Gene editing and gene-based therapeutics for cardiomyopathies

Abstract

With an increasing understanding of genetic defects leading to cardiomyopathy, focus is shifting to methods for correcting these underlying genetic defects. One approach involves treating mutant RNA through antisense oligonucleotides, and the first drug has now received regulatory approval to treat specific mutations associated with Duchenne muscular dystrophy. In contrast, gene editing targets DNA, and is currently being evaluated in the preclinical setting. For inherited cardiomyopathies, especially those that are caused by autosomal dominant mutations, genetic correction strategies require tight specificity for the mutant allele. Gene editing methods are currently in testing to create deletions that may be useful to restore protein expression by bypassing mutations and restoring protein production. Site specific gene editing is a less efficient process than inducing deletions. In both cases, it is essential to target the mutant allele while avoiding the non-mutant allele. Antisense oligonucleotide drugs have limited access to tissues, including the heart. Importantly, it is necessary to provide continual dosing with antisense oligonucleotides as their effect is temporary. For DNA-mediated gene editing, current research approaches rely on viral delivery of nucleases and guide RNAs to the heart. In principle, gene editing requires a one-time treatment as a permanent genetic correction. Gene editing targeting the germline is associated with ethical dilemmas.

Introduction

Cardiomyopathy and heart failure are under genetic influence. Genetic correction technologies are rapidly emerging, providing the tools to correct underlying genetic defects responsible for human disease, including cardiomyopathy. Gene editing strategies like CRISPR/Cas9 act on DNA. There are also RNA targeted approaches, including antisense oligonucleotides or even gene editing, which are used to suppress mutations or promote expression of functional molecules. Genetic correction strategies are designed to reverse specific mutations responsible for disease, or these same methods can also be applied to modulate normal sequences in order to improve heart function. Most genetic correction approaches are gene and mutation specific, and therefore require knowledge of the underlying genetic mutations responsible for cardiomyopathy and heart failure, further underscoring the importance of genetic diagnosis.

Technical advances enabling genetic editing in DNA

Tools for genetic engineering remain at the crux of scientific discovery and are poised for therapeutic application in heart failure. The Human Genome Project, and the subsequent efforts to define human genetic variation using larger scale efforts, have revolutionized the way in which we approach heart failure and cardiomyopathy ^1–3. Simultaneous with these efforts, the Xenopus-derived Zinc Finger Nuclease (ZFN) emerged as an early technology to change genome sequences ⁴. This DNA binding motif was designed to recognize DNA sequences with high specificity ⁵, but required engineering protein motifs to recognize DNA sequences of interest. Despite the broad application of these engineered nucleases for gene correction, obstacles in targeting complex sequences and interactions compromising the specificity of this system limited the widespread use of this for therapeutic genetic correction. Yet, ZNF served as a gateway for the development of new and improved gene editing technologies.

Transcription Activator-like Effector Nucleases (TALENs) represented an advance for genomic engineering ^6,7. These nucleases are engineered constructs that contain a DNA-binding domain and a nonspecific nuclease domain ⁸. TALENs generate double strand breaks (DSBs) at specific sites of interest within a given DNA sequence, which are subsequently repaired by nonhomologous end joining (NHEJ). This process induces the formation of insertion and deletion mutations. Unlike ZFNs, TALENs’ comparatively fast and easy construction allowed for a more precise and efficient method for genomic targeting ⁹. The emergence of TALENs advanced pre-existing methods for gene editing, and these nucleases served as a guide for the onset of innovative tools for gene-based therapies.

Found in both bacteria and archaea, Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/Cas9 emerged as the next generation in genome editing as a system that modifies DNA by generating site specific cleavage events, which may then be followed by a template driven repair process ^4,10. With a template-driven repair process, specific sequences can be changed or new sequences can be inserted or deleted. Similar to previous mechanisms, CRISPR/Cas9 functions by creating DSBs at precise sites within the DNA. The site specificity is driven by guide RNAs, which by virtue of having homology to the site of interest, use this homology to carry the Cas9 nuclease to specific sites. After cleavage by Cas9, DSBs can be repaired by NHEJ, which most commonly results in deletions of varying size and length. Alternatively, in the presence of a template, homology-directed repair (HDR) occurs, in which the DSB is repaired to resemble the template. HDR can be exploited to yield site-specific precise genetic correction from a mutation to a normal allele. HDR can also be exploited to add sequences of interest or more precisely delete selected regions (reviewed in ¹¹). In contrast to NHEJ, an error-prone system that fuses together blunt ends of DNA without the use of a repair template, HDR is more accurate by involving the recombination of a homologous template strand. However, HDR is a less efficient method of DSB repair compared to NHEJ. CRISPR/Cas9 was first described in 2012 and then adapted for its use in mammalian cells ^12–14. Since this discovery, CRISPR/Cas9 has been making headways in genomic engineering, providing useful tools in the laboratory setting and also in its development for therapeutic genetic correction.

DNA gene editing using CRISPR/Cas9

In bacteria, the CRISPR/Cas9 system is an endonuclease important for cleaving viral DNA. The adaptation of this endonuclease system ultimately relies on guide RNAs to direct the endonuclease, Cas9, to specific sequence sites that are then cleaved. Single guide RNA sequences (sgRNAs) are short synthetic RNA molecules that direct the Cas9 endonuclease protein to generate DSBs near the site of interest (Figure 1). The precision of this cleavage depends on the accuracy of the sgRNAs. The region of target homology in sgRNAs is ~20 bp, and even single base pair (bp) mismatch reduces target engagement. It is possible to utilize more than one sgRNA, for example to create more than one cleavage. However, a drawback of using more than one sgRNA is decreased efficiency, since two guides must find their cognate target, and an increased number of potential off-target effects may result ¹⁵. An additional limitation of this technology is the long term expression of Cas9 nuclease that can increase toxicity through multiple mechanisms including off-target mutations. Modifications to guide RNAs or using Cas9 nickases, which generate single-strand DNA breaks instead of double-strand DNA breaks, can reduce off-target mutations ¹⁶. Furthermore, destabilizing domains can be added to Cas9 to provide an additional level of regulation, whereby the Cas9 can be inactivated to reduce off-target effects by limiting the lifespan of Cas9 activity ¹⁷.

Figure 1.

CRISPR/Cas9 for gene editing. CRISPR/Cas9 is a bacterially derived system that can be applied to specifically modify the mammalian genome. A guide RNA (green) is used to carry the Cas9 nuclease to a region of homology, dictated by an approximately 20 base pair region within the guide that recognizes the complementary sequences in the target DNA. Cas9 then creates a double stranded break which is then ligated using NHEJ, generating insertions or deletions of varying sizes. Alternatively, in the presence of a template, HDR can be used to generate site specific nucleotide changes in the target DNA.

The small size of Cas9 contributes to the scalable nature of its application, making gene delivery by viral vectors possible. Cas9 endonucleases derive from bacterial species and differ in size and functionality ¹⁸. Staphylococcus aureus Cas9 (called SaCas9) is 1 kb smaller than Cas9 from Streptococcus pyrogenes Cas9 ¹⁹, providing a more compact Cas9. Additionally, the longer CRISPR RNA spacer sequence of Neisseria meningitidis Cas9 (referred to as NmCas9) limits its targeting range, but decreases off-target activity ²⁰. Streotococcus thermophiles (StCas9) has proven to be a more specific form of Cas9, albeit less efficient than SpCas9 ²¹. In tandem with the findings of new Cas9 species, ongoing studies are exploring delivery channels as a means for better transportation of these molecules intracellularly ²².

Application of gene editing in cardiomyopathy

Applications of gene editing for treating heart failure are multifold. One use of gene-editing is aimed at correcting underlying genetic mutations responsible for causing cardiomyopathy. Cardiomyopathies are often attributed to genetic mutations resulting in familial or inherited forms of cardiomyopathy. Hypertrophic cardiomyopathy (HCM) is often linked to sarcomere gene mutations ²³. Dilated Cardiomyopathy (DCM) is far more heterogeneous with mutations in more than 100 genes ²⁴. Arrhythmogenic right ventricular cardiomyopathy (ARVC or AVC), restrictive cardiomyopathy (RCM), and left ventricular noncompaction cardiomyopathy (LVNC) each have a significant genetic component as well ²⁵. For each of these genetic cardiomyopathies, autosomal dominant mutations are the most frequent mode of inheritance, although X-linked recessive, autosomal recessive, and mitochondrial inheritance each contribute. The most common genetic mutations linked to cardiomyopathies are small mutations, either single nucleotide variants (SNVs) or small insertions/deletions (indels), on the order of 1–30 bp. SNVs can result in the substitution of one amino acid for another, referred to as nonsynonymous SNVs (nsSNV) or result in the loss of a stop codon or insertion of a new premature stop codon (stop loss or stop gain SNVs, respectively).

Indels can produce in-frame or out-of-frame effects on the protein they encode. In-frame deletions produce internally truncated, but often functional proteins. Out-of-frame deletions may or may not be associated with residual protein expression depending on the degree of nonsense mediated decay, which reduces the amount of mRNA, and the stability of the amino-terminal protein domains. Similarly, premature stop codons or loss of stop codons each can result in varying degrees of residual protein expression. Predicting the effect of premature stop codons and out of frame mutations on residual protein expression is not straightforward and highly dependent on the protein’s normal structure and function. Even small amounts of amino-terminal protein expression can result in gain-of-function, dominant negative activity. As gene correction strategies move forward to treat genetic cardiomyopathies, it is critically important to determine the precise mode of action of a given mutation, as off-target effects of gene editing could render the molecular pathology worse. For example, genetic editing that shifts an in-frame mutation to an out-of-frame mutation could result in dominant negative activity, thus making the outcome worse. At present, most gene editing results in multiple events that differ from cell to cell. Although HDR may be intended to substitute one amino acid for another, many cells may actually have undergone NHEJ, resulting in a range of results within the heart.

In addition to ensuring high fidelity of genetic correction of the mutant gene, it is important that there is specificity for the mutant gene to avoid introducing unwanted mutations into the normal gene copy. The application of gene editing to an autosomal dominant disease like most cardiomyopathies mutations would require correcting the mutated copy and at the same time, leaving the normal copy intact. Because cardiomyopathy-causing mutations often span only 1–2 bp, this leaves little sequence-based differences between the mutant and normal copy, necessitating tight specificity in applying corrective technologies. Correcting these types of mutations is likely to require HDR over NHEJ. With the comparatively lower efficiency of HDR compared to NHEJ, this increases the challenge for gene editing.

NHEJ-based gene editing for Duchenne Muscular Dystrophy

Given the higher efficiency of NHEJ, there have been more efforts aimed at exploiting this approach where small deletions may be useful to provide partial correction. For example, for Duchenne Muscular Dystrophy (DMD), an X linked recessive disorder that affects both heart and skeletal muscle, gene editing generates additional deletions that restore the reading frame-resulting in internal truncated but functional proteins. Dystrophin, the protein product of the DMD gene, is a large internally repetitive protein composed of 24 spectrin repeats throughout its mid-section ^26,27. The DMD gene itself spans 2.5M bp and includes 79 coding exons. Deletions that disrupt the reading frame ablate expression of the dystrophin protein, leading to DMD. Approximately 80% of DMD mutations are large mutations that span large intervals of thousands of base pairs, and the majority of these are deletions ²⁸. In contrast, in-frame deletions typically lead to dystrophin protein production, which can be detected by immunoblotting or immunofluorescence detection using anti-dystrophin antibodies. In-frame deletions affect both the stability and function of dystrophin. For example, the middle portion of dystrophin has been implicated in binding nitric oxide synthase, which is important for regulating vascular tone and blood flow to muscle ²⁹. Nonetheless, restoring dystrophin’s reading frame converts DMD to the milder Becker Muscular Dystrophy (BMD). However, BMD has an associated cardiomyopathy, and a detailed analysis of BMD-associated mutations found that onset of cardiomyopathy was earliest for mutations disrupting the amino-terminal actin binding domain. BMD patients with mutations in spectrin repeats 17–19 had an intermediate age of cardiomyopathy onset, while those with mutations disrupting the 3^rd hinge and spectrin repeats 20 and 21 had the latest onset of cardiomyopathy ³⁰. These findings suggest that different domains may be more critical for some aspects of cardiac function compared to domains needed to rescue skeletal muscle function.

A detailed understanding of dystrophin’s domain function is important since CRISPR/Cas9 can be deployed to create new dystrophin gene deletions with the goal of restoring the reading frame. Approximately 13% of DMD patients have mutations in exons 48–50, resulting in loss of dystrophin ³¹. Gene editing is designed to target exon 51 by generating deletions that disrupt the splicing inclusion sequences in exon 51, which then restores the reading frame and converts a DMD mutation into a BMD mutation (Figure 2). This rationale also underlies the antisense oligonucleotide (AON) drug eteplirsen (see below, ³²). AON-based therapies require repeated weekly intravenous dosing for DMD. Gene editing, if successful, in principle, should require only a one-time application ³³.

Figure 2.

CRISPR/Cas9 strategy for restoring the reading frame due to deletion mutations in the dystrophin gene. Deletions in the dystrophin gene disrupt the reading frame in DMD, and deletions of exons 47, 48, 49 and 50 affect 13% of DMD patients. Dystrophin is a large protein with an actin-binding domain at its amino terminus. There are four hinge (H) domains and 24 spectrin repeats that comprise to the middle portion of the protein. The cysteine-rich (cys rich) region is needed for dystrophin function. The carboxy-terminal (COOH term) domain is dispensable. Guide RNAs directed towards splicing regulatory sequences in exon 51 (marked with an X) can be used to exclude this exon in the mRNA, which then results in exon 46 joining to exon 52. This restores the dystrophin reading frame, creating an internally truncated, but functional dystrophin protein. Similar strategies can be used to restore the reading frame in other regions of dystrophin.

Preclinical studies in mice have documented the success of systemic gene editing to restore dystrophin expression in the mdx model of DMD ^34–38. Virally-mediated systemic delivery of Cas9 and guide RNAs also corrected the dystrophin gene in the heart, resulting in dystrophin protein expression ^34,36,38,39. Several features make systemic in vivo genetic correction of DMD in humans appealing. First, DMD as a severe disease has a greater risk-benefit ratio that milder disorders. Second, the X-linked recessive inheritance of DMD necessitates the correction of only one X-chromosome. Lastly, NHEJ, rather than HDR, increases the feasibility of success. The drawbacks of applying CRISPR/Cas9 in human DMD are the limited data on safety of this approach in humans and the need to conduct this using widespread viral delivery in children. Gene editing in DMD, by design, will also result in cardiac gene editing, and therefore may become the first use of gene editing in the human heart.

Antisense oligonucleotide (AON) based genetic correction for DMD

AON-based therapy is an alternative genetic corrective strategy that aims to manipulate RNA, rather than DNA. Eteplirsen is an AON drug that targets exon 51 of the DMD gene. As an AON, eteplirsen contains complementary sequences that hybridize to the RNA prior to splicing, and promotes exon exclusion from the mature mRNA ³². This approach is conceptually similar to gene editing for its end effect, but does so by engaging RNA. In addition to Food and Drug Administration approval of eteplirsen, the AON drug nusinersen was recently approved for the treatment of the autosomal recessive spinal muscular atrophy, a neuromuscular disorder that affects young children ^40,41. AONs rely on chemical modifications of oligonucleotides, and these chemical modifications prolong drug half-life by avoiding endogenous nucleases that cleave double stranded nucleic acid moieties ⁴².

AON application for other disorders

Several other disorders with systemic features, including neuromuscular disease, have also been identified for application of AON therapy including limb girdle muscular dystrophy type 2C (LGMD 2C) ⁴³. LGMD 2C is an autosomal recessive disorder caused by loss of function mutations in SCGC, which encodes γ-sarcoglycan, a dystrophin-associated protein. LGMD 2C has similar findings to DMD, including progressive skeletal muscle wasting and weakness along with cardiomyopathy ⁴⁴. In this case, AON- directed therapy is designed to promote retention of exons 4, 5, 6 and 7 from this eight exon gene, which would remove nearly half the final protein product. This product, termed “mini-gamma”, was shown to be partially protective in animal models of LGMD 2C.

Mutations in the LMNA gene lead to a variety of disorders including cardiomyopathy. One disorder, Hutchinson Gulford Progeria, arises from a mutation that alters splicing and promotes the production of prelamin A, which is thought to underlie the toxicity related to this premature aging disorder. An AON based approach has been tested in human cells and has demonstrated the capacity to alter production of prelamin A, which should be therapeutic for this disease ^45,46. Other approaches manipulating LMNA splicing using ASOs have also been developed ⁴⁷.

Autosomal dominant MYBPC3 mutations lead to HCM ⁴⁸. Using a mouse model of Mybpc mutations, it was shown that AON-induced exon skipping could produce an alternative transcript that restored protein expression and cardiac function ⁴⁹. Moreover, additional studies of MYBPC3 suggests that trans-splicing events can happen, where splicing between two independent RNAs takes place ⁵⁰. In this case, splicing between the mutant and normal alleles could be used to bypass a specific mutation. However, at present the efficiency of trans-splicing is very low and would need to be significantly augmented to have therapeutic benefit. These studies and others underlie the complexity of RNA splicing, which may be even more complicated in cardiomyopathy, as heart failure itself leads to alternative splicing patterns. For example, TTN truncating mutations are a common cause of DCM, and exon skipping has been tested for 3′ end TTN mutations ⁵¹. In theory, it may be possible to use AON-directed strategies to re-frame more TTN truncations, but the sheer size and complexity of TTN makes this task daunting ⁵².

RNA-based gene editing for microsatellite repeat expansion disorders

Recently, CRISPR/Cas9 has been directed against microsatellite repeat expansion disorders including myotonic dystrophy type 1 and 2 ⁵³. Microsatellites are small units of repetitive DNA sequences. Expansion of these microsatellites repeats, typically on a single allele, is sufficient to cause disease; both myotonic dystrophy type 1 and 2 are associated with arrhythmias and cardiomyopathy ^54,55. In myotonic dystrophy type 1, the CTG trinucleotide in the DMPK gene on one allele expands to greater than 70 copies. Myotonic dystrophy type 2 arises from expansion of the CCTG repeat within the CNBP/ZNF9 gene and in type 2 the repeat expansion is many thousands of copies. These repeat expansions themselves have been the target of ASO-based treatment, but CRISPR/Cas9 has been engineered to recognize these repeats when expressed as RNA ⁵⁶.

Mutation-independent genetic correction for heart failure

In each of the above discussed strategies, genetic correction was directed at mutations directly responsible for disease. In applying this to human cardiomyopathy, many different, site-specific corrective strategies must be designed and tested. This feat is challenging from a clinical trial and regulatory perspective. Each AON or guide RNA would be useful for a very small number of individuals, which makes placebo-controlled trials, the regulatory standard, near impossible. This has prompted some to evaluate the possibility for broader genetic corrective avenues that would target normal genes in order to boost cardiac function. For example, gene therapy to upregulate SERCA was tested in heart failure with reduced ejection fraction ^57,58. Reducing phospholamban, even partially, may upregulate SERCA activity and thereby modulate Ca²⁺ mishandling that underlies heart failure ⁵⁹. It may be possible to use AON or gene editing to upregulate specific splice forms of sarcomere proteins that increase actomyosin interactions, and in effect make the heart more energetically efficient.

Ethical considerations of gene editing

Gene editing is a powerful technique, and its application to the human genome has raised ethical questions owing to the capacity to alter the human germ line and therefore future generations ^60,61. Somatic correction of mutations or manipulation of a normal gene with the goal of treatment is distinct from germline correction (Figure 3). Most recently, the capacity for germline correction was demonstrated when a fertilized oocyte carrying an MYBPC3 allele was corrected using HDR gene editing ⁶². Remarkably, the authors found the mutation correction rate was reasonably efficient in fertilized eggs, including template and nontemplate-directed HDR. The corrected zygotes were not implanted. However, it should be emphasized that the first step to collect and identify eggs relied on preimplantation genetic diagnosis (PGD) to define which zygotes carried the MYBPC3 mutation and which did not. As an autosomal dominant disorder, 50% of the zygotes were mutation free simply by inheriting the nonmutant allele. These mutation-free zygotes can be implanted, as is routinely accomplished with PGD, and PGD does not require gene editing, thus avoiding off-target mutations ⁶³.

Figure 3.

Strategies to conduct gene editing in humans for therapeutic purposes. Somatic gene editing for heart failure is expected to use adeno-associated virus (AAV) to deliver Cas9 and guide RNAs to the heart. This can be used for either systemic or intracardiac delivery. Germline gene editing delivers Cas9 protein and guide RNAs to a zygote to correct genetic mutations that cause disease. Cas9 can also be delivered using RNA encoding Cas9 directly into fertilized eggs.

HDR in non-dividing cardiomyocytes

The degree to which cardiomyocytes can complete HDR has been debated. In principle, HDR is thought to require a round of cellular replication. Since nearly all cardiomyocytes are post-mitotic, this would suggest that HDR could only be used to address dividing cells in the heart. However, gene editing in skeletal myofibers, which are terminally differentiated, was achieved using HDR ³⁸. Notably these studies used a Cas9 expressed under the control of a modified muscle creatine kinase promoter active only in mature myofibers and inactive in muscle stem cells. More recently, several studies have shown HDR in postnatal mouse hearts, indicating that HDR may not require cell replication ^64,65 and suggesting that genetic correction of site specific mutations may be more feasible than predicted.

Delivering gene editing machinery to the heart

Viral vectors are the primary delivery modality for Cas9. Specifically, adeno-associated virus (AAV) is a preferred virus since specific strains display high tropism for cardiomyocytes ⁶⁶. The major limitation of AAV is its small cargo capacity. As noted above, certain Cas9s fit more readily within the carrying capacity of AAV. However, even SaCas9 has reached its limit with AAV, necessitating second viruses to carry the guide RNAs ⁶⁷. With viral delivery, it may be necessary to terminate Cas9 activity either through guides that target the Cas9 gene or through drug regulatable domains that target Cas9 for proteolytically digestion ¹⁷. Immunity is known to develop against AAV capsids and Cas9 itself may be immunogenic. While this characteristic could help limit its activity, it would also limit redelivery of Cas9.

Fidelity of gene editing

Cas9 as a nuclease can produce unwanted off-target and on-target mutations. Off-target mutations are created at sites remote from the intended target. Estimating the frequency of off-target mutations for somatic gene correction requires detection of low frequency events. Where gene editing is carried out on single cells, that subsequently produce clonal expansion of the corrected cell, it is possible to conduct whole genome or whole exome sequencing to assess off-target mutations. However, in applying gene editing to the heart, the off-target mutations may differ from cell to cell and would not be detected by even whole genome sequencing. Estimating off-target mutations in zygotes has been done in genetically-engineered mice, but this too is complicated by the rate of new mutations that occurs in every zygote ⁶⁸.

In addition to off-target mutations, NHEJ produces a range of deletions, rather than single events. Deep sequencing across these sites can be used to estimate the fidelity of on-target events. From studies in the mdx mouse, it appears that only a small population of properly corrected cells can produce a greater percentage of corrected mRNA species ^38,39,69. Ultimately, a relatively small percentage of corrected cells produces a higher than expected amount of dystrophin protein. This may relate to the selective advantage of corrected cells over uncorrected cells, as uncorrected cells may be more prone to injury and necrosis.

Future considerations

The advent of gene editing has already made a significant impact in the laboratory setting, where it is now routinely used to make highly useful models of disease or to demonstrate the feasibility of gene correction. Transitioning this laboratory-based tool to in vivo human gene editing for the treatment of heart failure is on the horizon, but will require showing that this process is both safe and efficacious. The cardiac tropism of AAV as a delivery vehicle, and the ability to use catheter-based approaches to introduce AAVs carrying Cas9 and guide RNAs into the heart makes it an attractive target for gene editing.

KEY POINTS.

• Genetic editing targets DNA to correct or change DNA sequences, which can correct underlying genetic mutations.

• Delivering gene editing machinery to the heart uses viral vectors

• RNA directed correction approaches use antisense to modulate splicing, and this approach requires repeated dosing.