Incomplete reactivation of Oct4-related genes in mouse embryos cloned from somatic nuclei

After historic experiments by Briggs and King(Briggs and King, 1957;Briggs and King, 1952), animal cloning has emerged as a key approach to explore functional changes in the nuclear genome during differentiation(Gurdon, 1974;Gurdon, 1999;Kikyo and Wolffe, 2000;Rideout et al., 2001;Solter, 2000). Studies in amphibians have revealed that the potential of the nucleus to direct normal embryonic development decreases progressively with the developmental stage of the donor cell (Briggs and King,1960; Gurdon,1962; Gurdon et al.,1975). Similarly, in mammals, a direct relationship has been observed between the state of differentiation of a donor nucleus and its ability to generate cloned animals upon transfer into the oocyte(Campbell et al., 1996;Solter, 2000;Tsunoda et al., 1987;Willadsen, 1986). This was established most clearly in mouse cloning experiments using donor nuclei from somatic and pluripotent embryonic stem (ES) cells(Eggan et al., 2001;Rideout et al., 2000). Somatic and ES cell clones that develop to the blastocyst stage in vitro develop very differently in utero after their transfer to recipient animals. As in all other mammals where nuclear cloning has been successful, only a small fraction(1-3%) of mouse blastocysts derived from somatic donor nuclei develop to term,and even fewer reach adulthood (Wakayama et al., 1998; Wakayama and Yanagimachi, 1999; Wakayama and Yanagimachi, 2001). By contrast, up to a third of blastocysts from mouse clones made by nuclear transfer of ES cell donor nuclei develop to birth (Eggan et al., 2001;Rideout et al., 2000). These observations reinforce the notion that a progressive decrease or loss of nuclear potency during cell differentiation is a general characteristic of vertebrate development.

Although as many as 70% of somatic clones can develop to the blastocyst stage in vitro after nuclear transfer, most arrest soon after implantation(Wakayama and Yanagimachi,1999). Homozygous mutant embryos lacking Oct4(Pou5f1 – Mouse Genome Informatics), a gene that encodes a POU-domain transcription factor that has an essential role in the control of developmental pluripotency, form morphologically normal but developmentally deficient blastocysts (Nichols et al.,1998; Niwa et al.,2000). In particular, the cells of Oct4-deficient embryos fail to form the inner cell mass (ICM) that gives rise to all adult tissues(Nichols et al., 1998). The superficially normal development of Oct4-deficient embryos to the blastocyst stage suggests that the combined activity of maternal, zygotic and housekeeping genes might be sufficient to produce a phenocopy of normal preimplantation development. The subsequent failure of embryonic cells in the blastocyst to form the epiblast leads to the manifestation of developmental defects observed in Oct4-deficient embryos. The similar developmental phenotypes observed in Oct4-deficient and cloned blastocysts suggest that developmental failure in both cases results from the absence of a pluripotent ICM.

We have examined the hypothesis that the limited developmental potency of cloned embryos is a result of incomplete reactivation of genes that, likeOct4, function specifically in pluripotent embryonic cells. To test this hypothesis, we have identified 10 candidate genes with an expression pattern similar to Oct4 (designated as Oct4-related genes in this paper) and have compared expression of these genes in normal preimplantation embryos to embryos cloned from somatic cumulus cells and pluripotent ES cells.

Oct4-related genes were identified via electronic analysis of the mouse dataset of UniGene, a regularly updated database of electronically partitioned clusters of expressed sequence tags (ESTs) sequenced from a variety of organisms and their tissues(Schuler, 1997). Previously,electronic analysis of UniGene was used to identify genes whose expression is enriched in the germ cell lineage(Rajkovic et al., 2001). The UniGene mouse dataset can be found atftp://ncbi.nlm.nih.gov/repository/UniGene. Each individual UniGene cluster is a collection of expressed sequence tags(ESTs) that are grouped together based on sequence similarities. Each EST record is linked to information about the cDNA library from which it was sequenced and its biological source (available athttp://www.ncbi.nlm.nih.gov/UniGene/lbrowse.cgi?ORG=Mm). We examined biological source descriptions of all cDNA libraries that contribute to the UniGene mouse dataset and identified 105 cDNA libraries of interest prepared from pre-implantation embryos, gastrulating embryos,embryonic and adult gonads, posterior regions of early and late gestation embryos that also contained developing gonads and germ cells, and ES and embryonic carcinoma (EC) cells. Few cDNA libraries lacking sufficient biological source descriptions were excluded from our analysis.

The UniGene mouse dataset partitions 2.5 million ESTs sequenced from over 500 cDNA libraries into about 88,000 EST clusters. The 105 cDNA libraries of interest to us contributed over 350,000 ESTs to the dataset. Taking these parameters into account, we estimated that a UniGene cluster of six or more ESTs derived from cDNA libraries of interest has a statistically significant chance of identifying a gene specific to tissues containing pluripotent cells. We further restricted electronic selection criteria by requiring that contributing ESTs be derived from at least two distinct stages of development,including one from the preimplantation embryo. This latter criterion was essential to improve selection for genes with a continuous Oct4-like expression pattern and to avoid selecting highly expressed genes specific to a particular cell-type or developmental stage, such as genes specifically involved in spermatogenesis or meiosis.

We then used a custom Perl script to electronically identify 102 UniGene clusters that satisfied the selection criteria. EST contigs were assembled using Sequencher 4.1.2 (Gene Codes Corporation). Subsequent sequence analysis and expression analysis by RT-PCR narrowed the number of clusters of interest to 66. We then performed a more detailed RT-PCR-based expression analysis on embryonic and germ cell samples to identify genes with developmental expression profiles closely matching that of Oct4. Finally, we assessed the reproducibility of detection of candidate genes in single preimplantation embryos and chose 10 Oct4-related genes for further analysis.

Embryos were collected in 100 μl of TRIzol reagent (GibcoBRL) in non-stick, RNAse-free microtubes (Ambion), extracted twice with 20 μl of chloroform using Phase Lock Gel Heavy 0.5 ml tubes (Eppendorf) spun at 5000g. Total RNA was precipitated with an equal volume of isopropanol in the presence of 1.5 μl of GlycoBlue (Ambion). Residual genomic DNA was removed using DNA-free reagent (Ambion). RNA was re-precipitated, washed in 70% ethanol and dissolved in 2 μl of RNAse-free water (Ambion). First strand cDNA synthesis and universal amplification (four cycles) were performed using the SMART PCR cDNA synthesis kit (Clontech). cDNAs was purified using the QIAquick PCR purification kit (Qiagen), eluted in 100 μl of water. PCR products were resolved in 2% agarose gels.

PGCs purified from day 10 (E10) of embryonic development ofOct4-GFP transgenic mouse line (A.B. and D.C.P., unpublished) were collected directly into TRIzol reagent using a Becton Dickinson cell sorter. Epiblasts of early gastrulation embryos were dissected cleanly from extra-embryonic tissues. Embryonic gonads were collected from E12 embryos of both sexes by carefully dissecting them from mesonephroi. RNA was prepared from collected tissues using TRIzol reagent and processed for cDNA synthesis as described above.

Nuclear transfer of ES cell and cumulus cell nuclei into enucleated metaphase II oocytes was carried out as previously described(Eggan et al., 2000;Wakayama et al., 1998;Wakayama and Yanagimachi,2001). Three hours after nuclear transfer, oocytes were activated for 5 hours with 10 mM Sr²⁺ in Ca²⁺-free CZB media in the presence of 5 μg/ml Cytochalasin B. Embryos were cultured in vitro for 96 hours, preimplantation development was assessed by Hoffman contrast microscopy, and individual embryos were isolated for gene expression studies.

All embryo culture was carried out in microdrops on bacterial petri dishes(Falcon) under mineral oil (Squibb). HEPES-buffered CZB was used for room temperature operations, while long-term culture was carried out in bicarbonate buffered KSOM at 37°C under an atmosphere of 5% CO₂ in air.

cDNA sequences for examined genes: Pramel4, AF490340;Pramel5, AF490341; Pramel6, AF490342; Pramel7,AF490343; Ndp52l1, AF490344; Dppa1, AF490345;Dppa2, AF490346; Dppa3, AF490347; Dppa4, AF490348;Dppa5, AF490349.

Primer sequences and RT-PCR conditions for mouse genes: Pramel4,G73527; Pramel5, G73528; Pramel6, G73529; Pramel7,G73530; Ndp52l1, G73531; Dppa1, G73532; Dppa2,G73533; Dppa3, G73534; Dppa4, G73535; Dppa5,G73536; Ptgs2, G73537; Hsh, G73538; Tnfsg6, G73539;Oct4, G73540; Gapd, G73541.

Oct4 is the best-studied gene essential for the maintenance of a pluripotent state in early embryonic and ES cells(Pesce and Scholer, 2001).Oct4 encodes a POU-domain transcription factor and is expressed in pluripotent cells of the early embryo and in the embryonic and adult germ cells. We were interested in identifying genes whose expression patterns are like that of Oct4 and that might be involved in establishing or maintaining developmental pluripotency. We used the following criteria for electronic selection: (1) expression in cleavage-stage embryos, blastocysts and in the embryonic germline, and (2) the genes should not be expressed in somatic cells.

An electronic analysis of the UniGene mouse dataset generated a list of 66 candidate genes, ten of which were selected for this study based on theirOct4-like temporal expression pattern and reproducibility of detection in single preimplantation embryos (for details see Materials and Methods). Fig. 1 shows that these 10 candidate genes were expressed during cleavage in early epiblasts and germ cell-containing samples, but they were not expressed in any of the somatic tissues tested. Expression analysis of normal four-cell, morula- and blastocyst-stage embryos established that all but one of theseOct4-related genes were detectable, beginning with the four-cell stage of development (Fig. 2). Strong expression of Dppa1 was readily seen at the morula-to-blastocyst transition.

Comparison of predicted ORFs for the 10 genes revealed sequence similarities to previously characterized human and mouse proteins (summarized in Table 1). Of particular interest are four Oct4-related genes that encode proteins similar to the PRAME antigen detected in a wide range of human tumors, including melanomas, leukemias, sarcomas and renal carcinomas(van Baren et al., 1998). Two additional mouse Pramel genes, Pramel1 and Pramel3, have previously been shown to exhibit male germ stem cell-specific expression(Wang et al., 2001). Products of three other genes, Dppa2, Dppa3 and Dppa4, share a SAP putative DNA-binding motif often found in proteins implicated in chromosomal organization, as well as in various aspects of RNA biology(Aravind and Koonin, 2000).Dppa3 was described recently as a primordial germ cell(PGCs)-specific gene PGC7 (Sato et al., 2002), as was Stella(Saitou et al., 2002).Stella has been implicated in the initiation of the germ cell lineage, and its expression marks the emerging primordial germ cells in the epiblast (Saitou et al.,2002).

The results of expression analysis summarized in Figs1 and2 indicate that in silico selection identified a class of genes that were transcribed in pluripotent embryonic but not in differentiated somatic cells. The homologies, expression pattern and limited functional data suggest that these genes may play important roles in early development. We compared the expression of these genes in normal blastocysts to blastocysts cloned from different donor nuclei.

For expression analyses, total RNA was isolated from individual control and cloned embryos as well as from cumulus donor cells. Expression ofOct4 and related genes was not detected in cumulus cells, although sensitive RT-PCR analysis occasionally revealed small amounts of mRNA contamination that were probably the result of oocyte lysis during enzymatic and mechanical disruption of oocyte-cumulus-cell complexes(Fig. 3). Parthenogenetic and in vitro fertilized embryos served as controls for the nuclear transfer procedure and culture conditions. All tested Oct4-related genes were expressed in all normal control, in vitro fertilized and parthenogenetic blastocysts (Fig. 2,Table 2 and data not shown).

We examined expression of the Oct4 and 10 Oct4-related genes in a total of 48 individual cumulus cloned preimplantation embryos that developed to phenotypically normal blastocysts, after transfer of a donor cumulus cell nucleus into enucleated oocytes. While each of 15 control embryos derived from fertilized zygotes expressed all 11 genes, 18 of 48 cloned blastocysts failed to correctly reactivate expression of at least one of the 11 test genes (Figs 3,4). Of the tested genes,Oct4 was incorrectly expressed more than any other, and 25% of all examined cloned blastocysts failed to express detectable levels ofOct4. However, other Oct4-related genes also failed to be activated in multiple cloned embryos. For example, Pramel4 was not expressed in 17% of cloned blastocysts, while Dppa1 expression was undetectable in 13% of all cloned blastocysts. Overall, 11 out of the 18 (or 62%) clones with incomplete gene reactivation failed to express one gene,while the remaining cumulus clones failed to activate between 2 and 7 of the test genes. Two of the genes, Pramel6 and Pramel7, are transcribed in opposite directions from their respective promoters, which are positioned within 16 kb from each other on chromosome 2 (data not shown). Despite this close chromosomal proximity, we conclude that reactivation of one of the two genes occurred independently of the other, as several blastocysts expressed one but not the other of the two genes(Fig. 4).

Clones that arrested in development before reaching the blastocyst stage were also analyzed. As shown in Fig. 3 and summarized in Fig. 4 and Table 2, only 25% of clones arrested at the morula stage and none of those arrested at the eight-cell stage expressed all 11 tested Oct4-related genes.

We also examined the expression of three genes that are active in the cumulus donor cells but not in normal blastocysts. These genes are prostaglandin-endoperoxide synthase 2 (Ptgs2)(Joyce et al., 2001),hyaluronan synthase homologue (Hsh)(Fulop et al., 1997b) and TNF-stimulated gene 6 (Tnfsg6)(Fulop et al., 1997a). In contrast to incomplete reactivation of the Oct4-related genes, all cumulus cell-derived clones uniformly lacked expression of the three `somatic'genes regardless of the extent of development in vitro(Fig. 3).

The hypothesis that incomplete reactivation of the Oct4-related genes may contribute to the early lethality of somatic clones predicts that these genes should be expressed faithfully in blastocysts derived from nuclei of donor cells that are expressing these genes. To test this possibility, we examined expression of the 11 test genes in donor ES cells and in individual ES cell clones. As expected from the electronic selection criteria, all tested genes were expressed in undifferentiated ES cells(Fig. 5). In contrast to cumulus donor cell-derived clones, all ES cell nuclei-derived clones expressed all eleven tested genes (Figs 4and 5,Table 2 and data not shown). In addition, unlike somatic clones, ES cell clones correctly recapitulated temporal regulation of the Dppa1 gene. Like normal embryos in whichDppa1 expression is upregulated at the morula-to-blastocyst transition (Fig. 2),Dppa1 was expressed in blastocyst-stage ES clones but not in morula-stage clones (Fig. 5). Thus, the ES cell-derived clones appear to have faithfully recapitulated expression of the Oct4-related genes with a similar kinetics to the normal embryos.

In this study, we explored the biological causes of early embryonic lethality observed in somatic cell clones. Based on the similarities between the previously described phenotypes exhibited by Oct4-deficient embryos and somatic cloned blastocysts, we posited that somatic clones fail because of incomplete reactivation of genes functioning in the pluripotent cells of the preimplantation embryo. The goal of this study was to identify genes that can serve as suitable molecular markers for developmental pluripotency, and to examine their expression in both normal and cloned embryos derived from differentiated somatic and undifferentiated cells.

We used an in silico selection method to identify a class of embryonicOct4-related genes whose expression is limited to the pluripotent cells of the early embryo and the germline. Although the biological functions of these genes have yet to be elucidated, we have found that they serve as informative markers for pluripotent cell types and thus can be validly used in the analysis of cloned embryos. The electronic selection that led to the identification of this set of genes was modeled after the properties and expression pattern of the Oct4 gene, which has been shown to have an essential role in early development. Recently, our observations have been corroborated by the results of an expression profiling study that identifiedDppa2, Dppa4, Dppa5 and Oct4 among genes found at high levels in mouse ES cells (Ramalho-Santos et al., 2002). In addition, expression of theOct4-related genes is downregulated during ES cells differentiation(data not shown) in strong agreement with a prior description ofDppa5 as a cDNA downregulated during EC cell differentiation(Astigiano et al., 1991). Finally, one of the Oct4-related genes,Dppa3/Stella/PGC7, has been implicated in the earliest steps of germline initiation, further strengthening the argument that these genes might represent factors with specific roles in early development(Saitou et al., 2002).

Analysis of expression of the Oct4-related genes in somatic and ES cell-derived clones revealed that successful reactivation of the full set correlates with development of cloned embryos to term. Mouse clones derived from nuclei of ES cells, where the genes are already active, recapitulate their expression pattern with the same kinetics as observed during normal embryogenesis. In contrast to ES clones, almost 40% of the cumulus cell-derived blastocysts, though morphologically normal, failed to reactivate these genes faithfully. Activation of these genes in embryos that were arrested at the pre-blastocyst stage was even less efficient. The significance of this result is difficult to assess as impaired viability and earlier developmental arrest of these embryos might be have been caused by physical damage after nuclear transfer, leading to RNA degradation.

Incomplete reactivation of Dppa genes in cumulus clones strongly agrees with a recent independent study of Oct4 expression in somatic clones(Boiani et al., 2002). In that study, 82% of clones reactivated Oct4. In situ mRNA analysis also revealed that some of these Oct4-positive clones exhibited aberrantOct4 expression in the trophoectoderm. These findings suggest that somatic clones capable of reactivation of Oct4 may also fail to reproduce its normal spatial pattern of expression. Our work confirms these prior findings and attributes incomplete recovery of pluripotency in somatic clones to a larger set of specific genes.

Our observations could plausibly reflect a more profound transcriptional dysregulation in cumulus cell-derived cloned embryos. Future experiments using single-embryo expression profiling on microarrays will examine the entirety of mouse transcriptome and reveal which genes consistently fail to be expressed in somatic cell-derived cloned embryos. We emphasize, however, that all examined cumulus cell-derived cloned blastocysts expressed Gapd,while the expression of three genes, Ptgs2, Hsh and Tnfsg6,which are normally abundantly expressed in donor cumulus cells, was undetectable in cloned embryos (Fig. 3). These observations indicate that mechanisms controlling gene expression in cumulus cell-derived clones selectively affect whether housekeeping and `somatic' genes are expressed. By contrast, they can only achieve incomplete reactivation of expression of Oct4 and 10Oct4-related genes.

Why somatic clones fail to reactivate the Oct4-related genes remains to be determined. It is now generally accepted that epigenetic mechanisms play important roles in the control of gene expression during embryogenesis (Reik and Walter,2001; Renard,1998; Rideout et al.,2001; Surani,2001). However, little is known about the molecular mechanism of epigenetic reprogramming in the context of normal development and somatic cloning. Recently, initial studies of bulk genomic DNA methylation in normal and cloned embryos have been reported(Barton et al., 2001;Bourc'his et al., 2001;Dean et al., 2001;Fairburn et al., 2002;Kang et al., 2001;Mann and Bartolomei, 2002;Santos et al., 2002). These studies have revealed significant differences in the pattern and dynamics of DNA methylation between normal and cloned embryos. However, current approaches for analysis of epigenetic modifications in single pre-implantation embryos have significant limitations. Importantly, they lack the resolution to allow straightforward identification of individual genes whose expression and epigenetic make-up could be meaningfully correlated with the development of clones. In this regard, the Oct4-related genes identified in our work represent a set of loci whose epigenetic control should be examined in the future. Such analysis might provide specific examples of the role of epigenetic control of gene expression in the development of cloned and normal embryos.

We thank Glenn Paradis of the MIT Flow Cytometry Core Facility for his assistance with FACS, Robert Latek for electronic analysis of Dppa proteins,and Britt Rideout, Konrad Hochedlinger and Douglas Menke for discussions and comments on the manuscript. A.B. was Leukemia & Lymphoma Society Special Fellow. This work was supported by Howard Hughes Medical Institute and the National Institutes of Health.