Combinatorial testing of viral vector and CRISPR systems for precision genome editing Zhen Li
Combinatorial testing of viral vector and CRISPR systems for precision genome editing Zhen Li ISBN: 978-94-6522-568-5 The research presented in this thesis was supported by the China Scholarship Council - Leiden University joint program, the Prinses Beatrix Spierfonds, the Duchenne Parent Project NL, the Dutch Research Council (NWO) - Open Technology Program, and EU Marie Skłodowska-Curie Doctoral Network Actions. Cover: concept and design by Zhen Li Layout: Zhen Li Printing: Ridderprint, www.ridderprint.nl Copyright © Zhen Li, 2025, Leiden, the Netherlands. All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means without written permission of the copyright owner.
Combinatorial testing of viral vector and CRISPR systems for precision genome editing Proefschrift ter verkrijging van de graad van doctor aan de Universiteit Leiden, op gezag van rector magnificus prof.dr.ir. H. Bijl, volgens besluit van het college voor promoties te verdedigen op woensdag 8 oktober 2025 klokke 10:00 uur door Zhen Li geboren te Sanya, Hainan, China in 1993
Promotor: Prof. Dr. R.C. Hoeben Co-promotor: Dr. M.A.F.V. Gonçalves Leden Promotiecommissie: Prof. Dr. S.M. Chuva de Sousa Lopes Prof. Dr. W.W.M. Pijnappel (Erasmus MC, Rotterdam) Prof. Dr. M.H.M. Heemskerk Prof. Dr. P. ten Dijke Dr. Z.Y. Lei (University Medical Center Utrecht)
敬谢家⼈⻓久相伴 愿春祺夏安，秋绥冬禧 To my family for their love and trust
Contents Chapter 1 General introduction 9 Chapter 2 “Soft” genome editing using CRISPR nickases as a potential source of safer cell products 15 Chapter 3 AAV-vectored base editor trans-splicing delivers dystrophin repair 31 Chapter 4 Precision genome editing using combinatorial viral vector delivery of CRISPR-Cas9 nucleases and donor DNA constructs 39 Chapter 5 Selector AAV-CRISPR vectors purge off-target chromosomal insertions and promote precise genome editing 97 Chapter 6 Summary and general discussion 151 Chapter 7 Nederlandse Samenvatting Curriculum Vitae List of Publications Acknowledgements 161
Chapter 1 General introduction
Chapter 1 —— 10 —— General introduction and outline of this thesis Nowadays, advanced tools deployed for genome editing (GE) mainly derive from the CRISPR/Cas9 system found to be in 2012 as a prokaryotic RNAguide antiviral system. Typically, Cas9 proteins or engineered Cas9 proteins, e.g., nucleases, can make endogenously site-specific double-stranded DNA breaks (DSBs) which are predominantly repaired via the non-homologous end joining (NHEJ) DNA repair pathway leading to the introduction of mutagenic insertions/deletions (indels). Hence, NHEJ-mediated GE can lead to the removal of pre-existing genetic information, i.e., knock-out (KO). Alternatively, in the presence of exogenous donor DNA templates, e.g., a transgene cassette flanked by DNA sequences identical to those present around the targeted DSB made by engineered RNA-guided nucleases, can result in gene knock-in (KI) through the homology-directed repair (HDR) DNA repair pathway. Differently from DSBs made by RNA-guided Cas9 nucleases, there are Cas9 protein variants that induce instead single-strand DNA breaks (SSBs), or nicks, by virtue of having the D10A or H840A mutations that result in the catalytic inactivation of their RuvC or HNH nuclease domains, respectively. Significantly, in contrast to DSBs, nicks are not substrates for error-prone NHEJ processes. Furthermore, nicks caused by nicking Cas9 proteins, i.e., nickases, can also serve as stimuli for HDR in the presence of exogenous donor DNA templates for the purpose of achieving KI at target chromosomal sites. Ideally, SSB-based KI strategies can alleviate or even avoid the buildup of indels caused by DSBs, and hence it can be regarded as a preferable GE strategy choice in the regards to therapeutical applications. In Chapter 2, the pros and cons of GE involving DSB-based homologydirected repair are elaborated. We discuss the pressing need for the development and application of less mutagenic GE procedures, namely, via using DSB-independent nickases which, as aforementioned, lead only to residual amounts of mutagenic indels and can be tailored together with matched donor DNA templates for precise SSB-mediated HDR. As such, SSB-mediated HDR can serve as a valuable approach for editing
Chapter 1 —— 11 —— therapeutically relevant cells, e.g., induced pluripotent stem cells (iPSCs) and T cells, leading to the manufacturing of potentially safer cell products for transplantation purposes. In Chapter 3, we further elaborate on the use of nickase variants in the form of base editors directed at therapeutic GE applications. In particular, we commented on the work conducted by Chai et al. (Mol. Ther. Nucleic Acids. 2023 32:522-535), investigating repair of defective DMD alleles causing Duchenne muscular dystrophy (DMD) via adeno-associated viral (AAV) vector delivery of trans-splicing base editors in vitro and in vivo settings. Instead of testing AAV/CRISPR-Cas9-based DMD reading frame restoration, associated with the potentially prevalent capture of Cas9-encoding AAV at on- and off-target nuclease sites, the authors investigated dual AAV delivery of trans-splicing adenine-base editor constructs yielding a Cas9 nickase fused to an adenine deaminase that, together with a cognate gRNA, mediates adenine (A) to guanine (G) substitutions. The resulting substitutions disrupt splice site motifs leading to exon skipping of DMD exon sequences bearing premature stop codons. On the top of overcoming the limited cargo capacity of AAV vectors (i.e., 4.7 kb), such dual AAV method provides for a DSB-free DMD reading frame correction option for treating DMD patients whose disease is caused by outof-frame deletions. Even though, conceptually, one expects installing desired therapeutic gene edits with the aid of different GE tools, it has become patently clear the challenge of achieving efficient and safe delivery of the necessary tools into target cells, especially those cells resistant to transfection or relevant to therapeutical applications. Moreover, one needs to contend with the fact with the increasing precision of the most advanced GE tools there is a concomitant trend towards larger size increase in the associated machineries, which further hampers their delivery efficiency. Taken these delivery and safety aspects together, we reasoned that combining distinct engineered viral vectors lacking all viral genes could serve as sources of GE tools upon robust transduction of target cells. In particular, based on the complementary characteristics of
Chapter 1 —— 12 —— adenoviral (AdV) and adeno-associated viral (AAV) vectors we postulated that a robust and precise GE system could be assembled by combining the former and latter platforms for delivering CRISPR/Cas9 reagents and donor DNA templates into human cells, respectively. Indeed, contrary to linear uncapped AAV genomes, terminally capped double-stranded AdV genomes are not prone to off-target DNA insertions making them suitable for the delivery of Cas9 enzyme constructs since it is especially important to prevent chromosomal integration of these constructs to minimize the buildup of off-target effects. In addition, also in contrast to AAV vectors, AdV vectors can accommodate full-length Cas9 constructs together with single or multiple gRNAs. Conversely, AAV genomes are proficient substrates for HDR possibly owing to the AAV DNA structure featuring secondary-structured inverted terminal repeats flanking singlestranded DNA. Besides allocating AdV and AAV systems for the delivery of, respectively, Cas9 constructs and distinct types of donor DNA templates, the role of distinct AAV donor DNA structures on the efficiency and accuracy of genome editing was investigated. Hence, in Chapter 4, we put forward and systematically evaluate different iterations of this dual viral vector system in HeLa cells, human mesenchymal stem cells (hMSCs) and skeletal muscle progenitor cells (myoblasts) in terms of their efficiencies, specificities, and accuracies. In this context, we extend earlier observations showing that 3rd generation, fully viral gene-deleted AdV vectors have a notoriously dampened cytotoxicity profile when compared to that of their 2nd generation counterparts. In Chapter 5, we build on the dual viral vector platform introduced in Chapter 4 by investigating its compatibility with a marker-free co-selection system for selecting gene-edited cells and, simultaneously, purging imprecisely edited cells via ouabain selection. The main sub-unit of the ubiquitously expressed sodium/potassium pump (Na+/K+ ATPase) is encoded by ATP1A1 whose ATPase product is responsible for many physiological
Chapter 1 —— 13 —— functions including osmotic regulation. Interestingly, specific point mutations in ATP1A1 conferring ouabain resistance are naturally found in the human population without disrupting the regular physiological functions of the Na+/K+ ATPase. Hence, aiming at improving the performance of AAV-based gene editing procedures, we sought to investigate AAV donor constructs harboring marker-free co-selection components (selector AAV vectors) permitting ouabain-dependent enrichment for genome-edited cells. We demonstrate that combining selector AAV vectors with ouabain treatments, in addition to enriching for genome-edited cell populations, eliminates imprecise on-target edits and off-target and/or random donor DNA insertions from said populations. Importantly, selector AAV vector titration experiments showed that the highest fold-enrichment factors for genomeedited cell fractions are associated with the lowest vector input amounts. This finding is expected to be beneficial for alleviating AAV vector production costs, off-target donor insertions and P53-dependent activation of the DNA damage response linked to AAV DNA, which is known to be particularly deleterious in stem cells with scientific and therapeutic relevance, e.g., induced pluripotent stem cells and hematopoietic stem cells.
—— 14 ——
Chapter 2 “Soft” genome editing using CRISPR nickases as a potential source of safer cell products Zhen Li1 and Manuel A.F.V. Gonçalves1 Cell & Gene Therapy Insights 9:1201-1210 (2023) 1Leiden University Medical Centre, Department of Cell and Chemical Biology, Einthovenweg 20, 2333 ZC, Leiden, the Netherlands.
Chapter 2 —— 16 —— Abstract The integration of the gene and cell therapy fields through the application of genome editing principles permits generating ex vivo transplantable grafts from stem cells or from their differentiated progenies (e.g., T and NK cells) with novel genetically-engineered function(s). As such, these technologies are offering new therapeutic avenues to previously intractable inherited and acquired disorders (e.g., malignant and infectious diseases). In this article, we discuss the main characteristics, advantages and limitations of genome editing involving the targeted chromosomal insertion of transgenes upon site-specific double-stranded DNA break (DSB) formation by programmable nucleases, namely, RNA-programmable CRISPR nucleases. Subsequently, building on this information and recent findings, we put forward the view that targeted transgene insertion strategies based on CRISPR nickases, as opposed to nucleases, address important limitations of conventional DSB-dependent genome editing approaches. In particular, the cytotoxicity and high genotoxicity resulting from DSBs especially in cell types highly sensitive to DNA damage, including pluripotent and hematopoietic stem cells. Background Genome editing or genome engineering is a fast-evolving field with growing impact on basic science, biotechnology, and medicine [1]. Particularly versatile genome editing strategies consist of inserting exogenous donor DNA constructs into specific genomic loci (knock-in) subjected to double-stranded DNA breaks (DSBs) made by engineered nucleases derived from class 2 type II CRISPR systems consisting of single guide RNA (gRNA) and Cas9 ribonucleoprotein complexes (CRISPR-Cas9 nucleases) [2]. This versatility stems from the robust activity and straightforward designing of these RNAprogrammable nucleases and the amenability of gene knock-in strategies to genomic modifications spanning entire transgenes, including those encoding chimeric antigen receptors (CARs) and T-cell receptors (TCRs) alone or together with auxiliary factors, such as positive-selection markers and safety
Chapter 2 —— 17 —— genetic switches [3,4]. Indeed, notwithstanding the growing mining for and adaption of CRISPR and CRISPR-like systems for genome editing purposes, engineered CRISPR-Cas9 nucleases based on the prototypic Streptococcus pyogenes CRISPR system and their molecularly evolved or structurallyguided designed variants (e.g., high-specificity and targeting range-expanded variants), remain leading tools for a wide variety of genome engineering applications [5,6]. Main attributes of CRISPR nuclease-assisted genome editing Chromosomal gene knock-in procedures often entail the delivery of CRISPRCas9 nucleases together with donor DNA constructs designed as substrates for either homology-independent or homology-dependent repair (HDR) pathways [7]. Generally, HDR-mediated transgene knock-ins are more precise than those resulting from homology-independent processes in that they are naturally inserted at the chromosomal target site in a predefined orientation and present neither multiple-copy insertions nor imprecise ‘foot- prints’ at the junctions between genomic and exogenous DNA sequences [8,9]. Importantly, as HDR takes place during the late G2 and S phases of the cell cycle, therapeutically relevant dividing cell types, such as induced pluripotent stem cells (iPSCs), natural killer (NK) cells and T lymphocytes, are amenable to precise HDR-mediated genome editing. For instance, in what valuable target cells is concerned, genetically engineered CAR-T cells, serving as personalized ‘living drugs’, are yielding impressive results in terms of treating CD19-positive hematological malignancies [3,10]. This is so despite their high costs that stem in part from the difficulties in generating the large amounts of the respective engineered cell products. Since 2017, a growing number of these CAR-T cell products have in fact started to be approved by the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) [11]. Building on the resulting CD19-targeted cancer therapy datasets, over 500 CAR-T cell therapies directed at different antigens in liquid and solid tumors are currently undergoing clinical testing worldwide [11,12]. Significantly, CAR-NK cells are also entering the
Chapter 2 —— 18 —— adoptive immunotherapy field as a potential alternative to CAR-T cells owing to their intrinsic tumor-cell killing activity and fewer adverse effects in patients [13]. Yet, regardless of the target cell type, in what the genetic modification procedures are concerned, the adoptive immunotherapy field is moving from randomly integrating retroviral vector and DNA transposon systems to targeted transgene insertion approaches using programmable nucleases [3,10]. In contrast to unpredictable CAR or TCR donor construct integration, programmable nuclease-assisted genome editing assures stable and homogeneous transgene expression while minimizing insertional mutagenesis risks inherent to randomly integrating vehicles. In fact, in contrast to random, targeted TCR transgene insertion leads to predictable Tcell function in vivo [14]. In this context, genomic loci generically dubbed ‘safe harbor’ are particularly appealing endogenous landing pads for CAR and TCR transgenes as insertions at these sites minimize the chances for gene silencing or variegated transgene expression while preserving the endogenous transcriptome of engineered cells [15,16]. Main limitations of CRISPR nuclease-assisted genome editing As aforementioned, programmable nucleases and HDR-tailored donor DNA constructs yield precise gene knock-ins. However, a major limitation regarding the use of programmable nucleases is the fact that, in mammalian cells, DSBs are prevalently repaired via mutagenic non-homologous end joining (NHEJ) or microhomology-mediated end joining (MMEJ) processes instead of accurate HDR [17,18]. Moreover, in contrast to HDR, end-joining processes take place throughout the cell cycle. As a result, amongst cells exposed to donor constructs and programmable nucleases, the vast majority contains one or both target alleles disrupted by NHEJ- or MMEJ-derived small insertions and deletions (indels). This mutagenic burden, in the form of indel ‘footprints’, can lead to target protein imbalances and cell fitness losses [19]. In addition, on-target DSB formation can also yield translocations and gross chromosomal rearrangements [19–22]. Recent studies have further uncovered that on-target DSBs are capable of triggering extensive
Chapter 2 —— 19 —— chromosome fragmentation followed by haphazard reassembly (chromothripsis) [23,24] and the partial or entire loss of chromosomes (aneuploidy) [25]. Notably, the chromothripsis and aneuploidy phenomena were readily detected in T cells and hematopoietic progenitor cells subjected to CRISPR-Cas9 nuclease reagents used in clinical trials [23,25]. Notwithstanding these phenomena, recent findings are more reassuring in that, contingent upon gRNA target site selection, chromosomal losses in particular can be substantially minimized by inducing DSB formation before, as opposed to after, the activation/stimulation of the primary T-cell populations [26]. Finally, on-target DSBs trigger P53-dependent cell cycle arrest and apoptosis which limits the efficacy of HDR-mediated genome editing in regular P53positive cells [27,28], and creates selective pressure for the emergence of mutations associated with tumorigenesis. Related to the latter matter, during sub-culturing, pluripotent stem cells can acquire ‘spontaneous’ tumorassociated P53 mutations in a recurrent fashion [29] which, by virtue of being more resistant to DSBs, are in principle more prone to expansion than their wild-type counterparts once exposed to programmable nucleases. Indeed, CRISPR-Cas9 nuclease activation of certain signaling pathways can lead to the selection of cells with potentially harmful loss-of-function or dominantnegative mutations in the tumor-suppressor P53 transcription factor or gainof-function mutations in the KRAS oncoprotein [27,30]. Furthermore, recent mouse modelling experiments indicate that p53 mutant cells, rather than proceeding to malignancy via a haphazard route, are instead subjected to an unexpectedly more deterministic set of genetic instability events [31]. Together, these cytotoxic and genotoxic effects raise tangible concerns on the use of programmable nucleases for the optimal generation of autologous genetically-corrected cell products.
Chapter 2 —— 20 —— Rationale for “soft” genome editing based on CRISPR nickases Although emerging high-specificity programmable nucleases can greatly minimize off-target DNA cleavage, e.g., eSpCas9(1.1) [32] and Cas9-HF1 [33], they are inherently incapable of eliminating the potentially deleterious effects resulting from on-target DSB formation. Therefore, the substantial genotoxicity and cytotoxicity profiles associated with conventional nucleaseassisted genome editing create a pressing need for the development of alternative genetic engineering systems that reliably generate safer and functionally robust cell products. Indeed, DSB-dependent genome editing is expected to be particularly risky in the context to cell therapies based on the transplantation of populations of genetically engineered pluripotent stem cells, T lymphocytes and NK cells. The reasons are twofold. Firstly, in the context of extensive ex vivo cell amplification protocols underpinning the generation of these cell transplantation products, DSB-derived mutations and/or chromosomal rearrangements can cooperate in cell transformation and clonal expansion. Secondly, in instances where targeting multiple genes is needed for achieving a robust anti-tumor effect, e.g., via combinatorial exogenous CAR transgene knock-ins and endogenous TCR or programmed cell death protein 1 (PD1) gene knockouts, simultaneous induction of the attendant DSBs at different genomic positions is expected to exacerbate the levels of undesirable genome editing by-products in the form of translocations and chromosomal rearrangements. In this context, investigations exploring alternative HDR-mediated gene knock-in approaches that rely on sequence- and strand-specific nucleases (‘nickases’) are valuable in that the resulting single-stranded DNA breaks (SSBs), or nicks, are substrates for neither NHEJ nor MMEJ. As a corollary, the balance between precise HDR to undesired end-joining events are dramatically biased towards the former. Moreover, although genomic SSBs are, per se, poor HDR stimuli, earlier experiments from our laboratory using the native adeno-associated virus Rep68/78 nickases demonstrated that concomitant SSB formation at acceptor sequences and donor DNA constructs fosters HDR-mediated gene knock-in at an
Chapter 2 —— 21 —— endogenous human locus, namely, the prototypic safe harbour locus AAVS1 [34]. The application of this generic in trans paired nicking (ITPN) principle was subsequently expanded to other genomic sequences through the use of more versatile RNA-programmable CRISPR-Cas9 nickases [35,36] that are simply obtained through site-directed mutagenesis of one of the two nucleases domains of the parental Cas9 protein (i.e., HNH or RuvC) (Figure 1) [37]. Indeed, by stimulating otherwise inefficient SSB-dependent HDR, ITPN approaches based on the delivery of nicking CRISPR-Cas9 complexes and matched nickase-susceptible HDR donor constructs, are valuable for seamless and scarless chromosomal editing, including at multiple-copy or essential genomic tracts [19,38]. Additional examples regarding the application of ITPN methodologies in various mammalian cell types, e.g., iPSCs, keratinocytes and organoids featuring normal or cancer traits, encompass: (i) repairing or installing predefined gene mutations [35,38-41], (ii) maximizing the integrity of unmodified alleles during allele-specific gene editing [42,43], and (iii) streamlining one-step biallelic gene editing or onestep multiplexing gene knock-in or tagging [35,44,45]. It is equally worth mentioning that, in contrast to regular and high-specificity CRISPR-Cas9 nucleases, CRISPR-Cas9 nickases constitute poor P53-dependent signalling triggers in human cells, including in DNA damage-sensitive iPSCs [38,40]. Hence, it is expected that the aforementioned growing mining for CRISPR systems buried in large genomic and metagenomic databases, will start unearthing enzymes that, via either their intrinsic or engineered nicking activities, enlarge the toolset for DSB-free genome editing. Examples include
Chapter 2 —— 22 —— Figure1. Standard versus in trans paired nicking genome editing. The relative weights of desired and undesired genome editing outcomes derived from, respectively, homologydirected repair (HDR) and imprecise events caused by competing end-joining DNA repair pathways, e.g., non-homologous end joining (NHEJ), are illustrated. DSB and SSB, doublestranded and single-stranded chromosomal breaks, respectively; ‘nickases’, sequence- and strand-specific nucleases. In contrast to standard donor constructs, modified donor constructs have nickase-susceptible target sites (TS) framing their targeting modules consisting of exogenous DNA (green bar) flanked by sequences homologous to the genomic target region (‘homology arms’). the HNH-negative IsrB nickase derived from the ancestral CRISPR-like system OMEGA and the RuvC-only CRISPR class 2 type V Cas12i nuclease that nick and preferentially nicks, respectively, double-stranded DNA substrates [46–48]. Moreover, often, newly discovered CRISPR systems also yield genome editing components whose small sizes renders them more fitting for delivery through commonly used adeno-associated viral (AAV) vectors [49].
Chapter 2 —— 23 —— Finally, another recent ‘soft’ HDR-mediated genome editing concept that might be particularly suited for the repair of heterozygous or dominant mutations involves allele-specific chromosome nicking for the stimulation of interhomolog recombination (IHR) in somatic cells [50,51]. Through this process of allelic conversion, a pathogenic mutation in one allele can, in principle, be corrected using as donor template the endogenous ‘healthy’ allele (Figure 2). This elegant concept of using CRISPR-Cas9 nickases and endogenous homologous chromosomal DNA as repairing templates has been demonstrated in Drosophila models [51] and human cell lines [50,52]. Regarding the application of such exogenous donor DNA-free genome editing principles in human cells, recent investigations argue for multiplexing approaches in which primary allelic-specific gRNAs act in concert with secondary gRNAs to direct in trans paired nicking of homologous chromosomes and ensuing allelic conversion via IHR (Figure 2) [52]. Further research will be instrumental to advance CRISPR-Cas9 nickase-induced IHR from enticing proof-of-concept studies in cell lines to its application in human stem and progenitor cells. Translation Insight & Outlook There is a pressing need for investigating and validating alternative DSB-free and precise genome editing tools and strategies in various stem and progenitor cell types, e.g., bona fide T and NK cells as well as precursor iPSCs from which different effector cells can be differentiated, including immunotherapeutic T and NK cell candidates. Genome engineering strategies covering targeted and precise chromosomal incorporation of genetic payloads with varying sizes will become ever-more relevant. In this regard, CRISPR nickases per se and fused to reverse transcriptases offer a complementary
Chapter 2 —— 24 —— toolbox for ‘soft’ genome editing involving HDR and prime editing, respectively. Contrary to HDR, prime editing does not require the transfer of donor DNA substrates and allows for genomic insertion of up to ∼44-bp of Figure2. Gene correction via interhomolog recombination between heterozygous allelic sequences. Interhomolog recombination (IHR) characteristic of meiosis in germ cells can be fostered in somatic cells subjected to allele-specific double-stranded DNA breaks (DSB), yet the major products are on-target mutagenesis in the form of NHEJderived small insertions and deletions (indels). In contrast, allele-specific single-stranded DNA breaks (SSB) can equally foster IHR in somatic cells especially when using multiplexing CRISPR-Cas9 nickases for in trans multiple nicking IHR (MN-IHR). In somatic cells with heterozygous mutations or compound heterozygous mutations (not shown) underlying genetic disorders, CRISPR-Cas9 nickase-induced IHR offers the prospect for new genetic therapy interventions via wild-type allele-templated gene repair.
Chapter 2 —— 25 —— foreign DNA despite the need for substantial optimization of extended primeediting gRNAs (pegRNAs) [53]. Moreover, in contrast to HDR-based genome editing, prime editing can take place in post-mitotic cells albeit to lower efficiencies than in cycling cells [54]. Recent prime editing developments include the combinatorial use of dual pegRNAs and sitespecific recombinases designed for replacing genomic sequences with up 250-bp of foreign DNA and inserting entire transgenes at prime editordefined recombinase target sites, respectively [53]. Despite powerful and versatile, such combinatorial strategies require the delivery of large and multicomponent reagents into target cells. An aspect warranting attention when considering multiplexing approaches concerns the importance of introducing balanced amounts of the attendant individual components to maximize the performance and precision of genome editing interventions [55]. In addition, prime editing involving the delivery of dual pegRNAs is not compatible with large edits whereas sequential prime editing and site-specific recombination is not amenable to subtle genomic edits underlying endogenous gene repair due to discontinuous ‘footprint’ installation in the form of recombinase target sites. In conclusion, considering the herein discussed findings and matters, one can submit that cell therapy products derived from the use of RNA-programmable nickases as such or with heterologous domains, will offer a complementary set of ‘soft’ genome engineering options whose safety profiles are potentially higher than those associated with the exposure of cells to programmable nucleases.
Chapter 2 —— 26 —— Contributions The named author takes responsibility for the integrity of the work as a whole, and has given his approval for this version to be published. Acknowledgements None Disclosure and potential conflicts of interest Dr. Gonçalves is a member of the European Reference Network - Neuromuscular Diseases (ERN EURO-NMD). Funding declaration Research in the authors’ laboratory is supported by the Prinses Beatrix Spierfonds, the Duchenne Parent Project NL, the Dutch Research Council (NWO) - Open Technology Program, and EU Marie Skłodowska-Curie Doctoral Network Actions. Dr. Li holds a PhD fellowship from the China Scholarship Council - Leiden University Joint Scholarship Program
Chapter 2 —— 27 —— References 1. Cornu, T.I.; Mussolino, C.; Cathomen, T.. Refining strategies to translate genome editing to the clinic. Nat. Med. 2017, 23, 415–423. 2. Wang, J.Y.; Doudna, J.A.. CRISPR technology: a decade of genome editing is only the beginning. Science 2023, 379, eadd8643. 3. Finck, A.V.; Blanchard, T.; Roselle, C.P.; Golinelli, G.; June, C.H.. Engineered cellular immunotherapies in cancer and beyond. Nat. Med. 2022, 28, 678–689. 4. Sahillioglu, A.C.; Schumacher, T.N.. Safety switches for adoptive cell therapy. Curr. Opin. Immunol. 2022, 74, 190–198. 5. Chen, X.; Gonçalves, M.A.F.V.. DNA, RNA, and protein tools for editing the genetic information in human cells. iScience 2018, 6, 247–263. 6. Huang, X.; Yang, D.; Zhang, J.; Xu, J.; Chen, Y.E.. Recent advances in improving gene-editing specificity through CRISPR-Cas9 nuclease engineering. Cells 2022, 11, 2186. 7. Jang, H.K.; Song, B.; Hwang, G.H.; Bae, S.. Current trends in gene recovery mediated by the CRISPR-Cas system. Exp. Mol. Med. 2020, 52, 1016–1027. 8. Holkers, M.; Maggio, I.; Henriques, S.F.; Janssen, J.M.; Cathomen, T.; Gonçalves, M.A.. Adenoviral vector DNA for accurate genome editing with engineered nucleases. Nat. Methods 2014, 11, 1051–1057. 9. He, X.; Tan, C.; Wang, F.; et al. Knock-in of large reporter genes in human cells via CRISPR/Cas9-induced homology-dependent and independent DNA repair. Nucleic Acids Res. 2016, 44: e85. 10. Globerson Levin, A.; Riviere, I.; Eshhar, Z.; Sadelain. M.. CAR T cells: building on the CD19 paradigm. Eur. J. Immunol. 2021, 51, 2151–2163. 11. Detela, G.; Lodge, A.. EU regulatory pathways for ATMPs: standard, accelerated and adaptive pathways to marketing authorisation. Mol. Ther. Methods Clin. Dev. 2019, 13, 205–232. 12. Ivica, N.A.; Young C.M.. Tracking the CAR-T revolution: analysis of clinical trials of CAR-T and TCR-T therapies for the treatment of cancer (1997–2020). Healthcare (Basel) 2021, 9, 1062. 13. Vogler, M.; Shanmugalingam, S.; Sarchen. V.; et al. Unleashing the power of NK cells in anticancer immunotherapy. J. Mol. Med. (Berl.) 2022, 100, 337–349. 14. Muller, T.R.; Jarosch, S.; Hammel, M.; et al. Targeted T cell receptor gene editing provides predictable T cell product function for immunotherapy. Cell Rep. Med. 2021, 2, 100374. 15. Pavani, G.; Amendola, M.. Targeted gene delivery: where to land. Front. Genome Ed.
Chapter 2 —— 28 —— 2021; 2, 609650. Corrigendum: Front. Genome Ed. 2021, 3, 682171. 16. Odak, A.; Yuan, H.; Feucht, J.; et al. Novel extragenic genomic safe harbors for precise therapeutic T-cell engineering. Blood 2023, 141, 2698–2712. 17. Lieber, M.R.; Ma, Y.; Pannicke, U.; Schwarz, K.. Mechanism and regulation of human non-homologous DNA end-joining. Nat. Rev. Mol. Cell Biol. 2003, 4, 712–720. 18. Xue, C.; Greene, E.C.. DNA repair pathway choices in CRISPR-Cas9-mediated genome editing. Trends Genet. 2021, 37(7), 639–656. 19. Chen, X.; Tasca, F.; Wang, Q.; et al. Expanding the editable genome and CRISPRCas9 versatility using DNA cutting-free gene targeting based on in trans paired nicking. Nucleic Acids Res. 2020, 48, 974–995. 20. Frock, R.L.; Hu, J.; Meyers, R.M.; Ho, Y.J.; Kii, E.; Alt, F.W.. Genome-wide detection of DNA double-stranded breaks induced by engineered nucleases. Nat. Biotechnol. 2015, 33, 179–186. 21. Kosicki, M.; Tomberg, K.; Bradley, A.. Repair of double-strand breaks induced by CRISPR-Cas9 leads to large deletions and complex rearrangements. Nat. Biotechnol. 2018, 36, 765–771. 22. Turchiano, G.; Andrieux, G.; Klermund, J.; et al. Quantitative evaluation of chromosomal rearrangements in gene-edited human stem cells by CAST- Seq. Cell Stem Cell 2021, 28, 1136–1147. 23. Leibowitz, M.L.; Papathanasiou, S.; Doerfler, P.A.; et al. Chromothripsis as an ontarget consequence of CRISPR-Cas9 genome editing. Nat. Genet. 2021, 53, 895–905. 24. Amendola, M.; Brusson, M.; Miccio, A.. CRISPRthripsis: the risk of CRISPR/ Cas9induced chromothripsis in gene therapy. Stem Cells Transl Med. 2022, 11, 1003–1009. 25. Nahmad, A.D.; Reuveni, E.; Goldschmidt, E.; et al. Frequent aneuploidy in primary human T cells after CRISPR-Cas9 cleavage. Nat Biotechnol. 2022, 40, 1807–1813. 26. Tsuchida, C.A.; Brandes, N.; Bueno, R.; et al. Mitigation of chromosome loss in clinical CRISPR-Cas9-engineered T cells. bioRxiv 2023, 36993359. 27. Ihry, R.J.; Worringer, K.A.; Salick, M.R.; et al. p53 inhibits CRISPR-Cas9 engineering in human pluripotent stem cells. Nat. Med. 2018, 24, 939–946. 28. Haapaniemi, E.; Botla, S.; Persson, J.; Schmierer, B.; Taipale, J.. CRISPR-Cas9 genome editing induces a p53-mediated DNA damage response. Nat. Med. 2018, 24, 927–930. 29. Merkle, F.T.; Ghosh, S.; Kamitaki, N.; et al. Human pluripotent stem cells recurrently acquire and expand dominant negative P53 mutations. Nature 2017, 545, 229–233. 30. Sinha, S.; Barbosa, K.; Cheng, K.; et al. A systematic genome-wide mapping of oncogenic mutation selection during CRISPR-Cas9 genome editing. Nat. Commun.
Chapter 2 —— 29 —— 2021, 12, 6512. 31. Baslan, T.; Morris, J.P.4th; Zhao, Z.; et al. Ordered and deterministic cancer genome evolution after p53 loss. Nature 2022, 608, 795–802. 32. Slaymaker, I.M.; Gao, L.; Zetsche, B.; Scott, D.A.; Yan, W.X.; Zhang, F.. Rationally engineered Cas9 nucleases with improved specificity. Science 2016, 351, 84–88. 33. Kleinstiver, B.P.; Pattanayak, V.; Prew, M.S.; et al. High-fidelity CRISPR-Cas9 nucleases with no detectable genome-wide off-target effects. Nature 2016, 529, 490495. 34. Gonçalves, M.A.; van Nierop, G.P.; Holkers, M.; de Vries, A.A.. Concerted nicking of donor and chromosomal acceptor DNA promotes homology-directed gene targeting in human cells. Nucleic Acids Res. 2012, 40, 3443–3455. 35. Chen, X.; Janssen, J.M.; Liu, J.; et al. In trans paired nicking triggers seamless genome editing without double-stranded DNA cutting. Nat. Commun. 2017, 8, 657. 36. Nakajima, K.; Zhou, Y.; Tomita, A.; Hirade, Y.; Gurumurthy, C.B.; Nakada, S.. Precise and efficient nucleotide substitution near genomic nick via noncanonical homologydirected repair. Genome Res. 2018, 28, 223–230. 37. Jinek, M.; Chylinski, K.; Fonfara, I.; Hauer, M.; Doudna, J.A.; Charpentier, E.. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science 2012, 337, 816–821. 38. Wang, Q.; Liu, J.; Janssen, J.M.; Gonçalves, M.A.F.V.. Precise homology-directed installation of large genomic edits in human cells with cleaving and nicking highspecificity Cas9 variants. Nucleic Acids Res. 2023, 51, 3465–3484. 39. Kocher, T.; Wagner, R.N.; Klausegger, A.; et al. Improved double-nicking strategies for COL7A1-editing by homologous recombination. Mol. Ther. Nucleic Acids 2019, 18, 496–507. 40. Hyodo, T.; Rahman, M.L.; Karnan, S.; et al. Tandem paired nicking promotes precise genome editing with scarce interference by p53. Cell Rep. 2020, 30, 1195–1207. 41. Rahman, M.L.; Hyodo, T.; Karnan, S.; et al. Experimental strategies to achieve efficient targeted knock-in via tandem paired nicking. Sci. Rep. 2021, 11, 22627. 42. Bollen, Y.; Stelloo, E.; van Leenen, P.; et al. Reconstructing single-cell karyotype alterations in colorectal cancer identifies punctuated and gradual diversification patterns. Nat. Genet. 2021, 53, 1187–1195. 43. Fortschegger, K.; Husa, A.M.; Schinnerl, D.; Nebral, K.; Strehl, S.. Expression of RUNX1-JAK2 in Human induced pluripotent stem cell-derived hematopoietic cells activates the JAK-STAT and MYC pathways. Int. J. Mol. Sci. 2021, 22, 1. 44. Bollen, Y.; Hageman, J.H.; van Leenen, P.; et al. Efficient and error-free fluorescent
Chapter 2 —— 30 —— gene tagging in human organoids without double-strand DNA cleavage. PLoS Biol. 2022, 20, e3001527. 45. Hamaker, N.K.; Lee, K.H.. High-efficiency and multilocus targeted integration in CHO cells using CRISPR-mediated donor nicking and DNA repair inhibitors. Biotechnol. Bioeng. 2023, 120, 2419–2440. 46. Hirano, S.; Kappel, K.; Altae-Tran, H.; et al. Structure of the OMEGA nickase IsrB in complex with ωRNA and target DNA. Nature 2022; 610, 575–581. 47. Yan, W.X.; Hunnewell, P.; Alfonse, L.E.; et al. Functionally diverse type V CRISPRCas systems. Science 2019, 363, 88–91. 48. Zhang, H.; Li, Z.; Xiao, R.; Chang, L.. Mechanisms for target recognition and cleavage by the Cas12i RNA-guided endonuclease. Nat. Struct. Mol. Biol. 2020, 27, 1069–1076. 49. Chen, X.; Gonçalves, M.A.. Engineered viruses as genome editing devices. Mol. Ther. 2016, 24, 447–457. 50. Davis, L.; Khoo, K.J.; Zhang, Y.; Maizels, N.. POLQ suppresses interhomolog recombination and loss of heterozygosity at targeted DNA breaks. Proc. Natl Acad. Sci. USA 2020, 117, 22900–22909. 51. Roy, S.; Juste, S.S.; Sneider, M.; et al. Cas9/Nickase-induced allelic conversion by homologous chromosome-templated repair in Drosophila somatic cells. Sci. Adv. 2022, 8, eabo0721. 52. Tomita, A.; Sasanuma, H.; Owa, T.; et al. Inducing multiple nicks promotes interhomolog homologous recombination to correct heterozygous mutations in somatic cells. Nat. Commun. 2023, 14, 5607. 53. Chen, P.J.; Liu, D.R.. Prime editing for precise and highly versatile genome manipulation. Nat. Rev. Genet. 2023, 24, 161–177. 54. Wang, Q.; Liu, J.; Janssen, J.M.; Tasca, F.; Mei, H.; Gonçalves, M.A.F.V.. Broadening the reach and investigating the potential of prime editors through fully viral genedeleted adenoviral vector delivery. Nucleic Acids Res. 2021, 49, 11986–12001. 55. Tasca, F.; Brescia, M.; Liu, J.; Janssen, J.M.; Mamchaoui, K.; Gonçalves, M.A.F.V.. High-capacity adenovector delivery of forced CRISPR-Cas9 heterodimers fosters precise chromosomal deletions in human cells. Mol. Ther. Nucleic Acids 2023, 31, 746– 762.
Chapter 3 AAV-vectored base editor transsplicing delivers dystrophin repair Zhen Li1 and Manuel A.F.V. Gonçalves1 Cell & Gene Therapy Insights 9:1201-1210 (2023) 1Leiden University Medical Centre, Department of Cell and Chemical Biology, Einthovenweg 20, 2333 ZC, Leiden, The Netherlands.
Chapter 3 —— 32 —— Duchenne muscular dystrophy (DMD) (MIM: 310200) is a severe and frequent neuromuscular disorder (incidence of ~1 in 5,500 boys) caused by mutations in the vast X-linked DMD gene (~2.2 Mb), whose largest product, the long rod-shaped 427-kDa dystrophin isoform, is a key structural component of the striated musculature [1]. Contributing to the urgency in the development of currently inexistent DMD therapies is the observation that ~1/3 of cases arise de novo through germline mutations, often intragenic deletions that disrupt the mRNA reading frame. Critically, naturally occurring DMD gene deletions resulting in in-frame transcripts coding for internally truncated, yet partially functional, dystrophins cause the milder Becker muscular dystrophy (MIM: 300376). Hence, DMD gene manipulations yielding Becker-like dystrophins via direct coding sequence reframing or exon skipping have the potential of offering long-lasting therapeutic effects [1]. Toward this end, among other technologies, CRISPR-Cas9 nucleases and adeno-associated viral (AAV) vectors are being investigated for rescuing dystrophin expression upon double-strand DNA break (DSB) formation and ensuing chromosomal end-joining [1]. These experiments demonstrate that AAV/CRISPR-Cas9-based dystrophin restoration can improve striated muscle function in mice; however, a potentially insidious outcome is the identification of prevalent capture of Cas9-encoding AAV at nuclease target sites, including at Dmd exons 51 and 53 [2,3]. Moreover, programmable nucleases can trigger other untoward effects, e.g., locus- or chromosomewide rearrangements [4]. There is, therefore, a pressing need to expand candidate DMD genetic therapies to those based on DSB-independent genome editing systems. In a timely study published in Molecular Therapy – Nucleic Acids, Chai and coworkers [5] identify adenine base editors (ABEs) and guide RNAs (gRNAs) (Figure 1) that, after implementing single basepair substitutions (i.e., A·T-to-G·C transitions) at splicing motifs, a process that they name “single-swap” editing, lead to genotype-specific DMD repair through exon skipping. Next, the authors assemble a dual AAV ABE transsplicing system to demonstrate in dystrophin-defective mice the amelioration
Chapter 3 —— 33 —— of dystrophic traits at the cellular and organismal levels upon intramuscular or systemic administrations. This study identifies ABE:gRNA complexes compatible with ~30% of DMD-causing genotypes and, notwithstanding its inherent complexity, establishes dual AAV ABE trans-splicing as a DSB-free DMD gene correction option. Figure 1. Adenine base editing. Adenine base editors (ABEs) catalyze A·T-to-G·C substitutions and consist of a fusion product between a disabled or nicking Cas9, or ortholog protein, and an evolved deoxyadenosine deaminase, e.g., Escherichia coli tRNA adenosine deaminase (TadA) derivatives. Upon PAM binding, ABE:gRNA complexes form an R-loop at a gRNA-defined target sequence exposing a region of single-stranded DNA. A nucleotides in this single-stranded protospacer “bubble” become targets for the ABE effector domain that converts A nucleotides to inosine (I) intermediates preferentially within an “activity window.” Subsequently, nicking of the unedited strand induces DNA repair that installs C nucleotides opposite I intermediates with additional DNA repair events (or replication) establishing the final A·T-to-G·C transitions. Chai and colleagues start by testing in HEK293T cells and DMD iPSCs, ABE:gRNA complexes that, depending on their ABE component, i.e., ABE8e [6] or ABEe-NG, recognize, respectively, canonical NGG or NG PAMs. DNA sequencing assays in DMD iPSCs identified ABE:gRNA complexes yielding high-frequency target-base editing at DMD exons 51 and 45 splice acceptor (SA) motifs (71.6% and 79.3%–83.3%, respectively). As a result,
Chapter 3 —— 34 —— robust expression of Becker-like dystrophins was detected in cardiomyocytes differentiated from base-edited DMD iPSCs. Interestingly, it was found that DMD exon 51 targeting serendipitously established an in-frame 11-nucleotide deletion instead of the intended exon skipping, presumably due to internal cryptic splice site usage. This finding per se stresses the importance of carefully assessing gene-edited products even when using subtle DSBindependent systems. Moreover, various amounts of bystander A·T-to-G·C transitions were also detected (range: 3.3%–91%). These bystander changes might have limited consequences as they map to either spliced-out intron or, if accompanied with the intended SA edits, to skipped exon sequences. Further investigations will be, nonetheless, necessary to probe for slight (or otherwise) splicing alterations in different cell types or contexts. Regardless, base editors with narrower “editing windows” should facilitate more favorable target-to-bystander ratios. AAV vectors are attractive in vivo gene-editing tool delivery vehicles owing to their lack of viral genes and serotype diversity of their parental viruses. Indeed, packaging vector genomes in AAV serotype capsids with a strong tropism for certain cell types (pseudotyping) facilitates tissue-directed transductions. However, the limited AAV packaging capacity (<4.7 kb) permits delivering neither base editing nor other large constructs. To obviate this limitation, researchers are developing base editors with compact architectures [7], testing alternative delivery systems or applying dual AAV strategies in which split constructs linked to N- and C-terminal intein domains are packaged in different AAV vectors (Figure 2). Upon target cell cotransductions, intein-mediated protein trans-splicing results in the in situ assembly of full-length proteins. Indeed, dual AAV-vectored base editor trans-splicing is currently undergoing intense investigation for tackling various disease-causing mutations, including DMD mutations [8,9]. In Chai et al. [5], a dual AAV ABE trans-splicing system is assembled to address frequent DMD deletions through exon skipping modulation. By exploiting high human-murine conservation over intron 44 to exon 45 junctions and
Chapter 3 —— 35 —— guided by their earlier in vitro experiments, the authors apply dual AAVs to deliver a split version of an ABE8e variant (i.e., ABE.TadA-8eV106W), selected for its reduced off-target nucleic acid deaminase activities [6], together with a gRNA targeting the exon 45 SA region (Figure 2). Figure 2. Dual AAV ABE trans-splicing system for dystrophin repair. Two AAV serotype-9 vectors expressing separated portions of an adenine base editor (ABE.TadA8eV106W) and a gRNA (gRNAEx45) lead to intein-mediated assembly of complete ABE:gRNA complexes. Base editing involving A·T-to-G·C transitions at the splice acceptor site of exon 45 establishes permanent exon 45 skipping in striated muscle cells. In dystrophin-defective muscle cells lacking exon 44, exon 45 skipping restores the reading of mature transcripts that code for a truncated Becker-like dystrophin with therapeutic potential for DMD patients. ITR, T-shaped hairpin-structured AAV serotype2 inverted terminal repeats (cis-acting elements needed for vector DNA replication and packaging in producer cells). CK8e and An, synthetic striated muscle-specific promoter and polyadenylation signal, respectively. The dystrophin diagram was generated via: http://edystrophin.genouest.org/index.php?page=home. To favor base editing in striated muscles over non-target organs of Dmd exon 44-deleted mice, the vector constructs were packaged in AAV serotype-9 capsids with the split ABE.TadA-8eV106W moieties being expressed through the tissue-specific CK8e promoter. Intramuscular dual AAV coadministrations of 1 x 1011 total vector genomes (VGs) per tibialis anterior
Chapter 3 —— 36 —— (TA) led to 29.5% ± 2.7% A·T-to-G·C edits with minimal indel formation (i.e., 0.2% ± 0.1%), as determined by deep sequencing at 3 weeks post injection. Systemic dual AAV co-administrations of 1.5 x 1014 VG kg-1 and 3 x 1014 VG kg-1 via the temporal facial veins of postnatal day 2 (P2) mice yielded, in TA muscles, 5.5% ± 1.2% and 8.1% ± 3.0% edits and, in hearts, 22.0% ± 2.2% and 26.2% ± 4.4% edits, respectively, with <0.1% indels detected at 8 weeks post injection. Critically, a general dose-dependent improvement of disease-associated molecular, cellular, and functional endpoints is reported. In this regard, up to 31% and 60% of wild-type dystrophin levels estimated in treated TA and heart muscles corresponded to over 76% and 95% of dystrophin-positive myofibers and cardiomyocytes, respectively. Partial dystrophin rescue translated, in turn, in noticeable reduction of myofiber central nucleation, diameter distribution, and fibrosis, all hallmarks of muscle degeneration. Finally, Dmd exon 44-deleted mice systemically treated at P2 with low and high doses of dual AAV ABE transsplicing particles registered 31% and 41% grip strength augmentation, respectively, when compared with their untreated counterparts. Follow-up studies will be instructive to determine the long-term effects of the local and systemic gene-editing procedures in the treated animals. Of notice, despite the aforementioned transductional and transcriptional targeting measures, significant base editing was detected in the liver (i.e., 11.1% ± 5.9%), which correlated with high VG copy numbers present specifically in this organ. These data confirm the importance of developing liver de-targeting protocols and strictly myotropic AAV capsids [10]. Indeed, as dose-dependent toxicity and immunological constrains have emerged during AAV clinical applications, optimization of tissue tropism and expression will be particularly important for dual AAV trans-splicing approaches due to the necessarily higher particle amounts required for maximizing co-expression and full-length protein assembly. Moreover, besides seeking the elimination of the observed editing at two of five top-ranked in silico-predicted candidate off-target sites [5], unbiased genome- and transcriptome-wide assessments of
Chapter 3 —— 37 —— off-target effects, including gRNA-independent deamination, will complement the safety profile of DMD-targeting ABE:gRNA complexes. Concluding, Chai and colleagues demonstrate that AAV-vectored ABE transsplicing can induce robust synthesis of Becker-like dystrophins in striated muscles of dystrophic mice upon DSB-free splice site knockout and exon skipping [5], The resulting improvement in pathological traits measured at the molecular, tissue, and functional levels validates the use of this platform for the efficacious testing and optimization of the growing number of ABE reagents in vivo and expands the range of potential treatment modalities for DMD patients. Acknowledgements Research in our laboratory is supported by the Prinses Beatrix Spierfonds (W.OR21-01), the Duchenne Parent Project NL, the Dutch Research Council (NWO) – Open Technology Program, and EU Marie Skłodowska-Curie Doctoral Network Actions. Z.L. holds a Ph.D. fellowship from the China Scholarship Council – Leiden University Joint Scholarship Program. Authors of this work are members of the European Reference Network – Neuromuscular diseases (ERN EURO-NMD). Declaration of interest The authors declare no competing interests.
www.ridderprint.nlRkJQdWJsaXNoZXIy MTk4NDMw