CRISPR-Cas9 gene editing: Delivery aspects and therapeutic potential

Erik Oude Blenke a, Martijn J.W. Evers a, Enrico Mastrobattista a, John van der Oost b,⁎ a Department of Pharmaceutics, Utrecht Institute of Pharmaceutical Sciences (UIPS), Utrecht University, Universiteitsweg 99, 3584 CG Utrecht, The Netherlands b Laboratory of Microbiology, Wageningen University, Dreijenplein 10, 6703 HB Wageningen, The Netherlands

a b s t r a c ta r t i c l e i n f o

Article history: Received 16 June 2016 Received in revised form 1 August 2016 Accepted 3 August 2016 Available online 4 August 2016

The CRISPR-Cas9 gene editing system has taken the biomedical science field by storm, initiating rumors about fu- ture Nobel Prizes and heating up a fierce patent war, but also making significant scientific impact. The Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR), together with CRISPR-associated proteins (Cas) are a part of the prokaryotic adaptive immune system and have successfully been repurposed for genome editing in mammalian cells. The CRISPR-Cas9 system has been used to correct genetic mutations and for replacing entire genes, opening up a world of possibilities for the treatment of genetic diseases. In addition, recently some new CRISPR-Cas systems have been discovered with interesting mechanistic variations. Despite these promising de- velopments, many challenges have to be overcome before the system can be applied therapeutically in human patients and enabling delivery technology is one of the key challenges. Furthermore, the relatively high off-target effect of the system in its current form prevents it from being safely applied directly in the human body. In this review, the transformation of the CRISPR-Cas gene editing systems into a therapeutic modality will be discussed and the currently most realistic in vivo applications will be highlighted.

© 2016 Elsevier B.V. All rights reserved.

Keywords: CRISPR-Cas CRISPR-Cas9 Gene editing Genome editing Delivery systems Therapeutic applications In vivo Ex vivo

1. Introduction

The CRISPR-Cas9 gene editing system has received a tremendous amount of attention ever since the discovery of relevant mechanistic features [1–4] in 2010–2011 and the first application in eukaryotes in 2012 [1]. CRISPR is short for Clustered Regularly Interspaced Short Pal- indromic Repeats that direct the gene editing to a certain target and Cas9 is the associated nuclease that cuts the DNA. Applications of the system appear to be nearly endless, ranging from improving crop resis- tance [5] to overcoming HIV infections [6] and the controversial human embryo editing [7]. The most captivating application is the prospect of being able to correct genetic defects in diseased tissues and cells [8], al- though this may currently still be out of reach [9]. However, the system being named the Science Magazine's Breakthrough of the Year 2015 [10] makes it undisputed that CRISPR-Cas9 is here to stay and it is al- ready speculated that its inventors may receive a Nobel Prize within the coming decade [7]. Similar to RNA interference, where a eukaryotic defense system against viral infections is exploited to modulate gene expression [11], this new genome editing system makes use of an adap- tive immune system found in prokaryotes. There is a multitude of such systems and CRISPR-Cas9 is certainly not the first one to be described [12,13], but its simplicity and ease of use have sparked the interest of

researchers in diverse fields and initiated a run to clinical applications, again very similar to the early days of RNAi [14,15]. To exploit this po- tential, development of carrier systems capable of delivering the CRISPR-Cas9 system to human cells is of utmost importance, taking les- sons from the RNAi field where possible. In this review, the basic mech- anism of CRISPR-Cas9 genome editing is explained and current and potential therapeutic applications are highlighted. A special focus will be on the delivery aspects of the system, discussing the requirements for delivery vehicles to allow safe and effective ex vivo and in vivo ma- nipulation for therapy in human patients.

2. CRISPR-Cas9 genome editing mechanism

Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR), together with CRISPR-associated proteins (Cas) are a part of the adaptive immune system found in bacteria and archaea. This adap- tive immune system can detect and destroy Mobile Genetic Elements (MGEs) such as unwanted viral and plasmid DNA in a highly specific manner. As mentioned before, there are other bacteria-derived targeted nucleases, like Meganucleases, TALEN (Transcription Activator-Like Ef- fector Nucleases) or ZFN (Zinc Finger Nuclease), that are already being translated into clinical application [16–19]. The CRISPR-Cas system is a family of proteins, subdivided in Class 1 (Types I, III and IV) and Class 2 (Types II, V, VI) [12], all consisting of specific endonuclease proteins (Cas) and a guide RNA molecule [20–23]. The guide RNA molecule guides the Cas protein to a very specific MGE related DNA target (Fig. 1). This bacterial molecular machinery can be adapted for use in higher

Journal of Controlled Release 244 (2016) 139–148

⁎ Corresponding author at: Laboratory of Microbiology, Wageningen University, Stippeneng 4, 6708 WE Wageningen, The Netherlands.

E-mail address: [email protected] (J. van der Oost).

http://dx.doi.org/10.1016/j.jconrel.2016.08.002 0168-3659/© 2016 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Journal of Controlled Release

journal homepage: www.elsevier.com/locate/jconrel

organisms, in particular for gene-editing. To this end, the endonuclease and the guide RNA have to be heterologously expressed. For this pur- pose, a specific subtype of CRISPR-Cas is preferred: Class 2. The Class 2 CRISPR-Cas systems generally consist of a single multi-domain protein, such as the Type II nuclease: Cas9 [21–23]. The relatively simple archi- tecture of Class 2 nucleases (Cas9) makes them so easy to apply, as com- pared to the large, multi-subunit protein Class 1 complexes.

2.1. CRISPR-Cas nucleases and guide RNAs

CRISPR-Cas based immunity in bacteria proceeds in three distinct stages. The three stages are acquisition, expression and interference [13,24].

2.1.1. Acquisition As an adaptive defense system, bacteria and archaea collect se-

quences of foreign (plasmid or virus) DNA of 30–45 nucleotides long and integrate them as new spacers in the repetitive CRISPR arrays. To allow self/non-self-discrimination, foreign target sequences (protospacers) are selected on the basis of a flanking motif, the protospacer-adjacent motif (PAM).

2.1.2. Expression During the expression stage, the CRISPR array is transcribed in one

large pre-crRNA and is subsequently processed into smaller CRISPR RNAs (crRNAs). Each crRNA corresponds to one acquired foreign DNA sequence, so expression will result in a pool of crRNAs that all recognize a particular genetic element. The enzymes involved in this step vary be- tween the different CRISPR-Cas subtypes. In the Cas9 system, the re- peats of the pre-crRNA first hybridizes with a second, conserved RNA, called the transactivating CRISPR RNA (tracrRNA), after which the dsRNA is specifically cleaved by a non-Cas ribonuclease (RNaseIII). In the system adapted for gene editing (Fig. 1B), these two RNAs are fused and expressed together as a single guide RNA (sgRNA) [1].

2.1.3. Interference In this stage, the Cas nucleases are guided by the mature crRNAs to

target and (in the presence of an adjacent PAM motif) cleave the corre- sponding protospacer sequences in invading MGEs when present.

Hybridization of the tracrRNA:crRNA/Cas9 complex - or the sgRNA/ Cas9 complex to the corresponding protospacer sequence results in double stranded breaks and thereby inactivation of the invading DNA [21,22,24]. For adaptation in eukaryotes, the sgRNA and the enzyme have to be expressed, either from a plasmid or from delivered mRNA. mRNA can be used for a more transient expression. For the same reason, the Cas9 enzyme and the sgRNA complex can be directly administered, as the half-life of the enzyme is even shorter than that of exogenous mRNA. These strategies can be chosen to minimize off-target effects as will be discussed later. Expression of multiple sgRNAs from the same construct is called multiplexing, and can be used to target multiple genes or to enhance the knock-out by targeting multiple sites in the same gene [8].

2.2. PAM sequences

It should be noted that after integration of the invading DNA in the CRISPR locus, the ‘foreign’ sequence is also present in the bacterial ge- nome. To avoid cleavage of the DNA in the CRISPR locus, a safety mech- anism is built into the crRNA-sequence. The PAM-sequence is part of the MGE DNA, but is not copied into the CRISPR locus. Cas9-mediated cleav- age of the target DNA only occurs when the PAM-sequence is present at the 3′ end. When there is base-pair complementarity but no PAM-se- quence, it indicates that the crRNA is bound to the CRISPR locus itself and the sequence is then not degraded [25]. PAM sequences vary per bacterial species [26]. The most widely used Cas9 nuclease is derived from Streptococcus pyogenes (SpyCas9) and has GG as its PAM-se- quence, meaning that every target protospacer sequence is located ad- jacent to two guanine bases (protospacer-NGG) [1]. In the unlikely case that this sequence is not present in the intended target DNA, anoth- er Cas9 species could be used that binds to a different PAM sequence.

2.3. Non-homologous end joining (NHEJ) and homology-directed repair (HDR)

Cleavage by the targeted nuclease results in a double stranded break (DSB) at a desired sequence-specific location in the target DNA. In eu- karyotes, this DSB can be repaired by two distinct mechanisms: Non-ho- mologous end joining (NHEJ) or homology directed repair (HDR). NHEJ

Fig. 1. Mechanism of CRISPR-Cas9 in prokaryotes and the adapted mechanism in eukaryotes. A: In prokaryotes, the protospacer sequences acquired from invading pathogens are stored as spacers in the CRISPR-loci, in the DNA flanked by CRISPR repeats. These are transcribed into a precursor (pre-crRNA) after which the repeats hybridize with anti-repeat sequences within the tracrRNA. This dsRNA is recognized and cleaved by a housekeeping ribonuclease (RNaseIII), resulting in a mature crRNA/tracrRNA hybrid that forms a stable complex with Cas9. Upon a viral invasion, it guides the nuclease to the target sequence in the DNA for cleavage. B: In eukaryotes, a sgRNA is used that combines the function of the crRNA and tracrRNA. This can be expressed from a plasmid or from mRNA, alongside the Cas9 enzyme which is not naturally present in eukaryotes. Alternatively, the sgRNA/Cas9 complex can be administered as a whole. After translocation across the nuclear membrane (due to an engineered Nuclear Localization Signal; NLS) the heterologous complex cleaves the target sequence in the chromosomal DNA.

140 E. Oude Blenke et al. / Journal of Controlled Release 244 (2016) 139–148

generates small insertions or deletions (indels) which can inactivate the target sequence by inducing a frame-shift or introducing a pre-mature stop codon. This method is applied for the rapid generation of knock- out cell lines or animal models [27], functional genomic screens [28] and other applications of transcriptional modulation/gene silencing [29]. Alternatively, HDR repairs the DNA strands based on structural ho- mology. In the naturally occurring situation, this could be homology to a nearby located (and structurally related) gene or in therapeutic gene editing, a co-expressed or co-delivered repair template. When two DSBs are created this can result in complete excisions of a target gene and even a provided donor DNA template sequence could be precisely inserted into a specific target site [30–32] (Fig. 2). It should be noted that the process of HDR is much less efficient than NHEJ and the knock-in or replacement of a gene happens with much lower efficiency than the knock-out of a gene using CRISPR-Cas9. This is also reflected by the number of applications of NHEJ vs. HDR, as will be discussed in Section 3.

To summarize, the Cas9 system is the most flexible and easiest sys- tem to adapt, because it uses only a single enzyme that mediates both the crRNA processing as well as the DNA cleavage. The specificity of the targeted nuclease can be simply altered by replacing the guide RNA, unlike ZFNs and TALENs that require protein engineering for every new target. With CRISPR-Cas9, any 22 nucleotides long DNA se- quence can be targeted as long as it is flanked by the NGG motif (when SpyCas9 is used). In the simplest and most widely used applica- tion of this system, where a genetic knock-out is made, only two com- ponents need to be expressed in the host cell to cleave the target gene: the Cas9 nuclease and the sgRNA. Guide RNAs can be expressed from the same plasmid as the Cas9 protein, which is (human) codon op- timized and contains a nuclear localization sequence (NLS) when ap- plied in eukaryotes. Alternatively, the enzyme can be expressed from an mRNA for more transient expression. In the more complex situation where a gene is replaced, a DNA template has to be co-delivered to re- place the excised gene. Examples of these therapeutic applications will be discussed in the following paragraphs.

3. Biomedical applications of the CRISPR-Cas9 system

There are a huge number of applications in a wide variety of research fields and various organisms. This review will be limited to applications in mammals/mammalian cells that could be of use in the biomedical

field. The large majority of publications utilize a NHEJ strategy to induce knock-outs of the target gene or gain-of-function mutations in the tar- get gene. However, the number of reports of successfully replaced genes with HDR is also growing.

Since the first report of the prokaryotic CRISPR-Cas9 system being programmable to cut isolated DNA at a desired location [1] it has literal- ly been a race to adapt the system for use in human cells with publica- tions from four independent groups coming out almost back-to-back [8, 33–35]. Jennifer Doudna and Emmanuelle Charpentier (University of California, Berkeley and that time at Umeå University, respectively) [1] and Feng Zhang from the Broad Institute and Massachusetts Institute of Technology, Cambridge are generally regarded as the front-runners of adapting this technology, which has led to a debate about who owns the intellectual property [36]. In the mean-time however, the Zhang lab has reported on another Type II CRISPR effector called Cpf1 [37] which could cause the patent battle to settle down, now that it has become clear that there can be alternative systems to achieve the same goal. The Cpf1 system may even have some slight advantages over the Cas9 system when it comes to gene replacement, as will be discussed briefly in the section on improvements to the system later.

3.1. Genetic knock-out animals

The first biomedical application of the CRISPR-Cas9 system was the generation of genetic knock-out mice and rats by co-injecting Cas9 mRNA and sgRNAs in one-cell stage embryos, again published by differ- ent groups very shortly after each other [27,38,39]. Conditional knock- outs could also be generated in mice and rats by integrating a donor template containing a Cre/lox recombination site as well as by a knock-in of an 11 kb template, showing that the HDR pathway could also be used for embryo engineering [40–42]. Genetic knock-outs of rabbits [43] and cynomolgus monkeys [44] were also generated by injecting one-cell stage embryos, demonstrating that this approach is feasible in the full range of preclinical animal models.

3.2. Quick genome screening and drug target identification and validation

The emergence of CRISPR-Cas9 has made the generation of knock- out animals for drug screening and target validation a routine proce- dure. Before, RNA interference has made an important addition to this, but gene silencing using short-hairpin RNAs (shRNAs) has several

Fig. 2. Two different repair mechanisms of a double stranded break. After the targeted nuclease has created a double stranded break, there are two possible repair pathways. The first is the error-prone non-homologous end joining (NHEJ) which results in insertions or deletions in the gene, often creating a frameshift and thereby inactivating the gene. The second pathway is homology directed repair (HDR) which only takes place in the presence of a homologue part of DNA. When such a donor or repair template is co-administered, this can be used to replace or correct the gene.

141E. Oude Blenke et al. / Journal of Controlled Release 244 (2016) 139–148

drawbacks. First of all, gene silencing based on mRNA degradation is transient. In addition, it often results in only a partial knock-down of the intended target [45]. Apart from that, it turned out there is signifi- cant off-target effect due to extensive modulation of micro-RNAs [46], although admittedly, CRISPR-Cas9 in its original form is not completely free of off-target effects either. However, the ease of applicability and low costs of the CRISPR-Cas9 has removed many of the barriers to high-throughput knock-out screens for gene function [45]. Targeted en- donucleases can now be expressed by lentiviruses encoding Cas9 and a genome wide array of sgRNAs [47,48]. This approach was validated by screening for resistance to lethal doses of the nucleotide analog 6- thioguanine and etoposide, and the sgRNA screen correctly identified all known genes resulting in resistance. Furthermore, through a nega- tive selection screening, other gene sets involved in fundamental pro- cesses could be identified [47]. A similar screening against BRAF inhibitor vemurafenib led to the correct identification of all known genes involved in resistance mechanisms to that drug as well as some novel hits [48]. This Cas9-based screen was repeated with a comparable library of shRNAs and interestingly, only a fraction of the shRNAs ap- peared to hit the Cas9 targets, demonstrating the superiority of the CRISPR-based screening approach [48]. For a more complete overview on genome-scale knock-out screening using CRISPR-Cas9, see reviews [28,45,49].

3.3. Human embryo editing and high off-target effects

A logical follow-up to the animal embryo knock-out experiments was the editing of genes in human embryos, but ethical issues prohibit the use of normal embryos for such studies. Alternatively, tripronuclear (3PN) zygotes were used, that contain one oocyte nucleus and two sperm nuclei. These polyspermic zygotes are common byproducts of in vitro fertilization and are normally discarded in clinics because they are nonviable in vivo but do form blastocysts in vitro. In these zygotes, the β-subunit of the human β-globin (HBB) gene that is mutated in β- thalassemia, was cut by the CRISPR-Cas9 enzyme and a sgRNA. A replac- ing donor template was successfully integrated in approximately 15% of all cases, but also Homology Directed Repair with a very similar and closely located gene, HBD (delta-subunit of β-globin) was seen in 25% of the cases. Apart from that, there was a high degree of off-target cleav- age and the authors concluded that the CRISPR-Cas9 system has to be improved significantly before embryo editing can be applied in a clinical setting [50]. It may have to do with the controversy around this subject [51], but it is striking how critical the authors are on these results, while such high off-target cleavage and low HDR efficiency are also seen in other studies and are a well-known limitation of the current system [52,53]. Off-target cleavage can lead to a heterogenic population of edited cells, including cells that are not cut at the intended site and more importantly, cells that are cut at the wrong site in the genome. Given the ‘permanent’ nature of the edit, the latter is of course a huge safety concern, if therapeutic gene editing is ever to be applied in human patients. The off-target effects can be partially addressed by re-designing the sgRNAs as will be discussed in the following paragraph, or by making changes to the enzyme itself (see Section 5).

Initial reports found that a single mismatch between the sgRNA and the target DNA abolished nuclease mediated cleavage [8]. This indicated that the system is highly specific and that one mismatch would lead to inactivation of the nuclease but this turned out to be highly dependent on the position in the sgRNA and on the target sequence. A more sys- tematic investigation revealed that multiple mismatches are tolerated, at different positions depending on the sequence, the number, position and distribution [53]. Mismatches are better tolerated when they are lo- cated at the 5′ end of the sgRNA (that is, further away from the PAM se- quence at the 3′, where hybridization with the target DNA is initiated). Off-target sites with as many as five mismatches were identified that were mutagenized to a comparable extent as the intended target site [52,53]. Apart from that, indel frequency at the target site is not 100%

and therefore not even every cut at the intended site leads to a gene inactivating frameshift by NHEJ. This was also seen in all the experi- ments with the embryos. In a later stage, the embryos displayed mosa- icism, indicating that cleavage in the multi-cell stages occurred with varying frequencies and efficiencies in daughter cells. Of course, in the case of the knock-out animals, several rounds of selection and cross- breeding will follow [42] but the need to select successfully engineered cells hampers the direct application of the CRISPR-Cas9 system in human patients (Fig. 3). It is impossible to say right now what an ac- ceptable off-target cleavage rate is, because it is not known to which de- gree the specificity can be further optimized. Significant progress has already been made, but one could argue that it has to be 0% if it ever is to be safely applied in humans [7]. Therefore, many of the current ther- apeutic applications aim to engineer the target cells ex-vivo, thereby also avoiding the recurring delivery problem.

3.4. Ex-vivo modifications of T-cells

A cell type that is particularly suitable for ex-vivo engineering is the T-lymphocyte, because it can be easily harvested from the patient's blood, modified and expanded outside the body and then re-adminis- tered without any immunogenicity (Fig. 3B). This could give the im- mune system a boost in conditions where the body's defense mechanism is compromised such as in HIV-infection or cancer [54].

A recombinant Cas9 enzyme was pre-incubated with a sgRNA targeting CXCR4, a co-receptor for HIV entry and then electroporated into isolated human CD4+ T-cells. This resulted in knock-out of CXCR4 in ~40% of the cells, which could be sorted and enriched based on CXCR4 expression [55]. Other studies by the group of Carl June con- firmed that editing of CCR5 (another HIV co-receptor) with a Zinc Finger Nuclease is safe and reduces viral DNA in HIV-infected patients [56]. In the same CRISPR-Cas study, the cell surface receptor PD-1 (PDCD1 gene) was also targeted. PD-1 is a so-called “immune check-point” that inhibits the cancer cell killing signaling in exhausted or chronically activated T-cells, and blocking the PD-1 protein has made a dramatic improvement in cancer immunotherapy. It was shown that by electroporating Cas9/PD-1 sgRNA and a HDR repair template into CD4+ cells, a gene replacement was induced in ~20% of the cells. Note, in this study a defective repair template was used to inactivate the PD-1 gene. This is not necessary as it was shown in the same study (and another [57]) that a knock-out could be generated by NHEJ as well, but this shows that ex-vivo gene replacement is also possible using the CRISPR-Cas9 system which could have potential benefit in other applications.

For example, another approach of cancer immunotherapy is endowing T-cells with a chimeric antigen receptor (CAR), usually consisting of the single-chain variable fragment (scFv) from a monoclo- nal antibody and a co-stimulatory signaling domain. This has the theo- retical advantage that a broad repertoire of receptors with high affinity can be used, that these are applicable in every patient, and that there is minimal risk of graft-to-host immunity because the patient's own cells are used [58]. CAR T-cells are currently being evalu- ated in the clinic [59] and an “off-the-shelf” approach for T-cell receptor engineering was recently described using TALEN nucleases [60]. For many gene editing applications, the way has been paved by other pro- grammable nucleases like ZFN and TALEN and it is likely that because of the low costs and tailorability, also CRISPR-Cas based applications will emerge.

In fact, while this paper was under review, the Recombinant DNA Advisory Committee (RAC) at the U.S. National Institutes of Health ap- proved a clinical trial that proposed to use CRISPR-Cas9 for the first time to edit human T-cells in which some of the elements described above will be combined [61,62]. In the trial, led by Dr. Carl June and funded by the Parker Institute for Cancer Immunotherapy, autologous T-cells will be harvested and engineered to express an affinity enhanced T-cell receptor (TCR) recognizing the tumor antigen NY-ESO-1. This

142 E. Oude Blenke et al. / Journal of Controlled Release 244 (2016) 139–148

approach earlier showed promising, however very short-lived re- sponses in clinical trials with myeloma patients, when using viral trans- duction of the TCR [63]. To boost the effect, the team now wants to knock-out two different parts of the primary TCR, so that the engineered receptor becomes more potent. Additionally, they want to knock-out the PD-1 gene as described above, to further potentiate the immuno- therapeutic response. The simultaneous knock-down of three different gene segments demonstrates the strength of the CRISPR-Cas9 system and helps to make this already very complicated type of therapy a little less challenging [61,62].

3.5. In vivo applications

In vivo application of CRISPR-Cas9 is conceivable for indications where the current low efficacy of gene editing is sufficient to show a phenotypic - and most importantly - a clinical effect. One such example is the knock-out of Proprotein convertase subtilisin/kexin type 9 (PCSK9), involved in the Low Density Lipoprotein (LDL) clearance path- way. It was found that rare individuals that have an inactivating muta- tion in the PCSK9 gene not only have extremely low plasma LDL levels, but also appear to be protected against cardiovascular heart dis- ease. Surprisingly, this knock-out did not lead to any apparent other symptoms or adverse events, making it an attractive drug target [64]. Clinical trials with PCSK9 targeting siRNA lipid nanoparticles are cur- rently ongoing, as the protein is predominantly expressed in the liver, making it a suitable target for treatment with nanoparticles [65]. PCSK9 knock-out by adeno-associated virus-delivered CRISPR-Cas was demonstrated in mice, showing mutagenesis in ~50% of hepatocytes which resulted in decreased plasma PCSK9 levels, increased LDL recep- tor levels, and a 35–40% decrease in plasma cholesterol levels [66]. This study demonstrates that for certain indications, it is not essential to reach all cells, nor effectively edit all of them to show an effect, making the current incomplete targeting not necessarily an obstacle (Fig. 3C).

In fact, much lower editing efficiency was shown to still be clinically relevant. In a mouse model of hereditary tyrosinemia type I (HTI), the underlying Fah gene was corrected in the liver by hydrodynamic

injection of CRISPR-Cas plasmid and a HDR repair template. This result- ed in the initial expression of the wild-type Fah protein in ~1/250 cells (0.4%) but this was enough to rescue the bodyweight-loss phenotype [30]. In this particular disease, it was shown that correction of 1/ 10,000 hepatocytes could already reverse the disease progression, suggesting that there could be many other indications that could benefit from even such low gene editing frequencies [67]. Further- more, it appeared that there is positive selection for the edited cells, as after 30 days, ~33% of all hepatocytes in the treated mice were expressing the corrected protein [30]. This is explained by the fact that hepatocytes that are deficient for the Fah gene are poisoned by toxic metabolites, allowing selective outgrowth of the corrected, resistant, cells.

A similar phenomenon was seen in the muscle cells of mice in a Duchenne's muscular dystrophy (DMD) model in which the defective dystrophin gene was corrected. Previously, correction of the dystrophin gene by HDR in mouse embryos was shown [68]. But as germline editing in humans is currently not feasible [51] and the homology di- rected repair pathway is not active in postmitotic tissues such as heart and skeletal muscle a NHEJ strategy had to be applied in adult mice. In two back-to-back papers, an AAV-CRISPR-Cas mediated excision of the defective dystrophin exon 23 was reported, which skips the premature stop-codon and restores the reading frame …