Characterization of Glis2, a Novel Gene Encoding a Gli-related, Krüppel-like Transcription Factor with Transactivation and Repressor Functions

In this study, we describe the characterization of a gene encoding a novel Krüppel-like protein, named Glis2. Glis2 encodes a relatively proline-rich, basic 55.8-kDa protein. Its five tandem Cys2-His2 zinc finger motifs exhibit the highest homology to those of members of the Gli and Zic subfamilies of Krüppel-like proteins. Confocal microscopic analysis demonstrated that Glis2 localizes to the nucleus. Analysis of the genomic structure of the Glis2 gene showed that it is composed of 6 exons separated by 5 introns spanning a genomic region of more than 7.5 kb. Fluorescence in situ hybridization mapped the mouse Glis2 gene to chromosome 16A3-B1. Northern blot analysis showed that the Glis2 gene encodes a 3.8-kb transcript that is most abundant in adult mouse kidney. By in situ hybridization, expression was localized to somites and neural tube, and during metanephric development predominantly to the ureteric bud, precursor of the collecting duct, and inductor of nephronic tubule formation. One-hybrid analysis using Glis2 deletion mutants identified a novel activation function (AF) at the N terminus. The activation of transcription through this AF domain was totally suppressed by two repressor functions just downstream from the AF. One of the repressor functions is contained within the first zinc finger motif. The level of transcriptional activation and repression varied with the cell line tested, which might be due to differences in cell type-specific expression of co-activators and co-repressors. Our results suggest that Glis2 behaves as a bifunctional transcriptional regulator. Both the activation and repressor functions may play an important role in the regulation of gene expression during embryonic development.

DNA sequence-specific transcription factors are one group of nuclear proteins involved in the transcriptional regulation of gene expression. These proteins generally contain, in addition to a DNA-binding domain, one or more domains involved in activation or repression of transcription. DNA-binding domains often consist of conserved motifs, such as helix-turn-helix, helix-loop-helix, and zinc finger motifs (1)(2)(3). In proteins containing zinc finger motifs, the residues in the motif fold around a central zinc ion with a tetrahedral arrangement of cysteine and/or histidine metal ligands. Zinc finger proteins constitute one of the biggest families of transcriptional factors and can be divided into many subclasses based on, among other things, the number and type of zinc fingers they contain. For example, members of the nuclear receptor superfamily contain two Cys 4type zinc fingers (4,5). The Krü ppel-like zinc finger proteins (6), named after the Drosophila segmentation gene Krü ppel (7,8), contain two or more Cys 2 -His 2 -type zinc fingers. These proteins contain in addition a conserved consensus sequence, (T/S)GEKP(Y/F)X, between adjacent zinc finger motifs. Members of the Krü ppel-like zinc finger family have important regulatory functions during embryonic development and have been implicated in a variety of diseases (9).
The Gli protein subclass of Krü ppel-like zinc finger proteins consists of three members, Gli1, Gli2, and Gli3 (10 -12). Each member contains five tandem Cys 2 -His 2 zinc finger motifs. Gli proteins are related to Cubitus interruptus (Ci), 1 a Krü ppellike zinc finger protein important in wing development in Drosophila melanogaster (13). Ci and Gli proteins function as downstream regulators of transcription in the Sonic hedgehog (SHH)-Patched signal transduction pathway in Drosophila and vertebrates, respectively (12, 14 -17). Gli proteins can function as activators and repressors of gene transcription and play multiple roles in the regulation of prenatal mammalian development particularly in limb and craniofacial development (18 -20). Gli2 and Gli3 have been reported to be essential in the organogenesis of various tissues, including lung, trachea, and esophagus (18,21,22), and Gli3 has been implicated in several genetic disorders in humans (11,23,24). Gli1 has been demonstrated to function as a proto-oncogene. Aberrant expression and amplification of Gli1 is associated with basal cell carcinomas of the skin, sarcomas, and glioblastomas (10,20,25,26). been reported to play an essential role in body pattern formation during embryonic development (27)(28)(29)(30).
In this study, we determine the genomic structure of the mouse Gli-similar 2 (Glis2) gene, also referred to as NKL (31), encoding a novel member of the Krü ppel-like zinc finger family. Glis2 is most closely related to members of the Gli and Zic subfamilies of Krü ppel-like zinc finger proteins. As members of the Gli and Zic family, Glis2 contains five Cys 2 -His 2 zinc finger motifs and functions as a nuclear protein. Analysis of the genomic structure of Glis2 showed that it consists of 6 exons and 5 introns. The splice sites within the zinc finger domain are different from those of Gli genes in agreement with the conclusion that Glis2 forms a separate subfamily. Murine and human Glis2 mapped to chromosome 16A3-B1 and 16p13.3, respectively (32). Of the adult mouse tissues examined, Glis2 was found to be most highly expressed in kidney. In situ hybridization analysis localized Glis2 mRNA expression during metanephric development predominantly to the ureteric bud, precursor of the collecting duct, and inductor of nephronic tubule formation. These results suggest a possible role for Glis2 in the regulation of kidney morphogenesis. Its chromosomal localization and its expression kidney is of particular interest because the 16p13.3 locus has been implicated in several kidney diseases in humans (33,34). Analysis of its transcriptional activity using mono-hybrid and deletion mutant analysis identified a strong transactivation function (AF) in the N terminus of Glis2. The activation of transcription through this domain was totally suppressed by two repressor functions just downstream from the AF. One of the repressor functions was retained in a region that encompasses the first zinc finger motif. Both activation and repressor functions may play an important role in the regulation of gene expression during embryonic development and adult kidney.

EXPERIMENTAL PROCEDURES
cDNA Library Screening-During screening of a TriplEx cDNA library (4 ϫ 10 5 independent plaques) prepared from mouse kidney RNA (CLONTECH) with a Glis1 cDNA probe, 2 a cDNA clone was isolated containing a 1.5-kb-long insert with a sequence that was different from Glis1. The insert encoded a region highly homologous to the C 2 H 2 -type zinc finger motifs present in Glis1 and a flanking region that exhibited low homology to Glis1. GenBank TM searches indicated that the insert encoded a novel gene that is different from but related to Glis1. We named this gene Glis2.
DNA Sequencing-Plasmids were purified using Wizard miniprep or midiprep kits from Promega. Manual sequencing was performed using the Sequenase Quick-denature plasmid sequencing kit (Amersham Biosciences). Automatic sequencing was carried out using a Dynamic ET Terminator Cycle Sequencing Ready reaction kit (PerkinElmer Life Sciences) and an ABI Prism 377 automatic sequencer. DNA and deduced protein sequences were analyzed by the seqWEB and MacVector sequence analysis software packages.
Northern Blot and RT-PCR Analysis-Total RNA was isolated from 16 different mouse tissues using Trizol (Sigma) according to the manufacturer's protocol. Total RNA (30 g) was electrophoresed through formaldehyde 1.2% agarose gel, blotted to Hybond N ϩ membrane (Amersham Biosciences), and UV cross-linked. The expression of the mGlis2 gene was also examined by RT-PCR using total RNA (1 g) from different mouse tissues as template and ZFP1 (5Ј-TTCTGGTGGGGCT CTGCAC) and ZFP2 (5Ј-GCACTGAGGTCAAGGGGGC) as primers. The RT-PCR was carried out at the following conditions: 45 min at 48°C and 2 min at 94°C (1 cycle), followed by 0.5 min at 94°C, 1 min at 60°C, and 2 min at 68°C (25 cycles). Finally, after a 7-min incubation at 68°C, samples were analyzed on a TBE gel (0.8%) and transferred to a HyBond-N ϩ membrane. The blots were hybridized to a 32 P-labeled probe for Glis2. Hybridizations were performed at 68°C for 1 h, and the membranes were then washed twice with 2ϫ SSC, 0.1% SDS at room temperature for 30 min and 0.1% SSC, 0.1% SDS at 50°C for 40 min. Autoradiography was carried out with Hyperfilm-MP Amersham Biosciences) at Ϫ70°C.  ACACAG/gtgagcc 1512 ttatag/GTGAGA VI Ͼ1500

FIG. 1-continued
In Situ Hybridization (ISH)-Tissues from 16-and 19-dpc rat metanephrol were fixed in 4% buffered paraformaldehyde overnight and embedded in paraffin following ethanol dehydration. Hybridizations were performed according to Wilkinson and Green (35), using antisense or sense 35 S-labeled riboprobes generated from two different plasmids encoding either the 5Ј-end (nucleotides 21-515) or the 3Ј-end (nucleotides 1127-1562) of the cDNA sequence (35). Both probes yielded similar results. Thin section and whole mount ISH with 13-dpc rat embryos were performed as described (35,36).
Isolation of Genomic Clones-Two primer sets were used to screen by PCR a library of BAC vectors containing 50 -100-kb inserts of mouse genomic DNA. One primer set consisting of MGE1P (5Ј-CTATTTGGAT-GGTGTCCCA) and MGE1R (5Ј-ATCGACACACCGCTGCTTG) amplified a 150-bp fragment of mouse genomic DNA. The other primer set consisting of MGE4P (5Ј-TCAGATCATCATCCCGAACC) and MGE4R (5Ј-CGTCTTGTGATTTTCCAGGC) amplified a 300-bp fragment of mouse genomic DNA. Two positive clones were obtained. BAC vector DNA was isolated using a KB-500 kit from Incyte Genomics. DNA was digested with several restriction enzymes, including BamHI, HindIII, EcoRI, XbaI, and XhoI, and fragments were analyzed by Southern analysis using different regions of the mGlis2 cDNA as probes. One of two BAC vector clones that contained the full coding region of Glis2 was used for further investigation. The HindIII fragments hybridizing to the Glis2 probes were cut from the gel, purified with QIAEX II kit (Qiagen), and subcloned into pZErO-1 vector. One 7.0-kb HindIII fragment containing all the exons of Glis2 was sequenced.
Fluorescence in Situ Hybridization (FISH)-The regional chromosomal localization was determined by FISH using a 7-kb fragment of genomic DNA containing the Glis2 gene as a probe. FISH was carried out by Incyte Genomics. DNA from the mouse clone was labeled with digoxigenin dUTP by nick translation. Labeled probe was combined with sheared mouse DNA and hybridized to normal metaphase chromosomes derived from mouse embryonic stem cells in a solution containing 50% formamide, 10% dextran sulfate, and 2ϫ SSC. Specific hybridization signals were detected by incubating the hybridized slides in fluoresceinated anti-digoxigenin antibodies followed by counterstaining with 4,6-diamidino-2-phenylindole. The initial experiment resulted in specific labeling of the proximal region of a small sized chromosome that was believed to be chromosome 16 on the basis of 4,6-diamidino-2-phenylindole staining. A second analysis was conducted in which a probe that is specific for the telomeric region of chromosome 16 was co-hybridized with the Glis2 genomic fragment. This experiment re-FIG. 2. A, schematic comparison between Glis2, Glis1, Gli1, Gli2, Gli3, and Zic1. The percent identity of the various ZFDs with that of Glis2 is indicated. Little homology was observed in other regions. B, comparison of sequence within the zinc finger domain of Glis2 with those of Glis1, Gli1, Gli2, Gli3, and Zic1. Bold residues indicate amino acids conserved with Glis2. The asterisks indicate the Cys and His residues involved in the tetrahedral configuration in the zinc finger motifs. The amino acids between brackets indicate putative regions in the 4th and 5th zinc finger motif (ZF) interacting with DNA. Ϫ indicates gap. C, the second half of each zinc finger motif consists of an ␣-helix as determined by GOR3 secondary structure prediction analysis. sulted in the specific labeling of the telomere and the proximal portion of chromosome 16.
Plasmids-The reporter plasmid pG5-CAT containing five copies of the GAL4 upstream activating sequence (UAS) upstream of the E1B minimal promoter and referred to as (UAS) 5 -CAT, pM, encoding Gal4(DBD), and the pCMV␤ reporter vector were purchased from CLONTECH. The reporter plasmid pFR-LUC (referred to as (UAS) 5 -LUC) was obtained from Stratagene. The different pM-Glis2 deletion mutants were created by placing the Gal4-DNA binding domain at the N terminus of various Glis2 fragments. These fragments were generated by PCR using primers based on the polylinks in the pM vector. Glis2-specific 5Ј-and 3Ј-primers included either a BamHI or HindIII restriction site, respectively, to allow the PCR fragments to be subcloned into the BamHI and HindIII sites of the pM vector. Details on the length of each deletion are described in the text and figures. pcDNA4-Glis2 was constructed by inserting the full-length mGlis2 cDNA, generated by PCR, into BamHI and XbaI sites of pcDNA4/HisMax C (Invitrogen). pEGFP-Glis2 was generated by cloning full-length Glis2 into XhoI and BamHI sites of the pEGFP-C3 vector (CLONTECH). The plasmid pEGFP-Glis2⌬N containing an N-terminal truncated Glis2 (from Val 30 ) and pEGFP-Glis2⌬C (containing Val 30 to Ser 322 ) was generated by PCR. The integrity of all constructs was confirmed by restriction digestion and automatic DNA sequencing.
Transient Transfection Assays-Green monkey kidney fibroblast CV-1, Chinese hamster ovary CHO, dog kidney epithelial MDCK, and transformed human kidney 293 cells were obtained from ATCC and routinely maintained in DMEM or Ham's F-12 supplemented with 10% fetal bovine serum. Cells were plated in 6-well dishes at 2 ϫ 10 5 cells/well (6 ϫ 10 5 cells for 293 cells) and 20 h later co-transfected (1.0 -1.9 g DNA total) with 0.2-0.3 g of (UAS) 5  RNA was isolated from various mouse tissues and examined by Northern blot analysis (A) using a 32 P-labeled probe for Glis2 as described under "Experimental Procedures." Glis2 mRNA expression was also examined RT-PCR analysis (B).
FIG. 5. Expression of Glis2 mRNA in rat metanephroi. A-D, bright and dark field images of 16-dpc embryonic kidneys, which show high levels of expression in the branching ureteric bud/collecting duct (arrows), moderate levels in the cortical area or nephrogenic zone (NZ), and low levels in S-shaped bodies (arrowheads), which are precursors for the epithelia of the nephron. E and F, bright and dark field images of 19-dpc rat fetal kidneys, which reveal an intense signal in the medullary branches of the ureteric bud/collecting duct but reduced expression relative to the ducts in the nephrogenic zone. G and H, higher magnification of the inset from E which demonstrates weak expression of Glis2 in ampullae of the ureteric bud (UB), stroma (St), and glomeruli (Gl). Bar ϭ 0.1 mm. 0.2 g of pCMV␤ or ␤-actin-LUC expression plasmid, which served as an internal control to monitor transfection efficiency. CV-1, CHO, and MDCK cells were transfected in Opti-MEM (Invitrogen) using 3.0 l of FuGENE 6 (Roche Molecular Biochemicals) although 293 cells were transfected using 10 l of PolyFect transfection reagent (Qiagen). Cells were incubated for 48 h and then assayed for ␤-galactosidase and luciferase activity. The level of ␤-galactosidase activity was determined using a luminescent ␤-galactosidase detection kit (CLONTECH) according to the manufacturer's instructions. Luciferase activity was assayed with a luciferase assay kit from Promega. Transfections were performed in triplicate, and each experiment was repeated at least two times. The level of CAT protein was determined by the CAT enzyme-linked immunosorbent assay kit (Roche Molecular Biochemicals) according to the manufacturer's instructions.

Cloning of the Mouse Glis2
Gene-During screening of a mouse kidney cDNA library with a probe for Glis1, 2 a 1.5-kb cDNA fragment was cloned encoding a novel zinc finger protein, referred to as Glis2, that was related to but different from Glis1. The full-length coding region of Glis2 was obtained by assembling fragments obtained by cDNA library screening and 5Ј-RACE.
The genomic structure of the Glis2 gene was determined from a 7.5-kb HindIII genomic fragment as described under "Experimental Procedures." This analysis revealed that the Glis2 gene spans more than 7.5 kb and is composed of 6 exons separated by 5 introns (Fig. 1). A summary of the various sizes of the exons and introns and the sequence of the splice junctions is shown in Table I. The sequences of these junctions are consistent with the consensus A 62 G 77 g 100 t 100 a 60 a 74 g 88 for the 5Ј-donor and y 87 ny 97 a 100 g 100 G 55 (where y is pyrimidine) for the 3Ј-acceptor site for known splice sites within eukaryotic genes (37).
Comparison with Gli and Zic-The full coding and the deduced amino acid sequences of mouse Glis2 are shown in Fig. 1.
The putative translation initiation sequence CCACCATGC is, except for one nucleotide, identical to the Kozak consensus sequence CC(A/G)CCATGG (38). The open reading frame encodes for a protein of 521 amino acids with a calculated mass of 55.8 kDa. Glis2 is a basic protein with a theoretical pI of 9.03 and is relatively proline-rich with 13% of its residues consisting of proline. Glis2 contains a zinc finger domain (ZFD) consisting of five tandem zinc finger motifs with the consensus sequence X-Cys-X 2,4 -Cys-X 12,15 -His-X 3,4 -His-X (Fig. 2). The zinc finger motifs are connected by sequences that exhibit different degrees of homology with the consensus motif (T/S)GEKP(Y/F)X, typically found as interfinger spacer in members of the Krü ppel-like family (3). The ZFD of Glis2 exhibited the highest identity (49 -57%) with members of the Gli and Zic subfamilies of Krü ppel-like proteins (Fig. 2, A and B) (29, 39 -41). As Glis2, Gli and Zic proteins contain five zinc finger motifs. The 3rd and 5th zinc finger motif of Glis2 are most highly conserved with those of Gli proteins, exhibiting 70 and 88% identity, respectively. Analysis of the secondary structure of the Glis2 ZFD using the GOR3 secondary structure prediction method indicated that the second half of each zinc finger motif consists of an ␣-helix (Fig. 2C) and therefore resembles the structure of the zinc fingers present in Gli proteins (6). Little homology was observed between Glis2 and Gli and Zic proteins in regions outside the zinc finger domain.
Nuclear Localization of Glis2-To examine the cellular localization of Glis2, CV-1 cells were transfected with pEGFP-Glis2 plasmid DNA encoding the fusion protein EGFP-Glis2(fulllength) and its cellular localization analyzed by confocal microscopy. This analysis demonstrated that EGFP-Glis2 is primarily localized to the nucleus (Fig. 3A). The spatial distribution of Glis2 occurred in a typical speckled pattern as observed for many transcription factors (42), and no fluores-FIG. 6. Expression of Glis2 in somites and neural tube. Whole mount in situ hybridization was performed in 13dpc rat embryos as described (36). Glis2 mRNA expression was observed grossly in the caudal somites (arrows) and in the neural tube up to and including the 4th ventricle (arrowheads) in 13-dpc rat embryo. A, side view; B, dorsal view; C, dark field image of thin section hybridization of section of 13-dpc rat embryo.
FIG. 7. Regional mapping of the Glis2 gene by fluorescence in situ hybridization to mouse chromosome 16A3-B1. A, the location of the Glis2 gene was identified on metaphase chromosomes from murine embryo fibroblast cells using a digoxigenin dUTP-labeled, genomic fragment containing the Glis2 gene (arrowhead). Chromosome 16 was identified by co-hybridization with a probe specific for the telomeric region of chromosome 16 (arrow). A total of 80 metaphase cells were analyzed with 72 exhibiting specific labeling. B, the idiogram indicating the localization of the mouse Glis2 gene. Measurements of 10 specifically labeled chromosomes demonstrated that Glis2 is located at a position which is 8% of the distance from the heterochromatic-euchromatic boundary to the telomere of chromosome 16, an area that corresponds to band 16A3-B1. The arrowhead indicates the interval within which the hybridization signal was detected. cence was observed within the nucleoli. Although PEPSORT prediction II analysis of the Glis2 sequence did not identify any nuclear localization signal, visual inspection of the Glis2 sequence showed a sequence between residues 21 and 26 with homology to the nuclear localization signal consensus sequence XXK(K/R)X(K/R (43). However, this sequence appeared not to be important for the nuclear localization of Glis2, because EGFP-Glis2⌬N, in which the N terminus up to Val 30 was deleted, still localized exclusively to the nucleus (Fig. 3B), as did EGFP-Glis2⌬C in which the C terminus was deleted (Fig.  3C). These observations demonstrate that neither the N nor the C terminus is required for nuclear localization of Glis2 suggesting a role for the ZFD in nuclear localization.
Tissue-specific Expression of Glis2-To examine the tissuespecific expression of Glis2, total RNA was prepared from different mouse tissues and examined by Northern blot analysis using a radiolabeled probe for Glis2. The Glis2 probe hybridized to a single transcript about 3.8 kb in size. Of the adult mouse tissues examined, Glis2 mRNA was most highly expressed in kidney with low expression in several other tissues (Fig. 4A). This pattern of expression was confirmed by RT-PCR; highest expression was observed in kidney, moderate expression in heart and lung, and low expression in prostate, colon, and brain (Fig. 4B).
By ISH, Glis2 mRNA expression was observed in both 16and 19-dpc rat metanephrol (Fig. 5). At 16-dpc, high levels of expression were detected throughout the branching ureteric bud, including the tips of the bud located in the cortical aspect of the metanephros or nephrogenic zone. Moderate expression occurred throughout the nephrogenic zone, suggesting that blastemal and stromal populations derived from the metanephric mesenchyme both expressed the gene. Newly formed tubules originating from the mesenchyme in the form of S-shaped bodies were devoid or exhibited very low levels of Glis2 expression. By day 19, however, expression was clearly reduced in the nephrogenic zone, including the tips of the ureteric bud. Prominent expression was still observed in the medullary aspect of the ureteric bud/collecting duct as well as the ureter, and newly formed tubules remained negative. Expression patterns were comparable for both antisense riboprobes derived from distinct regions of the gene, whereas sense riboprobes from the same regions yielded background levels of expression (not shown).
By whole mount ISH, Glis2 mRNA expression was observed particularly in caudal somites (arrows, Fig. 6, A and B) and in the neural tube (arrowheads, Fig. 6B) up to and including the 4th ventricle of 13-dpc rat embryos. At this stage development of kidney is just beginning. Thin section ISH (Fig. 6C) confirmed and extended the expression profiles noted in whole mount ISH studies. The most prominent Glis2 expression in intact 13-dpc rat embryos was observed in neuroepithelia, especially in the developing spinal cord and in tissues surrounding the 4th and 3rd ventricles in the brain. The somites also showed moderate levels of expression, whereas most other tissues, e.g. heart and lung, exhibited lower levels of Glis2.
Chromosomal Localization of the Glis2 Gene-The regional chromosomal localization was determined by FISH using a genomic fragment of the Glis2 gene as a probe. Measurements of 10 specifically labeled chromosomes demonstrated that Glis2 is located at a position that is 8% of the distance from the heterochromatic-euchromatic boundary to the telomere of chromosome 16, an area that corresponds to band 16A3-B1 (Fig. 7).
Comparison of the Glis2 coding sequence with high throughput genomic sequences (htgs) in GenBank TM showed that the murine Glis2 sequence was contained in the human htgs clone AC012676 (working draft). The murine Glis2 coding region exhibited 93% homology with several regions in AC012676. We concluded that these genomic fragments encode the human homologue of murine Glis2. This htgs fragment, and therefore human Glis2, maps to chromosome 16p13.3 (32).
Identification of Transactivation and Repressor Functions-Members of the Gli and Zic families have been demonstrated to contain multiple domains with functions in activation or repression of transcription (19, 44 -50). To assess the transcriptional activity of Glis2, we first examined the effect of a series of C-terminal deletions on the transcriptional activity of Glis2 by mammalian one-hybrid analysis. As shown in Fig. 8A, human kidney 293 cells co-transfected with pM-Glis2(1-521) containing the full-length Glis2 coding region expressed similar levels of LUC activity as cells transfected with pM, indicating that full-length mGlis2 did not activate transcription. C-terminal deletions of Glis2 up to Glu 327 did not change the level of LUC activity; however, further deletion up to Arg 171 (pM-Glis2(1-171)) caused an 8 -11-fold increase in LUC activity. These observations indicated that the N terminus from Met 1 to Arg 171 contained a transcriptional activation function and that this activity was totally suppressed by the presence of a downstream repressor domain. Consistent with the locations of the activation and repressor domains, none of the N-terminal deletion constructs had a substantial effect on basal transactivation (Fig. 8B).
To map more precisely the activation domain, several additional N-and C-terminal deletions were examined. As shown in Fig. 9A, deletion at the C terminus up to Ser 148 (pM-Glis2(30 -148)) or Gly 137 (pM-Glis2(30 -137)) increased transcriptional activity about 8-fold compared with that of pM-Glis2 (30 -171). Further C-terminal deletions up to Ser 129 , Leu 120 , Asp 90 , or Asp 62 totally abolished transcriptional activity. N-terminal deletions up to Gly 71 (pM-Glis2(71-148)) slightly diminished transcriptional activity, whereas additional deletions greatly reduced transactivation (Fig. 9B). These observations demonstrate that the activation domain is located between Gly 71 and Gly 137 . Our results also indicate the presence of a repressor function between Ser 148 and Arg 171 .
The First Zinc Finger Contains a Major Repressor Function-Although the repressor function between Ser 148 and Arg 171 suppressed transactivation about 10-fold, Glis2 (30 -171) or Glis2(1-171) was still able to enhance transcription 8 -11fold, whereas induction of transactivation was totally repressed in Glis2(1-327) (Figs. 8A and 9A). These results suggested the presence of a second repressor function within the zinc finger domain of Glis2. To determine whether this repressor function is associated with a particular zinc finger motif of Glis2, we serially deleted each single zinc finger and determined its effect on transcriptional activity. As shown in Fig. 9C, consecutive removal of the fifth, fourth, third, and second zinc finger did not relieve this repression. However, the additional deletion of the first zinc finger motif restored transcriptional activation of the LUC reporter indicating that the first zinc finger contains a major repressor function.
The Transactivation Activity Is Cell Type-dependent-To determine whether the transcriptional activation by Glis2 is a general phenomenon or varies with cell type, we compared the induction of transactivation in four different cell lines, 293, MDCK, CV-1, and CHO cells. The results in Fig. 10 show that although Glis2(30 -148) induced transcriptional activation in all four cell lines, the fold increase in LUC activity differed greatly, from 7-fold in MDCK to about 200-fold in 293 cells. Inclusion of the first repressor domain (Glis2(30 -171)) almost totally suppressed transactivation in MDCK cells and reduced to a 3-4-fold induction in CV-1 and CHO cells, whereas in 293 cells a 10 -25fold increase in transactivation was still observed. Inclusion of the second repressor domain (Glis2(30 -521)) totally repressed transcriptional activation in all cell lines (Fig. 10). The different degrees of repression and transactivation observed in the cell lines may be due to different levels of expression of co-activators and co-repressors able to interact with Glis2.
In CV-1 and MDCK cells, Gal4(DBD)-Glis2(30 -521) repressed transcription at levels below those observed for Gal4(DBD), suggesting that Glis2 represses basal transcriptional activation and may function as an active repressor of transcription. pM-Glis2 encoding Gal4(DBD)-Glis2(full-length) fusion protein repressed basal transcriptional activation of the LUC reporter gene in CV-1 cells in a dose-dependent manner (Fig. 11A); basal transactivation was repressed by more than FIG. 9. Mapping of the activation and repressor domains. A and  B, kidney 293 cells were co-transfected with (UAS) 5 -LUC, pCMV␤, pM, pM-Glis2 containing various C-(A) and N-terminal (B) deletions, as indicated. After 48 h cells were assayed for LUC and ␤-galactosidase activity as described under "Experimental Procedures." The relative LUC activity was calculated and plotted. These results indicate that the activation domain of Glis2 is located between Gly 71 and Gly 137 . C, the first zinc finger in Glis2 contains a major repressor function. CV-1 cells were co-transfected with (UAS) 5 -LUC, pCMV␤, pM, pM-Glis2(1-171) or pM-Glis2 containing various deletions within the zinc finger domain as indicated. Zf1-5 refers to the presence of zinc fingers 1-5. After 48 h cells were assayed for LUC and ␤-galactosidase activity as described under "Experimental Procedures." The relative LUC activity was calculated and plotted.

95%.
Co-transfection with increasing amounts of pcDNA4-Glis2 expression vector reversed the transcriptional repression by Gal4(DBD)-Glis2 in a dose-dependent manner (Fig. 11B). This reversal may be due to competition between Glis2 and Gal4(DBD)-Glis2 for the limiting amounts of corepressors in the cell (squelching).

DISCUSSION
In this study, we describe the genomic cloning of the mouse Glis2 gene encoding a novel member of the Krü ppel-like family of zinc finger proteins. Based on the number, type, and sequence homology of the zinc finger motifs, Glis2 is most closely related to members of the Gli and Zic subfamilies of Krü ppellike proteins (10,12,29,30,51,52). However, Glis2 likely constitutes a distinct subclass of Krü ppel-like zinc finger proteins because the ZFDs among Gli or Zic family members are more highly conserved. This is supported by the fact that the locations of the splice junctions in the ZFD in Glis2 are not conserved with those of Gli genes (51,53) and the absence of several small highly conserved regions outside the ZFD of Gli proteins that are important in repressor and activation functions of these proteins (12,39).
The ZFDs of Gli and Zic proteins have been demonstrated to be involved in the recognition of specific DNA elements (54,55). Analysis of the crystal structure of the Gli zinc finger domain in complex with its DNA-binding site has indicated that the first three zinc fingers make little or no contact with DNA, whereas the ␣-helices in last two zinc fingers maintain extensive contacts with DNA (6,56). The residues in zinc finger 4 and 5, implicated in making DNA contacts, are identical among Gli proteins (brackets in Fig. 2B). Corresponding residues in zinc finger 4 and 5 of Glis2 exhibit an 80% identity with those of Gli proteins. These similarities suggest that Glis2 may interact with DNA elements similar to those for Gli and Zic proteins.
Although Glis2 was found primarily in the nucleus, GOR3 secondary structure prediction analysis did not identify a consensus nuclear localization signal. Deletion of the N or C ter-minus had little effect on its nuclear localization of Glis2 suggesting that its ZFD may be important for nuclear localization. In this context it is interesting to mention that Gli and Zic proteins have been reported to interact physically with each other through their ZFDs and that Gli proteins through their interaction with Zic are able to translocate to the nucleus (57). However, nuclear transport of Gli proteins is also determined by nuclear export signals and interactions with other proteins, such as the mammalian homologue of Drosophila Suppressorof-Fused (SUFUH). SUFUH can physically interact with Gli1 and retain it in the cytoplasm (58 -60). Future studies have to determine the precise mechanism of nuclear transport of Glis2.
Members of the Gli and Zic subfamilies act as transcriptional factors that have critical functions in the regulation of embryonic development (19, 29, 44 -50, 55). Although Gli and Zic proteins contain multiple domains that regulate their transcriptional activity, these functions are still poorly characterized. Analysis of a series of deletions mutants of Glis2 identified an activation domain between Gly 71 and Gly 137 , a region just before the ZFD. Transactivation domains usually serve as an interface for the recruitment of co-activators that mediate their interaction with the basic transcriptional machinery. The co-activator CREB-binding protein has been shown to mediate the transcriptional activation of Ci, Gli3, and Gli2 (47,49,61); however, we have been unable to demonstrate an interaction between Glis2 and CREB-binding protein. 3 An activation domain with similarity to the putative TAF II 31 interaction motif was identified in Ci and Gli proteins (50); however, this motif showed little homology with the activation domain of Glis2.
Deletion mutant analysis further demonstrated that the region downstream from the activation domain of Glis2 was able to totally suppress transcriptional activation. This repression was due to the presence of two adjacent repressor functions that together totally block transactivation activity. The first 3  repressor function was associated with a region extending from Ser 148 to Arg 171 . The second repressor function was found to be associated with the first zinc finger motif. Recently, a repressor function was identified within the ZFD of the transcription factor AEBP2, a repressor containing three Krü ppel-type zinc finger motifs. In this case, the repressor function was associated with the second zinc finger motif (62). Repression can be mediated by various mechanisms. Passive repression involves competition for a common DNA binding site or co-activator (quenching), whereas active repression involves binding to DNA and recruitment of specific co-repressors that mediate the interaction of the repressor with proteins of the basic transcriptional machinery. The results in Fig. 11 show that Glis2 can function as an active repressor.
A number of different repressor domains, including Krü ppelassociated box (KRAB) and SCAN domain, have been identified in Krü ppel-like zinc finger proteins. KRAB is an evolutionary conserved motif of 75 amino acids that can be found in about 30% of the Krü ppel-like zinc finger proteins (63,64). The SCAN domain is enriched in hydrophobic and negatively charged residues and contains an LX 6 LL motif (65). These regions have been shown to mediate repression by recruiting co-repressor complexes (66,67). Recently, transcriptional repression by the Krü ppel-like zinc finger proteins KRAZ1 and -2 has been reported to involve recruitment of the co-repressor TIF1 to their KRAB repressor domain (66). As in Gli and Zic proteins, the repressor domains identified in Glis2 do not show any homology with KRAB or SCAN and appear to be novel repressor functions.
Although Glis2 protein is in the nucleus, it is unable to activate transcription in the cell systems tested. The lack of transcriptional activation is therefore not due to retention of Glis2 in the cytoplasm as reported for Gli (49,58). For Glis2 to function as an enhancer of transcription may require activation by a specific signaling pathway that could involve deletion of the repressor domain by proteolytic cleavage, (de)phosphorylation of Glis2, conformational changes in Glis2, and/or inactivation of co-repressors. It is interesting to note that members of the Gli and Zic family also contain a composite of repressor and activation domains (47,49). The activity of Gli3 has been reported to be positively and negatively regulated by SHH and cAMP-dependent protein kinase signaling pathways, respectively (47). Truncation of the activation domain at the C terminus transforms Gli2 into a repressor, whereas deletion of the repressor domain at the N terminus converts Glis2 into a strong activator (49). Similarly, in Drosophila, the full-length Ci protein acts as a transcriptional activator, whereas a 75-kDa proteolytic product acts as a repressor of transcription (68). Studies are in progress to determine the molecular mechanisms that activate Glis2.
Northern and PCR analysis demonstrated that Glis2 mRNA is highly expressed in kidney. Kidney expresses a number of other Krü ppel-like zinc finger proteins that play critical roles during kidney development and in renal tumorigenesis. For example, the Wilms' tumor 1 (WT1) protein contains four Cys 2 -His 2 Krü ppel-like zinc fingers (69) and has been implicated in kidney development and in renal cancer. Kid-1, -2, and -3 are Krü ppel-like transcription factors that contain 13 Cys 2 -His 2 zinc finger motifs and a KRAB repressor domain (70,71). They appear to play an important role during nephrogenesis. The recently described kidney-enriched Krü ppel-like factor appears to regulate the expression of two kidney-specific chloride channels (72). Although Glis2 is only distantly related to these transcription factors, it may also control important functions in kidney. This is supported by in situ hybridization analysis of Glis2 expression in developing kidney. In addition, whole mount ISH demonstrated expression of Glis2 mRNA in caudal somites and neural tube suggesting additional roles for Glis2 in development and overlaps the expression pattern observed for Gli and Zic (73)(74)(75). The expression pattern for Glis2 in the developing kidney is quite intriguing as it correlates well with the pattern described previously for SHH (76,77). Expression of both genes is absent from the tips of the 19-dpc rat ureteric bud, where induction of tubular morphogenesis in the metanephric mesenchyme is mediated, suggesting it is not involved in nephronic differentiation. However, it is detected in the stalks or medullary branches of the ureteric bud, which forms the collecting duct structure in the adult tissue. Thus Glis2 may have a regulatory function in branching morphogenesis and cellular proliferation similar to that of SHH in the epithelia of the developing lung (78). It is unclear at this point whether there is a relationship between the expression of these two genes. As already mentioned, Gli and Ci proteins are downstream regulators of SHH signaling. Because SHH signaling is generally regarded as paracrine in nature, it may be unlikely that SHH regulates Glis2 expression, although it is possible that SHH can indirectly regulate Glis2 through interactions with the surrounding mesenchyme. Alternatively, Glis2 may function upstream of SHH.
of Glis2, a novel gene that is most closely related to members of the Gli and Zic subfamilies of Krü ppel-like zinc finger proteins and show that Glis2 mRNA is detected in the stalks or medullary branches of the ureteric bud in developing kidney. These observations suggest that Glis2 may have an important role in regulating kidney development. In this regard, it is interesting to note that human Glis2 maps to chromosome 16p13.3 (32). This locus is a hotspot for several kidney diseases and has been implicated in autosomal polycystic kidney disease, the most common inherited form of cystic kidney disease (33) and renal cancer (34). A recent study (31) described the cloning of a cDNA encoding a novel Krü ppel-like protein, referred to as NKL, that is identical to Glis2. NKL/Glis2 was shown to be expressed in Xenopus primary neurons and to promote neuronal differentiation in vertebrates suggesting a regulatory role for this protein in neurogenesis. The observed expression of Glis2 in the neural tube of 13-dpc rat embryos is in agreement with this conclusion. Thus, Glis2 appears to play an important role in the regulation of several developmental processes. Both the repressor and transactivator functions of Glis2, identified in this study, could be involved in this regulation.