LLC-PK1 cell growth is repressed by WT1 inhibition of G-protein alpha i-2 protooncogene transcription.

The temporal expression of the early growth response gene (EGR-1) is one molecular mechanism for both maximal activation of the G alpha i-2 gene and accelerated growth in mitotically active predifferentiated LLC-PK1 renal cells. These events are dependent on an enhancer area in the 5'-flanking region of the G alpha i-2 gene that contains an EGR-1 motif (5'-CGCCCCCGC-3'). However, acquisition of the polarized phenotype in LLC-PK1 cells is accompanied by loss of EGR-1 expression and occupancy of the EGR-1 site by nuclear binding proteins other than EGR-1. We now demonstrate that one of these binding proteins is the Wilms' tumor suppressor (WT1). Furthermore, the temporal expression of WT1 in LLC-PK1 cells acquiring the polarized phenotype represses both G alpha i-2 gene activation and growth in these cells. These findings suggest the existence of differentiation-induced pathways in LLC-PK1 cells that alternatively abrogates EGR-1 and promotes WT1 gene expression, thereby modulating a target protooncogene G alpha i-2 that is participatory for growth and differentiation in renal cells. These studies emphasize the usefulness of the LLC-PK1 renal cell as a model to elucidate normal programs of genetic differentiation in which WT1 participates.

Wilms' tumor (WT) 1 is a pediatric nephric neoplasm arising from the continued proliferation of embryonic blastemal cells that fail to differentiate (1). WT occurs in both sporadic and hereditary forms and as part of the WAGR syndrome (Wilms' tumor, aniridia, genitourinary abnormalities, and mental retardation) (2). Chromosome 11p13 contains a region encoding the tumor suppressor gene WT1, which is commonly deleted in heritable WT (3,4). Internal deletions and mutations of the WT1 gene are found in some WTs (reviewed in Ref. 5). Although WT1 is expressed in uterus, spinal cord, spleen, abdominal wall musculature, and the mesothelial lining of organs within the thoracic cavity, its highest expression is in the developing urogenital system (6). The WT1 gene encodes four splice variants of a transcription factor containing a glutamine/ proline-rich N terminus and four carboxyl Cys 2 His 2 type zinc fingers, which recognize 5Ј-GCGGGGGCGGTG-3Ј and redundant 5Ј-TCC-3Ј cis motifs (7)(8)(9). Alternative splice I allows the variable insertion of exon 5 between the transactivating and DNA binding domains of WT1. Alternative splice II allows the variable insertion of nine nucleotides encoding lysine, threonine, and serine (ϩKTS or ϪKTS) between the third and fourth zinc fingers (7). WT1 cis motifs are also recognized by another Cys 2 His 2 type zinc finger transcription factor EGR-1 also known as NGF1-A, Krox 24, TIS-8, and Zif 268, which is induced after mitogenic and differentiation cues in several cell types (reviewed in Ref. 10). In contrast to EGR-1, WT1 appears to predominantly function as a transcriptional repressor of several growth-related genes including EGR-1, insulin-like growth factor II, platelet-derived growth factor ␣ chain, insulin-like growth factor I receptor, colony-stimulating factor 1, transforming growth factor ␤1, and Pax2 (reviewed in Ref. 11). WT1 mutations prevent DNA binding or transcriptional repressive functions that are postulated to allow the constitutive expression of growth factors leading to renal neoplasia. Consistent with this possibility, suppression of colony formation occurs in WT cells transfected with wild type WT1 isoforms (12).
Further insights into molecular cascades linking WT1 to target genes in renal cells would clearly be advantageous. WTs occur with greater frequency in pediatric kidneys containing persistent renal stem cells, a condition designated nephroblastomatosis or nephrogenic rests. WT1 mutations occur in these rests, suggesting they may represent a transitional cell preceding malignancy (13). Embryonal kidney cell tumors reminiscent of WT can be induced in rats given the alkylating agent N-nitroso-NЈ-methyl urea (14). However, cultured cells have yet to be developed from these sources. We have previously identified genes participatory for epithelial cell growth and differentiation events in LLC-PK 1 cells, an extensively characterized cultured epithelial cell line derived from juvenile male pig kidney (15). These studies indicated that as in lower eukaryotic organisms such as Dictyostelium (16 -18) and Drosophila (19), heterotrimeric guanine nucleotide binding (G) proteins are involved in signal transduction pathways required for both growth and cellular differentiation programs in renal cells.
G proteins are composed of individual ␣, ␤, and ␥ subunits that are encoded by gene superfamilies that have been conserved by eukaryotes throughout evolution (reviewed in Ref. 20). Most of the transducing activities of G proteins in mammalian cells are associated with the state of activation of the ␣ subunit, which is involved in GDP/GTP exchange and GTP hydrolysis (reviewed in Refs. 21 and 22). G proteins can alter cell growth or differentiation by participation in growth factor receptor signaling pathways that converge in the nucleus to alter gene expression. Mutations in G␣ i-2 comparable with Ha-ras GP21, which decrease GTPase activity, are found in tumors of the adrenal cortex and ovary (23). Such mutations, which convert the G␣ i-2 gene into the oncogene gip2 induce increased growth and oncogenic transformation in Rat-1a cells (24). Increased growth may be a consequence of persistent activation of pathways coupled to mitogen-activated protein kinase (25). The G␣ i-2 subunit interacts with pathways required for differentiation of F9 teratocarcinoma cells (26). Even modest repression of G␣ i-2 expression is associated with renal developmental and morphologic abnormalities in transgenic mice, underscoring its important role in renal differentiation events (27). Polarized LLC-PK 1 cells contain two G␣ i isoforms, G␣ i-2 and G␣ i3 , which are involved, respectively, in the regulation of hormone-stimulated adenylyl cyclase and constitutive proteoglycan secretion through the Golgi complex (28,29). The genes encoding both G␣ i subunits are transcriptionally activated in these cells in a coordinated manner during growth and differentiation but differ in response to glucocorticoids and cAMP (30 -32). We recently determined in mitotically active predifferentiated LLC-PK 1 cells that the temporal expression of EGR-1 is one molecular mechanism for both maximal activation of the G␣ i-2 gene and accelerated growth in these cells. These events were dependent on an enhancer area in the 5Јflanking region of the G␣ i-2 gene that contains an EGR-1 motif (5Ј-CGCCCCCGC-3Ј) (33). Notably, acquisition of the polarized phenotype in LLC-PK 1 cells was accompanied by loss of EGR-1 expression and occupancy of the EGR-1 site by nuclear binding proteins other than EGR-1. In the present study we determine whether one of these binding proteins is WT1.

Cell Culture Cells
Wild type LLC-PK 1 cells are a polarized epithelial cell line derived from pig kidney. Cells were grown as confluent monolayers and maintained in Dulbecco's modified Eagle's medium containing 10 or 0.1% fetal calf serum in a 5% CO 2 atmosphere as described previously (15). Cells were plated at a density of 1 ϫ 10 6 /10 cm 2 , achieving confluence at approximately culture day 7.

Cellular Transfections
Plasmids-pRSV (WT1) and pRSV (G␣ i-2 ) are plasmids containing, respectively, the entire coding sequence of human WT1 and the entire coding sequence of rodent G-protein ␣ i-2 subunit driven by a Rous sarcoma virus promoter. Two plasmids m-14 LUC or m-4 LUC containing or lacking a putative EGR-1 binding sequence (5Ј-CGCCCCCGC-3Ј) were generated from nested deletions of plasmid 10 -4 LUC (XmaIII), which contains an ␣ i-2 5Ј-flanking sequence fused to a firefly luciferase reporter gene as described previously (30). A third plasmid-mutated m-14 LUC that replaced the nucleotides 1-5 of the EGR-1 site with adenosines was generated by polymerase chain reaction mutagenesis as described previously (33) utilizing the mutagenizing primer, 5Ј-CGGGCTACGAGATCCGCCAAAAACCGCCGTCGGGCAGCGGAG-3Ј. Plasmids containing this DNA segment were confirmed by dideoxynucleic acid sequencing.
Transient Transfections-Plasmids were transfected in equimolar amounts into LLC-PK 1 cells by calcium phosphate precipitation as described previously (30). Optimum transfection efficiency was obtained by the addition of 20 g of total plasmid DNA/55-cm 2 p10 plate (Falcon) followed by incubation for 20 h without glycerol shock. When required, this amount of DNA was achieved by the addition of "carrier plasmid" Bluescript II KSϩ. Transfection efficiency was normalized by co-transfection with 2.5 g of pSV2Apap, a plasmid carrying a human placental alkaline phosphatase reporter gene driven by a Rous sarcoma virus promoter (generously provided by T. Kadesch, University of Pennsylvania).
Transfection Assays-Forty-eight to 98 h after transfection, LLC-PK 1 cells were washed twice in phosphate-buffered saline (without calcium or magnesium) and then lysed by addition of 1.0 ml of lysis buffer A (1% Triton, 25 mM glycylglycine, pH 7.8, 15 mM MgSO 4 , 4 mM EGTA, and 1 mM fresh dithiothreitol). Scraped lysates were transferred to Eppendorf microfuge tubes and centrifuged at 10,000 ϫ g for 5 min at 4°C. The supernatants were transferred to fresh Eppendorf tubes and briefly vortexed prior to each assay. In some experiments the cell number of each plate was determined by direct cell count of trypsinized cells utilizing inverted phase microscopy.
Firefly Luciferase and Human Placental Alkaline Phosphatase Assays-These were performed as described previously (30). Results are expressed as percent increase Ϯ S.E. in luciferase activity normalized for heat-insensitive alkaline phosphatase activity. Data were analyzed by the paired Student's t test.
Protein Assay-This was performed by the dye binding assay of Bradford as described by the manufacturer (Bio-Rad).

Mobility Shift Assays
Nuclear Extract Preparation-Nuclear proteins were extracted from LLC-PK 1 cells as described previously (33).

RNA Gel Blots
RNA from LLC-PK 1 cells was separated by electrophoresis in 1.0% formaldehyde/agarose gels and then transferred to GeneScreen Plus membranes (DuPont NEN). Membranes were prehybridized for 2 h at 42°C in the presence of 1% SDS, 1 M NaCl, 10% dextran sulfate, and 50% deionized formamide. Hybridization was performed under similar conditions for 24 h in the presence of a human WT1 cDNA labeled with [␣-32 P]ATP by priming with random hexamers followed by extension of these primers with the Klenow fragment of DNA polymerase (30). After hybridization, membranes were washed twice for 30 min each in 0.3 M NaCl, 0.03 M sodium citrate at 23°C, then in the same buffer with 1% SDS at 65°C, followed by 15 mM NaCl, 1.5 mM sodium citrate at 65°C. The membranes were dried and autoradiographed with Kodak XAR film at Ϫ80°C for 6 -96 h with or without Cronex Lightning Plus intensifying screens. Quantification of hybridization signals was performed by densitometry of the autoradiograms with an LKB ultroscan XL enhanced laser densitometer.

Immunoblotting and Immunofluorescence of EGR-1 and WT1
LLC-PK 1 cells were washed twice in phosphate-buffered saline (without calcium or magnesium) and then lysed by addition of 1.0 ml of lysis buffer A. Scraped lysates were solubilized by boiling in sample buffer (1% SDS, 30 mM Tris, pH 6.8, 12% glycerol) and loaded onto a 10% acrylamide gel with 150 g of protein loaded per lane. Following SDSpolyacrylamide gel electrophoresis, proteins were transferred onto Immobilon membrane (Millipore), and the membrane was then stained with Coomassie Blue to ensure that all lanes contained equivalent amounts of transferred protein. The destained membrane was then blocked in blotting buffer (5% nonfat dry milk in 20 mM Tris, pH 7.4, with 0.15 M NaCl and 1% Triton X-100), incubated with either preimmune or immune rabbit IgG anti-EGR-1 Wi 21 (alpha 1) (which detects the non-zinc finger region of the EGR-1 protein corresponding to residues 29 -117) or rabbit polyclonal ␣6F or murine monoclonal H7 antibody (which detects the 173-residue amino-terminal non-zinc finger region of the WT1 protein) diluted 1/1000 in blotting buffer and washed.
In other experiments cell lysates were initially reacted with these antisera followed by precipitation with protein A prior to electrophoresis as described previously (8). EGR-1 and WT1-bound proteins were reacted with an enhanced chemiluminescent detection system as described by the manufacturer (Amersham Corp.) followed by autoradiography.
LLC-PK 1 cells plated on glass coverslips were fixed for immunofluorescent staining on Days 1-7. Cells were fixed in 4% paraformaldehyde for 1 h, permeabilized in Triton X-100 for 4 min, and then incubated in PBS containing 0.1% bovine serum albumin for 5 min to reduce nonspecific background staining. The cells were incubated for 2 h in anti-EGR-1 Wi 21 (alpha 1), anti-WT1 ␣6F or H7, preimmune rabbit IgG, or non-immune murine ascites at 1:50 or 1:100 dilutions, washed three times in 0.1% bovine serum albumin in PBS, and then incubated for 1 h with goat anti-rabbit or anti-murine IgG conjugated to fluorescein isothiocyanate (Kirkegaard and Perry). Cells were washed three times in PBS, then mounted in 100 mM Tris-HCl:glycerol, 50:50, 2% n-propyl gallate, pH 8, and viewed on an Olympus BHS photomicroscope equipped for epifluorescence.

Autoradiography
For mobility shift and immunoblotting studies the dried gels or membranes were autoradiographed with Kodak XAR film at Ϫ80°C for 0.5-96 h with Cronex Lightning Plus intensifying screens (DuPont). Quantification of signals was performed by densitometry of the autoradiograms with an LKB Ultroscan XL enhanced laser densitometer (Pharmacia Biotech Inc.).

RESULTS AND DISCUSSION
During culture, LLC-PK 1 cells differentiate from a rounded cell type to a fully polarized epithelium. Prior to their polarization and tight junction formation, these cells undergo several rounds of cell division coincident with the maximal EGR-1 expression and activation of the G␣ i-2 gene (31). To determine whether the WT1 gene also participates in these events a full-length human WT1 cDNA and a polyclonal (␣6F) or monoclonal (H7) antibody to the 173-residue amino-terminal non-zinc finger region of the WT1 protein were used to detect its expression in LLC-PK 1 cells. As seen in Fig. 1C, immunofluorescence of dividing non-confluent non-polarized cells with this antibody revealed a bright nuclear staining pattern in virtually every cell for EGR-1 whereas WT1 was not detected. By contrast, in fully polarized confluent cells this staining pattern was reversed. Quantification of EGR-1 and WT1 proteins in cells at Days 1-7 after plating by immunoblotting revealed a reciprocal pattern of expression for each protein in these cells. EGR-1 was barely detectable immediately after plating whereas WT1 was well expressed. However, by 36 -48 h, when the cells were actively dividing, EGR-1 protein was maximally expressed whereas WT1 was not detectable. On later culture days 5-7, following the full polarization of these cells, EGR-1 expression fell whereas WT1 protein and mRNA expression increased (Fig. 1, A and B). In day 7 cultures there was sporadic lifting of some quiescent cell monolayers (data not shown). Ingrowth of cells occurred in the margins of denuded areas of the culture plates with a corresponding re-expression of EGR-1 protein. The pattern of maximal expression of the EGR-1 protein coincides with the temporal maximal activation of the G␣ i-2 gene during growth, whereas its repression coincided with the maximal expression of the WT1 protein during cell quiescence (31).
In other cell types the overexpression of the G␣ i-2 subunit can activate signaling pathways that contribute to accelerated cell growth. Likewise overexpression of G␣ i-2 or EGR-1 also accelerates LLC-PK 1 cell growth. As WT1 expression corresponded to culture times when LLC-PK 1 cells were both differentiated and growth arrested, we questioned whether expression of WT1 in predifferentiated LLC-PK 1 cells would antagonize the effects of EGR-1 on cell growth during this period. LLC-PK 1 cells were transiently transfected with Bluescript or plasmids encoding cDNAs for G␣ i-2 (pRSV G␣ i-2 ) or WT1 (pRSV WT1), driven by a viral Rous sarcoma promoterenhancer to overexpress each protein. As seen in Fig. 2, by day 3 of culture predifferentiated cells transfected with plasmids encoding G␣ i-2 had growth rates that were 1.8 times faster than cells transfected with Bluescript, whereas LLC-PK 1 cells transfected with WT1 had growth rates that were 35% lower than Bluescript-transfected cells. Transfection efficiency in all these experiments was comparable at approximately 10 -20% efficiency. Despite the initial altered growth rates of G␣ i-2 and WT1-transfected cells, they still developed a normal polarized phenotype upon achieving confluence. These data suggested that both G␣ i-2 and WT1 participate in signaling pathways that normally contribute to renal cell growth.
We previously documented in rapidly dividing non-polarized cells a 135-bp enhancer area in the Ϫ200 to Ϫ335 region of the FIG. 1. Detection of WT1 gene expression in cultured LLC-PK 1 cells. A, Northern blot analysis of a tRNA (Ϫ), Wilms' tumor (ϩ), and LLC-PK 1 RNA from culture days 0 (day of trypsinization) through day 7 with a full-length human WT1 cDNA detected a predominant ϳ3kilobase transcript. B, quiescent LLC-PK 1 cells from day 10 of culture were trypsinized (day 0) and cultured for successive days. Immunoblotting of total cell extract from each of these cultures was performed to compare the relative amounts of EGR-1 or WT1 protein present. Each lane represents 150 g of protein quantitated. Autoradiograms of these proteins from the same acrylamide gel show in the upper panel EGR-1 induction (ϳ80 kDa) on culture day 1. The lower panel II shows maximal induction of WT1 protein (ϳ50 kDa) on culture days 5-7. Immunoprecipitation of WT1 was concurrently performed in these cultures, which also showed maximal induction of WT1 protein on culture days 5-7 seen in panel I. C, cell monolayers at days 1-10 were fixed, permeabilized, and stained by immunofluorescence with immune rabbit IgG anti-EGR-1 Wi 21 (alpha 1) (which detects the non-zinc finger region of the EGR-1 protein corresponding to residues 29 -117) or ␣6F and H7 antibodies (which detect the non-zinc finger amino-terminal region of the WT1 protein corresponding to residues 1-173). Representative images of cell cultures at days 2 (left panel) and 7 (right panel) are shown. No staining was observed with preimmune rabbit IgG or non-immune mouse ascites (data not shown). G␣ i-2 gene. This region contains a binding site (5Ј-CGC-CCCCGC-3Ј) for the EGR-1 transcription factor that provides a genomic signaling pathway for mitogenesis (33). To assess whether WT1 was repressing cell growth by repressing maximal transcriptional activation of the G␣ i-2 gene, cells were co-transfected with the pRSV WT1 plasmid, and plasmids encoding firefly luciferase reporter genes fused to 5Ј-flanking areas of the G␣ i-2 gene with (M14) or without the EGR-1 site (mutated M14 and M4). As seen in Fig. 3, a 60% repression of G␣ i-2 transcription was only found in renal cells following their transfection with the M14 plasmid that contained both an intact EGR-1 binding site and also overexpressed a functional WT1 protein. These data suggested that the WT1 protein was contributing to the temporal transcriptional repression of the G␣ i-2 gene in LLC-PK 1 cells during culture.
To determine whether the WT1 protein was directly contributing to the activation of the G␣ i-2 gene, a double-stranded 23-bp DNA segment derived from the 5Ј-flanking sequence of the gene, which also contained the EGR-1 consensus sequence (5Ј-ATCCGCC CGCCCCCGCCGTCGGG-3Ј), was synthesized and 32 P end-labeled for direct binding studies in mobility shift assays. Nuclear extracts from LLC-PK 1 cells that were either actively dividing and non-polarized (culture day 1) or LLC-PK 1 cells that were relatively quiescent and fully polarized (culture day 7) were examined. As seen in Fig. 4, the binding patterns of nuclear extracts from culture day 1 were different from those on culture day 7. Nuclear extracts from day 7 cells consistently demonstrated an additional faster mobility complex. We previously identified the EGR-1 protein as a predominant component of nuclear binding proteins in day 1 dividing cells but not day 7 quiescent cells (33). Based on our immunocytochemical studies it would be anticipated that the WT1 protein should be present in nuclear extracts of quiescent fully polarized LLC-PK 1 cells on culture day 7. To determine whether WT1 was one of the proteins interacting with the 23-bp probe, nuclear extracts from culture days 1 and 7 were preincubated with H7 antibody. Following electrophoresis, retarded mobility of only the additional complex in nuclear extracts from culture day 7 was found. The specificity of this interaction was demonstrated by competition with the 28-kDa protein expressing the aminoterminal 173 residues of the WT1 protein. These data demonstrate that the WT1 protein was one component of these nuclear complexes. Detectability of the WT1 protein was consistent with its pattern of maximal expression in quiescent polarized LLC-PK 1 cells that also have a corresponding repression of the G␣ i-2 gene.
These findings suggest the existence of differentiation-induced pathways in LLC-PK 1 cells that alternatively abrogrates EGR-1 and promotes WT1 gene expression, thereby modulating a target protooncogene G␣ i-2 that is participatory for growth and differentiation in these renal cells. These studies emphasize the usefulness of the LLC-PK 1 renal cell as a model to elucidate normal programs of genetic differentiation in which WT1 participates. Further examination these pathways may provide significant insights into the molecular events involved in renal hypertrophy, nephrogenesis, and oncogenesis.