ABSTRACT
The Ski oncoprotein has been found to bind non-specifically to DNA in association with unindentified nuclear factors. In addition, Ski has been shown to activate transcription of muscle-specific and viral promoters/enhancers. The present study was undertaken to identify Ski's DNA binding and transcriptional activation partners by identifying specific DNA binding sites. We used nuclear extracts from a v-Ski-transduced mouse L-cell line and selected Ski-bound sequences from a pool of degenerate oligonucleotides with anti-Ski monoclonal antibodies. Two sequences were identified by this technique. The first (TGGC/ANNNNNT/GCCAA) is the previously identified binding site of the nuclear factor I (NFI) family of transcription factors. The second (TCCCNNGGGA) is the binding site of Olf-1/EBF. By electophoretic mobility shift assays we find that Ski is a component of one or more NFI complexes but we fail to detect Ski in Olf-1/EBF complexes. We show that Ski binds NFI proteins and activates transcription of NFI reporters, but only in the presence of NFI. We also find that homodimerization of Ski is essential for co-activation with NFI. However, the C-terminal dimerization domain of c-Ski, which is missing in v-Ski, can be substituted by the leucine zipper domain of GCN4.
v-Ski is a 49 kDa nuclear protein that is truncated by 20 amino acids (aa) at its N-terminal end and 292 aa at its C-terminal end with respect to the full length c-Ski protein (1 -3 ). Analysis of the deduced amino acid sequence of v-Ski reveals a number of sequence motifs which suggest that its biochemical activity may involve DNA binding and transcriptional regulation (2 ). These motifs include a proline rich region, a potential helix-loop-helix and several highly conserved histidine and cysteine residues that could constitute metal binding domains. In addition to these elements, the c-Ski protein contains a C-terminal dimerization domain that mediates Ski homodimerization as well as its heterodimerization with the related protein SnoN (4 ,5 ). It has been suggested (5 ) that efficient dimerization mediated by this domain underlies c-Ski's more potent transforming activity compared to v-Ski which is lacking this region (6 ).
Both v-ski and c-ski are capable of inducing transformation and myogenesis when overexpressed via retroviral vectors (6 ,7 ). In addition, published work implicates ski in the transformation of both erythroid and myeloid cells (8 ,9 ) and in the differentiation of megakaryocytes (10 ) and neuronal cells (11 ). It is not clear how ski influences these varied cellular processes but some published work suggests that it functions as a regulator of transcription. Ishii and co-workers have shown that purified, bacterially-expressed human c-Ski (hu-Ski) does not bind DNA cellulose unless it is mixed with a nuclear extract from Molt 4 cells (12 ). Kelder and co-workers (13 ) have extended these findings by showing that v-Ski expressed in mouse L-cells binds DNA cellulose and activates transcription from several viral enhancers. In studies related to ski's role in muscle differentiation, Rosenthal and co-workers showed that Ski activates transcription from the muscle-specific myosin light chain enhancer (14 ). In this case, Ski appears to act in combination with the muscle regulatory protein, MyoD. Thus a simple explanation of Ski's multiple activities is to propose that it interacts with a variety of transcription factors and alters their roles in transcriptional regulation.
In this report we describe studies undertaken to investigate this possibility by determining whether, on its own or in combination with associated nuclear proteins, Ski possesses specific DNA binding activity. To accomplish this goal, we have employed cyclic amplification and selection of targets technique (CASTing) (15 ,16 ) which is a modification of the method developed by Oliphant et al. (17 ). In this technique one starts with a pool of oligonucleotides containing a central region of random sequence and performs repetitive cycles of DNA binding and PCR amplification of bound sequences to identify specific DNA binding sites. We have carried out CASTing with the use of immunoaffinity chromatography to purify the Ski-bound oligonucleotides. This approach has allowed us to identify one of Ski's partners and to assess its ability to regulate transcription in vivo as a consequence of these DNA and protein interactions. We show that the major binding site identified is identical to that previously found for the NFI family of transcription factors (18 -21 ). We present evidence for the involvement of Ski in protein complexes that bind this site and we show that Ski binds NFI proteins and activates transcription of a reporter gene with upstream NFI binding sites. Our data show that Ski dimerization is essential for cooperation with NFI, and this requirement is discussed with regard to the proposed role for dimerization in cellular transformation.
The v-Ski-specific monoclonal antibodies (mAbs), G8 and M6, were covalently linked to protein A agarose using the Immunopure IgG orientation kit (Pierce) according to the protocol supplied by the manufacturer. Ascites fluid (1 ml) containing either mAb G8 or M6 was used per 2 ml of protein A agarose. The protein A agarose-coupled mAbs were shown to specifically immunoprecipitate v-Ski from [35S]methionine labeled cell lysates (data not shown). The binding capacity of 1:1 mixture of the two matrixes was measured by competitive immunoprecipitation using a bacterial GST-v-Ski competitor (5 ) and found to be ~150 ng v-Ski/[mu]l matrix (data not shown).
The oligodeoxynucleotides (oligos) used for identification of v-Ski binding sites contain 22 bases of random sequence (CASTing oligo). On the 5'-end, this sequence is flanked by 16 bases of known sequence including the recognition site for the restriction enzyme BamHI. The 3' flank consists of 14 bases of known sequence including a terminal 10 base self complementary sequence and the site for the restriction enzyme PstI. The primer used for PCR amplification of Ski bound oligos was identical to the 16 base sequence at the 5'-end of the CASTing oligo. The sequences of these oligos and of those used as probes and competitors in electrophoretic mobility shift assays are shown below. CASTing oligo: ggcggatccacctaca(n22)tgtgcactgcagtg; PCR primer: ggcggatccacctaca; NFI probe (top strand): ggcggatccacctacaTGGCAacgTGCCAtgtgcactgcagtg; Mutant competitor (top strand): ggcggatccacctacaTTTCacGAAAtgtgcactgcagtg.
Oligos to be used for CASTing or as competitors in binding assays were made double stranded by annealing their self complimentary 3'-ends and extending with the Klenow fragment of DNA polymerase I. To produce 32P-labeled oligos for use as probes 20 ng of the top strand was 5'-end-labeled with [[gamma]-32P]ATP and polynucleotide kinase and annealed to a 10-fold excess of the complimentary strand. The double stranded probes were then gel purified (5% polyacrylamide gel electrophoresis and elution from minced gel slices in binding buffer).
A 1.5 kb BglII fragment from plasmid pCRpolski containing the entire v-ski coding region plus 145 bp of downstream gag sequence (2 ) was expressed in a clonal line (P28#3) of transfected mouse L-cells (13 ) or in chicken embryo fibroblasts (CEFs) infected with a recombinant avian retrovirus RCASBP(A) (22 ). Lysates of these cells and of control L-cells were analyzed for v-Ski production by sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and western blotting with anti-v-Ski monoclonal antibodies and the chemiluminescent detection system of Tropix (6 ). The relative amounts of v-Ski in these lysates were determined by densitometry of the exposed film. The intracellular location of v-Ski in P28#3 cells was determined by indirect immunofluorescence as described previously (23 ).
Nuclear extracts were prepared from P28#3 cells using the method of Dignam (24 ). An aliquot (100 [mu]l) of ` Dignam extract' (DE) was added to 600 [mu]l buffer C (100 mM NaCl, 50 mM Tris-HCl pH 7.9, 5 mM MgCl2, 1 mM DTT) containing 2.5 [mu]g double stranded CASTing oligo, 5 [mu]g yeast tRNA, 5 [mu]g Poly (dI-dC) and 100 [mu]g/ml BSA. This reaction was incubated for 1 h at 4oC with gentle mixing. Then 200 [mu]l of a 70% suspension of anti-Ski mAb affinity beads (prepared with an equal mixture of Ski-specific monoclonal antibodies G8 and M6), was added and incubation continued for 1 h at 4oC with gentle mixing. The mixture was transferred to a chromatography column and the buffer solution allowed to drain out. The resulting column was then washed with 25 column vol of ice cold buffer C. Bound oligos were eluted with 2.5 column vol of buffer C containing 1 M NaCl. The eluted oligos were extracted with phenol/chloroform, diluted to 0.25 M NaCl and precipitated with 2 vol of ethanol in the presence of 10 [mu]g glycogen. The precipitate was dissolved in H2O, adjusted to 0.3 M NaOAc and reprecipitated with ethanol.
One half of the oligonucleotide DNA eluted from the mAb affinity matrix was amplified by PCR using the PCR primer oligo. Amplified DNA was resolved on a 5% polyacrylamide gel and the full length reaction product was excised from the gel. Gel slices were crushed and incubated in 1 M KAc overnight at 37oC. The eluted DNA was precipitated twice with ethanol and used for a second round of selection. This purified product was digested with PstI and subjected to a final round of binding and affinity purification.
Oligonucleotide DNA isolated as described above was digested with BamHI and cloned into a modified pUC18 vector which had been previously digested with PstI and BamHI. The vector modification entailed digesting with HindIII, blunting the sticky ends with the Klenow fragment of DNA polymerase and religating. This produced a translational reading frame shift which rendered the pUC-encoded [alpha]-peptide of [beta]-galactosidase non-functional. Insertion of the isolated oligos restores the correct reading frame and thus provides a screen for recombinants (17 ). When plated on bacterial plates containing X-gal and IPTG, colonies which contain plasmid with inserted oligo DNA will have a blue color and those lacking inserts will remain white. Positive colonies (blue) were picked at random and the inserted DNA sequenced by the dideoxy chain termination method (25 ). We also cloned and sequenced oligos from the starting pool and confirmed that the central degenerate region consisted of random nucleotide sequences with no consensus and no relatedness to the sequences selected (data not shown).
Binding reactions (20 [mu]l) containing 3-6 [mu]g P28#3 nuclear extract were performed in Dignam buffer D (20 mM HEPES pH 7.9, 100 mM KCl, 0.5 mM DTT, 0.2 mM EDTA, 0.2 mM PMSF, 20% glycerol) containing 2 [mu]g poly (dI-dC), 300 [mu]g/ml BSA and 50 000 c.p.m. (0.5-1 ng) 32P-labeled oligonucleotide probe. Binding reactions were incubated at room temperature for 20 min and loaded directly onto a 5% polyacrylamide gel which had been pre-electrophoresed for 90 min at 150 V. Electrophoresis was continued at the same voltage for 3 h. The gel and chamber buffers used were either 50 mM Tris, 0.38 M glycine and 2 mM EDTA (TGE) or 90 mM Tris, 90 mM boric acid and 2.5 mM EDTA (TBE) as indicated in the figure legends. Following electrophoresis the gel was dried and autoradiographed overnight at -70oC with an intensifying screen. For competition assays the binding conditions were as described above except that unlabeled oligonucleotide competitors (amounts described in figure legends) were added and incubated for 15 min at room temperature prior to addition of the 32P-labeled probe. Monoclonal anti-Ski antibodies G8, G37 or anti-[beta]-galactosidase mAb B20 (2 [mu]l ascites fluid) were added to binding reactions after addition of all other components and the reactions were then incubated and analyzed as above.
The chloramphenicol acetyl transferase (CAT) reporter plasmids were derived from RSV-CAT (26 ) by replacing its NdeI to EcoRI (in RSV-LTR) fragment with the NdeI to EcoRI fragment from pUC18 containing the multi-cloning site, to generate pENCAT. This deletes the RSV-LTR enhancer, and places the pUC multi-cloning site sequence immediately upstream of the minimal LTR promoter. A reporter (pNF7CAT) with seven tandem copies of the Ski (NFI) binding site was generated by cloning self-ligated copies of these sequences at the BspMI site of pENCAT.
Plasmids for expression of Ski and NFI in Drosophila cells were constructed in pac5c-pl which uses the Drosophila actin 5c promoter to drive expression (27 ). A full length chicken c-ski cDNA (FB29) from pMEX29 (3 ,28 ) was cloned into pac5c-pl as an XbaI fragment to yield paccski29. A v-ski containing BglII fragment from pCRpolski was cloned into the BamHI site of pac5c-pl to yield pacvski. A v-ski-like gene with an added 3' leucine zipper (LZ) coding sequence was constructed by PCR amplification of the GCN4 LZ domain (codons 248-281) from plasmid Ycp88-GCN4 (29 ) using a 5' primer with an added NdeI site and a downstream primer with a SalI site. The amplified fragment was digested with NdeI and SalI and cloned between the NdeI site at codon 547 in c-ski (the 3'-end of v-ski corresponds to codon 458) and a SalI site in the downstream multi-cloning site region of plasmid KS29cski (3 ,30 ) to yield KS29ski-GCN4. A DraIII-RsaI fragment from KS29ski-GCN4, containing the 3'-end of ski and the GCN4 sequence of this fused gene was substituted for the DraIII-EcoRV fragment of pacvski to yield pacski-GCN4. A related plasmid expressing the same portion of ski but without the GCN4 sequence (pacski-dN) was generated by truncating ski at the NdeI site.
Plasmids for in vitro transcription-translation were constructed in p5'EETM1 or p3'EETM1 (gifts of D.Templeton) which contain a promoter and a transcriptional terminator for phage T7 RNA polymerase and the internal ribosome entry sequence from encephalomyocarditis virus. A v-ski-like fragment (NcoI-NdeI) of c-ski from pMEX29 was cloned into NcoI-StuI digested p5'EETM1 to generate pEEski-dN. The ski-GCN4 fragment (NcoI-SalI) from KS29ski-GCN4 was cloned at the cognate sites in p5'EETM1 to generate p5'EEski-GCN4. A deleted form of ski-GCN4 was constructed by replacing the NcoI-NdeI fragment of p5'EEski-GCN4 with the cognate fragment from Cla12-Ncoski[Delta]1 to generate pEETM[Delta]1skiGCN4. The deleted segment encodes aa 16-119 of c-Ski which includes the epitope for the G8 monoclonal antibody.
Subclones of chicken NFI cDNAs (31 ,32 ) in Bluescript KS+ were used to generate expression plasmids. NFIA1 cDNA (SpeI-HincII fragment from p144.1) and NFIB2 cDNA (BamHI-HincII fragment from pKSTC1) were isolated by digestion with the indicated restriction nuclease and cloned into pac5c-pl which had been cut with either SpeI-HpaI or BamHI-HpaI, to yield pacNFI-A1 and pacNFI-B2, respectively.
Plasmids for expression of a glutathione S-transferase (GST) fusion with v-ski and for in vitro transcription-translation of c-ski were described previously (5 ). Plasmids for in vitro translation of NFI-A1 and NFI-B2 for use in GST fusion protein binding assays were constructed by subcloning the respective cDNAs into pRB0.5 as described previously (33 ). For in vitro co-translation with c-Ski, NFI-A1 was isolated as an EcoRI fragment from p144.1 (34 ) and inserted into EcoRI cut p3'EETM1. A GST-NFI-A1 fusion was made by cloning the same fragment into the EcoRI site of pGEX-2T to yield pGEX-NFI-A1.
Drosophila Schneider Line 2 (SL2) cells were seeded at a density of 3 * 106 cells per 35 mm plate in Ex-cell 400 medium supplemented with 6.8 mM glutamine and cultured overnight at 25oC prior to transfection. Triplicate plates were co-transfected with each combination of CAT reporter (600 ng) and the indicated expression plasmid (activator plus empty pac5c-pl vector to total 600 ng) by the DEAE dextran (100 [mu]g/ml) method for 4-5 h at 25oC (35 ). Transfection medium was removed, the cells washed, re-fed with 2 ml of medium and cultured for 72 h. Cells were then harvested into 1 ml TNE (10 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA) and spun for 1 min at 13 000 r.p.m. in a microcentrifuge. The cell pellet was resuspended in 150 [mu]l of 0.25 M Tris-HCl pH 7.8 and lysed by three freeze-thaw cycles using an ethanol-dry ice bath and a 37oC water bath. Following cell lysis the sample was heated at 65oC for 10 min, cooled on ice, clarified by centrifugation at 13 000 r.p.m. for 5 min and the supernate used to determine CAT activity. Where indicated samples of each lysate were used to determine the amount of protein and these values, which never varied by more than 10%, were used to normalize the CAT activity data. We did not employ an internal control for transfection efficiency because our earlier work had shown that Ski activates expression of all of the commonly used control plasmids (13 ).
CAT assays were performed by the liquid scintillation method (36 ) in 100 [mu]l reactions containing 50 [mu]l cell lysate plus 50 [mu]l 0.2 M Tris-HCl pH 8.0, 0.5 mg/ml butyryl CoA and 0.2 [mu]Ci [3H]chloramphenicol. Reactions were incubated at 37oC for 2-3 h and extracted with 200 [mu]l xylene. After two back extractions with 10 mM Tris-HCl pH 7.5, 1 mM EDTA, the xylene phase was added to 4 ml scintillation fluid (Eco-Lite) and counted to determine the amount of butyrylated chloramphenicol formed.
[35S]methionine labeled Ski and NFI proteins were produced by coupled transcription-translation in vitro using the p5'EETM1 or pRP0.5 plasmids described above according to the suppliers protocol (Promega). Binding of in vitro translated proteins to GST-v-Ski was performed and analyzed as described previously (5 ). GST binding studies were also performed in the presence of 100 and 200 [mu]g/ml ethidium bromide to eliminate DNA-protein interactions (37 ). For studies involving co-immunoprecipitation two expression plasmids were employed in a single transcription-translation reaction. Following translation a sample was set aside for analysis of the total translation of the two proteins while the remainder was subjected to immunoprecipitation in buffer containing 50 mM Tris-HCl pH 7.8, 250 mM NaCl, 0.1% NP-40 and 0.1 mM PMSF. The untreated translation products and immunoprecipitates were analyzed by SDS-PAGE and fluorography. The ability of anti-NFI anti-serum to co-precipitate Ski protein was used to estimate protein-protein association. To assess the efficiency of the anti-serum to precipitate NFI and its background binding of the Ski protein, the two proteins were produced individually in separate transcription-translation reactions and subjected to analysis by immunoprecipitation and SDS-PAGE. To test the ability of an added LZ to affect dimerization of v-Ski, a plasmid (pEETM[Delta]1skiGCN4) encoding a v-Ski-like protein containing the GCN4 LZ but deleted of the region encoding the epitope for the G8 mAb was co-transcribed-translated with either plasmid pEEski-dN which encodes the undeleted v-Ski-like protein lacking the LZ or plasmid p5'EEski-GCN4 which encodes the same Ski segment plus the LZ. Background binding of G8 mAb to the [Delta]1skiGCN4 protein was determined by performing and analyzing an equal amount of this protein alone produced in a separate translation.
Human c-Ski protein was shown to require additional factors present in nuclear extracts to bind DNA cellulose (12 ). We have confirmed this finding with both v-Ski (13 ) and chicken c-Ski (data not shown). We therefore chose to use nuclear extracts from cells over-expressing v-Ski rather than using purified bacterially expressed Ski to isolate DNA binding sites. To this end, we chose a cell line established from clone P28#3 of transfected L-cells that expresses v-Ski at approximately five times the level seen in CEFs infected with an avian retrovirus expressing v-Ski (Fig. 1 , compare CEFvski, 20 [mu]g to P28#3, 7 [mu]g). Western blotting also shows that the v-Ski expressed in P28#3 cells is the correct apparent molecular mass of 55 kDa and that our antibody is specific for this protein as no v-Ski-like protein is detectable in extracts from control L-cells (Fig. 1 ). Immunofluorescence shows that v-Ski in P28#3 cells is located in the nucleus but is excluded from the nucleolus (Fig. 1 ). This is identical to the localization seen in chicken embryo fibroblasts (CEFs) infected by a v-ski retrovirus (1 ).
We have found that all of the v-Ski protein in nuclear extracts of P28#3 cells binds DNA cellulose and is eluted at NaCl concentrations >0.1 M (13 ). This result indicates that cellular factors which are required for Ski to bind DNA (12 ) are expressed at non-limiting levels in P28#3 cells. To determine if this result reflects the potential for sequence-specific DNA binding by v-Ski-containing complexes, we have employed P28#3 nuclear extracts in the CASTing technique to purify Ski bound sequences from a pool of synthetic, random-sequence oligonucleotides as described in Materials and Methods. The sequences of 35 Ski-bound oligos are presented in Figure 2 . No two of the 35 sequenced clones contain the same insert indicating that selective protein binding and not selective PCR amplification is responsible for the enrichment of the isolated sequences. These sequences readily fall into two distinct groups that possess either of two consensus sequence motifs (Fig. 2 ). Fourteen clones contain a close match to the motif 1A sequence (tTGGC/AxxxxxT/GCCAa) which has perfect dyad symmetry and is the previously identified binding site for the NFI family of transcription factors (18 -21 ). Another five clones contain motif 1B which is a NFI half site. Five clones contain the second consensus sequence TCCCxxGGGA (TC3 dyad) which also possesses dyad symmetry and has been previously identified as the DNA binding site for the transcription factor, Olf-1/EBF (38 ,39 ). An additional eight clones contain the motif 2B sequence (ATCCC) which is probably a half site of motif #2A.
To confirm that v-Ski protein is present in the complexes which bind to either the TC3 dyad or the NFI site, we have attempted to either supershift the complexes or inhibit their formation using Ski specific mAbs. We first screened a set of v-Ski mAbs generated in this lab for their ability to alter complex formation or mobility. This screen has identified two mAbs, G8 and G37, that produce specific supershifted bands in the NFI EMSA (Fig. 4 A). The two mAbs produce a supershifted band of similar mobility suggesting that they are recognizing Ski in the same complexes. To be certain that the production of supershifted complexes with the G8 mAb is due to interaction with Ski in specific DNA-protein complexes, we performed two controls. In the first, we determined whether the supershifted species was subject to specific competition by an unlabeled NFI oligo. In the second, we used an anti-[beta]-galactosidase mAb (B20), which is the same isotype as G8 and G37 ([gamma]I) and was generated simultaneously with the other mAbs as a result of immunizing against a [beta]-galactosidase-v-Ski fusion protein. The absence of the supershifted band in the presence of excess competitor and in the presence of B20 mAb supports the conclusion that this band results from specific interactions with the NFI binding site and with Ski protein (Fig. 4 A). On the other hand, we failed to identify an anti-Ski mAb that has a reproducible effect on the mobility or formation of complexes with the TC3 oligo and we have also failed in attempts to detect Ski in these complexes by western analysis (data not shown). Because of this uncertainty, and because the TC3 sequence was under-represented in the pool of bound sequences, we have limited the remainder of the analyses reported here to the studies involving the NFI site.
The EMSA results suggest that Ski is associated with a specific NFI binding site complex, but do not establish that this occurs by direct interaction between Ski and NFI proteins. To establish this, we have asked whether Ski proteins bind directly with members of the NFI family. As shown in Figure 5 A (lanes 2 and 5) we find that GST-v-Ski binds in vitro translated NFI-A1 and NFI-B2. The efficiency of binding the NFI proteins is about the same as the binding of v-Ski to itself (Fig. 5 A, lane 8). The binding of these three proteins by GST alone is detectable (Fig. 5 A, lanes 3, 6 and 9) but insignificant compared to that of GST-v-Ski. To test whether these associations are direct or mediated by binding of the proteins to contaminating DNA, we performed similar assays in the presence of ethidium bromide (37 ). As shown in Figure 5 B, GST-NFI-A1 binds efficiently to itself and to c-Ski in the presence of ethidium bromide at either 100 or 200 [mu]g/ml. These results indicate that if our reticulocyte lysate contains DNA it is not responsible for the association of Ski and NFI-A1.
Having found that Ski and NFI proteins participate in both protein-protein and protein-DNA interactions, we sought to investigate the effect of these interactions on transcriptional regulation. Initial experiments using NFI reporters in both L-cells and chicken fibroblasts have been inconclusive, probably due to the expression of endogenous Ski as well as multiple forms of NFI (data not shown). To surmount these problems we have turned to Drosophila SL2 cells as they lack both Ski and NFI but support transcription activation by exogenous NFI (40 ). Co-transfection of SL2 cells with either paccski or pacvski does not result in transcriptional activation of a reporter containing upstream NFI sites (pNF7CAT) above that seen with the empty pac5c-pl vector (Fig. 6 A). However, co-transfection of pacNFI-A1 or pacNFI-B2 with the pNF7CAT reporter results in an impressive 40- or 10-fold activation of CAT expression, respectively (Fig. 6 A). This activation is dependent on the upstream NFI binding sites in the reporter since CAT expression from the same reporter lacking the binding sites is not increased by NFI or Ski (Fig. 6 A).
We have shown that, in association with other nuclear proteins, v-Ski binds to DNA as a result of sequence-specific interactions. Using nuclear extracts as the source of v-Ski and monoclonal antibodies to Ski to purify bound sequences we have identified two DNA consensus sequences bound by Ski-containing complexes. One of these is the previously identified binding site of the NFI proteins. Electrophoretic mobility shift assays with an NFI binding site probe reveal multiple specific complexes reflecting the expression of several NFI proteins which are known to bind the same site as both homodimers and heterodimers. The ability of a Ski-specific mAb to elicit a supershift of at least one of these complexes, suggests that v-Ski is a component of that complex. The observation that pure Ski protein does not bind the NFI site (data not shown) indicates that NFI proteins are required for Ski to bind this sequence and the resulting complexes consist of NFI dimers plus Ski.
This work was supported by grant CA-43600 to E.S. from the National Cancer Institute of the USPHS and by grants from Deutsche Forshungsgemeinschaft (SFB 388), the Fonds der Chemischen Industrie and the German-Israeli Foundation for Research and Development to A.E.S. J.J.K. was supported in part by Sensus Corp. and the State of Ohio's Eminent Scholar's Program.
Namciu, S., Lieberman, M. A. and Stavnezer, E. (1994) Oncogene, 9, 1407-1416.
*To whom correspondence should be addressed. Tel: +1 216 368 3353; Fax: +1 216 368 3419; Email: exs44@po.cwru.edu
REFERENCES
Present addresses: +Department of Genetics, University of Wisconsin, Madison, WI, USA and §Department of Molecular and Experimental Medicine, The Scripps Research Institute, La Jolla, CA, USA