CAMP-dependent Transcription of the Human CYP2lB (P-450cz1) Gene Requires a Cis-regulatory Element Distinct from the Consensus CAMP-regulatory Element*

By utilizing chimeric genes constructed from 5’- flanking sequences of the human CYP21B (P-450& gene and reporter genes (chloramphenicol acetyltransferase or rabbit /3-globin), a 34-nucleotide sequence has been found to be required for CAMP-dependent transcription. This sequence (-1291-96 base pairs) shows no homology to that of the consensus (CRE) CAMP-regulatory element. Gel retardation analysis shows that a protein-DNA complex is formed between this DNA sequence and nuclear proteins from mouse adrenal Y 1 tumor cells or bovine adrenal cortical cells or human fetal adrenal tissue and that formation of this complex cannot be competed by DNA containing the consensus CRE sequence. Even though CAMP-en-hanced accumulation of P-450~~~ in primary cultures of bovine adrenocortical is by the enhanced by the CAMP-re- -129/-96-base pair fragment of kinaseor poly(dI-dC), cpm DNA

tochrome P-450 superfamily of genes and appears to be expressed exclusively in the adrenal cortex, thereby playing an essential role in both glucocorticoid and mineralocorticoid synthesis. In humans and other species investigated to date, there are two closely related CYP21 genes (Nebert et al., 1989), one being the functional gene and the other being a pseudogene. These genes are located on human chromosome 6 (White et al., 1985), and upon their isolation and characterization (Higashi et al., 1986;White et al., 1986) it has been shown that human gene CYP2lB encodes the active enzyme while human gene CYP2lA is the pseudogene. In addition to the tissue-specific regulation of human CYPPlB which appears to limit its expression to the adrenal cortex, its expression in cattle (and presumably humans) is regulated transcriptionally by adrenocorticotropin via CAMP (John et al., 1986a). This mechanism maintains optimal levels of 21-hydroxylase activity in the adrenal cortex and, consequently, optimal capacity for synthesis of glucocorticoids and mineralocorticoids. Furthermore, the CAMP-dependent increase in P-450~~~ mRNA has been shown to require on-going protein synthesis in primary cultures of bovine adrenal cortical cells and to require several hours to become manifest following initiation of treatment (John et al., 198613). In the mouse it is found that CypBlA is the functional gene and Cyp21B is the pseudogene . Parker, Schimmer, and colleagues have carried out detailed studies to identify the cis-regulatory elements associated with the mouse Cyp2lA gene which are required for its transcription (Handler et al., 1988;Handler et al., 1989). These studies have shown that both tissue-specific and hormone-inducible expression of the mouse Cyp2lA gene require sequences between -330 bp in the 5'-flanking region and the promoter. Deletion of sequences between -230-and -180-bp abolishes constitutive expression of this gene , while deletion of sequences between -330 and -156 bp abolishes CAMP-enhanced transcription (Handler et al., 1988 Saiki et al. (1988) using a Perkin-Elmer/ Cetus thermal cycler. All enzymatic manipulations were done according to standard procedures (Maniatis et al., 1982). All newly made  plasmid  constructs  were confirmed  by dideoxy  chain-termination  sequence  analysis  @anger  et al., 1977). Plasmids-The published numbering system of the 5'-flanking sequences of the human CYPZlB gene begins with -1 being the first base to the 5'-side of the ATG initiation of translation (Higashi et al., 1986;White et al., 1986), while for the bovine P-450& gene (Chunget al., 1986) and the mouse CypZlA gene  it begins with the first nucleotide to the 5' side of the initiation of transcription.
Since the site of initiation of transcription of the human CYPZlB gene is known to be nine nucleotides 5' from the ATG (Higashi et al., 1986), in this paper we have used the convention of numbering the first nucleotide to the 5' side of this site of initiation of transcription as -1, and thus the numbering of the 5'-flanking sequences of the human, bovine, and mouse genes cited herein all begin at the same relative position. (S&I), and lOOCAT (KpnI). Plasmids in which the CAT gene was fused to the SV40 early promoter, SV2CAT (Gorman et al., 1982a) or to the 3' long terminal repeat of Rous sarcoma virus, RSVCAT (Gorman et al., 1982b), were used as transfection controls. The plasmids OVEC and SV40-OVEC were kind gifts from Drs. Thomas Gerster and Walter Schaffner (University of Zurich) (Westin et al., 1987). The plasmid 3 x CRE, which contained three tandemly repeated CAMP-responsive elements of the human chorionic gonadotropin-a gene fused to the OVEC fl-globin sequences, was made by Johan Lund (Karolinska Institute, Huddinge, Sweden) as described previously .  (Gospodarowicz et al., 1977;Funkenstein et al., 1983). Primary cultures of human fetal adrenal cells were prepared as described by Mason et al. (1985). Cells were transfected by the calcium phosphate precipitation method (Graham and Van der Eb, 1973) using 10 pg of plasmid/lOO-mm dish. After 4 h the cells were treated with 15% glycerol for 3 min and then were incubated with fresh medium overnight. The next day the cells were maintained in the presence or absence of 25 pM forskolin in fresh medium for the indicated time periods. CAT Assay-Cell lysates were prepared and CAT assays performed as described by Gorman et al. (1982aGorman et al. ( , 1982b. The acetylated products of ['4C]chloramphenicol (Amersham Corp.) were analyzed on thin layer chromatography plates (PE SIL G, Whatman). After autoradiography the spots were excised, and the amount of radioactivity was determined by liquid scintillation counting.
The plasmid RSV-GPT in which the 3'-long terminal repeat of the RSV was fused to the GPT gene was cotransfected with the CAT constructs to normalize the transfection efficiencies using the GPT assay (Chu and Berg, 1985).
Preparation of RNA and Sl Protection Assay-The preparation of RNA -and Sl nuclease analysis of transcription from the OVEC nlasmids were nerformed as described previously . Cytoplasmic RNA (50 pg) was hybridized with a 5' end-labeled oligonucleotide probe extending between positions -18 and +75 bp on the non-coding strand in the rabbit P-globin gene (Westin et al., 1987). The hybridization mixtures were digested with Sl nuclease, resolved on 10% polyacrylamide, 7.5 M urea gels, and visualized by autoradiography.
The bands corresponding to the correctly initiated transcripts were cut out of the gels, and the radioactivity was determined by liquid scintillation counting.

RESULTS
Transient Expression of Human P-450~~~ Reporter Gene Fu.sion.s-In order to investigate the transcriptional regulation of the P-450cs1 gene, we constructed two series of plasmids, OVEC and CAT. In experiments utilizing the OVEC constructs, the rabbit /3-globin TATA box (promoter) was used to initiate transcription. All of the CAT constructs contain P-450~~~ sequences beginning from -10 bp at 3' end in order to utilize the TATA box (promoter) of the P-450~~~ gene. Mouse adrenocortical tumor Yl cells were transfected with both the OVEC and CAT plasmids, and cells were incubated with 25 pM forskolin which raises intracellular CAMP levels (Seamon et al., 1981). When the plasmid 1.3OV21 was transfected into Yl cells, a 4-fold increase in correctly initiated globin transcripts is observed upon forskolin treatment (Fig. 1). Transfection with the CRE-containing OVEC plasmid leads to a 13-fold increase in correctly initiated transcripts upon forskolin treatment. From cells transfected with OVEC (OV, containing only the minimal globin promoter) very low levels of protected correctly initiated transcripts are observed, both in the presence and absence of forskolin treatment. The plasmid SV40-OVEC in which the SV40 enhancer is inserted upstream of the minimal globin promoter (Westin et al., 1987) was used as a positive transfection control, and this plasmid yields high levels of globin transcripts both in the presence and absence of forskolin treatment. A 6-h treatment with forskolin was shown to be optimal for CAMP-enhanced transcription directed by the CYPPlB 5'-flanking region (data not shown) just as shown previously for that directed by the 5'-flanking region of the bovine P-450,,, gene .
Identification of Regulatory Elements between -1697 and -30 bp of the CYP21B Sequences-The location of the CAMPregulatory region(s) was identified using a series of plasmids deleted from the 5' end as described under "Materials and Yl cells were transfected with 1.3OV21, OV, 3XCREOV, and SV4OOV, treated with forskolin for 6 h, and cytoplasmic RNA was isolated. Incorrectly initiated transcripts arising prior to the &globin start site (i.i) and correctly initiated transcripts (ci.) in the Sl nuclease analysis are indicated.
Methods." When the sequences from -1305 to -389 bp are removed from 1.3OV21, the level of expression of the globin gene induced with forskolin was reduced gradually as was the basal level of expression, although the fold stimulation by forskolin was maintained at approximately the same level (Fig. 2). The plasmid 25OOV21 in which the sequences from -1305 to -250 bp were deleted from 1.3OV21 gives the highest levels of both basal and induced expression without any significant change of the stimulation ratio. These results suggest that there may be one or more weak negative transcriptional regulatory sequences between the positions -1305 to -250 bp. When an additional 18 bp are removed from 25OOV21, producing 232OV21, the expression of the globin gene is reduced substantially both in the absence and presence of forskolin although moderate levels of expression remain and the fold stimulation by forskolin remains unchanged. This substantial change in both the basal and CAMP-enhanced expression indicates that the deletion from -250 to -232 bp removes sequences which enhance human CYPBlB expression.
When the plasmid 129OV21 containing the P-450~2, sequences from -129 to -96 bp is transfected into Yl cells, the expression upon forskolin treatment again increases, while the basal expression does not show a significant increase. The resulting stimulation ratio, therefore, is increased up to lofold. This observation locates a CAMP-responsive cis-regulatory element of the human CYP21B gene between -129 and -96 bp. The plasmid 100/3OOV21 containing the P-450~~~ sequences from -100 to -30 bp fused to P-globin gene yields levels of expression as low as those with OVEC alone which only contains a minimal globin promoter. No enhancement of transcription by forskolin treatment is observed with 1001 3OOV21 indicating that the major CAMP-responsive element in this gene is located between -129 and -96 bp.
In the experiments involving transient transfection of the OVEC constructs, the effects of P-450c21 sequences to enhance transcription by the globin promoter are analyzed by measuring the expression of globin RNA using an Sl nuclease protection assay. In order to compare the results using the globin promoter with results using the P-450~~~ promoter, we Transcripts were analyzed by the Sl nuclease protection method followed by polyacrylamide gel electrophoresis. Radioactivity corresponding to correctly initiated transcripts was determined by liquid scintillation counting and is represented as relative activity.
Open boxes indicate results from untreated cells and the hatched boxes results from forskolin-treated cells. The relative expression of OV in the absence of forskolin treatment is arbitrarily set as 1.0. Each value represents the mean of at least three separate transfection experiments, and the variation from the mean was always less than 20%. The results obtained upon expression of 100/3OOV21 and 2 X 129/ 96OV21 in human fetal adrenal cells (WA) are also shown. In human fetal adrenal cells, 8 pg of cytoplasmic RNA was used for Sl protection assay, this being the single case described in this paper where 50 rg of RNA was not used. also investigated the transient transfection of CAT constructs in which CYP21B sequences containing P-450~~~ TATA box are fused to the CAT gene. After transfection, cells were treated with 25 pM forskolin for 14 h, and the CAT enzymatic activities of the cell lysates were analyzed. As shown in Table  I, a series of plasmids containing the P-450czl-CAT fusion genes gives a similar pattern of CAMP enhancement as does the series of OVEC constructs. Sequences required for CAMPinduced CAT activity are found to be located between -389 and -100 bp. In contrast to the results from the series of OVEC constructs, all of the basal activities of the CAT constructs are as low as that of BSCAT which was used as a negative control. This difference in basal expression between the OVEC constructs and the CAT constructs may arise from the different sensitivities of the assay methods or from the fact that they utilize different promoters.
From this series of experiments, the presence of two short sequences in the 5'-flanking region of the human P-450~2~ which serve as positive regulatory elements of expression is suggested. Because removal of sequences between -250 and -232 bp from 25OOV21 yields a significant decrease of expression in the presence and absence of forskolin treatment, it is suggested that the sequence between -250 to -232 bp includes a positive regulatory element. This element may only be involved in basal expression since the fold increase observed upon forskolin treatment is not changed between 25OOV21 and 232OV21. On the other hand, comparison of the results obtained using the plasmids 129OV21 and 100/3OOV21 suggests the presence of a CAMP-responsive cis-acting element within the sequence from -129 to -96 bp.
Nevertheless, since the removal of the sequence between -250 and -232 from 25OOV21 results in significant decreases of expression in the presence and absence of forskolin treatment, we then tested whether this region alone might confer a CAMP responsiveness to a globin promoter. We constructed the plasmids lX250/225OV21 and 2X250/225OV21, containing one copy of the -250/-225-bp sequence and two tandemly repeated copies of this sequence, respectively, upstream of the globin promoter. The plasmids lx129/96OV21 and 2x129/ 96OV21 which contain one copy and two tandemly repeated copies of the -129/-96 bp sequences, respectively, were also tested. As can be seen in Fig. 2, the plasmids lX250/225OV21 and 2X250/225OV21 do not yield any significant increase on the expression of globin gene upon forskolin treatment. Unexpectedly, the basal level of expression in the presence of these sequences was the same as with the OVEC plasmid and considerably less than with 25OOV21. In order to see whether the sequence from -250 to -225 bp might not contain a complete CAMP-responsive sequence, the plasmids 250/ 2010V21,250/1800V21 and 250/159OV21 which contain the sequences from -250 to -201, -250 to -180, and -250 to -159 bp, respectively, but not the CAMP-responsive sequence between -129/-96, were tested. These plasmids also yield a background level of expression and no significant increase upon treatment with forskolin (data not shown). These results suggest that the sequence from -250 to -232 functions in a positive fashion on basal transcriptional regulation only in the presence of the CAMP-responsive element -129/-96. The plasmid 2X129/96OV21 gives higher levels of globin gene expression than does the plasmid lx129/96OV21. The 2x129/ 96OV21 construct was also tested in primary cultures of human fetal adrenal cells. As seen in Fig. 2, forskolin treatment dramatically enhances correctly initiated globin transcripts from this construct in human fetal adrenal cells, just as it does in mouse Yl tumor cells and primary bovine adrenocortical cells (data not shown).

The Effect of Cycloheximide
Treatment in Transient Transfection-Cycloheximide, an inhibitor of translation, has been shown to inhibit the adrenocorticotropin-induced accumulation of bovine P-450cZ1 RNA in bovine adrenocortical cells in primary culture (John et al., 1986b). As noted above, the two sequences -2501-232 and -1291-96 bp are suspected to confer basal expression and CAMP responsiveness, respectively, to the human CYP21B gene. We therefore examined the effect of cycloheximide on the transient expression of the globin gene in Yl cells transfected with 25OOV21 or 129OV21. The increased expression of globin transcripts upon forskolin treatment is not inhibited by cycloheximide in Yl cells transfected with either plasmid but rather was moderately increased (Fig. 3). Treatment with cycloheximide alone also moderately increases the basal expression of the globin gene.
Nuclear Proteins Bind to the -250/-225 and -1291-96 bp Regions of CYPZlB-To investigate the interaction of nuclear proteins with the two regions which we have identified as being important for CYP2lB transcription, -250 to -225 and -129 to -96 bp, these fragments end-labeled with 32P were incubated with nuclear extracts prepared from Yl cells in the presence of the nonspecific polynucleotide competitor, poly(dI-dC). Resulting protein-DNA complexes were analyzed by non-denaturing polyacrylamide gel electrophoresis and visualized by autoradiography. The DNA-protein complexes of the fragments show no significant difference in mobility between nuclear extracts from Yl and bovine adrenocortical cell cultures and human fetal adrenal tissue (Fig. 4A), suggesting not only the presence of nuclear proteins which bind specifically to fragments -129/-96 and -2501-225 bp but Yl cells were transfected with 1 X 129/96OV21 and 25OOV21 and incubated for 6 h without any treatment or with 25 FM forskolin or with 40 pM cycloheximide or with 25 pM forskolin and 40 PM cycloheximide. Sl nuclease protection analysis was performed on the cytoplasmic RNA isolated from transfected cells, i.i., incorrectly initiated transcripts; c.i., correctly initiated transcripts. also that the protein factors are conserved among these three different species.
Since the sequences between -129/-96 bp of the P-450cZ1 gene confer CAMP responsiveness to the globin promoter we tested whether an unlabeled DNA fragment containing the CRE of the human n-chorionic gonadotropin gene (Deutsch et al., 1987) or the CAMP-responsive -243/-225-bp fragment of bovine P-450,;,, gene  would compete for Y 1 nuclear proteins binding to a labeled fragment containing the -129/-96 bp region. The complex formation between a protein factor and the -129/-96-bp fragment was not inhibited by either of these sequences even though 200.fold excess amounts of unlabeled fragments were used (Fig. 4B). Also we have observed that -250/-225 does not inhibit protein binding to -129/-96 (data not shown).

DISCUSSION
In this paper we report the sequences within the 5'.flanking region of the human CYP21B gene which are involved in its transcriptional regulation by CAMP. The sequence from -129 to -96 bp of the P-450c.Z, gene can confer increased CAMP responsiveness to a reporter gene. This sequence, 5'-GGCCACTCTGTGGGCGGGTCGGTGGGAGGGTACC-3', shows no apparent similarity to the CRE consensus sequence, TGACGTCA (Montminy et al., 1986;Short et al., 1986) or the AP-2-binding sequence, CCCCAGGC, (Imagawa et al., 1987;Roesler et al., 1988). A series of experiments by Parker, Schimmer, and their colleagues examining the expression of the mouse Cyp2lA gene also indicates the absence of known CAMP-responsive sequences in the 5'-flanking region of the functional P-450cZ, gene in this species Chaplin et al., 1986;Handler et al., 1988). Additionally, the consensus CRE sequence does not inhibit the DNA-protein complex formation between the sequence -129/-96 bp of P-450(.,, and nuclear extracts from Yl cells (Fig. 4R). Thus, it is concluded that the P-45Oc,, gene is regulated by a mechanism involving transcription factors which are different from those required for genes involving consensus CRE sequences. However, it should be noted that the sequence -243/-225 bp which confers CAMP responsiveness to the bovine P-450,:,, gene is also without any apparent similarity to the consensus CRE sequence, yet the CRE from the human n-chorionic gonadotropin gene competes binding of nuclear proteins to this sequence . The P-450,;,, -243/-225bp sequence shows no apparent similarity to the sequence -1291-96 bp of human P-450(.,,, and as seen in Fig. 4B it does not inhibit the DNA-protein complex formation of this sequence of human P-450(.,,. Thus, the CAMP-dependent transcription of the P-450cX, gene is regulated by a transacting protein factor which is different from that of the other adrenocortical microsomal steroid hydroxylase, P-4501;,,. The sequence within -129/-96 bp of the human CYPBlB gene is somewhat conserved at approximately the same position in the mouse and bovine genes (Fig. 5). From gel shift assays, the DNA-protein complexes formed between this sequence and nuclear extracts from mouse Yl cells, bovine adrenocortical cells, and human fetal adrenal tissue show about the same mobility (Fig. 4A). Also, the 5'-flanking sequence of the human P-450c,, gene confers increased expression to a reporter gene in primary cultures of both bovine adrenal cells and human fetal adrenal cells, as well as in mouse Yl cells. Thus, it can be suggested that this region of the P-450c,, gene is involved in CAMP responsiveness in mouse, bovine, and human adrenals.
Studies by Parker, Schimmer, and colleagues have indicated that the sequence required for adrenocorticotropin (CAMP)-dependent enhanced transcription resides between -330/-156 bp of the mouse P-450c2, gene (Hindler et al., 1988;Handler et al., 1989), although the precise location of the CAMP-responsive element within this sequence remains to be elucidated. These results argue against the mouse sequence in Fig. 5, which is related to the -129/-96, being the CAMP-responsive element in the mouse Cyp21A gene. Therefore, differences may exist between the human CYPBlB and the mouse CypBlA gene, with respect to CAMP-dependent regulation. It must be noted, however, that the CAMP-responsive element in the human gene (-129/-96 bp) has been localized by analysis of RNA transcripts using a heterologous promoter while localization ' (Chung et al., 1986) and mouse   of the region of the mouse gene containing the CAMP-responsive element (-330/-156 bp) involved use of the homologous promoter and measurement of protein products (growth hormone or CAT). Also, constitutive expression of the human CYP21B gene and the mouse Cyp21A is clearly different when comparing results of Parker et al. (1986) and those herein. Deletion of sequences between -230 and -180 bp in the mouse Cyp2lA gene dramatically reduced constitutive expression and suggested the presence of an enhancer for constitutive expression within this sequence. The results in Fig. 4 show no evidence for such a sequence in the human CYP21B gene. Finally, Parker et al. (1986) demonstrated the presence of a 40 bp highly conserved sequence in the 5'-flanking region of the P-450~~~ gene among different species (mouse, bovine, and human). This sequence is located between -231 and -191 bp in the human P-450cz1 gene (nucleotide numbering system used herein) and is not found to be required for either basal or CAMP-dependent expression in the present study. Thus, there appear to be differences in regulatory elements associated with both basal and CAMP-dependent expression between the human and mouse P-450~~~ genes which can not be explained simply by the fact that experiments have been carried out in different laboratories. At present we can only conclude that the sequence -129/-96 bp of the human CYP2lB gene contains within it a CAMP-responsive element which serves to enhance transcription of a reporter gene in human, bovine, and mouse adrenal cells. How this element relates to similar elements in the mouse and bovine P-450cz1 genes is not clear at this time. The sequence -129/-96 in the human P-450~~~ gene contains within it an almost perfect Spl-binding site (G/TGGGCGGG/AG/AC/T; Kadonaga, et al., 1986), and the corresponding sequence in the bovine P-450cz1 gene contains a perfect Spl-binding site. However the relationship between an Spl-binding site and CAMP-dependent transcription is not yet apparent. 450,,, gene a CAMP-responsive sequence is also found . There is some sequence relatedness between a segment within this 100 bp region of the P-450,, gene and the -243/-225-bp sequence of the P-450i7, gene, but it is not yet known whether this region of homology is required for CAMP-dependent P-450,,, expression. No consensus CRE is found in the P-450,,, -183/-83-bp sequence. We now report another CAMP-responsive sequence, this from the human P-450~~~ gene (-129/-96 bp), which does not contain a consensus CRE and is apparently not related to the other CAMPresponsive sequences associated with the steroid hydroxylase genes. It is becoming evident that each steroid hydroxylase gene contains its own set of CAMP-responsive sequences, and presumably, there exist a number of different transcription factors, each conferring CAMP responsiveness to a particular sequence element on a particular steroid hydroxylase gene.
Although the sequence -2501-225 bp in human CYP2lB forms a DNA-protein complex with nuclear extracts from Yl and bovine adrenocortical cells, and human fetal adrenal tissue, the function of this sequence remains unknown. Deletion of this sequence from 25OOV21 reduces the basal expression dramatically. However, the plasmids 1~250/225OV21 and 2X250/225OV21 show neither elevated basal expression nor CAMP responsiveness suggesting that this sequence requires the other cis-elements to function. As shown in Fig. 2, the plasmid 25OOV21 produces maximum expression both in the presence and absence of forskolin treatment, while 232OV21 and 21OOV21 produce lower levels of CAMP-dependent expression than does 129OV21 which contains only the sequence -129/-96 bp of P-450~~~. Perhaps there is a modest negative regulatory element with respect to CAMPresponsive transcription located between -232 and -129 bp. If so, we could speculate that the protein factor(s) binding to the sequence -250/-225 bp release the negative regulatory effect located between -232 and -129 bp. Clearly, further investigation is required to elucidate the function of the sequence between -250 and -232 bp in transcription of the human CYP2lB gene.
In addition to the P-450~~~ gene, transcription of the P-450,, gene, the P-45017, gene, the P-450ne gene, and the adrenodoxin gene is responsive to CAMP (John et al., 1986a). The mouse P-45O11p gene has been found to be regulated by a modified CRE, TGACGTGA (Mouw et al., 1989;. The bovine P-45017, gene is found to contain two CAMP-responsive sequences, neither of which is homologous to the other nor to the consensus CRE . Within a 100-bp sequence (-183/-83 bp) of the bovine P-