Making sense of advanced glycation endproducts (AGE)
Kristian F. Hanssen
Department of Endocrinology, Aker Diabetes Research Center, Aker University Hospital, Oslo, Norway

It is now well established that high mean blood glucose is the most important factor behind the development and progression of microvascular complications in Type 1 diabetes [1, 2]. It is also likely that the microvascular complications seen in Type 2 diabetes are also due to chronic high blood glucose levels. Furthermore, there is increasing evidence that hyperglycaemia predicts cardiovascular complications in diabetes [3, 4]. However, the exact mechanism of the deleterious effect of hyperglycaemia on the small and large blood vessels is not known. There are at present some three possible contenders which alone or in some subtle combination may be responsible for the microvascular complications of diabetes:
- aldose reductase activation,
- protein kinase C,
- advanced glycation endproducts (AGE).
This article will concentrate on the latter, AGE. Recent reviews have focused on the first two possibilities [5, 6].

Advanced glycation endproducts formation
AGE-modified proteins are formed from the covalent reaction between endamino acids, such as lysine or arginine, and aldehyde groups of glucose molecules to create, first, Schiff base and then Amadori products of which the best known is HbA 1c (Fig. 1). Fructose lysine, instead of glucose lysine, is formed because of
Fig. 1: General scheme for the formation of AGEs. Equilibrium levels of the reversible Schiff base and Amadori products are reached within hours and days, respectively. AGEs form over a longer period of time but remain irreversibly bound to amino groups. R, amino acid or lipid backbone; AFGP, antifreeze glycoprotein; CML, carboxymethyl lysine

the shift in the position of the carbonyl group (from C1 to C2) during the Amadori rearrangement. AGE formation from fructose lysine involves the non-oxidative dissociation of fructose lysine to form a new reactive intermediate that again modifies protein to form AGE. Afterwards these protein glucose adducts undergo slow changes to form a group of AGE compounds of which only a handful have so far been characterized. Most of these compounds crosslink with AGE-modified collagen to form stiff collagen [7]. Alternatively, fructose lysine may decay, releasing its carbohydrate moiety either as glucose or as more reactive hexoses, such as 3-deoxyglucosone, which themselves may modify proteins. Alternatively, it may be involved directly in further advanced glycation reactions involving free radical oxidation. It has recently been found that glucose can probably auto-oxidize to form reactive carbonyl products (glyoxal) and methylglyoxal (reactive oxygen species) (Fig. 2) which may react with protein to form glycoxidation products [8-10].

Fig. 2: Reactive oxygen species (ROS) attack protein, glucose and fructose lysine (FL) lipids, yielding reactive carbonyl-con-taining species that mediate the formation of glycoxidation and lipoxidation products. In the process, more ROS are generated, establishing further vicious and interactive cycles of molecular damage. (Reproduced from [9] with permission.)

In this case, fructose lysine is bypassed altogether. Thus, carboxymethyl lysine (CML) moieties of proteins may be formed either from oxidative modification of fructose lysine or from reaction of glyoxal, the main dicarbonyl-containing autoxidation product of glucose, with protein.

Arguments for AGE in microvascular complications
What are the arguments for AGE's role in microvascular diabetic complications? There are three lines of evidence:
- association between the accumulation of AGE-modified proteins and the severity of microvascular complications in both diabetic animals and man [11-13];
- development of typical microvascular complications following injections of AGE-modified proteins in non-diabetic animals [14];
- inhibition of the development and progression of microvascular complications by aminogua-nidine [15, 16].
Early diabetic microangiopathy is characterized by vasodilatation, increased blood flow and increased capillary permeability. AGE-modified proteins may lead to all these changes: AGE-modified proteins can also impair the binding of heparan sulphate to the extracellular matrix, which results in loss of anionic sites and thus an increase in endothelial permeability.

Oxidative stress
What is the relationship between oxidative stress and AGE in the creation of micro- and macro-vascular complications? Oxidative metabolism is designed to extract energy by controlled oxidation of substrates and, at the same time, to prevent uncontrolled oxidative damage - to use fire for warmth, but not get burnt [10].
As many of the identified AGE products are glycoxidation products, the question of whether there is increased oxidative stress in uncomplicated diabetes is important. The level of oxidizable substrate (glucose, Amadori adducts, reactive carbonyl and dicarbonyl compounds and polyunsaturated fatty acids) is increased in blood and tissues in diabetes and the changes in ascorbate, atocopherol and glutathione concentration are consistent with decreased antioxidant protection. Thus, increased levels of oxidative stress may be present in diabetes, but may be both a cause and an effect of tissue damage [10].
What markers of glycoxidation are there in diabetes? CML modification of proteins is one of the major glycoxidation products formed in vitro by the reaction between glucose and protein [17]. Many studies have investigated the accumulation of different AGE products (CML, pentosidine, pyrraline and 'total AGE') in diabetic animals and man and correlated their results cross-sectionally with diabetic microvascular complications. Most studies, but not all, have found a correlation between the severity of microvascular complications and the quantitative accumulation of AGE mostly in skin [11-13]. However, there is a large variation in accumulation, and such an association does not necessarily imply causality. The relevance of skin measurements to diabetic complications is also uncertain.
Immunohistochemistry has demonstrated different AGE products in both glomerular and tubular cells in experimental and human diabetic nephropathy [18, 19]. Recently, AGE products have been shown to increase retinal vascular endothelial growth factor expression, a factor which is important in the development of proliferative retinopathy [20].

AGE receptors
The deleterious effects of AGE may be due to increased production or decreased removal of AGE. At least four different AGE receptors have been characterized, of which two are scavenger receptors [21]. One of the scavenger receptors is very similar, if not identical, to the one that internalizes acetyl LDL (and oxidized LDL). Immunoreactive AGE material has been demonstrated in atheromas both in patients with and without diabetes.
One AGE receptor, RAGE, which has a wide tissue distribution, when interacting with AGE, results in the generation of cellular redox stress manifested by the appearance of malondialdehyde- reactive isotopes and activation of the transcription factor NF - [22]. Can variation in AGE explain the difference in propensity to complica-tions? This is unknown, but theoretically genetic differences in the AGE receptor may explain it.

AGE and atherosclerosis
Lipids may also be modified by AGE. AGE-modified substances, especially CML, have been found in atheromas by immunohistochemistry and chemical analysis [19].
LDL is modified by glycation, but it is still not clear whether the moderate modification due to moderate hyperglycaemia which is found in most people with diabetes is sufficient to render it more susceptible to oxidation [23]. However, as mentioned above, AGE-modified proteins may be taken up by scavenger receptors. AGE-modified LDL has been identified in serum. Thus, AGE-modified proteins may play a role in the pathogenesis of atherosclerosis.

Serum measurements of AGE products in human diabetes
Most AGE research has been performed in vitro or in diabetic animals with or without diabetic complications. It is therefore important to study human diabetes.
A small number of investigators have addressed the following questions:
- Do measurements of serum AGE predict the progression of microvascular complications?
- Can measurements of glycoxidation products in diabetes elucidate whether they play a role in diabetes?
Measurement of serum AGE in human diabetes has been difficult to carry out and, as there is no recognized standard, different groups may not be measuring in exactly the same way. Most measurements are based on a polyclonal AGE antibody [24]. One group [25] measured haemoglobin AGE in Type 2 patients before and after initial insulin treatment and found that haemoglobin AGE decreased in parallel accord- ing to blood glucose levels. One serum AGE measurement has been shown to predict the progression of morphological changes (basal membrane thickness, mesangial fraction) in the kidney [26]. AGE, and recently serum CML, have been found to be elevated in Type 1 [27] and Type 2 diabetes [28], but few data are avail-able on the relationship between serum AGE/CML and diabetic complications. How-ever, an interesting observation is that although serum AGE measurements predicted morpho-logical progression in the kidney in Type 1 dia-betes, serum CML measurements did not (TJ Berg et al., 1998, unpublished data). As CML is a dominant epitope also in serum AGE, the non-CML part in serum might be relevant to diabetic complications.
Recently, imidazolone has been identified (which is a reaction product between 3-deoxy-glucosone and arginine residues in proteins). Specific immunoreactivity was detected in nodular lesions and expanded mesangial matrix of glomeruli as well as in atheromas [29]. It is a glycation, but not a glycoxidation, product. Thus, imidazolone, or a related compound, may be a very important AGE product for the devel-opment of diabetic complications.
Thus, it may be that in diabetes, glycation products are most important in microvascular complications, and glycoxidation products are most important in macrovascular complications. Recent research has produced much suggestive evidence of the role of AGE in diabetic complications. However, in my opinion it has not yet been proven beyond doubt that AGE (in itself a very heterogenous group of modified proteins) is the mediator of the deleterious effects of glucose in long-term diabetes.
This is the fundamental problem with diabetes: the disease itself changes a whole array of biochemical pathways and, given the long time span for developing complications, a causal relationship is extremely difficult to prove. It is therefore necessary to do longitudinal studies in animals and humans before the development of diabetic complications and follow the subjects through the complications, measuring AGE in serum and tissue.

Biochemical inhibition of AGE formation
Pharmacological interventions, of which the best known is aminoguanidine, can inhibit some microvascular complications in diabetic animals. Although we do not yet understand aminoguanidine's mechanism of inhibition, it seems mostly to influence the early stage (pre-Amadori) of glucation and can scavenge reactive carbonyl compound [30].
Ongoing trials with aminoguanidine in diabetic patients, unfortunately only in kidney failure (the Type 2 trial with kidney failure has been stopped due to side effects), may cast some light on the role of AGE in diabetes. New pharmacological interventions such as crosslink breakers [31] also look promising.

Role of protein kinase C
Recently, microvascular complications have been linked to the activation of protein kinase C by hyperglycaemia-induced increases in diacyl-glycerol (DAG). A specific inhibitor of PKCb (LY 333531) prevents haemodynamic changes in the retina and renal glomeruli and reduces albuminuria in diabetic rats. Administration of vitamin E, which decreases DAG levels possibly through the activation of DAG kinase, have similar effects. However, it remains to be clarified whether the action of vitamin E on DAG kinase can be due to its antioxidant effect or to non-specific modulation of lipids in the plasma membrane. However, hyperglycaemia may activate the DAG/protein kinase C-ß system via gly-cation or glycoxidation (Fig. 3). Thus, AGE and the DAG/protein kinase C system may interact to create microvascular complications.

Fig. 3: Outline of some potential changes that can be caused by hyperglycaemia-induced activation of the DAG-protein kinase C (PKC) pathway. Note that both AGE and oxidation may induce DAG formation. Some possible inhibitors of this process are ami-noguanidine, vitamin E (an antioxidant) and LY333531 (an inhibitor of PKC- ß isoform).TGF- ß, transforming growth factor- ß; PAI-1, plasminogen activator inhibitor type 1; ICAM, intracellular adhesion molecule;VEGF, vascular endothelial growth factor; ANP, atrial natriuretic peptide. (Reproduced from [6] with permission.)

Conclusion
There is accumulating evidence of a pivotal role for AGE-modified proteins in the development and progression of microvascular and macrovascular complications in diabetes. However, the details of this putative mechanism are still unknown.

Acknowledgements
This article was written during a leave of tenure at the Medical University of South Carolina, Charleston, SC, USA. I would like to thank Drs John Colwell, Tim Lyons and Alicia Jenkins for ideal and stimulating working conditions. The author's research was supported by the Norwegian Medical Research Council, Norwegian Diabetes Association, Novo Nordisk Foundation and the Aker Diabetes Research Fund.