Gene Therapy for Restoring Heart Rhythm

Abstract

Efforts to use gene therapy to create a biological pacemaker as an adjunct or replacement of electronic pacemakers have been ongoing for about 15 years. For the past decade, most of these efforts have focused on the hyperpolarization-activated cyclic nucleotide gated-(HCN) gene family of channels alone or in combination with other genes. The HCN gene family is the molecular correlate of the cardiac pacemaker current, I_f. It is a suitable basis for a biological pacemaker because it generates a depolarizing inward current primarily during diastole and is directly regulated by cyclic adenosine monophosphate (cAMP), thereby incorporating autonomic responsiveness. However, biological pacemakers based either on native HCN channels or on mutated HCN channels designed to optimize biophysical characteristics have failed to attain the desired basal and maximal physiological heart rates in large animals. More recent work has explored dual gene therapy approaches, combining an HCN variant with another gene to reduce outward current, increase an additional inward current, or enhance cAMP synthesis. Several of these dual gene therapy approaches have demonstrated appropriate basal and maximal heart rates with little or no reliance on a backup electronic pacemaker during the period of study. Future research, besides examining the efficacy of other gene combinations, will need to consider the additional issues of safety and persistence of the viral vectors often used to deliver these genes to a specific cardiac region.

Introduction

In the United States, around 200 000 pacemakers are implanted annually. ¹ Of these, 5% result in complications requiring invasive treatment. ² These include lead dislodgement or fracture, pacemaker extrusion and associated skin erosion, pocket infection, pocket bleeding, pneumothorax, and myocardial perforation. Other issues with regard to electronic pacing include (1) inadequate autonomic modulation (eg, lack of immediate heart rate increases when needed with physical activity), (2) suboptimal cardiac output and remodeling, which results from initiation of electrical activation from a nonphysiological site, the right ventricular apex, ³ (3) inappropriateness for some pediatric patients, ⁴ and (4) the need for multiple replacement procedures because of the limited battery life. In the United States, annual costs of pacemaker implantations currently are estimated at more than 2 billion USD, and these costs are expected to increase because of aging of the population and the associated cardiac morbidity. ¹

Solutions to the problems with electronic pacing can in part be provided by the development of “biological pacemakers,” a therapeutic intervention utilizing biological material to create an automatic focus which initiates the heartbeat. Biological pacemakers are expected to have several advantages, including better autonomic responsiveness, improved cardiac output, no need for periodic resizing in a child during maturation, and the possibility of lifelong cure instead of temporary palliation.

Biological pacemakers typically incorporate a gene therapy component, a cell therapy component, or both. Several cell-based approaches have generated pacemaker function in proof-of-concept studies. These approaches range from the use of undifferentiated mesenchymal stem cells (MSCs) as a gene therapy vehicle to introduce transmembrane ion channels ^{5 –7} to methods that coax stem cells into a lineage of spontaneously active cells which can then pace the heart. ^8,9 A general concern with stem cell-based strategies is the difficulty in generating long-term outcomes. ¹⁰ Beyond the need to improve long-term efficacy, there are additional concerns dependent on the cell type employed that relate to cell migration, neoplasia, and rejection. Progress and unresolved issues with cell-based biological pacemakers have been the subject of several recent reviews. ^10,11 Here we primarily focus on gene therapeutic approaches which—while having their own set of challenges—may be closer to implementation.

Viral Vector Systems

Gene therapy uses designated vehicles to deliver genetic information to specific target cells. In the heart, such targeting may be achieved by injecting vectors into specific areas or infusing them into the coronary vasculature. Multiple viral vectors have been considered for gene delivery to the heart, and the key characteristics of each are summarized in Figure 1. Adenoviral vectors are useful candidates because they are easily produced, transduce myocardium efficiently, and have a large insert capacity allowing investigation of a broad range of candidate genes. However, because the gene is not integrated into the host genome and adenoviral gene products trigger an inflammatory response, gene expression generally only persists for weeks. ^12,13 Therefore, these vectors are suitable for proof-of-concept studies but their clinical value is limited. Long-term myocardial gene transfer is typically generated by adenoassociated viral vectors (AAVs). Because all the viral genes have been removed from this vector, they only induce a minimal inflammatory response. Moreover, myocardial gene transfer has been shown to persist for at least 12 months, ¹⁴ and in a recent study as long as 31 months, ¹⁵ and studies outside the heart support the expectation for long duration of transgene expression. ¹⁶ The AAV-based gene transfer has also been found safe in clinical testing of patients with heart failure. ¹⁷ At this stage, AAV therefore provides a useful platform to develop novel gene therapies for the heart. However, because AAV-based gene transfer remains largely episomal, it is uncertain whether expression will be maintained over the longer term (5-10 years). In addition, a significant downside of AAV is the limited insert capacity (<5.0 kb). ¹⁸ This may generate significant problems in biological pacing when transfer of larger genes (eg, sodium channel genes) is required or when gene combinations need to be carried by the vector. A final concern with the AAV-based approach is the presence of neutralizing antibodies in about 35% to 70% of the human population. These antibodies now exclude significant numbers of patients from receiving efficient gene transfer. Yet efforts are on their way to modify the AAV capsid and eliminate such an antibody-mediated immune response. ¹⁹ Alternatively, other systems such as lentiviral vectors are being explored. Lentiviral vectors are derived from the human immunodeficiency virus and, similar to this virus, they integrate into the host genome. ²⁰ They efficiently transduce cardiac myocytes ^21,22 and probably have the best potential for very long-term gene expression but their gene integration also comes with safety concerns regarding insertional mutagenesis. Another issue with lentiviral gene transfer is the efficient transduction of antigen-presenting cells that can prime cytotoxic T lymphocytes to target and eliminate transduced tissue. ²³ Because of these concerns and a more laborious production process, lentiviral vectors are now primarily used clinically for the ex vivo modification of stem cells. ²⁴ Yet, the first clinical trials using in vivo lentiviral gene transfer targeting the immune system, the eye and the brain ^{25 –27} have been initiated and it appears likely that, over time, these studies will extend to other organs including the heart.

Figure 1.

Major gene therapy vehicles used in cardiovascular applications. A, Scaled cartoons of viruses. From left to right, particles of adenovirus, adenoassociated virus (AAV), and lentivirus. B, Summary of important properties of these viral vectors. ds indicates double strand; ss, single strand.

Approaches to Gene Therapy-Based Biological Pacing

In developing gene therapies to create biological pacemakers, it is helpful to first consider how the normal heartbeat is initiated by the sinoatrial node (SAN). Although there is continuing controversy concerning the relative contributions of different currents to SAN automaticity, ²⁸ some basic principles are readily summarized. First, automaticity requires a net inward current to flow during diastole to drive the membrane potential to threshold. Second, autonomic regulation of automaticity is most readily achieved if cell membrane conductance during diastole is low such that a small current change has a marked effect on diastolic depolarization. These 2 considerations lead one to expect that outward current during diastole should be small, and in fact there is relatively little background potassium current in SAN cells. ²⁹ Third, the major currents contributing to automaticity should be responsive to autonomic agonists so that heart rate changes with alterations in sympathetic and parasympathetic activity. Finally, one might expect redundancy via multiple currents and pathways contributing to automaticity to insure robustness of such a critical function.

An autonomically regulated inward current that flows during diastole and that is well suited to contribute to spontaneous beating is the so-called pacemaker current, I_f, which derives from the hyperpolarization-activated cyclic nucleotide gated-(HCN) gene family. The HCN channels open after action potential repolarization, generating a depolarizing current during diastole. Further, these channels directly bind cyclic adenosine monophosphate (cAMP), resulting in activation occurring at less negative voltages ³⁰ to produce more inward current during diastole. The SAN also has a higher basal cAMP level than other cardiac regions, ³¹ which favors the activation of I_f. This high basal cAMP level creates a substrate wherein the autonomic nervous system can both decrease (parasympathetic) and increase (sympathetic) cellular cAMP to reduce and enhance automaticity, respectively. ³²

The recent development of a clinically approved I_f selective blocker and identification of patients with bradycardia having HCN mutations ³³ have confirmed that in humans I_f contributes to normal automaticity. At the same time, these findings have demonstrated that I_f is not the sole basis of spontaneous activity. Much of the recent effort exploring alternative contributors has focused on calcium homeostasis pathways and the manner in which these lead to activation of the Na/Ca exchanger, which generates an inward current while removing Ca²⁺ from the cell. ²⁸ Disruption of intracellular calcium stores leads to both reduced basal heart rate and reduced responsiveness to adrenergic agonists, highlighting the importance of normal calcium homeostasis to stable and robust pacemaker function. In the SAN, L-type Ca²⁺ current (I_Ca,L) is primarily responsible for the action potential upstroke and contributes to setting the threshold potential. It provides an obvious path connecting transmembrane events to calcium homeostasis. Further, L-type channel activity is enhanced by cAMP through protein kinase A-dependent phosphorylation, providing a link to autonomic responsiveness. However, it is now apparent that the communication between transmembrane and calcium homeostasis events may be 2-way. That is, not only do events at the membrane alter calcium homeostasis but changes in cytosolic Ca²⁺ levels can also alter events at the membrane. Recent studies found that SAN cells differ from working myocardium with respect to expression of adenylyl cyclase (AC) isoforms. Unlike ventricular cells, the SAN expresses Ca²⁺-activated isoforms of AC (types 1 and 8). ^34,35 Thus, increases in cytosolic Ca²⁺ can alter cyclase activity at the membrane and the function of cAMP-dependent channels, including the channels responsible for I_f and I_Ca,L.

Finally, various other inward currents have been suggested as contributing to automaticity. ²⁹ Among these, the most relevant to the topic of biological pacemakers is the fast inward Na current, I_Na. The cardiac isoform of this channel, even if present in SAN, would be largely inactivated by the relatively positive maximum diastolic potential of these cells. However, neuronal isoforms are expressed in the SAN of several species, including adult mouse ³⁶ and newborn rabbit. ^32,37 Unlike the cardiac isoform, these neuronal Na channel isoforms are not fully inactivated at the membrane potentials typical of the SAN. They can contribute during diastole and shift threshold potential negative.

Establishing Proof of Concept for Biological Pacing

Candidate gene therapy approaches are often first tested in vitro, and the most common model system for such studies is the monolayer primary culture of newborn rat ventricular (NBRV) cells. These cells are electrically coupled and beat spontaneously and synchronously, so the effect of an intervention is readily apparent by measuring the beating rate. These cells express adenovirus and lentivirus carrying genes with high efficiency and also can be transfected with reasonable efficiency by electroporation. ^{38 –40} This latter trait permits candidate genes to be prescreened for efficacy before a viral vector is prepared. The automaticity of these cultures derives in part from relatively low background potassium current (I_K1). This favors automaticity and interventions that can alter automaticity and in a limited way may represent a key characteristic of cardiac conducting tissue. Regardless, NBRV cultures are certainly not fully representative of either the conducting tissue or the working myocardium of the adult heart. Therefore, alternative testing systems are continuously being explored ^41,42 and some of the more relevant approaches to biological pacemakers have been pursued further via in vivo studies. In some cases, additional proof-of-concept studies are conducted in small animal models prior to large animal testing. However, due to the relatively rapid heart rates of smaller animals, these types of studies have sometimes been bypassed.

The approaches that have been explored can be separated into several categories. These include altering cellular programming to “create” a pacemaker-like cell, altering net current flow during diastole to introduce phase 4 depolarization, enhancing adrenergic responsiveness to unmask/increase endogenous pacemaking, and introducing additional ion channel genes to improve excitability and thereby enhance automaticity (eg, by altering action potential threshold). Each of these approaches is discussed subsequently.

Generating pacemaker cells by cellular reprogramming

Conceptually, one way to create a biological pacemaker would be to create SAN-like cells within the heart. In an effort to generate autologous pacemaker cells ex vivo, readily accessible cells (eg, skin fibroblasts or hair keratinocytes) have been isolated and reprogrammed toward a pluripotent state with the overexpression of a cocktail of transcription factors (eg, Oct3/4, Sox2, c-Myc, and Klf4). These pluripotent cells can then be used to generate cardiac myocytes, including pacemaker cells, via standard methods of stem cell differentiation. In a 15-day in vitro study, these pacemaker cells appeared capable of generating spontaneous activity with SAN-like properties such as intrinsic beating rate variability and intact autonomic responsiveness. ⁹

Alternatively, a gene therapy approach can be used to reprogram myocardial cells in vivo to generate pacemaker activity. In this case, without going through the pluripotent state, ventricular myocytes are transdifferentiated toward pacemaker-like cells. Initial proof of concept for this approach came from studies that employed overexpression of the T-box transcription factor TBX3. This factor was selected because of its well-established role in SAN formation during embryonic development. ⁴³ In adult mice harboring a working myocardial-specific, tamoxifen-inducible transgenic cassette, an efficient switch toward an expression profile of pacemaker myocardium could be induced upon the initiation of TBX3 expression. This resulted in the suppression of the gap junction genes encoding Cx40 and Cx43, the cardiac sodium channel gene SCN5a, and the inward rectifier K-channel genes. Although these changes enforced specific nodal properties such as a low I_K1 and slow impulse propagation, they did not initiate spontaneous activity. As an alternative to the tamoxifen-inducible transgenic mouse system, the phenotypic consequences of lentiviral TBX3 expression in neonatal cardiac myocytes were also explored. Here, TBX3 induced a variety of phenotypes including depolarized and spontaneously active cardiac myocytes. The results from both these biological systems indicate reductions in intercellular coupling and I_K1 to be critical reprogramming properties of TBX3. This type of reprogramming may be useful to enhance spontaneous activity, for example, in biological pacing based on HCN2. ⁴⁴

More recently, Cho and colleagues ⁴⁵ tested a similar approach with adenoviral gene transfer of various transcription factors. Using cultures of neonatal rat ventricular myocytes, they found TBX18 to have the strongest potential for increasing spontaneous activity (Figure 2A). This factor was therefore further studied after injection into the left ventricular apex of guinea pigs. Sinus bradycardia and complete atrioventricular (AV) block were induced by methacholine, revealing pacemaker activity from the TBX18-injected area (Figure 2B). Molecular and electrophysiological studies showed remarkably diverse reprogramming as indicated by reductions in Cx43 and I_K1 in combination with increases in HCN4, intracellular cAMP, and spontaneous Ca²⁺ cycling (Figure 2C). In the isolated Langendorff-perfused heart, TBX18 generated autonomically sensitive beating rates of ∼160 bpm in the setting of cryoablation-induced AV block. Although the TBX18-based approach is conceptually appealing, remaining issues include the documentation and interpretation of beating rate (ie, 24-hour stability of pacemaker performance was not recorded and relevance of the attained rate to outcome in large animals is unknown) and the heterogeneity of its effects (eg, only 13% of the transduced cells showed an upregulation in HCN4 expression).

Figure 2.

Reprogramming ventricular myocytes toward pacemaker myocardium. A, In a beating rate assay of adenovirally transduced NBRV cultures, TBX18 showed largest potential to increase spontaneous activity. B, Vector ECGs during methacholine-induced bradycardia show ectopic pacemaker activity originating from the left ventricular apex injection site in guinea pigs that received Ad-TBX18. Control animals that received Ad-GFP show slow idioventricular rhythms originating from the His-Purkinje system. C, Confocal line-scan images of intracellular Ca²⁺ concentrations show local Ca²⁺ release events (LCR) preceding whole cell Ca²⁺ transients (left panel). In control GFP-transduced cells, less pronounced Ca²⁺ sparks were detected (right panel). Reprinted from ⁴⁵ with permission. NBRV indicates newborn rat ventricular; ECGs, electrocardiograms; Ad, adenovirus; GFP, green fluorescent protein. *P < .05 vs control and other transcription factors

Increasing net inward current during diastole

Reducing background potassium current transforms ventricular myocytes into spontaneously active cells. A dominant negative construct of the Kir2.1 subunit (Kir2.1-DN) of the inward rectifier potassium channel forms nonfunctional channels by coassembling with endogenous subunits. The concept here was that net inward current would be increased and pacemaker activity enhanced by blocking the major outward current responsible for stabilizing resting membrane potential. Proof of concept for this approach was generated in guinea pigs, resulting in idioventricular rhythms that were not seen in control animals. ⁴⁶ However, reducing I_K1 affects multiple phases of the action potential including a potentially proarrhythmic prolongation of action potential duration. ⁴⁷

We pursued an alternative approach of enhancing inward current during diastole. The obvious candidate was the HCN gene family, with functional channels consisting of tetramultimers derived from 1 or more of the 4 HCN isoforms (HCN1-HCN4). We studied channels composed entirely of the HCN2 isoform, which has both relatively fast activation kinetics and strong cAMP responsiveness. ³⁰ Proof-of-concept studies in NBRV cultures resulted in a significant increase in spontaneous rate (Figure 3A and B). This increase was associated with both a greater slope of diastolic depolarization and a less negative maximum diastolic potential (Figure 3C), ³⁸ the latter reflecting the hyperpolarization-activated nature of HCN channels. Importantly, HCN2 expression did not induce automaticity in quiescent adult ventricle cells in culture because the channels activated more negatively in these cells as previously reported for native I_f in newborn versus adult ventricle. ⁴⁸ This latter observation emphasizes the importance of the substrate to which the vector is delivered, with different outcomes possible when these vectors are delivered to different cardiac regions.

Figure 3.

Spontaneous activity of NBRV cultures in the presence and absence of HCN2 overexpression. A, Spontaneous action potentials in nonexpressing NBRV. B, Spontaneous action potentials in HCN2-expressing NBRV. C, Summary data for control (n = 16-17), HCN2-expressing NBRV (n = 12-16), and GFP-expressing (n = 6) NBRV. Rate, diastolic depolarization (phase 4 slope), and maximum diastolic potential (MDP) are shown. Asterisk indicates significant difference relative to control. Reprinted from ³⁸ with permission. NBRV indicates newborn rat ventricular; Ad, adenovirus; HCN, hyperpolarization-activated cyclic nucleotide gated; GFP, green fluorescent protein. *P < .05 vs Control

We and others have studied additional HCN isoforms (HCN1 and HCN4) as well as mutated and chimeric HCN channels. ^{49 –52} The rationale behind these latter efforts was to fine-tune the biophysical characteristics in terms of voltage dependence, kinetics, and cAMP responsiveness for optimal biological pacemaker function. Some of these constructs, which were studied in large animals, are discussed later in this review. Although to date none of these single vectors based on the HCN gene family has achieved the ideal in vivo outcome with respect to basal and maximal heart rate, the result with the HCN212 chimeric channel is noteworthy in that it resulted in tachycardia when overexpressed in the canine left bundle branch (LBB). ⁵³ Thus, HCN212 demonstrates that a single gene approach based on the HCN family can drive the heart of a large mammal at least fast enough to achieve the desired physiological range. However, the difficulty in identifying the appropriate HCN-based construct for such a single gene therapy has led to dual gene therapy studies, where HCN is expressed along with another gene. These dual gene therapy studies have for the most part been carried out in the canine or porcine heart and are discussed later. However, the rationale for many of these efforts derives from earlier studies of the efficacy of these non-HCN genes alone to modulate automaticity or excitability, as described in the next sections.

Enhancing adrenergic signaling

The first gene therapy study of a biological pacemaker involved overexpression of the β₂-adrenergic receptor (β₂AR) and was initially tested in vitro and in mice. In adult mice, β₂AR plasmid DNA was injected into the right atrium (including the SAN) which increased cardiac chronotropy by ∼40%. ⁵⁴ Subsequent porcine studies tested β₂AR overexpression in the SAN, confirming the outcome in mice. ⁵⁵ A downside of this approach was the dependence on the native pacemaker system. This encouraged others to explore different strategies with the potential of generating de novo pacemaker activity. One such approach, tested only in vivo in the porcine heart, was overexpression of AC type 6. ⁵⁶ This is one of the endogenous ventricular AC isoforms and is not Ca²⁺ activated. Pacemaker activity following AC6 overexpression was only observed as an escape rhythm after rapid electrical stimulation in the setting of adrenergic agonist infusion.

Given that the SAN expresses Ca²⁺-activated AC isoforms 1 and 8, ^34,35 we considered the contribution that these isoforms might make to automaticity. To explore this, we overexpressed AC1 along with HCN2 in NBRV cultures. ⁵⁷ The results are summarized in Figure 4, which demonstrates that adding a Ca²⁺-activated AC isoform is sufficient to reproduce the Ca²⁺ sensitivity of pacemaker current adrenergic responsiveness that is seen in SAN. The HCN2 was overexpressed via adenovirus in NBRV along with flag-tagged AC isoforms, either AC6 or Ca²⁺-activated AC1. Both AC1 and AC6 coimmunoprecipitate with HCN2 (Figure 4A) and express comparably (equivalent maximal cAMP response to forskolin; Figure 4B). However, there are different functional outcomes with these ACs. In AC1 but not AC6 overexpressing cultures, both rate and basal cAMPs, are elevated (Figure 4C and D). Further, the ability of isoproterenol to shift HCN2 activation to less negative voltages is lost in NBRV when the Ca²⁺ chelator Bapta-AM clamps intracellular Ca²⁺ at diastolic levels (Figure 4E). That is, as with I_f in SAN, ⁵⁸ the response of HCN2 to isoproterenol is sensitive to intracellular Ca²⁺ in NBRV overexpressing AC1 (but not AC6). Taken together, these data show that AC1 introduces Ca²⁺ sensitivity to pacemaking where it is otherwise lacking, and importantly, results in elevated basal cAMP and elevated basal automaticity.

Figure 4.

AC1 and AC6 overexpression in NBRV cultures. A, Co-IP of AC1 and AC6 with HCN2. B, equivalent maximal cAMP after forskolin in NBRV-expressing AC1 and AC6. C, AC1-expressing NBRV exhibits higher spontaneous rate than AC6- or GFP-expressing NBRV. D, Total cAMP is greater in AC1-expressing NBRV compared to other conditions; *P < .05 versus GFP. E, AC1 but not AC6 introduces Ca²⁺ sensitivity to the β-adrenergic effect to shift HCN voltage dependence. *P < .05 versus corresponding no ISO. Modified from ⁵⁷ with permission. NBRV indicates newborn rat ventricular; AC, adenylyl cyclase; Co-IP, coimmunoprecipitation; HCN, hyperpolarization-activated cyclic nucleotide gated; GFP, green fluorescent protein; cAMP, cyclic adenosine monophosphate.

Enhancing excitability through addition of other inward currents alone or in combination

Automaticity in cardiac cells that are generating slow spontaneous rhythms (eg, the ventricular bundle branches) can be enhanced by altering various action potential parameters, including the slope of diastolic depolarization and the threshold potential. The former, as discussed earlier, can be achieved by increasing net inward current during diastole, for example, by expressing an HCN-channel isoform and/or by decreasing outward current. The latter can be achieved by enhancing Ca²⁺ influx and Na/Ca exchange current during late diastole. However, it can also be achieved by increasing available current for triggering the action potential, in particular by the addition of a current that activates at more negative voltages than I_Ca,L. A candidate for this is the family of fast inward Na channels.

As discussed earlier, the SAN has a largely Ca²⁺-dependent action potential but Na⁺ current contributes in some cases. ^32,36,37 In these cases, the current is not generated by the “cardiac” Na channel isoform (SCN5a), since this isoform is largely inactivated at the typical maximum diastolic potential in SAN. Rather, other isoforms, with a more favorable inactivation relation, are present. Even in the context of creating a biological pacemaker outside the SAN by expressing an Na channel along with HCN, SCN5a would not be the first choice since expression of HCN itself results in a less negative maximum diastolic potential. ³⁸ This was demonstrated in NBRV cultures by overexpressing either SCN5a or the skeletal Na channel isoform SkM1. This skeletal muscle isoform, like the “neuronal” isoforms present in some SAN tissue, has a less negative inactivation relation such that some channels are still available at resting potentials in the −60 to −70 mV range. In NBRV, SkM1 but not SCN5a is able to “rescue” the action potential upstroke from the depressing effect of potassium depolarization (10 mmol/L). ³⁹ In the next section, we discuss the efficacy of SkM1 coexpressed with HCN2 in the canine heart as a biological pacemaker.

Promising Large Animal Studies

Several of the gene therapy approaches to biological pacing have been studied in large animal models of bradycardia. The first studies used vagal stimulation to induce temporal bradycardia. ^59,60 However, the most extensive studies have been performed in the canine AV-block model where constructs were injected in the LBB or in the epicardium of the left ventricular free wall. ^49,61,62 Similarly, pigs have been made bradycardiac by RF ablation of the SAN or AV node, with subsequent construct injection into the left atrial appendage or His bundle. ^63,64

Testing of different HCN isoforms and mutants

The initial construct tested in large animals was wild-type HCN2. Proof of concept for the HCN2-based approach came from canine studies where vagal stimulation induced transient sinus arrest, revealing ectopic activity originating from the left atrium, where the construct had been injected. ⁵⁹ These studies were followed by experiments that confirmed enhancement of pacemaker activity after injecting the construct into the canine LBB; first in the setting of vagal stimulation-induced bradycardia ⁶⁰ and later during permanent AV block induced by RF ablation. ⁴⁹ Taken together, these studies showed the utility of an HCN2-based biological pacemaker and also indicated room for improvement in terms of basal beating rate, response to autonomic modulation, and dependence on electronic backup pacing.

The first efforts to improve HCN-based biological pacing involved employing different HCN isoforms or mutated channels with constitutive activity. For example, the HCN4 isoform is the predominant isoform in the SAN of most species, and patch-clamp studies indicate stronger sensitivity to cAMP stimulation than HCN2 (intermediate–high) or HCN1 (low). ⁶⁵ However, when HCN4 was overexpressed in the left ventricular free wall of pigs, acute AV-block studies indicated pacemaker performance that was relatively comparable to HCN2. ⁶⁶

An alternative hypothesis for improving functionality was to shift activation kinetics toward more positive potentials to activate more current during diastole. This has been studied with the point mutation HCN2-E324A ⁴⁹ and the HCN1-ΔΔΔ construct in which amino acids 235 to 237 (EVY) were deleted. ⁶³ HCN2-E324A did not perform better than HCN2 in vivo despite the favorable channel gating, perhaps because the mutated gene product did not express as well as HCN2. ⁴⁹ The HCN1-ΔΔΔ construct was injected into the left atrial appendage of porcines with SAN dysfunction based on RF-ablation. In this setting, HCN1-ΔΔΔ generated beating rates in the 60 to 65 bpm range and electronic backup pacing averaging around ∼15% of the beats. ⁶³ A follow-up in vitro article showed that around 60% of HCN1-ΔΔΔ-transduced myocytes had spontaneous activity while myocytes that overexpressed the wild-type HCN1 channel remained quiescent. ⁵¹

The observation that HCN1 is only minimally sensitive to cAMP stimulation led to the design of the chimeric construct HCN212, containing the transmembrane domain of HCN1 (fast gating) and the N- and C-terminal cytoplasmic domains of HCN2 (including the cyclic nucleotide-binding domain). Unfortunately, when HCN212 was injected into the LBB of AV-blocked canine, outcomes were excessive, with bursts of ventricular tachycardia exceeding 200 bpm. ⁵³ In vitro and in silico studies later revealed potential mechanisms contributing to this burst pacing behavior of HCN212 related to the nonequilibrium properties of the channel. ⁵⁰ These studies suggest that channels with properties intermediate between HCN212 and HCN2 may generate more favorable outcomes.

Increasing intracellular cAMP

In vitro studies suggested AC1 to be a prime candidate to increase intracellular cAMP, thereby potentially impacting on multiple pacemaker mechanisms including I_f and Ca²⁺-cycling pathways. ⁵⁷ Proof of concept for this approach was generated in AV-blocked canine that received adenoviral AC1 in the LBB, either alone or in combination with HCN2. In this setting, AC1 overexpression alone generated robust biological pacing at around 60 bpm, minimal dependence on electronic backup pacing (<2%), and high sensitivity to parasympathetic modulation (short-term beat rate variability was higher in AC1 than in HCN2). Sensitivity to sympathetic modulation (as assessed by long-term beat rate variability) was comparable between HCN2 and AC1. The combination strategy HCN2/AC1 generated faster beating rates and higher sensitivity to sympathetic modulation; however, these exceeded the physiologically desirable ranges. In sum, these experiments indicated clinically applicable biological pacing based on AC1 and potential improvement with the addition of HCN2, although the latter would require further titration of gene expression to balance the outcome. ⁶² A potential concern with the AC1-based approach is that increases in cAMP may give rise to problems in relation to Ca²⁺ overload and triggered activity. This concern remains at this time theoretical but merits further investigation.

Other dual gene therapies

Two other approaches to improve HCN2-based biological pacing have been tested in large animals. One of these is the combination of HCN2 with Kir2.1-DN, the dominant negative construct that targets the inward rectifier K current. The rationale here was that in the setting of reduced outward I_K1 less inward I_HCN2 would be needed to depolarize the membrane. Combined overexpression of HCN2 and Kir2.1-DN in the bundle of His of AV-blocked porcine indeed generated efficient biological pacemaker activity with basal beating rates in the 90 to 95 bpm range and minimal electronic back-up pacing. ⁶⁴ Although autonomic modulation was not reported, this may have been compromised as a result of I_K1 suppression-inducing membrane depolarization, reducing the overall contribution of cAMP-sensitive HCN2. An additional downside of approaches based on I_K1 suppression is the potentially proarrhythmic prolongation of action potential duration. ⁴⁷

Another dual gene therapy approach is based on HCN2 overexpression in combination with SkM1 (Nav1.4). This combination was selected based on the observation that HCN2 overexpression depolarizes the membrane ³⁸ (although not as strongly as I_K1 suppression), inactivating the voltage-gated cardiac sodium channel (Nav1.5, encoded by SCN5a). The hypothesis was that restoring/increasing Na-channel availability could hyperpolarize the action potential threshold and thereby improve pacemaker activity. SkM1 was selected because of its positively shifted kinetics of inactivation compared to Nav1.5, ⁶⁷ which results in greater channel availability at depolarized potentials. ³⁹ To test the impact of SkM1 overexpression in the setting of HCN2-based biological pacing, we injected both constructs into the LBB of dogs in complete heart block. The outcome was highly efficient biological pacing at basal beating rates of ∼80 bpm, maximal beating rates of ∼130 bpm, and a complete elimination of electronic backup pacing. ⁶¹

Maximal beating rates of HCN2/SkM1-injected animals were significantly faster than that of animals injected with HCN2 or SkM1 alone (Figure 5A and B). Moreover, when examining the 24-h Holter recordings, we detected a gradual warming up and cooling down of pacemaker rhythms in HCN2-injected animals (Figure 5C), whereas SkM1-treated animals frequently showed bigeminal rhythms at slow heart rates which converted to regular rhythms at higher rates (Figure 5D). In comparison, animals injected with the combination of HCN2 and SkM1 showed stable beating rates at baseline with robust responses in rate acceleration and cooling down (Figure 5E).

Figure 5.

Maximal beating rates in HCN2/SkM1-based biological pacing. A, Maximal beating rates as recorded by 24-hour Holter recording throughout a 7-day study of AV-blocked canine that received construct injections into the left bundle branch (LBB). ⁺ P < .05 for HCN2/SkM1 versus HCN2 and SkM1. B, Summary data of the pooled data of days 5 to 7 in LBB-injected animals compared to animals that received construct injections into left ventricular free wall epicardium. ⁺ P < .05 for LBB injected HCN2/SkM1 versus respective HCN2 and SkM1; ^‡ P < .05 for LBB versus subepicardially injected HCN2/SkM1. C-E, Typical examples of beating rate recordings during 8 minutes around an episode of maximal beating rate (left panels) and associated Holter ECGs (right panels). Reprinted from ⁶¹ with permission. HCN indicates hyperpolarization-activated cyclic nucleotide gated; AV, atrioventricular; ECGs, electrocardiograms.

In exploring the mechanisms underlying the highly efficient outcome with HCN2/SkM1, we found that injection into the LBB was critically important. This was evident by the outcome of HCN2/SkM1-injected animals that received the construct subepicardially. In this case, basal beating rates, maximal beating rates (Figure 5B), and the dependence on electronic back-up pacing were all inferior to animals that received HCN2/SkM1 in the LBB. ⁶¹ The latter appears to result from specific LBB features such as the presence of endogenous pacemaker function and lower I_K1 current density. ⁶⁸ Second, we isolated myocardial bundles from the injected animals and showed via microelectrode recordings that the action potential threshold was hyperpolarized as a result of SkM1 overexpression. ⁶¹ As a result, SkM1 brings the unchanged maximum diastolic potential closer to the action potential threshold. In conclusion, the high efficiency of HCN2/SkM1-based biological pacing appeared attributable to HCN2-induced diastolic depolarization, SkM1-induced hyperpolarization of the action potential threshold, and injection in a bundle branch environment.

Cellular delivery of ion-channel function

As an alternative to direct gene transfer, cellular delivery of pacemaker function-related genes has also been explored. Proof of concept for this approach has been provided by HCN2 overexpressing MSCs. These cells abundantly express connexins and are thereby able to form gap junctions with adjacent cardiac myocytes. ⁵ This coupling allows for HCN channel activation via hyperpolarization induced by the cardiac myocyte. Next, the inward current generated by the HCN channel drives membrane depolarization in the myocyte. When these HCN2 expressing MSCs were injected into the left ventricular free wall of AV-blocked dogs, demand pacemaker activity was generated with outcomes comparable to the adenoviral approach. ⁶ Moreover, separate studies showed efficient cellular delivery of SkM1 ^67,69 suggesting the MSC-based approach as an attractive carrier for HCN2/SkM1. Yet, a setback to the use of gene-modified stem cells was the finding that HNC2-based pacemaker function started to vanish after approximately 8 weeks. Current studies therefore focus on the stabilization of the cellular substrate via the use of biomaterials and different stem cell sources.

A Work in Progress

For the strategies that have proven effective in large animal studies, the next phase of development will focus on long-term delivery. Most of the approaches will likely be followed up by gene transfer via AAV. An exception here is HCN2/SkM1, which will need to be delivered via lentiviral vectors or gene-modified stem cells ⁶⁹ because the relatively large gene size of SkM1 challenges incorporation into AAV. If function can be maintained stable over a period of months to years, extensive safety testing will become a priority. In this respect, potential issues that need to be addressed include susceptibility to proarrhythmia, off-target gene expression, toxicity, and potential for insertional mutagenesis.

Additional challenges on the road ahead are titration, delay of function, and control of gene expression. Titration studies in the setting of biological pacing have not yet been performed but this will likely be an integral part of the long-term expression studies with lentiviral vectors and AAV. Moreover, there is at least a delay of several days (lentiviral vectors) or several weeks (AAV) before robust gene expression is established. In most cases, such a delay needs to be bridged by a temporal or permanent electronic pacemaker. The most logical solution here would be that of tandem biological-electronic pacing, which has been the protocol for many of the experimental studies. This takes care of the delay in gene expression and provides for the proven safety of electronic pacing. At the same time, such an approach takes advantage of benefits of biologics such as improved sensitivity to autonomic modulation and stimulation from a more physiological implantation site (eg, LBB). Finally, maintenance of stable gene expression after its initial plateau will largely depend on tolerance of transduced cells to high transgene expression and may require further optimization.

When considering potential application areas, it should be noted that all of the most relevant strategies have been tested in settings of demand pacing of the atrium or ventricle. As such, they cannot provide for AV-sequential pacing which is needed in the large majority of patients (∼75%) ^1,70 Although studies are being undertaken to repair AV conduction (primarily by cell therapy), ^{70 –72} these efforts are still in their infancy. Initial application of biological pacing in a clinical trial is therefore expected to target ventricular-demand pacing of patients who do not require AV-sequential pacing (eg, patients who are in permanent atrial fibrillation). Whether or not the field will get to this stage will in part depend on the advancements in electronic pacing (which could remove the need for biologics), but more importantly it critically relies on the success of experimental long-term safety trials that are scheduled for the coming years.

Conclusions

The development of gene therapy-based biological pacemakers is entering into a new engineering phase. Several constructs and combinations have reached a level of pacemaker function that holds strong potential for clinical application. The most relevant of these strategies have been tested in clinically relevant large animal models of bradycardia and employed sophisticated catheter-based delivery systems that are readily becoming available for application in humans. Ongoing efforts therefore focus on safe delivery of long-term biological pacemaker function based on AAV- or lentiviral-mediated gene transfer.