AU2485792A - Purine-region dna binding protein - Google Patents

Description

PURINE-REGION DNA BINDING PROTEIN

BACKGROUND OF THE INVENTION

Technical Field

The present invention relates, in general, to a DNA binding protein and to a DNA sequence encoding same. In particular, the invention relates to a GA binding protein and to DNA segments encoding the subunits thereof.

Background Information

Herpes simplex virus 1 (HSV1) immediate early (IE) genes are induced at the outset of the lytic infection by a virion associated protein termed VP16 (Post et al, Cell 24, 555 (1981)). At least two classes of cis-regulatory elements qualify HSV IE genes for induction by VP16. The most essential VP16 σis-response element is characterized by the nonanucleotide sequence 5*-TAATGARAT-3'

(Mackem et al, Proc Natl. Acad. Sci. U.S.A. 79, 4917 (1982); Mackem et al, J. Virol. 44, 939 (1982); Cordingley et al. Nucleic Acids Res. 11, 2347 (1983); Kristie et al, Proc. Natl. Acad. Sci. U.S.A. 81, 4065 (1984) ; Gaffney et al, Nucleic Acids Res. 13, 7847 (1985); Bzik et al, ijid. 14, 929 (1986); O'Hare et al, J. Virol. 61, 190 (1987); and Triezenberg et al. Genes Dev. 2, 730 (1988)). VP16 binds tightly to this DNA sequence in a complex with the cellular transcription factor Octl (Preston et al. Cell 52, 425 (1988); O'Hare et al, ibid. p. 435 (1988) ; and Gerster et al, Proc. Natl. Acad. Sci. U.S.A. 85, 6347 (1988)). A second cis-regulatory element required for VP16-mediated induction of HSV IE genes consists of three imperfect repeats of the purine-rich hexanucleotide 5*-CGGAAR-3' (Triezenberg et al, Genes Dev. 2, 730 (1988) and Spector et al, ibid. 87, 5268 (1990)). A protein complex capable of avid interaction with the purine-rich repeats (GA repeats) has been identified in soluble preparations of rat liver nucleic (Triezenberg et al, Genes Dev. 2, 730 (1988)). This GA binding protein (GABP) consists of two separable subunits. Purified samples of either subunit do not interact with the GA repeats, yet regain potent DNA binding activity when mixed (LaMarco et al, Genes Dev. 3, 1372 (1989)).

Applicants have isolated cDNA clones encoding both subunits of GABP and have revealed that one (GABPα) is related to the Ets transforming protein, while the other (GABPB) contains a series of 33-amino acid repeats related in sequence to a variety of proteins including Notch of Drosophila melanoσaster. Linl2 and Glpl of Caenorhabditis eleqans and SW14 and SW16 of Saccharormyces cerevisiae ( harton et al, Cell 43, 567 (1985); Greenwald, i id 43, 583 (1985); Yochem et al, Nature 335, 547 (1988); Yochem et al. Cell 58, 553 (1989); Austin et al, ibid. , p. 565 (1989); Breeden et al. Nature 329, 651 (1987); and Andrews et al, ibid. 342, 830 (1989)). In addition, Applicants have demonstrated that these two protein sequence motifs, the Ets-related domain of GABPα and the 33-amino acid repeats of GABPB, contribute the surfaces that form a multiprotein complex capable of stable and specific interaction with DNA. These findings, which form the basis of the present invention, provide insight into the problem of regulatory specificity and define, for the first time, a discrete function for the 33-amino acid repeat motif.

SUMMARY OF THE INVENTION

It is a general object of the invention to provide DNA segments encoding the subunits of GABP.

In one embodiment, the present invention relates to a DNA segment encoding GABPα:, GABPB1 or GABPB2, or portion thereof.

In a further embodiment, the present invention relates to a construct comprising at least one of the above-described segments and to a host cell transformed therewith.

Further objects and advantages of the present invention will become clear from the description that follows.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure l. Amino acid sequences of tryptic peptides derived from GA binding proteins. GABP (20 μg) was purified to homogeneity (inset) as described (LaMarco et al. Genes Dev. 3, 1372 (1989)) except that boiled salmon sperm DNA (20 μg/ml) was included as a non-specific competitor in the DNA affinity chromatographic step. Approximately 500 picomoles of protein was lyophilized, reduced, acetylated, and subjected to cleavage by trypsin (Boehringer

Mannheim) . The resulting peptides were separated by reverse-phase HPLC as described (Stone et al. Laboratory Methodology in Biochemistry, Fini et al eds (RC Press, Boca Raton, FL, 1991)). Amino acid sequence analysis was performed on a Vydac C18 column (Applied Biosystems 477-A) . The amino acid sequences derived from peaks 1-13 were: 1, SLFDQGVIEK; 2, 7AWALEGY; 3, DEIS?VGDEGEFK; 4, ELESLNQEDFFQR; 5, LQESLDAHEIELQDIQL?P?R; 6, DQISIVGDEGEFK; 7, MAELV; 8, YVQASQLQQMNEIVTIDQP; 9, TPLHMWASEGHA; 10, GEILWS; 11, LIEIEIDGTEK; 12, ILMANGAPFTTD; 13, TGNNGQIQL?QFLLEL?TDR.

Figure 2. Nucleotide and deduced amino acid sequences of cDNAs encoding GABP subunits. (A) Sequence of GABPα. (B) Sequences of GABPBl and &2. An unamplified cDNA library prepared from mouse adipocyte mRNA was screened with a mixture of degenerate oligonucleotides derived from the amino acids sequences of peptides 3, 4, 5, and 8 (Fig. 1) labeled with ³²P using polynucleotide kinase. The basic SSC protocol was used (Ausubel et al, Current Protocols in Molecular Biology (Wiley & Sons, NY) , 1989) . Hybridization was performed at 48°C for

16 hrs. GABPBl and B2 were isolated by screening a day-8.5 mouse embryo cDNA library (Lee, Mol. Endocrinol. 4, 1034 (1990)) with degenerate oligonucleotides corresponding to peptides 9 and 12 (Fig. 1) . Kinased oligonucleotide probes were hybrided in 6X SSC, IX Denhardt's, 0.05% sodium pyrophosphate, and 100 μg/ml yeast tRNA at 50°C for 14 hours. Washing conditions were 6X SSC, 0.05% sodium pyrophosphate at 55°C. A total of five clones were isolated that hybridized with both oligonucleotide probes. Four of the clones were approximately 2.6 kb and differed only slightly in the length of the 5' untranslated region; these cDNA clones encoded GABPBl.. The fifth cDNA clone was approximately 1.4 kb and differed from the other four at its 3' end; this cDNA clone encoded GABPB2. Four additional cDNA clones corresponding to GABPB2 were subsequently identified. DNA sequencing was by the dideoxy chain termination method (Sanger et al, Proc. Natl. Acad. Sci. USA 74, 5463 (1977)) using Sequenase (U.S. Biochemicals) under conditions suggested by the manufacturer. The complete nucleotide sequences of GABPα, £1 and 62 were determined on both DNA strands using deleted templates or synthetic oligonucleotide primers. Deletions were made using exonuclease III

(Pharmacia) under conditions specified by the manufacturer. The sequences for GABPBl and GABPB2 were identical up to nucleotide 1130 except for a three nucleotide insertion (GTA) at position 828 of GABPBl. Sequencing of four other independent isolates of GABPBl were identical to GABPB2 at this site. Peptides identified by amino acid sequencing of purified GABP are underlined in the deduced amino acid sequences. The dashed lines indicate the sequence in GABPBl not found in GABPB2. The sequence for B2 is shown from the point at which it diverges from Bl.

Figure 3. Tissue distribution of GABP mRNAS. RNA was isolated from various rat tissues (Chingwin et al. Biochemistry 18, 5294 (1979)) and mouse L cells (Chomczynski et al. Anal. Biochem. 162, 156 (1987)). lOμg of poly A+ RNA was separated on a 1% agarose- formaldehyde gel, transferred to Nytran (Schleicher and Schuell) and hybridized with a random-primed probe prepared from GABPα (A) or GABPBl (B) .

Figure 4. Requirement of GABPα and GABPBl for sequence specific DNA binding. A) In vitro translation of GABP proteins. Sense and anti-sense RNAs from GABPα and GABPB were transcribed in vitro and translated in rabbit reticulocyte lysates (Krieg et al, Nucl. Acids Res. 12, 7057 (1984)). Plas ids with cDNAs inserted in the Eco RI and Xho I sites of Bluescript (Stratagene) were linearized with Asp 718 and transcribed with T3 RNA polymerase to generate sense strand RNA, or linearized with Bam HI and transcribed with T7 RNA poly erase to generate anti- sense RNA. RNAs were used to program rabbit reticulocyte lysates in the presence of ³⁵S- methionine under conditions specified by the manufacturer (Promega Biotec) . Unlabeled protein was used for DNA binding experiments. The ³⁵S- methionine labeled products were separated on a 12.5% SDS-polyacrylamide gel and visualized by fluorography; (-) , anti-sense RNA; (+) , sense strand RNA. Positions of molecular weight markers are indicated in kD. (B) Electrophoretic mobility shift assays with in vitro translated GABP proteins. Proteins were incubated in the presence of a ³²P- labeled DNA fragment from the HSV ICP4 promoter and subjected to electrophoresis on a non-denaturing 5% polyacrylamide gel in .5X TBE (Garner et al, Nucl. Acids Res. 9, 3047 (1981); Fried et al, ibid, p. 6505 (1981)). For DNA binding assays, samples containing in. vitro translated protein were incubated in 25 mM Tris pH 8.0, 10% glycerol, 50 mM CK1, 3 mM MgC12, 0.5mM EDTA, ImM DTT, 50 μg/ml poly dldC on ice for 10 minutes, then probe was added and incubation continued at room temperature for 10 minutes. The probe was a 180 bp Nco I-Sal I fragment excised from the herpes simplex virus ICP4 promoter. The fragment was labeled by fill-in with the Klenow fragment of DNA polymerase I in the presence of "P-dCTP. Protein:DNA complexes were subjected to electrophoresis on 5% (30:1) polyacrylamide gels in 0.5X TBE. Radioactive DNA and DNA: rotein complexes were visualized by autoradiography. "B" indicates GABP bound DNA, "E" indicates DNA bound by proteins endogenous to reticulocyte lysates.

Figure 5. Schematic diagram of GABP subunits showing regions of amino acid sequence similarity to related proteins. (Top) GABPα is represented as a rectangle with the NH.-terminus on the left and the COOH-terminus on the right. The region of sequence similarity to Ets-related proteins is shaded (amino acids 316-400) and compared with the sequences of

Ets-1 (Gunther et al. Genes Dev. 4, 667 (1990)), Erg (Reddy et al, Proc. Natl. Acad. Sci. U.S.A. 84, 6131 (1987)), Elk (Rao et al. Science 244, 66 (1989) and E74A (Burtis et al. Cell 61, 85 (1990). Residues that are common to GABPα and other proteins are boxed in black. (Bottom) GABPBl is represented as a rectangle with the NH₂-terminus on the left and the COOH-terminus on the right. The 33-amino acids repeats are shown as shaded rectangles. The unique COOH-terminal segment of GABPBl relative to GABPB2 is indicated in black (333-382) . The sequence of the four 33 amino acid repeats in GABPBl are shown below; residues that are common to two or more repeats are boxed in black and used to derive the GABPB consensus. Similar criteria were used to derive consensus sequences for the 33 amino acid repeats of cdc 10/SW14,6 (Ares et al, EMBO J. 4, 457 (1985); Andrews et al. Nature, 342, 830 (1989); Breeden et al. Nature 329, 651 (1987)), Notch (Wharton et al. Cell 43, 567 (1985); Greenwald, iJid. 583 (1985)), glpl (Yochem et al. Cell 58, 553 (1989); Austin et al ibid. p. 565 (1989)), linl2 (Yochem et al. Nature 335, 547 (1988)), ankyrin (Lux et al ibid. 344, 36 (1990)), NF B Kieran et al. Cell 62, 1007 (1990); Bours et al. Nature 348, 76

(1990)), feml (Yochem et al. Cell 58, 553 (1989; Austin et al ibid. , p. 565 (1989)), and bσl-3 (Ohno et al ibid. , p. 991 (1990)). The repeats from cdclO, SW14 and SW16 were combined to determine the consensus. The consensus for ankyrin was taken from Lux et al. Nature 344, 36 (1990). The overall consensus was defined as residues present in at least 6 of the individual consensus repeat sequences.

Figure 6. DNA binding by GABP expressed in bacteria. Purified proteins were incubated with a "P-labeled oligonucleotide containing the GABP binding site derived from the enhancer of the herpes simplex virus ICP4 gene (5*-

TGCGGAACGGAAGCGGAAACCGCCGGATCG-3') (Triezenberg et al, Genes Dev. 2 (1988); LaMarco et al, ibid . 3, 1372 (1989)). Free and protein-bound DNA samples were subjected to electrophoresis on 5% polyacrylamide gels in either 0.5X TBE (A) or 0.25X TBE buffer (B) .

Figure 7. Characterization of the DNA binding site for GABP. (A) Increasing concentrations of GABPα, either in the absence (left panel) or presence (right panel) of GABPBl were mixed with a "P-labeled DNA fragment derived from the herpes simplex virus ICP4 enhancer. Free and protein-bound complexes were partially digested with DNase I and subjected to electrophoresis on an 8% polyacrylamide sequencing gel. The positions of three purine-rich repeats within the region of DNA protected from digestion by GABP are indicated by arrows. Lanes l- 6 (left panel) show digestion patterns resulting from GABPα concentrations starting at 1.5 nM and decreasing in 3-fold increments to 0.005 nM. Lanes 1-6 (right panel) show patterns resulting from addition of the same concentrations of GABPα that had been supplemented with 0.5 nM of GABPBl. (B)

Methylation protection (left panel) and interference (right panel) assays of DNA binding by GABP. The same DNA fragment used in (A) was incubated with GABPα, GABPBl, or an equimolar mixture of the two subunits, and exposed to dimethyl sulfate (DMS) . Partially methylated DNA was recovered, cleaved with piperidine, and run on an 8% polyacrylamide sequencing gel. For methylation interference assays, DNA was partially methylated, incubated with an equimolar mixture of GABPα and GABPBl, and subjected to electrophoresis on a non-denaturing polyacrylamide gel as described in Fig. 6. Free and protein bound DNA species were recovered, cleaved with piperidine, and electrophoresed on an 8% polyacrylamide sequencing gel. Nucleotide residues closely contacted by GABP are shown in the lower part of (B) . Filled circles identif guanine residues that were protected from DMS by GABP. Methylation of the same four guanine residues also inhibited DNA binding by GABP. Open circles identify adenine residues where methylation is enhanced in the presence of both GABPα and GABPBl.

Figure 8. Measurements of DNA binding stability of complexes formed by various mixtures of GABP subunits. A ^MP-labeled oligonucleotide containing a GABP binding site (Fig. 6) was incubated with GABPα alone, or together with equimolar amounts of either of the two B subunits. After a 10 minute incubation at 24°C, protein:DNA complexes were challenged with a 500-fold excess of unlabeled oligonucleotide.

Protein-bound and free DNA were separated by non- denaturing gel electrophoresis as described in Fig. 6A. Protein-bound and free DNAs were located by autoradiography, excised, and quantitated by scintillation spectrometry. Results are presented as fraction of probe bound, normalized to 1.0 at the start point (time = 0) .

Figure 9. UV-mediated crosslinking of GABP subunits to DNA. Isolated or mixed GABP subunits were incubated with a ^MP-labeled oligonucleotide containing a GABP binding site (Chodosh in Current Protocols in Molecular Biology, Vol. II, Ausubel et al eds (Greene Wiley, New York, 1988)) then exposed to ultraviolet light for varying lengths of time. UV crosslinking was performed using an oligonucleotide composed of a GA binding site flanked by 10 bp of non-specific sequence (5' AACCAAGCTTGCGGAACGGAAGCGGAAACCG 3') corresponding to residues located between 280 and 300 bp upstream of the herpes simplex virus gene encoding ICP4.

Oligonucleotides were labeled to high specific activity by fill-in reaction with the Klenow fragment of DNA polymerase I in the presence of all four ³²P-labeled dNTPs. DNA binding reactions were performed as described in a 96-well culture dish, followed by exposure to ultraviolet light. Samples were boiled in SDS-sa ple buffer and subjected to electrophoresis on SDS-polyacrylamide gels. Crosslinked protein species were visualized by autoradiography. Samples were denatured by boiling in SDS sample buffer and subjected to electrophoresis on a denaturing 12.5% polyacrylamide gel. Following electrophoresis the gel was dried and exposed to X-ray film. Time of exposure to UV light (minutes) is indicated above each gel lane.

Migration of molecular weight marks (kD) is shown on the left.

Figure 10. Glutaraldehyde crosslinking of GABPBl and GABPB2 subunits. Bacterially synthesized proteins were incubated in phosphate buffered saline with varying concentrations of glutaraldehyde as indicated below each lane for five minutes at room temperature. Samples were denatured by boiling in SDS sample buffer and subject to electrophoresis on a denaturing 10% polyacrylamide gel. Following electrophoresis the gel was stained with Coomassie brilliant blue. Proteins present in crosslinking reactions are indicated above each lane. BN110 is a truncated version of GABPBl missing 110 NH₂-terminal residues (see Fig. 12B) .

Figure 11. DNA binding and complex formation assays of deleted variants of GABPα. Top panel shows schematic representation of GABPα deletion mutants. Individual mutants are designated according to the position of deletion end points with respect to the amino acid sequence of GABPα. Prefix "N" designates deletions missing residues starting at the NH₂- terminus of GABPα, prefix "C" designates deletions missing COOH-terminal residues, numbers indicate the position of the amino acid at which the deletion terminates. The Ets-related segment of GABPα is highlighted by grey stippling. Bottom panel shows an autoradiographic image of a non-denaturing gel used to separate DNA:protein complexes formed between variants of GABPα, GABPBl and a ^MP-labeled oligonucleotide that contained a GABP binding site. Each variant of GABPα was tested for DNA binding in the absence and presence of GABPBl as indicated above the individual lanes.

Figure 12. Complex formation and UV crosslinking assays of deleted variants of GABPBl. Top panels of (A) and (B) show schematic representations of GABPBl deletion mutants. Individual mutants are designated according to the positions of deletion end points with respect to the amino acid sequence of GABPBl. Prefix "N" designates deletions from the NH₂- terminus of GABPBl (B) , prefix "C" designates deletions missing COOH-terminal residues. Repeated sequences 33 or 32 amino acids in length that are related to similarly sized repeats in the Notch protein of Drosophila melanoσaster are highlighted by grey stippling. Unique parts of GABPBl and GABPB2 are indicated by black and hatched rectangles at their respective COOH-termini. Deleted formation with GABPα as shown in the lower left panels of (A) and (B) . Each deletion mutant was also tested in UV crosslinking assays shown in the lower right panels of (A) and (B) . All complex formation and UV crosslinking assays were conducted in the presence of GABPα and a "P-labeled oligonucleotide containing a GABP binding site.

Figure 13. Model depicting complex formed between GABP and DNA. The sequence of the GABP binding site consists of two hexanucleotide repeats of the sequence 5'-CGGAAR-3' as in lower part of Fig. 13. Oval spheres directly above guanine residues of each hexanucleotide correspond to GABPα subunits, elongated rectangles correspond to 33 amino acid repeats of GABPB subunits. Smaller rectangles shown at top correspond to the region of GABPBl required for formation of stable homodimers. Circular arrows designate flexible regions inferred to occur between the dimer forming region of GABPBl and the 33-amino acid repeats located at its NH₂-terminus.

DETAILED DESCRIPTION OF THE INVENTION

The present invention relates to a DNA segment encoding all (or a unique portion) of the heteromeric transcriptional regulatory protein termed GA binding protein (GABP) . The invention further relates to the encoded proteins (or polypeptides) . A "unique portion" as used herein consists of at least five (or six) amino acids or, correspondingly, at least 15 (or 18) nucleotides. The present invention further relates to a recombinant DNA molecule comprising the above DNA segment and to host cells transformed therewith.

In particular, the present invention relates to a DNA segment that encodes the entire amino acid sequence of GABPα, GABPBl or GABPB2 given in Figure 2 (the specific DNA segments given in Figure 2 being only examples) , or any unique portion thereof. DNA segments to which the invention relates also include those encoding substantially the same protein subunits as shown in Figure 2 , including, for example, allelic and species variations thereof and functional equivalents of the amino acid sequences of Figure 2. The invention further relates to a DNA segment substantially identical to one of the subunit sequences shown in Figure 2. A "substantially identical" sequence is one the complement of which hybridizes to one of the sequences of Figure 2 at 50°C and 6X SSC (saline/sodium citrate) and which remains bound when subjected to washing at 55°C with 6X SSC (note: 20 x SSC = 3M sodium chloride/0.3 M sodium citrate). The invention also relates to nucleotide fragments complementary to such DNA segments. Unique portions of the DNA segment, or complementary fragments, can be used as probes for detecting the presence of respective complementary strands in DNA (or RNA) containing samples.

The present invention further relates to GABP, and subunits thereof, substantially free of proteins with which it is normally associated, and more especially, to unique peptide fragments of the subunits of that protein. The GABP protein (or functionally equivalent variations thereof) , or peptide fragments thereof, to which the invention relates, also includes those which are chemically synthesized using known methods. The proteins and peptides of the present invention can be modified, for example, phosphorylated, or unmodified.

The present invention also relates to recombinantly produced GABP, or subunits thereof, having the amino acid sequence shown in Figure 2 or functionally equivalent variation thereof. The recombinantly produced protein can be modified, for example phosphorylated, or unmodified. The present invention, more particularly, relates to recombinantly produced unique peptide fragments of GABP subunits.

The present invention also relates to a recombinant DNA molecule (or construct) and to a host cell transformed therewith. Using standard methodologies, well known in the art, a recombinant DNA molecule comprising a vector and a DNA segment encoding at least one GABP subunit, or a unique portion thereof, can be constructed. Vectors suitable for use in the present invention include plasmid and viral vectors. The vector can be selected so as to be suitable for transforming prokaryotic or eukaryotic cells. Advantageously, the recombinant molecule includes a promoter operably linked to the GABP encoding segment.' The recombinant DNA molecule of the invention can be introduced into appropriate host cells by one skilled in the art using method well known in the art. Suitable host cells include prokaryotic cells, such as bacteria, lower eukaryotic cells, such as yeast, and higher eukaryotic cells, such as mammalian cells. These cells can serve as a source of GABP when cultured under appropriate conditions.

As noted at the outset, and as will be further described in the Examples that follow, significant amino acid sequence similarity exists between GABPα and the products of the estl and ets2 (Watson et al, Proc. Natl. Acad. Sci. U.S.A. 85, 7862 (1988); Gunther et al, Genes Dev. 4, 667 (1990)) proto-oncogenes. The Ets-related region of GABPα is located close to the COOH-terminus of the subunit. Biochemical studies of Etsl (Gunther et al. Genes Dev. A , 667 (1990)), as well as the related proteins PU (Klemsz et al, Cell 61, 113 (1990)) and E74 (Urness et al, Cell 63, 47 (1990)), have demonstrated direct, sequence-specific DNA binding. These proteins, as well the products of several additional eukaryotic genes, share sequence similarity in an 85-amino acid region that is required for DNA binding (Karim et al, Genes Dev. 4, 1451 (1990)). The region of GABPα that is related to this family of proteins coincides with the 85 amino acid DNA binding domain (Fig. 5) .

The amino acid sequences of GABPBl and GABPB2 contain four repeats of a related amino acid sequence located at the NH₂-termini of both subunits (Fig. 5) . The first two repeats are 32 amino acids in length and the second two contain 33 amino acids. Similar repeats occur in the Notch protein of Drosophila melanoqaster (Wharton et al, Cell 43, 567 (1985); I. Greenwald, ijid. , 583 (1985)), and the Linl2 and Glpl proteins of Caenorhabditis eleαans

(Yochem et al. Nature 335, 547 (1988), Yochem et al, Cell 58, 553 (1989); Austin et al, iJid. , p. 565 (1989)). These "33-amino acid repeats" were first recognized in studies of the yeast protein SW16, which regulates gene expression involved in mating type switching (Breeden et al. Nature 329, 651 (1987)). Similar repeats have been identified in ankrin, a multifunctional protein associated with the membrane of red blood cells (Lux et al. Nature 344, 36 (1990)), several vaccinia virus encoded proteins of unknown function (Gillard et al, Proc. Natl. Acad. Sci. U.S.A. 83, 5573 (1986)), and the transcription factor NFKB (Kieran et al. Cell 62, 1007 (1990) ;_ Bours et al. Nature 348, 76 (1990)).

The two subunits of GABP exhibit primary sequence motifs typical of proteins normally found in different cellular compartments. Accordingly, transcriptional regulatory proteins, such as members of the Ets family, might interact with membrane bound proteins that contain the 33-amino acid repeats present in GABPB. The Notch, Glpl and Linl2 proteins might sequester transcription factors at the plasma membrane which could be released in response to appropriate extracellular signaling events. Alternatively, the cytoplasmic segments of these transmembrane proteins might be proteolyzed in response to an extracellular signal, allowing the 33-amino acid repeats to be translocated to the nucleus where they could abet the action of a second subunit. Either scenario would offer a direct pathway of signal transduction. The results set forth in the Examples that follow demonstrate the reliance of competent DNA binding complexes on multiple subunits. By separating functional components onto different polypeptide chains, critical subunits might be differentially expressed or sequestered, generating useful strategies for regulation. For example differential expression of the mRNAs encoding the Bl and B2 subunits of GABP would be expected to impact substantially on the function of the resulting complex. In this regard, it is interesting to note that in cells undergoing replication, the B2 subunit predominates whereas in non-dividing cells, subunit Bl predominates.

It will be clear to one skilled in the art from a reading of this disclosure that advantage can be taken of information provided herein to effect alterations of both normal and abnormal expression patterns regulated by the binding complexes described abpve. Such alterations can be effected, for example, using a variety of gene therapy protocols. It is contemplated that by altering the relative amounts of the Bl and B2 subunits, disease states characterized by rapid cell division, for example, cancer, can be controlled.

Certain aspects of the invention are described in greater detail in the non-limiting Examples that follow.

Example 1

Isolation of Recombinant cDNA Clones

GABP (20 μg) was purified from rat liver nuclear extracts and cleaved with trypsin. Proteolyzed fragments were separated by high performance liquid chromatography (HPLC) , recovered, and subjected to gas-phase amino acid sequencing (Fig. 1) . Partial sequences were derived from 13 tryptic peptides. Degenerate oligonucleotides capable of encoding four of the thirteen peptide sequences were synthesized and used as hybridization probes to screen an adipocyte cDNA library. Degenerate oligonucleotides were labeled with "P using polynucleotide kinase. The basic sodium chloride/sodium citrate (SSC) protocol was used for screening (F.M. Ausubel et al. Current Protocols in Molecular Biology (Wil-ey and Sons, NY), 1989). Hybridization was performed at 48°C for 16 hours. One recombinant bacteriophage that contained a cDNA insert of 2 kb gave a positive signal when hybridized with each of the four oligonucleotide probes.

The insert of this recombinant was sequenced and found to contain an opening reading frame that encoded a protein of 454 amino acids (Fig. 2A) . The predicted molecular weight of this polypeptide (51.3 kD) corresponded to the size of the GABPα subunit purified from rat liver nuclei (LaMarco et al, Genes Dev. 3, 1372 (1989); Fig. 1). Inspection of the deduced amino acid sequence revealed segments that corresponded to eight of the 13 peptides isolated by trypsin digestion of intact GABP. On the basis of the latter two observations, this 454 residue polypeptide was tentatively identified as GABPα.

Degenerate oligonucleotides capable of encoding two of the tryptic peptide sequences not present in GABPα were synthesized and used as hybridization probes to search for a cDNA clone that encoded GABPB (Lee, Mol. Endocrinol. 4, 1034 (1990) ) . Kinased oligonucleotide probes were hybrided in 6X SSC, IX Denhardt's, 0.05% sodium pyrophosphate, and 100 μg/ml yeast tRNA at 50°C for 14 hours. Washing conditions were 6X SSC, 0.05% sodium pyrophosphate at 55°C. A total of five clones were isolated that hybridized with both oligonucleotide probes. Four of the clones were approximately 2.6 kb and differed only slightly in the length of the 5' untranslated region; these cDNA clones encoded GABPBl. The fifth cDNA clone was approximately 1.4 kb and differed from the other four at its 3' end; this cDNA clone encoded GABPB2. Four additional cDNA clones corresponding to GABPB2 were subsequently identified. Five recombinant bacteriophage were identified according to their capacity to hybridize with both oligonucleotide probes. One of the cDNA clones differed at the 3• end from the other four. The largest cDNA insert of the four (2.6 kb) and the variant (1.4 kb) were sequenced (Fig. 2B) . Both DNA sequences revealed long open reading frames specifying highly similar polypeptides. One cDNA encoded a protein of 382 amino acids, the other encoded a protein of 349 residues. Starting at their respective NH₂-termini, the two proteins exhibited identical sequences for 333 amino acids. At this point the sequences to the two protein diverged such that the longer one contained an additional 50 residues before its terminus. The divergent COOH-terminal segments bore no apparent amino acid sequence similarity. The open reading frames of both polypeptides contained segments that corresponded to the two tryptic peptides used to design hybridization probes. Moreover, the predicted molecular weights of the two polypeptides (41.3 and 37 kD) corresponded closely with the size of the GABPB subunit purified from rat liver nuclei (LaMarco et al. Genes Dev. 3, 1372 (1989) ; Fig. 1) . The 41 kD polypeptide was therefore provisionally designated as GABPBl and the 37 kD polypeptide as GABB2.

Example 2

Tissue Distribution of mRNA Encoding GABPα, GABPBl and GABPB2

Northern (RNA) blot assays were used to determine the sizes and tissue distributions of mRNA encoding GABPα, GABPBl and GABPB2 (Fig. 3) . The cDNA corresponding to GABPα identified three mRNAs of roughly 5.0, 2.8 and 2.6 kb, which were expressed in a variety of tissues. The GABPα cDNA, which consisted of slightly less than 2.0 kb (Fig. 2A) , represents a partial copy of any of the three mRNAs. Two mRNAs measuring 2.7 and 1.5 kb were identified in northern blots probed with GABPBl cDNA. Like GABPα mRNAs, those encoding GABPBl had a wide tissue distribution. Because the cDNAs that encoded GABPBl and GABPB2 measured 2.6 and 1.4 kb, respectively (Fig. 2B) , they probably represent nearly full- length copies of the respective mRNAs. Moreover, because the nucleotide sequences of the two cDNAs are identical from their respective 5' termini to the point of abrupt divergence 1.1 kb internal to the mRNA, they likely represent alternatively spliced transcripts derived from the same gene. Consistent with this interpretation is the presence of a potential splice donor site (AG dinucleotide) immediately preceding the point of divergence.

Example 3

GABP DNA Binding Activity

To test whether the recombinant DNA clones described above possessed GABP DNA binding activity, reticulocyte lysates were programmed with RNA synthesized from the cDNAs that encode GABPB, GABPBl and GABPB2. Each RNA was translated to form a protein product of the expected size (Fig. 4A) . Individual lysates or mixtures thereof were tested for DNA binding to a fragment from the HSV1 ICP4 promoter that contained three GA repeats. Protein:DNA mixes were subjected to electrophoresis on nondenaturing polyacrylamide gels to separate free DNA from that complexed with protein.

Reticulocyte lysate that had not been programmed with exogenous RNA contained protein(s) capable of forming a complex with the oligonucleotide probe that migrated more rapidly than the complex formed by GABP. Other than background activity endogenous to the reticulocyte lysate, specific protein:DNA complexes were not observed when lysates programmed with GABPα, GABPBl, or GABPB2 were tested in electrophoretic mobility shift assays. Likewise, no new DNA binding activity was observed with lysate that had been used to co-translate RNAs encoding GABPα and GABPB2. However, co-translation of RNAs encoding GABPα and GABPBl caused the lysate to form a DNA binding activity that could be distinguished from background (Fig. 4B) . The interdependency of GABPα and GABPBl observed in these assays is consistent with earlier observations that tested subunits purified from rat liver nuclei (LaMarco et al, Genes Dev. 3, 1372 (1989)).

Example 4

DNA Binding Properties of GABP

Recombinant cDNA copies of the mRNAs that encode GABPα and GABPBl were introduced into bacteriophage T7 based vectors that allowed synthesis of the corresponding proteins in Escherichia coli (Studier et al, J. Mol. Biol., 189, 113 (1986)). Polymerase chain reaction was used to introduce a Bam HI site at the 5' end of the open reading frames encoding GABPα or GABPBl. cDNAs lacking the 3• untranslated region were inserted into a modified pT5 vector, which adds two amino acids (gly-ser) at the NH₂-terminus of the encoded protein. Each subunit was expressed and purified using conventional chromatographic techniques.

Proteins were expressed in bacteria as described (Shuman et al. Science 249, 771 (1990)). GABPα was precipitated from the soluble fraction by the addition of one volume of 2M ammonium sulfate in buffer A (10 mM Tris-HCl, pH 7.9, 100 mM KCl, 0.5 mM EDTA, 1 mM MgCl₂, 2 mM DTT, 0.1 mM phenylmethylsulfonyl fluoride) with 2 mM CaCl₂. The ammonium sulfate pellet was resuspended in 25 ml buffer B (25 mM Tris-HCl, pH 8.0, 0.75 mM EDTA, 10% (v/v) glycerol, 1 mM DTT) with 75 mM NaCl and dialyzed against the same buffer. The dialysate was loaded onto a column of Q-Sepharose Fast Flow (Pharmacia) . GABPα was eluted with a 75-500 mM NaCl gradient in buffer B. Peak fractions were pooled, dialyzed against buffer B and loaded onto a salmon sperm DNA-sepharose column. GABPα was eluted with a 0-400 mM NaCl gradient. GABPα was judged by Coomassie Blue staining of SDS polyacrylamide gels to account for greater than 90% of the total protein. GABPBl was solubilized from the particulate fraction of bacterial extracts by sonication in buffer A supplemented with 7 M urea. The urea solubilized fraction was dialyzed against buffer B with 75 mM NaCl and centrifuged at 16,300 x g for one hour. The supernatant was applied to a Q- Sepharose column and eluted with a gradient of 75- 500 mM NaCl. GABPBl was judged to account for greater than 90% of total protein by Coomassie Blue staining of SDS-polyacrylamide gels.

The DNA binding properties of the two individual polypeptides and mixtures thereof were first studied by gel retardation using a DNA substrate derived from the enhancer of an immediate early gene of herpes simples virus. Consistent with earlier studies (LaMarco et al, Genes Dev., 3, 1372 (1989)), binding was not observed when DNA was incubated with either of the isolated subunits. When GABPα and GABPBl were incubated with DNA simultaneously, a DNA:protein complex exhibiting substantially retarded mobility relative to free DNA was observed (Fig. 6, left panel) .

The multi-subunit dependence of DNA binding by GABP was relieved in gel retardation assays conducted at lower ionic strength (Fig. 6, right panel) . GABPα formed two retarded complexes that migrated at positions between free DNA and the complex formed with both subunits. Under these conditions, the mixture of GABPα and GABPBl again led to the formation of a slowly migrating DNA:protein complex. The retarded mobility of the latter complex, relative to those formed by GABPα alone, reflected the presence of the GABPBl subunit. The validity of this interpretation was confirmed by the use of antisera specific to each subunit. Antiserum specific to GABPα further retarded the migration of complexes formed with GABPα alone or the mixture of GABPα and GABPBl. Antiserum specific to GABPBl did not affect the mobility of complexes formed between GABPα DNA, but retarded the complex formed in the presence of both subunits. Polyclonal antisera were generated by rejecting rabbits with purified GABPα or GABPBl. Antisera were added to gel shift reactions at a dilution of 1:20. Pre- immune sera did not effect the migration of protein:DNA complexes. The HSVl-derived DNA fragment used in binding assays of GABP contains three imperfect repeats of the hexanucleotide sequence 5'-CGGAAR- 3• , which were shown in earlier studies to be protected from DNase I digestion when bound by GABP (Triezenberg et al. Genes Dev. 2 (1988) ; LaMarco et al, ibid. 3 1372 (1989)). DNase I footprinting assays were performed using bacterially synthesized proteins under conditions that allowed interaction of GABPα alone. As shown in Fig. 7A, GABPα was capable of protecting .the repeated hexanucleotide motifs from DNase I digestion when added at a concentration of 0.15 nM. When the GABPBl subunit was added, protection was observed at a 10-fold more dilute concentration of GABPα (0.015 nM) . In addition, the pattern of nuclease protection was extended slightly beyond the adenine residues of the third repeat. The segment of DNA protected from DNase I digestion by GABP encompassed all three hexanucleotide repeats, yet was not centered over the repeats. Methylation protection and interference assays were undertaken in order to gain a more refined image of the sites of close contact established between GABP and DNA. Methylation protection assays conducted with GABPα showed a pattern of protection that included both guanine residues of the second and third hexanucleotide repeats. The same sets of guanines were protected when GABPBl was added to the binding reaction. In the latter case, however, accentuated methylation was observed at adenine residues located adjacent to the guanine dinucleotides of the second and third repeats. Sites of methylation interference were mapped by separating protein-bound DNA molecules from those inactivated by partial methylation. Methylation of guanine dinucleotides in the second and third hexanucleotide repeats inhibited binding by the mixture of GABPα and GABPBl (Fig. 7B) .

The results of methylation protection and interference assays indicate that GABP binds to sites on DNA corresponding to two of the three purine-rich hexanucleotide repeats. The pattern of protection of guanine residues by GABPα was similar to that observed with the mixture of both subunits, indicating that GABPα, when added at a sufficiently high concentration, can bind specifically to DNA in the absence of GABPBl._. Such observations offer an explanation for the two retarded bands observed when DNA was challenged with GABPα alone under conditions of low ionic strength (Fig. 6, right panel) . The less retarded of the two bands is interpreted to represent a complex wherein GABPα is associated with only one of the two hexanucleotide repeats, while the more retarded complex is interpreted to contain GABPα subunits associated with two hexanucleotide repeats. Binding assays that tested DNA probes containing a single hexanucleotide repeat supported this interpretation. When incubated with GABPα and assayed in low ionic strength gels, such DNA probes generated only one retarded complex.

Several observations indicate that the mixture of GABPα and GABPBl forms a complex that binds DNA more stably than the α subunit alone (LaMarco et al. Genes Dev. 3, 1372 (1989)). To further investigate the effect of GABPBl on DNA binding by GABPα, the rate at which variously mixed proteins dissociate from DNA was measured (Fig. 8) . The dissociation rate of GABPα alone was too rapid to be accurately measured. Less than ten percent of the DNA remained bound to GABPα after a 10 second challenge with excess, unlabeled competitor DNA. In contrast, when both GABPα and GABPBl were present, the dissociation rate was much slower (T_1/2 = 1.5 min) . Similar assays were performed with a mixture of GABPα and GABPB2, which, in earlier experiments, failed to form a stable complex with DNA. When used at nM concentrations, GABPB2 was capable of forming a DNA binding complex with GABPα (see Fig. 12A) . The B2 isoform of GABP also stabilized DNA binding by GABPα, yet yielded a complex that dissociated more rapidly (T_I/2 = 30 s) than that formed with GABPBl. Because the Bl and 62 isoforms differ only at their COOH-termini, this part of the protein may be involved in stabilizing DNA binding.

The observations outlined thus far indicate that the Bl and B2 isoforms of GABP do not bind to DNA alone, but associate with the α subunit to augment DNA binding. The question arises, therefore, as to whether the B subunits cause a conformational change in α leading to its more avid interaction with DNA. Alternatively, or in addition to causing a conformational change, the B subunits might, in association with GABPα, establish direct contact with DNA.

To determine whether GABPBl contacted DNA when complexed with GABPα, DNA:protein complexes were exposed to ultraviolet (UV) light under conditions expected to permit covalent crosslinking between DNA and intimately bound proteins (L. A. Chodosh in Current Protocols in Molecular Biology. vol II, F. M. Ausubel et al., eds. (Greene/Wiley, New York, 1988) . UV crosslinking was performed using an oligonucleotide composed of a GA binding site flanked by 10 bp of non-specific sequence ((5' AACCAAGCTTGCGGAACGGAAGCGGAAACCG 3') corresponding to residues located between 280 and 300 bp upstream of the herpes simplex virus gene encoding ICP4. Oligonucleotides were labeled to high specific activity by fill-in reaction with the Klenow fragment of DNA polymerase I in the presence of all four ³²P-labeled dNTPs. DNA binding reactions were performed as described in a 96-well culture dish, followed by exposure to ultraviolet light. Samples were boiled in SDS-sample buffer and subjected to electrophoresis on SDS-polyacrylamide gels. Crosslinked protein species were visualized by autoradiography. The GABP subunits were incubated with "P-labeled DNA that contained the purine-rich hexanucleotide repeats, exposed to UV light, and subject to electrophoresis on a denaturing polyacrylamide gel. When DNA was incubated with

GABPα and exposed to UV light, a crosslinked product was observed bearing an electrophoretic mobility close to that of GABPα (Fig. 9) . The appearance of this product was dependent on the presence of GABPα and increased in a time-dependent manner upon exposure to UV light. Moreover, it was eliminated by the inclusion of excess, unlabeled DNA that contained the purine-rich hexanucleotide repeats, but not by excess non-specific DNA.

No evidence of protein:DNA crosslinking was observed when ³²P-labeled DNA was mixed with GABPBl and exposed to UV light. However, when the mixture of GABPα and GABPBl was complexed with DNA and irradiated, new crosslinked products were observed. In addition to GABPα, two closely migrating polypeptide bands, slightly larger than the native size of GABPBl, became covalently attached to the radioactive DNA substrate (Fig. 9) . Although GABPBl was incapable of binding DNA on its own, when present in a ternary complex it appeared to be even more susceptible to UV-mediated crosslinking than GABPα. These data provide evidence that the Bl subunit of GABP associates closely with DNA when complexed with GABPα.

Example 5

Formation of a Stable Complex between GABPα and GABPB in the Absence of DNA

Having found that the α and B subunits of GABP formed a heteromeric complex when exposed to their specific DNA substrate, gel filtration chromatography was used to determine whether these subunits might associate in the absence of DNA. Gel filtration chromatography was performed using a Superose-6 column (10 x 300 cm, Pharmacia) in buffer B supplemented with 0.4M NaCl. The column was calibrated with molecular weight standards thyroglobulin, apoferritin, catalase, bovine serum albumin and ribonuclease. 50-100 μg of each protein was chromatographed at 0.5 ml/min. Elution volume was converted to K„ by the equation K_av = (V_e-V₀)/V_t- V₀) where V₀ = void volume = 8.1 ml; V_t + total bed volume = 24.0 ml; V - eluted volume. The Stokes radius was calculated from a plot of (-log K_av) ⁿ versus Stokes radius (Ackers, Adv. Prot. Chem. 24, 343 (1970)). GABPα eluted as a single peak at 15.2 ml; GABPBl at 14.0 ml; GABPB2 at 15.8 ml. A mixture of equal amounts of GABPα and Bl chromatographed as a single peak at 12.1 ml. The mixture of GABPα and GABPB2 chromatographed as a single peak at 13.9 ml. Loaded separately, GABPα and GABPBl eluted as single peaks at K_av values of 0.45 and 0.38 respectively.

However, when loaded together at an equimolar ratio, both subunits eluted as a single peak at K_av 0.26 (Table 1) . Analysis of column fractions by polyacrylamide gel electrophoresis confirmed that the peak at K_av 0.26 contained both subunits.

TABLE 1

Determination of molecular weights of GABP subunits. Purified GABPα, GABPBl and GABPB2 produced in E. coli were analyzed by gel filtration and sedimentation velocity as described (17,18). K_av values were calculated from the elution volume of a Superose-6 FPLC column. Apparent molecular weights were determined from K_av vs. log MW for the column. Stokes radii were determined from a plot of (-log K_av) . Sedimentation coefficients together with measured Stokes radii were used to calculate native molecular weights.

29 Evidence confirming the stable association of GABP subunits in the absence of DNA was also obtained from measurements of sedimentation velocity (Table 1) . The gel filtration and sedimentation properties of a protein or protein complex are affected both by size and molecular shape. However, by using the analytical methods of Siegel and Monty (Martin et al., J. Biol. Chem. 236, 1372 (1961)), it was possible to calculate native molecular weights of the various protein species. Sedimentation coefficients were determined on 4.5 ml 10-30% glycerol gradients in 25 mM Tris-HCl pH 8.0, 75 mM NaCl, 0.75 mM EDTA, 1 M DTT. 30 μg of each protein was loaded in 0.1 ml together with catalase, bovine serum albumin, and cytochrome c as internal standards. Gradients were centrifuged at 4°C for 40 hours at 39,000 rpm. Fractions (0.25 ml) were collected and analyzed by SDS-PAGE with Coomassie blue staining. The S value for each sample was determined by its sedimentation relative to the BSA and cytochrome c standards. Native molecular weights were derived using the Stokes radius together with measured sedimentation coefficients as described in Siegel et al, (Biochim. Biophys. Acta 112, 346 (1966)). Partial specific volume was calculated using the predicted amino acid sequences of each GABP subunit as described by Cohn et al. Proteins, Amino Acids and Peptides as Ions and Dipolar Ions, E. J. Cohn and J. T. Edsall, eds. (Reinhold Publishing Co., New York, 1943), pp. 370- 381) . As shown in Table 1, the calculated molecular weights of GABPα and GABPB2 corresponded closely to their predicted sizes (51.3 and 37 kD, respectively) . In contrast, GABPBl exhibited a native molecular weight (82 kD) roughly twice its expected size (41.3 kD) . The complex formed between GABPα and GABPBl eluted from the gel filtration column prior to the largest molecular weight standard. Thus, the calculated molecular weight of this complex (170 kD) represents a provisional assignment. Since GABPBl existed as a stable dimer on its own, the very large complex formed between it and GABPα was tentatively identified as a tetramer composed of two molecules of each subunit. Polyacrylamide gel analysis of the constituents of the GABPα:GABPBl complex were consistent with this interpretation, showing that the subunits existed in equal stoichometries. This interpretation was also consistent with the properties of the complex formed between GABPα and the B subunits was interpreted to reflect the dimer forming property of GABPBl.

Gel filtration and gradient sedimentation assays indicated that GABPBl might exist as a dimer. This interpretation was tested using glutaraldehyde crosslinking assays. Bacterially expressed GABPBl and GABPB2 were exposed to glutaraldehyde and subjected to electrophoresis on a denaturing polyacrylamide gel. Incubation of GABPBl with glutaraldehyde led to the formation of a second polypeptide band exhibiting an apparent molecular weight roughly double that of the monomeric form of the protein (Fig. 10) . Similar experiments conducted with GABPB2 failed to yield an analogous product. Additional evidence suggesting that GABPBl exists as a dimer resulted from crosslinking experiments with the intact polypeptide and a truncated form lacking the 110 NH₂-terminal residues (BN110, see Fig. 12B) . Crosslinking of the truncated protein led to the formation of an additional species roughly double the size of the monomeric form. When the truncated protein was mixed with intact GABPβl and exposed to glutaraldehyde, three crosslinked protein species were observed. Two species corresponded to crosslinked, homodimeric complexes that had been observed upon glutaraldehyde treatment of active GABPBl and the NH₃-terminal truncated derivative. The third species migrated between the presumed homodimeric forms and probably represented a heteromeric complex consisting of one GABPBl polypeptide and one truncated polypeptide.

Therefore, both molecular weight measurements and crosslinking assays showed that GABPBl but not GABPB2 exists as a stable homodimer.

Example 6

Mapping of Functional Domains of GABPα and GABPB

Experimental results described above indicate that GABPα should contain at least two functional components, one that facilitates DNA binding and another that allows complex formation with GABPB. The GABPBl polypeptide should contain at least three components, facilitating self- dimerization, heterodimerization with GABPα, and direct contact with some part of the purine-rich DNA substrate. Recombinant copies of the genes that encoded each subunit were systematically deleted to localize these components. Deletion mutants of GABPα were generated by polymerase chain reaction and expressed in pT5 as described (Breeden et al. Nature 329, 651 (1987)). Soluble bacterial extracts containing deleted variants of GABPα were used for binding reactions. NH₂-terminal deletions of GABPBl were generated by exonuclease III digestion, followed by digestion with SI nuclease and ligation of Bam HI linkers. All deletions were sequenced and subcloned into the appropriate pET3 vector (Rosenberg et al., Gene 56, 125 (1987)) to maintain the proper reading frame. COOH-terminal deletions were generated using 3' deletions of the cDNA inserted in Bluescript (Stratagene) by subcloning

Bst EII-Asp 718 or Sac I-Asp 718 fragments into the pET-GABPBl plasmid that had been digested with the appropriate enzymes. Translation termination codons were provided by vector sequences so that in some cases extra amino acids are appended to the open reading frame. All GABPB derivatives were insoluble and re-solubilized in 8M urea followed by dialysis against 10 mM Tris pH 8.0, 75 mM KCl or NaCl, 1 mM DTT, 0.2mM PMSF, ImM benzamidine, 10% glycerol prior to use in binding reactions. All derivatives were expressed equivalently as determined by Coomassie staining of SDS polyacrylamide gels.

Deletion variants of GABPα that were missing as many as 313 residues from the NH₂-terminus retained the capacities to bind DNA and complex with GABPB (Fig. 11) . A GABPα variant further missing 17 residues (αN313/437) from the COOH-terminus also retained both functions. More extensive deletion from the COOH-terminus, to amino acid 407 (αN313/C407) , yielded a protein that was capable of binding to DNA, but had lost the ability to complex with GABPBl. These results showed that the Ets- related domain of GABPα was sufficient for DNA binding. The region of GABPα required to form a complex with GABPB included the Ets-related segment, as well as 37 amino acids located on the immediate COOH-terminal side of the Ets-related domain.

In order to define regions of GABPBl that interact with GABPα and contact DNA, systematically deleted variants were produced and tested in gel retardation and UV-crosslinking assays (Fig. 12). Variants that lacked up to 228 residues (BC154) from the COOH-terminus of GABPBl proved to be functional in both assays. Deletion of an additional 33 residues (BC121) yielded a protein that failed to function in either assay. The boundary defined by these experiments corresponded to the location of the most COOH-terminal of four 33-amino acid repeats that are present in both isoforms of GABPB.

Although about 70% of GABPBl could be deleted from its COOH-terminus without eliminating interaction with GABPα and DNA, removal of only a small segment from the NH₂-terminus resulted in deleterious effects. A variant of GABPBl that lacked 19 NH₂-terminal residues (BN19) was slightly less effective in converting GABPα-derived complexes into the very slowly migrating heteromeric complex. When tested in the UV crosslinking assay, BN19 yielded a reduced amount of crosslinked product relative to the intact Bl isoform. Variants that lacked 47 and 67 residues (BN47 and BN67) were progressively more defective in the complex formation assay and failed to be crosslinked to the radioactive DNA probe as efficiently as the intact protein. Finally, variants missing 80 or more residues from the NH₂-terminus were completely defective in both assays. The progressive loss of function observed in deleted forms of GABPBl corresponded to the progressive loss of the 33-amino acid repeats (Fig. 12) . BN19 was truncated within the first of the repeats, BN47 within the second, BN67 after the second,, and BN80 within the third repeat. The functional properties of the GABPBl deletion mutants indicate that the 33-amino acid repeats are important both for complex formation with GABPα and DNA contact. Example 7

A Model for DNA-Bound GABP

The foregoing observations have been incorporated into a provisional model of the complex formed when GABPα and GABPBl associate with a directly repeated set of purine-rich hexanucleotides (Fig. 13) . Each hexanucleotide repeat is hypothesized to be contacted by both GABPα and GABPBl. The linear order of contact, wherein GABPα is associated with guanines on one side of each hexanucleotide and GABPB with adenines on the other side, was deducted from three separate observations. First, GABPα was alone capable of protecting both guanines from methylation by dimethysulfate. Second, the DNase I footprint generated by the mixed subunits, relative to that resulting from GABPα alone, was extended in a direction toward the adenine residues of the hexanucleotide repeat. Third, addition of GABPB caused an enhanced pattern of methylation of adenine residues relative to the pattern generated in binding reactions that contained GABPα alone.

The most stable complexes formed between GABP and DNA were observed with the mixture of GABPα and GABPBl. The βl subunit, unlike B2, was observed to exist as a stable homodimer. Moreover, when mixed with GABPα, the Bl subunit generated a high molecular weight complex probably consisting of two polypeptides of each subunit. This heteromeric tetramer is believed to bind in a concerted manner to two purine-rich hexanucleotide repeats. If that is the case, a flexible region should exist between the dimerization domain of GABPBl and the surfaces located near its NH₂-terminus that facilitate interaction with GABPα and DNA. Without such flexibility a linked set of polypeptides would not likely be capable of binding simultaneously to a DNA substrate that is not rotationally symmetric.

Example 8

Isolation of cDNA Clones Encoding Human GABP

Alpha and Beta Subunits

cDNA clones encoding human GABP alpha, beta3 and beta4 were isolated by screening a human fetal brain cDNA library. cDNAs for human GABP betal and beta2 were isolated from a HeLa cell cDNA library. The probes used to screen for human GABP alpha were the 865 bp Ava 1-Sst 1 and 678 bp Bam Hl- Sst 1 fragments of the mouse GABP alpha cDNA. The probe used to isolate the human beta was an 850 base pair fragment from the 5' end of the mouse GABP beta2 cDNA.

Purified DNA fragments used as probes were radiolabeled with 32P by random priming reactions. Hybridization conditions were 6X SSC, IX Denhardt's, 0.05% sodium pyrophosphate, 100 μg/ml yeast tRNA. The final wash buffer was 2 X SSC. The hybridization and washing temperature were 65° C. cDNAs were confirmed as the human homologs of mouse GABP by determination of their nucleotide sequence.

Six different cDNAs have been deposited with the American Type_. Culture Collection (ATCC) , 12301 Parklawn Drive, Rockville, MD 20852.

1A = human GABPB, Eco RI fragment common to all beta isoforms, in Bluescript KS+, isolated from HeLa cDNA library. A = human GABPBl Eco RI fragment of Bluescript KS+, isolated from HeLa cDNA library. 5A = human GABPB2, Eco Rl fragment in Bluescript KS+, isolated from HeLa cDNA library.

F = human GABPB3, Eco Rl-Xho l fragment in Bluescript SK-, isolated from human fetal brain cDNA library.

J = human GABPB4, Eco Rl- Xho 1 fragment in Bluescript SK-, isolated from human fetal brain cDNA library.

G = human GABPα, Eco Rl- Xho 1 fragment in Bluescript SK-, from human fetal brain cDNA library.

The entire contents of all documents cited herein are hereby incorporated by reference.

While various aspects of the invention have been described in some detail for purposes of clarity and understanding, one skilled in the art, from a reading of this disclosure will appreciate that various changes can be made in form and detail without departing from the true scope of the invention. One skilled in the art will also appreciate that the invention includes the human counterparts of the sequences specifically disclosed herein as well as the encoded amino acid sequences, and fragments thereof. The invention also relates to the deposited human sequences and fragments thereof as well as to the full length sequences of which the partial sequences form a part.

Claims (13)

WHAT IS CLAIMED IS:

1. A DNA segment encoding a subunit of GA binding protein (GABP) , or an epitope specific thereto, or a DNA fragment complementary to said DNA segment.

2. The DNA segment according to claim 1, wherein said GABP is human GABP.

3. The DNA segment according to claim 1 wherein said subunit is GABPα.

4. The DNA segment according to claim 1 wherein said subunit is GABPBl.

5. The DNA segment according to claim 1 wherein said subunit is GABPB2.

6. The DNA segment according to claim 3 wherein said subunit has the amino acid sequence shown in Figure 2A.

7, The DNA segment according to claim 4 wherein subunit has the amino acid sequence shown in Figure 2B.

8. The DNA segment according to claim 5 wherein said subunit has the amino acid sequence shown in Figure 2B.

9. A recombinant DNA molecule comprising: i) said DNA segment according to claim 1; and ii) a vector.

10. A host cell stably transformed with said recombinant DNA molecule according to claim 9.

11. The host cell according to claim 10 wherein said cell is a procaryotic cell.

12. The host cell according to claim 10 wherein said cell is a eucaryotic cell.

13. A method of producing a recombinant GABP subunit protein, or portion thereof defining at least an epitope specific thereto, comprising culturing said host cell according to claim 10 under conditions such that said segment is expressed and said protein thereby produced, and isolating said protein.