E2A transcription factors limit expression of Gata3 to facilitate T lymphocyte lineage commitment

T-lymphocyte specification and commitment are associated with dynamic alterations in the transcriptome which are regulated by the complement of expressed and functional transcription factors.¹  Although numerous transcription factors are essential for T-cell development, the transcriptional networks in which they operate and their essential targets are not well understood.

One critical regulator of T-lymphocyte specification is NOTCH1, a transmembrane receptor and transcriptional coactivator. The intracellular domain of NOTCH1 (ICN1), which translocates to the nucleus and regulates gene expression,²  activates multiple transcription factors in thymocytes, including Notch1, Hes1, Tcf7, and Gata3. TCF7 (TCF1) induces a subset of NOTCH1 target genes, including Gata3, and limits alternative fates from multipotent progenitors (MPPs) in vitro in the absence of NOTCH1,³  although its role as a T-cell commitment factor in vivo is disputed.⁴  Although GATA3 is essential for T-cell specification, high concentrations of GATA3 block T-lymphocyte differentiation in vitro at the DN2 stage and prevent the development of T lymphocyte–committed DN3 cells.⁵  Moreover, overproduction of GATA3 reprograms DN2 cells into mast cells, but this reprogramming is restrained by NOTCH1.⁶  Thus, GATA3 expression should be regulated to promote T-cell specification and commitment while not achieving sufficient expression to block T-cell development or divert cells to an alternative fate. However, it remains controversial whether GATA3 can achieve the concentration necessary to block T-cell development under physiologic conditions and whether mechanisms exist to limit Gata3 during T-cell development.

The E2A transcription factors are required for proper T lymphopoiesis but, unlike NOTCH1 and GATA3, they are essential at earlier stages of lymphopoiesis.⁷  Mice lacking E2A have reduced numbers of lymphoid-primed multipotent progenitors and common lymphoid progenitors; therefore, they should have fewer thymus-seeding cells.⁷  In addition, E2A^−/− lymphoid-primed multipotent progenitors fail to express CCR9, a chemokine receptor involved in thymic homing,⁸  and have reduced expression of Notch1.⁹  Thus, E2A^−/− bone marrow cells should be compromised in their ability to seed the thymus. In vitro, E2A^−/− hematopoietic progenitors fail to differentiate into T lymphocytes unless they are forced to express ICN1, suggesting that their reduced expression of Notch1 prevents T-cell differentiation.¹⁰  Nonetheless, previous studies showed that the earliest T-cell population (DN1) is present in the thymus of E2A^−/− mice, although differentiation appears to be partially arrested just before the DN2 stage (DN1.5).^11,12  These older studies used staining strategies that are now known to be insufficient to identify the earliest thymic progenitors (ETPs) within the heterogeneous DN1 compartment.^13,14  Therefore, a requirement for E2A at the earliest stages of T lymphocyte lineage specification or commitment in the thymus has not been demonstrated.

Here, we demonstrated that E2A was required for the efficient differentiation of DN2 cells into T lymphocyte lineage-restricted DN3 cells. We found that E2A played a critical role in limiting Gata3 expression specifically in DN2 cells. With the use of both Gata3 small interfering RNA (siRNA) and heterozygous deletion of Gata3, we demonstrated that GATA3 was responsible for the arrested differentiation of E2A^−/− DN2 cells. Moreover, we demonstrated that the heightened expression of GATA3 caused an aberrant proliferation of DN2 and DN3 thymocytes in vivo. In the absence of E2A, DN2 cells have an increased propensity to adopt non–T-cell fates that was not influenced by Gata3 dose. However, reducing Gata3 resulted in an overall reduction in the number of non–T-cell progenitors in vivo. We propose a model in which the overproduction of GATA3 extends the self-renewal capacity of DN2 cells and inhibits T lineage restriction, whereas E2A deficiency, independent of the overproduction of GATA3, allows DN2 cells to access alternative fates. Thus, T-lymphocyte commitment requires passage through a checkpoint in DN2 cells in which positive and negative inputs to Gata3 must be balanced to allow further differentiation.

This work was approved by the Institutional Animal Care and Use Committee at the University of Chicago under protocol ACUP 71 119. Mice were housed at the University of Chicago Animal Resource Center. Genotyping for the E2A and Gata3 alleles was performed as described.^15,16  Tie2Cre transgenic mice were genotyped by PCR. Adult mice were analyzed at 6-8 weeks of age.

The antibody clones and fluorochrome combinations used in this study are available on request. All antibodies were purchased from eBiosciences or BD Biosciences.

The lineage cocktail for adult thymocytes included CD3ϵ, CD8α, TCRβ, TCRγδ, NK1.1, CD11c, Ter-119, CD11b, Gr1, B220, and CD19 and for fetal liver (FL) included Ter119 and Gr1. For in vitro culture of retrovirus-transduced MPPs, cells were sorted 48 hours after transduction as GR1⁻Ter119⁻CD117⁺CD27⁺GFP⁺ cells. Cells were examined on a FACSCanto or FACSAria and analyzed with FlowJo Version 9.5 (TreeStar). For sorting, thymocytes were depleted of Lin⁺ cells on a magnetic column (Miltenyi Biotec) before staining for sorting.

Mice were given an intraperitoneal injection of 1 mg of BrdU per 6 g of body weight (BD Biosciences) 24 hours and 12 hours before the analysis. BrdU staining was performed with the FITC BrdU Flow Kit (BD Biosciences).

OP9-DL1 stromal cells were maintained in OPTI-MEM and plated 1 day before use to achieve a near confluent monolayer of cells. Cells were cultured on OP9-DL1 cells in the presence of 5 ng/mL Flt3 ligand, IL-7, and CD117 ligand. The retroviral vectors and transduction procedures were described previously.¹⁷ 

RNA was extracted with the use of the RNeasy Micro Kit (QIAGEN) and reverse transcribed with SuperscriptIII reverse transcriptase (Invitrogen) with the use of random hexamers. Quantitative PCR (qPCR) was performed in triplicate with the iQ SYBR Green Supermix (Bio-Rad) and detected by the MyiQ Single color Real-Time PCR System (Bio-Rad). Expression levels were calculated for each gene relative to Hprt with the use of the ΔC_T method. The primers used in this study are available by request. Microarray procedures (GEO accession no. GSE43224) were as described previously.⁹ 

A rigorous analysis of E2A^−/− thymocytes by flow cytometry showed that E2A was required for normal ETP, DN2, and DN3 thymocyte numbers in vivo (supplemental Figure 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). The previously defined “DN1.5” population identified among E2A^−/− Lin⁻ thymocytes^11,12  was not detected when ETPs/DN2 cells were identified with CD117. However, a Lin⁻CD25^loCD117^lo population was present that included CD44^lo cells that lacked IL7Rα and CD122, and CD44^hi cells that expressed IL7Rα but not CD122. In vitro culture of Lin⁻CD25^loCD117^lo subsets showed that they included progenitors of non–T-cell lineages, including natural killer (NK) cells.

On the basis of these observations, we hypothesized that E2A was necessary to bias ETPs/DN2s toward the T-lymphocyte fate. To test this hypothesis, we compared the ability of cells from E2A^+/+, E2A^+/−, and E2A^−/− mice to generate Lin⁻CD44⁻CD25⁺ cells (DN3-like cells) in vitro. When cultured under T cell–supportive conditions the progeny of wild-type ETPs and DN2s were predominantly DN3-like, and only a few cells expressed the NK-cell marker NK1.1 (Figure 1A-B). Further, wild-type (WT) ETPs and DN2s underwent an approximate 10 000-fold increase in numbers (not shown). In contrast, E2A^−/− ETPs and DN2s expanded poorly and generated few DN3-like cells (Figure 1A,C). Most E2A^−/− cells remained CD44^lo/+CD25^lo, and a substantial portion of these cells expressed NK1.1 (Figure 1A-B). E2A^+/− DN2s, and to a lesser degree ETPs, also generated fewer DN3-like cells, indicating a dose-dependent requirement for E2A (Figure 1A-C).

We next tested the frequency of DN2s with T-lineage potential with the use of single-cell analysis. A reduced frequency of E2A^−/− DN2 cells gave rise to DN3-like cells compared with WT DN2s, although the cloning efficiencies were identical (Figure 1D). Further, single E2A^−/− DN2s produced more non–T cells (mostly NK1.1⁺) than DN3-like cells, indicating that in the absence of E2A these cells were not efficient T-lymphocyte progenitors (Figure 1D-E). Taken together, our data indicated that E2A^−/− and E2A^+/− ETPs and DN2s had a diminished capacity to differentiate into committed T lymphocytes but had increased potential for alternative lineages such as NK cells. Our findings are consistent with a previous study showing that deletion of E2A and Heb in early T-lineage cells arrests differentiation at the DN2 stage in vitro.¹⁸ 

A plausible explanation for the reduced T-cell potential of E2A^−/− DN2 cells is that they expressed less Notch1.¹⁰  qPCR analysis of E2A^−/− DN thymocytes showed that Notch1 mRNA was decreased by only 50% compared with WT (Figure 2A). Given that Notch1^+/− mice can generate DN3-like cells¹⁹  and that E2A^−/− thymocytes express the Notch target gene Deltex1 (Figure 2A), we concluded that reduced Notch signaling did not explain the decreased ability of E2A^−/− DN2s to generate DN3-like cells.

To gain insight into the mechanism underlying the loss of T-cell potential in E2A^−/− DN2s, we performed a global analysis of gene expression. Remarkably, only 5.87% of the total DN2 cell transcriptome was altered by the loss of E2A, and most of the differentially expressed genes were increased in E2A^−/− compared with E2A^+/+ DN2 cells (Figure 2B-C). Multiple T cell–associated genes were decreased in E2A^−/− DN2s, whereas NK cell-associated genes were increased (Figure 2D-E; supplemental Table 1). The altered expression of some of these genes may be a direct consequence of the loss of E2A (eg, Gfi1 and Cd3e).²⁰  However, some genes may be differentially expressed because E2A^−/− DN2 cells initiated differentiation toward alternative cell fates. Three genes were of particular interest, Bcl11b, Gfi1, and Gata3 (Figure 2D-E). Preliminary screens showed that Bcl11b expression was not affected by E2A deficiency in MPPs cultured under T-cell conditions. Moreover, ectopic expression of BCL11b did not rescue T-cell differentiation from E2A^−/− MPPs in vitro (not shown). Gfi1 produces the transcriptional repressor GFI1 that is highly related to GFI1b, and both proteins were reported to be differentially expressed in E2A^−/− T-cell lymphomas, and they repress Gata3.^17,21  However, although ectopic expression of GFI1 or GFI1b promoted differentiation of E2A^−/− MPPs to the DN2 stage in vitro, DN3 cells failed to develop, and GFI1-deficiency, unlike E2A-deficiency, did not compromise T-cell differentiation from MPPs in vitro (supplemental Figure 2). Therefore, we focused on GATA3 as the probable mediator of arrested T-cell development in E2A^−/− mice.⁶ 

To confirm the differential expression of Gata3, we performed qPCR analysis on highly purified Lin⁻ thymocytes from E2A^+/+, E2A^+/−, and E2A^−/− mice. Gata3 transcripts were increased selectively in the DN2s, but not in ETPs or DN3 cells, of E2A^+/− or E2A^−/− mice (Figure 3A). The elevated Gata3 mRNA was not a consequence of altered Gata3 promoter usage because most mRNA initiated at the Gata3-1b promoter, as is seen in WT thymocytes (data not shown).²²  Moreover, in E2A^−/− MPPs cultured in vitro, Gata3 mRNA remained Notch-signaling dependent and increased progressively between days 4 and 8 (Figure 3B). GATA3 protein was elevated in E2A^−/− DN2s and DN4s, and at varying amounts in DN3 thymocytes as well as in E2A^−/− MPPs cultured in vitro for 10 days (Figure 3C-D). However, E2A^−/− NK1.1⁺ cells did not express high levels of GATA3, indicating that deregulation of GATA3 was restricted to cells with T-cell potential. Taken together, our data indicate that E2A was required to limit Gata3 during commitment to the T-cell fate.

We hypothesized that the dysregulation of Gata3 in E2A^+/− and E2A^−/− DN2 cells underlies their reduced DN3 cell potential. If this were the case, then lowering Gata3 expression should relieve E2A^−/− cells from this developmental block. To test this hypothesis, we transduced E2A^+/+ and E2A^−/− MPPs with a retrovirus that produced Gata3 siRNA^17,23  and assessed the ability of these cells to differentiate into DN3-like cells. When expressed in WT MPPs, the Gata3 siRNA induced a dose-dependent arrest at the DN2 stage (supplemental Figure 3; Figure 4A-B). In contrast, when Gata3 siRNA was expressed in E2A^−/− MPPs, CD25⁺ cells developed and DN3-like cells were evident, whereas the control retrovirus failed to promote the development of these cells (supplemental Figure 3; Figure 4A-C). E2A^−/− DN3 stage cells were enriched in the GFP^low population, whereas GFP^high cells were arrested at the DN2 stage, indicating that low expression of Gata3 siRNA facilitates differentiation of E2A^−/− MPPs to the DN3 stage, whereas high expression of Gata3 siRNA inhibits T-cell differentiation. The DN1 and CD25⁺ cells in E2A^−/−Gata3 siRNA-expressing cultures had less Gata3 and Id2 mRNA compared with E2A^−/− DN1 or CD25⁺ cells that expressed the control retrovirus (Figure 4D). These cells also expressed the T cell–associated genes Ptcra and Gfi1, indicating that they were specified to the T-cell fate (Figure 4D). Moreover, DN3-like cells in E2A^−/−Gata3 siRNA-expressing cultures expressed Notch1 and Rag1 mRNA, indicating that they are T-lineage cells (supplemental Figure 3). Therefore, Gata3 siRNA was sufficient to allow some E2A^−/− MPPs to reach a DN3-like stage.

Our data indicated that in the absence of E2A excess GATA3 caused a block in T-cell differentiation. However, the Gata3 siRNA was efficient at reducing Gata3 mRNA and reduced the total number of cells generated from WT and E2A^−/− MPPs under T-cell conditions. To confirm that the excess GATA3 was responsible for this differentiation block, we generated mice deficient in E2A and heterozygous for a mutated allele of Gata3 in hematopoietic cells (Gata3^+/Δ).²⁴  We reasoned that this decrease in Gata3 may be sufficient to at least partially rescue the E2A^−/− DN2 stage arrest. We first examined the effect of the Gata3^+/Δ genotype on E2A^−/− MPPs cultured in vitro. Remarkably, although few E2A^−/−Gata3^fl/+ MPPs acquired a DN3-like phenotype, as many as 15% of the cells derived from E2A^−/−Gata3^+/Δ MPPs were DN3-like on day 10 of culture (Figure 5A-B). As predicted, Gata3 mRNA was decreased by approximately 50% in E2A^−/−Gata3^+/Δ cells compared with E2A^−/− DN2-like cells (Figure 5C). Further, the E2A^−/−Gata3^+/Δ DN3-like cells underwent T cell–lineage specification as indicated by the increased expression of Notch1 and Rag1 mRNA (Figure 5D). These data indicated that inactivation of 1 Gata3 allele was sufficient to augment T cell–lineage commitment from E2A^−/− MPPs during in vitro culture.

We next analyzed T-cell development in the thymus of E2A^−/−Gata3^+/Δ mice. Thymocyte populations in Gata3^+/Δ mice were indistinguishable from WT, and there was no selection against the deleted allele (supplemental Figure 4). In accordance with previous reports,^11,25 E2A^−/− mice had a decreased percentage of double-positive (DP) thymocytes and an increased percentage of single-positive (SP) cells (Figure 6A-B). In contrast, the percentage of DP and SP cells among E2A^−/−Gata3^+/Δ thymocytes was indistinguishable from that of WT mice (Figure 6A-B). The number of DP thymocytes was only slightly increased by heterozygous deletion of Gata3, suggesting that GATA3 influenced differentiation but did not augment thymocyte expansion in the absence of E2A (Figure 6C). As expected, the number of ETP and DN2 cells in E2A^−/−Gata3^+/Δ mice was similar to that in E2A^−/− mice (data not shown). However, E2A^−/−Gata3^+/Δ mice had an increased frequency and number of DN3 cells compared with E2A^−/− mice (Figure 6D-F). More dramatically, the frequency and number of Lin⁻CD117^lo/−CD25^lo (and CD44^hi, not shown) cells in E2A^−/−Gata3^+/Δ mice was lower than in E2A^−/− mice, indicating that the reduced expression of Gata3 prevented the development of these cells while simultaneously augmenting DN3 cell differentiation in E2A^−/− mice (Figure 6G-H). Given that Lin⁻ numbers were decreased in E2A^−/−Gata3^+/Δ mice compared with E2A^−/− mice (data not shown), the increased number of DP and DN3 cells was striking, and the increased frequency of these populations is probably a better measure of differentiation potential than number compared with E2A^−/− mice. The restoration of DP, SP, and DN3 cell frequencies and the partial rescue of DN3 cell numbers combined with the decreased frequency and number of Lin⁻CD117^lo/−CD44^hiCD25^lo NK cell progenitors in E2A^−/−Gata3^+/Δ mice supports the hypothesis that elevated expression of Gata3 in E2A^−/− DN2 and/or DN4 cells perturbs T-cell development in vivo.

One plausible explanation for the increased number of DN3 thymocytes in E2A^−/−Gata3^+/Δ mice is that these cells proliferated more than E2A^−/− cells. Previous studies showed that E2A^−/− or E2A/HeB double-deficient DN3 cells incorporate more BrdU than WT DN3 cells, suggesting that these cells proliferate more rapidly than their WT counterparts.^18,26  We found that E2A^−/− DN2s and DN3s incorporated more BrdU in vivo than WT cells (Figure 7A). In contrast, BrdU incorporation into E2A^−/−Gata3^+/Δ DN2s and DN3s was comparable with WT (Figure 7A). Therefore, heterozygous deletion of Gata3 limited proliferation of E2A^−/− DN2 and DN3 thymocytes. These observations are consistent with the hypothesis that heterozygous deletion of Gata3 restored DN2 cell differentiation rather than promoting the expansion of DN3 cells.

To determine whether the increased number of E2A^−/−Gata3^+/Δ DN3 cells was the consequence of restored T cell–lineage differentiation or survival/accumulation of DN3 cells, we quantified the frequency of DN2 cells capable of generating DN3-like cells. As expected, the frequency of E2A^−/− DN2 cells that gave rise to DN3-like cells was lower than from control DN2 cells (Figure 7B). Moreover, the E2A^−/− DN2 cells that produced DN3-like cells generated only a low percentage of DN3 seconds and produced a large population of CD44⁺CD25^lo cells (Figure 7C-D). In contrast, the frequency of E2A^−/−Gata3^+/Δ DN2 cells that progressed to the DN3-like stage was similar to WT cells (Figure 7A-C). Moreover, individual E2A^−/−Gata3^+/Δ DN2s generated more DN3-like cells in any given well than E2A^−/− DN2s (Figure 7D). Nonetheless, similar to E2A^−/− DN2s, E2A^−/−Gata3^+/Δ DN2s consistently gave rise to cells that also adopted the alternative fates (Figure 7C). Taken together, our data indicated that heterozygous deletion of Gata3 augmented the ability of E2A^−/− DN2 cells to generate T cell–committed DN3 cells without affecting the frequency of cells that give rise to NK cells.

We identified a novel role for the E2A transcription factors in controlling the T cell–differentiation potential of DN2 thymocytes. E2A impeded NK-cell differentiation from DN2 thymocytes and limited the expression of GATA3, thereby restraining DN2 cell proliferation and promoting T-cell differentiation. Our study is the first to identify a physiologic mechanism to limit GATA3 at this stage of development and to implicate GATA3 in the excessive proliferation of E2A^−/− DN thymocytes. The ability of GATA3 to derail T-cell development indicates a critical difference between the regulatory networks that drive B-lymphocyte and T-lymphocyte commitment. In B lymphocytes, E2A initiates a self-sustaining feed-forward pathway in which each transcription factor activates the downstream component, which then feeds back to enforce the expression of the upstream factor while cooperatively activating B cell–specific genes and promoting the repression of non–B-cell genes. In contrast, T lymphocyte lineage commitment involves both feed-forward (induction of Gata3 by NOTCH1) and repressive mechanisms (restriction of Gata3 by E2A) that ensure the appropriate expression of a transcription factor that can promote T-cell development only within a narrow concentration range.

A previous study implicated Notch1 as the essential E2A target whose deregulated expression prevents T-cell differentiation in E2A^−/− mice.¹⁰  Nonetheless, we found that Notch1 mRNA was expressed at sufficient levels in E2A^−/− thymocytes to promote expression of Deltex1 and Gata3. Therefore, Notch signaling is functional in E2A^−/− DN2 cells and sufficient to support T-cell differentiation. Notch1 is a target of Notch signaling; therefore, although E2A^−/− thymus-seeding cells express low levels of Notch1,⁹  sufficient Notch-signaling may occur in the thymus to up-regulate Notch1 mRNA. NOTCH1 can restrain the ability of high GATA3-expressing DN2s to adopt the mast cell fate,⁶  and E2A^−/− DN2s did not generate mast cells in vitro, suggesting that the slight decrease in Notch1 may have had no consequence. However, these cells may have failed to adopt the mast cell fate because E2A:SCL heterodimers are necessary for mast cell differentiation.^6,27  In contrast, E2A^−/− DN2s readily generated NK cells when cultured on OP9-DL1, suggesting that the level of Notch signaling was insufficient to restrain NK-cell differentiation or, alternatively, that E2A was necessary to prevent NK-cell differentiation even in the presence of Notch. On the basis of the known requirement for ID2 in NK-cell differentiation²⁸  and previous studies that show the dose responsiveness of NK-cell inhibition by NOTCH1,^10,29  we favor the latter explanation.

We conclude that the overproduction of GATA3 in E2A^−/− DN2s antagonized T-cell development and caused an increase in NK-cell progenitors in vivo even though E2A deficiency was the primary mechanism that allowed for NK-cell development. The hypothesis that the overproduction of GATA3 underlies the arrested T-cell development in E2A^−/− progenitors is supported by the increased frequency of DN3 cells generated in vitro from E2A^−/− MPPs after knocking down Gata3 with an siRNA and after heterozygous deletion of Gata3 in E2A^−/− hematopoietic cells. Although heterozygous deletion of Gata3 had little effect on the frequency of E2A^−/− DN2 cells that generated alternative fates, in vivo the frequency and number of Lin⁻CD117^loCD44^hiCD25^lo cells was reduced. We propose that GATA3 influences the number of cell divisions a DN2 cell undergoes before making the choice between T-cell differentiation and alternative fates. Thus, reducing GATA3 decreased DN2 cell “self-renewal,” augmented T-cell differentiation, and resulted in fewer Lin⁻CD117^loCD44^hiCD25^lo cells in the thymus. We propose that in vitro E2A^−/−Gata3^+/Δ DN2 cells retained the potential for NK-cell/alternative fate differentiation but made a cell fate choice after fewer divisions. If this model is correct, then E2A^−/−Gata3^+/Δ DN2 cells must have undergone some, although limited, “self-renewal” in vitro before T-cell commitment such that most wells contained both DN3 and NK cells. A role for GATA3 in augmenting self-renewal is consistent with the ability of ectopic GATA3 to promote T-cell transformation in transgenic mice.^30,31 

In vivo, E2A^−/−Gata3^+/Δ DN3 cells did not recover to the number of WT DN3 cells, although the ratio of Lin⁻CD117^lo/−CD44⁺CD25^lo to DN3 cells changed dramatically. The multiple functions of E2A in DN3 cells probably contributed to this observation. E2A is required for proper rearrangement and expression of TCRβ³²  and probably controls the expression of multiple DN3 cell genes.²⁰  In addition, E2A^−/− DN3 cells are able to pass β-selection without expressing a functional pre-TCR and may rapidly differentiate to the DP stage.^33,34  Another potential explanation for the failure of E2A^−/−Gata3^+/Δ DN3 cells to recover to WT numbers is that heterozygous deletion of Gata3 did not reduce GATA3 sufficiently to fully restore DN3 numbers. It is possible that deletion of 1 allele of Gata3 resulted in a < 50% reduction in GATA3 because GATA3 stability can be affected by interactions with other transcription factors^35,36  and GATA3 can be regulated translationally.³⁷  In vitro we found that heterozygous deletion of Gata3 was sufficient to restore the frequency of E2A^−/− DN2 cells that could differentiate to the DN3 stage. Therefore, a 50% reduction in Gata3 alleles is sufficient to at least partially alleviate the differentiation block in E2A^−/− thymocytes.

The question of how E2A limits Gata3 expression warrants further investigation. Gata3 may be a direct target of E2A because E2A binds near the Gata3 gene in T cells.^20,38  E proteins can function as transcriptional repressors when associated with the class II bHLH protein Tal1/SCL, which is expressed in DN2 cells.^39,40  However, direct regulation of Gata3 by E2A:SCL is unlikely because transgenic expression of SCL in T cells phenocopies E2A deficiency, implying that SCL inhibits E2A function.^41,42  Another possibility is that E2A target genes such as Gfi1 and Gfi1b function as repressors of Gata3.^17,21  However, Gfi1^−/− MPPs generated DN3 cells in vitro, showing that these cells do not phenocopy E2A^−/− MPPs. Therefore, although GFI1 and/or GFI1B may play a role in limiting Gata3 expression, GFI1 alone is not essential. We hypothesize that multiple targets of E2A participate in restricting Gata3 expression in DN2 cells. An analysis of the cis-regulatory elements that control Gata3 will be necessary to determine how Gata3 is controlled during T-cell development.

The mechanisms by which GATA3 arrests T-cell development have not been elucidated. A previous study suggested that GATA3 induces the expression of the Notch signaling antagonist DELTEX1, which can reduce Notch signaling and block T-cell development.⁴³  However, E2A^−/− MPPs cultured on OP9-DL1 had lower expression of Deltex1 mRNA than WT cells.¹⁰  In addition, Deltex1 mRNA was expressed at near WT levels in E2A^−/− DN2 or DN3 cells. We did observe high expression of Deltex1 and Id2 in E2A^−/− Lin⁻CD117^loCD44⁺CD25^lo thymocytes, which differentiated into NK cells in vitro. These findings suggest that this population may have confounded the previous studies. Nonetheless, removing Deltex1 from Id1 transgenic mice, which have reduced E protein activity, had a modest effect on early T-cell development, consistent with previous data showing that an increase in Notch signaling in E2A^−/− cells can augment T-cell development.^10,43  However, DELTEX1 does not appear to be a main contributor to GATA3-induced developmental arrest in E2A^−/− DN2 cells.

It remains to be determined whether high concentrations of GATA3 target the same genes (altering their level of expression) or a unique set of genes (resulting in ectopic expression) from those regulated by lower concentrations of GATA3. Ectopic expression of GATA3 alters the expression of multiple genes, including Il7ra, Sfpi1, Ptcra, and Rag1, but GATA3 only binds directly to Rag1 in T-lymphocyte progenitors.^5,6,40,44  Interestingly, a comparison of the genes that are differentially expressed in E2A^−/− DN2s with genes that are bound by GATA3 in DN1, DN2, or DP thymocytes showed that only a few E2A-dependent DN2 genes were bound by GATA3, and these were largely T cell–associated genes, including Rag1, Bcl11b, Gfi1, and Cd3e.⁴⁰  In contrast, a larger portion of the genes that increased in E2A^−/− DN2 cells were bound by GATA3, including Lgals3, Rora, Sema4f, Ccr6, Icos, Id2, Zbtb16, Il4, Itgb2, and Itgb3.⁴⁰  These observations raise the possibility that GATA3 activates genes in E2A^−/− cells and that lower concentrations of GATA3 may fail to activate these genes or even repress them. This hypothesis is consistent with previous studies that reported that GATA3 functions are concentration dependent and may be influenced by interactions with other transcription factors. For example, high concentrations of GATA3 antagonize PU.1 function in DN2s, whereas low concentrations of GATA3 synergize with PU.1 to produce a gene expression signature that allows for T-cell lineage commitment.^6,45,46  Additional studies will be necessary to determine how the elevation of GATA3 influences the differentiation and commitment of T-lymphocyte progenitors.

The authors thank Dr Avinash Bhandoola for assistance with identification of ETPs and helpful suggestions; Dr Jose Alberola-Ila for the Banshee-Gata3 siRNA vector; Dr Ivan Moskovitz for the Tie2-Cre mice; Drs I-Cheng Ho and Sun-Yun Pai for the Gata3^+/fl mice; Drs Sheila Dias and Mary Yui for help with the GATA3 staining; Dr Renee de Pooter for comments on the manuscript; and members of the Kee laboratory for helpful discussions.

This work was supported by grants from the National Institutes of Health (R01 CA099978, R01 AI079213, and R21 AI096530). B.L.K. is supported by a Leukemia & Lymphoma Society Scholar Award. S.M. was supported by the University of Chicago Medical Scientist Training Program T32 GM07281, and K.R. was supported by T32 GM38663 and R01 CA099978-S1.

National Institutes of Health

Contribution: W.X. designed and performed experiments and helped write the manuscript; T.C. and S.M. contributed to some experiments; K.R. analyzed microarray data; M.S. performed microarray analysis; and B.L.K. designed experiments, wrote the manuscript, and obtained funding for the project.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Barbara L. Kee, Department of Pathology, The University of Chicago, 5841 S Maryland Ave, MC1089, Chicago, IL 60637; e-mail: bkee@bsd.uchicago.edu.