modulation of soluble receptor for advanced glycation end products by angiotensin-converting...

Modulation of Soluble Receptor for Advanced Glycation EndProducts by Angiotensin-Converting Enzyme-1 Inhibition inDiabetic Nephropathy

Josephine M. Forbes,* Suzanne R. Thorpe,‡ Vicki Thallas-Bonke,* Josefa Pete,*Merlin C. Thomas,* Elizabeth R. Deemer,‡ Sahar Bassal,† Assam El-Osta,† David M. Long,*Sianna Panagiotopoulos,§ George Jerums,§ Tanya M. Osicka,§ and Mark E. Cooper**Danielle Alberti Memorial Centre for Diabetes Complications, Wynn Domain, and †Epigenetics Laboratory, BakerHeart Research Institute, Melbourne, Australia; ‡Department of Chemistry and Biochemistry, University of SouthCarolina, Columbia, South Carolina; and §Department of Endocrinology, Austin and Repatriation Medical Centre,Heidelberg, Australia

Recent studies have identified that first-line renoprotective agents that interrupt the renin-angiotensin system not only reduceBP but also can attenuate advanced glycation end product (AGE) accumulation. This study used in vitro, preclinical, andhuman approaches to explore the potential effects of these agents on the modulation of the receptor for AGE (RAGE). Bovineaortic endothelial cells that were exposed to the angiotensin-converting enzyme inhibitor (ACEi) ramiprilat in the presence ofhigh glucose demonstrated a significant increase in soluble RAGE (sRAGE) secreted into the medium. In streptozotocin-induced diabetic rats, ramipril treatment (ACEi) at 3 mg/L for 24 wk reduced the accumulation of skin collagen-linkedcarboxymethyllysine and pentosidine, as well as circulating and renal AGE. Renal gene upregulation of total RAGE (all threesplice variants) was observed in ACEi-treated animals. There was a specific increase in the gene expression of the splicevariant C-truncated RAGE (sRAGE). There were also increases in sRAGE protein identified within renal cells with ACEitreatment, which showed AGE-binding ability. This was associated with decreases in renal full-length RAGE protein fromACEi-treated rats. Decreases in plasma soluble RAGE that were significantly increased by ACEi treatment were also identifiedin diabetic rats. Similarly, there was a significant increase in plasma sRAGE in patients who had type 1 diabetes and weretreated with the ACEi perindopril. Complexes between sRAGE and carboxymethyllysine were identified in human and rodentdiabetic plasma. It is postulated that ACE inhibition reduces the accumulation of AGE in diabetes partly by increasing theproduction and secretion of sRAGE into plasma.

J Am Soc Nephrol 16: 2363–2372, 2005. doi: 10.1681/ASN.2005010062

T reatments that involve blockade of the renin-angioten-sin system (RAS) have revolutionized strategies to op-timize renoprotection in patients who are at risk for or

have diabetic nephropathy (1–3). These benefits have been at-tributed not only to interruption of the hemodynamic but morerecently also to the nonhemodynamic effects of angiotensin II(3–5).

Advanced glycation end products (AGE) have been impli-cated as major contributors to the pathogenesis of diabeticnephropathy, with therapies that inhibit their formation prov-ing beneficial in experimental diabetic nephropathy (6–8). Inthis context, the recent finding that angiotensin-converting en-zyme inhibitors (ACEi) are potent inhibitors of the formation ofAGE provides a potential link between hemodynamic and met-

abolic factors in mediating diabetic complications, includingnephropathy (9,10). The mechanisms by which these reductionsin AGE occur in response to interruption of the RAS, however,remain to be fully elucidated.

The receptor for AGE (RAGE) is a member of the Ig superfamily (11,12) and plays a central role in the development ofexperimental diabetic complications (13,14). Particularly, ad-ministration of the exogenous soluble form of RAGE (sRAGE)has shown benefits in diabetes-associated atherosclerosis (15)and nephropathy (13,16). Recent studies have identified novelgene splice variants of the human RAGE receptor, confirmingthe liberation of three distinct forms: Full-length RAGE, whichhas full signaling and AGE-binding potential; N-truncatedRAGE, a membrane-bound isoform that contains no AGE-bind-ing domain; and C-truncated (soluble) RAGE, which has AGE-binding properties in the absence of a signaling cascade (17).

The general aim of this study was to identify whether ACEinhibition modulates expression of the splice variants of RAGE,thereby partly providing an explanation for the effects of theseagents on AGE. This was investigated in a series of studies thatincluded a cell culture model and a long-term model of exper-

Received January 17, 2005. Accepted April 26, 2005.

Published online ahead of print. Publication date available at www.jasn.org.

Address correspondence to: Dr. Josephine Forbes, Danielle Alberti MemorialCentre for Diabetes Complications, Baker Heart Research Institute, P.O. Box 6492,St. Kilda Road, Melbourne, Victoria, 8008, Australia. Phone: 61-3-8532-1456; Fax:61-3-8532-1288; E-mail: [email protected]

Copyright © 2005 by the American Society of Nephrology ISSN: 1046-6673/1608-2363

imental diabetic nephropathy and in samples from patientswho had type 1 diabetes and participated in a clinical trial thatexplored early renoprotection with the ACEi perindopril (18).

Materials and MethodsCell Culture Experiments

Bovine aortic endothelial cells (BAEC) (19) were cultured in modifiedEagle’s medium (MEM), supplemented with nonessential amino acids,gentamicin, 10% FBS, and a low-glucose environment (1.1 mmol/Lglucose) until confluence. BAEC then were randomized to either low orhigh glucose (30 mmol/L) at 0.5% FBS in the presence and absence ofthe active metabolite of the ACEi ramipril, ramiprilat (150 �mol/L;Aventis, Frankfurt, Germany); the AT1R antagonist valsartan (10�mol/L); or the mitochondrial reactive oxygen species scavenger ide-benone (10 �mol/L) for 7 d. On day 7, the cells then were lysed withice-cold RIPA buffer (150 mmol/L NaCl, 1 mmol/L EDTA, 50 mMTris-Cl [pH 7.5], 1% NP-40, and 0.25% deoxycholic acid) on ice. TheRIPA buffer also contained the phosphatase inhibitors 1 mmol/LNa2VO3, 1 mmol/L NaF, and 30 mmol/L NaPP and a complete pro-tease inhibitor cocktail (Roche Molecular Biochemicals, Mannheim,Germany). The protein concentrations of the lysates and cell culturesupernatants (CCSN) were measured using the BCA protein assay(Pierce Biotechnology, Rockford, IL). Three independent experimentswere carried out with essentially similar results.

Experimental Rodent ModelDiabetes was induced in male Sprague Dawley rats (200 to 250 g) by

streptozotocin (intravenously, 50 mg/kg) in sodium citrate buffer (pH 4.5)after an overnight fast (20,21). Animals with plasma glucose concentra-tions �15 mmol/L 1 wk after induction of diabetes were included in thestudy. Sham-injected control animals (buffer only) were followed concur-rently. Diabetic and control animals were randomized into groups (n � 10)that received either no treatment (D and C) or ramipril at a dose of 3 mg/Lin drinking water (DACEi and CACEi) and followed for 24 wk. This dosewas based on previous in vivo studies showing inhibition of the RAS asassessed by an increase in plasma renin activity as well as a reduction inAGE accumulation (10). Two units of Ultralente insulin (Ultratard HM;Novo Industries, Bagsvaerd, Denmark) were administered daily to dia-betic animals to prevent ketoacidosis and to improve survival. Bodyweight, mean systolic BP by tail-cuff plethysmography, GFR using 99Tc-DTPA, albumin excretion rate (AER) by RIA, and glycated hemoglobin(HbA1c) were measured every 8 wk as described previously (22). Allanimal procedures were in accordance with guidelines set by the AustinHospital Animal Ethics Committee and the National Health and MedicalResearch Council of Australia.

Human Plasma SamplesMelbourne Diabetic Nephropathy Study Group. A subset of pa-

tients in the Melbourne Diabetic Nephropathy Study Group study ofpatients with type 1 diabetes (18) were randomized to receive placebotreatment (n � 11), the ACEi perindopril (Servier, Neuilly, France; n �

11), or the dihydropyridine calcium channel blocker nifedipine (Bayer,Wuppertal, Germany; n � 11) for 24 mo. The placebo was a single tabletin the image of perindopril. The initial dose of perindopril was 2 mgadministered upon waking and then titrated up to 4 mg at 2-wkintervals until lowering of 5 mmHg supine diastolic BP or greater wasseen. The initial dose of nifedipine was 10 mg twice daily titratedfortnightly to achieve lowering of 5 mmHg in supine diastolic BP.Plasma samples were obtained at the initiation of the study and at 24mo. The criteria for inclusion were patients who were aged 16 to 65 yrand had a duration of type 1 diabetes for �5 yr. All patients had

microalbuminuria, diagnosed on three separate occasions (AER 20 to200 �g/min) and supine BP of �160/90 mmHg if older than 40 yr ofage or 140/90 if younger than 40 yr (18). The exclusion criteria werenondiabetic renal disease, evidence of poor diabetic control (HbA1c

�10%), cardiac failure, and systemic disease (18). All human proce-dures were in accordance with guidelines set by the Austin HospitalHuman Ethics Committee and the National Health and Medical Re-search Council of Australia.

Plasma Low Molecular Weight AGE AnalysisLow molecular weight (LMW) AGE fluorescence was assayed in dupli-

cate 20-�l plasma aliquots as described previously (22). Briefly, sampleswere deproteinized by addition of 0.15 M TCA and then delipidated usingchloroform (23), followed by centrifugation and removal of the upperaqueous phase. Fluorescence of the supernatant was determined (Ex 370nm, Em 440 nm) using an on-line HPLC injector (Waters, Milford, MA)and expressed per arbitrary protein unit at 280 nm.

Isolation of Skin Collagen and Analysis of AGE/ALERat skin collagen was prepared as described previously (7). In brief,

insoluble collagen was isolated from 1.5-cm2 pieces of skin after re-moval of adventitious tissue with a razor blade and subsequent sequen-tial extractions with 1.0 mol/L NaCl, 0.5 M acetic acid, and delipidationwith chloroform:methanol (1:2). The collagen then was lyophilized andstored at �20°C until analysis of AGE/advanced lipoxidation endproduct (ALE) content.

The AGE/ALE N�(carboxymethyl)lysine (CML) and N�(carboxyethyl)-lysine were quantified by isotope dilution, selected ion monitoring gas chro-matography–mass spectrometry (24), and normalized to their parent aminoacid lysine. Pentosidine was analyzed by reverse phase–HPLC and was alsonormalized to lysine content (24).

Renal Cortical FluorescenceRenal cortical tissue samples (100 mg wet weight) were acid hydro-

lyzed (6 M HCL for 24 h at 110°C) using the procedure of Stegemannand Stalder (25). After clarification with DEAE resin and neutralization,AGE fluorescence was determined as described previously (22) via aflow injection system adapted from Wrobel et al. (23) with a Waters 470spectrophotometer (Ex 370 nm/Em 440 nm; Waters). AGE peptidefluorescence was expressed per protein density unit at 280 nm asdescribed previously (22).

Immunohistochemistry for Renal CMLA modification of the ABC Ig enzyme bridge technique (26) was used

for immunohistochemistry as described previously (22). The monoclonalAGE antibody 4G9, which was used in this protocol (1:500; Alteon Inc.,Ramsey, NJ), recognizes CML (27). Negative control sections had theomission of the primary antibody. Positive control tissues were also in-cluded. Quantification of renal cortical immunostaining was completed bycomputer-aided densitometry (MCID-Video Pro-32; Bedford Park, SA,Australia), whereby a total of 20 fields (�100) were counted per sectioncorresponding to a total kidney area of 6.08 mm2. Ten animals per groupwere counted. Results were expressed as proportional area of positivestaining (28).

Reverse Transcription–PCRThree micrograms of total RNA extracted from each kidney cortex

was used to synthesize cDNA with the Superscript first-strand synthe-sis system for reverse transcription–PCR (RT-PCR; Life TechnologiesBRL, Grand Island, NY). Gene expression for each of the sequencesidentified in Table 1 was analyzed by real-time quantitative RT-PCR

2364 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 2363–2372, 2005

performed with the TaqMan system based on real-time detection ofaccumulated fluorescence (ABI Prism 7700; Perkin-Elmer, Foster City,CA) as described previously (29). The amplification was performedwith the following time course: 2 min at 50°C and 10 min at 95°C and40 cycles of 20 s at 94°C and 1 min at 60°C. Each sample was tested intriplicate; the average interassay coefficient of variation was 2.1%.Results were expressed relative to control kidneys, which were arbi-trarily assigned a value of 1.

ImmunoprecipitationSamples of renal kidney cortex (100 mg wet weight) were microdis-

sected and homogenized in 1 ml of cold neutral salt buffer (50 mmol/LTris-Cl [pH 7.4], 150 mmol/L NaCl, and 5 mmol/L EDTA) that con-tained the protease inhibitors PMSF (1 mmol/L), leupeptin (10 �g/ml),and aprotinin (1 �g/ml) and then centrifuged at 3000 rpm for 1 h at4°C. Protein concentrations in supernatants were determined by theBCA protein assay (Pierce Biotechnology). Tissue pools then weremade for each of the four groups (C, CACEi, D, and DACEi; n � 6samples per pool).

A total of 500 �g of tissue supernatant was added to 2 �g of C-terminal RAGE antibody (which recognizes the full-length and N-RAGE splice variants; gift of Dr. E. Boel, Novo Nordisk), N-terminalRAGE antibody (which recognizes the full-length RAGE and sRAGEsplice variants; gift of Dr. E. Boel, Novo Nordisk), or total RAGE (whichrecognizes all three splice variants; gift of Dr. Mike Neeper) diluted to1 ml in binding buffer (50 mmol/L phosphate buffer [pH 7.2]) androlled for 2 h at 4°C. Protein G slurry (50%) then was added, and thesamples were rotated overnight. Samples then were centrifuged at 3000rpm for 3 min, the supernatant was removed, and the gel slurry pelletswere washed at least six times in wash buffer (50 mmol/L Tris-Cl [pH7.4], 500 mmol/L NaCl, and 0.1% Tween 20). After the final wash, thegel slurry was resuspended in Laemmli S20 sample buffer (30), and�-mercaptoethanol was added to a final concentration of 0.1 mol/L.Suspensions were denatured at 95°C for 10 min before being subjectedto SDS-PAGE (see below).

Western BlottingFor Western blotting, 5 �g of plasma protein or 50 �g of cell lysate

protein was incubated at 65°C in S20 sample buffer that contained 60mmol/L dithiothreitol for 10 min. CCSN were concentrated �10 withAmicon Microcon Filters (10000 MWCO; Millipore Corp., Bedford,MA). These samples or 10 �l of immunoprecipitated sample preparedpreviously was separated on a 10% SDS–polyacrylamide electrophore-sis gel (30). With the use of a semidry transfer tank (Bio-Rad Labora-tories, Hercules, CA), proteins were transferred to a polyvinylidenedifluoride membrane (Amersham Biosciences, Buckinghamshire, UK).Nonspecific binding sites were blocked for 1 h with 5% (wt/vol) nonfat

milk powder in Tris-buffered saline (pH 7.4) and 0.05% Tween 20(BLOTTO) followed by overnight incubations in primary antibody tototal RAGE (goat anti-human RAGE; 1:1000, which recognizes all threesplice variants). Bound antibodies were visualized by sequential incu-bation with biotinylated secondary antibody (1:15,000; DAKO, Carpin-teria, CA), horseradish peroxidase–conjugated streptavidin (1:15,000;DAKO), and then autoradiography using an enhanced chemilumines-cence kit (Pierce Biotechnology). Total band density from which load-ing control had been corrected was quantified using an imaging devicewith Optimas 6.2 Software (Optimas 6.2, Video Pro-32) associated witha JVC video camera and Olympus microscope. Loading controls forplasma samples was anti-albumin antibody (1:10,000; Sigma ChemicalCo, St. Louis, MO). Results were expressed relative to control animals,which were arbitrarily assigned a value of 1.

AGE/RAGE ComplexesPooled plasma samples (400 �l; n � 6) from untreated diabetic rats

and patients with type 1 diabetes were concentrated �10 with AmiconMicrocon Filters (10000 MWCO; Millipore Corp.). Immunoprecipita-tion was performed as outlined above using either the N-terminalRAGE antibody (which recognizes the full-length RAGE and sRAGEsplice variants; gift of Dr. E. Boel, Novo Nordisk) or rabbit anti-CML(31). The subsequent precipitated complexes then were separated bySDS-PAGE and determined by Western blotting as above. CML-asso-ciated precipitates were probed for RAGE with a goat anti-humanRAGE (1:5000), which recognizes all three splice variants. ExtracellularRAGE–associated immunoprecipitates were probed for the presence ofCML, with rabbit anti-CML antibody (1:400,000).

Far Western ImmunoblottingThe AGE-binding capacity of protein that was immunoprecipitated

with RAGE antibodies was determined. Immunoprecipitates were sub-jected to SDS-PAGE and transferred onto polyvinylidene difluoridemembrane as above. Filters were blocked overnight in 1% gelatin inTBST (BLOTTO, with the omission of skim milk) and then incubated in0.1 mg/ml AGE-modified BSA (20) for 2 h. After extensive washing inTBST, affinity-purified rabbit anti–AGE-BSA (1:2000) (31) was appliedfollowed by detection with a biotinylated secondary antibody(1:15,000), horseradish peroxidase–conjugated streptavidin, and chemi-luminescence as per Western immunoblotting protocol above.

Statistical AnalysesResults are expressed as mean � SD unless otherwise specified. Data

for albuminuria were not normally distributed and therefore wereanalyzed following logarithmic transformation and are expressed asgeometric mean �/� tolerance factors. Analyses were performed by

Table 1. Real time RT-PCR probe and primer designsa

Probe Sequence5�FAM-3�TAMRA

Forward Primer5�-3�

Reverse Primer5�-3�

ProbeStart

Position(bp)

GeneBankAccessionNumber

Total RAGE TGTGCCATCTCTGC TCCTGGTGGGACCGTGAC GGGTGTGCCATCTTTTATCCA 822 L33413INT-RAGE ATCCCAATTCAACCTC AGACGAGCTCCCCCACTCTAC GGGCTCTGGTTGGAGAAGAAA 1341 L33413ABS-RAGE TGGTCAGAACATCACAGC CTGGGTACTGGTTCTTGCTCTGT CTAGCTTCTGGGTTGGCTTCTTAG 118 L33413

aReal-time reverse transcription–PCR probes and primers designed for the Taqman system by Primer Express for receptorfor advanced glycation end products (RAGE) splice variants. INT-RAGE, probes and primers to the intracellular portion ofRAGE; ABS-RAGE, probes and primers to the extracellular active binding site of RAGE; Total RAGE, probes and primers tothe center of RAGE that recognize all three splice variants.

J Am Soc Nephrol 16: 2363–2372, 2005 ACE Inhibition Increases sRAGE 2365

ANOVA followed by post hoc analysis using Tukey least significantdifference method, correcting for multiple comparisons. P � 0.05 wasconsidered to be statistically significant.

ResultsRamiprilat Increases Secretion of sRAGE in CulturedEndothelial Cells

Western immunoblotting on BAEC lysates revealed a signif-icant decrease in RAGE band density for groups that were

treated with ramiprilat, with two bands resolved by SDS-PAGEat approximately 55 and 40 kD (Figure 1A). Only the bandidentified at 55 kD had AGE-binding affinity (FL-RAGE) asidentified by Far Western immunoblotting, suggesting that thelower molecular weight band was N-RAGE (Figure 1B). Inaddition, there was a significant increase in C-truncated RAGE(sRAGE) at approximately 50 kD identified in the CCSN ofBAEC that were treated with ramiprilat as compared with

Figure 1. Bovine aortic endothelial cells (BAEC) were exposed to low- and high-glucose environments in 0.5% FBS, in the presence andabsence of ramiprilat. (A) Western immunoblotting of cell lysates with an antibody that recognizes all three splice variants of thereceptor for advanced glycation end products (RAGE). (B) Far Western immunoblotting of cell lysates to assess the AGE-bindingcapacity of bands that were detected with RAGE antibody. (C) Western immunoblotting of cell culture supernatants (CCSN) fromBAEC cultures with an antibody that recognizes all three splice variants of RAGE. (D) Quantification of soluble RAGE (sRAGE) inBAEC culture supernatants, representative of three separate experiments, arbitrary units per mg protein. (E) Far Western immuno-blotting of CCSN from BAEC cultures to assess the AGE-binding capacity of bands that were detected with RAGE antibody. FL-RAGE,full-length RAGE; Ideb, idebenone; AT1a, valsartan. *P � 0.001 versus low glucose; †P � 0.001 versus high glucose; #P � 0.01 low-glucoseangiotensin-converting enzyme inhibitor (ACEi) versus high-glucose ACEi; §P � 0.005 versus high-glucose ACEi.


untreated cells. This band also had AGE-binding capacity andwas further increased upon exposure to high-glucose condi-tions (Figure 1C). This band was also identified in CCSN ofcells that were treated with valsartan or idebenone under high-glucose conditions but only appeared significantly increased byidebenone as compared with the high-glucose control (Figure1D). No other bands were identified on these membranes.These findings were confirmed in three separate experiments.

Experimental Diabetic NephropathyFunctional and Biochemical Parameters. Diabetes was as-

sociated with increases in plasma glucose and HbA1c levels (Table2). There was a significant decrease in body weight in diabeticanimals, which was attenuated modestly by ACE inhibition. Nochanges in diabetes-induced increases in kidney to body weightratios were seen with ACEi treatment. Untreated diabetic animalshad increased systolic BP (SBP) and AER (Table 2). ACE inhibitionreduced both SBP and AER to control levels. No increase inplasma cholesterol was seen at any time (Table 2).

Tissue AGE Concentrations Are Increased with Experi-mental Diabetes and Reduced by ACEi. Renal cortical CMLlevels were significantly increased (Figure 2A) in associationwith increased renal fluorescence (Figure 2B) in diabetic ascompared with control animals. Treatment with ramipril pre-vented the diabetes-induced increases in renal CML and alsoattenuated the increase in renal fluorescence.

Both CML (Figure 2C) and pentosidine (Figure 2D) were in-creased in skin collagen of diabetic animals, whereas levels ofN�(carboxyethyl)lysine were unchanged (data not shown). ACEinhibition attenuated the increase in skin collagen CML withoutaffecting skin collagen pentosidine levels (Figure 2, C and D).

ACEi Can Modulate the Expression of Renal RAGE Splice Vari-ants. Real-time RT-PCR analysis of renal cortical gene expressiondemonstrated significant increases in the total expression of RAGE(Figure 3A) in both the untreated and the ACEi-treated rats.

Gene analysis for INT-RAGE (which corresponds to the C-terminus of RAGE, recognizing both full-length RAGE andN-truncated RAGE) showed increases with diabetes, which

were prevented in the DACEi group (Figure 3B). No significantdifferences were detected between control groups.

Increases in ABS-RAGE (which recognizes both full-lengthRAGE and C-truncated/sRAGE at the N-terminus) were ob-served with diabetes in both untreated and ACEi-treated ani-mals (Figure 3C). In addition, the control group that wastreated with ACEi had significant increases in the expression ofABS-RAGE (Figure 3C).

Renal RAGE Protein Isoform Expression Is Altered byACEi. Immunoprecipitation of renal cortical pools with anantibody that recognizes the C-terminal intracellular domain ofRAGE yielded two major bands on SDS-PAGE (Figure 4A). Thefirst band (approximately 58 kD) was representative of full-length RAGE as it had AGE-binding capacity as assessed by FarWestern immunoblotting (Figure 4B). The second band at ap-proximately 35 kD had no AGE-binding capacity and thereforewas recognized as N-RAGE.

We used an antibody raised against the N-terminal extracellu-lar domain of RAGE to immunoprecipitate full-length RAGE andC-truncated/sRAGE from renal cortical homogenates. This pro-duced two major bands by SDS-PAGE, one at approximately 58kD (Figure 4C) and a second, more intense band at approximately40 kD. Each of these bands had AGE-binding capacity (Figure 4D).

Increases in Circulating AGE Levels Are Accompanied byDecreases in Circulating sRAGE

Experimental Diabetes. Experimental diabetes induced asignificant increase in plasma concentrations of LMW-AGEwhen compared with levels in control animals (Figure 5A). Thisincrease in LMW-AGE was attenuated by ACE inhibition. In-creases in LMW-AGE were accompanied by decreases in thelevels of sRAGE (Figure 5B) in the plasma of diabetic animals.

Patients with Type 1 Diabetes. As has been describedpreviously (18), perindopril provides superior protection fromdeclining renal disease as compared with nifedipine for equiv-alent BP lowering. Plasma taken from patients with type 1diabetes in the placebo group had significantly higher LMW-AGE concentrations at the completion of the study period than

Table 2. Physiologic parameters for STZ rats at 24 wka

Control(n � 10)

Control ACEi(n � 10)

Diabetes(n � 10)

Diabetes ACEi(n � 10)

Plasma glucose (mmol/L) 6.9 � 0.3 6.9 � 0.2 27.3 � 2.5b 27.2 � 1.0b

Glycated hemoglobin (%) 3.1 � 0.1 3.2 � 0.1 14.3 � 0.8b 14.6 � 0.6b

Mean SBP (mmHg) 127 � 3 123 � 1 147 � 4b 123 � 2c

AER (mg/24 h) 3.2 �/� 0.8 2.0 �/� 0.6 18.7 �/� 10.0b 2.2 �/� 0.5c

Body weight (g) 743 � 19 700 � 61 312 � 21b 381 � 70b,d

KW:BW ratio (�10�3) 5.8 � 0.5 5.8 � 0.4 13.5 � 3.8b 12.9 � 2.6b

Total plasma cholesterol (mmol/L) 2.1 � 0.1 1.8 � 0.1 1.7 � 0.1 1.9 � 0.1aPhysiologic parameters for streptozotocin-induced diabetic rats at 24 wk. ACEi, angiotensin-converting enzyme inhibitors;

SBP, systolic BP; AER, albumin excretion rate; KW, kidney weight; BW, body weight. Data are shown as mean � SEM exceptfor AER, which are shown as geometric mean �/� tolerance factors.

bP � 0.001 versus control.cP � 0.001 versus diabetic.dP � 0.05 versus diabetic.


did patients who were treated with the ACEi perindopril (Fig-ure 6A). By contrast, plasma sRAGE was increased in patientswho received perindopril (Figure 6A). The patients who re-ceived placebo or nifedipine had no increase in plasma sRAGE;in fact, this was significantly decreased in these groups. Therewas an inverse correlation (R2 � 0.52, P � 0.047; Figure 6B)between increasing plasma LMW-AGE and circulating levels ofsRAGE in patients who were treated with perindopril.

Complexes of RAGE- and CML-Modified Proteins WereIdentified in Plasma

Plasma from both diabetic rats and patients with type 1diabetes demonstrated significant complexes between RAGE-and CML-modified proteins (Figure 6C). These complexes weredetected with either CML or RAGE immunoprecipitation fol-lowed by detection with the alternate antibody. There werethree bands evident at approximately 85, 50, and 33 kD.

DiscussionWith the recent finding that there are in fact three splice variants

of RAGE in humans (17), the evaluation of the role of this receptorin disease states has become even more complex. Our study hasdemonstrated the presence of each of the three splice variants inrenal tissues from diabetic animals. The sum of total renal RAGEgene expression was found to be similar in diabetic and controlrats. However, the relative distribution of the RAGE variants wascomplex, with increases in the full-length splice variant encodingthe complete RAGE receptor (AGE-binding and downstream sig-naling ability) but with a decrease in tissue sRAGE gene expres-

sion. This pattern of RAGE receptor gene expression is likely tolead to an excess AGE “burden” whereby LMW-AGE were notbeing removed from the circulation by sRAGE and therefore areavailable for uptake by tissues, increasing end-organ AGE accu-mulation with subsequent activation of the full-length RAGE re-ceptor. Indeed, diabetic animals in this study had significant in-creases in their AGE burden both in plasma and in tissues, inassociation with evidence of evolving renal disease. ACE inhibi-tion altered the expression of renal RAGE splice variants with areduction in the full-length splice variant mRNA in the context ofincreased expression of the splice variant encoding for sRAGE.This was associated with decreases in tissue CML and circulatingLMW-AGE, thereby decreasing the overall AGE burden.

The nonhemodynamic effects of blockade of the RAS haveincreasingly been appreciated to play a role in mediating some ofthe renoprotection afforded by these agents. Indeed, our recentfinding that ACE inhibition reduced the accumulation of AGE inexperimental diabetic nephropathy has been confirmed in a num-ber of further experimental studies (10,32). The mechanism bywhich this occurs remains unresolved. One mechanism has beenexcluded, because studies show that ACEi do not trap reactivecarbonyl AGE precursors such as methylglyoxal (33,34), a com-mon characteristic of the AGE formation inhibitors aminoguani-dine, OPB-9195, and pyridoxamine (33,34). In part, effects on AGEreduction have been attributed to inhibition of the generation offree radicals that participate in the production of reactive carbon-yls (9). In support of this, we have shown that ramipril reducedtissue nitrotyrosine levels in experimental diabetes (10), although

Figure 2. Tissue quantification of AGE concentrations in streptozotocin (STZ)-induced diabetic rats at 24 wk. (A) Percentage areaof renal cortical N�(carboxymethyl)lysine (CML) by immunohistochemical morphometric analysis. (B) Renal cortical fluorescenceexpressed as arbitrary units corrected per OD 280 nm protein (Ex 370 nm, Em 440 nm). (C) Skin collagen–associated CML inmmol/mol lysine as assessed by selected ion monitoring gas chromatography–mass spectrometry. (D) Skin collagen–associatedpentosidine in mmol/mol lysine as assessed by RP-HPLC. *P � 0.001 versus C; †P � 0.001 versus D; #P � 0.05 versus D.


this is also seen with the AGE cross-link breaker ALT-711 (22).This is also supported by the in vitro experiments within our studyin which treatment of endothelial cells with the mitochondrialreactive oxygen species scavenger idebenone showed some in-crease in sRAGE secreted into the CCSN. Therefore, part of theeffect of ramipril on sRAGE may be via its antioxidant properties.

In this study, ramipril attenuated diabetes-induced increases inrenal and tissue CML, a major AGE. CML has been identified as aligand for the RAGE receptor (35). The relationship between AGElevels and RAGE expression has not been examined in detail. Typi-

cally, AGE and RAGE co-localize, and in states of increased AGE,there is also an increase in RAGE expression (36). Indeed, circulatingcomplexes between sRAGE and RAGE were found within this studyin both human and rodent plasma. Whether AGE per se promoteRAGE expression is not known, although there has been some evi-dence supporting this in human vascular endothelial cells throughNF-�B (37). Our group showed previously that NRK-52E cells thatare exposed to CML-BSA overexpress RAGE, which in turn causestransition of these cells from their epithelial phenotype to myofibro-blasts (20), a cell type that plays a major role in progressive renalscarring and fibrosis (38). This ability of AGE to modulate RAGEexpression is likely to contribute to a pathologic state of chronicactivation of the full-length RAGE receptor in disease states withelevated AGE levels, such as diabetes.

With evidence from the Diabetes Control and ComplicationsTrial of patients with type 1 diabetes showing that skin CMLlevels are a major predictor of progression to diabetic complica-tions (39), it is important to delineate the exact regulation not onlyof plasma but also of tissue components of the AGE pathway.Indeed, assessment of tissue collagen–associated AGE provides uswith a more accurate record of historical levels of circulating AGEand glucose. The relative importance not only of sRAGE in regu-lating the level of circulating AGE but also of its postulated effecton tissue AGE levels remain to be determined. It therefore istempting to speculate that the reduction seen in our study in theAGE, CML with ACE inhibition may ultimately influence theexpression of RAGE.

Figure 3. Real-time reverse transcription–PCR for the gene expres-sion of RAGE splice variants in renal tissues from STZ-induceddiabetic rats at 24 wk. (A) Gene expression of total RAGE withprobes and primers that recognize all three splice variants. (B)Gene expression of RAGE intracellular domain, with probes andprimers that recognize full-length RAGE and N-truncated RAGE.(C) Gene expression RAGE extracellular domain with probes andprimers that recognize full-length RAGE and C-truncated (solu-ble) RAGE. Gene expression is in arbitrary units with the controlgroup assigned a value of 1. *P � 0.001 versus C; §P � 0.05 versusC; †P � 0.001 versus D; #P � 0.05 versus D.

Figure 4. Protein analysis and AGE-binding characteristics ofRAGE isoforms in renal tissues from STZ-induced diabetic ratsat 24 wk. (A) Immunoprecipitation of renal homogenates withintracellular RAGE antibody that recognizes both full-lengthRAGE and N-truncated RAGE. (B) Far Western analysis ofimmunoprecipitates of intracellular RAGE antibody and renalhomogenates taken from A. (C) Immunoprecipitation of renalhomogenates with extracellular RAGE antibody that recognizesboth full-length RAGE and C-truncated (soluble) RAGE. (D)Far Western analysis of immunoprecipitates of extracellularRAGE antibody and renal homogenates taken from C.


A number of studies have suggested that administration of sRAGEis protective in a range of diabetic complications (13,15,16). Our studyhas identified a novel pathway for the production and secretion ofsRAGE into the circulation. Untreated patients with type 1 diabeteshad a significant decrease in circulating plasma sRAGE over thestudy duration in association with increases in circulating LMW-AGE when compared with patients who were treated with the ACEiperindopril. Although patients who were treated with nifedipine alsohad little increase in circulating AGE over the study period, it wasinteresting to note that these patient had a drop in circulating sRAGEand did not do as well as the patients who were on ACEi forequivalent BP lowering. These findings were also associated withprogression of diabetic renal disease as assessed by an increase inAER over the study duration. By contrast, patients who were treatedwith perindopril had less renal disease with a significant reduction inalbuminuria over the study period. Furthermore, plasma concentra-tions of LMW-AGE in these ACEi-treated patients had decreasedover the 2-yr follow-up duration, whereas levels of “protective”sRAGE were significantly increased by perindopril therapy. We hy-pothesize that this increase in sRAGE could act as a mechanism todivert CML-modified peptides and other AGE from binding to thefull-length RAGE receptor, acting as a “decoy” because this truncatedform of the receptor has no downstream signaling capacity. One mayspeculate, at least in part, that the secretion of sRAGE into the circu-lation may originate from endothelial cells, a major site of RAGEexpression (11), consistent with the significant increase in the concen-tration of sRAGE in the supernatants of bovine endothelial cells that

were treated with the active metabolite of the ACEi ramiprilat. Therewas, however, significant sRAGE protein identified within renal cellhomogenates particularly from ACEi-treated animals, which mayalso provide part of the pool of circulating sRAGE.

Figure 5. Circulating levels of low molecular weight (LMW) AGEand sRAGE in rat plasma. Diabetic rodent plasma samples at 24wk. (A) LMW-AGE fluorescence (Ex 370 nm/Em 440 nm) inarbitrary units. (B) sRAGE receptor in arbitrary units per milli-gram of plasma protein.

Figure 6. Circulating levels of LMW AGE and sRAGE in humanplasma. Human plasma samples from the Melbourne Diabetic Ne-phropathy Study Group representative of the change () in levelsbetween times 0 and 24 mo. (A) in LMW-AGE fluorescence (Ex 370nm/Em 440 nm) in arbitrary units and in sRAGE receptor inarbitrary units per milligram of plasma protein. (B) Correlation curvebetween plasma sRAGE and LMW AGE in perindopril-treated pa-tients (R2 � 0.52, P � 0.047). (C) Plasma AGE-RAGE complexes inpatients and rodents with type 1 diabetes. The left-hand sampleswere immunoprecipitated (IP) with sRAGE and Western immuno-blotted (WB) for CML. The right-hand samples were immunopre-cipitated intraperitoneally with CML and WB with an antibody tototal RAGE. *P � 0.01, placebo- versus perindopril-treated patients;#P � 0.05, placebo- versus perindopril-treated patients; †P � 0.05,perindopril- versus nifedipine-treated patients; ‡P � 0.001, perindo-pril- versus nifedipine-treated patients.


The AGE pentosidine can be generated from nonoxidative path-ways that may not be blocked by ACE inhibition (40). In vitrostudies have identified some effects of ACE inhibition on pento-sidine (9), but these findings have not been confirmed in vivo. Inaddition, pentosidine is not a major ligand of RAGE; therefore,increasing levels of protective sRAGE in the plasma would morelikely have no effect on reducing tissue pentosidine levels. It isunclear whether the lack of effect of ACEi in vivo on certain AGEsuch as pentosidine could partly explain why these agents do nottotally prevent but only retard the rate of progression to ESRD indiabetes (41). There is also intracellular glycation occurring indiabetes, and this phenomenon is likely not to be readily affectedby increases in circulating sRAGE levels. This is supported by thefact that although circulating LMW-AGE were normalized withACE inhibition, tissue levels of CML in the diabetic animals wereonly partially attenuated by this treatment.

These findings lead us to conclude that ACEi play a pivotal role inreducing the burden of diabetic nephropathy as a result of advancedglycation and activation of RAGE. These results suggest that in thefuture, it is likely that therapies will confer superior renoprotectionindependent of their previously established mechanism of action,thereby further preventing progression to ESRD in diabetes.

AcknowledgmentsThis work was completed with support from the Juvenile Diabetes

Research Foundation (JDRF), the National Health and Medical Re-search Council of Australia, and United States Public Health ServiceGrant DK-19971. J.F. is a JDRF Post-Doctoral Research Fellow. M.T. is arecipient of a Don & Lorraine Jacquot Fellowship. A.E.-O. is supportedby a fellowship from the FRAXA Research Foundation.

We thank Gavin Langmaid for expert care of the animals throughout thestudy and Maryann Arnstein for technical expertise.

References1. Brenner BM, Cooper ME, de Zeeuw D, Keane WF, Mitch WE,

Parving HH, Remuzzi G, Snapinn SM, Zhang Z, Shahinfar S:Effects of losartan on renal and cardiovascular outcomes inpatients with type 2 diabetes and nephropathy. N Engl J Med345: 861–869, 2001

2. Lewis EJ, Hunsicker LG, Bain RP, Rohde RD: The effect ofangiotensin-converting-enzyme inhibition on diabetic ne-phropathy. The Collaborative Study Group. N Engl J Med 329:1456–1462, 1993

3. Heart Outcomes Prevention Evaluation Study Investigators:Effects of ramipril on cardiovascular and microvascular out-comes in people with diabetes mellitus: Results of the HOPEstudy and MICRO-HOPE substudy. Lancet 355: 253–259, 2000

4. Wolf G, Neilson EG: Angiotensin II as a renal growth factor.J Am Soc Nephrol 3: 1531–1540, 1993

5. Ruiz-Ortega M, Lorenzo O, Ruperez M, Egido J: ACE inhib-itors and AT(1) receptor antagonists-beyond the haemody-namic effect. Nephrol Dial Transplant 15: 561–565, 2000

6. Soulis-Liparota T, Cooper M, Papazoglou D, Clarke B, JerumsG: Retardation by aminoguanidine of development of albu-minuria, mesangial expansion, and tissue fluorescence instreptozocin-induced diabetic rat. Diabetes 40: 1328–1334,1991

7. Degenhardt TP, Alderson NL, Arrington DD, Beattie RJ, Bas-gen JM, Steffes MW, Thorpe SR, Baynes JW: Pyridoxamine

inhibits early renal disease and dyslipidemia in the strepto-zotocin-diabetic rat. Kidney Int 61: 939–950, 2002

8. Forbes JM, Cooper ME, Oldfield MD, Thomas MC: Role ofadvanced glycation end products in diabetic nephropathy.J Am Soc Nephrol 14[Suppl]: S254–S258, 2003

9. Miyata T, van Ypersele de Strihou C, Ueda Y, Ichimori K,Inagi R, Onogi H, Ishikawa N, Nangaku M, Kurokawa K:Angiotensin II receptor antagonists and angiotensin-convert-ing enzyme inhibitors lower in vitro the formation of ad-vanced glycation end products: Biochemical mechanisms.J Am Soc Nephrol 13: 2478–2487, 2002

10. Forbes JM, Cooper ME, Thallas V, Burns WC, Thomas MC,Brammar GC, Lee F, Grant SL, Burrell LA, Jerums G, OsickaTM: Reduction of the accumulation of advanced glycationend products by ACE inhibition in experimental diabeticnephropathy. Diabetes 51: 3274–3282, 2002

11. Neeper M, Schmidt AM, Brett J, Yan SD, Wang F, Pan YC,Elliston K, Stern D, Shaw A: Cloning and expression of a cellsurface receptor for advanced glycosylation end products ofproteins. J Biol Chem 267: 14998–15004, 1992

12. Schmidt AM, Vianna M, Gerlach M, Brett J, Ryan J, Kao J,Esposito C, Hegarty H, Hurley W, Clauss M, Wang F, PanY-CE, Tsang TC, Stern D: Isolation and characterization oftwo binding proteins for advanced glycosylation end prod-ucts from bovine lung which are present on the endothelialcell surface. J Biol Chem 267: 14987–14997, 1992

13. Wendt TM, Tanji N, Guo J, Kislinger TR, Qu W, Lu Y,Bucciarelli LG, Rong LL, Moser B, Markowitz GS, Stein G,Bierhaus A, Liliensiek B, Arnold B, Nawroth PP, Stern DM,D’Agati VD, Schmidt AM: RAGE drives the development ofglomerulosclerosis and implicates podocyte activation in thepathogenesis of diabetic nephropathy. Am J Pathol 162: 1123–1137, 2003

14. Schmidt AM, Stern DM: RAGE: A new target for the preven-tion and treatment of the vascular and inflammatory compli-cations of diabetes. Trends Endocrinol Metab 11: 368–375, 2000

15. Park L, Raman KG, Lee KJ, Lu Y, Ferran LJ Jr, Chow WS,Stern D, Schmidt AM: Suppression of accelerated diabeticatherosclerosis by the soluble receptor for advanced glycationendproducts. Nat Med 4: 1025–1031, 1998

16. Flyvbjerg A, Denner L, Schrijvers BF, Tilton RG, MogensenTH, Paludan SR, Rasch R: Long-term renal effects of a neu-tralizing RAGE antibody in obese type 2 diabetic mice. Dia-betes 53: 166–172, 2004

17. Yonekura H, Yamamoto Y, Sakurai S, Petrova RG, AbedinMJ, Li H, Yasui K, Takeuchi M, Makita Z, Takasawa S, Oka-moto H, Watanabe T, Yamamoto H: Novel splice variants ofthe receptor for advanced glycation end-products expressedin human vascular endothelial cells and pericytes, and theirputative roles in diabetes-induced vascular injury. Biochem J370: 1097–1109, 2003

18. Jerums G, Allen TJ, Campbell DJ, Cooper ME, Gilbert RE,Hammond JJ, Raffaele J, Tsalamandris C: Long-term compar-ison between perindopril and nifedipine in normotensivepatients with type 1 diabetes and microalbuminuria. Am JKidney Dis 37: 890–899, 2001

19. Nishikawa T, Edelstein D, Du XL, Yamagishi S, MatsumuraT, Kaneda Y, Yorek MA, Beebe D, Oates PJ, Hammes HP,Giardino I, Brownlee M: Normalizing mitochondrial super-oxide production blocks three pathways of hyperglycaemicdamage. Nature 404: 787–790, 2000

20. Oldfield MD, Bach LA, Forbes JM, Nikolic-Paterson D,


McRobert A, Thallas V, Atkins RC, Osicka T, Jerums G,Cooper ME: Advanced glycation end products cause epithe-lial-myofibroblast transdifferentiation via the receptor for ad-vanced glycation end products (RAGE). J Clin Invest 108:1853–1863, 2001

21. Rumble JR, Cooper ME, Soulis T, Cox A, Wu L, Youssef S,Jasik M, Jerums G, Gilbert RE: Vascular hypertrophy in ex-perimental diabetes—Role of advanced glycation end prod-ucts. J Clin Invest 99: 1016–1027, 1997

22. Forbes JM, Thallas V, Thomas MC, Founds HW, Burns WC,Jerums G, Cooper ME: The breakdown of preexisting ad-vanced glycation end products is associated with reducedrenal fibrosis in experimental diabetes. FASEB J 17: 1762–1764, 2003

23. Wrobel K, Garay-Sevilla ME, Nava LE, Malacara JM: Novelanalytical approach to monitoring advanced glycosylationend products in human serum with on-line spectrophotomet-ric and spectrofluorometric detection in a flow system. ClinChem 43: 1563–1569, 1997

24. Dyer DG, Dunn JA, Thorpe SR, Bailie KE, Lyons TJ, McCanceDR, Baynes JW: Accumulation of Maillard reaction productsin skin collagen in diabetes and aging. J Clin Invest 91: 2463–2469, 1993

25. Stegemann H, Stalder K: Determination of hydroxyproline.Clin Chim Acta 18: 267–273, 1967

26. Hsu SM, Raine L, Fanger H: Use of avidin-biotin-peroxidasecomplex (ABC) in immunoperoxidase techniques: A compar-ison between ABC and unlabeled antibody (PAP) proce-dures. J Histochem Cytochem 29: 577–580, 1981

27. van der Pijl JW, Daha MR, van den Born J, Verhagen NA,Lemkes HH, Bucala R, Berden JH, Zwinderman AH, BruijnJA, van Es LA, van der Woude FJ: Extracellular matrix inhuman diabetic nephropathy: Reduced expression of hepa-ran sulphate in skin basement membrane. Diabetologia 41:791–798, 1998

28. Forbes JM, Hewitson TD, Becker GJ, Jones CL: Ischemic acuterenal failure: Long-term histology of cell and matrix changesin the rat. Kidney Int 57: 2375–2385, 2000

29. Candido R, Forbes JM, Thomas MC, Thallas V, Dean RG,Burns WC, Tikellis C, Ritchie RH, Twigg SM, Cooper ME,Burrell LM: A breaker of advanced glycation end productsattenuates diabetes-induced myocardial structural changes.Circ Res 92: 785–792, 2003

30. Laemmli UK: Cleavage of structural proteins during the as-sembly of the head of bacteriophage T4. Nature 227: 680–685,1970

31. Forbes JM, Soulis T, Thallas V, Panagiotopoulos S, Long DM,Vasan S, Wagle D, Jerums G, Cooper ME: Renoprotectiveeffects of a novel inhibitor of advanced glycation. Diabetologia44: 108–114, 2001

32. Nangaku M, Miyata T, Sada T, Mizuno M, Inagi R, Ueda Y,Ishikawa N, Yuzawa H, Koike H, van Ypersele de Strihou C,Kurokawa K: Anti-hypertensive agents inhibit in vivo theformation of advanced glycation end products and improverenal damage in a type 2 diabetic nephropathy rat model.J Am Soc Nephrol 14: 1212–1222, 2003

33. Dunn JA, McCance DR, Thorpe SR, Lyons TJ, Baynes JW:Age-dependent accumulation of N epsilon-(carboxymethyl-)lysine and N epsilon-(carboxymethyl)hydroxylysine in hu-man skin collagen. Biochemistry 30: 1205–1210, 1991

34. Dunn JA, Patrick JS, Thorpe SR, Baynes JW: Oxidation ofglycated proteins: Age-dependent accumulation of N epsi-lon-(carboxymethyl)lysine in lens proteins. Biochemistry 28:9464–9468, 1989

35. Valencia JV, Weldon SC, Quinn D, Kiers GH, DeGroot J,TeKoppele JM, Hughes TE: Advanced glycation end productligands for the receptor for advanced glycation end products:Biochemical characterization and formation kinetics. AnalBiochem 324: 68–78, 2004

36. Soulis T, Thallas V, Youssef S, Gilbert RE, McWilliam BG,Murray-McIntosh RP, Cooper ME: Advanced glycation endproducts and their receptors co-localise in rat organs suscep-tible to diabetic microvascular injury. Diabetologia 40: 619–628, 1997

37. Tanaka N, Yonekura H, Yamagishi S, Fujimori H, YamamotoY, Yamamoto H: The receptor for advanced glycation endproducts is induced by the glycation products themselvesand tumor necrosis factor-alpha through nuclear factor-kappa B, and by 17 beta-estradiol through Sp-1 in humanvascular endothelial cells. J Biol Chem 275: 25781–25790, 2000

38. Pedagogos E, Hewitson T, Fraser I, Nicholls K, Becker G:Myofibroblasts and arteriolar sclerosis in human diabetic ne-phropathy. Am J Kidney Dis 29: 912–918, 1997

39. Monnier VM, Bautista O, Kenny D, Sell DR, Fogarty J, DahmsW, Cleary PA, Lachin J, Genuth S: Skin collagen glycation,glycoxidation, and crosslinking are lower in subjects withlong-term intensive versus conventional therapy of type 1diabetes: Relevance of glycated collagen products versusHbA1c as markers of diabetic complications. DCCT SkinCollagen Ancillary Study Group. Diabetes Control and Com-plications Trial. Diabetes 48: 870–880, 1999

40. Miyata T, van Ypersele de Strihou C: Angiotensin II receptorblockers and angiotensin converting enzyme inhibitors: Im-plication of radical scavenging and transition metal chelationin inhibition of advanced glycation end product formation.Arch Biochem Biophys 419: 50–54, 2003

41. Perkins BA, Ficociello LH, Silva KH, Finkelstein DM, WarramJH, Krolewski AS: Regression of microalbuminuria in type 1diabetes. N Engl J Med 348: 2285–2293, 2003

See related editorial, “Angiotensin-Converting Enzyme Inhibition in Diabetic Nephropathy: It’s All the RAGE,” onpages 2251–2253.


modulation of soluble receptor for advanced glycation end products by angiotensin-converting...

Documents