Cardiac Gene Therapy with Adeno-Associated Virus-Based Vectors

Abstract

Purpose of Review

Cardiac gene therapy with adeno-associated virus (AAV)-based vectors is emerging as an entirely new platform to treat, or even cure, so far intractable cardiac disorders. This review describes our current knowledge of cardiac AAV gene therapy with a particular focus on the biggest obstacle for the successful translation of cardiac AAV gene therapy into the clinic, namely the efficient delivery of the therapeutic gene to the myocardium.

Recent Findings

We summarize the significant recent progress that has been made in treating heart failure in preclinically relevant animal models with AAV gene therapy and the recent results of clinical trials with cardiac AAV gene therapy for the treatment of heart failure. We also discuss the benefits and shortcomings of the currently available delivery methods of AAV to the heart. Finally, we describe the current state of identifying novel AAV variants that have enhanced tropism for human cardiomyocytes and that show increased resistance to pre-existing neutralizing antibodies.

Summary

Here we describe the successes and challenges in cardiac AAV gene therapy, a treatment modality that has the potential to transform current treatment approaches for cardiac diseases.

Introduction

Cardiac gene therapy with adeno-associated virus (AAV)-based vectors holds great promise for the treatment of both inherited cardiac disorders and acquired cardiac diseases, such as heart failure (HF). At present, for all the inherited diseases and HF the only curative treatment is heart transplant, and for most cardiac illnesses the currently available symptomatic treatment modalities are inadequate. In contrast, cardiac gene therapy with AAVs has the potential to treat the underlying cause of cardiac diseases and can — at least in principle — be curative, especially for inherited disorders. Here, we discuss the current progress and challenges in cardiac AAV gene therapy with a particular focus on the major obstacle to successful cardiac AAV gene therapy, namely the efficient delivery of the therapeutic transgene to the cardiomyocytes.

Basic AAV Biology

Adeno-associated viruses are small, non-enveloped viruses of the parvoviridae family and are generally considered to be non-pathogenic, a highly controversial publication [1] describing a potential link between hepatocellular carcinomas and wild-type AAV notwithstanding.

This lack of pathogenicity might be, at least in part, attributable to the fact that AAV is a defective virus that is unable to replicate on its own. Instead, AAV depends on co-infection with a helpervirus such as adenovirus or herpesvirus [2].

The genome of AAV is a linear, single-stranded DNA, of ~4.7 kb in length (flanked by two inverted terminal repeats (ITRs)) that is packaged into an icosahedral capsid composed of 60 capsid proteins. The genome contains only two genes, Rep and Cap. The Rep gene encodes three proteins that are involved in DNA replication and the packaging of the viral DNA [2], whereas the Cap gene encodes three capsid proteins (VP1, VP2 and VP3) with overlapping reading frames. A third viral protein, the so-called assembly-activating protein, is translated from an alternative reading frame within the Cap gene [2]. Most importantly for AAV gene therapy, the only cis-elements required to produce recombinant AAV vectors (rAAVs) are the ITRs.

What makes rAAVs particularly attractive for cardiac gene therapy is the long-term persistence of the viral genome in an extrachromosomal form, which leads to the durable expression of the therapeutic protein (at least in non-dividing cells such as cardiomyocytes) [3,4,5].

Inherited Cardiomyopathies

Although a clear classification of inherited cardiomyopathies (CMs) can be challenging at times because of significant phenotypic overlap, inherited CMs are typically classified into five groups: hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), arrythmogenic right ventricular cardiomyopathy (ARVC), restrictive cardiomyopathy (RCM) and left ventricular non-compaction cardiomyopathy (LVNC) [6,7,8,9]. Inherited cardiomyopathies are fairly common; the prevalence of HCM alone is estimated be approximately 1 in 500 [10], and inherited cardiomyopathies are likely underdiagnosed. The etiology of inherited cardiomyopathies, including the more common HCM, DCM and ARVC, is complex with varying penetrance, at times involving multiple mutations and a high prevalence of private mutations [7,8,9]. Nonetheless, inherited cardiomyopathies offer a number of attractive targets for AAV gene therapy. Some of the most prominent targets are listed in Table 1.

Table 1.

Common Genes Associated with Inherited Cardiomyopathies

Gene Symbol	Protein	Function	Type of CM	cDNA Size	OMIM
TTN	Titin	Scaffolding Protein	DCM HCM	~82 kb	188840
LMNA	Lamin-A/C	Structural Protein of the Nuclear Lamina	DCM AVC	~1.7-2kb kb	150330
MYH7	Myosin 7 Heavy Chain	Muscle Contraction	DCM HCM	~5.8 kb	160760
MYH6	Myosin 6 Heavy Chain	Muscle Contraction	DCM HCM	~5.8 kb	160710
SCN5a	Sodium Channel Protein Type 5, α Subunit	Sodium Transport	DCM	~6 kb	600163
MYBPC3	Cardiac-Type Myosin-Binding Protein C	Muscle Contraction	DCM HCM	~3.8 kb	600958
TNNT2	Cardiac Muscle Troponin T	Muscle Contraction	DCM HCM	~0.9 kb	191045
RBM20	RNA-Binding Protein 20	Regulation of Splicing of Several Cardiac Genes	DCM	~3.7 kb	613171
TNNI3	Cardiac Troponin I	Muscle Contraction	DCM HCM	~0.6 kb	191044
MYL2	Regulatory Light Chain of Cardiac Myosin β	Muscle Contraction	HCM	~0.5 kb	160781
MYL3	Myosin Light Chain, Ventricular Isoform	Muscle Contraction	HCM	~0.6 kb	160790
PKP2	Plakophilin 2	Cardiomyocyte Cohesion	AVC	~2.5 kb	602861
DSP	Desmoplakin	Cardiomyocyte Cohesion	AVC	~8.6 kb	125647
DSG2	Desmoglein 2	Cardiomyocyte Cohesion	AVC DCM	~3.4 kb	125671
DSC2	Desmocolin 2	Cardiomyocyte Cohesion	AVC	~2.5 kb	125645
JUP	Junction Plakoglobin	Cardiomyocyte Cohesion	AVC	~2.2 kb	173325

DCM: Dilated Cardiomyopathy, HCM: Hypertrophic Cardiomyopathy, AVC: Arrhythmogenic Ventricular Cardiomyopathy. For recent reviews please see [2, 3, 47]

For autosomal recessive CMs, or whenever the mutations result in haploinsufficiency, overexpression of the target protein is expected to be sufficient to achieve therapeutic efficacy. Unfortunately, for approximately one third of the genes listed in Table 1 the size of the respective cDNA exceeds the packaging capacity of AAV, which is ~5 kb (including ITRs and regulatory elements; see [11]). Consequently, only the smaller CM proteins can be expressed from a single AAV vector. However, with the exception of titin, the overexpression of the other proteins should be feasible with a two AAV vector strategy (reviewed in [11]). A promising approach to correct mutations in titin is likely gene editing with the CRISPR/CAS9 system [12]. This method has, in fact, been used with some success in a murine model of Duchenne’s muscular distrophy [13*,14*]. It is also noteworthy that allele-specific knock-down of a dominant-negative MYH6 variant with an AAV9 vector resulted in the amelioration of hypertrophic cardiomyopathy in a mouse model [15*]. However, because most mutations causing CMs are not the result of the expression of a dominant negative protein, this approach is likely only useful for a small subset of all inherited CMs.

Overall, the treatment of inherited cardiomyopathies with AAV gene therapy is conceptually simple and doubtlessly promising. However, absent AAV vectors that efficiently transduce human cardiomyocytes and the availability of suitable large animal models, the translation into the clinic could prove to be challenging.

Heart Failure Targets

HF is one of the most common causes of morbidity and death in the Western world [16]. Even with the best available treatments, upon diagnosis of HF, the 5-year survival rate is only about 50% [16]. However, with our increasing understanding of the biological processes underlying the progression of HF, gene therapy has emerged as a promising tool to reverse specific molecular changes of this devastating disease. Because the potential targets for the treatment of HF have been reviewed extensively before, we refer the reader to these publications [17,18,19] and provide only a table with promising targets for AAV gene therapy that have been validated in preclinically relevant animal models (Table 2).

Table 2.

Potential Therapeutic Proteins to Treat Heart Failure that Have been Tested in Large Animal Models

Therapeutic Protein	Animal Model	AAV serotype/variant	Pathway	Ref.
SERCA2a	Pig	AAV1	Calcium Handling	[20, 21**, 22]
SERCA2a	Sheep	AAV6	Calcium Handling	[23]
SERCA2a	Sheep	AAV1	Calcium Handling	[24, 25]
SERCA2a	Sheep	AAV1	Calcium Handling	[26*, 27]
SERCA2a	Sheep	AAV9	Calcium Handling	[28]
SERCA2a	Dogs	AAV6	Calcium Handling	[29]
SUMO1	Pigs	AAV1	Calcium Handling	[22]
S100A1	Pigs	AAV9	Calcium Handling	[30*]
S100A1	Pigs	AAV6	Calcium Handling	[31]
I1c	Pigs	AAV9	Calcium Handling	[32]
I1c	Pigs	AAV2i8 (BNP116)	Calcium Handling	[33]
shRNA-PLB	Dogs	AAV6	Calcium Handling	[34]
SERCA2a	Dogs	AAV1	Calcium Handling	[35]
βARKct	Pigs	AAV6	Beta-Adrenergic Pathway	[36]
VEGF/Ang1	Pigs	AAV1	Angiogenesis	[37]
VEGF-A/PDGF-B	Pigs	AAV9	Angiogenesis	[38]
VEGF-B	Dogs	AAV9	Angiogenesis	[39]
Heme Oxygenase-1	Pigs	AAV9	Inflammation	[40]

As can be seen from Table 2, in preclinically relevant, large animal models the vast majority of cardiac AAV gene therapeutic approaches to treat HF have focused on improving calcium handling, which is disturbed in patients with HF. Among the targets the cardiac specific isoform of the sarcoplasmatic calcium ATPase, SERCA2a, has been the therapeutic protein that has been studied most intensely.

Clinical Trials

Based on the extensive studies in large animal models, it is not surprising that the delivery of SERCA2a with an AAV vector to treat patients with advanced HF lead to the first-in-man study of cardiac AAV gene therapy [41,42*]. Phase 1 of the CUPID (calcium upregulation by percutaneous administration of gene therapy in cardiac disease, NCT02346422) was an open label dose-escalation study that enrolled a total of 12 patients [43]. No treatment related adverse events were observed in this study. Phase 2a was a double blind randomized placebo controlled study in 39 patients with three different doses of AAV1 vectors carrying SERCA2a. In this study, the rate of clinical events was significantly lower at the end of the study at 3 years [42*]. Based on these promising results, a double-blind, placebo-controlled, multicenter, multinational trial to treat patients with advanced HF was designed (CUPID2b trial, NCT01643330, [44]). In this study, patients with advanced HF received either placebo or 10¹³ vector genomes (vg) of AAV1.SERCA2a. Unfortunately, despite the promising results in phase 1/2a, neither the primary endpoint, time to recurrent cardiac event (HF-related hospitalization or ambulatory treatment for worsening HF), nor the secondary efficacy endpoint, time to first terminal event (all-cause death, heart transplant or implantation of a left ventricular assist device) showed an improved outcome in treated vs. placebo patients [45*]. Based on these disappointing results, patient recruitment for two additional trials using AAV1. SERCA2a to treat HF, the AGENT-HF trial (NCT01966887) and the SERCA-LVAD trial (NCT00534703), have been halted. According to the reported values of vg in the heart of patients from whom samples could be obtained, the most likely reason for the negative results of the CUPID trial was a failure of AAV1 to deliver efficiently SERCA2a to cardiomyocytes [45*]. In fact, the numbers reported suggest that <1% of all cardiomyocytes contained a vg [45*]. These results are in striking contrast to the results in large animal models (Table 2) and point to potential species differences in AAV1 tropism or deficiency in dosing.

Currently, two trials using a different AAV capsid are being designed. One trial aims to deliver S100A1 with an AAV9 vector, whereas the second trial plans to deliver a constitutively active form of the protein phosphatase 1 inhibitors, Il-c, with a chimeric capsid between AAV2 and AAV8 [33,46].

AAV Tropism

Most commonly, recombinant AAVs have capsids that are composed of capsid proteins of the serotypes 1 to 9. However, additional serotypes and isolates — as well as “designer variants” [46] — play an increasing role in AAV gene therapy. Collectively, these AAVs have a very broad but distinct tropism. It must be pointed out, however, that despite their distinct tropism none of the currently available AAVs triggers the exclusive expression of the transgene in just one tissue or cell-type.

Nonetheless, for the purpose of cardiac gene transfer three AAV serotypes — AAV1, AAV6 and AAV9 — have emerged as the most promising AAVs. In particular, AAV9 has proven to be the most powerful AAV serotype to transduce efficiently cardiomyocytes in mice and rats when injected systemically [47].

Unfortunately, our knowledge regarding the cardiac tropism of AAVs in large animals, let alone in humans, is very limited. Moreover, the variety of vector delivery methods used in different studies renders the comparison of the efficiency of cardiac transduction between serotypes difficult, at the best. Further complicating matters is the fact that the numerous research groups studying cardiac gene therapy in clinically relevant large animal studies use different animals; dogs, sheep and pigs being the most commonly used species. Clearly, much work remains to be done to determine what specific roles the AAV serotype/variant, the vector delivery approach and the species play in the tissue distinct expression pattern of AAVs.

Neutralizing Antibodies

AAVs are naturally occurring viruses; hence, it is not surprising that humans can harbor pre-existing neutralizing antibodies against AAVs (reviewed in [48]). Because neutralizing antibodies will prevent efficient transduction, the presence of antecedent neutralizing antibodies is a major exclusion criterion in all AAV gene therapy trials. In the only cardiac gene therapy trial, the CUPID trial (NCT02346422, NCT01643330), the presence of neutralizing antibodies varied according to country and among states in the US [49], ranging from a prevalence of 42% in Ohio to a staggering 79% in Poland and Hungary [49].

That non-human primates harbor neutralizing antibodies against several AAV serotypes is unsurprising as several AAV serotypes have been isolated from rhesus macaques [50**]. But, astonishingly, neutralizing antibodies against the common AAV serotypes are also present in many small and large animal species [51,52*,53]. Presumably because of the comparatively high vector doses relative to their weight, pre-existing neutralizing antibodies appear not to be an issue in mice. In larger animals, on the other hand, pre-screening of the animals for the presence of antecedent neutralizing antibodies is critical.

AAV Vector Delivery Methods for Cardiac Gene Therapy

Delivery of AAV vectors to the myocardium can be broadly divided into two approaches 1) direct intramyocardial injection and 2) transvascular administration.

Intramyocardial Injection

Direct intramyocardial injection has several advantages: 1) the virus can be delivered at a very high local concentration, 2) direct injection bypasses the endothelial barrier, which is a formidable hurdle for efficient gene transfer, 3) off-target organ transduction is minimized, though not eliminated and 4) the neutralizing effect of pre-existing antibodies is stunted. But, for global cardiac disorders, such as HF and genetically inherited cardiomyopathies, intramyocardial injection is clearly a suboptimal delivery method as the expression of the transgene is restricted to a very small area surrounding the injection site. Nonetheless, for certain applications, for instance for the stimulation of angiogenesis following a myocardial infarction [54] or for the delivery of biological beta-blockers or pace makers [55,56], direct intramyocardial injection might be a desirable approach.

Intramyocardial injection can be done via direct injection after thoracotomy, which offers the greatest flexibility and precision of injection; but this is an invasive technique that is not clinically desirable. Percutaneous, catheter based injections, on the other hand, are minimally invasive and as such are more desirable from a safety point of view [57].

Transvascular Delivery

Ideally, for most clinical applications, the recombinant AAV vector could be injected peripherally and trigger the near-homogeneous expression of the therapeutic protein. Unfortunately, this is only readily achievable in rodents but not in large animals, let alone in humans. The reasons for the discrepancy between rodents and large animals and humans are unclear. But the development of AAV vectors that could efficiently transduce human cardiomyocytes upon intravenous injection would offer tremendous therapeutic benefits.

Antegrade Intracoronary Injection

Antegrade intracoronary infusion of AAV vectors has been used extensively to treat HF with AAV gene therapy in large animal models [21**,22,32,33], and percutaneous access to the coronaries is routinely performed during angioplasty. Because of its minimally invasive nature, antegrade intracoronary injection is the only delivery method that has been used to deliver a potentially therapeutic AAV vector clinically [41,42*,45*]. However, the published results from the CUPID trial suggest that, at least with AAV1 and a dose of 10¹³ vg, this delivery approach results in inefficient delivery of the vg to cardiomyocytes.

To increase the efficiency of delivering SERCA2a to the myocardium Kaye and colleagues used a percutaneous cardiac recirculation delivery approach that allows pressure control [24,58] to deliver AAV1.SERCA2a in a pacing induced, ovine model of HF [24]. Strikingly, the SERCA2a overexpression in the high-dose (10¹³ vg) recirculation group was ~29-fold higher than in the group that received 2.5x10¹³ vg via intracoronorary infusion [24]. This suggests that, at least in an ovine model, antegrade recirculating delivery of rAAV might lead to superior transduction efficiencies when compared to slow antegrade intracoronary delivery.

Retrograde Injection

In HF, stenotic coronary arteries pose a significant hurdle to the successful delivery of rAAVs via the antegrade route. This barrier to efficient transduction can potentially be overcome by delivery of the vector into the coronary vein. This delivery approach could result in longer dwell times of the vector the coronary vasculature [59] increasing the chance of extravasation of the vector. However, to increase gene transfer a temporary occlusion of the left anterior descending coronary artery, to increase the intravascular pressure, was required [59]. While this increased vasculature pressure enhances gene transfer, it carries significant clinical risks, especially in patients with advanced HF.

Molecular Cardiac Surgery with Recirculating Delivery (MCARD)

In a surgical vector delivery technique developed by Bridges and colleagues the animal is put on cardio-pulmonary bypass. This allows the complete isolation of the cardiac circulation from the systemic blood circulation and the recirculation of vector-containing blood exclusively through the cardiac circulation [26*]. This system has several advantages: 1) it allows the prolonged circulation of the AAV vector containing the therapeutic transgene through the coronary system, 2) the closed nature of the system might allow its use in patients with pre-existing antibodies and 3) off-target transduction is reduced because the system can be flushed from remaining vector before completion of the coronary bypass. Remarkably, 12 weeks after delivering 10¹³ vg of AAV1.SERCA2a with the MCARD (Molecular Cardiac Surgery with Recirculating Delivery) system >400,000 vg per µg host DNA could be detected in the myocardium [26*]. This compares favorably to the delivery of 2.5x10¹³ vg via antegrade intracoronary delivery, which resulted in <2,000 vg per µg genomic DNA [24].

Unfortunately, MCARD is a highly invasive method that is likely only applicable in situations where an on-pump coronary bypass surgery or valve surgery is required, regardless of vector delivery

Isolation of Vectors with Increased Tropism for Cardiomyocytes and Resistance to Neutralizing Antibodies

As has been described above, today no combination of a cardiotropic vector and vector delivery system has been described that allows the efficient transduction of the human myocardium and that is minimally invasive. Hence, intensive efforts by numerous groups are underway to isolate AAV variants with increased tropism for cardiomyocytes and resistance to antecedent neutralizing antibodies. So far, essentially two approaches have been explored: 1) the “rational design” of novel AAV variants and 2) directed evolution to isolate cardiotropic AAV variants from AAV libraries with diverse capsids. For a review of the directed evolution approaches we refer to the review by Kotterman and Schaffer [60].

Using directed evolution, Xiao and colleagues isolated an AAV variant (M41) that, when tail vein injected into mice, transduced cardiomyocytes with an approximately 2-fold lower efficiency when compared to AAV9. Liver transduction, on the other hand, was almost ≥20-fold lower with M41 compared to AAV9 [61*]. But, interestingly, in pigs M41 did not transduce cardiomyocytes efficiently, at least when delivered by antegrade intracoronary infusion (Hajjar RJ, personal communication) pointing to specific differences in AAV tropism. In light of a report by Mark Kay's group showing that AAV8 transduces murine hepatocytes much more efficiently than human hepatocytes [62*], species specific differences in tropism are of significant concern.

Asokan et al. followed a “rational design” approach to isolate AAV variants with cardiac tropism [46]. Specifically, they replaced a hexapeptide in the receptor-binding region of AAV2 with the corresponding hexapeptides of seven other serotypes [46]. When they swapped the hexapeptide of AAV2 with the corresponding peptide of AAV8, the resulting vector (AAV2i8) displayed broad muscle tropism and efficiently transduced the myocardium. Moreover, liver transduction by AAV2i8 was drastically reduced when compared to AAV2 and AAV8 [46]. Using AAV2i8 (a.k.a. BNP116), Hajjar and colleagues demonstrated that delivery of a constitutively active inhibitor of protein phosphatase 1, I1c, ameliorated HF symptoms in a porcine MI model of HF[33]. Based on these promising results a phase 1/2 trial is currently being designed.

Conclusions

Gene therapy with AAVs shows great promise for the treatment of numerous inherited and acquired cardiovascular diseases. However, the efficient delivery of the therapeutic payload to human cardiomyocytes — and other target cells of the cardiovascular system, for this matter — remain a formidable roadblock to realize the great potential of cardiac AAV gene therapy. Both refined delivery methods and, even more importantly, improved AAV vectors will be essential to bring cardiac AAV gene therapy to the clinic.

Key Points.

• Cardiac AAV gene therapy is a novel and promising approach to treat a variety of cardiac disorders for which currently no satisfactory treatment modalities are available.

• The efficient delivery of the therapeutic gene to cardiomyocytes is the main roadblock to the successful implementation of cardiac AAV gene therapy.

• Currently available AAV vector delivery approaches need to be refined further to optimize cardiomyocyte transduction while minimizing the invasiveness of the procedure

• Above all, novel AAV variants that efficiently transduce human cardiomyocytes and display a reduced sensitivity to pre-existing neutralizing antibodies need to be developed.