Aldose reductase and cardiovascular diseases, creating human-like diabetic complications in an experimental model

Abstract

Rationale

Hyperglycemia and reduced insulin actions affect many biological processes. One theory is that aberrant metabolism of glucose via several pathways including the polyol pathway causes cellular toxicity. Aldose reductase (AR) is a multifunctional enzyme that reduces aldehydes. Under diabetic conditions AR converts glucose into sorbitol, which is then converted to fructose.

Objective

This article reviews the biology and pathobiology of AR actions.

Results

AR expression varies considerably among species. In humans and rats, the higher level of AR expression is associated with toxicity. Flux via AR is increased by ischemia and its inhibition during ischemia reperfusion reduces injury. However, similar pharmacologic effects are not observed in mice unless they express a human AR transgene. This is because mice have much lower levels of AR expression, probably insufficient to generate toxic byproducts. Human AR expression in LDL receptor knockout mice exacerbates vascular disease, but only under diabetic conditions. In contrast, a recent report suggests that genetic ablation of AR increased atherosclerosis and increased hydroxynonenal in arteries. It was hypothesized that AR knockout prevented reduction of toxic aldehydes.

Conclusion

Like many in vivo effects found in genetically manipulated animals, interpretation requires the reproduction of human-like physiology. For AR, this will require tissue specific expression of AR in sites and at levels that approximate those in humans.

Introduction

During the past decade, studies in genetically modified mice have been used to define processes that contribute to atherosclerosis. Alterations in macrophage and endothelial cell biology by genetic overexpression or deletion can greatly modify vascular lesion extent and also the composition of plaques; these latter effects are viewed as ways to uncover processes leading to plaque instability. Although many processes might alter inflammation and hence lesion extent and complexity, the observation that diabetes/hyperglycemia/insulin-deficiency alone does not always result in greater vascular disease (reviewed in ¹) suggests that basic biological differences between humans and mice may alter the vasculo-toxicity of diabetes.

Many genes are expressed in rodent tissues at levels that are disproportionate to those in humans, and these lead to changes in physiology and response to drugs that are totally different from those of humans. A case in point is the expression of PPAR transcription factors. Rodent livers express much higher levels of both PPARα and PPARγ than do humans ². For this reason, while activation of PPARα in rats led to marked proliferation of peroxisomes, a result that led to the naming of this molecule, similar drugs do not produce the same effects in humans. Similarly, treatment of mice with PPARγ agonists leads to steatosis ^{3, 4}, while the same drugs in humans are used to treat fatty liver because they cause more triglyceride distribution into the adipose tissue ³. For this reason, wild type animals do not faithfully model human biology.

In the case of diabetic complications, this might be most important. Efforts to replicate the pathology of diabetic nephropathy have been frustrating ⁵. Similarly, vascular disease, ischemia reperfusion injury that is a surrogate for myocardial infarction, and cardiomyopathies associated with diabetes using rodent models have utilized genetic modifications that allow similar end-points as found in humans; although many of the phenotypes are accelerated suggesting that the effects are either pharmacologic or represent extremes. Similarly, while genetic deletion can sometimes unmask signaling pathways or provide a guide for pharmacology, they also represent non-physiologic extremes or allow for compensatory changes that are often unexpected, e.g. loss of aquaporin 1 and integrinα1β1, genes associated with urinary concentration, leads to downregulation of aldose reductase (AR) ^{6, 7}.

Hyperglycemia in myocardial cells

Chronic hyperglycemia exerts its damaging effects on cardiovascular tissue by multiple mechanisms. Nonenzymatic glycation reactions of extracellular molecular species and advanced glycation end-products (AGEs), which result from spontaneous chemical rearrangements of glycation products ^8–10, are one mechanism by which hyperglycemia exerts its damaging effects. These derivatives can bind to pre-existing glycated protein ¹¹ or AGE ¹² cell surface receptors. The interactions of such ligands with cell surface membrane receptors can lead to generation of reactive oxygen species (ROS) ¹³.

Intracellular effects of chronic hyperglycemia are linked to the key observation that cardiovascular tissue is at least partly independent of insulin for uptake of glucose from the extracellular environment ^{14, 15}. However, in the presence of chronic hyperglycemia, there is an inadequate downregulation of the non-insulin dependent transporters ¹⁶ and the cells are consequently subjected to continuous influx of abnormally high amounts of glucose into their cytosol. Chronic elevation of cytosolic glucose levels and/or metabolic flux ¹⁷ is associated with generation of excess intracellular superoxide and other mediators of oxidative stress, an insult now generally acknowledged to play an important role in the pathogenesis of diabetic complications ^18–21.

Chronic elevation of glucose alters the biochemical homeostasis of cardiovascular cells by impacting a number of key biochemical pathways - including the polyol pathway, the cytoplasmic redox state, the protein kinase C (PKC) pathway, and the glucosamine biosynthesis pathway - and production of intracellular glycating species ^{8, 17, 20, 22, 23}. In some cell types such as cultured bovine aortic endothelial cells, chronic hyperglycemia causes enhanced glycolytic and mitochondrial oxidative metabolism ²⁴. In other cell types such as in rat cardiac tissue, chronic hyperglycemia inhibits glycolytic rates and alters substrate use of cardiovascular cells ²⁵. The primary emphasis in this review will be on AR, the first enzyme of the polyol pathway, which sits high in the biochemical cascade that follows the entry of excess glucose into the cytosol of cardiovascular cells.

Characteristics of AR activity

AR (E.C. 1.1.1.21; AKR1B1, ALD2) is a member of the aldo-keto reductase superfamily ²⁶ and has been extensively studied (27, 28). It is a monomeric, cytoplasmic ^{29, 30} enzyme of ~35,900 Daltons with a triose phosphate isomerase structural motif that contains ten peripheral alpha-helical segments surrounding an inner barrel of beta-pleated sheet segments ³¹. The enzyme preferentially and reversibly binds NADPH in an extended conformation and uses the hydride of the C4 of the nicotinamide ring of NADPH to reduce an aldehydic molecule to the corresponding alcohol, e.g., straight-chain aldehydic glucose to sorbitol. AR lacks structural carbohydrate and no catalytic or structural metal ion has been detected. As originally shown by Hers ³², AR reduces a variety of aldehydic substrates with varying affinities ^{33, 34}. However, its “natural” substrate remains elusive. Human AR efficiently utilizes 4-hydroxynonenal (4-HNE), 2-methylpentanal, glyceradehyde, retinoids, and methylglyoxal, with K_m values in the 8-50 µM range ^{35, 36}. Most studies allude to K_m for glucose in mM range. However, these studies often ignore key properties of glucose. Inagaki et al. ³⁷ and Grimshaw ³⁸ investigated the glucose anomer specificity of AR and calculated that AR acts on the aldehyde form of D-glucose with a K_m of 0.66 µM, i.e. it is a higher affinity substrate than many others.

The AR catalytic center contains a key cysteine residue, Cys 298, which when oxidized causes AR to exhibit altered properties and inhibitor sensitivity ³⁹. The enzyme was recently reported to be catalytically altered by S-nitrosothiols ⁴⁰, activated by nitric oxide under ischemic conditions in non-diabetic heart ⁴¹, and inhibited by exposure of diabetic rat brain, heart, and nerve to elevated nitric oxide levels ⁴². In human tissues AR has primarily a single, reduced enzyme form ⁴³. There is wide inter-individual variability of AR levels in a particular tissue, likely due to genetic allelic differences.

Polymorphisms of ALD2 and Risk of Diabetic Complication

The human AR gene (ALD2 or AKR1B1) is on locus q35 of human chromosome 7 ⁴⁴. ALD2 is approximately 18 kilobases (kb) and includes ten exons coding for 316 amino acids ⁴⁵. The ALD2 promoter has a TATA box (at -37), a CCAAT box (-104), and an androgen-like response element (-396 to -382) ⁴⁶. At about 1200 bp upstream of the transcription start site, there is a 132 basepair (bp) region containing three osmotic response elements: OreA, OreB and OreC ⁴⁷. ALD2 pseudogenes have also been described ⁴⁸.

Genetic polymorphisms associated with the human ALD2 gene have been linked to diabetic complications ⁴⁹. The first reported microsatellite polymorphism was an (AC)_n repeat region located ~ 2.1 kb upstream of the transcription start site ⁵⁰. Two single nucleotide polymorphisms (SNP) have been detected in the basal promoter region of the ALD2 gene, C(-106)T ⁵¹ and C(-12)G ⁵². In addition, a BamHI site consisting of an A to C substitution was reported at the 95^th nucleotide of intron 8 ⁵³. The (AC)_n and C(-106)T polymorphisms are closely linked ^{54, 55}, but their effects may be distinguishable in different patient populations. The “Z-2” (AC)_n microsatellite polymorphism, i.e., (AC)²³, has been associated with high expression levels of AR ⁵⁶ and with rapid progression or increased prevalence of diabetic retinopathy ^{50, 54}, diabetic nephropathy ^55–58, and, less strongly, with diabetic neuropathy ⁵⁸. In the last case, a relatively clear association was detected between microvascular complications and a decrease in the “protective,” low AR-expressing “Z+2” allele, i.e., (AC)₂₅ ⁴⁸. An equal ⁵⁵ or even stronger ⁴⁹ link with the C(-106)T promoter SNP has been found in several studies. In Chinese type 2 diabetic patients, genetic polymorphisms of AR independently predicted onset of cardiorenal complications ⁵⁹. In Japanese type 2 diabetic patients, genetic polymorphism of AR has been associated with diabetic macroangiopathy. Clearly, larger studies are warranted to establish the association between AR gene polymorphism and diabetic cardiovascular disease ⁶⁰.

Expression of AR in tissues

AR is widely distributed in many tissues ⁶¹. In humans, AR immunoreactivity is most concentrated in the inner medulla of kidney ^{17, 62}, consonant with its role in renal osmoregulation. AR is also abundant in human sciatic nerve, lens, testis and heart, and cornea, with lesser concentrations in liver, renal cortex, stomach, spleen, lung, small intestine, and colon ⁶². Unlike rats, mice have much lower levels of AR expression and activity ⁶³ than humans ^{61, 62}. A recent report suggests that in mice, AR protein expression is especially robust in vascular endothelial cells and within macrophages of atherosclerotic lesions ⁶⁴.

Physiological Functions of AR

Despite decades of experimental studies and the creation of genetically modified mice, the physiologic function of AR is poorly defined. Originally discovered in seminal vesicles ⁶⁵, the polyol pathway was hypothesized to be a biosynthetic route for producing fructose for the energy needs of spermatozoa. In the inner medulla of the kidney, AR synthesis of intracellular sorbitol may help protect cells from the locally high osmotic forces associated with antidiuresis ^{66, 67}. Perhaps for this reason, pharmacological suppression of AR activity upregulates other components of the renal osmolyte system to compensate for the loss of sorbitol ⁶⁸. Genetic deletion of AR in mice also supports a role for AR in normal kidney function and, as noted above, shows that at least in kidney AR expression is regulated by expression of other genes in the renal medulla.

AR activity alters metabolism of glucose and produces a number of downstream carbohydrates. These include sorbitol-6-phosphate ⁶⁹, sorbitol-3-phosphate ⁷⁰, and fructose-3-phosphate ⁷¹, formed in part via the polyol pathway activity; the physiologic roles of these molecules is unknown. AR has been proposed to function as a “fuel switch”, diverting excess glucose away from energy metabolism ¹⁷. It may metabolize steroids ⁷² and catecholamines ⁷³, and detoxify aldehydes ⁷⁴ or their glutathionylated derivatives ⁷⁵. Because the broad substrate specificity of AR overlaps with that of ubiquitous, structurally related enzymes like aldehyde reductase ⁷⁶ and aldehyde dehydrogenase ⁷⁷, it is difficult to define the role of AR simply on the basis of substrate preference.

AR is proposed to also function as an antioxidant defense enzyme ⁷⁸. Such actions need to be carefully analyzed and interpreted since many in vivo anti-oxidant defense pathways exert cytoprotective properties. Overexpression of AR in vivo has been linked to increased, rather than decreased, oxidative stress ^{79, 80}. In addition, in cultured endothelial cells exposed to hyperglycemia, as well as in diabetic and galactosemic peripheral nerve and retina in vivo, structurally distinct AR inhibitors (ARIs) suppress and reverse (and do not accentuate) markers of aldehydic and oxidative stress ^79–81. These results appear to contrast with total deletion of the AR gene.

Adding to the AR functional conundrum are data from AR “knockout” mice ⁸². Despite the fact that AR is absent from all tissues, these genetically-engineered mice have minimal phenotype; they have normal structural, biochemical, reproductive and physiological properties. Consistent with the known osmoregulatory function of AR, their only abnormality is mild polyuria compensated by mild polydipsia ⁸². Urine and blood divalent cation concentration was also slightly altered, but it remains unclear if the changes were secondary to mild chronic diuresis or if AR plays a role in maintaining systemic divalent cation levels ⁸³. Other processes that lead to generalized increases in oxidant stress are associated with aging ^{84, 85}, this is not found in these mice. In part, this might be because the relatively low AR expression in the mouse has allowed it to develop other defenses. Nerve conduction velocity, which is reduced in diabetic rats by the overexpression of AR, remains completely normal in the AR knockout mouse ⁸². Similarly, cardiac contractile function appears normal with AR genetic deletion, an observation consistent with the lack of change in heart function with AR pharmacological inhibition ^{25, 41, 63, 86, 87}. Nonetheless, in the presence of biological stress important actions of AR could emerge despite the lack of phenotype under non-stressed conditions.

Polyol pathway: prevailing hypothesis & mechanisms in diabetes

In order to understand the potential role of AR in mediating diabetic cardiovascular complications, an overview of prevailing hypotheses and proposed mechanisms are essential. The “Osmotic Hypothesis” is that high levels of glucose are metabolized through AR and sorbitol dehydrogenase (SDH) to generate high intracellular levels of polyhydroxylated sorbitol and fructose (polyols). The polyol pathway for aberrant glucose metabolism has been viewed as a potential cause of several diabetic complications (Figure 1). First described almost 50 years ago ⁶⁷, the pathway is comprised of two oxidoreductases, AR and SDH. In the presence of coenzyme NADPH, AR reduces glucose to sorbitol, while SDH then uses NAD⁺ to oxidize sorbitol to fructose. In rat and mice lens, rapid intracellular accumulation of polyols results in osmosis-driven water influx, swelling, imbalances in ion and metabolite homeostasis, and triggers formation of the “sugar cataract” ^{88, 89}. It is unlikely that the osmotic hypothesis can address the data on the role of AR in cardiovascular complications.

Figure 1. Flow diagram illustrating the interplay between polyol pathway, glycolysis, advanced glycation end-product (AGE) precursor generation, and diacylglycerol (activator of protein kinase C) generation.

It is important to note that arrows are drawn to show competition between SDH and GAPDH for cytosolic NAD+, thus impacting glycolysis.

A potential pathogenic role for increased metabolic flux through the polyol pathway independent of osmotic stress has been proposed ⁹⁰. There are a number of interactions of the polyol pathway and its coenzymes with other metabolic pathways. Cheng and Gonzalez ⁹¹ demonstrated that increased flux via the polyol pathway in rat lens increased turnover of NADPH and that AR and the antioxidant enzyme glutathione (GSH) reductase competed for the same pool of cytoplasmic NADPH. Williamson and colleagues demonstrated a strong linkage between polyol pathway flux and the ratio of free cytosolic NADH to NAD⁺, a factor critical to vascular function ^{92, 93}. Excess flux of glucose through AR impacts a variety of important metabolic pathways such as glycolysis, oxidative stress, intracellular nonenzymatic glycation and protein kinase C (PKC) activation^{8, 17, 20, 22, 23, 25}.

Diabetes-induced cardiovascular disease in the mouse

One of the burning questions in AR biology relates to its role in diabetic vascular disease. Because mice do not normally develop atherosclerosis, they require genetic manipulation to produce the hypercholesterolemia that is the essential ingredient in atherosclerosis. Although the plasma cholesterol levels are often at extremes that rarely occur in humans, this allows atherosclerosis within an acceptable timetable. However, extremely high cholesterol levels will “swamp out” many modifying effects that are seen when the hyperlipidemic causes of disease are less intense. An additional issue with the usual methods of studying atherosclerosis in mice is the use of juvenile animals. A recent report showed that older mice, more representative of middle-aged humans who develop atherosclerosis, develop more inflammation and disease when placed on an atherogenic diet ⁹⁴.

Another extreme condition is the method to create diabetes. The most common technique is the destruction of islet cells using streptozotocin. This creates severe insulin deficiency, but also can lead to marked hypercholesterolemia in genetically modified mice ⁹⁵. In contrast, diets lead to more mild phenotypes. Genetic modifications such as loss of leptin signaling lead to hyperglycemia and hyperlipidemia, but are associated with defects in lymphocytes that actually reduce atherosclerosis ⁹⁶.

There are a number of pathways of aberrant glucose metabolism that are postulated to lead to vascular toxicity. While deflection of glucose into any of these pathways could exacerbate toxicity, AR is an enzyme that is expressed at extremely low levels in the mouse (see ⁹⁷ and Figure 2). In an effort to determine if AR deficiency was responsible, in part, for the failure of many mouse models to accelerate atherogenesis with diabetes a transgenic mouse line was created in which human AR (hAR) was expressed via a histocompatibility gene promoter, leading to generalized AR overexpression ⁹⁸. This transgene led to enzyme levels that were not dissimilar to those in humans. The mice have no obvious phenotype. When crossed onto the atherogenic LDL receptor knockout background, the hAR transgene had no effect on atherosclerosis in non-diabetic mice ⁹⁷. However in streptozotocin-induced diabetic mice, the transgene accelerated disease. This was associated with evidence of alteration in the GSH anti-oxidant system. In contrast, expression of hAR in high fat diet-fed mice with mild insulin resistance without hyperglycemia had no effect on vascular lesions ⁹⁹. Thus, it appears that hyperglycemia is needed and might need to be sufficiently increased to provide substrate for AR. In contrast, in the presence of higher cholesterol levels, AR effects were “swamped” ¹⁰⁰, and AR inhibitors (ARIs) also did not alter lesion extent. Others have found increased AR expression in activated macrophages and suggested that this enzyme is inflammatory because AR inhibition led to reduced oxidative stress ^{101, 102}.

Figure 2. Comparison of AR activity in humans, and wild type and genetically modified rodents.

Cardiac aldose reductase (a) activity and (b) content in rats, wild type mice (WT), and human AR-expressing mice (ARTg).

In contrast, a recent study in apoE knockout mice found increased early lesion size in control and diabetic mice in the AR knockout mice and greater lesion progression with ARI treatment ⁶⁴. Lesion size was correlated with the presence of 4-HNE, which the authors postulated was due to defective removal of toxic phospholipid aldehydes. Surprisingly, the larger lesions were also associated with more collagen, a marker of greater lesion stability. The differences in the studies of AR transgenics and knockouts are reminiscent of those with alterations of eNOS in which knockout and overexpression are both associated with greater lesion formation, and lead to the following questions: Is knockout associated with compensatory regulation, including upregulation of other genes that alter vascular biology? Is genetic overexpression a pharmacologic effect or a reproduction of human pathophysiology? Furthermore, emerging studies on the impact of aldehyde dehydrogenase-2 (ALDH-2) in detoxifying 4-HNE ¹⁰³ adds further complexity to interpretation of lipotoxoic aldehyde levels in AR overexpressing and AR knockout mice. Clearly, comprehensive investigations on the interplay between AR and ALDH-2 in detoxifying 4-HNE and other lipotoxic aldehydes in hAR and AR knockout mice is critical to resolving these contrasting findings.

AR and vascular injury

AR is implicated in excess smooth muscle cell (SMC) growth; a model of vascular repair. AR inhibition prevents SMC growth in culture and in situ in balloon-injured carotid arteries ^104–109 Inhibiting the increased glucose flux via the AR pathway attenuated high-glucose–induced diacylglycerol accumulation and PKC activation in SMCs ¹⁰⁵, prevented high-glucose–induced stimulation of the extracellular signal–related kinase/mitogen-activated protein kinase and phosphatidylinositol 3-kinase ¹⁰⁶, activated nuclear factor-κB ¹⁰⁷, and decreased SMC chemotaxis, vascular inflammation, and adhesion. These data provide a rationale for evaluation of ARIs in diabetic patients undergoing angioplasty.

AR and the heart

Glucose flux via AR increases under ischemic conditions, even in the absence of diabetes ^{25, 41, 63, 86, 87}. The effects of the polyol pathway on ischemia/reperfusion (I/R) injury have been demonstrated in multiple studies using ex-vivo isolated perfused hearts and in in vivo transient occlusion and reperfusion of the left anterior descending coronary artery in mice and rats ^{41, 63, 110, 111}. In such studies, inhibition of AR reduced ischemic injury and was associated with attenuation of the rise in cytosolic redox (NADH/NAD+ ratio), improved glycolysis, increased ATP levels, and maintained normal sodium and calcium ion homeostasis in the heart post-I/R ^{41, 63, 110, 111}.

Transgenic mice expressing human-relevant levels of AR demonstrated more heart injury after I/R than wild-type mice ⁶³. Consistent with the premise that injury was increased directly via AR, inhibitors of AR in these transgenic mice reduced I/R injury ⁶³. Importantly, mice expressing human AR had several fold greater activity than wild type mice and the amounts of human AR expressed in these transgenic mice were similar to those seen in humans ⁶³.

Studies addressing mechanisms by which AR influences cardiac I/R injury have demonstrated key roles for this pathway in opening the mitochondrial permeability transition pore (MPTP) ¹¹¹. Compared to wild-type hearts, AR transgenic hearts had higher MPTP opening, increased generation of hydrogen peroxide, and reduced levels of antioxidant GSH ¹¹¹. Antioxidants or ARIs significantly reduced generation of ROS and inhibited MPTP opening in AR transgenic mitochondria after I/R ¹¹¹. Taken together, these studies implicate the AR pathway as a key player in mediating I/R injury in the heart.

In rabbit hearts inhibition of AR was protective ¹¹², although it has also been reported to abolish the cardioprotective effects of ischemic preconditioning ¹¹³. Though others ¹¹⁴ showed increases in AR activity during ischemia consistent with our earlier publication ⁴¹, they were unable to demonstrate cardioprotection with ARIs in a glucose perfused isolated rat heart I/R model. Reasons for these contrasting findings are not clear but may be due to model-dependent variations and substrate availability.

AR has been proposed to also detoxify aldehydes such as 4-HNE that accumulate during I/R. However, Chen et al. demonstrated that activation of ALDH2 reduces 4-HNE accumulation and protects hearts from ischemic damage ¹⁰³. We ⁶³ and Iwata et al ¹¹⁵ have demonstrated that AR overexpressing mouse hearts exhibit increased injury and poor functional recovery after myocardial I/R, and have changes associated with increased oxidative stress. Further, mice expressing human AR had greater injury and greater malonyldialdehyde (MDA) content than wild type mice with lower AR activity ⁶³. In contrast, AR null mice were reported to have reduced oxidative stress and protection against ischemic injury ¹¹⁶. In rat hearts subjected to I/R, increases in polyol pathway activity exacerbate oxidative damage ¹¹⁷. Furthermore, AR inhibition in animals does not cause increases in lipid peroxidation products such as MDA ^{79, 117–120}. Comprehensive measurements of 4-HNE and the role of ALDH2 will help resolve the role of AR as detoxifying enzyme in ischemic hearts.

Changes in AR expression have been demonstrated in diabetic and failing hearts. In Type 2 diabetic rat hearts, increased substrate flux via AR and SDH was observed in acute diabetics and in chronic diabetics both increases in expression of AR and SDH and greater flux via these enzymes was observed ¹²¹. In dogs, AR expression was attenuated in pacing induced heart failure ¹²². In humans, greater than 1.7 fold increases in AR expression were observed in patients with ischemic cardiomyopathy and diabetic cardiomyopathy ¹²³. These important data underscore the importance of addressing the role of AR using rodent models of heart failure.

Clinical applications of ARIs in humans

Experimental evidence in rodents supports the benefits of AR inhibition in the diabetic heart ^{102, 121}. Two classes of AR inhibitors have been extensively tested, the carboxylic acid class and the hydantoins, although novel classes of ARIs are in development ^124–126. In a study by Johnson and colleagues, diabetic subjects (with neuropathy) treated with zopolrestat for one year displayed increased left ventricular ejection fraction (LVEF), cardiac output, left ventricle stroke volume and exercise LVEF ¹²⁷. In contrast, placebo-treated subjects demonstrated decreased exercise cardiac output, stroke volume and end diastolic volume ¹²⁷. In another clinical study, ARI treatment was associated with improved autonomic variability in diabetic patients with autonomic neuropathy ¹²⁸. These relatively small but key studies in human subjects with established diabetic complications underscore the promising potential of inhibiting AR in the heart in long-term diabetes.

Inhibitors of AR have been extensively studied in animal models and in early human diabetic nephropathy. In animal models of diabetes, both classes of ARIs show protective effects; they reduced albuminuria, mesangial expansion and thickening of the glomerular basement membrane ^129–134. Genetic modulation of AR alters development of nephropathy . Galactose fed AR transgenic mice developed pathological changes in the kidney consistent with nephropathy ⁹⁸. In other studies, AR activity and TGF-β1 and type IV collagen mRNA levels were significantly increased in glomeruli from transgenic mice (vs wild type mice) exposed to AGE-BSA, in a manner suppressed by the ARI zopolrestat ¹³⁵. Mesangial cells cultured from AR transgenic mice exposed to AGE-BSA demonstrated greater increases in AR activity, and TGF-β1 and type IV collagen mRNA and protein versus wild-type cells. These increases were suppressed by either zopolrestat or AR antisense oligonucleotides¹³⁵.

Positive proof-of-concept studies in diabetic patients with nephropathy treated with ARIs have been reported. Administration of tolrestat to type 1 diabetic subjects for a six month period reduced urinary albumin excretion ¹³⁶. Administration of epalrestat to type 2 diabetic subjects for five years prevented increases in urinary albumin excretion rates ¹³⁷. In other studies, zopolrestat was administered to normotensive type 1 diabetic subjects for one year. After 3 months of treatment, zopolrestat administration caused a 34% reduction in urinary albumin excretion that was not correlated with changes in glycosylated hemoglobin or blood pressure; this effect persisted through 6, 9 and 12 month time periods ¹³⁸. Taken together, these data strongly suggestthat AR promotes diabetic cardiovascular and renal complications. A large randomized multicenter human trial using an ARI, that is relatively free from skin and liver side effects, will help establish its therapeutic potential in diabetic cardiovascular and renal complications.

Summary

AR is a central enzyme in the polyol pathway implicated in aberrant glucose metabolism and diabetic complications. We have proposed a working model (Fig 3) based on the comprehensive analysis of the data on AR in diabetic complications. While there are on-going studies of ARIs in humans, additional animal data are required to determine the role(s) of this enzyme under normal and pathological conditions. Both inhibitor studies and knockout mice attest to the relatively benign nature of AR loss. Indeed, the relatively low expression of this enzyme in control mice suggests that AR inhibition, unless total, is not likely to be toxic and is likely to be protective during hyperglycemia. Recent studies in AR knockout mice do suggest that caution is warranted. How best to examine AR’s physiologic actions in humans is open to debate. However, best efforts to reproduce AR human biology in experimental animals are needed to mimic the conditions representing human disease.

Figure 3. Overview of the proposed impact of increased AR pathway flux in diabetic cardiovascular tissue.

Non-standard Abbreviations and Acronyms

AR

aldose reductase

ARI

AR inhibitor

hAR

human AR

AGE

advanced glycation end-product

ROS

reactive oxygen species

SDH

sorbitol dehydrogenase

PKC

protein kinase C

I/R

ischemia reperfusion

SMC

smooth muscle cell

MPTP

mitochondria permeability transition pore

GSH

glutathione

4-HNE

4-hydroxy trans-2-nonenal (HNE)

MDA

malonaldialdehyde