Introduction

Peroxisome proliferator-activated receptor γ (PPARγ) is a member of the nuclear hormone receptor superfamily. Activation of the PPARγ pathway requires heterodimerisation of ligand-bound PPARγ with the retinoid X receptor, which binds specific peroxisome proliferator response elements in the promoter regions of target genes [1]. Despite extensive literature on the beneficial effects of synthetic PPARγ agonists in limiting insulin resistance and protecting pancreatic beta cell function, and our own data suggesting a protective effect on nephropathy [2–5], salt and water retention pose a key limitation to the uptake of these agents in clinical practice.

The proximal tubule mediates 50–75% of total tubular sodium and water reabsorption, with the sodium–hydrogen exchanger-3 (NHE3) and aquaprorin-1 (AQP1) being the key transporter and channel, respectively, mediating transcellular salt and water reabsorption from the tubular ultrafiltrate. We have demonstrated that PPARγ is produced in human proximal tubular cells (PTCs; HK2 cells) and is activated by exposure to high glucose [6], as well as by the insulin-sensitising agents, thiazolidinediones (TZDs) [7]. We have also demonstrated that the serum/glucocorticoid kinase receptor gene, SGK1, is a target gene of PPARγ and a mediator for PPARγ agonist-induced NHE3 and AQP1 upregulation in human PTCs [8].

High glucose has been shown to transactivate the EGF receptor (EGFR), and the activation of EGFR through tyrosine phosphorylation occurs in the absence of extracellular ligands in response to PPARγ agonists [9–11]. We have demonstrated that high glucose regulates NHE3 through an EGFR-mediated pathway [11]. This suggests that excessive sodium reabsorption in response to PPARγ activation may occur through regulated downstream activation of EGFR in the proximal tubule. The role of the EGFR in PPARγ-mediated NHE3 and AQP1 production in human PTCs is not known and will be addressed in this study.

Methods

Primary culture of human PTCs

Segments of macroscopically and histologically normal renal cortex were obtained under aseptic conditions from patients undergoing nephrectomy for small (<0.06 m) tumours. Patients were accepted for inclusion in the study if there was no history of renal or systemic disease known to be associated with tubulointerstitial pathology. Written informed consent was obtained from each patient before surgery, and ethics approval for the study was obtained from the Royal North Shore Hospital Human Research Ethics Committee. The methods for primary culture of human PTCs are described in detail elsewhere [12]. The ultrastructure, growth and immunohistochemistry of these cells have been well characterised in our laboratory and shown to reproducibly reflect the biology and physiology of their in vivo counterparts [12–14].

Cell culture

Human PTCs were grown in 0.1 m tissue culture dishes (Becton Dickinson, North Ryde, NSW, Australia) and exposed to 5 mmol/l d-glucose (control medium), 25 mmol/l d-glucose, 5 mmol/l d-glucose + 20 mmol/l l-glucose (osmotic control) or 10 μmol/l pioglitazone in 5 mmol/l d-glucose for 24 h. l- and d-glucose (ICN Biomedical, Irvine, CA, USA) and pioglitazone (Cayman Chemical, Ann Arbor, MI, USA) were used to determine the specific effects of PPARγ activation. Initial ‘dose–response’ experiments were undertaken to determine the concentration at which pioglitazone maximally stimulated PPARγ protein expression. On the basis of these studies, we used 10 μmol/l pioglitazone in the experimental protocols as previously described [4, 6, 8]. As pioglitazone is dissolved in DMSO, 0.13% DMSO was used in the control medium for all the pioglitazone-related experiments. The more selective PPARγ agonist, L-805645 (Merck Biosciences, Darmstadt, Germany), was used in some experiments to determine the specific effects of PPARγ activation [8].

To determine the role of EGFR in mediating the observed changes, the EGFR tyrosine kinase inhibitor of the pyrrolopyrimidine class, PKI166 (4 μmol/l), was used as previously described [11], in cells exposed to either high glucose or pioglitazone. PKI166 was synthesised in the laboratories of NIBR (Novartis Institutes for Biomedical Research, North Ryde, NSW, Australia) and was provided by NIBR for the purpose of this study. PKI166 is considered highly specific, with the inhibition constant (IC50) for EGFR being 0.7 nmol/l [15].

Western blotting

Western blots were performed on Triton X-100-soluble fractions. PPARγ-specific antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA), AQP1 and NHE3 antibodies (Chemicon International, Temecula, CA, USA), total EGFR (t-EGFR) antibody (Cell Signaling, Danvers, MA, USA), phospho-EGFR (P-EGFR) antibody (pY1068; Invitrogen, Carlsbad, CA, USA) or actin antibody (Sigma-Aldrich, St Louis, MO, USA) were used overnight followed by incubation with anti-rabbit or anti-mouse antibody (Amersham Pharmaceuticals, Rydalmere, NSW, Australia) for 1 h at room temperature. Data were normalised to actin to ensure equal protein loading. The bands corresponding to AQP1 (28 kDa), PPARγ (67 kDa), NHE3 (85 kDa), t-EGFR or P-EGFR (170 kDa) and actin (42 kDa) were quantified using NIH Image software version 1.60.

Real-time RT-PCR

RNA was extracted using an RNeasy Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s instructions. cDNA was generated by reverse transcribing 1 μg total RNA in a reaction volume of 20 μl using VILO cDNA Synthesis Kits (Invitrogen). Sequence-specific primers for human EGFR and β-actin are described in electronic supplementary material (ESM) Table 1. Primer specificity in real-time PCRs was confirmed using RT-PCR. A 25 μl real-time PCR included Brilliant SYBR Green QRT–PCR Master Mix according to the manufacturer’s instructions (Stratagene, La Jolla, CA, USA). Quantitative real-time PCR was performed using an ABI Prism 7,900 HT Sequence Detection System (Applied Biosystems, Mulgrave, VIC, Australia). Reactions were performed in at least triplicate and analysed by relative quantification using RQ Manager software, version 1.2 (Applied Biosystems). All data are presented as fold change compared with control after normalisation to the housekeeping gene, β-actin.

PPARγ silencing

Small interfering RNA (siRNA) was designed to specifically target PPARγ mRNA sites. PPARγ siRNA target is as follows: GCCTCATGAAGAGCCTTCCAACTCCCTCA (Ambion, Mulgrave, VIC, Australia). PTCs were transfected with 8 nmol/l PPARγ siRNA using Ribojuice (Novagen, Madison, WI, USA) as per the manufacturer’s instructions. Protein levels were determined using western blot. All siRNA experiments included a non-specific control (NSC siRNAs; Ambion). Once PPARγ knockdown was confirmed, subsequent experiments were conducted to determine the effect of exposure to high glucose (25 mmol/l) on AQP1, NHE3 and P-EGFR levels 24 h after PPARγ knockdown. Cell lysates were then collected. Western blotting experiments were performed as described above.

In vivo experiments

To determine P-EGFR levels in an in vivo model of diabetes mellitus, we used the streptozotocin-induced hypertensive Ren-2 transgenic rat (mRen2)27 model. These animals were confirmed to have demonstrable hyperfiltration, proteinuria and pathological signs of diabetic nephropathy at the time of study. This model has been previously validated for studies in diabetic nephropathy [16–20]. Briefly, 6-week-old female, heterozygous Ren-2 rats, weighing 0.125 ± 5 kg, were randomised to receive either 55 mg/kg streptozotocin (STZ; Sigma) (diabetic) or buffer alone (non-diabetic) by tail vein injection after an overnight fast. Diabetic Ren-2 rats were subsequently treated with 5 mg/kg pioglitazone (Cayman Chemical) fed by gavage or sham treatment for 12 weeks starting 7 days after STZ injection. Experimental procedures adhered to the guidelines of the National Health and Medical Research Council of Australia’s Code for the care and use of animals for scientific purposes and have been approved by St Vincent’s Hospital Animal Ethics Committee, Melbourne, Australia. Serial sections of 4 μmol/l thickness were fixed in 4% paraformaldehyde, and paraffin sections were prepared for immunohistochemistry. Levels of P-EGFR, PPARγ, NHE3 and AQP1 protein were determined.

Immunohistochemistry

Paraffin-embedded sections were dewaxed in xylene and rehydrated in graded concentrations of ethanol. Endogenous peroxidase activity was blocked by immersion in 3% H2O2 for 5 min. Epitope retrieval was required for P-EGFR, AQP1 and NHE3, but not PPARγ, staining. Heat retrieval was performed in 0.01 mol/l citrate buffer, pH 6.0, using a pressure cooker set at 121°C for 30 s. Non-specific protein binding was blocked with a protein block (Dako, Troy, MI, USA). Incubation with primary antibodies was performed using a Sequenza vertical cover-plate immunostaining system (Thermofisher Scientific, Scoresby, VIC, Australia) using 0.2 μg P-EGFR (Tyr1173) antibody, 4 μg PPARγ antibody (both from Santa Cruz), 0.2 μg NHE3 antibody (Novus Biologicals, Littleton, CO, USA) or 4 μg AQP1 antibody (Santa Cruz) overnight at 4°C. The primary antibodies were subsequently localised using biotinylated secondary anti-mouse or anti-rabbit IgG antibodies. Horseradish peroxidase-conjugated streptavidin was subsequently used to visualise the tissue immune complexes using the LSAB+detection system (Dako). Antigen–antibody reactions were visualised with the chromogen, diaminobenzidine, and counterstaining was performed using Mayer’s Haematoxylin followed by Scott’s Blue staining (Fronine, Taren Point, NSW, Australia). Control sections were also prepared in which the authentic primary antibodies were replaced with an irrelevant isotype-matched IgG. The tissue specimens were examined by bright field microscopy using a Leica photomicroscope linked to a DFC 480 digital camera.

Quantification of histological variables

In brief, 10 random non-overlapping fields from three stained sections were captured and digitised using an AxioImager.A1 microscope (Carl Zeiss AxioVision, Kirchdorf, Germany) attached to an AxioCam MRc5 digital camera (Carl Zeiss AxioVision). Areas of brown staining reflecting P-EGFR, PPARγ, AQP1 or NHE3 production were quantified and calculation of the proportional area stained brown was then determined using image analysis software AIS (Analytic Imaging Station version 6.0, Imaging Research, St Catherines, ON, Canada).

Chromatin immunoprecipitation assay

A chromatin immunoprecipitation (ChIP) assay was performed using an EZ ChIP Kit (Millipore, Billerica, MA, USA) according to the manufacturer’s instructions. In brief, cells were crosslinked in 1% formaldehyde after different cell culture treatments. Cells were then lysed and sonicated. Sonication was optimised to achieve 200 to 1000 bp DNA fragments. Equal amounts of protein were incubated with 4 μg/ml PPARγ antibody (Santa Cruz Biotechnology). The crosslink of immunoprecipitated protein–DNA samples was then reversed, and DNA samples were purified using spin columns. PPARγ-binding sites of the EGFR promoter were quantified using real-time PCR. Primers used for the promoter region are described in ESM Table 2.

EGFR promoter activity assay

The promoter activity of EGFR was determined by the Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA) as we have described previously [21]. In brief, the DNA fragment (1,900 bp) of the EGFR promoter was amplified using PfuUltra II HS Fusion DNA Polymerase (Agilent Technologies, Santa Clara, CA, USA). The DNA fragment of the EGFR promoter containing the PPARγ-binding site (−1271: AGGTCCTAGTGAA) was subcloned into pGL3 firefly luciferase vector (Promega). The plasmid, pGL3-EGFR promoter, or the pGL3 basal plasmid was introduced into PTCs using Lipofectamine 2000 (Invitrogen). pRL-SV40 Renilla vector (Promega) was cotransfected into cells, and its luciferase activity was used for normalisation of transfection efficiency. Cells were then exposed to pioglitazone or L-805645 for 24 h, and luciferase activity was detected by POLARstar (BMG Labtech, Offenburg, Germany).

Statistical analysis

Results are expressed as a percentage or fold change compared with the control. Experiments were performed in at least three different culture preparations derived from each patient, and at least three data points for each experimental condition were measured in each preparation. Six animals in each group were used in the in vivo study. Results are expressed as mean ± SEM, with n reflecting the number of culture preparations. Statistical comparisons between groups were made by ANOVA, with pairwise multiple comparisons made by Fisher’s protected least significant difference test. Analyses were performed using the software package, Statview version 4.5 (Abacus Concepts, Piscataway, NJ, USA). p values ≤0.05 were considered significant.

Results

High glucose and pioglitazone induce PPARγ

We have previously demonstrated that high glucose and pioglitazone increase PPARγ protein expression in human PTCs (HK2 cells) [6]. We have confirmed that PPARγ protein content was significantly increased to 199 ± 21.6% in human primary PTCs when exposed to 25 mmol/l glucose (HG) (p < 0.05). Exposure to the osmotic control (OC) had no effect on PPARγ levels (109 ± 23.4%; p = 0.7) (Fig. 1a). Exposure to 10 μmol/l pioglitazone also increased PPARγ protein levels to 142 ± 6.2% (p < 0.01) (Fig. 1b).

Fig. 1
figure 1

High glucose and pioglitazone induce PPARγ protein expression. PTCs were incubated for 24 h with 5 mmol/l glucose medium (Control), 25 mmol/l glucose (HG), 5 mmol/l glucose medium with 20 mmol/l glucose (osmotic control; OC) or 10 μmol/l pioglitazone (Piog) in control medium. Representative western blotting images are shown for PPARγ and actin bands after high-glucose (a) or pioglitazone (b) exposure. Normalised results are expressed as mean ± SEM (n = 3). *p < 0.05 and **p < 0.01 vs control

Full size image

Combined effect of high glucose and pioglitazone on EGFR

High glucose increased t-EGFR levels to 414 ± 150% of control, most of which (360 ± 2%) was phosphorylated (p < 0.05 vs control) (Fig. 2a, b). P-EGFR is produced at low levels in human PTCs, as we have previously described [11] and as is also shown in Fig. 5. The combination of high glucose and pioglitazone increased t-EGFR protein content to 1102 ± 329% (p < 0.01), in addition to increasing EGFR phosphorylation to 575 ± 89% (p < 0.001) (Fig. 2a, b). High glucose in the absence and presence of pioglitazone increased EGFR mRNA levels to 1.3 ± 0.03-fold (p < 0.001) and 1.7 ± 0.1-fold (p < 0.01), respectively (Fig. 2c).

Fig. 2
figure 2

Effect of high glucose (HG) in the presence and absence of pioglitazone on EGFR. PTCs were incubated for 24 h with 5 mmol/l glucose medium (Control) in the presence or absence of pioglitazone (Piog), and levels of P-EGFR were quantified. Representative western blotting images for t-EGFR (a) and P-EGFR (b) are shown. EGFR mRNA expression is shown in (c). Normalised results are expressed as mean ± SEM (n = 3). *p < 0.05, **p < 0.01 and ***p < 0.001 vs control and † p < 0.05 vs HG

Full size image

Pioglitazone increases NHE3 and AQP1 through an EGFR pathway

We have confirmed that NHE3 is significantly increased when PTCs are exposed to pioglitazone, as we have previously demonstrated [8], to 167 ± 16.1% (p < 0.01). To determine the role of EGFR in increased NHE3 production, we exposed cells to the EGFR tyrosine kinase inhibitor, PKI166, in the presence of pioglitazone, and NHE3 protein content was assessed. The combination of PKI166 with pioglitazone abrogated the observed increase in NHE3 levels with the PPARγ agonist, pioglitazone, to 103 ± 6.1% (Fig. 3a). These results suggest that the upregulation of NHE3 after exposure to pioglitazone is EGFR mediated.

Fig. 3
figure 3

PPARγ agonist-induced increase in NHE3 and AQP1 is P-EGFR mediated. PTCs were incubated for 24 h with 5 mmol/l glucose medium (Control) or pioglitazone (Piog) with or without PKI166 (4 μmol/l), and levels of NHE3 (a) or AQP1 (b) were determined by western blotting. Representative images for NHE3, AQP1 and actin bands are shown. Normalised results are expressed as mean ± SEM (n = 4). **p < 0.01 vs control

Full size image

As we have previously demonstrated [8], pioglitazone significantly increased AQP1 protein production to 201 ± 23% of control values (p < 0.01). The combination of PKI166 with pioglitazone abrogated the observed increase in AQP1 levels with the PPARγ agonist, pioglitazone, to 95 ± 21% (p = 0.7 vs control) (Fig. 3b). These results suggest that the upregulation of AQP1 after exposure to pioglitazone is also mediated through the EGFR.

High-glucose-induced P-EGFR, NHE3 and AQP1 is PPARγ mediated

Since pioglitazone has PPARγ-mediated and PPARγ-independent effects, to demonstrate the specific role of PPARγ in the high-glucose-mediated increase in P-EGFR, we performed experiments in the presence of specific PPARγ siRNA. As clearly demonstrated, we were able to significantly block PPARγ protein expression using specific PPARγ siRNA to 27 ± 4% of control values (p < 0.001) (Fig. 4a). Phosphorylation of EGFR after high glucose exposure was confirmed using immunoprecipitation. High glucose significantly induced P-EGFR to 120.5 ± 1.5% (p < 0.01). This increase was abolished in the presence of PPARγ siRNA to 96.5 ± 9.5% (Fig. 4b). This suggests that the effect of pioglitazone on P-EGFR is PPARγ mediated.

Fig. 4
figure 4

High-glucose-induced increase in P-EGFR, NHE3 and AQP1 is PPARγ mediated. Decreased content of PPARγ was confirmed in gene-silenced PTCs using PPARγ siRNA vs non-specific siRNA control (NSC) (a). PTCs were incubated for 24 h with 5 mmol/l glucose medium (Control) or high glucose (HG) in the presence of PPARγ siRNA or NSC siRNA. P-EGFR (b), NHE3 (c) or AQP1 (d) protein content was determined by western blotting. Representative images for P-EGFR, NHE3, AQP1 and actin bands are shown. Normalised results are expressed as mean ± SEM (n = 3). *p < 0.05, **p < 0.01 and ***p < 0.001 vs NSC siRNA

Full size image

To confirm that the high-glucose effect on NHE3 and AQP1 is PPARγ mediated, we repeated experiments in the presence of PPARγ siRNA. High-glucose-induced increases in NHE3 and AQP1 levels were significantly inhibited to basal levels in the presence of PPARγ siRNA (to 91 ± 17% and 78 ± 12.8%, respectively) (Fig. 4c, d). These data show that the high-glucose increase in NHE3 and AQP1 in human PTCs is PPARγ mediated.

Effect of high glucose and pioglitazone on P-EGFR, AQP1, NHE-3 and PPARγ levels in vivo

The expression of P-EGFR and the effect of high glucose and pioglitazone on P-EGFR expression in the proximal tubule were confirmed in an in vivo model of diabetes mellitus. P-EGFR levels were very low in the proximal tubule of non-diabetic rats. However, its levels were increased in the proximal tubule membranes and cytoplasm of diabetic rats and further increased in diabetic rats treated with pioglitazone (Fig. 5). As expected, nuclear and cytoplasmic abundance of PPARγ was shown in control rats (Fig. 6a). However, this was decreased in the diabetic rats. Interestingly only cytoplasmic staining of PPARγ was shown in the diabetic rats (Fig. 6b). In diabetic rats, cytoplasmic and nuclear PPARγ levels were significantly increased in the presence of pioglitazone (Fig. 6c), suggesting increased PPARγ activity. NHE3 is found at a low level in the tubules of control rats (Fig. 7a). Its levels were more concentrated in the luminal membrane of the tubule. This was increased in the diabetic rats and further increased in the presence of pioglitazone with increased perinuclear staining (Fig. 7b, c). Similarly, increased levels of AQP1 were demonstrated in the diabetic rats compared with controls (Fig. 8a, b). AQP1 is clearly produced in the luminal membrane of the tubules. Its content was significantly increased in the presence of pioglitazone. Interestingly, pioglitazone also increased AQP1 levels in the apical brush border as well as in the luminal membrane (Fig. 8c).

Fig. 5
figure 5

P-EGFR levels in diabetic rats with and without pioglitazone (Piog). Immunohistochemistry of P-EGFR in control Ren-2 rats (a) and diabetic (b) and diabetic+pioglitazone-treated (c) animals. Negative IgG was performed to confirm staining specificity (not shown). Magnification×400. Quantification of P-EGFR immunostaining is shown (d). Values are represented as mean ± SEM. ***p < 0.001 and ****p < 0.0001 vs control and †† p < 0.01 vs diabetic

Full size image
Fig. 6
figure 6

PPARγ levels in diabetic rats with and without pioglitazone (Piog). Immunohistochemistry of PPARγ in control Ren-2 rats (a) and diabetic (b) and diabetic+pioglitazone-treated (c) animals. Negative IgG was performed to confirm staining specificity (not shown). Magnification×400. Quantification of PPARγ immunostaining is shown (d). Values are represented as mean ± SEM. *p < 0.05 vs control and †† p < 0.01 vs diabetic

Full size image
Fig. 7
figure 7

NHE3 levels in diabetic rats with and without pioglitazone (Piog). Immunohistochemistry of NHE3 in control Ren-2 rats (a) and diabetic (b) and diabetic+pioglitazone-treated (c) animals. Negative IgG was performed to confirm staining specificity (not shown). Magnification×400. Quantification of NHE3 immunostaining is shown (d). Values are represented as mean ± SEM. **p < 0.01 and ***p < 0.001 vs control and † p < 0.05 vs diabetic

Full size image
Fig. 8
figure 8

AQP1 expression in diabetic rats with and without pioglitazone (Piog). Immunohistochemistry of AQP1 in control Ren-2 rats (a) and diabetic (b) and diabetic+pioglitazone-treated (c) animals. Negative IgG was performed to confirm staining specificity (not shown). Magnification×400. Quantification of AQP1 immunostaining is shown (d). Values are represented as mean ± SEM. **p < 0.01 and ****p < 0.0001 vs control and †††† p < 0.0001 vs diabetic

Full size image

PPARγ regulates EGFR expression and activity by binding to EGFR promoter, and thiazolidinediones promote this binding

Using a motif scan program, we determined that the EGFR promoter contains a PPARγ-binding site at −1271: AGGTCCTAGTGAA. The ability of PPARγ to bind the EGFR promoter was tested using a ChIP assay. The data show that pioglitazone significantly increased PPARγ binding to the EGFR promoter 1.66 ± 0.2-fold (p < 0.05) (Fig. 9a). In addition, the more selective PPARγ agonist, L-805645, further increased PPARγ–EGFR promoter binding 8.8 ± 2.6-fold (p < 0.05) (Fig. 9a). This suggests that PPARγ can bind to the EGFR promoter to regulate its transcriptional activity.

Fig. 9
figure 9

EGFR promoter binding activity and Renilla luciferase assay. Cells were exposed to control medium, pioglitazone (Piog; 10 μmol/l) or the more selective PPARγ agonist, L-805645 (8 μmol/l) for 24 h. (a) The ChIP assay was performed using the ChIP Kit according to the manufacturer’s instructions. Real-time PCR was performed on purified DNA samples which were immunoprecipitated with PPARγ antibody. (b) Luciferase activity was measured using the Renilla Luciferase Assay System according to the manufacturer’s instructions. Results are expressed as mean ± SEM and shown as fold change compared with control. Three independent cell culture preparations were performed. *p < 0.05 vs control

Full size image

Since PPARγ can directly bind the EGFR promoter, and in order to establish if this binding induces EGFR activation, we determined the promoter activity of EGFR after PPARγ activation by a Dual-Luciferase Reporter Assay. Pioglitazone increased luciferase activity 1.5 ± 0.1-fold (p < 0.05), and L-805645 further increased luciferase activity 5.5 ± 1.5-fold (p < 0.05) (Fig. 9b).

Discussion

Cellular sodium and water transport are dysregulated in patients with diabetes mellitus, which are considered to be at least in part responsible for the high incidence of hypertension observed in these patients. EGFR is activated/transactivated by EGF, high glucose and angiotensin II, all factors implicated in the pathogenesis of diabetic nephropathy. To date, only a few studies have focused on the role of EGFR in enhancing tubular fluid reabsorption in the kidney proximal tubule. We have demonstrated that high glucose transactivates EGFR, which then regulates the key transporter of proximal tubular sodium reabsorption, NHE3, through downstream regulation of Sgk1 [11]. In addition, we have shown that NHE3 is regulated through Sgk1 by TZDs [8]. The possible role of EGFR in the TZD-mediated effects on sodium and water retention in the proximal tubule was examined in the present study. We have confirmed that PPARγ is produced in human PTCs and its level is increased after exposure to high glucose and the clinically available PPARγ ligand (pioglitazone). This is consistent with our previous findings using the HK2 primary human PTC line [6]. We have also confirmed, using in vivo models of STZ-induced diabetes mellitus, that PPARγ is produced in the proximal tubule and upregulated in diabetic animals in the presence of pioglitazone. Interestingly, diabetic animals had reduced levels of nuclear PPARγ. Our data show that, in diabetic animals, PPARγ nuclear translocation was diminished, suggesting reduced PPARγ activity, as has been previously reported [22]. In addition, we have specifically determined that a combination of high glucose with pioglitazone increased total EGFR mRNA and protein expression and EGFR phosphorylation. Although high glucose increased P-EGFR levels, there was no change in EGFR mRNA levels. In vivo, diabetic Ren-2 rats also showed increased P-EGFR expression, which was further potentiated by treatment with pioglitazone.

One of the most important functions of the kidney is to maintain electrolyte and metabolic homeostasis. Our results show that blockade of EGFR phosphorylation using PKI166, a novel and highly specific EGFR tyrosine kinase inhibitor of the pyrrolopyrimidine class, attenuates the increased abundance of NHE3 and AQP1 in PTCs in response to pioglitazone. This novel finding suggests a role for EGFR in TZD-mediated sodium and water transport in the proximal tubule. Drumm et al have demonstrated that aldosterone-induced NHE3 cell surface expression and activity in the proximal tubule is EGFR-dependent [23]. EGFR activation leads to a specific decrease in the levels of the tight junction integral protein, claudin-2, which results in modulation of paracellular sodium transport in MDCK cells [24, 25]. Under physiological conditions, EGFR activation appears to play an important role in the regulation of renal haemodynamics and electrolyte handling by the kidney, whereas, in different pathophysiological states, EGFR activation may mediate either beneficial or detrimental effects on the kidney [26]. The interdependence of PPARγ and EGFR in the kidney has not been previously shown. However, in other tissues such as cancer and endothelial cells, EGFR is modulated by PPARγ agonists in a PPARγ-dependent and -independent manner.

To determine if the high-glucose effect on NHE3 and AQP1 is PPARγ mediated, we used PPARγ siRNA. Our data show that P-EGFR is regulated downstream of PPARγ in response to high glucose. We have shown that high-glucose-induced P-EGFR is PPARγ mediated, and the high-glucose-mediated increase in NHE3 and AQP1 is also through PPARγ. In addition, we have clearly demonstrated that NHE3 and AQP1 are increased in diabetic rats and further increased in the presence of pioglitazone and that pioglitazone increased AQP1 expression in the brush border. Increased levels of NHE3 and AQP1 represent increased activity, as previously described [8, 27]. This explains at least in part the enhanced salt and water retention observed with pioglitazone use in patients with diabetes mellitus. Our data suggests cross-talk between PPARγ and EGFR in human PTCs. PPARγ and EGFR signalling is known to intersect in some cell types. For example, PPARγ regulation of urothelial differentiation is modulated by downstream EGFR signalling [28]. Coordinated activity between PPARγ and EGFR has been documented in the induction of caveolin-1 by rosiglitazone treatment in human colon cancer cells [29]. Interestingly, Lewis et al have recently reported that patients treated with pioglitazone have an increased risk of urinary bladder cancer if the drug is taken for more than 2 years and at a high dose [30]. The risk of increased bladder cancer is potentially mitigated by the use of EGFR inhibitors.

We have uniquely demonstrated that EGFR has a PPARγ-binding site and that TZD not only increased PPARγ levels but also its binding to the EGFR promoter and EGFR activation. Our findings suggest that high glucose and TZDs activate/transactivate PPARγ, which, through direct binding and downstream activation of EGFR, then induce NHE3 and AQP1 in the proximal tubule. Hence, the abnormalities in tubular cell growth known to occur in diabetic nephropathy, cellular sodium and water transport which are dysregulated in diabetes mellitus, and the increase in sodium reabsorption known to be responsible for the observed fluid retention with the use of TZDs are likely to occur through an EGFR-dependent mechanism. These data clearly highlight the importance of EGFR in renal sodium reabsorption and suggest that PPARγ agonists in combination with the clinically available EGFR blockers may be beneficial in regulating the nephromegaly and excessive sodium reabsorption observed in diabetic nephropathy. In fact, EGFR tyrosine kinase inhibition ameliorates the early changes that occur in diabetic nephropathy such as tubular epithelial cell proliferation, glomerular enlargement and changes in kidney weight in diabetic rats [31] and suppresses TGFβ-mediated matrix protein production in rat kidney interstitial fibroblasts [32]. Diabetic rats administered an EGFR tyrosine kinase inhibitor showed attenuated kidney and glomerular enlargement and a reduction in albuminuria, in association with podocyte preservation [31, 33]. These studies, in addition to our data, suggest that inhibition of the EGFR may provide an attractive therapeutic target for the treatment of diabetic nephropathy as well as limiting salt and water retention, which currently restricts the use of PPARγ agonists.