|Home | About | Journals | Submit | Contact Us | Français|
microRNA (miR)-133a regulates cardiac and skeletal muscle differentiation and plays an important role in cardiac development. Because miR-133a levels decrease during reactive cardiac hypertrophy, some have considered that restoring miR-133a levels could suppress hypertrophic remodeling.
To prevent the ‘normal’ down-regulation of miR-133a induced by an acute hypertrophic stimulus in the adult heart.
miR-133a is downregulated in transverse aortic constriction (TAC) and isoproterenol-induced hypertrophy, but not in two genetic hypertrophy models. Using MYH6 promoter-directed expression of a miR-133a genomic precursor, increased cardiomyocyte miR-133a had no effect on postnatal cardiac development assessed by measures of structure, function, and mRNA profile. However, increased miR-133a levels increased QT intervals in surface electrocardiographic recordings and action potential durations in isolated ventricular myocytes, with a decrease in the fast component of the transient outward K+ current, Ito,f, at baseline. Transgenic (TG) expression of miR-133a prevented TAC-associated miR-133a downregulation, and improved myocardial fibrosis and diastolic function without affecting the extent of hypertrophy. Ito,f downregulation normally observed post-TAC was prevented in miR-133a-TG mice, although action potential duration and QT intervals did not reflect this benefit. miR-133a-TG hearts had no significant alterations of basal or post-TAC mRNA expression profiles, although decreased mRNA and protein levels were observed for the Ito,f auxiliary KChIP2 subunit, which is not a predicted target.
These results reveal striking differences between in vitro and in vivo phenotypes of miR expression, and further suggest that mRNA signatures do not reliably predict either direct miR targets or major miR effects.
Hearts respond to injury through stress-response mechanisms leading to specific molecular adaptations 1, 2. Genetic reprogramming directs changes in protein make-up that modify cardiomyocyte geometry, function, and viability, termed “cardiac remodeling”. microRNAs (miRs) can orchestrate cardiac remodeling by binding to complementary sequences in messenger RNAs (mRNA) and either destabilizing the transcript or suppressing translation 3. Bioinformatics analysis indicates that 30% of all mRNAs are likely miR targets 4, stimulating interest in miR-directed therapeutics 5.
miR-133a is highly expressed in cardiac and skeletal muscle, 6–8 and is regulated in cardiac hypertrophy and failure 9–13. TargetScan 14 prediction reveals 400–500 putative miR-133a mRNA targets, and numerous functional roles have been proposed, including suppressing embryonic cardiomyocyte proliferation 15, preventing genetic cardiac hypertrophy 16, inhibiting apoptosis 17, repressing HCN2 pacemaker channel 18, 19 and ERG K+ channel expression 18,20, and decreasing connective tissue growth factor expression 12. For these reasons, therapeutic overexpression of miR-133 in vivo has been proposed to prevent cardiac remodeling 12, 21. However, most proposed benefits of miR-133a are based upon tissue culture experimentation, and the long-term consequences of forced miR-133a expression in adult hearts have not been examined 3. Indeed, miR-133a expression in the embryonic heart directed by the β-myosin heavy chain (MYH7) promoter generates lethal cardiac developmental defects 15.
Here, we determined the consequences of increasing miR-133a expression on normal and pressure-overloaded adult mouse hearts. miR-133a did not alter normal post-natal cardiac growth or pump function. However, miR-133a overexpressing hearts exhibited modest QT prolongation, increases in cardiomyocyte action potential durations, and reduced functional expression of Kv4-encoded fast transient outward (Ito) K channels. Indeed, by preventing the typical hypertrophy-associated down-regulation of miR-133a, hypertrophy-induced downregulation of Ito,f was abrogated, although repolarization abnormalities were not prevented. Additional benefits conferred by miR-133a in pressure overloaded hearts included reduced myocardial fibrosis and cardiomyocyte apoptosis, with improved diastolic function.
A 735 bp fragment flanking the mouse miR-133a-1 locus on chromosome 18 (519 bp 5’ upstream sequence, 21 nt miR-133a, and 195 bp 3’ sequence) was cloned from mouse genomic DNA into the cardiac αMHC (MYH6) promoter. The current results derive from the higher-expressing of two independent lines.
Total RNA was prepared using Trizol (Invitrogen). Reverse transcription of miR-133a and 5S rRNA was performed using NCode miRNA reverse transcription reagents (Invitrogen). Quantitative PCR assays used the generic NCode universal reverse primer with the following forward primers: 5’-TTGGTCCCCTTCAACCAGCTGT-3’ for miR-133a, 5’-AATACCGGGTGCTGTAGGCTTT-3’ for 5S rRNA. Relative levels of each product were compared using the ΔΔC(t) method. Other mRNA levels were assessed using Affymetrix Mouse Gene 1.0 ST arrays, and miR expression was analyzed using arrays at LC Sciences (Houston, TX). Partek software was used to compute significance of mRNA expression changes using 1-way ANOVA at P<0.001 (false discovery rate of ≤0.03). The threshold level for regulation was 1.3-fold 13.
A subset of mRNA transcripts encoding Kv channel genes was measured using RT-qPCR and SYBR Green. Channel transcript expression was normalized to HPRT and relative levels reported using the ΔΔC(t) method. 22
Surgical transverse aortic coarctation (TAC) and isoproterenol infusion through subcutaneously implanted osmotic mini-pumps (ISO) were performed as described 23, 24. Invasive hemodynamic studies and M-mode echocardiography were performed using standard techniques. Electrocardiography was performed on anesthetized mice and analyzed after signal averaging using the ECG module in LabChart 6 (AD Instruments, Colorado Springs, CO).
TUNEL positivity was assessed using Promega’s DeadEnd Fluorometric system and reported as % TUNEL-positive myocyte nuclei compared to DAPI-stained myocyte nuclei. An average of 1500 nuclei were examined per heart. Caspase-3 cleavage was measured using immunohistochemistry with antibody from Cell Signaling together with the VectaStain ABC-alkaline peroxidase kit 23. Fibrosis was measured using picrosirius red staining with pixel counting to quantitate % collagen.
Whole-cell current- and voltage-clamp recordings were obtained from myocytes isolated from the left ventricular (LV) apex of adult (10–12 week) mouse hearts within 12 hrs of cell isolation as previously described 25, 26.
Membrane fractions from mouse hearts were analyzed using primary antibodies from Abcam (Kv4.3, ab65794; KChIP2, ab66741) or Millipore/Upstate (Kv4.2, #07-491). HRP-coupled secondary antibodies were from Cell Signaling. Expression was quantitated using ImageJ (NIH) and normalized to Ponceau-S loading.
All data are presented as means ± SEM (standard error of the mean). The statistical significance of differences between groups was assessed using Student’s t test, or when indicated, ANOVA. P values are presented in the text and Figures.
Some, but not all miR profiling studies, have reported that miR-133a is downregulated in murine cardiac hypertrophy 9–12, 27. Absence of uniformity in these results may derive from different experimental models and miR-133a assays. Therefore, we measured miR-133a by RT-qPCR in four forms of murine hypertrophy matched for age and genetic background. Compared to non-transgenic (ntg) mice, myocardial miR-133a levels decreased by 50% one week after surgical transverse aortic coarctation (TAC), but returned to normal by three weeks (Figure 1a, left panel). miR-133a levels also decreased with continuous isoproterenol infusion (Figure 1a, middle panel), but not in genetic hypertrophies caused by cardiomyocyte expression of constitutively activated phosphatidyl-inositol-3-kinase (caPI3K) 28 or Gαq 29 (Figure 1a, right panel). Indeed, distinct regulation of miRs in PI3K and Gαq hypertrophy was the rule (Figure 1b and Supplemental Table S1). Gαq overexpressing and TAC hearts shared ~40% of their regulated mRNAs, revealing similar “pathological” transcriptional profiles, whereas caPI3K was more like ntg (Figures 1c and 1d). These and prior findings 10 suggest that miR-133a downregulation is a transient response to reactive hypertrophy; experimental manipulation of miR-133a may therefore best be evaluated after a temporally-defined hypertrophic stimulus.
A goal of the current studies was to evaluate whether downregulation of miR-133a after TAC contributes to pathological aspects of pressure overload hypertrophy 16, 21. We used the MYH6 (αMHC) promoter to drive expression of a 735 bp genomic precursor of miR-133a-114 (Figure 2a). Northern blotting and reverse transcription-polymerase chain reaction (RT-PCR) showed abundant expression of the transgenic miR-133a genomic precursor (Figure 2b). As these two assays examine expression of the miR-133a transgene, we performed RT-qPCR of the mature 21 base miR-133a to determine that the level of overexpression was ~13-fold ntg (Figure 2c).
Reported functions of miR-133a include controlling embryonic cardiac growth and development 15. However, with postnatal overexpression, heart size and weight, echocardiographic left ventricular mass, and cardiomyocyte cross sectional area (CSA) were normal (Figure 2d and white bars in Figure 3a). Myocardial histology (not shown) and cardiac function were normal in αMHC-miR-133a mice, assessed as echocardiographic fractional shortening (64+/−1% vs 62+/−1% ntg, n=8 pairs), and by invasive hemodynamics (Figure 2e). Thus, increasing miR-133a in postnatal mouse hearts did not alter normal cardiac growth.
Another reported function for miR-133a is regulation of ion channels related to electrocardiographic (ECG) QT prolongation 19, 20. We compared signal-averaged ECGs in ntg and αMHC-miR-133a mice. Heart rate, P wave duration, PR interval, or QRS duration were normal (Table 1), but QT duration corrected for heart rate (QTc) was significantly increased in αMHC-miR-133a hearts (Table 1, Figure 2f). Accordingly, our subsequent studies in TAC hypertrophy included a careful analysis of cardiac repolarizing channels (see below).
We determined the consequences of miR-133a on reactive cardiac hypertrophy by performing TAC on 12-week old mice (transverse aortic gradient of 73+/−2 mmHg in αMHC-miR-133a mice, 76+/−4 mm Hg in ntg). Whereas miR-133a is normally downregulated by ~60% after TAC (see Figure 1a), miR-133a in TAC αMHC-miR133a hearts was maintained at 2.8±0.3-fold that of non-operated controls.
Figure 3a–c (black bars) shows the results of hypertrophy measures 3 weeks after TAC in ntg and αMHC-miR-133a mice (white bars are non-operated). There were no differences in heart weight indexed to tibial length (Figure 3a, left), echocardiographic left ventricular mass (Figure 3a, middle), or CSA (Figure 3a, right). Left ventricular ejection performance decreased to a similar extent (Figure 3b, left), and peak positive and negative dP/dt of TAC hearts were also similar (Figure 3b, middle and right).
mRNA profiling using Affymetrix mouse arrays showed generally comparable mRNA signatures in non-operated αMHC-miR-133a and ntg hearts. Only 12 of 12,148 cardiac-expressed transcripts were increased >1.3-fold in miR-133a hearts13, and only 25 were decreased by the same amount (Figure 3d, Supplemental Table S2). Roughly equal numbers of mRNAs were upregulated (n=155) and downregulated (n=173) by any amount at p<0.05 in miR-133a hearts (Supplemental Table S3). Among these are the hypertrophy-associated genes, Mef2 (decreased by 30%; P=0.00002), Myh7 (beta-myosin heavy chain, βMHC, decreased by 27%; P=0.0121) and atrial natriuretic peptide (Nppa, increased approximately 2-fold; P=0.037). Also significantly downregulated was the Ito,f K channel accessory subunit KChIP2 (Kcnip2, decreased by 27%, p=0.0132). RT-qPCR showed no significant effects of miR-133a on regulated expression of the usual molecular markers of hypertrophy after TAC (Figure 3c). Importantly, when the post-TAC mRNA signatures were compared, the identities of those mRNAs that were most upregulated and downregulated by TAC did not change with miR-133a overexpression (Figure 3d). Thus, miR-133a overexpression has relatively subtle effects on baseline and hypertrophied cardiac mRNA expression profiles.
As with TAC, there was no difference in heart weight, echocardiographic left ventricular mass, or left ventricular ejection performance between isoproterenol-treated ntg and αMHC-miR-133a mice (not shown).
Myocardial fibrosis occurs after TAC; picrosirius red staining of myocardial collagen increased ~3-fold after TAC in ntg, but not in miR-133a hearts (Figure 4a). The clue that miR-133a might regulate myocardial fibrosis was its reported downregulation of connective tissue growth factor (CTGF) 12. However, CTGF mRNA levels actually increased <2-fold (P=0.0046, Supplemental Table S3) in miR-133a hearts. Accordingly, we examined another mechanism for TAC-associated fibrosis, cardiomyocyte apoptosis 23, 30. Non-transgenic mice showed a ~3-fold increase in cardiomyoctye TUNEL positivity (to ~1%) one week after pressure overloading, whereas TUNEL positivity did not increase in TAC αMHC-miR-133a hearts (Figure 4b, left). Caspase 3 activity qualitatively paralleled the TUNEL results (Figure 4b, right panels). Although described in other contexts 15, 17, we found no change in myocardial SRF, caspase 9, or cyclin D2 mRNA (Supplemental Table S3) or protein (Supplemental Figure S1) by miR-133a in adult hearts.
TAC of ntg and miR-133a mice did not produce mortality through 4 weeks. However, fibrosis contributes to loss of ventricular compliance, and absence of myocardial fibrosis in TAC miR-133a hearts could therefore improve diastolic function. Indeed, the typical increase in left ventricular end-diastolic pressure (LVEDP) after TAC was diminished with miR-133a overexpression, and pressure-volume analysis revealed improved myocardial stiffness in miR-133a hearts (Figures 4c and 4d). Together, these studies reveal that preventing “normal” downregulation of miR-133a after pressure overloading can decrease fibrotic myocardial remodeling and improve diastolic performance, without altering the extent of hypertrophy.
Whole-cell patch clamp recordings from myocytes isolated from the apex of the left ventrices31–33 revealed that action potential durations (APDs) were significantly longer in αMHC-miR-133a than ntg (Figures 5a and 5b), similar to QT prolongation (Figure 5c). As previously observed 33, action potentials and QTc are also prolonged one week following TAC (Figures 5a–c). Indeed, the effects of TAC were greater than those of miR-133a (Figures 5a and 5b). In contrast, resting membrane potentials and action potential amplitudes were similar in ntg and αMHC-miR-133a, control and TAC, myocytes (Figure 5d).
To examine the relationship between observed repolarization abnormalities and repolarizing K+ currents, whole-cell voltage-clamp recordings were obtained. Although waveforms of voltage-gated outward K+ (Kv) currents in miR-133a cardiomyocytes were similar to those of ntg cells (Figure 5e), peak Kv current densities were significantly lower (Figure 5f, white bars). To quantify Ito,f, IK,slow and Iss expression, the decay phases of the Kv currents were fitted with the sum of two exponentials, and the amplitudes and τdecay values of individual Kv components derived 31–33. IK,slow (Figure 5h, white bars) and ISS densities (Figure 5g, white bars) were similar, but Ito,f density was significantly lower in miR-133a myocytes (Figure 5i, white bars).
Ventricular Kv current densities are decreased after TAC (Figure 5f, black bars), reflecting reduced Ito,f, IK,slow and ISS densities34 (Figures 5g,h, and i, black bars). (TAC does not affect τdecay values for Ito,f and IK,slow, or voltage-dependent properties of Kv currents34.) Cardiac repolarization is orchestrated by several K currents, and TAC-associated decreases in IK,slow and ISS densities were similar between ntg and miR-133a cardiomyocytes (Figures 5g and 5h), consistent with similar changes in APD and QTc. Intriguingly, miR-133a overexpressing cells were protected against the hypertrophy-induced decrease in Ito,f (Figure 5i).
Potential mechanisms underlying Kv current remodeling by miR-133a were explored. As noted above, Affymetrix microarrays had detected a ~27% decrease in Kcnip2 (P=0.0132, Supplemental Table S3) that encodes the Ito,f channel accessory subunit, KChIP238, but not in genes encoding other K channel subunits (Ito,f subunits Kv4.2 (Kcnd2) and Kv4.3 (Kcnd3); IKslow subunits Kv1.5 (Kcna5) and Kv2.1 (Kcnb1); and pore-domain K+ (K2P) channel subunits TASK1 (Kcnk3), TASK2 (Kcnk5) and TREK1 (Kcnk2) in Iss channels). Both the positive and negative microarray findings were confirmed by RT-qPCR (Figures 6a and 6c). Immunoblot analysis of two independent cohorts of ntg and miR-133a cardiac membranes at baseline showed that protein content followed both the respective mRNA levels and current densities (Figures 6b and 6d), and in a single cohort of TAC mice, confirmed that KChIP2 levels did not change after pressure overloading of miR-133a hearts.
Fulfilling the promise of miR-directed therapeutics requires knowledge of the consequences of individual cardiac-expressed miRs under normal and pathological circumstances. Accumulating evidence supports dynamic regulation of specific miRs in human heart disease 13, 27, 41–44. However, it has been difficult to establish consequences of modified miR expression. Our results suggest that miR-133a can be a double-edged sword, depending upon pathophysiological context. Increasing already abundant miR-133a 13-fold had no measurable effect in normal postnatal hearts except to modestly delay electrical repolarization. Preventing miR-133a downregulation in actively hypertrophying hearts protected against myocardial fibrosis and TAC-mediated downregulation of Ito,f. These results indicate that miR-133a has a defined range of expression that is essential to preventing pathology, and suggest that the therapeutic window for any miR-directed therapeutic may be narrow. A limited physiological range for miR-133a expression seems also to exist for cardiac development, as miR-133a gene ablation and embryonic cardiac overexpression both cause lethal cardiac structural abnormalities 15.
Our experimental design avoided cardiac developmental effects. Indeed, the only abnormality of non-stressed adult hearts overexpressing miR-133a was modest QT prolongation and down-regulation of Ito,f, a prominent repolarizing Kv current in adult mouse ventricle. This is the first demonstration of miR-133a-mediated regulation of cardiac Kv4-encoded Ito,f channels. Regulation of ERG-encoded IKr channel expression by miR-133a has also been suggested in diabetic rabbit hearts and transfected human SKBr3 cells 20. However, the contribution of IKr to action potential repolarization in the intact mouse ventricle is negligible 36 and modulation of Ito,f by miR-133a is more likely to contribute to observed changes in action potential duration 25, 26.
Kv4-encoded Ito,f (Kcnip2) has not been previously reported as a target of miR-133a, and is not suggested to be an miR-133a target in PicTar, TargetScan or miRanda. Interestingly, Kcnip2 is predicted by TargetScan to be a target of the miR-29 family of miRs, whose upregulation also decreases the extent of fibroblast collagen production 45. These findings suggest that regulation of Kcnip2 transcript levels in miR-133a transgenic hearts occurs by an indirect mechanism.
Because miR-133a levels decrease during reactive cardiac hypertrophy, some have considered that restoring miR-133a levels could suppress hypertrophic remodeling. 3, 21 Care, et al 16 reported that miR-133a levels declined in experimental and human cardiac hypertrophy, and that forced miR-133a overexpression reduced hypertrophy of cultured cardiomyocytes and Akt transgenic mice. Here, we found that postnatal cardiomyocyte miR-133a overexpression did not alter reactive hypertrophy in pressure overloaded or isoproterenol-treated hearts, although it protected against myocardial fibrosis. There are a number of possible explanations for differing results of miR-133a between the two studies. Rather than a genetic hypertrophy model, we chose transverse aortic banding as a model of pressure overload hypertrophy, and isoproterenol as a model of failure-associated hypertrophy, because we could induce hypertrophy at will, and because of their relevance to human disease. Furthermore, because miR-133a is normally expressed only in cardiac and skeletal muscle 6–8, we utilized cardiomyocyte-specific overexpression; our results represent cell-autonomous effects. Finally, our conceptual approach was to prevent the “normal” down-regulation of miR-133a induced by an acute hypertrophic stimulus.
Given that miRs act in part through destabilization of target mRNAs 3, we were surprised that miR-133a overexpression had minimal effects on myocardial mRNA signatures. Even using selection criteria (P<0.05) that sacrificed specificity for increased sensitivity, only 328 of over 22,000 mRNAs were regulated in miR-133a transgenic hearts. The expected result of increased miR-133a is to decrease steady-state levels of its mRNA targets. However, almost equal numbers of transcripts were downregulated (173; 53%) as were upregulated, and TargetScan identified miR-133a binding sites in only 19 regulated mRNAs (5 upregulated and 14 downregulated; P=0.10). Indeed, expression of the validated miR-133a target, CTGF, which is known to be downregulated by forced expression of miR-133a in cultured fibroblasts 12, was increased two-fold in miR-133a hearts. Other regulated hypertrophy-associated transcripts, Myh7 (decreased by 35%) and Nppa (increased by 2.3-fold), also do not contain miR-133a binding sites, and so we believe their regulation, and that of most mRNAs, was indirect.
One noteworthy miR-133a-regulated mRNA is the muscle-specific transcription factor Mef2, which was downregulated by ~30%. MEF2 positively regulates miR-133a 8, 15. Its downregulation in miR-133a hearts therefore negatively feeds back on cardiac gene expression directed by MEF2 46 and provides another example of indirect regulation by this miR. It is also increasingly evident that the dominant mechanism of miR action in vertebrates is translational silencing, and not mRNA destabilization 47, 48. For these reasons, the majority of mRNA targets cannot be reliably identified through examination of steady-state mRNA levels, even when the data are informed by bioinformatics detection of putative miR binding sites.
Sources of Funding. Supported by NIH NHLBI P50-HL077101 and R01-HL034161, UL1 RR024992 from the National Center for Research Resources and an American Heart Association Postdoctoral Fellowship to WW.
Subject codes: 138 (cell signaling/signal transduction), 145 (genetically altered mice), 152 (ion channels), 131 (apoptosis), 155 (physiological and pathological control of gene regulation)