Salutary Effect of Pigment Epithelium–Derived Factor in Diabetic Nephropathy: Evidence for Antifibrogenic Activities

Brown Norway (BN) rats were purchased from Charles River Laboratories (Wilmington, MA). Care, use, and treatment of all animals in this study were in strict agreement with the guidelines in the care and use of laboratory animals set forth by the University of Oklahoma.

Experimental diabetes was induced by an intraperitoneal injection of STZ (50 mg/kg in 10 mmol/l of citrate buffer; pH 4.5) into anesthetized BN rats (8 weeks of age) after an overnight fast. Age-matched control rats received an injection of citrate buffer alone. Blood glucose levels were measured 24 h after the STZ injection and monitored weekly thereafter. Only the animals with glucose levels >350 mg/dl were considered diabetic.

A construct custom built for the expression of a full-length human PEDF and a green fluorescent protein (GFP) under a cytomegalovirus promoter in an adenoviral vector were obtained from a commercial source (InVivoGen, San Diego, CA). The construct sequence provided by the manufacturer was confirmed by DNA sequencing (OMRF, Oklahoma City, OK). The recombinant virus was amplified, purified, and titered in the 293A cell line following the procedure recommended by the manufacturer.

One week after the STZ injection, diabetic rats were randomly assigned into three groups. Group 1 received no virus injection (n = 5); groups 2 and 3 received an intravenous injection of adenovirus expressing human PEDF (Ad-PEDF) (n = 7) and a control adenovirus expressing GFP under the CMV promoter (adenovirus expressing GFP [Ad-GFP]; Qbiogene, Montreal, Canada) (n = 5), respectively, at a dose of 4 × 10¹⁰ viral particles per rat.

The 24-h urine was collected from each rat in individual metabolic cage and centrifuged at 2,000g for 5 min. To avoid the variations caused by increased urine volumes in diabetic rats, the total albumin in the 24 h was normalized by creatinine concentrations. Urine creatinine levels were measured using the QuantiChrom Creatinine Assay Kit (BioAssay Systems, Hayward, CA), following the manufacturer’s protocol. Urine albumin was measured by enzyme-linked immunosorbent assay (ELISA) (Bethyl Laboratories, Montgomery, TX). Urine albumin excretion (UAE) was normalized by creatinine excretion and expressed as milligrams albumin per milligrams creatinine in 24-h urine.

The protein levels of PEDF, TGF-β1, and fibronectin in kidney tissue homogenate, urine, or cultured cells were quantified using the commercial Quantikine PEDF ELISA kit (Chemicon, Temecula, CA), the TGF-β1 ELISA kit (R&D Systems, Minneapolis, MN), and the fibronectin ELISA kit (Assaypro, Winfield, MO), respectively, following the protocols from manufacturers. The ELISA results were normalized by total protein concentrations measured using the BioRad Protein Assay reagent.

Western blot analysis was performed as described previously (16). Briefly, 50 μg of protein from each sample was blotted by an anti-CTGF antibody (R&D Systems). The same membranes were stripped and reblotted with an anti–β-actin antibody (Santa Cruz Biotechnology, Santa Cruz, CA).

Total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. The quality of the RNA samples was evaluated by electrophoresis through agarose gels with ethidium bromide staining. Primers specific for matrix metalloproteinase-2 (MMP-2) (MMP-2 forward, 5′-CTGAGCTCCCGGAAAAGATTG-3′; MMP-2 reverse, 5′-CCTGCGAAGAACACAGCCTT-3′) were used for real-time RT-PCR. PCR was performed using the GeneAmp RNA PCR kit and SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). The efficiency of real-time PCR was 99.1%. The average threshold cycle (C_T) of fluorescence units was used to analyze the mRNA levels. The mRNA levels of target genes were normalized by 18S ribosomal RNA levels. Quantification was calculated as mRNA levels (percent of control) = 2^Δ(ΔC_T) with ΔC_T = C_{T target gene} − C_{T 18S RNA} and Δ(ΔC_T) = ΔC_{T normal sample} − ΔC_{T STZ-diabetic sample}.

Primary HMC were cultured on 4-chamber slides (Nalge Nunc, Naperville, IL) to reach 80% confluence. After exposure to 1.25 ng/ml TGF-β1 with or without 160 nmol/l PEDF for 1 h, the cells were immediately fixed with 4% paraformaldehyde. The cells were incubated with an anti-Smad2/3 antibody (1:200; Upstate, Lake Placid, NY) for 2 h and then incubated with a Cy3-conjugated donkey anti-rabbit IgG antibody for 1 h. The slide was examined and analyzed under a fluorescent microscope (Olympus, Humburg, Germany).

Data were calculated and expressed as group means ± SD. Statistical analyses were performed using Student’s t test, ANOVA, and Bonferroni’s multiple comparison test. Statistical difference was considered significant at a P value of <0.05.

To confirm the expression of PEDF mediated by Ad-PEDF, 293A cells were infected with Ad-PEDF. Three days following the viral infection, the cultured cells showed high levels of GFP, a marker of the Ad-PEDF vector. PEDF protein levels were measured in the culture media and cell lysates of the cells infected by Ad-PEDF using a commercial ELISA. The results showed significantly higher PEDF levels in the conditioned media from the cells infected by Ad-PEDF (P < 0.0001, n = 3). The lysates from the cells infected with Ad-PEDF also showed significantly increased PEDF levels over the control (P < 0.001, n = 3) (Fig. 1).

Either Ad-PEDF or the control virus Ad-GFP was injected into the tail vein of STZ-induced diabetic rats 1 week following the onset of diabetes. The 24-h urine was collected from each rat individually at 0, 1, 2, 3, 4, and 5 weeks following the injection of Ad-PEDF. Total amounts of PEDF protein in 24-h urine was measured by ELISA and normalized by urine creatinine. The PEDF levels in the urine were significantly increased in diabetic rats injected with Ad-PEDF versus those injected with Ad-GFP, at 2 and 3 weeks after the virus injection (Bonferroni’s post hoc test, P < 0.001, t = 4.700 and 4.273, respectively; Fig. 2A). PEDF levels in the urine declined to control levels after 4 weeks (P > 0.05; Fig. 2A), consistent with the course of transient gene expression from adenoviral vectors.

Three weeks following the virus injection, rats were killed, and PEDF levels in the kidney were measured using ELISA. Renal PEDF levels in untreated diabetic rats were significantly reduced versus age-matched nondiabetic normal rats (ANOVA, P = 0.01, F = 6.774; Fig. 2B), corroborating our previous report (16). Ad-PEDF injection restored PEDF levels in the kidney of diabetic rats to a level approaching that of normal controls (Bonferroni’s post hoc test, P > 0.05, t = 1.646 for diabetic rats treated with Ad-PEDF versus normal controls; Fig. 2B). In contrast, injection of the same titer of the control virus, Ad-GFP, did not normalize the decrease of PEDF levels in diabetic rats (Bonferroni’s post hoc test, P < 0.05, t = 3.556 for diabetic rats treated with Ad-GFP versus normal controls; Fig. 2B), suggesting that the restoration of PEDF levels by Ad-PEDF is not a nonspecific effect from the adenovirus infection.

Compared with age-matched normal rats, the STZ-induced diabetic rats had significantly elevated blood glucose concentrations (442.00 ± 57.67 mg/dl) 48 h after STZ injection and thereafter. At 1 week after diabetes onset, no significant difference in body weight was observed between the diabetic rats (128.55 ± 5.37 g) and age-matched control rats (130 ± 9.21 g). At 10 weeks after diabetic onset, the body weights of diabetic rats (138.7 ± 27.4 g) were significantly lower than that of age-matched control rats (192.8 ± 3.7 g, P < 0.01). Ad-PEDF injection did not alter the hyperglycemia or the body weight when compared with diabetic rats treated with Ad-GFP (ANOVA, P > 0.05, F = 0.9492 and 1.095, respectively; Table 1). As expected, the diabetic rats had polyuria versus the normal controls. At 1 week after diabetes onset, the urine volume in diabetic rats (46.7 ± 12.9 ml) was significantly higher than that in age-matched normal controls (5.2 ± 0.8 ml, P < 0.001). There was, however, no difference in 24-h urine volume between diabetic rats treated with Ad-PEDF and those treated with Ad-GFP (P > 0.1, n = 4–7; Table 2).

Microalbuminuria is an early clinical manifestation of diabetic nephropathy (17). In the present study, albumin in 24-h urine was measured using ELISA specific for rat albumin and normalized by the amount of urine creatinine. The results demonstrated that diabetic rats developed a significant microalbuminuria with onset as early as 1 week after the STZ injection (Bonferroni’s post hoc test, P < 0.05, t = 2.535; Fig. 3). The 24-h UAE continued to increase with time through the 5 weeks analyzed in rats treated with the control virus (ANOVA, P < 0.0001, F = 20.09; Fig. 3). The diabetic rats injected with Ad-PEDF showed significantly reduced UAE at 2, 3, and 4 weeks after the injection of the Ad-PEDF versus diabetic rats injected with Ad-GFP (ANOVA, P < 0.0001, F = 15.3; Bonferroni’s post hoc test, P < 0.01, t = 4.525, 4.336, and 4.740; n = 4–7; Fig. 3).

It is well known that the accumulation of ECM proteins in the kidney and subsequent mesangial expansion are typically and consistently present in diabetic nephropathy, parallel with microalbuminuria (4,5). Four weeks after the STZ injection, fibronectin levels in the kidney were significantly increased in the diabetic rats treated with the control virus Ad-GFP when compared with nondiabetic rats (ANOVA, P < 0.0001, F = 54.99; Fig. 4A). At the same time, the Ad-PEDF injection decreased kidney fibronectin levels in the diabetic kidney to the normal range (Bonferroni’s post hoc test, P > 0.05, t = 1.192 for diabetic rats treated with Ad-PEDF versus normal controls; Fig. 4A).

It has been shown that both a decreased ECM protein degradation and an increased ECM protein expression contribute to the accumulation of mesangial ECM in the diabetic kidney (18). As MMP-2 is a major enzyme responsible for the degradation of ECM (19), we determined whether the PEDF gene delivery affects the expression of MMP-2 in the kidney. The results showed that in the diabetic kidney, MMP-2 expression was decreased to 60% of that in the kidney of normal controls (P < 0.05, n = 4; Fig. 4B). The PEDF gene delivery significantly increased MMP-2 mRNA levels in the diabetic kidney when compared with that in the Ad-GFP control group (P < 0.05, n = 4; Fig. 4B).

TGF-β1 is a major growth factor mediating the remodeling of ECM and mesangial expansion in diabetic nephropathy. In this study, we investigated the effect of PEDF on urine TGF-β1 excretion and TGF-β1 protein levels in diabetic kidneys. The results showed that the TGF-β1 excretion in the urine was significantly attenuated at 2 and 3 weeks after the Ad-PEDF injection (Fig. 5A).

TGF-β1 levels in the kidney, as measured by ELISA, were significantly elevated in the diabetic rats treated with the control virus, Ad-GFP, compared with that in the nondiabetic normal rats. Such overexpression of TGF-β1 was effectively prevented by the PEDF gene delivery (P < 0.05, n = 4; Fig. 5B).

CTGF is an important mediator of fibrosis by promoting the glomerular matrix accumulation in diabetic nephropathy (20). In the present study, the expression of CTGF was measured at the protein level. As shown by Western blot analysis, CTGF expression was significantly increased in diabetic kidney at the protein level (P < 0.001; Fig. 6). Ad-PEDF treatment significantly downregulated CTGF expression when compared with the diabetic rats treated with the control virus (P < 0.001; Fig. 6). These findings suggest that the reduction of the ECM protein accumulation by the PEDF gene delivery is mediated at least in part by inhibition of CTGF expression in the diabetic kidney.

To determine the direct effect of PEDF on fibrosis, we used primary HMCs treated with angiotensin II, which is known to play a pathogenic role in diabetic nephropathy and induce TGF-β1 expression. As measured by ELISA, PEDF, at concentrations of 40 and 160 nmol/l, significantly abrogated the stimulation of TGF-β1 expression by angiotensin II in cultured HMCs (Fig. 7). Similarly, PEDF also significantly prevented the induction of fibronectin secretion by angiotensin II in HMCs (Fig. 7), suggesting a direct effect of PEDF on fibrosis in kidney cells.

To gain insights into the antifibrogenic activity of PEDF, we have measured the direct effect of PEDF on fibronectin overexpression stimulated by TGF-β1 in cultured HMCs. As shown in Fig. 8, TGF-β1 treatment induced significant fibronectin overexpression in HMCs. At concentrations of 40–640 nmol/l, PEDF prevented the stimulation of fibronectin expression by TGF-β1 in a concentration-dependent manner.

To elucidate the mechanism responsible for the antifibrogenic activity of PEDF, we measured the activation of Smad3 pathway, which is known to mediate the effect of TGF-β1 on fibrosis (21,22). As Smad3 nuclear translocation is a crucial step in the activation of the pathway, we measured the PEDF effect on Smad3 nuclear translocation induced by TGF-β1. As shown by immunocytochemistry using an antibody specific for Smad2/3, Smad2/3 signal was diffusely distributed in the cytoplasm of the control HMCs. Upon treatment with TGF-β1, Smad2/3 levels were apparently increased and translocated into the nucleus. PEDF effectively blocked the translocation of Smad2/3 into the nucleus in HMCs treated with TGF-β1, suggesting that the antifibrogenic action of PEDF may be through interference with the Smad signaling pathway of TGF-β1.

PEDF is a major angiogenic inhibitor in the eye (13). The pathogenic role of the decreased retinal PEDF levels in vascular leakage and retinal neovascularization in models of diabetic retinopathy has been established (23,24). We recently reported that PEDF is highly expressed in the normal kidney but reduced in kidneys of diabetic rats, suggesting a potential implication of decreased PEDF levels in diabetic nephropathy (16). However, the protective effect of PEDF against abnormalities of diabetic nephropathy has not been demonstrated in vivo. In the present study, we provide the first in vivo evidence that a PEDF gene delivery ameliorates proteinuria in rats with STZ-induced diabetes. Toward potential mechanism for the PEDF effect, we found that PEDF inhibits the expression of pathogenic factors TGF-β1 and CTGF and suppresses ECM protein production in diabetic kidney, suggesting an antifibrogenic activity. Therefore, these observations revealed a new antifibrogenic activity of PEDF, which may be responsible for its salutary effect in diabetic nephropathy.

Microalbuminuria is a characteristic feature of early stages of diabetic nephropathy. It has been shown to be a good predictive feature for the progression to overt diabetic nephropathy (3,25,26). Recent studies indicate that microalbuminuria is an independent risk factor for cardiovascular complications and mortality in type 2 diabetic patients (27,28). Therefore, the prevention and treatment of microalbuminuria are important therapeutic targets in halting the progression of microvascular and macrovascular diabetes complications. The present study demonstrated that PEDF gene delivery significantly decreased microalbuminuria in diabetic rats. Although this PEDF effect lasted only for 3 weeks due to the inherently transient expression mediated by the adenoviral vector, the result suggests that the PEDF treatment may be promising in preventing progressive renal injury and consequently reducing the risk for overt nephropathy.

Mesangial matrix expansion is the most prominent pathological feature of diabetic nephropathy, which is characterized by the accumulation of ECM proteins such as collagen IV and fibronectin in the glomeruli (29–31). In diabetes, both the accelerated production and the decreased degradation of ECM proteins are known to contribute to the accumulation of mesangial matrix (29,32). Accumulating evidence suggests that MMP-2 is an important protease degrading ECM (18,33). In the kidney, MMP-2 is highly expressed in glomerular mesangial cells and moderately in epithelial cells (19). In high-glucose medium conditions, the expression and activity of MMP-2 were decreased in glomerular mesangial cells, which could lead to reduced ECM degradation and subsequent mesangium accumulation (18,33). Our study demonstrated that the upregulation of tissue fibronectin in the diabetic kidney is effectively reversed after the PEDF treatment. These results are consistent with but extend our previous findings showing that PEDF inhibits high glucose–induced fibronectin secretion in cultured HMCs (16), suggesting that PEDF is a potent inhibitor of ECM production by mesangial cells. Additionally, we observed that MMP-2 expression was significantly downregulated in the diabetic kidney but restored to nondiabetic control levels by the PEDF gene delivery. These findings suggest that PEDF suppresses glomerular mesangial matrix accumulation possibly by two mechanisms, inhibiting ECM production and enhancing ECM degradation.

TGF-β refers to three isoforms of secreted homodimeric proteins encoded by different genes (34). In the process of tissue development and wound healing, TGF-β plays a crucial role in controlling the ECM deposition and remodeling (22). On one hand, TGF-β stimulates the synthesis of major components of ECM proteins, such as collagen and fibronectin (29,35). On the other hand, TGF-β inhibits ECM degradation by decreasing protease secretion and activating protease inhibitors (36) and tissue inhibitor of metalloproteinase-1 (37). In diabetic kidneys, the overexpression of TGF-β is believed to be the major mediator responsible for the early pathological changes of diabetic nephropathy, including the glomerular basement membrane thickening and mesangial matrix expansion (9,20,34,38–40). Blockade of TGF-β expression or function has been shown to ameliorate pathological damage or to prevent functional renal insufficiency in diabetic nephropathy (9,10,29,41). The present study demonstrated that the PEDF gene delivery effectively decreased the urine TGF-β1 excretion and TGF-β1 expression in the kidney, providing the first in vivo evidence suggesting that PEDF functions as an endogenous inhibitor of TGF-β expression. These data suggest that the reduction of ECM accumulation induced by PEDF may be mediated, at least in part, by the downregulation of TGF-β1.

The Smad pathway is known to mediate the functions of TGF-β on renal fibrogenesis and subsequent ECM accumulation in diabetic nephropathy (13). The present study demonstrated that TGF-β induces Smad2/3 nuclear translocation, a key step in the activation of the Smad pathway. Moreover, PEDF blocks the Smad2/3 nuclear translocation induced by TGF-β in HMCs. This result suggests that PEDF not only downregulates TGF-β expression but also blocks its function, likely through the Smad pathway.

A number of studies have indicated that CTGF, a 349–amino acid cysteine-rich peptide, is an important downstream mediator of the fibrogenic effect of TGF-β in regulating matrix metabolism (42–44). All of the causal factors of diabetic glomerulosclerosis, including TGF-β and high glucose concentration, significantly induced CTGF expression in glomerular mesangial cells (45). In the early stage of diabetic nephropathy in db/db mice, the CTGF expression in the glomeruli is significantly increased (45). In a recent study in type 1 diabetic patients with nephropathy, the urine CTGF excretion was found to correlate with the severity of diabetic nephropathy (46). These findings support the notion that CTGF is an important mediator of diabetic kidney disease. In the present study, we demonstrated the upregulation of CTGF expression in the kidneys of STZ-induced diabetic rats, consistent with the previous observations from db/db mice and diabetic patients (45,46). Furthermore, we noted that the PEDF treatment essentially abrogated the overexpression of CTGF in diabetic kidneys, concomitant with the normalization of TGF-β1 and fibronectin levels to the nondiabetic range. These findings suggest that PEDF may suppress mesangial matrix accumulation through the blockade of hyperglycemia-induced TGF-β1 and CTGF expression.

In summary, our data provide the first evidence suggesting that PEDF blocks fibrogenesis in diabetic kidneys, via inhibition of TGF-β and CTGF expression and function. The antifibrogenic activity of PEDF may be responsible, at least in part, for its salutary effects in diabetic kidney complication in a type 1 diabetic rat model. As previous studies have shown that anti–TGF-β therapies can ameliorate the functional and structural abnormalities of diabetic nephropathy, we propose that PEDF, by virtue of its anti–TGF-β and antifibrogenic properties, may have a therapeutic potential in diabetic nephropathy.

This study was supported by National Institutes of Health Grants EY015650, EY012231, EY014427, and EY12190 and by National Institutes of Health Grant P20 RR17703 from the Centers of Biomedical Research Excellence program of the National Center for Research Sources; an unrestricted grant from Research to Prevent Blindness to the Department of Ophthalmology; a grant from Oklahoma Center for the Advancement of Science and Technology; Juvenile Diabetes Research Foundation Grant 1-2003-627; and American Diabetes Association Grant 1-05-RA-75.