MicroRNA-208a is a regulator of cardiac hypertrophy and conduction in mice (original) (raw)
Cardiac myosin heavy chain genes encode intronic miRNAs. During development, the αMHC and βMHC isoforms are expressed in a developmental stage-specific manner (18). In mouse hearts, the slower isoform βMHC is fetal specific, while the faster isoform αMHC becomes the predominant isoform in the adult heart. In humans and other large mammals, βMHC expression continues into adulthood. However, increased βMHC expression is a common feature of cardiac hypertrophy and heart failure in both mouse and human hearts (19, 20). An intron from each of the Myh6 and Myh7 genes host a conserved miRNA — miR-208a and miR-208b, respectively (Figure 1A). miR-208a expression was detected specifically in the adult mouse heart (Figure 1B) and could be detected at very low levels in the heart as early as E13.5 (Figure 1C). The switch from fetal isoform βMHC to the adult isoform αMHC in the mouse occurs shortly after birth (Figure 1D). We found that a similar switch from miR-208b to miR-208a expression also occurs, suggesting they are cotranscribed with their MHC host genes (Figure 1D). miR-208a and miR-208b are of similar sequence with identical seed regions (Figure 1A), which suggests they might be functionally redundant (21). However, miR-208b was not detectable in the adult heart, indicating that if miR-208a and miR-208b target the same mRNAs, they do so at different developmental stages.
Expression of miR-208a and miR-208b parallels the expression of their respective host genes Myh6 and Myh7. (A) miR-208a is encoded by intron 29 of the Myh6 gene, while miR-208b is encoded by intron 31 of the Myh7 gene. miR-208a and miR-208b are highly conserved and share similar sequence identity (indicated by asterisks). (B) Detection of mature and precursor miR-208a in adult mouse tissues. Sk. muscle, skeletal muscle; tRNA, transfer RNA. (C) Detection of mature and precursor miR-208a in E13.5, E16.5, and neonatal tissues using Northern blot analysis. (D) Top left: αMHC and βMHC transcripts were detected in E16.5, P0, P5, P10, and adult mouse hearts using RT-PCR. Bottom left: miR-208a and miR-208b expression was detected in the samples using Northern analysis. Right: Relative levels of miR-208a and miR-208b during heart development (E) Top left: αMHC and βMHC transcripts were detected using RT-PCR in isolated rat neonatal cardiomyocytes following treatment with thyroid hormone (T3). Bottom left: miR-208a and miR-208b were detected using Northern analysis. Right: Quantitative analysis. Fold change is relative to no T3 treatment (which was set at 1). *P < 0.01, compared with no T3 treatment.
Thyroid hormone signaling is a well-known regulator of αMHC and βMHC transcription (22). A surge of circulating thyroid hormone that occurs shortly after birth represses βMHC expression and activates αMHC expression though negative and positive cis-acting elements within their respective promoters. We treated isolated rat cardiomyocytes with thyroid hormone and observed reduced βMHC/miR-208b expression while dramatically inducing αMHC/miR-208a expression (Figure 1E). Together, those data suggest that the intronic miR-208 family and their MHC host genes are co-expressed and regulated by common transcriptional events and signaling pathways.
Cardiac overexpression of miR-208a is sufficient to induce cardiac hypertrophy. In an effort to understand the function of miR-208a in the adult heart, we overexpressed miR-208a specifically in the heart under the control of the α-myosin heavy chain (α_MHC_) promoter using a bigenic system. An advantage of this strategy is that miR-208a is overexpressed specifically at the same time and place it would normally be expressed. The overexpression strategy consisted of a transgene encoding miR-208a downstream of a tetracycline-responsive promoter (TRE–miR-208a) and a second transgene encoding the tetracycline-controlled transactivator protein driven by the αMHC promoter (α_MHC-tTA_) (23). Using this system, we found that cardiac-specific overexpression of miR-208a did not cause embryonic lethality, and thus administration of tetracycline to delay transgene expression was unnecessary. Multiple founder TRE–miR-208a Tg lines were established. Primary analyses indicated that miR-208a was overexpressed at similar levels; therefore we combined results from the studies of different Tg lines. Throughout our studies, we compared heterozygous mice carrying the α_MHC-tTA_ and TRE–miR-208a transgenes (referred to hereafter as “miR-208a Tg mice”) with mice heterozygous for αMHC-tTA (referred to hereafter as “control mice”).
Northern blot analysis showed miR-208a levels were approximately 4-fold higher in miR-208a Tg hearts compared with control hearts (Figure 2A). In situ hybridization using a DIG-labeled miR-208a probe confirmed that miR-208a was uniformly overexpressed in most cardiomyocytes (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI36154DS1). The gross heart morphology of 4-month-old miR-208a Tg hearts was dramatically larger relative to control littermates (Figure 2B). Accordingly, the heart weight to body weight ratios of miR-208a Tg animals were significantly higher than in control animals (Figure 2C). Histological sectioning and H&E staining revealed the appearance of enlarged chambers and thickened ventricular walls in the miR-208a Tg hearts, suggestive of hypertrophic growth (Figure 2D and Supplemental Figure 2). Analysis of desmin, an intermediate filament found near the sarcomeric Z line, revealed no changes in the integrity in the sarcomeric structure of miR-208a Tg cardiomyocytes (Figure 2E). Quantitative measurement of miR-208a Tg cardiomyocytes revealed a 52% increase in cell size relative to controls (Figure 2F). The distribution of the cell size measurements from control and Tg cardiomyocytes clustered around distinct peaks, indicating that the hypertrophy in miR-208a Tg heart is fairly uniform (Figure 2G). Together, these results indicate that miR-208a overexpression in the mouse heart induced hypertrophic growth.
Hearts of miR-208a Tg mice undergo hypertrophic growth. (A) Left: Northern blot analysis showing an approximately 4-fold increase of miR-208a expression in hearts of miR-208a Tg animals compared with control littermates. U6 served as loading control. Right: Quantitative analysis of fold change in miR-208a expression compared with control animals. *P < 0.001. (B) Gross morphology of miR-208a Tg hearts was enlarged compared with control hearts. Scale bar: 1 mm. (C) Heart weight to body weight ratios (Hw/Bw) of 4-month-old miR-208a Tg mice (n = 22) were significantly higher than controls (n = 19) (P < 7 × 10–7). (D) Macroscopic view of H&E-stained histological sections (upper, sagittal; lower, transverse) from control and miR-208a Tg hearts. Scale bars: 1 mm. (E) Sarcomeric structure of cardiomyocytes visualized by desmin staining of histological sections. Original magnification, ×200. (F) Histological sections were stained with wheat germ agglutinin–TRITC conjugate to determine cell size. Mean cardiomyocyte size of miR-208a Tg hearts (n = 930) was significantly larger than control cardiomyocytes (n = 926) (*P < 9 × 10–50). Original magnification, ×200. (G) Distribution of control and miR-208a Tg cardiomyocyte cell area measurements were compared. (H) Representative M-mode echocardiographs from conscious control and miR-208a Tg mice. IVSTD, interventricular septal thickness in diastole; IVSTS, interventricular septal thickness in systole; PWTD, posterior wall thickness in diastole; PWTS, posterior wall thickness in systole; LVEDD, left ventricular end-diastolic dimension; LVESD, left ventricular end-systolic dimension.
Cardiac hypertrophy is the heart’s primary response to stress caused by pathological and physiological hemodynamic overload, abnormal hormonal signaling, and certain inherited disorders involving particular transcription factors and contractile proteins (19). Hypertrophic growth involves enhanced protein synthesis, increased sarcomeric density, and increased cardiomyocyte size that culminates into structural remodeling of the heart. Although cardiac hypertrophy is considered an adaptive mechanism to sustain cardiac output, prolonged pathological hypertrophy has adverse consequences associated with heart failure and sudden death (24). Analysis of cardiac function by echocardiography on 3-month-old animals revealed that miR-208a Tg hearts displayed thickening of the ventricular walls (anterior wall in diastole and systole, posterior wall in diastole), an increase in left ventricular diameter (left ventricular diameter in diastole and systole), and deterioration in cardiac function, as indicated by decreased fractional shortening (Figure 2H and Table 1). We also measured cardiac function in 7-month-old animals and obtained similar results (Supplemental Table 1).
Echocardiography of dimensions and function of miR-208a Tg mice
A molecular hallmark of cardiac hypertrophy is the upregulation of βMHC and the cardiac hormone atrial natriuretic factor (ANF) in the adult heart (19, 20). Consistent with hypertrophic growth, we observed increased expression of βMHC transcripts and proteins, by real-time PCR and Western blot analyses, respectively, in miR-208a Tg hearts (Figure 3, A and B). Unexpectedly, no significant changes in ANF transcript levels were detected (Figure 3A).
miR-208a overexpression induces hypertrophic gene expression. (A) Transcripts for αMHC, βMHC, and ANF were detected by real-time PCR in 4-month-old hearts from control and miR-208a Tg mice (n = 5 per genotype). Data are the mean fold change in expression ± SEM. *P < 0.01. (B) Left: Western blot analysis of total MHC and βMHC protein levels in adult control and miR-208a Tg hearts. Right: Quantitative analysis of fold change in protein levels. *P < 0.01. (C) Top left: RT-PCR was used to detect αMHC, βMHC, and ANF transcripts in wild-type hearts following 3 weeks thoracic aortic banding (TAB) or in surgical sham hearts, which were used as controls. Bottom left: miR-208a and miR-208b were detected by Northern blot analysis. U6 served as a loading control. Right: Quantitative analysis of fold change in expression of mRNAs and miRNAs. *P < 0.01. (D) Northern blot and quantitative analysis of expression of miRNAs using hearts from control and miR-208a Tg mice. (E–H) Isolated rat neonatal cardiomyocytes were transduced with miR-208a and control adenoviruses (Ad-208 and Ad-Cntl, respectively) or transfected with oligonucleotides antisense to miR-208a or control oligonucleotides (2′Ome-208a and 2′Ome-Cntl, respectively). (E) Cardiomyocytes were stained for α-actinin or βMHC proteins. Original magnification, ×200. (F) Fold change in mean cell area ± SEM of α-actinin–immunostained cardiomyocytes were treated with adenoviruses or oligonucleotides (n = 100 cells/condition; **P < 4 × 10–12). (G) Fold change in mean fluorescent intensity ± SEM of βMHC-immunostained cardiomyocytes treated with adenoviruses or oligonucleotides (n = 100 cells/condition; #P < 3 × 10–7). (H) Cardiomyocytes were treated with adenoviruses or antisense 2′O-methyl oligonucleotides and were scored for ANF staining (n ≈ 425 cells/condition).
Changes in the expression levels of specific miRNAs have been reported in diseased human hearts and in animal models of heart disease, pointing to their potential roles in cardiomyopathies (15, 16, 25–27). Since βMHC expression is a hallmark of cardiac hypertrophy and because βMHC and miR-208b appear to be co-regulated, we surmised that miR-208b expression would also increase during cardiac hypertrophy. Using a mouse model of cardiac hypertrophy in which the aorta was surgically constricted to produce chronic pressure overload, we indeed found miR-208b expression induced during hypertrophic growth (Figure 3C). Furthermore, miR-208b expression was found induced in adult miR-208a Tg hearts (Figure 3D). We also analyzed the expression of miRNAs whose expression levels have been reported altered in cardiac hypertrophy. While previous studies reported decreased miR-1, miR-133, and miR-29a expression levels and increased miR-125b expression levels in cardiac hypertrophy (15, 16, 25–27), in our study the expression levels of those miRNAs were not significantly affected in miR-208a Tg hearts relative to control hearts (Figure 3D). Together, those data demonstrate that miR-208a induced hypertrophic growth without affecting all aspects of the hypertrophic growth response pathway.
We also determined whether the effects of miR-208a overexpression on hypertrophy could be recapitulated in vitro using isolated rat neonatal cardiomyocytes. Cardiomyocytes were transduced with miR-208a–expressing or control adenoviruses, then immunostained for α-actinin or βMHC (Figure 3E). Consistent with the role of miR-208a in the induction of cardiac hypertrophy in vivo, overexpression of miR-208a in isolated cardiomyocytes increased cell size and βMHC expression but did not significantly affect ANF levels (Figure 3, E–H). Conversely, knockdown of miR-208a by introducing chemically modified oligonucleotides (2′O-methyl modified) into isolated cardiomyocytes resulted in decreased βMHC expression, but the size of cardiomyocytes and ANF expression were not affected (Figure 3, E–H). Taken together, these results using in vivo and in vitro strategies suggest that miR-208a influences a subset of genes important in cell growth rather than activating a broader hypertrophic pathway.
Spatial distribution of βMHC in miR-208a Tg hearts is focal. Increased βMHC expression during cardiac hypertrophy is a well-established phenomenon and is thought to contribute to the overall poor functioning of the hypertrophic heart (28–30). To better assess the effects of miR-208a on the expression of βMHC, we employed a mouse strain harboring a βMHC indicator allele, in which the yellow fluorescent protein (YFP) sequence is fused to the Myh7 gene (31). We bred this allele into the miR-208a Tg line. The YFP-βMHC was highly expressed in neonatal cardiomyocytes, but was essentially unexpressed in adult hearts (ref. 31, Figure 4A, and data not shown). We observed dramatically increased YFP-βMHC protein levels in the miR-208a Tg hearts (Figure 4A and Supplemental Figure 3). However, YFP-βMHC expression did not increase in all cardiomyocytes of miR-208a Tg hearts. Rather it was intensely upregulated only in areas associated with interstitial fibrosis (Figure 4, B and C, and Supplemental Figure 3). This observation is consistent with an earlier report in which the distribution of βMHC was analyzed and found correlated with fibrosis in an established mouse model of cardiac hypertrophy (31). Inhibition of thyroid hormone synthesis by propylthiouracil (PTU) results in hypothyroidism and increases βMHC expression due to loss of thyroid hormone–mediated repression at the βMHC promoter (22). PTU treatment of adult miR-208a Tg mice resulted in uniform upregulation of YFP-βMHC expression throughout the myocardium after 4 weeks, indicating that expression of the βMHC indicator allele recapitulates βMHC expression (Supplemental Figure 4). Thus, even though in situ hybridization analyses for mature miR-208a revealed uniform overexpression in miR-208a Tg cardiomyocytes (Supplemental Figure 1), βMHC re-expression occurred only in a subset of cardiomyocytes associated with fibrosis.
Distribution of YFP-βMHC fusion protein in miR-208a Tg hearts. (A) Confocal fluorescent images of coronal sections from control and miR-208a Tg hearts. Overlapping images were stitched together using ImageJ. Original magnification, ×40. (B) Papillary muscle from control and miR-208a Tg hearts imaged for YFP-βMHC (green) expression and wheat germ agglutinin–TRITC staining (red). Original magnification, ×200. (C) Representative fluorescent images of YFP-βMHC expression (green) in an area of interstitial fibrosis (red) in miR-208a Tg hearts. Original magnification, ×200. (D) Mean cell area ± SEM of cardiomyocytes from miR-208a Tg; YFP-βMHC and control; YFP-βMHC hearts. Cells measured for area were also scored for presence or absence of YFP-βMHC expression (n = 100/genotype; *P < 0.001).
We next tested whether βMHC re-expression correlated with the miR-208a–induced hypertrophic growth of individual cardiomyocytes. To do this, we compared the cell areas of miR-208a Tg;YFP-βMHC cardiomyocytes with the areas of control YFP-βMHC cardiomyocytes lacking the miR-208a transgene. No association between the state of YFP-βMHC expression and cell area increase was observed (Figure 4D). Thus, cardiomyocytes from miR-208a Tg hearts were significantly larger than the cardiomyocytes from control hearts independently of whether they were positive or negative for YFP-βMHC expression (Figure 4D). Taken together, these observations demonstrate that βMHC expression is not an obligate component of miR-208a–induced hypertrophic growth and that overexpression of miR-208a alone is sufficient to induce hypertrophic growth in cardiomyocytes even when they show no re-activation of Myh7 gene expression.
Targeted deletion of miR-208a alters cardiac gene expression. Having demonstrated that miR-208a is sufficient for hypertrophic growth, we sought to examine miR-208a function using a loss-of-function approach in miR-208a knockout mice. We replaced the genomic sequence encoding miR-208a by homologous recombination with a neomycin selection cassette flanked by loxP sites (Figure 5, A and B). The selection cassette was subsequently excised by Cre-mediated recombination, leaving only a small footprint of exogenous DNA in place of miR-208a (Figure 5C). Since miR-208a is located within an intron of the Myh6 gene, we confirmed that the splicing pattern of the αMHC transcript was unaffected by the mutant miR-208a allele (Supplemental Figure 5).
Expression of βMHC is decreased in Mir208a–/– hearts. (A) Strategy to delete miR-208a from intron 31 of the Myh6 gene by homologous recombination. The miR-208a coding sequence (green box) was replaced by a neomycin selection cassette flanked by loxP sites (red triangles). The selection cassette was excised from the germline by mating with mice that ubiquitously expressed Cre recombinase, thus creating a mutant allele that contained a single loxP sequence in place of miR-208a. Genotyping PCR primers and 5′ probes are indicated. NdeI, restriction enzyme site; P1, primer binding site 1. (B) The occurrence of the intended recombination event in mouse ES cells was confirmed by PCR and Southern blot analyses. (C) The increased length of the mutant allele was the basis for a PCR-based genotyping strategy. (D) Genotypes of progeny from mating miR-208a mice were born at a Mendelian ratio (n = 128). (E) Northern blot analysis for miR-208a expression in hearts from wild-type (Mir208a+/+), Mir208a+/–, and Mir208a–/– mice. *P < 0.01. (F) Heart weight to body weight ratios of 4-month-old wild-type and Mir208a–/– mice (n = 25/genotype). (G) Transcripts for αMHC, βMHC, and ANF were detected by real-time PCR in hearts from wild-type and Mir208a–/– mice (n = 5/genotype). Values are presented as the fold change in mean expression ± SEM. *P < 0.01. (H) Western blot analysis of total MHC and βMHC protein levels in hearts from wild-type and Mir208a–/– mice. **P < 0.05. (I) Northern blot analysis of miRNA expression using hearts from wild-type, Mir208a+/–, and Mir208a–/– mice.
The progeny that resulted from mating Mir208a+/– mice were viable and born at an expected Mendelian ratio (Figure 5D). miR-208a expression was halved in Mir208a+/– hearts compared with the wild-type hearts and was undetectable in Mir208a–/– hearts (Figure 5E and Supplemental Figure 1). Hearts of 12- to 16-week-old Mir208a–/– mice did not display any gross morphological abnormalities and appeared normal compared with wild-type littermates (data not shown). Furthermore, no differences in heart weight to body weight ratios were observed when comparing Mir208a–/– and wild-type mice (Figure 5F). Those results are consistent with a recent report in which miR-208a was shown to not be required for normal heart development and function (17).
Consistent with the role of miR-208a in the regulation of cardiac hypertrophic growth and βMHC expression, we found that βMHC transcript and protein levels were significantly reduced, while αMHC and ANF transcript levels were unchanged in Mir208a–/– hearts (Figure 5, G and H). This result complements the elevation of βMHC transcript and protein levels observed in miR-208a Tg hearts (Figure 3, A and B). Together, those genetic data provide convincing evidence that miR-208a is important for the regulation of βMHC expression. We also examined the expression of miRNAs that were downregulated in miR-208a Tg hearts (Figure 3D). Surprisingly, we found that the expression of those miRNAs appeared unchanged in Mir208a–/– hearts, indicating that their expression is not dependent upon miR-208a (Figure 5I).
miR-208a and miR-208b repress the expression of thyroid hormone–associated protein 1 and myostatin. Utilizing the web-based TargetscanHuman database (http://www.targetscan.org/), we selected several predicted miR-208a target genes for experimental scrutiny (32, 33). A target site located in the 3′ UTR of thyroid hormone–associated protein 1 (Thrap1) is among the most heavily weighted target sites for miR-208a and was chosen for study since thyroid hormone signaling is a known repressor of βMHC transcription (34, 35). Upon closer inspection of the Thrap1 3′ UTR, we identified a second conserved miR-208a target site located approximately 60 bp downstream of the first target site (Figure 6A and Supplemental Figure 6).
miR-208a and miR-208b repress the expression of Thrap1 and myostatin. (A) Sequence alignment between miR-208a and candidate binding sites in the 3′ UTR of Thrap1 and myostatin. (B) Northern blot analysis demonstrated that miR-208a, miR-208b, and miR-124 expression plasmids produced mature miRNAs when transfected into 293T cells. Total RNA from mouse brain and neonatal and adult hearts was included as a control. U6 served as a loading control. (C) 293T cells were transfected with a luciferase reporter designed to detect miR-208a expression (208a sensor), along with the indicated miRNA expression plasmids. A Thrap1 3′ UTR (luc-Thrap1) and a mutated Thrap1 3′ UTR (luc-Thrap1 mutant) were also tested. Values are mean luciferase activity ± SD relative to the luciferase activity of reporters cotransfected with an empty expression plasmid. The dotted line indicates the basal level of luciferase activity in control (i.e., the luciferase vector alone). (D) A luciferase reporter with duplicated Thrap1 binding sites (luc-Thrap1 4×) was cotransfected with miRNA expression plasmids and luciferase activity determined. (E) A luciferase reporter with 4 repeats of the putative myostatin binding site was cotransfected with miRNA expression plasmids and luciferase activity determined. (F) Western blot analysis for Thrap1 and myostatin protein levels in hearts from 4-month-old miR-208a Tg versus control animals and Mir208a–/– versus wild-type animals. GAPDH served as a loading control. *P < 0.05.
In addition, myostatin is also predicted to be a miR-208a regulatory target, with a single miR-208a target site in its 3′ UTR (Figure 6A and Supplemental Figure 6). Myostatin stood out as an interesting candidate, given its role as a repressor of hypertrophic growth in skeletal muscle (36, 37). Myostatin is also expressed in cardiac muscle, although to a much lesser degree than it is found in skeletal muscle, and genetic inactivation of myostatin in mice has been recently linked to cardiac hypertrophy (38–41).
As a first step toward determining which genes are targeted by miR-208a and miR-208b, we cloned genomic fragments encoding miR-208a, miR-208b, and miR-124 into plasmids for overexpression in cultured cells (Figure 6B). We hypothesized that similar sequences and identical seed regions shared by miR-208a and miR-208b would enable them to repress similar sets of genes, while miR-124 is a brain-specific miRNA and served as a control miRNA for specificity.
Thrap1 is part of the thyroid hormone nuclear receptor complex and can positively and negatively influence transcription (42, 43). Thus we reasoned that repression of Thrap1 by miR-208a might account for the increased βMHC expression in miR-208a Tg hearts (Figure 3, A and B). In agreement with this notion, cotransfection of a luciferase gene with the Thrap1 3′ UTR cloned immediately downstream (luc-Thrap1) and the miR-208a expression plasmid in cultured cells resulted in repressed luciferase activity (Figure 6C). Expression of miR-124 with luc-Thrap1 had no such effect upon luciferase activity, which indicated that miR-208a repression of luc-Thrap1 was specific. To further confirm this specificity, we mutated the candidate miR-208a target sites (luc-Thrap1 mutant), which resulted in the complete loss of miR-208a–mediated repression (Figure 6C and Supplemental Figure 6). As we predicted from the sequence similarity shared by miR-208a and miR-208b, we found that miR-208b also repressed the luc-Thrap1 luciferase activity (Figure 6C). As another demonstration of miR-208a and miR-208b targeting of Thrap1 and to test whether increasing the number of target sites would also increase the degree of repression, we duplicated the Thrap1 target sites downstream of the luciferase gene (luc-Thrap1 4×). Indeed, increasing the target site number resulted in a pronounced decrease in luciferase activity, indicating that target site number is an important factor in miRNA-mediated repression (Figure 6D).
In order to directly test whether miR-208a could repress the expression of myostatin, we constructed 4 repeats of the myostatin target sequence downstream of a luciferase gene (luc-myostatin 4×) and cotransfected with miRNA expression plasmids. Transfection of either miR-208a or miR-208b plasmids repressed luc-myostatin 4× luciferase activity (Figure 6E). Transfection of the miR-124 plasmid caused no decrease in luc-myostatin 4× luciferase activity and confirmed that miR-208a and miR-208b specifically target the myostatin 3′ UTR (Figure 6E).
To determine whether Thrap1 and myostatin were regulated in vivo by miR-208a, we tested whether their expression was altered in our miR-208a gain- and loss-of-function mouse models. The transcript levels of Thrap1 and myostatin appeared unchanged in the miR-208a Tg and Mir208a–/– hearts (Supplemental Figure 6). However, the protein levels of Thrap1 and myostatin were repressed in miR-208a Tg hearts compared with control hearts (Figure 6F). Conversely, the protein levels of Thrap1 and myostatin were elevated in Mir208a–/– hearts compared with hearts from wild-type littermates (Figure 6F). Taken together, these observations demonstrated that Thrap1 and myostatin are bona fide miR-208a targets.
miR-208a is a regulator of the cardiac conduction system. To determine whether miR-208a overexpression disturbed cardiac conduction, we recorded surface ECGs of 4-month-old miR-208a Tg and control mice. Analysis of the ECG recordings showed significantly prolonged PR intervals in miR-208a Tg mice compared with control mice (Table 2 and Figure 7A). The PR interval is the period of time between the onset of atrial depolarization and the onset of ventricular depolarization; abnormal prolongation of the PR interval is clinically considered first-degree AV block. No significant differences were detected in other ECG parameters, such as QRS, QT, or QTc intervals (Table 2). We also recorded and analyzed surface ECGs from 1- and 6-month-old animals and obtained similar results (Supplemental Table 2). Interestingly, approximately 30% of the miR-208a Tg mice also had second-degree AV blocks, in which one or more of the electrical impulses from the atria failed to pass through the AV node to the ventricles, causing failures in ventricular contraction (Figure 7B). Second-degree AV blocks appeared on the ECG tracings as P waves (atrial depolarizations) without subsequent occurrence of QRS complexes (ventricular depolarizations) (Figure 7B). Taken together, the development of progressive heart blocks in miR-208a Tg mice demonstrates that miR-208a overexpression results in cardiac conduction abnormalities and suggests that miR-208a regulates cardiac conduction system components.
miR-208a is sufficient to induce arrhythmias and is required for proper cardiac conduction. (A) Representative waveforms in lead I indicate the location and relative duration of PR intervals in 4-month-old miR-208a Tg and control mice. (B) Representative ECGs in lead I of 4-month-old miR-208a Tg and control mice. Asterisks mark missing QRS complexes and indicate occurrences of second-degree atrioventricular block. (C) Representative waveforms in lead I from Mir208a–/– and wild-type mice. Asterisk indicates the normal position of the P wave, which was absent in Mir208a–/– mice. (D) Representative ECGs in lead I from 4-month-old Mir208a–/– and wild-type mice. Asterisks mark the presence of the P wave.
Summary of 4-month surface ECG findings
We and others have found that genetic deletion of miR-208a does not affect viability or cause gross morphological heart defects but is required for stress-dependent heart growth (Figure 5 and ref. 17). To determine whether miR-208a is also required for proper heart electrophysiology during normal conditions, we monitored heart function of 4-month-old Mir208a–/– and wild-type littermates by surface ECGs and found that miR-208a is necessary for proper cardiac conduction. The ECGs of approximately 80% of the Mir208a–/– mice lacked P waves preceding QRS complexes, suggesting that Mir208a–/– mice suffer atrial fibrillation (Figure 7, C and D). Consistent with a defect in atrial conduction, the PR intervals in Mir208a–/– mice were significantly prolonged compared with wild-type animals (Table 2). Collectively, the ECG analyses of miR-208a Tg and Mir208a–/– mice demonstrate that miR-208a is an important component of the cardiac conduction system.
Normal conduction is mediated by the orderly propagation of electrical impulses from one cardiomyocyte to the next. The connexin proteins are gap junction proteins required for this propagation, and their altered expression is a common feature in a variety of chronic human heart diseases associated with increased risk of arrhythmias and sudden death (44–47). Studies of mouse models have demonstrated that deficiencies in either connexin 43 (Cx43) or Cx40 result in cardiac conduction defects (48). Cx43 is expressed in cardiomyocytes throughout the heart, whereas Cx40 expression is restricted to the atria and the specialized cardiomyocytes that constitute the His bundle and Purkinje fibers (48). Previous studies suggested that abnormal connexin protein expression might account for, at least in part, the cardiac conduction defects induced by altered miR-208a levels (49–51).
We therefore evaluated the expression of Cx43 and Cx40 in hearts from 4-month-old miR-208a Tg and Mir208a–/– mice. Transcript analysis by real-time PCR did not reveal any readily apparent changes to Cx43 transcript levels in either miR-208a Tg or Mir208a–/– hearts (Figure 8, A and B). Though Cx40 transcript levels were not affected in miR-208a Tg hearts (Figure 8C), they were markedly decreased in Mir208a–/– hearts compared with wild-type (Figure 8D), indicating that miR-208a is required for Cx40 expression. Furthermore, Western blot and immunohistological analyses of hearts from Mir208a–/– mice showed decreased Cx40 protein levels compared with wild-type hearts (Figure 8E and Supplemental Figure 7). Consistent with the Mir208a–/– phenotype, mice lacking Cx40 suffer cardiac conduction abnormalities, including first-degree AV block (49).
miR-208a is required for proper expression of gap junction protein connexin 40 and cardiac transcription factors GATA4 and Hop. (A and B) Transcripts for Cx43 were detected by real-time PCR in hearts from (A) miR-208a Tg and control mice (n = 5/genotype) and (B) wild-type and Mir208a–/– mice (n = 5/genotype). (C and D) Transcripts for Cx40 were detected in hearts from (C) miR-208a Tg and control mice (n = 5 each genotype) and (D) wild-type and Mir208a–/– mice (n = 5/genotype). Values in A–D are presented as the fold change in expression ± SEM. *P < 0.01. (E) Western blotting for Cx40 proteins using hearts from 4-month-old wild-type and Mir208a–/– mice. β-Tubulin served as a loading control. (F) Transcripts for Hop were detected by real-time PCR in hearts from wild-type and Mir208a–/– mice (n = 5/genotype). (G) Western blotting for Hop proteins using hearts from 4-month-old wild-type and Mir208a–/– mice. (H) 293T cells were transfected with a luciferase reporter designed to detect miR-208a expression and with miRNA expression plasmids, and luciferase activity was determined. A luciferase reporter with 4 repeats of the putative miR-208a binding site was also cotransfected with miRNA expression plasmids and luciferase activity determined. Values are mean luciferase activity ± SD relative to the luciferase activity of reporters cotransfected with empty expression plasmid. (I) Western blotting for GATA4 proteins using hearts from 4-month-old wild-type and Mir208a–/– mice. **P < 0.05.
miR-208a maintains expression of cardiac transcription factors. The transcription factor homeodomain-only protein (Hop) is highly expressed within the adult murine conduction system, and Hop genetic deletion results in postnatal conduction defects accompanied by a loss of Cx40 expression (52). Therefore, we speculated that the decreased Cx40 expression observed in Mir208a–/– hearts might partially stem from reduced Hop expression. We evaluated Hop transcripts levels by real-time PCR using hearts from 4-month-old animals and found Hop expression to be abolished in Mir208a–/– mice (Figure 8F). Accordingly, Hop protein was also undetectable in Mir208a–/– hearts (Figure 8G). Our observations indicate that Hop is not directly regulated by miR-208a, but instead that miR-208a targets a transcription factor associated with Hop and/or required for Hop expression. GATA4 and Nkx2.5 are transcriptional cofactors expressed within the cardiac conduction system of the adult heart and coordinately transactivate the expression of serum response factor–dependent promoters, including Cx40 and Hop (53, 54). Interestingly, the 3′ UTR of Gata4 mRNA contained a predicted miR-208a target site (Supplemental Figure 8), and so we reasoned that the miR-208a gain- and loss-of-function phenotypes might partially result from irregular GATA4 protein expression (55). In order to directly test whether miR-208a could repress the expression of GATA4, we constructed 4 repeats of the GATA4 target sequence downstream of a luciferase gene (luc-GATA4) and cotransfected with miRNA expression plasmids. Cotransfection of miR-208a with the luc-GATA4 reporter repressed luciferase activity (Figure 8H). However, cotransfection of control miR-124 caused no decrease in luciferase activity, demonstrating that miR-208a specifically targeted the 3′ UTR of GATA4 (Figure 8H). To determine whether GATA4 is regulated in vivo by miR-208a, we tested whether GATA4 expression was altered in the Mir208a–/– hearts. Indeed, GATA4 protein levels were elevated in Mir208–/– hearts compared with hearts from wild-type littermates (Figure 8I). Consistent with posttranscriptional regulation, the GATA4 transcript levels were unchanged in Mir208a–/– hearts (data not shown). Taken together, these observations demonstrate that miR-208a directly targets the cardiac transcription factor GATA4.