Exercise Training in Diabetes: Start Earlier or Exercise Harder

Cardiac dysfunction and coronary vascular dysfunction are major complications of diabetes and exercise training appears effective in protecting against these cardiovascular complications¹. However, in most studies, exercise is initiated after cardiovascular complications are evident. It remains unclear whether early exercise training, before cardiac and coronary dysfunction, or higher intensity exercise regimens, would be more effective in preventing these cardiovascular complications.

In this issue of Circulation Research, Lew et al.² investigated the effects of exercise intensity and timing on cardiac and coronary structure and function in diabetes. Diabetic mice were subjected to moderate- (MIE) and high- (HIE) intensity exercise beginning at different stages of disease. At eight weeks of age, db/db diabetic mice displayed an obese phenotype with hyperinsulinemia and hyperglycemia but no cardiac dysfunction or impairment of coronary perfusion. By 16 weeks of age, db/db hearts exhibited systolic and diastolic dysfunction and structural remodeling which was associated with impaired coronary perfusion and dysfunction. Lew et al.² found that both MIE and HIE initiated at eight weeks in db/db diabetic mice were effective in preventing cardiac dysfunction and remodeling as well as coronary dysfunction assessed at 16 weeks. However, when initiated at 16 weeks, only HIE, but not MIE, mitigated these phenotypes. These results indicate that exercise is beneficial to cardiac and vascular function in diabetes, and these benefits depend on when exercise is initiated as well as training intensity (Figure). The authors are to be congratulated on rigorously delineating the impact of these crucial parameters, which have not previously been defined.

Figure.

Schematic illustrating the change in the exercise intensity required for cardiovascular benefits as a function of diabetes progression, along with possible mechanisms underlying the cardiovascular benefits of exercise in diabetes.

Another important question addressed by Lew et al. concerns the molecular mediators of exercise-benefits. MicroRNAs (miRNAs) are short noncoding RNAs that have been implicated in cardiovascular disease and identified as mediators of the cardiovascular benefits of exercise^{3, 4}. However, whether the roles of these miRNAs differ for different exercise intensities and at different stages of cardiac dysfunction during the development of diabetes, is unclear. In this study, Lew et al.² focused on miRNAs previously implicated in regulating angiogenesis (miR-126)⁵, apoptosis (miR-499)⁶, and fibrosis (miR-15b and miR-133)^{7, 8}. They found that these miRNAs were down-regulated in db/db hearts at the age of 16 weeks, and this was associated with up-regulation of their respective downstream targets, including anti-angiogenic SPRED1/VEGF (miR-126 target), pro-apoptotic cleaved caspase-3 (miR-499 target), pro-fibrotic CTGF (miR-15b target) and TGF-β (miR-133a target). These results suggest that reduction of these miRNAs and the subsequent up-regulation of their targets could contribute to the cardiac and coronary dysfunction in this model of diabetes. Interestingly, circulating miR-126 was also reduced in diabetic mice, and positively correlated with coronary arteriole density and capillary density, suggesting that circulating miR-126 could potentially serve as a biomarker of cardiovascular complications. Lew et al. further found that, when exercise was initiated in eight-week-old diabetic mice, both MIE and HIE were able to increase cardiac expression of these miRNAs and down-regulate their targets. Interestingly, when exercise was started at the age of 16 weeks, although both MIE and HIE were able to increase these miRNAs, down-regulation of their downstream targets was only seen in HIE. These results indicate that miR-126, miR-499, miR-15b, and miR-133a may serve as mediators of exercise cardioprotection in diabetic hearts and suggest that this effect is intensity-dependent (Figure). Of note, the four candidate miRNAs studied were selected from previous reports. There may well be other miRNAs contributing that could be uncovered through comprehensive transcriptional profiling. In addition, the authors showed that exercise reduced cardiomyocyte and endothelial cell apoptosis in diabetes. However, exercise has also been shown to induce cardiomyocyte and endothelial cell proliferation^9–11, which was not examined in the current study and could also be contributing to the benefits observed.

To further confirm the cardioprotective role of exercise-regulated miR-126 in diabetes, the authors enhanced (by injecting miR-126 mimic) or inhibited (by injecting miR-126 anti-miR) miR-126. In eight-week-old diabetic mice, silencing of miR-126 led to a reduction in miR-126 in the heart and circulation, resulting in loss of MIE’s beneficial effects on coronary vasculature in diabetic mice. In contrast, in 16-week-old diabetic mice, MIE combined with overexpression of miR-126 produced coronary benefits similar to those observed with MIE in 8-week-old diabetic mice. Interestingly, this combination of MIE and miR-126 silencing or overexpression had a limited effect on the heart, with inhibition diminishing the anti-fibrotic effect but failing to block the MIE-induced improvement in cardiac function, and overexpression reducing fibrosis and improving certain cardiac functional parameters. This suggests miR-126 is required for the beneficial effect of exercise on the coronaries but not the heart. Of note, in this study, anti-miR inhibition of miR-126 led to a significant reduction in miR-126 in the heart and circulation, not only blocking the MIE-induced increase but decreasing expression to below the diabetic animals’ baseline. It is unclear whether the loss of MIE benefits after miR-126 inhibition is mainly due to the cancellation of MIE-induced miR-126 or due to this drastic reduction of miR-126. It would be interesting to examine the effect of miR-126 inhibition in the absence of exercise. Similarly, in 16-week-old diabetic mice, overexpression of miR-126 in combination with MIE mimics the protective effects of HIE on the coronary but not on the heart. However, the level of miR-126 in the heart achieved by the combination of miR-126 mimic and MIE was much lower than that seen in HIE (10-fold in mimic + MIE vs. 20-fold in HIE), despite comparable plasma level of miR-126. The dose relationship of exercise-mediated cardiovascular benefits to miR-126 levels remains unclear.

This interesting work² provides new insights into the cardiovascular benefits of exercise in diabetes. First, the effects of exercise on the heart and coronary depend on timing of initiation and intensity (Figure). Specifically, after the onset of dysfunction in the heart and coronary in diabetes, higher intensity exercise is needed for full cardiovascular benefit likely due in part to severe reduction of four miRNAs (i.e., miR-126, miR-499, miR-15b, and miR-133a) after the onset of cardiac and coronary dysfunction in diabetic hearts. However, if initiated earlier, before the establishment of cardiac and coronary dysfunction in diabetes, even moderate intensity exercise is effective in preventing dysfunction in the heart and coronary during the development of diabetes. Second, this study provides evidence that four miRNAs with various effects on cardiomyocyte apoptosis, fibrosis and angiogenesis contribute to the beneficial effects of exercise in diabetes. Among them, an important role for miR-126 was suggested by overexpression and inhibition studies. Third, circulating miR-126 was shown to correlate with capillary and arteriolar density, but whether this is a clinically useful biomarker was not addressed. Nevertheless, this work highlights the concept that exercise models can serve as a platform for identifying new candidate therapeutic targets in combating diabetes.

As with all interesting and provocative studies, multiple questions are raised by the Lew et al.² study that warrant further investigation. Since the db/db mouse is a severe, genetic model of diabetes, it is unclear whether the effects of exercise seen in the current study would be similar in other models more characteristic of common forms of diabetes seen clinically, such as diet induced obesity. It is possible that in these milder models the benefits of moderate exercise would extend to later time-points. Although moderate-intensity exercise initiated early in diabetic mice was beneficial, the ‘dose’ and nature of exercise required to achieve this effect remains unclear, although it is noteworthy that we found cardiac miR-126 is induced in multiple forms of exercise³. The age of animals used in this study was also relatively young, while diabetes is more prevalent in older adults¹², who commonly manifest reductions in cardiac function¹³, and it will be important to assess whether similar benefits apply to older animals. Indeed, previous work from our group has shown that exercise can improve cardiac function in older, non-diabetic mice, and even reverses age-related changes in cell-cycle and other signaling pathways¹³.

It will also be important to evaluate the clinical relevance of these observations, and the feasibility of implementing comparable exercise interventions in patients. Extrapolating from the HIE and MIE protocols used here to humans, may prove challenging. Even in the absence of overt cardiovascular complications, diabetic subjects often have limited exercise capacity¹⁴ and achieving sustained adherence to lifestyle modifications is notoriously challenging, which could limit the feasibility of implementation.

These mouse studies suggest that circulating miR-126 could be a potential biomarker of structural and functional changes in diabetic heart disease. An important question for future studies is whether the same correlation can be detected in humans and whether miR-126 is sensitive and specific enough to serve as a clinically useful biomarker. It is also worth exploring the upstream effectors governing the different response of the heart and coronary to MIE and HIE, and whether factors previously demonstrated to be important in the beneficial effects of exercise¹³ may be involved in this process. The finding that moderate exercise restores, at least in part, miRNA expression regardless of when it is initiated, yet only rescues cardiac function and miRNA target protein expression when initiated early raises the possibility of additional critical changes in signaling during disease progression, in addition to the downregulation of these miRNAs. What these changes are, and the mechanism by which HIE overcomes them, remains unclear, although it is tempting to speculate on the possible roles of long non-coding RNAs and circular RNAs, which have been shown to regulate miRNA and to contribute to cardiovascular disease.

Despite these issues, Lew and colleagues have admirably provided evidence of the impact of the onset of cardiac and coronary dysfunction on the efficiency of exercise-induced cardiovascular benefits in diabetes. Their observations provide valuable insight into the molecular correlates of diabetic heart disease and mechanisms of exercise’s benefits, with possible implications for both therapeutic intervention and detection of diabetic heart disease. Exploring the clinical implications of this work for the titration of exercise interventions and potential other therapeutic interventions in diabetic patients will be exciting avenues for future investigation.