Allele-Specific Silencing of Mutant Myh6 Transcripts in Mice Suppresses Hypertrophic Cardiomyopathy

See allHide authors and affiliations

Science  04 Oct 2013:
Vol. 342, Issue 6154, pp. 111-114
DOI: 10.1126/science.1236921

Silencing a Silent Killer

Hypertrophic cardiomyopathy (HCM) is a leading cause of sudden death in young athletes. HCM is caused by dominant mutations in genes encoding constituents of the cardiac sarcomere, the contractile unit that keeps the heart pumping. Studying a mouse model that recapitulates a severe form of HCM caused by a mutation in a β myosin heavy chain gene, Jiang et al. (p. 111) investigated whether sarcomere dysfunction could be corrected by selectively silencing expression of the mutant allele. Mice treated shortly after birth with a viral vector encoding an appropriately designed RNA interference cassette did not develop cardiac hypertrophy or myocardial fibrosis—the pathologic manifestations of HCM—for at least 6 months.


Dominant mutations in sarcomere proteins such as the myosin heavy chains (MHC) are the leading genetic causes of human hypertrophic cardiomyopathy (HCM) and dilated cardiomyopathy. We found that expression of the HCM-causing cardiac MHC gene (Myh6) R403Q mutation in mice can be selectively silenced by an RNA interference (RNAi) cassette delivered by an adeno-associated virus vector. RNAi-transduced MHC403/+ mice developed neither hypertrophy nor myocardial fibrosis, the pathologic manifestations of HCM, for at least 6 months. Because inhibition of HCM was achieved by only a 25% reduction in the levels of the mutant transcripts, we suggest that the variable clinical phenotype in HCM patients reflects allele-specific expression and that partial silencing of mutant transcripts may have therapeutic benefit.

Hypertrophic cardiomyopathy (HCM) is an autosomal dominant disease characterized by an increase in left ventricular wall thickness (LVWT), disorganization of cardiomyocytes, and expansion of myocardial fibrosis that occurs in the absence of systemic disease (13). HCM is the leading cause of nonviolent sudden death in young adults and the most common cause of sudden death on the athletic field (4). HCM is caused by mutations in genes that encode protein constituents of the cardiac sarcomere, the contractile unit of muscle (5, 6). More than 1000 distinct pathogenic mutations have been identified, and more than half of these occur in MYH7 (encoding β myosin heavy chain) and MYBPC3 (encoding cardiac myosin binding protein C) (7). Most HCM mutations, and all that occur in MYH7, are missense mutations, producing amino acid substitutions in myosin that perturb the sarcomere’s contractile function.

The human HCM-causing mutation MYH7 R403Q (Arg403 → Gln) causes particularly severe disease that is characterized by early-onset and progressive myocardial dysfunction, with a high incidence of sudden cardiac death (8). Heterozygous MHC403/+ mice express the R403Q mutation in Myh6 under the control of the endogenous Myh locus. Myh6 and MYH7 are highly homologous in sequence and encode the predominant myosin isoforms in the adult hearts. MHC403/+ mice recapitulate human HCM and develop hypertrophy, myocyte disarray, and increased myocardial fibrosis (9). Analyses of mutant myosins isolated from MHC403/+ mice showed that HCM mutations cause fundamental changes in sarcomere functions, including increased acto-myosin sliding velocity, force generation, and adenosine triphosphate hydrolysis (10). These changes in turn alter calcium cycling and gene transcription in myocytes and ultimately induce pathologic remodeling of the heart in vivo (1113). Understanding this pathogenic cascade has led to the identification of secondary signaling molecules as potential therapeutic targets (13, 14), but no strategies have been defined that correct the primary biophysical and biochemical abnormalities of sarcomeres with HCM mutations.

Selective reduction in the expression of the mutant protein would be the most direct approach for correcting sarcomere dysfunction. As a first step in pursuing this strategy, we determined whether in vivo allele-specific repression of Myh6 R403Q was feasible. Because mice hemizygous for a normal Myh6 gene are viable, fertile, and have essentially normal cardiac function (15), we reasoned that inactivation of the mutant sarcomere protein allele is unlikely to have adverse cardiovascular effects. We used an RNA interference (RNAi) construct because this powerful tool has successfully reduced gene expression in many systems and can distinguish between genes that differ by a single nucleotide (16). We selected adeno-associated virus packed with serotype 9 capsid (AAV-9) as a delivery vehicle because this vector has strong tropism for cardiac tissues (17, 18). To enhance the cardiac tropism, we engineered the vector so that AAV-9 expression was under the control of the cardiac-specific troponin T (cTnT) promoter.

We produced 17 unique RNAi constructs, each cotransfected with a plasmid carrying the Myh6 R403Q mutant gene into 293T human embryonic kidney cells (fig. S1A). One RNAi construct, designated 403m, significantly reduced Myh6 R403Q expression (Fig. 1, A and B). To assess its specificity, we transfected wild-type or mutant Myh6 into 293T cells with 403m constructs. Because there was significant silencing (~80%) of both wild-type and mutant Myh6 expression, we introduced an additional mismatch into the 403m RNAi construct (designated 403i; Fig. 1A). The 403i construct had modest reduction (~20%) of wild-type Myh6 expression but retained ~80% reduction in the expression of Myh6 R403Q transcript in 293T cells (Fig. 1B).

Fig. 1 Selective silencing of Myh6 R403Q expression by AAV-9–mediated RNAi.

(A) Schematic representation of FVB, 129SvEv, and 129SvEv mutant (R403Q) transcript and RNAi sequences. (B) Quantitative real-time polymerase chain reaction (PCR) analysis of wild-type Myh6 (white bar) and mutant Myh6 R403Q (black bar) expression after transduction of the 403m and 403i constructs (n = 4). Levels of the transcripts were normalized to control LacZ RNAi. (C) Quantitative real-time PCR analysis of FVB Myh6 (white bar) and 129SvEv Myh6 R403Q (black bar) expression after transduction of the 129i construct (n = 4). Levels of the transcripts were normalized to control LacZ RNAi. Data are means ± SD.

To ascertain the cardiac selectivity of AAV-9–cTnT vector, we used enhanced green fluorescent protein (EGFP) (fig. S1B). Virus was injected [5 × 1013 vector genomes (vg)/kg] into the thoracic cavity of 1-day-old mice (see supplementary materials) and after 3 weeks, all organs were dissected and EGFP expression was assessed by fluorescence microscopy. EGFP expression occurred exclusively in the heart and was absent in other organs, including the brain, lung, and spleen (fig. S2). EGFP expression was present within 48 hours after virus transduction and remained robust for 12 months (figs. S2 and S3).

We next engineered 403i and control shRNAs (short hairpin RNAs), respectively designated 403i RNAi and control RNAi, into the AAV-9–cTnT-EGFP RNAi vector so that all cells expressing EGFP would also express shRNAs. To assess the efficacy of 403i shRNA in vivo, we injected variable amounts of 403i RNAi–encoding viruses (5 × 109, 5 × 1011, and 5 × 1013 vg/kg) into the thoracic cavity of 1-day-old mice. Two weeks after viral transduction, total RNA extracted from each left ventricle (LV) was individually analyzed by RNA-seq (19). Sequencing reads that corresponded to Myh6 R403Q or wild-type Myh6 were counted and visualized using Integrative Genomics Viewer (Broad Institute, Cambridge, MA). The expression of Myh6 was comparable in LV tissues after transduction with control RNAi (12,118 reads per million transcripts) and 403i RNAi (11,675 reads per million transcripts), indicating that the wild-type allele was not silenced in vivo. In contrast, the ratio of Myh6 R403Q reads to Myh6 wild-type reads varied between 403i RNAi titers. Only the highest titer (5 × 1013 vg/kg) resulted in a significant reduction (28.5%) in the relative expression of Myh6 R403Q compared to wild-type Myh6 transcripts (P = 2.5 × 10–5) (fig. S1C).

To assess the impact of silencing Myh6 R403Q on HCM development, we injected virus encoding the 403i RNAi cassette (n = 8) or control RNAi cassette (n = 7) into the thoracic cavity of 1-day-old male MHC403/+ mice. At 5 to 6 weeks of age, all mice were given cyclosporine A (CsA) for 3 weeks to accelerate the emergence of HCM histopathology, as described (13). Mice were serially evaluated by echocardiography; after killing, hearts were analyzed by histopathology. After CsA treatment, control RNAi–transduced mice had LV hypertrophy and severe HCM histopathology (Fig. 2A), similar to nontransduced, CsA-treated MHC403/+ mice (13). In contrast, CsA-treated MHC403/+ mice transduced with 403i RNAi did not develop HCM (Table 1). The LVWT of 403i RNAi–transduced mice (0.84 ± 0.10 mm) was significantly less than that of mice transduced with control RNAi (1.52 ± 0.25 mm, P = 1.9 × 10–5) and comparable to the LVWT of wild-type mice (0.74 ± 0.05 mm, n.s.) (12). Myocardial disarray (Fig. 2B) was absent, and fibrosis (Fig. 2C) was significantly reduced, in 403i RNAi–transduced mice (0.43 ± 0.11%) relative to control RNAi–transduced hypertrophic MHC403/+ mice (2.12 ± 0.57%, P = 0.003). QRS interval prolongation, an electrocardiographic feature of LV hypertrophy, was present in mice transduced with control RNAi (20.5 ± 1.2 ms) but not in mice transduced with 403i RNAi (16.9 ± 1.4 ms, P = 0.001) (Fig. 2D). Additionally, the expression of prototypic LV hypertrophy markers Nppa and Nppb in mice transduced with control RNAi was higher than in mice transduced with 403i RNAi by a factor of 2.5 (Fig. 2E).

Fig. 2 In vivo effect of Myh6 R403Q silencing.

(A) Cardiac histopathology from MHC403/+ mice transduced with control RNAi (left) and 403i RNAi (right). Masson trichrome staining reveals marked fibrosis (blue) in MHC403/+ mice transduced with control RNAi. Scale bar, 1 mm. (B) Hematoxylin and eosin staining shows myocyte disarray in MHC403/+ mice transduced with control RNAi (left) and normal myocyte architecture in mice transduced with 403i RNAi (right). Scale bar, 100 μm. (C) Quantification of myocardial fibrosis in MHC403/+ mice transduced with control RNAi (black bar, n = 4) and 403i RNAi (white bar, n = 4). (D) Representative electrocardiograms of MHC403/+ mice transduced with control RNAi (left) and 403i RNAi (right). Mice transduced with control RNAi have prolonged QRS (ventricular conduction) interval and high-voltage P waves consistent with LV hypertrophy and atrial enlargement. Measurements of QRS intervals from mice transduced with control RNAi (black bar, n = 5) and 403i RNAi (white bar, n = 6) are shown. (E) Quantitative real-time PCR analysis of Nppa (left) and Nppb (right) expression after transduction of control RNAi (black bar) and two different doses of 403i constructs (white bar) (n = 5). Levels of the transcripts were normalized to transcript levels from age-matched wild-type hearts. Data are means ± SD.

Table 1 RNAi effects on cardiac morphology and function in HCM mice.

To accelerate hypertrophic remodeling in MHC403/+ mice, we administered CsA for the number of weeks indicated either after RNAi transduction on day 1 (post) or for 3 weeks before RNAi transduction on day 42 (pre). Age denotes time of cardiac evaluation; LVDD, left ventricular diastolic dimension; FS, fractional shortening. Cardiac dimensions and function with associated P values, calculated by t test, reflect comparisons to MHC403/+ transduced with control RNAi. Values for wild-type 129SvEv mice not treated with CsA are shown for comparison. Data are means ± SD.

View this table:

To assess whether the early age at transduction and/or viral titer influenced HCM development, we injected high-titer (5 × 1013 vg/kg) and low-titer (5 × 1012 vg/kg) viruses of 403i RNAi into 3-week-old MHC403/+ mice (n = 5). At 4 weeks, mice were treated with CsA for three additional weeks, followed by echocardiography to assess LVWT and diastolic (relaxation) performance [left atrial diameter normalized to the aortic root diameter (20)], which becomes abnormal early in HCM (21). Mice transduced with high viral titers of control RNAi or low viral titers of 403i RNAi had both LV hypertrophy and diastolic dysfunction (Table 2). In contrast, mice transduced with high-titer 403i RNAi virus had neither hypertrophy (LVWT = 0.72 ± 0.05 mm, P = 1.9 × 10–6 compared to control RNAi) nor diastolic dysfunction (left atria dimension normalized to the aortic root = 1.17 ± 0.09, P = 0.009 compared to control RNAi).

Table 2 Viral dosage needed for RNAi effects on cardiac morphology and function in HCM mice.

MHC403/+ mice were transduced with high-titer (5 × 1013 vg/kg) and low-titer (5 × 1012 vg/kg) RNAi on day 1 and treated with CsA for 3 weeks to accelerate hypertrophic remodeling. Left atria (LA) dimensions normalized to the aortic root (Ao) are provided. Cardiac dimensions and function with associated P values, calculated by t test, reflect comparisons to MHC403/+ transduced with control RNAi. Data are means ± SD.

View this table:

Using the high-titer virus, we next investigated whether 403i RNAi transduction could alter established HCM by pretreating MHC403/+ mice with CsA for 3 weeks to induce hypertrophy (LVWT = 1.40 ± 0.11 mm) before viral transduction. Echocardiography assessments at 2 months after 403i RNAi (n = 3) transduction showed no change in LVWT (Table 1), which suggests that the treatment was ineffective in reversing established disease.

To determine whether 403i RNAi transduction affected the pathologic LV remodeling that slowly emerges in MHC403/+ mice with age in the absence of CsA, we monitored LV hypertrophy in mice transduced with a single high-titer dose of 403i RNAi (n = 5) or control RNAi (n = 6) during the first day of life. At 6 months, mice transduced with control RNAi had LV hypertrophy (LVWT = 0.93 ± 0.11 mm). There was no LV hypertrophy in mice transduced with 403i RNAi (LVWT = 0.68 ± 0.09 mm, P = 0.005), and LVWT was indistinguishable from that of wild-type mice (LVWT = 0.74 ± 0.05 mm, n.s.).

The protective effect of the 403i RNAi vector dissipated over time. LV hypertrophy emerged in 403i RNAi–transduced mice by 11 months of age (LVWT = 0.87 ± 0.11 mm) and was comparable to that observed in control RNAi–transduced mice. The inability to fully suppress hypertrophy in MHC403/+ mice during later life is presumably due to diminished AAV-mediated transgene expression that occurs 7 months after transduction (22) and/or subtherapeutic 403i RNAi levels.

Finally, we considered whether a single RNAi might silence different patient-specific mutations in the same gene by targeting a nearby single-nucleotide polymorphism (SNP) that distinguished the mutant from wild-type alleles. To test this model, we produced male F1 offspring from 129SvEv MHC403/+ and wild-type FVB crosses. We constructed an RNAi that targeted a 129SvEv SNP on the Mhy6 allele (designated 129i; Fig. 1A) and transfected this with Mhy6 R403Q (129SvEv) or wild-type Myh6 (FVB) plasmids into 293T cells. The 129i RNAi decreased Mhy6 R403Q levels by 75% and reduced wild-type Myh6 (FVB) by only 15% (Fig. 1C). We produced AAV-9–cTnT-EGFP-129i virus and transduced (5 × 1013 vg/kg) 1-day-old male F1 MHC403/+ mice with 129i RNAi (n = 4) or control RNAi (n = 5). At 4 weeks of age, mice were treated with CsA for 2 weeks and studied by echocardiography. Control RNAi–transduced mice developed LV hypertrophy (LVWT = 1.37 ± 0.03 mm), whereas MHC403/+ mice transduced with 129i RNAi did not (LVWT = 0.73 ± 0.03 mm; P = 1.6 × 10–6) (Table 1). We conclude from these studies that one RNAi construct targeting a SNP that demarcates mutant and wild-type alleles could be used to silence distinct HCM mutations in a gene or to augment mutation-specific RNAi.

AAV-9 mediated RNAi preferentially suppresses the expression of the Myh6 R403Q allele in a mouse model of HCM by directly targeting the mutation or a nearby SNP. It is noteworthy that in a mouse model characterized by accelerated onset and severity of HCM, reduction in the expression levels of the mutant allele by only 28.5% (fig. S1C) was sufficient to abrogate hypertrophy and histopathologic remodeling for several months. Indeed, the LVWT in 403i-treated MHC403/+ mice, with or without CsA, was comparable to that in wild-type mice. On the basis of these findings and evidence for unequal expression of mutated and wild-type MYH7 mRNAs in human HCM hearts (23), we suggest that variable penetrance and severity of HCM may reflect, at least in part, the proportion of wild-type and mutant transcripts and proteins.

The capacity for RNAi to attenuate the expression of Myh6 R403Q transcripts and abrogate HCM in mice raises the possibility that mutation silencing might benefit human cardiomyopathy patients. Although our study does not address the many important potential problems associated with viral-mediated gene therapy (including the potential for immune response and long-term off-target effects), some AAV protocols are sufficiently safe for human clinical trials (24). Moreover, the potential development of RNAi molecules that target allele-specific common SNPs (rather than each patient’s specific mutation) could be a way to avoid the daunting challenge of producing the thousands of RNAi molecules that would be required to silence each unique HCM or dilated cardiomyopathy (DCM) mutation. There are also new opportunities to deliver gene silencing therapies to the heart. For example, the septal perforating artery can be selectively cannulated to target interventions to the interventricular septum (25), which in the majority of patients is the most profoundly hypertrophied segment in the HCM heart and can cause obstructed ventricular outflow of blood, increasing the risk of sudden death. Adaptation of this approach could facilitate prophylactic, localized silencing of HCM mutations in this region, where its benefit could be measurable and clinically meaningful. With continued advances in viral and other delivery systems (26, 27), allele-selective silencing holds considerable potential to retard the onset and progression of HCM and other genetic cardiomyopathies.

Supplementary Materials

Materials and Methods

Figs. S1 to S4

References and Notes

  1. Acknowledgments: We thank J. Gorham, D. Conner, and S. DePalma for technical and bioinformatic assistance, and C. Cepko for assistance with viral protocols. Supported by NIH grants U01HL098166 and R01HL084553 (J.G.S. and C.E.S.) and the Howard Hughes Medical Institute (C.E.S.). The authors (J.J., C.E.S., J.G.S., and H.W.) and Harvard Medical School plan a patent application related to the use of short hairpin RNAs as therapies for human genetic cardiomyopathies. C.E.S. and J.G.S. are founders of and own shares in MyoKardia, a biotechnology company developing small molecules that target the sarcomere for treatment of inherited cardiomyopathy.
View Abstract

Navigate This Article