c-Jun Represses Transcription of the Human Chorionic Gonadotropin Q! and p Genes through Distinct Types of CREs*

Chorionic gonadotropin (CG) is a heterodimeric pla-cental hormone encoded by separate (Y and p subunit genes that is essential for the maintenance of pregnancy. The production of CG is stimulated by DNA synthesis inhibitors and by CAMP. The present study demon- strates that the proto-oncogene c-jun represses transcription of the human CGa and CGP promoters. c-dun repressed the CGrv promoter through a canonical CAMP response element (CRE) that is known to bind c-Jun and other members of the B-Zip transcription factor family. In the CGP promoter, two adjacent sites, CREl (-299 to -289) and CRE2 (-240 to -219), conveyed CAMP respon- siveness via sequences that are distinct from the canonical element, TGACGTCA. Mutations within CGP CREl or CRE2 reduced or abolished, respectively, c-Jun-medi-ated repression. Although the CGP CREs do not contain consensus sequences previously described to bind c-Jun, CRE2 bound c-Jun and c-Fos in electrophoretic mobility shift assays. Supershift assays, using anti-JUN antibody, demonstrated that Jun formed part of the na- tive complex that binds the CRE2 in JEG-3 cells. A series of c-Jun mutants were used to analyze the transcription factor domains required for repression of the CG subunit promoters. The DNA binding and leucine zipper domains of c-Jun as well as the amino terminus,

genes that is essential for the maintenance of pregnancy. The production of CG is stimulated by DNA synthesis inhibitors and by CAMP. The present study demonstrates that the proto-oncogene c-jun represses transcription of the human CGa and CGP promoters. c-dun repressed the CGrv promoter through a canonical CAMP response element (CRE) that is known to bind c-Jun and other members of the B-Zip transcription factor family.
In the CGP promoter, two adjacent sites, CREl (-299 to -289) and CRE2 (-240 to -219), conveyed CAMP responsiveness via sequences that are distinct from the canonical element, TGACGTCA. Mutations within CGP CREl or CRE2 reduced or abolished, respectively, c-Jun-mediated repression. Although the CGP CREs do not contain consensus sequences previously described to bind c-Jun, CRE2 bound c-Jun and c-Fos in electrophoretic mobility shift assays. Supershift assays, using anti-JUN antibody, demonstrated that Jun formed part of the native complex that binds the CRE2 in JEG-3 cells. A series of c-Jun mutants were used to analyze the transcription factor domains required for repression of the CG subunit promoters. The DNA binding and leucine zipper domains of c-Jun as well as the amino terminus, were required for repression of both subunit promoters.
Thus, both the CGa and CGP genes are repressed by c-Jun through promoter regions that convey cAMP-induced transcription, although these DNA sequences are unrelated.
The kinetics of CGa and CGP induction by CAMP treatment are distinguishable. The induction of CGa is rapid, occurring within a few hours, whereas CGP is induced in a delayed manner increasing up to 16 h after treatment with CAMP (5,6). The minimal DNA elements mediating CAMP responsiveness of the CGa (3, 4) and CGP (5) genes have quite different sequences. Several members of the CREB/ATF/AP-1 superfamily (CAMP responsive element binding proteins/activating transcription factor/activator protein-1), including c-dun, bind to the CGa CAMP responsive element (CRE) in vitro (14,15), raising the possibility that c-Jun may function in transcriptional regulation of the CGa subunit. Although the sequences of the CGP promoter are protected by nuclear proteins in DNase 1 footprinting studies (5, 16), the factors that bind to the CG@ CREs in native extracts are currently unknown.
c-Jun is a member of the transcription factor complex, AP-1, and was originally described as a DNA-binding activity that recognized the enhancer elements of the SV40 and human metallothionein promoters (17)(18)(19). AP-1 mediates basal level enhancer function and transcriptional activation in response to serum or phorbol esters (17)(18)(19). c-dun is important in promoting progression of the cell-cycle including the DNA synthesis or S phase (20). In addition to its role as a positive trans-acting factor, AP-1 is a negative trans-acting factor of several genes including c-Fos (21-23), human osteocalcin (24), adipocyte P2 (25), the muscle creatinine kinase enhancer, the MyoD promoter (26,27), major histocompatibility class 1 gene expression (28), and the human insulin promoter (29).
The molecular mechanisms governing transcriptional regulation by c-Jun are complex. The locations and functional roles of different activation surfaces remains an area of controversy. The c-Jun protein structure consists of a basic domain, a leucine zipper region and several functionally distinct transcriptional activation domains, referred to as Al, A 2 , and epsilon (30). These activation surfaces may be targets of distinct kinase pathways (31). Critical regulation of AP-1 activity involves post-translational modification (31) through phosphorylation by several kinases including casein kinase I1 (32, 33), mitogen-activated protein kinase, protein kinase C, and cell cycle regulatory kinases such as cyclin B/~34"~'' (31,33). In addition, the transcriptional effect of c-Jun is determined by cellular regulatory proteins (34) including inhibitory protein 1 (IP-1) (35), redox-dependent factor , and c-Jun-interacting factors (Jifs) (37). The factors which determine whether AP-1 behaves as a positive or negative regulator of gene transcription vary with the cell type and promoter, but together these interactions allow c-Jun to occupy a pivotal role in dictating the cellular genetic response to an array of diverse agents (38).
The capacity of c-Jun to act as a transcriptional repressor may be determined in part by its capacity to interact with Danscriptional Regulation of CG Genes by AP-1 31091 several different types of nuclear proteins. Specific regions of the c-Jun protein are required for interactions with these transcription factors. The c-Jun leucine zipper, for example, is required for interaction with members of the CREBIAP-1 superfamily (39) that bind to the CRE (14), with several steroid hormone receptors including the glucocorticoid receptor (40) and with members of the NFKB family (41,42). The amino terminus of c-Jun, on the other hand, is required for dimerization with helix-loop-helix proteins such as MyoD (26,27).
Given the pivotal role of c-Jun in mediating signal transduction by a wide variety of different stimuli (38), we hypothesized that c-Jun or components of the AP-1 complex may participate in transcriptional regulation of the CG genes. We demonstrate that c-Jun is a specific negative regulator of both subunit promoters and that this effect is conveyed through the CAMP responsive regions.

MATERIALS AND METHODS
Construction of Plasmid Vectors-The a subunit constructs have recently been described (43). The -846 aLUC construct contains 846 bp of 5"flanking sequence and 44 bp of exon 1 of the human glycoprotein a-gene linked to the luciferase gene in the plasmid p&LUC (44,45). The vector p&LUC contains a trimerized SV40 poly(A) termination site, which abolishes transcriptional readthrough (45), and does not contain AP-1 responsive vector plasmid backbone sequences (46). The a enhancer sequences from -195 to -98 were subcloned as a HindIILKpnI fragment into HindIIIIKpnI restricted TK81LUC (thymidine kinase) (47) to create the vector -195/-98 a TK LUC. Aseries of point mutations of the human CGa CRE were created by subcloning a HindIII fragment containing -172 to +44 bp of the a promoter or the site directed mutants of the CRE previously described (15) (see Fig. lB) into HindIII restricted pA,LUC (see Fig. 1B). The site directed mutants created in the context of -172 of the a subunit promoter are -172ml (one wild type CRE (TGACGTCA) and a deletion of one CRE), -172m2 (a deletion of one CRE with the second CRE mutated to cTGAtGTCA), -172m3 (a deletion of one CRE with the remaining sequence TGAgGTCA), -172m4 (a deletion of one CRE with the second CRE mutated to TGAtGTCA).
Three additional mutations of the a CRE were created in the context of the -846 a promoter. -846m1, includes a deletion of one CRE with the second CRE mutated to cTGAtGTCA, -846m2, includes a deletion of one CRE with the remaining sequence TGAgGTCA, and -846m3 includes a deletion of one CRE with the remaining sequence TactagtA.
The series of 5' deletions of the CGP subunit gene promoter in the plasmid p&LUC (44) were derived from the 3.7 kilobases of 5"flanking region and 362 bp of the 5"untranslated tract (UT) of the CGP gene (48). Constructs were made in the context of either the full-length (+362), +lo4 bp or +4 bp of the CGP UT to examine whether elements within the UT were involved in transcriptional regulation by c-dun.
The expression vectors CMV c-Fos and CMV c J u n (50), the wild type and mutant c-dun protein expression vectors (34), the AP-1 responsive reporter containing three AP-1 sites 3TpLUX (51) and the wild-type and mutant catalytic subunit of protein kinase A (52), have been previously described. The c-dun mutants N22, N51, and N91 encode cJun proteins that are truncated at the amino-terminal end by 22, 51, and 91 amino acids. The c-dun mutant that is defective in dimerization ( d u n dimermutant) has amino acid substitutions at positions 303 and 317. c-dun DNA-has double amino acid substitutions at positions 277 and 278 (see Fig. 2). pGEM c-Fos was constructed by subcloning the BamHIIHindIII insert containing the complete c-Fos cDNA from CMV c-Fos, into BamHIIHindIII restricted pGEM3Z (Promega, Madison, WI). pGEM c-Jun was constructed by subcloning the EcoRI insert containing the complete c-dun cDNA from CMV c-dun into EcoRI restricted pGEM3Z (Promega).
Cell culture, DNA transfection, and luciferase assays were performed as described previously (5). Cells were transfected by calcium phos-phate precipitation, the media was changed after 6 h and luciferase activity was determined after a further 24 h. At least two different plasmid preparations of each construct were used. In co-transfection experiments a dose response was determined in each experiment with 40, 60, 80, 100, and 200 ng of expression vector and the CGP or a subunit promoter reporter plasmids (1.6 pg). The -fold effect was determined for 80 or 100 ng of expression vector and statistical analyses was performed using the Mann Whitney U test. Significant differences were established a s p < 0.05.
Electrophoretic Mobility Gel Shift Assays-The wild type and mutant CGp CREs, the wild type consensus Ap-1 (AP-lw) site and a mutant Ap-1 (AP-lm) site were synthesized for EMS& as complementary oligodeoxyribonucleotide strands. The sequence of the sense strands of the CGP promoter CREl region oligodeoxyribonucleotides was GCC CCG TGG GCA GGA CAC ACC TCC TGC GGG CC, CGp promoter CRE2 region oligodeoxyribonucleotides, CRE2, was 5"CAC GCA TTT CCG GGG ACC GCT CCG GGC ATC CTG GCT TGA GGG TAG AGT-3'. The sequence of the oligodeoxyribonucleotide of the 5' portion of CRE2, CRE2m4 (-244 to -220) was CAT TTC CGG GGA CCG CTC CGG GCA T and the mutant of this sequence CRE2m5 was CAT TTC aGt GGA CCG CTC aGt GCA T. The AP-lw oligodeoxyribonucleotide sequence (AP-1 site is shown in boldface) was TCC ATT CTG ACT CAT T' IT TTT TAA, and AP-lmt was TCC ATT CTG cCg CAT T l T n ' T TAA. The in EMSA were CGP-4 5' GAT CTC AAT CCA GCA CTT TGC TCG GGT sequences of additional oligodeoxyribonucleotides used as competitors CAC GGC CTC CTC CG and the AP-3 oligodeoxyribonucleotide was 5' CTA CTG GGA CTT TCC ACA CAT C 3'.
The preparation of JEG-3 cell nuclear extracts and EMSA using nuclear extracts (5) or in vitro translated proteins (53) were performed essentially as described previously. The cDNAs were transcribed in vitro and translated using the TNT-coupled reticulocyte lysate system according to the protocol of the suppliers (Promega, Madison, WI). The programmed lysates (5 pl) were incubated in a reaction mix (20 pl) consisting of 20 mM HEPES, pH 7.8, 50 mM KCl, 1 mM EDTA, 10% glycerol, 10 mM dithiothreitol, and 50 &ml poly(d1-dC) at room temperature for 15 min. 32P-Labeled oligonucleotides (50 fmol, 50,000 cpm) were added to the reaction and incubated at room temperature for a further 15 min. The protein-DNA complexes were analyzed by electrophoresis through a 5% polyacrylamide gel, with 0.5 x Tris borate, EDTA buffer (TBE: 0.045 M Tris borate, 0.001 M EDTA) and 2.5% glycerol.

RESULTS
The a Subunit Promoter CRE Is a Target of c-Jun repression-Co-transfection experiments were conducted using the 5' deletions of the a subunit promoter in JEG-3 cells (Fig. 1). The basal activity of the -846 aLUC reporter was repressed 5-7fold by co-expression of c-Jun. Using a series of 5' a promoter deletions, the minimal region repressed by c-Jun was localized between -132 and -99 bp (Fig. IA). Repression of the -132 aLUC reporter construct was observed using either the human or rat c-Jun driven by either the RSV or CMV promoters, respectively. This effect was not observed using equal amounts of the RSV or CMV constructs without c-Jun (data not shown).
The -99 aLUC (Fig. lA) and -59 aLUC reporters (not shown), which do not contain the CRE, were not repressed by c-Jun. The AP-1 responsive reporter, 3TpLUX, was activated 7-fold by c-Jun, but the promoterless vector pA,LUC was not affected by co-transfection with the c-Jun expression vectors or by CAMP (not shown). These findings indicate that c-Jun represses CGa, and that sequences including the CRE are required for repression.
Point mutations within the CRE, in the context of the a subunit promoter, reduced the magnitude of repression by c-Jun. The deletion mutant, -172ACREaLUC (which deletes one of the repeated CREs), was repressed 90% of wild type, indicating that a single CRE is sufficient for repression by d u n . Point mutations of the remaining CRE reduced the magnitude of repression to less than 30% of wild type (-172ACREm1, -172ACREm2) (Fig. LB). When the CRE was mutated in the context of the -846 a subunit promoter, a single CRE conveyed repression, and mutations of this site (-846ACREm1, -846ACREm2) reduced repression.
Deletion of both CREs  (34). Previous studies demonstrated that the DNA-binding activity of these mutants, (other than DNA-), are similar to wild type (54). Using Western blot analyses these mutant proteins have been shown to be expressed similarly in transfected cells (34).

B-anscriptional Regulation of CG Genes by AP-1
Co-transfection experiments were conducted with the c-Jun mutants in conjunction with either -132aLUC or -195/98a TK81LUC. The amino-terminal mutants, N22 and N51, were impaired in their ability to repress -132 aLUC ( Fig. 2A). The extent of repression was reduced by 90% upon deletion to N91. A similar loss of repression was observed when examined using -195/98a TKLUC (Fig. 2B). Mutation in either the DNA binding or dimerization domains (34) abolished repression of either -132aLUC or -195/98a TK81LUC, indicating that both domains are required for repression of CGa transcription. The repression of either -132aLUC or -195/98a TK81LUC by c-Jun was unaffected by deletion of the A2 activation domain (aa 197-248) (Fig. 2). c-Jun Represses the CGP Subunit Promoter through the CREs-cAMP stimulated the -3700 CGP promoter activity by 8-10-fold. To determine whether CAMP responsive elements were present in the 5"untranslated region (5'-UT), 3' deletions of the 5'-UT were constructed within the context of the 345-bp promoter fragment. There was no significant difference in CAMP-induced transcription with sequential 3' deletions of the 5'-UT (Fig. 3B). Using the CGP subunit 5'promoter fragments, CAMP-induced LUC activity was reduced 4-5-fold upon deletion from -311 to -282 (Fig. 3B). An additional region conveying CAMP responsiveness was illucidated using site-directed mutagenesis of two regions between -311 and -210 that bind JEG-3 nuclear extracts as demonstrated by footprinting and EMSA (5, 16). Mutation of either of the two footprinted domains, designated CREl (-310/-279) (-345mlpLUC) and CRE2 (-250/-200) (-345m2PLUC and -345m3PLUC) caused reduced CAMP-induced expression (Fig. 3C). Similar results were obtained when the CGP 5'-promoter fragments and CGP CRE mutants were examined in the presence of co-transfected PKA catalytic subunit, but not with the PKA mutant vector (data not shown). These findings confirm that the CGP CAMP responsive region is a composite response element consisting of two subdomains, referred to as CREl and CRE2.
c-Jun repressed the activity of the -3700 CGP promoter 4-5fold (expressed as wild type 100% repression in Fig. 4-41. There was no significant loss of c-Jun-mediated repression upon sequential 3' deletions of the 5'-UT (Fig. 4-4). Using a series of 5' deletion mutants, the minimal region required for c-Jun-medi- ated repression was localized between -345 and -282, with further deletions resulting in c-Jun mediated stimulation. The promoter fragment, -210, was stimulated approximately 2-fold by c-Jun, however -187 and -87 PLUC were stimulated minimally by c-Jun. In contrast with the wild type -345 PLUC, which was repressed by c-Jun, mutations within the CREs either reduced or reversed this effect (Fig. a). c J u n repressed the clustered point mutant in the CREl region (345mlpLUC) only 50% ofwild type. In contrast, two different mutants in the CREB domain (-345m2 and -345m3PLUC) lost all repression and were stimulated by c-Jun. Thus CREB appears to be the primary locus of c-Jun-mediated repression of the CGP promoter.
The Domains of c-Jun Required for Repression of the CGP Promoter-The series of mutations in the human c-Jun cDNAs (34) were also used to examine the domains required for c-Jun repression of the CGP subunit. Mutation of t h e c J u n DNA binding domain abolished repression, and caused 4-5-fold activation of the construct, -345PLUC (and -3700 PLUC, data not shown) (Fig. 5). Similarly, mutation of the dimerization domain abolished repression, demonstrating the requirement for these regions of c-Jun in modulating CGP subunit basal transcription. Repression by the amino-terminal mutants, N22 and N51, was reduced (Fig. 5), and there was a 90% loss of repression upon deletion to N91. In contrast to the effect of c-Jun on the a subunit promoter (above), deletion of the A2 activation domain   may be involved in repression of the a and P subunit promoters (Fig. 5).
c-Jun Binds the CGP CRE2-The results of the co-transfection experiments implicated the CREs in c-Jun mediated repression of CGP transcription. The nucleotide sequences of these two regions are dissimilar and do not resemble previously defined AP-1 sites. Because repression by c J u n required the c-Jun DNA binding domain, the ability of c-Jun t o bind the CREs was assessed using EMSA. Equivalent amounts of c-Jun or c-Fos, as determined by polyacrylamide gel electrophoresis, were incubated with the CGP CRE probes. c-Jun alone, or c-Jun and c-Fos, bound to the CGP CRE2 (Fig. 6 A ) or a consensus AP-1 site (not shown), but not to the CGP CREl (Fig. 6A). The CGP CRES complex was competed by the addition of unlabeled CGP CRES competitor (Fig. 6 A ,  teins, EMSA were performed with CRE2m4 (-244 to -220) or mutated sequences within this region (CRE2m5). CRE2m4 bound d u n and c-Fodc-Jun, however, CRE2m5 did not (Fig.   6B ). These findings indicate that sequences between -244 and -220 bp are involved in the formation of the AP-1 complex on CRE2 in vitro.
Protein binding to the CGP CREl and CRE2 probes was also performed using native JEG-3 nuclear extracts. The CREl probe formed two major complexes which were competed by the CREl site, but not mutant sequences (Fig. 7A). The CRE2 probe formed two major complexes that were competed by a 50-fold molar excess of wild type, but not unrelated sequences. The upper complex was partially super-shifted by the addition of anti-JUN antibody (Fig. 7B, lane 9). This JUN antibody cross-reacts with several different Jun proteins including JunB, junD, and d u n . * These studies indicate that a component of the proteins binding this site in JEG-3 cells consists of proteins antigenically related to JUN.

DISCUSSION
These studies demonstrate that transcription of the human CGa and P subunit promoters is repressed by c-Jun. The sequences responsible for repression by c-Jun in both the a-and P-promoters co-localized with the regions conveying cAMP-induced transcription. The c-Jun DNA binding and amino-terminal activation domains were required for repression of both subunit promoters, while the A2 activation domain was involved in repression of the P, but not the a subunit promoters.
The finding that the c-Jun DNA binding domain is required for repression of the a subunit is consistent with a model in which c-Jun interacts directly with the CRE of the CGa promoter. Members of the AP-1 transcription factor family bind the V. J. Baichwal, personal communication. (14), and several different and as yet uncharacterized proteins bind a variant a subunit CRE that is found in non-primate species (15). Although the c-Jun homodimer binds the CGa CRE poorly when compared with an AP-1 site in vitro, several other members of the ATF family, such as ATF-2, can heterodimerize effectively with c-Jun (39, 55, 56). ATF-2 may also dictate the ability of c-Jun to bind the CGa CRE as d u n homodimers and c-Junk-Fos heterodimers preferentially bind the AP-1 site, but cJudATF-2 heterodimers bind the CRE with greater affinity than an AP-1 site (39,5547). As CREB is an important basal level enhancer of the CGa promoter through the CRE, it would be expected that displacement of CREB from the aCRE by a c-JunlATF complex would reduce basal level a-promoter activity as observed in these studies.

CGa CRE in vitro
An element in the human c-jun promoter at -72 (TGA-CATCA), which resembles the a CRE core element TGACGTCA, is also a target for dual regulation by CREB and AP-1 (58). In contrast with the a subunit promoter, which is positively regulated by CAMP and negatively regulated by d u n , the cjun promoter is negatively regulated by CREB and positively regulated by AP-1. In this manner, a single cis element provided positive or negative regulation depending upon the relative abundance of active CREB/ATF or AP-1 binding at the site. In an analogous manner, co-transfected c-Jun may compete with other members of the CREB/ATF family for binding to the a CRE, thereby reducing basal level expression. The dynamic equilibrium between the relative abundance of CREB/ ATF and c-Jun could thereby modulate a subunit expression in a given cell.
Dimerization between cdun and members of the FodJunl CREB family occurs through noncovalent bonds between homologous leucine zipper motifs that are adjacent to the basic DNA-binding domains (39). The dependence upon the leucine zipper of cdun for repression of CGa implies that dimerization  of c-Jun is required to form a functional transcriptional complex, probably because dimerization is required for high aftinity DNA binding. The interaction of c-Jun with other B-Zip proteins such as ATF-2 (39), the helix-loop-helix protein MyoD (26) or the high mobility group (HMG1) domain (59) requires an intact leucine-zipper domain. The abolition of CGa repression by mutations in the c-Jun zipper domain is consistent with a model in which transcriptional repression by c-Jun is mediated by the formation of a transcriptional repressor complex with another protein.
The effect of the d u n DNA-and zipper-mutants was to activate transcription rather than simply abolish repression. The binding of the d u n DNA-mutant to endogenous d u n may form nonproductive complexes, preventing endogenous c J u n from repressing CG transcription resulting in the activation of gonadotropin subunit transcription. Alternatively, if endogenous c-Jun requires a cellular factor for repression, cotransfected d u n zipper-may still be able to bind such a repressor, converting the endogenous d u n aCRE bound complex into a transcriptionally activating complex. Although it is currently unknown whether co-transfected c-Jun zipper-could titrate out such a repressor, several accessory factors which bind c J u n and modulate its transcriptional activity in a cell type specific manner have been described (34-37).
Co-transfection of the amino-terminal cdun mutants N22 and N51 resulted in a small reduction in repression of either the -132aLUC or -195/98a TK81LUC reporter (Fig. 21, with a further reduction in repression upon deletion to N91. Although previous studies using Western blot analyses demonstrated that equal amounts of mutant protein were expressed in transfected cells (341, we cannot entirely exclude the possibility that these mutant proteins are not expressed as well as wild type in all cell lines which contain endogenous d u n . Dose response curves, however, with the DNA-, Dimer-, N21 and N51 mutant proteins demonstrated no increase in repression with a &fold increase in the amount of transfected vector compared with the wild type protein (data not shown). Together these findings suggest several amino-terminal domains of d u n , in particular the region between amino acids 51 and 91, are involved in repression of the a subunit.
The mechanisms by which the c-Jun amino terminus modulates transcription is controversial. The amino-terminal mutants sequentially delete the trans-activation domains (TAD) I, 11, and I11 (60). The region of c-Jun between amino acids 51 and 91 includes the "homology box 1" (HOB1) activation domain (amino acids 67-77) (61). Phosphorylation of Ser-63 and Ser-73 which occurs after activation by protein kinase C, by the expression of several proto-oncogenes including v-sis and v-src, and by a specific Jun-kinase (62), results in increased AP-1 activity (63-66). Deletion of the amino-terminal57 amino acids of c J u n abrogated phosphorylation of c-Jun at Ser-63 and Ser-73 by ~3 4 '~'~ (33) and deletion of the amino terminus (amino acids 1-87) prevented transcriptional activation by Ha-Ras (67). It seems likely that one activation domain required for repression of CGa corresponds to the HOB1 domain. This domain may be the target of one of these kinases described above.
Two domains of the CGP promoter within the region -311 to -210, which are footprinted by JEG-3 cell nuclear extracts, designated CREl and CRE2, each contribute to full CAMP responsiveness and repression by c-Jun (Figs. 3 and 4). The relative importance of the CREs in regulation by either CAMP or by c-Jun were distinguishable. CREl appears to be the primary site of regulation by CAMP and CREB the primary site of d u nmediated repression although there are clearly interactions between the two CREs that denote these sequences as composite response elements. Thus CAMP responsiveness determined using the 5'-promoter deletions demonstrated a significant reduction in CAMP responsiveness upon deletion of CREl but not CRE2, however internal deletion of CREZ reduced CAMP responsiveness. CREI therefore conveys a component of CAMP responsiveness in the presence of CRE1, perhaps by interacting with factors binding to these sequences.
The analyses of the 5'-promoter deletions demonstrated that deletion of CREl reduced dun-mediated repression 90%, consistent with a role for CREl in this effect. However, internal deletion of CREl reduced d u n repression only 50% and internal deletion of CREB abolished c-Jun-mediated repression, implying that both CREs were required for repression by c-Jun. c-Jun was capable of binding the CREB in EMSA, and the minimal sequence required for binding is located between -244 and -220. Based on super-shift studies it appears that JUNrelated proteins form part of the complex that bind to CREB in JEG-3 cells. As both CREs are capable of modulating the effect of c-Jun on CGP transcription the effect through CREl appears to be indirect. Although neither CREl nor CRE2 resemble a known AP-1 site, an increasingly wide variety of sequences have been implicated in transcription by AF"1 and related proteins. The DNA elements responsible for repression by cJun in other promoters include AP-1 related sequences (28), the helixloop-helix site (26,27), the NF-1 site (68), the nuclear receptor binding site (40, 69) and the CRE (29).