N ε-(Carboxymethyl)Lysine Adducts of Proteins Are Ligands for Receptor for Advanced Glycation End Products That Activate Cell Signaling Pathways and Modulate Gene Expression*

Recent studies suggested that interruption of the interaction of advanced glycation end products (AGEs), with the signal-transducing receptor receptor for AGE (RAGE), by administration of the soluble, extracellular ligand-binding domain of RAGE, reversed vascular hyperpermeability and suppressed accelerated atherosclerosis in diabetic rodents. Since the precise molecular target of soluble RAGE in those settings was not elucidated, we tested the hypothesis that predominant specific AGEs within the tissues in disorders such as diabetes and renal failure,N ε-(carboxymethyl)lysine (CML) adducts, are ligands of RAGE. We demonstrate here that physiologically relevant CML modifications of proteins engage cellular RAGE, thereby activating key cell signaling pathways such as NF-κB and modulating gene expression. Thus, CML-RAGE interaction triggers processes intimately linked to accelerated vascular and inflammatory complications that typify disorders in which inflammation is an established component.

and, indeed, natural aging, albeit to a lesser degree, has suggested their potential contribution to the pathogenesis of complications that typify these conditions (3)(4)(5)(6)(7). Our previous studies demonstrated that both in vitro and in vivo derived heterogeneous AGEs ligate cell surface RAGE on endothelium (ECs), mononuclear phagocytes (MPs), vascular smooth muscle (VSMC), and neurons to activate cell signaling pathways such as ERK1/ERK2 kinases and NF-B (8 -9), thereby redirecting cellular function in a manner linked to expression of inflammatory and prothrombotic genes important in the pathogenesis of chronic disorders as apparently diverse as diabetic macrovascular disease and amyloidosis (10 -20).
Our recent studies suggested that interruption of the interaction of AGEs with RAGE in vivo, by administration of soluble RAGE (sRAGE), the extracellular ligand-binding domain of RAGE, reversed vascular hyperpermeability and suppressed accelerated atherosclerotic lesion development and complexity in diabetic rodents (19 -20). In the latter studies, analysis of plasma demonstrated evidence of an sRAGE⅐AGE complex; immunoprecipitation of plasma obtained from diabetic sRAGEtreated mice with anti-RAGE IgG yielded species immunoreactive with both anti-RAGE IgG or affinity purified anti-AGE IgG, suggesting that sRAGE might bind up AGEs and limit their interaction with and activation of cell surface RAGE. The beneficial effects of sRAGE were independent of alterations in other risk factors, such as hyperglycemia and hyperlipidemia, implicating a role for AGE-RAGE interaction in the development of vascular dysfunction in diabetes (20).
These past studies, however, did not elucidate the precise AGE(s) that trigger signal transduction mechanisms upon engagement of RAGE. We thus sought to test specific AGE structures for their ability to bind RAGE on the surface of cells such as ECs, MPs, and VSMCs in order to determine their role in cellular activation.
In this context, recent biochemical and immunohistochemical studies suggested that N ⑀ -(carboxymethyl)lysine (CML) modifications of proteins are predominant AGEs that accumulate in vivo (21)(22)(23)(24). Elevated serum levels of CML were demonstrated in patients with diabetes (24 -25) and renal failure (26). Importantly, enhanced accumulation of CML was shown in vascular tissue, atherosclerotic lesions, and glomerular tissue retrieved from diabetic rodents and human subjects (25,(27)(28)(29)(30). In these settings, CML adducts co-localized with oxidation epitopes, such as malondialdehyde and 4-hydroxynonenal. These observations are consistent with the concept that beyond processes mediating glycoxidation of proteins (31), lipid oxidation itself triggers generation of CML (32), thereby establishing a likely link between enhanced glycation observed in diabetic hyperglycemia and disturbances of lipid metabolism, common to both types 1 and 2 diabetes (33).
Furthermore, recent findings suggested that CML modifications may form directly as a consequence of activation of the myeloperoxidase-hydrogen peroxide-chloride system, thereby providing a mechanism for conversion of hydroxy amino acids into glycoaldehyde, a precursor in the steps leading to formation of CML (34). These findings may have direct implications for inflammatory processes that characterize certain complications of diabetes and renal failure, for example, such as atherosclerosis, impaired wound healing, aggressive and inflammatory periodontal disease, and dialysis-related amyloidosis (DRA) (6,(35)(36).
Consistent with these concepts, previous studies placed RAGE in the vascular and inflammatory milieu characteristic of disorders in which AGEs, such as CML, accumulate; indeed, expression of RAGE is enhanced in these settings, beyond that observed in normal adult tissues (37)(38)(39)(40)(41). Taken together, therefore, these considerations suggested that examination of CML adducts of proteins as potential specific AGE ligands for RAGE was a logical step.
Here we report that CML-modified proteins engage cellular RAGE in vitro and in vivo to activate key cell signaling pathways such as the transcription factor NF-B, with subsequent modulation of gene expression. Together, these findings link CML-RAGE interaction to the development of accelerated vascular and inflammatory complications that typify disorders in which inflammation is an established component.

EXPERIMENTAL PROCEDURES
Synthesis and Characterization of CML-modified Proteins-Keyhole limpet hemocyanin (KLH), bovine serum albumin (BSA), or ovalbumin (OVA) (Sigma) (176 mg in each case), and sodium cyanoborohydride (28.3 mg; 0.45 M) were dissolved in sodium phosphate buffer (0.2 M; pH 7.8); to this was added glyoxylic acid (14.3 mg; 0.155 M) in a total volume of 1 ml per reaction. The mixture was incubated for 24 h at 37°C. Control proteins were prepared under the same conditions, except that glyoxylic acid was omitted (23). These published methods were modified in order to prepare proteins with a range of CML modifications as follows: protein was incubated in buffer exactly as above, except that varying amounts of glyoxylic acid were added as follows: 10 mg (0.108 M), 7 mg (0.076 M), 4 mg (0.043 M), 2 mg (0.0217 M), and 1 mg (0.0108 M); concomitantly varying amounts of sodium cyanoborohydride were added as follows: 19.8 mg (0.3152 M), 14.1 mg (0.225 M), 7.1 mg (0.1125 M), 3.5 mg (0.0556 M), and 2 mg (0.0318 M), respectively, as described previously. Preparations of CML-modified proteins were extensively dialyzed versus phosphate-buffered saline (PBS) and characterized by percent modification as determined both by employment of 2,4,6-trinitrobenzene sulfonic acid to determine the difference in lysine residues of modified versus unmodified preparations (42) and by gas chromatography-mass spectroscopy (43). To test CML-modified adducts, 0.8 mmol of CML/mol of Lys, CML-OVA, 33 mmol of CML/mol of Lys, were diluted 1:40 with native OVA in PBS. All CML-modified native proteins, antibodies/F(abЈ) 2 fragments, and control proteins were devoid of endotoxin prior to experiments by chromatography onto Detox-igel columns (Pierce). The level of endotoxin in all protein preparations (concentration range, 2-6 mg/ml) was less than 3 pg/ml (Sigma).
Preparation and Characterization of Antibodies-Anti-human RAGE IgG and affinity purified anti-CML IgG were prepared and characterized as described previously (44 and 24, respectively). In the latter case, CML-BSA and CML-KLH prepared as above were employed to immunize New Zealand White rabbits (24). After 6 -12 weeks, serum was obtained and IgG prepared (Pierce). Affinity purification of the IgG fractions was performed as follows: IgG fractions were chromatographed onto Affi-Gel 15 resin (Bio-Rad), which had previously adsorbed KLH. Material that did not adhere to the resin was collected and subsequently chromatographed onto Affi-Gel 15 (Bio-Rad) which had previously been adsorbed CML-BSA. After extensive washing in Trisbuffered saline (Tris, 0.02 M; pH 7.4, and NaCl, 0.1 M) containing Tween 20 (0.05%), fractions were eluted in buffer containing glycine, 0.02 M; pH 2.5. A 280 nm of each fraction was determined; positive fractions were immediately neutralized and dialyzed versus Tris-buffered saline. Enzyme-linked immunosorbent assays (ELISA) were performed to char-acterize the affinity purified antibodies. To wells of plastic dishes (Maxisorp, Nunc, Naperville, IL) was added 5 g of the following proteins in bicarbonate/carbonate buffer, pH 9.6: BSA, KLH, CML-BSA, or CML-KLH. After incubation for 16 h at 4°C, wells were washed in PBS and then unoccupied sites on the wells blocked in the presence of PBS containing BSA (1%) and goat serum (5%; Sigma) for 1 h at 25°C. Wells were washed in PBS containing Tween 20 (0.05%) and primary antibody, and affinity purified anti-CML IgG was added at a concentration of 3 g/ml in blocking buffer as above for 3 h at 25°C. Wells were washed five times in PBS, and then sites of primary antibody binding were identified upon incubation with blocking buffer containing peroxidase-conjugated goat anti-rabbit IgG according to the manufacturer's instructions (Sigma) for 1 h at 25°C. Wells were washed five times and developed in citrate buffer containing o-phenylenediamine/H 2 O 2 as instructed (Sigma). In these assays, only CML modifications of proteins were highly reactive; no reactivity was identified versus native, unmodified proteins. Furthermore, upon incubation of affinity purified anti-CML IgG with excess CML-BSA, immunoreactivity of CML-modified adducts was abolished.
Isolation of Human Serum Albumin and Immunoblotting-Plasma samples were obtained from human donors with diabetes, renal failure, or age-matched healthy controls in accordance with the procedures of the Institutional Review Board of Columbia University. Samples (2 ml) were dialyzed for 16 h at 4°C in phosphate buffer, 0.02 M; pH 7.1 (total volume, 4,000 ml). After dialysis, samples were subjected to filtration, 0.8 m, and then chromatographed onto columns containing Affi-Gel blue resin (Bio-Rad) previously equilibrated in phosphate buffer, 0.02 M; pH 7.1; 5 ml of resin were employed per ml of plasma. The resin was washed in 2.5 column volumes of phosphate buffer as above, and human serum albumin was eluted by application of phosphate buffer containing NaCl (1.4 M). The A 280 nm for each fraction was determined (Ultraspec Plus, Amersham Pharmacia Biotech), and positive fractions were determined by SDS-PAGE followed by staining with silver (Bio-Rad). Immunoblotting was performed employing the above material; to each lane of SDS-PAGE gels (8%), 30 g of protein was added. Simultaneously, marker proteins were added as a means to assess ϷkDa (Amersham Pharmacia Biotech). After electrophoretic separation, contents of the gels were transferred to nitrocellulose membranes (Bio-Rad). Unoccupied sites on the membranes were blocked in the presence of nonfat dry milk in TBS (13.5%) for 4 h at room temperature. Immunoblotting was performed using affinity purified anti-CML IgG as above (14.2 g/ml) in milk buffer (5%) for 1.5 h at 37°C. Membranes were washed extensively in TBS containing Tween 20 (0.1%) and membranes incubated with goat anti-rabbit IgG labeled with horseradish peroxidase (Sigma) for 1 h at 37°C. Membranes were washed extensively in the above buffer, and visualization of antibody binding was performed employing the ECL detection system (Amersham Pharmacia Biotech). Quantitative evaluation of band intensity was performed using Molecular Dynamics/ImageQuant (Foster City, CA).

Synthesis of Advanced Glycation End Products (AGEs) and
Determination of Components That Bind RAGE-Human IgG (Sigma), 5 mg/ml in PBS, was subjected to nonenzymatic glycation by incubation in PBS containing D-ribose, 0.025 M (Sigma). The solution was sterile-filtered (0.2 m) and then incubated at 37°C for 6 weeks under aerobic conditions. At the end of that time, the mixture was extensively dialyzed versus PBS at 4°C to remove unreacted ribose. This material was then chromatographed onto resin containing Affi-Gel 10 (Bio-Rad) to which had previously been adsorbed recombinant human soluble RAGE. After incubation, the resin was washed extensively with 10-column volumes of Tris-buffered saline (Tris, 0.02 M, pH 7.4; NaCl, 0.1 M) and elution of bound components performed employing glycine, 0.02 M; pH 2.5. The A 280 nm of each fraction was determined; positive fractions were immediately neutralized and dialyzed versus TBS. ELISA was performed essentially as described above employing varying concentrations of CML-BSA as standard reference. Wells of plastic dishes were coated with AGE-IgG, material eluted from the above column, and volume control (buffer retrieved after elution of RAGE-binding components subsequent to washing with 10-column volumes of glycine buffer). Primary antibody, affinity purified anti-CML IgG was employed (3 g/ml) to identify CML-immunoreactivity in the various fractions. Finally, goat anti-rabbit IgG was added to detect sites of binding of primary antibody, and plates were developed as above. A 490 nm was detected.
Radioligand Binding Assays-Human soluble recombinant RAGE was radiolabeled with 125 I by incubation with IODO-BEADS (Pierce) to a specific activity of Ϸ4,000 to 5,000 cpm/ng protein. In all cases, after radioiodination, precipitation of the radiolabeled material in trichloroacetic acid exceeded 90%. CML modifications of proteins as indicated or native, unmodified proteins were loaded onto the wells of plastic dishes (Maxisorp, Nunc) (5 ng/well) in bicarbonate/carbonate buffer, pH 9.6, and incubated for 16 h at 4°C. Material in the wells was aspirated, and unoccupied sites were blocked by incubation with PBS containing BSA (1%) for 2 h at 37°C. Wells were washed twice with PBS containing octyl ␤-glucoside (0.005%) (Roche Molecular Biochemicals). A radioligand binding assay was performed in PBS containing BSA (0.2%) with the indicated concentration of 125 I-human sRAGE alone or in the presence of a 50-fold molar excess of unlabeled human sRAGE or the indicated competitor for 2 h at 37°C. At the end of that time, wells were washed rapidly five times with washing buffer as above; elution of bound material was performed in a solution containing heparin, 1 mg/ml. Solution was aspirated from the wells and counted in a gamma counter (Amersham Pharmacia Biotech). Equilibrium binding data were analyzed according to the equation of Klotz and Hunston (45): B ϭ nKA/1 ϩ KA, where B indicates specifically bound ligand (total binding, wells incubated with tracer alone, minus nonspecific binding, wells incubated with tracer in the presence of excess unlabeled material), n indicates sites/cell, K indicates the dissociation constant, and A indicates free ligand concentration) using nonlinear least squares analysis (Prism; San Diego, CA). Specific binding of CML-BSA to radiolabeled RAGE was further determined by subtraction of nonspecific binding (counts obtained upon binding of radiolabeled sRAGE to immobilized BSA) from that obtained upon binding of radiolabeled sRAGE to immobilized CML-BSA. In the case binding to immobilized BSA, counts were negligible and less than 10% that observed in the presence of CML-BSA. Where indicated, pretreatment with either antibodies, soluble RAGE, or the indicated potential competitor was performed for 2 h prior to binding assay. In certain experiments, material eluted from the RAGE-Affi-Gel 10 columns (after passage of AGE-IgG) was tested as unlabeled competitor in the binding assay; controls were performed with equal volumes of glycine buffer. In other experiments, isolated RAGE domains were employed as unlabeled competitors. Human RAGE cDNA encoding the V, C1, or C2 domain was inserted into the EcoRI site of pGEX4T vector containing GST. Fusion proteins, V-GST, C1-GST, and C2-GST, were expressed in Escehrichia coli, purified on a glutathione-Sepharose column, and cleaved with thrombin (Amersham Pharmacia Biotech). RAGE domains were then purified to homogeneity using glutathione-Sepharose and characterized by SDS-PAGE and amino-terminal sequencing (1-2) prior to testing in the radioligand binding assay.

Assessment of Cell Surface VCAM-1 and Binding of Radiolabeled Molt-4 Cells to Stimulated Human Umbilical Vein Endothelial Cells (HUVEC)-Human umbilical vein endothelial cells (HUVECs)
were isolated and characterized as described (46). Cells were cultured in 96-well tissue culture-treated wells (Corning, Corning, NY) in endothelial cell growth medium (Clonetics, San Diego, CA) until achieving confluence; medium was then changed to F-12 without serum (Life Sciences) immediately prior to stimulation with the indicated concentrations of CML-ovalbumin or native ovalbumin. Where indicated, cells were pretreated with rabbit anti-human RAGE IgG, nonimmune rabbit IgG for 2 h; in certain cases, CML-OVA was pretreated with the indicated concentration of sRAGE for 1 h prior to cell stimulation. After 6 h, cells were fixed in medium containing paraformaldehyde (2%) for 10 min followed by incubation in fresh paraformaldehyde (2%) for 16 h at 4°C. Wells were washed twice with PBS and incubated with H 2 O 2 (0.3%) for 10 min at room temperature. Wells were again washed twice with PBS and incubated in PBS containing BSA (2%) and non-fat dry milk (4%) to block nonspecific binding sites on the cell surface for 30 min at room temperature. After washing once in PBS, cell surface ELISA employing anti-VCAM-1 IgG (Santa Cruz Biotechnologies, Santa Cruz, CA) (2 g/ml) was performed for 2 h at 37°C. Wells were washed four times and incubated with peroxidase-conjugated rabbit anti-goat IgG (Sigma) for 1 h at 37°C. Wells were washed five times and sites of antibody binding detected with OPD as above; measurements of A 490 nm per well were obtained. Assessment of functional VCAM-1 activity was determined using 51 Cr-labeled Molt-4 cells (ATCC) as described (14).
Chemotaxis Assays-Chemotaxis assays were performed as described (10) in 48-well microchemotaxis chambers (Neuro-Probe, Bethesda, MD) containing a polycarbonate membrane (8 m; Nucleopore, Pleasanton, CA). Molt-4 cells, which bear cell surface RAGE, were grown in suspension in medium containing RPMI 1640, fetal bovine serum (10%), and antibiotics (Life Sciences). The lower chamber contained the chemotactic stimulus as indicated. N-Formyl-Met-Leu-Phe (Sigma) was employed as positive control. Molt-4 cells were added to the upper chamber (5 ϫ 10 4 cells/well). After incubation for 4 h at 37°C, nonmigrating cells on the upper surface of the membrane were gently scraped and removed; the membrane was then fixed in methanol (100%), and cells that had migrated through the membrane were stained with Giemsa (Sigma). Cells in nine high-powered fields were counted and mean Ϯ S.E. reported. Where indicated, cells were incubated with the indicated F(abЈ) 2 for 2 h at 37°C prior to assay; in certain cases, CML-modified proteins were incubated with the indicated molar excesses of human soluble RAGE for 1 h at 37°C prior to assay. In other experiments, cells were transfected with a construct encoding human RAGE in which the cytosolic domain (tail) was deleted employing superfect (Qiagen, Valencia, CA) (2 g DNA/ml medium); pcDNA3 (Invitrogen) was employed as vector. Stimulation experiments were performed 48 h after transfection. In cells transfected with RAGEtail deletion construct, the ability to bind ligand is retained as RAGElacking cytosolic domain is firmly anchored in the membrane (41).
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assay (EMSA)-HUVEC, murine macrophage BV-2 cells (47), and rat vascular smooth muscle cells (VSMC; generously provided by Dr. Abraham Rothman) (9) were incubated with the indicated CML-modified or native protein for 6 h at 37°C in the presence or absence of preincubation with the indicated IgG or sRAGE/control protein for 2 and 1 h, respectively, at 37°C. In certain cases, transient transfection was performed employing mock, or RAGE tail deletion, constructs as above, except that DNA, 0.4 g/ml medium, was employed. Nuclear extracts were prepared (48) and EMSA performed employing 32 P-labeled probe for NF-B (49) as described (8 -9). Supershift assays were performed by preincubating nonimmune anti-p50, anti-p65, or both IgG (Santa Cruz Biotechnology) with nuclear extract for 45 min at room temperature prior to addition of radiolabeled oligonucleotide probe. Relative intensity of the bands was determined by densitometric analysis as above.
In Vivo Studies: Infusion of CML-modified and Native Proteins-BALB/c mice (Charles River), approximately 6 weeks of age, were injected intravenously via the tail vein with CML-modified native protein or lipopolysaccharide. Where indicated, animals were pretreated with the indicated IgG 24 h prior to and at the time of infusion of CML-modified proteins. In other cases, CML-modified protein was incubated with excess sRAGE for 1 h prior to infusion. Twelve h later, lungs were rapidly harvested. Lung tissue was homogenized in Trisbuffered saline containing protease inhibitor (Roche Molecular Biochemicals) and subjected to centrifugation at 8,000 rpm for 10 min. Supernatant was centrifuged for 1 h at 4°C at 40,000 rpm and the pellet dissolved in TBS containing protease inhibitors and octyl ␤-glucoside (2%) for 4 h at 4°C. The suspension was subjected to centrifugation for 10 min at 14,000 rpm and supernatant assessed for protein concentration (Bio-Rad). Immunoblotting was performed after electrophoresis of 50 g of protein/lane of SDS-PAGE gels and transfer of gel components to nitrocellulose. Anti-VCAM-1 IgG (0.4 g/ml) was employed for immunoblotting as above. In other cases, total RNA was isolated using TRIZOL (Life Technologies, Inc.) and the concentration of total RNA determined by absorption at A 260/280 . Total RNA (30 g/lane) was subjected to electrophoresis. After electrophoretic separation on a formaldehyde-agarose gel (0.8%), RNA was transferred onto a GeneScreen hybridization filter (NEN Life Science Products) and linked by a UV cross-linker (Stratagene, La Jolla, CA). The filter was then prehybridized in QuikHyb hybridization buffer (Stratagene) for 30 min, followed by hybridization with rat VCAM-1 cDNA probe (generously provided by Dr. Tucker Collins) or glyceraldehyde-3-phosphate dehydrogenase (GADPH) cDNA probe labeled with [␣-32 P]dCTP using the random primer labeling system (Stratagene) in the above hybridization buffer for 1 h. After washing twice at room temperature (15 min/wash) with SSC (2 times) containing SDS (0.1%) and then with SSC (0.1 time) containing SDS (0.1%) at 60°C for 30 min, the filter was subjected to autoradiography at Ϫ80°C. cDNA for glyceraldehyde-3-phosphate dehydrogenase was used in control studies to normalize counts in the VCAM-1 message by densitometry.
Statistical Analysis-Statistical comparisons were determined using one-way analysis of variance; where indicated, individual comparisons were performed using Student's t test. In all cases, statistical significance was ascribed to the data when p Ͻ 0.05.

CML Modifications Are Present to Enhanced Degrees in Human Serum Albumin in Diabetes and Renal Failure and Bind RAGE
As a first step in elucidating whether CML modifications occurred in typical proteins such as albumin, we isolated human serum albumin from age-matched subjects with diabetes, renal failure, or controls by chromatography of plasma on Affi-Gel blue. Albumin preparations, analyzed by SDS-PAGE and immunoblotting with affinity purified anti-CML IgG, displayed Ϸ2.0and Ϸ3.3-fold increased CML immunoreactivity of samples from diabetic and renal failure patients, respectively, compared with normal controls (Fig. 1A). These findings were consistent with earlier work indicating enhanced CML accumulation in the serum of subjects with diabetes or renal failure and suggested the relevance of albumin for modification by CML in our studies (24 -26).
Thus, CML-BSA was prepared and tested for its ability to bind RAGE. When immobilized onto the wells of plastic dishes, CML-BSA bound 125 I-sRAGE in a dose-dependent manner, with K d Ϸ76.2 Ϯ 35 nM (Fig. 1B). The affinity of CML-modified BSA for RAGE was quite similar to that obtained previously employing heterogeneous AGEs, 61 Ϯ 23 nM (1). That the observed binding was dependent on RAGE was illustrated by inhibition of 125 I-sRAGE interaction with CML-BSA in the presence of excess sRAGE, anti-CML IgG, or anti-RAGE IgG (Fig. 1C). In contrast, preincubation with preimmune rabbit IgG was without effect (Fig. 1C). A polypeptide backbone for CML modification was required for recognition by RAGE, as free CML did not effectively compete for binding of radiolabeled sRAGE to immobilized CML-BSA (Fig. 1C).
Further evidence for selectivity of binding was sought by identifying structural determinants within RAGE that mediated interaction with CML-BSA. Three distinct RAGE domainspecific fusion proteins were prepared with glutathione S-transferase (GST) by expression in E. coli followed by thrombin treatment to remove GST. RAGE domains were then purified to homogeneity. By amino-terminal amino acid sequencing and SDS-PAGE, purified V domain, C1 domain, and C2 domain (1-2) were obtained (data not shown). By using purified RAGE domains, competitive binding studies were performed with 125 I-sRAGE and immobilized CML-BSA; addition of a 50-fold molar excess of unlabeled V domain significantly suppressed binding, whereas C1 and C2 domains were without effect (Fig.  1D). These data implicated the V domain of RAGE in specific binding to CML-modified adducts.
Multiple studies have indicated that a dominant species in the heterogenous AGEs prepared upon incubation of protein with reducing sugar was CML (22). We thus prepared heterogenous AGE by incubation of human IgG with ribose, 0.025 M (a relevant concentration of reducing sugar in diabetes) for 6 weeks at 37°C under aerobic conditions. After retrieval of this material and extensive dialysis to remove unreacted ribose, AGE-IgG demonstrated significant immunoreactivity with affinity purified anti-CML IgG by ELISA (data not shown). Upon chromatography of AGE-IgG onto Affi-Gel 10 resin to which had previously been adsorbed human RAGE, material immunoreactive with anti-CML IgG (by ELISA) was eluted in the presence of glycine buffer. In contrast, equal volumes of glycine buffer control lacked immunoreactivity with anti-CML-IgG (data not shown). In radioligand binding assays, eluate from this Affi-Gel 10-sRAGE resin substantially suppressed binding of radiolabeled sRAGE to immobilized CML-BSA; in contrast, glycine buffer control was without effect (Fig. 2). These data thus provided further evidence for the specific interaction of CML-modified adducts with RAGE.
Taken together, these studies indicated that chemically synthesized CML-modified protein, or CML adducts that form Results of densitometric analysis are shown; densitometry units in immunoblotted albumin from normal subjects was arbitrarily defined as "1". A representative immunoblot is shown here; this experiment was performed on samples from 2 subjects per group with analogous results. B, CML-modified adducts bind RAGE. CML-BSA, 310 mmol of CML/mol of Lys, was immobilized onto the wells of plastic dishes (5 ng/well) and a radioligand binding assay performed with the indicated amount of 125 I-sRAGE alone or in the presence of a 50-fold molar excess of unlabeled sRAGE as described above. Data were analyzed by nonlinear least squares analysis and fit to a one-site model according to the equation of Klotz and Hunston: K d Ϸ76.2 Ϯ 35 nM and capacity 2.9 Ϯ 0.6 fmol/well. C and D, effect of anti-RAGE IgG, sRAGE (C) or RAGE-specific domains (D) on the binding of 125 I-sRAGE to immobilized CML-BSA. Radioligand binding assays were performed as above except that where indicated wells were preincubated with anti-RAGE IgG, anti-CML IgG, or preimmune IgG (70 g/ml), or radiolabeled sRAGE was preincubated with a 10-fold molar excess of sRAGE or free CML prior to binding assay employing 125 I-sRAGE, 100 nM. D, binding assays were performed in the presence of a 50-fold molar excess of RAGE-specific domain, V, C1, or C2 as described. C and D, percent maximal specific binding Ϯ S.E. of triplicate determinations is reported. These experiments were performed at least three times. upon incubation of protein with relevant concentrations of reducing sugar under aerobic conditions, bind RAGE in a specific manner. We next sought to extrapolate these findings to those previously observed in in vitro cell culture systems to determine if CML-modified adducts ligated RAGE and modulated cellular properties.

CML-modified Adducts Mediate Cellular Activation, in Vitro Analyses
Endothelial Cells-We previously demonstrated that heterogeneous AGEs, either those prepared in vitro or those isolated from human diabetic subjects, modulated endothelial function via engagement of RAGE (14). Since in vitro, CML-modified adducts bound RAGE, we sought to determine if these modifications were important components of heterogeneous AGEs in effecting cellular activation.
A central means by which endothelial function is modulated in the presence of experimental diabetes is by early enhanced expression of vascular cell adhesion molecule-1 (VCAM-1), a cell adhesion molecule that mediates binding of mononuclear cells bearing VLA-4 to the vessel wall (50). Incubation of HU-VEC with CML-modified ovalbumin (CML-OVA), 310 mmol of CML/mol of Lys, resulted in significantly increased cell surface expression of VCAM-1 compared with incubation with native OVA (Fig. 3A). The central role of RAGE in mediating interaction of CML adducts with endothelium was demonstrated by significant suppression of CML-mediated VCAM-1 expression in the presence of either anti-RAGE IgG or excess sRAGE (Fig.  3A). The functional significance of VCAM-1 expression by CML-treated endothelium was shown by enhanced binding of Molt-4 cells, which bear counterligand for VCAM-1, VLA-4 (Fig. 3B). Adherence of Molt-4 cells to stimulated endothelium was dependent on the concentration of CML-OVA (Fig. 3B) and required interaction of CML-modified adduct with the receptor, as shown by dose-dependent inhibition of Molt-4 binding in the presence of anti-RAGE F(abЈ) 2 (Fig. 3B). Similar inhibitory effects were observed in the presence of excess sRAGE (data not shown). In contrast, pretreatment of CML-stimulated endothelium with preimmune F(abЈ) 2 had no effect (Fig. 3B).
Expression of VCAM-1 is subject, at least in part, to regulation at the transcriptional level by nuclear factor-B (49). We previously demonstrated that ligation of RAGE by heterogeneous AGEs enhanced nuclear translocation of NF-B, as dem-onstrated by electrophoretic mobility shift assay (EMSA) (8,9). Compared with HUVEC incubated with native OVA, nuclear extracts from cells exposed to CML-OVA, 310 mmol of CML/ mol of Lys, demonstrated a Ϸ4-fold increase in activation of NF-B as measured by EMSA employing 32 P-labeled NF-B probe (Fig. 3C, lanes 2 and 1, respectively). That these effects of CML-OVA were largely mediated by interaction with RAGE was confirmed by the dose-dependent inhibitory effect of anti-RAGE IgG added to the endothelium prior to CML-OVA (Fig.  3C, lanes 6 and 5). Similar inhibitory effects were observed in the presence of excess sRAGE (data not shown). In contrast, treatment with nonimmune IgG was without effect on CML-OVA-mediated activation of NF-B (Fig. 3C, lane 4). CMLmediated induction of NF-B nuclear translocation required intact RAGE intracellular signaling, as shown by experiments using a truncated form of the receptor, termed tail-deletion RAGE, from which the cytosolic tail was deleted. Endothelium transfected with tail deletion RAGE display marked suppression of NF-B activation (Fig. 3C, lane 8) upon engagement of CML-OVA compared with cells transfected with vector alone (Fig. 3C, lane 7), a dominant negative (DN) effect.
Since previous studies demonstrated that the extent of CML modification in vivo is varied (51), it was important to show that a range of extents of protein modification by CML was capable of mediating endothelial activation. We thus prepared and characterized CML-OVA containing 33 mmol of CML/mol of Lys, and we tested its ability to activate NF-B. Incubation of HUVEC with this material resulted in Ϸ2.5-fold increase in nuclear translocation of NF-B by EMSA, compared with incubation in the presence of native OVA (Fig. 3D, lanes 1 and 2,  respectively). That this was dependent on ligation of RAGE was demonstrated by suppressed activation of NF-B by CML-OVA in the presence of anti-RAGE IgG but not by nonimmune IgG (Fig. 3D, lanes 3 and 4, respectively). Similar inhibitory effects were noted with excess sRAGE (data not shown).
Taken together, these data suggested that CML-modified proteins were specific AGEs that mediate cellular activation in endothelium, at least in part, via RAGE.
Vascular Smooth Muscle Cells-We next sought to test the hypothesis that CML-modified proteins interacted with RAGE on vascular smooth muscle cells (VSMC), as such cells have been implicated in vascular perturbation in diabetes and renal failure. In VSMC, activation of NF-B is a potent means by which regulation of cytokines and vasoregulatory mediators is affected (52)(53).
We first tested varied extents of CML modification in order to determine if their interaction with VSMC RAGE resulted in activation of NF-B. Incubation of VSMC with CML-BSA (90, 160, or 310 mmol CML/mol Lys) resulted in Ϸ3.2-, Ϸ2.6-, and Ϸ3.2-fold induction of nuclear translocation of NF-B (Fig. 4A, lanes 3-5, respectively) compared with incubation in the presence of native BSA (Fig. 4A, lane 2). In VSMC treated with CML-BSA, 90 mmol of CML/mol of Lys, activation of NF-B was dependent on interaction with RAGE, as demonstrated by suppression in the presence of either anti-RAGE IgG or excess sRAGE (Fig. 4B, lanes 5 and 6, respectively). In contrast, nonimmune IgG was without effect (Fig. 4B, lane 4). Furthermore, incubation of VSMC with CML-OVA, 33 mmol of CML/mol of Lys, resulted in Ϸ3.1-fold increase in activation of NF-B compared with cells incubated with native OVA (Fig. 4C, lanes 3  and 2, respectively). That these effects were mediated by RAGE was indicated by marked suppression of CML-OVA-mediated activation of NF-B in the presence of anti-RAGE IgG (Fig. 4C,  lane 5). Similar suppressive effects were noted in the presence of excess sRAGE (data not shown). In contrast, incubation with nonimmune IgG was without effect (Fig. 4C, lane 4). after washing, elution with glycine buffer revealed material immunoreactive with anti-CML-IgG by ELISA. Eluate and glycine buffer control were then tested for their ability to inhibit the binding of radiolabeled sRAGE (100 nM) to immobilized CML-BSA. Percent maximal specific binding Ϯ S.E. of triplicate determinations is reported. This experiment was performed twice with analogous results.
These data indicated that at low levels of CML modification, interaction of these modified adducts with RAGE on VSMC resulted in activation of NF-B, thereby providing a potent mechanism by which VSMC function may be altered in the presence of CML adducts of proteins.
Mononuclear Phagocytes-Previous studies from our laboratory indicated that MP properties are significantly modulated upon ligation of RAGE by heterogeneous AGEs, either those prepared in vitro or those derived from in vivo sources, such as AGEs isolated from diabetic plasma, or AGE-␤ 2 -microglobulin isolated and purified from the urine/plasma of subjects with DRA (16). It was thus important to determine if CML adducts, present in these preparations, could activate RAGE on MPs causing cellular migration and activation in this setting (54). Furthermore, since recent studies have indicated that CML modifications may result not only from glycation/oxidation of proteins, but also secondary to modification of lipids, a possible role CML-RAGE interaction in lipid-rich foam cells and atherosclerotic plaques, even in euglycemia, might occur (5).
We thus tested the ability of CML-modified adducts to activate monocytes. In modified chemotaxis chambers, compared with native OVA, addition of CML-OVA, 310 mmol of CML/mol of Lys, resulted in enhanced migration of Molt-4 cells, mononuclear type cells that bear cell surface RAGE (Fig. 5A, lines 1  and 2, respectively). The effects of CML-OVA were mediated by cellular RAGE as preincubation of Molt-4 cells with anti-RAGE F(abЈ) 2 resulted in suppression of migration in a dose-dependent manner (Fig. 5A, lines 4 and 5); preincubation with nonimmune F(abЈ) 2 was without effect (Fig. 5A, line 3). Similarly, preincubation of CML-OVA with excess sRAGE resulted in dose-dependent suppression of Molt-4 migration (Fig. 5A, lines  6 and 7). Intact RAGE signaling was required for these CMLmediated effects on migration as transient transfection of Molt-4 cells with a construct expressing tail deletion RAGE resulted in significant suppression of Molt-4 migration in response to CML-OVA; mock transfection was without effect (Fig.  5A, lines 9 and 8, respectively).
It was important to test lower extents of modification of CML-OVA in order to determine their ability to modulate Molt-4 migration. These studies revealed that addition of CML-OVA, 33 mmol/mol Lys, to chemotaxis chambers resulted in increased migration of Molt-4 cells compared with native OVA Varying concentrations of CML-OVA, 310 mmol of CML/mol of Lys, were employed or cells were preincubated with the indicated F(abЈ) 2 prior to CML-OVA. After incubation, 51 Cr-labeled Molt-4 cells were bound to the monolayer for 1 h, and cells were then disrupted in the presence of Triton X-100 (1%); the resulting material was counted in a beta counter. Results are reported as fold increase above control (treatment with native OVA). A and B, results are reported as mean Ϯ S.E. of at least triplicate determinations. These experiments were performed three times with analogous results. C and D, electrophoretic mobility shift assays. HUVEC were treated with the indicated mediators for 6 h. In certain cases, HUVEC were preincubated with anti-RAGE IgG or nonimmune IgG prior to treatment with CML-OVA, 310 mmol of CML/mol of Lys (C) or CML-OVA, 33 mmol of CML/mol of Lys (D). Where indicated, HUVEC were transiently transfected with constructs expressing either dominant negative RAGE (RAGE tail deletion) or vector alone (mock). Nuclear extract was prepared and EMSA performed employing radiolabeled probes for NF-B. Lanes designated excess unlabeled NF-B indicate that a 100-fold molar excess of unlabeled NF-B probe was added to incubation mixtures of nuclear extracts from CML-OVA-treated cells. Results of densitometric analysis after normalization to native OVA, arbitrarily defined as "1" in these experiments, are shown. EMSA were performed three times with analogous results. (Fig. 5B, lines 2 and 1, respectively). These effects were mediated by ligation of RAGE as demonstrated by suppression of migration in the presence of either anti-RAGE F(abЈ) 2 or excess sRAGE (Fig. 5B, lines 4 and 5, respectively). In contrast, incubation with nonimmune F(abЈ) 2 was without effect (Fig. 5B,  line 3).
Recent reports have indicated that in lens protein and skin collagen retrieved from elderly human subjects, extent of modification by CML adducts is in the approximate range of 4.95 and 1.70 mmol of CML/mol of Lys, respectively (51). The extent of CML modification is likely to be higher in atherosclerotic plaques and inflamed joints in DRA, for example, due to the higher concentration of highly bioactive species involved in CML formation compared with slow protein turnover observed in lens or skin collagen protein. We tested the ability of CML-OVA, 0.8 mmol of CML/mol of Lys, to engage RAGE on mononuclear cells and to redirect cellular function. Compared with native ovalbumin, addition of these minimally modified CML-OVA adducts to chemotaxis chambers caused a significant increase in cell migration (Fig. 5C, lines 1 and 2, respectively). These effects were mediated by RAGE, as indicated by suppression of CML-OVA-mediated migration in the presence of either anti-RAGE F(abЈ) 2 or excess sRAGE (Fig. 5C, lines 4 and 5, respectively).
Consistent with these findings, exposure of MP-like BV2 macrophages to CML-OVA resulted in Ϸ3.5-fold increase in nuclear translocation of NF-B by EMSA, compared with incubation in the presence of native OVA (Fig. 5D, lanes 1 and 2,  respectively). Activation of NF-B by CML adducts was due to ligation of RAGE as exemplified by significant suppression of NF-B in the presence of anti-RAGE IgG, 70 g/ml (Fig. 5D,  lane 6), but not in the presence of either lower concentrations of anti-RAGE IgG, 0.7 g/ml, or nonimmune IgG (Fig. 5D, lanes 5 and 4, respectively). Supershift assays with anti-p50 and anti-p65 IgG demonstrated that the NF-B complex activated upon ligation of BV-2 RAGE by CML-OVA was composed of both p50 and p65 (Fig. 5D, lanes 8 -10). Similar inhibitory effects were observed in the presence of transient transfection of construct encoding DN-RAGE (Fig. 5E, lane 2) or excess sRAGE (data not shown) but not by nonimmune IgG (Fig. 5E, line 3).

CML-modified Adducts Mediate Cellular Activation, in Vivo Analyses
The results of in vitro analyses suggested that CML adducts, at physiologically relevant levels of modification, were capable of mediating cellular activation in a range of cell types. A central test of this hypothesis was the ability of CML-modified adducts to modulate cellular properties in vivo. As a first test of these concepts, we infused CML-BSA, 310 mmol of CML/mol of Lys, into immunocompetent, non-diabetic mice. Initial studies indicated that in comparison to infusion of native BSA, CML-BSA induced a Ϸ46-fold increase in mRNA for VCAM-1 in lung tissue (Fig. 6A, lanes 2 and 1, respectively). To determine if this was associated with modulation of VCAM-1 protein levels in the lung, immunoblotting was performed. Indeed, infusion of CML-BSA into mice resulted in a Ϸ3.6-fold increase in expression of VCAM-1 protein compared with infusion of native BSA (Fig. 6B, lanes 2 and 3, respectively). That this was dependent on ligation of vascular RAGE in the lung by CML-BSA was demonstrated by experiments in which mice were pre-infused with either anti-RAGE IgG or excess sRAGE; in both cases, significant suppression of CML-mediated expression of VCAM-1 in the lung was noted (Fig. 6B, lanes 4 and 6, respectively). In contrast, infusion of nonimmune IgG had no effect on CML-mediated modulation of VCAM-1 expression (Fig. 6B,  3-5). B, VSMC were treated with CML-BSA, 90 mmol of CML/mol of Lys in the presence of anti-RAGE IgG, nonimmune IgG, or excess sRAGE as above. C, VSMC were treated with CML-OVA, 33 mmol of CML/mol of Lys, in the presence of anti-RAGE IgG or nonimmune IgG. In all cases, after incubation as indicated, nuclear extracts were prepared and EMSA performed employing radiolabeled probes for NF-B. Lanes designated excess unlabeled NF-B indicate that a 100-fold molar excess of unlabeled NF-B probe was added to incubation mixtures of nuclear extracts from CML-adduct-treated cells.
Results of densitometric analysis after normalization to native OVA or native BSA, arbitrarily defined as "1" in these experiments, are shown. EMSA were performed three times with analogous results. lane 5). These findings illustrate a mechanism by which engagement of vascular RAGE by CML-modified adducts enhances expression of VCAM-1 and, likely, increases adherence of proinflammatory mononuclear cells to the vessel wall.

DISCUSSION
Increased recognition of specific AGEs and the multiple means by which they might form in diverse settings such as hyperglycemia, renal failure, and inflammation has highlighted the importance of determining mechanisms by which each might modulate cellular properties, either by receptor-dependent or -independent pathways. In this context, previous studies indicated that heterogenous AGEs, especially those derived from in vivo sources, were signal-transducing ligands of RAGE. Specifically, interaction of AGE-␤ 2 -microglobulin, isolated from the urine of subjects with renal failure and DRA, with cellular RAGE resulted in enhanced MP migration and generation of proinflammatory cytokines such as tumor necrosis factor-␣ (16). Since recent studies have shown AGE-␤ 2 M itself to be quite heterogeneous, composed at least in part of AGEs such as pentosidine, CML, and imidizalone (55)(56)(57), it was important to begin to delineate those specific AGEs that interacted with RAGE.
In the present work, we have identified that CML adducts, dominant AGEs within the tissues, are signal-transducing ligands for RAGE in vitro, in endothelial cells, mononuclear phagocytes, and vascular smooth muscle, and in vivo, upon infusion into mice. Importantly, we tested varied extents of CML modification to demonstrate that the CML adducts thus formed exerted specific and receptor-dependent effects. In this context, physiologically relevant levels of CML modification of ovalbumin were capable of modulating MP function. Such levels of modification parallel those reported in relatively inert tissue derived from elderly human subjects, such as skin collagen and lens protein (51). We speculate that in a highly biologically active milieu, such as lipid-laden atherosclerotic fatty streaks and plaques and chronically inflamed joints, levels of CML modification will likely exceed those tested here.
In proper context, however, levels of CML-modified adducts are low, even in tissue from subjects with diabetes or renal failure (51). We hypothesize that the inexorable accumulation of AGEs, via their interaction with RAGE, causes a change in cellular homeostasis priming mechanisms capable of triggering full-blown cellular activation. Such chronic AGE engagement of RAGE causes a shift in the basal state, favoring cellular activation with subsequent environmental challenge. Thus, AGE-RAGE interaction may be considered the first hit in a "two-hit" . In all cases, after incubation as indicated, nuclear extracts were prepared and EMSA performed employing radiolabeled probes for NF-B. Lanes designated excess unlabeled NF-B indicate that a 100-fold molar excess of unlabeled NF-B probe was added to incubation mixtures of nuclear extracts from CML -adduct-treated cells. Results of densitometric analysis are reported after normalization to native OVA-(D) or CML-OVA (E)-treated BV-2 cells, arbitrarily defined as "1" in these experiments. EMSA were performed three times with analogous results. model of diabetic complications. In wound repair, the second hit comprises the presence of foreign bodies and bacterial pathogens. In atherosclerotic vasculature, the second hit is an environment enriched in modified lipoproteins and macrophages. Juxtaposition of these two mechanisms results in an exaggerated cellular response, ultimately compromising reparative processes in diabetic tissues. Thus, as opposed to septic shock, in which highly potent lipopolysaccharide released by Gramnegative bacteria engages cellular sites, resulting in striking rapid activation and destabilization of homeostasis, in settings such as diabetes, renal failure, and chronic arthritis, long term, yet unrelenting, chronic cellular activation ensues, eventuating in irreversible tissue injury.
An important means by which AGEs modulate cellular properties is via generation of enhanced oxidant stress, manifested by increased transcripts for heme oxygenase and nuclear translocation of NF-B (8,9), the latter a central cell signaling molecule whose activation is linked to transcriptional regulation of a range of proinflammatory genes, including RAGE itself (58).
We speculate that these events, significantly suppressed in the presence of antioxidants (8,9) and mediated at least in part via RAGE, are early and highly influential steps in processes which, long term, contribute to cellular dysfunction. Consistent with these concepts, in experimental diabetes, AGEs begin to form within weeks of hyperglycemia (Refs. 19, 20, and 59). Thus, although evidence does not exist for globally enhanced oxidant stress in diabetes, as measured by lack of increased ortho-tyrosine and methionine sulfoxide in diabetic collagen (60), we speculate that in tissues such as the vessel wall and periodontium, sustained exposure to damaging substrates typified by high levels of modified lipid and Gram-negative bacteria, respectively, may yield CML adducts and thus mediate chronic local oxidant stress via RAGE. These local changes critically modulate cellular properties. Consistent with this view, therefore, lack of increased CML adducts in diabetic urine (61) may be insensitive in detecting locally relevant enhanced accumulation and oxidant stress. Indeed, support for the concept that chronic generation of CML-modified adducts at locally relevant sites in diabetic organisms contributes, at least in part, to mechanisms underlying complications, may be deduced from the findings that the extent of diabetic complications, such as those in the vasculature and retina, correlate with the degree of CML accumulation (62)(63).
The findings presented here do not rule out other specific AGE products of glycation or oxidation, such as pentosidine, pyralline, methylglyoxal, and imidizalone (64 -66) as signaltransducing ligands of RAGE. Indeed, although certainly present in the tissues at even lower degrees than CML, it is possible that locally relevant concentrations of these modified adducts may engage cellular RAGE to redirect cellular properties. Similarly, our findings do not rule out the presence of other receptors or cellular interaction sites for CML adducts. Although our studies suggest that RAGE is a receptor for CML, it is possible that other receptors for AGE (67-69) may also engage CMLmodified adducts. Indeed, it will be of particular interest to determine if the macrophage scavenger receptor plays a role in removal/detoxification of CML adducts, as recent views have emerged suggesting that overtaxed detoxification pathways lead to accelerated accumulation of toxic AGEs and highly reactive carbonyl compounds. Ultimately, it is speculated that these products exacerbate tissue injury, especially in the later stages of diabetes or renal failure (70 -71). Alternatively, different cellular receptors for AGEs may recognize diverse AGEs. For example, the macrophage scavenger receptor has been reported to bind glycoaldehyde (72), which does not bind RAGE.
Certainly, in complex disorders such as renal failure and diabetes, numerous factors likely converge in the development of cellular abnormalities. In diabetes, hyperglycemia directly results in activation of protein kinase C, especially the ␤II isoform, leading to cellular dysfunction (73)(74). In addition, accelerated generation and accumulation of products of the sorbitol/polyol pathway are implicated in the pathogenesis of diabetic complications, especially those of the retina and peripheral nerve (75). Recent observations suggesting that polyol metabolites may lead to generation of AGEs (76) suggest that, not unexpectedly, seemingly diverse pathways may converge and thus collaboratively contribute to cellular perturbation. These findings indeed parallel the diverse means by which FIG. 6. Infusion of CML adducts into mice increases mRNA and protein expression of VCAM-1 in lung tissue. A, Northern analysis. BALB-C mice were injected by tail vein with CML-BSA, 310 mmol of CML/mol of Lys, native BSA, or lipopolysaccharide. Twelve h later, lungs were harvested, RNA prepared, and Northern blotting for VCAM-1 and GADPH (for normalization of amount RNA loaded) performed. B, Western blotting. BALB-C mice were treated with the indicated mediators by tail vein infusion. In certain cases, pretreatment was performed with anti-RAGE IgG, nonimmune IgG, or excess sRAGE (20-fold molar excess) prior to infusion of CML-BSA. Twelve h later, lung tissue was retrieved, extract prepared as above, and SDS-PAGE/immunoblotting performed employing anti-VCAM-1 IgG as described. A and B, results of densitometric analysis are reported after normalization to native BSA, arbitrarily defined as "1" in these experiments. These experiments were performed twice with analogous results. CML adducts may form: from glycation/oxidation of proteins to lipid oxidation and myeloperoxidase-driven modification of amino acids and related structures.
Taken together, our observation that physiologically relevant extents of CML modification in proteins are signal-transducing ligands for RAGE transforms these adducts from inert biomarkers of aging and chronic disease to bioactive species capable of altering properties of cells critically perturbed in inflammatory milieu. Our findings provide a contributory mechanism for the chronic, yet unrelenting progressive vascular and inflammatory cell dysfunction that typifies the complications of diverse disorders, such as diabetes, renal failure, chronic inflammation and Alzheimer's disease, in which CML adducts accumulate and the expression of RAGE is enhanced. Thus, identification of specific molecular structures to inhibit interaction of CML-modified adducts with RAGE may provide a potent means to suppress cellular activation and, thereby, limit long term tissue injury in chronic diseases.