Hypoxia-inducible factor-2α and TGF-β signaling interact to promote normoxic glomerular fibrogenesis

Abstract

Hypoxia-inducible factors (HIFs) are transcription factors consisting of an oxygen-sensitive α-subunit binding to a stable β-subunit. HIFs regulate multiple signaling pathways that could contribute to fibrogenesis, supporting their potential role in hypoxia-mediated renal fibrosis. We previously reported that HIF-1 is upregulated and required for transforming growth factor (TGF)-β induction of collagen in renal tubular cells. Here, we performed in vitro and in vivo studies of potential glomerular crosstalk between TGF-β and normoxic HIF signaling. HIF-α has two major isoforms, HIF-1α and HIF-2α with different target gene sets. In cultured human mesangial cells, TGF-β1 treatment increased both HIF-1α and HIF-2α expression in normoxia. TGF-β1 did not increase HIF-1α/2α mRNA levels nor decrease the rate of protein degradation, suggesting that it enhances HIF-1α/2α expression through translation. TGF-β receptor (ALK5) kinase activity was required for increased, TGF-β-stimulated HIF-α expression in response to TGF-β, and inhibiting PI3-kinase markedly decreased HIF-α expression. Blocking HIF-1α/2α expression using siRNA decreased basal and TGF-β1-stimulated type I collagen expression, while overexpressing nondegradable HIF-α increased the collagen response, with HIF-2α being significantly more effective than HIF-1α. In adriamycin-induced mouse glomerulosclerosis, HIF-2α target genes were upregulated in sclerosing glomeruli. Taken together, our data demonstrate potential signaling interaction between TGF-β and HIFs to promote renal fibrogenesis in normoxia and suggest that the HIF-2α isoform is more important during glomerulosclerosis.

the final pathway of progression to end-stage kidney disease involves a common set of events that include renal fibrosis (24). Experimental evidence suggests that hypoxia may promote subsequent kidney fibrogenesis (11). Cellular adaptation to hypoxia is mediated by hypoxia-inducible factors (HIFs), heterodimeric transcription factors consisting of an oxygen-sensitive α-subunit binding a β-subunit, leading to subsequent activation of HIF-responsive elements (HRE) and promoting the expression of prosurvival genes (4, 14, 27). Under normoxic conditions, the HIF α chain is rapidly eliminated by oxygen-dependent prolyl hydroxylase (PHD) modification of HIF-α, enabling its interaction with the von Hippel-Lindau tumor suppressor, subsequent ubiquitin tagging and proteosomal degradation. Under hypoxic conditions, this oxygen-requiring hydroxylation is inhibited, thereby stabilizing the expression and activity of HIF-α (10). HIF-α has three isoforms in the human, of which HIF-1α and HIF-2α are closely related. HIFs promote myriad target gene transcriptions, most being common to both HIF-1 and HIF-2, while some are isoform specific, contributing to different cellular phenotypes (12, 22).

Transforming growth factor-β (TGF-β) is a pleiotropic cytokine regulating multiple cell functions including growth, differentiation, proliferation, migration, epithelial-to-mesenchymal transition, and extracellular matrix balance (5, 19, 25). Indeed, TGF-β is a major contributor to collagen accumulation in progressive kidney disease (28). We previously showed that TGF-β1 increases renal tubular cell (HKC) HIF-1α expression even in normoxia through a mechanism dependent on the kinase activity of the type I TGF-β receptor (TβRI), via the Smad-3 pathway that mediates expression of TGF-β target genes, including collagen I (2). The mechanism of TGF-β1-dependent HIF-1α expression in renal tubular cells is distinct from, and additive to, hypoxia-dependent stabilization of HIF-1α protein. Overexpression of nondegradable (ND)-HIF-1α also enhanced TGF-β induction of collagen I promoter activity in cultured, normoxic epithelial cells.

These data suggest that HIFs may contribute to renal fibrosis even in the absence of hypoxia. Because the tubulointerstitium is a relatively hypoxic microenvironment, we felt that it was appropriate to test this hypothesis in glomerular models of kidney injury leading to fibrosis, which less clearly involve hypoxia.

Potential mechanisms of HIF-1α and HIF-2α interaction with TGF-β in mediating cellular collagen expression are not well understood. Furthermore, differences between the two HIF-α isoforms in promoting fibrogenesis have not been elucidated. Here, we present data suggesting that both HIF-α isoforms contribute to human mesangial cell (HMC) collagen expression in response to TGF-β. As was the case for tubular epithelial cells (2), TGF-β stimulated expression of both HIF-α isoforms in HMC in culture via a mechanism unrelated to protein stabilization or gene transcription. In the present studies of HMC, activation of the collagen promoter by TGF-β was partly HIF dependent, with HIF-2α having a significantly greater impact than HIF-1α. Knockdown of either HIF-α isoform blocked TGF-β-stimulated collagen protein expression. In murine, adriamycin (ADR)-induced glomerulosclerosis, HIF-2α target genes, but not HIF-1α targets, were upregulated. Taken together, our data show that TGF-β upregulates HMC expression of HIF-1α and HIF-2α in normoxic culture, promoting mesangial cell collagen expression. Our findings also suggest that HIF-2α may be the more significant HIF isoform during glomerulosclerosis.

MATERIALS AND METHODS

Materials.

Except where otherwise indicated, reagents were purchased from commercial sources. Recombinant human TGF-β1 (R&D Systems, Minneapolis, MN) was reconstituted as a 4-μg/ml stock solution in 4 mM HCl with 1 mg/ml bovine serum albumin (BSA). Vehicle control for TGF-β1 contained equivalent amounts of HCl and BSA. Antibodies were purchased from the following vendors: anti-HIF-1α from BD Lifesciences (San Jose, CA); anti-HIF-2α from Novus Biologicals (Littleton, CO); anti-β-actin from Sigma (St. Louis, MO); and horseradish peroxidase-conjugated secondary antibodies for rabbit IgG from Promega (Madison, WI) and for mouse IgG from Santa Cruz Biotechnology (Santa Cruz, CA). The TβRI inhibitor SB431542 and PI3K inhibitor LY294002 were purchased from CalBiochem/EMD (La Jolla, CA).

Plasmid constructs.

The -378COL1A2-Luc construct containing the sequence 378 bp of the α2 (I) collagen promoter and 58 bp of the transcribed sequence fused to the luciferase reporter gene was constructed as previously described (19). The HRE-Luc construct (trimeric units of HRE sequences cloned into the pGL2 basic vector) was a gift from J.M. Leiden (6). ND-HIF mutants, where proline residues 402 and 564 in HIF-1α (plasmid #18955) or 405 and 531 in HIF-2α (plasmid # 18956) are substituted with alanines, were obtained through Addgene (http://www.addgene.org) (30). Control plasmid pcDNA3 vector was purchased from Invitrogen (Carlsbad, CA). siRNAs directed against HIF-1α or HIF-2α and scrambled control were purchased from Thermo Scientific/Dharmacon (Waltham, MA). Primers for quantitative PCR analyses were custom-synthesized by Integrated DNA Technologies (Coralville, CA), and the sequences are shown in Table 1.

Table 1.

Primer sequences for mRNA analysis by PCR

Primer	Forward Sequence	Reverse Sequence
HUMAN
COL1A2	5′-CAATGCTGCCCTTTCTGCTCCTTT-3′	5′-CACTTGGGTGTTTGAGCATTGCCT-3′
HIF-1α	5′-AGCCGAGGAAGAACTATGAAC-3′	5′-ATTTGATGGGTGAGGAATGGG-3′
HIF-2α	5′-ATTCACCAAGCTAAAGGAGGAGCC-3′	5′-ACTCCTCGAAGTTCTGATTCCCGA-3′
β2-Microglobulin	5′-TGTCTGGGTTTCATCCATCCGACA-3′	5′-TCACACGGCAGGCATACTCTT-3′
MOUSE
COL1A2	5′-CTGGAACAAATGGGCTCACTG-3′	5′-CAGGCTCACCAACAAGTCCTC-3′
BNIP3	5′-TTAAAGGGTGCGTGCGGGTTATCT-3′	5′-AAGGCGAGAATCCTCATCCTGCAA-3′
HMOX1	5′-TAGCCCACTCCCTGTGTTTCCTTT-3′	5′-TGCTGGTTTCAAAGTTCAGGCCAC-3′
DLL4	5′-TGCCACTTCGGTTACACAGTGAGA-3′	5′-TGGCAATCACACACTCGTTCCTCT-3′
Cyclin D1	5′-ATTGGTCTTTCATTGGGCAACGGG-3′	5′-GGCCAATTGGGTTGGGAAAGTCAA-3′
18S	5′-AGTTCCAGCACATTTTGCGAG-3′	5′-TCATCCTCCGTGAGTTCTCCA-3′

BNIP3, bcl2/adenovirus #1B interacting protein 3; HMOX1, heme oxygenase 1; DLL4, delta-like 4; HIF, hypoxia-inducible factor.

Cell culture.

HMCs were propagated from glomeruli isolated from normal margins of kidneys that were removed from patients under sterile surgical conditions, as described (26). The cells were grown in Dulbecco's modified eagle medium/F12 supplemented with 10% heat-inactivated Hyclone cosmic calf serum (CCS) obtained from Thermo Fisher Scientific, glutamine, penicillin-streptomycin, and HEPES buffer and maintained in normoxic condition (21% O₂, 5% CO₂) unless otherwise described.

Cell treatments.

Cells were changed to medium containing 1% CCS (1% CCS-HMC-M) 24 h before treatment. Cells were exposed to inhibitors, as described, for 30 min before the addition of TGF-β1 or vehicle. For hypoxic experiments, dishes were transferred to a hypoxia chamber (1.5% O₂, 5% CO₂, 93.5% N₂) for the indicated periods and lysed after the treatment within the chamber.

Preparation of cell lysates and Western blot analysis.

Cells were rinsed twice with ice-cold PBS lysed on ice in RIPA buffer (50 mM Tris·HCl, pH 7.5; 150 mM NaCl; 1% Nonidet P-40; 0.5% deoxycholate; 0.1% SDS) containing EDTA, protease, and phosphatase inhibitors (Sigma). Cleared cell lysates were subjected to SDS-PAGE (8 or 10% polyacrylamide gels), transferred onto PVDF membranes (Millipore, Bedford, MA), and immunoblotted with antibodies as indicated in the figure legends. Densitometric analysis was performed using the ImageJ 1.44 program for Windows.

Transfection.

For reporter assays, HMC were plated in triplicate in six-well plates, at 1.5 × 10⁵ cells per well. Eighteen to twenty-four hours later, cells were switched to 1% CCS-HMC-M and transfected with 0.5 μg each of the indicated constructs, or siRNA along with CMV-SPORT-β-galactosidase (Invitrogen) as a control for transfection efficiency, using the X-tremeGENE HP transfection reagent (Roche Applied Science, Indianapolis, IN) according to the manufacturer's instruction. After 3–24 h of transfection, 1 ng/ml of TGF-β1 or vehicle was added to the culture medium, and lysates were harvested after 24 h for measurement of luciferase and β-galactosidase activities with a commercial kit (Promega), as directed (19). For protein analysis, cells were plated at 50% confluence in 60-mm dishes, switched to 1% CCS HMC-M, and transfected with siRNAs for 48 h. Media were replaced with fresh 1% CCS HMC-M containing vehicle or 2 ng/ml TGF-β1. After 24 h, protein was harvested for Western blot analysis.

RNA isolation and analysis.

Total RNA was isolated from HMC cultures by using an RNA isolation kit from USB/Affymetrix (Santa Clara, CA). One microgram of the RNA was then reverse-transcribed with the iScript cDNA synthesis kit (Bio-Rad Laboratories, Hercules, CA), and mRNA expression was analyzed by real-time quantitative PCR (iCycler iQ real-time PCR detection system) with iQ SYBR Green Super mix (Bio-Rad). Primer sequences were provided in Table 1. Each reaction was run in triplicate and fold changes over the control values were calculated using the ΔΔCt method.

Animal models.

Animal protocols were reviewed and approved by the Northwestern University Institutional Animal Care and Use Committee. Glomerulosclerosis was induced, as previously described (8), by a single tail vein injection of ADR (15 mg/kg; Research Product International, Mt Prospect, IL) in 8-wk-old male 129x1/Svj mice (Jackson Laboratory, Bar Harbor, ME). Some mice also received pimonidazole hydrochloride (60 mg/kg; Hypoxyprobe, Burlington, MA) intraperitoneally 1 h before death and adducted probe was immunohistochemically detected as a marker for local hypoxia. A portion of the harvested kidneys was fixed in 10% neutral buffered formalin for 24 h and further processed by the Northwestern University Mouse Histology and Phenotyping Laboratory. Glomeruli were isolated by laser-capture microdissection (LCM) from cryosectioned kidney tissue preserved in −80°C embedded in OCT compound (Torrance, CA) until being processed (7). Approximately 200 glomerular sections per mouse were collected with the PALM laser catapulting microdissection system (Carl Zeiss MicroImaging GmbH, Munich, Germany) at the Cell Imaging Facility of the Northwestern University, and total RNA was extracted with RNeasy micro kit (Qiagen, Valencia, CA). HIF-1/2α target gene expression in the glomeruli was then analyzed by real-time quantitative PCR (primer sequences were provided in Table 1).

Statistics.

All data are expressed as means ± SE. Statistical analyses were performed using GraphPad Prism version 4.0 for Macintosh (GraphPad Software, San Diego, CA). Student's t-test or ANOVA followed by Fisher's post hoc analysis was used to evaluate differences between groups. P < 0.05 was considered significant.

RESULTS

TGF-β1 increases HIF-1α and HIF-2α expression in HMC in normoxia.

First, we measured HIF-1α and HIF-2α protein expression in response to TGF-β1 (1.0 ng/ml) in normoxic conditions (21% O₂, 5% CO₂). As anticipated from our previous findings in HKC, HIF-1α expression was significantly increased compared with the control in HMC after 6-h treatment with TGF-β1 (Fig. 1A). We observed a similar increase in HIF-2α expression. TGF-β-stimulated HIF-α expression was maximal at 4–6 h for HIF-1α (Fig. 1B) and at 4–8 h for HIF-2α (Fig. 1C).

Fig. 1.

Transforming growth factor (TGF)-β1 stimulation increases normoxic hypoxia-inducible factor (HIF)-1α and HIF-2α protein expression. A: expression of HIF-α isoforms. Human mesangial cells (HMC) were treated with TGF-β1 (1 ng/ml) or vehicle control for 6 h in normoxia (21% O₂, 5% CO₂). Whole cell lysates were assessed for HIF-1α or HIF-2α protein expression by Western blot. Representative blots (top) and graphs with densitometric analyses representing summary of repeated experiments and normalized to β-actin controls (bottom) are shown. TGF-β1 significantly increased both HIF-1α and HIF-2α protein expression in normoxia after 6 h of treatment by 3.6- and 3.2-fold, respectively. *P < 0.01 for HIF-1α, n = 6; *P < 0.01 for HIF-2α, n = 4, compared with the control. B: time course of HIF-1α protein expression after TGF-β1 (1 ng/ml) treatment in HMC. Representative blot of 3 separate experiments is shown. C: time course of HIF-2α protein expression after TGF-β1 treatment in HMC. Representative blot from 1 of 3 separate experiments is shown.

TGF-β does not alter HIF-1α or HIF-2α protein degradation or mRNA expression.

To study the mechanism by which TGF-β increases HIF-α expression, we next examined the effect of TGF-β1 on HIF-1α and HIF-2α protein degradation. HMC were subjected to hypoxia to maximize HIF expression and then switched to normoxia and treated with cycloheximide to prevent new protein synthesis. Protein disappearance curves generated in the presence or absence of TGF-β1 (Fig. 2A) were parallel, indicating that, unlike hypoxia, TGF-β does not increase stability of the HIF-1α or HIF-2α protein. TGF-β1 also did not increase HIF-1α or HIF-2α mRNA levels (Fig. 2B). These results suggest that TGF-β-mediated HIF-α protein upregulation does not require transcriptional activation or protein stabilization, but it likely involves translational regulation.

Fig. 2.

TGF-β1 does not affect HIF-α protein stability or mRNA expression in normoxia. A: disappearance curves for HIF-1α and HIF-2α. HMC were treated with vehicle or 1 ng/ml TGF-β1 for 6 h in hypoxia (1.5% O₂) before 5-min pretreatment with 10 μg/ml cycloheximide (CHX). At time 0, cells were transferred to normoxia (21% O₂) and lysates were harvested at the indicated time points for evaluation of HIF-1α or HIF-2α expression by Western blot. Representative blots (top) and densitometric analyses normalized to β-actin controls from 3 separate experiments (bottom) are shown in the graph. B: expression of HIF isoform mRNA. HMC were stimulated with vehicle or TGF-β1 for 6 h in normoxia, and COL1A1 and HIF-1α or HIF-2α mRNA expression were assessed by qPCR. Means from 3 independent experiments are shown as graphs. TGF-β1 significantly increased COL1A1 expression (*P < 0.05) without significantly affecting HIF-1α or HIF-2α expression.

TGF-β1-induced increase in HIF-α expression requires TβRI kinase activity and PI3-kinase activity.

We then examined whether TβRI kinase (ALK5) activity and/or PI3-kinase activity mediated the normoxic increase in HIF-1α or HIF-2α expression induced by TGF-β1. TGF-β1-stimulated HIF-1α protein expression was prevented by an ALK5 kinase inhibitor, SB431542 (SB4; Fig. 3A), indicating that TGF-β1-enhanced HIF-1α expression requires TβRI kinase activity. On the other hand, a PI3-kinase inhibitor, LY294002 (LY), markedly reduced both basal and TGF-β-stimulated HIF-1α protein expression. TGF-β significantly increased relative HIF-1α expression even in the presence of LY, although absolute expression of HIF-1α was significantly lower compared with the steady-state control. These data suggest that, while PI3-kinase activity promotes normoxic HIF-1α expression, as has been suggested (9), TGF-β could enhance HIF-1α expression by a mechanism independent of PI3-kinase. Similar effects of the ALK5- and PI3-kinase inhibitors were also observed on TGF-β stimulation of HIF-2α expression under normoxia (Fig. 3B). SB4 completely blocked Smad3 phosphorylation, and LY completely blocked AKT phosphorylation (Fig. 3C), confirming the efficacy of either inhibitor. Thus, these data suggest that both the TβRI/ALK5- and PI3-kinase activities are essential for TGF-β1-stimulated, increased HIF-α expression in normoxia.

Fig. 3.

HIF-α expression in normoxia is type I TGF-β receptor (TβRI)- and PI3-kinase-dependent. HMC were treated with DMSO, the TβRI receptor kinase inhibitor SB431542 (SB4; 5 mM), or PI3-kinase inhibitor LY-294002 (LY; 20 μM) for 30 min before 6-h TGF-β1 treatment (1 ng/ml). Whole cell lysate was analyzed by Western blot for HIF-1α (A) or HIF-2α (B). Representative blots for HIF-1α and HIF-2α, along with a β-actin expression (top) and graph of densitometric analyses (bottom), corrected for β-actin loading controls from 3 separate experiments are shown. TGF-β1 significantly increased HIF-1α and HIF-2α expression by 2.9- and 2.8-fold, respectively (*P < 0.05), compared with the respective control. The TβRI inhibitor abrogated HIF-1α and -2α induction by TGF-β1. While the PI3-kinase inhibitor significantly reduced the basal expression of HIF-α (P < 0.001 for either HIF-1α or HIF-2α), TGF-β still significantly increased HIF-1α expression (*P < 0.05, control vs. TGF-β within the LY-treated group). By 2-way ANOVA, the inhibitors significantly affected TGF-β1 induction of HIF-1α or -2α expression (P < 0.01 and P < 0.01, n = 3, respectively). C: same lysates used in the HIF-α analyses in A and B were probed for COOH-terminal phospho (P)-Smad3, Smad3, phospho (P)-AKT, and AKT. SB4 to verify efficacy of inhibitors.

HIF-α is required for TGF-β1-stimulated collagen expression in HMC.

We previously found that HIF-1α participates in TGF-β1-stimulated collagen expression in a human proximal tubular cell line. To determine whether HIFs have a similar role in glomerular cells, we investigated the influence of HIF-α on TGF-β1-stimulated collagen expression by HMC in normoxia, and examined whether there was a difference in effects between the two HIF-α isoforms. Constructs encoding siRNA for HIF-1α or HIF-2α, or a control siRNA, were each cotransfected with a COL1A2 promoter-luciferase reporter construct into HMC, and the cells were treated with TGF-β1 (1 ng/ml) or vehicle. HIF-1α siRNA reduced basal and TGF-β-stimulated promoter activity by about two-thirds (Fig. 4A); similar results were obtained in HMC transfected with siRNA HIF-2α. To examine the effect of increasing HIF-1α or HIF-2α expression levels on COL1A2 gene activation, we next transfected constructs that express either ND-HIF-1α or ND-HIF-2α along with the COL1A2-luc construct. In the ND constructs, prolines that are hydroxylated by PHD in normoxia, and thus target HIF-α for degradation by the von Hippel-Lindau E3 ligase, are replaced with alanine that cannot be hydroxylated (29). While overexpression of ND-HIF-1α or ND-HIF-2α enhanced the HRE minimal promoter activities to a similar degree (Fig. 4C), TGF-β stimulation of the COL1A2-promoter activity was enhanced significantly more by the expression of ND-HIF-2α compared with that observed with ND-HIF-1α overexpression (P < 0.05, ND-HIF-1α vs. ND-HIF-2 by post hoc Fisher's test; Fig. 4B). To examine their effects on collagen I protein, control, HIF-1α or HIF-2α siRNAs were transfected into HMC and the cells were then treated with 2 ng/ml TGF-β1 for 24 h before the cell layer was harvested for protein analysis (Fig. 5A). TGF-β1 significantly increased collagen I protein expression (1.36x, P < 0.05) in control siRNA-treated HMC (Fig. 5B). Transfection with HIF-1α or HIF-2α siRNA blocked TGF-β1-stimulated collagen I protein increase over vehicle-treated controls, with siRNA to HIF-2α slightly decreasing both basal and stimulated collagen I compared with vehicle-treated control siRNA (Fig. 5B). These data suggest that, in HMC cultured in normoxic conditions, HIF-2α plays the major role in enhancing TGF-β induction of collagen expression.

Fig. 4.

HIF-1α and HIF-2α enhance collagen promoter activity. A: effect of HIF isoform knockdown. HMC were transfected with siRNA to HIF-1α or HIF-2α, or scrambled control, along with a COL1A2 promoter-luciferase reporter construct and then treated with TGF-β1 (1 ng/ml) or vehicle. Luciferase assay results, corrected for transfection efficiency using β-galactosidase expression, are shown as means ± SE of fold change from the control siRNA from 3 separate experiments performed in triplicate. TGF-β1 increased the reporter activity 3.3-fold in cells transfected with the scrambled siRNA (*P < 0.05, compared with the control treated with vehicle to TGF-β). siRNA to HIF-1α and -2α significantly affected TGF-β induction of the COL1A2 promoter activity (P < 0.01, 2-way ANOVA). The inhibitory effects of siRNAs to HIF-1α or HIF-2α were equally significant (†P < 0.05 and P < 0.05, by post hoc Fisher's test, comparing with the scrambled siRNA group, respectively). B: effect of overexpressing HIF. HMC were transfected with pcDNA3 empty vector or nondegradable mutants of HIF-1α (ND-HIF-1α) or HIF-2α (ND-HIF-2α) where proline PHD target sites were mutated to alanines, along with the COL1A2 promoter-luciferase reporter construct. TGF-β1 increased the reporter activity 2.9-fold in cells transfected with the pcDNA3 (*P < 0.05, compared with the vehicle-treated control). Promoter activity was significantly induced by TGF-β in the presence of control vector or by ND-HIF-1α or -2α overexpression (*P < 0.05, control vs. TGF-β within the same group). Both ND-HIF-1α and ND-HIF-2α significantly increased TGF-β1-stimulated COL1A2 promoter activity (P < 0.05, 2-way ANOVA). The effect of ND-HIF-2α on COL1A2 promoter activity was significantly greater than that of ND-HIF-1α (†P < 0.05), by post hoc Fisher's test. C: effect of HIF overexpression on HRE-luc activity. Overexpression of ND-HIF-1α or HIF-2α significantly enhanced the HIF-responsive element (HRE) minimal promoter activities and there was no significant difference between the 2 isoforms.

Fig. 5.

HIF-1α and HIF-2α enhance collagen protein expression. HMC were seeded in 60-mm dishes at ∼50% confluence and transfected with control (scrambled) or siRNA to HIF-1α or HIF-2α for 48 h before media change and another 24-h incubation with vehicle or 2 ng/ml TGF-β1. Protein was harvested, subjected to SDS-PAGE, and transferred before Western analysis. A: representative blots probed for collagen I, HIF-1α, and HIF-2α with β-actin used as a loading control. B: densitometric analysis of collagen I and β-actin blots from 3 separate experiments (*P < 0.05). Transfection with siRNA to either HIF-α isoform blocked TGF-β1-stimulated increases in collagen I expression.

HIF-2 target genes are upregulated in ADR-induced murine glomerulosclerosis.

ADR induces proteinuria and kidney fibrosis by 2 wk after the ADR injection in BALB/c (17) or 129x1/Svj (8) strain mice. In a recent report, we showed that the fibrotic changes in this model are mediated by TGF-β (8). Hypoxyprobe staining shows that glomeruli are relatively spared from hypoxic changes despite substantial hypoxia detected in the tubulointerstitium in this model (Fig. 6A). To determine the potential in vivo relevance of our in vitro findings, we evaluated HIF-target gene expressions that are known to be relatively specific for either the HIF-1 or the HIF-2 isoform (13), in glomeruli isolated from the ADR mouse kidneys. In isolated glomeruli, COL1A2 mRNA expression was significantly elevated in ADR-treated mice (Fig. 6B). Expression of HIF-1 target genes (heme oxygenase 1 and Bcl2/adenovirus #1B interacting protein 3) was not significantly different between the control- and ADR-treated mouse glomeruli, whereas HIF-2 target gene delta-like 4 was significantly upregulated in ADR-treated glomeruli (P < 0.05 compared with the control). Expression of another HIF-2 target gene, cyclin D1, also appeared to increase (Fig. 6C).

Fig. 6.

HIF-2α target gene, but not HIF-1α target gene, expression is upregulated in glomeruli in adriamycin (ADR)-induced glomerulosclerosis. A: detection of hypoxia in ADR-induced glomerulosclerosis in 129x1/Svj mice. Glomerulosclerosis was induced by a single tail-vein injection of ADR (15 mg/kg) in 8-wk-old male mice. We recently reported that proteinuria and kidney fibrosis establish by 2 wk after the ADR administration (11). Hypoxyprobe (60 mg/kg) was given intraperitoneally 30 min before death at day 14, and the adducted products were detected immunohistochemically in formalin-fixed, paraffin-embedded kidney sections. DAB detection (brown) of the hypoxyprobe products demonstrates marked tubulointerstitial hypoxia, whereas the cortex and glomeruli are relatively spared. Nuclei were counterstained with hematoxylin. B and C: glomerular COL1A2 (B) and HIF-target gene (C) expression in ADR nephropathy. Glomeruli were isolated by laser-capture microdissection of sections from cryopreserved mouse kidneys and mRNA expression of COL1A2 (B) and HIF target genes (C) were analyzed by real-time quantitative PCR. Relative expression of the gene of interest was analyzed using 18S expression as a reference. Numbers of mice evaluated in each group are shown in parenthesis. *P < 0.05, compared with the control. HMOX1, heme oxygenase 1; BNIP3, Bcl2/adenovirus #1B interacting protein 3; DLL4, delta-like 4.

DISCUSSION

TGF-β is a potent, well-characterized mediator of tissue fibrosis, including renal glomerulosclerosis. However, progress in treating fibrosis by modulating TGF-β signaling has been slow, suggesting that important pathways remain uncharacterized. Roles for HIF in tubulointerstitial disease models have been suggested (11, 15), but the mechanisms involved are not well-understood. We previously demonstrated that HIF-1α promotes fibrogenesis in cultured kidney proximal tubular epithelial cells even under normoxic conditions (2). Seeking models that are more specific to glomeruli and thus may be less dependent on hypoxia for progression, we chose here to determine the potential interaction between TGF-β and HIF signaling in glomerular mesangial cells and in a mouse model of glomerulosclerosis. TGF-β1 increases both HIF-1α and HIF-2α expression in HMC under normoxic conditions in a TGF-β receptor-dependent manner. While basal expression of HIF-α in normoxia largely depends on PI3-kinase activity, TGF-β can stimulate HIF-α expression independent of PI3-kinase activity. We also found that the TGF-β1-dependent mechanism is distinct from that of hypoxia-dependent HIF stabilization.

Our present studies found an important role for PI3-kinase as well as TβR/Smad3 signaling in HIF-α expression. Since PI3-kinase signals through mTOR to promote protein translation, and we observed no effect of TGF-β on HIF-α mRNA expression or protein stability, our results are consistent with the notion that TGF-β increases HIF expression by enhancing translation of HIF-1α or -2α mRNA into protein, as has been described for enhanced HIF activity in response to other growth factors (1, 20). For example, insulin augments HIF-1α protein expression by enhancing its translation in an mTOR-dependent manner (29), and angiotensin II enhances translation of HIF-1α in a reactive oxygen species-dependent manner (18, 21). We recently found that both TGF-β/Smad3 and phosphatidylinositol-3-kinase promote the translation of HIF-1α in an mTOR-dependent manner (23).

Although both HIF-1α and HIF-2α subunits share similar domain structure, heterodimerize with HIF-β, and bind to the same DNA sequence, the HRE, their effects on the expression of some genes may vary because each HIF-α isoform binds a different spectrum of transcriptional cofactors (12). Since potential effects of these differences in promoting renal fibrogenesis have not been elucidated, it was important to determine how inhibiting or overexpressing each specific isoform affects TGF-β-stimulated collagen expression and subsequent glomerulosclerosis. Silencing each HIF-α isoform expression equally decreased TGF-β1-induced COL1A2-Luc activation. On the other hand, overexpression of HIF-2α showed a significantly greater effect than that of HIF-1α on collagen promoter activity, despite similar enhancement of the HRE-luc response by the two isoforms. Because it is difficult to stain HIF-α protein in tissue sections, we examined glomerular expression of HIF target genes by LCM of frozen tissue from a mouse model of glomerulosclerosis. These studies showed an increase in glomerular mRNA for targets of HIF-2α, but not HIF-1α. Differences in their actions on collagen expression could play a significant role in fibrogenesis.

Our hypothesis that HIF-2 is more important for collagen expression is supported by the suggestion of others that HIF-1 is the likely mediator of acute, hypoxia-based HIF responses, whereas HIF-2 is more associated with chronic responses (16). It is also consistent with observations indicating that HIF-1α may play a beneficial role in the renal response to acute kidney injury (3).

The present results offer an important distinction from previous studies by defining a potential role for HIFs in glomerular, rather than tubulointerstitial, disease. Our findings thus support the hypothesis that the HIF and TGF-β signaling pathways interact to promote fibrogenesis even in normoxia, and raise the possibility that HIFs play a role independent of the development of chronic renal hypoxia/ischemia in chronic kidney disease, at least for type I collagen expression. Our finding of HIF expression and actions beyond those limited to hypoxia could have broad implications for renal and other organ fibrogenesis and may provide a therapeutic target in a broad spectrum of fibrotic kidney disease. The understanding of the biology of both isoforms will be important for better diagnosis of pathological conditions, and for the development of new drugs acting selectively on either isoform.

Taken together, our data demonstrate the cooperation between TGF-β and HIF-1/2α and suggest a new approach in which HIFs are necessary for normoxic, TGF-β-stimulated glomerulosclerosis, with HIF-2α playing the significant role in normoxic HMC collagen expression. Further studies are needed to determine the exact mechanism by which TGF-β and HIFs interact and the implications of these interactions for gene transcription.

GRANTS

This study was supported in part by National Institute of Diabetes and Digestive and Kidney Diseases Grants R01-DK049362 and R01-DK075663 to H. W. Schnaper and National Heart, Lung, and Blood Institute Grant R01-HL35440 to P. T. Schumacker.

The Cell Imaging Facility and Mouse Histology and Phenotyping Laboratory of the Northwestern University are supported by NCI CCSG P30 CA060553 awarded to the Robert H. Lurie Comprehensive Cancer Center.

Present address of C. Hanna: Univ. of Minnesota Amplatz Children's Hospital, East Bldg., MB684, 2450 Riverside Ave., Minneapolis, MN 55454.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: C.H., S.H., X.L., and T.H. performed experiments; C.H., S.H., and T.H. analyzed data; C.H., S.H., and T.H. prepared figures; C.H. drafted manuscript; S.H., P.T.S., T.H., and H.W.S. conception and design of research; S.H., B.R.-Z., T.H., and H.W.S. interpreted results of experiments; S.H., X.L., B.R.-Z., P.T.S., T.H., and H.W.S. edited and revised manuscript; S.H., X.L., B.R.-Z., P.T.S., T.H., and H.W.S. approved final version of manuscript.