Attenuation of microRNA-1 derepresses the cytoskeleton regulatory protein twinfilin-1 to provoke cardiac hypertrophy

The heart responds to physiological stimuli, tissue injury or endocrine disorders by undergoing hypertrophic growth to sustain cardiac output (Frey et al., 2004). Functionally, cardiac hypertrophy is an initial adaptive response, but prolonged hypertrophy will lead to heart failure and sudden death (van Rooij et al., 2008b). Cardiac hypertrophy is a complicated process accompanied with an increase in the protein content of cardiomyocytes, the expression of fetal cardiac genes, and reorganization of the cytoskeleton. The actin cytoskeleton has a crucial role in the pathogenesis of cardiomyopathy (Arber et al., 1997; Heling et al., 2000; Wei et al., 2001). It not only contributes to morphological integrity and mechanical resistance of the cell, but also participates in the transmission of stress signals leading to changes in cardiac gene expression and function (Kawamura et al., 2003).

The dynamics of the actin cytoskeleton are regulated by multiple actin-binding proteins, which control many different aspects of actin dynamics. Twinfilin is an evolutionarily conserved actin-binding protein, which regulates the cytoskeletal structure and dynamics in organisms from yeast to mammals (Goode et al., 1998). Twinfilin binds actin monomers with high affinity and prevents their association with actin-filament barbed ends (Ojala et al., 2002). It also interacts with actin-filament barbed ends and the heterodimeric capping protein, another conserved regulator of cytoskeletal dynamics (Falck et al., 2004; Helfer et al., 2006; Paavilainen et al., 2007). In cells, twinfilin shows diffuse cytoplasmic localization, but also concentrates to the dynamic cortical actin cytoskeleton (Palmgren et al., 2001). Deletion of twinfilin in budding yeast caused enlarged cortical actin patches and defects in the bipolar budding pattern (Goode et al., 1998). Strong hypomorphic mutation in the Drosophila twinfilin gene results in defects in actin-dependent development processes (Wahlstrom et al., 2001). Two twinfilin isoforms exist in mammals: twinfilin-1 (TWF1) is predominantly expressed in non-muscle tissues and twinfilin-2 (TWF2), is the most abundant isoform in skeletal muscle and the heart (Nevalainen et al., 2009; Vartiainen et al., 2003). In cultured murine non-muscle cells, overexpression of TWF1 leads to fewer stress fibers and the appearance of abnormal wormlike actin-filament structures (Vartiainen et al., 2000). In cultured neonatal rat cardiomyocytes, TWF1 displays a punctate cytoplasmic localization and is also enriched along myofibrils (Nevalainen et al., 2009). However, the regulation and the function of TWF1 in heart disease are not clear.

Recently, a new gene regulator, microRNA (miRNA), was reported to have an important role in heart disease (Chen et al., 2009; Condorelli et al., 2010; Wang et al., 2010), including cardiac hypertrophy. miRNAs are a class of conserved, single-stranded non-coding RNAs consisting of 18-25 nucleotides. They negatively regulate target gene expression through mRNA cleavage or translation repression (Bartel, 2004). Cardiac-specific deletion of an RNase III endonuclease Dicer, which is essential for miRNA maturation, induces sudden death and mild cardiac remodeling in juvenile mice, and results in cardiac dysfunction with induction of hypertrophic marker genes in the adult mice (da Costa Martins et al., 2008). Recent studies also indicate that several miRNAs, including miR-133 (Care et al., 2007; Luo et al., 2008), miR-208 (van Rooij et al., 2007), miR-195 (van Rooij et al., 2006) and miR-1 (Care et al., 2007; Ikeda et al., 2009; Luo et al., 2008; Sayed et al., 2007) actively participate in cardiomyocyte hypertrophy. Among them, miR-1 is a muscle-specific miRNA, which has an important role in cardiac development and differentiation (van Rooij et al., 2008a; Zhao et al., 2007; Zhao et al., 2005). It also regulates cardiac hypertrophy by controlling cardiac growth, conduction and calcium-dependent signaling (Care et al., 2007; Ikeda et al., 2009; Luo et al., 2008; Sayed et al., 2007). However, it is not known whether miRNAs participate in the regulation of the cytoskeleton during cardiac hypertrophy.

Here, we show that TWF1, a cytoskeleton regulatory protein, is a direct downstream target of miR-1. Accompanied with the downregulation of miR-1, levels of TWF1 were increased in cardiac hypertrophy, both in vivo and in vitro. Overexpression of miR-1 in cardiomyocytes inhibited the hypertrophic phenotype induced by phenylephrine (PE), whereas silencing the endogenous miR-1 resulted in cellular hypertrophy. We further demonstrate that upregulation of TWF1 promotes cardiomyocyte hypertrophy.

Previous studies have provided variable results concerning the identity of the most abundantly expressed miRNAs in the heart (Lagos-Quintana et al., 2002; Landgraf et al., 2007). Thus, we performed microarray-based miRNA profiling analysis with total RNA from the human heart. Of the 719 human miRNAs probed, only 237 were detectable. The top 23 miRNAs with signal intensity above 4000 are presented in Fig. 1. Among them, miR-1 was most abundant in heart. Several families including let-7, miR-26, miR-126, and miR-133 were also among the enriched miRNAs in human heart, consistent with a previous report (Landgraf et al., 2007). A complete list of the expression profile of miRNAs in the heart is presented in supplementary material Table S1.

To verify miRNAs enriched in the heart, we constructed a small RNA library with human heart total RNA and sequenced the obtained small RNA clones. Among the 251 sequenced clones, 113 (45%) were annotated to known miRNAs (see supplementary material Table S2). The remaining small RNA clones corresponded to fragments of rRNA, tRNA, mRNA and repeat sequences from the human genome (data not shown). Among the miRNA clones, we identified 27 distinct miRNAs, 15 of which were cloned several times and the remaining 12 were cloned just once. Most of the miRNAs (22/27) obtained were present in the top 50 miRNAs of the above microarray data, suggesting the consistency of the results obtained by the two methods. It is important to note that miR-1 accounted for 24% (27/113) of the total miRNAs clones, indicating that miR-1 is, at least based on our study, indeed the most abundant miRNA in the heart.

Detailed analysis of our data revealed that compared with the mature miRNA sequences annotated in miRBase (release 13.0), the miRNAs cloned here from the human heart showed an extensive degree of heterogeneity, including deletions or extensions at 3′ and 5′ ends. Some of the variations are shown in Table 1. For instance, five of 27 miR-1 clones had an AU deletion at 3′ end, and two contained an extra A at the 5′ end. The most prevalent variation of miR-126 was a deletion of U at the 5′ end. Overall, the 3′-end variations (46.9% of total clones) were more abundant than those of the 5′ end (21.2% of total clones); 3′-end deletions were especially typical when A or U was the last nucleotide (supplementary material Table S3).

Next, seven miRNAs presented in both microarray and cloning data were selected for Northern blot verification with the total RNA extracted from mouse tissues (Fig. 2 and supplementary material Fig. S1). MiR-1 and miR-133 were found to be specific to the heart and skeletal muscle, as reported previously (Liu et al., 2007), and miR-125b was highly expressed in lung, heart and stomach; miR-23b, miR-26a, miR-27b and let-7b displayed ubiquitous expression in all detected tissues. In addition, verification was also performed with rat tissues. Consistently with the results obtained in mice, these miRNAs showed a similar expression pattern in the rat tissues (data not shown).

Northern blot analysis revealed that miR-1 was specifically expressed in heart and skeletal muscle (Fig. 2A). Real-time PCR analysis also confirmed the cardiac-enriched expression of miR-1 both in rat (Fig. 2B) and mouse (data not shown). To further investigate the expression pattern of miR-1 in various cell types of the heart, we isolated cardiac myocytes and fibroblasts from the rat heart and detected the expression of miR-1. miR-1 was detected predominantly in cardiac myocytes, but was undetectable in cardiac fibroblasts by Northern blot analysis (Fig. 2C). Similar results were obtained by real-time PCR analysis (Fig. 2D). We also found that miR-1 expression was below the detectable level in primary cultured cardiac vascular cells, including human coronary artery smooth muscle cells, endothelial cells (Fig. 2C), as well as in several non-cardiac cell lines: NIH3T3, HEK293T and HeLa cells (supplementary material Fig. S2A). Therefore our data indicated that miR-1 is a myocyte-specific miRNA in the heart.

It has been established that miRNA exerts its function by repressing the expression of its target genes. Therefore, we searched for the potential targets of miR-1. Combined with several bioinformatic algorithms, TWF1 was selected as a putative target of miR-1. Two miR-1-binding sites (labeled MBS1 and MBS2 in Fig. 3A) within the TWF1 3′ untranslated region (UTR) were identified using the RNA-hybrid program (Rehmsmeier et al., 2004) (Fig. 3A). The nucleotides complementary to the miR-1 seed sequences are highly conserved in human, mouse and rat, indicating the generality of the relationship between miR-1 and TWF1. There is a TWF1 homolog in mammalian cells, TWF2, which has no putative MBS in its 3′ UTR. Interestingly, the two isoforms have distinct expression patterns (Fig. 3B): TWF1 is the major isoform in non-muscle tissues including liver, lung and kidney, but expressed at low levels in heart and skeletal muscle, which was inversely related to the expression pattern of miR-1; by contrast, TWF2 is strongly expressed in cardiac and skeletal muscle and displayed no similar relationship with miR-1.

To examine whether miR-1 regulates TWF1 expression through its 3′ UTR, we generated luciferase reporters with insertion of the mouse Twf1 or Twf2 3′ UTRs. Overexpression of miR-1, but not miR-126, dramatically decreased the activity of the luciferase reporter bearing the 3′ UTR of Twf1. By contrast, neither miR-1 nor miR-126 inhibited the activity of the luciferase reporter containing the Twf2 3′ UTR (Fig. 4A). To figure out which one of the two putative MBSs in Twf1 3′ UTR was responsible for the expression suppression, we performed site-directed mutagenesis to destroy the MBSs individually or simultaneously (Fig. 4B). Our result showed that each MBS mutation could partially rescue the luciferase activity suppressed by miR-1. Inactivation of both MBSs almost completely rescued the suppression of the luciferase activity by miR-1 (Fig. 4C). Therefore, the luciferase activity suppression conferred by miR-1 was mediated mainly by the two MBSs, which contributed equally, although the presence of additional potential MBSs cannot be excluded.

To verify whether miR-1 could repress endogenous TWF1 protein expression, NIH3T3 cells which express high level of endogenous TWF1 (supplementary material Fig. S2B) were infected with lentivirus overexpressing miR-1 (Lenti-miR-1) or miR-126 (Lenti-miR-126). Several clones stably expressing miR-1 or miR-126 were established. The expression level of miR-1 in the selected miR-1-expressing clones was about 1% of that in the rat hearts (supplementary material Fig. S3A). Real-time PCR analysis indicated that neither Twf1 nor Twf2 mRNA levels showed obvious change in the miR-1-expressing clones compared with that in the control cells (Fig. 4D). However, infection of NIH3T3 cells with adenovirus expressing miR-1 (Ad-miR-1), which induced a similar expression level of miR-1 as that in the rat heart (supplementary material Fig. S3A), reduced the mRNA level of Twf1 compared with that infected with the adenovirus expressing the vector only (Ad-vector), whereas the mRNA level of Twf2 was not affected by Ad-miR-1 infection (supplementary material Fig. S3B). Western blot analysis showed that the protein level of TWF1 was decreased in the clones infected with Lenti-miR-1, but not in those with Lenti-miR-126. Overexpression of either miR-1 or miR-126 had no effect on the endogenous TWF2 protein level (Fig. 4E,F). Collectively, our results indicate that Twf1 is directly targeted by miR-1 and this inhibitory effect is mediated through its 3′ UTR. A mild change in the level of miR-1 did not affect the amount of Twf1 mRNA. However, robust expression of miR-1 (at a similar level to that in the heart) could reduce the amount of Twf1 mRNA.

Previous studies have linked miR-1 to cardiac hypertrophy, and the expression level of miR-1 was dynamically and inconstantly regulated during the early process (Care et al., 2007; Ikeda et al., 2009; Luo et al., 2008; Sayed et al., 2007). To address this issue, we first verified the changes of cardiac miR-1 expression in a rat cardiac hypertrophy model generated by abdominal aorta constriction (AAC). After constriction for 1 week, the ratio of heart weight to body weight was significantly increased (supplementary material Fig. S4A), and the hearts showed apparent hypertrophic growth compared with those of sham operated animals (Fig. 5A). Several hypertrophic markers, including the atrial natriuretic peptide (Nppa), skeletal muscle and cardiac α-actin (Acta1 and Actc1, respectively), and α- and β-myosin heavy chain (Myh6 and Myh7, respectively) were also examined. As expected, Acta1, Myh7 and Nppa were upregulated, whereas Myh6 was downregulated (supplementary material Fig. S4B). With real-time PCR analysis, the level of miR-1 in the hypertrophic left ventricle was decreased by 36% compared with that in the sham controls (Fig. 5B). Northern blot analysis further confirmed the downregulation of miR-1 in hypertrophic hearts (Fig. 5C). To investigate whether the repression of TWF1 by miR-1 could have pathological role in this process, we detected the expression of TWF1 in hypertrophic left ventricle. Real-time PCR analysis showed that the mRNA level of Twf1 was slightly increased, though without statistical significance (Fig. 5D). Also the protein level of TWF1 was increased as determined by western blot analysis (Fig. 5E). We additionally induced the primary cultured neonatal rat cardiomyocyte hypertrophy in vitro by stimulation with phenylephrine (PE). After 48 hours of treatment, the cardiomyocytes developed hypertrophy, which was made evident by the increased cell surface area (Fig. 6A,B). Real-time PCR analysis revealed that the expression of miR-1 in hypertrophic cardiomyocytes was decreased by 30% compared with that in the controls (P<0.05; Fig. 6C). In agreement with the reduction of miR-1, the endogenous TWF1 protein level was upregulated in PE-induced hypertrophic cardiomyocytes (Fig. 6D). However, there was no obvious change in the protein level of TWF2 (supplementary material Fig. S5). These results suggest that TWF1 is de-repressed by miR-1 during the pathogenesis of cardiac hypertrophy.

To further investigate the possible effects of miR-1 on cardiomyocyte hypertrophy, we infected neonatal rat cardiomyocytes with Ad-miR-1, and examined the miR-1 expression levels. The Ad-vector was used as a control. The infection of cardiomyocytes with Ad-miR-1 induced a tenfold increase of miR-1 over the endogenous level as determined by real-time PCR analysis (supplementary material Fig. S6A), and the protein level of TWF1 was decreased (supplementary material Fig. S7B). Neonatal cardiomyocytes were infected with Ad-miR-1 or Ad-vector for 12 hours, followed by treatment with PE for additional 36 hours. Overexpression of miR-1 impaired the increase in cell size induced by PE stimulation as measured by cell surface area (Fig. 7A,B). Furthermore, overexpression of miR-1 reduced the basal level of cell size without PE-stimulation (supplementary material Fig. S6B,C). The results indicate that miR-1 overexpression in cardiomyocytes affects the morphology and structure of the cell, perhaps through the regulation of the cytoskeleton. To determine the effect of miR-1 overexpression on hypertrophic cardiomyocytes at the molecular level, the hypertrophic markers, Acta1, Myh7 and Nppa were examined. Expression of all these genes was significantly reduced in Ad-miR-1-infected cells (Fig. 7C). These results indicate that miR-1 overexpression could inhibit cardiomyocyte hypertrophy. To access whether silencing of the endogenous miR-1 would result in hypertrophy in vitro, we generate an adenoviral vector in which a 3′UTR with tandem miR-1-binding sites is linked to the enhanced green fluorescent protein (EGFP) reporter gene. The complementary sequence acts as a sponge to sequester the endogenous miR-1. Infection of Ad-sponge (Ad-spg) rescued the luciferase activity of the reporter suppressed by miR-1 in HEK293T cells (supplementary material Fig. S7A), indicating that Ad-spg could silence the miR-1. Then the cardiomyocytes were infected with Ad-spg. Western blot analysis showed that the protein level of TWF1 was increased by Ad-spg infection (supplementary material Fig. S7B). Importantly, infection of Ad-spg significantly increased the cell size (Fig. 7D,E) and upregulated the expression of the hypertrophic markers (Fig. 7F). These results indicate that silencing of the endogenous miR-1 de-represses the expression of TWF1, and induces cardiomyocyte hypertrophy. Collectively, these data suggests that elevated level of miR-1 inhibits cardiac hypertrophy; however, blocking the function of miR-1 directly induces cardiac hypertrophy, which might occur through the de-repression of TWF1. Our results thus support the notion that miR-1 has anti-hypertrophic activities.

To test whether the increase of TWF1 is sufficient to induce cardiac hypertrophy, we constructed an adenovirus overexpressing TWF1 (Ad-twf-1), and infected the neonatal cardiomyocytes. After infection for 48 hours, the cell size as measured by cell surface area was increased with Ad-twf-1 infection compared with that with the Ad-vector infection (Fig. 8A,B). The expression of the hypertrophic marker genes was subsequently examined in cardiomyocytes infected with Ad-twf-1 or Ad-vector by real-time PCR analysis. The result showed that Nppa expression was significantly increased by Ad-twf-1 infection (Fig. 8C), indicating that the upregulation of TWF1 directly promotes the pathogenesis of cardiomyocyte hypertrophy.

MiRNAs were recently shown to have a fundamental role in the regulation of a wide range of cardiac functions. In this study, we profiled the miRNA expression in the heart by miRNA microarray, and obtained a collection of miRNAs expressed in human heart. The heart-enriched miRNAs were verified using small RNA cloning and Northern blot analyses. We revealed that miR-1 is the most abundant miRNA in the heart and that its expression is limited to cardiomyocytes. Importantly, our work identified an actin-binding protein TWF1 as a direct target of miR-1, and provided evidence that downregulation of miR-1, together with the consequent increase in the cellular TWF1 levels, might result in cardiac hypertrophy.

Interestingly, our work revealed that mature miRNAs in the heart exhibit an extensive degree of sequence end polymorphism. Such end variations have also been noted in other species or tissues (Lagos-Quintana et al., 2002; Landgraf et al., 2007; Ruby et al., 2006; Wu et al., 2007), but the polymorphism observed here in the heart appears to be more dramatic than previously reported: 68.1% of the obtained clones showed 3′- or 5′-end heterogeneity. The miRNA end polymorphism might reflect imprecise cleavage by Drosha or Dicer. Furthermore, the 3′-end polymorphism appears more frequent than that at the 5′ end, which might be due to less precise cleavage by these two enzymes or preferential degradation at 3′ end. The untemplated addition of nucleotides indicates that there are enzymes in the heart that generate end modification after digestion by Drosha and Dicer. Notably, the end polymorphism might affect the function of the mature miRNAs. It is clear that the changes in the 5′-end seed sequence can influence target recognition; the 3′-end polymorphism might affect the stability, subcellular localization and functional efficacy of miRNA (Hwang et al., 2007; Li et al., 2005). Recently, a new method called deep sequencing was developed to extensively explore miRNA discovery. This high-throughput approach can acquire large quantities of sequence readout to identify miRNAs with low abundance, as well as to classify sequence polymorphism for miRNAs.

MiR-1 was identified as the most abundant miRNA in the heart. Our data further demonstrated that miR-1 was specifically present in cardiomyocytes, indicating that miR-1 has an essential role in the regulation of cardiomyocyte morphology and/or function. In the present study, we show that miR-1 was apparently downregulated in the AAC hypertrophic rat left ventricle and PE-induced hypertrophic cardiomyocytes. Our result that overexpression of miR-1 in cardiomyocytes attenuated the hypertrophic phenotype, whereas the reduction of miR-1 resulted in cellular hypertrophy, together with previous reports (Care et al., 2007; Ikeda et al., 2009; Sayed et al., 2007), clearly indicate the inhibitory effect of miR-1 on cardiac hypertrophy. The re-expression of fetal genes such as Acta1, Myh7 and Nppa is characteristic of cardiac hypertrophy, and they are used most commonly as molecular markers of hypertrophy. The expression of these genes is regulated by several transcription factors, including the GATA family members, transcriptional enhancer factor-1 (TEF-1), serum response factor (SRF) and Nkx2-5, which can be activated by the calcium-dependent calcineurin-NF-AT3 signal pathway, MAPK pathway and the PI3K cascade. It has been shown that miR-1 directly targets calmodulin and Mef2a (Ikeda et al., 2009) during cardiac hypertrophy. By negatively regulating the expression of calmodulin and Mef2a, as well as another important transcription factor GATA4, miR-1 could attenuate the calcineurin-NF-AT3 signal pathway and downregulate the expression of the downstream hypertrophic markers. Since induction of the fetal genes is the integrated result of several signal pathways, miR-1 might also regulate the expression of other upstream signal molecules to affect the expression of the hypertrophic markers.

As commonly accepted, miRNA exerts its function through the regulation of several target genes. Therefore, identification of the targets is crucial for understanding the mechanism of miRNAs in regulating diverse biological processes. In the present study, by using several approaches, we demonstrate that the cytoskeleton regulatory protein TWF1 is a novel target of miR-1. First, there are two putative MBSs in the 3′ UTR of TWF1 analyzed by bioinformatic algorithms. Second, the expression level of TWF1 was low in the heart, which is inversely related with the high expression level of miR-1. Third, overexpression of miR-1 significantly reduced the luciferase activity of the reporter containing the 3′ UTR of TWF1. In addition, increased expression of miR-1 in NIH3T3 cells reduced endogenous levels of TWF1 protein. More importantly, in hypertrophic heart and cardiomyocytes, the upregulation of TWF1 was in accordance with the downregulation of miR-1, but the TWF1 homolog protein TWF2, which contains no miR-1-binding sites in the 3′UTR of its mRNA did not show obvious change in the cellular hypertrophic model. Taken together, our results suggest that in the pathological process of cardiac hypertrophy, TWF1 is directly regulated by miR-1. Furthermore, overexpression of TWF1 in cardiomyocytes induced cellular hypertrophy.

TWF1 contributes to actin dynamics by sequestering actin monomers, capping actin-filament barbed ends, and by interacting with the heterodimeric capping protein. TWF1 promotes actin-based motility in vitro and regulates actin dynamics at the leading edge of motile cells (Helfer et al., 2006; Vartiainen et al., 2000; Vartiainen et al., 2003), but the exact mechanism by which the combination of its three distinct biochemical activities drive actin dynamics in cells is not understood. In PE-treated cardiomyocytes, TWF1 displays mainly a diffuse cytoplasmic localization but also concentrates to the non-sarcomeric actin filament structures at the cell protrusions (supplementary material Fig. S8). It is thus possible that, analogously to morphogenesis and motility of non-muscle cells, TWF1 regulates the turnover of the cortical actin cytoskeleton also in cardiomyocytes to promote their expansion during hypertrophy. However, the exact molecular mechanism by which TWF1 regulates cardiac hypertrophy warrants further investigation. It is important to note that also TWF2 is expressed in the heart, but its expression is not regulated by miR-1. TWF1 and TWF2 bind actin and capping protein though conserved sites (Nevalainen et al., 2009; Paavilainen et al., 2008) and thus competition between these two proteins for actin or capping protein might also contribute to the morphology of cardiomyocytes.

Recent work showed that a signature of miRNAs is dysregulated during cardiac hypertrophy, and these tiny RNAs were reported to regulate diverse aspects in the pathogenesis of the disease (Care et al., 2007; Ikeda et al., 2009; Luo et al., 2008; Sayed et al., 2007; van Rooij et al., 2007). MiR-208 influences cardiac contractility by regulating myosin heavy chain, and miR-133 inhibits cardiac hypertrophy by regulating the Rho signaling and automaticity of the heart. miR-1 has also been verified to target several genes in cardiac hypertrophy (Ikeda et al., 2009; Luo et al., 2008; Sayed et al., 2007): calmodulin and Mef2a, which are crucial mediators of the calcium-dependent signal pathway during cardiac hypertrophy; growth-related genes, including those encoding RasGAP, Cdk9, fibronectin and Rheb; and HCN2 and HCN4, which are involved in the automaticity regulation of the heart. Our data showing that the cytoskeleton regulatory protein TWF1 is a new target of miR-1, suggest that miR-1 could prevent cardiac hypertrophy by regulating the cytoskeleton, and thus further clarifies the function of miR-1 during cardiac hypertrophy.

Collectively, we speculate that downregulation of miR-1 by hypertrophic stimuli results in the increase of TWF1, which in turn induces cardiac hypertrophy through the regulation of cardiac cytoskeleton. Our results reveal cardiac cytoskeleton regulated by miRNAs as a new direction for further investigation concerning the mechanism of cardiac hypertrophy. Therefore, appropriate manipulation of the expression of TWF1 in cardiomyocytes might help to prevent the initiation and progression of cardiac hypertrophy.

Primary cultured human coronary artery smooth muscle cells, endothelial cells and human umbilical vein endothelial cells are purchased from Cascade Biologics (Portland, OR). The cells were maintained in cell-specific medium supplemented with growth supplement provided by the manufacturer. HEK293T, HEK293A, HEK293FT and NIH3T3 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum.

The total RNA of the human heart was purchased from Ambion, Inc. The donor is a 71 year old Caucasian male who died of head trauma. The total RNA from mouse, rat tissues, or cultured cells, was extracted using TRIzol reagent according to the protocol of the manufacturer (Invitrogen).

The miRNA expression profile of the human heart was determined by miRNA microarray analysis using the human miRNA array (LC Sciences, Houston, TX), which includes 718 mature human miRNA probes (Sanger miRBase, release 10.1). 5 μg total RNA was size-fractionated (<200 nucleotides) and labeled with Cy5 fluorescent dyes. Then the labeled sample was hybridized to microarray. The microarray data are based on five replicates for each miRNA. Results have been deposited in the public database Gene Expression Omnibus.

Cloning of miRNAs was performed as described (Lagos-Quintana et al., 2002). Human heart total RNA was separated on a 15% denaturing polyacrylamide gel, and RNA of 18- to 25-nt range in size was recovered. A 5′ phosphorylated 3′ adaptor oligonucleotide (5′-pUUUaaccgcgaattccagx: uppercase, RNA; lowercase, DNA; p, phosphate; x, 3′-amino-modifier C-7) and a 5′ adaptor oligonucleotide (5′-acggaattcctcactAAA: uppercase, RNA; lowercase, DNA) were ligated to the recovered RNAs. RT-PCR was performed with 3′ primer (5′-GACTAGCTGGAATTCGCGGTTAAA-3′) and 5′ primer (5′-CAGCCAACGGAATTCCTCACTAAA-3′). After recovery, the 3′ ends of the PCR product were filled in by incubating for 15 minutes at 72°C with Taq polymerase, and then this solution was used for ligation into T vectors. Clones were randomly picked and screened by PCR for inserts, and then subjected to sequencing.

Total RNA (40 μg) was resolved in a 15% denaturing polyacrylamide gel, and then electro-transferred to a Nylon membrane Hybond-N+ (Amersham Biosciences). After UV crosslinking, the membrane was pre-hybridized for 30 minutes, and then hybridized with γ-³²P-labeled specific probes for 12 hours. Probes was washed twice for 10 minutes each and exposed to X-ray film at −80°C.

The total RNA (500 ng) extracted from tissues or cells was used to generate cDNA by using SuperScript II reverse transcriptase (Invitrogen) with special stem-loop primer for miRNA and oligo-dT or random primer for mRNA. Real-time quantitative PCR was performed on a LightCycler 480 quantitative PCR system (Roche) using SYBR Green (TOYOBO Co). Primers used in the amplification reaction are as follows (shown 5′ to 3′): 28S rRNA F, AGCAGCCGACTTAGAACTGG and R, TAGGGACAGTGGGAATCTCG; U6 F, CTCGCTTCGGCAGCACA and R, AACGCTTCACGAATTTGCGT; mouse Twf1 F, ATTTGCTGTCCCAGTCTTCC and R, TACGCCTTGGAGTGTCTGGT; mouse Twf2 F, GCTCTTCGGGACAGTAAAGGA and R, CGGATCTGCTGGAGTTCTCT; rat Gapdh F, AACGACCCCTTCATTGACCTC and R, CCTTGACTGTGCCGTTGAACT; rat Twf1 F, ATGCCCGGATACACATGC and R, CATCCCCATTGTCTATCTCAATC; rat Twf2 F, TCCTTGACTCTGTGGAGCAG and R, TCGCCGATCTCAATCTTCTT; rat Acta1 F, GGTTATGCCCTGCCACAC and R, ATGTCGCGCACAATCTCAC; rat Myh7 F, GAGCCTCCAGAGTTTGCTGAAGGA and R, TTGGCACGGACTGCGTCATC; rat Nppa F, CGGAAGCTGTTGCAGCCTA and R, GCCCTGAGCGAGCAGACCGA.

The miRanda (John et al., 2004) algorithms were used to identify potential targets of miR-1. The RNA-hybrid program (Rehmsmeier et al., 2004) was used to predict multiple potential miR-1 binding sites in the 3′ UTR of mRNA.

The precursor sequence for miR-1 or miR-126 was amplified by PCR using human genomic DNA as a template, and the PCR products were cloned into the pSuper vector to generate miRNA expression plasmids. For construction of the luciferase reporter plasmid, the full-length 3′ UTR of Twf1 or Twf2 were amplified from mouse genomic DNA by PCR, and inserted into the 3′ UTR of the firefly luciferase gene. The mutated 3′ UTR luciferase reporter plasmids were generated by site-directed mutagenesis using the QuikChange site-directed mutagenesis kit, according to the manufacturer's instructions (Stratagene).

HEK293T cells were transfected with luciferase reporter plasmids and the miRNA expression plasmid, and a Renilla luciferase plasmid was cotransfected as an internal control. Cells were harvested 24 hours after transfection. Luciferase activities were measured with a dual luciferase reporter assay kit (Promega) on a luminometer (Lumat LB9507), as described previously (Luo et al., 2008).

Western blotting was performed as described previously (Jing et al., 2005). Antibodies against TWF1 and TWF2 have been described previously (Vartiainen et al., 2003). Antibody against GAPDH was obtained from GenScript.

Left-ventricle hypertrophy was induced in 150-180 g male Sprague-Dawley rats by abdominal aorta constriction (AAC), as described previously (Ruetten et al., 2005). Briefly, rats were anesthetized with chloral hydrate (0.2 mg/g), followed by exposure of the abdomen, and then the suprarenal abdominal aorta was isolated and tightened with a 4-0 nylon suture against 24 gauge needle. After removing the needle, the incision was closed. A control group underwent a sham operation involved all the procedures except aorta constriction. After surgery, each rat was given penicillin twice a day for the first 3 days. Rats were sacrificed 1 week after surgery. All animal protocols were approved by the Institute of Health Sciences Institutional Animal Care and Use Committee.

For histological analysis, sham or AAC operated rat hearts were fixed with 4% paraformaldehyde overnight, and embedded in paraffin. Paraffin sections (4 μm) were stained with hematoxylin and eosin (H&E).

Recombinant adenoviruses were constructed, propagated and titered as previously described (He et al., 1998). Briefly, the precursor sequence for miR-1 was amplified from rat genomic DNA, and the PCR product was inserted into the pAdTrack-CMV vector. For generating the adenovirus expressing TWF1, the full length of mouse Twf1 cDNA fragment was digested from the plasmid pPL144, which has been described previously (Vartiainen et al., 2003). For the construction of the sponge adenoviral vector, we annealed, ligated, gel purified and cloned oligonucleotides for miR-1 binding sites with bulge (CGACACACT TGAGACATTCCA) into the modified pEGFPC1 vector (the stop code sequence was inserted before the multiple cloning sites) digested with BglII and HindIII, then the sequence coding for EGFP together with the miR-1-binding sites was cloned into the pShuttle vector. Then the pAdTrack-CMV vectors with miR-1 precursor sequence or Twf1 cDNA fragment and the pShuttle-sponge vectors were recombinated with pAdEasy in BJ5183 to generate the adenovirus plasmids. After digestion with PacI, the linearized adenovirus plasmids were transfected into HEK293A cells to produce recombinant adenovirus. Titering was performed on HEK293A cells overlaid with DMEM plus 5% FBS and 1.25% low-melt agarose. For lentivirus construction, the precursor sequence for miR-1 or miR-126 was inserted into the modified pLentiLox 3.7 vector (Sun et al., 2007), and then cotransfected with VSVG and PAX2 plasmids into HEK293FT cells to produce lentivirus overexpressing miR-1 or miR-126.

Neonatal rat cardiomyocytes were isolated from 1- to 3-day-old Sprague-Dawley rats. Briefly, isolated hearts were digested with 0.2% trypsin at 37°C under 100 r.p.m. rotation. After dissociation, the cells were subjected to centrifugation to enrich the myocytes, followed by differential preplating to separate from the non-myocytes. Cells were plated in DMEM with 20% fetal bovine serum and BrdU at a density of 3-5×10⁵/ml. 48 hours after plating, serum was removed and the cells were infected with recombinant adenovirus at a multiplicity of infection (MOI) of 100. 48 hours later, cells were harvest for morphology observation, and RNA or protein analysis.

Cells were fixed in 4% paraformaldehyde for 15 minutes and permeabilized with 0.1% Triton X-100 in PBS, followed by blocking with 5% goat serum in PBS for 1 hour at room temperature. They were then incubated with anti-actinin antibody (Sigma) at 1:500 overnight, after washing with PBS three times, the second antibody coupled with Alexa Fluor 555 (Molecular Probes) was added to the cells. The nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). After washing, the slides were mounted using fluorescent mounting medium. Cell surface area was analyzed using software AxioVision 4.7.1 (Carl Zeiss). Images were captured using an Olympus confocal microscope (FV1000) or a Leica confocal microscope (TCS SP5).

All experiments were performed at least three times. Data are expressed as means ± s.e.m. For relative gene expression, the mean value of vehicle control group is defined as 1. Unpaired Student's t-test was used for statistical comparison of the data. Values of P<0.05 were considered to be significant.

We thank all members of the laboratory for helpful discussions and comments on the manuscript. We also thank Sarah E. Stanton for critically reading the manuscript. This work was supported in part by the Ministry of Science and Technology (2005CB724602, 2007CB947002, 2009CB521902), National Natural Science Foundation of China (30770457, 30828006, 30670437), Chinese Academy of Sciences (KSCX2-YW-R-096, KSCX2-YW-R-233, KSCX1-YW-R-64) and Shanghai Pujiang Program (05PJ14105).