Oct-4: Gatekeeper in the Beginnings of Mammalian Development


  • Maurizio Pesce,

    1. Laboratorio di Patologia Vascolare, Istituto Dermopatico dell' Immacolata, Rome, Italy
    Search for more papers by this author
  • Hans R. Schöler Ph.D.

    Corresponding author
    1. Center for Animal Transgenesis and Germ Cell Research, Department of Animal Biology, School of Veterinary Medicine, University of Pennsylvania, Kennett Square, Pennsylvania, USA
    • Director, Center for Animal Transgenesis and Germ Cell Research, Germline Development Group, New Bolton Center, 382 W. Street Rd., Kennett Square, Pennsylvania 19348, USA. Telephone: 610-444-5800×2289; Fax: 610-925-8121
    Search for more papers by this author


The Oct-4 POU transcription factor is expressed in mouse totipotent embryonic stem and germ cells. Differentiation of totipotent cells to somatic lineages occurs at the blastocyst stage and during gastrulation, simultaneously with Oct-4 downregulation. Stem cell lines derived from the inner cell mass and the epiblast of the mouse embryo express Oct-4 only if undifferentiated. When embryonic stem cells are triggered to differentiate, Oct-4 is downregulated thus providing a model for the early events linked to somatic differentiation in the developing embryo. In vivo mutagenesis has shown that loss of Oct-4 at the blastocyst stage causes the cells of the inner cell mass to differentiate into trophectoderm cells. Recent experiments indicate that an Oct-4 expression level of roughly 50%-150% of the endogenous amount in embryonic stem cells is permissive for self-renewal and maintenance of totipotency. However, upregulation above these levels causes stem cells to express genes involved in the lineage differentiation of primitive endoderm. These novel advances along with latest findings on Oct-4-associated factors, target genes, and dimerization ability, provide new insights into the understanding of the early steps regulating mammalian embryogenesis.

Oct-4: The Regulator of Totipotency in the Mouse Embryo

POU transcription factors were originally identified as DNA-binding proteins that are able to activate the transcription of genes bearing cis-acting elements containing an octameric sequence called the octamer motif, within their promoter or enhancer region. The consensus-binding motif recognized by POU factors is the sequence ATGCAAAT that was originally found as an element controlling the immunoglobulin heavy chain enhancer [1].

The POU domain is a bipartite DNA-binding domain present in all POU proteins. It consists of two subdomains, called the POU-specific (POUS) and the POU homeo-domain (POUHD), which are connected by a flexible linker, variable in length. Flexibility of the linker region engendered between the two subdomains enables the POUS and the POUHD to contact the DNA-binding site independently of each other. Due to the particular configuration of the two POU subdomains, POU proteins have an intrinsic ability to adopt several binding configurations on the DNA. This results in an exceptional transactivation flexibility and interaction with different sets of coactivators [1]. In addition, POU factors possess an intriguing capability to form homo- and heterodimers that can bind to octamer motif variants (see below).

Mouse Oct-4 is a 352 amino acid protein belonging to class V of POU proteins. It is expressed in totipotent embryonic cells [2]. Initially expressed in all blastomeres of the developing mouse embryo, Oct-4 gene expression becomes restricted to the inner cells of the blastocyst forming the inner cell mass (ICM), and is downregulated in the trophectoderm and the primitive endoderm. Later in development, Oct-4 expression is maintained in the embryonic ectoderm at the egg-cylinder stage and is downregulated at gastrulation in an anterior-posterior pattern. The only cells maintaining Oct-4 expression after this stage are primordial germ cells (PGCs) arising within the extraembryonic mesoderm at 7.2 days of mouse embryo development. Germ cells maintain the expression of Oct-4 until the initiation of sexual differentiation of the gametes and meiosis, in the female at day 13-14 post-coitum (dpc) and at the beginning of spermatogenesis in the newborn males [3].

Oct-4 ortholog genes share a high degree of genomic structural organization and sequence conservation and have been identified in other mammalian species including human, bovine, and rat [3]. Sequence comparison of the promoter/ enhancer regions of the mammalian Oct-4 genes with that of the mouse ortholog revealed a common organization of cis-regulatory elements that are supposed to function in a similar fashion during embryonic development [4]. The analysis of the endogenous Oct-4 protein expression shows that in bovine and porcine preimplantation embryos, Oct-4 is not restricted to totipotent cells [5, 6]. In addition, it was found that the expression of green fluorescent protein controlled by the promoter/enhancer regions of the murine Oct-4 gene microinjected into fertilized bovine and porcine oocytes, was not restricted to cells of the ICM [6]. In contrast, the expression of human Oct-4 is comparable to the pattern in the mouse, suggesting that it may have a similar function in preventing human totipotent embryo cells from differentiating [7].

The unique Oct-4 expression pattern in the mouse embryo has led to the hypothesis of the “totipotent cycle” [8]. According to this postulate, cells losing Oct-4 during embryonic development differentiate into somatic lineages whereas cells maintaining Oct-4 expression retain totipotency and have the competence to develop into germ cells.

Oct-4 function has recently been abolished by homologous recombination [9]. Oct-4–/– embryos die at the time of implantation due to a failure to form the ICM. In vitro culture of Oct-4–/– embryos revealed that the loss of Oct-4 changes the fate of cells destined to become ICM cells to differentiate to a trophectoderm lineage. Interestingly, suppression of Oct-4 function does not influence the maintenance of the totipotent phenotype of the blastomeres prior to the formation of the blastocyst. This suggests that Oct-4-dependent transactivation in the early mouse embryo begins to be essential for maintaining a totipotent cell phenotype when the first somatic lineage (the trophectoderm) splits from the totipotent compartment (the ICM). Thus, the decision to maintain Oct-4 in the inner cells of the compacted morula and downregulate it in the outer cells appears to be one of the crucial events that enables the proper development of the preimplantation embryo [10] (Fig. 1).

Figure Figure 1..

Events in the formation of the mouse blastocyst with respect to Oct-4 expression levels.A) Prior to compaction, all the cells of the morula express similar amounts of Oct-4 protein (orange color). B) The formation of the trophoblast tissue is accompanied by downregulation of Oct-4 in the outer cells (yellow). C) Differentiation of primitive endoderm cells is preceded by transient upregulation of Oct-4 and subsequent shutdown (red).

It has not yet been demonstrated whether Oct-4 functions in the same manner at other stages of mouse development, that involve a decision between the trigger to differentiate and the escape from differentiation, such as the formation of the epiblast and the segregation of germ cells. However, it can be predicted that loss of Oct-4 function at these crucial stages may cause epiblast cells and germ cells to exit the totipotent cycle and be abolished from the embryos.

Oct-4 has recently been purported to have an even more important function in early mouse embryonic development. It was shown that it is not only involved in the maintenance of totipotency and self-renewal of the stem cell population of the preimplantation embryo, but it may also play a role in the initiation of the pathways controlling the early differentiation of the primitive endoderm from totipotent ICM cells (Fig. 1).

In a study examining Oct-4 protein expression at early stages of mouse development, it was reported that Oct-4 is transiently increased in the primitive endodermal cells (EC) of the blastocyst, prior to final downregulation [11]. The function of Oct-4 transient upregulation in primitive EC was suggested from the evidence that Oct-4 is implicated in the regulation of the Osteopontin gene (OPN), encoding an extracellular matrix component involved in cellular migration. Botquin et al. [12] found that retinoic acid treatment of F9 EC cells, that are usually used as a model for differentiation of the primitive endoderm, results in a transient increase of Oct-4, paralleled by an increased expression of OPN. A search for Oct-4 potential binding sites within the cis-regulatory sequences of the OPN gene identified a novel palindromic sequence named palindromic-oct-regulatory-element (PORE) within the first intron of OPN. This sequence contains a consensus octamer motif and an inverted half-site CAAAT spaced by two nucleotides to which Oct-4 binds in vivo as a monomer and a homodimer [12]. A comparison of the transcriptional activities of the mono- and dimeric configurations was performed by transient transfections of reporter vectors bearing mutated OPN Oct-4-binding sites of which one binding configuration was abolished leaving the other unaffected. This analysis revealed that the Oct-4 dimer plays a major role in OPN enhancer activity whereas the monomer has only weak effects [12]. This in turn suggests that the canonical octamer motif is not sufficient, per se, to create a strong Oct-4 activation site unless it is enclosed in a sequence context allowing Oct-4 to crosstalk with appropriate coactivators or form dimers (see below).

Recently, a strategy using inducible Oct-4 expression coupled to homologous recombination of the endogenous Oct-4 gene was applied to demonstrate dose-dependent phenotypic effects of Oct-4 in embryonic stem (ES) cells. Niwa et al. [13] introduced a tetracycline-tTA Oct-4 transgene into ES cells in which one copy of the endogenous Oct-4 gene was knocked-out [14]. Transfection of the cells in the absence of tetracycline (Tc) caused up to a twofold increase in Oct-4 expression relative to endogenous levels. Under these conditions, ES cells differentiated and formed clones of primitive endoderm cells as shown by upregulation of the GATA-4 gene [15]. A targeting of the second endogenous copy of Oct-4 was then performed in the same cells. Thus, ES cells were obtained which expressed a sufficient level of Oct-4 to maintain totipotency due to the introduced transgene. In addition, the possibility of repressing the expression of the transgene by treating the cells with different amounts of Tc provided a precise evaluation of the minimal Oct-4 level required for the maintenance of stem cell totipotency. As in the case of Oct-4–/– embryos [9], loss of Oct-4 caused ES cells to differentiate into trophectoderm. Interestingly, this only occurred when less than 50% of the normal expression level of Oct-4 was achieved by treating the cells with increasing amounts of Tc. Altogether, these results suggest that either an increase above 150% or a decrease below 50% of the endogenous Oct-4 levels can serve as a trigger for the differentiation of the two somatic lineages in mouse preimplantation embryos [16] (Fig. 1).

What Lies Upstream?

Restriction of Oct-4 expression during early mouse embryogenesis appears to be regulated by a positional effect which is likely due to the establishment of a differential gene expression pattern in the preimplantation embryo. Only cells that are located in the ICM of the blastocyst seem to have the competence to maintain Oct-4 expression, whereas cells that are excluded from this position are fated to differentiate into trophectoderm. In addition, cells of the ICM that differentiate into primitive endoderm transiently upregulate Oct-4 and activate a gene expression cascade distinct from that of the outer cells.

A unique cellular identity of the blastomeres is maintained until compaction of the morula. At this stage, an inner compartment of cells is segregated from the outer layer of cells that will form the trophectoderm. It has been found that a differential gene expression pattern in the outer cells of the compacted morula is established prior to the differentiation of these cells into trophectoderm [17]. One possibility is that this gradient may be due to the different polarity and cell contacts established between outer and inner cells of the compacted morula. Therefore, genes like Oct-4 that are initially equally expressed in all cells of the embryo may be repressed as a consequence of specific cell-cell adhesion, thereby allowing the upregulation of genes involved in the formation of the trophectoderm.

Another possible explanation as to the differences in Oct-4 expression between inner and outer cells of the preimplantation embryo suggests that repressors that are unequally distributed in the oocytes may be differentially inherited by inner and outer blastomeres during cleavage. However, the only report showing that a similar mechanism may act in the mouse was the observation that a gradient of the leptin/STAT-3 proteins is formed in the mouse oocyte and may be conserved throughout the early cleavage stages [18]. This makes it unlikely that differential expression of Oct-4 in the compacted morula may be regulated by an unequal distribution of determinants laid down in the oocyte.

Recent observations have suggested that the main symmetry of the mouse embryo could be laid down long before the morphological appearance of the antero-posterior axis at gastrulation. It has been shown that the position of the second polar body that is formed after fertilization and remains associated to the embryo up to the blastocyst stage correlates with the prospective bilateral symmetry axis of the mouse embryo [19]. An asymmetrical distribution or proliferation of cells forming the nascent visceral endoderm (VE) between the region of the ICM close to the polar body and the region away from it is established as early as day 5.5 dpc [20]. Interestingly, VE cells arising from cells of the ICM close to the polar body cover the embryonic ectoderm and migrate from the prospective posterior pole towards the anterior whereas VE cells originating from distal ICM cells mainly cover the extraembryonic ectoderm [20]. VE cells covering the epiblast and migrating towards the anterior have been found to express a number of genes that are supposed to play a role in the establishment of the A-P axis prior to the beginning of gastrulation [21]. Therefore, the finding that the VE is formed according to an A-P fashion prior to 5.5 dpc suggests that the overall A-P structure of the embryo could be established during preimplantation development.

How does Oct-4 fit into this process? Up to now there is no evidence supporting the notion that Oct-4 could be differentially expressed in individual cells of the ICM. However, the transient upregulation of Oct-4 protein [11] that is now correlated to OPN overexpression [12] and initiation of the genetic program of the primitive endoderm differentiation [13, 16] strongly suggest that this may be the case. Since in situ hybridization studies have failed to detect differences in Oct-4 mRNA levels in the ICM [11], it is likely that post-translational mechanisms and/or an increased stability of the Oct-4 protein may be involved in Oct-4 regulation.

What We Know AboutOct-4 Regulation

Oct-4 mRNA is present as a maternal transcript in fully grown oocytes. A low level of Oct-4 protein is found in the blastomeres at the early cleavage stages until the eight cell-stage, when a high amount of newly synthesized Oct-4 protein appears in the nuclei of embryonic cells [11].

Transgenic mice and studies in ES cell lines have addressed Oct-4 regulation in stem cells and in developing mouse embryos. Analysis of transgenic mice carrying the LacZ reporter under the control of an 18 kb fragment from the Oct-4 genomic sequence allowed the identification of elements that play important roles in Oct-4 gene expression [22]. Two elements named proximal enhancer (PE) and distal enhancer (DE), based on their position with respect to the transcription initiation site, regulate the stem cell-specific activity of Oct-4. Mice carrying the Oct-4 transgene lacking a 1-kb fragment containing the PE did not express LacZ in the epiblast whereas deletion of a 3.2-kb fragment including the DE caused abolition of LacZ activity in the ICM and germ cells [22]. The differential in vivo activity of the Oct-4 enhancers is recapitulated in ES cells and EC that resemble cells of the ICM and the epiblast, respectively. The DE is active in ES cells whereas the PE is active in EC. An in vivo footprinting study has allowed the identification of the precise binding sites within the two enhancers that are bound by transcription factors [23]. One site named 1A was found to be located within the PE and another site named 2A was identified in the DE [23]. Strikingly, these sites exhibit nearly identical sequence homology to the GC-box, and are both protected in undifferentiated ES and EC cells, and released from such protection upon retinoic acid differentiation of either cell type [23]. Electrophoretic mobility shift assays have shown that both ES and EC cells express proteins that are able to bind these sites in an identical fashion [K. Hübner and H.R.S., unpublished]. Thus, although sites 1A and 2A are crucial for the activity of the PE and the DE, the occupancy data do not support their involvement in the stem cell-specific activities of the two enhancers in vivo. This suggests that other unknown cis-acting elements within the DE and PE act in concert with sites 1A and 2A in determining the specific activity of the two enhancers in different stem cell lines and during embryogenesis.

Epigenetic Modifications of the Chromatin Structure May Play a Role inOct-4 Expression and Downregulation During Embryonic Development

Oct-4 segregation in totipotent cells may reflect the presence of specific sets of transcriptional activators and chromatin remodeling factors. For example, the murine protein mbrm, that is homologous to the human brm, the Drosophila brahma, and the yeast SWI2, is confined to ICM cells of the mouse embryo [24]. Similar to the human, Drosophila, and yeast homologues, mbrm is supposed to be comprised in large chromatin remodeling complexes that are known to facilitate access of promoter and enhancer elements to activators, thereby enabling transcription. One intriguing possibility is that the maintenance of a steady-state expression level of Oct-4 in totipotent cells may be a consequence of the establishment of a general state of active chromatin configuration in these cells rather than an outcome of the function of specific Oct-4 activators. In this case, downregulation of Oct-4 in the differentiating trophectoderm could result from a negative perturbation of the stem cell basic Oct-4 activity.

Oct-4 gene expression is potentially affected by methylation although it is not subject to imprinting. Ben-Shushan et al. reported that Oct-4 gene activity extinction in stem cell × fibroblast hybrid cells is accompanied by a rapid methylation of CpG islands in the promoter and the PE [25]. Methylation of regulatory sequences such as the PE and the DE may play a role in Oct-4 shutdown occurring during gastrulation, when a wave of de novo methylation is reported to occur in the somatic cells of the embryo [26]. Segregation of PGCs into an extraembryonic tissue such as the extraembryonic mesoderm may prevent methylation of their genome and germ cell differentiation [8]. Again, this suggests that maintenance of Oct-4 expression in PGCs, and therein of mammalian germline totipotency, may be a consequence of germ cells escaping from general epigenetic reprogramming of the chromatin that occurs in epiblast cells at the time of gastrulation.

The Function of Oct-4: Activation/Suppression of Transcription, Dimerization Potential, and Functional Partners

The function of Oct-4 as a suppressor of differentiation in mammalian totipotent cells bears similarity to factors involved in the pre-set of a gene expression pattern restricted to a totipotent phenotype. In the C. elegans embryo, the Pie-1 zinc-finger protein [27] is already confined to the progenitor cell of the germline at the four-cell stage. All cells that lose Pie-1 expression during early cleavage stages are restricted to differentiate into somatic cell types, whereas Pie-1+ cells are fated to become germ cells and maintain totipotency. Pie-1 function appears to be involved in a general transcriptional repression mechanism although it has not yet been implicated in direct binding to DNA or with interference of RNA Pol II enzyme activity [28].

Oct-4 was reported to act as a repressor of specific target genes in two cases. Octamer motifs located within the proximal promoters of the α and β subunit of the human chorionic gonadotropin (hCG) genes mediate Oct-4 transrepression of both genes in choriocarcinoma cells [29, 30]. Alleviation of hCG gene repression due to Oct-4 downregulation in the outer cells of the compacted morula may be one of the initial events which establish a gene expression pattern leading to the trophectoderm lineage.

Repression is by no means the most effective or the only way of Oct-4 target gene regulation in stem cells. Oct-4-dependent transcriptional transactivation from either proximally or remotely located binding sites in totipotent cells has been reported. Distance-dependent Oct-4 transactivation is a unique property of embryonic cells that is mediated by specific sets of coactivators connecting a remotely bound Oct-4 molecule to the transcriptional machinery.

The adenoviral E1A protein was the first protein found to functionally interact with Oct-4 [31]. In differentiated cells, E1A substitutes for the stem cell-specific bridging factor(s) enabling distance-dependent transactivation. An interaction between the Oct-4 POU domain and the zinc-finger motif contained within the constant region 3 (CR3) of E1A is responsible for this functional cooperation [31]. In addition, we have recently found that regions other than CR3 contribute to the interaction between Oct-4 and E1A [32, 33]. E1A and E1A-like proteins in stem cells mediate Oct-4 distance transactivation by bridging Oct-4 molecules to the basal transcription machinery. It has been suggested that this activity does not require binding of this coactivator to DNA [31] (Fig. 2).

Figure Figure 2..

Three models of Oct-4 transactivation in stem cells.A) Distance-dependent transactivation is mediated by bridging factors connecting a remotely bound Oct-4 molecule and the transcription machinery. E1A, E7, and HMG-1 enable Oct-4 transactivation in this manner and do not bind to DNA. B) Cooperative binding of Oct-4 and Sox-2 to adjacent sequences localized in the 3 ′ -UTR of the FGF-4 and Utf1 genes enables transactivation via conformational changes of both the Sox-2 and Oct-4 transactivation domains due to protein-protein interactions. C) Oct-4 dimer formed on the PORE localized in the first intron of the OPN gene mediates distance-dependent transactivation by crosstalking with yet unknown coactivators.

Other models of Oct-4 transactivation are based on the synergistic action of factors that bind the DNA in the proximity of the octamer motif. Factors binding in the vicinity of Oct-4 are probably responsible for the activation of Oct-4 N- and C- transactivation domains through conformational changes. The best example of this type of Oct-4 functional modulation comes from the analysis of the stem cell-specific FGF-4 growth factor gene activity [4, 10]. An enhancer element localized within the 3′-untranslated region (UTR) is responsible for the stem cell-specific expression of FGF-4 [34]. The FGF-4 enhancer contains an octamer motif and an adjacent binding site to which Oct-4 and the high mobility group (HMG) transcription factor Sox-2 bind cooperatively and activate transcription synergistically [35]. In the absence of Sox-2, Oct-4 neither is sufficient for FGF-4 enhancer activity, even in the presence of E1A, nor does it even bind to the octamer motif [36]. It has been shown that Oct-4/Sox-2 synergism is mediated by direct protein-protein interaction that is dependent on the spacing between the Oct- and Sox-binding sites [37]. In addition, formation of an Oct-4/Sox-2 complex on the FGF-4 enhancer likely causes allosteric changes that are necessary for unmasking latent activation domains in both proteins leading to transcriptional activation [38]. Thus, Oct-4 and Sox-2-mediated transcriptional synergism depends on a reciprocal modeling of both proteins onto DNA in a binding site-dependent manner.

An Oct-4/Sox-2 complex positively regulates the expression of another stem cell-specific gene. Utf1 is a transcriptional activator expressed specifically in the ICM of the mouse embryo [39]. Like FGF-4, Utf1 is regulated by an enhancer element localized in the 3′-UTR to which Sox-2 and Oct-4 bind and form a complex [40]. Cooperative fixation of both Sox-2 and Oct-4 to their adjacent binding sites is necessary for the activity of the Utf1 enhancer, again demonstrating that none of the two transcription factors is able to transactivate on their own [40] (Fig. 2).

Sox-2 represents an important sequence-dependent Oct-4+ coregulator but it can also negatively regulate Oct-4 activity. The formation of the Oct-4 dimer on the PORE (see above), can be counteracted by Sox-2-binding onto one of the half sites of the PORE, thus resulting in the formation of an Oct-4/Sox-2 complex [V. Botquin and H.R.S., unpublished observations]). Accordingly, cotransfection of Oct-4 expression vectors with increasing amounts of Sox-2 expression vector does repress the activity of the Oct-4 dimer in 293 cells [12]. Since Sox-2 is reported to be downregulated in stem cells prior to Oct-4 shutdown [12], it is possible that relief of Sox-2 interference resulting in Oct-4 dimer formation may cause the transient OPN overexpression that marks the lineage of the primitive endoderm. Thus, it appears that in addition to increased Oct-4 protein levels, as suggested by the experiments performed by Niwa et al. [13], a differential stoichiometry of the interaction of Oct-4 with functional partners could also regulate Oct-4 transactivation in totipotent cells (see below).

HMG proteins have been reported to interact and functionally cooperate with POU proteins in a binding site-independent fashion [4]. Using the novel approach of phage display, we have recently identified HMG-1 and HMG-2 as two factors that are able to interact with Oct-4 in coimmunoprecipitation assays. Transfection of HMG-1 into EC cells causes an increase in Oct-4 transactivation of reporter vectors bearing six copies of the consensus octamer motif cloned at a distance from a minimal TK promoter (6W enhancer) [31], thus providing the first example of a nonviral factor that contributes to Oct-4 activity in stem cells [33]. Although it is not likely that HMG-1 represents a coactivator like E1A, which connects Oct-4 with the basal transcription machinery, it is possible that the interaction between HMG-1 and Oct-4 contributes to the stabilization of the Oct-4/DNA complex and/or to the activation of Oct-4 transactivation domains.

PORE and MORE: Two Alternative Oct-Dimer Configurations Reflecting the Functional Plasticity of POU Proteins

The binding of POU proteins to DNA sites in different dimerization configurations may represent a potent transcriptional activity modulation mechanism in stem cells and differentiating somatic cells. Recently, a novel high affinity Oct dimer-binding site (MORE) has been compared to the known PORE [12] (Fig. 2). The MORE sequence has been obtained by inserting the POUS ATGC subsite within the Pit-1 dimer-binding sequence [41]. MORE mediates the formation of homo- and heterodimers of virtually all mouse Oct factors [42]. The important difference between the PORE and the MORE dimer configurations revolves around the arrangement of the POUS and POUHD. In the case of the PORE, the two subdomains of the same Oct molecule occupy one half-site whereas in case of the MORE, the POUS and POUHD domains of two different Oct molecules make contacts with each half-site. Strikingly, in the presence of the coactivator OBF-1 (OCA-B, BOB-1) [43], Oct-1 activates transcription only when it is bound to the PORE whereas it is not active on the MORE. Indeed, the crystal structure of the Oct-1/MORE dimer revealed that the binding configuration of Oct-1 to the MORE locks the docking site of OBF-1 on the Oct-1 protein, thus impairing interaction and functional cooperation [42]. It is possible that other coactivators may be differentially associated with different POU protein dimer configurations based on the particular arrangement of the DNA binding site.

The MORE and PORE provide two alternative ways of forming POU protein dimers resulting in differential availability of interfaces for the interaction with associated factors. This issue opens the possibility of another level of complexity of transcriptional regulation by POU proteins.

Can There Be Mammalian Life without POU Proteins?

Our previous hypothesis that Oct-4 is a mammalian-specific transcription factor [8] has been supported by the now known absence of homologs in the genomes of C.elegans and Drosophila and probably chicken [44]. Possible reasons that were discussed are the obvious lack of germ plasm or the formation of the trophoblast, an extraembryonic lineage specific for mammalian species.

The differentiation of ES cells to trophoblast cells in which Oct-4 expression levels are reduced can be explained in terms of a simple model of transactivation involving the crosstalk with a bridging factor like E1A or a cofactor such as Sox-2. This model implies that a reduction of Oct-4 protein expression causes a reduction of the activity of Oct-4 target genes (such as FGF-4) [10], or the de-repression of other target genes (such as α and β hCGs, see above) due to a limiting amount of Oct-4. However, the same model cannot account for the primitive endoderm phenotype that results from an increase in Oct-4 expression level in ES cells. In fact, it has been shown that Oct-4 overexpression causes a reduction of 6W enhancer transactivation due to squelching of E1A [31]. Therefore, an increase in Oct-4 expression levels should be linked to a reduction of the ability of Oct-4 to activate target genes and result, in principle, in a similar phenotype of the cells in which Oct-4 protein is downregulated. One possible explanation as to why this is not the case is that an increase in Oct-4 protein levels may modify the balance between Oct-4/Oct-4 homodimers and Oct-4/Oct-6 and/or Oct-4/Oct-1 heterodimers. This may result in a change of the overall dimeric Oct-dependent transactivation in stem cells. Similar to the explanation offered for the increase in PORE activity in primitive EC [12] (see above), the activity of Oct-4 target genes may be increased by forcing the formation of Oct-4 homodimers.

To date, the existence of a transcriptional “equilibrium” established by the maintenance of precise levels of dimers containing Oct-4 and other Oct proteins driving the activity of genes involved in the self-renewal of ES cells is only a matter of speculation. However, the possibility of artificially modifying Oct-4 expression levels using inducible expression systems has now provided the first evidence that the perturbation of this equilibrium can result in specific phenotypic changes of stem cells. This has definitely led to new ways of looking at the differentiation of totipotent cells as a process that is dynamically driven by the relative expression levels of a unique transcription factor, prior to the establishment of a specific gene expression pattern.

Given the striking feature of POU proteins to adopt several binding configurations involving differential interaction with coactivators, it can be predicted that modifications of the relative levels of other Oct factors that are known to be expressed in a tissue-specific manner during mouse embryogenesis [45] may play a major role in the initial events of tissue modeling during embryogenesis. Experiments using systems to control up- and downregulation of the Oct proteins in vivo are therefore necessary to determine whether, as in the case of Oct-4, modifications of their relative levels determine a major “reprogramming” of the cell fate during mouse development.


The authors thank Areti Malapetsa (Syllabus, Canada) for editing the manuscript and Prof. Massimo De Felici and Katharina Lins for critical suggestions. The Marion Dilley and David George Jones Funds and the Commonwealth and General Assembly of Pennsylvania supported this work.