Circulating MicroRNA-30d Is Associated With Response to Cardiac Resynchronization Therapy in Heart Failure and Regulates Cardiomyocyte Apoptosis: A Translational Pilot Study

Abstract

Background: Biomarkers that predict response to cardiac resynchronization therapy (CRT) in heart failure patients with dyssynchrony (HFDYS) would be clinically important. Circulating extracellular microRNAs (miRNAs) have emerged as novel biomarkers that may also play important functional roles, but their relevance as markers for CRT response has not been examined.

Methods and results: Comprehensive miRNA polymerase chain reaction arrays were used to assess baseline levels of 766 plasma miRNAs in patients undergoing clinically indicated CRT in an initial discovery set (n=12) with and without subsequent echocardiographic improvement at 6 months after CRT. Validation of candidate miRNAs in 61 additional patients confirmed that baseline plasma miR-30d was associated with CRT response (defined as an increase in left ventricular ejection fraction ≥10%). MiR-30d was enriched in coronary sinus blood and increased in late-contracting myocardium in a canine model of HFDYS, indicating cardiac origin with maximal expression in areas of high mechanical stress. We examined the functional effects of miR-30d in cultured cardiomyocytes and determined that miR-30d is expressed in cardiomyocytes and released in vesicles in response to mechanical stress. Overexpression of miR-30d in cultured cardiomyocytes led to cardiomyocyte growth and protected against apoptosis by targeting the mitogen-associated kinase 4, a downstream effector of tumor necrosis factor. In HFDYS patients, miR-30d plasma levels inversely correlated with high-sensitivity troponin T, a marker of myocardial necrosis.

Conclusions: Baseline plasma miR-30d level is associated with response to CRT in HFDYS in this translational pilot study. MiR-30d increase in cardiomyocytes correlates with areas of increased wall stress in HFDYS and is protective against deleterious tumor necrosis factor signaling.

Keywords: biological markers; heart failure; microRNAs; stress, biological.

Figure 1

Differential expression of miRNAs between CRT responders and non-responders in validation cohort. A. Box plot of change in LVESV or LVEF from baseline to follow-up study as measured by available volumetric data (N=29). (LVEF measured in same 29 patients with standard two-dimensional or visual analysis.) P values reflect Wilcoxon testing. B. miR 142-5p, miR-30d, and miR-766 are significantly increased in responders versus non-responders. C Only miR-29c was significantly increased in responders relative to non-responders. **denotes p<0.05, Benjamini Hochberg correction for multiple hypothesis testing. D-G, Levels of miRNAs identified in A-B in responders, non-responder or control patients. miRNA levels are in arbitrary units, normalized to spike-in C. Elegans miR-39. Ctrl: control patients, NR: non-responders, R: Responders.

Figure 2

Plasma miRNAs as biomarkers of response to CRT. A. Univariate logistic regression to determine miRNAs and clinical variables associated with favorable LV function post-CRT (increase in LVEF ≥ 10%). B. Correlation between EF increase and miR-30d level. Associations is statistically significant with Spearman coefficient of 0.39, p=0.01 for miR-30d. Correlation between miR-30d and change in EF remained significant with exclusion of either outlier point (p=0.01), but was borderline significant with exclusion of both extreme values (Spearman rho=0.31, p=0.058). C. Probability of 10% improvement in EF plotted against log (miR-30d concentration) Logistic regression remained significant even after removal of extreme miR-30d values noted in 2C (p=0.03). D. ROC curve for prediction of CRT response (as assessed by 10% improvement in EF) by miR-30d levels compared to prediction of response by QRS duration. E miR-30d levels significantly decrease by six months with CRT response but are unchanged in non-responders among 21 additional heart failure patients.

Figure 3

Origin and regulation of miR-30d. A. miR-30d levels normalized to C-elegans spike-in miR-39) in samples from the coronary sinus and peripheral plasma in the same human subjects at the time of CRT implantation (n=6). B. Cellular miR-30d level in cultured neonatal rat ventricular CMs versus cardiac fibroblasts (normalized to U6snRNA). C. miR-30d is significantly enriched in the extracellular vesicle fraction (purified by ultracentrifugation followed by FACS sorting) relative to cellular lysates (n=3 experiments). Inset FACS sorting of EV fraction following ultracentrifugation identifies a population of acridine orange positive particles with size <200nm. D. miR-30d levels in the myocardium in a canine model of dyssynchronous heart failure (HF_DYS) with subsequent resynchronization (CRT) compared to control. MiR-30d levels were measured in both the septal and lateral walls in control, HF_DYS and CRT dogs. MiRNA levels were compared in each individual canine lateral versus septal wall by paired t-test, and between CRT, DHF and control animals by unpaired-t test. * denotes p<0.05 relative to control septum; *** denotes p<0.05 relative to DHF septum. miR-30d levels were normalized to U6snRNA (see Fig. S4). E. After 6 hours of cyclic stretch at 10Hz on an elastic membrane, miR-30d levels are unchanged in CM cell lysate (n=4 experiments), but are significantly increased in the EV fraction. F. After 24 hours of stretch, levels of miR-30d remain elevated in the EV fraction. Levels of 30d are normalized to U6snRNA. For both EV fractions and cell lysates, fold change relative to unstretched fraction (EV or cell lysate) is shown. For all subfigures, ** denotes p<0.05 by T-test for normally distributed samples, and by Mann-Whitney for all others unless otherwise specified.

Figure 4

miR-30d mediates adaptive hypertrophy in cultured CMs. A. Cultured rat CMs were transfected with either a sham or a miR-30d mimic using a liposomal delivery system, stained with α-actinin, and imaged with confocal microscopy. CM size was traced and quantitated using Image J software. Bar graph shows average of 150-290 cells in each group. *: p < 0.05. Scale bar: 50 μm. B. mRNA levels of gene markers of hypertrophy/fibrosis as assessed by qRT-PCR. Results were normalized to house-keeping gene HPRT. **: p<0.05, n=3 experiments. C. miR-30d effects on hypertrophic signaling pathways: Protein lysates from CMs subjected to miR-30d mimic or scramble transfection for 24 hrs were harvested and assessed for levels of phospho- and total levels of kinases implicated in cardiac hypertrophy as indicated. n=4 independent experiments for each, ** denotes p<0.05 for comparison of pAkt in miR-30d and scramble transfected cells.

Figure 5

LIMMA and IPA analysis identifies potential miR-30d gene targets. A. Schema of bioinformatics analysis to identify miR-30d targets in HF_DYS. Microarray gene expression data from the HF_DYS and CRT lateral walls was subjected to LIMMA analysis to generate a set of differentially expressed (DGE Genes) genes. Within this set miR-30d putative targets were identified using Tarbase and Targetscan algorithms. IPA (Ingenuity Pathways Analysis) was then used to generate a set of predicted signaling pathways. B. List of pathways ranked by predicted magnitude of change. The grey box to the right identifies the putative miR-30d target gene. Bold and underlined molecules were selected for further investigation, based on the signaling pathways involved and/or their involvement in multiple pathways. C. miR-30d target genes of interest were evaluated in neonatal rat CMs that were transfected with scramble or miR-30d mimic. *: p<0.05. D. Western blot analysis and quantitation of MAP4K4 protein levels in the dog dyssynchronous heart model (average of n=4 dogs). Inset shows a representative Western blot. Levels were normalized to GAPDH.

Figure 6

miR-30d is cardio-protective against apoptosis by modulating signaling downstream of TNF-α. A. MAP4K4 mRNA levels in cultured rat CMs transfected with miR-30d mimic or scramble control and treated with TNF-α (25 ng/mL for 24 hours) (n=5 experiments), left panel; and level of apoptosis, as assessed by cell death assay ELISA kit in CMs transfected with miR-30d mimic or scrambled control and treated with TNF-α as described above (n=4 experiments), right panel. **: p<0.05 for comparison of TNF-α treated and untreated scramble transfected cells. B. Levels of ANP, β-MHC and TIMP mRNA levels in cultured rat CMs transfected with miR-30d mimic or scrambled control treated as in (A). C miR-30d levels are correlated with baseline TnT levels, r=−0.51, p=0.0001 for the association. D. Schematic for our proposed mechanism of miR-30d action in HF_DYS.