Long-Term Renal Effects of a Neutralizing RAGE Antibody in Obese Type 2 Diabetic Mice

Adult female db/db mice (C57BLKS/J-lepr^db/lepr^db) and their age-matched db/+ littermates (C57BLKS/J-lepr^db/+) (M&B, Ry, Denmark) were used. db/+ mice had an initial body weight of ∼19–20 g, whereas the db/db mice had an initial weight of 39–41 g. The db/db mice were included in the study 1–2 weeks after the development of diabetes, at the age of 8 weeks. The mice were housed six to eight per cage in a room with a 12:12-h artificial light cycle (7:00 a.m. to 7:00 p.m.), a temperature of 21 ± 1°C, and a humidity of 55 ± 5%. The animals had free access to standard diet (Altromin, Lage, Germany) and tap water throughout the experiment. The study complied with Danish regulations for care and use of laboratory animals.

The db/db mice were randomized into two groups of 10 animals. One group of db/db mice was treated with intraperitoneal injections of a neutralizing murine RAGE antibody, whereas the other half was treated with an isotype-matched irrelevant IgG (M3645; Sigma, St. Louis, MO). db/+ mice were randomized into two groups of six animals to receive RAGE antibody or irrelevant IgG. The RAGE antibody and irrelevant IgG were administered intraperitoneally in an initial bolus dose of 300 μg followed by doses of 100 μg three times weekly. The RAGE antibody and irrelevant IgG were dissolved in 0.154 mol/l NaCl and injected in a volume of 0.5 ml. A full characterization of the RAGE antibody and documentation of its neutralizing activity are described below.

Body weight, food consumption, and blood glucose levels were determined at initiation of the experiment and every 2 weeks. Blood glucose was measured in tail vein blood, as described below. After 8 weeks, mice were placed in metabolic cages to collect 24-h urine samples for urinary albumin excretion (UAE) and urinary creatinine determinations. Urine samples were stored at −20°C until assay was performed. When killed, mice were anesthetized with pentobarbital (50 mg/kg i.p.), and nonfasting blood samples were drawn from the retro-orbital venous plexus using heparinized capillary tubes. Serum samples were stored at −80°C until analysis was performed. In all animals, the right and left kidneys were removed and weighed. The middle piece of the right kidney (including the papilla) was fixed in 4% paraformaldehyde for determination of glomerular volume by light microscopy (see details below). The middle piece of the left kidney (including the papilla) was fixed in 0.1 mol/l cacodylate buffer with 1% glutaraldehyde and 2% paraformaldehyde for later determination of basement membrane thickness (BMT) and mesangial fraction by electron microscopy (see details below). Cortical renal tissue was obtained from the kidney poles and snap frozen in liquid nitrogen for measurements of nuclear factor κB (NFκB) DNA binding activity by electrophoretic mobility shift assay and endothelial nitric oxide synthase (eNOS) levels by immunoblotting (see below). In addition, liver and heart were removed, weighed, and snap frozen in liquid nitrogen.

The preparation and characterization of the neutralizing monoclonal RAGE antibody followed procedures previously described in detail for the preparation of other neutralizing monoclonal antibodies (18). The human RAGE extracellular domain encompassing residues 23–340 (sRAGE) was expressed and purified from E. coli using the pET thioredoxin system (Novagen, Madison, WI). Female 8-week-old BALB/c mice were immunized and boosted three times 21 days apart by intraperitoneal and subcutaneous injections of 100 μg sRAGE protein in Complete Freund’s adjuvant for the primary immunization and an additional 50 μg sRAGE in Incomplete Freund’s adjuvant for secondary immunizations. The mouse with the highest serum titer to sRAGE as measured by enzyme-linked immunosorbent assay was injected intravenously with an additional 30 μg immunogen in PBS, 21 days after the last immunization. Three days later, spleen cells were harvested for production of hybridomas to sRAGE using previously described techniques (19). The hybridoma cell line with highest antibody titer and neutralizing antibody activity was selected after cloning three to four times by limiting dilution in 96-well microtiter plates, grown in a Cellmax Bioreactor (Spectrum, Rancho Dominguez, CA) using Dulbecco’s modified Eagle’s medium. Purified IgG was prepared by Protein-A chromatography. The isotype (IgG₃) and light chain composition (κ) of the antibody were determined as previously described (18).

An NFκB reporter-gene assay using N^ε-(carboxymethyl)-lysine (CML)-modified human serum albumin (CML-HSA) as a ligand for RAGE was used to measure neutralizing activity of the monoclonal antibody. Details of the preparation of CML-HSA and the reporter-gene assay have been described previously (20). THP-1 cells were seeded at 5 × 10⁶ cells per 100-mm dish in 10 ml of serum-free medium (SFM) the day before transfection. Transient transfection was performed using the DEAE-dextran method, as previously described (21). Cells were washed once with SFM and resuspended in 1 ml of the same medium containing 2 μg NFκB-Luc reporter plasmid (Clontech, Palo Alto, CA) and 200 mg/l DEAE-dextran (Promega, Madison, WI). The cell-DNA mixture was incubated at room temperature for 20–30 min before washing, centrifugation, and resuspension into fresh SFM. Transfected cells were seeded into 96-well plates at 70,000 cells/well for recovery. After 24 h, cells were pretreated with 10–100 mg/l RAGE antibody for 1 h and then treated with 200–600 mg/l CML-modified albumin for 1–6 h before the reporter assay. Equivalent amounts of cell lysates, normalized for total protein (Bradford protein assay; Bio-Rad, Palo Alto, CA), were used for measurement of luciferase activity. Luciferase assays were performed using the Steady-Glo luciferase assay system according to the manufacturer’s instructions (Promega), and luminescence was detected in a TopCount microplate scintillation counter using a single-photon monitor program (Packard Instrument, Meriden, CT).

Blood glucose was measured at day 0 and every 2 weeks in tail vein blood by Precision Xtra Plus (Abbott Laboratories, MediSence Products, Bedford, MA), and urine was tested for glucose and ketone bodies by Combur (5) Test D (Roche Diagnostics, Mannheim, Germany). Serum insulin was measured by an ultrasensitive rat insulin enzyme-linked immunosorbent assay (DRG Diagnostics, Marburg, Germany). Semilog linearity of mouse serum and rat insulin was found at multiple dilutions, indicating antigen similarity between mouse and rat insulin. The intra- and interassay coefficients of variation were <5 and <10%, respectively, for the insulin assay.

The urinary albumin concentration was determined by radioimmunoassay using rat albumin antibody and rat albumin standard, as previously described (17). Semilog linearity of mouse urine and rat albumin (in the standard) was found at multiple dilutions, indicating antigen similarity between mouse and rat albumin. Serum and urinary creatinine concentrations were measured by an automated technique adapted from the method of Jaffé and corrected for the prevailing glucose content due to interference in the Jaffé reaction. The creatinine clearance (CrCl) was expressed in milliliters per hour. The intra- and interassay coefficients of variation were <5 and <10%, respectively, for both assays.

The middle part of the right kidney (containing the papilla) was embedded in paraffin for light microscopy examination. Two-micron thick sections were cut on a rotation microtome and stained with p-aminosalicylic acid and hematoxylin. The mean glomerular tuft volume (V_G) was determined from the mean glomerular cross-sectional area (A_G) at a magnification of 400×, as previously described (22,23,24). The areas were determined with a two-dimensional version of the nucleator (24) (CAST; Olympus, Copenhagen, Denmark) as the average area of a total of 40–50 glomerular profiles (tuft omitting the proximal tubular tissue within the Bowmann capsule). V_G was calculated as V_G = β/k × (A_G)^3/2 where β = 1.38, which is the shape coefficient for spheres (the idealized shape of glomeruli), and k = 1.1, which is a size distribution coefficient (22,23,24).

Small blocks of cortical tissue were embedded in Epon 825 for electron microscopy examination. Thin sections were cut on a Reichert Ultracut (Leica, Vienna, Austria) and stained with uranyle acetate and lead citrate. From an electron microscope (Tecnai 12; Phillips, Einthoven, Holland), images covering the whole glomerular profile were recorded with a MegaView video camera (Soft Imagimg System, Münster, Germany) onto a monitor. Measurements of mesangial regions were performed at a final magnification of 3,200×. Four to six glomeruli were measured from two blocks. Mesangial fractions were determined by point counting of mesangial regions as fraction of the tuft. The total mesangial volume was calculated by multiplying the mesangial fraction by the total glomerular volume. For measurements of BMT, randomized fields were recorded at a magnification of 30,000× from the same sections described above. BMT was measured applying the orthogonal intercept method, as previously described (25). About 60 measurements were performed per glomerulus, and BMT is given as a harmonic mean.

To assay for DNA-binding activity, 15 μg homogenate of kidney cortex in 5 μl PBS was mixed with 1.5 μg poly[d(I-C)], 6 μl electrophoretic mobility shift assay buffer (60 mmol/l HEPES, pH 7.9, 12 mmol/l Tris-HCl, pH 7.9, 30% glycerol, 80 mmol/l NaCl, 3 mmol/l DTT, 1.8 mmol/l EDTA, and 15 mmol/l MgCl₂), and 50,000 cpm ³²P-labeled probe to reach a volume of 18 μl. After 20 min of incubation at room temperature, the reaction mixture was subjected to electrophoresis in a nondenaturing 5% polyacrylamide gel in 0.5 × TBE buffer (45 mmol/l Tris base, 45 mmol/l boric acid, and 1 mmol/l EDTA). The gel was dried and analyzed by autoradiography. The probe was derived from the distal κB site of the murine NOS2 promoter 5′-TAG GGG GAT TTT CCC CTC-3′ (26,27). The bands were quantified by densitometry.

Approximately 50 mg of cortical kidney tissue was sonicated for 20 s in chilled 50 mmol/l Tris, 1 mmol/l EDTA, pH 7.5, buffer. Samples (100 μg protein per lane) of tissue homogenates were separated by SDS-PAGE (7.5% acrylamide) and transferred by electroelution onto nitrocellulose paper (Schleicher & Schuell, Munich, Germany). The blots were blocked overnight at 4°C in 5% nonfat milk and incubated for 12 h at 4°C with a monoclonal eNOS antibody (Transduction Laboratories, Lexington, KY) in a dilution of 1:1,000. After washing in TBS-Tween (0.1%), the membranes were incubated with an anti-mouse IgG antibody conjugated with horseradish peroxidase (Pierce, Rockford, Il) for 1 h at room temperature in a dilution of 1:10,000 and 1:75,000, respectively. The bands were visualized by chemiluminescence (BioWest; UPV, Upland, CA) and quantified by a UPV BioImaging System.

ANOVA for repeated measurements was used to evaluate differences with Student’s t test for unpaired comparisons. A P value <0.05 was considered statistically significant. For data not following a normal distribution, the Mann-Whitney rank-sum test was used. All data are expressed as means ± SE, with n indicating the number of mice studied. Statistical analysis was performed using the statistical package SPSS for Windows.

The db/db mice had a greater body weight than the db/+ mice, as was also the case for food consumption (Table 1). Mean blood glucose levels were ∼18–19 mmol/l in db/db mice throughout the study and 5–6 mmol/l in db/+ animals (Table 1). The db/db mice had severe hyperinsulinemia (Table 2). The placebo-treated db/db mice had greater liver weights than the db/+ mice, whereas heart weights were unchanged (Table 2). RAGE antibody administration did not affect any of the above parameters in either db/db mice or db/+ mice (Tables 1 and 2).

Placebo-treated db/db mice showed a 48% increase in kidney weight at day 60 (189 ± 6 vs. 128 ± 7 mg, P < 0.01) compared with the placebo-treated db/+ animals (Fig. 1). No difference in kidney weight was seen between the two db/+ groups (128 ± 7 vs. 118 ± 4 mg, NS) (Fig. 1). In RAGE antibody–treated db/db mice, a significantly smaller increase in kidney weight was observed versus placebo-treated db/db mice (160 ± 5 mg, P < 0.05), although the kidney weight was higher than that seen in db/+ animals (P < 0.01). RAGE antibody administration in db/+ animals had no effect on kidney weight (Fig. 1). The same pattern of changes was seen in glomerular volume (Fig. 1). Total glomerular volume increased by 31% in placebo-treated db/db mice versus placebo-treated db/+ animals (2.31 ± 0.06 vs. 1.76 ± 0.03 10⁵ μm³, P < 0.01). RAGE antibody treatment in db/db mice partially blocked the increase in glomerular volume compared with placebo-treated db/db mice (2.09 ± 0.04 10⁵ μm³, P < 0.01). The glomerular volume was, however, still elevated above that of placebo- and RAGE antibody–treated db/+ animals (P < 0.05). No difference in glomerular volume was seen between the two db/+ groups (1.76 ± 0.03 vs. 1.67 ± 0.04 10⁵ μm³, NS) (Fig. 1). BMT increased by 11% in placebo-treated db/db mice compared with db/+ animals (171 ± 5 vs. 154 ± 2 nm, P < 0.05), whereas this increase was blocked in RAGE antibody–treated db/db mice (147 ± 6 nm), with a value comparable with both placebo- and RAGE antibody–treated db/+ mice (Fig. 2). Placebo-treated db/db mice had an increase in total mesangial volume of 117% compared with placebo-treated db/+ animals (7.63 ± 0.38 vs. 3.51 ± 0.50 10⁴ μm³, P < 0.01). RAGE antibody administration in db/db mice partially blocked the increase in total mesangial volume (P < 0.01) (Fig. 2).

A pronounced increase in UAE was observed in placebo-treated db/db mice at day 60 versus nondiabetic db/+ mice (2.37 ± 0.29 vs. 0.29 ± 0.01 μg/24 h, P < 0.01), with a considerably lower level in db/db mice treated with the RAGE antibody (1.13 ± 0.20 μg/24 h versus placebo-treated db/db mice, P < 0.01) (Fig. 3). No effect of RAGE antibody treatment on UAE was seen in db/+ animals. Placebo-treated db/db mice showed a pronounced increase in CrCl compared with db/+ mice (5.10 ± 0.28 vs. 2.31 ± 0.30 ml/h, P < 0.01), an increase blocked in the RAGE antibody–treated db/db mice (2.13 ± 0.23 ml/h) (Fig. 3). No effect of RAGE antibody treatment was seen in db/+ animals.

Activated NFκB decreased in the cortical tissue of placebo-treated db/db mice (Fig. 4 and Table 3). No effect of RAGE antibody administration was seen on NFκB in db/db mice, whereas a significant decrease was seen in db/+ animals treated with the RAGE antibody (P < 0.01). Renal eNOS levels increased by 30% in placebo-treated db/db mice compared with db/+ animals (P < 0.01) (Table 3). The diabetes-associated increase in eNOS was not influenced by RAGE antibody administration in db/db mice, and no effect of treatment was seen in db/+ animals (Table 3).

The db/db mouse, which expresses a leptin receptor defect in the hypothalamus, is a genetic model of type 2 diabetes characterized by obesity, sustained hyperglycemia, hyperinsulinemia, and lack of ketonuria. Previously, this model has been shown to present with robust diabetic renal changes characterized by increased renal weight, glomerular volume, CrCl, BMT, UAE, and mesangial volume within 2 months of diabetes onset (12,13,14,15,16,17). The major new finding of the present study is a specific effect of RAGE antibody administration on the development of renal dysfunction in db/db mice with a diabetes duration of 2 months. RAGE antibody administration attenuated the increase in renal weight, glomerular volume, mesangial expansion, and UAE and prevented the increase in CrCl and BMT. RAGE antibody administration had no renal effects in nondiabetic animals. Furthermore, the renal effects in db/db mice were seen without any effect on metabolic control, insulin levels, body weight, or food consumption. These data directly indicate that RAGE plays a significant pathogenic role in the development of both early and late renal changes in a model of type 2 diabetes.

Nonenzymatic glycation occurs through a series of biochemical reactions between glucose and other reactive carbonyl compounds, proteins, lipids, or nucleic acids (28). AGEs interact with specific receptors, the best characterized being RAGE (10). Other identified AGE-binding proteins include the components of the AGE receptor p60 complex, p90, galectin-3 (29), and the macrophage scavenger receptor types I and II (30). In addition, AGEs also act in a direct nonreceptor mediated way by cross-linking proteins (10). By both mechanisms, AGEs may activate signal transduction pathways that involve cytokines and growth factors (5). The colocalization of AGEs and RAGE in renal glomeruli and other sites of diabetic microvascular injury suggests that this ligand-receptor interaction may represent an important mechanism in the pathogenesis of diabetes complications (8). A significant increase in ¹²⁵I-AGE binding is found in diabetic rat kidney and appears to be modulated by endogenous AGE levels (31). Furthermore, in the absence of hyperglycemia, exogenously administered AGEs induce extracellular matrix genes specific for early diabetic kidney disease in nondiabetic mice (6) and mimic the long-term effects of diabetes, such as increased mesangial expansion in nondiabetic rats (3). Both of these results were accompanied by upregulation of transforming growth factor (TGF)-β1 in mice and rats (3,6). In STZ-induced diabetic rats, renal AGE levels are increased after 3 weeks of diabetes (31) and over a period of 32 weeks (9,32), confirming that in vivo generation of AGEs in the kidney is time dependent and closely linked to the development of experimental diabetic glomerulopathy.

Although there is increasing interest in the identification and development of specific antagonists of a pathophysiologically enhanced AGE system in oncology, inflammatory responses, and macrovascular complications in type 1 diabetes (10), very few studies have appeared on the effect of specific RAGE blockade in diabetic kidney disease. Evidence for a role of RAGE in the renal changes seen in a model of type 1 diabetes (i.e., STZ-induced diabetic mice) has been published in a study using diabetic mice overexpressing human RAGE (11). These animals developed renal/glomerular hypertrophy, increased albuminuria, mesangial expansion, glomerulosclerosis, and increased serum creatinine robustly exceeding the levels seen in diabetic nontransgenic littermates (11). Furthermore, induction of STZ-induced diabetes in homozygous RAGE null mice (i.e., animals with a global deletion of RAGE) was followed by a diminished increase in renal hypertrophy, glomerular and mesangial area, and BMT compared with diabetic wild-type animals (12).

In the present study performed in a mouse model of type 2 diabetes, administration of a neutralizing RAGE antibody was shown to blunt the outcome in the classical early features of diabetic kidney disease, i.e., renal/glomerular hypertrophy, hyperfiltration measured as CrCl, and UAE, but also in the late renal changes by blocking the increase in BMT and minimizing the increase in total mesangial volume. In another recent study, in db/db mice treated chronically with sRAGE (i.e., a truncated form of RAGE), the glomerular area, UAE, mesangial area, and BMT were all minimized by treatment compared with placebo-treated db/db mice (12), which is consistent with the results obtained in the present study. The db/db mouse has previously been reported to develop decreased CrCl within 2 months after the onset of diabetes, suggesting a progressive diabetic kidney disease with loss of kidney function (12,16,33). However, the present study corroborates and extends several lines of evidence we previously reported (17), indicating that the renal hyperfunction in placebo-treated db/db mice was partially or fully normalized by RAGE antibody treatment, i.e., partial effect on kidney weight, glomerular volume, UAE, and normalization of elevated CrCl. The reason for this discrepancy in CrCl is unknown but may be explained by a variable susceptibility to diabetes in subbreedings of the db/db mouse strain.

The observation that RAGE antibody treatment abolished the increase in BMT and renal hyperfiltration and partially abolished the increase in UAE is interesting in view of the previously described actions of RAGE on vascular permeability (10) and the anatomical localization of RAGE in the glomerulus (i.e., the endothelial cell and the podocyte) (8,12,34). These results indicate that administration of a specific, neutralizing RAGE antibody in db/db mice prevents the abnormally increased albumin permeability in the diabetic kidney, which is believed to be caused by abnormalities in the filtration barrier due to increased membrane pore size and reduced anion charge. Furthermore, in vascular smooth muscle cells, AGEs induce RAGE-dependent induction of a p21 (ras)-dependent mitogen-activated protein kinase pathway, an effect that can be blocked by RAGE antibodies or sRAGE (35).

RAGE expression has been described in glomerular mesangial cells (7) and, intriguingly, correlates to the RAGE antibody–induced retardation of mesangial expansion shown in the present study. In cultured mesangial cells, anti-p60 antibodies prevent AGE-induced increases in IGF-I, TGF-β1, and extracellular matrix production and gene expression, suggesting that AGE-induced growth factor and extracellular matrix synthesis is AGE-receptor dependent and a likely mediator of the pathogenesis of hyperglycemia-induced mesangial expansion (7).

The present results suggest that the most prominent role of RAGE in the diabetic renal changes in type 2 diabetes is linked to the diabetes-associated permeability changes, although RAGE also seems to play a role in mesangial expansion. In this context, it is interesting that administration of a neutralizing vascular endothelial growth factor (VEGF) antibody in db/db mice has been shown exclusively to block the increase in BMT and elevated UAE but has no effect on glomerular mesangial expansion (17). Furthermore, administration of a TGF-β antibody in db/db mice has been shown specifically to block the diabetes-associated glomerular mesangial expansion without affecting either elevated UAE or renal VEGF expression (33). Additionally, administration of sRAGE in db/db mice or induction of STZ-induced diabetes in RAGE null mice are followed by a decrease in the diabetes-induced rise in both VEGF antigen expression and in TGF-β mRNA compared with placebo-treated animals or wild-type animals, respectively (12). Accordingly, these results support the notion that by blocking RAGE, downstream pathways responsible both for glomerular permeability (e.g., VEGF) and matrix accumulation (e.g., TGF-β) in diabetes are affected (12) and further suggest RAGE to be an important upstream regulator of well-described pathways leading to diabetic kidney disease.

Several studies performed in vitro in endothelial cells, monocytic cells, and vascular smooth muscle cells have shown that AGEs, through stimulation of RAGE, activate NFκB (20,36,37). Accordingly, NFκB has been suggested to play a central role in mediating the effects of AGEs on vascular dysfunction (20,37). However, so far very few studies have appeared on renal NFκB levels in studies performed in diabetic animal models. One study in STZ-diabetic rats (i.e., model of type 1 diabetes) showed a modest increase in renal cortical NFκB levels, and this increase was unaffected of AGE blockade by aminoguanidine and blockade of the ACE system by ACE inhibition (38). Furthermore, in another study in STZ-induced diabetic rats the elevated renal cortical NFκB levels were unaffected by blockade of the endothelin receptor inhibitor bosentan (39). In the present study, using a type 2 diabetic mouse model with more advanced renal changes than those seen in the two studies mentioned above (38,39), renal cortical levels of activated NFκB were decreased. Although RAGE antibody administration had no impact on NFκB in diabetic animals, RAGE antibody administration in nondiabetic mice significantly decreased NFκB. Activation of NFκB has been suggested to be driven by both hyperglycemia and hyperinsulinemia, but the exact role of NFκB as a “good or bad player” in diabetic renal complications is still largely unknown (37). Accordingly, the relationship between AGE, NFκB, and the diabetic kidney changes in vivo needs further investigation.

As AGEs are known to upregulate VEGF, which in turn stimulates the nitric oxide system, it is tempting to speculate that RAGE blockade might influence renal eNOS activity. In line with previously reported data, diabetic animals presented with a modest increase in renal eNOS activity. However, RAGE antibody administration had no impact on this increase. In a recent study, the expressions of RAGE and eNOS were examined in skin biopsies from healthy control subjects and type 2 diabetic patients (40). Interestingly, RAGE and eNOS expressions were not significantly increased in diabetic subjects, and no association was found between the two parameters (40).

In conclusion, the present data strongly support the hypothesis that RAGE is an important pathogenetic factor in the development of long-term renal changes in type 2 diabetes. Future studies are warranted to further elucidate the role of RAGE in diabetic kidney disease and to explore how agents with specific RAGE-blocking properties can be developed for clinical trials.

This work was supported by the Danish Medical Research Council (Grant 9700592), the Eva and Henry Frænkels Memorial Foundation, the Danish Diabetes Association, the Novo-Nordisk Foundation, the Nordic Insulin Foundation, and the Institute of Experimental Clinical Research, University of Aarhus, Denmark. B.F.S. is supported by a grant from the Institute for the Promotion of Innovation by Science and Technology in Flanders, Belgium. We thank Texas Biotechnology, Houston, Texas, for the generous gift of the RAGE antibody.

The excellent technical assistance by Birgitte Grann, Karen Mathiassen and Kirsten Nyborg is highly appreciated.

Preliminary data from the present study were presented in abstract form at the 35th Annual Meeting at the American Society of Nephrology in Philadelphia, Pennsylvania, in October 2002.