Different composite regulatory elements direct expression of the human alpha subunit gene to pituitary and placenta.

To identify elements of the human α subunit gene necessary for cell-specific expression, we generated an array of block mutations spanning approximately 400 base pairs (bp) of promoter proximal region and examined them using transient transfection analysis in pituitary (αT3) and placental (BeWo) cell lines. Comparison of promoter activity in the two cell types revealed both common and unique elements required for transcription in pituitary and placenta. Two strong elements, the cyclic AMP response element (CRE) and the upstream regulatory element (URE), regulate expression of the α subunit gene in BeWo cells. In contrast, promoter activity in αT3 cells requires an array of weaker elements. These include the CREs, the URE, as well as two previously described elements, pituitary glycoprotein hormone basal element (PGBE) and gonadotrope-specific element (GSE), and two new elements we designated as the α basal elements 1 and 2 (αBE1 and αBE2). These new elements reside between −316 and −302 bp (αBE1) and −296 and −285 bp (αBE2) of the human α subunit promoter and bind distinct proteins designated αBP1 and αBP2, respectively. Southwestern blot analysis revealed that αBE1 specifically binds 54- and 56-kDa proteins. Additional studies disclosed several potential interactions between proteins that bind the CRE and proteins that occupy PGBE, αBE1, and αBE2, suggesting that gonadotrope-specific expression occurs through a unique composite regulatory element that includes components of the placenta-specific enhancer.

The glycoprotein hormones luteinizing hormone, folliclestimulating hormone, thyroid-stimulating hormone, and chorionic gonadotropin regulate a variety of key biological functions, including reproduction, pregnancy, and metabolism. These hormones belong to a family of heterodimeric proteins composed of a common ␣ subunit and a specific ␤ subunit (1). The ␣ subunit is expressed in the pituitaries of all mammals as the product of a single-copy gene (2). In this tissue, expression occurs in two distinct cell types: gonadotropes, where it is a subunit of both follicle-stimulating hormone and luteinizing hormone, and thyrotropes, where it is a subunit of thyroid-stimulating hormone (3). In addition, the ␣ subunit gene is expressed in the placenta of primates and horses (4,5). Here, it is expressed in trophoblast cells as a subunit of chorionic gonadotropin in primates or as pregnant mare serum gonadotropin in horses. Because the human ␣ subunit gene is expressed in both placenta and pituitary, examination of its promoter allows a comparative analysis of the mechanisms involved in directing expression of the same gene to two separate tissues. Such studies have been aided by the availability of both placenta and pituitary cell lines that express the ␣ subunit gene.
Analysis of the human ␣ subunit promoter in placental cell lines (BeWo or JEG3) has revealed the presence of a composite enhancer located within the first 180 bp 1 of the promoter (6 -13). This enhancer confers placenta-specific expression and consists of an upstream regulatory element (URE) and two juxtaposed (tandem) cyclic AMP response elements (CREs). Many earlier studies revealed a synergistic relationship between the URE and CREs in placenta, suggesting a functional interaction between the proteins that bind these elements (7,9). Additional elements are also required for optimal expression in placenta cells and include the junctional response element (JRE) (14) and the CCAAT box (15).
The emerging model for gonadotrope-specific expression builds on the current model of placenta-specific expression. Although incomplete at this stage, transfection studies with mouse and human promoters containing deletion and clustered mutations have revealed the location of several elements important for expression in gonadotropes (16 -18). These elements reside within the first 435 bp of the human promoter and 507 bp of the mouse promoter (17,18).
In the human promoter, several elements have been identified as important for promoter activity in the pituitary gonadotrope cell line ␣T3. These include the gonadotrope-specific element (GSE) (18), the CREs, and the ␣ACT element (17,19). An additional element in the mouse ␣ promoter, the pituitary glycoprotein hormone basal element (PGBE), has been identified as an important regulator of basal transcription in gonadotropes (17,20). The significance of the PGBE in the human promoter is not known.
Although these studies add to our understanding of ␣ subunit expression in gonadotropes, a more extensive analysis of the promoter is clearly needed to help identify the regions important for regulation of this gene. We therefore utilized an array of single and double block mutations spanning nucleotides Ϫ430 to Ϫ95 of the human ␣ subunit promoter to further define the regulatory elements necessary for expression. These studies unveiled several potential interactions that collectively define a new composite regulatory element that controls pituitary-specific expression and established the presence of two new elements, ␣BE1 and ␣BE2, important for expression in gonadotropes.
DNA-All plasmid DNAs were prepared from overnight bacterial cultures using Qiagen DNA plasmid columns according to the supplier's protocol (Qiagen, Chatsworth, CA). Oligonucleotides were purchased from Midland Scientific (Midland, TX). Double-stranded oligonucleotides were generated by heating complementary oligonucleotides to 95°C in the presence of 10 mM Tris (8.0), 100 mM NaCl, and 1 mM EDTA and allowing them to cool slowly to room temperature. Radiolabeled oligonucleotides were generated using T4 polynucleotide kinase and [␥-32 P]ATP.
Clones-The human ␣ subunit promoter was isolated from H␣(Ϫ1500/ϩ45)CAT plasmid (12) by digestion with HindIII and subcloned into the HindIII site of the luciferase reporter vector pGL2 Basic (Promega) to generate H␣(Ϫ1500)Luc. H␣(Ϫ845)Luc was generated by digesting H␣(Ϫ1500)Luc with BglII and religation. H␣(Ϫ485)Luc was generated by digestion of H␣(Ϫ1500)Luc with NsiI and XhoI, blunting the DNA using a fill-in reaction with the Klenow fragment of DNA polymerase I, gel isolation of the DNA fragment, and ligation of the DNA ends.
Generation of each of the mutant constructs was accomplished by bidirectional PCR using Deep Vent DNA polymerase (New England Biolabs, Beverly, MA). The reactions were performed using the buffer supplied by the manufacturer in the presence of 300 M each dNTP, 100 pmol of each primer, 2 ng of H␣(Ϫ1500)CAT template, and 1.5 units of Deep Vent polymerase per 100-l reaction. For each clone, two independent PCR reactions were performed: an upstream PCR reaction using a standard 5Ј-primer plus a 3Ј-mutant primer and a downstream PCR reaction using a 5Ј-mutant primer and a 3Ј-standard primer. The primer set for each mutant contained overlapping sequence within the mutation in which there was a common restriction enzyme site. The reaction mixtures were precipitated with ethanol, and the DNA was redissolved in the appropriate restriction enzyme buffer and digested with restriction enzymes compatible with both the H␣(Ϫ845)Luc cloning vector and the restriction enzyme common to both mutant primers. The products of the digestion were separated by electrophoresis on an 6 -8% polyacrylamide gel in TBE and 5% glycerol. After staining, the bands of the appropriate size were excised and isolated. The new clones were generated by a triple ligation with each digested PCR product and a H␣(Ϫ845)Luc vector that was digested with either NsiI/SnaBI or SnaBI/PstI.
To generate the H␣(Ϫ1500)Luc mutants, the upstream fragment, which extends from the BglII site at Ϫ845 to the HindIII site at approximately Ϫ1500, was subcloned into each mutant using the BglII site and an appropriate site within the multiple cloning site. All clones were confirmed by sequencing using the Sequenase 2.0 sequencing kit from U. S. Biochemical Corp.
Double mutants were made by subcloning DNA fragments containing one mutation into a vector having the second mutation. The double mutant H␣(Ϫ1500)7/14 was made by subcloning the 600-bp SnaBI-BglII fragment of 7 into the analogous site in H␣(Ϫ1500)14. H␣(Ϫ1500)8/14 contains the SnaBI-BglII fragment of 8 inserted in the analogous position in H␣(Ϫ1500)14. H␣(Ϫ1500)11/14 contains the 846-bp SauI fragment of H␣(Ϫ1500)14 cloned into the analogous site in H␣(Ϫ1500)11. H␣(Ϫ1500)7/11 contains the 600-bp SnaBI-BglII fragment of 7 into the analogous site in 11. H␣(Ϫ1500)8/11 contains the 600-bp SnaBI-BglII fragment of 8 cloned into the analogous site of 11. The 7/8 clone required an additional PCR reaction. The PCR reactions for this mutant were the same as with the original 8 clone, except that the 3Ј-primer of the upstream PCR reaction was changed to 5Ј-CGTAAGAGCTCGGTACCTCCCGGGCTCGGTACCT-GTACCTGATATTTAC-3Ј, and the template for the PCR reaction was H␣(Ϫ1500)7. The clone was generated as described above. The mutated sequence within region 8 is altered slightly from the original 8 mutation in that it has an 8-bp deletion. Removal of bases from region 8 is unlikely to alter the interpretation of the results, as the binding sites for the proteins in question are still fully disrupted, and there are no upstream elements that would be affected by the spacing change.
Transfection and Enzyme Analysis-␣T3 cells were seeded onto 35-mm plates at a density of 170,000 cells per plate. The day after seeding, the cells were transfected using lipofectamine reagent according to the manufacturer's suggestions (Life Technologies, Inc.). Each construct (1.25 g) was cotransfected with 0.42 g of Rous sarcoma virus-␤-galactosidase, to control for transfection efficiency, in triplicate using 5 l of lipofectamine. The cells were incubated with the lipofectamine/DNA solution overnight, which was then replaced with complete medium. Cells were harvested 48 h after removal of the DNA using 150 l of lysis buffer (Promega). After removal from the plates, the lysate was centrifuged at 16,000 ϫ g at 4°C, and the supernatant was assayed for both luciferase and ␤-galactosidase activities. BeWo cells were seeded into 35-mm plates at a density of 85,000 cells per well. The day after plating, the cells were transfected and harvested as above, except 8 l of lipofectamine was used for transfection.
Luciferase assays were performed using 10 -20 l of lysate and 100 l of luciferase assay reagent (Promega). Emitted light was quantified as residual light units in a Berthold Lumat LB9501 luminometer (WALLAC Inc., Gaithersburg, MD). The light units emitted for the human wild-type ␣ subunit promoter were in the range of 1-3 ϫ 10 6 residual light units. ␤-Galactosidase activity was quantified by luminescence as above using the Galacto-light assay system (Tropix, Bedford, MA). The light units emitted for ␤-galactosidase activity were in the range of 2-6 ϫ 10 6 residual light units. The linear range of both assays was determined, and all data presented fell within those ranges. For each transfection, the luciferase/␤-galactosidase activity of each construct was normalized to the luciferase/␤-galactosidase activity of the wild-type promoter. The values were then averaged over a minimum of three independent experiments.
Electrophoretic Mobility Shift Analysis-Nuclear extracts were prepared as described elsewhere (22). Electrophoretic mobility shift analysis (EMSA) was performed as described previously with slight modification (23). 7 g of ␣T3 nuclear extract was incubated with 25-50 fmol of radiolabeled double-stranded oligonucleotide in the presence of 10 mM MgCl 2 , 25 mM KCl, 0.5 mM dithiothreitol, 12.5 mM HEPES (7.9), 0.5% Triton X-100, 200 ng of Escherichia coli DNA, 200 ng of salmon sperm DNA, 10% glycerol, and 1.0 g of poly(dI-dC):(dI-dC) in a 20-l reaction volume. Reactions were incubated on ice 5 min prior to the addition of probe and then an additional 10 min after. Reactions were loaded on a 5% polyacrylamide gel in 0.25 ϫ TBE and subjected to 240 volts for 3-4 h. The gels were dried and analyzed by autoradiography using Dupont's Reflections autoradiography film (DuPont NEN).
Southwestern Blot Analysis-This analysis involved modification of a procedure described elsewhere (18). 40 g of nuclear extract was resolved on an SDS-polyacrylamide gel with a 4% stacking gel and a 10% resolving gel and transferred to nitrocellulose membrane (Schleicher and Schuell). Membrane-bound proteins were renatured in binding buffer (20 mM HEPES (pH 7.9), 50 mM KCl, 1 mM dithiothreitol, 10% glycerol, and 0.1% Nonidet P-40) for 45 min at room temperature, and the filters were then blocked for 3 h at room temperature in binding buffer plus 5% nonfat dry milk. The membrane was then rinsed with binding buffer and hybridized overnight at 4°C in binding buffer containing 0.5% Carnation nonfat dry milk, 1 g/ml salmon sperm DNA, and 1 ϫ 10 6 cpm/ml of probe. The membrane was then rinsed one time in binding buffer and washed for 30 min at room temperature in binding buffer, changing the buffer one time during the interval. The blot was analyzed by autoradiography as described above. Prestained molecular weight markers (Life Technologies, Inc.) were used to estimate the molecular weights of the proteins.

Unique Combinatorial Arrays of Regulatory Elements Are
Important for Activity of the Human ␣ Subunit Promoter in Placenta and Pituitary-To better define the regions of the ␣ subunit promoter important for pituitary and placental expression, we have generated an extensive array of block mutations spanning nucleotides Ϫ430 to Ϫ95 of the human ␣ promoter and placed each mutation within the context of the naturally occurring 1500-bp sequence located upstream of the transcrip-tional start site. Fig. 1 identifies the bases altered in each mutation. Several of these mutations span regions that contain elements previously described as important for either human or mouse ␣ promoter activity and are identified in the figure legend. Promoters containing the various mutations were placed upstream of the luciferase reporter gene and assayed by transient transfection analysis in both gonadotrope (␣T3) and placenta (BeWo) cell lines.
Results from transfection in BeWo cells confirmed that activity of this promoter is regulated by three main elements that lie within the regions disrupted by 13 (URE), 14 (tandem CREs), and 15 (JRE) ( Fig. 2A). All of these reside within the first 182 nucleotides upstream of the transcriptional start site. Mutations in either the URE or the CREs had a dramatic effect on promoter activity, whereas the effect of the JRE mutation was less pronounced. These elements have been identified previously as important for placenta-specific expression (6 -14).
In contrast to BeWo cells, expression in ␣T3 cells requires a more complex array of elements that lie within regions disrupted by 7,8,11,13,14, and to a lesser extent 9 (Fig.  2B). Elements disrupted by 11, 13, and 14 have previously been described for their role in regulating expression of the human ␣ subunit gene in gonadotropes and correspond to the GSE, the URE, and the CREs, respectively (17)(18)(19). 7 is a mutation through the PGBE, an element that was described previously only in the mouse promoter (17,20). This study therefore establishes a role for the PGBE in transcriptional regulation of the human promoter as well. More importantly, the presence of a new element within the bases altered by 8 has been revealed. Further analysis of this element is presented below.
An important attribute of the block replacement approach is the ability to compare the relative contributions that each element makes to promoter function while maintaining the normal sequence environment. In the placental cell line, the tandem CREs are clearly the dominant basal element. Mutation of this element dramatically reduced promoter activity to 2% of wild type ( Fig. 2A). Mutation of the URE reduced activity to 15% of wild type, while mutation of the JRE only reduced activity to 32%. In contrast, the main elements involved in gonadotrope-specific expression had less dramatic effects on promoter activity, with the majority of the mutations reducing promoter activity to between 30 and 50% of wild type (Fig. 2B).
Although not as clearly dominant as in BeWo cells, the CREs did have the strongest effect on promoter activity in gonado-tropes. Thus, while expression in the placenta appears to be controlled by two strong elements, expression in the pituitary is regulated by an array of weaker elements. Furthermore, these results show that distinct arrays of regulatory elements direct expression to placenta and pituitary.
A New Transcriptional Element Resides within Region 8 of the Human ␣ Subunit Promoter-The scanning mutagenesis shown in Fig. 2B disclosed the presence of an undefined element(s) within the region disrupted by 8 that is important for expression in ␣T3 cells. This element resides between bases Ϫ320 and Ϫ288 of the human promoter and is juxtaposed to the PGBE. To further define the site and to ensure that the diminished promoter activity is not the result of abrogating the binding of protein(s) to PGBE, several additional mutations were made (Fig. 3A). Transfection analysis indicated that three of the more restricted mutations within this region (8.1, 8.2, and 8.4) had considerable impact on promoter activity. A mutation that abuts 8, 9.1, also diminished promoter activity (Fig. 3B). These smaller mutations define a new region important for promoter activity that is distinct from the PGBE (20).
Sequence analysis of this region disclosed the presence of a potential Ets protein-binding site spanning positions Ϫ301 to Ϫ298. This family of transcription factors interacts specifically with sequences containing a common core tetranucleotide, GGAA, where both guanines are required for binding (24 -29). A mutation that alters the essential guanines of this site (8.3) had little influence on promoter activity, suggesting that under basal conditions the transcription factor involved in activation of this promoter via this new element is not a member of the Ets family.
A Single Protein Complex Binds between Bases Ϫ316 and Ϫ302 of the ␣ Subunit Promoter-EMSA using a radiolabeled probe spanning bases within region 8 revealed the presence of a single major shifted complex from ␣T3 nuclear extracts (Fig.  4A). Formation of this major protein complex, indicated on the gel as ␣BP1 (␣ binding protein 1), was inhibited by increasing concentrations of unlabeled homologous competitor DNA but not by unlabeled competitor DNAs containing the 8 mutation sequence (Fig. 1) or a nonspecific sequence. An oligonucleotidecontaining sequence corresponding to the PGBE was also an ineffective competitor (Fig. 4B). Thus, ␣BP1 binds specifically to region 8 of the ␣ subunit promoter and is distinct from proteins binding the PGBE.
contain mutations within the 8 region was used to identify the bases important for binding of ␣BP1 (Fig. 5). Oligonucleotides 2 through 5, which contained mutations between bases Ϫ313 and Ϫ302, failed to compete the ␣BP1 signal, indicating that bases within this region are important for formation of the ␣BP1 complex. In contrast, oligonucleotides containing mutations downstream of base Ϫ302 (6 and 7) effectively competed for ␣BP1 binding, while an oligonucleotide containing  Fig. 1 were inserted into 1500 bp of the human ␣ subunit promoter and used to drive expression of the luciferase reporter gene. Human ␣ subunit promoter constructs were cotransfected with Rous sarcoma virus-␤-galactosidase into either ␣T3 or BeWo cells. The luciferase activity of each construct was determined and normalized to ␤-galactosidase activity to control for transfection efficiency. The data represent the luciferase/␤-galactosidase activity of each mutant or truncated promoter normalized to the luciferase/␤-galactosidase activity of the wildtype construct. mutations upstream of Ϫ312, 1, was capable of only partial competition. This indicates that the bases important for ␣BP1 binding lie between Ϫ316 and Ϫ302, defining the ␣ basal element one (␣BE1).
A Second Protein Complex Binds between Bases Ϫ296 and Ϫ285 of the ␣ Subunit Promoter-As indicated above, ␣BP1 binding is not affected by mutations downstream of base Ϫ302. However, transfection analysis indicated that bases within this downstream region are important for promoter activity (8.4 and 9.1, Fig. 3). This suggested that a second element resides within this more downstream region and binds a transcription factor distinct from ␣BP1. EMSA, using a radiolabeled oligonucleotide spanning through the 3Ј-portion of region 8 into region 9 (wt2), revealed the binding of a second protein complex, ␣ binding protein 2 (␣BP2, Fig. 6). Binding of ␣BP2 is competed by increasing concentrations of unlabeled wild-type competitor but not by either an oligonucleotide corresponding to the PGBE or a nonspecific oligonucleotide. Thus, ␣BP2 binds specifically to promoter sequences within this downstream oligonucleotide and does not correspond to proteins binding the PGBE.
EMSA in the presence of unlabeled competitor oligonucleotides containing mutations that span bases Ϫ306 to Ϫ285 further defined the region of ␣BP2 binding (Fig. 7). Mutations within 3, which disrupted the binding of ␣BP1, had little or no effect on the binding of ␣BP2, as indicated by the ability of the 3 oligonucleotide to compete effectively. However, mutations in 7, which had no effect on the binding of ␣BP1, interfered with the ability of the ␣BP2 complex to bind DNA. Thus, the bases required for ␣BP2 binding are clearly distinct from those of ␣BP1. This conclusion is also supported by the observation that the wild-type oligonucleotide used as the probe to reveal ␣BP1 binding is not effective as a competitor for ␣BP2 binding, and conversely the oligonucleotide used to reveal ␣BP2 binding (wt2) does not compete for the binding of ␣BP1 (Fig. 5). Thus a second element, distinct from ␣BE1, resides between bases Ϫ296 and Ϫ285 and defines the ␣ basal element two, ␣BE2.
␣BE1 Binds Both 54-and 56-kDa Proteins-Southwestern blot analysis using a radiolabeled probe corresponding to the wild-type oligonucleotide (see Fig. 5) unveiled the binding of two proteins having an approximate molecular mass of 54 and 56 kDa (Fig. 8). Both proteins bound specifically to this probe as demonstrated by the ability of unlabeled homologous competitor (wt), but not a competitor containing the mutation sequence through region 8 (8, Fig. 1), to diminish binding to the probe. Furthermore, oligonucleotides 2 and 3 did not compete for binding to these proteins, while 7 did compete (see Fig. 5 for sequence of 2, 3, and 7). This binding profile mimics that of the ␣BP1 complex observed by EMSA (Figs. 4  and 5), indicating that the 54-and 56-kDa proteins are components of the ␣BP1 complex.
Mutations in Multiple Elements Suggests Transcription Factors Binding the PGBE, the ␣BE, and the CRE Influence the Activities of Each Other-As shown above, activity of the ␣ FIG. 4. Electrophoretic mobility shift analysis of proteins binding to region 8 of the human ␣ subunit promoter. A, radiolabeled probe corresponding to bases within region 8 (wt, 5Ј-TGTCTCT-TGTTTATAGGAAAGTGTCAGCT-3Ј) of the human ␣ subunit promoter was used in an electrophoretic mobility shift assay with ␣T3 nuclear extracts. Unlabeled competitors were added to the indicated assays and include wild type (wt, homologous competitor added in increments of 50ϫ, 100ϫ, 200ϫ, 500ϫ, 1000ϫ), 8 (oligonucleotide containing the sequence to which region 8 was mutated as shown in Fig. 1, 1000ϫ), and nonspecific competitor (NS, 1000ϫ). B, electrophoretic mobility shift assay using the radiolabeled wild-type oligonucleotide as probe and ␣T3 nuclear extracts. Unlabeled competitors were added to the indicated assays and include wild type (200ϫ), PGBE (5Ј-ATATCAGGTACT-TAGCTAATTAAATGTCT-3Ј, 200ϫ), and nonspecific competitor (NS, 200ϫ).

FIG. 5. Electrophoretic mobility shift analysis of ␣BP1 using wild-type and mutant competitors.
A, sequence of oligonucleotides used in electrophoretic mobility shift analysis. The top shows the wildtype promoter sequence from Ϫ319 to Ϫ277. The boxed area indicates the sequence of the oligonucleotide radiolabeled and used as a probe (wt). The base mutations are shown for each oligonucleotide. A dash indicates no change from wild type. B, a radiolabeled probe corresponding to bases Ϫ316 to Ϫ288 (wt) was used in an electrophoretic mobility shift assay with ␣T3 nuclear extracts. Competitors were added at a 100-fold molar excess. NS corresponds to an oligonucleotide having a nonspecific sequence.
FIG. 6. Electrophoretic mobility shift analysis of proteins binding to ؊306 to ؊277 of the human ␣ subunit promoter. A, radiolabeled probe corresponding to bases Ϫ306 to Ϫ277 bp (wt2, 5Ј-TTATAGGAAAGTGTCAGCTTTCAGGATGTT-3Ј) of the human ␣ subunit promoter was used in an electrophoretic mobility shift assay with ␣T3 nuclear extracts. Unlabeled competitors were added to the indicated assays and include homologous competitor (wt2) added in increments of 50ϫ, 100ϫ, and 200ϫ, PGBE (200ϫ), and nonspecific competitor (NS, 200ϫ).
subunit promoter in ␣T3 cells is not controlled by a single dominant element but by an array of weaker elements. This suggested that interactions between regulatory elements and their cognate transcription factors may play a key role in dictating cell-specific expression. To address this, a number of promoters with block mutations through two basal elements were evaluated for their effects on activity (Fig. 9).
Assessment of promoters with paired mutations in PGBE and GSE (7/11), ␣BE(1ϩ2) and GSE (8/11), and GSE and CRE (11/14) indicated that mutations through the GSE had the same relative impact on promoter activity in the presence or absence of PGBE, ␣BE(1ϩ2), or CRE. Thus, a mutation through the GSE decreased promoter activity by approximately 2.5-fold when mutated in the intact promoter or promoters with an additional mutation in PGBE, ␣BE(1ϩ2), or CRE. In contrast, a paired mutation through PGBE/␣BE (7/8) had no further impact on promoter activity than either of the individual mutations, suggesting that both sites must be present for either to be functional. Interestingly, paired mutations in PGBE/CRE (7/14) or ␣BE/CRE (8/14) resulted in a greater loss in promoter activity than the independent contributions of the individual elements. For example, in the intact promoter, a mutation through the CREs decreased promoter activity by approximately 6-fold, while in the absence of either PGBE or ␣BE(1ϩ2), a second mutation through the CRE decreased promoter activity by an additional 50-fold. Moreover, in the intact promoter, a mutation through PGBE or ␣BE(1ϩ2) decreased promoter activity by approximately 3-fold, while in the absence of the CRE an additional mutation in either of these elements decreased promoter activity by approximately 20-fold. Thus, mutations in multiple elements revealed that the transcription factors binding the PGBE, the ␣BE, and the CRE influence the transactivation abilities of each other, while the activity of the transcription factor binding the GSE does not appear to be influenced by the presence of the activators binding the other three elements. DISCUSSION A key objective of this study was to address how the ␣ subunit gene achieves its highly restrictive pattern of expression. Our comparative analysis in BeWo and ␣T3 cells has revealed a number of differences as well as similarities between the regulatory codes that direct expression to placenta and pituitary. While placenta expression is predominantly controlled by two very strong elements, the URE and the CREs, pituitary expression is regulated by an array of weaker elements. This suggests that no single element is absolutely essential for ␣ subunit gene expression in pituitary. Instead, mutational analysis of multiple elements indicated that there are a number of key interactions that are important for gonadotrope-specific expression.
It is clear from our studies and others that the CREs of the human ␣ subunit promoter play an important role in the expression of this gene in both placenta and pituitary (6,17,19,30). In placental cells, the CREs bind homodimers of the ubiquitous protein CREB. We have preliminary data indicating that CREB binds the CREs in ␣T3 cells as well (data not shown). The URE also plays a role in both placenta and pituitary expression of the human ␣ subunit gene. In placenta, the URE is actually a composite of several elements; the ␣ACT, the trophoblast-specific element (TSE), and the URE1 (31). Here, the ␣ACT element binds to hGATA-2 and hGATA-3 (19), while the TSE/URE1 forms an overlapping element that may bind two functionally interchangeable proteins, TSEB and UREB (31).
In ␣T3 cells, mutation of the URE had a considerably less dramatic effect on promoter activity than in BeWo cells, suggesting that it is less critical for pituitary expression than for placenta. Steger et al. (19) have shown that the activity associated with the URE in ␣T3 cells is due at least in part to the ␣ACT element. This element both binds to and is activated by the transcription factor hGATA-2.
Gonadotrope-specific expression of the ␣ subunit gene requires four additional elements that are not needed for expres- FIG. 7. Electrophoretic mobility shift analysis of ␣BP2 using wild-type and mutant competitors. A, sequence of oligonucleotides used in electrophoretic mobility shift analysis. The top shows the wildtype promoter sequence between Ϫ306 to Ϫ277 and indicates the sequence of the oligonucleotide radiolabeled and used as a probe (wt2). The base mutations are shown for each oligonucleotide. A dash indicates no change from wild type. Note that oligonucleotide 8 is not the same as the previous 8 which has the entire sequence of the region 8 mutated. B, a radiolabeled probe corresponding to bases Ϫ306 to Ϫ277 (wt2) was used in an electrophoretic mobility shift assay with ␣T3 nuclear extracts. Competitors were added at a 100-fold molar excess. wt corresponds to an oligonucleotide with the sequence given in Fig. 5. NS corresponds to an oligonucleotide having a nonspecific sequence.
FIG. 8. Southwestern blot analysis of ␣BE1 binding proteins. 40 g of ␣T3 nuclear extracts were resolved on an SDS-polyacrylamide gel with a 4% stacking gel and a 10% resolving gel and transferred to nitrocellulose membrane. Membrane-bound proteins were renatured, blocked, and probed with a radiolabeled probe corresponding to ␣BE1 (Ϫ316 to Ϫ288). Competitors, when included, were at a 200-fold molar excess to that of the probe. Unlabeled competitors include wild type (homologous), 8 (corresponding mutation through region 8 as indicated in Fig. 1), and 2, 3, and 7 (sequences of 2, 3, and 7 are given in Fig. 5). The results shown from Southwestern blots in the presence of oligonucleotides 2, 3, and 7 were from a different experiment than those done in the absence of competitor, wild-type competitor, or 8. In this case, the control blot in the absence of added competitor and presence of wild-type competitor was nearly identical to the one shown here and therefore was not included. Migrations of molecular weight markers are indicated on the right. sion in placenta: the GSE, the PGBE, the ␣BE1, and the ␣BE2. The GSE was described previously for its role in regulating the human promoter and binds the orphan nuclear receptor steroidogenic factor-1 (SF-1) (18,32,33). The PGBE has only been described for it role in regulating activity of the mouse promoter, and one of the proteins that binds this element has been identified as the Lim-homeodomain protein, LH2 (17,20,34). Our studies show that PGBE is involved in the regulation of the human promoter as well. More importantly, we have shown through transfection analysis and binding studies the presence of two new elements, ␣BE1 and ␣BE2. These elements bind two different transcription factors, ␣BP1 and ␣BP2, respectively, both of which contribute to the function of the ␣ subunit promoter. The binding of multiple transcription factors to this portion of the promoter is consistent with an earlier report indicating that these regions as well as PGBE are protected from digestion by DNase I (18).
Southwestern blot analysis revealed that ␣BE1 binds two proteins of similar molecular weights, 54,000 and 56,000. The binding profile of these proteins mimics that of the ␣BP1 complex observed by EMSA (Figs. 5 and 8), indicating that the 54and 56-kDa proteins are components of the ␣BP1 complex. The presence of two proteins by Southwestern blot analysis as opposed to the single ␣BP1 band observed by EMSA suggests that the 54-and 56-kDa proteins may comigrate as a single band on an EMSA gel or that ␣BP1 is a higher order complex consisting of both proteins. The relationship of the two proteins is not yet known, but it is possible that they are different forms of the same protein. Attempts to reveal ␣BP2 by Southwestern blot analysis were unsuccessful, suggesting that this protein may be a heterodimer incapable of binding ␣BE2 when resolved on a denaturing gel. It may also be that the conditions for Southwestern blot analysis are too harsh to reconstitute ␣BP2 binding activity.
Mutational analysis of multiple elements was used to identify potential interactions between regulatory elements and their cognate transcription factors that play a role in cellspecific expression. This analysis disclosed three types of interactions: independent, dependent, and compensatory (Fig. 9). The studies indicated that the GSE activates the promoter in a manner that is independent from PGBE, ␣BE(1ϩ2), or CRE. That is, when a mutation is present in PGBE (7), ␣BE(1ϩ2) (8), or the CREs (14), a second mutation in the GSE(11) has the same relative effect on promoter activity as it does when the rest of the promoter is intact. This implies that SF-1, the GSE binding protein, can activate transcription to the same degree in the presence or absence of the proteins binding the PGBE, the ␣BE(1ϩ2), or the CREs.
This contrasts to what was observed when mutations were made in both PGBE (7) and ␣BE(1ϩ2) (8). In this case, no further impact on promoter activity was observed with the double mutation (7/8) when compared to either mutation alone. This suggests that both sites must be present (and occupied) to get promoter activation, revealing a dependent or synergistic relationship between the proteins that bind these elements.
A third type of interaction was implicated by the results of the double mutations through the PGBE/CRE(7/14) and the ␣BE/CRE (8/14). In the absence of either PGBE (7) or ␣BE(1ϩ2) (8), a mutation in the CRE(14) had a much more dramatic effect on promoter activity than in the presence of these elements. The reverse is also true. That is, in the absence of the CRE, a mutation in either PGBE or ␣BE had a much more dramatic effect on promoter activity than in the presence of the CRE. This effect suggests that PGBE and ␣BE become more important in the absence of a functional CRE and vice versa. It appears that the PGBE/␣BE-binding proteins can partially compensate for the loss of the CRE-binding proteins and that the CRE-binding protein(s) can partially compensate for the loss of the PGBE and ␣BE-binding proteins. The fact that the binding of one protein can make up for the loss of the other suggests that the activation pathways of these proteins converge at some point to stimulate transcription and suggests the possibility that the proteins binding these elements share a common coactivator or member of the general transcription machinery.
A different way of interpreting these results is that proteins binding the CREs partially interfere with the ability of the proteins binding PGBE and ␣BE(1ϩ2) to activate transcription and vice versa. Thus, in the absence of one protein the other is a much stronger activator. Again, this suggests that proteins binding these elements interact with each other either directly or indirectly.
The double replacement mutagenesis strategy has disclosed several potential interactions between proteins that bind the CRE, PGBE, ␣BE1, and ␣BE2. This suggests that gonadotropespecific expression occurs through a unique composite element that, like the placenta-specific enhancer, includes the CRE. This provides incentive for future studies to identify the biochemical nature of these interactions.