---

An amino acid component, NFC-1, when formed in vitro by the reaction of ribose and protein was shown to comprise a complex mixture of high and low molecular AGE compounds. Two low-molecular-weight components have been successfully isolated and their structure determined. These were aNFC-1 [N^d-(4-oxo-5-dihydroimidazol-2-yl)--ornithine] and bNFC-1 a 4-imidazolon-2-yl derivative existing in three tautomeric forms. These imidazolone compounds have been shown to originate from the reaction of arginine with glyoxal and methylglyoxal, respectively. A third ninhydrin-positive AGE, gNFC-1, was shown to be composed of a number of chromatographically similar compounds which have not yet been characterized.

---

Abbreviations used: AGE, advanced glycation end product; PCEC, preparative cation-exchange chromatography; MALDI-MS, matrix assisted laser desorption ionization mass spectrometry; ESMS-CAD-MA, electrospray mass spectrometry-collisional activated dissociation-mass spectrometry; TFA, trifluoroacetic acid; HFBA, heptafluorobutyric acid.

¹ To whom correspondence should be addressed.

---

INTRODUCTION  

Carbonyl containing compounds such as glucose can react with the free amino groups on proteins via Maillard type reactions to form a variety of advanced glycation end (AGE) products. The formation of these compounds can have a profound effect upon protein integrity and function, the extent of which is most pronounced in long-lived proteins such as collagen [1]. This is of particular significance in diseases such as diabetes in which sugar levels are elevated [2,3].

AGEs can affect the interaction of proteins with cells, e.g. by altering the charge profile of the protein molecule and more importantly can modify the physiochemical properties of the protein by the formation of intermolecular cross-links. In the case of fibrous collagen this additional cross-linking has the effect of rendering the protein less soluble, more resistant to enzyme digestion and less flexible. As collagen constitutes the major structural protein in the body these changes have marked affects on vital tissue and organ function.

Isolation of the major AGE compounds formed in vivo has been hampered by their acid lability and the complex nature of protein hydrolysates. As a result, only a few have been isolated successfully from glycated proteins. Cross-link structures isolated and fully characterized to date are pentosidine [4], imidazolysine [5] and vesperlysines A, B and C [6]. Model structures formed by the reaction of glucose and blocked lysine have also been proposed as potential cross-links, i.e. Crosslines A and B [7]. Attempts to isolate the latter from in vivo glycated tissue have proved unsuccessful [6], however, some immunohistochemical evidence has been presented indicating their presence in situ [8,9].

We recently identified a non-fluorescent compound NFC-1 following in vitro incubation of collagen with glucose, and from collagenous tissues in vivo [10]. Although the component has not been characterized, the yield of this putative cross-link was consistent with the modification of the physical properties. The only other characterized AGEs from glycated tissues are carboxymethyllysine [11], pyrraline [12], lactatolysine [13], carboxyethyllysine [14] and a number of imidazolones [15–18]. Whilst none of these are cross-links, pyrraline can theoretically, undergo modifications to form ether or methylene bridges [19] although no evidence for the formation of such cross-links in vivo has been reported.

In this paper we describe the partial characterization of NFC-1 when formed by the reaction of ribose with collagen.

MATERIALS AND METHODS  

In vitro incubation of rat tail tendons

Tail tendons from 3- to 4-month-old rats were dissected free and washed extensively in 25 mM PBS, pH 7.4. Tendons were subsequently blot dried, weighed and incubated in PBS containing 250 mM ribose (Sigma, Poole, U.K.). Toluene/chloroform (1:1) was added as a bacteriostat (10 µl/10 ml) and the mixture was incubated at 37 °C for 15 days. At the end of the incubation period the tendons were washed in 0.9% saline to remove the excess ribose and stored at -20 °C until required.

Hydroxyproline assay

Tendons (53 g, wet weight) were weighed and then hydrolysed in 6 N HCl (1.5 l) for 24 h under reflux, freeze-dried, redissolved in water and hydroxyproline was determined colorimetrically using a Continuous Flow Autoanalyser (ChemLab, Cambridge, U.K.). The crude hydrolysate was then fractionated by a number of chromatographic techniques detailed below and summarized in Scheme 1.

---

Scheme 1

Overview of the chromatographic techniques used in the purification of ribose-derived NFC-1.

---

Preparative cation-exchange chromatography (PCEC)

The freeze-dried hydrolysate was redissolved in water (200.0 ml) the pH adjusted to 2.0 with concentrated HCl and the NFC-1 concentration was determined by amino acid analysis. Aliquots (15.0 ml) were then loaded on to a 1.2×60 cm column containing Duolite 225 cation-exchange resin. The column was standardized by analysis of pure amino acids using pyridine/formate buffers described previously [20]. Aliquots from 5.0 ml fractions were spotted on to Whatman No. 1 paper (Whatman International, Maidstone, U.K.) and amino acid peaks located by spraying with ninhydrin (0.25%, w/v in acetone). Fractions (65–74) eluting after lysine were freeze-dried, dissolved in 19.0 ml 0.01 M HCl and assayed for the presence of NFC-1 by amino acid analysis [10].

P₂ gel filtration

Separation was performed on a 100 cm×2.5 cm column packed with Bio-Gel P₂ (Bio-Rad Laboratories, Hemel Hempstead, U.K.) equilibrated in 10% (v/v) acetic acid. Material obtained from the PCEC column was freeze-dried, redissolved in 10% (w/v) acetic acid (0.5 ml) and applied to the P₂ column. The flow rate (0.5 ml/min) was controlled using an LKB HPLC system (Pharmacia, St. Albans, U.K.) and the effluent monitored by UV absorbance (205 nm). Fractions were freeze-dried and assayed for the presence of NFC-1.

Semi-preparative C₁₈ HPLC

Fractions containing NFC-1 separated by P₂ gel filtration were freeze-dried, dissolved in 1.0% heptafluorobutyric acid (HFBA) (6.0 ml) and loaded (500 µl) onto a 28×2.5 cm Spherisorb C₁₈ ODS2 column (Spherisorb, Queensferry, U.K.). The column was run isocratically in water for 10 min at 10 ml/min followed by a linear gradient to 20% acetonitrile over 40 min, before finally, being increased to 100% acetonitrile in 1 min. The gradient was controlled by a DuPont HPLC (DuPont Ltd., Herts, U.K.) fitted with 4× preparative pump heads. All buffers contained 0.05 M HFBA as an ion-pairing agent. Fractions were collected, freeze-dried, dissolved in 0.01 M HCl (0.5 ml) and the presence of amino acids identified by ninhydrin detection as described above. Ninhydrin-positive fractions were then assayed for the presence of NFC-1.

Superdex hydrophobic interaction chromatography

Fractions shown to contain NFC-1 were freeze-dried, dissolved in 1.0 ml 0.05 M acetic acid and 0.5 ml aliquots loaded onto a Pharmacia Superdex Peptide HF 10/30 column equilibrated in 0.05 M acetic acid attached to an FPLC system. Separation was performed at a flow rate of 0.75 ml/min and the effluent monitored at 214 nm. UV dense peaks were freeze-dried and assayed for the presence of NFC-1 [10].

HyperCarb S HPLC

NFC-1-rich fractions were dissolved in 0.1% TFA (500 µl) and aliquots (90 µl) fractionated by high-resolution chromatography using a 15×0.4 cm HyperCarb S column (Shandon Scientific Ltd., Runcorn, U.K.) equilibrated in 0.5% TFA in water. Amino acids were eluted using a gradient from 0–8% of 25% acetonitrile in 0.5% TFA as the limit buffer over 40 mins. The effluent was monitored at 205 nm and the UV dense peaks freeze-dried and assayed for the presence of NFC-1.

Superdex peptide gel filtration

Gel filtration chromatography was performed on a Pharmacia Superdex Peptide HR 10/30 column equilibrated in 45% acetonitrile/0.1% TFA to overcome hydrophobic interactions. NFC-1 rich fractions from the C₁₈ column were freeze-dried, dissolved in acetonitrile (0.3 ml) (45% in 0.1% TFA) and 0.05 ml aliquots were loaded onto the peptide column. Separation was performed on a Pharmacia FPLC system at a flow rate of 0.75 ml/min and the effluent monitored at 214 nm. UV dense peaks were freeze-dried and assayed for the presence of NFC-1.

Mass spectrometric analyses of NFC-1

Mass spectrometric analyses were performed by M-Scan Ltd. (Ascot, England).

(i) Matrix-assisted laser desorption ionization MS (MALDI-MS)

MALDI-MS was performed using a Voyager Elite Biospectrometry Research Station laser-desorption mass spectrometer coupled with delayed extraction. Sample aliquots (0.5 µl) were analysed using matrices of 2,5-dihydroxybenzoic acid and a-cyano-4-hydroxycinnamic acid. Glu-Fibrinopeptide was used to calibrate the instrument externally.

(ii) Electrospray MS–collisional activated dissociation MS (ESMS-CAD-MS)

Off-line ESMS was performed using the VG Bio-Q instrument (VG Biotech, Altrincham, U.K.) with quadropole analyser. Elution was carried out using a mixture of acetonitrile/0.1% aq. TFA/methoxyethanol/isopropanol in equal proportions at a flow rate of 10 µl/min. Sample aliquots were injected directly into the instrument source. Daughter ions of chosen signals were analysed following fragmentation using argon as the collision gas.

(iii) Accurate mass analysis

Positive fast atom bombardment (FAB) analysis was performed on a VG AutospecE mass spectrometer operating at a resolution of at least 10000. A VG caesium ion gun (operating at 30 kV) was used to generate spectra which were recorded on a VAX station M76 data system using VG Analytical Opus software. Aliquots of the sample were concentrated under a stream of nitrogen and loaded onto the FAB target, previously smeared with 2–4 µl of m-nitrobenzyl alcohol matrix containing 15-poly(ethylene glycol) (average molecular mass 200–400 Da) as an internal reference material.

Extended acid hydrolysis

Aliquots of NFC-1 were freeze-dried and subjected to an additional acid hydrolysis. This was performed in a Waters Pico-Tag work station at 110 °C using 6 N HCl in the vapour phase for 24 h. At the end of this period samples were freeze-dried, dissolved in 0.01 M HCl and analysed by complete amino acid analysis.

Synthesis of synthetic arginine derivatives

Methylglyoxal-derived imidazolone was prepared using a modification of the method described by Henle et al. [18]. Briefly, N^a-acetyl-L-arginine (1081 mg) and methylglyoxal (900 µl of a 40% aqueous solution; ICN Pharmaceuticals, Oxfordshire, U.K.) were dissolved in 50.0 ml sodium phosphate pH 7.4, and heated at 65 °C for 30 min. After cooling, concentrated HCl was added to a final concentration of 2.0 M and samples were incubated at 110 °C for 2 h to remove the blocking group. The solution was then freeze-dried, dissolved in 10.0 ml Milli-Q water and applied to the PCEC system (5.0 ml loadings). Fractions containing NFC-1 were pooled and freeze-dried as described above. This material was then dissolved in 45% acetonitrile in 0.1% TFA and 20 µl aliquots loaded onto a Superdex peptide column equilibrated in the same buffer. This was repeated 10 times and the fractions from each run containing NFC-1 were pooled, freeze-dried, dissolved in 1.5 ml Milli-Q water and subsequently analysed by mass spectrometry. A glyoxal/arginine derivative (Glarg) [21] was kindly donated to us and was analysed for any chromatographic similarities to NFC-1 and subsequently characterized by mass spectrometry.

RESULTS  

Preparative cation-exchange chromatography (PCEC)

Amino acid analysis of the crude hydrolysate estimated the NFC-1 concentration to be 4.2 mol/mol collagen, assuming a leucine equivalent of 2. This material was then applied to the PCEC column from which NFC-1 eluted in fractions (65–74) in conjunction with a number of contaminants C1 and C2 (Figure 1).

P₂ gel filtration

P₂ gel filtration showed that the material isolated by PCEC contained a broad range of molecular weight components. The bulk of the NFC-1 recovered from the column eluted in the region between 260 and 295 ml.

Semi-preparative C₁₈ HPLC

The NFC-1 fraction from the P₂ separation was then fractionated by semi-preparative HPLC using a C₁₈ column. Five major amino acid peaks, detected by ninhydrin spraying, were found to elute from the C₁₈ column. Amino acid analysis of these peaks identified three of them, eluting at 36–38, 47–49, 62–65 min respectively to be NFC-1 components (Figures 2A, 2C, and 2E). These components, a, b and gNFC-1, were obviously individual entities and were subsequently purified separately. The identity of the two other amino acid peaks (Figures 2B and 2D) which eluted from the C₁₈ column at approximately 45 (Cl) and 54 min (C2) is unknown.

Superdex hydrophobic interaction chromatography

aNFC-1 eluted in a series of peaks after the total column volume (Figure 3A). Amino acid analysis of the UV dense material indicated that NFC-1 eluted between 22–30 ml. The contaminant (X) from the C₁₈ column (Figure 2A) eluted earlier in the chromatogram (20–22 ml) and was effectively reduced by this method. As with aNFC-1, NFC-1 material from the C₁₈ column also eluted after the total column volume (21–27 ml), Figure 3B. Amino acid analysis of the UV dense material indicated that it contained NFC-1 as the major amino acid component.

HyperCarb S HPLC

aNFC-1 eluted from the HyperCarb column as a single peak at 25 min (Figure 3C). bNFC-1 eluted from the HyperCarb column as two peaks (Figure 3D). Fractions from the leading portion of the first peak, the middle region of both peaks and the trailing region of the second peak were pooled and freeze-dried.

Gel filtration

aNFC-1 material from the HyperCarb column was fractionated on a Pharmacia peptide column equilibrated in 45% acetonitrile/0.1% TFA, which minimized hydrophobic interactions. Under these conditions NFC-1 eluted between 19.5–20.5 ml (Figure 4A). Material isolated in this manner (Figure 4B) was then analysed by mass spectrometry. The three preparations of NFC-1 from the HyperCarb column were further purified on the Pharmacia peptide column. All three preparations produced a single UV peak between 18.0 and 19.5 ml (Figure 4C). Amino acid analysis of this region from all three preparations confirmed the presence of NFC-1 material (Figure 4D). Two of these preparations, leading and trailing, were then analysed by mass spectrometry. The small amounts of gNFC-1 material obtained from the HyperCarb S column (Figure 2E) were freeze-dried, dissolved in 45% acetonitrile (0.1% TFA) and loaded onto the Superdex peptide column equilibrated in the same buffer. Three main peaks (UV 214 nm) were detected eluting from the column (Figure 5). Fractions were pooled as indicated in Figure 5, freeze-dried and assayed for the presence of NFC-1 which revealed the presence of NFC-1 material in all regions examined (Figure 6). None of the preparations were present in sufficient quantities for further analysis.

Mass spectrometric analyses

(i) aNFC-1

The major peak after MALDI-MS was observed at m/z 215.2 (Figure 7A). Arginine was also run for comparative purposes and gave a protonated (M+H⁺) molecular ion at m/z 175.1 (Figure 7B). ESMS-CAD-MS was then performed on both. Collisionally activated decomposition of arginine using as the parent ion the protonated molecular ion (M+H⁺ m/z 175.1) produced the data shown in Figure 8A. The major fragment ions obtained were at m/z 60.4, 70.3, 116.3 with minor ions at 130.3 and 157.8. None of these ions were observed in the ESMS-CAD-MS spectrum of the eluant blank (parent ion m/z 175.1) which contained daughter ions at m/z 23.5 and 99.3. The ESMS-CAD-MS data produced by the NFC-1 sample using m/z 215.1 as the parent ion produced a very similar profile, the major difference being the disappearance of the m/z 60.4 fragment ion and the appearance of a fragment 40 Da higher at m/z 100.1 (Figure 8B). Accurate mass analysis of aNFC-1 was achieved by high resolution FAB analysis and gave a value of the protonated ion as 215.1163. The most likely elemental composition is C₈H₁₅N₄O₃.

(ii) bNFC-1

(a) Leading bNFC-1. Direct injection ESMS of the leading bNFC-1 peak (Figure 8C) gave a signal presumed to be the protonated molecular ion (M+H⁺) at m/z 229.3, approximately 54 Da higher than that of pure arginine. The additional peaks obtained were attributed to the matrix. ES-MS-CAD-MS data produced by the leading bNFC-1 peak using the m/z 229 as the parent ion are shown in Figure 8D. When compared to arginine (Figure 8A) and NFC-1 (Figure 8B) the major difference is the appearance of the m/z 114.0 fragment ion (approx. 54 Da higher than the signal at m/z 60.3 in the arginine daughter ion spectrum and 14 Da higher than the signal at m/z 100.1 in the aNFC-1 daughter ion spectrum).

(b) Trailing bNFC-1. Direct injection ESMS spectrum showed a signal presumed to be the protonated molecular ion (M+H)⁺ at m/z 229.3 (Figure 8E). The ES-MS-CAD-MS data produced by the trailing bNFC-1 sample using m/z 229 as the parent ion are displayed in Figure 8F and are almost identical to those produced by leading bNFC-1 (Figure 8D). Accurate mass analysis of bNFC-1 was achieved by high resolution FAB analysis and gave a measured value of the protonated ion as 229.1289. The most likely elemental composition is C₉H₁₇N₄O₃.

Extended acid hydrolysis

A mixture of the NFC-1 components described above was subjected to an additional hydrolysis procedure following which they decomposed to arginine and ornithine (Figure 9).

Analysis of synthetic arginine derivatives

A glyoxal/arginine derivative (Glarg) recently characterized [21] was analysed by amino acid hydrolysis and was shown to elute in the same position as the NFC-1 complex (Figure 10A). ES-MS-CAD-MS of the m/z 215.2 ion produced by this compound (Figure 10B) gave fragment ions almost identical to that of aNFC-1. The reaction of methylglyoxal and blocked arginine also produced a derivative which eluted in the same position as the NFC-1 complex (Figure 10C). ES-MS-CAD-MS of the m/z 229.0 ion produced by this compound (Figure 10D) gave fragment ions almost identical to that of bNFC-1.

DISCUSSION  

Although NFC-1 was originally described as a single amino acid peak, we have now shown that the NFC-1 peak when derived from ribose is in fact a complex mixture of high- and low-molecular-weight AGE compounds. Two low-molecular-weight components have now been characterized, i.e. NFC-1 and bNFC-1. Based upon their molecular weights, 229 and 215 respectively, their decomposition to arginine upon extended hydrolysis and their similarity to synthetic compounds the structures shown in Figure 11 have been proposed. A third component, gNFC-1, was not isolated in sufficient quantities for structural analysis, however, gel filtration chromatography indicated that it contained a mixture of NFC-1-like compounds, at least one of which was of a higher molecular weight than those described above. We have previously shown that glucose and ribose react differently [22]. Whether NFC-1 generated from glucose is also a multicomponent peak is currently being determined.

aNFC-1 is believed to be formed by the reaction of arginine with glyoxal. Two possible structures for this compound have been proposed, i.e. N^d-(4-oxo-5-dihydroimidazol-2-yl)--ornithine (Figure 11A) and 1-(4-amino-4-carboxybutyl)-2-imino-5-oxo-imidazolidine (Figure 11B). This latter structure was originally described by Schwarzenbolz et al. [21] based upon detailed NMR analysis of Glarg. bNFC-1 is believed to be formed by the reaction between methylglyoxal and arginine. This methyl imidazolone may be present in three tautomeric forms, N^d-(5-methyl-4-oxo-5-hydroimidazol-2-yl)--ornithine and N^d-(4-methyl-5-oxo-4-hydroimidazol-2-yl)--ornithine and 2-imino imidazolidinone (Figure 11C). An identical imidazolone has been isolated from the reaction of blocked arginine and methylglyoxal [17] and was initially purified from bakery products [18]. Both of these groups have also remarked upon the acid lability of this type of compound and the presence of ornithine in amino acid analysis chromatograms of acid hydrolysates of methylglyoxal-modified proteins and confirms the observation in the present study that these hydroimidazolones are susceptible to breakdown upon acid hydrolysis.

The formation of AGE compounds in vivo and in vitro has long been recognized to involve reactive dicarbonyl compounds derived from the reaction of reducing sugars and proteins. They may form directly from the autoxidation of the sugar or be present as reactive intermediates in Maillard-type reactions as a result of oxidative fragmentation of Schiff bases or Amadori products [2,3]. In recent years both glyoxal and methylglyoxal have been shown to be elevated in diabetes. Glyoxal is formed by lipid peroxidation and degradation of glucose and glycated protein whereas the major source of methylglyoxal, even in diabetes, is from the non-oxidative degradation of triosephosphates, acetone and ketone body metabolism and aminoacetone in threonine catabolism [23]. Consequently, a number of investigators have studied their role in ageing and disease [24–27]. In general this has involved the incubation of blocked amino acids (-amino) in the presence of the dicarbonyl followed by purification and characterization of the major reaction products. Using this approach a number of compounds have been isolated in vitro that may or may not have an important role in vivo. These include imidazolones formed by the reaction of arginine with methylglyoxal [17] and glycosylamine and biglycosylamines formed by the reaction of methylglyoxal and lysine [17]. Imidazolium and pyrimidinium cross-links derived from lysine derivatives and methylglyoxal have also been described [5,28,29] as well as hemithioacetal adducts formed by the reaction of cysteine with methylglyoxal [17]. However, only one of these compounds, imidazolysine, has actually been purified from modified protein [5]. The presence of other imidazolones formed by the reaction of 3-deoxyglucosone and arginine has also been indicated by amino acid analysis of in vitro glycated tissue hydrolysates [15]. These structures differ from those described above in that they have a trihydroxybutyl group attached to the imidazolone ring. Immunochemical studies using antibodies against these imidazolones have also indicated the presence of imidazolones in diseased tissue [30].

The presence of imidazolone adducts on protein may have profound effects upon its properties, e.g. large alterations in the charge profile of the protein and consequently its interaction with cells or other extracellular components, and its effectiveness as a substrate for proteinase enzymes. One of the most remarkable effects of hydroimidazolones is the recent discovery that they displace methylglyoxal-modified proteins from cell surface AGE receptors [31] and suggests that they are monocyte/macrophage receptor recognition factors which, when further investigated, may be one of the most profound effects of hydroimidazolone modifications of collagen. The isolation and characterization of glyoxal and methylglyoxal arginine derivatives from both in vivo aged and diseased tissues is currently under way in our laboratory. The identification of the major imidazolone formed in vivo, i.e. derived from glyoxal, methylglyoxal, 3-deoxyglucosone or some other dicarbonyl will give an invaluable insight into the mechanism by which AGE compounds are formed and provide targets for efficient pharmaceutical intervention.

We thank Uwe Schwarzenbolz for his advice and for very kindly donating Glarg. The work was supported by Wellcome grant no. 041073.

REFERENCES  

1 Paul, R. G. and Bailey, A. J. (1996) Int. J. Biochem. Cell Biol. 28, 1297–1310
Medline   1st Citation

2 Baynes, J. W. and Monnier, V. M. (1989) The Maillard Reaction in Ageing, Diabetes and Nutrition, A. R. Liss, New York
1st Citation  2nd

3 Labuza, D. V., Reineccius, G. A., Monnier, V. M., O'Brien, J. and Baynes, J. W. (1994) Maillard Reactions in Chemistry, Food, and Health, The Royal Society of Chemistry, Cambridge
1st Citation  2nd

4 Sell, D. R. and Monnier, V. M. (1989) J. Biol. Chem. 264, 21597–21602
Medline   1st Citation

5 Nagaraj, R. H., Shipanova, I. N. and Faust, F.M. (1996) J. Biol. Chem. 271, 19338–19345
Medline   J. Biol. Chem.   1st Citation  2nd  3rd

6 Nakamura, K., Nakazawa, Y. and Ienaga, K. (1997) Biochem. Biophys. Res. Commun. 232, 227–230
Medline   1st Citation  2nd

7 Nakamura, K., Hasegawa, T., Fukunaga, Y. and Ienaga, K. (1992) J. Chem. Soc. Chem. Commun. 992–994
1st Citation

8 Ienaga, K., Katita, H., Hochi, T., Nakamura, K., Nakazawa, Y., Fukunaga, Y., Aoki, S., Hasegawa, G., Tsutsumi, Y., Kitagawa, Y. and Nakano, K. (1996) Proc. Jpn. Acad., Ser. B 72, 79–84
1st Citation

9 Obayashi, H., Nakano, K., Shigeta, H., Yamaguchi, M., Yoshimori, K., Fukui, M., Fujii, M., Kitagawa, Y., Nakamura, N., Nakamura, K., Nakazawa, Y., Ienaga, K., Ohta, M., Nishimura, M., Fukui, I. and Kondo, M. (1996) Biochem. Biophys. Res. Commun. 226, 37–41
Medline   1st Citation

10 Bailey, A. J., Sims, T. J., Avery, N. C. and Halligan, E. P. (1995) Biochem. J. 305, 385–390
Medline   1st Citation  2nd  3rd

11 Ahmed, M. U., Thorpe, S. R. and Baynes, J. W. (1986) J. Biol. Chem. 261, 4889–4894
Medline   1st Citation

12 Hayase, F., Nagaraj, R. H., Miyata, S., Njoroge, F. G. and Monnier, V. M. (1989) J. Biol. Chem. 264, 3758–3764
Medline   1st Citation

13 Ahmed, M. U., Dunn, J. A., Walla, M. D., Thorpe, S. R. and Baynes, J. W. (1988) J. Biol. Chem. 263, 8816–8821
Medline   1st Citation

14 Ahmed, M. U., Frye, E. B., Degenhardt, T. P.,Thorpe, S. R. and Baynes, J. W. (1997) Biochem. J. 324, 565–570
Medline   Biochem. J.   Biochem. J.   1st Citation

15 Konishi, Y., Hayase, F. and Kato, H. (1994) Biosci. Biotech. Biochem. 58, 1953–1955
1st Citation  2nd

16 Hayase, F., Konishi, Y. and Kato, H. (1995) Biosci. Biotech. Biochem. 59, 1407–1411
1st Citation

17 Lo, T. W. C., Westwood, M. E., McLellan, A. C., Selwood, T. and Thornalley, P. J. (1994) J. Biol. Chem. 269, 32299–32305
Medline   1st Citation  2nd  3rd  4th  5th

18 Henle, T., Walter, A. W., Haessner, R. and Klostermeyer, H. (1994) Z. Lebensm. Unters. Forsch. 199, 55–58
1st Citation  2nd  3rd

19 Nagaraj, R. H., Porterootin, M. and Monnier, V. M. (1996) Arch. Biochem. Biophys. 325, 152–158
Medline   1st Citation

20 Light, N. D. and Bailey, A. J. (1982) Methods Enzymol. 82, 360–372
1st Citation

21 Schwarzenbolz, U., Henle, T., Haebner, R. and Klostermeyer, A. (1997) Z. Lebensm. Unters. Forsch. 205, 121–124
1st Citation  2nd  3rd

22 Paul, R. G., Avery, N. C., Sims, T. J., Hayes, C. and Bailey, A. J. (1996) Int. J. Exp. Pathol. 77, A1–A125
1st Citation

23 Thornalley, P. J. (1996) Endocrinol. Metab. 3, 149–166
1st Citation

24 Haik, G. M., Lo, T. W. C. and Thornalley, P. J. (1994) Exp. Eye Res. 59, 497–500
Medline   1st Citation

25 Thornalley, P. J., McLellan, A. C., Lo, T. W. C., Benn, J. and Sonksen, P. H. (1996) Clin. Sci. 91, 575–582
Medline   1st Citation

26 Thornalley, P. J., McLellan, A. C., Lo, T. W. C., Benn, J. and Sonksen, P. H. (1996) Diabetes 45, 996
1st Citation

27 McLellan, A. C., Thornalley, P. J., Benn, J. and Sonksen, P. H. (1994) Clin. Sci. 87, 21–29
Medline   1st Citation

28 Alabed, Y., Mitsubashi, T., Ulrich, P. and Bucala, R. (1996) Bioorg. Med. Chem. Letts 6, 1577–1578
1st Citation

29 Brinkmann, E., Wellsknecht, K. J., Thorpe, S. R. and Baynes, J. W. (1995) J. Chem. Soc. Perkin Trans., 2817–2818
1st Citation

30 Niwa, T., Katsuzaki, T., Miyazaki, S., Momoi, T., Akiba, T., Miyazaki, T., Nokura, K., Hayase, F., Tatemichi, N. and Takei, Y. (1997) Kidney Int. 51, 187–194
Medline   1st Citation

31 Westwood, M. E., Argirov, O. K., Abordo, E. A. and Thornalley, P. J. (1997) Biochim. Biophys. Acta Mol. Cell Res. 1356, 84–94
1st Citation

---

Received 25 June 1997/12 November 1997; accepted 28 November 1997

---

The Biochemical Society, London © 1998