Novel dual repressor elements for neuronal cell-specific transcription of the rat 5-HT1A receptor gene.

The level of expression of the 5-HT1A receptor in the raphe and limbic systems is implicated in the etiology and treatment of major depression and anxiety disorders. The rat 5-HT1A receptor gene is regulated by a proximal TATA-driven promoter and by upstream repressors that inhibit gene expression. Deletion of a 71-base pair (bp) segment between -1590/-1519 bp of the 5-HT1A receptor gene induced over 10-fold enhancement of transcriptional activity in both 5-HT1A receptor-expressing (RN46A raphe and SN48 septal) cells and receptor-negative (L6 myoblast and C6 glioma) cells. A 31-bp segment of the repressor was protected from DNase I digestion by RN46A or L6 nuclear extracts. Within the 31-bp segment, a single protein complex was present in receptor-expressing cells that bound a novel 14-bp DNA element; in receptor-negative cells, an additional complex bound an adjacent 12-bp sequence. In receptor-positive but not receptor-negative cells, mutation of the 14-bp element to eliminate protein binding abrogated repression to nearly the same extent as deletion of the -1590/-1519 bp segment. Additional mutation of both 14-bp and 12-bp elements abolished protein binding and repressor activity in receptor-negative cells. Thus a single protein-DNA complex at the 14-bp element represses the 5-HT1A receptor gene in 5-HT1A receptor-positive neuronal cells, whereas adjacent DNA elements provide a dual repression mechanism in 5-HT1A receptor-negative cells.

The level of expression of the 5-HT1A receptor in the raphe and limbic systems is implicated in the etiology and treatment of major depression and anxiety disorders. The rat 5-HT1A receptor gene is regulated by a proximal TATA-driven promoter and by upstream repressors that inhibit gene expression. Deletion of a 71base pair (bp) segment between ؊1590/؊1519 bp of the 5-HT1A receptor gene induced over 10-fold enhancement of transcriptional activity in both 5-HT1A receptor-expressing (RN46A raphe and SN48 septal) cells and receptor-negative (L6 myoblast and C6 glioma) cells. A 31-bp segment of the repressor was protected from DNase I digestion by RN46A or L6 nuclear extracts. Within the 31-bp segment, a single protein complex was present in receptor-expressing cells that bound a novel 14-bp DNA element; in receptor-negative cells, an additional complex bound an adjacent 12-bp sequence. In receptor-positive but not receptor-negative cells, mutation of the 14-bp element to eliminate protein binding abrogated repression to nearly the same extent as deletion of the ؊1590/؊1519 bp segment. Additional mutation of both 14-bp and 12-bp elements abolished protein binding and repressor activity in receptor-negative cells. Thus a single protein-DNA complex at the 14-bp element represses the 5-HT1A receptor gene in 5-HT1A receptor-positive neuronal cells, whereas adjacent DNA elements provide a dual repression mechanism in 5-HT1A receptor-negative cells.
The serotonergic neurons of the raphe nuclei are the primary site of serotonin synthesis in the brain. They send projections to a wide variety of brain regions, including the hippocampus, cortex, limbic system, and hypothalamus (1). Activation of postsynaptic receptors in the above regions is associated with serotonergic regulation of memory, motivation, emotion, neuroendocrine stress response, etc. (2)(3)(4). The activity of serotonergic neurons of the raphe nuclei is regulated in part by presynaptic autoreceptors. The 5-HT1A autoreceptor is located at the cell body and dendrites of raphe serotonergic neurons (5,6) and mediates negative feedback inhibition of the firing rate through recurrent activation of potassium channels via pertussis toxin-sensitive G proteins (7) to decrease serotonin release. Thus the 5-HT1A receptor plays a major role in controlling serotonergic outflow to the wide variety of brain regions that are innervated by the raphe nuclei.
Abnormal regulation of 5-HT1A receptor expression is implicated in depression and anxiety disorders. 5-HT1A receptor knockout mice display increased anxiety-related behaviors (8 -10), suggesting that a loss of the 5-HT1A autoreceptors is correlated with symptoms of anxiety (3). On the other hand, 5-HT1A receptor levels are increased in the midbrain of suicide victims with major depression compared with nondepressed suicides (11). Down-regulation of the 5-HT1A autoreceptor by antidepressants (12,13) disinhibits action potential firing of the raphe neuron, thereby enhancing serotonergic neurotransmission (14 -17). The prolonged (2)(3) week) time course required for antidepressant action suggests an alteration in transcriptional activity of the 5-HT1A receptor.
To investigate the mechanisms that regulate cell-specific and basal regulation of the 5-HT1A receptor, we have identified the transcriptional start site and examined the regulation of the transcriptional activity of a 2.719-kb 1 fragment of the rat 5-HT1A receptor gene in several cell lines, including 5-HT1A receptor-positive RN46A raphe and SN48 septal cells (18 -20) and receptor-negative L6 myoblast and C6 glioblastoma cells (21). We identified a region of the rat 5-HT1A receptor gene located upstream of an ubiquitously active promoter region that reduces transcriptional activity. In the present study, we have identified a 14-bp element in the 5Ј-flanking region of the 5-HT1A gene that mediates transcriptional repression in raphe cells, but is dispensable in receptor-negative cells where an adjacent 12-bp element maintains repression of the gene. In contrast to the single repressor-DNA complex present in 5-HT1A receptor-expressing cells that may modulate basal levels of receptor expression, the presence of two protein-DNA complexes in receptor-negative cell lines provides a dual mechanism to repress 5-HT1A receptor expression.

EXPERIMENTAL PROCEDURES
Plasmid Construction-The 2.7-kb fragment of rat 5-HT1A receptor promoter-luciferase pGL3-Basic (Promega, Madison, WI) reporter construct (Ϫ2719 (21)) was digested with XbaI to generate a plasmid that was ligated internally with a 1.590-kb 5-HT1A gene fragment (Ϫ1590) containing 71 bp from the repressor region (between Ϫ1519 and Ϫ1590) and two other resultant fragments of 0.4 and 0.8 kb, respectively. The Ϫ1590 construct was digested with SmaI and ligated internally to obtain construct Ϫ1519, which lacks the repressor segment. To obtain constructs of Ϫ2300 and Ϫ2300/Ϫ1519, the fragment of 0.8 kb was * This research was supported by the Medical Research Council (MRC) of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  ligated with Ϫ1590 and Ϫ1519 constructs, respectively. All of the constructs were verified by dideoxynucleotide DNA sequencing.
Construction of Ϫ2300 Mutants-The Ϫ2300 mutations were made by site-specific mutagenesis of the Ϫ2300 sequence contained in the pGL3-Basic plasmid using a mutagenesis kit (U.S.E., Pharmacia Biotech, Milwaukee, WI). Three mutants (designated Ϫ2300 m1, Ϫ2300 m2, and Ϫ2300 m3) were made to incorporate the nucleotide substitutions (underlined in Table I) in the mut14, mut12, and mut14 ϩ mut12 electrophoretic mobility shift assay (EMSA) primers (Table I, lanes 4 and 8) using the oligonucleotides listed in Table I (lanes 5, 9, and 10, respectively). The nucleotide substitutions preserved the nucleotide content, but were found to interfere with protein binding to the 31-bp element (Ϫ1555/Ϫ1524). The resulting plasmids were verified by dideoxynucleotide DNA sequencing.
Cell Culture and Transient Transfection-Cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Gaithersburg, MD) supplemented with 10% fetal calf serum at 37°C in 5% CO 2 except for the RN46A cell line, which was maintained in Neurobasal medium (Life Technologies) supplemented with 10% fetal calf serum at 33°C in 5% CO 2 . All cell lines were grown to 50 -60% confluence, and the media were replaced 12 h before transfection. Cells were transfected transiently by calcium phosphate coprecipitation (18) using 20 g/plate of luciferase constructs and 10 mg/plate of pCMV-lacZII plasmid (ATCC) except for RN46A cells. RN46A cells were transfected in 3.5-cm Primaria 6-well plates (Falcon, Franklin Lakes, NJ) using 1.5 g/plate luciferase plasmid and 0.5 g/plate pCMV-lacZII with 12 g/plate pFx-7 (Invitrogen, San Diego, CA) at 33°C for 8 h and harvested at 48 h following transfection (21). All plasmids used for transfection were purified using columns (Maxiprep, Qiagen, Inc., Chatsworth, CA) and quantified by spectrophotometric analysis and by ethidium bromide staining.
Luciferase and ␤-Galactosidase Assays-Cells were extracted with 100 l/well of Reporter lysis buffer (Promega), and the supernatants were collected and assayed for luciferase activity using a Promega luciferase assay system using a luminometer (model 1250, BioOrbit, Finland). To correct for variations in transfection efficiency, plasmid pCMV-lacZII was cotransfected with each luciferase reporter construct. A portion of the harvested cell extract (10%) was assayed for ␤-galactosidase activity based on the conversion of 4-methylumbelliferyl-␤-Dgalactoside (Sigma) to the highly fluorescent molecule methylumbelliferone and quantitated using a Perkin-Elmer LS50 spectrofluorometer, at 350-nm excitation and 450-nm emission. The ratio of luciferase to ␤-galactosidase activity was determined in triplicate samples and normalized to vector-transfected extracts. All data are presented as the mean Ϯ S.E. of at least three independent experiments normalized to pGL3-Basic activity.
DNase I Protection Assay-The Ϫ1590 construct was digested with MluI. The resulting linear DNA was end-labeled using the Klenow fragment of DNA polymerase and 3000 Ci/mmol [␣-32 P]dCTP. Labeled DNA was digested with HpaI to generate a 208-bp fragment, which was purified on an agarose gel. The 60-l reactions contained 20,000 cpm of labeled probe, 3 g of poly(dI-dC) (Roche Molecular Biochemicals), 30 g of nuclear extract from RN46A and L6 cells in binding buffer containing 20 mM HEPES, 0.2 mM EDTA, 0.2 mM EGTA, 100 mM KCl, 5 mM MgCl 2 , 5% glycerol, and 2 mM dithiothreitol, pH 7.9 and were incubated for 25 min at room temperature. Following incubation samples were digested with DNase I (Pharmacia) for 2 min at room temperature with addition of 15 g of yeast tRNA to stop the reaction. After phenol/ chloroform extraction, samples were resuspended in formamide/dye solution and electrophoresed through 6% polyacrylamide/urea gel. Total nuclear protein for all extracts was quantitated according to the Bradford method with bovine serum albumin as standard.
EMSA-Complementary oligonucleotides with 5Ј overhangs were designed according to the nucleotide sequence of the proximal repressor region of the 5-HT1A receptor gene (see Table I) and labeled with [␣-32 P]dCTP using the Klenow fragment of DNA polymerase. Nuclear extract (15 g/sample) was preincubated with or without competitor DNA in a 20-l reaction containing DNA binding buffer (20 mM HEPES, 0.2 mM EDTA, 0.2 mM EGTA, 100 mM KCl, 5 mM MgCl 2 , 5% glycerol, and 2 mM dithiothreitol, pH 7.9) and 3 g of poly(dI-dC) at room were transfected in RN46A raphe cells or L6 myoblast cells as indicated, and luciferase activity was measured as described in "Experimental Procedures." To correct for variation in transfection efficiency among the constructs, a CMV-␤galactosidase plasmid was cotransfected with each construct. pGL3-Promoter (pGL3P) containing the SV40 promoter was measured as a positive control. Data are expressed as the ratio of luciferase to ␤-galactosidase activity of triplicate samples normalized to the control vector (pGL3-Basic) activity and represent the mean Ϯ S.E. of at least three independent experiments. temperature for 25 min. 32 P-Labeled probe (50,000 cpm/sample) was added and incubated for an additional 20 min at room temperature. The reaction was separated on a 5% polyacrylamide gel at 4°C. The gels were dried and then exposed to film overnight at Ϫ80°C with an intensifying screen.

Upstream Repressor and Enhancer Activities in RN46A and
L6 Cells-Transcription of the rat 5-HT1A receptor gene is initiated at a major start site located Ϫ967 bp upstream from the translational initiation ATG sequence and is regulated by a TATA-containing promoter/enhancer region between Ϫ1519 and Ϫ967 bp (21). A region of strong repressor activity is located upstream from this promoter. To identify specific gene regulatory elements, a series of finer deletions of the rat 5-HT1A receptor gene fused to the luciferase reporter gene were transfected in 5-HT1A receptor-positive rat raphe RN46A and receptor-negative rat L6 myoblast cell cultures and the relative luciferase activity was quantitated (Fig. 1). Luciferase activities of the 2300-luciferase gene were 5.66 Ϯ 0.49-and 4.6 Ϯ 0.36-fold basal in RN46A and L6 cells, respectively. The results show that the 5Ј region of the 5-HT1A receptor gene contains a negative regulatory region between Ϫ1519 and Ϫ1590 bp. Specific deletion of the putative repressor region between Ϫ1590/Ϫ1519 in the 2300/1519-luciferase construct revealed a striking increase in luciferase activity that attained 47.5 Ϯ 6.6-and 52.2 Ϯ 7.2-fold basal in RN46A and L6 cells, respectively, which represents an enhancement of over 10-fold compared with the Ϫ2300 construct. This result indicates the presence of a strong repressor activity between Ϫ1590/Ϫ1519 bp that inhibits the activity of the 5-HT1A receptor gene. The 1590-luciferase construct displayed weak luciferase activity that was 1-to 1.5-fold basal, suggesting the presence of enhancer elements between Ϫ2300 and Ϫ1590 bp. Luciferase activity of the 5-HT1A receptor gene was markedly enhanced in the 1519-luciferase construct to 6.2 Ϯ 0.45-and 5.1 Ϯ 0.58-fold basal in RN46A cells and L6 cells, respectively. Therefore, upon removal of the Ϫ1590/Ϫ1519 region in the Ϫ2300/Ϫ1519 or Ϫ1519 constructs, the 5-HT1A promoter was derepressed compared with the Ϫ2300 and Ϫ1590 constructs, respectively.
Cell-specific Protein Binding to the Proximal Repressor Region of 5-HT1A Receptor Gene-To determine whether nuclear proteins interact with specific sequences within the proximal repressive element (Ϫ1590/Ϫ1519 bp), DNase I protection and EMSA were conducted. For the DNase I protection assay a 208-bp fragment that contained the 71-bp sequence between Ϫ1590 and Ϫ1519 was end-labeled and incubated with nuclear proteins prepared from L6 and RN46A cells. A discrete segment spanning from Ϫ1524 to Ϫ1555 bp of the 5-HT1A receptor gene was protected from DNase I digestion at low but not high concentrations of DNase I (Fig. 2). The protected region was similar for nuclear extracts from both RN46A cells and L6 cells. The sequence of this 31-bp region (Table I, lane 1) did not match known DNA elements and therefore may correspond to a novel element.
To further investigate the DNA binding activities of proteins from RN46A and L6 cells, double-stranded 31-bp oligonucleotides were designed according to the nucleotide sequence of the segment of the 5-HT1A repressor that exhibited a clear DNase I footprint (Ϫ1555/Ϫ1524 bp; Table I). The 31-bp probe was labeled and incubated with nuclear protein extracts from RN46A and L6 cells and examined by EMSA. A single specific protein complex was identified from RN46A extracts that was competed by excess unlabeled oligonucleotide but was not dis- 5Ј-ATTTAAGTTTCGCGCCTTTTC-3Ј  d and g). A protein-DNA-interacting segment spanned from Ϫ1524 to Ϫ1555 bp upstream from the translational start site. placed by the unrelated E2F binding site sequence (Fig. 3A). This specific interaction provides further evidence that the 31-bp segment spanning from Ϫ1524 to Ϫ1555 is recognized by a nuclear protein or protein complex from RN46A cells. To further localize the site of protein binding, competition experiments were done using double-stranded oligonucleotides targeted within the 31-bp segment ( Table I). The 15-bp segment (Fig. 3B) competed weakly for the protein binding site of the 31-bp oligonucleotide, whereas the 14-bp primer competed as effectively as the unlabeled 31-bp primer to eliminate protein binding to the labeled 31-bp oligonucleotide (Fig. 3C). Thus the DNA element bound by RN46A proteins is located primarily in the region Ϫ1554/Ϫ1540 bp upstream from the translational start site. Because no consensus DNA element has been found so far, this 14-bp element may represent a novel protein interaction site.
To examine whether the same protein complex was present in L6 myoblasts, EMSA was performed using nuclear extracts from L6 and RN46A cells and the 31-bp oligonucleotide as probe (Fig. 4A). Interestingly, the L6 lanes displayed two nuclear protein complexes: the lower complex had an identical migration to that in the RN46A extracts, whereas the upper complex was not detected in RN46A extracts. The 14-bp primer was equally effective in competition of the lower protein com-  Table I for sequences). A, identification of a single specific RN46A protein complex with the 31-bp probe. A single specific protein complex was competed by a 20-or 100-fold molar excess of cold 31 oligo but not by a 100-fold excess of the unrelated E2F oligonucleotide. Note that the lower bands were not competed by cold 31 and represent nonspecific protein-DNA interactions. B, competition assay with 15-bp oligo. The specific protein complex was competed by cold 31 oligo, but only weakly competed by 100-fold but not 20-fold cold 15 bp. C, effective competition of the RN46A protein-31-bp probe complex by the unlabeled 14-bp oligonucleotide. The labeled 31-oligo probe was used and the nuclear extract was from RN46A cells. The specific protein complex was entirely competed by cold 31 oligo and a 20-or 100-fold excess of 14 bp indicating that the 14-bp element competes with high affinity for binding of the labeled 31-bp element to the protein complex.  Table I) as indicated. A, nuclear extracts from L6 cells display two components of binding to the 31-bp probe. The lower L6 protein-DNA complex is identical in migration to the single species from RN46A nuclei. Note that both mut21 and 14-bp primers competed effectively the lower complex but not the upper complex in L6 cells. B, identical binding of L6 or RN46A proteins to the 14-bp probe. In the presence of both L6 and RN46A nuclear proteins, a single major complex bound the labeled 14-bp element in the absence of competitor, and was competed by both cold 31 and cold 14 but inefficiently or not at all by mut14. This indicates that the DNA binding site lies in the 14-bp sequence and is destroyed by the mutations incorporated in mut14.
plex but was relatively ineffective in competing with the upper complex. The 14-bp oligonucleotide was labeled and used as a probe to conduct the EMSA using protein extracts from both the RN46A and L6 cell lines (Fig. 4B). One specific band was detected in both L6 and RN46A cells and was competed by the unlabeled 14-bp oligonucleotide. This result indicates that the 14-bp element is essential and sufficient for DNA binding of an identical regulatory protein complex from both RN46A cells and L6 cells (the lower complex, Fig. 4A). However, the upper complex is absent from RN46A cells and represents protein binding to an adjacent site on the 31-bp probe, because it is weakly competed for by the 14-bp primer and does not bind to the labeled 14-bp primer. Thus nuclear proteins from L6 cells bind to the two adjacent complexes, one of which is identical to the raphe RN46A nuclear complex.
To generate inactivating mutations of the 14-bp element, two mutants within the Ϫ1554/Ϫ1540-bp region (mut21 and mut14, Table I) were designed to preserve nucleotide content.
The mut21 mutant converted AGCA to CAAG, while mut14 rearranged eight nucleotides within the element. These primers were examined in competition assays for protein binding to labeled 31-bp or 14-bp oligonucleotides, respectively. Like the 14-bp nonmutated primer, the mut21 mutant did not compete with the upper complex of L6 nuclei, but competed equally well for the lower complex present in both L6 and RN46A cells (Fig.  4A). By contrast, rearrangement of 8 of 14 nucleotides in the mut14 oligonucleotide completely (for RN46A) or substantially (for L6) impaired the ability to compete for protein binding to the intact 14-bp primer (Fig. 4B). Thus, the mutation in mut14 inactivates protein binding to the lower complex in L6 cells and completely blocks protein interaction in RN46A cells.
The location of the upper protein-DNA binding complex in L6 cells was probed using double-stranded oligonucleotides directed at the 3Ј portion of the 31-bp repressor located adjacent to the 14-bp site. Both 15-bp and 12-bp probes were used as competitors in EMSA using L6 extracts, and they failed to compete the two protein-DNA complexes (Fig. 5). The 14-bp sequence competed only the lower complex in L6 cells (Fig. 4A). However, the combination of the 14-and 12-bp sequences competed effectively and eliminated the two protein complexes associated with the labeled 31-bp oligonucleotide (Fig. 5A). Mutation of the 14-and 12-bp sequences (mut14 ϩ mut12, Fig.  5A) substantially impaired their ability to compete for protein binding. Because both 12-mer and 14-mer oligonucleotides were necessary to compete the upper complex, it suggested that the upper protein complex interacts with the 12-mer site and a small portion of the adjacent 14-mer site. To further examine the importance of these segments, mutations of the 14-mer, 12-mer, or both were incorporated in the 31-mer oligonucleotide as competitors in EMSA using L6 extracts. The 31-mer with the mutated 14-mer site selectively displaced the upper complex, the 31-mer with the mutated 12-mer site blocked the lower complex as expected, and the 31-mer with both mutations was ineffective in competition (Fig. 5B). These results indicate that the 12-mer segment is required for the upper protein-DNA complex, and the 14-mer sequence is essential for formation of the lower complex.
Cell-specific Repressor Activity of the 14-bp Element of the 5-HT1A Receptor Gene-We examined whether protein interaction with the 14-bp sequence mediates repression of the 5-HT1A receptor gene by mutating the 2300-luciferase construct (Ϫ2300) to incorporate the changes of the mut14 oligonucleotide (Ϫ2300 m1, Fig. 6), which failed to displace protein binding to the labeled 14-bp oligonucleotide. When transfected in RN46A cells, mutation of the 14-bp element (Ϫ2300 m1) resulted in a 36.8 Ϯ 6.2-fold basal induction of luciferase activity that was 80% of the activity observed upon deletion of the entire region between Ϫ1590 and Ϫ1519 bp (Ϫ2300/Ϫ1519). However, upon transfection in L6 cells luciferase activity was not altered between the wild type and mutated constructs, with average luciferase activities of 4.6 Ϯ 0.8-and 4.7 Ϯ 0.7-fold basal for the 2300-luciferase and 2300 m1-luciferase constructs, respectively. In contrast to the RN46A cell line, intact protein-DNA interaction at the 14-bp site was not essential for repression of the transcriptional activity of the 5-HT1A gene in L6 cell. Therefore, the 14-bp element suppresses the basal expression of 5-HT1A receptor in the raphe RN46A cell line. However, additional DNA elements maintain repression of the 5-HT1A receptor gene in L6 cells. This additional repressor may correspond to the second protein-DNA binding complex identified by EMSA using L6 nuclear extracts. In the same way, the 2300-luciferase construct was mutated at the 12-mer site (Ϫ2300 m2). Upon transfection in RN46A or L6 cells, luciferase activities were not significantly different from the  Table  I) were unlabeled double-stranded 12-bp (Ϫ1543/Ϫ1532; 12), 15-bp (15), 14-and 12-bp (14ϩ12), or combined mutated versions (mut14 ϩ mut12) as indicated. Note that the combination of 14-and 12-bp primers competed effectively the upper and lower bands in L6 cells but 12-and 15-bp alone or mut14 ϩ mut12 did not compete. B, the competition oligonucleotides were the double-stranded 31-mer incorporating mutations of the14-bp, 12-bp, or both sites (m1, m2, and m3, respectively) as shown. Note that m1 displayed only the upper complex, while m2 displaced the lower complex. nonmutated Ϫ2300 construct (Fig. 6, lane 4). Because this mutation eliminated only the upper complex in EMSA (Fig. 5), these functional data indicate that the 14-mer complex is sufficient to repress the 5-HT1A receptor gene in both RN46A and L6 cells.
We examined whether combined mutation to disrupt protein binding to the 14-bp and 12-bp elements would derepress the 5-HT1A receptor gene in L6 cells. The 2300-luciferase construct (Ϫ2300) was mutated to inactivate both elements (Ϫ2300 m3, Table I), and luciferase activity was measured in transfected RN46A and L6 cells (Fig. 6, lane 5). The Ϫ2300 m1 (14-bp mutant only) and Ϫ2300 m3 (14-and 12-bp elements mutated) constructs displayed equivalent activities in RN46A cells, hence, no additional derepression was observed in the RN46A cells upon mutation of the 12-bp element. In contrast, although the 2300 m1 mutant remained repressed in L6 cells, the Ϫ2300 m3 mutant displayed a 37.2 Ϯ 3.1-fold basal induction of luciferase activity that was nearly equivalent to the induction observed for the deletion mutant (Ϫ2300/Ϫ1519). Therefore, additional protein-DNA interactions at the 12-bp element in L6 cells maintain repression of the 5-HT1A receptor gene upon inactivation of the 14-bp element, suggesting that the upper protein-DNA complex provides a novel dual repression mechanism to regulate 5-HT1A gene transcription in the receptornegative cell line.
Analogous Repressor Activities in SN48 and C6 Cells-We investigated whether the single versus dual site repression could be extended to other 5-HT1A receptor-positive (SN48) or -negative (C6 glioma) cell lines, respectively. To examine protein-DNA interactions, nuclear extracts from RN46A, L6, C6, and differentiated SN48 cells (18) were incubated with labeled 31-bp probe and analyzed by EMSA (Fig. 7A). SN48 and C6 cells displayed similar protein-DNA complexes as those of RN46A and L6 cells, respectively, suggesting that a single repressor operates in SN48 cells, and a dual repressor is present in C6 cells. To examine the importance of the 14-and 12-bp elements in these cells, the luciferase activities of the 5-HT1A receptor gene constructs used above were measured in SN48 and C6 cell lines (Fig. 7B). A similar pattern of activities was observed. Mutation of the 14-bp element (2300 m1) derepressed the gene to nearly the same extent as the deletion mutant (Ϫ2300/Ϫ1519) in SN48 cells as observed in RN46A cells. However, mutation of both 14-and 12-bp elements (2300 m3) was required to derepress the gene in receptor-negative C6 and L6 cell lines. Thus, the 14-bp element appears to be crucial for regulation of 5-HT1A receptor expression in receptor-positive cells, whereas a dual mechanism appears to repress the gene in receptor-negative cells.

A Novel 14-bp Repressor Element of the 5-HT1A Receptor
Gene-Repression of gene transcription appears to be an important mechanism for determining neuronal specificity of expression (22,23). Hence we sought to identify novel DNA elements that repress the 5-HT1A receptor gene in neuronal and non-neuronal cells. In the present study we have identified a proximal repressor region located between Ϫ1590 and Ϫ1519 bp and a putative positive regulatory region located between Ϫ2300 and Ϫ1590 bp upstream of the translational initiation site using a series of finer deletions of the 5-HT1A receptor gene fused to luciferase. Specific deletion of the putative repressor region between Ϫ1590 and Ϫ1519 bp revealed a strik-FIG. 6. Derepression of the 5-HT1A receptor gene upon mutation of the 14-bp element in raphe RN46A but not in L6 cells. Luciferase constructs of the 5-HT1A receptor gene: Ϫ2300-luciferase; Ϫ2300 construct containing the mutated 14-bp element (Ϫ2300 m1), mutated 12-bp element (Ϫ2300 m2), or mutations of both 14-bp and adjacent 12-bp elements (Ϫ2300 m3) (see Table I for sequences); and 2300/1519-luciferase, the deletion mutant lacking the 71-bp repressor region (Ϫ1590/Ϫ1519) are shown. Luciferase activities were measured after transfection of these constructs in RN46A or L6 cells as indicated. Note that complete derepression is observed upon mutation of the 14-bp element in either the 2300 m1 or 2300 m3 constructs for RN46A cells (lanes 3 and 5), but only upon mutation of both 14-bp and 12-bp sites in the 2300 m3 construct in L6 cells (lane 5). The activity of the pGL3-promoter (pGL3P) construct was measured as a positive control. Data are expressed as luciferase over ␤-galactosidase activity of triplicate samples normalized to pGL3-Basic activity and represent the mean Ϯ S.E. of at least three independent experiments.
ing 10-fold increase in luciferase activity (Fig. 1). The site of protein binding was further localized to a region spanning from Ϫ1524 to Ϫ1555 bp (Fig. 2). A 31-bp oligonucleotide based on this sequence interacted with a single protein complex of identical migration in EMSA from nuclei of 5-HT1A receptor-expressing cells; however, an additional complex was present in receptor-negative cell lines (Figs. 4 and 7). The minimal element essential for protein binding was identified within a 14-bp sequence (Ϫ1554/Ϫ1540 bp), although the DNase I protection assay (Fig. 2) suggested that other interactions with adjacent sequences may also occur. The sequence of the 14-bp element is novel and does not contain any recognizable repeats or palindromes. Sequence analysis of the 14-bp element aligned with two half-sites (CATAAG and GCAAGCC) for nuclear factor 1 (NF1) binding. Apart from its traditional role as a transcriptional activator, NF1 functions as a silencer in the Pit-1 gene (24). However, supershift using anti-NF1 antibody in EMSA failed to identify NF1 in protein complexes bound to the 31-bp probe (data not shown). A search of the GenBank TM data base identified a matching sequence for the 14-bp element (CATAAGCAAGCC) located at Ϫ1357/Ϫ1368 bp relative to the transcription start site in the human neurotrophin-3 promoter (25). This location is within a strong repressor region (Ϫ2914/ Ϫ514 bp) that preferentially silences the neurotrophin-3 gene in nonexpressing leukemia cells, but also partially in neurotrophin-3-positive glioma cells. Thus the 14-bp element may mediate repression of other genes in addition to the 5-HT1A receptor. We have also identified a sequence with high homology to the 14-bp element located in a repressor region of the human 5-HT1A receptor gene. 2 Preliminary results using the 31-bp element to probe a cDNA expression library indicate that a novel protein interacts with this sequence. 3 Further identification of this protein is undergoing. Thus the 14-bp repressor element appears to represent a novel element that interacts with a new class of DNA binding proteins.
Single-versus Dual-Site Repression of the 5-HT1A Receptor Gene-An intriguing finding was the crucial role of the 14-bp element in repression of the 5-HT1A receptor gene selectively in 5-HT1A receptor-expressing neuronal cells. Upon mutation to inactivate protein interaction with the 14-bp element the extent of transcriptional derepression was nearly identical to that observed with deletion of the entire 71-bp repressor region, suggesting that the 14-bp element largely accounts for the inhibitory activity of this region in these neuronal cells. The RN46A and SN48 cell lines examined are derived from different neuronal populations that were transformed by different means. The RN46A cells are derived from transformation of e13 raphe cells using the temperature-sensitive SV40 T antigen (19), whereas the SN48 cells are hybrid p21 septal X N18TG2 neuroblastoma fusion cells (26). However, both cell lines share the property of expressing a low level of 5-HT1A receptors, which can be increased upon differentiation of the cells (18,19). Our results suggest that endogenous expression of 5-HT1A receptors is correlated with the presence of a single FIG. 7. Cell-specific repressor activities in 5-HT1A receptor-positive SN48 septal cells and receptor-negative C6 glioma cells. A, cell-specific protein-DNA complexes. Nuclear extracts from RN46A, L6, C6, and differentiated SN48 cells were incubated with 32 P-endlabeled 31-oligo for 25 min and were separated on a 5% nondenaturing polyacrylamide gel at 4°C. Nuclear extracts from SN48 and C6 cells display single and dual protein-DNA binding complexes similar to those of RN46A raphe and L6 myoblast cells, respectively. B, cell-specific repressor activity. Luciferase constructs of the 5-HT1A promoter were transfected into SN48 cells and C6 cells, and luciferase activity was measured. Shown is an average of at least three independent experiments and luciferase over ␤-galactosidase activity is normalized to the pGL3-Basic vector activity. Lanes: pGL3-promoter (1), 2300-(2), 2300 m1-(3), 2300 m3-(4), 2300/1519 (5), 1590-(6), and 1519-luciferase (7). Note that the 2300 m1 construct was derepressed only in receptor-positive SN48 cells (as for RN46A cells), whereas mutation of both 14-bp and 12-bp sites in the 2300 m2 construct was necessary for derepression in the receptor-negative C6 cells.
protein-DNA complex at the proximal repressor. The low level of 5-HT1A receptor expression in these transformed cell lines may reflect the presence of elevated levels of the repressor protein compared with postmitotic raphe or septal neurons.
In contrast to the neuronal cell lines examined above, the non-neuronal L6 myoblast and C6 glioma cells lack 5-HT1A receptors (21). In these cell lines there was no change in the extent of repression of the 5-HT1A receptor gene upon mutation of the 14-bp element. The maintenance of repression in L6 and C6 cells upon mutation of the 14-bp element appears to result from a second protein complex in these cells that binds to a 12-bp segment located adjacent to the 14-bp element, because mutation of both 14-and 12-bp segments blocked repression in receptor-negative L6 and C6 cells. We postulate that two repressor elements are involved, because the 14-bp element selectively competed one, but not both, protein-DNA complexes in EMSA with the 31-bp probe. The upper protein complex may result not only from direct protein binding at a second DNA site but also from indirect protein-protein interactions to form a stable complex with an enlarged DNA recognition domain, because both 14-and 12-bp oligonucleotides were required to compete for protein binding to the 31-bp probe. Elucidation of the exact mechanism for repression in 5-HT1A receptor-negative cells will await the identification of the proteins involved.
Gene Repression as a Regulatory Mechanism in Neurons-Both positive and negative cis-acting elements and factors are required for correct tissue-specific expression (22,27). Previously, repression of neuron-specific genes was thought to occur only in non-neuronal tissues that robustly express neural restriction proteins such as the silencer protein REST/NRSF (28 -30). The concept that gene repression plays important roles in neurons is gaining wider attention. For example, mutation of the RE-1/NRSE in the BDNF promoter does not influence extra-neuronal BDNF expression, but enhances both basal and kainate-induced induction of BDNF in hippocampal neurons (31). Furthermore, REST is expressed in brain (e.g., midbrain, pons/medulla, and hippocampus) and its expression is induced by kainic acid treatment (32). This indicates that REST may regulate gene expression in neurons as well as silencing expression in non-neuronal cells. Another example of repressor function in neurons is the SNOG element that restricts the expression of the GAP-43 gene to neurons (33). However this element also exerts repression in cortical neurons, and a binding protein that associates with the SNOG element is present in neurons, although at lower apparent levels than in non-neuronal cells. By analogy, our results indicate that the 14-bp repressor of the 5-HT1A receptor gene plays a crucial role in the inhibitory regulation of the 5-HT1A receptor in receptor-expressing raphe and septal cells. The precise role of the 14-bp element in repression of the 5-HT1A receptor in vivo remains to be elucidated. However, a reduction of 5-HT1A receptor gene repression at the raphe could lead to elevated levels of 5-HT1A autoreceptors, which have been correlated with major depression (11). On the other hand, an increase in gene repression would reduce the number of 5-HT1A receptors expressed in the raphe and limbic system, a condition that is thought to underlie generalized anxiety syndromes (8 -10). Unlike cell-specific repression models involving non-neuronal expression of a single repressor (such as REST), we find that in 5-HT1A receptor-negative cells there is a dual repressor involving an additional protein complex that mediates repression of the gene. This suggests that the neural specificity of 5-HT1A receptor expression may be achieved in large part by differential repression using dual DNA elements.
In summary, our results provide evidence for an alternate paradigm of single and dual repressors for cell-specific regula-tion of the rat 5-HT1A receptor gene. We have identified a novel 14-bp repressor element within the proximal repressive region that mediates selective repression of the 5-HT1A receptor gene in raphe and septal cells that express the 5-HT1A receptor. By contrast, in cells that do not express the 5-HT1A receptor, two adjacent repressors and their associated protein complexes provide a novel dual repression mechanism. The 14-bp element is likely to play an important role in the regulation of basal 5-HT1A receptor expression in neuronal cells that express the receptor. Conversely, the dual repression mechanism helps to ensure that the gene is inactivated in cells that do not express the receptor, such as non-neuronal cells. Because alteration in the level of 5-HT1A receptors in the raphe nuclei and limbic system is associated with mental disorders, mechanisms that selectively regulate 5-HT1A receptor gene transcription in the raphe nuclei could provide a molecular basis for illnesses such as major depression or anxiety disorders.