We report here that overexpression of SOX2 or SOX2-D, an inhibitory form of SOX2, in 1-cell embryos results in developmental arrest, with SOX2-expressing embryos arresting at the 2-cell stage and SOX2-D-expressing embryos arresting between the 3-cell and 8-cell stages. SOX2 overexpression has a substantial effect on the reprogramming of gene expression that occurs during the course of zygotic gene activation, affecting approximately 20% of zygotically expressed genes, with approximately 70% of the affected transcripts being down-regulated. In contrast, overexpression of SOX2-D affects expression of a smaller number of zygotically expressed genes (108 vs. 453), a finding consistent with SOX2:POU5F1 inhibition requiring the carboxyl terminus of SOX2, which contains the trans
-activation domain [39
] and is not present in SOX2-D. Such differences in gene expression likely account for the differences when developmental arrest is observed (i.e., earlier arrest following SOX2 overexpression is due to a more pronounced effect on reprogramming gene expression). Moreover, the dominant effect of SOX2 overexpression is repression of zygotically expressed genes, a situation not observed following overexpression of SOX2-D. These findings are consistent with the small decrease in transcription following overexpression of SOX2, but not SOX2D. In addition, as noted in the Results
, the small amounts of biological material preclude ChIP analyses that would support a direct role for SOX2 in regulating expression of the zygotically expressed genes for which expression is perturbed when SOX2 activity is experimentally manipulated. Ascertaining whether the changes in transcript abundance occur in the presence of cycloheximide is a possible way to circumvent this difficulty. We found, however, that the long-term exposure to cycloheximide required to conduct such experiments resulted in nonspecific effects on transcript abundance (data not shown). Nevertheless, that more than 90% of the affected genes contain a SOX2-binding consensus sequence within −3 kb of their promoter provides indirect support for the idea that the affected genes may, indeed, be directly regulated by SOX2.
Consistent with our published microarray data [3
, based on the increase in transcript abundance between the 1-cell and 2-cell stages, is a zygotically expressed gene. Also consistent with the microarray data is the dramatic increase in Sox2
transcript abundance between the 8-cell/morula and blastocyst stages. Although cytoplasmic staining of SOX2 protein is observed, SOX2 is found in the nucleus at all stages of development, which is consistent with a recent report [24
]. In contrast to our finding that SOX2 is localized in both the cytoplasm and the nucleus in oocytes, a previous study reported that SOX2 is primarily localized in the cytoplasm as assessed by immunohistochemistry [14
]; those authors did observe a nuclear localization for SOX2 in 2-cell embryos when immunocytochemistry was employed. Our finding that the nuclear fluorescent signal is markedly increased following injection of a cRNA encoding SOX2-EGFP is consistent with SOX2 localizing to the nucleus. The low resolution of immunohistochemistry is a possible explanation for the observed differences.
To date, the function of only a few genes (e.g., Smarca4
, and Ube2a
) has been identified as being required for development beyond the 2-cell stage [40
]. Of these, only SMARCA4 and TRIM24 are transcription factors for which function in early development was identified by loss-of-function approaches. RNAi-mediated depletion of SOX2 protein was unsuccessful because of the stability of SOX2 protein, a finding consistent with a previous study [14
]. The dominant-negative approach, however, did unmask a phenotype that is less severe than that observed when SOX2 is overexpressed. Although SOX2-D can totally suppress SOX2-driven expression of a reporter gene in somatic cells (Supplemental Fig. S2), in the absence of such data in early embryos, it is possible that a more severe phenotype would be observed if higher levels of SOX2-D expression were achieved. Cdkn1a
(p21), which inhibits cell-cycle progression [45
], is a potential candidate for the observed cleavage arrest between the 3-cell and 8-cell stages, because it is up-regulated in SOX2-D-overexpressing embryos (Supplemental Table S8). Oocyte-specific targeting of Sox2
could, in principle, deplete maternal SOX2 protein and, thereby, provide an opportunity to assess better the consequences of loss of SOX2.
A concern with overexpression studies is that higher concentrations of protein will lead to promiscuous interactions that would not normally occur. For SOX2, such interactions could take the form of binding to weak SOX2-binding sites, squelching effects caused by titrating normal binding partners [46
], or changing promoter accessibility as a result of SOX2′s ability to bend DNA [47
]. It is very difficult to exclude the contribution from any of these scenarios to the observed compromised embryo development. Nevertheless, the endogenous pool of SOX2 was only expanded by 70%, and a similar inhibition of development was observed when the endogenous pool was expanded by approximately 5-fold. These results minimize the likelihood that the observed affects on development are not the result of the aforementioned scenarios.
The response of ES cells and embryos to SOX2 overexpression is strikingly different. Overexpressing SOX2 in ES cells does not inhibit cell proliferation but, rather, biases differentiation toward neuroectoderm [19
], whereas overexpressing SOX2 in embryos results in cleavage arrest at the 2-cell stage. The decrease in expression of the genes Sox2
, and Nanog
, the functions of which appear to be critical to stem cell renewal and maintenance of pluripotency [9
], following overexpression of SOX2 in ES cells [39
] likely contributes to their differentiation; each of these genes is a SOX2:POU5F1 target gene [39
]. Although Sox2
expression is increased in SOX2-overexpressing embryos (Supplemental Table S5), no apparent change is observed in expression of Pou5f1
in SOX2-overexpressing embryos (Supplemental Tables S3 and S5), and Nanog
, which is not expressed in 2-cell embryos [3
], is not precociously expressed. Sox2
expression in ES cells via Nr5a2
) and Nr2f2
]. This pathway, however, appears not to be functional in early embryos, because Nr5a2
expression is not affected in SOX2-overexpressing embryos (Supplemental Tables S3 and S5). The pronounced effect on reprogramming gene expression observed in SOX2-overexpressing embryos—namely, expression of approximately 20% of zygotically activated genes is perturbed—likely underlies the inability of SOX2-overexpressing embryos to develop beyond the 2-cell stage. Of note is that depletion of maternal SMARCA4 also results in arrest at the 2-cell stage and misexpression of about one third of the zygotically expressed genes [40
Why is the major outcome of SOX2 overexpression in early embryos decreased expression of the majority of zygotically expressed genes? In ES cells, binding of SOX2, POU5F1, and NANOG to closely localized sites activates and maintains expression of genes involved in stem cell renewal [51
]; NANOG most often colocalizes with SOX2 [38
]. In contrast, promoters of genes involved in differentiation and, therefore, not expressed in ES cells tend to be occupied by a single factor [53
]. We found at least one consensus SOX2-binding sequence within −3 kb of the promoter in more than 90% of zygotically expressed genes for which expression is perturbed in SOX2-overexpressing embryos. We searched TTTGCATXACAA(A/T)G for adjacent SOX2- and POU5F1-binding sequences within −10 kb of the promoters of the aforementioned genes and noted that this motif was not found in the promoter of any affected gene. This finding, coupled with the absence of NANOG expression in 2-cell embryos, could account for repression of gene expression as a major outcome of SOX2 overexpression.
The ChIP experiments using mouse ES cells identified approximately 1000 genes for which the promoters are occupied by SOX2 [38
]. Remarkably, only 25 of the zygotically expressed genes for which expression is perturbed in SOX2-overexpressing embryos are common (Supplemental Table S14), and this may reflect differences in function of pluripotency factors, such as SOX2, during early development. As noted above, these factors are critical for stem cell renewal and maintenance of pluripotency, and presumably, the gene expression profile of these cells is relatively constant, except for changes linked to cell-cycle progression. Blastomeres of the early embryo, however, do not self-renew; rather, this population of cells undergoes a series of reductive cell divisions in which the pattern of gene expression is changing [1
]. Whether a 2-cell blastomere is truly totipotent or is on a dedifferentiation pathway yet to be completed—the oocyte is a highly differentiated cell—remains an open question. The lack of a substantial set of common genes between ES cells and 2-cell embryos that could potentially be directly regulated by SOX2 is consistent with the concept of 2-cell blastomeres being an intermediate stage on the path to totipotency. Alternatively, this difference in gene expression could reflect distinctions between the totipotent (blastomere) and pluripotent (ES cell) states or indicate that several patterns of gene expression are compatible with a totipotent or pluripotent state.