|Home | About | Journals | Submit | Contact Us | Français|
Delineating the role of microRNAs (miRNAs) in the posttranscriptional gene regulation offers new insights into how the heart adapts to pathological stress. We developed a knockout of miR-22 in mice and investigated its function in the heart.
Here, we show that miR-22–deficient mice are impaired in inotropic and lusitropic response to acute stress by dobutamine. Furthermore, the absence of miR-22 sensitized mice to cardiac decompensation and left ventricular dilation after long-term stimulation by pressure overload. Calcium transient analysis revealed reduced sarcoplasmic reticulum Ca2+ load in association with repressed sarcoplasmic reticulum Ca2+ ATPase activity in mutant myocytes. Genetic ablation of miR-22 also led to a decrease in cardiac expression levels for Serca2a and muscle-restricted genes encoding proteins in the vicinity of the cardiac Z disk/titin cytoskeleton. These phenotypes were attributed in part to inappropriate repression of serum response factor activity in stressed hearts. Global analysis revealed increased expression of the transcriptional/translational repressor purine-rich element binding protein B, a highly conserved miR-22 target implicated in the negative control of muscle expression.
These data indicate that miR-22 functions as an integrator of Ca2+ homeostasis and myofibrillar protein content during stress in the heart and shed light on the mechanisms that enhance propensity toward heart failure.
Cardiac hypertrophy may initially develop as an adaptive response aimed at increasing heart function and normalizing wall stress to hemodynamic (pressure or volume) overload generated by pathological stresses such as hypertension or other myocardial injuries. Although compensatory cardiac hypertrophy seems beneficial at first, prolonged cardiac hypertrophy is correlated with poor clinical prognosis, eventually leading to maladaptive organ-level changes such as thinning of ventricular walls, dilatation, and diminished cardiac contractility, eventually leading to heart failure.1
Pathological cardiac hypertrophy is characterized by transcriptional reactivation of a fetal gene program, induction of hypertrophic genes, and adaptive and maladaptive changes in expression levels of sarcoplasmic reticulum Ca2+ ATPase activity (SERCA2) and other important contractile proteins.1–4 At the molecular level, reduced expression or activity of SERCA2a is one of the hallmarks of heart failure.3 SERCA2a is a critical determinant of cardiac contractility responsible for Ca2+ reuptake from cytosol during excitation-contraction coupling.5,6 SERCA2a expression and fetal/cardiac hypertrophy genes are under direct transcription control by the serum response factor (SRF), a MADS box transcription factor.7,8 SRF controls gene expression by binding to discrete DNA sequences known as CArG elements [CC(A/T)6GG] usually located in the upstream regions of genes.7–9 The ability of SRF to stimulate gene expression depends largely on number of CArG elements, the binding sites of other transcription factors, and the association of SRF with a large number of positively or negatively acting cofactors.7–9 SRF cofactors like GATA4, MEF2, and myocardin (Myocd) can form ternary complexes with SRF in the heart and activate gene expression. Similarly, negatively acting or repressive SRF cofactors such as four and a half LIM domains 2 (FHL2), Ying-yang 1 transcription factor (YY1), Hop homeobox (HOPX), and purine-rich element binding protein B (PURB) repress SRF gene expression.9–13
New evidence has established the importance of microRNAs (miRNAs) in controlling cardiac hypertrophy, contractility, and regulation of SRF expression.14 Studies in cultured cells and genetically engineered mouse models revealed a fundamental function for miR-133a as an inhibitor of SRF expression/activity in the heart.15 Other studies revealed the miR-208a/miR-208b/miR-499 family as essential regulators of pathological cardiac remodeling and stress-mediated re-expression of the sarcomeric myosin gene β-cardiac muscle myosin heavy chain (Myh7).16,17 Furthermore, miR-208a/miR-208b/miR-499 family–mediated regulation of Myh7 appears to involve posttranscriptional co-regulation of SOX6, PURB, and SP3.17
Several expression-based studies in human diseased hearts and animal models have demonstrated either increased or reduced expression of miR-22, suggesting its role in the pathophysiology of cardiomyopathy.18,19 Moreover, miR-22 expression was found to be upregulated after phenylephrine or angiotensin II stimulation of cultured myocytes.20 Our studies aimed at defining the functional role of miR-22 in heart through the analysis of a loss-of-function mutation in mice. Although a subset of miR-22–null (miR-22–/–) mice die before birth, the surviving mutant mice displayed no overt cardiac pathology under baseline conditions. However adult cardiac myocytes isolated from miR-22–/– mice showed reduced sarcoplasmic reticulum (SR) Ca2+ load and reduced SERCA2a transporting activity. Furthermore, miR-22–/– mice did not adapt well to hemodynamic stress by β-adrenergic stimulation or pressure overload. The absence of miR-22 accelerated the progression to cardiac dysfunction and maladaptation during pressure overload and impaired SERCA2 and SRF expression/activity. In-depth molecular assessment of miR-22 phenotypes led to the identification of PURB as a potentially important downstream target.
Animal procedures were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee. An expanded Methods section appears in the online-only Data Supplement.
Quantitative data are expressed as mean±SEM. Statistical differences between experimental groups were determined with the 2-tailed Student t test, ANOVA with the Dunnett or Tukey post hoc test (using Sigma Plot software), or a linear mixed statistical model obtained with lme4 package in R. The Mann–Whitney U test was used to compare continuous variables with a skewed distribution. A value of P<0.05 was considered statistically significant.
We previously demonstrated that a single miR-22 family member transcribed from an independent noncoding RNA transcript exists in mice/humans.21 To define the role of miR-22, we established miR-22–null mice for analysis (see the Methods section and Figure IA–IC in the online-only Data Supplement). Although miR-22–heterozygous mice developed and reproduced normally, intercrosses between these mice yielded miR-22–homozygous (miR-22–/–) mutant animals at approximately half of the expected mendelian ratio after birth (data not shown). We also noted that a subset of miR-22–/– embryos developed cardiac morphogenetic defects (unpublished data), but surviving adult miR-22–/– mice were fertile and appeared normal.
We then investigated the expression of miR-22 in stressed hearts and normal expression pattern in adult organs. Expression profiling from various adult tissues/organs by quantitative PCR indicated that miR-22 expression is enriched in heart and skeletal muscle (Figure ID in the online-only Data Supplement). We assayed miR-22 cardiac expression in the context of pathological stress induced by pressure overload by transverse aortic constriction (TAC) or in activated calcineurin transgenic (CnA-Tg) mice. miR-22 cardiac expression was transiently increased by 1.5-fold over sham mice at 1 week after TAC operation. However, after 4 weeks of TAC, miR-22 expression was similar to that of Sham mice (Figure IE in the online-only Data Supplement). Furthermore, miR-22 expression was transiently enhanced 2.5-fold in 10-day-old CnA-Tg but was unchanged in 6-week-old CnA-Tg mice (Figure IF in the online-only Data Supplement). Thus, miR-22 expression appears to be temporally induced by pathological stress during early phase(s) of cardiac remodeling.
Initially, we searched for evidence of abnormal heart pathology in the miR-22–/– mice that survived to adulthood. Examination of paraffin-embedded heart sections stained with Masson trichrome or hematoxylin and eosin did not reveal any changes in heart morphology from 4-month-old miR-22–/– mice (data not shown). Resting cardiac function and anatomy, as assessed by echocardiography and pulsed Doppler analysis, were similar between miR-22–/– and wild-type (WT) mice at 4 or 8 months of age (Table I in the online-only Data Supplement). In addition, closed-chest invasive hemodynamic analysis detected no difference in blood pressure, contractility, relaxation, or heart rate in miR-22–/– mice at baseline (Table). However, cardiac inotropic and lusitropic reserve, revealed by infusion of dobutamine (a β-adrenergic receptor agonist), was compromised in miR-22–/– mice as evidenced by a decrease in left ventricular (LV) maximal contraction (dP/dtmax) and relaxation rate (–dP/dtmax) in miR-22–/– mice (the Table). The miR-22–/– mice also demonstrated impaired active relaxation as demonstrated with prolongation of τ. A similar increase in heart rate was observed after dobutamine infusion, suggesting that the chronotropic responsiveness to β-adrenergic stimulation was not impaired in these mice (Table). The reduced response to dobutamine suggests that miR-22–/– mice have a defect in the adrenergic signaling pathway, which could occur at the level of Ca2+ homeostasis and/or in G-protein–coupled receptor signaling.
To explore potential intrinsic abnormalities in calcium handling, we examined calcium transients and SR Ca2+ loading in isolated mutant adult cardiac myocytes. Although the magnitude of the calcium transients was comparable to that in WT cardiac myocytes during phasic electric stimulation, there was a prolongation in caffeine-evoked transient decay in miR-22–/– myocytes compared with WT controls (Figure 1A). We considered that this prolongation in transient decay could be due in part to reduced Ca2+ uptake by SERCA2. To determine if this was the case, the rate constant of Ca2+ decay and the SR load were estimated in Fluo-4–loaded adult cardiac myocytes (Figure 1B and 1C). Indeed, SERCA2a transporting activity was significantly reduced in miR-22–/– myocytes (Figure 1B). Furthermore, the SR Ca2+ load, assessed by caffeine application, was 25% lower than in controls (Figure 1C and 1D). Collectively, these results suggest that the absence of miR-22 impairs intracellular calcium homeostasis under baseline conditions.
To determine whether the prolongation in transient decay seen miR-22–/– myocytes was due in part to reduced SERCA2a expression, we quantified Serca2a mRNA expression and protein abundance in miR-22–/– mice compared with controls. Serca2a transcript levels were subtly repressed in miR-22–/– hearts (Figure 1E). However, Western blot analysis of SERCA2 found comparable protein abundance in miR-22–/– hearts (Figure 1F). These data suggest that the observed subtle downregulation in Serca2a mRNA might originate from a transcriptional mechanism or that there is a subtle alteration in SERCA2a protein expression that is beyond the limit of detection. We also investigated the expression of other genes involved in Ca2+ excitation-contraction coupling and cardiac contractility. We did not detect a difference in expression levels of phospholamban, sodium-calcium exchanger, calsequestrin, and ryanodine receptor 2 in miR-22–/– hearts at baseline (Figure 1E and 1F, Figure IIA and IIB in the online-only Data Supplement, and data not shown). Additionally, no difference was seen in the phosphorylation state of phospholamban at Ser-16 or Thr-17 (Figure IIB in the online-only Data Supplement).
The prolonged transient decay and reduced SR Ca2+ load prompted us to ask whether miR-22–/– mice had increased propensity to cardiomyopathy by pressure overload. To explore this possibility, we subjected miR-22–/– and WT mice to pressure overload by TAC for 1, 2, or 4 weeks. Echocardiographic analysis revealed a comparable degree of LV chamber geometry and systolic function in TAC-operated miR-22–/– and WT mice at both 1 and 2 weeks after TAC in a typical pattern of concentric hypertrophy (Table II in the online-only Data Supplement). Furthermore, as expected, WT mice subjected to TAC had increased LV posterior wall thickness at end systole without significant changes in fractional shortening as an indication of compensatory hypertrophy (Figure 2B and Table II in the online-only Data Supplement). However, after 4 weeks of TAC, miR-22–/– mice developed LV chamber dilation and a marked deterioration in cardiac contractile function, as indicated by increased LV end-systolic dimension and impaired fractional shortening (Figure 2B).
Internally consistent with the notion that the absence of miR-22 sensitizes to cardiomyopathy by pressure overload, histomorphometric analysis revealed increased fibrosis in miR-22–/– mice after 1 or 4 weeks of TAC (Figure 2A and 2C and Figure IIIA and IIIC in the online-only Data Supplement). Pathological assessment of cardiac histological sections from TAC-operated mice revealed that WT mice had mild interstitial and perivascular fibrosis, whereas miR-22–/– had multifocal interstitial and replacement fibrosis and areas evocative of cardiomyocyte “dropout” (Figure 2A and Figure IIIC in the online-only Data Supplement). To obtain additional evidence for myocyte dropout, we investigated apoptosis and necrosis in mutant hearts. There was no detectable difference in terminal deoxynucleotidyl transferase dUTP nick-end labeling–positive apoptosis in miR-22–/– compared with WT mice at 1 or 4 weeks after TAC (Figure IIIB and IIID in the online-only Data Supplement). Myocyte cell death by necrosis was indirectly evaluated by assessing Ca2+ deposits through von Kossa staining of histological sections.22 In contrast to sham or TAC WT mice, which showed no von Kossa staining, miR-22–/– mice showed multifocal areas within the LV wall that reacted with the Ca2+-specific von Kossa stain after 1 or 4 weeks of TAC (Figure 2D and Figure IIIC in the online-only Data Supplement). Finally, TAC-operated miR-22–/– mice showed heart weight/tibia length ratios and myocyte cross-sectional areas comparable to those in WT mice, suggesting that miR-22 is dispensable for regulating the growth response of the heart (Figure 2E and Figure IIIE and Table II in the online-only Data Supplement). However, miR-22–/– hearts were impaired in fetal gene program (ie, Myh7, miR-208b, Acta1) activation after TAC, indicating that some aspect of hypertrophic signaling were suppressed (Figure 2F and 2G, Figure IIIF and IIIG in the online-only Data Supplement, and data not shown).
To connect the observed hemodynamic stress–induced accelerated cardiac decompensation phenotype of miR-22–/– mice to a potential abnormality in calcium homeostasis, we initially focused on SERCA2 expression. Loss of miR-22 resulted in decreased production of Serca2a mRNA in the sham group of miR-22–/– mice (Figure 3A), and this reduction in expression was further amplified after 1 week of TAC in mutants (Figure 3A). Western blot analysis from cardiac extracts showed that SERCA2a protein abundance was reduced by 2.4-fold in miR-22–/– versus WT mice after 1 week of TAC (Figure 3B). Quantitative Western blotting revealed that phosphorylation of phospholamban at Ser-16 or Thr-17 was also diminished in miR-22–/– versus WT TAC-operated mice (Figure 3B). Phospholamban and calsequestrin also appeared repressed (Figure 3A and 3B). The activity of SERCA2 is regulated by phospholamban and the status of phospholamban phosphorylation.3,23 This decrease in the phosphorylation of phospholamban would be expected to be detrimental because it would further inhibit SERCA2 activity. Thus, the profound decrease in phosphorylation status of phospholamban in conjunction with the loss of SERCA2a expression would be predicted to render SERCA2a less active, which in turn could accelerate the onset of impaired cardiac pump functional performance to pressure overload in miR-22–/– mice (see Discussion).
To explore the mechanistic basis for stress-dependent cardiac phenotype(s) in the absence of secondary effects of hemodynamic stress, we compared global mRNA transcript levels in 9-week-old miR-22–/– and WT mice hearts by microarray analysis. A total of 131 transcripts were significantly upregulated and 133 were downregulated in miR-22–/– hearts under basal conditions (≥1.2-fold-change; adjusted P<0.05; Table III in the online-only Data Supplement). Application of the gene set enrichment analysis tool on the microarray revealed significant repression of essential muscle-restricted genes encoding Z-disk/sarcomeric proteins such as actin, alpha 1, skeletal muscle (Acta1), titin, integrin beta 1 binding protein 2 (Melusin), and cysteine and glycine-rich protein 3 (Mlp) within the gene ontology cellular component, contractile fiber category, in miR-22–/– hearts (Figure 4A and Table IV in the online-only Data Supplement).
A large proportion of repressed contractile fiber genes are implicated in cardiomyopathy and are known to contribute to contractile dysfunction, myocyte loss, and fibrosis in animal models and humans.4,24,25 In accordance with microarray, we confirmed that mRNA expression levels of genes implicated in cardiomyopathy such as myozenin 2 (Calsarcin-1), dystrophin, LIM domain binding 3 (Ldb3), Melusin, Mlp, and titin were coordinately subtly repressed in sham and TAC groups of miR-22–/– mice (Figure 4D). Sarcomeric myosin genes and their intron-encoded miR-208a/miR-208b/miR-499 family were also subtly repressed in miR-22–/– hearts (Figure 4B and 4C and data not shown). These genes are not predicted targets of miR-22; hence, changes in their expression likely arise secondarily from the absence of miR-22.
Since the majority of miRNAs inhibit gene expression, our prediction was that the absence of miR-22 in the heart would lead to upregulation of direct targets that could contribute to impaired cardiac adaptation to hemodynamic stress. Sylamer analysis of the microarrays conducted from miR-22–/– and WT mice showed that the heptamers complementary to the seed region of miR-22 were enriched within the 3′ untranslated region of upregulated genes in miR-22–/– hearts (Figure 5A and Figure IVA and Table III in the online-only Data Supplement). We applied TargetScan 6.0 and also searched for miR-22 seed matches in upregulated genes in miR-22–/– hearts. This yielded a list of 73 genes, including caveolin-3 (Cav3), chloride intracellular channel protein 4 (Clic4), Purb, and transformation-related protein 53 inducible nuclear protein 1 (Trp53inp1) as potentially important miR-22 targets with seeds in their 3′ untranslated regions (Figure 5B and 5G and Figure IVB–IVD in the online-only Data Supplement).
Overexpression of CAV3 in the heart was linked to cardiomyopathy in mice.26Clic4 and Trp53inp1 promote growth arrest and stress-induced cell death.27,28 However, of special interest to the miR-22–/– cardiac phenotype, PURB was implicated an as antagonist of SRF-dependent transactivation of sarcomeric and hypertrophic gene expression in muscle cells.12,13 Furthermore, Myh6 and Myh7 are subject to negative transcriptional regulation by PURB in muscle.12,13,17,29
To determine whether Purb expression levels and other probable target genes were enhanced in vivo, we performed quantitative polymerase chain reaction and Western blots on miR-22–/– and WT hearts. Indeed, Purb mRNA expression and protein expression were elevated by 2.5- and 2.7-fold, respectively, in miR-22–/– sham and 4-week TAC-stimulated mice compared with WT control groups (Figure 5C and 5D; for information on the PURB antibody, see the work by Kelm et al30). In addition, compared with WT, there was a nearly 4-fold increase in Purb mRNA levels in purified adult cardiomyocytes isolated from 6-week old miR-22–/– ventricles (Figure 5E). This suggests that accumulation of PURB in miR-22–/– myocardium derives, for the most part, from cardiomyocytes. Cav3, Clic4, and Trp53inp1 mRNA expression levels were also elevated in miR-22–/– mice (Figure IVE in the online-only Data Supplement). On the basis of these observations, we cloned the WT 3′ untranslated region for these genes into luciferase reporter plasmids and tested for direct and specific miR-22 repression by virtue of mutagenesis of seed sites in cotransfection assays. Intact miR-22 seed sites were required for efficient repression of luciferase reporters, confirming these mRNAs are targeted by miR-22 (Figure 5G, Figure IVF and IVG in the online-only Data Supplement, and data not shown).
Previous studies suggest that PURB antagonizes muscle expression in part by binding to repressive purine nucleotide–rich (PNR) promoter/intronic/exonic elements with the consensus sequence (GGN)n and, in some cases, through competitive binding to the purine-rich motifs located inside and adjacent to SRF-binding, CArG box–containing enhancers.10,12,13,29 We thus considered the hypothesis that expression derangements of muscle-restricted genes in miR-22–/– hearts relate in part to inappropriately enhanced expression/activity of PURB. To explore this notion, we scanned both murine and human 10-kb genomic regions upstream of repressed genes in miR-22–/– hearts for cis-regulatory control regions representing potential PNR and CArG elements. Comparisons of genomic sequences in human and mouse with MAT inspector identified PNR and CArG elements in the upstream regions of many downregulated genes, including Myh6, Myh7, Serca2a, and Melusin (Table V and Figure VA in the online-only Data Supplement). Next, we compared PURB binding activity in heart homogenates obtained from WT or miR-22–/– mice by semiquantitative ELISA. Single-stranded biotin-labeled sense oligos containing functional PURB-binding motifs in the promoter regions of Myh6 or Myh7,12,13,29 respectively, were more highly bound in cardiac extracts prepared from miR-22–/– compared with WT mice (Figure VB in the online-only Data Supplement). As evidence of specificity, PURB did not bind appreciably to oligos corresponding to the antisense strand (Figure VB in the online-only Data Supplement). Similar results were seen in the AKR-MEF cell line that shows high intrinsic PURB activity (Figure VC in the online-only Data Supplement).
To determine whether loss of SRF expression and activity in miR-22–/– hearts might also partially explain the loss of CArG gene expression, we examined Srf and Myocd expression levels at baseline or after TAC. Quantitative polymerase chain reaction analysis did not find changes in cardiac Srf mRNA expression in miR-22–/– mice (Figure VIA in the online-only Data Supplement and data not shown). In addition, SRF protein levels were unchanged at baseline, and there was mild accumulation of SRF after TAC (Figure VIB in the online-only Data Supplement). In contrast, expression levels of Myocd mRNA were significantly lower in miR-22–/– than in WT TAC mice (Figure VIC in the online-only Data Supplement). Interestingly, cardiomyocyte-specific depletion of either SRF or Myocd in mice results in dilated cardiomyopathy and heart failure in association with repression of cardiac/muscle expression.31,32 Collectively, these data suggest that reduced expression of CArG genes in miR-22–/– hearts comes about in part because increased PURB and impaired SRF-Myocd transcription activity (see Discussion).
miR-22 was recently identified as a inducer of cardiac hypertrophy in gain-of-function experiments conducted in primary cultures of neonatal cardiomyocytes.20,33 To expand on these observations, we engineered transgenic mice that express miR-22 under the control of the myocyte-specific Myh6 promoter (miR-22-Tg). We established 2 independent transgenic mouse lines (low, high) that expressed miR-22 at levels ≈4- and 9-fold higher, respectively, than WT mice at 5 weeks of age (Figure 6A). Both transgenic mouse lines showed significant elevation in heart weight/tibia length ratios and greater cross-sectional areas compared with their WT littermates by 5 weeks of age (Figure 6B–6D). The more pronounced hypertrophy phenotype correlated with transgenic miR-22 expression levels, the reciprocal relationship with its downstream target PURB, and combined effects on SRF-dependent expression (Figure 6E-F), a phenotype that is in large part opposite that of miR-22–/– hearts. Transcript levels of Cav3, a known inhibitor of cardiac hypertrophy,34 were severely compromised in miR-22-Tg high hearts (Figure 6H).
Here, we present data indicating that miR-22 is a novel gene acting to influence cardiac performance during acute and chronic hemodynamic stress. Our data reveal an underlying defect in contractile reserve in miR-22–/– mice after dobutamine stress. Defects in cardiac response to dobutamine may emanate from defective β-adrenergic coupling and/or intrinsic abnormalities at the level of calcium handling in miR-22–/– mice. Our data suggest that reduced cardiac response to dobutamine relates in part to an intrinsic defect at the level of calcium handling. We speculate that miR-22–/– mice may be incapable of increasing SR Ca2+ content to the same extent as in WT hearts. The consequence of impaired SR Ca2+ content during dobutamine stimulation would be a loss of calcium transients/inotropic response.6,35 Supportive of this interpretation, adult cardiac myocytes isolated from miR-22–/– mice showed reduced SR Ca2+ load and mild loss in SERCA2 transporting activity and Serca2a mRNA expression. Previous studies have shown that a reduction in SERCA2 expression/activity beyond a certain threshold can lead to impaired calcium transients and loss of cardiac reserve to adrenergic stimulation.3,36 In the future, it will interesting to decipher mechanism(s) acting upstream in the adrenergic signaling cascade that contribute to impaired adrenergic response and SERCA activity in the absence of miR-22.
We found that genetic deletion of miR-22 accelerated the transition to eccentric remodeling and loss of cardiac contractility after TAC. In agreement with this interpretation, chronic pressure overload brought about more rapid onset of LV dilation and loss of ventricular pump function as denoted by echocardiographic analysis. Histological analysis revealed an increased propensity of stressed miR-22–/– myocardium to enhanced fibrosis and calcium deposits evocative of myocyte dropout. Our data suggest that these defects arise as a result of intrinsic defects in cardiac myocytes (see below).
Some genes such as Serca2a, Mlp, and Dystrophin appeared to be downregulated in miR-22–/– hearts. The reactivation of some cardiac fetal genes such as Myh7, miR-208b, and Acta1 to hypertrophic signals was prevented in miR-22–/– hearts, whereas other genes that typically are upregulated or that have an expression that is not typically modified during cardiac hypertrophy such as Calsarcin-1, Casq2, Ldb3, Melusin, Titin appeared downregulated in mutants during pressure overload. This raises the interesting question of why SRF target genes are coordinately downregulated in miR-22–/– hearts under basal and stress conditions and, more important, whether this defines a primary or secondary mechanism to LV chamber dilation and loss of cardiac contractility. It appears unlikely that stress alone and/or defects in the β-adrenergic pathway account for such a phenotype; instead, these data suggest miR-22 acts both to sustain steady-state cardiac transcription and as a switch allowing adaptive stimulation of some SRF targets genes during pathological cardiac hypertrophy. Thus, we would like to propose a model in which the impaired cardiac gene transcription in miR-22–/– hearts represents a central disease mechanism for the underlying propensity to cardiac decompensation and impaired contractile reserve to biomechanical stress. This defect may give rise to abnormalities in intracellular SR loading in conjunction with loss of force generation and transmission related to progressive loss of myofilament protein content. Furthermore, compensatory mechanisms during initial hypertrophic stimuli would be expected to be surmounted by defects in these pathways, eventually leading to dilated cardiomyopathy and heart failure.
Supportive of a potential contribution of abnormal calcium homeostasis to the phenotype, miR-22–/– myocytes have lower-than-normal SR Ca2+ load and a characteristic prolongation of the calcium transient. Furthermore, there was a reduction in cardiac protein abundance of SERCA2 and calsequestrin and lower-than-normal phospholamban–Ser-16 and phospholamban–Thr-17 phosphorylation status in miR-22–/– mice before the advent of LV chamber dilation and loss of ventricular pump function. Clinical and animal model studies indicate that SERCA2a activity and/or SR Ca2+ uptake are impaired in the failing heart.6,36 Reduced SERCA2a expression levels combined with pressure overload sensitizes to a propensity to contractile heart failure in mice in association with altered excitation-contraction coupling.36,37 Impaired SERCA2a expression/activity may relate in part to loss of steady-state Serca2a transcription in miR-22–/– hearts. However, it is possible that defective β-adrenergic signaling or an increase in protein phosphatase activity may also contribute to loss of phospholamban phosphorylation (SERCA2 activity) in mutants.3,38
Propensity to LV chamber dilation and loss of contractility in miR-22–/– mice to TAC could also relate to impaired force generation and transmission caused by progressive loss in abundance of sarcomeric myosins, Dystrophin, Ldb3, Mlp, and Titin. Genetic mutations in these genes are associated with loss of myofibril organization, altered intercalated disk architecture, increased interstitial fibrosis, and/or altered mechanical stress sensor pathways in animal models.24,25,39 Importantly, these genes have also been linked to genetic susceptibility to hypertrophic cardiomyopathy and/or dilated cardiomyopathy in human patients.4,25
We suggest that defective SRF target gene expression in miR-22–/– hearts may be explained in part by at least 2 distinct mechanisms: (1) PURB transcript and protein levels are increased under basal and stress conditions in miR-22–/– mice as a result of lack of posttranscriptional regulation by miR-22 and (2) loss of Myocd expression/activity during cardiac hypertrophy through a primary or secondary mechanism. Previous studies have shown that PURB represses transcription via CArG box– and/or PNR element–specific nucleoprotein interactions, which either directly or indirectly inhibit muscle gene activation by SRF.12,13,29 PURB has been reported to physically interact with SRF in both the presence and absence of DNA.10,12 Furthermore, absence of miR-22 impaired Myh7 expression and transgenic overexpression of PURB in skeletal muscle of mice led to reduced expression of Myh7.17 In addition, we detected increased PURB binding activity within functional PNR elements found in promoter regions of target genes in miR-22–/– cardiac extracts. In addition, many repressed CArG genes in miR-22–/– hearts, including Serca2a and Melusin, contain evolutionarily conserved putative PNR binding sequences that may mediate transcriptional repressive activity by PURB. Finally, we demonstrated that Myocd, a central activator of SRF activity, was downregulated in miR-22–/– hearts during TAC. It has been shown previously that Myocd forms a ternary complex with SRF at CArG boxes and potentiates target gene expression.8,31 These data support the view that enhanced PURB, in conjunction with cardiac depletion of Myocd, synergistically impairs muscle-restricted gene expression and accelerates the transition to eccentric remodeling in miR-22–/– stressed hearts (see the model in Figure 7).
miRNAs are evolutionarily optimized to titrate the expression of hundreds or even thousands of genes. Although we assigned a role to PURB in the phenotype of miR-22–/– mice, miR-22 has numerous other potential targets in the heart. We found that genes such as Clic4 and Trp53inp1, implicated in cell growth and cell death,27,28 are direct targets of miR-22 and upregulated in miR-22–/– hearts. Hence, it can be predicted that inappropriate expression of these genes may combinatorially impair cardiac myocyte survival. In addition, although these data favor the interpretation that miR-22 exerts its primary effects on myocytes, it is possible that miR-22 also influences the compensatory response to pathological stress by acting within nonmyocytes. Additional studies are required to establish a direct role of PURB in negative transcriptional regulation of Serca2a and to identify other targets of miR-22 that increase propensity to heart failure. Regardless of mechanism(s), this study has important implications and provides insights into the role of miR-22 in the pathogenesis of heart failure.
The identification of genes and pathways involved in progression to heart disease is a major challenge and focus in cardiac research. New evidence has established the importance of microRNAs (miRNAs) in controlling cardiac pathophysiology. The relevance of miR-22 to human heart failure has been suggested by the recent demonstration that miR-22 expression was either downregulated or upregulated in human dilated cardiomyopathy and diseased hearts. In the present report, we demonstrate that genetic ablation of the non–protein-coding gene miR-22 in mice impairs cardiac reserve to β-adrenergic stimulation with dobutamine. Here, we show that the absence of miR-22 increases vulnerability to pressure-overload–induced cardiac decompensation characterized by left ventricular dilation and loss of contractile function.The inability of mutant mice to adapt to biomechanical stress was caused in part by restrained expression of genes such as sarcoplasmic reticulum calcium ATPase (Serca2a), LIM domain binding 3 (Ldb3), cardiac LIM protein (Csrp3/Mlp), and Titin encoding calcium handling and other contractile/myofibrillar proteins implicated in human dilated cardiomyopathy. This phenotype was attributed in part to inappropriate inhibition of serum response factor–dependent gene expression in mutant hearts. In addition, we demonstrated that miR-22 inhibits the expression of purine-rich element binding protein B, a transcription factor that opposes control of sarcomeric/cardiac expression by serum response factor. These results demonstrate the importance of miR-22 as a homeostatic keeper of cardiac gene expression and modulator of cardiac contractile reserve to acute and hemodynamic stress. These results are the first to demonstrate a novel disease mechanism of cardiac pathogenesis involving miR-22.
We thank Thuy Pham for technical support and Chad Shaw and David Chiang for statistical analysis.
Sources of Funding
Dr Rodriguez was supported by scientist development grant 10SDG2640137 from the American Heart Association (national), a Gillson-Longenbaugh Foundation award, and new investigator startup funds at Baylor. Dr Wehrens was supported by US National Institutes of Health (NIH) grants HL089598 and HL091947, Muscular Dystrophy Association grant 69238, and the Fondation Leducq (Alliance for CaMKII Signaling in Heart). Dr Entman was supported by NIH grant HL089792 and by the Hankamer Foundation. Dr Reddy was supported by NIH grants K25 HL73041 and R01 HL22512. Dr Kelm was supported by American Heart Association grant 09GRNT2170060. Dr Lee was supported by the Ronalette and Berdon Lawrence Bone Disease Program.