A role for the homeobox protein Distal-less 3 in the activation of the glycoprotein hormone alpha subunit gene in choriocarcinoma cells.

Synthesis and secretion of chorionic gonadotropin in trophoblast cells of the placenta is required for establishment of early pregnancy in primates. Chorionic gonadotropin is a heterodimeric glycoprotein hormone consisting of alpha and beta subunits. Regulation of the alpha subunit gene within the placenta requires an array of cis elements within the 5'-flanking region of the promoter. Within this array of elements, the junctional regulatory element (JRE) putatively binds a placental-specific transcription factor. The aim of our studies was to determine the identity and role of the transcriptional regulator that binds to the JRE in choriocarcinoma cells (JEG3 cells). Mutations within the JRE resulted in reduction in basal expression of an alpha subunit reporter gene, suggesting that the JRE binding factor was necessary for full basal activity. Using electrophoretic mobility shift assays, we determined that the JRE was capable of serving as a homeobox factor-binding site. The homeobox factor, Distal-less 3 (Dlx 3) was found to be expressed in JEG3 cells and in the trophoblast layer of human chorionic villus but not in a gonadotrope cell line that also expresses the alpha subunit gene. Electrophoretic mobility shift assays revealed that recombinant Dlx 3 could bind specifically to the JRE and endogenous Dlx 3 was present in JRE/JEG3 nuclear protein complexes. Overexpression of Dlx 3 resulted in activation of an alpha subunit reporter gene. A JRE mutation resulted in attenuated activation of the alpha subunit reporter via an adjacent cis element, suggesting that JRE/Dlx 3 interactions may facilitate regulation of the alpha subunit gene at sites immediately upstream of the JRE. Our studies support the conclusion that Dlx 3 is a placental-specific transcriptional regulator that binds to the JRE and contributes to expression of the alpha subunit gene in cells of trophoblast origin.

The glycoprotein hormones are a family of heterodimeric peptides that consist of an ␣ subunit noncovalently linked to a hormone-specific ␤ subunit (1). Within the glycoprotein hormone family, the ␣ subunit is a common component of all members, and it is the ␤ subunit of these hormones that confers biological specificity to the heterodimer. Glycoprotein hor-mones derived from the anterior pituitary gland include luteinizing hormone, follicle-stimulating hormone, and thyroid-stimulating hormone, produced by cells of the gonadotrope (luteinizing hormone and follicle-stimulating hormone) and thyrotrope lineages (thyroid-stimulating hormone). These peptide hormones play an integral role in the regulation of reproduction and metabolic homeostasis in mammals. In humans and nonhuman primates, an additional member of the glycoprotein hormone family is synthesized and secreted from the placenta. Human CG 1 is produced and secreted by the cells of the trophoblast lineage during the first trimester of pregnancy and is necessary for the maintenance of progesterone secretion from the ovarian corpus luteum to ensure the establishment of an appropriate uterine environment and pregnancy (2)(3)(4)(5). Furthermore, CG may play a paracrine role in the uterine endometrium to facilitate uterine receptivity to implantation in primates (6).
The expression of the ␣ subunit gene is restricted to cells of pituitary and placental origin. The regulatory region of the ␣ subunit gene is unique in that it confers cell-specific expression via a complex array of cis-acting elements and their cognate binding factors (7,8). In the gonadotrope of the anterior pituitary gland, expression of the human ␣ subunit gene requires dual CREs that bind CREB family members, an element that binds a LIM-homeobox factor, and an SF-1 binding site, all of which contribute to cell-specific expression. Two additional elements, referred to as ␣ basal elements 1 and 2, also contribute to pituitary-specific ␣ subunit expression; however, the factors that bind these sites have not yet been identified (9,10). An additional element within the murine ␣ subunit promoter has also been identified as an Ets factor-binding site that contributes to inducible expression of the ␣ subunit by phorbol esters and gonadotropin-releasing hormone (11,12). Together, these elements and the transcription factors that bind them represent a combinatorial code required for expression of the ␣ subunit in the anterior pituitary gland. An overlapping but separate combinatorial code exists for expression of the ␣ subunit in cells of placental trophoblast origin. The dual CREs, in conjunction with an upstream regulatory element, a junctional regulatory element, and a unique CCAAT box provide for trophoblast-specific expression (13)(14)(15)(16)(17)(18)(19). Portions of the URE and JRE putatively bind cell-specific factors; however, the identity of these factors is not currently known (19). We have recently demonstrated that the dual CREs are a focal point for regulation via multiple signal transduction cascades including the protein kinase A and multiple mitogen-activated protein kinase pathways in choriocarcinoma cells (20). Thus, with the exception of the dual CREs, we know relatively little about the cell-specific factors that bind to and transactivate the ␣ subunit gene in trophoblast cells. The aim of our studies was to determine the identity and role of the transcriptional regulator that binds to the JRE.
Regulation of morphogenic changes and differentiation of cell lineages within many endocrine organs occurs in association with complex expression patterns of cell-specific transcriptional regulators including homeodomain-containing transcription factors. Two notable examples are the organogenesis of the anterior pituitary gland and the pancreas. Of the many transcription factors described in the ontogeny of pituitary gland development (21), the LIM homeobox factor Lhx3 (20,22) and the POU-homeobox factor Pit 1 (23) play key roles in early determination of pituitary cell lineages that give rise to multiple endocrine cell types. Furthermore, the homeobox factor Pdx 1 is necessary for differentiation of multiple cell lineages within the pancreas (24,25). Targeted gene disruption of Lhx3 and Pdx 1 lead to developmental failure of the anterior pituitary gland and pancreas, respectively, underscoring the importance of these factors to organogenesis. Naturally occurring mutations in Pit 1 have generated important mouse models for pituitary dwarfism (26 -28). Interestingly, these homeobox factors also contribute to cell-specific regulation of gene products that define the differentiated character of cell lineages within the developing endocrine organ. For example, Lhx3 and Lhx2 have been shown to regulate the ␣ subunit gene in the anterior pituitary (29,30). Pit 1 is a key regulator for the growth hormone, prolactin, and thyroid-stimulating hormone ␤ genes (23,(31)(32)(33). Pdx 1 functions in the transactivation of the insulin gene (34). The studies presented in this investigation of trophoblast-specific gene regulation are completely consistent with this model. It has been previously reported that the Distal-less class homeobox factor, Dlx 3, is necessary for normal placental development in a mouse knockout model (35). Mouse knockout models are not sufficient to determine a role for Dlx factors in the cell-specific regulation of CG subunit genes since the mouse does not express a CG. Our studies demonstrate that Dlx 3 is localized to the trophoblast layer of human chorionic villus, and Dlx 3 binds to and transactivates the human ␣ subunit gene via interaction at the JRE. JRE/Dlx 3 interactions contribute functionally to the combinatorial code necessary for placental-specific regulation of the ␣ subunit gene.

MATERIALS AND METHODS
Plasmids-All plasmids were prepared by two cycles through cesium chloride using standard methodologies. The glycoprotein hormone ␣ subunit reporter has been described previously (20). Mutagenesis of this reporter was carried out using oligonucleotide-directed mutagenesis as described (20). The JRE mutation consisted of a four-nucleotide substitution (i.e. wild type 5Ј-TAATTACA-3Ј; the JRE mutation 5Ј-TGGCCACA-3Ј). This mutation was confirmed by nucleotide sequence analysis. The Dlx 3 coding sequence was amplified using High Fidelity polymerase chain reaction (Roche Molecular Biochemicals). The polymerase chain reaction product was subcloned into the TA cloning vector (Invitrogen, Carlsbad, CA), and orientation and fidelity were corroborated by DNA sequencing. A FLAG epitope (DYKDDDDK) was inserted at the 3Ј end of the coding sequence immediately upstream of the TGA stop codon using the site direct mutagenesis kit (Stratagene, La Jolla, CA). The Dlx 3 sequence was then excised by specific restriction endonuclease digestion (EcoRI-NotI) and cloned into the pBK-CMV vector (Stratagene, La Jolla, CA) to create the pCMV/Dlx 3 FLAG construct. DNA sequence analysis was performed to confirm the Dlx 3 sequence.
Cell Culture and Transfection Studies-The human choriocarcinoma cell line, JEG3, was cultured in monolayers using Dulbecco's modified Eagle's medium supplemented with fetal bovine serum (10%). All culture media were purchased from Sigma. Fetal bovine serum was purchased from Life Technologies, Inc. Before all studies, cell cultures were split to fresh medium, and sub-confluent cultures were used. All transient transfection studies were conducted using electroporation as described previously (20). In transfection studies requiring agonist administration, cells were treated with forskolin (1 M) and/or EGF (50 ng/ml) for 6 h before collection (18 h after electroporation). Forskolin was purchased from Sigma and resuspended at a stock concentration of 1 mM in dimethyl sulfoxide. Epidermal growth factor was purchased from Life Technologies, Inc. and resuspended at a stock concentration of 1 mg/ml in Dulbecco's phosphate-buffered saline. After cell collection, lysates were prepared by three freeze-thaw cycles and clarified by centrifugation, and luciferase activity was determined as described (11,20,36). Luciferin was purchased from Promega (Madison, WI). Luciferase activity was standardized by total protein amount, and all transfection studies were conducted in triplicate on at least three separate occasions with similar results. Data shown are reported as a mean (n ϭ 3) Ϯ S.E of the mean.
Preparation of Recombinant Lhx2 Homeodomain, Dlx 3, and JEG3 Cell Nuclear Extracts-The homeodomain for Lhx2 (coding sequence from lysine 147 through the carboxyl termini of the protein) was prepared as a polyhistidine fusion protein in bacteria as described (29). Lhx2 homeodomain was partially purified from bacterial lysates using a nickel chelate-agarose. Full-length Dlx 3 was prepared using a rabbit reticulocyte lysate system. A transcription and translation reticulocyte lysate kit was purchased from Promega and was used as per instructions.
Subconfluent JEG3 cells were used for the preparation of nuclear extracts. Plates were placed on an ice bed, and cells were washed with ice-cold 10 mM Hepes (pH 7.4) and 150 mM NaCl (Hepes-buffered saline). Cells were collected by scraping in ice-cold Hepes-buffered saline and pelleted by centrifugation (2000 rpm for 15 min). Cells were resuspended in a hypotonic buffer and lysed by douncing, and nuclei were isolated using the sucrose cushion method as described previously (20). Nuclei were resuspended in a buffer containing 10 mM Tris (pH 7.5), 50 mM NaCl, 5% glycerol, 1 mM EDTA, 1 mM dithiothreitol, 5 mM benzamadine, and 0.2 mM phenylmethylsulfonyl fluoride. This buffer was referred to as binding buffer. Nuclear proteins were extracted by adding additional NaCl (in binding buffer) to a final concentration of 450 mM and incubating the extract for 30 min at 4°C with constant rocking. Nuclear debris was removed by centrifugation, and protein concentration was determined by Bradford assay. Nuclear extracts were separated into aliquots and stored at Ϫ80°C until later use.
Electrophoretic Mobility Shift Assay-Electrophoretic mobility shift assays were conducted essentially as described (11,29). Briefly, recombinant Lhx2 homeodomain, recombinant Dlx 3, or binding activity from nuclear extracts was mixed with binding buffer, 1 g of poly(dI-dC), and in some reactions, antiserum or competition oligonucleotides in a total reaction volume of 25 l. Reactions were maintained at room temperature for 20 to 30 min. JRE oligonucleotides were radiolabeled using polynucleotide kinase and [␥-32 P]ATP. Labeled JRE probe (ϳ10,000 -20,000 cpm) was then added, and the incubation was continued for 20 -30 min at room temperature. The binding reactions were resolved on native polyacrylamide gels. The gels were dried, and DNA-protein complexes were visualized by autoradiography. All DNA binding studies were conducted at least twice with similar results.
Western Blot Analysis-JEG3 or ␣T3-1 cell nuclear extracts were suspended in an equal volume of 2ϫ SDS loading buffer (100 mM Tris (pH 6.8), 4% sodium dodecyl sulfate, 20% glycerol, and 200 mM dithiothreitol). Protein samples were boiled for 5 min and chilled briefly on ice before loading on gels. Proteins were resolved by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidine difluoride membranes by electroblotting. Membranes were blocked with nonfat dried milk (5%) in Tris-buffered saline (10 mM Tris (pH 7.5), 150 mM sodium chloride) containing 0.1% Tween 20 (TBST). Lhx2 antiserum was obtained from rabbits immunized with a purified glutathione S-transferase fusion of the homeodomain region of Lhx2 as previously reported (37). The Lhx2 antibody was used at 1:15,000 in TBST, 5% nonfat dried milk. Rabbit anti-Dlx3 polyclonal antiserum was generated by Berkeley Antibody Company (Berkley, CA), against a 16-mer synthetic peptide containing amino acids 242 through 256 of the murine Dlx3 protein.
Sequence comparison with other Dlx family members of several species showed homology in this area only between Dlx 3 orthologs. There is a very high amino acid conservation between the mouse Dlx 3 and human Dlx 3. There are only eight amino acid substitutions throughout the entire protein when comparing murine and human Dlx 3. IgG anti-Dlx3 antibodies were obtained by running the polyclonal antiserum through a protein A column, washing, and eluting with a change in pH and salt concentration as directed by Pierce (ImmunoPure IgG Purification Kit). The Dlx 3 antibody was used at 1:500 in TBST, 5% nonfat dried milk. Proteins were visualized by chemiluminescence using reagents purchased from PerkinElmer Life Sciences.
Immunocytochemical Analysis-JEG3 cells were plated on poly-Llysine-coated glass slides and cultured overnight. The cells were fixed in ice-cold methanol and allowed to dry for ϳ20 min. Human chorionic villi from placenta obtained at 8 weeks of gestation were fixed in 3% paraformaldehyde, infiltrated with 5-15% sucrose, placed in embedding medium, and frozen in liquid nitrogen. Sections (5-7 m) were cut using a cryostat and transferred to charged glass slides as described (38). The slides were stored at Ϫ80°C until use. For immunocytochemistry, slides and sections were allowed to dry for ϳ20 min, then dehydrated using 70% ethanol for 10 min at room temperature. Slides and sections were blocked for endogenous peroxidase in a solution of 1.5% hydrogen peroxide in absolute methanol for 10 min at room temperature. Antigen retrieval was not necessary for slides of fixed JEG3 cells. Antigen retrieval for chorionic villus sections included boiling the sections in 0.01 M sodium citrate (pH 6.0) for ϳ5 min. The sections were cooled for 20 min and then washed 3 times (5 min each) with deionized water. The sections were then treated with a trypsin solution (0.1% trypsin, 0.1% CaCl 2 , 20 mM Tris (pH 7.8)) for 5 min at room temperature followed by washing in deionized water as described above. The sections were equilibrated in 0.01 M PBS (pH 7.5) for 5 min.
The slides and sections were then incubated with serum-blocking solution (Histostain SP Kit, Zymed Laboratories Inc., South San Francisco, CA) for 10 min at room temperature to eliminate nonspecific background staining. Rabbit-derived anti-Dlx3 in a dilution of 1:50 in PBS was used as a primary antibody. In chorionic villus sections, murine-derived pan-cytokeratin antibody (Zymed Laboratories Inc.) was used as a marker for cells of the trophoblast lineage. Normal rabbit serum (preimmune) was used as negative control. Incubation with primary antisera was carried out for 4 h at room temperature followed by washing with 0.01 M PBS (pH 7.5) three times (2 min each). The samples were next incubated with biotinylated secondary antibodies for 20 min at room temperature and then washed with PBS as described above. After exposure to secondary antibody, the samples were incu-bated with a strepavidin peroxidase conjugate for 10 min at room temperature and washed with PBS as described above. A chromogen solution was applied to the samples. The reaction was observed under a microscope and blocked with deionized water after 8 min.

FIG. 2. The JRE can serve as a homeobox-binding site in vitro.
Panel A depicts the similarity between the nucleotide sequences of the JRE (JRE wt ), the PGBE, and a mutant JRE containing a four-nucleotide substitution mutation (JRE mut ). EMSA was used to determine whether the JRE could bind a recombinant homeodomain from Lhx2 (Lhx2 homeobox ; panel B). Lhx2 homeobox was prepared in bacteria and partially purified. Shown are binding reactions containing radiolabeled JRE (JRE Probe) and Lhx2 homeobox in the absence or presence of a 200-fold molar excess of unlabeled oligonucleotides for wild type JRE (JRE WT ), the PGBE, or the JRE mutant (JRE mut ). In some binding reactions, antiserum directed against Lhx2 (Lhx2 ab) was used to determine the specificity of JRE binding compared with NRS. Lhx2 antibody induced a supershifted complex. Endogenous JEG3 cell nuclear extracts were used with the labeled JRE to determine binding activity present in choriocarcinoma cells (panel C). Similar competition studies were conducted using the unlabeled oligonucleotides described above. The JRE binding reactions resulted in the formation of two discrete complexes, complex 1 and 2, identified by the arrows.

Mutations within the Junctional Regulatory Element Reduce
Basal Expression of an ␣ Subunit Reporter Gene-The regulatory regions of the human ␣ subunit gene promoter necessary for expression in trophoblast cell models such as choriocarcinoma cells are depicted in Fig. 1A. The dual CREs play a central role in basal and cAMP-induced regulation of the ␣ subunit promoter. However, the contribution of the dual CREs to basal expression of the ␣ subunit also depends upon interaction among several other cis regulatory elements including the URE and the JRE (10,19). Portions of the URE and the JRE putatively bind tissue-specific factors thought to be necessary for placental-specific gene activation (10,19,39). A four-nucleotide mutation was placed within the JRE and used in the context of an ␣ subunit reporter in transient transfection studies in JEG3 choriocarcinoma cells. These studies revealed that the JRE was required for full basal expression of the ␣ subunit reporter gene despite the presence of wild type dual CREs (Fig. 1B).
The Junctional Regulatory Element Can Serve as a Homeobox Factor-binding Site-Examination of the nucleotide sequence of the JRE revealed marked similarity to the downstream half-site of the pituitary glycoprotein hormone basal element ( Fig. 2A). The PGBE is an imperfect palindromic sequence required for basal and GnRH-induced activation of the human and murine ␣ subunit genes in the anterior pituitary gland (10,12,29). The PGBE is a binding site for a LIM class of homeobox factor, where Lhx3 and Lhx2 have been shown to bind and induce transcriptional activation (29,30). Based upon this similarity, we speculated that the JRE could serve as a homeobox-binding site. To test this hypothesis, EMSA was performed using a radiolabeled JRE oligonucleotide and the Lhx2 homeodomain (recombinant DNA binding domain) prepared and partially purified from a bacterial expression system ( Fig. 2B; Ref. 29). Nonradioactive competitor oligonucleotides were added to some reactions at a 200-fold molar excess. Competitors included wild-type JRE, a JRE probe containing a four-nucleotide substitution mutation that disrupts basal expression of the ␣ subunit reporter (Fig. 1B) or the PGBE. Competition studies revealed that Lhx2 homeodomain binding to the JRE was specific since the wild type JRE and PGBE oligonucleotides successfully competed for binding, whereas the mutant JRE was ineffective. The JRE-Lhx2 homeodomain complex was confirmed using an Lhx2-specific antibody (37) that resulted in the formation of a "supershift" (Fig. 2B). These studies provide direct evidence that the JRE can serve as a homeobox-binding site in vitro.
We characterized the JRE binding activity in JEG3 cells by EMSA. Nuclear extracts were prepared from JEG3 cells and used in DNA binding reactions containing the radiolabeled JRE binding site. Again, competition studies were used to determine the specificity of binding as described above. EMSA revealed that JRE binding with JEG3 cell nuclear extracts resulted in the formation of two distinct DNA-protein complexes (designated complex 1 and 2). Competition with wild type JRE or the PGBE oligonucleotides abolished formation of both complexes. In contrast, the mutant JRE failed to disrupt JRE binding to JEG3 cell nuclear extracts, suggesting that the complexes formed were specific to the central core of the JRE binding site (Fig. 2C). Dlx3 Activates the CG ␣ Subunit Gene Dlx3 Is Present in Human Choriocarcinoma Cells-A number of homeobox transcription factors have been localized to the developing murine and human placenta and are thought to be important in placental cell differentiation and function (40,41). Recently, targeted deletion of a member of the Distal-less class of homeobox factors, Dlx 3, resulted in embryonic lethality characterized by a failure in the development of the murine placenta (35). The reported consensus-binding site for Dlx 3 is identical to the JRE (42). Fig. 3A depicts the known structural domains of Dlx 3 including the homeodomain that likely serves as the DNA binding domain and two putative transcriptional activation domains (42). Dlx 3 is expressed in JEG3 cells as measured by Western blot analysis of nuclear extracts but not in nuclear extracts from ␣T3-1 cells, a clonal gonadotrope cell line that expresses the ␣ subunit gene (Fig. 3B). As a positive control for ␣T3-1 nuclear extracts, Lhx2 immunoreactivity was determined and was readily apparent in Western blots. Dlx 3 has an apparent molecular mass of 38 -39 kDa, which is slightly larger than predicted by the amino acid sequence. Additional immunocytochemical studies revealed that Dlx 3 expression in JEG3 cells was compartmentalized to the nucleus of this trophoblast cell model (Fig. 3C).

Dlx 3 Expression Is Restricted to the Nuclear Compartment of Trophoblast in 8-Week Human Chorionic Villus-We sought to localize expression of Dlx 3 in human chorionic villus obtained
during the first trimester of gestation (Fig. 4), a time when ␣ subunit expression and CG secretion are relatively high. Eightweek human chorionic villus samples were obtained, fixed, and sectioned. The chorionic villus was visualized by hematoxylin staining, demonstrating a cytokeratin-positive trophoblast layer surrounding the central villus core. Dlx 3 expression was restricted primarily to the trophoblast of the villus and was clearly compartmentalized to the nucleus (Fig. 4). Villi stained with preimmune normal rabbit serum were essentially devoid of signal. Consistent with the murine model (35), these immunocytochemical studies provide evidence that Dlx 3 is present in human trophoblasts at a time during early gestation when CG is synthesized and secreted.
Dlx 3 Binds to the Junctional Regulatory Element of the ␣ Subunit Promoter-Recombinant full-length Dlx 3 was synthesized in reticulocyte lysates and subjected to EMSA to determine whether Dlx 3 could bind to the JRE of the ␣ subunit promoter. Dlx 3 formed a single DNA-protein complex over a range of doses of recombinant Dlx 3 (Fig. 5A). Additionally, EMSA performed with the JRE probe and control reticulocyte lysates expressing luciferase did not produce any DNA-protein complexes (data not shown), suggesting that JRE binding is specific to recombinant Dlx 3 expression in this system. Competition studies in EMSA using recombinant Dlx 3 (Fig. 5B) revealed a similar pattern of DNA-protein interactions compared with EMSA conducted with JEG3 cell nuclear extracts  (Fig. 2). Unlabeled, wild type JRE and PGBE effectively competed for Dlx 3 binding to the radiolabeled JRE. The mutant JRE oligonucleotide did not compete for JRE binding. These studies suggested that Dlx 3 binding to the JRE was specific. Direct comparison of JRE binding to recombinant Dlx 3 and nuclear extracts from JEG3 cells revealed that the recombinant Dlx 3-JRE complex migrated with the same relative electrophoretic mobility as complex 1 from JRE binding to JEG3 cell nuclear extracts (Fig. 5C).
The Dlx 3 Antibody Recognizes an Epitope within the Endogenous JRE Binding Complex-The appropriateness of the Dlx 3 antibody for use in EMSA was initially investigated using recombinant Dlx 3 and the JRE. Binding reactions included normal rabbit serum (NRS), the Dlx 3 antibody, or an antibody directed against Lhx2 (37). Consistent with previous control studies, interactions between the JRE, recombinant Dlx 3, and NRS resulted in the formation of a single complex (Fig. 6A). Replacement of NRS with the Dlx 3 antibody generated a complex with slower electrophoretic mobility, characterized as a marked supershift in the JRE-Dlx 3 complex. The addition of the Lhx2 antibody gave results similar to the NRS control. These studies provide evidence for the specificity of the Dlx 3 antibody in EMSA. We then examined whether the Dlx 3 antibody would alter JRE-nuclear protein interactions using extracts from JEG3 cells. Again, JRE-nuclear protein interactions resulted in the formation of two complexes (Fig. 6B). The addition of the Dlx 3 antibody specifically disrupted the formation of complex 1. The addition of either Lhx2 antibody or NRS did not reduce complex 1 and, in the case of Lhx2 antibody, slightly enhanced binding. It is unclear from these studies whether a supershift formed in the presence of the Dlx 3 antibody since the supershifted complex that formed with recombinant Dlx 3 had a similar electrophoretic mobility as complex 2 from JEG3 nuclear extracts. There did appear to be a slight increase in activity present in complex 2 in the presence of the Dlx 3 antibody, suggesting that this may have been due to a supershift of complex 1. These experiments provide compelling evidence that Dlx 3 or a highly related epitope is present in the JRE binding complex associated with complex 1.
Overexpression of Dlx3 Is Sufficient to Activate the ␣ Subunit Promoter in JEG3 Cells-To determine whether Dlx 3 could function as a transcriptional regulator in JEG3 cells, overexpression studies were conducted using an expression vector for full-length murine Dlx 3. Overexpression of Dlx 3 has been reported to activate transcription of a multimer of a consensus Dlx 3 binding site cloned upstream of a minimal promoter in a heterologous system (42). JEG3 cells were transiently cotransfected with a wild-type ␣ subunit reporter (wild type) or an ␣ subunit reporter containing a mutation within the JRE (Fig.  7A, JRE Mut) and control plasmid or increasing doses of the Dlx 3 expression vector. The following day, luciferase activity was determined (Fig. 7A). Transcription of the wild-type ␣ subunit reporter increased with increasing doses of the Dlx 3 expression vector. Similar increases were not evident using the ␣ subunit reporter containing the mutation within the JRE. Thus, these studies provide evidence of the potential for Dlx 3 to regulate transcription of the ␣ subunit promoter via the JRE. These studies certainly do not discount the possibility that in addition to endogenous Dlx 3, other uncharacterized factors (contributing to complex 2) may also be involved in the regulation of the ␣ subunit promoter in JEG3 cells.
We have recently demonstrated that expression of the ␣ subunit promoter is potentiated by activation of multiple signal transduction pathways induced by EGF and forskolin (20). The dual CREs of the ␣ subunit promoter are located immediately upstream of the JRE (separated by two nucleotides; see Fig.   1A) and are required for activation of the ␣ subunit by EGF and forskolin (20). Our aim was to determine whether overexpression of Dlx 3 would interfere with transcriptional activation of the ␣ subunit reporter via inducible factors that bind to the immediately adjacent CREs. Overexpression of Dlx 3 increased basal expression of the ␣ subunit reporter but did not markedly alter response to EGF, forskolin, or the combination of these two agonists (Fig. 7B). These studies support the conclusion that Dlx 3 could function as a transcriptional regulator in this system concurrent with inducible transcriptional activation of the ␣ subunit reporter via the immediately adjacent CREs. Thus, the JRE and dual CREs can be functionally occupied at the same time.
A Mutation within the JRE Alters EGF-and Forskolin-stimulated ␣ Subunit Gene Regulation-The previous overexpression studies provide evidence that Dlx 3 has the potential to serve as a transcriptional regulator in transfected JEG3 cells. We sought to determine whether the JRE is required for transcriptional activation of the ␣ subunit by EGF and forskolin. JEG3 cells were transiently transfected with a wild type ␣ subunit reporter or a reporter containing the mutation within the JRE. Transfected cells then received control solution, EGF, forskolin, or the combination of EGF and forskolin (Fig. 7C). With the wild-type ␣ subunit reporter, EGF administration induced a 90% increase in ␣ subunit promoter activity. Forskolin administration resulted in a 3.7-fold induction of the ␣ subunit reporter. The combined actions of EGF and forskolin on the ␣ subunit reporter resulted in a 7.2-fold activation. A mutation within the JRE of the ␣ subunit promoter reduced basal activity as observed previously. Interestingly, the JRE mutation resulted in a reduction in activation by the combined effects of EGF and forskolin. Thus, JRE interactions with Dlx 3 and other potential factors (JRE-binding proteins in complex 2) appear to contribute to activation of the ␣ subunit gene by EGF and forskolin. The effects of Dlx 3 are likely permissive since overexpression of Dlx 3 did not enhance the effects of EGF and forskolin on the ␣ subunit reporter. DISCUSSION Expression of Dlx 3 is required for development of a normal murine placenta (35). Our studies provide novel evidence to support the conclusion that Dlx 3 is likely involved in placental-specific activation of the gene encoding the human ␣ subunit of CG, a hormone critical for maintenance of early pregnancy. Establishment of pregnancy in mammals requires appropriately timed endocrine communication between the mother and conceptus. In human and nonhuman primates, secretion of CG in early pregnancy is critical as a luteotropic signal to maintain progesterone secretion from the ovarian corpus luteum. Mistimed or reduced rate of CG synthesis and secretion have been associated with increased potential for early pregnancy failure and repeated miscarriage (3,5,(43)(44)(45)(46). The promoter of the ␣ subunit gene contains a complex array of cis-acting elements that define a transcription factor "code" required for cell-specific and hormone-inducible gene regulation. Among the elements required for expression of the ␣ subunit in cells of the trophoblast lineage are the URE, dual tandem CREs, and the JRE (10,19). The present studies extend our understanding of the transcription factor code by providing direct evidence that the JRE supports basal expression of the ␣ subunit gene. Furthermore, Dlx 3 is an excellent candidate as a homeobox factor that binds to and contributes to the transactivation of the ␣ subunit promoter by direct interaction with the JRE.
The Distal-less family of homeobox factors currently has six members, Dlx 1-6 (for a comprehensive review, see Ref. 47). A unique quality of this family of homeobox factors is that they Dlx3 Activates the CG ␣ Subunit Gene are linked as contiguous pairs within the genome. For example, Dlx 3 and Dlx 4 colocalize on human chromosome 17q21, a region closely linked with the HOXB cluster (48,49). In mammals, expression of the Dlx 3 gene has been shown to be restricted to the branchial arches, dental tissues, epithelial derivatives, and the placenta (47). The Dlx 3 expression pattern partially overlaps with Dlx 4, which is also found in the placenta (40,50). Dlx 3 promoter activity was induced by calcium coincident with keratinocyte differentiation (51). In addition, overexpression of Dlx 3 to the epidermis induced cessation of proliferation and premature or accelerated differentiation, supporting a role for Dlx 3 during epidermal differentiation (52). Recently, a naturally occurring mutation in Dlx 3 was identified and correlated with a condition known as trichodento-osseus (TDO) syndrome (53). This frameshift mutation resulted in premature termination of Dlx 3, leading to compromised function.
Within the developing mouse placenta, expression of Dlx 3 is restricted to the ectoplacental cone cells, the chorionic plate and the labyrinthine trophoblast layer. Targeted deletion of Dlx 3 resulted in embryonic death between embryonic day 9.5 and 10, due to failure in the appropriate morphogenesis of the placenta (35). Interestingly, Dlx 3 Ϫ/Ϫ mice also have reduced expression of a paired class homeobox factor, Esx 1 (35). In normal mice, Esx 1 has been shown to be expressed in the labyrinthine trophoblasts in a pattern consistent with Dlx 3 expression, suggesting that Dlx 3 expression may be a prerequisite for up-regulation of Esx 1 (41). Our immunocytochemical studies document expression of Dlx 3 compartmentalized to the nucleus in human choriocarcinoma cells and human trophoblasts during the first trimester of pregnancy, providing evidence that in vivo Dlx 3 is expressed at the appropriate time and location to contribute to ␣ subunit gene regulation via the JRE. It is tempting to speculate a developmental role for Dlx 3 in the differentiation of the early human placenta.
A hierachary of regulatory "importance" exists among the array of cis elements required for ␣ subunit gene transcription in placental cells (10,19). The dual CREs are the principal regulatory elements that confer basal and cAMP inducibility to the ␣ subunit gene in placental cells (12,14,18,54,55). Mutations within the CREs reduce basal expression quite dramatically. The JRE was first described nearly a decade ago (14). At that time, the JRE binding activity was believed to be required for placental regulation of the ␣ subunit gene; however, the binding activity was not believed to be specifically expressed in the trophoblast cell lineage. Subsequent Southwestern blot analysis revealed that the JRE binding activity was characterized as a 39 -40-kDa protein and was specific to choriocarcinoma cells and not cells of pituitary lineage that also express the ␣ subunit (19). Our identification of Dlx 3 as a JRE binding factor is completely consistent with this estimate of molecular size and cell-specific expression. The Dlx 3 antibody predominantly recognizes a 38 -39-kDa protein in JEG3 cells. Furthermore, expression of Dlx 3 appears to be restricted to cells of the trophoblast lineage but not ␣T3-1 cells that are derived from the gonadotrope lineage of the pituitary gland.
Mutations within the JRE reduced basal ␣ subunit expression despite the presence of intact dual CREs (19). Thus, despite a principal role for the dual CREs, basal expression via the CREs requires potential contributions from the JRE. Overexpression studies revealed that JRE/Dlx 3 interactions increased basal expression but did not interfere with agonistinduced ␣ subunit activation via CREB family member dimers present on the dual CREs immediately upstream of the JRE. These studies suggest that the JRE and CREs can be functionally occupied at the same time. Interestingly, mutations within the JRE interfered with induction of ␣ subunit transcription by the combined actions of EGF and forskolin. After activation by EGF and forskolin, the CRE binding complex consists of multiple transcriptional regulators including CREB and AP-1, which are required for activation of the ␣ subunit by this combination of agonists (20). In contrast, treatment with forskolin alone results in recruitment of CREB alone to the CRE binding complex. It is reasonable to speculate that the formation of the CRE binding complex associated with the combined actions of EGF and forskolin on ␣ subunit gene expression may require recruitment of additional factors such as coactivators, whose presence may depend at least in part on JRE interactions with factors such as Dlx 3. This study underscores the notion that the JRE binding complex may influence transcriptional mechanisms mediated at adjacent cis elements (i.e. the dual CREs). A similar mechanism has been described in the regulation of the murine ␣ subunit promoter by GnRH in cells of pituitary lineage (11,12). Basal expression of the mouse ␣ subunit gene in the pituitary depends on a Lhx homeobox factor binding to the PGBE. Regulation of mouse ␣ subunit expression by gonadotropin-releasing hormone requires the PGBE and a more distal site, the GnRH-responsive element. The GnRH-responsive element is an Ets factor binding site capable of binding a putatively ubiquitous Ets factor(s) that is inducible via signaling cascades activated by GnRH. Thus, a composite transcriptional unit exists where the PGBE binds to a cell-specific factor (Lhx homeobox factor), and the GnRHresponsive element binds a non-cell-specific factor, whose activity depends on activation via signaling molecules induced by GnRH. In the present studies, the JRE serves as the cellspecific homeobox binding site, whereas the dual CREs bind more general factors (like CREB and AP-1) that are regulated by signaling cascades induced by EGF and forskolin. This conserved strategy for gene regulation may reflect an important mechanism by which homeobox proteins contribute to inducible, cell-specific gene regulation in differentiated cells.
Our studies have only accounted for a single factor, Dlx 3, within the JRE binding complex. However, EMSA reveals that two complexes form at this site, suggesting that proteins in addition to Dlx 3 likely contribute to ␣ subunit gene regulation via the JRE. Thus, we cannot yet rule out the possibility of additional factors involved in the JRE binding complex that may have critical importance to the regulation of the ␣ subunit gene. Several possibilities may account for two complexes. A key observation that led to our studies of Dlx 3 was that the consensus binding site for Dlx 3 ((A/C/G)TAATT(A/G)(C/G)) was identical to the JRE (bold; Ref. 42). The central core of the Dlx 3 binding site is essentially the preferred recognition sequence for another family of homeobox factors, the Msx proteins (for review, see Ref. 47). Msx 1, Msx 2, and Dlx 3 are putatively capable of binding identical nucleotide sequence motifs. Msx 2 is expressed in regions of the placenta consistent with Dlx 3 during mouse development (40). Most importantly, Msx proteins have been shown to be transcriptional repressors that functionally antagonize transcriptional activation via Dlx proteins (56). Based upon this, the possibility exists for potential antagonistic interactions between Msx factors and Dlx 3 in placental cells. In our studies, overexpression of Dlx 3 may have simply altered this functional antagonism to favor transcriptional activation. Alternatively, the possibility exists that additional factors are capable of binding the JRE directly or indirectly by physical interaction with Dlx 3, independent of DNA binding. Our studies provide clear evidence that Dlx 3 or a highly related epitope represents the binding activity in complex 1. JRE binding complex 2 may reflect binding of a second protein(s) directly with the JRE or facilitated by pro- tein-protein interactions with Dlx 3. Since the Dlx3 antibody was developed against a peptide, it is possible that Dlx3 may be present in complex 2 but with the epitope functionally blocked due to protein-protein interactions. Additional studies are necessary to resolve these possibilities.
The present studies support the conclusion that Dlx 3 or a highly related epitope binds to and contributes to trophoblastspecific regulation of the ␣ subunit gene. Consistent with findings for the murine placenta, Dlx 3 is expressed in the cells of the trophoblast lineage within the human placenta during the first third of gestation, a time when expression of CG subunit genes and CG secretion is high. Our studies provide novel evidence that overexpression of Dlx 3 can serve to increase basal transcription of the ␣ subunit gene in JEG3 cells. Furthermore, JRE interactions with factors such as Dlx 3 contribute to ␣ subunit gene regulation after administration of EGF and forskolin to induce potentiated activation. Consistent with developmental determination of the anterior pituitary gland and the pancreas, key homeobox proteins (such as Dlx 3) that direct early developmental decisions in the placenta are also important to differentiated endocrine cell function by contributing to expression of cell-specific target genes (such as the ␣ subunit) that define the differentiated character of that cell type.