Long-Term Cardiac-Targeted RNA Interference for the Treatment of Heart Failure Restores Cardiac Function and Reduces Pathological Hypertrophy
Background— RNA interference (RNAi) has the potential to be a novel therapeutic strategy in diverse areas of medicine. Here, we report on targeted RNAi for the treatment of heart failure, an important disorder in humans that results from multiple causes. Successful treatment of heart failure is demonstrated in a rat model of transaortic banding by RNAi targeting of phospholamban, a key regulator of cardiac Ca2+ homeostasis. Whereas gene therapy rests on recombinant protein expression as its basic principle, RNAi therapy uses regulatory RNAs to achieve its effect.
Methods and Results— We describe structural requirements to obtain high RNAi activity from adenoviral and adeno-associated virus (AAV9) vectors and show that an adenoviral short hairpin RNA vector (AdV-shRNA) silenced phospholamban in cardiomyocytes (primary neonatal rat cardiomyocytes) and improved hemodynamics in heart-failure rats 1 month after aortic root injection. For simplified long-term therapy, we developed a dimeric cardiotropic adeno-associated virus vector (rAAV9-shPLB) to deliver RNAi activity to the heart via intravenous injection. Cardiac phospholamban protein was reduced to 25%, and suppression of sacroplasmic reticulum Ca2+ ATPase in the HF groups was rescued. In contrast to traditional vectors, rAAV9 showed high affinity for myocardium but low affinity for liver and other organs. rAAV9-shPLB therapy restored diastolic (left ventricular end-diastolic pressure, dp/dtmin, and τ) and systolic (fractional shortening) functional parameters to normal ranges. The massive cardiac dilation was normalized, and cardiac hypertrophy, cardiomyocyte diameter, and cardiac fibrosis were reduced significantly. Importantly, no evidence was found of microRNA deregulation or hepatotoxicity during these RNAi therapies.
Conclusions— Our data show for the first time the high efficacy of an RNAi therapeutic strategy in a cardiac disease.
Received April 10, 2008; accepted November 21, 2008.
RNA interference (RNAi) has been investigated as a new treatment option in infectious diseases and cancer. We report here on a strategy for the treatment of a cardiac disease by locally induced RNAi. Heart failure (HF) remains a leading cause of mortality in the developed world. Current drug treatment has limited efficacy, and in advanced HF, left ventricular (LV) assist devices or heart transplantation are the ultimate options. For long-term treatment of HF, novel approaches are currently being explored, including gene and cell therapies, whereas the use of RNAi to modulate cardiac gene functions has not yet been evaluated. Whereas gene therapy rests on recombinant protein expression as its basic principle, RNAi therapy instead uses small regulatory RNAs to achieve its effect. Targeting, kinetics, and toxicity of these RNAs in vivo are grossly different from those of recombinant proteins and not yet well characterized.
Clinical Perspective p 1252
Although HF may result from multiple causes, defective cardiac Ca2+ homeostasis has been identified as an important final common pathway. In the present study, we show successful treatment of HF by RNAi targeting of a key regulator of cardiac Ca2+ homeostasis. Malfunction of the failing heart is due in part to dysfunction of the phospholamban-controlled sarcoplasmic reticulum Ca2+ ATPase pump (SERCA2a), which results from reduced SERCA2a expression and/or phospholamban phosphorylation.1 Unphosphorylated phospholamban keeps the Ca2+ affinity of SERCA2a low, which results in decreased sarcoplasmic reticulum (SR) Ca2+ uptake, slowed relaxation, and decreased SR Ca2+ load, whereas phospholamban phosphorylation in response to β-adrenergic stimulation relieves this inhibition.2 Germline ablation of the phospholamban gene, gene transfer for dominant-negative phospholamban mutants,3,4 phospholamban antisense RNAs,5 and intracellular inhibitory phospholamban antibodies6,7 have been used to increase SERCA2a activity and to rescue HF models.8 RNAi mediated by chemically synthesized small interfering RNAs in cardiomyocytes showed very low efficacy and stability even in vitro,9 and pharmacological approaches to phospholamban modulation have failed thus far. Fundamental limitations of synthetic small interfering RNAs are their rapid degradation in plasma and target cells and the unsolved problem of achieving adequate transfer and targeting in vivo. Viral vectors have the potential to overcome these limitations, and we previously showed highly efficient phospholamban ablation in primary neonatal rat cardiomyocytes (NRCMs) by an adenoviral RNAi vector.10 No change in the expression of other cardiac proteins, including Ca2+ handling proteins, occurred, which indicates high target specificity.
In the present study, we evaluated the principle of RNAi against phospholamban for short-term and long-term treatment of HF in vivo. Functional characterization of a series of vectors and the determinants of their efficacy was followed by investigation of both an optimized recombinant adeno-associated virus pseudotype 9 vector (rAAV9) alongside a traditional adenoviral vector in an animal model of HF and cardiomyocyte microRNAs (miRNAs) during RNAi therapy.
Development of Recombinant Adenoviral and Adeno-Associated Virus Vectors
rAAV vectors were developed for the in vitro studies as pseudotyped rAAV6 and for the in vivo work as rAAV9. Throughout all in vitro and in vivo studies, we used only self-complementary adeno-associated virus genomes owing to their enhanced performance compared with single-stranded adeno-associated virus vectors. Vector maps are shown in Figure 1⇓A. For details of all methods used, see the online-only Data Supplement.
Vector Production and Purification, Quality Assessment, and Titration
rAAV9-shGFP (rAAV9 generating small hairpin RNA [shRNA] to silence green fluorescent protein [GFP]) and rAAV9-shPLB (rAAV9 generating shRNA to silence phospholamban) were produced by a 2-plasmid protocol described previously11 with the following modifications: 293T cells were grown in triple flasks for 24 hours (DMEM, 10% FBS) before the addition of calcium phosphate precipitate. After 72 hours, the virus was purified from Benzonase-treated (Merck, Darmstadt, Germany) cell crude lysates over an iodixanol density gradient, followed by heparin-agarose type I affinity chromatography. Finally, viruses were concentrated and formulated into lactated Ringer solution with a Vivaspin 20 centrifugal concentrator (ISC BioExpress, Kaysville, Utah) with 50K molecular weight cutoff and stored at −80°C. Vector stock biochemical purity (>95%) was assessed by silver staining after electrophoresis. Genome-containing particles were determined by a real-time polymerase chain reaction approach.
Vector Evaluation in Primary Cardiomyocytes
Primary NRCMs are suitable to pretest any RNAi-based cardiac therapy before its definitive test in vivo because although developmentally regulated, the SERCA2a/phospholamban system functions well in NRCMs, and adenoviral gene transfer strategies that target the SERCA2a/phospholamban system have been successful in both neonatal and adult cardiomyocytes. Although both cell types are suited for in vitro pretesting, a number of other differences between cultured cardiomyocytes and the intact heart in vivo render any in vitro study of RNA-based therapies in cultured cells preliminary.
NRCMs were prepared from ventricular tissue of 1- to 3-day-old Wistar rat pups and grown in 6-well dishes. Phospholamban, troponin I, sodium-calcium exchanger, and SERCA2a mRNA or protein expression were determined by Northern and Western blot analyses as described previously.10 [Ca2+]i transients were measured during electrical stimulation at 1 Hz after loading of NRCMs with 8 μmol/L Fluo-4 AM for 20 minutes (image capture at 120 Hz, 8.3 ms per image). Five treatment groups of NRCMs were studied: AAV9-shPLB (n=26 cells), AAV9-shGFP (n=26 cells), adenovirus (AdV)-shPLB (n=71 cells), AdV-shGFP (n=49 cells), and untreated control cells (n=32). The amplitude of the transient (systolic [Ca2+]; F/F0), its time to peak, and the time constant (τ) of its decay were measured.
Induction of Hypertrophy
Phenylephrine at a concentration of 100 μmol/L was used in portions of the in vitro studies as a hypertrophic stimulus. TaqMan assays (Applied Biosystems, Foster City, Calif) to quantify the cellular miRNAs were performed in NRCMs, either under baseline conditions or in the presence of phenylephrine, or in rat hearts. The agent was added on day 2 of culture, either alone or together with the respective RNAi vector.
In search of possible influences of vector-derived shRNAs on cardiomyocyte miRNAs, we used TaqMan assays to quantify 2 miRNAs with known cardiac functions.12–16
Transaortic Banding and Serial Echocardiographic Assessment
Four-week-old Sprague Dawley rats (weight 70 to 80 g) were anesthetized with pentobarbital (65 mg/kg IP) and placed on a ventilator. A suprasternal incision was made to expose the aortic root, and a tantalum clip with an internal diameter of 0.58 mm was placed on the ascending aorta. Animals in the sham group underwent a similar procedure without insertion of a clip. In the animals that were aortic-banded, we waited 25 to 30 weeks for the animals to develop LV dilatation and a decrease in ejection fraction by 25% before cardiac gene transfer. Of the initial 56 that underwent pressure-overload hypertrophy, only 40 animals survived; these were further grouped to receive either AdV-shGFP (n=10), AdV-shPLB (n=10), rAAV9-shGFP (n=10), or rAAV9-shPLB (n=10). Operators performing the echocardiographic studies were blinded in terms of the animal groups they were studying.
Cardiac Distribution of rAAV9 Vectors After Intravenous Injection
Rats were transduced either with a vector rAAV9-GFP that expresses the marker protein GFP or with saline. One month after delivery of rAAV9-GFP or saline, hearts were removed and visualized under a fluorescence system (Maestro In Vivo Imaging, Cambridge Research & Instrumentation, Inc, Woburn, Mass) at 510 nm with a single excitation peak at 490 nm of blue light. In addition to this visualization of GFP expression, GFP immunohistological staining was performed 1 month after intravenous injection of rAAV9-GFP to evaluate vector distribution on a microscopic scale.
Experimental Protocol for RNAi Therapy In Vivo
The adenoviral delivery system has been described in detail previously by our group.17–20 Briefly, after the rats were anesthetized and a thoracotomy was performed, a 22-gauge catheter containing 200 μL of adenoviral (3×1010 pfu) solution was advanced from the apex of the LV to the aortic root. The aorta and main pulmonary artery were clamped for 40 seconds distal to the site of the catheter, and the solution was injected; then, the chest was closed, and the animals were allowed to recover. For experiments with rAAV9, a simple tail-vein injection was performed with 5×1011 genomes of either rAAV9-shRNA vector. Animals in the sham group were injected with saline.
Hemodynamics and Cardiac Histology During RNAi Therapy
Rats in the different treatment groups and at different stages after adenoviral gene transfer were anesthetized with pentobarbital 40 mg/kg and mechanically ventilated. A small incision was then made in the apex of the LV, and a 2.0F high-fidelity pressure transducer (Millar Instruments, Houston, Tex) was introduced into the LV. Pressure measurements were digitized at 1 KHz and stored for further analysis. The operators performing the hemodynamic studies were blinded in terms of the animal groups they were studying.
Data in Figures 1C through 1F, 2G, and 3A through 3⇑⇓⇓⇓F and in online-only Data Supplement Figures Ia through Ie, IIIa through IIIe, and IVa through IVc are presented as mean (columns)±SD (error bars). We analyzed the groups of rats that received interventions using a 2-step procedure. First, we performed an overall F test to determine whether any significant difference existed among any of the means. We then selected 2 means and performed a Tukey test for each mean comparison. We next determined whether the Tukey score was statistically significant with Tukey probability/critical values. Statistical significance was accepted at the level of P<0.05.
Drs Poller, Hajjar, Fechner, and Suckau designed and performed research, analyzed data, and wrote the article. The remaining authors performed research or contributed analytical tools. The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
Optimization of RNAi Systems
The determinants of the silencing efficacy of viral RNAi vectors were investigated because we observed that rAAV-shPLB vectors with apparently minor structural differences (Figure 1⇑A) had grossly different shRNA production rates and target silencing in NRCMs in vitro (Figure 1B and 1⇑C). For these initial studies in NRCMs, the rAAV6 pseudotype was used, which has higher transduction efficacy than rAAV9 in vitro. For the later RNAi therapeutic investigations reported in Figures 2 and 3⇑⇑, only rAAV9 was used, which displays superior cardiac transduction in vivo.
The in vitro experiments showed that coexpression of GFP marker to tag cells harboring the shPLB vector nearly abolished shPLB production (Figure 1⇑C) and phospholamban silencing (Figure 1⇑B). The presence of a cytomegalovirus (CMV) promoter in the expression cassette that contained the U6 promoter used for shRNA transcription reduced shRNA production strongly if GFP was driven by the CMV, but also if CMV was linked to a β-intron (online-only Data Supplement Figure I). Figure 1⇑ shows that by far the highest efficacy was displayed by an rAAV construct previously considered too short for efficient packaging. Comparison of the shRNA transcription by AdV-shPLB versus rAAV6-shPLB in NRCMs showed a decline to one third by day 10 for the adenoviral vector but constant expression for the rAAV vector (Figure 1⇑C). Ablation of phospholamban expression at the mRNA level was >98% for both vectors at a dose of 4×103 vector particles per cell. Interestingly, incorporation of a CMV-GFP cassette to allow detection by in vivo imaging unexpectedly led to a vector unable to silence its target. CMV promoter–driven marker gene expression is apparently unsuitable for use in U6-shRNA vectors, and we therefore chose only the simplest and most efficient U6-shRNA vectors (Figure 1⇑) for in vivo RNAi. rAAV9-shRNA was used for long-term therapy in vivo because of its highly stable shRNA production compared with AdV-shPLB and its long-term stability in vivo. For short-term therapy, the adenoviral vector was used. In vitro, a lag of several days in phospholamban ablation was found at the protein level compared with the mRNA level, with protein leveling off at 9% (AdV-shPLB) and 12% (rAAV6-shPLB) of baseline, respectively, on day 7 (Figure 1⇑D).
Measurement of [Ca2+]i transients during AAV9-shPLB treatment of NRCMs (Figures 1E and 1⇑F) showed that this vector led to significantly higher amplitude and accelerated transient kinetics (shortened time to peak and τ) compared with the AAV9-shGFP group, with transients indistinguishable from untreated cells. AdV-shPLB treatment also resulted in a significantly higher amplitude than with AdV-shGFP. In contrast to the AAV9 groups, time to peak [Ca2+]i transient was prolonged in AdV-shPLB versus AdV-shGFP, and no difference was found in τ. Studies of SR Ca2+ loading in the adenoviral groups showed increased SR Ca2+ loading and fractional Ca2+ release from the SR in the AdV-shPLB versus AdV-shGFP group (online-only Data Supplement Figure Id). With respect to cell-to-cell variability of transduction in vitro, online-only Data Supplement Figure Ib shows grossly homogeneous GFP expression in NRCMs treated with rAAV-GFP marker vector, which serves as the best possible approximation of a direct demonstration of homogeneous shPLB expression in vitro. With current technology, the latter cannot be visualized directly, because coexpression of GFP together with shPLB extinguishes its silencing capacity (Figure 1B and 1⇑C). Homogeneous spatial and temporal distribution of the RNAi vectors in rat hearts in vivo is also indirectly inferred by fluorescent imaging (Figure 2B and 2C) of a GFP vector of the same type (rAAV9) as used for RNAi therapy (Figure 3⇑). Figure 2D and online-only Data Supplement Figure IIa through IIc show GFP immunohistochemical staining of the heart and other organs after intravenous injection of rAAV9-GFP. Strong and grossly homogenous expression in the heart contrasts with weak staining of liver and skeletal muscle and no visible staining of the lungs (for quantitation, see online-only Data Supplement Figure IIc).
Efficacy of RNAi Therapy In Vivo
Transaortic banding led to HF in rats after 30 weeks. The experimental protocol for in vivo RNAi therapy is outlined in Figure 2A. Fluorescent imaging and immunohistological analysis of a GFP vector of the same type (rAAV9) as used for the RNAi therapies showed grossly homogeneous cardiac GFP expression 1 month after intravenous injection on the macroscopic and microscopic scale and may be assumed to approximate the cardiac shRNA expression levels generated by the RNAi vectors (Figures 2B through 2D). Direct measurement of shPLB production in vivo is unfeasible with current technology. Figures 2F and 2G show significantly decreased cardiac phospholamban protein after treatment with either AdV-shPLB or rAAV9-shPLB. SERCA2a protein was decreased in failing hearts, whereas shPLB therapy was accompanied by an increase in cardiac SERCA2a protein. The sodium-calcium exchanger was not changed significantly. Treatment by aortic root injection of AdV-shPLB compared with AdV-shGFP control vector (generating an irrelevant shRNA sequence directed at xenogenic GFP) served as a model of short-term treatment of acute (and potentially reversible) HF. One month after injection, the LV end-diastolic pressure, rate of LV pressure decrease (−dP/dt), and isovolumetric relaxation time constant (τ) as measures of diastolic function (Figure 3⇑A and online-only Data Supplement Figure IIIa) were significantly (P<0.05) better in the AdV-shPLB group than in the control group. LV systolic pressure, rate of LV pressure increase (+dP/dt), and fractional shortening (FS) as parameters of systolic function (Figure 3⇑B; online-only Data Supplement Figure IIIb) were likewise improved. Beyond these beneficial effects on hemodynamics, the massive cardiac hypertrophy and dilation after transaortic banding were reduced significantly at 1 month (Figure 3⇑C; online-only Data Supplement Figure IIIc). Postmortem morphometric data (LV–body weight ratio, LV–tibia length ratio) correlated with echocardiography (Figure 3⇑D). Histology showed reduced cardiomyocyte size after 1 month of AdV-shPLB therapy, whereas cardiac collagen content was unchanged (Figures 3E and 3⇑F). Survival rates were 8 of 10 versus 9 of 10.
Treatment by intravenous injection of the most efficient rAAV9-shPLB compared with the rAAV9-shGFP vector served as a model of long-term therapy of chronic HF. Three months after injection, diastolic function (LV end-diastolic pressure, −dP/dt, τ; Figure 3⇑A; online-only Data Supplement Figure IIIa) was improved significantly by rAAV9-shPLB therapy compared with control and was no longer significantly different from the sham-operated non-HF group. Systolic function (LV systolic pressure, +dP/dt, fractional shortening; Figure 3⇑B; online-only Data Supplement Figure IIIb) was also restored, although less so than diastolic functional parameters. Beyond hemodynamics, this treatment reduced LV hypertrophy (LV–body weight ratio) and dilation (LV–tibia length ratio) at 3 months (Figure 3⇑C; online-only Data Supplement Figure IIIc). Echocardiography corroborated the reduction of LV wall thickness and dilation (Figure 3⇑D; online-only Data Supplement Figure IIId). Histology showed reduction of both cardiomyocyte size and cardiac collagen after 3 months of rAAV9-shRNA therapy (Figures 3E and 3⇑F). Survival of rAAV9-shPLB–treated animals after 3 months was 9 of 10 versus 6 of 10 in the control group.
Cardiac miRNAs and RNAi Treatment
shRNAs exploit the cellular machinery of RNAi to mediate therapeutic effects by mimicking the endogenous process but may disturb cellular miRNA pathways.21 Because miRNAs play important roles in cardiac morphogenesis,12 hypertrophy,13,14 arrhythmogenesis,15 and failure,16,22 we searched for possible side effects of the RNAi vectors at the miRNA level in NRCMs, under standard culture conditions and in the presence of the hypertrophy-inducing drug phenylephrine. In the absence of phenylephrine, no significant effects of any vector on miRNA-1 or miRNA-133 were found. In the presence of phenylephrine, a marked reduction was found on day 5 that was reversed in NRCMs treated with shPLB vectors. Rat hearts treated with shPLB vector had higher miRNA levels than the shGFP group (online-only Data Supplement Figure IIa through IIc). With respect to possible side effects of the RNAi vectors, hematoxylin-and-eosin stains of the liver (Figure 2E) and other organs after vector injection revealed no pathological findings.
RNAi Therapy of HF
Building on recent landmark articles by other groups describing highly cardiotropic vector systems23–26 and after work on rAAV-mediated downregulation of phospholamban,27–29 the present study demonstrates for the first time the in vivo restoration of cardiac function and reduction of pathological hypertrophy and dilation in an HF animal model by RNAi. Comparison of the 2 vector systems used here suggests that for intermediate time scales, adenoviral vectors may suffice and even provide advantages over long-term stable rAAV,23–26 because RNAi may be desirable only temporarily in acute and potentially reversible HF. In fact, the significant improvement of diastolic and systolic function and LV morphology 1 month after AdV-shPLB treatment is evidence of at least a functional therapeutic benefit from the adenoviral system. What has been shown previously for classic gene transfer therapy18–20,30 may obviously work for RNAi-based strategies, too, although additional constraints exist for RNAi vector structure to avoid loss of therapeutic efficacy (Figures 1A through 1⇑C) and disturbance of miRNA pathways.21 The experiments further suggest that rAAV-based RNAi may be suitable for the long-term treatment of chronic HF by RNAi strategies. In classic gene therapy studies, rAAV vectors have supported stable transgenic protein expression for more than 1 year, which was never achieved with adenoviral systems in immunocompetent hosts. Although shRNA production from rAAVs is different in several aspects from classic gene transfer (Figure 1⇑), the data from the rAAV arm of the present study provide the first evidence that cardiac rAAV9-based shRNA production remains stable for a period of time sufficient for long-term improvement of cardiac function and possibly also survival in HF.
We have developed a method for gene delivery by recombinant adenovirus using cross-clamping of the aorta and the pulmonary arteries in rat hearts, which yields homogeneous transgene expression17 and has been used by several groups during the past 10 years. In terms of rAAV9, we now show by fluorescent imaging and immunohistochemistry that a single intravenous injection of an rAAV9-GFP marker vector induces strong and grossly homogeneous cardiac GFP expression 1 month after treatment (Figures 2B through 2D). The important finding that coexpression of GFP together with shPLB extinguishes its silencing capacity (Figures 1B and 1⇑C) prevents, at the present time, direct demonstration of spatial and temporal uniformity of cardiac shRNA synthesis in vivo by GFP coexpression. However, uniformity may be deduced by inference from our analogous in vivo imaging of GFP expression from adenoviral and rAAV9 vectors. Using rAAV9-GFP, we further show that the transduction rate increases over time and reaches 90% at 1 month. Improvement of HF by both RNAi therapeutic protocols is mediated via ablation of phospholamban protein, which occurs in cultured NRCMs10 (Figure 1⇑D) and in rat hearts in vivo (Figures 2F and 2G).
Cardiac phospholamban protein was decreased significantly after RNAi therapy. Although sodium-calcium exchanger expression was unchanged, SERCA2a expression was more complex. In failing hearts treated with control vectors, SERCA2a expression was decreased, as expected in any experimental HF model compared with sham. After RNAi therapy, SERCA2a expression was found to be increased compared with HF groups, consistent with the fact that RNAi therapy normalized LV function. Because SERCA2a expression is a well-known marker of the degree of HF,31 its increase after RNAi therapy by shPLB reflects the improved status of cardiac function. Both RNAi therapies induced a decrease in cardiomyocyte size (Figure 3⇑E). In contrast, adenovirus-based therapy over 1 month did not influence the fibrosis induced in failing hearts, whereas long-term rAAV9-based treatment resulted in significantly reduced fibrosis after 3 months (Figure 3⇑F). This could be a consequence of basic differences between the traditional adenoviral and the new AAV9 vectors, as suggested by the calcium transient studies in vitro. Adenovirus per se could induce changes in the transcriptional program of target cells different from those of AAV9 (see discussion of possible side effects below). Irrespective of such differences in vitro, however, the effect of the RNAi vectors in vivo on phospholamban ablation, hemodynamics, and morphology confirm that the doses chosen for both systems are within the therapeutic range.
The AAV9 vector used in the present study fulfills 1 initial requirement for application in human HF, because it is cardiotropic in primates.29 Beyond its long-term stability, rAAV9 offers further advantages of clinical interest through cardiac targeting after intravenous injection (Figures 2B through 2D) and low immunogenicity.32 In contrast to rodents, regulatable phospholamban modulation is most likely required in humans, because permanent phospholamban deficiency or phospholamban dysfunction due to genomic mutations has been associated with cardiomyopathies.33–35 Drug-regulatable RNAi appears possible, however, on both vector platforms used here. Transcriptional control of shRNA expression is technically more demanding than the use of the tissue-specific promoters employed for traditional gene therapy, because these promoters are unable to support proper formation and cleavage of shRNA. New promoters compatible with shRNA biosynthesis have been developed and may add an additional safety feature to organ-targeted RNAi therapy.
Independent of its therapeutic potential, organ-targeted RNAi may be of use to identify novel gene functions in that organ by functional ablation, analogous to classic tissue-specific inducible knockout models. The extent of ablation observed in the present study may suffice to recognize unknown gene functions, and the efficacy of this novel analytical approach is likely to increase with the advent of more sophisticated RNAi delivery systems. Importantly, gene ablation by RNAi may be induced at any desired age or disease state and would be more rapid and inexpensive than the traditional models.
Possible Side Effect of RNAi Therapy
The cellular machinery of RNAi evolved over millions of years and is the most efficient and versatile mechanism known for specific gene silencing. shRNAs exploit this machinery to mediate therapeutic effects by mimicking the endogenous process and achieve silencing at far lower concentrations than antisense RNAs, but they may disturb cellular miRNA pathways21 and thereby cause hepatotoxicity. When using a cardiotropic rAAV9 serotype with low affinity for the liver, we observed no histological evidence of acute or chronic liver damage. We also studied the cardiac-expressed miRNAs 1 and 133 during RNAi treatment. Because malignant arrhythmias are important complications in HF, deregulation of an arrhythmia-related miRNA such as miRNA-1,15 by a novel treatment should be considered as a possibly serious adverse effect. None of the RNAi vectors changed miRNA-1 levels in NRCMs, but interestingly, shPLB treatment was instead associated with rescue of the miRNA-1 depression induced by phenylephrine in these cells, and rat hearts that underwent shPLB therapy had higher miRNA levels than controls (online-only Data Supplement Figure IVa through IVc). In conjunction with the trend toward improved survival in the AAV-shPLN treatment group, no evidence was thus found for arrhythmogenic side effects.
Although improved contractile function during shPLB therapy results immediately from its influence on excitation-contraction coupling via the cytoplasmic Ca2+ transients, the marked reduction of LV hypertrophy and dilation (Figures 3C, 3D, and 3⇑F) is not deduced as easily, because it involves major reprogramming of the cardiac transcriptome. The RNAi-induced changes of Ca2+ homeostasis may affect not only the cytoplasmic Ca2+ transients but also the separate and insulated Ca2+ signals generated in the perinuclear space36–38 that influence transcription. With respect to hypertrophy, it is of interest that shPLB treatment was associated with rescue of the phenylephrine-induced miRNA-133 depression in NRCMs and higher levels in rat hearts on shPLB treatment (online-only Data Supplement Figures IVb and IVc). The rescue of miRNA-133, which plays a critical role in the control of cardiomyocyte size,13,14 appears to be linked to RNAi at the cell level, because it also occurs in vitro, where hemodynamic stress, neurohumoral, or cytokine activation, as in HF in vivo, cannot play any role.
Although our rationale to measure miRNA expression was to test whether shPLB may “poison” the RNAi machinery, differential miRNA induction can also be viewed as “master switches” that control reinduction of fetal genes during HF.22 A clear distinction between side effect and master-switch aspects of miRNA regulation is very difficult and certainly beyond the scope of the present report. In addition, clarification of the mechanism by which shPLB-RNAi is linked to the observed changes in miRNA-1 and miRNA-133 (passive association or component of the therapeutic process) requires future studies.
The present study demonstrates for the first time the high efficacy of a locally targeted RNAi therapeutic strategy in a cardiac disease. Whereas classic gene therapy rests on recombinant protein expression as its basic principle, RNAi therapy instead uses regulatory RNAs with grossly different targeting, kinetics, and toxicity. We provide evidence that under the precondition of careful vector adaptation to the specific requirements of RNAi, and provided that regulatory RNA with high intrinsic activity and target specificity is selected, RNAi therapy may achieve long-term cardiac benefit without apparent toxicity. When functionally optimized RNAi vectors were used, and aortic root vector injection or cardiotropic rAAV9 capsids were used to target RNAi to the heart, no evidence was found of side effects. Short-term phospholamban silencing led to improved cardiac function 1 month after aortic root injection of an adenoviral RNAi vector. Long-term RNAi after simple intravenous injection of an optimized rAAV9 vector resulted in restored cardiac function and reduction of cardiac dilation, hypertrophy, and fibrosis after a period of 3 months. The rAAV9 approach uses a vector known to target the heart in primates, thus offering potential for clinical translation. Specifically, for targets such as phospholamban in which pharmacological approaches have failed thus far, the RNAi approach may enhance the therapeutic repertoire for cardiac diseases.
Sources of Funding
This work was supported by the Deutsche Forschungsgemeinschaft through grant Po 378/6-1 to Dr Poller and through SFB Transregio 19 projects C5 (Drs Poller and Fechner), C1 (Drs Kurreck and Erdmann), A2 (Drs Westermann and Schultheiss), and Z3 (Dr Tschope). This study was also supported in part by grants from the National Institutes of Health (R01 HL078691, HL071763, HL080498, and HL083156), a Leducq Transatlantic Network grant (Dr Hajjar), and National Institutes of Health grant K01 HL076659 (Dr Lebeche).
Iwanaga Y, Hoshijima M, Gu Y, Iwatate M, Dieterle T, Ikeda Y, Date MO, Chrast J, Matsuzaki M, Peterson KL, Chien KR, Ross J Jr. Chronic phospholamban inhibition prevents progressive cardiac dysfunction and pathological remodeling after infarction in rats. J Clin Invest. 2004; 113: 727–736.
Eizema K, Fechner H, Bezstarosti K, Schneider-Rasp S, van der Laarse A, Wang H, Schultheiss H-P, Poller W, Lamers J. Adenovirus-based phospholamban-antisense-mRNA expression as a novel approach to improve cardiac contractile dysfunction: comparison of a constitutive viral versus an endothelin-1-responsive cardiac promoter. Circulation. 2000; 101: 2193–2199.
Dieterle T, Meyer M, Gu Y, Belke DD, Swanson E, Iwatate M, Hollander J, Peterson KL, Ross J Jr, Dillmann WH. Gene transfer of a phospholamban-targeted antibody improves calcium handling and cardiac function in heart failure. Cardiovasc Res. 2005; 67: 678–688.
Meyer M, Belke DD, Trost SU, Swanson E, Dieterle T, Scott B, Cary SP, Ho P, Bluhm WF, McDonough PM, Silverman GJ, Dillmann WH. A recombinant antibody increases cardiac contractility by mimicking phospholamban phosphorylation. FASEB J. 2004; 18: 1312–1314.
Minamisawa S, Hoshijima M, Chu G, Ward C, Frank K, Gu Y, Martone M, Wang Y, Ross J, Kranias E, Giles W, Chien K. Genetic complementation identifies chronic phospholamban-sarcoplasmic reticulum calcium ATPase interaction as a critical calcium cycling defect in the progression of dilated cardiomyopathy. Cell. 1999; 99: 313–322.
Fechner H, Suckau L, Kurreck J, Sipo I, Wang X, Pinkert S, Loschen S, Rekittke J, Weger S, Dekkers D, Vetter R, Erdmann V, Schultheiss H-P, Paul M, Lamers J, Poller W. Highly efficient and specific modulation of cardiac calcium homeostasis by adenovector-derived short hairpin RNA targeting phospholamban. Gene Ther. 2007; 14: 211–218.
Care A, Catalucci D, Felicetti F, Bonci D, Addario A, Gallo P, Bang ML, Segnalini P, Gu Y, Dalton N, Elia L, Latronico M, Hoydal M, Autore C, Russo MA, Dorn G II, Ellingsen O, Ruiz-Lozano P, Peterson K, Croce C, Peschle C, Condorelli G. MicroRNA-133 controls cardiac hypertrophy. Nat Med. 2007; 13: 613–618.
van Rooij E, Sutherland L, Qi X, Richardson J, Hill J, Olson E. Control of stress-dependent cardiac growth and gene expression by a microRNA. Science. 2007; 316: 575–579.
van Rooij E, Sutherland L, Liu N, Williams A, McAnally J, Gerard R, Richardson J, Olson E. A signature pattern of stress-responsive microRNAs that can evoke cardiac hypertrophy and heart failure. Proc Natl Acad Sci U S A. 2006; 103: 18255–18260.
Hajjar R, Schmidt U, Matsui T, Guerrero J, Lee K-H, Gwathmey J, Dec G, Semigran M, Rosenzweig A. Modulation of ventricular function through gene transfer in vivo. Proc Natl Acad Sci U S A. 1998; 95: 5251–5256.
del Monte F, Lebeche D, Guerrero JL, Tsuji T, Doye AA, Gwathmey JK, Hajjar RJ. Abrogation of ventricular arrhythmias in a model of ischemia and reperfusion by targeting myocardial calcium cycling. Proc Natl Acad Sci U S A. 2004; 101: 5622–5627.
Sakata S, Lebeche D, Sakata N, Sakata Y, Chemaly ER, Liang LF, Tsuji T, Takewa Y, Del Monte F, Peluso R, Zsebo K, Jeong D, Park WJ, Kawase Y, Hajjar R. Restoration of mechanical and energetic function in failing aortic-banded rat hearts by gene transfer of calcium cycling proteins. J Mol Cell Cardiol. 2007; 42: 852–861.
Sakata S, Lebeche D, Sakata Y, Sakata N, Chemaly ER, Liang L, Nakajima-Takenaka C, Tsuji T, Konishi N, del Monte F, Hajjar RJ, Takaki M. Transcoronary gene transfer of SERCA2a increases coronary blood flow and decreases cardiomyocyte size in a type 2 diabetic rat model. Am J Physiol Heart Circ Physiol. 2007; 292: H1204–H1207.
Thum T, Galuppo P, Wolf C, Fiedler J, Kneitz S, van Laake L, Doevendans P, Mummery C, Borlak J, Haverich A, Gross C, Engelhardt S, Ertl G, Bauersachs J. MicroRNAs in the human heart: a clue to fetal gene reprogramming in heart failure. Circulation. 2007; 116: 258–267.
Pačak C, Mah C, Thattaliyath B, Conlon T, Lewis M, Cloutier D, Zolotukhin I, Tarantal A, Byrne B. Recombinant adeno-associated virus serotype 9 leads to preferential cardiac transduction in vivo. Circ Res. 2006; 99: e3–e9.
Champion HC, Georgakopoulos D, Haldar S, Wang L, Wang Y, Kass D. Robust adenoviral and adeno-associated viral gene transfer to the in vivo murine heart: application to study of phospholamban physiology. Circulation. 2003; 108: 2790–2797.
Schmitt J, Kamisago M, Asahi M, Li G, Ahmad F, Mende U, Kranias E, MacLennan D, Seidman J, Seidman C. Dilated cardiomyopathy and heart failure caused by a mutation in phospholamban. Science. 2003; 299: 1410–1413.
Haghighi K, Kolokathis F, Gramolini AO, Waggoner JR, Pater L, Lynch R, Fan G, Tsiapras D, Parekh R, Dorn G II, MacLennan D, Kremastinos D, Kranias E. A mutation in the human phospholamban gene, deleting arginine 14, results in lethal, hereditary cardiomyopathy. Proc Natl Acad Sci U S A. 2006; 103: 1388–1393.
Zhao W, Yuan Q, Qian J, Waggoner JR, Pathak A, Chu G, Mitton B, Sun X, Jin J, Braz J, Hahn H, Marreez Y, Syed F, Pollesello P, Annila A, Wang H, Schultz Jel J, Molkentin J, Liggett S, Dorn G II, Kranias E. The presence of Lys27 instead of Asn27 in human phospholamban promotes sarcoplasmic reticulum Ca2+-ATPase superinhibition and cardiac remodeling. Circulation. 2006; 113: 995–1004.
Kockskämper J, Seidlmayer L, Walther S, Hellenkamp K, Maier L, Pieske B. Endothelin-1 enhances nuclear Ca2+ transients in atrial myocytes through Ins(1,4,5)P3-dependent Ca2+ release from perinuclear Ca2+ stores. J Cell Sci. 2008; 121: 186–195.
RNA interference (RNAi) has the potential to be a novel therapeutic strategy in diverse areas of medicine. Whereas gene therapy (already used for cardiac disorders) rests on recombinant protein expression as its basic principle, RNAi therapy instead uses regulatory RNAs to achieve its effect. Fundamental limitations of the use of chemically synthesized small interfering RNAs to mediate RNAi are their rapid degradation in plasma and target cells and the unsolved problem of adequate targeting in vivo. The present study shows for the first time the high efficacy of an RNAi therapeutic strategy in a cardiac disease in vivo. It demonstrates successful treatment of heart failure in a rat model by cardiac-targeted RNAi ablating phospholamban, a key regulator of cardiac Ca2+ homeostasis. A novel vector was developed based on a cardiotropic adeno-associated virus (rAAV9) that carries RNAi activity to the heart after intravenous injection. Over a period of 3 months, this therapy restored cardiac function, reversed cardiac dilation and hypertrophy, and reduced cardiac fibrosis. In recent years, adeno-associated virus–based vectors have overcome key challenges to gene therapy, such as stability, safety, and host immune response. The present study shows that under the precondition of careful adeno-associated virus vector adaptation to the specific requirements of RNAi, if regulatory RNA sequences with high intrinsic activity and target specificity are selected, they may also serve as valuable tools for cardiac RNAi therapy and offer clinical potential. Specifically, for targets such as phospholamban, for which pharmacological approaches have failed so far, the RNAi approach may enhance the therapeutic repertoire.
↵*The last 2 authors contributed equally to this article.