br Results br Discussion Many small molecules including two

Results
Discussion Many small molecules, including two reported by our group (Lee et al., 2013b), have been suggested to be selectively toxic to PSCs. However, as described previously (Knoepfler, 2009), despite the high potency and efficacy of small molecules that target hPSC-specific genes (e.g., BIRC5) (Lee et al., 2013b), there may be unexpected side effects on certain differentiated cell types owing to unexpected expression of the particular target protein (e.g., BIRC5 expression in hematopoietic stem cells, see Leung et al., 2007). Although the small molecules used in previous studies had no significant effects on cellular functions, at least those of dopaminergic neurons and smooth muscle ubiquitin e3 ligase (Lee et al., 2013b), such possible side effects need to be carefully examined for other small molecules. As selective cell death of undifferentiated hPSCs is crucial to guarantee functional safety of differentiated cells during a treatment, development of alternative strategies based on unique gene expression and properties of undifferentiated PSCs is required. In this respect, KR, which produces ROS upon exposure to visible light of 540–580 nm (Bulina et al., 2006), was synergized with the unique features of PSCs, namely, their high sensitivity to oxidative stress (Han et al., 2008) and high susceptibility to mitochondrial-dependent apoptosis (Lee et al., 2013b) to develop an alternative strategy. To this end, KR was expressed at the mitochondria (Figure 1B), and this expression was tightly controlled by EOS-C(3+), an artificial pluripotent-specific promoter (Okumura-Nakanishi et al., 2005) (Figure 2). ROS generated upon exposure of KR to light specifically induced mPSC cell death (Figure 5), and the PSCs specific cell death by KR expression blocked teratoma formation in a mouse model (Figures 5 and 6). More importantly, EC-KR-mESCs were functional in vitro and in vivo (Figure 6) after exposure to light at an intensity that was sufficient to inhibit teratoma formation (Figures 6G and 6H). This approach does not use additional chemicals, unlike the typical suicide gene-based approach, which uses thymidine kinase and ganciclovir, which may have cell-type-dependent cytotoxicity (Rong et al., 2012; Janoly-Dumenil et al., 2009) or hepatotoxicity (Shea et al., 1987). Despite numerous studies demonstrating that teratomas do not form following xenotransplantation of cells differentiated from hPSCs into a rodent model (Cho et al., 2007; Kriks et al., 2011), review articles continue to warn of the risk of teratoma or tumor formation in hPSC-based therapies (Cunningham et al., 2012; Lee et al., 2013a). Cells derived from mPSCs are apparently more prone to develop tumors in a mouse model (Arnhold et al., 2004; Moon et al., 2013), consistent with host-dependent bias (Erdö et al., 2003). Thereby, it is most appropriate to verify a teratoma inhibition strategy using mPSCs in a mouse model to exclude the host-dependent bias. Once the strategy is validated, it can be applied to hPSCs. To this end, we also established KR-hESCs to examine selective cell death by light exposure. Of interest, we noticed that cell death rate by light exposure in KR-hESCs was significantly lower compared to ∼100% cell death achieved in mESCs under the identical condition (Figure 3). We reasoned that, unlike in mESCs (Figure S7D), KR protein in hESCs underwent a consistent proteasome-dependent degradation despite its high mRNA expression level (Figure S7C). This notion was further supported by pretreatment of MG132, a proteasome inhibitor, which significantly increased KR protein level in KR-hESCs (Figure S7E). Recently, it has been reported that hESCs show high proteasome activity due to high expression of 19S proteasome subunit (Vilchez et al., 2012). Considering a variety of cellular and molecular differences between hESCs and mESCs have been addressed by distinct characteristics of the primed (for hESCs) and naive (for mESCs) state, respectively (Nichols and Smith, 2009), the different protein stability of KR between hESCs and mESCs might result from the increased proteasome activity in hESCs, which may be linked to distinct characteristics of the primed state of ESC.