Versatile and Efficient Genome Editing in Human Cells by Combining Zinc-Finger Nucleases With Adeno-Associated Viral Vectors

Abstract

Zinc-finger nucleases (ZFNs) have become a valuable tool for targeted genome engineering. Based on the enzyme's ability to create a site-specific DNA double-strand break, ZFNs promote genome editing by activating the cellular DNA damage response, including homology-directed repair (HDR) and nonhomologous end-joining. The goal of this study was (i) to demonstrate the versatility of combining the ZFN technology with a vector platform based on adeno-associated virus (AAV), and (ii) to assess the toxicity evoked by this platform. To this end, human cell lines that harbor enhanced green fluorescence protein (EGFP) reporters were generated to easily quantify the frequencies of gene deletion, gene disruption, and gene correction. We demonstrated that ZFN-encoding AAV expression vectors can be employed to induce large chromosomal deletions or to disrupt genes in up to 32% of transduced cells. In combination with AAV vectors that served as HDR donors, the AAV-ZFN platform was utilized to correct a mutation in EGFP in up to 6% of cells. Genome editing on the DNA level was confirmed by genotyping. Although cell cycle profiling revealed a modest G2/M arrest at high AAV-ZFN vector doses, platform-induced apoptosis could not be detected. In conclusion, the combined AAV-ZFN vector technology is a useful tool to edit the human genome with high efficiency. Because AAV vectors can transduce many cell types relevant for gene therapy, the ex vivo and in vivo delivery of ZFNs via AAV vectors will be of great interest for the treatment of inherited disorders.

Introduction

I n the last two decades various platforms have been developed that allow for efficient and precise modification of the human genome. A very promising approach is based on zinc-finger nuclease (ZFN) technology (Urnov et al., 2010; Rahman et al., 2011). ZFNs are designer nucleases that consist of two functional domains: a customized zinc-finger array that specifies DNA binding and the endonuclease domain of the restriction enzyme FokI that contains the catalytic activity (Smith et al., 2000). Each zinc-finger array in a ZFN subunit confers binding to the respective target half-site, and upon dimerization of the two subunits in correct spacing and orientation, the nuclease domain introduces a DNA double-strand break (DSB) within the spacer sequence that separates the two target half-sites (Smith et al., 2000). ZFNs have been shown to cleave target DNA with high specificity and efficiency in a variety of human cell lines and primary cells, including T cells (Perez et al., 2008), hematopoietic stem cells (Lombardo et al., 2007; Holt et al., 2010), mesenchymal stromal cells (Benabdallah et al., 2010), embryonic stem cells (Lombardo et al., 2007; Hockemeyer et al., 2009; Zou et al., 2009), and induced pluripotent stem cells (Hockemeyer et al., 2009; Zou et al., 2009).

A major issue for therapeutic applications of designer nucleases is the balance between ZFN activity and nuclease-associated toxicity (Händel and Cathomen, 2011). ZFNs have undergone several critical improvements in design in the last decade, including the development of novel platforms to generate specific zinc-finger arrays (Maeder et al., 2008; Urnov et al., 2010; Kim et al., 2011; Sander et al., 2011), improved architecture of the dimer interface to prevent homodimerization of ZFN subunits (Miller et al., 2007; Szczepek et al., 2007; Söllü et al., 2010; Doyon et al., 2011; Ramalingam et al., 2011), and customized interdomain linkers that connect the DNA-binding domain with the nuclease domain (Bibikova et al., 2001; Händel et al., 2009). After ZFN-mediated DNA cleavage of the target site, the resulting DSB leads to activation of the cellular DNA damage response, involving nonhomologous end-joining (NHEJ) and homology-directed repair (HDR) (Shrivastav et al., 2008). NHEJ is an error-prone repair pathway that can lead to small insertions or deletions (indels) at the junction site. If the DSB occurs within coding sequences, indels can result in frame-shift or nonsense mutations. HDR, on the other hand, is based on homologous recombination and is the basis of ZFN-induced gene targeting. Although NHEJ is the more prominent repair pathway in mammalian cells, when high numbers of donor DNA are delivered to the nucleus, the HDR pathway becomes more prominent for DSB repair (Lombardo et al., 2007; Gellhaus et al., 2010).

An important aspect in ZFN-based genome engineering is vectorization of the nuclease and the donor. Ideally, ZFNs as well as the donor DNA are present only transiently at high concentrations. Episomal DNA-based ZFN expression systems, such as plasmid DNA or viral vectors, usually harbor strong promoters to ensure high ZFN levels. Thus, the duration of ZFN expression is limited through rapid dilution during cell division in mitotic cells. In addition to plasmid DNA, transient nuclease expression in human cells has been reported from integrase-deficient lentiviral vectors (Cornu and Cathomen, 2007; Lombardo et al., 2007), adenoviral vectors (Perez et al., 2008), and vectors based on adeno-associated virus (AAV) (Porteus et al., 2003; Gellhaus et al., 2010). The use of AAV vectors is of particular interest because these vectors function efficiently as a substrate for HDR, even if the target locus contains only a DNA single-strand nick (Metzger et al., 2011) or no DNA damage at all (Hendrie and Russell, 2005). In proof-of-concept studies using AAV as a donor for HDR and an I-SceI–induced DSB in the target locus to stimulate gene targeting, we and others showed that gene targeting at a chromosomally integrated marker gene can be achieved in up to 65% of cultured human cells (Miller et al., 2003; Porteus et al., 2003; Gellhaus et al., 2010; Hirsch et al., 2010). Moreover, in a recent study ZFN-stimulated in vivo gene targeting was reported after AAV-based gene transfer in a mouse model for factor IX deficiency (Li et al., 2011).

The goals of this study were to demonstrate the versatility of combining AAV vectors with the ZFN platform for rational editing of the human genome and to assay platform-associated toxicity. We show that ZFN encoding AAV expression vectors can be employed (i) to delete a lentiviral provirus, (ii) to disrupt an expression cassette, or (iii) to correct a mutational insertion.

Materials and Methods

Plasmids

Plasmids were assembled by polymerase chain reaction (PCR)-directed cloning and standard molecular biology procedures. Relevant parts of the plasmids and the resulting vectors are shown in Figs. 1 and 2. ZFN pairs EB1/BA1 (here termed EB/BA) (Cornu et al., 2008), E292 and E502 (Maeder et al., 2008), and their respective recognition sites have been described before. For this study, the corresponding zinc-finger arrays were first transferred into a pRK5 plasmid backbone (Alwin et al., 2005) that contained either of the obligate heterodimeric FokI variants KV or EA under control of a cytomegalovirus (CMV) promoter (Szczepek et al., 2007; Söllü et al., 2010), and then subcloned into the AAV backbone plasmids pFB-GFPR (Urabe et al., 2002) or pTR-UF (gift of Nicholas Muzyczka, Gainsville, FL). Plasmid pAV.LHA-Sce1Δ encodes a CMV-driven, C-terminally truncated version of I-SceI that served as a negative control. The AAV donor ∂GFPiNwpre and the lentiviral vector encoding the mGFPiNwpre target locus (Fig. 2) have been described before (Gellhaus et al., 2010; Söllü et al., 2010). The lentiviral vector encoding target locus dsEGFP (Fig. 1) was generated by adding two PEST domains at the C-terminus of EGFP of vector pLV-CMV.EGFPiNwpre (Cornu and Cathomen, 2007) and inserting the EB/BA target site in the ΔU3 region of the 3′ long terminal repeat (LTR). Maps and sequences of all plasmids are available upon request.

FIG. 1.

Adeno-associated virus (AAV)-mediated genome editing by nonhomologous end-joining (NHEJ). (A) Targeted deletion of lentiviral provirus. U2OS.693 cells carry an integrated copy of a destabilized enhanced green fluorescence protein (dsEGFP) expression cassette that was introduced by lentiviral transduction (target). The lentiviral long terminal repeats harbor a target site for zinc-finger nuclease (ZFN) pair EB/BA (fat arrows). Expression of both ZFN subunits (ZFN_EB and ZFN_BA) upon infection of cells with 3×10³ (gray) or 10⁵ genome copies (gc)/cell (black) of the respective AAV vectors leads to excision of the expression cassette (product). The graph shows the percentage of EGFP-negative cells, as assessed by flow cytometry; * indicates statistically significant increase in EGFP-negative cells as compared to cells infected with AAV-ZFN_EB alone (p<0.05). (B) Polymerase chain reaction (PCR)-based molecular characterization. Genomic DNA of single EGFP-negative clones (lanes 1–8) was extracted and used as a template to amplify a 1-kb fragment contained in the provirus. The positions of the primers are indicated as arrows in (A). Amplification of a 150-bp fragment in the PTBP2 gene served as a control. NT, nontransduced U2OS.693 cells; U, U2OS parental cell line; H₂O, PCR control. (C) Targeted gene disruption. U2OS.693 were infected with 10³, 10⁴, or 10⁵ gc/cell of AAV vectors that express ZFNs targeting either position 292 or 502 (fat arrows) of the EGFP reading frame, respectively. Error-prone repair of the DNA double-strand break leads to disruption of the coding sequence (product). The graph shows the fraction of EGFP-negative cells, as assayed 5 days post-infection by flow cytometry; * indicates statistically significant increase in EGFP-negative cells, as compared to cells infected with a control vector expressing a nonfunctional nuclease (p<0.02). (D) T7 endonuclease I (T7E1)-based genotyping. Genomic DNA of cells transduced with E502-specific ZFN expression vectors was used as a template to amplify a 544-bp fragment containing the target site. The amplicon was subjected to digestion with mismatch-sensitive T7E1 to verify the presence of insertions/deletions at E502. The position of the expected 446-bp band is indicated. (E) PCR strategy to detect AAV-ZFN integration into E502 site in antisense orientation. The positions of the PCR primers to detect the 5′- and 3′-junctions, respectively, are shown. (F) Qualitative assessment. Genomic DNA of U2OS.693 cells transduced with E502-specific ZFN expression vectors was subjected to PCR analysis. DNA from nontransduced (NT) cells served as a control. The position of the expected ∼800-bp band is indicated. H₂O, PCR control.

FIG. 2.

AAV-mediated genome editing by homology-directed repair (HDR). (A) Schematic of gene correction in U2OS.893 cells. The target locus consists of a mutated EGFP (mGFP) gene under control of a cytomegalovirus (CMV) promoter, followed by an ires-Neo^R-wpre cassette. The mutation in mGFP is based on a 43-bp insertion that includes three in-frame stop codons and a recognition site for ZFN pair EB/BA (fat arrow). The AAV donor vector contains a 5′-truncated EGFP gene (∂GFP) followed by the ires-Neo^R-wpre cassette. HDR (indicated by two crosses) between target locus and AAV donor leads to expression of a functional EGFP gene (product). (B) AAV-ZFN mediated gene targeting. U2OS.893 cells were transduced with the indicated vector dose of AAV donor and ZFN expression vectors, and assessed by flow cytometry 6 days post-transduction. The graph displays the average percentage of EGFP-positive cells; * and ** indicate statistically significant increase in EGFP-positive cells as compared to mock infected cells (p<0.01 and p<0.002, respectively). (C) AAV-mediated ZFN expression. U2OS cells were transduced separately with AAV-ZFN_EB, AAV-ZFN_BA, or a control vector at 10⁴ or 10⁵ gc/cell, harvested after 72 hr, and ZFN expression levels detected using an HA tag specific antibody. (D) Genotyping. Genomic DNA of U2OS.696 cells infected with 10⁴ gc/cell of AAV donor or AAV-ZFN, as indicated, was isolated 30 days post-transduction. The DNA was subjected to nested PCR analysis (primer positions shown in panel A) to detect the corrected target locus; 10⁴ and 10⁵ copies of an EGFP plasmid served as positive controls (cto1, cto2).

AAV-mediated genome engineering

All experiments were performed with recombinant AAV type 2 vectors. With the exception of AAV.ZFN_EB and AAV.ZFN_BA, all AAV vectors were produced and purified as previously described (Gellhaus et al., 2010). Vectors AAV.ZFN_EB and AAV.ZFN_BA were generated using the baculovirus system, as described in detail elsewhere (Smith et al., 2009). AAV vector titers were determined by quantitative real-time PCR (LightCycler, Roche) using SYBR Green (DyNAmo^™ Capillary SYBR Green, Finnzymes) and appropriate primers, as previously described (Gellhaus et al., 2010). For AAV-mediated genome editing, 5×10⁴ target cells were infected with vector doses ranging from 10³ to 10⁵ genome copies (gc)/cell in 500–2000 μL of standard medium. To analyze AAV-mediated EGFP expression kinetics, 10⁵ U2OS cells were transduced with either 50 or 500 gc/cell of AAV2.EGFP (Schwartz et al., 2007). At the indicated time points, cells were analyzed by flow cytometry (FACSCalibur, BD Biosciences). U2OS-based cell lines were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and penicillin/streptomycin (Invitrogen). The polyclonal target cells were generated by lentiviral transduction (Cornu et al., 2008), followed by selection with 0.4 mg/mL geneticin (Händel et al., 2009).

Genotyping

Genomic DNA from target cells was extracted with the Blood MiniKit (Qiagen). To amplify part of the proviral DNA containing the EGFP locus, 100 ng of genomic DNA was used as a template, along with 10 μM of each primer (#1270 5′-tacatcaatgggcgtggata and #598 5′-gaactccagcaggaccatgt), 10 mM dNTP, and 0.125 U of Phusion High-Fidelity DNA Polymerase (Finnzymes) in 1× reaction buffer for 25 cycles. A 150-bp amplicon of the PTBP2 allele (primers #1274 5′-tctccattccctatgttcatgc and #1275 5′-gttcccgcagaatggtgaggtg) served as a control. T7 endonuclease I (T7E1) assays were carried out as described previously (Hoher et al., 2011; Mussolino et al., 2011). In brief, 100 ng of the purified 544-bp PCR amplicon containing part of the EGFP gene (primers #76 5′-taaacggccacaagttcagcgt and #185 5′-gtgctcaggtagtggttgtcg) was melted and re-annealed to allow the formation of heteroduplex DNA, treated with 5 U of T7E1 (New England BioLabs) for 20 min at 37°C, and separated on a 2% agarose gel. To detect AAV vector integration into the E502 site, primers recognizing the EGFP gene and the AAV-ZFN expression vector were used for the 5′-junction primers #76 and #1401 (5′-accatggcccaacttgttta; SV40 pA) and for the 3′-junction primers #185 and #14 (5′-aatggggcggagttgttacgac; CMV). Correction of the EGFP target locus was verified using a nested PCR. For the first PCR, 100 ng of genomic DNA was used as a template, along with 300 μM of each primer (#136 5′-caagggcgaggagctggt and #559 5′-ctcggcgcgggtcttgtag), 200 mM dNTP, 0.125 U of Taq polymerase (PEQLAB) in 1× reaction buffer for 13 cycles. PCR products were purified using QIAquick PCR Purification Kit (Qiagen) and 1 μL (out of 20 μL) was used as a template for a second amplification round with primers #361 5′-gaggagctgttcaccggg and #559 for 25 cycles.

Analysis of ZFN expression

Western blot analysis to detect ZFN expression was performed as previously (Shimizu et al., 2011). To assess ZFN expression at the RNA level, 10⁵ U2OS cells were transduced with 10⁵ gc/cell of AAV.ZFN_EB and AAV.ZFN_BA each, and harvested in TRIzol reagent (Invitrogen) at indicated time points. The cDNA was generated using QuantiTect reverse transcription kit (Qiagen) and used as a template to amplify a 162-bp fragment of the FokI nuclease using primers #1436 5′-agtcaagagcgagctggaag and #1437 5′-cggtagccgtacaccttcat. The PCR products were separated on a 2% agarose gel and quantified from nonsaturated gel images with Quantity One V.4.6.9 (BioRad) software (Hoher et al., 2011; Mussolino et al., 2011).

Cell cycle profiles and apoptosis

U2OS target cells were synchronized by serum starvation for 1 day. Then, 10⁵ cells were transduced with 10⁵ gc/cell each of both AAV-ZFN_subunit expression vectors or treated with 100 nM vinblastine for 4 hr. After 4 days, cells were fixed overnight in ethanol and stained with propidium iodide (PI) staining solution (3.8 mM sodium citrate, 25 μg/mL of PI, 0.2 mg/mL of RNase A in phosphate-buffered saline). Before analysis by flow cytometry (FACSCalibur), low molecular weight DNA was released by incubating cells in DNA extraction buffer (200 mM Na₂HPO₄, 100 mM citric acid, pH 7.8). The fractions of cells in G1, S, and G2/M phases were determined by plotting FL3-W versus FL3-A. The number of cells containing a fragmented genome due to apoptosis was quantified by determining the percentage of subG1 events by plotting the whole cell population as FL2-H versus cell counts.

Statistical analysis

All experiments were performed at least three times, with the exception of data point E502 in Fig. 4. Error bars represent standard deviation. Statistical significance was determined using a one-sided Student's t-test with unequal variance.

Results

AAV-ZFN–mediated gene disruption and provirus excision

As a paradigm for inactivating and/or deleting a retroviral sequence in the human genome, we generated a reporter line based on human osteosarcoma U2OS cells that expresses a destabilized EGFP (U2OS.693) by lentiviral transduction. Quantitative PCR established that U2OS.693 cells contain between three and seven copies of the provirus (data not shown). The lentiviral LTRs of the lentivector used contain a target site for the previously characterized ZFN pair EB/BA (Cornu et al., 2008). Upon expression of both ZFN subunits the entire provirus should be excised, leaving just a single LTR signature in the genome (Fig. 1A). To this end, U2OS.693 cells were infected with 3×10³ or 1×10⁵ gc of AAV particles per cell, either individually with a single ZFN subunit or in combination, and the percentage of EGFP-expressing cells was determined 5 days later. As expected, successful excision of the provirus was dependent on the expression of both ZFN subunits and on the vector dose. After co-transduction with 10⁵ gc/cell of AAV vectors encoding the EB and BA subunits, 11% of U2OS.693 cells turned EGFP-negative. To verify ZFN-mediated excision of the provirus on the genome level, single clones were expanded and the genomic DNA of EGFP-negative cells isolated. PCR-based amplification of a region spanning most of the EGFP gene and part of the CMV promoter revealed that in the majority of clones all proviral genomes were excised (Fig. 1B). In three out of the eight clones at least one provirus copy remained untouched. Since all clones were EGFP negative, the respective expression cassettes must have been silenced before or after AAV transduction.

As a proof of principle that demonstrates AAV-ZFN mediated inactivation of functional gene expression, ZFN pairs targeting the EGFP marker gene at positions 292 (E292) and 502 (E502), respectively (Maeder et al., 2008), were employed. U2OS.693 cells were infected with 10³, 10⁴, or 10⁵ gc/cell of each ZFN expression vector and evaluated after 5 days. The percentage of EGFP-negative cells was dependent on both the vector dose and the ZFN pair used (Fig. 1C). Whereas 9% of cells lost functional EGFP expression upon infection with 10⁵ gc/cell of the AAV vectors encoding the left and right subunits of the E292 ZFN pair, 32% of cells transduced with 10⁵ gc/cell of the E502 encoding AAV vectors turned EGFP-negative. Molecular characterization of genomic DNA from pooled cells using the mismatch-sensitive T7E1 confirmed the presence of indels at the E502 site after expression of the respective ZFNs (Fig. 1D). To assess whether the AAV-ZFN expression vectors can integrate into a ZFN-induced DSB, a PCR strategy to detect vector–host genome junctions at the E502 site was applied (Fig. 1E). Both PCR amplifications produced the expected ∼800-bp fragments (Fig. 1F), confirming AAV integration into nuclease-induced DSBs at low frequencies as previously shown (Gellhaus et al., 2010).

Together these experiments demonstrated that AAV vectors encoding ZFNs can be efficiently employed to modify the human genome by harnessing the cellular NHEJ pathway for DNA repair. The AAV-ZFN platform was successfully used to either disrupt a gene or to delete an entire expression cassette, as illustrated by the excision of a lentivirus vector genome.

AAV-ZFN–mediated gene correction

To demonstrate AAV-mediated gene correction by HR, we established an U2OS-based cell line harboring a promoter-proximal 43-bp insertion that contains the recognition site for the ZFN pair EB/BA at position 327 of the EGFP marker gene (U2OS.893). Because promoter-free donors are advantageous, such a target configuration requires the generation of AAV donors with asymmetric homology arms. The AAV donor contained a 5′-truncated EGFP (∂GFP) and a homologous 3′-sequence stretch (ires-Neo^R-wpre). Of note, the 5′-homology arm between the donor and target was only 268 bp long (Fig. 2A).

U2OS.893 cells were co-infected with different amounts of AAV donor vector, AAV-ZFN_EB, and AAV-ZFN_BA, and the percentage of EGFP-positive cells was determined 10 days post-transduction (Fig. 2B). Co-infection of target cells with 10⁴ gc/cell of donor vector and 10⁵ gc/cell of both AAV-ZFN_EB and AAV-ZFN_BA led to correction of 6% of transduced cells. The percentage of EGFP-positive cells dropped significantly to 0.9% when the amount of donor was increased to 10⁵ gc/cell, suggesting that the ratio between donor and ZFN expression vector is crucial. Immunoblot analysis confirmed equal expression of both ZFN subunits upon transduction (Fig. 2C).

To verify that EGFP-positive cells were the result of a genuine HR event between the target locus and the AAV donor vector, genomic DNA from U2OS.893 cells was extracted 30 days post-transduction. The amplification products of an allele-specific PCR, in which primers that specifically recognize the corrected target locus but neither the AAV donor DNA nor the mutated target locus were employed (Gellhaus et al., 2010), confirmed that efficient gene targeting was dependent on the presence of all three vectors (Fig. 2D).

Expression kinetics

Transient expression of ZFNs is crucial for many applications because prolonged expression of the designer nucleases may be toxic to the cell. To characterize expression kinetics of AAV-based transgene expression both at the protein and RNA levels, we transduced U2OS cells either with an AAV2-based EGFP expression vector or the AAV-ZFN vectors. U2OS cells transduced with 50 or 500 gc/cell of AAV.EGFP, respectively, were assessed by flow cytometry at 2, 5, and 8 days post-transduction (Fig. 3A). EGFP fluorescence dropped more than threefold between days 5 and 8, suggesting that the vector is rapidly diluted out in these cultured cells. Total RNA was isolated from cells transduced with 10⁵ of AAV-ZFN_EB and AAV-ZFN_BA each at days 2, 5, and 8 post-transduction. Reverse-transcription PCR confirmed rapid loss of ZFN expression over the timeframe of 1 week (Fig. 3B). Together these results confirmed the transient, short-lived nature of ZFN expression in mitotic cells.

FIG. 3.

AAV expression kinetics. (A) Protein expression. U2OS cells were transduced with an AAV-EGFP vector at 50 or 500 gc/cell, respectively, and subjected to flow cytometric analysis at the indicated time points. The graph displays mean fluorescence intensity measured in channel FL-1. (B) RNA expression. U2OS cells were transduced with AAV-ZFN_EB and AAV-ZFN_BA at 10⁵ gc/cell. RT-PCR amplifying a fragment contained in the ZFN FokI domain was performed on RNA extracted from transduced cells at indicated time points. The position of the expected 162-bp band is indicated. Numbers below the gel indicate signal intensities relative to day 2. NT, nontransduced cells; AAV, cells transduced with an AAV control vector.

AAV-ZFN–associated toxicity

In the present study we applied three ZFN pairs that were generated by selection-based approaches (Hurt et al., 2003; Maeder et al., 2008). To assess cytotoxicity associated with AAV-based expression of these ZFNs, U2OS cells were co-infected with 10⁵ gc/cell of each vector encoding either ZFN subunit. After 2 and 4 days cells were fixed and stained with PI to perform a cell cycle analysis and determine the percentage of apoptotic cells by quantification of the subG1 fraction. The cell cycle profiles at day 2 (not shown) and day 4 post-transduction (Fig. 4A) revealed a modest increase in the G2/M populations for cells expressing the E292 and E502 ZFN pairs. In agreement with this modest effect on the cell cycle, the percentage of the subG1 fraction, representing apoptotic cells, did not increase above background for all samples analyzed (Fig. 4B). Also Annexin V staining of these cells did not reveal any significant signs of apoptosis (data not shown). Together, these results suggest that AAV-mediated expression of ZFNs can induce a mild G2/M arrest but does not induce measurable levels of apoptosis.

FIG. 4.

Platform-associated toxicity. U2OS.693 cells were infected with 10⁵ gc/cell of AAV-ZFN vectors, as indicated, fixed in ethanol 4 days after infection, stained with propidium iodide (PI) and analyzed by flow cytometry. Cells treated with 100 nM vinblastine (Vin) were assayed after 1 day and served as a control. (A) Cell cycle profile. The plots show the respective cell populations in G1 (gray), S (white), and G2/M phase (black). The average percentage of cells in G2/M phase is indicated. (B) Apoptosis. The graph displays the average fraction of cells in the subG1 population at day 4 post-transduction.

Discussion

An important safety measurement when developing platforms for therapeutic ZFN expression in human cells is to make sure that nuclease expression quickly reaches a high peak value after transfection/transduction but that expression lasts only transiently. The latter point is of utmost importance because long-lasting high ZFN expression may increase ZFN-associated off-target activity and hence cytotoxicity (Alwin et al., 2005; Doyon et al., 2008; Meng et al., 2008; Foley et al., 2009; Pruett-Miller et al., 2009). Nucleofection of plasmid DNA and transduction with adenoviral or integrase-deficient lentiviral vectors (IDLVs), respectively, have been successfully used to achieve ZFN-mediated gene editing in primary human cells (Lombardo et al., 2007; Perez et al., 2008; Holt et al., 2010). Moreover, microinjection of ZFN-encoding mRNA into one-cell zebrafish and rat embryos was shown to allow for efficient gene disruption (Doyon et al., 2008; Meng et al., 2008; Geurts et al., 2009). In this study we present AAV vectors as a potentially valuable vector platform for ZFN-mediated genome engineering based on NHEJ and HDR. We demonstrated that combining the two platforms allows for efficient genome editing in human cells, such as the targeted deletion of an entire expression cassette, the targeted disruption of an expressed gene, and the correction of mutations in an open reading frame.

The concentration and configuration of the donor DNA is a crucial parameter in HDR-based genome editing (Hirata and Russell, 2000; Moehle et al., 2007; Gellhaus et al., 2010; Orlando et al., 2010). It has been reported previously that AAV vectors serve as efficient templates for HDR in human cells, both in the absence (Chamberlain et al., 2004; Khan et al., 2010) and the presence (Miller et al., 2003; Porteus et al., 2003; Gellhaus et al., 2010; Hirsch et al., 2010) of a DSB. For the experimental systems used in this and in our previous studies, however, we were not able to detect AAV-based gene targeting in the absence of a DSB-inducing enzyme. The reason for this discrepancy likely resides in the donor architecture, which in our case was optimized to work efficiently for DSB-stimulated gene targeting.

Co-applying the AAV donors with AAV-ZFN expression vectors resulted in functional gene correction in up to 6% of cells. Although this number is therapeutically relevant, it is about four times smaller as compared with a similar setting in which U2OS target cells were co-infected with the same donor and an AAV I-SceI expression vector (25% EGFP-positive cells) (Gellhaus et al., 2010). Moreover, when the same cells were infected with a single AAV vector that combined donor function and I-SceI expression, 65% of cells revealed a corrected target locus (Gellhaus et al., 2010). Hence, the lower gene targeting activity cannot be solely reduced to differences in nuclease activities. This is in agreement with previous comparisons between I-SceI and ZFNs pairs, in which we did not detect more than twofold differences in the ability of these nucleases to stimulate gene targeting when plasmid transfection was used (Cornu et al., 2008; Szczepek et al., 2007). We therefore speculate that the drop in promoting gene targeting is a vector-associated phenomenon, and reducing the number of different AAV vector constructs from three to two may be crucial to further increase the gene targeting frequency. Future experiments will consequently focus on generating AAV vectors that harbor a so-called two-in-one ZFN construct; i.e., a combined expression cassette in which the two ZFN subunits are separated by a 2A autoproteinase (Söllü et al., 2010).

High peak expression of ZFN is a prerequisite to reach the critical nuclear concentration of the enzyme necessary to induce a targeted DSB. As shown previously for IDLVs and adenoviral vectors, sufficient ZFN expression can only be achieved with relatively high vector doses (Lombardo et al., 2007; Perez et al., 2008). In many experiments we observed efficient NHEJ and HDR-based genome editing only when applying 10⁵ AAV-ZFN gc/cell. Although this is a large number, 10⁵ copies of an expression cassette per cell correspond to about 0.5 μg of a 4-kb plasmid vector per 10⁶ cells, which is typically used for ZFN expression vectors in nucleofection protocols. On the other hand, we sometimes noticed inhibitory effects on gene targeting when using very high AAV doses. Whether this is related to saturation of the cellular transport machinery that shuttles the capsids from the cell surface to the nucleus or saturation of the DNA repair pathways that have to deal with high numbers of incoming single-stranded DNA vectors, needs to be addressed in future experiments.

The use of self-complementary AAV (scAAV) vectors (McCarty et al., 2003; Wang et al., 2003) may effectively reduce the required ZFN vector dose. scAAVs harbor a double-stranded DNA genome and are therefore not dependent on conversion into duplex forms, either via second strand synthesis or re-association kinetics, leading to faster and usually higher transgene expression. In a proof-of-concept approach Hirsch et al. (2010) showed that scAAV-based I-SceI expression vectors performed better in a gene targeting assay than single-stranded AAV counterparts. On the downside, the limited vector genome size of scAAV will not permit the packaging of a two-in-one ZFN expression cassette or a combination of donor and ZFN. The high AAV doses required to see biological effects in our studies suggest that high vector genome concentrations in the nucleus may allow for pairing of complementary genomes of (+) and (−) polarity and as a result bypass the requirement of DNA second strand synthesis.

ZFN and vector-associated toxicity is of utmost importance when transferring the AAV-ZFN platform to the clinic. It is known that episomal DNA vectors can integrate into naturally occurring DSBs (Miller et al., 2004), and we and others have also shown that nuclease-induced DSB serve as a prominent integration site for viral vectors, including IDLVs and AAV (Cornu and Cathomen, 2007; Gellhaus et al., 2010; Petek et al., 2010). Also in this study, we detected integration of AAV vectors into the ZFN-induced DSB. However, although AAV vectors can integrate into nuclease-induced DSBs at low frequency, the majority of integrations seem to occur outside the target locus (Gellhaus et al., 2010), most likely in naturally occurring DSBs or nuclease off-target sites (Petek et al., 2010; Gabriel et al., 2011). Moreover, we also reported that a targeted DSB shifts the balance from illegitimate recombination to gene targeting (Gellhaus et al., 2010), which strongly supports the use of designer nucleases to promote HDR-based genome engineering.

Quantitative assessment of ZFN-associated toxicity has been challenging, especially since the platforms to generate ZFNs improved in terms of producing designer nucleases with higher activity and lower toxicity (Händel and Cathomen, 2011). In this study we applied two assays to quantify AAV-ZFN–induced apoptosis but did not detect signs of toxicity. Cell cycle analysis, on the other hand, demonstrated a mild arrest of cells expressing ZFN pairs E292 and E502 in the G2/M phase. A G2/M phase arrest is indicative of DNA damage, likely caused by the combined on- and off-target activities of the respective ZFNs. Previous studies have further shown that transduction of cells with wild-type AAV2 or UV-inactivated AAV2 led to cell cycle arrest in the G2/M phase (Saudan et al., 2000; Raj et al., 2001). Under the experimental conditions used in our study, we only observed minimal effects of AAV transduction on the cell cycle. The absence of both Rep expression and UV irradiation of our AAV vectors could explain the different experimental outcomes. In summary, this study emphasizes the importance of a comprehensive and thorough analysis of the activities and toxicities triggered by the designer nucleases and/or the vector system.

In conclusion, this study validates AAV vectors as a valuable platform to deliver designer nucleases to target cells. Given the safety record of AAV vectors in clinical and preclinical applications, one can even envisage in vivo gene editing therapies, as recently demonstrated in the liver of neonatal mice (Li et al., 2011). AAV vectors have also been used to efficiently transduce human iPS cells or mesenchymal stem cells ex vivo, opening the door for a wide range of applications in regenerative medicine by combining cell therapy with targeted genome engineering.

Footnotes

Acknowledgments

We thank Jessica Wenzl and Eva Guhl for technical assistance, Shamim H. Rahman and Sylwia Bobis-Wozowicz for help with experiments and critical discussions, and Nicholas Muzyczka for plasmids. This work was supported by grants ITCF–01GU0618 of the German Ministry of Education and Research, PERSIST–222878 of the European Commission's 7th Framework Programme, and SFB/TR19–TPC5 of the German Research Foundation to T.C., and a fellowship of the German Academic Exchange Service (DAAD) to K.K.

Author Disclosure Statement

No competing financial interests exist.

References

Alwin

, Gere

M.B.

, Guhl

et al. 2005. Custom zinc-finger nucleases for use in human cells. Mol. Ther., 12:610–617.

Benabdallah

B.F.

, Allard

, Yao

et al. 2010. Targeted gene addition to human mesenchymal stromal cells as a cell-based plasma-soluble protein delivery platform. Cytotherapy, 12:394–399.

Bibikova

, Carroll

, Segal

D.J.

et al. 2001. Stimulation of homologous recombination through targeted cleavage by chimeric nucleases. Mol. Cell Biol., 21:289–297.

Chamberlain

J.R.

, Schwarze

, Wang

P.R.

et al. 2004. Gene targeting in stem cells from individuals with osteogenesis imperfecta. Science, 303:1198–1201.

Cornu

T.I.

, Cathomen

2007. Targeted genome modifications using integrase-deficient lentiviral vectors. Mol. Ther., 15:2107–2113.

Cornu

T.I.

, Thibodeau-Beganny

, Guhl

et al. 2008. DNA-binding specificity is a major determinant of the activity and toxicity of zinc-finger nucleases. Mol. Ther., 16:352–358.

Doyon

, Mccammon

J.M.

, Miller

J.C.

et al. 2008. Heritable targeted gene disruption in zebrafish using designed zinc-finger nucleases. Nat. Biotechnol., 26:702–708.

Doyon

, Vo

T.D.

, Mendel

M.C.

et al. 2011. Enhancing zinc-finger-nuclease activity with improved obligate heterodimeric architectures. Nat. Methods, 8:74–79.

Foley

J.E.

, Yeh

J.R.

, Maeder

M.L.

et al. 2009. Rapid mutation of endogenous zebrafish genes using zinc finger nucleases made by Oligomerized Pool ENgineering (OPEN) PLoS One, 4:e4348.

10.

Gabriel

, Lombardo

, Arens

et al. 2011. An unbiased genome-wide analysis of zinc-finger nuclease specificity. Nat. Biotechnol., 29:816–823.

11.

Gellhaus

, Cornu

T.I.

, Heilbronn

, Cathomen

2010. Fate of recombinant adeno-associated viral vector genomes during DNA double-strand break-induced gene targeting in human cells. Hum. Gene Ther., 21:543–553.

12.

Geurts

A.M.

, Cost

G.J.

, Freyvert

et al. 2009. Knockout rats via embryo microinjection of zinc-finger nucleases. Science, 325:433.

13.

Händel

E.M.

, Cathomen

2011. Zinc-finger nuclease based genome surgery: it's all about specificity. Curr. Gene Ther., 11:28–37.

14.

Händel

E.M.

, Alwin

, Cathomen

2009. Expanding or restricting the target site repertoire of zinc-finger nucleases: the inter-domain linker as a major determinant of target site selectivity. Mol. Ther., 17:104–111.

15.

Hendrie

P.C.

, Russell

D.W.

2005. Gene targeting with viral vectors. Mol. Ther., 12:9–17.

16.

Hirata

R.K.

, Russell

D.W.

2000. Design and packaging of adeno-associated virus gene targeting vectors. J. Virol., 74:4612–4620.

17.

Hirsch

M.L.

, Green

, Porteus

M.H.

, Samulski

. 2010. Self-complementary Aav mediates gene targeting and enhances endonuclease delivery for double-strand break repair. Gene Ther, 17:1175–1180.

18.

Hockemeyer

, Soldner

, Beard

et al. 2009. Efficient targeting of expressed and silent genes in human ESCs and iPSCs using zinc-finger nucleases. Nat Biotechnol, 27:851–857.

19.

Hoher

, Wallace

, Khan

et al. 2011. Highly efficient zinc-finger nuclease-mediated disruption of an eGFP transgene in keratinocyte stem cells without impairment of stem cell properties. Stem Cell Rev.[Epub ahead of print].

20.

Holt

, Wang

, Kim

et al. 2010. Human hematopoietic stem/progenitor cells modified by zinc-finger nucleases targeted to CCR5 control HIV-1 in vivo. Nat. Biotechnol., 28:839–847.

21.

Hurt

J.A.

, Thibodeau

S.A.

, Hirsh

A.S.

et al. 2003. Highly specific zinc finger proteins obtained by directed domain shuffling and cell-based selection. Proc. Natl. Acad. Sci. U. S. A., 100:12271–12276.

22.

Khan

I.F.

, Hirata

R.K.

, Wang

P.R.

et al. 2010. Engineering of human pluripotent stem cells by AAV-mediated gene targeting. Mol. Ther., 18:1192–1199.

23.

Kim

, Lee

M.J.

, Kim

et al. 2011. Preassembled zinc-finger arrays for rapid construction of ZFNs. Nat. Methods, 8:7.

24.

, Haurigot

, Doyon

et al. 2011. In vivo genome editing restores haemostasis in a mouse model of haemophilia. Nature, 475:217–221.

25.

Lombardo

, Genovese

, Beausejour

C.M.

et al. 2007. Gene editing in human stem cells using zinc finger nucleases and integrase-defective lentiviral vector delivery. Nat. Biotechnol., 25:1298–1306.

26.

Maeder

M.L.

, Thibodeau-Beganny

, Osiak

et al. 2008. Rapid “open-source” engineering of customized zinc-finger nucleases for highly efficient gene modification. Mol. Cell, 31:294–301.

27.

McCarty

D.M.

, Fu

, Monahan

P.E.

et al. 2003. Adeno-associated virus terminal repeat (TR) mutant generates self-complementary vectors to overcome the rate-limiting step to transduction in vivo. Gene Ther., 10:2112–2118.

28.

Meng

, Noyes

M.B.

, Zhu

L.J.

et al. 2008. Targeted gene inactivation in zebrafish using engineered zinc-finger nucleases. Nat. Biotechnol., 26:695–701.

29.

Metzger

M.J.

, McConnell-Smith

, Stoddard

B.L.

, Miller

A.D.

2011. Single-strand nicks induce homologous recombination with less toxicity than double-strand breaks using an Aav vector template. Nucleic Acids Res., 39:926–935.

30.

Miller

D.G.

, Petek

L.M.

, Russell

D.W.

2003. Human gene targeting by adeno-associated virus vectors is enhanced by DNA double-strand breaks. Mol. Cell Biol., 23:3550–3557.

31.

Miller

D.G.

, Petek

L.M.

, Russell

D.W.

2004. Adeno-associated virus vectors integrate at chromosome breakage sites. Nat. Genet., 36:767–773.

32.

Miller

J.C.

, Holmes

M.C.

, Wang

et al. 2007. An improved zinc-finger nuclease architecture for highly specific genome editing. Nat. Biotechnol., 25:778–785.

33.

Moehle

E.A.

, Rock

J.M.

, Lee

Y.L.

et al. 2007. Targeted gene addition into a specified location in the human genome using designed zinc finger nucleases. Proc. Natl. Acad. Sci. U. S. A., 104:3055–3060.

34.

Mussolino

, Morbitzer

, Lütge

et al. 2011. A novel TALE nuclease scaffold enables high genome editing activity in combination with low toxicity. Nucleic Acids Res.[Epub ahead of print].

35.

Orlando

S.J.

, Santiago

, Dekelver

R.C.

et al. 2010. Zinc-finger nuclease-driven targeted integration into mammalian genomes using donors with limited chromosomal homology. Nucleic Acids Res., 38:e152.

36.

Perez

E.E.

, Wang

, Miller

J.C.

et al. 2008. Establishment of HIV-1 resistance in CD4+ T cells by genome editing using zinc-finger nucleases. Nat. Biotechnol., 26:808–816.

37.

Petek

L.M.

, Russell

D.W.

, Miller

D.G.

2010. Frequent endonuclease cleavage at off-target locations in vivo. Mol. Ther., 18:983–986.

38.

Porteus

M.H.

, Cathomen

, Weitzman

M.D.

, Baltimore

2003. Efficient gene targeting mediated by adeno-associated virus and DNA double-strand breaks. Mol. Cell Biol., 23:3558–3565.

39.

Pruett-Miller

S.M.

, Reading

D.W.

, Porter

S.N.

, Porteus

M.H.

2009. Attenuation of zinc finger nuclease toxicity by small-molecule regulation of protein levels. PLoS Genet, 5:e1000376.

40.

Rahman

S.H.

, Maeder

M.L.

, Joung

J.K.

, Cathomen

2011. Zinc-finger nucleases for somatic gene therapy: the next frontier. Hum Gene Ther.[Epub ahead of print].

41.

Raj

, Ogston

, Beard

. 2001. Virus-mediated killing of cells that lack p53 activity. Nature, 412:914–917.

42.

Ramalingam

, Kandavelou

, Rajenderan

, Chandrasegaran

2011. Creating designed zinc-finger nucleases with minimal cytotoxicity. J. Mol. Biol., 405:630–641.

43.

Sander

J.D.

, Dahlborg

E.J.

, Goodwin

M.J.

et al. 2011. Selection-free zinc-finger-nuclease engineering by context-dependent assembly (CoDA) Nat. Methods, 8:67–69.

44.

Saudan

, Vlach

, Beard

2000. Inhibition of S-phase progression by adeno-associated virus Rep78 protein is mediated by hypophosphorylated pRb. EMBO J., 19:4351–4361.

45.

Schwartz

R.A.

, Palacios

J.A.

, Cassell

G.D.

et al. 2007. The Mre11/Rad50/Nbs1 complex limits adeno-associated virus transduction and replication. J. Virol., 81:12936–12945.

46.

Shimizu

, Sollu

, Meckler

J.F.

et al. 2011. Adding fingers to an engineered zinc finger nuclease can reduce activity. Biochemistry, 50:5033–5041.

47.

Shrivastav

, De Haro

L.P.

, Nickoloff

J.A.

2008. Regulation of DNA double-strand break repair pathway choice. Cell Res., 18:134–147.

48.

Smith

, Bibikova

, Whitby

F.G.

et al. 2000. Requirements for double-strand cleavage by chimeric restriction enzymes with zinc finger DNA-recognition domains. Nucleic Acids Res., 28:3361–3369.

49.

Smith

R.H.

, Levy

J.R.

, Kotin

R.M.

2009. A simplified baculovirus-AAV expression vector system coupled with one-step affinity purification yields high-titer rAAV stocks from insect cells. Mol. Ther., 17:1888–1896.

50.

Söllü

, Pars

, Cornu

T.I.

et al. 2010. Autonomous zinc-finger nuclease pairs for targeted chromosomal deletion. Nucleic Acids Res., 38:8269–8276.

51.

Szczepek

, Brondani

, Buchel

et al. 2007. Structure-based redesign of the dimerization interface reduces the toxicity of zinc-finger nucleases. Nat. Biotechnol., 25:786–793.

52.

Urabe

, Ding

, Kotin

R.M.

2002. Insect cells as a factory to produce adeno-associated virus type 2 vectors. Hum. Gene Ther., 13:1935–1943.

53.

Urnov

F.D.

, Rebar

E.J.

, Holmes

M.C.

et al. 2010. Genome editing with engineered zinc finger nucleases. Nat. Rev. Genet., 11:636–646.

54.

Wang

, Ma

H.I.

, Li

et al. 2003. Rapid and highly efficient transduction by double-stranded adeno-associated virus vectors in vitro and in vivo. Gene Ther., 10:2105–2111.

55.

Zou

, Maeder

M.L.

, Mali

et al. 2009. Gene targeting of a disease-related gene in human induced pluripotent stem and embryonic stem cells. Cell Stem Cell, 5:97–110.