INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Tau interferons (IFN-) are structurally related to IFN-, IFN-, and IFN- (with approximately 25, 50, and 75% primary sequence identities, respectively) (38, 39). Although these IFNs exhibit typical properties, including the ability to induce an antiviral state in cells expressing the appropriate IFN receptor, their function is in reproduction and not in disease prevention (2, 11, 37, 38). The preattachment conceptuses of ruminant species, such as cattle, sheep, and deer, produce IFN- in order to prevent the regression of the maternal corpus luteum, an event that would normally occur at the end of the estrous cycle if the animal were not pregnant (2, 11, 37). Members of the multigene IFN- family are expressed weakly in the trophectoderm of cattle beginning when this epithelium differentiates at blastocyst formation (11, 14, 16, 37). The level of production of IFN- per cell increases markedly as the blastocyst enlarges and begins to elongate. Production is down-regulated when the trophectoderm forms contacts with the uterine wall and as the process of placentation is initiated. The expression of IFN- is confined to a single cell layer, the trophectoderm, for 1 to 2 weeks during the critical period when the corpus luteum of pregnancy must be rescued and progesterone production must be maintained. Neither the inner cell mass nor the embryo that is derived from it expresses this IFN (2, 11, 37).

An analysis of the nucleotide sequences of the IFN- genes (IFNT genes) has indicated that the progenitor gene arose by duplication of a closely related gene, one encoding IFN-, approximately 35 million years ago (38, 39). This event appeared to involve a disruption in the upstream promoter region of the new gene relative to its parent and presumably led to the observed loss of viral inducibility (20) and a gain in the ability to be expressed constitutively in trophoblasts. A major divergence in sequence identity between the promoter sequences of IFNT genes and IFN- genes (IFNW genes) begins just proximal to a presumptive Ets binding site (79 to 70) in the IFNT genes which is lacking in the IFNW genes (38, 39). This region of the IFNT promoter has been implicated in controlling high-level IFN- expression in the trophectoderm (10, 20).

In experiments that used the yeast one-hybrid screen, Ets-2 was confirmed as the transcription factor that most likely bound to the Ets binding site in the IFNT proximal promoter region (13). Ets-2 strongly transactivates IFNT-luciferase (luc) gene reporters, and IFNT genes that lack the Ets site are poorly expressed. The Ets-2 protein is also expressed in the trophectoderm of ovine conceptuses when this tissue expresses IFNT genes. A role for Ets-2 in controlling the differentiation of trophoblast derivatives has been proposed because deletion of the murine ets-2 gene leads to embryonic death by day 8.5 as the result of defective placental development (56). Further, a number of genes in addition to IFNT genes, which are expressed in trophoblasts but not in the inner cell mass (i.c.m.), are regulated by the Ets factor (18, 30, 36, 50, 52).

It was recently demonstrated that two human genes expressed in trophoblasts, the  and  subunits of human chorionic gonadotropin (hCG), are silenced in human choriocarcinoma cells by the transcription factor Oct-4 (22, 23). Oct-4 (48), also called Oct-3 (29), belongs to the family of POU transcription factors that contain a bipartite DNA binding domain (POU specific and POU homeodomain) (43). In mouse embryos, it is expressed predominantly in blastomeres, pluripotent early embryo cells, and the germ cell lineage (34, 42, 47). Its mRNA and protein are expressed throughout the cleavage stages of embryo development but not in the trophectoderm of blastocysts (32). The situation is slightly different in bovine embryos (19, 54). There, Oct-4 is expressed strongly in the i.c.m. However, the expression of Oct-4 can also be detected in the trophectoderm until day 10, approximately 3 days after blastocyst formation but before the massive up-regulation of IFN- that accompanies trophoblast elongation (14, 37). Oct-4 has been predicted to play a defining role in maintaining cells in the pluripotent state and in preventing differentiation of the i.c.m. of mouse embryos into the trophectoderm (28). Day 4.5 mouse embryos with their Oct-4 gene deleted lack a true i.c.m. and consist entirely of trophectoderm-like cells (26). It is not yet clear how Oct-4 prevents differentiation, but it likely activates certain key genes (3, 31, 41, 58) while repressing others (22, 23). Down-regulated Oct-4 expression presumably lifts the constraints operating on the differentiation process.

These data together suggest that Ets-2 and Oct-4 might have counteracting roles in the functional differentiation of trophoblasts. While the former may have a promoting action, the role of Oct-4 may be in restraint. Here we have examined whether the two transcription factors interact in controlling IFNT gene expression, which is a biochemical marker for the functional differentiation of trophoblasts in bovine embryos.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Reporter gene constructs and expression plasmids. A boIFNT1 promoter fragment (bp 457 to +66) (10) was subcloned into the KpnI site of the luciferase reporter plasmid, pGL2-Basic (Promega, Madison, Wis.) (457luc). Site-directed mutagenesis of the Ets-2 binding site at 79 on the 457luc promoter was achieved with primers mets1 and mets2 (Table 1) and standard PCR procedures (8). The mutated promoter was subcloned into the SacI site of pGL2 Basic Vector. A series of deletion mutants of boIFNT1 promoter fragments was generated from the 457luc construct by PCR with promoter primers, 353se, 228se, and 91se, and vector primer, GL2 (Table 1). The PCR fragments were subcloned from the pBluescript SK() (Stratagene, La Jolla, Calif.) vector by SacI and XbaI digestion and inserted into the luciferase reporter plasmid. The 126luc construct (containing the promoter fragment from bp 126 to bp +50) has been described previously (13). 126luc was mutated by introduction of an StuI site at 50. The 49luc construct was generated from mutated 126luc by NotI and StuI digestion, blunt ending, and self-ligation. A construct consisting of three tandemly repeated octamer motifs (3×Oct) was generated by ligation of annealed oligonucleotides, OCTf and OCTr (21) (Table 1), and was cloned into the SpeI site of pBluescript SK(). By using the BamHI and XbaI sites of the vector polylinker, the 3×Oct fragment was cloned into the same sites of pTKCAT (51), upstream of the thymidine kinase (tk) promoter (105 to +51). The 3×Oct-tk fragment was released by XbaI and XhoI digestion and was cloned into the luc reporter plasmid. The fidelity of all the constructs was verified by DNA sequencing.

Cell cultures and transfections. JAr cells (HTB-144; American Type Culture Collection) were maintained in RPMI 1640 supplemented with 10% fetal bovine serum, penicillin (50 U/ml), and streptomycin (50 µg/ml) and transfected by the calcium phosphate method as described previously (13). The boIFNT1 promoter-luc reporter construct (3.2 µg) was cotransfected with either pCMV-Oct4 (0.2 to 0.4 µg) or pCGNEts-2 (0.2 to 0.4 µg). All transfections included 50 ng of pRSVLTR-gal or 10 ng of pRL-CMV. Total amounts of transfected DNA were kept constant by adding corresponding empty vectors. luc reporter assays were conducted 36 h after transfection. Enzyme assays for analyses of transfection experiments were carried out as described previously (13), except when pRL-CMV was used. The activities of both firefly and Renilla luciferases were measured with a dual-luciferase reporter assay system (Promega). Firefly luciferase activity was normalized to -galactosidase or Renilla luciferase activity.

Western blot and immunoprecipitation analyses. JAr cells were scraped from 6-cm dishes and lysed in 0.2 ml of the buffer-reagent M-PER (Pierce, Rockford, Ill.)/plate. After centrifugation, 75 µg of cleared cell lysate was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene difluoride membranes (Immobilon-P; Millipore, Bedford, Mass.). The antisera were commercially prepared in rabbits. Anti-Oct-1, -2, and -4 (sc-231, sc-233, and sc-9081), diluted 1:1,000, and anti-Ets-2 (sc-351), diluted 1:4,000, were all products of Santa Cruz Biotechnology, Santa Cruz, Calif. The second antibody was alkaline phosphatase-conjugated anti-rabbit immunoglobulin G diluted 1:10,000, which was used in conjunction with a Western-Star chemiluminescence kit from Applied Biosystems, Bedford, Mass.

Electrophoretic mobility shift assay. A 1-kb Oct-4 cDNA fragment (48) amplified by PCR with primers Oct4se and Oct4as (Table 1) was subcloned into the EcoRI-BamHI site of vector pGBT9. The EcoRI-SalI cDNA fragment from construct pGBT9 was cloned into pGEX-4T-1 and used to transform Escherichia coli DH5. The fidelity of the construct was verified by DNA sequencing, and bacterial extracts were processed as described elsewhere (13). Annealing of oligonucleotides OCTf and OCTr (Table 1) produces a double-stranded canonical octamer site (23). Reaction mixtures included 2 µg of E. coli protein containing either GST-Oct-4 or GST plus 10 fmol of ³²P-labeled, double-stranded octamer DNA probe (~18,700 cpm). For competition binding assays, competitor DNA (250-fold molar excess, 2.5 pmol) was added before incubation with the labeled probe. DNA binding conditions and the electrophoretic analysis have been described previously (13). Synthetic double-stranded oligonucleotides used as competitors are listed in Table 1 (353/333, 267/238, 201/182, and 170/146). A DNA fragment from bp 126 to bp34 of the boIFNT1 promoter was amplified by PCR by using primers 126se and 34as (Table 1). The amplified DNA was purified by agarose gel electophoresis and excised by using a QIAEX II kit (Qiagen, Valencia, Calif.). The quantity of DNA was measured by ethidium bromide fluorescence quantitation (45).

GST pull-down assay. Each GST fusion protein was expressed in E. coli DH5 and allowed to bind to glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) under conditions recommended by the manufacturer. The size and quantity of bound fusion protein were estimated by SDS-PAGE and Coomassie blue staining. Human Ets-2 cDNA and murine Oct-4 cDNA were subcloned into pBluescript SK() to generate a template for the TNT-coupled transcription-translation system (Promega). A slurry of GST protein-beads (20 µl), which contained approximately 1 µg of protein, was suspended in 50 µl of binding buffer (20 mM Tris-HCl [pH 7.5], 0.12 M NaCl, 10% [vol/vol] glycerol, 0.055% [vol/vol] 2-mercaptoethanol, 1 mM EDTA, 0.1 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, 0.5% [vol/vol] Nonidet P-40) (12). An aliquot (5 µl) of in vitro-translated [³⁵S]-methionine-labeled protein was mixed with GST protein-beads and suspended for 1 h at 4°C. The beads were washed four times with 420 µl of washing buffer (0.12 M NaCl in the binding buffer was replaced with 0.1 M NaCl). The final, washed pellet was resuspended in SDS-PAGE sample buffer, boiled, and analyzed by SDS-PAGE. Radiolabeled proteins in the gel were visualized by fluorography by using EN³HANCE (New England Nuclear, Boston, Mass.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Inhibition of boIFNT1 reporter expression by Oct-4. As in earlier studies (13), transfections were performed with JAr choriocarcinoma cells, which are of human origin, because appropriate bovine or ovine cell lines are not available. A preliminary study was designed to determine first whether Oct-1, Oct-2, and Oct-4 were expressed in JAr cell and second whether the pCMV expression plasmid gave equivalent levels of expression of Oct-1, Oct-2, and Oct-4. Western blot analysis of untransfected JAr cells (Fig. 1A) showed that, while small amounts of Oct-1 were present, neither Oct-2 nor Oct-4 could be detected in the cell extracts. Importantly, different levels of apparent expression of the three Oct proteins were observed 36 h after transfection. Detection of Oct-4 was achievable only when the largest amount of expression plasmid (4 µg) was transfected and then required long exposure times to X-ray film. Although these data may be explained in part by the relative avidities of the antisera used in the chemiluminescence procedure, they also suggest that Oct-4 either is poorly expressed or has a short half-life in JAr cells. The Western blots revealed two bands of Oct-4 protein in transfected cells (Fig. 1A, right panel), consistent with data obtained by others, who indicated that the higher-molecular-weight form was phosphorylated (4).

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Expression of Oct proteins in transfected JAr cells. (A) Western blotting of JAr cell extracts (75 µg) before and after transfection with expression plasmids for Oct-1, Oct-2, and Oct-4. The amount of plasmid used in the transfection is shown above each panel; the exposure times for detecting chemiluminescence are shown below. Arrowheads show the positions of the respective proteins. The numbers to the left of each gel blot are molecular size markers (in thousands). (B) Structures of the Oct expression constructs that were used in transfection experiments. All plasmids share the same expression system [CMV promoter-tk leader and -globin gene poly(A) signal]. The numbers represent the lengths (in amino acid residues) of the Oct proteins.

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Oct-4 acts solely as a repressor of the boIFNT1 promoter. (A) The 126luc construct is suppressed specifically by Oct-4 and not by other octamer proteins. JAr cells were transfected with 3 µg of 126luc, 0.2 µg of Ets-2, and/or 0.2 µg of Oct expression plasmid. Reporter activities were normalized to the activity of cotransfected plasmid pRL-CMV. (B) Dose-dependent repression of the boIFNT1 promoter by Oct-4. JAr cells were transfected with 3.2 µg of 126luc, 0.4 µg of Ets-2, and 0 to 400 ng of Oct-4 expression plasmid. Reporter activities were normalized to the activity of cotransfected plasmid pRSVLTR-gal. Error bars show standard errors (SE). (C) Oct-4 transactivation of an artificial promoter, 3×Oct-tk. JAr cells were transfected with 3 µg of 3×Oct-tk reporter, 0.2 µg of Oct-4, and 0.2 µg of Ets-2 expression plasmid. Reporter activities were normalized to the activity of cotransfected plasmid pRL-CMV. Results are means and SE from three independent experiments. Activity is expressed as fold activation relative to basal activity. (D) Oct-4 represses E.18 reporter activity in either the absence () or the presence (+) of Ets-2. JAr cells were transfected with 3.2 µg of E.18-luc reporter and 0.4 µg of Ets-2 expression plasmid. Reporter activities were normalized to the activity of cotransfected plasmid pRSVLTR-gal. Data are reported as described for panel C.

Oct-4 suppresses a simple promoter that contains Ets-2 binding sites. To determine whether Oct-4 inhibits Ets-2 transactivation of a simple promoter, the E.18 reporter (57), which consists of two inverted Ets-2 consensus binding sites located upstream of a c-fos minimal promoter (56 to +119), was cotransfected into JAr cells either alone or in combination with Oct-4 and Ets-2 expression plasmids (Fig. 2D). Ets-2 transactivation of the E.18 promoter was evident but was relatively modest compared with the effect on the more complex IFNT enhancer (Fig. 2A). Oct-4 suppressed the activity of the E.18 promoter, both in its basal state and after transactivation by Ets-2 (Fig. 2D), but the extent of silencing was limited (only about 50% in each instance). These experiments show that Oct-4 suppression of Ets-2 transactivation can occur through a simple Ets-2-responsive promoter that lacks obvious octamer binding sequences. Oct-4 therefore might be able to interact with Ets-2 without binding to DNA itself. Moreover, suppression seems somewhat independent of the promoter involved.

Oct-4 suppression of the boIFNT1 promoter after deletion of all presumptive octamer binding sites. IFNT genes are intronless and occur as a multigene family (38, 39). The first 400 bases of their promoter regions are highly conserved both within and between species, as might be anticipated from their relatively recent origin. A 450 boIFNT1 promoter is also sufficient to direct the full expression of reporter genes in JAr cells (10). Consistent with earlier data (13), Ets-2 was able to transactivate a 457luc reporter 23-fold (Fig. 3). As anticipated, the activity increase observed in the presence of Ets-2 was suppressed more than 90% by cotransfection with 0.4 µg of Oct-4 plasmid. When the Ets-2 binding site on the 457luc promoter was inactivated by mutation, the ability of Ets-2 to transactivate the promoter was almost but not completely lost (Fig. 3). Oct-4 effects were also partially abolished when the mutated promoter was used, although only about one-half of the low residual activity noted in the presence of Ets-2 was inhibited by cotransfection with Oct-4.

View larger version (19K): [in this window] [in a new window]   FIG. 3.   Oct-4 suppresses the promoter activities of boIFNT1 promoter constructs with progressive deletions removing potential Oct binding sites. The IFNT promoter deletions (from 457 to 49 bp) linked to luc reporters were transfected into human JAr cells in either the absence (Basal) or the presence of the expression vectors for human Ets-2 (+Ets-2) and Ets-2 and murine Oct-4 together (+Ets-2 +Oct-4). As a control, expression from the 457luc construct, with its Ets binding site mutated, was measured in the presence and absence of Ets-2 with and without cotransfection with Oct-4. Reporter activities are expressed relative to the activity of cotransfected plasmid pRSVLTR-gal. Results are means and standard errors from three independent experiments. Activity is shown as fold activation relative to the basal activity for each construct, except in two instances. For the mutated 457luc construct, fold changes in expression are shown relative to the basal expression of nonmutated 457luc. Similarly, expression from 49luc that lacks the Ets-2 binding site is compared to the basal expression of 126luc. A mutation targeted to the Ets binding site is indicated at the bottom of the figure. wt, wild type; mets, mutated Ets site.

View larger version (31K): [in this window] [in a new window]   FIG. 4.   Failure of Oct-4 to bind the octamer-like sites in the boIFNT1 promoter efficiently. (A) Structure of the IFNT promoter and the locations of the octamer-like sites, the Ets site, and the TATA box. (B) DNA competition binding assays with a GST-Oct-4 fusion protein. Recombinant GST (negative control) or GST-Oct-4 fusion protein was incubated with a 32P-labeled, double-stranded canonical octamer (Table 1) in the presence of either no competitor DNA () or a 250-fold excess of unlabeled probe (c. octamer), double-stranded oligonucleotides identical in sequence to the octamer-like sites in the boIFNT1 promoter (lanes 1 to 4), and the promoter fragment from 126 to 34 (last lane). (C) Comparisons of the putative octamer sites in the boIFNT1 promoter with canonical octamer sequences. Mismatched sequences are shown with lowercase letters.

DNA competition binding assay with Oct-4. To identify whether Oct-4 suppresses IFNT promoter activity by binding to promoter DNA, a GST-Oct-4 fusion protein was used in a mobility shift assay for DNA competition analysis. A DNA fragment from the mouse immunoglobulin heavy-chain gene enhancer, which contains a canonical octamer site (49), was used as the DNA probe (Fig. 4C). Although the GST protein itself did not form a DNA-protein complex with the probe, the Oct-4 fusion protein did form a complex (Fig. 4B). A 250-fold molar excess of unlabeled probe competed efficiently with labeled probe, thereby confirming the specificity of the interaction (Fig. 4B).

Domain specificity of Oct-4 for the suppression effect on the boIFNT1 promoter. To determine what parts of the tripartite Oct-4 protein contributed to the suppression of the IFNT promoter, various domain deletion mutants were cotransfected with the Ets-2 expression construct and the 126luc reporter. In this series of experiments, Ets-2 activation of the IFNT promoter averaged about 12-fold, and Oct-4 suppression of reporter activity was always greater than 80% (Fig. 5). Only one truncated Oct-4 construct, the one with the N-terminal and POU domains intact but with the C terminus largely deleted, was an effective silencer. Neither the POU domain (aa 135 to 286 plus aa 349 to 352), the N-terminal domain alone, nor the POU domain plus the C-terminal domain was able to suppress Ets-2 transactivation of the 126 boIFNT1 promoter effectively. These data suggest that silencing requires both the N-terminal and the POU domains of Oct-4.

View larger version (12K): [in this window] [in a new window]   FIG. 5.   Domain specificity of Oct-4 in the suppression of the boIFNT1 promoter. The 1261uc reporter was transfected into JAr cells in either the absence () or the presence of expression vectors for Ets-2 (+) and for both Ets-2 and Oct-4 deletion constructs. The structures of the various Oct-4 deletion proteins are shown diagrammatically on the left, with shaded rectangles representing the POU domains. Reporter activity was normalized to the activity of cotransfected plasmid pRSVLTR-gal. Results are means and standard errors. Activity is expressed as fold activation relative to basal activity. Data from four independent transfections, each run in triplicate, were log transformed to limit the heterogeneity of variance and were analyzed by least-squares analysis of variance (PC-SAS version 6.12; Statistical Analysis System Institute, Cary, N.C.). Pairwise comparisons among treatments were completed by using F test statistics (PC-SAS). Values marked with different letters (a, b, c, and d) differ significantly (P < 0.05).

Recombinant Ets-2 protein specifically binds to the POU domain of Oct-4 in vitro. To explore the manner whereby Oct-4 interacts with Ets-2, GST pull-down assays were used. GST-Oct-4 and various truncation products were coupled to glutathione-Sepharose beads and tested for their abilities to trap Ets-2 that had been labeled with [³⁵S]methionine by in vitro translation. As expected, ³⁵S-labeled Ets-2 did form a complex with full-length GST-Oct-4 (Fig. 6A). In order to determine which domain of Oct-4 was responsible for the interaction, a series of Oct-4 truncations were tested in pull-down assays (Fig. 6B). All fusion proteins that contained an intact POU domain bound to Oct-4, but there was no interaction with either the C-terminal (aa 280 to 352) or the N-terminal (aa 1 to 136) peptides. Clearly, the POU domain of Oct-4 was sufficient to provide binding to Ets-2.

View larger version (29K): [in this window] [in a new window]   FIG. 6.   Recombinant Ets-2 protein binds specifically to the POU domain of Oct-4 in vitro. (A) Ets-2 protein interacts with Oct-4 protein in a GST pull-down assay. E. coli-expressed GST-Oct-4 fusion protein immobilized on Sepharose beads was mixed with [35S]methionine-labeled, in vitro-translated Ets-2 protein. Protein bound specifically was analyzed by SDS-10% PAGE. The first lane shows an analysis of the input protein (10% of in vitro-translated Ets-2). The second lane shows that immobilized GST failed to bind in vitro-translated Ets-2, while the third lane shows that GST-Oct-4 bound the radioactive protein. (B) Summary of the data from the pull-down assay, in which a series of Oct-4 truncations were tested for their ability to bind 35S-labeled Ets-2. The truncated proteins (shown diagrammatically on the left) were synthesized as GST fusion proteins and coupled to Sepharose. The ability of the proteins to bind 35S-labeled Ets-2 was then assessed in a pull-down assay. The data are consistent with the conclusion that the POU domain (aa 127 to 282) is required for Oct-4 to bind Ets-2.

Recombinant Oct-4 binds to the transactivation domain of Ets-2 in vitro. In this experiment, the experiment shown in Fig. 6 was reversed. GST-fused Ets-2 deletion mutants were coupled to glutathione-Sepharose and probed with ³⁵ S-labeled Oct-4 (Fig. 7A). The results of a typical experiment out of three performed are shown in Fig. 7B. Polypeptide fragments representing the C terminus of Ets-2 (aa 322 to 465), which contains the DNA binding domain, and the first 130 or 209 aa (including the "pointed" domain; aa 68 to 168) of the N terminus of the protein failed to bind to Oct-4. However, two polypeptides (aa 1 to 326 and 45 to 465), both of which included a central sequence of Ets-2 between the pointed and DNA binding domains, exhibited binding activity.

View larger version (27K): [in this window] [in a new window]   FIG. 7.   Recombinant Oct-4 binds to a central domain of Ets-2 in vitro. (A) Oct-4 protein interacts with Ets-2 protein in a GST pull-down assay. GST-Ets-2 fusion protein and GST immobilized on Sepharose beads were mixed with [35S]methionine-labeled, in vitro-translated Oct-4. Bound protein was analyzed by SDS-PAGE as described in the legend to Fig. 6. The first lane is input protein (10% of the in vitro translation mixture of Oct-4). (B) Summary of the data from the pull-down assay, in which a series of Ets-2 truncations (shown on the left) were tested for their ability to bind 35S-labeled Oct-4. The data indicate that a central domain within Ets-2 is required for binding to Oct-4.

Ets-2 associates with Oct-4 in vivo. Transfection of JAr cells with the Ets-2 expression plasmid, followed 36 h later by direct analysis of whole-cell lysates by Western blotting with anti-Ets-2 antiserum, revealed a strong immunopositive band that was not detectable in untransfected control cells (Fig. 8, fifth and seventh lanes). However, the Ets-2 signal was observed in the untransfected cell lysate when larger amounts of the lysate were analyzed (data not shown). The same level of Ets-2 expression after transfection was present whether or not Oct-4 was coexpressed (Fig. 8, sixth lane).

View larger version (11K): [in this window] [in a new window]   FIG. 8.   Oct-4 binds Ets-2 in vivo. Whole-cell extracts were prepared from JAr cells that had been transiently transfected with 10 µg of pCGNEts-2 in either the presence or the absence of 10 µg of pCMV-Oct4. Protein G cross-linked with antiserum against Oct-4 (-Oct-4) or nonimmune serum (NRS) was then added. Immune complexes were collected and analyzed by SDS-PAGE and Western blotting with anti-Ets-2 antiserum (first through fourth lanes). Blots show the presence of Ets-2 protein in whole-cell lysates (10 µg) of transfected cells (fifth and sixth lanes) and in nontransfected cells (seventh lane). The fourth lane shows that Ets-2 is immunoprecipitated with anti-Oct-4 antiserum only in cells coexpressing both proteins.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Oct-4 probably holds a key position in maintaining cells in a pluripotent state (33-35). Until the blastocyst stage in the mouse, cells that give rise to the trophoblast and those that will form the embryo proper express Oct-4. The consensus view is that Oct-4 prevents differentiation by maintaining the expression of key embryonic genes. As postulated here and elsewhere (22, 23), they may also silence the transcription of genes that are associated with differentiation into the trophectoderm. The positively regulated genes in the murine i.c.m. include FGF-4 and the transcriptional coactivator UTF1 (1, 27). A third potential target is the Glut-3 glucose transporter, whose expression mirrors that of Oct-4 (44). In contrast, two genes up-regulated in the trophectoderm of human embryos (the - and -subunit genes of hCG) are silenced by Oct-4 (22, 23). Surprisingly, Oct-4 control of the hCG and hCG genes, two genes that must be expressed coordinately to produce the active hormone, is different. In the case of the hCG gene, an atypical octamer site localized at about 270 binds Oct-4 and is responsible for the silencing (so-called direct silencing; see below) (23). With hCG, Oct-4 exerts its effects within a relatively narrow (170 to 148) region of the promoter, but no direct binding to DNA is required (22). Instead, the suppression of transcription appears to be dependent upon interactions with other, presently unknown, protein factors. There is therefore a close parallel between the silencing of the hCG gene and that of the IFNT gene described here.

Chorionic gonadotropin is a hormone confined to the primate order (40). In ruminants, the corpus luteum of pregnancy is prevented from undergoing luteolysis by the release of IFN- from the trophoblast prior to the formation of the placenta. Timely production of the hormone is crucial if the pregnancy is to be maintained (11). As with hCG, IFN- can first be detected as the blastocyst forms, although maximal production per cell is not reached for a few days, at a time when Oct-4 is fully down-regulated (54). Although initial experiments to demonstrate the effects of Oct-4 on the expression of transfected IFN- promoters in JAr cells indicated only a modest effect (9), which may be through endogenously expressed Ets-2, Oct-4 is clearly a potent silencer of IFN- promoters when the transactivator, Ets-2, is overexpressed (Fig. 2A and B).

Although there are several octamer-like binding sites in the IFNT promoter, none of these could bind Oct-4 strongly, and all could be eliminated without any effect on silencing (Fig. 3 and 4). These results rule out a direct mechanism of repression in which Oct-4 would cause silencing by docking to the promoter and interfering with the transcriptional process. Similarly, it seems unlikely that Oct-4 competes with Ets-2 or some other transcription factor for a site on the promoter (the so-called competition mechanism of repression). There is, for example, no obvious octamer binding site either overlapping or close to the Ets-2 site, and the entire proximal IFNT promoter region (126 to 34) fails to compete with Oct-4 for binding to a double-stranded canonical octamer sequence (Fig. 4B).

As observed previously for the shorter 126 promoter (13), a modest transactivation of the 457 boIFNT1 promoter by Ets-2 was observed after the main Ets-2 binding site had been mutated (Fig. 3). This increase in activity was partially reversed when Oct-4 was expressed. Conceivably, there is a cryptic, although much weaker, Ets-2 enhancer site on the promoter. Alternatively, the effects could be indirect and mediated through the action of Ets-2 and Oct-4 on another gene.

Two other mechanisms have been proposed to cause the repression of eukaryotic genes, squelching and quenching (21). Squelching results from the ability of the silencer to sequester either the activator itself or some other molecule necessary for transactivation of the gene (6). Such a mechanism appears unlikely for Oct-4 repression of IFNT. Ets-2 did not, for example, have a reciprocal effect and prevent Oct-4 from transactivating a reporter gene through the 3×Oct enhancer (Fig. 2C). A second observation that appears to rule out squelching as the silencing mechanism is that Oct-4 interacts with Ets-2 through its POU domain (Fig. 6), yet the POU domain alone is ineffective as a silencer (Fig. 5). Clearly, Oct-4 is able to interact with Ets-2 both in vitro (Fig. 6 and 7) and in vivo (Fig. 8).

The fourth type of silencing, quenching, occurs when the silencer interferes with the ability of the DNA-bound transactivator to interact with the basal transcriptional machinery (21). Such a mechanism seems most consistent with the observed Oct-4 suppression of the IFNT promoter (Fig. 9). First, Oct-4 seems not to be required to bind to promoter DNA to exert its effect (Fig. 3 and 4). Instead, Oct-4 binds Ets-2, with its POU domain targeting a region between the pointed and DNA binding regions of Ets-2 (Fig. 6 and 7). Presumably, this interaction places the N terminus in a position to block transcription (Fig. 9). It may be significant that a functional transactivation domain of Ets-2 has been mapped to a region within aa 1 to 293 (7) and that the POU domain of octamer proteins forms functional complexes with several transcription factors, including C/EBP (15), Sox-2 (1, 3, 27), and the viral proteins EIA and E7 (5), although in each of these instances the outcome is up-regulation rather than silencing of the targeted promoters. It remains to be seen whether the experiments reported here can be mimicked with other types of cells and whether either the POU or the C-terminal domain of Oct-4 can be individually replaced by the homologous regions from other Oct proteins and still provide effective silencing.

View larger version (11K): [in this window] [in a new window]   FIG. 9.   Oct-4 likely silences IFNT gene transcription through a quenching mechanism involving binding to the transactivator Ets-2.

There is considerable similarity between the mechanism of Oct-4 suppression of the IFNT promoter and the silencing of myelin P₀ gene transcription by a related octamer protein, Oct-6 (also known as SCIP and Tst-1), which is expressed in premyelinating Schwann cells and oligodendrocytes (17, 24, 55). In both instances, repression of transcription occurs through protein-protein interactions involving the POU domain and regardless of whether or not octamer binding sites in the promoters are deleted. Moreover, the N-terminal domain of the Oct protein appears to provide the specificity for the repression (24). Finally, just as Oct-4 is down-regulated as trophoblast cells form, Oct-6 expression is lost as myelinated cells begin to appear (17).

In conclusion, the findings of this study provide additional evidence that Oct-4 can function as a silencer as well as a transactivator. Oct-4 likely contributes to maintaining cells in an undifferentiated, pluripotent or totipotent state in two ways, by activating certain key genes and silencing others. This silencing, which for IFNT involves quenching of Ets-2 transactivation, may be a key mechanism that prevents cells of the i.c.m. from expressing products that direct differentiation toward the trophectoderm.