Sequences Required for Induction of Neurotensin Receptor Gene Expression during Neuronal Differentiation of N1E-115 Neuroblastoma Cells*

The promoter region of the mouse high affinity neurotensin receptor (Ntr-1) gene was characterized, and sequences required for expression in neuroblastoma cell lines that express high affinity NT-binding sites were characterized. Me2SO-induced neuronal differentiation of N1E-115 neuroblastoma cells increased both the expression of the endogenousNtr-1 gene and reporter genes driven by NTR-1 promoter sequences by 3–4-fold. Deletion analysis revealed that an 83-base pair promoter region containing the transcriptional start site is required for Me2SO activation. Detailed mutational analysis of this region revealed that a CACCC box and the central region of a large GC-rich palindrome are the crucial cis-regulatory elements required for Me2SO induction. The CACCC box is bound by at least one factor that is induced upon Me2SO treatment of N1E-115 cells. The Me2SO effect was found to be both selective and cell type-restricted. Basal expression in the neuroblastoma cell lines required a distinct set of sequences, including an Sp1-like sequence, and a sequence resembling an NGFI-A-binding site; however, a more distal 5′ sequence was found to repress basal activity in N1E-115 cells. These results provide evidence that Ntr-1 gene regulation involves both positive and negative regulatory elements located in the 5′-flanking region and thatNtr-1 gene activation involves the coordinate activation or induction of several factors, including a CACCC box binding complex.

Neurotensin (NT) 1 is a 13-amino acid peptide (1) that is expressed in a complex pattern in the limbic regions of the brain and in the gastrointestinal tract (2). NT and the related peptide neuromedin N are generated from a common precursor protein and are thought to have overlapping signaling functions (3,4). There is considerable anatomical and functional evidence indicating that NT functions as a neuromodulator in the dopamine (DA) pathways in the central nervous system. The majority of midbrain DA neurons express the cloned high affinity NT receptor (NTR-1), and there is evidence that at least some DA neurons are directly contacted by NT-positive axons (5,6). NT appears to excite midbrain DA neurons both directly (7,8) and through inhibition of DA D 2 autoreceptors (9) resulting in locomotor activation (10,11). However, NT also attenuates D-amphetamine locomotor activation after intracerebroventricular administration or after direct application in the ventral striatum possibly through the inhibition of postsynaptic D 2 signaling (12)(13)(14). These results suggest that the expression of NTR-1 in midbrain DA neurons is important for appropriate regulation of DA-mediated behaviors.
NTR-1 was cloned using an expression assay in frog oocytes, and sequence analysis revealed that it is a member of the G protein-coupled receptor superfamily (15). A lower affinity levocobastine-sensitive receptor (NTR-2) was subsequently cloned by low stringency hybridization using an NTR-1 cDNA probe (16,17). Recent evidence suggests that NT does not stimulate signaling through NTR-2, suggesting that NT acts mainly through NTR-1 (18). NTR-1 is expressed at high levels in midbrain DA neurons, and dopamine regulates NTR expression in corticolimbic structures in the rat brain (19 -21). The ability of DA to modulate NTR expression suggests that changes in DA signaling result in plastic changes in NT signaling. This hypothesis is further supported by the observations that the indirect DA agonists cocaine and methamphetamine and D 2 antagonists stimulate NT gene expression in the dorsal and ventral striatum (22,23). Long term NTR-1 blockade and continuous infusion of NT also result in alterations in NTR-1 expression (24,25). These results collectively indicate that DA, NT, and perhaps other signals can result in plastic changes in NTR-1 expression; however, the underlying mechanisms controlling NTR-1 expression are poorly understood.
Tissue culture cell lines that express NTR-1 in a regulated or constitutive manner potentially provide model systems for understanding the mechanisms controlling Ntr-1 gene expression. The N1E-115 neuroblastoma cell line was isolated as part of a screen for catecholamine-producing neuronal cell lines from a mouse brain neuroblastoma (26), and Me 2 SO-induced neuronal differentiation of these cells (27) is accompanied by increased expression of high affinity NT-binding sites (28). The transition of these cells to a post-mitotic stationary phase also results in the induction of NTR expression (29). A neuroblastoma ϫ glioma cell line has also been described that expresses high constitutive levels of high affinity NT-binding sites (30). The N1E-115 cells are a particularly attractive model to probe Ntr-1 gene regulation since the Ntr-1 gene is expressed in catecholaminergic neurons in vivo (31). They should also be useful for identifying signaling mechanisms that activate the Ntr-1 gene during neuronal differentiation (28).
To investigate the pathways controlling Ntr-1 gene expression in these and other cell types, we have cloned the mouse Ntr-1 gene and sequenced the promoter region. Detailed mutational analysis of the Ntr-1 promoter has revealed sequence elements that are crucial for Me 2 SO induction and basal expression in N1E-115 cells that are conserved in the rat and human promoters. A CACCC sequence appears to be the most critical sequence element for Me 2 SO responsiveness, and gel shift and DNase I footprinting experiments indicate that an Me 2 SO-inducible complex binds to this site. Several sequence elements contribute to basal expression, including an Sp1related site and a sequence that is similar to the initiator element first identified in the terminal deoxynucleotidyltransferase gene (32). We also present evidence that a transcriptional silencer controls the activity of this positive regulatory region. These results provide evidence that the Ntr-1 gene is transcriptionally activated during Me 2 SO-induced neuronal differentiation of N1E-115 cells most likely through a mechanism involving the induction of a complex that binds to a sequence that includes a CACCC motif.

EXPERIMENTAL PROCEDURES
Isolation of Mouse Ntr-1 Genomic Clones-A mouse genomic library was constructed by ligating partially digested high molecular weight DNA isolated from D3 embryonic stem cells (15-20-

kb fragments) into
BamHI-digested EMBL4 using standard methods (33). The library was screened using a 32 P-labeled SmaI fragment of the rat NTR-1 cDNA (15) that contains the 5Ј 1.3 kb of the cloned sequences and standard filter lift procedures (33). Positives were plaque-purified and used to prepare phage DNA. Two positives were further characterized by restriction mapping (mNTR2-2 and mNTR10 -1). Promoter fragments from mNTR2-2 were subcloned into pGEM4 (Promega), and the promoter region was sequenced using the chain terminator method.
Cell Culture-N1E-115 cells were passaged in Dulbecco's modified Eagle's medium containing 4.5 g of glucose/liter and supplemented with 10% fetal bovine serum (Sigma) and 2 mM L-glutamine. NG108 cells were passaged in the medium described above for N1E-115 cells except supplemented with hypoxanthine, aminopterin, and thymidine (Life Technologies, Inc.) as described (30). Mouse embryonic fibroblast (MEF) cells were passaged in Dulbecco's modified Eagle's medium containing 1.0 g of glucose/liter and supplemented with 10% fetal bovine serum and 2 mM L-glutamine.
Transfections-N1E-115 and MEF cells were subcultured from confluent dishes by diluting the cells 1:4 with fresh medium 3 days prior to transfection. All lines were subcultured the day before transfection at a density that resulted in 1 ϫ 10 6 cells per 10-cm dish at the time of transfection. CsCl-purified plasmid DNA was transfected by calcium phosphate precipitation using 10-cm dishes. Briefly, for each reporter plasmid, two 10-cm dishes were fed with 8 ml of fresh medium just prior to transfection, and 25 g of plasmid DNA as a calcium phosphate precipitate were added to each dish. DNA precipitates were prepared by diluting 50 g of plasmid DNA (luciferase reporter and pPGK␤-gal standardization plasmids) into 700 l of NTE (150 mM NaCl, 10 mM Tris-HCl, pH 7.4, 1 mM EDTA), 100 l of 2 M CaCl 2 , and 800 l of transfection buffer (50 mM HEPES, pH 7.1, 180 mM NaCl, 2 mM NaPO 4 ), and the mixtures were incubated at room temperature for 20 -30 min. The calcium phosphate precipitate (800 l) was added dropwise to the cells, and after 16 h the medium was replaced, and the cells from one 10-cm dish were resuspended by trituration and plated into two 60-mm dishes. The cells were allowed to adhere to the dish for 1 h; 1.5% Me 2 SO was added to one dish of each pair, and the cells were incubated for an additional 72 h. Cells were incubated for 72 h, and cell extracts were prepared after washing the cells with ice-cold phosphatebuffered saline by lysis in buffer containing 1% Triton X-100 as described (34). Luciferase and ␤-galactosidase activities were determined as described previously (35).
Luciferase Reporter Constructs-The initial NTR luciferase reporter construct was created by ligating a mouse Ntr-1 gene 1.2-kb BamHI fragment, containing sequences Ϫ1425 to Ϫ185 relative to the NTR methionine initiator codon, into a BglII site just upstream of the luciferase gene in the pXP2 reporter plasmid (36). Convenient restriction sites (SmaI and SacI) and exonuclease III digestion were used to create a series of 5Ј deletion mutants in pXP-2. Additional 5Ј and 3Ј deletion constructs were created by PCR using appropriate primers. The PCR primers contained 17 nucleotide regions that were complementary to deletion end points and resulted in the introduction of a BamHI site at the 5Ј end and a SacI site at the 3Ј end. The PCR fragments were digested with BamHI and SacI and cloned into the SacI 5Ј deletion construct described above digested with BamHI and SacI. This minimal promoter fragment was initially selected based on the positions of the transcriptional start sites that had been determined in rat and human (37,38); however, the start point in the mouse actually lies 5Ј to the SacI site used to create the deletion clones. Clustered point mutations were created using the PCR overlap extension protocol (39). Briefly, two overlapping PCR fragments were generated using either a 5Ј primer that ends at the promoter SmaI site and introduces a BamHI site and a 3Ј primer containing the clustered point mutations flanked by 10 and 6 nucleotide stretches of complementary sequence on the 3Ј and 5Ј sides, respectively, or a 5Ј primer containing clustered point mutations as described above and a 3Ј primer with the promoter SacI site as an end point. The two fragments were isolated by electrophoresis on a low melting temperature gel; the gel slices were melted, mixed, and a third PCR reaction was performed using the outer 5Ј and 3Ј primers described above. The resulting fragment was digested with BamHI and SacI, gel-isolated, and ligated into the SacI deletion clone in pXP-2 described above digested with BamHI and SacI. The promoter regions of the mutant constructs were completely sequenced to verify that only the intended changes had been introduced.
RT-PCR-Mouse brain poly(A) ϩ RNA was treated with DNase I to remove trace amounts of contaminating genomic DNA and RT-PCR reactions using avian myeloblastosis virus reverse transcriptase (Promega) and Taq polymerase (Roche Molecular Biochemicals) using the conditions specified by the manufacturers, except that RT reactions were performed at 50°C and betaine was added to the PCR reactions. The reverse transcription reaction was initiated after heat inactivation of the DNase I (70°C for 5 min) by the addition of 5 pmol of a genespecific primer (PEX-411c 5Ј-TGGCGCAGAGAGGAGCGCACGCACG-GCTGCCCAGC-3Ј) complementary to nucleotides Ϫ411 to Ϫ376 of the mouse Ntr-1 gene and 25 units of avian myeloblastosis virus reverse transcriptase (Roche Molecular Biochemicals). PCR reactions were performed using either mouse brain cDNA or embryonic stem cell genomic DNA templates mixed with 10 pmol of the PEX-411c primer and 10 pmol of one of three different 5Ј primers (Ϫ513, 5Ј-GTGGAAGCGCGA-GGAGCCCG-3Ј; Ϫ548, 5Ј-TTTTGGATCCACTGCTGGGCGCGCC-3Ј; Ϫ566, 5Ј-CTCCAACACCCACCCTCCTCCACTG-3Ј, the numbers correspond to the position of the 5Ј end of the oligonucleotide on the mouse NTR-1 promoter). PCR reactions were supplemented with 2.5 M betaine (Sigma) due to the high GC content of the promoter region, and 30 amplification cycles (94°C, 15 s; 62°C, 1 min; 72°C, 1 min; the extension time was increased to 4 min on the last cycle) were performed after an initial 1-min denaturation step at 94°C (41).
DNase I Footprinting and Methylation Interference Assays-32 P-Endlabeled probes spanning the promoter region were prepared by diges-tion of a 1.2-kb BamHI gene fragment subcloned into pGEM4 with either EcoRI or BamHI followed by Klenow fill-in with [ 32 P]dATP. The plasmids were subsequently digested with either BamHI or EcoRI, respectively; the resulting fragments were separated on a 1% agarose gel, and the 496-bp labeled EcoRI/BamHI fragments labeled at either the EcoRI or BamHI sites were recovered by centrifugation through a Gen Elute column (Supelco) and ethanol precipitation. Nuclear extracts were prepared from N1E-115 cells that had either been treated with 1.5% Me 2 SO for 48 h or grown under control conditions as described previously (42) with minor modifications (43) using approximately 20 dishes of cells grown to confluency for each preparation. Protein concentrations were determined using the Bradford method. DNase I footprint reactions were performed as described (35) by mixing 20,000 cpm probe, 160 g of nuclear extract, and treatment with various amounts of DNase I (DNase I, pure, ribonuclease-free, Worthington). Reactions were also performed without the addition of nuclear extract to identify protected regions. The DNase I-treated reactions were phenol/chloroform-extracted, ethanol-precipitated, analyzed on sequencing gels, and visualized by autoradiography with an intensifying screen. Chemical sequencing reactions (G, G ϩ A) were also run so that the footprinted regions could be precisely aligned with the DNA sequence.
Methylation interference reactions were performed as described (44) using a 32 P-labeled probe spanning nucleotides Ϫ589 to Ϫ507 generated by PCR and labeled by Klenow fill-in of a BamHI site at the 5Ј end. The probe (10 l, 5 ϫ 10 5 cpm) was partially methylated by incubation in 200 l of dimethyl sulfate reaction buffer (sodium cacodylate, 1 mM EDTA, pH 8.0) containing 1 l of dimethyl sulfate for 5 min at room temperature followed by the addition of dimethyl sulfate stop buffer (1.5 M sodium acetate, 1 M ␤-mercaptoethanol, pH 7.0) and ethanol precipitation. The methylated probe was then reacted with 50 g of nuclear extract from N1E-115 cells that had been induced with 1.5% Me 2 SO for 72 h, and bound complexes were separated from free probe by electrophoresis on a non-denaturing gel. The complexes were visualized by autoradiography, and the region of the gel containing complexes 1 and 2 and unbound probe were excised, the DNA was electroeluted, phenol/ chloroform-extracted, and recovered by ethanol precipitation. The DNA was resuspended in 1 M piperidine, incubated at 95°C for 30 min, lyophilized several times, and analyzed on a sequencing gel.
Gel Shift Analysis-N1E-115 cells were propagated as described for transfection analysis and treated with 1.5% Me 2 SO or grown under control conditions in 10-cm dishes. The cells were washed twice with ice-cold phosphate-buffered saline, resuspended in 400 l of buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 2 mM aprotinin, 2 mM pepstatin, and 2 mM leupeptin), and incubated on ice for 15 min. The cells were lysed by the addition of 25 l of 10% Nonidet P-40 and vigorous vortexing for 10 s followed by centrifugation for 30 s. The supernatant was discarded; the pellet was washed once with buffer A on ice, centrifuged, and resuspended in 30 l of buffer C (20 mM HEPES, 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 2 mM aprotinin, 2 mM pepstatin, and 2 mM leupeptin) on ice by vortexing for 10 s every 5 min for 30 min. Nuclear extracts were stored at Ϫ80°C. Binding reactions were performed by mixing (10 g, 1-2 l) N1E-115 nuclear extract in binding buffer (12.5 mM HEPES, pH 7.9, 100 mM KCl, 10% glycerol, 0.1 mM EDTA, pH 8.0, 1.5 mM dithiothreitol) supplemented with 3 g of poly(dI-dC), 1 g of acetylated bovine serum albumin, and in some cases unlabeled competitor DNA (16-l final volume). The reactions were incubated on ice for 10 min, and 1 ϫ 10 5 cpm probe was added followed by incubation for 15-20 min at room temperature. The probe was the same as that used for methylation interference. Binding reactions were analyzed on low ionic strength Tris acetate acrylamide gels as described (44).

Cloning and Characterization of the Mouse Ntr-1 Gene-
Mouse NTR-1 clones were isolated by screening a genomic library constructed from D3 mouse embryonic stem cell DNA with a 32 P-labeled 1.3-kb SmaI 5Ј fragment of NTR-1 cDNA (15), and two were further analyzed (2-2 and 10-1) by restriction mapping and Southern blotting. Exon one was localized to an ϳ7-kb HindIII fragment that contained sequences extending ϳ2.0 kb 5Ј to the initiator methionine codon. A portion of the first exon and the 5Ј-flanking region were sequenced (Fig.  1A), and comparison with the corresponding regions of the rat (37) and human (38) NTR-1 genes revealed a region that is highly conserved between all three species (Fig. 1B). This region contains Sp1-, CRE-, CACCC box-, and initiator elementrelated sequences as well as a large 40-bp palindromic sequence (mPAL) that includes the initiator element-related sequences (Fig. 1B). This conserved sequence is located within a 200-bp region that is required for expression.
RNase protection experiments were performed to map the approximate location of the transcriptional start site. Two 32 Plabeled antisense riboprobes (depicted schematically in Fig.  2A) were hybridized to total brain, kidney, spleen, or yeast RNA, followed by digestion with a mixture of nuclease P1 and ribonuclease T1, and the resulting products were analyzed on a sequencing gel. An antisense probe transcribed from a subcloned SacI fragment (Sac293, Ϫ448 to Ϫ149 relative to the AUG start codon) was fully protected after hybridization with mouse brain RNA but not the control RNAs, indicating that the start site is located upstream of the region encompassed by this probe (Fig. 2B). Hybridization of mouse brain RNA with an antisense probe that extends 233 nucleotides further in the 5Ј direction (B/E 497) resulted in three closely spaced protected fragments (Fig. 2C). The estimated sizes of these protection products indicate that transcription starts within the mPAL region (in the vicinity of position Ϫ523). Identical protection results were obtained with RNA isolated from Me 2 SO-induced N1E-115 cells (data not shown). We attempted to confirm this result using several different oligonucleotides for primer extension experiments using poly(A) ϩ brain RNA and control RNAs isolated from several different tissues, but no specific products were detected. Therefore, we employed RT-PCR to delineate the 5Ј boundary of the gene using a common 3Ј primer and several closely spaced 5Ј primers located at successively further 5Ј positions. Control PCR reactions using the same primer sets to amplify mouse embryonic stem cell genomic DNA were also performed ( Fig. 2D). Specific products were detected by electroblotting the RT-PCR products and hybridizing with a 32 Plabeled rat Ntr-1 gene probe from within the amplified region. Primers with 5Ј end points at Ϫ513 and Ϫ548 produced products of the expected size; however, much less product (at least 10-fold less) was detected after RT-PCR with the Ϫ566 primer (Fig. 2D). These results indicate that the major transcriptional start site is close to position Ϫ548 (most likely at Ϫ545) near the 5Ј end of the mPAL within a sequence that is strikingly similar to the consensus initiator element (32,45).
Sequence analysis revealed that an approximately 500-bp region surrounding the transcriptional start site has the structural features of a CpG island (Fig. 3). CpG islands have been identified in the promoter regions of a variety of housekeeping and tissue-specific genes from higher eukaryotes (46,47). The CpG island in the mouse NTR-1 promoter region has close to the expected frequency of CpG dinucleotides (CpG/GpC Ն 0.6) in a region that has greater than 50% G ϩ C content and encompasses at least 500 bp (Fig. 3). Although the function of CpG islands remains uncertain, they are commonly found in promoter regions, including a large number of promoters that lack TATA elements like the NTR-1 promoter (46,47).
Cell Density and Me 2 SO Treatment Increase Ntr-1 Gene Expression in N1E-115 Cells-N1E-115 cells plated at low density are devoid of high affinity NT-binding sites; however, large increases in binding activity are observed when the cells are grown to high density (29) or treated with Me 2 SO (28). To examine whether these culture conditions result in increased Ntr-1 gene expression, NTR-1 mRNA levels were quantitated by nuclease protection assay using a 32 P-labeled antisense riboprobe corresponding to nucleotides 296 -632 of the rat NTR-1 cDNA (PstI 300). N1E-115 cells were plated at low density (5 ϫ 10 5 cells/15-cm dish) and propagated without subculture but with frequent medium changes in the first set of experiments as described (29). Cells were harvested from duplicate cultures at the indicated times; RNA was prepared, and NTR-1 mRNA levels were quantified using an RNase protection assay (Fig. 4A). Culturing the cells at low density resulted in an initial decline in Ntr-1 gene expression, followed by a gradual increase during growth to stationary phase. NTR-1 mRNA levels increased Ͼ20-fold compared with day 3 levels after 17 days of culture but declined precipitously thereafter as the cultures deteriorated (Fig. 4). To examine Ntr-1 gene expression following Me 2 SO treatment, cells were plated in du-plicate in medium containing reduced serum and treated with 1.5% Me 2 SO for the indicated times (Fig. 4B). Me 2 SO treatment resulted in a 4-fold increase in Ntr-1 gene expression over the course of 72 h. These results indicate that increased Ntr-1 gene expression underlies the previously observed increases in NT-binding sites in these cells (28,29).
Promoter Sequences Required for Me 2 SO Induction in N1E-115 Neuroblastoma Cells-To determine what promoter sequences are required for the response to Me 2 SO, a series of promoter deletion constructs in which up to 1.4 kb of 5Ј-flanking sequences were fused to a luciferase reporter gene were FIG. 2. Mapping the transcriptional start site by RNase protection and RT-PCR. RNase protection experiments were performed by hybridizing either 32 P-labeled Sac293 (B) or B/E 497 (C) antisense riboprobes with total RNA (10 g) from the indicated tissues, subsequent digestion with nucleases P1 and T1, and analysis of the products on a sequencing gel. 32 P-Labeled RNA markers were prepared by transcription of a mixture of pGEM4-NT1.8 templates linearized at different sites to produce transcripts of the indicated lengths as described under "Experimental Procedures." A, schematic diagram of the antisense riboprobes used. Thick black lines represent Ntr-1 gene sequences, and open lines represent pGEM4 vector sequences. The sizes of the regions of the transcript derived from NTR-1 sequences and vector sequences are indicated. The Sac293 probe encompasses a 293-nucleotide region located between the SacI sites at Ϫ448 and Ϫ155. The B/E 497 probe spans 497 nucleotides of the Ntr-1 gene between the EcoRI site at Ϫ682 and the BamHI site at Ϫ185. B and C, RNase protection products resulting from hybridization to the Sac293 (B) or B/E 497 (C) probes were analyzed on sequencing gels and visualized by autoradiography. Specific protection products were detected only in reactions containing brain RNA and are indicated by arrows. D, RT-PCR reactions were performed as described under "Experimental Procedures" using a common 3Ј primer and three different 5Ј primers with 5Ј end points at the indicated positions (Ϫ566, Ϫ548, Ϫ513) to map the transcriptional start site more precisely. RT-PCR reactions in which reverse transcriptase was omitted from the initial reaction (no RT) were performed to ensure that PCR-amplified products were derived from reverse-transcribed mRNA templates. Control PCR reactions were also performed with the same primers using mouse embryonic stem cell DNA as template. The reaction products were analyzed on a 5% polyacrylamide gel, electroblotted, and probed with a 32 P-labeled fragment of the mouse Ntr-1 gene (Ϫ514 to Ϫ448, see A, Southern probe). The gel was stained with ethidium bromide and photographed prior to electroblotting (top panel), and hybridization signals were detected by film autoradiography (bottom panel). The results indicate that the major transcriptional start site is in the vicinity of Ϫ548, since much less RT-PCR product was generated with the Ϫ566 primer than with the Ϫ548 and Ϫ513 primers.
transfected into N1E-115 cells, and the cells were either grown under control conditions or treated with 1.5% Me 2 SO for 72 h. To control for variations in transfection efficiency, a PGK-␤gal plasmid was co-transfected, and ␤-galactosidase activity was used to standardize luciferase activity. The initial series of constructs revealed two interesting features of the regulatory region. First, sequences upstream of Ϫ640 (we have used the numbering system previously used for the rat and human NTR-1 genes (37,38) where ϩ1 is the A of the initiator methionine codon) appear to suppress basal promoter activity (Fig.  5A). Second, promoter elements required for Me 2 SO induction are located between a SmaI site at Ϫ640 and a SacI site at Ϫ448 (Fig. 5A). The Ϫ640 deletion construct is induced 3-5-fold upon Me 2 SO treatment; however, deletion to the SacI site at Ϫ448 completely abolishes Me 2 SO induction, and this construct is actually repressed after Me 2 SO treatment (Fig. 5A). The effect of Me 2 SO was selective, since the expression of a reporter gene controlled by the Rous sarcoma virus-long terminal repeat was not affected by Me 2 SO treatment (data not shown). These experiments define an Me 2 SO-responsive region between Ϫ640 and Ϫ448 and indicate that sequences upstream of Ϫ640 suppress basal promoter activity.
To determine more accurately the boundaries of the regulatory region, additional 5Ј and 3Ј deletion mutants were constructed and tested in N1E-115 cells (Fig. 5B). The response to Me 2 SO was maintained through deletion to Ϫ589; however, deletion to Ϫ569 nearly abrogated the response. In contrast, basal activity decreased in a graded fashion as sequences were deleted from the 5Ј end, except for a small increase when sequences between Ϫ589 and Ϫ569 were deleted, perhaps indicating that a weak repressor element is located in this region. Deletion in the 3Ј direction from the SacI site at Ϫ448 to Ϫ526 had a severe impact on both basal activity and Me 2 SO induc-tion. Further 3Ј deletions had no effect until removal of an Sp1-like sequence (compare constructs 3Ј-590 and 3Ј-603 that have deletion end points on either side of the Sp1-like site), which essentially eliminated expression, indicating that this Sp1-like site is important for basal expression. These results indicate that the Me 2 SO regulatory region lies in the 83-bp region located between Ϫ589 and Ϫ507 and that the region between Ϫ640 and Ϫ507 is required for full basal expression. This regulatory region is highly conserved in the rat and human NTR-1 promoters (Fig. 1B), indicating that it is functionally important in vivo.
The mPAL was examined in more detail by evaluating the effects of clustered mutations along the length of the palindrome (Fig. 6A, mPal mut-1 to -6). Several mutations reduced basal activity; however, mPal mut-6 had the largest effect, reducing activity by ϳ70%. Removal of this sequence in the 3Ј-526 deletion construct greatly reduced both basal activity and Me 2 SO induction (Fig. 5, A and B). Mutations near the 5Ј end of the mPal all had similar effects, reducing basal activity by about half (Fig. 6A, mPal mut-1, -3, and -5). The transcriptional start site lies within the region affected by mPal mut-3. Deletion of the entire mPAL (Fig. 6A, dmPal)  The sequence of the promoter region was analyzed in 100 nucleotide blocks for G/C content and the frequency of CpG and GpC dinucleotides. The ratio of CpG/GpC was calculated, and values of Ն0.6 in regions that had Ն50% G ϩ C content were defined as meeting the criteria for CpG islands as described (43). A 500-bp region extending from approximately 140 bp upstream of the transcriptional start site to 350 bp into exon one has the characteristics of a CpG island. A graph of the ratio of CpG/GpC is shown, and the positions of the GpC and CpG dinucleotides are depicted schematically below along with a schematic of the mouse NTR-1 promoter region. The transcriptional start site is denoted by an arrow, and the coding region is depicted as a black box.
FIG . 4. Cell density and Me 2 SO induce Ntr-1 gene expression in N1E-115 cells. A, cell density experiments were performed in duplicate by plating N1E-115 cells at low density and incubation without further subculturing but with frequent medium changes as described under "Experimental Procedures." Cells were harvested at the indicated times for RNA preparation. Total RNA isolated from each duplicate culture dish (10 g) was hybridized with 32 P-labeled PstI 300 riboprobe and nuclease-treated, and the protected products were analyzed on sequencing gels. The relative levels of NTR-1 mRNA were quantitated using a PhosphorImager, and autoradiographs of the gels are shown below the graphs. The two lanes under each time point represent the protection products obtained with RNA preparations from each of the duplicate culture dishes. B, Me 2 SO induction was analyzed by treating N1E-115 cells with 1.5% Me 2 SO in duplicate for the indicated times and quantitation of NTR-1 mRNA as described in A. expression by ϳ60%, consistent with this sequence being important for basal expression; however, the residual basal expression of this construct indicates that transcription can also initiate outside this region. A mutation near the center of the mPAL (mPal mut-2) had little effect on basal activity but severely curtailed Me 2 SO induction (Fig. 6A). Two other mutants that reduced basal expression had no effect on Me 2 SO induction (mPal mut-4 and-5), whereas the remaining mutations (mPal mut-1, -3, -6) reduced both basal and Me 2 SOinducible expression. These results indicate that the mPAL is composed of multiple cis-regulatory sequences that contribute to basal and Me 2 SO-induced expression. The mPAL core (affected by mPal mut-2) appears to be most critical region involved in Me 2 SO induction, although sequences at the 5Ј border of the mPAL are also important.
To examine the regulatory properties of the mPAL in more detail, one (mP12-4) or two (mP10 -1) copies of an oligonucleotide spanning the mPAL were cloned upstream of the SacI deletion construct (5Ј-448) that displays very low basal expression and is actually inhibited by Me 2 SO treatment (Fig. 6A). Both constructs conferred some Me 2 SO responsiveness to the deleted promoter, but two copies were required to increase basal expression, although neither construct had full activity.
To determine whether inclusion of sequences just upstream of the mPal could restore full Me 2 SO responsiveness, a construct containing sequences between Ϫ589 and Ϫ507 cloned upstream of the SacI promoter deletion was tested (Fig. 6A,  URmP-1). Inclusion of these more 5Ј sequences restored Me 2 SO induction nearly completely, although the basal activity of this construct was increased only marginally (similar to the mP10-1 construct). These results provide additional evidence that sequences immediately upstream of the mPAL and the mPAL itself are required for full Me 2 SO responsiveness.
To define the additional sequences located in the upstream region that are required for full Me 2 SO induction, a series of mutations in this region were analyzed by transfection in N1E-115 cells (Fig. 6B, mut-1 to -4). Mutants 2-4 had little effect on Me 2 SO induction (Fig. 6B), and none of the mutants had a large effect on basal expression (data not shown). In contrast, mut-1, which specifically alters the conserved CACCC box, nearly abolished Me 2 SO inducibility (Fig. 6B). These results indicate that the conserved CACCC box that lies immediately upstream of the mPAL is required for Me 2 SO induction. This result was somewhat surprising in view of the fact that 5Ј deletion to Ϫ569 results in a large drop in Me 2 SO responsiveness (Fig. 5, 5Ј-569), and this deletion removes the sequences mutated in mut-2 to -4. This apparent discrepancy could result from either functional redundancy within the region mutated in mut-2 to -4 (e.g. the sequence GTGGC is directly repeated), the presence of a cis-regulatory element within the deleted region that is not inactivated by any of the individual mutations, or the creation of junction sequences during cloning that have an adventitious effect on activity. Thus, although it remains possible that one or more sequences in the region affected by mut-2 to mut-4 contribute to Me 2 SO responsiveness, the CACCC box is clearly an important cis-regulatory element required for Me 2 SO

Ntr-1 gene promoter region deletion constructs in N1E-115 cells.
A, an initial series of deletion constructs was constructed by cloning the indicated restriction fragments upstream of the luciferase reporter gene in pXP-2 (33). The plasmids were co-transfected with a PGK-␤gal standardization plasmid into N1E-115 cells using calcium phosphate precipitation, and the cells were either grown under control conditions (open bars) or in the presence of 1.5% Me 2 SO (black bars) for 72 h. The cells were harvested, extracts were prepared, luciferase and ␤-galactosidase activities were determined, and ␤-galactosidase activity was used to correct for variations in transfection efficiency. The constructs are depicted schematically, and their activity in light units is plotted. B, an extensive series of 5Ј and 3Ј deletion constructs was generated using PCR methods and tested as described above. The constructs are depicted schematically and activity is plotted as either fold induction or relative basal expression. Transfections were performed in duplicate, and the luciferase and ␤-galactosidase activities were determined in duplicate for each transfection. To correct for variations in transfection efficiency, luciferase activity was divided by ␤-galactosidase activity. Fold induction was calculated by dividing the corrected induced value by the corrected control value. Relative basal expression was calculated by dividing the corrected mutant control value by the corrected wild type control value. The mean values are plotted and the error bars indicate the S.E. (n ϭ 3-8). Promoter fragments generated by PCR were completely sequenced to verify that only the intended changes had been introduced.

induction.
Similar Promoter Elements Are Required for Expression in NG108 Neuroblastoma Cells-NG108 neuroblastoma ϫ glioma cells display high constitutive levels of high affinity NT-binding sites (30). To determine whether similar promoter elements are required for constitutive expression and to examine possible regulation by Me 2 SO, these cells were transfected with a series of promoter constructs using the same methods as were used for N1E-115 cells. Me 2 SO treatment reproducibly increased expression of the wild type promoter construct (5Ј-640) by 2-3fold (Fig. 7A). The results obtained in these cells were similar to those obtained in N1E-115 cells. The 5Ј-and 3Ј-deletion mutants define a region between Ϫ589 and Ϫ507 that is required for full induction, and the analysis of selected clustered point mutations indicates that the CACCC box (URmut-1) and the center of the mPAL (mPalmut-2) are critical sequence elements required for induction (Fig. 7A). The only real difference was that mPal mut-6 had a modest negative effect on Me 2 SO induction in N1E-115 cells but slightly enhanced the response in NG108 cells. The region required for constitutive expression (Ϫ622 to Ϫ507) was also similar to that found in N1E-115 cells. The major differences were that 5Ј deletion to Ϫ622 had no effect on constitutive expression in NG108 cells but decreased basal expression in N1E-115 cells by about onehalf, and several mPal mutants that reduce basal expression in N1E-115 cells had no effect on expression in NG108 cells (mPalmut-1, -3, and -5). These results indicate that similar promoter elements are required for expression in these two neuroblastoma cell lines and provide further evidence that the CACCC box and the core of the mPAL are critical elements required for Me 2 SO induction.
Distinct Promoter Requirements for Expression in Mouse Em-bryo Fibroblasts-To determine whether the NTR-1 promoter is expressed in a cell type-specific manner, the same series of constructs that were tested in NG108 cells were also transfected into mouse embryo fibroblasts that do not express the endogenous Ntr-1 gene (data not shown). Preliminary experiments revealed that Me 2 SO treatment of these cells had no effect on the expression of the wild type promoter, indicating that Me 2 SO induction is cell type-specific. The wild type promoter (5Ј-640) was active in these cells; however, this basal expression was dependent on largely different promoter elements compared with the neuroblastoma cell lines. There were several major differences. First, the Sp1-like sequence that is removed in the 5Ј-589 promoter construct appears to be unimportant for expression in MEF cells (Fig. 7B). Second, deletion of the region between the Sp1-like site and the mPAL (5Ј-548) essentially abolished expression in MEF cells but had only a modest effect in NG108 cells (Fig. 7B) and essentially no effect in N1E-115 cells (Fig. 5B). The CACCC box appears to be critical for basal expression in these cells (Fig. 7B), in contrast to the situation in the neuroblastoma cell lines where the CACCC box is required for Me 2 SO induction and makes at most a modest contribution to basal expression (data not shown). The mPAL was also found to be important for basal expression in MEF cells; however, the core of the mPAL appears to be extremely important for basal expression in these cells (Fig. 7B, mPal mut-2) but again is mainly involved in Factors Binding to the Regulatory Region-DNase I footprint analysis was used to identify protein-binding sites in the Ntr-1 gene regulatory region. Nuclear extracts were prepared from N1E-115 cells that were either grown under control conditions or treated with 1.5% Me 2 SO for 48 h. 32 P-End-labeled promoter FIG. 7. Transfection analysis of Ntr-1 gene promoter constructs in NG108 and MEF cells. A subset of the reporter constructs diagrammed in Figs. 5 and 6 were transfected into either NG108 cells or MEF cells together with a PGK-␤-gal standardization plasmid as described for N1E-115 cells. Initial experiments indicated that the full-length Ntr-1 gene promoter construct was induced by Me 2 SO in NG108 cells but not in fibroblasts. Therefore, reporter constructs were transfected into NG108 cells and assayed exactly as described for N1E-115 cells in the presence or absence of 1.5% Me 2 SO, but Me 2 SO induction was not further analyzed in MEF cells. A, NG108 cells were transfected with the indicated constructs and grown either under control conditions or in the presence of 1.5% Me 2 SO for 72 h. The data are plotted as fold induction as described for N1E-115 cells (n ϭ 3-4). B, the basal activity of the indicated NTR-1 promoter constructs in either NG108 (solid bars) or MEF (diagonal fill bars) cells is plotted relative to the activity of the Ϫ640 wild type promoter construct (n ϭ 3-4). C, the relative activity of the Ϫ640 wild type Ntr-1 gene promoter construct in the different cell lines was calculated by dividing luciferase activity by ␤-galactosidase activity. In each case, cells were transfected with 10 g of the Ϫ640 NTR-1 promoter luciferase and 2.5 g of the PGK-␤-gal reporter constructs. The data are plotted as light units/␤-galactosidase units, and the standard error is indicated (n ϭ 3). fragments were mixed with protein from either control or induced cells, incubated on ice, treated with different concentrations of DNase I, phenol-extracted, and analyzed on a sequencing gel (Fig. 8A). A probe labeled at the BamHI site at Ϫ175 that extends to an EcoRI site at Ϫ680 revealed that the functional promoter, defined by the mutational analysis described above, is extensively protected by nuclear extracts from Me 2 SO-induced cells (Fig. 8A). The specific protections are indicated by circles and fall into six footprinted regions (FP1-6). Inducible or partially inducible protections are represented by open and gray circles, respectively, and constitutively protected sites are represented by black circles. Comparison of the patterns obtained with Me 2 SO-induced and control N1E-115 nuclear extracts indicates that FP1-3 and FP6 are induced by Me 2 SO; FP4 is better protected by Me 2 SO-induced cell extracts, and FP5 is constitutive. PC12 cell nuclear extract resulted in a protection pattern that was nearly identical to that obtained with control N1E-115 cell extracts (data not shown), indicating that the pattern obtained with uninduced control cell extracts is not cell type-specific. PC12 cells express no detectable NTR-1 mRNA (data not shown). A cluster of constitutive hypersensitive sites (indicated by arrows in Fig. 8A) was observed just downstream of a putative Sp1 site, indicating that this region is occupied in both Me 2 SO-induced and control cells. The significance of FP5 and FP6 is uncertain since deletion of this region has no effect on either basal expression or Me 2 SO induction. However, the Me 2 SO-inducible footprints FP3 and FP4 lie over functionally important sequences.
To examine FP-1 in more detail, the EcoRI/BamHI promoter fragment was labeled at the EcoRI site and subjected to footprint analysis using Me 2 SO-induced and control N1E-115 cell nuclear extracts (Fig. 8B). The footprint was found to consist of a region of Me 2 SO-induced protections (FP1A) and an adjacent region that was constitutively protected (FP1B). The constitutive footprint at least partially overlaps a half-CRE site (CGTCA), and many CREs bind cAMP-response element family proteins constitutively (48). The Me 2 SO-inducible FP1A is within a region that the 5Ј-622 deletion construct indicates is required for full basal expression in N1E-115 cells but does not affect Me 2 SO induction (Fig. 5B); thus, the significance of this site is uncertain. Since there is a relative paucity of strong DNase I cleavages 5Ј of FP1A, it remains possible that additional proteins may bind to this region.
To characterize further the factors binding to the promoter region, gel shift experiments were performed using a probe that spans the mPAL and CACCC element (Ϫ589 to Ϫ507). Nuclear extracts were prepared from N1E-115 cells that were either treated with 1.5% Me 2 SO for 72 h or grown under control conditions and incubated with the 32 P-labeled promoter fragment, and the resulting complexes were analyzed on non-denaturing acrylamide gels (Fig. 9A). At least four complexes (Fig. 9A, complexes 1-4) were identified that were specifically competed by cold wild type probe fragment (longer exposure times were required to detect complexes 3 and 4, Fig. 9A, bottom panel). To examine whether the functional sequence elements defined in the transfection experiments described above were required for the formation of these complexes, competition experiments were performed with wild type and mutant promoter fragments. An oligonucleotide corresponding to the collagenase TRE (49) was used as a nonspecific competitor. Competition with a cold mutant promoter fragment containing clustered point mutations in the CACCC element (mut-1 in Fig. 6B) identified two complexes that require this sequence for binding (Fig. 9A, 4th lane, complexes 1 and 2). Comparison between the induced and control lanes indicates that Me 2 SO treatment results in the specific induction of complex 1 (Fig. 9A, compare 2nd and 6th lanes). Close DNA contacts in the inducible complex were analyzed using a methylation interference assay (Fig. 9B). Complexes were formed by mixing partially methylated 32 P-end-labeled probe (the same probe used in the experiment depicted in Fig. 8A) with nuclear extract from Me 2 SO-induced N1E-115 cells, separated on a native acrylamide gel, and the regions corresponding to complex 1/2 and free probe were excised for analysis of methylated G residues on sequencing gels. Several bands were underrepresented in complex 1/2 compared with free probe, specifically the Gs that are complementary to last three Cs in the CACCC sequence (Fig. 9B). Analysis of the other strand was not informative most likely due to the absence of G residues in the region containing the CACCC sequence. These results indicate that the Me 2 SO-inducible complex makes specific DNA contacts in the CACCC element, since complex 1 is the major component of the complex 1/2 band in Me 2 SO-induced cell extracts (see Fig. 9A).
The CACCC box is a binding site for a number of zinc finger FIG. 8. DNase I footprint analysis of the mouse Ntr-1 gene regulatory region. DNase I footprinting assays were performed using an EcoRI/BamHI mouse NTR-1 promoter fragment 32 P-end-labeled at either the BamHI (A) or EcoRI (B) site and nuclear extracts (160 g) from either control (Ϫ) or Me 2 SO-induced (48 h, ϩ) N1E-115 cells. A series of reactions containing increasing amounts of DNase I were performed, and equal counts (2,000 cpm) were loaded on a sequencing gel for analysis. The products of a reaction in which the probe was treated with DNase I in the absence of protein (NP) were also analyzed. Lanes in which the extent of probe digestion was similar were compared and are depicted. Chemical sequencing reactions (lanes G/A, G) were used to generate markers for the alignment of footprinted regions on the promoter sequence. DNase I protections are indicated by either open circles (Me 2 SO-inducible), gray circles (partially inducible), or black circles (constitutive). Footprinted regions are indicated schematically by boxes shaded like the circles and are numbered 1-6. The mouse NTR-1 promoter sequence in the region analyzed is depicted to the right of the autoradiographs and promoter schematics, and the positions of the mPAL and CACCC box are indicated. B, the antisense strand was labeled, but the coding sequence is shown for clarity. transcription factors, including Sp1-related and Krü ppel-like proteins (50 -52). To examine whether complexes 1 and 2 were Sp1-related factors, competition and antibody detection experiments were performed using the same probe (Fig. 9C). A competitor oligonucleotide containing a consensus Sp1 site effectively competed for binding of complex 2 but not complex 1, indicating that complex 2 has a binding specificity similar to Sp1. To determine whether complex 2 contains Sp1, a specific Sp1 antibody was added to the binding reaction at two different concentrations (1-4 l), and for comparison a Jun B-specific antiserum was added to control reactions. Complex 2 was abolished by addition of 4 l of Sp1 antiserum but was not affected by the same amount of JunB antiserum, indicating that complex 2 contains Sp1 (Fig. 9C). These results provide evidence that complex 2 is due to binding of Sp1 to the CACCC box; however, complex 1 appears to be due to a distinct factor(s), perhaps a Krü ppel-like protein, that is specifically induced during Me 2 SO-induced neuronal differentiation of N1E-115 cells.

DISCUSSION
Several Sequences, Including a CACCC Box, Are Required for Me 2 SO Induction of Ntr-1 Gene Expression-DNA transfection experiments were used to define promoter elements required for Ntr-1 gene induction in neuroblastoma cells. Previous work has shown that Me 2 SO-stimulated neuronal differentiation of N1E-115 neuroblastoma cells is accompanied by increased expression of high affinity neurotensin-binding sites (28). We demonstrate here that Me 2 SO treatment induces the expression of the endogenous Ntr-1 gene from 3-to 5-fold and has a similar effect on the expression of a luciferase reporter gene driven by NTR-1 promoter sequences. Me 2 SO also increased FIG. 9. An Me 2 SO-inducible complex binds to the CACCC element in the NTR-1 regulatory region. A, a 32 P-labeled probe spanning the mPAL and immediate 5Ј-flanking sequences (Ϫ589 to Ϫ507) was incubated with nuclear extracts prepared from N1E-115 cells that had either been treated with 1.5% Me 2 SO for 72 h or grown under control conditions as indicated, and the resulting complexes were analyzed on a non-denaturing acrylamide gel and visualized by autoradiography with an intensifying screen for 16 h. A longer exposure (72 h) of the upper part of the gel is shown below to visualize complexes 3 and 4. Binding reactions were performed either in the absence of unlabeled competitor DNA (2nd and 6th lanes) or in the presence of a 100-fold molar excess of either wild type (WT), CACCC mutant (Mut 1), or collagenase TRE (Col TRE) DNA. Complexes 1 and 2 require an intact CACCC element for binding. B, nucleotide contacts required for binding of complexes 1 and 2 were analyzed by methylation interference. The probe described above was partially methylated with dimethyl sulfate, mixed with nuclear extract from Me 2 SO-induced cells, and the resulting complexes were separated on a low ionic strength polyacrylamide gel. The regions of the gel containing complexes 1 and 2 (lane B, bound) and free probe (lane F, free) were excised, the DNA was purified, cleaved with piperidine, and the products were analyzed on a sequencing gel. The input probe was also analyzed (lane P, probe). C, complex 2 is closely related to Sp1. The probe described above was incubated with nuclear extracts of cells treated with 1.5% Me 2 SO for 72 h, and the resulting complexes were analyzed on a non-denaturing acrylamide gel. Binding reactions were performed either in the absence of competitor (No comp) or in the presence of a 100-fold molar excess of oligonucleotides containing either a consensus Sp1-binding site (Sp1) or the collagenase TRE (Col TRE). To examine whether complexes 1 and 2 were immunologically related to Sp1, two different amounts (1 and 4 l) of either an Sp1-specific or control JunB-specific antibody were added to the binding reactions as indicated. The free probe was also loaded (Probe).
Ntr-1 gene expression in NG108 neuroblastoma ϫ glioma hybrid cells that have previously been shown to express high affinity NT-binding sites (30). This was a selective effect since the expression of a luciferase reporter gene controlled by the Rous sarcoma virus-long terminal repeat was not affected by Me 2 SO treatment of either neuroblastoma cell line. In contrast, Me 2 SO had no effect on NTR-1 reporter constructs transfected into fibroblasts. Mutational analysis of the NTR-1 promoter has defined an 80-bp region encompassing a CACCC box and large GC-rich palindrome as critical for Me 2 SO induction. The CACCC box is bound by an Me 2 SO-induced complex in N1E-115 cells. Several regions within the mPAL are required for full Me 2 SO induction; however, the central core appears to be the most critical Me 2 SO-responsive sequence. The transcriptional start site was mapped to the 5Ј end of the mPAL to a region that closely matches the initiator element that was first identified in the terminal deoxynucleotidyltransferase gene (32). The regulatory sequences identified through transfection analysis here are highly conserved between mouse, rat, and human (Fig. 1B), indicating that they are functionally important in vivo. These results provide evidence that a conserved region surrounding the transcriptional start site is critical for Ntr-1 gene activation during a program of neuronal differentiation initiated by Me 2 SO treatment in N1E-115 cells.
The CACCC box was first identified through sequence comparisons of ␤-globin genes and has been shown to be important for expression in erythroid cells. This site binds a number of related zinc finger transcription factors, including Sp1-and Krü ppel-related proteins (50,53). The majority of the characterized Krü ppel-related genes are expressed in tissue-restricted patterns. Gene targeting approaches have revealed that the erythroid Krü ppel-like factor is essential for high level erythroid-specific ␤-globin gene expression (54,55) and that lung Krü ppel-like factor is essential for T-cell development (56), indicating that specific Krü ppel-like proteins play key roles in transcriptional control and terminal differentiation. The ␤-globin CACCC boxes are required for activation during Me 2 SO-induced differentiation of mouse erythroleukemia cells (57), possibly through the phosphorylation of erythroid Krü ppel-like factor (58). The CACCC box in the NTR-1 promoter is clearly required for Me 2 SO induction and is bound by both constitutive and Me 2 SO-inducible complexes in N1E-115 neuroblastoma cells. The major constitutively expressed complex is closely related to Sp1 (Fig. 9C, complex 2). The inducible complex clearly has a binding specificity different from that of the Sp1-related proteins and makes close contacts within the CACCC box, similar to Krü ppel-like proteins (see for example Ref. 53). These data suggest a model in which the induction of one or more Krü ppel-like factors is required for Ntr-1 gene activation during N1E-115 cell differentiation. Intriguingly, both gut-enriched (52) and neuron-enriched (59 -61) Krü ppellike factors have been described, and these are the major sites of Ntr-1 gene expression in vivo (15,31).
The mutational analysis of the mPAL revealed that it most likely consists of several independent cis-active elements that influence Me 2 SO-induced and basal expression. Several mutations along the length of the mPAL reduced Me 2 SO inducibility in both N1E-115 and NG108 cells (Figs. 6A and 7A). Alteration of sequences near the center of the mPAL greatly reduced Me 2 SO responsiveness (Figs. 6A and 7A, mPal mut-2). This sequence is a perfect palindrome consisting of alternating G and C residues and is conserved in both the rat and human genes; however, comparisons with known transcription factorbinding sites did not reveal any close similarities, indicating that either it is a novel binding site or functions as a DNA structural element. Mutation of an additional sequence located at the 5Ј end of the palindrome (mPal mut-3) also decreased Me 2 SO activation in both N1E-115 and NG108 cells but appears to affect a distinct functional element. This sequence is nearly identical to the consensus initiator element, and mPal mut-3 also reduced basal activity in N1E-115 and MEF cells but surprisingly not NG108 cells (Figs. 6B and 7B). The major transcriptional start site was also mapped to this location making it likely that this is a functional initiator. DNase I footprint analysis indicates that factor binding to this site is regulated by Me 2 SO in N1E-115 cells (Fig. 8A). This could be due to either increased expression of a factor that binds to this site or cooperative interactions with the Me 2 SO-inducible CACCC box-binding factor. Thus, the mutational analysis indicates that there are at least two distinct cis-regulatory elements within the mPAL that are required for full Me 2 SO responsiveness, and DNase I footprint analysis indicates that Me 2 SO-inducible complexes bind in the mPAL region (Fig. 8, A  and B). The central portion of the palindrome contains few DNase I-sensitive sites; thus, further analysis will be required to determine whether Me 2 SO-inducible complexes bind to the mPAL core region. These results coupled with the results discussed above for the CACCC box indicate that Me 2 SO induction of Ntr-1 gene expression requires cooperative interactions between multiple sites, including the CACCC box, the mPAL core, and the initiator element.
Different Promoter Sequences Required for Basal Expression in Neuroblastoma and MEF Cells-There were several differences in the sequences required for basal expression in the three cell lines, although the results obtained in N1E-115 and NG108 cells were similar. A notable difference was that the CACCC box and the central region of the mPAL are required for high level basal expression in MEF cells but are principally involved in Me 2 SO responsiveness in the neuroblastoma cell lines. The contribution of the CACCC box to basal expression in MEFs is most likely due to the expression of one or more CACCC box-binding proteins in these cells, for instance BKLF is expressed in fibroblasts (62). A CACCC box has been shown to be important for ␤-globin gene expression in transient expression assays in non-erythroid cell lines (63), although this element mediates cell-specific expression through the binding of erythroid Krü ppel-like factor in erythroid cells (54,55,57). The endogenous Ntr-1 gene is most likely repressed in most cell types and only accessible to Krü ppel-like proteins and perhaps other CACCC box-binding proteins in neurons and a restricted set of other cell types. The induction of these proteins during neuronal differentiation or in response to environmental cues could underlie Ntr-1 gene activation in specific neuronal populations.
There were also differences in the requirement for the Sp1like sequence located between Ϫ590 and Ϫ603 and a sequence near the 5Ј end of the positive regulatory region for basal expression. The Sp1-related sequence was clearly important for basal activity in the neuroblastoma cell lines but not in MEF cells (compare 3Ј-590 and 3Ј-603 in Figs. 5B and 7B). Sp1 is constitutively expressed in N1E-115 cells (see Fig. 9A, complex 2), and nuclear extracts from both control and Me 2 SO-induced cells create DNase I-hypersensitive sites just downstream of the Sp1-related sequence indicating that this site is constitutively occupied. Sequences near the 5Ј end of the positive control region were found to be important for basal expression in N1E-115 cells (Fig. 5B, compare 5Ј-640 and 5Ј-622) but not in NG108 or MEF cells (Fig. 7B). DNase I footprinting experiments indicate that this region binds both constitutive and Me 2 SO-inducible factors (Fig. 8B), and sequence comparisons indicate that this region is similar to a neural specific regulatory element identified in the Drosophila dopadecarboxylase (Ddc) gene (64). The Ntr-1 gene is expressed at high levels in midbrain dopamine neurons (31), and the homology to the Drosophila neural element raises the intriguing possibility that this region may be required for expression in catecholamine-producing neurons (N1E-115 cells produce catecholamines).
Although there were significant differences in the promoter elements required for basal expression, certain regions of the mPAL are important in all the lines examined. The most critical sequence is defined by mPal mut-6 which had a severe impact on basal expression in all lines but had only a modest effect on Me 2 SO induction in N1E-115 cells (Figs. 6A and 7, A  and B). The sequence affected by this mutation is similar to the consensus NGFI-A/Egr-1/Krox 24-binding site (65), and DNase I footprint analysis indicates that it is bound by a factor that is regulated by Me 2 SO but also expressed in uninduced N1E-115 cells (Fig. 8A). Neurotrophic factors and other stimuli induce ngfi-A gene expression (65); however, further work will be required to determine whether NGFI-A plays a role in Ntr-1 gene expression.
Transcription Initiates within the mPAL-A combination of RNase protection experiments and RT-PCR were used to map the transcription initiation site of the Ntr-1 gene to a position that is either at or near a consensus initiator sequence at position Ϫ545 (see Fig. 1A). This position is well upstream of the sites previously described for the rat and human NTR-1 genes that were mapped solely by primer extension analysis (37,38). The RNase protection data presented in Fig. 2B strongly indicate that the sites mapped in the rat and human are not utilized in the mouse, since a probe spanning these sites was fully protected. The sequences of the mouse and rat promoter regions are closely similar (Fig. 1B) making it unlikely that the transcriptional start site would differ substantially. Primer extension within the GC-rich promoter region is problematic, and the rat start sites were viewed as tentative for this reason (37). Our own attempts to map the transcription start site by primer extension using multiple primers were unsuccessful. The RT-PCR analysis (Fig. 2D) that was performed to map more accurately the transcriptional start site required the use of high concentrations of betaine (41) to allow PCR to proceed through the GC-rich promoter region. The recent characterization of a mouse NTR-1 cDNA (GenBank TM accession number AB017127) with a 5Ј end corresponding to position Ϫ451 of the mouse gene sequence (Fig. 1A) provides additional support that transcription starts considerably upstream of the sites proposed for the rat, although this is most likely not a full-length cDNA. There is considerably more sequence divergence between the mouse and human NTR-1 gene promoters, although the regulatory region identified here is closely conserved (Fig. 1B). Transcription could initiate further downstream in the human as indicated by primer extension experiments (38). We conclude that the major transcriptional start site of the mouse gene is located at or near the initiator consensus sequence near the 5Ј end of the mPAL, a sequence that is closely conserved in the rat and human genes.
The NTR-1 and DA D 2 Receptor Regulatory Regions Share Considerable Sequence Homology-Previous studies have demonstrated that the majority of midbrain DA neurons express NTR-1 (5), and these neurons also express DA D 2 autoreceptors (66). Comparison of the promoter regions from the DA D 2 receptor and Ntr-1 genes (Fig. 1C) revealed a striking similarity (66% sequence identity over a 60-bp region of the D 2 receptor (Ϫ53 to ϩ4) that includes the transcription start site (67)) and extends through the mPAL and into the CACCC box of the Ntr-1 gene (Ϫ559 to Ϫ507). These sequences are required for DA D 2 receptor gene expression in a neuroblastoma cell line (NB41A3) and are negatively regulated by upstream sequences (67,68). This homology suggests that transcriptional initiation and perhaps other aspects of the regulation of these two genes are similarly controlled. The D 2 receptor promoter also contains a sequence composed of three TGGG repeats that is similar to the upstream Sp1-related sequence in the NTR-1 promoter (Ϫ602 to Ϫ590) required for full basal promoter activity. The TGGG repeat region of the D 2 receptor is paired with a consensus Sp1 site in a negative modulatory region of the D 2 receptor gene promoter (67,69). The negative regulatory action of these sites may be conferred by a protein that binds these sites but is not Sp1 by several criteria (69). These sequence similarities suggest that these two genes that are coexpressed in DA neurons could rely on similar regulatory strategies; however, additional regulatory elements are also likely to be required to generate the specific complex patterns of expression characteristic of these two genes.