Expression of Advanced Glycation End Products and Their Cellular Receptor RAGE in Diabetic Nephropathy and Nondiabetic Renal Disease

Patients

Renal biopsies from patients with diabetic nephropathy (n = 26), hypertensive nephrosclerosis (n = 7), idiopathic focal segmental glomerulosclerosis (FSGS) (n = 11), secondary FSGS attributable to obesity-induced hyperfiltration (n = 7), and lupus nephritis (n = 11) and from normal control subjects (n = 2) were studied by immunohistochemical analysis, using affinity-purified polyclonal antibodies raised to RAGE and two subclasses of AGE, i.e., CML and PENT. The mean ages of the patients in each diagnostic category were 58.7 ± 5.1 yr for hypertensive nephrosclerosis, 29.0 ± 5.0 yr for idiopathic FSGS, 39.0 ± 3.1 yr for secondary FSGS, 26.0 ± 2.4 yr for lupus nephritis, and 20.5 ± 4.5 yr for the normal control subjects.

The group with diabetic nephropathy consisted of 26 patients with adult-onset diabetes mellitus (18 male and 8 female patients, of whom 11 were Caucasian, 11 were African American, 2 were Hispanic, and 2 were Asian American), with a mean age of 59.7 ± 2.2 yr. Cases were chosen retrospectively from the files of the Nephropathology Laboratory of Columbia University to represent a range of severity of diabetic nephropathy, from mild to advanced. This spectrum is reflected in the wide range of serum creatinine levels, from 0.8 to 11.4 mg/dl (mean, 2.9 ± 0.4 mg/dl), and in the broad range of urinary protein excretion values, from 0.3 to 19.5 g/d (mean, 6.7 ± 0.2 g/d). All renal biopsies were performed between August 1997 and March 1998, and specimens were studied with light microscopy, immunofluorescence microscopy, and electron microscopy, using standard techniques. Diagnoses of diabetic nephropathy were based on the characteristic findings of glomerular basement membrane (GBM) thickening and mesangial sclerosis, as detected using light and electron microscopy (17). Cases included 10 examples of nodular diabetic glomerulosclerosis, 15 cases of diffuse diabetic glomerulosclerosis, and 1 example of the hypertrophic phase of diabetic nephropathy with no recognizable mesangial sclerosis at the light microscopic level. All cases exhibited linear staining of GBM and tubular basement membranes (TBM) with antisera to IgG and albumin. The light and electron microscopic features are listed in Table 1.

Generation of Specific Antibodies to RAGE, CML, and PENT

Monospecific polyclonal antibodies were raised against human soluble RAGE. Human soluble RAGE was prepared in a baculovirus expression system, purified to homogeneity, subjected to aminoterminal sequence analysis, and used to immunize New Zealand White rabbits. IgG from immunized rabbits was purified and tested for recognition of RAGE in both enzyme-linked immunosorbent assay (ELISA) and immunoblotting studies, as described previously (7,8). Antibody specificity was confirmed by Western blotting performed with homogenates of normal human kidney (data not shown).

Affinity-purified antibodies recognizing CML adducts of proteins or PENT-modified adducts were prepared and characterized according to previously published procedures (18,19). CML- or PENT- modified keyhole limpet hemocyanin (KLH) (Sigma Chemical Co., St. Louis, MO) was used to immunize rabbits. Antibodies specific for the KLH backbone were removed by passage of the immune IgG through KLH-preadsorbed Affi-gel resin (Bio-Rad, Hercules, CA). The material not adhering to this column was then chromatographed on Affi-gel resin bearing CML- or PENT-modified bovine serum albumin (BSA) and was eluted with 1 M NaCl. Specificity was tested using ELISA and immunoblotting assays, according to previously published procedures (18,19).

Immunohistochemical Analysis of Renal Biopsy Specimens

Formalin-fixed, paraffin-embedded tissues were cut at 3 μm on 3-aminopropyltriethoxysilane (Sigma)-coated slides, deparaffinized, and rehydrated in graded alcohols. Sections to be stained with the antibody to RAGE were heated in a microwave for 25 min before immunostaining. Sections to be stained with the antibodies to CML or PENT were pretreated with trypsin for 90 min. After blocking with 10% normal goat serum (Vector Laboratories, Burlingame, CA), serial sections were stained with the antibodies to CML (0.7 μg/ml), PENT (10 μg/ml), or RAGE (4 μg/ml) and were incubated overnight at 4°C in a humidified chamber. After washing with phosphate-buffered saline (PBS), sections were stained with biotinylated secondary goat anti-rabbit antibody (1:100; Vector Laboratories). Sections were washed with PBS and incubated with avidin-biotin complex (Vector Laboratories) for 30 min, followed by 3,3′-diaminobenzidine solution containing 0.003% H₂O₂. Sections were counterstained with hematoxylin and mounted. Negative controls consisted of serial sections stained with equivalent concentrations of preimmune IgG in place of the primary antibody. To prove the podocytic distribution of RAGE immunostaining, serial sections were stained with a monoclonal antibody to synaptopodin (Maine Biotechnology, Portland, ME), a podocyte-specific, actin-associated protein. After microwave heating, sections were overlaid sequentially with 10% normal horse serum (Vector Laboratories), synaptopodin-specific antibody (1:1), and biotinylated horse anti-mouse antibody (1:100; Vector Laboratories). The combined intensity and distribution of immunostaining were determined on a scale of 0 to 3+ (0, absent; 1+, low intensity and <25% of structures stained; 2+, moderate intensity and <50% of structures stained; 3+, high intensity and >50% of structures stained) for the GBM, mesangium, TBM, interstitium, and vessels.

Correlations between the prevalence of CML and PENT immunostaining and severity of proteinuria and serum creatinine levels were performed using t tests. Correlations between the grade of immunostaining (0 to 3+) for CML and PENT and the severity of diabetic glomerulosclerosis (assessed histologically) were performed using a doubly ordered contingency table (Jonckheere-Terpstra test) and the linear-by-linear association test.

Laser Capture Microdissection

For the studies of RAGE mRNA expression, snap-frozen, archival, renal biopsy tissue (which had been stored at -80°C) from patients with diabetic nephropathy and from normal control subjects was used. The frozen tissue was cut at 8 μm onto noncoated slides and was rehydrated briefly in graded alcohols diluted with diethyl pyrocarbonate (DEPC)-treated water. The sections were stained with Weigert's hematoxylin for 20 s. rinsed briefly in DEPC-treated water for 5 s, dehydrated in graded alcohols diluted with DEPC-treated water, immersed in xylene, and air-dried for 20 min. The PixCell I LCM system (Arcturus Engineering, Mountain View, CA) was used for laser microdissection. The laser spot size and beam intensity were adjusted to microdissect pure populations of glomeruli, proximal tubules, and vessels under direct microscopic observation. For each specimen, 60 individual glomeruli, 60 individual tubules, and 10 individual vessels were captured sequentially on separate polymer films. For negative controls, caps were placed on the tissue sections in the same way but without activation of the laser pulse.

RNA Extraction and Reverse Transcription (RT)-PCR

Total RNA was extracted using TRI Reagent (Sigma), according to the single-step method reported by Chomczynski and Sacchi (20). The RNA was redissolved in 4 μl of DEPC-treated water, and RT was performed at 60°C for 30 min, using a GeneAmp EZ rTth RNA PCR kit (Perkin Elmer, Foster City, CA), in a final volume of 50 μl, with 2 μl of total RNA solution, 0.3 mM levels of each dNTP, 2.5 mM manganese diacetate, 0.45 μM levels of each primer, and 5 U of rTth DNA polymerase. The sense primer for RAGE corresponded to bp 1801 to 1820 (5′-GCCCTCCAGTACTACTCTCG-3′), and the antisense primer corresponded to bp 2043 to 2062 (5′-TGTGTGGCCACCCATTCCAG-3′). The sense and antisense primers for glyceraldehyde-3-phosphate dehydrogenase corresponded to bp 204 to 226 (5′-CAATGGAAATCCCATCACCATCT-3′) and bp 908 to 930 (5′-AATGAGCTTGACAAAGTGGTCGT-3′), respectively. The entire RT reaction product was used for PCR amplification. Forty cycles of amplification were performed with a thermal programmer (Perkin Elmer), as follows: denaturation at 94°C for 15 s and annealing and extension at 60°C for 30 s. The PCR products were then electrophoresed on a 1.5% agarose gel stained with ethidium bromide.

Characterization of Antibodies to CML and PENT

Characterization of the antibodies was performed by ELISA, using affinity-purified anti-CML IgG and affinity-purified anti-PENT IgG. In the former case, CML-BSA or native BSA (both at 5 μg total protein/well) was coated onto wells of Maxisorp plates (Nunc, Naperville, IL) in buffer (pH 9.6). After overnight incubation, wells were washed with PBS (0.02 M, pH 7.4), and unoccupied sites were blocked with PBS containing 1% BSA. ELISA was then performed using affinity-purified anti-CML IgG (0.88 μg/ml), followed by peroxidase-conjugated goat anti-rabbit IgG. Results in OD₄₉₀ nm were as follows: native BSA, 0.048; CML-BSA, 1.781. In other studies, immunoreactivity to CML-ovalbumin was 1.742, that for CML-ribonuclease was 1.090, and that for PENT-BSA was 0.051. Similarly, ELISA was performed using affinity-purified anti-PENT IgG (1 μg/ml), as described above. Results in OD₄₉₀ nm were as follows: native BSA, 0.042; PENT-BSA, 0.542; CML-ovalbumin, 0.047; CML-BSA, 0.047; CML-ribonuclease, 0.042.

Immunohistochemical Detection of AGE and RAGE in Normal Control Samples

With the exception of small amounts of CML and PENT in the intima of medium-sized and large arteries, normal control samples exhibited negative immunostaining for CML and PENT in the glomerular, tubular, and interstitial compartments (Figure 1, A and B). Diffuse low-level expression of RAGE was restricted to the podocytes of normal control subjects (Figure 1C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Normal kidney. (A) Staining for N^ϵ-carboxymethyllysine (CML) is negative in the glomeruli. Magnification, ×420. (B) Staining for pentosidine (PENT) is negative in the glomeruli. Magnification, ×420. (C) Low-level positivity for the receptor for advanced glycation end products (RAGE) is identified in the podocytes. Magnification, ×420.

Immunohistochemical Detection of AGE and RAGE in Diabetic Nephropathy

The immunohistochemical staining results for diabetic nephropathy are itemized in Table 1 and summarized in Table 2. CML was the major AGE identified in the GBM (42% of cases), mesangium (96%), TBM (85%) of both atrophic and nonatrophic tubules, and vessels (96%) (Table 2; Figure 2). Although PENT was identified in the mesangium in 77% of cases, it was rarely identified in the GBM (4%) (Figure 3A). Both CML and PENT were readily identified in mesangial nodules and in areas of glomerular hyalinosis (“fibrin cap lesions”). In contrast, PENT was the major AGE identified in areas of subcapsular fibrosis (65%) and interstitial fibrosis (92%) (Figure 3, B and C). PENT was less frequently identified in TBM (31%) and vessel walls (54%), compared with CML. Thus, in diabetic nephropathy, there was differential localization of PENT to interstitial collagen as well as mesangium and vascular basement membranes, whereas CML was more restricted to renal basement membranes known to contain collagen IV. Weak focal tubular epithelial staining for CML and PENT was identified in patients with moderate or severe diabetic nephropathy. There was marked uniform upregulation of RAGE expression in the podocytes of diabetic nephropathy (Figure 4A). The distribution of RAGE at the base of the podocytes (Figure 4B) was confirmed in serial sections stained for the podocyte-specific marker synaptopodin (Figure 4C). No RAGE expression was identified in other renal cell types.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Diabetic glomerulosclerosis. (A) Mild diabetic glomerulosclerosis. There is global staining for CML in the mesangium, glomerular basement membrane (GBM), and Bowman's capsule. Magnification, ×420. (B) Moderate diabetic glomerulosclerosis. There is more intense staining for CML throughout the mesangium, GBM, and Bowman's capsule. Magnification, ×420. (C) Moderate diabetic glomerulosclerosis. There is intense and diffuse tubular basement membrane immunoreactivity for CML. Magnification, ×420. (D) Moderate diabetic glomerulosclerosis. CML staining is present in the intima and media of a large artery. Magnification, ×180.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Diabetic glomerulosclerosis. (A) Severe diabetic glomerulosclerosis. PENT immunostaining is seen in the outer layers of a mesangial nodule. Magnification, ×550. (B) Severe diabetic glomerulosclerosis. Immunoreactivity for PENT is largely confined to the areas of subcapsular fibrosis, with only focal mesangial positivity. Magnification, ×420. (C) Moderate diabetic glomerulosclerosis. Prominent staining for PENT is identified in areas of interstitial fibrosis and glomerular subcapsular fibrosis. Magnification, ×250.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Diabetic glomerulosclerosis. (A) In mild diabetic glomerulosclerosis, RAGE immunoreactivity is markedly upregulated in the podocytes. Magnification, ×550. (B) With high-power magnification, staining for RAGE in a diabetic glomerulus has a punctate distribution at the base of the podocytes, above the GBM. Magnification, ×700. (C) The same field as in B reveals an almost identical distribution of synaptopodin immunoreactivity at the base of the podocytes. Magnification, ×700.

Correlations between the severity of diabetic nephropathy and the semiquantitative assessments of CML immunostaining in mesangium (P = 0.004), GBM (P < 0.05), and TBM (P = 0.003) were observed. There was no correlation between the severity of diabetic nephropathy and CML immunostaining in globally sclerotic glomeruli, interstitium, or vessel walls (P = NS). In contrast, there were significant correlations between the severity of diabetic nephropathy and the semiquantitative assessments of PENT staining in TBM (P = 0.035) and interstitium (P = 0.037) but not globally sclerotic glomeruli, GBM, mesangium, or vessel walls (P = NS). Among the patients with diabetic nephropathy, no correlation was found between the immunohistochemical quantification of AGE deposition and the severity of proteinuria or renal insufficiency.

RAGE mRNA Expression

The results of RT-PCR for RAGE are shown in Figure 5. RAGE mRNA expression was seen in glomeruli, but not tubules or vessels, in renal biopsy tissue from patients with diabetic nephropathy. No band was identified in glomeruli, tubules, or vessels from normal control kidney samples (data not shown). No expression was observed in the negative control samples.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Reverse transcription-PCR results for microdissected glomeruli, tubules, and vessels from a diabetic glomerulosclerosis specimen. A band corresponding to the predicted RAGE PCR product (262 bp) is identified in the microdissected glomeruli from a diabetic nephropathy specimen but not in the diabetic tubular or vascular compartments. Glyceraldehyde-3-phosphate dehydrogenase (G3PDH) was used as an internal positive control (727 bp). No expression is identified in the negative control sample. The molecular marker used was a 100-bp ladder.

Immunohistochemical Detection of AGE and RAGE in Nondiabetic Renal Diseases

In primary and secondary forms of FSGS and in hypertensive nephrosclerosis, CML and PENT were commonly detected in the distribution of partially or globally sclerotic glomeruli (Figure 6, A and B). However, in contrast to their distribution in diabetic nephropathy, these AGE were infrequently identified in the mesangium or GBM of nonsclerotic glomeruli (Table 2). CML and PENT were also present in the TBM of atrophic tubules but were rarely identified in nonatrophic tubules. In all of these conditions, CML and PENT staining was common in the vessel walls, particularly the intima (Figure 6C). CML was universally identified (100%) in the arteries of hypertensive nephrosclerosis samples, compared with the 86% prevalence of PENT in vessel walls. PENT, but not CML, was typically identified in the fibrotic interstitium. In all three of these sclerosing conditions, the levels of podocyte RAGE were indistinguishable from those of normal control samples (Figure 6D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Nondiabetic glomerulosclerosis. (A) Primary focal segmental glomerulosclerosis (FSGS). CML immunoreactivity is identified in an area of segmental glomerulosclerosis. Magnification, ×380. (B) Primary FSGS. The same glomerulus as in A also shows positive staining for PENT in the area of segmental sclerosis. Magnification, ×380. (C) Primary FSGS. There is no upregulation of podocyte RAGE in the same glomerulus. Magnification, ×380. (D) Hypertensive nephrosclerosis. CML staining is seen in globally sclerotic glomeruli, arterial walls, and the basement membranes of atrophic tubules. Magnification, ×380.

The differences between diabetic nephropathy and primary and secondary forms of FSGS and hypertensive nephrosclerosis were best observed when these categories were dichotomized into subgroups of 2+ or greater immunostaining versus 1+ or no immunostaining. As shown in Table 2, no patient with primary or secondary FSGS or hypertensive nephrosclerosis exhibited 2+ or greater immunostaining for either CML or PENT in the mesangium or GBM of nonsclerotic glomeruli. In contrast, 18 and 31% of diabetic nephropathy biopsies exhibited this level of CML or PENT immunostaining in the GBM and mesangium, respectively, of nonsclerotic glomeruli.

In lupus nephritis, PENT and, to a lesser extent, CML were found in active glomerular necrotizing and inflammatory lesions, including cellular crescents (Figure 7, A to C). However, the uniform distribution of mesangial and GBM staining characteristic of diabetic nephropathy was not observed. The intensity of staining in these active lesions was particularly high, often exceeding that observed for scarred sclerotic glomeruli. Podocyte RAGE expression was focally upregulated in the glomeruli of patients with active lupus nephritis (World Health Organization class IV) (Figure 7D). As shown in Table 2, lupus nephritis was the only category of nondiabetic renal disease studied in which CML or PENT staining of 2+ or more was identified in the mesangium or GBM of nonsclerotic glomeruli. This pattern of staining was most frequently observed in the distribution of active glomerular lesions with endocapillary hypercellularity, infiltrating neutrophils, and necrotizing features.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Lupus nephritis. (A) Lupus nephritis, class IV. There is strong PENT immunoreactivity in the distribution of the crescents and areas of active glomerular endocapillary proliferation. Magnification, ×80. (B) Lupus nephritis, class IV. PENT immunoreactivity is present in the mesangium and GBM. Magnification, ×420. (C) Lupus nephritis, class IV. Low-intensity CML immunoreactivity is present in the mesangium and some glomerular capillary walls. Magnification, ×420. (D) Lupus nephritis, class IV. RAGE expression is upregulated in the podocytes. Magnification, ×420.