Both the basal and inducible transcription of the tyrosine hydroxylase gene are dependent upon a cAMP response element.

The cAMP response element (CRE) mediates cAMP responsiveness in many eukaryotic genes (Roesler, W. J., Vandenbark, G. R., and Hansen, R. W. (1988) J. Biol. Chem. 263, 9063-9066). The tyrosine hydroxylase gene (TH) contains a single copy of a consensus CRE at -45 to -38 base pair (bp) upstream of the transcription initiation site. Deletional and mutational analyses of the upstream 2400-base pair region of the rat TH gene using transient transfection assay demonstrated that the CRE was essential for both cAMP-mediated induction and basal transcription of the TH gene. Another domain between -365 and -151 bp, containing the AP1 site, contributed to transcription to a smaller degree. Thus, the CRE appears to play an important dual role as a basal promoter element and an inducible enhancer for TH transcription. Interactions between the DNA binding factors in nuclear extract and CRE-containing oligonucleotides were investigated by gel retardation and competition assays. Oligonucleotides corresponding to the CRE regions of the TH or somatostatin gene gave rise to a pair of distinct protein-DNA complexes with identical mobilities in the gel retardation assay, suggesting that similar nuclear factor(s) might bind to the CREs of the TH and somatostatin genes. This study emphasizes a fundamental role of the CRE in transcriptional activation of the TH gene in catecholaminergic cells.

limiting step in the biosynthetic pathways for dopamine, norepinephrine, and epinephrine by catalyzing the conversion of L-tyrosine to 3,4-dihydroxy-~-phenylalanine (Nagatsu et al., 1964). A variety of trans-synaptic and hormonal stimuli, acting via different signal transduction pathways, can produce acute (minutes to hours) increases in TH activity or somewhat delayed (one to several days) elevations in levels of TH (Thoenen et al., 1969;Black et al., 1987;Zigmond et al., 1989). The latter changes correlate with elevations in mRNA for TH, indicating that transcriptional regulation of TH gene expression plays an important role in the long term regulation of catecholaminergic transmission.
Among the important regulators of TH, cAMP has been shown to stimulate increases in T H activity by promoting phosphorylation of T H enzyme by CAMP-dependent protein kinase (PKA) (Joh et al., 1978). More recently, cAMP has also been shown to mediate increased transcription of mRNA for T H . cAMP induces the transcription of a variety of genes via a consensus octamer, 5'-TGACGTCA-3', the cAMP response element (CRE), found in the 5"upstream regions of genes regulated by cAMP (Roesler et Goodman, 1990). The protein that binds the CRE of the somatostatin gene has been characterized and cloned (Montminy and Bilezikjian, 1987;Gonzalez et al., 1989). This protein, known as CRE-binding protein (CREB), activates the transcription of responsive genes after it is phosphorylated by PKA (Yamamoto et al., 1988;Gonzalez and Montminy, 1989;Lee et al., 1990). Inspection of the TH gene and functional analysis of fusion gene constructs have identified a putative CRE at -45 to -38 bp upstream of the transcription initiation site (Harrington et al., 1987;Fader and Lewis, 1990;Carroll et al., 1991). However, the function of this element in the context of the TH upstream region has been questioned. Indeed, some investigators have suggested that the CRE plays a relatively minor role in TH gene transcription (Cambi et al., 1989;Fung et al., 1992).
The present article presents data characterizing the upstream promoter region of the rat TH gene using the highly transfectable human neuroblastoma SK-N-BE(2)C cell line (Ross et al., 1981) and the less transfectable rat PC12 cell line (Green and Tischler, 1976). Both cell lines express high levels of T H endogenously. Functional analysis of fusion gene constructs, containing deletion mutations of the 5"upstream region of TH, localizes the minimal promoter elements to within 60 bp of the transcription initiation site (CAP), a region which includes the CRE. Deletion of the 5"region to -39, which removes 6 bp of the CRE, results in a dramatic loss of transcriptional activity. Furthermore, site-directed mutational analyses indicate that the TH CRE, in the proximal region of the TH upstream sequence (-45 to -38), is not only responsible for CAMP-mediated gene induction but is essential for the basal transcription of the T H gene in these catecholaminergic cell lines. In vitro binding studies detect two nuclear protein-DNA complexes of similar size using the CREs of the TH gene and somatostatin gene. In addition, the interaction of nuclear factors is analyzed further by competition assay using wild type and mutant oligonucleotides.

EXPERIMENTAL PROCEDURES
Cell Culture-Human neuroblastoma SK-N-BE(2)C cells were passaged in Dulbecco's modified Eagle's medium supplemented with 10% newborn calf serum. PC12 cells were grown in RPMI 1640 medium supplemented with 10% horse serum and 5% fetal calf serum. Each serum was used after heat inactivation. All culture media contained 100 units/ml penicillin and 100 pg/ml streptomycin.
Construction of Reporter Plasmids-TH(-503/+27)CAT plasmid was constructed in the pOCAT vector (Carroll et aL, 1991). This construct contains 503 bp of the 5"flanking sequence of the rat TH gene, the transcription initiation site, and the first 27 bases of untranslated sequence of the TH transcript. To facilitate further constructions, the plasmid backbone was replaced with that of pBLCAT3 (Luckow and Schutz, 1987) by ligating the 4.0-kb BamHI-EcoRI fragment of pBLCAT3 with the 800-bp BamHI-EcoRI fragment of TH(-503/+27)CAT plasmid. Then, BglII-BamHI genomic fragment ranging from -2400 to -503 was inserted into the BamHI site of TH503CAT, resulting in TH2400CAT. A series of deletion constructs were made utilizing the unique PstI site in the pBLCAT3 and appropriate restriction sites in the upstream region of the T H gene. The junctions between TH upstream region and CAT gene of the fusion constructs were confirmed by sequencing analysis. To introduce mutations into the TH CRE, oligonucleotide-derived sitedirected mutagenesis was performed. An M13 mp19 subclone containing the 700-nucleotide BamHI-EcoRI fragment of TH2400CAT plasmid was used as a template. Two 23-mer oligonucleotides of the sequence 5' AGGCCAGGCTGAAAGCCCCTCTG 3' and 5' GCCAGGCTGACCTCAAAGCCCCT 3' were utilized as described to produce deletion and point substitution mutations (Nakamaye and Ekstein, 1986). The presence of mutations was confirmed by sequence analysis. The same 700-bp BamHI-EcoRI fragments containing mutations were used to reconstitute the final mutant reporter plasmids. For this step, the additional EcoRI site present in the 3"polylinker site of TH2400CAT and TH503CAT plasmids was removed by partial digest and fill-in reaction. This step did not affect the expression of CAT.
Transfection and CAT Assay-Transfection of reporter plasmids into SK-N-BE(2)C and PC12 was performed by the calcium phosphate co-precipitation method (Gorman et al., 1983). In all experiments, pRSV8gal plasmid containing the @-galactosidase gene linked to the RSV promoter/enhancer (Edlund et al., 1985) was included as an internal control for the different transfection efficiencies between experiments. When cells reached approximately 50% confluency, each 6-cm Falcon tissue culture dish received 4 pg of TH2400CAT plasmid and 1 pg of pRSVPgal plasmid. For shorter reporter plasmids, the amount of DNA was adjusted to the same molar ratio, and the total amount was maintained at 5 pg by supplementing with PUC19 DNA. Cells were exposed to the precipitate for 16 h and media was replaced with Dulbecco's modified Eagle's medium, 10% newborn calf serum. After a further 24-h incubation, cells were harvested by scraping with a rubber policeman in phosphate-buffered saline. Forskolin and 3isobutyryl-1-methylxanthine (IMX) were added directly into media 16 h prior to harvest. Extracts were prepared by resuspending cells in 200 pl of 0.25 M Tris-HC1 (pH 8.0), exposure to three freeze-thaw cycles, and then heating at 60 "C for 10 min to inactivate endogenous acetylase. All plasmid DNA was separated on CsCl gradients by ultracentrifugation twice and further purified by phenolchloroform extraction twice and by EtOH precipitation twice in the presence of 2.5 M ammonium acetate. CAT activity was measured using 0.5 pCi of [14C]chloramphenicol, n-butyryl coenzyme A (200 p~) , and 10-20 p1 of cell extracts at 37 "C for 30 min. Expression of pRSVPgal did not vary with forskolin treatment and was used to correct for any variations in transfection efficiency. Cell extracts corresponding to the same 8-galactosidase activity were used in the CAT assay. In most experiments, the variation of 8-galactosidase activity was within the range of 20%. Butyrylated reaction products were separated on TLC plates, exposed overnight at -7O'C with intensifying screen, and visualized by autoradiography.
Gel Retardation Assay-Nuclear extracts were made from SK-N-BE(2)C and PC12 cells based on described procedure (Dignam et al., 1983). The pellet was resuspended in Dignam's Buffer D and quickto 50% saturation with (NH4),S04. In PC12 cells, this step made the frozen in liquid N1. In certain cases, the prepared extract was brought formation of complex I1 more apparent. Sense and antisense strands of oligonucleotide (Fig. 4A) were annealed and labeled using T4 polynucleotide kinase and [w32p]ATP. Alternatively, oligonucleotides were labeled by fill-in reaction using Klenow and [y-3$]dCTP. Nuclear protein-DNA binding was carried out at room temperature for 20 min. Nuclear extract (1-4 pg of protein) was incubated with 20,000-40,000 cpm of labeled probe (0.04-0.1 ng) and 1 pg of poly(d1-dC) .poly(dI-dC) in binding buffer (10 mM Tris (pH 7.5), 100 mM NaCl, 1 mM dithiotbreitol, 1 mM EDTA, 4% glycerol). The complexes were resolved on nondenaturing 6% polyacrylamide gels. Gels were prewarmed by electrophoresis for 1 h at 10 V cm" prior to loading samples. Samples containing bromphenol blue and xylene cyano1 were electrophoresed for 2 h. Gels were dried and visualized by autoradiography.

RESULTS
TO study T H gene regulation, rat T H genomic clones containing 5"flanking sequences were isolated (Carroll et al., 1991). Potential cis-acting motifs such as AP1, AP2, POU/ OCT, SP1, and CRE are located in the 5'-proximal region of the rat (Harrington et al., 1987;Carroll et al., 1991) and the human genes (Kobayashi et al., 1988) (Fig. L4). Both the nucleotide sequences of these motifs (under boxes) and their relative distance (numbers above boxes) from the CAP site are highly conserved in both species. The 5'-proximal sequence contains a single copy of the palindromic consensus CRE (5'-TGACGTCA-3'); located only 7 and 8 base pairs upstream of the TATA box in the human and rat genes, respectively ( Fig. lA). To localize important cis-acting elements, a series of plasmids were constructed containing different lengths of the rat TH upstream region fused to the bacterial CAT gene as a reporter gene (Fig. 1B). SK-N- BE(2)C cell line expresses levels of TH message similar to those of human adrenal medulla, as assessed by Northern blot analysis (Carroll et al., 1991). The Capo4 co-precipitation method reproducibly provided a high transfection efficiency in the SK-N-BE(2)C cell line as quantitated by the /3-galactosidase histochemistry (>20%; data not shown). Although PC12 cells were transfected with a much lower efficiency (1-2%) than SK-N-BE(B)C, this was still significantly higher FIG. 1. Rat TH-CAT fusion genes and determination of transcription initiation site in the fusion gene construct. A, schematic diagram of putative cis-acting elements residing in the upstream region of the rat and human TH genes. Nucleotide sequences and the relative position of the proximal nucleotide in each motif in relation to the transcription initiation are shown below and above each box, respectively. B, CAT fusion constructs containing different length of TH 5"upstream region. Restriction enzymes which were used for deletion construction are indicated above. Arrow represents the transcription initiation site. than the transfection efficiency reported previously (0.1%) in PC12 cells (Gandelman et al., 1990). Primer extension analysis using poly(A+) RNA prepared from SK-N-BE(2)C cells transfected with TH2400CAT fusion construct verified that the CAT activity measured in our transient transfection assay resulted from proper transcription initiation (data not shown; Fig. L4).

Deletional Analysis of 5"Flanking Sequence Reveals Two Potentially Important Regions for TH Gene Transcription-
The expression of different fusion constructs was compared using the transient transfection method in SK-N-BE(2)C cells. Deletions from the -2400-bp region to the -773-, -503-, or to -365-bp region did not influence CAT activity (data not shown). Thus, upstream sequences ranging from -2400 to -365 bp exerted no significant influence on basal transcription as measured by this transient transfection assay. When the upstream sequence was trimmed down further to -151 bp, a 30-40% drop in CAT activity was observed, suggesting the presence of a positive regulatory element or elements in the region spanning -365 to -151 bp. As depicted in Fig. l.4, several putative cis-acting elements such as AP2, AP1, and POU/OCT are located here. Further deletion to -108 or to -60 bp had little effect on residual transcriptional activity. Thus, the TH60 CAT construct, containing only 60 bp of proximal sequence, retained approximately 60% of the activity observed with the 2400-base pair construct, localizing the minimal upstream promoter to within -60 bp of the CAP site. Further deletion of 21 bp to -39 bp, however, abolished all remaining transcriptional activity, indicating that important basal promoter element(s) reside between -60 and -39 bP.
The response of these fusion constructs to forskolin treatment was tested in the presence of the phosphodiesterase inhibitor, IMX. In the SK-N-BE(2)C cell line, all fusion constructs except TH39CAT demonstrated approximately 3fold stimulation of CAT activity, localizing cAMP responsiveness to the proximal 60 bp (Fig. 2). TH39 CAT and the promoterless plasmid, pBLCAT3 (Luckow and Schutz, 1987), showed negligible basal expression and minimally responded to forskolin treatment. These data strongly indicate that the CRE, located a t -45 to -38 bp, is necessary and sufficient for forskolin induction.
It is possible that differences across species could influence the expression of the rat upstream region in the human cell line. To address this issue, the promoter analysis was replicated using rat PC12 cells as the host. Overall, a very similar profile of CAT activities was observed using the same series of deletional constructs in PC12 cell line. Again, removal of the -365 to -151 upstream regions decreased expression by 30-4096, TH6OCAT expressed approximately 60% of TH2400CAT activity, and further deletion to -39 bp decreased the CAT activity to the level of pBLCAT3 (data not shown). Also, forskolin treatment stimulated the transcriptional activity of all fusion constructs, with an exception of TH39CAT, by 3-4-fold (data not shown).
The TH CRE Is an Essential Promoter Element of TH Gene Transcription-The TH39CAT fusion construct is transcriptionally silent and unresponsive to cAMP induction. In the -60 to -39-bp region, there is no identifiable cis-acting motif besides the CRE. The fact that the TH39CAT lacks 6 out of 8 bases of the CRE motif led to the hypothesis that the CRE may not only be a CAMP-response element but also a core promoter element for T H gene transcription. T o test this possibility, we constructed deletion as well as base substitution mutants in the CRE and compared them with wild type constructs using transient transfection assays (Fig. 3). Strikingly, deletion of 5 internal bases of the TH CRE (GACGT out of TGACGTCA) abolished all transcriptional activity  4 0.7 14.8 46.1 16.1 43.2 12.2 33.6 11.3 26.7 4.4 20.7 0.4 0 from the rat 2400 and 503-bp upstream regions both in SK-N-BE(2)C (Fig. 3 B ) and PC12 cell line (data not shown). Furthermore, these mutant constructs were not responsive to forskolin treatment. Thus, the T H CRE appears essential both for basal transcription and induction in response to elevated CAMP. The effect of a single base change within the CRE was tested while maintaining the spatial and contextual surroundings of the TH upstream region. Based on the previous observation that two guanosines on the sense strand (-42 and -39 bp) and a third guanosine on the opposite strand (-43 bp) of T H CRE motif are protected from methylation by dimethyl sulfate through the interaction with the CRE-binding protein (ATF) , we mutagenized one of these residues. TH2400(42C + G)CAT, which has a single base substitution mutation a t -42 position in the context of the whole 2400 bp upstream, displayed a dramatic loss of transcriptional activity when assayed by transient transfection in SK-N-BE(2)C cell line (Fig. 3B). First, the basal transcription was profoundly decreased (>80%). Second, induction by forskolin treatment was substantially reduced when compared with wild type constructs (from 3.0 x down to 1.6 x). In PC12, TH2400(42C + G)CAT appeared to be as inactive as pBLCAT3 (data not shown).

Binding of Nuclear Proteins to the Wild Type and Mutant CRE in the TH
Gene-The interaction between the DNA binding factors in crude nuclear proteins and CRE-containing oligonucleotides was investigated by gel retardation assay. Oligonucleotides (23-mer) corresponding to the CRE of the rat T H ( W T oligo) and rat somatostatin genes (SOM oligo) were synthesized (Fig. 4A). These oligonucleotides do not share any apparent homologies except an identical octamer CRE motif. TH CRE oligonucleotides containing deletions ( D E L oligo) as well as point mutations were also made (Fig.  4A). Both the wild type oligonucleotide and the somatostatin oligonucleotide formed two distinct protein-DNA complexes ( I and I I ) with identical mobility in the gel shift assay (Fig.  4 B ) , suggesting that the same protein factor(s) may bind to the CRE in each gene. The same pattern of complex formation was observed when the probe was labeled by kinase or by fillin reactions, indicating that these two complexes were not related to single strand DNA binding activities. Deletion of 5 out of 8 bases in the TH CRE motif substantially reduced the affinity to nuclear proteins. Intriguingly, the same complexes re-emerged when excess amounts of crude nuclear proteins were used (Fig. 4 R ) . The relative affinity of wild type and mutant oligonucleotides was assessed by incubating with molar excess of unlabeled DNA in a parallel competition experiment using crude nuclear extracts isolated from PC12 cells (Fig. 4 C ) and SK-N-BE(2)C cells (Fig. 4 0 ) . In both experiments, it appears that approximately 25-fold excess of unlabeled deletion oligonucleotides over wild type oligonucleotides is required to displace the same amount of nuclear proteinswild type oligonucleotide complex, indicating that approximately 95% of the relative binding affinity is lost by the deletion mutation (Bokar et al., 1988). Thus, the in vitro binding assay of the deletion mutant is consistent with the in vivo transfection analysis (Figs. 3 and 4). When the point mutation oligonucleotide ( M 4 2 ) was employed as a competi- C, competition comparison between wild t-ype and mutant oligonucleotides using nuclear proteins from PC12 cells. 4 pg of nuclear proteins was incubated with y-:"P-labeled wild type oligonucleotide in the presence of the indicated molar excess of unlabeled wild type oligonucleotide, deletion oligonucleotide, M42 oligonucleotide, or 1-kb ladder DNA (Life Technologies, Inc.). Lane 1, no nuclear extracts; lane 2, no competitor DNA. 11, competition comparison between wild type oligonucleotide and deletion oligonucleotide using nuclear protein from SK-N-RE(2)C. Gel was run for a longer period of time to facilitate the separation of complexes I and 11. tor, 2-3-fold excess was required, which represents a moderate decrease (50-70%) in the binding affinity ( Fig. 4C; data not shown).

DISCUSSION
Using transient transfection assay, several investigators have suggested that the upstream region of the rat TH gene can drive transcription in a cell type-specific manner (Harrington et al., 1987;Cambi et al., 1989;Gandelman et d., 1990). Recently, Kaneda et al. (1991) showed that an 11-kb DNA isolate of the human T H gene containing the 2.5-kb upstream region, the entire exon-intron structure, as well as the 0.5-kb 3"flanking region can direct expression of human T H mRNA in the brains and adrenal glands of transgenic mice in a tissue-specific manner. This result indicates that the 5"upstream sequences involved in tissue specificity may largely reside within -2500 bp of the CAP site. In fact, detailed functional analyses of the 5"flanking region of the T H gene from different laboratories have produced conflicting results, leaving this issue unresolved. Chikaraishi and her colleagues (Cambi et al., 1989; proposed that the AP1-surrounding domain was essential for rat TH gene transcription by showing a dramatic decrease of expression (>go%) upon removal or mutation of sequences between -212 and -187 bp. However, other investigators have detected no decrease in basal expression after deletion of this AP1-surrounding region as measured by RNase protection or CAT assay (Gizang-Ginsburg and Ziff, 1990;Fader and Lewis, 1990). In addition, Gandelman et al. (1990) reported that the upstream sequence between -749 and -505 bp of the rat TH gene was also important for transcriptional activity in PC12 cells.
We examined expression driven by the 2.4-kb upstream region of the rat TH gene in both the human SK-N-BE(2)C and rat PC12 cell lines. The human neuroblastoma cell line, SK-N-BE(2)C, was reproducibly transfected with extremely high efficiency (>20%) and provided statistically reliable transfection analysis. Primer extension analysis demonstrated that CAT activity in SK-N-BE(2)C human cell line arises from correctly initiated transcription by the rat upstream sequence (data not shown). When the upstream sequence was deleted up to -365, no discernable effect on transcriptional activity was observed. Further deletion to -150 bp resulted in a 30-40% decrease in CAT activity in both the human SK-N-BE(2)C and the rat PC12 cells, suggesting the presence of positive element(s) between -365 and -150 bp. Our data are thus qualitatively consistent with those of other groups which showed that the AP1 site, located in this region (Fig. lA), is an important cis-acting element in T H gene transcription (Gizang-Ginsberg and Ziff, 1990;. Quantitatively, however, our results differed from some previously reported results (Cambi et al., 1989).
In our experiments, shorter constructs containing 108-or 60-bp upstream regions still retained 60% transcriptional activity compared with the 2400-bp construct in both the SK-N-BE(2)C and PC12 cells, thus localizing the minimal upstream promoter of the TH gene to the 60-bp region. Further deletion down to -39 bp abolished all residual activity, indicating the presence of critical basal regulatory elements in the region between -60 and -39 bp. Our deletional analysis therefore indicated that the consensus CRE, located between -45 and -38, might be crucial for both basal and inducible transcription of the TH gene. We tested this dual role of the CRE further by performing oligonucleotide-derived site-directed mutagenesis of the CRE motif in the context of intact 2400-bp and 503-bp upstream sequences. Deletion of 5 bases out of the octamer CRE motif rendered the entire 2400-bp upstream sequence as silent as the promoterless plasmid both for basal and inducible transcription. Even a single base substitution (42C + G) within the TH CRE severely reduced basal transcription (>80%) and forskolin induction. Previous investigators have suggested that the TH CRE may function only as a CAMP-inducible element Fader and Lewis, 1990;Huang et al., 1991;Fung et al., 1992). Our data, for the first time, clearly implicate the CRE as an essential basal element for T H gene transcription. All fusion constructs except TH39CAT showed about 3-fold induction by forskolin treatment. Notably, deletion of the AP1 and AP2 sites did not influence the inducibility of the upstream region in response to forskolin treatment. In contrast, Fung et al. (1992) showed that either the AP1 or the CRE would confer cAMP responsiveness when placed in front of the TATA box. One possible explanation for their observation is that the APl motif can function as a CRE-like element under certain circumstances (Comb et al., 1990). Clearly, our results show that in the native context, the CRE suffices for cAMP responsiveness of the TH gene. Thus, our deletional and mutational analyses indicate that the TH CRE is crucial for both basal and CAMP-inducible transcription of the TH gene in TH-expressing cells. These data clearly contrast with previous reports by other investigators which suggested that the CRE contributes little to basal transcription of the TH gene (Cambi et al., 1989;Fung et al., 1992).
The CRE motif has been found in the upstream regions of many genes, including neuropeptide genes, which are transcriptionally inducible by the elevation of intracellular cAMP concentration (Table I). Most CREs are located within the first 170 bp of the upstream region of their respective genes (the tyrosine aminotransferase gene being an exception), suggesting that the CRE may also function as a basal transcription element. Indeed, the dual role of the CRE as a basal and inducible transcription element has been suggested in several genes (Short et al., 1986;Andrisani et al., 1987;Roesler et al., 1988). While it remains to be determined how commonly the CRE performs a dual role in basal and inducible expression, our recent study on gene regulation of dopamine P-hydroxylase, which converts dopamine to norepinephrine, indicated that the CRE plays a similar dual role in transcriptional regulation of this gene (Ishiguro et al., 1993).
Gel shift assays demonstrated that the CREs of the somatostatin (SOM oligo) or T H genes ( WT oligo) incubated with crude nuclear extract of PC12 cells form similar patterns of DNA/protein complexes (Fig. 4B). These oligonucleotides do not share any sequence identities beyond the CRE and are presumably the targets of the same or similar CRE-binding proteins. Yamamoto et al. (1988) previously observed the same pattern of complex formation using the somatostatin CRE and nuclear extracts of PC12 cells. Hyman et al. (1988) suggested that several neuronally expressed genes, i.e. somatostatin, tyrosine hydroxylase, vasoactive intestinal polypeptide, and proenkephalin genes, might be co-regulated by a common trans-acting element. At present, definite identification of the specific transcription factor involved in TH gene regulation, via the CRE, awaits further investigation. IR vitro competition experiments showed that >95% of the relative affinity is lost by deletion mutation, in general agreement with the in vivo transfection result (Figs. 3 and 4). Nevertheless, it was rather surprising to observe that the deletion oligonucleotide retained some binding activity (Fig. 4B). In the deletion oligonucleotide, 5 out of the 8 bp in the CRE motif are missing. The resulting oligonucleotide contains an almost intact half-site of the palindromic octamer (3 out of 4 bp) as well as identical surrounding sequences. Yamamoto et al. (1988) showed previously that the half-site of the CRE   Boshart et al. (1990) retained some binding affinity to the CREB protein. We surmise, thus, that it is possible that the deletion oligonucleotide retains <5% binding affinity. The point mutation oligonucleotide (M42 oligo) represented a 50-70% decrease in the relative affinity. This contrasts with the more severe loss of transcriptional activity detected by the transfection analysis (Fig. 3). This may be due to the fact that the oligonucleotides used in the gel shift assays lack the functional context of the promoter region. For instance, the unusual structure of the TATA box of the TH gene (Fig. lA) might render its transcription more dependent upon the intact CRE sequence. Thus, in the native context, a point mutation of the CRE could lead to a profound impairment of transcriptional activation due to a lack of proper interactions with basic transcription factors, e.g. TFIID (Horikoshi et al., 1988).
We propose that the TH CRE is a key mediator for transcriptional activation of the TH gene. First, the TH CRE appears to be crucial for basal transcription in TH-expressing cells, since the transcriptional activity of the upstream sequence of the TH gene was severely impaired by deletion or single-base mutation of the CRE motif. Second, the TH CRE could mediate altered expression of the TH gene by activated neuronal signal transduction pathways which regulate the transcriptional activity of CRE-binding protein(s). Considering the essential role of the CRE for basal and CAMPinducible expression of the TH gene, it is tempting to speculate that the CAMP-dependent protein kinase (PKA)-signaling pathway might exert a dual role for T H gene regulation. Indeed, our recent analyses of several PKA-deficient PC12 cell lines demonstrate that the PKA system regulates both the basal and CAMP-inducible expression of the TH gene (Kim et al., 1993). Finally, the CRE might contribute to the tissue-specific transcription of the TH gene via interactions with other DNA-binding proteins at further upstream or downstream regions, as has been suggested for the a-gonadotropin gene (Delegeane et al., 1987). It was noteworthy that deletion of the TH CRE did not leave any transcriptional activity despite the presence of all the other 2400-bp upstream sequences (Fig. 3). Thus, the protein factor(s) that binds to the putative positive element(s) residing at -365 to -150 bp, e.g. AP1, might contribute to TH transcription in concert with CRE-binding protein (Gizang-Ginsberg and Ziff, 1990).