Bcl11b represses a mature T-cell gene expression program in immature CD4⁺CD8⁺ thymocytes

Abstract

Bcl11b is a transcription factor that, within the hematopoietic system, is expressed specifically in T cells. Although Bcl11b is required for T-cell differentiation in newborn Bcl11b-null mice, and for positive selection in the adult thymus of mice bearing a T-cell-targeted deletion, the gene network regulated by Bcl11b in T cells is unclear. We report herein that Bcl11b is a bifunctional transcriptional regulator, which is required for the correct expression of approximately 1000 genes in CD4⁺CD8⁺CD3^lo double-positive (DP) thymocytes. Bcl11b-deficient DP cells displayed a gene expression program associated with mature CD4⁺CD8⁻ and CD4⁻ CD8⁺ single-positive (SP) thymocytes, including upregulation of key transcriptional regulators, such as Zbtb7b and Runx3. Bcl11b interacted with regulatory regions of many dysregulated genes, suggesting a direct role in the transcriptional regulation of these genes. However, inappropriate expression of lineage-associated genes did not result in enhanced differentiation, as deletion of Bcl11b in DP cells prevented development of SP thymocytes, and that of canonical NKT cells. These data establish Bcl11b as a crucial transcriptional regulator in thymocytes, in which Bcl11b functions to prevent the premature expression of genes fundamental to the SP and NKT cell differentiation programs.

Introduction

T-cell differentiation is a complex and dynamic process that leads to the production of functionally distinct populations within the thymus – γδ and αβ T-cell subsets, the latter of which include helper CD4⁺ T cells, cytotoxic CD8⁺ T cells, Treg cells, and NKT cells. Hematopoietic progenitor cells enter the thymus as CD4⁻CD8⁻ double-negative (DN) cells and proceed through successive steps of maturation. DN thymocytes are further divided into at least four developmental stages based on the differential expression of CD44 and CD25: CD44⁺CD25⁻ (DN1), CD44⁺CD25⁺ (DN2), CD44⁻CD25⁺ (DN3), and CD44⁻CD25⁻ (DN4). γδ T cells differentiate from DN3 thymocytes, following rearrangement of the β, γ, and δ TCR chains. αβ T cells develop from DN4 thymocytes that further differentiate into CD4⁺CD8⁺ double-positive (DP) CD3^loαβTCR^lo thymocytes. Positive selection events between the TCR expressed by DP cells and MHC molecules expressed by thymic stromal cells lead to the appearance of mature CD4⁺ and CD8⁺ single-positive (SP) CD3^hi/TCR^hi thymocytes, and NKT cells, all presumably resulting from large-scale changes in gene expression programs.

Transcription factors essential for the αβ T-cell developmental programs have been identified [1–3]. In particular, Zbtb7b (also known as ThPok) is required for CD4⁺ T-cell differentiation [4, 5]. Zbtb7b is not expressed in DP thymocytes, but is activated downstream of TCR signaling by TOX [6, 7] and GATA3 [8, 9], the latter of which appears to function with Zbtb7b in a positive, self-reinforcing loop that is dependent on the duration and intensity of the TCR signal [10–12]. Zbtb7b is believed to function primarily as an enforcement factor to lock down the CD4⁺ phenotype by repressing CD8⁺ T-cell-associated genes [13–16]. Runx3, which has been implicated in CD8⁺ T-cell differentiation [17], also appears to be induced by TCR signaling in DP cells, and is required for the silencing of CD4 and Zbtb7b genes in CD8⁺ SP thymocytes [18, 19].

Bcl11b (also known as Ctip2) is highly and specifically expressed within T cells, and to a lesser extent in NK cells [20], suggesting that Bcl11b could function as a T-cell-specific regulator. Bcl11b has been shown to bind to GC-rich target sequences, and is involved mostly in gene repression [21–23]. It recruits the class III histone deacetylase SIRT1 [22] and/or the class I histone deacetylases to promoters [23, 24]. Genetic analyses have shown that Bcl11b is crucial at several stages of T-cell development. Germline deletion of Bcl11b results in a complete block of T-cell differentiation at the DN stage, associated with impaired TCRβ rearrangement [25]. Bcl11b inactivation at the DP stage strongly blocks the maturation of DP thymocytes into SP cells and impairs positive selection, possibly through defective TCR signaling [26]. Here, we further investigated Bcl11b function in T cells by generating new T-cell-specific deletions of this gene.

Results

Bcl11b is essential for the early steps of αβ T-cell differentiation

We previously generated a germline deletion of exon 4 of the Bcl11b locus, Bcl11b^L−/L− [27], which is lethal just after birth [27]. These mice exhibited a tenfold decrease in thymic cellularity (0.9±0.2 × 10⁶ cells for Bcl11b^L−/L− versus 9.3±2.3 × 10⁶ cells for Bcl11b^L−/+ or Bcl11b^+/+ mice). The majority of Bcl11b^L−/L− thymocytes were large cells lacking CD4 and CD8 expression, whereas a smaller proportion expressed CD8 (Supporting Information Fig. 1A). Bcl11b^L−/L− thymocytes lacked αβTCR but most expressed γδTCR, including those expressing CD8 ( Supporting Information Fig. 1A, and data not shown).

To circumvent the perinatal lethality and to analyze the role of Bcl11b in adult T cells, we combined the floxed Bcl11b alleles (Bcl11b^L2/L2) with a transgenic allele expressing Cre recombinase under the transcriptional control of the Lck promoter, which initiates T-cell-specific expression in DN2 and DN3 cells [28]. Bcl11b^L2/L2Lck^cre/+ mice appeared healthy and indistinguishable from littermates and were analyzed at 6 wk of age. The thymuses from these mice were very small and contained low numbers of thymocytes (an average of 3 × 10⁵ cells; control littermates had an average of >10⁸ cells). T cells from Lck-Cre-deleted mice exhibited a phenotype reminiscent of that found in null newborn mice: most cells were large DN (48%) or CD8⁺ (30%) cells, and few DP cells (10%) were detected (Supporting Information Fig. 1B). In addition, as was observed in Bcl11b^L−/L− newborns, a large proportion of cells, including most CD8⁺ cells, expressed γδTCR ( Supporting Information Fig. 1B; 46% of total thymocytes on average). Although these γδTCR⁺ cells were present in absolute numbers similar to WT, the phenotype of these cells was clearly abnormal, as CD8-expressing TCRγδ⁺ cells were not detected in control mice (Supporting Information Figs. 1B and 2). These data confirm that Bcl11b acts early in T cells to promote differentiation toward the αβ lineage.

Block of CD4 and CD8 SP differentiation in CD4-Cre-deleted mice

To determine if Bcl11b plays a role at later stages of T-cell maturation, we crossed the floxed mice with transgenic mice expressing Cre recombinase under the control of the CD4 promoter [29]. Bcl11b^L2/L2CD4^cre/+ (Bcl11b^dp−/− hereafter) mice were also viable, fertile, and lived well into adulthood. PCR analysis of mice heterozygous for the Bcl11b mutation showed that deletion of the floxed sequences was initiated at the DN3 stage and completed in DP cells (Fig. 1A), with low amounts of Bcl11b protein in mutant DP cells (Fig. 1B compare lanes 1 and 4). Bcl11b was undetectable in more mature, mutant SP populations. Thus, CD4-Cre-mediated deletion leads to a profound reduction in Bcl11b protein levels at the DP stage. However, the presence of residual Bcl11b protein suggests that Bcl11b function may not be completely abrogated in all DP cells.

Figure 1.

CD4-Cre-mediated excision of Bcl11b exon 4 in T-cell populations. (A) Genomic PCR analyses of sorted T cells from a heterozygous (Bcl11b^dp+/⁻) mouse to determine timing of excision. (B) Bcl11b immunoblot in sorted DP and thymic SP cells of indicated mice. Actin was used as a loading control (lower panel).

Thymic cellularity was reduced by more than half in Bcl11b^dp−/− mice compared with control animals (average of 66 × 10⁶ cells compared with 152 × 10⁶ cells for control mice; Supporting Information Fig. 3). Strikingly, CD4⁺ and CD8⁺ SP thymocytes were almost completely absent in these mice (Fig. 2A), and only a reduced proportion of CD3^hi cells was detected in Bcl11b^dp−/− thymuses (8.3% compared with 19% in control mice). Most of the mutant CD3^hi cells had a DP phenotype, with slightly downregulated CD4 and CD8 levels (Fig. 2B, compare right with left panel). The mutant DP population was flanked by CD4⁺CD8^lo and CD4^loCD8⁺ cells expressing high levels of CD3 and CD24 (Fig. 2A and B), suggesting that they had not fully matured (note that these cells lacked detectable Bcl11b protein; Fig. 1B). Expression of TCRαβ and CD69 was also detected in a small proportion of mutant DP thymocytes (Fig. 2C). These analyses indicated that CD4-Cre-deleted DP cells completed some aspects of differentiation but failed to mature to the SP stage. Our results are consistent with those previously reported by Albu et al. [26], although the differentiation block observed previously appears to be more severe than that observed here (Discussion).

Figure 2.

Loss of Bcl11b in DP thymocytes impairs differentiation to CD4⁺ and CD8⁺ SP cells. (A) Thymocytes were stained for CD4, CD8, CD3, and CD24. The CD4/CD8 profile is shown in left panels; right panels show CD3/CD24 expression on the populations gated in the left panels. (B) Same stainings as in (A), CD4 and CD8 expression is shown on CD3^lo- and CD3^hi-gated cells, respectively. (C) TCRαβ and CD69 staining on total thymocytes. Data are representative of >10 independent experiments.

Low numbers of unconventional NKT cells in peripheral lymphoid organs

Spleen and lymph node cellularity was similar between Bcl11b^dp−/− and control mice (Supporting Information Fig. 3). However, Bcl11b^dp−/− organs contained markedly reduced populations of T cells, which expressed lower levels of CD4 and CD8 than WT T cells (Supporting Information Fig. 4A). All mutant peripheral T cells expressed high levels of CD44, and variable levels of CD62L, suggesting activated or memory phenotypes (Supporting Information Fig. 4B). However, most of these cells (>60%) also expressed NK1.1, suggesting that they might be related to NKT cells (Supporting Information Fig. 4C). Indeed, these cells were reminiscent of the unconventional CD44^hi NKT cells, which have been described in other systems where T-cell differentiation is severely impaired [30]. In agreement with this notion, CD3⁺ splenocytes from Bcl11b-deficient mice expressed the NK cell markers CD94, Ly49A, Ly49C/I/F/H, and NKG2D (Supporting Information Fig. 4D), previously shown to be hallmarks of unconventional NKT cells [30].

Absence of canonical NKT cells in Bcl11b^dp−/− mice

The above data revealed that CD4-Cre-deleted mice exhibited more NK1.1-expressing T cells in the periphery and thymus than WT mice (Supporting Information Fig 4C and Fig. 3A, respectively). Although NK1.1 is frequently expressed by NKT cells, binding to CD1d tetramers loaded with the glycosphingolipid antigen α-galactosylceramide (α-GalCer) is considered the best criterion to identify conventional NKT cells, as these cells express a T-cell receptor bearing an invariant Vα14-Jα18 chain that is specific for CD1d molecules loaded with α-GalCer [31]. However, CD1d tetramers loaded with α-GalCer failed to label cells within the thymus and the peripheral lymphoid organs of Bcl11b^dp−/− mice (Fig. 3B). Because NKT cells have been shown to differentiate from DP thymocytes, Bcl11b expression at the DP stage appears thus to be essential for promoting the differentiation of canonical NKT cells.

Figure 3.

NKT cell populations in Bcl11b^dp−/− mice. (A) Expression of CD3 and NK1.1 on thymic and splenic cells from control and CD4-Cre-deleted mice. (B) Detection of Vα14⁺ NKT cells by staining with CD3 and an αGalCer-loaded CD1d tetramer. Representative of three independent experiments.

Bcl11b controls T-cell differentiation intrinsically

To distinguish if the block in T-cell differentiation in Bcl11b^dp−/− mice was due to a cell-intrinsic defect, or an indirect effect from the thymic microenvironment, we performed single and mixed BM chimeras to allow the development of Bcl11b^dp−/− progenitors in a WT environment. Lethally irradiated B6.Ly5^SJL mice (which express the Ly5^SJL allele) were reconstituted with BM cells from Bcl11b^dp−/− or undeleted mice (single chimeras where both types of donor cells express the Ly5^B6 allele), or with 50:50 mixes of WT BM cells (B6.Ly5^SJL-positive) and BM cells from Bcl11b^dp−/− or control mice (double chimeras). Both single and double chimeras exhibited the same block in Bcl11b^dp−/− T-cell and NKT cell differentiation as described above (Fig. 4). These results demonstrate that the T- and NKT cell phenotypes observed in Bcl11b^dp−/− mice are due to a cell-intrinsic activity of Bcl11b in DP thymocytes, which could not be rescued by the presence of either T cells or stromal cells from WT mice.

Figure 4.

Cell-intrinsic requirement of Bcl11b in T- and NKT-cell differentiation. BM cells from Bcl11b^L2/L2 or Bcl11b^dp−/− mice were used to reconstitute lethally irradiated B6.LY5^SJL mice, either alone (left panels), or in combination with competitor BM cells from B6.LY5^SJL mice. T cells from the Bcl11b^L2/L2 or Bcl11b^dp−/− donors express the Ly5.2 haplotype of Ly5, whereas cells from recipient or competitor mice express the Ly5.1 haplotype. In the right panels, donor-derived cells were identified by gating on Ly5.2-positive cells. (A) CD4 and CD8 expression on thymic and splenic cells. (B) Staining with anti-CD3 antibody and αGalCer-loaded CD1d tetramer. Representative of three independent experiments.

Early expression of a differentiated gene expression program in Bcl11b-deficient DP cells

Bcl11b-deficient DP cells were previously shown to exhibit alterations in the expression of a small set of genes involved in positive selection and programmed cell death, such as CD5, PD1, or Pik3r3 [26]. We performed a global gene expression analysis by comparing the transcriptome profiles of CD4⁺CD8⁺CD3^lo thymocytes sorted from Bcl11b^L2/L2 and Bcl11b^dp−/− mice (two independent samples for each genotype), using Affymetrix 430 2.0 arrays. We studied the more immature CD3^lo DP population because the differentiation of CD3^hi DP cells appeared to be severely perturbed in the mutants. As shown in Fig. 5A, there was a clear dysregulation of global gene expression in Bcl11b-deficient cells, as evidenced by the degree of dispersion in the expression values between the control Bcl11b^L2/L2 and the Bcl11b^dp−/− samples. The expression of 835 probe sets was increased >1.4-fold, whereas that of 608 probe sets was decreased by the same magnitude in all possible mutant/WT comparisons (Fig. 5B, left panel; see Supporting Information Tables 1 and 2 for a complete list of all dysregulated probe sets). The dysregulated probe sets corresponded to 1130 unique genes. Mutant DP cells displayed more increases than decreases of gene expression when compared with WT cells, and this was particularly striking among genes with the highest magnitude of dysregulation (Fig. 5B, right panel). Most of the dysregulated genes were causally dysregulated by the deletion of Bcl11b, estimated by the low number of false positives (“nonspecific” in Fig. 5B, estimated by performing nonspecific comparisons of the various combinations of groups comprising each one WT and one mutant sample). Thus, taking into account the low rate of false discovery and the redundancy among probe sets, our results indicate that loss of Bcl11b in DP cells leads to the altered expression of approximately 1000 genes. The dysregulation of several genes identified with the Affymetrix arrays was also confirmed by RT-qPCR using independent samples (Fig. 6 and Table 1). In several cases (Zbtb7b, Runx3, CD160, and Itgb7), the real magnitude of the dysregulation was even higher than that observed by microarray profiling (Table 1). It should be noted that lower fold changes detected by microarrays are likely to underestimate the real magnitude of the changes, especially for genes, such as Zbtb7b, which are expressed at low levels in the control samples.

Figure 5.

Large-scale dysregulation of gene expression in Bcl11b^dp−/− DP cells. (A) Scatter plots comparing gene expression from Bcl11b^L2/L2 (labeled “WT”) and Bcl11b^dp−/− (labeled “mut”) DP cells. Red and blue dots indicate up and downregulated genes. (B) Bar charts indicating number of genes with expression levels changing by at least 1.4-fold (left panel) or three-fold (right panel) in all four paired WT versus mutant comparisons. To estimate the false discovery rate, nonspecific comparisons were performed between groups comprising one WT and one mutant sample. The numbers of genes that changed in the four possible nonspecific comparisons are shown in grey bars. Note that when the more stringent criterion was used (right panel), nonspecific changes were found in only one of the nonspecific comparisons. (C) The present Affymetrix data were merged and normalized with those from splenic CD4⁺ and CD8⁺ cells. The 835 upregulated and 609 downregulated genes shown in (B) and Supporting Information Tables S1 and S2 were clustered according to their expression across all samples. Red and green colors represent expression levels that are increased and decreased with respect to the average expression across samples. The position of selected genes is highlighted.

Figure 6.

Dysregulated expression of Zbtb7b and Runx3 in Bcl11b^dp−/− DP cells. RT-qPCR analyses were conducted using the indicated sorted DP cells from two WT mice (WT3-4) and three mutant mice (MUT3-5). Three technical replicates were generated from each sample. ^***p<0.005 and *p<0.05.

Table 1.

RT-qPCR verification of the dysregulation of selected genes

Gene symbol	Gene name	Fold change (RT-qPCR)	Fold change (microarray)
Upregulated
Klf2	Kruppel-like factor 2	1.8±0.7	4.8
Lmo4	LIM domain only 4	4.6±0.5	4.4
Id2	Inhibitor of DNA binding 2	5.1±3.2	12.5
CD9	CD9 antigen	11±2	4.2
Itgb7	Integrin β7	14±3	6.7
CD160	CD160 antigen	137±29	8.2
Downregulated
Scn4b	Sodium channel, type IV, β	45±27	9.1

Multiple, technical replicates were generated using sorted DP cells from two Bcl11b^L2/L2 or three Bcl11b^dp−/− mice. Average values of the fold change measured by the microarrays are given in the right column.

Pathway analyses using the Ingenuity Pathway Analysis software indicated that several gene networks were affected by Bcl11b deficiency. These included genes involved in G2/M transition, as well as signaling pathways centered on ERK, NFκB, TCR, JAK/STAT, and PI3K/AKT (Supporting Information Fig. 5). In addition, many of the genes affected by Bcl11b deficiency encode transcription factors/cofactors, which were either upregulated (Zbtb7b/ThPok, Runx3, Id2, Jun, Klf2, Lmo4, OBF-1/ Pou2af1, Foxo1, Klf10, Ikzf2, NFATc2, STAT4, Lyl1, MTA1, MTA3, and the Groucho-related corepressors TLE2, TLE3 and TLE6) or down-regulated (TOX3, Ikzf3, SATB1, Klf3, Zbtb4, Jmjd3, and Sin3B), suggesting that some of the dysregulations might be secondary to the mis-expression of these factors.

Among the genes strongly induced in Bcl11b-deficient DP cells, several were known to be expressed at high levels in SP T cells and low levels in WT DP thymocytes, such as Zbtb7b and Runx3. To determine if a mature T-cell gene expression program was prematurely induced in Bcl11b-deficient DP cells, we compared the above transcriptomic profiles with those from mature splenic CD4⁺ and CD8⁺ T cells [20]. Strikingly, these analyses revealed that more than half of the probe sets dysregulated in Bcl11b-deficient DP cells, induced or repressed, displayed an expression profile closer to that of WT SP cells than DP cells (Fig. 5C, and Supporting Information Tables 1 and 2). In particular, several of the upregulated genes encode transcriptional regulators known to be critical for SP cell differentiation and/or function. These include Zbtb7b and Runx3; Id2, an inhibitor of E protein function that is critical for T-cell differentiation and CD8⁺ T-cell effector function [32]; Klf2, a Krüppel-like zinc finger protein that is essential for SP cell survival and migration [33, 34]; c-Jun, a component of the AP1 factor that is involved in TCR-dependent transcriptional responses [35–37]; Pou2af1 (also known as Bob-1 or OcaB), a cofactor for Oct factors important during T-cell activation and T-helper cell polarization [38, 39] and FoxO1, a factor important for T-cell homeostasis [40].

SP gene expression in mutant CD3^lo DP cells is unlikely to reflect contamination from mature cells

A potential caveat of the above results is that the CD3^lo DP cells from Bcl11b^dp−/− mice may not represent a pure population of immature, unselected, DP cells, and might contain cells derived from more mature populations, possibly owing to the difficulty to resolve the mutant cell populations with the CD8, CD4, and CD3 markers. To address this issue, we analyzed the expression of several genes previously found to be induced in WT DP cells during positive selection, using transcriptome data from a published comparison of gene expression profiles of unselected DP cells (CD69⁻ DP cells from Zap70-deficient mice) to selected, CD69^hi cells from WT animals [41] (data accessible at NCBI GEO database accession GSE2262). Although some selection-induced genes were indeed overexpressed in the CD3^lo DP cells from Bcl11b^dp−/− mice (Zbtb7b, Id2, Klf2, CD53, IL7r, and Irf7), several others were expressed at similar low levels in WT and mutant cells (Itm2a, Nr4a1, Bcl2a1a, Slfn1, Mapk11, Nr4a3, Tnfrsf9, Acvrl1, Ccr7, Ephx1, Ms4a4b, St6gal1, Tes, Nab2, and Ccl22), suggesting that the mutant CD3^lo DP cells do not exhibit a general induction of the gene expression program associated with thymocyte maturation. We selected five of these genes (Ccr7, Slfn1, Ephx1, Ms4a4b, and Mapk11) for further analysis, as these genes displayed strong differences in gene expression levels between unselected and selected cells in the data from Sun et al. [41] (>3 log induction), and were thus likely to be informative with respect to the selection/purity status of the analyzed populations.

We sorted CD3^loDP, CD3⁺DP, CD3⁺CD4⁺ SP, and CD3⁺CD8⁺ SP cells from two WT and two Bcl11b^dp−/− mice (see Supporting Information Fig. 6 for sorting gates and purity of the sorted populations) and analyzed the expression of the selected genes in these populations by RT-qPCR (Fig. 7). In WT samples, all five genes were expressed at low levels in CD3^lo DP cells and strongly induced in the CD3⁺ DP and SP populations, thus validating previous microarray results [41]. In agreement with our transcriptome data, all five genes were also expressed at very low levels in mutant CD3^lo DP cells. Two genes (Ephx1 and Ms4a4b) were strongly induced in the mutant CD3⁺DP and SP-like populations. This observation reveals that the phenotypically more mature cells from Bcl11b^dp−/− mice have retained the capacity to induce a subset of the genes normally upregulated during positive selection. Hence, the lack of expression of these genes in the mutant CD3^lo DP cells strongly suggests that this population is unlikely to be substantially contaminated with more mature cells. The Ccr7, Slfn1, and Mapk11 genes were weakly induced in mature cell populations from only one of the mutant mice, but remained at background levels across all populations in the samples derived from the second mouse. This observation suggests that some gene-specific variability exists across mutants in their ability to activate genes induced during positive selection, and is in agreement with the previous results demonstrating impaired expression of genes associated with positive selection in DP cells from Bcl11b-deficient thymocytes [26]. Collectively, these analyses indicate that the premature expression of SP-associated genes in Bcl11b^dp−/− DP cells reflects gene-specific dysregulation in cells that have not undergone positive selection.

Figure 7.

Expression of positive selection-induced genes in Bcl11b^dp−/− thymocyte subsets. The indicated subpopulations were sorted from two Bcl11b^L2/L2 mice (WT5, WT6) and two Bcl11b^dp−/− mice (MUT6, MUT7), and the expression of the indicated genes analyzed by RT-qPCR (see Supporting Information Fig. S6 for sorting gates). “n.s.” (not significant) indicates signals too low to measure. Data were normalized with respect to the expression of Gapdh. Error bars represent the standard deviation of two to three technical replicates.

Bcl11b binds to regulatory regions of dysregulated genes

To determine if Bcl11b directly controls the expression of some dysregulated genes, we mapped Bcl11b binding to regulatory sequences by performing ChIP-seq experiments on chromatin immunoprecipitated from WT thymocytes (a full bioinfomatic analysis of these data will be published elsewhere). We found that Bcl11b was present at multiple, specific regions in most of the genes that were dysregulated in our transcriptomic analyses (see Fig. 8 for a representative selection of binding profiles). Of particular interest, Bcl11b bound to several regions within the Zbtb7b locus, including the distal regulatory element, which has been reported to be the target of TCR signal(s) responsible for CD4 lineage commitment [42]. These data indicate that Bcl11b likely acts directly in DP cells to prevent the premature expression of genes encoding critical regulators of the SP differentiation programs.

Figure 8.

ChIP-seq analysis of Bcl11b binding to regulatory regions of dysregulated genes. Histogram of Bcl11b ChIP-Seq reads and gene prediction tracks from the USCS genome browser of the indicated genetic loci and upstream regions. ^*p<10⁻⁵, peaks of reads that are enriched significantly over the input sample.

Discussion

The results presented herein further establish Bcl11b as a key regulator of cellular differentiation in the αβ T-cell lineage. Bcl11b plays a critical role in at least two stages of T-cell development: progression of DN to DP cells, and differentiation of DP cells into CD4⁺ and CD8⁺ SP cells, and NKT cells. Although our results confirm the previous results with respect to the early T-cell block resulting from the germline deletion of Bcl11b [25], and the block of SP T-cell differentiation of CD4-Cre-deleted mice [26], our studies also bring new and important insights. Specifically, we show that Bcl11b is: (i) absolutely and intrinsically required in DP thymocytes for canonical NKT cell development, (ii) required for the correct expression of approximately 1000 genes in DP cells, acting as a bifunctional transcriptional regulatory protein, and (iii) required in CD3^lo DP cells to prevent the premature expression of a large number of SP-specific genes, including the key regulators Zbtb7b and Runx3. In addition, Bcl11b was found to bind to genomic regions of dysregulated genes, including the previously characterized distal regulatory element region of the Zbtb7b promoter, which is critical in the induction of this gene upon CD4 commitment. Together, these results identify Bcl11b as a central regulator of genes associated with T-cell maturation at the DP stage.

The phenotype of the Lck-Cre-excised mutants recapitulated that of mice with a germline disruption [25]. These mice exhibited a severe differentiation block in DN cells, accompanied by a dramatic reduction in thymic cellularity, consistent with a role of Bcl11b in the survival of immature thymocytes [25]. Importantly, loss of Bcl11b either in the germline (Bcl11b^L−/L) or in the DN1-DN2 cells (Bcl11b^L2/L2−Lck^cre/+) preferentially affected the αβ T-cell lineage while appearing to spare γδ T cells. In both cases, a large percentage of Bcl11b-null cells expressed TCRγδ, most notably in the CD8⁺ population. TCRγδ expression might reflect impaired TCRβ rearrangement [25], and subsequent attempts by the developing thymocyte to use a surrogate route of differentiation. Alternatively, Bcl11b may play a more active role in the cell-fate choice between the αβ and the γδ lineages. This possibility is supported by the strong upregulation of TCRγ transcripts in Bcl11b-deleted DP cells (>100 × compared to WT, Supporting Information Table S1), suggesting a possible role of Bcl11b in repressing TCRγ expression. Note, however, that DP cells from Lck-Cre- (or CD4-Cre-) deleted mice did not exhibit surface TCRγδ expression (Supporting Information Fig. 7).

As previously reported [26], disruption of the Bcl11b locus in DP cells resulted in a block in the differentiation into CD4⁺ and CD8⁺ SP cells. In addition, we observed a loss of canonical NKT cells in CD4-Cre-deleted mice, a T-cell population that has also been shown to differentiate from DP cells [43]. However, the block in T-cell differentiation in our mice appeared less severe than that reported by Albu et al. [26] – while we observed CD3^hi (Fig. 2B) cell populations that were at least partially engaged into an SP differentiation process, such cells were apparently not as abundant in the mice described by these authors [26]. These differences may possibly be attributed to differences in the timing of the deletion, as different CD4-Cre deleter lines were used in both studies, and/or genetic background differences.

The large-scale changes in the gene expression program of DP cells appear to be at the heart of the mutant phenotype. In addition to the large number of genes encoding transcription factors that are dysregulated in DP cells from Bcl11b^dp−/− mice (see above), Bcl11b also regulates expression of a variety of genes that play key roles in signaling cascades during T-cell differentiation (e.g. IL7R (up), Lck (down), Notch1 (up), and Jak1 (up)), and in ubiquitous pathways, such as ERK and PI3K/AKT (Supporting Information Fig. 5). Thus, Bcl11b appears to function as a master transcriptional regulator that is required for the harmonious interplay of numerous signaling cascades and transcriptional networks in DP thymocytes. Whether the aborted SP and NKT cell differentiation in Bcl11b^dp−/− mice occurs as the result of the dysregulation of one or several specific genes, or as the combined deregulation of many biological processes and pathways remains unknown. We note, however, that expression of RORγ and Runx1, two factors that are essential for NKT cell differentiation [43], was normal in Bcl11b^dp−/− mice, indicating that Bcl11b controls NKT cell development independently of these factors.

Our expression profiling analyses suggest that Bcl11b is required to prevent premature and inappropriate expression of many genes specifically expressed in mature CD4⁺ and/or CD8⁺ T cells. We speculate that Bcl11b may serve as a timing factor that holds cells in the immature, DP state until a constellation of factors is in place to support SP differentiation. It is likely that the premature SP gene expression program that is induced in the Bcl11b-deficient DP cells reflects both the direct loss of Bcl11b-dependent repression, and the precocious activity of SP-specific transcription factors (such as Klf2, Zbtb7b, Runx3, and Id2). Therefore, our data suggest that correct regulation of SP cell differentiation involves mechanisms not only to induce cell-specific gene expression programs, but also to prevent these programs from being inappropriately expressed in immature cells. Mechanisms that prevent early expression of differentiation-associated genes have also been described in other systems. For instance, Polycomb-dependent repression has recently been shown to prevent the premature expression of structural genes in differentiating keratinocytes [44].

It is of particular interest that that loss of Bcl11b in DP cells expressing low levels of CD3 results in the induction of genes encoding Zbtb7b and Runx3, which are required for, and strongly upregulated during, CD4 and CD8 SP differentiation programs, respectively [45, 46]. We found that Bcl11b bound to sequences in the regulatory regions of these genes, suggesting that Bcl11b directly represses expression of Zbtb7b and Runx3 in immature T cells. The regulation of Zbtb7b has been intensively investigated in recent years. Induction of Zbtb7b expression occurs downstream of TCR signaling and requires activation of GATA3 expression [47], whereas Runx3 contributes to Zbtb7b repression in CD8-committed cells [19]. The mechanisms that render Zbtb7b silent prior to TCR signaling are less well understood but may in part involve repression by Runx complexes [19]. Our present data suggest an essential role for Bcl11b in this early silencing, and thus identify another key player in the regulatory network controlling the dynamic regulation of Zbtb7b during T-cell differentiation. However, our results also raise several questions about how Bcl11b participates in Zbtb7b regulation. It will be important to identify activators responsible for Zbtb7b expression in Bcl11b-deficient DP cells, and determine how Bcl11b antagonizes these activators at the transcriptional level in WT cells. Furthermore, as Bcl11b protein levels remain unchanged between immature and differentiated T cells (Fig. 1B), it will be important to understand how Bcl11b-mediated Zbtb7b repression is modulated during T-cell maturation. In this respect, the region that binds GATA3 [47] is distinct from those that appear to bind Bcl11b, and TCR-induced GATA3 binding may thus simply override the repressive activity of Bcl11b on Zbtb7b expression. Alternatively, or in addition, the transcriptional regulatory activity of Bcl11b might itself be influenced by TCR signals.

In summary, the present and previously published data identify Bcl11b as a crucial regulator that is essential during both the DN and the DP stages of T-cell development. Bcl11b appears to act predominantly as a transcriptional repressor in DP cells, highlighting the importance of preventing premature and inappropriate gene expression in these cells prior to initiation of an SP differentiation program.

Materials and methods

Materials and methods are provided as Supporting Information on line.

Abbreviations

α-GalCer

α-galactosylceramide

DN

double negative

DP

double positive

SP

single positive