PDGF receptor-β uses Akt/mTORC1 signaling node to promote high glucose-induced renal proximal tubular cell collagen I (α2) expression

Abstract

Increased expression of PDGF receptor-β (PDGFRβ) has been shown in renal proximal tubules in mice with diabetes. The core molecular network used by high glucose to induce proximal tubular epithelial cell collagen I (α2) expression is poorly understood. We hypothesized that activation of PDGFRβ by high glucose increases collagen I (α2) production via the Akt/mTORC1 signaling pathway in proximal tubular epithelial cells. Using biochemical and molecular biological techniques, we investigated this hypothesis. We show that high glucose increases activating phosphorylation of the PDGFRβ, resulting in phosphorylation of phosphatidylinositol 3-kinase. A specific inhibitor, JNJ-10198409, and small interfering RNAs targeting PDGFRβ blocked this phosphorylation without having any effect on MEK/Erk1/2 activation. We also found that PDGFRβ regulates high glucose-induced Akt activation, its targets tuberin and PRAS40 phosphorylation, and finally, mTORC1 activation. Furthermore, inhibition of PDGFRβ suppressed high glucose-induced expression of collagen I (α2) in proximal tubular cells. Importantly, expression of constitutively active Akt or mTORC1 reversed these processes. As a mechanism, we found that JNJ and PDGFRβ knockdown inhibited high glucose-stimulated Hif1α expression. Furthermore, overexpression of Hif1α restored expression of collagen I (α2) that was inhibited by PDGFRβ knockdown in high glucose-stimulated cells. Finally, we show increased phosphorylation of PDGFRβ and its association with Akt/mTORC1 activation, Hif1α expression, and elevated collagen I (α2) levels in the renal cortex of mice with diabetes. Our results identify PDGFRβ as a driver in activating Akt/mTORC1 nexus for high glucose-mediated expression of collagen I (α2) in proximal tubular epithelial cells, which contributes to tubulointerstitial fibrosis in diabetic nephropathy.

one-third of patients with diabetes develop nephropathy (3). Manifestation of diabetic nephropathy involves hyperfiltration, followed by microalbuminuria that leads to macroalbuminuria. Although glomerular mesangial cells and podocytes play significant roles, the tubular system also contributes to the pathology of diabetic nephropathy initially via a tubuloglomerular feedback, which increases early glomerular filtration rate (41, 77). Early changes preceding the loss of renal function also include whole kidney hypertrophy, including glomerular hypertrophy; however, the primary site of initial hypertrophy is the proximal tubule, which is followed by thickening of the tubular basement membrane and interstitial fibrosis. In fact, proximal tubular epithelial cells undergo epithelial to mesenchymal differentiation, which increases extracellular matrix deposition causing fibrosis of the kidney (41, 58). The molecular mechanisms for proximal tubular phenotypic changes involve autocrine and paracrine action of multiple growth factors and cytokines including transforming growth factor-β (TGF-β) (41). Apart from TGF-β, platelet-derived growth factor (PDGF) has been shown to be present in the glomerular compartment in rodent models of diabetes and in patients with diabetic nephropathy (45, 57).

PDGFs are composed of four polypeptide chains that form five different isoforms by homo- or heterodimerization. The PDGF-A and -B chains are secreted as homo or heterodimers. In contrast, PDGF-C and -D chains are secreted as latent forms and are processed to produce mature homodimers (25). The PDGF dimers bind to two PDGF receptors (PDGFRα and PDGFRβ) with distinct binding affinity. For example, PDGF-AA and -CC bind to PDGFRα with high affinity, whereas PDGF-BB and -AB bind to both receptors. PDGF-DD serves as a high-affinity ligand for homodimeric PDGFRββ; however, both PDGF-DD and -CC can interact with PDGFRαβ heterodimer with low affinity (4).

The transmembrane PDGFRs contain IgG-like domains in their extracellular segment and two tyrosine kinase domains in the cytoplasm. The tyrosine kinase domains are separated by an interkinase stretch, which contains specific tyrosine residues for phosphorylation (4, 32). Binding of ligands to the extracellular domain of the receptor induces dimerization and a conformational change in the juxtamembrane domain of the receptor to relieve an autoinhibitory constraint on the intrinsic tyrosine kinase activity (7). Ligand-bound activated dimeric receptors then undergo transautophosphorylation at multiple tyrosine residues including the juxtamembrane, interkinase, and COOH-terminal domains (30). These phosphotyrosines act as docking sites for many SH2 domain-containing signaling proteins such as c-Src and phosphatidylinositol 3-kinase (PI3-kinase), which mediate many biological actions of the receptor.

PDGFs are predominantly expressed in renal mesenchymal cells, especially in mesangial cells (1). These cells constitute one-third of the population of glomeruli (1). In mice, deletion of PDGF-A chain did not show any phenotype (5). Similarly, mice deficient in PDGF-C and -D showed no obvious renal phenotype (4). On the other hand, mice in which the PDGF-B chain or PDGFRβ have been deleted showed defects in glomerular mesangial cell development, establishing a critical physiological role for this ligand-receptor system (48, 49). Importantly, mesangial cells respond to various forms of kidney injury encountered in lupus nephritis, IgA nephropathy, and mesangioproliferative glomerulonephritis (39). Mesangial cells predominantly express PDGFRβ with limited expression of PDGFRα (8, 73). Many growth factors increase proliferation of mesangial cells; however, the most potent mitogen for these cells is PDGF-BB, which acts through PDGFRβ (8, 9). In fact, neutralization of PDGFRβ prevents mesangioproliferative glomerulonephritis in rats, indicating its importance in proliferation of mesangial cells under pathological conditions (1, 39).

In patients with diabetic nephropathy, expression of PDGF-A and PDGFR-α in the glomeruli is similar to that in normal human kidney. In contrast, PDGF-B and its receptor-β expression were increased in diabetic glomeruli (70). Although glomerulosclerosis represents a main feature of diabetic nephropathy, upon injury, tubular epithelial cells produce abundant amounts of PDGF-AA and -BB (45). Proximal tubular epithelial cells also contribute to the progression of diabetic kidney disease (72). In the present study, we investigated the signal transduction pathways elicited by PDGFRβ in response to high glucose in human proximal tubular epithelial cells. We show that high glucose-induced activation of PI3-kinase and Akt requires PDGFRβ activity. We demonstrate the requirement of PDGFRβ for high glucose-stimulated mammalian target of rapamycin complex 1 (mTORC1) activity to increase Hif1α, which is necessary for proximal tubular cell collagen I (α2) expression. Finally, in the renal cortex of diabetic mice, we show increased phosphorylation of PDGFRβ, which is associated with enhanced Akt and mTORC1 activity.

MATERIALS AND METHODS

Reagents.

Tissue culture reagents, D-glucose, D-mannitol, Nonidet P-40, phenylmethylsulfonyl fluoride, Na₃VO₄, protease inhibitor cocktail, TRI reagent for RNA isolation, JNJ-10198409 (JNJ), and actin antibody were purchased from Sigma (St. Louis, MO). OPTIMEM medium for transfection was obtained from Life Technologies (Grand Island, NY). Antibodies for phospho-PDGFRβ (Tyr⁸⁵⁷), phospho-PDGFRβ (Tyr⁷⁴⁰), phospho-PDGFRβ (Tyr⁷⁵¹), PDGFRβ, phospho-Erk1/2 (Thr²⁰²/Tyr²⁰⁴), Erk1/2, phospho-MEK (Ser^217/221), MEK, phospho-p85 (Tyr⁴⁵⁸), p85, phospho-Akt (Thr³⁰⁸), phospho-Akt (Ser⁴⁷³), Akt, phospho-GSK3β (Ser⁹), GSK3β, phospho-tuberin (Thr¹⁴⁶²), phospho-PRAS40 (Thr²⁴⁶), PRAS40, phospho-4EBP-1 (Thr^37/46), phospho-4EBP-1 (Ser⁶⁵), 4EBP-1, phospho-S6 kinase (Thr³⁸⁹), S6 kinase, phospho-ribosomal protein s6 (Ser^240/244), ribosomal protein s6, phospho-eIF4E (Ser²⁰⁹), and eIF4E were obtained from Cell Signaling Technology (Danvers, MA). Antibodies against tuberin, Hif1α, collagen I (α2) and pooled silent interfering RNAs (siRNAs) for PDGFRβ and scrambled RNA were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Transfection reagent FuGENE HD and luciferase assay kit were purchased from Promega (Madison, WI). RT² real-time SYBR green/ROX PCR mix, collagen I (α2), and GAPDH primers were from Qiagen (Hilden, Germany). HA-tagged Hif1α, HA-tagged Myr Akt and FLAG-tagged constitutively active mTORC1 expression vectors have been described previously (14, 26, 52).

Cell culture and treatment.

HK2 human renal proximal tubular epithelial cells were grown in DMEM/F12 with 10% fetal bovine serum as described (18). The cells were starved in serum-free medium for 24 h before incubation with 27.5 mM D-glucose (high glucose, HG) for 24 h or indicated times. For osmotic control, 7.5 mM glucose (low glucose, LG) plus 20 mM mannitol was used. For PDGFRβ inhibitor, serum-starved cells were treated with 0.1 μM JNJ for 1 h before incubation with high glucose for 24 h.

Animals.

The monogenic Akita mouse model of type 1 diabetes was used. These mice are heterozygous for the missense Cys⁹⁶Tyr mutation in the insulin 2 (Ins2) gene, which produces a Cys⁹Tyr mutation in the A-chain of insulin (78). Male heterozygous Akita mice and their control littermates were purchased from The Jackson Laboratory (Bar Harbor, ME). These mice exhibit hyperglycemia within 2 wk of age and develop the pathologic features of diabetic nephropathy, including glomerular expansion, podocyte loss, and albuminuria (29, 59, 68). The mice were housed in the University of Texas Health Sciences San Antonio animal facility. They had free access to food and water. Mice were euthanized at 6 mo of age. Both kidneys were removed, and renal cortical tissues were isolated and frozen as described previously (19). The Institutional Animal Care and Use Committee approved the protocol.

Cell lysis, immunoblotting, and immunoprecipitation.

The cell monolayer was lysed in RIPA buffer (20 mM Tris·HCl pH 7.5, 5 mM EDTA, 150 mM NaCl, 1 mM Na₃VO₄, 1 mM NP-40, 1 mM PMSF, and 0.1% protease inhibitor cocktail) at 4°C for 30 min. In a similar fashion, the renal cortices from the control and Akita heterozygous diabetic mice were harvested in RIPA buffer. Lysed cells along with the debris were scraped from the dish. Cell and cortical extracts were centrifuged at 10,000 g for 20 min at 4°C. The supernatant was collected as cell and cortical lysates, and proteins were estimated. Equal amounts of cell lysates were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The membrane containing the separated proteins was immunoblotted with indicated antibodies. Protein bands were visualized with horseradish peroxidase-conjugated secondary antibodies using films (14). For immunoprecipitation, equal amounts of cell lysates or cortical extracts were incubated with the indicated antibody on ice for 30 min before adding protein G agarose beads. The mixture was rotated at 4°C overnight before washing with RIPA buffer. Immunobeads were suspended in SDS sample buffer and electrophoresed. The separated protein was then immunoblotted with the indicated antibody as described above.

RNA extraction and real-time quantitative RT-PCR.

Total RNAs were isolated using TRIzol reagent according to the protocol provided by the vendor and as described previously (19). RNA (1 μg) was used to synthesize first-strand cDNA using oligo(dT) and reverse transcriptase. Using a 96-well plate, the cDNA was amplified with human collagen I (α2) primer sets in a 7500 real-time PCR machine (Applied Biosystems, Foster City, CA). The PCR conditions were as follows: 94°C for 10 min; 45 cycles at 94°C for 30 s, 58°C for 30 s, and 72°C for 30 s. In the same sample, the level of GAPDH mRNA was measured and used for normalization of collagen I (α2) mRNA. Data were analyzed using the comparative C_t method as described (19).

Transfection.

Cells were transfected with 20 nM of pooled siRNAs against PDGFRβ or scramble RNA using FuGENE HD after the day of seeding. Briefly, the complete medium was removed and the cell monolayer was washed once with PBS. OPTIMEM medium was added to the cells. siRNAs and FuGENE mix in OPTIMEM was added according to the vendor’s instruction. Cells were incubated for 6 h at 37°C before adding complete medium. Transfected cells were grown to confluency and serum starved for 24 h before incubation with high glucose for 24 h. Where indicated, cells were transfected with 500 ng vector or HA-tagged Hif1α, HA-tagged constitutively active Myr Akt, or FLAG-tagged constitutively active mTOR expression plasmid using the same protocol described above.

Luciferase assay.

Proximal tubular epithelial cells were transfected with the Col-Luc reporter plasmid along with the siRNAs against PDGFRβ or plasmid expression vectors and treated with glucose as indicated. Luciferase activity was determined in cell lysates using a kit as described previously (17, 19). Values are presented as means ± SE of luciferase activity per microgram of protein (12).

Statistical analysis.

Data were analyzed by paired Student’s t-test or ANOVA followed by a Student-Newman-Keuls analysis. A value of P < 0.05 was considered significant (12, 16). Representative immunoblots of three to six independent experiments (indicated in figure legends) are shown. The bottom of each immunoblot shows the quantification of protein bands with statistical analysis. P values are described in the figure legends.

RESULTS

High glucose increases PDGFRβ autophosphorylation in proximal tubular epithelial cells.

Increased expression of PDGF-BB and PDGFRβ in the renal sections of diabetic rodents with nephropathy has been reported (57). Similarly, patients with diabetic nephropathy show elevated renal expression of PDGF-B predominantly in the glomeruli; however, expression of PDGF-B in the proximal tubules of patients with diabetes is also evident (45). The signaling role of PDGFRβ activation in renal cells by high glucose has not been investigated. Activation of PDGFRβ requires initial phosphorylation at Tyr⁸⁵⁷ in the activation loop (30). Therefore, we examined the phosphorylation of PDGFRβ in proximal tubular epithelial cells. High glucose increased the phosphorylation of PDGFRβ at Tyr⁸⁵⁷ in a time-dependent and sustained manner (Fig. 1, A and B). Because PDGFRβ undergoes autophosphorylation at this site, we used a specific ATP-competitive PDGFRβ inhibitor, JNJ (35). Incubation of proximal tubular epithelial cells with JNJ significantly inhibited high glucose-induced tyrosine phosphorylation of PDGFRβ (Fig. 1C).

Fig. 1.

High glucose (HG) promotes activating phosphorylation of platelet-derived growth factor receptor-β (PDGFRβ), which is not necessary for ERK1/2 MAPK activation. A, B, D, and E: proximal tubular epithelial cells were incubated with 27.5 mM glucose (HG) for the indicated time periods. As control (0 h), 20 mM mannitol was used with low glucose (LG) as described in materials and methods. Cell lysates were immunoblotted with phospho-PDGFRβ (Tyr⁸⁵⁷), PDGFRβ (A and B), phospho-Erk1/2 and Erk1/2 (D), and phospho-MEK and MEK (E) antibodies as indicated. C, F, and G: proximal tubular epithelial cells were treated with 0.1 μM JNJ for 1 h before incubation with HG or LG for 24 h. Cell lysates were immunoblotted with the indicated antibodies. H, I, and J: proximal tubular epithelial cells were transfected with scramble RNA or small interfering RNAs (siRNAs) against PDGFRβ as described in materials and methods. Cells were then incubated with LG or HG for 24 h (I and J). Cell lysates were immunoblotted with the indicated antibodies. A representative blot of four independent experiments is shown. Bottom in each blot represents quantification of the protein band. Values are means ± SE of four experiments. *P < 0.001 vs. LG; **P < 0.001 vs. HG or scramble in H.

We and others have shown that activation of PDGFRβ increases Erk1/2 MAPK activity by increasing phosphorylation of these kinases at Thr²⁰²/Tyr²⁰⁴ (4, 74). High glucose increased phosphorylation of Erk1/2 with similar kinetics (Fig. 1D) as autophosphorylation of PDGFRβ (Fig. 1A). Similarly, the upstream kinase of Erk1/2 activation, MEK, underwent phosphorylation in response to high glucose in a time-dependent manner (Fig. 1E). We examined whether activation of PDGFRβ is necessary for Erk1/2 and MEK phosphorylation. Interestingly, incubation of proximal tubular epithelial cells with JNJ did not have any effect on high glucose-induced phosphorylation of Erk1/2 and MEK (Fig. 1, F and G). To confirm these observations, we used PDGFRβ-targeted siRNAs, which inhibit expression of this receptor in proximal tubular epithelial cells (Fig. 1H). Similar to expression of JNJ, expression of siRNAs against PDGFRβ did not inhibit high glucose-stimulated Erk1/2 and MEK phosphorylation (Fig. 1, I and J). In contrast to these results in proximal tubular epithelial cells, PDGFRβ inhibition by JNJ, as judged by its activating phosphorylation (Fig. 2A), blocked high glucose-induced phosphorylation of Erk1/2 and its upstream kinase MEK in glomerular mesangial cells (Fig. 2, B and C). Similarly, siPDGFRβ significantly inhibited PDGF stimulated Erk1/2 and MEK phosphorylation (Fig. 2, D and E). These results indicate that high glucose activates PDGFRβ and Erk1/2 kinase cascade in both mesangial and proximal tubular epithelial cells. However, PDGFRβ activity is not required for high glucose-induced MAPK activation in proximal tubular epithelial cells. Thus, our data demonstrate differential regulation of signaling of PDGFRβ in these cells.

Fig. 2.

Inhibition of PDGFRβ blocks phosphorylation of Erk1/2 and MEK in glomerular mesangial cells in response to HG. A, B, and C: glomerular mesangial cells were incubated with 0.1 μM JNJ for 1 h before treatment with HG for 24 h as described in materials and methods. D and E: mesangial cells were transfected with scramble or siRNAs against PDGFRβ. Transfected cells were incubated with HG for 24 h. In all panels, cell lysates were immunoblotted with indicated antibodies. A representative blot of four independent experiments is shown. In A, *P < 0.01 vs. LG, **P < 0.01 vs. HG. In B, *P < 0.001 vs. LG, **P < 0.001 vs. HG. In C, *P < 0.001 vs. LG, **P < 0.01 vs. HG. In D, *P < 0.05 vs. LG, **P < 0.01 vs. HG.

High glucose-induced PI3-kinase/Akt signaling is dependent upon PDGFRβ.

Our data surprisingly showed no involvement of PDGFRβ in MAPK signaling by high glucose in proximal tubular epithelial cells. However, PDGFR is known to activate other signaling enzymes. One such pathway is PI3-kinase, which upon PDGFR activation associates with the receptor by its 85-kDa regulatory subunit (p85) (31). In fact, the SH2 domains of p85 physically bind to the phosphorylated PDGFRβ at Tyr⁷⁴⁰ and Tyr⁷⁵¹ after their phosphorylation by the activated receptor (55, 71). Therefore, we determined the phosphorylation at these sites in proximal tubular epithelial cells. High glucose time dependently increased the phosphorylation of PDGFRβ at Tyr⁷⁴⁰ and Tyr⁷⁵¹ (Fig. 3, A and B). This phosphorylation was significantly inhibited by JNJ (Fig. 3C). Next, we determined the association of p85 regulatory subunit with PDGFRβ. Incubation of proximal tubular epithelial cells with high glucose increased the association of p85 regulatory subunit of PI3-kinase with the PDGFRβ as judged by the coimmunoprecipitation experiment with p85 antibody followed by immunoblotting with PDGFRβ antibody (Fig. 3D). Treatment of cells with JNJ significantly blocked this association (Fig. 3D). Similarly, reciprocal coimmunoprecipitation showed inhibition of high glucose-stimulated PDGFRβ association with p85 by JNJ (Fig. 3E). Previously, it was reported that phosphorylation of p85 at Tyr⁴⁵⁸ is associated with the PI3-kinase activation (28, 65). High glucose increased the phosphorylation of p85 in a time-dependent manner (Fig. 3F). Incubation of cells with JNJ before high glucose stimulation significantly decreased the phosphorylation of p85 (Fig. 3G). Similarly, siRNA-mediated downregulation of PDGFRβ inhibited high glucose-induced phosphorylation of p85 (Fig. 3H). These results demonstrate the requirement of PDGFRβ for high glucose-stimulated activation of PI3-kinase.

Fig. 3.

HG-induced phosphorylation of PDGFRβ is required for phosphatidylinositol 3-kinase (PI3-kinase) phosphorylation. A, B, and F: proximal tubular epithelial cells were incubated with HG for the indicated times as described in Fig. 1A. Cell lysates were immunoblotted with phospho-PDGFRβ (Tyr⁷⁴⁰), phospho-PDGFRβ (Tyr⁷⁵¹), or PDGFRβ (A and B); and phospho-p85 (Tyr⁴⁵⁸) and p85 (F) antibodies as indicated. C–E, and G: proximal tubular epithelial cells were treated with 0.1 μM JNJ for 1 h before incubation with LG or HG for 24 h. Cell lysates were used to immunoblot with phospho-PDGFRβ (C) or phospho-p85 (G) antibodies as indicated. In D and E, cell lysates were immunoprecipitated with the p85 subunit of PI3-kinase (D) or with PDGFRβ (E) antibodies. Immunoprecipitates were immunoblotted with antibodies against PDGFRβ (D) and p85 (E). H: proximal tubular epithelial cells were transfected with scramble or siRNAs against PDGFRβ as described in materials and methods. Serum-starved transfected cells were incubated with HG for 24 h. Cell lysates were immunoblotted with phospho-p85, p85, PDGFRβ, and actin antibodies. A representative blot of three (A) or four (B–H) independent experiments is shown. Bottom in each blot represents quantification of the protein band. In A, values are means ± SE of three experiments. *P < 0.01 and 0.05 vs. unstimulated at left and right, respectively. In all other panels, values are means ± SE of four independent experiments. *P < 0.001 vs. unstimulated in B–D, F–H, **P < 0.001 vs. HG as indicated. In E, *P < 0.05 vs. LG, **P < 0.01 vs. HG.

Activation of PI3-kinase controls Akt kinase activity by inducing phosphorylation of this kinase at the catalytic loop Thr³⁰⁸ and hydrophobic motif Ser⁴⁷³ sites (53). High glucose increased phosphorylation of Akt at these sites in proximal tubular epithelial cells in a time-dependent manner (Fig. 4A). JNJ inhibited high glucose-induced phosphorylation at Thr³⁰⁸ and Ser⁴⁷³ (Fig. 4B). Because phosphorylation at both these sites increases the kinase activity of Akt, we determined the phosphorylation of one of its endogenous substrates, GSK3β. JNJ attenuated the phosphorylation of GSK3β by high glucose (Fig. 4C). Also, siPDGFRβ significantly inhibited the high glucose-stimulated phosphorylation of Akt and GSK3β (Fig. 4, D and E). These results indicate that activation of PDGFRβ by high glucose increases Akt kinase activity.

Fig. 4.

HG-stimulated Akt activity is dependent upon PDGFRβ activation. A: proximal tubular epithelial cells were incubated with HG for indicated periods of time. Cell lysates were immunoblotted with phospho-Akt (Thr³⁰⁸), phospho-Akt (Ser⁴⁷³), and Akt antibodies. B and C: proximal tubular epithelial cells were treated with JNJ before incubation with LG or HG. Cell lysates were immunoblotted with phospho-Akt or phospho-GSK3β as indicated. D and E: proximal tubular epithelial cells were transfected with siPDGFRβ. Cells were incubated with LG or HG. Cell lysates were immunoblotted with the indicated antibodies. A representative blot of four independent experiments is shown. Bottom shows quantification of protein bands. In A, *P < 0.05 to 0.001 vs. LG; in B–E, *P < 0.001 vs. LG; **P < 0.001 vs. HG.

High glucose-induced PDGFRβ controls mTORC1 activity.

We and other researchers have shown that high glucose increases mTORC1 activity in renal cells (13, 16, 38). Two proteins, tuberin and PRAS40, act as negative regulators of mTORC1 activity (46). Both these proteins are inactivated when phosphorylated by Akt kinase (79). We examined the involvement of PDGFRβ in this process. JNJ significantly inhibited the high glucose-induced phosphorylation of tuberin and PRAS40 (Fig. 5, A and B). Similarly, siPDGFRβ blocked phosphorylation of both these proteins (Fig. 5, C and D). Because phosphorylation of these proteins increases mTORC1 activity, we determined its activity by measuring the phosphorylation of its endogenous substrates, 4EBP-1 and S6 kinase (79). High glucose increased phosphorylation of 4EBP-1 and S6 kinase (Fig. 6, A and B). Both JNJ and siPDGFRβ significantly inhibited high glucose-stimulated phosphorylation of these proteins (Fig. 6, A–D). Phosphorylation of S6 kinase by mTORC1 increases its activity toward its substrate ribosomal protein s6 (rps6). JNJ and siPDGFRβ blocked the phosphorylation of rps6 in response to high glucose (Fig. 6, E and F). We have shown previously that phosphorylation and inactivation of 4EBP-1 is accompanied by phosphorylation of eIF4E in proximal tubular epithelial cells by high glucose (42, 54). We examined the sensitivity of phosphorylation of eIF4E to PDGFRβ inhibition. High glucose increased the phosphorylation of eIF4E. However, JNJ and siRNAs against PDGFRβ did not have any effect on this phosphorylation (Fig. 6, G and H). These results conclusively demonstrate that activated PDGFRβ by high glucose contributes to mTORC1 activation to phosphorylate 4EBP-1 and S6 kinase; however, PDGFRβ activation does not affect eIF4E phosphorylation in response to high glucose.

Fig. 5.

HG-induced PDGFRβ activation regulates tuberin and PRAS40 phosphorylation. A and B: proximal tubular epithelial cells were treated with JNJ before incubation with HG or LG. C and D: cells were transfected with siPDGFRβ. Cells were then incubated with LG or HG. Cell lysates were immunoblotted with phospho-tuberin (Thr¹⁴⁶²) and tuberin (A and C), phospho-PRAS40 (Thr²⁴⁶), and PRAS40 (B and D). A representative blot of four independent experiments is shown. Bottom shows quantification of protein bands. *P < 0.001 vs. LG; **P < 0.001 vs. HG.

Fig. 6.

PDGFRβ regulates HG-induced mammalian target of rapamycin complex 1 (mTORC1) activation. A, B, E, and G: proximal tubular epithelial cells were treated with JNJ for 1 h before incubation with LG or HG. Cell lysates were immunoblotted with phospho-4EBP-1 (Thr^37/46), phospho-4EBP-1 (Ser⁶⁵), and 4EBP-1 (A); phospho-S6 kinase (Thr³⁸⁹) and S6 kinase (B); phosphoribosomal protein s6 (Ser^240/244) and s6 (E); and phospho-eIF4E (Ser²⁰⁹) and eIF4E (G) antibodies as indicated. C, D, F, and H: proximal tubular epithelial cells were transfected with siPDGFRβ and incubated with HG as described in materials and methods. Cell lysates were then immunoblotted with the antibodies as indicated. In A–G, representative blots of four independent experiments (six independent experiments in H) is shown. Bottom shows quantification of protein bands. *P < 0.001 vs. LG; *P < 0.001 vs. HG.

Activation of PDGFRβ by high glucose regulates proximal tubular cell matrix protein synthesis.

Recently, we have shown that mTORC1 regulates matrix protein expression in proximal tubular epithelial cells (11). Therefore, we determined the effect of PDGFRβ inhibition on the expression of the matrix protein collagen I (α2) in proximal tubular epithelial cells. As expected, high glucose increased the expression of collagen I (α2) protein (Fig. 7, A and B). Incubation of cells with JNJ or transfection of siPDGFRβ significantly inhibited high glucose-stimulated collagen I (α2) expression (Fig. 7, A and B). Similarly, JNJ and siPDGFRβ suppressed the expression of collagen I (α2) mRNA (Fig. 7, C and D). Expression of collagen I (α2) is known to be regulated by a transcriptional mechanism (61). Therefore, we used a reporter construct in which the luciferase gene is driven by the collagen I (α2) promoter (11). High glucose significantly increased the transcription of collagen I (α2) (Fig. 7, E and F). Both JNJ and siRNAs against PDGFRβ markedly inhibited the high glucose-induced transcription of collagen I (α2) (Fig. 7, E and F). Together these results indicate that high glucose-stimulated PDGFRβ contributes to matrix protein collagen I (α2) expression.

Fig. 7.

PDGFRβ controls HG-induced collagen I (α2) expression. Proximal tubular epithelial cells were treated with JNJ (A, C, and E) or transfected with siPDGFRβ (B, D, and F) before incubation with LG or HG as described in materials and methods. In A and B, cell lysates were immunoblotted with collagen I (α2), PDGFRβ, and actin antibodies as indicated. Representative blots of four independent experiments are shown. *P < 0.001 vs. LG, **P < 0.001 vs. HG. C and D: total RNAs were used in real-time RT-PCR to detect collagen I (α2) mRNA as described in materials and methods (19). Values are means ± SE of six measurements. *P < 0.001 vs. control, **P < 0.001 vs. HG alone. E: proximal tubular epithelial cells were transfected with collagen I (α2) promoter-driven luciferase plasmid. Transfected cells were incubated with JNJ. F: proximal tubular epithelial cells were transfected with collagen I (α2) promoter-driven luciferase plasmid along with scramble or siPDGFRβ. Cell lysates were used to assay luciferase activity as described in materials and methods (17, 19). Values are means ± SE of five replicates. *P < 0.001 vs. control, **P < 0.01 vs. HG alone. Bottom in D and F shows expression of PDGFRβ.

PDGFRβ-stimulated Akt/mTORC1 regulates high glucose-induced collagen I (α2) expression.

We have shown above that high glucose-stimulated Akt activity is mediated by PDGFRβ (Fig. 4). Previously, we showed that Akt regulates matrix protein expression by high glucose (17, 19). We determined the involvement of cross talk between PDGFRβ and Akt in high glucose-induced collagen I (α2) expression. We used a constitutively active (CA) Myr Akt (26). Expression of the Myr Akt significantly reversed the siPDGFRβ-induced inhibition of collagen I (α2) protein and mRNA expression (Fig. 8, A and B). Similarly, Myr Akt restored the siPDGFRβ-mediated inhibition of transcription of high glucose-induced collagen I (α2) (Fig. 8C). We showed earlier that high glucose activates mTORC1 through PDGFRβ (Fig. 6). Myr Akt alone stimulated collagen I (α2) expression, similar to that of high glucose (Figs. 8, A–C). We have previously shown a role for mTORC1 in matrix protein expression in renal cells (17, 63). We examined whether PDGFRβ-regulated mTORC1 controls collagen I (α2) expression by using a CA mTORC1 (12, 16). Expression of siPDGFRβ inhibited the high glucose-induced expression of collagen I (α2) protein and mRNA (Fig. 8, D and E). Similarly, siPDGFRβ blocked the transcription of collagen I (α2) in response to high glucose (Fig. 8F). Interestingly, expression of constitutively active mTORC1 reversed the inhibition of collagen I (α2) expression and transcription by siPDGFRβ (Fig. 8, D–F). Together, these results demonstrate the requirement of Akt/mTORC1 activation downstream of PDGFRβ in high glucose-induced collagen I (α2) expression.

Fig. 8.

PDGFRβ-stimulated Akt/mTORC1 axis controls HG-induced collagen I (α2) expression. A, B, D, and E: proximal tubular epithelial cells were transfected with constitutively active Myr Akt and siPDGFRβ (A and B) or constitutively active (CA) mTOR and siPDGFRβ (D and E). Cell lysates were immunoblotted with antibodies against collagen I (α2), PDGFRβ, HA (to detect Myr Akt), FLAG (to detect CA mTOR), and actin (A and D). Representative blots of four independent experiments are shown. *P < 0.01 or 0.001 vs. LG, **P < 0.01 or 0.001 vs. HG; @P < 0.01 or 0.001 vs. HG plus siPDGFRβ. In B and E, total RNAs were used in real-time RT-PCR to detect collagen I (α2) mRNA. Values are means ± SE of quadruplicate measurements. *P < 0.001 or 0.01 vs. control, **P < 0.001 vs. HG alone, @P < 0.001 vs. HG with siPDGFRβ. Bottom: expression of PDGFRβ, HA Myr Akt, and FLAG CA mTOR in parallel samples. C and F: proximal tubular epithelial cells were transfected with collagen I (α2) promoter-driven luciferase plasmid along with Myr Akt (C) or CA mTOR (F) and siPDGFRβ. Cell lysates were used to assay luciferase activity as described in materials and methods (17, 19). Values are means ± SE of quadruplicate measurements. *P < 0.001 or 0.01 vs. control, **P < 0.001 vs. HG alone, @P < 0.001 vs. HG with siPDGFRβ. Bottom: expression of PDGFRβ, HA Myr Akt and FLAG CA mTOR in the lysates.

PDGFRβ regulates Hif1α to promote high glucose-induced collagen I (α2) expression.

We and others have recently shown that Hif1α controls the expression of collagen I (α2) in renal cells (2, 11). Incubation of proximal tubular epithelial cells with high glucose increased the expression of Hif1α in a time-dependent manner (Fig. 9A). To examine the involvement of PDGFRβ, we used its inhibitor, JNJ. Incubation of cells with JNJ significantly inhibited expression of Hif1α by high glucose (Fig. 9B). Similarly, siRNAs against PDGFRβ markedly inhibited Hif1α expression (Fig. 9C), indicating involvement of PDGFRβ in Hif1α expression. Also, inhibition of Hif1α expression using siRNAs blocked the expression of collagen I (α2) protein (Fig. 9D). To investigate the cross talk between PDGFRβ and Hif1α, we transfected an expression vector for Hif1α along with siRNAs against PDGFRβ into proximal tubular epithelial cells. Expression of Hif1α reversed the siRNA-mediated inhibition of high glucose-induced collagen I (α2) protein and mRNA expression (Fig. 9, E and F). Similarly, Hif1α prevented the inhibitory effect of PDGFRβ siRNAs on high glucose-stimulated transcription of collagen I (α2) (Fig. 9G). These results conclusively demonstrate that Hif1α downstream of PDGFRβ contributes to collagen I (α2) expression in response to high glucose.

Fig. 9.

HG-induced PDGFRβ regulates Hif1α-dependent collagen I (α2) expression. A: cells were incubated with HG as described in Fig. 1A. Cell lysates were immunoblotted with Hif1α and actin antibodies. B: proximal tubular epithelial cells were treated with JNJ before incubation with HG for 24 h. Cell lysates were immunoblotted with Hif1α and actin antibodies. C and D: cells were transfected with scramble RNA or siPDGFRβ or Hif1α siRNAs as indicated. Cells were then incubated with HG for 24 h. Cell lysates were immunoblotted with Hif1α, collagen I (α2), PDGFRβ, and actin antibodies as indicated. E and F: cells were transfected with Hif1α along with siPDGFRβ as indicated. Transfected cells were incubated with HG. In E, cell lysates were immunoblotted with collagen I (α2), PDGFRβ, HA (to detect HA-tagged Hif1α), and actin antibodies. In A–E, representative blots of four independent experiments are shown. *P < 0.001 vs. control, **P < 0.01 or 0.001 vs. HG. In F, total RNAs were used to detect collagen I (α2) mRNA as described in materials and methods (19). Values are means ± SE of quadruplicate determinations. *P < 0.001 vs. control, **P < 0.001 vs. HG alone, @P < 0.001 vs. HG plus siPDGFRβ. G: cells were transfected with collagen I (α2) promoter-driven luciferase construct along with siPDGFRβ and Hif1α as indicated. Transfected cells were incubated with HG. Cell lysates were assayed for luciferase activity as described (17, 19). Values are means ± SE of quadruplicate measurements. *P < 0.01 vs. control, **P < 0.01 vs. HG alone, @P < 0.01 vs. HG plus siPDGFRβ, #P < 0.05 vs. control. Bottom in F and G shows expression of PDGFRβ and HA-tagged Hif1α in parallel samples.

Phosphorylation of PDGFRβ in the kidneys of Akita type 1 diabetic mice.

Increased expression of the PDGF-B chain has been reported in patients with diabetes (45). Our results described above show activation of PDGFRβ in response to high glucose. Also, activated PDGFRβ is necessary for collagen I (α2) expression in proximal tubular epithelial cells. To examine the relevance of our observations in vivo, we used the Akita mouse model, which displays type 1 diabetes within 2 wk of age (78). The heterozygous Akita mouse exhibits the pathologic features of diabetic nephropathy, including glomerular expansion, podocyte loss, and albuminuria (29, 59, 68). We determined the phosphorylation of PDGFRβ at Tyr⁸⁵⁷, which is known to be required for increased tyrosine kinase activity, in the renal cortex of diabetic Akita mice. The results show a significant increase in the phosphorylation of PDGFRβ at Tyr⁸⁵⁷ (Fig. 10, A and B). Similarly, levels of tyrosine phosphorylation at Tyr⁷⁴⁰ and Tyr⁷⁵¹ were markedly increased in the renal cortex of Akita diabetic mice (Fig. 10, C and D). Levels of PDGFRβ were slightly increased in diabetic animals (Fig. 10, A–D). These phosphotyrosines of the activated PDGFRβ bind to the p85 subunit of PI3-kinase (71). Thus, our results indicate that PI3-kinase may associate with the PDGFRβ in the diabetic kidney. We directly examined the association of these two proteins. PDGFRβ was immunoprecipitated from the cortical lysates of control and diabetic mice followed by immunoblotting with p85 antibody. The results show significantly elevated association of p85 with the PDGFRβ in the diabetic renal cortex (Fig. 10, E and F). Reciprocal immunoprecipitation/immunoblotting confirmed this observation (Fig. 10, G and H). The association of PI3-kinase with the PDGFRβ induces its tyrosine phosphorylation and activation (28, 65, 71). In fact, in renal cortices of diabetic mice, we detected increased tyrosine phosphorylation of p85 regulatory subunit compared with that in the kidney cortices of control mice (Fig. 10, I and J). These results indicate that increased activation of PDGFRβ is associated with phosphorylation of PI3-kinase in the diabetic kidney.

Fig. 10.

Activation of PDGFRβ signaling in the kidney cortex of type 1 diabetic Akita mice. A and C: lysates of renal cortexes from Akita diabetic mice (D) and control littermates (C) were immunoblotted with phospho-PDGFRβ (Tyr⁸⁵⁷) (A), or phospho-PDGFRβ (Tyr⁷⁴⁰ and Tyr⁷⁵¹) (C) PDGFRβ and actin antibodies as indicated. Each lane represents an individual animal. B and D: quantification of the data presented in A and C, respectively. Cortical lysates were immunoprecipitated with PDGFRβ (E) or p85 (G) antibodies. Immunoprecipitates were immunoblotted with p85 (E) or PDGFRβ (G) antibodies as indicated. F and H: quantification of data presented in E and G, respectively. Values are means ± SE of four animals. I, K, M, O, and Q: cortical lysates were immunoblotted with phospho-p85 (Tyr⁴⁵⁸), p85, phospho-Akt (Thr³⁰⁸ and Ser⁴⁷³), Akt, phospho-S6 kinase (Thr³⁸⁹), S6 kinase, Hif1α, collagen I (α2), and actin antibodies as indicated. J, L, N, P, and R: quantification of data presented in I, K, M, O, and Q, respectively. Values are means ± SE of four animals per group.

We have shown that PDGFRβ promoted phosphorylation of Akt in proximal tubular epithelial cells (Fig. 4). Our results demonstrate that enhanced phosphorylation of PDGFRβ was associated with increased phosphorylation of Akt at both catalytic loop and hydrophobic motif sites in the cortex of diabetic mice (Fig. 10, K and L). Because activation of Akt results in mTORC1 activation, we determined its activation by measuring the phosphorylation of its substrate, S6-kinase. As shown in Fig. 10, M and N, phosphorylation of S6-kinase was significantly increased in the diabetic renal cortex, indicating activation of mTORC1 in the kidneys of Akita mice.

Our results in proximal tubular epithelial cells demonstrate a role for PDGFRβ-mediated Hif1α in collagen I (α2) expression (Fig. 9). Therefore, we examined the expression of Hif1α in the renal cortexes of diabetic mice. The data show a significant increase in expression of Hif1α in diabetic kidneys (Fig. 10, O and P). Furthermore, this elevated Hif1α was associated with a significant increase in collagen I (α2) levels in the renal cortexes of mice with type 1 diabetes (Fig. 10, Q and R). These results demonstrate a possible role for activated PDGFRβ in the pathology of diabetic kidney disease.

DISCUSSION

In the present study, we show that high glucose-stimulated PDGFRβ contributes to matrix protein collagen I (α2) expression in proximal tubular epithelial cells. We provide novel evidence that in these cells, the PI3-kinase/Akt pathway activation but not MAPK signaling in response to high glucose is mediated by activated PDGFRβ. Our novel observations demonstrate that PDGFRβ-activated Akt and mTORC1 kinases are necessary for collagen I (α2) expression. Finally, we show that activation of PDGFRβ by high glucose increases Hif1α to induce collagen I (α2) expression (Fig. 11). To our knowledge, this is the first demonstration of involvement of PDGFRβ in Hif1α expression in response to high glucose. Furthermore, our data show the requirement of PDGFRβ signal transduction for high glucose induction of matrix protein; these findings are relevant to mechanisms underlying renal tubulointerstitial fibrosis in diabetes.

Fig. 11.

Schematic summary of results of the present study.

Tubular changes are early features of diabetic nephropathy (72). The molecular signature that contributes to proximal tubular matrix protein expansion in diabetic kidney disease involves high glucose-induced signal transduction, which leads to tubular basement thickening and tubulointerstitial fibrosis. In the diabetic kidney milieu, increased expression of many growth factors and cytokines promotes early structural changes. Among many growth factors, PDGFs have been shown to be expressed in rodent and human diabetic kidneys (45, 57). Although contradictory results exist about the expression of PDGF-A mRNA in diabetic human kidneys (45, 70), significant renal expression of both PDGF-B and receptor-β has been reported in humans (70). Similarly, increased expression of PDGF-B chain and receptor-β were observed in diabetic renal glomeruli in rodents (57). Also, the proximal tubular compartment showed increased levels of both PDGF-A and -B (45). In fact, using proximal tubular epithelial cells incubated with high glucose for 7 days, Fraser et al. (24) showed PDGFRα-mediated expression of TGF-β. In the present study, we demonstrate increased activation of PDGFRβ by high glucose (Fig. 1, A and B). It is possible that in proximal tubular epithelial cells, increased expression of the PDGF B-chain by high glucose increases the abundance of PDGF-BB, resulting in activation of PDGFRβ. Moreover, increased urinary excretion of PDGF-BB has been reported in patients with diabetic nephropathy (22, 75). Thus, in vivo, due to the low molecular weight of the PDGF-BB homodimer, its reabsorption by the injured diabetic kidney may expose proximal tubular cells to high concentrations of this growth factor, rendering sustained stimulation of PDGFRβ.

PDGFRβ is known to stimulate Erk1/2 MAPK. Interestingly, activated PDGFRβ by high glucose did not promote MEK/Erk1/2 phosphorylation in proximal tubular epithelial cells (Fig. 1, G–J). To elucidate the cell-specific nature of this phenomenon, we used glomerular mesangial cells, which abundantly express PDGFRβ (1). Interestingly, in contrast to the observation in proximal tubular epithelial cells (Fig. 1, F–J), inhibition of PDGFRβ by JNJ or siPDGFRβ significantly blocked the high glucose-induced phosphorylation of MEK and Erk1/2 in glomerular mesangial cells (Fig. 2). These novel results demonstrate that although high glucose activates PDGFRβ in both cell types, it selectively activates Erk1/2 in mesangial cells but not in proximal tubular epithelial cells. Activated PDGFRβ contains 11 autophosphorylated tyrosine residues that serve as the docking sites for different SH2-domain containing signaling enzymes and adaptor proteins (30). One of the signaling enzymes is the heterodimeric PI3-kinase, which contains the regulatory subunit p85 with the SH2 domains in its NH₂-terminus. The p85 binds to phosphotyrosine residues 740 and 751 of activated PDGFRβ. Our data for the first time show increased phosphorylation of these tyrosine residues in PDGFRβ in response to high glucose. We also showed increased association of the p85 subunit of PI3-kinase with the activated receptor. In fact, this association resulted in activation of PI3-kinase as shown by increased tyrosine phosphorylation of p85 and phosphorylation and activation of its downstream kinase, Akt (Figs. 3 and 4). These results conclusively provide evidence for the requirement of PDGFRβ in high glucose-induced Akt activation.

Recent reports indicate that hyperglycemia-induced activation of mTOR is necessary for renal matrix protein expansion, a pathologic feature of diabetic nephropathy (21, 50, 56, 63). mTOR forms two kinase complexes containing common and distinct protein subunits. mTORC1 has raptor, mLST8/GβL, Sin1, PRAS40, and deptor; whereas mTORC2 contains rictor, mLST8/GβL, Sin1, protor, and deptor (79). We have shown recently that both mTORC1 and mTORC2 are required for renal cell matrix protein expansion (13, 15). mTORC2 contributes to the activating phosphorylation of Akt at both Thr³⁰⁸ and Ser⁴⁷³ sites (16). Activated Akt phosphorylates the two negative regulators of mTORC1, tuberin, and PRAS40. Phosphorylation of tuberin inhibits its GTPase-activating protein activity toward Rheb GTPase (37). Rheb binds to mTORC1 at a low affinity, and when charged with guanosine 5′-triphosphate, promotes mTORC1 activity toward its substrates (79). An additional mechanism of mTORC1 activation includes PRAS40. Its phosphorylation by activated Akt kinase at Thr²⁴⁶ was originally identified by insulin (44). However, later it was found that PRAS40 represents an intrinsic subunit of mTORC1 and is a negative regulator of its activity. When phosphorylated at Thr²⁴⁶, it undergoes inactivation, resulting in activation of mTORC1 (62). We show that high glucose-mediated phosphorylation of tuberin and PRAS40 is dependent upon PDGFRβ (Fig. 5). Also, we previously showed a requirement for PI3-kinase in high glucose-stimulated phosphorylation of PRAS40 in glomerular mesangial cells (20). In the present study, we demonstrate activation of PI3-kinase by PDGFRβ in high glucose-treated proximal tubular epithelial cells. Furthermore, our novel data for the first time demonstrate the requirement of PDGFRβ for the activation of mTORC1 in response to high glucose (Fig. 6).

Phosphorylation of 4EBP-1 by mTORC1 promotes its dissociation from the eIF4E-4EBP-1 complex (42, 67). Free eIF4E binds to eIF4G and eIF4A to form the eIF4F complex, which initiates the translation of the 5′ cap-containing mRNAs (67). Also, eIF4E undergoes phosphorylation at the Ser²⁰⁹ residue during initiation of protein synthesis (76). Although the Ser²⁰⁹ phosphorylation site of eIF4E is located in the cap-interaction region, its role for eIF4E binding to mRNA Cap is controversial. Shibata et al. (66) showed increased affinity of eIF4E for Cap binding when Ser²⁰⁹ was changed to an acidic residue. In contrast, an acidic mutation of the same site resulted in reduction in the interaction (64). Our results in the present study demonstrate increased phosphorylation of eIF4E in response to high glucose. Interestingly, high glucose-stimulated PDGFRβ did not regulate this phosphorylation (Fig. 6, G and H). This observation is inconsistent with our data demonstrating lack of any effect of PDGFRβ activation by high glucose on MEK/Erk phosphorylation, the upstream kinase cascade that regulates the phosphorylation of eIF4E, in proximal tubular epithelial cells (Fig. 1, F–J). Although eIF4E retains its phosphorylation when PDGFRβ is inhibited (Fig. 6, G and H), high glucose-induced collagen I (α2) expression were significantly attenuated (Fig. 7). One interpretation of these data is that high glucose-stimulated phosphorylation of eIF4E may not contribute to collagen I (α2) expression by high glucose. Also, our results suggest that PDGFRβ regulates mTORC1 activity in response to high glucose by the activation of the PI3-kinase/Akt cascade (Fig. 3). Thus inhibition of PDGFRβ results in attenuation of mTORC1-dependent phosphorylation of the translation repressor 4EBP-1, leading to stabilization of the 4EBP-1-eIF4E complex. Furthermore, we conclusively demonstrate that PDGFRβ-stimulated Akt and mTORC1 activities contribute to proximal tubular cell collagen I (α2) expression (Fig. 8). Previously, we reported that the Akt/mTORC1 pathway is involved in high glucose-induced mesangial cell hypertrophy, another pathological feature of diabetic nephropathy (19, 20, 51). We investigated the role of PDGFRβ in this process. Both JNJ and siRNA against PDGFRβ significantly inhibited the protein synthesis and hypertrophy of proximal tubular epithelial and mesangial cells induced by high glucose (data not shown). These data demonstrate the involvement of the PDGFRβ signaling pathway in mediating the same outcome in both cell types. Thus, use of a specific inhibitor of PDGFRβ may be beneficial to block hyperglycemia-induced pathological features of kidney.

In patients with diabetic nephropathy, increased expression of the basic helix-loop-helix transcription factor Hif1α was detected in the tubular compartment, which correlated with elevated expression of Hif1α target genes (34). In a mouse model of unilateral ureteral obstruction (UUO), stable expression of Hif1α increased renal fibrosis (43). Similarly, mice with deletion of Hif1α in proximal tubular cells showed attenuated development of fibrosis in UUO kidneys (33). Furthermore, the anti-Hif1α agent YC-1 prevented the progression of kidney fibrosis in both UUO and in a type 1 diabetic model of nephropathy (43, 60). These results conclusively demonstrate a role for Hif1α in tubular fibrosis. In the present study, we show increased expression of Hif1α in high glucose-treated proximal tubular epithelial cells (Fig. 9A). Also, expression of collagen I (α2) expression was dependent upon Hif1α (Fig. 9D). Because the normoxic level of Hif1α is regulated by mTORC1 (6, 14, 52) and PDGFRβ downstream of high glucose regulates mTORC1 activity (Fig. 6), our data show a role for PDGFRβ in high glucose-induced Hif1α expression (Fig. 9, B and C). This novel observation is the first demonstration of involvement of PDGFRβ in the upregulation of Hif1α by high glucose. We also provide the first evidence for the requirement of Hif1α downstream of PDGFRβ in the expression of collagen I (α2) in response to high glucose (Fig. 9, E–G).

We have recently shown that use of rapamycin in diabetic mice attenuated renal hypertrophy, albuminuria, and matrix protein expansion, suggesting a role for mTORC1 in renal pathology (21, 63). In the current study, our results demonstrate the requirement for PDGFRβ in high glucose-induced mTORC1 activation and the downstream collagen I (α2) expression in proximal tubular cells. Moreover, phosphorylation of PDGFRβ and activation of Akt/mTORC1 signaling is necessary for Hif1α expression, and increased collagen I (α2) levels are confirmed in the renal cortex of Akita mice with type 1 diabetes (Fig. 10). However, a significant adverse effect was found with chronic treatment with rapamycin, which produced glucose intolerance in mice and insulin resistance in rats (10, 23, 36). Similar observations may be possibly true in humans (40, 69). Also, complete loss of proximal tubular mTORC1 activity itself is sufficient to induce progressive renal fibrosis (27). Thus, although mTORC1 inhibition may show a beneficial effect in preclinical studies in models of rodent diabetic nephropathy, its adverse effect may limit its efficacy. Therefore, an alternative target that uses the same signaling pathway may be of great importance in treating diabetic nephropathy. In fact, in an accelerated form of streptozotocin-induced diabetes in apolipoprotein E-knockout mouse, use of imatinib, a broad-spectrum tyrosine kinase inhibitor of c-Abl, c-kit, and PDGFR, showed attenuation of both glomerular and tubular pathology, albuminuria, and PDGF-B chain and PDGFRβ expression (47). Using a specific PDGFRβ inhibitor, JNJ, in the present study, we demonstrate amelioration of proximal tubular epithelial cell collagen I (α2) by high glucose. Thus, we outline a signaling circuit involving PDGFRβ signaling in response to high glucose that can be therapeutically exploited by using a highly specific inhibitor of this tyrosine kinase.

GRANTS

Support for this work was provided by Department of Veterans Affairs Biomedical Laboratory Research and Development Merit Review Award 2 I01 BX000926 and by National Institute of Diabetes and Digestive and Kidney Diseases Grant RO1 DK-50190 to G. G. Choudhury. G. Ghosh-Choudhury is also a recipient of a Department of Veterans Affairs Biomedical Laboratory Research and Development Service Senior Research Career Scientist Award. N. Ghosh-Choudhury and B. S. Kasinath are supported by Department of Veterans Affairs Merit Review Grants I0 1 BX 000150 and I01 BX 001340, respectively. N. Ghosh-Choudhury is a recipient of a research grant from the San Antonio Area Foundation.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

DISCLAIMERS

The contents do not represent the views of the US Department of Veterans Affairs or the United States Government.

AUTHOR CONTRIBUTIONS

G.G.C. conceived and designed research; F.D. and B.V. performed experiments; F.D. analyzed data; G.G.C. interpreted results of experiments; G.G.C. prepared figures; G.G.C. drafted manuscript; N.G.-C., B.S.K., and G.G.C. edited and revised manuscript; F.D., N.G.-C., B.V., and B.S.K. approved final version of manuscript.