Fibrosis of the diabetic heart: clinical significance, molecular mechanisms, and therapeutic opportunities

Abstract

In patients with diabetes, myocardial fibrosis may contribute to the pathogenesis of heart failure and arrhythmogenesis, increasing ventricular stiffness and delaying conduction. Diabetic myocardial fibrosis involves effects of hyperglycemia, lipotoxicity and insulin resistance on cardiac fibroblasts, directly resulting in increased matrix secretion, and activation of paracrine signaling in cardiomyocytes, immune and vascular cells, that release fibroblast-activating mediators. Neurohumoral pathways, cytokines, growth factors, oxidative stress, advanced glycation end-products (AGEs), and matricellular proteins have been implicated in diabetic fibrosis; however, the molecular links between the metabolic perturbations and activation of a fibrogenic program remain poorly understood. Although existing therapies using glucose- and lipid-lowering agents and neurohumoral inhibition may act in part by attenuating myocardial collagen deposition, specific therapies targeting the fibrotic response are lacking. This review manuscript discusses the clinical significance, molecular mechanisms and cell biology of diabetic cardiac fibrosis and proposes therapeutic targets that may attenuate the fibrotic response, preventing heart failure progression.

Graphical abstract:

The cell biology of diabetic cardiac fibrosis. The primary metabolic perturbations of type 1 and type 2 diabetes, hypeglycemia and insulin resistance, activate cardiac fibroblasts, the main cellular effectors of cardiac fibrosis. In addition to their direct effects on cardiac fibroblasts, metabolic changes may promote fibrosis through actions on other myocardial cell types, such as cardiomyocytes, immune cells and vascular cells. In the diabetic heart, cardiomyocytes may exhibit activation of oxidative pathways, resulting in release of fibrogenic cytokines and growth factors, and local activation of angiotensin II. Inflammatory stimulation may also induce Transforming Growth Factor-β (TGF-β) synthesis by cardiac macrophages and lymphocytes, whereas mast cells may secrete chymase, activating an alternative pathway involved in angiotensin II generation. Endothelial cells may contribute to the fibrotic response by producing fibrogenic growth factors and endothelin-1 (ET-1). In the diabetic heart, macrophages, vascular cells and fibroblasts may produce specialized matrix proteins with fibrogenic properties (such as fibronectin (FN) and thrombospondin-1 (TSP-1)). Although resident cardiac fibroblasts are the predominant matrix-producing cells in diabetic hearts, contributions of endothelial cells converting to fibroblasts through endothelial to mesenchymal transition (EndMT), and of mural cells that may acquire a fibroblast phenotype, have been suggested, but not convincingly documented. Increased deposition of interstitial and perivascular collagens and matrix crosslinking, increase myocardial stiffness and may play an important role in development of diastolic dysfunction, and in the pathogenesis of diabetes-associated heart failure with preserved ejection fraction (HFpEF). IL-1 = interleukin-1, ROS = reactive oxygen species, TNF-α = tumor necrosis factor-α.

1. Introduction

Diabetes is associated with an increased risk of heart failure and is a predictor of adverse outcome in heart failure patients [1]. Fifty years ago, the Framingham study demonstrated that male patients with diabetes have a 2.5-fold increase in the incidence of heart failure. The impact of diabetes on heart failure was even higher in female patients who exhibited a 5-fold increase in risk [2]. Moreover, extensive clinical evidence suggests that diabetes has adverse prognostic implications in heart failure patients, in both heart failure with reduced ejection fraction (HFrEF) and in heart failure with preserved ejection fraction (HFpEF) populations [3]. In several heart failure clinical trials, including the Candesartan in Heart failure-Assessment of Reduction in Mortality and morbidity (CHARM) [4], the Digitalis Investigation Group (DIG) [5], the Irbesartan in Heart Failure with Preserved Ejection Fraction (I-PRESERVE) [6], and the Phosphodiesterase-5 Inhibition to Improve Clinical Status and Exercise Capacity in Heart Failure with Preserved Ejection Fraction (RELAX) [7] trials, diabetes was consistently associated with worse symptoms, increased risk of hospitalization, and higher morbidity and mortality [3], in both groups with systolic dysfunction and with preserved ejection fraction [4].

The basis for the increased risk of heart failure in patients with diabetes is not only due to the high incidence of macrovascular complications (that may cause hypertension and early onset of severe atherosclerotic coronary disease) [8], but also reflects the development of „diabetic cardiomyopathy“, a distinct pathologic entity that is independent of coronary disease or other concomitatnt conditions and is associated with structural and functional perturbations [9],[10],[11],[12]. Diabetic cardiomyopathy is characterized by a constellation of myocardial abnormalities that involve not only cardiomyocytes, but also interstitial, immune and vascular cells [13]. Cardiac interstitial and perivascular fibrosis are prominent characteristics of diabetic cardiomyopathy and may contribute to the development of diastolic dysfunction, thus explaining the high prevalence of HFpEF in subjects with diabetes [14], while promoting arrhythmogenesis. This review manuscript deals with the characteristics, mechanisms and implications of diabetes-associated cardiac fibrosis. We discuss the cellular effectors and molecular mechanisms of fibrosis in the diabetic heart, and we examine the functional implications of fibrotic remodeling in patients with diabetes. Finally, we propose therapeutic strategies that may attenuate dysfunction and improve remodeling in diabetes-associated heart failure by targeting fibrogenic pathways.

2. Fibrosis in patients with diabetes-associated myocardial disease

2.1. Evidence of fibrotic remodeling in patients with diabetes

Extensive evidence suggests that subjects with diabetes exhibit significant myocardial fibrosis (Table 1). Increased extracellular matrix (ECM) accumulation in the hearts of patients with type 1 and type 2 diabetes has been suggested by imaging studies, or documented directly in tissues collected during cardiac biopsy and autopsy [15],[16],[17],[18],[19]. Fibrosis in subjects with diabetes may be interstitial and perivascular, or extend to replacement fibrotic lesions (in which cardiomyocytes are replaced by fibrous tissue), even in the absence of overt clinical cardiac disease [20],[18],[21]. Fibrotic changes involve collagen deposition in both left [22],[21] and right ventricle [15]. In addition to the histological evidence of fibrosis, patients with diabetes may also have changes in the composition of the myocardial interstitial ECM, exhibiting increased levels of type III vs. type I collagen [23]. The extent of myocardial fibrotic changes in subjects with diabetes is dependent on the duration and severity of metabolic dysregulation and on the presence or absence of concomitant conditions. Several studies have demonstrated that glycosylated hemoglobin (HbA1c) levels are positively correlated with fibrotic remodeling [20],[24], possibly reflecting a relation between glycemic control and myocardial ECM accumulation. Moreover, cardiac fibrosis was even more pronounced when diabetes was accompanied by arterial hypertension or renal disease [21],[9],[25]. In contrast, relatively healthy patient populations with pre-diabetes and without concomitant conditions exhibited no significant fibrotic changes [26]. Diabetes-associated fibrotic myocardial changes are typically accompanied by cardiomyocyte hypertrophy [9] and by microvascular alterations, characterized by thickening of the media of small intracardiac vessels and narrowing of the vessel lumen due to the deposition of acid mucopolysaccharide material [9].

Table 1:

Clinical evidence on the association between diabetes and cardiac fibrosis

Patient populations	Method of myocardial fibrosis assessment	Findings	Ref
Patients with type II diabetes (T2D) (n=80)	Cardiac magnetic resonance (CMR)	Positive correlation between HbA1c levels and myocardial fibrosis.	[24]
Patients with or without diabetes undergoing cardiac catheterization without significant coronary disease (T2D, n=12; subjects without diabetes, n=6)	Myocardial biopsy	Subjects with diabetes had increased content of collagen III in the perimysium and perivascular regions. No significant differences in collagen I and VI levels.	[23]
Patients with diabetic glomerulosclerosis (n=27)	Autopsy	Patients with diabetic glomerulosclerosis exhibited a cardiomyopathy associated with diffuse fibrotic strands, and subintimal thickening and media hypertrophy of intramural coronary arterioles.	[9]
Patients with T2D (n=100) with/without microalbuminuria, vs. controls (n=30)	CMR	Highest extracellular volume fraction (ECV) as surrogate for extracellular matrix (ECM) deposition in patients with diabetic microalbuminuria, followed by individuals with diabetes in the absence of microalbuminuria and controls. Diastolic dysfunction detected in patients with diabetic microalbuminuria.	[25]
Patients with arterial hypertension, diabetes, or both (n=67)	Autopsy	Highest levels of interstitial, perivascular and replacement fibrosis in patients with diabetes and arterial hypertension compared to subjects with isolated diabetes, or hypertension.	[21]
Patients with T2D (n=231) vs. 945 healthy controls (n=945)	CMR	Increased ECV in patients with diabetes. Association between ECV and increased mortality and higher hospitalization rates for heart failure (HF). Renin-angiotensin-aldosterone system (RAAS) inhibitor use linked to reduced ECV.	[28]
Patients with T2D (n=70), and patients with pre-diabetes (n=76), vs. individuals without diabetes (n=296)	CMR	Increased ECV was an independent predictor of cardiovascular outcomes and all-cause mortality in patients with diabetes. Patients with diabetes and increased ECV had worse outcome, in comparison to subjects with diabetes, but without increased ECV, and to individuals without diabetes.	[27]
Patients with congestive heart failure (CHF); one-fourth of them had T2D	Serum markers of cardiac fibrosis	Patients with diabetes had increased circulating levels of collagen synthesis markers. Increased markers of cardiac fibrosis were associated with higher mortality, and increased incidence of death and hospitalization.	[399]
Patients with T2D (n=53) without evidence of coronary artery disease, but with hypertension and hyperlipidemia	CMR	Elevated ECV despite normal echocardiographic diastolic function and normal left ventricular ejection fraction (LVEF). Positive association between ECV and 24-hour urinary aldosterone excretion.	[19]
Obese adolescents with T2D and obese individuals without diabetes with normal LVEF and atrium size vs. healthy volunteers	CMR	Highest levels of ECV in patients with obesity and T2D, followed by obese patients and healthy individuals. Association between ECV and BMI for the entire population, and between ECV and high-sensitivity C-reactive protein, serum triglycerides and HbA1c in the obese group.	[444]
343 subjects: pre-diabetes (n=78), diabetes (n=47), controls (n=218) all with preserved ejection fraction and without history of cardiovascular disease	CMR	No increase in ECV in subjects with pre-diabetes or diabetes vs. healthy population.	[26]
T2D patients (n=15) vs. healthy controls (n=9)	Autopsy and myocardial biopsy during heart surgery	Subjects with diabetes had increased collagen content and increased left ventricular end-diastolic pressure (LVEDP).	[16]
Type 1 diabetes (T1D) patients (n=741)	CMR	Positive correlation between HbA1c and myocardial scar. Myocardial scar was linked to adverse cardiac remodeling.	[18]
Asymptomatic patients with insulin-dependent diabetes (n=26)	Echocardiography with integrated backscatter	Subjects with diabetes had increased myocardial echodensity (presumed due to higher collagen content). However, abnormal Doppler signals were noted in only 3 patients.	[20]
T2D patients without cardiac symptoms (n=67)	CMR, echocardiography with integrated backscatter, collagen biomarkers	Diffuse myocardial fibrosis as indicator of diastolic dysfunction in early diabetic cardiomyopathy. Positive association between collagen biomarkers and diastolic dysfunction.	[17]
Patients with HFrEF (n=36, 10 with T2D) vs. HFpEF patients (n=28, 16 with T2D)	Myocardial biopsy	Increased myocardial fibrosis and AGE deposition in HFrEF vs. HFpEF patients.	[35]
T2D patients with normal LVEF and without coronary artery disease (n=50) vs. 19 healthy control subjects (n=19)	CMR	In subjects with diabetes, there was a positive correlation between increased interstitial fibrosis and impaired LV systolic and diastolic function.	[29]
T2D patients without hypertension or coronary artery disease (n=9) vs. healthy individuals (n=7)	Myocardial biopsy	In patients with diabetes, there was an inverse correlation between right ventricular interstitial fibrosis and LVEF.	[15]
T1D patients (n=714)	CMR	Negative correlation between nonischemic myocardial scar and LVEF.	[445]
Patients with T2D (n=12) vs. 14 subjects without diabetes (n=14).	Cell culture of atrial fibroblasts (CFs) collected during heart surgery	Activation of CFs indicated by increased type I collagen production and morphological changes with greater variation in CF size in patients with diabetes.	[51]

2.2. The clinical consequences of diabetes-associated cardiac fibrosis

Myocardial fibrosis has adverse prognostic implications in patient populations with or without diabetes [27]. Independently of other factors, fibrosis in subjects with diabetes is associated with higher hospitalization rates for heart failure and with increased mortality [28]. Development of cardiac fibrosis in patients with diabetes has a major impact on both contractile function and electrophysiology. In the early stages, diabetes-associated fibrosis is predominantly associated with diastolic dysfunction and HFpEF. In more advanced stages, fibrosis may also cause systolic dysfunction, and may play an important role in the pathogenesis of atrial and ventricular arrhythmias and in sudden cardiac death [29],[30],[31],[32].

2.2.1. Fibrosis and diabetes-associated heart failure

The link between diabetic fibrosis and diastolic dysfunction/HFpEF is intuitive, and is supported by a significant amount of experimental and clinical evidence. Increased deposition of cross-linked ECM in subjects with diabetes reduces ventricular compliance, increasing filling pressures. In animal models of diabetic cardiomyopathy, the extent of fibrotic changes is associated with the severity of diastolic dysfunction [33]. Moreover, genetic approaches disrupting matrix synthesis and preservation improve ventricular compliance and attenuate diastolic dysfunction [34]. Some, but not all clinical investigations support the involvement of diabetes-associated fibrosis in the pathogenesis of HFpEF. A study examining patients with type II diabetes, but without symptomatic myocardial disease, showed that fibrosis detected by Cardiac Magnetic Resonance (CMR) and/or echocardiography, and serum procollagen type III levels (reflecting active collagen synthesis) were linked with impaired left ventricular relaxation, suggesting that ECM deposition may contribute to the early development of diabetes-related diastolic dysfunction [17]. However, in other investigations, the link between fibrosis and diastolic dysfunction is less consistent. In asymptomatic subjects with insulin-dependent diabetes, increased collagen deposition suggested by augmented myocardial echodensity was inconsistently associated with a perturbed left ventricular filling pattern, a characteristic of diastolic dysfunction [20]. Moreover, a study examining endomyocardial biopsies from patients with diabetes without coronary artery disease suggested that myocardial fibrosis may be a major determinant of diastolic stiffness in HFrEF patients, whereas in HFpEF, cardiomyocyte resting tension may be the predominant cause of diastolic dysfunction [35]. It should be emphasized that the use of unreliable strategies for assessment of myocardial fibrosis in some studies (such as non-specific imaging indicators of interstitial expansion, or circulating biomarkers that may originate from several different organs) and the very limited evidence based on histological samples challenge the validity of any conclusions on the role of diabetic fibrosis in the clinical setting.

The involvement of fibrosis in diabetes-associated HFrEF and systolic dysfunction is also supported by some clinical evidence. Patients with type 2 diabetes exhibited shorter global contrast-enhanced time T1 on CMR imaging, a surrogate for increased fibrosis, that was linked to impaired myocardial systolic function [29]. Moreover, interstitial fibrosis in biopsied samples from diabetic right ventricles correlated inversely with ejection fraction [15]. Whether the association between diabetic myocardial fibrosis and systolic dysfunction is due to direct effects of fibrotic remodeling on contractility, or simply reflects loss of cardiomyocytes (resulting in replacement fibrosis) is unclear. Potential mechanisms underlying the effects of fibrosis on systolic function may include disruption of excitation-contraction coupling by the expanding ECM network, prominent activation of proteases that may perturb key molecular links between the matrix and the sarcomeric contractile apparatus [36], or promote cardiomyocyte death through anoikis [37], generation of matrix fragments in protease-rich fibrotic lesions that may stimulate pro-inflammatory signaling [38], and reduced myocardial perfusion due to perivascular fibrosis.

2.2.2. Relations between fibrosis and diabetes-associated dysrhythmias

The link between diabetes and arrhythmogenesis may involve, at least in part, enhanced fibrosis [39]. Diabetes is associated with an increased incidence of atrial fibrillation (AF) and predicts progression to permanent AF after the first episode [40]. Increased susceptibility of subjects with diabetes to AF may be due to perturbed electrophysiologic properties related to actions on ion channels and gap junctions, effects on autonomic or microvascular function, and accentuated ECM deposition [41]. Diabetes-associated fibrosis may reduce atrial myocyte coupling, leading to generation of reentry circuits, in which sustained electrical activity and propagation of action potentials along closed-loop circuits promotes AF [42]. Direct documentation of a causative role of atrial fibrosis in diabetes-associated AF is lacking; however, several lines of associative evidence support the notion that diabetic fibrotic remodeling of the atria may contribute to atrial arrhythmias. First, in animal models of both type I and II diabetes, atria exhibit accentuated collagen deposition [43],[44],[45],[46],[47],[48] that may enhance atrial arrhythmogenicity [49],[47],[45]. Second, increased deposition of fibro-fatty tissue in human atria correlated with a longer P wave duration, an indicator of perturbed atrial conduction [50]. Third, fibroblasts from atrial tissue from patients with diabetes undergoing coronary artery bypass grafting (CABG) surgery showed increased production of type I collagen [51]. In addition to its effects on atrial tachyarrhythmias, diabetes-induced fibrosis may also promote sinus node dysfunction [52].

The increased susceptibility of individuals with diabetes to ventricular arrhythmias and sudden cardiac death may also involve fibrogenic actions [32]. Unfortunately, evidence supporting this notion is lacking. Considering that diabetes also exerts arrhythmogenic actions on cardiomyocytes by enhancing mitochondrial oxidative stress [53], the relative contribution of fibrosis in diabetes-related ventricular arrhythmias is unknown.

3. Pathophysiologic mechanisms of fibrosis development in diabetic cardiomyopathy

3.1. Animal models of diabetic cardiac fibrosis

Animal models of type 1 and type 2 diabetes are powerful tools for the study of the cellular and molecular mechanisms involved in diabetic cardiomyopathy (Table 2) [54],[55]. In models of type 1 diabetes, destruction or dysfunction of pancreatic β-cells, induced through pharmacologic or genetic interventions, cause hypoinsulinemia and severe hyperglycemia. Administration of agents with β–cell toxicity, such as streptozotocin (STZ) is a commonly used strategy for induction of type I diabetes in both rodents and in large animals. Development of interstitial and perivascular cardiac fibrosis has been reported in rodent models of STZ-induced diabetes. However, interpretation of data on cardiac function is often hampered due to the volume depletion that may accompany the severe hyperglycemia noted in these animals. Genetic models of type I diabetes include the OVE26 model, induced through overexpression of calmodulin in pancreatic ß-cells, leading to ß-cell injury and insulin deficiency [56] and the Akita (Ins2+/−) model, induced through a spontaneous mutation in the insulin 2 gene causing severe β-cell dysfunction [57],[44],[58]. Myocardial fibrosis has been reported in both these models.

Table 2:

Cardiac fibrosis in animal models of diabetes.

Model	Metabolic perturbations	Myocardial fibrotic changes	Functional changes	Limitations of the model	Ref
Type I diabetes
Pancreatic β-cells destruction by streptozotocin (STZ) injection in rodents or large animals.	Hyperglycemia, hypoinsulinemia, hypertriglyceridemia, polyuria, polydipsia, weight loss	Myocardial fibrosis was noted in some, but not all studies. In rats, cardiac fibrosis, accompanied by immune cell infiltration was noted 2 and 6 weeks after induction of diabetes [446]. However, in another rat study, no significant effect of diabetes on collagen content was noted 3 weeks after STZ injection [447].	Impaired systolic and diastolic function, cardiac electrical disturbances	Toxic and carcinogenic effects of STZ. Very high glucose levels may lead to dehydration. Female mice are less sensitive to STZ.	[446], [448], [447], [449], [450], [406]
Pancreatic β-cells destruction by alloxan injection	Hyperglycemia, hypoinsulinemia, no changes in serum lipids, decreased body weight	Extensive interstitial fibrosis, mitochondrial swelling 12 weeks after injection (rabbit model). Interstitial fibrosis, accompanied by cardiomyocyte hypertrophy, without an increase in cardiomyocyte apoptosis at 8 weeks (rat model)	Increased atrial size, hypertrophy, and diastolic dysfunction with preserved systolic function	Fluctuations in blood glucose levels, high mortality rate due to initial hypoglycemic shock, Model does not recapitulate the autoimmune nature of type 1 diabetes.	[384] [451] [452]
OVE26 transgenic mouse (β-cell injury by calmodulin overexpression)	Hyperglycemia, hypoinsulinemia	OVE26 mice on the FVB background exhibited cardiac fibrosis at 24 weeks of age, accompanied by inflammation and cardiomyocyte hypertrophy.	Cardiac remodeling, reduced LVEF	Strain-dependent effects. More pronounced fibrosis in FVB vs. C57Bl6 mice	[56] [453]
Akita (Ins2+/−) mouse (severe pancreatic β - cell dysfunction due to mutation in the insulin 2 gene)	Pronounced hyperglycemia, reduced plasma insulin and decreased body weight	Conflicting results have been reported in various studies: - augmented cardiac fibrosis at 12 weeks of age - enhanced collagen I and III amounts in cardiac atria at 16–20 weeks of age - no increase in cardiac fibrosis or cardiomyocyte size at 12 and 24 weeks of age	Atrial remodeling, LVEF modestly increased, decreased or preserved (depending on the study), diastolic dysfunction.	Conflicting evidence on the presence and severity of fibrosis	[57] [44] [58]
Type II diabetes
db/db mice (point mutation in the gene encoding the leptin receptor resulting in central leptin resistance)	Severe obesity (related to hyperphagia), hyperglycemia, hyperinsulinemia, hyperglucagonemia, hyperlipidemia, hyperleptinemia	Development of cardiac interstitial and perivascular fibrosis, accompanied by cardiomyocyte hypertrophy has been demonstrated in mice of different ages: 10, 16, 20, 22 weeks, 6–12 months of age. Severity of fibrosis is dependent on age, sex and genetic background of animals.	Effects on cardiac function are dependent on genetic background, age and sex. In C57Bl6J background only evidence of diastolic dysfunction is noted.	The most widely studied model of type II diabetes. Signaling through the truncated leptin receptor in peripheral cells (including fibroblasts) may complicate interpretation. Severity of fibrosis and diastolic dysfunction are dependent on genetic background, age and sex of the mice.	[382] [454] [80] [64] [455] [176],[62], [68]
Leptin-deficient ob/ob mouse	Severe hyperglycemia, hyperinsulinemia, hypertriglyceridemia, obesity	For mice on the black and tan brachyuric (BTBR) background: increased collagen I and III deposition, myocyte apoptosis at 20 weeks of age For mice on the C57BL/6 background under high-fat diet: no changes in atrial matrix deposition at 16 weeks of age	Ob/ob mice on the BTBR background exhibited cardiac remodeling with systolic dysfunction	Cardiac fibrosis development depends on the genetic background. Fibrosis may be absent in the C57BL/6 background.	[374] [456]
Zucker diabetic fatty (ZDF) rats (leptin receptor mutation)	Obesity, insulin-resistance, hyperinsulinemia, pronounced hyperglycemia, dyslipidemia	Increased interstitial and perivascular collagen deposition documented at 6, 16 and 45 weeks of age	Cardiac remodeling, mild diastolic dysfunction, accompanied by preserved or decreased systolic function.	May not recapitulate human obesity, considering the rarity of leptin receptor mutations in humans	[457] [377] [458] [459]
Goto-Kakizaki (GK) rat (polygenic non-obese)	Hyperglycemia, hyperlipidemia, proteinuria	Interstitial and perivascular fibrosis, cardiac hypertrophy, myocyte, fibroblast and vascular cell apoptosis at 26 weeks of age. Diffuse interstitial fibrosis with atria remodeling at 24 and 40 weeks of age	Diastolic dysfunction, elevated LVEF, increased atrial arrhythmogenicity	This genetic non-obese model is associated with early depletion of β-cell mass and hyperglycemia and may not recapitulate the pathophysiology in the majority of patients with type 2 diabetes.	[383] [47] [460]
Otsuka Long-Evans Tokushima Fat (OLETF) rat, (polygenic obese)	Mild obesity, late onset of hyperglycemia, postprandial hyperglycemia, hyperinsulinemia,	Interstitial and perivascular fibrosis, has been reported at 15–22 weeks of age.	Diastolic dysfunction	Diabetes affects mostly male animals. Cardiomyopathic changes may be milder, in comparison to other models.	[238] [245] [73] [461]
Kuo-Kondo (KK) mice (polygenic obese) under high-calorie diet	Hyperglycemia, hyperinsulinemia, hyperlipidemia	Increased cardiac fibrosis (collagen IV and fibronectin) and cardiac hypertrophy at 24 weeks of age. Augmented deposition of collagen I and III at 18 weeks of age	Enhanced NT-pro-BNP serum concentrations as a surrogate of impaired cardiac function Reduced LVEF, diastolic dysfunction	Strain-dependent effects are likely. Most of the evidence on effects in the heart were derived from studies in KK-Ay mice.	[204] [372]
Mice fed high-fat or high-fat and high-sugar diet (Western diet)	Hyperglycemia, hyperinsulinemia, hypertriglyceridemia, obesity	Mild cardiac fibrosis, accompanied by cardiac hypertrophy has been reported after long-term feeding. Time of onset of fibrosis is dependent on the type of diet.	Diastolic dysfunction	Although the model recapitulates the human disease better than genetic models, the extent of fibrosis depends on the type of the diet and the genetic background. The development of fibrosis often requires prolonged feeding [75]	[456] [462] [463]
Low-dose STZ and high-calorie diet (model of advanced type II diabetes with insulin resistance and β-cell failure)	Hyperglycemia, hyperinsulinemia, mild obesity	Development of fibrosis has been reported; timing is dependent on the diet used	Diastolic dysfunction with preserved LV systolic function	The onset and severity of cardiac fibrosis depends on the type of the diet.	[81] [464]

Animal models of type II diabetes, on the other hand, are either genetic, or induced through feeding with hypercaloric diets. db/db mice have a point mutation in the gene encoding the leptin receptor, resulting in a protein with a truncated cytoplasmic domain that is functionally inactive [59]. This defect results in resistance to the central effects of leptin, hyperphagia, severe obesity and development of overt diabetes. Both histochemical staining techniques and biochemical assays have consistently documented cardiac fibrosis in db/db mice at 4–12 months of age [60],[61],[34],[62], accompanied by cardiomyocyte hypertrophy and microvascular rarefaction [62]. Studies examining cardiac function in db/db mice have produced conflicting results. The bulk of the evidence suggests that db/db mice develop predominant diastolic dysfunction [63],[34],[62],[64],[65] a finding that may reflect fibrotic remodeling or alterations in cardiomyocyte phenotype. However, some studies have demonstrated systolic dysfunction in db/db mice, evidenced by significant reduction in ejection fraction [66], [67]. The differences may reflect effects of the genetic background [68] on the severity of the metabolic perturbations, but also sex-specific effects and functional assessment performed at different timepoints.

Fibrotic cardiac remodeling has also been reported in other models of type 2 diabetes. Zucker diabetic rats (ZDF) develop hyperphagia and obesity due to a leptin receptor missense mutation, and exhibit significant myocardial fibrosis [55]. In contrast, evidence on the development of fibrosis in leptin deficient ob/ob mice is conflicting. Although some studies have reported significant pericoronary fibrosis in ob/ob mice [69], other investigations found very mild [70], or even negligible fibrotic changes in the ob/ob myocardium [71]. Several polygenic models of type 2 diabetes, including the Otsuka Long-Evans Tokushima Fat (OLETF) rat model [72],[73], and the non-obese Goto-Kalizaki (GK) rat [74] exhibit cardiac fibrosis.

Models of diet-induced obesity and type 2 diabetes may recapitulate the pathophysiology of human disease better than genetic models. Development of myocardial fibrosis in models of diet-induced metabolic dysfunction is dependent on sex and genetic background of the mice, and on the type and duration of the diet used. Considering the relatively mild fibrotic changes observed in diet-induced models, prolonged feeding and highly sensitive methodology may be needed to detect fibrotic changes. In male C57BL/6J mice, feeding with a high-fat diet for 8–16 months caused a significant increase in myocardial collagen I and collagen III levels [75]. High-fat diets with a high content of simple carbohydrates have been shown to accelerate development of fibrosis. Thus, male C57BL/6 mice on a high-fat diet that contained large amounts of simple carbohydrates exhibited increased deposition of cross-linked collagen and diastolic dysfunction after 6 months of feeding [76]. Moreover, male C57BL/6 mice fed a high fat diet containing high-fructose corn syrup developed myocardial fibrosis after only 16 weeks of feeding [77]. It has been suggested that female mice may be more susceptible to fibrosis and diastolic dysfunction when fed a high-fat/high-fructose diet, exhibiting fibrotic changes and reduced ventricular compliance after 8 weeks of feeding [78].

Some studies have used complex pathophysiologic models, combining metabolic perturbations with neurohumoral activation. For example, db/db mice infused with angiotensin II exhibited accentuated cardiac hypertrophy and fibrosis [79],[80]. Although these models may be useful to understand the combined effects of diabetes and hypertension on the myocardium, their value in dissecting signals specifically activated by metabolic disease is limited. Moreover, in order to mimic late stages of type 2 diabetes, characterized by insulin resistance and β-cell failure, rodents on high-fat diet have been injected with a low-dose STZ [81]. In this model, metabolic changes were associated with myocardial fibrosis and with development of diastolic dysfunction [81].

3.2. Fundamental metabolic perturbations that induce diabetes-associated myocardial fibrosis

3.2.1. Hyperglycemia

Extensive in vitro evidence supports the notion that hyperglycemia may be implicated in diabetes-associated cardiac fibrosis. High glucose concentrations have been reported to stimulate fibroblast proliferation, promote myofibroblast conversion, and activate transcription and secretion of ECM proteins in cardiac fibroblasts in many [82] [83],[84], but not in all studies [85] (Table 3, Figure 1). The stimulatory effects of glucose have been attributed to oxidative mechanisms [86], increased ion channel activity leading to Ca2+ influx [87], stimulation of angiotensin II and TGF-β signaling cascades [88],[89], production of advanced glycation end-products (AGEs) [90], and activation of Erk [82] pathways. Moreover, fibrogenic effects of hyperglycemia may also involve actions on other cell types. High glucose concentrations have been reported to induce synthesis of ECM proteins, such as fibronectin in endothelial cells and may even trigger endothelial to mesenchymal transition (EndMT) [91], [92]. Moreover, under hyperglycemic conditions, cardiomyocytes and immune cells may acquire a fibrogenic phenotype [93],[94].

Table 3:

Effects of high glucose on cardiac fibroblast phenotype and function

Source of cardiac fibroblasts	Protocol of glucose stimulation	Effects of glucose	Proposed mechanism	Ref
15–16-week old diabetic (db/db) and lean male mice	25mM (high) vs. 5mM (normal) glucose concentration for 24h	Increased production of collagen I, TIMP-2, MMP-2, TGF-β, RAGE, PAI-1 in CFs from db/db mice under both hyperglycemic and normoglycemic conditions. In contrast glucose had no significant effects on CFs from lean animals.	N/A	[139]
8–10-week old male C57BL/6J mice and 19-week old male C57BL/6J mice following 11 weeks of diabetogenic diet	25mM (high) vs. 5.5mM (normal) glucose concentration for 72h	Increased collagen I and αSMA expressions and unchanged production of hyaluronan in CFs from mice fed diabetogenic diet under high glucose. No effects of high glucose on CFs from lean mice in terms of proliferation, hyaluronan or collagen I production and αSMA expression	N/A	[85]
Adult male wistar rats	33mM (high) vs. 5mM (normal) glucose concentration for 6–24h	Increased synthesis of fibronectin	Reduction in PPARδ	[383]
1- to 3-day old Wistar rats	25mM (high) vs. 5mM (normal) glucose concentration for 4–72h	Increased synthesis of collagen I and III	Activation of extracellular signal regulated kinase (ERK) 1/2	[82]
3- to 40-day old Sprague Dawley rats	25mM (high) vs. 5mM (normal) glucose concentration for 48h	Decreased proliferation and migration of CFs, increased collagen gel remodeling and contraction, enhanced collagen synthesis and α1β1 integrin expression. No changes in α-SMA expression	N/A	[250]
1- to 3-day old Sprague-Dawley rats	30mM (high) vs. 5mM (normal) glucose concentration for 48h	Apoptosis of CFs, increased synthesis of collagen I and III	Increased oxidative stress due to decreased acetaldehyde dehydrogenase 2 (ALDH2)	[270]
0- to 2-day old Sprague-Dawley rats	High glucose (12.5mM, 17mM, 25mM) vs. 5.5mM (normal) glucose concentration for 24h, 48h or 72h	Significantly increased production of collagen 1 and TGF-β under high glucose (25mM) for 24h	Activation of intracellular RAS	[88]
1- to 3-day old Wistar rats	40mM (high) vs. 5.5mM (normal) glucose concentration for 48h	Increased proliferation, upregulated α-SMA, collagen-I/-III, and MMP-2/-9 protein expression	Endoplasmic reticulum stress and activation of Wnt/β-catenin	[465]
1- to 3-day old Wistar rats	40mM (high) vs. 5.5mM (normal) glucose concentration for 24h	Increased viability, migration, proliferation and production of collagen I/III and MMP-2, -9	Activation of calcium-sensitive receptor (CaSR)/intracellular Ca2+/TGF-β1/Smads pathway	[258]
Neonatal Sprague-Dawley rats	33.3mM (high) vs. 5.5mM (normal) glucose concentration for 24h	Increased proliferation, α-SMA and collagen-I expressions	Activation of Methyl CpG binding protein 2 (MeCP2)/Ras association domain family 1 isoform A (RASSF1A)/Ras/ERK1/2 pathway	[466]
neonatal Sprague–Dawley rats	30mM (high) vs. 5.5mM (normal) glucose concentration for 48h	Increased proliferation and production of collagen I	Increased transient receptor potential vanilloid (TRPV4) channel responsible for regulation of Ca2+ influx to CFs with consecutive activation of TGF-β1/Smad3 pathway	[87]
1-day old Sprague–Dawley rats	25mM (high) vs. 5.6mM (normal) glucose concentration for 24h	Increased proliferation, collagen synthesis	Activation of TGF-β1/pSmad2/3 pathway through increased oxidative stress	[467]
Neonatal C57BL/6 mice	30mM (high) vs. 5.5mM (normal) glucose concentration for 24h	Enhanced collagen I and collagen III expressions	Activation of Kcnq1ot1/miR-214-3p/NLRP3 inflammasome/caspase-1/Il-1β/TGF-β1/Smad pathway	[326]
Right atrium of patients with arterial hypertension and dyslipidemia undergoing cardiac surgery	15mM and 25mM (high) vs. 5.0mM (normal) glucose concentration for 24h, 48h and 72h	Increased proliferation	N/A	[468]
Human cardiac fibroblasts	25mM (high) vs. 5.0mM (normal) glucose concentration for 48h	Increase in proliferation and apoptosis, acquisition of α-SMA, increase in expression of collagens I and III, and TGF-β1 Bigger changes in higher glucose	miR-32-5p-dependent decrease in expression of dual specificity phosphatase 1 (DUSP1)	[469]
Primary human cardiac fibroblast cells (ACBRI 5118)	25mM and 100mM (high) vs. 5.5mM (normal) glucose concentration for 1, 2, 4, 6, 12, 24, and 48 h	Upregulation of MMP-2 and TIMP-1	Upregulation of SGLT1	[470]
Primary human cardiac fibroblast cells	25mM (high) vs. 5.5mM (normal) glucose concentration for 24h	Increased cell viability and proliferation, elevated α-SMA expression, production of collagens I, III and VI	Downregulation of miR-9 and upregulation of TGFβRII	[471]

Figure 1: The effects of high glucose on cardiac fibroblast phenotype and function.

In vitro studies have suggested that exposure to high glucose has profound effects on fibroblast activation, survival, proliferation, responsiveness to growth factors (such as TGF-β), matrix-synthetic capacity, and matrix-remodeling properties. High glucose may also modulate fibroblast:matrix interactions by stimulating integrin synthesis. It should be emphasized that various studies have produced inconsistent findings, depending on the protocols used, the duration of glucose stimulation (acute vs. chronic), and the culture conditions. Although some studies have suggested that high glucose may promote myofibroblast conversion, in vivo investigations did not show significant myofibroblast conversion in diabetic hearts. coll-1 = collagen I, coll-III = collagen III, CTGF = Connective Tissue Growth Factor, FN = fibronectin, IL-1β = interleukin-1β, MMP-2 (−9) = matrix metalloproteinase-2 (−9), PDGF-β = platelet-derived growth factor-β, TGF-β = transforming growth factor-β.

Despite the abundant in vitro evidence supporting the fibrogenic effects of high glucose, the in vivo role of hyperglycemia in diabetes-associated myocardial fibrosis is unclear. Although in some clinical studies, levels of HbA1c were associated with indicators of myocardial fibrosis [24], to what extent glycemic control attenuates progression of cardiac fibrotic changes remains unknown. The findings on the impact of optimal glycemic control on diabetic complications are conflicting [95],[96],[97]. Randomized controlled trials do not support the notion that intensive glycemic control reduces the incidence of diabetes-associated heart failure [1]. Moreover, any beneficial effects of antidiabetic drugs on cardiac fibrotic remodelling may not necessarily reflect their glucose-lowering actions [98] [99]. In animal models, very limited evidence is available on the effects of glucose lowering on tissue fibrosis. In a rat model of STZ-induced type 1 diabetes, tight glycemic control had no effects on renal fibrosis [100], and studies examining the effects of glucose lowering on myocardial collagen deposition have not been performed.

3.2.2. Hyperinsulinemia/insulin resistance

Insulin resistance is the hallmark of type 2 diabetes. Clinical studies suggest an association between insulin resistance and myocardial fibrosis. An investigation studying volunteers without diabetes with varying body mass index showed that insulin resistance was independently associated with myocardial extracellular volume assessed through CMR, an indicator of interstitial cardiac fibrosis [101]. The basis for the link between insulin resistance and cardiac fibrotic remodeling has not been systematically investigated. Although the involvement of perturbed insulin signaling in diabetic fibroblast activation has not been systematically studied, the accentuated oxidative stress associated with insulin resistant states may promote a fibrogenic response [102]. Moreover, insulin resistance may also stimulate fibrosis by inducing lipotoxic cardiomyocyte injury, causing accumulation of harmful lipids that perturb organelle function, leading to cardiomyocyte injury and subsequent activation of inflammatory and fibrogenic programs [103],[104].[105],[106]. Elevated circulating insulin levels, typically associated with insulin resistance, may induce activation of cardiac fibroblasts, upregulating collagen synthesis, and promoting fibrosis [107].

3.2.3. Lipotoxicity and dyslipidemia

Insulin resistance is associated with lipotoxicity and hyperlipidemia, alterations that may play an important role in cardiac fibrosis. Insulin resistance attenuates the anti-lipolytic effects of insulin, resulting in triglyceride breakdown and free fatty acid (FFA) release from adipose tissue. Circulating FFAs are taken up by several different tissues, including the heart, where they accumulate as triglycerides. Perturbations in myocardial fatty acid oxidation and direct activation of lipogenesis may further accentuate myocardial lipotoxic injury [108],[109], leading to lipid accumulation, that is not limited to triglycerides, but also involves ceramides, diacylglycerols, medium chain acyl-carnitines and saturated long-chain fatty acids (such as palmitate) [109], [110]. In cardiomyocytes, lipid excess causes mitochondrial dysfunction and increases endoplasmic reticulum (ER) stress; these perturbations may trigger release of pro-inflammatory and fibrogenic mediators, activating fibroblasts and accentuating interstitial matrix deposition. Diabetes-induced lipotoxicity was found to induce fibrosis through Forkhead Box O3 (FoxO3) and downstream activation of mammalian Ste20-like kinase 1 (Mst1)/mitogen-activated protein kinase kinase kinase 1 (MEKK1)/c-Jun N-terminal kinase (JNK) cascades [111]. The significance of myocardial lipotoxicity as a fibrogenic stimulus is supported by associative data and by the anti-fibrotic effects of strategies attenuating lipotoxic injury [112],[113]. For example, deficiency of general control nonderepressible 2 (GCN2), a regulator of lipid metabolism, reduced cardiac fibrosis and improved cardiac function [114]. Moreover, cardiac overexpression of cardiac hormone-sensitive lipase in STZ-induced diabetic mice induced intracellular lipolysis and reduced cardiac acylglyceride and lipid peroxide content, leading to downstream reduction in TGF-ß, matrix metalloproteinase (MMP)-2 and collagen levels [115]. In addition to their role in activation of paracrine cardiomyocyte-driven fibrogenic responses, toxic lipids may also directly stimulate fibroblasts. Evidence supporting this concept is limited to in vitro experiments. In cardiac fibroblasts, palmitate stimulation was found to induce pro-inflammatory activation, while attenuating collagen synthesis [116]. These findings suggest that any direct effects of lipotoxic injury on fibroblasts may be dependent on the biochemical profile of lipid accumulation and may not be uniformly fibrogenic. Expansion and activation of epicardial adipose tissue, often found in insulin-resistant subjects, may also be involved in diabetic cardiac fibrosis. In volunteers without diabetes, epicardial adipose tissue volume was associated with cardiac interstitial matrix expansion, assessed through CMR [101].

Patients with diabetes often develop serum lipid abnormalities, typically characterized by hypertriglyceridemia and low HDL levels [117]. To what extent circulating lipids contribute to myocardial fibrosis is unclear. Lipid lowering agents, such as 3-hydroxy-3-methylglutaryl (HMG)-CoA reductase inhibitors or fibrates, attenuated myocardial fibrosis in models of diabetic cardiomyopathy [118], [119]. However, these anti-fibrotic effects may be due to direct inhibitory actions of these agents on inflammatory and fibrogenic cascades, including direct targeting of cardiac fibroblasts rather than reflecting lipid-lowering effects [120], [121], [122], [123].

Changes in the profile of circulating lipoproteins may also contribute to diabetes-associated myocardial fibrosis. In vitro, oxidized LDL triggers fibroblast activation, inducing a matrix-synthetic phenotype [124],[125]. HDL, on the other hand may exert anti-inflammatory and anti-fibrotic actions [126]. Hyperglycemic conditions in subjects with diabetes may trigger HDL modifications, such as glycation, that attenuate the anti-inflammatory effects of HDL and may promote fibrogenic responses [127]. The in vivo significance of this mechanism in mediating fibrosis associated with metabolic disease has not been tested.

3.3. The cellular mechanisms of diabetes-associated fibrosis

In addition to cardiomyocytes, the adult mammalian heart also contains large populations of vascular cells and fibroblasts, and smaller but significant populations of macrophages, mast cells and dendritic cells [128]. The metabolic perturbations associated with diabetes may induce fibrotic changes either directly, by activating cardiac fibroblasts, the main matrix-producing cells in the heart, or indirectly by stimulating paracrine pathways involving cardiomyocytes, vascular cells or immune cells (graphical abstract) [129].

3.3.1. Fibroblasts, the main matrix-producing cells in the diabetic heart

All cardiac fibrotic conditions involve activation of interstitial fibroblasts [130],[131],[132],[133]. In infarcted and in failing hearts, injury-associated release of fibrogenic growth factors, neurohumoral stimulation, activation of mechanosensitive signaling cascades and deposition of specialized matrix proteins induce conversion of fibroblasts into myofibroblasts, activated cells that express contractile proteins, such as α-smooth muscle actin (α-SMA), and synthesize large amounts of ECM proteins [134],[135],[136],[137]. Although abundant evidence has doccumented activation of fibroblasts in diabetic hearts [51], whether diabetes-associated cardiac fibrosis involves myofibroblast transdifferentiation remains controversial. Cardiac fibroblasts harvested from obese diabetic Zucker rat hearts, a model of type 2 diabetes, exhibited greater ability to contract gels and elevated α-SMA expression, consistent with a myofibroblast phenotype [138]. In contrast, in db/db mouse hearts, fibroblasts exhibited increased expression of collagens and protease inhibitors [139], but no significant expresion of the myofibroblast markers α–SMA and periostin [140]. Thus, in contrast to the prominent role of myofibroblasts in myocardial infarction and pressure overload-induced heart failure, diabetic fibrosis may involve an alternative mechanism of fibroblast activation.

Extensive evidence has demonstrated that diabetes causes structural and functional perturbations within mitochondria, including perturbed mitochondrial metabolism, redox imbalance, defects in electron transport chain complex structure and activity, reduced mitochondrial Ca²⁺ uptake, mitochondrial degradation, and suppression of mitophagy. These diabetes-associated mitochondrial perturbations increase ROS release, and have been associated with increased myocardial fibrosis and dysfunction [141],[142],[143],[144],[145]. Most experimental studies have studied either whole hearts, or mitochondria harvested from cardiomyocytes and microvascular endothelial cells. Evidence documenting mitochondrial changes in diabetic cardiac fibroblasts is scarce. Upon neurohumoral activation, impaired calcium uptake into fibroblast mitochondria reprogrammed fibroblasts to glycolytic metabolism, promoting a myofibroblast conversion and cardiac fibrosis [146]. However, the significance of this mechanism in fibrotic remodeling of the diabetic heart has not been investigated.

3.3.2. Cells indirectly involved in diabetes-associated cardiac fibrosis by regulating fibroblast activation and matrix remodeling

3.3.2.1. Cardiomyocytes

Cardiomyocytes are major targets of metabolic dysregulation and may contribute to the fibrotic response by releasing pro-inflammatory and fibrogenic mediators in response to hyperglycemic or lipotoxic injury. High glucose levels activate reactive oxygen species (ROS)-mediated synthesis of inflammatory and fibroblast-activating cytokines and growth factors (such as IL-1β and TGF-β) in cardiomyocytes [93], [147]. Moreover, prolonged and severe metabolic perturbations may cause irreversible cardiomyocyte injury leading to cell death [148],[149], and triggering replacement fibrosis. Although in animal models of type 1 and type 2 diabetes, that typically involve young animals, replacement fibrosis is generally absent [140], in older patients with diabetes, the combined effects of glycolipotoxic injury, microvascular dysfunction and atherosclerotic changes may result in death of significant numbers of cardiomyocytes, thus contributing to the development of fibrosis.

3.3.2.2. Vascular cells

Considering the rich capillary network of the myocardium, it is not surprising that endothelial cells are the most abundant non-cardiomyocytes in the adult mammalian heart [128]. In the infarcted and failing heart, endothelial cells may be implicated in the pathogenesis of cardiac fibrosis through conversion to fibroblasts, by secreting fibroblast-activating growth factors and cytokines, or by regulating matrix metabolism through release of proteases. The relative contribution of the endothelial to mesenchymal transition (EndMT) process in expansion of the fibroblast population in remodeling hearts [150],[151] has been challenged by systematic lineage tracing studies in models of myocardial infarction [152] and left ventricular pressure overload [153]. Similar studies in models of diabetes are lacking; thus, any support for a role of EndMT in diabetic cardiac fibrosis is based on associative findings [154]. Secretion of fibrogenic mediators by activated endothelial cells may contribute to the pathogenesis of diabetic cardiac fibrosis. A study using endothelial cell-specific loss-of-function approaches suggested that endothelial cell-derived endothelin-1 stimulates myocardial fibrosis in STZ-induced diabetic mice [154]. Moreover, in a hyperglycemic environment, endothelial cells may contribute to fibrotic remodeling by secreting specialized extracellular matrix proteins, such as fibronectin [91], or proteases involved in matrix remodeling [155]. Diabetes-related endothelial cell activation may also indirectly promote cardiac fibrosis by inducing fibrosis of the aorta and large arteries, thus increasing vascular resistance and systemic blood pressure. In a model of high fat-induced metabolic dysfunction, endothelial cell-specific mineralocorticoid receptor signaling was implicated in the pathogenesis of aortic fibrosis [156].

Vascular mural cells (pericytes and vascular smooth muscle cells) are capable of myofibroblast conversion and may also contribute to diabetes-associated fibrosis by secreting mediators that activate adventitial fibroblasts, stimulating perivascular collagen deposition [157],[158],[159]. Moreover, in subjects with diabetes, vascular smooth muscle cells may respond to high glucose concentrations by producing collagens and pro-fibrotic matricellular proteins, such as thrombospondin (TSP)-1 [160]. Fibrogenic activation of vascular smooth muscle cells may promote fibrosis of the aorta and of resistance arteries [161], indirectly causing myocardial fibrotic remodeling by increasing the pressure load.

3.3.2.3. Macrophages

The heart contains a population of resident macrophages [162] that has been implicated, at least in mouse models, in cardiac electrophysiology [163] and in cardiomyocyte mitochondrial homeostasis [164]. Myocardial injury triggers recruitment of monocytes that replace the resident macrophage population [165],[166], and exert a wide range of functions, phagocytosing dead cells, regulating the inflammatory response, and stimulating fibroblasts and vascular cells [167]. Considering the abundant evidence suggesting a role for macrophages in fibroblast activation in infarcted and in pressure-overloaded hearts [168],[169],[170],[171], and the pro-inflammatory effects of hyperglycemia that induces adhesion molecule and CC chemokine expression [172], [173], it is plausible to hypothesize that macrophage recruitment and activation may be involved in diabetic myocardial fibrosis. Several lines of evidence support this notion. First, infiltration of the diabetic myocardium with monocytes and macrophages has been consistently demonstrated in models of type 1 and type 2 diabetes [174],[175],[176],[177]. Second, high glucose-induced cytokine and chemokine synthesis in macrophages [178] may activate fibroblasts, promoting secretion of fibrogenic mediators and proteases. Third, exosomes released by activated macrophages have been suggested to stimulate fibroblasts in diabetic hearts [94]. Fourth, genetic deletion of the chemokine receptor CCR2 (a key mediator in recruitment of inflammatory monocytes), or administration of a CCR2 inhibitor attenuated fibrosis in a model of STZ-induced diabetes [179]. However, considering the systemic effects of these approaches, whether the reported protective effects were due to attenuated myocardial macrophage recruitment is unclear.

3.2.3.4. Lymphocytes and mast cells.

T lymphocytes may be implicated in fibrotic conditions through secretion of fibrogenic cytokines, such as the Th2 cytokines IL-4, IL-10 and IL-13, or through effects on macrophage phenotype and function. Limited information is available on the potential role of T cells in diabetic cardiac fibrosis. Genetic depletion of circulating CD4+ and CD8+ lymphocytes in a type 1 diabetes model attenuated cardiac fibrosis [180],[181],[182]. Whether these effects were specific to myocardial complications of diabetes, or reflected broad effects of lymphocyte targeting on diabetic complications remains unknown. Moreover, the mechanisms responsible for diabetes-associated T cell activation and the T cell-derived fibrogenic mediators that may induce diabetic myocardial fibrosis have not been explored. Induction of adhesion molecules and chemokines in diabetic hearts may promote T cell recruitment through mechanisms similar to the ones involved in macrophage infiltration [183]. Moreover, it has been suggested that chronic hyperglycemia may activate T cells through RAGE-dependent pathways that induce expression of Th1, Th2 and Th17 cytokines [184].

Mast cells have also been implicated in fibrotic conditions, serving as a source of a wide range of fibrogenic mediators (such as TGF-β, TNF-α and tryptase), and producing proteases that remodel the ECM. Moreover, mast cells express chymase, an enzyme involved in generation of angiotensin II. High glucose may activate mast cells, inducing fibrogenic pro-inflammatory mediators, such as TNF-α, and triggering release of granular contents [185]. In vivo studies showed that myocardial mast cell density is increased in a mouse model of type 1 diabetes and that pharmacologic mast cell stabilization attenuates myocardial fibrosis [186]. Mast cell deficient mice exhibited attenuated ventricular and atrial fibrosis in STZ-induced models of type 1 diabetes [187],[188]. Whether these protective actions are due to abrogation of local fibrogenic actions of mast cells on the myocardium, or reflect systemic effects of mast cell depletion remains unknown.

3.4. Molecular signals involved in diabetes-associated myocardial fibrosis

3.4.1. Secreted mediators

3.4.1.1. Neurohumoral pathways

Activation of the renin-angiotensin-aldosterone system (RAAS) is one of the predominant mechanisms responsible for progression of diabetic cardiomyopathy and has been suggested to contribute to both fibrosis and cardiomyocyte dysfunction. Diabetes is associated with activation of several components of the RAAS, including increased renin levels, enhanced angiotensin II and aldosterone, and increased angiotensin II type 1 receptor (AT1R) density [189],[190],[191] (Figure 2). Moreover, reduced levels of angiotensin-converting enzyme 2 (ACE2) in diabetic hearts have been proposed to contribute to fibrosis, by increasing fibrogenic angiotensin II effects and by decreasing levels of its antagonist Ang-(1–7) [192]. Hyperglycemia does not only induce systemic activation of the RAAS, but may also stimulate local synthesis of angiotensin II in cardiomyocytes [193] and in cardiac fibroblasts.

Figure 2: The Renin-Angiotensin-Aldosterone System (RAAS) in diabetes-associated cardiac fibrosis.

Diabetes-associated cardiac fibrosis involves both systemic and myocardial activation of the RAAS system. A: Systemic effects of the RAAS in diabetic myocardial fibrosis. Diabetes is associated with induction and activation of all components of the RAAS, including increased synthesis of angiotensinogen, accentuated renin release, increased activity of angiotensin converting enzyme (ACE) and chymase to generate angiotensin II, and enhanced aldosterone secretion by the adrenal cortex. Diabetes-associated activation of the RAAS results in volume overload (by increasing renal sodium reabsorption) and pressure overload (by inducing vasoconstriction and arterial fibrosis). These effects contribute indirectly to myocardial fibrosis. B: Local fibrogenic effects of Angiotensin II and aldosterone in the myocardium. Although systemic release of angiotensin II and angiotensin II-stimulated secretion of aldosterone are prominent in patients with diabetes, local myocardial expression of angiotensin II has also been reported. Angiotensin II and aldosterone directly activate fibroblasts (F), promoting their proliferation, increasing extracellular matrix (ECM) synthesis, and modulating the protease:antiprotease balance, thus regulating ECM remodeling. The fibrogenic actions of RAAS activation in the diabetic heart may also involve indirect effects on cardiomyocytes (CM) and macrophages (Ma) which, upon activation, may secrete fibrogenic cytokines and growth factors, stimulating fibroblast-derived matrix synthesis. Although effects of angiotensin II on endothelial to mesenchymal transition (EndMT) have been suggested, the endothelial origin of activated fibroblasts has not been convincingly documented in diabetic hearts. ACE = angiotensin converting enzyme, ADR = adrenal gland, ART = artery, AT1R = Angiotensin II receptor type 1, Erk = extracellular signal-regulated kinase, K = kidney, L = liver, miRs = microRNAs, MR = mineralocorticoid receptor, ROS = reactive oxygen species, TGF-β = transforming growth factor-β.

The critical role of the RAAS in the pathogenesis of diabetic cardiac fibrosis has been suggested predominantly by pharmacologic inhibition studies. In rodent models of type 2 diabetes, treatment with ACE inhibitors attenuated interstitial and perivascular fibrosis [194],[63]. Administration of AT1R blockers had similar effects in both type 2 and type 1 diabetes models, suggesting that the profibrotic actions of angiotensin II in diabetic hearts are AT1R-dependent [69],[195],[193]. Aldosterone antagonism also attenuated fibrosis in rodent models of type 1 diabetes [196] and diet-induced obesity/insulin resistance [197]. The fibrogenic effects of the RAAS in diabetic hearts are supported by clinical evidence. AT1R blockade reduced levels of biomarkers associated with collagen synthesis in patients with type 2 diabetes [198]. Moreover, in a prospective randomized controlled clinical study, treatment of obese patients with the aldosterone antagonist spironolactone for 6 months reduced levels of serological markers of collagen synthesis. The anti-fibrotic effects of spironolactone were associated with improved myocardial compliance and diastolic function [199].

The fibrogenic actions of the RAAS in diabetic hearts are mediated through several different mechanisms. First, both angiotensin II and aldosterone exert pro-inflammatory actions [200],[201] that may contribute to recruitment and activation of fibrogenic macrophages. Second, angiotensin II may stimulate fibroblast activation and proliferation through AT1R-mediated generation of reactive oxygen species [63],[202], or through Erk-dependent pathways [203]. Third, angiotensin II may induce expression of fibrogenic microRNAs, such as miR-21 [204]. Finally, the fibrogenic actions of angiotensin II are mediated in part through induction and activation of TGF-β [205], and downstream activation of Smad-dependent signaling cascades [193],[88],[206].

3.4.1.2. Pro-inflammatory cascades involved in diabetes-associated cardiac fibrosis

The metabolic perturbations associated with diabetes, such as hyperglycemia and lipotoxicity, activate pro-inflammatory signaling cascades that may contribute to the development of cardiac fibrosis. High glucose-mediated inflammatory activation may involve the NLR Family Pyrin Domain-Containing 3 (NLRP3) inflammasome, the molecular platform required for generation of bioactive Interleukin (IL)-1β [207]. NLRP3 activation in response to high glucose is mediated, at least in part, through oxidative stress [208],[209]. Several lines of evidence suggest that inflammasome-mediated inflammatory activation may promote fibrosis of the diabetic heart. First, silencing of the NLRP3 inflammasome in a rat model of type 2 diabetes induced through a high fat diet and low dose STZ attenuated myocardial inflammation and fibrosis [210]. Second, in a rodent model of type 1 diabetes, tumor necrosis factor (TNF)-α antagonism protected the myocardium from fibrosis [211]. Third, in a model of STZ-induced type 1 diabetes, IL-6 loss attenuated myocardial interstitial fibrosis [212], and in a model of diet-induced type 2 diabetes, IL-6 neutralization reduced myocardial collagen deposition [213]. Fourth, IL-17 disruption reduced cardiac fibrosis in a model of STZ-induced diabetes [214]. Fifth, administration of a pharmacologic inhibitor of Nuclear Factor (NF)-κB, a cytokine-activated transcription factor with a key role on in inflammatory activation, attenuated myocardial fibrosis in a rat model of STZ-induced type 1 diabetes [215]. Considering that the NRLP3 inflammasome and pro-inflammatory cytokines are also critically involved in the pathogenesis of metabolic dysfunction in type 1 and type 2 diabetes [216, 217],[218], anti-fibrotic effects of inflammasome inhibition on the myocardium may reflect, at least in part, an improved metabolic profile, rather than direct actions on cardiac fibrogenic signaling.

3.4.1.3. The potential role of the anti-inflammatory cytokines IL-4, IL-13 and IL-10

The anti-inflammatory cytokines IL-4, IL-10 and IL-13 are synthesized by lymphocyte and macrophage subsets in remodeling hearts [219] and have been implicated in regulation of inflammation and fibrosis. In models of heart failure, IL-4, IL-10 and IL-13 have been suggested to play a dual role, inhibiting inflammatory macrophage activation [220],[221], while stimulating fibroblasts [222],[223],[171] and fibrogenic mononuclear cells [224],[225], and promoting matrix preservation through induction of TIMPs [219]. The long-term effects of these cytokines on fibrotic cardiac remodeling are dependent on the balance between their anti-inflammactory and fibrogenic actions. Although endogenous IL-4 and IL-10 have been suggested to stimulate fibrosis in some experimental models [226],[171], in other studies the impact of their fibrogenic actions may have been outweighed by their anti-inflammatory effects, ultimately resulting in no significant net changes in fibrotic remodeling [227], or even in inhibition of fibrosis [220],[221]. The patterns of regulation and the potential involvement of endogenous IL-4, IL-10 and IL-13 in diabetes-associated cardiac remodeling have not been systematically studied. In a model of type 1 diabetes, endogenous IL-10 was found to be protective, attenuating diabetes-associated sinus node fibrosis and dysfunction [52]. Moreover, infusion of exogenous IL-10 has been suggested to attenuate atrial fibrosis in models of high-fat diet (HFD)-induced obesity and type 1 diabetes [228],[52].

3.4.1.4. The fibrogenic role of TGF-βs in diabetic cardiac fibrosis

TGF-βs activate a matrix-preserving phenotype in cardiac fibroblasts and are central effectors of myocardial fibrosis [229]. The adult heart contains latent TGF-β that can be activated following injury, mechanical or metabolic stress, resulting in release of active growth factor. Subsequently, the active TGF-β dimer binds to complexes of constitutively active type 2 receptors and activatable type 1 receptors, transducing downstream signals mediated through a family of transcription factors, the receptor-activated Smads (R-Smads), or through non-Smad pathways [230],[231],[232],[233],[234]. Increased myocardial expression of TGF-β and activation of downstream Smad2 and Smad3 signaling cascades have been consistently reported in models of type 1 and type 2 diabetes and are associated with cardiac fibrosis [235],[236],[237],[194],[34],[61],[238]. Accentuated TGF-β signaling responses in diabetic hearts may be mediated through angiotensin II- or cytokine-mediated de novo synthesis of TGF-β isoforms [194], via stimulation of oxidative pathways and proteases that activate local stores of TGF-βs [239], or through direct effects of high glucose on TGF-β transcription, secretion, and activation [240],[241],[242]. Moreover, high glucose has been suggested to induce synthesis and externalization of both type 1 and type 2 TGF-β receptors on the cell surface [239], thus accentuating fibrogenic TGF-β/R-Smad signaling.

TGF-βs exert a wide range of pro-fibrotic actions, inducing myofibroblast conversion, stimulating synthesis of integrins on the fibroblast surface, accentuating ECM synthesis and promoting matrix preservation through suppression of MMPs and upregulation of protease inhibitors [243],[244]. Although the involvement of TGF-βs in diabetic cardiac fibrosis is plausible, direct evidence documenting the significance of TGF-β signaling cascades in fibrotic remodeling and dysfunction of the diabetic heart is limited. The relative role of the 3 TGF-β isoforms remains unknown. A study using the db/db mouse model of type 2 diabetes demonstrated that germline Smad3 haploinsufficiency reduced cardiac fibrosis and improved myocardial compliance, attenuating oxidative stress [34]. However, the cellular mechanisms of TGF-β/Smad actions in the diabetic heart have not been investigated and may involve direct actions on cardiac fibroblasts, or effects on cardiomyocytes, immune and vascular cells.

3.4.1.5. FGFs and VEGF

Changes in myocardial expression of the angiogenic growth factors VEGF and FGFs have been reported in diabetic hearts; however, the findings have been inconsistent, and their potential role in the pathogenesis of diabetes-associated fibrosis is unclear. Some studies have demonstrated increased myocardial expression of FGF2 and VEGF in the diabetic myocardium [245],[246]. In contrast, other studies showed that myocardial VEGF expression is downmodulated in a model of STZ-induced type 1 diabetes and suggested that VEGF loss may result in capillary rarefaction and accentuated fibrotic remodeling [247]. The conflicting findings may be due to assessment at different stages in the progressive development of diabetic cardiomyopathy, or reflect the lack of systematic assessment of VEGF isoform levels in models of diabetes-associated heart disease. Investigations on the effects of FGFs in diabetic hearts have suggested anti-fibrotic actions. Endothelial cell-specific signaling through FGFR1 (the receptor for FGF1 and FGF2) was found to attenuate EndMT in diabetic hearts, protecting from the development of myocardial perivascular fibrosis [248]. These anti-fibrotic effects are consistent with findings suggesting inhibition of diabetic fibrosis upon administration of FGF1 [249].

3.4.1.6. Metabolically activated integrin signaling cascades

Integrins are cell surface receptors involved in cellular adhesion, migration and signaling. In stimulated fibroblasts, activated integrins transduce mechanosensitive, matrix-derived and growth factor-mediated signals that regulate proliferation, survival and matrix protein synthesis. The potential role of integrin activation as a molecular mechanism involved in diabetic fibrosis is supported predominantly by in vitro evidence. Exposure to high glucose induced expression of α1ß1 integrin in cardiac fibroblasts, stimulating collagen synthesis and matrix remodeling [250]. In renal mesangial cells, elevated glucose was found to activate β1 integrin, promoting fibronectin assembly [251]; whether a similar mechanism is involved in diabetes-associated cardiac fibrosis is unknown. A glycated matrix was also found to enhance growth factor-induced α11 integrin expression in cardiac fibroblasts, which in turn may stimulate myofibroblast conversion and increase collagen production [252]. The pro-fibrotic effects of α11 integrin are supported by findings showing that mice with germline loss of α11 integrin had reduced cardiac fibrosis in a model of STZ-induced type 1 diabetes. Surprisingly, attenuated fibrosis in α11 integrin null mice was associated with impaired cardiac function and perturbed morphology of cardiomyocyte myofibrils [253]. The basis for the homeostatic effects of α11 integrin on the myocardium remains unknown, but may reflect the critical role of the fibroblast-ECM axis in preservation of cardiomyocyte function.

3.4.1.7. Metabolically activated ion channels and receptors

Several members of the Transient Receptor Potential (TRP) family of cationic channels, such as TRPC6, TRPM7 and TRPV4 have been implicated in fibroblast activation and myofibroblast conversion [254],[255],[256],[257], predominantly in pathologic conditions associated with accentuated mechanical stress. Although exposure to high glucose also activates ion channels, information on their role in diabetic myocardial fibrosis is scarce. Diabetes-associated activation of TRPV4 increases intracellular Ca2+ influx in fibroblasts, promoting their proliferation and accentuating growth factor-mediated collagen synthesis [87]. High glucose was also found to activate calcium-sensitive receptor (CaSR), increasing intracellular calcium in cardiac fibroblasts and leading to increased fibrogenic signaling through TGF-ß/R-Smad-dependent pathways [258],[259]. Moreover, high glucose induces expression of several members of the TRP family and enhances calcium influx in monocytes promoting their activation [260]. Whether these effects may promote a fibrogenic macrophage phenotype is unknown. It should be emphasized that the role of ion channels in diabetic cardiomyopathy is not limited to effects on the fibrotic response. In a model of STZ-induced type 1 diabetes, TRPC6 loss was associated with heart failure [261], likely reflecting actions on other cell types that may outweigh any beneficial anti-fibrotic effects.

3.4.1.8. Reactive oxygen species (ROS)

Reactive oxygen species are generated by all myocardial cells and play a critical role in homeostasis and repair. However, perturbations in the balance between cellular ROS and antioxidant defensive mechanisms results in injury and has been implicated in the pathogenesis of many fibrotic conditions. Hyperglycemia and lipotoxicity, the main metabolic perturbations associated with diabetes, cause an imbalance between reactive oxygen/nitrogen species (e.g. malondialdehyde (MDA), 3-nitrotyrosine (3-NT), 4-hydroxy-2-nonenal (4-HNE)) and antioxidant enzymes, resulting in nitroso-oxidative stress [262],[263],[264]. Increased myocardial oxidative stress is consistently found in animal models of type 1 and type 2 diabetes [265]. In a rat model of STZ-induced type 1 diabetes, increased myocardial glutathione oxidation was accompanied by increased lipid hydroperoxide levels [266]. Moreover, in db/db type 2 diabetic mice, myocardial mitochondrial ROS increase was associated with lipid and protein peroxidation [267]. Diabetic hearts also exhibit attenuated myocardial activation of antioxidant enzymes, such as manganese superoxide dismutase and glutathione peroxidase 1 [268],[269], suggesting that diabetes-associated myocardial oxidative stress reflects both increased ROS generation and defective free radical scavenging.

Enhanced ROS generation in fibroblasts of the diabetic heart has been implicated in stimulation of a matrix-synthetic [270] and matrix-remodeling program. Moreover, the fibrogenic effects of ROS may also reflect oxidative DNA damage in cardiomyocytes. High glucose has been reported to perturb the ability to repair damaged DNA [271], thus resulting in persistent injury, inflammation and progressive fibrosis [272]. The in vivo role of oxidative stress in diabetes-associated cardiac fibrosis is supported predominantly by studies using pharmacologic interventions. In models of type 1 or type 2 diabetes, various approaches that inhibit ROS generation, or attenuate oxidative modifications, reduced myocardial fibrosis [273] and decreased diastolic dysfunction [274],[63].

3.4.1.9. Intracellular pathways involved in diabetic cardiac fibrosis: an overview

The metabolic perturbations associated with diabetes stimulate activation of intracellular fibrogenic cascades, either directly by increasing oxidative stress, or indirectly by inducing release of cytokines and growth factors, and by triggering activation of neurohumoral pathways. These intracellular cascades transduce a matrix-synthetic or matrix-remodeling program in cardiac fibroblasts. Moreover, metabolically-mediated activation of intracellular pathways in cardiomyocytes and immune cells may exert fibrogenic actions through secretion of fibroblast-activating mediators. Considering the complex interactions of the intracellular cascades that regulate fibrosis, and the multiple mechanisms involved in their activation, their role in diabetes-associated fibrosis remains poorly understood. In addition to the involvement of pathways activated by specific cytokines and growth factors (such as the NF-κB system and the R-Smads), cascades stimulated by a wide range of extracellular and intracellular signals, such as the MAPKs, the RhoA/ROCK system, and the YAP/TAZ pathway have been implicated in diabetes-associated fibrogenic responses [275]. Moreover, the cardioprotective effects of AMP-activated protein kinase (AMPK), a key regulator of metabolic function, may involve, at least in part, attenuation of fibrosis.

3.4.1.9.1. AMPK

AMPK serves as a cellular fuel gauge that rapidly senses energy deprivation, and upon activation orchestrates protective metabolic responses [276],[277]. In diabetic hearts, AMPK has profound effects on cardiomyocyte metabolism, increasing insulin sensitivity, and stimulating glucose uptake and glycolysis. In addition to its effects on cardiomyocytes, AMPK may also exert anti-fibrotic actions by modulating fibroblast phenotype. AMPK has been suggested to inhibit myofibroblast conversion by disrupting the TGF-β/Smad3 cascade [278], and was found to suppress MMP9 synthesis in embryonic fibroblasts, attenuating their matrix-degrading capacity [279]. Although the role of endogenous AMPK on fibroblast activation in diabetic hearts remains poorly understood, several studies have suggested that the protective effects of metformin and sodium glucose cotransporter 2 inhibitors in diabetic cardiomyopathy may involve, at least in part, anti-fibrotic actions due to AMPK activation [280],[281].

3.4.1.9.2. MAPKs

Myocardial MAPK activation has been consistently reported in models of type 1 and type 2 diabetes [282]. In diabetic hearts, MAPKs may be activated in cardiomyocytes, interstitial fibroblasts, immune cells and vascular cells in response to hyperglycemia-mediated oxidative stress, angiotensin II, or secreted cytokines and growth factors. Limited information is available on the role of MAPKs in diabetes-associated fibrosis. In a model of STZ-induced type 1 diabetes, mice with cardiomyocyte-specific expression of a dominant negative mutant form of p38a MAPK had attenuated myocardial fibrosis [283], suggesting a paracrine p38-mediated mechanism involved in fibroblast activation. In contrast, another study using a model of STZ-induced diabetes showed that treatment with a p38 MAPK inhibitor did not affect myocardial fibrosis and diastolic dysfunction, despite attenuating inflammatory activity [284]. p38α MAPK is considered the major isoform expressed in cardiac fibroblasts [285], and has been implicated in fibroblast and myofibroblast activation following ischemic myocardial injury, mechanical stress, or neurohumoral stimulation [286],[287]. However, the role of fibroblast-specific activation of p38 MAPK in diabetic hearts has not been examined.

Erk MAPK and JNK MAPK are also prominently activated in diabetic hearts and in cardiac fibroblasts stimulated with high glucose [288],[282],[289],[290] and may contribute to the pathogenesis of diabetic cardiac fibrosis. In vitro, effects of angiotensin II and TGF-β on fibroblast-like cells have been attributed to Erk stimulation, triggered by activation of the small G protein Ras [291]. In vivo, glucolipotoxicity in diabetic hearts, has been suggested to activate an AGEs-Erk1/2 axis, promoting pro-apoptotic cascades in cardiomyocytes, while stimulating pro-inflammatory and fibrogenic pathways in immune cells and fibroblasts respectively [292]. Considering the broad range of cellular targets of Erk and JNK MAPKs, the cellular basis for their effects in diabetic fibrosis remains unknown.

3.4.1.9.3. The RhoA/ROCK (Rho-associated coiled-coil containing kinases) pathway

Signaling through tyrosine kinases, G-protein coupled receptors and integrins leads to activation of the RhoA/ROCK cascade, a pathway with a central role in generation of actin-myosin contractility and in regulation of cytoskeletal dynamics. The RhoA/ROCK system has been implicated in the pathogenesis of fibrosis of the pressure-overloaded myocardium through effects on fibroblasts, cardiomyocytes, immune cells and vascular cells [293]. The evidence on the role of RhoA/ROCK in diabetic fibrosis is limited to in vitro studies and non-specific pharmacologic inhibition experiments. RhoA/ROCK signaling has been implicated in activation of a matrix-synthetic phenotype in cardiac fibroblasts exposed to high glucose [294],[295]. Moreover, in a rat model of type 2 diabetes induced through high fat diet and low dose STZ injection, treatment with the Rho kinase inhibitor fasudil attenuated myocardial fibrosis and improved dysfunction [296]. Whether these effects reflect abrogation of fibroblast-specific Rho signaling remains unknown.

3.4.1.9.4. YAP/TAZ

The homologous transcriptional coactivators YAP and TAZ are the main effectors of the Hippo pathway, and have been implicated in the pathogenesis of fibrosis in many tissues, serving as key regulators of cellular responses to mechanical stress [297],[298],[299]. Although emerging evidence suggests a role for components of the Hippo pathway in regulation of metabolic function in homeostasis and disease [300], whether YAP/TAZ signaling plays a role in diabetic cardiac fibrosis remains virtually unknown. YAP was found to be activated in diabetic hearts, and in a mouse model of heart failure induced through pressure overload and a high fat diet, YAP signaling was implicated in the pathogenesis of heart failure [301]. However, the potential role of YAP/TAZ as a link between metabolic stress and cardiac fibrosis has not been explored.

3.4.1.10. Epigenetic mechanisms in diabetes-associated cardiac fibrosis

Epigenetic mechanisms that may be involved in the pathogenesis of diabetic fibrosis through effects in regulation of cardiac fibroblast gene transcription [302],[303], include post-transcriptional modifications of histone tails [304], chemical nucleosomal DNA processing [305], and the regulatory effects of non-coding (nc) RNAs [306],[214]. Increased histone acetylation, presumably induced through the effects of histone acetyltransferases (HATs), has been associated with myocardial fibrosis in STZ-induced type 1 diabetic rats [307]. Hyperglycemia-mediated Smad2 acetylation by factor acetyltransferase 300 (FATp300) has been implicated in TGF-ß-stimulated diabetic cardiac fibrosis [305]. Histone deacetylases (HDACs) on the other hand, remove acetyl groups from histone tails and have also been implicated in fibrotic diabetic cardiomyopathy [308],[309]. Whether these epigenetic fibrogenic actions are due to direct activation of fibroblasts, or reflect actions on other myocardial cell types, such as cardiomyocytes and vascular cells, remains unknown.

Emerging evidence suggests that microRNAs (miRNAs) and long non-coding RNAs (lncRNAs), may also play an important role in diabetic myocardial fibrosis [310]. Experimental evidence suggests that exposure to high glucose may shift the balance of miRNAs in the diabetic heart, increasing levels of pro-fibrotic miRNAs, such as miR-21 [306] and miR-155 [311], and downregulating the antifibrotic miRNAs miR15a/b [312] and miR133a [313]. The effects of high glucose on the miRNA profile are mediated at least in part through upregulation of Dicer [314], an enzyme crucial for miRNA maturation. miRNAs modulate fibrosis either directly through effects on key fibroblast-activating signals, such as ion channels, TGF-β cascades, angiotensin II signaling, RhoA/ROCK, MAPK and Myocardin-related Transcription Factor (MRTF)/Serum-response Factor (SRF) [315],[316],[317],[318], or indirectly through effects on cardiomyocyte survival [319], and on vascular and immune cell phenotype and function [320],[321].

Several lncRNAs have also been implicated in diabetic cardiac fibrosis. Emerging evidence suggests that diabetes perturbs the lncRNA expression profile in the myocardium, favoring synthesis of pro-fibrotic lncRNAs and reducing levels of transcripts with anti-fibrotic properties. The basis for the distinct patterns of regulation remains unknown. In a model of STZ-induced type 1 diabetes, the lncRNA ANRIL was found to mediate myocardial ECM synthesis in the diabetic myocardium [322]. Diabetes was also associated with increased expression of several other fibrogenic lncRNAs, including metastasis-associated lung adenocarcinoma transcript 1 (lncR-MALAT1), which has been involved in activation of the TGF-ß/Smad axis [323], myocardial infarction-associated transcript (MIAT), which may act by inhibiting miR-214–3p [324], and CircRNA_000203, a circRNA that may increase synthesis of fibrosis-associated genes by sponging miR-26b-5p [325]. LncKCNq1ot1 was also upregulated in diabetic hearts and was found to activate the TGF-ß cascade, promoting collagen deposition [326]. On other hand, diabetes-associated suppression of anti-fibrotic lncRNAs, such as colorectal neoplasia differentially expressed (Crnde) [327] and HOX transcript antisense RNA (HOTAIR) has also been implicated in the pathogenesis of myocardial fibrosis [328]. It should be emphasized that the cellular basis for the fibrosis-regulating effects of lncRNAs in the diabetic heart remains unknown, and may involve both direct effects on fibroblasts and actions on other myocardial cell types.

3.4.1.11. The role of the matricellular proteins

Tissue injury is associated with secretion of matricellular proteins, a family of structurally unrelated glycoproteins that enrich the ECM and do not play a primary structural role, but bind to ECM proteins and cell surface receptors, transducing or modulating signaling cascades, and regulating activity of cytokines, proteases and growth factors [329]. Some experimental evidence suggests that metabolic perturbations associated with diabetes may induce expression of matricellular macromolecules, generating a matrix-driven fibrogenic environment. The prototypical matricellular protein TSP-1 enriches the extracellular matrix in diabetic tissues [61],[330],[331] and has been suggested to mediate fibrogenic actions by activating TGF-β [332],[333], by stabilizing the ECM through inhibition of MMP activity and by promoting vascular rarefaction [61]. Tenascin-C, a matricellular macromolecule with fibrogenic and inflammatory properties was significantly upregulated in biopsied myocardium from patients with dilated cardiomyopathy and concomitant type 2 diabetes, and its expression was associated with increased collagen deposition [334]. Whether these associations reflect a causative role for tenascin-C in diabetic cardiac fibrosis has not been examined. Osteopontin was also found to be upregulated in diabetic heart and may promote fibrosis by mediating angiotensin II and TGF-β -induced fibroblast activation and by promoting macrophage recruitment [335].

3.4.1.12. The role of advanced glycation end-products in diabetic cardiac fibrosis

The fibrogenic actions of hyperglycemia are mediated, at least in part, through modifications in the biochemical properties of collagen. High concentrations of glucose react non-enzymatically with amino groups in proteins, lipids and nucleic acids, thus generating AGEs. AGEs promote irreversible cross-linking of collagen making it stiffer, less soluble and resistant to degradation [336],[337]. Thus, in patients with diabetes, hyperglycemia-induced collagen glycosylation may reduce ventricular compliance, promoting diastolic dysfunction. In addition to their effects on the mechanical properties of collagens, AGEs may also contribute to fibrotic cardiac remodeling by binding to the cell surface receptors RAGE and AGE-R3, which are also upregulated in diabetes [338], thus transducing downstream fibrogenic signaling cascades. The pro-fibrotic effects of AGE-RAGE signaling involve direct activation of Erk MAPK [339],[340], activation of TGF-β [341], deposition of the matricellular macromolecule CCN2 [342], and stimulation of NF-κB-mediated collagen synthesis [343]. N(ε)-carboxymethyllysine (CML), a major AGE formed by combined reactions of non-enzymatic glycation and oxidation (glycoxidation), was found to be increased in the hearts of patients with diabetes [344]. The fibrogenic actions of the AGE/RAGE axis are supported by experiments showing that exposure to AGEs enhanced expression of markers of fibrosis (TGF-ß, α-SMA, CTGF, coll I-α1, coll III-α1, Fn1) in engineered cardiac tissue in a RAGE-dependent manner [263].

4. Targeting diabetes-associated cardiac fibrosis

Considering the involvement of fibrosis in the pathogenesis of diastolic dysfunction and the high prevalence of HFpEF in patients with diabetes, it is tempting to hypothesize that anti-fibrotic strategies may be beneficial in preventing diabetes-associated heart failure progression. In fact, the protective effects of optimal glycemic control, and the improved outcome in subjects with diabetes receiving anti-diabetic medications, ACE inhibitors or statins may reflect, at least in part attenuation of myocardial fibrosis.

4.1. Does tight glycemic control prevent cardiac fibrosis in subjects with diabetes?

The role of hyperglycemia in the pathogenesis of diabetic cardiac fibrosis is supported by extensive in vitro and in vivo evidence. Moreover, in patients with type 1 and type 2 diabetes, levels of glycosylated hemoglobin are associated with cardiac magnetic resonance indicators of myocardial fibrosis [18],[24],[345]. Although these associative findings are consistent with the notion that tight glycemic control may delay progression of fibrosis in patients with diabetes, direct evidence demonstrating that protective effects of glucose lowering on clinical outcome are mediated through attenuation of cardiac fibrotic remodeling is lacking. Although several clinical trials demonstrated that intensive anti-hyperglycemic treatment delayed progression of microvascular complications in patients with diabetes, effects on cardiovascular complications were less consistent [346],[347]. In a meta-analysis, tight glycemic control did not reduce heart failure events in patients with diabetes [348], and in the ACCORD study, intensive glucose lowering did not affect incident atrial fibrillation, a condition typically associated with atrial fibrotic changes [349]. There are several potential explanations for the negative findings. First, heart failure progression in patients with diabetes may be slow, reflecting the late development of fibrotic changes in diabetic hearts. Thus, long-term follow up may be required to document the benefits of glucose lowering. Second, in clinical trials, intensive glycemic control is compared to standard therapy, and not to uncontrolled hyperglycemia. Although this approach is obviously therapeutically relevant, negative findings do not exclude effects of hyperglycemia on cardiac remodeling and fibrosis, but may suggest the absence of additional benefit with aggressive glucose lowering. Third, glucose-independent mechanisms may contribute to the pathogenesis of diabetic myocardial fibrosis.

4.2. Antifibrotic effects of glucose-lowering drugs

In addition to reducing fibrosis through their glucose lowering effects, several classes of anti-diabetic agents (including metformin, thiazolidinediones, SGLT2 inhibitors and incretin-based drugs) have been suggested to exert direct anti-fibrotic actions (Table 4). Thus, direct inhibition of fibrogenic signaling may be important in mediating the protective actions of glucose-lowering agents in patients with diabetes.

Table 4:

Animal model-based evidence on the effects of glucose-lowering agents in diabetes-associated cardiac fibrosis

Agent	Animal model	Effects on cardiac fibrosis, myocardial structure and function	Proposed mechanism of protection	Ref
Biguanide
Metformin	STZ-induced type I diabetes	Reduction in collagen I and III accumulation, reduced arteriolar wall thickness, accompanied by attenuated cardiomyocyte apoptosis and decreased hypertrophy	Stimulation of liver kinase B-1 (LKB-1)-mediated phosphorylation of AMP-activated protein kinase (AMPK), Akt activation, suppression of GSK-3β and p38MAPK	[351]
Metformin	Spontaneously hypertensive, insulin-resistant rats (SHHF)	Reduction in perivascular fibrosis, and cardiac lipid accumulation, accompanied by attenuated left ventricular (LV) remodeling, and improved systolic and diastolic function.	Activation of AMPK, increase in endothelial nitric oxide synthase (eNOS) and vascular endothelial growth factor (VEGF), reduced tumor necrosis factor (TNF)-α expression.	[98]
Metformin	Alloxan-induced type 1 diabetes in mongrel dogs	Reduction in diastolic LV stiffness. Attenuated formation of AGE-mediated collagen crosslinks without reduction in collagen levels.	N/A	[354]
Thiazolidinediones
Pioglitazone	STZ-induced type I diabetes in rats	Reduction in cardiac fibrosis	Reduction in oxidative stress, inflammation and fibrosis, attributed to inhibition of Ca2+/calmodulin-dependent protein kinase II (CaMKII)/NFkB/TGF-β1 axis, and activation of PPARγ	[358]
Rosiglitazone	Otsuka Long–Evans Tokushima Fatty (OLETF) rats	Reduction in cardiac fibrosis accompanied by improved diastolic function	Attributed to suppressed RAGE and CCN2 expression.	[356]
Rosiglitazone	Type 2 diabetic (induced through high fat diet and low dose STZ) and hypertensive (renal artery occlusion) rats	Attenuated fibrosis evidenced by reduced expression of collagen I and III.	Attributed to decreased TGF-β and TNF-α, and increased MMP-2 expressions.	[357]
Sodium-glucose co-transporter-2 (SGLT2) inhibitors
Dapagliflozin	STZ-induced type 1 diabetes in mice	Reduction in myocardial fibrosis and cardiomyocyte apoptosis, accompanied by improved cardiac function	Suppression of reactive oxygen species and endoplasmic reticulum (ER) stress in cardiomyocytes	[365]
Empagliflozin	STZ-induced type 1 diabetes in mice	Reduction in collagen III deposition, accompanied by improved systolic LV function	Restoration of insulin signaling leading to reduction in cardiac stromal cell prosenescent phenotype through Akt activation and p38 MAPK inhibition.	[373]
Empagliflozin	STZ-induced type 1 diabetes in rats subjected to myocardial infarction	Reduction in cardiac fibrosis and hypertrophy, accompanied by attenuated post-infarction remodeling	Increase in cardiac GTP enzyme cyclohydrolase 1 (cGCH1) resulting in increased NO levels and decreased O2− and nitrotyrosine levels	[472]
Empagliflozin	db/db mice	Reduction in myocardial fibrosis, accompanied by attenuated diastolic dysfunction.	Attributed to reduction in serum/glucocorticoid regulated kinase 1 (SGK1) and epithelial sodium channel (ENaC)	[64]
Empagliflozin	db/db mice	Amelioration of cardiac interstitial and pericoronary fibrosis, reduced coronary arterial thickening and decreased cardiac macrophage infiltration	Attenuation of oxidative stress and impaired vascular dilating function	[371]
Dapagliflozin	db/db mice + angiotensin II infusion	Reduction in collagen I deposition and attenuated inflammation, accompanied by improved LV systolic function	Reduction in intracellular calcium load, decreased ROS, protection of mitochondria, attenuated inflammation.	[79]
Dapagliflozin	BTBR ob/ob mice	Decrease in myocardial collagen I and III deposition, attenuated myocyte apoptosis and reduced inflammation	Attributed to activation of AMPK/p-mTOR/Akt, activation of a FOXO3a pathway, and subsequent NLRP3 downregulation	[374]
Empagliflozin	KK-Ay mice fed a high-fat diet	Reduction in collagen I, III accumulation and αSMA expression, accompanied by improved systolic and diastolic LV function	Attributed to inhibition of transforming growth factor (TGF) β/Smad pathway and reduction in oxidative stress	[372]
Dapagliflozin	STZ- and high fat diet-induced type 2 diabetes in rats	Decrease in myocardial fibrosis and attenuated cardiomyocyte apoptosis	Attributed to inhibition of oxidative stress, decreased inflammation, and reduced sympathetic nerve fiber activity	[473]
Empagliflozin	Model of metabolic syndrome with prediabetes, (SHRcp rats)	Reduction in interstitial fibrosis and cardiac hypertrophy	Attributed to attenuated cardiac oxidative stress and reduced inflammation.	[364]
Incretin-based drugs: glucagon-like peptide (GLP)-1 agonists
Liraglutide	STZ-induced type 1 diabetes in rats	Reduction in myocardial collagen I and III levels	Attributed to decreased MMP-1, -9 expression and reduced prolyl-4-hydroxylase (a key enzyme in collagen synthesis).	[379]
Liraglutide	Zucker diabetic fatty (ZDF) rats	Reduction in cardiac fibrosis, accompanied by improved cardiac function.	Attributed to activation of AMPK and downregulation of mTOR	[377]
Liraglutide	STZ- and high fat diet-induced type 2 diabetes in rats	Reduction in perivascular and interstitial fibrosis and attenuation of cardiac hypertrophy	N/A	[474]
Liraglutide	Mice fed a high-fat diet	Reduction in cardiac fibrosis	Attributed to attenuated inflammation and reduced oxidative stress	[380]
Liraglutide	Mice fed a high-fat diet	Reduction in markers of cardiac fibrosis and hypertrophy, accompanied by improved cardiac function	Attributed to prevention of insulin resistance, activation of AMPK, attenuated inflammation and endothelial nitric oxide synthase (eNOS) downregulation.	[378]
Incretin-based drugs: inhibitors of dipeptidyl peptidase-4 (DPP-4)
Sitagliptin	STZ-induced type 1 diabetes	Reduction in cardiac fibrosis, decreased thickness of arteriolar walls, reduced cardiomyocyte apoptosis and attenuated hypertrophy.	Attributed to AMPK stimulation, Akt activation and p38 MAPK inhibition.	[351]
Vidagliptin	STZ-induced type 1 diabetes	No impact on fibrotic changes. Reduced cardiomyocyte injury and improved microvascular function	N/A	[475]
Alogliptin	Alloxan-induced type 1 diabetes in rabbits	Alleviation of interstitial fibrosis, and attenuation of ventricular hypertrophy, accompanied by reduced diastolic dysfunction	Attributed to improved mitochondrial function.	[384]
Sitagliptin	db/db mice	Reduction in cardiac fibrosis	Attributed to reduced TGF-β, levels, attenuated AGE-and ROS-mediated effects.	[382]
Sitagliptin	Goto-Kakizaki (GK) rats	Reduction in cardiac fibrosis (evidenced by decreased fibronectin expression), and attenuated hypertrophy.	Attributed to reduced TGF-β1 and CCN2 and to activation of PPAR-δ.	[383]
DPP-4 inhibition with MK0626	STZ- and high fat diet-induced type II diabetes in mice	Modest cardiac hypertrophy, impair ment of cardiac function, and dysregulated expression of genes and proteins controlling inflammation and cardiac fibrosis	N/A	[476]
DPP-4 inhibition with MK0626	Mice fed a high-fat/high-fructose diet	Decrease in cardiac fibrotic changes, accompanied by normalization of cardiac diastolic relaxation	Attributed to reduced oxidative stress and decreased insulin resistance.	[381]

4.2.1. Metformin

One of the oldest and most effective antidiabetic drugs, metformin, attenuates diabetes-associated myocardial fibrosis [350] both by lowering glucose levels and through direct anti-fibrotic actions. Metformin-induced inhibition of fibrosis may involve AMPK activation [351] and AMPK-independent effects [98]. The anti-fibrotic actions of metformin have been attributed to inhibition of neurohumoral fibroblast activation [352], to disruption of the TGF-β/Smad3 signaling cascade [353], and to attenuation of the formation of AGE-mediated collagen cross-links [354]. A meta-analysis of heart failure studies suggested that metformin may have beneficial effects in HFpEF patients [355]. These protective actions may involve inhibition of fibrotic cardiac remodeling.

4.2.2. Thiazolidinediones

Thiazolidinediones act as peroxisome proliferator-activated receptor γ (PPARγ) agonists and thus exert favorable metabolic effects on glucose and lipid metabolism. In addition to their metabolic actions, thiazolidinediones have been suggested to inhibit myocardial fibrosis by disrupting RAGE signaling, through attenuation of oxidative stress, and via suppression of TGB-β1 and CCN2-mediated fibrogenic responses [356], [357], [358]. Thiazolidinediones have been shown to exert direct modulatory effects on fibroblasts and myofibroblasts, inhibiting proliferation and attenuating matrix synthesis through PPAR-γ-mediated and PPAR-γ-independent actions [359],[360].

The clinical significance of these effects is unclear. In fact, clinical trials have demonstrated that members of the thiazolidinedione family (including rosiglitazone and pioglitazone) increased the risk for hospitalization due to heart failure [361]. These deleterious effects are related to reduced renal excretion of sodium and fluid retention, and do not involve cardiac interstitial alterations.

4.2.3. Antifibrotic effects of SGLT2 inhibitors

The SGLT2 inhibitors (empagliflozin, canagliflozin, dapagliflozin and erugliflozin) reduce cardiovascular mortality and heart failure hospitalization rates in patients with type 2 diabetes [362]. In addition to their glucose-lowering properties, SGLT2 inhibitors have non-glycemic effects such as natriuresis, body weight and visceral fat reduction, blood pressure- and lipid-lowering effects that may contribute to their cardioprotective actions [99]. Several lines of evidence suggest that SGLT2 inhibitors may also act by reducing diabetes-associated myocardial fibrosis. First, in vitro studies showed that SGLT2 inhibitors inhibit functional activation of cardiac fibroblasts [363]. Second, in rodent models of diabetes and metabolic syndrome empagliflozin and dapagliflozin reduced cardiac interstitial and peri-coronary fibrosis [364],[64],[365],[366] through effects that may be independent of changes in blood pressure and glycemic control [64],[364]. Third, in patients with type 2 diabetes, empagliflozin treatment decreased left ventricular mass and diastolic dysfunction [367], and reduced myocardial extracellular volume assessed through CMR, findings consistent with inhibition of the myocardial fibrotic response [368].

The antifibrotic effects of SGLT2 inhibitors on the heart may be due to attenuated metabolic cardiomyocyte injury [79], reduced inflammatory macrophage activation [369],[370] or direct inhibition of cardiac fibroblast activity. SGLT2 inhibition may suppress fibroblast function through several mechanisms, including attenuation of oxidative stress [371], inhibition of TGF-ß/R-Smad signaling cascades [372], downmodulation of p38 MAPK signaling [373], AMPK activation and inhibition of NLRP3-driven inflammation [374]. Unfortunately, although attenuation of fibrosis is a plausible mechanism for the protective effects of SGLT2 inhibitors on the diabetic heart, the molecular basis for these effects is currently unclear, and the evidence supporting candidate pathways is correlative.

4.2.4. Incretin-based drugs

The two families of incretin-based drugs, the glucagon-like peptide (GLP)-1 receptor agonists and the dipeptidyl peptidase-4 (DPP-4) inhibitors, act by stimulating insulin secretion by pancreatic β-cells. In the LEADER (Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results) trial, the GLP-1 receptor agonist liraglutide was found to reduce cardiovascular mortality in patients with type 2 diabetes [375]. Analysis of the data from the LEADER trial suggested that improved glycemic control (assessed through HbA1c levels) may be the major mediator of cardiovascular benefit in liraglutide-treated patients [376]. Inhibition of cardiac fibrosis, either through optimization of glycemic control, or via direct antifibrotic actions may be partially responsible for the protective effects of liraglutide. In several studies using experimental models of type 2 diabetes, treatment with liraglutide attenuated cardiac fibrosis and improved dysfunction [377],[378]. The antifibrotic effects may involve protective actions on cardiomyocytes [377] and direct inhibition of fibroblast activity [379], and have been attributed to attenuated oxidative stress [380] and accentuated AMPK activation [377].

Large clinical trials using DPP-4 inhibitors (alogliptin, linagltiptin, saxagliptin and sitagliptin) in patients with diabetes did not show significant effects on cardiovascular outcome [346]. Although some studies have suggested antifibrotic effects of DPP-4 inhibitors that may improve dysfunction in diabetic cardiomyopathy models [381],[351],[382],[383],[384] the relevance of any such effects in the clinical context may be limited. A growing body of clinical evidence suggests that some members of the DPP-4 inhibitor family may worsen heart failure in patients with pre-existing ventricular dysfunction, or even precipitate heart failure in high-risk individuals [385],[386],[387]. It has been suggested that these detrimental effects may involve increased expression of the CXC chemokine Stromal Cell Derived Factor (SDF)-1/CXCL12 (due to attenuated DPP-4-mediated proteolysis) [388] that may enhance inflammation and promote fibrosis in the myocardium [389]. However, robust experimental evidence supporting this concept, and dissecting the mechanisms responsible for heart failure development upon DPP-4 inhibition is lacking.

4.3. Anti-fibrotic effects of RAAS inhibition in patients with diabetes.

Inhibition of the RAAS has protective effects in patients with diabetes [390]. ACE inhibition reduced mortality, cardiovascular events and the incidence of heart failure in a high-risk patient population that included individuals with diabetes with at least one additional risk factor [391]. Angiotensin receptor blockers had similar protective effects in high-risk populations with diabetes [392], and were found to reduce the rate of first hospitalizations for heart failure [393]. Some of the protective effects of RAAS inhibition in patients with diabetes patients may involve anti-fibrotic actions and attenuation of myocardial fibrosis. In animal models, ACE inhibitors, AT1 receptor blockers and aldosterone antagonists ameliorated diabetes-induced cardiac fibrosis, through effects attributed to reduced expression and activity of pro-fibrotic growth factors (such as TGF-ß1), attenuated oxidative stress, suppressed inflammatory activity, and alterations in the MMP:TIMP balance (Table 5) [193],[213],[394],[395],[204],[396],[397]. However, considering the broad effects of the RAAS on cardiomyocyte function and in arrhythmogenesis, and its role in regulation of systemic blood pressure and renal function, the relative contribution of attenuated myocardial fibrosis to the clinical benefit of RAAS disruption in individuals with diabetes is unclear. In patients with obesity and metabolic syndrome, correlative data suggested an association between the effects of aldosterone inhibition on ventricular function and the reduction in circulating fibrotic markers [199],[398]. Moreover, in the RALES heart failure trial (which included patients with diabetes), patients exhibiting the highest benefit from aldosterone antagonism had higher baseline levels of collagen synthesis serum biomarkers, which were significantly reduced after treatment with spironolactone [399].

Table 5:

Effects of inhibitors of the renin-angiotensin aldosterone system (RAAS) in animal models of diabetic cardiac fibrosis.

Agent	Animal model	Effects on cardiac fibrosis, myocardial structure and function	Proposed mechanism of protection	Ref
(Pro)renin receptor (PRR) inhibition by RNA interference (RNAi) silencing	STZ-induced type 1 diabetes in rats	Reduction in cardiac collagen I and III, accompanied by improved LV systolic and diastolic function	Inhibition of oxidative stress and ERK signaling, downregulation of TGF-β	[206]
Renin inhibition (aliskiren), ACE inhibition (benazepril), AT1R blockade (candesartan)	STZ-induced type I diabetes in rats	Reduction in cardiac fibrosis (partial by candesartan and benazepril alone, complete by aliskiren) and myocyte apoptosis (partial by all three agents)	Inhibition of oxidative stress (partial by candesartan and benazepril alone, complete by aliskiren)	[193]
ACE inhibition (ramipril)	STZ- and high-fat diet- induced type 2 diabetes in rats	Reduction in perivascular and interstitial fibrosis and cardiac hypertrophy	N/A	[474]
AT1R blockade (irbesartan)	STZ-induced type 1 diabetes in rats	Reduction in cardiac fibrosis and cardiac remodeling	Downregulation of TGF-β/Smad2/3 pathway	[394]
AT1R blockade (telmisartan)	STZ-induced type 1 diabetes in rats	Reduction in cardiac fibrosis and improvement of cardiac function	Activation of PPARδ/STAT3/CTGF/MMP-9 pathway	[396]
AT1R blockade (valsartan)	KK-Ay mice with diabetic nephropathy	Reduction in cardiac fibrosis, (suggested by decreased deposition of collagen IV and fibronectin)	Attributed to downregulation of miR-21	[204]
AT1R blockade (losartan)	Rats fed a high-fat diet	Attenuated collagen I and III synthesis	Attributed to disruption of TGF-β1/Smad3.	[213]
AT1R blockade (irbesartan)	STZ- and high-calorie diet-induced type 2 diabetes in rats	Attenuated cardiac collagen I and collagen III levels, decreased ratio of collagen I/collagen III, and reduced cardiac hypertrophy	Attributed to inhibition of the AGE-RAGE pathway and to alterations in MMP/TIMP expression profile.	[395]
Aldosterone receptor antagonism (spironolactone)	STZ-induced type 1 diabetes in rats	Reduction in cardiac fibrosis and hypertrophy	Reduction in oxidative stress, attenuated inflammation and reduced mitochondrial dysfunction	[397]
Ang-(1–7)	db/db mice	Reduction in cardiac fibrosis, hypertrophy and apoptosis. Improvement in LV systolic function	Attenuation of inflammation, oxidative stress, lipid accumulation	[200]
Ang-(1–7)	db/db mice	Reduction in cardiac fibrosis, hypertrophy. Improvement in LV diastolic function	Reduction in inflammation and cardiac lipotoxicity	[477]
AT1R blockade (telmisartan)/neprilysin inhibition (thiorphan)	STZ-induced type 1 diabetes in rats	Reduction in cardiac fibrosis and apoptosis	Reduction in inflammation and histone acetylation	[307]
AT1R blockade (valsartan)/neprilysin inhibition (sacubitril)	STZ-induced type 1 diabetes in mice with reperfused myocardial infarction	Reduction in cardiac fibrosis, accompanied by improved LV systolic function	Attributed to suppression of TGF-β	[478]

Modulation of the natriuretic system through inhibition of neprilysin, the enzyme that inhibits and degrades natriuretic peptides is an effective therapeutic strategy in patients with heart failure [400]. The combination of angiotensin receptor blockade and neprilysin inhibitor (ARNi) (valsartan and sacubitril) reduced risk of death and hospitalization more effectively than a standard monotherapy with ACE inhibitor alone in patients with HFrEF [401]. Atrial natriuretic peptide (ANP) and brain (B-type) natriuretic peptide (BNP) are released by failing and remodeling myocardium and exert protective actions by promoting diuresis and natriuresis, by stimulating vasodilation, and by inhibiting the RAAS [402]. Moreover, BNP has been demonstrated to exert anti-fibrotic actions, attributed to disruption of fibrogenic TGF-β/Smad2/3 signaling. BNP actions on cardiac fibroblasts involve activation of a cGMP-PKG-MEK-ERK pathway [403]. Although the exact mechanism for ERK-mediated inhibition of the TGF-β response is unclear, it has been suggested that ERK may phosphorylate Smad2 and Smad3 in the linker region, thus inhibiting TGF-β-induced C-terminal Smad activation, disrupting nuclear translocation of Smads, and downregulating transcription of profibrotic genes [404]. Considering the antifibrotic effects of BNP, which are attributed to disruption of the TGF-β system [405],[403], it is tempting to hypothesize that ARNi therapy may have more pronounced effects on diabetic myocardial fibrosis than AT1 receptor blockade alone. Experimental evidence in a rat model of type 1 diabetes showed that ARNi therapy with telmisartan/thiorphan had more pronounced effects on cardiac fibrosis than the monotherapies alone, associated with improved cardiac function [307]. The protective effects were accompanied by reduced activation of inflammatory and TGF-β-mediated fibrogenic pathways [403].

4.4. The anti-fibrotic effects of lipid lowering agents

Statins (HMG-CoA reductase inhibitors) and cholesterol absorption inhibitors (such as ezetimibe) may benefit patients with diabetes, not only through their lipid lowering and anti-inflammatory actions, but also by exerting antifibrotic effects. In experimental models of diabetic cardiomyopathy, treatment with statins, or ezetimibe attenuated myocardial interstitial fibrosis, suppressing inflammation and decreasing oxidative stress [406],[176]. These protective actions were associated with improved ventricular function. In vitro experiments suggest direct effects of statins on cardiac fibroblast activation. Simvastatin and atorvastatin inhibited proliferation of atrial and to a lesser extent of ventricular human cardiac fibroblasts [120]; their anti-proliferative effects were attributed to inhibition of the TGF-β1/RhoA/ROCK-1 axis [407],[408]. Statins may also exert anti-fibrotic actions by inducing fibroblast apoptosis, by attenuating fibroblast migration [121], and by reducing collagen synthesis [122]. To what extent the direct effects of statins on fibroblasts contribute to their anti-fibrotic actions in models of diabetes is unclear. Although extensive evidence suggests that statins attenuate renal fibroblast activation in a high glucose environment [409], similar evidence in cardiac fibroblasts is lacking. However, AGE-induced proliferation and differentiation of cardiac fibroblasts into myofibroblasts was found to be reduced by atorvastatin through PPARγ-mediated inhibition of an AGE-RAGE-ERK1/2 pathway [123].

4.5. Novel therapeutic approaches

4.5.1. Inflammatory cytokines as therapeutic targets in diabetes-associated fibrosis

Experimental studies suggest the involvement of the pro-inflammatory cytokines IL-1β, TNF-α, and IL-6 in the pathogenesis of diabetes-associated cardiac fibrosis [212]. Targeting IL-1β through administration of an interleukin converting enzyme inhibitor, was found to attenuate dysfunction, reducing fibrosis in a rat model of STZ-induced type 1 diabetes [410]. Moreover, inhibition of several other inflammatory signals, including TNF-α, IL-6 and CXCR4 (the receptor for the chemokine CXCL12/SDF-1), has been suggested to reduce diabetes-associated myocardial fibrosis in animal models [213],[211],[411]. Inflammatory cascades have also been implicated in the pathogenesis of metabolic dysfunction [216, 217]. Thus, administration of biologics neutralizing pro-inflammatory cytokines is an attractive and promising strategy that may exert beneficial systemic and cardiac effects. Despite the availability of effective agents to neutralize IL-1β, TNF-α and IL-6 and the promising experimental evidence, clinical data supporting this approach are lacking. In the Canakinumab Antiinflammatory Thrombosis Outcome Study (CANTOS) trial, administration of the monoclonal anti-IL-1β antibody canakinumab in a population of high-risk cardiovascular patients reduced the incidence of cardiovascular events [412] and decreased hospitalization rates for heart failure and heart failure–related mortality [413]. In contrast to reports suggesting reduction in HbA1c levels in individuals with diabetes treated with canakinumab [414], in the CANTOS study, only transient effects on glycemic control were noted [415].

4.5.2. The TGF-β superfamily as a therapeutic target in diabetic fibrosis

Considering the central role of the TGF-β signaling pathway in cardiac fibrosis and in diabetes-associated fibrogenic responses, targeting TGF-β/Smad3 signaling may be effective in attenuating diabetic fibrotic myocardial remodeling and dysfunction. However, due to the wide range of their effects on many different cell types, and their context-dependent functions, TGF-βs are challenging therapeutic targets. Although clinical studies testing the effectiveness of TGF-β inhibition in diabetic cardiac fibrosis have not been performed, the findings from a small clinical study targeting TGF-β1 in diabetic nephropathy were disappointing. In a group of patients with advanced disease, anti-TGF-β1 antibody neutralization did not affect renal function and proteinuria, and did not reduce biomarkers reflecting matrix remodeling [416]. Whether the negative results reflect ineffective neutralization, actions of other TGF-β isoforms, or the focus on a population with advanced disease (that may exhibit irreversible fibrosis) is unclear. In addition to approaches specifically targeting the TGF-β system, such as neutralizing anti-TGF antibodies and TβR kinase inhibitors [417], pharmacological agents with broad anti-fibrotic actions may act predominantly by disrupting the TGF-β system. Pirfenidone is used for the treatment of idiopathic pulmonary fibrosis and exerts antifibrotic actions, at least in part, by suppressing TGF-β responses [418]. In a model of STZ-induced type 1 diabetes, treatment with pirfenidone reduced myocardial fibrosis and attenuated diastolic dysfunction [196].

4.5.3. Modulating the growth factor environment to target diabetic myocardial fibrosis

In addition to TGF-β, several other growth factors may be implicated in the pathogenesis of diabetic cardiac fibrosis. Although the relative contribution of these growth factors remains poorly understood, some experimental studies have suggested that modulating the growth factor environment in the diabetic heart may attenuate fibrosis, ameliorating cardiac function. The downregulation of the expression of cardiac insulin-like growth factor (IGF)-1 receptor (IGF-1R) in diabetes by berberine, a natural alkaloid, was found to attenuate fibrosis in a model of type 2 diabetes induced through low dose STZ and a high-fat diet [419].

On the other hand, therapeutic administration of certain growth factors may inhibit diabetic cardiac fibrosis by limiting metabolic injury, or by accentuating angiogenic responses. The stress-inducible hormone Fibroblast Growth Factor 21 (FGF21) is an important regulator of energy balance, glucose and lipid homeostasis [420]. In a model of type 2 diabetes, administration of FGF21 reduced cardiac fibrosis and attenuated dysfunction, exerting AMPK-mediated protective effects on cardiomyocytes and suppressing inflammation [421]. Administration of angiogenic growth factors, such as VEGF has also been suggested as a promising antifibrotic cardioprotective strategy, acting indirectly by preserving the microvasculature, thus preventing fibrogenic remodeling [247],[422].

4.5.4. Antioxidant strategies to inhibit diabetic cardiac fibrosis

Glucolipotoxic oxidative injury plays a central role in fibroblast activation in diabetic hearts. Thus, administration of antioxidants may be a promising approach to halt the progression of diabetic myocardial fibrosis and to attenuate development of ventricular dysfunction. In animal models of diabetic heart disease, several different antioxidants, including N-acetylcysteine (NAC), the xanthine oxidase inhibitor allopurinol and alpha lipoic acid were found to reduce interstitial myocardial fibrosis [423],[263],[424],[425]. However, despite the promising animal model data, in the clinical setting, antioxidant approaches, (using vitamin C, E and beta-carotene administration), failed to improve cardiovascular outcome in high-risk patients, including subjects with diabetes [426],[427].

4.5.5. Targeting AGE-mediated fibrosis and matrix crosslinking

The important role of the AGE/RAGE axis in diabetic cardiac fibrosis suggests that strategies reducing AGE formation, accentuating AGE degradation and breaking AGE-mediated crosslinks may hold promise as therapy for diabetic fibrotic cardiomyopathy associated with diastolic dysfunction [428]. Aminoguanidine (also known as pimagedine) is the most extensively used inhibitor of AGE generation. In animal models of diabetes, aminoguanidine administration was found to attenuate vascular protein crosslinking, and reduced myocardial collagen deposition [429],[336]. These antifibrotic effects were associated with improved diastolic function. Moreover, in a model of STZ-induced type 1 diabetes, treatment with the AGE cross-link breaker ALT-711 (alagebrium) reduced de novo expression of collagen in the myocardium and restored collagen solubility [338]. Beneficial effects of AGE breakers on the diabetic heart may also involve potentiation of the blood pressure-lowering effects of antihypertensive medications due to attenuated vascular fibrosis [430]. Unfortunately, translation of these promising observations to the clinical context has been challenging. Although some clinical studies suggested protective effects of ALT-711 in improving arterial compliance in aged patients [431], the BENEFICIAL clinical trial in 102 HFrEF patients failed to show effects on systolic and diastolic function, suggesting that ALT-711 did not affect peripheral AGE accumulation [432]. The absence of beneficial effects may reflect the limited enrollment of patients with diabetes, and the lack of studies in HFpEF patients, who may be better candidates for AGE breakers.

4.5.6. Cell therapy approaches

Over the last 20 years, a large body of experimental work has suggested anti-fibrotic effects of various cell therapy approaches in models of myocardial injury. The mechanistic underpinnings of these effects are not always clear, and interpretation of the findings is often challenging due to the differences in the characteristics of the cells used. In experimental models of diabetes, bone marrow-derived progenitor cells have been reported to exert anti-fibrotic actions in the diabetic myocardium [433]. Similar protective actions associated with attenuated myocardial fibrosis were found in several studies using infusion of mesenchymal stem cells (MSCs) in rodent models of type 1 and type 2 diabetes [434],[435]. The anti-fibrotic effects of bone marrow progenitors and of MSCs have been attributed to secretion of anti-fibrotic growth factors, such as hepatocyte growth factor (HGF) [433] and prostaglandin E2 [434], that may lead to attenuation of downstream fibrogenic signaling cascades [435]. Some studies have suggested that MSCs may be an important source of exosomes [436] that may inhibit fibrosis by delivering miRNAs [437]. It should be noted that concerns have been raised regarding the long-term effects of MSC cell therapy in myocardial diseases. MSCs are capable of converting to fibroblasts in response to a pro-inflammatory or fibrogenic microenvironment [438]. Thus, depending on the specific cell type profile used for cell therapy, and the cellular environment, MSC infusion could exacerbate, rather than inhibit, myocardial fibrosis [389].

Recently, a chimeric antigen receptor T-cell (CAR-T)-based strategy has been proposed to target cardiac fibroblasts in failing hearts [439]. In a model of cardiac pressure overload, CAR-T cells targeting fibroblasts expressing the activation marker fibroblast activation protein (FAP) attenuated cardiac fibrosis and reduced dysfunction [439]. Concerns related to the need for early and continuous therapy may challenge clinical implementation of this intriguing strategy. Moreover, in contrast to the effects of pressure overload or myocardial infarction, diabetes does not induce FAP or myofibroblast markers in cardiac fibroblasts [62], thus limiting the potential for cell-specific targeting.

4.6. Challenges of antifibrotic approaches in patients with diabetes

Decades of experimental research have identified several promising therapeutic targets for patients with diabetes and prominent fibrotic cardiac remodeling. However, clinical translation is challenging for several major reasons. First, the contribution of cardiac fibrosis to ventricular dysfunction in patients with diabetes is unclear. Diabetes is associated with a constellation of ventricular pathologic changes involving the cardiomyocytes, microvasculature and the interstitium. The relative significance of each one of these compartments in the perturbations in organ function remains poorly understood. Intuitive thinking and some experimental evidence suggest that established fibrosis can contribute to dysfunction and arrhythmogenesis. However, as a reparative response to injury in a non-regenerative organ, cardiac fibroblast activation may also play protective roles, even in the absence of significant cardiomyocyte death [38]. It has been postulated that certain subsets of fibroblast-like cells may protect the injured myocardium by preserving the endomysial extracellular matrix network, which is important in transducing pro-survival signals in cardiomyocytes [440],[441],[38]. Thus, broad and non-selective antifibrotic therapies may abrogate important reparative processes. Second, considering the slow progression of diabetes-associated fibrotic cardiac remodeling, effective protection would require early initiation of treatment and prolonged therapy. Unless the strategy used is very selective for the maladaptive fibroblast functions, patients may be exposed to potential adverse effects, due to chronic disruption of fibroblast-mediated repair. Third, most of the information on the effectiveness of various therapeutic strategies is derived from animal models. Although these models are very useful in dissecting cellular responses and molecular mechanisms involved in regulation of diabetes-associated cardiac fibrosis, their translational value is much more limited. In particular, animal models cannot recapitulate the heterogeneity of HFpEF, the main condition associated with diabetic myocardial fibrosis [442],[443].

5. Conclusions

Interstitial/perivascular myocardial fibrosis is a prominent feature of diabetic cardiomyopathy and may be involved in morbidity and mortality in patients with diabetes by contributing to the development of heart failure and by promoting arrhythmogenesis. The pathogenesis of diabetic cardiac fibrosis involves metabolic perturbations which in turn activate a series of events in the myocardium, including oxidative stress, cytokine-driven inflammation, neurohumoral activation, growth factor release, and changes in the extracellular matrix network. These alterations activate cardiac interstitial and perivascular fibroblasts, leading to increased deposition of structural matrix proteins, and generation of crosslinks that increase ventricular stiffness. The phenotypic changes in diabetic cardiac fibroblasts seem to be distinct from those noted upon reparative or mechanical activation in conditions associated with myocardial infarction or pressure overload. Thus, there is an urgent need to understand the fundamental signals involved in diabetes-associated cardiac fibroblast activation and to dissect potential interactions with other cell types that may contribute to the fibrotic response. These insights are essential for identification of new therapeutic strategies for diabetes-associated heart failure.