murine bone marrow.
Blood
105
:
4407
-4415.
28
Lu, M., H. Kawamoto, Y. Katsube, T. Ikawa, Y. Katsura.
2002
. The common myelolymphoid progenitor: a key intermediate stage in hemopoiesis generating T and B cells.
J. Immunol.
169
:
3519
-3525.
29
Katsura, Y..
2002
. Redefinition of lymphoid progenitors.
Nat. Rev. Immunol.
2
:
127
-132.
30
Kawamoto, H..
2006
. A close developmental relationship between the lymphoid and myeloid lineages.
Trends Immunol.
27
:
169
-175.
31
Singh, H..
1996
. Gene targeting reveals a hierarchy of transcription factors regulating specification of lymphoid cell fates.
Curr. Opin. Immunol.
8
:
160
-165.
32
Anderson, M. K., A. H. Weiss, G. Hernandez-Hoyos, C. J. Dionne, E. V. Rothenberg.
2002
. Constitutive expression of PU.1 in fetal hematopoietic progenitors blocks T cell development at the pro-T cell stage.
Immunity
16
:
285
-296.
33
Taniguchi, M., K. Seino, T. Nakayama.
2003
. The NKT cell system: bridging innate and acquired immunity.
Nat. Immunol.
4
:
1164
-1165.