Miz-1 Is Required To Coordinate the Expression of TCRβ and p53 Effector Genes at the Pre-TCR “β-Selection” Checkpoint

T cell development starts from early thymic precursors (ETP), which progress through four CD4⁻CD8⁻ double-negative (DN1-DN4) subsets that differentially express c-Kit, CD25, and CD44 (reviewed in Ref. 1). After cytokine-dependent steps of development and commitment, which also require signaling through the Notch, IL-7R, and Wnt pathways, early DNs differentiate into DN3 cells that express T cell-specific genes such as Rag-1, Rag-2, pTα, CD3ε, Lck, and TdT (reviewed in Ref. 2). DN3 thymocytes can be subdivided into DN3a and DN3b based on their size and CD27 expression (3, 4). DN3a cells rearrange the TCRβ-chain locus and, if productive, become DN3b blast cells that express a TCRβ-chain on the cell surface together with an invariant pTα-chain forming the pre-TCR (3, 4). DN3b cells become activated by TCR-dependent signals referred to as the “β-selection” checkpoint, downregulate CD25, and generate DN4 cells, which, in turn, differentiate into the largest thymic subpopulation, the CD4⁺CD8⁺ double-positive (DP) cells (reviewed in Refs. 5, 6). DN3a cells that fail to undergo productive V(D)J recombination at the TCRβ locus are eliminated by apoptosis.

The survival signals that assure proper T cell differentiation are promoted by increasing the expression of antiapoptotic Bcl2 and Bcl-x_L, and by redistributing the cell death proteins Bax and BAD (reviewed in Ref. 7). In addition, a role for the p53 tumor suppressor protein in the regulation of cell cycle progression and apoptosis in response to the physiological DNA damage generated by V(D)J recombination at the DN3 stage of pre-TCR cells has been proposed (8). Interestingly, a differentiation block from DN3/DN4 to DP cells caused by defective pre-TCR signaling, for instance, in the absence of Rag-1/2, DNA-PK (SCID), or the CD3γ-chain, can be rescued, to some extent, by the simultaneous loss of p53, suggesting that the β-selection at the DN3 stage depends on a balance between pre-TCR signaling and p53 activation. Importantly, p53 activation has to be contained in the presence of ongoing V(D)J recombination to avoid that developing pre-T cells immediately undergo apoptosis (9–12). However, the mechanisms to control a p53-dependent DNA damage response in the presence of V(D)J strand breaks have yet to be fully elucidated.

A number of studies showed that the oncoprotein c-Myc plays a role in many cell lineages, including B and T cells, and the generation of conditionally deleted c-Myc alleles in mice has allowed researchers to assess the role of c-Myc in T cell development more precisely (13, 14). According to these findings, c-Myc is required to ensure pre-TCR–induced proliferation and expansion of both DN3 and DN4 cells (13, 14). It is known that c-Myc protein forms a heterodimeric complex with the transcription factor Max, and that the c-Myc/Max complex can activate transcription by binding to E-boxes (CACGTG) in upstream enhancer elements of target genes (15–20). Moreover, c-Myc also interacts with the Broad-complex, Tramtrack and Bric-à-brac/pox virus zinc finger (BTB/POZ) domain transcription factor Myc-interacting zinc finger protein-1, and a ternary c-Myc/Max/Miz-1 complex has been shown to occupy target gene promoters in an E-box–independent manner. In contrast with c-Myc, which is a helix-loop-helix protein and contains a leucine zipper, Miz-1 is composed of 13 zinc finger domains at its C terminus and a BTB/POZ domain at its N terminus (21, 22). Whether Miz-1 activates or represses the transcription of its target genes depends on its interacting partner, but an intact POZ domain is required for both activities. The genes that encode the negative cell cycle regulators Cdkn2b (22, 23) or Cdkn1a (24, 25) have been validated as direct Miz-1 targets. These target genes are activated by Miz-1 and the positive cofactors histone acetyltransferase EP300 (p300) and L23-nucleophosmin, and repressed by the c-Myc/Miz-1 complex (22, 23, 26, 27). It has also been shown that c-Myc is recruited to the Cdkn1a promoter by Miz-1, and this interaction blocks Cdkn1a induction by p53 and other activators in colon cancer cells. As a result of Miz-1 actions, c-Myc switches the cell fate from cell cycle arrest to apoptosis in response to p53-dependent activation (24).

Because the full knockout of Miz-1 arrests development at an early stage of gastrulation (28), we have generated mice carrying a conditional allele of Miz-1, which produces a truncated protein lacking the functionally critical N-terminal BTB/POZ domain (hereafter named Miz-1^ΔPOZ). Using these mice, we were previously able to demonstrate a new c-Myc–independent function of Miz-1 in early steps of B- and T-lymphoid development, where IL-7R signaling regulates survival and commitment (29, 30). We now report that Miz-1^ΔPOZ mice have an additional defect in pre T-cell development at the DN3/DN4 pre-T cell transition. Our data suggest that Miz-1 is important for generating survival signals by counteracting a p53-dependent pathway and assuring the expression of the pre-TCR β-chain to support the proliferative burst of DN3 cells.

Mice had been bred on C57BL/6 background for at least 10 generations and were maintained in Specific-Pathogen-Free Plus environment. All mice used in this study were previously described (29, 30). OTI mice expressing TCRαβ transgenes were purchased from The Jackson Laboratory [C57BL/6-Tg(TCRaTCRb)1100Mjb/J]. The Institutional Review Board approved all animal protocols, and experimental procedures were performed in compliance with the Institut de Recherches Cliniques de Montréal guidelines.

OP9DL1 cultures and P6D4 SCID.adh murine thymic lymphoma were used as previously described (30). All Abs were from BD Bioscience except when indicated. To analyze DN thymic subsets, we selected CD25 (PC61.5 from eBioscience) and CD44 (IM7) plus lineage marker-negative cells (Lin⁻) by staining thymocytes with the biotinylated Abs against CD3ε (145-2C11), CD4 (RM 4-5), CD8α (53-6.7), CD45/B220 (RA3-6B2), Gr-1 (RB6-8C5), CD11b (Mac-1, M1/70), Ter-119 (Ly-76), NK1.1 (PK136 from eBioscience), Pan-NK (DX5), TCRγδ (GL3), followed by Streptavidin-PerCPCy5.5 or PECy5. When OTI transgenic (Tg) mice were analyzed, the same lineage mixture without anti-CD3ε was used. Additional staining was performed using CD24 (heat stable Ag [HSA], M1/69), CD27 (LG.3A10), pTα (2F5), and TCRβ (H57-597) Abs. Ab incubation was performed at 4°C for 20 min in PBS buffer with 1% FBS. Cells were analyzed with a FACSCalibur, FACScan, or LSR (Becton-Dickinson). Cell sorting was performed using a MoFlo cell sorter (Cytomation).

Cells were fixed with formaldehyde (BD cytofix 554655), blocked with 5 μg Armenian Hamster Ig2, λ1 (Ha4/8), and washed and permeabilized with BD Cytofix Fixation/Permeabilization kit (Becton Dickinson). Intracellular TCRβ staining was performed and surface staining was done thereafter. For immunoblot analysis, cells were sorted and lysed in 1% Nonidet P-40 (NP-40) containing 20 mM Tris-HCl pH 7.5, 420 mM NaCl, 2 mM EDTA, 1 mM MgCl₂, and 1 mM EGTA in the presence of protease and phosphatase inhibitors. Immunoblotting was performed using anti-Cdkn1a (556431; BD Biosciences) or anti–β-actin (I-19; Santa Cruz).

A total of 2 × 10⁶ cells was incubated in HBSS buffer containing 0.001 M CaCl₂, 0.001 M MgCl₂, 0.001 M HEPES, 0.1 g BSA. Cells were loaded with Indo1-AM dye (Invitrogen) at 31°C. After washing and surface staining, the cells were coated with biotinylated anti-CD3 (145-2C11) or biotin anti-TCRβ. Warm HBSS buffer was added and cells were acquired for 30 s. Avidin (Sigma) was added for the crosslinking (represented by the arrow in Supplemental Fig. 2C), and data recording was performed for a total time of 5 min. As positive control for Ca²⁺ mobilization, ionomycin was used at 4 μM and cells were acquired for 2 min total after a first fluorescence at 405 (bound Ca²⁺) versus 510 nm (free Ca²⁺).

For pERK1/2 staining, after stimulation (CD3/CD28 biotin at 10 μg/ml followed by avidin crosslinking at 20 μg/ml, or 25 ng/ml PMA/1 μM ionomycin), cells were fixed with formaldehyde (BD cytofix 554655) and additionally permeabilized with methanol (BD phosflow Perm III 558050). Samples were then stained with anti–phospho-p44/42 MAPK (Erk1/2) Thr202/Tyr 204 mouse Ab (E10), Alexa Fluor 488 conjugate (Cell Signaling), or isotype control. Surface staining was performed after the intracellular staining.

Cell cycle analysis was performed on sorted DN3-DN4. Cells were directly sorted in modified Krishan buffer (0.1% sodium citrate, 0.3% NP-40) containing 0.05 mg/ml propidium iodide and 0.02 mg/ml RNase, and analyzed after 30 min of incubation. Apoptosis rates were measured by Annexin V staining (Annexin V-allophycocyanin kit; BD Pharmingen).

For RNA isolation, cells were FACS-sorted directly in TRIzol Reagent (Invitrogen). RT-PCR was performed using Superscript II (Invitrogen). Real-time PCR was performed in triplicates on the Invitrogen Mx3005 in 20-μl reactions using TaqMan Universal PCR Master Mix (Applied Biosystems). The expression of the gene of interest was calculated relative to GAPDH mRNA (ΔC_T) and is presented as fold induction relative to values obtained with the respective control (set as 1-fold). Primers used for the experiments are available upon request. FITC-Morpholino oligo against Cdkn1a mRNA (5′-GTCGGACATCACCAGGATTGGACAT-3′ fluorescein) or a control Morpholino oligo (Gene Tools) were used on DN3a sorted cells or P6D4 clone, as previously described (30).

DN3 from wild-type (WT) and Miz-1^ΔPOZ thymi were sorted in 0.05% Tween 20, 0.05% NP-40, 50 μg/ml proteinase K, and incubated overnight at 56°C. After proteinase K inactivation, genomic DNA was amplified by PCR with TaKaRa Ex Taq (TAKARA BIO). The amplification protocol (1 min at 94°C, 1 min at 63°C, 2 min at 72°C) for Dβ2-Jβ2, Vβ5-Jβ2, Vβ8-Jβ2, Vβ11-Jβ2, and eF1 (primers described in Ref. 31) was repeated 31 cycles. The amplification protocol (1 min at 94°C, 1 min at 63°C, 2 min at 72°C) for Dβ1.1-Jβ1.7 and Dβ2.1-Jβ2.7 was first performed with external primers and repeated 20 times. A total of 0.5 μl from the first amplification was used for a second PCR reaction with nested internal primers (1 min at 94°C, 1 min at 63°C, 2 min at 72°C) for an additional 31 cycles (primers described in Ref. 32). An aliquot of the PCR product for each reaction was fractionated on a 1.8% agarose gel.

A total of 10 μg cRNA from sorted DN3 cells (triplicates) was hybridized on Affymetrix GeneChip MG-430_2.0 arrays (GPL1261). After washing and staining, GeneChips were scanned using the Affymetrix GeneChip scanner 3000 (G7 update), and data were analyzed with GCOS 1.4 software using Affymetrix default analysis settings and global scaling as normalization method. The data have been deposited in the public database Gene Expression Omnibus repository (National Center for Biotechnology Information, accession number GSE28342; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE28342).

Two-tailed Student t test was used to calculate p values where indicated. A p value ≤0.05 was considered statistically significant: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

Flow cytometric analysis showed a strongly reduced thymic cellularity affecting all T cell subsets and an increase in the relative percentages of DN cells at the expense of DP and CD4 or CD8 single-positive (SP) cells in Miz-1^ΔPOZ mice compared with control littermates (Fig. 1A, 1B). In addition, it revealed a block at the transition from DN3 to DN4 cells in Miz-1^ΔPOZ mice, most noticeable because of the relative increased frequencies of DN3 at the expense of DN4 cells and a strong relative accumulation of DN3a versus DN3b cells (Fig. 1A). In absolute cell counts and compared with WT controls, Miz-1–deficient DN3a cells were reduced by 13-fold, and DN3b and DN4 cells were reduced by 100- and 180-fold, respectively (Fig. 1B). This loss of DN3 and DN4 cells and the block at DN3/DN4/DP transition, in combination with the recently described early ETP/DN1/DN2 differentiation block (30), dramatically reduced the absolute cell counts of DP, CD4, and CD8 SP cells in Miz-1^ΔPOZ mice >100-fold compared with WT controls (Fig. 1B).

DN3b cells, which are larger in size than DN3a cells and upregulate CD27, were less frequent in Miz-1^ΔPOZ mice than in WT controls (Fig. 1A). In addition, DN3b cells expressing surface TCRβ were decreased in frequency and numbers (Fig. 1B), and their TCRβ expression levels were reduced, evident by a lower mean fluorescence intensity (MFI) compared with WT DN3b cells (Fig. 2A). Consequently, few Miz-1^ΔPOZ DN4 cells expressed TCRβ on their surface compared with the respective WT subset (Fig. 2A). Also, the frequencies of Miz-1–deficient DN3b and DN4 cells expressing a cytoplasmic (intracellular) TCRβ-chain were reduced by 2- and 2.5-fold compared with WT controls (Fig. 2B). The expression of CD3ε on the surface of the different thymocyte subsets (Supplemental Fig. 1), as well as the expression of TCRβ and pTα mRNA, and of other genes required for V(D)J recombination such as Rag-1, Rag-2, and Tdt, were at WT levels in Miz-1^ΔPOZ thymic subsets (Fig. 2C–E). In addition, no defects in Dβ2-Jβ2, Dβ1.1-Jβ1.7, or Dβ2.1-Jβ2.7 rearrangement or in Vβ5-Jβ2, Vβ8-Jβ2, or Vβ11-Jβ2 recombination were detected (Fig. 2F), indicating that all prerequisites for a proper pre-TCR selection are present in Miz-1–deficient cells.

DP and CD4 and CD8 SP cells are still found in Miz-1^ΔPOZ mice, albeit at drastically reduced numbers (Fig. 1B), and expressed normal levels of CD3 and TCRβ on their surface (Supplemental Fig. 1A, 1B). Also, Miz-1–deficient thymocytes did not show defects in TCR-dependent signaling, as evaluated by the phosphorylation of the MAPK ERK1/2 (Supplemental Fig. 2A, 2B) and the mobilization of intracellular Ca²⁺ (Supplemental Fig. 2C). Finally, the frequency of CD4 and CD8 cells within selected populations defined by the expression of CD69 and TCRβ (33, 34) was comparable between WT and Miz-1^ΔPOZ mice (Supplemental Fig. 3). This indicated that commitment to T cell lineage, mRNA expression of T cell-specific genes including the TCRβ-chain, V(D)J recombination, TCR signaling, and positive/negative selection are intact in Miz-1^ΔPOZ mice.

To further investigate the block at the Miz-1–dependent DN3/DN4 transition, we generated Miz-1–deficient mice expressing a transgene encoding for rearranged TCRαβ-chains (OTI Tg × Miz-1^ΔPOZ). In these mice, the frequencies and absolute numbers of DN3b, DN4, and DN4TCRβ⁺ cells were significantly increased compared with Miz-1^ΔPOZ thymocytes (Fig. 3), but the overall cellularity of the thymus and of DN4 or DN4 TCRβ⁺ cells did not reach WT or OTI Tg levels (Fig. 3B). This indicated that a functional TCRαβ receptor can be expressed and processed from the cytoplasm to the cell surface of Miz-1^ΔPOZ pre-T cells and confirmed that TCR signals driving DN3 to DN4 differentiation can be initiated in Miz-1–deficient mice. It also suggests that the absence of TCRβ surface expression is not solely responsible for the DN3/DN4 transitional block in Miz-1^ΔPOZ mice. In addition, the positive and negative selection was evaluated by the expression of CD69 and TCRβ in OTI Tg × Miz-1^ΔPOZ mice expressing a less diverse TCR repertoire compared with Miz-1^ΔPOZ mice. The FACS analyses of CD4 and CD8 expression did not reveal any obvious defects in the positive or negative selection processes (Supplemental Fig. 3). The efficiency of the positive/negative selection was also evaluated in previously described Lck-cre Miz-1^ΔPOZ mice, where the deletion of the BTB/POZ domain of Miz-1 starts at the DN2/DN3 transitional stage of T cell development but is complete only at the DN4 and subsequent stages (30). Consistent with a normal DN/DP T cell development in Lck-cre Miz-1^ΔPOZ mice (30), the positive/negative selection was intact in mice expressing Miz-1^ΔPOZ at a later point of T cell differentiation (Supplemental Fig. 3).

Annexin V staining that allows for detection of cells undergoing programmed cell death was used to evaluate DN3a cells from Miz-1–deficient mice. Our data showed that 37.6% of Miz-1^ΔPOZ DN3a cells were apoptotic compared with only 6.8% of WT cells (Fig. 4A). In contrast, the DN3b subpopulation showed only 9.2% of Annexin⁺ cells, and the DN4 subsets did not contain any apoptotic cells (Fig. 4A). Consistent with this, DN3a cells sorted from Miz-1^ΔPOZ or OTI Tg × Miz-1^ΔPOZ mice were unable to differentiate on OP9DL1 cultures and showed low frequencies of live cells, as determined by the forward scatter/side scatter (FSC/SSC) gate, which is indicative of apoptosis (Supplemental Fig. 4A). This suggested that enhanced spontaneous cell death occurs in Miz-1^ΔPOZ DN3a at the moment where V(D)J recombination and the β-selection checkpoint take place.

Next, we compared genome-wide expression profiles from Miz-1^ΔPOZ and WT DN3 cells, and observed that differently regulated genes between those two cell populations belonged to specific gene ontology and KEGG pathways specific for senescence, cell cycle arrest, and cell death. As illustrated by a heat map, we noticed, in particular, that the p53 target genes Bax, Noxa (Pmaip1), Atr, Apaf1, Atm, and PUMA (Bbc3) that induce apoptosis in response to a DNA damage signal were upregulated in Miz-1–deficient DN3 cells (Fig. 4B). We reanalyzed a subset of this group by RT-PCR and confirmed upregulation of PUMA, Noxa, and Bax in Miz-1–deficient DN3a cells (Fig. 4C). Because the expression of transformation-related protein 53 itself was at comparable levels with WT (Fig. 4C), these findings suggested that Miz-1 might be involved in controlling the activity of p53 or the expression of its effector genes that mediate cell death in DN3 pre-T cells. We reasoned that constitutive Bcl2 expression should counteract the enhanced cell death and restore the numbers of Miz-1–deficient DN3 and DN4 subsets. Indeed, coexpression of a previously described H2K-Bcl2 transgene that confers pan-hematopoietic expression of Bcl2 (35) significantly increased frequencies (Fig. 5A) and numbers (Fig. 5B) of DN3 and DN4 cells, but did not restore the numbers of DN3, DN4 (Fig. 5B), and DN4TCRβ⁺ cells (not shown) to WT levels, or rescued the block at the DN3/DN4 transition in Miz-1^ΔPOZ mice. Thus, although accelerated apoptosis may cause the low numbers of DN3 cells in Miz-1–deficient mice, it does not explain the block at the DN3/DN4 transition.

Our expression array data also showed that the G1-specific, cyclin-dependent kinase inhibitor Cdkn1a, which is a downstream target of p53, was upregulated in Miz-1^ΔPOZ DN3 cells (Fig. 4B). This could be confirmed by RT-PCR and Western blot analysis on Miz-1–deficient DN3 cells (Fig. 6A, 6B). We also found significant defects in cell cycle progression in Miz-1^ΔPOZ DN3 and DN4 cells (Fig. 6C). To test whether the differentiation block observed at the DN3/DN4 transition in Miz-1^ΔPOZ mice could be attributable to the defect in cell cycle progression and the upregulation of Cdkn1a expression, we generated combinatorial mutants deficient for both Miz-1 and Cdkn1a (Cdkn1a^−/− × Miz-1^ΔPOZ). However, we found that the DN3/DN4 differentiation block and the accumulation of DN3a cells at the expense of DN3b cells were not rescued in Cdkn1a^−/− × Miz-1^ΔPOZ mice (Fig. 6D).

To test whether a combination of Cdkn1a inhibition and ectopic TCR expression could rescue the Miz-1–deficient phenotype, we sorted DN3a cells from OTI Tg and OTI Tg × Miz-1^ΔPOZ, transfected them with FITC-labeled morpholino-oligonucleotide against Cdkn1a mRNA, or a control morpholino, and cocultured them on OP9DL1 cells. Although the knockdown of Cdkn1a was efficient (Supplemental Fig. 4B), OTI Tg × Miz-1^ΔPOZ DN3a cells treated with the morpholino oligo against Cdkn1a were still arrested, as assessed by the intensity of the FITC labeling, the poor survival on OP9DL1 cells, and the generation of only 5.6% of DP cells compared with 20.1% from OTI Tg DN3a cells (Supplemental Fig. 4C, 4D). Similarly, a rescue of the DN/DP differentiation block was not obtained when we treated DN3 cells sorted from Bcl2 × Miz-1^ΔPOZ mice with the morpholino oligo against Cdkn1a (Supplemental Fig. 4C, 4D). Thus, high levels of Cdkn1a are not responsible for the observed block of differentiation at the β-selection checkpoint seen in Miz-1^ΔPOZ DN3 cells.

We next examined whether the block at the β-selection checkpoint can be restored in Miz-1^ΔPOZ mice by providing both the prosurvival protein Bcl2 and the rearranged TCRαβ transgenes. Indeed, coexpression of H2K-Bcl2 and OTI TCRαβ transgenes relieved the DN to DP block observed in Miz-1^ΔPOZ mice and significantly enhanced the transition from DN3 to DN4 cells compared with controls (Fig. 7A). Both frequencies and absolute numbers of DN3b and DN4 cells were increased in Bcl2 Tg × OTI Tg × Miz-1^ΔPOZ compared with OTI × Miz-1^ΔPOZ mice, and >90% of gated DN4 cells now expressed TCRβ on their cell surface (Fig. 7A). The full rescue of the defect in pre-T cell differentiation by the introduction of both transgenes was also evident by the overall cellularity of the thymus, and most importantly by the reappearance of DN3b and DN4 cells, which were almost undetectable in Miz-1^ΔPOZ mice and now reached OTI Tg × Bcl2 Tg levels (Fig. 7B). To further show that the rescue provided by the expression of both Bcl2 and OTI transgenes was at the β-selection checkpoint, we compared HSA expression in DN3, DN4, and DP cells, because HSA is a marker that gets downregulated as thymocytes mature from DN4/DP to SP (36). The downregulation of HSA was much more noticeable, both in percentages and MFIs, as DN3 cells mature to DN4 and DP cells in Bcl2 Tg × OTI Tg × Miz-1^ΔPOZ mice compared with OTI Tg × Miz-1^ΔPOZ mice (Fig. 7C). The accelerated transition from DN3b to DN4 in Bcl2 Tg × OTI Tg × Miz-1^ΔPOZ mice suggested that the introduction of the TCRαβ transgenes allowed the cells to deliver a differentiation and proliferation signal, and that, together with Bcl2, which counteracts apoptosis, these signals were sufficient to neutralize the effect of Miz-1 deficiency on cell survival.

We have previously shown that the BTB/POZ transcription factor Miz-1 regulates IL-7R signaling by monitoring the expression levels of both SOCS1 and Bcl2. This protects ETP/DN1/DN2 cells from apoptosis and enables their differentiation (30). In this study, we show that Miz-1 has an additional function later in pre-T cell differentiation, at the β-selection stage of DN3 cells, which is the first critical checkpoint in the maturation of pre-T cells. DN3 cells passing this checkpoint actively rearrange the TCRβ locus and thus have to tightly control DNA damage response pathways such as the one induced by the activation of p53. After productive rearrangement, DN3 cells activate allelic exclusion (37) and become pre-TCR⁺ DN3b cells that undergo a massive proliferative expansion, differentiate into DN4 cells (3), and escape apoptosis. In this article, we present evidence that the transcription factor Miz-1 is essential to coordinate the steps that assure survival of DN3 cells and the expansion and differentiation of DN3 and DN4 cells. Our data suggest that Miz-1 coordinates expression of the pre-TCR and may be involved in controlling p53 target genes possibly induced by DNA double-strand breaks initiated on V(D)J recombination in DN3 cells.

FACS analyses demonstrated that DN cell differentiation is blocked in Miz-1^ΔPOZ mice at the β-selection checkpoint. However, Rag-1, Rag-2, Tdt, and TCRβ genes are expressed, and pTα and CD3ε are present in Miz-1^ΔPOZ pre-T cells. Moreover, V(D)J recombination and TCR-mediated signaling appeared to be intact in the absence of a functional Miz-1, at least in the cells that still emerged in Miz-1–deficient mice. Despite this, only few DN3b or DN4 cells that express the TCRβ protein in the cytoplasm or at the surface were present in Miz-1–deficient mice. This phenotype is detected only in Vav-cre Miz-1^ΔPOZ mice and not in the previously described Lck-cre Miz-1^ΔPOZ model, where the deletion of the BTB/POZ domain starts occurring in the DN2/DN3 transitional stage of T cell development (38). We reported that the deletion in this mouse strain started at the DN2 stage, reached 50% in DN3a cells, and was complete at the DN4 stage (30). The residual expression of Miz-1 in the DN3a/b cells of Lck-cre Miz-1^ΔPOZ mice may be sufficient to overcome the T cell differentiation block at the β-selection checkpoint. Consistent with this report, no obvious DN/DP defects in T cell development were noted (30), and the positive and negative selection processes were not affected by Miz-1 deficiency in Lck-cre Miz-1^ΔPOZ mice.

TCRβ gene is expressed at normal levels in Miz-1–deficient DN3 or DN4 cells, possibly implicating Miz-1 in the control of a posttranscriptional step affecting translation, stability, or membrane transport of the TCRβ protein. Transport to the membrane and surface expression of TCRβ is necessary to trigger proliferation and the differentiation of DN to DP cells (39). However, the introduction of a rearranged TCR transgene led to a stable, high-level surface expression of ectopic TCRαβ in Miz-1–deficient cells. In addition, the DP and SP cells that still develop in Miz-1^ΔPOZ mice expressed normal levels of TCRβ-chain on their surface. Furthermore, the frequency of CD4 and CD8 cells within selected populations defined by the expression of CD69 and TCRβ (33, 34) was comparable between WT and Miz-1^ΔPOZ mice. Because selected thymic populations could be better evaluated in mice expressing a less diverse repertoire like TCR Tg mice, we could show that positive/negative selection is not affected by Miz-1 deficiency in OTI Tg × Miz-1^ΔPOZ mice. Finally, Miz-1–deficient cells did upregulate CD69 and CD5 in response to TCR stimulation at comparable levels with WT controls (data not show). These data rather support the view that Miz-1 deficiency does not alter the expression, stability, or processing of the pre-TCR per se.

The presence of a Tg OTI TCR led to an expansion of DN3 cells in Miz-1^ΔPOZ mice but failed to rescue the DN3/DN4 block, indicating that even if the TCRβ-chain is expressed at the cell surface and is able to transmit the appropriate signal, the cells still encounter a proliferative block mediated by other signals. One of the most striking findings that could possibly explain this observation was the significant upregulation of Cdkn1a expression in Miz-1^ΔPOZ DN3 cells. At high Cdkn1a levels, Miz-1–deficient DN3 cells would be unable to react to the proliferative burst induced by the pre-TCR signaling. Although compelling, a rescue attempt showed that the DN3/DN4 block and the accumulation of DN3a cells at the expense of DN3b persisted in Cdkn1a^−/− × Miz-1^ΔPOZ mice. Therefore, although the Cdkn1a gene is a bona fide Miz-1 target and is occupied by Miz-1 on its promoter in DN thymocytes (data not shown), the deregulation of Cdkn1a by Miz-1 cannot be the only cause of the Miz-1^ΔPOZ DN3 cell arrest at the β-selection checkpoint. The Cdkn1a upregulation may rather be a consequence of the cell cycle progression arrest that is caused by both the lack of surface pre-TCR expression and the upregulated p53-target genes at this stage.

Analysis of genome-wide expression array data indicated that Miz-1–deficient DN3 cells overexpressed not only genes mediating cell cycle arrest such as Cdkn1a but also a whole set of p53 target genes that can initiate apoptosis such as PUMA, Noxa, and Bax. It is known that p53 is required for the maintenance of genomic stability, and regulates apoptosis and growth arrest in response to DNA damage, in particular to DNA double-stranded breaks (DSBs; reviewed in Refs. 40, 41). Moreover, it has been shown that pre-TCR signaling inhibits the p53 response after a successful V(D)J recombination. Such regulation is crucial for pre-TCR selection and DN/DP differentiation and survival (9–12).

Because Miz-1–deficient cells fail to express TCRβ proteins on their cell surface, it is possible that these cells cannot induce pre-TCR–dependent inactivation of p53 when they have completed a productive rearrangement that generates physiological DSBs. One effect of this failure to inactivate p53 could result in the induction of proapoptotic genes such as Bax (42) and the regulation of Bcl2 (43). Miz-1–deficient DN3 cells express high Bcl2 levels (data not shown) that should decline in developing αβ DN3 cells and also display high Bax expression, which most likely counteracts the Bcl2 function and provokes higher apoptosis rates. Interestingly, the introduction of the antiapoptotic Bcl2 protein partially restored DN3a numbers but did not rescue the DN3/DN4 block in Miz-1^ΔPOZ mice, which is consistent with the finding that Bcl2 expression alone does not promote differentiation of pre-TCR–deficient SCID DN3 cells (44).

During V(D)J recombination, DSBs are generated, and any p53 response that may initiate apoptosis has to be contained until a pre-TCR signal shuts down the p53 pathway and enables selected cells to expand. In this model, cells that fail to receive a pre-TCR signal die, which is in line with evidence showing that loss of p53 in pre-TCR–deficient thymocytes can restore their development and survival (9–12). How the p53 response is controlled during V(D)J recombination is not fully understood, but our data point to the possibility that Miz-1 is involved in this process. Although alternative possibilities exist and further experimentation is needed to fully support this hypothesis, our finding that the ectopic expression of Bcl2 and OTI Tg TCR fully rescued Miz-1 deficiency supports this view, because both Bcl2 and TCR can counteract the p53-initiated apoptosis and growth arrest in DN3 cells. A balance between the role of p53 in inducing apoptosis versus promoting cell survival has been shown to contribute to the normal development of the cells (45). Accordingly, low levels of p53 are maintained under physiological conditions and normal cell proliferation to promote the induction of temporary cell cycle arrest in stress situations to allow, for instance, the repair of DNA damage. If a damage signal persists, a p53-mediated apoptosis program is activated to eliminate these cells. Because the expression levels of p53 are comparable with WT in Miz-1^ΔPOZ cells and because it has been shown that Miz-1 can bind to p53 (46), it is possible that Miz-1 is one of the factors that dampens the part of p53 response that promotes cell death. The precise regulation of p53 activity by Miz-1 is presently unknown. It is possible that in the absence of a functional Miz-1 protein, p53-mediated cell death response may be induced, and the disturbed expression levels of Bax and PUMA as seen in DN3 cells from Miz-1^ΔPOZ mice are consistent with this view. Experiments that enable the detection of a co-occupation of DNA damage response gene promoters by p53 and Miz-1 may help clarify this point.

A recent report that directly implicates Miz-1 as a mediator of the p19ARF-p53 pathway would also be consistent with this hypothesis. In this study, evidence show that Miz-1 is able to bind to p19ARF and to interfere with p53 stability. The same study also demonstrated that Miz-1 interacts directly with p53, and that this interaction diminishes the binding of p53 to its target promoters and inhibits p53-mediated gene transcription (46). It is therefore conceivable that Miz-1 plays a role in the regulation of p53 in pre-T cells, and that a functional form of Miz-1 is necessary to control p53 activity at the specific β-selection checkpoint to allow V(D)J recombination to occur without initiating apoptosis. In the absence of a functional form of Miz-1, particularly lacking its BTB/POZ domain, the p53 response to the V(D)J recombination events is enhanced, causing cell cycle arrest and apoptosis, and this prevents TCRβ⁺ cells from expanding and initiating their differentiation into DP thymocytes. More experiments to understand this regulation need to be done to identify the particular p53 target genes that may be dependent on the expression of a functional Miz-1 protein. Taken together, the data presented in this report help explain the DN3/DN4 block observed in Miz-1–deficient mice and would establish the BTB/POZ domain protein Miz-1 as a new regulator of the pre-TCR β-selection checkpoint. Whether this regulation is mediated by cofactors or by posttranscriptional events remains to be explored.

We are grateful to Ellen V. Rothenberg and Juan-Carlos Zuniga-Pflücker for providing SCID.adh clones and OP9DL1 cells, Martin Pelletier for critical reading of this manuscript, Mathieu Lapointe and Rachel Bastien for technical assistance, and Eric Massicotte and Julie Lord for flow sorting.

The authors have no financial conflicts of interest.