PiggyBac Transposon-Mediated, Reversible Gene Transfer in Human Embryonic Stem Cells

Abstract

Permanent and reversible genetic modifications are important approaches to study gene function in different cell types. They are also important for stem cell researchers to explore and test the therapeutic potential of stem cells. The piggyBac transposon from insects is a rising nonviral system that efficiently mutagenizes and mediates gene transfer into the mammalian genome. It is also characterized by its precise excision, leaving no trace sequence behind so that the genomic integrity of the mutated cell can be restored. Here, we use an optimized piggyBac transposon system to mediate gene transfer and expression of a bifunctional fluorescent reporter in human embryonic stem (ES) cells. We provide molecular evidence for transposase-mediated piggyBac integration events and functional evidence for successful expression of a transferred fluorescent protein genes in human ES cells and their in vitro differentiated derivatives. We also demonstrate that the integrated piggyBac transposon can be removed and an undisrupted insertion site can be restored, which implies potential applications for its use in gene therapy and genetics studies.

Introduction

Human embryonic stem (ES) cells hold great promise in regenerative medicine [1]. They are characterized by their ability to self-renew and differentiate into >200 cell types and tissues of the body. They are also different from other types of stem cell because they can be expanded in vitro almost indefinitely [2]. Therefore, they provide a potentially unlimited source of a variety of human tissues, not only for their potential applications in cell therapy, toxicity tests, and new drug development, but also for basic studies of stem cell and human developmental biology.

The culture and manipulation of human ES cells provide an important system for human biological studies. Over the last 2 decades, the development of genetic manipulation technologies in mouse ES cells has had a significant impact on biomedical research. These mouse studies provide a foundation for similar studies in human ES cells. Transgenic and gene-targeted human ES cell lines have been useful for studying gene function in maintaining differentiation potential, examining differentiation states, and facilitating the purification of a target cell type after in vitro differentiation [3–13]. In addition to gene ablation, there is also the potential to correct genetic defects in human ES cells. Thus, other than correlative evidence deduced from clinical studies, the genetic cause of a human disease can be determined by loss-of-function or gain-of-function evidence in a genetically controlled model [14,15]. The etiology of congenital human diseases can also be studied using in vitro differentiation assays from an undifferentiated state to terminally differentiated, specialized cell types.

Permanent genetic modifications such as introducing a transgene, targeted gene deletion, point mutation, conditional ablation, even engineering chromosomes are widely carried out in mouse ES cells [16]. However, because of an extremely low cloning rate (˜0.1%), progress on permanent genetic modifications in human ES cells has been limited. Recent discoveries on the use of neurotrophic factors or a rho-associated kinase (ROCK) inhibitor to significantly reduce human ES cell apoptosis have increased the clonal survival rate up to 30% [17,18]. These improvements make human ES cells more amendable to cell culture manipulations, thus genetic engineering of human ES cells is becoming more feasible. Although protocols for targeted gene manipulation in human ES cells have been developed, including gene targeting via homologous recombination, adeno-associated virus type 2 (AAV2) or phiC31 integrase-mediated site-specific integration, and Zn finger nuclease-mediated gene editing, experiments for permanent gene modifications in human ES cells were mainly delivered by lentiviral vector random integrations [3, 6, 7, 19–23]. However, due to the transcriptional regulatory sequences encoded in the virus terminal repeats, the viral integration events not only disrupt but can also activate gene expression around the integration sites thus complicating biological interpretation or clinical translation of these transduced cells.

Nonviral vectors such as transposable elements introduce stable integration in the genome and are a safer choice compared to virus-based systems. Recently, the Sleeping Beauty transposon was reported to mediate stable gene transfer in human ES cells [24]. For a variety of reasons, piggyBac transposons are becoming the leading transposon/transposase system for use in mammals [25,26]. Despite the low specificity of its insertion site preference (TTAA), there are several advantages for using piggyBac as a vehicle to introduce genetic components. First, the cargo size it can carry is relatively large. PiggyBac carries DNA fragments up to 14 kb yet transposition efficiency is not significantly affected [27]. Second, transposition efficiency is less affected by transposase levels compared to Sleeping Beauty and the Tol2 transposon systems [25]. Another important feature of piggyBac transposition is that its excision is precise, leaving no trace sequence behind [28,29]. This provides an option for the removal of an unwanted genetic modification made by piggyBac insertion and leaves no change in the genome as long as a re-integration event does not occur. It also allows one to revert a phenotype caused by piggyBac-mediated mutagenesis so that gene correction evidence can be provided to argue a previously introduced mutation is truly responsible for the observed phenotypic consequence.

In this report, we demonstrate that it is feasible to use the piggyBac transposon/transposase system for permanent gene transfer in human ES cells. We first provide statistical evidence to show an increased transfection efficiency of a piggyBac transposon-containing plasmid in the presence of piggyBac transposase. Second, we show sequence information of transposon–chromosomal junctions, providing molecular evidence that PBase-mediated transposition events are truly responsible for the elevated stable transfection rate. Third, we performed confocal microscopy and fluorescence-activated cell sorting (FACS) analysis to examine the expression of a fluorescent reporter cassette delivered by piggyBac to a genomic context. Fourth, we demonstrated that the transgenic reporter expression is stable in the human ES cell derivatives during an in vitro differentiation process. Finally, we removed the integrated transposon by re-introducing a helper plasmid. Genomic Southern analysis and sequences of genomic PCR products revealed a restored, intact insertion site.

Materials and Methods

Plasmid construction

Sleeping Beauty inverted repeats from pT3 [30] were sub-cloned into the multiple cloning site of the piggyBac vector pXL-BacII [31]. A human ubiquitin C promoter (UBC) [32]-driven bi-color fluorescent protein-labeling cassette, H2B-EGFP-2A-mCherry-GPI (Stewart et al., in press), flanked by 2 copies of chicken beta-globin insulators [33] was inserted between the Sleeping Beauty inverted repeats. A floxed PGKneobpA cassette [34] was cloned 5′ to the UBC-H2B-EGFP-2A-mCherry-GPI cassette in the transposon.

Human ES cell culture and electroporation

The H9 human ES cell line was obtained from the National Stem Cell Bank (WiCell Research Institute, Madison, WI). H9 cells were maintained on mitomycin C-treated SNLPB76/7 feeders [35], using human ES cell culture medium containing 20% Knockout Serum Replacement (Invitrogen, Madison, WI), 1 mM l-glutamine (Invitrogen), 0.1 mM 2-mercaptoethanol (Sigma, St. Louis, MO), 1% nonessential amino acids, and 40 ng/mL recombinant zebra fish basic fibroblast growth factor (bFGF) in DMEM-F12 (Invitrogen). Cells were incubated at 37°C in 5% CO₂ and passaged every 5–7 days using a collagenase IV (Invitrogen) protocol [3]. Immunostaining of SSEA-1, SSEA-4, TRA-1-60, and TRA-1-81 using an ES Cell Characterization Kit (Chemicon, Rosemont, IL) confirmed that cells remained undifferentiated throughout the experiment. Karyotypic analysis was performed by Clinical and Research Cytogenetic Laboratory at Texas Children's Hospital, Houston, TX.

To increase the clonal survival rate, a rho-associated kinase (ROCK) inhibitor (Y-27632; Calbiochem, San Diego, CA) was used to prevent apoptosis throughout the electroporation process. H9 cells were cultured with 10 μM ROCK inhibitor 2–4 h before electroporation. To harvest the cells, 1.5 mL of 0.25% trypsin was added to each 10-cm plate and incubated at 37°C in 5% CO₂ for 15 min. A 3.5 mL of trypsin inhibitor (1 mg/mL)-containing medium was then added, and all floating cell clumps were broken into a single cell suspension by repeated pipetting. Cells were counted and resuspended at a density of 1.1 × 10⁷ cells/mL. For each electroporation, 10 μg donor plasmid and 10 μg helper plasmid (pCMV-mPB-pA [36] or pCMV-HSB3-pA [30]) were mixed with 0.9 mL cell suspension (about 10⁷ cells). Electric pulses at 230 V with a capacitance of 500 μF were given by a Biorad GenePulser (Bio-Rad Laboratories, Hercules, CA). The time constant was between 7.2 and 8.2. After electroporation, each cuvette of cells was seeded onto a 10-cm plate with a feeder layer and was cultured with medium containing 10 μM ROCK inhibitor. On day 2 post-electroporation, cells were cultured with medium without ROCK inhibitor. On day 3–7, cells were cultured with a selection medium containing G418 at a concentration of 50 μg/mL. On day 8–13, the G418 dosage was increased to 100 μg/mL. On day 14, surviving colonies were picked individually into wells of a 96-well plate and cultured in medium without G418.

For PBase-mediated seamless removal experiment, 1 × 10⁷ transfected human ES cells (clone G4 and clone H6) were electroporated with 20 μg helper plasmid (pCMV-mPB-pA) [36]. After electroporation, cells were seeded in low density (1,000 to 3,000 cells/10-cm plate). Individual clones were identified and picked 2 weeks after. Each subclone was examined under a fluorescent microscope and expanded for G418 resistance test, Southern analysis, and genomic PCR experiments.

Southern analysis and inverse polymerase chain reaction

Ten G418-resistant ES cell clones were propagated and their genomic DNA were extracted for Southern analysis, using standard methods. Southern blots of EcoRI-digested genomic DNA were hybridized at 65°C with a ³²P-labeled DNA fragment from the GFP-coding region. The numbers of transposons that had integrated in each human ES cell clone ranged from 1 to 5.

To identify the genomic locations of the transposon integration events in the human ES cell clones, we used an inverse PCR (iPCR) strategy to clone the transposon insertion sites. Either Spiel or Pavli was used to digest 1.5 μg of genomic DNA and inactivated at 70°C for 10 min. The locations of primers within the transposon and their sequences used for iPCR are shown in Figure 1 and Table 1, respectively. UGm primers (UGm-I and UGm-N) and Bac primers (BacE and BacF) were used to amplify clone G4 Spiel-digested DNA. Neo primers (Neo-I and Neo-N) and Bac primers were used to amplify clone H6 Pavli-digested DNA. The digested genomic DNA was self-ligated using T4 DNA ligase (Roche, Nutley, CA). An aliquot of the self-ligation mixture was used as template in a PCR. The first round of PCR was performed using the following conditions: 94°C for 1 min followed by 35 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 3 min. A second round of PCR was performed using 2 μL of the PCR products from the first round of amplification as the template using an annealing temperature of 58°C. Amplified DNA fragments were fractionated and purified for direct sequencing using primer BacF. The chromosomal location of the transposon was analyzed using the Human BLAT Search program provided by the University of California, Santa Cruz (http://genome.ucsc.edu) [37].

FIG. 1.

Diagram of piggyBac transposon. Top, pXL-T3-Neo-UGm-cHS4X. Open arrowheads, piggyBac inverted repeats; filled arrows, Sleeping Beauty inverted repeats; filled arrowheads, loxP sites; Ins, chicken beta-globin insulator sequence; UBC, human ubiquitin C promoter sequence; H2B-EGFP-2A-mCherry-GPI, a bifunctional transgenic construct encoding a human histone 2B-enhanced green fluorescence protein fusion and a glyco-sylphosphatidylinositol anchor-linked mCherry fluorescent protein separated by a self-cleavable 2A peptide from the foot-and-mouth disease virus. Bottom, the relative positions of restriction sites and inverse PCR primers used in this study are indicated.

Table 1.

Inverse PCR Primer Sequences Used in This Study

Primer name	Sequence
BacE	5′-AGT GAC ACT TAC CGC ATT GAC AAG-3′
BacF	5′-TCC TAA ATG CAC AGC GAC GGA TTC-3′
Neo-I	5′-CGG CCA TTG AAC AAG ATG GAT TGC-3′
Neo-N	5′-ATG ACT GGG CAC AAC AGA CAA TCG-3′
UGm-I	5′-AAT TCC TGC AGC CCA ATT CCG ATC-3′
UGm-N	5′-TCT GAA GAG GAG TTT ACG TCC AGC-3′

Fluorescence microscopy and image processing

After G418 selection, eight human ES cell clones were picked and plated onto MatTek 35-mm glass bottom culture dishes (MatTek, Ashland, MA), allowed to grow for 5–7 days, and then examined under a Nikon 2000U Inverted Microscope with a 60x/NA1.45 objective lens (Nikon Inc., Melville, NY). Images stacks were acquired at 0.2 μm intervals using a Photometric CoolSnap HQ camera (Roper Scientific, Trenton, NJ) and then deconvoluted using AutoQuant's AutoDeblur software (Meyer Instruments, Houston, TX). For undifferentiated marker analysis, embryoid body formation, and neuronal lineage differentiation experiments, we used a Leica DMRE equipped with a Leica DFC 490 cool-CCD and a Leica MZ16FA dissection microscope. Images were captured using an XnView software.

In vitro differentiation assays

Differentiation of human ES cells into embryoid bodies and neuronal lineages was carried out using a previously established protocol [38]. In briefly, undifferentiated human ES cells (clones #9) and H9 cells were mechanically detached from the cell culture plate surface using a glass needle and then cultured on a Petri dish with human ES cell medium containing no bFGF. Medium was changed every 2 days. For neuronal lineage differentiation, day 4 embryoid bodies (EBs) (from clone #9 and H9) were seeded onto poly-d -lysine (50 ng/mL) (Sigma, St. Louis, MO)-coated dishes with differentiation medium composed of DMEM/F12 supplemented with 1% N2 (Invitrogen, Madison, WI). For the first 4 days, 10 ng/mL bFGF (Invitrogen, Madison, WI) was added to the differentiation medium. The differentiation results were examined 13 days after EB attachment.

Flow cytometry

In all cases, fresh human ES cell colonies were first trypsinized for 10 min and then neutralized with DMEM medium with 10% FBS. The cells (10⁵ to 5 × 10⁶ cells) were washed once and resuspended in phosphate-buffered saline (PBS). These specimens were strained through BD Falcon cell strainers (REF 352235; Becton Dickinson [BD], San Jose, CA) and then kept on ice before acquisition. The analysis was performed with a FACSAria II (Becton Dickinson [BD], San Jose, CA). In all samples, >10,000 cellular events from the defined cell cluster were analyzed per tube. Events with very low forward scatter, which were likely to include degenerated cells, were excluded from the analysis by forward versus side scatter gating. Propidium iodide (PI) staining was applied in control (untransfected cells) to help dead cells exclusion.

Results and Discussion

Structure of the piggyBac transposon used in this study

Previously, the piggyBac transposon was shown to be functional in human transformed cell lines in the presence of a helper plasmid encoding piggyBac transposase (PBase) [25]. Whether PBase-mediated piggyBac transposition requires the presence of other tissue type-specific cofactors from the host is not clear [26]. To test whether PBase-mediated transposition can occur in human ES cells, we utilized a previously described optimized piggyBac system [36]. Our piggyBac transposon construct is shown in Figure 1. We engineered the transposon to carry a bifunctional fluorescent protein expression cassette that includes a human ubiquitin C (UBC) promoter for strong expression of a fluorescent protein transgene, composed of a human histone 2B and enhanced green fluorescence protein (H2B-EGFP) fusion gene linked to a monomeric Cherry fluorescent protein (mCherryFP) and glycosylphosphatidylinositol anchor sequence (GPI) fusion gene by a foot-and-mouth disease virus 2A peptide [32,39,40]. The 2A peptide-mediated self-cleavage dissociates the H2B-EGFP fusion protein, which marks the nucleus or the chromosomes during mitosis, from the mCherryFP-GPI protein, which marks the plasma membrane and vacuoles [41,42]. Four copies of chicken beta-globin insulators (2 on each side of the fluorescent reporter cassette) were included to minimize the effects of chromosomal integration sites on transgene expression [33]. The transposon also has a loxP-floxed PGKneobpA cassette that allows selection of stable transposon integration events by G418 [34]. The helper plasmid contains the PBase coding sequence optimized for translation in mammalian cells (mPB) [36]. In addition to the piggyBac inverted terminal repeats (ITRs), we also placed Sleeping Beauty ITRs on both ends of the transgene expression cassette and the drug selectable cassette. This allows a comparison between the 2 transposon systems since the cargos carried are almost identical except the piggyBac transposon is slightly bigger than the Sleeping Beauty transposon.

PBase-mediated transposition in human ES cells

A pilot experiment to test whether the piggyBac transposon system is functional in human ES cell was performed by co-electroporating the pXL-T3-Neo-UGm-cHS4X transposon plasmid and a pCMV-mPB-pA helper plasmid at different quantities. A pCMV-HSB3-pA helper plasmid expressing HSB3 was also used to do a comparison between the piggyBac and the Sleeping Beauty systems [30]. The results are shown in Supplementary Figure 1 (Supplementary materials are available online at www.liebertonline.com/scd). In our preliminary test, the mPB transposase mediated a significant increase in the permanent transfection events (∼120 CFU/electroporation) whereas the HSB transposase was moderate (∼40 CFU/electroporation). We also observed that increasing the quantities of both the transposon and the helper plasmids used for electroporation could result in significantly more transposition events (Supplementary Fig. 1B).

To show that the co-electroporation of mPB-expressing plasmid does change the permanent transfection rate, uncut piggyBac transposon plasmid (20 μg) alone or with mPB helper plasmid (20 μg) was co-electroporated into the H9 human ES cell line. After G418 selection for 2 weeks, G418-resistant human ES cell clones were either stained by methylene blue directly on the plates for colony counts (Fig. 2B) or picked and propagated for further study. Figure 2A summarizes the numbers of G418-resistant colonies from 3 independent electroporation experiments. The data revealed that electroporating the piggyBac transposon alone results in rare plasmid insertion events (1.02 + 0.15 CFU/pM of plasmid; n = 3), whereas co-electroporating mPB helper plasmid resulted in a significant increase (∼90-fold) in the transfection rate (90.99 + 18.80 CFU/pM of plasmid; n = 3).

FIG. 2.

Evidence for mPB-mediated transposition in H9 human embryonic stem (ES) cells. (A) Number of G418-resistant human ES cell colonies after electroporation of 20 μg of pXL-T3-Neo-UGm-cHS4X in the absence or presence of 20 μg of mPB. Error bars, standard deviation; pM, pico mole. (B) Methylene blue staining of tissue culture plates after electroporation of the transposon plasmid with or without the mPB helper plasmid in human ES cells after 2 weeks of G418 selection. (C) Southern analysis of piggyBac transposition events in human ES cells. EcoRI genomic digests and an EGFP probe were used. Variation of the EcoRI fragment size illustrated independent piggyBac transposition events. (D) Molecular evidence for PBase-mediated transposition provided by sequence analysis of inverse PCR products from 12 independent human ES cell clones. Sequence in lower case indicated inconsistency with the published human genome sequence (hg19).

To verify that the significant increase in the transfection rate with the helper plasmid was achieved by PBase-mediated transposition, we isolated genomic DNA from 34 different G418-resistant clones. EcoRI genomic digests and an enhanced green fluorescence protein (EGFP)-coding sequence probe (or BamHI genomic digests and a neomycin phosphatase (neo)-coding sequence probe) were used for Southern analysis. Transposon insertions ranging from 1 to 5 copies were observed in the clones (Fig. 2C). The transposon-chromosomal junctions were cloned by inverse PCR using primer sequences complementary to the transposon sequence near the 5′ terminal repeats (5′-TR) (Fig. 1, Table 1). The PCR products were gel-purified and directly sequenced using a BacF oligonucleotide primer. A total of 12 insertion sites were partially cloned. The sequencing results revealed an intact donor plasmid terminal repeat sequence and a TTAA piggy-Bac transposon insertion site followed by junctional genomic sequence (Fig. 2D). Sequence tag information was used to locate the transposon insertion site within the genome by Human BLAT Search (http://genome.ucsc.edu) [37]. The Blat results are summarized in Table 2. Within the 12 insertion sites cloned, 10 of them were unambiguously assigned to unique genomic locations. Three of the 10 insertion sites were located in regions where no genes or spliced transcripts have been annotated. Whereas 5 of the other 7 inserted in Ensembl annotated genes and the other 2 hit spliced ESTs. This observation of a high proportion of intragenic insertions (5/10 or 7/10) was consistent with a previous report stating that there is a significant bias toward integration in intragenic regions in the piggyBac transposon system [43].

Table 2.

Summary for Human Blat Results of Partially Cloned Insertion Sites

Clone name	Insertion sitea	Gene or transcript disrupted	Orientation
#4	Chr 5: 44061378-81	BG334794	(+)
#9	Chr 1: 180338211-14	—	(+)
#10	No hit
#11	Chr 6: 55360211-14	GDNF family receptor alpha-like (GFRAL)	(+)
#13	Chr 3: 69917943-46	Microphthalmia-associated transcription factor (MITF)	(+)
#14	Chr 8: 93706172-75	AA885105	(+)
#16	Chr 2: 86182284-87	Polymerase (RNA) I polypeptide A (POLR1A)	(−)
#18	Chr 10: 124117808-11	—	(−)
#22	Chr 10: 84695927-30	Neuregulin 3 (NRG3)	(−)
#17	Chr 11: 71138024-27	LTR repeat	(−)
H6	Chr 18: 73749986-89	—	(+)
G4	Chr 3: 180692952-55	Fragile X mental retardation, autosomal homolog 1 (FXR1)	(−)

^aUCSC Genome Browser (http://genome.ucsc.edu) on Human February 2009 Assembly (hg19).

To show that piggyBac transposition events do not cause chromosomal abnormality, 2 clones, G4 and H6, carrying single copy transposon insertion, were used for insertion site sequence and karyotypic analysis. Genomic sequences of both ends of the insertion sites were obtained (Fig. 3D, data not shown). In both clones, the junctional genomic sequences from the 2 ends can reconstitute an intact genomic sequence and the characteristic TTAA insertion site could be recognized. Karyotypic analysis results appeared normal in both clones (Supplementary Fig. 2). The insertion of a piggyBac transposon did not influence the differentiation ability of the human ES cells because the immunostaining of ES cell markers of a transposed clone (clone #9) revealed an undifferentiated human ES cell expression pattern (Supplementary Fig. 3).

FIG. 3.

PBase-mediated transposon excision in human embryonic stem (ES) cells. (A) PCR primers used to clone both ends of the transposon–chromosomal junctions and a 3-primer genotyping strategy for clone G4. (B) PCR genotyping using a mixture of G4_F2, G4_R1, and PRIR_F2 primers on H9 human ES cell, transgenic human ES cell clone G4, and a G418-sensitive sub-clone (G4∆39) recovered after a transient transfection of a helper plasmid. Details are provided in Supplementary Figure 6. (C) Genomic sequence analysis at the clone G4 insertion site of a H9 human ES cell. (D) and (E) Transposon–chromosomal junction sequence analysis on both the 3′ terminal repeat (3′-TR) side and the 5′ terminal repeat (5′-TR) side of clone G4. (F) Genomic sequence analysis at the clone G4 insertion site of a G4∆39 subclone provided evidence for a seamless removal of the inserted piggyBac transposon in a human ES cell. Color images available online at www.liebertonline.com/scd.

PiggyBac-mediated stable gene expression in human ES cells

To demonstrate that piggyBac-transferred genes can be active in a human ES cell genomic context, we used confocal microscopy to examine the expression of the dual fluorescent protein reporter. We randomly picked 24 G418-resistant clones for clonal expansion. Although fluorescence was observed in all clones examined, the intensity of fluorescence varied between clones (Supplementary Fig. 4). The fluorescence intensity variance was at least not only due to differences in the copy number of transposon insertion(s) but also due to the incomplete sheltering effect of the insulator sequences since we observed higher intensity in a clone (#14) carrying a single copy transposon than many of those carrying multiple copies. As shown in Figure 4A, G418-resistant human ES cells (clone B9) displayed red fluorescence (mCherry-GPI) in their plasma membranes and vacuoles, highlighting the shape of the cell and internal membranous structures. At the same time, the histone moiety of the H2B-EGFP fusion localizes the green fluorescence in nuclei. During mitosis, H2B-EGFP fusion highlights chromosomes, providing a tool to monitor cell division (Fig. 4A). When used in the fluorescence-activated cell sorting (FACS) assays, the pXL-T3-Neo-UGm-cHS4X- transfected cells displayed a characteristic correlation between the green and the red fluorescence intensities (Fig. 4B). This unique green/red fluorescence intensity ratio was consistent with a fixed molar ratio between the EGFP and mCherry fusions that resulted from a post-translational self-cleavage mediated by the 2A peptide [41,42]. It also allows the separation of the transfected cells emitting dual fluorescence from those auto-fluorescent dead cells without the assistance of a propidium iodide staining control (Supplementary Fig. 5A and 5B). The FACS experiment further indicated that the fluorescence intensity was a clone-specific character (Supplementary Fig. 5C). It is most likely influenced by extrinsic factors introduced by the transposition events such as position effects around the insertion site(s) or the transposon copy numbers rather than intrinsic factors such as cell cycle or other cellular phenomena.

FIG. 4.

Functional evidence for piggyBac-mediated gene transfer in undifferentiated H9 human embryonic stem (ES) cells. (A) An optical slice documents the expression of the H2B-EGFP-2A-mCherry-GPI cassette in H9 human ES cell transferred by the piggyBac transposition. Green fluorescence from H2B-EGFP defines the shape of nuclei and condensed chromatin when the cells undergo mitosis and red fluorescence from mCherryFP-GPI defines the cell shape and membrane vesicles in the cytoplasm. (B) Fluorescence-activated cell sorting (FACS) analysis on clone #9. The narrow, straight, linear distribution on the 2-dimensional plot indicated a characteristic correlation between the green and the red fluorescence intensities. Color images available online at www.liebertonline.com/scd.

Stable gene expression after in vitro differentiation

One important feature of human ES cells is their ability to form embryoid bodies (EBs) and differentiate into all 3 germ layers. To further evaluate whether the ubiquitin C promoter introduced by the piggyBac system was able to direct a universal transgene expression pattern in the ES cell derivatives, we performed a random in vitro differentiation experiment as well as a directed in vitro differentiation toward the neural lineage. In our random in vitro differentiation experiment, embryoid bodies were generated and cultured using clone #9 transgenic human ES cells. Day 11 EBs, in which normally molecular markers for all 3 germ layers are detectable, were collected and examined. Under a fluorescent microscope, all the cells in day 11 EBs observed expressed the characteristic nuclear green and membranous red fluorescence no matter where they were located within the EB (Fig. 5A and 5B). When the EBs were dissociated using an enzymatic method and used for FACS analysis, all the cells within the EBs emitted green fluorescence at a intensity similar to their undifferentiated parental ES cell clone (#9) that can be easily distinguished from the H9 human ES cells (Fig. 5C). The stability of the piggyBac-mediated transgene expression was further evaluated by differentiating transfected clones toward neural lineage directly. Thirteen days after the EBs attached to the poly-d-lysine-coated plates, which was followed by a 4-day 10 ng/mL bFGF treatment, we started to see an early sign of neural differentiation characterized by the formation of a neural rosette structure (Fig. 5D). All the ES cell derivatives, whether they were involved in the neural rosette formation or not, presented dual fluorescence (Fig. 5E). When the directly differentiated cells were trypsinized and used in a FACS experiment, all of them emitted green fluorescence of intensity close to those from randomly differentiated EBs. But differentiated cells from an H9 EB did not emit green fluorescence (Fig. 5F). These results demonstrated that our gene expression cassette delivered by the piggyBac system was able to mediate a universal, stable dual fluorescence gene expression in both undifferentiated human ES cells and their differentiated derivatives. These dual fluorescence-labeled ES cells and their derivatives should be useful in lineage tracing to discuss their in vivo differentiation potentials after transplantation. It could also be used in embryoid body formation experiments to study the dynamics in a directed or controlled in vitro differentiation process [44].

FIG. 5.

Morphological change and stable transgene expression of piggyBac transposed human embryonic stem (ES) cells during in vitro differentiation. (A) Cell morphology in day 11 embryoid bodies (EBs) generated from clone #9 was evaluated by phase-contrast microscopy. (B) And their ability to emit dual fluorescence was examined by fluorescent microscopy. (C) The fluorescence-activated cell sorting (FACS) analysis on enzymatically disrupted EBs illustrated that all the cells within the EBs (#9 EB d11) emitted green fluorescence of intensity similar to their undifferentiated parental ES cell clone (#9). (D) On directed neural differentiation day 17, an early sign of neural differentiation characterized by the formation of a neural rosette structure were found. (E) All the ES cell derivatives, whether they were involved in the neural rosette formation or not, presented dual fluorescence. (F) The FACS experiment revealed that almost all of them (#9 ND d17) emitted green fluorescence at intensity close to those from randomly differentiated EBs (#9 EB d11). Color images available online at www.liebertonline.com/scd.

PBase-mediated transposon excision in human ES cells

One of the most attractive features to use piggyBac in stem cell studies is its ability to “jump” out of the genome without leaving a trace [28,29]. Recently, piggyBac transposon was used to generate induced pluripotent stem (iPS) cells from mouse embryonic fibroblasts. Its subsequent removal from the genome was achieved by transient transfection of a helper plasmid [45,46]. Although piggyBac-mediated human iPS cell generation was reported, subsequent seamless removal was not documented [45,47] (Supplementary Fig. 6). To test whether piggyBac-mediated gene transfer could be reversed without causing any damage to the genome, we transiently transfected a helper plasmid to catalyze the jumping out events in 2 transgenic human ES cell clones (G4 and H6). The integrity of the transposon–genomic junctions of clone G4 was checked by obtaining transposon–chromosomal junctional sequences from both ends of the inserted transposons (Supplemental Material 6; Fig. 3D and 3E). A CMV promoter-driven mPB expression vector was electroporated into the ES cells to allow a transient piggyBac transposase expression. The electroporated ES cells were then seeded onto feeder plates at a density of 1,000 to 3,000 cells/10-cm plate. Two weeks later, individual ES cell colonies were picked, expanded, and checked for their ability to emit dual fluorescence and their resistance to G418 selection. Ten colonies were picked and tested for the seamless removal experiment using the G4 clone, eight of them emitted fluorescence and the other 2 became sensitive to G418. Both the Southern analysis and a 3-primer PCR genotyping strategy confirmed that these 2 G418-sensitive subclones (G4Δ25, G4Δ39) no longer carry a transposon in the previously cloned insertion site (data not shown; Fig. 3B). Analysis on the genomic PCR product surrounding the insertion site in these subclones revealed an intact, reconstituted genomic sequence (Fig. 3F). The experiment on clone H6 resulted in similar conclusion except the pop-out rate is higher (all 12 colonies picked did not emit fluorescence and became sensitive to G418). These results suggested that the piggyBac transposon system could mediate reversible gene transfer in human ES cells.

In this report, we provide evidence that piggyBac transposase-mediated transposition can occur in human ES cells. Thus, the piggyBac transposon can serve as a vehicle to transfer exogenous expression cassettes into human ES cell. The piggyBac system is known to handle large cargo size (9–14 kb) and its transposition efficiency is less dependent on transposase expression levels [25]. In our study, we used a cargo of ∼11 kb. Although the piggyBac insertion site provides low sequence specificity (TTAA), piggyBac-mediated genetic modification is reversible because its transposition leaves no footprint behind [28,29]. Therefore, it has been considered to be a potentially useful tool to correct genetic deficiencies in isolated stem cells [26]. To combine this gene transfer technology with other regulated gene expression systems, such as a tissue-specific promoter or tetracycline-regulated systems, RNA interference technologies to knockdown endogenous transcripts will provide alternative approaches to study human ES cell biology [22,45,48]. Engineering novel piggyBac transposases with the ability to recognize unique sequences in the human genome by adding DNA-binding domains, such as GAL4 DBD or other Zn finger motifs, should be of great interest to extend the use of the piggyBac system for therapeutic purposes [21,25,26,49,50].

Acknowledgments

We thank Dr. Allan Bradley for providing the optimized piggyBac transposon system, Drs. Thomas P. Zwaka and Marion Dejosez for technical guidance on human ES cell culture and genetic manipulation. We are also grateful to Ms. Cordelia P. Conley for her administrative support. H.F. was supported by the M.D. Anderson Summer Science Program for College Students. DNA sequencing was provided by the M.D. Anderson Cancer Center DNA Analysis Core Facility supported by the National Institutes of Health (NIH) Cancer Center Support Grant CA16672. This study was supported by NIH grants HD30284, GM81627 and the Ben F. Love endowment to R.R.B. This study was also supported by grants from National Science Council in Taiwan to Y.T.C (NSC 98-2314-B-002-043-MY2 and NSC 99-2627-B-002-005-) and to C.L.C.(NSC 97-3111-B-002-004-).

Author Disclosure Statement

The authors declare that they have no competing interests.