Despite being first described 45 years ago, the existence of a distinct diabetic cardiomyopathy remains controversial. Nonetheless, it is widely accepted that the diabetic heart undergoes characteristic structural and functional changes in the absence of ischaemia and hypertension, which are independently linked to heart failure progression and are likely to underlie enhanced susceptibility to stress. A prominent feature is marked collagen accumulation linked with inflammation and extensive extracellular matrix changes, which appears to be the main factor underlying cardiac stiffness and subclinical diastolic dysfunction, estimated to occur in as many as 75% of optimally controlled diabetics. Whether this characteristic remodelling phenotype is primarily driven by microvascular dysfunction or alterations in cardiomyocyte metabolism remains unclear. Although hyperglycaemia regulates multiple pathways in the diabetic heart, increased reactive oxygen species (ROS) generation is thought to represent a central mechanism underlying associated adverse remodelling. Indeed, experimental and clinical diabetes are linked with oxidative stress which plays a key role in cardiomyopathy, while key processes underlying diabetic cardiac remodelling, such as inflammation, angiogenesis, cardiomyocyte hypertrophy and apoptosis, fibrosis and contractile dysfunction, are redox sensitive. This review will explore the relative contributions of the major ROS sources (dysfunctional nitric oxide synthase, mitochondria, xanthine oxidase, nicotinamide adenine dinucleotide phosphate oxidases) in the diabetic heart and the potential for therapeutic targeting of ROS signalling using novel pharmacological and non-pharmacological approaches to modify specific aspects of the remodelling phenotype in order to prevent and/or delay heart failure development and progression.
- heart failure with preserved ejection fraction
- myocardial disease basic science
- vascular biology
Statistics from Altmetric.com
- heart failure with preserved ejection fraction
- myocardial disease basic science
- vascular biology
Diabetes is becoming alarmingly prevalent and is characterised by elevated cardiovascular risk, and specifically chronic heart failure (CHF), largely due to hypertension or ischaemia, with poor prognosis and survival.1 While the existence of distinct cardiomyopathy remains controversial, the diabetic heart displays major structural changes independent of coronary disease which underlie CHF progression and enhanced susceptibility to stress.2 Prominent features are inflammation and collagen accumulation linked with major extracellular matrix (ECM) changes, underlying subclinical diastolic dysfunction observed in many optimally controlled type 1/2 diabetics (T1D/T2D).3 Despite appreciation of distinct cardiac remodelling, CHF management remains similar between diabetics and non-diabetics, following European Society of Cardiology guidelines.4 Although some standard cardiac drugs (eg, angiotensin receptor blockers) confer direct benefit against fibrosis,5 and novel antidiabetic therapies (eg, liraglutide, empagliflozin) reduce cardiovascular mortality and hospitalisation,6 they are not specifically indicated in diabetes, highlighting a need for tailored strategies.
Cardiac remodelling in diabetes
Pathogenesis of diabetic cardiac remodelling
Remodelling of the non-diabetic heart in response to pathophysiological stimuli (eg, hypertension, ischaemia) is largely driven by neurohumoral activation, for example, renin-angiotensin-aldosterone system (RAAS) and sympathetic nervous system.7 Chronic or acute stress induces remodelling of cardiomyocytes, fibroblasts, endothelium, smooth muscle, and inflammatory cells, and distinct (mal)adaptation (eg, hypertrophy, apoptosis, collagen accumulation, ECM turnover, cytokine/chemokine signalling, autophagy, excitation-contraction coupling, oxidative metabolism), ultimately driving development/progression of CHF (figure 1).7 While these processes are evident in diabetes, they generally occur subclinically, thereby predisposing to stress. The diabetic heart is particularly characterised by ECM remodelling, the main drivers of which are matrix metalloproteinases (MMPs) and TIMPs (tissue inhibitors of MMPs), which are dysregulated, leading to collagen accumulation.8 In diabetes, collagen synthesis and cross-linking are specifically promoted by advanced glycation end-products (AGEs), which upregulate profibrotic signalling (eg, angiotensin II, transforming growth factor-β), largely dependent on cytokine/chemokine secretion (eg, tumour necrosis factor-α, interleukin-1β), by infiltrating inflammatory cells.9 Normally, such pathways interact with MMPs/TIMPs to maintain ECM turnover, which becomes dysregulated in diabetes largely due to RAAS activation. Such alterations promote ECM expansion/dysregulation, which underlies cardiac stiffness and diastolic dysfunction, ultimately causing elevated wall stress, chamber dilatation, contractile dysfunction and CHF.3 Significantly, diabetes promotes cardiomyocyte hypertrophy by reactivating fetal genes (eg, β-myosin heavy chain, atrial natriuretic peptide).10 Associated contractile dysfunction may be explained by altered cytoskeletal organisation and calcium handling involving dysregulation of sarcoplasmic reticulum (SR) Ca2+-ATPase 2a (SERCA2a) and calcium/calmodulin-dependent protein kinase II, and their endogenous regulators, phospholamban and ryanodine receptor 2. Post-translational maladaptive changes in major calcium transport mechanisms in diabetes, possibly mediated by AGEs and reactive oxygen species (ROS), cause delayed SR reuptake and enhanced SR leakage, leading to elevated intracellular calcium, impaired excitation-contraction coupling and cardiac relaxation, and systolic/diastolic dysfunction.3 Cardiomyocyte hypertrophy in diabetes may also occur consequent to reduced cell number due to decreased proliferation or enhanced apoptosis, the latter primarily driven by hyperglycaemia-induced caspase signalling involving, for example, ROS, cytokines and RAAS activation.3 Similarly, autophagy, which degrades dysfunctional intracellular proteins/organelles, is dysregulated in diabetes, although whether this is protective or deleterious is unclear.11–13 The diabetic heart also displays major metabolic remodelling, particularly characterised by impaired mitochondrial oxidative metabolism, leading to reduced energy generation. These key cellular remodelling processes, which ultimately become detrimental in both the diabetic and non-diabetic heart, are summarised in figure 1.
Microvascular versus metabolic drivers of cardiac remodelling in diabetes
Although the diabetic heart is recognised to undergo distinct (mal)adaptive remodelling, whether this occurs secondary to cardiomyocyte or microvascular alterations is unclear (figure 2). Reports that metabolic disturbances (eg, GLUT4/peroxisome proliferator-activated receptor α (PPARα) dysregulation) impair cardiomyocyte insulin sensitivity, causing reduced glucose uptake, increased free fatty acid (FFA) utilisation and cardiac steatosis/lipotoxicity, support the former.3 14 By nature, excess FFA oxidation causes mitochondrial electron transport chain (ETC) dysfunction, ROS-mediated activation of mitochondrial uncoupling proteins (UCP), enhanced proton leak, uncoupling of oxidative phosphorylation, reduced ATP production and, consequently, increased myocardial oxygen demand and impaired efficiency.3 Elevated intracellular FFAs also drive accumulation of cardiotoxic lipid intermediates (eg, ceramide), reduced mitochondrial calcium uptake and mitochondrial permeability transition pore (MPTP) opening.14 Together, such changes decrease enzymatic activity (eg, pyruvate dehydrogenase), causing impaired ATP production and caspase-mediated cell death, while aberrant myocardial insulin signalling drives further mitochondrial dysfunction.15 In addition to metabolic abnormalities, the diabetic heart displays major CHF-independent microvascular changes (eg, perivascular/subendothelial fibrosis, endothelial dysfunction, abnormal capillary permeability/density, microaneurysms, aberrant angiogenesis),2 highlighting likely contribution to associated remodelling. Indeed, in diabetics, impaired coronary microvascular function is linked with diastolic dysfunction,16 while microvascular complications predict cardiovascular disease,2 and retinopathy confers increased CHF risk.17 Similarly, systemic and tissue inflammation, observed in experimental/clinical diabetic CHF, is associated with endothelial dysfunction and may drive reduced coronary microvascular vasodilation in diastolic CHF.15 18 Indeed, vascular endothelial growth factor blockade exacerbates CHF in experimental diabetes,19 while clinical microvascular complications predict CHF progression,2 further highlighting inflammation-mediated endothelial dysfunction as a key CHF promoter in diabetes. Nonetheless, it seems likely that both metabolic and microvascular alterations contribute synergistically to diabetic cardiac remodelling.
Mediators of diabetic cardiac remodelling
Hyperglycaemia and associated metabolic disturbances are positively correlated with CHF in diabetes (eg, UK Prospective Diabetes Study, Heart Outcomes Prevention Evaluation Study).1 Mechanistically, hyperglycaemia primarily activates AGEs, cytokines, oxidative stress and RAAS, thereby driving ECM remodelling via cardiomyocyte and/or endothelium-specific actions. Other mediators include abnormal calcium homeostasis, mitochondrial dysfunction and protein kinase C (PKC)/PPARα activation secondary to FFA metabolism.3 15 Although intensive glycaemic control confers some protection against diabetes-associated CHF, incidence remains increased,1 highlighting other key mechanisms, particularly ROS, which mediate glucose-dependent and independent actions.20 Indeed, experimental and clinical diabetes are linked with oxidative stress, characterised by increased cardiac ROS and impaired antioxidant activity,20 while important processes underlying cardiomyopathy (eg, inflammation, angiogenesis, hypertrophy/apoptosis, fibrosis, contractile dysfunction) are redox sensitive.3
Cardiac ROS signalling in diabetes
ROS are broadly divided into oxygen radicals (eg, superoxide) or non-radicals (eg, hydrogen peroxide (H2O2)), and low-level production governs redox regulation of physiological signalling. Indeed, ROS regulate activity of numerous molecules and signalling pathways, thereby determining precise phenotypic alterations, mediated by post-translational modification of specific amino acids and thereby protein function (see review).21 Superoxide is generated by electron donation to molecular oxygen, and is short lived so only acts locally, before dismutation to H2O2 (figure 3). At high concentrations, superoxide reacts with nitric oxide (NO) to form peroxynitrite, which is directly cytotoxic and reduces NO bioavailability.21 While endogenous ROS scavengers prevent H2O2 overproduction, at elevated levels it generates highly reactive hydroxyl radicals via Fenton or Haber-Weiss reactions. Notably, H2O2 is more stable than superoxide and can diffuse widely, likely accounting for distinct effects on redox regulation and underlying established specificity of ROS signalling.21 In diabetes, oxidative stress ensues due to imbalance between ROS production and degradation, which contributes to cardiac dysfunction/remodelling. Indeed, endogenous antioxidant overexpression is cardioprotective in experimental diabetes.22 Elevated cardiac ROS may react rapidly and non-specifically with DNA, proteins, lipids and carbohydrates, causing irreversible cytotoxicity. ROS-mediated DNA damage also stimulates poly(ADP-ribose) polymerase, thereby inhibiting glyceraldehyde 3-phosphate dehydrogenase and causing accumulation of glycolytic intermediates, activation of, for example, aldose reductase, PKC, AGE signalling and cardiac dysfunction.21 Although ROS are implicated in cardiac remodelling associated with diabetes and other stresses, clinical antioxidant-based approaches have proved unsuccessful, so it is evident that detailed understanding of major sources is required to inform more selective and effective strategies. Here, we consider the specific roles of mitochondria, dysfunctional NO synthase (NOS), xanthine oxidase (XO) and nicotinamide adenine dinucleotide phosphate (NADPH) oxidases, which are present in cardiomyocytes, fibroblasts and endothelium, and linked with diabetic cardiac remodelling.
Mitochondria represent the main cardiomyocyte energy source due to highly efficient metabolism of glucose/FFA to acetyl-CoA, providing high-energy electrons to the mitochondrial ETC, thereby generating abundant ATP.15 Normally 1%–2% of electrons leak from the mitochondria which are scavenged or donated to molecular oxygen to form superoxide, although ROS increase significantly with mitochondrial dysfunction in diabetes.23 Indeed, excess ROS may themselves drive mitochondrial uncoupling by directly impairing the ETC while also promoting electron leak, causing reduced myocardial energy generation, increased superoxide, contractile dysfunction and CHF.24 With insulin resistance, mitochondria are progressively unable to use glucose, resulting in switching to FFA oxidation and further ROS generation.15 One suggested mechanism involves upregulation of proton leak channels, UCP2/UCP3, causing reduced mitochondrial membrane potential, mitochondrial uncoupling/dysfunction and impaired myocardial efficiency.15 Elevated mitochondrial ROS in diabetes may also drive adverse remodelling, particularly cardiomyocyte apoptosis, secondary to MPTP opening and mitochondrial swelling, which is exacerbated by PKC-dependent activation of redox-sensitive KATP channels.21 Indeed, direct linkage between oxidative stress, mitochondrial dysfunction, DNA damage and CHF is established, with similar effects reported in diabetes which are reversible by antioxidant overexpression.24 However, given high myocardial energy demand, the main influence of mitochondrial ROS in diabetes likely relates to metabolic rather than structural remodelling. Indeed, characteristic switching from glucose to FFA metabolism in T2D causes impaired mitochondrial respiration, increased ROS generation, ETC uncoupling, reduced ATP synthesis and cardiac dysfunction. Conversely, when energy demand is low, excess FFAs are preferentially oxidised to superoxide by the mitochondrial ETC.15 Notably, increased FFA uptake/metabolism is regulated by PPARα, which controls oxidative phosphorylation and mitochondrial biogenesis, so activation in diabetes could represent an adaptive response to prevent cardiomyocyte FFA accumulation,15 although persistent upregulation, for example, PPARα-overexpressing mice fed high-fat diet, causes cardiac dysfunction.14 Specifically, Peroxisome proliferator-activated receptor gamma coactivator (PCG-1α)-induced PPARα activation may upregulate UCP2/UCP3, thereby reducing ATP production and promoting mitochondrial ROS generation and cardiac dysfunction.15 Superoxide may also activate UCP3 directly in diabetes, thus driving myocardial lipid peroxidation and propagating ROS generation.15 Notably, excess ROS-producing mitochondria are normally destroyed by cytoprotective mitophagy, which is impaired in diabetes, further increasing mitochondrial dysfunction.15
NO is synthesised by vascular cells and cardiomyocytes via NOS-mediated enzymatic conversion of L-arginine, catalysed by tetrahydrobiopterin. The heart expresses all three NOS isoforms: NOS1 in cardiomyocyte SR, while NOS2/NOS3 are more ubiquitous. NOS2 is induced by pathological stresses (eg, cytokines, calcium), with NOS3 localised to endothelium and cardiomyocyte caveolae where it serves anti-inflammatory, vasoactive and vasoprotective actions.25 Physiological H2O2 may activate endothelial NOS3, thereby promoting NO signalling and downregulating ROS-producing enzymes.25 However, in disease, such cardioprotective mechanisms are overwhelmed and increased superoxide causes reduced NO bioavailability and peroxynitrite generation, which oxidises tetrahydrobiopterin, resulting in NOS3 uncoupling and superoxide production, thereby exacerbating oxidative stress and dysfunctional NO signalling associated with diabetes.25 Peroxynitrite is particularly harmful due to combination with tyrosine to form nitrotyrosine which promotes DNA damage, endothelial dysfunction and cell death.25 Indeed, NOS3 uncoupling and decreased NO availability are widely reported in experimental diabetes and linked with endothelial dysfunction and impaired myocardial relaxation,26 while dysfunctional NOS3 expression/activity may be enhanced, thereby generating peroxynitrite, superoxide and H2O2.25 Myocardial inflammation, which is characteristic of diabetes, also drives NOS2-mediated oxidative stress via nuclear factor kappa B, contributing to associated cardiovascular dysfunction and directly modulating cardiomyocyte calcium signalling.27 However, the most likely impact of dysfunctional NOS in diabetes is impaired coronary microvascular function causing reduced myocardial perfusion, increased leucocyte infiltration and dysregulation of redox-dependent signalling, thereby driving adverse cardiac remodelling.25
Xanthine oxidoreductase comprises two interconvertible forms, xanthine dehydrogenase and XO, which generate superoxide by catalysing conversion of hypoxanthine to xanthine and then uric acid, during terminal purine metabolism.28 While basal XO levels are low, myocardial XO-dependent superoxide production is increased in CHF.20 Indeed, XO activation in experimental myocardial infarction-induced remodelling is reversed by the XO inhibitor, allopurinol, while patients with CHF exhibit elevated uric acid levels.20 Furthermore, increased XO-dependent ROS generation may contribute to abnormal energy metabolism and endothelial dysfunction in human CHF, the latter linked with xanthine oxidoreductase upregulation in coronary endothelium, suggesting that XO influences key aspects of diabetic cardiomyopathy.28 Notably, XO is a major contributor to vascular oxidative stress and dysfunction in diabetes, while allopurinol increases NO-dependent vasodilatation in diabetic humans with high uric acid.29 With specific regard to cardiac remodelling in diabetes, XO is activated in diabetic mice, while allopurinol attenuates hyperglycaemia-induced hypertrophy, fibrosis, cardiac dilatation and dysfunction.30 However, enthusiasm for this research area has been dampened by disappointing outcomes of recent XO inhibitor CHF trials.31
NADPH oxidases are a family of membrane-bound enzymes which are widely expressed in the cardiovascular system and whose primary function is ROS generation.21 The classical NADPH oxidase is composed of a membrane-bound catalytic core (Nox2/p22phox) and several regulatory cytosolic subunits, with the former now known to comprise a family of Nox isoforms (Nox1–5) encoded by different genes and with varying sequence heterogeneity; the main cardiac isoforms are Nox2 and Nox4. In contrast to Nox2, Nox4 is localised to the perinuclear endoplasmic reticulum and mitochondria and its activity regulated by protein level.21 Nox normally produce low levels of ROS which modulate redox-sensitive signalling (eg, growth, differentiation, proliferation). Although their principal product is superoxide, Nox4 may generate primarily H2O2, due to a highly conserved E-loop histidine residue which dismutates superoxide before leaving the complex.32 Superoxide is relatively unstable and rapidly degraded within the compartment it is produced, while H2O2 is longer lived and can diffuse widely, which may underlie distinct effects on redox regulation and specificity of Nox signalling.21 In disease, Nox2-derived superoxide may be largely damaging while Nox4-derived H2O2 (does not react with NO) may confer cardioprotection. Nox-derived ROS have multiple stimuli, with hyperglycaemia, hypoxia and cytokines, most relevant to diabetes.21 Indeed, activity/expression of cardiovascular NADPH oxidases is elevated in clinical/experimental diabetes. However, while they are established in cardiac remodelling associated with, for example, ischaemia and hypertension, their function in diabetes is uncertain. Nonetheless, tissue concentrations of known NADPH oxidase stimuli (angiotensin II, aldosterone, endothelin-1) central to cardiac remodelling are raised in experimental diabetes, while associated cardiovascular dysfunction, apoptosis and fibrosis are reduced by pharmacological or gene targeting of Nox signalling.21 Notably, mice with cardiomyocyte and endothelial-specific Nox4 overexpression demonstrate cardioprotective H2O2-dependent myocardial angiogenesis and vasodilatation, respectively, while those with endothelial-specific Nox2 overexpression show superoxide-driven cardiovascular dysfunction, macrophage recruitment and adverse remodelling,33–35 highlighting differential actions relevant to diabetes. Indeed, mice with combined endothelial-specific insulin resistance and Nox2 deletion display improved endothelial function,36 while endothelial-specific Nox2 overexpression promotes cardiac fibrosis, diastolic dysfunction, myocardial inflammation and endothelial-mesenchymal transition,35 suggesting that Nox2 may be largely damaging in diabetes. However, the role of Nox4 is less clear. One study reported Nox4 downregulation in experimental diabetes and atherosclerosis with proinflammatory and profibrotic signalling exacerbated by pharmacological Nox inhibition, implying cardioprotection,37 while another study found that Nox4 activation in experimental diabetes promoted cardiac fibrosis via ROS-dependent extracellular signal-regulated kinase 1/2-mitogen-activated protein kinase signalling,38 suggesting detrimental actions. Indeed, localisation of Nox4 to mitochondria, which are prone to dysfunction in metabolic disease, suggests a likely important but complex role in the diabetic heart.
Therapeutic targeting of ROS signalling in diabetes
Although cardiac oxidative stress and ROS are increased in diabetes, several non-enzymatic antioxidant trials (eg, vitamin C/E) have proved ineffective against associated cardiovascular disease.20 This may not be surprising given the complexity of ROS signalling, which appears to be highly dependent on species, subcellular localisation, source and local concentration. Failure of non-specific antioxidant therapies is therefore likely due to limited appreciation of the physiological role of ROS signalling, together with its temporal and compartment-specific nature, and emerging tissue-protective functions with regard to CHF.39 The exception is TACT (Trial to Assess Chelation Therapy) in which non-specific iron chelation using edetate disodium, which likely inhibits conversion of H2O2 to peroxynitrite, significantly reduced cardiac events postmyocardial infarction, with particular benefit in diabetes, findings currently being substantiated by TACT2.40 Nonetheless, there remains ongoing research using non-specific antioxidants (eg, resveratrol, N-acetylcysteine), which reduce cardiac ROS generation and remodelling in experimental diabetes,20 although likelihood of clinical benefit seems low. Therefore, attention has turned to more sophisticated source-specific approaches, largely targeting mitochondrial ROS. For example, coenzyme Q10 (CoQ10), a key component of the mitochondrial ETC with antioxidant properties, protects against diastolic dysfunction, myocardial fibrosis and cardiomyocyte remodelling in experimental diabetes, while myocardial CoQ10 levels are diminished in patients with CHF. CoQ10 administration to hypertensive patients also reduces ROS, blood pressure and endothelial dysfunction.20 However, the clinical benefit of CoQ10 in CHF remains unclear, further to heterogeneity of study design, sample size and outcomes, with meta-analyses reporting conflicting conclusions.41 Nonetheless, a recent randomised CHF trial (Q-SYMBIO) indicated that CoQ10 reduced cardiovascular end-points,41 highlighting potential therapeutic application, although no studies have specifically focused on diabetes. Other preclinical approaches targeting mitochondrial ROS include mitoTEMPO, a superoxide dismutase mimetic directed against mitochondria, which improves systolic/diastolic function in diabetic cardiomyopathy and reduces NADPH oxidase expression,42 although this has not been trialled in patients. Of potential ROS targets, Nox are perhaps most exciting given their specificity of signalling, although current preclinical inhibitors (eg, apocynin, VAS2870) remain non-selective for individual isoforms. Notably, well-established CHF medications (eg, ACE inhibitors, angiotensin receptor blockers, mineralocorticoid receptor antagonists and also statins) possess antioxidant properties which may be mediated via suppression of Nox signalling.21
While research towards specific pharmacological targeting of ROS signalling is relatively advanced, recent interest has turned towards novel non-pharmacological approaches focused on microRNAs, epigenetics and gene therapy. With regard to CHF, dysregulation of specific microRNAs has been implicated in key redox-sensitive components, for example, fibrosis, hypertrophy and calcium signalling, while ROS generation may be both determined by hypoxia-mediated DNA hypermethylation and drive downstream epigenetic DNA/histone modifications, thereby impacting gene transcription and maladaptive remodelling.43–45 Indeed, preclinical studies have shown anti-miRs, which inhibit specific microRNAs, and DNA methylation inhibitors, to protect against ROS-mediated tissue damage/dysfunction,46 47 providing proof of concept. Human gene therapy is now approved for several non-cardiac conditions, so may hold more immediate potential for CHF.48 Indeed, SERCA2a has been successfully delivered to cardiomyocytes using viral vectors to restore impaired calcium handling and cardiac contractility in experimental CHF, while clinical safety and beneficial effects on acute cardiovascular outcomes are reported.49 50 Taken together, such approaches appear to hold vast potential for modulating specific redox-sensitive aspects of CHF, particularly in diabetes which is typified by distinct remodelling.
Summary and future perspectives
Diabetic cardiac remodelling is increasingly recognised as a key determinant of susceptibility to cardiac stress and CHF. It is critical that specific therapies are developed to target these often subclinical myocardial alterations which promote systolic/diastolic dysfunction and likely explain increased CHF and poor outcomes in diabetes. In this regard, therapeutic targeting of ROS, which are dysregulated in diabetes, represents a viable approach to selectively modulate detrimental aspects of cardiac remodelling. While non-specific antioxidants are ineffective, novel methods to specifically target ROS species, sources and cellular compartments hold clear potential to reduce CHF progression in diabetes.
Contributors AJW, EKG, RAA, KSE and CJW drafted the manuscript. DJG revised the manuscript critically for important intellectual content. All authors have approved the final submitted version.
Funding This commissioned review article was written with the support of the British Society for Cardiovascular Research. The authors’ work is funded by the British Heart Foundation.
Competing interests None declared.
Provenance and peer review Commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.