Early dysregulation of cardiac-specific microRNA-208a is linked to maladaptive cardiac remodelling in diabetic myocardium

The human myocardium study was approved by the Health and Disability Ethics Committee of New Zealand and the Human Ethics Committee at the University of Otago, New Zealand. All the patients provided written consent for the collection and use of samples in this study. The animal study using type 2 diabetes was approved by the Animal Ethics Committee at the University of Otago, New Zealand.

Right atrial appendage (RAA) and epicardial left ventricular (LV) biopsies were collected from type-2 diabetic ischaemic heart disease (D-IHD) (n = 8–10) and non-diabetic ischaemic heart disease (ND-IHD) (n = 10) patients with preserved ejection fraction undergoing on-pump coronary artery bypass graft surgery. Myocardial tissue samples from non-diabetic healthy (ND-H, n = 5) were collected from cadavers without any known history of ischemic heart disease to be used as healthy controls. Tissue samples were snap-frozen and stored at − 80 °C immediately after collection for molecular analysis.

BKS.Cg-m +/+Leprdb/J mice (db/db) and their non-diabetic C57BL/ksJ-lepr + littermates (db/+) (Jackson laboratory), were used as the model of type 2 diabetes and age-matched controls respectively. Under terminal anaesthesia, myocardial tissue samples were collected from both the groups at 7 different time points (8, 12, 16, 20, 24, 28, 32 weeks (W) of age, n = 10 each. Immediately after excision, the hearts were snap frozen in liquid nitrogen for molecular analyses.

Evolution of cardiac dysfunction in the diabetic myocardium was monitored every 4 weeks from 8 to 32 weeks of age in mice by measuring the changes in cardiac function using a Vivid E9 cardiovascular ultrasound system (GE Vingmed Ultrasound, Horten, Norway). Left ventricular wall thickness, systolic function (internal diameters, ejection fraction, fractional shortening) and diastolic function (E/A ratio, deceleration time (DecT)) were assessed as previously described [10‐13]. LV mass was calculated from both mouse and human echocardiography data using the following equation: LV mass (mg) = [0.8 (1.04[([LVAWd or IVSd + LVIDd + LVPWd]³ − LVIDd³)]) + 0.6], where, LVAWd is left ventricular (LV) anterior wall thickness during diastole, IVSd is interventriculat septal thickness during diastole, LVIDd is LV internal diameter during diastole and LVPWd is LV posterior wall thickness during diastole [14].

Correlations between miR-208a and LV mass was done using Pearson′s correlation equations. A probability value (P value) less than 0.05 was considered statistically significant.

Clinical characteristics of the study participants is illustrated in Table 1. Diabetic patients had significantly higher HbA1c compared to the non-diabetic patients undergoing surgery. Quantitative RT-PCR analysis showed significant upregulation of miR-208a in both RAA and LV regions of human type 2 diabetic myocardium in patients with IHD (P < 0.0001 vs. ND-IHD, Fig. 1a). Intriguingly, while ischemia alone (ND-IHD) significantly upregulated the expression of miR-208a (P < 0.001 and P < 0.0001 vs. ND-H, Fig. 1a), diabetes further upregulated the expression level of miR-208a in both RAA and LV (P < 0.05 and P < 0.01 vs. ND-IHD, Fig. 1a), thereby demonstrating the independent detrimental effects of diabetes on myocardium. Our previous study showed a similar increase in miR-208a in human type-2 diabetic heart with normal ejection fraction (EF) [23], however,  this study also included samples from patients with reduced EF and myocardial samples from those died due to non-cardiac reason. Echocardiography showed a non-significant increase in LV mass in the diabetic heart (P = 0.07, Table 1), however, pearson correlation analysis showed a singificant positive correlation between miR-208a and LV mass irrespective of diabetes (P = 0.03, Fig. 1b). Diabetic heart also showed a marked upregulation of pro-hypertrophic protein β-MHC (Fig. 1c) while its negative regulator TR-α was downregulated (Fig. 1d).

Having demonstrated the dysregulation of cardio-enriched miR-208a in the human diabetic myocardium, our next question was to determine if there is any correlation of miR-208a and cardiac dysfunction. For this, we used a mouse model of type 2 diabetes and serially monitored the changes in cardiac function, miR-208a expression and its target proteins α- and β-MHC and TR-α expression. Echocardiography analysis showed a significant decrease in E/A ratio at 20 W (P < 0.0001, Fig. 2a). Furthermore, a pseudonormalized pattern of diastolic dysfunction was observed at 24 W, evident by E/A ratio > 1 along with decrease in DecT time when compared to age-matched non-diabetic littermates (Fig. 2a, b). Importantly, a restrictive filling pattern of LV was observed in 28 W and 32 W old diabetic mice, evident by a significant increase in E/A ratio > 2 (P < 0.001 vs. non-diabetic, Fig. 2a) as well as a marked reduction in DecT (P < 0.001 vs. non-diabetic, Fig. 2b).

M-mode echocardiography analysis revealed the onset and progression of LV systolic dysfunction in type 2 diabetic mice as evidenced by reduced ejection fraction (EF, Fig. 2c), fractional shortening (FS, Fig. 2d) and increased LV internal diameter at systole (LVIDs, Fig. 2e) and diastole (LVIDd, Fig. 2f) from 20 W of age worsening with the evolution of the disease (P < 0.001 vs. age-matched non-diabetic mice, Fig. 2c–f). There was a significant increase in LV mass in db/db mice at 16 W (Fig. 2k). With the progression of the disease there was a gradual decrease in anterior wall thickness during systole (LVAWs, Fig. 2g) and diastole [LVAWd, Fig. 2h)], posterior wall thickness in systole (LVPWs, Fig. 2I) and diastole [LVPWd, Fig. 2j)] and LV mass (Fig. 2k). This was also accompanied by a decrease in relative wall thickness (RWT) in diabetic mice, an index of cardiac remodeling (Fig. 2l).

To determine if changes in cardiac function are associated with dysregulation in the expression pattern of miR-208a, RT-PCR analysis was conducted at each time point. As shown in Fig. 3a, there was a marked upregulation of miR-208a in the diabetic heart as early as 8-weeks of age (Fig. 3a), when echocardiography showed normal cardiac function (Fig. 2). Intriguingly, miR-208a remained upregulated until the 16 W time point (Fig. 3a) after which there was a significant downregulation in its expression (Fig. 3a). The initial upregulation of miR-208a in the mouse  diabetic heart was consistent with its upregulation in expression which was  observed in human myocardium (Fig. 1a). Interestingly, like in the human heart, there was a significant positive correlation between miR-208a and LV mass, further supporting the role of miR-208a in hypertrophy (Fig. 3b).

To determine the functional significance of the differential expression of miR-208a, we next measured the expression levels of the contractile proteins α- & β-MHC in the type 2 diabetic mice hearts at 8 weeks, 16 weeks and 32 weeks of age by western blot analysis. β-MHC is a slow myosin contractile protein, and its increased expression is a common feature of pathological cardiac hypertrophy and remodelling.

Results showed a significant increase in α-MHC protein expression as early as 8 weeks in the diabetic heart (Fig. 3c) corresponding to an increase in expression of miR-208a (p < 0.0001 vs. non-diabetic, Fig. 3a). Follow-up analysis showed a significant decrease in the expression of α-MHC at 16 and 32 weeks of age in the diabetic heart (Fig. 3d). This was associated with a significant increase in the expression of pro-hypertrophic protein β-MHC at 16 and 32 weeks of age (Fig. 3e), indicating myosin switching in the contractile apparatus. Interestingly, we observed a sustained expression of miR-208a even after downregulation of α-MHC in the diabetic heart at 16 weeks of age  (Fig. 3a) which is in agreement with a previous report, suggesting a longer half-life of miR-208a relative to α-MHC [9]. Further, we also observed a marked decrease in the expression level of thyroid hormone receptor-α (TR-α) in the diabetic heart (Fig. 3d). Previous studies have suggested that miR-208a potentiates βMHC expression through a mechanism involving the TR-α. Of note, the marked upregulation of miR-208a in diabetic mice preceded the switch of α/β-MHC isoforms as observed at 16 weeks in the diabetic heart (Fig. 3).

Finally, to determine if therapeutic modulation of miR-208a can ameliorate the detrimental effects of high glucose on cardiac hypertrophy, we conducted a gain or loss of function study using HL-1 adult cardiomyocytes. Treating HL-1 cardiomyocytes with high glucose for 24 h and 48 h showed a significant upregulation of miR-208a at both the time points (Fig. 4a). Therefore, the 24 h time point was chosen to carry out all subsequent experiments on HL-1 cells. This also confirms that response of HL-1 cardiomyocytes to high glucose is comparable to that  observed in type 2 diabetic mouse and human hearts.

To confirm the pro-hypertrophic effect of miR-208a in adult cardiomyocytes, we next overexpressed miR-208a in normal glucose cultured HL-1 cardiomyocytes using miR-208a mimic under normal conditions. One advantage of this strategy was that it enabled us to determine if cardiac hypertrophy is due to the direct effect of miR-208a overexpression or a result of HG stress. The RT-PCR analysis confirmed upregulation of miR-208a, 24 h after transfection (Fig. 4b). Importantly, overexpression of miR-208a was sufficient to provoke cardiomyocyte hypertrophy (Fig. 4c).

Next to determine if inhibition of miR-208a activity in the high glucose treated cardiomyocytes is beneficial, high glucose treated cardiomyocytes were transfected with anti-miR-208a to inhibit its activity. As expected, silencing miR-208a activity significantly attenuated the increase in surface area of HL-1 cardiomyocytes reflecting cardiomyocytes hypertrophy in vitro (Fig. 4d).

Further, western blot analysis revealed a significant increase in the expression of the β-MHC protein in high glucose treated HL-1 cardiomyocytes, while the expression of α-MHC remained unchanged (Fig. 4e). Interestingly, inhibition of miR-208a activity prevented this increase in β-MHC (Fig. 4f) which was in parallel to the reduction in cardiomyocyte hypertrophy (Fig. 4d). Together, these data suggest that upregulation of miR-208a was required to activate the switch in the isoforms arrangement from α- to β-MHC and therefore, contributed to the development of cardiomyocyte hypertrophy. This was consistent with the data shown in diabetic mice heart, where the upregulation of miR-208a at 16 weeks was sufficient to activate the switch of α to β-isoform of MHC at 20 weeks.

Cardiac contractility is typically maintained by a tight balance between the fast isoform of myosin heavy chain (α-MHC) and the slow isoform of myosin heavy chain (β-MHC) levels [27]. Of note, one of the molecular hallmarks of pathological cardiac hypertrophy and remodelling is the upregulation of β-MHC. Experimental studies have demonstrated the specific induction of β-MHC mRNA and protein levels by disease-related hypertrophic stimuli in vivo and by hypertrophic agonists in culture [28‐30] but not by any physiological stimuli such as exercise [31]. A marked increase in the expression of β-MHC in the diabetic heart in our study, therefore, suggests the initiation of molecular alterations that induce hypertrophy. Previous reports demonstrated a crucial role for miR-208a in the stress-dependent hypertrophic growth of cardiomyocytes [9], and that miR-208a is essential to induce the upregulation of β-MHC by directly targeting thyroid hormone-associated protein 1 (THAP1) and myostatin, two negative regulators of muscle growth and hypertrophy [32]. However, this is the first study to demonstrate that marked upregulation of β-MHC is concordant with the upregulation of miR-208a in both the human and mouse type 2 diabetic heart. We further showed significant downregulation of TR-α, a downstream regulator of THAP1.

Another interesting finding of the current study was that the upregulation of miR-208a in the type 2 diabetic mouse heart preceded the activation of the fetal cardiac gene program and was associated with a switch of α/β-MHC isoforms with ensuing marked cardiac dysfunction. These findings are in line with existing reports where it was shown that the switch of α/β-MHC isoforms occurs in diabetic myocardium before the echocardiographic detection of ventricular hypertrophy [33, 34]. Consistently, we observed a similar isoform switch in type 2 diabetic mouse heart, although there was no clear evidence for hypertrophy in the diabetic heart by echocardiography. A steep increase in LV mass at 16 weeks of age in the diabetic heart was suggesting the heart is undergoing hypertrophic remodelling, however, this need to be cautiously interpreted as we did not observe any significant changes in the wall thickness. In contrast, the functional data from the human participants were clearly pointing towards hypertrophic remodelling in the diabetic heart showing a significant increase in interventricular sepatal thickness and posterior wall thickness during diastole which was associated with marked increase in the LV mass. Moreover, both the mouse and human study showed a positive correlation between miR-208a and LV mass. While more studies are required, these data clearly suggest a critical role of miR-208a in remodeling of the diabetic myocardium. Furthermore, using the in vitro cardiomyocytes, we also confirmed that the regulation of β-MHC was directly linked to miR-208a expression level. The cellular content of β-MHC was specifically increased in hypertrophic cardiomyocytes, and therapeutic inhibition of miR-208a activity was able to attenuate not only the hypertrophic response but also decreased the expression of β-MHC. These observations provide new insight into the mechanistic role of miR-208a in promoting cardiac remodelling through β-MHC-dependent hypertrophic activity, indicating activation of fetal cardiac gene programming in diabetic cardiomyocytes.

Given that miR-208a is encoded by an intron of the α-MHC gene [35], its expression is dependent on the expression of α-MHC. Intriguingly, in our study, the expression of the α-MHC gene was decreased at 16 W in the type 2 diabetic mouse heart, while miR-208a remained upregulated. This observation suggests a longer half-life of mature miR-208a even after downregulation of its host gene. This was supported by an earlier time course study showing persistent expression of miR-208a even after the downregulation of α-MHC [9]. Furthermore, in the compensated phase of in vivo studies, the expression of miR-208a was downregulated in the diabetic heart, while the expression of β-MHC persistently remained at a higher level. While the precise reason for this discrepancy is not known, it is possible that the early upregulation of miR-208a in the diabetic heart leads to persistent downregulation of TR-α thereby releasing its inhibitory switch on β-MHC.