Growing evidence that the CD3γ gene is specifically targeted in some T cell diseases focused our attention on the need to identify and characterize the elusive elements involved in CD3γ transcriptional control. In this study, we show that while the human CD3γ and CD3δ genes are oriented head-to-head and separated by only 1.6 kb, the CD3γ gene is transcribed from an independent but weak, lymphoid-specific TATA-less proximal promoter. Using RNA ligase-mediated rapid amplification of cDNA ends, we demonstrate that a cluster of transcription initiation sites is present in the vicinity of the primary core promoter, and the major start site is situated in a classical initiator sequence. A GT box immediately upstream of the initiator binds Sp family proteins and the general transcription machinery, with the activity of these adjacent elements enhanced by the presence of a second GC box 10 nt further upstream. The primary core promoter is limited to a sequence that extends upstream to −15 and contains the initiator and GT box. An identical GT box located ∼50 nt from the initiator functions as a weak secondary core promoter and likely generates transcripts originating upstream from the +1. Finally, we show that two previously identified NFAT motifs in the proximal promoter positively (NFATγ1) or negatively (NFATγ1 and NFATγ2) regulate expression of the human CD3γ gene by their differential binding of NFATc1 plus NF-κB p50 or NFATc2 containing complexes, respectively. These data elucidate some of the mechanisms controlling expression of the CD3γ gene as a step toward furthering our understanding of how its transcription is targeted in human disease.
Dynamic regulation of TCR/CD3 complexes on the cell surface plays a critical role in controlling Ag-dependent T cell-mediated immune responses. This multisubunit receptor is composed of six different transmembrane chains: the Ag recognition components TCRα and TCRβ (or TCRγ and TCRδ in <10% of T cells) and the signal transduction module containing CD3γ, CD3δ, CD3ε, and CD3ζ. A defect or mutation affecting the expression of a single chain is sufficient to inhibit surface expression (1). Defective TCR/CD3 expression and function have been increasingly reported for a number of clinical conditions, including tumor-infiltrating T cells from a wide variety of cancers (recently reviewed in Ref. 2) and after infection of CD4+ T cells with DNA (3, 4) or RNA viruses (5, 6, 7).
Experimental data from our laboratory has shown that TCR/CD3 surface receptors are down-modulated after infection with HIV-1 (5, 8, 9), HIV-2 (10), and human T cell leukemia virus I, as well as in patients with the lymphocytic variant of hypereosinophilic syndrome, leading to T cell lymphoma due to a specific defect in human CD3γ (hCD3γ)3 gene transcription. 4 The CD3γ-chain plays several roles in TCR/CD3 expression and function, including a role in correct receptor assembly (11), its required presence for pre-TCR function during T cell development (12), a quantitative contribution to signal transduction (13), and a principal role in the constitutive recycling of surface receptor complexes (14), and the comodulation of nonengaged receptors (15) via a di-leucine motif in its cytoplasmic tail. Despite our growing understanding of the protein and its role within the TCR/CD3 complex, our knowledge of hCD3γ gene expression remains very limited.
The human CD3ε, CD3γ, and CD3δ genes lie in a 50-kb cluster on chromosome 11 (chromosome 9 in mice) (16) with CD3γ and CD3δ oriented in a head-to-head position and separated by only 1.6 kb (1.4 kb in mice) (17, 18). Calculations based on sequence divergence indicate that CD3γ, CD3δ, and CD3ε arose from a common ancestral gene in a two-step process of gene duplication (19). Originally, it was postulated that the CD3γ and CD3δ genes are controlled by common bidirectionally active regulatory elements, a conclusion based on their high sequence homology coupled with the orientation and the close proximity of their upstream regions (17, 18). However, this is not compatible with evidence that the CD3γ gene is expressed before CD3δ and CD3ε in thymocyte development (20), nor with our data showing specific transcriptional down-modulation by HIV (5, 10).
The transcriptional control regions governing expression of the TCR/CD3 genes were studied intensely and characterized for the TCRα (21), TCRβ (22), TCRγ (23), TCRδ (24), CD3δ (25), CD3ε (26), and CD3ζ (27) genes but not for the CD3γ gene. Examination of the human sequence revealed that CD3γ, as with CD3δ and CD3ε, does not contain a TATA or CCAAT box immediately upstream of the cap site. A study of the intron/exon organization of the CD3γ gene identified three major transcription start sites and arbitrarily designated the CD3γ promoter as extending ∼250 nt upstream from the start sites (17). A DNase I-hypersensitive site ∼100 nt upstream from exon 1 was used to support the assignment of this region as the hCD3γ gene promoter, although this was never verified experimentally (18, 28, 29).
In our efforts to define the mechanisms whereby HIV progressively down-modulates surface complexes from TCR/CD3high to TCR/CD3low to TCR/CD3− by targeting the CD3γ gene, we found that the defect is initiated early after infection with a continuous erosion of transcripts until a threshold is reached (loss of >90%) whereby the normal number of complete TCR/CD3 complexes can no longer be assembled and exported to the cell surface (30). This led us to search the upstream region of the hCD3γ gene for elements similar to those involved in controlling HIV gene expression. Two NFAT consensus sequences (NFATγ1 and NFATγ2) were located upstream from the first transcription initiation site with a third (NFATγ3) positioned in intron 1 (30). We demonstrated that NFATc2 alone binds to the NFATγ2 motif; however, complexes containing either NFATc2 or NFATc1 plus NF-κB p50 bind to the NFATγ1 and NFATγ3 sites. Furthermore, an increase in the binding of nuclear NFATc2 is correlated with a progressive loss of CD3γ transcripts after HIV infection (30).
In this study, we provide experimental evidence that the hCD3γ promoter is lymphoid specific, initiates transcription from multiple start sites, and contains two core promoters capable of recruiting the general transcription machinery through specificity protein (Sp)-binding motifs, with an Initiator (Inr) element present in the primary core promoter. In addition, the NFAT-binding motifs in the proximal promoter were investigated for their regulatory activity and found to positively or negatively regulate CD3γ gene expression via the differential binding of NFATc1 plus NF-κB p50- or NFATc2-containing complexes, respectively.
Materials and Methods
Cell culture conditions and reagents
Nuclear extracts were prepared from 2 × 107 cells, and EMSA experiments were performed as described previously (30). The radiolabeled oligonucleotide probes used for nuclear protein binding were as follows.
Oligonucleotides encoding wild-type, deleted, and mutated Spγ1/CD3γInr binding sites: Spγ1/CD3γInrwt, 5′-GTGATGGGTGGAGCCAGTCTAG-3′; Spγ1/CD3γInrdel 87, 5′-GTGATGGGGGAGCCAGTCTAG-3′; Spγ1/CD3γInrdel 82, 5′-GTGATGGGTGGAGCAGTCTAG-3′; Spγ1/CD3γInrmut 87, 5′-GTGATGGGGGGAGCCAGTCTAG-3′; Spγ1/CD3γInrmut 86/85, 5′-GTGATGGGTAAAGCCAGTCTAG-3′; Spγ1/CD3γInrmut 83/79, 5′-GTGATGGGTGGAACCAATCTAG-3′; and Spγ1/CD3γInrmut 89/88, 5′-GTGATGAATGGAGCCAGTCTAG-3′; oligonucleotides encoding the wild-type and mutated Spγ2 binding site: Spγ2wt, 5′-CTCAAAGGCCCCAGCCCCAACA-3′; and Spγ2mut, 5′-CTCAAAGGCCCCAGAACCAACA-3′; oligonucleotide encoding the wild-type Spγ3 binding site: Spγ3wt, 5′-AGCAGAGGGTGGAGGCTCTGGG3′; oligonucleotides encoding the wild-type and mutated Sp1 binding site in the human IL-12R2β promoter: IL-12Rβ2wt, 5′-CTCCAGTGGGCGGTCTTGTG-3′; and IL-12Rβ2mut, 5′-CTCCAGTGTTCGGTCTTGTG-3′.
The oligonucleotide bound complexes were separated on a 6% Tris-glycine-EDTA polyacrylamide gel migrated overnight at 50 V, and the radiolabeled protein complexes were detected by autoradiography.
The supershift assay, performed as previously described (30), used the following Abs: Sp1 (SC-59X), Sp2 (SC-643X), Sp3 (SC-644X), Ap2 family proteins AP2α (SC-184X), AP2β (SC-6310X), AP2γ (SC-8977X), TFIID (SC-204X), and TFIIB (SC-225X) (all from Santa Cruz Biotechnology). Abs were preincubated with nuclear extracts for 1 h on ice before the addition of the radiolabeled probe.
Plasmid and reporter constructs
A 1068-nt fragment (see Fig. 1) was amplified by PCR from human DNA using the following primer pair: 5′-ATTGTTCCACCTATTGCCTTCC-3′ (forward) and 5′-GAAGGGCAAAATGGAGGCT-3′ (reverse) and inserted into the Sfi site of pPCR-Script (Stratagene). This recombinant plasmid was digested with SacI and HindIII, and the resulting fragment was cloned into pGL3-BV (Promega).
To obtain the various 5′-truncated constructs, fragments were generated by PCR using oligonucleotide primers pairs with a 5′-linked SacI site in the forward primer and a 3′-linked HindIII site in the reverse primer (underlined). The forward primers used were as follows:
pHγ3(−15/+279) 5′- GCGCGAGCTCGTGATGGGTGGAGCCAGTC-3′; pHγ3(−59/+279) 5′- GCGCGAGCTCAGGCTCTGGGTTCTTGCCT-3′; pHγ3(−87/+279) 5′-GCGCGAGCTCTGCTGCTCACACTTGCAGC-3′; pHγ3(−123/+279) 5′-GCGCGAGCTCAAAAGGCATCTGCACCTGC-3′; pHγ3(−149/+279) 5′-GCGCGAGCTCACCTTCACCCTCCTTAACGGA-3′; pHγ3(−199/+279) 5′-GCGCGAGCTCGATGAGTCTCTGAGTGGGAATCC-3′; pHγ3(−239/+279) 5′-GCGCGAGCTCACCATCTCCCACCCAGCA-3′; pHγ3(−309/+279) 5′-GCGCGAGCTCAACAACTGGCTACGATCCTAACAA-3′; and pHγ3(−419/+279) 5′-GCGCGAGCTCCCTGAATGAAGGCCTGGACT-3′.
The reverse primer paired with all of the above forward primers was 5′- GCGCAAGCTTAGCCTCCATTTTGCCCTTC-3′.
All PCR products were digested with SacI and HindII and cloned into pGL3-BV. The 3′-deletion construct pHγ3-PstI (−789/−103) was derived by removing the 3′-PstI fragment from pHγ3(−789/+279).
Mutations and/or deletions in the NFAT, Sp, and transcription start sites were derived from pHγ3-wt using the QuikChange site-directed mutagenesis kit (Stratagene) and the following primers: pHγ3-NFATγ1mut, 5′- CCTTCACCCTCCTTAACCCTTAAACAAAAGGCATCTGC-3′; pHγ3-NFATγ2mut, 5′-CTGGACTGAGGTGGCTAAGGATTTGGAGGTCCAGCC-3′; pHγ3-Spγ2mut, 5′-CCCACCCAGCATCCATTTCTTTTCCCTGTGCAAGATG-3′; pHγ3-Spγ2del, 5′-GGGTTCTTGCCTTCGTGATGGGTGGAGC-3′; pHγ3-del87, 5′-CCAACAGTGATGGGGGAGCCAGTCTAGC-3′; pHγ3-del82, 5′-GTGATGGGTGGAGCAGTCTAGCTGC-3′; and pHγ3-del67, 5′-GCCAGCCAGCCTGTCAGCAGCTAGAGTGG-3′.
Jurkat, SupT1, WE17/10, Molt 4, Raji, GM607, HeLa, and 293T cells were transiently transfected using standard DEAE-dextran protocols. Exponentially growing cells (3 × 106 for Jurkat, SupT1, Molt 4, WE17/10 Raji, and GM607 cells and 2 × 106 for HeLa and 293T cells) were washed once with STBS 1× (25 mM Tris-HCl (pH 7.5), 1.37 mM NaCl, 5 mM KCl, 500 μM CaCl2, 500 μM MgCl2, and 600 μM Na2HPO4) and resuspended in 410 μl of STBS 1× containing 450 μg/ml DEAE-dextran, 0.5 μg of the reporter plasmid (containing the regulatory region being tested), and 0.06 μg of an internal control plasmid containing the Renilla luciferase gene under control of the herpes simplex virus-1 thymidine kinase promoter (pRL-TK vector; Promega). In cotransfection experiments, an additional 0.085 μg of the p-REP-NFATc1 or pREP-NFATc2 (kindly provided by T. Hoey (Tularik)) (32) was added with the reporter plasmid constructs. Luciferase was detected using a dual-luciferase reporter assay system (Promega), according to the manufacturer’s instructions. Each experiment was performed in triplicate, and the experiments show data representative of at least three independent experiments.
RLM-RACE clones were generated using First Choice RLM-RACE (Ambion) with poly(A)+ RNA from human spleen, total RNA from human thymus, (BD Clontech), or extracted from Jurkat, SupT1, and WE17/10. The RNA was reverse transcribed either at 42°C with Moloney murine leukemia virus reverse transcriptase (Ambion) and random decamers or at 52°C with Thermoscript RT-PCR (Invitrogen Life Technologies), 2.6% DMSO, and oligo(dT). The RLM-RACE clones were amplified by nested PCR from the cDNA using the 5′-RACE- and 3′-hCD3γ-specific primers shown in Fig. 4. The cycling conditions were as follows: 95°C at 12 min; cycle: 94°C at 30 s, 65°C at 30 s 72°C at 30 s, and 72°C at 10 m. The PCR product DNA were then inserted into pCR2.1 TOPO (Invitrogen Life Technologies) and sequenced.
Chromatin immunoprecipitation (ChIP) assay
The ChIP assay was performed using the kit purchased from Upstate Biotechnology generally following the manufacturer’s protocol. Uninfected and HIV-1-infected WE17/10 cells were fixed with 1.5% formaldehyde for 10 min at 37°C. Chromatin was isolated, sheared using a Bioruptor (Diagenode), and immunoprecipitated with Abs directed to Sp1 (SC-59X), Sp3 (SC-644X), TFIID (SC-204X) (all from Santa Cruz Biotechnology), NFATc2 (67.1; kindly provided by A. Rao (33)), or control rabbit IgG (Upstate Biotechnology). Cross-linking was reversed by heating, and the proteins were removed subsequently by proteinase K digestion. The presence of selected DNA sequences in the immunoprecipitated DNA was assessed by PCR using the following primer pairs (their location is shown in Fig. 1): NFATγ2 (200-bp product), forward, 5′-CAGCCTGGGCAACAAGT-3′, reverse, 5′-GTTGTTAGGATCGTAGCCAGTTG-3′; and Spγ1, CD3γInr, and Spγ2 (205-bp product), forward, 5′-GGGTTCTTGCCTTCTCTCTCAA-3′, reverse, 5′-CCCCTAGTAGGCCCTTACCTT-3′.
The amplified 32P-labeled PCR product was separated on a 6% acrylamide gel and detected by autoradiography.
The hCD3γ gene promoter sequences
We cloned a fragment from the 5′-flanking sequence of the hCD3γ gene (National Center for Biotechnology Information (NCBI) no. X06026; Ref. (17) upstream from a luciferase reporter gene to produce the pHγ3-wt construct. This 1068-nt sequence extends 869 nt upstream and 199 nt downstream from the ATG (Fig. 1) and includes the Alu element, the CD3γ half of the hCD3γ-δ intergenic region, exon 1, and 143 nt of intron 1. Transient transfection of pHγ3-wt in a variety of human T cell lines, including Jurkat, WE17/10 (Fig. 2,A) and Molt 4 (data not shown), revealed that this sequence contains weak promoter activity with levels increasing from 2.5- to 5-fold over the empty pGL3-basic vector (pGL3-BV). The 1068-nt sequence cloned in the antisense direction in the pGL3-enhancer vector has no activity when transfected in Jurkat cells (data not shown), demonstrating that this is not a bidirectionally active promoter. Interestingly, promoter activity was consistently found to be much lower in SupT1 cells (1.25- to 2-fold increase), an immature CD3lowCD4−CD8− T cell line expressing the pre-TCR (Fig. 2 A) (34).
Tissue specificity of the hCD3γ gene promoter
The T cell specificity of pHγ3-wt was investigated in a transient reporter assay using two B cell lines, Raji and GM607, as well as the epithelial cell line HeLa and the fibroblast cell line 293T (Fig. 2 A). In B cells, promoter activity was similar to that of T cells, with Raji generally displaying higher activity than GM607 relative to pGL3-BV. Raji is considered to be a pre-B cell line expressing intracellular IgM (35), whereas GM607 secretes IgM and thus is a terminally differentiated plasma cell, possibly accounting for this difference in activity.
The nonlymphoid cell lines, HeLa and 293T, exhibited a 3-fold increase in pHγ3-wt activity compared with pGL3-BV when the same concentrations of the Renilla internal control and hCD3γ promoter construct were transiently transfected (data not shown). However, in these experiments, the pGL3 promoter vector (pGL3-PV; used as a positive control) was one to two orders of magnitude higher in the adherent cells compared with all of the suspension grown lymphoid cells. Experiments where the Renilla plasmid concentration was increased progressively demonstrated that the pHγ3-wt activity detected in nonlymphoid cells results from increased transfection efficiency rather than true promoter activity (Fig. 2 A) and suggests that the hCD3γ promoter is lymphoid but not T cell specific.
A construct containing the 5′-PstI fragment of pHγ3-wt (pHγ3-PstI; PstI site shown in Fig. 1), which includes the 5′-proximal promoter sequence starting 183 nt upstream the ATG, was investigated for its activity. Transient transfection of pHγ3-PstI in various cell types revealed that while there is no detectable activity in any of the T cell lines, this construct is active in B cell and nonlymphoid cell lines (Fig. 2 B). The activity in Raji was 2-fold higher than pGL3-BV but 2.5-fold lower than the full-length promoter construct pHγ3-wt. In HeLa, where the full-length promoter construct has no measurable activity, the pHγ3-PstI construct is 2.5-fold more active, suggesting that lymphoid-specific and/or T cell-specific regulatory elements may be located in the region 3′ of the PstI site.
Transcriptional initiation of the hCD3γ gene
Three major transcription initiation sites for the hCD3γ gene were identified previously by S1 nuclease and primer extension analysis using cytoplasmic RNA extracted from the T cell lines Jurkat and SupT1 (located 87, 82, and 67 nt from the ATG; Fig. 1) (17). We explored the functional activity of these initiation sites by deleting each nt individually in the 1068-nt pHγ3-wt construct, with the resulting 1067-nt constructs called pHγ3-del87, pHγ3-del82, and pHγ3-del67. Transient transfection experiments in Jurkat, SupT1, and Raji (all three were similar; Jurkat is shown in Fig. 3 A) revealed that deletion of the cytidine at 67 does not have a significant effect on promoter activity in comparison with pHγ3-wt. Alternatively, deletion of the thymidine at 87 or the cytidine at 82 has a severe deleterious effect on hCD3γ promoter activity, although the deletion at 82 consistently displays slightly more activity than the deletion at 87 relative to pGL3-BV. These results were surprising considering that removal of a single nucleotide corresponding to the start site at position 67 (or randomly at position 79, data not shown) had no effect. Deletion of one among multiple transcription initiation sites should modulate promoter activity as a function of site usage.
These data led us to ask whether this loss of activity was related to a critical transcription factor complex whose binding was disrupted by the removal of a single nucleotide in the vicinity of nt 87 and 82. We examined binding to this region and found that the sequence surrounding nt 87 and 82 (Spγ1/CD3γInr probe) binds two specific nuclear protein complexes (bands A and B, Fig. 3,B). These complexes could be specifically competed with a wild-type but not a mutated version of an Sp consensus sequence from the IL-12Rβ2 promoter (data not shown; Ref. 36). EMSA supershift experiments revealed that the majority of the A complex can be shifted with an anti-Sp1 Ab, whereas the entire B complex is shifted with an anti-Sp3 Ab. Multiple exposures of this gel revealed that portions of both the A and B complexes are independently shifted with an anti-TFIID Ab, whereas anti-Sp2 and anti-TFIIB Abs have no effect. Thus, two complexes containing either Sp1 and TFIID (band A) or Sp3 and TFIID (band B) bind to the sequence surrounding nt 87 and 82. Chromatin immunoprecipitation experiments confirmed the in vivo binding of Sp1, Sp3, and TFIID binding to a 205-bp sequence (primer locations are indicated in Fig. 1) that extended upstream to −59 and contained the Spγ1/CD3γInr motif (Fig. 3 C).
The use of mutated or deleted EMSA probes revealed the importance of specific nucleotides for binding of the Sp/TFIID complexes (Fig. 3, D and E). Deletion of the thymidine at position 87 severely reduces the binding of both the A and B complexes. Mutation from G→A of the two preceding nucleotides at positions 89 and 88 (lane 6) or the two succeeding nucleotides at positions 86 and 85 further reduces binding compared with the deletion of nt 87; however, the mutation 86/85 is associated with the appearance of a higher m.w. complex. Mutation of nt 87 from T→G severely reduces binding (a low m.w. band appears), but the effect is not as dramatic as when this nucleotide is deleted. Alternatively, deletion of the position 82 start site only reduces binding by ∼50%, perhaps explaining the low level of activity observed in the pHγ3-del82 construct (Fig. 3,A). Mutation of the nucleotides at positions 83 and 79 from G→A only moderately affects A and B binding. These data (summarized in Fig. 3,E) show that the 5′-GGGTGGAG-3′ motif at nt 90–81, designated Spγ1 (Fig. 1), is critical for Sp1, Sp3, and TFIID binding and positive promoter activity and that any alteration to the architecture of this critical element eliminates promoter function.
The hCD3γ gene lacks a classical upstream TATA or CCAAT box, which could be used to help position the start site(s). Furthermore, the previously identified hCD3γ transcription initiation sites (17) were not confirmed either by the cDNA (37) nor by human mutant mRNA sequences (NCBI nos. X60491, X60492, and X60493; Ref. 38), which begin 38 and 30 nt upstream from the ATG, respectively (Fig. 1). Although this discrepancy could be explained because cDNA clones frequently lack sequences corresponding to the 5′-end of the mRNA transcript due to stable stem-loop structures that inhibit the processivity of the reverse transcriptase, our data suggest that the transcription initiation sites may actually be located at alternate nucleotides.
We used RLM-RACE in an attempt to resolve these inconsistencies and precisely define the 5′-nucleotide at the boundary of full-length capped hCD3γ mRNA transcripts. We used RNA prepared from human lymphoid tissues and found that the majority of clones from the spleen terminated at nt 80, with a significant number also ending at nt 68 or 66 upstream from the AUG (Fig. 4). In the thymus, the most prevalent clones terminated at nt 66. Based on these data, we have designated nt 80 as the +1, which is an adenine positioned in a classical Inr consensus sequence (5′-PyCA+1NTPyPy-3′; CD3γInr; Figs. 1 and 4) (39). Inr sequences do not surround the G at +13 (nt 68) or the A at +15 (nt 66). In addition to the three major start sites, we found transcripts terminating at other nucleotides in the vicinity. The longest spleen clone terminated at −44 and the shortest ended at +23, with others extending to −19, −16, +6, +10, +17, +19, +21, and +22. In the thymus, the longest clone ran to −39 and the shortest stopped at + 15, with others extending to −14, −6, −5, +4, +6, and + 14. To resolve whether these other initiation sites were due to RNA secondary structure, we used a thermostable reverse transcriptase and ran the reaction at 52°C with or without DMSO. These experiments reinforced the most frequent use of + 1, +13, and + 15 start sites but also confirmed that they were not exclusive.
We also examined RLM-RACE clones from SupT1, Jurkat, and WE17/10, which represent immature, intermediate, and mature T cell lines, respectively. We found that the majority of clones again initiated at the three principal sites in Jurkat and WE17/10, with +1 most frequently used in WE17/10 and the +13 and +15 in Jurkat. Interestingly, initiation in SupT1 occurred most frequently at −19 and −16. Regarding the previously identified start nucleotides (17), we did not find any transcripts that initiated at nt 87 or 82 (corresponding to −8 and −2; Fig. 1) and only one transcript beginning at nt 67 (+14) in the thymus. Overall, these experiments show that a cluster of transcription initiation sites surrounds the hCD3γ core promoter, defined by the Spγ1 motif and the CD3γInr (Figs. 1 and 4).
The hCD3γ gene contains a second core promoter
The 5′-GGGTGGAG-3′ Spγ1 sequence, which binds TFIID, Sp1 and Sp3, is present as an identical repeat at −65 to −58 (Spγ3; Fig. 1). In light of our RLM-RACE data, detecting clones that extended as far back as −44, we asked whether a second core promoter surrounds the Spγ3 motif. Binding experiments show that Sp1 or Sp3 and TFIID can also bind to the Spγ3 probe; however, their abundance is significantly lower compared with the Spγ1/CD3γInr probe (Fig. 5; this gel was exposed for 7.5 days compared with 2 days for Spγ1/CD3γInr in Fig. 3,B). Again, these complexes could be specifically competed with a wild-type but not a mutated Sp consensus sequence (data not shown). EMSA supershift experiments confirmed that the A complex can be shifted with an anti-Sp1 Ab, the B complex with an anti-Sp3 Ab, and portions of the A and B complexes with the anti-TFIID Ab as for Spγ1/CD3γInr. No Inr sequence is located adjacent to the Spγ3 motif; however, two degenerate Inr-like sequences were detected 12–17 nt downstream from Spγ3 (Figs. 1 and 4). Thus, a second but weaker core promoter is located in the region of Spγ3 (−65 to −58) and likely generates the transcripts originating upstream of the +1.
Regulatory activity of the hCD3γ promoter
Potential positive and negative regulatory sequences within the hCD3γ promoter were investigated by cloning increments of the 1068-bp fragment in pGL3-BV. All of the fragments begin at +279 (downstream from the exon 1 splice donor site) and terminate at the various upstream locations highlighted in Fig. 1. Analysis of the truncated constructs in a transient reporter assay reveals that both positive and negative regulatory regions are present in the full-length pHγ3-wt construct (= pHγ3(−789/+279)). The activity of the full-length pHγ3-wt construct shown (Fig. 6 A) is at the lower end of the 2.5- to 5-fold increase normally observed; however, the relative activity (increase or decrease) of the individual constructs was reproducible in 15 experiments performed in Jurkat and other T cell lines.
In T cells, the shortest construct pHγ3(−15/+279) possesses 3-fold more activity than the full-length wild-type promoter and 10-fold more compared with the empty vector. This increased activity was detected in all cell lines tested, including T cells, B cells, fibroblasts, and epithelial cells (Fig. 6 B). Thus, isolated from its upstream regulatory elements, the pHγ3(−15/+279) construct, which contains the primary core promoter (Spγ1/CD3γInr), exhibits unregulated expression. The successive pHγ3(−59/+279) construct, truncated in the middle of the secondary core promoter Sp motif (Spγ3), has only <50% of wild-type promoter activity. Furthermore, a 6-nt (GGGTGG) Spγ3 deletion in the full-length wild-type construct completely eradicates promoter activity in T cells but not B cells (data not shown), suggesting that complexes binding in this region do play a critical role in promoter activity. Elements present in the pHγ3(−87/+279), pHγ3(−123/+279), and pHγ3(−149/+279) constructs, the latter one containing the NFATγ1 motif (30), bring promoter activity back in proximity to wild-type levels.
An important positive element appears to be contained in pHγ3(−199/+279), whose activity is suppressed by a negative regulatory element present in pHγ3(−239/+279). The positively acting downstream sequence in pHγ3(−199/+279) contains potential Ikaros and Ets binding sites, both of which are important positive transcription factors. The sequence from −224 to −202, present in pHγ3(−239/+279) binds an abundant and specific complex whose binding is abrogated by mutating the −215, −213, and −212 nt from G→T, a mutation known to affect Sp protein binding (36). Introduction of the same mutation in the pHγ3-wt construct increased activity 2-fold over the wild-type vector (data not shown), suggesting a negative role for this complex. Although this nuclear protein complex can be specifically competed with a Sp consensus sequence, a range of different anti-Sp Abs were unable to supershift this band, perhaps due to epitope inaccessibility (data not shown).
A positive effect is again apparent in the pHγ3(−309/+279) construct, which appears to be negated in the pHγ3(−419/+279) construct. A search for transcription factor motifs detected potential GATA and Oct binding sites located in the sequence from −309 to −239. The GATA family of transcription factors includes GATA3, which is expressed exclusively in T cells, and while Oct 1 is ubiquitous, Oct 2 is lymphoid specific and is expressed in CD4+ T cells but not CD8+ T cells. Potentially one or both could play a role in regulating CD3γ promoter activity in T cells. The pHγ3(−419/+279) construct contains the NFATγ2 motif, which will be discussed further below (functional activity of the NFATγ1 and NFATγ2 motifs), and the Alu element described above. The 3′-deletion in pHγ3-PstI (−789/−103) eliminates the transcription start sites and the two core promoters, demonstrating that the promoter activity detected in T cells is derived from the hCD3γ sequence.
The GC/GT rich sequence of the hCD3γ gene promoter contains a third Sp-binding motif
Computational analysis of the promoter sequence using Transcription Element Search System (〈 www.cbil.upenn.edu/tess/〉) detected a third potential Sp-binding motif at −29 to −20 (Spγ2; Fig. 1). EMSA experiments show that a major nuclear protein complex binds to Spγ2 (Fig. 7,A, lane 1, band A), which can be specifically competed with the homologous probe and an Sp consensus sequence but not probes mutated to abrogate Sp protein binding (data not shown). Binding of this complex is observed in nuclear extracts from T and B cell lines but not from HeLa. Using Abs to the Sp family members, Sp1, Sp2, and Sp3, and the AP-2 family members, AP-2α, AP-2β, and AP-2γ, which also bind GC rich sequences, and TFIID (Fig. 7 A) and TFIIB (data not shown), only the anti-Sp1 Ab shifted the A complex, demonstrating Sp1 specifically binds to the Spγ2 motif.
To understand whether Sp1 binding to Spγ2 has a functional role, we produced a full-length promoter construct (pHγ3-Spγ2mut) containing the mutation (AGCC→AGAA) that was found to abrogate Sp binding in EMSA experiments. A transient reporter assay revealed that in pHγ3-Spγ2mut the loss of Sp1 binding reduces promoter activity by ∼40–50% (Fig. 7 B). This activity was reduced an additional 10–15% in an Spγ2-deleted construct (Δ−29 to −20; pHγ3-Spγ2del). Thus, Sp1 plays a significant role in regulating hCD3γ promoter activity, and its presence may also stabilize the binding of other protein:DNA complexes to adjacent elements involved in positive expression of the hCD3γ gene.
Functional activity of the NFATγ1 and NFATγ2 motifs
We previously identified two NFAT consensus sequences in the hCD3γ promoter at −131 to −127 (NFATγ1) and −391 to −387 (NFATγ2) from the first transcription initiation site (Fig. 1) (30). We further demonstrated that NFATc2 alone binds to the NFATγ2 motif while complexes containing either NFATc2 or NFATc1 plus NF-κB p50 bind to the NFATγ1 site, with an increase in NFATc2 binding associated with the loss of CD3γ gene transcripts after HIV-1 infection (30). Recent ChIP experiments confirmed the in vivo binding of NFATc2 containing complexes to the NFATγ1 (data not shown) and NFATγ2 (Fig. 8 A) motifs and further support their increased binding in HIV-1-infected cells.
An investigation into the functional role of these motifs in regulating the hCD3γ promoter was accomplished by producing NFAT binding mutants (GGAA→CCTT; Ref. (40) in the pHγ3-wt construct. In a transient reporter assay, hCD3γ promoter activity was consistently increased 2-fold when the NFATγ2 site was mutated (pHγ3-NFATγ2mut) (Fig. 8,B). This data confirms the lower level of promoter activity in the truncated pHγ3(−309/+279) (NFATγ2 is present) compared with pHγ3(−419/+279) (NFATγ2 is absent) constructs (Fig. 6). In addition, cotransfection of an NFATc2 expression vector with the full-length pHγ3-wt construct results in a reduction of promoter activity by 40% (Fig. 8 C).
In contrast, a NFATγ1-binding mutant (pHγ3-NFATγ1mut) decreases promoter activity by ∼15% (Fig. 8,B). Cotransfection of an NFATc1 expression vector with the pHγ3-wt construct (Fig. 8,C) increases promoter activity up to 20% over the wild-type vector alone, which reflects the differences observed between the truncated pHγ3(−149/+279) (NFATγ1 is present) and pHγ3(−123/+279) (NFATγ1 is absent) constructs (Fig. 6). Additional investigation into the role these factors play in regulating the NFATγ1 motif alone was conducted using the NFATγ2-mutated full-length construct (pHγ3-NFATγ2mut; Fig. 8,D). Cotransfection with either an NFATc1 or NF-κB p50 expression vector increased promoter activity by 25 and 30%, respectively (Fig. 8,D), but had no apparent effect on constructs containing a mutated NFATγ1 motif (pHγ3-NFATγ1mut or pHγ3-NFAT(γ1+γ2)mut) (data not shown). Cotransfection of NFATc1 together with the NF-κB p50 expression vector had a synergistic effect, increasing promoter activity by 55%. In contrast, cotransfection of the NFATγ2-mutated construct with an NFATc2 expression vector decreases full-length promoter activity by 21% (Fig. 8 D), confirming the lower level of negative activity conferred by the NFATγ1 compared with the NFATγ2 motif. Cotransfection with an NFATc4 expression vector, which is not expressed in T cells and has no effect on promoter activity, was used to compensate for the variable amounts of DNA present in each mixture.
These experiments suggest that the NFATc1 plus NF-κB p50 and the NFATc2 containing complexes compete for binding to the NFATγ1 site with the former exerting a positive influence and the latter a pronounced negative effect on promoter activity. A construct where the NFATγ1 and NFATγ2 motifs are both mutated appears to neutralize the negative activity of the individual sites with activity generally higher than in the wild-type construct, likely due to the presence of other positive acting elements (Fig. 8 B). Overall, our data suggest that NFATγ2 plays a negative role and NFATγ1 a positive or negative cis-regulatory role in the hCD3γ promoter depending upon the complexes bound.
The data presented here show for the first time that the hCD3γ gene is controlled independently by a weak, non-T cell-specific initiator-based promoter with some similarities to those previously described for the highly homologous CD3ε and CD3δ genes (41, 42). Originally, both the CD3ε and CD3δ promoters were found to lack tissue specificity, with T cell-specific expression achieved through their enhancer elements (26, 43). A recent study (44) has shown that specific upstream elements in the mouse CD3δ promoter also contribute to its T cell-specific expression. We found the hCD3γ proximal promoter to be equally active in B cells; however, studies with the constructs truncated 5′ at −15 or 3′ at the PstI site (−103) suggest that this intervening region does contain T cell-specific activity. However, this does not exclude the possibility that other critical T cell-specific elements, including an enhancer, are located elsewhere and contribute to tissue-specific expression of the CD3γ gene.
The hCD3γ promoter is located downstream from an Alu element that forms a natural partition between the CD3γ and CD3δ promoters and accounts for the increased length of the intergenic region in humans compared with mice. Comparison of the wild-type promoter with a construct truncated at the 3′-end of the Alu element revealed that it does have a positive influence on hCD3γ promoter activity, and potential binding sites for LyF-1, Ets-1, and GATA are present in the hCD3γ-δ Alu element. The presence of this natural partition in the CD3γ-δ intergenic region provided our rationale for cloning incremental segments of the proximal promoter to the upstream −789 nt. We also extended our constructs downstream into intron 1 (+279) to avoid expression from cryptic promoter sites in the pGL3 vector, which can be significant when short promoter sequences are inserted (45).
Analysis of constructs containing incremental lengths of the hCD3γ promoter sequence revealed that the proximal promoter contains an alternating series of positive and negative regulatory elements. An important negative regulatory element is located from −419 to −309, a region we demonstrated contains the NFATc2-binding NFATγ2 motif (30). Mutational analysis, together with cotransfection experiments, confirmed a negative regulatory role for NFATc2 binding to the NFATγ2 motif and to a lesser extent the NFATγ1 motif. Additional support for the negative role of NFATc2 comes from the decrease in CD3γ gene transcripts observed after HIV-1 infection in concert with increased NFATc2 binding (30), recently confirmed using chromatin immunoprecipitation experiments. In addition, we found that the CD3−D4+ T cell phenotype of a premalignant clone associated with hypereosinophilic syndrome was correlated with a defect in CD3γ gene transcripts and a substantial increase in nuclear NFATc2 binding. 4 The presence of a silencer element that extends beyond the NFATγ2 motif is supported by the more pronounced negative effect observed between the constructs truncated at −419 and at −309 relative to the NFATγ2 mutant construct, and our preliminary experiments indicate that other specific nuclear protein complexes bind in this region. Interestingly, the NFATγ1 motif appears to have a positive or negative effect on promoter activity, depending upon whether complexes containing NFATc1 plus NF-κB p50 or NFATc2 are bound, respectively. Therefore, the balance between the positive and negative activity of the NFATγ1 motif may depend upon the nuclear abundance of these NFAT and NF-κB family proteins. The role of NFATγ1 in the hCD3γ promoter may be to achieve full positive or negative activity, depending upon the differential binding of other transcription factors to elements located immediately upstream or downstream.
Recent studies suggest that there are two distinct classes of NFAT target genes, one positively controlled by an NFAT:AP-1 complex and the other regulated by NFAT alone or in cooperation with other partners (46). Although NFAT family proteins frequently bind to composite binding sites with AP-1 (47); however, we previously demonstrated that the hCD3γ NFAT motifs do not bind the AP-1 proteins c-Fos or c-Jun (30). NFAT proteins can also cooperate with other factors bound to nearby elements, including GATA, Ets, C/EBP, ICER, c-Maf, MEF2, and Sp1/Sp3 (48, 49). Preliminary experiments indicate that a potential Ets motif located ∼30 nt upstream from NFATγ1 does bind specific nuclear factors, whereas up to 10 different nuclear protein complexes bind to a 40-nt sequence upstream of the NFATγ2 motif. Our data on the hCD3γ NFAT motifs provides another example of a gene where members of the NFAT family are involved in differential regulation of the promoter.
In recent years, an increasing number of regulated or tissue-specific genes have been found to lack a TATA or CCAAT box sequence and to recruit the general transcription machinery to the promoter via an Inr and/or downstream-positive element (DPE). TFIID specifically interacts with the Inr via TATA box binding protein-associated factors (TAF)-1 and -2, and an Inr is sufficient for strong and accurate initiation stimulated by the transcription factor Sp1 (50). We located a classical Inr sequence (CD3γInr, Fig. 1) (51) in the hCD3γ promoter surrounding the most frequently used transcription start site (+1). Additionally, we identified a contiguous upstream Sp binding site (Spγ1) along with a second Sp site (Spγ2) located at −29 to −20. Our experimental data demonstrate that the contiguous Spγ1/CD3γInr element can recruit and bind Sp1/TFIID- and Sp3/TFIID-containing complexes, which were found to be essential for promoter function. In addition, the CD3γ Inr not only depends upon the contiguous Spγ1 site but also on the upstream Spγ2 site, which increases promoter activity 2-fold. Sp1 can directly stabilize the binding of TFIID to core promoter elements (52), and physical interactions have been detected between Sp1 and the TFIID-associated TAFs, TAF-1 and TAF-2 (53).
Sp3 can repress the positive role of Sp1 in transcription initiation, particularly in genes with multiple adjacent binding sites (54). Sp3 is thought to compete with Sp1 for binding to the promoter (55, 56) and was found to negatively regulate TCRα promoter expression through repression of Sp1-mediated promoter activation (57). Alternatively, Sp1 and Sp3 were shown to interact with NFATc1 and play a positive role in transcription of a metalloproteinase gene (49). Sp1 can also interact with other transcription factors such as YY1 (58, 59), and recently, YY1 was shown to play a negative role in mouse CD3δ core promoter activity (44). An initial examination suggested there may also be Sp-binding GT boxes in the CD3δ core promoter similar to those we identified for CD3γ, and preliminary experiments indicated that an Sp1, Sp3, and TFIID complex does bind to this element. Although our data show that Sp1 plays a positive role in regulating hCD3γ promoter activity, it is still unclear whether Sp3 binding with TFIID enhances or inhibits activity of the general transcription machinery.
In many promoters, the binding and correct positioning of the TFIID transcription machinery depends upon the TATA box or DPE (located 25–30 nt upstream or downstream, respectively) to precisely position TFIID relative to a single transcription initiation site. We have shown that the hCD3γ core promoter can bind TFIID and initiate transcription in the absence of these upstream or downstream elements. Additionally, we found that in the full-length pHγ3-wt construct: 1) deletion of a single nucleotide in the Spγ1 motif destroys promoter activity; 2) deletion or mutation of the Spγ2 motif reduces promoter activity by 50%; 3) deletion of the Spγ3 motif eradicates promoter activity in T cells but not B cells; and 4) a construct truncated at −59 (in the Spγ3 motif) has only 50% of wild-type promoter activity. This differential reduction in activity suggests that the primary core promoter requires the presence of all three Sp motifs for its full function in T cells. The 50% activity of the Spγ3-truncated or the Spγ2-deleted constructs compared with the completely inactive Spγ3-deleted construct suggests that elements located further upstream exert a significant influence on their cooperation with the Inr.
Transcription is not required to start at the +1 for an Inr to function, and in TATA-containing promoters, the location of the start site is dictated normally by the location of the TATA box rather than the Inr (60). Although multiple initiation sites are commonly observed in TATA-less promoters, many function through a DPE to more accurately initiate transcription. Deletion of the sequence from +4 to +43 did not reduce the activity of the construct containing the primary core promoter in the absence of upstream elements (pHγ3(−15/+279); data not shown), making it unlikely that the hCD3γ promoter contains a DPE. RNA polymerase II is generally thought to recognize DNA nonspecifically, and the absence of a TATA box or DPE to stabilize the large TFIID complex on the promoter would thus lead to erratic transcription initiation. We detected tremendous heterogeneity in the hCD3γ transcription initiation sites in the region of the three Sp motifs, which suggest significant imprecision by the hCD3γ Inr. Such an extensive cluster of transcription start sites has not been reported previously; however, this may simply reflect the sensitivity of the RLM-RACE technique we used compared with the S1 nuclease or primer extension techniques. An alternative explanation for our data is that a stable RNA secondary structure inhibits reverse transcriptase processivity with premature termination occurring at relatively high frequency. Using a thermostable reverse transcriptase and DMSO to melt any potential RNA secondary structures did increase the frequency of the +1 and +15 start sites but did not completely eliminate transcripts with alternate terminal nucleotides, suggesting that this is not the entire explanation.
The presence of multiple start sites is associated frequently with alternate promoters and/or splicing of the transcripts to produce different protein isoforms. In addition to the primary core promoter, we found that the Spγ3 motif apparently can function as a second albeit weaker core promoter. The Spγ3 sequence is identical to Spγ1 and both bind Sp1/TFIID- and Sp3/TFIID-containing complexes, although there is no contiguous Inr sequence for Spγ3. It has been shown that if a large number of pyrimidines are present surrounding the start site, then low levels of Inr activity can be detected in the absence of either the A at +1 or the T at +3 (51). The transcripts initiating with the G at −44 and the C at −39 are each surrounded by a degenerate Inr sequence, which lacks the A (+1) and T (+3) but contains correctly positioned pyrimidines (5′-PyPy+1NNPyPy-3′; Figs. 1 and 4). The nearby upstream Spγ3 motif may function through these Inr-like elements with lower affinity for the binding of the TFIID-associated TAFs and thus weaker promoter function. The RLM-RACE experiments detected transcripts initiating upstream of the +1 at −44, −19, and −16 in the spleen and −39, −14, −6, and −5 in the thymus, suggesting that they arise from the upstream core promoter. Additional support for two functional core promoters comes from the alternating positive and negative control regions identified in the upstream proximal promoter, which perhaps independently influence expression from a given core promoter. Alternatively, the two TFIID-binding Sp motifs could function in cooperation to recruit the general transcription machinery to this region of the promoter with transcripts initiating in a relatively random fashion over a region of ∼60 nt.
Our rationale for comparing transcripts in immature and mature T cells stemmed from a study showing that the mouse pre-TCR contains an alternatively spliced CD3γ gene product (61). This observation suggested that these alternate gene products could be derived from transcripts generated independently by the two core promoters. Support for this hypothesis is provided by our experiments showing that transcripts originating from the secondary core promoter (−19 and −16) were those principally detected in the immature T cell line SupT1, which expresses the pre-TCR. However, the observed preference for +1 in the spleen and +15 in the thymus, which potentially is also linked with a distinct role in T cell differentiation, has not been explored. Alignment of the hCD3γ proximal promoter with the similar mouse CD3γ gene sequence revealed a high degree of sequence conservation in the proximal promoter and >90% homology between each core promoter region. Our preliminary experiments have determined that the mouse gene has a similar architecture with two putative core promoters, which bind Sp1-, Sp3-, and TFIID-containing complexes, and additional studies are underway to investigate the similarities and differences of CD3γ gene transcripts from various T cell subpopulations.
Identification of the hCD3γ core and proximal promoter sequences provides a beginning to our understanding of what appears to be a divergence in the control of this immune response gene, as well as preliminary insight into how and why this gene is targeted in abnormal T cells (3, 5, 7, 10, 62). During normal T cell processes, the hCD3γ promoter could modulate CD3γ levels and thereby production of CD3εδTCRαTCRβCD3εγ complexes in the endoplasmic reticulum as a means of promoting or limiting the assembly and export of newly formed receptor complexes to the cell surface (63, 64). A transient decrease in the number of receptor complexes on the cell surface is characteristic of the TCR/CD3-directed Ag response and parallels the induction of cytokine genes, such as IL-2. Our findings that NFAT, NF-κB, and Sp family proteins play an important role in regulating CD3γ promoter function (30) suggests that a nuclear environment favoring up-regulation of the immune response genes could exert a negative influence on CD3γ gene expression, which is temperate during normal immune responses and amplified in abnormal T cells, such as those infected with HIV. Although the identification and characterization of other transcription factor-binding motifs in the CD3γ proximal promoter and its putative enhancer element is not complete, it is intriguing that parallels can be drawn between the elements identified and transcriptional control of both cytokine genes and the HIV-long terminal repeat.
We are indebted to Dr. T. Hoey and Tularik, Inc. (San Francisco, CA) for the NFAT expression plasmids and to Dr. F. Fuks for advice and the use of his Bioruptor for the ChIP experiments. We also thank Makram Merimi for his help with the ChIP experiments.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grants from the Belgian Fonds National de la Recherche Scientifique (FNRS) (Fonds de la Recherche Scientifique Médicale (Belgium) Grant no. 3.4584.01 and Télévie Grant nos. 7.4554.01 and 7.4584.01), the European Commission (Grant QLK2-2000-01040), U.S. National Institutes of Health Grant HD37356, the Fonds Medic, Friends of the Bordet Institute, the Fondation David and Alice Van Buuren and a collaborative grant from the International Brachet Foundation (R 97/8-05). K.E.W.-G. is a scientific collaborator of the FNRS-Télévie, B.M.B. a fellow of the FNRS-Télévie, and H.A. is a fellow of the Fonds pour la Formation a la Recherche dans L’Industrie et dans L’Agriculture.
Abbreviations used in this paper: hCD3γ, human CD3γ; Inr, Initiator; Sp, specificity protein; ChIP, chromatin immunoprecipitation; DPE, downstream positive element; RLM-RACE; RNA ligase-mediated rapid amplification of cDNA ends; TAF; TATA box binding protein-associated factors.
K. E. Willard-Gallo, B. M. Badran, M. Ravoet, A. Zerghe, A. Burny, P. Martiat, M. Goldman, F. Roufosse, and C. Sibille. Submitted for publication.