Fibroblasts can be directly reprogrammed into neurons

Fibroblasts can be directly reprogrammed into neurons (Vierbuchen et al., 2010), cardiomyocytes (Ieda et al., 2010), and NSCs (Han et al., 2012) using tissue-specific combinations of transcription factors. However, the pluripotent reprogramming cocktail of Oct4, Sox2, Klf4, and Myc can also convert fibroblasts into cardiomyocytes (Efe et al., 2011) and NSCs (Kim et al., 2011; Thier et al., 2012) under suitable culture conditions. In addition, the overexpression of only Oct4 has been reported to reprogram fibroblasts toward the hematopoietic lineage (Szabo et al., 2010). In these studies, the overexpression of Oct4 alone or in combination with other transcription factors is thought to force the cells into an unstable and plastic intermediate (Orkin and Hochedlinger, 2011) that can be pushed toward the desired final cell type by specific environmental cues. We speculate that the initial stochastic phase of the reprogramming process (Buganim et al., 2012; Hanna et al., 2009) involves a transient transcriptionally unstable state that may be a prerequisite to the hierarchical events that take place during the late steps of reprogramming (Buganim et al., 2012). In addition, the reprogramming transgenes are usually expressed at higher levels during reprogramming than their corresponding endogenous counterparts in ESCs. These nonphysiological conditions may promote nonspecific binding to low-affinity stavudine and contribute to transcriptional chaos and instability. Our data suggest that Oct4 overexpression at early stages of reprogramming plays a role in the initiation of such an unstable intermediate through the inhibition of the somatic cell-type-specific program rather than through the activation of the pluripotent transcriptional network. Our results are consistent with previous four factor studies showing that fibroblast-specific genes are downregulated at early stages and that pluripotency-related genes are not upregulated until the late stages of reprogramming (Brambrink et al., 2008; Stadtfeld et al., 2008). Although MYC was suggested to be the reprogramming factor primarily responsible for suppressing the cell-type-specific program (Sridharan et al., 2009), our data show that OCT4 alone can also inhibit somatic transcriptional networks. Finally, current available data suggest that OCT4, SOX2, and KLF4 act synergistically, as OCT4 heterodimerizes with SOX2 in order to maintain ESC self-renewal (Boyer et al., 2005) and OCT4, SOX2, and KLF4 co-occupy the promoters of many reprogramming-related genes (Soufi et al., 2012; Sridharan et al., 2009). In contrast, we show that OCT4 alone is able to initiate certain steps in reprogramming, such as the activation of Mgarp, and that OCT4 and KLF4 play an antagonistic role on Mgarp transcriptional regulation. Whereas each cell type exhibited a unique pattern of up- and downregulated genes, four genes were upregulated by OCT4 in all examined somatic cell types, including Mgarp. Our results show that the process of inducing de novo pluripotency does not simply consist of the activation of ESC-specific genes to the levels present in ESCs. Instead, some genes such as Mgarp need to be first downregulated and subsequently upregulated during the reprogramming process. We have elucidated the mechanism underlying this counterintuitive temporal expression pattern. In fact, KLF4 alone completely abolishes Mgarp expression whereas OCT4 alone induces high levels of Mgarp. This competitive interplay ensures that appropriate expression levels of Mgarp are maintained at different time points, which appear to be crucial for a successful reprogramming process, as either the permanent inhibition or the premature activation of Mgarp prevents the efficient generation of iPSCs. To our knowledge, Mgarp is the first gene described to date upon which KLF4 and OCT4 exhibit antagonistic effects. The mechanism that regulates the switch from KLF4- to OCT4-regulated Mgarp expression remains to be identified.