exposed to high glucose, and they are also elevated in diabetic plasma (31). Dicarbonyls can react directly with protein to yield many of the same structures derived from the Amadori product, and they constitute an important source of intraand extracellular AGEs (32). For example, fructose-lysine can undergo metal-catalyzed oxidative cleavage giving rise to the irreversible “glycoxidation” product, CML which can also be formed from direct reaction of GO with lysine, independent of the presence of glucose. GO also reacts with arginine residues on protein to form carboxymethyl-arginine (CMA) (33), while MGO can give rise to the AGEs N(-(carboxyethyl)lysine (CEL) and arginine-hydroimidazolone (28, 32). MGO is also derived from spontaneous elimination of phosphate from triose phosphates, the concentrations of which are increased during hyperglycemia because of the increased flux of glucose through glycolysis.

Cells may have endogenous protection against intracellular reactivity of AGEforming dicarbonyls and several such “detoxifying” enzymes have been identified. For example, a glutathione-dependent glyoxalase complex (formed from glyoxalase I and glyoxalase II components) acts as an effective detoxification system for GO and MGO (34). The enzyme system catalyzes conversion of MGO to s-D-lactoylglutathione which is subsequently converted to D-lactate by glyoxalase II. Cells that overexpress this enzyme show less accumulation of MGO-derived AGEs (35).

Dyslipidemia is often overlooked as a pathogenic force in diabetic retinopathy (36). Lipid peroxidation reactions can also form a class of Maillard products called advanced lipoxidation endproducts (ALEs), and these are linked to diabetes and dyslipidemia (37) (Fig. 2). ALEs may represent an important source of protein modification especially in lipid-rich, highly oxidative environments, such as in the retina. Although understanding about the role of ALEs in diabetic retinopathy lags far behind that which is known about AGEs, these pathogenic adducts deserve more investigation.

PATHOGENIC ROLE OF AGES IN DIABETIC RETINOPATHY

AGEs and Clinical Correlation of Diabetic Retinopathy

The methods for AGE quantification in biological systems are based on analytical and/or immunocytochemical analysis that differ between researchers, and this has produced variable outcomes in a wide range of studies. With this proviso, patient-based studies have demonstrated that the levels of AGEs in serum correlate with the clinical progression of diabetic retinopathy (38). While many reported studies measured a range of ill-defined AGE moieties, others evaluated specific adducts such as CML, pentosidine, crossline, or hydroimidazolone (39–41) and found association with diabetic retinopathy. At the same time, some studies have reported no correlation between AGE levels and retinopathy in diabetic patients (39, 42), although the apparent disparity with other studies may be related to variations in patient populations, presence of nephropathy, and/or the nonuniform assays for plasma AGE quantification.

AGE-modified proteins in serum get readily cleared (except during renal dysfunction) and thus may not always provide robust biomarkers for disease. By contrast, modification of extracellular matrix proteins isolated from skin biopsies often provides more meaningful data (43). This is well illustrated by recent investigation of skin AGEs by the Diabetic Control and Complications Trial (DCCT) skin collagen ancillary study group

(44) which has added weight to the assertion that glycation and AGE modifications on long-lived proteins could be associated with progression of diabetic retinopathy. This study followed over 200 patients from the original DCCT for a further 10 years under the auspices of the Epidemiology of Diabetes Interventions and Complications (EDIC) trial (45). It revealed that levels of diabetic retinopathy were significantly less in the group initially maintained under “tight” glycemic control and that these benefits extended far beyond the period of intensive insulin therapy (45). The patients under “conventional” control for the first 10 years maintained a so-called hyperglycemic or metabolic memory and retained a strong association with retinopathic progression. Interestingly, the same memory phenomenon was also shown for retinopathy nearly 20 years previously by Engerman and Kern in variously controlled diabetic dogs (46). CML-modified skin collagen significantly predicted the progression of retinopathy (and nephropathy) even after initiation of intensive insulin therapy (45). Furthermore, the predictive effect of hemoglobin A1c (HbA1c) vanished after adjustment for furosine and CML, suggesting that accumulation of these adducts is an excellent marker for retinopathic progression and could offer a basis for the metabolic memory phenomenon (44).

AGE Accumulation in the Eye

AGEs have been extensively quantified in ocular tissues and shown to be elevated in diabetics when compared to nondiabetic controls. This includes cornea (47) and vitreous (48), where the levels of adducts may form an association with diabetic retinopathy (49). In the retina, AGEs and/or late Amadori products have been localized to vascular cells, neurons, and glia of diabetics (50–55). This would be expected to have pathogenic implications for the individual cells and retinal function. Although differential accumulation of AGEs exists in the diabetic retina over the course of life, diabetes significantly enhances the occurrence of these adducts in the vascular and neural tissue components (54).

Effect of AGEs on Retinal Cells

Demise of the retinal microvasculature remains a hallmark lesion of retinopathy in both diabetic animal models and patients (11). Retinal capillaries could be the principal targets for AGE-induced toxicity by several routes (Fig. 3). AGEs induce toxic effects on retinal pericytes by inducing apoptotic death linked to increased oxidative stress (56) and depleted superoxide dismutase (SOD) activity. Increases in such oxidative stress along with enhanced levels of ceramide and DAG further contribute to pericyte loss in the retinal capillaries (57). In addition, some studies have indicated that AGEs cause osteoblastic differentiation and calcification in retinal pericytes by the activation of alkaline phosphatases eventually leading to apoptosis (58). Pericytes exposed to AGE-modified basement membrane undergo MAPK-dependent apoptosis (59) and show acute attenuation in endothelin-1 (ETA receptor mediated)-induced contraction with subsequent downregulation of ETA receptor signaling suggesting that substrate-derived AGE cross-links could influence pericyte physiology (60). A more recent study shows that the intrinsic glyoxalase I detoxification system is critical for pericyte survival, and under high glucose conditions these cells may undergo rapid apoptosis possibly by the inactivation of glyoxalase by nitric oxide (NO) (61).