These studies address the role of PU.1 in T cell development through the analysis of PU.1−/− mice. We show that the majority of PU.1−/− thymocytes are blocked in differentiation prior to T cell commitment, and contain a population of thymocyte progenitors with the cell surface phenotype of CD44+, HSAbright, c-kitint, Thy-1−, CD25−, Sca-1−, CD4−, and CD8−. These cells correspond in both number and cell surface phenotype with uncommitted thymocyte progenitors found in wild-type fetal thymus. RT-PCR analysis demonstrated that PU.1 is normally expressed in this early progenitor population, but is down-regulated during T cell commitment. Rare PU.1−/− thymi, however, contained small numbers of thymocytes expressing markers of T cell commitment. Furthermore, almost 40% of PU.1−/− thymi placed in fetal thymic organ culture are capable of T cell development. Mature PU.1−/− thymocytes generated during organ culture proliferated and produced IL-2 in response to stimulation through the TCR. These data demonstrate that PU.1 is not absolutely required for T cell development, but does play a role in efficient commitment and/or early differentiation of most T progenitors.
PU.1 is an ets family, winged helix-turn-helix-containing transcription factor expressed at high levels in myeloid and B lymphoid cells. Low level expression has also been detected in erythroid progenitors and in thymus (1, 2). PU.1 has been shown to regulate many lymphoid- and myeloid-specific genes important for cellular function (reviewed in 3), and therefore PU.1 is likely to play a role in the execution of both lymphoid and myeloid differentiation programs. In support of this proposed PU.1 function, targeted disruption of the PU.1 gene results in a complete block in development of B lymphoid and macrophage lineages (4). All myeloid lineages, including granulocytes, osteoclasts, and myeloid dendritic cells,3 were found to be severely affected by the PU.1 mutation. B lymphoid cells were blocked at an early pre-commitment stage, since no rag, μ, λ, or B220 expression, and no Ig rearrangements were detected. No thymocytes positive for the T markers Thy-1 or CD2 were detected, and the thymus was relatively hypocellular. In addition, the PU.1 deficiency resulted in late embryonic lethality due to unknown causes. These results strongly suggest that both lymphoid and myeloid cells require PU.1.
A second strain of PU.1−/− mice was independently generated by McKercher et al. (5). Where examined, the embryonic phenotype of the two mutations is nearly identical. If, however, McKercher et al. maintained PU.1−/− pups on antibiotics, they could survive up to 2 wk. Surviving animals were found to contain increasing numbers of thymocytes expressing CD4 and CD8 and TCR. By day 12 after birth, many PU.1−/− thymi contained thymocyte numbers within fivefold of wild type, and were relatively normal with respect to expression of CD4, CD8, and TCR. Mature phenotype T cells were also found in the spleen. Interestingly, older antibiotic-treated mice also contained small but significant numbers of B220, rag-expressing cells, possibly aberrant B progenitors, and rare immature and nonfunctional neutrophils. The basis for the differences in the two models is currently unexplained. Possibilities include strain differences and/or differences in targeting constructs. In light of these discrepancies in phenotype, we have reexamined T cell development in PU.1−/− fetal thymus.
The thymus is seeded on an ongoing basis with hemopoietic progenitors derived from fetal liver or bone marrow. These incoming progenitors do not self-renew within the thymus but rather become rapidly committed for differentiation and expansion to the T and associated (NK, thymic dendritic) cell lineages. One hypothesis of lineage commitment suggests that initial specification events induce the expression of particular transcription factors that replace or supplement transcription factors already present in the pluripotent precursors (6). The key role played by large transcription factor families in lineage-specific gene expression supports this hypothesis. Transcription factor families couple similar DNA-binding specificities with differences in transactivation ability and protein-protein interactions. Divergent patterns of expression within a family can provide the necessary regulation of a specific developmental program. A second feature of this hypothesis is the combinatorial nature of gene regulation that requires multiple transcription factors to work in concert to express a given gene. Specific protein-protein interactions within a higher order protein complex can have as much influence on gene expression as DNA-binding specificity. The ets family of transcription factors is a good example in support of this hypothesis (7). The ets family is composed of at least 17 different members with similar DNA-binding specificities due to related DNA-binding domains. The nonmammalian family members have been implicated in a variety of developmental programs including steroid hormone-induced metamorphosis in Drosophila (8, 9) and oocyte pattern formation in frogs (10). In birds and mammals, ets proteins play an important role in regulating hemopoiesis (3, 7, 11).
Studies from many groups have characterized early T cell development and commitment with respect to a variety of cell surface markers. The earliest, uncommitted progenitors are known to express CD44, c-kit, and lack expression of TCR, CD4, CD8, and other T-specific markers. Commitment of these progenitors to T, NK, and lymphoid dendritic cell lineages has been correlated with expression of a variety of markers. For example, Hattori et al. have determined that expression of FcγRII/III on E12.5-E14.5 thymocytes is correlated with the expression of T-specific factors, including TCF-1, GATA-3, and intracellular CD3ε (12, 13). Carlyle and Zuniga-Pflucker (14, 15) have correlated early thymic expression of the NK1.1 marker with transition to the T/NK bipotential stage. Finally, up-regulation of CD25 and down-regulation of CD44 are associated with commitment to the αβ T cell lineage to the exclusion of other cell types, including γδ T cells, NK cells, and thymic dendritic cells (16, 17, 18). Using this battery of markers, we have characterized the extent of differentiation and commitment among PU.1-deficient fetal thymocytes. We find that the majority of PU.1-deficient fetal thymocytes are blocked at an apparently uncommitted progenitor stage. However, we also find that PU.1-deficient thymus contains an extremely rare population of T committed and early differentiating T progenitors, which go on to develop mature, functionally competent thymocytes with time in fetal thymic organ culture.
Materials and Methods
PU.1-deficient mice were of mixed 129/C57BL/6 background (backcrossed to 129 a maximum of three times) and were bred in our own facility.
Abs, staining, flow cytometric analysis
Thymic lobes were mechanically dissociated and stained using standard procedures. mAbs were: anti-CD4 FITC (Caltag, South San Francisco, CA), anti-CD8 Tricolor (Caltag), and anti-CD3ε PE (Life Technologies, Rockville, MD) or anti-CD4-FITC, anti-CD8-FITC (Caltag), anti-CD25-PE (PharMingen, San Diego, CA), biotinylated-CD44 (a gift of Dr. E. Pure), and Streptavidin Red 670 (Life Technologies), anti-Thy-1 FITC (Becton Dickinson), Sca-1-PE (PharMingen), NK1.1-FITC (PharMingen), HSA-FITC (PharMingen), anti-FcγRII/III (a gift of S. Carding). Flow cytometric analysis was performed on a FACScan and analyzed using CellQuest software (Becton Dickinson). In some cases, flow cytometry was used to quantify the relative cell yield from each thymic lobe. In these cases, thymic lobes were dissociated in equal volumes, 200 μl, which provides a cell density within the limit of accurate detection by the cytometer. Samples were collected for a fixed time, generally 1 min, with the sample tube-washing system disabled. Samples were gated as usual for live and dead cells using forward and side scatter, and the number of live events compared between different samples. The reliability of the technique was confirmed by direct counting of trypan blue excluding cells using a hemacytometer (∼10% of the samples were directly counted).
Thymocyte fractionation and RT-PCR
Thymi were harvested from C57BL/6 mice and single cell suspensions prepared by manual disruption. Thymocyte populations were fractionated via magnetic bead separation as per the manufacturer’s instruction (MACS, Miltenyi Biotec, Auburn, CA) as follows. Thymocytes were stained with B cell (anti-B220) and myeloid (anti-CD11b) specific Abs (PharMingen) and subjected to two rounds of negative selection. The remaining thymocytes were stained with anti-CD4 and anti-CD8 Abs followed by two rounds of positive selection to generate a mixed population of CD4+CD8+ (double-positive), CD4+8−, and CD4−8+ (single-positive) cells. The thymocytes were then stained with anti-CD25 and subjected to two rounds of positive selection to generate the CD25+, CD4−, CD8− population. Finally, the remaining cells were stained with anti-CD44 and subjected to two rounds of positive selection to generate the CD44+, CD25−, CD4−, CD8− population. CD4+ or CD8+ splenic T cells were also selected in a similar manner. All populations were confirmed >95% pure by flow cytometry (data not shown). RNA was prepared using RNAzol (Teltest, Friendswood, TX) following the manufacturers instructions. RNA was also prepared from whole spleen to serve as a positive control. Semiquantitative PU.1 RT-PCR was performed as described previously (19) using the following primers: PU.1, 5′-GAGTTTGAGAACTTCCCTGAG-3′, 5′-TGGTAGGTCATCTTCTTGCGG-3′; ETS-1, 5′-GGAATTCCTGAATACACAGTATAGTGAG-3′, 5′-CGGTCGACTCCTGTGTAGCCAGCCAG-3′; FLI-1, 5′-GGAATTCGGGTCAATGTGTGGAATATTGG-3′, 5′-CGGTCGACCCCAGGGTTTGCTAGGCG-3′; HPRT, 5′-CACAGGACTAGAACACCTGC-3′, 5′-GCTGGTGAAAAGGACCTCT-3′. PCR conditions were established for each transcript to ensure that analysis was undertaken in the linear amplification range. For the abundant HPRT message, 25 amplification cycles were used. For the transcription factor cDNAs, 28 rounds of amplification were used. RT-PCR for the myeloid-specific transcript CD11b was used to confirm the purity of the fractionated thymocyte subset RNA (data not shown). No template reactions also served as a negative control (data not shown).
Fetal thymic organ cultures (FTOCs)4
FTOCs were performed according to procedures previously described (20). Embryos were generated by heterozygous crosses, timed from the first day of plug observation (day 0.5), and embryos dissected and placed in ice-cold RPMI plus 10% FCS. A small piece of liver was dissected from each embryo, dissociated, and stained using anti-CD11b Ab and flow cytometry to identify mutants. The reliability of the CD11b staining to detect mutants was confirmed by Southern analysis of embryo DNA (data not shown). During the staining and analysis (<2 h), all control and mutant littermate embryos were held on ice. Dissected thymic lobes were either pooled and dissociated for staining or cultured in RPMI (Life Technologies) plus 10% FCS (Cansera, Ontario, Canada), glutamine (Life Technologies), gentamicin, and suspended on nuclepore filter rafts (Costar, Cambridge, MA) supported by gelfoam (Upjohn, Kalamazoo, MI). Culture medium was partially replaced every 3 days.
The number of cells plated was normalized such that each well received 2 × 104 TCRhigh thymocytes. A total of 2 × 105 irradiated spleen cells were added to provide costimulation, and cells were exposed to plate-bound anti-CD3ε Ab for 48 h, then pulsed for 12 h with [3H]thymidine and harvested and counted. Since PU.1-deficient thymuses contained fewer TCRhigh thymocytes, many more cells were added. To control for this, background levels of 3H incorporation were determined for equivalently high numbers of wild-type thymocytes, and found to be indistinguishable from controls.
Differences between two means were evaluated using Student’s t test (Microsoft Excel, Redmond, WA).
Expression of PU.1 and other ets family members in thymus and T cells
The ets family members, PU.1, Ets-1, Ets-2, Spi-B, Fli-1, and Elf-1, are all expressed at some point during thymocyte development (7). Ets-2 is ubiquitously expressed, but transgenic overexpression leads to abnormal thymic histology, among other defects (21, 22, 23, 24). Fli-1 is expressed at high levels in the spleen, thymus, and lung with lower levels of expression in skeletal muscle (25). Spi-B is expressed in fetal thymus and spleen with low level expression found in some T cell lines (26). Targeted disruption of PU.1 leads to defects in early stages of lymphoid development as discussed (4, 27), while disruption of Ets-1 leads to reduced mature thymocyte and peripheral T cell numbers, increases in T cell apoptosis, and decreased T cell activation (28, 29). Targeted disruption of Fli-1 results in thymic hypocellularity (30). Therefore, we examined the expression of PU.1, Ets-1, and Fli-1 at different stages of thymocyte development. To produce subsets at different developmental stages, thymocytes were fractionated based on expression of CD4, CD8, CD25, and CD44, and RNA was prepared. Expression levels of the ets family transcription factors PU.1, Ets-1, and Fli-1 were determined by semiquantitative RT-PCR (Fig. 1). PU.1 was found to be expressed in the most immature CD4−8−25−CD44+ thymocytes, at lower levels on CD4−8−CD25+ thymocytes, and was undetectable in more mature mixed populations of CD4+8+, CD4−8+, or CD4+8− thymocytes or splenic T cells. Thus PU.1 expression is inversely correlated with T cell developmental progression. Expression of Ets-1, however, had a nearly opposite pattern of expression with no detectable expression in the most primitive CD4−8−25− CD44+ thymocytes, followed by up-regulated expression as thymocytes become committed to the T cell lineage. Fli-1 was found to be expressed early, and increased as T cells matured. Both Ets-1 and Fli-1 expression peaked in the mixed CD4+8+, CD4+8−, CD4−8+ thymocyte populations and remained high in circulating splenic T cells.
Phenotype of PU.1-deficient thymocytes
To determine precisely the stage at which T cell development is arrested in PU.1−/− animals, embryos were harvested at E14.5, and E16.5 and thymic cellularity and phenotype assessed. When thymic lobes from E14.5 embryos were pooled and counted, mutant thymus contained on average 0.5 × 105 cells per lobe, compared with 1.65 × 105 cells per lobe in wild-type littermate controls. Cell numbers were observed to expand in vivo in wild-type embryos, so that by E16.5 wild-type lobes contained 6.7 ± 1.9 × 105 cells per lobe, while mutants still contained just 0.2 ± 0.1 × 105 cells per lobe. Therefore, at E16.5, wild-type and mutant cell numbers significantly differed (p = 0.001) by over 30-fold.
The expression of the markers Thy-1, Sca 1, NK1.1, HSA, CD44, CD25, FCγRII/III, CD4, and CD8 were determined using flow cytometry. Neither wild-type nor PU.1−/− thymocytes expressed CD4 or CD8 at E14.5, as expected (data not shown). The majority of PU.1−/− thymocytes were smaller on average than wild-type thymocytes as determined by forward scatter (Fig. 2,A). These smaller PU.1−/− thymocytes were Thy-1 and CD25 negative, and expressed intermediate levels of CD44 (second column, Fig. 2,B). A similar population of Thy-1− CD25− CD44+ thymocytes was found in wild-type thymus at E14.5, and these were enriched in the smaller cell gate (third column, Fig. 2,B). The majority of wild-type thymocytes at E13.5 also showed a similar Thy-1− CD25− and CD44+ phenotype (fourth column, Fig. 2 B). Besides this major population, ∼2% of PU.1−/− thymocytes were larger by forward scatter. Among these larger cells, the majority expressed Thy-1 and about half of these expressed CD25.
The majority of PU.1−/− thymocytes stained brightly for HSA and expressed intermediate levels of c-kit (Fig. 2,C). Again, a similar population was observed in wild-type thymus, which was especially enriched if thymocytes were gated for the smaller size (Fig. 2,C). HSAbright c-kitint cells were also enriched in wild-type thymus at E13.5 (Fig. 2 C). Again, PU.1−/− thymi contained a low but reproducible population of HSAlow c-kitbright cells consistent with further developmental maturation.
About half of wild-type thymocytes at E14.5 express FcγRII/III, which has been shown to correlate with the beginning of T differentiation (12, 13, 31) (Fig. 2,D). In PU.1−/− thymus, about 3% of thymocytes were found to express FcγRII/III. In wild-type thymus, the smaller cell gate was depleted of FcγRII/III-expressing cells (Fig. 2 D), as were E13.5 wild-type embryos. Sca-I expression was observed on a subset of E14.5 wild-type thymus, but was not seen in PU.1-deficient thymus and was dramatically depleted on small wild-type cells.
NK1.1 is expressed on committed T/NK progenitors in fetal thymus, and also on mature NK T cells, which have been shown to arise early in ontogeny (16). We found that PU.1−/− thymi contained a small fraction of NK1.1+ thymocytes, similar to that observed in E14.5 wild-type controls gated for smaller size (Fig. 2 E).
When individual E14.5 thymic lobes were examined, Thy-1+CD25+ cells were present in just one out of three lobes. This suggests that the small percentage of Thy-1+CD25+ cells in pooled samples was due to increased development in one or a few lobes, rather than a low percentage of development present in all lobes.
Individual lobes were also examined in the case of E16.5 thymi. The majority of thymocytes expressed the Thy-1− CD25− FcγRII/III− Sca I− NK1.1− CD4− CD8− CD44+ HSAbright c-kitint phenotype described above for PU.1−/− thymocytes at E14.5. In addition, in two cases a small percentage (8–11%) of HSAlowCD8+ cells was observed (one shown in Fig. 3). Although by E16.5 many of the wild-type cells were clearly expressing CD4 and CD8, we were still able to distinguish a small population of HSAbright, CD4−8− wild-type cells phenotypically similar to the predominant PU.1−/− phenotype (boxed population shown in Fig. 3).
In summary, the majority of PU.1−/− thymocytes expressed the following phenotype: Thy-1− CD25− FcγRII/III− Sca I− NK1.1− CD4− CD8− (hereafter abbreviated T-Lin−) CD44+ HSAbright c-kitint. Based on the total cell counts and the percentage of cells in this subset, the absolute number of cells with this phenotype was calculated to be ∼0.3–0.4 × 105 per thymic lobe for both PU.1−/− and wild-type embryos at E14.5. Therefore, PU.1−/− mice do not have reduced numbers of this progenitor subset. At E16.5, the absolute number of these progenitors has dropped to between 0.1 to 0.2 × 105 in both mutant and wild-type embryos.
T cell development in PU.1-deficient FTOCs
The existence of a low percentage of Thy 1+, CD25+, and FcγRII/III+, and NK1.1+ cells in PU.1−/− fetal thymi suggested that some of the earliest steps of T cell development were possible in the absence of PU.1. Since the PU.1 defect is embryonic lethal, we tested this possibility further by incubating E14.5 FTOC for 12–16 days. The thymic lobes were harvested and analyzed by flow cytometry for expression of CD4, CD8, and TCR. We found that 4 out of 20 (20%) mutant lobes developed >1000 thymocytes. Furthermore, in 9 out of 20 lobes (45%), greater than 10% of thymocytes developed to the CD4+8+ stage (Fig. 4). In addition, a few lobes developed mature CD4+TCR+ thymocytes (see below).
Although T cell development was variable from experiment to experiment, in any given experiment wild-type thymus developed more cells and progressed farther developmentally than PU.1−/− thymus. Representative examples from a single experiment demonstrate the range of phenotypes observed from PU.1−/− thymuses relative to control, including high, intermediate, and low progressors, and nonprogressors (Fig. 5). High progressors had similar patterns of expression of CD4 and CD8 compared with wild-type littermate controls, but often showed higher percentages of CD4−8− and CD25+ thymocytes. Intermediate and low progressors were progressively more enriched for CD4−8− and CD25+ progenitors. In addition, even high progressors maintained a higher level of overall HSA and Thy-1 expression consistent with relative immaturity (Fig. 5). Among those PU.1−/− lobes that developed fewer CD4 or CD8 expressing cells, (intermediate and low progressors) the high percentages of CD25 indicates that development had proceeded beyond the T cell commitment stage of development. Nonprogressors failed to develop at all and few live thymocytes were recovered (data not shown). Of the 20 lobes shown in Fig. 4, 3 were classified as high progressors (CD4, CD8, TCR expression comparable with wild-type), 7 were intermediate progressors (greater than 5% CD4+8+ thymocytes), 5 were low progressors (<5% CD4+8+, but CD25 expression detected) and 5 did not develop (fewer than 100 cells recovered).
Although the PU.1−/− mice are interbred on the C57BL/6 and 129 backgrounds, the lobe-to-lobe variability in development could not be explained genetically since the two lobes from the same embryos were observed to express different phenotypes, for example, high and low progressors, or low and nonprogressors. Therefore, it appears that developmental progression in PU.1−/− thymus may be a randomly determined process.
In some experiments, the appearance of γδ TCR+ T cells was assessed. All three mutant lobes examined showing some T commitment also contained up to 11% CD4−8− γδ TCR-positive, CD4− thymocytes (data not shown).
To determine whether PU.1-deficient thymocytes were functionally competent, cultures of E14.5 thymic lobes were harvested after 14 days and tested for expression of CD4, CD8, TCR, and, subsequently, for proliferative responses to anti-CD3. By normalizing the number of wild-type and PU.1−/− thymocytes plated based on expression of TCR, we found that PU.1−/− thymocytes proliferated as well as wild-type thymocytes in response to anti-CD3ε cross-linking (Fig. 6). In addition, 48 h after stimulation, culture supernatants were harvested and assessed for IL-2 production in bioassays. IL-2-dependent proliferation of CTLL-2 cells (>10 times over background) was observed in response to both mutant and wild-type culture supernatants from stimulated, but not unstimulated cells. The proliferation was inhibited by neutralizing Abs against IL-2 (data not shown).
We have shown that the majority of thymocytes in PU.1−/− thymus have characteristics of early, precommitment T cell progenitors. These thymocytes (the majority population) show no aberrant expression of markers, and closely resemble a normal population enriched in E13.5 and smaller E14.5 wild-type thymocytes. Furthermore, at E14.5, the absolute number of thymocytes with this phenotype is approximately equal in both mutant and wild type. This suggests that the homing and survival of progenitors to the thymus occurs relatively normally in PU.1−/− animals. In vivo studies support this finding (32). It further argues that the requirement for PU.1 activity occurs when thymic progenitors begin their intrathymic commitment and/or differentiation.
The phenotype of the major PU.1−/− population, Thy-1− CD25− FcγRII/III− Sca I− NK1.1− CD4− CD8− (T-Lin−) CD44+ HSAbright c-kitlow, is consistent with previous identifications of the most immature thymic progenitors before commitment to the T lineage (12, 13, 31, 33, 34, 35, 36). Therefore, the major population of thymic progenitors present in PU.1−/− thymus is likely blocked at an early stage of development.
Our data further demonstrate that PU.1 is not absolutely required for fetal T cell development, since even as early as E14.5, rare Thy-1+, CD25+ and at E16.5, CD8+ thymocytes are detected. We found that committed T cell progenitors in PU.1-deficient embryos occurred as a low percentage of total thymocytes, and stochastically in rare thymic lobes, features which likely contributed to their lack of detection in earlier experiments. We further found that, when placed in FTOC, a significant fraction of PU.1−/− thymi developed beyond the point of T cell commitment and in some cases developed to maturity. These mature T cells were shown to proliferate and produce IL-2 in response to stimulation through the TCR. We have also detected γδ TCR expressing CD4− thymocytes in PU.1−/− FTOCs, demonstrating that this lineage also does not absolutely require PU.1 activity. NK1.1 expression was observed in PU.1−/− thymi, suggesting that thymic NK cell development may also occur. However, further experiments are necessary to determine whether mature, functional NK T cells are present in the PU.1 mutant, since the NK1.1 marker is present on a subset of immature progenitors. Furthermore, we also detect significant numbers of lymphoid dendritic cells in both fresh E16.5 thymus and after FTOC (see footnote 3). This demonstrates that PU.1 activity is not absolutely essential for T cell development and function.
PU.1−/− T cell development differed from wild-type in several major aspects. First, developmental progression, and stage of progression was variable in PU.1−/− thymuses. Two thymic lobes isolated from the same individual differed in their extent of development in FTOC, suggesting that differences in genetic background do not contribute to developmental variability. Interestingly, T cell development in neonatal PU.1 mutants (5) also showed considerable variability. This suggests that PU.1 may contribute to whatever process ensures consistent developmental timing of T cell development. Second, T cell development in PU.1-deficient thymus was delayed. Third, even when cultured PU.1−/− thymocytes progressed comparably to wild type, recovered cell numbers were still substantially lower. These results are most consistent with a model in which PU.1 function is most essential in an early progenitor, before the αβ T, γδ T, and thymic dendritic cell lineages split. For example, if PU.1 deficiency reduced the efficiency of T cell commitment, fewer progenitors would be able to differentiate, reducing cellular recovery. In addition, a lowered efficiency of T cell commitment could also explain the variability and delayed developmental kinetics seen in PU.1−/− thymus cultured in vitro.
T cell development may also provide direct evidence to support the overlapping functional nature of transcription factor families during development. We examined the expression of selected ets family transcription factors during thymocyte development. PU.1, Ets-1, and Fli-1 were chosen because gene targeting has demonstrated that all three play a role in the T lineage (7). Fli-1 was expressed at all stages of thymocyte development examined, both pre- and post-commitment. Its early expression pattern is consistent with its probable function in an early thymic progenitor, contributing to the reductions in thymic cellularity observed in Fli-1 gene disruption experiments (30). Since Fli-1 is also expressed later, it may also play a role in other aspects of T cell function. PU.1 was expressed in noncommitted thymocytes, but was down-regulated at the point when T cells become committed, consistent with its role in early progenitors prior to T cell commitment. In contrast, Ets-1 has a later expression pattern consistent with its importance for the survival and/or maturation of single-positive thymocytes and peripheral T cells, and in T cell activation, as shown by gene targeting (28, 29). The promoters for both PU.1 and Ets-1 have functionally important ets family-binding motifs that are thought to play a role in auto-regulation (7, 37, 38, 39). Perhaps low level gene expression of Ets-1 is initiated by PU.1, which then leads to a true auto-regulatory loop for high level expression. Thus, overlapping expression of transcription factor families may explain the “chicken or egg” paradox of initiating gene expression. Since T cell development can occur in the absence of PU.1, Ets-1, or Fli-1, these transcription factors are not individually required for T cell commitment. All, however, appear to be required for efficient T cell development. This suggests an incomplete functional redundancy among the ets family transcription factors expressed during T cell development. The production of double and triple mutants for transcription factors within the T lineage will be required to address this issue.
Lisa M. Spain thanks D. Izon and A. Caton for comments on the manuscript.
This work was supported by National Institutes of Health Grant P01 DK53558 (E.W.S. and L.M.S., subproject leaders). A.G. was partially supported by National Institutes of Health Grant HL58716 (to E.W.S.); S.K. and L.M.S. were partially supported by National Institutes of Health Grant AI36453 and National Aeronautics and Space Administration Grant NAG9-832 (to L.M.S.). E.W.S. is a Leukemia Society of America Scholar.
A. Guerriero, L. M. Spain, and E. W. Scott. PU.1 is required for myeloid-derived but not lymphoid-derived dendritic cells. Submitted for publication.
Abbreviations used in this paper: FTOC, fetal thymic organ culture; HPRT, hypoxanthine phosphoribosyltransferase.