Purification of PO-B, a Protein That Has Increased Affinity for the Pro-opiomelanocortin Gene Promoter after Dephosphorylation*

The region -15 to -3 of the pro-opiomelanocortin (POMC) gene promoter specifically binds a transcription factor previously designated PO-B. This region of the POMC gene is involved in the control of constitutive POMC gene expression since mutation of the PO-B DNA-binding site severely reduces transcription from the POMC promoter both in vivo and in vitro (Riegel, A. T., Remenick, J., Wolford, R., Berard, D., and Hager, G . (1990) Nucleic Acids Res. 18, 4513- 4521). We have now purified PO-B from HeLa cells approximately 25,000-fold to greater than 90% ho- mogeneity by a combination of ion exchange and reversed phase chromatography. In addition we have studied post-translational modifications that alter the affinity of purified PO-B for its cognate DNA binding site. In Southwestern analysis of column fractions, two bands of apparent molecular masses of 54 and 56 kDa bound specifically to the PO-B recognition sequence. The two copurified components have indistinguishable amino acid composition, are highly hydrophobic, and are heat and acid stable. DNA-binding specificity studies suggest that PO-B does not represent any previ- ously described

The region -15 to -3 of the pro-opiomelanocortin (POMC) gene promoter specifically binds a transcription factor previously designated PO-B. This region of the POMC gene is involved in the control of constitutive POMC gene expression since mutation of the PO-B DNA-binding site severely reduces transcription from the POMC promoter both in vivo and in vitro (Riegel, A. T., Remenick, J., Wolford, R., Berard, D., and Hager, G . (1990) Nucleic Acids Res. 18, 4513-4521). We have now purified PO-B from HeLa cells approximately 25,000-fold to greater than 90% homogeneity by a combination of ion exchange and reversed phase chromatography. In addition we have studied post-translational modifications that alter the affinity of purified PO-B for its cognate DNA binding site. In Southwestern analysis of column fractions, two bands of apparent molecular masses of 54 and 56 kDa bound specifically to the PO-B recognition sequence. The two copurified components have indistinguishable amino acid composition, are highly hydrophobic, and are heat and acid stable. DNA-binding specificity studies suggest that PO-B does not represent any previously described transcription factor. In addition, dephosphorylation of both species with acid phosphatase induced an about 30-fold increase in DNA binding but failed to produce any significant change in electrophoretic mobility. We conclude that the purified PO-B species represent products of the same gene and suggest that the in vivo function of PO-B may be regulated by its phosphorylation status.
The pro-opiomelanocortin (POMC)' gene is predominantly expressed in the anterior pituitary. The primary gene product is a large precursor protein which is processed in the pituitary into smaller polypeptide species such as ACTH and @-endorphin (1). In pituitary tissue and in pituitary tumor cell lines , POMC mRNA is easily detectable (Z), reflecting a high level of constitutive, tissue specific transcription. Hormonal repression or induction of POMC transcription can be produced by administration of glucocorticoids or corticotropin-releasing factor, respectively (3). Approximately 780 base pairs of the POMC promoter upstream of the transcription initiation site appear to be required for high basal transcription in uiuo* (4) and in uitro.' Furthermore this area also appears to encompass all the sequence information required for corticotrophin-releasing factor induction (5) and glucocorticoid repression of POMC transcription* (6). Despite detailed knowledge of the endocrine control of POMC gene expression, the transcription factors involved in high basal, tissue specific and hormonally regulated control of POMC transcription have not been clearly defined.
We have previously described at least three areas within the first 480 base pairs of the POMC promoter which contribute to high basal transcription of the POMC gene. These are situated between -480 to -320, -70 to -50 and -15 to -3 relative to the POMC cap site ' (7). The more distal sites upstream of -50 appear to be involved in glucocorticoid repression of POMC gene expression.2 In contrast, mutation of the -15 to -3 region does not abolish the glucocorticoid repressive effect (7), but reduces constitutive transcription severely in vivo and in uitro. In nuclear extracts from a number of cell lines a specific binding factor for the -15 to -3 recognition sequence can be detected and has been designated PO-B (7). This factor is particularly intriguing since it appears to exert its effect a t a site positioned between the POMC TATA box and cap site.
However, the PO-B recognition sequence does not appear to have any of the characteristics of previously described initiator elements (8) since it is unable to direct transcription initiation in the absence of a functional TATA box (7). Furthermore, PO-B is not analogous to "general" transcription factors involved in initiation complex formation since these proteins do not exhibit sequence-specific DNA binding (9,10). Indeed this region of eukaryotic promoters is infrequently associated with sequence-specific DNA binding of transcriptional activators. Whether PO-B is an ubiquitous cellular factor or is involved in one of the many hormonal (11) and immune (12,13) signal transduction pathways that regulate POMC gene expression in the pituitary is not known. However, PO-B does represent the first characterized transcriptional activator of POMC whose function has been verified by mutational analysis.
As a first step in elucidating the mechanism of action of PO-B and its cellular control mechanisms, we have prepared a highly purified form of PO-B by a combination of ionexchange and reversed phase chromatography. In this study we demonstrate that two protein species of approximately 54 and 56 kDa bind to the PO-B cognate binding site. These two proteins are highly related by a number of criteria such as amino acid composition, DNA binding specificity, hydrophobicity, acid and heat stability. Intriguingly, binding of PO-B to its cognate sequence is regulated by the phosphorylation status of the protein: dephosphorylation leads to a >30-fold increase in specific binding, suggesting that the signal transduction pathway that regulates PO-B may act through phosphataselkinase interactions.

MATERIALS AND METHODS
Preparation of Nuclear Extracts-Nuclear extracts were prepared from the AtT-20 mouse pituitary tumor, C127 mouse mammary adenocarcinoma, and HeLa cell lines, maintained in Dulbecco's modified Eagle's medium + 10% fetal calf serum, by the method of Dignam et al. (14). Large scale preparation of nuclear extracts was from HeLa cells in suspension culture maintained in minimum Eagle's medium spinner culture media + 5% horse serum. Briefly the cells were washed and lysed in Hepes (10 mM pH 7.9) a t 4 "C, containing 1.5 mM MgCl,, 10 mM KC1,0.5 mM DTT, and nuclei were prepared. Nuclear proteins were extracted in high salt buffer (20 mM Hepes, pH 7.9,25% glycerol, 0.42 M NaC1, 1.5 mM MgC12, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 mM DTT). The final dialysis of the high salt nuclear extract was performed in buffer D (20 mM Hepes, pH 7.9, 20% v/v glycerol, 100 mM KC1, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 0.5 mM DTT). The dialysate was cleared of precipitate by centrifugation at 25,000 X g for 20 min and the supernatant was frozen at -80 "C. The protein concentration of the extract, assayed by the Bradford procedure (15), was in the range of 1-20 mg/ml. Electrophoretic Mobility Shift Assay-Protein-DNA complexes were resolved on nondenaturing 4% low ionic strength polyacrylamide gels as described previously (16). In brief, nuclear extracts (1-20 pgj were incubated with 1-10 fmol of PO-B consensus oligonucleotide end-labeled with polynucleotide kinase (oligonucleotide sequence 5'-GGTATAAAAGAAGAGAGAAGAGTGACAGGGACCAAACTCGA GA- 3') in the presence of 1 pg of poly(d1-dC) .poly(dI-dC) nonspecific competitor in a final volume of 20 pl for 20 min a t 25 "C. Reactions contained 10 mM Tris, pH 7.4, 100 mM KCI, 5 mM MgCl,, 5% glycerol, and 1 mM DTT. Specific binding of the radiolabeled oligonucleotide was determined by competition with either unlabeled PO-B oligonucleotide competitor or by a GA-rich oligonucleotide designated FS (17) (5'-GAAAGGAGAAACGAAAGGAGAAC-3') included in the initial reaction (up to 200-fold molar excess). Binding reactions were subjected to gel electrophoresis for 3-5 h a t 10 V/cm in 12.5 mM Tris borate/EDTA running buffer. Following electrophoresis, gels were dried onto Whatman No. 3MM paper and subjected to autoradiography.
Southwestern Blot Analysis of Protein-DNA Interactions-The method for Southwestern analysis has been described previously (18). Crude or pure nuclear proteins were resolved by SDS-PAGE using either linear (10%) or gradient (10-20%) gels (10 cm X 10 cm X 1 mm) using standard procedures (19). Prior to electrophoresis acetonitrile was removed from the samples by brief vacuum drying and disulfide bonds reduced in the presence of P-mercaptoethanol accompanied by heating a t 100 "C. After electrophoresis proteins were transferred to nitrocellulose for 1 h a t 150 mA in 6.25 mM Tris, 1.3% glycine, and 20% methanol. After transfer, the nonspecific sites on the membrane were blocked by three washes of 45 min in 10 mM Tris, pH 7.5, 5% nonfat dry skim milk, 10% glycerol, 2.5% Nonidet P-40,O.l mM DTT, 150 mM NaCl at 25 "C. The membrane was then rinsed briefly in binding buffer (10 mM Tris, pH 7.5, 40 mM NaC1, 1 mM EDTA, 1 mM DTT, 8% glycerol, 0.125% nonfat dry skim milk) and was incubated in 3 ml of binding buffer containing 5 mM MgC12, 500,000 cpm/ml end-labeled specific oligonucleotide probe, and 10 pg/ml poly(d1-dC) .poly(dI-dC). The presence of specific binding was determined by the addition of a 100-200-fold molar excess of unlabeled specific PO-B oligonucleotide to a separate hybridization bag. To further analyze the specificity of PO-B binding Southwestern blots of crude extract and purified PO-B were also probed with the FS oligonucleotide (17j, the PO-Bm oligonucleotide (5'-AAGAAGTTCCTGTCCCACTGGGGACCAA-3') which does not bind PO-B (7) and an oligonucleotide harboring the SP1 consensus sequence (5'GATCGGGGCGGGGCTGGGGCGGGGC- 3'). Incuba-tions were at 25 "C for 16 h with gentle agitation, followed by 3 X 15min washes in 10 mM Tris, pH 7.5, 50 mM NaC1. The blot was then exposed for autoradiography.
Ion Exchange Chromatography-A Mono-S ion-exchange column (0.5 X 5 cm; Pharmacia LKB Biotechnology Inc.) was equilibrated with ice-cold buffer D (final dialysis buffer of nuclear extraction procedure; see above) at a flow rate of 0.5 ml/min. Nuclear extracts in buffer D were cleared from precipitated protein by a brief centrifugation, and about 5 mg of total protein were loaded in 2 ml onto the Mono-S column. The column was washed with ice-cold buffer D until the 280-nm absorbance reading returned to base line, and bound proteins were eluted with a 20-min linear gradient of 100 mM KC1 to 1 M KC1 in buffer D. Fractions of 0.3 ml were collected and stored frozen a t -70 "C. 5-10-p1 aliquots of each fraction were analyzed for the presence of PO-B by Southwestern blotting analysis.
Reversed Phase HPLC-Two subsequent reversed phase HPLC runs were carried out. The C-3 reversed phase HPLC columns (Ultrapore RPSC, Beckman) were equilibrated with H20 containingO.l% trifluoroacetic acid a t a flow rate of 1 ml/min. In the first HPLC run, fractions from the ion-exchange chromatography containing PO-B were pooled and loaded directly onto a semi-prep column (10 X 250 mm) using multiple injections of 2 ml of trifluoracetic acid pretreated sample. After a 5-min wash of the column, bound proteins were eluted with a linear (0-60%) gradient of acetonitrile + 0.1% trifluoroacetic acid over 60 min. Fractions of 1 ml were collected and 10-pl aliquots of each were analyzed for the presence of PO-B by Southwestern blotting.
In the second reversed phase HPLC run, fractions from the first run containing PO-B were frozen a t -70 "C and subjected to a 20min vacuum extraction of residual acetonitrile. After thawing, the active fractions were pooled and loaded onto an analytical C-3 column (5 x 75 mm) a t a flow rate of 1 ml/min. Bound proteins were eluted with a flat gradient (35-65%) of acetonitrile + 0.1% trifluoroacetic acid over 60 min. Fractions of 1 ml were collected and analyzed for the presence of PO-B using Southwestern blot analysis. Proteins were visualized on trichloracetic acid-pretreated SDS-PAGE gels or on nitrocellulose membranes by staining with silver or silver/gold, respectively, using commercially available assays (enprotech-ISS, Hyde Park, MA and Amersham Corp.). Protein concentrations in crude and fractionated extracts were estimated using the Bradford procedure with BSA as a standard.
Renaturation of Purified PO-B-BSA (molecular biology grade 1 pg/pl) was added to pooled samples from fractions after the second reverse phase purification step containing PO-B activity. The samples (100 plj were vacuum-dried briefly to remove acetonitrile resuspended in buffer and dialyzed against 120 volumes of 10 mM Tris, pH 7.5, 100 mM KC1, 1 mM DTT, 1 mM EDTA, 5% glycerol for 24 h at 4 "C. Redialyzed samples were stored a t -80 "C prior to use. Amino Acid Analysis-The 54-and 56-kDa PO-B species were resolved on 10-20% SDS-polyacrylamide gradient gels and transferred onto a PVDF membrane (Millipore) using 10 mM CAPS buffer, pH 11, with 10% methanol. Protein bands were detected on the membrane with a brief Coomassie staining. The bands were excised from the membrane, acid hydrolyzed in 6 M HC1 for 24 h at 110 "C, and analyzed on a Beckman 6300 amino acid analyzer (20) by Dr. K. Williams, Yale University.
Phosphatase Treatment of Puri/ieied PO-B-Purified, redialyzed PO-B protein was incubated with BSA (0.1 pg/plj in 50 mM PIPES (pH 6.0) 0.2 M NaCl, 5 mM DTT, 1.5 mM MgCl,, 10% glycerol, v/v in the presence of 0.5 unit of potato acid phosphatase/ml (Sigma; type 111) in a total reaction volume of 25 pl. The reaction was allowed to proceed for 16 h a t 25 "C. An aliquot of the incubation solution, equivalent to 10 p1 of the column fraction, was denatured, separated by 10-20s SDS-PAGE followed by Southwestern blot analysis (see above). Control samples contained either potato acid phosphatase alone or preboiled potato acid phosphatase.

RESULTS
Purification of PO-B Assay of PO-B Activity-We have previously demonstrated that the region between -15 to -3 of the POMC promoter specifically binds a transcription factor designated PO-B (7). Mutation of the PO-B-binding site reduced POMC basal transcription by approximately 70% using transient transfection assays of the AtT-20 pituitary tumor cell line. This effect was also observed in in vitro transcription assays with whole cell extracts from HeLa and AtT-20 cells (7). Now we describe the purification and characterization of t h e PO-R prot,ein. As described previously, specific PO-R binding to its cognate site can be detected by electrophoretic mobility shift assay ( 7 ) ( Fig. 1 A ) . T h e PO-R binding is competed effectively by the PO-R oligonucleotkle but not by a GA-rich oligonucleotide (FS) that has some sequence homology to PO-H (Fig. In).
Although the electrophoretic mobi1it.y shift assay is a n effective method of detecting PO-R, one drawback is t h a t D N A binding is highly sensitive to minor pH and salt changes. Thus, fractionation procedures t,hat involved changes in these parameters would require buffer adjustment of each fraction prior to the DNA-binding assay. To circumvent this problem we decided to develop the Southwestern detection assay (18) for PO-R. Using t.his technique two protein species of a n apparent molecular mass of 56 and 54 kDa bound to a radiolabeled PO-R probe in AtT-20, C127, and HeLa nuclear extracts (Fig. 1H). The small differences between the mobility of the bands for each cell type (r.g. Innrs 2 vrrsus .?) were due to salt differences in the extracts and were not observed consistent.ly. Competition by a 200-fold molar excess of nonlabeled PO-R probe significantly reduced t.he signal observed in the Southwestern analysis (Fig.   1H, lanc 4 ) . To further validate Sout,hwestern analysis as a specific detection method, a n oligonucleotide harboring a PO-R mutation (PO-Rm) which does not bind 1'0-€3 in vitro and suppresses POMC gene expression both in vivo and in vitro ( 7 ) was used. This mutat,ed PO-R probe did not det,ect. any proteins similar to the wild-type probe (Fig. lC, lnnr 1 ). Furthermore an Sl'lbinding site oligonucleotide detected only a predominant species at approximately 100 kDa ( Fig. I(', lonc 2 ) which is the expected molecular mass range of SPI-binding species (21 ). We have repeated this analysis with a number of oligonucleotide-binding sites for characterized transcript ion factors (AP1 (22), AP2 (23). IRE (24). and ISKE (17, 2.5). None o f these oligonucleotides was able to compete for PO-H binding (analyzed by electrophoretic mobility shift or Southwestern analysis) nor bind to the 54-or 56-kDA species (data not shown). Although we have not defined the limits of sensitivity of the Southwestern analysis, it is possible to detect 1'0-H in 1 pl of crude HeLa nuclear extract (1-20 pg of protein) and <I0 fmol of purified protein are sufficient to give a signal in this assay. Thus it appears that Southwestern analysis of PO-R is sufficiently sensitive and specific to be wed as a detection method for PO-R.
Ion Exchangr Chromatography-As an initial step we decided to use a cation exchange column because PO-H had previously been fractionated ef'l'ectivelv on an analytical gravit.y flow phosphocellulose column ('7). We decided to use H e h cells in suspension culture as a source of PO-H because they grow rapidly and are less demanding in terms of serum and media requirements than AtT-20 cells. The 54-and 56-kDA PO-H species are detected in both these cell lines (Fig.   I N ) . Purification of Transcriptiol Dignam and Roeder procedure (14) was applied directly to a Mono-S cation exchange column and proteins were eluted with a linear gradient of 0.1-1.0 M KCI. The active fractions eluted at approximately 0.25-0.3 M KC1 as indicated in Fig.  2 A . We determined that PO-R present in these fractions remained active a t room temperature for at least 24 h. Therefore all subsequent purification steps were performed at 22 "C.
Reoersed Phase Chromatography-The PO-B-containing fractions retrieved from the Mono-S column were pooled and were loaded direct,ly onto a semipreparative C-3 column. Proteins hound to the column were eluted with a linear 0-60% acetonitrile gradient (Fig. 2R). Protein concentration was monitored by absorbance a t 280 nm (Fig. 2R). The 54and 56-kDa PO-R protein species were eluted a t approximately 50% acetonitrile in four fractions as assayed by Southwestern analysis (Fig. 2R, lower panel). We subjected the pooled fractions that contained PO-B to a second analytical scale reverse phase chromatography. The samples were loaded after residual acetonitrile had been removed by vacuum extraction. In this second reversed phase run prot,eins were eluted by a shallow linear gradient of 35-65S;I acetonitrile ( Fig. 2C). Using this met.hod the bulk of the protein eluted a t the beginning of the run, whereas PO-B protein once again eluted in four fractions at approximately 50-55% acetonitrile (Fig. 2C, lower panel). The protein concentration in these fractions was below the detection limit of the 280 nm UV absorbance (Fig. 2C).
Esfimate of the -Fold Purification-To assess the purity of Tal Activator of POMC Gene 12237 PO-R after the second reverse phase we compared the protein profile of aliquots of peak fractions from each stage of purification. The amounts of protein loaded on the SIX-PAGE gel were determined from pilot experiments to give approximately equal PO-R signals after Southwestern analysis. The protein in the gel was assayed using silver stain analysis (Fig.  3). The most purified fraction after the second reversed phase step contains predominantly two silver-stained proteins with apparent molecular masses of 56 and 54 kDa. Approximatelv 2 ng of protein was loaded into this lane. Therefore if we assume that the lower detection limit of protein in these gels is 50 pg/hand (26) we can conclude that this preparation of PO-R is greater than 90% homogeneous. To estimate the -fold purification obtained, we titrated with a crude and a purified preparation, how much of either was required to give a signal of equal strength in the Southwestern analysis. In these experimenh, 20 pg of total protein from crude extract (Fig. 4, lane I ) and 600 pg of purified protein of either PO-R species (Fig. 4, lune 2 ) gave a signal of equal strength. These protein concentrations were assessed by the Bradford assay of crude extract and bv total amino acid analysis of pure protein (see "Materials and Methods"). We thus estimate that we achieved about a 25,000-fold purificntion. In a t-ypical purification run the start-up material would contain 100 mg of total nuclear protein which we estimate contains about 3 pg of PO-R (from Southwestern annlvsis above). Our final preparation of PO-R from this run contained active frartions from the Mono-S column were pooled and applied directly onto n semipreparative column ( Y ) O x 100 mm), and hound proteins were eluted from the column with a linear gradient of ( 1 -W ' ; acetonitrile as (iescritwd under "Materials antl Methods." Eluted proteins were monitored hv ahsorhance at 2x0 nm. antl 1 0 p1 o f the I-ml frartions were assayed for 1'0-R activity 1)v Southwestern analysis (see Fig. 1H). T h e hoxrd nrcn in the upprrpnnd represents the peak fractions shown in the Southwestern analysis in the hwrr pnnrl. (', serond reversed phase H1 '1,C: active fractions from the first reversed phase were pooled. vacuum-dried. and applied to a second revcmrd phase analytical column (details under "Materials and Methods"). Hound proteins were eluted with a shallow linear grdient of :15-65"h acetonitrile as descrihed under "Material and Methods." Eluted proteins were detected a t 2x0 nm but were I)elow sensitivity in most of the fractions. Each fraction ( 1 ml) was analyzed lor 1'0-13 n.; desrrihed al)ove. Southwestern analysis of fractions 48-58 (hmxrd arm, upprr pnnrd) is indicated in lorwr p n r l (Innrs 1-10), analysis. Therefore our recovery of PO-R using this protocol is approximately 3%. In four separate purification runs we recovered hetween 1 and 5% of PO-H present in the initial preparation after the final Southwestern blot of purified material. The major loss of PO-H occurs during the ion exchange chromatography step. A rough calculat.ion of the abundance of PO-R can be made based on the amount of protein recovered and the numher of cells used initially to prepare the nuclear extract. From n pproximately 10"' cells we recovered 5 p g of purified 1'0-13. Assuming an average molecular mass of 55 kI>A we estimate a cellular abundance of approximately 6001) molecules o f I'O-€3 per cell.

-H I)NA Hindin,c
To determine if the specificity of purified 1'0-H was similar to that of the crude extracts we assayed the ability o f a series of transcription factor hinding sites (Sl'l. A P I , A P 2 ) t o bind t o t h e purified protein. In a t?lpical experiment the FS consensus sequence was unable t o compete for 1'0-13 binding (Fig. I n ) hut hinds to a numher of species in the 40-50-kI);t range in crude extracts (Fig. 4, lnnc 3 ) . However no FS binding species was detected in the PO-H purified fractions (Fig. 4.   lnnc 4 ) . In fact only the PO-H oligonucleotide was ahle t o detect the 54-and 56-kI)a species in crude and purified fractions (Fig. 4, lone 1 L'crsus lane 2 ) .
Finally we tested if the two protein species ohserved on the silver-stained gel corresponded to the two hands observed after binding in the Southwestern procedure. Three lanes o f the purified sample were run on a 10-20'; SDS-polyacrylamide gradient gel and hlotted nnto nitrocellulose. The h l o t was cut in half dividing the middle lane longitr~dinally and protein was detected on one half by silver-golt! staining (Fig.  5, lanrs 1 and 2 ) and on the other by the Sout hwestern procedure (Fig. 5, lnncs 2 and 3 ) . The bands detected by each procedure were coincident indicating that the protein specics visualized by silver staining are most likely t o be the ones binding to the PO-R recognition sequence.
Also consistent with this conclusion is the observation that a single I)NA binding species is detected after each hand has heen cut o u t of a Coomassie-stained gel, and protein electroeluted and analyzed by Southwestern hlotting (data not shown).
Experiments using electrophoretic mohility shift assnqs with purified PO-R indicated that DNA-binding activity o f PO-R was undetectable after the protein had been in contact with acetonitrile. However. this process was not irrwersihle since Southwestern analysis of the reversed phase fractions det.ected specific PO-R DNA binding, Therefore we attempted to renature purified PO-H by overnight dialysis. After this renaturation step the electrophoretic mohility of the purified PO-R/DNA complex was indistinguishable from that oht.ained with crude extracts (Fig. 6). 'I'hese data taken together with the results presented in Fig. 5 strongly suggest that the 56-and 54-kDa proteins that we have purified are responsible for specific binding to the PO-I3 recognition sequence.

Role of Phosphorylation in PO-B D N A Binding
We were particularly interested to determine if the two PO-B forms observed were due to post-translational modifications. I t has previously been demonstrated that glvcosylation (27) and phosphorylat,ion (28-33) of transcription factors can be important modifications for the activity of these proteins. Furthermore, these post-translational modifications can produce considerable differences in the electrophoretic mobility of proteins (29). Initial experiments using digestion of purified PO-B with Nand 0-endoglycosidases did not alter the mobility of either species (data not shown). To test for differences in phosphorylation status, purified protein was digested with potato acid phosphatase and probed by Southwestern analysis. No significant change in the electrophoretic mobility of the proteins was observed after dephosphorylation. However, phosphatase treatment increased the PO-R DNA-binding act.ivity hy over 30-fold as analyzed by laser scanning densitometry of the two species (Fig. 7). The increase in DNA binding was observed for both the 54-and 56-kDa species (Fig.  7). T h e effect was not due to a prot,ein species fortuitously present in the potato acid phosphatase preparation since this enzyme gives no signal on a Southwestern when run in the absence of purified PO-R. Roiled enzyme did not produce an increase in DNA binding (data not shown). We have also obtained an increase in DNA binding affinity using acid phosphatases from different sources (data not shown).

Relatedness of the 54-and fj6-kDa Species
T h e 54-and 56-kDa PO-R species copurify throughout all biochemical fractionation procedures and the binding specificity of both forms is identical. Furthermore, dephosphorylation of both species increases DNA binding. These data taken together indicate that these species probably represent modified forms of the same protein. To investigate this further we performed total amino acid analysis on the 54-and 56-kDa species (Table I). For each amino acid or group of amino acids analyzed, the percent composition for each species is almost identical within the precision of the HPLC analysis (20). Although these data do not prove that these proteins are the same it does not preclude this conclusion and is consistent with the other physicochemical data that indicates that these proteins are highly related.

DlSCllSSlON
In the present study we have used Southwestern blot anal?sis to follow the purification of PO-H, a trnnscription factor that binds to the -15 to -3 region of the POMC promoter. We have previously demonst.rated that mutation of this site causes a 70% decrease in POMC basal transcription both in uiuo and in in vitro transcription extracts from a number o f cell lines (7). PO-R activitv was particularlv pronounced in HeLa cell extracts (7) and hence our decision to purifv and characterize PO-B further from this cell line. We had previously demonstrated that the retarded complex formed with the cognate PO-B DNA binding site was similar with nuclear extracts from the AtT-20 pituitary tumor, C127 mouse mammary adenocarcinoma and HeLa cell lines (7). This is consistent with the observation in this study of similar 54-and 56-kDa species detected by Southwestern analysis in these different cell types. Mutation of the PO-B binding site reduces POMC transcription by about 3-fold in HeLa, ATt-20, and C127 whole cell extracts (7). Furthermore, purified PO-B is capable of producing the same electrophoretic mobility shift complex as that observed with crude nuclear extracts. Taken together these data imply that the 56-and 54-kDa proteins are components of the specific retarded DNA-protein complex obtained with the PO-B DNA binding site.
A separate issue is whether the 54-and 56-kDa PO-Bbinding species represent modified forms of the same protein or different gene products. The current data strongly suggest that the former view is correct. First, the 56-and 54-kDa species copurify in all fractionation procedures that we have used. Second, the amino acid analysis of the two species is highly similar. Third, the binding specificity of both species is identical (discussed below). Finally the dephosphorylation of the purified fractions leads to a concomitant increase in DNA binding of the 54-and 56-kDa species but no change in the electrophoretic mobility of either species. Our preliminary data also imply that the differences are not due to glycosylation. We are currently determining if other post-translational modifications are involved in producing the two PO-B species. Conversely they may be products of alternate splicing as has been shown to occur for the CREB and ACREB transcription factors (33). Interestingly under nondenaturing conditions only a single species of approximately 110 kDa is observed with Southwestern analysis." This observation might suggest that the two forms of PO-B copurify as a dimer and are separated during denaturing electrophoresis. Homo-and heterodimer formation has been described for a number of cellular transcription factors (34, 35).
The rare abundance of PO-B is consistent with that of other described cellular transcription factors (36, 37). Furthermore, PO-B fractionates in high (50-60%) acetonitrile implying that this is a highly hydrophobic protein, also a characteristic of a number of other transcription factors (37). In fact, a number of these proteins have molecular masses in the range of 45-60 kDa (37) leading us to question whether PO-B was actually one of these previously characterized proteins. In addition, homology searches of other DNA consensus sequences had revealed some similarities to the PO-B DNA binding site. However, in extensive cross competition and Southwestern analysis with the AP1 (22), AP2 (23), SP1 (37) and a variety of interferon and viral response elements (17,24,25), no competition for PO-B DNA binding site in either crude or purified fractions was observed. This was true for both the 56-and 54-kDa species. This suggests that PO-B is a novel protein that has not been previously characterized. In fact we have now obtained an N-terminal amino acid sequence from the 54-kDa species. Homology searches with available sequences in GenBankTM indicate no significant similarity with any protein in this database.
The role of dephosphorylation in PO-B mechanism of action is intriguing especially since phosphorylation of other transcription factors is required for high DNA affinity and transcriptional activation (28-32). Some of these phosphorylation events are accompanied by significant changes in mobility on SDS-PAGE (29). Phosphorylation of other factors, as is the case with PO-B, causes no appreciable change (38). Some recent studies demonstrate that progressive phosphorylation in vitro is accompanied by progressively larger differences in SDS-PAGE mobility (29). This might imply that PO-B is undergoing dephosphorylation at relatively few amino acids. Clear conclusions on this point await further i n vitro studies. However, our data do imply that there must be significant overlap in the purification profiles of the phosphorylated and dephosphorylated forms of PO-B. In fact the phosphorylated and dephosphorylated forms of the 54-and 56-kDa species must have the same mobility on SDS-polyacrylamide gels since no other major protein species is observed on the silver stains of the purified product. Whether the variable levels of the phospho and dephospho versions of PO-B alter its transcriptional activating ability is currently under investigation. Currently we do not know the signal transduction pathway that regulates PO-B i n vivo. Therefore we can only speculate that PO-B may be activated by a specific cellular phosphatase or conversely deactivated by a specific cellular kinase. The number of cellular phosphatases characterized is small (39) especially compared with the extensive knowledge on cellular kinases. Recently a number of specific tyrosine phosphatases have been characterized and cloned (40, 41). However, we failed to detect PO-B using anti-phosphotyrosine antibodies in Western blot analysis, indicating that PO-B may be a substrate for serine/threonine kinase a~t i v i t y .~ T h e availability of more specific enzymes should make experiments on the role of dephosphorylation in PO-B mechanism of action more feasible. It has been postulated that for each specific cellular kinase there exists an opposing specific phosphatase (42); thus one might predict that a specific cellular kinase may reduce PO-B DNA binding. We are currently determining if any of the known cellular kinases are able to alter PO-B DNA-binding affinity in vivo and i n vitro.