Glycoxidation products (GOPs), such as N epsilon-(carboxymethyl)lysine (CML) and pentosidine, are formed during reaction of glucose with protein under oxidative conditions in vitro. It is uncertain whether these GOPs are derived from oxidation of Amadori adducts on protein or from oxidation of glucose or intermediates formed prior to the Amadori rearrangement. To address this question, we reacted collagen with 250 mM glucose in 200 mM phosphate buffer, pH 7.4, under antioxidative conditions, yielding a protein rich in Amadori adducts, but with only traces of GOPs. This "preglycated" collagen was then exposed to [13C6]glucose under oxidative conditions, producing both natural and [13C2]-CML. At 200 mM phosphate buffer, [13C2]-CML was the major product, even at low (5 mM) [13C6]glucose concentration, indicating a limited role for Amadori compounds in formation of CML in high phosphate. The relative yields of natural and [13C2]-CML varied with phosphate concentration, becoming similar at more physiological (10 mM) phosphate. We conclude that during glycation of proteins at high phosphate concentrations in vitro, GOPs are formed primarily by oxidation of free glucose or rapidly-formed intermediates preceding the Amadori rearrangement, such as carbinolamine or Schiff base adducts. In contrast, at lower phosphate and glucose concentrations in vivo, the Amadori adduct may be the more significant precursor of GOPs. The fact that glycoxidation reactions proceed by multiple routes must be considered in the development of therapeutic approaches for inhibiting the Maillard reaction in diabetes.

The amount of advanced glycation end-products (AGE) in tissue proteins increases in diabetes mellitus, and the concentration of a subclass of AGEs, known as glycoxidation products, also increases with chronological age in proteins. The rate of accumulation of glycoxidation products is accelerated in diabetes and age-adjusted concentrations of two glycoxidation products, N epsilon-(carboxymethyl)lysine (CML) and pentosidine, correlate with the severity of complication in diabetic patients. Although AGEs and glycoxidation products are implicated in the development of diabetic complications, these compounds are present at only trace concentrations in tissue proteins and account for only a fraction of the chemical modifications in AGE proteins prepared in vitro. The future of the AGE hypothesis depends on the chemical characterization of a significant fraction of the total AGEs in tissue proteins, a quantitative assessment of their effects on protein structure and function, and an assessment of their role as mediators of biological responses. In this manuscript we describe recent work leading to characterization of new AGEs and glycoxidation products. These compounds include: (1) the imidazolone adduct formed by reaction of 3-deoxyglucosone with arginine residues in protein; (2) N epsilon-(carboxyethyl)lysine, an analogue of CML formed on reaction of methylglyoxal with lysine; (3) glyoxal-lysine dimer; and (4) methyl-glyoxal-lysine dimer, which are imidazolium crosslinks formed by reaction of glyoxal or methylglyoxal with lysine residues in protein. The presence of 3-deoxyglucosone, methylglyoxal and glyoxal in vivo and the formation of the above AGEs in model carbonyl-amine reaction systems suggests that these AGEs are also formed in vivo and contribute to tissue damage resulting from the Maillard reaction.

The glycoxidation products N ⑀ -(carboxymethyl)lysine and pentosidine increase in skin collagen with age and at an accelerated rate in diabetes. Their age-adjusted concentrations in skin collagen are correlated with the severity of diabetic complications. To determine the relative roles of increased glycation and/or oxidation in the accelerated formation of glycoxidation products in diabetes, we measured levels of amino acid oxidation products, distinct from glycoxidative modifications of amino acids, as independent indicators of oxidative stress and damage to collagen in aging and diabetes. We show that ortho -tyrosine and methionine sulfoxide are formed in concert with N ⑀ -(carboxymethyl)lysine and pentosidine during glycoxidation of collagen in vitro, and that they also increase with age in human skin collagen. The ageadjusted levels of these oxidized amino acids in collagen was the same in diabetic and nondiabetic subjects, arguing that diabetes per se does not cause an increase in oxidative stress or damage to extracellular matrix proteins. These results provide evidence for an age-dependent increase in oxidative damage to collagen and support previous conclusions that the increase in glycoxidation products in skin collagen in diabetes can be explained by the increase in glycemia alone, without invoking a generalized, diabetes-dependent increase in oxidative stress. ( J. Clin. Invest. 1997. 100:839-846.)

To investigate the role of modified low-density lipoproteins (LDL) in the pathogenesis of diabetic retinopathy, we studied the cytotoxicity of normal and mildly modified human LDL to bovine retinal capillary endothelial cells and pericytes in vitro. Pooled LDL was incubated (in phosphate-buffered saline-EDTA, 3 days, 37 degrees C) under 1) nitrogen with additional chelating agents and 2) air, to prepare normal and minimally oxidized LDL, respectively. Similar conditions, but with the addition of 50 mM D-glucose, were used to prepare glycated and glycoxidized LDL. None of the LDL preparations was recognized by the macrophage scavenger receptor, confirming limited modification. Retinal capillary endothelial cells and pericytes were grown to confluence and then exposed for 2 or 3 days to serum-free medium (1% albumin) supplemented with normal or modified LDL (100 mg/l) or to serum-free medium alone. Cytotoxicity was assessed by cell counting (live and total cells) and by cell protein determination. Compared with normal LDL, modified LDL were cytotoxic to both cell types at both time points, causing highly significant decreases in live and total cell counts (P < 0.001) (analysis of variance). Reductions in cell protein also were significant for pericytes at day 3 (P = 0.016) and of borderline significance for endothelial cells at day 2 (P = 0.05) and day 3 (P = 0.063). Cytotoxicity increased as follows: normal < glycated < or = minimally oxidized < glycoxidized LDL. We conclude that, in diabetes, mild modification of LDL resulting from separate or combined processes of glycation and oxidation may contribute to chronic retinal capillary injury and thus to the development of diabetic retinopathy.

No abstract