Promoter analysis of Zfp-36, the mitogen-inducible gene encoding the zinc finger protein tristetraprolin.

The gene encoding the putative zinc finger protein tristetraprolin (TTP), Zfp-36, is rapidly induced by a variety of mitogens and growth factors. We show here that 77 base pairs 5' of the transcription start site are sufficient for full serum inducibility of the mouse Zfp-36 promoter. This region of the promoter includes consensus sequences for the binding of the transcription factors EGR-1, AP2, and Sp1. In addition, we have identified a previously undescribed element, TTP promoter element 1 (TPE1); this 10-base pair sequence includes a palindrome and is identical in the human, bovine, and mouse genes. Each of the three binding elements, EGR-1, AP2, and TPE1, contribute to the serum induction of Zfp-36 and can confer serum-inducible expression on a heterologous minimal promoter. Gel mobility shift assays demonstrated the formation of complexes consisting of this region of the promoter and cellular nuclear proteins and demonstrated that the extent of complex formation could be altered by treatment of the cells with serum or insulin. These results suggest that the response of Zfp-36 to serum and other mitogens is mediated by a series of cis-acting elements acting in concert to confer full inducible transcription of this gene.

2) a zinc finger protein also known as Nup475 (3) and TIS11 (4). This protein, which is encoded by the gene Zfp-36 (2), is expressed at very low levels in quiescent fibroblasts but is rapidly induced by serum, polypeptide growth factors, and phorbol 12-myristate 13-acetate (PMA) (1,3,4). This expression is transient, with most mRNA disappearing after 2 h. In these cells, induction of Zfp-36 expression is independent of protein synthesis, and treatment with cycloheximide results in superinduction of the gene (1,(3)(4)(5)(6) The mechanisms that control the activation of Zfp-36 transcription have not been defined.
To identify sequence elements involved in the transcriptional control of TTP synthesis, we have characterized the 5Ј-flanking region of Zfp-36 using site-directed mutagenesis and deletional analysis. Several potential binding sites for known transcription factors have been identified in this study as contributing to the serum-stimulated activity of this promoter. In addition, we have identified and characterized a previously unknown promoter element that appears to function as a transcriptional enhancer and that participates in the regulation of seruminduced Zfp-36 transcription.

Library Screening and Subcloning
A Balb/c mouse genomic library (Clontech, Palo Alto, CA) was screened using the mouse TTP cDNA (1) as a probe. Pure plaques positively hybridizing to the probe were subjected to Southern mapping analysis. Restriction fragments from two clones overlapping within the intron and 1 kb of TTP protein coding sequence (clone 13 extended 5Ј, clone 7 extended 3Ј from the overlapping sequence) were subcloned into pBS ϩ (Stratagene, La Jolla, CA). An 8-kb SstI fragment from clone 13 was ligated into the SstI cloning site in pBS ϩ , and another plasmid was made by ligating a 3-kb XhoI fragment from clone 7 into the SalI site of pBS ϩ . The inclusion of the entire mRNA coding region was confirmed by dideoxy sequencing. Five kb of 5Ј-flanking region was also sequenced. The entire mouse genomic sequence has been deposited in GenBank TM (accession number L42317). A second mouse genomic library (129 sv, Stratagene) was also screened to obtain a 3.8-kb Sau3aI-SalI fragment, which contained 0.9 kb of the 5Ј-flanking region, the entire mRNA coding sequence, and a single intron. DNA sequence analysis revealed only a 2-bp difference in the 5Ј-flanking region when compared with the clone from the Balb/c library. This clone was used for promoter activity analysis as well as for the construction of a genetargeting vector to create knockout mice. 2 A human genomic clone was obtained by screening a human placental genomic library (Clontech) with the human TTP cDNA (2) (Gen-Bank TM accession number for the human genomic sequence: M19844). The bovine TTP cDNA was obtained from a bovine aorta epithelial cell cDNA library (Stratagene) using the human TTP cDNA as a probe (2). The bovine cDNA was then used to screen a bovine liver genomic library * 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.
The (Stratagene). The bovine genomic sequence has been deposited in Gen-Bank TM (accession number L42319).
Base pair numbers in this report refer to the transcription initiation site in the mouse genomic clone as determined by primer extension assays (data not shown). The position of the single intron was determined by sequence comparison of the genomic clones with the cDNAs and had the conventional sites for splice donors and acceptors (7). Southern mapping, subcloning, and DNA sequencing were performed by standard techniques (8).

Plasmid Constructions
Reporter Constructs-Zfp-36 promoter-driven human growth hormone reporter plasmids were created by ligating restriction fragments from genomic clone 13 to the 5Ј end of a promoterless human growth hormone gene. Three unique restriction fragments extending upstream from the TTP mRNA translation start site (NcoI, 5 kb; XhoI-NcoI, 2.1 kb; and SstII-NcoI, 137 bp) were excised, filled in with dNTPs, and inserted into the blunted BamHI cloning site of pØGH (Nichols Institute Diagnostics, San Juan Capistrano, CA) to produce 5kb GH , 2.1kb GH , and 137bp GH , respectively. These plasmids contained genomic DNA beginning at Ϫ5026, Ϫ2057, and Ϫ77 relative to the transcription initiation site and extending to ϩ60.
TTP mRNA Expression Constructs-Genomic Zfp-36 constructs in vector pBS ϩ were made by ligating 5Ј TTP fragments from clone 13 and a 3Ј fragment from clone 7 at the SstI site (located at base pair 330 of the mouse TTP cDNA (1). Correct annealing was confirmed by comparing the sequence with that of a Sau3AI-SalI fragment of the 129sv genomic clone, which contains a 0.9-kb 5Ј-flanking region, the intron, and the entire mRNA coding sequence that includes the endogenous polyadenylation signal and three consensus sequences for mRNA instability (1). Plasmids TTP 1.7kb and TTP 137bp constructed in such a manner contained 1.7 kb and 77 bp, respectively, of 5Ј-flanking sequences. Intronless constructs TTP 1.7kb -Int and TTP 137bp -Int were made using a 1.7-kb EcoRV-NcoI or a 137-bp SstII-NcoI promoter fragment ligated to the NcoI site of the cDNA.
Mutant Constructs-Site-specific deletions or substitutions in the TTP upstream sequence of the TTP 137bp construct were generated by using the Site-directed in vitro Mutagenesis System (Amersham Corp.). Deletion primers for each specific site are: dEGR-1, vector Sequences in parenthesis were to be deleted. Additional site-directed mutations (underlined) were introduced for cloning purposes. All constructs were sequenced to confirm that the deletions and mutations had taken place.

Cell Culture and Transfections
Primary embryonic chicken fibroblasts (CEF) were isolated by trypsinizing tissues (except for the internal organs and the head) of 11-14-day chicken embryos as described by Bergsma et al. (13) and Carroll et al. (14) and maintained in minimal essential medium (Life Technologies, Inc.) supplemented with 10% tryptose phosphate broth, 5% fetal bovine serum (FBS, Life Technologies, Inc.), 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin.
Transient transfections were performed with plasmid DNA in calcium-phosphate precipitates prepared in a transfection solution containing 140 mM NaCl, 25 mM HEPES, 0.12 mM CaCl 2 , and 0.75 mM sodium phosphate. CEF cells were plated 1 day before the transfection in 10-cm tissue culture dishes at a density of 3 ϫ 10 6 cells/dish. After removing the culture medium, 1 ml of transfection mixture containing 15 g of test plasmid and 5 g of pXGH5 was added to each dish for 30 min at room temperature. The cells were then incubated at 37°C following the addition of 9 ml of culture medium. Four h after incubation with the plasmids, the cells were treated for 4 min with 4 ml of 10% glycerol in HEPES-buffered saline and washed two times with phosphate-buffered saline to remove the remaining precipitate. A 24-h incubation in complete culture medium followed to allow the transfected DNA to be expressed. The cells were then incubated for 24 h in medium containing 0.5% FBS to synchronize cells into quiescence and were harvested to prepare total cellular RNA. Plasmid pXGH5 encoding human growth hormone driven by the mouse metallothionein-1 promoter (17) was co-transfected as an internal control for transfection efficiency. Growth hormone released was measured by the HGH transient gene expression assay (Nichols Institute Diagnostics). HIR3.5 cells were plated at the density of 1 ϫ 10 6 cells/dish and were transfected with 10 g of test plasmid as described above.

RNA Preparation and Northern Blot Analysis
Total cellular RNA was prepared following the method of Chomczynski and Sacchi (18) with modifications by Xie and Rothblum (19). Northern blots were prepared as described before (1). Blots were hybridized to random primed ␣-32 P-labeled (Stratagene) mouse TTP cDNA, and the presence of TTP mRNA was quantified using a Phos-phorImager (Molecular Dynamics, Sunnyvale, CA).
The following synthetic oligonucleotides containing potential sequences for DNA binding factors were used in this study. For EGR1-AP2, two complementary synthetic oligonucleotides were annealed to form a double-stranded oligonucleotide corresponding to mouse Zfp-36 nucleotides Ϫ77 to Ϫ37. A 5-base single-stranded tail (SalI site) was included at the 5Ј end, and a 6-bp double-stranded EcoRV site was included at the other. For TPE1, two complementary synthetic oligonucleotides were annealed to form a 22-bp double-stranded oligonucleotide corresponding to bases Ϫ68 to Ϫ55 of mouse Zfp-36, with cloning sequences for HindIII and SalI site at either end.
To ensure sequence fidelity, each oligonucleotide pair was cloned into Bluescribe and sequenced. Double-stranded oligonucleotides were released by digesting the plasmid with appropriate restriction enzymes, and the ends were filled in with [␣-32 P]dCTP (DuPont NEN) and unlabeled dATP, dGTP, and dTTP (Life Technologies, Inc.). Unlabeled dCTP was subsequently added to the reaction. The labeled DNA was separated from unincorporated radioactivity and the cloning vector by acrylamide gel purification.

RESULTS
TTP Promoter Analysis-For convenience, we will refer to the 5Ј-flanking region of Zfp-36 as the "TTP promoter." TTP promoter activity upstream of the translation start site was first tested by fusing 5-kb NcoI and 2.1-kb XhoI-NcoI fragments of the mouse genomic sequence to the human growth hormone gene, followed by assay of growth hormone (GH) released into the medium following transient transfection of the plasmids into HIR3.5 cells. Both the 5-kb and the 2.1-kb promoter sequences efficiently drove the expression of the GH gene (Fig. 1A). GH secretion from these constructs was 60 -100-fold greater than that of the promoterless plasmid pØGH. Sequential deletions of 5Ј sequences from the 2.1-kb promoter were then made to determine the minimal promoter length conferring full expression activity on the reporter construct. A 137-bp fragment (SstII-NcoI) containing only 77 bp upstream of the transcription initiation site was found to promote GH expression with similar effectiveness to the longer promoter constructs (Fig. 1B). These results suggested that this region of the TTP promoter contained the essential elements for high level expression under these experimental conditions.
Promoter Sequences Required for TTP Expression-Both the GH mRNA and protein are stable when expressed in the cell lines we have tested. This fundamental difference with the very labile TTP mRNA (1) made it difficult to measure the responsiveness of the TTP promoter to extracellular stimulation using the GH reporter system, since GH mRNA levels remained high even when the transfected cells were made quiescent by serum deprivation. A series of constructs containing the complete mouse Zfp-36 genomic sequence 3Ј from the transcription start site and varying lengths of promoter was created (Fig. 2). We chose CEF cells to study the expression of transiently transfected mammalian Zfp-36 because the endogenous chicken TTP mRNA was not detectable by the mammalian TTP cDNA probes under the hybridization conditions used (see below). Mouse Zfp-36 expression in transfected CEF cells showed a time course of mRNA accumulation and degradation (data not shown) similar to that of the endogenous mouse mRNA in HIR3.5 cells (1). Truncation of the TTP promoter to 1.7 kb, 0.9 kb, or 137 bp resulted in no decrease in serum-induced mRNA expression (Fig. 2B). In addition, transfection of human ZFP-36 constructs containing 1.4-kb and 151-bp 5Ј sequences resulted in similar patterns of serum-inducible expression (data not shown). From these findings, we concluded that the mouse TTP promoter sequences that are essential for serum responsiveness are located within a 137-bp region 5Ј of the translation start site or 77 bp 5Ј of the transcription initiation site.
Computer analysis of the 137 bp 5Ј of the translational start site revealed a consensus H2A (ϩ46 to ϩ51) binding sequence for the eukaryotic core histone dimer H2A-H2B (21)(22), and a consensus c-fos.5 binding site (ϩ52 to ϩ59) (23). Neither the H2A site nor the c-fos.5 site were present in the human and bovine TTP sequences, and neither mutation of the putative H2A binding site (from CCATTC to GTCGAC) nor the putative c-fos.5 sequence (from GCGCCACC to GGTCGACC) affected either the basal or serum-stimulated expression of mouse TTP (data not shown).
The human, bovine, and mouse sequences were highly conserved in the proximal 5Ј-flanking region (Fig. 3A). Sequence analysis revealed consensus motifs for several transcription factors in the 5Ј-proximal region of all three TTP promoters, including sites for the TATA binding protein and Sp1 binding protein (Fig. 3A). Deletion of these sites from the mouse construct containing the 137-bp 5Ј-proximal sequence produced a predictable inhibition of TTP mRNA expression (Fig. 3B). Consensus sequences for the binding of transcription factors EGR-1 (11) and AP2 (12,24) were also found in the proximal promoter region of the three animal species (Fig. 3A). To assess the contribution of the EGR-1 and AP2 elements to the induction of TTP by serum, deletion of each of these elements was made in the mouse TTP construct containing the 137-bp 5Ј-proximal sequence. The effects of these deletion mutations on serum induction of TTP mRNA expression by these plasmids were measured in starved CEF cells after transient transfection. Deletion of either of the two elements resulted in a reduced level of TTP mRNA accumulation induced by serum. Serumstimulated TTP mRNA expression in cells expressing the deleted EGR-1 or AP2 constructs was decreased to 35 and 18%, respectively, of the wild-type level (Fig. 3C).
TTP mRNA expression from constructs lacking the TTP intron was also greatly attenuated when compared with the corresponding constructs containing the same length of 5Ј sequence (Fig. 3, B and C). Full expression of human and bovine TTP constructs also required the presence of the intron (data not shown). The relationship between the promoter and the intron in the expression of Zfp-36 is not understood at present; these potential interactions are the subject of ongoing experimentation and will not be discussed further here.
Analysis of promoter sequences in this region also revealed a previously undescribed 10-bp sequence, TPE1, that was identical in the promoters of all three animal species (Fig. 3A). When this region was deleted from the mouse construct containing 137 bp of 5Ј DNA sequence, serum-induced TTP expression was decreased by 75% (Fig. 4A). In order to rule out the possibility that the diminished expression was due to nonspecific effects resulting from a shortened construct, we created a site-specific mutation at the palindromic sequence of TPE1 from TCCCGGA to TAACTTA. Serum-induced expression of the substitution mutant construct was decreased to a similar extent as that of the deletion mutant (Fig. 4A). Mutation of the same 4 bp also had profound effects on the expression of TTP from longer constructs containing 1.7 kb or 0.9 kb of promoter sequence (Fig. 4B, Table I, Group 1), indicating that this palindrome is one of the key sites for the activation of the TTP gene during serum stimulation.
The above results suggest that the EGR-1, TPE1, and AP2 elements within the first 77 bp upstream of the mouse TTP transcription start site all contribute to transcriptional activation of Zfp-36 by serum in CEF cells (Table I, Group 2). Since deletion or mutation of each single element reduced, but did not abolish, serum-induced expression, we examined whether any of these elements was able to impart serum-responsiveness to a minimal promoter. We therefore constructed plasmids containing a single copy of EGR-1, AP2, or TPE1 5Ј of a hybrid insert, Glo48TTP. This hybrid insert was constructed of a 48mer 5Ј-flanking sequence from the human ␤-globin gene and the mouse TTP cDNA. Function of this minimal human ␤-globin promoter has only been observed in vectors that include an enhancer element (9,25). The 5Ј 48-mer only contained a TATA box and no other promoter elements. We did not include the TTP intron in these plasmids because this intron may also contain enhancer activities (Fig. 3, B and C). A plasmid with the c-fos SRE sequence inserted 5Ј to Glo48TTP served as a positive control, since the SRE has been demonstrated to be sufficient for the induction by serum in heterologous constructs (10). Using transient transfection in CEF cells, we tested the ability of these putative elements, EGR-1, TPE1 and AP2, to confer serum responsiveness on the silent human ␤-globin promoter. The expression of Glo48TTP plasmids containing each of the test sequences was induced by serum as compared with the plasmid lacking these sequences (Fig. 5), both in the presence and absence of cycloheximide. In addition, the presence of FIG. 3. Deletion analysis of the mouse TTP promoter. Panel A, sequences of the first 100 bp of TTP promoter from human, bovine, and mouse are compared. The consensus sequences for the EGR-1, TPE1, AP2, and Sp1 elements and the TATA box are indicated. The mouse TPE1 sequence shown was from the Balb/c clone. The analogous sequence from the 129sv mouse genomic clone was identical to the human and bovine sequences. Panel B, TTP plasmids containing 77 bp of genomic sequence 5Ј of the TTP mRNA transcription start site (TTP 137bp ) deleted of the TTP intron (-Int), SP1, or TATA sequences, or deleted of the TTP intron (-Int), EGR-1, or AP2 sequences (panel C) were transfected into CEF cells. 24 h after transfection, the cells were deprived of serum for 24 h and then treated with control conditions (C) or 10% FBS (S) for 60 min. Northern blot analyses were performed as described in the legend to Fig. 2.   FIG. 4. Effect of deletion or mutation of the TPE1 sequence on the expression of mouse TTP. A, CEF cells were transfected with the mouse TTP plasmid containing the wild-type 137 bp 5Ј-proximal sequence or the same sequence with the TPE1 site deleted (delTPE1) or mutated from TCCCGGA to TAACTTA (⌬TPE1). B, CEF cells were transfected with mouse TTP plasmids containing varying lengths of promoter with (⌬TPE1) or without (Wt) the analogous mutation of the TPE1 site described above. Cell treatments, abbreviations, and other details are as described in the legend to Fig. 2. all three putative enhancer elements together resulted in a 1.5-fold increase in expression compared with the sum of expression from each element individually (Fig. 5). Finally, the reverse orientation of the TPE1 element resulted in a similar increase in serum-induced expression to that of the forward construct (data not shown).
Binding of TPE1 by Nuclear Extracts-Our transient transfection analysis of the 77-bp minimal TTP promoter identified three DNA elements that contribute to its serum responsiveness other than the intron and general factor binding sites such as the TATA box. These include two known DNA-protein binding motifs (EGR-1 and AP2) as well as a previously undescribed element (TPE1), all of which are highly conserved in three mammalian species (Fig. 3A). To determine if nuclear proteins could associate with these elements, we performed DNA mobility shift analyses using nuclear extracts from both serum-or insulin-treated and quiescent cells. When a radiolabeled probe (EGR1-AP2) containing the EGR-1, TPE1, and AP2 elements was used in gel shift assays, four major protein-probe complexes were identified (Fig. 6A). One of these complexes (C4) showed a reproducible decrease in intensity following 10 min of treatment of the cells with insulin or serum, whereas complexes C1 and C2 increased following insulin or serum stimulation. Double-stranded oligonucleotides encoding the individual EGR-1, TPE1, and AP2 sites were then used as competitors in gel shift assays with the EGR1-AP2 probe. No specific DNAprotein complex was displaced by the EGR-1 oligonucleotide (not shown). The top three complexes (C1, C2, and C3) were decreased in the presence of the AP2 competitor, although higher concentrations of the AP2 competitor were required to completely abolished C3. The TPE1 competitor oligonucleotide specifically abolished C4. We also observed that the intensity of C1 and C2 increased when C4 was competed by the TPE1 oligonucleotide competitor.
Wild-type and mutant TPE1 double-stranded oligonucleotides were next radiolabeled for use as probes in gel shift assays. Nuclear extracts from both serum-treated and control cells produced a single DNA-protein complex with the wildtype TPE1 probe, which corresponded in eletrophoretic mobility to complex C4 seen with the EGR1-AP2 probe. As with the EGR1-AP2 probe, formation of this complex was decreased in extracts from serum-treated cells. A mutant ⌬TPE1 oligonucleotide competitor containing base changes from TCCCGGA to TAACTTA had no effect on formation of the TPE1 complex (Fig. 6B). The mutant ⌬TPE1 oligonucleotide probe was also directly radiolabeled and shown to be unable to form this complex (Fig. 6C). These results suggested that the TPE1 element was recognized by one or more specific nuclear proteins. The small but consistent decrease in the intensity of the TPE1 complex following serum treatment, seen with both the EGR1-AP2 (C4) and the TPE1specific probe, suggests that serum treatment might modify these binding proteins in such a way as to decrease their binding to the TPE1 element. Similar decreases in intensity of complex C4 were seen when HIR3.5 cells were treated with either insulin (70 nM) (Fig. 6A) or PMA (1.6 M) (not shown) for 10 min.
FIG. 5. Effect of TTP promoter elements on the activity of a silent promoter. CEF cells were transfected with Glo48TTP alone (Ϫ), or with Glo48TTP constructs containing a single copy of the indicated TTP promoter elements or the c-fos SRE. Cells were deprived of serum for 24 h after 24 h of recovery from transfection; FBS (10%) was then added to the treatment group (S) for 60 min at 37°C, compared with control conditions (C). The position of the TTP mRNA is indicated. Other details describing the Northern blotting are contained in the legend to Fig. 2.   FIG. 6. Gel mobility shift assay for nuclear proteins binding to the TTP promoter. Panel A, nuclear extracts (12 g of protein) from HIR3.5 cells treated for 10 min with 70 nM insulin (I) or control conditions (C) were then allowed to bind to the EGR1-AP2 probe containing Ϫ35 to Ϫ77 of the mouse TTP promoter. These assays were performed with or without specific oligonucleotide competitors comprising either the AP2 or TPE1 sites, as indicated at the top of the gel. C1-C4 represent DNA-protein complexes 1-4 as described in the text. Panel B, nuclear extracts (5 g of protein) from CEF cells treated for 10 min with 10% FBS (S) or control conditions (C) were allowed to bind to the TPE1 probe. Panel C, nuclear extracts (5 g of protein) from HIR3.5 cells treated with control conditions (C) were allowed to bind to the wild-type or the mutant (⌬TPE1) TPE1 probes. All lanes contained 20 ϫ 10 3 cpm of 32 P-labeled double-stranded oligonucleotide, in the presence of poly(dI-dC) at a final concentration of 50 g/ml. The unlabeled competitors were present at 1 g/reaction when indicated. Assay conditions are described under "Experimental Procedures." The sequences of the probes used were: EGR1-AP2, ctagaGCGGGGGCGCGTCCCG-GAAGCTCTAGTGGCCACGCCCCCAGGCgatatc; TPE1, tcgacGTC-CGGGAAGCGtcga; ⌬TPE1, tcgacGTAAGTTAAGCGtcga, where the underlined bases indicate the core sequences of the consensus protein binding sites, and the lowercase bases indicate portions of the restriction sites used for subcloning. These studies establish that the first 77 bp 5Ј upstream of the transcriptional start site are sufficient for maximal serum induction of the mouse TTP gene (Zfp-36) when expressed in CEF cells; deletions 3Ј of this point dramatically decrease serum inducibility of this gene. Within this minimal effective promoter, we have also identified several putative transcription factor binding sites in the mouse TTP promoter, all of which are present in the human and bovine genes. The presence of each is necessary for the full, serum-inducible expression of the gene. Finally, we have identified binding proteins in cell nuclear extracts that bind specifically to some of these DNA motifs and whose binding is altered by prior treatment of the cells with mitogens. These studies have begun to evaluate the mechanisms by which serum and other mitogens rapidly and dramatically stimulate the transcription of this immediateearly response gene.
One conserved transcription factor binding site identified in the present study is the Sp1 site, located at Ϫ35 to Ϫ30 5Ј of the transcription start site in the mouse promoter. Sp1 is a well characterized zinc finger-containing transcription factor, which enhances transcription by RNA polymerase II from promoters that contain at least one properly positioned GGGCGG hexanucleotide (for review, see Ref. 26). Sp1 has been shown to regulate transcription of certain proto-oncogenes (27,28) and growth factor genes (29). When the consensus binding sequence for Sp1 was deleted from the TTP 137bp plasmid, an 80% decrease in TTP expression resulted, indicating that the TTP promoter is Sp1-responsive (Fig. 4B); however, we have not demonstrated directly that Sp1 binds to its hexanucleotide binding site in the TTP promoter. The TTP promoter construct with the Sp1 site deleted remained serum-responsive but to a lesser extent than the wild-type construct, implicating other promoter elements in the serum-induced expression of TTP mRNA.
EGR-1 is another zinc finger-containing transcription factor, also known as NGF1-A (30), KROX24 (31), TIS-8 (4), and Zif268 (32). It binds to a GC-rich consensus sequence, GCGGGGGCG, that is found in the 5Ј-flanking regions of many genes involved in cell growth such as proto-oncogenes and genes encoding mitogens and mitogen receptors. Several immediate-early response genes also have EGR-1 binding sites in their promoters (33). Deletion of the EGR-1 binding sequence from the TTP promoter decreased its serum-stimulated expression by 65%. Although we found no direct evidence of EGR-1 binding to the TTP promoter in our gel shift assays, our deletional analysis indicates that EGR-1 may contribute to the regulation of TTP expression.
The AP2 consensus binding sequence GCCNNNGGC (34) is present in the minimal effective promoter of TTP from all three animal species tested. This sequence, when bound by AP2 homodimers, has been identified as a control element for several viral and cellular genes (24,34). AP2 mediates regulation of gene expression in response to a number of different signal transduction pathways (35). The activity of AP2 is increased in response to treatment of cells with phorbol esters and agents that elevate cAMP levels (34 -36). When the AP2 consensus sequence in the mouse TTP promoter was deleted, induction of TTP expression by serum was decreased by 72%. We have previously shown that PMA could induce TTP expression (1), making it possible that the AP2 binding site was involved in PMA-induced TTP expression. However, when the AP2 binding sequence was deleted from the TTP 137bp construct, PMA still induced TTP expression to a similar extent as the seruminduced response (data not shown). These results indicate that the PMA-stimulated TTP expression does not depend solely upon the consensus AP2 binding sequence in the TTP promoter.
We also identified a previously undescribed promoter element at Ϫ66 to Ϫ60 5Ј of the cap site in the mouse gene that we have called TPE1. It contains a palindromic element with dyad symmetry, TCC(C/G)GGA. It is perfectly conserved in the TTP promoter from all three animal species we have studied. Deletion or mutation of this element led to a 75% decrease in serum-induced TTP expression. Mutation of the TPE1 palindrome also severely impaired serum responsiveness of the TTP promoter when introduced into longer promoter constructs. Both orientations of the TPE1 element could also confer serum responsiveness to the silent promoter Glo48, indicating that this element behaves as a transcriptional enhancer.
The TPE1 palindrome also appears to represent a binding site for a nuclear protein, as demonstrated by gel shift analysis. Mutations within the palindrome that impaired serum responsiveness of the promoter also abolished binding of this nuclear protein. In addition, nuclear extracts from cells stimulated with serum or other mitogens for 10 min showed a small but consistent decrease in band intensity in the gel shift assays. This suggests, but does not prove, that the TPE1 element binding protein is a target of the signaling cascade responsible for serum induction of the TTP promoter.
We have located consensus TPE1 binding sites in the promoters of a number of genes, including c-ha-ras1 (27), mdm2 (37), and fosb (38). fosb is induced by serum with early response gene characteristics. It will be interesting to investigate the possibility that this motif is involved in the regulated expression of these and other genes.
Because the serum-responsive region in the TTP promoter contains multiple potential promoter elements, it is likely that a number of nuclear proteins interact in its regulation. The DNA mobility shift assay using the EGR1-AP2 probe provides evidence to suggest that several proteins interact with the Ϫ77 to Ϫ35 sequence to mediate the serum response. We observed small but consistent changes in the intensity of a number of protein-DNA complexes using nuclear extracts from serumtreated and control cells. Serum-induced transcriptional activation of TTP is likely to result from post-transcriptional modification of pre-existing nuclear protein complexes, since TTP gene transcription is very rapidly stimulated by insulin or serum, even when cells have been pretreated with the protein synthesis inhibitor cycloheximide (1). Post-transcriptional modification of such factors following serum treatment could result in changes in the affinity of these factors for their DNA binding sites. Isolation and characterization of the TPE1 protein should lead to a better understanding of the transcription factor interactions that are involved in the regulation of TTP expression.
Our results suggest cooperative interactions among nuclear proteins binding to closely positioned cis-acting elements in the TTP promoter. For example, the EGR-1, TPE1, and AP2 elements from the TTP promoter each conferred only low levels of serum-inducible expression of ␤-globin-TTP hybrid constructs, but together they produced greater than additive expression. The gel mobility shift data also suggest the possibility of cooperative interactions among several transcription factors; for example, there was a reciprocal increase in the binding of nuclear proteins to the AP2 binding site and decreased protein binding to the TPE1 site. Finally, the presence of the single intron in the TTP gene markedly enhanced serum induction of the gene; we suspect this is due to enhanced transcription, although we cannot exclude effects of the intron on processing or stability of the mRNA (39). Further experimentation will be necessary to determine whether the intron and/or proteins that bind to it interact in some way with the other serum-responsive elements of the TTP promoter.