Poly(ADP-Ribose) Polymerase Inhibitors Ameliorate Nephropathy of Type 2 Diabetic Lepr^db/db Mice

PJ-34 and INO-1001 were synthesized at Inotek Pharmaceuticals (Beverly, MA). PJ-34 was purchased from Alexis Biochemicals (Axxora, San Diego, CA).

Male db/db (Lepr^db/db) mice, together with nondiabetic control db/m mice on C57BLKs/J background (Jackson Laboratories, Bar Harbor, ME), were used. INO-1001 and PJ-34 treatment were initiated at 5 weeks of age. In sterile water that was sweetened with NutraSweet, 4.8 g/l INO-1001 and 2.4 g/l PJ-34 was dissolved. Control animals received sweetened water only. The average inhibitor consumption was 60 mg/kg INO-1001 and 30 mg/kg PJ-34. All animal protocols and procedures were approved by the institutional animal care and use committee at the Albert Einstein College of Medicine and at the Mount Sinai School of Medicine.

For determination of urinary albumin excretion, mice were placed in individual metabolic cages (Nalgene Nunc, Rochester, NY) and urine collected for 16 h. Urinary albumin was measured using an enzyme-linked immunosorbent assay (ELISA) specific for mouse albumin (Albuwell; Exocell, Philadelphia, PA), and urine creatinine was determined using Creatinine Companion (Exocell). Blood glucose was measured with Glucometer Elite (Bayer) after 6 h of fasting. Animals were killed at 17–21 weeks of age, when significant albuminuria and mesangial expansion occurs.

For analysis of glomerular histology, formalin-fixed, paraffin-embedded kidney tissue sections were stained with periodic acid Schiff (PAS) reagent and coded and read by an investigator who was unaware of the experimental protocol. Mesangial matrix expansion was evaluated semiquantitatively on 50 glomeruli per kidney, using a score of 1 for minimal, 2 for mild mesangial matrix expansion, 3 for moderate, and 4 for diffuse mesangial matrix expansion.

Podocyte number per glomerular cross section was determined on 4-μm frozen sections double stained with WT-1 (rabbit) (Santa Cruz Biotechnology, Santa Cruz, CA) and synaptopodin (mouse) as described earlier (30) and developed with fluorescein isothiocyanate–conjugated anti-mouse and Cy3-conjugated anti-rabbit secondary antibodies (Jackson Immunologicals). Only cells stained positive for both synaptopodin and WT-1 were counted. NFκB p50 and synaptopodin double immunostaining was performed similarly using NFκB p50 (rabbit) (Santa Cruz Biotechnology) and anti synaptopodin (mouse) antibodies.

Poly(ADP) ribosylated staining was performed using mouse anti–poly(ADP)-ribose (Alexis) antibody. Briefly, after deparaffinization, the tissue was treated in 10 mmol/l sodium citrate and microwaved. After blocking, sections were incubated with the primary antibody 1:50 dilution overnight at 4°C. Signal was developed using ABC Vectastain Elite kit (Vector Laboratories, Burlington, CA).

Whole-kidney samples were homogenized in radioimmunoprecipitation assay buffer. Proteins were separated on 10% SDS gels and transferred to polyvinylidine fluoride membranes. Blots were incubated using mouse anti–poly(ADP)-ribose antibody (1:1,000 dilution), followed by peroxidase-conjugated goat anti-mouse antibody and developed with ECL chemiluminescence kit (Pierce). Blots were reprobed with mouse anti-αtubulin antibody (Sigma Aldrich) as a loading control. Western blot analyses on podocyte lysates were performed similarly by using anti-PAR (Alexis), phospho-IκBα (inhibitor of κBα; Santa Cruz Biotechnology), and tubulin (Sigma) antibodies.

Cultivation of conditionally immortalized mouse podocytes was performed as described (31). Briefly, cells were propagated in the undifferentiated state on type 1 collagen at 33°C in RPMI-1640 in the presence of 10% fetal bovine serum (Hyclone) and 20 units/ml interferon-γ (Sigma Chemical, St. Louis, MO). To induce differentiation, cells were maintained at 37°C without interferon for 10 days. All experiments were performed using differentiated podocytes. Before the experiment, cells were incubated overnight in RPMI with 0.2% fetal bovine serum containing 5 mmol/l glucose.

Apoptotic nuclei of cultured podocytes were detected on paraformaldehyde-fixed cells using DAPI (4′,6-diamidino-2-phenylindole) staining (1 μg/ml) for 10 min. Cells were analyzed under a fluorescence microscope and assessed for chromatin condensation and segregation. Caspase-3 activity was measured in podocyte extracts using EnzCheck Caspase-3 Assay kit (Molecular Probes) following the manufacturer’s protocol.

Nuclear extracts were prepared using Transfactor DB Nuclear extraction kit (Clontech BD Bioscience) following the manufacturer’s protocol. Western blotting was performed using NFκB p50 (Santa Cruz Biotechnology) antibody. NFκB p50 nuclear binding assay was performed using TransFactor Colorimetric kit (Clontech BD) according to the manufacturer’s protocol.

For kinetic studies, podocytes were plated on 24-well plates, loaded with 50 μmol/l carboxymethyl-H₂-dichlorofluorescein diacetate (CM-H₂DCFDA; Invitrogen) for 30 min at 37°C, washed twice with PBS, and incubated with PJ-34 (3 μmol/l) or INO-1001 (200 nmol/l) for 30 min, followed by stimulation with 30 mmol/l d-glucose or 30 mmol/l l-glucose. Fluorescence was analyzed using Wallac Victor² Fluorescence Plate Reader (7).

Podocytes, grown on coverslips, were loaded with 1 μg/ml JC-1 (Invitrogen) for 20 min at 37°C. Coverslips were rinsed with PBS five times, and cells were maintained at 37°C and either left untreated or pretreated with INO-1001 (200 nmol/l) or PJ-34 (3 μmol/l) for 30 min, followed by incubation in 5 or 30 mmol/l d-glucose. To view J-aggregates, excitation and emission were recorded with a charged-coupled device camera at 543–560 nm and at 480–530 nm for monomeric. For each coverslip, five images of each condition were analyzed (32). Flow cytometry cells were incubated in 5 or 30 mmol/l d-glucose in the presence or absence of the inhibitors, trypsinized, loaded with 5 μg/ml JC-1, and washed twice with PBS, and 10,000 cells/condition were analyzed with FACSscan (AECOM Shared Scientific Facilities).

All results are expressed as means ± SE. The data were analyzed by Student’s t test. Differences were considered statistically significant when the P values were <0.05.

We evaluated the effect of two structurally unrelated PARP inhibitors, PJ-34 and INO-1001, on the development of diabetic nephropathy of Lepr^db/db mice. The Lepr^db/db mice suffer from obesity and develop diabetes around 8 weeks of age (33,34). PARP inhibitors were administered in the drinking water, starting at 5 weeks of age. Blood levels of INO-1001 were also determined in serum samples using high-performance liquid chromatography method, and it ranged between 46 and 109 nmol/l. Treatment with the PARP inhibitors did not cause any excess mortality or any obvious phenotype change in diabetic and control mice; there was no significant difference in the body weight of control and inhibitor-treated db/db mice (Table 1). Neither PJ-34 nor INO-1001 inhibited the development of diabetes in this type 2 diabetes model (defined as fasting blood glucose >250 mg/dl). However, PJ-34–treated mice had lower serum glucose values than control db/db animals (501 ± 19 vs. 595 ± 2.5 mg/dl, respectively) (Table 1). There was no statistical difference in the serum glucose values in the control and INO-1001–treated db/db mice. In summary, oral administration of both inhibitors were well tolerated by the animals.

Increased PARP activity has been described in various tissues in diabetes. Using an antibody that specifically detects poly(ADP) ribosylated proteins, products of the active enzyme, various poly(ADRP) ribosylated proteins can be visualized as multiple bands on Western blots. We found a marked increase in the amount of poly(ADP) ribosylated proteins in renal extracts of 17- and 21-week-old diabetic (db/db control) mice compared with nondiabetic db/m mice (Fig. 1A). Diabetic mice treated with the PARP inhibitors had lower amount of poly(ADP) ribosylated proteins in kidney extracts compared with untreated diabetic mice. Quantification/densitometry of the poly(ADP) ribosylated proteins compared with the loading control tubulin is shown in Fig. 1B. To determine in which cell type PARP is activated, we performed immunostaining with an anti–poly(ADP) ribosylated antibody in kidney sections of control (db/m), diabetic db/db, and db/db mice treated with PJ-34. There was an overall increase in the amount of poly(ADP) ribosylated–modified proteins in kidneys of diabetic mice. As shown in Fig. 1C, poly(ADP) ribosylated–modified proteins were also increased in glomerular podocytes of diabetic mice (Fig. 1C). This increase was already evident at 8 weeks of age (online appendix Fig. 1 [available at http://diabetes.diabetesjournals.org]) at the time of development of diabetes. PJ-34–and INO-1001–treated animals had lower glomerular poly(ADP) ribosylated staining; however, some tubular cells remained positive. In summary, we found increased PARP activity in the diabetic kidneys, particularly in podocytes.

Next, we examined the development of diabetic nephropathy in control and PARP inhibitor–treated db/db mice. As shown in Fig. 2, diabetic db/db mice treated with INO-1001 or PJ-34 had significantly less mesangial expansion than control db/db mice. Semiquantitative analysis of PAS-stained kidney sections showed that 21-week-old diabetic mice treated with INO-1001 had a mesangial expansion score (on a scale of 1–4) of 2.3 ± 0.4 vs. 3.6 ± 0.2 of control db/db mice (Table 2). Similarly, 17-week-old diabetic mice treated with PJ-34 also had significantly less glomerular disease than untreated diabetic mice (1.4 ± 0.3 vs. 2.9 ± 0.3 arbitrary units). Albuminuria, a hallmark of diabetic renal disease, was also decreased in mice treated with PARP inhibitors. As quantitated by mouse albumin-specific ELISA, control db/db mice had 1,036 ± 180 μg/mg (albumin/creatinine) albuminuria compared with mice treated with INO-1001, which had only 440 ± 71 μg/mg (albumin/creatinine) albuminuria (>50% reduction). A similar reduction was observed following PJ-34 treatment (Table 2). In summary, both inhibitors ameliorated diabetic mesangial expansion and albuminuria, the two hallmarks of diabetic nephropathy in the db/db model of type 2 diabetes.

Since depletion of glomerular podocytes correlates with albuminuria and disease progression in diabetic nephropathy (7,35), we next evaluated podocyte number in control and inhibitor-treated diabetic mice. Diabetic db/db mice have significantly lower podocyte number per glomerular profile than age-matched control db/m mice; their numbers amounted at 7.5 ± 1 in 17-week-old db/db mice vs. 10.2 ± 0.4 in db/m animals (Table 2). Diabetic mice treated with PJ-34 had an average of 9.6 ± 1.4 podocyte number per glomerular profile, which was statistically different from untreated db/db mice. Similar results were obtained with INO-1001 treatment (Table 2). Thus, both PARP inhibitors prevented podocyte depletion in diabetic animal models.

Next, we determined whether high glucose leads to PARP activation in vitro by using cultured murine podocytes. We found increased levels of poly(ADP) ribosylated–modified proteins in podocytes as early as 60 min after incubation in 30 mmol/l d-glucose (Fig. 3). Both INO-1001 and PJ-34 significantly decreased the amount of poly(ADP) ribosylated–modified proteins. In summary, we found increased poly(ADP) ribosylated–modified proteins, likely indicating increased PARP activity, in podocytes following incubation in hyperglycemic milieu.

We have previously reported that high glucose leads to increased apoptosis of cultured glomerular epithelial cells. Since PARP is an important regulator of apoptosis, we determined the effect of PARP inhibitors INO-1001 or PJ-34 on glucose-induced apoptosis. We observed an increased podocyte apoptosis rate made evident by increased nuclear condensation of cells incubated in a medium containing 30 mmol/l d-glucose (4.2 ± 0.8%) compared with cells incubated in 5 mmol/l d-glucose (1.3 ± 0.2%). Preincubation of cells with 200 nmol/l INO-1001 or 3 μmol/l PJ-34 significantly decreased the degree of glucose-induced apoptosis (Fig. 4A). Similarly, both PJ-34 and INO-1001 inhibited hyperglycemia-induced caspase-3 activation (Fig. 4B). In summary, PARP inhibitors blocked high-glucose–induced podocyte apoptosis.

PARP has emerged as an important factor in the pathophysiology of reactive species–mediated cell injury. Recent data suggest that PARP could be either a target or an upstream regulator of ROS generation (36,37). Therefore, we next investigated the effect of INO-1001 and PJ-34 on the glucose-induced oxidative stress of glomerular epithelial cells. Incubation of podocytes in high-glucose medium led to an increase of DCF (2′,7′-dichlorofluorescein) fluorescence, which was evident as early as 20 min; however, we observed a slight increase even at earlier time points (Fig. 5B). We found that 30 mmol/l d-glucose increased intracellular ROS amount by ∼60% (after 4 h incubation) compared with cells incubated in 5 mmol/l glucose with or without 25 mmol/l l-glucose, an osmotic control (Fig. 5B). We, similar to other investigators (38), observed an increase in baseline fluorescence of DCF over time, as the dye is light sensitive; however, the glucose-induced increase in fluorescence was fully sensitive to the NADPH oxidase inhibitor dihydrophenyl iodonium (Fig. 5A). We found that INO-1001 (200 nmol/l) and PJ-34 (3 μmol/l) treatment abolished glucose-induced intracellular ROS generation of cultured murine podocytes (Fig. 5A and B).

We next investigated the effect of INO-1001 and PJ-34 on mitochondrial function by determining their effect on mitochondrial membrane potential (32). We found that the incubation of podocytes in 30 mmol/l d-glucose for 4 h caused depolarization of the mitochondria made evident by a shift to the green color of the mitochondrial membrane potential dye JC-1 (Fig. 5C and D). Preincubation of cells with INO-1001 or PJ-34 prevented mitochondrial membrane depolarization of high-glucose–treated glomerular epithelial cells (Fig. 5E and F). Quantification of the red and green fluorescence was performed via fluorescence-activated cell sorter analysis on 10,000 JC-1–loaded podocytes. The mean red-to-green fluorescence ratio was 3.6 in control cells, which decreased to 1.2 in cells incubated in 30 mmol/l d-glucose for 18 h, indicating the shift toward the green fluorescence. Preincubation of cells with PJ-34 or INO-1001 prevented the decrease in red fluorescence (Fig. 5G), indicating that both inhibitors protect from mitochondrial dysfunction induced by elevated extracellular glucose concentration.

Close association between NFκB and PARP has been described in various cell types (22). We examined whether or not we could observe NFκB activation in vivo in podocytes of diabetic db/db mice. We performed immunostaining on kidney sections of control and PJ-34–treated diabetic and nondiabetic samples with NFκB p50 antibody. We found increased nuclear NFκB p50 staining in diabetic db/db mice compared with nondiabetic db/m mice, indicating an increase of NFκB activity. Double immunostaining with anti-synaptopodin and anti–NFκB p50 showed increased nuclear p50 staining in control db/db mice (Fig. 6A). Diabetic mice treated with PJ-34 showed less p50 staining compared with control db/db mice (Fig. 6A).

Next, we determined whether hyperglycemia leads to increased NFκB activation of podocytes in vitro. We detected an increase in the amount of phosphorylated IκBα protein levels in cultured podocytes after 60 min incubation in 30 mmol/l d-glucose (Fig. 6B), indicating the activation of the NFκB pathway. Preincubation with PJ-34 or INO-1001 blocked IκBα phosphorylation. In addition, preincubation of cells in apocynin (25 μmol/l), an inhibitor of NADPH oxidase shown to completely block glucose-induced ROS generation (7,38), also prevented the increase in IκBα phosphorylation. Next, we tested the effect of INO-1001 on nuclear translocation NFκB p50 subunit. Nuclear and cytosolic extracts were prepared from podocytes treated with d-glucose for 18 h in the presence or absence of INO-1001. We found increased nuclear NFκB p50 protein expression in cells treated with d-glucose compared with control l-glucose. Similarly, treatment with transforming growth factor-β also resulted in increased nuclear NFκB p50 levels (Fig. 6C). Pretreatment with INO-1001 decreased nuclear NFκB p50 levels, likely inhibiting the translocation of this protein in the absence of phosphorylated IκBα. To further confirm the glucose-induced NFκB p50 activation, we performed an ELISA-based NFκB p50 DNA binding assay (Fig. 6D). Our data indicate that NFκB is a target of hyperglycemia-induced PARP activation in podocytes.

Based on in vivo and in vitro evidence presented in this report, we propose that PARP activation plays an important role in the development of diabetic glomerulopathy in the type 2 diabetes model of Lepr^db/db mice. PARP activity is increased in db/db mice, which was made evident by an increase in poly(ADP) ribosylated proteins in the kidneys, specifically in glomerular podocytes. Pharmacological inhibition of PARP by two structurally different orally available potent PARP inhibitors (PJ-34 and INO-1001) decreased the development of diabetic glomerular expansion and albuminuria, which are two major hallmarks of diabetic nephropathy.

We propose that prevention of podocyte loss by PARP inhibitors contributes to their observed protective effect. Reports from us and other groups (39–41) indicate that podocyte density plays a central role in the development and progression of diabetic and focal segmental glomerulosclerosis. Podocyte apoptosis and dysfunction can be detected early in the development of diabetic nephropathy, and inhibition of apoptosis was associated with lower albuminuria and less mesangial expansion (7). While we focused our experiments around the dysfunction and apoptosis of glomerular epithelial cells, the effects of PARP inhibitors on other cells types (mesangial and endothelial cells) might also be important in the disease pathogenesis.

PARP inhibitors appear to protect podocytes by multiple mechanisms. First, PJ-34 and INO-1001 prevented high-glucose–induced podocyte apoptosis in vitro. This effect could be related to normalization of the mitochondrial membrane potential, which could protect from mitochondrial collapse. Nuclear translocation of mitochondrial apoptosis–inducing factor, in response to excess PARP-1 activation triggering apoptotic cell death, was found to represent a major cell death pathway in various neuronal and cardiovascular excitotoxicity and oxidative stress models (42–45). NAD⁺ depletion and induction of mitochondrial permeability transition were implicated as intermediate steps linking PARP-1 activation to apoptosis-inducing factor translocation. Our results could be consistent with these findings.

PARP inhibitors also blocked high-glucose–induced ROS generation in cultured podocytes. It has long been thought that DNA breaks induced by nitrosative or oxidative stress are the obligatory triggers of PARP activation. Overexpression of uncoupling protein 1 or manganese superoxide dismutase prevents PARP activation in endothelial cells (14). Therefore, PARP activation was placed downstream of intracellular ROS generation. However, here we report that catalytic inhibitors of PARP block high-glucose–induced ROS release. Therefore, it is possible that in high-glucose–treated podocytes, PARP might be activated via an early, and so far unidentified, mechanism and that its activation contributes to an intracellular ROS increase or local ROS generation (not easily assessed by conventional whole-cell ROS assays) that contributes or leads to PARP activation. INO-1001 and PJ-34 are catalytic inhibitors of PARP. (They bind to the active sites.) Therefore, it is unlikely that their ROS inhibitory effect can be attributed to an antioxidant effect, as such an effect is usually observed in much higher concentrations (millimolar). In addition, similar protective results were obtained by other investigators in thymocytes with genetic deletion of PARP-1 and in high-glucose–subjected Schwann cells (36,46).

An additional mechanism that may contribute to the protective effects of PJ-34 and INO-1001 could be related to the suppression of NFκB activation. Regulation of the NFκB pathway by PARP has been reported in multiple previous studies and has been either attributed to the enzymatic activity of PARP (24) or the physical association of NFκB with PARP (22). In this study, we propose that the catalytic activity of PARP is important in this process. Persistent NFκB activation has been shown in many complication-prone tissues in diabetes, including vascular smooth muscle cells, endothelial cells, pericytes, and renal tubular epithelial cells (47,48). Here, we report that NFκB is activated in high-glucose–treated podocytes in vitro and in podocytes of diabetic mice in vivo. This was made evident by an increase of immunostaining of nuclear NFκB p50 in db/db diabetic animals and an increase of IκBα phosphorylation and p50 nuclear translocation following glucose treatment of podocytes. NFκB is a multifunctional transcriptional factor that can regulate the expression of genes involved in cell survival, apoptosis, and inflammatory response. Recent reports indicate that NFκB might play an important role in the pathogenesis of diabetic nephropathy of streptozotocin-induced diabetic rats (48), although the exact mechanism of activation and its downstream effectors are not fully established. In the case of HIV-infected podocytes, Fas and FasL was shown to be directly regulated by NFκB (49). We have also observed an increase of FasL mRNA in high-glucose–treated podocytes (K.S., unpublished observations). Since FasL plays an important role in the apoptosis pathway, it might also represent a link by which PARP inhibitors protect podocytes from high-glucose–induced apoptosis. However, other effects of NFκB should not be excluded.

In summary, we propose that PARP activation plays an important role in the development of diabetic glomerulopathy, as inhibitors of this enzyme attenuate the development of glomerular changes in Lepr^db/db mice, a model of type 2 diabetes. We propose that PARP activation is an important central mediator of hyperglycemia-induced podocyte dysfunction. PARP inhibitors block high-glucose–induced podocyte apoptosis, ROS generation, and the activation of the NFκB pathway and can subsequently protect from podocyte loss. This conclusion may be particularly significant, as PARP inhibitors are currently in clinical trials for various other indications (15). Based on our findings, we suggest that they may also be beneficial in diabetic nephropathy.

This work was supported by grants from the Juvenile Diabetes Foundation, the National Kidney Foundation, the National Institutes of Health, the Hungarian Research Fund OTKA, and the European Foundation for the Study of Diabetes.

Cell culture and histology service were provided by the O’Brian Kidney Center Imaging Core Facility.

We thank Mr. Chih-Kang Huang and Chun-Yang Xiao for their technical assistance with the animal experiments. We also thank Dr. Peter Mundel (Mount Sinai School of Medicine) for providing the murine podocyte cell line and Dr. Thomas Hostetter for a critical reading of the manuscript.