Purification and Characterization of a 43-kDa Transcription Factor Required for Rat Somatostatin Gene Expression*

A 43-kDa DNA binding protein which recognizes the TGACGTCA element of the rat somatostatin promoter has been purified from rat brain. Purification of the protein involved initial separation of three sequence- specific binding activities, bl-b3, from each other using DEAE-Sepharose chromatography. The protein corresponding to the b2 complex was further purified to apparent homogeneity by two cycles of sequence- specific DNA affinity chromatography, yielding a single species with an apparent mass of 43,000 daltons on a silver-stained polyacrylamide gel. Sequence-specific DNA binding of this purified protein was demonstrated by Southwestern blotting, renaturation, and DNase I footprinting studies. The 43-kDa protein was phosphorylated on serine residue(s) by the catalytic subunit of CAMP-dependent protein kinase, as shown by phospho- amino acid analysis. Furthermore, the purified protein specifically stimulated transcription from the rat somatostatin promoter in an in vitro transcription sys- tem. These results indicate that this 43-kDa protein is a transcription factor required for somatostatin gene expression. Complex variations in gene expression are regulated, in part, at the level of transcription initiation. One approach to the study of transcription regulation has been to isolate and characterize trans-acting factors that can activate specific genes (1, 2). Some of these examples include the activator protein (3), the CAAT-binding transcription factor (4), and the chicken ovalbumin gene upstream promoter binding protein (5).

7 To whom reprint requests should be addressed. DK 07532.
plexes, designated bl, b2, and b3, have been identified from extracts of HeLa cell, rat brain, as well as CA-77 cells (10). bl, b2, and b3 show competitive binding with the nonlabeled probe, but not with a variety of unrelated DNA fragments including the adenovirus major later promoter and a-globin promoter (-70 to -51) or other DNA sequences found within the 5' region of the somatostatin gene. The proteins corresponding to bl, b2, and b3 bind to the same region of somatostatin promoter from -59 to -38, which includes a CAMP consensus sequence from -48 to -41 (TGACGTCA). The G and C residues within the consensus sequence are important for protein-DNA interaction, as determined by methylation interference analyses and further confirmed by single residue mutagenesis (10). Single mutation in any of G or C residues in the region from -48 to -41 resulted in a total loss of binding activity, while substitution T to G in position -50 had no apparent effect on its DNA-protein interaction.
The importance of developing a purification procedure for a single TGACGTCA binding protein is underscored by noting that this DNA structural motif has been found in numerous other CAMP-responsive genes, such as human proenkephalin (ll), rat phosphoenolpyruvate carboxykinase (12), and human chorionic gonadotropin (13). We have purified a protein corresponding to b2 activity from rat brain by DEAEchromatography and DNA affinity column. The affinity-purified material showed a single band of 43 kDa on a silverstained SDS' gel. The specific DNA binding activity of this 43-kDa protein was confirmed by Southwestern blotting, renaturation, and DNase I footprinting analyses. Furthermore, the purified protein specifically stimulated transcription from the rat somatostatin promoter in an in uitro transcription system. These results demonstrate that the 43-kDa protein is a transcription factor required for rat somatostatin gene expression.

EXPERIMENTAL PROCEDURES
Gel Retardation Assay-Gel retardation assays were performed as previously described (10). The labeled probe used in the assay was a double-stranded synthetic oligonucleotide corresponding to positions -70 to -29 of the rat somatostatin promoter. The binding reactions were incubated in a volume of 20 pl at room temperature for 30 min, followed by running a 4% polyacrylamide gel electrophoresis. The gels were then fixed, dried, and autoradiographed.
Quuntitation of Specific DNA Binding Activity by Southwestern Blot Analysis-The southwestern blot analysis was performed as described by Silva et al. (17) with the following modifications. Briefly, after protein samples were electrophoresed on SDS-PAGE, the gel was incubated for 2 h in renaturation buffer (50 mM NaCI, 10 mM Tris-HCI, pH 7.2, 20 mM EDTA, 0.1 mM DTT, and 4 M urea). The proteins were electrophoretically transferred onto a nitrocellulose The abbreviations used are: SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; DTT, dithiothreitol; Hepes, 442-hydroxyethy1)-1-piperazineethanesulfonic acid; MLP, major late promoter.

6551
filter. The filter was then incubated with binding buffer (5% nonfat dry milk, 0.1 mM KC1, 10 mM Hepes, pH 7.9, 1 mM EDTA, 5 mM MgC12, and 0.1% Nonidet P-40) for 1 h. 1 X lo6 cpm of 32P-labeled DNA probe (1 X 10' cpm/pg DNA) was then added and gently agitated at room temperature for 3 h (the probe was the same as that used for the gel retardation assays). After DNA binding, the nitrocellulose filter was washed at room temperature in the binding buffer for 30 min with three buffer changes and autoradiographed overnight. The Southwestern blot has been used to quantitate the increase in specific activity of the 43-kDa protein during purification procedure. The overnight-exposed films were scanned to quantitate the intensity of bands corresponding to the 43-kDa protein by using LKB2222-010 ultroscan XL laser densitometer and 2400 gel scan XC program. One activity unit (unit) is arbitrarily defined as the intensity of a 43-kDa DNA band on the Southwestern blot from 1 mg of protein of the (NH4)2S04-fractionated rat brain extract.
Preparation of Oligonucleotide-Sephrose-The oligonucleotide-Sepharose was prepared according to the procedure described by Kadonaga and Tjian (3). The sequence of the monomer is shown in Fig. 1B. The monomer was ligated to form concatemers ranging in size from 3-mers to approximately 30-mers. The efficiency of DNA coupling to the resin was 50"70%. The affinity matrix was estimated to contain 12 pg of DNA/ml of Sepharose resin.
Preparation of Cell Extracts-Rat brain cell extracts were prepared according to Manley et al. (14) with slight modifications. Approximately 100 g of brain tissue was obtained from 50 adult rats. After addition of 400 ml of 0.01 M Tris-HC1, pH 7.9, 1 mM EDTA, and 5 mM DTT, the tissue was homogenized sequentially using a Polytron and Dounce homogenizer. Subsequently, 400 ml of 0.05 M Tris-HC1, pH 7.9, 10 mM MgC12, 2 mM DTT, 25% sucrose, and 15% glycerol was added, and the mixture was stirred gently. To this suspension, 100 ml of saturated (NH4)2S04 was added dropwise. After stirring for 20 min the extract was centrifuged at 27,000 X g for 30 min. The supernatant was removed and fractionated again by the addition of solid (NH&SO4 (0.33 g/ml of supernatant). After stirring for 30 min, the precipitate was collected by centrifugation under the same conditions, resuspended, and dialyzed overnight against 30 mM KC1 in buffer A (10 mM Hepes, pH 7.9,l mM DTT, 1 mM EDTA, and 5 mM MgCM. Chromatography-The above extract (135 ml) was loaded onto a 90-ml DEAE-Sepharose CL-GB column equilibrated with 30 mM KC1 in buffer A. The column was eluted with a linear KC1 gradient ranging from 30 to 300 mM. Finally, the column was washed with 500 mM KC1 in buffer A. Protein concentration of each fraction was measured by absorbance at 280 nm and DNA binding activity was examined by gel retardation assays.
The within the somatostatin promoter is indicated by the numbers (+1 site is the transcription initiation point). The underlined region is a CAMP-responsive sequence, which is involved in the binding with trans-acting factor(s), as indicated by methylation interference and footprint studies (10). and dialyzed against 0.2 M KC1 in buffer B (10 mM Hepes, pH 7.9,lO mM EDTA, 5 mM DTT, 5 mM MgC12,0.1% Nonidet P-40, and 20% glycerol). Sonicated and denatured calf thymus DNA was added to the sample at a final concentration of 10 pg/ml. The material was then loaded onto an oligonucleotide-Sepharose column that had been equilibrated with 0.2 M KC1 in buffer B. After the column was washed with 10 column volumes of 0.2 M KC1 in buffer B, the specific DNA binding activity was recovered with a 1 M KC1 in buffer B step elution. The eluate was diluted with 4 volumes of buffer B and reapplied to the same affinity column. The specific DNA binding activity, which was eluted with 1 M KC1 in buffer B, was then dialyzed against 0.1 M KC1 in buffer B and stored at -80 "C.
DNase Z Footprinting Assay-The procedure for the DNase I footprinting assay has been previously described (10). The HindIII-  ' One activity unit (unit) is arbitrarily defined as the DNA binding activity of 43-kDa protein from 1 mg of the (NH4)2S04-fractionated rat brain extract. The DNA binding activity was measured by scanning the intensity of the hybridization signal on the Southwestern blot (also see Fig. 5A).
* The cell extract was derived from 50 rat brains.
noncoding strand 3. DNase I footprinting assay. Footprinting assays were performed in the presence ( l a n e 2) and absence (lanes 1  BamHI fragment used in the footprinting reactions extended from -100 to +50 of the somatostatin promoter. The DNA, labeled on the noncoding strand, was incubated with the purified 43-kDa protein followed by treatment with DNase I. The reaction products were analyzed on an 8% sequencing gel. SDS-PAGE-SDS-PAGE was performed as described by Laemmli (15). High molecular mass protein standards (ranging from 18,400 to 200,000 daltons) were used as markers. Proteins were visualized after the gel was silver-stained (16).
Renaturation of the 43-kDa Protein from SDS-PAGE-The affinity column-purified material was mixed with 1% SDS and heated at 54°C for 15 min before being subjected to electrophoresis. The gel slices spanning the entire lane were cut out and crushed following extraction overnight in 50 mM Tris-HC1, pH 7.9, 0.1 mM EDTA, 5 mM D m , 150 mM NaCl, 0.1% SDS, and 0.1 mg/ml bovine serum albumin. After the extract was precipitated with 5 volumes of acetone, the precipitates were then dissolved in 6 M guanidine HCl. The mixtures were incubated at room temperature for 20 min and dialyzed overnight against 20 mM Hepes buffer, pH 7.9, containing 20% glycerol, 0.1 mM EDTA, 5 mM MgC12, 2 mM DTT, and 100 mM KCl. The renatured samples were tested for the DNA binding activity by DNase I footprinting assay.
Phosphorylation Assay-Protein samples were incubated with varying amounts of the catalytic subunit of CAMP-dependent protein kinase in 10 mM Hepes, pH 7.9, and 10 mM MgCl2 at 30°C for 5 min. The reaction was stopped by the addition of 1% SDS, and the product was subjected to SDS-PAGE.
In Vitro Transcription-Preparation of HeLa cell extracts and somatostatin promoter transcription activity-depleted extracts will be described elsewhere (19). Briefly, the HeLa cell extract for in vitro transcription was prepared according to Shapiro et al. (20). The TGACGTCA binding activity depleted extract was prepared by passing the HeLa extract through a DNA affinity column. The column was made by coupling biotinylated oligonucleotides containing the -60 to -35 region of the somatostatin promoter to an avidin-Sepharose support. Transcripts from the somatostatin promoter were monitored by primer extension analysis. The adenovirus major late promoter was used as an internal control.

RESULTS
Purification of the 43-kDa Protein-Previous studies have identified three sequence-specific DNA-protein complexes formed when extracts of CA-77, HeLa, or rat brain cells were incubated with the -70 to -29 region of the rat somatostatin promoter (10). The availability of large quantities of rat brain made it the tissue of choice in purification of the DNA binding proteins. The purification procedures are schematically illustrated in Fig. 1A. Rat brain extract was prepared according to the procedure of Manley et al. (14). The 10-65% (NH4)2S04-fractionated material, which contained all DNA binding activities, was subject to a DEAE-Sepharose column eluted with a linear 30-300 mM KC1 gradient ( Fig. 2A). Gel retardation assays were used to detect DNA binding activities after the DEAE-chromatography. While b l activity was in the flow-through part, b2 and b3 were eluted off the column at about 100 and 180 mM KCl, respectively (Fig. 2B). The b2containing fractions gave a highly resolved DNA-protein band in the gel retardation assay and also appeared to be the major binding activity. Therefore, subsequent fractionation steps focused upon purification of the b2 protein by means of DNA affinity chromatography.
Synthetic oligonucleotides used for the affinity column spanned the CAMP-responsive element, TGACGTCA, and contained 8 base pairs of flanking DNA on each side of the sequence exhibiting dyad symmetry (Fig. 1B). The DNA affinity column was constructed by self-ligating the doublestranded monomer to molecules ranging in size from 3-mers to approximately 30-mers. This mixture was coupled to cyanogen bromide-activated Sepharose (3). The coupling efficiency was approximately 50-70%, resulting in the covalent attachment of about 12 pg of DNA/ml of resin.
The DEAE-Sepharose fractions containing the b2 activity were applied to the DNA affinity column in the presence of 10 pg/ml calf thymus DNA added as a nonspecific competitor. The column was extensively washed with 10 column volumes of 0.2 M KC1 in buffer B. The presence of 0.2 M KC1 significantly reduced the amount of minor protein contaminants in the final purification step. The column was then washed with a stepwise gradient of KC1 in buffer B. The DNA binding protein was eluted in the 1 M KC1 fraction. Further purification was obtained by reapplying the sample to the affinity column. Nonspecific competitor DNA was not added during the second affinity chromatographic step, because the addition of the nonspecific competitor caused a loss of specific DNA binding activity. After two passes over the affinity column the final material was shown to be a single band on silver-stained SDS-PAGE with a molecular mass of approximately 43,000 daltons (Fig. 2C). Because of low quantity of the purified material, its concentration was estimated by the ratio of its intensity and that of a known protein standard on a silver-stained gel. The amount of purified 43-kDa protein was thus estimated to be approximately 3.8 pg obtained from 670 mg of rat brain extract (50 rats), corresponding to about a 1.8 X 106-fold purification on a protein basis.
To quantitate the increase in specific DNA binding activity during purification Southwestern blots were performed. The intensities of labeled DNA. 43-kDa protein complex from each sample of the purification steps were scanned by densitometry (Fig. 5A). One unit of activity is defined as the intensity of the band shown by 1 mg of protein in rat brain extract. Using the arbitrarily defined activity unit, 20% of total DNA binding activity was recovered through the purification and a 3.5 x 104-fold increase in specific activity was achieved (Table I).
DNA Binding Activity of the 43-kDa Protein-To demonstrate the specificity of binding and to define the DNA sequence protected by the affinity-purified material, DNase I footprinting experiments were performed using the DNA fragment spanning the region from -150 to +50 of the rat somatostatin gene. The protected region spans the sequence between nucleotides -59 to -35, identical to that protected by rat brain extracts (Fig. 3). To demonstrate that DNA protection resulted from the binding of the 43-kDa protein -----2 5 K 1rather than from minor contaminants in the affinity-purified material, the DNA affinity-purified material was resolved on a 7.5% gel. Each gel slice was cut and extracted and proteins from the gel slices were precipitated with acetone and renatured in the presence of 6 M guanidine HC1. The renatured protein samples were then examined in DNase I protection assays described above. DNA protection was observed from the gel slice four, which contained the 43-kDa protein (Fig.  4).
Southwestern blots have been performed to further confirm the specific DNA binding activity of the 43-kDa protein in the DNA affinity-purified material. As shown in Fig. 5A, multiple bands from rat brain extract were visible on the Southwestern blot, while the b2-containing fraction of the DEAE-column and the affinity-purified material showed only one DNA-protein interaction band corresponding to a protein of 43 kDa. This experiment, along with gel retardation assay (Fig. 2 A ) , indicates that the DEAE-chromatography gave a good separation of b2 activity from other DNA binding activities present in the rat brain extracts, and that the 43 kDa is the only protein in the affinity-purified material which shows DNA binding activity. When a mutated DNA fragment shown below was used in the Southwestern blot, no DNAprotein interaction was visualized (Fig. 5B). In contrast, the wild type probe gave the expected DNAprotein interaction, suggesting a specific binding of the 43-kDa protein to the somatostatin promoter fragment. Phosphorylation of the 43-kDa Protein-The 43-kDa protein binds specifically to the CAMP-responsive sequence of the rat somatostatin promoter. Montminy et al. (21) have shown that CAMP induction of the somatostatin promoter in the rat pheochromocytoma cell line, PC-12, is dependent upon this sequence element. Functional CAMP-responsive sequences have been found in the genes for human proenkephalin ( l l ) , cu-subunit of chorionic gonadotropin (13), and vasoactive intestinal polypeptide (22). Since it has been suggested that the catalytic subunit of protein kinase is required for CAMP-inducible transcription (23), it was of interest to see whether the 43-kDa protein could be phosphorylated by this kinase. Using the purified catalytic subunit of CAMP-dependent kinase, the 43-kDa protein was shown to be a substrate for phosphorylation, as indicated in Fig. 6A. In addition, the 43-kDa protein was not autophosphorylated in the absence of CAMP-dependent protein kinase (data not shown). The phosphorylated amino acid residue(s) in the protein were identified by phosphoamino acid analysis (18). Phosphorylated proteins were excised from the SDS gel, hydrolyzed with HC1, and phosphoamino acids resolved by thin layer chromatography. Phosphorylated residue(s) of the protein were identified by comparison to standard phosphoamino acids. Phosphoryla- The intensities of these bands were scanned to estimate the increase in specific activity throughout the purification. Lane 1, (NH&SO~-fractionated extract; lane 2, b2-containing fraction from the DEAE-column; lane 3, the affinity-purified material. The asterisks indicate three other DNA binding activities present in the rat brain extract. B, the affinity-purified material on the filter was hybridized with wild type ( l a n e 1) or mutated ( l a n e 2) somatostatin promoter sequence.
tion was shown to occur on serine residue(s) (Fig. 6B). Phosphorylation of threonine or tyrosine residues was not observed. The number of phosphorylated serine residues has not been determined, and the biological function of the phosphorylation remains unknown.
Transcriptional Activity of the 43-kDa Protein in Vitm-Although the 43-kDa protein had been purified to apparent homogeneity and shown to bind specifically to the CAMP- Lane 1 (from Ieft to right), HeLa cell extract; lanes 2 and 3, HeLa cell extract depleted of somatostatin promoter transcription activity by affinity procedure. Addition of the 43-kDa protein to depleted cell extract ( l a n e 3) restored the somatostatin promoter transcription. MLP is the adenovirus major later promoter used as an internal control.
structural gene. This MLP-chloramphenicol acetyltransferase construct served as an internal control in the in vitro transcription system. Transcripts from the somatostatin promoter and MLP were detected by a primer extension assay using an oligonucleotide which is complementary to the chloramphenicol acetyltransferase structural gene. The anticipated length of the primer-extended DNA is 137 nucleotides for the somatostatin gene and 110 nucleotides for the MLP. Fig. 7 demonstrates that the extract correctly initiated transcription from the somatostatin promoter and MLP.
To examine the transcriptional activity of the purified 43-kDa protein, endogenous somatostatin promoter binding protein(s) were depleted from the HeLa cell extract by passing the extract through a biotinylated oligonucleotide-avidin column. With the depleted extract, it is evident that transcription from the MLP was unaffected, while transcription from the somatostatin promoter was greatly reduced. This observation suggests that the factor(s) which had been depleted from the transcription extract were essential for somatostatin gene transcription. Addition of the purified 43-kDa protein to the depleted extract restored the transcriptional activity of the somatostatin promoter (Fig. 7). These data clearly indicate that the 43-kDa protein purified from rat brain extract is a component required for the in vitro transcription of the somatostatin gene.

DISCUSSION
An increasing number of eukaryotic DNA binding proteins have been purified using the technique of DNA recognition site affinity chromatography (5,24). We describe in this paper the purification from rat brains of a DNA binding protein specifically recognizing the -59 to -35 region of the rat somatostatin promoter. The experimental data presented here show that the 43-kDa protein-purified by DNA affinity chromatography is a factor required for somatostatin gene tran-scription. DNase I footprinting assays demonstrated that the affinity-purified protein bound specifically to the -59 to -35 region of the somatostatin promoter. Southwestern blot, and renaturation experiments confirmed that the DNA binding activity was derived from the 43-kDa protein. In addition, purified 43-kDa protein restored transcription initiating from the somatostatin promoter in an in vitro transcription/complementation assay.
The buffers used for the affinity column were found to be important for the successful purification. Gel retardation assays were used to optimize the conditions for protein-DNA interaction prior to performing the DNA affinity column. The presence of 10 mM EDTA effectively inhibited the endogenous nuclease activity present in the DEAE-fractionated samples. Lower concentration of EDTA in the buffer reduced the capacity of the DNA affinity column (data not shown). In addition, the inclusion of 0.2 M KC1 in buffer B effectively reduced the nonspecific binding of proteins to the column, while specific protein-DNA interactions were retained.
Previous work demonstrated the same gel retardation pattern with extracts prepared from somatostatin-producing CA-77 cells, as well as from HeLa cells which do not express somatostatin (IO), suggesting that the same DNA binding factor(s) exist in both cell lines. Thus, a possible explanation for selective expression of the somatostatin gene in CA-77 cells would involve tissue-specific post-translational modification(s) of pre-existing transcription factor(s). Since the 43-kDa protein has been shown to be phosphorylated in vitro by the catalytic subunit of CAMP-dependent protein kinase, it will be interesting to study the effect of phosphorylation of this protein on the DNA binding and in vitro transcriptional activities of the somatostatin promoter. Preliminary data indicate that incubation of the phosphorylated 43-kDa protein with CA-77 cell extract results in a selective dephosphorylation of the 43-kDa protein?
Montminy's group (25,26) has purified a nuclear protein, CREB, which selectively binds to the -55 to -32 region of the somatostatin gene. Their experimental data indicated that this factor has a molecular mass of 43,000 daltons on SDS-PAGE. In addition, a cellular transcription factor, ATF, has been reported to bind to both E1A-and CAMP-inducible promoters, including the -55 to -35 region of the somatostatin gene (27). UV cross-linking experiments demonstrate that ATF has a molecular mass of 45,000 daltons. These data suggest that CREB and ATF are probably related to the 43-kDa protein we have purified from rat brain. Further biochemical evidence is required to confirm their identity.
Protein-protein interactions have been proposed to play a central role in the control of prokaryotic gene transcription (28). This type of interaction may also represent an important mechanism for gene expression in more complex eukaryotic systems. Recent studies have indicated that multiple cisacting elements and multiple trans-acting factors are involved in the expression of many RNA polymerase 11-transcribed genes. One advantage of such a combinatorial mechanism is that a high degree of diversity in gene regulation can be achieved with a limited number of trans-acting elements. The 43-kDa protein purified from rat brains corresponds to b2, one of three DNA-protein complexes identified with gel retardation assays (10) plexes are observed in gel retardation assays. The in uitro transcription experiment demonstrates that the purified 43-kDa protein is necessary for the transcription of the somatostatin gene. However, it is not clear whether other factors are also involved in this transcription machinery. Further characterization of this 43-kDa protein should enhance our understanding of the nature of the b l and b3 DNA-protein complexes. A detailed molecular understanding of the 43-kDa protein and somatostatin gene transcription must await the isolation of the gene for the trans-acting factor.