METHODS FOR MEASURING AND ELIMINATING ADVANCED GLYCATION ENDPRODUCTS Technical Field of the Invention The present invention provides an advanced glycation endproduct (AGE) crosslink that exhibits immunological crossreactivity with AGEs formed in vivo and in vitro. Specifically, the present invention provides a method for standardizing an antibody-based diagnostic assay that measures AGE levels in a sample, comprising adding a defined amount of a certain AGE condensation product.

Background of the Invention The glycation reaction is manifest by the appearance of brown pigments during the cooking of food. identified by Maillard in 1912. Maillard observed that glucose or other reducing sugars react with amino-containing compounds, including amino acids and peptides, to form adducts that under go a series of dehydrations and rearrangements to form stable brown pigments. Heat-treated foods undergo non-enzymatic browning as a result of a reaction between glucose and a polypeptide chain. Thus, pigments responsible for the development of brown color that develops as a result of protein glycosylation possessed characteristic spectra and fluorescent properties.

Following the initial reaction of a reducing sugar with an amino-bearing substrate, subsequent reactions in the non-enzyme Malliard reaction pathway (including various dehydrations, oxidations, eliminations, condensations, cleavages and other chemical changes) occur to produce a vast array of"early"and"late"glycation adducts. More advanced glycation adducts are sometimes described as a class of yellow-brown, fluorescent pigments with intra-and intermolecular crosslinking activity. Specific glycation entities are thought to occur at low abundance within a widely divergent pool of advanced glycation endproducts (or AGEs). Despite significant research activity, the molecular structures of only a few of the later glycation adducts and products have been determined. Moreover, the contribution of these identified, in vivo-formed advanced glycation structures to biological processes is poorly understood. Therefore, there is a need in the art to identify AGEs and determine their biological properties.

The process of advanced glycation leads from the reversible interaction of reducing sugars with amino groups to the formation of more complex, irreversibly-bound structures with varied spectral and covalent cross-linking properties. These later products, termed advanced glycation endproducts or AGEs, form in vitro or in vivo by chemical principles first described for the Maillard reaction (Ledl and Schleicher, Angew. Chem. Int. Ed. Engl. 29: 565,1990; and Maillard, C. R. Hebd. Seances Acad. Sci. 154: 66,1912). The potential significance of Maillard-type reactions in living systems however, has been appreciated only over the last 15 years and the term advanced glycation has come to refer specifically with those aspects of Maillard chemistry that involve macromolecules and which occur under physiological conditions. It is evident that AGEs form in living tissues under a variety of circumstances, and that they play an important role in protein turnover, tissue remodeling, and the pathological

sequelae of diabetes and aging (Bucala and Cerami, Adv. Pharm. 23: 1,1992).

The initial event in protein glycation is the reaction of a reducing sugar such as glucose with the N-terminus of a protein or the s-amino group of a lysine to form an aldimine, or Schiff base. The Schiff base can hydrolyze back to its reactants or undergo an Amadori rearrangement to form a more stable N6- (1-deoxy-1-fructosyl) lysine (Amadori product, AP).

The reaction pathway leading to reactive crosslinking moieties (i. e. AGE formation) commences by further rearrangement or degradation of the AP. Possible routes leading from AP precursors to glucose-derived protein crosslinks has been suggested only by model studies examining the fate of the AP in vitro. One pathway proceeds by loss of the 4-hydroxyl group of the AP by dehydration to give a 1,4-dideoxy-1-alkylamino-2, 3-hexodiulose (AP-dione). An AP-dione with the structure of an amino-1,4-dideoxyosone has been isolated by trapping model APs with aminoguanidine, an inhibitor of the AGE formation (Chen and Cerami, J.

Carbohydrate Chem. 12: 731,1993). Subsequent elimination of 5-hydroxy then gives a 1.4,5- trideoxy-1-alkylamino-2,3-hexulos-4-ene (AP-ene-dione), which has been isolated as a triacetyl derivative of its 1,2-enol form (Estendorfer et al., Angew. Chem. Ent. Ed. Engl.

29: 536,1990). Both AP-diones and AP-ene-diones would be expected to be highly reactive in protein crosslinking reactions, for example, by serving as targets for the addition of a guanidine moiety from arginine or an s-amino group from lysine.

Dicarbonyl containing compounds, such as methylglyoxal, glyoxal and deoxyglucosones, participate in condensation reactions with the side chains of arginine and lysine. For example, the addition of methylglyoxal to the guanidine moiety of arginine leads to the formation of imidazol-4-one adducts (Lo et al., J. Biol. Chem. 269: 32299,1994) and pyrimidinium adducts (Al-Abed et al., Bioorg. Med. Chem. Lett. 6: 1577,1996). In one study, Sell and Monnier isolated pentosidine, an AGE fluorescent crosslink which is a condensation product of lysine, arginine, and a reducing sugar precursor (Sell and Monnier, J. Biol. Chem.

264: 21597,1989) from human dura collagen. The mechanism of pentosidine formation remains uncertain but crosslinking requires that the lysine-bound, glucose-derived intermediate contain a dicarbonyl functionality that can react irreversibly with the guanidinium group of arginine.

Several lines of evidence have established that AGEs exist in living tissue (Bucala and Cerami, Adv. Pharm. 23: 1,1992), yet the identity of the major AGE crosslink (s) that forms in vivo remains uncertain. Recent pharmacologically-based data nevertheless have affirmed the importance of the AP-dione pathway in stable crosslink formation (Vasan et al., Nature 382: 275,1996). The lack of precise data concerning the structure of AGEs has been attributed to the lability of AGE crosslinks to the standard hydrolysis methods employed to remove the protein backbone, and to the possible structural heterogeneity of the crosslinks themselves.

Moreover, there is data to suggest that the pathologically-relevant crosslinks may not themselves be fluorescent (Dyer et al. J. Clin. Invest. 91: 2463,1993), a property that has been historically associated with AGE formation and almost universally used as an indicator of the

Maillard reaction in vivo.

Hyperimmunization techniques directed against an AGE-crosslinked antigen produced both polyclonal and monoclonal antibodies that recognize in vivo formed AGEs (Makita et al., J. Biol. Chem. 267: 1997,1992). These antibodies made possible the development of immunohistochemical and ELISA-based technologies that were free of specificity and other technical problems associated with prior fluorescence-based assays, and provided the first sensitive and quantitative assessment of advanced glycation in living systems. These anti- AGE antibodies were found to recognize a class of AGEs that was prevalent in vivo but immunochemically distinct from previously characterized structures such as FFI, pentosidine, pyrraline, CML, or AFGP (Makita et al., J. Biol. Chem. 267: 1992,1992). The specific AGE epitope (s) recognized by these antibodies increased as a consequence of diabetes or protein age on various proteins such as collagen, hemoglobin, and LDL (Makita et al., J. Biol. Chem.

267: 1997,1992 ; Makita et al., Science 258: 651,1992; Wolffenbuttel et al., The Lancet 347, 513,1996; and Bucala et al. Proc. Natl. Acad. Sci. U. S. A. 91: 9441,1994). One particular polyclonal antibody species, designated"RU,"has been employed in an ELISA assay tested in human clinical trials. Immunoreactive AGEs were found to be inhibited from forming by administration of the pharmacological inhibitor, aminoguanidine (Makita et al., Science 258: 651,1992; and Bucala et al. Proc. Natl. Acad. Sci. U. S. A. 91: 9441,1994), and to provide important prognostic information correlated to diabetic renal disease (Beisswenger et al.

Diabetes 44: 824,1995).

Despite the increasing body of data implicating the advanced glycation pathway in the etiology of such age-and diabetes-related conditions as atherosclerosis, renal insufficiency, and amyloid deposition, elucidation of the structure (s) of the pathologically important AGE- crosslinks that form in vivo has been a challenging problem. Investigations of AGEs that form in vivo have necessarily relied on chemical methods to purify the crosslinking moieties away from their macromolecular backbones. These studies have led to a recognition that the major crosslinks which form in vivo are largely acid-labile and non-fluorescent (Bucala and Cerami, Adv. Pharm., 23: 1,1992; Sell and Monnier, J. Biol. Chem. 264: 21597,1989; and Dyer et al. J.

Clin. Invest. 91 : 2463,1993). However, in view of a predictive antibody-based (ELISA) diagnostic assay, there is a need in the art to isolate and identify immunogenic AGEs that can be used to both standardize and improve such diagnostic assays. The present invention was made in an effort to achieve the foregoing goals. Further, there is a need in the art to measure formation of advanced glycosylation endproducts in all applications where protein aging is a serious detriment. This includes, for example, the area of food technology (i. e., determination of the amount of food spoilage), perishability or shelf-life determination of proteins and other amino-containing biomolecules and pharmaceutical compositions.

Summary of the Invention The present invention provides a means for standardizing a kit that provides a means for measuring the formation of AGEs as a diagnostic assay. The present invention further

provides a novel isolated AGE that is antigenic and useful for forming antibodies having utility in diagnostic assays and for standardizing diagnostic assays.

The invention provides an advanced glycation endproduct (AGE) condensation product comprising a lysine component, an arginine component and a reducing sugar component.

Preferably, the condensation product is an AGE according to formula I: wherein the lysine component is indicated by the box labeled"K" ; the arginine component is indicated by the box labeled"R" ; and the reducing sugar component is not boxed; and wherein Rl and R4 are independently H or an amide bond to an amino acid residue or a peptide chain; R2 and R3 are, independently, OH or an amide bond to an amino acid residue or a peptide chain; Rs is H, CH20H or CHOHCH20H; and wherein if more than one of Rl, R2, R3 or R4 is an amide bond, then the lysine"K"component and the arginine"R"component may be amino acid residues of the same or a different peptide chain. Most preferably, the condensation product is an ALI (arg-lys-imidazole link) having the structure II shown at the left of the reaction scheme below: wherein Z is H, carboxybenzoyl, or the remainder of the polypeptide linked to the Arg and Lys groups; and Y is OH or the remainder of the polypeptide.

The present invention further provides a method for increasing macrophage recognition and elimination of advanced glycosylation endproducts, comprising administering to a mammal a therapeutic amount of a compound of formula I.

Brief Description of the Drawings Figure 1 shows the dose-dependent activity of synthetic ALI in an anti-AGE antibody-

based competitive ELISA to measure AGE content. The binding curve is steep and shows 50% inhibition at 500 nmoles. For comparison purposes, the reactivity of FFI, CML, AFGP, pentosidine and cyclic pentosidine, and ligands with related epitope structure, such as Na-Z- arg-lys, Na-Z-arg-lys-AP, an imidazolium adduct, a pyrimidinium adduct, histidine, and lys- his, were studied (wherein"Z"designates carboxybenzoyl). ELISA competition curves for the polyclonal anti-AGE antibody"RU"followed the methods described by Makita et al. (J. Biol.

Chem. 267: 1992,1992). Assays employed glucose-derived, AGE-BSA as the absorbed antigen. FFI: 4-furanyl-2-furoyl-lH-imidazole (Ponger et al,. Proc. Natl. Acad. Sci. U. S. A.

81: 2684,1984), CML: carboxy-methyllysine (Ahmed et al., J. Biol. Chem. 261: 4889,1986), AFGP: 1-alkyl-2-formyl-3,4-diglycosylpyrrole (Farmaretal., J. Org. Chem. 53: 2346,1988), PY: pyrraline (Njoroge et al., Carbohydrate Res. 167: 211,1987), P: pentosidine (Sell and Monnier, J. Biol. Chem. 264: 21597,1989), CP: cyclic pentosidine (CP) (Al-Abed et al., Bioorg. Med. Chem. Lett. 5: 2929,1995), AL: Na-Z-arg-lys (AL), ALAP: Na-Z-arg-lys-AP, H: histidine, LH: lys-his, PA: pyrimidinium adduct (Al-Abed et al., Bioorg. Med. Chem. Lett.

6: 1577,1996), and IA: imidazolium adduct (Brinkmann et al., J. Chem. Soc. Perkin Trans.

1: 2817,1995) showed no detectable crossreactivity with the RU anti-AGE antibody. With the exception of ALI, none of these haptens showed detectable crossreactivity with anti-AGE antibodies shown previously to react with in vivo-formed AGEs.

Figure 2 shows a schematic to form the novel 2-amino-4,5 dihydroxyimidazol adduct as an intermediate and then the ALI AGE product through a dehydration reaction.

Detailed Description of the Invention The present invention was made by exploiting the specificity of anti-AGE antibodies reactive with in vivo-formed AGEs to identify novel crosslinking moieties contained within a synthetic mixture of AGEs. This selection method found a single, immunoreactive AGE that formed in 0.6% yield in a synthetic mixture consisting of glucose and a Na-block dipeptide, Na-Z-arg-lys, as reactants. ALI was highly reactive with the anti-AGE antibody"RU"and showed a steep binding curve at nmole amounts.

The present invention found a novel genus of AGEs, designated ALI, based upon intramolecular crosslinking of adjacent residues in a model system for the formation of pentosidine-type AGEs, using a lys-arg-type dipeptide. In particular, the present invention provides a means for standardizing a kit that provides a means for measuring the formation of AGEs as a diagnostic assay. The present invention further provides a novel isolated AGE that is antigenic and useful for forming antibodies having utility in diagnostic assays and for standardizing diagnostic assays.

The invention provides a condensation product advanced glycation endproduct (AGE) comprising a lysine component, an arginine component and a reducing sugar component.

Preferably, the condensation product is an AGE according to formula I:

wherein the lysine component is indicated by the box labeled"K" ; the arginine component is indicated by the box labeled"R" ; and the reducing sugar component is not boxed; and wherein Rl and R4 are independently H or an amide bond to an amino acid residue or a peptide chain; R2 and R3 are, independently, OH or an amide bond to an amino acid residue or a peptide chain; R5 is H, CHzOH or CHOHCH20H; and wherein if more than one of Rl, R2, R3 or R4 is an amide bond, then the lysine"K"component and the arginine"R"component may be amino acid residues of the same or a different peptide chain. Most preferably, the condensation product is an ALI having the structure: wherein Z is H, carboxybenzoyl, or the remainder of the polypeptide linked to the Arg and Lys groups; and Y is OH or the remainder of the polypeptide. The compounds of formula I are prepared by incubating one or more polypeptides containing arg and lys with a reducing sugar, such as ribose, glucose, fructose, ascorbate or dehydroascorbate, at physiological pH for periods of 10-300 hours and optionally at elevated temperatures for shorter periods of time. A preferred compound, such as ALI of formula II, is prepared by incubating a N'-block dipeptide (e. g., the N «-Z-arg-lys dipeptide, N «-CBZ-arg-lys) with a reducing sugar, such as ribose, glucose, fructose, ascorbate or dehydroascorbate, at physiological pH for periods of 10- 300 hours and optionally at elevated temperatures for shorter periods of time. A preferred amine-protecting group is CBZ (carboxybenzyl) group due to its ease of removal and retention of stereochemistry during manipulations. The formed AGEs are purified, for example, by HPLC to provide the AGEs of formula I or formula II.

Among the biological activities of AGEs, the formation of stable crosslinks may be

considered their most important pathological manifestation. The imidazole-based AGE of formula I is a major species of the pathologically-important AGE crosslinks that form in vivo.

The mechanism of formation of ALI adducts points to the importance of the AP-dione as a critical, reactive intermediate and further affirms prior, pharmacologically-based studies that have implicated this intermediate, as well as its dehydration product, the AP-ene-dione in irreversible, protein-protein crosslinking (Vasan et al., Nature 382: 275,1996). These intermediates also have been implicated in the formation of the cyclization product cypentodine (Zhang and Ulrich, Tetrahedron Lett. 37: 4667,1996), which may display sufficient redox potential to participate in the oxidative reactions associated with phospholipid advanced glycation (Bucala et al., Proc. Natl. Acad. Sci. U. S. A. 90: 6434,1993; and Bucala, Redox Reports 2,291,1996).

There are several possible in vivo synthetic routes leading from Amadori product precursors to glucose-derived protein crosslinks. Applicants have examined models examining the fate of the Amadori products in vitro. For instance, the Amadori product can undergo dehydration to give 1,4-dideoxy-l-alkylamino-2,3-hexodiulose (AP-dione) (Figure 2). The dipeptide Na-Z-arg-lys was used as exemplarplary starting material for AGE formation. The proximity of arginine and lysine residues to each other promoted stable intramolecular crosslink formation. A N «-Z-arg-lys dipeptide was incubated with 10 equivalents of glucose in 0.2 M phosphate buffer (pH 7.4) at 37 °C for five weeks. This reaction mixture produced at least 25 distinct reaction products, recognized after fractionation of this mixture by HPLC.

Each fraction was isolated, concentrated, and analyzed for its reactivity with a polyclonal anti- AGE antibody (RU) that has been shown previously to recognize a class of AGEs that increase in vivo as a consequence of hyperglycemia, and which are inhibited from forming in human subjects by treatment with the advanced glycation inhibitor aminoguanidine. The products present within one fraction (1.5% yield) were found to block antibody binding in a competitive ELISA assay for AGEs in a dose-dependent fashion. Therefore, the compound of formula I is useful to standardize AGE-based diagnostic assays as either a standard target or a standard competitor, or both. Further purification of this fraction by HPLC revealed the presence of one major (0.6% yield) immunoreactive compound. Characterization of this adduct by UV, ESMS and'H-NMR spectra revealed the presence of an intramolecular arg-lys-imidazole crosslink (ALI of formula II). This crosslink is non-fluorescent and acid labile and may represent an important class of immunoreactive AGE-crosslinks that form in vivo.

Various haptens, antigens, and conjugated immunogens, corresponding to the AGEs of the present invention, including without limitation AGE ALI described in Example 1, can conveniently be prepared, either by isolation from incubation mixtures or by direct synthetic approaches. An N'-Z-arginine-lysine dipeptide, for example, can be incubated with a reducing sugar such as glucose in vitro, essentially as described in Examples 1 and 2, to form a mixture of AGEs, from which the specific AGE ALI of formula II can be purified by one of several convenient methods including, for instance, the HPLC method described in Examples 1

and 2. This AGE ALI then may be used as an immunogen to raise a variety of antibodies which recognize specific epitopes or molecular features thereof. In a preferred embodiment, AGE ALI itself is considered a hapten, which is prepared with or without the CBZ moiety and coupled to any of several preferred carrier proteins, including for instance keyhole limpet hemocyanin (KLH), thyroglobulin, and most preferred, bovine serum albumin (BSA), using any of a number of well-known divalent coupling reagents such as carbodiimide or other coupling reactions, according to protocols widely circulated in the art. Alternatively, the desired AGE can be synthesized ab initio. Irrespective of the source, the ALI or related AGE, whether alone or coupled to a carrier protein, may be employed in any well-recognized immunization protocol to generate antibodies and related immunological reagents that are useful in a number of applications owing to the specificity of the antibodies for molecular features of the ALI hapten or ALI-bearing immunogen.

Following a preferred protocol, any of several animal species may be immunized to produce polyclonal antisera directed against the ALI-carrier protein conjugate, including for instance mice, rats, hamsters, goats, rabbits, and chickens. The first three of the aforesaid animal species are particularly desired choices for the subsequent production of hybridomas secreting hapten-specific monoclonal antibodies. The production of said hybridomas from spleen cells of immunized animals may conveniently be accomplished by any of several protocols popularly practiced in the art, and which describe conditions suitable for immortalization of immunized spleen cells by fusion with an appropriate cell line, such as a myeloma cell line. Such protocols for producing hybridomas also provide methods for selecting and cloning immune splenocyte/myeloma cell hybridomas and for identifying hybridoma clones that stably secrete antibodies directed against the desired epitope (s).

Animal species, such as rabbit and goat, are more commonly employed for the generation of polyclonal antisera, but regardless of whether polyclonal antisera or monoclonal antibodies are desired ultimately, the hapten-modified carrier protein typically is initially administered in conjunction with an adjuvant such as Complete Freund's Adjuvant.

Immunizations may be administered by any of several routes, typically intraperitoneal, intramuscular or intradermal; certain routes are preferred in the art according to the species to be immunized and the type of antibody ultimately to be produced. Subsequently, booster immunizations are generally administered in conjunction with an adjuvant such as alum or Incomplete Freund's Adjuvant. Booster immunizations are administered at intervals after the initial immunization; generally one month is a suitable interval, with a blood samples taken between one and two weeks after each booster immunization. Alternatively, a variety of so- called"hyperimmunization"schedules, which generally feature booster immunizations spaced closer together in time, are sometimes employed in an effort to produce anti-hapten antibodies preferentially over anti-carrier protein antibodies.

The antibody titers in post-boost blood samples can be compared for hapten-specific immune titer in any of several convenient formats including, for instance, Ouchterlony

diffusion gels and direct ELISA protocols. In a typical direct ELISA, a defined antigen is immobilized onto the assay well surface, typically in a 96-well or microtiter plate format, followed by a series of incubations separated by rinses of the assay well surface to remove unbound binding partners. By way of example, the wells of an assay plate may receive a dilute, buffered aqueous solution of the hapten/carrier conjugate, preferably wherein the carrier protein differs from that used to immunize the antibody-producing animal to be tested; such as serum from ALI/KLH conjugate-immunized animals might be tested against assays wells decorated with immobilized ALI/BSA conjugate. Alternatively, the assay surface may be decorated by incubation with the hapten alone. Generally, the surface of the assay wells is then exposed to a solution of an irrelevant protein, such as casein, to block unoccupied sites on the plastic surface. After rinsing with a neutral buffered solution that typically contains salts and a detergent to minimize non-specific interactions, the well is then contacted with one of a serial dilution of the serum prepared from the blood sample of interest (the primary antiserum).

After rinsing again, the extent of test antibodies immobilized onto the assay wells by interaction with the desired hapten or hapten/carrier conjugate can be estimated by incubation with a commercially available enzyme-antibody conjugate, wherein the antibody portion of this secondary conjugate is directed against the species used to produce the primary antiserum; such as if the primary antiserum was raised in rabbits, a commercial preparation of anti-rabbit antibodies raised in goat and conjugated to one of several enzymes, such as horseradish peroxidase, can be used as the secondary antibody. Following procedures specified by the manufacturer, the amount of this secondary antibody immobilized on the assay plate by interaction with the primary antibody can then be estimated quantitatively by the activity of the associated conjugate enzyme in a colorimetric assay. Many related ELISA or radioimmunometric protocols, such as competitive ELISAs or sandwich ELISAs, all of which are well known in the art, may optionally be substituted, to identify the desired antisera of high titer; that is, the particular antisera which give a true positive result at high dilution (e. g., greater than 1/1000 and more preferably greater than 1/10,000).

Similar immunometric protocols can be used to estimate the titer of antibodies in culture supernatants from hybridomas prepared from spleen cells of immunized animals. In so characterizing antisera or hybridoma supernatants, it is desirable to employ a variety of control incubations, such as with different carrier proteins, related but structurally distinct haptens or antigens, and omitting various reagents in the immunometric procedure in order to identify reliable determinations of antibody specificity and titer from false positive and false negative results. The types of control incubations to use in this regard are well known. Also, the same general immunometric protocols subsequently may be employed with the antisera identified by the above procedures to be of high titer and to be directed against specific structural determinants in the AGEs of the present invention. These desired antibodies may be used in a variety of applications, and particularly to quantify the presence and extent of AGEs on biological samples, foodstuffs, or other amine-bearing substances and molecules of interest.

Such latter applications of the desired anti-ALI antibodies, whether polyclonal or monoclonal, together with instructions and optionally with other useful reagents and diluents, including, <BR> <BR> <BR> without limitation, a set of molecular standards of the AGE ALI, may be provided in kit form for the convenience of the operator.

Example 1 This example illustrates a synthesis to prepare Arg-Lys-Imidazole (ALI). To a solution of Na-Z-arg-lys (l g, 0.013 mmol) in 10 ml of aqueous 0.2 M phosphate buffer (pH 7.4) was added D-glucose (0.13 mmol). The reaction mixture was stirred at 37 °C for five weeks. At intervals, 10 pl of the reaction mixture was analyzed by HPLC using an analytical Primesphere column (5C18 MC, 5 micron, 250 x 4.6 mm, Phenomenex, Torrance, CA) and a binary solvent gradient consisting of 0.05% TFA in H20 (solvent A), and methanol (solvent B). Solvent was delivered at a flow rate of 1 ml/min as follows. From 0-30 min: a linear gradient from A: B (95: 5) to A: B (25: 75); from 30-45 min: a linear gradient from A: B (25: 75) to (0: 100). Detection was by monitoring UV absorption at X 214,254,280,320, and 350 nm.

At least 25 distinct reaction products were identified upon fractionation of this mixture by HPLC. Larger amounts of these products were fractionated using a similar HPLC method (Primesphere column 5C18 MC, 5 micron, 250 x 21.2 mm, Phenomenex, Torrance, CA).

Solvent was delivered at a flow rate of 10 mL/min using the same gradient as described above.

The AGE crosslink eluted as a mixture of three components at 34.0 min. Further purification of this subfraction using the same method gave the desired compound in a high purity (>95%).

Example 2 This example illustrates the discovery and isolation of the AGE of formula I. The dipeptide N°-Z-arg-lys was selected as a target, because a close association of the arginine and lysine residues provides a significant proximity effect that promotes crosslink formation.

Moreover, a synthetic strategy employing an arg-lys dipeptide was used successfully in the past to isolate a cyclic pentosidine in a high yield (Beisswenger et al., Diabetes 44: 824,1995).

Na-Z-arg-lys (13 mmoles) was incubated together with 10 equivalents of glucose in 0.2 M phosphate buffer (pH 7.4) for five weeks at 37 °C. At least 25 distinct reaction products were identified upon fractionation of this mixture by reverse-phase HPLC. Each fraction was isolated, concentrated, and analyzed for its reactivity with anti-AGE antibody by ELISA (Makita et al., J. Biol. Chem. 267: 1992,1992). Briefly, HPLC fractions and purified compounds were analyzed by an AGE-specific ELISA following methods described previously (Makita et al., J. Biol. Chem. 267: 1992,1992). This ELISA employed a polyclonal anti-AGE antibody raised by hyperimmunization against a heavily AGE-crosslinked preparation of ribonuclease. Total IgG was prepared by protein-G affinity chromatography and the ribonuclease backbone specificities removed by immunoabsorption. For assaying AGE immunoreactivity, 96-well round bottom microtitre plates (EIA/RIA plate, Costar, Cambridge, MA) first were coated with AGE-BSA (3 mg/ml, dissolved in 0.1 M sodium bicarbonate, pH

9.6) by incubation overnight at 4 °C. After washing, the unbound sites were blocked with SuperBlockTM following the manufacturer's recommendations (Pierce, Rockford, IL). After washing, dilutions of test antigen, together with anti-AGE IgG, were added and the plates incubated at room temperature for 1 hr. The plates then were washed again and incubated with a secondary antibody (alkaline phosphatase-conjugated anti-rabbit IgG) at 37 °C for 1 hr. The unbound antibodies were removed by extensive washing and bound antibodies were detected by incubation withp-nitrophenyl phosphate (pNPP) substrate for 30-60 min, and recording the optical density at 405 nM by an ELISA reader (EL309, Bio-Tek Instruments Inc., Burlington, VT). Results were expressed as B/Bo, calculated as [experimental OD-background OD (i. e. no antibody)]/ [total OD (i. e. no competitor)-background OD].

The product (s) contained within one distinct fraction, present in 1.5% yield, were found to block antibody binding in a dose-dependent fashion. Further purification of this fraction by HPLC revealed the presence of one major, immunoreactive compound (0.6% yield), together with two minor ones. The UV and fluorescence spectrum of the isolated, major product was unremarkable and similar to that of the starting material. The ESMS spectrum displayed a molecular ion of m/z 545 [MH] +, an increase of 108 daltons compared to the starting material, Na-Z-arg-lys (MW: 436). A'H-NMR spectrum in D20 showed, in addition to the N'-Z-arg- lys protons, five aliphatic protons that resonate as a multiplet between 2.35-4.65 ppm (5H), and an olefinic proton that resonates at 7.3 ppm within the Z-group. Overall, these data are consistent with the structure of a cyclic, arg-lys imidazole crosslink (ALI, formula II, Figure 2).

Example 3 This example illustrates a proposed mechanism of the AGE formation (Figure 2). First, dehydration of the lysine-derived AP (which has been determined to form in 18% yield under these reaction conditions) gives an obligate, AP-dione reactive intermediate. Reversible addition of the guanidine moiety to the dicarbonyl yields the 2-amino-4,5-dihydroxyimidazole adduct, which then undergoes dehydration to deliver the stable cyclic ALI. Of importance, this crosslink is non-fluorescent, acid-labile, and can be inhibited from forming by aminoguanidine.