Rex-1, a Gene Encoding a Transcription Factor Expressed in the Early Embryo, Is Regulated via Oct-3/4 and Oct-6 Binding to an Octamer Site and a Novel Protein, Rox-1, Binding to an Adjacent Site

Abstract

The Rex-1 (Zfp-42) gene, which encodes an acidic zinc finger protein, is expressed at high levels in embryonic stem (ES) and F9 teratocarcinoma cells. Prior analysis identified an octamer motif in the Rex-1 promoter which is required for promoter activity in undifferentiated F9 cells and is involved in retinoic acid (RA)-associated reduction in expression. We show here that the Oct-3/4 transcription factor, but not Oct-1, can either activate or repress the Rex-1 promoter, depending on the cellular environment. Rex-1 repression is enhanced by E1A. The protein domain required for Oct-3/4 activation was mapped to amino acids 1 to 35, whereas the domain required for Oct-3/4 repression was mapped to amino acids 61 to 126, suggesting that the molecular mechanisms underlying transcriptional activation and repression differ. Like Oct-3/4, Oct-6 can also lower the expression of the Rex-1 promoter via the octamer site, and the amino-terminal portion of Oct-6 mediates this repression. In addition to the octamer motif, a novel positive regulatory element, located immediately 5′ of the octamer motif, was identified in the Rex-1 promoter. Mutations in this element greatly reduce Rex-1 promoter activity in F9 cells. High levels of a binding protein(s), designated Rox-1, recognize this novel DNA element in F9 cells, and this binding activity is reduced following RA treatment. Taken together, these results indicate that the Rex-1 promoter is regulated by specific octamer family members in early embryonic cells and that a novel element also contributes to Rex-1 expression.

The development of an organism from a single cell is a complex process. One substance known to influence various aspects of embryogenesis is retinoic acid (RA). RA regulates the transcription of genes in part by acting through two types of nuclear receptors, RA receptors and retinoid X receptors (6, 11, 33, 44). F9 teratocarcinoma stem cells, the malignant stem (undifferentiated) cells of the mouse teratocarcinoma, resemble cells of the murine blastocyst inner cell mass (ICM), which give rise to the entire fetus and the extraembryonic endoderm and extraembryonic mesoderm components of the placenta. F9 cells are a widely used in vitro model of differentiation of the early mouse embryo because they can differentiate into nonmalignant cells resembling the extraembryonic endoderm of the mouse blastocyst (19, 63). This differentiation is induced by RA. The levels of expression of many genes increase or decrease during F9 differentiation. Many of the RA-activated genes have been analyzed extensively, but less is known about the regulation and the roles of genes whose expression is down-modulated during F9 differentiation. The Rex-1 and Oct-3/4 genes are two genes whose mRNA levels are high in undifferentiated embryonal carcinoma (EC) cells and in the ICM and diminish during EC differentiation and normal embryonic development (20, 40, 47, 49, 56, 58).

The Rex-1 gene is a developmentally regulated acidic zinc finger gene (Zfp-42) (20). The presence of a zinc finger motif in Rex-1 suggests that the Rex-1 protein binds DNA and regulates transcription. This possibility is supported by the recent identification of a protein which possesses zinc finger motifs very similar to those in Rex-1 and which binds to transcriptional regulatory elements in a broad range of cell types. This protein, generally called YY1, has been studied in a large number of cell types (5, 50, 60). In contrast, Rex-1 mRNA is detected in a limited range of cells and tissues: undifferentiated embryonic stem (ES) and EC cells, mouse embryos at the blastocyst stage, trophectoderm, and meiotic germ cells of the adult mouse testis (47). Transcription of the Rex-1 gene is reduced when F9 cells are induced to differentiate with RA (20, 47). The Rex-1 promoter contains an octamer motif (ATTTGCAT) at position −220 which is required for the activity of the Rex-1 promoter in F9 stem cells and contributes to the RA-induced down-regulation of the gene (21). The octamer motif is a binding site for octamer transcription factor members of the POU domain family of DNA-binding proteins. The members of the POU family of transcription factors share two regions of homology: a highly conserved POU-specific domain and a more divergent homeodomain (55). EC and ES cells express three members of the POU family of transcription factors: Oct-3/4, its alternative spliced form Oct-5, and Oct-6 (40, 49, 56, 58, 59), in addition to the ubiquitously expressed Oct-1. When F9 EC cells differentiate in response to RA, the expression of Oct-3/4, Oct-5, and Oct-6 genes decreases (26, 37, 40, 56).

Oct-3/4 is the earliest-expressed gene known to encode a transcription factor which is developmentally regulated during mammalian embryogenesis. It is expressed very early in development in the totipotent and pluripotent stem cells of the pregastrulation embryo, including oocytes, early cleavage stage embryos, and the ICM of the blastocyst (49, 58). Oct-3/4 mRNA expression is down-regulated in the embryo during differentiation to endoderm and mesoderm. In the adult, Oct-3/4 expression is detected in both ovary and testis, where it is confined to oocytes and to primordial germ cells. Oct-3/4 mRNA is expressed in EC and ES cells, and its expression is down-regulated when these cells are induced to differentiate by RA (40, 49, 59). Therefore, it is very likely that expression of Oct-3/4 plays an important role in determining early steps in embryogenesis and differentiation. Support for this notion was gained by showing that in EC × fibroblast somatic cell hybrids, Oct-3/4 expression is suppressed, and reexpression of Oct-3/4 in these cells correlated with differentiation potential (2, 61).

Oct-3/4 gene expression is regulated by RA through its enhancer and promoter elements (3, 38, 41, 45, 54, 66, 74). Whereas no RA-responsive elements (RARE) were identified in the enhancer region, the Oct-3/4 promoter contains a RARE motif (designated RAREoct) which contributes to transcriptional activation in EC cells and mediates the RA-induced repression in RA-differentiated EC cells (45, 54). Inhibition of Oct-3/4 expression was shown to occur through the binding of ARP-I/COUP-TFII and EAR-3/COUP-TFI orphan receptors to the RAREoct site. These orphan receptors bind to the RAREoct site with a high affinity and actively silence the promoter activity (3, 54, 66).

The Oct-6 gene is expressed in EC and ES cells, in glial progenitor cells, and in a restricted set of neurons in the central nervous system (16, 37, 39, 65). Both the Oct-3/4 and Oct-6 proteins have been shown to function as positive or negative regulators of transcription, depending on the cellular environment and/or the exact promoter architecture (17, 26, 36, 39, 57, 65).

Given the coexpression of Oct-3/4, Oct-6, and Rex-1 in the early embryo and in EC, ES, and testis cells, their down-regulation by RA treatment in EC and ES cell lines, and data showing that the Rex-1 promoter contains an octamer motif which is crucial both for Rex-1 expression in EC cells and for Rex-1 negative regulation by RA, we were interested in analyzing the effects of Oct-3/4 and Oct-6 on Rex-1 promoter activity. In this report, we have shown that Oct-3/4 and Oct-6 specifically regulate the Rex-1 promoter through the Octa site in a dose-dependent manner. This regulation requires both the N-terminal and the DNA binding domains of these Oct proteins. However, different amino acids in the Oct-3/4 N-terminal region are required for repression or activation of Rex-1. Moreover, we have identified a novel positive regulatory element which is contained within an 11-bp sequence, located immediately 5′ to the octamer motif. Mutations in this element severely compromise the activity of the Rex-1 promoter, suggesting that this element plays a key role in the activation of Rex-1 gene transcription. We show that this sequence binds a protein(s) (designated Rox-1), which is specifically expressed in EC and ES cells and which exhibits reduced expression in RA-differentiated F9 cells.

MATERIALS AND METHODS

Cells.

Cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum, 2 mM glutamine, 100 U of penicillin/ml, and 100 mg streptomycin/ml. Differentiation was induced by the addition of 10⁻⁶ M RA (Sigma) for 2 to 3 days.

Plasmids.

The expression plasmid Oct-3/4 was made by cloning the full-length Oct-3/4 cDNA into a SmaI site in the cytomegalovirus-based expression vector pFCS, kindly provided by P. Brulet (9). Deletions of the Oct-3/4 cDNA were created by using naturally occurring restriction sites. The different constructs were cloned in frame into the expression vector pFCS, pEVRF0, or pEVRF1 (34). The amino-terminus-deleted plasmids ΔN35 and ΔN126 were generated by deletion of 131 and 403 bp, respectively, of the 5′ end of Oct-3/4 cDNA. The number following the N refers to the number of codons removed from the Oct-3/4 open reading frame. The carboxy-terminus-deleted plasmid ΔC75 was constructed by deletion of 400 bp of the 3′ end of the Oct-3/4 cDNA. The number following the ΔC indicates the number of codons removed from the carboxy terminus of the Oct-3/4 protein. These deletion constructs were tested for expression in COS-1 cells. Whole-cell extracts (WCEs) prepared from transfected COS-1 cells were used in electrophoretic mobility shift assays (EMSAs), which indicated that all Oct-3/4 mutants were stably expressed and bound to the labeled Octa oligonucleotide. Plasmid ΔDB was generated by excision of the StuI-StuI internal fragment (positions 571 to 1033) of the Oct-3/4 cDNA. The numbers represent the codons at which the Oct-3/4 open reading frame is fused. Analysis of the protein extract generated from the ΔDB-transfected COS-1 cells on sodium dodecyl sulfate-polyacrylamide gels indicated that this mutant encodes a stable protein.

The Oct-6 expression vectors, described in detail elsewhere (36), were a gift from Dies Meijer. The previously described (70) N-terminal deletion mutant 5′P was a gift from Louis H. Staudt. Plasmid pLA8, containing the adenovirus type 2 E1A region, was obtained from M. Horowitz. The Oct-1 expression vector was provided by W. Herr (68).

The reporter plasmids 0.3Rex-CAT and pRoxOcta*-CAT (previously designated Boct-CAT) were previously described (21). Plasmid px3-Octa-CAT was constructed by inserting three copies of double-stranded Octa oligonucleotide sequence and surrounding sequence (5′GATCCGTACTAATTTGCATTTCTA3′) into the BamHI site located upstream of the TATA box of plasmid px3-Octa-CAT, provided by T. Kadesch. The reporter plasmid pPκEκ-CAT was described previously (62). The pRox*Octa-CAT plasmid was made by site-directed mutagenesis using the Rox*Octa oligonucleotide described in Table 1. Plasmids pmRox1, pmRox2, pmRox3, pmRox4, pmRox5, pmRox6, and pmRox1,5,6 were made by site-directed mutagenesis using the oligonucleotides described in Table 1. The mutated fragment was digested with KpnI-ClaI and cloned into the reporter plasmid pBCO (8). The mutation in the sequence was confirmed by DNA sequencing.

TABLE 1.

Oligonucleotides used

Name	Description	Sequencea
RoxOcta	WTb Rex-1 promoter from −234 to −204	GATTCAGAAGAGGCATTTGCATAACTGAGCA
RoxOcta*	Rex-1 promoter with WT Rox-1 binding site; mutations in the octamer sequence (same mutations as in the previously described pBoct-CAT)	GATTCAGAAGAGGCATgTtCATAACTGAGCA
Rox*Octa	Rex-1 promoter with mutations in the Rox-1 binding site; WT octamer sequence (same mutations as in pRox*Oct-CAT)	GATTCctcctctGCATTTGCATAACTGAGCA
mRox1	Rex-1 promoter with mutations in 2 base pairs in the Rox-1 binding site; WT octamer sequence	GATgaAGAAGAGGCATTTGCATAACTGAGCA
mRox2		GATTCctAAGAGGCATTTGCATAACTGAGCA
mRox3		GATTCAGccGAGGCATTTGCATAACTGAGCA
mRox4		GATTCAGAAtcGGCATTTGCATAACTGAGCA
mRox5		GATTCAGAAGAttCATTTGCATAACTGAGCA
mRox6		GATTCAGAAGAGtaATTTGCATAACTGAGCA
mRox1,5,6	Rex-1 promoter with mutations in 5 base pairs in the Rox-1 binding site; WT octamer sequence	GATgaAGAAGAttaATTTGCATAACTGAGCA
Octa	Octamer sequence from the μ-chain enhancer used for Fig. 1B	GATCCGTACTAATTTGCATTTCTA
Octa-3′	Rex-1 promoter with octamer and WT sequence 3′ of it	GGCATTTGCATAACTGAGCA
Octa*-3′	Rex-1 promoter with mutated octamer and WT sequences 3′ of it	GGCcagTtCATAACTGAGCA
TCRα enhancer	Nonspecific oligonucleotide used as a competitor	CGTAGGGCACCCTTTGAAGCTCTCCC

^aLowercase letters indicate mutations; boldface letters indicate the octamer motif. 

^bWT, wild type. 

DNA transfections.

All transfections were performed by the calcium phosphate precipitation procedure as described previously (72). Transfections were carried out with 5 × 10⁵ cells, 5 to 10 μg of chloramphenicol acetyltransferase (CAT) reporter plasmid, 2 μg of β-galactosidase (β-Gal) internal control plasmid (pRSV-βgal), and the amounts of effector plasmids indicated below. In all cases, the total amount of DNA was adjusted with pBluescript DNA. For transfection into RA-treated cells, 10⁻⁶ M RA was added at the time of transfection. Medium with or without RA was refreshed 16 to 20 h after transfection. After an additional 20 to 24 h, cells were harvested for CAT assays. CAT activity was measured by using [¹⁴C]chloramphenicol (53 mCi/mmol; Amersham International) as the substrate in the presence of acetyl coenzyme A at 37°C for 16 h. [¹⁴C]chloramphenicol was separated from its acetylated forms by silica thin-layer chromatography and quantitated on a PhosphorImager by using ImageQuant software. The results are expressed as mean percent conversion of chloramphenicol to acetylated forms, on the basis of three independent transfections.

WCEs.

WCEs were prepared by lysing the cells in 100 μl of high-salt extraction buffer (400 mM KCl, 20 mM Tris-HCl [pH 8.0], 20% [vol/vol] glycerol, 2 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 1 mg of leupeptin/ml, 0.3 mg of antipain/ml, 0.5 mg of trypsin inhibitor/ml). Cells were lysed by three cycles of freeze (−70°C)-thaw (ice), and the cellular debris was removed by centrifugation at 12,000 × g for 15 min at 4°C.

DNA probes and EMSAs.

For EMSAs, the following probes were used: the RoxOcta oligonucleotide containing a region from the Rex-1 promoter between positions −234 and −204; the RoxOcta* oligonucleotide containing sequences between −234 and −204 with two point mutations in the octamer site; the Rox*Octa oligonucleotide containing sequences between −234 and −204 with seven point mutations in the Rox binding site; the mRox1, mRox2, mRox3, mRox4, mRox5, and mRox6 oligonucleotides containing sequences between −234 and −204 with two point mutations in the Rox-1 binding site; the mRox1,5,6 oligonucleotide containing sequences between −234 and −204 with five point mutations in the Rox binding site; the Octa oligonucleotide containing sequences from the μ-chain enhancer; the Octa-3′ oligonucleotide containing sequences between −223 and −204 of the Rex-1 promoter; and the Octa*-3′ oligonucleotide containing sequences between −223 and −204 with four point mutations in the octamer site. All probes are described in Table 1.

EMSAs were performed by incubating 15 μg of WCEs with 0.3 ng of ³²P-labeled probe at room temperature for 20 min in the presence of 2 μg of poly(dI-dC), 10 mM Tris-HCl (pH 7.8), 14% glycerol, 74 mM KCl, and 4 mM dithiothreitol. Samples were electrophoresed on a 4% polyacrylamide gel (19:1 acrylamide/bisacrylamide) in 0.25× Tris-borate-EDTA buffer. When competitor oligonucleotide was used, the competitor was added in 100-fold molar excess and preincubated in the reaction mixture described above for 10 min prior to the addition of the radiolabeled probe. For supershift assays, the appropriate antiserum was added to the reaction 15 min prior to addition of the probe.

DNase I footprinting assays.

For DNase I footprinting assays, an SpeI-HindIII fragment of 490 bp was isolated from plasmid p0.3Rex-CAT and labeled at the HindIII site, using the Klenow fragment and [α-³²P]dCTP to a specific activity of greater than 10,000 cpm/ng of DNA. The probe was incubated with 30 μg of F9 WCE in 40 μl of reaction mixture containing 10 mM Tris-HCl (pH 7.8), 14% glycerol, 4 mM dithiothreitol, 50 mM KCl, and 100 ng of poly(dI-dC). After incubation for 30 min at room temperature, 0.2 to 0.3 U of DNase I (Boehringer Mannheim) diluted in 50 mM MgCl₂–10 mM CaCl₂ was added for 1 min. The reaction was stopped by addition of 150 μl of stop solution containing 200 mM NaCl, 20 mM EDTA, 1% sodium dodecyl sulfate, and 33 μg of yeast tRNA/ml. DNA was extracted with phenol-chloroform, ethanol precipitated, and analyzed on a denaturating 6% polyacrylamide gel. Gels were dried and autoradiographed with an intensifying screen at −70°C. Sequencing lanes of the same probe were generated by the Maxam-Gilbert procedure (35).

RESULTS

The RoxOcta sequence located in the Rex-1 promoter binds Oct-1, Oct-3/4, Oct-6, and a novel factor designated Rox-1.

To characterize the binding sites involved in regulation of the Rex-1 promoter, we performed DNase I footprinting experiments, using the Rex-1 promoter as a probe, with nuclear extracts derived from undifferentiated F9 stem cells. This analysis identified a region that was protected on the sense (−226 and −204) and antisense (−229 and −209) DNA strands (Fig. 1A). This region contains not only the previously identified (48) Octa sequence (ATTTGCAT) but also sequences located 5′ and 3′ of it. EMSAs carried out with the labeled oligonucleotide (designated RoxOcta) containing the protected area resulted in the formation of four specific complexes using nuclear extracts prepared from F9 cells (Fig. 1B). As previously published (48), these complexes contain Oct-1, Oct-3/4, and Oct-6 (lanes 1 and 5); they were specifically inhibited by an unlabeled RoxOcta oligonucleotide (lane 2) and by an oligonucleotide containing the octamer sequence (lane 3) but not by an unrelated oligonucleotide (lane 4). Interestingly, we observed an additional complex, designated Rox-1, which migrated faster than Oct-6 and slower than Oct-3/4. This Rox-1 complex was specifically competed by the RoxOcta unlabeled probe but not by a 100-fold molar excess of unrelated oligonucleotide containing the sequence from the T-cell receptor alpha-chain (TCRα) enhancer (lanes 2 and 4). However, the Rox-1 complex was not competed by the addition of the Octa oligonucleotide, suggesting that the Rox-1 complex does not contain POU proteins (lane 3). To further support the notion that the Rox-1 binding site is not included in the octamer sequence, we performed EMSAs with the labeled Octa oligonucleotide. This analysis showed that while binding of Oct-1, Oct-3/4, and Oct-6 to the Octa probe was comparable to that observed with the RoxOcta probe, the Rox-1 complexes were not apparent (compare lanes 5 and 6).

FIG. 1.

Binding to the Rex-1 promoter region. (A) DNase I footprinting of the murine Rex-1 promoter. A ³²P-end-labeled Rex-1 promoter fragment (SpeI-HindIII) was incubated with 75 μg of WCE prepared from F9 cells (lanes 3, 4, 8, and 9) and in the absence of WCE (lanes 1, 2, 6, and 7). Lanes 5 and 10, Maxam-Gilbert A+C and C+T, respectively, sequencing ladders of the sense (lanes 1 to 5) and antisense (lanes 6 to 10) probes. The protected regions are boxed and the corresponding sequences are indicated. (B) The indicated ³²P-end-labeled oligonucleotide probes were incubated with WCE prepared from F9 cells. Binding reactions were performed in the absence of competitors (lanes 1, 5, and 6) or in the presence of the indicated competitors (lanes 2 to 4). Binding reactions shown in lanes 2 to 4 were performed in the presence of a 100-fold molar excess of the indicated unlabeled oligonucleotides. The arrows indicate the known Oct-1, Oct-3/4, and Oct-6 complexes and the novel Rox-1 complex.

As an additional piece of evidence that the Rox-1 protein neither binds to the octamer sequence nor is bound to an Oct-3/4 protein, we showed that the Rox-1 complex was competed by the unlabeled RoxOcta* oligonucleotide, which contains two mutations in the octamer sequence (Fig. 2A, lane 2; the mutations are depicted in Fig. 4). Furthermore, anti-Oct-3/4 antibodies supershifted the Oct-3/4 band but did not change the intensity or the position of the Rox-1 complex (lane 5). Thus, the Rox-1 complex is a novel complex which binds to the footprinted region of the wild-type Rex-1 promoter and does not contain Oct proteins.

FIG. 2.

The Rox-1 complex. (A) ³²P-end-labeled RoxOcta oligonucleotide was incubated with WCE prepared from F9 cells. Binding reactions were performed in the absence of a competitor (lanes 1 and 4), in the presence of a 100-fold molar excess of the indicated competitors (lanes 2 and 3), or in the presence of 1 μl of anti-Oct-3/4 (lane 5) or nonspecific (n.s.; lane 6) antibody. The oligonucleotides are depicted in Table 1. The arrows indicate the Oct-1, Oct-3/4, and Rox-1 complexes. Oct-6 is detected only in freshly prepared WCEs, and the extracts used in this experiment were frozen and subsequently thawed. (B) ³²P-end-labeled RoxOcta probe was incubated with WCE prepared from F9 cells (lanes 1 and 4), F9 cells treated with RA for either 2 days (lane 2) or 3 days (lane 3), ES cells (lane 5), F9 × L somatic cell hybrids (1) (lane 6), L cells (lane 7), P19 cells (lane 8), HL60 cells (lane 9), NIH-9 cells (lane 10), BW5417 T cells (lane 11), S194 myeloma cells (lane 12), M12 lymphoma cells (lane 13), WEHI 3B myeloid cells (lane 14), βTC6 insulinoma cells (lane 15), and H4 hepatoma cells (lane 16).

FIG. 4.

Functional importance of the Rox-1 binding site for Rex-1 promoter activity. (A) The p0.3Rex-CAT, pRoxOcta*-CAT, pRox*Octa-CAT, pmRox-1 to pmRox6, and pmRox1,5,6 reporter plasmids (10 μg) were cotransfected with β-Gal-containing reference plasmid (2 μg) into F9 stem cells. After 48 h, cells were harvested and lysed, and CAT activities were determined. The percent conversion to the acetylated forms of each separate transfection was normalized to the β-Gal activity. The values for percent conversion, presented as means ± standard deviation, corresponding to p0.3Rex-CAT, pRoxOcta*-CAT, pRox*Oct-CAT, pmRox1, pmRox2, pmRox3, pmRox4, pmRox5, pmRox6, and pmRox1,5,6 are 15 ± 2.06, 1.23 ± 0.4, 5.00 ± 0.14, 5.22 ± 0.2, 15.26 ± 1.5, 21.39 ± 3.2, 5.13 ± 0.4, 28.98 ± 0.3, 3.84 ± 0.07, and 2.63 ± 0.3, respectively. CAT activity of p0.3Rex-CAT was arbitrarily set at 100%. Relative CAT activity represents CAT activity relative to that obtained with p0.3Rex-CAT. (B) Sequences of wild-type RoxOcta and the mutant oligonucleotides. The RoxOcta probe is the sequence of the wild-type Rex-1 promoter from −234 to −204 (21). The octamer motif is boxed, and the mutant bases are underlined.

Rox-1 is differentially expressed in EC and ES cells.

To characterize further the Rox-1 complex, nuclear extracts were prepared from RA-differentiated F9 cells and from a number of cell lines and incubated with the labeled RoxOcta probe containing the binding sites for Oct-related proteins and for the Rox-1 protein. Interestingly, Rox-1 activity was down-regulated in nuclear extracts derived from RA-differentiated F9 cells (Fig. 2B, lanes 1 to 3). As shown in Fig. 2B, Rox-1 binding activity was detected only in WCEs derived from the EC cell lines, F9 and P19, and in WCE generated from ES cells. Rox-1 activity was not observed in L-fibroblast cells, F9 × L somatic cell hybrids, or HL60, NIH-9, BW5147 T, S194 myeloma, M12, WEHI 3B myeloid, βTC6 insulinoma, and H4 hepatoma cells (Fig. 2B, lanes 6, 7, and 9 to 16). Thus, Rox-1 activity appears to be present in cell lines that express the Rex-1 gene, while cell lines that do not contain Rox-1 do not transcribe Rex-1. These data are consistent with the notion that Rox-1 is required for Rex-1 promoter activity.

Rox-1 binds to the site adjacent to the octamer site and contributes to Rex-1 promoter activity.

To identify the nucleotides responsible for Rox-1 binding in the RoxOcta probe, we prepared a series of double-stranded oligonucleotides with a wild-type Octa sequence and sequences located either 5′ or 3′ of the octamer sequence. The results show that the Oct-1 and Oct-3/4 retarded complexes were efficiently competed by the Octa-3′ oligonucleotide, which contains the octamer wild-type sequence and sequences located 3′ of it, but not by the unlabeled Octa*-3′ oligonucleotide (containing a mutated Octa sequence) (Fig. 3A). In contrast, the Rox-1 complex was not affected by the addition of any of these oligonucleotides (Fig. 3A). Thus, we concluded that sequences located 3′ of the octamer binding site are not required for the formation of the Rox-1 complex.

FIG. 3.

Identification of the Rox-1 binding site. (A) ³²P-end-labeled RoxOcta oligonucleotide was incubated with F9 WCE. Binding reactions were done in the absence (−) or presence of Octa*-3′ (an oligonucleotide carrying mutations in the Octa site, same mutations as in the Octa oligomer shown in Fig. 4, and no mutation in the sequence located 3′ of the Octa) (lane 2), and Octa-3′ oligonucleotide containing wild-type Octa and wild-type sequences located 3′ to it (lane 3). (B) ³²P-end-labeled RoxOcta oligonucleotide was incubated with F9 WCE. Binding reactions were done in the absence (−) or presence of the indicated oligonucleotides.

To examine whether the sequences 5′ of the Octa site were required for Rox-1 complex formation, two double-stranded oligonucleotides with mutations either in the 5′ region or in the Octa sequence were analyzed for the ability to compete for Rox-1 binding. The oligonucleotide RoxOcta, which harbors the wild-type Octa and Rox-1 sequence, was able to inhibit the Rox-1 binding (Fig. 3B; compare lanes 1 and 2), while oligonucleotide Rox*Octa competes to a similar extent as an irrelevant oligonucleotide (Fig. 3B, lanes 3 and 5). The Rox*Octa oligonucleotide contains a wild-type Octa motif and mutations in the sequences located 5′ of the octamer site. Moreover, binding to the labeled Rox*Octa oligonucleotide resulted in Oct-1, Oct-6 (migrating slower than Rox-1), and Oct-3/4 complex formation only. Rox-1 complex was not detected (data not shown). Thus, sequences located immediately upstream of the Octa site in the Rex-1 promoter are recognized by the factor Rox-1. Knowing that the sequence located 5′ to the Octa site is responsible for Rox-1 binding to the RoxOcta probe, we generated a series of oligonucleotides, each mutated in two base pairs in the Rox-1 binding site, designated mRox1 to mRox6 (Table 1). An additional oligonucleotide containing five point mutations present in mRox1, mRox5, and mRox6 (designated mRox1,5,6) was prepared. All the mutant oligonucleotides harbor a wild-type Octa motif (Table 1). Analyzing the mRox1-mRox6 in EMSA showed that mutating two nucleotides affected the binding of Rox-1 to these oligonucleotides to a small extent (data not shown). In contrast, Rox*Octa and mRox1,5,6 containing multiple point mutations either in the core Rox-1 binding site or at the borders of the site competed very poorly or not at all for Rox-1 binding (Fig. 3B, lanes 3 and 4).

To define the functional contribution of Rox-1 binding to Rex-1 promoter activity, we introduced the point mutations in the Rox-1 binding site described above into the Rex-1 promoter region directing CAT reporter gene activity. These constructs were designated pRox*Octa-CAT, pmRox1 to pmRox6, and pmRox1,5,6. These plasmids have the advantage of maintaining the Rox-1 binding site within its natural promoter context and avoiding the alteration of spacing between transcription factor binding sites. p0.3Rex-CAT (containing the wild-type Rex-1 promoter), pRoxOcta*-CAT (containing the Rex-1 promoter mutated at the Octa site; previously designated pBoct-CAT), and the above-described reporter plasmids mutated at the Rox-1 site were individually transfected into F9 stem cells, and CAT activities were determined. Consistent with previous results (21), mutation of the Octa site reduced promoter activity by greater than 90% (Fig. 4). Mutation at the core of the Rox-1 site, pRox*Octa, reduced promoter activity to ∼30% of wild-type activity in F9 cells (Fig. 4), clearly indicating that the Rox-1 site contributes to Rex-1 promoter activity. Interestingly, mutating only two bases at a time in pmRox1, pmRox4 and pmRox6, lowered Rex-1 promoter activity considerably. Furthermore, mutating five bases located at the borders of the Rox-1 binding site, pmRox1,5,6, reduced promoter activity to a larger extent (<18% of wild-type activity [Fig. 4]). Taken together, these results demonstrate that the Rox-1 sequence, located immediately upstream of the Octa site in the Rex-1 promoter, contributes to promoter activity and is recognized by the novel protein(s), Rox-1. Mutating the Rox-1 binding site severely compromises the activity of the Rex-1 promoter in undifferentiated F9 stem cells. The Rox-1 binding site is a compound element in which nucleotides located at the borders of the site, mutated in pmRox1 and pmRox6, and the bases located at the core of the recognition element, mutated in pmRox4, are important for Rex-1 promoter activity.

Oct-3/4 activates the Rex-1 promoter in RA-treated, differentiated P19 teratacarcinoma cells.

The parallel decline of Rex-1 and Oct-3/4 mRNAs during RA-induced differentiation of EC and ES cells initially suggested to us that Rex-1 mRNA expression is sustained by positive regulation exerted by the Oct-3/4 gene product. We searched for a cell culture system which would resemble F9 teratocarcinoma cells but in which there was no detectable endogenous Oct-3/4 activity; in such a system, we would be able to assess the effect of only the exogenously introduced Oct-3/4 on the Rex-1 promoter. Since almost 6 days of RA treatment is required to achieve a very low level of Oct-3/4 protein in F9 cells (data not shown), we chose to use RA-differentiated P19 (P19/RA) cells. We could not detect any Oct-3/4 mRNA or protein following 24 to 48 h of RA treatment of the P19 cells, whereas Rox-1 protein was still expressed, albeit at a lower level (3). However, these cells express the Oct-6 protein, which prevented us from testing the effect of Oct-6 on Rex-1 activity in P19/RA cells.

We cotransfected the p0.3Rex-CAT construct containing the Rex-1 promoter fragment, together with the Oct-3/4 expression vector and the control plasmid pRSV-βgal, into P19/RA cells. CAT activity in these cells was determined, normalized to the corresponding β-Gal activities, and expressed as bar graphs. The Oct-3/4 expression vector up-regulated CAT activity driven by the wild-type Rex-1 promoter about sixfold above the baseline level, which is defined as the level of expression in the absence of cotransfected Oct-3/4 (Fig. 5). This up-regulation occurs through the Octa site, since Oct-3/4 activates an octamer-mutated Rex-1 promoter to a lesser extent (compare p0.3Rex-CAT to pRoxOcta*-CAT, in which the Octa site is mutated) (Fig. 5). Interestingly, mutations at the Rox-1 site (we chose pRox*Octa as a representative of Rox-1-mutated Rex-1 promoter) also affect the ability of the exogenous Oct-3/4 expression vector to activate Rex-1 promoter activity (Fig. 5). Thus, Oct-3/4 up-regulated the activity of the Rex-1 promoter only when both the Octa and the Rox-1 binding sites are intact.

FIG. 5.

Oct-3/4 activates Rex-1 promoter activity in P19/RA cells. The p0.3Rex-CAT, pRoxOcta*-CAT, and pRox*Octa-CAT reporter plasmids (10 μg) were transfected with β-Gal-containing reference plasmid (2 μg) into P19 cells which had been cultured in the presence of RA for 36 h. Transfections were performed in the absence (−) or presence (+) of 10 μg of wild-type Oct-3/4 expression vector. CAT activities were determined and normalized to β-Gal activity. The values for percent conversion, presented as means ± standard deviations, corresponding to p0.3Rex-CAT, pRoxOcta*-CAT, and pRox*Octa-CAT each in the absence and presence of Oct-3/4 are 8.1 ± 1.5, 47.7 ± 5, 5.4 ± 1.1, 9.72 ± 1.6, 16.9 ± 2.2, and 16.2 ± 1.5, respectively. CAT activity of each construct alone, in the absence of Oct-3/4, was arbitrarily set at 1. Relative CAT activity represents CAT activity relative to that obtained from p0.3Rex-CAT, pRoxOcta*-CAT, and pRox*Octa-CAT, respectively. Fold activation was calculated as the ratio between CAT activities in the absence and presence of Oct-3/4.

The Rex-1 promoter is inhibited by Oct-3/4 and Oct-6 through the Octa site in F9 stem cells.

We determined the effects of the Oct-3/4 and Oct-6 proteins on Rex-1 expression in F9 stem cells, where the two proteins are coexpressed. We cotransfected the p0.3Rex-CAT construct and either an Oct-3/4 or Oct-6 expression vector into undifferentiated F9 cells. Cotransfection of the Oct-3/4 or Oct-6 expression vector inhibited six- or sevenfold, respectively, the CAT activity driven by the Rex-1 promoter (Fig. 6A and 7A). The observed inhibition did not result from nonspecific effects, since cotransfection of equal amounts of a vector without Oct-3/4 or Oct-6 did not lead to inhibition of Rex-1 promoter activity. To ensure that the repression of Rex-1 promoter activity by Oct-3/4 or Oct-6 occurred through the Octa site, we cotransfected the pRoxOcta*-CAT reporter gene driven by the Rex-1 promoter in which the Octa site was mutated with or without the Oct-3/4 or Oct-6 expression vector. The results clearly demonstrate that mutation of the Octa site attenuates the ability of Oct-3/4 or Oct-6 to down-regulate the Rex-1 promoter activity (Fig. 6A and 7A). Thus, transfection of Oct-3/4 into cells which express a high level of endogenous Oct-3/4 mRNA (e.g., F9 stem) results in repression of the p0.3Rex-CAT reporter; transfection into RA-treated P19 cells, which express a low or undetectable level of Oct-3/4 mRNA, results in activation of the Rex-1 promoter.

FIG. 6.

Oct-3/4 inhibits Rex-1 promoter activity in F9 cells. (A) The p0.3Rex-CAT (containing the wild-type Rex-1 promoter), pRoxOcta*-CAT (containing the Octa-mutated Rex-1 promoter), and pRox*Octa-CAT (harboring mutations in the Rox-1 binding site) reporter plasmids (10 μg) were transfected with β-Gal-containing reference plasmid (2 μg) into F9 stem cells, with (+) or without (−) the Oct-3/4 expression vector. The values for percent CAT conversion, presented as means ± standard deviation, corresponding to p0.3Rex-CAT, pRoxOcta*-CAT, and pRox*Octa-CAT each in the absence or presence of Oct-3/4 are 58.7 ± 9.8, 10.11 ± 1.0, 1.69 ± 0.2, 1.36 ± 0.2, 14.8 ± 1.5, and 12.9 ± 0.3, respectively. CAT activity of each construct alone, in the absence of Oct-3/4, was arbitrarily set at 100. Relative CAT activity represents CAT activity relative to that obtained from p0.3Rex-CAT, pRoxOcta*-CAT, and pRox*Octa-CAT, respectively. Fold repression was calculated as the ratio between CAT activities in the absence and presence of Oct-3/4, and the data are averages of at least four independent experiments. (B) Effect of increasing amounts of Oct-3/4 on Rex-1 promoter activity. F9 stem cells were cotransfected with p0.3Rex-CAT and increasing amounts of Oct-3/4, and CAT activity was measured and assayed as described for panel A. The values for percent CAT conversion corresponding to p0.3Rex-CAT in the absence and presence of 0.25, 1, 5, 10, and 20 μg of Oct-3/4 are 13.46, 10.67, 5.17, 4.33, 2.0, and 1.71, respectively.

FIG. 7.

Oct-6 inhibits Rex-1 promoter activity. (A) The p0.3Rex-CAT and pRoxOcta*-CAT reporter plasmids (10 μg) were transfected with the β-Gal-containing reference plasmid (2 μg) into F9 stem cells with (+) or without (vector backbone only; −) the Oct-6 expression vector. After 48 h, CAT activities were measured and assayed as described for Fig. 6. The values for percent CAT conversion, presented as means ± standard deviation, corresponding to p0.3Rex-CAT and pRoxOcta*-CAT each in the absence or presence of Oct-6 are 18.2 ± 2.5, 2.5 ± 0.2, 0.46 ± 0.09, and 0.38 ± 0.08, respectively. (B) F9 stem cells were cotransfected with p0.3Rex-CAT and increasing amounts of the Oct-6 expression vector. Values for percent CAT conversion corresponding to p0.3Rex-CAT in the absence or presence of 0.25, 1, 2.5, 5, and 10 μg of Oct-6 are 17.2, 9.1, 6.8, 5.8, 4.3, and 3.1, respectively. (C) F9 cells were cotransfected with p0.3Rex-CAT in the absence or the presence of Oct-3/4 (1 or 10 μg), Oct-6 (1 or 10 μg), or both Oct-3/4 and Oct-6 expression vectors at the indicated amounts. The values for percent CAT conversion, presented as means ± standard deviation, corresponding to p0.3Rex-CAT are as follows: in the absence of Oct-3/4 and Oct-6, 23.4 ± 0.2; in the presence of 1.0 and 10 μg of Oct-3/4, 10.2 ± 1 and 4.4 ± 0.5; in the presence of 1 and 10 μg of Oct-6, 11.3 ± 0.15 and 5.5 ± 0.6; in the presence of 1 μg of Oct-3/4 and 1 μg of Oct-6, 6.5 ± 1.2; in the presence of 1 μg of Oct-3/4 and 10 μg of Oct-6, 4.3 ± 0.5; in the presence of 10 μg of Oct-3/4 and 1 μg of Oct-6 2.5 ± 0.3; and in the presence of 10 μg of Oct-3/4 and 10 μg of Oct-6, 1.4 ± 0.2.

To determine whether the inhibition of the Rex-1 promoter by Oct-3/4 or Oct-6 was dose dependent, a constant amount (10 μg) of p0.3Rex-CAT construct was cotransfected with increasing amounts of either the Oct-3/4 or Oct-6 expression vector. CAT activity was progressively decreased by increasing amounts of Oct-3/4 or Oct-6 expression vector (Fig. 6B and 7B). Cotransfection of the Oct-3/4 expression vector together with the Oct-6 expression vector inhibited Rex-1 promoter activity to a greater extent than transfection of either Oct-3/4 or Oct-6 separately (Fig. 7C).

Since the Rox-1 binding site was found to participate in the Oct-3/4 activation of the Rex-1 promoter in P19/RA cells, we wished to ascertain whether it played a role in Rex-1 repression by the Oct-3/4 protein. F9 cells were transfected with pRox*Octa-CAT with and without the Oct-3/4 expression vector. Surprisingly, mutation in the Rox-1 binding site did not affect the ability of Oct-3/4 to down-regulate Rex-1 promoter activity (Fig. 6A). Thus, the Rox-1 binding site is not essential for Oct-3/4 to repress Rex-1 promoter activity in F9 cells but is crucial for activation of Rex-1 promoter by Oct-3/4 in P19/RA cells, as described above (Fig. 5).

Deletion mapping of the Oct-3/4 trans-repression regions.

It was previously shown that the N-terminal proline-rich region of Oct-3/4 functions as a transcriptional activating domain (22, 40). More recently it was shown that a transcriptional activation domain resides also in the C-terminal region of Oct-3/4 (70). To map the domain(s) of Oct-3/4 involved in regulation of the Rex-1 promoter, we generated four Oct-3/4 constructs: ΔN35, which lacks the first 35 amino acids; ΔN126, which lacks 126 amino acids from the N-terminal region; ΔC75, a deletion of the 3′ 75 amino acids which constitute almost the entire carboxy-terminal domain; and ΔDB, in which the binding domains located between amino acids 182 and 336 were deleted (Fig. 8A). The first three deletion mutants have an intact POU domain and thus were checked for the approximate protein size, stability, and amount by EMSA of COS-1 extracts transfected with the appropriate expression vector. ΔDB was analyzed by Western blotting using polyclonal rabbit anti-mouse Oct-3/4. All constructs express Oct-3/4 mutant proteins of the expected sizes at high levels (data not shown). We also analyzed plasmid 5′P, which harbors a deletion in the N-terminal proline-rich region from Pro13 to Pro60 (70). The transcriptional repression potential of the mutant Oct-3/4 proteins was tested in cotransfection experiments with the p0.3Rex-CAT reporter gene in F9 stem cells (Fig. 8A). Deletion of the first 60 amino acids did not reduce the ability of Oct-3/4 to repress the Rex-1 promoter, whereas deletion of the 126 amino acids from the N terminus abolished repression. The C-terminus-truncated Oct-3/4 mutant (ΔC75) was able to repress Rex-1 promoter activity in F9 cells. Thus, the Oct-3/4 C-terminus region is dispensable for Rex-1 repression. The ΔDB expression vector, containing the N- and C-terminal domains of Oct-3/4 but lacking the DNA binding domain, not only was unable to repress Rex-1 promoter activity but, rather, activated it twofold. It most probably acts similarly to a dominant negative gene. Therefore, we concluded that the N-terminal portion of Oct-3/4, located between amino acids 61 and 126, and the DNA binding domain are required for the down-regulation of the Rex-1 promoter in F9 cells.

FIG. 8.

Deletion mapping of the repressive region of Oct-3/4 and Oct-6. (A) F9 stem cells were cotransfected with p0.3Rex-CAT reporter construct (10 μg) and a β-Gal-containing reference plasmid (1 μg) in the absence (−) or presence of 10 μg of wild-type Oct-3/4, N-terminal-deleted Oct-3/4 (ΔN35, 5′P, and ΔN126), C-terminus-deleted Oct-3/4 (ΔC75), and DNA-binding-domain-deleted Oct-3/4 (ΔDB). CAT activity was measured and quantitated as described in the legend to Fig. 6. The values for percent CAT conversion, presented as means ± standard deviation, corresponding to p0.3Rex-CAT in the absence or presence of Oct-3/4, ΔN35, 5′P, ΔN126, ΔC75, and ΔDB are 34.8 ± 5.1, 6.8 ± 0.7, 5.4 ± 0.6, 43.86 ± 4.5, 6.5 ± 0.7, and 69.4 ± 7.2, respectively. (B) P19/RA cells were cotransfected with 0.3Rex-CAT reporter constructs (10 μg) and a β-Gal-containing reference plasmid (1 μg) in the absence (−) or presence of the expression plasmids described above. The values for percent CAT conversion, presented as means ± standard deviation, corresponding to p0.3Rex-CAT in the absence or presence of Oct-3/4, ΔN35, ΔN126, ΔC75, ΔDB are 8.5 ± 1.1, 49.3 ± 5.1, 5.1 ± 0.4, 6.8 ± 0.7, 48.45 ± 6.3, and 3.4 ± 0.6, respectively. (C) F9 stem cells were cotransfected with p0.3Rex-CAT reporter construct (10 μg) and a β-Gal ΔDB-containing reference plasmid (1 μg) in the absence (−) or presence of 10 μg of wild-type Oct-6, N-terminus-deleted Oct-6 (N157 or N229), C-terminus-deleted Oct-6 (N2C52), and DNA-binding-domain-deleted Oct-6 (N229C52). These vectors were previously described (35). The values for percent CAT conversion, presented as means ± standard deviation, corresponding to p0.3Rex-CAT in the absence or presence of Oct-6, N157, N229, N2C52, and N229C52 are 18.2 ± 0.3, 2.5 ± 0.3, 12.3 ± 0.1, 12.3 ± 0.2, 1.2 ± 0.2, and 16.6 ± 0.3, respectively.

Deletion mapping of the Oct-3/4 activation domains.

We have mapped the domain(s) of the Oct-3/4 protein involved in this trans activation of the Rex-1 promoter in P19/RA cells by using the same set of Oct-3/4 mutants as described above. Similarly to the repression of the Rex-1 promoter in undifferentiated F9 cells, the C-terminal region is dispensable for Rex-1 activation. Activation of the Rex-1 promoter in the P19/RA cells requires the DNA binding domain as well (Fig. 8B). However, in contrast to the data described above showing that the first 35 N-terminal amino acids are dispensable for Rex-1 repression by Oct-3/4, these amino acids play a crucial role in the ability of Oct-3/4 to activate the Rex-1 promoter.

Deletion mapping of the Oct-6 trans-repression domains.

To map the domain(s) of Oct-6 involved in Rex-1 repression, we used a set of deletions obtained from D. Meijer (36). As can be seen in Fig. 8C, deletion of 157 amino acids from the N terminus considerably reduced the ability of Oct-6 to repress Rex-1 promoter activity. In contrast, the N2C52 deletion, in which the entire carboxy-terminal domain is removed, displayed a wild-type level of repression. The POU domain alone (N229C52) did not influence Rex-1 promoter activity. Thus, the C-terminal region of Oct-6 is dispensable for Rex-1 repression, whereas the N-terminal region and most likely the DNA binding domain are necessary to mediate the inhibition of Rex-1 promoter activity.

Oct-3/4 specifically regulates the Rex-1 promoter.

Since Oct-3/4 is the major POU-specific protein expressed in the embryo ICM, we chose to concentrate on assessing the specificity of Rex-1 promoter repression by Oct-3/4. To assess the specificity of Rex-1 repression, we cotransfected p0.3Rex-CAT with increasing amounts of Oct-1 expression vector (provided by W. Herr) into undifferentiated F9 and P19/RA cells. The results (Fig. 9A) clearly indicate that whereas 5 μg of Oct-3/4 inhibits Rex-1 activity in F9 cells, the Oct-1 expression vector (1 to 10 μg) did not affect CAT activity driven by the Rex-1 promoter. Similarly, whereas Oct-3/4 activates the Rex-1 promoter in P19/RA (Fig. 5), Oct-1 (1 to 10 μg) did not have a significant effect (although Oct-1 activated an Octa-dependent luciferase reporter expression [data not shown]).

FIG. 9.

Oct-3/4 specifically regulates the Rex-1 promoter. (A) F9 and P19/RA cells were transiently transfected with the reporter plasmid p0.3Rex-CAT (10 μg), a β-Gal-containing reference plasmid (1 μg), in the absence (−) or presence of the indicated increasing amounts of Oct-1 expression vector or Oct-3/4 (5 μg) expression vector. Transfection efficiency and CAT activity were monitored and assayed as described in the legend to Fig. 6. The values for percent CAT conversion in F9 cells, presented as means ± standard deviation, corresponding to p0.3Rex-CAT in the absence or presence of 1, 2.5, 5, and 10 μg of Oct-1 are 37.2 ± 4.1, 34.3 ± 3.5, 45.0 ± 5.0, 40.2 ± 6.1, and 55.4 ± 6.5, respectively; the value in the presence of 5 μg of Oct-3/4 is 6.2 ± 0.5. The values for percent CAT conversion in P19/RA cells (a representative experiment) corresponding to p0.3Rex-CAT in the absence or presence of 1, 2.5, 5, and 10 μg of Oct-1 are 9.0, 9.7, 9, 6.5, and 5.3, respectively. (B) p0.3Rex-CAT and pPκEκ-CAT (containing the Ig κ-chain promoter and intronic enhancer) reporter plasmids were transfected with β-Gal-containing reference plasmid into F9 and P19/RA cells in the absence (−) or presence (+) of the Oct-3/4 expression vector. The values for percent CAT conversion in F9 cells, presented as means ± standard deviation, corresponding to p0.3Rex-CAT and pPκEκ-CAT each in the absence or presence of Oct-3/4 are 52.8 ± 0.6, 10.2 ± 1.2, 5.2 ± 0.6, and 6.0 ± 0.7, respectively. The values for percent CAT conversion in P19/RA cells (a representative experiment) corresponding to p0.3Rex-CAT and pPκEκ-CAT each in the absence or presence of Oct-3/4 are 8.1, 47.7, 2.95, and 3.82, respectively.

To determine whether Oct-3/4 can regulate the activity of any promoter harboring an octamer site, we cotransfected a CAT reporter plasmid driven by the immunoglobulin (Ig) κ-chain promoter (pPκEκ [62]) with the Oct-3/4 expression vector. As can be seen in Fig. 9B, in contrast to the Rex-1 promoter, which was inhibited by Oct-3/4 in F9 cells and activated by Oct-3/4 in P19/RA cells, the κ promoter was not regulated by the Oct-3/4 expression vector under the same conditions. Similarly, a synthetic promoter containing three tandem repeats of an octamer motif in front of a TATA minimal promoter (px3-Octa-CAT) was also affected by Oct-3/4 (data not shown). The specificity of the response suggests that (i) the effect of Oct-3/4 on Rex-1 promoter activity did not result from general blockage of RNA-polymerase II-dependent transcription machinery and (ii) there are additional important elements in the Rex-1 promoter that collaborate with Oct-3/4 in regulating Rex-1 activity.

Excess of N-terminal or DNA-binding-domain-deleted Oct-3/4 activates theRex-1 promoter in F9 cells.

Since wild-type Oct-3/4 specifically represses Rex-1 promoter in F9 cells, which express high levels of endogenous Oct-3/4 protein, and activates it in differentiated P19/RA cells, which lack endogenous Oct-3/4, we decided to study how modulation of the endogenous Oct-3/4 protein in F9 cells affects the transfected Rex-1 promoter activity. We cotransfected a constant amount of p0.3Rex-CAT construct with increasing amounts of wild-type, ΔN126 or ΔDB Oct-3/4 expression vector. While the entire wild-type Oct-3/4 represses Rex-1 promoter activity in F9 cells, the ΔN126 and ΔDB Oct-3/4 expression vectors were shown to be inactive when 10 μg of DNA was transfected (Fig. 10A). However, higher amounts of both Oct-3/4 deletion plasmids activate the Rex-1 promoter in F9 cells. Therefore, we conclude that the ΔN126 and ΔDB Oct-3/4 expression vectors may act as dominant negative proteins, effectively lowering the level of the endogenous Oct-3/4 transcription factor in F9 cells. This results in the activation of the transfected Rex-1 promoter (Fig. 10A).

FIG. 10.

Modulation of Oct-3/4 activity. (A) ΔN126 and ΔDB expression vectors activate Rex-1 promoter in F9 cells. F9 cells were transiently transfected with the p0.3Rex-CAT reporter plasmid and a β-Gal-containing reference plasmid (1 μg) in the absence (−) or presence of 10 μg (grey bars) and 20 μg (black bars) of Oct-3/4, ΔN126, and ΔDB expression vectors. The values for percent CAT conversion, presented as means ± standard deviation, corresponding to p0.3Rex-CAT in the absence or presence of 10 and 20 μg of Oct-3/4, ΔN126, and ΔDB are 37.2 ± 4.1, 18.2 ± 0.3, 5.46 ± 0.7, 3.87 ± 0.4, 31.43 ± 3.0, 53.14 ± 4.9, 52.37 ± 5.1, and 58.74 ± 6.1, respectively. (B) Effect of E1A on Rex-1 expression. F9 cells were transiently transfected with either the p0.3Rex-CAT or pRoxOcta*-CAT reporter plasmid and a β-Gal-containing reference plasmid (1 μg) in the absence (−) or presence of the indicated E1A and Oct-3/4 expression vectors. The values for percent CAT conversion, presented as means ± standard deviation, corresponding to p0.3Rex-CAT are as follows: in the absence of Oct-3/4 and E1A, 27.1 ± 0.3; in the presence of Oct-3/4, 7.6 ± 0.8; in the presence of 2.5, 5, and 10 μg of E1A, 5.9 ± 0.7, 3.2 ± 0.4, and 2.1 ± 0.2, respectively; in the presence of Oct-3/4 and 2.5, 5, and 10 μg of E1A, 2.6 ± 0.3, 1.6 ± 0.2, and 1.0 ± 0.2, respectively. The values for percent CAT conversion, presented as means ± standard deviation, corresponding to pRoxOcta*-CAT in the absence or presence of E1A are 1.3 ± 0.2 and 1.56 ± 0.2, respectively.

In F9 cells, E1A represses the Rex-1 promoter through the Octa site.

It was previously shown that E1A represses the IgH enhancer, containing an octamer motif, in lymphoid cells, but activates it in fibroblasts (4, 7, 18). More recently, it has been shown that Oct-3/4 and E1A can bind to each other and stimulate transcription synergistically from a distance (57). Given this information, we decided to assess the effect of E1A on repression of Rex-1 activity. We cotransfected the p0.3Rex-CAT construct with increasing amounts of E1A expression vector (Fig. 10B). CAT activity was progressively inhibited by increasing amounts of E1A. Thus, in F9 cells the Rex-1 promoter is inhibited by E1A. Moreover, this inhibition is mediated through the octamer site, since mutations of this element in the pRoxOcta*-CAT construct completely abolished inhibition of Rex-1 promoter activity by E1A. We also studied the effect of E1A on the ability of Oct-3/4 to repress the Rex-1 promoter. We cotransfected F9 cells with the p0.3Rex-CAT reporter construct, a constant amount of Oct-3/4 (5 μg), and increasing amounts of E1A. The results show that the magnitude of the inhibition of the Rex-1 promoter by Oct-3/4 was increased by the addition of E1A (Fig. 10B). Thus, E1A can enhance the ability of Oct-3/4 to down-regulate the Rex-1 promoter in F9 cells.

DISCUSSION

Rox-1 binding site activity.

We have demonstrated that the regulation of Rex-1 promoter activity through the octamer site at position −220 is dependent on the presence of specific members of the octamer protein family. While an expression vector containing the Oct-1 protein did not influence CAT activity driven by the Rex-1 promoter, both Oct-3/4 and Oct-6 expression vectors regulated the Rex-1 promoter. Regulation of the Rex-1 promoter by the Oct-3/4 and Oct-6 proteins most likely depends on sequences in addition to the octamer site, since activities of promoters that contain the octamer motif, such as the Ig κ-chain promoter or a synthetic octamer-containing promoter, were not regulated by the Oct-3/4 expression vector under the same conditions. Indeed, through the use of DNase I footprinting, gel shift assays, and mutation analyses, we identified a protein, designated Rox-1, that binds to the sequence TCAGAAGAGGC, located immediately 5′ of the octamer site. While this sequence of the Rox-1 element has some similarity to ETS binding sequences (69, 73), a PEA3/ETS sequence (CCAGGAAGTGAC) did not compete for the Rox-1 element (data not shown). The Rox-1 binding site is required for a high level of Rex-1 promoter activity in F9 stem cells. The novel factor Rox-1 displays several interesting properties. Rox-1 binding activity is not observed in various cell types such as fibroblasts, B, T, insulinoma, and hepatoma cells, cells of the myeloid lineage such as HL60, and NIH-9 cells. This finding is of interest because it suggests that the Rox-1 protein may be specifically required for Oct-3/4 to function in undifferentiated, early embryonic cell types. It is also of interest that the Rox-1 binding activity is reduced in cellular extracts from RA-treated F9 cells, again suggesting that the expression of the Rox-1 protein may be a critical feature of Rex-1 promoter regulation in early embryonic cells.

Oct-3/4 and Oct-6 repression of the Rex-1 promoter.

We have shown that in F9 cells which express high levels of endogenous Oct-3/4 and Oct-6, the exogenously transfected Oct-3/4 and Oct-6 repress Rex-1 promoter activity. This repression of Rex-1 promoter activity is in apparent conflict with the deletion analyses published by Hosler et al. (21), in which mutation or deletion of the octamer site resulted in a large decrease in Rex-1 promoter activity. However, in the study by Hosler et al. (21), the activity of the transiently transfected Rex-1 promoter was assessed in the presence of the endogenous Oct-3/4 and Oct-6 proteins, and no additional Oct-3/4 or Oct-6 proteins were added to the cells. Further experiments have provided evidence that the addition of high levels of exogenous Oct-3/4 and Oct-6 proteins via transient transfection into the F9 stem cells, which already express high levels of endogenous Oct-3/4 and Oct-6 proteins, results in the inhibition of the cotransfected reporter gene driven by the Rex-1 promoter. Oct-3/4 and Oct-6 proteins exert this repression through the Octa sequence, not via the Rox-1 binding site. In accordance with the ability of Oct-3/4 and Oct-6 to inhibit Rex-1 promoter activity, it was shown that these POU proteins inhibit the expression of a number of additional promoters such as the human chorionic gonadotropin (30), the involucrin (71), and the myelin Po gene promoters (17, 39).

Repression of transcription at RNA polymerase II promoters may occur by multiple mechanisms involving either steric hindrance or inhibitory protein-protein interactions (reviewed in references 23 and 27). Our results argue against repression mechanisms involving either displacement of an octameric bound activator protein or repression of basal transcription machinery. Since promoters such as the κ-chain promoter or a synthetic promoter containing the Octa oligonucleotide are not repressed by the Oct-3/4 or Oct-6 protein, our data suggest that Oct-3/4 and Oct-6 repress the Rex-1 promoter by either a quenching or a squelching mechanism. The ability of Oct-3/4 to interact with other proteins such as Oct-1 (70) suggests that the regulatory properties of Oct-3/4 could be influenced by the presence of other interacting proteins within the cell. Thus, it is also possible that Oct-3/4 forms a complex with an as yet unidentified transcription factor (different from Rox-1) which binds to the Rex-1 promoter and that such dimers gain novel repressing functions which differ from the individual function of each partner. Our data (Fig. 10A), indicating that transfection of ΔDB into F9 cells results in activation of Rex-1 promoter, support this possibility. We have shown that repression of the Rex-1 promoter is achieved through the joint action of Oct-3/4 amino-terminal and POU domains. Repressors have not been characterized to the same extent as activators, although recent assays have identified discrete repression domains (12, 14, 15, 28). No common signature has emerged. Interestingly, the first 35 amino acids of Oct-3/4 that contain the activation domain are rich in glutamine, glycine, and alanine, whereas amino acids 61 to 126, which harbor the repression domain, are proline rich (40).

Oct-3/4 activation of the Rex-1 promoter.

In contrast to the ability of exogenous Oct-3/4 to repress Rex-1 expression in F9 cells, in P19/RA cells, in which expression of Oct-3/4 is not detectable (40), exogenous Oct-3/4 activates the wild-type Rex-1 promoter. Mutations either in the Octa site or in the Rox-1 binding sequences compromise the ability of Oct-3/4 to activate the Rex-1 promoter. Oct-3/4 protein possesses the ability to enhance expression of several additional genes, including the platelet-derived growth factor α receptor (PDGFαR [25]) and fibroblast growth factor (FGF-4) genes (10, 31, 54, 75). Interestingly, nucleotides juxtaposed to the octamer motif were found to play a key role in positively regulating the expression of two additional cellular genes, encoding PDGFαR and FGF-4, both of which are controlled by the Oct-3/4 gene product in EC cells (1, 10, 25, 75). Transcriptional activation of the FGF-4 gene depends on a synergistic interaction between the Oct-3/4 and Sox2 (a member of the Sry-related Sox factors family) proteins, both binding sequences located very close to each other. The Sox2 binding sequence, CTTTGTT, differs from the Rox-1 site sequence. The Oct-3/4 and Sox2 transcription factors form a ternary complex with the FGF-4 enhancer sequences (75). In contrast to these data, we could not show a ternary complex between the Rox-1 protein and either Oct-1 or Oct-3/4 and the Rex-1 promoter sequences, using EMSA. However, our preliminary data indicate that Rox-1 binds preferentially to the RoxOcta sequence containing a wild-type octameric motif, compared to a RoxOcta* oligomer harboring a mutated Octa sequence; this finding suggests that Oct-3/4 and Rox-1 protein may cooperatively bind to the Rex-1 promoter sequences (data not shown). Through future cloning and characterization of the Rox-1 gene, we will be able to approach this question in a more direct manner. Analysis of the functions of the Rox-1 protein should provide new insights into the regulation of genes in early ES cells, into the earliest differentiation events of the morula, which forms the trophectoderm and the ICM, and into the roles of retinoids such as RA in the embryo.

Oct-3/4 as a dual regulator of a promoter.

Since Oct-3/4 protein possesses both activating and repressive potentials (26, 40, 57), we interpret our results to indicate that the level of expression of Oct-3/4 protein is critical with respect to whether the Oct-3/4 protein will activate or inhibit the Rex-1 promoter. Our data suggest that Oct-3/4 protein at low levels activates the Rex-1 promoter and that high levels of Oct-3/4 repress Rex-1 promoter activity. Our data also correlate with the in situ hybridization experiment results, showing that Oct-3/4 expression is indeed higher in 6.5 day embryos than in 4.5 day blastocysts (42), whereas the opposite is true for the Rex-1 gene, which is expressed in the embryo ICM up to day 4.5 (47). Thus, it is possible that in vivo under physiological conditions, Oct-3/4 behaves as both an activator and a repressor, most likely depending on the level of its expression. Precedence to dose-dependent regulation has been previously reported. For example, the Drosophila Krüppel (29, 51, 52, 53) and ATF-1 (46) transcription factors were shown to both activate and repress gene expression in a concentration-dependent manner.

The fact that the function of Oct-3/4 depends on the cellular environment also suggests that the key to understanding how the Oct-3/4 protein activates or represses transcription of cellular genes will also rely on the identification of additional cell-restricted activities, such as the B-cell-specific coactivator OCA-B/OBF-1/BOB1, which can specifically activate transcription through interactions with either Oct-1 or Oct-2 (13, 64), or the OTX-related homeobox transcription factor that interacts with Pit-1 (67). It is possible that Oct-3/4 can switch between an activator and a repressor due to its interactions with auxiliary adapter proteins which Oct-3/4 recruits to the Rex-1 promoter. Such an auxiliary activator is the adenovirus E1A protein, which is needed for Oct-3/4-activating transcription through distal octamer sites. Since E1A and Oct-3/4 can bind to each other, E1A probably serves as a bridging factor between Oct-3/4 and the basal initiation complex (57). Indeed, we show that E1A enhances the ability of Oct-3/4 to repress the Rex-1 promoter. It is possible that multiple E1A-like cellular transcription factors exist in different cell types, and they may determine whether Oct-3/4 activates or represses the Rex-1 promoter. In line with this suggestion, it was shown that other dual-function regulators are induced to switch activity by interacting with either coactivators or corepressors, such as the Par-4 and p53 proteins, which modulate the activity of the tumor suppressor WT1 protein, and the N-CoR or SMRT proteins, which interact with the thyroid hormone and RA receptors (24, 32; reviewed in reference 43).

Our analysis of the Rex-1 promoter has defined a compound element which contains both the octamer and the Rox-1 binding sites. The specificity of the expression of this gene in early embryogenesis is achieved via the specific functions provided by combination of the Oct-3/4 and Rox-1 factors. Activation and repression of the Rex-1 gene by Oct-3/4, which depends on the promoter architecture, cellular environment, and amount of this octamer-binding protein, may have important consequences for the early stages of development.