Viruses are highly infectious simple microorganisms. They have evolved with extraordinary skills to enter our body. It is expected that vectors based on viruses may lead to much more robust gene transfer than plasmid-based nonviral vectors . Several viral vectors have been explored for heart gene delivery. These include lentiviral, adenoviral and adeno-associated virus (AAV) viral vectors (Figure 1).
Lentiviral vectors for heart gene delivery
Lentiviral vectors are originally developed from HIV, a RNA virus . They can package an approximately 8 kb genome. In contrast to traditional retroviral vectors, lentiviral vectors mediate stable gene transfer in terminally differentiated nondividing cells.
Initial studies in cultured cardiomyocytes suggest that lentiviral vectors are extremely powerful. The transduction efficiency reached 70% in adult cardiomyocytes and 100% in neonatal cardimyocytes [22, 23]. Using a heterotopic heart transplantation model, Zhao et al.  observed robust transduction at 7 days after direct injection of a green fluorescence protein (GFP) lentiviral vector. In a separate study, Fleury et al.  showed persistent heart gene transfer up to 10 weeks after direct myocardial injection of the lentiviral GFP vector. Transduction efficiency reached a peak at 3–7 days post-injection but declined thereafter. Additional studies suggested that the drop of gene expression was likely a result of gene transfer-associated myocardial inflammation at the injection site .
More recently, Niwano et al.  tested whether lentivirus-mediated SERCA2a expression could protect against heart failure in a rat myocardial infarction model. At 21 days after hypothermic intracoronary injection (Figure 2), transduction efficiency reached approximately 40% in the heart but was essentially not detectable in the liver and spleen. Importantly, favorable myocardial remodeling was observed. Echocardiography and a cardiac catheter assay showed significant functional improvement. The survival rate was almost doubled in treated animals when challenged with myocardial infarction 6 months later.
Adenoviral vectors for heart gene delivery
Adenovirues are double-stranded DNA viruses. They carry an approximately 30-kb genome. Adenoviral vectors have been shown to effectively transduce several tissues such as the airway epithelium [25, 26]. The use of recombinant adenoviruses for heart gene delivery was initiated in early 1990s (Figure 1). Today, adenoviral vectors have been tested for cardiac gene therapy in rodents, large animals and in humans.
In 1992, Stratford-Perricaudet et al.  showed that intravenous injection of an adenoviral LacZ vector to neonatal mice resulted in widespread gene transfer in several tissues, including the heart and skeletal muscle (Figure 2). Sustained LacZ expression was detected for at least 12 months. However, similar injection protocol resulted in minimal LacZ expression at 3 weeks after gene transfer in adult mice . This result suggests that the mammalian heart is amenable to adenovirus transduction but there are important limitations. Several studies published in the subsequent 2 years confirmed the transient nature of adenoviral mediated heart gene transfer [28, 29]. Guzman et al.  showed that adenoviral vectors were significantly more efficient than plasmids after direct myocardial injection. Robust expression was detected at the injection site during the first week but it diminished to the background level by 30 days. Kass-Eisler et al.  obtained similar results with a different adenoviral vector. These results suggest that adenoviral vectors represent an efficient but unstable gene delivery vehicle for the heart.
To further extend the findings in rodents, French et al.  explored adenoviral gene transfer in the heart of domestic swine. Adenoviral vectors were injected at multiple locations in the ventricular wall. Similar to that reported in rodents, expression peaked at 7 days and declined thereafter. A clear dose response was also noted. Nevertheless, expression did not spread far from the injection site and, furthermore, a marked leukocytic infiltration was observed near transduced cardiomyocytes .
Heart transplantation has become a standard procedure. Ex vivo manipulation of the donor heart by viral gene transfer may further enhance therapeutic outcome. Gojo et al.  tested whether a hypothermically preserved heart graft could be transduced by adenovirus. The vector was perfused through the coronary artery of the donor heart at 4 °C before heterotopic transplantation. Interestingly, the heart infected with low-dose virus (1 × 109 plaque forming units, pfu) showed high transduction for 4 weeks. However, strong inflammation was observed around the transduced cardiomyocytes in the heart that was infected with high-dose adenovirus (≥ 1 × 1010 pfu). Furthermore, minimal expression was detected at 2 weeks after operation in the high-dose group .
Direct myocardial injections only lead to limited transduction at the injection site. To obtain broader myocardial transduction, Fromes et al.  developed an intrapericardial injection strategy (Figure 2). Collagenase and hyaluronidase were injected together with adenovirus to the pericardial sac in rats. This protocol significantly increased adenovirus diffusion into the myocardium. Up to 40% myocardium was efficiently transduced at 7 days post-injection. Nevertheless, expression declined thereafter and became undetectable by 28 days .
To achieve global myocardial gene transfer, Hajjar et al.  developed a catheter-based delivery method (Figure 2). Briefly, a catheter was inserted through the apex of the left ventricle until it reached just above the root of the aortic valve. The aorta and pulmonary artery were briefly blocked using a clamp while adenovirus was injected. This technique allowed efficient circulation of adenovirus through the coronary arteries. Almost homogeneous cardiac gene transfer was achieved with several different adenoviruses. Importantly, the application of this method to an adenovirus carrying the phospholamban gene resulted in significant change of left ventricular function . Although adenoviral transduction of the lung and liver was also noted, the study demonstrated for the first time that a gene transfer technique could be used to modulate overall heart function. Subsequently, a modified catheter-based technique was investigated in rabbits . To perfuse the coronary beds, Maurice et al.  rapidly injected 1.5 ml of adenovirus via a catheter placed in the chamber of the left ventricle. During injection, they transiently clamped the aorta. Diffuse multi-chamber expression was achieved, with peak expression at 6 days post-injection. Using this system, it was also shown that a therapeutic β-adrenergic receptor adenovirus significantly enhanced heart contractility and hemodynamic performance in rabbits .
The first successful demonstration of adenovirus-mediated gene therapy in a cardiomyopathy model was performed by Ikeda et al.  2002 . δ-sarcoglycan deficient hamster is a naturally occurring model of dilated cardiomyopathy. Interestingly, only limited gene transfer (approximately 5%) was achieved when a brief aortic and pulmonary artery occlusion was applied. However, a combination of core temperature reduction, partial cardioplegia and transient aorta/pulmonary artery blocking resulted in an impressive transduction efficiency of approximately 77% . Similar to previous adenovirus studies, inflammation was observed approximately 1 week later and transduction remained short-term. Nevertheless, left ventricular function was significantly improved at 3 weeks after the δ-sarcoglycan adenovirus was delivered.
To achieve persistent adenoviral gene transfer in the heart, Christensen et al.  explored embryonic and neonatal intracardiac delivery. Embryonic adenovirus injection appeared to have minimal impact on pregnancy. Adenovirus injection in 1-day-old mice resulted in stable myocardial transduction that lasted for several months. Importantly, there was no inflammatory reaction in the heart of neonatally injected mice. This technique has provided an excellent platform for studying gene function in murine cardiomyocytes through in vivo gene transfer.
Accumulatively, adenovirus-mediated heart gene transfer has been conducted in a relatively large number of human patients by many research groups [37-41]. A particular focus of these studies is to treat coronary artery disease with an adenovirus that expresses VEGF121. In most trials, 4 × 1010 particle units of adenoviral vectors were delivered to the heart. The procedure was well tolerated [37-41]. However, controversy exists regarding the therapeutic efficacy. Although some studies have demonstrated reduced myocardial ischemia during exercise, as well as increased new blood vessel formation , others have failed to show an improvement in exercise capacity and myocardial perfusion .
Recombinant AAV for heart gene delivery
Basic biology of AAV
One of the greatest challenges in heart gene transfer is to obtain long-term myocardial transduction in adult animals. This apparently impossible task was accomplished with the development of AAV vectors. AAV is a single-stranded DNA virus. It belongs to the dependovirus genera of the parvoviridae family. In the mid 1960s, AAV was discovered as a small contaminating DNA virus in adenovirus preparations (hence, the name adeno-associated virus) . Mature AAV virion is an icosahedrally symmetrical particle approximately 20–25 nm in size. AAV has three overlapping capsid proteins including viral protein (VP) 1, VP2 and VP3. VP3 is the major capsid protein. All three structural proteins are produced from the same open reading frame by alternative splicing and the use of different translation starting codon. AAV is classified into different serotypes based on amino acid sequence of the capsids protein.
Although AAV was initially considered as an integrating virus, recent studies show that both wild-type AAV and recombinant AAV vectors may mainly persist as episomal circular molecules [43, 44]. The interest in developing AAV vector originated from a unique biological property of wild type AAV serotype-2 (AAV-2). Most of our current understanding on AAV biology comes from studying AAV-2. Kotin et al.  found that AAV-2 genome specifically integrated in a defined region in human chromosome 19. Site-specific integration may significantly enhance the safety profile of a gene therapy vector. Because of the potential safety advantage, and also because the transduction biology of AAV-2 is well studied, early development of AAV vectors has mainly focused on AAV-2.
Pre-clinical studies of AAV-mediated cardiac gene delivery in rodents
Application of AAV vector for heart gene transfer was first demonstrated by Svensson et al.  in 1999 (Figure 1). An AAV-2 LacZ vector was delivered to the adult mouse heart either by direct intramyocardial injection or by trans-coronary perfusion before syngeneic transplantation (Figure 2). Minimal expression was seen at the 2-week time point. Unexpectedly, fairly strong expression was noted at 8 weeks after gene transfer. In the case of trans-coronary perfusion, 50% of cardiomyocytes were transduced . Li et al.  further extended this exciting finding in a hamster heterotopic heart transplantation model. Widespread myocardial transduction was obtained with AAV-2 for more than 1 year via ex vivo trans-coronary perfusion .
The success of long-term heart gene transfer with AAV-2 has prompted the testing of this vector in animal models of inherited cardiomyopathy. Hoshijima et al.  applied the same hypothermic-cardioplegia technique that was used by Ikeda et al.  for adenovirus. Although adenovirus showed transient expression , AAV-2 resulted in broad and persistent heart transduction for at least 30 weeks in hamsters . Importantly, expressing a pseudophosphorylated mutant of human phospholamban with AAV-2 improved global cardiac function and suppressed progressive heart failure in the hamster model of dilated cardiomyopathy .
The identification and development of novel AAV capsids is one of the greatest advances in AAV gene transfer technology over recent years . Various strategies have been explored to obtain new AAV serotypes that may exhibit unique transduction profiles for specific applications. Several AAV serotypes have subsequently been tested for heart gene delivery. Yue et al.  showed that AAV-5 could mediate efficient and persistent myocardial gene transfer in the mdx model of Duchenne muscular dystrophy, the most common lethal muscle disease caused by dystrophin deficiency. Importantly, it was found that a highly miniaturized micro-dystrophin gene could restore the critical dystrophin–glycoprotein complex and improved sarcolemmal integrity in the mdx heart .
Novel AAV serotype results in efficient heart gene transfer via peripheral vein injection
Traditionally, heart gene transfer is achieved by direct injection of vectors to the heart (myocardium, cardiac chamber or pericardial sac) or through coronary artery perfusion. These procedures are usually quite invasive and risky. A strategy that allows therapeutic level myocardial transduction via peripheral veins would offer significant advantages (Figure 2). Accordingly, several newly developed AAV serotypes have shown great promise. The first breakthrough was reported by Gregorevic et al. , who injected AAV-6 via tail vein to adult mice and found extensive gene transfer in the heart and all skeletal muscles in the body. With co-administration of VEGF, systemic gene delivery was achieved even at lower AAV doses. Despite uniform myocardial transduction, significant CD4-positive cell infiltration was noted in the heart after AAV LacZ infection. Additional studies suggest that this immune response may relate to LacZ expression rather than AAV-6 capsid. Another AAV serotype AAV-8 was also found to cross the vascular barrier effectively . Wang et al. compared AAV-1, 2, 5, 6, 7 and 8 in mice and hamsters via intraperitoneal and/or intravascular injection . Superior heart and skeletal muscle transduction was achieved with AAV-8. Subsequent studies from several independent laboratories included AAV-9 in the comparison [53-56, 70]. Pacak et al.  and Inagaki et al.  showed that AAV-9 was log-fold more robust than AAV-8 in transducing the mouse heart. Their finding was further substantiated by Zincarelli et al.  and Bish et al. . Collectively, these results clearly established AAV-9 as a cardiac tropic vector superior to all the other serotypes in rodents. However, AAV-9 also displays a strong tropism for other organs such as the liver. This may result in untoward side-effects. To achieve tissue-specific AAV gene transfer in the heart, Yang et al.  constructed a random library by shuffling the capsid genes of AAV-1–9. Subsequent in vivo screening identified AAV-M41 as a heart-specific AAV variant. This new recombinant capsid retained the heart tropism of AAV-9 but showed substantially diminished transduction in skeletal muscle and the liver. A more recent study suggests that random mutagenesis of the original AAV-9 capsid gene may also yield new heart tropic variants with diminished transduction in other tissues .
AAV for heart gene delivery in large animals
Experimentation in large animal models is a bottleneck for translating success in rodents to human patients. Kaspar et al.  tested direct intracoronary delivery of AAV-2 in pigs. Transduction was observed in most pigs for 8 weeks and no inflammatory response was noted. Raake et al.  achieved long-term heart specific expression with pressure-regulated coronary vein retroinfusion using AAV-6. However, AAV genome was also detected in the liver and lung . At the same time, Su et al.  reported that AAV-1 was more efficient than AAV-2 in the pig heart after direct myocardial injection. More recently, Hardri et al.  demonstrated that intracoronary delivery of AAV-1 SERCA2a significantly improved systolic function and coronary blood flow in a swine model of heart failure.
Dog is another commonly used species in large animal studies. Encouraged by robust gene transfer of AAV-9 in the rodent heart, Yue et al.  tested systemic peripheral vein injection of AAV-9 in newborn dogs. Although they observed whole body skeletal muscle transduction, unexpectedly, the heart was minimally transduced. Two groups investigated AAV-6 for canine heart transduction [64, 65]. Gregorevic et al.  showed that direct infusion of AAV-6 into the external jugular vein of a 2-month-old dog (after a transient immune suppression) resulted in frequent cardiomyocyte transduction. Bish et al.  compared AAV-6, 8 and 9 in 2–3-month-old dogs using a percutaneous transendocardial delivery method. This is a relatively non-invasive procedure and a similar protocol has been used in human patients. Consistent with the observations of Yue et al. , AAV-9 also performed poorly in young dog hearts. Indeed, both AAV-8 and AAV-9 were ten-fold less efficient than AAV-6 . Nevertheless, the results reported by Pacak et al.  suggest that AAV-9 may still hold promise for transducing the heart of nonhuman primates.
Dual vector strategies expand AAV capacity for heart gene therapy
A unique hurdle for AAV-mediated gene transfer is the small viral packaging capacity. A single AAV virion cannot handle a vector genome larger than 5 kb . Several dual vector strategies have been developed to increase AAV packaging capacity. These include cis-activation, trans-splicing, overlapping and hybrid methods [67, 68]. The fundamental principle is to split a complete vector genome into two independent virions. Reconstitution is achieved by homologous recombination and/or AAV inverted terminal repeat-mediated viral genome concatamerization. Ghosh et al.  and Bostick et al.  first showed that trans-splicing AAV-9 vectors could reach the same transduction efficiency as a single intact AAV-9 vector in the heart of normal neonatal mice. Subsequent studies from the same group suggest that trans-splicing AAV-9 vectors were equally effective in the heart of adult dystrophic mice. More recently, Odom et al.  reported that overlapping AAV-6 vectors also resulted in robust cardiac transduction in dystrophin-deficient mdx mice.
AAV heart gene therapy yields promising results in early clinical trials
Recent work has greatly advanced AAV-mediated heart gene therapy in human patients [72-74]. Specifically, an AAV-1 SERCA2a vector was delivered to heart failure patients via percutaneous intracoronary delivery. The same vector has been shown to significantly enhance heart function in a swine heart failure model . In an open-label phase 1 study, nine patients received AAV-1 SERCA2a gene therapy at low (1.4 × 1011 DNase resistant particles, DRP), middle (6 × 1011 DRP) and high (3 × 1012 DRP) doses, with three patients in each group . One patient in the medium dose group died for reasons unrelated to gene therapy. No adverse effects were observed in the remaining eight patients. Six patients showed some level of improvement. Two patients (one in the low-dose group and another in the high-dose group) had high pre-existing anti-AAV1 neutralizing antibodies and they did not respond to therapy as expected . Recently, a double-blind placebo-controlled phase 2 trial was conducted . Thirty-nine patients were recruited in the study and AAV1 SERCA2a were delivered at the dose of 6 × 1011 DRP (low-dose), 3 × 1012 DRP (middle dose) and 1 × 1013 DRP (high-dose). Functional studies at 6 months post AAV therapy revealed an improvement in multiple end points (such as 6-min walk test, oxygen consumption, N-terminal pro-B-type natriuretic peptide levels, left ventricular end-systolic volume, New York Heart Association heart failure classification and Minnesota Living with Heart Failure questionnaire). Importantly, no adverse events were observed .