Silencing of the gene for the alpha-subunit of human chorionic gonadotropin by the embryonic transcription factor Oct-3/4.

The transcription factor Oct-3/4 may be important in maintaining embryonic cells in an undifferentiated state. It is probably down-regulated at about the time that human chorionic gonadotropin (hCG) is first expressed in embryonic trophectoderm. Here we report that Oct-3/4 strongly inhibits the hCGβ subunit (hCGβ) promoter in JAr choriocarcinoma cells. Oct-3/4 reduced chloramphenicol acetyltransferase (CAT) reporter expression from the −305hCGβ promoter by about 90% in transient co-transfection assays, but had no effect on expression from the −249hCGβ promoter. The −305/−249 hCGβ fragment specifically bound synthetic Oct-3/4 protein as measured in electrophoretic mobility shift assays, and the Oct-3/4-binding site was localized around −270 by methylation interference footprinting. Site-directed mutagenesis of this binding site abolished Oct-3/4 repression. When stably transfected into JAr cells, Oct-3/4 reduced the amounts of both endogenous hCGβ messenger RNA and hCG protein to less than 10% of controls. We suggest that silencing of Oct-3/4 in trophectoderm is a prerequisite for hCG up-regulation in early human embryos at the time of maternal recognition of pregnancy.

The transcription factor Oct-3/4 may be important in maintaining embryonic cells in an undifferentiated state. It is probably down-regulated at about the time that human chorionic gonadotropin (hCG) is first expressed in embryonic trophectoderm. Here we report that Oct-3/4 strongly inhibits the hCG␤ subunit (hCG␤) promoter in JAr choriocarcinoma cells. Oct-3/4 reduced chloramphenicol acetyltransferase (CAT) reporter expression from the ؊305hCG␤ promoter by about 90% in transient co-transfection assays, but had no effect on expression from the ؊249hCG␤ promoter. The ؊305/ ؊249 hCG␤ fragment specifically bound synthetic Oct-3/4 protein as measured in electrophoretic mobility shift assays, and the Oct-3/4-binding site was localized around ؊270 by methylation interference footprinting.

Site-directed mutagenesis of this binding site abolished Oct-3/4 repression. When stably transfected into JAr cells, Oct-3/4 reduced the amounts of both endogenous hCG␤ messenger RNA and hCG protein to less than 10% of controls. We suggest that silencing of Oct-3/4 in trophectoderm is a prerequisite for hCG up-regulation in early human embryos at the time of maternal recognition of pregnancy.
Human chorionic gonadotropin (hCG) 1 is crucial for preventing regression of the corpus luteum during early pregnancy. It is first secreted by trophectoderm, the precursor cell layer of the placenta, as the blastocyst forms and begins to implant (1)(2)(3). The timing and quantity of hCG release are considered to be key factors in determining whether a human pregnancy succeeds or fails (4 -6).
hCG is a heterodimer containing an ␣ subunit (hCG␣), common to all the glycoprotein hormones, and a distinct ␤ subunit (hCG␤) responsible for the biological specificity of the hormone. Whereas there is only a single gene for the ␣ subunit, there are six hCG␤ subunit genes or pseudogenes (7,8). Of the latter, hCG␤5 is the one expressed predominantly in placenta and choriocarcinoma cells (7,9).
Most studies on the control of expression of the hCG genes have concentrated on their transactivation. The upstream promoter region of the hCG␣ gene includes two tandem repeats of a cyclic AMP response element (CRE), a complex upstream regulatory element (URE), the ␣-activator element, the junctional regulatory element, and the CCAAT region (10 -16). Although much less studied, the hCG␤5 gene also contains multiple regulatory regions that contribute toward expression in choriocarcinoma cells (17)(18)(19)(20). These elements include several within the Ϫ310 to Ϫ200 region and further ones more than 1 kilobase upstream of the transcription start site. Although expression of hCG has been studied extensively, little is known about what controls its onset at a time when the corpus luteum must be rescued if the pregnancy is to proceed.
The transcription factor Oct-3/4, characterized by its conserved POU DNA-binding domain, is a strong candidate for a regulator of early embryogenesis (21)(22)(23)(24)(25). It is expressed in totipotent/pluripotent embryonic cells and germ cells and in undifferentiated embryonic stem cells and embryonal carcinoma cells, but is rapidly down-regulated when these cells differentiate. Fusion of embryonal carcinoma cells and fibroblasts results in loss of Oct-3/4 expression and neuronal differentiation of the hybrid cells, while introduction of Oct-3/4 transactivating function back into such hybrid cells causes partial dedifferentiation (26). The expression pattern of Oct-3/4 and its correlative relationship with cell pluripotency suggest that Oct-3/4 may be important in maintaining cells in an undifferentiated state and that silencing of its expression could contribute to the process of differentiation. No natural target genes for Oct-3/4 have been unequivocally identified, and it remains unclear whether Oct-3/4 is an activator or repressor of gene expression (27)(28)(29)(30).
During the study of the transcriptional regulation of a trophoblast interferon gene (IFNT) (31) in choriocarcinoma cells, an hCG␣-CAT construct was included as an internal control in transient transfection experiments and, surprisingly, was found to be completely silenced by Oct-3/4 co-transfection. 2 Both the Oct-3/4 messenger RNA (mRNA) and the protein have been detected in early stage trophectoderm but not in trophoblast cells after the blastocyst has hatched from the zona pellucida in mouse (22,23,25). The down-regulation of Oct-3/4 in the human embryo, therefore, probably coincides with the first appearance of hCG in trophectoderm. Here we demonstrate that Oct-3/4 is an inhibitor of hCG␤ expression in JAr human choriocarcinoma cells and suggest that the loss of Oct-3/4 expression in developing trophectoderm may be a prerequisite for the onset of hCG expression. * This work was supported by Grants R37 HD21896 and HD29483 from the National Institutes of Health (to R. M. R.) and a fellowship from the Molecular Biology Program of the University of Missouri-Columbia (to L. L.). This paper is a contribution from the Missouri Agricultural Experiment Station, Journal Series Number 12,431. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-CAT expression plasmids pϪ305hCG␤-CAT, pϪ279hCG␤-CAT, and pϪ249hCG␤-CAT were provided by Dr. Pamela L. Mellon (19). Oligonucleotide primers Ϫ305hCG␤ and CATr (Table I) were used in conjunction with pϪ305hCG␤-CAT to produce the mutant Ϫ305hCG␤-CAT fragment by PCR. The PCR product was digested with XbaI and XhoI. This fragment (Ϫ305/ϩ66 hCG␤) was used to replace its wild type hCG␤ counterpart in pϪ305hCG␤-CAT to form the pϪ305hCG␤-CAT plasmid.
The Oct-3/4 coding region was synthesized from the pCMV-Oct4 expression plasmid (23) by PCR with the primers Oct45Ј and Oct43Ј (Table I). The PCR product was digested with SalI and BglII, blunted, and cloned into the KpnI site of pcDNA3 (Invitrogen) via blunt-end ligation. The hCG␤-CAT and pcDNA3-Oct4 constructs were confirmed by DNA sequencing.
A 417-base pair hCG␤ cDNA fragment (Ϫ28 to ϩ389 base pairs, relative to the first nucleotide of the translated sequences) was synthesized from the hCG␤ cDNA clone, pCG␤474 (32), by PCR with the oligonucleotide primers hCG␤5Ј and hCG␤3Ј (Table I). The hCG␤ cDNA fragment was then cloned into the pCR TM II plasmid (Invitrogen Corp.). The orientation of the hCG␤ fragment relative to the T7 promoter in the pCRII-hCG␤ plasmid was determined by DNA sequencing.
Transient Transfection, CAT Assay, and hGH Radioimmunoassay-JAr cells were cultured to about 40% confluence in 6-well tissue culture plates in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Either pcDNA3-Oct-3/4 (2 g) or pcDNA3 (2 g) were co-transfected with the phCG␤-CAT construct (2 g) and pTKGH (0.2 g) into cells by the calcium phosphate precipitation method (34). The amounts of hGH in the culture media after 40 -48 h were measured by a radioimmunoassay (Nichols Institute Diagnostics). The cell lysates were measured for CAT activity by phase extraction (twice) and liquid scintillation counting of the CAT enzyme reaction products (34). CAT activity was normalized by hGH expression. Analysis of variance and a standard Student's t test were used for statistical analyses (35).
Electrophoretic Mobility Shift Assay (EMSA)-Single-stranded oligonucleotides Ϫ305hCG␤f and Ϫ249hCG␤r (Table I) were labeled at their 5Ј ends by using T4 polynucleotide kinase and [␥-32 P]dATP (3000 Ci/ mmol, DuPont NEN). They were annealed to each other and blunted with Klenow fragment of Escherichia coli DNA polymerase I (34). The resulting double-stranded Ϫ305/Ϫ249hCG␤ probe (20,000 cpm, 0.2 ng) and the Oct-3/4 protein (1 l of the synthesis mixture) were used in electrophoretic mobility shift assays (28). The double-stranded OCT oligonucleotide that had been produced by annealing oligonucleotide OCTf and OCTr (Table I) was included (10 ng) as a competitor in one reaction. The synthetic Oct-3/4 was also used with the 32 P-labeled OCT oligonucleotide in an electrophoretic mobility shift assay performed similarly to that described above. The double-stranded ␣CRE oligonucleotide that had been produced by annealing oligonucleotides ␣CREf and ␣CREr (Table I) was included as a competitor in one reaction.
Methylation Interference Analysis-Double-stranded Ϫ305/ Ϫ249hCG␤ probes were made as described above, except that only one strand was 32 P-end-labeled. After partial methylation at guanines with dimethyl sulfate (36), the probe was used with Oct-3/4 in the preparative electrophoretic mobility shift assay. The free probe and the retarded probe-Oct-3/4 complex in EMSA were identified by autoradiography and excised from the gel. After being purified by the crush and soak method (36), the DNA fragments were cleaved with piperidine, analyzed on a DNA sequencing gel, and detected by autoradiography (36).
Stable Transfection of Oct-3/4 and Northern Blot Analysis-JAr cells were transfected with pcDNA3-Oct-3/4 or pcDNA3 by the calcium phosphate precipitation method, and stably transfected cells were selected by G418 (250 mg/liter, Sigma) and clonally propagated. Poly(A) RNA was partially purified from stable clones and from normal JAr cells with the QuickPrep Micro mRNA purification kit (Pharmacia Biotech Inc.). Expression of the Oct-3/4 messenger RNA in these cells was tested by Northern blot analysis (34) with a 32 P-labeled mouse Oct-3/4 cDNA probe.
Ribonuclease Protection Assay-Ribonuclease protection assays of the total RNA (10 g) isolated from several stable clones were performed using the Hybspeed TM RPA kit from Ambion Inc. (Austin, TX). Antisense hCG␤ RNA probe was synthesized from pCRII-hCG␤ using T 7 bacteriophage RNA polymerase in the presence of [␣-32 P]CTP (800 Ci/mmol, DuPont NEN). Antisense human ␤-actin RNA probe was synthesized to an 80-fold lower specific activity (by diluting the original 800 Ci/mmol [␣-32 P]CTP with a concentrated solution of unlabeled CTP) from the pTRI-␤-actin-125-human antisense control template (Ambion Inc.). The hCG␤ probe (50,000 cpm) and the human ␤-actin probe (10,000 cpm) were used for each protection assay. After ribonuclease digestion, the protected duplexes were resolved in 5% Long Ranger TM gels (AT Biochem, Malvern, PA), and relative intensities of hCG␤ and ␤-actin bands were measured by densitometry. The amount of hCG␤ mRNA was then normalized relative to the amount of ␤-actin mRNA in each lane.
Radioimmunoassay of hCG-The amount of hCG secreted by cultured JAr cells (ϳ2 ϫ 10 6 cells/culture) over a 24-h period was measured by a radioimmunoassay (Nichols Institute) that employed monoclonal antibodies specific to the hCG␤ subunit. Intact hCG (Nichols Institute) was used to standardize the assay. hCG concentrations were normalized according to the cell number.

Inhibition of hCG␤-CAT Expression by Oct-3/4 -To determine whether
Oct-3/4 influences hCG␤ gene transcription, the promoter of hCG␤5 gene was fused to the CAT reporter, and the resulting plasmid (Ϫ305hCG␤-CAT) co-transfected into JAr cells with the Oct-3/4 expression plasmid, pcDNA3-Oct-3/4, in which the coding region of a murine Oct-3/4 cDNA had been placed under the control of the cytomegalovirus promoter (Fig.  1). CAT expression was markedly reduced compared to values obtained when the control plasmid vector (pcDNA3) was the co-transfection partner. By contrast, when Ϫ305hCG␤-CAT was co-transfected with an Oct-2 expression plasmid, CAT expression was not affected (94% Ϯ 17% of the control value). CATr

5Ј-GTGCTTCAGGTGATTTAACTGATTATTGAAT-3Ј
Ϫ305hCG␤f Human growth hormone expression driven either by a viral thymidine kinase promoter or by a mouse metallothionein-I promoter (pTKGH and pXGH5, respectively) was not affected by pcDNA3-Oct-3/4 co-transfection (data not shown). In all subsequent experiments, pTKGH was used as the internal control to normalize transfection efficiencies. As expected (19), 8-Br-cAMP increased CAT expression from the Ϫ305hCG␤ promoter approximately 2-fold. It did not, however, affect repression by Oct-3/4 (12.8 Ϯ 2.4% expression of control in presence of 8-Br-cAMP, 11.9 Ϯ 2.0% in its absence).
hCG␤ promoters with 5Ј truncations were employed to define the region that responded to Oct-3/4. Expression of CAT from the shortest construct (Ϫ249hCG␤-CAT) was unchanged in the presence of pcDNA3-Oct-3/4, while expression from Ϫ279hCG␤-CAT was reduced to about 40% of control values (Fig. 1). Clearly, sequences within the Ϫ305/Ϫ249 promoter region were responsive to Oct-3/4 and probably included position Ϫ279.
Binding of Oct-3/4 to the hCG␤ Promoter in Vitro-Oct-3/4 protein ( Fig. 2A), produced by coupled in vitro transcription and translation in a reticulocyte lysate, was able to interact with the 32 P-labeled Ϫ305/Ϫ249hCG␤ promoter fragment in an electrophoretic mobility shift assay (Fig. 2B, lane 2). This complex was abolished when excess unlabeled Ϫ305/Ϫ249hCG␤ fragment was added to the reaction mixture (data not shown) or when an oligonucleotide competitor (OCT) that contained the consensus octamer motif was present (Fig. 2B, lane 3). The formation of the complex was unaffected by an excess of unrelated oligonucleotide (not shown). In another set of electrophoretic mobility shift assays, radioactive Oct-3/4, which had been synthesized in the presence of L-[ 35 S]methionine, was used in combination with various non-radiolabeled DNA fragments (Fig. 2C). A single band of 35 S-Oct-3/4 was observed (lane 1) when it was incubated with poly(dI-dC) before electrophoresis. No additional labeled bands were observed when either the Ϫ249/ϩ66hCG␤ or Ϫ60/ϩ44hCG␣ fragments were included in the reaction mixture (lanes 4 and 5). In contrast, a faster moving 35 S-Oct-3/4 complex appeared when either OCT or the Ϫ305/Ϫ249hCG␤ fragment were used (lanes 2 and 3). These observations confirm that Oct-3/4 could bind the Ϫ305/ Ϫ249hCG␤ promoter region directly. As expected, Oct-3/4 also bound to a 32 P-labeled OCT oligonucleotide (Fig. 2D), and this binding was reduced in the presence of unlabeled Ϫ305/ Ϫ249hCG␤ (lane 4), but not by an unrelated oligonucleotide (lane 3). Clearly, an Oct-3/4 binding site was present in the Ϫ305/Ϫ249 region of the promoter.
Methylation interference analysis was employed to define the binding site for Oct-3/4 on Ϫ305/Ϫ249hCG␤ more precisely (Fig. 3A). Probe that had been methylated in the antisense strand at Ϫ276 or Ϫ269 clearly bound Oct-3/4 less well than probe that had not been methylated at those positions. The Ϫ275/Ϫ268 region (Fig. 3B) was identical at seven nucleotides out of eight of the optimal POUS motif, to which the POUspecific domain of Oct-1 would bind (39). Methylation of the sense strand of Ϫ305/Ϫ249hCG␤ did not interfere with Oct-3/4 binding (data not shown).

Mutation of the Oct-3/4 Binding Site Negates Repression of the hCG␤ Gene Promoter by Oct-3/4 -When
the sequence 5Ј-AATC (Ϫ272 to Ϫ269) within the Oct-3/4 binding region of the hCG␤ promoter was mutated to 5Ј-ccag, CAT expression from this mutant promoter was not significantly different from that observed from the nonmutated Ϫ305hCG␤ promoter in the absence of Oct-3/4 co-transfection (Fig. 4). However, in contrast to the strong Oct-3/4 inhibition of Ϫ305hCG␤-CAT (11.9% of the control), CAT expression from the Ϫ305hCG␤ promoter was only slightly decreased (73.4% Ϯ 15 of the control and statistically nonsignificant) when pcDNA3-Oct-3/4 was cotransfected. Therefore, the Oct-3/4 binding site identified in vitro was critical for the ability of this transcription factor to repress the activity of the hCG␤ promoter in JAr cells.

Inhibition of Endogenous hCG Production in JAr Cells by Oct-3/4 Stable Transfection-To study the effect of Oct-3/4 on endogenous hCG␤ gene expression, JAr cells were transfected
with pcDNA3-Oct-3/4. Stably transfected cells were selected by antibiotic G418 and clonally propagated. Such cells expressed Oct-3/4 mRNA, whereas control JAr cells did not (Fig. 5A). They did not differ morphologically from either wild type cells or from cells that had been stably transfected with the pcDNA3 vector lacking the Oct-3/4 gene. No differences were detected in the rates of protein synthesis (as assessed by incorporation of label from [ 35 S]methionine over 24 h of culture), and analysis of the radiolabeled proteins in the medium by one-dimensional SDS-PAGE could not distinguish the transfected and control cells (data not shown). It should be emphasized that the subunits of hCG are produced in such small amounts that they could not be readily detectable by this procedure. The conclusion drawn from these experiments was that stable transfection with Oct-3/4 had no major effect on the phenotype of JAr cells.
RNA was isolated from both stable Oct-3/4 clones and stable control clones and subjected to a ribonuclease protection assay in the presence of an antisense hCG␤ RNA probe expected to hybridize to the first 389-bp part of the coding region of all hCG␤ transcripts (7,32,37). As shown in Fig. 5B, hCG␤ mRNA was barely detectable in either of the stable Oct-3/4 clones tested (S1 and S4) but was present in both control clones (C1 and C2). The content of ␤-actin mRNA was comparable among all clones, whether they expressed Oct-3/4 or not. When quantitated by densitometry and normalized to ␤-actin mRNA, the hCG␤ mRNA content of the clones expressing Oct-3/4 was about 6% of that in the controls (Fig. 5C).
Production of hCG protein, as determined by a radioimmunoassay specific for the hCG␤ subunit, was markedly reduced in clones expressing Oct-3/4 (Fig. 6). For clones S1 and S4, the amount of hCG␤ was 8.3% and 3.2%, respectively, of that produced by two control lines C1 and C2.

FIG. 1. Inhibition of hCG␤-CAT by Oct-3/4 co-transfection in JAr cells.
The hCG␤-CAT constructs (2 g) was cotransfected into JAr cells with 2 g of either the pcDNA3-Oct-3/4 plasmid (ϩOct-3/4) or the pcDNA3 vector alone (ϪOct-3/4) CAT expression is shown as a percentage of that from the Ϫ305hCG␤-CAT construct in the absence of Oct-3/4 co-transfection. The results were the means (ϮS.E.) of at least four independent experiments. Values marked with different letters were statistically different (p Ͻ 0. 01). DISCUSSION Here we have demonstrated the transcription factor Oct-3/4 to be a potent repressor of hCG␤ gene expression in JAr choriocarcinoma cells. Stable expression of Oct-3/4 reduced the amounts of both endogenous hCG␤ messenger RNA and hCG␤ protein by over 90% in these cells. Oct-3/4 also strongly inhibited reporter expression from the hCG␤5 gene promoter in transient transfection assays. Furthermore, an Oct-3/4 binding site was present in the hCG␤ promoter and was necessary for Oct-3/4 inhibition.
Oct-3/4 can specifically bind to the hCG␤5 promoter in vitro.
When the Oct-3/4 binding site in the hCG␤ promoter was mutated or deleted, Oct-3/4 repression was lost. This Oct-3/4 binding site (Ϫ275/Ϫ268; AATAATCA) differed markedly from the previously described octamer consensus sequence (ATG-CAAAT) (21,38). Despite its unconventional sequence, this site is identical at seven out of eight nucleotides to the motif described as optimal for binding of the Oct-1 POU-specific domain (POU S ) (39) and is placed just one base pair upstream from a stretch of six A/T nucleotides (Fig. 3B) that might be capable of interacting with the POU homeodomain (POU HD ) of Oct-3/4. It has been found that the arrangement of the binding sites for the POU S and POU HD critically influences the orientation of the two DNA binding domains to each other and to the promoter (40). The Ϫ277/Ϫ268 region (TCAATAATCA) of the hCG␤5 promoter is also identical at seven out of ten nucleotides to a weak Oct-3/4 binding sequence (TTAAAATTCA) described by Okamoto et al. (21). This sequence and the consensus ATGCAAAT motif have each been found in two enhancerpromoter units of the mouse genome that are active in undifferentiated P19 embryonal carcinoma cells but inactive in differentiated P19 cells (41). As indicated by competition experiments, Oct-3/4 may possess lower affinity for the recognition sequence in the hCG␤ promoter than it does for the consensus motif (Fig. 2), but it is not uncommon for a low affinity binding site to confer regulatory activity as effectively as one of high affinity (42). In addition, the POU transcription factor Pit1 binds well to the octamer consensus motif (ATGCAAAT), but fails to transactivate promoters containing that motif (42). Such a consensus sequence is not present in any of the known genes activated by Pit1 (43).
It seems unlikely that Oct-3/4 merely competes for binding with some transcriptional activator whose response element overlaps the Oct-3/4 binding site in the hCG␤ gene. If such were the case, expression from Ϫ305hCG␤-CAT after Oct-3/4 co-transfection would be anticipated to be at least as high as from Ϫ249hCG␤-CAT, a construct from which the response element for the putative activator had been deleted. Instead, expression from the Ϫ305 construct was much lower than from Ϫ249hCG␤-CAT in the presence of Oct-3/4 but about the same in its absence (Fig. 1). Oct-3/4, therefore, seems to have an intrinsic ability to repress the hCG␤ promoter. It is unclear whether once bound it directly inhibits the general transcriptional machinery or whether it recruits some other inhibitory factor. Repression of herpes simplex immediate-early promoter by neuronal forms of Oct-2 probably occurs through such a secondary recruitment process (44), and there are several other examples where POU domain proteins function cooperatively with other proteins to regulate transcription (45,46).
All functional hCG␤ genes possess TATA-less promoters (18), and it could be for this reason that the hCG␤5 gene tested here was repressed by Oct-3/4 while another octamer-containing promoter, the one for thymidine kinase, which possesses a conventional TATA box (47), was not affected by Oct-3/4 cotransfection. Some special transcription factor required specifically for proper functioning of such TATA-less promoters (48,49) could be the target of Oct-3/4 inhibition.
Oct-3/4 dramatically reduced overall expression from the endogenous hCG␤ genes of JAr cells. The mRNA for hCG␤5 accounts for about 64% of the total hCG␤ mRNA in firsttrimester placenta, whereas hCG␤3 and hCG␤8 each accounts for about 18% (7). Expression levels of the hCG␤ genes in choriocarcinoma cells are probably similar to those in placenta (7,9). Therefore, ectopic expression of Oct-3/4 seems likely to inhibit expression not only from the hCG␤5 gene but from other hCG␤ genes as well.
The association of Oct-3/4 expression with the totipotent/ pluripotent state of cells has been a subject of considerable speculation, but its target genes have not been unequivocally identified, and whether it is an activator or repressor of gene FIG. 3. Methylation interference footprint of the hCG␤ promoter in presence of Oct-3/4. A, the Ϫ305/Ϫ249hCG␤ fragment was radiolabeled at the 5Ј end of its antisense (bottom) strand and partially methylated at guanines. The free (F) probes and the Oct-3/4-bound (B) probes were cleaved at the methylated sites and resolved on an 8% sequencing gel. The sites where methylation apparently interfered with Oct-3/4 binding are indicated with asterisks. B, the hCG␤ promoter sequence around the methylation interference sites. The boxed region represents a motif with close similarity (7 nucleotides out of 8) for the optimal binding sequence of the POU-specific domain (POU S ) established for Oct-1 (39). expression remains unclear (27)(28)(29)(30). For example, although Oct-3/4 has recently been shown capable of activating transcription from an enhancer of the fibroblast growth factor 4 gene, it requires the cooperation of the Sox2 gene product (30). It remains unclear whether the fibroblast growth factor 4 gene is a target for Oct-3/4 in vivo, since the expression of fibroblast growth factor 4 and Oct-3/4 only coincide in the early stages of mouse embryo development prior to gastrulation (22,23,50). In this report, we demonstrate that the hCG␤ genes that are first expressed in the human blastocyst are repressed by Oct-3/4. Thus, down-regulation of Oct-3/4 may permit the transcriptional activation of previously silent genes and the emergence of a more differentiated phenotype.
In summary, the Oct-3/4 has been shown to repress expression of the hCG␤ genes in choriocarcinoma cells. This silencing likely results from direct interaction of Oct-3/4 with the hCG␤ gene promoter. These results, together with the probable reciprocal expression pattern of Oct-3/4 and hCG in trophectoderm, strongly suggest that Oct-3/4 inhibits transcription from the hCG␤ genes in early stages of trophectoderm formation. As Oct-3/4 expression declines, so the hCG␤ genes are probably relieved from repression.  1 (ϳ0. 8 g) and lane 2 (ϳ20 ng) contain partially purified poly(A) RNA from the stable clone S4 that had been transfected with pcDNA3-Oct-3/4. Lane 3 (ϳ2 g) contains partially purified poly(A) RNA from normal JAr cells that had not been transfected with any plasmid. The lower panel is a portion of the ethidium bromide-stained gel showing the 28 S rRNA that was present in the preparation. It is only evident in the two heavily loaded lanes. B, ribonuclease protection assays were carried out as described under "Experimental Procedures" to determine relative amounts of hCG␤ mRNA in stable JAr clones. Clones S1 and S4 had been stably transfected with pcDNA3-Oct-3/4, C1 and C2 with pcDNA3. The protected hCG␤ fragment and the internal ␤-actin control are indicated by arrows. The signals were then quantitated by densitometry. The exposure time used to obtain appropriate optical densities in x-ray film was 30 min for ␤-actin, and the densitometric values were 0.51 (C1), 0.21 (S1), 0.73 (C2), and 0.46 (S4), respectively. The exposure times for hCG␤ were 2 h in C1 and C2 and 5 h in S1 and S2. The densitometric values were 0.82 (C1), 0.04 (S1), 0.85 (C2), and 0.08 (S4), respectively. All hCG␤ values were then normalized by comparison with ␤-actin (C).
FIG. 6. Reduced endogenous hCG production in JAr cells stably transfected with pcDNA3-Oct-3/4. Stable clones C1, C2, S1, and S4 are the same as in Fig. 5. Amounts of hCG secreted by the stable clones were measured by a radioimmunoassay that utilized monoclonal antibodies specific to the hCG␤ subunit. The results are the means (ϮS.E.) of four independent experiments. Values marked with different letters are significantly different (p Ͻ 0.001).