Role of the Hepatocyte Nuclear Factor-1β (HNF-1β) C-terminal Domain in Pkhd1 (ARPKD) Gene Transcription and Renal Cystogenesis*

Hepatocyte nuclear factor-1β (HNF-1β) is a homeodomain-containing transcription factor that regulates tissue-specific gene expression in the kidney and other epithelial organs. Mutations of HNF-1β produce congenital cystic abnormalities of the kidney, and previous studies showed that HNF-1β regulates the expression of the autosomal recessive polycystic kidney disease (ARPKD) gene, Pkhd1. Here we show that the C-terminal region of HNF-1β contains an activation domain that is functional when fused to a heterologous DNA-binding domain. An HNF-1β deletion mutant lacking the C-terminal domain interacts with wild-type HNF-1β, binds DNA, and functions as a dominant-negative inhibitor of a chromosomally integrated Pkhd1 promoter. The activation of the Pkhd1 promoter by wild-type HNF-1β is stimulated by sodium butyrate or coactivators CREB (cAMP-response element)-binding protein (CBP) and P/CAF. The interaction with CBP and P/CAF requires the C-terminal domain. Expression of an HNF-1β C-terminal deletion mutant in transgenic mice produces renal cysts, increased cell proliferation, and dilatation of the ureter similar to mice with kidney-specific inactivation of HNF-1β. Pkhd1 expression is inhibited in cystic collecting ducts but not in non-cystic proximal tubules, despite transgene expression in this nephron segment. We conclude that the C-terminal domain of HNF-1β is required for the activation of the Pkhd1 promoter. Deletion mutants lacking the C-terminal domain function as dominant-negative mutants, possibly by preventing the recruitment of histone acetylases to the promoter. Cyst formation correlates with inhibition of Pkhd1 expression, which argues that mutations of HNF-1β produce kidney cysts by down-regulating the ARPKD gene, Pkhd1. Expression of HNF-1α in proximal tubules may protect against cystogenesis.

The hepatocyte nuclear factor-1 (HNF-1) 1 family comprises two nuclear transcription factors, HNF-1␣ and HNF-1␤, that regulate tissue-specific gene expression in the kidney, liver, and other epithelial organs (1)(2)(3)(4). HNF-1␣ and HNF-1␤ have a similar overall structure consisting of an N-terminal dimerization domain, a POU (Pit-1, Oct-1/2, Unc-86)-specific domain and a POU-homeodomain that mediate sequence-specific DNA binding, and a C-terminal domain. The sequences of the dimerization domains and DNA-binding domains are highly similar, but the C-terminal domains are divergent (5)(6)(7). HNF-1␣ and HNF-1␤ bind to DNA as obligatory homodimers or heterodimers and recognize an identical DNA sequence (3,8). They were first discovered in hepatocytes because of their roles in liver-specific expression of genes such as albumin and transthyretin, but they also regulate gene expression in epithelia of the pancreas, lung, intestine, and kidney. In general, HNF-1␣ and HNF-1␤ function as transcriptional activators, but transcriptional repression has also been observed depending on the promoter, cell type, and alternative splicing (7,9,10). The C-terminal domain of HNF-1␣ contains two discrete activation domains, ADI and ADII (11). The C-terminal domain of HNF-1␤ also has transcriptional activity but has not been characterized in detail (10).
HNF-1␣ and HNF-1␤ are both highly expressed in the kidney, but their patterns of expression are distinct. HNF-1␣ is restricted to the proximal tubules, whereas HNF-1␤ is expressed in all segments of the nephron and in renal collecting ducts (12)(13)(14). Neither protein is expressed in glomeruli. In the embryonic kidney, HNF-1␤ is expressed in the ureteric bud that will give rise to the renal collecting system and ureter as well as comma-and S-shaped bodies that will form the nephron proper. HNF-1␣ appears later and only in developing proximal tubules (12)(13)(14). Despite their prominent expression, the functions of HNF-1␣ and HNF-1␤ in the kidney are incompletely understood, and relatively few downstream target genes have been identified in this organ.
Mutations of the human HNF-1␣ gene (TCF1) produce the autosomal dominant disorder maturity-onset diabetes of the young, type 3 (MODY3) (15). MODY3 is associated with renal glycosuria in the absence of hyperglycemia (16). HNF-1␣ knock-out mice exhibit glycosuria, amino aciduria, and phosphaturia and show decreased expression of genes encoding glucose and phosphate transporters in the proximal tubule (12,17,18). These findings suggest that HNF-1␣ may be involved in the regulation of proximal tubule-specific gene expression. Mutations of HNF-1␤ (TCF2) produce maturity-onset diabetes of the young, type 5 (MODY5) and are invariably associated with congenital abnormalities of the kidney and/or genitourinary tract (19). The most common kidney abnormality is the formation of cysts in the renal tubules, and the acronym RCAD (renal cysts and diabetes) has been used to describe this syndrome (20). The spectrum of renal cystic abnormalities includes simple cysts, multicystic renal dysplasia, and hypoplastic glomerulocystic kidney disease. Renal involvement is often severe and can produce kidney failure. In addition to renal cysts, other kidney abnormalities caused by mutations of HNF-1␤ include oligomeganephronia, familial juvenile hyperuricemic nephropathy, and renal agenesis or hypoplasia (21).
To elucidate the mechanism by which mutations of HNF-1␤ produce renal cystic disease, we have previously generated transgenic mice expressing a mutant form of mouse HNF-1␤ that corresponds to a naturally occurring human mutation (A263fsinsGG) (22). The A263fsinsGG mutation prevents DNA binding and produces a dominant-negative mutant. Transgenic mice expressing mutant HNF-1␤ develop kidney cysts that are similar to those found in humans with MODY5. The kidney cysts show decreased expression of Pkhd1, the gene that is mutated in humans with autosomal recessive polycystic kidney disease (ARPKD) (23,24). Similarly, tissue-specific inactivation of HNF-1␤ in the kidney using Cre/loxP recombination results in the formation of kidney cysts and down-regulation of Pkhd1 expression (25). The Pkhd1 promoter contains an evolutionarily conserved binding site for HNF-1␤ that is located 49 bp upstream to the transcription start site within a region of cell-specific chromatin remodeling (22). Mutations of the HNF-1␤ binding site inhibit promoter activity in renal epithelial cells. Expression of wild-type HNF-1␤ transactivates the Pkhd1 promoter, whereas expression of a dominant-negative HNF-1␤ mutant inhibits promoter activity.
Taken together, these studies suggest that two renal cystic diseases, MODY5 and ARPKD, are linked in a common transcriptional pathway and that the mechanism of cyst formation in humans with mutations of HNF-1␤ involves down-regulation of PKHD1 (22,25). HNF-1␤ directly regulates the transcription of Pkhd1, but the mechanism remains unclear. The aim of the current study was to elucidate the molecular mechanism involved in the regulation of the Pkhd1 promoter by HNF-1␤, focusing on the role of the C-terminal domain. Additionally, we studied the consequences of expression of an HNF-1␤ mutant lacking the C-terminal domain on mammalian kidney development as a model for naturally occurring human mutations affecting this region.

MATERIALS AND METHODS
Plasmids-Expression plasmids encoding wild-type mouse HNF-1␤ (pHNF1␤) and HNF-1␤ containing C-terminal V5 and FLAG epitope tags (pHNF1␤-V5 and pHNF1␤-FLAG) were described previously (22). A plasmid containing 444 bp of the Pkhd1 promoter linked to a Photinus luciferase reporter gene (pPkhd1-luc) was described previously (22). The plasmid pHNF1␤⌬C-V5 encoding an HNF-1␤ mutant lacking 236 amino acids at the C terminus was produced by PCR amplification and cloning into pcDNA3.1/V5/His-TOPO (Invitrogen, Carlsbad, CA). The V5-epitope tag was replaced with a FLAG epitope tag to generate pHNF1␤⌬C-FLAG. The plasmid pGal4 (originally named pM1) encoding the Gal4 DNA-binding domain (DBD) under the control of the SV40 promoter was a gift from Ivan Sadowski (University of British Columbia). Plasmids encoding fusion proteins containing the Gal4 DBD and C-terminal fragments of HNF-1␤ were generated by inserting appropriate PCR fragments in-frame into pGal4. The plasmid pcDNA-Pod1-FLAG was generated by inserting a PCR product encoding mouse Pod1 (26) and a C-terminal FLAG epitope tag into pcDNA3.1. The plasmid pPkhd1-Gal4-VP16 was generated by replacing the luciferase coding region of pPkhd1-luc with DNA encoding a Gal4-VP16 fusion protein. In addition, a neomycin resistance cassette was inserted into this plasmid. The pG5-luc reporter plasmid containing 5 repeats of the Gal4 binding site and a TATA box upstream of the Photinus luciferase gene was a generous gift from Richard Baer (Columbia University) (27). The plasmids pCX-FLAG-P/CAF, pFLAG-Scr1␣, and pFLAG-Rac3 were generous gifts from Iannis Talianidis (Institute of Molecular Biology and Biotechnology, Herakleion, Crete, Greece) (28). The plasmid pRc/Rsv-mCBP-HA was a gift from Roland Kwok (University of Michigan) (29). The plasmids pcDNA-FLAG-P/CAF and pcDNA-mCBP-HA were obtained by subcloning the corresponding coding sequences into pcDNA3.1-V5/His.
Immunoprecipitation-In vitro translated proteins were generated using TNT Quick-coupled transcription/translation kits (Promega, Madison, WI) according to the manufacturer's directions. 1 g of plasmid and 40 l of rabbit reticulocyte lysate were used in each reaction. Reaction products were diluted with 1 ml of Tris-buffered saline (30 mM Tris, pH 7.4, 120 mM NaCl) containing 0.2% bovine serum albumin and protease inhibitor mixture (Sigma). After addition of 25 l of FLAGagarose or HA-agarose (Sigma), samples were rotated for 3 h and then washed three times with Tris-buffered saline. Bound proteins were analyzed by immunoblot analysis.
Immunoblot Analysis-HeLa cells were grown in 100-mm dishes and transfected with 2 g of plasmid DNA using Effectene according to the manufacturer's instructions (Qiagen, Valencia, CA). After 30 h, cells were lysed in 100 l of lysis buffer (Tris-buffered saline, 0.5% Triton X-100, protease inhibitor mixture (Hoffmann-La Roche Inc., Indianapolis, IN)), and 20 l of the lysates were separated by SDS-PAGE. Immunoblots were prepared as described previously (22) and incubated with anti-Gal4 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) or horseradish peroxidase-conjugated anti-V5 (Invitrogen). Immune complexes were detected using chemiluminescence. Signal intensity was quantified with a FluorImager and expressed as a percentage relative to the highest signal measured in each group.
Cell Culture, Transfection, and Reporter Gene Assays-Mouse inner medullary collecting duct cells (mIMCD-3) and HeLa cells were grown as described previously (30). Cells were plated in 6-well dishes (3 ϫ 10 5 cells/well) and were cotransfected with 0.3 g of pPkhd1-luc or pG5-luc and 0.01-0.3 g of effector plasmids. Transfection was performed using Effectene (Qiagen). Luciferase assays were performed as described previously (22). To control for transfection efficiency, cells were cotransfected with 0.05 g of phRL-TK(IntϪ) encoding Renilla luciferase (Promega), and relative luciferase activity was calculated as the ratio of Photinus and Renilla luciferase. To produce cell lines stably expressing a Gal4-VP16 fusion protein under the control of a chromosomally integrated Pkhd1 promoter, mIMCD-3 cells were transfected with pPkhd1-Gal4-VP16 and grown in the presence of G418 (0.8 mg/ml) for 17 days. Six independent cell lines that expressed Gal4-VP16 were identified and used in subsequent experiments.
Electrophoretic Mobility Shift Assay (EMSA)-EMSA was performed as described previously (22) using in vitro translated proteins and a 32 P-labeled DNA fragment containing the mouse Pkhd1 promoter (Ϫ81 bp to Ϫ38 bp) including the HNF-1 binding site.
Animals-The plasmid pKsp-HNF1␤⌬C-V5 was generated by inserting the coding sequence of HNF1␤⌬C-V5 into a plasmid 3Ј to the 1.3-kb Ksp-cadherin promoter and ␤-globin TATA box (29). A DNA fragment containing the Ksp-cadherin promoter, HNF1␤⌬C-V5 coding region, and bovine growth hormone polyadenylation site was excised with BamHI and Asp 718 and purified as described previously (29). Transgenic mice were produced by the University of Texas Southwestern Transgenic Core Facility and analyzed at ages 30, 52, and 100 days. Tissues were harvested from anesthetized mice and were fixed by perfusion with 4% paraformaldehyde. Paraffin embedding, sectioning, and histochemical staining were performed using standard procedures (31).
In Situ Hybridization-Fluorescence in situ hybridization was performed as described previously (22). Isotopic in situ hybridization was performed using 35 S-labeled Pkhd1 riboprobes essentially as described previously (32). Following hybridization and washing, the sections were dipped in K.5 nuclear emulsion gel (Ilford, UK) and exposed for 21-28 days.

HNF-1␤⌬C Mutant Is Capable of Dimerization and DNA
Binding-To determine whether the C-terminal domain of HNF-1␤ is required for transcriptional activation of the Pkhd1 promoter, we produced a C-terminal truncated mouse mutant (HNF-1␤⌬C). The mouse HNF-1␤⌬C mutant lacks 236 amino acids at the C terminus, similar to a naturally occurring human MODY5 mutant (P328L329fsdelCCTCT) (20). The HNF-1␤⌬C mutant retains the N-terminal dimerization domain as well as the POU-specific domain and POU-homeodomain that mediate DNA binding (Fig. 1A). The human P328L329fsdelCCTCT mutant has previously been shown to bind to an HNF-1 site in the HP1 promoter (33). To test the ability of mouse HNF-1␤⌬C to interact with DNA, EMSA was performed using programmed reticulocyte lysates and a radiolabeled DNA fragment containing the HNF-1␤ binding site from the mouse Pkhd1 promoter. Lysates programmed with full-length FLAG-tagged HNF-1␤ produced a DNA-protein complex that was not seen in lysates programmed with empty expression plasmid ( To verify the composition of the complexes, EMSA was performed in the presence of antibodies directed against the FLAG and V5 epitope tags that were appended to wild-type and mutant HNF-1␤, respectively. The presence of supershifted bands (lanes 8 -15) indicated that the upper complex contained HNF-1␤ by itself, the lower complex contained HNF-1␤⌬C by itself, and the middle complex contained both proteins. These results demonstrated that the HNF-1␤⌬C mutant competed with wild-type HNF-1␤ for binding to the HNF-1 site in the Pkhd1 promoter.
C-terminal Domain of HNF-1␤ Contains an Autonomous Activation Domain-The transcriptional activity of the C-terminal domain of HNF-1␤ was tested using mammalian one-hybrid assays. An expression plasmid encoding the yeast Gal4 DBD fused to the C-terminal domain of HNF-1␤ (amino acids 298 -532) was co-transfected into mIMCD-3 cells with a Gal4responsive luciferase reporter plasmid. mIMCD-3 cells are derived from the renal collecting duct and endogenously express HNF-1␤ and Pkhd1 (22). Expression of the Gal4-HNF1␤-(298 -532) fusion protein produced a dose-dependent increase in luciferase activity, whereas expression of the Gal4 DBD by itself had no effect ( Fig. 2A). These results indicated that the Cterminal domain of HNF-1␤ contained an autonomous activation domain that was functional when fused to a heterologous DNA-binding domain.
To more precisely define the activation domain, Gal4 fusion proteins containing various deletions of the HNF-1␤ C terminus were tested in the one-hybrid assay. In addition to mIMCD-3 cells, the studies were performed in HeLa cells, which do not express endogenous HNF-1␤ but can support transactivation by exogenous HNF-1␤ (30). The region containing amino acids 298 -432 did not activate transcription when fused to the Gal4 DBD. Gal4 fusion proteins containing amino acids 311-483 induced the highest activity in HeLa cells, and further addition of the C-terminal sequence did not increase but instead decreased activation. In mIMCD-3 cells, further C-terminal additions led to an increase in reporter gene activity. Progressive deletions from either end of the C-terminal domain produced gradual reductions in transcriptional activation, indicating that the sequences contributing to transcriptional activation were distributed over a relatively broad region. Gal4 fusion proteins containing the core region of the activation domain from amino acids 352 to 483 retained 83% (HeLa cells) and 59% (mIMCD-3 cells) of the maximal activity. The integrity of this region was required for transcriptional activation because further deletion to amino acids 388 -471 reduced the activity to 9.5% (HeLa cells) and 3.7% (mIMCD-3 cells).
Deletion of the C-terminal Domain of HNF-1␤ Produces a Dominant-negative Mutation-Deletion of the C-terminal domain of HNF-1␤ produces a dominant gain-of-function mutation when assayed with an episomal promoter-reporter gene (30,33). To evaluate the effects of mutations of HNF-1␤ in the context of native chromatin, we performed reporter gene assays using a chromosomally integrated promoter-reporter gene. The effects of mutant transcription factors on integrated promoters can be difficult to detect, especially if the efficiency of transfection is low. Expression of dominant-negative mutants and inhibition of reporter gene activity in a minor population of transfected cells may not be detected because of continued activity in untransfected cells. To circumvent this limitation we have developed a novel two-step reporter gene assay that can be generally applied to study the effects of mutant transcription factors on an integrated promoter independent of transfection efficiency. mIMCD-3 cells were stably transfected with the Pkhd1 promoter linked to the coding region of Gal4-VP16, a synthetic transcriptional activator containing the Gal4 DBD and herpes simplex virus activation domain. The cells were then transiently transfected with the pG5-luc reporter plasmid and expression plasmids encoding wild-type or mutant HNF-1␤. Expression of HNF-1␤ alters Pkhd1 promoter activity and Gal4-VP16 production, which is quantified with the pG5-luc reporter plasmid. Because luciferase is only produced in transfected cells, and untransfected cells are silent, inhibition of Pkhd1 promoter activity can be readily detected.
Six independent cell lines containing the Pkhd1 promoter linked to the Gal4-VP16 coding region were transiently transfected with pG5-luc and expression plasmids encoding wildtype HNF-1␤, the HNF-1␤⌬C mutant, or Pod1 as a negative control. Relative luciferase activity was measured 48 h after transfection. Expression of the HNF-1␤⌬C mutant inhibited Pkhd1 promoter activity compared with an irrelevant protein (Pod1), consistent with a dominant-negative effect (Fig. 3A). In contrast, expression of wild-type HNF-1␤ did not produce significant inhibition. Similar results were obtained in all six cell lines, indicating that the inhibition by HNF-1␤⌬C was independent of position effects. Transfection with increasing amounts of HNF-1␤⌬C expression plasmid produced a dose-dependent decrease in promoter activity (Fig. 3B). Transfection with increasing amounts of wild-type HNF-1␤ expression plasmid initially produced a dose-dependent stimulation of promoter activity, and transfection with higher amounts of plasmid decreased activity consistent with squelching (Fig. 3B). Transfection with increasing amounts of wild-type HNF-1␤ expression plasmid in the presence of the histone deacetylase inhibitor, sodium butyrate, resulted in a dose-dependent stimulation of promoter activity and prevented squelching (Fig.  3C). The effect of sodium butyrate was specific, because inhibition was not prevented by treatment with 5-aza-2Ј-deoxycytidine, an inhibitor of DNA methylation.
Histone Acetyltransferases (HATs) Interact with the HNF-1␤ C Terminus-Because the preceding results suggested that histone acetylation might be important for regulation of Pkhd1 promoter activity, we tested whether coactivators that have intrinsic HAT activity interacted with the C-terminal domain of HNF-1␤. mIMCD-3 cells were cotransfected with pHNF1␤-FLAG, the pPkhd1-luc reporter plasmid, and expression plasmids encoding CBP, P/CAF, steroid receptor coactivator-1␣, and RAC3, which are HATs that have been shown to interact with HNF-1␣. Co-expression of P/CAF or CBP augmented the activation by HNF-1␤ by 8-and 5.3-fold, respectively, whereas expression of steroid receptor coactivator-1␣ or RAC3 had no significant effect (Fig. 4A, gray bars). The stimulation by P/CAF and CBP required the C-terminal domain of HNF-1␤ because no augmentation of activity was observed when the proteins were co-expressed with the HNF-1␤⌬C mutant (Fig.  4A, black bars). To test whether the C-terminal domain was sufficient to enhance transcriptional activity, we studied the effects of P/CAF and CBP on the activity of the Gal4-HNF1␤-(298 -532) fusion protein. Co-expression of P/CAF or CBP increased the activation mediated by Gal4-HNF1␤-(298 -532) by 65 and 43%, respectively (Fig. 4B). P/CAF and CBP had no effect on the Gal4 DBD by itself, indicating that the HNF-1␤ C-terminal domain was required.
To determine whether the C-terminal domain of HNF-1␤ directly interacted with P/CAF or CBP, co-immunoprecipitation was performed. HNF-1␤ and coactivators containing two different epitope tags were translated in vitro, and proteins were immunoprecipitated with antibodies against one of the epitope tags. The immunoprecipitates as well as the proteins remaining in the supernatant were analyzed by immunoblot analysis. Immunoprecipitation of FLAG-tagged P/CAF resulted in co-precipitation of full-length HNF1␤-V5, whereas HNF1␤⌬C-V5 lacking the C-terminal domain was not co-precipitated (Fig. 4C, lanes 1 and 2). Similarly, immunoprecipitation of HA-tagged CBP resulted in co-precipitation of HNF1␤-V5 but not HNF1␤⌬C-V5 (Fig. 4C, lanes 3 and 4). HNF-1␤⌬C was found in the supernatant indicating that equivalent amounts of the proteins were produced. As a positive control, immunoprecipitation of HNF1␤-FLAG resulted in the co-precipitation of HNF1␤-V5 (Fig. 4C, lane 5).
Kidney-specific Expression of HNF-1␤⌬C Produces Renal Cysts Similar to Kidney-specific Deletion of HNF-1␤-Kidneyspecific deletion of HNF-1␤ in transgenic mice leads to the formation of kidney cysts (25). Similarly, kidney-specific expression of an HNF-1␤ mutant that is unable to bind DNA causes dilatation of renal tubules and cyst formation (22). To determine the effects of deletion of the C-terminal domain on the function of HNF-1␤ in vivo, transgenic mice expressing the HNF-1␤⌬C mutant were produced. Because constitutive gene inactivation of HNF-1␤ is embryonic lethal (34,35), a tissuespecific expression system was used. Transgenic mice expressing HNF1␤⌬C-V5 under the control of the kidney-specific Kspcadherin promoter were produced as described previously (22). The Ksp-cadherin promoter directs expression in renal tubules in adult mice and in the developing kidney, ureter, and sex ducts in embryos (29). Individual founders were genotyped and sacrificed at postnatal days 30 (P30), 52, or 100. To test for transgene expression, proteins were extracted from the left kidney of each animal, whereas the right kidney was used for morphological analyses. Fig. 5 shows the histology of the kidneys and ureters from transgenic mice at P52 and P100 compared with a non-transgenic littermate at P52. Similar to mice with kidney-specific deletion of HNF-1␤, HNF-1␤⌬C transgenic mice developed extensive renal cystic disease. Cysts were found throughout the cortex and medulla of the kidney (Fig. 5, A-C). Staining with Masson trichrome showed extensive interstitial fibrosis in the kidneys of transgenic mice but not in the non-transgenic animals (Fig. 5, D-F, blue staining). Cysts were identified in all animals expressing HNF-1␤⌬C (n ϭ 6) but were not found in any of the control non-transgenic littermates (n ϭ 6), which indicated that cyst formation was not because of position effects. Additional phenotypic similarities between HNF-1␤⌬C transgenic mice and mice with kidney-specific deletion of HNF-1␤ included the presence of multilayered cysts lined by a hyperplastic epithelium (Fig. 5I) and marked dilatation of the ureter (Fig. 5, K-L).
Expression of Pkhd1 Is Down-regulated and Cell Proliferation Is Increased in Cyst Epithelia-To test whether expression of the HNF-1␤⌬C mutant affected transcription of Pkhd1 in vivo, mRNA in situ hybridization was performed. Fluorescence in situ hybridization showed that cyst epithelial cells in HNF-1␤⌬C transgenic mice lacked Pkhd1 transcripts (Fig. 6, A-B,  arrowheads), whereas Pkhd1 was expressed in surrounding non-cystic tubules (arrows). Similar down-regulation of Pkhd1 was previously seen in kidney-specific HNF-1␤ knock-out mice Luciferase activity is shown relative to cells that were not transfected with coactivators. B, mIMCD-3 cells were cotransfected with 0.2 g of pG4-luc, 0.1 g of pGal4 or 0.1 g of pGal4-HNF1␤ (amino acids 298 -532), and 0.1 g of pcDNA3.1, pCX-FLAG-P/CAF (P/CAF), or pRc/ Rsv-mCBP-HA (CBP). Luciferase activity is shown relative to cells cotransfected with pPkhd1-luc and pGal4-HNF1␤ (amino acids 298 -532) alone. Data are the mean Ϯ S.E. of three independent transfections. C, reticulocyte lysates were programmed with 0.5 g of plasmids encoding the indicated proteins and were precipitated with anti-FLAG (lanes 1-2 and 5-6) or anti-HA antibodies (lanes 3-4). The immunoprecipitates (upper panels) and supernatants (lower panel) were subjected to immunoblot analysis with anti-V5 antibody. (22). To quantify the expression, isotopic in situ hybridization was performed (Fig. 6, C-E). The silver grains overlying epithelial cell nuclei were counted in 10 different fields, and the values were expressed as grains/nucleus (Fig. 6F). The number of Pkhd1 mRNA transcripts in cyst epithelial cells was reduced by 76% compared with non-cystic tubules in the same animal and by 78% compared with tubules in non-transgenic littermates.
Cyst formation in polycystic kidney disease involves increased proliferation of renal tubular epithelial cells. Inactivation of HNF-1␤ has been associated with increased cell proliferation (25). To measure cell proliferation in HNF-1␤⌬C transgenic mice, kidney sections were stained with an antibody against PCNA. Abundant PCNA-positive cells were observed in the cystic kidneys, whereas only isolated positive cells were seen in control non-transgenic kidneys (Fig. 7). PCNA-positive cells were found in multilayered cysts as well as simple cysts, but interestingly, the multilayered cyst epithelia showed PCNA staining primarily in the basal cell layer (Fig. 7C). 51% of the epithelial cells in simple cysts were PCNA-positive, whereas only 1.8% of cells in the kidneys of non-transgenic mice were PCNA-positive (Fig. 7E). Increased numbers of PCNA-positive cells (12%) were also observed in non-cystic tubules of transgenic mice suggesting that proliferative activity may increase prior to tubular dilation.
Absence of Cyst Formation and Pkhd1 Inhibition in Proximal Tubules-Most cysts in the kidneys of HNF-1␤⌬C transgenic animals stained positive for aquaporin-2, a water channel that is specifically expressed in renal collecting ducts (Fig. 8A, red). The aquaporin-2-positive cysts were lined with cells that expressed the HNF-1␤⌬C transgene, as identified by staining with an antibody to the V5 epitope tag (Fig. 8A, green nuclei). In contrast, none of the renal cysts in HNF-1␤⌬C transgenic animals stained positive with LTA, a lectin that specifically interacts with carbohydrates on proximal tubules. The absence of cysts derived from proximal tubules was not because of the lack of transgene expression in this nephron segment. Fig. 8B shows that the expression of the transgene in LTA-positive proximal tubules was comparable with the expression in cystic collecting ducts. Proximal tubule-derived cysts were also not observed after kidney-specific deletion of HNF-1␤ or in transgenic mice expressing an HNF-1␤ mutant that is unable to bind DNA (22,25). Arrowheads indicate the absence of hybridization signal in the cyst epithelium. Nuclei were counterstained blue with 4Ј-6-diamidino-2phenylindole in A. C-E, isotopic in situ hybridization of Pkhd1 mRNA transcripts in kidney sections from HNF-1␤⌬C transgenic mice at P30 (D-E) and a non-transgenic littermate (C). Silver grains are absent in cysts (arrowheads) but are present in surrounding non-cystic tubules (arrows). Bars, 10 m. F, quantification of Pkhd1 expression in HNF-1␤⌬C transgenic mice and non-transgenic littermates. The number of silver grains per nucleus was counted in cystic and non-cystic tubules. Data are the mean Ϯ S.E. of measurements in 10 different fields.
One possibility was that cyst formation in proximal tubules was prevented by continued expression of the paralogous protein, HNF-1␣, which recognizes the same DNA sequence as HNF-1␤ (2). Staining of wild-type kidney sections with an antibody to HNF-1␣ and co-staining with LTA showed that HNF-1␣ was expressed exclusively in the nuclei of renal proximal tubule cells (Fig. 8C). In contrast, HNF-1␤ was expressed in the nuclei of cells in all segments of the nephron including proximal tubules, loops of Henle, distal tubules, and collecting ducts (Fig. 8D, and data not shown). Similar results have previously been obtained using in situ hybridization and lacZ knock-in mice (12,14). In HNF-1␤⌬C transgenic mice, HNF-1␣ was expressed in non-cystic proximal tubules but was absent from the cysts (Fig. 8, E-G). To test whether the expression of the HNF-1␤⌬C transgene affected Pkhd1 transcription in proximal tubules, Pkhd1 transcripts were visualized by fluorescence in situ hybridization. Pkhd1 transcripts were present in LTA-positive, non-cystic proximal tubules (Fig. 8H). The LTApositive cells expressed Pkhd1 despite the expression of HNF-1␤⌬C in the same cells, indicating that expression of HNF-1␤⌬C in proximal tubules did not interfere with Pkhd1 transcription (Fig. 8I). In contrast, cystic collecting duct cells that expressed HNF-1␤⌬C also lacked Pkhd1 RNA transcripts (Fig. 8J). DISCUSSION Previous studies have shown that HNF-1␤ directly regulates Pkhd1 gene expression and established a novel link between two renal cystic diseases, MODY5 and ARPKD (22,25). The current studies were undertaken to elucidate the molecular mechanism by which HNF-1␤ regulates the Pkhd1 promoter. Here we show that the C-terminal domain of HNF-1␤ is required for activation of the Pkhd1 promoter. Deletion of the C-terminal domain does not interfere with DNA binding but blocks transactivation. The C-terminal part of HNF-1␤ can activate transcription when fused to a heterologous DNA-binding domain, and deletion analysis showed that the active region is primarily located between amino acids 352 and 483. The integrity of this region is essential because further deletion to amino acids 388 -471 reduced activity by Ͼ90%. The location of an activation domain in the mid-portion of the C-terminal part of HNF-1␤ contrasts with HNF-1␣, in which two activation domains (AD) have been identified at either end of the Cterminal domain: ADI consists of a serine-rich region located at the extreme C terminus of HNF-1␣ and ADII consists of a proline-rich region located adjacent to the POU-homeodomain (11). Neither region is conserved in HNF-1␤, which may in part explain the differing biological properties of the two proteins.
We produced a deletion mutant, HNF-1␤⌬C, which lacks 236 amino acids at the C terminus of mouse HNF-1␤. The mouse HNF-1␤⌬C mutant is similar to a human mutant lacking 230 amino acids at the C terminus that is predicted to be produced by a naturally occurring mutation (P328L329fsdelCCTCT) of the human HNF-1␤ (TCF2) gene. The P328L329fsdelCCTCT mutation was identified in a family with inherited diabetes, renal cysts, and impaired renal function (20). We have previously shown that the A263insGG mutation of human HNF-1␤ encodes a dominant-negative mutant that is unable to bind DNA (22). In contrast, the HNF-1␤⌬C mutant is able to bind DNA either as a homodimer or heterodimer with wild-type HNF-1␤. The HNF-1␤⌬C mutant may therefore compete with wild-type HNF-1␤ for binding to the Pkhd1 promoter.
Previous studies by our group and others have suggested that deletion of the C-terminal domain of HNF-1␤ produces a dominant gain-of-function mutation (30,33). This conclusion was based on studies performed with transiently transfected promoter-reporter genes. However, regulatory proteins that affect the chromatin structure of a promoter might be more accurately studied with reporter gene assays using a chromosomally integrated promoter. Therefore, we produced cell lines in which the expression of a Gal4-VP16 fusion protein is controlled by a stably integrated Pkhd1 promoter and then transfected the cells with a Gal4-responsive luciferase reporter plasmid and expression plasmids encoding wild-type or mutant HNF-1␤⌬C. Under these conditions, the effects of the proteins on promoter activity are only measured in transfected cells, whereas non-transfected cells remain experimentally silent. This novel assay facilitates the detection of mutants that have inhibitory activity and can be generally applied to the study of any transcription factor and promoter of interest. In contrast to the results obtained with an episomal promoter, the HNF-1␤⌬C mutant produces a dose-dependent inhibition of Pkhd1 promoter activity in cells that endogenously express wild-type HNF-1␤. Consistent with this finding, expression of the HNF-1␤⌬C mutant in transgenic mice inhibits Pkhd1 mRNA expression in kidney tubules. Collectively, these studies indicate that the HNF-1␤⌬C mutant functions as a dominantnegative rather than a gain-of-function mutant. The human P328L329fsdelCCTCT mutant is likely to behave similarly.
The observation that the HNF-1␤⌬C mutant inhibits the activity of a chromosomally integrated promoter, but not an episomal promoter, suggests that the regulation of Pkhd1 promoter activity may be chromatin-dependent. The involvement of chromatin remodeling in Pkhd1 transcription was initially suggested by studies in which a DNase-hypersensitive site was identified in the promoter (22). The DNase-hypersensitive site is located near the HNF-1␤ binding site, and its presence is cell-specific and correlates with Pkhd1 expression. DNA-binding proteins can induce chromatin remodeling by recruiting enzymes that alter histone acetylation. In the present study we found that sodium butyrate, an inhibitor of histone deacetylase, stimulates the activation of the Pkhd1 promoter by wildtype HNF-1␤. Two coactivators that have intrinsic HAT activity, CBP and P/CAF, also augment the activation by HNF-1␤. The coactivators have no effect on the HNF-1␤⌬C mutant, even though the latter is able to bind to the promoter, which indicates that the C-terminal domain is required. One-hybrid assays showed that the C-terminal domain by itself is sufficient to recruit CBP and P/CAF.
Co-immunoprecipitation confirmed that HNF-1␤ interacts with CBP and P/CAF and that the C-terminal domain is required for the interaction. However, when expressed by itself as a monomer the C-terminal domain is unable to interact with CBP or P/CAF (data not shown). One possibility is that the HNF-1␤ C-terminal domain only interacts with HATs as a dimer, such as in the native protein or in the Gal4 fusion protein, whereas monomers are unable to interact. Taken together, these studies elucidate the molecular mechanism of the dominant-negative action of the HNF-1␤⌬C mutant. The HNF-1␤⌬C mutant is able to bind to the Pkhd1 promoter, but the recruitment of coactivators is impaired. This mechanism likely also applies to humans with the P328L329fsdelCCTCT muta-tion and other mutations that involve the C-terminal domain. It remains unclear whether the recruitment of coactivators is interrupted in heterodimers containing HNF-1␤⌬C and the wild-type protein or whether the dominant-negative action is based solely on occupancy of the DNA binding site by transcriptionally inactive HNF-1␤⌬C mutants.
The above findings point to additional functional differences between HNF-1␤ and HNF-1␣. HNF-1␣ has previously been implicated in chromatin remodeling through its interaction with HATs (28). The C terminus of HNF-1␣ but not HNF-1␤ is able to interact with Rac3 and steroid receptor coactivator-1␣. The difference in the interaction potential between HNF-1␣ and HNF-1␤ is likely explained by the divergence of their C-terminal sequences. Moreover, CBP interacts with the Nterminal region of HNF-1␣, whereas the interaction with HNF-1␤ is limited to the C-terminal domain. During the preparation of this manuscript, Barbacci et al. (36) reported the interaction of HNF-1␤ with coactivators. Similar to our results, their study showed that HNF-1␤ interacts with CBP and P/CAF. However, Barbacci et al. (36) found that CBP interacts with the N-terminal part of HNF-1␤ and P/CAF interacts with both the N-terminal and C-terminal parts, whereas we found that CBP and P/CAF interact with the C-terminal domain of HNF-1␤. These differences may reflect the use of different cell types, promoters, or splice variants in the two studies.
Tissue-specific expression of the HNF-1␤⌬C mutant in transgenic mice produces renal cysts and ureteral dilatation. A similar phenotype is observed in mice with kidney-specific deletion of HNF-1␤, which provides additional evidence for a dominantnegative action of the HNF-1␤⌬C mutant. Similar to HNF-1␤ knock-out mice, HNF-1␤⌬C transgenic mice show increased Anti-V5 staining (green) shows that the HNF-1␤⌬C transgene is expressed in the nuclei of cystic collecting ducts that co-express aquaporin-2 (AQP2) (A, red) as well as non-cystic proximal tubules that are co-labeled with LTA (B, red). Cy, cyst. C and D, expression of HNF-1␣ and HNF-1␤ in wild-type mice. HNF-1␣ (C, red) is expressed in proximal tubules that are co-labeled with LTA (green). HNF-1␤ (D, red) is expressed in all renal tubules. E-G, expression of HNF-1␣ in HNF-1␤⌬C transgenic mice. HNF-1␣ (red) is absent from cysts but is expressed in surrounding non-cystic tubules. The HNF-1␤⌬C transgene (green) is expressed in both the cysts and surrounding non-cystic tubules. H-J, expression of Pkhd1 in cystic and non-cystic tubules. H, arrows indicate expression of Pkhd1 mRNA transcripts (green dots) in proximal tubules of wild-type mice that were stained with LTA (red). I, Pkhd1 mRNA transcripts (green dots) are also expressed in LTA-positive proximal tubules (red) of HNF-1␤⌬C transgenic mice. Arrows indicate transgene expression (green nuclei). J, arrowheads indicate the absence of Pkhd1 mRNA transcripts (green dots) in cystic collecting ducts that express aquaporin-2 (red). Arrows indicate Pkhd1 expression in surrounding non-cystic tubules. Bars, 10 m; Cy, cyst; G, glomerulus. epithelial proliferation and form multilayered cysts that are lined by a hyperplastic epithelium. The mechanism of cyst formation is likely to involve decreased transcription of Pkhd1, because homozygous mutations of this gene produce renal cysts in humans with ARPKD. The mechanism of ureteral dilatation is less clear, because this abnormality is not typically seen in ARPKD. Fetal urinary tract obstruction can produce hydroureter and multicystic renal dysplasia. However, the presence of normal glomeruli and proximal tubules in HNF-1␤⌬C transgenic mice argues against this possibility. Both HNF-1␤ and Pkhd1 are expressed in the ureteric bud that gives rise to the adult ureter, so it is likely that ureteral dilatation is a primary abnormality resulting from decreased transcription of Pkhd1 or other HNF-1␤ target genes during embryonic development.
Down-regulation of Pkhd1 transcription and accompanying cyst formation are observed in renal collecting ducts but not in proximal tubules of HNF-1␤⌬C transgenic mice. The absence of cyst formation and lack of inhibition of Pkhd1 expression in proximal tubules, despite transgene expression in this segment of the nephron, strongly suggests that cyst formation in this model requires down-regulation of Pkhd1 transcription. Proximal tubule-derived cysts are also absent in mice with kidneyspecific deletion of HNF-1␤, and only slight dilatation of proximal tubules is found in A263insGG transgenic mice with the highest levels of transgene expression (22). One possible explanation is that HNF-1␣ expressed in proximal tubules is able to compensate for the loss of HNF-1␤ or the expression of dominant-negative HNF-1␤ mutants in this nephron segment. Expression of HNF-1␣ may prevent the repression of the Pkhd1 promoter and subsequent cyst formation induced by the HNF-1␤⌬C mutant. Consistent with this possibility, HNF-1␣ is expressed in non-cystic proximal tubules of HNF-1␤⌬C transgenic mice. Inhibition of HNF-1␤ has been shown to inhibit HNF-1␣ expression in embryoid bodies, suggesting that HNF-1␤ may be upstream to HNF-1␣ in a transcriptional pathway, but the apparently normal expression of HNF-1␣ indicates that this may not be the case in the kidney. Previous studies have shown that HNF-1␣ can also bind to the Pkhd1 promoter and stimulate its activity (22,25). Further studies will be required to determine whether the ability of the HNF-1␤⌬C mutant to inhibit HNF-1␤ but not HNF-1␣ is because of differences in DNA affinity, transcriptional activity, or recruitment of coactivators such as steroid receptor coactivator-1 or RAC3.