Regulation of the Human Chorionic Gonadotropin α- and β-Subunit Promoters by AP-2*

Production of the placental hormone, chorionic gonadotropin (CG), increases dramatically as cytotrophoblasts fuse to form syncytiotrophoblasts. The CG α- and β-promoters are both responsive to cAMP, although the kinetics of cAMP stimulation are different. In an effort to understand the mechanisms of coordinate induction of these genes, AP-2 binding sites were identified in the promoter regions of the α and CGβ genes. AP-2 bound to the upstream regulatory element (−186 to −156 base pairs (bp)) in the α-promoter and to several different regions of the CGβ promoter, including footprints 2 and 4B (FP2, −311 to −279 bp; FP4B, 221 to −200 bp). AP-2 antibodies induced supershifts of these complexes, confirming the identity of the protein-DNA complex. In JEG-3 cells, which contain abundant AP-2, mutations in these CGβ AP-2 sites reduced basal activity and decreased cAMP stimulation. In AP-2-deficient Hep-G2 cells, co-transfection of AP-2 stimulated expression of the CGβ promoter 10–20-fold, and the α-promoter was induced by 3–6-fold. Mutations that eliminate AP-2 binding to CGβ FP4B reduced AP-2 stimulation by more than 80%, whereas mutations in FP2 reduced AP-2 stimulation by less than 50%. Analyses of AP-2 mutants revealed a requirement for the DNA binding/dimerization domain and the amino-terminal proline-rich and acid-rich transactivation domains for stimulation of the CGβ promoter. Primary cultures of placental cytotrophoblasts were differentiated into syncytiotrophoblasts in vitro to examine AP-2 expression by reverse transcriptase-polymerase chain reaction. AP-2 mRNA levels increased by day 2 and continued to rise in parallel with a marked increase in α and CGβ gene expression. We conclude that both the α and CGβ promoters contain binding sites for AP-2 and suggest that this transcription factor provides a mechanism for coordinating the induction of these genes during placental cell differentiation.

Human chorionic gonadotropin (hCG) 1 is a heterodimeric placental hormone encoded by separate ␣and CG␤-subunit genes (1)(2)(3). It is a member of a family of hormones that are expressed in the pituitary (luteinizing hormone (LH), folliclestimulating hormone, and thyroid-stimulating hormone) and the placenta (CG). The ␣-subunit is common to these hormones, and it is expressed in both pituitary and placenta. The CG␤ gene is expressed almost exclusively in the placenta (3). The function of CG is to stimulate the corpus luteum in the ovary to produce progesterone during the early stages of pregnancy.
The dramatic exponential increase in CG expression in early pregnancy correlates with the formation of differentiated placental cells (4,5). Trophoblast progenitor cells convert to proliferative cytotrophoblasts that invade the endometrium of the uterus. The cytotrophoblasts fuse to form nonmitotic syncytiotrophoblast cells. The production of CG is greatly enhanced upon the formation of syncytiotrophoblasts, which also produce a variety of other hormones including placental lactogen (5).
The cellular pathways that lead to activation of the ␣ and CG␤ genes have not been clearly established, although cAMP is able to induce expression of these genes in both placental cells (6) and choriocarcinoma cell lines (7,8). Cyclic AMP-responsive DNA sequences have been characterized in the promoters of both genes (for review, see Ref. 9). In the ␣-gene, two identical repeats of a consensus cAMP response element (CRE) are located between Ϫ146 and Ϫ111 bp of the promoter (10 -12). These CREs bind cAMP response element-binding protein (12)(13)(14) along with other members of the B-Zip family of transcription factors (15,16). An adjacent element, termed the upstream response element (URE, Ϫ180 to Ϫ151), also contributes to basal expression and appears to contribute to placenta-specific expression of the ␣-promoter (12,13,(17)(18)(19)(20). The URE contains three overlapping protein binding sites referred to as the trophoblast-specific element (TSE, or URE2 (Ϫ187 to Ϫ159)) (12,13,18,21), downstream domain (Ϫ172 to Ϫ151) (13,19), and GATA (␣-ACT, URE1) (Ϫ165 to Ϫ140) (22). Protein binding to the TSE and downstream domain are mutually exclusive (13,19).
The cAMP-responsive region in the CG␤ promoter encompasses several protein binding domains between Ϫ311 and Ϫ200 bp (23)(24)(25). Maximal expression in placental cell lines and stimulation by cAMP requires this entire region, suggesting that it functions as a composite regulatory element (23).
Nuclear extracts footprint two major regions (Ϫ311 to Ϫ274; Ϫ250 to Ϫ200) within this domain (23,24). Neither region binds transcription factor cAMP response element-binding protein (23), nor are they competed by the CRE derived from the ␣-promoter (23,24), suggesting distinct pathways for cAMP control of the two genes. Another B-Zip protein, c-Jun, negatively regulates both promoters, and it binds to the ␣-CRE and to the CG␤ gene promoter between Ϫ245 and Ϫ220 bp (16). However, based upon antibody-mediated supershift studies, c-Jun does not appear to represent the major protein that binds to this region of the CG␤ gene (16). Distinct sequences within the cAMP-responsive region of the CG␤ promoter may share common binding proteins, since they cross-compete for protein interactions (24). In addition, the TSE region from the ␣-promoter also competes for these proteins, suggesting that the same transcription factors might be involved in the coordinate regulation of the two promoters (24).
Transcription factor AP-2 has been shown to mediate cAMP responses in other genes including metallothionein IIA (26), acetyl-CoA carboxylase (27), insulin-like growth factor-binding protein-5 (28), and RII protein kinase ␤ (29) among others (16,30). AP-2 has also been implicated in developmentally regulated gene expression in a variety of cell types including NT-2 and P19 teratocarcinoma cells (31,32), primary neural cells (33), keratinocytes (34 -36), and adipocytes (27). AP-2 is a 52-kDa protein that binds to DNA as a homodimer (37) and can associate with c-Myc, inhibiting transactivation by c-Myc (38). Several AP-2 variants have been identified (37,39), some of which act as inhibitors of AP-2 gene activation through an undetermined mechanism (37). In this report, we examine a potential role of AP-2 in the regulation of the ␣ and CG␤ genes. We find that AP-2 binds to regulatory elements in both promoters and regulates the expression of these genes. AP-2 is also induced during placental cell differentiation.

MATERIALS AND METHODS
Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared by the Shapiro method (40) modified by the addition of protease inhibitors (1 g/ml aprotinin, 1 g/ml pepstatin, 1 g/ml leupeptin, and 1 g/ml p-aminobenzamide (Sigma) to the final dialysis buffer. Nuclear extracts (5-10 g) were added to a 20-l reaction containing: 2 l of 10 ϫ buffer (200 mM HEPES, pH 7.9, 400 mM KCl, 10 mM MgCl 2 , and 1% Nonidet P-40), 500 ng of dI-dC, and 1 g of AP-2 or Sp-1 antibodies as indicated. Reactions were preincubated on ice for 30 min before the addition of 30 fmol of radiolabeled probe with or without unlabeled competitor DNA. Reactions were incubated at room temperature for 30 min before electrophoresis (180 V, 3 h) through nondenaturing 5% polyacrylamide gels in 0.5 ϫ TBE (45 mM Tris borate, 1 mM EDTA).
Oligonucleotides for electrophoretic mobility shift assays are listed in Table I. Hybridized oligonucleotides were labeled by Klenow end filling. Klenow reactions included 5 pmol of annealed oligonucleotides, 1 l of 10 ϫ Klenow buffer (Promega, Madison, WI), 4 l of 25 mM deoxynucleoside triphosphates (without dATP), 3 l of [␣-32 P]dATP (3000 Ci/ml, DuPont NEN), and 1 l of Klenow fragment (Promega). Reactions were incubated at 37°C for 30 min before the addition of 5 l of 25 mM dNTPs and incubation for an additional 10 min. After stopping the reactions, the solution was passed through Centricep columns (Adelphia, NJ) to remove unincorporated nucleotides.
Reporter Genes and Construction-The ␣ and CG␤ luciferase constructs in the pA 3 LUC plasmid have been described previously (23,25). Site-directed mutagenesis was performed using sequences that correspond to the mutant oligonucleotides used in gel mobility shift assays (Table I). Polymerase chain reactions were used to incorporate the mutations within the ␣ and CG␤ promoters (25). All site-directed mutations were sequenced to verify the mutation as well as the correct native promoter sequence. Expression vectors containing wild type AP-2 and AP-2 mutants were driven by the Rous sarcoma virus promoter (41).
Western Blots-Nuclear extracts are described above (40). Whole cell placental extracts were kindly provided by Dr. T. Woodruff (Northwestern University). Primary antibodies included 1:1000 AP-2 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and 1:1000 Sp-1 (Santa Cruz Biotechnology) and were added to nitrocellulose membranes for 60 min in solution containing albumin (5 mg/ml). After washing three times in 0.1% Tween phosphate-buffered saline, membranes were incubated with 1:10,000 secondary anti-rabbit antibody (Santa Cruz Biotechnology) in 3% milk phosphate-buffered saline. After washing, membranes were subjected to enzyme-linked chemiluminescence as described by the manufacturer (Amersham Corp.).
RT-PCR Assays for mRNA Expression in Placental Cells-Cytotrophoblast cells were isolated from human term placenta and cultured under conditions that allow fusion into syncytiotrophoblast cells (46). RNA was isolated from placental cells every 2 days over a 12-day period of culture (46). Total RNA (1 g) was reverse transcribed (37°C, 2 h) by the addition of 15 units of reverse transcriptase (Promega) in the presence of 10 pmol of random hexamer primers, 25 mM deoxynucleoside triphosphates (dNTPs) in 1 ϫ MI buffer (67 mM Tris, pH 8.8, 6.7 mM MgCl 2 , 16 mM (NH 4 ) 2 SO 4 , 10 mM ␤-mercaptoethanol) in a total volume of 20 l. PCR reactions included specific primers for the ␣ and CG␤ genes, AP-2, and internal controls, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and ribosomal protein 19 (RPL19). The final FIG. 1. Identification of AP-2 binding sites in the ␣ and CG␤ genes. A, schematic representation of regulatory elements and protein binding sites in the ␣ and CG␤ genes. GATA, GATA sequence binding site; ACT, activator; JRE, junctional regulatory element. B, gel mobility shift assays for AP-2 binding. Radiolabeled DNA sequences are indicated below the gel and include ␣URE (Ϫ186 to Ϫ156 bp), hCG␤ FP2 (Ϫ311 to Ϫ279 bp), hCG␤ FP4A (Ϫ240 to Ϫ221 bp), hCG␤ FP4B (Ϫ221 to Ϫ200 bp), hMTII A AP-2 site (Ϫ186 to Ϫ165 bp), and the Sp-1 consensus site. Competition with a 100-fold excess of the AP-2 sequence from the hMTIIA promoter is indicated at the top of the gel.
PCR reaction (100 l) included a 1-l aliquot of the RT reaction product, 50 pmol of sense and antisense primers for AP-2, ␣, and CG␤, along with one of the controls, GAPDH or hRPL19, in 1 ϫ MI buffer (25 M dNTPs, 10 l of Me 2 SO, 0.5 l of Taq DNA polymerase (Promega), and 0.1 l of [ 32 P]dATP (DuPont NEN). Cycle conditions were 96°C for 30 s, 94°C for 1 min, 58°C for 1 min, and 72°C for 1 min. An aliquot (20 l) of each reaction was subjected to polyacrylamide (6%) gel electrophoresis, and the specific products were quantitated using a Fujix 2000 phosphoimager (Fuji Medical Systems, Stamford, CT). Data were analyzed by one-way analysis of variance using Dunnett's test.

AP-2 Binds to Regulatory DNA Elements in the ␣ and CG␤
Promoters-The ␣ and CG␤ regulatory elements that have been shown to bind proteins in JEG-3 nuclear extracts are depicted in Fig. 1A (12,13,23,24). Previous studies have shown cross-competition by a subset of these elements, including the ␣URE/TSE and several of the CG␤ footprinted elements, suggesting that these sequences may share common transcription factors (24,47). The size of the URE/TSE binding protein (24) and the GC-rich nature of the DNA sequences shared by these elements raised the possibility that AP-2 might interact at these sites.
Electrophoretic mobility shift assays were performed using JEG-3 nuclear extracts to assess whether a consensus AP-2 sequence competed for binding to the ␣ and CG␤ elements (Fig.  1B). Excess (100-fold) unlabeled AP-2 oligonucleotide inhibited binding to the ␣URE and CG␤ FP4B. The lower part of the complex that binds to CG␤ FP2 was also reduced by the AP-2 competitor. As a positive control, the AP-2 competitor was shown to compete well for protein binding to the homologous AP-2 sequence derived from the human metallothionein IIA (hMTIIA) promoter (Ϫ185 to Ϫ167) (26,48). However, excess unlabeled AP-2 oligonucleotide did not compete with complexes that bind to the Sp-1 binding site or to CG␤ FP4A, indicating that the competition is specific. These results suggest that AP-2 binds to sequences within both the ␣ and CG␤ promoters.
AP-2 antibody was used in supershift assays to confirm whether AP-2 was present in the protein complexes that bind to the ␣ and the CG␤ promoter elements (Fig. 2A). The AP-2 antibody supershifted the major complex binding to the ␣URE and to CG␤ FP4B as well as proteins that bind to the control AP-2 element, hMTIIA. In contrast, the AP-2 antibody had no effect on protein binding to an Sp-1 element. The Sp-1 antibody had no effect on protein binding to the putative AP-2 sites, whereas it caused a supershift of the Sp-1 complex.
Protein interactions with different domains within CG␤ FP2 are illustrated in Fig. 2B. When the full-length FP2 fragment was used, multiple protein complexes were observed. The addition of either AP-2 or Sp-1 antibody, but not preimmune AP-2 sera, appeared to shift part of the major protein complex, suggesting that both proteins might bind to FP2. The FP2 fragment was divided into overlapping segments FP2A, FP2B, and FP2C. AP-2 competitor DNA inhibited protein binding to the FP2A and FP2C fragments, and these complexes were also supershifted by AP-2 antibody. On the other hand, Sp-1 competitor inhibited protein binding to FP2B, and Sp-1 antibody supershifted this complex. Additional experiments confirmed that these interactions were specific, since the Sp-1 antibody did not alter binding to FP2A and FP2C and the AP-2 antibody had no effect on the FP2B complex (data not shown). These experiments indicate that AP-2 binds to the 5Ј-and 3Ј-ends of the FP2, whereas Sp-1 binds to the central region (Ϫ306 to Ϫ285 bp).
The interaction of AP-2 with the FP2 region was resolved further by performing combinations of oligonucleotide competitions and antibody supershift experiments (Fig. 2C). In the presence of Sp-1 competitor oligonucleotide, several FP2 protein complexes were diminished, and the amount of the putative AP-2 complex was increased. The addition of AP-2 antibody in the presence of Sp-1 competitor supershifted the residual protein complex. These findings support the idea that both Sp-1 and AP-2 bind to FP2 and raise the possibility that these binding sites overlap partially.
AP-2 Is Present in JEG-3 Nuclear Extracts and in Placenta-Western blots were performed to determine whether AP-2 is present in cells that express the ␣ and CG␤ genes (Fig. 3). AP-2 antibody detected a 52-kDa protein in JEG-3 cells and extracts from whole placenta and trophoblasts. As controls, the same band was seen in HeLa cells but not in Hep-G2 cells, which

FIG. 5. AP-2 binding to mutant CG␤ FP4B regulatory elements.
A, AP-2 binding to hCG␤ FP4B was examined in the presence of increasing amounts of the indicated competitor DNA sequences. B, AP-2 binding to the ␣URE is shown on the left with competition by the native or 172M␣ mutant. Competition of various FP2 mutants for binding to a consensus AP-2 site is shown on the right. The sequences of the various mutants are shown in Table I. have been shown previously to be deficient in AP-2 (41). The same blot was reprobed with an Sp-1 antibody and verified that the nuclear extracts from each of the cell lines contained similar levels of Sp-1 protein (Fig. 3B). However, little Sp-1 was detected in the whole cell extracts from placenta or trophoblast cells (Sp-1 was seen in these extracts with longer exposure, data not shown). These results indicate that AP-2 is abundant in the placenta.

Role of the AP-2 Binding Sites in the Function of the ␣ and CG␤ Promoters-Transient expression assays were performed
in Hep-G2 cells to examine the effects of co-transfected AP-2 on the ␣ and CG␤ promoters (Fig. 4) (41). AP-2 stimulated Ϫ3700 CG␤ promoter activity 22-fold (Fig. 4A). Deletions to Ϫ1700 or Ϫ345 bp decreased stimulation to 10-fold, and subsequent deletions to Ϫ248 bp essentially eliminated AP-2 stimulation. The deletion between Ϫ345 and Ϫ248 bp includes FP2 but not FP4. An internal deletion of FP4 (345 d-␤FP4) also eliminated AP-2 stimulation. AP-2 increased Ϫ846 ␣-promoter activity by 6.8-fold. This effect was reduced to 3-4-fold by deletion to Ϫ290 or Ϫ180 bp. Deletion to Ϫ156 bp eliminates the ␣URE and the AP-2 binding site. Deletion to Ϫ132 bp eliminates one of the two CREs. These deletions both reduced basal expression and also reduced AP-2 stimulation to 1.3-and 2.5-fold, respectively. A single point mutation (172M␣), shown previously to disrupt binding to the URE (19) (see below), decreased AP-2 stimulation further (0.9-fold).
Because deletion of CG␤ FP2 has been shown previously to eliminate cooperative interactions with more proximal sequences (23,25), the roles of CG␤ FP2 and FP4B were examined further by creating point mutations within the individual domains. Electrophoretic mobility shift assays were used to determine the effects of the mutations on protein binding (Fig.  5A). Relative to the wild-type FP4B sequence, each of the CG␤ FP4B mutants (Table I) competed poorly for AP-2 binding. FP4B-m1 and FP4B-m3 showed little or no competition, whereas FP4B-m2 competed partially for AP-2 binding. The effects of mutations in CG␤FP2 or the ␣URE are shown in Fig.  5B. Consistent with the AP-2 supershift studies in Fig. 2B, fragments FP2A and FP2C, but not FP2B competed for AP-2 binding. The FP2-m1 mutation, which alters the binding site in fragment FP2A did not compete for AP-2 binding. The 172 M in the ␣URE eliminated AP-2 binding to this fragment of the ␣-promoter (Fig. 5B).
The effects of the CG␤ FP4 and FP2 mutations were examined in the context of the Ϫ345 bp construct (Fig. 6A). In Hep-G2 cells, each of the point mutations in FP4 reduced AP-2 induction by 80% or more, confirming that AP-2 exerts functional effects through the FP4 element (Fig. 6B). In contrast to the FP4B mutations, the FP2 mutations reduced basal activity, but AP-2 induction was reduced by only 30%.
JEG-3 cells have been used extensively for studies of ␣ and CG␤ gene expression (9). Because JEG-3 cells express abundant amounts of AP-2 (Fig. 3), they were used to assess the effects of the AP-2 mutations in the presence of the endogenous protein (Fig. 7). Each of the FP4 mutations greatly reduced basal activity, consistent with a role for endogenous AP-2 in the regulation of this site (Fig. 7A). The FP2 mutations caused an even greater decrease in basal activity, suggesting that the AP-2 and/or the Sp-1 sites in this region are also involved in basal expression.
AP-2 has been implicated in cAMP regulation of gene expression (26 -28), and cAMP is known to induce the CG␤ gene. The effect of mutations of the AP-2 binding site on cAMP stimulation of CG␤ promoter activity were assayed after 18 h of treatment (Fig. 7B). The wild-type Ϫ345 CG␤ promoter was induced 42-fold by cAMP (Fig. 7B). The AP-2 mutations in FP4 reduced cAMP stimulation to a variable extent (50 -75% decrease). The FP2-m1 mutation, which eliminates AP-2 binding to the FP2A sequence, also decreased cAMP stimulation. In contrast, the FP2-m2 and FP2-m3 mutations, which disrupt Sp-1 binding, reduced basal activity but had little effect on cAMP stimula- tion. The AP-2 mutant in the ␣-promoter did not affect cAMP stimulation (data not shown), consistent with the presence of other consensus CREs in this promoter (9).
Domains of AP-2 Required for Induction of the CG␤ Promoter-Several functional domains have been delineated in AP-2, including a DNA binding domain, a dimerization domain, and transactivation domains (41,49). Transient co-transfection studies were conducted in Hep-G2 cells comparing wild type and mutant AP-2 expression vectors using the Ϫ345 hCG␤ promoter as the reporter gene (Fig. 8). Deletion of the aminoterminal 50 amino acids of AP-2 (⌬N51) did not affect transactivation, but further deletion of the amino-terminal 165 amino acids (⌬N165) decreased CG␤ promoter activation by 70%. The region of difference between the two deleted stretches includes the proline-rich and acidic activation domains. A further deletion that also removes the DNA binding domain (⌬N278) was inactive. Carboxyl-terminal deletion from 437 to 413 (⌬C413) had no effect, whereas deletion into the dimerization domain (⌬C390) completely eliminated AP-2 induction of the CG␤ promoter. These results suggest that AP-2 induction of the CG␤ promoter requires dimerization and DNA binding together with the transactivation domains, similar to studies performed with the hMTIIA AP-2 site (41).
AP-2 Gene Expression Increases during in Vitro Differentiation of Trophoblast Cells-Cytotrophoblast cells were isolated from human placenta and induced to undergo differentiation into syncytiotrophoblasts (46). RNA was extracted over the course of 12 days of differentiation, and the levels of AP-2 and of ␣ and CG␤ mRNA were analyzed by RT-PCR (Fig. 9). AP-2 mRNA levels increased 3-fold after 2 days of differentiation and continued to increase gradually during the 12-day period (8fold increase). The ␣ and CG␤ mRNA levels also increased during the first 2 days and showed more marked stimulation between days 2 and 4 before reaching a plateau (maximal -fold increase for ␣ was 16-fold and for CG␤ was 45-fold). GAPDH and hRPL19 were used to normalize expression of the other mRNAs, and they did not change substantially during the differentiation process (data not shown). The finding that AP-2 mRNA levels increase in conjunction with the stimulation of the ␣ and CG␤ genes is consistent with a role for AP-2 in the regulation of these genes in the placenta. DISCUSSION Chorionic gonadotropin gene expression in the placenta is a relatively recent evolutionary event, since its expression occurs almost exclusively in higher primates (9). In the case of the ␣-gene, modifications in the CRE sequence and in adjacent upstream regulatory elements appear to account for the ability of the ␣-gene to be expressed in the placenta as well as in the pituitary gland (17,18). The CG␤ genes appear to have duplicated and diverged from an ancestral LH␤ gene (2). Although LH␤ expression is restricted to the pituitary gland, the CG␤ genes are expressed preferentially in the placenta, presumably reflecting the acquisition of new regulatory DNA sequences that direct placenta-specific expression (50).
The DNA regulatory elements that control CG␤ gene expression have been challenging to define. Mutational studies have suggested that several distinct elements may interact in an interdependent manner, a phenomenon that has made it difficult to clearly delineate discrete functional domains (23). Protein binding studies have therefore proven quite helpful for defining potential regulatory elements. DNase I footprinting analyses delineated several discrete binding sites, particularly between Ϫ311 and Ϫ200 bp (23,24). More recently, it was found that several of these sites competed with one another, raising the possibility that a common protein was binding to multiple sites (24). Moreover, evidence that a key regulatory element (URE/TSE) for placental expression of the ␣-promoter also competed for binding to the CG␤ elements suggested that this protein might be involved in the coordinate expression of the two genes (24). These findings have underlined the importance of identifying the factors that bind to the CG␤ regulatory elements.
In this report, we provide several lines of evidence that a transcription factor that is immunoreactive with AP-2 antibodies binds to the CG␤ regulatory elements. Consensus AP-2 elements compete for binding to FP4B and for two of the complexes that bind to FP2. The identity of this protein as AP-2 is strengthened by the fact that an AP-2 antibody supershifts the complexes that bind to these sites. Similar data were found for the ␣URE site that competes for protein interactions with the CG␤ elements, supporting the notion that this protein is shared in common by these sequences. The molecular mass of AP-2 is similar (52 kDa) to the size of the protein previously purified by affinity chromatography using the ␣URE sequence (24). Last, in AP-2-deficient Hep-G2 cells, co-expression of AP-2 stimulated expression of the ␣-promoter and, to a greater degree, the CG␤ promoter. Taken together, these data suggest AP-2 may be a regulator of ␣ and CG␤ gene expression in the placenta.
AP-2 appears to interact with several CG␤ elements, including FP2 and FP4B. Additional AP-2 sites were also identified several kilobase pairs upstream in the CG␤ promoter. 2 The proximal part of the CG␤ promoter is very G-C-rich and has yet to be tested for AP-2 binding. FP2 is also bound by Sp-1, and it appears that AP-2 and Sp-1 may bind to partially overlapping elements, since competition for Sp-1 facilitated the binding of AP-2 (Fig. 2C). Previous studies revealed that c-Jun binds close to FP4A, which is adjacent to one of the AP-2 sites (FP4B) (25). Thus, Jun, AP-2, and Sp-1 have now been shown to interact with the CG␤ promoter. Given evidence for combinatorial interactions among these sequences, an important question for future studies is to understand the mechanisms by which these regulatory elements interact. One possibility is that they may share transcriptional co-activators.
AP-2 has been suggested to mediate cAMP responsiveness in a variety of promoters (26 -30). cAMP regulation of hCG␤ was partially reduced by AP-2 mutants in FP4B (50 -75% decrease). In FP2, the mutation that eliminates AP-2 binding (FP2-m1) decreased cAMP stimulation (65% decrease), whereas the mutations within the Sp-1 binding site (FP2-m2, FP2-m3) had less effect on cAMP stimulation. These findings support the idea that AP-2 plays a role in cAMP stimulation of the CG␤ promoter. However, the fact that cAMP responsiveness is not eliminated by these mutations indicates that other sequences are probably involved in cAMP stimulation of the CG␤ promoter (50).
AP-2 has been implicated in developmental regulation in several cell types, and it is intriguing to consider the possibility that AP-2 may participate in a developmental cascade during trophoblast differentiation. For example, the expression of keratin genes in the developing epidermis correlates with the presence of cells that express high levels of AP-2 (35). In NT-2 teratocarcinoma cells, retinoic acid induces AP-2 expression as these cells undergo differentiation (31). We found that AP-2 mRNA levels increased early in the process of trophoblast differentiation in vitro, and AP-2 protein levels are high in normal placenta. Because the CG genes are expressed very early during embryogenesis and implantation, it is of interest to determine whether AP-2 is already expressed at this time of development. Recently, the AP-2 gene was disrupted by targeted mutagenesis in mice, revealing that it is necessary for neural tube closure and craniofacial development (51,52). Potential effects on placental development were not evaluated, but one can speculate that there were no severe abnormalities as pregnancies proceeded to completion. It should be noted, however, that two additional AP-2-related genes (AP-2␤, AP-2␥) have recently been identified (37,39), and it is possible that these genes may exert redundant functions in the placenta and other tissues. In addition, because the CG␤ genes are not expressed in mice, it is not possible to evaluate potential effects of the null mutants on CG␤ gene expression in the murine model.
In conclusion, we have shown that AP-2 expression increases as trophoblasts differentiate in an in vitro model. AP-2 appears to be a major regulator of the CG␤ gene, and to a lesser degree, the ␣-gene. As such, it may represent one of several factors that coordinate the expression of these genes. Future studies will help to define other factors that function in conjunction with AP-2 to regulate the combinatorial elements in these genes.