Fig. S1. qRT-PCR analysis of marker gene expression in embryo-derived EpiSCs compared with ES cells. ES-SL, ES cells in serum/Lif; ES-2iL, ES cells in 2i/Lif. Epi6 and Epi7 are two independent embryo-derived EpiSC lines. y-axis, relative expression to Gapdh, then normalized to ES-SL.

Fig. S2. qRT-PCR analysis of marker gene expression in ES cell-derived EpiSCs (ES-Epi) compared with ES cells. (A) qRT-PCR analysis of ES cell differentiation into EpiSCs upon culture in Fgf2 and activin. ES-Epi3 and ES-Epi10 indicate cells cultured in Fgf2 and activin for three and ten passages, respectively. y-axis is relative expression to ES cells in 2i/Lif (ES-2iL). (B) ES cells constitutively expressing Klf4 retain a EpiSC marker profile in Fgf2 and activin. MT, empty vector transfectants. P0, P2 and P10 indicate passage numbers in Fgf2/activin. y-axis is relative expression to MT-P0, control vector-transfected ES cells.

Fig. S3. pGG137Klf4-transfected EpiSCs in Fgf2 plus activin retain the X chromosome silencing mark of trimethylated H3K27. The white arrowhead indicates the me3H3K27 nuclear body. Immunostaining for Oct4 (green) and me3H3K27 (red) is shown.

Fig. S4. Time lines of transfection and Oct4-GFP-positive iPS cell colony formation after transfer to 2i/Lif. (A) Cells plated in 2i/Lif 48 hours after transfection. (B) Cells plated in 2i/Lif 72 hours after transfection. (C) Klf4 stable transfectants, transferred into 2i/Lif on day 0. Dashed line, time in activin and Fgf2. Black line, 2i/Lif. Blue line, 2i/Lif culture, GFP-positive colonies expanding.

Fig. S5. qRT-PCR analysis of marker gene expression in Epi-iPS cells compared with embryo-derived EpiSCs and ES cells. K4C1, K4C12, K4C3, K4C5 are clones with the Klf4 transgene; C3-A3, C3-D4, C5-A5, C5-B4 are clones with Klf4 deleted from K4C3 and K4C5. y-axis, relative expression to Gapdh, normalized to EpiSCs.

Fig. S6. Aggregations of Oct4-GFP EpiSC and Epi-iPS cells with morula-stage embryos. (A-C) Oct4GiP reporter EpiSCs or derivative Epi-iPS cells were combined with E2.5 MF1 morulae and maintained in culture for 48 hours before transfer to pseudopregnant recipients. Cultured embryos were examined by fluorescence microscopy at 24-28 and 44 hours. Representative embryos are shown for both time points. For both EpiSC and Epi-iPS cell aggregations, the GFP-expressing cells are predominantly localised internally. However, there is an obvious reduction in Oct4-GFP levels at 44 hours in the EpiSC aggregates. Following transfer of 18 blastocysts from each group, we recovered embryos at E6.5. None of seven normal-looking embryos obtained from the EpiSC aggregates showed any detectable GFP in the embryonic region. By contrast, five out of six embryos generated from Epi-iPS cell aggregates showed GFP expression in the egg cylinder. The lower panels show representative embryos from both types of aggregation. These data indicate that EpiSCs can intermingle with ICM cells during blastocyst formation, consistent with expression of the adhesion molecule E-cadherin, but differentiate or die in this environment as reflected in the loss of Oct4-GFP. Consequently, EpiSCs, unlike ES cells or iPS cells, cannot participate in subsequent embryogenesis and fail to produce chimaeras (Tesar et al., 2007).