Zik1, a transcriptional repressor that interacts with the heterogeneous nuclear ribonucleoprotein particle K protein.

The heterogeneous nuclear ribonucleoprotein particle (hnRNP) K protein is comprised of multiple modular domains that serve to engage a diverse group of molecular partners including DNA, RNA, the product of the proto-oncogene vav, and tyrosine and serine/threonine kinases. To identify additional K protein molecular partners and to further understand its function, we used a fragment of K protein as a bait in the yeast two-hybrid screen. The deduced primary structure of one of the positive clones revealed a novel zinc finger protein, hereby denoted as Zik1. In addition to the nine contiguous zinc fingers in the C terminus, Zik1 contains a KRAB-A domain thought to be involved in transcriptional repression. Zik1 and K protein bound in vitro and co-immunoprecipitated from cell extracts indicating that in vivo their interaction is direct. Expression of Gal4 DNA-binding domain-Zik1 fusion protein repressed a gene promoter bearing Gal4-binding elements, indicating that from cognate DNA elements Zik1 is a transcriptional repressor. The known diverse nature of K protein molecular interactions and now the identification of a K protein partner that is a transcriptional repressor lends support to the notion that K protein is a remarkably versatile molecule that may be acting as a docking platform to facilitate communication among molecules involved in signal transduction and gene expression.

The heterogeneous nuclear ribonucleoprotein particle (hnRNP) K protein is comprised of multiple modular domains that serve to engage a diverse group of molecular partners including DNA, RNA, the product of the proto-oncogene vav, and tyrosine and serine/threonine kinases. To identify additional K protein molecular partners and to further understand its function, we used a fragment of K protein as a bait in the yeast two-hybrid screen. The deduced primary structure of one of the positive clones revealed a novel zinc finger protein, hereby denoted as Zik1. In addition to the nine contiguous zinc fingers in the C terminus, Zik1 contains a KRAB-A domain thought to be involved in transcriptional repression. Zik1 and K protein bound in vitro and co-immunoprecipitated from cell extracts indicating that in vivo their interaction is direct. Expression of Gal4 DNA-binding domain-Zik1 fusion protein repressed a gene promoter bearing Gal4-binding elements, indicating that from cognate DNA elements Zik1 is a transcriptional repressor. The known diverse nature of K protein molecular interactions and now the identification of a K protein partner that is a transcriptional repressor lends support to the notion that K protein is a remarkably versatile molecule that may be acting as a docking platform to facilitate communication among molecules involved in signal transduction and gene expression.
K protein was first discovered as a component of the heterogeneous nuclear ribonucleoprotein particle (hnRNP) 1 from which it derives its name (1,2). As a component of hnRNP, K protein binds poly(C) RNA, but not other RNA homopolymers (1), and might be involved in mRNA processing. K protein binds RNA via the highly conserved KH (K homology) domains (3) that are present in a variety of other RNA-binding proteins (4,5). In vitro, K protein also recognizes the c-myc promoter CT element (6 -8) and the B enhancer motif (9,10) suggesting that it might also be involved in transcription. In fact, expression of K protein can both activate and repress RNA polymerase II promoters (8,11), and addition of K protein stimulates transcription by purified RNA polymerase II in vitro (12,13). It appears that the function of K protein is regulated by the cognate RNA and DNA motifs (14,15).
Since its original discovery in hnRNP, K protein has been shown to be a highly versatile molecule that engages not only nucleic acids but also an array of proteins, it interacts in vivo and in vitro with the product of the proto-oncogene vav, the protein-tyrosine kinase, Src (14,16,17), and the TFIID TATAbinding protein (8). In vivo and in vitro K protein forms a complex and is phosphorylated by an interleukin-1-responsive serine/threonine K protein kinase (s) (14,15,18). K protein can simultaneously engage multiple molecular partners, and, in the context of K protein, the serine/threonine K protein kinase can be reactivated in vitro by Src (14) providing evidence that K protein may act as a docking platform to facilitate molecular interactions.
Clearly, K protein is a very interactive molecule that appears to be involved in signal transduction, gene transcription, and mRNA processing. Considering K protein's highly interactive nature, we reasoned that it could be used as a probe to uncover novel processes and factors involved in signal transduction and gene expression. Therefore, we used K protein as a bait in the yeast two-hybrid screen (19) and have isolated a novel gene that encodes a transcriptional repressor.

MATERIALS AND METHODS
Cell Lines-The rat glomerular epithelial cells, the murine pre-B lymphocytes 70Z/3, the human epidermoid KB, and the monkey kidney COS cell lines were grown as described before (20).
Reagents-The bacterial expression vector pGEX-KT was provided by Dr. J. Dixon (University of Michigan). The mammalian expression vector pM2 was kindly provided by Dr. I. Sadowski (University of British Columbia, Vancouver, Canada). Glutathione agarose beads were obtained from Sigma.
RNA Extraction and Northern Blot Analysis-Total RNA was extracted essentially as described previously (21). Cells or animal tissues was washed with PBS. 2.0 ml of Solution D (4 M guanidinium thiocyanate, 25 mM sodium citrate, 0.5% sarkosyl, 0.1 M ␤-mercaptoethanol) was added to each plate or to 100 mg of animal tissue to lyse the cells. The final RNA was dissolved in 25 l of water and used for Northern blot analysis. RNA was analyzed essentially as described previously (22). After first denaturing in formaldehyde and formamide at 65°C for 15 min, the RNA sample was cooled to 4°C on ice. 10 g of total RNA per lane was resolved by electrophoresis for 4 h at 70 V using 1.4% agarose gel containing 2.2 M formaldehyde. RNA was transferred overnight in 20 ϫ SSC to Hybond N ϩ membrane (Amersham, Arlington Heights, IL), and dried at 80°C for 45 min. The membranes were prehybridized for 2 h at 42°C in hybridization buffer (50% formamide, 5 ϫ Denhardt's solution, 5 ϫ SSC, 0.5% SDS, 0.1 mg/ml denatured salmon sperm DNA, and 0.1 mg/ml yeast tRNA). After prehybridization, 32 P-labeled cDNA probe (2 ϫ 10 6 cpm/ml) was added and hybrid-* This work was supported by National Institutes of Health Grants GM42508, GM45134, and DK45978 and the Northwest Kidney Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U69133.
ization was carried out overnight at 42°C. Following hybridization, the membranes were washed twice in 2 ϫ SSC with 0.1% SDS at 22°C for 10 min, then twice in 0.1 ϫ SSC with 0.1% SDS at 55°C for 20 min, and autoradiographed.
Yeast Two-hybrid System Library Screen-The procedure used here was based on methods previously described (19,23). Briefly, for the screen, the L40 strain of yeast was used, which is auxotrophic for his, trp, leu, and ade (generated by Dr. Stanley Hollenberg). The L40 yeast strain contains both lacZ and HIS3 marker genes under the control of minimal GAL1 promoters fused with multimers of LexA DNA-binding sites. Yeast strains were grown at 30°C in rich medium containing 1% yeast extract, 2% Bacto-peptone, 2% glucose, and 0.1 mg/ml adenine.
To construct the bait, the open-reading frame of a K protein was used as a template for PCR. The PCR primers were programmed to contain a BamHI site on the 5Ј end and stop codon and a SalI site on the 3Ј end. After digestion, the PCR-generated fragments were subcloned in-frame with LexA in the pBTM116 vector, the DNA-binding domain plasmid (pBTM116 was constructed by Drs. Paul Bartel and Stan Fields). The cDNA library for the yeast two-hybrid screen was generated by randomprimed cDNA synthesis from a mouse 9 -10-day-old embryo (Dr. Stanley Hollenberg), and size-selected to have inserts in the range of 350 -700 nucleotides. This library was inserted into the transcription activation domain vector, pVP16, to generate VP16-cDNA hybrid proteins.
Yeast was first transformed with the bait plasmid, and then the L40 LexA-K protein strains were transformed with the cDNA library. Screening for positive clones containing the two plasmids, LexA-K fusion protein and the activation domain-library plasmids, was done by growth on His selective media and by testing the colonies for ␤-galactosidase activity (23). To eliminate false positive clones, we mated a transformant containing the positive cDNA library plasmid with a strain transformed with LexA-lamin fusion protein plasmid, pLAM5 (generated by Dr. R. Sternglanz), as a nonspecific "bait" (24).
Screening of Murine cDNA Library and DNA Sequencing-The partial cDNA labeled by random priming using [␣-32 P]dCTP (Amersham) was used to screen approximately 10 6 plaque-forming units of a cDNA library in gt10 from 22D6 mouse pre-B cells (provided by Dr. A. Bothwell, Yale University) as described previously (9,25). Several clones were plaque-purified and subcloned into pBluescript vector (Stratagene) for sequencing. Sequencing was performed using DyeDeoxy Terminator Cycle Sequencing Kit (Perkin Elmer).
In Vitro Transcription and Translation-The partial or complete cDNAs were used as template for in vitro transcription by SP6 or T7 DNA-dependent RNA polymerases to generate mRNAs for in vitro translation as described previously (26). In vitro translation in rabbit reticulocyte cell-free system was performed according to the manufacturer's protocol (Promega, Madison, WI).
In Vitro Binding Studies-5 l of the 35 S-labeled translation products were added to a suspension of 10 l of glutathione beads bearing either the full-length or a fragment of K protein fused to GST in 100 l of binding buffer (10 mM Tris-HCl, pH 7.5, 100 mM KCl, 1.0 mM MgCl 2 , and 0.1% Nonidet P-40). After mixing for 60 min (4°C), the beads were washed three times with 400 l of binding buffer and were boiled with 30 l of SDS sample buffer, and released proteins were analyzed by SDS-PAGE.
Plasmid Constructs-StuI-SmaI fragment containing five Gal4-binding elements and TATA box from pG5bCAT plasmid (27) was introduced into the SmaI site of pGL3 enh plasmid (Promega) to create a reporter plasmid pG5GL3 enh. Expression plasmid pM2Zik1 was constructed by insertion of the full-length coding region of Zik1, which was modified by PCR to create appropriate cloning sites, into the BamHI site of pM2 plasmid (28). The final construct contained an insert encoding the Gal4 DNA-binding domain-Zik1 fusion protein under the control of SV40 promoter. The plasmid pM2ZikDBH was derived from pM2Zik1 by cutting at BstXI-HindIII and blunt-ligating the linearized plasmid creating the N-terminal 54 amino acids of Zik1 fused to the Gal4 DNA-binding domain.
For immunoprecipitation, 50 l of the cell extracts diluted 2ϫ with extraction buffer (no inhibitors) were incubated with 1 g of anti-Gal4 DNA-binding domain monoclonal antibody (Clontech, Palo Alto, CA).
After 2 h (4°C), 5 l of protein A-Sepharose beads (Pharmacia Biotech Inc.) were added for another hour of incubation and were then washed three times with 400 l of TBST (10 mM Tris, pH 7.5, 150 mM NaCl, 0.1% Triton X-100). Prior to use, protein A-Sepharose beads were blocked for 1 h with 1 mg/ml bovine serum albumin in TBST. Proteins were eluted from the beads with SDS loading buffer and were resolved by SDS-PAGE.
Transient Transfection and Luciferase Reporter Gene Assay-Rat glomerular epithelial and COS cells were grown in Dulbecco's minimal essential medium supplemented with 10% fetal calf serum to approximately 60 -75% of confluence in 100-mm diameter dishes and were transfected using the DEAE-dextran method (29). Briefly, cells were treated with a total of 15-20 g of plasmid DNA in 5 ml of Dulbecco's minimal essential medium and 0.16 mg/ml DEAE-dextran. After a 4-h incubation, cells were shocked for 1 min at room temperature with 10% dimethyl sulfoxide in 1 ϫ PBS, were washed twice with PBS, and fresh medium was added. After 48 h, transfected cells were washed twice with PBS, scraped with a rubber policeman, and spun down in a microcentrifuge. The cell pellet was resuspended in 150 l of ice-cold 1 ϫ lysis buffer (Promega). The supernatant was assayed for luciferase activity by the standard Promega protocol using a luminometer.

RESULTS AND DISCUSSION
Cloning of a Novel Protein That Interacts with hnRNP K-The N-terminal two-thirds of K protein (amino acids 1-318) (14) was used as a bait in two-hybrid screen to isolate K protein molecular partners. A partial cDNA of 366 base pairs in length was isolated and sequenced, and this cDNA represented a fragment of a novel gene. This partial cDNA was used to screen a gt10 cDNA library (30) derived from the murine pre-B cell line, 22D6, and 3 putative clones were isolated and subcloned into a pBluescript vector. The longest clone was used for sequencing of both strands.
The complete nucleic acid and deduced amino acid sequences of the 3042-long cDNA clone are shown in Fig. 1. The ATG located at position ϩ245 is an initiation codon which starts the longest open reading frame. This clone encodes a protein of 463 amino acids, with a calculated molecular weight of 52,778 and a pI of 8.5. The cloned cDNA contains a short 5Ј (244-base pair) and a long 3Ј (1406-base pair) untranslated region (Fig. 1.). Comparison of the deduced primary structure of the protein with sequences in the data base revealed that the cDNA encodes a protein with the N-terminal segment (amino acids 14 -47) similar to Kid1, a zinc finger protein enriched in the rat kidney (31, 32) (Fig. 2C). Notably, this region contains the Kruppel-associated type A box (KRAB-A). This highly conserved motif is present in approximately one-third of all zinc finger proteins and has been shown to mediate transcriptional repression (32). The C terminus of the cloned protein (amino acids 215-463) consists of nine contiguous repeats that are similar or identical to the C 2 H 2 zinc finger consensus sequence (YxCxxCxxxFxxxxxLxxHxxxHTGEKP) (Fig. 2B) (33). The cDNA fragment isolated in the two-hybrid screen represented the 59 -180-amino acid stretch of the full-length protein. Therefore, this 122-amino acid-long stretch contains the K proteinbinding domain (Figs. 1 and 2A). We have designated this novel protein Zik1, for zinc-finger protein interacting with K protein.
Tissue Distribution of Zik1 Transcripts-The full-length cDNA was used as a probe to define the tissue distribution of Zik1 transcripts. RNA was isolated from a variety of mouse tissues and cells grown in culture, separated by size on an agarose-formaldehyde gel, and blotted onto a nylon membrane. It is interesting to note that Kid1, the zinc finger protein that shares most sequence similarities to Zik1, is also expressed preferentially in the kidney (31).
Interaction of Zik1 with K Protein Is Direct-To determine whether or not the interaction of K protein with Zik1 is direct, we tested the ability of 35 S-labeled Zik1 in vitro translated to bind to glutathione beads bearing the GST-K fusion protein (14). After incubation with Zik1, beads were washed, boiled in the presence of SDS, and proteins were resolved by SDS-PAGE. The Coomassie Blue-stained and autoradiographed gel is illustrated in Fig. 4. In agreement with the yeast two-hybrid screen, these results revealed that Zik1 binds to the beads bearing the full-length K protein fused to GST (lane 1). Zik1 also binds to several K protein deletion fragments, the GST-K13 (amino acids 1-337) (lane 2) and GST-K3 (amino acids 171-337) (lane 5), but not to the GST-K8 (amino acids 1-108) (lane 4) or the GST-K12 (amino acids 1-209) (lane 3) deletion mutants. These results suggest that the binding of Zik1 to K protein is direct and specific and allow us to map the Zik1-binding site to the 209 -337-amino acid domain of K protein.

Interaction of Zik1 with K Protein Is Regulated by Nucleic
Acids-K protein binds to RNA and to single-and doublestranded DNA in a sequence-selective manner (6,9,15). Because the function of K protein might be regulated by its binding to RNA and DNA, we next tested the effect of cognate nucleic acid motifs on the interaction of Zik1 with K protein.
Results of this series of experiments are illustrated in Fig. 5. The binding assay was performed as described above (Fig. 4) in the presence or absence of RNA or DNA. These results showed that poly(C) RNA, which binds K protein, blocked Zik1 binding to K protein, while poly(A), which does not bind K protein (1), did not affect binding of Zik1 to glutathione beads bearing K protein (compare lanes 3 and 4 with lane 2). Similarly, a double-stranded synthetic oligonucleotide containing the B motif and the single-stranded synthetic oligonucleotide containing the antisense B motif, both of which bind K protein (9), blocked Zik1 binding (compare lanes 7 and 9 with lane 6), while the single-stranded sense B, which does not bind K protein, did not alter Zik1 binding (compare lane 8 with lane 6). These results demonstrate that cognate RNA and DNA sequences both block the interaction of Zik1 with K protein. This experiment, in conjunction with the observation that K protein phosphorylation is regulated by nucleic acid binding (9, 14, 15), indicates that K protein function is highly regulated by cognate RNA and DNA motifs.
Co-immunoprecipitation of Zik1 and K Protein from Cell Extracts-To determine whether Zik1 and K protein interact in vivo in mammalian cells, we used monoclonal antibody directed against Gal4 DNA-binding domain to determine whether K protein co-immunoprecipitated with either Gal4 or Gal4-Zik1 fusion proteins when they were co-expressed in COS cells. Western blot analysis using an antibody directed against K protein (Fig. 6) showed that the anti-Gal4 antibody precipitated K protein (lane 2) when it was co-expressed with Gal4-Zik1 fusion protein, but not when it was co-expressed with Gal4 protein alone (lane 1). This experiment provides further evidence that Zik1 and K protein can interact in vivo.
Zik1 Contains Transcriptional Repressor Activity-Zik1 contains the KRAB-A domain and shares sequence similarity with the transcriptional repressor Kid1, suggesting that Zik1 might also affect transcription. To determine the transcriptional activity of Zik1, either the full-length Zik1 (amino acids 1-463) or its KRAB-A domain (amino acids 1-54) were fused in-frame to the Gal4 DNA-binding domain (amino acids 1-147) in the mammalian expression vector pM2 (36). These constructs were transiently co-expressed with the reporter luciferase gene plasmids in either rat glomerular epithelial or COS cell lines. The results of luciferase activity assay are shown in Fig. 7. The full-length Gal4-Zik1 and more so its KRAB-A domain substantially repressed transcription of the luciferase gene driven by a promoter containing 5 Gal4 DNA elements (shaded bars), but did not diminish activity of the same promoter lacking the Gal4-binding motifs (clear bars). These results demonstrate that, like Kid1 (31), Zik1 can act as a transcriptional repressor. The observation that K protein can engage both a transcriptional repressor and the TFIID TATA-binding protein may explain why expression of K protein can both activate and repress gene promoters (8,11).
Identification of a transcriptional repressor that directly interacts with K protein further expands the repertoire of K protein molecular partners: sequence-specific RNA and DNA, Vav, both tyrosine and serine/threonine kinases, TATA-binding protein, and now Zik1. What are the possible implications of this versatility? Not only is K protein very interactive, but it is also abundant, and is present both in the nucleus and the cytoplasm (9,10,15). The diversity of K protein molecular interactions, K protein abundance, and wide intracellular distribution would fit a model in which K protein acts as a universal docking platform. The yeast two-hybrid screen revealed that K protein can interact with itself at two different sites (data not shown), suggesting that it has the potential to form oligomers. If so, this would provide the K protein with a great capacity to permit simultaneous docking of a number of proteins and thus provide an environment for multilateral protein cross-talk. As exemplified by the K protein-Zik1 (Fig. 5) and K protein-K protein kinase (9,14,15) interactions, a cross-talk among the docked proteins could be regulated by sequencespecific nucleic acids. The potential of K protein to simultaneously engage and allow cross-talk between signaling molecules such as Src or Vav, with transcriptional repressors such as Zik1, opens up an exciting avenue to investigate. FIG. 6. K protein co-immunoprecipitates with Zik1 from COS cell extracts. K protein was transiently co-expressed in COS cells with either Gal4 DNA-binding domain (Gal4) or Gal4 DNA-binding domain-Zik1 fusion protein (Gal4-Zik1) using mammalian expression vectors. 48 h following transfection, cells were harvested and extracts were prepared for immunoprecipitation. 50 g of cell extracts were incubated with 1.0 g of anti-Gal4 monoclonal antibody. The immune complexes were recovered with protein A beads and centrifugation. After several washes, proteins were eluted from the beads by boiling in loading buffer, and proteins were resolved by SDS-PAGE and were electrotransferred to polyvinylidene difluoride membrane for immunoblotting with anti-K protein antibody (14). K protein is marked by the arrow. and rat glomerular epithelial (GEC) cells. 10 g of expression plasmid with 5 g of luciferase reporter plasmid were used for cotransfections in both cell lines. 48 h following transfection, cells were pelleted and were lysed in reporter lysis buffer. Luciferase activity was measured using a luminometer (37) and was normalized for protein concentration. The data shown are representative of one out of five experiments.