MicroRNAs in β-Cell Biology, Insulin Resistance, Diabetes and Its Complications

Pancreatic β-cells play a fundamental role in glucose homeostasis, releasing insulin in response to glucose levels in the bloodstream. Insulin then triggers glucose uptake in its target tissues, such as the liver, kidney, skeletal muscle, and cardiac muscle. Absence or malfunction of β-cells leads to diabetes due to lack of insulin producing cells (T1D), or to the inability to increase insulin levels to sufficiently stimulate glucose uptake in the face of insulin resistance (T2D).

In T1D, the lack of insulin is primarily caused by the absence or destruction of pancreatic β-cells, which is driven by developmental errors or immune malfunction, respectively. A large suite of miRNAs has been implicated in pancreas (and therefore β-cell) development including miR-15a/b, miR-16, miR-195 (7), miR-503, miR-541, miR-214 (8), miR-9 (9), miR-124a (10), miR-7 (9,11), miR-376 (9) and miR-375 (9,12), among others. There remains a need for detailed studies of the role of these miRNAs in diabetes, but it is clear that mutations or misexpression of these species could lead to β-cell pathologies (see below).

Likewise, miRNAs have been implicated in the autoimmune destruction of β-cells, also leading to T1D. Recently, Hezova et al. (13) measured changes in miRNA expression in regulatory T cells (T-reg cells) of T1D-affected individuals. These cells are of special interest as they are critical regulators of autoimmune disease. They found that miR-510 was significantly upregulated, and miR-191 and miR-342 were significantly downregulated in adult peripheral T-reg cells of diabetic patients compared with healthy individuals (13). miR-342 is also known to show altered expression profiles in hematological disease (14,15). These observations suggest a role for these miRNAs in the autoimmune destruction of β-cells.

The role of miRNAs in the function of β-cells in T2D patients has been extensively studied but is not yet fully understood, as exemplified by miR-375. In adult β-cell islets miR-375 levels are decreased when high levels of glucose are available (6). Low levels of miR-375 induce insulin secretion by de-repression of its targets Mtpn (6,16) and PDK1 (17,18), while overexpression of miR-375 attenuates proliferation and insulin gene transcription while reducing glucose-induced insulin secretion (16,18,19). Indeed, ectopic expression of miR-375 in diabetic pancreatic β-cells results in increased susceptibility to fatty acid induced apoptosis (16). Consistent with these studies, high levels of miR-375 are present in pancreatic islets of obese diabetic mouse models (ob/ob) (20) and T2D affected individuals (21). Additionally, when miR-375 is deleted in ob/ob mice, they develop a marked decrease in β-cell mass, which results in a severe insulin-deficient diabetes not found in ob/ob mice (20). Overall, it is becoming clear that miR-375 targets a suite of genes that negatively regulate cellular growth and proliferation (20), and that aberrant loss of this miRNA leads to dramatic reduction of β-cell mass, leading to low levels of insulin, hyperglycemia, and thus diabetes.

At least three additional miRNAs play a critical role in insulin exocytosis in β-cells. For example, miR-9 positively regulates glucose-induced insulin secretion in β-cells by directly repressing Onecut-2, the granuphilin (synaptotagmin-like 4) repressor, a protein that significantly enhances basal, but strongly inhibits K⁺-induced insulin secretion (22). Likewise, miR-96 also negatively regulates insulin exocytosis by targeting synaptotagmin-like 4 (23), and reduces levels of Noc2, which impairs β-cell ability to respond to secretagogues (23). The third well-described β-cell miRNA, miR-124a, intriguingly, appears to link β-cell and neural biology through the exocytosis pathway. miR-124a is upregulated in β-cells in the presence of glucose (24), but was first described as a brain-specific miRNA implicated in neurogenesis (25). High levels of miR-124a in β-cells lead to exaggerated insulin release at low-glucose concentrations but reduced glucose-induced secretion (24). This is likely to the result of targeting of Foxa2, which in turn downregulates Sur-1, Kir6.2, and Pdx1, the latter of which directly regulates the expression of the insulin gene (10). miR-124a also decreases levels of Rab27a and Noc2 and upregulates Snap25, Rab3a, and Synapsin1a, facilitating tight regulation of the insulin exocytosis pathway (23).

Interestingly, two other miRNAs, miR-7 (26) and miR-375 (12), are expressed both in brain and in β-cell islets. Both β-cells and neurons share similar secretion mechanisms and are responsive to signals in the bloodstream including glucose and insulin. Taking into account the fact that insulin stimulates glucose metabolism in the brain, mainly in the cerebral cortex (27), it is likely that miRNA expression in the brain is affected by diabetes, which could therefore have profound neurologic consequences.

Obesity, hyperlipidemia (elevated levels of blood lipids), and insulin resistance (reduced glucose uptake in response to the insulin signal) are strongly associated with the onset of T2D (28). Obesity itself is characterized by adipocyte dysfunction with abnormalities in adipokine secretion and in energy metabolism. Adipogenic abnormalities lead to an adipocyte phenotype distinct from the “ideal” which stores excess energy benignly in the triglyceride droplet. Among other downstream effects, this leads to an accumulation of fat in ectopic sites, such as liver, muscle, pancreas, and kidneys (29). It also leads to the secretion of chemoattractants that promote macrophage infiltration of adipose tissue, leading to inflammation and excessive release of free fatty acids into the bloodstream (29,30).

miRNAs in adipose tissue are strongly dysregulated in response to obesity-induced molecular changes and environmental signals. For instance, the expression of the miR-29 family is induced by hyperglycemia and hyperinsulinemia in adipose tissue (31), and miR-29a is highly upregulated in 3T3-L1 adipocytes in response to high glucose (32). Likewise, miR-320 increases insulin sensitivity of insulin resistant adipocytes (33) and miR-27b impairs human adipocyte differentiation (34). These miRNAs target conserved core cell-regulatory pathways that are affected both locally and systematically by obesity and by diabetes generally. Indeed, miR-320 targets p85, which plays a critical role in cell growth by increasing Akt phosphorylation and thus the level of Glut4 (33). Similarly, miR-29 targets p85α in HeLa cells, and is thus likely to affect Akt signaling in adipocytes (35). Additionally, miR-27b targets peroxisome proliferator–activated receptor γ (34), the nuclear receptor targeted by thiazolidinediones, an insulin-sensitizing agent used for treating T2D (36). Just as miRNAs are diagnostic of broad changes in cancer (37,38), they also reflect the degree of disturbance in diabetes.

As mentioned above, obesity triggers macrophage infiltration and cytokine release in adipose tissue. This is closely followed by changes in miRNA expression, which in turn affect lipid levels and adipogenesis. Indeed, many cytokines, such as tumor necrosis factor (TNF)-α, interfere with insulin signaling and inhibit adipogenesis (39,40) and several miRNAs that are induced during adipogenesis are downregulated in obesity (41). For example, inhibition of miR-143 has been shown to regulate adipocyte differentiation and results in reduced adipogenesis (42). Similarly, miR-103 and miR-107 have been shown to accelerate adipogenesis (43), and are predicted to target pathways that regulate cellular acetyl-CoA and lipid levels (44). Levels of miR-143, miR-103 and miR-107 are reduced after one-day treatment with TNF-α in adipocytes (43), suggesting that macrophage infiltration and cytokine release are a contributing factor to reduced adipogenesis in obesity. The connection between miRNA expression and cytokine exposure may eventually be leveraged into treatment options for morbidly obese diabetic patients.

Tissues adversely affected by diabetes may include cardiac and skeletal muscle, liver, kidney, and endothelium. Hyperglycemia and hyperlipedemia damage these tissues causing conditions such as fatty liver, stroke, kidney failure, neuropathy, and blindness. Not surprisingly, the expression of miRNAs is altered in these tissues in diabetic individuals.

A precondition for the onset of T2D is insulin resistance of skeletal muscle. Abnormally circulating fatty acids accumulate in skeletal muscle and disrupt important signaling pathways (29). Normally, insulin activates the transcription of sterol regulatory element–binding protein 1, which represses transcription of miR-1 and miR-133 by inhibiting muscle specific factor myocyte enhancer factor (MEF) 2C (45,46). The repression of miR-133a and miR-1 in response to insulin levels is impaired in T2D patients (45), which contributes to impaired muscle function (47). Indeed, a recent study has shown that the expression of miR-133a is strongly downregulated in this tissue in T2D-affected individuals (48).

The miRNAs miR-1 and miR-133 also play a primordial role in normal cardiac function and cardiogenesis (49,50). In the normal heart, cardiomyocytes exposed to high levels of glucose develop hypertrophy and, similar to skeletal muscle, have low levels of miR-133a (51). Indeed, glucose-induced cardiomyocyte hypertrophy is associated with increased levels of MEF2A and MEF2C (52). In diabetes, however, the relationship between MEF2A/C is impaired. miR-133 is in fact overexpressed in the rabbit diabetic heart, where it induces prolongation of the QT interval, a known phenotype of T2D (53). However, and surprisingly, both miR-1 and miR-133a are downregulated in the hearts of mice with insulin-deficiency induced by streptozotocin (STZ), and in cardiac-hypertrophy and heart failure (54). This strongly suggests that the relationship between miR-133 and miR-1 and cardiomyocte responses to insulin and glucose is complex. Indeed, a recent study has shown that these miRNAs are upregulated in cardiomyocytes after high-glucose exposure (55), which accelerates cardiomyocyte apoptosis (55), a key factor in the development of diabetic cardiomyopathy. These observations suggest that miR-1 and miR-133 are excellent candidates for further functional studies. Unraveling their seemingly contradictory behavior will undoubtedly shed light on the complicated biology underpinning muscle cell responses to physiological stimuli.

In the late stages of diabetes, liver and kidney function may be impaired, and this is reflected in several abnormal miRNA expression profiles. For example, inhibition of a miRNA crucial for liver function, miR-122, which is downregulated in the livers of STZ-induced diabetic mice (56), results in low plasma cholesterol levels, increased hepatic fatty acid oxidation, and decreased hepatic fatty acid and cholesterol synthesis rate (57). miR-21 also plays an important role in hepatocytes, where it is induced by nuclear factor-κB, leading to the downregulation of phosphatase and tensin homolog (PTEN), a protein that inhibits Akt activation (58). High-fat diets result in the upregulation of miR-21 in rats (58), and liver biopsies of obese human patients also show an increase in the expression of miR-21 and a decreased level of PTEN expression in comparison with normal controls (58). In the diabetic kidney, tranforming growth factor-β and miR-192 induce expression of miR-216a and miR-217 (59,60), which leads to activation of Akt through targeting of PTEN (60).

A tissue that has recently come to attention in diabetes is vascular endothelium, which changes in response to diabetic inflammatory signals. There is a strong negative correlation between miR-126 levels and the onset of diabetic vascular complications (61). However, this miRNA is significantly increased in patients suffering of coronary artery disease (62). Interestingly, delivery of this miRNA by apoptotic bodies protects against diet-induced atherosclerosis (62). This indicates that miR-126 can be used as both a biomarker for early detection of vascular complications of diabetes, and as a possible RNA-based therapeutic for diabetes-induced atherosclerosis.

There are many areas of research that remain unexplored in relation to miRNAs and diabetes. One is the interplay between the brain and insulin target tissues. It is known that the brain is fueled by glucose and, despite previous reports that this tissue is insulin insensitive, recent articles have shown insulin-induced glucose uptake in specific brain regions (27). Indeed, the similarities between β-cell and neuronal exocytosis systems are remarkable (see above). Due to the fact that the brain is responsible for behavioral control, it is likely that it reacts directly to the availability or depletion of glucose in the bloodstream by regulating caloric uptake. miRNAs have already been shown to carry out important regulatory functions in the brain, and therefore a study focusing on the miRNA changes in brain in response to hyperglycemia and hyperlipidemia is essential for a complete systemic understanding of diabetes.

It is now clear that the miRNA profile of insulin resistance tissues changes years before the onset and/or diagnosis of T2D. Indeed, a study by Zampetaki et al. (61) has shown that a plasma signature of five miRNAs (miR-15a, miR-29b, miR-126, miR-223, and miR-28–3p) can accurately differentiate patients with a high likelihood of developing diabetes from healthy controls. In a review, Regazzi (63) proposed that these miRNAs might not only be indicators of early disease onset but are also responsible for its progression and thus a good target for early intervention. One possibility is that these miRNAs actually act distant from their sites of biogenesis, and are transmitted from one tissue to another (64). The recent observation of miR-150 transmission through the bloodstream and uptake by target endothelial cells (65) makes this possibility likely. If diabetes-related miRNAs are trafficked in trans, it may be possible to interfere in the progression of the disease with minimally invasive RNA therapeutics.

Because of the high genetic predisposition of some individuals to develop T2D, identification of genetic variants that alter the levels of key miRNAs may become a clinically powerful tool, undoubtedly aided by the increasing affordability of personal genomic sequencing. We predict that the availability of this technology will provide clinicians with an invaluable tool to identify at-risk individuals, to prompt their patients to take preventive action, and, eventually, to prescribe tailored therapeutics (66). It will also provide a baseline for beginning to rigorously identify environmental factors that promote epigenetic variations, particularly for genetically susceptible individuals.

It is important to emphasize the need for integrating current and future miRNA studies into existing resources like the Beta Cell Biology Consortium (http://www.betacell.org/). Many recent genome-wide studies have provided a snapshot of the miRNA content of different tissues involved in this disease, including the developing pancreas (67) and subcutaneous adipocytes of T2D patients (68). However, it is difficult to integrate this information if it is not summarized in a central repository. As highlighted by Gallagher et al. (48), even though the community hoped to gain insight into miRNA mechanisms of action by studying them in isolated model systems, it is now clear that this information must be centralized and integrated with clinical data in order to obtain the insights necessary to conquer this illness.

This work was supported by the Australian National Health and Medical Research Council (Australia Fellowship 631668 to J.S.M.), the Australian Research Council (Australian Postgraduate Awards Fellowship to S.L.F.-V.), and the University of Queensland (University of Queensland Postdoctoral Fellowship to R.J.T.).

No potential conflicts of interest relevant to this article were reported.

S.L.F.-V. surveyed the literature and wrote and developed the manuscript. R.J.T. and J.S.M. wrote sections of the manuscript and provided editorial feedback on its structure and text.

The authors apologize to those colleagues whose work could not be cited because of space constraints. The authors thank John Prins (University of Queensland Centre for Diabetes and Endocrine Research) and Brad Marsh (Institute for Molecular Bioscience, University of Queensland) for insightful comments and criticisms.