|Home | About | Journals | Submit | Contact Us | Français|
Mutations in DMD disrupt the reading frame, prevent dystrophin translation, and cause Duchenne muscular dystrophy (DMD). Here we describe a CRISPR/Cas9 platform applicable to 60% of DMD patient mutations. We applied the platform to DMD-derived hiPSCs where successful deletion and non-homologous end joining of up to 725kb reframed the DMD gene. This is the largest CRISPR/Cas9-mediated deletion shown to date in DMD. Use of hiPSCs allowed for evaluation of dystrophin in disease relevant cell types. Cardiomyocytes and skeletal muscle myotubes derived from reframed hiPSC clonal lines had restored dystrophin protein. The internally deleted dystrophin was functional as demonstrated by improved membrane integrity and restoration of the dystrophin glycoprotein complex in vitro and in vivo. Furthermore, miR31 was reduced upon reframing, similar to observations in Becker muscular dystrophy. This work demonstrates the feasibility of using a single CRISPR pair to correct the reading frame for the majority of DMD patients.
Duchenne muscular dystrophy (DMD) is the most common fatal genetic disease of childhood, affecting ~1 in 3500–5000 boys. In DMD, progressive muscle degeneration generally leads to death in the twenties, and there are currently no highly effective therapies. DMD is often caused by frameshifting exonic deletions in DMD, which encodes dystrophin. Dystrophin stabilizes the dystrophin glycoprotein complex (DGC) at the sarcolemma; loss of functional dystrophin leads to the degradation of DGC components which results in muscle membrane fragility and leakage of creatine kinase (CK) (Pearce et al., 1964). Approximately 60% of mutations causing Duchenne occur between DMD exons 45–55 (Beroud et al., 2007). Multiple independent clinical reports in patients and dystrophic mice have revealed that in-frame deletions of exons 45–55 produce an internally deleted dystrophin protein and are associated with a very mild Becker muscular dystrophy (BMD) disease course, with some patients still asymptomatic in their 60s (Beroud et al., 2007; Echigoya et al., 2015; Nakamura et al., 2008; Taglia et al., 2015). Thus, genetic manipulation to create a large deletion of exons 45–55 is a therapeutic strategy to restore the reading frame for 60% of DMD patients with mutations in this region.
One promising approach to induce genetic correction of DMD is through the use of the bacterially acquired immune surveillance system known as clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated nuclease (Cas) 9. In this system a short guide RNA (gRNA), which is complimentary to a specific site in the genome, is used to target the Cas9 nuclease and induce double stranded breaks (DSBs). The DSB can be repaired through non-homologous end joining (NHEJ) or homology directed repair.
Previous work has shown that CRISPR/Cas9 components can modify the DMD gene (Li et al., 2015; Long et al., 2015, 2014; Nelson et al., 2015; Ousterout et al., 2015; Tabebordbar et al., 2015; Wojtal et al., 2016; Xu et al., 2015). In this investigation, we describe a therapeutically relevant CRISPR/Cas9 platform that we designed to modify DMD. Our platform involves excision of exons 45–55 and NHEJ to reframe dystrophin through creation of an internally deleted protein that is stable and functional. The internally deleted protein mimics the naturally occurring exon 45–55 deletion observed in mild BMD patients and encompasses 60% of DMD patient mutations.
For the first time, we demonstrate CRISPR/Cas9-mediated deletion and NHEJ of up to 725kb of the DMD gene in human induced pluripotent stem cell (hiPSC) lines. We show that CRISPR/Cas9 reframed hiPSC-derived skeletal and cardiac muscle cells express stable dystrophin that improves membrane stability and restores a DGC member, β-dystroglycan. We also demonstrate reduced microRNA 31 (miR31) levels after the reading frame is restored, consistent with the observations made in BMD patients (Cacchiarelli et al., 2011). Furthermore, we show restoration of dystrophin and β-dystroglycan in vivo after engraftment of reframed hiPSC-derived skeletal muscle cells into a mouse model of DMD. This work sets the stage for use of reframed DMD hiPSC-derived cells or in vivo correction strategies using CRISPR/Cas9 for direct translation to patients with DMD.
We have developed several xenobiotic-free hiPSC lines derived from wild type and DMD patient fibroblasts using current good manufacturing practice protocols. Each DMD hiPSC line harbors a unique frame-shifting DMD mutation within the exon 45–55 hotspot region. All hiPSC lines (Center for Duchenne Muscular Dystrophy (CDMD) 1003, 1006 and 1008) express pluripotency markers (NANOG and SOX2) and are karyotypically normal (Figures 1A and 1B). CDMD hiPSCs maintain pluripotency, as they form teratomas in vivo that represent all three germ layers (Figure 1C), and each harbor unique mutations (Figure 1D).
In order to delete exons 45–55 of DMD, gRNAs were designed to target introns 44 and 55. gRNA sites were chosen to only retain ~500bp of the intron next to each of the flanking exons (44 and 56). The rationale for this design is to develop gRNAs applicable to as many patient mutations as possible and to ensure a small functional chimeric intron is generated. During NHEJ, the 3′ end of intron 44 and the 5′ end of intron 55 join to create a ~1kb chimeric intron (Figure 2A). We expect introns generated in this manner are functional and splice correctly to create an in-frame transcript, with exon 44 joined with exon 56.
Since hiPSCs are challenging to genetically manipulate, human embryonic kidney (HEK) 293FT cells were used to screen five gRNAs at each intronic region. All gRNAs demonstrated individual cutting activity on Surveyor assay up to 34% (Figures S1A and S1B). Using multiplex PCR, gRNAs transfected in pairs were shown to effectively delete the entire 708kb region encompassing exons 45–55 (Figures S1C and S1D).
In order to assess the feasibility of an exon 45–55 deletion across different patient mutations, we applied our gRNAs to three DMD hiPSC lines. The lines (CDMD 1003, 1006, 1008) require ~530kb, 670kb, or 725kb for successful deletion and NHEJ of DMD respectively. The gRNAs used were shown to be active in all three lines and effectively deleted exons 45–55 (Figures S2 and S3). Transient puromycin selection of cells nucleofected with the CRISPR plasmids improved the efficiency of deletion in CDMD 1003 and 1006 hiPSCs (Figure S3D).
Stably deleted DMD hiPSC lines were generated from CDMD 1003 and 1006 by clonal selection after nucleofection with the gRNA pair 44C4 and 55C3 (Figure 2B and 2C) and are pluripotent (Figures 2C and S4B). All reframed lines were karyotypically normal except for one clone (CDMD 1003-81), which was found to contain a 1q32 amplification confirmed via FISH analysis (Figure S4A), also observed in the original parental line and in all daughter clones after post-hoc analysis. The 1q32 amplification is common in hPSCs after extended propagation in culture (Dekel-Naftali et al., 2012), and thus was not a result of CRISPR-mediated off target activity. To determine off target activity of our gRNAs, the top 10 homologous sites per guide were determined by COSMID (Cradick et al., 2014) and sequenced in all clonal and parental lines. No off target mutations were observed at any site (Table S2). All variants, besides a heterozygous SNP in chromosome 11, were detected in less than 1% of reads which is consistent with error in the sequencing method.
CRISPR/Cas9-mediated deletion of DMD should result in an internally deleted dystrophin protein lacking exons 45–55 (hereafter referred to as DYSΔ45–55). As hiPSCs do not express dystrophin, we differentiated the reframed DMD hiPSC clonal lines to two disease-relevant cell types, cardiomyocytes and skeletal muscle myotubes, using directed differentiation or overexpression of MyoD to evaluate rescue of DYSΔ45–55. PCR and sequencing of the exon 44/56 boundary in cDNA from the reframed cardiomyocyte clones demonstrated correct splicing of the dystrophin transcript (Figures S4C and S4D). Additionally, both the reframed cardiac and skeletal muscle cell lines restored dystrophin expression as assayed by immunocytochemistry and Western blot (Figures 3A–3C). Compared to wild type CDMD 1002 or human skeletal muscle myotubes (HSMM), the band was truncated by ~66kDa as expected.
Cardiomyocytes or skeletal myotubes lacking dystrophin demonstrate membrane fragility in vitro and respond to osmotic stress by releasing elevated levels of CK (Guan et al., 2014; Menke and Jockusch, 1995), as is seen in human patients (Pearce et al., 1964). To determine whether DYSΔ45–55 could restore stability to dystrophic plasma membranes, we subjected differentiated cardiomyocytes and skeletal muscle myotubes derived from reframed and out-of-frame hiPSCs to hypoosmotic conditions. Cells were stressed by incubation in hypoosmolar solutions (66–240mosmol) and CK release into the supernatant was measured to show functional improvement after dystrophin restoration. Both the reframed CDMD 1003-49 cardiomyocytes and skeletal muscle cells demonstrated reduced CK release, similar to wild type (CDMD 1002), versus the out-of-frame CDMD 1003 cells, indicating that DYSΔ45–55 was capable of reducing membrane fragility (Figure 4A). The same trend was also observed with CDMD 1006/1006-1 cardiomyocytes (Figure S4E). After normalizing and pooling all experiments, we observed that significantly less CK was released at 93, 135, and 240mosmol in the reframed and wild type cells compared to out-of-frame (Figure S4F).
Elevated levels of miR31 have been observed in DMD patient biopsies compared to wild type or BMD (Cacchiarelli et al., 2011). We measured levels of miR31 using droplet digital PCR (ddPCR) after differentiation of out-of-frame and reframed CDMD hiPSCs to skeletal myotubes. Reframing DMD reduced levels of miR31 (similar to wild type cells), compared to out-of-frame DMD, as is observed in human dystrophinopathies (Figure 4B). Thus, reframing the DMD gene normalizes miR31 levels similar to BMD, demonstrating functional rescue of the dystrophic phenotype to a BMD phenotype.
As a third assay of DYSΔ45–55 functionality, we evaluated its ability to restore the DGC in vitro and in vivo. The DGC member, β-dystroglycan, was restored and detected at the membrane of reframed hiPSCs, but not out-of-frame hiPSCs, after directed differentiation to skeletal muscle in vitro by immunostaining and Western blot (Figure 4C and 4D). Additionally, skeletal muscle cells derived from a wild type (CDMD 1002), out-of-frame (CDMD 1003), or reframed (CDMD 1003-49) hiPSC line were injected into the tibialis anterior of NSG-mdx mice. Correctly localized dystrophin and β-dystroglycan was only observed in engrafted human cells (demarked by human lamin A/C and spectrin) from the reframed or wild type lines (Figure 4E and 4F). These studies taken together with the hypoosmotic stress assays demonstrate the ability of DYSΔ45–55 to functionally reassemble the DGC and restore membrane stability in vitro and in vivo.
Using CRISPR/Cas9 gene editing, we have induced the largest deletion accomplished to date in DMD hiPSCs and restored a functional dystrophin protein. Deletion of DMD exons 45–55 has the potential to be therapeutically relevant to 60% of Duchenne patients. Since this internal deletion has been associated with a very mild disease course in multiple independent patients, a therapy utilizing this approach should create a highly functional dystrophin. We showed successful deletion of exons 45–55 using a single gRNA pair and did not identify any off target activity at the top 10 homologous sites, however a more comprehensive and unbiased approach should be undertaken such as whole genome sequencing. Importantly, removal of exons 45–55 resulted in stable dystrophin protein (DYSΔ45–55) in both cardiomyocytes and skeletal myotubes in vitro. Functionality of DYSΔ45–55 was tested in cardiomyocytes and skeletal muscle derived from reframed DMD hiPSCs and demonstrated improved membrane stability by a physiologically relevant measure of CK release, similar to wild type. The ability to evaluate cardiomyocyte functionality is an advantage of using hiPSCs, as some current preclinical and clinical studies for DMD therapies do not efficiently target the heart (e.g. exon skipping; Arechavala-Gomeza et al., 2012). Additionally, we demonstrated a normalization in miR31 levels, a microRNA that inhibits dystrophin, after reading frame restoration similar to what is observed in human BMD patients (Cacchiarelli et al., 2011). Finally, we show restored DGC localization in vitro and in vivo, which further validates the functionality of DYSΔ45–55.
Previous work by Ousterout et al. (2015) demonstrated that multiplexed gRNAs can restore the DMD reading frame in primary myoblasts. However, myoblasts do not provide a renewable source of stem cells, which is a requirement for long-term therapeutic efficacy (Partridge, 2002). In contrast, we used hiPSCs which offer the opportunity to evaluate the internally deleted dystrophin protein in multiple cell types that are affected in DMD, and in future studies, may provide a renewal source of corrected progenitor cells. Our work is further distinguished from previous studies as we are the only group to show restoration of dystrophin function on membrane integrity, miR31 expression, and the DGC in cardiac and skeletal muscle cells following CRISPR-mediated gene editing.
An advantage of our CRISPR platform is the therapeutic potential of a single pair of gRNAs to treat the majority of DMD patients. By designing gRNAs that accomplish a deletion that encompasses the majority of DMD mutations, this approach is optimized for future clinical studies. It would be unreasonable to design, validate, and evaluate off targets for every new CRISPR pair tailored for each individual patient. Additionally, CRISPR/Cas9 is advantageous over exon skipping, as it results in permanent restoration of reading frame as opposed to transient effects on RNA splicing. Previously, Li et al. (2015) used CRISPR/Cas9 to induce exon skipping, frameshifting, or exon knockin to restore dystrophin in a DMD hiPSC line with an exon 44 deletion; however, their platform is only applicable to 3–9% of DMD patients (Bladen et al., 2015), and two of their strategies relied on creation of indels which would be difficult to apply consistently to each patient. While Ousterout et al. deleted exons 45–55, they removed significantly less of the intervening region (336kb) and thus their approach would cover fewer patient mutations within the hotspot region. This is because many mutations extend into the intronic region, thus by designing gRNAs that encompass more of the intron, our platform is applicable to more patients.
Another benefit of using this platform to delete a large portion of DMD, as opposed to single exons, is the known correlation of DYSΔ45–55 with a mild BMD phenotype. Large deletions in the rod domain of dystrophin often produce a more functional (more like wild type) protein, than even very small deletions (Harper et al., 2002). Larger deletions, which remove hinge III (exons 50–51), are believed to lead to a milder BMD phenotype than smaller deletions, or those which retain hinge III (Carsana et al., 2005). Thus, in many cases larger deletions are more therapeutically beneficial than smaller ones, due to the way they affect the secondary structure of the protein.
In summary, we have developed a potentially therapeutic gene editing platform for DMD to permanently restore dystrophin reading frame in multiple patient-derived hiPSCs. Our approach using CRISPR/Cas9 and NHEJ deletes up to 725kb of DMD encompassing exons 45–55 and restores dystrophin protein function in both cardiomyocytes and skeletal muscle cells derived from corrected hiPSCs. A current limitation of this platform is that clinical protocols still need to be developed that allow for rapid clonal line derivation, and the utilization of hiPSC-derived cardiac and skeletal muscle progenitors combined with gene correction. Alternatively, CRISPR/Cas9 to restore the reading frame in DMD mouse models has been delivered directly in vivo (Long et al., 2015; Nelson et al., 2015; Tabebordbar et al., 2015). Thus, applications of this platform in the future will allow for the development of an in situ gene correction strategy or ex vivo gene correction followed by autologous cell transplantation, either of which offers tremendous potential for DMD.
Skeletal muscle differentiation from hiPSCs was induced using overexpression of a tamoxifen inducible MyoD-ERT lentivirus or an adapted 50 day directed differentiation protocol where NCAM+ HNK1− cells were fluorescently activated cell sorted at day 50. Cardiomyocytes were derived through aggregates over 30 days. See Supplemental Experimental Procedures.
NOD scid IL2Rgamma (NSG) immunodeficient mice (Jackson Laboratory) were crossed to mdx scid mice (Jackson Laboratory) to generate NSG-mdx mice, see Supplemental Experimental Procedures. 5–7 week-old NSG-mdx mice were pretreated with 50μl of 10μM cardiotoxin (Sigma-Aldrich) injected into the right tibialis anterior (TA) 24hrs prior to engraftment. For MyoD OE cells, 100μl of 5mg/ml tamoxifen (Sigma-Aldrich) was IP injected for 5 days beginning the day prior to engraftment. 1×106 cells in HBSS were injected intramuscularly and the TA was harvested after 30 days. See Supplemental Experimental Procedures.
Terminally differentiated skeletal muscle cells and cardiomyocytes plated in duplicate were stressed by incubation in hypoosomolar solutions ranging from 66–240mosmol, see Supplemental Experimental Procedures, for 20mins at 37°C. CK was measured in triplicate from the supernatant and cell lysate with the Creatine Kinase-SL kit (Sekisui Diagnostics) according to the manufacturer’s instructions.
We would like to thank Doug Black, Richard Wang, and Haibin Xi for suggestions and Diana Becerra and Jane Wen for technical assistance. The following cores were utilized: CDMD Muscle Phenotyping and Imaging Core, HTS and Cell Repository Core, and Bioinformatics and Genomics Core (supported by NIAMS-P30 AR057230), the UCLA Broad Stem Cell Research Center (BSCRC) Flow Cytometry Core Resource, the UCLA Jonsson Comprehensive Cancer Center (JCCC) and Center for AIDS Research (CFAR) Flow Cytometry Core Facility (supported by NIH P30 CA016042, 5P30 AI028697), the UCLA Integrated Molecular Technologies Core (supported by CURE/P30 DK041301), the UCLA Humanized Mouse Core (supported by the JCCC, BSCRC, CFAR, NIH/NIAID AI028697), and the UCLA GenoSeq core. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship Program under Grant No. DGE-1144087 (CSY). Funding was provided by NIAMS of the NIH (P30 AR057230-MJS, R01AR064327-ADP) and the Eli & Edythe Broad Center of Regenerative Medicine and Stem Cell Research at UCLA (MJS, ADP) and Rose Hills Foundation Research Award (ADP).
Author contributionsConceptualization, Methodology, Writing-Original Draft, Visualization, Project Administration - CSY, MJS, ADP; Validation- CSY, MRH, NVE; Formal Analysis- CSY; Investigation- CSY, MRH, NVE, HN, MJ, SY, SK CK-C, DW; Resources- AN, SFN, MCM, MJS, ADP; Writing- Review and Editing- CSY, MRH, NEV, SK, CK-C, DBK, AN, SFN, MCM, MJS, ADP; Supervision- JAZ, DBK, AN, MCM, MJS, ADP; Funding Acquisition- MJS, ADP.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.