Retargeting plasmids used in this study and identification of retargeting events. The platform line generated in WA09 contained a docking site inserted at chromosome 13q32.3, which harbored a wild-type R4 attP site upstream of a Zeocin-resistance gene lacking a eukaryotic promoter. Insulated retargeting vectors pJTI-cHS4-R4-EG (ie, iEG, A) and pJTI-cHS4-R4-CAGG (ie, iCAGG, B) had 2 double-cHS4 insulator sequences flanking the EF1α or CAG promoter-driven GFP cassette. A wild-type R4 attB site was included to mediate recombinations between the plasmids and the platform line, which had an attP site already inserted. In the retargeting vector pJTI-cHS4-R4-CAGG (B), the EF1α promoter was replaced by the CAG promoter (a combination of the cytomegalovirus early enhancer element and chicken β-actin promoter) to obtain higher GFP expression levels in hESCs and their differentiated progeny. The retargeted hESC lines had double-insulator sequences flanking the expression cassette as indicated in (C) to prevent potential silencing events from interfering with the expression of inserted genes. Correct recombination events were screened by PCR as described in [7] and verified by Southern blot analysis (D). A predicted 11 kb single band was detected in representative clones obtained from retargeting with the pJTI-cHS4-R4-EG (iEG) vector (D), and a predicated 9.7 kb single band was shown in clones retargeted with the pJTI-cHS4-R4-CAGG (iCAGG) vector (E). To ensure that the generation of the platform at chromosome 13 did not cause any disruption of expression of endogenous genes, and that the CLYBL gene remained intact, RT-PCR was performed in the parental WA09 line and the retargeted iEG and iCAGG lines. GAPDH was used as an internal control. CLYBL expression was similar in all 3 lines examined (F). EF1α, elongation factor 1 alpha; GFP, green fluorescent protein; hESC, human embryonic stem cell.

Double-insulated pJTI-cHS4-R4-EG (iEG) and pJTI-cHS4-R4-CAGG (iCAGG) vectors were successfully retargeted to the chromosome 13q32.3 site, and expressed GFP. Immunocytochemistry results of clones from both vectors are similar, and only those of iCAGG clones are shown here. A representative clone displayed the typical human hESC colony morphology (C inset), expressed GFP, and coexpressed characteristic hESC markers, including OCT4 (A–C), SSEA4 (D–F), Tra1-60 (G–I), and Tra1-81 (J–L). The GFP expression profile was identical in prolonged culture for over 20 passages for both iEG (M) and iCAGG (N). x-axis represents the GFP intensity; y-axis represents percentage of Max, an indicator of the number of cells. Red curves represent the parental cell line WA09, which did not express GFP. Green curves represent clones at passage 10. Blue curves represent clones at passage 30. Scale bar, 100 μm. Color images available online at www.liebertonline.com/scd

iEG and iCAGG clones differentiated into NSCs, neurons, and glial cells without any silencing of GFP expression. The iEG and iCAGG clones were specified toward NSCs. Representative iCAGG clones showed robust and constitutive GFP expression along the time course of differentiation, together with NSC markers PAX6 (A–C) and NESTIN (D–F), neuronal marker β3 tubulin (J–L), glial markers OLIG2 (G–I) and NKX2.2 (M–O), and astrocyte marker GFAP (P–R). Flow cytometric analysis was applied to determine the percentage of cells that retained GFP expression during differentiation (S–U). Cells differentiated after the protocol described in Materials and Methods were harvested, fixed, and stained using a Pax6 antibody that did not stain undifferentiated ESCs. Samples were gated on the basis of the appropriate forward scatter (FSC) and side scatter (SSC) to exclude cell debris (S). The majority of the cells underwent directed neural differentiation were Pax6+ (T), and 96% of which continued to express GFP (U). The shaded gray area represented the negative control cells. Quantification of coexpressed cells of GFP and neural lineage markers is shown in (V). Note that in all experiments, the GFP shown was from a native signal without any additional immunostaining using GFP antibodies, indicating the robustness of the transgene expression in our system. Arrowheads indicated representative double-labeled cells. Scale bars, 100 μm (A–F), 50 μm (G–R). NSCs, neural stem cells. Color images available online at www.liebertonline.com/scd

iEG and iCAGG clones differentiated into dopaminergic neurons, while maintaining robust GFP expression. TH staining (A) revealed the relatively high differentiation efficiency of iCAGG cells toward dopaminergic lineage. These cells continued to express GFP (B, C). (D) Higher magnification of inset in (C). Please refer to Fig. 3V for quantification of coexpression of TH and GFP. TH, tyrosine hydroxylase. Color images available online at www.liebertonline.com/scd

Double-insulated retargeted clone iCAGG differentiated into pancreatic endoderm retained expression of GFP throughout the entire differentiation process. iCAGG hESCs were subjected to a multistep differentiation protocol, and analyzed for expression of GFP (native signal, B, E, H, K, N) and coexpression of the stage-specific transcription factors (red, C, F, I, L, O) OCT4 (A) for undifferentiated cells, SOX17 (D) for definitive endoderm, HNF1B (G) and SOX17 (G, inset) for primitive gut tube, PDX1 (J) for posterior foregut, and NKX6.1 (M) for pancreatic endocrine precursors. The individual images were merged (right-hand column) to show colocalization. Flow cytometric analysis was applied to determine the percentage of cells that retained GFP expression during differentiation (P–S). No SOX17+ cells were detected in undifferentiated cells (P) or cells that were allowed to differentiate without growth factors (Q). Three days after treatment with growth factors, 17% of the cells had differentiated into SOX17+ definitive endoderm (R), ∼93% of which were positive for GFP (S). Scale bars 100 μm. Color images available online at www.liebertonline.com/scd

GFP expression was retained in insulated clones after directed differentiation toward mesodermal lineages. Markers that characterized mesodermal lineages such as SMA (A–C) colabeled with GFP after differentiation. Cells expressing cardiovascular markers PDGFRα (D–F) and Brachyury (G–I) also retained robust GFP expression. PECAM1 (CD31), a common marker for endothelial cells, was detected to be coexpressed with GFP (J–L, inset in L shows the phase image of J–L). In addition, single cells dissociated from beating clusters were analyzed by flow cytometry and showed coexpression with GFP (M). Black curve represents a control cell line that does not have GFP expression; green curve represents the iCAGG-derived beating cells. Quantification of coexpressed cells of GFP and mesodermal lineage markers is shown in (N). Scale bar, 100 μm. Color images available online at www.liebertonline.com/scd

Double-insulator elements prevented unbalanced expression of transgenes in cis. Uninsulated and insulated dual-reporter plasmids pJTI-R4-EF1α-GFP-EF1α-RFP and pJTI-R4-EF1α-GFP-(cHS4)₂-EF1α-RFP were used to transfect HEK293 cells and the expression level of GFP and RFP was compared (A). Expression of GFP and RFP was biased with GFP being expressed at a much higher level (D–F) in uninsulated clones. However, when 2 copies of cHS4 were placed between these 2 cassettes, expression of GFP and RFP was almost identical as revealed by direct fluorescence microscopy of the native signals (G–I). This was also confirmed by retargeting to the platform hESC line generated in an abnormal hESC line BG01V (J–L) and a widely used normal hESC line WA09 using an episomal expression backbone (M–O). In addition to immunocytochemistry assays, flow cytometric analysis also confirmed that similar percentages of cells expressed both reporters (∼93% for GFP+, as shown in B, ∼90% for TagRFP+, shown in C). x-axis represents the intensity of GFP or TagRFP; y-axis represents the relative number of cells. Nonengineered ESCs were used as negative controls (black curves in B and C). EG:EF1α-EmGFP, ER: EF1α-RFP. EmGFP, emerald green fluorescent protein. Color images available online at www.liebertonline.com/scd

DNA methylation status of EF1α promoters. Location of PCR amplicons for the endogenous EF1α promoter at chromosome 6 and the exogenous EF1α promoter at the chromosome 13 locus is shown in (A). GFP and phase images of a representative uninsulated EG clone named G2R and an insulated EG clone (iEG) at undifferentiated (B, C; H, I) and differentiated NSC stage (D, E; J, K) are also shown. While endogenous EF1α promoter remained unmethylated at both ESC stage and after neural differentiation, the exogenous EF1α that drove GFP had higher methylation status (F–G). An increase of DNA methylation after differentiation was also observed in insulated iEG clone, but to a lesser degree when compared with G2R (L, M). Primer sets EEF1A1 P1 and EEF1A1 P2 were used to detect methylation status of endogenous EF1α promoter. The primer set EF1α GFP P2 was used to detect methylation status of the exogenous EF1α promoter. x-axis represents elution fractions with differential methylation levels. There was a direct correlation between salt concentration in elution buffer and DNA methylation. y-axis represents percentage of total amplified DNA signal from all fractions collected. Higher bars in higher salt concentrations represent higher DNA methylation. Bars in the unbound section represented fractions that did not have any DNA methylation. Color images available online at www.liebertonline.com/scd