K. Susztak, K. Sharma, and E. Böttinger designed the study. K. Susztak, E. Ciccone, P. McCue, K. Sharma, and E. Böttinger performed the experiments. K. Susztak, P. McCue, K. Sharma, and E. Böttinger analyzed the data and contributed to writing the paper.
The authors have declared that no competing interests exist.
Diabetic nephropathy (DNP) is a common complication of type 1 and type 2 diabetes mellitus and the most common cause of kidney failure. While DNP manifests with albuminuria and diabetic glomerulopathy, its progression correlates best with tubular epithelial degeneration (TED) and interstitial fibrosis. However, mechanisms leading to TED in DNP remain poorly understood.
We found that expression of scavenger receptor CD36 coincided with proximal tubular epithelial cell (PTEC) apoptosis and TED specifically in human DNP. High glucose stimulated cell surface expression of CD36 in PTECs. CD36 expression was necessary and sufficient to mediate PTEC apoptosis induced by glycated albumins (AGE-BSA and CML-BSA) and free fatty acid palmitate through sequential activation of src kinase, and proapoptotic p38 MAPK and caspase 3. In contrast, paucity of expression of CD36 in PTECs in diabetic mice with diabetic glomerulopathy was associated with normal tubular epithelium and the absence of tubular apoptosis. Mouse PTECs lacked CD36 and were resistant to AGE-BSA-induced apoptosis. Recombinant expression of CD36 in mouse PTECs conferred susceptibility to AGE-BSA-induced apoptosis.
Our findings suggest a novel role for CD36 as an essential mediator of proximal tubular apoptosis in human DNP. Because CD36 expression was induced by glucose in PTECs, and because increased CD36 mediated AGE-BSA-, CML-BSA-, and palmitate-induced PTEC apoptosis, we propose a two-step metabolic hit model for TED, a hallmark of progression in DNP.
The CD36 receptor may have a key role in the degeneration of the renal tubular epithelium, the first step in diabetic nephropathy.
The kidneys are often affected in people with diabetes. Around one in three people with type 1 (juvenile, or insulin-dependent) and one in ten people with type 2 (late onset, or non-insulin-dependent) diabetes will develop kidney disease (called diabetic nephropathy). Diabetic nephropathy is one of the leading complications of diabetes and is the leading cause of kidney failure worldwide. Some risk factors make it more likely that certain people with diabetes will develop kidney disease—for example, kidney disease occurs more often in patients from South Asian or African backgrounds, in men, in patients with poor control of their blood sugar levels, and in those with high blood pressure or who smoke. However, the details of how, exactly, diabetes damages the kidneys are not clear.
They studied samples taken from the kidneys of humans and mice with and without diabetes and looked at the effects of high glucose concentrations on the cells in the kidneys. They found that in one part of the human kidneys high glucose caused a change in the cell surface causing an increase in a protein called CD36. This change occurred in the samples from people with diabetes, but did not occur in the samples from mice with diabetes. The investigators also found that some substances that are often found in the blood of people with diabetes could join to CD36; in doing so, these substances triggered the death of these cells, which is one of the first steps that occurs in diabetic nephropathy.
This particular protein (CD36) could have a central role in triggering diabetic nephropathy. Although there are no immediate clinical implications of this research for the treatment of people with kidney problems, this research helps in understanding how high glucose damages the kidney. In particular, it highlights how important it is to keep blood glucose levels as normal as possible.
Medline Plus's article on diabetic nephropathy:
Diabetes UK's online information centre:
National Diabetes Information Clearinghouse:
National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Animal Models of Diabetic Complications Consortium (AMDCC):
Diabetic nephropathy (DNP) is a serious and common complication of type 1 and type 2 diabetes mellitus, leading to end-stage renal failure in up to 30% of individuals with diabetes. Early abnormalities of DNP affect glomeruli and include an increase in glomerular filtration rate, microalbuminuria, glomerular hypertrophy, and thickening of the glomerular basement membrane, followed by expansion of mesangial extracellular matrix and glomerulosclerosis [
Combining detailed renal phenotype analysis with gene expression profiling of hyperglycemic mouse models of type 1 (streptozotocin [STZ]) and type 2 (db/db) diabetes, we recently reported that decreased mRNA levels of CD36 in kidneys were strongly correlated with albuminuria [
Here we report a novel functional role for CD36 scavenger receptor and AGE and FFA palmitate (PA) in tubular epithelial apoptosis associated with TED and progression of DNP. Specifically, we show that glucose stimulates CD36 cell surface expression in proximal tubular epithelial cells (PTECs), and increased CD36 renders PTECs susceptible to both AGE- and PA-induced PTEC apoptosis by mediating sequential activation of src kinase, proapoptotic p38 MAPK, and caspase 3. Based on these findings, we propose a new two-step metabolic hit model for TED in the progression of DNP.
Kidneys were obtained from 28-wk-old C57BLKS/J-leprdb/db, STZ-treated C57BL/6J, or STZ-treated 129SvJ mice and from age-matched control C57BLKS/J-leprdb/m, C57BL/6J, and 129SvJ mice as described [
Human proximal tubular cell line HK-2 and murine collecting duct cell line M1 were purchased from American Type Culture Collection (Manassas, Virginia, United States) and cultured according to the vendor's instructions. Mouse proximal tubular cell line MCT was provided by Fuad Ziyadeh (University of Pennsylvania, Philadelphia, Pennsylvania, United States). Transfections were performed with Fugene 6 (Roche Diagnostics, Indianapolis, Indiana, United States) according to manufacturer's protocol. CD36-containing plasmid was a kind gift of Nada Abumhrad (SUNY at Stony Brook, New York, United States). Cells were also co-transfected with EGFP (Clontech, Franklin Lakes, New Jersey, United States) to assess transfection efficiency. Cells were serum starved in 0.2% serum containing DMEM (1 g/l glucose) for at least 24 h prior to stimulation with AGE–bovine serum albumin (BSA), glucose, or FFA.
Quantitative real-time PCR analysis of mouse and human CD36, HPRT1, and beta actin was performed as described previously [
Primary antibodies specific for the following proteins were used: monoclonal mouse anti-CD36 antibody, clone FA 6–152 (IgG) (Immunotech, Fullerton, California, United States), clone SMO (IgM) (Santa Cruz Biotechnology, Santa Cruz, California, United States), rabbit polyclonal anti-CD36 (Santa Cruz Biotechnology), rabbit polyclonal anti-aquaporin1, anti-aquaporin2, anti-Na/K/2Cl (Chemicon, Temecula, California, United States), rabbit polyclonal phospho38/MAPK and mouse monoclonal p38 (Cell Signaling Technology, Beverly, Massachusetts, United States), rabbit polyclonal p-src (Y418) (Biosource, Camarillo, California, United States), and mouse monoclonal anti-tubulin (Sigma, St. Louis, Missouri, United States). Immunostaining was performed on frozen sections with FITC- and Cy3-labeled secondary antibodies (Jackson Laboratories, USA), or on paraffin-embedded sections with immunoperoxidase, as described earlier [
Cells were incubated in 0.5 mM EDTA in PBS at 37 °C for 20 min, scraped, and then washed with 1% fetal bovine serum. Cells were then exposed to monoclonal anti-CD36 IgG FA6 (5 μg/ml), or control mouse IgG1 (5 μg/ml) (Sigma), for 45 min on ice in the presence of 10% fetal bovine serum then washed with PBS. This was followed by an incubation with phycoerythrin-conjugated goat anti-murine secondary antibody (Southern Biotechnology, Birmingham, Alabama, United States) 1:50 for 45 min on ice. Cells (1 × 104) were analyzed by using a SCAN flow cytometer (BD, Franklin Lakes, New Jersey, United States), with appropriate gating. Flow cytometry data were analyzed using Cellquest (BD).
Briefly, to prepare AGE-BSA, essentially fatty-acid-free and endotoxin-free BSA (250 mg/ml) was incubated at 37 °C for 2, 5, and 10 wk with D-glucose (500 mM) in a 0.4-M phosphate buffer containing EDTA, ampicillin, Fungazone, polymixin B, and protease inhibitors. Control preparations were treated identically except that glucose was omitted. Carboxymethyl-lysine (CML)–BSA was prepared as described earlier [
Palmitic acid (P5585), oleic acid, and FFA-free low-endotoxin BSA (A8806) were purchased from Sigma. Palmitic acid was dissolved at 12 mM in PBS containing 11% fatty-acid-free BSA, sonicated for 5 min, shaken overnight at 37 °C, and sonicated for 5 min again [
In situ detection of DNA fragmentation was performed using the ApoTag TUNEL assay following the manufacturer's protocol (Intergen, Purchase, New York, United States) [
Human kidney tissues (ten controls, ten with diabetic nephropathy, and ten with focal segmental glomerulosclerosis [FSGS]) were obtained from archived kidney biopsy samples or from discarded nephrectomy specimens. All diabetic samples were from patients with biopsy-proven advanced DNP with serum creatinine ranging from 1.7 to 5.6 mg/dl (151 to 444 μM/l), heavy proteinuria (3+ by dipstick or 3–6 gr/d), and hypertension. All patients with FSGS were from patients with creatinine levels of 1.7 to 4.9 mg/dl (151 to 435 μM/l), heavy proteinuria (3+ by dipstick), and hypertension. The diagnosis of FSGS was made on Periodic acid–Schiff staining in the absence of immunodeposits on electron microscopy. The diagnosis of diabetic nephropathy was based on the presence of diabetes, proteinuria, and the characteristic light microscopy findings. Institutional Review Board approval was obtained for procurement of kidney specimens at the Thomas Jefferson University Hospital.
Data are reported as mean and standard error of the mean (SEM) for continuous variables. All cell culture experiments were performed at least three times and summarized. Standard software packages (SPSS and Excel for Windows) were used to provide descriptive statistical plots (unpaired
Using microarray-based gene expression profiling on whole kidney RNA together with supervised clustering methods, we previously identified and validated gene expression patterns for molecular classification of diabetic mice with albuminuria and mesangial expansion [
(A and B) Indirect double-immunofluorescence labeling of kidney sections from non-diabetic control (A) and diabetic (B) mice with anti-CD36 (green) and proximal tubular marker anti-aquaporin1 (red).
(C and D) Double labeling of non-diabetic control mice with anti-CD36 (green) and loop-of-Henle marker sodium potassium chloride cotransporter anti-NKCC (red) (C) and collecting duct marker aquaporin2 (red) (D) (arrow depicts colocalization of anti-CD36 and anti-aquaporin2 staining).
(E and F) Double labeling of human kidney sections from control individuals (E) and individuals with diabetes with DNP (F) using anti-CD36 (green) and anti-aquaporin1 (red).
(G) Higher-magnification image of (F) with arrows depicting colocalization of anti-CD36 and anti-aquaporin1. (Note that anti-CD36 labeling is heterogeneous: staining is isolated proximal tubular cells.)
(H–J) Representative images of anti-CD36 immunoperoxidase staining of sections of normal human kidney (H), human kidney with DNP (I), and human kidney with FSGS (J). Arrow in (I) depicts proximal tubular epithelial staining.
(K) CD36 PTEC expression score derived from blinded, semi-quantitative analysis of distribution and intensity of proximal tubular CD36 staining of human biopsy samples from ten normal control, ten DNP, and ten FSGS kidneys and the result shown on a dot plot. Significance was calculated by Wilcoxon Rank Sum Test, and PTEC scores for DNP kidneys were significantly different from those of FSGS kidneys and normal human kidneys.
Periodic acid–Schiff–stained sections of kidneys from mice exposed to type 2 diabetes (db/db mice) for 20 wk (
(A and B) Periodic Acid–Schiff staining of diabetic mouse kidney (28-wk-old C57BLKS/J-leprdb/db) (A) and human DNP kidney (B). Arrowheads denote glomeruli with advanced mesangial expansion and glomerulosclerosis; arrows depict normal proximal tubule in diabetic mouse (A) and TED in human with DNP (B).
(C) TUNEL assay (green) and anti-CD36 (red) double labeling of human DNP. Arrows indicate apoptotic, CD36-positive tubular epithelial cells.
(D) TUNEL assay (green) and anti-aquaporin1 (red) double labeling of human DNP. Arrows depict TUNEL-positive and aquaporin1-positive PTECs.
(E) Dot plot indicates the number of TUNEL-positive tubular cells per 100 total tubular cells in kidneys of control (CTL) and diabetic (DM) mice and humans, as indicated.
CD36 has been shown to mediate apoptosis signaling induced by TSP-1 in endothelial cells [
High ambient glucose has been shown to induce CD36 protein synthesis in macrophages [
(A) Relative CD36 mRNA abundance determined by quantitative real-time PCR in human PTEC line HK-2 treated with 30 mM D-glucose (open bars) or control L-glucose (black bars) for 4 and 24 h following maintenance of cells in 5 mM D-glucose medium. Bars represent mean ± SEM of three to five repeat experiments. Numbers on top of bars indicate significant
(B) Bar graphs show experiment as described under (A), using mouse collecting duct cell line M1 instead of human HK-2 PTECs. The relative expression of CD36 was normalized to internal control housekeeping genes
(C) Relative cell surface expression of CD36 protein determined by FACS in M1 cells (open bars) and HK-2 cells (black bars) maintained in 5 mM D-glucose medium (CTL), or in medium containing 30 mM D-glucose (D-gluc) or L-glucose (L-gluc) for 72 h. (Original FACS histograms are provided in
(D) Immunoblot showing CD36 protein levels in human HK-2 PTECs maintained in control 5 mM D-glucose (CTL), or after stimulation for 72 h with 30 mM L-glucose (L-gluc) or D-glucose (D-gluc), as indicated. Tubulin is shown for loading control. All data represent at least four independent repeat experiments.
AGE albumin [
Bar graphs show mean ± SEM of apoptotic nuclei, visualized by DAPI staining and normalized to 100 total cells, in human HK-2 PTECs. Data are derived from three independent repeat experiments. Numbers on top of bars indicate significant
(A) Cells were treated for 48 h with control BSA (40 μM), TSP-1 (1 μg/ml), and AGE-BSA modified for 2, 5, or 10 weeks (AGE-BSA2, AGE-BSA5, and AGE-BSA10, respectively) in the absence or presence of control IgG (10 μg/ml) or anti-CD36 neutralizing antibody (10 μg/ml), as indicated.
(B) Cells were treated with control BSA (40 μM), or CMLmin-BSA at 0.5, 1, 2, 5, and 10 μM, in the absence or presence of anti-CD36 neutralizing antibody, as indicated.
(C) Cells were treated with monounsaturated FFA oleic acid (OA) or PA at increasing concentrations, in the absence or presence of control IgG (10 μg/ml) or anti-CD36 neutralizing antibody (10 μg/ml), as indicated.
Among the most abundant glucose-modified proteins detectable in the plasma of diabetic individuals are CML proteins [
CD36 has been shown to transport fatty acids in adipocytes [
CD36 has previously been shown to trigger the activation of p59fyn, p38 MAPK, and caspase 3 (GeneID: 836) in response to thrombospondin in endothelial cells [
(A and C) Immunoblots show levels of (A) phosphorylated (Y418) src kinase and tubulin or (C) phosphorylated p38 MAPK (pp38) and total p38 MAPK (p38) in HK-2 cells treated with AGE-BSA5 (40 μM) in the absence or presence of control IgG or anti-CD36 neutralizing antibody (10 μg/ml) for different time periods, as indicated.
(B and D) As shown in (A) and (C), except HK-2 cells were treated with PA (150 μM) instead of AGE-BSA5.
(E and F) Bar graphs demonstrate mean ± SEM of caspase 3 activity in three independent repeat experiments. Caspase 3 activity was measured by quantitative ELISA in HK-2 cells after 18 h of stimulation with AGE-BSA5 and PA, as per manufacturer's protocol. Numbers on top of bars indicate significant
(G) Bar graphs demonstrate number of apoptotic nuclei of HK-2 cells, normalized to 100 total cells, treated with AGE-BSA5 (40 μM) or PA (150 μM) in the absence (black bars) or presence of pan-caspase inhibitor (z-VAD-fmk [100 μM]; open bars), caspase 3 inhibitor (z-DEVD-fmk [20 μM]; first striped bars), caspase 9 inhibitor (z-LEHD-fmk (20 μM); gray bars), or chemical inhibitors of p38 MAPK (SB203580 [10 μM]; second striped bars). Mean ± SEM of three independent repeat experiments is presented. Numbers on top of bars indicate the significant
In contrast with CD36-positive human HK-2 PTECs, we found that treatment of CD36-negative mouse MCT PTECs with AGE-BSA had no effect on rates of apoptotic nuclei (data not shown). To test whether CD36 was sufficient to mediate AGE-BSA-induced apoptosis, we transfected CD36-negative mouse MCT PTECs with a plasmid expressing human CD36 or empty control vector, followed by treatment with control BSA or AGE-BSA5. AGE-BSA5 treatment had no significant effect on rates of apoptotic nuclei in MCT PTECs transfected with control vector (
(A–D) Representative images show DAPI (A and C) and FITC (B and D) labeling of CD36-negative MCT cells treated with 40 μM AGE-BSA5 for 24 h after co-transfection with green fluorescent protein plasmid pEGFP and pcDNA3.1 empty control vector (A and B), or pEGFP and CD36 expression plasmid pcDNA3.1/CD36 (C and D).
(E) The dot plot shows results of four independent experiments where apoptotic nuclei per 100 total cells were quantitated in transfected cell cultures with or without treatments as indicated.
Advanced diabetic nephropathies in humans with type 1 or type 2 diabetes are uniformly characterized by TED, or tubular atrophy, and IF leading to renal failure [
Our conclusions are supported by several key observations. First, we identify a new functional role for CD36 as an essential mediator of proximal tubular epithelial apoptosis, inducible by AGE-BSA, CMLmin-BSA, and FFA PA. Previous reports demonstrated a role for CD36 in mediating apoptosis induced by TSP-1 in endothelial cells and ox-LDL in macrophages [
Almost all TUNEL-positive apoptotic tubular epithelial cells showed increased expression of CD36, suggesting a strong correlation between upregulation of CD36 expression and increased apoptosis in PTECs specifically in human diabetic kidney in vivo. Importantly, biopsy samples from cases of FSGS that were matched for degree of proteinuria, renal function, and hypertension were characterized by TED, IF, and increased tubular epithelial apoptosis; however, proximal tubular CD36 expression was similar to that in normal human control kidney. Therefore, CD36 expression in PTECs is specifically associated with the diabetic condition and appears to be independent of degree of proteinuria and renal failure. Indeed, increased CD36 expression in PTECs in human DNP in vivo may be caused by hyperglycemia, as we show that high glucose concentration stimulates CD36 expression in vitro. It is intriguing that CD36 expression was not detected in PTECs in diabetic mice with longstanding hyperglycemia in vivo, although underlying mechanisms for the species-dependent differential regulation of CD36 in PTECs in vivo and in vitro between mouse and human remain unclear at this time. Comparisons of human
Third, to our knowledge, our report provides the first controlled study demonstrating increased apoptosis specifically in PTECs in DNP with TED and IF. These findings are consistent with a recent uncontrolled case series of five patients with DNP [
In conclusion, we report a new functional role for CD36 scavenger receptor in tubular epithelial apoptosis associated with tubular degeneration and progression of DNP. Specifically, we show for the first time that both AGE and FFA PA induce PTEC apoptosis through CD36-mediated activation of src kinase, p38 MAPK, and caspase 3. Because high glucose stimulates CD36 expression in human PTECs and because CD36 expression is increased in apoptotic tubular epithelial cells in human DNP, we propose a two-step metabolic hit model relevant for TED, a hallmark of progression of human DNP.
Flow cytometric analysis of (A) human (HK-2) and (B) murine (M1) tubular epithelial cells incubated with control IgG (green curve) or with anti-CD36 antibody (FA6) (black curve) in medium containing 5 mM glucose (empty curve) or in medium containing 30 mM glucose (red curve) for 3 d.
(45 KB PPT).
The LocusLink (
We thank Dr. Michael Brownlee for helpful discussions and advice in preparing AGE-BSAs, Dr. Paul J. Thornalley for his advice on glycated proteins, and Dr. Nada Abumrad for providing CD36 expression plasmid. We are grateful to Mr. Chih-Kang Huang and Mrs. Yutong Zhang for expert technical support. K. Susztak was supported by the National Kidney Foundation and by an Albert Einstein College of Medicine Clinical Investigator Fellowship. This work was supported by National Institutes of Health grants R01 DK056077 and DK060043 (E. P. Böttinger), R01 DK053867 (K. Sharma) and U01 DK060995 (E. P. Böttinger and K. Sharma).
advanced glycation end
advanced glycation end–bovine serum albumin glycated for 5 wk
bovine serum albumin
carboxymethyl-lysine
bovine serum albumin with minimal carboxymethyl-lysine content
diabetic nephropathy
long-chain free fatty acid
focal segmental glomerulosclerosis
interstitial fibrosis
MAP kinase
modified (oxidized) low-density lipoprotein
palmitate
proximal tubular epithelial cell
standard error of the mean
streptozotocin
tubular epithelial degeneration
thrombospondin-1