The effects of the Maillard reaction on the physical properties and cell interactions of collagen

Pathologie Biologie 54 (2006) 387–395 http://france.elsevier.com/direct/PATBIO/

The effects of the Maillard reaction on the physical properties and cell interactions of collagen Effets de la réaction de Maillard sur les propriétés physiques et les interactions cellulaires du collagène N.C. Avery, A.J. Bailey* Collagen Research Group, University of Bristol, Langford, BS40 5DU Bristol, UK Available online 07 September 2006

Abstract The non-enzymic glycation of collagen occurs as its turnover decreases during maturation, with complex carbohydrates accumulating slowly and the end-products of these reactions being permanent. The nature of these advanced glycation end-reaction products (AGEs) can be categorised as: 1) cross-linking: intermolecular cross-linking may occur between two adjacent molecules and involve lysine to lysine or lysine to arginine residues. Several compounds have been characterised. They are believed to be located between the triple helical domains of adjacent molecules in the fibre resulting in major changes of the physical properties, primarily, fibre stiffness, thermal denaturation temperature and enzyme resistance, all of which increase slowly with age but the rate is accelerated in diabetes mellitus due to high glucose levels: 2) sidechain modifications: these changes alter the charge profile of the molecule affecting the interactions within the fibre and if they occur at specific sites can affect the cell–collagen interaction. Modification of arginine within the sites RGD and GFOGER recognised by the two specific integrins (α1β2 and α2β1) for collagen reduce cell interactions during turnover and for platelet interactions (α1β2). These changes can ultimately affect repair of, for example, vascular damage and dermal wound healing in diabetes mellitus. Both types of modification are deleterious to the optimal properties of collagen as a supporting framework structure and as a controlling factor in cell matrix interactions. Glycation during ageing and diabetes is therefore responsible for malfunctioning of the diverse collagenous tissues throughout the body. © 2006 Elsevier Masson SAS. All rights reserved. Résumé La glysosylation non enzymatique (glycation) du collagène se produit lorsque son renouvellement diminue au cours de la maturation, alors que des sucres complexes s’accumulent lentement et que les « produits terminaux » de leur réaction deviennent permanents. La nature de ces produits avancés de glycation (AGEs) peut être divisée en : 1) pontage : des pontages intramoléculaires peuvent se produire entre deux molécules adjacentes et impliquer des résidus de lysine–lysine ou lysine–arginine. Plusieurs composés ont été caractérisés mais n’ont pas été identifiés dans la fibre de collagène. Ils sont localisés entre les domaines en triple hélice des molécules voisines dans la fibre avec pour résultat des changements majeurs des propriétés physiques, principalement, rigidification de la fibre, dénaturation thermique et résistance aux enzymes, qui augmentent lentement au cours du vieillissement, mais dont la vitesse s’accélère lors du diabète à cause de taux de glucose élevé ; 2) modifications des chaînes latérales : ces changements modifient le profil de la charge de la molécule en affectant les interactions dans la fibre, et si elles surviennent en des sites spécifiques, peuvent affecter les interactions cellule–cellule. Des modifications de l’arginine dans les sites RGD et GFOGER reconnus par les deux intégrines spécifiques pour le collagène (α1β2 et α2β1), réduisent les interactions cellulaires au cours du renouvellement et les interactions plaquettaires (α1β2). Ces changements peuvent affecter la réparation et, par exemple, les lésions vasculaires et la cicatrisation du derme au cours du diabète. Les deux types de modifications sont délétères à cause des propriétés du collagène en tant que réseau structural et facteur de contrôle dans les interactions cellules–matrice. Au cours du vieillissement et du diabète, la glycation est responsable du mauvais fonctionnement des divers tissus collagéniques du corps. © 2006 Elsevier Masson SAS. All rights reserved.

* Corresponding

author. E-mail address: [email protected] (A.J. Bailey).

0369-8114/$ - see front matter © 2006 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.patbio.2006.07.005

388

N.C. Avery, A.J. Bailey / Pathologie Biologie 54 (2006) 387–395

Keywords: Collagen; Glycation cross-linking; Cell interactions; Integrins; Platelets Mots clés : Collagène ; Pontage par glycation ; Interactions cellulaires ; Intégrines ; Plaquettes

1. Introduction Collagen is the most abundant structural protein of vertebrae varying from bone through skin and tendon to thin basement membranes. The collagen family is synthesised by fibroblasts, osteoblasts and chondrocytes, and over the past few years molecular sequence studies have identified a number of genes resulting in 27 genetically distinct types of collagen molecule, which readily accounts for its remarkable diversity of tissue distribution and function from sponge to man. Originally thought to be solely a mechanical support, more recent studies have demonstrated its role as a bioactive surface, playing a role in stimulating and controlling growth and development. The characteristic feature of all collagens is the triple helical domain, which results from the repeating sequence (Gly-X-Y)n where X is frequently proline and Y hydroxyproline in the amino acid sequence of all three α-chains. These triple helical molecules can spontaneously aggregate to form extracellular assemblies to perform a particular function. The major types of aggregates are the fibrillar collagens, the non-fibrillar basement membrane collagens and the FACIT collagen which are often involved with the surface of fibrillar collagens thereby providing additional functions [1]. The five fibrillar collagens are types I, II, III, V and XI, the latter two being present as minor components in tissues. Recently two additional fibrillar collagens have been identified, types XXIV and XXVII. Phylogenetic analyses of different invertebrates indicates that the formation of fibrils of variable structure and function arose step by step during invertebrate evolution [2], a finding that rather contradicts the evolutionary molecular incest model previously proposed for fibrillar collagens [3]. It has now become clear that fibrils are generally formed from a mixture of major and minor fibrillar collagens, for example I, III and V are present in most fibrillar tissue and collagens II and IX in the fibres of articular cartilage. The relative composition and concentration of these various types in the fibre determines the ultimate structure and consequently the biomechanical properties primarily due to the specific intermolecular cross-links formed. The subclass of FACIT collagens are characterised by interruptions in the triple helix and they interact with existing fibrils to produce a further variation in properties. For example, type IX on the surface of type II fibres in cartilage may link type II fibres [4] whilst types XII and XIV are localised at the surface of the collagen type I and III fibres in skin and tendon and may control fibre size. An additional eight members of this FACIT group have been identified but not yet assigned a role. The major non-fibrous collagen of basement membrane is type IV, the molecules binding head to head to form an open network to support cells and act as a permeable membrane. Type I collagen is the major supporting structure in skin, tendon and bone, type II in articular cartilage and types I and

III in the vascular system. The fibres are strong and virtually inextensible and this is achieved by the formation of intermolecular cross-links between the molecules within the fibre. They are formed by the action of the enzyme lysyl oxidase (LO) which oxidises the single lysines in the amino and carboxy telopeptides. The lysine-aldehyde formed then reacts with a specific hydroxylysine in the triple helix due to the precise quarter-stagger and head to tail end-overlap alignment of the molecules in the fibre. As the tissue matures these divalent cross-links react spontaneously to form stable trivalent crosslinks between fibrils [5] thereby increasing the mechanical strength and denaturation temperature of the fibre (Fig. 1). Non-fibrous basement membrane type IV collagen forms similar aldehyde cross-links in the 7S region of the aggregated non-triple helical carboxy-terminal domain. In contrast the non-helical amino-terminal (NC1) does not form these lysylaldehyde cross-links although there is some controversy as to the presence of an alternative cross-link. Than et al. [6] proposed that the NC1 was stabilised by a covalent met-lys cross-link. Although this was initially disputed by Vanacorre et al. [7] these authors subsequently proposed a similar crosslink in the NC1 between met and hydroxylysine [8]. The formation of the trivalent mature cross-links increases in the fibre as the turnover of collagen decreases. The low turnover of the collagen varies between tissues, from about 100 years for type II collagen in articular cartilage, about 10 years for type I in skin and about 1–2 years for bone collagen. Exceptionally, periodontal ligament has a rapid turnover of about 1 day. However the biological half-life for collagen is generally long and it is therefore susceptible to interaction with metabolites, primarily glucose and other aldehydes in what is referred to as the Maillard reaction. In this short review we will focus on the effect of the Maillard reaction on collagens where its normal functions are well established, that is, the fibrillar and basement membrane collagens. 2. The Maillard reaction and collagen The Maillard reaction, originally the research field of food chemists, has over the last few decades been extensively investigated primarily because of its relevance to the complications due to ageing and diabetes [9–11]. Since collagen provides the functional properties of the most vulnerable tissues, such as renal basement membrane, the cardiovascular system and retinal capillaries, the effect of the Maillard reaction on fibrous and non-fibrous collagen has been extensively investigated; [12]. Glycated haemoglobin has been used as a marker for the management of diabetic patients, but other markers are needed to evaluate the risk of specific diabetic complications. Fluorescence of skin correlates with retinopathy and arterial stiffness in some individuals, but more specific chemical mar-

N.C. Avery, A.J. Bailey / Pathologie Biologie 54 (2006) 387–395

389

Fig. 1. The effect of the Maillard reaction on the physical properties of collagen fibres: (a) changes in the cross-link profile of the enzyme LO cross-linking of skin collagen fibres. DHLNL, reduced keto-imine in embryonic collagen: HLNL, reduced dehydrohydroxylysinonorleucine the major cross-link in immature animal skin; HHL, histidinohydroxy-lysinonorleucine, the major cross-link of mature skin collagen.; the glycation cross-links increase rapidly following maturation; (b) typical increase in tensile strength following glycation; (c) typical increase in thermal denaturation following glycation; (d) typical increase in hydrothermal isometric tensile strength, that is, the tension generated during heating at constant length, following glycation.

kers are required. Pentosidine relates to certain complications but other markers relate to different complications. Clearly more chemical investigations are crucial as this is an important approach to ascertain risk, even if it involves more than one or two markers. The initial reaction of glucose with collagen is with the εamino group of peptide bound lysine to form a Schiff base, glucosyl–lysine [13] which undergoes an Amadori rearrangement to fructose–lysine. Both these adducts are susceptible to oxidation to deoxyglucosone and fragmentation to smaller sugar aldehydes such as methylglyoxal, which are many times more reactive than glucose although only present in minute quantities. Studies by numerous groups have shown that it is not these initial glucose adducts that result in the dysfunction of the collagenous tissues but the variety of subsequent oxidation pro-

ducts, now referred to as advanced glycation end-products (AGEs) The AGEs may be inert products that do not affect the physical properties of collagen, for example carboxymethyl–lysine, but some AGEs form intermolecular crosslinks between the collagen molecules within the fibre. The effect of these cross-links on the fibre is to modify the physical properties leading to an increase in stiffness and breaking load, denaturation temperature, solubility and a decrease in susceptibility to degradative enzymes [14–16]. Resistance to enzymic degradation is due to two effects, blocking the enzyme sites such as lysine and arginine and the closer packing of the molecules in the fibre due to cross-linking. Similar physical changes occur during the normal ageing of collagen but are accelerated in diabetes. Such changes are also deliberately further accelerated in laboratory incubations with high glucose concentrations to ensure a result within a reason-

390

N.C. Avery, A.J. Bailey / Pathologie Biologie 54 (2006) 387–395

able time scale, but they may not be a strictly accurate representation of in vivo produced AGEs. Direct evidence for the importance of AGEs was neatly demonstrated by Vlassara et al. [17] by the administration of AGEs (a mixture formed by glycation of BSA) to normal rats. These rats subsequently revealed typical changes observed in diabetes, basement thickening, glomerula hypertrophy and an increase in mesangial volume in the absence of hyperglycaemia. These results suggest that the BSA–AGEs were not all inert and stable ‘end-products’ but a mixture containing some very reactive components, probably on the pathway from fructose–lysine to stable AGEs. It would be interesting to see what effect a descrete and stable AGE would produce. There are two different aspects of the glycation of collagen that need to be considered: ● intermolecular Cross-linking. Reaction with the amino acid side-chain and subsequent further reaction to form a crosslink with an adjacent collagen molecule resulting in a modification of the physical properties of the collagen; ● collagen–cell interactions. Reaction with specific amino acid side-chains to form stable AGEs, which lead to the dramatic modification of the interaction of collagen with cells. 2.1. Intermolecular glycation cross-linking The best characterised glycation cross-link is pentosidine [18], Fig. 2. However, the concentration of pentosidine in aged or diabetic tissue is extremely low, of the order of one pentosidine per several hundred molecules, and is therefore unlikely to be responsible for the stiffening of the collagen fibre. Other cross-links have therefore been sought, primarily based on simplified reactions in vitro, but they have not been demonstrated to be present in collagenous tissue in vivo. The cross-links may be formed between lysines from two adjacent molecules, for example, GOLD and MOLD derived from the initial reaction with glyoxal and methylglyoxal, respectively. Other cross-links are formed between lysine and arginine, for example, pentosidine, glucosepane and DOGDIG [19] (Fig. 2). Glucosepane forms under non-oxidative conditions and is present in much larger quantities, reported to be 20 times that of pentosidine, and therefore the single most important cross-link to date (Fig. 2). At this level glucosepane is likely to have a significant effect on the physical properties of the collagen fibre. The location of these cross-links in collagen must be between the triple helical regions but they need to be identified in a collagen peptide to confirm their role as an intermolecular cross-link affecting the physical properties of the fibre. There are probably some preferential reaction sites with specific lysines and arginines along the collagen molecule (as in the case of BSA [20] and glyoxal-modified arginines in ribonuclease [21] but at the present time there is little evidence of such specificity apart from a preliminary study suggesting lysine in the cyanogens bromide peptide CB7 (unpublished data).

Fig. 2. Structure of the two most probable intermolecular glycation cross-links formed from the initial Maillard reaction product fructose–lysine. Pentosidine formation involves oxidation and reaction with ribose and collagen bound arginine; glucosepane is formed non-oxidatively by reaction with collagen bound arginine.

The consistent detection of AGEs in vivo is difficult in view of the different products arising from pathways dependent on the conditions of pH, ROS, free radicals and metal-ions. Similarly one cannot determine the relative importance of an individual cross-link in effecting a change in the physical properties of the collagen fibre. 2.1.1. Ageing The formation of AGE cross-links in the collagen of the dermis, vascular and renal and ophthalmic tissues causes a reduction in optimal function and is therefore a major contribution to ageing. It is now clear that ageing due to intermolecular crosslinking takes place in two stages, the enzymic (LO) crosslinks stabilise the growing fibre and provide its optimal functional capacity. Following maturity these enzymic cross-links do not increase, but the low turnover of the collagen allows the accumulation of the products of the Maillard reaction of glucose, some of which ultimately form intermolecular crosslinks. These later non-enzymatic glycation cross-links further increase the mechanical properties of the fibres ultimately resulting in a rigid and brittle fibre that cannot fulfil its particular role in the collagenous tissue. Glycation of collagen is therefore a true ageing effect [22]. In this context it is worth noting that FL, the precursor to AGEs, may be deglycated by enzymes known as amadoriases [23]. Apparently there are significant differences in the level of these amadoriases in different tissues [24], consequently the concentration of FL may not be as dependent on the level of collagen turnover as currently assumed. It has been reported that there is an increase in LO crosslinks in the skin of subjects with diabetes mellitus. Other evidence is based on the increased level of DHLNL, (the borohydride reduced product of the keto-imine cross-link) [25]. This cross-link is not normally present to any extent in skin collagen, the major cross-link being dehydro-lysinonorleucine (HLNL) together with its mature product histidinohydroxylysinonorleucine (HHL). These apparent LO-induced increases in DHLNL revealed a correlation with the duration of the dia-

N.C. Avery, A.J. Bailey / Pathologie Biologie 54 (2006) 387–395

betes and the complications that included thickening and tightening of the skin [25]. This is surprising since the accumulation of glycation cross-links and the reported tightening of the skin generally indicates a low turnover of the skin collagen rather than an increase in the immature keto-imine cross-links which would be derived from newly synthesised collagen. Certainly an increase in the keto-imine (DHLNL) has been observed in fibrotic conditions such as pulmonary fibrosis, hypertrophic scars or Dupuytren’s contracture [26] which are analogous to the presence of the keto-imine during the rapid synthesis of embryonic skin collagen [27]. It has also been suggested that the AGEs can induce the Strecker degradation of the ε-amino group of lysine and that the resulting lysyl-aldehyde can form the typical LO Schiff base cross-link. The levels of lysyl-aldehyde in normal rat serum albumin was reported to be four to fivefold higher in diabetic rats [28]. These “lysine-aldehyde cross-links” would be formed between two regions of the triple helix of adjacent collagen molecules rather than via the telopeptide and triple helix s determined by LO. They could therefore occur anywhere along the molecule, unless specific lysines are involved. If this suggestion can be born out by further studies then these “enzyme-like” cross-links would make a significant contribution to the ageing of collagen. This is an interesting proposal, demonstrating that the ε-NH2 of lysine was converted to lysine-aldehyde by ribose, ascorbic acid and methylglyoxal, but only in the presence of copper and oxygen which might suggest a role for superoxide and H2O2. The mechanism may therefore be a simple Fenton type reaction rather than the Strecker reaction proposed as operating in vivo. No α-amino γ-semialdehyde derived cross-links have been detected in plasma proteins to support this proposal. Ultraviolet irradiation of AGEs from BSA and melanoidins can lead to bleaching of the brown colour, that is, they must serve as photosensitisers leading to formation of reactive oxygen species [29]. Thus, AGEs accumulating in the skin collagen can promote ageing by photoxidation [30]. 2.1.2. Cleavage of glycation cross-links Obviously the use of competitive agents for the initial Maillard reaction products would ultimately prevent AGE crosslinking. Various inhibitors have been investigated, antioxidants, including Green tea and carnosine [31], metal chelators and competitive compounds such as aminoguanidine [32] and more recently pyridoxamine [33] which also inhibits lipoxidation [34]. Pyridoxamine (Fig. 3) may inhibit post-Amadori stages, scavenge reactive carbonyls or inhibit the effects of ROS. Further mechanistic investigations are necessary in addition to the on-going clinical trials However, the problem is already on-going in normal ageing and in diabetics before presentation to the clinician. The cleavage of existing glycation cross-links is therefore an important approach to reducing the deleterious effects of glycation of collagen. Vasan et al. [35] employed phenyl-thiazolium compounds Fig. 4) to cleave dicarbonyl AGE cross-links but the mechanism of action is controversial. There is no direct evidence for the presence of

391

Fig. 3. Structures of the glycation inhibitors pyridoxamine, which competitively inhibits formation of the initial Maillard reaction product fructose lysine, and phenyl-thiazolium compounds which are generated to cleave the existing glycation cross-links, although the latter’s mechanism of action is currently in doubt.

Fig. 4. Glycation inhibition of the cell surface integrin binding to collagen. The major specific collagen binding sites along the triple helix are R–G–D and G– F–O–G–E–R, where R represents Arginine.

α-diketones, which would in any case be very reactive compounds and therefore of short half-life, rather than a stable AGE cross-links. Despite the controversy over the mechanism these compounds have been clearly shown to be effective in restoring the flexibility of large arteries in animals with experimentally induced diabetes [36]. The hydrolysis products of the thiazolium compounds are known to be scavengers of metalions and this may be their mode of action [37]. However, in the case of collagen the nature of the AGE cross-links should be identified and characterised as the first step in any attempt to devise techniques to cleave the existing glycation cross-links. 2.1.3. Additional in vivo glycation agents Glycated collagen has been reported to accelerate the oxidation of unsaturated lipids, leading to the formation of malondialdehyde, 4-hydroxynonenal, glyoxal and methylglyoxal. If glycation studies are extended to include the action of these aldehyde products then the end-products would be known as advanced lipid end-products (ALEs). The formation of some AGEs, such as carboxymethyllysine, could also be classed as ALEs [38]. 2.1.4. Methylglyoxal and glyoxal These reactive dialdehydes are formed from the oxidative fragmentation of fructose–lysine as discussed above, but can also be formed from triosephosphate, acetone from ketone

392

N.C. Avery, A.J. Bailey / Pathologie Biologie 54 (2006) 387–395

bodies, or degradation of unsaturated lipids, and circulate in the system where they are generally removed by the glyoxalase system. They are consequently present in minute quantities, however, their reactivity is several thousand times greater than glucose. The importance of the triosephosphate derived glyoxal appears to be significantly greater than the minor amount derived from fructose–lysine fragmentation. The glyoxals react selectively with the arginine side-chain to form hydroimidazolones. 2.1.5. Malondialdehyde The breakdown of lipids also produces malondialdehyde (MDA) which, if the lipid and collagen are in close contact, as in the vascular system, could account for the stiffening of the large arteries in ageing and diabetes [39]. In vitro incubation of collagen fibres with MDA leads to rapid stiffening of the collagen fibre indicating the formation of cross-links. The MDA does not act through the formation of a double Schiffs base as previously reported, but forms a potential cross-link as a dihydropyridine [40]. More recently we have reported the formation of a potential cross-link involving malondialdehyde and histidine [41] The new inhibitor pyridoxamine readily reacts with malondialdehyde preventing ALEs and the formation of lipofuscin-like fluorescence reactions of malondialdehyde [42]. We also recently reported that the incubation products of MDA with fibrous collagen produced a different series of glycation products to those of globular BSA [41]. We have now extended these observations with glucose by comparing glycation of bovine serum albumin and fibrous type I collagen under near physiological conditions and found a much faster reaction with BSA and differences in the glycation products formed [43]. We suggest that the many model reactions carried out with BSA do not necessary provide correct data for the glycation of insoluble fibres such as collagen. 2.1.5.1. AGEs and the diet. A heterogeneous group of AGEs may also be formed externally, for example, during the heat processing of foods and in tobacco smoke [44,45], and are therefore often ingested. The nature of these AGEs as bioactive derivatives has recently been reviewed [46]. Recently it has become apparent that dietary AGEs can contribute significantly to the body’s AGE pool. Foods high in lipid and protein content exhibit the highest AGE levels, for example fat and meat, which contain 10–20-fold the AGE content of carbohydrate foods such as bread and fruit. Cooked collagen, for example, in skin and sausage casings, would contribute to this AGE pool. The presently known AGEs in foodstuffs almost certainly represent only a fraction of the total AGEs present. Previous studies by Vlassara et al. [47] demonstrated that the administration of AGEs produced the same affects as seen in diabetes and these observations have been extended to AGEs in the normal diet. The gastrointestinal (GI) absorption of AGEs is thought to be low, but recent studies have demon-

strated significant levels of AGEs in plasma following an AGE rich meal, which suggests that they could play a role in a variety of disease states, or relate to inflammation markers. Indeed, animal studies suggest that diet-derived AGEs have an impact on tissues such as vascular wall and kidney, nephropathy and wound healing that could equal that due to glucose in diabetes mellitus [48,49]. Interestingly, diabetic subjects respond to low AGE diets by a considerable reduction in markers or vascular dysfunction and inflammation. Foods rich in AGEs and the ingestion of tobacco smoke lead to deposition of tissue AGEs before those forming during hyperglycaemia or renal failure. These high levels are therefore likely to overwhelm the natural defence system against AGE accumulation, such as the AGE receptor system. Food derived AGEs, like endogenous AGEs, display protein cross-linking activity. As discussed above the AGEs are clearly a mixture of active glycation products and stable AGEs. The effects of an AGE restricted diet were tested in chronically fatfed mice. After 6 months on the low AGE diet, but high fat regime, mice showed significantly lower fasting insulin compared to high AGE high fat fed controls. AGE restriction in these studies was found to be as effective as caloric restriction, (currently the only effective means of prolonging life-span), in extending median life-span, without restricting caloric intake. It would be interesting to know if human subjects react in the same way as these animals. Taking a positive approach the food industry have attempted to exploit the Maillard reaction to improve food taste and stability [50]. 2.1.6. Formation of toxic acrylamide in foods Over the past 3 years there has been a series of publications related to the Maillard reaction and acrylamide. The discovery of acrylamide in thermally processed potatoes initiated projects to inhibit the formation of such a toxic substance. Asparagine is the primary amino acid responsible for the formation of acrylamide, possibly through the decarboxylation of Schiff bases prior to the Amadori rearrangement [51,52]. Presumably proteins such as thermally processed collagen that contain asparagines could also be a source of acrylamide in foods. 2.2. Collagen–cell interactions Once thought to provide solely a mechanical function, the collagen fibre has now been shown to take part in the regulation of cell differentiation, cell migration, and stimulating and controlling development and growth. The factors involved are also important in wound repair, inflammation and homeostasis. Clearly a modification of specific side-chains on the molecule by glycation could lead to disasterous consequences. 2.2.1. Effect of formation of AGEs on the collagen side-chains amino acids Cells interact with collagen through cell surface receptors known as integrins which interact with specific amino acid sequences along the collagen molecule. Thus AGE side-chain

N.C. Avery, A.J. Bailey / Pathologie Biologie 54 (2006) 387–395

modification of lysines and arginines along the molecule alters their charge profile and at the same time modifies specific reactive sites thereby reducing cell–collagen interaction. It is now clear that there are several sites along the molecule that readily react with integrins, the major sites being RGD and GFOGER sequences. Integrins are heterodimeric cell surface receptors containing an alpha subunit non-covalently bound to a beta subunit. To date there are at least 18 known alpha subunits and eight beta subunits, which can assemble into 24 distinct integrins [53]. A subunit of the integrins that contains an inserted domain or I-domain within their α-subunit is the only class that interacts directly with collagen. Four α-subunits α1, α2, α10 and α11 associate with β1 and constitute the native collagen binding integrin family, the commonest being α1β1 and α2β1 [54]. The most widespread and well characterised integrin receptor for collagen I is α2β1 which is also a receptor for epithelial and endothelial cells, smooth muscle cells, fibroblasts, leukocytes and mast cells. Both α1β1 and α2β1 integrins recognise the GFOGER amino acid sequences, (where O is hydroxyproline), in collagen as a high affinity binding site in collagen types I and IV [55]. Human fibrillar type I collagen contains three triplet integrin binding motifs, which have GER as their second triplet, that is, GFOGER, GLOGER and GASGER have been identified as binding sites for α2β1. The loci of these sites has been mapped along the collagen fibre molecule by Farndale et al. [58]. There is a conserved spacing of these sequences within collagens I, II, III, V and XI. Type XXVII has one large triple helical domain which also contains three GXXGER sequences. Clearly there are differences between collagens in respect of the α2β1 integrin reactivity. There is also a differential α2β1 dependence of platelets and cell function interaction on collagen monomers compared with the aggregated fibrils. It is probable that these sites are no longer accessible in the closely packed molecules of the intact fibril, only the exposed surface residues being available to the integrin. It would be interesting to determine which sites remain exposed. On the other hand the internal residues may become exposed following degradation or remodelling of the fibres. Clearly the effect of glycation of the arginine residues at these sites on collagen would inhibit integrin interaction. For example, we have shown that the specific modification of arginine side-chains of type I collagen by glyoxal to form an hydroimidazolone reduces both the adhesion and spreading of MG63 and HT1080 cells [57]. These reactions are clearly of particular importance in basement membrane (type IV collagen) since these membranes support various type cell types and influence their differentiation during remodelling, and in the case of vascular membranes reduced interaction with endothelial cells would lead to vascular dysfunction. 2.2.2. Platelets Collagen-platelet interaction is central to haemostasis and appears to be a critical determinant of arterial thrombosis. Recent studies have shown that the glycoprotein VI, a dimer on the platelet surface, is another important signalling receptor for collagen [58]. α2β1 is the only collagen binding integrin in platelets, and is crucial for deposition on collagen exposed in

393

damaged arterial walls. This integrin binds collagen through its metal-ion dependent adhesion site with the α1 inserted domain. Knight et al. [55] identified the sequence (GFO)GER as the minimal recognition motif for α1I, α2I and α11I. The α2β1 mediated platelet adhesion was poor to GER type sequences apart from GFOGER. In platelets the integrin α2β1 acts as the primary adhesion receptor following vascular trauma. It has been suggested that it is activated following vascular trauma by a shear-dependent mechanoreceptor [56] from its normal resting state regulated by the divalent cation concentration in the blood. The integrin binding motifs GXXGER all contain the arginine residue; hence glycation of the Arg in these sequences, GFOGER being the major site, would inhibit binding of the α2β1 integrin, with crucial consequences for deposition of platelets on collagen exposed in damaged arterial walls. 2.2.3. AGE specific receptors Macrophages were first shown to possess receptors for AGEs believed to be capable of targeting and removing senescent AGE modified proteins by Vlassara et al. [59]. These receptors have now been identified on a number of cell types [60] and are referred to as RAGE. RAGE is up-regulated in all tissues susceptible to diabetic complications, in a diverse array of cell types from epithelial, endothelial, lymphocytes and smooth muscle cells. RAGE is a multi-ligand receptor and is a member of the immunoglobulin superfamily of cell surface molecules. RAGE stimulates cellular activation towards dysfunction and tissue destruction. The defective removal of these glycated proteins was believed to be important in ageing and diabetes. The increased permeability of the collagenous tissues, skin, intestine and kidney of diabetic rats is completely suppressed by treatment with soluble RAGE (sRAGE) [61]. The increased expression of RAGE in atherosclerotic plaques in diabetes correlated with the level of glycated haemoglobin. RAGE blockade with sRAGE reduced atherosclerosis and vascular inflammation and is up-regulated in diabetes [62]. However, the cloning and characterisation of RAGE demonstrated that the binding of AGEs to RAGE did not accelerate their clearance and degradation, the interaction actually inducing complex post-receptor signalling including activation of P21 ras, MAP kinases with NF-kB pathway. Thus, the role of RAGE as a clearance receptor for AGEs has had to be revised. It appears to be more important to the central aspects of inflammation response and oxidative stress. Although it was assumed that AGE receptor system would be a key defence against accumulation of AGEs, these more recent studies have revealed a very complex system. Indeed, it has been suggested that many of the deleterious effects caused by glycation are in fact mediated via the AGE receptors, such as RAGE, galaectin-3 and CD36, but their actual contribution has as yet to be defined [63]. Interaction with these RAGE receptors are thought to play a central role in the onset of vascular disease and diabetes through chemotaxis, generation of cellular oxidative stress

394

N.C. Avery, A.J. Bailey / Pathologie Biologie 54 (2006) 387–395

and associated increases in the levels of the vascular adhesion molecule (VCAM-1). The interplay of AGEs and cellular receptors is complex and elicits multiple effects which have been reviewed [64,65]. However, a clear understanding of AGE-receptor studies is hampered by the diverse array of AGE compound mixtures used in the various studies, the effect of a specific AGEs would be interesting. 3. Concluding remarks Stabilisation of the collagen fibre occurs through selective intermolecular attachment through lysine-aldehydes providing collagen with its optimal functional properties from immature divalent bonds to mature trivalent bonds up to maturation of the tissue. However, further increases in age are characterised by deleterious changes whereby the tissues lose their elasticity due to cross-linking via the Maillard reaction. Similar changes occur at twice the rate in diabetes mellitus. These AGE crosslinks accumulate since the turnover of the collagen has decreased dramatically in maturity. Several glycation crosslinks have been proposed but are all present in minute quantities except the recent identification of glucosepane, which could make a significant change to the biomechanics of the collagen fibre and basement membrane. A major task is to confirm its ability to form intermolecular cross-links in the collagen fibre at a significant level to account for the observed changes in biomechanical properties. A second important question is to resolve whether the formations of lysine-aldehyde residues are formed within the triple helical domain through the Strecker reaction, and, if so, what would be their relative contribution to the stiffening of the fibre. Cleavage of existing glycation cross-links is still an important approach and the development of new highly specific chemical agents is crucial. The role of the RAGE receptor as a key defence against accumulation of AGEs appears to be no longer viable, but the complex signalling pathways require elucidation and may yet prove valuable to inhibition of the Maillard reaction. Equally important questions are the relative importance of RGD and GFOGER-type sites along the collagen fibre rather than the molecule in solution following glycation of the arginine residue to an imidazolone in these sequences. Such information is crucial to our understanding of vascular damage and repair involving platelets. Recent studies have revealed the surprising fact that high AGEs accumulate in normoglycaemic diseases such as atherosclerosis, Alzheimer’s disease, rheumatoid arthritis, cancer and obesity and chronic inflammation. It has been suggested that the increased levels in normo-glycemic conditions result from oxidative stress. Further, the Amadori-product has been reported to directly activate cellular pro-inflammatory events in the absence of AGEs, thus, Amadori glycated HSA but not AGE-HAS, stimulated the pro-inflammatory marker protein PAI-1 in human peritoneal mesothelial cells [66]. The study of AGEs is now clearly opening new fields in biomedical research; it is no longer restricted to glycation pro-

ducts but is directly involved in stimulating a variety of cellular reactions. The relationship of these non-glycemic AGEs to col-

lagenous tissues remains to be elucidated. References [1] Ricard-Blum S, Ruggierro F. The collagen superfamily: from the extracellular matrix to the cell membrane. Path Biol 2005;53:430–42. [2] Aouaderia A, Cluzel C, Lethias C, Gouy M, Garrone R, Exposito JY. Invertebrate data predict an early emergence of vertebrate fibrillar collagen clades and an anti-incest model. J Biol Chem 2004;279:47711–9. [3] Boot-Handford RP, Tuckwell DS. Fibrillar Collagen: the key to vertebrate evolution? A tale of molecular incest. Bioessays 2003;35:142–51. [4] Miles CA, Knott L, Sumner IG, Bailey AJ. Differences between the thermal stabilities of the three triple helical domains of type IX collagen. J Mol Biol 1998;27:135–44. [5] Bailey AJ. Molecular mechanisms of aging in connective tissue. Mech Aging Develop 2001;122:735–55. [6] Than ME, Heinrich S, Huber R, Ries N, Mann K, Timple R, et al. The 1.9 Å crystal structure of the non-collagenous (NC1) domain of human placenta collagen IV shows stabilization via a novel type of covalent crosslink between lysine and methionine. Proc Natl Acad Sci US 2002; 99:6607–12. [7] Vanacorre RM, Shanmugasundararj, Friedman DB, Boudar O, Hudson BG. Sundaramoorthy. The α1:α2 network of collagen IV. Reinforced stabilization of the non-collagenous domain-1 by non-covalent forces and the absence of met-lys cross-link. J boil Chem 2004;279:44723–30. [8] Vanacorre RM, Friedman DB, Ham AJL, Sundaramoorthy M, Hudson BG. Identification of S-hydroxylysyl-methionine as the covalent crosslink of the non-collagenous (NC1) hexamer of the α1α1α2 collagen IV network: a role for the post-translational modification of lysine 211 to hydroxylysine 211 in hexamer assembly. J Biol Chem 2005;280(32): 29300–10. [9] Brownlee M. Biochemistry and molecular cell biology of diabetic complications. Nature 2001;414:813–20. [10] Ulrich P, Cerami A. Protein glycation, diabetes and aging. Recent Prog Horm Res 2001;56:1–21. [11] Ann N Y Acad Sci 2005:1043. [12] Paul RG, Bailey AJ. Glycation of collagen: the basis of its central role in the late complications of ageing and diabetes. Int J Biochem Cell Biol 1996;28:1297–310. [13] Robins SP, Bailey AJ. Age-related changes in collagen: the identification of reducible lysine-carbohydrate condensation products. Biochem Biophys Res Communs 1972;48:76–84. [14] Andreassen TT, Seyer-Hansen K, Bailey AJ. Thermal stability, mechanical properties and reducible cross-links of rat tail tendon in experimental diabetes. Biochim Biophys Acta 1981;677:313–7. [15] Monnier VM, Cerami A. Nonenzymatic browning in vivo, possible process for aging of long-lived proteins. Science 1981;211:491–3. [16] Kent MJC, Light ND, Bailey AJ. Evidence for glucose-mediated covalent cross-linking of collagen after glycosylation in vitro. Biochem J 1985;225:745–52. [17] Vlassara H, Striker LJ, Teichberg S, Fuh H, Li YM, Steffes M. Advanced glycation end-products induce sclerosis and albuminurea in normal rats. Proc Natl Acad Sci USA 1994;91:11704–8. [18] Sell DR, Monnier VM. Structure elucidation of a senescent cross-link from human extracellular matrix. Implications of pentoses in the aging process. J Biol Chem 1989;264:21597–602. [19] Biemel KM, Reihl O, Conrad J, Lederer MO. Formation pathways for lysine–arginine cross-links derived from hexoses and pentoses by the Maillard process. J Biol Chem 2001;276:23405–12. [20] Hinton DJ, Ames JM. Site Specificity of Glycation and Carboxymethylation of BSA by Fructose Amino Acids 2006;30:425–37. [21] Cotham WE, Metz TO, Ferguson PL, Brock JWC, Hinton DJS, Thorpe SR, et al. Proteomic analysis of arginine adducts in glyoxalmodified ribonuclease. Mol Cell Proteomics 2004;3:1145–53.

N.C. Avery, A.J. Bailey / Pathologie Biologie 54 (2006) 387–395 [22] Paul RG, Bailey AJ. Glycation of collagen. The basis of its central role in the late complication of ageing and diabetes. Int J Biochem Cell Biol 1996;28:1297–310. [23] Wu X, Monnier VM. Enzymic deglycation of proteins. Arch Biochem Biophys 2003;419:16–24. [24] Brown SM, Smith DM, Alt N, Thorpe SR, Baynes J. Tissue specific variation in glycation of proteins in diabetes: evidence for a functional role for amadoriase enzymes. Ann N Y Acad Sci USA 2005;1043:817–23. [25] Buckingham B, Reiser KM. Relationship between the extent of lysyl oxidase-dependent cross-links in skin collagen, nonenzymatic glycosylation and long-term complications of type I diabetes mellitus. J Clin Invest 1990;86 (1946–1054). [26] Bailey AJ. Collagen changes in Dupuytrens disease. In: Berger, et al., editors. Dupuytrens disease. Berlin: Springer-Verlag; 1994. p. 37–48. [27] Bailey AJ, Robins SP. Embryonic skin collagen. Replacement of the type of aldimine cross-link during early growth period. FEBS Lett 1972;21: 330–4. [28] Akagawa M, Sasaki T, Suyama K. Oxidative deamination of lysine residues in plasma proteins of diabetic rats. Novel mechanism via the Maillard reaction. Eur J Biochem 2002;269 (5451–5420). [29] Masaki H, Okano Y, Sakurai H. Generation of active oxygen species from AGEs during ultraviolet light irradiation and a possible mechanism for cell damage. Biochim Biophys Acta 1999;1428:45–56. [30] Agirov M. Photosensitizer activity of model melanoidins. J Agric Food Chem 2005;52:1210–4. [31] Monnier VM. Intervention against the Maillard reaction in vivo. Arch Biochem Biophys 2003;419:1–15. [32] Brownlee M, Vlassara H, Kooney A, Ulrich P, Cerami A. Aminoguanidine prevents diabetes-induced arterial wall protein cross-linking. Science 1986;232:1629–32. [33] Voziyan PA, Hudson BG. Pyridoxamine. The many virtues of a Maillard reaction inhibitor. Ann N Y Acad Sci USA 2005;1043:807–16. [34] Metz TD, Alderson NL, Thorpe SR, Baynes JW. Pyridoxamine, an inhibitor of advanced glycation and lipoxidation reactions: a novel therapy for treatment of diabetic complications. Arch Biochem Biophys 2003; 419:41–9. [35] Vasan S, Foiles P, Founds H. Therapeutic potential of breakers of advanced glycation end-products—protein cross-links. Arch Biochem Biophys 2003;419:89–96. [36] Wolfenbuttle BHR, Boulanger CM, Crjin FRL, Huijbets MSP, Poitevin P, Swennen GNM, et al. Breakers of advanced glycation end-products restore large artery properties in experimental diabetes. Proc Natl Acad Sci USA 1998;95:4630–40. [37] Price DL, Rhett M, Thorpe SF, Baynes JW. Aminoguanidine and pyridoxamine are weak metal chelators and are given in high dose, the hydrolysis products of AGE cross-link breakers are also potent chelators. Chelating activity of AGE inhibitors. J Biol Chem 2001;276:48967–72. [38] Baynes JW, Thorpe SR. Glycoxidation and lipoxidation in atherogenesis. Free Radic Biol Med 2000;28:1708–16. [39] Sims TJ, Rasmussen LM, Oxlund H, Bailey AJ. The role of glycation cross-linking in diabetic vascular stiffening. Diabetologia 1996;39:946– 51. [40] Slatter DA, Paul RG, Murray M, Bailey AJ. Reaction of lipid-derived malondialdehyde with collagen. J Biol Chem 1999;274:19661–9. [41] Slatter DA, Avery NC, Bailey AJ. Identification of a new lipid derived cross-link and unique histidine adduct from bovine serum albumin incubated with malondialdehyde. J Biol Chem 2004;279:61–9. [42] Kang Z, Li H, Li G, Yin D. Reaction of pyridoxamine with malondialdehyde; mechanism of inhibition of formation of advanced lipoxidation end-products. Amino Acids 2006;30:55–61. [43] Slatter DA, Avery NC, Bailey AJ. Differences in the glycation products from globular serum albumin and fibrous collagen under near physiological conditions, unpublished. [44] Cerami C, Founds H, Nicholl I, Mitsuhashi T, Giordano D, Vanpathea S, et al. Tobacco smoke is a source of toxic reactive glycation products. Proc Natl Acad Sci USA 1997;94:13915–20.

395

[45] Nicholl ID, Stitt AW, Moore JE, Archer DB, Bucala R. Increased levels of advanced glycation end-products in lenses and blood vessels of cigarette smokers. Mol Med 1998;4(9):594–601. [46] Henle T. Protein-bound advanced glycation end-products (AGEs) as bioactive derivatives in foods. Amino Acids 2005;29:313–22. [47] Vlassara H, Striker LJ, Teichberg G, Fuh H, Li YM, Steffes M. Advanced glycation end-products induce glomerular sclerosis and albuminuria in normal rats. Proc Natl Acad Sci USA 1994;91:11704–8. [48] Oliver CM, Melton LD, Stanley RA. Creating protein with novel functionality via the Maillard reaction: a review. Crit Res Food Sci Nutr 2006;46:337–50. [49] Vlassara H. Advanced glycation in health and disease. Ann N Y Acad Sci 2005;1043:452–60. [50] Urribari J, Peppa M, Cai WJ, Goldberg T, Lu M, He CJ, et al. Restriction of dietary glycotoxins reduces excessive advanced glycation products in renal failure patients. J Am Soc Nephrol 2003;14(3):728–31. [51] Mottram DS, Wedzicka BL, Dodson AT. Acrylamide is formed in the Maillard reaction. Nature 2002;419:448–9. [52] Zyzak D, Sanders RA, Stojanovic M, Tallmade DH, Eberhart BL, Ewald DK, et al. Acrylamide formation mechanisms in heated foods. J Agric Food Chem 2003;51:4782–7. [53] Humphries MJ. Integrin structure. Biochem Soc Trans 2000;28:311–39. [54] Tuckwell W, Humphries M. Integrin-collagen binding. Semin Cell Dev Biol 1996;7:649–57. [55] Knight CG, Morton LF, Peachey AR, Tuckwell DS, Farndale RW, Barnes JM. The collagen-binding A-domains of integrins α1β1 and α2β 1 recognize the same specific amino acid sequence GFOGER in native (triple helical) collagen. J Biol Chem 2000;275:35–40. [56] Siljander PRM, Hamaia S, Peachey AR, Slatter DA, Smethurst PA, Ouwehand WH, et al. Integrin activation state determines selectivity for novel recognition sites in fibrillar collagen. J Biol Chem 2004;279: 47736–72. [57] Paul RG, Bailey AJ. The effect of advanced glycation end-product formation upon cell-matrix interaction. Int J Biochem Cell Biol 1999;31: 653–60. [58] Farndale RW, Siljander PRM, Onley DJ, Sundaresan P, Knight CG, Barnes MJ. Collagen-platelet interactions: recognition and signalling. Biochem Soc Symp 2003;70:81–94. [59] Vlassara H, Brownlee M, Cerami A. High affinity receptor mediated uptake and degradation of glucose modified protein. A potential mechanism for the removal of senescent macromolecules. Proc Natl Acad Sci USA 1985;82:5588–92. [60] Schmidt AM, Vianna M, Gerlach M, Brett J, Ryan J, Kao J, et al. Isolation and characterisation of 2 binding-proteins for advanced glycosylation end-products from bovine lung which are present on the endothelial-cell surface. J Biol Chem 1992;267(2):14987–97. [61] Wautier JL, Zoukourian C, Chappey O, Wautier MP, Guillausseau PJ, Cao R, et al. Receptor-mediated endothelial cell dysfunction in diabetic vasculopathy-soluble receptor for advanced glycation end-products blocks hyperpermeability in diabetic rats. J Clin Invest 1996;97(1):238– 43. [62] Park L, Raman KG, Lee KJ, Lu Y, Ferran LJ, Chow WS, et al. Suppression of accelerated diabetic atherosclerosis by soluble receptor for advanced glycation end-products. Nat Med 1998;4(9):1025–31. [63] Pugliese G, Pricci F, Iacobini C, Leto G, Amadio L, Barsotti P, et al. Accelerated diabetic glomerulopathy in galectin-3/AGE receptor 3 knockout mice. FASEB J 2001;15(13):2471–9. [64] Bierhaus A, Humpert PM, Marcos M, Wenet T, Chavakis T, Arnold B, et al. Understanding RAGE, the receptor for advanced glycation endproducts. J Mol Med 2005;83:876–86. [65] Hudson BG, Schmidt AM. RAGE: a novel target for drug intervention in diabetic vascular disease. Pharm Res 2004;21:1079–86. [66] Mandle-Weber S, Haslinger B, Schalkwijk CG, Sitter T. Early glycation albumin, but not advanced glycated albumin, methylglyoxal, or deoxyglucosone increases expression of PAI-1 in human peritoneal mesothelial cells. Perit Dial 2000;21:487.