HNF4α and the Ca-Channel TRPC1 Are Novel Disease Candidate Genes in Diabetic Nephropathy

Human intestinal Caco-2 cells, a colorectal adenocarcinoma cell line, were obtained from the DSMZ (Deutsche Sammlung von Mikroorganismen und Zellkulturen, Braunschweig, Germany).

Nuclear extracts from rat liver and kidney were prepared as described by Gorski et al. (15), whereas nuclear extracts from Caco-2 cells were isolated by the method of Dignam et al. (16) with minor modifications, as detailed previously (17). Details for Western blot analysis and electrophoretic mobility shift assays have been published previously by Niehof and Borlak (17). The antibody directed against HNF4α (sc-8987x) was purchased from Santa Cruz Biotechnology (Heidelberg, Germany). The oligonucleotides were purchased from MWG Biotech (Ebersberg/Muenchen, Germany) and used as double-stranded ³²P-labeled probes (for sequence information, see online appendix table 1 [available at http://dx.doi.org/10.2337/db007-1065]). Super shift experiments were done with an HNF4α-specific antibody (sc-6556x; Santa Cruz Biotechnology), details previously published (17).

ChIP experiments were carried out with Caco-2 cells. All ChIP procedures and cloning and sequence analyses were carried out as described previously (17). Sequences were identified and annotated by database searches (GenBank Version Build 36.2, maintained by NCBI). Detailed sequence information is given in Table 2. Cloned fragments as well as proximal promoter sequences were interrogated for HNF4α binding sites by applying a genetic algorithm, as we have recently described (18). Accordingly, primer pairs were designed to confirm predicted sites experimentally (for primer sequence information, see online appendix table 2). Notably, for a novel candidate gene we demanded at least three independent confirmations from a series of ChIP experiments until that candidate gene was considered confirmed.

The transcription start site (TSS, +1) of the NCBI mRNA reference sequence (RefSeq) was aligned using the University of California Santa Cruz Genome Browser (http://genome.ucsc.edu/) for promoter annotation of the respective clones. Cloned fragments and respective proximal promoters (−1 to −10,000 bp) were checked for putative HNF4α binding sites with two different bioinformatic weight matrix-based tools, i.e., V$HNF4 01 with cutoff core similarity 0.75 and matrix similarity 0.78, Transfac matrix (Biobase, Wolfenbüttel, Germany [www.biobase.de]) and V$HNF4 01 with cutoff core similarity 0.75 and matrix similarity 0.82, or V$HNF4 02 with cutoff core similarity 0.75 and matrix similarity 0.76, Genomatix matrix (Genomatix software, München, Germany [www.genomatix.de]).

Human HNF4α-specific siRNA probes were purchased from Qiagen (Hilden, Germany). Caco-2 cells (1.5 × 10⁵ cells/well in 24-well plates) were transfected in triplicate for 48 h with 25 nmol/l of the siRNA duplex using HiPerFect transfection reagent (Qiagen). Alexa-Fluor488 labeled siRNA (Qiagen) was used as negative siRNA and as positive control for transfection efficiency.

Total RNA was isolated using the nucleospin RNA Isolation Kit (Macherey-Nagel) according to the manufacturers recommendations. Four micrograms of total RNA from each sample was used for reverse transcription (Omniscript Reverse Transcriptase, Qiagen). Real-time RT-PCR measurement was done with the Lightcycler (Roche Diagnostics, Mannheim, Germany) with the following conditions (online appendix table 3): denaturation at 94°C for 120 s, annealing at different temperatures for 8 s, extension at 72°C for different times, and fluorescence at different temperatures. The reaction was stopped after a total of 42 cycles, and at the end of each extension phase, fluorescence was observed and used for quantification within the linear range of amplification. Exact quantification was achieved by serial dilution with cDNA produced from total RNA extracts using 1:5 dilution steps. A detailed oligonucleotide sequence information is provided in online appendix table 3. Gene expression levels were normalized to mitATPase6, which was found to be stably expressed (online appendix Fig. 1).

Liver and kidney of fa/fa obese ZDF rats and of +/fa lean nondiabetic control rats, aged 14 weeks, were kindly provided by W. Linz and H. Ruetten (sanofi-aventis, Frankfurt, Germany) (14). All rats were male with mean body weight of 398.8 ± 30.2 kg (obese) and 334.2 ± 19.3 kg (lean). Representative phenotype data are provided in online appendix table 4.

Liver and kidney of STZ-administered Wistar rats (2 months of treatment) were kindly provided by P. Rösen (German Diabetes Research Institute, Düsseldorf, Germany). Experimental diabetes was induced in 6-week-old male rats by single intraperitoneal injection of 60 mg STZ/kg body wt. All animals developed hyperglycemia until 60 h after STZ injection. The animals did not receive an antidiabetic treatment. Diabetic rats had stable moderate hyperglycemia throughout the 2 months (mean blood glucose concentrations >400 mg/dl) (19). After 2 months the experiment was terminated.

Paraffin-embedded slices of human kidney biopsies (control biopsies, n = 3, and biopsies with nodular glomerulosclerosis derived from type 2 diabetic patients with diabetic nephropathies, n = 3; for details, see online appendix table 5) were kindly provided by Dr. J. Fries (Department of Pathology, University of Cologne, Cologne, Germany). The sections were deparaffinized, demasked with 1 mol/l citrate buffer, incubated with 0.6% H₂O₂ in methanol for 30 min, and subsequently incubated with protein block serum-free reagent (Dako, Glostrup, Denmark) for 8 min. Incubation with polyclonal antibodies (Santa Cruz Biotechnology) against HNF4α (sc-6556x, 1:50 dilution), TRPC1 (sc-15055, 1:10 dilution), or PLCB1 (sc-205, 1:10 dilution) was performed for 45 min. The sections were rinsed with Tris-buffered saline, incubated with biotinylated universal secondary antibodies (Dako) for 15 min, and subsequently incubated with horseradish peroxidase–conjugated streptavidin solution (Dako) for 15 min. Labeling was detected using a diaminobenzidine (DAB) chromogen solution (Dako) for 5 min. The sections were counterstained with hematoxylin before examination under light microscope. To confirm the specificity of the immunohistochemical localization, antibodies preabsorbed for 2 h with a fivefold excess of antigens for HNF4α (sc-6556P), TRPC1 (sc-15055P), or PLCB1 (sc-205P) (all Santa Cruz Biotechnology) were used.

We studied regulation of HNF4α in kidney and liver of the ZDF rat, an established disease model for diabetic nephropathy (14). By quantitative real-time RT-PCR, we found HNF4α to be significantly reduced in kidney and liver (Table 1). In addition, we used the ChIP cloning protocol to search for novel disease candidate genes targeted by HNF4α after formaldehyde crosslinking of nucleo-protein complexes. Notably, highly differentiated human epithelial Caco-2 cell cultures have been used successfully to study HNF4α function (17). Initially we screened immunoprecipitated DNA for enriched promoter sequences of HNF1α to ensure specificity of the ChIP assay (5,6). By PCR amplification, we confirmed immunoprecipitated DNA to contain promoter sequences of HNF1α and therefore evidence reliability of our experimental approach (Fig. 1A). Furthermore, HNF4α protein binding to the A-site of the HNF1α promoter (HNF1pro) was investigated by the electrophoretic mobility shift assay (EMSA) assay. We used a specific HNF4α antibody to supershift bands and observed strong binding of HNF4α as depicted in Fig. 1B. For each ChIP assay, identification of the HNF1pro site served as control, and ChIP assays of immunoprecipitated DNA yielded clones with inserts of up to 1,800 bp. The inserts were sequenced by capillary electrophoresis and by amplification with vector-specific primers. The genomic sequences were identified by database searches (GenBank, maintained by NCBI) for the human genome (Table 2). Then, cloned fragments as well as proximal promoter sequences were interrogated for HNF4α binding sites by use of two different bioinformatic matrices. Accordingly, primer pairs were designed to confirm predicted sites experimentally. In independent ChIP experiments followed by PCR analyses with gene-specific primers, robust identification of the new HNF4α gene targets was enabled. Notably, we confirmed candidate genes in at least three independent ChIP experiments. HNF4α in vivo binding to DNA was confirmed for clone 460 (TRPC1) and clone 264 (PLCB1) (Fig. 1A). We further analyzed target gene enrichment versus β-actin in ChIP DNA versus total input DNA by real-time PCR. Promoter sequences of known HNF4α target genes were enriched by approximately sevenfold for phosphoenolpyruvate carboxykinase (PEPCK) and ∼100-fold for HNF1α. Promoter sequences of PLCB1 were enriched approximately fourfold, whereas promoter sequences of TRPC1 were enriched approximately twofold (online appendix Table 6).

We also studied the ability of HNF4α to bind to cognate recognition sites with ³²P-labeled probes encompassing predicted HNF4α sites located in TRPC1 (probe 131) and PLCB1 (probe 05). We obtained supershifted bands with a specific HNF4α antibody, therefore demonstrating binding of HNF4α to both sites (Fig. 1B). A diagram of human TRPC1 and human PLCB1 indicating the location of ChIP clones, ChIP primers, and gel shift oligonucleotides is depicted in Fig. 1C. In the past, we demonstrated that Aroclor 1254 treatment of cell cultures induced HNF4α gene expression (20). After Aroclor 1254 treatment of Caco-2 cells, binding of the HNF4α protein to HNF1pro as well as to the newly identified binding sites within TRPC1 (probe 131) and PLCB1 (probe 05) was increased (online appendix Fig. 1A). Likewise, treatment of rats with Aroclor 1254 resulted in increased HNF4α protein binding (online appendix Fig. 1B). This provides additional evidence for the newly identified gene targets to be regulated by HNF4α. Alignment of human and rat TRPC1 and PLCB1 genes did not identify common HNF4α binding sites in these orthologous genes (see online appendix Table 7 for PhastCons sequence conservation among mammals). Nonetheless, oligonucleotides were designed to confirm predicted sites in rat genes experimentally. Note, we analyzed HNF4α binding with nuclear extracts obtained from rat kidney and of human origin using oligoprobes designed for the rat TRPC1 and PLCB1. We observed supershifted bands with a specific HNF4α antibody, therefore demonstrating binding of either rat or human HNF4α to rat TRPC1 and PLCB1 (Fig. 2A, B). A diagram of rat TRPC1 and rat PLCB1 indicating the location of gel shift oligonucleotides is depicted in Fig. 2C.

A summary of the cloned HNF4α gene targets is provided in Table 2. Clone 460 was ChIP verified from the first intron and identified as TRPC1 and codes for a nonselective cation channel. Clone 264 contained a ChIP-verified HNF4α binding site in the promoter region (around −919) and was identified as PLCB1. Furthermore, we studied TRPC1 and PLCB1 gene expression in kidney and liver extracts of ZDF diabetic rats and found its expression to be significantly reduced, as was expression of HNF4α (Table 1). Notably, gene expression of HNF4α and of TRPC1 (Fig. 3A, P = 0.0012) or PLCB1 (Fig. 3B, P = 0.005) in kidney of ZDF diabetic rats is significantly correlated (Fig. 3 and online appendix Fig. 4). Furthermore, protocols using streptozotocin (STZ) to cause β-cell toxicity and to result in diabetes are well known. Thus, administration of STZ to rats caused diabetic nephropathy and neuropathy (21,22), and we observed significant reduction of HNF4α transcript level in kidney and liver as reported previously (17). Likewise, TRPC1 gene expression in kidney and liver of STZ-administered rats was repressed, as was HNF4α itself (Table 3). Significant repression of TRPC1 in kidney to ∼80% of the control is only a marginal effect, but it is consistent with the results found in the ZDF model. Notably, PLCB1 was not regulated in STZ kidney (online appendix Table 8). To further probe for the role of HNF4α in TRPC1 gene regulation, we used an siRNA approach. Specifically, siRNA-mediated functional knock down of HNF4α in the human Caco-2 cell line resulted in significantly decreased gene expression of TRPC1 (Table 4; for transfection efficiency, see online appendix Fig. 2, and stable expression of mitATPase6 after transfection is shown in online appendix Fig. 5). These results further confirm TRPC1 to be a direct gene target of HNF4α.

We also determined protein expression of HNF4α, TRPC1, and PLCB1 in human kidney biopsies of type 2 diabetic patients diagnosed with nodular glomerulosclerosis. Kidney biopsies of three control patients and three patients diagnosed with diabetic nephropathy were used for immunohistochemistry (HNF4α, Fig. 4A and C; TRPC1, Fig. 4E and G; and PLCB1, Fig. 4I and K). To confirm specificity, antibodies preadsorbed with an excess of antigen were used but did not display any staining (HNF4α, Fig. 4B and D; TRPC1, Fig. 4F and H; PLCB1, Fig. 4J and L). Immunohistochemistry clearly evidenced reduction of HNF4α (Fig. 4C) and TRPC1 (Fig. 4G) protein expression in diseased kidneys. For PLCB1, results were not consistent (Fig. 4K). Overall, we demonstrate reduced HNF4α and TRPC1 gene and protein expression in diabetic nephropathy.

The orphan nuclear receptor HNF4α is an essential transcription factor and master regulatory protein in the control of gene expression of a wide range of metabolic enzymes. Dysfunction of HNF4α leads to metabolic disease, and identification of genes targeted by this factor will improve our understanding of disease-causing mechanisms. Notably, HNF4α regulates several genes involved in glucose metabolism and participates in the glucose-dependent insulin secretory pathways (5–7). There is evidence of HNF4α dysfunction resulting in multifactorial type 2 diabetes (9). Here we report significant reduction in HNF4α gene expression in kidney and liver of two accepted rodent disease models of diabetes. Furthermore, we found HNF4α protein expression to be reduced in kidney of patients diagnosed with diabetic nephropathy. By use of a ChIP cloning protocol, we identified TRPC1 and PLCB1 as novel HNF4α gene targets. Furthermore, TRPC1 gene expression was reduced in kidney and liver of ZDF and STZ-administered rats as determined by quantitative real-time RT-PCR. Indeed, gene expression of HNF4α and TRPC1 correlated well in kidney of individual animals. Similarly, TRPC1 protein expression was reduced in human patients diagnosed with diabetic nephropathy. We then used bioinformatics to search for HNF4α binding sites and applied a genetic algorithm to proximal promoter sequences. In vivo binding of HNF4α to novel gene targets was confirmed in separate ChIP experiments. We also confirmed in vitro binding of HNF4α to recognition sites of candidate genes (rat and human) by use of the EMSA assay. Notably, siRNA-mediated knock down of HNF4α resulted in repressed TRPC1 gene expression. Taken collectively, we propose HNF4α and TRPC1 as novel disease candidate genes in diabetic nephropathy.

Specifically, TRPC1 codes for a nonselective cation channel with six transmembrane segments and assembles as a tetramer. It does not form functional homodimers, rather heterodimers with TRPC3, -C4, -C5, and/or -C7 (23,24). Furthermore, TRPC1 can associate with polycystic kidney disease 2 (23,25,26) and is embedded in a protein complex that may include caveolin, which contributes to plasma localization of TRPC1 (27). Gene expression of all TRPC1 interaction partners was unchanged in kidneys of ZDF rats, except for reduced expression of TRPC3 (online appendix Table 9). TRPs may be the primary mode of non–voltage gated Ca²⁺ entry in the cell, and TRPC1 allows PLC-dependent Ca²⁺ influx in kidney, in the central nervous system, and in other peripheral tissues (23). Specifically, TRPC1 responds to general stimuli, such as depletion of intracellular Ca²⁺ stores, receptor activation, and membrane stretch (23). Thereof, plasma membrane Ca²⁺ influx occurs in response to intracellular Ca²⁺ depletion. This impacts regulation of store operated channels (SOCs), which are defined as channels that open in response to depletion of internal Ca²⁺ stores. Unlike in previous reports, where uncertainties regarding the molecular identity of SOCs have been discussed, contribution of TRPC proteins to store-operated Ca²⁺ entry is clearly evident (28). Notably, activated PLC converts phosphatidylinositol-bisphosphate (PIP₂) to diacylglycerol (DAG) and inositol-triphosphate (IP₃) and releases Ca²⁺ from the endoplasmatic reticulum via the IP₃ receptor. These events lead to SOC activation, which is dependent on store depletion in addition to released Ca²⁺ and DAG (28). It is of considerable importance that TRPC1 is localized in compartments of the plasma membrane in close proximity to superficial endoplasmic reticular structures. A mechanism involving association/dissociation between TRPC1 in the plasma membrane and IP3 receptor in the membrane of the stores has been proposed for the coupling of depleted stores to TRPC1 (23). The channel is embedded in a protein complex that includes PLCB1 (see below) in addition to IP₃ receptor and other proteins (23). Furthermore, TRPC1 was identified in glomeruli (29), and dysfunction of SOCs for diabetic nephropathy has been proposed (28). Here we report transcriptional and translational dysregulation of TRPC1 in animal disease models of diabetic nephropathy and in human kidney biopsies of patients diagnosed with diabetic nephropathy. Specifically, HNF4α and TRPC1 gene and protein expression were reduced in kidney of ZDF and STZ-administered rats and in human diabetic kidneys with nodular glomerulosclerosis as confirmed by immunohistochemistry. We propose TRPC1 to be a disease candidate gene in diabetic nephropathy. Recently, TRPC1 was identified in glomerular mesangial cells (29,30) to contribute to contractile function (31). Essentially, mesangial cells are located within glomerular capillary loops and play a role in the physiological regulation of glomerular hemodynamics and filtration (28,30). The function of these cells is controlled by a variety of ion channels in the plasma membrane, including SOCs (30). Impairment of SOC-mediated Ca²⁺ influx via reduced number of TRPC1 containing channels alters intracellular Ca²⁺ homeostasis and signaling and thus might contribute to the glomerular hemodynamic changes in diabetic nephropathy. Strikingly, linkage analysis in siblings with diabetic nephropathy pointed to chromosome 3q21-25 as a disease hotspot (32,33). Notably, TRPC1 is located within this region on human chromosome 3q22-24, and we identified a disease candidate gene in a chromosomal region linked to diabetic nephropathy.

In addition, we identified PLCB1 as a molecular target of HNF4α. This protein is involved in cell communication and signal transduction and belongs to one of the several PLCB isoforms. Specifically, PLCBs catalyze the hydrolysis of PIP₂ to generate the second messengers DAG and IP₃ with subsequent Ca²⁺ mobilization and protein kinase C (PKC) activation (34,35). PLCB1 is basically expressed in all human tissues tested so far. Within the plasma membrane, PLCB1 is activated by members of the α-q family of G-proteins. However, phospoinositide metabolism also occurs in the nucleus, and PLCB1 represents the main nuclear phospholipase form (34,35). Upregulation occurs in response to IGF-1 or insulin stimulation. Upon occupancy of the IGF-1 receptor, phosphorylated mitogen-activated protein kinase translocates in the nucleus, phosphorylates nuclear PLCB1, and triggers its signaling (34,35). Nuclear PLCB1 has been linked with either cell proliferation (G1 progression) or differentiation (34,35). It mediates the mitogenic effect of IGF-1 in Swiss 3T3 cells through activation of PKC-α, which elevates cyclin D1 and cyclin E, resulting in a higher proliferation rate (34,35). Note, nuclear PLCB1 activity markedly increased during myoblast differentiation through stimulation of the IGF-1 receptor but was repressed during erythroleukemia differentiation (34,35). Therefore, PLCB1 is important in cell proliferation and differentiation events. Recently, we reported that HNF4α targets two kinases (RSK4 and PAK5) that participate in the regulation of cell cycle. We suggested a novel role for HNF4α in repressing cell-cycle progression to enable cellular differentiation (17). Because of its role in epithelial differentiation, it is probable that HNF4α functions in the control of cell proliferation as well. Targeting PLCB1 might be part of this mechanism. Abnormal activation of the DAG-PKC pathway plays a pathogenic role in diabetic nephropathy (12,36–38); therefore, increased PLC activity might be expected. However, differences in the activity of PLCB isoforms might be important as well. Recently, expression of PLCB3 was reported to be decreased in diabetes (39). We report reduction of PLCB1 expression in kidney of ZDF rats with PLCB1 interacting with TRPC1 (as discussed above), therefore reinforcing the impact of HNF4α dysfunction in Ca²⁺ signaling during diabetic nephropathy. In conclusion, we propose HNF4α to be a transcriptional regulator for TRPC1 and PLCB1. We further suggest HNF4α and TRPC1 to be disease candidate genes in diabetic nephropathy.

This work was supported by a grant of the Lower Saxony Ministry of Culture and Science to J.B.

We thank S. Marschke, A. Pfanne, and A. Schulmeyer for valuable technical assistance, A. Kel and S. Reymann for assistance in bioinformatics and advice on design of PCR primers and gel shift oligonucleotides, Dr. J. Fries for providing us with paraffin-embedded slices of human kidney biopsies, Dres. W. Linz and H. Ruetten for providing kidney and liver of ZDF rats, and P. Rösen for providing tissue of STZ-administered diabetic rats.