Aminoguanidine inhibits protein browning without extensive Amadori carbonyl blocking

23

Diabetes Research and Clinical Practice, 19 (1993) 23-30 0 1993 Elsevier Scientific Publishers Ireland Ltd. All rights reserved. 0168-8227/93/$06.00

DIABET 00704

Aminoguanidine

inhibits protein browning without extensive Amadori carbonyl blocking

Jestis R. Requenaa, Pablo Vidalb.* and JosC Cabezas-Cerrato” “Laboratorio Central, Hospital General de Galicia. Santiago, Spain, bProtein Chemistry Department, Hagedorn Research Laboratory, Gentofte. Denmark and CServicio de Endocrinologia y Nutricidn (Departamento de Medicina), Hospital General de Galicia. Clinic0 Universitario, Universidad de Santiago, Santiago, Spain

(Received 24 February 1992; revision accepted 8 September 1992)

Summary

It has been proposed that aminoguanidine reacts extensively with Amadori carbonyl groups of glycated proteins thus blocking them and inhibiting the further reactions which lead to browning and fluorescence development. We have glycated bovine serum albumin in the presence of 1, 5, 10 and 25 mM aminoguanidine and measured fluorescence development at 440 nm upon excitation at 370 nm, free (unblocked) Amadori groups as fructosamine with a calorimetric assay and furosine by HPLC, as an index of total Amadori products. Aminoguanidine significantly inhibited fluorescence development at all the tested concentrations (31%, 65%, 69% and 82% inhibitions, respectively) (P c 0.001). Blocking of Amadori groups was demonstrated by decreased fructosamine and unchanged furosine yields but only at the higher concentrations and to a very limited extent (13% and 27% blocking, respectively) (P < 0.01). Incubation of Aminoguanidine with albumin produced the appearance of 320 nm absorbing yellow chromophores, quite increased in the presence of glucose. These results suggest that Aminoguanidine is able to block Amadori groups, as previously hypothesized, but question the importance of this mechanism as an explanation of its capacity to inhibit browning. Scavenging of glucose seems to have no impact on glycation as seen by unchanged furosine yields.

Key words: Aminoguanidine;

Protein browning; Glycation; Fructosamine;

Introduction

Glycation of proteins involves a complex series of reactions including the initial attachment of Correspondence to: Prof. J. Cabezas-Cerrato, Servicio de Endocrinologia y Nutrition, Hospital General de Galicia, Galeras, s/n, Santiago, Spain. *Present address: Departamento de Endocrinologia, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain.

Furosine; Amadori products

glucose to protein amino groups by Schiff base formation, Amadori rearrangement to generate a ketoamine and further poorly characterized steps yielding brown, fluorescent compounds known as advanced glycation end-products (AGE) [ 1,2]. Many studies have established links between glycation, particularly AGE production and the appearance and progression of the late complications of diabetes mellitus (reviewed in Refs. 3-5).

24

Development of specific fluorescence with emission at 440 nm upon excitation at 370 nm has been used as a measure of AGE generation. Although some complications seem to be more related to the appearance of protein cross-links whose nature and fluorescence properties are still unknown, fluorescence development has been shown to be correlated to these complications [6]. A considerable effort has been devoted to finding ways to inhibit protein glycation and browning. Different approaches have been tried. Protective blocking of amino groups with agents such as acetylsalicylic acid has proved successful in vitro [7], although factors such as pH changes during incubation must be carefully considered [8]. Alternatively, antioxidant agents like diethylenetriaminepentaacetic acid (DETAPAC), glutathione and tocopherol have been also used with different results [9,10]. The mechanism of action of these compounds is not clear; some proposed browning reactions have been found to be oxidation-dependent [ 111. Alternatively, these products might inhibit the auto-oxidation of glucose to generate highly reactive dicarbonyls, a reaction which has been proposed as an alternative mechanism of conventional glycation [ 121. Brownlee et al. reported that aminoguanidine nucleophilic com(AG), a hydrazine-derived pound, inhibits the development of fluorescence and cross-linking of proteins in vitro and in vivo [13], making this compound a focus of active research as an agent able to inhibit the development of AGE and their consequences in diabetes mellitus. The mechanism of action that they hypothesized for AG consists in the blocking of the Amadori carbonyl groups of glycated proteins, which would them become unable to participate in further reactions generating AGE; however, no experimental proofs were presented to support this idea. More recently, two new proposals were made to explain the mechanism of action of AG: the direct reaction of AG with glucose, competing with protein for it and inhibiting glucose attachment [14,15] and the reaction of AG with active dicarbony1 intermediates of the browning reaction, such as 3-deoxyglucosone [16,17] (Fig. 1). We have aimed to investigate the existence of

GLYCATION

BROWNING

+

t-Q

Q-NH2

y--P INTERMEDIATE

CHO

PRaxlcTs

H-C-OH

+s

__j

dL”

’ i

AGE

H-C=0

AG

Lo NH

Cl-i=N-_~t+~ I

Ii-y

-

OH

2

I NH2

I

AG

.

6LC tfH-Q

SCAVENGED GLUCOSE

&i* H-&-OH Id-OH I CH2c+l 2-DEOXYGLLICOSONE

CH2 L=N-N+F

3

AG

ALU BLOCKED AMADORI

GROUP

LOW MOLECULAR WEIGHT YELLOW PRODUCTS ??

Fig. 1. Proposed mechanisms of action of AG. (1) competitive reaction with glucose (scavenging of the glycating agent); (2) Amadori carbonyl blocking; (3) trapping of dicarbonyl browning intermediates.

blocked Amadori carbonyl groups and their importance in the inhibition of glucose derived fluorescence development in protein by AG. Materials and Methods In vitro glycation

Bovine serum albumin (BSA, Sigma, St Louis, MO, USA) at a concentration of 40 mg/ml and 50 mM glucose (Sigma) were incubated in 0.1 M phosphate buffer, pH 7.4, at 37°C in a shaking bath for two weeks. 3 mM sodium azide (Merck, Darmstadt, Germany) was added as preservative. AG (Bicarbonate salt, Sigma), or acetylsalicylic acid (Merck) were added at different concentrations to some samples. All the complementary and control incubations were performed under the same buffer and temperature conditions. The pH of all samples was measured and corrected with dilute HCl or NaOH when necessary and did not change during the course of the experiments. Fluorescence measurements

Samples were dialysed overnight against phosphate buffer and diluted with this buffer to a concentration of 1.5 mg/ml. Fluorescence was

25

measured at 440 nm. upon excitation at 370 nm. Fluorescence was expressed as arbitrary units/mg albumin. Albumin was determined in undiluted samples using the Bromocresol Green method [ 181 adapted to a Hitachi 747 analyzer (Hitachi, Tokyo, Japan) following the manufacturer’s instructions. Fructosamine

Sephadex G-25 minicolumn (PD-10, Pharmacia, Bromma, Sweden). Fractions of 0.5 ml were collected and absorbance at 320 nm determined. Statistical

analysis

The Student’s t-test was used with data from fluorescence, fructosamine and furosine determinations.

measurement

Fructosamine was determined with a commercial kit (Fructosamine, Boehringer Mannheim, Mannheim, Germany) based in the reduction of nitroblue tetrazolium by Amadori ketoamine groups [20] and adapted to a Hitachi 747 analyzer. Results were expressed as mol fructosamine/mol albumin. Samples were dialysed overnight against 0.1 M phosphate buffer prior to analysis. Intraassay C.V. for fructosamine was 2% (n = 30). Furosine measurements

Furosine was determined by the method of Schleicher and Wieland [19] with minor modifications. Briefly, protein was precipitated with 40% trichloroacetic acid and after centrifugation, the pellet was washed twice with 8% acid. Hydrolysis was performed in 6 N HCl under argon atmosphere for 20 h at 100°C in a hot block. Finally, the hydrolysate was lyophilized and redissolved in distilled water. Samples were analyzed in a Waters chromatograph (Waters, Millipore, Milford, USA) using a reverse phase C-18 column (Techopack, HPLC Technology, Macclesfield, UK) and isocratic elution with 7 mM orthophosphoric acid at 1 ml/min. Furosine and tyrosine peaks were detected at 280 nm. The area under the furosine peak was normalized with respect to that of tyrosine in each sample (fur/tyr); results were expressed as percentage of fur/tyr with respect to that of samples of albumin incubated without glucose. Interassay C.V. was 6% (n = 15). Characterization

of yellow chromophores

Samples were diluted to l/4 and their scanning spectra were obtained with a HP8452 spectrophotometer (Hewlett Packard, Madrid, Spain). Samples containing 40 mg/ml BSA + 50 mM glucose or 40 mg/ml BSA + 50 mM glucose + 25 mM AG were diluted to l/4 and applied to a

Results Effect of AG on fluorescence

development

After 15 days samples of BSA incubated with glucose alone (0 mM AG) developed a fluorescence of 20.4 i 0.8 arbitrary units/mg albumin. Samples containing 25 mM AG, 10 mM AG, 5 mM AG and 1 mM AG developed, respectively, 10.8 f 0.6, 12.3 i 0.8, 12.8 f 0.8 and 16.8 f 1.0 arbitrary units/mg albumin. Samples containing BSA alone had 8.7 f 0.5 arbitrary units/mg albumin. The differences of all AG containing incubation mixtures are significant with respect to the 0 mM AG mixture (n = 4, P < 0 0.001). Table 1 represents the percentage of inhibition calculated as 100 - P, where P is the percentage of net fluorescence development of each incubation mixture with respect to the 0 mM AG mixture; net fluorescence development is calculated by subtraction of the fluorescence of BSA incubated without glucose. These percentages are 82%, 69%, 65% and 31%, respectively. Effect of AG on fructosamine

After 15 days, samples containing 0 mM AG had 2.26 f 0.08 mol fructosamine/mol albumin, whereas samples containing 25 mM AG, 10 mM AG, 5 mM AG and 1 mM AG had, respectively, 1.66 f 0.06, 1.97 f 0.09, 2.28 f 0.08 and 2.24 f 0.09 mol fructosamine/mol albumin. The differences were statistically significant for 20 mM and 10 mM AG (n = 4, P < 0.01). Samples of BSA incubated without glucose (0 mM AG) had 0.25 f 0.03 mol fructosamine/mol albumin. Incubation of BSA with 20 mM AG in absence of glucose had no effect on fructosamine. Percentages of Amadori group blocking were calculated as 100 - P, where P is the percentage of fructosamine with respect to that of 0 mM AG samples. Percen-

26 TABLE 1 Effect of AG on fluorescence development and fructosamine. a.u. stands for arbitrary units. Calculation of inhibition percentages is explained in Materials and Methods.

Oa

0 1 5 10 25

Fluorescence (a.u./mg albumin)

% inhibition

Fructosamine (mol/mol albumin)

% inhibition

8.7 20.4 16.8 12.8 12.3 10.8

31% 65% 69% 82%

0.25 2.26 2.24 2.28 1.97 1.66

-

f zt f i f zt

0.5 0.8 1.0 0.8 0.8 0.6

zt 0.03 zt 0.08 f 0.09 f 0.08 f 0.09 ztz 0.06

0% 0% 13% 27%

YSamples incubated without glucose.

tages of blocking were: 27%, 13%, 0% and O%, respectively for samples containing 25, 10, 5 and 1 mM AG (Table 1).

to inhibit glucose attachment, was used for the sake of comparison and produced a dose dependent decrease of fur/tyr.

Effect of AG on furosine

Generation of new chromophores

After 15 days, 5, 10 and 25 mM AG had no significant effect on the fur/tyr ratio with respect to 0 mM AG containing control samples (Fig. 2). A higher AG concentration (50 mM) was tried and a slight though significant 15% decrease was found (P < 0.001). Acetylsalicylic acid, which is known

An unexpected observation, made in the course of the experiments, was that samples containing BSA + glucose + AG quickly turned yellow while samples containing BSA + glucose progressed more slowly to the expected yellow-brown colour. Scanning spectra of both kinds of samples showed a constant increase of absorbance with two shoulders in the tailing of the 280 nm peak at about 320 and 370 nm (Fig. 3A). Samples containing BSA + AG also became yellow, but to a lesser extent than BSA + glucose + AG containing samples (data not shown). Scanning spectra of 50 mM lysine incubated with glucose plus 25 mM AG show the absorbance characteristics of the new chromophores without interference from protein: a round peak at 320 nm with a shoulder at 370 nm (Fig. 3B). Samples containing 25 mM AG and 25 mM AG + 50 mM lysine maintained baseline absorbances. Samples containing 50 mM lysine + 50 mM glucose and 50 mM glucose + 25 mM AG developed some absorbance, but with a maximum at around 290-300 nm. Gel filtration of mixtures containing 40 mg/ml BSA, 50 mM glucose and 25 mM AG produced two 320 nm absorbing peaks, one, the expected high molecular weight albumin peak eluting with

01

,

,

0

5

10

25

50

mM Fig. 2. Effect of AG and acetylsalicylic acid on furosine yield. Samples of BSA at 40 mg/rnl incubated for 2 weeks with 50 mM glucose and different concentrations of AG (0) or acetylsalicylic acid (0).

27

Discussion A

Fig. 3. Absorbance spectra. Samples were incubated for 2 weeks days in phosphate buffer (pH 7.4). Spectra were recorded after dilution to l/4. A: -, 40 mg/ml BSA+SO mM glucose + 25 mM AG, ... .... 40 m@rnl BSA + 50 mM glucose; --, 40 mg/ml BSA. B: -, 50 mM lysine + 50 mM glucose + 25 mM AG; x x x, 50 mM lysine + 50 mM glucose; ...... 50 mM glucose + 25 mM AG; - - -, 50 mM lysine + 25 mM AG; -.-.-, 25 mM AG.

the void volume and the other, a peak corresponding to low molecular weight products. Mixtures of 40 mg/ml BSA and 50 mM glucose only produced the protein peak (Fig. 4).

Fig. 4. Gel filtration. Samples were incubated for 2 weeks and diluted to l/4. A: 40 mg/ml BSA + 50 mM glucose + 25 mM AG. B: 40 mg/ml BSA + 50 mM glucose.

Aminoguanidine is a very efficient inhibitor of the development of fluorescent AGE when it is incubated with glucose and protein. As an hydrazine derivative with strong nucleophilic properties, it might be expected to block extensively Amadori carbonyl groups. Lewis et al. [ 151 incubated radioactive aminoguanidine with glycated and non-glycated proteins and found a non-specific binding to both. From their data, we can calculate that the specific binding of AG to glycated protein (the amount of AG bound to glycated protein after subtraction of the amount of AG bound to nonglycated protein) accounts for 0.043 mol of AG per mol of glycated protein. Since glycation represents 1.4 mol of attached carbohydrate per mol of protein and assuming that most of this carbohydrate represents Amadori product due to the short glycation periods used, it can be deduced that most Amadori groups are present in an unblocked form. We determined furosine and fructosamine in order to assess the presence of blocked Amadori groups in samples of glycated BSA whose fluorescence development had been inhibited by AG. Blocking of Amadori groups renders them unable to react with NBT (blocked carbonyl groups cannot enolyse) and, therefore, fructosamine values decrease. However, decreased fructosamine values could also be interpreted as decrease in the actual concentration of Amadori groups. The unchanged yields of furosine when up to 25 mM AG is added to incubation samples rules out this possibility: if the decreases of fructosamine were due to decreases in Amadori product concentration, furosine yield would also decrease, as it represents a measure of fructoselysine, the overwhelmingly most abundant Amadori compound in our experimental system [19]; this is exemplified by the action of acetylsalicylic acid, which competes with glucose for attachment sites and therefore causes a decrease in Amadori product concentration which leads to a decreased furosine yield (Fig. 2). In the other hand, blocking of Amadori products does not affect the yield of furosine, as the acid

28

hydrolysis step breaks any hydrazonic bond between AG and Amadori carbonyls. We found that AG caused decreases of fluorescence development of up to 65% without concomitant decrease of free Amadori group concentration (1 and 5 mM AG, Table I). At higher AG concentrations (10 and 25 mM), extended fluorescence development inhibition was paralleled by a decrease of free Amadori groups (Table 1). However, this blocking of Amadori groups was relatively small. These results suggest that mechanisms other than Amadori carbonyl blocking are responsible of a considerable part of the inhibitory capacity of AG towards browning. Our results are testimony against inhibition of glucose attachment through a competitive reaction with glucose. A 82% inhibition of fluorescence development was achieved without any decrease of Amadori product (the first stable product upon glycation) generation (Fig. 2). Brownlee et al. also found similar results [ 131. However, Khatami et al. reported a strong inhibition of glucose attachment to protein in the presence of AG [14]. Both authors used radioactive glucose to trace glucose incorporation to protein; but Khatami et al., unlike Brownlee et al., did no pre-purify radioactive glucose to eliminate the fast reacting components present in it [21] which might be very reactive towards AG, thus explaining the different results. The early steps of glycation have been shown to be very sensitive to acid-base catalysis; different amino groups have very different reactivities according to their pK and microenvironment [22]. In human albumin, lys 525 is the first group to be extensively glycated. Therefore, AG is probably unable to prevent the glycation of some of these very reactive sites [23]. Recently, Oimomi et al. introduced a third hypothesis concerning the mechanism of action of AG. They showed that AG inhibits in vitro the fluorescence development and cross-linking induced in protein by 3-deoxyglucosone (3-DG). 3DG is a dicarbonyl compound which can be generated by decomposition of Amadori products; it is a powerful browning agent and is thought to play a key role as an intermediate in glucose induced browning [24]. These authors proposed that AG traps 3-DG, adding this effect to other possi-

ble actions. Our results are compatible with this hypothesis since trapping of post-Amadori intermediates seems to be the only possible explanation of the observed capacity of AG to inhibit up to a 65% of fluorescence generation without any concomitant effect on Amadori groups. Other putative post-Amadori Browning intermediates have been described. Monnier et al. hypothesized the existence of a dicarbonyl carbohydrate adduct bound to protein which would be a precursor of the recently described AGE compound Pentosidine [25]. AG inhibits the formation of Pentosidine [26] and reaction of AG with this intermediate could explain this finding. Farmar et al. described another protein-bound browning intermediate, AFGP [27]. This compound was blocked and stabilized by sulfite, with concomitant development of a yellow chromophore with absorbance at 300 nm. These authors proposed that sulfite exerts its action through a nucleophilic attack to AFGP which is blocked and rendered unavailable for further Browning reactions. AG, a powerful nucleophile, could act in a similar way. We have also described the development of yellow chromophores with 320 nm absorbance in AG containing albumin + glucose samples. Yellowing also ocurred in absence of glucose, though to a lesser extent. Gel filtration showed that most of the 320 nm absorbing material consisted in low molecular weight products. Replacement of BSA by lysine also gave place to generation of yellow chromophores, but only in the presence of glucose. These data are suggestive of the generation of low molecular products from reaction of AG and some carbohydrate-derived compound. In the case of BSA + AG, there is a minor fraction of glycated lysines which could generate a small amount of these intermediates. However, direct reaction of AG with protein cannot be excluded as a cause of the observed yellowing. The occurrence of 3DG/AG adducts responsible for the 320 nm absorbance is an attractive hypothesis which needs further research. To conclude, we have demonstrated that, in vitro, AG is able to block Amadori groups only to a minor extent and that most of its inhibitory action on protein browning is exerted through mechanisms which do not involve reduction of

29

number or blocking of Amadori products. We have described the generation of 320 nm absorbing low molecular compounds which might represent adducts of AG with Browning dycarbonyl intermediates. In vivo, other circumstances such as low concentrations of AG, longer reaction periods, inhibition of lysyl oxidase and scavenging of lipid oxidation-derived aldehydes surely complicate the mechanism of action of AG [28]. In this respect, we have found that AG is a powerful inhibitor of malondialdehyde induced browning (data not shown). However, our findings might be of importance when considering the possible therapeutic use of AG. While this manuscript was being prepared, a report has appeared questioning the blocking of Amadori products by AG [29].

8

9

10

I1 12 I3

14

Acknowledgements We are grateful to Dr. B. Welinder for excellent advice concerning HPLC analyses and Dr. I. Alonso for allowing us the use of fluorimetry equipment. Thanks to Concha Rodriguez for graphic design.

IS

I6

17

References 18 Hodge, J.E. and Rist, C.E. (1953) Amadori Rearrangement under new conditions and its significance for nonenzymatic Browning. J. Am. Chem. Sot. 75, 316-322. Monnier, V.M. (1989) Toward a Maillard reaction theory of aging. In: J.W. Baynes and V.M. Monnier (Eds.), The Maillard Reaction in Aging, Diabetes and Nutrition. Alan L. Liss, New York, NY, pp. l-22. Brownlee, M., Vlassara, H. and Cerami, A. (1984) Nonenzymatic glycosylation and the pathogenesis of diabetic complications. Ann. Int. Med. 101, 527-537. Brownlee, M., Cerarni, A. and Vlassara, H. (1988) Advanced glycosylation end products and the biochemical basis of diabetic complications. N. Engl. J. Med. 318. 1315-1321. Cerami, A., Vlassara, H. and Brownlee, M. (1988) Role of advanced glycosylation end products in complications of diabetes. Diabetes Care I I (Suppl. l), 73-79. Monnier, V.M., Vishwanath, V., Frank, K.E., Elmets, C.A. and Kohn. R.R. (1986) Relation between complications of type I diabetes mellitus and collagen-linked fluorescence. N. Engl. J. Med. 314, 403-408. Swamy, MS. and Abraham, E. (1989) Inhibition of lens crystallin glycation and high molecular weight aggregate

19

20 21

22

23

24

25

formation by aspirin in vitro and in vivo. Invest. Ophthalmol. Vis. Sci. 30, 1120-l 126. Vidal, P., Requena, J.R. and Cabezas-Cerrato, J. (1989) Inhibition of protein glycation: a warning about the pH factor. Med. Sci. Res. 17, 21-22. Huby, R. and Harding, J.J. (1988) Non-enzymic glycosylation (glycation) of lens proteins by galactose and protection by aspirin and reduced glutathione. Exp. Eye Res. 47, 53-59. Ceriello, A., Giugliano, D., Quatraro, A., Del10 Russo, P. and Torella, R. (1988) A preliminary note on the inhibiting effect of a-tocopherol (vit E) on protein glycation. Diabete and Metabolisme 14, 40-42. Baynes, J.W. (1991) Role of oxidative stress in development of complications in diabetes. Diabetes 40, 405-412. Wolff, S.P. and Dean, R.T. (1987) Glucose autoxidation and protein modification. Biochem. J. 245, 243-250. Brownlee, M., Vlassara, H., Kooney, A., Ulrich, P. and Cerami, A. (1986) Aminoguanidine prevents diabetesinduced arterial wall protein cross-linking. Science 232, 1629-1632. Khatami, M., Suldan, Z., David, I.. Li, W. and Rockey, J.H. (1988) Inhibitory effects of pyridoxal phosphate, on nonenzymatic ascorbate and aminoguanidine glycosylation. Life Sci. 43, l725- 1731. Lewis, B. and Harding, J.J. (1990) The effects of aminoguanidine on the glycation (non-enzymic glycosylation) of lens proteins. Exp. Eye Res. 50. 463-467. Oimomi, M. and Igaki, N. (1989) Aminoguanidine inhibits 3-Deoxyglucosone during the advanced Maillard reaction. Diab. Res. Clin. Pratt. 6, 31 l-313. Igaki, N.. Sakai, M., Hata, H., Oimomi, M., Baba, S. and Kato, H. (1990) Effects of 3-deoxyglucosone on the Maillard reaction. Clin. Chem. 36. 631-634. Boehringer Mannheim (1987) Technical Sheet I 187-E 7701. Schleicher, E. and Wieland, O.H. (1981) Specific quantitation by HPLC of protein (lysine) bound glucose in human serum albumin and other glycosylated proteins. J. Clin. Chem. Clin. Biochem. 19, 81-87. Schleicher, E. and Vogt, B.W. (1990) Standardization of serum fructosamine assays. Clin. Chem. 36, 136-139. Triieb, B., Holenstein, C.G.. Fischer, R.W. and Winterhalter, K.H. (1980) Nonenzymatic glucosylation of proteins: a warning. J. Biol. Chem. 255, 6717-6720. Garlick, R.L. and Mazer. J.S. (1983) The principal site of nonenzymatic glycosylation of human serum albumin in vivo. J. Biol. Chem. 258. 6142-6146. Acharaya. A.S.. Cho, Y.J. and Manjula, B.N. (1988) Cross-linking of proteins by aldotriose: reaction of the carbonyl function of the keto amines generated in situ with amino groups. Biochemistry 27, 4522-4529. Kato, H., Shin, D.B. and Hayase. F. (1987) 3Dcoxyglucosone cross-links proteins under physiological conditions. Agric. Biol. Chem. 51. 2009-201 I. Grandhee, S.K. and Monnier. V.M. (1991) Mechanism of formation of the Maillard protein cross-link pentosidine. J. Biol. Chem. 266, 11649-11653.

30 26

27

Dyer, D.G., Blackledge, J.A., Thorpe, S.R. and Baynes, J.W. (1991) Formation of pentosidine during nonenzymatic browning of proteins by glucose. J. Biol. Chem. 266, 11654-I 1660. Farmar, J.G., Ulrich, P.C. and Cerami, A. (1988) Novel pyrroles from sulfite-inhibited Maillard reactions: insight into the mechanism of inhibition. J. Org. Chem. 53, 2346-2349.

28

29

Richard, S., Tamas, C., Sell, D.R. and Monnier, V.M. (1991) Tissue-specific effects of aldose reductase inhibition on fluorescence and cross-linking of extracellular matrix in chronic galactosemia. Diabetes 40, 1049-1056. Edelstein, D. and Brownlee, M. (1992) Mechanistic studies of advanced glycosylation end product Inhibition by aminoguanidine. Diabetes 41, 26-29.