|Home | About | Journals | Submit | Contact Us | Français|
Patient-specific induced pluripotent stem cells (iPSCs) derived from somatic cells provide a unique tool for the study of human disease, as well as a promising source for cell-replacement therapies. However one crucial limitation has been the inability to perform experiments under genetically defined conditions. This is particularly relevant for late age-onset disorders where in vitro phenotypes are predicted to be subtle and susceptible to significant effects of genetic background variations. By combining zinc-finger nuclease (ZFN)-mediated genome editing and iPSC technology we provide a generally applicable solution to this problem by generating sets of isogenic disease and control human pluripotent stem cells that differ exclusively at either of two susceptibility variants for Parkinson’s disease by modifying the underlying point mutations (A53T/E46K) in the α-synuclein gene. The robust capability to genetically correct disease-causing point mutations in patient-derived hiPSCs represents not only a significant progress for basic biomedical research but also a major advancement towards hiPSC-based cell-replacement therapies.
Extraordinary excitement over progress in the ability of genomic medicine to connect specific genotypes to disease predisposition has been tempered by the host of challenges in translating such correlations to specific treatments. There is consensus in the field that modeling “diseases in a dish” is one of the most promising approaches to address this crucial problem (Vogel, 2010). Human induced pluripotent stem cell (hiPSC) technology, which enables the epigenetic reprogramming of human somatic cells into an embryonic stem cell-like state followed by differentiation into any cell type of the body, is being developed as a key component of such in vitro disease modeling (Dimos et al., 2008; Park et al., 2008; Soldner et al., 2009; Takahashi et al., 2007; Yu et al., 2007). In principle, patient-specific iPSCs that carry all disease-relevant genetic alterations could provide researchers with a unique opportunity to study the cellular and molecular mechanisms of monogenic and complex diseases in relevant cell types in vitro with the potential to identify alternative treatments (Daley and Scadden, 2008; Saha and Jaenisch, 2009).
However, so far only a few such studies have identified disease-related phenotypes, mostly in rare, early age-onset or metabolic diseases (Brennand et al., 2011; Ebert et al., 2009; Itzhaki et al., 2011; Lee et al., 2009; Marchetto et al., 2010; Rashid et al., 2010). Due to the robust and rapid manifestation of these disorders, in vitro models are more likely to display clear differences when compared to healthy donor controls. In contrast, late age-onset disorders, such as Parkinson’s and Alzheimer’s disease with long latency and slow progression of cellular and pathological changes in vivo are expected to show only subtle phenotypes in vitro. To distinguish these subtle but disease-relevant phenotypical changes from unpredictable background-related variations could prove difficult due to the lack of genetically matched controls. Commonly used control cells from healthy donors represent an approximate solution at best because individual hESC and hiPSC lines display highly variable biological characteristics such as the propensity to differentiate into specific functional cells (Bock et al., 2011; Boulting et al., 2011; Soldner et al., 2009). The basis for these profound differences is manifold and includes (i) differences in genetic background; (ii) the process of cell derivation (Lengner et al., 2010) and (iii) in the case of hiPSCs, variegation effects and residual transgene expression of the viral vectors used to induce reprogramming (Soldner et al., 2009) and genetic alterations introduced during the reprogramming process (Gore et al., 2011; Hussein et al., 2011). Variable genetic background presents a particularly significant impediment to in vitro disease modeling approaches because it is not possible to control for effects from genetic modifier loci. Even mutations that cause the most prevalent “monogenic” diseases, including sickle-cell anemia, cystic fibrosis and dominant forms of familial Parkinson’s disease, are susceptible to significant epistatic effects of genetic background which result in incomplete penetrance, and variable age of onset and disease progression (Lees et al., 2010; Summers, 1996).
Therefore, for the “disease in a dish” approach to be successful, it is essential to set up experimental systems in which the disease-causing genetic lesion of interest is the sole modified variable. However, the unresolved problem of genetically manipulating pluripotent human cells has prevented the creation of such genetically defined human model systems. Recently, we and others used engineered zinc finger nucleases (ZFNs) to drive efficient targeted integration of selectable markers into hESCs and hiPSCs (Hockemeyer et al., 2009; Zou et al., 2009). Employing this technology (referred to as “genome editing”), we present here a generally applicable solution to this key problem by demonstrating the generation of a panel of isogenic disease and control cell lines from hESCs and hiPSCs that differ exclusively at well validated susceptibility variants for Parkinson’s disease (PD) by genetically modifying single base pairs in the α-synuclein gene.
PD, the second most common late age-onset neurodegenerative disorder, is characterized primarily by major loss of nigrostriatal dopaminergic neurons and the presence of proteinacious inclusion bodies (Lewy Bodies) in affected cells. The discovery of mutations linked to rare forms of familial PD, such as dominant mutations in α-synuclein (A53T, E46K, A30P), the major component of Lewy bodies, has provided vital clues in understanding the molecular pathogenesis not only of the rare familial but also of the more prevalent sporadic forms of the disease (Lees et al., 2010; Schulz, 2008). However, cellular and transgenic animal models expressing such mutants only partially recapitulate PD pathology (Dawson et al., 2010). In order to develop a genetically defined human in vitro model of PD, we sought to generate a panel of control and disease-related cell lines by either deriving hiPSCs from a patient carrying the A53T (G209) α-synuclein mutation followed by the correction of this mutation or, alternatively, by generating either the A53T (G209A) or E46K (G188A) mutation in the genome of wild-type hESCs.
Genome editing with engineered ZFNs relies on a double-strand break (DSB) introduced by the nucleases. The ability to precisely target DSB to an investigator-specified site is critical because point mutations are transferred with maximal efficiency from episomal donors into the position of the DSB itself (Elliott et al., 1998; Goldberg et al., 2010). We engineered a panel of ZFNs that introduces a DSB precisely at nucleotide base 209 (site of A53T mutation) in exon 3 of the α–synuclein gene (Figure 1A,B and Table S1 and S2). Screening of the ZFNs in transformed human cells demonstrated editing of up to 18% of α-synuclein alleles precisely at the intended site (Figure 1A).
Guided by our previous results on highly efficient targeted integration of genes in hESCs and hiPSCs (Hockemeyer et al., 2009), we initially considered a drug-selection-based strategy to introduce the PD-causing A53T (G209A) mutation into the endogenous α-synuclein locus in hESCs. However, for many disease-related mutations that are located within protein coding exons, including α–synuclein A53T (G209A), the same approach cannot be used, since insertion of the selection marker would disrupt expression of the targeted gene. We therefore devised an alternative strategy based on the insertion of a loxP site flanked puromycin resistance gene in the adjacent intron 23 bases downstream of the A53T α–synuclein mutation and the DSB (Figure 1B). For this gene editing strategy, a correct targeting event followed by Cre-recombinase mediated excision of the selection cassette is expected to result in a single base pair change that creates the A53T (G209A) mutation in exon 3 of α-synuclein, with a remaining single loxP site in the following intron (Figure 1B).
The targeting donor construct (Syn-A53T-loxP-pGK-puro-loxP) comprising approximately 600bp homology on each side of the ZFN targeted site carrying the A53T (G209A) mutation (Figure 1B), together with 4 distinct ZFN pairs (Table 1, Figure 1A and Table S1 and S2) were electroporated into two different hESC lines (BGO1 and WIBR3). Southern blot analysis of individual single-cell-derived puromycin-resistant clones using probes 5′ and 3′ external to the donor homology region demonstrated the disruption of the genomic locus and integration of the targeting donor vector with a frequency of at least 25% (Figure 1C, Figure S1A and Table 1). Further analysis using an internal probe against the 3′ targeting arm of the donor vector (Figure S1B) and against the ampicillin resistance gene (Figure S1C) revealed integrations of additional donor-derived vector sequences into the target locus, presumably via a hybrid homology directed repair (HDR)-end joining based process such as described previously (Richardson and Jasin, 2000).
Three out of 336 puromycin resistant clones showed the correct modification of the targeted genomic locus by southern blot (Figure 1C, Figure S1 and Table 1), which was further confirmed by sequencing after Cre-mediated excision of the selection cassette (Figure 1D). Two out of the three clones carried a small deletion in the second allele as a result of ZFN mediated gene disruption. The correctly targeted clone with a non-disrupted wild-type allele (WIBR3-SNCAA53T/WT-1A/-1C) displayed a normal karyotype (Figure S1D) and maintained a pluripotent state as indicated by the uniform expression of pluripotency specific marker proteins (Figure 1E) and the ability to form teratomas comprised of cell types originating from all three developmental germ layers (Figure 1F). The stable integration of the transiently transfected ZFN or Cre-recombinase expressing plasmids was excluded by southern blot analysis (Figure S3). Furthermore, using an embryoid body (EB) based protocol to induce neural differentiation, we were able to efficiently derive dopaminergic thyrosine hydroxylase (TH) expressing neurons from the targeted hESC line (Figure 1G). To verify that the loxP site remaining after Cre-mediated excision of the selection cassette does not interfere with the splicing or gene expression of α-synuclein, we differentiated the parental and targeted hESC lines into neurons in order to induce expression of α-synuclein. Mutation analysis RT-PCR (Polymeropoulos et al., 1997) confirmed that the levels and ratio of expression of the wild-type and the A53T-mutated transcript in the targeted cell line were similar to those observed in neurons derived from A53T-patient-specific hiPSCs (WIBR-iPS-SNCAA53T(1lox)) (Figure 1H).
In order to increase the targeting efficiency by reducing non-targeted integrations of the donor vector, as well as integration of donor vector sequences that are outside of the homology arms at the site of ZFN cleavage, we employed a positive-negative selection strategy (Capecchi, 1989) by incorporating the herpes simplex virus thymidine kinase (HSV-TK) and diphteria toxin A-chain (DT-A) into the vector backbone (Figure 1B). Using this strategy, 9 out of 41 puromycin- and ganciclovir-resistant colonies resulted in a correctly targeted allele (Figure 1I and Table 1). Four out of these clones had no disruption of the second, wild-type, allele and 1 out of the 41 clones resulted in correct targeting and insertion of the A53T (G209A) mutation into both alleles (WIBR3-SNCAA53T/A53T). The targeted clones initially identified by southern blot analysis were confirmed by sequencing of the genomic locus (data not shown). None of the clones integrated the transiently transfected ZFNs (Figure S3). This single step biallelic modification of a disease-relevant locus represents a unique tool to study the role of mutant α-synuclein in the absence of the wild-type protein. Individuals homozygous for this mutation have not been described and the study of homozygous mutant cells may provide new insights into the pathogenesis of PD.
The experiments described so far, while successful in transferring a desired point mutation to the native locus, required the integration of a selectable marker into a neighboring intron. Dependent on the location of the desired editing events relative to intron/exon junction, such strategy may not be applicable to all genes; in addition, while the selectable marker can be excised using Cre-recombinase, the remaining loxP site represents an additional non-required genetic alteration with possible unpredictable effects. We therefore turned to an approach aimed at generating genetically pristine hESCs that contain no exogenous sequences other than the edited base (Urnov et al., 2005) for introducing the disease-causing point mutation (Figure 2A). Given the high gene editing activity of the ZFNs (Figure 1A, Figure S1A and Table 1), we constructed a donor vector lacking a selection cassette, consisting only of ~1 kb homology flanking the ZFN cleavage site carrying the A53T (G209A) point mutation in order to insert the mutation in the endogenous α-synuclein locus in hESCs (Figure 2A). The hESC line BGO1 was electroporated with the donor construct together with ZFNs and an eGFP expressing plasmid, which allows transfected cells to be enriched by fluorescence-activated cell sorting (FACS). Colonies derived from single eGFP expressing cells were screened by southern blot analysis using an A53T (G209A) allele specific Tsp45I restriction digest. Three out of 240 BGO1 clones showed the A53T (G209A) allele specific restriction pattern, indicative of an accurate genetic alteration event resulting in a A53T (G209A) mutation at the endogenous genomic locus (Figure 2B). Further analysis by PCR genotyping (Figure 2C) and sequencing of the genomic locus (Figure 2D) confirmed one correctly targeted clone (Table 1) with the expected single base pair change of nucleotide 209 on one allele and an unaffected second allele, resulting in a A53T mutated cell line on a genetic BGO1 background (BGO1-SNCAA53T/WT). The targeted cell line maintained a pluripotent state as indicated by the uniform expression of pluripotency specific markers (Figure 2E), the ability to form teratomas (Figure 2F) and to differentiate into dopaminergic neurons in vitro (Figure 2G). The stable integration of the ZFNs and GFP expression plasmids was excluded by southern blot analysis (Figure S3).
Recent studies suggest, that short single-stranded oligodeoxynucleotides (ssODNs) instead of double-stranded donor plasmids can be used as an alternative DSB repair-template for ZFN driven genome editing (Radecke et al., 2006; Radecke et al., 2010). As it has been shown, that point mutations can be transferred distant from the ZFN induced DSB, (Elliott et al., 1998; Goldberg et al., 2010) we wanted to test the possibility to introduce a second PD-causing mutation (E46K/G188A) into Exon 3 of α-synuclein (Zarranz et al., 2004) and therefore designed a 114bp ssODN carrying the disease relevant G188A base pair change, located 21 bases upstream of the A53T (G209A) mutation (Figure 3A). The hESC lines BGO1 and WIBR3 were electroporated with ZFNs and an eGFP expressing plasmid together with the ssODNs instead of a double-stranded donor vector (Figure 3A). Individual single cell derived clones from FACS sorted eGFP expressing cells were screened by PCR followed by the mutation specific StyI restriction digest. Four out of 480 WIBR3 clones and one out of 240 BGO1 clones showed the E46K (G188A) allele specific loss of the StyI restriction site as verified by PCR genotyping (Zarranz et al., 2004) and southern blot analysis (Figure 3B,C), indicative for the accurate genetic alteration of the endogenous α-synuclein genomic locus. Further analysis by sequencing of the genomic locus confirmed the correct insertion of the E46K (G188A) mutation without any additional alteration of the wild type or mutated allele adjacent to the ZFN cleavage site (Figure 3D,E). The targeted cell lines maintained a pluripotent state as indicated by the uniform expression of pluripotency specific marker proteins (Figure 3F,G). The efficiency of correctly genome edited clones using ssODNs as template was comparable to that of using double-stranded donor plasmids (Table 1).
Fibroblasts obtained from a patient carrying the A53T (G209A) α-synuclein mutation were reprogrammed using previously described doxycycline-inducible and Cre-recombinase excisable lentiviral vectors (Hockemeyer et al., 2008; Soldner et al., 2009). The resulting hiPSCs before (WIBR-iPS-SNCAA53T(2lox)) (Figure 4) and after Cre-mediated excision of the reprogramming factors (WIBR-iPS-SNCAA53T(1lox)) (Figure S2) displayed all basic properties of pluripotent cells as indicated by the uniform expression of the pluripotency marker proteins (Figure 4A and Figure S2B), the ability to form teratomas (Figure S2C) and a normal karyotype (Figure 4B). Sequencing of the genomic α-synuclein locus confirmed the A53T (G209A) mutation in the patient derived hiPSCs (Figure 4D). Furthermore, the cells were shown to differentiate into TH expressing dopaminergic neurons (Figure 4E).
In order to genetically repair the A53T (G209) mutation in the PD patient-derived hiPSCs we employed a selection-free targeting strategy as described above for hESCs with the only difference of using a wild-type sequence containing donor vector (Figure 2A). Six out of 240 WIBR-iPS-SNCAA53T clones demonstrated the loss of the A53T specific Tsp45I restriction site by southern blot screening, which is either the result of a ZFN-induced DSB followed by non-homologous error-prone end-joining or HDR based correction of the allele (Figure 5A). Further analysis by PCR genotyping (Figure 5B) and sequencing of the genomic locus (Figure 5C) confirmed one correctly repaired patient-derived hiPSC lines (WIBR-iPS-SNCAA53T-Corr, Table 1) with the expected single base pair change of nucleotide 209 of α-synuclein. Finally, to prevent uncontrollable effects from residual expression of the reprogramming transgenes (Soldner et al., 2009), we excised the reprogramming vectors from the corrected patient derived hiPSCs (Figure S2D), which subsequently displayed a normal karyotype (Figure 5E), maintained a pluripotent state as indicated by the uniform expression of the pluripotency markers (Figure 5F) and the ability to form teratomas (Figure 5G). Stable integration of ZFNs, Cre-recombinase and GFP expression plasmids was excluded by southern blot analysis (Figure S3). The genetic repair of the A53T mutation in the patient-derived hiPSCs did not compromise the ability to differentiate into TH expressing dopaminergic neurons (Figure 5H). To further validate accurate editing of the α-synuclein locus in the repaired patient-derived hiPSC line (WIBR-iPS-SNCAA53T-Corr), we performed mutation analysis RT-PCR of α-synuclein after neuronal differentiation confirming the loss of expression of the mutated A53T (G209A) transcript (Figure 5D).
A potential limitation of ZFNs mediated genome editing is the induction of DNA strand breaks at sequences other than the intended target site. To examine off-target modifications in the edited cell lines we initially determined the DNA binding specificity for each ZFN employed in this study by SELEX analysis as described previously (Hockemeyer et al., 2009; Perez et al., 2008) (Table S3). This allows for the identification of the most probable off-target cleavage sites genome-wide (Table S4). A Surveyor endonuclease (Cel-1) assay was subsequently performed to reveal any potential nonhomologous end joining (NHEJ)-mediated indels for a large panel of putative off-target sites in several of our edited cell lines (Table 1, Figure S4). While this analysis has revealed rare bona fide off-target modification by ZFNs in other studies (Hockemeyer et al., 2009; Perez et al., 2008) we saw no evidence of off-target genome disruption at any of the examined loci (Table 1, Figure S4).
It is well established that prolonged culture of hESCs can lead to adaption such as increased growth rate, reduced apoptosis and the acquisition of chromosomal abnormalities such as copy number variations (CNVs) (Laurent et al., 2011; Narva et al., 2010). More recently, it has been proposed that the reprogramming process itself compromises genomic integrity and leads to the accumulating of CNVs and somatic mutations (Gore et al., 2011; Hussein et al., 2011). Although all our tested cell lines showed a normal karyotype after genome editing as determined by conventional karyotyping, the low resolution of this technology excludes the detection of smaller CNVs, which are considered a major source for human genome variability and particularly important in the context of genome editing. Such genetic changes involve the induction of DNA double strand breaks and clonal events, which are thought to increase the chance for additional genomic alterations. We therefore performed high-resolution genome wide CNV analysis using an Affymetrix SNP 6.0 array as described previously (Hussein et al., 2011; Narva et al., 2010) on 3 pairs of isogenic parental and genetically modified cell lines. We identified on average 77 CNVs with an average size of 158 kb per cell line (Table S5). For human ES cell lines this is slightly higher than described previously and may be due to technical variability such as low sample or higher passage number (between passage P25 and P60 for hESCs and P22 and P40 for hiPSCs) of the cell lines used in this study. 63% of the identified CNVs (number and total genomic area) overlapped between isogenic pairs using pairwise comparison before and after ZFNs mediated genome modification (Figure S5A–E). In contrast we observed only 35% overlap of CNVs between genetically unrelated samples (Figure S5A–E). This degree of overlap is comparable to previously published hESC data comparing identical cell lines at different passage numbers (Figure S5D,E) (Narva et al., 2010). Furthermore, comparing average number and total genome area of CNVs before and after ZFN mediated gene targeting did not reveal any additional change other than that observed during regular hESCs in culture (Figure S5F). Consistent with previous reports (Narva et al., 2010), our analysis confirms that hESCs and hiPSCs contain a higher number of CNVs than the normal human genome independently of the genome editing procedure, probably acquired during hESC derivation, the reprogramming process and prolonged cell culture. Moreover we conclude that the Cre-recombinase mediated excision of the reprogramming factors and ZFNs mediated genome editing did not substantially increase the level of genomic alterations. To further validate that the genome editing approach did not induce major genetic or epigenetic alterations which would result in aberrant gene expression profiles, we performed whole genome expression analysis of undifferentiated pairs of parental and genetically modified cell lines. Despite very similar gene expression of all pluripotent cell lines, hierarchical cluster analysis showed that the influence of genetic background on gene expression is more significant than the genome editing, since gene expression patterns of pairs of parental and genome edited cell lines are more closely correlated than genetically independent cell lines (Figure S5G).
While transgenic animal models of disease and studying human cultured cells in a Petri dish are core technologies driving basic biomedical research, there are many instances where these approaches only partially recapitulate the molecular and cellular changes observed in the patient. The ability to generate an unlimited supply of patient-derived, disease-relevant cell types using hiPSC technology has the potential to transform biomedical research of human disease. However, numerous challenges on the path to well defined experimental in vitro systems must be resolved before realizing the full promise of this technology. One of the crucial limitations has been the inability to perform experiments under genetically defined conditions. Here we present an elegant solution to this problem by combining ZFNs mediated gene editing with hESC and hiPSC technology.
Table 2 compares the strengths and the weaknesses of the four strategies described in this study. Although a gene-targeting vector using positive-negative selection represents the most efficient genome editing strategy, this approach requires the deletion of the selection cassette and therefore necessitates a second single-cell cloning step adding time to generate the final targeted cell line. Another unavoidable consequence of the selection-based strategy is that it leaves a genetic footprint, i.e. a loxP site, close to the targeted genetic alteration. In contrast, the selection-free strategy, though less efficient, leaves no genetic footprint and involves only one cell cloning step and thus shortens the time required to generate the targeted cell line. The advantage of simple synthesis of ssODN-based donors combined with the ability to transfer genomic alterations adjacent to the ZFN induced DSB greatly increases the versatility and applicability of the no-selection based approach and may represent the most advantageous option for genome editing.
Here, we generated a panel of isogenic control and disease cell lines on several defined genetic backgrounds by either engineering the PD related A53T or E46K mutation into hESCs or, inversely, to repair the mutation in PD-patient-derived hiPSCs by exclusively changing a single base pair without the need to alter the genome in any other way. Considering the broad influence of genetic background and the profound biological differences between individual hESCs and hiPSCs such as propensity to differentiate towards specific cell lineages (Bock et al., 2011; Boulting et al., 2011), this experimental system may overcome some of the shortcomings of conventional hiPSC approaches in identifying disease-related phenotypes. Robust disease-relevant in vitro phenotypes are fundamentally important to identify new drug targets or allow large-scale screening of genetic and small molecule disease modifiers. The availability of genetically defined pairs of disease and control cell lines becomes even more significant in the context of unbiased genome wide analysis in order to distinguish between genetic background noise and disease relevant effects, considering that the expression of approximately 5% of genes is thought to be linked to genetic variation (Montgomery et al., 2010; Pickrell et al., 2010).
While biomedical research is likely to benefit substantially from in vitro disease modeling, the ultimate promise of iPSC technology, albeit at a very early stage of development, is the concept of cell replacement therapy in degenerative diseases using autologous cells (Daley and Scadden, 2008). Proof of principle experiments in the mouse (Hanna et al., 2007) have established the feasibility of treating monogenic diseases using the combined approach of derivation of disease-specific iPSCs followed by in vitro repair of the underlying genetic alteration and subsequent transplantation of corrected iPSC derived cells. Similar approaches with human cells are currently hindered by the problem of efficient gene targeting required for repair of disease-causing genomic alterations (Daley and Scadden, 2008). The approach of genetically correcting a disease-causing point mutation in patient-derived hiPSCs thus represents not only a significant progress for basic biomedical research but also a major advancement towards hiPSC-based cell replacement therapies.
HESC lines BG01 (NIH Code: BG01; BresaGen, Inc., Athens, GA) and WIBR3 (Whitehead Institute Center for Human Stem Cell Research, Cambridge, MA) (Lengner et al., 2010) and hiPSCs were derived and maintained as described previously (Soldner et al., 2009) and in detail in the Extended Experimental Procedures. The patient biopsied harbors the A53T α-synuclein mutation and has been described previously (Golbe et al., 1996). All protocols were approved by the relevant Institutional Review Boards (Boston University Medical Campus; Massachusetts Institute of Technology) and Embryonic Stem Cell Research Oversight Committees (Whitehead Institute) and written informed consent was obtained before biopsy. All pluripotent cell lines have been characterized by conventional methods, which are described in detail in the Extended Experimental Procedures.
ZFNs against the human α-synuclein loci were designed using an archive of pre-validated 2-finger modules as described (Doyon et al., 2008; Miller et al., 2007; Perez et al., 2008; Urnov et al., 2005). Complete sequences target sites of the zinc finger proteins are provided in Table S1 and Table S2. All ZFNs were linked to wild-type FokI, except pair SNCA-L1/R3, which was linked to an obligate heterodimer form of the Fok1 endonuclease (ELD-KKR) (Doyon et al., 2010; Miller et al., 2007; Perez et al., 2008). The ZFNs were tested by transient transfection into K562 cells to test for disruption of the wild-type α-synuclein allele. Disruption efficiency at the target locus was determined by Surveyor (Cel-1) endonuclease-based measurement of non-homologous end joining as described (Miller et al., 2007; Perez et al., 2008) (Primers used in Cel-1 analysis: SNCA-Cel1-FW: 5′-AAACTAGCTAATCAGCAATTTAAGGC-3; SNCA-Cel1-RW: 5′-AGCCCTCATTATTCTTGGCA-3). Analysis of off-target cleavage by ZFNs, which results in NHEJ-mediated indels, was performed essentially as described before (Doyon et al., 2008; Hockemeyer et al., 2009; Perez et al., 2008) and described in detail in the Extended Experimental Procedures.
HESCs and hiPSCs were cultured in Rho-associated protein kinase (ROCK)-inhibitor (Calbiochem; Y-27632) 24 hours prior to electroporation. Cells were harvested using 0.05% trypsin/EDTA solution (Invitrogen) and resuspended in phosphate buffered saline (PBS). For drug selection based gene editing 1 × 107 cells were electroporated with 35 μg of donor plasmids (designed and assembled by F.S.) and 7.5 μg of each ZFN encoding plasmid (Gene Pulser Xcell System, Bio-Rad: 250 V, 500μF, 0.4 cm cuvettes). Cells were subsequently plated on DR4 MEF feeder layers in hESC medium supplemented with ROCK inhibitor for the first 24 hours. Ganciclovir (1X10−6 M) and/or puromycin selection (0.5 μg/ml) was started 72 hours after electroporation. For drug selection-free gene editing, 1 × 107 cells were electroporated with 30 μg of donor plasmids (designed and assembled by F.S.), 7.5 μg of each ZFN encoding plasmid and 10 μg of pEGFP-N1 (Clontech). For gene editing using ssODNs, 1 × 107 cells were electroporated with 7.5 μg of each ZFN encoding plasmid and 15 μg of pEGFP-N1 (Clontech) together with 30 μg of ssODNs (5′-GACTTATGTCTTGAATTTGTTTTTGTAGGCTCCAAAACCAAGA AGGGAGTGGTGCATGGTGTGGCAACAGGTAAGCTCCATTGTGCTTATATCCA AAGATGATATTTAAAGTAT-3′; Integrated DNA Technologies, Iowa). Cells were maintained on MEF feeder layers for 72 hours in the presence of ROCK Inhibitor followed by FACS sorting (FACS-Aria; BD-Biosciences) of a single cell suspension for eGFP expressing cells and subsequently plated at a low density in hESC medium supplemented with ROCK inhibitor for the first 24 hours. Individual colonies were picked and expanded 10 to 14 days after electroporation.
We thank Ping Xu and Tenzin Lungjangwa for technical support and Jessica Daussman, Kibibi Ganz, Ruth Flannery, and Dongdong Fu for their help with animal husbandry and processing of teratomas. We would like to thank Tom Volkert and Jeong-Ah Kwon of the Whitehead Genome Technology Core for their help with the copy CNV analysis, Nicki Watson of the W.M. Keck Microscopy Facility for help with confocal microscopy and Patti Wisniewski and Chad Araneo of the Whitehead Institute FACS facility for their help with cell sorting. We thank Gladys Dulay for construction of ZFN vectors and Elo Leung for computational analysis for the off-target study. We thank all the members of the Jaenisch lab for helpful discussions and comments on the manuscript. R.J. was supported by NIH grants R01-CA084198 and R37-HD045022. This research was supported in part by a Collaborative Innovation Award from the Howard Hughes Medical Institute. RJ is an adviser to Stemgent and a cofounder of Fate Therapeutics. L.K.F., L.Z., F.D.U., P.D.G., H.S.Z., D.J., J.V., X.M., E.J.R. are full-time employees of Sangamo BioSciences, Inc. Authors’ contributions are provided in the Supplemental Information.
Supplemental Information includes Extended Experimental Procedures, five figures and 5 tables and can be found with this article online at
Expression Microarray data and SNP 6.0 CNV array data have been submitted to the Gene Expression Omnibus (GEO) and are available under accession number
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.