pi3k inhibitors We speculated that MEKi might decrease DNA m

We speculated that MEKi might decrease DNA methylation levels through the downregulation of the DNMT3 family. MEKi significantly diminished global DNA methylation levels and, surprisingly, was sufficient to induce high 5hmC levels. Because the TET family of dioxygenases that catalyze the conversion of 5mC into 5hmC plays important roles in the induction of DNA hypomethylation (Hill et al., 2014; Kohli and Zhang, 2013), we hypothesized that JMJD2C might interact with TET family members. Thus, the absence of Jmjd2c may disrupt MEKi-induced DNA hydroxymethylation. Indeed, when Jmjd2c was knocked out, 5hmC levels were not altered by MEKi, suggesting that TET1 activity toward 5mC requires JMJD2C. Consequently, MEKi decreases 5mC levels via DNMT3A/B reduction and increases 5hmC levels via the enhancement of TET1 activity. Our findings support the idea that MEKi promotes a ground state in ESCs by lowering global DNA methylation through two different mechanisms: passive DNA demethylation via reduction of the DNMT3 family and active DNA demethylation via JMJD2C-mediated TET1 activation. GSK3i reduced Dnmt3 family transcript levels, thus decreasing DNMT3A/B protein levels. Remarkably, GSK3i decreased global 5mC levels without altering 5hmC levels, indicating that GSK3i regulates DNA demethylation by a mechanism different than that of MEKi. Thus, our results show that GSK3i-induced DNA demethylation does not involve active DNA demethylation by the TET family. PRDM14 regulates the maintenance of naive pluripotency in ESCs through a dual mechanism: the inhibition of FGFR signaling and the repression of Dnmt3 pi3k inhibitors (Grabole et al., 2013; Hackett et al., 2013a; Yamaji et al., 2013). Although PRDM14 contains a PR/SET domain, we could not detect PRDM14 methyltransferase activity toward the DNMT family. Instead, we discovered that G9a interacts with PRDM14. Consistent with recent reports (Grabole et al., 2013; Hackett et al., 2013a; Yamaji et al., 2013), PRDM14 suppressed the expression levels of DNMT3A/B and, moreover, interacted with DNMT3A/B but not with DNMT1. G9a interacted with and methylated the DNMT3 family. Based on those findings, we postulated that PRDM14 might serve as a scaffold to accommodate both G9a and DNMT3A/B and facilitate G9a-mediated DNMT3A/B protein degradation when PRDM14 is strongly induced in ESCs under 2i/LIF conditions. In accordance with that idea, DNMT3A/B levels in G9a/GLP double-KO cells were high under normal conditions and were less affected than those in wild-type cells under 2i conditions. We propose that DNMT3A/B methylation by G9a leads to the degradation of those proteins through a mechanism similar to the one we reported previously (Jung et al., 2015). Therefore, DNMT3A/B may contain a methyl degron targeted by G9a. This idea is supported by our findings that 248-5 cells express high levels of DNMT3A/B mRNA and protein compared with wild-type TT2 cells. MEK1/2 and GSK3α/β could have many phosphorylation targets in ESCs. Thus, we focused on the roles of targets changed by PD0325901 and CHIR99021. However, researchers have largely ignored protein modifications induced by 2i thus far. 2i may affect the protein and mRNA levels of those targets differently, as we observed with JMJD2C, OCT4, and TET1. In agreement with that conclusion, UHRF1 was decreased on the protein level during the serum-to-2i transition (von Meyenn et al., 2016). We could discriminate between the effects of the inhibitors, either alone or in combination, on transcriptional and translational control; thus, the DNMT3 family was reduced by 2i through transcriptional suppression and protein degradation. Our findings suggest that PD0325901 alone might be sufficient to maintain a ground state in ESCs. Likewise, PD0325901 alone induced the formation of PGC-like cells from ESCs (Kimura et al., 2014). But, CHIR99021 alone sustained an ESC-like morphology (Figure 1A), probably due to suppression of Wnt signaling (Atlasi et al., 2013). Thus, the combination of MEKi and GSK3i must have a synergistic effect on the maintenance of ESCs in a naive ground state. Taken together, our data support a model wherein 2i maintains the ground state in ESCs through JMJD2C-enhanced TET1 activation and PRDM14/G9a-dependent DNMT3A/B protein degradation (Figure 7E).