Bone Morphogenetic Protein 4 and Smad1 Mediate Extracellular Matrix Production in the Development of Diabetic Nephropathy

Diabetic nephropathy is a life-threatening complication of diabetes and the leading cause of end-stage renal disease (1). The structural features of diabetic nephropathy include thickening of the glomerular basement membrane (GBM) and mesangial matrix expansion (2,3). Mesangial matrix expansion is pathologically important because it leads to glomerulosclerosis accompanied by various tubulointerstitial damages and subsequent nephron loss (4,5). In addition, the severity of mesangial matrix expansion is clinically important because it is closely associated with the decline of the glomerular filtration rate (6).

Mesangial matrix expansion is characterized by increased amounts of extracellular matrix (7), particularly α1/α2 type IV collagen (Col4a1/a2) (8). Although various peptides or growth factors are shown to mediate the regulation of this key component, the protein responsible for its direct regulation remains to be determined.

Because various injuries of epithelial, endothelial, and mesangial cells converge on the accumulation of Col4a1/a2 in the mesangium, mesangial cells presumably play a central role for the regulation of Col4a1/a2, even if they are not the primary target of injury (9). Therefore, we attempted to elucidate the direct regulation of Col4a1/a2 under diabetic conditions and demonstrate that Smad1 can transcriptionally regulate Col4a1/a2 in the presence of advanced glycation end products in mesangial cells (10).

Smad1 is an intracellular molecule originally cloned as a signal transducer of the transforming growth factor (TGF)-β superfamily (11). In response to these stimuli, Smad1 is phosphorylated at the COOH-terminal SSXS motif followed by accumulation in the nucleus where it regulates the transcription of specific target genes (12). In vivo, Smad1 is essential for the development of the kidney (13), but it is not detected in adult murine glomeruli (14). Previously, we reported that Smad1 is induced and phosphorylated by advanced glycation end products and binds to the promoter of Col4a1/a2, thus upregulating its transcriptional activity in mesangial cells (10). We also found that Smad1 is highly expressed in human diabetic glomeruli and that glomerular expression of Smad1 is closely correlated with the severity of mesangial matrix expansion in a rodent model of diabetic nephropathy (15). However, the functional role of Smad1 in diabetic nephropathy in vivo remains unknown.

To this end, we sought to demonstrate that transgenic mice overexpressing Smad1 can accelerate mesangial matrix expansion under diabetic conditions and to identify a regulatory factor of Smad1. We also focused on bone morphogenetic proteins (BMPs) because they are potent stimulators of Smad1. Of note, we demonstrate that BMP4 is increased in diabetic glomeruli, so to prove the involvement of BMP4 in diabetic nephropathy, we treated diabetic mice with an anti-BMP4 antibody and demonstrate that the neutralization of BMP4 prevents the phosphorylation of Smad1, accumulation of Col4a1/a2, and mesangial expansion in both diabetic Smad1 transgenic (Smad1-Tg) mice and their littermates.

All animal experiments in this study were performed in accordance with institutional guidelines, and the review board of Kyoto University granted ethical permission. The pCAGGS-SMAD1 vector was constructed by inserting human SMAD1 cDNA into the mammalian expression vector pCAGGS (provided by J. Miyazaki, Osaka University). Although a previous report indicated that this promoter could be transactivated in glomerular epithelial cells (16), another report showed that the transgene under this promoter is expressed ubiquitously in glomeruli (17); therefore, it could still work in mesangial cells. Transgenic mice expressing SMAD1 were generated as described (18). Integration of the transgene into host genome was confirmed by Southern blot analysis of DNA using a ³²P-labeled SMAD1 cDNA fragment as a probe (Fig. 1A).

To generate inducible Smad1-Tg mouse lines, we used the tamoxifen-regulated Cre-loxP system (TaconicArtemis GmbH, Köln, Germany). This system consists of two transgenes. The first transgene is the inducible Smad1 expression cassette pMacII-floxed GFP pA-BMP4 using expression vector pMacII consisting of a cytomegalovirus enhancer and mouse β-actin promoter (Supplementary Fig. 2A). The second transgene is a construct for the expression of a fusion protein of mutated murine estrogen receptor (Mer) and Cre recombinase (MerCreMer [MCM]), with the pCAGGS vector (Supplementary Fig. 2A). MCM cDNA was a gift from M. Reth (Max Planck Institute of Immunobiology and Epigenetics) (19). Each transgene was microinjected into the pronuclei of C57BL/6J fertilized eggs to create a transgenic mouse line. For induction of SMAD1 gene expression, 8-week-old transgenic mice were fed a diet containing tamoxifen citrate 400 mg/kg.

Diabetes was induced in 8-week-old mice by injecting streptozotocin (STZ) 50 mg/kg i.p. (Sigma, St. Louis, MO) for 5 consecutive days. Control animals received 0.1 mmol/L sodium citrate buffer (pH 4.5) alone.

The right-side kidney was divided into fragments, each of which was fixed with Carnoy solution, neutral buffered formalin, and 2% glutaraldehyde or was frozen immediately in optimal cutting temperature compound.

The recombinant human BMP4 peptide was purchased from R&D Systems (Minneapolis, MN). For immunization with the peptide, 6-week-old mice were injected subcutaneously with 50 μg conjugated peptide once a week for 4 weeks followed by an interval of 2 weeks. Three days after final immunization, spleen cells were harvested for production of hybridomas to recombinant human BMP4. Monoclonal antibodies were generated using established procedures.

Ten milligrams/kilogram of neutralizing anti-BMP4 antibody was injected subcutaneously into each group of mice once every 2 weeks from 24 until 36 weeks after the induction of diabetes. As a negative control, mouse IgG (MP Biomedicals, Solon, OH) was injected at the same time points.

Two-micrometer sections embedded in paraffin were collected through the largest axial section and stained with periodic acid silver methenamine (PASM). To quantify mesangial expansion, all tissues were sectioned and stained by one professional pathology technician (H. Uchiyama, Taigenkai Hospital) to control the thickness of sections and the intensity of silver staining of the individual slides. Sections were further coded and read by an observer (M.A.) blinded to the experimental protocol applied (15).

Portions of the cortex were fixed in 2% glutaraldehyde and postfixed in 1% osmic acid. After embedding, ultrathin sections were stained (20). The average GBM thickness was measured using Image-Pro Plus software (Media Cybernetics, Bethesda, MD).

Kidneys were processed as previously described (15). The primary antibodies used in this experiment are listed in Supplementary Table 3. For immunostaining of Smad1, a mouse monoclonal antibody was used as described (21).

Immunofluorescence staining was performed as described (15). Fluorescein isothiocyanate–labeled or biotinylated secondary antibody followed by avidin-labeled Alexa Fluor 594 (Molecular Probes, Carlsbad, CA) was applied. For immunofluorescence staining of type IV collagen, 2-μm sections of formalin-fixed paraffin-embedded tissue blocks were used. The antitype IV collagen antibody used in this experiment reacted mainly with mesangium, which includes Col4a1/a2, but not with the GBM, which includes α3/α4/α5 type IV collagen. The immunoreactivity of type IV collagen was quantified as described (15).

Glomeruli were isolated by DynaBeads (22). For the quantification of α-smooth muscle actin (αSMA), laser-manipulated microdissection was performed because some of the glomeruli isolated by DynaBeads had the afferent and/or efferent arteriole still attached, which contained abundant αSMA (23).

Total RNA was extracted from isolated glomeruli using the guanidinium thiocyanate-phenol-chloroform method (TRIzol reagent, Invitrogen) in 20 μL RNase-free water or from microdissected glomeruli using a PicoPure RNA Isolation Kit (Arcturus, Foster City, CA) in 15 μL elution buffer.

For RNA from isolated glomeruli, real-time RT-PCR was performed (24). Specific primers are listed in Supplementary Table 2 except Col1a2, which was synthesized commercially (ABI primers and probes; Applied Biosystems). For RNA from microdissected glomeruli, the primers and probes were obtained from Applied Biosystems.

Tissues were homogenized in radioimmunoprecipitation assay buffer and subjected to immunoblotting (25). The anti–β-actin antibody (#4967), anti-Smad1 antibody reactive only to human or monkey (#9512), and anti-Smad1 antibody reactive to both human and mouse (#9743) were obtained from Cell Signaling Technology (Danvers, MA). The antiphospho-Smad1/5/8 antibody was obtained from Chemicon (Millipore, Billerica, MA). The anti-αSMA antibody was obtained from Sigma. The anti-GAPDH antibody was obtained from BD Biosciences (San Jose, CA).

The reporter plasmid that contained the Smad1 responsive element (3GC2-Lux) was provided from K. Miyazono (University of Tokyo). The αSMA promoter reporter plasmid (SMA-Luc) contains 219 base pairs of the proximal 5′-flanking region of the αSMA gene subcloned into the luciferase reporter vector (Promega, Madison, WI). Expression vectors for wild-type (WT) and mutant Smad1 have been described previously (26).

Murine mesangial cells were established as described previously (25). After 12-h incubation, cells were starved in DMEM containing 0.5% FCS followed by the stimuli. Treatment with dorsomorphin (Tocris Bioscience, Ellisville, MO), anti–BMP4-neutralizing antibody, or control IgG antibody was performed 30 min before the stimulation.

Mesangial cells or Cos7 cells (1.0 × 10⁵/mL) were seeded into 12-well plates (Nunc). After 6 h, the cells were transfected with 375 ng SMA-Luc or 3GC2-Lux and 37.5 ng pRL-CMV (Promega) along with Smad1-DVD, Smad1-ΔC expression vector, or a mock vector. Transfection was performed with FuGENE 6 (Roche Diagnostics, Indianapolis, IN). The medium was changed 12 h after transfection to 0.5% FCS in DMEM. Twenty-four hours after medium change, cells were harvested and luciferase activity measured (15).

Mesangial cells (1.0 × 10⁵/mL) were seeded in chamber slides (Nalge Nunc, Roskilde, Denmark). Twenty-four hours after transfection and medium change, cells were fixed in 4% paraformaldehyde and treated with anti-Smad1 antibody at 1:100 (T-20; Santa Cruz Biotechnology, Dallas, TX). An appropriate fluorescein isothiocyanate–conjugated secondary antibody was used.

All analyses were performed using JMP 11 software (SAS Institute, Cary, NC). Normal distribution assumptions were verified using the Shapiro-Wilk test. For type II Bmp receptor (BmprII), type I Bmp receptor (Alk3), Smad1, and albuminuria, logarithmically normal distribution assumptions were verified. Analyses were performed using MANOVA for the time course of albuminuria and one-way or two-way ANOVA for other variables followed by Tukey honest significant difference (HSD) test for multiple comparisons. Data are presented as mean ± SD. P < 0.05 was considered significant.

We constructed a transgene consisting of a fragment of the chicken β-actin promoter and human Smad1 cDNA (Fig. 1A). Transgenic founder lines carrying the human Smad1 transgene were identified by Southern blot analysis (Fig. 1B). Human SMAD1 mRNA was detected in only two founders (Tg#59 and Tg#60). Moreover, because of their poor fertilizing ability, only one line could be finally established from the male founder (Tg#59) (Fig. 1C). Western blot analyses for Smad1 revealed that Smad1 from the transgene was mainly expressed in skeletal muscle, heart, and testis and was slightly expressed in whole kidney, whereas endogenous Smad1 was expressed ubiquitously (Fig. 1D). On the other hand, Smad1 was increased ∼50% in Tg mice compared with their WT littermates in isolated glomeruli (Fig. 1E). Immunohistochemical analysis revealed that Smad1 from the transgene was detected mainly in the mesangial area (Fig. 1F). To confirm that the Smad1 transgene was expressed in mesangial cells, primary cultures were established from glomeruli isolated from normal 4-week-old Smad1-Tg#59 mice. Smad1 from the transgene was detected in primary cultured mesangial cells from Tg#59 mice (Supplementary Fig. 1A). We also confirmed no contamination of glomerular epithelial cells in this primary culture from Tg#59 mice by demonstrating that cell lysate from Tg#59 cells did not contain E-cadherin (Supplementary Fig. 1B). However, there were no significant histological changes in the kidney between Smad1-Tg#59 mice and their WT littermates. The glomerular density and glomerular surface area were also comparable between the two groups (10.8 ± 1.9 vs. 11.3 ± 2.2 μm² and 1,980 ± 290 vs. 1,910 ± 260 μm², respectively).

Because we could obtain only one line using this construct, we established another line of Smad1-Tg mice with inducible Smad1 expression using the tamoxifen-regulated Cre-loxP system (Supplementary Fig. 2A). The transgene was expressed mainly in epithelial cells (Supplementary Fig. 2B–J) and partly in the mesangial area (Supplementary Fig. 2K). After induction by tamoxifen, Smad1 was significantly increased in various tissues, including kidney (Supplementary Fig. 2L).

To examine the role of Smad1 overexpression in diabetic nephropathy, we induced diabetes by STZ. Body weights, HbA_1c, and blood pressure in diabetic Smad1-Tg mice did not differ from their diabetic littermates (Table 1).

First, we analyzed mesangial matrix expansion of each group of mice 36 weeks after STZ treatment. Morphometric analysis revealed a significant increase in the mesangial matrix expansion of diabetic Smad1-Tg#59 mice (Fig. 1H) and inducible Smad1-Tg#5 mice (Supplementary Fig. 2M) compared with their diabetic littermates. Histologically, most glomeruli exhibited a widespread increase in PASM-positive material within the mesangium, termed “diffuse lesion” (Fig. 1G). Diabetic MCM-Tg mice, which did not express the Smad1 transgene, exhibited similar mesangial expansion to their WT littermates (Supplementary Fig. 2M).

Next, we measured albuminuria in each group of mice. Both WT and Smad-Tg#59 diabetic mice exhibited more albuminuria than nondiabetic mice at 24 weeks, and these increases were sustained through 32 weeks during the experimental period (Fig. 1I). However, there was no difference in albuminuria levels between WT and Smad1-Tg#59 mice, although diabetic Smad1-Tg#59 mice tended to exhibit slightly more albuminuria than WT mice. Inducible Smad1-Tg#5 also exhibited a tendency toward slightly extended albuminuria relative to their littermates or MCM-Tg mice 32 weeks after the induction of diabetes, but the trend was not statistically significant (Supplementary Fig. 2N).

Previously, we reported that Smad1 transcriptionally regulates Col4a1/a2 and other extracellular matrix proteins, such as type I collagen, in vitro (10). Because mesangial matrix expansion was accelerated in diabetic Smad1-Tg mice compared with their diabetic littermates, we quantified the glomerular expression of Col4a1/a2 and Col1a1/a2 in these mice. As shown in Fig. 2A, the glomerular expression of these molecules was increased in diabetic mice compared with nondiabetic mice. Moreover, the expression was significantly increased in diabetic Smad1-Tg mice relative to their diabetic littermates. Immunohistochemistry revealed that these molecules were accumulated mainly in the mesangial area (Fig. 2B). αSMA is another key molecule in diabetic glomerulopathy and a marker of mesangial phenotypic changes. Therefore, we examined the glomerular expression of αSMA in each group of mice. In diabetic mice, the expression of αSMA was significantly increased after STZ treatment. Of note, Smad1-Tg mice exhibited significant expression of αSMA at the same stage (Fig. 2A). Immunohistochemistry revealed that increased αSMA was also localized mainly in the mesangial area (Fig. 2B).

We further investigated the thickness of the GBM by electron microscopy. Diabetic Smad1-Tg mice exhibited significant GBM thickening relative to their diabetic littermates at the same stage (Fig. 2C and D). Glomerular expression of Col4a3, a major molecular component of the GBM, was also increased in diabetic Smad1-Tg in parallel with GBM thickening (Fig. 2E).

These data demonstrate that the overexpression of Smad1 per se does not exacerbate nephropathy. Smad1 is activated by the phosphorylation of its carboxyl terminus. Therefore, we quantified Smad1 activation by counting cells with positive staining of phosphorylated Smad1. Although glomerular expression of Smad1 was increased in Smad1-Tg mice relative to their littermates (Fig. 2F), phosphorylated Smad1 was barely detectable in both Smad1-Tg mice and their littermates before STZ treatment. After STZ treatment, however, Smad1 was phosphorylated and translocated into the nucleus in both Smad1-Tg mice and their littermates (Fig. 2G and H). Of note, nuclear translocation of phosphorylated Smad1 was more evident in diabetic Smad1-Tg mice than in their littermates (Fig. 2H) and was largely localized in mesangial and/or endocapillary cells (Fig. 2I).

The data demonstrate that Smad1 is phosphorylated under diabetic conditions. BMPs/BMPRs are generally accepted as potent stimulators of Smad1. Among the BMPs, BMP2, -4, and -7 are expressed at various sites in different embryonic stages of renal development (27–29). We observed that Bmp7 was abundantly expressed both in nondiabetic and diabetic glomeruli, but there was no difference in its expression between the nondiabetic and diabetic groups of mice (Fig. 3A). Of note, glomerular expression of Bmp4 was increased approximately twofold after STZ treatment in both WT and Smad1-Tg mice (Fig. 3B), although Bmp2 was not detected by quantitative RT-PCR (RT-qPCR) from the same aliquot (data not shown). These data suggest that BMP4 is involved in the progression of diabetic nephropathy. BMPs induce Smad1 phosphorylation by forming heterotetrameric complexes with two major types of membrane-bound serine/threonine kinase receptors: the type I ALK receptors and the type II receptors (30). In vitro binding assays suggest that ALK3/6 are type I receptors for BMP4 (31,32). Therefore, we examined the glomerular expression of BMPRII, Alk3, and Alk6 in these mice. The glomerular expression of BmprII and Alk3 but not Alk6 was increased after STZ treatment. Next, we studied the localization of BMP4 and ALK3 by immunohistochemistry. In nondiabetic mice, BMP4 was barely detected in the glomeruli (data not shown). In contrast, 36 weeks after diabetes induction, BMP4 was extensively expressed in the podocytes and partly in the mesangium (Fig. 3B). Double immunostaining for ALK3 and desmin, platelet-derived growth factor (PDGF) receptor β (PDGFRβ), or nephrin revealed that increased ALK3 in diabetic glomeruli was mainly localized in the mesangium (Fig. 3C). These data suggest that BMP4/ALK3/Smad1 signaling contributes to the progression of diabetic nephropathy.

To further delineate the role of BMP4 in the development of diabetic nephropathy, we administered a neutralizing antibody against BMP4 to both diabetic Smad1-Tg mice and their diabetic littermates. First, we evaluated the specificity of the neutralizing activity of the antibody using an assay measuring Smad1 phosphorylation in mesangial cells induced by BMPs and TGF-β. The addition of the antibody completely inhibited the phosphorylation of Smad1/5/8 induced by BMP4 but not by BMP2, BMP7, or TGF-β (Fig. 4A), indicating its specificity. Next, we administered the neutralizing antibody or control IgG to each group of mice (Table 2). The administration of the neutralizing antibody attenuated the nuclear translocation of phosphorylated Smad1 (Fig. 4B), ameliorated the glomerular accumulation of type IV collagen (Fig. 4D), and inhibited mesangial matrix expansion (Fig. 4C and F) in both diabetic Smad1-Tg mice and littermates. Furthermore, RT-qPCR of RNA from glomeruli isolated by laser-manipulated microdissection revealed that treatment with the BMP4-neutralizing antibody also improved glomerular expression of αSMA in both diabetic Smad1-Tg mice and littermates (Fig. 4E). However, the BMP4-neutralizing antibody did not have any effect on albuminuria (Fig. 4G).

In this study, we demonstrate that overexpression and subsequent phosphorylation of Smad1 results in an increase in the glomerular expression of αSMA, whereas the inhibition of Smad1 phosphorylation using an anti-BMP4 antibody leads to the improvement of glomerular expression of αSMA. To further elucidate the relationship between Smad1 phosphorylation and αSMA, we generated Smad1 mutants: a constitutively active mutant in which two serine residues at the carboxyl termini were substituted with aspartic acid (Smad1-DVD) (26) and a dominant negative mutant in which the carboxyl termini were lacking (Smad1-ΔC) (Fig. 5A). The expression of Smad1-DVD increased the transcriptional activity of 3GC2-Lux, a Smad1-dependent reporter, whereas the expression of Smad1-ΔC did not (Fig. 5B). In mesangial cells, Smad1-DVD was localized in the nuclei, whereas Smad1-ΔC was localized mainly in the cytoplasm (Fig. 5C). Therefore, we asked whether the transcriptional activity of αSMA is modulated by the constitutive activation of Smad1 in mesangial cells. As expected, the expression of Smad1-DVD but not Smad1-ΔC increased the transcriptional activity of αSMA in mesangial cells (Fig. 5D). Finally, we examined whether dorsomorphin, a small molecule inhibitor of BMP signaling (33), can affect the expression of αSMA by inhibiting the phosphorylation of Smad1 in mesangial cells. As shown in Fig. 5E and F, BMP4 induced the phosphorylation of Smad1 along with increased αSMA expression, which was blocked by dorsomorphin. These data indicate that BMP4 mediates αSMA expression through the phosphorylation of Smad1 in mesangial cells.

We further investigated whether the reduction of Smad1 expression improves diabetic glomerular changes by using heterozygous SMAD1 knockout (KO) mice, which were provided from Anita B. Roberts (National Cancer Institute, Bethesda, MD) (34). After the induction of diabetes, body weights, blood pressure, and HbA_1c in diabetic SMAD1 heterozygous KO mice did not differ from diabetic WT mice (Supplementary Table 4). The glomerular expression of Smad1 was reduced by ∼30% in both diabetic and nondiabetic SMAD1 heterozygous KO mice (Fig. 6A). We observed a partial attenuation of albuminuria (Fig. 6B) and mesangial matrix expansion (Fig. 6C and D) along with improvement of the glomerular expression of Col4a1/a2 and Col1a1/a2 (Fig. 6E) in diabetic SMAD1 heterozygous KO mice compared with diabetic WT mice. These data suggest that Smad1 is a critical determinant for the development of diabetic nephropathy.

In this study, we demonstrate that Smad1 plays a critical role in the development of diabetic nephropathy. Smad1 is the only molecule shown in vivo to be a direct regulator of Col4a1/a2; therefore, it seems reasonable to consider Smad1 or its modifier as therapeutic targets for diabetic nephropathy. In this regard, BMP4 could be a potential candidate. We reported that heterozygous BMP4 KO mice exhibit attenuated diabetic nephropathy (18). However, because these mice display renal abnormalities (35), this attenuation might be partly due to the effect of the congenital kidney anomalies. Therefore, in the present study, we used a more physiological method for modulating BMP4 action using the BMP4-blocking antibody and demonstrated that the inhibition of BMP4 ameliorates diabetic glomerular injuries.

Although BMP7 is known to protect against diabetic nephropathy, the data suggest that BMP4 could aggravate diabetic nephropathy. BMP4 could have several distinct roles from BMP7 in its structure and function. First, BMP4 has only 58% homology in peptide sequence to BMP7. Second, the functional targets differ between BMP4 and BMP7 during kidney development. For example, BMP4 promotes glomerulogenesis, but BMP7 does not. On the other hand, BMP7 but not BMP4 contributes to mesenchymal survival (36). Finally, the downstream target of BMP4 is Smad1 in mesangial cells, as demonstrated in the present study, whereas the downstream target of BMP7 in mesangial cells is Smad5 (37). Thus, we suggest that each BMP has its own function and signaling targets in kidney development and disease progression.

It remains to be determined whether BMP4/Smad1 signaling is specific for diabetic nephropathy. We previously reported increased glomerular expression of BMP4/Smad1 in another diabetic mouse model (38) and identified PDGF-β as another activator of Smad1 in mesangial cells through Src in murine experimental glomerulonephritis (39). Smad1 also operates during chronic stages of fibrosis in scleroderma (40). Thus, we speculate that the Smad1 signaling is a common pathway among various glomerular injuries and in organ fibrosis. To determine the specificity of BMP4 in diabetic nephropathy, the glomerular expression of BMP4 should be tested in other disease models in future experiments.

In this study, albuminuria did not reflect mesangial expansion in BMP4 antibody–treated diabetic mice and diabetic Smad1-Tg mice, although diabetic Smad1 heterozygous KO mice exhibited slightly improved albuminuria. The mechanisms responsible for these results remain unknown. Previous reports have indicated that albuminuria is not correlated with the severity of mesangial expansion in incipient diabetic nephropathy both in humans (41) and in a rodent model (15). Treatment with an anti–TGF-β antibody does not attenuate albuminuria in db/db mice despite its beneficial effects on glomerular matrix expansion (42). These data suggest that distinct mechanisms may underlie albuminuria and mesangial matrix expansion in diabetic nephropathy.

In conclusion, this study of Smad1-overexpressing mice reveals that both the induction and the phosphorylation of Smad1 play critical roles in the development of diabetic glomerulopathy in vivo. BMP4 might be responsible for the phosphorylation of Smad1 and represents a novel therapeutic target for diabetic nephropathy.

Acknowledgments. The authors thank Hideo Uchiyama and Kazumasa Usami (Taigenkai Hospital) for preparing tissue sections and PASM staining and Norihiko Suzuki (Nagoya University) for electron microscope sample preparation. They also thank Ayumi Hosotani and Maki Watanabe (Kyoto University) and S. Hayashi and A. Sakurai (Tokushima University) for excellent technical assistance. The authors are grateful to H. Kanamori (Fukuchiyama City Hospital); M. Matsuura, T. Murakami, and T. Araoka (Tokushima University); K. Torikoshi (Kitano Hospital); A. Fukatsu (Yachiyo Hospital); M. Yanagita (Kyoto University); and T. Kimura (Kyoto University) for helpful discussion. Finally, they thank Chugai Research Institute for Medical Science, Inc., colleagues Yousuke Kawase, Takanori Tachibe, Toshio Hani, and Hiromi Tateishi for manipulation of mouse embryos; Toshio Mori for technical support in pathological examination; and Satomi Uchida and Yumiko Nakajima for excellent technical assistance.

Funding. This study was supported by Grants-in-Aid for Young Scientists (B) (21790808), Grants-in-Aid for Scientific Research (21591033), Grants-in-Aid for Scientific Research (C) (25461221), the Kidney Foundation of Japan (JKFB09-41), and Takeda Science Foundation.

Duality of Interest. Funding was provided in part by an AstraZeneca Virtual Research Institute research grant. T.D. received collaborative research funds from Chugai Pharmaceutical Co., Ltd. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. T.M. contributed to the data analysis and writing of the manuscript. M.A. contributed to the experiments and data analysis. H.Ab. contributed to the experiments and writing of the manuscript. O.U. contributed to establishing the transgenic mice and to the experiments. K.-i.J. contributed to establishing the transgenic mice. A.M. contributed to the data analysis. C.G. contributed to the experiments. T.T., S.K., K.N., N.I., and T.K. contributed to the data analysis. M.K. established the neutralizing antibody and contributed to the data analysis. N.F. contributed to the study concept and design, experiments, and data analysis. H.Ar. and T.D. contributed the study concept and design and writing of the manuscript. T.M. and H.Ab. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 71st Scientific Sessions of the American Diabetes Association, San Diego, CA, 24–28 June 2011.