Functional Map of a Placenta-specific Enhancer of the Human Leukemia Inhibitory Factor Receptor Gene*

We recently reported a placenta-specific enhancer in the human leukemia inhibitory factor receptor (LIFR) gene and now show detailed characterization of the 226-base pair enhancer (−4625/−4400 nucleotides). Four of twenty-two mutants in linker analysis showed reduced promoter activities to 45, 30, 10, and 10%, respectively. Specific binding of region A (−4617/−4602) with nuclear extract was competed by a known Oct-1 oligo and supershifted by Oct-1 antibody. Specific binding of region B (−4549/−4535) was competed by a GATA oligo, but could not be supershifted by four GATA antibodies. Nevertheless, mutagenesis showed that critical bases in region B were identical to the GATA core motif, indicating that region B may bind to a novel GATA family transcription factor. The other two adjacent regions designated as region C (−4464/−4445) showed no known consensus binding sites, and their specific placental JEG-3 nuclear extract binding was not evident in nonplacental nuclear extracts and was not competed by a trophoblast specific element (TSE), indicating that region C is a novel placenta-specific element (PSE, CATTTCCTGAACTAGTTTTT). Footprinting localized the binding boundary of PSE-binding protein (PSEB), and three Gs were found to be important for specific PSE binding. UV cross-linking showed that PSEB had a molecular mass of ∼160 kDa, substituting the PSE with two previously reported placenta elements TSE or chorionic somatomammotropin enhancer factor 1 (CSEF-1) motifs resulted in markedly different promoter activities, indicating that PSEB is indeed different from TSE binding protein or CSEF-1. These results are the first demonstration that a novel PSE is the major element for placenta-specific enhancer activity in humanLIFR gene.

The leukemia inhibitory factor receptor (LIFR), 1 a member of the hematopoietic cytokine receptor family, includes receptors for cytokines functioning in immune and hematopoietic systems such as interleukins 2-7 and 9, erythropoietin, granulocyte macrophage colony-stimulating factor, and granulocyte colony-stimulating factor (1)(2)(3). Additional receptors for factors normally functioning outside the immune and hematopoietic systems, such as growth hormone, prolactin, ciliary neurotropic factor, and leptin are also members of this group (1,4,5). The LIFR was found to contain 1097 amino acid residues (a 44residue signal sequence, a 789-residue extracellular domain, a 26-residue transmembrane domain, and a 238-residue cytoplasmic domain) (6), and exhibits the characteristic structure of this receptor family, including two folding domains in the extracellular region and a Trp-Ser-X-Trp-Ser motif in the Cterminal (1). The LIFR heterodimerizes with gp130 to mediate intracellular signaling of interleukin-6 (IL-6) cytokine family members and is a substrate for mitogen-activated protein kinase (7)(8)(9)(10)(11)(12). The phenotype of the LIFR knockout mouse demonstrates that LIFR is essential for animal survival (13,14), whereby homozygote animals died within 24 h of birth (13,14), and severely affected tissues include placenta, bone, liver, and neurons (13). Disrupted placental architecture results in poor intrauterine nutrition; and reduced bone volume and increased osteoclast numbers lead to imbalanced bone development and severe perinatal osteopenia, excessive fetal hepatic glycogen storage, and significant neuronal losses (13,14).
In humans, endometrial LIF mRNA is significantly increased in the mid and late secretory phase compared with the proliferative phase of the menstrual cycle (22,23). Although the LIFR is undetectable in nonpregnant endometrium, LIFR mRNA is highly expressed in the chorionic villus during the first trimester and in term placenta (24), suggesting that LIFR may play a regulatory role in trophoblast growth and differentiation in the human placenta. These observations indicate the important functions of LIF/LIFR system in placental development and pregnancy, and are supported by the LIF/LIFR knockout results. These findings also suggests that specific gene transcriptional regulation of the LIF/LIFR system occurs during these reproductive stages, and malfunctioning of specific regulation would impair and/or lead to failure of the normal pregnancy process.
We have recently demonstrated that the human LIFR gene utilizes alternative promoters to regulate its expression in placental and nonplacental tissues (25). Interestingly, a placentaspecific enhancer was identified ϳ4.6 kb upstream to the placenta-specific promoter (25). The minimal enhancer, 226 bp, increases heterologous promoter activity 10 -35-fold when placed upstream or downstream of the reporter gene in either orientation (25). Within this 226-bp region, several potential transcription factor binding sites were predicted, but site-directed mutagenesis failed to correlate two transcription factors, Sp1 and NF-B, with enhancer activity (24). Here we report a detailed characterization of this minimal enhancer with three critical elements contributing to the enhancer activity, an Oct-1 binding site at Ϫ4617/Ϫ4602 site; a GATA-like element at Ϫ4549/Ϫ4535 site, and a novel placenta-specific element (PSE) at Ϫ4464/Ϫ4445.
Oligos-Oligos used for electrophoretic mobility shift assay (EMSA) assays are listed in Table I.
Linker Analysis and Mutagenesis-A total of 22 10-bp replacement (AGCTTAAGCT) mutants in linker analysis were generated by ExSite (Stratagene) or Kunkel's method (26). Site-directed mutagenesis disrupting Oct-1 binding site or GATA binding site and substitution mutagenesis with TSE or chorionic somatomammotropin enhancer factor 1 (CSEF-1) binding sites at PSE position were also performed. The mutagenic primer sequences are available upon request. All mutants were verified by sequencing.
Transient Cell Transfection-All plasmids were prepared using Maxi-prep kit (Qiagen). pCMV ␤-galactosidase was co-transfected as an internal control. JEG-3 were transfected using standard Lipo-fectAMINE method (Life Technologies Inc.). Transfections were performed in triplicate. For linker analysis, all 22 mutant plasmids to-gether with control pGL3PX-(SK) 1 (Ϫ4400/Ϫ4625 nt) were transfected in the same experiment and repeated twice. 48 h after transfection, cell lysates were prepared for measurement of luciferase and ␤-galactosidase activities.
Nuclear Extract Preparation, EMSA, and Supershift Assay-Crude nuclear extracts from JEG-3, HeLa, TC1, MCF-7, U-2 OS, AtT20, and GC cells were prepared as previously reported (25). The protein concentrations of the nuclear extracts were quantitated by Bio-Rad assay (Bio-Rad). Equal amounts of nuclear extract from different cells were used in EMSA. Labeled oligonucleotide duplex (20,000 -30,000 cpm) was mixed with ϳ5 g of nuclear extract, 1 g of poly(dI-dC) in 25 l of reaction buffer. For Oct-1 and GATA-like elements, the binding buffer was 10 mM Tris-Cl, pH7.5, 100 mM KCl, 1 mM dithiothreitol, 8% glycerol. For the novel placental element, the binding buffer was 10 mM Hepes, pH 7.6, 50 mM KCl, 0.5 mM dithiothreitol, 0.1 mM EDTA, 10% glycerol. For competition assay, 200-fold excess cold competitor oligos were added prior to the addition of labeled probe. For supershift assay, antibody was added after addition of labeled probe and incubated at room temperature for 1 h or 4°C overnight. EMSA samples were resolved on 5% nondenaturing polyacrylamide gel electrophoresis gel.
Mutant oligos disrupting G residues and their 3Ј-neighboring bases in this region were also used as competitor oligos to test the contribution of Gs to the binding complex and are listed in Table I. For location of the minimal region of PSE, nested oligos were synthesized as in Table  I and used as competitor oligos.
DNase I Footprinting-DNase I footprinting was used to localize the binding boundary of placenta-specific element binding protein. The DNA probe was generated by a polymerase chain reaction in which only one primer was labeled with [␥-32 P]ATP. Probes for both plus and minus strands were mixed with 30 g of JEG-3 nuclear extract in the binding buffer as used in EMSA. After incubation on ice for 30 min, various amounts of DNase I were added and allowed to incubate at room temperature for 1 min. Stop buffer (200 mM NaCl, 2 mM EDTA, 1% SDS, 50 g/ml tRNA) was added to terminate the reaction. For the competition assays, 200-fold placenta-specific element characterized in EMSA was used as cold competitors. Samples were analyzed on 9% 8 M urea-PAGE gels. A GA ladder reaction was performed according to standard methods (26).
UV Cross-linking-For UV cross-linking, 50,000 cpm oligo duplex probe was incubated with 10 g of JEG-3 nuclear extract and resolved as for EMSA. The wet gel was then irradiated with a 312-nm UV transilluminator, and the corresponding gel slice excised after autoradiography, eluted at room temperature in 10 mM Tris, pH 7.2, 0.1% SDS, 0.5 mM EDTA, 1 mM dithiothreitol, 100 mM NaCl for 2 h, and precipitated with 3 volumes of acetone. The protein precipitate was then resuspended in 1ϫ SDS-PAGE loading buffer (26) and loaded onto a 10% SDS-PAGE along with protein marker. The gel was dried and autoradiographed.

RESULTS
Linker Analysis Reveals Three Critical Regions within the LIFR Placenta-specific Enhancer-Previous transfection experiments demonstrated that the cloned human LIFR gene enhancer was active in choriocarcinoma JEG-3 and JAR cells, and inactive in cell lines derived from nonplacental tissues such as liver, bone marrow, thyroid, breast, and lung (25). At that time, we did not specifically test pituitary cell lines, but placental and pituitary cells share considerable similarities for transcriptional regulation of several genes, including growth hormone, glycoprotein hormone ␣ subunit, chorionic somatomammotropin B (CS-B, or placental lactogen, PL) (27)(28)(29). We therefore chose several available pituitary cell lines, namely ACTH-producing AtT20 and growth hormone secreting GH4 and GC cell lines to further test the specificity of the cloned LIFR gene enhancer and two LIFR promoters. The transfection results shown in Fig. 1 depict the inactive placenta-specific LIFR promoter in pituitary cells, while the alternative LIFR promoter (25) is strongly active in these pituitary cells. Also, the LIFR gene enhancer, active in placental cell lines, is not active in the three pituitary cell lines tested, further demonstrating its stringent tissue specificity.
Data base searching predicted several potential transcrip-tion factor binding sites within this minimal enhancer region (25), including cyclic AMP-response element (CRE), Sp1, GATA, and NF-B. Notably, three NF-B/c-Rel binding sites and two Sp1 binding sites were revealed (25). Since both Sp1 and NF-[kapppa]B participate in enhancer functions, we per-    (Table I). Lane 1, oligo a only, without any competitor; in lanes 2-4, cold competitor oligos as indicated were added at 200-fold excess concentrations; lane 5 was the same as lane 1 except that an Oct-1 antibody was also added in the mixture.
formed site-directed mutagenesis to disrupt critical bases within the corresponding Sp1 or NF-B consensus regions within this enhancer, but failed to observe loss of enhancer activity (25). Thus, it appears that the predicted Sp1 and NF-B/c-Rel binding sites are irrelevant to enhancer activity in LIFR gene.
To characterize this placenta-specific enhancer from a functional point of view, we performed a comprehensive linker analysis with the introduction of 10 bp of replacement mutation covering the entire enhancer. Twenty-two mutants were thus obtained, and their sequences were confirmed. These mutants together with an enhancer control were transfected into JEG-3 cells, and the result is as depicted in Fig. 2. Four mutants demonstrated consistent loss of enhancer activity. Mu-tants 2, 9, 17, and 18 have base substitutions from Ϫ4614 to Ϫ4605 nt, Ϫ4544 to Ϫ4535 nt, Ϫ4464 to Ϫ4455 nt, and Ϫ4454 to Ϫ4445 nt, respectively, and their corresponding enhancer activities are reduced to 45, 30, 10, and 10%, respectively. Also, mutant 17 and mutant 18 have adjacent base substitutions. Taken together, these results indicate that three elements in this enhancer, namely A (Ϫ4614/Ϫ4605 nt), B (Ϫ4544/Ϫ4535 nt), and C (Ϫ4464/Ϫ4445 nt) are critical to the full enhancer activity. Subsequently, a new data base search was performed specifically for these three elements, and the result is as shown in Fig. 3. Potential transcription factors capable of binding these regions were identified including Oct-1 and GATA.
Region A Is an Oct-1 Binding Site-Region A (Ϫ4614/Ϫ4605 nt) was found to have considerable homology to a known tran-  Table I), was used as probe; for lanes 11-18, a known GATA oligo (Table I)  scription factor Oct-1 consensus binding site (30), although a previous data base search using higher stringency failed to predict a potential binding site in this locus. To test whether Oct-1 does in fact bind to region A, both competition EMSA and Oct-1 antibody supershift assay were performed. An oligo a (Ϫ4617/Ϫ4602 nt) covering region A was used as a probe to perform these assay (Fig. 4). This oligo produced a specific binding complex as determined by competition assay, including using itself as a cold competitor (lanes 1-3). Specific binding of the complex was also dissipated when a known Oct-1 oligo was added separately as competitor; moreover, this binding complex was supershifted with the addition of an Oct-1 antibody (lane 5). These results demonstrate that the transcription factor Oct-1 binds to element A. Also, cotransfection of Oct-1 expression vector into JEG-3 cells slightly increased enhancer activity, similar to previous observations that although TEF-5 is required for hCS-B enhancer activity and binds to two elements necessary for hCS-B enhancer activity, overexpression of TEF-5 does not activate hCS-B enhancer activity (31).
Region B Is a GATA-like Element-Region B (Ϫ4544/Ϫ4535 nt) exhibited high homology to the GATA family member consensus binding site (WGATAR), as demonstrated both in a previous search (25) and a current search after linker analysis. Since GATA family members are implicated in the placentaspecific regulation of glycoprotein hormone ␣ subunit and adenosine deaminase (ADA) (32,33), it was important to determine whether region B is GATA-related.
Competition experiments with EMSA showed that the specific binding complex formed by an oligo b (Ϫ4549/Ϫ4535 nt) covering region B and JEG-3 nuclear extract was competed by a known GATA oligo at 200-fold excess concentration (Fig. 5A,  lanes 1-4), while specific complexes formed by the known GATA oligo probe and JEG-3 nuclear extract could only partially be competed by the test oligo b at 200-fold excess concentration (Fig. 5A, lanes 11-14), suggesting that region B is a GATA-like element; GATA oligo efficiently binds to the same proteins that this element can bind, but the element cannot efficiently bind to all proteins to which the GATA oligo binds. This hypothesis was supported by supershift assays using antibodies to GATA-1, -2, -3, and -4 as shown in Fig. 5A, lanes 5-8 and lanes 15-18. Specific complexes between the known GATA oligo with JEG-3 nuclear extract were supershifted by antibody to GATA-2, -3, or -4, indicating the capability of the known GATA oligo to bind to GATA-2, -3, or -4 proteins, which are known to be present in JEG-3 cells (32,34). In contrast, specific complexes between the oligo b and JEG-3 nuclear extract could not be supershifted by antibody to GATA-2, -3, or -4, indicating that these specific complexes involved a protein other than GATA-2, -3, or -4 recognizing region B.
Previous observations revealed that mutation of two bases within the GATA consensus binding site (GA 3 CT in WGA-TAR) abolishes binding with GATA family members. To test whether these two bases are also critical within this GATA-like element, we performed site-directed mutagenesis to change GA in region B to CT, and in transient transfection assays (Fig. 5B) showed that this new mutant lost enhancer activity to a similar extent as the 10-bp replacement mutant 9, demonstrating that this GA is a critical contact site between region B and its corresponding binding protein.
The tissue distribution pattern of this corresponding protein was examined by EMSA using nuclear extracts from cell lines of different tissue origin (Fig. 5C). As depicted, this protein is also present in all nuclear extracts tested, showing that it is not placenta-specific; however, it seems to be more abundant in placental JEG-3 cells than other nonplacental cells.
Region C Is a Novel PSE-Another critical region determined by linker analysis spanned from Ϫ4464 to Ϫ4445 nt, which possesses a stronger effect on enhancer activity loss than does region A or B. Data base searching also failed to indicate corresponding transcription factor(s) for this region. From the results shown above, it is apparent that binding proteins for region A and B are present in many cell types and therefore placenta specificity of the enhancer cannot be attributed to these elements.
Using an oligo c (Ϫ4464/Ϫ4445 nt) covering region C as a probe for EMSA with nuclear extracts from several cells, including JEG-3, HeLa, GC, U-2 OS, AtT20, and MCF-7, a specific binding complex was observed only in placental JEG-3 cells (Fig. 6). Furthermore, this specific complex could not be competed by a previously identified placenta-specific trophoblast specific element (TSE) oligo (35) at 200-fold excess concentration (Fig. 6, lane 4). These results indicate that region C contains a novel PSE, and its binding with corresponding protein(s) is a major factor contributing to the tissue specificity and activity of the LIFR gene enhancer.
DNase I footprinting was subsequently performed to determine the boundary of the corresponding binding protein on the enhancer region. As shown in Fig. 7, specific footprints were observed on both plus (Ϫ4455 to Ϫ4436 nt) and minus (Ϫ4462 to Ϫ4437 nt) strands. Analysis of the base composition of oligo c (Ϫ4464/Ϫ4445 nucletides) revealed a total of 8 Gs in both plus and minus strands. Since G has been found to be involved in various specific DNA-protein interactions, and most mutations for disrupting specific DNA-protein interactions usually utilize simultaneous mutations at two neighboring bases we designed a total of 6 oligos mutating every G and its 3Ј-neighboring base for use in the competition assay (Fig. 7C, lanes 4 -9). As depicted, lanes 5, 6, and 9 using mutant oligos cM2, cM3, and cM6 exhibit the same specific binding complex as lane 1 without FIG. 6. Placenta-specific binding complex between region C and JEG-3 nuclear extract. In EMSA, oligo c (Ϫ4464/Ϫ4445 nt) was used as probe (Table I). The same amount (4 g) of nuclear extracts from several indicated cell lines was used. In lanes 2 and 3, cold competitor oligos, oligo c, and AP-1, respectively, were added at 200-fold excess concentration, while in lane 4 cold TSE oligo (a reported placenta-specific oligo) was added at 200-fold excess concentration. The placenta-specific binding complex is arrowed. added competitor, suggesting that mutant oligos cM2, cM3, and cM6 used in lanes 5, 6, and 9 contained mutations at critical bases, as underlined in CATTTCCTGAACTAGTTTTT. To determine the minimal region of PSE required for specific binding to this protein, we designed a set of nested oligos and used them as competitor oligos to test their ability to compete the specific binding. A typical competiton assay to determine this minimal region is shown in Fig. 7C (lanes 10 -12). The 14-bp oligo from Ϫ4459 to Ϫ4446 nt (lane 11) competed specific binding, while a 12-bp oligo from Ϫ4458 to Ϫ4447 nt (lane 12) did not compete. Thus the 14-bp element (Ϫ4459/Ϫ4446 nt) was defined as the minimal functioning element for region C.
Two previously reported DNA binding proteins, TSE binding protein (TSEB) and chorionic somatomammotropin enhancer factor 1 (CSEF-1), are implied to play important roles in regulating several placenta-specific transcripts and appear to be detected mainly in placental cell lines. In order to see whether this PSE bind protein (PSEB) is indeed different from TSEB or CSEF-1 and whether these PSE, TSE, and CSEF-1 binding sites are functionally interchangeable, we made two mutants with substitutions at the PSE region in the enhancer to the TSE and CSEF-1 binding sites, respectively. Transfection into placental JEG-3 cells and nonplacental U-2 OS cells were performed, and the results are as shown in Fig. 8A. As depicted, substitution of PSE with TSE or CSEF-1 binding sites resulted in 1 ⁄5-and 4-fold difference of promoter activity, respectively, as compared with the wild-type construct. And all three constructs showed no promoter induction in U-2 OS cells, demonstrating the inactiveness of these constructs in nonplacental cells. We also performed UV cross-linking to determine the molecular mass of this PSEB. As shown in Fig. 8B, the free probe recovered from EMSA showed no signal in SDS-PAGE after UV cross-linking, while the placenta-specific binding complex recovered from EMSA revealed a molecular mass band of approximately 160 kDa in SDS-PAGE after UV cross-linking. Since TSEB and CSEF-1 have molecular masses of 56 and 30 strand of LIFR placenta-specific enhancer. A corresponding single end-labeled probe was used; in lane 1, no nuclear extract was added and 0.1 unit of DNase I was added; in lanes 2-5, 40 g of JEG-3 nuclear extract were added with DNase I at 0.2, 0.5, 0.8, and 1 unit, respectively; in lanes 6 and 7, 40 g of JEG-3 nuclear extract were added with 0.5 unit of DNase I and 200-and 100-fold cold oligo c competitor included, respectively. Sequences of the protected region are shown by the break in the brackets. C, determinations of G residue involvement in PSEB binding and minimal functional region for PSE. In EMSA, oligo c was used as a probe, with JEG-3 nuclear extract. Different cold competitor oligos were used as indicated. Lanes 1-3 demonstrate specific complexes formed between oligo c and nuclear protein as arrowed; lanes 4 -9 show the use of cold competitor oligos (Table I) harboring base substitutions at G residues and their 3Ј-neighboring bases; lanes 10 -12 show the use of shortened cold competitor oligo (Table I) to determine the minimal functional PSE region. In lanes 4 -12, disruption of critical bases results in the competitor oligos losing their competing abilities and thus the specific complexes persist, as depicted in lanes 5, 6, 9, and 12.   FIG. 8. PSEB is a novel placentaspecific transcription factor. A, substitutions of PSE with two previously reported placenta-specific elements result in different promoter activity. Mut17 and mut18 are the same mutants obtained in linker analysis (Fig. 2), mutX and mutY have a TSE and CSEF-1 binidng site, respectively, in the PSE position. Plamsids, including pGL3-promoter control, were co-transfected with pCMV-␤-galactosidase into placental JEG-3 cells and nonplacental U-2 OS cells, and luciferase activity was normalized to ␤-galactosidase activity. Values shown represent mean Ϯ S.E. of triplicate determinations. B, UV cross-linking between oligo c and JEG-3 nuclear extract. UV cross-linking was performed as indicated under "Experimental Procedures." Protein marker sizes are indicated. Lane 1, gel extract eluant from the free probe band; lane 2, gel extract eluant from specific binding complex between oligo c and JEG-3 nuclear extract. kDa, respectively (36,37), it is therefore apparent that PSEB is a novel protein rather than TSEB or CSEF-1.

DISCUSSION
Placenta-specific transcriptional controls fall into two broad categories: first, they may share the same promoter as other tissues, but utilize placenta-specific enhancers or upstream elements as observed for glycoprotein hormone ␣ subunit, CS-B, adenosine deaminase (ADA), and leptin genes (28,29,32,38,39); second, they may utilize both placenta-specific promoters and enhancers, as previously shown for the aromatase gene (33) and as we now show for the LIFR gene.
A composite enhancer (Ϫ180/Ϫ111 bp) of the glycoprotein hormone ␣ subunit gene promoter is responsible for its placental expression rather than pituitary gonadotropic or thyrotropic expression (29). This enhancer was composed of two CREs and an upstream regulatory element (URE) (40,41), whereas URE could be further subdivided into three overlapping sites, ␣ activator element, TSE, and URE1 (42). It was also found that the ␣ activator element binds to hGATA-2 and hGATA-3. TSE/URE1 forms an overlapping element that may bind two functionally interchangeable proteins, TSEB and UREB (42,43).
The ADA gene promoter is active in several cells, but a 770-bp enhancer located Ϫ5.4 kb upstream of the murine ADA gene conferred its placenta-specific expression. Within this enhancer, two motifs similar to TSE were found in a strong placenta-specific footprint (FP1) (33). Two GATA motifs, two CREs, and five basic helix-loop-helix sequence motif (CANNTG) were predicted to be involved in enhancer activity (33).
For the leptin gene, although the same promoter is used for adipose and placental transcription, an upstream enhancer (Ϫ1951/Ϫ1546 nt) functions in JEG-3 and JAR choriocarcinoma cells but not in adipocytes or HeLa cells (39). Three elements were suggested to participate in DNA-protein interaction within the enhancer by DNase I footprinting, including two motifs PLE1 (Ϫ1948/Ϫ1913) and PLE3 (Ϫ1909/Ϫ1874), which appeared to be placenta-specific (39).
In the second group of placenta-specific gene transcripts, aromatase expression is under placenta-specific control for its placental transcripts, albeit mature aromatase protein is present in multiple tissues. DNA sequences (Ϫ301/Ϫ115 nt) upstream to placenta-specific exon I increased reporter expression 20-fold in its natural orientation, and two important elements in this region contained a TSE consensus sequence and are able to bind to TSEB (47).
Although LIFR placenta-specific transcription appears characteristic of the same group as aromatase, LIFR gene enhancer utilizes a different set of DNA elements and binding proteins for its placenta-specific function. An Oct-1 binding site, a GATA-like element, and a novel PSE were found to be critical. Nevertheless, these elements share some resemblance to other placenta-specific elements. Although GATA-2, -3, and -4 antibody produced supershifts, it is clear that these proteins do not account for all the specific binding complexes containing GATA oligo; on the other hand, the binding protein for he GATA-like element was competed by the known GATA oligo, suggesting this novel GATA family member also has the potential to bind with previously detected GATA motifs in other placenta-specific enhancers. For the Oct-1 binding site, an interesting report has shown that TEF-1 binds to an Oct binding site (CAAAGCAT) in the SV40 enhancer (30) and also an Oct element overlaps with one of the TEF-1 binding sites in the HPV-16 enhancer (30). In our Oct-1 supershift assay, the specific binding complex was only partially supershifted, implying that the remaining binding complex might be derived from the Oct-1 binding site with another transcription factor such as TEF-1 (48) or TEF-5 (31). The core sequence of the novel placenta-specific element is CATGGCCTGAACTAGTTTTT. It does not share homology to TSE, CSEF binding site, PLE1, or PLE3, and thus this might be a new mechanism for regulation of placenta-specific gene expression.
One intriguing observation in studying placenta-specific transcription of these genes is that no hormone response element has been shown to be involved in their placenta-specific transcription, given the fact that placenta is the major expituitary organ that can produce hormones and also an organ under the influence of several hormones during pregnancy. For the human LIFR placenta-specific enhancer and promoter, this also seems to be the case; although some glucocorticoid response element, progesterone response element, retinoic acid response element, and estrogen response element were predicted in the data base search, neither dexamethasone, retinoic acid, progestrerone, nor 17-␤-estradiol treatment of transfected JEG-3 cells showed significant effects on LIFR promoter activity (data not shown). Nor did the Northern analysis of JEG-3 cells treated with these steroids reveal any effect on LIFR mRNA levels (data not shown). These observations appear to be in constrast to the observation that human LIFR mRNA levels increase during the first trimester and term placenta. A possible explanation is that JEG-3 cells, as a differentiated cell line secreting high level of human chorionic gonadotropin hormone, may not represent a suitable placental cell type which is responsive to hormone treatment.
Since TSE was found to be involved in several placentaspecific enhancers such as ␣ subunit, hCS-B, aromatase, and ADA, it was felt that TSE might be a master switch in placenta cell differentation that simultaneously regulates a range of placenta-specific gene expression (36). However, LIFR is not under the same control because no TSE motif was detected. Leptin gene transcription also does not appear to be under TSE control. These observations suggest that the placenta may not possess a universal master switch for all placenta-specific gene expression, although TSE is involved in placenta-specific expression of several genes.
Several knockout mice targeting placenta-specific genes and transcription factors involved in their regulations have been produced. Glycoprotein hormone ␣ subunit homozygous knockout mice did not exhibit placental abnormality, although they are hypogonadal and exhibit profound hypothyroidism and dwarfism (49). Mice lacking a functional leptin gene not only became massively obese (50), but also are infertile (51), but no placental abnormality was reported. ADA-deficient fetuses lacking ADA in their adjoining placenta die during late fetal development (52), while genetically restoring ADA to placentas of ADA-deficient fetuses rescued them from perinatal lethality (53). While GATA-3 knockout mice have severe abnormalities in the nervous system and fetal liver hematopoiesis (54), and GATA-2 knockout mice are defective in early hematopoietic cell proliferation and mast cell formation (55), both GATA-2 and GATA-3 knockout mice did not show placental abnormalities, thus supporting the presence of other GATA family members functioning in the placenta. TEF-1 null mice show defects only in the heart, but not in other tissues, including the placenta (56). Notably, these knockout results, except those of the ADA knockout, showing no murine placental abnormalities, are in marked contrast to the LIFR knockout mice which demonstrated severe placental dysfunction (13).
We previously hypothesized that in placenta and/or during pregnancy, placenta-specific transcription factor(s) are active and interact with the LIFR gene placenta-specific enhancer and promoter, leading to enhanced transcription of LIFR mRNA restricted to the placenta. Identification of this novel PSE in LIFR gene enhancer strongly supports this hypothesis, and the cDNA cloning of the 160-kDa PSEB will further elucidate the specific placental transcriptional control of the LIFR.