Neuron-specific expression of the human dopamine beta-hydroxylase gene requires both the cAMP-response element and a silencer region.

Dopamine beta-hydroxylase (DBH), the enzyme catalyzing the conversion of dopamine to norepinephrine, is specifically expressed in adrenergic and noradrenergic neurons in the central nervous system. DNase I hypersensitive sites were found in the 5'-flanking region of the DBH gene in noradrenergic human neuroblastoma SK-N-BE(2)C cells, but not in DBH-nonexpressing HeLa cells. We report here that the 4.3-kilobase upstream sequence of the human DBH gene confers cell type-specific expression as assessed by transient expression assay. Furthermore, deletional and mutational analyses revealed two genetic regulatory elements required for the regulation of cell type specificity. First, deletion of the cAMP-response element (CRE) abolished > 95% of the transcriptional activity by the DBH upstream promoter, thus implicating the CRE as an essential positive genetic element. Second, deletion of a region between -490 and -263 base pairs resulted in 10-fold increase of reporter gene activity only in HeLa cells, indicating that this region contains a cell-specific silencer. A 13-base pair fragment residing within that region shows 77% sequence identity with the neuron-specific silencer motif recently identified in two neuronal genes, i.e. SCG10 and type II sodium channel genes. We propose that the interplay between the CRE and this neuron-specific silencer region plays an important role in the tissue-specific expression of the DBH gene in noradrenergic cells.


Neuron-specific Expression of the Human Dopamine @-Hydroxylase
Gene Requires Both the CAMP-Response Element and a Silencer Region* (Received for publication, February 3, 1993, andin revised form, April 30,1993) Hiroshi Ishiguro, Kyong Tai  Dopamine &hydroxylase (DBH), the enzyme catalyzing the conversion of dopamine to norepinephrine, is specifically expressed in adrenergic and noradrenergic neurons in the central nervous system. DNase I hypersensitive sites were found in the 5"flanking region of the DBH gene in noradrenergic human neuroblastoma SK-N-BE(2)C cells, but not in DBH-nonexpressing HeLa cells. We report here that the 4.3-kilobase upstream sequence of the human DBH gene confers cell type-specific expression as assessed by transient expression assay. Furthermore, deletional and mutational analyses revealed two genetic regulatory elements required for the regulation of cell type specificity. First, deletion of the CAMP-response element (CRE) abolished ~9 5 % of the transcriptional activity by the DBH upstream promoter, thus implicating the CRE as an essential positive genetic element. Second, deletion of a region between -490 and -263 base pairs resulted in 10-fold increase of reporter gene activity only in HeLa cells, indicating that this region contains a cell-specific silencer. A 13-base pair fragment residing within that region shows 77% sequence identity with the neuron-specific silencer motif recently identified in two neuronal genes, Le. SCGlO and type I1 sodium channel genes. We propose that the interplay between the CRE and this neuron-specific silencer region plays an important role in the tissue-specific expression of the DBH gene in noradrenergic cells.
The differential expression of catecholamine biosynthetic enzymes Eundamentally distinguishes the biochemical phenotypes of both catecholamine-producing neurons and neurosecretory cells. Thus, dopaminergic neurons express only the first two enzymes in the catecholamine pathway, tyrosine hydroxylase (TH; EC 1.14.16.2.)' and L-aromatic amino acid decarboxylase (EC 4.1.1.28.), and thereby synthesize only dopamine. Noradrenergic and adrenergic cells express dopamine @-hydroxylase (DBH, EC 1.14.17.1), enabling them to synthesize norepinephrine. Finally, only adrenergic cells express phenylethanolamine N-methyltransferase (EC 2.1.1.28.), rendering them the only aminergic cells capable of producing epinephrine (Goodall and Kirshner, 1957). Neither the mechanisms governing differential expression of the foregoing enzymes within neurons and neurosecretory cells nor the mechanisms underlying neuron-specific gene expression are well understood. Those overlapping areas of inquiry are among the most important themes in molecular neurobiology. We report here the results of experiments to elucidate molecular mechanisms regulating expression of the DBH gene in the norepinephrine-synthesizing human neuroblastoma SK-N-BE(2)C cell line (Ciccarone et al., 1989), as well as those contributing to suppression of DBH expression in the nonneuronal HeLa cell line.
Recently, two laboratories independently demonstrated that a silencer region is directly involved in neuron-specific expression of genes encoding SCG10, a neuron-specific growth-associated protein (Mori et al., 1990), and the type I1 sodium channel protein (Maue et al., 1990). 5'-proximal sequences from each of those genes directed transcription in a variety of non-neuronal cell lines, but only after deletion of a putative silencer region. The silencer region thus appeared in large part to determine cell-specific expression of these neuronal genes (Maue et al., 1990;Mori et al., 1990). Surprisingly, further analyses of the silencer function by both groups revealed that the silencers working in these unrelated neuronal genes shared striking homologies. Moreover, the same sequence-specific protein(s), present only in nuclear extract from non-neuronal cells, bind to both sequences (Kraner et al., 1992;Mori et al., 1992). The foregoing studies suggest that negative genetic elements play a critical role in determining tissue-specific expression of various neuronal genes.
Numerous in vivo and in vitro experiments indicate that the TH and DBH genes share common features of regulation in response to various extracellular stimuli such as nerve growth factor (Acheson et al., 1984;Badoyannis et al., 1991), glucocorticoids (Otten and Thoenen, 1976;Kim et al., 1993c), CAMP analogues (Sabban et al., 1983;Lamouroux et al., 1993), and reserpine (Biguet et al., 1986;Wessel and Joh, 1992). Consistent with these observations, sequence analyses of 5'flanking regions demonstrated that the TH and DBH genes share various relevant cis-acting motifs, e.g CAMP-response element (CRE), glucocorticoid response element, APL, AP2 (Kobayashi et al., 1989;Cambi et al., 1989). In the present report, we isolated a genomic clone containing the 5"flanking sequence of the human DBH gene and characterized the 4.3kb upstream region by deletional and mutational analyses using transient expression assay in DBH-positive and DBHnegative human cell lines. We describe the identification of two genetic elements, one positive element and one silencer Neuron-specific Expression of the Human DBH Gene region, required for cell-type specific expression of the human DBH gene.

Northern Blot Hybridization
Poly(A+) RNA was prepared by oligo(dT)-cellulose affinity chromatography (Badley et at., 1988) and was subjected to Northern blot hybridization as described (Kim et al., 1993a). A DNA fragment corresponding to the nucleotide sequence from 1324 to 1695 bp of the human DBH cDNA (Lamouroux et al., 1987) was isolated by polymerase chain reaction, confirmed by sequence analysis, and used as a probe.
Analysis for DNase Z Hypersensitivity Sites Nuclear suspensions were prepared as described previously (Ip et aL, 1989) with the following modifications. Cells were grown to a concentration of -5 X lo6 cell/lOO-mm plate and washed twice with phosphate-buffered saline containing 2.5 mM EDTA. Nuclei were prepared by lysing the cells in 0.5 ml of ice-cold cell lysing buffer (10 mM HEPES, pH 8.0,50 mM NaCl, 1 mM EDTA, 0.25 mM EGTA, 0.5 mM spermidine, 0.15 mM spermine, 0.5% Triton X-100, and 0.25 M sucrose) for 3 min at 4 "C. Different concentrations of DNase I (0, 25, 50, 100 units) in 4.5 ml of digestion buffer (10 mM Tris, pH 7.4, 50 mM NaCl, 10 mM MgC12, 1 mM CaC12, and 0.25 M sucrose) was directly added to each 100-mm dish and incubated at 20 "C for 5 min. Each sample was transferred to a centrifuge tube, and the reaction was stopped by adding SDS to 1%, EDTA to 50 mM, and proteinase K to 200 pg/ml at 37 "C for 1 h. DNA was purified by standard methods (Sambrook et ul., 1989). 10 pg of purified genomic DNA prepared from DNase I-treated nuclei were digested with EcoRI, separated by 1% agarose gel electrophoresis, and transferred to the nylon transfer membrane using the capillary transfer method for further Southern blot analysis. The filter was prehybridized for 3 h at 42 "C in 50% formamide, 5 X SSC, 10 X polyvinylpyrrolidone, 1% SDS, and 100 pg/ml of sonicated herring sperm DNA. A 0.9-kb HindIII-BamHI fragment encompassing exon 1 of the human DBH gene ( Fig. 1) was purified, labeled by the random priming method to a specific activity of lo9 cpm/pg (Sambrook et al., 1989), and used as a probe. Hybridization was carried out for 1 week at -80 "C with intensifying screen.
Isolation of the 5'-Upstreum Sequence of the Human DBH Gene Two oligonucleotides were synthesized and used as primers in polymerase chain reaction as described (Saiki et al., 1988), 5'-GGCTGGGGTGAGCTCTCT-3' and 5"TCTACTTGCGGGAGA-GGA-3', correspond to locations from -9 to +9 and from -978 to -961 bp of the human DBH upstream region, respectively (Kobayashi et al., 1989). Human genomic DNA prepared from placenta was used as a template. The resulting 987-bp fragment was subcloned at the HincII site of pUC19 plasmid and confirmed by sequence analysis. This fragment DNA was then labeled by the random primer method and used as a probe to screen a human genomic library (Clontech). Restriction site mapping and Southern hybridization analyses demonstrated that one of four positive clones, XhuD-2, contained >4.3 kb of the DBH upstream sequence (data not shown). The insert DNA of AhuD-2 was utilized for constructing various CAT-fusion plasmids.

Plasmid Construction
Construction of 4.3CAT and Other CAT-fusion Plasmids Containing Shorter Upstream Sequences of the Human DBH Gene"978CAT plasmid was constructed by subcloning the 978-bp DNA fragment in pBLCAT3 plasmid (Luckow and Schutz, 1987). Then, a 4.2-kb HindlII-Hind111 fragment isolated from XhuD-2 phage DNA was substituted for the corresponding 870-bp HindIII-Hind111 fragment of 978CAT plasmid, resulting in 4.3CAT. Orientation of the 4.2-kb insert DNA in this plasmid was confirmed by restriction mapping and sequence analyses. 4.3CAT plasmid contains the upstream sequence of the human DBH gene ranging from -4.3 kb to +9 bp fused to the chloramphenicol acetyltransferase gene. A series of plasmids containing different lengths of upstream sequence were made utilizing the unique SphI site in pBLCAT3 and appropriate restriction sites residing in the upstream region of the human DBH gene (Figs. L4 and 2 A ) .
Construction of 231CAT, 202CAT, and 189CAT Plasmids-To construct plasmids containing upstream sequence between -262 and -175 bp, 978CAT plasmid DNA was digested with PstI and subsequently treated with Ba131 enzyme. The plasmid DNA was then digested again with SphI, rendered blunt by Klenow enzyme, and self-ligated. Candidate plasmids were isolated after restriction analysis and confirmed by nucleotide sequence determination.
Construction of CRE-debted Mutant Plasmids"978CAT plasmid was partially digested with AatlI and treated with T4 DNA polymerase, resulting in ACRE978CAT plasmid. Sequence analysis demonstrated that 14 bases between -189 and -176 bp, spanning the CRE, was deleted in this plasmid probably due to nonspecific exonuclease activity of T4 DNA polymerase. ACRE978CAT plasmid was digested with PstI and SphI, rendered blunt by Klenow enzyme, and self-ligated to make ACRE262CAT plasmid.
Construction of AhDAP604CAT-604CAT plasmid was digested with ApaI and PstI, made blunt-ended, and self-ligated to construct AhDAP604CAT plasmid.
Construction of hDAPCAT-1 and hDAPCAT-2-A 224-bp DNA fragment, ranging from -486 to -263 bp of the human DBH gene, was isolated from 486CAT plasmid by complete HindIII and partial ApaI digestion. This DNA fragment was placed in the 5'-position of the tk promoter in both orientations, using the HindIII site of pBLCAT2 plasmid (Luckow and Schutz, 1987) to produce hDAPCAT-1 and hDAPCAT-2. The orientation of the insert DNA was confirmed by restriction analysis.

Transient Transfection Experiments
For transfection analyses, construct plasmids were purified by CsCl gradient ultracentrifugation twice, pheno1:chloroform (1:l) extraction twice, and EtOH precipitation twice in the presence of 2.5 M ammonium acetate. An equimolar amount of each plasmid construct (0.7 pmol for SK-N-BE(2)C cells and 1.4 pmol for other cells) was introduced into cells by the calcium phosphate co-precipitation method (Gorman et al., 1982a) as described (Kim et aL, 1993b). All transfection experiments were repeated two to seven times with similar results utilizing construct plasmids that were independently prepared at least twice. Cells in a 60-mm dish received a total of 5 pg of DNA for SK-N-BE(2)C cell line and 10 pg of DNA for other cell lines. To control for differences in transfection efficiency from dish to dish, 1 or 2 pg of pRSV-@gal plasmid containing the @-galactosidase gene under control of the RSV promoter/enhancer (Edlund et at., 1985) was included in each transfection and used for normalization. PUC19 plasmid was used as a carrier to make up the total amount of DNA. To compare the promoter activities across different cell lines, RSV-CAT plasmids of the equal molar concentration to the CAT fusion construct were introduced into cells in parallel dishes to serve as positive controls (Gorman et al., 1982b;Kraner et al., 1992). 40 h after transfection, cells were extracted and CAT, and @-galactosidase activities were determined as described (Gorman et al., 1982a;An et al., 1982).

DNase I Hypersensitive Sites Reside in the 5"Flanking
Region of the DBH Gene in SK-N-BE(2)C Cells, but Not in HeLa Cells-We chose cell lines suitable to study of DBH gene regulation after screening mRNA from several different lines by Northern blot hybridization, using a human DBH cDNA probe. The human neuroblastoma cell line, SK-N-BE(2)C, exhibited robust intrinsic expression of DBH mRNA, whereas HeLa cells showed no detectable signal (Fig. 1B). When the same blot was hybridized with a cDNA probe for the a-tubulin gene, both lanes produced equivalent signals, indicating similar amounts of mRNA were present in each lane.
We then compared the chromatin structure of the DBH gene promoter in nuclei isolated from SK-N-BE(2)C and HeLa cell lines by treating nuclei with increasing amounts of DNase I prior to DNA isolation. The purified DNA was digested with EcoRI, fractionated by agarose gel electropho- Putative cis-acting motifs such as the CRE, AP1, and TATA box are indicated. Experimental data in this study demonstrate that the CRE of the human DBH gene is a critical genetic element. In contrast, the functionality of the corresponding CRE-like motif (denoted "CRET' in the figure) in the rat sequence is not yet clear. Dashed lines between two sequences show the sequence identity between human and rat DBH genes.  (HSSs). Two prominent DNase I HSSs appeared at about 1.35 and 1.5 kb upstream of the reference EcoRI site in the nuclei isolated from SK-N-BE(2)C cells (Fig. 1, A and C).
In addition, there was another DNase I HSS with weaker signal at 2.5 kb upstream of the EcoRI site. None of these DNase I HSSs were detected in nuclei from HeLa cells, suggesting that the proximal promoter region of the human DBH gene maintains an open chromatin structure accessible to DNase I only in DBH-expressing cells. We hypothesized I I I I I I I I I I I I I I  I I I I I I  I I I I I I that the three DNase I HSSs represented active promoter elements involved in cell type-specific expression of the human DBH gene. To test this hypothesis, we performed mutational analyses using SK-N-BE(2)C and HeLa cell lines, as models exhibiting cell type-specific expression or repression of the DBH gene, as follows.

4.3and 2.6-kb Upstream Sequence of the Human DBH Gene Direct the Expression of the Reporter Gene in a Cell
Type-specific Manner-After isolating and partially characterizing a genomic clone containing >4.3-kb upstream region (see "Experimental Procedures"), we used HindIII and SacII sites (Fig. lA) to fuse the upstream 4.3-and 2.6-kb sequence 5' to the reporter gene encoding CAT in the promoterless plasmid, pBLCAT3 (Luckow and Schutz, 1987). Fusion was made at position +9 in the untranslated sequence of the human DBH mRNA (Fig. 2A). The resultant constructs, 4.3CAT and 2.6CAT, also contain t intron and polyadenylation signal in the 3"nontranslated region which originated from SV40 (Luckow and Schutz, 1987). When introduced into cell lines which differed with respect to DBH gene expression, both 4.3CAT and 2.6CAT exhibited cell-type specificity in transient expression assays (Fig. 2B). DBH-expressing human neuroblastoma SK-N-BE(2)C transiently expressed substantial CAT activity, whereas DBH-nonexpressing cell lines HeLa and mouse Ltk-showed only minimal CAT activity, not greater than that driven by pBLCAT3. In SK-N-BE(2)C cells, CAT activity driven by pBLCAT3 was less than 3% of that by 4.3CAT or 2.6CAT. Another DBH-nonexpressing cell line, rat C6 glioma, also expressed CAT activity no greater than that by pBLCAT3 (data not shown). These data, in conjunction with the observation that all three DNase I HSSs identified in the active DBH gene were localized within 2.6 kb upstream of the transcriptional initiation site suggested that this region contained the information necessary and sufficient for cell type-specific expression (Figs. 1C and 2B).

Both a Positive Genetic Element and a Silencer Region
Contribute to Cell Type-specific Expression of DBH-We performed deletional analyses of this 2.6-kb sequence using the same transient expression assay to identify regulatory elements required for cell-specific expression (Fig. 3A). In DBHexpressing SK-N-BE(2)C cells, deletion of the 5'-most region up to -1556 bp did not alter promoter activity (Fig. 3B). Further deletion to -978 and -604 bp progressively increased the CAT activity by 40 and 110%, respectively, compared with longer fusion constructs. Notably, this region coincided with one of the DNase I HSS identified in the same cell line (Fig.   IC).
Deletions from -604 to -486, and to -262 bp, did not produce any significant change in CAT activity. However, further deletion of 87 base pairs up to -175 bp or to -114 bp virtually eliminated CAT activity, rendering it transcriptionally as inert as pBLCAT3 (Fig. 3B). These data suggest that essential positive element(s) reside in the upstream region between -262 to -175 bp, a region that contains one of the strong DNase I HSS identified above (Fig. IC). To define further the sequence motif of the positive genetic element, we produced smaller deletions within this region using Ba131 endonuclease digestion (Fig. 4A). All three deletional constructs between -262 and -175 bp, i.e. 231CAT, 202CAT, and 189CAT, retained transcriptional activity similar to that of 262CAT, indicating that an essential positive element resides in the region between -189 and -175bp (Fig. 4B). This area contains a potential CAMP response element, -181 TGACGTCC -174 (Fig. ID), with a single base deviation from the consensus CRE sequence (TGACGTCA; Montminy et al., 1986). We thus performed site-directed mutagenesis of this putative CRE in the intact upstream sequence to test if (i) the CRE is an essential genetic element for the expression of the human DBH gene and (ii) this CRE motif is functional in response to elevated levels of CAMP. ACRE978CAT, in which the 14 base pairs between -189 and -176 bp are deleted in the context of 978 bp of intact upstream sequence, reduced expression of 978CAT plasmid by >95% (Fig. 4C). Moreover, the response of ACRE978CAT plasmid to treatment with dibutyryl CAMP was decreased compared to that of 978CAT ( Fig. 4C). We concluded that the CRE residing between -189 and -174 bp is a functional CRE. These data define the CRE, residing at -181 to -174 bp upstream of the transcription initiation site, as an essential positive element in the expression and regulation of the human DBH gene.
We also performed parallel transient expression assays in HeLa cells in an attempt to identify any regulatory element involved in the cell type-specific transcriptional suppression of the DBH gene in this cell line. None of the fusion constructs containing the promoter sequences deleted between -4.3 kb and -486 bp showed CAT activity higher than that directed by pBLCAT3 (Fig. 3C), indicating that the 486-bp upstream sequence contains sufficient information to suppress expression in HeLa cells. Further deletion to -262 bp, however, resulted in derepression of the CAT activity approximately by 10-fold compared with that by 486 bp upstream (Fig. 3C). Thus, the upstream region between -486 and -262 bp appears to contain one or more negative elements that are required for effective suppression of the DBH gene in HeLa cell line. The derepressed promoter activity of 262CAT plasmid in HeLa cells relative to the RSV promoter was approximately 10% of 262CAT activity in SK-N-BE(2)C cells (Fig. 3, B and   C ) . To further characterize the negative region, we deleted the sequence between -490 and -262 bp in the 604CAT plasmid (Fig. 5A) and compared the promoter activity of the resulting construct in both cell lines (Fig. 5B). As expected, deletion of this region in the context of 604-bp upstream sequence did not alter promoter activity in SK-N-BE(2)C cell line. In sharp contrast, however, this deletion caused about 10-fold derepression of CAT activity in HeLa cells (Fig. 5B) as well as in mouse Ltk-cells (data not shown), confirming that this region contains a cell type-specific repressing sequence. Interestingly, the fusion constructs lost transcriptional activity upon deletion to -175 bp or to -114 bp, indicating that the upstream region between -262 and -114 bp contains positive element(s) required for the derepressed

FIG. 4. Identification of the CRE region as an essential regulatory element.
A, schematic diagram illustrating detailed deletion constructs and a CRE-deleted construct. B, location of an essential positive element by detailed deletion analysis. These constructs were prepared using Ba131 exonuclease as described under "Experimental Procedures." C, site-directed mutagenesis of the CRE region. Fourteen bases spanning the CRE were in ACRE978CAT. Sixteen hours prior to harvest, cells were treated with dibutyryl CAMP to assess the functionality of this putative CRE motif (as illustrated by + in parentheses next to each plasmid construct). ACRE978CAT construct lost not only the basal expression but most of its responsiveness to treatment of dibutyryl CAMP. This experiment was performed two times in triplicate with similar results. transcriptional activity of the 262CAT plasmid in HeLa cells. Based on the essential functional role of the CRE for DBH expression in the SK-N-BE(2)C cell line, we surmised that the CRE plays an important role in the derepressed ectopic transcriptional activity of the 262CAT plasmid in the HeLa cell line.
To test this possibility, we constructed ACRE262CAT, another 262CAT plasmid in which the CRE is deleted (Fig. 5 A ) . When introduced into the HeLa cells, this plasmid, unlike 262CAT, was as inert as pBLCAT3 (Fig.  5B). These data indicate that the CRE of the DBH gene is important both for the cell-specific expression in neural cells and for the ectopic expression in non-neural cells which occurs upon deletion of the negative region.

The Silencer Region, Residing between -486 and -263 bp
ative result is shown here. pRSV-@gal was included in transfection as an internal control to normalize variation of transfection efficiencies between different DNA precipitates. Cell extracts corresponding to the same @-galactosidase activity were used in the CAT assay. Since pRSV-CAT had very high CAT activity, a much smaller amount, 1/ 10 compared with other fusion plasmid, was used for measurement of the CAT activity.

Neuron-specific Expression of the Human DBH Gene
FIG. 5. The upstream sequence of the human DBH gene contains a silencer region required for cell typespecific expression. A, the schematic diagram showing the deletions of the CRE or hDAP. Thick bars represent upstream regions of the DBH gene, and thin bars represent deleted regions. B, transient expression assays demonstrate the functionality of the CRE or hDAP. This experiment has been performed at least two times in triplicated with the same pattern. Representative autoradiograms are shown. C, sequence comparison of the hDAP region reveals two nucleotide patches sharing significant homologies with previously characterized neuron-specific cis-acting elements. Sequence similarities are indicated by dashed lines to 1) Drosophila Dopa decarboxylase element I on the opposite strand (Scholnick et al., 1986;Bray et ai., 1988) ((I)) and silencer element of rat SCGlO (Mori et al., 1992) and of type I1 sodium channel (Kraner et ai., 1992) ((2)). Nucleotide sequences of element I and the silencer elements are underlined. In addition, a putative core sequence of the silencer element is shown below.

Upstream of Human DBH Gene, Contains Two Sequence Patches Highly Homologous to Previously Identified Neuronspecific Sequence
Motifs-Our analyses of the 4.3-kb upstream region of the human DBH gene revealed a 223-bp fragment which conferred a repressing effect to the native DBH promoter in a cell type-specific manner. We determined whether this region of the human DBH gene contained sequences similar to previously described neuron-specific sequence elements by comparing the nucleotide sequences. These searches revealed two sequence patches sharing significant homologies with such sequence motifs previously identified in other neuronal genes (Fig. 5C). First, the sequence between -456 and -444 bp shared 77% identity with a cis-acting element initially described in the Drosophila Dopa decarboxylase gene. This element, designated element I, was protected in DNase I footprinting by embryonic nuclear extract and was necessary but not sufficient for neuron-specific expression in the Drosophila central nervous system (Scholnick et al., 1986;Bray et al., 1988). Another sequence was found at -400 and -384 bp, which is homologous to the silencer sequence motifs independently identified in two different mammalian neuron-specific genes, i.e. rat type I1 sodium channel and SCGlO genes (Kraner et al., 1992;Mori et al., 1992; Fig. 5C). This silencer sequence predominantly determines the neuron-specific expression of these genes via a presumably identical silencer-binding protein(s). The 21 nucleotide sequences, which retained silencer consensus: CAGCTCCTCGGAC A A A G function by themselves in type I1 sodium channel and SCG 10 genes, shared 81% sequence identity with each other (Kraner et al., Mori et al., 1992). The sequence found in the human DBH gene shows less homology with these motifs: 48% with SCGlO and 57% with type I1 sodium channel gene sequences (Fig. 5C). However, a subdomain of 13 nucleotides (-400 to -388 bp) within this putative silencer region of the human DBH gene shared 77% identity to the corresponding fragments from the SCGlO gene or from the type I1 sodium channel gene.

The Negative Element Region Exerts the Silencing Effect on a Heterologous Promoter in a Cell Type-specific Manner-To
assess whether the foregoing putative silencer region can confer negative regulation upon a heterologous promoter independent of orientation, as would be expected if it functions analogously to the silencer present in SCGlO and type I1 sodium channel genes (Mori et al., 1992;Kraner et al., 19921, we placed this 223-bp fragment of putative silencer region in front of the thymidine kinase (tk) promoter using pBLCAT2 in either orientation (Fig. 6A, Luckow and Schutz, 1987). When introduced into SK-N-BE(2)C cells, the properly oriented fragment did not inhibit transcriptional activity of the tk promoter; however, in the opposite orientation, this fragment inhibited tk promoter-supported CAT activity by 50% compared with the wild-type tk promoter. In contrast, in HeLa cells, this sequence suppressed tk-supported CAT activity to less than 50% of control in either orientation (Fig. 6 B ) . Thus, the region between -486 and -263 bp of the human DBH gene appears to contain an element that acts as a silencer in non-neural cells, but not in neural cells. The suppression effect of this sequence was more effective when linked to its native promoter than when attached to a heterologous t k promoter (compare Fig. 3, B and C, with Fig. 6B).

DISCUSSION
The transcriptional regulation of many eukaryotic genes has been intensively studied during the last two decades, and tissue-specific enhancers for different cell types, e.g. liver (Cereghini et al., 1987), lymphoid cells (Grosschedl and Baltimore, 1985), pancreas (Cockell et d., 1989), and anterior pituitary (Lefevre et al., 1987) are well established. Nevertheless, the molecular mechanisms underlying neuron-specific ' gene expression are poorly understood. In the present study, we examined the cell type-specific expression of the human DBH gene by comparing its promoter function in a human noradrenergic cell line to that in non-neuronal non-catecholaminergic cell lines.
As a first step in identifying important genetic elements involved in the cell type-specific DBH expression, we tested whether the 5"flanking sequence of the human DBH gene contained cell type-specific nuclease-sensitive sites. DNase I HSS mapping analysis revealed three sites in the 5'-proximal sequence in SK-N-BE(2)C cells but not in HeLa cells (Fig.  1C). The two proximal sites of these DNase I HSS, located at about -25 and -180 bp, respectively, corresponded to the locations of the TATA box and CRE. Since active promoter elements often exist as nuclease-sensitive structures in chromatin, our data indirectly suggest that the proximal area of the human DBH gene contains several cis-acting elements which may be important for cell type-specific expression. Thus, we next sought to address whether the upstream sequence of the human DBH gene can direct expression in a cell-specific manner. When transiently expressed in different cell lines, both the 4.3-and 2.6-kb upstream sequences directed the expression of the reporter gene in a tight cell typespecific manner (Fig. 2), consistent with previous results obtqined in transgenic mice (Mercer et al., 1991), demonstrating that the 5.8-kb upstream sequence of the human DBH gene can direct tissue-specific expression. Our data further suggest that the sufficient information for cell type-specific expression may be confined to the 2.6-kb upstream region.
The 2.6-kb upstream sequence was further examined using extensive deletional and mutational analyses. Transient expression assays in SK-N-BE(2)C cells revealed that the 5'flanking sequence of the DBH gene may contain several regulatory cis-acting elements. First, upon deletion of the sequence from -1556 bp to -978, and to -604 bp nucleotide position, CAT activity progressively increased about 2-fold ( Fig. 3B), suggesting that this region might contain negatively acting element(s) active in the SK-N-BE(2)C cell line. Second, and more importantly, the nucleotide sequences between -189 to -176 bp appeared to contain a critical positive element supporting transcriptional promoter activity of the human DBH upstream sequence (Fig. 4). This region contains a consensus CRE with a single base deviation (A to C at -174 bp position; Fig. 1D). Deletion of the 14 bp region (from -189 to -176), including the CRE, in the context of 978 bp of intact upstream sequence, not only caused a severe decrement in the basal promoter activity but also reduced its responsiveness to the elevated cAMP (Fig. 4C). In contrast, the CAT activity driven by the wild-type promoter, 978CAT, increased about 2.5-fold upon treatment with dibutyryl CAMP. These data strongly suggest that the CRE, residing at -181 to -174 bp, plays an important dual role in the DBH promoter: (i) it supports the uninduced, basal expression of DBH and (ii) mediates, at least in part, CAMP-inducible expression in noradrenergic cells.
While this paper was being prepared, two laboratories reported the results of analyzing the 5'-promoter region of the DBH gene. First, Shaskus et al. (1992) examined the 395-bp upstream sequence of the rat DBH gene and showed that a 30-bp region between -180 and -151 bp might contain a transcriptional enhancer which had a dual function for cell type specificity and second messenger responsiveness. Second, Lamouroux et al. (1993) analyzed the 1247-bp upstream sequence of the human DBH gene and showed that the upstream region between -267 and -115 bp was crucial for cell-specific and cAMP regulation, suggesting the importance of a nearconsensus CRE located in this region. These data generally agree well with ours. On close inspection of the nucleotide sequences in these overlapping promoter regions; however, it became evident that the human and rat genes contain different compositions of nucleotide sequences in putative cisacting elements in this area (Fig. 1D). The rat gene contains a sequence motif, TGCGTCA, which is a canonical AP1 motif (Shaskus et al., 1992). The corresponding position of the human gene has a similar motif, TGTGTCA, with a single base variation (Fig. 1D). As both 175CAT and ACRE978CAT retain this sequence and do not show CAT activity any greater than pBLCAT3 (Fig. 4), it appears that this motif, TGTGTCA, in and of itself, does not exert transcriptional activity. We suggest here that the CRE with a single base deviation, which in the human gene is just proximal to this AP1-like motif, is functionally essential for transcriptional activity. The corresponding rat sequence contains a CRE-like motif which varies by 2 bases from the consensus CRE motif (Fig. 1D). In contrast to the functional importance of the CRE in the human gene, transient expression analysis by Shaskus et al. (1992) suggests that the CRE-like motif is functionally silent in the rat DBH gene. Clear delineation of the functional importance of these sequence motifs in regulation of the human and the rat DBH genes awaits further investigation.
Both of the rat and human T H genes contain the consensus CRE in the proximal region (Kobayashi et al., 1988;Cambi et aL, 1989). Recent data from our laboratory suggest that the CRE plays an important dual role in both basal and CAMPinducible expression in the T H gene (Kim et al., 1993b). Thus, it appears that two enzyme genes of the catecholamine pathway, T H and DBH, adopt similar molecular strategies by utilizing the same cis-acting element (CRE) for their expression and regulation. The CRE has been inferred to be important for the basal expression of several other genes (Deutsch et al., 1987, Delegeane et al., 1987, Quinn et al., 1988, Andrisani et ab, 1989. It is plausible that CAMP-dependent protein kinase signaling pathway might be directly involved in regulating the expression of these genes. Indeed, we recently demonstrated that the CAMP-dependent protein kinase system regulates both the basal and CAMP-inducible expression for the TH gene (Kim et al., 1993a) as well as the DBH gene.' Recent studes have identified a number of different genes where silencers play important roles in transcriptional regulation (Renkawitz, 1990). Our transient expression analyses indicate that an upstream region of the human DBH gene might play an important role for the cell type-specific expression of this gene. The promoter activity, however, reached only lo%, upon deletion of the putative silencer region, in HeLa cells compared with SK-N-BE(2)C cells (Fig. 3). This contrasts with the SCGlO and type I1 sodium channel genes where the proximal sequence devoid of the silencer displayed similar promoter strength in neuronal and non-neuronal cells (Mori et al., 1992;Kraner et al., 1992). These data suggest that the proximal 262 bp of the DBH promoter contributes to cell type-specific expression as indicated by other investigators (Shaskus et al., 1992, Lamouroux et al., 1993. The CRE appears to be an essential positive element which is required for transcriptional activity by the upstream sequence of the human DBH gene. Consequently, the corresponding CRE-binding proteins appear to play a central role in transcriptional activation of the human DBH gene in noradrenergic and adrenergic cells. In DBH-nonexpressing cells, the action of such transcription factors, if present, should be neutralized by the silencer residing farther upstream, thus rendering the DBH gene transcriptionally silent. Our observation, that the 262CAT plasmid which lacks the silencer region loses the derepressed expression in HeLa cells upon deletion of the CRE (Fig. 5), supports that contention. Thus, in the absence of the silencer region, derepressed expression by the upstream sequence of the human DBH gene requires a functional CRE. These data suggest that the interplay between the CRE-binding protein(s) and silencer-binding protein might represent, at least in part, an underlying mechanism for tissue-specific expression of the human DBH gene.
A clearer understanding of the foregoing mechanisms will be possible once the proteins that bind to the CRE and the silencer motif of the human DBH gene have been identified, cloned, and characterized. That would also make it possible K. S. Kim, unpublished data. to test whether the same silencer-binding protein or a family of similar proteins is involved in determining both neuronspecific expression and neuronal subtype-specific distribution of a variety of genes.