|Home | About | Journals | Submit | Contact Us | Français|
In this study, we describe the generation and partial characterization of Krüppel-like zinc finger protein Glis3 mutant (Glis3zf/zf) mice. These mice display abnormalities very similar to those of patients with neonatal diabetes and hypothyroidism syndrome, including the development of diabetes and polycystic kidney disease. We demonstrate that Glis3 localizes to the primary cilium, suggesting that Glis3 is part of a cilium-associated signaling pathway. Although Glis3zf/zf mice form normal primary cilia, renal cysts contain relatively fewer cells with a primary cilium. We further show that Glis3 interacts with the transcriptional modulator Wwtr1/TAZ, which itself has been implicated in glomerulocystic kidney disease. Wwtr1 recognizes a P/LPXY motif in the C terminus of Glis3 and enhances Glis3-mediated transcriptional activation, indicating that Wwtr1 functions as a coactivator of Glis3. Mutations in the P/LPXY motif abrogate the interaction with Wwtr1 and the transcriptional activity of Glis3, indicating that this motif is part of the transcription activation domain of Glis3. Our study demonstrates that dysfunction of Glis3 leads to the development of cystic renal disease, suggesting that Glis3 plays a critical role in maintaining normal renal functions. We propose that localization to the primary cilium and interaction with Wwtr1 are key elements of the Glis3 signaling pathway.
Gli-similar 1 to 3 (Glis1-3) constitute a subfamily of Krüppel-like zinc finger proteins (4, 25, 27, 28, 30, 39, 56, 57). Glis proteins contain a DNA binding domain consisting of five C2H2-type zinc finger motifs that share a high degree of homology with members of the Gli and Zic subfamilies of transcription factors (1, 24). Glis proteins regulate gene transcription by interacting with a specific nucleotide sequence, referred to as the Glis-DNA binding site (Glis-BS), in the promoter region of target genes (3, 4). Glis1-3 proteins are expressed in a spatial and temporal manner during embryonic development, suggesting that they regulate specific physiological processes (25, 27, 28, 30, 39, 56). Loss of Glis2 function in mice and mutations in GLIS2 have been associated with nephronophthisis (2, 26), while genetic alterations in the GLIS3 gene have been linked to a syndrome characterized by neonatal diabetes and congenital hypothyroidism (NDH) (45, 47).
To obtain greater insights into the physiological functions of Glis3 and its role in disease, we generated Glis3 mutant mice (Glis3zf/zf) in which the fifth zinc finger (ZF5) is deleted. ZF5 is critical for the binding of Glis3 to Glis-BS and therefore for its transcriptional activity (3). We show that Glis3 mutant mice exhibit abnormalities very similar to those displayed by NDH1 patients (45, 47), including a greatly reduced life span and development of polycystic kidneys and neonatal diabetes. These similarities suggest that Glis3zf/zf mutant mice provide an excellent model to study this syndrome.
This study focuses on the cystic renal phenotype of Glis3zf/zf mutant mice. Cystic renal disease represents a heterogeneous group of genetic disorders, characterized by the development of multiple cystic lesions that could involve any segment of the nephron (36, 49). Autosomal dominant polycystic kidney disease (PKD), autosomal recessive PKD, and nephronophthisis are the most studied variants of cystic renal disease. Interestingly, a large number of genes implicated in cystic renal disease encode proteins that are either localized to the primary cilium or are part of a signaling pathway associated with ciliary function (7, 12, 17, 36, 49, 50, 52, 54). These findings led to the hypothesis that dysfunction of the primary cilium and defects in cilium-associated signal transduction pathways are key factors in the etiology of cystic renal disease. Although the precise molecular mechanisms responsible for cyst development have yet to be established, it is thought that changes in cell-matrix and cell-cell interactions, Ca2+ signaling, cell proliferation and differentiation, apoptosis, and cell polarity play critical roles in this process (11, 29, 46).
In this study, we identify two key elements in the Glis3 signaling pathway that are relevant to the development of cystic kidney disease. We demonstrate that Glis3 is associated with the primary cilium, suggesting that activation of Glis3 involves a primary cilium-associated signal pathway. In addition, we show that Wwtr1, a WW domain-containing protein (also named TAZ) that functions as a modulator of several transcription factors (9, 19, 37, 51), interacts with and functions as a coactivator of Glis3. Interestingly, Wwtr1 null mice themselves have been reported to develop cystic renal disease that resembles that with Glis3 (20, 34, 48). Our results indicate that Glis3 and Wwtr1 are part of overlapping transcription regulatory networks that play a critical role in the maintenance of normal renal architecture and function.
Glis3 genomic flanking regions were generated by PCR amplification using 129/Sv genomic DNA as a template. A 4.7-kb XbaI/ClaI fragment of intron 3 and a 3.0-kb BamHI/NotI fragment of intron 4 were inserted into the NheI/ClaI and BamHI/NotI sites of pOSdupdel. The resulting pOSdupdel-Glis3 plasmid DNA was linearized by NotI and electroporated into 129/Sv embryonic stem (ES) cells (Randy Thresher, UNC Transgenic Facility). Gene-targeted ES cells were microinjected into blastocysts from C57BL/6 mice that were then returned to pseudo-pregnant B6D2 mice. Chimeric mice were crossed with C57BL/6 mice to identify transmitting chimeras and to obtain mice heterozygous for the mutant allele. Heterozygous mice were intercrossed to obtain animals homozygous for the mutant allele and to obtain wild-type (WT) littermate controls. All animal studies followed guidelines outlined in the NIH Guide for the Care and Use of Laboratory Animals, and protocols were approved by the Institutional Animal Care and Use Committee at the NIEHS and UNC.
DNA (8 μg) prepared from ES cells and tail biopsy specimens was digested with KpnI or XbaI and separated by electrophoresis in a 1% agarose gel containing 1× Tris-acetate-EDTA buffer. After transfer to an Immobilon-Ny+ membrane (Millipore), DNA was hybridized to [α-32P]dCTP-labeled 5′-end or 3′-end probes, respectively. Routine genotyping was carried out with the following primers: Glis3-F, 5′-AGCTAGTGGCTTTCGCCAACA-3′; Glis3-R, 5′-GAACAAGATAGAATCATGGTATATCC-3′; and Neo-pro, 5′-ACGCGTCACCTTAATATGCG-3′.
Kidneys were fixed in 10% neutral buffered formalin for 24 h, subsequently transferred to 70% ethanol, processed, embedded in paraffin, sectioned at 5 μm, and stained with hematoxylin/eosin. To identify proximal tubules, collecting ducts, distal tubules, or primary cilium, frozen sections were stained with fluorescein isothiocyanate (FITC)-labeled Lotus tetragonolobus lectin (LTA), FITC-labeled Dolichos biflorus agglutinin (DBA) (Vector Laboratory, Burlingame, CA), calbindin-D-28K, or anti-acetylated α-tubulin antibodies (Sigma, St. Louis, MO), respectively. To examine cell polarity, sections were stained with antibodies against epidermal growth factor (EGF) receptor (Sigma), β-catenin (BD Bioscience), ZO-1 (Zymed, San Francisco, CA), or Na-K-ATPase α (Santa Cruz Biotechnology). Alexa Fluor-488- or -594-conjugated antibodies (Molecular Probes, Eugene, OR) were used as secondary antibodies. Fluorescence was observed with a Leica DMRBE microscope (Leica, Wetzlar, Germany). To analyze proliferating cells lining the glomeruli cyst, 50 mg/kg of bromodeoxyuridine (BrdU) was injected into postnatal day 3 (PND3) pups. After 2 h, kidneys were collected and fixed. Sections were selected randomly from three mice for each group and stained with a BrdU antibody. The percentage of BrdU-positive cells was calculated from six to seven glomeruli in each section and plotted.
In situ hybridization on frozen tissue sections was carried out by Phylogeny (Columbus, OH) using a 35S-UTP-labeled, Glis3-specific probe as described previously (28).
Total RNA from cells or tissues was isolated using mini or midi RNA isolation kits (Qiagen, Valencia, CA) according to the manufacturer's instructions (23). In Glis3 knockdown experiments, TKPTS cells were transfected with Glis3 small interfering RNAs (siRNAs) from Ambion (Austin, TX) using Lipofectamine 2000 (Invitrogen, Carlsbad, CA), and 48 h later RNA was isolated. Equal amounts of total RNA were reverse transcribed using a high-capacity cDNA archive kit (Applied Biosystems) and then examined by real-time quantitative PCR (QRT-PCR) analysis. The PCRs were carried out in triplicate in a 7300 QRT-PCR system (Applied Biosystems, Foster City, CA) as previously described (4). All results were normalized to the 18S transcript. Primers were designed using Primer Express 2.0 software and synthesized at Sigma/Genosys (St. Louis, MO). Primers and probes are shown in Table S1 in the supplemental material.
The reporter plasmid pFR-Luc (Stratagene), p-TAL-Luc-(Glis-BS)6, and several p3×FLAG-CMV-Glis3 expression vectors were described previously (3, 4). p3×FLAG-CMV-Glis1 and p3×FLAG-CMV-Glis2 were described previously (25, 27). p3×FLAG-CMV-Glis3M4 containing the point mutations P843A and Y845A was generated using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). pcDNA3.1-Wwtr1(TAZ) was kindly provided by Ikramuddin Aukhil (University of Florida, Gainesville, FL). pCMV-myc-Wwtr1 was generated by PCR amplification using pcDNA3.1-Wwtr1 as a template, and the PCR product was inserted into the EcoRI and KpnI sites of pCMV-myc. The sequence of each insert was verified by restriction enzyme mapping and DNA sequencing.
The subcellular localization of Glis3 and Wwtr1 was examined by confocal microscopy largely as described previously (3). COS-1 cells were plated and transfected with pEGFP-Glis3, pEGFP-Glis3M4 or pEGFP-Glis3ΔZF5 with or without p3×FLAG-CMV-Wwtr1. After 30 h of incubation, cells were fixed and stained with anti-FLAG M2 mouse monoclonal antibody (Sigma) and subsequently with a goat anti-mouse Alexa Fluor-594 antibody (Molecular Probes). Fluorescence was observed in a LSM 510 UV Zeiss confocal microscope.
The mouse kidney proximal tubule cell line TKPTS was kindly provided by Elsa Bello-Reuss (Texas Tech University, Lubbock, TX) and cultured in Dulbecco's minimal essential medium/F-12 medium supplemented with 7% fetal bovine serum and 50 μg/ml insulin. To analyze Glis3 transcriptional activity, HEK293T or COS-1 cells (ATCC) were cotransfected with p-TAL-Luc-(Glis-BS)6, p3×FLAG-CMV-Glis3, p3×FLAG-CMV-Glis3M4, pCMV-myc-Wwtr1, and pCMVβ using Fugene 6 transfection reagent (Roche) as described previously (3). For mammalian two-hybrid analysis, cells were transfected with the reporter plasmid pFR-Luc, pCMVβ, pM-Wwtr1, pVP16-Glis3, or pVP16-Glis3M4 as indicated. Cells were incubated for 30 h and then assayed for reporter activity. Luciferase and β-galactosidase activity were assayed with a luciferase kit (Promega) or a luminescent β-galactosidase detection kit (Clontech). Transfections were performed in triplicate, and each experiment was repeated at least twice.
HEK293 cells were transiently transfected with pCMV-Myc-Wwtr1 and p3×FLAG-CMV-Glis3 expression vectors containing Glis3, Glis3ΔN463, Glis3ΔC756, Glis3M4, or Glis3ΔZF5. Forty-eight hours after transfection, cells were harvested and lysed in radioimmunoprecipitation assay buffer (Upstate, Charlottesville, VA) containing protease inhibitor cocktails I and II (Sigma) as described previously (25). Myc-Wwtr1 protein complexes were then isolated using agarose A and Myc antibody (Invitrogen, Carlsbad, CA). Bound protein complexes were then solubilized in sample buffer and analyzed by Western blot analysis using mouse anti-Flag M2 antibody (Sigma).
Equal amounts of glutathione S-transferase (GST)-Wwtr1 or GST protein were incubated with glutathione-Sepharose 4B beads and then washed in phosphate-buffered saline. [35S]methionine-labeled Glis3 and Glis3M4 were generated using a TNT quick coupled transcription/translation system (Promega). The GST- and GST-Wwtr1-bound beads were then incubated with [35S]methionine-labeled Glis3 in 0.5 ml of binding buffer (20 mM Tris-HCl, pH 7.6, 100 mM KCl, 0.05% Nonidet P-40, 0.1 mM EDTA, 10% glycerol, 0.2% Tween 20, 1 mM phenylmethylsulfonyl fluoride). After 1 h of incubation at room temperature, beads were washed five times in binding buffer and boiled in 15 μl 2× sodium dodecyl sulfate-polyacrylamide gel electrophoresis loading buffer. Proteins were separated by 4 to 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and visualized by autoradiography.
To obtain greater insights into the physiological functions of Glis3 and its role in NDH, we generated mice, referred to as Glis3zf/zf, in which the Glis3 gene was disrupted. We selected a strategy that would result in the deletion of exon 4, which encodes ZF5, and parts of introns 3 and 4 (Fig. (Fig.1A;1A; see Fig. S1 in the supplemental material). Deletion of ZF5 is predicted to disrupt the DNA binding ability and, therefore, the transcriptional activity of Glis3 (3). The latter was confirmed by experiments showing that, in contrast to WT Glis3, Glis3 lacking ZF5 (Glis3ΔZF5) was unable to induce Glis-BS-dependent transcriptional activation of a luciferase reporter (Fig. (Fig.1B).1B). However, in transiently transfected cells, deletion of ZF5 did not affect the ability of Glis3 to translocate to the nucleus (Fig. 1C and D). The loss of transcriptional activity of Glis3ΔZF5 is consistent with our recent observations showing that ZF5 is critical for the interaction of Glis3 with Glis-BS (3).
Glis3+/zf mating pairs produced offspring with all three genotypes with the expected Mendelian distribution, suggesting that deletion of exon 4 did not cause embryonic lethality (data not shown). Homozygous Glis3zf/zf pups were smaller (about 10% reduction in weight; see Fig. S1D in the supplemental material) and displayed a dramatically shortened life span (Fig. (Fig.1E).1E). At PND3, only about 50% of the Glis3zf/zf pups survived, while none of them lived longer than 11 days. The general appearance of PND3 Glis3zf/zf mice was normal and milk bands were observed, suggesting that mice did not appear to exhibit any severe abnormalities that interfered with their ability to drink or swallow (Fig. (Fig.1F).1F). Heterozygous Glis3+/zf mice displayed no obvious phenotype and were indistinguishable from WT mice, as judged by general appearance and life span.
Histopathological analysis of PND3 Glis3zf/zf mice showed that the corticomedullary junction is consistently abnormal in Glis3zf/zf kidneys (Fig. (Fig.2;2; also see Fig. S2 in the supplemental material). In addition to there being multiple, fluid-filled cysts in both the cortex and medulla, the corticomedullary junction appeared less complex and showed an increased abundance of mesenchymal cells. At PND3, all Glis3zf/zf mice developed cystic tubules and glomerular cysts; however, not all tubules and glomeruli were cystic. Male and female Glis3zf/zf mice were about equally affected, while no renal cysts were observed in Glis3zf/+ mice, suggesting the recessive nature of this mutation. The severity of the cystic phenotype was variable, and the size and number of cysts increased with age (Fig. 2B, D, and F). Abnormalities in Glis3zf/zf kidneys were first observed around embryonic day 14.5 (E14.5) to E15.5; dilations of Bowman's spaces were the most obvious (Fig. (Fig.2).2). Cystic glomeruli could often be identified by the presence of glomerular tufts. These cysts became increasingly larger in diameter at PND0 and PND3 (Fig. (Fig.2).2). At PND3, Glis3zf/zf mice did not develop proteinurea (see Fig. S2C in the supplemental material) and no change was observed in urinal pH or ketone levels (data not shown). In addition to polycystic kidneys, Glis3zf/zf mice developed overt neonatal diabetes, evidenced by hyperglycemia and hypoinsulinemia caused by a deficiency in the generation of pancreatic β-cells (data not shown; H. S. Kang, Y.-S. Kim, J. Y. Beak, G. Kilic, B. Sosa-Pineda, J. Jensen, J. Foley, and A. M. Jetten, submitted for publication).
We reported previously that during early rat kidney development, Glis3 is highly expressed in the ureteric bud (28). In situ hybridization analysis revealed that at PND10, Glis3 is most highly expressed in kidney (see Fig. S3A to C in the supplemental material), consistent with the reported Northern blot analysis of Glis3 expression (28). Glis3 expression was detected in the epithelia of the collecting ducts and tubules and in the parietal layer of the glomerulus (see Fig. S3 in the supplemental material). These observations suggest that the cystic phenotype in Glis3zf/zf mice is related directly to alterations in parietal cells and epithelial cells lining the tubules and collecting ducts. Thus, expression of Glis3 overlaps with that of many other genes that have been implicated in renal cystic disease, including Pkd1 and -2, Pkhd1, Wwtr1, and Hnf1β (15, 20, 21, 34, 36, 48-50).
To determine the origins of the cysts, kidneys were stained with LTA, calbindin-D-28K antibody, and DBA, which specifically label the proximal tubules, distal tubules, and the collecting ducts, respectively. At PND3, most of the larger cysts originated from glomeruli, while many proximal and distal tubules and collecting ducts in kidneys of Glis3zf/zf mice were dilated (Fig. 2J to O). Of the total cystic volume, about 65% originated from glomeruli, 4% from collecting ducts, 4% from proximal tubules, and 11% from distal tubules. The origin of the remaining 16% could not be determined.
Increased cell proliferation has been proposed to be involved in cyst formation. We therefore analyzed proliferation in kidneys from WT and Glis3zf/zf mice by BrdU labeling. Because the largest cysts originated from glomeruli, we compared the percentage of BrdU-positive cells lining Bowman's spaces with that of glomerulocysts in Glis3zf/zf mice. The percentage of BrdU-positive cells was significantly higher in the epithelium lining glomerulocysts than that of normal Bowman's spaces (Fig. (Fig.2I).2I). Moreover, immunofluorescent staining showed that cells in normal tubules contained significantly higher levels of the cyclin-dependent kinase inhibitor cdkn1a (p21cip1), which is known to be associated with growth arrest/inhibition, than in cystic tubules (see Fig. S4A in the supplemental material). Together these data suggest that Glis3 dysfunction leads to increased cell proliferation of cells lining Bowman's spaces and renal tubules. Whether the observed increase in proliferation is a consequence or cause of cyst formation needs to be determined further.
A large number of genes implicated in cystic renal disease encode proteins that either localize to the primary cilium or are part of a signaling pathway linked to ciliary function (7, 12, 49, 54). This raised the question of whether Glis3 is also associated with the primary cilium and whether there is a link between the cystic renal phenotype observed in Glis3zf/zf mice and the primary cilium. This idea was strengthened by recent observations demonstrating that another member of the Glis subfamily, Glis2, and the closely related Gli proteins have been reported to be associated with the primary cilium (2, 14, 33). Primary cilia were visualized by immunofluorescent staining with an anti-acetylated α-tubulin antibody. Immunofluorescent microscopic analysis showed that Glis3 localized to the primary cilia of tubule epithelial cells in sections from mouse kidney, as indicated by its colocalization with acetylated α-tubulin (Fig. 3A to C). This was further supported by experiments examining the subcellular localization of Glis3 in cultured mouse kidney proximal tubule TKPTS cells transiently expressing enhanced green fluorescent protein (EGFP)-Glis3. To promote formation of primary cilia in these cells, cultures were maintained at confluence and in low serum for several days. Confocal microscopy showed that EGFP-Glis3 is localized to the primary cilium in addition to the nucleus (Fig. 3D to F; see Fig. S5A to C in the supplemental material). In the primary cilium, the fluorescent staining of EGFP-Glis3 overlapped that of acetylated α-tubulin, and in several instances, Glis3 appeared to localize preferentially to the tip of the primary cilium (Fig. 3D to F). These observations suggest that Glis3 may be part of a primary cilium-associated signaling pathway.
Cystic renal disease has been linked to various abnormalities in the formation of the primary cilium or ciliary function (29, 46, 54). Therefore, we examined whether the formation of primary cilia was affected in Glis3zf/zf mice. Most epithelial cells lining the tubules and collecting ducts of kidneys of PND3 WT mice contained a primary cilium as indicated by staining for acetylated α-tubulin (Fig. (Fig.3G).3G). In Glis3zf/zf mice, the percentage of cells with cilium lining normal renal tubules was not significantly altered compared to that in WT mice; however, this percentage was greatly decreased in dilated tubules and further reduced in cysts (Fig. 3H to J).
Next, we examined the epithelial lining of renal cysts in Glis3zf/zf mice by scanning electron microscopy. This analysis confirmed that many cells in renal cysts contained a normal primary cilium; however, a significant number of cells lacked a primary cilium and a few contained a truncated primary cilium (Fig. (Fig.3K).3K). These observations are consistent with our conclusion that Glis3 dysfunction leads to a significant reduction in renal epithelial cells possessing a primary cilium. However, many cells still contained what appeared to be a structurally normal cilium. These results suggest that Glis3 is not essential for the formation of the primary cilium. Consistent with this is the observation that TKPTS cells in which Glis3 expression was downregulated by Glis3 siRNAs are still able to form a primary cilium (see Fig. S5D in the supplemental material). The reduction in cells with a primary cilium may be due to an indirect mechanism, such as increased proliferation or changes in cell-cell interactions.
The development of renal cysts has been reported to be associated with changes in cell-cell and cell-matrix interactions and cell polarity (36, 46, 49). We therefore compared the localizations and expression level of β-catenin, zona occludens 1 (ZO-1), and EGF receptor (EGFR), proteins that either are part of junctions or function as markers of cell polarity in the kidneys of WT and Glis3zf/zf mice. In normal tubules, β-catenin localized to the basolateral membrane (Fig. 4A to C), but in regions of cystic distal tubules of Glis3zf/zf mice, it was found to a significant degree in the cytoplasm as well as in the apical membrane (Fig. 4D to F). ZO-1, which in normal distal tubules localized at tight junctions in the apical membrane (Fig. 4G to I), was highly localized in the basolateral membrane in Glis3zf/zf cystic tubules (Fig. 4J to L). In some PKD models, including Kif3a mutant mice (32), alterations in cell polarity are associated with localization of the EGFR to the apical membrane instead of the basal membrane; however, in Glis3zf/zf mice, EGFR remained largely confined to the basal membrane (Fig. 4M to R). Na+-K+ ATPase α, which is normally restricted to the basal membrane, exhibited a basolateral pattern of subcellular distribution in tubules of Glis3zf/zf kidneys (see Fig. S4B in the supplemental material). These results suggest that Glis3 dysfunction results in changes in cell-cell interactions. This was supported by experiments examining the effect of Glis3 knockdown on the distribution pattern of EGFR, ZO-1, and E-cadherin in TKPTS cells. All three markers appeared absent or were weaker at the leading edge of migrating control TKPTS cells than in TKPTS cells in which Glis3 was downregulated (see Fig. S6 in the supplemental material). Interestingly, expression of polycystin-1 (PC-1, which is encoded by Pkd1) in MDCK cells has been reported to induce a pattern of ZO-1 and E-cadherin distribution that is very similar to that of control TKPTS cells expressing Glis3 (8). Thus, expression of Glis3 or Pkd1 has similar effects on the distribution of several junctional proteins, including ZO-1 and E-cadherin.
Since Glis3 functions as a regulator of gene transcription, it is likely that the formation of cysts in Glis3zf/zf mice is caused by changes in the expression of specific Glis3 target genes. We therefore compared the expression levels of several genes previously reported to be associated with cystic renal disease and ciliary function in kidneys of WT and Glis3zf/zf mice. The expression of most of the genes examined, including Pkhd1, Nphp1-4, Kif3α, Umod1, Pkd2, Wwtr1, and Hnf1β, was not significantly altered in kidneys of Glis3zf/zf mice (Fig. (Fig.5A).5A). In contrast, the expression of Dctn5 and Pkd1 was decreased (by 60 and 20%, respectively) in kidneys of Glis3zf/zf mice. The downregulation of Dctn5 and Pkd1 was confirmed in TKPTS cells in which Glis3 expression was downregulated by Glis3-specific siRNA (Fig. (Fig.5B5B).
The cystic phenotype of Glis3zf/zf mice exhibit a number of similarities with cystic renal diseases in humans and in transgenic mice caused by mutations in other genes, including HNF1β, Wwtr1, and PKHD1 (15, 20, 21, 34, 35, 48). The similarity between the renal phenotype of Glis3zf/zf and Wwtr1−/− mice was particularly intriguing because cysts originating from glomeruli are prominent in the kidneys of both mutant mice. In addition to the apparent similarity in renal phenotype, Wwtr1 and Glis3 expression in the kidney overlap each other, and both Wwtr1 and Glis3 have been reported to have a stimulatory role in osteogenesis while inhibiting adipogenesis (4, 19, 20, 28, 34). This comparison raised the question of whether there was a functional link between the two proteins. This idea was strengthened by reports showing that Wwtr1 acts as a modulator of gene transcription by interacting with several transcription factors (9, 18, 38, 42). These interactions are mediated through the WW domain of Wwtr1, which recognizes a P/LPXY motif (in which X is any amino acid) in these transcription factors (22). We hypothesized that Wwtr1 might interact with Glis3 and enhance its transcriptional activity by acting as a coactivator of Glis3. This concept was supported by examination of the Glis3 sequence, which identified four potential P/LPXY motifs in Glis3, three at the N terminus before the zinc finger domain, and one at its C terminus (Fig. (Fig.6A).6A). To determine whether Wwtr1 functions as a transcriptional mediator of Glis3, we first examined the interaction between Glis3 and Wwtr1 by coimmunoprecipitation analysis. These results showed that Glis3 was able to coimmunoprecipitate Wwtr1, suggesting that Glis3 and Wwtr1 are part of the same protein complex (Fig. (Fig.6B).6B). To establish which, if any, of the putative P/LPXY motifs are involved in the interaction of Glis3 with Wwtr1, we examined the effect of several deletion mutations within Glis3 on its interaction with Wwtr1. As shown in Fig. Fig.6B,6B, although deletion of the N terminus, including three of the four P/LPXY motifs of Glis3, had some effect on the interaction with Wwtr1, deletion of the C terminus, including the fourth motif, abolished this interaction almost completely, indicating a major role for the C terminus in the interaction of Glis3 with Wwtr1. Moreover, these observations suggested that of the four motifs, the fourth P/LPXY motif at the C terminus is the most important for this interaction. This was supported by data showing that mutant Glis3M4, which contains the P841A, Y843A double point mutation in the C-terminal P/LPXY motif, was unable to interact effectively with Wwtr1. Deletion of the ZF5 motif, which matches the mutation in Glis3zf/zf mice, did not greatly affect the interaction of Glis3 with Wwtr1 (Fig. (Fig.6B),6B), indicating that Glis3ΔZF5 can interact with Wwtr1.
The interaction between Glis3 and Wwtr1 was further investigated by mammalian two-hybrid analysis in 3T3-L1 cells transfected with a (UAS)5-Luc reporter, pM-Wwtr1, and pVP16-Glis3 plasmid DNA. As shown in Fig. Fig.6C,6C, cotransfection of pM-Wwtr1 with increasing amounts of pVP16-Glis3 significantly enhanced the transcriptional activation of the Luc reporter in a concentration-dependent manner, whereas cotransfection with pVP16-Glis3M4 had little effect. These data support the conclusion that the interaction of Glis3 and Wwtr1 is dependent on the C-terminal P/LPXY motif.
Subsequently, we examined the interaction of Glis3 with Wwtr1 by in vitro pull-down analysis using purified GST-Wwtr1 fusion protein and 35S-labeled Glis3. This analysis showed significant binding of Glis3 to Wwtr1, while GST alone did not bind Glis3 (Fig. (Fig.6D).6D). To examine whether the C-terminal P/LPXY motif of Glis3 was required for this interaction, GST pull-down analysis was also performed with the Glis3M4 mutant. As shown in Fig. Fig.6D,6D, Glis3M4 did not interact efficiently with GST-Wwtr1. These results are in agreement with the conclusion that the fourth P/LPXY motif plays a major role in the interaction between Glis3 and Wwtr1.
Next, we examined the subcellular localization of Glis3 and Wwtr1 in COS-1 cells. In virtually all cells (95%) transfected with pEGFP-Glis3 or pEGFP-Glis3M4 only, Glis3 protein localized predominantly to the nucleus (Fig. (Fig.6E).6E). In contrast, in cells transfected with 3×FLAG-CMV-Wwtr1 only, Wwtr1 localized predominantly to the cytoplasm (58%) or was equally distributed between the nucleus and cytoplasm (30%) and localized predominantly to the nucleus in only 12% of the cells. Coexpression of both EGFP-Glis3 and FLAG-Wwtr1 had little impact on the distribution of Glis3. However, the percentage of cells in which Wwtr1 predominantly localized to the nucleus increased dramatically (from 12 to 61%), while a concomitant decline was observed in the percentage of cells in which Wwtr1 predominantly localized to the cytoplasm (Fig. (Fig.6F).6F). In contrast to what occurred with EGFP-Glis3, coexpression of EGFP-Glis3M4 and Wwtr1 did not change the subcellular distribution of Wwtr1. These observations are consistent with the concept that Glis3 and Wwtr1 interact with each other in a P/LPXY-dependent manner and that this interaction promotes localization or retention of Wwtr1 to the nucleus (Fig. 6E and F). Glis3ΔZF5, which is predominantly localized to the nucleus, was also able to enhance the nuclear localization of Wwtr1, supporting our conclusion that it is able to interact with Wwtr1 (Fig. 6E and F).
Previous studies (3, 28) demonstrated that Glis3 can function as an activator of transcription by binding a specific DNA sequence (Glis-BS) in the regulatory regions of target genes. To determine whether Wwtr1 is able to modulate Glis3-regulated gene transcription, we analyzed the effect of Wwtr1 on (Glis-BS)6-dependent transcriptional activation of the Luc reporter by Glis3. Figure Figure7A7A shows that increased Wwtr1 expression significantly enhanced the induction of Luc reporter activity by Glis3, indicating that Wwtr1 functions as a coactivator of Glis3. The closely related Yes kinase-associated protein was also able to enhance Glis3 activity (data not shown).
We previously showed that the transactivation domain of Glis3 is located within the 230 amino acids at its C terminus, a region that encompasses the C-terminal P/LPXY motif (Fig. (Fig.6A)6A) (3, 28). We therefore examined the effect of the P841A, Y843A double point mutation on the transcriptional activity of Glis3. As shown in Fig. Fig.7A,7A, these point mutations totally abolished the transcriptional activity of Glis3, indicating that this motif is critical for the transcriptional activity of Glis3 and part of its transactivation domain.
To determine whether Wwtr1 was able to interact with other members of the Glis subfamily, we analyzed the sequence of Glis1 and Glis2 for the presence of P/LPXY motifs. This analysis identified two putative PPXY motifs in Glis1, while no such consensus motif was present in Glis2. First, we compared the effects of Wwtr1 on the transcriptional activities of Glis1-3. Figure Figure7B7B shows that Wwtr1 had only a minor effect on Glis1 transcriptional activity, whereas it had no apparent effect on the transcriptional activity of Glis2. These observations suggest that Wwtr1 functions only as an effective coactivator of Glis3 and does not significantly affect the transcriptional activity of Glis1 or Glis2. Next, we examined the interaction between Wwtr1 and Glis1-3 by coimmunoprecipitation analysis. These data showed that Wwtr1 formed a complex with Glis3 but not with Glis1 or Glis2 (see Fig. S7 in the supplemental material).
Recently, mutations in human GLIS3 have been linked to NDH (45, 47). In addition to neonatal diabetes and congenital hypothyroidism, NDH is associated with facial anomalies, PKD, congenital glaucoma, and liver fibrosis. To further study the role of Glis3 in this disease, we generated for the first time mutant mice that are impaired in Glis3 function. Glis3 mutant mice develop polycystic kidneys and neonatal diabetes and exhibit a very short life span. The development of hyperglycemia and hypoinsulinemia, caused by an insufficiency of pancreatic β cells (Kang et al., submitted), rather than the development of polycystic kidneys, which at PND3 is rather moderate in severity, may be the major cause of the shortened life span in Glis3zf/zf mice. Whether these mice develop hypothyroidism, glaucoma, and liver fibrosis has yet to be determined. Development of polycystic kidneys was reported only for NDH patients with the most severe abnormalities (NDH1 patients) (45). As with Glis3zf/zf mice, NDH1 patients have a very short life span (10 days to 16 months) and neonatal diabetes. NDH1 patients have a frameshift mutation that results in a loss of the C-terminal activation domain of GLIS3 (3, 45). Thus, the phenotype of Glis3zf/zf mutant mice appears to be the most similar to the abnormalities observed in NDH1 patients. This analogy suggests that Glis3zf/zf mutant mice will provide an excellent model to study this syndrome at a molecular and mechanistic level.
Development of renal cysts in Glis3zf/zf mice was observed as early as E15.5 and increased with age. Cyst formation originated from glomeruli, tubules, and collecting ducts, corresponding to the observed expression of Glis3 in kidney in parietal cells and epithelial cells lining the tubules and collecting ducts. Thus, Glis3 expression overlaps that of many other genes implicated in PKD. Study of cystic renal diseases in humans and in transgenic mice revealed that many proteins implicated in these diseases localize to the primary cilium and are either a structural component of the primary cilium or function as part of a signal transduction pathway associated with the primary cilium (36, 49, 53, 54). These observations led to the hypothesis that ciliary dysfunction plays a key role in the etiology of cystic renal disease. This link raised the possibility that the development of renal cysts in Glis3zf/zf mice may also have a connection to the primary cilium. This hypothesis was supported by our data showing that Glis3 is associated with the primary cilium in renal tubules and in confluent TKPTS renal tubule epithelial cells expressing EGFP-Glis3. Interestingly, the family member Glis2, which has been implicated in nephronophthisis, an autosomal recessive cystic kidney disease (2, 7, 26), was also found to be associated with the primary cilium. Moreover, members of the closely related Gli family have been reported to localize to the primary cilium as well (14, 40, 44). The primary cilium plays a critical role in the activation of the sonic hedgehog (Shh)/Gli signal transduction pathway. In the absence of Shh, its receptor Patched-1 (Ptch1) prevents the accumulation of smoothened (Smo) in the primary cilium. Binding of Shh to Ptch1 inactivates the receptor and results in the activation of Smo, which then accumulates within the cilium and activates Gli. Subsequently, Gli translocates to the nucleus, where it regulates the transcription of target genes. In addition to Shh/Gli, the primary cilium plays a critical role in several other signal transduction pathways (10, 12, 54). Its association with the primary cilium suggests that Glis3 may be part of a cilium-mediated signal transduction pathway and requires activation before it is translocated to the cytoplasm and nucleus (Fig. (Fig.8).8). Because Shh does not affect Glis3-mediated transactivation (our unpublished observations), regulation of Glis3 activity likely involves another signal.
Although many proteins associated with the primary cilium have been implicated in the development of cystic renal disease, the precise molecular mechanisms underlying these diseases have not yet been fully established. Planar cell polarity and oriented cell division have been reported to play an important role in the postnatal development of nephrons, while misoriented cell division appears to be part of the mechanism leading to cystogenesis (6, 11, 41, 43, 46). It has been proposed that the primary cilium may guide oriented cell division (11, 41, 43). Our observations indicate that Glis3 dysfunction or Glis3 knockdown does not prevent the formation of the primary cilium, suggesting that Glis3 does not regulate the expression of an essential structural component of the cilium. However, dilated and cystic tubules in Glis3zf/zf mice contain relatively fewer cells with a primary cilium. Because cell proliferation negatively affects the formation of primary cilia, this decrease might involve an indirect mechanism and be due to increased proliferation, as shown by the increase in BrdU-positive cells and the reduced level of cdkn1a observed in renal cysts of Glis3zf/zf mice. In addition to increased proliferation, renal tubule epithelial cells in Glis3zf/zf mice exhibit changes in cell-cell and cell-matrix interactions, as indicated by alterations in the distribution of β-catenin, ZO-1, and Na+-K+ ATPase α. Changes in the distribution patterns of EGFR, ZO-1, and E-cadherin were also observed in mouse proximal tubule TKPTS cells in which Glis3 was downregulated. Interestingly, the effects of Glis3 on junctional proteins resemble those of PC-1. Expression of PC-1 (Pkd1) in MDCK cells induces cell migration and reabsorption of ZO-1, E-cadherin, and β-catenin at the leading edge of migrating cells, while Pkd1−/− mouse embryo fibroblasts exhibit a reduced migratory capability (8). It was concluded that PC-1 might be a regulator of epithelial plasticity by controlling cell polarity, cell migration, and cell-cell and cell-matrix interactions, functions important in the formation, elongation, and maintenance of renal tubules. Such a mechanism may also play a role in the control of renal functions by Glis3. Interestingly, the observation that Dctn5 and Pkd1 expression were decreased in the kidneys of Glis3zf/zf mice and in Glis3 downregulated TKPTS cells would be consistent with this hypothesis (Fig. (Fig.5).5). Also notable was that in Glis3zf/zf mice, pancreatic tubules were also found to be dilated, suggesting a common mechanism for renal and pancreatic cyst formation (Kang et al., submitted). Whether such a common mechanism is responsible for the insufficiency in pancreatic β cells observed in Glis3zf/zf mice needs further study.
Because of the significant similarities between different cystic renal phenotypes and the connection between ciliary proteins and cystic renal diseases (49, 53), it is not surprising that functional links are being found between some of the proteins implicated in these diseases. For example, PKHD1 is a target gene of hepatocyte nuclear factor HNF-1β, and mutations in both genes have been implicated in cystic renal disease (5, 13, 16, 36, 49). In this study, we identify a novel link between Glis3 and the transcriptional modulator Wwtr1. In both Glis3 mutant mice and Wwtr1 null mice, the development of glomerulocysts are prominent (20, 34, 48) and both Glis3 and Wwtr1 promote osteogenesis and repress adipogenesis (4, 19). Wwtr1 can function as a corepressor as well as a coactivator (19). It represses peroxisome proliferator-activated receptor γ (PPARγ)-mediated transcription while it enhances transcriptional activation by Cbfa1/Runx2, T-box transcription factor 5 (TBX5), paired box homeotic gene 3 (Pax3), and thyroid transcription factor 1 (TTF1) (9, 18, 38, 42). Wwtr1 interacts through its WW domain directly with these transcription factors by recognizing a P/LPXY motif. Glis3, which functions as a positive regulator of Glis-BS-dependent transcription (3, 4, 28), contains four putative P/LPXY motifs (Fig. (Fig.6A).6A). These observations raised the possibility that Wwtr1 and Glis3 might interact with each other and that Wwtr1 might function as a coactivator of Glis3-mediated transcription. Coimmunoprecipitation and mammalian two-hybrid analyses demonstrated that Wwtr1 and Glis3 are part of the same protein complex, while in vitro pull-down analysis indicated that Glis3 interacts with Wwtr1 directly. Deletion of its N terminus affected the interaction of Glis3 with Wwtr1 to some extent, suggesting that the N terminus may contribute to the interaction either through direct binding or through the mediation of other proteins within the Glis3-Wwtr1 complex. Deletion of the C terminus, containing the fourth P/LPXY motif, or mutation of the C-terminal motif (Glis3M4) abolished the interaction of Glis3 with Wwtr1 almost completely. These data indicate that the P/LPXY motif is a major requirement for the interaction and are consistent with the conclusion that Wwtr1 recognizes Glis3 directly via this motif. We further showed that increased expression of Wwtr1 significantly enhanced Glis3-mediated transcriptional activation, consistent with the concept that it acts as a coactivator of Glis3. Although the family member Glis1 contains two putative P/LPXY motifs, Wwtr1 did not interact with Glis1 and did not significantly enhance Glis1 transcriptional activity. Glis2, which does not contain any P/LPXY motif, also did not interact with Wwtr1. These observations indicate that the interaction of Glis3 with Wwtr1 is specific and that Wwtr1 functions as an effective coactivator of Glis3 but not of Glis1 or Glis2. We previously showed that the transcriptional activation domain is located in the C-terminal region of Glis3, a region that includes the fourth P/LPXY motif (3, 28). This, together with the observation that this motif is required for the interaction with the coactivator Wwtr1, suggested that this motif may be part of the transactivation function of Glis3. This conclusion was supported by data showing that mutations in this motif (PPHY to PAHA) greatly diminished the transcriptional activity of Glis3 (Fig. (Fig.7A).7A). This is consistent with the finding that a base insertion in the GLIS3 gene of NDH1 patients results in a frameshift and that deletion of the C terminus, including the fourth P/LPXY motif, causes loss of Glis3 transcriptional activity (3, 45). Because Glis3ΔZF5, which matches the mutation in Glis3zf/zf mice, can interact with Wwtr1 but is unable to bind Glis-BS, the development of the renal and pancreatic phenotype in Glis3zf/zf mice is related to its inability to induce GLIS-BS-dependent transactivation and not due to lack of Wwtr1 interaction.
The interaction between Wwtr1 and Glis3 was further supported by observations showing that coexpression of Glis3 with Wwtr1 promoted the nuclear localization of Wwtr1. Previous studies reported that Wwtr1 shuttles in and out of the nucleus (22, 31, 51, 55). This shuttling seems to be controlled at several levels. In the cytoplasm, phosphorylation of Wwtr1 by the Hippo kinase signaling cascade promotes interaction with 14-3-3 proteins and retention in the cytoplasm. In the nucleus, Wwtr1 can interact with several transcription factors (9, 18, 38, 42, 51). It has been suggested that competition may exist between the transcriptional machinery and 14-3-3 proteins for Wwtr1 binding. Interaction of Glis3 with Wwtr1 may result in nuclear retention of Wwtr1 (Fig. (Fig.6E),6E), as has been proposed for the interaction between Wwtr1 and the transcription cofactor ARC105 (51). However, we cannot rule out that Glis3 forms a complex with Wwtr1 in the cytoplasm and then translocates to the nucleus.
In summary, this study demonstrates that Glis3 mutant mice have a very short life span and develop polycystic kidneys and neonatal diabetes, abnormalities that are very similar to those observed in patients with NDH1. Therefore, these mice provide an excellent model to study the molecular mechanisms underlying this syndrome. We further identify two important elements of the Glis3 signal transduction pathway. We show that Glis3 localizes to the primary cilium and propose that upon activation of Glis3 by a primary cilium-associated signaling pathway, Glis3 is transported by intraflagellar transport into the cytoplasm and subsequently into the nucleus (Fig. (Fig.8).8). In the nucleus, Glis3 interacts with several coactivators, including Wwtr1, that mediate the transcriptional activation of Glis3 target genes. Thus, Glis3 and Wwtr1 are part of overlapping transcription regulatory networks that are critical in maintaining normal renal structure and homeostasis.
We thank Gregory Germino, Johns Hopkins University, for his advice on the polycystic kidney phenotype; Darlene Dixon and Tina Teng, NIEHS, for their comments on the manuscript; and Randy Thresher, UNC Transgenic Facility, and Laura Miller, NIEHS, for their support and advice with generating the mutant mice.
This research was supported by the Intramural Research Program of the NIEHS (Z01-ES-101485).
We declare no conflicts of interest.
Published ahead of print on 9 March 2009.
‡Supplemental material for this article may be found at http://mcb.asm.org/.