The Tcra enhancer (Eα) is essential for pre-TCR–mediated activation of germline transcription and V(D)J recombination. Eα is considered an archetypical enhanceosome that acts through the functional synergy and cooperative binding of multiple transcription factors. Based on dimethylsulfate genomic footprinting experiments, there has been a long-standing paradox regarding Eα activation in the absence of differences in enhancer occupancy. Our data provide the molecular mechanism of Eα activation and an explanation of this paradox. We found that germline transcriptional activation of Tcra is dependent on constant phospholipase Cγ, as well as calcineurin- and MAPK/ERK-mediated signaling, indicating that inducible transcription factors are crucially involved. NFAT, AP-1, and early growth response factor 1, together with CREB-binding protein/p300 coactivators, bind to Eα as part of an active enhanceosome assembled during pre-TCR signaling. We favor a scenario in which the binding of lymphoid-restricted and constitutive transcription factors to Eα prior to its activation forms a regulatory scaffold to recruit factors induced by pre-TCR signaling. Thus, the combinatorial assembly of tissue- and signal-specific transcription factors dictates the Eα function. This mechanism for enhancer activation may represent a general paradigm in tissue-restricted and stimulus-responsive gene regulation.
The generation of αβ T-lymphocytes requires the construction of a TCR complex through a highly ordered series of somatic gene rearrangement events at the TCRα and TCRβ loci (Tcra and Tcrb) during T cell development. Thymocytes mature through a series of stages that are identified by the expression of surface receptors. Most immature thymocytes, known as double-negative (DN) thymocytes, are CD4−CD8−. DN thymocytes can be classified into four populations (DN1–4) based on the expression of CD25 and CD44. Tcrb rearrangements are completed at the DN3 stage. Based on the expression of CD27, DN3 thymocytes can be subdivided into two populations: those that have not yet undergone TCRβ-selection (DN3a) and those that have (post-TCRβ–selected DN3b) (1). DN3a thymocytes that have successfully rearranged a Tcrb allele differentiate into DN3b, DN4, and CD4+CD8+ double-positive (DP) thymocytes in a process known as β-selection. This process is driven by signaling through the pre-TCR, which is composed of TCRβ and the invariant pre-Tα, and through cooperating NotchRs. Pre-TCR signaling is sufficient for Tcrb allelic exclusion and the activation of Tcra transcription and rearrangement but not for DN to DP differentiation (2). Among the transcription factors induced by pre-TCR signaling, it has been well established that NFAT, AP-1, and early growth response factors (Egr) are essential for β-selection (3–7). NFAT2 and Egr-1/3 factors are especially interesting because they are sufficient for traversal of the β-selection checkpoint (5, 8–10) and functionally collaborate to perform this function (11). At present, little is known about the molecular targets of these inducible transcription factors during thymocyte differentiation.
Tcra germline transcription is required for Vα-to-Jα recombination (12–15). All VαJα rearrangements, with the exception of a few Vα2 rearrangements that occur in DN cells by the action of the Tcrd enhancer (Eδ) (16), depend on the Tcra enhancer (Eα) (17). Eα influences chromatin structure across a 500-kb region that includes the 65-kb Jα array and the proximal third part of the 1.5-megabase Vα array (18). Initial Tcra germline transcription depends on the T early-α promoter (TEA) and the Jα49 promoter (Jα49p), which are activated by the action of Eα and required for the activation of primary Vα-to-Jα recombination (19). A notable aspect of Eα as a critical regulator of Tcra locus rearrangements is its strict regulation during thymocyte development. Eα is inactive in DN1 to DN3a thymocytes when they attempt to successfully rearrange their Tcrd locus; its activity is first detected after pre-TCR signaling in DN4 thymocytes, coinciding with the detection of Tcra germinal transcripts (1, 20, 21). Because the Tcrd locus is positioned between the Vα and Jα gene segments at the combined Tcrad locus, Vα-to-Jα rearrangements cause the Tcrd locus to be deleted from the chromosome. Hence, Eα-dependent developmental control of Tcra rearrangement is a critical component of αβ and γδ T cell lineage commitment.
Eα, defined as a 275-bp fragment containing four protein-binding elements (Tα1–Tα4), is the minimal fragment required for proper developmental regulation (22). Tα1-Tα2 is considered the core enhancer because it is the smallest fragment with transcriptional activity in vitro (23), and it is thought to be controlled by a compact nucleoprotein structure, the enhanceosome, formed by functional synergy and cooperative binding to enhancer DNA among specific transcription factors (24, 25). Known factors bound to Tα1–Tα4 comprise lymphoid-specific factors that are constitutively present during thymocyte development (Fig. 1). Previous genomic footprinting experiments using dimethylsulfate (DMS) have shown that the occupancy of Eα is indistinguishable between DN3a and DP thymocytes, indicating that proteins bound in both cell stages must be identical or very closely related (21, 26, 27). However, an extensive opening of Eα chromatin was detected during the DN3a to DP transition, suggesting that the protein complex formed at the Eα is more compact in DP than in DN3a cells (22, 26). To study the requirements for the induction of Tcra transcription during β-selection, we have employed both an SCID mouse-derived DN3a cell line that can be induced to differentiate to the DN4 stage through CD3ε-transduced signaling (Scid.adh [TAC:CD3ε] [Scid.adh]) and thymocytes from Rag2−/−, CD3ε mAb-injected Rag2−/−, and Rag2−/− × Tcrb (Rxβ) mice (20, 28, 29) as models that mimic pre-TCR signaling while maintaining Tcra in germline configuration. We have found that the activation of phospholipase Cγ (PLCγ) elicits a set of inducible transcription factors that assemble into a functional enhanceosome by binding to multiple sites within Eα, thereby activating germline transcription at Tcra. Our data are consistent with a model that proposes that in DN3a thymocytes, Eα is occupied by lymphoid-restricted and constitutively expressed transcription factors. Eα in this preactivated and occupied state serves as a regulatory platform to which additional pre-TCR–inducible transcription factors are recruited in a transitory fashion, thereby activating and inducing Tcra germline transcription from the TEA and Jα49p in DN4/early proliferating DP thymocytes. Although our study has focused on the analysis of the transcriptional activity of Eα, but not its recombinational activity, as an indicator of enhancer function, these germline Tcra transcripts initiated at Jα gene segments are required for initiating VαJα recombination (12–15). This mechanism of Eα activation involving the combinatorial assembly of tissue- and signal-specific transcription factors not only provides an explanation for the long-standing paradox regarding why the activation of this enhancer is not accompanied by obvious detectable differences in the DMS footprinting of the bound proteins, but it may also represent a general paradigm in tissue-restricted and stimulus-responsive gene regulation. Furthermore, our data on the functional regulation of Eα by inducible transcription factors during thymocyte development might have important implications for T cell maturation.
Materials and Methods
Cell lines and mice
The Scid.adh cells have been described previously (20) and were maintained in RPMI 1640 supplemented with 10% FBS and standard amounts of glutamine, penicillin, streptomycin, nonessential amino acids, sodium pyruvate, and 50 μM 2-ME. Jurkat cells were maintained in RPMI 1640 supplemented with 10% FBS and standard amounts of glutamine, penicillin, and streptomycin. HEK-293T cells were maintained in DMEM supplemented with 10% FBS and standard amounts of glutamine, nonessential amino acids, penicillin, and streptomycin.
Rag2−/− and Rxβ mice (28, 29) were purchased from Taconic Farms and maintained in pathogen-free conditions within the Animal House at the Instituto de Parasitología y Biomedicina “López-Neyra.” Five- to 8-wk-old mice were used in all experiments. Animal use adhered to Instituto de Parasitología y Biomedicina “López-Neyra” and Consejo Superior de Investigaciones Científicas Bioethical Guidelines.
In vitro and in vivo cell stimulation
Scid.adh cells (1 × 105 cells/ml) were stimulated in culture with human CD25 (TAC) mAb from the hd245/332 hybridoma for 24 h at 37°C with 5% CO2 (9, 30) or stimuli such as PMA (50 ng/ml), ionomycin (1 μg/ml), PMA (20 ng/ml) and ionomycin (0.5 μg/ml), or thapsigargin (50 nM) for 3.5 or 6 h at 37°C with 5% CO2. For inhibition assays with cyclosporine A (CsA; 0.5–1 μg/ml), EGTA (10 mM), UO124 (10 μM), UO126 (10 μM), or PD98059 (30 μM) (Calbiochem), the inhibitors were added 1 h before stimulation. For the inhibition assays using U73122, 10 μM inhibitor was added in cultures at 0.2–0.5 × 106 Rxβ thymocytes/ml at 37°C with 5% CO2. For in vivo thymocyte stimulation, Rag2−/− mice were injected i.v. with 50 μg purified CD3ε (2C11) mAb as described (31) and sacrificed after 16 h.
RNA was isolated using TRIzol (Invitrogen) from unstimulated or 6 h-stimulated cells. A total of 200 U M-MLV RT (Invitrogen) and 300 ng hexaprimers was used to synthesize cDNA from 100–400 ng total RNA. The PCR conditions were performed as described (9, 12). The amplified fragments were detected using radiolabeled probes. For quantitative RT-PCR, templates (equivalent to 20 ng RNA) were assessed with reported primers (1) using IQ SYBR Green Supermix (Bio-Rad) in a Bio-Rad iCycler thermocycler (Bio-Rad). PCR conditions were: 3.5 min at 95°C, followed by 40 cycles of 30 s at 94°C, 30 s at 60°C, and 20 s at 72°C, followed by a final extension step of 5 min at 72°C. Expression levels of the transcripts were normalized to levels of Actb in each sample. Paired sample t tests were used to determine statistical significance between values. The p values are represented as follows: *p < 0.05, **p = 0.005–0.0005, and ***p < 0.0005.
rDNA-binding domains of NFAT1 (rNFAT-DBD) was purified as previously described (32), and recombinant Egr-1 (rEgr-1) was obtained from Enzo Life Sciences. For mapping the Egr binding sites in Eα, extracts from HEK-293T cells transfected with a human Egr-1 expression plasmid (pEFBOST7/Egr-1) were used. To generate this plasmid, human Egr-1 cDNA was obtained by PCR with the oligonucleotides 5′-CTCATGATCTAGACCAGCTCGCTCGTCCAG-3′ (containing an XbaI site) and 5′-CTGATGTTTCGAATTAGCAAATTTCAATTGTCCTGGGAG-3′ (containing a BstBI site), using the pCMVXLS–Egr-1 plasmid obtained from Enzo Life Sciences as a template. Egr-1 cDNA was then subcloned into the XbaI- and BstBI-digested expression vector pEFBOST7 (33). The pEFBOST7/Egr-1 was transfected into ∼6 × 105 cells seeded 20 h before at 60–70% confluence using calcium phosphate in 100-mm diameter plates. Approximately 48 h after transfection, the cells were harvested, washed in PBS, and lysed in 1 ml 20 mM HEPES (pH 7.9), 150 mM NaCl, 5 mM EDTA, 1% Nonidet P-40, 1 mM PMSF, 1 mM DTT, and protease inhibitors (1× Complete Protease Inhibitors; Roche) for 30 min on ice. The lysates were centrifuged at 12,000 × g for 5 min at 4°C to the separate soluble material from the debris, and glycerol was added to a concentration of 20%. The protein concentration was determined by Bradford assay using a protein assay (Bio-Rad), and Egr-1 expression was analyzed by Western blotting with Egr-1 Ab (Santa Cruz Biotechnology) and T7-epitope Ab (Bethyl Laboratories). Nuclear extracts from unstimulated Scid.adh cells or Scid.adh cells stimulated with PMA+ionomycin for 6 h were prepared as described previously (34), with modifications. Ten million cells were washed in 10 ml PBS and incubated for 15 min in 300 μl ice-cold buffer A (10 mM Tris-HCl [pH 7.9], 60 mM KCl, 1 mM EDTA, 0.5 mM PMSF, and a mixture of protease inhibitors [Roche]) in a 1.5-ml tube. Nonidet P-40 was added from a 10% stock solution to a final concentration of 0.4%, and the samples were vortexed for 10 s. After a 3-min incubation on ice, the nuclei were centrifuged for 3 min at 15,000 × g at 4°C, washed in ice-cold buffer A, resuspended in 100 μl ice-cold buffer C (20 mM Tris-HCl [pH 8], 0.4 M NaCl, 1.5 mM MgCl2, 1 mM EDTA, 1 mM DTT, 0.5 mM PMSF, and protease inhibitors), and incubated at 4°C under rotation for 30 min. The nuclear extracts were separated from the debris by centrifugation at 15,000 × g for 15 min at 4°C. Glycerol was added to a final concentration of 25%. The protein concentration was determined by Bradford assay (Bio-Rad).
Human Tα1-Tα2 and Tα3-Tα4 fragments were obtained as reported (22).
The consensus binding site (CS) and mutated CS oligonucleotide sequences were obtained from Santa Cruz Biotechnology. The Eα-derived oligonucleotides used in the EMSAs are listed in Supplemental Table I.
For EMSAs, Scid.adh nuclear extracts (2 μg), HEK-293T cell extracts (3 μg), NFAT-DBD, or rEgr-1 proteins were incubated with 1 μg poly(deoxyinosinic-deoxycytidylic) acid sodium salt carrier and 1 μg BSA in a 30-μl mix containing 10 mM Tris-HCl (pH 7.9), 50 mM NaCl, 1 mM DTT, and 5% glycerol for 30 min at 4°C in presence or absence of 12.5-fold excess of unlabeled CS or mutated CS competitors. Some of the reactions also included 1 μg indicated Ab and were incubated for 30 min at room temperature. All Abs were purchased from Santa Cruz Biotechnology except for the c-Jun Ab (BD Biosciences). The 32P-radiolabeled probes (80 fmol, ≈5–10 × 104 cpm) were added for an additional 20 min of incubation at 4°C. The samples were electrophoresed through a 4% PAGE gel containing 22.5 mM Tris-borate and 0.5 mM EDTA at 4°C. The abundance of DNA/protein complexes was quantified using a phosphorimager (Storm 820; Molecular Dynamics).
Chromatin immunoprecipitation experiments
Chromatin immunoprecipitation (ChIP) experiments were conducted as previously described with modifications (27). Unstimulated Scid.adh cells or Scid.adh cells (2 × 106/ml) stimulated with PMA+ionomycin for 3.5 h at 37°C or thymocytes (3–5 × 106/ml) were cross-linked in culture medium by the addition of 1% formaldehyde (v/w) and incubated for 10 min at room temperature. Alternatively, the cells to be used in ChIPs with the NFAT4 Ab were cross-linked for 20 min at room temperature. The reaction was stopped by adding glycine to 0.125 M and incubating for 5 min at room temperature. After two washes in PBS, Scid.adh cells (107) or thymocytes (107) were resuspended in 500 μl 1% SDS, 10 mM EDTA, 50 mM Tris-HCl (pH 8.1), 1 mM PMSF, and protease inhibitors and incubated for 10 min on ice. The cell suspension was sonicated using a Branson sonicator at 50% of capacity, alternating between 20 s on and 2–5 min off for 13 cycles (for a 10-min fixed sample) or 18 cycles (for a 20-min fixed sample), while the sample was immersed in an ice/water bath. Chromosomal DNA was reduced to an average size of 200–500 bp as determined by gel analysis. The lysate was centrifuged for 30 min at 20,000 × g at 4°C, and the supernatant was diluted 10-fold by adding 4.5 ml 0.01% SDS, 1.1% (v/v) Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl (pH 8.1), 167 mM NaCl, 1 mM PMSF, and protease inhibitors. Five percent of the starting material (250 μl) was saved as input for PCR detection. Chromatin was precleared by incubation for 2 to 3 h at 4°C with 150 μl 50% salmon sperm DNA/protein A-agarose slurry, prepared as recommended by the manufacturer (Upstate Biotechnology). Precleared chromatin corresponding to 5 × 106 Scid.adh cells or 7 × 106 thymocytes (∼2.3 ml) was used for ChIP by incubation for 16 h at 4°C with 10 μg specific or isotype-matched controls Abs in polystyrene tubes, followed by the addition of 70 μl protein A-agarose slurry, 600 μg BSA, and 40 μg sonicated salmon sperm DNA for an additional 1–1.5 h incubation at 4°C. All Abs were purchased from Santa Cruz Biotechnology except for the c-Jun Ab (BD Biosciences). Immunoprecipitates were transferred to 1.5-ml tubes and washed by rocking for 5 min at room temperature with the following ice-cold buffers (containing protease inhibitors): 1) buffer A; 2) 0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8), and 150 mM NaCl; 3) 0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8), and 500 mM NaCl; 4) 1% Nonidet P-40, 1% deoxycholic acid, 2 mM EDTA, 20 mM Tris-HCl (pH 8), and 250 mM LiCl; and twice with 5) 10 mM Tris-HCl (pH 8) and 1 mM EDTA. The DNA/protein/Ab complexes were eluted twice from the protein A-agarose slurry by heating at 65°C for 2 min and rocking for 15 min at room temperature in 250 μl 50 mM NaHCO3 and 1% SDS. The complexes were treated with 20 μg RNase A at 37°C for 1 h and 20 μg proteinase K at 45°C for 1 h, extracted once with phenol/chloroform and twice with chloroform, and precipitated with ethanol to purify the DNA. The DNA was resuspended in 30 μl water for subsequent PCR analysis as a template to evaluate the presence of Eα and Oct2 sequences. For semiquantitative PCR analysis, 4 μl Ab-bound DNA (equivalent to 13.3% of the total Ab-bound material) or 4 μl 1/100, 1/200, and 1/400 diluted input DNA (equivalent to 0.0033–0.0008% of the starting material used for each ChIP) was amplified using previously described primers (35). The PCR conditions were as follows: 5 min at 95°C, followed by 25 cycles of 25 s at 94°C, 25 s at 55°C, 50 s at 72°C, and a final extension step of 2 min at 72°C. The PCR products were resolved by 1.5% agarose gels, blotted, hybridized with 32P-labeled probes, and quantified using a phosphorimager (Storm 820; Molecular Dynamics). For quantitative PCR analysis, 3–5 μl Ab-bound DNA was amplified using LightCycler FastStart DNA Masterplus SYBR Green I (Roche) in a Roche LightCycler 1.5 thermocycler (Roche). The primer sequences used to amplify Eα are 5′-CCCTGAAATGGGTAAGCTGG-3′ and 5′-TGTTCAGACCCAAACACCTG-3′, and the primer sequences used to amplify Oct2 are 5′-CGGGTGTGAGAGGTGTGG-3′ and 5′-CGAGTCTGAAGCAAGCCAGT-3′. The PCR conditions were: 3 min at 95°C, followed by 55 cycles of 5 s at 95°C, 20 s at 55°C, and 1 s at 72°C. Paired sample t tests were used to determine the statistical significance between values. The p values are represented as follows: *p < 0.05, **p = 0.005–0.0005, and ***p < 0.0005.
Construction of reporter plasmids and luciferase assays
The Jα49-firefly luciferase (LUC), LUC-Eα, and Jα49–LUC-Eα (referred to as Jα49p, Eα386, and Jα49p-Eα386, respectively) pXPG plasmids have been previously described (19). The Eα440 fragment was obtained by PCR with the primers 5′-GTCTCCGAATTCAGACATTGAGTCAGTAGCC-3′ and the same 3′ primer previously used to clone Eα386 into pXPG (19), which includes the introduced EcoRI sites. After digestion with EcoRI, Eα440 was cloned into the unique EcoRI site downstream of LUC in pXPG and in Jα49p to generate LUC-Eα440 and Jα49p-LUC-Eα440 (in this study called Jα49p-Eα440). The TEA fragment was obtained by PCR with the primers 5′-CTGATGTCCCGGGCCCTACCTCTG-3′ (with an introduced SmaI site) and 5′-GACAGTCAAGCTTCTAGTGTCCTGTATTCATC-3′ (with an introduced HindIII site). To generate TEA-LUC (called TEA), TEA-LUC-Eα386 (called TEA-Eα386), and TEA-LUC-Eα440, the 673-bp SmaI-HindIII fragment was cloned into the unique SmaI and HindIII sites upstream of LUC in either pXPG. To generate the Vδ1 promoter (Vδ1p)-LUC, Vδ1p-LUC-Eα386, and Vδ1p-LUC-Eα440 (called Vδ1p, Vδ1p-Eα386, and Vδ1p-Eα440, respectively), the 1.6-kb Vδ1p fragment was excised from the Vδ1/CAT construct (36) by digestion with ScaI and HindIII and cloned into the ScaI and HindIII sites of pXPG or derivative plasmids. The structures of all the PCR products were confirmed by sequencing. The 3xNFAT (37) and 9xNFAT (38) have been previously described. The 3xNFAT-LUC contains three copies of the NFAT–AP-1 site of the murine IL-2 promoter (37), and the 9xNFAT-LUC contains nine NFAT sites from the IL-4 promoter (38). Jurkat cells (4 × 106) were transfected with 5 μg LUC reporter plasmid plus 10 ng pRL-TK (renilla luciferase plasmid; Promega) by electroporation in 300 μl ice-cold RPMI 1640 medium at 260 V, 80 Ohms, and 1500 μF in 0.4-cm gap BTX cuvettes. Five hundred microliters ice-cold inactivated FBS was added immediately to the cuvette, and the cells were kept for 15 min on ice, diluted in 3–5 ml complete medium in 25 cm2 flasks or six-well plates, and cultured at 37°C with 5% CO2 for 24 h. Afterwards, the cells were lysed, and luciferase activity (firefly/renilla) was measured with a Dual Luciferase Kit (Promega). Paired sample t tests were used to determine statistical significance between the values. The p values are represented as follows: *p < 0.05, **p = 0.005–0.0005, and ***p < 0.0005.
Both the calcineurin/Ca2+ and MAPK/ERK signaling pathways are required for induction of germline Tcra transcription
Tcra transcription and recombination depend on Eα activation by pre-TCR signaling. This enhancer acts through the binding of multiple transcription factors constitutively present in DN3a and DP thymocytes (Fig. 1). To study the molecular mechanisms for the induction of germline transcription at Tcra by pre-TCR signaling, we sought to establish an in vitro system that would allow us to analyze the proximal signaling events involved in this process. Scid.adh (TAC:CD3ε) cells (in this study denoted as Scid.adh cells) have been previously established as an excellent cellular system for studying the molecular mechanisms of pre-TCR signaling by stimulation through a chimeric human CD25:CD3ε surface protein (20). These cells resemble the phenotype of DN3a thymocytes and express pre-Tα transcripts. CD3ε-induced changes in these cells include the downregulation of Il2ra (CD25), Rag1, Rag2, and Pre-Ta transcripts and the upregulation of CD5, CD27, CD28, Egr-1, Egr2, Egr-3, and germline Tcra-Cα transcripts (9, 20). Consistent with previous data, stimulation of these cells with TAC Ab induced the reported changes, including the induction of Egr-1, Egr-3, and germline Tcra-Cα transcripts (Fig. 2A). These changes are identical to those induced in normal DN3a thymocytes during β-selection (1, 39).
In an attempt to dissect the signaling requirements for this induction during β-selection, we evaluated the total germline Tcra-Cα transcripts after cell stimulation with phorbol esters, such as PMA, and/or Ca2+ ionophores, such as ionomycin or thapsigargin (Fig. 2B). Germline Tcra-Cα transcription was activated by all three stimuli. Quantitative analysis of these transcripts induced by the different treatments revealed that Ca2+-mediated signaling was the major contributor to the activation of germline Tcra transcription, as Cα transcripts were induced much more robustly by ionomycin or thapsigargin treatment than by PMA treatment. As a control for treatment specificity, the induction of Egr-1 and Egr-3 transcripts was also analyzed (Fig. 2C). Egr-1 transcription was specifically induced by sustained protein kinase C activation by PMA treatment, whereas Egr-3 transcription was more dependent upon activation of Ca2+-mediated signaling as expected (40, 41). These data indicate that Ca2+-mediated signaling is by far the most relevant pathway involved in inducing Tcra germline transcription in Scid.adh cells. To investigate whether Ca2+-mediated signaling activates the Cα transcripts through a calcineurin-dependent pathway, we pretreated cells with CsA before stimulation (Fig. 2B, 2C). In agreement with previous studies that found that CsA completely inhibits TAC Ab-induced Egr-3 and Cα transcript induction in Scid.adh cells (11), we found that CsA totally blocked the induction of Egr-3 and germline Tcra-Cα transcripts in cells stimulated with ionomycin or thapsigargin. These results demonstrate that the calcineurin/NFAT signaling pathway is essential for activation of Tcra germline transcription.
To evaluate whether the MAPK/ERK signaling pathway is required in addition to calcineurin/NFAT for the induction of germline Tcra transcripts, we pretreated cells with specific inhibitors of MAPK/ERK- or Ca2+-mediated pathways before stimulating the cells with PMA and ionomycin (Fig. 2D). Chelation of extracellular Ca2+ with EGTA, which inhibits the entry of Ca2+ into the cell and the nuclear translocation of NFAT, totally abrogated the induction of germline Tcra-Cα transcripts. In addition to specific inhibitors of Ca2+-mediated signaling, pretreating the cells with PD98059 (MAPK inhibitor) and UO126 (ERK inhibitor), but not with the control UO124, also inhibited the induction of Cα transcripts, suggesting that transcription factors induced through the MAPK/ERK pathways are also involved in the activation of Tcra germline transcription. As expected, the PLCγ inhibitor U73122 did not inhibit germline Tcra-Cα transcripts because stimulation with PMA and ionomycin bypasses PLCγ activation. The induction of Tcra germline transcripts by independent MAPK/ERK or Ca2+ signaling pathways in Scid.adh cells, as seen in Fig. 2B, is most likely to be due to the bioavailability of low levels of endogenous NFAT or Egr factors in some cells before their activation (11). Altogether, our data indicate that both calcineurin/NFAT- and MAPK/ERK-mediated pathways are essential for induction of germline Tcra transcription during β-selection.
NFAT, AP-1, and Egr-1 factors bind specifically to multiple sites within Eα
Our functional data suggest an essential role for Ca2+/calcineurin-dependent transcription factors, such as NFAT, together with MAPK/ERK-dependent transcription factors, such as AP-1 and Egr-1, in the germline transcriptional induction of Tcra. By comparing the nucleotide sequences of the human and murine Eα with rVista and MatInspector programs (42–44), we identified 10 conserved putative sites for NFAT (GGA motifs with 3′-adenine tracts), 4 for AP-1 within the 5′ Tα1 and Tα1-Tα2 regions, and 7 for Egr factors within the 5′ Tα1, Tα1-Tα2, and 3′ Tα4 regions. The concentration of previously undetected NFAT, AP-1, and Egr binding sites present within Tα1-Tα2 and the flanking regions is very striking because Tα1-Tα2 constitutes the previously reported core Eα, which is extensively occupied by multiple transcription factors during thymocyte development (21, 26, 27)(Fig. 1, Supplemental Fig. 1, Supplemental Table I).
To validate the sites that can readily support NFAT binding, we performed EMSAs with Eα-derived probes and rNFAT-DBD (Supplemental Fig. 2A). Binding experiments to 5′ Tα1, Tα1, and Tα2 demonstrated highly efficient binding of rNFAT-DBD to the three regions (data not shown). The 5′ Tα1 can accommodate simultaneous binding of two NFAT molecules to the same DNA molecule (Supplemental Fig. 2A); however, further experiments using sites that can support the formation of NFAT dimers (45) demonstrated noncooperative NFAT binding to 5′ Tα1 (B. del Blanco and C. Hernández-Munain, unpublished observations). To clearly identify those sites that were actually bound to NFAT from among the potential sites, we performed EMSAs using Eα probes containing single binding sites (Supplemental Fig. 2A). Our dissection of the NFAT sites indicated that the relevant Eα NFAT sites in vivo might include the following four sites: 5′ Tα1 sites I and II, Tα1 site III, and Tα2 site IV (Fig. 3A). Furthermore, incubation of the single site-containing probes with increasing amounts of rNFAT-DBD allowed us to compare the relative affinity among the several binding sites (Supplemental Fig. 2A). Our data indicate that sites I and III can afford very strong NFAT binding, whereas sites II and IV can afford moderate NFAT binding.
To evaluate binding of endogenous NFAT factors to Eα during β-selection, we first generated nuclear extracts from unstimulated and PMA+ionomycin-stimulated cells as a source of such inducible factors. The efficient induction of NFAT1, NFAT2, NF-κB, AP-1, and Egr-1/3 factors after cell stimulation was evidenced by binding to their corresponding CSs in the EMSA experiments (Supplemental Fig. 2B and B. del Blanco and C. Hernández-Munain, unpublished observations). No induction of NFAT4 was detected in these experiments, which is consistent with the relative expression of NFAT factors in Scid.adh cells, given that NFAT1 and NFAT2 are far more abundant that NFAT4 in these cells (11). We then asked whether the endogenous NFAT factors present in PMA+ionomycin-stimulated cell extracts could bind to the NFAT sites present in the Eα-derived probes (Fig. 3B–D). Two specific complexes (marked NFAT1 and NFAT2) based on competition experiments were formed with each of the Eα probes used (data not shown). Inclusion of NFAT1 or NFAT2 Abs resulted in a shift of the correspondent complexes (marked as NFAT1* and NFAT2*, respectively). These experiments demonstrate that endogenous NFAT1 and NFAT2 can bind in vitro to each of their sites within Eα. It is interesting to note that Ets factors also bind to GGA core sequences (46). In fact, the 5′ Tα1 NFAT sites I and II coincide exactly with previously described binding sites for Ets-1 and Fli-1 (Fig. 1, Supplemental Fig. 1) (26, 47). Both NFAT and Ets factors require intact GG nucleotides within their binding sites in 5′ Tα1, indicating that these factors share the same contacts with DNA (B. del Blanco and C. Hernández-Munain, unpublished observations). To further discriminate between complexes containing NFAT and Ets-1 factors formed with the 5′ Tα1 probes, EMSAs in the presence of an Ets-1 Ab were performed (Fig. 3E). These experiments identified a complex containing Ets-1 for which formation was inhibited in the presence of the Ets-1 Ab. The Ets-1–containing complex can be easily discriminated from complexes containing NFAT1 or NFAT2 due to its increased mobility. These experiments revealed that in contrast to the induction of NFAT-containing complexes, the formation of the Ets-1–containing complex was inhibited upon cell stimulation. These results are consistent with a previously reported Ca2+-dependent and protein kinase C-mediated phosphorylation of Ets-1, which inhibits its binding to DNA (48).
We next analyzed the binding of AP-1 to its putative Eα sites. We asked whether AP-1 complexes present in the nuclear extracts of stimulated cells could bind to the putative Eα AP-1 sites. Our experiments showed that only one site present in Tα1 was bound efficiently by AP-1 (Fig. 3A, 3F, 3G, Supplemental Fig. 2C). This complex (marked AP-1) was judged to be specific based on competition experiments (Supplemental Fig. 2C). Inclusion of Jun Abs inhibited complex formation, confirming the identity of this complex (Fig. 3F). The JunB Ab was more efficient than JunD and c-Jun Abs in inhibiting the formation of this complex (Fig. 3F and data not shown). These results are consistent with the relative expression of Jun proteins upon Scid.adh cell stimulation: JunB transcripts are more abundant than JunD and c-Jun transcripts in stimulated cells (B. del Blanco and C. Hernández-Munain, unpublished observations). This site coincides with a previously described essential site for CREB (24). In agreement with this, we have detected a CREB-containing complex in nonstimulated and stimulated cells, which was supershifted in the presence of a CREB Ab (Fig. 3F). CREB is mostly present in an unphosphorylated form in resting thymocytes, and its phosphorylation is induced by different pharmacological stimuli that activate different signaling pathways resembling TCR engagement (49). Consistent with this, we have detected a p-CREB–containing complex bound to this sequence (p-CREB*) in stimulated cells (Fig. 3G) as previously reported (22, 26). Interestingly, the abundance of the induced AP-1 complex exceeds that of the p-CREB* complex, suggesting the possibility that AP-1 might outcompete p-CREB for binding to Tα1.
To verify whether the putative Egr binding sites present at Eα can support Egr-1 binding, we performed EMSAs using Eα-derived probes. Because the use of nuclear extracts did not allow us to clearly discriminate the specific binding of Egr factors to these probes, we performed these experiments with rEgr-1 (Fig. 3H) and extracts from HEK-293T cells overexpressing Egr-1 (Supplemental Fig. 2D, 2E). Analysis of rEgr-1 binding to 5′ Tα1, Tα1-Tα2, and 3′ Tα4 revealed the formation of complexes of identical mobility to the complex formed with the CS (data not shown). To verify that Egr-1 is present in these complexes, we used an Egr-1 Ab (Fig. 3H). Inclusion this Ab resulted in a supershift of the Egr-1–containing complex (Egr-1*). As expected, we did not observe any binding of rEgr-1 to Tα3-Tα4 demonstrating the specificity of the assays (B. del Blanco and C. Hernández-Munain, unpublished observations). Our dissection of the Egr binding sites present in Eα indicated that the relevant sites in vivo might include five sites: 5′ Tα1 site I, Tα1 site II, Tα2 sites III and IV, and 3′ Tα4 site V (Fig. 3A, Supplemental Figs. 1, 2D, 2E). Our comparison of Egr-1 binding to Eα probes containing a single Egr-binding site with respect to that to the CS allowed us to compare the relative affinities of the different Egr binding sites (Supplemental Fig. 2E). Our data indicate that Egr-1 has a very low affinity for binding to the Egr binding sites present in Eα. Among the five confirmed Egr binding sites, only the Tα1 site II affords the strongest Egr-1 binding to Eα, and it represents 11.725 ± 1.421% of the binding of this factor compared with that of the CS. The capability of Egr-1 to bind to the other Eα Egr binding sites is much weaker compared with Egr site II, and each of them represents <1% of the binding of Egr-1 to the CS. The percent of Egr-1 binding to these individual sites versus to the CS is as follows: 3′ Tα4 Egr site V (0.820 ± 0.139%), Tα2 Egr site III (0.203 ± 0.035%), 5′ Tα1 Egr site I (0.079 ± 0.016%), and Tα2 Egr site IV (0.044 ± 0.008%). We conclude that Egr-1 can bind in vitro specifically to several sites present in 5′ Tα1, Tα1-Tα2, and 3′ Tα4 with low affinity (Tα1 Egr site II) or very low affinity (5′ Tα1 site I, Tα2 sites III and IV, and 3′ Tα4 site V). These results suggest that the Tα1 Egr site II might be the only relevant Egr binding site present in Eα. Taken together, our data suggest that NFAT, AP-1, and Egr proteins are inducible factors that might function in the activation of Eα during β-selection. A summary of the confirmed binding sites for these factors present at Eα is presented in the Fig. 3A and Supplemental Fig. 1.
NFAT, AP-1, and Egr-1 factors bind transiently to Eα to assemble an active enhanceosome, together with GATA-3, E47, Ets-1, CREB, and CREB-binding protein/p300 factors, during thymocyte development
To confirm the presence of NFAT factors, AP-1, and Egr-1 at the Eα enhanceosome in vivo in PMA+ionomycin-stimulated versus unstimulated Scid.adh cells, we performed ChIP experiments and analyzed them by quantitative PCR (Fig. 4) or semiquantitative PCR (Supplemental Fig. 3). As a control for gene specificity, factor binding to a presumably negative control sequence present in the Oct2 gene (35) was analyzed in all experiments.
GATA-3 was originally identified as a positive regulator of Eα (50), which is required for β-selection (51). We have previously shown that it participates in the formation of the Eα enhanceosome assembled in vitro (22) and in vivo (27). Because the expression of GATA-3 is upregulated in thymocytes by TCR signaling as a result of the additive inputs from the Ras/MAPK and calcineurin pathways (52, 53) and because binding of GATA-3 to Eα depends on its bioavailability in cell nuclei (54), we decided to evaluate possible changes in GATA-3 binding to Eα in unstimulated and stimulated Scid.adh cells. As is shown in Fig. 4A, GATA-3 binding to Eα was induced by ∼3-fold in stimulated versus unstimulated cells. These results are consistent with previous reports of GATA-3 inducibility upon cell stimulation and suggest that this factor might have an important role in the assembly of a functional Eα enhanceosome.
We next analyzed binding of NFAT, AP-1, and Egr-1 factors to Eα upon cell stimulation (Fig. 4B–D). As shown in Fig. 4B, NFAT2 binding to Eα was induced by ∼5-fold in stimulated cells but not in unstimulated cells. In fact, NFAT2 was not present in the Eα enhanceosome in unstimulated cells. Our analysis of NFAT4 binding to Eα revealed that the NFAT4 binding pattern parallels that of NFAT2 upon cell stimulation (Supplemental Fig. 3A), whereas we did not find clear evidence of NFAT1 binding to Eα (data not shown). These results suggest that NFAT2 and NFAT4 are the relevant NFAT factors present on an active Eα enhanceosome.
Analysis of the AP-1 components revealed that the binding of JunB and JunD to Eα was induced upon cell stimulation (Fig. 4C), which involved significant differences in their binding to the enhancer in unstimulated and stimulated cells. Analysis of c-Jun binding also highlighted an increase in c-Jun binding to Eα upon cell stimulation with binding kinetics that are initiated 3.5 h after cell stimulation and increase for 6 h after cell stimulation (Supplemental Fig. 3B). These results suggest AP-1 complexes containing different Jun components can all be part of an active Eα enhanceosome.
Analysis of the presence of Egr-1 on the Eα demonstrated that binding of this factor was dramatically increased upon cell stimulation (Fig. 4D). Egr-1 Ab binding was enriched by ∼16-fold with respect to the control Ab, and its binding was induced by ∼6-fold in stimulated cells compared with unstimulated cells. No clear binding of Egr-3 to Eα was detected, probably due to its low expression compared with Egr-1 in these cells (data not shown).
Furthermore, consistent with the dependence of the proper Eα-mediated functional activation of both transcription and V(D)J recombination on the recruitment of histone acetylases (HATs) (27, 35), we found that the levels of p300 and CREB-binding protein (CBP) in the Eα enhanceosome were induced upon cell stimulation (Fig. 4E and data not shown). Binding of p300 to Eα was increased by ∼3-fold in stimulated cells compared with unstimulated cells. These results are consistent with the fact that the binding of p300 constitutes a common marker of activated enhancers (55).
All transcription factors assayed exhibit increased binding to Eα in stimulated versus unstimulated cells. To formally exclude potential biases with regard to specific Ab binding in stimulated cells, we analyzed the behavior of a transcription factor for which binding does not change before and after early stimulation. For that purpose, we performed ChIP experiments analyzing the binding of E-protein E47 to Eα (Fig. 4F), which is present at the Eα enhanceosome in DP thymocytes (27). Our ChIP experiments indicated that the binding of E47 to Eα is comparable in unstimulated and stimulated cells (Fig. 4F). Although E2A-binding site occupancy generally decreases during β-selection, a recent genome-wide analysis of E2A occupancy revealed that 247 out of 939 of those sites are shared in DN3a and DN4 cells, suggesting that occupancy of ∼25% of E-boxes is not altered by pre-TCR signaling (56). Similar data were obtained in regard to CREB binding; there were no obvious differences in binding between unstimulated and stimulated cells (Fig. 4H, see below). These results validate our ChIP data and demonstrate the specific recruitment of inducible transcription factors to Eα upon DN3a cell stimulation.
Because our in vitro binding data suggest that the binding of Ets-1 to 5′ Tα1 might be outcompeted by the binding of NFAT factors and that the binding of CREB to Tα1 might be outcompeted by the binding of AP-1 factors upon cell stimulation, we evaluated these possibilities by performing ChIPs to analyze Ets-1 and CREB binding to Eα before and after cell stimulation (Fig. 4G, 4H). In agreement with our previous analysis that demonstrated the presence of both Ets-1 and CREB in the Eα enhanceosome assembled in DP thymocytes in vivo (27), we confirmed the binding of these factors to Eα in both unstimulated and stimulated Scid.adh cells. In contrast to our expectations, we found that Ets-1 binding was increased by ∼3-fold upon cell stimulation (Fig. 4G), whereas CREB binding was unaffected (Fig. 4H). Hence, the results obtained from the ChIP analysis of Ets-1 argue against competition between Ets-1 and NFAT for binding to the same sequences at 5′ Tα1. These data suggest that the relevant NFAT sites present in Eα might be 5′ Tα1 site II, Tα1 site III, and Tα2 site IV, which do not coincide with reported Ets binding sites (Supplemental Fig. 1). In the case of CREB, it is relevant to note that CREB can also bind in vitro to an additional ATF/CRE site present in 5′ Tα1 (B. del Blanco and C. Hernández-Munain, unpublished observations) (Fig. 1, Supplemental Fig. 1).
Taken together, our ChIP experiments support our hypothesis that an active Eα enhanceosome is assembled de novo at the transition from DN3a to DN4 by the binding of inducible factors, including NFAT2 and NFAT4, AP-1 (JunB, JunD, and c-Jun), and Egr-1 proteins, to multiple sites within the enhancer. Furthermore, these results suggest that CBP/p300 recruitment to the Eα enhanceosome may be important for its activation and for long-range Eα-dependent histone acetylation (35).
To evaluate whether the results that we have obtained in Scid.adh cells can be extended also to primary thymocytes, we first compared germline Tcra Jα-initiated transcripts present in PMA+ionomycin-stimulated Scid.adh cells to those in thymocytes from Rxβ mice (Supplemental Fig. 3C, 3D). We found abundant germline Tcra transcripts in both cases, suggesting that active transcription is occurring at this locus in each case. To evaluate whether continuous signaling through the pre-TCR is a requirement for the induction of germline Tcra-Cα transcription in DP thymocytes, we analyzed the effect of the PLCγ inhibitor U73122 on germline Tcra-Cα transcription in Rxβ thymocytes by quantitative RT-PCR (Fig. 5A). We found that germline Tcra-Cα transcription was diminished by 50% in Rxβ thymocytes after a 24-h in vitro treatment, indicating that constant PLCγ-mediated signaling by the pre-TCR is required to maintain the levels of germline Tcra transcription in thymocytes. Treatment with other inhibitors did not have any effect because they need to be added before the specifically targeted signaling pathway was initiated (57, 58).
To evaluate the presence of inducible factors at the Eα enhanceosome during the DN3a to DP thymocyte transition, we performed ChIP experiments in Rag2−/− and Rxβ thymocytes and analyzed them by quantitative (Fig. 5B–G) or semiquantitative PCR (Supplemental Fig. 3E). During pre-TCR signaling, two temporally distinct periods can be distinguished: an initial proliferative phase and a secondary nonproliferative phase (39, 59). During the first phase, the rapid and transient expression of Egr factors triggers the expression of Id3, which prevents E12/E47-dependent transcription (11, 39, 60, 61). During the second phase, Egr activity declines, and E-protein activity increases (39). Essentially all Rxβ thymocytes are blocked at the small resting DP stage due to their inability to rearrange the Tcra locus, which is equivalent to thymocytes that have reached the antiproliferative phase of pre-TCR signaling. In fact, Id3 transcripts are strongly diminished in Rxβ thymocytes compared with Rag2−/− thymocytes (B. del Blanco and C. Hernández-Munain, unpublished observations). Consistent with this notion, the in vivo binding of E47 to Eα in Rxβ thymocytes was robust and increased by ∼5-fold with respect to the levels in Rag2−/− thymocytes (Fig. 5B). Our ChIP analyses demonstrated no significative differences in NFAT1 and NFAT2 binding to Eα in Rxβ thymocytes compared with that in Rag2−/− thymocytes (Fig. 5C, 5D). In the analysis of AP-1 binding, we did not detect a significant induction of JunD, JunB, or c-Jun binding in Rxβ thymocytes compared with in Rag2−/− thymocytes (Fig. 5E and data not shown). Similarly, there was no apparent increase in Egr-1 binding in Rxβ versus Rag2−/− thymocytes (Fig. 5F). However, in vitro stimulation of Rxβ thymocytes induced the further binding of Egr-1 to Eα (B. del Blanco and C. Hernández-Munain, unpublished observations); this is probably related to a transient induction in synchronously stimulated cells (40). These results are consistent with a possible transient recruitment of inducible factors to Eα in stages prior to small resting DP thymocytes. In contrast, we found a strong increase in Ets-1 binding to Eα in Rag2−/− versus Rxβ thymocytes (Fig. 5G). These results are in agreement with the correlation between the occupancy of E2A sites and the enrichment of Ets-1 binding sites reported recently in a genome-wide analysis during β-selection (56), supporting a collaborative functional interaction between these factors. In agreement with the data obtained in Scid.adh cells, we found that pre-TCR signaling stimulated the recruitment of coactivators, such as p300 and CBP, to the Eα enhanceosome, as evidenced by their stronger binding in Rxβ thymocytes than in Rag2−/− thymocytes (Supplemental Fig. 3E and data not shown). These results are consistent with the requirement for Eα to recruit HATs to induce V(D)J recombination (35).
To evaluate whether the formation of an active Eα enhanceosome containing inducible transcription factors is indeed transient during DN3a to DP thymocyte differentiation, we performed ChIP experiments using thymocytes from Rag2−/− mice that were induced to differentiate to DN4 in response to CD3 stimulation by i.v. injections with CD3ε mAb. CD3ε mAb injection of Rag2−/− mice mimics pre-TCR signaling because it stimulates the proliferation and differentiation of DN3a to DP thymocytes and promotes allele exclusion at the Tcrb locus and transcription at the Tcra locus (62–64). Previous experiments have demonstrated that at 12 h after CD3ε mAb injection, the DN3a thymocytes begin to differentiate into the DN4 stage, and at 24 h after the CD3ε mAb injection, approximately half of the DN3a thymocytes have differentiated into DN4 thymocytes (4, 60). We have analyzed the factor binding to Eα in vivo using thymocytes isolated from Rag2−/− mice 16 h after CD3ε mAb injection. We found that the binding of JunD was dramatically increased in thymocytes from injected Rag2−/− mice compared with control Rag2−/− mice or Rxβ mice (Fig. 5I). Consistent with the results obtained in Scid.adh cells, E-47 binding was not affected upon in vivo Rag2−/− thymocyte stimulation (Fig. 5H). These results confirm our hypothesis that the assembly of an active Eα enhanceosome is transient during thymocyte development.
Eα-transcriptional induction is dependent on calcineurin-mediated signaling
All of our data indicate that Tcra germline transcription is dependent on signaling pathways mediated by calcineurin and MAPK/ERK, which culminate in the binding of NFAT2/4, AP-1, and Egr-1 factors to the Eα enhanceosome after β-selection. To formally investigate whether Eα is the element through which these signaling pathways activate germline Tcra transcription, we evaluated Eα inducibility after Jurkat cell stimulation in transient transfection experiments (Fig. 6). In addition, we also evaluated the functional contribution of NFAT sites I and II and Egr site I within 5′ Tα1 on Eα activation. To this end, we employed reporter constructs in which Eα-dependent TEA or Jα49p is controlled by Eα variants containing (Eα440) or lacking (Eα386) these sites (Fig. 6A). We note that these types of experiments, with Eα positioned within the context of artificial reporter constructs, allow the analysis of Eα function in the absence of cell stimulation and therefore do not recapitulate the in vivo situation, in which transcriptional activation of Tcra, with Eα positioned at large distances from its responding promoters, is cell stimulation dependent. However, this in vitro system allowed us to evaluate some essential aspects of Eα inducibility. Transient transfection of these constructs revealed that, consistent with previous results (19, 65), in the absence of Eα, neither TEA nor Jα49p was able to activate the transcription of the reporter gene, whereas Eα efficiently activated transcription from each of these promoters (Fig. 6B). The high basal transcriptional activity observed in the Jα49p/Eα386 construct compared with that in the Jα49p/Eα440 construct supports the presence of a negative regulatory site within 5′ Tα1 involved in this collaboration. The opposite situation was found in the comparison between the TEA/Eα386 and TEA/Eα440 constructs, indicating that Eα440 was a more efficient transcriptional activator of transcripts derived from TEA than Eα386. Cell stimulation with PMA+ionomycin further activated Eα440- and Eα386-dependent transcription from both promoters, demonstrating Eα inducibility in this system (Fig. 6B). It is interesting to note that both the Jα49p/Eα386 and Jα49p/Eα440 constructs had the same transcriptional activity, suggesting that the 5′ Tα1 NFAT sites I and II are dispensable for enhancer induction driven by Jα49p. This result is consistent with the lack of any observed effect on transcription driven by either Jα49p or TEA after the introduction of a point mutation at the 5′ Tα1 NFAT site I, the major NFAT site in the 5′ Tα1 region, that affects NFAT binding without disturbing Ets-1 binding (GGAA to GGTT instead of GGAA to TTAA) (data not shown). In contrast to data obtained from the Jα49p/Eα constructs, when using the TEA/Eα constructs, the Eα440-dependent transcriptional induction was twice as high as the Eα386-dependent transcriptional induction. The fact that Eα440 was also a more efficient transcriptional activator for transcripts derived from TEA than Eα386 in unstimulated cells suggests that the binding of noninducible factors might be involved in the functional collaboration between TEA and the 5′ Tα1 region. As a control for transcriptional induction upon cell stimulation, the activities of two reporter constructs containing several NFAT sites in tandem, 3xNFAT and 9xNFAT (37, 38), were analyzed (Fig. 6B). The transcriptional activity of both positive control constructs was efficiently induced upon cell stimulation (Fig. 6B). Pretreatment of cells with CsA before stimulation totally abrogated the ability of both Eα440 and Eα386 to activate cell stimulation-dependent transcription (Fig. 6B), suggesting a specific role for calcineurin and NFAT factors in this stimulation. The transcriptional activity of both the 3xNFAT and 9xNFAT control constructs was efficiently inhibited upon CsA treatment (Fig. 6B). Because it is possible that cell stimulation might also be involved in TEA and Jα49p function, we assessed whether Eα is specifically induced by cell activation. To address this possibility, Eα386- and Eα440-dependent transcriptional induction was assayed from the Vδ1p, which is an active promoter in DN3a cells prior to β-selection (Fig. 6C). Vδ1p alone or directed by Eδ or Eα386 did not respond to cell stimulation, whereas it strongly responded when directed by Eα440 (Fig. 6C and B. del Blanco and C. Hernández-Munain, unpublished observations). These results indicate that Eα itself responds to cell stimulation and that calcineurin-mediated signals are crucially involved in this function. Altogether, our data indicate that pre-TCR signals induce Tcra germline transcription through calcineurin- and MAPK/ERK-dependent signaling pathways that activate Eα through the assembly of a functional Eα enhanceosome formed by multiple inducible factors.
Previous experiments have established Tα1-Tα2 as a paradigm of an enhanceosome created by stereospecific interactions among activators bound to the enhancer (24, 25). This view has also been supported by previous genomic footprinting experiments using DMS that revealed that Tα1–Tα4 and the flanking areas are extensively occupied without major differences between DN3a and DP thymocytes from Rag2−/− and Rxβ mice, respectively (21, 26, 27). However, this notion has been challenged by the creation of a mutant version of Eα, EαMC, in which Eδ Myb and Runx binding sites were substituted for the Tα2 Runx and Ets binding sites (27). EαMC was a highly potent enhancer, indicating that the stereospecific interactions among proteins that form an Eα enhanceosome are rather flexible. These experiments suggested the possibility of the assembly of distinct sets of proteins on Eα that might represent a more flexible form of information processing during thymocyte development. In support of this hypothesis, analysis of DNase I sensitivity at the nucleotide resolution revealed that Eα chromatin is generally much more sensitive to digestion in DN3a than in DP thymocytes from Rag2−/− and Rxβ mice, respectively (22). This suggests a more compact enhanceosome structure in the latter cell population due to the assembly of a different multiprotein complex or as a consequence of physical interactions between Eα and its associated promoters, such as TEA (13, 66). In addition to these general changes in the Eα DNA structure, striking differences were also found at specific nucleotides, including the 5′ Tα1 and Tα1-Tα2 regions (26), suggesting that a new complex is formed in these regions by direct contact with Eα or be indirectly recruited through enhancer/promoter interactions. Therefore, Eα undergoes general chromatin structural changes at the DN3a to DP transition at specific regions that coincide with the regions where the NFAT, AP-1, and Egr-1 factors bind, suggesting that a new complex containing these inducible factors is formed at the DN4/early proliferating DP Eα enhanceosome to establish specific contacts with specific promoters. Based on these observations, together with the locations of the NFAT, AP-1, and Egr-1 factor binding sites described in this study, the requirement for an intact Tα2 for the assembly of an Eα enhanceosome in vivo (25), and our observation that in vivo Ets-1 binding seems not to be outcompeted by NFAT binding, we favor a scenario in which these inducible factors occupy their sites in DN3b, DN4, and early proliferating DP thymocytes, without disturbing the general nucleoprotein structure established by the binding of LEF-1/TCF-1-Runx1 and Ets-1 to Tα2 and of the Ets-1/Fli-1 factors to 5′ Tα1 in DN3a thymocytes (Fig. 7). In fact, the complex formed by the factors bound to Tα2 seems to be very stable based on in vitro (24) and in vivo studies (25, 47) because the binding sites at Tα2 facilitate cooperative protein binding to DNA (24, 25, 67). We propose that the binding of these lymphoid-restricted factors to Eα prior to its activation in DN3a thymocytes would constitute a regulatory landscape for recruiting factors induced by pre-TCR signaling (Fig. 7). In DN3b/DN4/early proliferating DP thymocytes, the binding of inducible transcription factors, such as NFAT, AP-1, and Egr-1, together with GATA-3, Ets-1, CREB, and E-proteins, to Eα in cells results in the recruitment of CBP/p300 coactivators to assemble an active enhanceosome. This enhanceosome is in turn able to activate TEA and Jα49p by intrachromosomal Eα/Jα physical promoter interactions to trigger first Tcra germline transcription and then primary Vα to Jα rearrangements (13, 66). Interestingly, specific interactions between inducible factors have also been demonstrated to be involved in other specific inducible enhancer/promoter interactions (68–70). In the later stages of β-selection, these inducible factors are not present in small resting DP nuclei (39, 71, 72), and the Eα enhanceosome composition presumably changes to an active DN3a-like state (Fig. 7). The active DP Eα enhanceosome differs from the inactive DN3a Eα enhanceosome by enhanced binding of Ets-1 to Tα2 and E-47 to the E-box III and the recruitment of HATs. The assembly of different transient pre- and post–β-selection enhanceosomes in DN3a, DN3b/DN4/early proliferating DP cells, and small resting DP cells (Fig. 7) provides a new explanation for the long-standing paradox regarding the activation of this enhancer without obvious differences in the footprint of bound proteins by DMS between the DN3a and small resting DP cells but accompanied by general changes in Eα chromatin structure (21, 22, 26, 27).
Enhanced binding of Ets-1 to Tα2 in DP thymocytes with respect to that in DN3a thymocytes is supported by the higher sensitivity of Tα2 chromatin to DNase I digestion detected in DN3a thymocytes and by an increased stability of the Ets-1/Runx-containing complex formed in vitro observed with Tα2- and DP-derived nuclear extracts (22, 26). These data suggest an increased stability in the formation of the DP-derived complex compared with that of the DN3a-derived complex in Tα2. In addition to this, it is possible that enhanced Ets-1 binding might be related with induced binding to specific GGA sequences present at the 5′ Tα1. Further experiments are required to clearly establish the differential occupancy of Eα E-boxes and 5′ Tα1 GGA sequences during thymocyte development.
The scenario proposed in this study is supported by the description of two distinguishable temporally distinct periods during pre-TCR signaling (39, 59). During the first 36-h period, or proliferative phase, defined by DN3b, DN4, and the early proliferating DP stages, there is a rapid activation of NFAT and AP-1 factors that drives the transient expression of Egr factors. This triggers expression of Id3 and prevents the induction of the E12/E47-dependent transcription factor retinoic acid-related orphan receptor γt and controls the expression of the antiproliferative gene mCPEB4 (11, 39, 60, 61). During the second 36–96-h period, or nonproliferative phase, defined in the small resting DP stage, the activity of the Egr factors declines, whereas E-protein activity increases and drives the induction of retinoic acid-related orphan receptor γt expression, triggering cell quiescence and the activation of Tcra locus rearrangement (39). Essentially all Rxβ thymocytes are blocked at the small resting DP stage, which is equivalent to thymocytes that have reached the nonproliferative phase. Consistent with this notion, the in vivo binding of E47 to Eα in Rxβ thymocytes was strong, whereas binding was not detected in Rag2−/− thymocytes. The reason why E47 binding to Eα is not detected in Rag2−/− thymocytes is unclear at present, but it might be related to the fact that different combinations of E-proteins seem to occupy the different E-boxes present in Eα (27). Eα contains three E-boxes: 5′ Tα1 E-box-I, Tα3 E-box II, and Tα4 E-box-III. The pattern of DMS in vivo footprints found that the Tα4 E-box III was affected when HEB−/− thymocytes were compared with Rxβ thymocytes (27). In fact, the pattern of occupation of this E-box observed in HEB−/− thymocytes resembled that of Rag2−/− thymocytes, with no occupation of the sequence (27). Together, these data suggest that the Tα4 E-box III might be occupied by an HEB/E47 heterodimer, which is the primary E-protein complex found in thymocytes (73, 74), whereas the 5′ Tα1 E-box-I and Tα3 E-box II might be occupied by dimers of E12 and/or E2-2 proteins (75, 76). Alternatively, it is also possible that E47 occupancy of Eα is induced during the DN3a to DP transition, as it has been shown that this occurs in ∼10% of E2A-occupied sites when comparing DN3a and DN4 cells by genome-wide analysis of E2A occupancy (56). Furthermore, this study has also revealed that the occupancy of E-boxes by E2A proteins is accompanied by an enrichment of binding sites for Runx-1 and Ets-1, as it is the case for Eα. Because we detected simultaneous robust recruitment of both E47 and Ets-1, our data suggest that collaborative interactions between these factors and presumably Runx-1 might be of particular interest for Eα function in small resting DP thymocytes.
Our model is supported by the fact that most Fos and Jun components, NFAT proteins, and Egr factors decreased dramatically in the transition from DN3b/DN4 (stimulated Scid.adh cells) to small resting DP thymocytes (Rxβ thymocytes) (39, 71, 72). Furthermore, the Scid.adh model allows the synchronous induction of signaling in DN3a-like thymocytes, allowing a clear detection of the transient interactions that occur in the DN3a/DN4/early proliferating DP transition that might be missed in Rxβ cells. Supporting our hypothesis that the assembly of an active Eα enhanceosome is highly transient during DN3a to DN4 thymocyte development, we found that: 1) in vivo c-Jun and Egr-1 binding to Eα increase upon Scid.adh cell and Rxβ thymocytes stimulation, respectively (Supplemental Fig. 3B and B. del Blanco and C. Hernández-Munain, unpublished observations), presumably due to the transienttranscriptional induction of c-jun and Egr-1 in synchronously stimulated cells (40); and 2) the binding of JunD is dramatically increased in thymocytes from injected Rag2−/− mice compared with noninjected Rag2−/− mice or Rxβ mice (Fig. 5I). Therefore, AP-1, NFAT, and Egr-1 factor recruitment to Eα seems to be highly transient during the DN3a to DN3b/DN4/early proliferating DP transition (Fig. 7). Although our ChIP experiments in Scid.adh suggest that AP-1 complexes containing different Jun components can all be part of an active Eα enhanceosome (Figs. 4C, 5F, Supplemental Fig. 3B), our data obtained both in Scid.adh cells and thymocytes indicate that the main Jun component of these complexes is JunD. These results contrast with the relative abundance of Jun upon Scid.adh cell stimulation, as JunB is more abundant than JunD and c-Jun in stimulated cells (Fig. 3F). Similarly, our ChIP data obtained both in Scid.adh cells and thymocytes indicate that the main NFAT components correlating with active Eα enhanceosomes in vivo are NFAT2 and NFAT4, despite the fact that NFAT1 seems to be the most abundant NFAT component in these cells (11). These data indicate that a pressure must exist for the binding of JunD-containing AP-1 complexes, together with NFAT2 or NFAT4, to Eα to form an active multiprotein complex on Eα in DN3b/DN4/early proliferating DP thymocytes.
Although our Egr-1 ChIP experiments indicate strong binding to Eα (Figs. 4D, 5E), our in vitro data indicate that the binding of Egr-1 to Eα is very weak compared with the binding to CS. The strongest Egr binding site located within the Eα (Tα1 site II) exhibited only ∼10% of the binding of Egr-1 to CS (Supplemental Fig. 2E). The presence of adjacent efficient binding sites for NFAT within the 5′ Tα1 and Tα1-Tα2 regions might compensate in vivo for weak Egr-1 binding because both factors could interact to form complexes with additional proteins to cooperate in regulating gene expression (77). In fact, NFAT sites I and II are immediately adjacent to the Egr binding site present within 5′ Tα1 (Egr site I), Tα1 NFAT site III overlaps with Egr site II, and the Tα2 Egr site III is located between NFAT sites III and IV. Alternatively, Egr-1 might be recruited to Eα by binding to Jα promoters via a process that loops out of the intervening DNA and creates a putative Jα promoter/Eα holocomplex (66), similar to the described interaction between the Dβ1 promoter and Eβ (78, 79). Further dissection of the relevant Egr binding sites within Eα and their collaboration with adjacent NFAT sites, as well as the possible contributions of putative Egr binding sites within the Jα promoters, in the formation of an active Eα enhanceosome will be addressed in future experiments.
Our analysis supporting the requirement for inducible factors for the activation of Eα has important novel implications about the functional regulation of this enhancer during T-lymphocyte development. Although our study has been limited to an analysis of the transcriptional activity of Eα, not its recombinational activity, as an indicator of enhancer function, germline Tcra transcription has an essential role in activating VαJα recombination (12–15). Our data strongly support the idea that Eα activity during T cell development is regulated through the ordered assembly of different enhanceosomes with an essential role in the induction of germline Tcra transcription to drive the activation of Vα-Jα rearrangements in DN4/DP thymocytes. This model for activation-dependent Eα function implies that its contributions might be dispensable for expression of a rearranged Tcra in resting naive T cells, as suggested in transgenic reporter constructs (80). Similarly, the different modes of Tcrb enhancer function have suggested the assembly of different enhanceosomes in early thymocytes to trigger initial Dβ-Jβ recombination and in peripheral T cells for the expression of the assembled Tcrb (81). In addition, our data prove the important role of Eα-binding inducible transcription factors in instructively regulating the choice between αβ and γδ T cell lineages through the activation of Tcra rearrangement and Tcrd recombinational deletion during β-selection. This molecular mechanism for Eα activation might be involved in the pre-TCR signaling-dependent repression of the γδ lineage (82). Future efforts will be focused on the molecular mechanisms controlling Eα activity during T lymphocyte maturation.
The authors have no financial conflicts of interest.
We thank T.A. Waldman for the hd245/332 αTAC hybridoma, F. Macián for the rNFAT1-DBD, J.M. Redondo for the 9xNFAT and 3xNFAT LUC reporter plasmids, M.S. Krangel for the pJα49p and Jα49p/Eα386 pXPG constructs, A.M. Ibáñez for generation of the Vδ1p-LUC plasmid, and C. Suñé and M.S. Krangel for critically reading the manuscript. We also thank members of the laboratories of C.H.-M. and C. Suñé for helpful discussions.
This work was supported by Spanish Ministry of Science and Innovation Grant BFU2009-08796, Junta de Andalucía Grants CTS-6587 and CVI-4526, and Consejo Superior de Investigaciones Científicas Grant 201020E060. This work was partly supported by the European Regional Development Fund. B.d.B. was partly supported by a predoctoral grant for formation of research personnel (FPI program) from the Spanish Ministry of Science and Innovation (BFU2005-01715/BCM).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- CREB-binding protein
- chromatin immunoprecipitation
- consensus binding site
- cyclosporine A
- Tcra enhancer
- Tcrd enhancer
- early growth response factor
- histone acetylase
- Jα49 promoter
- firefly luciferase
- phospholipase Cγ
- recombinant early growth response factor
- rDNA-binding domains of NFAT1
- Rag2−/− × Tcrb
- Scid.adh (TAC:CD3ε)
- T early-α promoter
- Vδ1 promoter.
- Received January 31, 2011.
- Accepted January 17, 2012.
- Copyright © 2012 by The American Association of Immunologists, Inc.